Datasets:
J0nasW
/
paperswithcode

Dataset card Data Studio Files
xet
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
8f536069-1bce-4f6f-aa79-34a2c4eb0253
fully-connected-tensor-network-decomposition
2110.08754
null
https://arxiv.org/abs/2110.08754v1
https://arxiv.org/pdf/2110.08754v1.pdf
Fully-Connected Tensor Network Decomposition for Robust Tensor Completion Problem
The robust tensor completion (RTC) problem, which aims to reconstruct a low-rank tensor from partially observed tensor contaminated by a sparse tensor, has received increasing attention. In this paper, by leveraging the superior expression of the fully-connected tensor network (FCTN) decomposition, we propose a $\textbf{FCTN}$-based $\textbf{r}$obust $\textbf{c}$onvex optimization model (RC-FCTN) for the RTC problem. Then, we rigorously establish the exact recovery guarantee for the RC-FCTN. For solving the constrained optimization model RC-FCTN, we develop an alternating direction method of multipliers (ADMM)-based algorithm, which enjoys the global convergence guarantee. Moreover, we suggest a $\textbf{FCTN}$-based $\textbf{r}$obust $\textbf{n}$on$\textbf{c}$onvex optimization model (RNC-FCTN) for the RTC problem. A proximal alternating minimization (PAM)-based algorithm is developed to solve the proposed RNC-FCTN. Meanwhile, we theoretically derive the convergence of the PAM-based algorithm. Comprehensive numerical experiments in several applications, such as video completion and video background subtraction, demonstrate that proposed methods are superior to several state-of-the-art methods.
['Ting-Zhu Huang', 'Yu-Bang Zheng', 'Guang-Jing Song', 'Xi-Le Zhao', 'Yun-Yang Liu']
2021-10-17
null
null
null
null
['video-background-subtraction']
['computer-vision']
[ 8.51354674e-02 -2.37460762e-01 1.19456105e-01 -4.15051430e-02 -6.30761683e-01 -2.17856184e-01 3.33978571e-02 -5.21825910e-01 -2.47295693e-01 4.21409279e-01 1.67353526e-01 -4.32430357e-01 -6.64021134e-01 -2.92979896e-01 -8.23916316e-01 -9.89918590e-01 -3.44946355e-01 -4.28879708e-02 -3.96673262e-01 -2.84190029e-01 8.46060812e-02 2.57764518e-01 -1.01786315e+00 8.52407441e-02 9.35159504e-01 1.38802826e+00 1.80958524e-01 4.19774920e-01 1.21004269e-01 1.03569031e+00 5.84878810e-02 -3.74599904e-01 5.18886983e-01 -2.44838834e-01 -7.52911270e-01 5.41094661e-01 3.77787203e-01 -4.98751700e-01 -6.95914388e-01 1.38340414e+00 4.43775170e-02 3.97783488e-01 3.55937451e-01 -1.16520667e+00 -6.28831744e-01 2.11429939e-01 -1.03254700e+00 2.56613165e-01 1.21582657e-01 -1.63372695e-01 9.79287386e-01 -1.50087893e+00 4.59979922e-01 1.03770292e+00 3.97441685e-01 3.30351144e-02 -1.07395184e+00 -5.66938758e-01 5.76152623e-01 1.75868019e-01 -1.58850718e+00 -1.70717120e-01 8.33745539e-01 -5.99398375e-01 2.88165122e-01 5.84116876e-01 5.51544189e-01 3.67598593e-01 -2.32294962e-01 9.72969949e-01 1.13193059e+00 -2.53691733e-01 -1.22455448e-01 -5.46374261e-01 1.62638783e-01 9.83330607e-01 4.16850030e-01 -1.36369318e-01 -4.76812422e-01 -1.14770576e-01 1.18759584e+00 3.03315729e-01 -6.19583726e-01 -6.11354671e-02 -1.49242353e+00 7.36575603e-01 3.77236307e-01 1.01845406e-01 -5.14122069e-01 2.69330770e-01 2.34485000e-01 9.38743576e-02 6.15682662e-01 -2.19303757e-01 -2.19082803e-01 7.13735670e-02 -8.68848026e-01 1.29699364e-01 4.12875354e-01 1.33756268e+00 7.49644041e-01 4.69370991e-01 -1.23825580e-01 9.02435958e-01 5.44868052e-01 8.91230643e-01 -1.28969401e-01 -1.50430155e+00 9.08452153e-01 3.75651181e-01 1.71594739e-01 -1.47924936e+00 -1.67171702e-01 -5.31161189e-01 -1.39141524e+00 -2.37294167e-01 2.65490919e-01 -1.20032862e-01 -3.81898731e-01 1.52554870e+00 4.95957226e-01 3.66610050e-01 -1.66748360e-01 1.41376412e+00 3.60247731e-01 7.89616227e-01 -5.06053269e-01 -6.02304578e-01 1.31130672e+00 -6.97983623e-01 -6.16769850e-01 1.27028540e-01 5.77595890e-01 -9.13219929e-01 7.47599006e-01 6.77331448e-01 -1.08240044e+00 -2.56738365e-01 -8.27836633e-01 2.13707029e-03 5.38481772e-01 4.76173848e-01 6.79011345e-01 3.78932714e-01 -6.66524947e-01 4.27525848e-01 -7.79719889e-01 8.43684375e-02 2.32578740e-01 3.54126841e-01 -4.27378535e-01 -6.29997969e-01 -6.70298100e-01 2.48450384e-01 -3.36473100e-02 1.04804003e+00 -1.04046440e+00 -5.18661559e-01 -5.89284778e-01 -2.38434508e-01 8.34202588e-01 -4.19187754e-01 7.09349751e-01 -5.90984225e-01 -1.37849343e+00 5.35716653e-01 -2.33304918e-01 6.76096976e-02 4.76361752e-01 -2.61096507e-01 -3.43534887e-01 6.36849463e-01 2.44537145e-01 -1.21866604e-02 1.21995962e+00 -1.17924249e+00 -2.80073166e-01 -3.94554347e-01 1.78703189e-01 1.33699030e-01 -3.83229852e-01 1.34207517e-01 -6.13878787e-01 -1.06820393e+00 8.09773147e-01 -1.01721466e+00 -3.45479816e-01 7.42802098e-02 -5.28267682e-01 1.31334603e-01 6.16882026e-01 -1.02742815e+00 1.24973822e+00 -2.17836142e+00 6.84327543e-01 3.59283239e-01 6.67442918e-01 1.60713211e-01 -7.07067177e-02 2.89575756e-01 -1.65586665e-01 -1.37175977e-01 -2.69253343e-01 -4.23052251e-01 -1.70491174e-01 3.08677793e-01 -2.18014434e-01 9.59841430e-01 -1.26525417e-01 4.93957669e-01 -8.38627040e-01 -4.13816005e-01 3.66000459e-02 3.55281979e-01 -5.72623134e-01 -4.23824741e-03 -5.57859195e-03 5.34168124e-01 -8.20627272e-01 7.56685257e-01 1.12151062e+00 -3.33579153e-01 2.06361189e-01 -6.27385676e-01 -4.04640526e-01 -4.04626518e-01 -1.55858767e+00 1.61961126e+00 -3.50437276e-02 1.78020090e-01 9.51151431e-01 -1.47905052e+00 6.46469295e-01 2.41944104e-01 9.21364367e-01 -4.61972773e-01 1.64977238e-01 4.22266304e-01 -3.20124567e-01 -4.63167518e-01 4.17648643e-01 -3.14596862e-01 1.51155144e-01 5.04605174e-01 -2.05941200e-01 3.18467319e-01 3.65643591e-01 5.47502875e-01 1.11034727e+00 2.29779705e-01 -3.56560141e-01 -5.88462114e-01 1.05721831e+00 -2.37390399e-01 9.55402672e-01 3.88229489e-01 -1.45724952e-01 4.53652829e-01 5.08709610e-01 -2.53165752e-01 -9.02207434e-01 -6.47191167e-01 4.51556295e-02 7.95952916e-01 1.31729499e-01 -3.33984524e-01 -6.12407804e-01 -5.50036013e-01 -2.52530038e-01 1.02342673e-01 -3.49165142e-01 2.09858835e-01 -7.28879929e-01 -9.15187180e-01 2.25428641e-01 1.86773732e-01 7.15963781e-01 -2.52398878e-01 4.11798581e-02 1.15710095e-01 -7.10096121e-01 -1.42624998e+00 -8.94054651e-01 -3.91871810e-01 -1.02266467e+00 -9.97940779e-01 -1.11777425e+00 -4.66035098e-01 1.14510024e+00 9.95315731e-01 5.35382748e-01 3.88758093e-01 9.99873728e-02 6.07914627e-01 -4.71163213e-01 2.09687114e-01 3.26582998e-01 -4.75017428e-01 3.58278006e-01 7.74245560e-01 -2.24321589e-01 -5.30477047e-01 -8.93295348e-01 6.30986631e-01 -1.20266891e+00 3.30488831e-01 5.55847347e-01 7.43163645e-01 9.25819337e-01 1.94675773e-01 1.32866958e-02 -5.66872060e-01 3.03058535e-01 -3.21722329e-01 -9.76187348e-01 1.02546908e-01 -5.44645190e-01 4.20345776e-02 6.98605955e-01 -3.04103225e-01 -8.42116416e-01 -1.19480863e-01 3.28205109e-01 -1.15192962e+00 6.87794328e-01 9.03257906e-01 -1.58858821e-01 -3.04128855e-01 8.53466392e-02 5.26966453e-01 -3.55544724e-02 -8.07168245e-01 4.71750617e-01 2.81952381e-01 5.83441138e-01 -1.05624533e+00 1.10055065e+00 9.39318478e-01 2.73863405e-01 -8.18888962e-01 -1.03546476e+00 -6.89425111e-01 -4.13777977e-01 -4.09224987e-01 6.83162808e-01 -1.01186907e+00 -1.18389797e+00 3.89001191e-01 -9.33021367e-01 -1.57908231e-01 7.23509043e-02 9.39394236e-01 -4.60571915e-01 9.25301909e-01 -5.87226689e-01 -9.60291326e-01 -2.18453497e-01 -1.09984994e+00 7.97586501e-01 -2.08230466e-01 5.82088113e-01 -7.17492819e-01 -2.38812149e-01 9.56246614e-01 1.77020907e-01 3.55088323e-01 5.42459190e-01 3.18392664e-01 -1.00588632e+00 -1.90722376e-01 -8.12379181e-01 8.46546113e-01 -1.70061827e-01 -3.48451555e-01 -2.97246575e-01 -5.76785207e-01 3.84924322e-01 2.05282301e-01 6.27745390e-01 2.93223053e-01 1.09241879e+00 -8.32201898e-01 5.71192093e-02 8.99123371e-01 1.49933612e+00 -2.81674325e-01 6.95106447e-01 1.10717848e-01 1.10156608e+00 4.71342713e-01 6.74189031e-01 7.31193364e-01 5.39122581e-01 5.14172673e-01 5.87406576e-01 4.79529910e-02 9.27226618e-02 -1.06438771e-02 4.84396368e-01 1.53703046e+00 -6.41570628e-01 2.80348390e-01 -5.84308684e-01 4.16689456e-01 -2.06452322e+00 -7.02294409e-01 -6.59502923e-01 2.32776976e+00 4.82637346e-01 -2.33427837e-01 -1.74910855e-02 2.08227023e-01 8.46899807e-01 1.84462577e-01 -4.40197229e-01 9.44915563e-02 -9.35692191e-02 1.05241515e-01 6.07008636e-01 4.23322618e-01 -8.20625484e-01 5.40142953e-01 3.96001506e+00 8.92584920e-01 -9.74593222e-01 3.99596304e-01 2.91121215e-01 -1.97849944e-01 -2.23047450e-01 3.28193963e-01 -3.65354419e-01 5.38494885e-01 5.21095335e-01 -6.17957348e-03 9.44777966e-01 6.45261049e-01 5.85849643e-01 -2.08860431e-02 -5.88788331e-01 1.26975417e+00 1.37906015e-01 -1.29239380e+00 -7.32369274e-02 5.07333815e-01 7.74008393e-01 -2.05388144e-01 7.55574331e-02 7.22545162e-02 8.91017690e-02 -5.44977725e-01 9.00932372e-01 4.82999057e-01 1.05184543e+00 -4.59130585e-01 3.52254033e-01 8.78789574e-02 -1.57921982e+00 -1.20796494e-01 -5.02155244e-01 -8.94172564e-02 1.96724638e-01 1.06957042e+00 1.41046479e-01 1.12620687e+00 8.17552388e-01 8.74724448e-01 -4.63019237e-02 7.51310289e-01 -8.70167166e-02 5.70324779e-01 -3.49608690e-01 3.70057762e-01 4.26841259e-01 -9.06531572e-01 9.08632278e-01 7.74352789e-01 3.84118795e-01 7.00995564e-01 3.61659497e-01 5.76319337e-01 -1.47759348e-01 2.24133059e-01 2.53696702e-02 -7.92275369e-02 1.01862825e-01 1.44415331e+00 -6.13081634e-01 -1.39497146e-01 -3.77365679e-01 9.69651699e-01 2.63449967e-01 7.53383636e-01 -7.53328145e-01 -2.00324491e-01 5.73045373e-01 4.94746715e-02 5.43518305e-01 -6.54546916e-01 1.40438104e-04 -1.75472355e+00 5.22859275e-01 -9.50276554e-01 3.50040346e-01 -7.80979037e-01 -1.14269876e+00 3.07941079e-01 -4.46145050e-02 -1.54090071e+00 5.80781817e-01 -6.51876569e-01 -2.90266126e-01 6.68682218e-01 -1.45967281e+00 -1.11385524e+00 -4.41797554e-01 1.12332821e+00 -1.51777733e-02 5.02957776e-03 2.43686318e-01 8.36825073e-01 -1.10283482e+00 2.02091366e-01 5.32231271e-01 3.10470670e-01 6.31298125e-02 -8.13175499e-01 -4.46581572e-01 1.27270293e+00 -2.71091819e-01 7.68480062e-01 3.63756269e-01 -5.12474358e-01 -2.12300301e+00 -1.25064623e+00 2.95267552e-01 2.23010689e-01 9.03445661e-01 -1.85688332e-01 -6.18245661e-01 8.13489139e-01 -2.86525875e-01 3.64978224e-01 5.71690977e-01 -1.57150716e-01 -5.22001386e-01 -4.11142200e-01 -9.70771194e-01 5.10214150e-01 1.15101349e+00 -5.39833069e-01 3.45674902e-02 6.12569869e-01 6.68459296e-01 -3.79557371e-01 -1.18356824e+00 3.08690280e-01 4.20992136e-01 -7.77269661e-01 1.13124049e+00 -4.42561299e-01 4.29372609e-01 -6.80560291e-01 -7.75809824e-01 -6.68578386e-01 -3.85004073e-01 -1.22709441e+00 -4.67811495e-01 9.44470763e-01 2.06589717e-02 -5.98072946e-01 6.19082570e-01 5.87542474e-01 -4.47516888e-01 -7.94654846e-01 -1.22963226e+00 -1.02522862e+00 -1.13392048e-01 -6.86040342e-01 1.80185199e-01 9.79221344e-01 -1.71867535e-01 -6.43605962e-02 -7.45014369e-01 3.97537082e-01 1.18852508e+00 4.08109762e-02 6.01785243e-01 -8.11291695e-01 -5.45668662e-01 -1.02859706e-01 -1.43445820e-01 -1.29933548e+00 -1.90002874e-01 -8.55420530e-01 -1.60156086e-01 -1.40784597e+00 3.00758481e-01 -7.47065604e-01 -3.50626141e-01 1.84726283e-01 -1.00202993e-01 2.60706574e-01 5.00417650e-01 7.49271810e-01 -7.85714209e-01 7.12435305e-01 1.49393046e+00 -1.43563628e-01 2.33168751e-01 -2.69447684e-01 -7.51553714e-01 6.00187182e-01 1.37010172e-01 -4.04628843e-01 -3.10208321e-01 -8.04839492e-01 5.68984747e-01 4.38407838e-01 4.20094937e-01 -6.00291193e-01 1.18854672e-01 -3.31501454e-01 -5.82416132e-02 -6.46347761e-01 5.53695440e-01 -6.71061635e-01 9.91060883e-02 3.61952960e-01 1.23538069e-01 -2.23131757e-02 -2.35221937e-01 7.50556767e-01 -4.96301353e-02 -1.35049328e-01 4.70970035e-01 -1.46166950e-01 -1.42376795e-01 7.41727829e-01 -2.14032322e-01 1.16634727e-01 6.36721313e-01 1.92672834e-02 -3.30387324e-01 -4.37902808e-01 -6.46526694e-01 3.86899740e-01 -4.33250815e-02 -1.05146386e-01 8.31400156e-01 -1.60861552e+00 -9.31358814e-01 2.11706311e-02 -1.64892405e-01 2.30401337e-01 5.88012457e-01 1.65560317e+00 -6.06048942e-01 5.39171875e-01 4.18614537e-01 -5.55013716e-01 -9.23838139e-01 5.65995514e-01 1.06578521e-01 -4.03076708e-01 -5.68263888e-01 8.78322184e-01 1.54095963e-01 -3.03166747e-01 -1.01635732e-01 -2.27855667e-01 2.67364472e-01 -3.21926683e-01 4.32522327e-01 1.01221502e+00 -1.09655000e-02 -9.67840195e-01 -2.45323792e-01 6.42788053e-01 1.23901218e-02 -2.91533023e-01 1.36512315e+00 -4.58327562e-01 -7.09131658e-01 -1.07112803e-01 1.48477817e+00 -9.89775360e-02 -1.25929272e+00 -4.70027387e-01 -2.88462132e-01 -8.53903711e-01 4.13880050e-01 -1.29361048e-01 -1.49082136e+00 5.95345318e-01 2.07244560e-01 2.18886863e-02 1.17727590e+00 -4.15634692e-01 9.37654316e-01 3.68824005e-01 6.32444799e-01 -1.01726377e+00 2.61391997e-01 4.38932210e-01 1.02231276e+00 -8.57677162e-01 5.42388201e-01 -6.68983936e-01 -3.57581615e-01 1.04478192e+00 1.88477352e-01 -2.61082977e-01 7.45406628e-01 -4.92300093e-01 -3.65899652e-01 -4.17578876e-01 -3.01089734e-01 -1.03274100e-01 3.42459112e-01 1.23858809e-01 2.42349610e-01 1.17086656e-01 -4.96968776e-01 3.24725032e-01 9.11000744e-02 4.65523452e-02 6.31165922e-01 9.24476981e-01 3.98040935e-02 -9.53104138e-01 -7.43772388e-01 4.77551401e-01 -4.52368975e-01 -8.30696076e-02 7.67099038e-02 3.17968756e-01 1.13023497e-01 1.20375085e+00 -4.60921884e-01 -3.05697143e-01 1.13858186e-01 -4.10039186e-01 6.66212738e-01 -2.64462411e-01 -1.48843348e-01 6.54134750e-01 -1.00446723e-01 -6.23259008e-01 -5.60254276e-01 -8.11482191e-01 -1.25311708e+00 -5.79750538e-01 -5.04471183e-01 2.31751576e-01 5.53212345e-01 9.59075391e-01 2.51389354e-01 1.97049722e-01 1.02891731e+00 -8.30459118e-01 -2.59822518e-01 -7.47536898e-01 -9.14997816e-01 3.76091838e-01 1.70794219e-01 -6.92522109e-01 -3.51619303e-01 1.82646029e-02]
[7.424502849578857, 4.465641021728516]
cc7c6acd-7dae-4f44-9755-226e162d3bf7
2-d-signature-of-images-and-texture
2205.11236
null
https://arxiv.org/abs/2205.11236v1
https://arxiv.org/pdf/2205.11236v1.pdf
2-d signature of images and texture classification
We introduce a proper notion of 2-dimensional signature for images. This object is inspired by the so-called rough paths theory, and it captures many essential features of a 2-dimensional object such as an image. It thus serves as a low-dimensional feature for pattern classification. Here we implement a simple procedure for texture classification. In this context, we show that a low dimensional set of features based on signatures produces an excellent accuracy.
['Samy Tindel', 'Guang Lin', 'Sheng Zhang']
2022-05-10
null
null
null
null
['texture-classification']
['computer-vision']
[ 8.72939453e-02 -1.35404110e-01 -3.29854846e-01 -5.88197887e-01 -3.85419965e-01 -2.23739967e-01 7.85807431e-01 1.11897267e-01 -2.55931228e-01 3.13324749e-01 -1.58542305e-01 -1.28377274e-01 -6.71775281e-01 -1.14298451e+00 -1.54469013e-01 -8.42255473e-01 -7.38922954e-01 3.91192794e-01 4.31945115e-01 -3.92594665e-01 5.17300189e-01 1.13135529e+00 -1.82074118e+00 3.79622906e-01 2.87292391e-01 1.47841144e+00 -1.53637707e-01 6.29693210e-01 -2.13992685e-01 1.81600660e-01 -1.33720309e-01 -6.97095171e-02 2.42608160e-01 -3.91952783e-01 -1.09966445e+00 2.07721502e-01 -5.39531261e-02 5.85232861e-02 -1.76385075e-01 1.08671451e+00 -1.26508236e-01 -2.77946498e-02 1.29256117e+00 -7.32240140e-01 -4.82437938e-01 -7.02878535e-02 -3.07515472e-01 3.48295979e-02 6.05084479e-01 -4.97137934e-01 7.84537554e-01 -1.01716256e+00 7.92996526e-01 1.26370597e+00 8.31312001e-01 1.80558085e-01 -1.04260707e+00 3.52929175e-01 -4.66962278e-01 4.35973227e-01 -1.19778860e+00 -2.13070646e-01 8.75866652e-01 -4.49300349e-01 -2.18981039e-02 5.68230987e-01 4.63669538e-01 6.16109550e-01 7.50335693e-01 8.13911021e-01 1.62143183e+00 -8.56551945e-01 3.19958687e-01 1.30526386e-02 6.38860285e-01 1.18971717e+00 4.10727859e-01 6.26191646e-02 2.55595762e-02 -3.20232123e-01 7.80444860e-01 1.44662529e-01 -1.11454297e-02 -5.92791319e-01 -7.96630442e-01 6.73502505e-01 2.77346373e-01 6.78361356e-01 -2.44405389e-01 -3.12616706e-01 9.68994424e-02 5.12533903e-01 4.19329822e-01 1.32926658e-01 2.49539339e-03 -5.55237234e-02 -4.35764521e-01 -1.01075009e-01 9.96393979e-01 5.12976468e-01 8.71938407e-01 -3.21921498e-01 3.15880291e-02 8.35123658e-01 2.05353037e-01 4.89743352e-01 1.81128129e-01 -6.47201002e-01 -3.69256943e-01 8.25784683e-01 9.53421444e-02 -1.37992656e+00 -7.32224226e-01 5.31463400e-02 -9.76003170e-01 6.81698382e-01 5.44350803e-01 6.52955532e-01 -8.67978394e-01 1.00671363e+00 4.16058153e-01 -4.30618897e-02 -1.93720207e-01 8.70347440e-01 6.69868410e-01 3.34402472e-01 -4.76116091e-01 -1.27642989e-01 1.29726756e+00 -3.17054182e-01 -6.26091659e-01 7.08287001e-01 1.93935782e-01 -6.13747239e-01 8.63875926e-01 6.48768127e-01 -7.86150157e-01 -3.42870802e-01 -1.15370190e+00 3.35277379e-01 -5.90088069e-01 -2.07197350e-02 8.31081927e-01 7.76643634e-01 -7.15773523e-01 1.08035254e+00 -5.91955364e-01 -5.80863893e-01 2.66297549e-01 9.55396965e-02 -5.29288650e-01 -2.97262847e-01 -7.95771539e-01 7.90863216e-01 -1.22866452e-01 2.43774310e-01 -3.93937945e-01 1.70367837e-01 -6.61492229e-01 -3.17472130e-01 3.52313742e-02 -1.70466095e-01 7.83791184e-01 -3.35215360e-01 -1.68979812e+00 1.22302842e+00 -3.39545399e-01 -9.60069820e-02 3.97144318e-01 8.82740542e-02 -6.12331927e-01 4.87776756e-01 -2.55770057e-01 -3.41481626e-01 1.28582585e+00 -1.41361880e+00 -3.06423843e-01 -6.33534908e-01 -2.12276801e-01 -3.81597012e-01 -1.46846756e-01 -9.04856920e-02 -3.46940942e-02 -7.01554537e-01 9.47894156e-01 -8.06249619e-01 -4.20143992e-01 8.13550651e-02 -3.08902025e-01 -2.77753294e-01 8.58106434e-01 -1.75892398e-01 1.00253284e+00 -2.04202676e+00 -2.57405322e-02 8.44725609e-01 3.18195373e-01 -1.73556805e-02 -7.74131808e-03 5.55252433e-01 1.83121413e-01 3.01319778e-01 -4.45236623e-01 1.79953113e-01 1.70112140e-02 2.51721233e-01 -1.90906844e-03 9.05021071e-01 1.30416200e-01 4.58990455e-01 -9.35201168e-01 -4.16882277e-01 9.75633711e-02 3.12517583e-01 -1.23002537e-01 -6.21415265e-02 1.60186335e-01 1.69187695e-01 -1.05181038e+00 9.54930425e-01 9.94276941e-01 -2.06917971e-01 1.14652224e-01 -1.31572530e-01 -1.71654612e-01 -8.96704942e-02 -1.19543171e+00 1.19452500e+00 -1.14130065e-01 5.23289621e-01 -4.79369797e-02 -1.20446098e+00 1.35822463e+00 1.23411819e-01 7.39606678e-01 -5.73406935e-01 2.05028728e-01 4.02932256e-01 -2.77126342e-01 -5.72891414e-01 3.37758511e-01 -2.48393968e-01 -2.60256410e-01 6.76783681e-01 -3.49633306e-01 -2.64758736e-01 -2.33734492e-02 -2.50105590e-01 1.27339768e+00 -6.06346875e-02 5.32181144e-01 -8.52629006e-01 9.12842393e-01 1.69546947e-01 3.62230271e-01 6.73378766e-01 -5.61645254e-02 5.78876436e-01 5.10649621e-01 -8.83832455e-01 -8.92818809e-01 -1.07417524e+00 -7.83466637e-01 4.62927729e-01 4.67055678e-01 -6.76760525e-02 -7.34526396e-01 -5.24863839e-01 2.69008070e-01 -2.49400914e-01 -9.01554048e-01 4.51693535e-02 -4.81600553e-01 -5.98356605e-01 3.09866458e-01 -2.96073332e-02 5.83030403e-01 -7.60046482e-01 -3.87739450e-01 4.55271155e-02 2.79472232e-01 -8.22392941e-01 8.33138376e-02 -2.41340138e-02 -1.18899190e+00 -1.44352639e+00 -5.76471686e-01 -7.10738599e-01 8.98429871e-01 5.08782923e-01 9.57383931e-01 3.58195007e-01 -6.25349641e-01 5.72871804e-01 -7.76939213e-01 -1.79589048e-01 -4.44749773e-01 -3.86620075e-01 4.37538832e-01 4.54206407e-01 5.40689886e-01 -3.65586489e-01 -4.75526005e-01 7.04272866e-01 -8.67505193e-01 -5.09378314e-01 4.28106815e-01 8.70783746e-01 8.79557073e-01 6.02695882e-01 3.81819487e-01 -8.03687990e-01 5.21585584e-01 -2.00249255e-01 -4.41499531e-01 3.70453566e-01 -3.68185580e-01 3.58692735e-01 6.15155995e-01 -6.22429028e-02 -8.61849666e-01 6.90070465e-02 -1.16073741e-02 1.83031812e-01 -4.21152771e-01 3.02143306e-01 7.26737678e-02 -6.89116061e-01 6.07395291e-01 1.70261905e-01 1.67128384e-01 -1.04904461e+00 2.42770657e-01 8.27814937e-01 2.16069385e-01 -7.62667716e-01 7.89630175e-01 1.25489581e+00 6.78249717e-01 -1.24455106e+00 -8.53657126e-01 -5.90154529e-01 -1.12390363e+00 -1.65955454e-01 6.01527929e-01 8.72355551e-02 -8.01130652e-01 6.00096643e-01 -8.40573251e-01 -1.74959209e-02 -3.71341914e-01 5.96686006e-01 -8.95993888e-01 4.76195812e-01 -3.86301100e-01 -8.45214248e-01 1.05839565e-01 -8.35456431e-01 7.38332868e-01 -2.01539248e-02 1.04788810e-01 -1.33084047e+00 2.50972003e-01 -2.93551207e-01 1.18861005e-01 5.66319525e-01 1.07311666e+00 -4.03490394e-01 -1.88312873e-01 -5.53390503e-01 -1.44681349e-01 3.32212001e-01 3.38249475e-01 6.49667680e-02 -7.36657500e-01 -6.86316658e-03 5.15494168e-01 -4.86504473e-02 1.02231228e+00 2.42438585e-01 1.22404337e+00 4.98985238e-02 -4.14582402e-01 6.68197393e-01 1.65857208e+00 5.80330752e-02 6.68624759e-01 4.88316774e-01 2.14304119e-01 7.30921865e-01 9.45070684e-01 5.02915561e-01 1.04672387e-01 5.59014201e-01 1.42675385e-01 -1.22459307e-01 -7.43038347e-03 1.20873973e-01 -1.70178667e-01 7.41469979e-01 -4.00458246e-01 2.59635329e-01 -9.53396499e-01 1.08264647e-01 -1.68579972e+00 -1.03552711e+00 -5.49038053e-01 2.27116156e+00 3.69235486e-01 1.89285830e-01 1.20940998e-01 5.53742588e-01 8.56689453e-01 -4.52329870e-03 -2.95292586e-02 -5.70636451e-01 -3.94732714e-01 3.64975929e-01 4.04859960e-01 5.53638995e-01 -1.09944820e+00 5.66978514e-01 7.93320274e+00 7.05615163e-01 -9.81438518e-01 -3.26236308e-01 2.84320444e-01 7.98750818e-01 -2.83574402e-01 -1.12833664e-01 -6.33134127e-01 1.73772842e-01 3.76150578e-01 -3.12794477e-01 -1.35904744e-01 7.27313101e-01 1.25237077e-01 -3.69626671e-01 -1.01697254e+00 9.37180936e-01 9.14111454e-03 -1.32469904e+00 2.37335294e-01 2.59864777e-01 6.30681217e-01 -4.81935650e-01 1.54612720e-01 -3.27271014e-01 -1.51427627e-01 -1.07917750e+00 4.36486751e-01 1.07627213e+00 9.20286357e-01 -4.85541344e-01 6.19763255e-01 2.09929004e-01 -1.35020375e+00 -1.19753808e-01 -6.72985554e-01 -2.07338527e-01 6.25648424e-02 1.09503591e+00 -5.60898423e-01 6.51458800e-01 4.24989581e-01 8.42979848e-01 -4.85966414e-01 1.43351531e+00 1.04449295e-01 1.64886415e-01 -4.31486666e-01 -1.76158860e-01 4.63806182e-01 -5.28678358e-01 6.33376837e-01 1.23850846e+00 3.73130202e-01 7.36700535e-01 1.82093680e-02 2.40709901e-01 3.23690861e-01 1.22593932e-01 -1.17854786e+00 2.22116664e-01 1.24805234e-01 1.23243296e+00 -1.19010186e+00 -2.93594182e-01 -1.77402347e-01 7.10591257e-01 -1.17321752e-01 -5.25918156e-02 -8.54296163e-02 -6.03047311e-01 6.47488654e-01 2.55023897e-01 1.66388288e-01 -4.62120622e-01 -5.24405301e-01 -9.47667658e-01 2.61069741e-02 -4.32186246e-01 3.70985061e-01 -1.86172724e-01 -1.44608521e+00 6.14244103e-01 -2.75186658e-01 -1.66318488e+00 5.32554276e-02 -1.30369914e+00 -6.93825901e-01 5.46235323e-01 -1.50368714e+00 -5.95323563e-01 -2.82305241e-01 7.93622315e-01 1.04001209e-01 -4.79445606e-02 1.10428405e+00 -2.11573437e-01 -1.39291346e-01 1.70381010e-01 6.48314059e-01 9.36923325e-02 3.40844482e-01 -1.27336776e+00 1.63233474e-01 3.96810204e-01 3.37681100e-02 3.76448870e-01 5.40322363e-01 -4.47497755e-01 -1.68273056e+00 -5.48929095e-01 7.49812961e-01 -5.24526000e-01 5.56801200e-01 7.61968940e-02 -9.49156284e-01 4.09204923e-02 -7.37506390e-01 2.57482171e-01 6.38725340e-01 6.43067509e-02 -1.55418724e-01 -2.45052859e-01 -1.50919783e+00 3.80672365e-01 1.15988517e+00 -4.88648117e-01 -8.06207716e-01 3.91923964e-01 -5.19316010e-02 1.38418004e-01 -1.16330767e+00 5.80052257e-01 1.00066316e+00 -1.38975608e+00 9.43800330e-01 -6.24596477e-01 -6.83250427e-02 -5.40897287e-02 -3.48891497e-01 -1.07780111e+00 -7.72349179e-01 -6.78384840e-01 5.33379503e-02 3.42337489e-01 -1.70061570e-02 -7.95439959e-01 7.94533014e-01 2.97063496e-02 7.44541362e-02 -9.02536631e-01 -9.62777138e-01 -1.37904620e+00 -9.67162699e-02 -4.38876808e-01 6.44172192e-01 7.21693039e-01 3.68062139e-01 -3.44997734e-01 3.58032584e-02 -4.66252007e-02 1.22529507e+00 3.44032764e-01 4.83862489e-01 -1.94046092e+00 2.89258033e-01 -7.64444053e-01 -1.08257341e+00 -9.45886493e-01 1.21138476e-01 -6.98254943e-01 -4.93175909e-03 -1.41019034e+00 5.08401096e-02 -1.23877144e+00 -4.45609093e-01 -4.57490124e-02 3.93585026e-01 5.57327628e-01 -2.95539439e-01 3.62821430e-01 -5.23397207e-01 2.29532570e-01 1.39188635e+00 7.62977377e-02 -2.26456709e-02 3.76735419e-01 -3.28148365e-01 1.14169860e+00 8.67975533e-01 -4.65693444e-01 1.79376483e-01 9.00398642e-02 -6.73722401e-02 -2.69666556e-02 2.23632902e-01 -8.37106109e-01 -1.40500907e-03 -4.32152152e-01 3.56322885e-01 -4.06518430e-01 3.91543627e-01 -7.83022940e-01 -1.60610780e-01 5.80630779e-01 1.52881533e-01 -3.44608247e-01 -3.04129273e-01 5.92829823e-01 -3.90281379e-01 -3.94956708e-01 1.08347058e+00 -2.47801363e-01 -1.09785581e+00 2.97452182e-01 -3.10939997e-01 -4.19777870e-01 9.69119787e-01 -7.36490130e-01 -3.49474162e-01 -1.16996422e-01 -9.74125981e-01 -3.06982756e-01 7.95809805e-01 9.77636203e-02 1.00572252e+00 -1.70798111e+00 -5.12877643e-01 4.35483009e-01 1.37042046e-01 -5.06666958e-01 -4.84886877e-02 9.81689990e-01 -7.60384917e-01 4.67786223e-01 -5.63863397e-01 -8.14589500e-01 -1.19100094e+00 4.79012132e-01 3.15513223e-01 -5.87609969e-02 -9.85053420e-01 5.89724362e-01 1.25991013e-02 -1.01039179e-01 -1.55524045e-01 -1.56901509e-01 -3.84224296e-01 -1.18619937e-03 7.99261808e-01 8.54693294e-01 1.39757171e-01 -6.96360230e-01 -5.57301641e-01 1.36451304e+00 3.41743946e-01 -1.56516105e-01 1.41967118e+00 -1.33640721e-01 -6.41152978e-01 7.58398235e-01 1.39582872e+00 -2.72795837e-02 -8.35227251e-01 -3.92513245e-01 4.54407603e-01 -8.78323138e-01 -7.96247944e-02 -2.23947808e-01 -3.53178620e-01 7.88549781e-01 4.28555787e-01 8.85268450e-01 1.06996775e+00 7.34363943e-02 5.16831994e-01 7.57904410e-01 9.49484169e-01 -1.18837810e+00 1.82130337e-01 7.08029926e-01 1.07979727e+00 -1.13880527e+00 2.18113303e-01 -5.54493070e-01 -2.47644827e-01 1.61361587e+00 -2.23028421e-01 -4.52898860e-01 1.39359927e+00 1.12899587e-01 7.29449419e-03 -4.24798429e-01 -3.59461337e-01 -5.25148273e-01 4.58921194e-01 7.01400816e-01 1.67351350e-01 4.02051657e-01 -4.47749436e-01 2.28911191e-01 -1.70672424e-02 -4.93779555e-02 5.69467545e-01 1.00575542e+00 -1.05500174e+00 -1.15503621e+00 -6.76667571e-01 5.14841616e-01 -3.55123013e-01 4.50075328e-01 -2.60032743e-01 5.62282205e-01 -2.08440676e-01 6.71589255e-01 -7.35108033e-02 -5.79583347e-01 3.23687702e-01 7.13910023e-03 9.20607805e-01 -5.14972329e-01 2.20049948e-01 -3.52365285e-01 -3.88722084e-02 -7.67196119e-01 -4.56406057e-01 -6.67746365e-01 -1.21706486e+00 -5.01544237e-01 -1.17818755e-03 2.86104500e-01 9.75428700e-01 8.17255080e-01 -8.21093470e-03 -1.86411262e-01 1.38166094e+00 -9.38004971e-01 -6.37349606e-01 -5.86673498e-01 -1.41572762e+00 3.52321148e-01 5.57578087e-01 -1.09175920e+00 -4.35178190e-01 -6.61512092e-02]
[10.063292503356934, -0.483588308095932]
136e2938-be35-4263-8b78-112ab9f7e512
unveiling-the-link-between-logical-fallacies
1304.3940
null
http://arxiv.org/abs/1304.3940v2
http://arxiv.org/pdf/1304.3940v2.pdf
Unveiling the link between logical fallacies and web persuasion
In the last decade Human-Computer Interaction (HCI) has started to focus attention on forms of persuasive interaction where computer technologies have the goal of changing users behavior and attitudes according to a predefined direction. In this work, we hypothesize a strong connection between logical fallacies (forms of reasoning which are logically invalid but cognitively effective) and some common persuasion strategies adopted within web technologies. With the aim of empirically evaluating our hypothesis, we carried out a pilot study on a sample of 150 e-commerce websites.
['Fabiana Vernero', 'Antonio Lieto']
2013-04-14
null
null
null
null
['persuasion-strategies', 'logical-fallacies']
['computer-vision', 'miscellaneous']
[ 2.60222256e-01 9.27306116e-01 -1.86397019e-03 -3.61400098e-01 2.33683735e-01 -5.56124270e-01 6.70466542e-01 5.33321142e-01 -5.41735768e-01 6.20406508e-01 3.82510245e-01 -1.18065310e+00 -3.29490811e-01 -8.25828671e-01 -3.80154103e-01 1.43193528e-01 3.25179428e-01 -3.64174992e-02 4.04181540e-01 -3.39928597e-01 1.02030492e+00 1.59588084e-01 -1.24966240e+00 2.85402238e-01 1.01467586e+00 1.79403409e-01 -1.48065746e-01 4.25820291e-01 -7.29896948e-02 7.29401827e-01 -3.74683678e-01 -6.33003294e-01 -3.17215413e-01 -6.76162839e-01 -1.04737544e+00 -3.08095906e-02 2.93300569e-01 -2.87609220e-01 2.38247290e-01 1.26450813e+00 3.67153049e-01 4.33097742e-02 1.98217466e-01 -1.00105941e+00 -8.31837356e-01 7.28225708e-01 1.20245330e-01 1.83724850e-01 9.91958320e-01 2.36490101e-01 5.40429831e-01 -1.99581653e-01 5.09689987e-01 1.45741248e+00 7.14664221e-01 6.32338881e-01 -1.38048470e+00 -3.84644240e-01 2.88666785e-01 3.53263855e-01 -1.06878352e+00 -2.67696679e-01 9.38171983e-01 -3.71310562e-01 9.60564375e-01 5.88757932e-01 6.79269671e-01 1.19039607e+00 2.47907668e-01 2.93720543e-01 1.66870105e+00 -1.05456400e+00 5.21575391e-01 8.79786730e-01 8.95778954e-01 3.71248454e-01 1.09125149e+00 -1.00615345e-01 -1.56651154e-01 -4.39140052e-01 3.41201931e-01 -5.10633767e-01 -3.24038833e-01 -1.63814545e-01 -6.23456359e-01 8.61447573e-01 3.24158370e-02 8.22883010e-01 -6.30376518e-01 -3.37274492e-01 1.74462289e-01 1.73316702e-01 -1.55089408e-01 6.89048648e-01 -4.56751823e-01 -5.22752106e-01 2.15057537e-01 8.43706653e-02 1.37488067e+00 3.90463471e-01 2.60171294e-01 -8.01084191e-02 -6.60207942e-02 2.25389704e-01 8.21715713e-01 3.98848712e-01 3.11406255e-01 -6.41386390e-01 -1.39725149e-01 7.56710589e-01 5.05612135e-01 -1.15530336e+00 -8.03506896e-02 8.38696435e-02 2.27585062e-02 4.44024920e-01 5.20251691e-01 -4.72926199e-01 -1.03937589e-01 1.47467422e+00 3.93456161e-01 -4.95784581e-01 -1.35267556e-01 7.70303309e-01 2.95239478e-01 3.27572227e-02 8.43358874e-01 -4.15912777e-01 1.35448527e+00 -8.69461335e-03 -1.02056360e+00 -8.32825899e-03 4.21345055e-01 -5.69185853e-01 1.77732348e+00 5.60256898e-01 -1.03687465e+00 -3.57564628e-01 -1.16174841e+00 2.00767919e-01 -4.97551858e-01 -6.47914886e-01 6.53256178e-01 1.52640891e+00 -9.07562137e-01 2.29316637e-01 -4.74183977e-01 -9.19625819e-01 6.97494000e-02 -1.92086399e-01 2.97631949e-01 3.07785630e-01 -1.36447966e+00 1.04581368e+00 3.89652878e-01 1.11685798e-01 4.10683379e-02 -2.93423206e-01 -5.72727501e-01 8.84799287e-02 5.45084059e-01 -6.16771698e-01 1.34251201e+00 -1.30877316e+00 -1.69081092e+00 7.06776381e-01 3.33114028e-01 -3.58898163e-01 4.54758167e-01 -3.49531531e-01 -6.93763614e-01 -1.09169357e-01 -1.46153212e-01 6.72034100e-02 2.78147757e-01 -1.34434450e+00 -7.31284499e-01 -7.68302798e-01 3.96979749e-01 1.06388569e-01 -5.55128157e-01 1.51557937e-01 3.43943864e-01 -1.07552566e-01 -2.22081989e-01 -6.02926373e-01 -1.74370989e-01 -4.76698875e-01 2.01228447e-02 -5.60652792e-01 6.60956323e-01 -5.39721906e-01 1.60218263e+00 -1.79569745e+00 -6.40189707e-01 3.72094363e-01 1.22438900e-01 6.50692344e-01 3.92682970e-01 5.96285224e-01 2.01828405e-01 6.95703387e-01 5.66005290e-01 9.74732161e-01 3.19233954e-01 -7.44370222e-02 1.39998779e-01 4.79562674e-03 -2.32060939e-01 6.11235380e-01 -8.67345691e-01 -5.73949277e-01 2.58620471e-01 5.63475490e-01 -4.54917401e-01 -1.21081538e-01 -1.81979865e-01 -1.91034943e-01 -8.69569361e-01 7.96423480e-02 4.90081340e-01 -1.76876128e-01 1.02666354e+00 1.56232297e-01 -6.52645707e-01 8.15038085e-01 -8.08729291e-01 8.10285985e-01 -1.81096777e-01 4.10527289e-01 -2.37994105e-01 -3.77117127e-01 5.84015906e-01 2.93611497e-01 -1.05997019e-01 -1.08210731e+00 4.46261376e-01 -1.95376948e-01 4.56919014e-01 -6.31617546e-01 4.30316180e-02 1.07556835e-01 2.89501160e-01 5.35261631e-01 -7.60141373e-01 1.81591585e-01 1.02986820e-01 1.70299783e-01 1.09456813e+00 2.47318238e-01 8.03145945e-01 -5.34606874e-01 6.12935185e-01 2.00434119e-01 8.55688006e-02 1.02096999e+00 -3.52476269e-01 -5.75767636e-01 3.08525026e-01 -4.59591359e-01 -5.48933625e-01 -9.01085973e-01 9.24610570e-02 6.42084539e-01 2.07634419e-01 -5.23133159e-01 -1.15971351e+00 -9.39875543e-01 -3.37947726e-01 1.85230565e+00 -4.44452822e-01 -3.65175307e-01 -1.51043370e-01 -1.07233971e-01 2.09605724e-01 -8.01259130e-02 1.00162435e+00 -1.27217674e+00 -1.48132896e+00 3.11006576e-01 -1.44751340e-01 -6.39223218e-01 1.03308842e-01 -2.96731353e-01 -1.16072166e+00 -1.21845293e+00 3.71748328e-01 -3.75954926e-01 6.38159156e-01 5.76312959e-01 9.15099978e-01 4.91011173e-01 2.36173704e-01 6.85407043e-01 -6.28947258e-01 -8.18271756e-01 -4.27181989e-01 -3.41820538e-01 -3.70690405e-01 -4.13901210e-01 9.86303926e-01 -2.68361896e-01 -6.06689692e-01 3.30209106e-01 -9.08811629e-01 9.30566415e-02 4.71130341e-01 -1.69063583e-02 -2.80049354e-01 2.34109133e-01 2.72862554e-01 -9.79712844e-01 1.37375891e+00 -3.44340175e-01 -4.10371304e-01 3.18316132e-01 -1.35246909e+00 -2.81372011e-01 2.63630748e-01 -5.98638833e-01 -1.62516582e+00 -5.46782970e-01 -3.84451114e-02 8.62932622e-01 -5.76398551e-01 7.14996219e-01 -4.09942836e-01 -3.41597758e-02 1.03623605e+00 -1.11464143e-01 2.28539065e-01 -2.08224267e-01 1.60583239e-02 7.68570781e-01 -2.60619987e-02 -5.13501167e-01 4.71157700e-01 5.76521084e-02 -4.80633408e-01 -9.38898385e-01 -6.51321292e-01 2.78115541e-01 -9.55729485e-02 -7.31792331e-01 5.67338824e-01 -1.31415546e-01 -1.40958405e+00 -1.50185511e-01 -7.06003785e-01 -5.39942682e-01 1.82084084e-01 5.83733201e-01 -2.06658617e-01 5.59786439e-01 -2.33637601e-01 -1.36128104e+00 -3.42728674e-01 -2.81280339e-01 9.79535207e-02 4.85819370e-01 -1.15881371e+00 -1.08580911e+00 3.48005295e-02 4.31335956e-01 5.34113765e-01 4.73084114e-03 1.18472052e+00 -6.93146348e-01 7.03730993e-03 -3.27496737e-01 1.36397347e-01 3.79964709e-04 2.70804942e-01 -7.99538046e-02 -4.34447497e-01 2.83295900e-01 3.72065693e-01 9.63829830e-02 -1.97435826e-01 2.30745018e-01 5.87643683e-01 -7.63897538e-01 -1.79680645e-01 -6.74490273e-01 1.39828026e+00 7.61896551e-01 1.16412938e+00 7.26488948e-01 -3.86634208e-02 7.15378404e-01 1.00342929e+00 2.06420869e-01 2.89273202e-01 5.15063226e-01 -4.06694636e-02 3.68431896e-01 4.28850427e-02 -5.40393829e-01 3.02342057e-01 -1.52944356e-01 -1.81742698e-01 -1.03902817e-01 -1.02776706e+00 2.37474758e-02 -1.78961921e+00 -1.02194297e+00 -6.30507231e-01 2.33972049e+00 7.90712237e-01 6.99073732e-01 5.20814257e-03 3.52767825e-01 7.68991113e-01 -3.61419469e-01 -3.78429256e-02 -5.96349955e-01 4.90466148e-01 -7.71942213e-02 6.49956167e-02 9.21383023e-01 -5.26809156e-01 7.26856291e-01 6.76246643e+00 -1.78683102e-01 -9.59943652e-01 -6.75964504e-02 5.43964148e-01 4.55415845e-01 -6.83880270e-01 7.97485933e-02 -3.97316337e-01 4.23386842e-01 1.02843308e+00 -2.43678123e-01 2.03147396e-01 6.70829058e-01 6.98980093e-01 -5.33339143e-01 -6.32426023e-01 4.73903120e-01 -1.07172407e-01 -7.75738358e-01 -5.74441031e-02 2.42697507e-01 1.46055385e-01 -1.00553644e+00 -9.45750177e-02 3.23527008e-01 2.89760083e-01 -4.62843686e-01 7.61591077e-01 5.11342406e-01 -3.07459906e-02 -4.73201752e-01 6.59427583e-01 3.50021899e-01 1.59484789e-01 1.70614198e-02 1.82887524e-01 -8.89470160e-01 1.28716096e-01 1.62370622e-01 -1.02546763e+00 -7.44549632e-02 8.12199473e-01 -2.91492581e-01 -4.79043990e-01 6.83161974e-01 -4.51978356e-01 9.63265657e-01 -6.19352795e-02 -8.45592916e-01 1.19247185e-02 -5.37614882e-01 2.86812574e-01 8.80546451e-01 -2.30223730e-01 5.31840980e-01 -4.41560209e-01 7.05640912e-01 7.66155660e-01 1.69169500e-01 -6.00010216e-01 -2.70786077e-01 4.95974749e-01 1.03228641e+00 -9.76705194e-01 -4.39483166e-01 -5.13340175e-01 6.86378598e-01 -4.61234860e-02 3.75040829e-01 -7.58451521e-01 -2.43797153e-01 5.09960413e-01 5.15232027e-01 1.04618154e-01 -1.41388252e-01 -8.81602526e-01 -6.90393686e-01 -7.10016340e-02 -1.14990807e+00 2.63862550e-01 -6.36564255e-01 -8.40337455e-01 3.89057398e-02 4.78452118e-03 -1.95763990e-01 6.31412640e-02 -5.27690172e-01 -5.38077772e-01 6.65492415e-01 -7.81911731e-01 -6.07106328e-01 -4.72353131e-01 1.28322333e-01 2.97213107e-01 5.35035372e-01 9.67764556e-01 -1.59770072e-01 -3.07501465e-01 9.94280130e-02 -6.11565650e-01 -4.53125149e-01 4.63925570e-01 -9.81485784e-01 1.12759300e-01 7.73002744e-01 -4.95862812e-01 1.20117772e+00 1.41271901e+00 -1.15156269e+00 -1.62063909e+00 -1.84129462e-01 1.31287503e+00 -3.79647195e-01 5.48217654e-01 -7.56536983e-03 -8.06799352e-01 6.53283775e-01 4.94442940e-01 -1.13677502e+00 9.46710050e-01 3.10948253e-01 -1.90856352e-01 2.17922866e-01 -1.52125001e+00 1.28881264e+00 1.19889009e+00 -2.18626037e-01 -9.70935702e-01 -4.77318615e-02 1.28859460e-01 3.29402179e-01 -5.33163428e-01 2.10879400e-01 7.54895806e-01 -1.10427630e+00 8.12462151e-01 -7.21418381e-01 8.68819747e-03 -5.11527695e-02 2.98317447e-02 -9.80664909e-01 -5.71781039e-01 -8.17305028e-01 5.08345179e-02 1.15916276e+00 2.71936208e-01 -1.01805687e+00 5.97575426e-01 1.36747849e+00 2.50725567e-01 -6.33280948e-02 -2.35592529e-01 -9.80552509e-02 -3.32262099e-01 -2.83823729e-01 4.86111194e-01 1.03099144e+00 8.18147659e-01 6.34694636e-01 1.60927504e-01 9.65200961e-02 4.86100703e-01 -4.82965022e-01 6.42071962e-01 -1.57517040e+00 -1.33900493e-01 -5.88627458e-01 -3.36081423e-02 -6.17271006e-01 -2.38593474e-01 -1.00416817e-01 -1.10711180e-01 -1.53103817e+00 2.35280972e-02 3.12596768e-01 -9.57601611e-03 3.27610821e-01 -2.36829862e-01 -4.41046506e-01 -2.00315028e-01 -3.72797817e-01 -5.87713718e-01 1.77270666e-01 8.45240653e-01 3.71128350e-01 -4.83167440e-01 -2.82765198e-02 -1.68146396e+00 8.27950418e-01 1.20092118e+00 -1.35881841e-01 -7.39862859e-01 6.48657829e-02 6.36677146e-01 -9.91930440e-02 3.02828521e-01 -5.29337585e-01 -1.82573050e-01 -6.47157252e-01 2.66680688e-01 4.35133539e-02 -4.03331429e-01 -1.28588176e+00 4.16911721e-01 9.33778048e-01 -4.61420357e-01 -1.22743353e-01 2.45697632e-01 3.10479313e-01 3.60108227e-01 -5.54932058e-01 3.24422181e-01 1.97453126e-01 -5.79110563e-01 -8.18737447e-01 -1.23054707e+00 -4.06187564e-01 1.26719582e+00 -5.77431321e-01 -2.79011369e-01 -4.51804966e-01 -3.65467429e-01 -7.50202835e-02 3.96722466e-01 5.61695039e-01 5.23925185e-01 -8.96270454e-01 -1.21985912e-01 -3.32382530e-01 -1.75027121e-02 -1.10508358e+00 5.52657843e-02 7.48310804e-01 -6.78034842e-01 8.48309219e-01 -3.97819638e-01 1.87003955e-01 -1.62059045e+00 6.36900067e-01 1.93943173e-01 3.68654072e-01 -5.01235545e-01 -6.90922886e-02 -3.42794061e-01 -2.58815289e-02 4.98950854e-02 1.34798989e-01 -5.81559062e-01 -3.01901162e-01 7.92748094e-01 7.52972960e-01 -1.48125902e-01 5.08170947e-02 -2.86077857e-01 -3.34421992e-01 -9.02296156e-02 -4.50131595e-01 8.62849891e-01 -5.12765050e-01 -1.26579732e-01 3.57658178e-01 2.46511161e-01 2.41623893e-01 -6.70720518e-01 2.39008456e-01 6.10189378e-01 -9.32282686e-01 4.62640114e-02 -1.48287225e+00 5.54116368e-02 1.47949815e-01 6.50209367e-01 9.26128089e-01 8.02443147e-01 -1.73772961e-01 -2.89399251e-02 6.01793945e-01 5.47358632e-01 -1.41209424e+00 -5.98503798e-02 1.40061140e-01 8.66216898e-01 -7.88604140e-01 -2.05006972e-01 -8.54844213e-01 -6.38086021e-01 1.05015945e+00 8.15395355e-01 3.51354659e-01 7.60433078e-01 -1.02673680e-01 1.51922807e-01 -4.31080967e-01 -6.24458611e-01 -9.24549401e-02 -1.48960620e-01 9.63157237e-01 1.17458963e+00 1.14812888e-01 -1.80754018e+00 5.53058267e-01 -9.65024307e-02 5.80232561e-01 6.78417325e-01 1.28127480e+00 -7.61468053e-01 -1.10271466e+00 -6.52298570e-01 1.65743247e-01 -4.97296929e-01 1.42338440e-01 -1.10718918e+00 1.07647312e+00 -3.79356593e-01 1.71232164e+00 -1.54081851e-01 -1.77854940e-01 6.52769029e-01 3.23921651e-01 4.34299499e-01 -2.43401706e-01 -7.83492804e-01 2.62223706e-02 7.97806442e-01 -4.73267078e-01 -2.12170348e-01 -9.84192908e-01 -1.15533721e+00 -5.43468535e-01 -4.50040074e-03 4.65554088e-01 7.32784808e-01 7.76234269e-01 3.80803645e-01 3.84722412e-01 1.19676605e-01 8.30177590e-02 -7.09523857e-01 -1.26968265e+00 -1.66124254e-01 6.58760667e-01 -5.93913615e-01 -2.48761594e-01 -1.59637049e-01 -2.22714126e-01]
[9.101140975952148, 6.308643341064453]
5c4b2f8f-e0ab-4edc-b2c3-115bb697742b
ccdn-checkerboard-corner-detection-network
2302.05097
null
https://arxiv.org/abs/2302.05097v1
https://arxiv.org/pdf/2302.05097v1.pdf
CCDN: Checkerboard Corner Detection Network for Robust Camera Calibration
Aiming to improve the checkerboard corner detection robustness against the images with poor quality, such as lens distortion, extreme poses, and noise, we propose a novel detection algorithm which can maintain high accuracy on inputs under multiply scenarios without any prior knowledge of the checkerboard pattern. This whole algorithm includes a checkerboard corner detection network and some post-processing techniques. The network model is a fully convolutional network with improvements of loss function and learning rate, which can deal with the images of arbitrary size and produce correspondingly-sized output with a corner score on each pixel by efficient inference and learning. Besides, in order to remove the false positives, we employ three post-processing techniques including threshold related to maximum response, non-maximum suppression, and clustering. Evaluations on two different datasets show its superior robustness, accuracy and wide applicability in quantitative comparisons with the state-of-the-art methods, like MATE, ChESS, ROCHADE and OCamCalib.
['Qi Zhang', 'Caihua Xiong', 'Ben Chen']
2023-02-10
null
null
null
null
['camera-calibration']
['computer-vision']
[ 1.94359139e-01 -5.97936690e-01 2.28015244e-01 -1.21219307e-01 -2.09542349e-01 -4.43457842e-01 2.97448903e-01 -5.74915819e-02 -4.94507998e-01 4.77170706e-01 -2.87894636e-01 -3.35391372e-01 -1.17089689e-01 -6.94070816e-01 -7.74719715e-01 -7.27317095e-01 1.68230549e-01 -2.13463590e-01 8.27713251e-01 7.75138661e-03 8.80253494e-01 5.41891575e-01 -1.64551508e+00 4.14240688e-01 8.00459683e-01 1.39714479e+00 -1.22850863e-02 7.17670262e-01 4.91529375e-01 7.23751962e-01 -6.28871024e-01 -2.09412396e-01 5.45599341e-01 -9.69365016e-02 2.63904274e-01 -1.74547825e-02 6.50853813e-01 -2.00290874e-01 -4.51024055e-01 1.30912149e+00 8.08107436e-01 -2.96172928e-02 7.88881123e-01 -1.13218927e+00 -5.69109797e-01 1.52004048e-01 -9.53953326e-01 1.98158577e-01 3.81335407e-01 3.95993322e-01 3.95283937e-01 -8.89211178e-01 3.82118672e-01 1.09368515e+00 1.04125464e+00 8.47313181e-02 -7.10328758e-01 -1.04733825e+00 -1.42291144e-01 4.13103551e-01 -1.40535831e+00 -1.35394260e-01 5.79589903e-01 -3.12870592e-01 5.18320978e-01 4.18371469e-01 4.58959669e-01 9.59061921e-01 5.17602444e-01 6.98011816e-01 1.18970764e+00 -5.37855685e-01 5.16729243e-02 -1.13810860e-01 4.27643955e-02 7.36002743e-01 5.90464294e-01 3.33568543e-01 -3.62224907e-01 1.31829217e-01 9.25510943e-01 3.48737799e-02 -4.20524359e-01 -4.75250989e-01 -1.13304210e+00 2.68510282e-01 6.99805796e-01 1.17256679e-01 5.83969317e-02 4.08363529e-02 3.05328876e-01 8.70697573e-02 -1.95355996e-01 4.51336384e-01 -3.33447129e-01 3.19827199e-01 -1.05058670e+00 2.05341235e-01 4.82476234e-01 1.05266213e+00 4.33587790e-01 -1.20517097e-01 -4.53679174e-01 3.60367566e-01 3.04149479e-01 5.75345576e-01 4.50255871e-01 -7.09262371e-01 4.08047646e-01 8.03009987e-01 2.76177287e-01 -1.58796501e+00 -7.03701019e-01 -7.60974884e-01 -1.21586359e+00 7.43042648e-01 3.10041428e-01 -1.38372138e-01 -1.14091885e+00 1.01959193e+00 3.12330216e-01 2.63229460e-01 -1.55819178e-01 1.10878813e+00 8.41022670e-01 3.99970293e-01 -5.20803809e-01 -1.34356782e-01 1.44170034e+00 -9.25507545e-01 -6.16484344e-01 -1.94587812e-01 2.64867749e-02 -1.11200798e+00 8.74357522e-01 8.27242374e-01 -7.09142506e-01 -8.91295671e-01 -1.46730673e+00 -4.01409604e-02 -5.48364222e-01 7.39781797e-01 5.09557009e-01 6.34940326e-01 -7.03748226e-01 2.64290839e-01 -4.09590930e-01 -5.56460619e-02 3.69844407e-01 3.55124146e-01 -2.55000979e-01 9.77313295e-02 -8.54850113e-01 6.91253662e-01 4.12459940e-01 3.74349743e-01 -5.57888031e-01 -4.38706875e-01 -7.48434961e-01 2.21782923e-01 4.32742089e-01 -6.05549693e-01 7.98697948e-01 -1.09383428e+00 -1.40677464e+00 6.07941866e-01 1.63304627e-01 -4.67992693e-01 9.78550494e-01 -1.79806426e-01 -4.53182131e-01 1.01815581e-01 -3.43277976e-02 5.03087044e-01 1.22668350e+00 -1.03824437e+00 -9.00163651e-01 -3.20582747e-01 -2.17954591e-01 1.71465263e-01 -1.21898867e-01 -1.18665844e-01 -6.91138744e-01 -5.11342585e-01 5.26145101e-01 -7.16078281e-01 -6.77706972e-02 2.36879274e-01 -5.29242814e-01 1.73721611e-02 9.66993511e-01 -5.75280666e-01 1.25536430e+00 -2.36370587e+00 -4.96939540e-01 4.69567209e-01 7.65699670e-02 5.23860276e-01 1.47823945e-01 1.73863601e-02 1.27715096e-01 -2.11712435e-01 -3.63919437e-02 1.48973361e-01 -2.07552776e-01 -4.68738765e-01 -9.05197039e-02 7.76907921e-01 1.49150729e-01 4.66889590e-01 -6.50407314e-01 -4.69242483e-01 4.55425024e-01 2.36294746e-01 -4.95787472e-01 -1.11047901e-01 -3.32193859e-02 9.57852229e-02 -1.44183591e-01 8.68752658e-01 1.09594023e+00 -1.49600610e-01 -1.84124693e-01 -4.16744202e-01 -4.02684242e-01 -5.38844168e-01 -1.77432692e+00 1.09367502e+00 -1.43549666e-01 8.29442203e-01 -5.65858325e-03 -4.30932730e-01 1.09258163e+00 -2.37415120e-01 -1.40476227e-01 -9.03358936e-01 5.02528667e-01 2.94379503e-01 1.49800420e-01 -6.36304617e-01 4.11167294e-01 5.44311523e-01 9.23864618e-02 -1.82566717e-01 -2.78423667e-01 1.40681162e-01 2.08700463e-01 -1.28295943e-01 9.10042703e-01 1.02673590e-01 1.37190104e-01 -1.44998655e-01 8.14247608e-01 -2.15977356e-01 6.30396068e-01 8.77920449e-01 -1.91223308e-01 9.02805865e-01 4.42872882e-01 -5.66894948e-01 -9.33888257e-01 -8.68276358e-01 -4.61544186e-01 5.51979303e-01 7.61010408e-01 6.68359846e-02 -6.27391934e-01 -2.93834597e-01 2.07469627e-01 2.05359459e-01 -6.50385320e-01 -1.75435349e-01 -4.39850926e-01 -8.61209571e-01 5.18640101e-01 6.83594525e-01 1.23524034e+00 -7.94249475e-01 -1.01991248e+00 -2.27924243e-01 1.54946610e-01 -1.09456706e+00 -3.98404688e-01 1.15626432e-01 -5.37081659e-01 -1.59862995e+00 -5.96959710e-01 -1.10648966e+00 8.89182508e-01 2.38574371e-01 6.70517147e-01 2.17539743e-01 -6.19166732e-01 -2.44079784e-01 -1.10962771e-01 -3.03371459e-01 1.15850829e-01 -2.42090464e-01 -1.15818791e-01 1.07678317e-01 3.36875439e-01 -1.38126805e-01 -9.98884857e-01 4.23546642e-01 -7.02032268e-01 3.37659456e-02 1.01543427e+00 9.44065273e-01 4.25214499e-01 2.65839100e-01 8.70204624e-03 -5.06779850e-01 5.48599601e-01 1.09624542e-01 -1.34721267e+00 1.21510081e-01 -6.31987453e-01 -2.61283547e-01 6.31156445e-01 -3.49519759e-01 -1.02075744e+00 2.97683835e-01 2.21521452e-01 -3.74633223e-01 -2.14872956e-01 9.95433237e-03 1.64683554e-02 -5.72049201e-01 9.07521784e-01 2.66802818e-01 -1.43676937e-01 -2.38431647e-01 -8.75380542e-03 9.23894584e-01 9.00660932e-01 9.70353410e-02 9.24303710e-01 5.85891426e-01 4.12106544e-01 -6.89762831e-01 -6.07253611e-01 -5.04087090e-01 -5.80606997e-01 -2.71885842e-01 6.31214917e-01 -9.15227592e-01 -1.22613871e+00 1.00276423e+00 -1.02910531e+00 1.54406920e-01 4.63257700e-01 4.45208460e-01 -2.21209511e-01 5.96240163e-01 -4.33159679e-01 -8.18425596e-01 -4.73965526e-01 -9.49359834e-01 8.61731291e-01 8.23429346e-01 3.34797353e-01 -4.78261560e-01 -2.83151597e-01 1.65506110e-01 2.29607940e-01 4.42761987e-01 5.94626904e-01 -3.30705911e-01 -9.75549638e-01 -5.49832046e-01 -5.52633166e-01 4.29492563e-01 -2.21891478e-01 4.20740277e-01 -8.18540514e-01 -3.22239667e-01 -2.10163355e-01 -2.05842312e-02 9.50996816e-01 4.96687979e-01 1.25145137e+00 -5.18692099e-02 -4.82111782e-01 1.05411446e+00 1.80197942e+00 3.07651877e-01 7.69409359e-01 6.95969343e-01 3.03363562e-01 -5.26949465e-02 7.15726554e-01 3.21886092e-01 -2.68158559e-02 2.45704651e-01 6.67326510e-01 -2.45722786e-01 -2.29736269e-01 -1.19406037e-01 4.00805809e-02 2.45526358e-01 1.76958919e-01 -1.89482510e-01 -7.81291723e-01 3.85025114e-01 -1.79655743e+00 -9.87258852e-01 -5.27372658e-01 2.23726296e+00 3.71724844e-01 6.56564176e-01 -2.37182632e-01 3.90869349e-01 9.73751545e-01 2.85246018e-02 -6.90325916e-01 -3.90601121e-02 -4.92656946e-01 -1.39451936e-01 9.03175890e-01 2.04887286e-01 -1.58961356e+00 7.47216880e-01 6.78599548e+00 7.08720446e-01 -1.30109990e+00 -4.10068989e-01 7.18181908e-01 3.08407158e-01 3.43635708e-01 -1.61654100e-01 -9.05476332e-01 6.54851377e-01 3.27300988e-02 4.41905588e-01 3.54189724e-01 8.43560398e-01 5.61496653e-02 -5.00138819e-01 -6.71992779e-01 1.12947285e+00 3.89095277e-01 -1.33061802e+00 -1.88980386e-01 -4.94652599e-01 1.00906098e+00 -1.09425284e-01 3.34301084e-01 1.43017232e-01 -1.00045957e-01 -9.29685891e-01 5.81349909e-01 7.87721336e-01 7.07023680e-01 -6.43510640e-01 1.12284863e+00 4.27096844e-01 -9.85446513e-01 -5.86761057e-01 -5.32649457e-01 -2.25072458e-01 -5.41130722e-01 5.95765054e-01 -4.86125469e-01 5.31850994e-01 8.17934334e-01 4.27229196e-01 -9.19244587e-01 1.76295173e+00 -4.61545020e-01 1.44487321e-02 -4.60071802e-01 -3.69115263e-01 2.50621766e-01 -1.76827714e-01 2.87940949e-01 1.23601604e+00 3.41965437e-01 -2.17088982e-01 9.53832362e-03 6.12250566e-01 1.32464185e-01 4.18767892e-02 -3.11252683e-01 7.15963185e-01 5.64663470e-01 1.34519851e+00 -8.33484948e-01 -2.40172938e-01 -4.12122965e-01 7.79633105e-01 -6.07709587e-02 3.66366059e-01 -8.94923210e-01 -1.00255167e+00 1.76990747e-01 2.45505765e-01 4.11495775e-01 -7.27698579e-02 -7.17751324e-01 -8.71486187e-01 2.77746737e-01 -9.46591198e-01 3.31756651e-01 -1.00482893e+00 -9.80463445e-01 3.31775010e-01 -2.84984499e-01 -1.48857546e+00 2.17139423e-01 -1.18884456e+00 -9.84239280e-01 5.00959873e-01 -1.64342272e+00 -9.49400008e-01 -8.30640614e-01 5.87742388e-01 2.45115682e-01 -3.99290770e-01 1.44092701e-02 4.08334345e-01 -5.35241306e-01 8.29248607e-01 3.04594487e-01 4.69623327e-01 8.48824084e-01 -1.03009379e+00 9.52749848e-02 1.25856173e+00 -2.95454323e-01 4.66131538e-01 6.81742609e-01 -5.24696171e-01 -1.37396109e+00 -9.90368187e-01 3.92461300e-01 -1.86235175e-01 2.72985905e-01 -2.60009140e-01 -4.45148319e-01 3.40073258e-01 4.77886312e-02 3.19266468e-01 -1.55907767e-02 -2.57921726e-01 -1.70244560e-01 -5.00980079e-01 -1.14053762e+00 7.71600425e-01 9.63276207e-01 -2.80496497e-02 -4.93973255e-01 1.95598125e-01 2.12905690e-01 -8.34113359e-01 -1.83393881e-01 8.07226896e-01 6.61011994e-01 -1.42657816e+00 8.72837603e-01 -2.43332740e-02 2.78934330e-01 -8.96132767e-01 2.03528568e-01 -7.65922964e-01 -6.41973615e-01 -5.98625839e-01 4.97338586e-02 8.12126458e-01 1.60808161e-01 -5.60941935e-01 7.49240100e-01 1.41740835e-03 -1.96953937e-01 -6.36185706e-01 -1.04014099e+00 -6.23169243e-01 -4.47125167e-01 -2.33691111e-02 4.39662427e-01 5.24578333e-01 -2.54606396e-01 -9.03012380e-02 -3.37018698e-01 6.96570218e-01 7.30474293e-01 3.04203957e-01 1.07111490e+00 -1.14459288e+00 2.19777469e-02 -4.10613537e-01 -8.54474366e-01 -9.95311797e-01 -3.66407961e-01 -3.45747620e-01 7.30534568e-02 -1.30740964e+00 1.33080423e-01 -1.02728009e-01 -4.38176185e-01 3.36181790e-01 -2.51119673e-01 5.61827362e-01 8.11652094e-03 -1.95814613e-02 -7.76870191e-01 1.98945999e-01 1.43676412e+00 -2.98052486e-02 -3.39243226e-02 1.35384932e-01 -2.38625482e-01 1.05115926e+00 4.97433811e-01 -2.27285609e-01 2.22348999e-02 -2.17160523e-01 4.34507042e-01 -3.01388472e-01 5.39246678e-01 -1.82237256e+00 8.01331580e-01 1.10911056e-02 1.25782740e+00 -9.32948530e-01 -2.79725087e-03 -9.09772217e-01 8.53943347e-04 9.35319901e-01 2.32230872e-03 1.07455263e-02 2.28674576e-01 5.31270206e-01 -2.07321241e-01 -2.46304914e-01 1.13288724e+00 -6.38144016e-02 -8.32243741e-01 -3.77081446e-02 -2.02096105e-01 -1.65225923e-01 1.34977472e+00 -5.53709626e-01 -6.31671429e-01 -1.19504869e-01 -2.72963434e-01 4.02984619e-01 5.51829100e-01 3.69001389e-01 7.25341976e-01 -1.29280412e+00 -4.15727377e-01 7.25793660e-01 6.10013567e-02 -5.41521423e-02 3.01980257e-01 7.92412460e-01 -1.05233276e+00 2.99588829e-01 -5.34047067e-01 -5.58598638e-01 -1.34399569e+00 6.56676948e-01 5.43951988e-01 1.28930524e-01 -4.99926090e-01 6.13390028e-01 -1.25038624e-01 -1.25599071e-01 5.82347214e-01 -6.11120582e-01 -2.62539476e-01 -1.54721990e-01 5.04935145e-01 5.15690863e-01 7.28573129e-02 -2.73301870e-01 -4.08182859e-01 9.85025823e-01 1.48911908e-01 4.13841099e-01 8.22453797e-01 1.91746727e-01 9.35966969e-02 3.15067433e-02 7.25912035e-01 1.35696575e-01 -1.48208523e+00 1.36004701e-01 -2.56677896e-01 -7.65785515e-01 4.07814868e-02 -9.95689690e-01 -9.86495733e-01 6.72384739e-01 1.03335440e+00 2.75260713e-02 1.21439981e+00 -7.27329075e-01 5.98667443e-01 4.95013267e-01 1.65183514e-01 -1.44283307e+00 1.23588890e-01 4.77913648e-01 8.70388925e-01 -1.35758269e+00 2.65446931e-01 -2.84129143e-01 -2.11698100e-01 1.45233977e+00 9.98842895e-01 -6.84279501e-01 5.97141087e-01 5.17116070e-01 1.93469003e-01 1.59156442e-01 -4.88425761e-01 -1.27035424e-01 3.86815399e-01 5.19684613e-01 -1.25204489e-01 -2.21473873e-01 -3.32715839e-01 4.71174389e-01 -3.70134786e-02 1.19385898e-01 5.21811545e-01 6.76738977e-01 -6.69491589e-01 -3.84920448e-01 -8.59643161e-01 4.48444217e-01 -4.75497395e-01 -1.59895979e-02 -3.40049386e-01 9.03621852e-01 5.77861965e-01 7.59386718e-01 1.97785482e-01 -4.32859987e-01 4.31870341e-01 -3.07784230e-01 3.06013495e-01 -1.08560205e-01 -7.15921104e-01 8.98688808e-02 -4.24298495e-01 -5.20532906e-01 -2.92761028e-02 -4.28096801e-01 -9.86341178e-01 -2.57999599e-02 -6.32979155e-01 -3.19562525e-01 5.30791819e-01 5.61204135e-01 3.26608419e-01 5.53200901e-01 5.95010579e-01 -7.58924544e-01 -5.96127748e-01 -9.73506391e-01 -4.39642042e-01 3.91547292e-01 4.64222997e-01 -5.49598575e-01 -6.41613901e-01 -2.12935448e-01]
[8.637683868408203, -0.8387932181358337]
d3e9db0e-9e33-4c46-bc72-e9ae0d008c2b
cross-domain-few-shot-learning-with-meta-fine
2005.10544
null
https://arxiv.org/abs/2005.10544v4
https://arxiv.org/pdf/2005.10544v4.pdf
Cross-Domain Few-Shot Learning with Meta Fine-Tuning
In this paper, we tackle the new Cross-Domain Few-Shot Learning benchmark proposed by the CVPR 2020 Challenge. To this end, we build upon state-of-the-art methods in domain adaptation and few-shot learning to create a system that can be trained to perform both tasks. Inspired by the need to create models designed to be fine-tuned, we explore the integration of transfer-learning (fine-tuning) with meta-learning algorithms, to train a network that has specific layers that are designed to be adapted at a later fine-tuning stage. To do so, we modify the episodic training process to include a first-order MAML-based meta-learning algorithm, and use a Graph Neural Network model as the subsequent meta-learning module. We find that our proposed method helps to boost accuracy significantly, especially when combined with data augmentation. In our final results, we combine the novel method with the baseline method in a simple ensemble, and achieve an average accuracy of 73.78% on the benchmark. This is a 6.51% improvement over existing benchmarks that were trained solely on miniImagenet.
['Sheng Mei Shen', 'John Cai']
2020-05-21
null
null
null
null
['cross-domain-few-shot', 'cross-domain-few-shot-learning']
['computer-vision', 'computer-vision']
[ 3.63048851e-01 2.23741785e-01 -1.55666456e-01 -4.92172956e-01 -6.89089119e-01 -1.10439315e-01 8.83488119e-01 1.04649045e-01 -6.01095498e-01 6.36403561e-01 2.92156547e-01 1.94869593e-01 7.78111145e-02 -8.77845585e-01 -7.11034775e-01 -4.24603701e-01 8.40099156e-02 5.14483154e-01 8.23208988e-01 -7.24876881e-01 1.40808344e-01 1.91439033e-01 -1.68525946e+00 5.91845155e-01 7.27933645e-01 8.94285679e-01 2.65164226e-01 4.64445084e-01 -1.32707760e-01 9.96562123e-01 -4.82259661e-01 -4.32216078e-01 7.47263506e-02 -5.67696214e-01 -8.52242529e-01 -7.75729865e-02 3.03526700e-01 -1.70067623e-01 -3.25997770e-01 7.35417247e-01 7.42458522e-01 6.53110206e-01 6.85237825e-01 -7.29680061e-01 -6.52572095e-01 6.66493714e-01 -4.87502456e-01 3.17528307e-01 -1.39673918e-01 2.34561160e-01 6.58186913e-01 -8.82538378e-01 8.22164714e-01 9.68965471e-01 7.09188998e-01 1.15628850e+00 -1.13373494e+00 -5.70788085e-01 1.33255556e-01 6.41038835e-01 -1.17124677e+00 -6.24470592e-01 9.03448224e-01 -3.25962573e-01 1.36993122e+00 -2.12733984e-01 3.62504512e-01 1.40651965e+00 6.65280744e-02 6.28055394e-01 9.60717320e-01 -7.23728359e-01 5.40671349e-01 1.84078574e-01 9.58826989e-02 4.86579955e-01 -7.74886906e-02 1.48615152e-01 -3.81832689e-01 9.15598031e-03 3.43802005e-01 -3.24764922e-02 9.16695297e-02 -6.43997848e-01 -7.70432174e-01 1.12916923e+00 5.69737494e-01 5.46303213e-01 -3.79474550e-01 -5.77722967e-04 6.66807473e-01 1.41128004e-01 6.03466272e-01 8.02320778e-01 -4.55547035e-01 -5.30444756e-02 -1.13710988e+00 1.04647912e-01 8.29222322e-01 6.64560735e-01 7.71867156e-01 8.39189216e-02 -7.29492962e-01 1.34390450e+00 9.85932536e-03 -2.66048133e-01 9.08644915e-01 -7.75118947e-01 2.38579154e-01 5.37112236e-01 -2.60178477e-01 -4.67762560e-01 -4.52158123e-01 -5.10334611e-01 -7.38428473e-01 9.17394534e-02 -1.05272643e-01 -1.86669528e-01 -1.40526497e+00 1.77467835e+00 2.96097964e-01 6.51054561e-01 4.72329929e-02 6.49004877e-01 8.07760954e-01 5.96448004e-01 4.02620792e-01 -9.26785544e-02 1.14649153e+00 -1.49672782e+00 -5.47267973e-01 -3.62361342e-01 6.82039738e-01 -3.06376129e-01 1.12212694e+00 1.94837630e-01 -9.07038808e-01 -8.92780066e-01 -1.41892672e+00 1.11108214e-01 -8.09951544e-01 -3.34210634e-01 3.17909181e-01 5.19359231e-01 -1.04885495e+00 9.29373860e-01 -6.42016411e-01 -5.83554983e-01 6.23241127e-01 1.05312169e-01 -2.62995567e-02 -1.64888382e-01 -1.58361471e+00 1.23817551e+00 6.53873622e-01 -5.75466633e-01 -1.02438104e+00 -9.47743893e-01 -8.95712376e-01 5.83011471e-02 4.48860437e-01 -8.21574986e-01 1.37697148e+00 -8.48480761e-01 -1.78942490e+00 8.17203104e-01 1.08733863e-01 -9.11512554e-01 2.92346418e-01 -1.80514883e-02 -5.66894293e-01 1.40742689e-01 -9.38762501e-02 8.07278812e-01 9.56825197e-01 -1.00995946e+00 -5.20479500e-01 -1.35789469e-01 7.26585910e-02 7.07080737e-02 -5.72451413e-01 -8.11606348e-02 -6.66717172e-01 -6.78567767e-01 -6.22563958e-01 -7.83128679e-01 -4.54198897e-01 -5.01492202e-01 1.80148676e-01 -2.02505842e-01 7.31458724e-01 -4.99478191e-01 1.40461993e+00 -1.97055340e+00 2.02907354e-01 -1.03362240e-01 -2.88841669e-02 8.44197750e-01 -4.47623491e-01 3.65633696e-01 -3.94074582e-02 -1.60500363e-01 -5.34978807e-01 -6.27913654e-01 -7.00307414e-02 2.40242243e-01 -6.46400526e-02 4.01556045e-02 5.10985672e-01 1.11984348e+00 -9.47917879e-01 -2.70717114e-01 3.50067258e-01 5.72021663e-01 -6.59019113e-01 2.58530140e-01 -4.27281052e-01 3.38987470e-01 -1.34625033e-01 3.39511931e-01 4.40173894e-01 -2.67161787e-01 1.06376827e-01 -1.58003464e-01 1.74583092e-01 3.62391993e-02 -8.78116667e-01 2.23446560e+00 -5.86614668e-01 3.42735171e-01 -5.36490858e-01 -1.29169953e+00 1.05170393e+00 1.81189656e-01 4.14427191e-01 -1.00229275e+00 1.53480589e-01 4.28253338e-02 -1.80330090e-02 -4.60395813e-01 4.63526696e-01 -1.82072297e-01 -2.65526641e-02 2.47708887e-01 7.78666198e-01 7.68077821e-02 3.39112967e-01 9.48679224e-02 1.35106099e+00 3.02400649e-01 5.21308184e-01 2.07866915e-02 5.97951293e-01 -1.65614672e-03 4.36246037e-01 8.23355913e-01 -3.29974622e-01 7.35630333e-01 5.43235280e-02 -4.23270971e-01 -1.20479202e+00 -7.29643703e-01 1.06158048e-01 1.62542570e+00 -1.40582174e-01 -3.44612926e-01 -8.59667063e-01 -9.31728482e-01 -2.87102044e-01 9.84483123e-01 -1.02268720e+00 -7.56182253e-01 -5.01992226e-01 -8.80523443e-01 4.29728597e-01 6.30421221e-01 6.51506305e-01 -1.19013667e+00 -6.27478302e-01 5.86070478e-01 7.00587705e-02 -1.03351438e+00 -3.37412536e-01 5.11334717e-01 -7.59078503e-01 -7.97694981e-01 -1.07382715e+00 -7.73626029e-01 2.10538879e-01 -3.91971432e-02 1.21188438e+00 -1.36228070e-01 -3.27696085e-01 4.03209507e-01 -7.81069398e-01 -4.10273105e-01 -4.38544571e-01 5.08301795e-01 -1.94006875e-01 1.07006513e-01 6.34751678e-01 -6.30319238e-01 -6.84092343e-01 1.60838664e-01 -9.80233431e-01 -1.24674343e-01 6.12620592e-01 1.14828181e+00 6.05371356e-01 -3.32993478e-01 8.36173415e-01 -1.25358963e+00 5.91788530e-01 -6.25103652e-01 -2.12911353e-01 4.01824594e-01 -8.52462113e-01 2.83437967e-01 5.83323419e-01 -6.29336536e-01 -1.37097967e+00 2.60192543e-01 -3.74651074e-01 -5.74892104e-01 -1.75988212e-01 5.06685615e-01 6.73519298e-02 -3.14990580e-01 1.18200088e+00 2.33085901e-01 -1.58754542e-01 -5.98641157e-01 7.29750574e-01 6.40018344e-01 4.86378193e-01 -2.37670764e-01 4.89837527e-01 3.01607996e-01 -3.74440730e-01 -6.02411687e-01 -1.11238611e+00 -6.37510657e-01 -8.04187000e-01 -9.22157019e-02 8.81412029e-01 -9.43163395e-01 7.41766542e-02 3.94733459e-01 -8.75697792e-01 -7.12853670e-01 -5.98731935e-01 2.15774432e-01 -8.25988889e-01 1.15448616e-01 -3.94726217e-01 -3.73451501e-01 -5.30363858e-01 -7.27486134e-01 7.43655026e-01 3.33332568e-01 -1.70444056e-01 -1.14490831e+00 6.90199792e-01 1.18332654e-01 8.28280509e-01 1.88860908e-01 6.00579083e-01 -1.11509395e+00 1.45306021e-01 3.43789905e-02 -1.15916640e-01 4.33717936e-01 -4.66526747e-02 -4.89659816e-01 -1.23162293e+00 -4.00279880e-01 1.14991859e-01 -5.84801555e-01 1.32752979e+00 3.90261710e-01 1.14987624e+00 1.04109555e-01 -3.67745668e-01 7.43327379e-01 1.46936083e+00 1.81247875e-01 8.82308304e-01 6.77335918e-01 4.51926410e-01 3.90593380e-01 5.94215035e-01 5.03041267e-01 5.55568159e-01 8.10125172e-01 2.74894148e-01 9.00338441e-02 -6.21462882e-01 -2.57812798e-01 1.59838736e-01 7.26366818e-01 -1.09895557e-01 1.75649002e-01 -1.01350176e+00 7.05409586e-01 -2.11775923e+00 -1.10210085e+00 4.94605035e-01 1.95651591e+00 9.69266653e-01 2.24453598e-01 1.93249270e-01 -2.12500528e-01 6.38336599e-01 3.21633250e-01 -7.30740607e-01 -5.88727057e-01 1.33424997e-01 6.94619477e-01 4.28864747e-01 2.01907575e-01 -1.27385020e+00 1.20050585e+00 6.15169096e+00 9.40434575e-01 -1.10488236e+00 5.48162401e-01 4.86743867e-01 -1.05050363e-01 5.97224683e-02 -1.75219461e-01 -9.56076324e-01 3.11572343e-01 1.50824189e+00 -3.40778641e-02 5.39637804e-01 8.90946865e-01 -2.47306690e-01 2.29675397e-01 -9.25354958e-01 7.33372808e-01 4.67025101e-01 -1.56587017e+00 4.03623059e-02 -2.82736331e-01 1.09592474e+00 3.89326334e-01 -8.40483308e-02 1.13521767e+00 2.74372041e-01 -7.96502233e-01 2.46105865e-01 6.73995733e-01 5.97175419e-01 -7.89399147e-01 6.54271841e-01 4.02229190e-01 -1.07917535e+00 -2.36618608e-01 -7.20566988e-01 2.99563147e-02 9.07349810e-02 2.85882622e-01 -9.24982607e-01 5.87743580e-01 6.56056523e-01 7.71363437e-01 -7.21064210e-01 1.25774229e+00 -1.93446577e-01 4.27148342e-01 1.04344733e-01 -3.54868285e-02 2.47365475e-01 3.99268299e-01 3.53673160e-01 1.32792556e+00 2.13693663e-01 5.68826832e-02 -2.67244820e-02 8.07516098e-01 -1.96725532e-01 6.34226352e-02 -5.31354845e-01 8.01608711e-02 3.18514138e-01 1.43691516e+00 -4.79173750e-01 -6.93503857e-01 -5.95068216e-01 1.07269478e+00 7.75516093e-01 2.62088716e-01 -8.62295151e-01 -6.11699104e-01 3.13256174e-01 -3.11838668e-02 8.65082979e-01 4.84363325e-02 -2.56372616e-02 -1.24876046e+00 -4.39908832e-01 -7.98125982e-01 5.25658429e-01 -5.93330741e-01 -1.50231266e+00 8.59681726e-01 2.05095503e-02 -1.12306237e+00 -6.01294935e-01 -4.21459556e-01 -7.22241223e-01 7.58695066e-01 -1.79559398e+00 -1.24911392e+00 -2.72509456e-01 7.36357272e-01 7.00214267e-01 -4.26591307e-01 1.07695174e+00 3.03374916e-01 -4.00590241e-01 7.58252203e-01 2.62796637e-02 1.62703684e-03 9.42917526e-01 -1.11628890e+00 7.04013467e-01 7.62200415e-01 2.25088730e-01 3.20319474e-01 5.78342259e-01 -5.68249762e-01 -9.21894729e-01 -1.45312524e+00 6.84227645e-01 -4.62821007e-01 6.67250156e-01 -2.71881312e-01 -1.17440104e+00 4.29913193e-01 3.42504680e-01 3.01538795e-01 7.45300174e-01 1.41330555e-01 -5.10612428e-01 -1.89319979e-02 -1.17282677e+00 3.70723993e-01 1.09822452e+00 -4.66915816e-01 -1.14916563e+00 -4.64792401e-02 9.19799745e-01 -3.28478813e-01 -8.91252756e-01 6.00302935e-01 3.24173063e-01 -7.93180466e-01 9.37120378e-01 -8.84237826e-01 4.16385293e-01 8.74384567e-02 -1.77733362e-01 -1.75665760e+00 -6.45047307e-01 -2.86976635e-01 -6.35580540e-01 1.31884885e+00 4.49503213e-01 -4.08109158e-01 6.79992199e-01 4.23407584e-01 -3.01317155e-01 -7.52909005e-01 -8.08856428e-01 -9.21728909e-01 1.59986809e-01 -2.85975128e-01 4.94866699e-01 9.54657912e-01 1.13683969e-01 6.06039882e-01 -5.84625602e-01 -3.52903992e-01 5.53357363e-01 -1.71463639e-01 6.88666761e-01 -1.27057147e+00 -5.47646344e-01 -3.31375211e-01 -2.98827529e-01 -5.29745698e-01 1.79285049e-01 -9.48145330e-01 1.08603969e-01 -1.46680140e+00 3.75627846e-01 1.39087904e-03 -8.39358151e-01 6.58086479e-01 -2.05564886e-01 3.24289382e-01 2.44996533e-01 5.60695305e-02 -8.25186312e-01 7.69214153e-01 1.03143597e+00 -2.71169484e-01 -5.28602779e-01 -5.92279322e-02 -8.34496558e-01 4.27247137e-01 9.23600554e-01 -4.39353049e-01 -4.19987231e-01 -1.79921761e-01 -1.14350379e-01 -3.83228868e-01 4.97977212e-02 -1.28419495e+00 3.99058223e-01 1.97182998e-01 4.51051176e-01 -1.06092498e-01 5.02351284e-01 -5.61121404e-01 -2.01166093e-01 5.19683838e-01 -3.29676628e-01 -5.07096112e-01 5.22133112e-01 6.42507315e-01 -2.06817791e-01 -4.08484370e-01 1.09978056e+00 -1.89040080e-01 -1.31169546e+00 4.83421534e-01 4.02694456e-02 1.84329420e-01 1.16723824e+00 3.01572420e-02 -5.26639581e-01 -8.12661052e-02 -9.72767174e-01 1.53558120e-01 2.91228920e-01 7.20542550e-01 5.23929238e-01 -1.49188769e+00 -6.44675732e-01 1.56767324e-01 6.36875868e-01 -5.15437901e-01 5.16425550e-01 7.50017405e-01 1.07217520e-01 1.87380046e-01 -5.37235677e-01 -3.82590383e-01 -7.73523688e-01 8.31800997e-01 3.90942991e-01 -5.83899200e-01 -5.33508122e-01 8.31829488e-01 -3.15419078e-01 -4.95300204e-01 1.54797226e-01 4.47471365e-02 -6.44332111e-01 3.32325667e-01 7.89911270e-01 3.00229192e-01 2.29122460e-01 -3.19875360e-01 -3.17065150e-01 3.88966024e-01 -3.99394512e-01 -1.57540858e-01 1.73810637e+00 6.75087358e-05 3.84632826e-01 4.91621464e-01 1.18138778e+00 -5.35065055e-01 -1.42905402e+00 -5.93281627e-01 1.19844437e-01 -1.53597165e-02 9.84846354e-02 -1.14009583e+00 -8.34810436e-01 8.59112918e-01 8.56924117e-01 4.73181009e-02 1.15778971e+00 1.06320240e-01 6.38451338e-01 3.37293297e-01 2.66434461e-01 -1.42155540e+00 2.42219284e-01 8.47614944e-01 7.00357080e-01 -1.49630105e+00 -2.20675930e-01 2.14983970e-01 -8.13369989e-01 9.91037369e-01 7.71425784e-01 -3.89844149e-01 7.09525824e-01 -1.52149588e-01 -2.98490971e-01 -1.09074377e-01 -1.07841063e+00 -4.66006160e-01 6.69238806e-01 8.45180631e-01 3.14148009e-01 -2.05880627e-01 -3.23011756e-01 9.10172701e-01 2.89503127e-01 4.06969100e-01 1.55309737e-01 9.07963455e-01 -7.04156101e-01 -1.20723379e+00 1.62776351e-01 4.80578810e-01 -1.74960822e-01 -1.70887351e-01 -2.56354928e-01 5.77025592e-01 7.84129649e-02 7.52955616e-01 1.10835563e-02 -6.72907233e-01 5.69355309e-01 5.06176949e-01 5.30097544e-01 -1.00457513e+00 -6.81826234e-01 -2.86392450e-01 -1.57072563e-02 -7.59952843e-01 -6.15751326e-01 -3.79167706e-01 -9.87886310e-01 6.23569787e-02 -1.87967625e-02 -1.23926438e-01 5.61832607e-01 1.09029675e+00 5.70039690e-01 9.87795353e-01 4.90565896e-01 -1.02728081e+00 -7.15155423e-01 -1.28123760e+00 -3.35180461e-01 3.91654789e-01 1.10552549e-01 -8.70447040e-01 -9.20531824e-02 -5.77836744e-02]
[9.938042640686035, 2.9019992351531982]
b0282133-cb6b-4451-b92f-c1f731d3c395
on-pitfalls-and-advantages-of-sophisticated
2303.17511
null
https://arxiv.org/abs/2303.17511v1
https://arxiv.org/pdf/2303.17511v1.pdf
On pitfalls (and advantages) of sophisticated large language models
Natural language processing based on large language models (LLMs) is a booming field of AI research. After neural networks have proven to outperform humans in games and practical domains based on pattern recognition, we might stand now at a road junction where artificial entities might eventually enter the realm of human communication. However, this comes with serious risks. Due to the inherent limitations regarding the reliability of neural networks, overreliance on LLMs can have disruptive consequences. Since it will be increasingly difficult to distinguish between human-written and machine-generated text, one is confronted with new ethical challenges. This begins with the no longer undoubtedly verifiable human authorship and continues with various types of fraud, such as a new form of plagiarism. This also concerns the violation of privacy rights, the possibility of circulating counterfeits of humans, and, last but not least, it makes a massive spread of misinformation possible.
['Anna Strasser']
2023-02-25
null
null
null
null
['misinformation']
['miscellaneous']
[ 1.96469530e-01 4.64307725e-01 1.65857166e-01 1.16923593e-01 -3.39138448e-01 -6.88721120e-01 7.42811382e-01 4.94064569e-01 -9.68845725e-01 9.82600391e-01 5.61667002e-05 -4.87834662e-01 2.82055456e-02 -9.53778625e-01 -3.17204088e-01 -2.33223170e-01 -4.59577590e-02 2.52099633e-01 4.72793318e-02 -1.01616092e-01 6.87257826e-01 5.51359177e-01 -1.04320884e+00 -8.86579528e-02 7.56591618e-01 8.50955307e-01 -4.44318444e-01 4.05918211e-01 -2.76612848e-01 1.25038004e+00 -8.93552244e-01 -1.34811985e+00 3.07867914e-01 -2.69522518e-01 -7.40680218e-01 -3.40398043e-01 1.97235093e-01 -4.91946787e-01 -1.36192441e-01 1.44594109e+00 9.76679102e-02 -3.97820286e-02 2.48740330e-01 -1.34262002e+00 -5.74206769e-01 4.72480148e-01 -6.38624251e-01 -2.96531599e-02 2.40213186e-01 1.60430700e-01 7.24991918e-01 -5.47177792e-01 7.61119187e-01 1.16107833e+00 7.17242539e-01 3.05895895e-01 -9.07145858e-01 -7.54720569e-01 -3.20020944e-01 2.26881336e-02 -1.22190332e+00 -2.60349125e-01 4.28433180e-01 -5.51020861e-01 6.01657391e-01 2.73065090e-01 6.78116560e-01 1.03026903e+00 4.71269548e-01 2.97027618e-01 7.63944745e-01 -3.58871222e-01 3.90253961e-01 2.85163671e-01 -1.37418509e-01 6.02742314e-01 1.10854936e+00 3.07973027e-01 -4.40962940e-01 -4.61369514e-01 5.42369604e-01 -1.86831087e-01 -1.87610731e-01 -1.56462282e-01 -1.23667443e+00 9.92712021e-01 1.36590183e-01 4.55394268e-01 -4.73363489e-01 -4.02290486e-02 5.37895143e-01 1.91301197e-01 9.12968516e-02 8.87110353e-01 -5.17644025e-02 -5.35092533e-01 -8.50192130e-01 1.77606374e-01 1.05048501e+00 2.95977920e-01 4.31661010e-01 -1.26996458e-01 5.97122490e-01 5.00805616e-01 3.04100513e-01 4.37808223e-02 5.00430226e-01 -8.59208703e-01 3.93823951e-01 6.33896708e-01 1.93568230e-01 -1.71568716e+00 -2.64145508e-02 -4.86474186e-01 -9.93432164e-01 4.17737246e-01 7.85448551e-01 -4.48032409e-01 -3.09433341e-01 1.57361889e+00 1.09768994e-01 -2.78197110e-01 -2.08101273e-02 8.77762556e-01 2.14303166e-01 5.54015577e-01 2.78936416e-01 -4.95010614e-02 1.16294110e+00 -3.24952871e-01 -6.40336931e-01 -6.12234533e-01 4.21826452e-01 -5.14261365e-01 4.35866654e-01 5.61361134e-01 -9.46277499e-01 1.56551048e-01 -1.09555709e+00 6.60446361e-02 -5.87772906e-01 -6.19018078e-01 8.45833361e-01 9.05388951e-01 -6.28533006e-01 7.24964321e-01 -3.67442280e-01 -4.78135467e-01 6.85594499e-01 2.87269384e-01 -6.52086556e-01 1.02210447e-01 -1.48915863e+00 1.18231654e+00 4.84449595e-01 4.43765402e-01 -4.84043062e-02 -3.41000080e-01 -6.94187045e-01 3.76371257e-02 5.57317138e-01 -4.15824920e-01 8.58141243e-01 -1.19982946e+00 -9.01775360e-01 9.74553347e-01 2.31439844e-01 -7.23704100e-01 9.65624154e-01 -1.51718453e-01 -3.69906992e-01 3.44060846e-02 1.18520729e-01 2.92501360e-01 6.19080067e-01 -9.19666708e-01 -5.61240613e-01 -4.57395107e-01 4.61768098e-02 1.47001565e-01 -4.03521955e-01 2.71948069e-01 2.03321487e-01 -4.60144699e-01 -2.52150595e-01 -6.26105547e-01 -3.92292947e-01 2.86219686e-01 -3.15543860e-01 -3.97244208e-02 3.15614372e-01 -5.29834867e-01 1.11015260e+00 -2.22632813e+00 -3.34918231e-01 1.56829208e-01 7.00718701e-01 5.49774468e-01 2.40236327e-01 4.12503123e-01 3.69231433e-01 7.97375500e-01 -2.82969046e-02 1.20053232e-01 -1.14901312e-01 -2.60321498e-01 -2.29041442e-01 3.67331266e-01 1.78349689e-01 8.97128046e-01 -1.00743186e+00 -3.85108173e-01 -1.51309028e-01 1.62011564e-01 -2.16720819e-01 -1.61180064e-01 2.21940756e-01 7.33122304e-02 -3.47420931e-01 2.77598977e-01 4.05689657e-01 -3.85568231e-01 2.15946689e-01 5.83881199e-01 -1.45038933e-01 2.02528536e-01 -9.19765890e-01 9.79770243e-01 -4.30693328e-02 8.92236710e-01 1.62274569e-01 -7.87819207e-01 6.22466326e-01 3.02894503e-01 1.05845161e-01 -4.83833849e-01 5.35094619e-01 2.54159659e-01 2.98047513e-01 -4.70452607e-01 7.92613983e-01 -4.07680750e-01 -1.47427976e-01 8.41154456e-01 -3.08159292e-01 1.32624730e-01 -6.47399947e-02 3.20445538e-01 1.19408107e+00 -3.41016978e-01 4.55650777e-01 -3.81871015e-02 1.36168823e-01 1.77015767e-01 6.53245747e-01 6.95817769e-01 -5.73580980e-01 2.61795700e-01 6.80711329e-01 -5.74040473e-01 -1.03771758e+00 -9.79586124e-01 1.70519739e-01 5.52785397e-01 -2.30375770e-03 -1.81364492e-01 -5.52263200e-01 -4.13670748e-01 1.03627443e-01 7.41014838e-01 -3.22194010e-01 -3.21500331e-01 -1.82571083e-01 -5.06149173e-01 8.96036804e-01 4.16963641e-03 6.11032069e-01 -1.09677625e+00 -1.10031998e+00 1.19450249e-01 -8.14224929e-02 -1.17875111e+00 -4.56992835e-02 -1.48914129e-01 -5.98452389e-01 -9.77825403e-01 -6.81325972e-01 -3.38035434e-01 6.39532506e-01 2.99640950e-02 7.49982297e-01 3.16192716e-01 -3.32446188e-01 -1.01676397e-01 -1.19884655e-01 -8.35880160e-01 -9.09185886e-01 -1.14355065e-01 -8.32137167e-02 -1.15880117e-01 6.13325417e-01 -4.12498146e-01 -3.78419369e-01 -1.94398448e-01 -9.10163581e-01 -1.35204699e-02 4.70676512e-01 4.59123284e-01 -3.62497598e-01 1.70331374e-01 9.64501441e-01 -1.22568798e+00 1.12452197e+00 -6.00256503e-01 -5.60393214e-01 2.42336586e-01 -8.06610286e-01 -2.96745479e-01 6.34835660e-01 -4.38710093e-01 -8.57857585e-01 -4.58540469e-01 2.83619761e-01 -1.96372345e-02 -3.47550899e-01 7.99877942e-01 1.43238217e-01 -4.28401716e-02 8.08460772e-01 -1.38496384e-01 3.28483254e-01 9.25209671e-02 1.81360111e-01 8.97002339e-01 3.06588918e-01 7.95338973e-02 8.95378411e-01 2.98175395e-01 -1.70828924e-01 -1.12455463e+00 -3.43923718e-01 9.76738557e-02 -2.75983840e-01 -4.83438253e-01 7.56786466e-01 -6.93536758e-01 -7.94921398e-01 6.73417568e-01 -1.24328732e+00 1.07595533e-01 -6.16911165e-02 3.60331625e-01 1.01813532e-01 7.76974261e-01 -6.97547138e-01 -8.36046934e-01 -1.28865734e-01 -7.20459938e-01 -2.91829854e-02 3.98473024e-01 -7.66152263e-01 -8.05257201e-01 -2.72721529e-01 6.97704434e-01 5.79428554e-01 4.04564053e-01 9.76177692e-01 -8.17412496e-01 -5.67016184e-01 -1.10679281e+00 -2.71170527e-01 3.06441247e-01 1.77611649e-01 -1.12523697e-01 -7.51298487e-01 -1.26431473e-02 1.31671414e-01 -4.13522720e-01 1.79023251e-01 -2.03512430e-01 4.09018993e-01 -6.72411740e-01 -8.78498182e-02 -1.49309888e-01 1.16380823e+00 5.91488123e-01 6.98013246e-01 4.29729819e-01 3.72102946e-01 7.82943070e-01 1.86856329e-01 4.04965580e-01 3.08242232e-01 -2.64845174e-02 1.91452026e-01 3.38282466e-01 6.20448947e-01 -5.28998256e-01 -2.79202443e-02 3.78648311e-01 6.64534234e-03 -3.51812929e-01 -1.16176140e+00 3.58848929e-01 -1.71843827e+00 -1.03315163e+00 8.09543952e-03 2.74130940e+00 5.44460237e-01 4.93583113e-01 -7.97116235e-02 9.16083977e-02 8.27424049e-01 3.51947583e-02 -4.69569087e-01 -7.20570624e-01 3.92901525e-02 -1.03671722e-01 5.57116926e-01 2.13650599e-01 -7.05512226e-01 7.98975229e-01 6.02562952e+00 3.42389852e-01 -1.22266567e+00 -1.07247196e-01 7.79987693e-01 1.11532494e-01 -1.12216748e-01 3.78260091e-02 -3.12382549e-01 5.97374260e-01 1.06735754e+00 -7.72998512e-01 3.99632007e-01 4.58159417e-01 7.55431429e-02 -6.47233009e-01 -7.42672086e-01 9.15274858e-01 1.39766499e-01 -1.29788554e+00 -7.82052651e-02 4.67711180e-01 1.44982278e-01 -7.76441842e-02 -2.99362153e-01 9.70182642e-02 4.40276772e-01 -1.18939912e+00 7.68226802e-01 2.34953016e-01 4.70308036e-01 -7.90417016e-01 8.79411817e-01 7.95287013e-01 -3.00460368e-01 -1.46165743e-01 -3.82536501e-01 -6.05354190e-01 1.46229416e-01 4.25343812e-01 -6.97268903e-01 5.19101769e-02 1.22716129e-01 2.30309367e-01 -3.84905398e-01 1.10860360e+00 -3.74413073e-01 1.64027095e-01 -2.11643621e-01 -5.35230458e-01 7.71247149e-02 -1.04575396e-01 4.34739023e-01 6.69344783e-01 1.22773096e-01 1.97944269e-01 -5.13723910e-01 8.56068194e-01 -4.43275422e-01 -1.53363366e-02 -1.05636096e+00 -8.98498774e-01 6.83960676e-01 1.14609444e+00 -9.27134812e-01 -5.43096401e-02 -5.16987622e-01 1.01525891e+00 3.29937428e-01 1.03963763e-01 -4.37884897e-01 -4.29871380e-01 3.75565320e-01 3.01839888e-01 -2.86335796e-01 -2.56180584e-01 -4.57053840e-01 -1.30152774e+00 -7.54962116e-02 -9.77086186e-01 1.42467186e-01 -5.04333258e-01 -1.25180793e+00 5.23232341e-01 -4.70821351e-01 -8.84157062e-01 -3.87421519e-01 -2.23508373e-01 -4.87424105e-01 6.09911025e-01 -8.36710989e-01 -6.31565213e-01 3.00461471e-01 -7.64463423e-03 1.34262800e-01 -2.20423535e-01 8.26720893e-01 2.80911416e-01 -1.97084174e-01 6.22155905e-01 4.94341403e-02 5.26043355e-01 5.26342332e-01 -4.77765054e-01 5.92349529e-01 7.98716187e-01 1.83498815e-01 7.24316597e-01 5.87784708e-01 -6.93159640e-01 -1.23950243e+00 -7.16516674e-01 1.40713203e+00 -4.67130452e-01 9.45021033e-01 -5.76699018e-01 -8.61690104e-01 6.93600357e-01 6.95584938e-02 -5.37549853e-01 7.14820921e-01 1.12023063e-01 -2.58080661e-01 2.74681628e-01 -1.23927879e+00 8.84791255e-01 5.52609146e-01 -7.38735616e-01 -6.19920254e-01 1.21433221e-01 4.13146853e-01 6.84980676e-02 -3.93379569e-01 -5.46955541e-02 8.02703440e-01 -8.35800827e-01 5.57364225e-01 -7.86938488e-01 7.05070019e-01 1.38762230e-02 3.18036079e-01 -9.11288202e-01 -3.06855273e-02 -7.03910530e-01 5.68564057e-01 1.13180280e+00 6.27103508e-01 -9.41991150e-01 9.07482147e-01 1.58234525e+00 6.43896818e-01 -3.83338600e-01 -1.06850541e+00 -6.46877348e-01 3.32214922e-01 -5.03578126e-01 2.12079540e-01 1.32287610e+00 6.42690003e-01 4.52499539e-01 -5.34854889e-01 -9.95843932e-02 8.41500401e-01 -3.06412369e-01 6.10927820e-01 -1.42150629e+00 7.60172606e-02 -5.45362175e-01 -8.19567978e-01 -4.30654854e-01 -1.40962794e-01 -5.69464207e-01 2.49231048e-02 -1.24752223e+00 1.55178279e-01 -8.53085071e-02 -4.83996645e-02 1.87980548e-01 1.74878448e-01 1.35521442e-01 6.25703275e-01 2.54088283e-01 -5.26091099e-01 1.33583829e-01 9.37987447e-01 1.49693877e-01 -2.63278056e-02 -3.15669999e-02 -9.60843682e-01 8.32555294e-01 7.95460105e-01 -4.69652891e-01 -2.16622069e-01 -1.69369176e-01 1.06885040e+00 3.14068317e-01 4.04622138e-01 -1.01848769e+00 5.33872128e-01 -1.94067642e-01 1.69403359e-01 1.33809462e-01 6.31555244e-02 -8.73043954e-01 2.45163292e-01 5.97033322e-01 -4.93681163e-01 -1.33712187e-01 2.36964166e-01 5.97640276e-01 -8.74055848e-02 -4.46488082e-01 7.10758567e-01 -3.63807976e-01 -2.94522524e-01 -5.70970587e-02 -9.93477702e-01 1.03472881e-01 1.10734236e+00 -3.62953067e-01 -5.32501161e-01 -6.83016658e-01 -3.53652298e-01 2.39535838e-01 6.63507700e-01 6.52874887e-01 5.19241452e-01 -7.96060026e-01 -2.95151085e-01 1.19199909e-01 -9.74892229e-02 -1.45480350e-01 1.24750637e-01 4.67896312e-01 -7.06023991e-01 2.99960226e-01 -3.03488374e-01 3.28582406e-01 -1.02960443e+00 4.29054737e-01 1.49079844e-01 2.98000779e-03 -5.21063685e-01 7.84287214e-01 -4.10376489e-02 -5.30912243e-02 2.97227055e-01 1.34872779e-01 -4.15653884e-02 1.16359487e-01 6.15554869e-01 3.01758498e-01 -2.54604399e-01 -6.74664497e-01 -3.08703870e-01 -2.45799422e-01 -3.95902544e-01 -2.47971758e-01 1.16014087e+00 -6.95587173e-02 -4.17618036e-01 6.60821438e-01 5.76632798e-01 1.67585358e-01 -7.47780323e-01 7.17857108e-02 2.27679536e-01 -4.89205092e-01 -1.79503694e-01 -1.06779468e+00 -6.25652850e-01 9.63534296e-01 1.99679285e-01 5.45296490e-01 4.82744604e-01 -3.34853649e-01 8.72327864e-01 4.34639841e-01 8.10755789e-01 -1.07436991e+00 -2.92535007e-01 4.02763605e-01 4.44729388e-01 -1.34765875e+00 2.80531198e-02 -3.40210809e-03 -6.37030363e-01 1.08207488e+00 4.46156561e-01 1.99738115e-01 5.97531140e-01 1.04136482e-01 1.12676062e-01 -1.47131145e-01 -5.47572076e-01 4.89184529e-01 -4.36246157e-01 3.80145758e-01 4.45081502e-01 7.44266585e-02 -8.35077763e-01 6.02692723e-01 -2.66182512e-01 4.27451640e-01 1.07507932e+00 9.94408607e-01 -4.47905034e-01 -1.08273458e+00 -4.07303423e-01 6.21878803e-01 -8.39689612e-01 -1.73783600e-01 -7.72085071e-01 8.71477425e-01 -1.33523913e-02 8.85962963e-01 -5.25821149e-02 -4.25757498e-01 -1.46769047e-01 6.24081455e-02 -1.02682495e-02 -4.27071422e-01 -5.75473070e-01 -4.98483002e-01 1.81057036e-01 -2.56637543e-01 1.27946585e-01 -6.84716523e-01 -1.20007491e+00 -8.11227083e-01 -2.93815374e-01 2.12877780e-01 6.81402385e-01 9.14969206e-01 5.49548805e-01 -3.15746158e-01 6.13856502e-02 -1.77733228e-01 -6.42495155e-01 -6.81611419e-01 -7.59408951e-01 2.85222977e-01 1.89375564e-01 -1.29843801e-01 -3.45535398e-01 -2.48278975e-01]
[8.935675621032715, 6.59321928024292]
880b4212-9baa-49b0-a712-c5918c0cd483
icface-interpretable-and-controllable-face
1904.01909
null
https://arxiv.org/abs/1904.01909v2
https://arxiv.org/pdf/1904.01909v2.pdf
ICface: Interpretable and Controllable Face Reenactment Using GANs
This paper presents a generic face animator that is able to control the pose and expressions of a given face image. The animation is driven by human interpretable control signals consisting of head pose angles and the Action Unit (AU) values. The control information can be obtained from multiple sources including external driving videos and manual controls. Due to the interpretable nature of the driving signal, one can easily mix the information between multiple sources (e.g. pose from one image and expression from another) and apply selective post-production editing. The proposed face animator is implemented as a two-stage neural network model that is learned in a self-supervised manner using a large video collection. The proposed Interpretable and Controllable face reenactment network (ICface) is compared to the state-of-the-art neural network-based face animation techniques in multiple tasks. The results indicate that ICface produces better visual quality while being more versatile than most of the comparison methods. The introduced model could provide a lightweight and easy to use tool for a multitude of advanced image and video editing tasks.
['Esa Rahtu', 'Soumya Tripathy', 'Juho Kannala']
2019-04-03
null
null
null
null
['face-reenactment']
['computer-vision']
[ 3.69394451e-01 2.05248863e-01 7.26185217e-02 -5.60289323e-01 6.89043989e-03 -4.38208491e-01 7.68723249e-01 -4.47121918e-01 -3.38837564e-01 5.66852391e-01 -2.55745530e-01 2.06262410e-01 1.00521453e-01 -3.02489221e-01 -7.40203142e-01 -6.90099418e-01 -4.65960940e-03 6.16995096e-01 -1.14750803e-01 -4.10928279e-01 1.45360425e-01 1.00978541e+00 -2.22372794e+00 1.57574099e-02 4.54964519e-01 9.46653306e-01 2.39473522e-01 7.97364056e-01 6.36030734e-02 8.12483370e-01 -7.09168613e-01 7.75235966e-02 1.25220522e-01 -4.89778131e-01 -2.58455575e-01 2.34918445e-01 7.03772664e-01 -4.09262359e-01 9.02534798e-02 9.50624943e-01 6.41365170e-01 1.17381364e-01 5.47646880e-01 -1.62113154e+00 -3.82487386e-01 1.77776158e-01 -4.32719409e-01 -1.19338311e-01 5.58013558e-01 2.19528794e-01 3.17912221e-01 -9.32322264e-01 9.39005196e-01 1.49033344e+00 2.41053388e-01 9.24973130e-01 -1.19231391e+00 -8.64605725e-01 1.04810312e-01 2.17093825e-01 -1.11423624e+00 -7.00235128e-01 1.11687839e+00 -4.78969872e-01 5.82700074e-01 4.32423115e-01 8.37454498e-01 1.18064713e+00 1.46990210e-01 3.39030594e-01 1.14819586e+00 -5.94323099e-01 2.52545029e-01 4.93517488e-01 -2.49279901e-01 9.18298960e-01 -2.23280683e-01 1.17626242e-01 -5.94975471e-01 8.90255421e-02 1.05379546e+00 -3.23478729e-01 -5.29407203e-01 -2.31773928e-01 -7.66325533e-01 7.22055793e-01 1.45167246e-01 1.04031548e-01 -4.19937074e-01 2.74105757e-01 3.84450346e-01 5.63825250e-01 1.97398692e-01 2.38045946e-01 -3.83527815e-01 1.26884803e-02 -8.39565575e-01 1.84048504e-01 8.78469765e-01 8.67674112e-01 6.22589052e-01 7.26383209e-01 -1.77448973e-01 5.71557224e-01 6.94629490e-01 5.99063516e-01 3.88248295e-01 -1.12238920e+00 -7.42600709e-02 2.86568820e-01 1.49234742e-01 -1.41059816e+00 -3.70601386e-01 1.06065154e-01 -6.52046561e-01 1.09229088e+00 3.40340883e-01 -3.23869407e-01 -9.11647081e-01 1.99522460e+00 6.67047560e-01 9.30052437e-03 -2.78863519e-01 1.01151395e+00 1.00345242e+00 6.92295551e-01 1.02299377e-01 -5.55121481e-01 1.35894251e+00 -7.27592528e-01 -1.20791233e+00 -6.23581707e-02 -3.22144985e-01 -6.91562712e-01 8.84686410e-01 6.06717348e-01 -1.03784919e+00 -9.63717222e-01 -1.12483335e+00 7.85194524e-03 -2.62190610e-01 5.09399056e-01 3.82778645e-01 5.34517765e-01 -1.36637449e+00 6.50066733e-01 -7.81802833e-01 -3.39564502e-01 -2.36522984e-02 8.60704184e-01 -6.21694446e-01 6.21250987e-01 -1.07986414e+00 9.62817371e-01 6.27569705e-02 4.64590311e-01 -9.53278899e-01 -3.74197394e-01 -8.24139118e-01 -1.37807444e-01 3.18282068e-01 -4.44945365e-01 1.11830473e+00 -1.83821523e+00 -2.23262119e+00 8.84705722e-01 -5.03363349e-02 4.73933481e-02 7.14276791e-01 -2.62366295e-01 -2.97461182e-01 5.76896787e-01 -2.69191474e-01 1.04648674e+00 1.69853687e+00 -1.37417531e+00 -2.02957138e-01 -4.57913160e-01 -1.42306507e-01 1.60425961e-01 -2.51169592e-01 3.44280511e-01 -6.39013946e-01 -6.64330363e-01 -5.26617944e-01 -1.11832833e+00 2.63188213e-01 7.98273802e-01 -1.84634820e-01 -5.17454855e-02 1.34943318e+00 -7.39863396e-01 9.60877836e-01 -1.79534292e+00 5.70163846e-01 1.51169017e-01 8.57822970e-02 5.62919021e-01 -6.84269667e-02 2.46768564e-01 -5.00416696e-01 -2.91508824e-01 1.54303566e-01 -5.47087610e-01 -2.72333056e-01 2.11805835e-01 2.24509582e-01 5.90566993e-01 3.25246423e-01 3.10707361e-01 -5.67126870e-01 -6.24391794e-01 4.94972020e-01 1.10864115e+00 -3.11593503e-01 5.86184680e-01 -3.71207625e-01 9.33609426e-01 -2.68884599e-01 4.43659335e-01 5.83339334e-01 2.59774655e-01 2.50216722e-01 -4.86921996e-01 -1.51031077e-01 -4.56760973e-01 -1.36338830e+00 1.32979476e+00 -5.93232334e-01 1.00561023e+00 5.86042106e-01 -6.50379241e-01 1.06283545e+00 6.71105266e-01 1.87084392e-01 -1.19469337e-01 7.10073650e-01 -1.50154367e-01 -1.88472811e-02 -8.07914257e-01 1.25326917e-01 7.99717903e-02 4.61033225e-01 3.72723818e-01 1.76417992e-01 -2.63145983e-01 3.42962109e-02 -2.07957566e-01 4.82691258e-01 5.03515065e-01 4.04268563e-01 -2.25095019e-01 1.03832006e+00 -4.32128698e-01 3.18783015e-01 1.50805876e-01 -1.32321119e-01 2.84765899e-01 4.31473464e-01 -4.63166833e-01 -9.62494493e-01 -5.25185943e-01 1.42744198e-01 1.16881669e+00 -1.69418678e-01 4.25896570e-02 -1.16730392e+00 -1.97114423e-01 -1.37852475e-01 4.07641351e-01 -8.12825203e-01 1.23786226e-01 -7.48778462e-01 2.94019431e-02 2.25186214e-01 3.37450027e-01 4.20530975e-01 -1.53540146e+00 -7.66530812e-01 6.98810583e-03 2.66092062e-01 -9.29136634e-01 -5.72784424e-01 -5.27322777e-02 -5.60785592e-01 -8.28846514e-01 -4.05252546e-01 -9.19135273e-01 9.32772398e-01 -2.61921585e-01 8.54560733e-01 1.91844940e-01 -2.57767081e-01 4.71957594e-01 -6.29345104e-02 -6.01755083e-01 -6.91505432e-01 -3.32686782e-01 2.97732502e-01 6.18108034e-01 4.81755622e-02 -5.82444489e-01 -3.96111131e-01 1.08281329e-01 -6.41411185e-01 1.38592944e-01 1.78393871e-01 7.41303742e-01 4.37424988e-01 -3.22576761e-01 4.94014293e-01 -8.57844830e-01 7.14083016e-01 -1.59517571e-01 -9.14045691e-01 5.13657890e-02 -4.06734377e-01 1.92117393e-01 8.90959263e-01 -8.49074483e-01 -1.27120245e+00 6.21236324e-01 1.81608088e-02 -8.25964153e-01 -3.50923508e-01 -6.91007897e-02 -3.30498308e-01 -4.99735564e-01 4.98849481e-01 -5.09481728e-02 6.08640909e-01 -2.42519826e-01 3.91555071e-01 8.17156792e-01 6.86660290e-01 -3.69360209e-01 9.15478110e-01 3.44507366e-01 1.58812657e-01 -1.12263429e+00 -6.48204833e-02 2.34099939e-01 -9.30409372e-01 -1.01164865e+00 9.10484314e-01 -7.49295473e-01 -1.17115748e+00 8.18971813e-01 -1.38392246e+00 -6.94210827e-02 1.53322563e-01 1.75585732e-01 -4.98989314e-01 8.59887749e-02 -5.89895606e-01 -9.05729353e-01 -6.06584489e-01 -1.40126014e+00 1.11125410e+00 5.15751600e-01 -4.22299683e-01 -9.30246592e-01 -2.26851121e-01 1.76552966e-01 3.61993402e-01 8.55977952e-01 6.43831730e-01 -2.64370859e-01 -3.93788278e-01 -3.02839369e-01 3.51051271e-01 2.93540746e-01 2.64500320e-01 7.15551734e-01 -1.07839549e+00 -3.42957795e-01 1.47355199e-01 -3.65486413e-01 1.75990239e-01 2.27418095e-01 9.14008677e-01 -5.95567346e-01 -8.09108019e-02 5.03557563e-01 1.25455213e+00 5.20873368e-01 4.60722476e-01 1.38572395e-01 7.62339771e-01 7.61840105e-01 3.36479604e-01 4.18456763e-01 -1.10352769e-01 1.02793515e+00 5.63862562e-01 -2.24976212e-01 -1.94855407e-02 -2.38799006e-02 5.46858132e-01 5.80690026e-01 -4.74784702e-01 -1.18409850e-01 -3.41033787e-01 -4.22026254e-02 -1.51521838e+00 -1.04393029e+00 -6.47092517e-03 2.03960943e+00 8.27978909e-01 -1.16658866e-01 1.06302969e-01 3.14313442e-01 1.00634480e+00 -5.67047112e-02 -6.73315763e-01 -9.34146702e-01 3.24165583e-01 4.00242686e-01 -7.62527296e-03 6.63407326e-01 -9.48662698e-01 7.29492307e-01 5.96867895e+00 4.95771050e-01 -1.70293939e+00 -7.59707391e-02 4.54310685e-01 -4.97336462e-02 6.16800524e-02 -5.39597571e-01 -5.92365265e-01 4.26685125e-01 9.52774167e-01 -2.45190854e-03 6.98349833e-01 1.08765936e+00 6.78396821e-01 -2.26852242e-02 -1.29415143e+00 9.93296027e-01 4.75494385e-01 -1.04584479e+00 1.26087219e-01 -2.90518582e-01 3.74798030e-01 -5.69282532e-01 1.61818102e-01 -6.84218407e-02 -2.79084563e-01 -1.06891537e+00 8.98619056e-01 6.67894006e-01 1.13192105e+00 -7.87172616e-01 3.45779359e-01 1.00953735e-01 -1.16916203e+00 -8.41005519e-02 9.72052813e-02 -5.24413325e-02 8.85642916e-02 -2.11966008e-01 -7.04870582e-01 6.36391267e-02 4.39051181e-01 3.86972666e-01 -3.79561394e-01 3.92489910e-01 -3.33465576e-01 3.32069129e-01 -3.93716156e-01 -1.62757367e-01 -1.57890648e-01 -3.09069127e-01 6.50174499e-01 1.03756106e+00 1.63219586e-01 4.00814526e-02 -9.60399359e-02 7.98626542e-01 3.80618945e-02 1.53436765e-01 -7.73067892e-01 9.19175427e-03 4.33144897e-01 1.58300591e+00 -5.64020157e-01 -2.29377925e-01 -2.10604876e-01 9.56762910e-01 7.21314475e-02 1.31534293e-01 -9.27676678e-01 -3.23280543e-01 6.61365032e-01 2.37177283e-01 3.28604549e-01 -5.29406182e-02 2.24642664e-01 -7.54287839e-01 -1.72252938e-01 -1.06932771e+00 -1.19444162e-01 -1.17212319e+00 -6.29388034e-01 1.14075410e+00 2.51259774e-01 -1.29185545e+00 -7.99344480e-01 -7.58492470e-01 -7.84373343e-01 7.13779986e-01 -1.17282379e+00 -1.20609295e+00 -6.15925252e-01 7.16982484e-01 7.08679080e-01 -5.17724097e-01 8.74322832e-01 2.46357486e-01 -6.42867208e-01 6.18668497e-01 -4.27476972e-01 -2.74737738e-03 6.61287069e-01 -1.05110443e+00 -5.64977303e-02 5.86857855e-01 -1.71053529e-01 5.55985630e-01 1.06188142e+00 -3.90318096e-01 -1.68854952e+00 -8.31325173e-01 4.99836951e-01 -2.89408535e-01 2.27143437e-01 -3.63922894e-01 -7.21872568e-01 6.21035099e-01 5.83206534e-01 1.60595716e-03 2.21240386e-01 -7.45053113e-01 1.31584078e-01 -3.64250124e-01 -1.42992198e+00 5.94833672e-01 5.35467505e-01 -3.00707042e-01 -2.84144640e-01 8.38340074e-02 3.25461894e-01 -5.23045421e-01 -6.20636404e-01 6.38172477e-02 7.82614291e-01 -9.21403527e-01 6.99327946e-01 -3.93246949e-01 3.16653341e-01 -4.70511585e-01 4.41749960e-01 -1.19625688e+00 -6.72680214e-02 -1.05693972e+00 -1.27171263e-01 1.33512187e+00 5.47595806e-02 -4.08120036e-01 5.11260033e-01 7.66314983e-01 3.56852651e-01 -4.76638228e-01 -9.12581146e-01 -2.51865298e-01 -4.82702404e-01 -9.50874761e-02 3.65574807e-01 7.22631156e-01 -1.66146263e-01 4.69986320e-01 -8.36334169e-01 2.40849212e-01 6.36667132e-01 -2.70116299e-01 8.42370510e-01 -1.36243188e+00 -1.79931402e-01 5.38281724e-03 -6.30268276e-01 -3.99392694e-01 4.20163512e-01 -5.45295119e-01 -1.04180528e-02 -7.59641051e-01 -4.09907192e-01 1.92704737e-01 3.49386394e-01 6.99422657e-01 1.61671087e-01 2.18219906e-01 4.94205624e-01 2.96714948e-03 2.24436298e-01 4.88865167e-01 1.35056543e+00 8.73777047e-02 -3.42921555e-01 -7.45920911e-02 -1.84668317e-01 8.97842824e-01 6.81810975e-01 -2.49638334e-01 -6.38530195e-01 -2.43174806e-01 -2.70916760e-01 4.34221983e-01 2.65958607e-01 -1.09285986e+00 2.59640962e-01 -3.54006328e-02 7.11320400e-01 -1.73439100e-01 6.08819306e-01 -1.17219019e+00 6.38839602e-01 3.48016560e-01 -3.25376093e-01 3.78061861e-01 2.95167148e-01 3.94831926e-01 -5.61767332e-02 -1.00633197e-01 1.06286895e+00 -2.05923304e-01 -5.85592091e-01 1.22486413e-01 -5.36553264e-01 -4.38650280e-01 1.18072963e+00 -3.16001534e-01 5.14521077e-02 -8.66947651e-01 -1.02880108e+00 -1.83404088e-01 2.16884434e-01 5.81166387e-01 5.72524071e-01 -1.31770587e+00 -4.90932226e-01 7.01395988e-01 -3.49155784e-01 -3.77533853e-01 -3.77235957e-03 4.74580318e-01 -7.93828607e-01 4.86546569e-02 -8.64908040e-01 -6.37745798e-01 -1.86352682e+00 4.31150615e-01 4.86036807e-01 3.66770655e-01 -4.34796065e-01 5.63604116e-01 -1.59664229e-01 -1.60407200e-01 4.00528640e-01 -1.19472869e-01 -8.10228944e-01 2.30038688e-01 6.64470255e-01 2.56943554e-01 -2.28011325e-01 -1.08913541e+00 -2.02396572e-01 8.81884515e-01 1.91040426e-01 -1.80232376e-01 1.42975295e+00 1.63708944e-02 -1.64995044e-01 3.55987132e-01 1.16707075e+00 -1.92486435e-01 -1.49374974e+00 4.07417566e-01 -7.00616181e-01 -4.12294686e-01 1.97878536e-02 -6.17946029e-01 -1.53533101e+00 7.21507013e-01 1.04430413e+00 -1.94845080e-01 1.32705355e+00 -5.01040995e-01 2.00580508e-01 2.92521119e-01 3.39419007e-01 -1.33817279e+00 8.40109289e-02 -6.60478696e-02 1.47770798e+00 -1.15151191e+00 -2.37878915e-02 -4.53528970e-01 -7.34231353e-01 1.56485713e+00 8.72839510e-01 -1.67010635e-01 6.93088472e-01 5.14643192e-01 3.76113534e-01 -1.67087972e-01 -7.50467479e-01 2.52266109e-01 2.02914849e-01 8.46279025e-01 3.26798588e-01 -3.01107645e-01 -1.97106764e-01 1.39382273e-01 -1.52773187e-01 3.60908657e-01 5.85413873e-01 6.94203019e-01 -2.31500968e-01 -9.35510576e-01 -6.73003852e-01 4.87910360e-02 -2.95593530e-01 3.72234613e-01 -3.42216492e-01 1.01677322e+00 2.57528603e-01 8.45179617e-01 1.72669832e-02 -2.81765044e-01 2.05961347e-01 2.20399722e-02 5.61883628e-01 -3.07331085e-01 -7.25463867e-01 1.27542391e-01 2.31417865e-02 -5.65398037e-01 -6.74074650e-01 -5.68814278e-01 -1.25339353e+00 -1.90267146e-01 -1.88856095e-01 -1.52634636e-01 8.03790033e-01 7.94022143e-01 2.86007404e-01 4.03269827e-01 7.93950021e-01 -1.65386355e+00 -1.72347367e-01 -9.66548145e-01 -3.71095628e-01 5.09390354e-01 4.02770638e-01 -9.49315548e-01 -4.92568344e-01 6.28206611e-01]
[13.046059608459473, -0.32509931921958923]
35640eba-00c5-49c9-a1ec-c7ab2686cd9c
dd-cisenet-dual-domain-cross-iteration
2305.00088
null
https://arxiv.org/abs/2305.00088v1
https://arxiv.org/pdf/2305.00088v1.pdf
DD-CISENet: Dual-Domain Cross-Iteration Squeeze and Excitation Network for Accelerated MRI Reconstruction
Magnetic resonance imaging (MRI) is widely employed for diagnostic tests in neurology. However, the utility of MRI is largely limited by its long acquisition time. Acquiring fewer k-space data in a sparse manner is a potential solution to reducing the acquisition time, but it can lead to severe aliasing reconstruction artifacts. In this paper, we present a novel Dual-Domain Cross-Iteration Squeeze and Excitation Network (DD-CISENet) for accelerated sparse MRI reconstruction. The information of k-spaces and MRI images can be iteratively fused and maintained using the Cross-Iteration Residual connection (CIR) structures. This study included 720 multi-coil brain MRI cases adopted from the open-source fastMRI Dataset. Results showed that the average reconstruction error by DD-CISENet was 2.28 $\pm$ 0.57%, which outperformed existing deep learning methods including image-domain prediction (6.03 $\pm$ 1.31, p < 0.001), k-space synthesis (6.12 $\pm$ 1.66, p < 0.001), and dual-domain feature fusion approaches (4.05 $\pm$ 0.88, p < 0.001).
['Gerardo Hermosillo Valadez', 'Zhigang Peng', 'Xiongchao Chen']
2023-04-28
null
null
null
null
['mri-reconstruction']
['computer-vision']
[ 2.19688267e-01 4.97357324e-02 2.15795636e-01 -2.67939299e-01 -1.08417165e+00 1.83867048e-02 1.74347639e-01 -2.36610830e-01 -3.65875661e-01 8.40180099e-01 3.42620879e-01 -6.66014031e-02 -5.91549337e-01 -3.16314697e-01 -5.36730945e-01 -9.85498071e-01 -5.27713954e-01 3.20299327e-01 -8.99279192e-02 1.02685332e-01 9.80319381e-02 5.07901013e-01 -7.75223732e-01 3.31724733e-01 1.09536040e+00 9.35105860e-01 7.12151051e-01 7.82667473e-02 1.71746865e-01 7.11625695e-01 -2.47345999e-01 -2.91243568e-02 1.53254732e-01 -4.68814641e-01 -7.84698009e-01 -1.68398783e-01 -1.35746859e-02 -4.31066453e-01 -7.43344486e-01 1.19684923e+00 7.93301523e-01 1.77696019e-01 4.83041853e-01 -6.29017532e-01 -3.60373676e-01 8.63357723e-01 -9.08115685e-01 6.51692629e-01 -2.51520723e-01 1.89492881e-01 1.05333909e-01 -8.94492030e-01 6.04658246e-01 5.39816976e-01 8.29785526e-01 3.77965122e-01 -1.39425516e+00 -1.09783626e+00 -2.00479805e-01 3.66379946e-01 -1.32579482e+00 -3.68200332e-01 8.67619336e-01 -4.95983779e-01 1.08372688e+00 1.05881497e-01 8.57309401e-01 7.43692160e-01 6.83468282e-01 4.17793393e-01 1.51080930e+00 -9.99228731e-02 1.01314746e-02 -3.82026017e-01 -4.26550716e-04 4.89178181e-01 9.28001255e-02 2.48615518e-01 -3.90519351e-01 -2.30229944e-01 1.19468617e+00 2.81766290e-03 -4.54914480e-01 4.16552648e-03 -1.58932841e+00 6.81291223e-01 5.92883885e-01 6.20450318e-01 -6.57861352e-01 -1.49578512e-01 4.39446718e-01 8.72788206e-02 4.23163086e-01 4.97581363e-01 -2.24746153e-01 -9.28879529e-02 -9.80689049e-01 8.87477100e-02 6.67295232e-02 3.24436843e-01 1.57796368e-01 4.40697342e-01 -4.60925326e-02 1.04271197e+00 1.77972894e-02 4.75097805e-01 9.66723561e-01 -1.05011582e+00 1.23128980e-01 -7.77507275e-02 -3.62653166e-01 -9.49308336e-01 -8.34612906e-01 -7.91437685e-01 -1.49236429e+00 -1.45919144e-01 1.10898241e-01 -2.88516760e-01 -7.80886948e-01 1.69701934e+00 1.18922912e-01 5.13326228e-01 -1.10785574e-01 1.20279872e+00 9.57660139e-01 2.97330350e-01 -1.06929801e-01 -6.47376299e-01 1.20662606e+00 -7.54639387e-01 -9.58221197e-01 -4.83744554e-02 4.56243187e-01 -7.71567464e-01 5.88224471e-01 5.43628156e-01 -1.33928847e+00 -3.57440472e-01 -7.25215852e-01 3.54493082e-01 4.47287858e-01 1.19568147e-01 7.28716493e-01 3.03165019e-01 -1.04777431e+00 6.07928038e-01 -1.10859323e+00 4.18098599e-01 6.01492047e-01 5.98066032e-01 -4.48844016e-01 -2.92325228e-01 -1.21919310e+00 1.04517615e+00 1.23383887e-01 1.25103787e-01 -8.37268889e-01 -1.36927152e+00 -5.93388677e-01 -3.91460955e-01 -1.11204147e-01 -4.82635140e-01 9.58893239e-01 -4.48238939e-01 -1.32862318e+00 8.06150258e-01 -6.97659515e-03 -5.10663211e-01 2.57921815e-01 -4.37445343e-02 -7.35404253e-01 7.24385679e-01 3.50395113e-01 6.63575351e-01 8.16243529e-01 -9.19844389e-01 -1.18175328e-01 -5.61027408e-01 -5.33067465e-01 3.31896156e-01 1.38970882e-01 9.02196169e-02 -2.97680646e-02 -1.01979649e+00 6.86938226e-01 -1.04855251e+00 -6.79985404e-01 -2.20766813e-01 -9.76490602e-02 3.38580012e-01 3.82010013e-01 -1.41914296e+00 8.92077386e-01 -2.04563546e+00 4.66652289e-02 2.92333633e-01 7.16315985e-01 1.90960869e-01 8.68571922e-03 -2.53900588e-01 -6.35446846e-01 -3.01900387e-01 -5.06017804e-01 9.98192877e-02 -8.29720140e-01 -2.40491942e-01 1.98248327e-01 7.99681306e-01 -2.64131039e-01 9.34439659e-01 -7.66432405e-01 -2.86472678e-01 3.12367201e-01 7.74711847e-01 -5.66422999e-01 -4.42795735e-03 7.72611856e-01 1.15140057e+00 -2.14718804e-01 3.62969786e-01 8.74696195e-01 -4.08931643e-01 2.97060519e-01 -7.31149733e-01 -1.10573329e-01 8.16550106e-02 -9.55677629e-01 2.15855789e+00 -2.06299499e-01 6.04626000e-01 2.29596585e-01 -1.35641229e+00 7.71186590e-01 5.89771926e-01 1.36922932e+00 -1.11187530e+00 1.59077719e-01 4.91915107e-01 6.32568598e-01 -3.85397226e-01 -6.85949177e-02 -6.23960972e-01 3.07423949e-01 3.92265499e-01 7.55402073e-02 -1.53222367e-01 -3.01671237e-01 2.26818115e-01 1.19782054e+00 -3.32123458e-01 -1.15372483e-02 -6.86800778e-01 2.56401718e-01 -1.19526826e-01 6.41385317e-01 6.36688590e-01 -3.09877694e-01 6.97010577e-01 1.19879559e-01 -4.57581460e-01 -9.81229246e-01 -1.13222075e+00 -6.94522440e-01 2.46153459e-01 9.36350133e-03 7.15162009e-02 -7.38176048e-01 -1.82794020e-01 -2.21029639e-01 3.65244716e-01 -2.31244192e-01 -2.54111558e-01 -9.29533124e-01 -1.05040169e+00 4.28390861e-01 4.01173234e-01 5.79493403e-01 -8.74140978e-01 -5.39059699e-01 4.96136993e-01 -5.74011862e-01 -1.06798756e+00 -4.62071061e-01 2.13970974e-01 -1.45960927e+00 -6.14185154e-01 -1.15643406e+00 -6.89804316e-01 7.76319265e-01 3.76821399e-01 7.92482257e-01 -2.42940962e-01 -6.03483438e-01 1.38280973e-01 -1.85564712e-01 6.71222955e-02 5.71068935e-02 -2.95982003e-01 3.40531915e-01 -2.83798814e-01 1.13444678e-01 -8.43217075e-01 -9.24206614e-01 5.81032969e-02 -8.05205703e-01 2.78673947e-01 7.89147735e-01 1.21362269e+00 9.72098053e-01 6.50553927e-02 7.45958745e-01 -5.47992408e-01 5.10618091e-01 -5.31084239e-01 -2.88271517e-01 -1.28829777e-01 -5.27825832e-01 4.09019254e-02 2.64130443e-01 -6.07029498e-01 -1.08128309e+00 -1.11500509e-01 -1.49865791e-01 -6.71110570e-01 9.76012740e-03 7.03541577e-01 3.46802831e-01 -4.75073367e-01 5.08025765e-01 4.67770934e-01 3.55036050e-01 -5.20377219e-01 -2.89718919e-02 2.55969197e-01 7.81448424e-01 -3.44171017e-01 3.44457418e-01 4.81630236e-01 7.31205102e-03 -8.23571801e-01 -3.43539536e-01 -1.91841513e-01 -5.35730481e-01 -2.37259120e-01 6.82358623e-01 -1.08720374e+00 -4.47019279e-01 3.83856505e-01 -7.06254125e-01 -2.00540349e-01 -2.49694716e-02 1.22589087e+00 -4.03114259e-01 5.13831019e-01 -8.42210948e-01 -1.65781721e-01 -8.16982925e-01 -1.59353030e+00 4.40018058e-01 9.66246182e-04 -2.38295570e-01 -6.21299148e-01 -1.90517217e-01 6.42879784e-01 6.56843424e-01 1.66611806e-01 8.00002158e-01 -2.75840223e-01 -3.34781945e-01 7.60319307e-02 -2.89970279e-01 3.35684448e-01 1.82851180e-01 -9.56091523e-01 -5.38994551e-01 -4.22709823e-01 7.18960285e-01 -7.65661672e-02 4.97733027e-01 1.14387310e+00 1.62652397e+00 -2.85835303e-02 -2.16189906e-01 9.33381438e-01 1.34736013e+00 6.96498454e-01 8.01129401e-01 3.45601022e-01 5.77839136e-01 3.45665634e-01 1.51640281e-01 4.33891535e-01 1.38190195e-01 5.71331203e-01 8.46640691e-02 -1.23700455e-01 -3.90819490e-01 2.04954728e-01 -7.09923953e-02 1.30268228e+00 -4.84673567e-02 7.23672032e-01 -1.13587618e+00 7.78925836e-01 -1.42430055e+00 -7.24697173e-01 -3.51122230e-01 1.85077190e+00 1.01565671e+00 -2.29497790e-01 -1.01070657e-01 6.84800074e-02 7.05313623e-01 -2.49136686e-01 -8.07204485e-01 9.79776680e-02 -1.24578513e-01 5.87476850e-01 4.67629880e-01 2.66079575e-01 -6.55384958e-01 3.01155061e-01 5.75983429e+00 9.34023738e-01 -1.43431675e+00 5.04665852e-01 8.14407945e-01 -4.12571609e-01 -2.08631635e-01 -3.01985323e-01 -1.89504012e-01 7.54444420e-01 9.45082784e-01 -2.35577464e-01 5.92122257e-01 4.19038057e-01 2.46530339e-01 -3.84441435e-01 -5.02743006e-01 1.59906983e+00 1.54327095e-01 -1.48736608e+00 -4.05527264e-01 1.85199797e-01 7.97459424e-01 4.39767271e-01 1.78440034e-01 -6.44668788e-02 -2.61531413e-01 -1.15451109e+00 3.40389550e-01 4.98292476e-01 1.21688187e+00 -8.07739675e-01 6.13025844e-01 9.12147909e-02 -9.56390500e-01 1.65812552e-01 -6.63671717e-02 4.78800014e-02 5.06475508e-01 1.08636200e+00 -5.61607957e-01 6.27909005e-01 9.76198733e-01 5.89941561e-01 -3.17321122e-02 1.06715059e+00 1.72938049e-01 5.29053211e-01 -2.40233615e-01 6.65169895e-01 1.69658571e-01 -3.48843604e-01 6.96346998e-01 8.60416234e-01 5.10532677e-01 7.29963839e-01 -7.58633763e-02 6.73516631e-01 2.55218983e-01 -1.88946187e-01 -3.30095559e-01 3.50612611e-01 2.42249206e-01 1.19572377e+00 -7.25278437e-01 -3.10655743e-01 -4.04300600e-01 7.42661119e-01 1.21682771e-02 2.81153947e-01 -8.57782900e-01 -3.95058393e-01 3.99675727e-01 2.43659735e-01 -6.52476549e-02 -2.49257341e-01 -6.30785942e-01 -1.35517180e+00 -1.39999567e-02 -9.76991296e-01 2.05578282e-01 -7.68126249e-01 -1.28719914e+00 8.19788635e-01 1.00493826e-01 -1.15347493e+00 -1.40654799e-02 -3.01976144e-01 -1.68849379e-01 1.13787067e+00 -1.22628474e+00 -6.84998214e-01 -1.13538884e-01 7.97044933e-01 1.91030517e-01 -3.55131716e-01 7.91163921e-01 6.70295954e-01 -3.56428832e-01 3.85491282e-01 2.28907466e-01 4.09189649e-02 3.92900735e-01 -5.51381767e-01 -2.40039617e-01 6.68064773e-01 -3.53863180e-01 7.54911363e-01 4.02609140e-01 -7.57606149e-01 -1.29909503e+00 -7.24080920e-01 5.49691677e-01 1.55409575e-01 6.15813613e-01 3.63020569e-01 -1.07480204e+00 5.24014533e-01 2.10725367e-01 3.12072635e-01 9.02574956e-01 -4.21083301e-01 2.33285695e-01 -1.94393009e-01 -1.61531162e+00 3.88972133e-01 9.59550261e-01 -4.58830893e-01 -6.42956436e-01 1.42596364e-01 4.79657412e-01 -7.20910549e-01 -1.46810067e+00 7.18593061e-01 5.42059541e-01 -6.42304778e-01 1.20691276e+00 -1.58587992e-01 3.97768945e-01 -1.18824303e-01 5.27949035e-02 -1.38485885e+00 -7.87369967e-01 -3.83169323e-01 -1.87116623e-01 4.56390381e-01 -5.06902263e-02 -6.54471457e-01 7.16076374e-01 7.58651435e-01 -6.00449443e-01 -8.81223738e-01 -1.45608687e+00 -5.68824828e-01 2.61465251e-01 -3.56312841e-01 3.15793782e-01 1.22020590e+00 1.23564519e-01 -1.94465127e-02 -3.76295954e-01 4.77848947e-02 1.09975445e+00 -7.35353604e-02 -2.73821384e-01 -8.54984581e-01 -2.04017520e-01 -4.79821622e-01 -2.25456953e-01 -7.02181041e-01 -1.21524008e-02 -1.19129992e+00 -3.42797816e-01 -1.44516277e+00 4.27322775e-01 -7.23777115e-01 -5.72598398e-01 2.49493092e-01 2.47636400e-02 2.02044949e-01 3.19325142e-02 4.20626551e-01 2.32656430e-02 5.63434303e-01 1.64611137e+00 -1.24179929e-01 8.84437747e-03 -4.40316886e-01 -7.69472718e-01 5.20563006e-01 1.21816564e+00 -4.88521069e-01 -4.79111940e-01 -7.43252397e-01 -3.31586152e-01 5.81880450e-01 3.35121363e-01 -1.12801468e+00 3.22540283e-01 7.13500082e-02 6.38987660e-01 -6.97095096e-01 3.50072414e-01 -5.98076642e-01 6.51960969e-01 8.53534639e-01 -1.42743945e-01 1.72960371e-01 4.26243931e-01 2.50859745e-02 -2.26439252e-01 -1.18981227e-01 1.06390166e+00 -4.10968304e-01 -4.99326229e-01 3.51805657e-01 -4.96984869e-01 -1.01135045e-01 8.67026329e-01 -2.05711156e-01 2.17841491e-01 7.44987559e-03 -1.03912580e+00 -1.50575578e-01 -2.10542068e-01 2.84408648e-02 1.06430769e+00 -1.51795828e+00 -6.69362128e-01 3.45446914e-01 -4.36305225e-01 -9.51321982e-03 1.21593559e+00 1.57336962e+00 -3.85848284e-01 5.59045196e-01 -5.25049746e-01 -8.05167079e-01 -8.86762321e-01 1.55873343e-01 5.35972595e-01 -3.63460183e-01 -1.22768176e+00 1.02414107e+00 8.89327973e-02 -2.91637331e-01 -1.01184256e-01 -1.93161428e-01 8.26460961e-03 -2.01577231e-01 6.83600128e-01 4.62251991e-01 4.07922775e-01 -6.87865853e-01 -5.16058087e-01 5.69529593e-01 -3.93648952e-01 -3.18747193e-01 1.68095756e+00 -9.58010778e-02 -4.28803533e-01 -6.41511604e-02 1.30340576e+00 -4.48448420e-01 -1.13721776e+00 -3.51687551e-01 -3.91459346e-01 -4.13019538e-01 5.89573920e-01 -9.41121638e-01 -1.51075244e+00 7.59122550e-01 1.03669786e+00 -4.86405671e-01 1.27720404e+00 -6.11685365e-02 1.02608144e+00 -1.65540755e-01 7.28307903e-01 -9.18970644e-01 -3.91574204e-02 3.12945962e-01 1.04577494e+00 -9.07327056e-01 4.57514040e-02 -1.31593570e-01 -7.72388041e-01 7.92664468e-01 5.52888691e-01 -2.36080930e-01 1.00939071e+00 3.13551426e-01 -2.27434158e-01 -3.76493007e-01 -3.10429603e-01 4.97331381e-01 2.35769168e-01 7.43060291e-01 6.60849094e-01 3.20698082e-01 -6.25241220e-01 8.53057265e-01 -2.03917608e-01 1.88222244e-01 2.69102126e-01 7.93198049e-01 2.10061595e-02 -7.87132025e-01 -5.24388492e-01 1.00545990e+00 -6.39387667e-01 -2.84434944e-01 5.91136754e-01 2.30810717e-01 1.35407031e-01 6.50148690e-01 -1.02401100e-01 -2.41769031e-01 6.10287562e-02 -2.32912764e-01 6.87576294e-01 -2.84669399e-01 -5.80618143e-01 4.78250891e-01 -2.03951597e-01 -5.46977103e-01 -5.72839797e-01 -8.42480004e-01 -1.54181445e+00 -2.47461617e-01 4.37515639e-02 1.86956897e-01 5.63430429e-01 7.87223577e-01 6.23266041e-01 8.89697194e-01 3.99759859e-01 -8.13486397e-01 -2.69697189e-01 -1.01056647e+00 -7.77659416e-01 1.83995917e-01 7.54211843e-02 -8.19762230e-01 -1.43346697e-01 -7.60538429e-02]
[13.598021507263184, -2.4191553592681885]
53b520a0-beec-426a-a62c-141cc12997b0
is-the-elephant-flying-resolving-ambiguities
2211.12503
null
https://arxiv.org/abs/2211.12503v1
https://arxiv.org/pdf/2211.12503v1.pdf
Is the Elephant Flying? Resolving Ambiguities in Text-to-Image Generative Models
Natural language often contains ambiguities that can lead to misinterpretation and miscommunication. While humans can handle ambiguities effectively by asking clarifying questions and/or relying on contextual cues and common-sense knowledge, resolving ambiguities can be notoriously hard for machines. In this work, we study ambiguities that arise in text-to-image generative models. We curate a benchmark dataset covering different types of ambiguities that occur in these systems. We then propose a framework to mitigate ambiguities in the prompts given to the systems by soliciting clarifications from the user. Through automatic and human evaluations, we show the effectiveness of our framework in generating more faithful images aligned with human intention in the presence of ambiguities.
['Rahul Gupta', 'Aram Galstyan', 'Richard Zemel', 'Kai-Wei Chang', 'Qian Hu', 'Varun Kumar', 'Jwala Dhamala', 'Apurv Verma', 'Palash Goyal', 'Ninareh Mehrabi']
2022-11-17
null
null
null
null
['common-sense-reasoning']
['reasoning']
[ 7.33807385e-01 5.04802227e-01 3.45024437e-01 -6.29119277e-01 -9.77830470e-01 -1.01471448e+00 9.02903378e-01 2.08451197e-01 -3.24004710e-01 8.89301240e-01 3.90700549e-01 -4.63640124e-01 -1.09822661e-01 -4.92437810e-01 -6.11437976e-01 -1.13436036e-01 6.99404359e-01 6.95492268e-01 1.67125478e-01 -3.53418678e-01 4.10222173e-01 4.05215807e-02 -1.43432748e+00 5.32014012e-01 1.22232008e+00 5.75826287e-01 5.67463040e-01 5.94323337e-01 -3.01467568e-01 8.67737830e-01 -1.07037795e+00 -5.05508304e-01 6.55190572e-02 -5.66414535e-01 -1.14766359e+00 5.64450741e-01 6.23787642e-01 -1.81811363e-01 3.92530143e-01 1.26680756e+00 6.51806071e-02 -4.56497371e-02 6.71385825e-01 -1.35134459e+00 -7.67044246e-01 3.38356972e-01 -1.25165403e-01 1.19165108e-01 8.14206719e-01 2.03418434e-01 9.76274610e-01 -7.77753413e-01 6.38905466e-01 1.44310367e+00 4.52334285e-01 4.73772258e-01 -1.21277320e+00 -2.99211293e-01 2.66755342e-01 2.24604756e-01 -1.23635590e+00 -5.66183627e-01 5.36835313e-01 -4.00306374e-01 8.38958919e-01 6.20746255e-01 9.54118222e-02 1.39670169e+00 2.47068197e-01 5.77541113e-01 1.33777320e+00 -4.75946397e-01 1.64600015e-02 3.96042287e-01 2.17623323e-01 4.15063530e-01 3.09190303e-01 -3.69454809e-02 -3.01310360e-01 -2.16373011e-01 6.99780822e-01 -3.23297381e-01 -3.49952489e-01 1.26662850e-01 -1.31732726e+00 8.53135049e-01 1.68709666e-01 4.63636190e-01 -5.85560381e-01 -2.38952085e-01 -6.42244518e-02 3.12393010e-01 5.62359877e-02 9.96641874e-01 -1.94500268e-01 -8.45286921e-02 -5.59444785e-01 3.33874881e-01 9.35608029e-01 1.00283349e+00 8.46470475e-01 -2.52872974e-01 -3.80841851e-01 7.81000078e-01 2.42738083e-01 5.40945828e-01 4.27764088e-01 -1.16292405e+00 2.82025516e-01 5.38897455e-01 6.11083686e-01 -1.48764098e+00 -9.76307392e-02 -3.38023454e-01 -6.35520875e-01 -9.82259661e-02 2.47377202e-01 -1.00317545e-01 -8.48948121e-01 1.59955156e+00 9.43018645e-02 -2.15376064e-01 2.80136168e-01 1.07525051e+00 7.16734946e-01 6.15828156e-01 2.37167865e-01 -3.53929788e-01 1.63427353e+00 -6.73015118e-01 -1.17463231e+00 -7.53725708e-01 1.58708319e-01 -1.27480328e+00 1.32060564e+00 2.54679561e-01 -8.36619020e-01 -5.25157928e-01 -5.89547217e-01 -1.50865212e-01 -2.08759978e-02 5.06838597e-03 4.55972925e-02 2.14713499e-01 -1.06656837e+00 6.36958927e-02 -3.62984210e-01 -3.66525799e-01 -1.94998622e-01 1.01549461e-01 -2.63973296e-01 -2.23306775e-01 -1.28246975e+00 1.03918970e+00 4.50819761e-01 2.04336450e-01 -5.91540396e-01 -2.04319060e-01 -1.06368411e+00 -1.22319907e-01 1.06871295e+00 -9.40341651e-01 1.82817638e+00 -1.29915357e+00 -1.01398349e+00 7.91922867e-01 -5.11410773e-01 -4.53143835e-01 5.88106692e-01 -3.80130112e-01 -2.38589868e-01 1.86950043e-02 7.55633414e-01 8.67709100e-01 8.81990731e-01 -1.62198770e+00 -7.78775632e-01 -1.27618313e-01 4.02242482e-01 2.09644824e-01 2.64431387e-01 3.14176902e-02 -3.01696867e-01 -8.21032286e-01 2.61877000e-01 -1.13469970e+00 -3.51540476e-01 -4.61062193e-01 -6.14708602e-01 -3.52401525e-01 8.62016797e-01 -7.20740795e-01 1.10537398e+00 -1.97688341e+00 -4.45316136e-02 -1.00955330e-01 2.63339162e-01 2.66790509e-01 -2.23530293e-01 3.56708050e-01 6.62663803e-02 4.36582655e-01 -1.72874197e-01 -5.28227240e-02 2.06079856e-01 7.29580045e-01 -5.12796998e-01 -2.84972280e-01 4.42559868e-01 9.70461190e-01 -9.10449326e-01 -4.98641759e-01 1.01194650e-01 2.41940126e-01 -3.44963133e-01 6.29863739e-01 -6.10681474e-01 5.42429924e-01 -4.00192976e-01 4.87663507e-01 5.22612393e-01 -6.23739004e-01 1.82715997e-01 -3.19517940e-01 1.66449234e-01 6.41606808e-01 -1.05983222e+00 1.10313189e+00 -5.23560405e-01 5.54747939e-01 -3.34935598e-02 -6.74471140e-01 5.57098866e-01 3.08696598e-01 -2.35043123e-01 -7.22599447e-01 -3.81037220e-02 4.72219624e-02 -1.93936914e-01 -8.64128351e-01 7.13054836e-01 -3.18253011e-01 -3.32145035e-01 6.46023870e-01 -6.28180504e-02 -2.68121421e-01 2.15654418e-01 3.50791603e-01 5.69375992e-01 -1.50422156e-01 5.95713258e-01 -1.56904787e-01 4.06493336e-01 3.04593623e-01 4.79079276e-01 1.20462704e+00 -2.25407779e-01 5.24816573e-01 4.68609482e-01 -6.69827580e-01 -7.56972075e-01 -9.69349742e-01 4.75944310e-01 9.52868342e-01 5.33329025e-02 -6.62493825e-01 -9.29054499e-01 -7.38836825e-01 -4.30454224e-01 1.10001266e+00 -3.55128855e-01 2.67859828e-02 -3.17265451e-01 -4.23248559e-01 1.49256900e-01 3.12556237e-01 7.11029232e-01 -1.06674111e+00 -8.26600015e-01 2.54447669e-01 -9.15810764e-01 -1.82142413e+00 -7.19755828e-01 -2.80889094e-01 -4.55287188e-01 -1.12739217e+00 -2.85225481e-01 -6.22095168e-01 8.76498103e-01 6.33940160e-01 1.44971800e+00 3.38763118e-01 -1.06130220e-01 7.20694125e-01 -3.92543226e-01 -5.21227300e-01 -1.01370919e+00 -1.77863762e-01 -2.44937167e-01 -1.36469185e-01 2.14874923e-01 -2.04598904e-01 -2.91096479e-01 7.96809435e-01 -1.38054609e+00 6.02432847e-01 5.79826474e-01 6.57637358e-01 3.80395561e-01 -1.09917447e-01 3.94210398e-01 -9.89903331e-01 1.18861139e+00 -1.45914301e-01 -4.94873255e-01 3.81949872e-01 -5.40290892e-01 3.72100234e-01 2.69913137e-01 -3.46004069e-01 -1.21118557e+00 6.03188202e-03 7.88858011e-02 -4.26148903e-03 -7.16817558e-01 5.32693267e-01 -1.59599662e-01 2.25313127e-01 7.44042873e-01 1.62139267e-01 -1.23213798e-01 -1.17083229e-01 3.81409794e-01 6.19130015e-01 7.40483165e-01 -8.01620960e-01 7.90206909e-01 1.15019113e-01 -5.92078030e-01 -9.49687421e-01 -1.29664493e+00 -2.15738341e-01 -3.70105654e-01 -2.01575398e-01 8.51557374e-01 -6.95428908e-01 -2.84609646e-01 6.15949184e-02 -1.63384080e+00 2.38197908e-01 8.33609030e-02 -6.21175021e-02 -4.76012588e-01 6.54491425e-01 -1.42239079e-01 -7.71451116e-01 -1.25713885e-01 -1.24653125e+00 1.27430451e+00 2.56899714e-01 -1.01327491e+00 -8.08954477e-01 -2.45020479e-01 9.05054331e-01 5.82049251e-01 5.13579994e-02 9.83067870e-01 -7.31107116e-01 -7.10108399e-01 1.47620916e-01 -1.70391634e-01 1.09748311e-01 4.86215800e-01 -2.40599275e-01 -8.01466465e-01 9.30855498e-02 2.00607359e-01 -2.22659960e-01 3.79084349e-01 8.75212438e-03 6.41680002e-01 -1.04553545e+00 -2.84989960e-02 -2.24473000e-01 1.01140738e+00 2.27236658e-01 6.52175546e-01 1.64459199e-01 3.87437314e-01 8.61668050e-01 8.39565456e-01 1.38305515e-01 7.64412940e-01 6.85303926e-01 3.11268926e-01 -1.91324241e-02 2.65390903e-01 -3.30315828e-01 2.68073659e-02 3.79990369e-01 1.36418045e-01 -4.24477637e-01 -1.05188704e+00 6.32821798e-01 -1.96664929e+00 -7.77073503e-01 -1.61576971e-01 1.75846326e+00 1.13630319e+00 8.26311707e-02 -2.36378655e-01 -2.12212235e-01 8.09850991e-01 2.95697711e-02 -3.95253152e-02 -2.20392093e-01 -5.76628745e-02 -2.86147565e-01 -2.00920075e-01 1.00678051e+00 -9.90684688e-01 1.19225550e+00 6.40134144e+00 4.83031869e-01 -8.76464367e-01 -6.24310561e-02 6.69237554e-01 5.26562035e-01 -7.35186219e-01 3.33370656e-01 -6.05859816e-01 1.23216517e-01 5.58172643e-01 -6.67458959e-03 2.90114999e-01 4.74078745e-01 3.22118491e-01 -5.63435614e-01 -1.13401914e+00 9.04792249e-01 3.69847983e-01 -1.18185282e+00 5.36448658e-01 -2.97095686e-01 5.58309317e-01 -4.67541784e-01 -6.73775449e-02 3.23003158e-02 4.69820023e-01 -1.08357859e+00 6.64680302e-01 4.79071766e-01 1.99388564e-01 -4.54230517e-01 7.01278329e-01 7.99821138e-01 -5.46951413e-01 2.34967262e-01 -5.32959737e-02 -3.41246843e-01 2.51325846e-01 5.05439103e-01 -1.47799850e+00 3.19075227e-01 3.88182819e-01 7.37453774e-02 -7.40394711e-01 4.98502463e-01 -7.54422069e-01 2.70678639e-01 -2.60494560e-01 -1.34970337e-01 2.42228925e-01 -1.35230704e-03 6.77086055e-01 1.10601258e+00 2.35780075e-01 4.87638056e-01 3.57079625e-01 1.11536169e+00 2.68951267e-01 -1.57473028e-01 -8.45644176e-01 -1.77216381e-01 2.92661667e-01 1.07745790e+00 -5.02050161e-01 -4.63662863e-01 -1.55907050e-01 1.20200920e+00 6.82677403e-02 5.68230808e-01 -6.03673160e-01 -2.97751650e-02 3.27061802e-01 3.25778574e-01 -1.50023177e-01 -3.60935092e-01 1.04513712e-01 -1.03589070e+00 1.73436746e-01 -1.47257984e+00 4.14162755e-01 -1.31364286e+00 -1.27168703e+00 8.41657519e-01 1.60546660e-01 -8.90493453e-01 -7.17222035e-01 -4.69341040e-01 -3.12075377e-01 7.53008604e-01 -1.54518950e+00 -8.53905380e-01 -3.43097717e-01 1.76627994e-01 8.88086677e-01 3.44642222e-01 8.74193013e-01 -5.27074635e-02 -9.90210399e-02 -1.92970131e-02 -1.07474887e+00 -2.34963354e-02 8.92972350e-01 -1.07409847e+00 4.18823034e-01 8.20312917e-01 3.82334620e-01 9.52058196e-01 1.36879647e+00 -7.46239424e-01 -1.00328827e+00 -1.03580606e+00 1.40625370e+00 -4.85264659e-01 5.03015339e-01 -1.28375709e-01 -1.16083777e+00 9.21268463e-01 6.79967284e-01 -4.52314883e-01 7.69829094e-01 -2.53289074e-01 -3.71278584e-01 5.08689702e-01 -8.81383538e-01 8.27100694e-01 8.73820543e-01 -6.26791954e-01 -1.18121016e+00 5.15655816e-01 7.32649565e-01 -5.82247198e-01 -1.08771928e-01 3.67917418e-01 5.09859547e-02 -8.73921394e-01 7.14397371e-01 -6.91762567e-01 4.14234847e-01 -5.16884148e-01 -2.55111814e-01 -1.21022427e+00 -1.07513972e-01 -9.42341566e-01 5.72159410e-01 1.07927012e+00 6.00132346e-01 -4.44743186e-01 1.13834150e-01 1.00560188e+00 1.51289895e-01 9.88046899e-02 -8.65224659e-01 -4.25471574e-01 -4.52434629e-01 -4.30815786e-01 3.96196365e-01 9.77334023e-01 -3.35420519e-02 9.01026666e-01 -5.06841481e-01 4.57577169e-01 2.34184667e-01 2.97054052e-01 8.90268922e-01 -1.06242669e+00 -6.61155209e-02 -4.51906659e-02 -4.75063324e-02 -1.07413948e+00 2.02251002e-01 -3.47784668e-01 5.31030893e-01 -1.60860574e+00 1.26047924e-01 -5.40625025e-03 3.76793504e-01 6.25073135e-01 -4.56787914e-01 -8.02288428e-02 3.62655222e-01 9.98784155e-02 -9.84192729e-01 2.66802937e-01 1.21990800e+00 -2.15914488e-01 4.85926196e-02 -1.21443391e-01 -1.13018882e+00 1.03168786e+00 7.54581809e-01 -4.97556210e-01 -4.39792395e-01 -7.06456542e-01 7.02735424e-01 3.00042361e-01 9.39963222e-01 -6.93176508e-01 2.82057285e-01 -5.23231566e-01 -1.59215942e-01 -3.08707595e-01 8.48290771e-02 -8.80623519e-01 3.28033358e-01 2.07929492e-01 -5.06283462e-01 4.59926575e-01 1.88451260e-01 4.99593228e-01 -3.99419814e-01 -3.89701039e-01 3.51151258e-01 -3.68037432e-01 -6.87179565e-01 -3.07443470e-01 -8.68972898e-01 2.44016588e-01 8.17148149e-01 -1.41746495e-02 -3.70057732e-01 -9.77488220e-01 -9.18500662e-01 3.26388747e-01 4.26990330e-01 7.41173327e-01 7.87608445e-01 -1.17179680e+00 -7.96681106e-01 8.38142782e-02 2.72931486e-01 -1.84983224e-01 5.85107245e-02 4.78485912e-01 -3.88198271e-02 5.32530963e-01 -1.93813518e-02 -8.38688731e-01 -1.46254802e+00 2.23819196e-01 3.79441798e-01 -9.10045281e-02 -7.94029161e-02 4.09465522e-01 4.10380900e-01 -4.45610046e-01 2.00205483e-02 -6.63202643e-01 -3.23810101e-01 -1.40586823e-01 5.13601959e-01 -1.59327209e-01 -8.28901678e-02 -7.75917411e-01 -2.21886545e-01 4.93583232e-01 -2.13613391e-01 -3.76536995e-01 7.18625784e-01 -5.15539527e-01 -1.62201747e-01 1.83192343e-01 4.82742310e-01 4.48783115e-02 -8.58990192e-01 -5.08423209e-01 2.38143086e-01 -5.15246630e-01 -4.01940256e-01 -1.12786901e+00 -3.16766649e-01 6.84148848e-01 1.74752653e-01 4.70402807e-01 9.57691729e-01 2.88054883e-01 5.92018783e-01 8.01439106e-01 3.36818278e-01 -9.14407670e-01 3.66049320e-01 6.95594788e-01 1.30350983e+00 -1.56449962e+00 -3.46950203e-01 -8.20504069e-01 -1.02531648e+00 9.27535594e-01 7.88284779e-01 4.87342089e-01 1.43791242e-02 2.33770102e-01 7.32273459e-01 -2.08492681e-01 -8.74020398e-01 1.03811480e-01 3.21954727e-01 5.70550084e-01 4.50484931e-01 3.56130339e-02 -3.91434014e-01 8.40432465e-01 -2.24130839e-01 -1.44181997e-01 7.87070155e-01 1.08481264e+00 -6.54812336e-01 -1.10719800e+00 -7.82656908e-01 -4.73609865e-02 -3.20505857e-01 -1.68453813e-01 -9.26962316e-01 4.92691785e-01 -1.36701772e-02 1.34837592e+00 -4.34758402e-02 1.89100020e-02 4.12889242e-01 -2.05733795e-02 3.00648123e-01 -7.82454610e-01 -4.21407074e-01 2.48819560e-01 4.63463992e-01 -3.84606481e-01 -5.19782484e-01 -4.41888064e-01 -1.15804422e+00 2.21656039e-01 -2.79611081e-01 3.87773693e-01 1.75965682e-01 1.43562007e+00 5.87018669e-01 3.50225538e-01 1.72494605e-01 -3.85545164e-01 -6.51724696e-01 -9.58289325e-01 1.63824588e-01 6.89787209e-01 4.25513148e-01 -3.26647252e-01 -2.08249182e-01 5.15522540e-01]
[11.074695587158203, 1.7350010871887207]
67791eb1-b662-4631-941d-1a2b8b2aca36
weakly-supervised-online-action-detection-for
2208.03648
null
https://arxiv.org/abs/2208.03648v1
https://arxiv.org/pdf/2208.03648v1.pdf
Weakly Supervised Online Action Detection for Infant General Movements
To make the earlier medical intervention of infants' cerebral palsy (CP), early diagnosis of brain damage is critical. Although general movements assessment(GMA) has shown promising results in early CP detection, it is laborious. Most existing works take videos as input to make fidgety movements(FMs) classification for the GMA automation. Those methods require a complete observation of videos and can not localize video frames containing normal FMs. Therefore we propose a novel approach named WO-GMA to perform FMs localization in the weakly supervised online setting. Infant body keypoints are first extracted as the inputs to WO-GMA. Then WO-GMA performs local spatio-temporal extraction followed by two network branches to generate pseudo clip labels and model online actions. With the clip-level pseudo labels, the action modeling branch learns to detect FMs in an online fashion. Experimental results on a dataset with 757 videos of different infants show that WO-GMA can get state-of-the-art video-level classification and cliplevel detection results. Moreover, only the first 20% duration of the video is needed to get classification results as good as fully observed, implying a significantly shortened FMs diagnosis time. Code is available at: https://github.com/scofiedluo/WO-GMA.
['Xiaowei Ding', 'Kang Dang', 'Guangjun Yu', 'Yuan Tian', 'Siheng Chen', 'Chuncao Zhang', 'Jia Xiao', 'Tongyi Luo']
2022-08-07
null
null
null
null
['online-action-detection']
['computer-vision']
[ 5.94323575e-02 5.46157435e-02 -5.00171661e-01 -2.15422332e-01 -8.94003153e-01 -2.81508714e-01 1.72284320e-01 9.66218635e-02 -3.96112055e-01 1.91752777e-01 1.92033127e-01 6.06353693e-02 -2.32640188e-02 -4.10679132e-01 -1.05381382e+00 -7.55802929e-01 -4.78256404e-01 1.46554098e-01 7.10630238e-01 2.19704032e-01 5.96728474e-02 3.62000346e-01 -1.62903798e+00 8.62224460e-01 7.08350897e-01 8.51699412e-01 4.70991462e-01 9.70851541e-01 1.02643661e-01 1.06319177e+00 -3.80201131e-01 -8.18773881e-02 2.00117856e-01 -3.82780075e-01 -7.12593019e-01 -4.38279435e-02 5.99931180e-01 -7.70630658e-01 -5.65072119e-01 1.00913048e+00 8.12142491e-01 6.33633509e-02 5.65029383e-01 -1.39944553e+00 -6.73346967e-02 6.16379559e-01 -6.60875142e-01 5.78193665e-01 7.26360559e-01 1.45677567e-01 5.61557472e-01 -1.01255596e+00 6.35579169e-01 9.42153037e-01 4.35195625e-01 7.79402018e-01 -7.35622644e-01 -7.10894227e-01 3.72245193e-01 8.66024792e-01 -1.17579520e+00 -5.69726050e-01 6.88688755e-01 -6.94864750e-01 9.45172012e-01 -7.05840737e-02 1.03586650e+00 1.15573025e+00 -2.51420856e-01 1.27114975e+00 5.25089025e-01 -5.13473272e-01 1.34360760e-01 -7.21850038e-01 -9.68747959e-02 8.64279687e-01 5.98967932e-02 -4.35850285e-02 -1.00275910e+00 2.94163764e-01 4.83691216e-01 6.64539039e-02 -5.04189909e-01 -3.67576897e-01 -1.16945958e+00 5.43265045e-01 1.00237839e-01 3.53530169e-01 -5.55171669e-01 6.08833171e-02 4.99706507e-01 3.36095005e-01 3.91078442e-01 -2.35316724e-01 -3.60983998e-01 -4.96664286e-01 -1.13189220e+00 4.41509895e-02 3.20349157e-01 1.11202979e+00 1.37993202e-01 -2.09830850e-01 -1.99541181e-01 7.58620739e-01 -4.42976095e-02 3.51463437e-01 5.74887633e-01 -1.01350868e+00 8.49230886e-01 3.45910728e-01 -1.21793859e-01 -8.37831616e-01 -8.13282013e-01 -1.28840163e-01 -3.24503690e-01 4.03665215e-01 6.94170892e-01 -2.81040311e-01 -9.31535959e-01 1.67510712e+00 5.82186401e-01 3.07024002e-01 -1.74114570e-01 9.15856481e-01 9.07689393e-01 3.48338872e-01 -9.03240964e-02 -3.96262735e-01 1.18642509e+00 -1.32827616e+00 -5.50966024e-01 -1.45285040e-01 9.02284980e-01 -4.88759607e-01 6.15403831e-01 6.87372684e-01 -1.30019259e+00 -3.48012596e-01 -8.04058552e-01 2.24264994e-01 -8.65516961e-02 3.89033943e-01 3.62572461e-01 2.37016991e-01 -1.19201553e+00 4.89292920e-01 -1.36159337e+00 -2.25401238e-01 7.38111854e-01 3.13858151e-01 -7.68164158e-01 -2.91853517e-01 -7.99669623e-01 7.85315990e-01 6.59349501e-01 -5.72642835e-04 -1.08656144e+00 -5.69481015e-01 -8.35587323e-01 -2.09184036e-01 5.78168035e-01 -2.11097121e-01 1.21611965e+00 -8.71365488e-01 -1.23033524e+00 9.15577412e-01 -1.71332642e-01 -4.09122825e-01 8.39826584e-01 -6.09079719e-01 -3.38578343e-01 1.14693654e+00 8.42636228e-02 7.42493808e-01 1.15361691e+00 -5.74847996e-01 -1.07353079e+00 -1.47404537e-01 1.19813986e-01 1.25156000e-01 -2.32601106e-01 4.99460667e-01 -6.38164520e-01 -8.48382831e-01 3.71607482e-01 -7.10798204e-01 3.90137404e-01 1.64733976e-01 -6.06906973e-02 -2.90068299e-01 5.40310144e-01 -1.16421998e+00 1.15608120e+00 -2.05126119e+00 2.63112485e-01 -1.77094176e-01 4.06599373e-01 3.93785775e-01 -2.29235113e-01 2.62608439e-01 -4.88153189e-01 -3.93945456e-01 1.42991632e-01 -3.44625413e-01 -3.56156528e-01 -2.28165373e-01 3.45567465e-01 8.43153477e-01 1.03268608e-01 6.84147716e-01 -1.09646869e+00 -5.59527516e-01 3.18303347e-01 3.70977581e-01 -8.50838602e-01 4.20440584e-01 8.00634399e-02 7.31172740e-01 -1.81073070e-01 1.01701379e+00 3.54932934e-01 5.32272980e-02 -2.00770915e-01 -1.40621066e-01 -1.71989128e-01 -5.56623526e-02 -9.38966453e-01 1.84919310e+00 9.40069631e-02 7.68890321e-01 9.38707814e-02 -1.44346082e+00 1.42229244e-01 6.84771299e-01 9.17244434e-01 -5.58804989e-01 3.76733392e-01 2.56802112e-01 1.68193057e-01 -1.26822948e+00 -3.98152649e-01 3.55915695e-01 5.41460276e-01 1.43920004e-01 3.09359819e-01 4.54646826e-01 6.62692070e-01 5.74555844e-02 1.49530816e+00 4.23088640e-01 2.35032290e-01 9.12684798e-02 4.57473189e-01 -9.95890275e-02 6.35152757e-01 6.49079919e-01 -2.96202362e-01 9.52128589e-01 3.88651252e-01 -2.50146300e-01 -6.45045340e-01 -7.09893644e-01 1.65415570e-01 1.32670057e+00 -1.97056636e-01 -3.03049356e-01 -1.25613117e+00 -7.03917325e-01 -3.33086818e-01 2.96965450e-01 -5.95219672e-01 -1.05508834e-01 -1.00199378e+00 -3.79896492e-01 3.19258243e-01 8.93146038e-01 1.22759409e-01 -1.36190689e+00 -1.01831067e+00 2.21853584e-01 -4.06794935e-01 -1.27315044e+00 -6.31879270e-01 -1.58103585e-01 -5.57147741e-01 -1.42987895e+00 -1.33873570e+00 -1.07182848e+00 9.42995429e-01 2.41623655e-01 3.21725726e-01 1.08069748e-01 -4.61183429e-01 5.07997990e-01 -1.08653927e+00 -2.84644067e-01 -3.68855834e-01 -3.34997892e-01 1.68585286e-01 5.64427525e-02 3.15176547e-01 -8.71106327e-01 -9.43848133e-01 1.73970431e-01 -5.27436316e-01 2.96426117e-01 5.29547513e-01 5.87184191e-01 3.33408982e-01 -1.98307008e-01 4.74784851e-01 -2.28161201e-01 -1.25736771e-02 -5.00495255e-01 -5.30523419e-01 1.82093427e-01 -1.27721250e-01 -4.10177618e-01 4.87525940e-01 -9.85174298e-01 -5.67719281e-01 4.25330177e-02 -2.66491562e-01 -9.35192347e-01 -4.72538233e-01 3.38238418e-01 -3.11362855e-02 -6.79813027e-02 3.40014696e-01 2.12486148e-01 -1.24430351e-01 -5.88077903e-01 8.72371048e-02 5.58012486e-01 9.80656326e-01 -4.08124924e-01 4.80568707e-01 3.74381810e-01 -2.27988124e-01 -6.58889174e-01 -6.60095811e-01 -7.83501923e-01 -8.59467804e-01 -9.00911689e-01 1.20438290e+00 -9.63909268e-01 -6.57265782e-01 8.05024981e-01 -1.11541700e+00 -5.44893444e-01 1.25726223e-01 1.03998446e+00 -7.74385571e-01 3.78513366e-01 -5.77576280e-01 -6.08630002e-01 -4.15349394e-01 -1.06384873e+00 8.05118382e-01 1.63255781e-01 -1.67931635e-02 -4.16828841e-01 -1.59522846e-01 6.29504800e-01 -9.47033614e-02 5.58285475e-01 6.68375671e-01 -6.35397434e-01 -3.44580054e-01 -4.89585429e-01 -6.23398945e-02 3.75795990e-01 -9.92269441e-03 3.73041369e-02 -8.61765921e-01 -5.38881660e-01 -7.38164559e-02 -1.21129818e-01 6.00859225e-01 6.51273608e-01 1.07207298e+00 -1.62274480e-01 -1.55671895e-01 6.97958648e-01 8.61522317e-01 6.47517800e-01 2.07981184e-01 3.10315251e-01 7.83798158e-01 7.25239396e-01 1.05229223e+00 4.66036618e-01 3.99744809e-01 7.10008502e-01 5.48548102e-01 9.30373222e-02 -5.14163494e-01 -1.15010865e-01 7.29733706e-01 1.10462940e+00 -6.38662696e-01 -1.63618788e-01 -1.06303608e+00 8.60445857e-01 -1.95901620e+00 -1.09857702e+00 2.92293765e-02 2.04205251e+00 6.64864480e-01 1.10811993e-01 5.49528658e-01 5.09504080e-01 9.11676168e-01 -9.65157226e-02 -4.63219136e-01 7.68758208e-02 2.00345084e-01 6.21583536e-02 3.39955002e-01 9.61077660e-02 -1.17554986e+00 6.32021010e-01 4.74815655e+00 7.67242908e-01 -1.09890866e+00 5.61652005e-01 6.52999505e-02 -6.38188124e-01 6.02982283e-01 -3.82480383e-01 -4.81161118e-01 7.58533120e-01 5.61379611e-01 2.58758605e-01 4.08877492e-01 9.63746488e-01 4.84278738e-01 -2.42397279e-01 -1.25488448e+00 1.26627624e+00 2.17211053e-01 -9.54263806e-01 -3.86296958e-01 -3.93156648e-01 5.02674699e-01 3.50582123e-01 -5.20810127e-01 6.14728071e-02 -4.28526312e-01 -6.69661582e-01 1.24423623e+00 6.28592491e-01 9.68050241e-01 -6.96942151e-01 4.64355469e-01 6.36224747e-01 -1.34480381e+00 -4.07157093e-01 -5.44831455e-02 1.63753420e-01 2.16825098e-01 6.21026345e-02 -6.01446092e-01 3.30789506e-01 1.05431008e+00 7.17624247e-01 -3.55238169e-01 1.36951280e+00 -5.18799722e-01 7.52111912e-01 -3.36590827e-01 2.94313848e-01 1.83268059e-02 1.64874777e-01 8.46971631e-01 1.31344819e+00 4.67045516e-01 3.28850627e-01 4.20933068e-02 1.79036498e-01 1.61444336e-01 2.86236882e-01 -2.93079227e-01 6.15135953e-03 1.78918481e-01 8.58432531e-01 -8.72942448e-01 -3.29770625e-01 -8.90635729e-01 9.37821209e-01 3.53325874e-01 1.82275996e-01 -9.25867081e-01 -2.20885351e-01 2.89379507e-01 4.15804297e-01 4.48851883e-01 6.13631718e-02 4.58052009e-01 -1.30459726e+00 3.61085832e-01 -9.27477837e-01 5.48739254e-01 -7.91733921e-01 -4.67321128e-01 2.56303012e-01 2.89153486e-01 -1.63273513e+00 -4.21845943e-01 -5.78055263e-01 -7.97238052e-01 1.28334314e-01 -1.13361931e+00 -9.35577095e-01 -4.68610525e-01 8.31525028e-01 1.01067853e+00 -8.96489322e-02 4.15753871e-01 7.49409437e-01 -6.94442272e-01 7.11402893e-01 -3.41390282e-01 4.23828632e-01 5.08504748e-01 -1.01055706e+00 1.16457202e-01 1.31305027e+00 6.68426082e-02 1.92803472e-01 6.89222813e-01 -7.63662457e-01 -1.08190560e+00 -8.33023727e-01 4.91511613e-01 -1.46366313e-01 4.77281451e-01 -2.15112269e-01 -7.91641772e-01 5.29684424e-01 -1.11163035e-01 3.35070103e-01 1.40183195e-01 -7.62938321e-01 1.41037302e-02 1.46037549e-01 -1.00642121e+00 3.34425747e-01 1.65094447e+00 -3.75776440e-01 -6.19740725e-01 6.04454935e-01 6.21213257e-01 -7.11186826e-01 -8.70929837e-01 7.39748001e-01 6.81693733e-01 -7.97058463e-01 8.45419705e-01 -4.68039781e-01 4.44645435e-01 -2.46740818e-01 1.20673992e-01 -9.10841227e-01 7.46515691e-02 -6.71354413e-01 -6.48748517e-01 9.83780563e-01 1.67676851e-01 -4.82217848e-01 6.58922851e-01 3.47313583e-01 -3.71895194e-01 -9.87444818e-01 -9.73374426e-01 -7.76443720e-01 -3.19894284e-01 -8.04941237e-01 2.69413322e-01 7.83660412e-01 3.43316972e-01 -4.09694850e-01 -3.32573354e-01 4.36494499e-01 5.73771834e-01 -6.45608231e-02 4.44294810e-01 -7.65397370e-01 -3.32426637e-01 -5.18810332e-01 -7.38320649e-01 -7.50445068e-01 -1.89715534e-01 -8.33201528e-01 1.01775773e-01 -1.48255789e+00 1.54938102e-01 2.66859770e-01 -3.13510805e-01 7.06189573e-01 6.32046089e-02 1.13990478e-01 1.08502567e-01 -2.30430812e-02 -5.83565295e-01 -6.54223934e-03 1.02794969e+00 1.00702085e-01 -1.09161474e-01 3.31081390e-01 9.70864892e-02 1.17706585e+00 7.87324667e-01 -6.97371542e-01 -4.26168710e-01 -4.94438678e-01 -5.79339489e-02 4.67875808e-01 3.20093662e-01 -1.18614852e+00 3.63790959e-01 2.40133516e-02 2.76857138e-01 -9.73473489e-01 2.24603370e-01 -5.79819202e-01 -2.30331123e-01 6.09093308e-01 -5.90567030e-02 -2.04155911e-02 -6.23284355e-02 2.29825497e-01 -2.06945732e-01 -3.90695304e-01 7.72466779e-01 1.68200694e-02 -7.32706249e-01 5.67169309e-01 -5.44098675e-01 2.87546694e-01 1.14673662e+00 -3.12956691e-01 -1.26005933e-01 -2.59975553e-01 -1.18800795e+00 2.50344306e-01 2.24319354e-01 6.09644949e-01 7.37232566e-01 -9.23749447e-01 -6.04546964e-01 2.28000879e-01 4.91468757e-02 2.08702996e-01 5.14373124e-01 1.41569054e+00 -7.71131277e-01 2.38348648e-01 -1.38808444e-01 -7.03989804e-01 -1.64923167e+00 7.12963223e-01 5.48862517e-02 3.67269278e-01 -1.29519832e+00 1.04562521e+00 1.89771742e-01 3.01967859e-01 8.10955763e-01 -6.46273673e-01 -5.65721869e-01 3.49816710e-01 9.41498578e-01 6.61303699e-01 6.32484034e-02 -8.70140672e-01 -4.45776165e-01 6.85188830e-01 5.16179502e-02 7.40627199e-02 1.47087014e+00 -5.56403473e-02 1.85144082e-01 1.95614267e-02 1.15212154e+00 -1.57432750e-01 -1.47865701e+00 -6.40265569e-02 -3.39088179e-02 -4.25952971e-01 -1.86936975e-01 -4.73725706e-01 -1.42904437e+00 1.13590086e+00 7.87569880e-01 -1.00581400e-01 1.41540813e+00 3.10831070e-01 7.72025108e-01 9.87610817e-02 6.17506802e-01 -1.12145174e+00 2.01344281e-01 -8.01612958e-02 1.21169186e+00 -1.10482454e+00 -2.01966882e-01 -2.86012888e-01 -4.25781429e-01 1.20895767e+00 5.65088034e-01 -6.90323999e-03 4.50908929e-01 2.77077138e-01 -2.98823323e-02 -7.10379183e-02 -5.63916922e-01 -3.30139846e-01 3.51614684e-01 6.27444446e-01 -1.61588624e-01 3.02102137e-02 -3.48084718e-01 8.01332235e-01 -2.87405346e-02 1.01212151e-01 3.62512767e-01 1.18324375e+00 -4.58177507e-01 -6.70035779e-01 -2.78141439e-01 4.76684958e-01 -9.51625168e-01 6.04400299e-02 2.99110532e-01 6.21446252e-01 4.19727832e-01 8.78811061e-01 -4.15934235e-01 -3.30962092e-01 3.29646617e-01 1.52562827e-01 6.79954171e-01 -6.76351666e-01 -1.03644989e-01 3.83911848e-01 1.31158948e-01 -1.13597012e+00 -5.46312988e-01 -1.15776825e+00 -1.56015968e+00 2.08482668e-01 -2.21875966e-01 -2.83205509e-01 5.33786535e-01 9.98624682e-01 -7.30563551e-02 4.81305242e-01 3.11552614e-01 -1.10398269e+00 6.56864643e-02 -9.75009263e-01 -2.29883701e-01 2.38701984e-01 6.62505329e-01 -9.44180012e-01 -3.68658572e-01 2.74963349e-01]
[8.548392295837402, 0.5182920098304749]
bab166c3-3548-419e-a312-6cfb4f82bc90
pu-gcn-point-cloud-upsampling-using-graph
1912.03264
null
https://arxiv.org/abs/1912.03264v3
https://arxiv.org/pdf/1912.03264v3.pdf
PU-GCN: Point Cloud Upsampling using Graph Convolutional Networks
The effectiveness of learning-based point cloud upsampling pipelines heavily relies on the upsampling modules and feature extractors used therein. For the point upsampling module, we propose a novel model called NodeShuffle, which uses a Graph Convolutional Network (GCN) to better encode local point information from point neighborhoods. NodeShuffle is versatile and can be incorporated into any point cloud upsampling pipeline. Extensive experiments show how NodeShuffle consistently improves state-of-the-art upsampling methods. For feature extraction, we also propose a new multi-scale point feature extractor, called Inception DenseGCN. By aggregating features at multiple scales, this feature extractor enables further performance gain in the final upsampled point clouds. We combine Inception DenseGCN with NodeShuffle into a new point upsampling pipeline called PU-GCN. PU-GCN sets new state-of-art performance with much fewer parameters and more efficient inference.
['Guocheng Qian', 'Abdulellah Abualshour', 'Guohao Li', 'Bernard Ghanem', 'Ali Thabet']
2019-11-30
null
http://openaccess.thecvf.com//content/CVPR2021/html/Qian_PU-GCN_Point_Cloud_Upsampling_Using_Graph_Convolutional_Networks_CVPR_2021_paper.html
http://openaccess.thecvf.com//content/CVPR2021/papers/Qian_PU-GCN_Point_Cloud_Upsampling_Using_Graph_Convolutional_Networks_CVPR_2021_paper.pdf
cvpr-2021-1
['point-cloud-super-resolution']
['computer-vision']
[-3.58194947e-01 -1.80289581e-01 -1.99419454e-01 -2.98646897e-01 -6.88172221e-01 -3.02455157e-01 8.37454438e-01 1.41314253e-01 6.44098148e-02 4.39847261e-01 9.64405853e-03 -1.28218561e-01 -2.41058301e-02 -1.73190141e+00 -1.05534244e+00 -2.08623439e-01 -2.50139713e-01 4.88143682e-01 4.98069167e-01 -2.54154414e-01 1.72692969e-01 1.31749642e+00 -1.87713528e+00 3.17076415e-01 9.11457658e-01 9.42352593e-01 1.29235193e-01 6.41121387e-01 -4.79263395e-01 7.17923120e-02 -1.41048819e-01 -5.45615405e-02 6.67080283e-01 5.07784784e-01 -5.51701546e-01 -2.48570815e-01 1.21259439e+00 -7.52174139e-01 -4.39970255e-01 7.96191990e-01 2.64295876e-01 1.40097618e-01 2.77156532e-01 -1.30481505e+00 -4.17327404e-01 4.65927720e-01 -6.16306543e-01 -3.07045672e-02 -1.13595054e-01 4.62403089e-01 9.03055727e-01 -1.35856342e+00 5.73237777e-01 1.49700117e+00 1.33443999e+00 2.79329605e-02 -1.15171969e+00 -8.69229257e-01 5.03943078e-02 -2.59415120e-01 -1.43132591e+00 -8.36494565e-02 8.42662096e-01 -3.44791681e-01 1.22035933e+00 2.60832310e-01 1.08036232e+00 3.11930358e-01 2.92036254e-02 7.18621433e-01 4.66294140e-01 4.59393784e-02 3.17121893e-01 -6.00009978e-01 -8.79304856e-03 8.95652711e-01 4.79145527e-01 2.14468554e-01 -4.89769459e-01 -5.61933875e-01 1.61215270e+00 6.64457142e-01 -6.54901490e-02 -5.17662764e-01 -1.26372743e+00 8.78611863e-01 1.32045960e+00 -7.04871444e-03 -4.64496195e-01 6.39047444e-01 2.05934599e-01 4.00968678e-02 9.16056991e-01 4.70645785e-01 -5.63070536e-01 8.38972256e-02 -1.48458970e+00 7.73097396e-01 4.00680780e-01 1.25949061e+00 1.60885739e+00 -5.53607009e-02 -2.00247720e-01 6.42448246e-01 2.28886440e-01 8.54705215e-01 -2.40855753e-01 -1.10472000e+00 5.47944784e-01 1.04841173e+00 -7.53605440e-02 -1.01709533e+00 -4.01951402e-01 -5.33288121e-01 -8.49978149e-01 5.74029505e-01 -2.32995138e-01 -4.24962379e-02 -1.06270075e+00 9.63908911e-01 6.43353522e-01 6.49465144e-01 -3.60507250e-01 7.08353460e-01 9.53223944e-01 6.71633303e-01 -1.25462130e-01 5.70502698e-01 1.24251795e+00 -9.80940163e-01 -1.63491875e-01 -8.53393674e-02 4.00285035e-01 -5.01491487e-01 9.57845807e-01 4.77320328e-02 -1.04244328e+00 -6.73617005e-01 -1.09045064e+00 -7.02694118e-01 -4.12366182e-01 2.93875299e-02 1.28348470e+00 -2.80937329e-02 -1.33658350e+00 1.22837234e+00 -1.11852562e+00 -1.09491460e-01 1.06081426e+00 4.79971856e-01 -3.17743212e-01 -2.83501327e-01 -7.71199882e-01 3.03446442e-01 9.54821631e-02 -1.84413224e-01 -5.68069100e-01 -1.57142735e+00 -1.05354977e+00 2.80587077e-01 3.24667767e-02 -1.36538815e+00 1.10243285e+00 -2.44074166e-01 -1.14049399e+00 4.84763205e-01 -4.03957695e-01 -4.64709073e-01 3.41585070e-01 -2.68899232e-01 -5.02385013e-02 3.10809672e-01 3.00981849e-01 9.34124887e-01 9.42413926e-01 -1.08747530e+00 -7.64177680e-01 -5.46983719e-01 -1.43558308e-01 -4.02588844e-02 1.09244958e-01 -5.08894324e-01 -4.75306600e-01 -4.86917287e-01 3.72741163e-01 -5.92554808e-01 -3.01791817e-01 5.62652767e-01 -2.08543912e-01 -4.13016528e-01 1.37600017e+00 -1.26666218e-01 8.79841924e-01 -2.17938948e+00 -3.52162868e-01 1.65555164e-01 8.05918694e-01 1.98433056e-01 -2.35232398e-01 5.85658431e-01 1.23836339e-01 8.39186385e-02 -3.42745125e-01 -7.09747612e-01 1.52928140e-02 3.14148694e-01 -4.74096656e-01 3.45223457e-01 6.78606212e-01 1.08240402e+00 -1.09651005e+00 -2.57243186e-01 7.54984617e-01 8.34839523e-01 -1.07768726e+00 -2.09219635e-01 -3.61816823e-01 3.15930769e-02 -7.27539539e-01 8.91942382e-01 1.48188889e+00 -4.14380372e-01 -6.65595949e-01 -3.78450841e-01 -4.97000426e-01 2.89485097e-01 -9.27645087e-01 1.98539913e+00 -4.53736186e-01 4.37198907e-01 1.24671400e-01 -1.46468014e-01 1.00878024e+00 -1.37502551e-01 6.15171671e-01 -5.75153017e-03 -2.38996491e-01 2.79868245e-01 -5.20668328e-01 8.34107921e-02 9.67850268e-01 -6.54588128e-03 3.14799398e-01 -4.86033894e-02 3.05567950e-01 -7.92129815e-01 -1.10494010e-01 4.66260314e-01 1.12183416e+00 2.56376743e-01 7.65119940e-02 -5.33973455e-01 1.97931349e-01 1.28081575e-01 4.67523754e-01 7.34013796e-01 1.11483566e-01 8.73389304e-01 2.30645776e-01 -7.88097143e-01 -1.22356725e+00 -1.01082301e+00 -3.82679462e-01 7.09716856e-01 -3.40881161e-02 -8.95197272e-01 -3.82694274e-01 -5.81083596e-01 8.18171203e-01 3.65694672e-01 -4.76777852e-01 2.42547944e-01 -5.20725906e-01 -2.54876912e-01 4.28981125e-01 7.08923399e-01 7.75397599e-01 -6.15964234e-01 -2.36453101e-01 1.02757558e-01 3.80208790e-01 -9.45640624e-01 -1.18298844e-01 -9.42837000e-02 -1.46693289e+00 -8.49950671e-01 -4.40130264e-01 -3.22451144e-01 5.92634439e-01 5.60862422e-01 1.54148543e+00 3.29388916e-01 -1.27010513e-02 2.35246141e-02 -2.26346686e-01 -6.12708747e-01 2.22230405e-01 5.69913328e-01 -4.95636053e-02 -4.16293502e-01 6.85976028e-01 -1.09686494e+00 -8.74242187e-01 -1.40977591e-01 -9.20759737e-01 1.83398515e-01 4.21529830e-01 3.92539710e-01 9.16278660e-01 -8.24606940e-02 1.71451002e-01 -7.32196033e-01 6.97972532e-03 -4.44610745e-01 -8.14720094e-01 -3.94330174e-01 -2.56819904e-01 2.19605625e-01 7.02762544e-01 1.19918033e-01 -5.26237905e-01 1.21481292e-01 -5.83941221e-01 -9.68568027e-01 -4.15708981e-02 6.80655837e-02 6.07046750e-05 -6.22227907e-01 6.05248272e-01 -1.31165653e-01 6.70169070e-02 -6.91235662e-01 6.40797913e-01 3.75767708e-01 5.38561165e-01 -5.60268164e-01 1.11280370e+00 1.10605097e+00 2.71771103e-01 -1.20758212e+00 -6.05678976e-01 -6.14672303e-01 -7.29389966e-01 -2.05807444e-02 5.56566596e-01 -1.27776289e+00 -5.41353643e-01 4.66558903e-01 -1.25345349e+00 -2.50457197e-01 -6.41736805e-01 2.70375520e-01 -4.32587773e-01 1.03112102e-01 -5.94076991e-01 -3.88449252e-01 -5.78541696e-01 -1.06497347e+00 1.96604562e+00 1.26291543e-01 2.07081333e-01 -8.01441610e-01 6.66717216e-02 -2.23645255e-01 5.42880416e-01 4.49461371e-01 5.78337252e-01 -2.34140486e-01 -1.29320168e+00 -2.46492371e-01 -6.13497496e-01 2.06281513e-01 1.27377495e-01 3.95283192e-01 -1.08622038e+00 -2.85208136e-01 -2.94881761e-01 8.68333504e-02 1.08234584e+00 4.98000264e-01 1.33420885e+00 -1.78714976e-01 -5.95934868e-01 1.44919837e+00 1.78945112e+00 -6.58866048e-01 7.74868369e-01 2.60689318e-01 1.26355815e+00 -1.04936078e-01 3.15105468e-01 5.03878236e-01 4.29775357e-01 2.67448455e-01 9.26691949e-01 -2.48112336e-01 -3.82305443e-01 -5.28803170e-01 -2.10240617e-01 7.77992427e-01 -3.18530232e-01 3.97162080e-01 -8.70182633e-01 5.27092457e-01 -1.73158586e+00 -8.69647801e-01 -4.48590428e-01 1.84703958e+00 5.32063901e-01 -2.25608885e-01 -1.62541434e-01 -1.81702286e-01 4.56271470e-01 6.56638920e-01 -5.51131606e-01 -4.43606824e-03 1.24327257e-01 7.86607325e-01 8.75851631e-01 5.92290461e-01 -1.09880650e+00 1.22644889e+00 5.94780731e+00 1.02207327e+00 -1.16916656e+00 -9.63181257e-02 1.92578778e-01 -1.71339720e-01 -6.92171514e-01 1.35585487e-01 -1.21023190e+00 2.60194153e-01 4.78807509e-01 6.61213771e-02 4.42390710e-01 1.34670293e+00 1.98745187e-02 1.81623518e-01 -9.55265939e-01 1.13379288e+00 -3.27091485e-01 -2.08178210e+00 3.95973802e-01 1.87943071e-01 8.19123626e-01 8.90583634e-01 -2.13028595e-01 5.17580330e-01 4.89693373e-01 -7.58736968e-01 3.59064996e-01 6.12660110e-01 1.03028178e+00 -1.06033027e+00 3.97109866e-01 1.99179262e-01 -1.54928684e+00 4.28852946e-01 -8.74020696e-01 -2.43226960e-01 1.95919827e-01 1.31706178e+00 -8.60226154e-01 9.40079331e-01 6.31555736e-01 1.10910308e+00 -6.70587242e-01 1.14956021e+00 -1.13733254e-01 2.55594313e-01 -8.42713177e-01 3.39077950e-01 4.18891311e-01 -1.83587268e-01 6.85718358e-01 1.11989605e+00 5.11848092e-01 -1.65718012e-02 1.85674533e-01 1.18323612e+00 -3.09221953e-01 -3.71691227e-01 -8.34146202e-01 2.99635977e-01 8.70226204e-01 1.47135007e+00 -5.60048103e-01 -6.53362453e-01 -3.35512370e-01 6.15285873e-01 6.50766253e-01 1.47482723e-01 -4.83018190e-01 -6.86888158e-01 1.37356889e+00 5.44211209e-01 6.72848940e-01 -4.82423395e-01 -4.63186026e-01 -1.28113997e+00 -2.57993847e-01 -4.27875340e-01 -2.55779922e-01 -8.67256105e-01 -1.32360935e+00 6.44262016e-01 1.83980092e-02 -1.43985796e+00 -2.26503834e-02 -5.96290827e-01 -6.09708250e-01 1.08339679e+00 -1.84008968e+00 -1.33855081e+00 -7.35433877e-01 6.22996211e-01 5.50208747e-01 2.22978324e-01 5.01490593e-01 2.05481008e-01 -1.26380920e-01 9.25356299e-02 -1.12642013e-01 8.23531225e-02 3.02411497e-01 -1.17912197e+00 1.42063379e+00 7.91048110e-01 -6.22442923e-02 9.37761068e-01 1.53731555e-01 -1.05708575e+00 -1.51734960e+00 -1.77871704e+00 4.53853130e-01 -4.09353346e-01 5.31394958e-01 -3.68097275e-01 -9.29971516e-01 7.42153287e-01 -2.84556568e-01 7.40721285e-01 7.03195408e-02 1.39812618e-01 -3.57974768e-01 -1.91224620e-01 -1.32534635e+00 4.34941143e-01 1.22017014e+00 -5.82721889e-01 -5.29908478e-01 3.97988766e-01 1.28194582e+00 -6.19803071e-01 -1.00917685e+00 8.02831352e-01 2.16492385e-01 -1.00667107e+00 1.29217660e+00 -2.88976640e-01 6.12705231e-01 -5.31310797e-01 -3.62898745e-02 -1.29344213e+00 -8.56768787e-01 -3.74129862e-01 -4.63813365e-01 9.03874218e-01 6.86931685e-02 -7.40430295e-01 9.53260064e-01 1.66781619e-01 -4.88632292e-01 -8.75714421e-01 -6.29756570e-01 -6.86618030e-01 5.01393788e-02 -6.01590157e-01 1.61082232e+00 1.00801504e+00 -4.76356655e-01 3.55758965e-02 2.30605885e-01 5.74093580e-01 9.37504470e-01 3.04213375e-01 1.17917669e+00 -1.57066834e+00 3.46698649e-02 -2.61628240e-01 -6.94913328e-01 -1.36010456e+00 -2.47524410e-01 -9.76665080e-01 -1.61099583e-01 -1.92279553e+00 -2.95575976e-01 -7.86742032e-01 2.50609875e-01 4.71699089e-01 -3.18934381e-01 3.92805606e-01 1.90352872e-01 4.28956509e-01 -1.58349097e-01 6.82748079e-01 1.37903225e+00 6.17645159e-02 -1.66048735e-01 -1.34775773e-01 -3.35540593e-01 9.03273225e-01 6.33761883e-01 -2.39833042e-01 -9.08094719e-02 -8.06093514e-01 1.37406856e-01 -2.69465476e-01 5.66307724e-01 -1.47534645e+00 1.83325440e-01 6.28170222e-02 5.76124787e-01 -1.48486114e+00 5.55953085e-01 -7.27312505e-01 1.82877004e-01 7.65242130e-02 3.72215837e-01 2.14727387e-01 4.29848313e-01 4.67936665e-01 -1.15405902e-01 9.47412550e-02 5.93935132e-01 -3.31717908e-01 -5.55328190e-01 1.14342892e+00 3.43367845e-01 -3.83482307e-01 6.13625169e-01 -1.18925981e-01 -4.47060674e-01 1.31290779e-01 -3.26515198e-01 3.01844388e-01 8.47202539e-01 1.70758963e-01 6.91598713e-01 -1.66597116e+00 -6.93066657e-01 7.01637030e-01 9.60688293e-02 1.00001848e+00 3.58529180e-01 3.84470433e-01 -1.17600691e+00 2.03088999e-01 7.64406696e-02 -1.00327837e+00 -7.84104526e-01 4.51884568e-01 6.28879219e-02 -2.19826639e-01 -1.14241803e+00 9.94736195e-01 2.44168621e-02 -7.55284131e-01 -3.71056050e-01 -1.14215147e+00 3.58630985e-01 -1.87196255e-01 6.30105972e-01 4.91865098e-01 4.88568008e-01 -3.60606492e-01 -2.20458373e-01 5.93326688e-01 6.46707863e-02 2.50306964e-01 1.58252060e+00 4.32616949e-01 -4.70227987e-01 2.66295642e-01 1.22273433e+00 2.99582779e-01 -1.50897110e+00 -1.94181442e-01 -5.55684030e-01 -9.08494115e-01 4.71271336e-01 -2.68976428e-02 -1.31530416e+00 7.64959216e-01 6.20651729e-02 2.14455053e-01 7.03633666e-01 -1.60210595e-01 1.03647768e+00 2.98181146e-01 9.65955555e-01 -5.25054395e-01 -6.38540268e-01 6.64629817e-01 8.83629739e-01 -1.05483544e+00 5.09393275e-01 -7.82792509e-01 6.29228801e-02 1.10400116e+00 4.89457905e-01 -1.05343962e+00 8.63013268e-01 4.68036085e-01 -4.45066005e-01 -7.22778320e-01 -5.04023015e-01 -2.53599375e-01 2.48406708e-01 5.78639984e-01 5.26826493e-02 2.04560310e-01 2.55273223e-01 2.97277179e-02 -6.51585340e-01 3.44590038e-01 -8.55818484e-03 7.79507637e-01 -6.16604030e-01 -9.32880580e-01 -5.16910374e-01 8.87737691e-01 1.37690693e-01 -4.14286584e-01 -1.81065910e-02 8.65756512e-01 1.77226946e-01 2.32783660e-01 5.64264178e-01 -4.40021545e-01 3.76040608e-01 -2.17630476e-01 2.65873522e-01 -7.53969491e-01 -7.71754742e-01 -2.18900487e-01 -3.02638263e-01 -1.21986878e+00 -2.19557732e-01 -4.89958972e-01 -1.22395754e+00 -7.42377400e-01 -2.93182611e-01 1.46776050e-01 7.25518882e-01 4.97821271e-01 1.12296081e+00 8.07045043e-01 4.77995783e-01 -1.82127786e+00 -1.77653193e-01 -9.92988944e-01 -6.59218371e-01 9.44472756e-03 6.91194415e-01 -5.83356023e-01 -3.48087996e-01 -5.73820412e-01]
[8.101410865783691, -3.5731799602508545]
9bbc94d5-ac91-4602-926d-2ecb7c504cb3
unsupervised-cross-dataset-person-re
1803.07293
null
http://arxiv.org/abs/1803.07293v1
http://arxiv.org/pdf/1803.07293v1.pdf
Unsupervised Cross-dataset Person Re-identification by Transfer Learning of Spatial-Temporal Patterns
Most of the proposed person re-identification algorithms conduct supervised training and testing on single labeled datasets with small size, so directly deploying these trained models to a large-scale real-world camera network may lead to poor performance due to underfitting. It is challenging to incrementally optimize the models by using the abundant unlabeled data collected from the target domain. To address this challenge, we propose an unsupervised incremental learning algorithm, TFusion, which is aided by the transfer learning of the pedestrians' spatio-temporal patterns in the target domain. Specifically, the algorithm firstly transfers the visual classifier trained from small labeled source dataset to the unlabeled target dataset so as to learn the pedestrians' spatial-temporal patterns. Secondly, a Bayesian fusion model is proposed to combine the learned spatio-temporal patterns with visual features to achieve a significantly improved classifier. Finally, we propose a learning-to-rank based mutual promotion procedure to incrementally optimize the classifiers based on the unlabeled data in the target domain. Comprehensive experiments based on multiple real surveillance datasets are conducted, and the results show that our algorithm gains significant improvement compared with the state-of-art cross-dataset unsupervised person re-identification algorithms.
['Qing Li', 'Jianming Lv', 'Weihang Chen', 'Can Yang']
2018-03-20
unsupervised-cross-dataset-person-re-1
http://openaccess.thecvf.com/content_cvpr_2018/html/Lv_Unsupervised_Cross-Dataset_Person_CVPR_2018_paper.html
http://openaccess.thecvf.com/content_cvpr_2018/papers/Lv_Unsupervised_Cross-Dataset_Person_CVPR_2018_paper.pdf
cvpr-2018-6
['unsupervised-person-re-identification']
['computer-vision']
[ 1.74134091e-01 -4.73593533e-01 -2.94230998e-01 -5.99915862e-01 -4.36419070e-01 -2.87451744e-01 5.59222639e-01 -1.19116299e-01 -5.24042785e-01 7.09589839e-01 2.74142116e-01 1.28338039e-01 -9.31400508e-02 -5.21678686e-01 -4.74518955e-01 -7.16320395e-01 -4.34468267e-03 5.14546692e-01 3.06071788e-01 2.54991323e-01 1.04066841e-01 1.59034565e-01 -1.68315101e+00 1.26070529e-01 9.24437046e-01 9.03986156e-01 1.32111087e-01 4.45472091e-01 1.51500493e-01 8.71374309e-01 -3.49946499e-01 -4.78319824e-01 3.63825679e-01 -1.88695669e-01 -5.63600540e-01 5.38496554e-01 5.82958877e-01 -5.97779393e-01 -5.80713689e-01 1.27823627e+00 4.47226405e-01 2.84303486e-01 7.06818879e-01 -1.36197901e+00 -5.49511552e-01 1.97323114e-01 -7.07951784e-01 2.37674668e-01 1.89341635e-01 1.70052290e-01 4.78513449e-01 -8.67100835e-01 1.96128935e-01 1.29573905e+00 7.80068994e-01 5.12327194e-01 -1.08613062e+00 -9.33466554e-01 3.74067545e-01 6.03253186e-01 -1.61139917e+00 -4.99481618e-01 8.37936699e-01 -5.07324457e-01 3.99812043e-01 -1.99695881e-02 5.54996848e-01 9.90342677e-01 -4.58223283e-01 8.04741979e-01 1.13497734e+00 -3.54463428e-01 -1.47165135e-01 3.69321197e-01 3.52243751e-01 7.61131406e-01 3.06351751e-01 5.49281180e-01 -3.99980515e-01 -3.06943357e-02 4.74177092e-01 4.60765183e-01 4.11141999e-02 -3.62849504e-01 -1.16824639e+00 5.00831246e-01 5.44626117e-01 5.39196916e-02 -1.55063465e-01 -3.05492520e-01 3.92320961e-01 1.08325630e-01 4.53984767e-01 -1.95631072e-01 -3.04307967e-01 2.13937208e-01 -8.81839454e-01 -6.28023893e-02 3.06136996e-01 9.36818063e-01 1.05087709e+00 -3.34633231e-01 -2.32335478e-01 1.12356699e+00 6.46670341e-01 6.93843901e-01 4.72172409e-01 -6.68368936e-01 6.68696105e-01 8.46713960e-01 1.45668998e-01 -1.01282120e+00 -2.19416529e-01 -3.58413160e-01 -1.04212224e+00 -8.00936744e-02 5.73220372e-01 -3.48875672e-01 -8.39783549e-01 1.62710178e+00 4.74602938e-01 8.28364491e-01 1.42040700e-01 9.39174771e-01 7.16730654e-01 6.29164398e-01 3.43271047e-01 -3.74750286e-01 1.26980937e+00 -1.12789261e+00 -3.25712234e-01 -2.89398670e-01 4.74460661e-01 -4.23503697e-01 4.50975031e-01 1.89968962e-02 -3.24741185e-01 -1.08741105e+00 -1.00971925e+00 5.20659566e-01 -3.08655053e-01 7.06833422e-01 2.06683904e-01 5.87598205e-01 -7.71507859e-01 6.06988706e-02 -5.58992147e-01 -7.22921193e-01 4.90985900e-01 4.41787452e-01 -5.77702820e-01 -3.20947349e-01 -9.93322849e-01 5.98490000e-01 7.98301101e-01 3.15943450e-01 -8.27880025e-01 -3.87401283e-01 -6.43871903e-01 -2.83810735e-01 3.74224663e-01 -5.02680898e-01 9.69827652e-01 -1.09051132e+00 -1.18867743e+00 8.93545985e-01 -3.11211199e-01 -3.83159965e-01 4.78898287e-01 -7.05748275e-02 -7.18024611e-01 5.74830770e-02 3.37372661e-01 5.90776026e-01 8.26770186e-01 -1.38718271e+00 -1.23327303e+00 -5.95992565e-01 -1.05524302e-01 3.69411945e-01 -8.33252728e-01 -7.45567679e-02 -7.40514338e-01 -6.10431552e-01 -9.51778367e-02 -1.01913846e+00 -2.00245246e-01 -2.46836722e-01 -2.79129833e-01 -3.07466477e-01 1.01686633e+00 -7.26966202e-01 9.80620563e-01 -2.01655126e+00 4.36898060e-02 3.44616830e-01 1.15222126e-01 5.58234155e-01 -4.59958240e-02 -6.25360087e-02 -7.32681975e-02 -4.22898889e-01 -5.25265224e-02 -3.44237626e-01 -4.31508690e-01 2.37003993e-02 -7.98016042e-02 5.00548363e-01 -1.26597300e-01 7.20986426e-01 -1.02432096e+00 -9.02484775e-01 5.22553742e-01 1.05265856e-01 -6.28285855e-02 6.27357066e-01 2.55923957e-01 6.18839502e-01 -4.57381040e-01 6.45935118e-01 7.42694139e-01 -2.83173263e-01 1.54210702e-01 -4.54492122e-01 1.78977363e-02 -5.38138390e-01 -1.17170191e+00 1.37404931e+00 7.75088072e-02 3.74263257e-01 -3.47925514e-01 -1.31655753e+00 9.22137558e-01 1.37477994e-01 6.81379318e-01 -6.18724585e-01 3.80748627e-03 -8.73908177e-02 -3.42275053e-01 -6.93463683e-01 1.02168128e-01 8.38305503e-02 -8.86258408e-02 4.06778544e-01 1.11824319e-01 8.09900820e-01 1.62414491e-01 1.64773241e-01 6.10660970e-01 1.19056031e-01 1.47508994e-01 -6.26455992e-03 1.00476444e+00 1.36826918e-01 8.42343032e-01 7.07625389e-01 -6.02179170e-01 3.17605317e-01 -2.56092697e-01 -7.51210988e-01 -9.53998268e-01 -1.05299544e+00 1.35865086e-03 1.18290532e+00 6.02685750e-01 -1.21308580e-01 -7.92821825e-01 -1.16710305e+00 -7.65092596e-02 1.66560322e-01 -6.31397724e-01 -2.29537547e-01 -3.74070495e-01 -1.12597752e+00 5.35692871e-01 6.52741492e-01 1.24192119e+00 -6.29401445e-01 -1.20449178e-01 -3.67656089e-02 -3.92994881e-01 -1.15902388e+00 -7.02131927e-01 -3.77516747e-01 -6.77561104e-01 -1.35343075e+00 -9.31747556e-01 -1.14396155e+00 1.08300304e+00 7.88760543e-01 5.17720461e-01 1.42577797e-01 -1.24508314e-01 6.29885018e-01 -4.36061949e-01 -5.23733385e-02 -3.87087874e-02 -9.19666961e-02 5.63054919e-01 6.90389097e-01 8.16246688e-01 -2.20388874e-01 -5.46100795e-01 7.58601189e-01 -5.22019446e-01 1.05251029e-01 6.39115751e-01 1.06386173e+00 5.30539989e-01 5.45915604e-01 4.58562851e-01 -7.29728997e-01 1.44811258e-01 -4.71650124e-01 -6.37178957e-01 7.62153089e-01 -6.14551842e-01 -1.31252915e-01 4.82448041e-01 -7.42591441e-01 -1.44470191e+00 4.27149028e-01 4.43853199e-01 -4.06663984e-01 -3.73393178e-01 3.04578424e-01 -3.61352295e-01 -1.44162863e-01 4.53033596e-01 3.72722954e-01 -2.08828840e-02 -4.09359545e-01 2.40087554e-01 9.59902048e-01 8.37949872e-01 -4.76719916e-01 1.10857022e+00 4.74954963e-01 -2.88929433e-01 -6.12163424e-01 -9.75615680e-01 -8.00754607e-01 -1.10921502e+00 -5.67151189e-01 9.17942941e-01 -1.24370921e+00 -4.77788687e-01 9.45864797e-01 -9.12932575e-01 -5.12471274e-02 1.97824407e-02 7.91179001e-01 -1.66067213e-01 5.99031985e-01 -2.70260721e-01 -8.69516313e-01 -1.40947104e-01 -9.71487701e-01 8.56626332e-01 6.41880989e-01 2.63192296e-01 -8.99895251e-01 1.66591868e-01 6.11317039e-01 -1.71630597e-03 -2.21956313e-01 2.54334420e-01 -7.21564174e-01 -4.56015646e-01 -3.75844985e-01 -6.20088756e-01 3.67941886e-01 4.95991260e-01 -2.33459055e-01 -8.72056723e-01 -6.46719992e-01 -3.15281391e-01 -3.02256256e-01 8.98806036e-01 1.66121140e-01 9.66819406e-01 -3.29138011e-01 -7.91616261e-01 7.10553646e-01 1.14750063e+00 1.97081164e-01 2.62079507e-01 3.77959490e-01 1.04729712e+00 7.89550841e-01 8.79766405e-01 3.57588589e-01 7.87964225e-01 7.02767074e-01 2.44772211e-02 -1.52219068e-02 -1.34073272e-01 -5.80369771e-01 3.75599802e-01 6.83363855e-01 -2.71010548e-01 1.15993857e-01 -7.77145982e-01 4.42531914e-01 -2.17814779e+00 -1.21789479e+00 2.00055391e-01 2.34657073e+00 3.78178924e-01 -8.93092975e-02 4.66708660e-01 -8.64989534e-02 1.36247933e+00 -5.72237931e-02 -6.72697306e-01 7.52332330e-01 -2.81614941e-02 -6.64582133e-01 6.44501626e-01 3.36138755e-01 -1.63785946e+00 9.00219858e-01 6.11136103e+00 7.21080661e-01 -1.01616943e+00 1.30223572e-01 7.87028790e-01 2.40062852e-03 4.95464265e-01 -7.98732638e-02 -1.00890148e+00 9.05397892e-01 6.93281710e-01 -1.70501992e-01 3.64607275e-01 8.35570037e-01 1.52013749e-01 4.92918268e-02 -9.55684125e-01 1.46072757e+00 3.91120285e-01 -8.15127075e-01 7.98650160e-02 1.89462863e-02 8.37787092e-01 -1.62496895e-01 4.59854826e-02 2.58265704e-01 5.39794981e-01 -7.01672018e-01 4.18856740e-01 8.17666590e-01 7.31967568e-01 -7.94160366e-01 8.62859905e-01 4.97536331e-01 -1.66739702e+00 -4.13250655e-01 -4.62468594e-01 2.54526198e-01 3.98419984e-02 2.40401059e-01 -8.22147071e-01 6.03548586e-01 1.00674069e+00 1.26795673e+00 -9.82903242e-01 1.27039587e+00 3.92238982e-03 5.65096319e-01 -2.08531022e-01 1.73687711e-01 -1.26735479e-01 -2.87184387e-01 3.83408815e-01 1.06745923e+00 3.24459612e-01 2.64873803e-01 6.22810602e-01 2.65341073e-01 5.27195111e-02 -4.16969284e-02 -5.75806975e-01 3.01566601e-01 5.11559129e-01 1.16205859e+00 -5.14398813e-01 -6.42798245e-01 -6.09886408e-01 1.13846195e+00 4.75472838e-01 5.80331564e-01 -8.27735603e-01 1.31345261e-02 3.35333854e-01 -1.26159281e-01 1.68015003e-01 -4.16531935e-02 1.47535235e-01 -1.39436388e+00 2.19278112e-02 -6.82860553e-01 8.61451209e-01 -5.28800070e-01 -1.90367889e+00 5.44345438e-01 2.25907162e-01 -1.69948602e+00 -2.27308333e-01 -4.92703885e-01 -6.00805461e-01 6.78794265e-01 -1.33048248e+00 -1.62637115e+00 -6.94196224e-01 9.99142766e-01 3.84212524e-01 -9.13836062e-01 4.48891044e-01 5.15024722e-01 -8.52012455e-01 9.00115967e-01 1.25898227e-01 6.04676187e-01 9.54889476e-01 -8.26640904e-01 -7.38848522e-02 1.10107207e+00 4.64307666e-02 4.68883932e-01 2.64528394e-01 -7.63277829e-01 -1.08676493e+00 -1.50184572e+00 5.98786473e-01 -4.29444045e-01 2.66397566e-01 -1.86805893e-02 -6.72200441e-01 8.05105269e-01 -6.14883713e-02 2.33768195e-01 8.36916387e-01 3.98067124e-02 -4.10041749e-01 -6.52026713e-01 -1.17432213e+00 3.45221221e-01 1.16119277e+00 -5.31436086e-01 -7.40147948e-01 3.05108070e-01 2.85205305e-01 2.43462268e-02 -5.70485413e-01 4.95386392e-01 6.50390744e-01 -6.16418719e-01 1.11392021e+00 -7.47233570e-01 -1.47550911e-01 -7.46114969e-01 -1.50051475e-01 -1.25391781e+00 -6.10596359e-01 -6.69688731e-02 -3.96184400e-02 1.61825240e+00 8.90693590e-02 -7.10055530e-01 8.49565864e-01 7.23240733e-01 3.65523249e-01 -8.90900008e-03 -9.51554716e-01 -6.77413404e-01 -4.74893928e-01 -1.92497894e-01 6.00560129e-01 9.84246254e-01 -5.17636724e-02 3.71285349e-01 -8.70639026e-01 5.48786581e-01 1.22787213e+00 -8.44639353e-03 1.02045107e+00 -1.40964198e+00 -8.38389471e-02 -4.96989526e-02 -6.16370559e-01 -1.28919625e+00 2.44155109e-01 -7.79755294e-01 1.25354439e-01 -1.20301652e+00 8.24103236e-01 -6.49101377e-01 -5.99130988e-01 4.26756650e-01 -5.18728673e-01 3.84166867e-01 9.77511778e-02 6.80709004e-01 -1.08910477e+00 5.34541190e-01 8.08393121e-01 -7.41405547e-01 -1.61621526e-01 1.44979388e-01 -5.64637899e-01 6.93238318e-01 4.49071020e-01 -2.73982853e-01 -5.06415367e-01 -4.11368132e-01 -6.07270181e-01 -1.43303171e-01 5.83067298e-01 -1.35338819e+00 7.78872609e-01 -1.69708282e-01 8.85563850e-01 -7.12192059e-01 2.31663406e-01 -1.06006539e+00 1.34026095e-01 3.53163123e-01 -3.00010175e-01 -1.15753360e-01 -1.22651882e-01 9.37780857e-01 -1.59366727e-01 -6.44136593e-02 8.13500345e-01 -9.61554348e-02 -1.12246454e+00 7.08333731e-01 -1.70394987e-01 -2.60084301e-01 1.35109651e+00 -3.98178160e-01 -2.71565944e-01 -2.08008334e-01 -6.86690509e-01 5.19055068e-01 4.76796269e-01 6.09495759e-01 7.21428454e-01 -1.71453321e+00 -8.34446430e-01 3.47064734e-01 4.37089920e-01 -3.05602342e-01 5.82647026e-01 4.72498775e-01 -1.11862004e-01 3.43312651e-01 -3.24140906e-01 -8.92159164e-01 -1.61671829e+00 9.00432706e-01 4.09395188e-01 -1.58027589e-01 -2.31243223e-01 6.40249908e-01 2.03350559e-01 -5.49232066e-01 3.18295926e-01 5.20668387e-01 -5.07417262e-01 1.21448068e-02 8.11686456e-01 4.72977579e-01 -3.95111084e-01 -1.23548794e+00 -4.57861245e-01 8.19812298e-01 -2.61783779e-01 -2.43375674e-02 1.07137859e+00 -4.52389807e-01 -5.88103244e-03 6.63407296e-02 1.17934394e+00 -3.40847969e-01 -1.55979955e+00 -8.23874056e-01 -2.10318323e-02 -6.97085619e-01 -2.30142295e-01 -5.12778580e-01 -1.21607435e+00 5.65333962e-01 1.21159673e+00 -2.10753784e-01 1.17613947e+00 -1.54143542e-01 7.26188958e-01 4.85794842e-01 6.00998640e-01 -1.42597151e+00 3.30242902e-01 2.51519620e-01 1.63490579e-01 -1.71996713e+00 6.86753765e-02 -1.92692444e-01 -7.22991228e-01 8.59401286e-01 9.13347721e-01 1.03464149e-01 8.03650677e-01 -1.99072987e-01 1.12291202e-01 2.33882830e-01 -3.02169770e-01 -3.95647973e-01 4.76992518e-01 9.48429883e-01 -1.81694373e-01 1.64649934e-01 1.73068076e-01 4.93319333e-01 2.38277167e-01 1.29946604e-01 -1.80622786e-01 6.01702154e-01 -3.89744878e-01 -1.18628085e+00 -7.18741953e-01 5.40249228e-01 9.54705551e-02 2.91338712e-01 -3.21405977e-01 3.35213661e-01 5.54427505e-01 1.28306746e+00 1.86909910e-03 -9.78590429e-01 1.96287394e-01 -1.54283866e-01 2.92504489e-01 -3.97698790e-01 -1.22038364e-01 -2.00868949e-01 -1.77604854e-01 -1.71399996e-01 -9.46940720e-01 -7.96139061e-01 -7.27011085e-01 -2.48069525e-01 -2.33129516e-01 1.86285898e-01 2.16175467e-02 1.12466371e+00 2.04352751e-01 1.04382567e-01 9.92676556e-01 -8.89411032e-01 -3.42438459e-01 -9.66885865e-01 -4.13098365e-01 7.69926429e-01 2.27137372e-01 -9.41393495e-01 -1.34623438e-01 4.77012545e-01]
[14.775957107543945, 1.0399638414382935]
f2ba7e88-8505-42ae-8798-4628f70c5eca
the-disrpt-2021-shared-task-on-elementary
null
null
https://aclanthology.org/2021.disrpt-1.1
https://aclanthology.org/2021.disrpt-1.1.pdf
The DISRPT 2021 Shared Task on Elementary Discourse Unit Segmentation, Connective Detection, and Relation Classification
In 2021, we organized the second iteration of a shared task dedicated to the underlying units used in discourse parsing across formalisms: the DISRPT Shared Task (Discourse Relation Parsing and Treebanking). Adding to the 2019 tasks on Elementary Discourse Unit Segmentation and Connective Detection, this iteration of the Shared Task included for the first time a track on discourse relation classification across three formalisms: RST, SDRT, and PDTB. In this paper we review the data included in the Shared Task, which covers nearly 3 million manually annotated tokens from 16 datasets in 11 languages, survey and compare submitted systems and report on system performance on each task for both annotated and plain-tokenized versions of the data.
['Sonia Badene', 'Chloé Braud', 'Philippe Muller', 'Mikel Iruskieta', 'Yang Janet Liu', 'Amir Zeldes']
null
null
null
null
emnlp-disrpt-2021-11
['discourse-parsing', 'connective-detection', 'relation-classification']
['natural-language-processing', 'natural-language-processing', 'natural-language-processing']
[ 4.28970724e-01 1.12109971e+00 -3.97425234e-01 -3.45124513e-01 -1.36305308e+00 -1.00272036e+00 9.14322734e-01 4.86650795e-01 -4.53505307e-01 1.32622766e+00 8.28275919e-01 -7.03306615e-01 1.09674893e-01 -6.84370279e-01 -4.08454508e-01 -2.47478038e-01 -1.15668893e-01 8.09972048e-01 7.06341982e-01 -5.35340726e-01 1.85896620e-01 -3.15785736e-01 -7.15317488e-01 1.28769147e+00 9.40253377e-01 6.87740088e-01 1.21520109e-01 1.20413899e+00 -4.51492846e-01 1.67267394e+00 -1.23240936e+00 -6.21015847e-01 -6.08185112e-01 -5.10300040e-01 -1.96133697e+00 -1.58994123e-01 2.82123983e-01 -4.57741730e-02 -1.60450965e-01 5.99654436e-01 3.80345553e-01 1.68151453e-01 5.84776819e-01 -5.74957669e-01 -5.35636604e-01 1.49764717e+00 -7.59725794e-02 7.16749430e-01 1.10397983e+00 -2.62187034e-01 1.57666063e+00 -4.57844317e-01 1.52772009e+00 1.51401305e+00 6.20638907e-01 4.47013527e-01 -1.09872139e+00 1.57711543e-02 2.29600638e-01 2.64987886e-01 -7.81670332e-01 -6.22271419e-01 2.43076503e-01 -5.43747067e-01 1.88224161e+00 5.88281333e-01 3.29531282e-01 1.25373530e+00 -2.71282464e-01 1.08508217e+00 1.07650793e+00 -7.50740528e-01 -2.10473254e-01 -1.44451112e-01 6.27929866e-01 4.67229843e-01 5.95381521e-02 -5.43771088e-01 -6.84061766e-01 -5.52273244e-02 2.50405908e-01 -1.62859392e+00 -3.55037332e-01 5.58217883e-01 -1.47843874e+00 7.02302635e-01 -1.18870378e-01 8.82617772e-01 2.66569972e-01 -4.29476202e-01 1.14823675e+00 3.12084466e-01 7.42547691e-01 5.89157939e-01 -5.77126503e-01 -5.62059283e-01 -3.48756552e-01 5.02948165e-01 1.38559020e+00 1.08005571e+00 3.57518077e-01 -3.42550993e-01 -5.78200042e-01 9.05502081e-01 1.16500683e-01 -1.34404957e-01 4.98168528e-01 -1.19360447e+00 1.35859048e+00 5.94652295e-01 -4.43411581e-02 -6.55105174e-01 -5.66479623e-01 5.47317326e-01 -2.19947755e-01 -4.75195289e-01 8.83265018e-01 -6.40009463e-01 -3.17939579e-01 1.29856431e+00 3.17823380e-01 -6.14732981e-01 5.56441128e-01 3.99616539e-01 1.72472596e+00 6.56873107e-01 2.99348801e-01 -4.86706614e-01 1.75497711e+00 -1.07379818e+00 -1.32364821e+00 -1.48472399e-01 1.36326301e+00 -9.82263267e-01 5.54480374e-01 1.24488011e-01 -1.38870597e+00 -2.09959388e-01 -1.04413843e+00 -7.79354215e-01 -3.72262806e-01 -2.19242603e-01 6.55065417e-01 5.25442719e-01 -7.36760318e-01 4.40720528e-01 -6.64727688e-01 -2.10568219e-01 2.16921657e-01 -1.78989574e-01 -2.38945544e-01 4.11879331e-01 -1.54480314e+00 1.47029829e+00 1.16437316e+00 -3.04631650e-01 -2.52944797e-01 -4.48109418e-01 -1.11331749e+00 -3.67316991e-01 6.65330172e-01 -2.81697214e-01 1.90031767e+00 -4.81151521e-01 -1.77894723e+00 1.54429770e+00 -1.57793790e-01 -7.77543068e-01 3.96684170e-01 -6.44900501e-01 -2.95432717e-01 -1.53401494e-02 2.04192236e-01 1.57512680e-01 -1.11440949e-01 -8.29720020e-01 -9.73987997e-01 2.00852185e-01 2.71373838e-01 4.30183887e-01 3.57354879e-01 9.33772087e-01 2.47041121e-01 -5.92530787e-01 -8.35350454e-02 -3.24380815e-01 2.20596373e-01 -9.64915991e-01 -7.60051966e-01 -1.00467575e+00 8.04021716e-01 -1.16797113e+00 1.38341773e+00 -1.75649774e+00 5.45780182e-01 -6.33821905e-01 1.36027738e-01 2.65369564e-01 3.79750580e-01 5.54633796e-01 -9.98077914e-02 2.74477214e-01 -3.85166436e-01 -4.94318962e-01 1.54036611e-01 6.07945979e-01 -3.16670299e-01 7.62850419e-02 3.86894166e-01 1.27543473e+00 -1.06946588e+00 -9.19316411e-01 8.88853222e-02 -2.82921314e-01 -5.39564043e-02 1.84396043e-01 -7.37823904e-01 5.52244902e-01 -3.16248566e-01 5.24354875e-01 -5.46070747e-03 -1.48954928e-01 8.96046340e-01 3.65279652e-02 -4.45155472e-01 1.39195502e+00 -8.56215477e-01 1.74204040e+00 -3.56902689e-04 1.05134857e+00 2.87947625e-01 -8.92084539e-01 6.21883273e-01 1.15124214e+00 2.76819557e-01 -3.45344901e-01 3.80332679e-01 4.30649608e-01 3.84891838e-01 -5.70824623e-01 1.01405656e+00 1.85252011e-01 -7.11347044e-01 5.89633286e-01 4.45845664e-01 -1.50523409e-01 5.88427126e-01 3.33169043e-01 1.15200853e+00 3.16708416e-01 9.89160776e-01 -3.78309608e-01 7.10640669e-01 4.89062726e-01 4.84578550e-01 3.29315752e-01 -2.50041634e-01 3.47983778e-01 9.63358164e-01 -2.96648651e-01 -8.02510798e-01 -7.01964140e-01 -4.68249202e-01 1.48340440e+00 -4.07266587e-01 -8.43463898e-01 -7.32402802e-01 -5.18726587e-01 -3.58750045e-01 6.38754606e-01 -5.67775607e-01 6.80904567e-01 -1.47388399e+00 -8.12052190e-01 1.40895760e+00 4.46250319e-01 8.25319588e-01 -1.32765472e+00 -6.18521571e-01 3.81720811e-01 -9.09860313e-01 -1.61397195e+00 1.17900319e-01 3.55364323e-01 -2.12250501e-01 -1.53957689e+00 -3.37367326e-01 -1.15862095e+00 -2.84317434e-01 -5.11346698e-01 1.41056120e+00 -2.22451717e-01 8.66296440e-02 -4.39553522e-02 -6.59617662e-01 -2.81206757e-01 -8.06682289e-01 4.19834375e-01 -8.02469790e-01 -7.03381360e-01 2.40639850e-01 7.02739283e-02 4.34606671e-01 -3.99299711e-01 -1.46288380e-01 1.88251927e-01 -2.43654564e-01 7.37495542e-01 2.66838446e-02 -7.48079062e-01 3.75143856e-01 -1.42635584e+00 9.72066879e-01 -3.21806610e-01 -2.04709932e-01 4.97526973e-01 4.06964362e-01 -1.89450264e-01 -1.61237732e-01 7.82407820e-02 -1.69053972e+00 -5.14600098e-01 -2.52285779e-01 8.70984018e-01 -6.51915073e-02 7.10288405e-01 -3.16881269e-01 5.36000073e-01 6.48689151e-01 -6.37200415e-01 -3.67295474e-01 -5.11801600e-01 7.04619288e-01 5.73657751e-01 9.16025281e-01 -1.07197464e+00 2.82401312e-02 -2.89976537e-01 -3.22852433e-01 -9.64518309e-01 -1.21528804e+00 -2.81703144e-01 -8.66690040e-01 8.04875866e-02 1.65334296e+00 -7.58964241e-01 -6.43601835e-01 5.05128145e-01 -1.96570778e+00 -8.06029320e-01 -5.03683090e-01 1.07007653e-01 -6.95248187e-01 4.63457525e-01 -1.41955554e+00 -7.90335119e-01 -2.70962656e-01 -7.72123754e-01 7.74672151e-01 3.21766571e-03 -9.76598322e-01 -1.45658815e+00 4.26390231e-01 4.16365862e-01 -2.04690278e-01 8.46526742e-01 1.09277952e+00 -9.90997612e-01 -3.09377074e-01 5.92355847e-01 -2.56516814e-01 9.07952636e-02 2.33901083e-01 1.34050608e-01 -9.58048105e-01 2.41390392e-01 -2.02588528e-01 -8.02750826e-01 8.20903063e-01 3.12435508e-01 3.84009540e-01 -3.02910239e-01 -2.65029430e-01 2.14450672e-01 7.52962649e-01 1.78870901e-01 4.93092000e-01 8.39445233e-01 5.28290331e-01 9.77091074e-01 6.70552015e-01 1.11295909e-01 6.93259120e-01 5.30552506e-01 -1.22027487e-01 3.33697468e-01 -4.41557407e-01 2.27045462e-01 3.56982648e-01 1.15738046e+00 -5.11281848e-01 -4.53565866e-01 -1.25678408e+00 7.25479186e-01 -2.10770679e+00 -8.36242795e-01 -7.07261622e-01 1.36654127e+00 1.52165842e+00 2.82577395e-01 2.15271533e-01 2.15284288e-01 7.07144737e-01 5.87954760e-01 1.77730188e-01 -1.00675941e+00 -7.02874124e-01 4.56563830e-01 1.01292714e-01 1.05990112e+00 -1.55211592e+00 1.33248818e+00 7.62785578e+00 5.35679698e-01 -1.97921753e-01 4.69422698e-01 4.77515578e-01 2.46298075e-01 -2.29238179e-02 1.36660963e-01 -8.85110736e-01 1.77150164e-02 1.28474808e+00 -5.65184951e-01 -1.57088041e-01 5.32451868e-01 -5.58992624e-01 -4.84546542e-01 -1.24097717e+00 1.38595521e-01 -1.75021067e-01 -1.73621356e+00 -3.81784379e-01 -2.32318401e-01 7.01865911e-01 5.31505831e-02 -4.62157577e-01 5.42643905e-01 8.67001355e-01 -1.13164318e+00 7.02365637e-01 4.58478928e-03 7.60166526e-01 -3.36007744e-01 9.13059950e-01 1.60201311e-01 -1.22984827e+00 4.30014431e-01 1.46228015e-01 -6.06745780e-01 7.11446524e-01 5.02021164e-02 -9.50908363e-01 9.22233999e-01 4.72770810e-01 7.39985585e-01 -2.68180341e-01 2.70755619e-01 -6.89566731e-01 6.12937808e-01 -1.19499952e-01 -6.92459196e-02 2.66649395e-01 2.28291541e-01 1.02035785e+00 1.84397554e+00 -5.82114935e-01 3.89253765e-01 4.03653741e-01 2.53385246e-01 -5.29924557e-02 4.36413735e-02 -3.34146559e-01 5.01173884e-02 6.88402057e-01 9.74289894e-01 -7.59432554e-01 -7.10587025e-01 -2.44275257e-01 4.61902410e-01 5.19961119e-01 -3.20931431e-03 -5.71875215e-01 -4.42080736e-01 3.65225673e-01 -5.67498624e-01 3.24665904e-01 -3.44581753e-01 -4.69593793e-01 -7.19555378e-01 -8.79696906e-02 -8.19369793e-01 8.55662048e-01 -2.90833443e-01 -1.24144721e+00 7.67001867e-01 4.84297752e-01 -4.19423014e-01 -7.11049736e-01 -8.60274017e-01 -5.12574732e-01 9.30443645e-01 -1.27068603e+00 -1.12468612e+00 1.18512236e-01 4.30161327e-01 5.10270834e-01 -9.85259097e-03 1.34813261e+00 2.36607954e-01 -5.07045746e-01 3.26207787e-01 -2.12135971e-01 7.78995931e-01 1.01924765e+00 -1.72364795e+00 7.82600641e-01 4.98671532e-01 -6.08237311e-02 3.72339338e-01 7.20724583e-01 -7.44222820e-01 -5.03962219e-01 -7.19806373e-01 1.53590477e+00 -7.71876574e-01 1.22547054e+00 -2.74257779e-01 -1.13194537e+00 1.35827053e+00 1.06203794e+00 -4.37206477e-01 7.98237503e-01 7.19432056e-01 -1.67973742e-01 8.43262434e-01 -9.26723242e-01 1.83310986e-01 1.08709335e+00 -6.52620375e-01 -1.48092771e+00 7.12168396e-01 1.26936138e+00 -1.42952025e+00 -1.41683722e+00 2.48653859e-01 1.81442529e-01 -6.92347288e-01 4.84863609e-01 -8.19248080e-01 6.95194542e-01 3.15060675e-01 -4.19302046e-01 -8.48337471e-01 1.22345634e-01 -1.01821423e+00 -4.12276864e-01 1.63112855e+00 7.39523768e-01 -6.90945089e-01 3.67159128e-01 3.18755776e-01 -8.53883803e-01 -6.26719475e-01 -1.47214389e+00 -4.60529864e-01 6.07560813e-01 -2.66170412e-01 2.19981536e-01 1.04150617e+00 8.68987322e-01 9.01855886e-01 2.80227810e-01 -5.25695086e-01 9.34488326e-02 4.14955243e-02 3.88619781e-01 -1.04813528e+00 -3.46736848e-01 -5.15342712e-01 -2.60156263e-02 -1.07371438e+00 2.73894161e-01 -1.01669157e+00 9.70699638e-02 -1.86057830e+00 -2.20062464e-01 -3.52891117e-01 3.66194606e-01 5.26101112e-01 -1.30302384e-01 -1.54272601e-01 3.42894793e-01 1.30515739e-01 -9.47544098e-01 2.35590190e-01 1.21360266e+00 -9.48638394e-02 -3.11360717e-01 -2.73895621e-01 -5.04763186e-01 1.04523325e+00 5.74738503e-01 -1.84054643e-01 1.00179538e-01 -7.22845078e-01 1.36343077e-01 2.44364947e-01 -4.02498916e-02 -5.16698062e-01 1.84035525e-02 -1.06522173e-01 -2.11804330e-01 -9.49155927e-01 4.00673330e-01 2.84058928e-01 -6.02831423e-01 2.37544000e-01 -4.41132993e-01 2.64609456e-02 4.05391395e-01 -1.20664917e-01 -4.22957450e-01 -3.62551183e-01 2.04768211e-01 -4.42839205e-01 -7.52717257e-01 -6.49918079e-01 -9.61453974e-01 8.58828783e-01 1.15609586e+00 -1.65228680e-01 -1.13201714e+00 1.11917838e-01 -1.08281624e+00 3.56681287e-01 -2.18128756e-01 2.51875758e-01 -3.52960229e-02 -9.24449921e-01 -1.09107041e+00 -6.81618154e-01 -3.27087253e-01 5.98970652e-01 -2.17664078e-01 6.20438337e-01 -5.51957369e-01 1.00199151e+00 -2.73491070e-02 -4.43152964e-01 -1.58038878e+00 6.27442747e-02 1.37097538e-01 -1.03698397e+00 -6.47238076e-01 1.14941192e+00 -3.68334472e-01 -3.75338048e-01 2.97739413e-02 -6.77540958e-01 -6.57902181e-01 4.79006082e-01 4.51249242e-01 4.51771408e-01 1.67642146e-01 -8.34459364e-01 -4.39882547e-01 -1.52694881e-01 -2.35385448e-01 -4.83933330e-01 1.07433224e+00 -7.12942854e-02 -6.21138990e-01 1.18792224e+00 6.81543112e-01 1.50448740e-01 -6.83276296e-01 -3.90011936e-01 9.47377741e-01 2.17781872e-01 -4.43895072e-01 -9.06108201e-01 1.22914821e-01 3.58642697e-01 -5.54056108e-01 6.75434947e-01 4.31615919e-01 4.10506576e-01 8.02866995e-01 4.59137499e-01 3.22104782e-01 -1.37146676e+00 -8.99238884e-02 1.52233481e+00 9.76193547e-01 -1.01769936e+00 2.20819771e-01 -9.42709625e-01 -7.88412929e-01 9.30711329e-01 6.55420005e-01 -6.70143515e-02 3.02519143e-01 4.32317108e-01 1.77377071e-02 -2.59418905e-01 -8.50671053e-01 -4.79082502e-02 1.65478840e-01 8.40321660e-01 1.26615691e+00 3.35662425e-01 -7.29305804e-01 7.93762445e-01 -7.07458615e-01 -6.89727426e-01 4.84743923e-01 1.36011207e+00 -5.96749842e-01 -1.54589415e+00 -3.02968293e-01 8.55552778e-02 -5.46686888e-01 -4.30223346e-02 -9.01185751e-01 1.22159982e+00 1.87388971e-01 1.13559246e+00 2.30706483e-01 1.08157158e-01 3.47486496e-01 3.82198662e-01 8.61315429e-01 -1.26610315e+00 -1.04783249e+00 -2.27353781e-01 1.62117636e+00 -2.72001326e-01 -9.69613612e-01 -1.12810421e+00 -1.46064460e+00 -3.23990405e-01 -2.26590216e-01 2.51717716e-01 -7.95248598e-02 1.33437371e+00 -3.38455409e-01 1.13497257e+00 -4.03010070e-01 -7.20920861e-01 -1.60901859e-01 -1.54266298e+00 -1.10644974e-01 -5.68584539e-02 2.34066218e-01 -5.42653680e-01 5.88562228e-02 2.10225046e-01]
[10.821043968200684, 9.458447456359863]
c62f5052-a09d-4092-bb90-46ef7ba42922
distributionally-robust-end-to-end-portfolio
2206.05134
null
https://arxiv.org/abs/2206.05134v1
https://arxiv.org/pdf/2206.05134v1.pdf
Distributionally Robust End-to-End Portfolio Construction
We propose an end-to-end distributionally robust system for portfolio construction that integrates the asset return prediction model with a distributionally robust portfolio optimization model. We also show how to learn the risk-tolerance parameter and the degree of robustness directly from data. End-to-end systems have an advantage in that information can be communicated between the prediction and decision layers during training, allowing the parameters to be trained for the final task rather than solely for predictive performance. However, existing end-to-end systems are not able to quantify and correct for the impact of model risk on the decision layer. Our proposed distributionally robust end-to-end portfolio selection system explicitly accounts for the impact of model risk. The decision layer chooses portfolios by solving a minimax problem where the distribution of the asset returns is assumed to belong to an ambiguity set centered around a nominal distribution. Using convex duality, we recast the minimax problem in a form that allows for efficient training of the end-to-end system.
['Garud N. Iyengar', 'Giorgio Costa']
2022-06-10
null
null
null
null
['portfolio-optimization']
['time-series']
[-2.89605737e-01 2.00985685e-01 -1.38138101e-01 -7.04067111e-01 -1.30232513e+00 -1.00643110e+00 3.42352957e-01 5.54221161e-02 -3.74470890e-01 4.96904641e-01 2.35611781e-01 -5.71055949e-01 -8.08778226e-01 -8.72220397e-01 -6.25133514e-01 -5.59287667e-01 -8.73360857e-02 7.51065910e-01 -4.53871101e-01 1.72793612e-01 1.39496952e-01 2.49135673e-01 -1.03749609e+00 4.52061556e-02 7.17190504e-01 1.47488761e+00 -1.82058379e-01 5.87823451e-01 3.20147723e-01 6.18832767e-01 -3.73580307e-01 -7.49020219e-01 1.00978434e+00 -1.68245248e-02 -6.74018711e-02 -1.71430141e-01 3.01297277e-01 -3.80091250e-01 2.11121533e-02 1.15031540e+00 7.44361699e-01 1.61121011e-01 9.17887092e-01 -1.15337503e+00 -3.10235471e-01 6.01685703e-01 -2.91960895e-01 -4.78997864e-02 -2.21775904e-01 5.11401407e-02 1.56289625e+00 -8.84082973e-01 2.21283183e-01 1.13573468e+00 6.43212438e-01 1.19911551e-01 -1.13126743e+00 -4.37849313e-01 2.29534000e-01 -4.53923345e-01 -9.43923712e-01 -4.16492134e-01 6.65553331e-01 -9.35559928e-01 5.71017921e-01 2.68798858e-01 2.12436199e-01 5.79210639e-01 1.74994513e-01 5.39131045e-01 8.32134783e-01 -2.07671136e-01 4.57012087e-01 2.97441691e-01 -1.26671866e-02 2.80180573e-01 2.44667724e-01 6.20325625e-01 -8.57083276e-02 -5.51464796e-01 4.41426814e-01 2.66153254e-02 -7.94247463e-02 -5.81554353e-01 -7.73533106e-01 9.92247283e-01 1.76210448e-01 -4.19036984e-01 -6.06890142e-01 2.74052143e-01 3.52785468e-01 3.96202236e-01 6.83006406e-01 4.94239181e-01 -5.99146962e-01 -4.31080908e-02 -1.13737249e+00 5.97124398e-01 9.70908940e-01 7.05518723e-01 2.14320436e-01 -4.41664346e-02 -6.30658090e-01 5.98950088e-01 6.84487164e-01 5.26008427e-01 1.33288484e-02 -1.11432314e+00 9.49247897e-01 2.46972919e-01 5.25445998e-01 -6.19492650e-01 -1.52498513e-01 -8.61686528e-01 -3.23946238e-01 5.80675244e-01 5.77208519e-01 -6.57792091e-01 -3.80894095e-01 1.90019333e+00 2.44204029e-01 -2.98221712e-03 1.95414484e-01 8.79000127e-01 -1.95060849e-01 3.95437598e-01 -4.52656969e-02 -2.11727843e-01 1.02428377e+00 -6.31431520e-01 -3.39606792e-01 2.29036273e-03 4.61128950e-01 -5.83674967e-01 8.68584096e-01 4.01880205e-01 -1.25092471e+00 -7.46662468e-02 -9.08908725e-01 3.83315057e-01 2.01147813e-02 1.15036324e-01 1.04406744e-01 7.92652071e-01 -6.42944634e-01 8.11243832e-01 -4.67169315e-01 5.28831184e-01 4.93257493e-01 4.52734590e-01 7.37225041e-02 4.19322580e-01 -1.13755417e+00 9.15758312e-01 3.17212522e-01 -1.31653538e-02 -8.42491388e-01 -1.11106277e+00 -4.71263707e-01 3.63633335e-01 4.52477604e-01 -9.30367887e-01 1.48405766e+00 -1.19799531e+00 -1.49808443e+00 2.77110875e-01 5.93734682e-01 -7.61413038e-01 1.13177943e+00 -2.88339198e-01 4.75375466e-02 -2.22061351e-01 2.93425377e-02 -1.74211323e-01 8.80835414e-01 -9.29894209e-01 -7.45810449e-01 -1.86544031e-01 6.42008614e-03 2.44440183e-01 -9.27934721e-02 -3.08713634e-02 -7.64303654e-02 -1.10973990e+00 -5.10890782e-01 -7.81988263e-01 -1.70278534e-01 1.00113027e-01 -3.16932082e-01 1.30149171e-01 3.54248613e-01 -8.99934053e-01 1.24066544e+00 -1.99720025e+00 -2.61515766e-01 6.97015464e-01 -2.02854395e-01 5.57778496e-03 -3.02029364e-02 3.56994599e-01 -2.62382895e-01 3.13306116e-02 -3.89507353e-01 -5.32331109e-01 7.65275419e-01 -2.08607376e-01 -8.13893735e-01 6.03354752e-01 1.28195152e-01 6.01889253e-01 -7.31666207e-01 5.48834167e-02 2.31759325e-01 1.92283645e-01 -7.90150821e-01 5.81657648e-01 -4.66024280e-01 1.32048707e-02 -5.76422691e-01 2.91182429e-01 5.25754631e-01 1.86184317e-01 2.96177175e-02 1.98331714e-01 3.01112309e-02 2.76552141e-01 -1.36287081e+00 9.10556912e-01 -6.38950765e-01 5.06283566e-02 1.41046658e-01 -8.47458482e-01 8.77793312e-01 1.85208932e-01 6.69439435e-01 -4.14176166e-01 2.70919017e-02 2.27747992e-01 -1.88625485e-01 -1.86102949e-02 1.60004452e-01 -6.88930511e-01 -3.65109742e-01 6.37531698e-01 -2.46728599e-01 2.15293929e-01 -1.63715661e-01 -3.18863600e-01 6.78892255e-01 1.21734448e-01 1.02952249e-01 -5.06407857e-01 2.25523099e-01 -3.36248547e-01 7.46429265e-01 6.27342165e-01 1.69489399e-01 7.56706953e-01 8.25684071e-01 -1.46919921e-01 -1.00807917e+00 -1.49428499e+00 -1.93981603e-01 9.79384005e-01 -4.97165084e-01 1.60905570e-01 -6.23337746e-01 -7.87775040e-01 6.00350797e-01 1.14369941e+00 -4.78575975e-01 -2.45915651e-02 -1.98498502e-01 -6.81416094e-01 1.50287449e-01 4.86926138e-01 1.65039971e-01 -5.95399737e-01 -4.74180847e-01 3.72968584e-01 2.92343974e-01 -5.20100236e-01 -9.23940897e-01 1.95573330e-01 -6.83312714e-01 -9.47642803e-01 -8.01259100e-01 -2.46323019e-01 5.71171820e-01 -5.60063779e-01 1.20426631e+00 -5.37692845e-01 1.35675386e-01 5.61255813e-01 2.12441549e-01 -7.35851824e-01 -1.60775721e-01 -2.00185388e-01 3.89804840e-02 3.04573447e-01 -2.20790520e-01 -4.12280381e-01 -7.25698233e-01 3.94458860e-01 -5.70116401e-01 -4.40614611e-01 2.49961540e-01 7.59500265e-01 9.03840184e-01 1.11822926e-01 1.14018214e+00 -1.02017546e+00 9.10553217e-01 -6.73861325e-01 -1.23881114e+00 5.52771091e-01 -9.25188482e-01 3.74716163e-01 6.05384588e-01 -2.98536509e-01 -1.06772506e+00 -9.43608582e-03 1.00924090e-01 -4.66269135e-01 4.94917810e-01 7.27694809e-01 -4.65019107e-01 3.70071411e-01 3.32057476e-01 -4.05473977e-01 -2.74800118e-02 -4.42364842e-01 5.38840473e-01 6.70950174e-01 5.26146829e-01 -9.39030409e-01 7.23370790e-01 -8.32122844e-03 1.19169382e-02 1.59651041e-02 -1.12475002e+00 -1.99072435e-01 -3.41946006e-01 2.44423598e-02 4.34842378e-01 -1.09350932e+00 -6.67589962e-01 1.51266545e-01 -6.72054887e-01 -4.40250665e-01 -9.37964857e-01 4.06495333e-01 -9.09774423e-01 -1.82303134e-02 8.32130143e-04 -1.23464859e+00 -7.40772426e-01 -7.74456918e-01 8.00665498e-01 -1.25533223e-01 -1.24242060e-01 -1.45921171e+00 2.32835546e-01 2.33461648e-01 3.51546347e-01 5.26262462e-01 9.35051501e-01 -9.07523990e-01 -3.86928320e-01 -5.47597408e-01 -1.51277736e-01 7.17218876e-01 -7.85499215e-02 -1.46240279e-01 -9.45110440e-01 -3.78718853e-01 4.08133894e-01 -2.41819303e-02 7.75103927e-01 6.97037280e-01 1.04962122e+00 -8.57430577e-01 2.91785479e-01 7.31848061e-01 1.38476694e+00 -2.32627410e-02 2.45650038e-01 4.03793573e-01 3.32198501e-01 8.46459329e-01 8.04208100e-01 7.23404169e-01 4.29928273e-01 5.55981278e-01 3.41117829e-01 3.14815789e-01 4.65916455e-01 -2.92051911e-01 5.70979893e-01 1.51595414e-01 5.24029255e-01 -2.67527342e-01 -8.97480905e-01 4.67081010e-01 -2.14104557e+00 -1.01079857e+00 5.25468946e-01 2.77560830e+00 7.58340001e-01 3.36633891e-01 4.70637441e-01 -2.77635843e-01 4.77281839e-01 2.36265734e-03 -8.28239977e-01 -4.49846178e-01 1.46558866e-01 -2.03836218e-01 9.30120885e-01 6.95741534e-01 -1.25793183e+00 1.63340107e-01 6.37488937e+00 7.36040711e-01 -9.30917680e-01 -2.18251497e-01 1.01023829e+00 -5.36619961e-01 -6.83646381e-01 1.00032344e-01 -7.95305550e-01 6.95963204e-01 1.05831456e+00 -6.73831344e-01 3.89107168e-01 1.01759303e+00 3.89225841e-01 2.87504077e-01 -1.33139884e+00 6.89671338e-01 -4.38234389e-01 -1.15540349e+00 -1.80405825e-01 3.28068823e-01 7.13552117e-01 -5.11910208e-02 5.66945553e-01 2.81174839e-01 4.44846123e-01 -1.10414779e+00 1.33856797e+00 1.12250113e+00 6.22325182e-01 -1.32472420e+00 8.55353773e-01 3.77526283e-01 -8.37063193e-01 -4.70534712e-01 -3.16282600e-01 1.66942161e-02 1.91477165e-01 9.76792634e-01 -7.70488024e-01 3.83123904e-01 8.56100097e-02 4.17180955e-01 -8.63626450e-02 1.11054575e+00 3.64384763e-02 5.22429407e-01 -4.68936026e-01 2.80464798e-01 8.07933062e-02 -4.59746480e-01 5.11638165e-01 1.05103230e+00 6.90240562e-01 -1.94768995e-01 3.25576752e-01 1.08789563e+00 -2.85084695e-01 2.08274852e-02 -4.23073441e-01 -3.03977132e-02 4.10829335e-01 9.28131402e-01 -1.31855682e-01 5.27886786e-02 -3.00853878e-01 3.07206213e-01 1.86774895e-01 2.81801641e-01 -7.46115863e-01 -2.89724618e-01 9.22164679e-01 1.98221445e-01 5.09315431e-01 6.92754164e-02 -7.16652334e-01 -9.38405097e-01 4.11087990e-01 -8.14018488e-01 7.04860747e-01 -2.21318275e-01 -1.80007815e+00 1.22570187e-01 3.39723416e-02 -1.20031583e+00 -6.14926517e-01 -6.99576080e-01 -1.02323878e+00 1.27773046e+00 -1.51476324e+00 -7.33784914e-01 5.13600767e-01 3.01824749e-01 -1.84774801e-01 -5.90097308e-01 5.38023710e-01 1.38775870e-01 -5.55196047e-01 8.33864629e-01 7.12092757e-01 2.14505047e-01 6.12641513e-01 -1.62894869e+00 2.60204703e-01 7.81495214e-01 -2.11217538e-01 3.92382830e-01 6.90216601e-01 -4.44705516e-01 -1.03859723e+00 -1.52578449e+00 6.17183745e-01 -6.37275457e-01 8.65948141e-01 -1.98771551e-01 -5.31668007e-01 4.47867334e-01 -2.58996904e-01 1.33157343e-01 7.41892099e-01 1.47698134e-01 -2.84453928e-01 -4.37028706e-01 -1.33666945e+00 3.13933134e-01 3.78293216e-01 -5.60423136e-01 -4.50929463e-01 2.12422445e-01 5.61126232e-01 -3.10326666e-01 -1.06851828e+00 1.36099935e-01 5.61458409e-01 -6.00059390e-01 1.07956529e+00 -5.98462045e-01 1.70839980e-01 -1.89521343e-01 -4.45036262e-01 -1.29543960e+00 -1.09548561e-01 -1.00627804e+00 -2.39365935e-01 1.49304533e+00 6.47262096e-01 -8.45355928e-01 6.23137236e-01 1.16881990e+00 1.41251117e-01 -8.73497367e-01 -1.09159839e+00 -8.45437288e-01 6.01773858e-01 -5.80935538e-01 7.54557014e-01 5.25608957e-01 -2.98697442e-01 1.83347128e-02 -4.02048498e-01 3.34927112e-01 1.10115063e+00 5.34265101e-01 4.03011829e-01 -1.23225081e+00 -7.29357779e-01 -6.67409360e-01 2.88613662e-02 -5.84239781e-01 2.39312351e-01 -9.55114484e-01 7.22414777e-02 -1.05227017e+00 -4.62041833e-02 -7.49822736e-01 -5.81885755e-01 3.35756004e-01 -1.56989574e-01 -5.88317513e-01 4.88986701e-01 6.22935826e-03 -4.04523075e-01 7.16726005e-01 8.83406222e-01 3.65350731e-02 -1.98334694e-01 5.90175807e-01 -1.24647307e+00 6.86728179e-01 7.41221607e-01 -6.90199852e-01 -6.31286323e-01 -2.62511730e-01 4.57069486e-01 4.32876319e-01 3.93810868e-01 -4.20365959e-01 2.84435898e-02 -5.31274855e-01 2.23606229e-01 -4.21504050e-01 -6.19983301e-02 -8.04027677e-01 1.27405420e-01 7.91867450e-02 -6.27066672e-01 -7.16479272e-02 1.04665244e-02 6.83582544e-01 -7.40587711e-02 -2.91999042e-01 9.49246585e-01 2.94346064e-01 1.56504765e-01 4.93657261e-01 3.53709869e-02 4.83721286e-01 1.07599318e+00 3.04408938e-01 8.64927843e-03 -6.28907561e-01 -7.12146342e-01 8.06386054e-01 3.57219100e-01 1.99478358e-01 3.62119555e-01 -1.47149372e+00 -9.76093113e-01 1.36360526e-01 -2.00739890e-01 6.62722513e-02 -1.24488063e-01 3.14049810e-01 -1.95728377e-01 1.90704614e-01 1.93838730e-01 -1.06259301e-01 -6.69189513e-01 3.97700638e-01 8.91725183e-01 -5.30069888e-01 -2.26089835e-01 6.79142773e-01 3.38564426e-01 -6.06742144e-01 5.51041842e-01 -3.95686865e-01 2.10309356e-01 2.30256766e-01 4.68456328e-01 4.16232318e-01 5.56132793e-02 -2.41985977e-01 -2.55741030e-01 3.66142601e-01 2.11350009e-01 -4.24434066e-01 1.66636324e+00 -1.00975752e-01 2.20754087e-01 1.85910344e-01 1.24816263e+00 -7.91769475e-02 -1.87097180e+00 -2.78844148e-01 2.55049825e-01 -4.86455619e-01 2.63168216e-01 -1.08307421e+00 -1.14459324e+00 8.13131690e-01 5.97162724e-01 -1.12676069e-01 1.05094492e+00 -4.35184360e-01 6.09424055e-01 3.68179291e-01 1.78367823e-01 -1.40480292e+00 -3.06629658e-01 5.24262547e-01 1.14835823e+00 -1.05009758e+00 1.07212573e-01 2.27657869e-01 -9.86953497e-01 1.11451817e+00 -7.80438539e-03 -5.52187026e-01 1.19940746e+00 5.25252461e-01 1.29114747e-01 1.96888611e-01 -9.32876885e-01 2.52972454e-01 7.84577310e-01 4.79097724e-01 1.14599936e-01 4.08471882e-01 1.99031711e-01 1.40661681e+00 -4.06377286e-01 -3.58460903e-01 2.00207844e-01 6.45858407e-01 -2.69416302e-01 -1.10842419e+00 -3.26869279e-01 5.78253806e-01 -7.88346887e-01 3.89489830e-02 -6.46704286e-02 2.28064507e-01 -6.74741119e-02 7.38644004e-01 1.34161815e-01 5.47161251e-02 7.06812263e-01 1.04437783e-01 1.18638158e-01 -4.88228738e-01 -6.62271857e-01 6.10778593e-02 1.35386571e-01 -3.93232912e-01 2.79782057e-01 -1.12202549e+00 -1.00372648e+00 -2.61690527e-01 -2.44078845e-01 1.12474658e-01 4.67093408e-01 7.89897740e-01 3.52267563e-01 4.17987853e-01 1.11800575e+00 -6.04664326e-01 -1.76113558e+00 -4.71675187e-01 -8.27178955e-01 2.53616273e-01 5.06617665e-01 -4.23950732e-01 -3.50283831e-01 -2.37695947e-01]
[5.058665752410889, 3.8072609901428223]
5dc311fc-5ed3-4ef4-91da-8a23df3f561d
a-deep-variational-bayesian-framework-for
2106.02884
null
https://arxiv.org/abs/2106.02884v1
https://arxiv.org/pdf/2106.02884v1.pdf
A Deep Variational Bayesian Framework for Blind Image Deblurring
Blind image deblurring is an important yet very challenging problem in low-level vision. Traditional optimization based methods generally formulate this task as a maximum-a-posteriori estimation or variational inference problem, whose performance highly relies on the handcraft priors for both the latent image and the blur kernel. In contrast, recent deep learning methods generally learn, from a large collection of training images, deep neural networks (DNNs) directly mapping the blurry image to the clean one or to the blur kernel, paying less attention to the physical degradation process of the blurry image. In this paper, we present a deep variational Bayesian framework for blind image deblurring. Under this framework, the posterior of the latent clean image and blur kernel can be jointly estimated in an amortized inference fashion with DNNs, and the involved inference DNNs can be trained by fully considering the physical blur model, together with the supervision of data driven priors for the clean image and blur kernel, which is naturally led to by the evidence lower bound objective. Comprehensive experiments are conducted to substantiate the effectiveness of the proposed framework. The results show that it can not only achieve a promising performance with relatively simple networks, but also enhance the performance of existing DNNs for deblurring.
['Deyu Meng', 'Qian Zhao', 'Zongsheng Yue', 'Hui Wang']
2021-06-05
null
null
null
null
['blind-image-deblurring']
['computer-vision']
[ 1.58644497e-01 -3.10606778e-01 2.29996309e-01 -3.01362455e-01 -4.27929878e-01 -1.93289965e-01 6.11536980e-01 -6.69353306e-01 -3.29400629e-01 7.50881910e-01 4.89838064e-01 2.31637321e-02 -2.02646479e-01 -2.31663212e-01 -8.24008107e-01 -1.14204848e+00 4.64929610e-01 -1.15325442e-02 -5.16933091e-02 2.70212650e-01 1.63471028e-01 3.36843804e-02 -1.17778254e+00 -2.54530013e-01 1.12045646e+00 1.05212808e+00 5.39525151e-01 6.63663149e-01 3.44844878e-01 9.21702087e-01 -4.50821757e-01 -1.83443412e-01 -2.06337329e-02 -3.49941790e-01 -4.18760180e-01 2.92800933e-01 5.08717716e-01 -9.62865829e-01 -8.87742400e-01 1.70971084e+00 3.50706577e-01 1.93598419e-01 6.13889098e-01 -7.51263618e-01 -1.07882106e+00 4.24799472e-01 -6.77805662e-01 2.24830464e-01 -1.46713644e-01 3.33164304e-01 7.05676854e-01 -8.65027428e-01 9.72819179e-02 1.18193126e+00 4.89124328e-01 3.97891402e-01 -1.06338680e+00 -3.83329988e-01 1.08511493e-01 5.72190225e-01 -1.37527454e+00 -7.30212390e-01 7.32484460e-01 -3.91741157e-01 3.43935072e-01 3.32984403e-02 2.99300641e-01 1.18717229e+00 1.92184687e-01 7.79798508e-01 1.19221747e+00 -6.39211461e-02 2.60891199e-01 -8.91772285e-02 1.72624022e-01 5.19078076e-01 3.80444974e-01 1.98018327e-01 -2.89003044e-01 8.06916654e-02 9.69069302e-01 3.20522666e-01 -9.23525572e-01 -2.28202492e-01 -1.11582577e+00 4.75248605e-01 6.90705657e-01 7.77687281e-02 -5.89421391e-01 3.76074225e-01 1.09234182e-02 -1.98845029e-01 5.02917528e-01 -2.10199170e-02 -2.42427617e-01 7.18183443e-02 -1.03669751e+00 1.17685020e-01 5.70090115e-01 7.41589725e-01 7.97169864e-01 6.48266897e-02 -4.40536320e-01 8.79297376e-01 8.04954112e-01 6.74826145e-01 2.87263632e-01 -9.95202661e-01 2.26067662e-01 -4.47419472e-02 6.39921963e-01 -9.02715981e-01 2.12215394e-01 -5.72535157e-01 -1.26952112e+00 1.71037614e-01 3.20667535e-01 -3.96419346e-01 -1.13840997e+00 1.77648544e+00 1.35434255e-01 7.78146327e-01 8.31934810e-03 1.44961226e+00 4.55219090e-01 6.79290891e-01 -3.21343273e-01 -4.42072630e-01 1.38926983e+00 -1.21483445e+00 -1.15599668e+00 -4.47247148e-01 -3.23198736e-01 -8.25417399e-01 6.65946782e-01 4.28769380e-01 -1.07303739e+00 -6.67672217e-01 -1.12025273e+00 -2.70396680e-01 3.38163435e-01 3.53963047e-01 3.86889130e-01 4.60115820e-01 -1.05807698e+00 4.58128631e-01 -1.02490079e+00 4.64916900e-02 6.05742335e-01 6.39873594e-02 -8.93112365e-03 -6.30773664e-01 -1.16132951e+00 1.05503893e+00 2.82863379e-01 8.45571816e-01 -1.56821704e+00 -3.91092718e-01 -7.68677354e-01 3.03537577e-01 2.86923110e-01 -9.13781524e-01 1.15894091e+00 -7.55164266e-01 -1.56963825e+00 3.50732714e-01 -2.95495480e-01 -4.14617747e-01 5.87735653e-01 -7.95629203e-01 -1.82403415e-01 1.34107187e-01 -1.17644869e-01 2.97244221e-01 1.59677696e+00 -1.39611804e+00 -3.95262122e-01 -2.99546450e-01 5.76682910e-02 2.25115702e-01 -1.35924414e-01 -1.71984658e-01 -6.32022977e-01 -7.01893270e-01 -8.06081444e-02 -6.11316741e-01 -1.26204908e-01 -4.95637627e-03 -3.40988010e-01 7.87331630e-03 6.65707707e-01 -1.16751301e+00 1.05337381e+00 -2.29328418e+00 5.31320632e-01 -4.21854764e-01 3.71649593e-01 3.70600700e-01 5.57606705e-02 -1.69019744e-01 1.18457019e-01 -2.90757507e-01 -5.38251877e-01 -4.98483837e-01 7.98259676e-02 2.28285581e-01 -5.75505257e-01 8.73479128e-01 2.08806172e-02 8.46034408e-01 -9.98615980e-01 -2.23198548e-01 4.25450772e-01 7.51684070e-01 -3.44908118e-01 6.25376999e-01 -2.48964131e-01 5.77000558e-01 -2.95606345e-01 1.69161752e-01 1.15476000e+00 -3.67351502e-01 -8.92811939e-02 -6.15663528e-01 -1.06926247e-01 -1.88119769e-01 -1.05819881e+00 1.89521921e+00 -4.08765197e-01 6.64550900e-01 5.56557894e-01 -9.35011566e-01 5.85966706e-01 3.78711879e-01 6.82246080e-03 -1.37263373e-01 1.97230548e-01 1.27441660e-01 -2.08888724e-01 -8.41044962e-01 2.21565813e-01 -2.88094670e-01 3.58721197e-01 2.69356549e-01 1.31709874e-01 -1.47235513e-01 -2.58911550e-01 -1.24731511e-02 7.13050961e-01 1.83824360e-01 -2.03357358e-02 -7.66436234e-02 5.12696087e-01 -5.10132313e-01 5.35012424e-01 9.42682147e-01 -2.36564323e-01 7.67196059e-01 1.04601374e-02 -4.42103148e-02 -9.52482939e-01 -1.14137614e+00 -1.31898478e-01 5.10986030e-01 5.74107587e-01 1.41614005e-01 -1.08312905e+00 -3.91428977e-01 -3.36568058e-01 7.53374159e-01 -3.67775768e-01 -2.76791066e-01 -3.12483847e-01 -1.01535523e+00 1.32397398e-01 1.33502007e-01 9.85221684e-01 -6.83640718e-01 -1.18825771e-01 1.21462904e-02 -3.15210134e-01 -1.20546782e+00 -7.70824134e-01 -1.73363790e-01 -7.64430165e-01 -8.22299480e-01 -1.33447087e+00 -7.32552409e-01 7.28349626e-01 6.08879328e-01 6.06395423e-01 -6.05809093e-02 -1.36132669e-02 1.27378091e-01 -9.05062631e-03 -2.76542213e-02 -2.16252282e-01 -3.97013366e-01 1.78485185e-01 5.21500289e-01 5.69002107e-02 -7.34699428e-01 -8.87951910e-01 2.06295356e-01 -1.10002899e+00 1.80501163e-01 8.88355196e-01 1.04688096e+00 1.36112496e-01 4.96023715e-01 2.34979525e-01 -3.27235013e-01 5.65682411e-01 -5.03385246e-01 -7.54022717e-01 2.55167782e-01 -5.32955766e-01 2.37172395e-01 4.40785855e-01 -3.74533832e-01 -1.71112239e+00 -2.10789993e-01 2.87758484e-02 -9.84458625e-01 -3.44906509e-01 5.32211900e-01 -3.93655479e-01 1.21105999e-01 3.26695532e-01 5.16455054e-01 -1.96793489e-03 -8.86213124e-01 5.66264629e-01 8.53799045e-01 9.34270740e-01 -4.03376877e-01 9.26118612e-01 6.20158374e-01 -2.44565174e-01 -5.89464605e-01 -1.18801892e+00 -2.59811580e-01 -4.13142592e-01 -7.03933239e-02 1.06890559e+00 -1.24432790e+00 -6.48077488e-01 1.21447253e+00 -1.42956710e+00 -2.10697129e-01 1.41238496e-01 7.09804237e-01 -2.92879492e-01 7.58780122e-01 -8.18385124e-01 -8.04995179e-01 -5.14181815e-02 -1.21559584e+00 8.38006616e-01 5.48697293e-01 4.41105098e-01 -1.10121763e+00 -1.25019133e-01 5.39415002e-01 5.49768448e-01 -1.67249858e-01 6.55949116e-01 4.81638722e-02 -9.29837346e-01 -7.26285577e-02 -8.59326482e-01 8.91835809e-01 4.78845805e-01 -5.03136277e-01 -1.26050735e+00 -3.68834317e-01 7.38372684e-01 -7.56513104e-02 1.17800558e+00 8.47049415e-01 1.10998106e+00 -5.55883706e-01 -9.97787341e-02 8.09262872e-01 1.43033993e+00 -1.95634633e-01 6.14050686e-01 -4.67623472e-02 1.05036354e+00 3.17667693e-01 2.19297528e-01 3.52885336e-01 2.63539761e-01 5.76213419e-01 6.91443443e-01 6.30016401e-02 -3.69552374e-01 -6.43494129e-02 3.92865062e-01 8.48488629e-01 -5.88011630e-02 -2.10180908e-01 -4.85836208e-01 6.01617098e-01 -2.05806470e+00 -7.95413733e-01 -5.92890242e-03 2.07745743e+00 1.15325761e+00 -5.03513217e-02 -5.61745644e-01 -3.97484124e-01 9.80598986e-01 4.60768998e-01 -8.23394179e-01 4.59592968e-01 -3.86280939e-02 -1.06441945e-01 3.79718125e-01 9.29100156e-01 -1.01570594e+00 7.45628178e-01 5.94902706e+00 9.04382825e-01 -1.02499020e+00 2.55242318e-01 5.41922390e-01 1.89587791e-02 -1.04271680e-01 1.18129969e-01 -5.51790476e-01 8.56321454e-01 6.99538410e-01 1.34459091e-02 1.05007350e+00 5.48225820e-01 5.99206984e-01 -1.82095692e-01 -9.60601211e-01 1.25449371e+00 3.20232473e-02 -1.12337935e+00 -7.03843310e-02 3.70055027e-02 8.64034593e-01 1.25025421e-01 2.39816546e-01 -3.12209968e-02 3.37960899e-01 -1.06958389e+00 7.57122695e-01 1.07596302e+00 6.62554502e-01 -2.66922265e-01 9.44577932e-01 4.85192597e-01 -5.96947551e-01 6.58757659e-03 -5.28441846e-01 -2.89871305e-01 3.30801874e-01 1.08229446e+00 -2.19213098e-01 5.24960995e-01 7.78774023e-01 1.11188567e+00 -2.30158508e-01 1.17602360e+00 -6.83940172e-01 7.33981729e-01 3.33578996e-02 4.38058078e-01 5.66987209e-02 -5.10970592e-01 7.29061723e-01 1.01329613e+00 3.00322294e-01 6.29861664e-04 -1.71031579e-01 1.25294185e+00 -1.93202317e-01 -7.43604481e-01 -2.15435363e-02 1.21651307e-01 2.08575055e-01 1.18975508e+00 -1.89822882e-01 -4.60833758e-01 -2.52940714e-01 1.32745814e+00 8.89079049e-02 8.63366008e-01 -9.47166026e-01 -3.44709814e-01 8.63567293e-01 -5.66599548e-01 6.33573294e-01 -2.07923114e-01 -2.35433310e-01 -1.64224041e+00 1.16638884e-01 -8.01779270e-01 -1.74857348e-01 -1.27313316e+00 -1.52816451e+00 3.56527299e-01 -5.20918518e-03 -1.03153789e+00 1.49214625e-01 -6.86558008e-01 -7.03429461e-01 1.43014419e+00 -1.90186870e+00 -8.95392716e-01 -6.04776144e-01 4.96017635e-01 4.83393401e-01 1.77324876e-01 2.80745775e-01 2.29180679e-01 -9.47233558e-01 -1.58080682e-02 5.03953636e-01 2.40786690e-02 8.20095301e-01 -1.27323377e+00 1.00881897e-01 1.26647675e+00 -1.86582461e-01 8.14098179e-01 1.05173802e+00 -4.52622622e-01 -1.27222824e+00 -1.02347124e+00 3.70278269e-01 -2.60738254e-01 7.28461623e-01 -1.33049890e-01 -1.16058278e+00 4.61187392e-01 6.18997276e-01 1.91065148e-01 -9.71581563e-02 -1.75119385e-01 -1.63064420e-01 -2.32994482e-01 -8.53686452e-01 4.74413842e-01 6.98428810e-01 -7.61223316e-01 -8.28187466e-01 3.10305417e-01 7.75552034e-01 -4.63636905e-01 -5.95568597e-01 1.84807613e-01 2.96833009e-01 -9.46728885e-01 9.69464600e-01 -3.18312109e-01 6.31419301e-01 -6.78121984e-01 1.90983173e-02 -1.59352970e+00 -4.72555339e-01 -7.12380469e-01 -6.02803886e-01 1.17211723e+00 -2.33888701e-01 -3.94208521e-01 3.65902364e-01 4.46553320e-01 -3.10626805e-01 -2.47744903e-01 -7.43970931e-01 -4.26505834e-01 -2.26387098e-01 -2.17139557e-01 3.89946491e-01 7.73978412e-01 -5.33206820e-01 5.19199848e-01 -9.97536123e-01 7.71749556e-01 1.18965161e+00 1.21560581e-02 3.58005881e-01 -8.21995020e-01 -5.58314323e-01 -3.89922708e-01 5.83820790e-02 -1.73640883e+00 1.02903962e-01 -3.97912443e-01 6.52849734e-01 -1.62940001e+00 4.80663955e-01 1.19867831e-01 -4.39644784e-01 1.24711785e-02 -7.04775810e-01 8.34745530e-04 -2.29160592e-01 5.32512367e-01 -5.06221056e-01 9.17514443e-01 1.35239291e+00 -3.49850476e-01 2.30659425e-01 4.19764481e-02 -7.00166702e-01 8.30263913e-01 2.49782696e-01 -3.04830521e-01 -4.47661936e-01 -1.08219421e+00 -5.96881881e-02 1.39556557e-01 8.53792667e-01 -6.42309904e-01 5.04638076e-01 -2.75283039e-01 3.80416155e-01 -1.97467580e-01 3.40642720e-01 -8.09334338e-01 1.33573962e-02 1.67644441e-01 -1.39491513e-01 -7.20847309e-01 -4.43549305e-02 1.04766130e+00 -2.74712354e-01 -4.26289290e-01 9.55764890e-01 -1.01754472e-01 -5.71290791e-01 4.10620421e-01 -1.28953129e-01 -1.92784652e-01 4.24804837e-01 1.33947298e-01 -3.73666316e-01 -4.54341590e-01 -6.82321548e-01 1.07361704e-01 3.73341113e-01 2.48276919e-01 6.60763383e-01 -1.16327751e+00 -7.06702352e-01 8.97276103e-02 -2.83387303e-01 3.21659833e-01 7.14302123e-01 9.32284236e-01 -3.13455790e-01 2.72289217e-01 -1.10869512e-01 -5.54408431e-01 -7.07409203e-01 7.48890042e-01 5.62216580e-01 1.00028798e-01 -4.15188640e-01 1.00242412e+00 6.79844916e-01 1.40510321e-01 2.08548963e-01 -3.56635541e-01 -6.76170066e-02 -3.05181175e-01 7.36953259e-01 4.36762184e-01 -2.72545546e-01 -5.57076991e-01 -9.77550671e-02 3.77263218e-01 -1.90654978e-01 -5.18456101e-02 1.32103920e+00 -7.05379784e-01 -4.97257888e-01 7.91156366e-02 1.12199283e+00 -1.44643798e-01 -2.06318641e+00 -5.46873808e-01 -2.34421670e-01 -6.42707705e-01 8.32623363e-01 -6.51866376e-01 -1.17617452e+00 1.06933987e+00 6.81947112e-01 1.02607375e-02 1.20124626e+00 -4.62706834e-02 8.30451846e-01 7.46881515e-02 4.09011133e-02 -7.24839270e-01 1.10841393e-01 3.14453423e-01 9.17264581e-01 -1.34486985e+00 2.26667505e-02 -3.42673995e-02 -2.68756539e-01 9.62617755e-01 3.20712417e-01 -3.57363254e-01 7.56765604e-01 -1.67917550e-01 -7.07302913e-02 -2.20611840e-02 -4.93505180e-01 9.37661454e-02 3.99703681e-01 3.72049779e-01 -7.98841007e-03 -1.70687586e-01 2.05678225e-01 6.72499239e-01 2.96288639e-01 3.37931067e-01 4.96293753e-01 3.20267141e-01 -4.52664614e-01 -6.05192184e-01 -4.87473607e-01 8.98161530e-02 -3.96510720e-01 -4.21147436e-01 1.59170613e-01 8.03975482e-03 2.57563204e-01 1.07900512e+00 -8.80500525e-02 2.91078333e-02 -5.55958673e-02 -3.35319072e-01 4.71341252e-01 -4.44978058e-01 2.38729253e-01 3.49331081e-01 -3.79897058e-01 -2.93791622e-01 -5.93065560e-01 -6.08788848e-01 -5.39022505e-01 -2.25842372e-01 -5.25780737e-01 7.32159615e-02 4.76022154e-01 1.31858909e+00 2.14993522e-01 5.56574285e-01 5.42411566e-01 -1.11936200e+00 -8.82413387e-01 -1.32275319e+00 -7.95550585e-01 2.88161337e-01 1.02614868e+00 -4.88563567e-01 -7.31306732e-01 4.65826064e-01]
[11.55589771270752, -2.6762313842773438]
df52a148-0b84-4790-baab-8839bfc66088
stock-market-prediction-from-wsj-text-mining
1406.7330
null
http://arxiv.org/abs/1406.7330v1
http://arxiv.org/pdf/1406.7330v1.pdf
Stock Market Prediction from WSJ: Text Mining via Sparse Matrix Factorization
We revisit the problem of predicting directional movements of stock prices based on news articles: here our algorithm uses daily articles from The Wall Street Journal to predict the closing stock prices on the same day. We propose a unified latent space model to characterize the "co-movements" between stock prices and news articles. Unlike many existing approaches, our new model is able to simultaneously leverage the correlations: (a) among stock prices, (b) among news articles, and (c) between stock prices and news articles. Thus, our model is able to make daily predictions on more than 500 stocks (most of which are not even mentioned in any news article) while having low complexity. We carry out extensive backtesting on trading strategies based on our algorithm. The result shows that our model has substantially better accuracy rate (55.7%) compared to many widely used algorithms. The return (56%) and Sharpe ratio due to a trading strategy based on our model are also much higher than baseline indices.
['Zhenming Liu', 'Mung Chiang', 'Felix Ming Fai Wong']
2014-06-27
null
null
null
null
['stock-market-prediction']
['time-series']
[-7.96215951e-01 -2.41550535e-01 -6.25689328e-01 -1.42850816e-01 -7.11838007e-01 -1.01926541e+00 1.19768322e+00 -7.18567595e-02 -1.86417177e-01 8.65701020e-01 5.81171036e-01 -5.78273058e-01 -7.15500563e-02 -1.09665442e+00 -8.08800638e-01 -3.16486537e-01 -2.12827638e-01 5.12913406e-01 4.69281226e-01 -3.15172344e-01 7.72646010e-01 9.50987116e-02 -1.18609142e+00 1.37604460e-01 5.20033717e-01 1.30064082e+00 3.49067897e-02 3.22004110e-01 -4.01704431e-01 1.37218308e+00 -3.16228449e-01 -6.86989844e-01 8.57781947e-01 -2.00378790e-01 -3.07372332e-01 -2.20156848e-01 -8.43346417e-02 -5.13003767e-01 -1.99219733e-01 1.14700437e+00 -3.54441375e-01 -2.54406214e-01 6.49142265e-01 -9.50022757e-01 -7.28978217e-01 1.29332411e+00 -7.47396529e-01 8.86294603e-01 3.54887098e-02 -2.14881375e-01 1.90578592e+00 -9.37778592e-01 5.41319013e-01 6.62864327e-01 7.21922874e-01 -4.87364471e-01 -1.08927548e+00 -6.89401090e-01 3.88777822e-01 -2.09852025e-01 -6.09356344e-01 -7.99548402e-02 6.76423490e-01 -6.69977188e-01 8.48648429e-01 3.70184690e-01 9.69860494e-01 9.95553613e-01 8.44131827e-01 6.84483469e-01 1.25688064e+00 -1.21194921e-01 3.38378176e-02 2.48034433e-01 1.49656564e-01 -7.76361302e-02 6.02009296e-01 3.00738662e-01 -5.56109011e-01 -5.67490995e-01 8.32830846e-01 4.33291167e-01 1.00600488e-01 6.03261106e-02 -1.41771960e+00 1.29427159e+00 3.72761698e-03 2.50340849e-01 -7.16243148e-01 5.71200848e-02 -8.62461608e-03 8.75553429e-01 1.01894379e+00 6.97307408e-01 -1.06911778e+00 -4.22616303e-01 -1.21080565e+00 7.64870942e-01 1.20687127e+00 6.50653303e-01 2.40804061e-01 1.64784655e-01 -9.93351489e-02 4.27095681e-01 5.09140909e-01 7.73891509e-01 7.57151008e-01 -5.26716828e-01 7.48342574e-01 3.30568254e-01 5.12684882e-01 -1.20876777e+00 -3.07123870e-01 -6.35276139e-01 -1.99422941e-01 3.92615504e-04 3.70843679e-01 -2.81030715e-01 -2.65545130e-01 1.29641974e+00 -3.90102506e-01 3.89663011e-01 1.01052120e-01 3.31586629e-01 5.76380789e-02 1.08170891e+00 -3.92870098e-01 -7.42919385e-01 1.17882752e+00 -1.03252661e+00 -6.42597079e-01 -2.75462508e-01 2.75820285e-01 -1.00828791e+00 5.53120375e-01 2.70045489e-01 -1.25617504e+00 -2.89167725e-02 -8.74338388e-01 6.10758007e-01 -2.70440906e-01 -2.27400720e-01 6.78745985e-01 1.91718459e-01 -8.22481632e-01 8.50055695e-01 -7.45922387e-01 5.11115491e-01 -2.01430127e-01 4.37279865e-02 4.96252626e-01 1.01682889e+00 -1.25201523e+00 8.66599083e-01 1.66993335e-01 -2.27168530e-01 -2.41777509e-01 -9.23235238e-01 -2.27185592e-01 4.60408106e-02 4.59855676e-01 -2.49308437e-01 1.42034101e+00 -5.90020537e-01 -1.67537296e+00 3.92284989e-01 2.38584206e-01 -1.07316279e+00 8.88654709e-01 -5.10658920e-01 -6.93600953e-01 -2.99704790e-01 3.63425910e-01 -2.61827618e-01 5.62218428e-01 -5.49350739e-01 -1.01280689e+00 3.72844711e-02 -1.61646411e-01 -6.45202398e-02 -1.73192590e-01 2.89585501e-01 3.26595991e-03 -1.38156426e+00 2.99320012e-01 -1.12126338e+00 2.07969453e-02 -9.12339151e-01 -3.90991896e-01 -6.68551475e-02 2.40713030e-01 -6.62771642e-01 1.54623878e+00 -1.71938491e+00 -3.77029449e-01 5.46907187e-01 -5.69669753e-02 -5.20017028e-01 3.83716464e-01 5.99913895e-01 -1.77993774e-01 1.96856499e-01 1.27304077e-01 7.88860247e-02 4.14495081e-01 -2.25623131e-01 -1.56462932e+00 4.40023571e-01 -3.20474923e-01 1.16835558e+00 -4.38134640e-01 3.01387131e-01 -2.40765348e-01 -4.09047544e-01 -4.49102193e-01 -6.79527521e-02 -3.98464799e-01 -1.39611199e-01 -3.27585816e-01 4.67473298e-01 4.39423561e-01 -5.87558448e-01 1.81885108e-01 4.16494399e-01 -6.53050900e-01 9.59398985e-01 -1.02936101e+00 5.72785616e-01 2.82564312e-02 4.70450491e-01 -5.87834120e-01 -5.39821029e-01 1.01743340e+00 1.98013067e-01 4.66179311e-01 -7.74915874e-01 -1.87172890e-02 5.00171959e-01 -6.21322580e-02 2.30011433e-01 5.23585081e-01 -2.80530453e-01 -2.69128740e-01 1.13099170e+00 -4.38832164e-01 2.83574104e-01 3.16851974e-01 1.72647517e-02 6.96176469e-01 -1.40533790e-01 3.31624210e-01 -6.49253190e-01 -1.39764445e-02 -9.01519358e-02 5.75784802e-01 9.16189909e-01 2.67790973e-01 1.40734598e-01 1.00681186e+00 -6.37836576e-01 -1.18673658e+00 -1.00795519e+00 -3.06171715e-01 1.09951627e+00 -1.68626338e-01 -3.59455734e-01 -7.24314898e-02 -3.01529229e-01 6.14944279e-01 9.47466791e-01 -7.92126894e-01 4.50492442e-01 -2.09604576e-01 -1.31544113e+00 4.04652953e-02 5.64194024e-01 3.37195665e-01 -9.50545192e-01 -4.82917607e-01 5.06818593e-01 1.86325863e-01 -8.74006808e-01 -4.66209084e-01 1.26756728e-01 -1.07685852e+00 -7.14875758e-01 -7.41916776e-01 -3.26075971e-01 7.86492303e-02 -7.71856159e-02 1.24555039e+00 -3.47986817e-01 6.47602677e-01 1.41950147e-02 -2.28749841e-01 -9.61416602e-01 -2.02853084e-01 -5.01023000e-03 1.99948698e-01 2.34914213e-01 3.79330039e-01 -4.27074403e-01 -4.84786451e-01 2.64128149e-01 -6.25154018e-01 -9.79695544e-02 5.20060956e-01 6.55376136e-01 3.63252014e-01 2.06072882e-01 5.61704159e-01 -1.05101049e+00 7.18594551e-01 -8.78621817e-01 -1.51979423e+00 2.19725251e-01 -1.28699946e+00 1.36871263e-01 2.61112452e-01 -3.55276436e-01 -9.19621646e-01 -4.62155849e-01 2.58744299e-01 1.01907946e-01 5.80941916e-01 9.83171344e-01 7.81787157e-01 5.25542855e-01 -4.49510552e-02 4.57651436e-01 2.45361421e-02 -7.20747173e-01 -1.72015224e-02 2.11601093e-01 3.52682680e-01 3.39125050e-03 1.13520908e+00 4.06167984e-01 -4.47755784e-01 -2.32767597e-01 -1.03896260e+00 -2.71170795e-01 -3.65211010e-01 2.55731940e-01 4.35472727e-01 -1.19445300e+00 -4.74191725e-01 6.01397753e-01 -5.86767912e-01 -1.53320625e-01 -1.18155383e-01 9.26921368e-01 -4.85377192e-01 -6.57197461e-02 -1.31221676e+00 -8.65947545e-01 -2.55803853e-01 -8.61249268e-01 3.88902098e-01 -4.94119152e-02 -4.19126451e-01 -1.42527092e+00 6.33386016e-01 6.44091591e-02 6.32594526e-01 2.09191307e-01 7.74658382e-01 -1.40701246e+00 -6.85256958e-01 -3.42943847e-01 -9.09887068e-03 -7.87874497e-03 2.01456383e-01 2.09727570e-01 -4.25178647e-01 -2.48439591e-02 5.24028242e-01 9.29556787e-02 9.02893960e-01 6.85204268e-01 2.28578746e-01 -8.06230307e-01 5.99901341e-02 4.48566675e-01 1.24441564e+00 3.93575698e-01 3.66757929e-01 1.22880471e+00 1.20770540e-02 3.42839420e-01 5.59557557e-01 8.37342680e-01 2.68376321e-01 4.61936980e-01 -6.75483868e-02 2.90770888e-01 8.86604667e-01 -4.93155748e-01 8.01551998e-01 1.16394150e+00 -1.54007778e-01 9.08217803e-02 -8.10554862e-01 2.53611892e-01 -1.89845347e+00 -1.38486445e+00 -2.07721055e-01 1.82116163e+00 8.04622233e-01 9.56582606e-01 5.72445035e-01 -3.09610754e-01 3.61739099e-01 6.47389591e-01 -4.78898138e-01 2.75148824e-02 -4.23080772e-01 -2.44272336e-01 1.17133844e+00 3.79948646e-01 -1.28213882e+00 7.06021488e-01 8.27427769e+00 2.78355032e-01 -1.01701093e+00 -2.86996543e-01 7.70710886e-01 -2.62400359e-01 -7.92880476e-01 3.10444180e-02 -1.28703320e+00 1.05294228e+00 1.18054640e+00 -9.26640868e-01 1.56458691e-01 1.01186037e+00 7.03679621e-02 1.37747034e-01 -6.85522616e-01 5.74633002e-01 -1.89157009e-01 -1.93155253e+00 1.42540351e-01 5.09492338e-01 1.08641505e+00 3.42405349e-01 5.92140853e-01 1.85335845e-01 6.65262520e-01 -4.59394366e-01 1.17622924e+00 8.18709970e-01 -3.85852880e-04 -8.17076564e-01 9.45661485e-01 4.76011425e-01 -9.27758753e-01 -2.73946255e-01 -4.30125803e-01 -4.75944608e-01 1.58702329e-01 7.88595498e-01 -4.22671348e-01 2.05892026e-01 6.67692602e-01 1.00857556e+00 -4.29768443e-01 6.99893296e-01 -1.69663548e-01 8.24618280e-01 -3.49392742e-01 -1.54532388e-01 6.18289053e-01 -6.38266146e-01 4.62165415e-01 7.16235101e-01 5.89984536e-01 -4.85950261e-02 -1.05438516e-01 9.27067816e-01 -6.21377528e-02 2.25636557e-01 -5.97593665e-01 -2.56816745e-01 2.22533464e-01 6.13515854e-01 -9.54805374e-01 -5.37138343e-01 -9.83794272e-01 5.08791625e-01 -2.12718874e-01 1.33577287e-01 -9.10764217e-01 -2.56167687e-02 5.73732376e-01 2.87002534e-01 5.86535692e-01 -1.55273780e-01 -5.49000204e-01 -1.52921700e+00 1.15821108e-01 -6.90113664e-01 3.70316148e-01 -4.64460999e-01 -1.63270426e+00 4.57770914e-01 2.08296757e-02 -1.48629642e+00 -5.08513033e-01 -6.26407444e-01 -8.52419555e-01 7.17690170e-01 -1.55840731e+00 -1.55723467e-01 8.29873383e-01 2.91220158e-01 4.60692614e-01 -8.50390196e-01 4.05341536e-01 -2.64934272e-01 -4.58967388e-01 7.08471239e-02 8.80327523e-01 3.47528577e-01 4.08918798e-01 -1.52036858e+00 1.09330881e+00 8.21614027e-01 6.71257555e-01 7.07204759e-01 6.15165055e-01 -1.20372748e+00 -8.40454698e-01 -7.67167687e-01 1.15158761e+00 -7.69457161e-01 1.82620466e+00 -1.42550971e-02 -7.22700357e-01 1.30299687e+00 3.20387363e-01 -6.25647843e-01 8.45492661e-01 2.96156168e-01 -5.15191317e-01 -6.31491914e-02 -3.82687002e-01 5.16292572e-01 2.56230295e-01 -3.27367932e-01 -1.33889461e+00 3.37960005e-01 8.45259488e-01 -1.09965265e-01 -8.44184160e-01 1.30588040e-01 7.30764031e-01 -1.15271926e+00 9.10092711e-01 -6.82086945e-01 5.28383613e-01 1.34370148e-01 -1.12045296e-01 -1.15950859e+00 -7.52126396e-01 -8.09984207e-01 -2.28773803e-01 9.74656403e-01 9.85191584e-01 -1.25621533e+00 5.85163832e-01 8.22853625e-01 3.86692673e-01 -4.92032290e-01 -6.35438561e-01 -9.71834660e-01 1.72009736e-01 -3.12198818e-01 8.08151305e-01 1.07335401e+00 -9.01712198e-03 1.14764385e-01 -7.69910693e-01 -4.20666784e-02 5.34282029e-01 1.05409777e+00 4.89907861e-01 -1.55330265e+00 -6.26585126e-01 -8.39685977e-01 1.12148494e-01 -1.08161879e+00 6.62498735e-03 -7.77153194e-01 -5.74392319e-01 -9.63196099e-01 2.57866144e-01 -1.92720681e-01 -7.26662576e-01 1.08920231e-01 5.55358529e-02 1.16950057e-01 2.29585379e-01 9.50158179e-01 -3.44502330e-01 3.06420207e-01 7.66989291e-01 1.97585821e-01 -3.58569503e-01 3.88837248e-01 -9.79931712e-01 1.13864768e+00 7.68079102e-01 -5.08453548e-01 7.60201812e-02 -1.59184113e-02 9.58410025e-01 3.69065285e-01 4.05586585e-02 -4.88840550e-01 2.81830132e-01 -5.70160329e-01 4.69555646e-01 -1.02245629e+00 -1.03338704e-01 -3.73839200e-01 3.81148428e-01 6.26941323e-01 -4.58214581e-01 8.46718788e-01 -1.30514830e-01 7.29148686e-01 -5.81296325e-01 -3.25675569e-02 3.76158178e-01 -2.45606601e-01 -5.54071426e-01 1.91125706e-01 -5.66988170e-01 1.27817169e-01 1.08600247e+00 3.18111151e-01 -3.32489312e-01 -7.16141701e-01 -6.67238414e-01 1.53725296e-01 2.57833809e-01 5.25351524e-01 1.65743783e-01 -1.31125963e+00 -8.78337860e-01 2.17756450e-01 -1.32113799e-01 -9.42708433e-01 -3.64860654e-01 9.01204467e-01 -7.33276069e-01 8.82398725e-01 -8.41310248e-02 9.09237750e-03 -4.53089774e-01 4.18622911e-01 9.79309622e-03 -7.15975761e-01 -8.82507086e-01 7.92038321e-01 1.18351601e-01 8.88231099e-02 4.35767658e-02 -5.60060143e-01 -3.65418822e-01 8.67891788e-01 8.27702403e-01 3.29946935e-01 -2.81437248e-01 -4.84702289e-01 -1.82596967e-01 5.52637339e-01 -3.31275970e-01 -3.83794993e-01 1.96106470e+00 -1.20322257e-01 -2.37046272e-01 1.26769400e+00 9.85611320e-01 5.72075963e-01 -1.23578918e+00 -3.73715103e-01 8.33917499e-01 -3.97794604e-01 -3.25976908e-02 -5.85289121e-01 -1.05269706e+00 -2.76668370e-02 -1.02128647e-01 7.41032064e-01 4.94454443e-01 -9.37166512e-02 1.02257228e+00 3.79095584e-01 4.07437861e-01 -1.32229328e+00 -1.50583252e-01 6.37433350e-01 8.83051515e-01 -1.03780532e+00 3.32205564e-01 9.26760510e-02 -9.76633847e-01 1.03851247e+00 -3.67331117e-01 -7.03130186e-01 1.27821541e+00 3.50955158e-01 3.02524388e-01 -2.57263899e-01 -1.32210267e+00 2.55590200e-01 3.48464429e-01 -6.19725227e-01 1.77094892e-01 1.42765298e-01 -3.07013214e-01 9.33916986e-01 -8.53708982e-01 -3.51951569e-01 8.37058663e-01 8.94359708e-01 -5.47249734e-01 -8.84766698e-01 -4.46365744e-01 9.33499038e-01 -1.25925589e+00 -4.10104394e-01 -2.25853324e-01 7.46064961e-01 -7.57438183e-01 5.56527555e-01 6.06719136e-01 -3.02206814e-01 2.78802305e-01 2.79557079e-01 -4.23100263e-01 -4.58242685e-01 -4.89249915e-01 6.95491076e-01 -1.65186197e-01 -4.20936555e-01 -3.15557539e-01 -1.12808943e+00 -7.18809545e-01 -6.00385070e-01 -1.45251602e-01 3.55524004e-01 2.29004338e-01 7.82604396e-01 7.59557337e-02 7.81357661e-02 1.13276219e+00 -3.68586391e-01 -1.16277134e+00 -7.76470780e-01 -1.39223194e+00 3.10124040e-01 4.04527456e-01 -7.67345250e-01 -8.94860923e-01 5.37798554e-02]
[4.460839748382568, 4.259009838104248]
7b1bada3-deff-4cc4-98d8-a612e1c6221c
sfe-ai-at-semeval-2022-task-11-low-resource
2205.14660
null
https://arxiv.org/abs/2205.14660v1
https://arxiv.org/pdf/2205.14660v1.pdf
SFE-AI at SemEval-2022 Task 11: Low-Resource Named Entity Recognition using Large Pre-trained Language Models
Large scale pre-training models have been widely used in named entity recognition (NER) tasks. However, model ensemble through parameter averaging or voting can not give full play to the differentiation advantages of different models, especially in the open domain. This paper describes our NER system in the SemEval 2022 task11: MultiCoNER. We proposed an effective system to adaptively ensemble pre-trained language models by a Transformer layer. By assigning different weights to each model for different inputs, we adopted the Transformer layer to integrate the advantages of diverse models effectively. Experimental results show that our method achieves superior performances in Farsi and Dutch.
['Qifeng Xiao', 'Benqi Wang', 'Xiandi Jiang', 'Xiaopeng Wang', 'Qizhi Lin', 'Guotong Xie', 'Peng Gao', 'Peng Jiang', 'Yixuan Qiao', 'Jun Wang', 'Changyu Hou']
2022-05-29
null
https://aclanthology.org/2022.semeval-1.219
https://aclanthology.org/2022.semeval-1.219.pdf
semeval-naacl-2022-7
['low-resource-named-entity-recognition']
['natural-language-processing']
[-3.29328835e-01 -2.39165664e-01 5.01430556e-02 -6.39756441e-01 -8.62485409e-01 -4.69173402e-01 5.63529611e-01 -2.02817217e-01 -1.18912756e+00 1.01499641e+00 3.51618737e-01 -9.90844220e-02 1.21928021e-01 -5.73361337e-01 -3.09349537e-01 -3.94073218e-01 2.26695046e-01 6.80564523e-01 1.61661550e-01 -4.09933686e-01 -9.70064476e-02 5.80067635e-01 -9.14728224e-01 4.33596760e-01 1.21510541e+00 4.23848152e-01 2.00933650e-01 5.18845081e-01 -5.95724225e-01 6.79464519e-01 -8.59227002e-01 -6.04648352e-01 -1.05693288e-01 -2.43783042e-01 -8.63178670e-01 -5.01257658e-01 9.63065401e-02 -3.86117003e-03 -2.10116759e-01 9.07232940e-01 8.69696319e-01 3.16271335e-01 6.28600657e-01 -8.18452001e-01 -6.13723040e-01 1.40687072e+00 -1.61584750e-01 3.91515404e-01 -6.39988035e-02 -4.55954701e-01 8.48710358e-01 -1.15628314e+00 5.98263919e-01 9.88632917e-01 8.15307736e-01 7.69609690e-01 -9.23811078e-01 -9.34594333e-01 7.97610655e-02 1.88724920e-01 -1.67180061e+00 -7.85963595e-01 4.55660313e-01 1.32224873e-01 1.34088051e+00 1.70150191e-01 8.88420492e-02 1.12886071e+00 -3.20254192e-02 9.64517772e-01 9.60045159e-01 -2.84344226e-01 -7.08917901e-02 2.06687838e-01 2.98918873e-01 2.22451270e-01 2.92979389e-01 -3.06010187e-01 -9.90813971e-02 -2.48190075e-01 5.82341850e-01 -1.53208107e-01 -1.48908749e-01 3.85887325e-01 -1.20466483e+00 5.11285365e-01 4.90251929e-01 9.51702833e-01 -6.45427167e-01 -1.79594532e-01 3.79004240e-01 3.53479058e-01 3.86323243e-01 7.13429570e-01 -1.05931818e+00 -3.85021269e-02 -1.01012444e+00 -1.45389199e-01 7.59687364e-01 8.97137761e-01 4.83524114e-01 1.72297657e-01 -1.75216764e-01 1.37252092e+00 1.32902279e-01 5.26017249e-01 7.63418972e-01 -5.23458600e-01 6.69130325e-01 5.71306109e-01 2.23671019e-01 -4.08963174e-01 -5.23999453e-01 -2.89792955e-01 -1.08338654e+00 -2.95959175e-01 2.39020273e-01 -7.94490218e-01 -1.09867251e+00 1.61218071e+00 7.37560242e-02 3.86670023e-01 3.70168418e-01 4.06513095e-01 1.06296182e+00 6.44450128e-01 7.27142751e-01 -6.10770956e-02 1.27470887e+00 -9.92489457e-01 -1.00247884e+00 -2.68557787e-01 8.95231426e-01 -8.59104931e-01 3.32798153e-01 -8.11120681e-03 -7.63410687e-01 -5.33825815e-01 -7.24503398e-01 -3.16653214e-02 -7.87618697e-01 6.00960970e-01 3.35322469e-01 6.45697415e-01 -9.15719807e-01 6.43621087e-01 -7.90482879e-01 -3.72004926e-01 2.12010115e-01 4.98428673e-01 -5.85734963e-01 1.17846109e-01 -1.68723404e+00 1.11913657e+00 9.86339390e-01 4.15635139e-01 -3.57516319e-01 -5.12329638e-01 -7.34894395e-01 1.10214822e-01 1.55725703e-01 -5.12425482e-01 1.26816225e+00 -7.04333425e-01 -1.40742099e+00 4.36625749e-01 -2.12662786e-01 -4.39942181e-01 3.05777967e-01 -4.00242716e-01 -1.10795057e+00 -3.51628512e-01 -8.56494233e-02 5.90017617e-01 2.44160667e-01 -1.03039789e+00 -6.57449782e-01 -2.61749327e-01 -1.77419290e-01 2.26170033e-01 -4.86651301e-01 4.95607913e-01 -1.77114382e-01 -6.18964255e-01 -1.74786508e-01 -7.56531954e-01 -5.55916488e-01 -1.01086140e+00 -3.11609030e-01 -5.41607380e-01 4.02787387e-01 -8.47714245e-01 1.72726142e+00 -1.90815151e+00 -1.26593709e-01 1.83496661e-02 9.11052674e-02 8.23586166e-01 -2.28089184e-01 4.18421715e-01 -2.55483836e-02 4.91671801e-01 -1.04997931e-02 -2.23390460e-01 -1.84371531e-01 2.78828651e-01 -6.42346740e-02 -1.14330165e-01 2.80960500e-01 8.01605701e-01 -8.80467355e-01 -7.15124547e-01 1.27082691e-01 6.94379628e-01 -1.06749117e-01 2.05936551e-01 1.95338398e-01 1.60933435e-01 -7.68754244e-01 3.68302286e-01 6.10352278e-01 -7.79353529e-02 2.68125147e-01 -2.53156811e-01 -2.12809384e-01 2.82591581e-01 -1.40311825e+00 1.52316022e+00 -6.45923972e-01 2.80460805e-01 -4.11888352e-03 -5.31920850e-01 8.46376300e-01 6.97866499e-01 1.88743174e-01 -3.89069498e-01 2.37889901e-01 5.67914844e-01 1.18777975e-01 -1.91863239e-01 7.38517284e-01 -2.37737164e-01 -2.81835854e-01 -3.03485505e-02 5.60278535e-01 3.05040747e-01 1.81590915e-01 3.62766087e-02 9.08516169e-01 -5.29984757e-02 6.56840622e-01 -1.77524596e-01 6.51024938e-01 -5.47693893e-02 8.96016121e-01 7.37384379e-01 -1.60830349e-01 4.36500669e-01 -2.11059213e-01 -2.59615153e-01 -8.73898566e-01 -7.50194788e-01 -3.24596256e-01 1.41068602e+00 -2.37880334e-01 -4.13220376e-01 -6.52621388e-01 -1.01032066e+00 -4.55101639e-01 1.09391594e+00 -4.99396026e-01 2.54940331e-01 -9.82544959e-01 -8.57828796e-01 1.28478777e+00 8.22236657e-01 7.26646721e-01 -1.39971030e+00 -8.57309438e-03 6.51464403e-01 -4.42913949e-01 -1.19195557e+00 -6.17929578e-01 4.34469998e-01 -9.42734778e-01 -7.05981910e-01 -9.06884670e-01 -8.37912083e-01 3.65854770e-01 -3.96561921e-02 1.06152678e+00 -2.02632487e-01 2.83464879e-01 -4.25642580e-02 -3.59673917e-01 -5.11969090e-01 -4.70297813e-01 6.83764219e-01 9.53141600e-02 -1.24087058e-01 5.82973421e-01 -2.51386434e-01 -4.91715185e-02 5.49075246e-01 -7.47643590e-01 -2.11821511e-01 6.86372399e-01 9.87196207e-01 4.33219135e-01 2.44784430e-02 7.75745809e-01 -1.27372789e+00 6.13438785e-01 -5.46678960e-01 -1.87783316e-01 8.12955737e-01 -5.10432422e-01 4.05348241e-01 8.25596094e-01 -4.93662506e-01 -1.70801532e+00 -8.27265978e-02 -6.41280174e-01 -7.99721107e-02 -4.09502000e-01 5.41480005e-01 -5.20095706e-01 1.39608622e-01 5.29401660e-01 -1.53543307e-02 -7.93943882e-01 -1.01233888e+00 5.21568596e-01 1.04744744e+00 4.11166400e-01 -5.86126924e-01 5.16603827e-01 -1.48150519e-01 -6.22840822e-01 -8.20688665e-01 -7.70078540e-01 -5.97973466e-01 -8.14955652e-01 1.57366991e-01 9.05443907e-01 -1.12011957e+00 -3.86888944e-02 7.01342940e-01 -1.40834224e+00 9.31074321e-02 -1.31265804e-01 7.47293532e-01 1.27668634e-01 7.11446032e-02 -7.95495570e-01 -7.17987180e-01 -5.74164331e-01 -7.08781183e-01 7.16870189e-01 7.73380935e-01 -6.37131631e-02 -1.23672247e+00 4.31141585e-01 1.28589213e-01 6.86097920e-01 -2.50252277e-01 5.02218962e-01 -1.57006705e+00 2.03112125e-01 -3.75506550e-01 6.39368147e-02 4.35413122e-01 1.20548017e-01 -7.78481662e-02 -1.22280526e+00 2.12785136e-02 -3.51249874e-01 -1.92989945e-01 9.58478570e-01 5.32085225e-02 9.27440047e-01 -3.57024632e-02 -5.52977681e-01 3.33641589e-01 1.38273299e+00 3.00213844e-01 7.15096891e-01 2.45319843e-01 8.89347315e-01 2.57263422e-01 3.22385162e-01 1.35358766e-01 3.89109164e-01 3.84420037e-01 -3.74917269e-01 -2.14643881e-01 4.34476696e-02 -3.00039798e-01 4.35214669e-01 1.25054932e+00 -5.70655227e-01 -4.92629468e-01 -1.06389761e+00 5.36271036e-01 -1.53417468e+00 -1.12522495e+00 -8.18993449e-02 1.95143032e+00 8.64595413e-01 1.17270388e-01 -1.58849731e-01 -4.89739567e-01 1.14641535e+00 5.18079437e-02 -3.69579643e-01 -4.44250703e-01 -4.26035345e-01 3.59381497e-01 6.16606891e-01 8.18844885e-02 -1.22546816e+00 1.17511177e+00 6.86363792e+00 1.24806905e+00 -7.80196071e-01 2.94861764e-01 3.72401714e-01 4.13854778e-01 -2.12818325e-01 -1.06677026e-01 -1.31336439e+00 2.37223864e-01 1.52041984e+00 -4.10960913e-01 2.55206954e-02 7.14444041e-01 -1.46617264e-01 4.06832069e-01 -8.08702111e-01 6.62755251e-01 -8.31919536e-02 -9.43307579e-01 1.99266538e-01 -5.28431125e-02 8.68663013e-01 6.19273782e-01 -4.87410277e-01 8.71503234e-01 7.98319876e-01 -9.88589048e-01 2.19021738e-01 5.47111809e-01 5.27655423e-01 -6.11133635e-01 1.23539436e+00 4.71111298e-01 -1.32070148e+00 1.39524639e-01 -5.97282231e-01 4.35182601e-01 4.05524433e-01 3.00745517e-01 -8.84111643e-01 1.01966751e+00 5.39275408e-01 2.97470212e-01 -5.13490438e-01 1.18562591e+00 -1.51508838e-01 8.11324358e-01 -5.44157207e-01 -3.39669213e-02 2.30943337e-01 8.15655570e-03 3.77664149e-01 1.49726164e+00 3.11405271e-01 2.25423262e-01 -1.58766117e-02 2.47947767e-01 -5.12756228e-01 5.73104918e-01 -5.03585398e-01 -3.14039201e-01 6.63125455e-01 1.42134964e+00 -4.84826148e-01 -6.96258426e-01 -5.06246030e-01 8.75197232e-01 6.03808045e-01 2.81599969e-01 -7.73031473e-01 -7.86426187e-01 3.81441385e-01 -2.53719091e-01 5.00965297e-01 -1.29209325e-01 -1.87205344e-01 -1.55081427e+00 -3.55096847e-01 -6.02308989e-01 8.67339849e-01 -5.79797447e-01 -1.48044908e+00 1.22498417e+00 -2.30929330e-02 -9.81631517e-01 -1.78760692e-01 -6.29248917e-01 -6.97808802e-01 9.62777734e-01 -1.41961420e+00 -1.07274330e+00 2.16096640e-01 4.76673394e-01 5.09643853e-01 -2.79330999e-01 1.14998960e+00 6.57383025e-01 -8.44740868e-01 7.64992297e-01 4.10073787e-01 7.10879564e-01 1.04920256e+00 -1.30487263e+00 4.79813457e-01 6.81219876e-01 4.32498634e-01 9.25619304e-01 3.25231671e-01 -6.03218734e-01 -6.45437539e-01 -1.08122587e+00 1.63527906e+00 -6.87098145e-01 6.22930646e-01 8.82005766e-02 -1.16643953e+00 8.46895397e-01 5.15519202e-01 -2.62058806e-03 1.06030393e+00 5.11503458e-01 -3.06793809e-01 -1.02709554e-01 -1.12280941e+00 5.50393164e-01 9.24791694e-01 -3.09742689e-01 -1.08054030e+00 -1.36315078e-01 7.09247649e-01 -2.26150215e-01 -1.32560515e+00 5.41200876e-01 4.06841338e-01 -3.77797812e-01 8.02122831e-01 -1.14298224e+00 -7.41747916e-02 -1.67700171e-01 -7.13831708e-02 -1.70819700e+00 -4.09222096e-01 -2.62663513e-01 2.46555150e-01 1.86408675e+00 9.03735757e-01 -6.08097255e-01 7.68378228e-02 7.19017804e-01 -9.29009020e-02 -2.26507679e-01 -1.00937581e+00 -6.20435655e-01 3.97058547e-01 -3.67976695e-01 8.82949710e-01 1.19881284e+00 -1.65187165e-01 7.70809412e-01 -4.24873412e-01 1.79979309e-01 3.21160078e-01 -3.72659594e-01 2.21277580e-01 -1.31977355e+00 1.03487529e-01 -3.35995436e-01 -1.46432444e-01 -7.46503890e-01 2.85208285e-01 -9.68348682e-01 1.28516123e-01 -1.50567937e+00 2.09316611e-01 -5.48175871e-01 -1.12992847e+00 8.02311122e-01 -5.31691790e-01 -6.75562443e-03 1.65175632e-01 1.10847041e-01 -9.64619577e-01 4.47794944e-01 8.79179180e-01 3.00366525e-02 -1.75473318e-01 -7.25052059e-02 -6.89059436e-01 7.30118275e-01 9.08354223e-01 -7.97544956e-01 1.19248942e-01 -7.03980982e-01 7.35400543e-02 -4.98871356e-02 -2.77281374e-01 -9.78606105e-01 4.62972194e-01 -6.63829222e-02 6.61608458e-01 -4.28770661e-01 -1.26331434e-01 -8.12668681e-01 4.12929416e-01 1.06534183e-01 -3.84624779e-01 1.69833511e-01 2.76922345e-01 3.45673919e-01 -3.19232732e-01 -2.81570375e-01 6.21565402e-01 -1.80233002e-01 -1.02453721e+00 2.41636589e-01 -2.63164312e-01 3.86716664e-01 6.28552079e-01 1.92497656e-01 -1.00368753e-01 1.81186795e-01 -8.58746231e-01 4.64990437e-01 3.73181212e-03 6.66002989e-01 3.07810336e-01 -1.42908990e+00 -1.14395022e+00 7.00623468e-02 2.02315018e-01 -3.46518129e-01 4.69756186e-01 4.88453925e-01 -1.23352729e-01 3.97354245e-01 -2.18703508e-01 -2.81464159e-02 -1.15455604e+00 1.69830352e-01 4.50785846e-01 -8.62976134e-01 -2.28122592e-01 7.82243669e-01 -1.33245438e-01 -9.95858490e-01 -1.32644743e-01 -1.06185354e-01 -9.01717484e-01 2.06177652e-01 6.55131519e-01 5.39871812e-01 2.26987153e-01 -8.71796131e-01 -5.22232115e-01 2.49770880e-01 -4.25097555e-01 -8.47300738e-02 1.36097658e+00 -8.40609372e-02 1.20532796e-01 6.83702946e-01 9.39755678e-01 1.42987326e-01 -5.17183661e-01 -4.37109709e-01 3.57099503e-01 1.41046643e-01 1.08718455e-01 -1.07542717e+00 -8.80998552e-01 6.87042594e-01 2.84068942e-01 5.92099801e-02 9.43616509e-01 -2.27991894e-01 9.62350786e-01 7.88321912e-01 5.59068561e-01 -1.36071599e+00 -8.21348429e-01 8.88032794e-01 3.92069906e-01 -1.31696355e+00 -4.42517221e-01 -7.34908655e-02 -9.04361010e-01 1.21423495e+00 6.43083215e-01 1.69732515e-02 7.82759964e-01 4.28219318e-01 4.23312753e-01 1.71285555e-01 -7.61545479e-01 -2.20259488e-01 4.16814625e-01 2.57395655e-01 7.40553498e-01 3.29438448e-01 -6.86316311e-01 1.16424310e+00 1.56360120e-01 3.14544924e-02 2.03541100e-01 6.93363070e-01 -4.13779318e-01 -1.41120446e+00 -2.61411309e-01 4.63742942e-01 -8.06786001e-01 -3.92629415e-01 -3.48498315e-01 7.64115810e-01 1.85751542e-01 1.02477884e+00 -2.95473605e-01 -5.02113938e-01 6.03190124e-01 5.80140233e-01 1.36228338e-01 -5.92254281e-01 -1.03308797e+00 -2.38667261e-02 5.72289944e-01 -1.56578571e-01 -5.97765803e-01 -4.56109434e-01 -1.33606684e+00 -4.95831408e-02 -7.18689919e-01 7.45104313e-01 5.87804675e-01 1.01837158e+00 4.71370995e-01 5.82418144e-01 5.21651506e-01 -3.40401977e-01 -8.62110317e-01 -1.32944191e+00 -6.92491889e-01 2.34690681e-01 -3.14529300e-01 -4.71190423e-01 -4.12201077e-01 -1.42707884e-01]
[9.80215835571289, 9.624140739440918]
088722e9-afea-4f08-9373-a598196edb59
faceswapnet-landmark-guided-many-to-many-face
1905.11805
null
https://arxiv.org/abs/1905.11805v2
https://arxiv.org/pdf/1905.11805v2.pdf
FReeNet: Multi-Identity Face Reenactment
This paper presents a novel multi-identity face reenactment framework, named FReeNet, to transfer facial expressions from an arbitrary source face to a target face with a shared model. The proposed FReeNet consists of two parts: Unified Landmark Converter (ULC) and Geometry-aware Generator (GAG). The ULC adopts an encode-decoder architecture to efficiently convert expression in a latent landmark space, which significantly narrows the gap of the face contour between source and target identities. The GAG leverages the converted landmark to reenact the photorealistic image with a reference image of the target person. Moreover, a new triplet perceptual loss is proposed to force the GAG module to learn appearance and geometry information simultaneously, which also enriches facial details of the reenacted images. Further experiments demonstrate the superiority of our approach for generating photorealistic and expression-alike faces, as well as the flexibility for transferring facial expressions between identities.
['Yong liu', 'Liang Liu', 'Yusu Pan', 'Yu Ding', 'Xianfang Zeng', 'Mengmeng Wang', 'Changjie Fan', 'Jiangning Zhang']
2019-05-28
freenet-multi-identity-face-reenactment
http://openaccess.thecvf.com/content_CVPR_2020/html/Zhang_FReeNet_Multi-Identity_Face_Reenactment_CVPR_2020_paper.html
http://openaccess.thecvf.com/content_CVPR_2020/papers/Zhang_FReeNet_Multi-Identity_Face_Reenactment_CVPR_2020_paper.pdf
cvpr-2020-6
['face-reenactment']
['computer-vision']
[ 2.82273799e-01 3.51305217e-01 2.68691927e-01 -6.76673472e-01 -8.08489323e-01 -7.07384050e-01 6.26414895e-01 -8.57425809e-01 -3.80936414e-02 7.08875179e-01 9.77897868e-02 3.69047374e-01 4.72381622e-01 -7.86594093e-01 -7.99207270e-01 -7.86296189e-01 4.43018228e-01 7.34464452e-02 -4.35730457e-01 -2.52858818e-01 -1.57203287e-01 8.20986927e-01 -1.56815219e+00 2.58754730e-01 5.95956326e-01 1.19067228e+00 -3.03562284e-01 4.14397538e-01 9.64981243e-02 6.11433625e-01 -4.52699751e-01 -9.23036754e-01 5.83019137e-01 -6.68527007e-01 -4.22628790e-01 2.19688654e-01 9.60920930e-01 -4.97720808e-01 -3.55134577e-01 9.62855458e-01 5.40573776e-01 5.07444069e-02 5.25401473e-01 -1.58837891e+00 -1.17910516e+00 -3.04074734e-02 -8.45344782e-01 -5.30481219e-01 4.07576144e-01 1.16075478e-01 7.98079312e-01 -9.88058627e-01 8.20346534e-01 1.66721320e+00 5.79726040e-01 8.84916902e-01 -1.21091092e+00 -1.18320763e+00 1.36735752e-01 -2.46617287e-01 -1.75094044e+00 -9.65316594e-01 9.79869783e-01 -2.68194556e-01 3.37617099e-01 1.05662614e-01 4.96692389e-01 1.13515913e+00 -1.96057628e-03 2.95228809e-01 1.11428452e+00 -2.47207075e-01 -1.59968734e-01 2.75575127e-02 -7.65141845e-01 9.83936191e-01 -1.52935416e-01 3.57382447e-01 -5.99041522e-01 6.00598641e-02 1.27503443e+00 -1.22744828e-01 -2.70800352e-01 -3.40307206e-01 -6.81780696e-01 6.25876427e-01 5.60822666e-01 -1.64978448e-02 -2.38554135e-01 2.07685992e-01 1.18108027e-01 1.93647116e-01 4.88770902e-01 1.35158837e-01 -9.52136591e-02 2.23985851e-01 -8.80668879e-01 9.60903391e-02 5.06569803e-01 1.16230261e+00 1.13483655e+00 2.15526447e-01 -3.50105166e-01 9.13841665e-01 4.17878926e-01 6.24247432e-01 5.11840433e-02 -1.19064045e+00 2.56996065e-01 6.17355108e-01 -9.99555066e-02 -1.16512287e+00 1.43184707e-01 -8.93921405e-02 -8.44922543e-01 4.05277371e-01 -3.93551476e-02 -2.80420959e-01 -7.72006333e-01 2.30564952e+00 4.87579018e-01 3.24324369e-01 1.20261848e-01 9.85104203e-01 9.90879238e-01 5.72020710e-01 9.64081809e-02 1.49126807e-02 1.23307526e+00 -9.75626111e-01 -5.44772148e-01 -5.41240275e-02 1.30682990e-01 -8.08566630e-01 9.44552481e-01 -1.30733386e-01 -1.33851707e+00 -6.57702982e-01 -9.63691652e-01 -6.05303049e-01 -1.10795395e-02 4.34473693e-01 5.17653883e-01 6.75081015e-01 -1.40595770e+00 3.05956125e-01 -4.22801375e-01 -8.10534656e-02 6.42540395e-01 5.00291109e-01 -7.93472767e-01 -3.36387120e-02 -1.02631176e+00 5.00967026e-01 -2.05115657e-02 2.50836015e-01 -9.09726918e-01 -9.11698461e-01 -1.33710051e+00 7.68074766e-02 -2.38046399e-03 -8.31038952e-01 9.80077565e-01 -1.64359939e+00 -1.98070776e+00 1.34927416e+00 -2.79861569e-01 2.57010490e-01 6.83753431e-01 2.44961515e-01 -3.73635501e-01 1.26054257e-01 2.37913072e-01 1.15629935e+00 1.17424321e+00 -1.43383884e+00 -3.64534885e-01 -4.63754356e-01 1.02708407e-01 3.10883701e-01 1.84502024e-02 -2.37565544e-02 -7.72007287e-01 -9.21244562e-01 -5.08228093e-02 -9.02988136e-01 1.74456090e-01 6.00396931e-01 -5.01519859e-01 1.44432142e-01 6.34475648e-01 -6.52435243e-01 5.44663668e-01 -2.36174583e+00 2.69835651e-01 2.50497758e-01 1.92576274e-01 -9.95026231e-02 -7.10460961e-01 1.82218850e-01 -4.78720009e-01 -8.78367946e-02 -1.15766101e-01 -8.48299384e-01 -7.96725303e-02 8.91089961e-02 -4.01506931e-01 5.03261149e-01 2.96457350e-01 1.28144646e+00 -6.26076221e-01 -4.76015508e-01 -1.45498505e-02 8.75358164e-01 -6.49939179e-01 4.22952354e-01 3.70387062e-02 7.90346146e-01 -2.69709945e-01 6.03521585e-01 1.17193460e+00 4.18858714e-02 1.47938505e-01 -5.17908454e-01 -7.21446276e-02 -2.46486098e-01 -8.18978429e-01 1.86418784e+00 -8.17881703e-01 3.27825725e-01 4.08761531e-01 -2.35538244e-01 1.25214458e+00 2.80191272e-01 4.00373280e-01 -7.41328180e-01 2.49750242e-01 8.03052355e-03 -4.48356390e-01 -7.53844604e-02 3.02166671e-01 -5.31081736e-01 1.06835049e-02 4.50121552e-01 2.06911504e-01 -1.82051003e-01 -1.66130796e-01 1.46789374e-02 4.97195154e-01 4.84122097e-01 -4.36367244e-02 -1.40497968e-01 9.53369319e-01 -7.16164351e-01 7.86158204e-01 -9.47724581e-02 6.27468824e-02 8.01865995e-01 5.04847229e-01 -3.51557761e-01 -1.01895273e+00 -1.40455449e+00 1.97086439e-01 9.98760164e-01 2.23696858e-01 -2.32867494e-01 -8.76892865e-01 -4.97313082e-01 -2.21564267e-02 5.07691979e-01 -9.05912399e-01 -2.61112511e-01 -4.02047098e-01 -1.20197751e-01 6.95186555e-01 4.32552576e-01 8.36360812e-01 -9.04893160e-01 -1.01437882e-01 -3.10888469e-01 -2.12948650e-01 -1.16471410e+00 -1.15074313e+00 -6.84219182e-01 -3.12712640e-01 -1.05892301e+00 -7.38277614e-01 -9.52185512e-01 1.18515098e+00 3.92351635e-02 7.42982268e-01 1.26300707e-01 -2.51522154e-01 2.59141594e-01 -1.07470108e-02 5.93006937e-03 -5.59185207e-01 -3.13820213e-01 2.54244246e-02 8.01180959e-01 -9.96708125e-02 -7.91857481e-01 -6.75678134e-01 3.37599427e-01 -8.22604835e-01 4.31964308e-01 4.24106747e-01 5.45256019e-01 5.57238162e-01 -3.58465135e-01 4.71534491e-01 -6.73323333e-01 3.60670298e-01 -1.19755939e-01 -7.03306496e-01 3.65231395e-01 -2.22842813e-01 3.04729887e-03 6.69257402e-01 -2.12351263e-01 -1.39685977e+00 2.13908166e-01 -2.33654618e-01 -7.90307522e-01 1.74577013e-01 -1.88473597e-01 -7.55457342e-01 -4.90057051e-01 2.50399172e-01 1.62039265e-01 3.32892478e-01 -3.16427916e-01 7.66039133e-01 4.72467035e-01 9.46475685e-01 -7.60913014e-01 1.04608512e+00 6.57202482e-01 6.76511377e-02 -5.22949398e-01 -6.63788795e-01 1.11006476e-01 -8.86357903e-01 -1.38279870e-01 8.81620288e-01 -1.13476944e+00 -9.30468738e-01 7.19579279e-01 -1.38246310e+00 -1.88924164e-01 -4.23915505e-01 -8.44526589e-02 -7.69100130e-01 1.82911441e-01 -5.86198747e-01 -3.84239763e-01 -4.43537533e-01 -1.15412474e+00 1.47278976e+00 5.25508821e-01 8.05399269e-02 -8.57285857e-01 6.20639436e-02 3.19595456e-01 2.37343520e-01 5.67129076e-01 8.28426003e-01 1.15674660e-01 -6.45381629e-01 -1.38279442e-02 -5.31726897e-01 4.94670421e-01 5.42132676e-01 7.79443979e-02 -1.21893632e+00 -4.69069958e-01 -2.50742167e-01 -5.03884792e-01 4.41449523e-01 4.42519225e-02 8.92995596e-01 -4.65656638e-01 -1.98809922e-01 1.17159867e+00 1.25748491e+00 1.31243497e-01 9.04088378e-01 -2.64889151e-01 8.27074409e-01 6.71284556e-01 1.92233130e-01 2.57380545e-01 5.24655700e-01 8.14511120e-01 8.92901495e-02 -5.25824010e-01 -4.22756910e-01 -6.78736985e-01 4.25452620e-01 5.23503482e-01 -1.02882445e-01 8.10127184e-02 -2.67346382e-01 1.73114970e-01 -1.29918647e+00 -8.82956028e-01 6.32048786e-01 2.00917983e+00 9.18051898e-01 -7.08228171e-01 -2.67950773e-01 -4.85083014e-01 8.31133544e-01 1.75043628e-01 -7.51231968e-01 -4.07448858e-01 -2.04143286e-01 4.29078758e-01 8.64395276e-02 6.66701198e-01 -8.59720469e-01 1.26387894e+00 5.94741344e+00 6.22796655e-01 -1.35806739e+00 1.25385284e-01 9.00794685e-01 -8.99114534e-02 -3.61308515e-01 -2.30702102e-01 -6.09820664e-01 2.51254529e-01 5.82866609e-01 -4.22675312e-01 4.92497802e-01 8.20556939e-01 2.04553455e-02 5.07716477e-01 -1.13960624e+00 1.36208916e+00 4.80428517e-01 -1.33830392e+00 4.79019403e-01 2.81362627e-02 7.98151731e-01 -6.11426234e-01 3.54140848e-01 1.84172243e-01 1.99593410e-01 -1.23674238e+00 8.05978060e-01 6.58529162e-01 1.73158419e+00 -9.54886973e-01 3.09622914e-01 -3.80048424e-01 -1.37478828e+00 2.12423965e-01 -2.86526561e-01 1.92637920e-01 2.07941592e-01 -1.24985561e-01 -5.77461660e-01 7.04094350e-01 4.25551713e-01 6.25633299e-01 -4.83768821e-01 3.99658442e-01 -4.86920863e-01 -1.57312170e-01 -9.88518447e-02 7.26881981e-01 3.45807319e-04 -6.02883995e-01 3.17532301e-01 7.60941327e-01 4.34006333e-01 1.62973106e-01 3.83900777e-02 1.35322464e+00 -5.75696230e-01 2.18735069e-01 -6.11004591e-01 2.88778692e-01 5.67408741e-01 1.58875978e+00 -2.73159385e-01 -7.86098167e-02 -3.49092662e-01 1.48966765e+00 2.88434893e-01 4.40010995e-01 -9.42232609e-01 -2.10279733e-01 1.08273137e+00 1.00638214e-02 8.25637951e-02 1.36898011e-01 7.29234293e-02 -1.07244885e+00 4.37297486e-02 -8.57967913e-01 3.46030332e-02 -1.12296641e+00 -1.24270225e+00 1.01462710e+00 -2.17119396e-01 -1.08881259e+00 -1.89416021e-01 -3.67765397e-01 -7.53479779e-01 1.29025400e+00 -1.41820908e+00 -1.95460522e+00 -6.35897338e-01 8.37343216e-01 1.19874090e-01 -2.29881987e-01 7.52022684e-01 3.25179726e-01 -6.37334585e-01 1.11376297e+00 -3.31511408e-01 4.14805293e-01 8.58101368e-01 -7.55969405e-01 5.72616160e-01 6.46024346e-01 2.37971712e-02 6.65663660e-01 5.80424443e-02 -5.01191974e-01 -1.14442813e+00 -1.51983750e+00 6.15807056e-01 -2.56285042e-01 1.46626920e-01 -6.15244806e-01 -6.01664066e-01 9.21002269e-01 1.58441067e-01 1.58243075e-01 7.76848257e-01 -3.40066701e-01 -8.18408072e-01 -2.36367360e-01 -1.31917369e+00 9.27851379e-01 1.09684038e+00 -7.89842010e-01 -1.80691853e-01 2.65945327e-02 6.55886769e-01 -5.72315156e-01 -8.35509300e-01 2.60775626e-01 8.47750783e-01 -1.03340137e+00 1.00742006e+00 -4.37668324e-01 5.43334246e-01 -2.83575386e-01 -1.67108759e-01 -1.26131558e+00 -2.44049862e-01 -9.93599117e-01 4.37897682e-01 1.57130277e+00 1.22124039e-01 -6.77544057e-01 7.62452126e-01 7.34091759e-01 2.62652826e-03 -7.22466946e-01 -1.12838817e+00 -5.87472916e-01 3.00861716e-01 -4.70777117e-02 1.12716568e+00 9.05487239e-01 -5.04491806e-01 4.27571654e-01 -4.97548729e-01 1.82772651e-01 5.79489946e-01 2.47728631e-01 8.96620333e-01 -9.43681359e-01 8.39725956e-02 -2.51641005e-01 -4.40446764e-01 -9.99349833e-01 7.51350164e-01 -1.28155529e+00 -1.46671772e-01 -9.90947425e-01 1.07643209e-01 -4.26872045e-01 7.56187737e-02 7.01780200e-01 9.48010832e-02 7.96560526e-01 4.08588856e-01 6.15838394e-02 -1.00908071e-01 1.06074607e+00 1.67043889e+00 -8.07847604e-02 -1.54469326e-01 -2.51251191e-01 -9.17067528e-01 6.71991706e-01 3.98778737e-01 -1.28132746e-01 -5.80918014e-01 -4.31522250e-01 -1.47462398e-01 5.32840863e-02 6.72427058e-01 -7.70955801e-01 3.24005485e-02 -4.50391620e-02 6.74245238e-01 -1.27750412e-01 6.17752790e-01 -5.83013892e-01 4.93923515e-01 -7.75300562e-02 -3.06182146e-01 7.33547732e-02 2.73244649e-01 2.48207361e-01 -1.53561354e-01 2.59403169e-01 1.21870041e+00 1.93369418e-01 -4.70582336e-01 8.25910389e-01 4.35216129e-01 -3.86282578e-02 1.12956798e+00 -3.83723557e-01 -4.08837944e-02 -5.36034763e-01 -6.69923723e-01 3.81912179e-02 8.77612293e-01 7.68035412e-01 7.78728068e-01 -1.72217762e+00 -9.02684569e-01 7.89951503e-01 1.32836074e-01 -1.10915452e-01 6.09934986e-01 5.12077272e-01 -5.92281997e-01 1.47526905e-01 -6.44151926e-01 -2.91368335e-01 -1.30679846e+00 4.41114157e-01 7.20996916e-01 2.26863295e-01 -5.68266273e-01 1.05062139e+00 8.67192328e-01 -6.49092555e-01 -1.74666837e-01 2.01045990e-01 2.17009574e-01 -1.00950010e-01 5.71571827e-01 5.19947447e-02 -2.98846722e-01 -1.35399795e+00 -2.81378329e-01 9.88619328e-01 -4.73701917e-02 -2.56177962e-01 1.01427412e+00 -2.63505101e-01 -4.79712605e-01 -7.26568475e-02 1.55154002e+00 2.24054873e-01 -1.62240505e+00 -9.74146500e-02 -6.27711713e-01 -7.01008677e-01 -1.93413585e-01 -6.96100950e-01 -1.57289338e+00 7.39065230e-01 5.36139369e-01 -8.23244452e-01 1.31809962e+00 -1.32633895e-01 8.33192587e-01 -1.60905033e-01 5.05152106e-01 -6.65275037e-01 2.77344346e-01 2.49877021e-01 1.35552943e+00 -8.70493054e-01 -2.84255594e-01 -6.85414195e-01 -7.56677568e-01 1.01659763e+00 8.08159053e-01 -6.40045106e-02 5.55430174e-01 1.05644688e-01 2.89674371e-01 -1.47372916e-01 -5.25677562e-01 6.27065077e-02 3.28106493e-01 8.42411160e-01 3.05127263e-01 -2.06421360e-01 2.76881278e-01 6.79332316e-01 -4.70412642e-01 6.42013028e-02 2.73861080e-01 3.26631010e-01 3.57349604e-01 -1.20851433e+00 -2.35093281e-01 -1.68781415e-01 -1.43657625e-01 -4.47533019e-02 -4.83841777e-01 7.99788058e-01 4.99743700e-01 6.66830063e-01 3.72492343e-01 -3.46607983e-01 4.02425349e-01 -4.76781130e-02 7.71275461e-01 -6.59529388e-01 -3.38557661e-01 4.19493206e-02 -3.58491153e-01 -7.72231579e-01 -2.02381581e-01 -2.77665943e-01 -1.12787604e+00 -4.15823370e-01 1.96393341e-01 4.06871438e-02 3.51051122e-01 5.00653028e-01 7.18284905e-01 2.11221799e-01 9.22358811e-01 -1.04306269e+00 -2.36185178e-01 -5.68586826e-01 -7.53651261e-01 5.54470658e-01 3.24538797e-01 -7.89200962e-01 -1.78538308e-01 3.44620824e-01]
[12.696833610534668, -0.11263283342123032]
c836843a-31df-48d6-83a4-a4f6a1d2ae2d
high-dimensional-causal-discovery-learning
2211.14221
null
https://arxiv.org/abs/2211.14221v2
https://arxiv.org/pdf/2211.14221v2.pdf
Learning Large Causal Structures from Inverse Covariance Matrix via Matrix Decomposition
Learning causal structures from observational data is a fundamental yet highly complex problem when the number of variables is large. In this paper, we start from linear structural equation models (SEMs) and investigate ways of learning causal structures from the inverse covariance matrix. The proposed method, called $\mathcal{O}$-ICID (for {\it Independence-preserving} Decomposition from Oracle Inverse Covariance matrix), is based on continuous optimization of a type of matrix decomposition that preserves the nonzero patterns of the inverse covariance matrix. We show that $\mathcal{O}$-ICID provides an efficient way for identifying the true directed acyclic graph (DAG) under the knowledge of noise variances. With weaker prior information, the proposed method gives directed graph solutions that are useful for making more refined causal discovery. The proposed method enjoys a low complexity when the true DAG has bounded node degrees, as reflected by its time efficiency in experiments in comparison with state-of-the-art algorithms.
['Michèle Sebag', 'Koji Maruhashi', 'Yusuke Koyanagi', 'Shuang Chang', 'Akito Fujii', 'Kento Uemura', 'Shuyu Dong']
2022-11-25
null
null
null
null
['causal-discovery']
['knowledge-base']
[ 4.46818054e-01 4.01936799e-01 -3.11320990e-01 -3.77776116e-01 -3.82485896e-01 -5.48697650e-01 3.00644308e-01 1.90433651e-01 -2.05662027e-01 7.79909194e-01 2.24214435e-01 -9.41819310e-01 -1.05967164e+00 -9.49482501e-01 -1.05088949e+00 -8.70385230e-01 -5.88765323e-01 6.35769010e-01 -7.57572427e-02 1.23139553e-01 1.37553468e-01 2.41387695e-01 -9.91097867e-01 -2.11967856e-01 9.62322712e-01 4.61907774e-01 4.91214618e-02 4.57370371e-01 1.48417249e-01 8.00523400e-01 -4.77036238e-02 -4.99325246e-01 2.46815190e-01 -6.72166586e-01 -7.15314925e-01 -2.15954363e-01 2.58201957e-01 4.21748161e-02 -4.27179486e-01 1.33573115e+00 1.26221746e-01 -1.44885793e-01 9.52663422e-01 -1.36136675e+00 -5.09833276e-01 1.07146347e+00 -8.03027332e-01 2.76220024e-01 2.93587178e-01 -2.92792827e-01 1.45344889e+00 -5.44618666e-01 5.92021644e-01 1.46800756e+00 3.47371608e-01 -1.19590166e-03 -1.82503843e+00 -9.82527614e-01 3.01501840e-01 3.85746002e-01 -1.40067840e+00 -9.45452154e-02 6.63734734e-01 -6.42721236e-01 4.37089384e-01 4.98860329e-01 1.67887568e-01 9.44690168e-01 1.64176375e-01 3.71149480e-01 1.31391621e+00 -5.02822697e-01 3.58654559e-01 -1.30379707e-01 5.04972696e-01 1.06905746e+00 1.02846229e+00 3.22779775e-01 -4.53090221e-01 -4.72855419e-01 8.09485018e-01 -1.71009555e-01 -2.42927358e-01 -5.83216488e-01 -9.76142108e-01 8.67122471e-01 3.08646649e-01 7.36105219e-02 -2.29605213e-01 3.24153781e-01 -2.10138932e-01 3.58784318e-01 4.62627187e-02 1.75725847e-01 -2.61490524e-01 2.48561651e-01 -5.14723778e-01 8.52831528e-02 1.02738404e+00 9.48826432e-01 6.39488697e-01 -7.93123320e-02 1.45592272e-01 3.46349835e-01 5.90194941e-01 8.74938667e-01 -4.59953129e-01 -7.84184277e-01 6.24911010e-01 7.72022009e-01 9.74757671e-02 -1.29415405e+00 -5.61679780e-01 -4.58862722e-01 -1.37036610e+00 -7.72410035e-02 6.50246441e-01 -4.29568142e-01 -7.34066188e-01 2.04902720e+00 4.13785100e-01 4.19671208e-01 -2.03517362e-01 7.19948947e-01 4.23899204e-01 4.50067610e-01 -9.48842838e-02 -5.67514420e-01 1.19100618e+00 -1.04641527e-01 -7.95377195e-01 -2.30942681e-01 2.93741643e-01 -2.77570605e-01 7.31051564e-01 3.72774571e-01 -6.42835259e-01 -2.50593014e-02 -9.30943608e-01 3.33981693e-01 9.72925648e-02 -1.91188738e-01 9.05894041e-01 1.09865332e+00 -7.84183681e-01 4.13526893e-01 -8.74546051e-01 -1.22494452e-01 1.61711127e-01 6.47763133e-01 -4.00721282e-01 -1.97267577e-01 -1.10379803e+00 3.46129537e-01 1.30002588e-01 1.51748657e-01 -1.13399601e+00 -7.22032726e-01 -7.57561803e-01 2.21962437e-01 9.26614523e-01 -7.01959610e-01 6.65031135e-01 -4.66611832e-01 -1.01230514e+00 3.39035124e-01 -3.74136508e-01 -1.78512380e-01 2.43982702e-01 1.19386073e-02 -2.08287433e-01 -8.86927769e-02 9.20656100e-02 -3.27225119e-01 9.50168908e-01 -1.31111968e+00 -3.82550538e-01 -8.42238486e-01 3.57599258e-02 -2.78886765e-01 -2.67377436e-01 -2.12274849e-01 -1.41237393e-01 -4.50691313e-01 4.53120053e-01 -1.08510268e+00 -4.99020070e-01 -4.60266262e-01 -8.08853328e-01 -3.97884026e-02 2.59012282e-01 -5.13758838e-01 1.56654024e+00 -1.82021022e+00 4.88737792e-01 8.25088203e-01 4.15180802e-01 -3.67592812e-01 -1.16562461e-02 6.43962562e-01 -3.83422554e-01 2.80247092e-01 -4.35177535e-01 1.09852940e-01 -1.21653244e-01 2.61751264e-01 -1.15679882e-01 7.82179952e-01 -2.16643028e-02 4.37671363e-01 -8.83942246e-01 -3.46087039e-01 -1.42649353e-01 -5.75274751e-02 -6.70000553e-01 1.96296573e-01 -9.47955698e-02 5.01616895e-01 -6.58376336e-01 2.09313735e-01 6.99638426e-01 -4.46771234e-01 1.01978719e+00 3.02261170e-02 -2.60442328e-02 8.21256936e-02 -1.92057645e+00 1.25354040e+00 -1.25400230e-01 1.72875658e-01 2.32941568e-01 -1.40179563e+00 6.50926292e-01 2.71536946e-01 4.04603362e-01 -4.10300404e-01 2.25800708e-01 -4.67673019e-02 4.92328733e-01 -5.15831590e-01 -2.77695686e-01 -1.02912009e-01 -1.95615992e-01 5.38253009e-01 8.78286436e-02 4.61612314e-01 4.69432294e-01 5.28711915e-01 1.57268107e+00 -3.15183252e-01 4.68290716e-01 -6.90152645e-01 4.45725083e-01 -3.34566116e-01 8.27252626e-01 1.16416216e+00 4.20522988e-01 -5.38892187e-02 1.30265582e+00 -7.72265121e-02 -8.18606198e-01 -1.34316778e+00 -6.73967227e-02 5.47757208e-01 2.72353813e-02 -5.05719662e-01 -6.69650137e-01 -6.36132538e-01 1.32118165e-01 6.92211628e-01 -8.99988532e-01 -1.21120997e-02 -3.44182432e-01 -1.13694990e+00 3.27473938e-01 2.45945886e-01 1.66858107e-01 -2.87809610e-01 2.99001113e-02 -9.12303478e-03 -1.57950819e-01 -8.74329925e-01 -1.73665717e-01 4.12704557e-01 -1.11648333e+00 -1.47180355e+00 -1.22590467e-01 -3.28210652e-01 9.21992362e-01 1.86992422e-01 8.42372298e-01 -2.71024257e-01 -7.45141134e-02 2.26488709e-01 -1.63984179e-01 -4.11339462e-01 -3.27664822e-01 -1.36367589e-01 1.21964224e-01 4.41222459e-01 8.98604766e-02 -1.10125709e+00 -3.38390589e-01 2.08972514e-01 -8.37810695e-01 -9.08595547e-02 7.15932369e-01 8.83840501e-01 5.65510929e-01 5.09350240e-01 3.64340365e-01 -1.48484516e+00 5.45114279e-01 -6.19435132e-01 -1.16229415e+00 2.57885456e-01 -9.78126526e-01 7.60657430e-01 5.58592558e-01 -6.56899512e-02 -1.20909333e+00 1.04523003e-01 3.58193099e-01 -9.94513929e-02 7.79094994e-02 8.42216253e-01 -4.40868527e-01 1.49984539e-01 6.28827035e-01 -1.43145993e-01 -2.52548575e-01 -7.23271847e-01 5.18731534e-01 2.48637289e-01 3.48950893e-01 -6.41400933e-01 9.43050861e-01 4.58070040e-01 8.32696736e-01 -6.89197600e-01 -7.67757416e-01 -2.87607372e-01 -6.57005072e-01 -2.59809499e-03 6.19874477e-01 -6.91476643e-01 -1.25180006e+00 -5.16438335e-02 -8.45710278e-01 -1.04463466e-01 3.07603747e-01 7.03811109e-01 -1.98156893e-01 3.43580782e-01 -3.09510946e-01 -1.14061117e+00 2.88034409e-01 -7.51904011e-01 5.99829257e-01 -9.40032750e-02 -7.95638189e-02 -1.06358385e+00 3.67348522e-01 3.47248465e-01 -4.30431813e-01 2.49416307e-01 1.49063253e+00 -4.10766423e-01 -8.54423225e-01 -1.09359540e-01 -3.68551105e-01 -1.06064662e-01 9.54061821e-02 -2.01565400e-01 -5.31266391e-01 -1.68262944e-01 -8.18081647e-02 2.82308966e-01 7.50183582e-01 6.21324360e-01 1.09771705e+00 -6.59125149e-01 -5.21983325e-01 4.20521736e-01 1.42797267e+00 2.32145965e-01 3.96459579e-01 -2.10286915e-01 8.32094252e-01 6.35172248e-01 2.97525287e-01 6.08720601e-01 3.51589590e-01 3.68034512e-01 6.92797244e-01 -8.65140185e-02 4.05140609e-01 -4.06433523e-01 1.61724418e-01 6.48259401e-01 -2.99066186e-01 -3.50711286e-01 -8.32196474e-01 3.96875232e-01 -2.09596062e+00 -9.93571460e-01 -9.29413736e-01 2.47662425e+00 7.32248247e-01 -1.21509610e-02 1.12612613e-01 1.72030598e-01 5.33534825e-01 -3.46262939e-02 -4.06143188e-01 -2.26654246e-01 -2.61905268e-02 4.63538289e-01 8.69907737e-01 7.90196478e-01 -7.76884854e-01 4.46698159e-01 5.80814457e+00 5.70244312e-01 -3.08158785e-01 2.25955293e-01 2.71113038e-01 1.69187352e-01 -4.75597233e-01 4.44920421e-01 -5.93957126e-01 3.37552637e-01 1.01242757e+00 -2.24686444e-01 5.34227729e-01 4.75627303e-01 5.58118224e-01 -3.04543614e-01 -1.20738053e+00 6.99744940e-01 -2.50625044e-01 -1.09273005e+00 -2.50312954e-01 5.27673602e-01 9.36446369e-01 -4.54666287e-01 -9.31530222e-02 -2.18344390e-01 1.09998441e+00 -1.02463377e+00 3.07020456e-01 4.35250789e-01 6.85251474e-01 -8.10613692e-01 5.91207743e-01 5.00680327e-01 -1.05292904e+00 -3.54276955e-01 -2.11136594e-01 -2.45770305e-01 9.41746905e-02 1.12725055e+00 -7.51077294e-01 9.34335768e-01 6.07030690e-01 5.25619864e-01 -3.44714731e-01 6.58365607e-01 -5.65804601e-01 1.12249327e+00 -4.27442640e-01 -1.57261845e-02 -1.45709366e-01 -6.46910548e-01 6.59863591e-01 9.64201331e-01 2.52879173e-01 4.51924443e-01 9.51180793e-03 8.67805302e-01 -1.89869687e-01 5.47758192e-02 -7.33125925e-01 -1.90671667e-01 4.08620685e-01 9.05134022e-01 -7.49623060e-01 -1.45974746e-02 -2.43345276e-01 5.44347405e-01 3.63448650e-01 3.58450472e-01 -7.11785734e-01 -3.30902375e-02 5.72252870e-01 2.44295876e-02 1.17546044e-01 -3.91429514e-01 -4.60765690e-01 -1.11767042e+00 -1.27772138e-01 -8.01459134e-01 8.59483361e-01 -2.25229383e-01 -1.17823017e+00 1.47430226e-02 3.20752859e-01 -7.33301103e-01 -1.42535299e-01 -5.63185453e-01 -2.43550792e-01 6.72933817e-01 -8.36097419e-01 -7.19362497e-01 -7.36326873e-02 7.32252836e-01 -8.76117721e-02 2.00613886e-01 7.42023468e-01 3.48855168e-01 -8.91572893e-01 3.70666206e-01 3.32281083e-01 2.26070508e-02 4.05150682e-01 -1.54345810e+00 -1.29881039e-01 1.14894772e+00 3.29107851e-01 7.71471739e-01 1.03448057e+00 -9.31523919e-01 -1.89597654e+00 -8.44958186e-01 9.35633063e-01 -3.94276321e-01 1.03833246e+00 -5.70364237e-01 -5.79455435e-01 8.91118288e-01 3.17898095e-02 -3.83700937e-01 7.17777491e-01 7.89256036e-01 -3.88181329e-01 -3.49182367e-01 -7.14906871e-01 6.70390606e-01 1.53472483e+00 -2.15296522e-01 -3.53520364e-01 3.70955884e-01 5.76927185e-01 8.05180520e-02 -9.50148761e-01 4.28783327e-01 5.07229209e-01 -7.79827952e-01 8.81479561e-01 -9.38838780e-01 3.67480248e-01 -3.62131834e-01 -1.49494946e-01 -1.10686290e+00 -6.50148928e-01 -7.12022483e-01 -2.38193981e-02 1.10768127e+00 7.13745832e-01 -6.57610714e-01 7.11811960e-01 5.02905369e-01 4.51797128e-01 -2.30071470e-01 -9.27795589e-01 -8.19387138e-01 -2.90211588e-01 -4.94374245e-01 2.03076541e-01 1.11523664e+00 -6.51337802e-02 7.57779241e-01 -6.31770551e-01 7.20334828e-01 1.28660524e+00 1.83477744e-01 8.75564396e-01 -1.58923304e+00 -5.98555207e-01 -8.76617506e-02 -2.81786650e-01 -7.82262981e-01 1.33696914e-01 -8.66829813e-01 -2.77426332e-01 -1.29503214e+00 5.51727295e-01 -6.01412475e-01 -3.20594400e-01 4.29137349e-01 -2.50636071e-01 -3.64953309e-01 -2.17739232e-02 -1.89821124e-01 -2.31868401e-01 3.24602127e-01 9.33618724e-01 -4.58274372e-02 -2.00178757e-01 2.15752900e-01 -8.07947218e-01 7.04792619e-01 3.79486620e-01 -1.12043107e+00 -7.14599073e-01 -1.48895890e-01 6.30188465e-01 7.06953287e-01 4.57609355e-01 -3.72489631e-01 2.43023276e-01 -5.30604720e-01 -7.53372312e-02 -3.83759797e-01 -5.48663624e-02 -9.37181234e-01 8.30216885e-01 6.11946285e-01 -5.02178371e-01 4.60125655e-02 -1.66933253e-01 1.02583051e+00 1.79132551e-01 -3.37365657e-01 4.11160558e-01 1.26676589e-01 -1.74237445e-01 1.29347861e-01 -3.59291345e-01 3.55617367e-02 6.21038616e-01 4.10902113e-01 -1.99282870e-01 -5.09319544e-01 -8.35184991e-01 2.63825148e-01 -7.09577352e-02 1.56620726e-01 3.18082839e-01 -9.78752196e-01 -7.78349757e-01 -4.46149632e-02 -1.88712135e-01 -2.93207705e-01 2.65257865e-01 1.15053105e+00 1.67894177e-02 5.78314006e-01 2.39618987e-01 -3.82064283e-01 -1.41480434e+00 7.27180541e-01 -6.60136417e-02 -5.77058315e-01 -2.43605897e-01 7.06205428e-01 6.35143816e-01 -2.84722239e-01 1.98165048e-02 -7.12609738e-02 -8.41344818e-02 -6.12999387e-02 2.24074230e-01 7.13072836e-01 -2.09811404e-01 -2.03072533e-01 -3.42506588e-01 2.82244325e-01 3.10129702e-01 -3.06735069e-01 1.44802463e+00 -2.55622417e-01 -6.43917084e-01 4.88457382e-01 9.55784202e-01 3.08587462e-01 -9.27202821e-01 -2.53331363e-01 3.53820086e-01 -5.88909566e-01 3.17404754e-02 -6.80146933e-01 -1.03586805e+00 6.74027324e-01 4.28101271e-01 3.13531071e-01 9.82334673e-01 1.35923237e-01 -1.14373282e-01 5.14017463e-01 4.44154650e-01 -5.38252771e-01 -2.87979692e-01 2.61472404e-01 7.01848149e-01 -9.74537790e-01 2.45357975e-01 -8.50026309e-01 -1.21965408e-01 7.92300880e-01 2.53615022e-01 -2.31116563e-01 9.03984666e-01 3.29276800e-01 -5.49570560e-01 -3.93324167e-01 -7.62811840e-01 -2.89237052e-01 2.49948725e-01 5.01657307e-01 2.17541128e-01 4.20450628e-01 -6.78940535e-01 7.92531133e-01 -1.48863316e-01 -3.21384043e-01 5.88851035e-01 4.12404865e-01 -7.77930841e-02 -1.27237427e+00 -5.50040185e-01 6.49147034e-01 -4.14189309e-01 -1.74867913e-01 -6.40164435e-01 8.53337407e-01 -6.61388636e-02 1.28537023e+00 -3.88666302e-01 -2.27285191e-01 3.54642570e-01 -1.42855898e-01 6.07274055e-01 -3.64803910e-01 2.11750269e-02 1.65340543e-01 3.02151650e-01 -5.23281932e-01 -3.42299849e-01 -9.73107934e-01 -9.94707346e-01 -4.30696040e-01 -3.25906068e-01 2.87861943e-01 4.49510545e-01 1.02137232e+00 2.89744288e-01 5.67411125e-01 7.60076284e-01 -1.52164087e-01 -5.54426432e-01 -7.09377825e-01 -8.94453883e-01 1.50572345e-01 2.02801041e-02 -8.57828856e-01 -3.95095050e-01 1.44090191e-01]
[7.749428749084473, 5.290071487426758]
eac7a9e5-bd5d-42e1-91d0-f5774061c62d
multi-oriented-scene-text-detection-via
1802.08948
null
http://arxiv.org/abs/1802.08948v2
http://arxiv.org/pdf/1802.08948v2.pdf
Multi-Oriented Scene Text Detection via Corner Localization and Region Segmentation
Previous deep learning based state-of-the-art scene text detection methods can be roughly classified into two categories. The first category treats scene text as a type of general objects and follows general object detection paradigm to localize scene text by regressing the text box locations, but troubled by the arbitrary-orientation and large aspect ratios of scene text. The second one segments text regions directly, but mostly needs complex post processing. In this paper, we present a method that combines the ideas of the two types of methods while avoiding their shortcomings. We propose to detect scene text by localizing corner points of text bounding boxes and segmenting text regions in relative positions. In inference stage, candidate boxes are generated by sampling and grouping corner points, which are further scored by segmentation maps and suppressed by NMS. Compared with previous methods, our method can handle long oriented text naturally and doesn't need complex post processing. The experiments on ICDAR2013, ICDAR2015, MSRA-TD500, MLT and COCO-Text demonstrate that the proposed algorithm achieves better or comparable results in both accuracy and efficiency. Based on VGG16, it achieves an F-measure of 84.3% on ICDAR2015 and 81.5% on MSRA-TD500.
['Xiang Bai', 'Wenhao Wu', 'Cong Yao', 'Pengyuan Lyu', 'Shuicheng Yan']
2018-02-25
multi-oriented-scene-text-detection-via-1
http://openaccess.thecvf.com/content_cvpr_2018/html/Lyu_Multi-Oriented_Scene_Text_CVPR_2018_paper.html
http://openaccess.thecvf.com/content_cvpr_2018/papers/Lyu_Multi-Oriented_Scene_Text_CVPR_2018_paper.pdf
cvpr-2018-6
['multi-oriented-scene-text-detection']
['computer-vision']
[ 0.14486589 -0.28210196 0.02535762 -0.16150261 -0.4880689 -0.501841 0.803701 0.12259955 -0.5311591 0.14245866 0.07277016 -0.1563022 0.29845828 -0.7831416 -0.460797 -0.630985 0.538759 0.6312291 0.8236843 0.02771872 0.5622947 0.3541291 -1.2464927 0.57859224 0.90514606 0.93079805 0.5445803 0.65998024 -0.7148428 0.68103886 -0.7330161 -0.13771398 0.08849257 -0.274068 -0.58410585 0.5121921 0.5767132 -0.38896167 -0.45057413 1.1327927 0.5166051 -0.06302268 1.005323 -0.9020962 -0.53309387 0.6550589 -1.1681606 0.28696662 0.02803698 -0.14405136 0.8745767 -1.192314 0.4666896 1.3638796 0.84621036 0.36679536 -0.8143871 -0.5020962 0.5072079 -0.19398873 -1.5423311 -0.1594349 0.69255716 -0.65647244 0.7854356 0.30703107 0.26064792 0.7823486 0.15117931 1.3264695 0.761532 -0.49983963 0.14307885 0.09206672 0.38920802 0.8678038 0.37146842 -0.6838358 -0.26929116 -0.02866305 0.7275446 0.26629645 -0.19692984 -0.17135072 -1.5092561 0.8306916 0.35084847 0.49474 -0.06798093 -0.08352613 0.61944646 -0.3656563 0.6240543 -0.08367231 -0.3068937 0.48123798 -1.3135223 0.28752795 0.3366271 1.1441504 0.47385374 0.24481004 -0.32499778 1.1084027 0.5615463 0.82197607 0.7263152 0.09844959 0.97231776 0.8510035 -0.14022146 -1.0534719 -0.5168607 -0.20352541 -1.0047485 -0.04776289 0.34258255 -0.13287878 -1.6239341 0.73748225 0.2723198 -0.25548327 -0.19319138 0.8544197 1.2339227 0.9355484 0.04169522 -0.01012574 1.5562646 -1.2552521 -0.79618686 -0.4501642 0.6353886 -1.0587603 1.1369271 0.5754076 -0.7764197 -0.7091892 -0.8752791 -0.2838402 -0.68435603 0.8844562 0.38596582 0.4726721 -0.86921495 0.16450202 -0.71373934 -0.5707501 0.5944577 0.17397794 0.17395985 0.2118328 -0.6611948 0.29426566 0.6158557 0.14682911 -0.7645861 -0.21585551 -0.74771225 0.014556 0.44475105 -0.37874436 1.028088 -0.739452 -1.1217242 1.0090845 -0.17511559 -0.2418006 0.85378146 -0.36629677 -0.29848078 0.14320858 0.40902963 0.79604286 1.0830231 -1.1441249 -0.9872832 -0.53958225 -0.6337016 0.4337388 -0.23301406 0.136567 -0.9081837 -0.8838725 0.65984666 -0.6086574 -0.08705272 0.04761807 -0.9328581 -0.49960738 1.47721 -0.63586915 1.0779115 -1.907145 -0.20581728 -0.08727039 0.23746578 0.34617355 0.24947104 0.25684816 0.15492082 0.28891027 -0.15716842 -0.43432245 0.06501225 -0.307209 -0.66866994 0.6723586 -0.11969228 0.7824473 -0.33334717 -1.1268336 0.76174235 0.29162398 -0.15991552 -0.12932219 -0.5331241 -0.09426446 -0.754204 0.9050606 0.91107273 -0.19861008 -0.20807886 -0.19341049 -0.18291731 -0.0396723 -1.3493356 1.4335172 0.08614565 0.9747554 0.06715977 -0.971006 1.0486002 -0.06026387 0.21928343 -0.43549585 0.4673787 -0.03351649 -0.40127704 -0.6164479 0.80442756 0.18791181 -0.09315592 0.1885469 -0.22296458 -0.28264162 0.16841048 0.17905381 0.58528817 0.14908536 0.2781436 -0.3469718 0.63651055 0.27448764 0.2125045 0.9635918 -0.19020799 0.97013444 0.28619158 -0.4887853 -1.0282156 -0.9962182 -0.46490178 1.1577599 0.26485965 -0.42358112 -0.86518264 -0.9392692 -0.09873545 0.47074258 -0.57940334 0.5664456 -0.6840567 -1.0562642 0.6717239 0.6127798 0.98893046 -1.1111337 -0.2903697 -0.12138056 -0.21172239 -1.2934527 -0.50065136 0.15495177 -1.0093133 -0.7416634 -1.1280037 -1.1949797 0.764019 0.577918 0.8973945 0.03487328 -0.48726568 -0.10391183 -0.48977992 -0.53719103 -0.06544729 0.09838971 -0.34178033 -0.02304021 0.43529403 0.20984033 -0.47626817 0.3225757 -0.88767165 0.24815221 0.6489871 0.55594385 0.78883976 0.31674922 0.1588392 -0.88416684 0.27640778 -0.12350103 -0.69883 0.05941778 -0.30781505 -0.3793434 0.7516049 -0.33742848 -1.1158485 0.40288883 -0.01932209 -0.31784302 -0.5868298 0.03252362 -0.23001112 0.25828037 0.5227192 0.6348935 -0.73938143 -0.5682919 0.21335924 1.193605 0.52285635 -0.224944 0.7407433 0.93489563 -0.31338236 -1.3986539 -0.8114405 -0.8180864 -0.80397576 -0.0389987 1.287464 -0.93346107 -0.34957406 0.81864256 -1.0258056 -0.31281012 0.21973205 0.28938437 -0.19423383 0.6948412 -0.67246145 -0.85797447 -0.6424505 -0.914468 1.7995229 0.1831124 0.20492037 -0.8760257 -0.3227641 0.3401318 -0.09371648 0.03583066 0.72870463 -0.75595033 -0.50762725 -0.35248438 -0.55014724 0.03627254 -0.07336658 0.1962163 -1.1253283 -0.0857014 -0.10743046 0.02207633 1.1567264 0.66359943 1.4561994 0.04936392 -0.72587967 0.5842337 1.5593898 0.35802492 0.5417537 0.51013905 1.0858123 0.42199883 0.8001223 0.2894243 0.09673804 0.46102038 0.29793113 -0.4317788 -0.05713157 -0.17057519 0.06234114 0.6194009 0.4007678 -0.72824925 -1.1405928 0.6587361 -1.8814933 -0.7164513 -0.8121151 1.6472964 0.29473442 0.44604558 0.185054 0.27248302 1.0811931 0.36961952 -0.45706773 -0.08113306 -0.23687087 -0.17544104 0.64999115 0.18508866 -1.687936 1.4520738 5.763081 1.1384275 -1.0445895 -0.12510577 0.7328628 0.11321668 0.38356012 -0.33136764 -1.2298493 0.5460821 0.1217206 0.31322554 -0.138286 1.0248715 0.3224657 -0.38375017 -0.7345248 1.1540657 0.4058358 -1.1975214 0.34052268 -0.27556193 0.753695 0.10077491 -0.00973921 0.28095537 0.10019594 -0.9480648 0.86391026 0.26015362 0.5771092 -0.5956106 0.7857631 0.4732062 -1.4177572 0.07764777 -0.79922473 0.32712558 -0.15860914 0.65561014 -0.96866137 0.41095206 0.9589821 0.72654575 -0.9144953 1.0474691 -0.14540154 0.71285963 -0.32193628 -0.45039397 0.45138296 -0.20615521 0.35417506 1.6495012 0.10980637 -0.17787029 0.4209191 0.8834039 -0.01762685 0.43766534 -0.44976252 0.15059522 0.27514136 1.3492604 -1.6354495 -0.74790734 -0.21337906 1.1532847 -0.1466369 0.2586994 -1.0095458 -0.81468314 -0.3065121 0.01168646 0.671123 -0.06500242 -0.6771175 -1.2149915 0.13619326 -0.7461584 0.38168952 -1.1201949 -0.94955015 0.5649419 -0.19794357 -1.105008 0.33508158 -0.8833748 -0.72286284 0.58525515 -1.0351245 -1.4655914 -0.52597785 0.56704646 1.2790747 -0.11498223 0.20524338 0.10668665 -0.91082907 0.40768948 0.45767072 0.6380583 0.61820567 -1.3784583 0.7177791 0.9665931 0.21282856 0.1797539 0.57446617 -0.8586821 -0.97792745 -1.4091097 0.7142158 -0.49455038 0.46569958 -0.84567624 -0.89937407 0.74568945 0.15109012 -0.21913844 -0.0915103 -0.2503516 -0.02537625 0.04943485 -1.0152737 0.8200257 0.78941655 -0.171204 -0.83284557 0.7767207 0.52838176 -0.54257953 -0.25110418 0.22939503 0.04799582 -0.859281 0.822765 -0.07288747 0.24198347 -0.5047446 -0.03141882 -0.62621766 -0.14393646 -0.22402339 0.35453928 1.42794 0.14334004 -0.3335433 1.0995854 -0.2399383 -0.33167332 -0.40283722 -0.631483 -0.5314064 0.17860383 -0.39499515 0.28061897 0.9522944 -0.4064571 0.46940213 -0.15811999 0.26707923 0.5464046 0.19061658 0.8972987 -1.1906017 0.09705766 -0.5998369 -0.2414252 -1.4657098 -0.12266142 -0.70836097 0.37798813 -1.8237618 0.3598459 -0.31791708 0.22891474 0.3284872 -0.24188833 0.30370203 0.05155366 0.38059753 -0.84687066 0.360956 1.1188635 -0.2771851 -0.24138303 -0.10879195 -0.29008684 1.3243386 0.79871774 -0.38088375 -0.23863317 -0.46303985 -0.02368825 -0.32150382 0.22571287 -1.1849937 0.32921904 -0.0494147 0.89537454 -1.7531369 0.14767091 -0.6326585 -0.6326696 0.3281431 -0.22375067 -0.14301702 0.24677745 0.7065782 0.0849762 -0.5044727 0.94023126 -0.12377137 -0.81893164 0.07000872 -0.75177747 0.10021612 0.9896113 -0.3473559 -0.42464817 0.03706627 -0.44437015 0.17580368 0.30121532 0.4079802 0.70357674 -0.8794956 -0.63762426 0.19554614 0.09860516 0.47156414 0.21272168 0.68079567 -0.95374125 0.85796684 0.31655347 -1.1403506 -1.4651363 0.6705642 0.41041863 -0.19996396 -1.002427 0.75738823 0.70903766 -0.37912515 0.51640373 -0.59678835 -0.3481936 0.06011205 0.4049645 0.26178688 0.09068963 -0.70556486 -0.41889223 0.99972945 -0.38331673 -0.01052138 0.8683287 -0.17020796 0.10076404 0.42940512 0.8411037 0.11629611 -0.8984895 -0.12278982 0.00738215 -0.37192088 0.22400004 -0.7731964 -0.9300272 1.1213137 0.6963958 0.44891715 1.0029961 0.13516872 0.7684423 0.50921446 -0.18326066 -1.3314652 0.27625963 0.63993984 0.6264249 -1.303784 0.29816663 -0.5724117 -0.64743066 1.2990289 0.77117956 -0.26055354 0.36142033 0.33173397 0.04736606 -0.33754066 -0.18302917 -0.35840032 0.30358103 0.43113455 0.40144128 -0.07337821 -0.21116352 0.35996357 -0.08959828 -0.55695343 0.4679076 0.92428154 -1.0269724 -0.3092894 -0.8690266 0.6113749 -0.7303062 -0.319003 -0.7760035 1.2004904 0.16922969 0.8826801 0.4681284 -0.13322106 0.17705952 -0.06363416 -0.00630089 -0.63672346 -0.3471817 0.78700477 -0.15380229 0.01576298 -0.15341812 -0.6348037 -1.4609264 -0.05476944 -0.7900269 -0.24992536 0.72509795 0.8758946 -0.04228396 0.568023 0.47496244 -0.8407463 -0.2468466 -1.1334791 -0.60072064 0.09315584 0.2563749 -0.35099056 -0.54955226 0.42440462]
[12.070450782775879, 2.3170721530914307]
a880d7a4-acdd-4192-83c2-cde49e21dd61
response-conditioned-turn-taking-prediction
2305.02036
null
https://arxiv.org/abs/2305.02036v1
https://arxiv.org/pdf/2305.02036v1.pdf
Response-conditioned Turn-taking Prediction
Previous approaches to turn-taking and response generation in conversational systems have treated it as a two-stage process: First, the end of a turn is detected (based on conversation history), then the system generates an appropriate response. Humans, however, do not take the turn just because it is likely, but also consider whether what they want to say fits the position. In this paper, we present a model (an extension of TurnGPT) that conditions the end-of-turn prediction on both conversation history and what the next speaker wants to say. We found that our model consistently outperforms the baseline model in a variety of metrics. The improvement is most prominent in two scenarios where turn predictions can be ambiguous solely from the conversation history: 1) when the current utterance contains a statement followed by a question; 2) when the end of the current utterance semantically matches the response. Treating the turn-prediction and response-ranking as a one-stage process, our findings suggest that our model can be used as an incremental response ranker, which can be applied in various settings.
['Gabriel Skantze', 'Erik Ekstedt', "Bing'er Jiang"]
2023-05-03
null
null
null
null
['response-generation']
['natural-language-processing']
[ 6.06312871e-01 5.71805775e-01 -2.97932059e-01 -9.22838330e-01 -1.14383495e+00 -9.04620707e-01 9.35497701e-01 2.42929161e-01 -1.99841917e-01 6.06837869e-01 9.24084246e-01 -6.83517992e-01 9.25163552e-02 -6.23995841e-01 -1.29131898e-01 -1.03992999e-01 4.46332306e-01 7.57819235e-01 2.27810875e-01 -6.44817770e-01 5.74632943e-01 1.16311505e-01 -1.30998528e+00 9.09687221e-01 5.03955781e-01 6.10278845e-01 1.08630821e-01 9.71976638e-01 -2.97384977e-01 1.19935703e+00 -6.47933185e-01 -3.85036111e-01 6.61491463e-03 -1.03073812e+00 -1.68999100e+00 9.68522355e-02 1.51085407e-01 -5.12729704e-01 9.80443060e-02 6.00664854e-01 3.56089056e-01 4.04162019e-01 3.92213792e-01 -1.08590794e+00 -2.95942146e-02 9.04655278e-01 3.04973453e-01 6.18221015e-02 1.15159631e+00 1.04153141e-01 1.32274556e+00 -8.03448796e-01 7.60524511e-01 1.55752301e+00 4.42105144e-01 6.65357292e-01 -1.03449690e+00 -2.43474603e-01 4.36305225e-01 8.32980946e-02 -5.82941353e-01 -6.88118994e-01 5.75224102e-01 -5.15638649e-01 9.78033960e-01 7.06810296e-01 3.57004046e-01 8.19938481e-01 5.09634688e-02 7.67727077e-01 1.21530330e+00 -5.30333102e-01 1.93787828e-01 2.22939536e-01 4.31337476e-01 1.10545784e-01 -7.86981463e-01 -1.94573775e-01 -5.73086143e-01 -5.13843596e-01 8.81611183e-03 -3.49097401e-01 -1.59807995e-01 5.17735302e-01 -1.20724618e+00 8.55351269e-01 6.93760365e-02 2.37326950e-01 -6.25555158e-01 -3.86305958e-01 3.65962684e-01 7.63070881e-01 4.71003920e-01 6.37098134e-01 -2.61375636e-01 -6.63516104e-01 -7.81244159e-01 5.82321227e-01 1.49520147e+00 6.14617169e-01 6.79331243e-01 -4.86152887e-01 -4.90076214e-01 9.34981167e-01 1.51641220e-01 3.57674458e-03 2.74587065e-01 -1.21120477e+00 5.58658302e-01 5.14124930e-01 5.54337084e-01 -8.47770393e-01 -3.92739743e-01 1.85895488e-01 -4.51027043e-02 -2.37618387e-01 6.28326654e-01 -3.89136940e-01 -3.88261318e-01 1.71437013e+00 5.63584566e-01 -4.27160174e-01 2.84432292e-01 1.02581143e+00 8.01369071e-01 8.23238194e-01 -7.41601735e-02 -6.10507607e-01 1.33926797e+00 -9.08856273e-01 -6.41603470e-01 -4.11915183e-01 7.07868874e-01 -1.27673244e+00 1.06312740e+00 2.94258714e-01 -1.00552893e+00 -4.73344535e-01 -6.65573180e-01 -5.53671131e-03 3.39966863e-01 -1.81791857e-01 4.32577044e-01 2.43347064e-01 -1.16095114e+00 3.51619571e-01 -1.39712289e-01 -6.42460167e-01 -7.46455789e-01 5.17842360e-02 -5.52380458e-02 -3.32947560e-02 -1.40382338e+00 9.72355723e-01 -1.38573945e-01 1.15595080e-01 -4.48267937e-01 7.39259552e-03 -4.76382941e-01 -1.18072093e-01 4.49761748e-01 -5.69505274e-01 2.05725574e+00 -1.23492372e+00 -1.95741487e+00 8.72416615e-01 -7.45345235e-01 -3.15460205e-01 5.87001026e-01 -1.16284721e-01 -2.21247286e-01 9.27883163e-02 1.81049570e-01 5.02262175e-01 4.70320612e-01 -9.97613847e-01 -9.46728170e-01 -1.30276650e-01 7.55677998e-01 6.35543883e-01 4.59488064e-01 4.21102375e-01 -7.87589625e-02 3.09405308e-02 3.68328273e-01 -1.23809469e+00 -8.11095238e-02 -6.94540083e-01 -4.21433389e-01 -8.53905439e-01 4.52354699e-01 -5.96083403e-01 1.43321753e+00 -1.69871593e+00 -2.09472358e-01 1.52567059e-01 -2.60401852e-02 -1.99254882e-02 -1.34518608e-01 1.20058548e+00 -1.37091294e-01 2.78482765e-01 1.94228426e-01 -4.02226597e-02 -8.34063888e-02 -1.23702481e-01 -7.30596662e-01 -4.78679277e-02 -4.84042317e-02 6.14801049e-01 -9.70214307e-01 -3.22765380e-01 -1.18750878e-01 -1.24861188e-01 -5.09095132e-01 6.03454709e-01 -4.32688087e-01 4.96522099e-01 -5.16806066e-01 1.01962432e-01 1.49900734e-01 -2.81501144e-01 5.17193437e-01 3.52600664e-01 -3.61265600e-01 1.35736251e+00 -7.67841935e-01 9.00430739e-01 -4.80337828e-01 6.72429562e-01 1.18229449e-01 -5.60057521e-01 8.18233013e-01 5.64506888e-01 9.12365764e-02 -4.35323298e-01 -2.28943393e-01 5.81770428e-02 5.73788464e-01 -5.76872587e-01 6.53639257e-01 -3.45973283e-01 -3.84028196e-01 1.12593365e+00 -5.68892956e-01 -4.90455300e-01 1.88475549e-01 5.08573294e-01 1.09659648e+00 -1.20400108e-01 4.97411013e-01 9.13945213e-02 6.35880768e-01 1.29543439e-01 4.79492188e-01 1.04724228e+00 -1.92093402e-01 4.32525277e-01 9.61444318e-01 -4.28754717e-01 -7.38815308e-01 -5.18269062e-01 2.94892341e-01 1.61999619e+00 -8.50058571e-02 -4.22541469e-01 -7.61112154e-01 -6.59043491e-01 -3.25540543e-01 1.16550803e+00 -3.46583664e-01 1.67557254e-01 -8.09799135e-01 -9.95919295e-03 2.48824954e-01 1.66301340e-01 9.94884893e-02 -1.28821242e+00 -4.15422887e-01 4.64057028e-01 -9.87416923e-01 -8.93612504e-01 -7.54565120e-01 -1.82790324e-01 -4.51922834e-01 -9.14468288e-01 -2.17171833e-02 -5.62517047e-01 3.30424070e-01 5.35772920e-01 1.04778445e+00 5.05605519e-01 6.55774951e-01 4.79425877e-01 -7.10793853e-01 -1.67499155e-01 -9.52920496e-01 -4.81884368e-02 -9.48015377e-02 9.57948267e-02 3.39044392e-01 -2.23630130e-01 -5.38733900e-01 7.22354710e-01 -4.53171849e-01 3.36621344e-01 1.85696185e-01 7.56210327e-01 -2.96417139e-02 -6.16628826e-01 9.62678373e-01 -1.17156398e+00 1.18848538e+00 -6.18029952e-01 -4.99380305e-02 3.93890709e-01 -4.73545432e-01 -1.32098764e-01 5.56427538e-01 -2.15580359e-01 -1.16283810e+00 -1.80205867e-01 -4.67123419e-01 4.90034163e-01 -2.33751655e-01 6.55667663e-01 7.85441101e-02 4.77582216e-01 6.64284050e-01 2.02026054e-01 1.54185027e-01 -7.92661086e-02 1.80665210e-01 1.01928830e+00 1.28674388e-01 -6.83988810e-01 1.88079000e-01 -7.62530565e-02 -5.50148427e-01 -6.79102361e-01 -1.06174183e+00 -7.12574005e-01 -2.84458995e-01 -7.26697206e-01 4.82315451e-01 -7.38189399e-01 -1.02616775e+00 6.65550381e-02 -1.46416569e+00 -5.32232285e-01 2.19740570e-01 4.21478897e-01 -5.79618812e-01 3.88451874e-01 -8.06753993e-01 -1.18226969e+00 -2.49440506e-01 -1.13336980e+00 7.27506876e-01 1.03403866e-01 -1.14622653e+00 -6.79656148e-01 -1.45483330e-01 5.91669500e-01 3.67273510e-01 -2.90460885e-01 1.08317220e+00 -1.24410546e+00 -3.43513072e-01 -2.49250844e-01 1.96386188e-01 8.79072919e-02 1.18066333e-01 8.84983167e-02 -7.67009556e-01 -2.52689105e-02 2.14553952e-01 -5.09940505e-01 3.29580367e-01 -2.95405369e-02 5.55124521e-01 -7.87336111e-01 -1.01420991e-01 -3.39587837e-01 6.16970658e-01 4.74544555e-01 5.15772223e-01 -1.03492932e-02 4.82273251e-02 1.30811286e+00 9.05934215e-01 2.19776571e-01 8.85064304e-01 7.48913467e-01 1.30848572e-01 3.53057295e-01 1.33552149e-01 -4.47048098e-01 4.97123897e-01 9.69672024e-01 2.79551536e-01 -6.06745481e-01 -9.53117132e-01 4.89576519e-01 -2.01180744e+00 -1.26337564e+00 -3.53016764e-01 2.21954870e+00 8.53648722e-01 2.92648166e-01 3.27210575e-01 -4.25287671e-02 7.51465917e-01 2.20650434e-01 -3.07594121e-01 -9.47006762e-01 2.45785862e-01 -2.70080000e-01 -3.55984628e-01 1.19262791e+00 -5.05316138e-01 1.00449538e+00 7.12851477e+00 8.14330205e-02 -1.25781584e+00 1.08625991e-02 8.74487340e-01 1.91218540e-01 -5.54220319e-01 5.11204422e-01 -9.12221253e-01 3.39254439e-01 1.09571743e+00 -2.53359824e-01 3.42186511e-01 5.95570862e-01 5.19002855e-01 -4.67314273e-01 -1.62089479e+00 3.37439895e-01 6.64383024e-02 -1.02019346e+00 -7.59060532e-02 -9.56051275e-02 3.70921940e-01 -4.57640737e-01 -2.40288690e-01 5.71634591e-01 5.16590297e-01 -7.54637003e-01 6.96145833e-01 4.45202559e-01 3.66369873e-01 -4.69554186e-01 6.96537495e-01 8.92559946e-01 -7.40093589e-01 -1.54141262e-01 1.57170370e-01 -6.40291929e-01 3.53387743e-01 2.62669772e-01 -1.69521964e+00 1.78674430e-01 9.30468738e-02 1.34670451e-01 6.70464337e-02 4.83394802e-01 -6.11477554e-01 9.80880082e-01 -1.10827968e-01 -4.20163989e-01 2.41914630e-01 -1.72967855e-02 7.09875226e-01 1.15061700e+00 8.07284787e-02 6.69096231e-01 7.25469589e-01 3.48950207e-01 7.34315291e-02 2.94067919e-01 -4.56776887e-01 8.04033652e-02 9.08246338e-01 1.05337512e+00 -5.19071698e-01 -5.82626879e-01 -1.89898655e-01 6.99094713e-01 2.61542380e-01 3.85924697e-01 -3.68033469e-01 -1.25271454e-01 5.68625510e-01 2.66135275e-01 -1.65795818e-01 2.13544413e-01 -8.70592818e-02 -7.57519186e-01 1.54003933e-01 -1.30403280e+00 3.13816994e-01 -8.27085435e-01 -9.67970490e-01 7.11126387e-01 -7.83038735e-02 -1.05499279e+00 -1.22332489e+00 -6.80347905e-02 -9.01663721e-01 9.49486434e-01 -1.03169036e+00 -6.94538414e-01 3.78751606e-02 -4.19148989e-03 9.38516021e-01 4.68537897e-01 9.40973818e-01 -1.40029356e-01 -1.89573526e-01 3.30683380e-01 -6.33872390e-01 8.23170021e-02 8.93990099e-01 -1.00668287e+00 4.37915474e-01 7.03881681e-01 -2.35957384e-01 9.13893282e-01 1.09386528e+00 -5.58535159e-01 -1.01743960e+00 -4.61995155e-01 1.76866341e+00 -5.67072153e-01 4.37691808e-01 -2.18800485e-01 -9.51783776e-01 8.03046167e-01 4.17999476e-01 -7.51422763e-01 9.50239003e-01 4.95985866e-01 -1.47386938e-01 1.02378190e-01 -9.50934052e-01 6.33288324e-01 7.18817532e-01 -7.95532644e-01 -9.98473704e-01 5.44186532e-01 8.03135395e-01 -5.16934097e-01 -3.65282178e-01 8.17459002e-02 6.89646065e-01 -1.12682998e+00 3.59283209e-01 -8.23078036e-01 5.45320988e-01 -6.75411895e-02 -1.89556047e-01 -1.29489291e+00 -3.17049414e-01 -1.04992568e+00 2.76074260e-01 1.08777905e+00 6.89202428e-01 -5.81594646e-01 3.22351128e-01 1.20701480e+00 -2.56752938e-01 -6.90326095e-01 -7.41514027e-01 -1.00643784e-01 -1.75369710e-01 -3.91596973e-01 7.43230760e-01 8.15321207e-01 6.83385313e-01 9.12903607e-01 -3.95289183e-01 -2.60422766e-01 -1.66023776e-01 4.33277428e-01 1.10284281e+00 -1.16848624e+00 -1.98023275e-01 -4.73971337e-01 2.68161833e-01 -1.73490524e+00 1.74732015e-01 -6.07395649e-01 7.09392130e-01 -1.63968325e+00 2.64495295e-02 -2.93167859e-01 1.72859296e-01 4.21870947e-01 -4.31013346e-01 -5.52897990e-01 3.05409938e-01 4.09432113e-01 -4.99937862e-01 -3.27243982e-03 9.72912490e-01 2.07169592e-01 -4.49022830e-01 7.09515810e-01 -9.03367817e-01 7.05969572e-01 6.56998575e-01 -3.70857775e-01 -4.24526870e-01 -5.46701737e-02 4.35181558e-01 9.97968972e-01 -3.84260453e-02 -4.44957614e-01 5.28703570e-01 -6.15080595e-01 -2.13773459e-01 -6.51752949e-01 3.93496484e-01 -2.75529772e-01 2.03889254e-02 2.02760011e-01 -1.13533199e+00 3.02223235e-01 -2.68796593e-01 3.52981210e-01 -2.62828231e-01 -2.80247360e-01 4.65628058e-01 -1.87543571e-01 -5.02486587e-01 -2.18589798e-01 -1.04813504e+00 -5.89231886e-02 7.29629993e-01 -4.23124284e-01 -3.21720272e-01 -1.24624038e+00 -6.83070540e-01 3.29005063e-01 2.71403611e-01 6.34350479e-01 5.28941333e-01 -1.01337731e+00 -8.92105162e-01 -1.91909254e-01 1.44695416e-01 -3.87302339e-01 3.60455923e-02 8.70613575e-01 -3.25996578e-02 6.13658607e-01 3.23663473e-01 -5.36159098e-01 -1.42581952e+00 7.54179880e-02 1.99425280e-01 -4.23010558e-01 -3.15850884e-01 9.21031177e-01 1.61690056e-01 -6.79377615e-01 1.44300878e-01 -3.12684983e-01 -4.15659457e-01 3.31483990e-01 8.16221893e-01 -1.93958562e-02 1.37080494e-02 -7.71080613e-01 -3.05982858e-01 -1.73727363e-01 -2.58497357e-01 -6.74477160e-01 9.06393111e-01 -4.22921300e-01 -3.56204927e-01 9.31611061e-01 9.13376212e-01 2.08873555e-01 -7.80775905e-01 -4.24255311e-01 1.39062509e-01 -5.34348428e-01 -5.50394356e-01 -9.41511333e-01 -7.71005377e-02 5.04430771e-01 -2.23628283e-01 7.24250734e-01 6.22172713e-01 1.16734272e-02 7.52597034e-01 5.37864625e-01 5.19377470e-01 -1.17710531e+00 2.57482320e-01 9.97417450e-01 1.10597336e+00 -1.15850556e+00 -2.09542781e-01 -3.28126490e-01 -1.03338778e+00 1.19118464e+00 7.71975696e-01 4.00540531e-01 2.96100348e-01 -2.01935098e-01 4.75326359e-01 -7.28569478e-02 -1.57625353e+00 5.46751171e-02 1.25480250e-01 6.19915351e-02 1.04620910e+00 3.33121240e-01 -7.33081281e-01 3.24120045e-01 -6.54121637e-01 -3.65716308e-01 7.73710847e-01 8.60974789e-01 -8.45097721e-01 -1.14094758e+00 -2.87176758e-01 5.48678458e-01 -4.26160812e-01 -1.09865338e-01 -1.18208528e+00 3.00837457e-01 -4.28111792e-01 1.85146749e+00 -4.20504101e-02 -6.27464533e-01 4.53933835e-01 5.17661154e-01 -5.80383725e-02 -1.03041494e+00 -1.04686224e+00 -8.26389119e-02 9.01108027e-01 -4.84207511e-01 -3.03395987e-01 -8.44887972e-01 -1.22794044e+00 -4.27201390e-01 -4.38348562e-01 6.48916304e-01 4.59458441e-01 1.31475806e+00 3.09094399e-01 -7.57020572e-03 1.15755212e+00 -4.56594378e-01 -8.33373368e-01 -1.16546071e+00 1.30016312e-01 4.83847737e-01 4.61432159e-01 -1.70190796e-01 -3.91931355e-01 -1.16629094e-01]
[12.755139350891113, 7.976809024810791]
b27b9b10-5127-41f0-b538-5e5b78261fdb
are-multimodal-models-robust-to-image-and
2212.08044
null
https://arxiv.org/abs/2212.08044v1
https://arxiv.org/pdf/2212.08044v1.pdf
Are Multimodal Models Robust to Image and Text Perturbations?
Multimodal image-text models have shown remarkable performance in the past few years. However, evaluating their robustness against distribution shifts is crucial before adopting them in real-world applications. In this paper, we investigate the robustness of 9 popular open-sourced image-text models under common perturbations on five tasks (image-text retrieval, visual reasoning, visual entailment, image captioning, and text-to-image generation). In particular, we propose several new multimodal robustness benchmarks by applying 17 image perturbation and 16 text perturbation techniques on top of existing datasets. We observe that multimodal models are not robust to image and text perturbations, especially to image perturbations. Among the tested perturbation methods, character-level perturbations constitute the most severe distribution shift for text, and zoom blur is the most severe shift for image data. We also introduce two new robustness metrics (MMI and MOR) for proper evaluations of multimodal models. We hope our extensive study sheds light on new directions for the development of robust multimodal models.
['Mu Li', 'Bo Li', 'Ding Zhao', 'Zhiqiang Tang', 'Florian Wenzel', 'Xingjian Shi', 'Yi Zhu', 'JieLin Qiu']
2022-12-15
null
null
null
null
['visual-reasoning', 'visual-reasoning', 'visual-entailment']
['computer-vision', 'reasoning', 'reasoning']
[ 4.91979599e-01 -3.83172423e-01 1.55765265e-01 -9.89480838e-02 -9.30983603e-01 -8.84210765e-01 1.03450060e+00 2.51333326e-01 -4.44338143e-01 4.34738308e-01 4.45257604e-01 -1.18918672e-01 4.87050600e-02 -6.77681640e-02 -9.16284919e-01 -7.29005575e-01 2.27144241e-01 2.28766724e-01 1.68619871e-01 -3.13963979e-01 6.26537681e-01 5.10969818e-01 -1.63821805e+00 5.80636322e-01 7.44715989e-01 8.06854486e-01 -8.91982839e-02 9.43330646e-01 1.70857519e-01 7.49802470e-01 -7.95855939e-01 -8.93902361e-01 8.53081197e-02 -1.08852535e-01 -5.34649551e-01 1.70142531e-01 8.62866282e-01 -2.04719543e-01 -5.49566746e-01 1.17906618e+00 1.08524609e+00 2.87260652e-01 1.04312778e+00 -1.53048348e+00 -9.32168841e-01 5.35290599e-01 -7.73526371e-01 9.98484120e-02 7.84207046e-01 6.70690000e-01 6.17839336e-01 -1.02334869e+00 6.76724017e-01 1.77603841e+00 3.96953821e-01 5.40044606e-01 -1.31714058e+00 -2.95511335e-01 -5.64001389e-02 3.33520323e-01 -1.26147211e+00 -6.10147178e-01 4.90501314e-01 -4.48039860e-01 8.01634669e-01 5.99030674e-01 -4.11709666e-01 1.66425431e+00 2.44324997e-01 8.13285768e-01 1.01370788e+00 -3.77953321e-01 -1.67158082e-01 2.27265567e-01 -1.27506420e-01 3.37946117e-01 2.43019477e-01 -1.52981803e-01 -5.20160854e-01 -2.60408670e-01 1.37087151e-01 -3.65329742e-01 -4.48837698e-01 -1.63323894e-01 -1.50004303e+00 4.86722082e-01 6.70245960e-02 1.12604909e-01 4.64696139e-02 1.95460156e-01 6.31879628e-01 3.08613598e-01 2.49106869e-01 4.23224062e-01 2.70514470e-02 -1.05522215e-01 -7.05570221e-01 5.17276585e-01 4.15539920e-01 9.84533429e-01 5.42880118e-01 -1.42061934e-01 -8.65453184e-01 1.11134064e+00 8.47276598e-02 1.10870433e+00 5.30669034e-01 -8.55588734e-01 8.34476948e-01 4.62716520e-01 2.59248465e-01 -1.48664200e+00 -3.12363863e-01 2.54589528e-01 -1.06290400e+00 -2.27605134e-01 3.93307596e-01 8.34148824e-02 -9.06075001e-01 1.49925387e+00 -7.57478103e-02 -3.34771931e-01 7.92709887e-02 8.17995131e-01 1.02894807e+00 5.63200474e-01 -2.38463525e-02 -2.31055468e-02 1.28093290e+00 -7.88728774e-01 -7.82127023e-01 -9.06727165e-02 2.91084439e-01 -1.32625175e+00 1.30112803e+00 1.05162352e-01 -1.28904819e+00 -4.07178909e-01 -7.08446443e-01 -1.26668349e-01 -5.82276225e-01 -6.91179708e-02 -2.29600534e-01 4.33774978e-01 -9.49040532e-01 2.89946109e-01 -2.26004288e-01 -7.65447557e-01 -7.47417100e-03 7.95671791e-02 -6.22537196e-01 -3.08909684e-01 -1.12219310e+00 1.13357317e+00 1.71083003e-01 -8.44216347e-03 -8.94227922e-01 -4.72867429e-01 -9.08925951e-01 -2.81642795e-01 2.36373335e-01 -6.69404626e-01 1.21838248e+00 -6.39005482e-01 -1.30227757e+00 1.09663236e+00 -1.30086303e-01 -3.58869731e-01 9.42106068e-01 -6.61525205e-02 -5.12075186e-01 3.91578525e-01 -1.13384992e-01 9.71446812e-01 1.28924203e+00 -1.76522017e+00 -1.63683623e-01 -2.44310081e-01 -2.07706586e-01 4.05179113e-01 -5.09489477e-01 3.36756915e-01 -8.95243645e-01 -8.22127223e-01 -2.22750515e-01 -1.03827822e+00 1.30779697e-02 -2.57979423e-01 -9.59978759e-01 5.41076623e-02 7.39938200e-01 -4.81709540e-01 1.28034556e+00 -2.17438674e+00 4.89154577e-01 1.21122845e-01 -1.07359260e-01 1.04424253e-01 -6.31890595e-01 8.92329574e-01 -2.12028295e-01 4.38390553e-01 -1.80525273e-01 -6.32113338e-01 4.12661195e-01 6.17515817e-02 -6.49473727e-01 4.54467863e-01 1.22401968e-01 1.07519948e+00 -4.06624675e-01 -7.07714438e-01 3.78745466e-01 3.31691206e-01 -2.23796129e-01 9.97319669e-02 -2.62079954e-01 3.41553926e-01 2.24083774e-02 7.73951232e-01 8.57437432e-01 2.08432432e-02 -4.46523547e-01 -4.79999244e-01 2.22932398e-01 -6.06721222e-01 -8.36864412e-01 1.27854145e+00 -1.00698687e-01 8.58301580e-01 -2.50011623e-01 -3.36873680e-01 4.47496533e-01 1.46856099e-01 1.07900560e-01 -9.70252395e-01 8.58566165e-02 -1.61692783e-01 -3.55393052e-01 -9.14665222e-01 9.36708331e-01 2.72505283e-01 -2.83112198e-01 4.52507108e-01 -2.41047442e-01 -3.54871511e-01 4.67698157e-01 5.76374888e-01 8.57432485e-01 -3.82317334e-01 -1.62771240e-01 6.26193881e-02 6.39526367e-01 -2.38426581e-01 -2.70838171e-01 1.11354780e+00 -3.20826113e-01 1.15818536e+00 7.91299105e-01 -2.35719904e-01 -1.19720852e+00 -1.09526682e+00 -5.31328991e-02 1.09442174e+00 3.07605714e-01 -3.14065307e-01 -8.82254183e-01 -5.31997740e-01 1.04860738e-01 6.91796243e-01 -6.07528448e-01 -2.52580196e-01 -2.12537467e-01 -9.95693862e-01 1.24802125e+00 1.24601550e-01 4.23072904e-01 -1.06247115e+00 -1.72492757e-01 -4.62784827e-01 -7.68929303e-01 -1.51331496e+00 -8.10737610e-01 -3.49273443e-01 -4.42025512e-01 -9.67199564e-01 -1.14481139e+00 -4.76497501e-01 6.99150026e-01 3.79736423e-01 1.09081137e+00 1.24997552e-02 -3.00295353e-01 8.46191227e-01 -4.91464585e-01 -1.40571669e-01 -7.47710288e-01 -2.24694297e-01 2.00766429e-01 7.50653744e-02 -1.00674085e-01 -6.43580481e-02 -6.10666394e-01 7.63095796e-01 -1.61020422e+00 -1.74768940e-01 4.35032338e-01 6.81934536e-01 2.88983375e-01 7.11633191e-02 -5.58438450e-02 -3.00425917e-01 1.05176115e+00 -2.72793710e-01 -2.40956351e-01 3.92298758e-01 -2.44988605e-01 1.33810952e-01 4.50978547e-01 -6.04491889e-01 -9.85613286e-01 -2.18494073e-01 1.88860580e-01 -5.93880415e-01 -4.57219929e-01 4.61445034e-01 -1.49457574e-01 -2.33507976e-01 1.04768014e+00 2.90398657e-01 -3.75424176e-02 -2.72684932e-01 6.06690466e-01 5.53617656e-01 7.55068302e-01 -6.63271904e-01 1.10261297e+00 5.18155217e-01 -7.42420480e-02 -1.13824797e+00 -3.13370317e-01 -2.51432121e-01 -2.29485303e-01 -1.87064633e-01 7.15564191e-01 -7.46615767e-01 -6.30561650e-01 8.47850263e-01 -1.29802310e+00 -2.45201558e-01 3.01185876e-01 -1.02306306e-01 -5.64455330e-01 1.18211794e+00 -5.48203111e-01 -6.20281816e-01 -2.62956440e-01 -1.49250138e+00 1.44123888e+00 -7.02438597e-03 -1.94285408e-01 -7.94412971e-01 8.51953700e-02 5.70111454e-01 3.30693066e-01 2.89383352e-01 1.08930993e+00 -3.91244709e-01 -3.66929829e-01 -2.81615019e-01 -4.95516986e-01 3.12216580e-01 -2.08666340e-01 5.42703807e-01 -9.45872307e-01 -4.10973340e-01 -3.82701725e-01 -5.07232070e-01 1.03366852e+00 3.35532904e-01 1.10804749e+00 -2.84099936e-01 -6.34208173e-02 5.12762249e-01 1.15020800e+00 -2.77278066e-01 1.06935811e+00 4.97507632e-01 8.08964431e-01 6.90464616e-01 5.98301589e-01 3.95633012e-01 2.72692412e-01 7.67993093e-01 6.13375247e-01 -4.96339388e-02 -1.51238203e-01 -8.76547098e-02 7.39325583e-01 3.42694908e-01 1.19455777e-01 -9.69445407e-01 -1.05081546e+00 3.23492736e-01 -2.01282644e+00 -1.01559055e+00 -1.71173543e-01 2.18682432e+00 6.70189381e-01 -1.42342523e-01 9.60729644e-03 5.31011112e-02 9.05663490e-01 1.94630384e-01 -4.08319354e-01 -3.17395389e-01 -7.40733743e-01 -5.21731436e-01 4.30296212e-01 3.58564258e-01 -1.29301322e+00 1.03926218e+00 6.74745560e+00 9.29077327e-01 -8.29000413e-01 -3.61912936e-01 6.53057575e-01 -1.42988786e-01 -3.24722052e-01 -3.70032370e-01 -5.93156278e-01 4.57389981e-01 6.91721678e-01 -1.40724376e-01 3.94539595e-01 3.66743952e-01 2.45311439e-01 -3.11291605e-01 -1.18134201e+00 1.33339155e+00 6.00644350e-01 -8.31041396e-01 8.10931802e-01 -1.37803972e-01 9.15192544e-01 -1.20014451e-01 7.06376672e-01 -1.12734027e-01 1.40669625e-02 -1.21460867e+00 6.67090893e-01 7.12843657e-01 9.12594497e-01 -7.33931422e-01 7.77608991e-01 -5.09864697e-03 -6.90370023e-01 5.50920852e-02 -3.79003227e-01 5.07675231e-01 3.11332885e-02 3.82655203e-01 -6.37071490e-01 5.24776042e-01 8.27945650e-01 4.52071846e-01 -1.27898598e+00 9.39439595e-01 1.06653720e-01 1.36090189e-01 -1.19275391e-01 1.10233411e-01 8.04504156e-02 3.03068042e-01 1.02925682e+00 1.57303333e+00 2.95400232e-01 -5.32567322e-01 -1.93270192e-01 5.15830398e-01 -1.56894848e-01 1.63034141e-01 -9.70784664e-01 -8.77765194e-02 1.69453636e-01 1.15683794e+00 -5.00396669e-01 -1.57371894e-01 -1.76213920e-01 1.43270397e+00 -8.67488310e-02 8.47111583e-01 -1.04505253e+00 -2.48807847e-01 5.59841037e-01 -2.47569099e-01 9.83689278e-02 -7.47657344e-02 -1.62951544e-01 -1.28360021e+00 5.36689535e-02 -1.51291013e+00 4.63575542e-01 -1.24653840e+00 -1.57109904e+00 5.52367330e-01 2.84149051e-01 -1.20071447e+00 -1.44533277e-01 -6.89715743e-01 -4.14373517e-01 6.46765292e-01 -1.28507781e+00 -1.09731722e+00 -4.17806536e-01 9.56046879e-01 4.73043710e-01 -2.74060935e-01 5.90027392e-01 1.53246954e-01 -8.51401687e-01 1.07506096e+00 2.30589300e-01 -8.47650617e-02 1.46718907e+00 -1.08436191e+00 4.94263589e-01 9.39289749e-01 -1.02706790e-01 4.71269697e-01 1.14035082e+00 -5.02801538e-01 -1.65154278e+00 -1.05418372e+00 4.98985976e-01 -8.67021918e-01 7.07231283e-01 -1.98678553e-01 -7.58014321e-01 4.99264657e-01 6.93638802e-01 -3.30615371e-01 4.36483949e-01 -5.01396000e-01 -5.58894455e-01 2.76077166e-02 -1.07907784e+00 1.30310476e+00 6.08137190e-01 -5.86286783e-01 -3.07939202e-01 3.94792438e-01 7.82010257e-01 -4.80787367e-01 -6.92592025e-01 5.18553019e-01 3.93013895e-01 -1.10177124e+00 1.09772539e+00 -4.79582965e-01 6.39695883e-01 -4.11529928e-01 -4.50047076e-01 -1.32599950e+00 -7.68349320e-02 -8.69348466e-01 3.76648493e-02 1.26986158e+00 4.80681688e-01 -3.47745508e-01 2.46021941e-01 6.48404896e-01 9.26339477e-02 -5.74314483e-02 -7.54879057e-01 -6.40054286e-01 5.29058389e-02 -4.91256714e-01 3.19521546e-01 7.37986982e-01 4.78707105e-02 1.80791482e-01 -6.49367690e-01 2.51284331e-01 5.78476727e-01 -3.83722484e-01 1.16439688e+00 -4.61845160e-01 8.69191960e-02 -8.13656986e-01 -5.68677425e-01 -6.60612583e-01 1.52242675e-01 -5.78166366e-01 4.07597832e-02 -1.23478079e+00 6.13813460e-01 4.34290975e-01 -6.93889409e-02 2.78681070e-01 -4.27641988e-01 6.20742559e-01 4.81464237e-01 1.91730544e-01 -8.41825724e-01 5.26314437e-01 1.32106769e+00 -4.94754523e-01 1.33373082e-01 -2.73587227e-01 -6.27391756e-01 5.96420348e-01 6.74200177e-01 -1.19711317e-01 -2.19562873e-01 -6.09169066e-01 5.27091503e-01 -2.38759652e-01 5.82220018e-01 -8.82077098e-01 3.41299623e-01 -1.29415676e-01 3.52462500e-01 -7.36282170e-01 3.22965056e-01 -7.35716105e-01 -1.09602384e-01 1.06136844e-01 -5.00803351e-01 4.89090651e-01 6.36916280e-01 5.86076260e-01 -2.76040047e-01 -1.83169529e-01 8.53745282e-01 8.47564489e-02 -4.95232761e-01 8.35010037e-02 -5.40293038e-01 2.38152742e-01 8.92156482e-01 -1.00588739e-01 -7.65384078e-01 -8.34916532e-01 -3.97774726e-01 3.36936474e-01 7.90403485e-01 7.83356786e-01 8.30542445e-01 -1.26221120e+00 -9.00604784e-01 -1.04145631e-01 6.19075239e-01 -5.27282059e-01 5.75023293e-01 9.12092745e-01 -5.63376725e-01 4.71164942e-01 -2.05512326e-02 -9.00067270e-01 -1.65027785e+00 8.06573153e-01 2.73548573e-01 -1.00487754e-01 -1.10876083e-01 5.23449719e-01 3.11255991e-01 -3.30453694e-01 7.75331378e-01 -1.90066919e-01 -6.88786581e-02 -6.64193416e-03 5.69207370e-01 5.40574074e-01 6.87814429e-02 -1.04779613e+00 -2.89869457e-01 8.37796748e-01 -2.48352140e-01 -3.78574282e-01 6.34168148e-01 -4.19092000e-01 -2.38981694e-01 3.24710250e-01 1.08659124e+00 -1.04390815e-01 -8.47634912e-01 -1.28626689e-01 -1.19162001e-01 -3.40643674e-01 -3.69599283e-01 -1.02775860e+00 -6.23010099e-01 9.07054067e-01 5.57987571e-01 3.15065861e-01 1.27763879e+00 -1.66019872e-01 5.44916332e-01 5.74181795e-01 -2.52059549e-01 -1.03302789e+00 5.43101072e-01 7.14260638e-01 1.41455555e+00 -1.47550201e+00 -1.66640431e-02 -2.97356285e-02 -1.12993896e+00 8.77027690e-01 5.93189001e-01 4.20738608e-01 1.22706927e-01 7.26756230e-02 2.49298602e-01 -2.34224610e-02 -7.91363597e-01 -1.01763315e-01 7.46316373e-01 6.65684581e-01 2.02710584e-01 -2.92063832e-01 1.64485946e-01 -6.53640255e-02 4.40690480e-02 -6.40973866e-01 5.66615760e-01 4.35580164e-01 -1.25070781e-01 -7.91202605e-01 -9.93827760e-01 -6.37592049e-03 -5.15140712e-01 -3.41126174e-01 -8.80612671e-01 6.67112768e-01 -2.92761832e-01 1.21481013e+00 -2.05902159e-01 -4.17782515e-01 5.92627704e-01 8.58080983e-02 5.88243842e-01 -1.78136364e-01 -5.07528126e-01 5.20260707e-02 1.03271892e-02 -6.65165007e-01 -2.51912802e-01 -5.92621446e-01 -6.87362373e-01 -5.60258150e-01 -1.01075783e-01 -3.58097494e-01 7.38314807e-01 6.63287163e-01 4.63011354e-01 1.67168528e-01 5.05057931e-01 -1.12142575e+00 -4.60589826e-01 -1.11537504e+00 -1.57975063e-01 9.30873692e-01 3.43498647e-01 -3.26022267e-01 -6.85777783e-01 3.03784460e-01]
[11.099872589111328, 1.2445402145385742]
3e7fae8f-fad1-4140-9783-3e1bfa5cce17
revisiting-automatic-evaluation-of-extractive
null
null
https://aclanthology.org/2022.findings-acl.122
https://aclanthology.org/2022.findings-acl.122.pdf
Revisiting Automatic Evaluation of Extractive Summarization Task: Can We Do Better than ROUGE?
It has been the norm for a long time to evaluate automated summarization tasks using the popular ROUGE metric. Although several studies in the past have highlighted the limitations of ROUGE, researchers have struggled to reach a consensus on a better alternative until today. One major limitation of the traditional ROUGE metric is the lack of semantic understanding (relies on direct overlap of n-grams). In this paper, we exclusively focus on the extractive summarization task and propose a semantic-aware nCG (normalized cumulative gain)-based evaluation metric (called Sem-nCG) for evaluating this task. One fundamental contribution of the paper is that it demonstrates how we can generate more reliable semantic-aware ground truths for evaluating extractive summarization tasks without any additional human intervention. To the best of our knowledge, this work is the first of its kind. We have conducted extensive experiments with this new metric using the widely used CNN/DailyMail dataset. Experimental results show that the new Sem-nCG metric is indeed semantic-aware, shows higher correlation with human judgement (more reliable) and yields a large number of disagreements with the original ROUGE metric (suggesting that ROUGE often leads to inaccurate conclusions also verified by humans).
['Shubhra Kanti Karmaker', 'Naman Bansal', 'Mousumi Akter']
null
null
null
null
findings-acl-2022-5
['extractive-summarization']
['natural-language-processing']
[ 1.19568229e-01 2.55436897e-01 -6.11319020e-03 -2.62080401e-01 -8.12253952e-01 -6.42278731e-01 7.96076477e-01 6.26111388e-01 -7.13734210e-01 9.84317601e-01 6.12477183e-01 -1.09172404e-01 -8.76279548e-02 -6.19119346e-01 -3.45458657e-01 -2.27300063e-01 2.62733489e-01 4.70306128e-01 3.34241658e-01 -4.36824203e-01 7.92491913e-01 2.38931164e-01 -1.68208694e+00 3.50448132e-01 1.14535761e+00 7.28400230e-01 1.86479300e-01 5.79830587e-01 -1.58188850e-01 5.95937669e-01 -1.11609185e+00 -5.15694916e-01 -1.02123767e-01 -7.14168370e-01 -1.14318621e+00 -1.81444392e-01 5.27490616e-01 -5.53658158e-02 2.42040396e-01 1.12899137e+00 6.27177775e-01 3.24015111e-01 7.77747571e-01 -9.27341461e-01 -6.83097363e-01 7.88861454e-01 -3.09482187e-01 2.48893425e-01 4.95392412e-01 -1.61411062e-01 1.35745561e+00 -7.10850239e-01 8.35104704e-01 1.32382345e+00 6.67169988e-01 3.34662169e-01 -9.21636939e-01 -2.67453164e-01 -9.14195776e-02 5.09021938e-01 -1.15798235e+00 -1.38370439e-01 5.61395288e-01 -2.37679571e-01 1.24106359e+00 3.75171721e-01 4.20636028e-01 9.15292144e-01 8.21382776e-02 8.13756347e-01 1.15930760e+00 -6.67289555e-01 2.88893819e-01 7.43279383e-02 3.16663444e-01 5.45812786e-01 5.30484080e-01 -4.19491082e-01 -3.26135248e-01 7.05641042e-03 2.73572505e-01 -4.13465023e-01 -4.71555531e-01 -1.64453149e-01 -1.44198847e+00 8.87689173e-01 2.40602732e-01 9.84333754e-01 -4.60937917e-01 -4.44191694e-02 7.19899893e-01 2.60215908e-01 5.07236302e-01 1.00352228e+00 -3.09146941e-01 -6.70922756e-01 -1.32012069e+00 4.49532688e-01 1.07538247e+00 5.40088952e-01 5.00696719e-01 -1.06502347e-01 -2.89540559e-01 9.48589623e-01 -9.99903958e-03 1.85682669e-01 7.97903180e-01 -9.96851206e-01 2.84856498e-01 6.61245942e-01 1.45645872e-01 -1.27639532e+00 -4.51152027e-01 -5.96716523e-01 -7.54004419e-01 -5.51945940e-02 3.87059331e-01 5.93387075e-02 -6.01051271e-01 1.50303853e+00 -2.49220565e-01 -2.36378744e-01 1.12489045e-01 9.83973622e-01 1.00963545e+00 4.94315028e-01 -3.37486230e-02 -1.62977740e-01 1.27487957e+00 -1.00713289e+00 -8.31555068e-01 -2.18548644e-02 8.04760218e-01 -9.46278274e-01 1.27129078e+00 6.13415837e-01 -8.56914341e-01 -4.10787255e-01 -1.36494517e+00 -7.52544552e-02 -5.97854912e-01 1.38452366e-01 4.67410922e-01 6.22249842e-01 -1.18533170e+00 9.52721298e-01 -4.57775563e-01 -1.01840043e+00 1.11607816e-02 4.87401113e-02 -4.32866156e-01 5.01736179e-02 -1.30484819e+00 1.25077665e+00 7.61247635e-01 -2.80988753e-01 -2.85217851e-01 -1.93291143e-01 -4.77245390e-01 1.11282162e-01 5.92429042e-01 -7.48985589e-01 1.42076921e+00 -9.49312449e-01 -1.34289384e+00 9.03663874e-01 -1.42455533e-01 -6.84690654e-01 4.91717160e-01 -2.53908932e-01 -3.66748303e-01 2.17879251e-01 1.76350519e-01 6.37880862e-01 2.78116018e-01 -1.25630081e+00 -4.84293818e-01 -2.88430929e-01 2.18155369e-01 2.31356129e-01 -3.34814757e-01 1.45289749e-01 -5.35652488e-02 -8.57156694e-01 -8.68664458e-02 -7.49130845e-01 1.95076749e-01 -7.44255126e-01 -4.04723704e-01 -5.77412963e-01 3.85154009e-01 -6.68242395e-01 1.52412939e+00 -1.69820750e+00 1.01036996e-01 -2.20772997e-01 2.94315070e-01 6.23707831e-01 -5.23555428e-02 9.10328329e-01 3.37026976e-02 4.05217201e-01 -5.11629283e-01 -3.04465234e-01 7.45588839e-02 8.81830379e-02 -2.78993636e-01 1.22800782e-01 -7.95116127e-02 9.79870558e-01 -1.28928304e+00 -6.25094712e-01 9.83898342e-02 3.42333555e-01 -2.31525108e-01 -1.19448312e-01 -1.24687038e-01 7.78047666e-02 -2.24066496e-01 3.03544283e-01 3.96758735e-01 -1.48125708e-01 1.21579608e-02 -3.16389024e-01 -2.02873379e-01 2.71523863e-01 -8.27291965e-01 1.76196837e+00 -3.34211409e-01 8.36289823e-01 -5.59607804e-01 -1.16642082e+00 1.01279438e+00 2.63611019e-01 2.74995625e-01 -5.87012887e-01 3.64589602e-01 6.02790952e-01 -2.88123228e-02 -3.61034125e-01 1.20852578e+00 -1.55000955e-01 -1.54663548e-01 4.73422170e-01 2.81878591e-01 -3.00047666e-01 5.94829679e-01 2.90031761e-01 1.05594552e+00 -1.07126450e-02 7.68347025e-01 -5.13126135e-01 7.13065863e-01 2.37351909e-01 4.07037079e-01 8.94555390e-01 -2.95824468e-01 9.97240245e-01 5.24068952e-01 -2.00985461e-01 -1.12113988e+00 -7.01075315e-01 2.27883801e-01 8.31889689e-01 2.64714975e-02 -7.33662546e-01 -1.08010018e+00 -7.90842831e-01 -2.59486556e-01 1.26000249e+00 -4.80177104e-01 9.91392359e-02 -5.22189021e-01 -6.21253192e-01 7.61257827e-01 3.93211097e-01 7.32820928e-01 -1.12550390e+00 -7.44794250e-01 2.40943983e-01 -6.39867961e-01 -1.05013192e+00 -2.04661489e-01 -2.23455638e-01 -1.00414860e+00 -1.13662541e+00 -8.47658396e-01 -3.79294097e-01 2.13991985e-01 3.58552337e-01 1.32982147e+00 1.83122680e-01 1.28528431e-01 3.33958119e-01 -9.37165916e-01 -5.99385440e-01 -5.65540195e-01 4.20929521e-01 -1.80313721e-01 -2.96657085e-01 6.93782985e-01 -3.06400269e-01 -5.44687748e-01 1.90168574e-01 -1.06768310e+00 -8.09832811e-02 6.60365105e-01 5.72970986e-01 2.97010928e-01 -6.79842159e-02 9.60172474e-01 -8.73865187e-01 1.20203578e+00 -1.75792739e-01 8.78590569e-02 3.01197916e-01 -8.45426142e-01 4.62039337e-02 6.90636814e-01 9.19789821e-02 -8.14765334e-01 -7.00019598e-01 -9.86049920e-02 9.08961445e-02 -2.86119252e-01 6.98975861e-01 1.62090927e-01 4.00103658e-01 8.35663319e-01 1.30522251e-01 -1.02969468e-01 -5.92630625e-01 4.65428084e-01 8.06296468e-01 6.32543802e-01 -3.28335732e-01 3.08001846e-01 4.98452842e-01 -1.10054269e-01 -1.05119658e+00 -9.95203555e-01 -7.06711411e-01 -6.30283535e-01 -3.73100072e-01 8.09431672e-01 -5.35808742e-01 -4.83315110e-01 4.38688904e-01 -1.39614189e+00 1.38176978e-01 -3.55044335e-01 5.19177675e-01 -6.73537493e-01 9.36259031e-01 -3.54563057e-01 -6.01992011e-01 -6.91982090e-01 -8.61786723e-01 9.20928419e-01 9.71275270e-02 -7.71523356e-01 -9.71395969e-01 6.02799915e-02 4.52831835e-01 5.31729698e-01 3.89570713e-01 6.23805285e-01 -9.17802870e-01 9.10550239e-04 -3.24123532e-01 -4.39199030e-01 6.03414953e-01 9.18349698e-02 3.77191305e-02 -8.47446561e-01 -5.01239300e-02 1.73187684e-02 -2.52688289e-01 1.27023184e+00 2.70206034e-01 8.68231833e-01 -1.87875807e-01 8.99750565e-04 -2.05615640e-01 1.46259654e+00 1.14475563e-02 7.96540678e-01 5.76270103e-01 4.76914942e-01 6.38083458e-01 9.20168281e-01 2.41972998e-01 5.15646338e-01 6.90593123e-01 3.15140635e-01 3.41482699e-01 -3.24191481e-01 -2.87960947e-01 2.88711637e-01 1.20789385e+00 -4.14488167e-01 -6.64064109e-01 -9.44359839e-01 6.74721241e-01 -2.14968729e+00 -9.40056264e-01 -2.45070636e-01 2.12082934e+00 6.54509306e-01 6.37877062e-02 1.82486653e-01 4.70465690e-01 7.84430325e-01 2.44774252e-01 -1.51822865e-02 -8.23452771e-01 -2.69933671e-01 2.86686253e-02 3.72842938e-01 4.82097715e-01 -9.26614761e-01 1.00456655e+00 5.89353895e+00 1.08567059e+00 -9.24684107e-01 6.46303594e-02 2.74728924e-01 1.20548144e-01 -2.62696087e-01 2.66657285e-02 -4.68838096e-01 3.18937987e-01 7.69075871e-01 -3.89185041e-01 1.64555505e-01 5.84895372e-01 3.27521563e-01 -3.64131093e-01 -9.66730297e-01 9.43454683e-01 5.94323099e-01 -1.16185856e+00 1.41337827e-01 -2.89203711e-02 6.79644883e-01 2.90775057e-02 -3.70510191e-01 8.23678598e-02 1.15500621e-01 -9.48163271e-01 6.98903143e-01 7.08123922e-01 4.52939838e-01 -8.17181528e-01 1.22704434e+00 2.88172632e-01 -7.08661497e-01 2.00644717e-01 -3.73645633e-01 -6.20746315e-02 2.68148512e-01 9.00976837e-01 -9.43825901e-01 8.21684659e-01 4.29505944e-01 7.25893736e-01 -7.03234792e-01 1.09485936e+00 -3.00702274e-01 6.73457980e-01 -8.58433470e-02 -4.91520107e-01 5.19160509e-01 -6.12761639e-02 1.01157820e+00 1.53191030e+00 4.68940884e-01 -2.66808003e-01 -1.41257077e-01 6.18459821e-01 -8.27764571e-02 5.71282268e-01 -6.62904978e-01 -1.91874281e-01 3.35627496e-01 1.18707764e+00 -1.02180052e+00 -4.60917115e-01 -2.39349946e-01 9.91411328e-01 2.12130293e-01 5.38652204e-02 -6.42009437e-01 -7.14060545e-01 1.77775607e-01 3.34664993e-02 1.59881830e-01 -3.19105923e-01 -4.63297635e-01 -1.12020445e+00 8.37303922e-02 -8.48586619e-01 4.33499992e-01 -8.03862929e-01 -1.14918327e+00 7.95146048e-01 1.83273688e-01 -1.18183327e+00 -3.61123115e-01 -3.39742154e-01 -5.21583498e-01 5.23568809e-01 -1.40099430e+00 -7.04168260e-01 -3.78802419e-01 4.24103476e-02 6.66720867e-01 -1.49329543e-01 8.76170397e-01 1.85904413e-01 -2.35013261e-01 3.12222809e-01 5.43400943e-02 -1.43704310e-01 8.92650664e-01 -1.49629056e+00 2.39376634e-01 7.64713764e-01 1.60859808e-01 5.22315919e-01 1.27958298e+00 -6.01820707e-01 -6.82931125e-01 -7.27898359e-01 1.37754214e+00 -3.83175641e-01 5.75230718e-01 2.63100505e-01 -9.41511393e-01 2.11247712e-01 4.98149067e-01 -8.63495648e-01 6.59586549e-01 -1.11279730e-02 -2.25292191e-01 1.25026450e-01 -1.06078553e+00 5.54644346e-01 9.49357867e-01 -2.01184884e-01 -1.14399493e+00 2.25896791e-01 7.43372738e-01 7.16596991e-02 -7.55739391e-01 5.40646851e-01 4.44393396e-01 -1.34781849e+00 5.91859162e-01 -3.54787260e-01 5.97870469e-01 -3.04064095e-01 -1.47167623e-01 -1.60415924e+00 -4.99638030e-03 -2.61732280e-01 5.85840568e-02 1.30716956e+00 2.81474859e-01 -6.05527341e-01 5.35489380e-01 1.64725393e-01 -3.16204756e-01 -7.57640481e-01 -7.13003814e-01 -1.02565932e+00 2.12193862e-01 -4.31902438e-01 5.83797216e-01 9.38304901e-01 2.36527830e-01 4.20552820e-01 -1.52375653e-01 -4.62582856e-01 3.50687653e-01 9.41115245e-02 7.51340210e-01 -1.37823904e+00 3.29068527e-02 -8.52169931e-01 -5.80144048e-01 -7.49926805e-01 4.69129235e-02 -9.35675919e-01 2.38669151e-03 -2.07544303e+00 2.16049403e-01 1.35890320e-01 -3.25712562e-01 3.19229543e-01 -1.55793399e-01 3.67037982e-01 3.23802143e-01 2.78392673e-01 -8.81833136e-01 5.43633223e-01 1.18930495e+00 6.15682118e-02 -5.58634438e-02 -4.17257726e-01 -9.86713469e-01 8.53223205e-01 1.12834454e+00 -4.12020028e-01 -2.94167370e-01 -3.64379138e-01 4.88870800e-01 -4.19215500e-01 3.22735757e-01 -1.16963458e+00 1.13744281e-01 2.39348993e-01 -2.47409329e-01 -6.63998842e-01 -8.38832185e-02 -3.64679039e-01 -2.10669488e-01 2.32778907e-01 -3.43141377e-01 1.50230750e-01 -3.90582159e-02 3.76636565e-01 -6.59844577e-01 -7.52085268e-01 5.55825830e-01 -2.96412468e-01 -9.41779971e-01 -3.77583206e-01 -3.70571017e-01 2.25010425e-01 7.45304465e-01 -4.43911105e-01 -4.06190276e-01 -5.82848728e-01 -3.47814083e-01 -6.47312552e-02 7.12927461e-01 6.27761006e-01 4.53119606e-01 -9.36396599e-01 -1.02724278e+00 -6.35065615e-01 2.91561961e-01 -3.06198180e-01 1.11898981e-01 8.60408068e-01 -8.39139283e-01 8.79745185e-01 -1.93639666e-01 -3.29841942e-01 -1.29512286e+00 4.37689126e-01 5.35288118e-02 -5.37390471e-01 -5.74698269e-01 3.72100145e-01 -3.62419277e-01 -2.90474087e-01 3.92088033e-02 -3.15111309e-01 -6.18377090e-01 4.66380060e-01 4.48065728e-01 8.80959630e-01 3.86349916e-01 -8.10836017e-01 -2.48228848e-01 3.99165332e-01 -7.66173825e-02 -3.47795784e-01 1.24681187e+00 -1.31227687e-01 -3.87413621e-01 6.23267293e-01 1.08173859e+00 -5.19781411e-02 -4.34285402e-01 -4.28529158e-02 3.11459988e-01 -4.31816310e-01 5.42467460e-02 -9.57290411e-01 -6.91993535e-01 8.98053944e-01 1.86684966e-01 6.57675624e-01 9.74857569e-01 -1.07295595e-01 9.88943100e-01 5.85082173e-01 4.33129728e-01 -1.43439353e+00 -3.93464714e-02 7.82210588e-01 1.23351073e+00 -9.80272412e-01 1.33280620e-01 -3.28257114e-01 -8.17711771e-01 1.13681185e+00 2.52840191e-01 -9.16798785e-02 1.61364704e-01 -3.79895300e-01 1.49834361e-02 -2.66821712e-01 -3.95900428e-01 -3.78643990e-01 3.96907628e-01 4.66927320e-01 7.76143610e-01 1.58077449e-01 -1.20472491e+00 3.56695950e-01 -6.37109220e-01 1.45627886e-01 8.10700059e-01 6.79981112e-01 -7.89787471e-01 -1.21278965e+00 -1.41335458e-01 5.94500661e-01 -6.70877099e-01 9.32651106e-03 -8.03632200e-01 7.06664503e-01 -3.12497355e-02 1.30899465e+00 -2.43461475e-01 -3.22539538e-01 3.78463507e-01 2.00937361e-01 5.53527653e-01 -7.03156233e-01 -6.79289818e-01 -3.57765675e-01 4.11855489e-01 -4.38041657e-01 -8.69844019e-01 -6.33985877e-01 -1.12935364e+00 -4.10940379e-01 -3.69787663e-01 4.75961626e-01 8.47501397e-01 8.68717909e-01 3.46592844e-01 4.87747908e-01 1.49766281e-01 -7.39580214e-01 -5.75284541e-01 -1.39506507e+00 -5.49142182e-01 6.71750724e-01 -1.35237917e-01 -5.73207021e-01 -4.09706354e-01 -1.29755840e-01]
[11.996589660644531, 9.321993827819824]
1e50303e-93fb-42fb-9c41-c865ebe0b87a
integrating-user-feedback-under-identity
null
null
https://openreview.net/forum?id=SygLHbcapm
https://openreview.net/pdf?id=SygLHbcapm
Integrating User Feedback under Identity Uncertainty in Knowledge Base Construction
Users have tremendous potential to aid in the construction and maintenance of knowledges bases (KBs) through the contribution of feedback that identifies incorrect and missing entity attributes and relations. However, as new data is added to the KB, the KB entities, which are constructed by running entity resolution (ER), can change, rendering the intended targets of user feedback unknown–a problem we term identity uncertainty. In this work, we present a framework for integrating user feedback into KBs in the presence of identity uncertainty. Our approach is based on having user feedback participate alongside mentions in ER. We propose a specific representation of user feedback as feedback mentions and introduce a new online algorithm for integrating these mentions into an existing KB. In experiments, we demonstrate that our proposed approach outperforms the baselines in 70% of experimental conditions.
['Andrew McCallum', 'Nicholas Monath', 'Ari Kobren']
2018-11-17
null
null
null
akbc-2019
['entity-resolution']
['natural-language-processing']
[-4.00987744e-01 9.01311278e-01 -3.42932969e-01 -3.21656108e-01 -1.00081658e+00 -7.18963325e-01 1.93815574e-01 7.15319693e-01 -4.10342753e-01 1.30388486e+00 4.55763251e-01 3.20434012e-02 5.18385582e-02 -7.92109370e-01 -8.70911002e-01 1.91353396e-01 4.06215750e-02 7.30314672e-01 6.04422212e-01 -4.95694816e-01 1.47094265e-01 2.82812715e-01 -1.18986857e+00 4.37697709e-01 9.65673029e-01 6.08835876e-01 -1.75219163e-01 3.93353760e-01 -2.98933864e-01 8.70181799e-01 -8.13376725e-01 -8.29689205e-01 7.51068145e-02 2.85839736e-01 -1.36095202e+00 -3.18630964e-01 3.86486918e-01 -2.80000478e-01 -1.57678857e-01 8.46632421e-01 4.80283022e-01 3.13622475e-01 5.71406245e-01 -1.00857365e+00 -1.05770612e+00 1.06581736e+00 -1.78129375e-01 3.98838855e-02 8.98405671e-01 -5.89227796e-01 1.27492583e+00 -1.39959049e+00 1.01680839e+00 1.06967831e+00 6.75577760e-01 4.68554616e-01 -1.27509999e+00 -5.20302296e-01 4.20384973e-01 5.44742346e-01 -1.76453698e+00 -7.31012464e-01 1.93236157e-01 -4.43740487e-01 1.15867245e+00 4.90700722e-01 1.93747416e-01 5.83554804e-01 -6.78860426e-01 7.19779491e-01 4.23016787e-01 -5.84993780e-01 2.06861660e-01 8.91828775e-01 5.84201932e-01 6.10782862e-01 6.32218421e-01 -3.44855011e-01 -9.02458668e-01 -5.41194081e-01 5.21323025e-01 -6.45444751e-01 -5.32979190e-01 -3.11656028e-01 -6.78267717e-01 3.48969430e-01 5.18519521e-01 -8.77792835e-02 -4.58718628e-01 -1.43886447e-01 6.07375689e-02 4.03415263e-01 3.71659875e-01 7.89430439e-01 -9.70759571e-01 1.16300397e-01 -3.72152269e-01 3.97315800e-01 1.16476464e+00 1.30239284e+00 1.11570108e+00 -6.95728242e-01 -2.94158876e-01 7.31893420e-01 3.70011687e-01 1.50743514e-01 2.90107280e-01 -8.51623714e-01 4.25305843e-01 1.04653227e+00 1.09115243e+00 -1.01387835e+00 -3.85366082e-01 -3.18096280e-01 -1.74950168e-01 -4.92223263e-01 1.87126219e-01 -1.91120967e-01 -8.01315665e-01 1.53017259e+00 7.63768673e-01 1.89283326e-01 3.23292047e-01 6.22075200e-01 1.16615999e+00 2.09011808e-01 1.28235623e-01 -3.99485528e-01 1.13010669e+00 -6.06083632e-01 -9.74639893e-01 -4.68444191e-02 9.29463923e-01 -4.47342068e-01 5.90509176e-01 1.99001990e-02 -8.93075943e-01 -2.25490391e-01 -7.79959977e-01 -1.55525014e-01 -5.96895397e-01 1.94505434e-02 6.12562418e-01 5.95133007e-01 -1.08967113e+00 5.03225327e-01 -4.87068444e-01 -3.37954193e-01 5.19410893e-02 3.72821569e-01 -4.23912644e-01 -1.77965146e-02 -1.83309674e+00 1.24373364e+00 6.12589478e-01 7.59519339e-02 -2.59559453e-01 -9.70649719e-01 -9.66269255e-01 4.90214787e-02 8.52571249e-01 -8.21949244e-01 1.54721379e+00 -2.60067701e-01 -9.03489470e-01 2.52008647e-01 -5.42638242e-01 -2.86819607e-01 1.86436534e-01 -6.23972714e-01 -6.04649127e-01 -3.28224361e-01 2.94133276e-01 2.39909753e-01 3.02078366e-01 -1.60056460e+00 -9.89959002e-01 -2.31401488e-01 3.23962927e-01 4.18616325e-01 -1.94491014e-01 8.53243768e-02 -6.65904820e-01 -2.16862500e-01 9.15661380e-02 -6.46120369e-01 -2.47986004e-01 -5.45402050e-01 -4.88531411e-01 -4.04638320e-01 4.21942472e-01 -9.66247201e-01 1.87018299e+00 -1.68263948e+00 1.06107414e-01 5.45752585e-01 3.34312141e-01 2.67576188e-01 2.20092610e-01 5.50956488e-01 1.48951158e-01 6.14647746e-01 2.17905685e-01 -1.52115494e-01 -2.44085994e-02 1.35499477e-01 -5.46395242e-01 -2.71034539e-01 1.66661695e-01 9.91811275e-01 -1.07877398e+00 -4.13114399e-01 -4.87430573e-01 1.62289232e-01 -3.94284248e-01 9.18211490e-02 -3.40632111e-01 1.38260230e-01 -4.62126851e-01 6.85922682e-01 3.72097433e-01 -3.03670913e-01 4.48521435e-01 -2.63360739e-01 7.25743994e-02 5.75899899e-01 -1.53583848e+00 1.23995852e+00 -1.75965995e-01 3.38095203e-02 -1.55174464e-01 -3.75581443e-01 6.06079638e-01 5.55950046e-01 1.75135285e-01 -2.36393407e-01 -4.71046120e-01 2.14793697e-01 -4.04193461e-01 -4.50303972e-01 1.07298660e+00 3.69419456e-01 -1.64397821e-01 4.30215597e-01 8.27716812e-02 2.69526988e-01 4.15528715e-01 9.23544288e-01 1.22042310e+00 -1.23875529e-01 7.08192408e-01 2.67587274e-01 5.90140402e-01 1.11060731e-01 6.75560713e-01 7.97245622e-01 7.60295838e-02 -6.01170622e-02 2.04895958e-01 2.47088484e-02 -5.71050227e-01 -8.51118505e-01 -6.54403716e-02 1.09660995e+00 2.04483584e-01 -9.34586883e-01 -1.86252221e-01 -1.05962253e+00 6.41671956e-01 8.42939198e-01 -7.68168986e-01 -1.49897173e-01 -2.62657106e-01 -4.02779967e-01 3.75916362e-01 7.15461195e-01 4.64071296e-02 -7.35668361e-01 7.34176412e-02 4.27275747e-01 -5.49470186e-01 -9.62713122e-01 -2.98794031e-01 -4.70602773e-02 -5.24169385e-01 -1.17606843e+00 -1.93374410e-01 -3.30146372e-01 7.20444202e-01 -3.81434103e-03 1.23363280e+00 1.54523239e-01 3.41214351e-02 7.00352609e-01 -4.15174723e-01 -6.07212603e-01 -3.60001415e-01 2.03179643e-01 2.13633254e-01 -2.82849401e-01 4.99513566e-01 -3.50259870e-01 -3.06178629e-01 3.19708318e-01 -5.01585186e-01 -2.43926153e-01 4.05409664e-01 5.97100317e-01 4.88649845e-01 -8.89310241e-02 1.08228636e+00 -1.61033738e+00 8.90349984e-01 -8.43016386e-01 -3.06195825e-01 8.60137641e-01 -1.13662148e+00 3.04707438e-01 7.23102391e-02 1.77098606e-02 -1.50539696e+00 -8.92747473e-03 1.34284317e-01 7.75181204e-02 1.19702525e-01 1.08701408e+00 -1.74816117e-01 -7.51158372e-02 1.10764813e+00 -3.53847325e-01 -8.07778060e-01 -6.40977859e-01 8.41418147e-01 7.77898192e-01 6.07062936e-01 -7.06635714e-01 7.39302695e-01 7.49823153e-02 -6.19154155e-01 -2.76758671e-01 -1.16126072e+00 -1.02841783e+00 -7.94993043e-01 -1.22194432e-01 1.04090333e-01 -1.24584985e+00 -5.69335938e-01 2.39634011e-02 -1.23162377e+00 2.48461232e-01 -4.77719396e-01 7.81240836e-02 -5.26441447e-03 3.91398102e-01 -6.43292606e-01 -1.08829272e+00 -1.07751906e-01 -4.99198169e-01 6.85034156e-01 3.83641839e-01 -5.67673862e-01 -7.31979012e-01 1.61175668e-01 4.87891316e-01 1.87583134e-01 -2.14323521e-01 7.72322416e-01 -9.29275036e-01 -7.77814448e-01 -4.28037077e-01 -2.75489986e-01 -1.17812537e-01 2.92098939e-01 -1.96761861e-01 -9.35923398e-01 6.79812655e-02 -5.84469020e-01 -3.20317060e-01 8.38330209e-01 -3.69573981e-01 6.11221373e-01 -5.62072098e-01 -6.64725959e-01 1.80856548e-02 1.10240102e+00 2.41937283e-02 5.44062853e-01 2.70403057e-01 6.65847659e-01 5.49593270e-01 9.24278975e-01 5.71422338e-01 9.94959295e-01 5.88016391e-01 1.52049318e-01 3.15807611e-01 2.22382307e-01 -3.60354900e-01 1.01166241e-01 5.37596166e-01 -2.10746437e-01 -1.66615859e-01 -7.33436942e-01 8.01809847e-01 -2.25941181e+00 -7.78793633e-01 -1.37437567e-01 2.35566092e+00 1.46344054e+00 -8.79756510e-02 -1.26722246e-01 -1.69907138e-01 6.59505844e-01 -5.49919128e-01 -7.40859509e-01 -1.37552932e-01 5.64551875e-02 1.05539985e-01 7.05561161e-01 7.93879151e-01 -8.96135449e-01 1.11407006e+00 6.55239010e+00 9.92071554e-02 -4.30276245e-01 -6.44978806e-02 4.45531942e-02 1.31028444e-01 -6.26759708e-01 6.76433519e-02 -1.40505230e+00 -1.60738807e-02 8.70597839e-01 -7.09602833e-01 2.01431051e-01 8.54557276e-01 -1.80523619e-01 -2.02977285e-01 -1.28988147e+00 5.27452648e-01 -1.87211812e-01 -1.37704813e+00 3.89660820e-02 -3.20781529e-01 6.36633635e-01 -1.61332935e-01 -3.70316923e-01 6.87152147e-01 1.06065118e+00 -6.50080860e-01 4.66356128e-01 6.77186906e-01 7.77584374e-01 -6.70176983e-01 8.13619435e-01 2.77488977e-01 -1.02864456e+00 1.00499596e-02 -2.16774359e-01 1.93879321e-01 1.52424023e-01 5.75062573e-01 -1.64523935e+00 1.03699696e+00 6.13008499e-01 4.62812126e-01 -6.73161328e-01 8.97564054e-01 -5.64213216e-01 3.59790534e-01 -1.58999607e-01 2.06833139e-01 -4.72661167e-01 3.48309070e-01 4.37940031e-01 1.29310799e+00 -6.38861954e-03 4.79076982e-01 2.11427972e-01 6.85351253e-01 -5.91284394e-01 3.62085223e-01 -2.13331506e-01 -1.25990555e-01 1.10302770e+00 1.09522510e+00 1.03715640e-02 -6.99613214e-01 -3.09862852e-01 9.31953549e-01 8.86228621e-01 5.44618547e-01 -3.09916884e-01 -3.35408598e-01 7.64055848e-01 2.44119596e-02 2.44625419e-01 2.61900336e-01 -2.26188991e-02 -1.36091828e+00 3.72746021e-01 -5.72337985e-01 8.48036051e-01 -6.91964865e-01 -1.23547459e+00 1.25014871e-01 -5.44388890e-02 -4.79720324e-01 -4.05139685e-01 -4.85211499e-02 3.93529870e-02 1.09868562e+00 -1.57753015e+00 -7.62556195e-01 -2.16970697e-01 3.36261362e-01 2.89416779e-02 3.49051744e-01 1.11275744e+00 5.04443169e-01 -3.37792635e-01 7.31138647e-01 -1.60955433e-02 2.80636609e-01 1.24262738e+00 -1.63320279e+00 5.16885579e-01 7.45072544e-01 1.92707866e-01 1.24236012e+00 7.84346223e-01 -1.25218880e+00 -9.25278068e-01 -1.19375706e+00 1.36371958e+00 -9.32354271e-01 5.86404860e-01 -1.11161083e-01 -1.36106086e+00 1.17278695e+00 -1.92664310e-01 -9.37006250e-02 1.00942075e+00 9.11060631e-01 -7.44697094e-01 2.16668323e-01 -1.30301940e+00 4.28940177e-01 1.11249125e+00 -6.41307592e-01 -7.98220277e-01 3.14435601e-01 9.55268681e-01 -7.76824772e-01 -1.12065196e+00 4.75017220e-01 5.78085005e-01 -3.51974100e-01 9.50706363e-01 -1.11566389e+00 -2.51973212e-01 -6.11095548e-01 6.22772649e-02 -1.52472842e+00 -5.12890577e-01 -5.27782738e-01 -9.35858846e-01 1.38094616e+00 9.60251510e-01 -4.83589172e-01 7.12386966e-01 1.50714064e+00 3.95461032e-03 -4.01572347e-01 -4.55012083e-01 -5.22135437e-01 -4.91842628e-01 -2.17946425e-01 6.80202246e-01 1.21062696e+00 6.72672391e-01 6.11623764e-01 -1.99504420e-01 7.17743933e-01 3.25638294e-01 -1.85742632e-01 6.77006602e-01 -1.59414494e+00 -1.09264314e-01 1.72052458e-01 -2.24617377e-01 -9.39256966e-01 -1.05656840e-01 -7.77152181e-01 -1.82697866e-02 -1.82763612e+00 1.32140219e-01 -7.61873007e-01 -4.94649410e-01 7.55284131e-01 -7.20849931e-01 -2.06314132e-01 -1.26046658e-01 2.62492031e-01 -8.94638956e-01 1.34152174e-01 5.28293371e-01 1.52706146e-01 -5.54741621e-01 2.64880285e-02 -1.25756109e+00 6.54270649e-01 3.71875942e-01 -4.96269494e-01 -3.44155282e-01 -1.55335873e-01 8.27085555e-01 1.40106678e-02 -9.93458256e-02 -4.70003337e-01 7.42206275e-01 -1.58112139e-01 3.01705241e-01 -6.48597836e-01 2.27919936e-01 -6.53864145e-01 3.10165584e-01 -1.65455729e-01 -6.19630992e-01 -2.13564217e-01 1.11782968e-01 8.33928406e-01 -1.44664988e-01 -2.10810050e-01 -4.11707396e-03 -2.39098519e-01 -1.01475704e+00 9.08798277e-02 -8.18661526e-02 1.41501576e-01 7.52488375e-01 1.30192384e-01 -4.76466775e-01 -3.68723571e-01 -1.28447139e+00 7.48679399e-01 3.76241714e-01 4.58044678e-01 4.69280750e-01 -1.20958233e+00 -5.26669919e-01 -2.40869552e-01 5.99290669e-01 3.26688081e-01 -3.81711796e-02 3.93382430e-01 2.09245712e-01 5.72593391e-01 2.76861072e-01 1.64802372e-01 -1.35792768e+00 4.12122875e-01 1.36599869e-01 -3.39462221e-01 -1.28486514e-01 1.15552700e+00 -1.20180681e-01 -6.84066415e-01 5.00031352e-01 -1.39005378e-01 -5.11664808e-01 3.29545796e-01 7.73070395e-01 2.43534550e-01 5.24149954e-01 -3.73817444e-01 -3.83023590e-01 -4.58713382e-01 -6.12440169e-01 -9.87865627e-02 1.16453207e+00 -5.42749763e-01 -1.51840270e-01 6.33789778e-01 3.63430351e-01 4.98335332e-01 -4.60652918e-01 -7.88594782e-01 7.74460971e-01 -5.30674458e-01 -2.90308505e-01 -1.38800192e+00 -4.81321275e-01 -8.14304352e-02 -1.20619196e-03 1.27460688e-01 6.47860885e-01 2.22526550e-01 5.66896021e-01 9.33614314e-01 4.98102397e-01 -1.23521042e+00 -4.45660889e-01 6.21200264e-01 8.76598477e-01 -1.22206759e+00 -3.45363319e-02 -9.23996329e-01 -6.60437226e-01 7.82517791e-01 1.05038679e+00 6.05171382e-01 5.79824746e-01 1.78714529e-01 3.47920544e-02 -2.51615465e-01 -1.10112619e+00 -4.46131319e-01 4.46716636e-01 7.38490164e-01 5.61310351e-01 4.88000549e-02 -4.70243663e-01 1.06368172e+00 1.24690579e-02 3.11492056e-01 6.99522376e-01 1.07493412e+00 -6.67943180e-01 -1.22470593e+00 -1.25045314e-01 6.17062569e-01 -3.52971822e-01 -3.00405234e-01 -9.43804622e-01 3.79243582e-01 3.07790320e-02 8.49116027e-01 -3.70341897e-01 -4.61298019e-01 7.42059112e-01 5.21139085e-01 2.94056863e-01 -1.17261481e+00 -6.50440931e-01 -6.50031209e-01 8.83299589e-01 -5.31363428e-01 -1.41285822e-01 -6.19353771e-01 -1.56760824e+00 1.07921436e-01 -8.77804697e-01 7.13499308e-01 5.09796858e-01 8.86068225e-01 7.55661309e-01 2.10170150e-01 2.71112889e-01 5.47171794e-02 -5.98506808e-01 -9.32020068e-01 -4.21247035e-01 3.89764667e-01 2.04561919e-01 -8.49451840e-01 -3.96347679e-02 1.94256887e-01]
[9.390728950500488, 8.780304908752441]
a6ad62e2-4373-4945-af8c-557724018986
understanding-robust-overfitting-of
2206.08675
null
https://arxiv.org/abs/2206.08675v2
https://arxiv.org/pdf/2206.08675v2.pdf
Understanding Robust Overfitting of Adversarial Training and Beyond
Robust overfitting widely exists in adversarial training of deep networks. The exact underlying reasons for this are still not completely understood. Here, we explore the causes of robust overfitting by comparing the data distribution of \emph{non-overfit} (weak adversary) and \emph{overfitted} (strong adversary) adversarial training, and observe that the distribution of the adversarial data generated by weak adversary mainly contain small-loss data. However, the adversarial data generated by strong adversary is more diversely distributed on the large-loss data and the small-loss data. Given these observations, we further designed data ablation adversarial training and identify that some small-loss data which are not worthy of the adversary strength cause robust overfitting in the strong adversary mode. To relieve this issue, we propose \emph{minimum loss constrained adversarial training} (MLCAT): in a minibatch, we learn large-loss data as usual, and adopt additional measures to increase the loss of the small-loss data. Technically, MLCAT hinders data fitting when they become easy to learn to prevent robust overfitting; philosophically, MLCAT reflects the spirit of turning waste into treasure and making the best use of each adversarial data; algorithmically, we designed two realizations of MLCAT, and extensive experiments demonstrate that MLCAT can eliminate robust overfitting and further boost adversarial robustness.
['Tongliang Liu', 'Mingming Gong', 'Chen Gong', 'Jun Yu', 'Li Shen', 'Bo Han', 'Chaojian Yu']
2022-06-17
null
null
null
null
['data-ablation']
['computer-vision']
[-6.53844848e-02 2.90677756e-01 1.52126640e-01 -3.43355417e-01 -7.73328960e-01 -1.01877916e+00 1.88918531e-01 -4.51222450e-01 -5.58929682e-01 8.54854167e-01 -3.31212464e-03 -6.08654499e-01 -1.75530925e-01 -9.69304562e-01 -1.16815472e+00 -9.94315624e-01 8.84910896e-02 1.97856188e-01 1.45734809e-02 -5.31934917e-01 -3.28589648e-01 5.29101491e-01 -8.65821540e-01 2.05958053e-01 1.08820200e+00 7.97448695e-01 -3.33786637e-01 4.30569082e-01 1.30350202e-01 8.72589648e-01 -1.03654206e+00 -8.47501874e-01 8.14032376e-01 -3.46270263e-01 -5.69130182e-01 -4.73203123e-01 5.02657712e-01 -6.45914972e-01 -8.02174807e-01 1.25781524e+00 7.80648410e-01 1.56236812e-02 4.87277955e-01 -1.38779342e+00 -6.99992239e-01 8.96904230e-01 -2.97938436e-01 2.19647795e-01 -2.17026576e-01 4.43010956e-01 6.40479565e-01 -6.89631939e-01 1.93450943e-01 1.16640818e+00 9.56069410e-01 1.07617307e+00 -1.00991583e+00 -1.19552636e+00 1.91609591e-01 -4.93844628e-01 -1.29561102e+00 -4.77594763e-01 1.01085198e+00 -2.50062704e-01 2.51689464e-01 6.80931985e-01 2.74291247e-01 1.53556871e+00 1.36495978e-01 8.59428048e-01 8.80481184e-01 -3.47832441e-02 1.01133130e-01 2.34228387e-01 -9.02428925e-02 2.84353763e-01 1.24554239e-01 5.11433959e-01 5.41635416e-02 -3.23503733e-01 7.41787910e-01 2.34769374e-01 -3.68071109e-01 -5.84516115e-02 -5.35902381e-01 9.04420674e-01 7.56303132e-01 5.35305403e-03 -3.72658558e-02 1.22806333e-01 4.67455685e-01 7.06256986e-01 4.64773327e-01 5.83233893e-01 -6.23440266e-01 2.16953591e-01 -7.25964308e-01 4.63478476e-01 5.33722758e-01 1.03388047e+00 6.35267973e-01 3.37618262e-01 5.98471723e-02 8.23226154e-01 5.56335971e-02 6.66701794e-01 4.51178819e-01 -9.16660547e-01 9.41905737e-01 3.83037388e-01 -1.82504341e-01 -8.62467408e-01 -1.45197079e-01 -6.16336465e-01 -1.10408044e+00 4.84657347e-01 7.84261584e-01 -5.93086839e-01 -8.13027203e-01 2.17347479e+00 5.50665148e-03 -4.22712415e-02 1.26963407e-01 7.86468148e-01 5.90648711e-01 3.58576357e-01 -1.73178054e-02 -1.14792667e-01 8.11085224e-01 -6.13879800e-01 -4.90586609e-01 -4.42832768e-01 5.45134842e-01 -5.57889163e-01 1.27639604e+00 1.83486000e-01 -1.05182695e+00 -4.69140917e-01 -1.10143185e+00 3.04442883e-01 -2.89101273e-01 -4.66414720e-01 4.88648891e-01 9.39626396e-01 -5.73220491e-01 8.56777728e-01 -5.11356175e-01 3.73155683e-01 8.71050656e-01 4.46326315e-01 -5.58810472e-01 -9.60379839e-02 -1.71602488e+00 5.95472336e-01 4.84182179e-01 1.72154292e-01 -1.03386927e+00 -1.03967118e+00 -5.93741238e-01 -1.59194008e-01 3.24212641e-01 -5.69543660e-01 8.28508198e-01 -1.34163308e+00 -1.10988832e+00 7.22719371e-01 5.26552975e-01 -5.35685480e-01 1.07337689e+00 -3.28705162e-01 -4.54704702e-01 -3.25001627e-01 -3.62782419e-01 3.16188723e-01 9.77398098e-01 -1.41921568e+00 -3.99974175e-03 -4.88312364e-01 2.30304793e-01 -7.95228556e-02 -7.32992530e-01 -4.30887975e-02 -1.71898797e-01 -1.15460312e+00 -1.23606823e-01 -9.75589991e-01 -2.55891651e-01 3.08978483e-02 -6.44304395e-01 2.36222655e-01 7.40536690e-01 -3.90012294e-01 1.37710214e+00 -2.44626164e+00 -3.37904125e-01 3.02492201e-01 4.76988256e-01 7.38074124e-01 -2.44710594e-01 1.06779523e-01 -4.05322582e-01 5.84591210e-01 -3.57303709e-01 -7.97126889e-02 4.26955707e-02 6.18904591e-01 -8.88285816e-01 3.95238459e-01 1.22593820e-01 8.90704155e-01 -8.21580529e-01 -1.68813944e-01 -1.49210855e-01 2.04805732e-01 -8.67087543e-01 4.36577648e-01 7.82123357e-02 3.09574902e-01 -7.17798710e-01 5.93978286e-01 1.16828167e+00 4.18055475e-01 -2.65174985e-01 -1.58350542e-01 3.40945154e-01 -1.23608381e-01 -9.68799531e-01 1.06245720e+00 -6.86536962e-03 4.10085291e-01 1.27814040e-01 -9.00862575e-01 8.58374774e-01 1.79044127e-01 3.67088914e-01 -5.99836767e-01 2.76419491e-01 2.71402538e-01 7.43731931e-02 -4.96687293e-01 1.35052741e-01 -5.58569252e-01 -2.11368039e-01 2.44942054e-01 -1.92103133e-01 5.41847609e-02 -4.81692225e-01 1.32802233e-01 1.25845337e+00 -1.47117838e-01 -9.38160568e-02 -6.01509884e-02 2.75026441e-01 -3.65889341e-01 9.72226202e-01 9.74539816e-01 -5.42903900e-01 7.87879765e-01 4.93359268e-01 -5.61389506e-01 -1.08100522e+00 -1.42632115e+00 -8.43785927e-02 1.22006345e+00 -7.30810389e-02 -1.31094500e-01 -9.91108716e-01 -1.26993072e+00 2.19385669e-01 5.86918116e-01 -7.38265216e-01 -6.89856946e-01 -7.91511774e-01 -9.95274067e-01 1.25929618e+00 6.16130710e-01 3.49795312e-01 -1.15219092e+00 2.70377696e-01 -9.07020047e-02 7.15140924e-02 -6.99370563e-01 -7.78529823e-01 3.78411323e-01 -8.07508647e-01 -1.01847386e+00 -5.27586162e-01 -4.62179333e-01 9.82005954e-01 -3.37689323e-03 1.15022838e+00 2.07575262e-01 2.81023588e-02 -1.93174243e-01 -2.50704825e-01 -7.88960516e-01 -6.95162714e-01 -1.09005928e-01 3.89278293e-01 -2.36733466e-01 -1.21796047e-02 -8.12001646e-01 -5.53440750e-01 5.11892796e-01 -1.21475911e+00 -5.90588450e-01 5.42367339e-01 8.63837719e-01 4.89530861e-01 3.60294580e-01 7.14621186e-01 -1.24983597e+00 6.65874898e-01 -7.29118168e-01 -2.92797983e-01 3.39255556e-02 -4.86073196e-01 1.87527551e-03 1.37277865e+00 -9.12938893e-01 -6.45529985e-01 -3.88980418e-01 -7.14358151e-01 -1.04526901e+00 -2.72890069e-02 -3.52205783e-02 -7.24829197e-01 -3.16537112e-01 8.58591259e-01 1.16738096e-01 1.32564694e-01 -4.71101999e-01 3.11726630e-01 4.21222985e-01 5.45893431e-01 -8.59092355e-01 1.31149781e+00 3.37056398e-01 -3.69106323e-01 -2.23727196e-01 -1.03012991e+00 2.40448266e-01 -2.18746513e-01 -5.72029725e-02 3.86520803e-01 -6.94012046e-01 -6.40202880e-01 7.55818605e-01 -7.85475016e-01 -4.40342575e-01 -7.46181846e-01 1.78357556e-01 -3.09048653e-01 1.75350085e-01 -3.65360916e-01 -7.52413154e-01 -4.78546262e-01 -1.23348057e+00 5.48686624e-01 4.02510986e-02 1.02126375e-02 -1.09494305e+00 -1.40483946e-01 4.54888731e-01 4.38879400e-01 5.95207989e-01 8.32103908e-01 -1.08421528e+00 -2.04883084e-01 -5.49262464e-01 5.95751069e-02 1.08382535e+00 -1.63301490e-02 -5.96348904e-02 -1.23939526e+00 -6.24222636e-01 3.18079591e-01 -5.96715868e-01 6.45367384e-01 1.39853865e-01 1.69913793e+00 -8.28670800e-01 -5.05034253e-03 1.16977799e+00 1.26743352e+00 -3.87620144e-02 8.69554877e-01 3.93421739e-01 9.47710395e-01 3.66799772e-01 2.53202796e-01 2.39680737e-01 -3.42302881e-02 3.49837631e-01 8.55190158e-01 -3.22748452e-01 1.30563423e-01 -5.20730615e-01 5.76918304e-01 6.23974681e-01 1.11401401e-01 -3.82292956e-01 -6.04768038e-01 2.21903503e-01 -1.46523154e+00 -1.15661085e+00 8.27326328e-02 2.18550992e+00 1.21188390e+00 6.10984266e-01 5.18549532e-02 3.06720495e-01 4.65161204e-01 2.20462993e-01 -8.11754167e-01 -3.25504184e-01 -3.47928852e-01 1.44136995e-01 8.91141355e-01 4.19282347e-01 -1.17903388e+00 7.91149020e-01 6.89252663e+00 1.34994876e+00 -9.19270098e-01 3.31746973e-02 8.77819717e-01 -3.35339993e-01 -7.34438479e-01 -1.67217597e-01 -6.20297909e-01 8.17509890e-01 5.68941712e-01 -7.34878425e-03 6.41781807e-01 9.56072509e-01 5.49456365e-02 7.88848042e-01 -1.02232683e+00 5.31003535e-01 -2.29829460e-01 -1.12473154e+00 2.63449490e-01 9.70861241e-02 6.84246302e-01 5.47914952e-02 3.44485193e-01 6.82290316e-01 6.48264885e-01 -1.18277967e+00 8.67281914e-01 5.08833230e-01 7.06573725e-01 -9.80385065e-01 8.47528756e-01 8.22858930e-01 -6.28734648e-01 -2.60895550e-01 -5.95341265e-01 1.58859357e-01 -2.53274083e-01 7.75555789e-01 -3.41394603e-01 4.78116125e-01 6.41388178e-01 1.68759063e-01 -4.07730937e-01 4.36422169e-01 -6.26364499e-02 9.24178958e-01 -3.80713522e-01 4.99343455e-01 1.83665246e-01 1.84530474e-03 7.60064960e-01 8.39133978e-01 3.39076892e-02 -1.02130570e-01 9.41422731e-02 9.41822827e-01 -6.09826565e-01 -2.38921687e-01 -7.99567163e-01 2.33881861e-01 7.20315993e-01 9.35616672e-01 -7.75486305e-02 1.13966227e-01 -9.43522975e-02 7.42254078e-01 4.58403677e-01 3.63410652e-01 -1.12726533e+00 -4.11263704e-01 9.12341058e-01 1.49331346e-01 -1.52581975e-01 2.77071059e-01 -5.02244592e-01 -9.78596270e-01 1.86219797e-01 -1.24189985e+00 5.13713717e-01 -4.23608094e-01 -1.81175709e+00 6.37529731e-01 -3.27155352e-01 -1.40877509e+00 1.43151000e-01 -4.94044900e-01 -1.05285656e+00 8.25033307e-01 -1.18615079e+00 -1.01375782e+00 -8.75709355e-02 8.56209159e-01 1.97593823e-01 -4.47778195e-01 6.77773654e-01 5.43031573e-01 -7.43924677e-01 1.69709039e+00 3.51196408e-01 5.57578385e-01 8.41787577e-01 -1.09778500e+00 4.40583616e-01 9.06564593e-01 -1.98442683e-01 8.50326896e-01 6.23553932e-01 -3.65540475e-01 -1.25242734e+00 -1.43860769e+00 3.08185399e-01 -8.50545406e-01 7.36446202e-01 -4.16842699e-01 -1.21137989e+00 7.03036845e-01 -3.24051231e-01 3.76616865e-01 6.68307841e-01 -1.96714297e-01 -6.10072017e-01 -5.32559276e-01 -1.68973148e+00 6.46283031e-01 1.07535326e+00 -4.84776735e-01 -5.82186341e-01 2.63472289e-01 1.06656921e+00 -5.00229418e-01 -1.00966108e+00 6.28617465e-01 3.57250333e-01 -9.16492820e-01 1.35840011e+00 -9.16184962e-01 3.21629345e-01 7.09470734e-03 -1.23752341e-01 -1.28741419e+00 -1.52930602e-01 -9.88180935e-01 -6.44456670e-02 1.49959481e+00 3.33222657e-01 -8.18641245e-01 9.42983150e-01 5.47218442e-01 -2.38734856e-01 -9.31295216e-01 -9.50436890e-01 -9.83992279e-01 9.18035328e-01 -6.16217792e-01 8.57511640e-01 1.16704047e+00 -4.53081369e-01 -3.06458682e-01 -5.65558612e-01 2.18734428e-01 6.92760050e-01 -4.91826206e-01 7.99718797e-01 -8.60403538e-01 -3.58142674e-01 -4.83884364e-01 -1.69477686e-01 -7.27381468e-01 1.93495616e-01 -9.69330907e-01 1.01742074e-01 -6.40519977e-01 -2.89487410e-02 -9.05964017e-01 -5.54747403e-01 7.33188629e-01 -6.80090308e-01 2.49827519e-01 2.47770086e-01 3.23698133e-01 -3.17427427e-01 5.33166230e-01 1.47614872e+00 -1.86581761e-01 5.08945659e-02 3.54585171e-01 -1.16915011e+00 8.25914383e-01 7.84924567e-01 -7.81494200e-01 -5.08617640e-01 -5.22654474e-01 2.38966510e-01 -2.67840356e-01 3.26289922e-01 -8.64567280e-01 -1.18941352e-01 -3.08995754e-01 4.56394970e-01 -1.00804634e-01 -1.80760249e-01 -1.05357587e+00 6.26442507e-02 3.63372207e-01 -4.05682325e-01 -1.47626206e-01 3.23350132e-01 3.89939815e-01 7.91126210e-03 -2.42644921e-01 1.14032388e+00 4.14191373e-02 -6.84443861e-02 6.81570292e-01 -4.75036167e-02 7.29944527e-01 9.81451511e-01 -5.28020710e-02 -4.30300057e-01 -1.51853308e-01 -6.02209032e-01 2.58414090e-01 4.56412107e-01 3.88765752e-01 5.40840924e-01 -1.52439213e+00 -7.16248810e-01 6.60993218e-01 -1.12104990e-01 3.76789987e-01 2.58202434e-01 3.89270067e-01 -2.56257921e-01 -3.92165124e-01 -1.00534201e-01 -3.33474912e-02 -9.69784319e-01 6.83141172e-01 7.02272534e-01 -3.09742421e-01 -4.61829662e-01 1.30548871e+00 5.26757300e-01 -8.59835923e-01 4.05237108e-01 1.54222804e-03 1.85430452e-01 -3.12471032e-01 4.99336392e-01 3.72804195e-01 -6.47019148e-02 -2.06261933e-01 -2.65602142e-01 2.11490422e-01 -3.72966111e-01 4.44636166e-01 1.21499872e+00 1.58477858e-01 -6.27473521e-04 -1.73808150e-02 1.40656888e+00 3.37046862e-01 -1.60860634e+00 -1.40124917e-01 -5.26337266e-01 -4.01697636e-01 -3.25349540e-01 -6.94156587e-01 -1.46438146e+00 6.78892970e-01 3.56437802e-01 4.04681027e-01 1.28762221e+00 -2.00517535e-01 9.76957083e-01 2.78631389e-01 8.73865560e-02 -9.19854462e-01 5.59550337e-02 7.09454298e-01 9.60186720e-01 -1.19840527e+00 -1.44384101e-01 -7.42374733e-02 -6.66683853e-01 7.38330781e-01 8.90889287e-01 -3.73645127e-01 7.42171288e-01 7.29465365e-01 3.22210789e-01 2.07126997e-02 -3.81920248e-01 4.25298780e-01 6.07656725e-02 7.10898578e-01 -1.43869475e-01 -5.68659641e-02 2.12189898e-01 1.22545826e+00 -5.28250396e-01 -2.89569765e-01 1.85663491e-01 6.65483952e-01 -1.47373572e-01 -1.07906961e+00 -5.16149759e-01 4.10238087e-01 -7.92847693e-01 -1.52947038e-01 -4.52275246e-01 8.14249635e-01 6.24357104e-01 6.65735126e-01 -1.64341554e-01 -8.65157843e-01 6.57117605e-01 -8.65508839e-02 1.51177227e-01 -1.06962509e-01 -8.95818949e-01 -3.17460269e-01 -2.58022398e-01 -4.90574002e-01 2.87193596e-01 -1.11991093e-01 -1.27615786e+00 -7.63190985e-01 -3.36043864e-01 1.39170691e-01 1.49185106e-01 1.00358760e+00 3.12375892e-02 8.02959085e-01 1.25951612e+00 -4.53161806e-01 -1.22798133e+00 -8.38466108e-01 -5.09753346e-01 9.27508593e-01 5.78696311e-01 -3.26335013e-01 -9.44631219e-01 -2.15965018e-01]
[5.601353168487549, 7.943862438201904]
81d02928-9346-49d9-ad17-b085bdad56b4
neural-marionette-unsupervised-learning-of
2202.08418
null
https://arxiv.org/abs/2202.08418v1
https://arxiv.org/pdf/2202.08418v1.pdf
Neural Marionette: Unsupervised Learning of Motion Skeleton and Latent Dynamics from Volumetric Video
We present Neural Marionette, an unsupervised approach that discovers the skeletal structure from a dynamic sequence and learns to generate diverse motions that are consistent with the observed motion dynamics. Given a video stream of point cloud observation of an articulated body under arbitrary motion, our approach discovers the unknown low-dimensional skeletal relationship that can effectively represent the movement. Then the discovered structure is utilized to encode the motion priors of dynamic sequences in a latent structure, which can be decoded to the relative joint rotations to represent the full skeletal motion. Our approach works without any prior knowledge of the underlying motion or skeletal structure, and we demonstrate that the discovered structure is even comparable to the hand-labeled ground truth skeleton in representing a 4D sequence of motion. The skeletal structure embeds the general semantics of possible motion space that can generate motions for diverse scenarios. We verify that the learned motion prior is generalizable to the multi-modal sequence generation, interpolation of two poses, and motion retargeting to a different skeletal structure.
['Young Min Kim', 'Hyungun Choi', 'Cheol-Hui Min', 'Hojun Jang', 'Jinseok Bae']
2022-02-17
null
null
null
null
['motion-retargeting']
['computer-vision']
[ 4.48307127e-01 4.17691648e-01 -3.79228950e-01 -6.23928010e-02 -5.80649137e-01 -8.06090653e-01 6.46792471e-01 -8.42420816e-01 2.34705746e-01 4.82922167e-01 7.52244413e-01 3.05980802e-01 -4.72559035e-02 -5.30733526e-01 -1.09086514e+00 -9.14765298e-01 -1.86565697e-01 7.57833660e-01 2.43293688e-01 -1.26410156e-01 8.04318711e-02 5.16533375e-01 -1.54034197e+00 2.17959419e-01 1.07938491e-01 1.89212635e-01 4.60544765e-01 1.17813385e+00 2.55288750e-01 8.28184247e-01 -3.40055525e-01 6.19746111e-02 5.65159321e-01 -5.33636510e-01 -8.65902305e-01 7.66263127e-01 4.75172490e-01 -6.15535438e-01 -7.22750306e-01 7.11966872e-01 1.13674387e-01 2.98811883e-01 7.30802178e-01 -1.03001928e+00 -2.48459160e-01 3.56915891e-01 -5.46714187e-01 -1.19168952e-01 7.82841325e-01 4.17736024e-01 1.03760612e+00 -6.55335605e-01 1.43754172e+00 1.35217237e+00 3.31427693e-01 9.87685382e-01 -1.09214628e+00 -1.35422587e-01 -2.70133577e-02 -3.44628654e-02 -1.10100234e+00 -3.77491415e-01 8.67477357e-01 -6.30640566e-01 5.35756886e-01 8.20243359e-02 1.01406050e+00 1.60290253e+00 2.29483783e-01 9.25757408e-01 1.62700877e-01 -2.46485963e-01 2.66191810e-01 -9.78753090e-01 -3.33160400e-01 8.09657335e-01 1.59572482e-01 2.04262048e-01 -9.06991780e-01 -3.17243487e-01 1.41418278e+00 2.16325317e-02 -4.47069079e-01 -7.20481932e-01 -1.84195006e+00 4.61957753e-01 -1.24591507e-01 -1.35356188e-01 -5.57361722e-01 6.60864592e-01 -2.19845120e-02 -5.29125221e-02 -5.39955758e-02 2.74872661e-01 -4.08828497e-01 -2.96993911e-01 -9.39415514e-01 7.28189170e-01 7.47222543e-01 1.13721907e+00 7.46396840e-01 4.07769799e-01 -1.50322411e-02 1.45703197e-01 5.35735250e-01 7.58183837e-01 5.10758638e-01 -2.02082920e+00 4.83287066e-01 2.47844815e-01 2.97616869e-01 -8.56900811e-01 -1.21295571e-01 -6.56397343e-02 -6.48542941e-01 9.46398824e-02 4.46610332e-01 -1.93924174e-01 -8.99976969e-01 1.89090085e+00 4.54127789e-01 5.92452645e-01 1.29360288e-01 1.04106712e+00 3.00741255e-01 6.41056716e-01 -3.77218038e-01 -5.20699564e-03 1.05416536e+00 -7.67684340e-01 -4.87267435e-01 -2.26984456e-01 3.90289187e-01 -5.18145621e-01 7.17689574e-01 1.29759029e-01 -1.21363676e+00 -4.31186706e-01 -1.02909100e+00 -1.21611590e-02 7.90731728e-01 -2.94511616e-02 3.57670933e-01 4.21086438e-02 -1.01475561e+00 7.50764608e-01 -1.33391154e+00 -3.98539096e-01 -1.46205708e-01 2.35902265e-01 -5.49344838e-01 8.73513054e-03 -8.62261176e-01 3.48145366e-01 5.24555147e-01 2.10895687e-02 -1.67405236e+00 -4.03899044e-01 -1.04967582e+00 -3.72117311e-01 2.53601760e-01 -1.58827615e+00 1.34602296e+00 -9.17224228e-01 -1.72542298e+00 6.48618937e-01 -4.91147697e-01 -2.63860404e-01 5.33149719e-01 -2.58157760e-01 1.61960959e-01 6.18780375e-01 1.54704347e-01 9.95615005e-01 1.13229489e+00 -1.37073088e+00 -3.60132992e-01 -2.46804714e-01 -9.09297466e-02 6.42868221e-01 2.48698428e-01 -3.72717142e-01 -7.04718053e-01 -1.04783630e+00 5.08397639e-01 -1.39832222e+00 -5.37995040e-01 1.00054093e-01 -5.84943235e-01 2.90203840e-01 8.74708712e-01 -6.30505264e-01 8.19852471e-01 -1.86474717e+00 1.05515528e+00 2.83639938e-01 -2.57053990e-02 -6.06102586e-01 -1.99297518e-01 5.55113077e-01 2.40568072e-02 -1.72558278e-01 -4.92945969e-01 -3.38384032e-01 -9.67072770e-02 8.12205672e-01 -4.41779286e-01 5.70954502e-01 9.00978446e-02 9.06673014e-01 -1.10857403e+00 -3.49950284e-01 -8.26635659e-02 2.69861102e-01 -7.41958380e-01 4.66575086e-01 -6.30073190e-01 1.29434419e+00 -7.85604477e-01 5.79260588e-01 6.10064231e-02 -4.11971897e-01 3.23366374e-01 -4.13748436e-02 2.92327583e-01 -3.78884003e-02 -1.27347147e+00 2.51690912e+00 2.29258358e-01 4.38214362e-01 -9.00635123e-03 -6.37132406e-01 6.63326740e-01 5.18703401e-01 8.39243650e-01 2.10975841e-01 -1.95942074e-01 4.92562316e-02 -1.00992978e-01 -9.65391278e-01 5.13628483e-01 -1.92904592e-01 -7.34421313e-02 6.18454516e-01 1.18054999e-02 -1.53515726e-01 -1.04077578e-01 2.14439735e-01 1.30545115e+00 8.88296306e-01 1.32895410e-02 -2.06819661e-02 2.27698565e-01 1.93237379e-01 7.23495007e-01 3.91873419e-01 2.46948019e-01 1.08043098e+00 1.39150292e-01 -4.21276361e-01 -1.50415027e+00 -1.43553650e+00 5.25129139e-01 6.20030582e-01 2.12843090e-01 -2.58622766e-01 -6.98228419e-01 -1.37187019e-01 -2.18051508e-01 2.28524357e-01 -4.64501381e-01 -1.47432610e-02 -1.10123587e+00 -3.33310455e-01 6.15264118e-01 6.20592952e-01 1.32323921e-01 -9.11883652e-01 -9.46557045e-01 1.89715981e-01 -6.63980186e-01 -1.26792955e+00 -7.62877345e-01 -4.66337293e-01 -1.33783686e+00 -1.22150862e+00 -9.21513617e-01 -7.97122896e-01 7.95307636e-01 1.31835848e-01 1.00047338e+00 -8.04508291e-03 -1.99477956e-01 9.06369984e-01 -1.21586882e-01 3.58449638e-01 -7.36604273e-01 -2.69800991e-01 1.92896172e-01 1.31850585e-01 -3.08591157e-01 -9.48520422e-01 -8.40764046e-01 2.67505676e-01 -1.01972449e+00 3.44904631e-01 3.48649085e-01 6.88099325e-01 6.66540444e-01 -2.58935899e-01 -1.08241111e-01 -4.50402468e-01 -5.62798046e-02 -5.99815011e-01 -5.31177558e-02 -5.95231466e-02 2.07184464e-01 5.14136612e-01 1.32127538e-01 -5.85012972e-01 -1.17097402e+00 6.34858012e-01 3.59726883e-02 -9.08631444e-01 -3.17294359e-01 2.72734642e-01 -2.30807051e-01 4.35974926e-01 6.13175035e-01 3.87525380e-01 1.26380652e-01 -5.17740667e-01 8.04374456e-01 -1.54206827e-01 1.08860707e+00 -1.14254797e+00 1.06958926e+00 1.01449633e+00 3.01834017e-01 -8.69955599e-01 -3.79085422e-01 -4.43492740e-01 -1.07450283e+00 -2.69458085e-01 1.23879230e+00 -1.14337790e+00 -4.56003636e-01 3.82598430e-01 -1.42164993e+00 -4.85166609e-01 -4.68235314e-01 7.31031239e-01 -1.24381864e+00 9.57291245e-01 -8.63114476e-01 -5.54923952e-01 9.36613828e-02 -1.15282845e+00 1.61450517e+00 -2.07214445e-01 -6.93675756e-01 -8.78367066e-01 5.16950488e-01 2.31144041e-01 -4.91729438e-01 9.02233899e-01 8.36792469e-01 1.00903481e-01 -1.12732673e+00 1.12853721e-01 7.07599461e-01 -1.47581160e-01 1.96698144e-01 4.10749823e-01 -5.29628932e-01 -3.29256862e-01 3.18128616e-02 -4.03663516e-02 4.80982423e-01 5.73692143e-01 6.68923199e-01 -5.92688739e-01 -3.24067980e-01 7.69328654e-01 1.12634516e+00 6.46433532e-02 5.82735360e-01 3.16035956e-01 9.26744640e-01 7.63044477e-01 3.19123298e-01 5.80134749e-01 4.05834019e-01 7.37450778e-01 6.66405380e-01 4.22196567e-01 -1.49541274e-01 -7.31162965e-01 8.02437782e-01 1.02988815e+00 -6.29502773e-01 -4.74436730e-02 -8.45198154e-01 6.70784235e-01 -2.06028843e+00 -1.28737819e+00 -1.18944747e-02 1.97491300e+00 6.40471637e-01 -2.40583435e-01 2.33714148e-01 -4.99510802e-02 8.19065511e-01 2.67798930e-01 -7.58892834e-01 2.91718632e-01 -1.59284011e-01 -3.41257937e-02 4.19307292e-01 4.97210503e-01 -6.82169497e-01 7.22596586e-01 7.38123465e+00 2.32076705e-01 -6.94979548e-01 -1.58721879e-01 -1.31200582e-01 -1.16203427e-01 -8.25950086e-01 2.58149713e-01 -4.89587992e-01 1.19902506e-01 7.93494642e-01 -1.50284827e-01 3.29867750e-01 6.03003621e-01 4.23212945e-01 3.49267662e-01 -1.34468937e+00 8.34344983e-01 -2.31994148e-02 -1.61803734e+00 5.66632152e-01 2.05522671e-01 9.76569295e-01 -1.47529885e-01 -1.62320957e-01 -3.75115663e-01 6.36183262e-01 -8.21204185e-01 1.16156638e+00 7.15490222e-01 7.06371188e-01 -2.88152307e-01 -2.58838441e-02 8.24905634e-01 -1.15752172e+00 -2.69419439e-02 -1.58745229e-01 -2.22516945e-03 6.52018785e-01 3.18072215e-02 -7.39962578e-01 6.59758627e-01 4.13332731e-01 1.01913047e+00 2.55698916e-02 6.40106976e-01 -4.43411499e-01 6.48872137e-01 -3.22351933e-01 6.30748332e-01 1.54833004e-01 -2.23816261e-01 1.22862649e+00 7.35763729e-01 7.34137058e-01 1.89481869e-01 2.45724708e-01 7.10275173e-01 9.25300792e-02 -3.33649218e-01 -8.85842800e-01 1.37384944e-02 2.61147320e-01 7.02242792e-01 -5.46451032e-01 -3.61878335e-01 -1.37133405e-01 1.30583525e+00 -6.32394031e-02 6.96120977e-01 -6.77660942e-01 6.04411542e-01 8.34599257e-01 1.87234715e-01 3.26375306e-01 -8.03131342e-01 2.35207961e-03 -1.53500986e+00 7.75754377e-02 -8.19837809e-01 2.96710789e-01 -9.86786664e-01 -9.68480170e-01 2.50966430e-01 3.54879349e-01 -1.65583456e+00 -9.99905288e-01 -3.21427435e-01 -6.20526016e-01 5.36840022e-01 -7.40032375e-01 -1.04694366e+00 -2.82243073e-01 7.47261882e-01 9.98216987e-01 -1.58504829e-01 7.27735162e-01 -1.51300445e-01 -6.71397895e-02 -8.61460716e-02 -1.64650455e-01 2.08043888e-01 4.35294122e-01 -1.09669244e+00 6.47292793e-01 9.20298040e-01 4.03262854e-01 5.68546951e-01 9.45513666e-01 -9.05800939e-01 -1.76804566e+00 -1.12300909e+00 2.67447412e-01 -8.64333510e-01 6.74077928e-01 2.46047024e-02 -6.99691832e-01 1.16691780e+00 -9.92200151e-02 -1.25430152e-01 5.05470097e-01 -7.46115983e-01 -2.63087898e-01 5.43473005e-01 -7.05643177e-01 8.98338675e-01 1.63919353e+00 -4.50407505e-01 -9.28577006e-01 1.26574755e-01 7.91062593e-01 -8.86312902e-01 -7.70508170e-01 3.41565043e-01 7.91299820e-01 -6.02991819e-01 1.23997748e+00 -1.00373602e+00 8.69653523e-01 -6.94514751e-01 -3.94302815e-01 -9.74429429e-01 -1.92705646e-01 -1.02771115e+00 -5.46027780e-01 5.64964831e-01 1.21963188e-01 1.31392583e-01 1.34830344e+00 4.44411755e-01 -2.24474579e-01 -2.74234802e-01 -8.85525823e-01 -6.96059108e-01 -1.20892569e-01 -3.47837090e-01 6.06380761e-01 1.05982614e+00 -5.09864569e-01 7.94072598e-02 -7.56951809e-01 5.07919967e-01 9.77830291e-01 2.04487905e-01 1.17144144e+00 -8.28533292e-01 -8.50563526e-01 8.27407911e-02 -7.79576242e-01 -1.79358578e+00 4.19352204e-01 -8.66293728e-01 2.59463727e-01 -1.40708709e+00 1.90189704e-01 1.98024869e-01 3.98301482e-01 1.30986035e-01 5.93749201e-03 -9.24712792e-03 2.04108119e-01 8.54797065e-01 -1.50472492e-01 6.56730890e-01 1.69191897e+00 -1.56155258e-01 -1.55484542e-01 -5.81194386e-02 -1.50133342e-01 1.14971590e+00 3.55239451e-01 -4.85566407e-01 -8.48362863e-01 -7.68281400e-01 1.65247336e-01 7.11771011e-01 5.82372606e-01 -7.62669086e-01 2.72987306e-01 -5.52643776e-01 3.24714273e-01 -6.81608975e-01 4.65558648e-01 -7.12538898e-01 9.59348142e-01 5.42210340e-01 -3.62512499e-01 3.04536074e-01 -3.28845024e-01 1.08528113e+00 -7.73624256e-02 -7.55246654e-02 1.82023376e-01 -5.06200969e-01 -8.76934767e-01 5.75225055e-01 -5.17233491e-01 -2.63252202e-02 8.45604181e-01 -6.82639241e-01 1.08551934e-01 -5.90403199e-01 -1.30712450e+00 5.06512150e-02 9.17551041e-01 3.96134466e-01 7.29983926e-01 -1.53759050e+00 -8.54782939e-01 1.10662200e-01 -7.50081316e-02 5.96072078e-01 2.00174958e-01 3.33158046e-01 -9.47428048e-01 -4.55571450e-02 -3.59129727e-01 -1.23027837e+00 -9.25346375e-01 2.38295570e-01 1.90353587e-01 -1.79608620e-03 -1.20525706e+00 5.40190279e-01 3.82129341e-01 -1.65964186e-01 -1.97369922e-02 -3.87236714e-01 1.39860421e-01 -5.96425474e-01 2.43520454e-01 4.39777195e-01 -6.08127177e-01 -1.03970945e+00 -1.62371080e-02 1.00430667e+00 6.04659379e-01 -6.77600980e-01 1.10127378e+00 -3.18186462e-01 -2.08596457e-02 6.63378060e-01 9.67330754e-01 1.05629064e-01 -1.88900352e+00 6.76460490e-02 1.93075597e-01 -4.40322787e-01 -8.83888185e-01 1.23728231e-01 -9.29789007e-01 6.00318372e-01 -1.27759323e-01 -4.90238994e-01 7.05852866e-01 1.91374227e-01 8.90616536e-01 5.76407433e-01 7.75869846e-01 -6.47106349e-01 4.37293768e-01 4.98426229e-01 9.27129626e-01 -6.53678954e-01 -4.71175835e-02 -4.92336929e-01 -6.39736295e-01 1.41656208e+00 3.65692735e-01 -3.14762801e-01 5.34705400e-01 2.70648390e-01 -2.52033235e-03 -2.42748559e-01 -9.00219023e-01 7.68254697e-02 2.65005767e-01 8.46825182e-01 -8.07903986e-03 -3.71762738e-02 7.66972303e-02 2.59589255e-02 -3.82278353e-01 4.95020719e-03 6.96912885e-01 9.83590484e-01 -3.24129850e-01 -1.05967391e+00 -5.73680937e-01 -2.56419837e-01 -2.40737334e-01 4.62560713e-01 -3.05264205e-01 6.14432096e-01 1.97570096e-03 4.70487267e-01 1.10618442e-01 -2.76808858e-01 1.71620622e-01 9.56902280e-02 8.32583427e-01 -8.30179751e-01 1.32007509e-01 4.07500386e-01 7.85602629e-02 -8.85019660e-01 -9.91845012e-01 -8.64161432e-01 -1.58068693e+00 4.07520272e-02 4.24415737e-01 -4.24300134e-03 1.81228444e-01 8.10319781e-01 1.67982399e-01 1.99583203e-01 3.89130473e-01 -1.36299706e+00 -3.84203553e-01 -4.96760905e-01 -4.43983704e-01 9.55389202e-01 5.31544566e-01 -5.69418907e-01 -3.46180290e-01 1.08591139e+00]
[7.356040000915527, -0.3012984097003937]
5b34f9fc-8051-4444-98ab-c87d825fab84
cross-lingual-training-with-dense-retrieval
2109.01628
null
https://arxiv.org/abs/2109.01628v1
https://arxiv.org/pdf/2109.01628v1.pdf
Cross-Lingual Training with Dense Retrieval for Document Retrieval
Dense retrieval has shown great success in passage ranking in English. However, its effectiveness in document retrieval for non-English languages remains unexplored due to the limitation in training resources. In this work, we explore different transfer techniques for document ranking from English annotations to multiple non-English languages. Our experiments on the test collections in six languages (Chinese, Arabic, French, Hindi, Bengali, Spanish) from diverse language families reveal that zero-shot model-based transfer using mBERT improves the search quality in non-English mono-lingual retrieval. Also, we find that weakly-supervised target language transfer yields competitive performances against the generation-based target language transfer that requires external translators and query generators.
['Jimmy Lin', 'He Bai', 'Rui Zhang', 'Peng Shi']
2021-09-03
null
null
null
null
['passage-ranking']
['natural-language-processing']
[-3.19385737e-01 -5.32185674e-01 -4.24742430e-01 -2.70033982e-02 -1.97109747e+00 -9.35222626e-01 9.14952815e-01 -4.20892984e-02 -8.77698481e-01 1.20211053e+00 5.69908679e-01 -2.97698349e-01 -1.78812206e-01 -7.59550989e-01 -5.53248167e-01 -2.19833925e-01 7.06479549e-02 9.82241511e-01 4.99634951e-01 -8.43028665e-01 2.91774571e-01 9.91018414e-02 -8.75183046e-01 6.93281412e-01 1.03551269e+00 1.81639075e-01 2.41283581e-01 7.78172374e-01 -1.10032931e-01 7.07764864e-01 -6.50025189e-01 -4.60299641e-01 -1.03089437e-01 -5.53173482e-01 -1.06655049e+00 -7.60101438e-01 2.92621136e-01 -3.41701835e-01 -3.80032361e-01 8.44056487e-01 1.06096160e+00 2.62828767e-01 1.05058992e+00 -4.72816676e-01 -1.28161645e+00 8.43609154e-01 -3.78037214e-01 3.38717252e-01 6.02083921e-01 -5.68580449e-01 1.18523097e+00 -1.38255167e+00 1.05425239e+00 1.44070971e+00 4.64766622e-01 7.31140733e-01 -7.82364368e-01 -4.39097762e-01 -4.37066823e-01 2.24523127e-01 -1.75489426e+00 -3.81555498e-01 2.43285179e-01 2.91080382e-02 1.43753231e+00 3.67358834e-01 1.25735812e-02 1.15815914e+00 1.07545197e-01 8.99573445e-01 8.94703567e-01 -1.06474757e+00 -2.08281815e-01 5.09066820e-01 7.64870457e-03 5.30995667e-01 1.28454611e-01 -1.34601057e-01 -7.59807765e-01 -2.47598276e-01 2.28477478e-01 -4.78627741e-01 -3.28010768e-01 1.07161745e-01 -1.08942544e+00 9.24208343e-01 1.78438872e-01 5.46138227e-01 4.25644331e-02 -1.81748793e-01 8.01650703e-01 7.33340919e-01 8.88231635e-01 6.39831960e-01 -8.45254004e-01 -3.11174728e-02 -8.75955880e-01 1.84930369e-01 8.35994303e-01 1.51143122e+00 6.03744745e-01 -1.98414996e-01 -5.69546819e-01 1.36825693e+00 2.58345008e-01 9.64159846e-01 9.67441261e-01 -1.95350617e-01 6.99847281e-01 2.52169549e-01 1.88117206e-01 -5.31032801e-01 6.58850521e-02 -3.10453087e-01 -5.62146425e-01 -6.33465111e-01 -7.24302903e-02 -3.57541367e-02 -9.71877813e-01 1.44701099e+00 -1.44913033e-01 -6.39535248e-01 7.27411389e-01 6.63181961e-01 1.12217140e+00 1.07692802e+00 4.67348695e-02 -1.21154226e-01 1.14434361e+00 -1.37625039e+00 -5.33837974e-01 1.70079887e-01 9.10282373e-01 -1.31600726e+00 1.44797814e+00 -1.48724154e-01 -1.05109024e+00 -5.19895434e-01 -5.99693179e-01 -4.42483634e-01 -7.66684592e-01 5.45090139e-01 2.86048263e-01 5.50705492e-01 -1.17738509e+00 3.32664810e-02 -5.91186225e-01 -9.33640599e-01 -2.13648692e-01 1.87336624e-01 -2.38257468e-01 -5.46062946e-01 -1.69896829e+00 9.79532540e-01 7.40241766e-01 -3.03976476e-01 -1.10724974e+00 -3.84982437e-01 -6.54306650e-01 6.59873188e-02 -3.24180536e-02 -5.76941907e-01 1.17933846e+00 -5.14637411e-01 -1.46229076e+00 8.95821869e-01 -8.26829746e-02 -2.90809512e-01 4.02795672e-01 -4.20150369e-01 -5.76853156e-01 1.37651205e-01 3.80895346e-01 6.94667339e-01 4.10486490e-01 -8.47156286e-01 -6.63079560e-01 -3.46504152e-02 5.76864891e-02 6.23551250e-01 -8.35232198e-01 6.57122731e-01 -9.48742568e-01 -6.08767509e-01 -4.17016178e-01 -9.14042056e-01 2.07298011e-01 -8.03029776e-01 -2.25047320e-01 -6.49227321e-01 5.89357734e-01 -8.66568506e-01 1.50496972e+00 -1.81390047e+00 1.32263554e-02 9.20524746e-02 -6.00670099e-01 3.81713837e-01 -3.87401521e-01 1.06811118e+00 4.77621973e-01 3.39991003e-01 1.27020061e-01 4.36738282e-02 -3.71949822e-02 -2.42156312e-02 -4.41757560e-01 -4.07490879e-02 -5.25449030e-03 1.17829680e+00 -1.21668458e+00 -1.02101576e+00 -2.55505681e-01 4.06312615e-01 -3.42847556e-01 8.12620521e-02 5.16735464e-02 -1.00254817e-02 -7.38148570e-01 1.00536513e+00 7.47525543e-02 2.53941547e-02 -2.52121557e-02 1.70722336e-01 4.13121507e-02 6.33263826e-01 -4.24554080e-01 1.95999014e+00 -6.68674290e-01 5.42268932e-01 -4.42248493e-01 -3.36442620e-01 7.24657357e-01 6.20565295e-01 5.05884700e-02 -9.76033509e-01 -2.15821579e-01 9.81127322e-01 -7.15297684e-02 -2.51819581e-01 1.00120592e+00 8.20444673e-02 -4.05886769e-01 4.66728628e-01 4.03271824e-01 -2.96353698e-01 8.36069345e-01 4.24212188e-01 1.02254164e+00 3.28421116e-01 1.07804283e-01 -4.46321636e-01 6.79881513e-01 2.88835138e-01 2.86632404e-02 9.73498106e-01 2.54634321e-01 6.53960943e-01 -3.31599861e-01 9.54267159e-02 -1.06732643e+00 -1.00458372e+00 -2.05065683e-01 1.87012041e+00 -1.37368396e-01 -5.74775457e-01 -6.36607409e-01 -6.92621469e-01 -1.69019654e-01 6.07210398e-01 -2.48721123e-01 -4.32534575e-01 -7.72692025e-01 -7.53880858e-01 9.71430063e-01 2.66231596e-01 3.57576728e-01 -1.48869133e+00 -5.75200915e-02 3.35613102e-01 -5.33601642e-01 -9.46200252e-01 -8.98674667e-01 -2.97214724e-02 -6.57682300e-01 -5.63267708e-01 -1.57101655e+00 -1.26993060e+00 6.44405484e-01 2.34960869e-01 1.35154772e+00 -1.30310766e-02 -1.33744150e-01 7.91566968e-01 -8.05820882e-01 -4.80071694e-01 -4.82325613e-01 9.38781083e-01 3.61646549e-03 -7.10173368e-01 6.92643464e-01 3.28859121e-01 -3.37213516e-01 1.71280220e-01 -8.77256155e-01 -5.16736150e-01 4.92490441e-01 1.00854445e+00 4.05377924e-01 -3.69369864e-01 7.88499832e-01 -8.65109861e-01 1.38602269e+00 -4.13348019e-01 -2.27984086e-01 9.52703893e-01 -5.82493246e-01 1.28765404e-02 4.01817232e-01 -4.39459920e-01 -1.28749704e+00 -5.57125747e-01 9.30120051e-02 1.78349078e-01 4.26506251e-01 9.73743021e-01 2.61680543e-01 9.32430401e-02 1.01585305e+00 3.69194180e-01 -7.18994379e-01 -6.42745435e-01 4.60340500e-01 9.96496439e-01 5.73682114e-02 -9.03553247e-01 6.02579355e-01 -7.98993185e-02 -5.82270503e-01 -9.24065828e-01 -7.38517046e-01 -6.67791605e-01 -5.61330140e-01 5.66190481e-03 6.58099174e-01 -1.18078780e+00 1.55665576e-01 1.88456178e-01 -1.39127982e+00 -1.74582914e-01 -2.59107322e-01 6.68470800e-01 -1.48367748e-01 1.37336463e-01 -1.23193550e+00 -4.73307192e-01 -1.10430479e+00 -7.79942095e-01 1.39765406e+00 4.74796444e-02 -7.32910037e-02 -1.10379541e+00 6.02886438e-01 2.17298530e-02 5.58787584e-01 -6.64041460e-01 1.11669433e+00 -8.53073061e-01 -3.97968858e-01 -4.98957515e-01 -1.37737572e-01 3.70618075e-01 1.31067917e-01 -1.24144316e-01 -5.21966636e-01 -7.76577055e-01 -6.46600306e-01 -1.11526918e+00 8.73336077e-01 9.45136696e-02 4.90453601e-01 -8.78959894e-02 -2.55197436e-01 2.79124118e-02 1.36415243e+00 2.38104135e-01 5.62053442e-01 5.35298109e-01 4.66846347e-01 4.89197373e-01 8.03245246e-01 -1.22048549e-01 2.64705718e-01 5.87396681e-01 -7.22410560e-01 -1.98441342e-01 -3.58003199e-01 -5.23105741e-01 6.38775766e-01 1.68553054e+00 -1.76043510e-01 -6.69177771e-01 -9.38548088e-01 7.40902185e-01 -1.62228107e+00 -6.85191095e-01 3.04908395e-01 2.34550643e+00 1.32279599e+00 -1.14518322e-01 -1.72636136e-01 -6.36704028e-01 4.51425880e-01 -3.01564842e-01 -1.75205663e-01 -4.02319312e-01 -4.75927889e-01 3.99680167e-01 3.67254108e-01 6.48975253e-01 -9.02716517e-01 1.73165715e+00 6.46998596e+00 1.28444266e+00 -9.12523150e-01 3.85283142e-01 4.19578582e-01 1.61076710e-01 -4.47358340e-01 -7.45089427e-02 -1.30152404e+00 1.17194571e-01 1.01530850e+00 -8.76955450e-01 1.92281112e-01 7.19842494e-01 -3.74902487e-01 3.07787359e-01 -9.67182159e-01 5.24591506e-01 4.62211818e-01 -8.30430746e-01 5.61681449e-01 -2.41662562e-01 1.07361090e+00 5.27097046e-01 -6.48213029e-02 1.03897309e+00 3.90731305e-01 -7.92043269e-01 4.23055530e-01 3.59705329e-01 1.16012490e+00 -8.09801638e-01 8.04820836e-01 3.81926656e-01 -9.48932767e-01 4.06780452e-01 -9.19020355e-01 5.32453358e-01 1.83316860e-02 -1.73083544e-01 -9.70370591e-01 5.42917728e-01 7.78034925e-01 3.86880934e-01 -6.89895868e-01 8.73486400e-01 -2.83575147e-01 6.65592432e-01 -2.19189465e-01 -5.05924463e-01 3.73594016e-01 -3.21537107e-02 4.84974027e-01 1.77963233e+00 5.96556664e-01 -2.24226475e-01 3.05619448e-01 3.08592528e-01 -3.58287930e-01 8.52896392e-01 -8.52157712e-01 -2.22438037e-01 1.92551672e-01 1.13535929e+00 -5.44397771e-01 -6.15587533e-01 -2.12045625e-01 1.20363617e+00 5.27728200e-01 7.37047911e-01 -2.76844263e-01 -7.91581452e-01 -2.27296278e-01 -2.84252763e-01 -4.26299162e-02 -1.59306362e-01 5.46230197e-01 -1.38259554e+00 3.30192765e-04 -1.12081575e+00 5.98983347e-01 -6.69881403e-01 -1.34006524e+00 1.01433897e+00 1.72090545e-01 -1.34587097e+00 -7.56664038e-01 -4.78778243e-01 -4.47128266e-02 1.10800719e+00 -1.57745683e+00 -1.34068120e+00 4.18967128e-01 6.40209794e-01 1.01185691e+00 -7.07248986e-01 1.33958435e+00 7.74900973e-01 -1.72284350e-01 8.72522771e-01 7.13375747e-01 3.00162941e-01 1.25219810e+00 -1.16875100e+00 1.97023332e-01 5.68612397e-01 5.63061893e-01 1.04421401e+00 2.26800621e-01 -6.36303902e-01 -1.24848497e+00 -8.51526797e-01 1.46854639e+00 -5.64194381e-01 7.03736842e-01 -3.43275249e-01 -8.11116278e-01 5.64243436e-01 6.72090888e-01 -2.30404928e-01 7.17169523e-01 2.98629016e-01 -2.26341456e-01 -3.95955630e-02 -7.67549634e-01 6.80160820e-01 6.17623985e-01 -1.04370260e+00 -6.10325158e-01 7.76765704e-01 8.04743886e-01 -2.87520617e-01 -7.53446460e-01 3.90549392e-01 5.08595347e-01 -1.39820175e-02 1.08437598e+00 -1.07522905e+00 5.67717671e-01 -4.99288328e-02 -3.31392527e-01 -1.39674056e+00 -1.93350911e-01 -3.13600957e-01 1.33555040e-01 1.57024574e+00 9.60443914e-01 -4.00800318e-01 2.31940299e-01 1.09727234e-01 -1.04924046e-01 -4.10680145e-01 -7.91919291e-01 -8.32326889e-01 6.47491217e-01 5.72699979e-02 1.28888875e-01 7.94614136e-01 2.10350174e-02 1.00379574e+00 -3.95724028e-01 -4.37515289e-01 9.75380689e-02 -4.97939857e-03 5.46380460e-01 -7.79254675e-01 -1.71705127e-01 -1.48993552e-01 6.57674298e-02 -9.76931453e-01 4.10563707e-01 -1.31239414e+00 2.55265445e-01 -1.64695692e+00 6.13406539e-01 -3.10031056e-01 -6.47676468e-01 4.02134567e-01 -2.93491572e-01 5.21764398e-01 -9.11941901e-02 3.58329833e-01 -1.01785374e+00 6.55122638e-01 1.09422541e+00 -3.69699806e-01 -9.57063511e-02 -2.06784859e-01 -3.63559932e-01 3.36643070e-01 5.53532779e-01 -6.68558121e-01 -4.47413653e-01 -9.94242966e-01 4.16055113e-01 -3.12607512e-02 -5.62861323e-01 -6.46130681e-01 1.74043149e-01 2.48399913e-01 2.38374978e-01 -5.08423269e-01 1.16716772e-01 -4.02719140e-01 -6.65827811e-01 3.64956379e-01 -7.60467112e-01 4.58512843e-01 2.77083308e-01 2.27133095e-01 -5.91794312e-01 -6.35065734e-01 2.01903880e-01 -4.57125664e-01 -7.43435919e-01 1.77772045e-01 -4.70700473e-01 5.07680714e-01 3.18440229e-01 3.55893314e-01 -3.55902314e-01 -4.62424815e-01 -1.92944586e-01 1.92528084e-01 1.41860649e-01 8.40050399e-01 5.55204570e-01 -1.49529445e+00 -1.36623800e+00 -2.31415927e-01 5.72643399e-01 -6.65656447e-01 -1.93594977e-01 3.69263768e-01 -6.55323625e-01 1.13448870e+00 2.48584077e-02 -2.51715034e-01 -1.31965113e+00 2.41560489e-01 -2.20753253e-02 -7.96566963e-01 -8.14321488e-02 8.65725279e-01 -1.47909552e-01 -9.30489659e-01 2.07810283e-01 1.77275628e-01 -3.65780741e-01 -5.72103113e-02 4.49712902e-01 4.12868351e-01 3.22299898e-01 -7.15226233e-01 -2.35433578e-01 5.64733505e-01 -6.28662229e-01 -6.69193387e-01 8.71646821e-01 -1.53960258e-01 -4.34899956e-01 5.81328630e-01 1.54697156e+00 3.30126643e-01 7.59254918e-02 -3.88098061e-01 5.82708180e-01 -1.48342818e-01 -9.69333351e-02 -9.26478684e-01 -3.68006796e-01 8.26235175e-01 5.96488714e-01 -1.23607025e-01 1.07531166e+00 2.18534637e-02 8.03857148e-01 1.12345362e+00 9.34140325e-01 -1.40807700e+00 6.46812201e-04 1.13640344e+00 1.08780849e+00 -1.28236949e+00 -2.27847658e-02 5.19636758e-02 -5.72428405e-01 9.39763308e-01 5.51303804e-01 1.93752766e-01 5.01859426e-01 -3.37021083e-01 3.04465294e-01 -1.41939390e-02 -7.83460081e-01 -2.84494847e-01 9.56495047e-01 3.30210179e-01 1.26038301e+00 -4.09245938e-02 -9.89723444e-01 5.84611148e-02 -1.84644625e-01 -6.27640858e-02 1.03837550e-01 1.12877250e+00 -2.48435706e-01 -1.51817453e+00 -1.36633784e-01 3.58020306e-01 -8.23192179e-01 -8.48038256e-01 -5.98521233e-01 8.64727318e-01 -4.30163652e-01 9.08263743e-01 -3.54149491e-01 3.44765186e-02 2.81448603e-01 1.21618718e-01 4.48515058e-01 -1.02116370e+00 -7.85004377e-01 5.98311663e-01 2.92755723e-01 3.02723013e-02 -3.54153752e-01 -2.79178351e-01 -8.94839287e-01 2.93075055e-01 -5.97850263e-01 8.85351241e-01 5.44720232e-01 6.37480855e-01 1.90500438e-01 1.83425218e-01 4.18118596e-01 -3.59093070e-01 -7.40712464e-01 -1.61928546e+00 -4.64421272e-01 2.97011435e-01 -5.56899048e-02 -8.73609334e-02 -1.42394856e-01 -4.00603712e-02]
[11.401276588439941, 9.769583702087402]
e8c484d9-d14c-442a-bdf8-7abfe44ce8b9
variational-likelihood-free-gradient-descent
null
null
https://openreview.net/forum?id=svH3klEbuXa
https://openreview.net/pdf?id=svH3klEbuXa
Variational Likelihood-Free Gradient Descent
In many scientific applications, we do not have explicit access to the likelihood function. However simulations of the process of interest, using different parameter settings, may give us access to the likelihood function implicitly. The methodology for approximating likelihoods and posterior distributions based on simulated observations can be described as simulation-based inference. In this paper, we propose a simulation-based inference algorithm in which we iteratively update particles to more closely resemble the posterior. Our approach utilises simulations to estimate a density ratio function at each iteration and then uses it to approximate the KL divergence between the particle density and the posterior density. By alternating between gradient descent and density ratio estimation, the approximated KL divergence is minimized. We benchmark the performance of our algorithm on a Gaussian mixture model and the M/G/1 queue process model and report promising results.
['Mark Beaumont', 'Song Liu', 'Jack Simons']
2021-11-22
null
null
null
pproximateinference-aabi-symposium-2022-2
['density-ratio-estimation']
['methodology']
[-1.44129723e-01 -2.93284923e-01 2.08322018e-01 -2.61586666e-01 -7.05183566e-01 -2.68316358e-01 6.67570651e-01 4.45506632e-01 -7.53854215e-01 1.12264240e+00 -2.74789870e-01 -6.93979800e-01 -5.78281805e-02 -8.69934201e-01 -6.33753121e-01 -7.77038217e-01 -1.31989524e-01 1.06609368e+00 2.74131030e-01 2.58977234e-01 4.58687395e-01 6.94811404e-01 -1.10516536e+00 -3.80749226e-01 8.68949413e-01 5.21514237e-01 2.05215961e-01 1.26860285e+00 -2.51125485e-01 4.01439756e-01 -7.80813098e-01 -3.04091930e-01 4.25809287e-02 -5.86102307e-01 -4.50086743e-01 1.47551879e-01 -2.40380213e-01 -3.65748256e-01 4.18455936e-02 1.03598547e+00 3.13252568e-01 2.02749804e-01 1.19046700e+00 -1.12754965e+00 2.21495867e-01 3.15207094e-01 -7.88180351e-01 4.84663993e-01 3.40578020e-01 1.43869355e-01 5.19164324e-01 -4.90427822e-01 2.35181764e-01 1.39715672e+00 6.99742854e-01 -1.27169788e-01 -1.74644732e+00 -4.43311036e-01 -1.94163561e-01 -2.47530028e-01 -1.73869586e+00 -2.44537428e-01 1.37270674e-01 -2.71624148e-01 6.33442283e-01 -1.43653154e-02 6.98338270e-01 6.22519732e-01 6.49528205e-01 6.50766253e-01 1.09584188e+00 -6.17608905e-01 7.07274675e-01 2.68757880e-01 2.05660295e-02 4.64136630e-01 5.20258963e-01 2.50339806e-02 -1.91600248e-02 -8.63011599e-01 1.05778277e+00 -3.85040119e-02 -1.88027799e-01 -3.32350880e-01 -8.91913891e-01 8.41658592e-01 -4.24349904e-01 -2.25098774e-01 -6.16838932e-01 6.61239505e-01 3.21412720e-02 8.85807276e-02 5.52201331e-01 -5.04809543e-02 -1.95458829e-01 -4.76146042e-01 -1.19241059e+00 5.81786335e-01 1.61178589e+00 7.89547920e-01 8.28306735e-01 -1.19652897e-01 -2.30199695e-01 6.64513886e-01 8.96992087e-01 8.89469802e-01 -1.06555775e-01 -1.27261126e+00 1.27973706e-02 -2.41840914e-01 7.66142964e-01 -4.22628224e-01 1.63436756e-01 -2.25292206e-01 -6.32997274e-01 4.09132004e-01 8.82713497e-01 -4.84452039e-01 -8.34300816e-01 1.51573050e+00 4.46999341e-01 7.02949643e-01 -5.06902672e-02 4.28836256e-01 -2.64891654e-01 9.27849948e-01 -2.76489556e-02 -5.48421621e-01 1.10678160e+00 -4.70622659e-01 -6.47391260e-01 1.81757480e-01 2.64149874e-01 -9.22659874e-01 6.38180971e-01 4.25113320e-01 -1.28629959e+00 -3.34381610e-01 -8.13095450e-01 7.04200208e-01 1.69159517e-01 -2.86875844e-01 1.94569513e-01 9.60587859e-01 -1.10077453e+00 8.33243787e-01 -1.34160793e+00 -3.71817529e-01 4.64356802e-02 3.22454393e-01 3.76248896e-01 1.77776530e-01 -7.17407048e-01 8.39376688e-01 3.01944375e-01 -7.07881600e-02 -9.49632049e-01 -5.58244944e-01 -3.36583823e-01 2.12559253e-01 2.33905688e-02 -8.22782815e-01 1.51110792e+00 -2.98573256e-01 -1.75400567e+00 2.07714945e-01 -4.44800675e-01 -5.07249057e-01 5.86251199e-01 -3.42810266e-02 -1.47308648e-01 4.48803529e-02 -2.29409337e-01 2.06968591e-01 8.91421258e-01 -1.35735261e+00 -6.84809089e-01 2.04638764e-01 -2.07729429e-01 1.84954405e-01 4.20419633e-01 -1.73127308e-01 -5.27606905e-01 -9.58756804e-02 -1.19671404e-01 -8.51001859e-01 -5.27172387e-01 -3.17268372e-02 -1.36512354e-01 7.26505965e-02 3.59828681e-01 -2.82252848e-01 1.11841178e+00 -1.89740562e+00 -3.12141269e-01 6.53530359e-01 1.55924499e-01 -1.01300806e-01 2.69178510e-01 7.37868845e-01 5.94060183e-01 -1.11796102e-02 -3.49183261e-01 -4.93260860e-01 1.15020916e-01 3.79694670e-01 -3.78466666e-01 8.73880148e-01 -7.85289854e-02 3.88019204e-01 -1.06368446e+00 -6.73331082e-01 2.25587517e-01 4.48247969e-01 -6.60640061e-01 3.85208875e-01 -1.62492335e-01 2.76184440e-01 -4.41647321e-01 3.53790857e-02 9.96115327e-01 -3.42033952e-01 4.50612694e-01 1.66088298e-01 -2.44026840e-01 -1.20152824e-01 -1.42512834e+00 9.91880834e-01 -5.02768219e-01 4.34810281e-01 4.08393174e-01 -8.08998466e-01 7.57083178e-01 2.06816182e-01 2.66704410e-01 1.32319212e-01 1.64450899e-01 1.55193537e-01 -5.59390858e-02 -8.87023509e-02 4.10854906e-01 -6.08572125e-01 1.40190497e-01 8.04694772e-01 -2.66960487e-02 -3.86046052e-01 1.69904038e-01 1.68225363e-01 9.47846115e-01 1.93629086e-01 4.68122810e-01 -4.06400084e-01 2.72292525e-01 -3.09874982e-01 1.89196706e-01 1.26506245e+00 -3.66023816e-02 3.27348918e-01 6.19653940e-01 4.61892560e-02 -1.25684714e+00 -1.82965744e+00 -4.19406831e-01 4.21917140e-01 1.54159755e-01 -2.00277880e-01 -7.65131712e-01 -1.80440769e-01 2.20232129e-01 9.55436170e-01 -3.13212574e-01 1.25241831e-01 -3.44580114e-01 -1.20486641e+00 5.28089285e-01 1.51659444e-01 1.29094094e-01 -5.65017343e-01 -4.97601032e-01 5.50484538e-01 1.64059341e-01 -8.16845119e-01 -3.75336617e-01 1.01872034e-01 -1.02039742e+00 -7.64198780e-01 -7.00343728e-01 -1.24940604e-01 6.44682825e-01 -1.58947557e-01 1.20422471e+00 -3.41704004e-02 -4.44064941e-03 5.35838544e-01 7.27834031e-02 -4.48168784e-01 -9.15741980e-01 -3.91072750e-01 1.74078822e-01 -1.09806433e-01 2.96075225e-01 -7.11811185e-01 -5.20715714e-01 1.47700176e-01 -8.02285254e-01 -3.27710032e-01 3.93224478e-01 4.90117610e-01 5.41190982e-01 5.28454268e-03 3.25203091e-01 -8.38783622e-01 9.69598055e-01 -7.18612611e-01 -9.13890481e-01 1.40606865e-01 -5.03533304e-01 3.83642673e-01 5.37464261e-01 -5.13424873e-01 -1.22718239e+00 -2.26894930e-01 -2.09819257e-01 -4.62179273e-01 -2.55619705e-01 4.66222942e-01 2.23741323e-01 1.84171721e-01 2.30207473e-01 2.65056252e-01 3.13481182e-01 -4.93613720e-01 2.19478846e-01 7.10122526e-01 4.65744376e-01 -8.74777794e-01 5.32854378e-01 6.45523727e-01 2.36391202e-01 -8.79458725e-01 -5.81033170e-01 -4.90741730e-01 -2.33607635e-01 -3.18879098e-01 4.76444155e-01 -6.78097725e-01 -1.06598139e+00 5.70588946e-01 -1.05898678e+00 -4.53047693e-01 -5.27538300e-01 1.01441455e+00 -9.28899467e-01 6.61157668e-01 -7.10201204e-01 -1.51163638e+00 8.08535144e-02 -8.73091161e-01 9.40316081e-01 3.63232762e-01 -1.71492279e-01 -1.36251473e+00 5.73464751e-01 -3.51686418e-01 3.98845673e-01 -1.50573105e-01 6.89276099e-01 -7.13371933e-01 -6.17348671e-01 -3.15179199e-01 -1.82662442e-01 1.15763664e-01 6.23396896e-02 3.40388983e-01 -5.78280032e-01 -5.52466333e-01 2.93757915e-01 2.02927098e-01 5.05399466e-01 8.48668754e-01 9.21019495e-01 -1.31542161e-01 -6.49600565e-01 5.75756371e-01 1.45857298e+00 -1.51733798e-03 7.09093690e-01 8.19542038e-04 1.10359274e-01 7.60957226e-02 5.62052131e-01 9.81909752e-01 1.50006756e-01 3.17828685e-01 -4.54390347e-02 2.61013299e-01 3.59230876e-01 -1.40298083e-01 3.35759133e-01 8.14669430e-01 1.33077547e-01 -6.60602748e-01 -9.64431643e-01 3.90035987e-01 -1.73075426e+00 -9.35679257e-01 -3.03775221e-01 2.56337166e+00 9.28549528e-01 4.12367672e-01 1.33557498e-01 -3.18736762e-01 8.85182261e-01 -4.01314884e-01 -4.21083808e-01 -4.66502577e-01 5.10206997e-01 3.84914875e-01 7.01515496e-01 9.86211061e-01 -6.54961765e-01 4.69037503e-01 7.99041462e+00 1.12000239e+00 -5.32621145e-01 -2.57015671e-03 4.79723811e-01 3.25545967e-02 -2.44560272e-01 3.72695833e-01 -1.06908345e+00 8.87268603e-01 1.55477762e+00 -4.39226657e-01 3.31877410e-01 1.54863745e-01 5.38856387e-01 -7.72327900e-01 -9.73520815e-01 6.88709378e-01 -4.97891724e-01 -9.58713412e-01 -7.09652454e-02 5.19861281e-01 5.21183312e-01 -9.35818627e-02 -1.77932248e-01 1.99462757e-01 9.92896855e-01 -9.15500283e-01 4.33227181e-01 1.15198255e+00 3.40748429e-01 -9.81684983e-01 8.67304623e-01 7.30624855e-01 -9.93458986e-01 4.08811241e-01 -4.93875861e-01 -1.95495769e-01 6.90936267e-01 9.49018896e-01 -1.02190590e+00 1.55260796e-02 2.79626369e-01 4.89687175e-03 2.65450310e-02 1.62580431e+00 2.05391482e-01 1.03361320e+00 -1.02785671e+00 -2.63031304e-01 7.10253641e-02 -8.74945879e-01 7.06257284e-01 1.45752692e+00 5.51019073e-01 -1.86999187e-01 2.77163804e-01 1.05297720e+00 1.98118448e-01 -1.19996548e-01 -1.57637358e-01 4.63233851e-02 7.37875283e-01 9.63126183e-01 -1.08248329e+00 -5.80884159e-01 -2.04970807e-01 7.85324454e-01 1.73098221e-01 6.81995511e-01 -9.59795535e-01 -3.31530094e-01 8.27866733e-01 3.17221209e-02 5.74779153e-01 -4.02697116e-01 2.37957492e-01 -8.42658997e-01 -4.88438576e-01 -4.58943665e-01 2.71806456e-02 -5.92787087e-01 -1.42928350e+00 -4.84851934e-03 5.51905930e-01 -8.12622905e-01 -7.25652397e-01 -2.33233750e-01 -7.96904802e-01 1.40674651e+00 -1.15241206e+00 -2.30870038e-01 1.81207791e-01 8.03009793e-02 1.52791977e-01 1.05936259e-01 5.48608363e-01 -3.44996005e-02 -3.04827064e-01 1.76358700e-01 9.44421470e-01 -2.01986730e-01 4.05066282e-01 -1.41591740e+00 5.16110599e-01 3.96084011e-01 -2.26431131e-01 6.41516507e-01 1.42829537e+00 -9.14588749e-01 -1.11960375e+00 -7.04636931e-01 4.57573086e-01 -3.76907676e-01 9.44409072e-01 -1.02334887e-01 -1.02230203e+00 5.37244678e-01 7.90358186e-02 -1.32766247e-01 6.54613376e-01 -3.92462388e-02 4.04780507e-01 2.33890668e-01 -1.25191331e+00 5.45305192e-01 4.52277899e-01 -3.04610938e-01 -2.66768008e-01 2.10622489e-01 1.51515409e-01 -2.64064401e-01 -1.00105524e+00 -1.39042968e-02 5.28314829e-01 -7.80529201e-01 9.28099394e-01 -3.75258654e-01 -2.82954186e-01 -4.55144376e-01 -1.94469661e-01 -1.37706268e+00 -7.11763278e-02 -7.04708755e-01 -4.62348014e-01 1.12164509e+00 2.69198120e-01 -7.15739071e-01 6.85497820e-01 3.34473670e-01 4.63121921e-01 -5.97920179e-01 -8.53759229e-01 -8.34813178e-01 2.46697009e-01 -5.36516488e-01 3.60439241e-01 1.42803997e-01 -2.95398027e-01 3.90270613e-02 -6.99261576e-02 3.68520856e-01 1.25477159e+00 7.86781535e-02 8.03781271e-01 -1.09436512e+00 -7.75615692e-01 -3.08971345e-01 -1.15358584e-01 -1.54215682e+00 1.24763235e-01 -4.86246347e-01 3.41181517e-01 -1.36895931e+00 4.89135891e-01 -4.23112959e-01 1.31159741e-02 -4.74098861e-01 -2.84379900e-01 3.31076048e-02 -1.77373394e-01 2.52596170e-01 -6.24501288e-01 5.19583821e-01 9.03660834e-01 4.12418276e-01 -1.64037310e-02 3.78326595e-01 -1.82484388e-01 8.27168882e-01 8.54120553e-01 -7.39122272e-01 -2.89912164e-01 1.79224193e-01 2.57693022e-01 3.75731349e-01 4.05113548e-01 -9.79105711e-01 3.35616052e-01 -3.17095429e-01 3.22692454e-01 -7.00790286e-01 4.73015487e-01 -4.55464333e-01 4.34733003e-01 4.10936207e-01 -1.19991474e-01 -7.50750303e-02 1.51221618e-01 1.06224465e+00 6.41082153e-02 -8.92309248e-01 7.88978457e-01 -7.60020837e-02 -2.49655508e-02 8.65052938e-02 -1.18091059e+00 1.11011900e-01 7.51937151e-01 -1.50652304e-01 1.29888788e-01 -8.66794288e-01 -8.55330467e-01 2.96398163e-01 8.81299198e-01 -5.97800910e-01 5.28565705e-01 -9.63964164e-01 -8.48973155e-01 1.60069037e-02 -4.79395896e-01 -3.14864904e-01 -9.37039778e-02 7.89806664e-01 -8.99858177e-01 -5.14564700e-02 2.47200191e-01 -7.78777778e-01 -8.77212346e-01 2.87406683e-01 5.66730678e-01 -4.43606317e-01 -2.78281063e-01 4.69632745e-01 -6.79460615e-02 -3.14614683e-01 -6.20738342e-02 -9.10398588e-02 2.53697425e-01 -3.54249209e-01 5.30557394e-01 6.53415024e-01 -3.83945435e-01 -2.00615793e-01 -1.64945394e-01 2.25286022e-01 6.52769506e-02 -7.91856825e-01 8.89887154e-01 -5.03052950e-01 -1.15770735e-01 8.75468373e-01 1.01948440e+00 7.92455003e-02 -1.60827768e+00 2.39068773e-02 -1.04382351e-01 -5.09211779e-01 8.22305530e-02 -3.38780314e-01 -6.16697192e-01 7.31701016e-01 4.40900117e-01 5.46039462e-01 5.73525369e-01 -5.51516265e-02 5.40283322e-01 3.02641600e-01 2.87141502e-01 -9.50382352e-01 -2.59097248e-01 6.44063652e-01 7.69520774e-02 -7.69645870e-01 2.95632482e-01 -1.83604911e-01 -1.96812049e-01 9.25356984e-01 1.00878842e-01 -4.40373570e-01 1.10414755e+00 8.58662844e-01 -1.81797877e-01 2.58946195e-02 -9.32882905e-01 -7.33502060e-02 -9.36691388e-02 5.25309324e-01 4.74845141e-01 4.01706509e-02 -3.55621874e-01 -1.32460549e-01 -4.64983582e-02 1.24985635e-01 7.81964481e-01 1.02135384e+00 -7.90726423e-01 -1.24229109e+00 -6.70287669e-01 7.78562725e-01 -6.42578602e-01 -1.34231597e-01 2.49240845e-01 5.68246901e-01 -4.91118044e-01 1.00777709e+00 5.88539362e-01 4.12294120e-01 -3.64076230e-03 9.66879800e-02 8.24544668e-01 -4.26168680e-01 -8.08079243e-02 3.65712553e-01 9.90055576e-02 -1.74100220e-01 -2.25269973e-01 -9.34700131e-01 -1.19420767e+00 -8.64673138e-01 -4.32469994e-01 7.16103852e-01 6.97933793e-01 9.89698350e-01 -1.06049232e-01 3.25974435e-01 3.63861471e-01 -9.38762486e-01 -9.97513354e-01 -8.64807129e-01 -1.03351271e+00 2.17888709e-02 1.86154470e-01 -5.08415520e-01 -5.88526368e-01 -4.52435128e-02]
[6.757779598236084, 3.9725563526153564]
ab294254-a580-470c-bd99-91e0eae44b40
safe-mutations-for-deep-and-recurrent-neural
1712.06563
null
http://arxiv.org/abs/1712.06563v3
http://arxiv.org/pdf/1712.06563v3.pdf
Safe Mutations for Deep and Recurrent Neural Networks through Output Gradients
While neuroevolution (evolving neural networks) has a successful track record across a variety of domains from reinforcement learning to artificial life, it is rarely applied to large, deep neural networks. A central reason is that while random mutation generally works in low dimensions, a random perturbation of thousands or millions of weights is likely to break existing functionality, providing no learning signal even if some individual weight changes were beneficial. This paper proposes a solution by introducing a family of safe mutation (SM) operators that aim within the mutation operator itself to find a degree of change that does not alter network behavior too much, but still facilitates exploration. Importantly, these SM operators do not require any additional interactions with the environment. The most effective SM variant capitalizes on the intriguing opportunity to scale the degree of mutation of each individual weight according to the sensitivity of the network's outputs to that weight, which requires computing the gradient of outputs with respect to the weights (instead of the gradient of error, as in conventional deep learning). This safe mutation through gradients (SM-G) operator dramatically increases the ability of a simple genetic algorithm-based neuroevolution method to find solutions in high-dimensional domains that require deep and/or recurrent neural networks (which tend to be particularly brittle to mutation), including domains that require processing raw pixels. By improving our ability to evolve deep neural networks, this new safer approach to mutation expands the scope of domains amenable to neuroevolution.
['Jay Chen', 'Joel Lehman', 'Kenneth O. Stanley', 'Jeff Clune']
2017-12-18
null
null
null
null
['artificial-life']
['miscellaneous']
[ 4.48908776e-01 1.26681656e-01 3.30245972e-01 4.42503244e-02 1.33397132e-01 -4.92924094e-01 3.06149215e-01 7.75934830e-02 -7.00699449e-01 9.07659948e-01 -4.48920220e-01 -3.41956228e-01 -1.74255416e-01 -1.09369206e+00 -6.92494214e-01 -9.38402414e-01 -1.89829573e-01 1.47251397e-01 4.73403573e-01 -7.14135528e-01 2.23798290e-01 6.64332092e-01 -1.96634996e+00 -1.86134413e-01 1.02851653e+00 7.62088895e-01 -4.97945696e-02 5.06889820e-01 -7.18623251e-02 2.48399839e-01 -1.04570425e+00 -1.34114370e-01 3.89276892e-01 -7.62717247e-01 -2.84250110e-01 -2.65482754e-01 1.77164257e-01 1.57910809e-01 9.11299288e-02 1.17265522e+00 7.41714478e-01 3.04865301e-01 4.24676090e-01 -9.64301586e-01 -6.74324036e-01 5.38890243e-01 -2.79209554e-01 2.75163651e-01 9.65567827e-02 3.71093571e-01 5.54046929e-01 -3.45309943e-01 6.31180644e-01 1.18843186e+00 1.04565942e+00 8.21283996e-01 -1.17783797e+00 -4.23941851e-01 1.27189606e-01 -1.99750140e-01 -1.17252302e+00 -7.04096779e-02 5.83824933e-01 -5.93283623e-02 1.17225814e+00 3.50329578e-01 1.17369914e+00 7.02957988e-01 4.44072038e-01 3.62775058e-01 7.02844560e-01 -5.63912570e-01 6.76540375e-01 -3.71087082e-02 -4.32381064e-01 8.24438512e-01 3.84894758e-01 2.76577175e-01 -1.79816797e-01 -1.23580046e-01 6.79732978e-01 -1.24109074e-01 -2.76690990e-01 -5.80132782e-01 -7.04070091e-01 8.79102886e-01 5.23898423e-01 5.35767198e-01 -3.74123752e-01 1.54265121e-01 3.01142156e-01 6.24794960e-01 1.11396357e-01 1.20682549e+00 -6.19625509e-01 -3.38632911e-01 -7.00500190e-01 3.90842110e-01 7.19154358e-01 8.22731256e-02 8.53502870e-01 5.62078476e-01 1.03840336e-01 8.78384471e-01 -2.32204184e-01 5.27011901e-02 1.01923048e+00 -7.92050600e-01 7.02927560e-02 9.72960770e-01 -2.74704814e-01 -1.14527678e+00 -4.71702665e-01 -7.54253328e-01 -6.59770012e-01 8.67612898e-01 4.43799466e-01 -5.04162073e-01 -9.55352485e-01 1.99364221e+00 4.28792149e-01 -1.31665513e-01 3.89809795e-02 5.44196546e-01 1.94254518e-01 3.88269663e-01 -2.65831769e-01 -1.59598351e-01 6.13153875e-01 -6.81232691e-01 -2.58616269e-01 -3.16769123e-01 6.71969414e-01 -1.05656423e-01 1.25332165e+00 3.46090764e-01 -1.26596057e+00 -3.09759825e-01 -1.36262310e+00 5.14304280e-01 -7.31037617e-01 -6.24682069e-01 5.53170502e-01 9.77686584e-01 -1.21240580e+00 9.14403141e-01 -6.92961156e-01 -3.73837739e-01 3.96378219e-01 6.25714362e-01 -3.10090967e-02 3.74372870e-01 -1.36937630e+00 1.26659465e+00 6.48190081e-01 1.91929117e-01 -4.38958943e-01 -6.36447489e-01 -6.11783981e-01 1.88482419e-01 2.41668642e-01 -5.89885652e-01 8.30529213e-01 -1.60066879e+00 -1.81997919e+00 4.48172748e-01 2.04655126e-01 -6.31968319e-01 7.12556601e-01 2.78854728e-01 -2.38535002e-01 -2.09673822e-01 -5.14084637e-01 9.48426664e-01 1.05351353e+00 -9.21359241e-01 -4.22826678e-01 -2.91371822e-01 3.60261425e-02 2.21022919e-01 -8.52858782e-01 -3.01031083e-01 -4.08122018e-02 -8.13764215e-01 5.44492789e-02 -1.04553950e+00 -3.20539892e-01 1.99788705e-01 1.83473364e-01 2.70775799e-02 8.87299955e-01 -1.78640217e-01 1.23209167e+00 -2.17489815e+00 3.71841550e-01 2.85587937e-01 -3.61987688e-02 6.89242959e-01 -3.27344120e-01 2.57015079e-01 -1.00510977e-01 2.65377522e-01 -7.32906699e-01 2.64878839e-01 -1.75050884e-01 4.03955758e-01 1.95318647e-02 1.35068133e-01 4.30839330e-01 9.79153037e-01 -8.51638019e-01 4.95606400e-02 -2.17658252e-01 6.60463333e-01 -8.24726224e-01 -2.94719726e-01 -3.32590610e-01 4.18494195e-02 -1.85735315e-01 4.12090838e-01 2.93016016e-01 2.03112543e-01 6.79325089e-02 4.71609354e-01 -1.31312922e-01 -5.08046187e-02 -1.19025159e+00 1.32492197e+00 -2.44901985e-01 6.29088223e-01 -1.64339229e-01 -1.00161874e+00 1.18674433e+00 6.00494305e-03 2.77975500e-01 -7.60336399e-01 1.53312102e-01 2.01894552e-01 4.85656202e-01 -2.02599794e-01 2.39743352e-01 -6.67121485e-02 6.54818714e-02 5.15575290e-01 -2.20103562e-01 -4.24090177e-01 3.85697842e-01 -4.40526843e-01 1.30604017e+00 9.15118530e-02 2.51803715e-02 -8.04313049e-02 3.00987303e-01 -2.73541026e-02 8.98242891e-01 7.63515353e-01 -1.65786743e-01 3.84515882e-01 6.26785517e-01 -5.65011919e-01 -1.10901594e+00 -9.00306165e-01 -7.28535408e-04 9.86972034e-01 -2.63711989e-01 5.13317622e-02 -9.31693852e-01 -4.58654970e-01 1.25290811e-01 6.60561681e-01 -8.59385967e-01 -8.28777492e-01 -7.49766707e-01 -1.04305887e+00 8.81861210e-01 2.44664118e-01 5.18613636e-01 -1.25632739e+00 -1.20366955e+00 3.38533700e-01 6.81389868e-01 -1.52951464e-01 4.79073748e-02 6.73855424e-01 -1.23344493e+00 -7.23779142e-01 -7.35208929e-01 -7.43170798e-01 7.59724677e-01 -3.33680630e-01 8.35553646e-01 4.19277519e-01 -6.96188331e-01 -8.09757411e-02 -1.62922770e-01 -5.19944429e-01 -5.20340443e-01 8.67723152e-02 1.19869590e-01 -4.18935835e-01 1.60139799e-01 -8.96921575e-01 -2.37442940e-01 7.54791126e-02 -1.08198404e+00 -3.85984749e-01 4.90149051e-01 1.10752618e+00 3.02172840e-01 4.57729101e-01 7.34991252e-01 -5.01302958e-01 1.19290519e+00 -2.84027576e-01 -5.54802120e-01 1.78846121e-01 -8.78072739e-01 2.39957660e-01 7.71235585e-01 -9.14363325e-01 -6.88505173e-01 -2.22306728e-01 -1.46059506e-03 -2.69337147e-01 1.05062678e-01 3.98968339e-01 1.87758416e-01 -3.85391355e-01 9.77720141e-01 1.96076185e-01 3.60582739e-01 -2.17056096e-01 2.01942161e-01 1.01451822e-01 3.64954293e-01 -3.44563007e-01 6.26566350e-01 -4.37936466e-03 4.46702242e-02 -5.63048065e-01 -8.96347985e-02 4.70962048e-01 -4.92266208e-01 -2.00319991e-01 5.87458849e-01 -7.41122961e-02 -5.83450973e-01 5.11713445e-01 -6.89654589e-01 -4.68498886e-01 -7.49746799e-01 8.64610896e-02 -3.32590491e-01 4.81364913e-02 -3.83892596e-01 -6.20805919e-01 -2.50378042e-01 -9.96042728e-01 1.94639355e-01 5.54154873e-01 -3.49936396e-01 -1.00487769e+00 5.18537611e-02 -4.46641028e-01 7.49883235e-01 5.14599979e-01 1.31247008e+00 -4.18775707e-01 5.86464480e-02 -3.90831977e-01 3.96802068e-01 5.45672119e-01 2.28919566e-01 3.24015051e-01 -5.53707361e-01 -3.67660999e-01 -3.84341665e-02 -2.88495392e-01 7.30733573e-01 2.54112929e-01 9.57456172e-01 -1.71321824e-01 -6.40915334e-02 5.96287370e-01 1.20621705e+00 7.91068196e-01 7.40036130e-01 9.12213206e-01 3.17023724e-01 4.43812162e-01 4.44269031e-02 3.04826826e-01 -2.52505779e-01 5.04307032e-01 4.62574720e-01 -1.53955847e-01 1.48410331e-02 3.12879384e-02 3.35340768e-01 4.64166611e-01 -3.15525644e-02 -2.50449091e-01 -9.46792245e-01 2.58963466e-01 -1.64242232e+00 -1.00130665e+00 4.40613598e-01 2.17661619e+00 9.83927488e-01 4.85252768e-01 8.61671865e-02 4.62555677e-01 7.54087627e-01 -1.17878221e-01 -1.11116266e+00 -1.15216649e+00 -4.02878731e-01 4.58400846e-01 2.68224865e-01 2.26280823e-01 -5.77870786e-01 7.84937382e-01 7.06078339e+00 4.85309124e-01 -1.46164250e+00 -3.44736785e-01 4.42171752e-01 -4.53634411e-01 -3.50310862e-01 -2.20850915e-01 -4.65550244e-01 5.94007671e-01 8.12067151e-01 -5.81450462e-02 7.23668635e-01 7.48334229e-01 1.50127774e-02 -1.96087852e-01 -7.89138079e-01 5.40405989e-01 -7.48675540e-02 -1.28998983e+00 6.97127655e-02 -9.37645063e-02 9.68325078e-01 -2.09112570e-01 3.53029281e-01 2.26604640e-01 3.62555325e-01 -1.17396450e+00 5.79382122e-01 4.05557603e-01 3.16633433e-01 -1.16310263e+00 5.33752739e-01 3.53192836e-01 -4.77993339e-01 -5.36461949e-01 -4.21286613e-01 -2.29095012e-01 -1.16647966e-01 4.95672852e-01 -7.14162886e-01 -1.68273985e-01 8.13292861e-01 9.82618630e-02 -7.28372037e-01 1.13463581e+00 -1.60521716e-01 2.89755970e-01 -3.78512949e-01 -5.17532289e-01 3.82637948e-01 -1.50266558e-01 6.78997457e-01 8.98924112e-01 5.13632059e-01 -2.82158464e-01 -4.26221102e-01 7.17884183e-01 2.15916913e-02 -1.41600147e-01 -6.28551543e-01 -1.26031071e-01 6.41836226e-01 7.35142112e-01 -7.80077875e-01 -5.57758249e-02 7.78001994e-02 9.98649418e-01 3.48476946e-01 3.41213465e-01 -6.29956126e-01 -7.63558447e-01 7.89514661e-01 -1.03616916e-01 4.89501536e-01 -2.12777317e-01 -4.52703983e-01 -5.59493959e-01 1.01178899e-01 -1.23682845e+00 1.41678497e-01 -4.89302486e-01 -8.97507071e-01 6.81462288e-01 -4.25136566e-01 -8.75392377e-01 -4.73734558e-01 -3.61153245e-01 -6.97046041e-01 8.38360310e-01 -1.05830240e+00 -3.71060491e-01 1.57378018e-02 3.47626597e-01 2.80438066e-01 -4.24419165e-01 8.89053166e-01 1.25068367e-01 -8.06590915e-01 7.17150629e-01 2.42835492e-01 -3.44042182e-01 3.36844802e-01 -9.87558246e-01 4.51719910e-01 7.15388954e-01 -1.84984952e-01 6.96361184e-01 9.10178423e-01 -5.31612158e-01 -1.36534762e+00 -7.13756442e-01 5.17197728e-01 3.41781694e-03 4.06515360e-01 -1.77939415e-01 -9.88887787e-01 1.59793496e-01 -6.39401451e-02 -1.25680149e-01 2.64929891e-01 -7.01966360e-02 -1.48424745e-01 -1.53683215e-01 -1.46761799e+00 9.55893457e-01 9.89537656e-01 -1.24548621e-01 -4.06296879e-01 -1.21090002e-01 6.08972132e-01 -2.73376822e-01 -6.45525336e-01 6.45897925e-01 5.77858388e-01 -1.15391886e+00 8.36366355e-01 -5.89032471e-01 1.91102684e-01 -3.99738610e-01 9.92267281e-02 -1.60898447e+00 -3.55154186e-01 -6.01678252e-01 -1.23317122e-01 9.54286575e-01 6.94213629e-01 -1.07949960e+00 9.03592527e-01 7.02283084e-01 -2.74112951e-02 -1.14334178e+00 -1.06471884e+00 -9.41573203e-01 3.26896340e-01 6.67957636e-03 9.11115468e-01 1.04228258e+00 -1.80255145e-01 -2.65741944e-01 7.55679086e-02 -2.86697030e-01 1.04078732e-01 -3.78763169e-01 3.87928486e-01 -1.28585303e+00 -5.74456990e-01 -1.09845674e+00 -7.67182469e-01 -3.26675057e-01 -6.56708479e-02 -5.43110013e-01 2.66654864e-02 -1.16882527e+00 -4.12863016e-01 -5.84403813e-01 -2.77387470e-01 6.13111019e-01 -1.55879557e-01 2.31934443e-01 2.04111964e-01 -1.56547412e-01 3.03464513e-02 5.38931727e-01 1.07743728e+00 -1.31416246e-01 -6.14810348e-01 -1.26118347e-01 -6.28586531e-01 7.23786235e-01 9.63369071e-01 -5.84727883e-01 -4.57085937e-01 -5.40867567e-01 5.26824832e-01 -5.20157993e-01 3.71205620e-02 -1.31920791e+00 1.15268908e-01 -1.11361928e-01 6.20209932e-01 2.47106314e-01 1.98991820e-01 -4.89650220e-01 2.53355891e-01 1.03792012e+00 -4.23567742e-01 5.46541572e-01 4.91426766e-01 2.94564575e-01 -1.68183029e-01 -5.88704765e-01 9.95266795e-01 -3.23477209e-01 -6.48919463e-01 -1.69490859e-01 -6.63810134e-01 3.15845385e-02 1.19097519e+00 -8.46738696e-01 -1.07091576e-01 -5.69240861e-02 -6.13592863e-01 -4.91243377e-02 8.37646246e-01 4.81170654e-01 4.70466703e-01 -1.02363074e+00 -3.46805841e-01 3.87161583e-01 -3.99768710e-01 -8.28313231e-02 -1.77542746e-01 4.11450028e-01 -6.76635027e-01 -4.38077971e-02 -5.87258995e-01 -3.56170088e-01 -1.07534254e+00 4.00410324e-01 8.33854675e-01 1.32264942e-01 -6.09584093e-01 1.33064330e+00 -3.81112278e-01 -5.51105082e-01 4.35298979e-01 -7.64398649e-02 -1.19577900e-01 1.72027871e-01 4.56419528e-01 4.18806821e-01 3.84377420e-01 3.45054902e-02 -3.33960295e-01 4.93896365e-01 1.80684328e-02 -1.72671691e-01 1.55632234e+00 3.95586401e-01 -2.52568603e-01 3.79691720e-01 9.16911066e-01 -3.86585534e-01 -1.50504482e+00 3.57316911e-01 1.36234146e-02 -2.30777338e-01 1.62192002e-01 -9.96219218e-01 -1.11512685e+00 6.56514645e-01 8.27650189e-01 3.58404517e-01 1.48554981e+00 -6.53666317e-01 6.75722063e-01 7.03478038e-01 2.74809599e-01 -1.32185018e+00 2.02145740e-01 7.07112134e-01 6.74781561e-01 -6.06606245e-01 -2.20261380e-01 4.04842973e-01 -3.89992923e-01 1.22785616e+00 7.70347357e-01 -1.10857926e-01 3.36257100e-01 5.50452054e-01 1.57004103e-01 4.06924598e-02 -7.64290452e-01 8.19477625e-03 -1.13299794e-01 7.39452004e-01 2.58344978e-01 -3.16092640e-01 -5.38005769e-01 -2.45844916e-01 -3.01640600e-01 -7.26875514e-02 3.63204747e-01 1.32769942e+00 -7.44748890e-01 -1.21150303e+00 -3.20773512e-01 3.61428559e-01 -2.01318473e-01 -4.65749539e-02 -5.30790806e-01 6.00109994e-01 5.15419602e-01 5.83315253e-01 2.23432228e-01 -2.73803145e-01 3.30849260e-01 1.12510361e-01 5.59638023e-01 -3.05103511e-01 -1.03778672e+00 -4.34622109e-01 -1.92104623e-01 -3.56200248e-01 2.72874609e-02 -6.57938659e-01 -1.49435103e+00 -3.47230583e-01 -2.35033557e-01 1.50029525e-01 8.33716273e-01 7.19973981e-01 3.53606164e-01 6.81029081e-01 4.67562348e-01 -8.43731046e-01 -5.72493732e-01 -4.75848496e-01 -4.08937126e-01 1.54436380e-01 2.81117290e-01 -6.41319633e-01 -3.42700094e-01 -3.00701618e-01]
[8.243386268615723, 3.2333173751831055]
4e8220a7-7372-4332-b48b-9ce75330f80e
transductive-linear-probing-a-novel-framework
2212.05606
null
https://arxiv.org/abs/2212.05606v1
https://arxiv.org/pdf/2212.05606v1.pdf
Transductive Linear Probing: A Novel Framework for Few-Shot Node Classification
Few-shot node classification is tasked to provide accurate predictions for nodes from novel classes with only few representative labeled nodes. This problem has drawn tremendous attention for its projection to prevailing real-world applications, such as product categorization for newly added commodity categories on an E-commerce platform with scarce records or diagnoses for rare diseases on a patient similarity graph. To tackle such challenging label scarcity issues in the non-Euclidean graph domain, meta-learning has become a successful and predominant paradigm. More recently, inspired by the development of graph self-supervised learning, transferring pretrained node embeddings for few-shot node classification could be a promising alternative to meta-learning but remains unexposed. In this work, we empirically demonstrate the potential of an alternative framework, \textit{Transductive Linear Probing}, that transfers pretrained node embeddings, which are learned from graph contrastive learning methods. We further extend the setting of few-shot node classification from standard fully supervised to a more realistic self-supervised setting, where meta-learning methods cannot be easily deployed due to the shortage of supervision from training classes. Surprisingly, even without any ground-truth labels, transductive linear probing with self-supervised graph contrastive pretraining can outperform the state-of-the-art fully supervised meta-learning based methods under the same protocol. We hope this work can shed new light on few-shot node classification problems and foster future research on learning from scarcely labeled instances on graphs.
['Huan Liu', 'Jundong Li', 'Kaize Ding', 'Song Wang', 'Zhen Tan']
2022-12-11
null
null
null
null
['product-categorization']
['miscellaneous']
[ 5.31495214e-01 7.98958898e-01 -8.83624375e-01 -3.59080255e-01 -5.35045266e-01 -9.52423066e-02 5.19132853e-01 5.29506207e-01 -6.40333369e-02 4.47199196e-01 1.15635851e-02 -2.90446132e-01 -1.17195353e-01 -1.24432397e+00 -4.64148790e-01 -8.69974196e-01 -2.11475492e-01 5.42155862e-01 9.25261676e-02 -4.89249438e-01 -3.63475293e-01 1.97407097e-01 -1.30583191e+00 8.95760357e-02 6.65084362e-01 8.23415518e-01 -6.61671236e-02 4.04452056e-01 -3.05074573e-01 6.83278978e-01 -2.86326744e-02 -7.12651849e-01 1.09388076e-01 -4.82994467e-01 -8.65804017e-01 2.19640285e-01 3.09743971e-01 2.63723508e-02 -6.47894263e-01 1.30552447e+00 4.47631210e-01 1.46118328e-01 7.32145369e-01 -1.61097240e+00 -8.44182730e-01 7.01463401e-01 -5.40664256e-01 2.29160786e-01 5.59973493e-02 3.43262614e-03 1.27769625e+00 -6.49110436e-01 9.81689036e-01 9.52713788e-01 9.05974925e-01 8.71439278e-01 -1.29806674e+00 -4.68151867e-01 8.13956037e-02 3.19928765e-01 -1.14886940e+00 -2.11903259e-01 1.08884323e+00 -1.86706141e-01 6.31962001e-01 1.43957749e-01 5.89775860e-01 1.58511078e+00 8.60228166e-02 5.96234143e-01 7.80045748e-01 -4.31422293e-01 4.64365721e-01 4.60185230e-01 3.83363634e-01 1.14575970e+00 3.37714642e-01 3.97007763e-02 -2.73141176e-01 -3.00845176e-01 3.87225389e-01 6.66415155e-01 -1.33276775e-01 -8.31606328e-01 -1.06189048e+00 1.17599869e+00 8.07564259e-01 5.59394538e-01 -3.12681407e-01 3.98724712e-02 7.10880280e-01 4.78353947e-01 1.17001975e+00 4.82211053e-01 -4.72040772e-01 1.76672682e-01 -7.01112270e-01 -3.80598903e-01 9.21282232e-01 1.16441774e+00 1.03444946e+00 4.54337075e-02 -4.07577306e-02 7.10031748e-01 1.09390296e-01 6.34377152e-02 5.33884823e-01 -4.72113132e-01 2.01548219e-01 9.24605846e-01 -5.05030632e-01 -9.50639963e-01 -6.36624873e-01 -8.22495461e-01 -1.12146354e+00 -1.88548222e-01 1.46276325e-01 -1.26819462e-01 -1.07048786e+00 1.55811977e+00 6.09437823e-01 7.83450663e-01 -1.80420168e-02 5.94733775e-01 1.14411664e+00 4.18763906e-01 1.17864810e-01 -4.43449795e-01 1.18263495e+00 -1.20934391e+00 -5.41127264e-01 -1.14350088e-01 1.33526516e+00 1.01105375e-02 1.07914889e+00 -1.38903394e-01 -4.10855114e-01 -3.68704706e-01 -9.81599987e-01 1.61507621e-01 -7.18565941e-01 -6.18095279e-01 1.15992951e+00 7.23019958e-01 -9.39416945e-01 8.15686524e-01 -8.15172076e-01 -8.97038221e-01 8.42455208e-01 1.77407652e-01 -5.00183403e-01 -5.93551517e-01 -1.15892422e+00 5.44741929e-01 2.77385890e-01 -1.20296352e-01 -7.71140695e-01 -9.40098643e-01 -1.12576497e+00 1.49073377e-01 9.00678813e-01 -7.72504926e-01 7.37342715e-01 -7.92601228e-01 -1.20043635e+00 1.08108819e+00 3.41093570e-01 -3.09769720e-01 1.65512457e-01 7.36050725e-01 -5.08701742e-01 6.33830205e-02 2.25059614e-01 4.71622378e-01 1.02825785e+00 -1.12374783e+00 -2.69880921e-01 -6.68215156e-01 3.03614177e-02 -3.84955630e-02 -9.04504836e-01 -7.59476185e-01 -1.47230729e-01 -5.16883314e-01 3.70968506e-02 -9.86350238e-01 -4.88004327e-01 9.22882035e-02 -6.22652233e-01 -4.58818644e-01 8.43117952e-01 4.80521992e-02 8.72847140e-01 -2.04520535e+00 2.45467335e-01 4.85890843e-02 6.96774781e-01 2.19543815e-01 -4.95945185e-01 6.03700399e-01 -1.65561721e-01 5.82599305e-02 -2.45515525e-01 -4.09504205e-01 -1.04541242e-01 3.68541121e-01 -3.72011736e-02 5.76523483e-01 5.88327944e-02 1.40632701e+00 -1.46433461e+00 -6.60417020e-01 1.78513497e-01 3.13195020e-01 -3.44002932e-01 1.22448564e-01 -1.87714860e-01 1.12589933e-01 -4.60177809e-01 1.01454008e+00 4.68172640e-01 -7.97211647e-01 2.46020943e-01 -2.37559021e-01 6.24013841e-01 -3.93885940e-01 -6.01700485e-01 1.98685491e+00 -5.03780246e-01 3.06843579e-01 -1.62939861e-01 -1.72124815e+00 7.67100632e-01 2.25066930e-01 7.02055514e-01 -4.31363136e-01 3.02486718e-01 -9.52033792e-03 -1.37017488e-01 -5.80810368e-01 2.43483260e-01 -5.03129721e-01 -1.08849294e-01 3.17646235e-01 6.24806464e-01 -8.45836103e-02 -1.08312210e-02 4.68015343e-01 1.59687626e+00 -2.86489964e-01 4.98486638e-01 -7.89085627e-02 -4.20428589e-02 2.45041102e-01 4.34494108e-01 7.37988472e-01 -4.35849875e-01 4.51555520e-01 4.30604100e-01 -5.05050004e-01 -6.79820716e-01 -9.71711338e-01 -1.33603290e-01 1.47432292e+00 1.53531641e-01 -5.47550857e-01 -4.18394148e-01 -1.36439884e+00 1.69684649e-01 4.39082146e-01 -1.11912096e+00 -6.69426680e-01 -1.52732179e-01 -1.01291192e+00 2.23857060e-01 4.28987294e-01 8.23984817e-02 -9.16018188e-01 4.01510671e-02 2.03292504e-01 3.54387313e-01 -1.12594020e+00 -7.92589858e-02 4.89664465e-01 -1.10881281e+00 -1.29316258e+00 -8.98309112e-01 -1.13686347e+00 8.65007401e-01 6.19715810e-01 1.11921597e+00 2.93389171e-01 -6.46299124e-01 6.81382120e-01 -6.49414480e-01 -1.54878259e-01 -4.42892909e-01 4.92916912e-01 4.26851548e-02 6.09638281e-02 4.91846114e-01 -7.71020830e-01 -4.72951353e-01 1.08106032e-01 -7.83889711e-01 4.08320362e-03 5.03973365e-01 1.15717268e+00 5.15293539e-01 6.45953566e-02 8.72987509e-01 -1.80206728e+00 4.15202409e-01 -1.18736279e+00 -4.85255271e-02 3.07169855e-01 -9.23870146e-01 -1.24197528e-01 8.71487677e-01 -5.78585565e-01 -6.50714517e-01 -1.25789195e-01 2.82366145e-02 -6.15413666e-01 2.50066966e-02 7.56841540e-01 1.29825681e-01 -2.02326849e-01 9.31960881e-01 -5.05926162e-02 8.65170360e-02 -2.54967749e-01 7.40164220e-01 6.47318065e-01 2.77857691e-01 -6.29696250e-02 9.71109271e-01 6.14393532e-01 3.77003670e-01 -9.40395832e-01 -1.09464562e+00 -6.99721098e-01 -6.29179955e-01 -5.63342981e-02 6.96494102e-01 -7.39703715e-01 -3.34260911e-01 3.50322835e-02 -6.52455032e-01 -4.94081348e-01 -7.92037606e-01 2.28389964e-01 -4.14973587e-01 3.18510115e-01 -7.12059319e-01 -4.51516658e-01 -3.16973656e-01 -7.01780796e-01 1.02192366e+00 9.75759178e-02 1.62960529e-01 -1.56306815e+00 3.02751601e-01 2.36366689e-01 3.46618444e-01 3.85869592e-01 1.09193587e+00 -1.00272918e+00 -2.01798886e-01 -5.78752875e-01 -1.57501757e-01 6.87410012e-02 5.48746169e-01 -4.08631206e-01 -1.03076005e+00 -5.47661543e-01 -2.19601855e-01 -6.67959869e-01 1.09486735e+00 7.51677901e-02 9.74769711e-01 -1.60635427e-01 -8.66070449e-01 6.25298023e-01 1.39656627e+00 -4.56431746e-01 4.55319099e-02 -7.43365511e-02 1.10760236e+00 7.41894960e-01 4.53312337e-01 2.75356889e-01 4.32820916e-01 5.06126046e-01 5.93638718e-01 -2.40200803e-01 -3.44350308e-01 -3.76469046e-01 -2.14644343e-01 9.01617765e-01 2.68161863e-01 -3.24975193e-01 -9.16399240e-01 4.90526676e-01 -1.94234455e+00 -9.32573318e-01 1.93256646e-01 1.81225872e+00 5.54218233e-01 1.86945140e-01 1.01315282e-01 9.72314626e-02 9.27154005e-01 4.87604171e-01 -9.26670313e-01 -3.11662052e-02 2.76331067e-01 1.84131086e-01 3.25063884e-01 -5.17095299e-03 -1.14215040e+00 9.37587738e-01 5.28915215e+00 7.80400574e-01 -1.03652549e+00 6.63952351e-01 8.09156418e-01 2.78794646e-01 -4.53410149e-01 -1.10465638e-01 -6.19780242e-01 2.54951358e-01 1.05649412e+00 -3.62586528e-02 2.30726138e-01 1.28858125e+00 -3.79499286e-01 4.23308849e-01 -1.43629086e+00 1.11716986e+00 3.74690920e-01 -1.56337583e+00 -1.00669660e-01 2.66736031e-01 9.98685598e-01 2.47051358e-01 8.91276449e-02 9.31756377e-01 2.35406026e-01 -8.51898313e-01 -2.19114631e-01 1.42596468e-01 1.00814283e+00 -3.39632660e-01 7.59829998e-01 5.12165546e-01 -1.29158485e+00 -2.08184361e-01 -6.61213040e-01 3.42420451e-02 3.80643085e-02 7.42265880e-01 -1.08598101e+00 5.48551798e-01 3.86137128e-01 1.18267357e+00 -6.25076890e-01 6.96109653e-01 6.58654943e-02 7.00312018e-01 -3.89947779e-02 -1.73515603e-01 2.07132414e-01 5.08353338e-02 3.79830867e-01 8.47844124e-01 1.19976990e-01 7.47984946e-02 3.88965994e-01 6.48802459e-01 -4.97213393e-01 1.90440461e-01 -1.03720117e+00 -3.06897998e-01 3.51838052e-01 1.75105679e+00 -1.03449667e+00 -4.51173127e-01 -6.54674053e-01 1.12632668e+00 6.77884281e-01 2.58698583e-01 -4.68249112e-01 -3.21008861e-01 2.13711709e-01 3.78025055e-01 6.56436831e-02 2.83867329e-01 -3.29937378e-04 -1.36566699e+00 -4.15536225e-01 -3.55376273e-01 6.89619184e-01 -2.73575664e-01 -1.93002105e+00 6.93557560e-01 -3.60675454e-02 -1.35166764e+00 -3.33964318e-01 -5.95519006e-01 -8.17678928e-01 8.18009824e-02 -1.61459434e+00 -1.51847947e+00 -5.13867974e-01 4.93753672e-01 6.45046413e-01 -3.06901783e-01 1.14858270e+00 1.63323984e-01 -7.30119705e-01 7.89531052e-01 3.21077779e-02 1.70460120e-01 6.40493989e-01 -1.31412804e+00 3.60553235e-01 3.53242636e-01 4.19446409e-01 3.31712544e-01 5.41874349e-01 -7.08852232e-01 -1.65032256e+00 -1.50408792e+00 4.15353745e-01 -3.16214651e-01 8.84233356e-01 -6.71910584e-01 -9.68872070e-01 7.17269361e-01 -2.81064026e-02 1.05143237e+00 9.64250982e-01 4.30438012e-01 -3.81262660e-01 -2.09456891e-01 -1.31006992e+00 4.61104363e-01 1.44789886e+00 -6.47742510e-01 -2.32410237e-01 8.32725942e-01 1.01871753e+00 7.32406899e-02 -1.04316092e+00 5.44620514e-01 9.05463696e-02 -4.79663402e-01 8.49428952e-01 -9.82651711e-01 2.43610814e-01 3.92290682e-01 -5.14913350e-02 -1.64326358e+00 -4.18506712e-01 -5.55399477e-01 -5.00307262e-01 1.08114493e+00 4.60387766e-01 -6.99428678e-01 1.46842289e+00 4.02662992e-01 -2.88382798e-01 -1.04120660e+00 -9.14196372e-01 -7.86508620e-01 -4.91770729e-02 -2.07936853e-01 2.69103110e-01 1.51912344e+00 4.40496713e-01 6.83367193e-01 -3.80507171e-01 -1.13244191e-01 8.72519910e-01 1.63239628e-01 6.14566922e-01 -1.64244914e+00 -5.14215112e-01 -1.03822254e-01 -8.86743367e-01 -5.60440361e-01 6.15811825e-01 -1.54098308e+00 -1.42712191e-01 -1.57676423e+00 4.10174251e-01 -6.92116022e-01 -4.60591674e-01 6.33308470e-01 -1.98130429e-01 5.73810577e-01 -7.60586187e-02 1.17697995e-02 -8.90827477e-01 7.37970829e-01 1.21678352e+00 -6.73069119e-01 -1.52344063e-01 1.92629069e-01 -7.25460470e-01 5.03506601e-01 5.96471727e-01 -7.14505315e-01 -7.17199385e-01 1.94183946e-01 1.88364238e-01 1.70911491e-01 1.05281562e-01 -6.88124597e-01 3.40439320e-01 -3.50971892e-02 -6.68228418e-02 -9.62773561e-02 4.74725544e-01 -9.01425004e-01 -2.13196844e-01 5.51628828e-01 -2.17864543e-01 -3.44438612e-01 -4.06431526e-01 1.16748750e+00 1.23363718e-01 -3.24633688e-01 4.73111778e-01 -3.55984479e-01 -8.16311598e-01 9.27588642e-01 1.30810380e-01 2.64624834e-01 1.30900180e+00 -2.77031839e-01 -5.23704112e-01 -1.35013476e-01 -1.04264283e+00 1.34615391e-01 3.85448247e-01 4.55831766e-01 6.63452089e-01 -1.38673258e+00 -5.27224481e-01 1.09360307e-01 7.01366544e-01 -2.24913597e-01 3.81329387e-01 7.88629472e-01 8.48406404e-02 7.99981132e-03 7.09959567e-02 -6.02727354e-01 -8.26264858e-01 1.12504673e+00 1.15495780e-02 -3.91615301e-01 -7.87082255e-01 8.25368345e-01 4.00462523e-02 -7.51392663e-01 3.29072714e-01 -6.08995929e-02 -7.47096911e-02 3.12370777e-01 2.36153424e-01 4.67641026e-01 1.44444168e-01 -4.49475706e-01 -2.79257447e-01 2.54518539e-01 -2.57200658e-01 5.39358795e-01 1.59464681e+00 2.70004012e-02 2.01997638e-01 6.90777838e-01 1.64230490e+00 -5.14051974e-01 -8.75746489e-01 -5.85653663e-01 -6.31152019e-02 -2.02214345e-01 1.41765922e-01 -4.03467119e-01 -1.32883489e+00 8.53201568e-01 6.17556691e-01 5.29091299e-01 6.94260299e-01 4.33291554e-01 9.18634117e-01 5.51690221e-01 6.37525439e-01 -8.37552011e-01 4.90905136e-01 -8.32454190e-02 2.07323968e-01 -1.78757763e+00 -7.91929439e-02 -7.71508038e-01 -4.96817589e-01 1.01600182e+00 5.57384551e-01 -2.20394149e-01 1.19439960e+00 -6.51319474e-02 -2.22762778e-01 -6.20455563e-01 -7.58518159e-01 -3.67365539e-01 1.87847868e-01 8.71368945e-01 1.68177918e-01 1.80390581e-01 6.16427399e-02 5.25253415e-01 2.48949111e-01 -6.91180378e-02 4.95977432e-01 9.52302575e-01 -3.28491539e-01 -1.07357109e+00 3.50530714e-01 1.12264156e+00 -4.75731678e-02 -1.02891557e-01 -1.08336255e-01 6.96929097e-01 -9.96993110e-02 8.77214134e-01 -2.06176005e-02 -6.18840337e-01 2.07741693e-01 1.55628622e-02 4.52557474e-01 -1.38098121e+00 -3.20776224e-01 -3.76788557e-01 -7.48055801e-02 -2.95927137e-01 -4.01785523e-01 -1.78018212e-01 -9.83946502e-01 -2.56715685e-01 -6.19492531e-01 2.08863318e-01 4.05441642e-01 8.19349170e-01 4.04321343e-01 5.42131424e-01 8.99087191e-01 -1.00618291e+00 -8.09628427e-01 -9.30136681e-01 -8.74162793e-01 5.19291639e-01 1.56066760e-01 -7.55275130e-01 -4.99221742e-01 -5.03452539e-01]
[7.379175186157227, 6.155055046081543]
be3f84c9-ecb0-4b5e-bed0-d24cb972e20e
object-pose-estimation-from-monocular-image
1809.00553
null
http://arxiv.org/abs/1809.00553v1
http://arxiv.org/pdf/1809.00553v1.pdf
Object Pose Estimation from Monocular Image using Multi-View Keypoint Correspondence
Understanding the geometry and pose of objects in 2D images is a fundamental necessity for a wide range of real world applications. Driven by deep neural networks, recent methods have brought significant improvements to object pose estimation. However, they suffer due to scarcity of keypoint/pose-annotated real images and hence can not exploit the object's 3D structural information effectively. In this work, we propose a data-efficient method which utilizes the geometric regularity of intraclass objects for pose estimation. First, we learn pose-invariant local descriptors of object parts from simple 2D RGB images. These descriptors, along with keypoints obtained from renders of a fixed 3D template model are then used to generate keypoint correspondence maps for a given monocular real image. Finally, a pose estimation network predicts 3D pose of the object using these correspondence maps. This pipeline is further extended to a multi-view approach, which assimilates keypoint information from correspondence sets generated from multiple views of the 3D template model. Fusion of multi-view information significantly improves geometric comprehension of the system which in turn enhances the pose estimation performance. Furthermore, use of correspondence framework responsible for the learning of pose invariant keypoint descriptor also allows us to effectively alleviate the data-scarcity problem. This enables our method to achieve state-of-the-art performance on multiple real-image viewpoint estimation datasets, such as Pascal3D+ and ObjectNet3D. To encourage reproducible research, we have released the codes for our proposed approach.
['Rahul M. V.', 'Jogendra Nath Kundu', 'R. Venkatesh Babu', 'Aditya Ganeshan']
2018-09-03
null
null
null
null
['viewpoint-estimation']
['computer-vision']
[-1.03901282e-01 -1.83191523e-01 -1.49195280e-03 -3.68113726e-01 -8.04959655e-01 -6.73062027e-01 5.65380514e-01 -1.06542021e-01 -2.88190186e-01 1.46999255e-01 -8.07671472e-02 2.90948331e-01 -1.03299804e-01 -7.24534273e-01 -9.28287327e-01 -5.16279638e-01 3.23641092e-01 8.15990150e-01 4.67521518e-01 -8.99175033e-02 4.78184074e-01 1.17443776e+00 -1.78116190e+00 1.79765709e-02 2.73660779e-01 1.35595155e+00 5.22399306e-01 3.67326021e-01 1.22517149e-03 2.33839050e-01 -3.38185757e-01 -2.38080710e-01 5.33014655e-01 1.59659326e-01 -5.14961720e-01 2.69348145e-01 9.71810043e-01 -6.34788692e-01 -3.28914732e-01 8.20706069e-01 5.21916866e-01 -6.82635829e-02 4.99121755e-01 -1.27681112e+00 -1.45683289e-01 -1.14293359e-01 -5.57348788e-01 -2.35727102e-01 6.09146953e-01 -1.48263352e-03 9.09499228e-01 -1.15583372e+00 8.17018926e-01 1.28121340e+00 5.23259103e-01 2.62959003e-01 -9.19418216e-01 -6.36574626e-01 6.40563760e-03 2.37938732e-01 -1.56512737e+00 -2.06220821e-01 1.16747606e+00 -2.61936754e-01 9.49403286e-01 1.92034140e-01 9.41314697e-01 8.51356745e-01 1.30670413e-01 7.83842683e-01 9.99902904e-01 -2.42787600e-01 -6.31333292e-02 -3.45279686e-02 -2.83724606e-01 7.81846225e-01 4.42414172e-02 2.61891559e-02 -5.58757961e-01 6.26724958e-02 1.02318227e+00 3.95914644e-01 -1.47547916e-01 -1.29317617e+00 -1.36969805e+00 5.06418288e-01 7.13045061e-01 -1.29900172e-01 -4.39519614e-01 1.76207051e-01 3.84200439e-02 -1.82662383e-01 3.65291715e-01 3.54910344e-01 -6.25685811e-01 -8.34835693e-02 -4.30195600e-01 4.24705356e-01 5.24374068e-01 1.23562860e+00 1.01536930e+00 -3.48918438e-01 3.20121765e-01 6.10453248e-01 4.87246573e-01 8.06244850e-01 1.25248209e-01 -1.14786315e+00 4.96553540e-01 1.10454512e+00 1.64847821e-01 -1.18326139e+00 -6.44571185e-01 -3.82624090e-01 -4.62075144e-01 2.01228470e-01 2.30614170e-01 6.34484351e-01 -8.27248335e-01 1.21773183e+00 6.50085270e-01 -2.59525418e-01 -2.86259949e-01 1.00581288e+00 6.03306234e-01 3.37166250e-01 -4.87484813e-01 2.29898721e-01 1.42291355e+00 -7.13804126e-01 -1.94287166e-01 -2.62679875e-01 3.03603917e-01 -8.80332112e-01 7.03963637e-01 4.36433285e-01 -9.98214006e-01 -7.11593151e-01 -9.71609354e-01 -2.25503787e-01 -5.12318909e-01 2.94213414e-01 5.28248191e-01 2.36860737e-01 -7.07735956e-01 2.80810535e-01 -8.15245330e-01 -3.61214131e-01 5.13576150e-01 5.72086811e-01 -8.04162025e-01 -2.85978734e-01 -6.62237287e-01 1.18569112e+00 5.70688307e-01 1.79865509e-01 -8.97863328e-01 -6.90916359e-01 -1.05996275e+00 -2.47658283e-01 5.65709710e-01 -7.27074087e-01 1.15367794e+00 -3.50171119e-01 -1.25048304e+00 1.16232383e+00 6.51847497e-02 5.49093913e-03 6.54542923e-01 -5.14284372e-01 1.85549688e-02 3.31043065e-01 1.51082695e-01 9.29796338e-01 8.46317708e-01 -1.46146619e+00 -6.71904385e-01 -8.95635545e-01 1.57874927e-01 4.53322351e-01 -2.73871161e-02 -1.55027598e-01 -7.65421391e-01 -1.00622922e-01 7.31722176e-01 -1.14058352e+00 -2.06775926e-02 4.26849812e-01 -3.05253446e-01 -2.37081334e-01 1.09503090e+00 -4.29568559e-01 5.15335262e-01 -1.99844456e+00 2.64622420e-01 3.99937518e-02 1.18913285e-01 7.62314945e-02 -5.58391102e-02 3.81197602e-01 4.99111265e-02 -3.27846050e-01 2.19667315e-01 -3.56635302e-01 -8.14753678e-03 1.84801623e-01 -9.00639817e-02 7.00710356e-01 2.58978575e-01 9.19431686e-01 -7.14002073e-01 -3.72257710e-01 7.39981771e-01 6.74402416e-01 -5.06863117e-01 4.02847797e-01 -2.73262590e-01 3.64282221e-01 -5.29799819e-01 8.21094394e-01 9.77327883e-01 -6.17141835e-02 -2.92292297e-01 -7.53456533e-01 -8.96736011e-02 2.77832616e-02 -1.38369918e+00 2.12434030e+00 -5.53462267e-01 2.00308785e-01 -1.94034576e-01 -6.11858726e-01 9.76182878e-01 1.12845689e-01 6.18746877e-01 -4.19112235e-01 1.40091151e-01 2.70033538e-01 -3.87885898e-01 -3.17988813e-01 4.68949735e-01 1.98714688e-01 3.32679451e-02 2.51450390e-01 2.21189812e-01 -7.53369391e-01 -7.27863908e-02 -1.37830973e-01 6.41877174e-01 6.69664919e-01 4.07346606e-01 1.73522413e-01 6.02709770e-01 -8.82238075e-02 2.66235560e-01 2.89980590e-01 -2.07862863e-03 9.45347846e-01 1.16916656e-01 -7.58868694e-01 -1.30896294e+00 -1.23814750e+00 -2.65707582e-01 4.68032539e-01 4.42798734e-01 -4.03730124e-01 -3.93060505e-01 -6.68708265e-01 2.84468919e-01 2.32679054e-01 -5.43158233e-01 -7.59178549e-02 -7.23663628e-01 -2.77745664e-01 9.46107805e-02 6.64183795e-01 6.77185833e-01 -6.78287625e-01 -1.06209099e+00 -4.31894101e-02 -1.57637849e-01 -1.38370502e+00 -2.65407830e-01 9.66643021e-02 -8.75511706e-01 -1.22344530e+00 -6.04209065e-01 -5.00108063e-01 7.70865560e-01 6.56966448e-01 1.03655970e+00 -1.28436744e-01 -4.54720140e-01 6.64975882e-01 -2.57642984e-01 -4.07363355e-01 -1.53310061e-01 1.34668378e-02 2.13554204e-01 -1.80859827e-02 3.87315422e-01 -4.52876925e-01 -8.57590914e-01 6.48583412e-01 -7.40386784e-01 1.44330442e-01 8.07644069e-01 4.73465055e-01 8.38486850e-01 -2.43103132e-01 4.69182199e-03 -2.35583961e-01 -1.05859011e-01 -3.08565870e-02 -8.89203787e-01 1.59732461e-01 -1.93837568e-01 2.45361775e-03 2.60450304e-01 -3.45741272e-01 -8.36767375e-01 6.61939263e-01 9.91396010e-02 -7.61619806e-01 -3.35224837e-01 9.88050550e-02 -3.84668678e-01 -3.60082358e-01 4.08357143e-01 2.43572339e-01 1.44242823e-01 -4.86657500e-01 3.48028362e-01 6.27854586e-01 3.92563879e-01 -4.19968069e-01 1.01647854e+00 7.29873180e-01 3.69747818e-01 -6.78147793e-01 -8.92453253e-01 -6.94636583e-01 -1.26256061e+00 -3.41507673e-01 9.11936581e-01 -1.18576694e+00 -9.89220500e-01 5.39283574e-01 -1.37932456e+00 2.22881645e-01 3.96471284e-02 6.63475335e-01 -8.64529729e-01 2.40337834e-01 -1.06820114e-01 -5.51803768e-01 -1.57907218e-01 -1.29944706e+00 1.80507398e+00 7.57605061e-02 3.23480293e-02 -6.23214185e-01 -1.09758176e-01 6.40554130e-01 8.37337896e-02 2.89977163e-01 6.06025517e-01 -3.55613768e-01 -1.20909476e+00 -5.23538649e-01 -2.22964108e-01 2.60863394e-01 8.49299133e-02 -8.15621316e-02 -1.07722604e+00 -2.08308488e-01 -2.23961566e-03 -4.30387378e-01 3.05282027e-01 2.30357140e-01 1.13285470e+00 3.02938312e-01 -3.84305388e-01 7.66504228e-01 1.48868525e+00 -3.50507535e-02 3.34615856e-01 4.49007303e-01 1.01480854e+00 5.53903162e-01 9.44176495e-01 4.87306803e-01 5.24576068e-01 1.07266784e+00 1.04430890e+00 2.60315444e-02 -5.69503987e-03 -4.47725117e-01 9.68447477e-02 6.75689399e-01 -1.70923680e-01 2.22511902e-01 -1.06139493e+00 2.39624962e-01 -1.46372569e+00 -5.57217538e-01 -4.28400040e-02 2.22549248e+00 3.95508975e-01 7.99023733e-02 -3.66447144e-03 -7.11950706e-03 4.38392788e-01 1.86300755e-03 -6.14711225e-01 1.98702186e-01 3.09419632e-02 -1.17317274e-01 5.12310445e-01 1.59639865e-01 -1.11673224e+00 9.23860848e-01 5.11945391e+00 5.50776720e-01 -1.09056556e+00 -2.26721302e-01 1.14770271e-01 1.67748943e-01 -3.48969214e-02 1.76921114e-03 -1.02859557e+00 -2.80667990e-02 4.02826697e-01 1.40997842e-01 1.89231113e-01 1.14288390e+00 -1.65530100e-01 -2.82752246e-01 -1.43278813e+00 1.32918477e+00 4.60256457e-01 -1.16089857e+00 1.27861366e-01 2.85293370e-01 5.87791443e-01 1.53631449e-01 -1.40134040e-02 -3.14843059e-02 -2.27449074e-01 -6.49070442e-01 8.48198473e-01 4.59285557e-01 6.32036209e-01 -8.50099564e-01 7.20516205e-01 6.07582033e-01 -1.20982385e+00 -1.32992923e-01 -5.03763855e-01 4.04723287e-02 1.08708171e-02 2.37538218e-01 -1.06741107e+00 7.99916863e-01 7.51707494e-01 8.23301733e-01 -9.13181126e-01 9.90191936e-01 -1.49490237e-01 -3.21264356e-01 -4.47426289e-01 1.76780403e-01 6.21738732e-02 -1.51843846e-01 4.18400794e-01 5.33575535e-01 3.02086741e-01 -1.02959843e-02 1.07971750e-01 8.16192329e-01 3.73535678e-02 -1.03918806e-01 -8.27883720e-01 2.77885050e-01 4.46441442e-01 1.49515414e+00 -8.53838563e-01 -3.07762008e-02 -3.36831510e-01 9.00040984e-01 2.70926684e-01 -5.35288565e-02 -7.05383539e-01 -9.87143144e-02 5.14185846e-01 1.00776121e-01 4.47836876e-01 -5.80094457e-01 1.24203868e-01 -1.20586574e+00 3.01481754e-01 -7.51642585e-01 -6.30345047e-02 -1.33662558e+00 -9.07482028e-01 5.59428036e-01 4.41114217e-01 -1.63164127e+00 -2.79665828e-01 -9.48294520e-01 -5.54073900e-02 8.05543125e-01 -1.37549353e+00 -1.63575304e+00 -7.31860816e-01 4.65888113e-01 5.82025766e-01 -2.13588160e-02 8.25588286e-01 1.45637598e-02 5.75922541e-02 1.87483534e-01 -2.44526416e-01 3.25486995e-02 5.78885257e-01 -1.09877682e+00 3.76140296e-01 4.48730260e-01 4.12996411e-01 5.63351393e-01 3.35449308e-01 -4.15569305e-01 -1.89151180e+00 -8.78445268e-01 4.26796108e-01 -1.09843802e+00 3.09873015e-01 -5.60977936e-01 -5.58772087e-01 5.76573014e-01 -3.18105817e-01 3.28322858e-01 2.21626639e-01 -1.96058869e-01 -3.93957913e-01 -4.07862008e-01 -1.00992143e+00 4.44821805e-01 1.10818982e+00 -6.67557597e-01 -5.72155714e-01 3.77913922e-01 5.74734926e-01 -8.87651384e-01 -1.06828690e+00 5.38662612e-01 8.02385271e-01 -1.01239610e+00 1.35206509e+00 -2.74205238e-01 2.74903923e-01 -4.67696726e-01 -4.04015541e-01 -1.17070651e+00 -2.30085663e-02 -1.53890565e-01 -1.27848491e-01 9.30193365e-01 -1.88701093e-01 -3.72023731e-01 9.14074719e-01 4.37777042e-01 -7.20277950e-02 -7.57516861e-01 -9.64506209e-01 -6.57981336e-01 -3.60582262e-01 -4.10385221e-01 6.34101808e-01 4.98886019e-01 -6.20622218e-01 9.44513679e-02 -5.06939106e-02 4.48551416e-01 7.74181485e-01 5.24767041e-01 1.25710988e+00 -1.39930880e+00 1.07517824e-01 -1.05540253e-01 -1.06301010e+00 -1.23348355e+00 2.64638990e-01 -7.55983472e-01 -8.18775445e-02 -1.23974895e+00 2.05287814e-01 -3.93820792e-01 3.45489904e-02 2.89049685e-01 1.16209991e-01 5.80857754e-01 3.99146885e-01 2.47527614e-01 -5.50706863e-01 5.91523826e-01 1.47920763e+00 1.25036374e-01 7.44777173e-02 5.82302846e-02 -2.17127696e-01 9.07860041e-01 5.96166193e-01 -3.94969910e-01 -2.57032454e-01 -5.16273677e-01 3.29782218e-01 1.02836400e-01 7.72777557e-01 -1.11200500e+00 1.57769009e-01 -1.41856149e-01 8.21941853e-01 -1.25620329e+00 8.45970273e-01 -1.34472966e+00 2.23520562e-01 3.92021179e-01 -2.96218470e-02 2.79797792e-01 2.91212481e-02 5.15180051e-01 -1.35523230e-01 -4.42100130e-02 5.96140802e-01 -4.22595650e-01 -7.90400326e-01 6.18786216e-01 3.72900307e-01 -2.42983580e-01 1.17349994e+00 -5.56990087e-01 -1.16316481e-02 -2.43991166e-01 -3.90182853e-01 3.53173986e-02 8.01979125e-01 6.43691897e-01 9.04318929e-01 -1.43724716e+00 -4.75322753e-01 6.24182105e-01 4.67727602e-01 5.64448595e-01 1.54834598e-01 7.75676847e-01 -8.31249714e-01 5.93988121e-01 -4.45372462e-01 -1.28001833e+00 -1.30387974e+00 5.66939831e-01 3.70355129e-01 2.72575796e-01 -5.32445252e-01 6.78517938e-01 3.01076591e-01 -7.89347351e-01 1.39017373e-01 -5.57787180e-01 1.45006016e-01 2.20054784e-03 3.11329126e-01 3.30602020e-01 3.09893399e-01 -1.01364672e+00 -5.16460776e-01 1.27333486e+00 -1.42920405e-01 6.94367066e-02 1.50690913e+00 -1.73995510e-01 -9.51163918e-02 3.64944160e-01 1.55758870e+00 -2.79365499e-02 -1.60067177e+00 -3.28665406e-01 -2.35080749e-01 -8.28581274e-01 -2.28905454e-02 -5.31912506e-01 -8.87917399e-01 1.11886692e+00 7.35715985e-01 -3.05303931e-01 9.37814295e-01 1.95376068e-01 4.93406534e-01 6.16801679e-01 8.00106227e-01 -9.16441441e-01 3.15258056e-01 4.33249056e-01 1.06172597e+00 -1.51078832e+00 3.10275435e-01 -4.12616849e-01 -3.58313650e-01 1.36772203e+00 7.59151101e-01 -1.09900013e-01 3.88722539e-01 -8.53690058e-02 6.97363913e-02 -2.79116184e-01 -3.76451015e-01 -8.95874202e-02 6.13456190e-01 7.08566964e-01 9.92299523e-03 -2.14993462e-01 3.91805828e-01 -3.73482145e-02 -2.17089772e-01 -2.27510571e-01 2.33230591e-01 8.82960558e-01 -2.99854249e-01 -1.01492143e+00 -5.86502016e-01 2.00846270e-02 -1.20457277e-01 2.72516012e-01 -3.90103638e-01 1.04300690e+00 2.08369792e-01 3.52695584e-01 1.36624947e-01 -2.97609001e-01 5.27870536e-01 -7.34632984e-02 9.50798571e-01 -6.28048956e-01 -2.42242143e-01 9.93925631e-02 -3.55520934e-01 -8.86366665e-01 -5.77293575e-01 -6.10895813e-01 -9.40297723e-01 6.47492185e-02 -4.19921428e-01 -3.23064983e-01 1.07087815e+00 8.34948659e-01 3.74523222e-01 1.52084947e-01 6.03844047e-01 -1.51448941e+00 -7.66112089e-01 -6.34242654e-01 -4.05443460e-01 5.23258209e-01 2.77362406e-01 -1.01038682e+00 -1.61199376e-01 -7.08465949e-02]
[7.5038604736328125, -2.6286838054656982]
1af92796-592d-4aeb-b5fa-af38b7698ca4
a-cross-task-flexible-transition-model-for
null
null
https://aclanthology.org/W13-4904
https://aclanthology.org/W13-4904.pdf
A Cross-Task Flexible Transition Model for Arabic Tokenization, Affix Detection, Affix Labeling, POS Tagging, and Dependency Parsing
null
['Stephen Tratz']
2013-10-01
null
null
null
ws-2013-10
['transition-based-dependency-parsing']
['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]
[-7.444147109985352, 3.649477005004883]
3c51d313-6475-4fe6-bbb1-99f5e3c2dfa8
on-the-fundamental-limits-of-matrix
2109.05408
null
https://arxiv.org/abs/2109.05408v1
https://arxiv.org/pdf/2109.05408v1.pdf
On the Fundamental Limits of Matrix Completion: Leveraging Hierarchical Similarity Graphs
We study the matrix completion problem that leverages hierarchical similarity graphs as side information in the context of recommender systems. Under a hierarchical stochastic block model that well respects practically-relevant social graphs and a low-rank rating matrix model, we characterize the exact information-theoretic limit on the number of observed matrix entries (i.e., optimal sample complexity) by proving sharp upper and lower bounds on the sample complexity. In the achievability proof, we demonstrate that probability of error of the maximum likelihood estimator vanishes for sufficiently large number of users and items, if all sufficient conditions are satisfied. On the other hand, the converse (impossibility) proof is based on the genie-aided maximum likelihood estimator. Under each necessary condition, we present examples of a genie-aided estimator to prove that the probability of error does not vanish for sufficiently large number of users and items. One important consequence of this result is that exploiting the hierarchical structure of social graphs yields a substantial gain in sample complexity relative to the one that simply identifies different groups without resorting to the relational structure across them. More specifically, we analyze the optimal sample complexity and identify different regimes whose characteristics rely on quality metrics of side information of the hierarchical similarity graph. Finally, we present simulation results to corroborate our theoretical findings and show that the characterized information-theoretic limit can be asymptotically achieved.
['Changho Suh', 'Soheil Mohajer', 'Adel Elmahdy', 'Junhyung Ahn']
2021-09-12
null
null
null
null
['stochastic-block-model']
['graphs']
[ 2.18450814e-01 5.30832171e-01 -2.19470471e-01 7.29465187e-02 -4.17025208e-01 -9.15282607e-01 1.81335986e-01 2.54600972e-01 -1.68374062e-01 5.79654932e-01 1.38784826e-01 -5.06520510e-01 -8.34253728e-01 -8.92316461e-01 -7.49443233e-01 -8.57806981e-01 -6.53361499e-01 1.96196780e-01 -1.27531663e-01 -3.73666942e-01 5.23192175e-02 2.75855929e-01 -1.10089576e+00 -3.52509558e-01 7.43675351e-01 8.75714302e-01 6.85613304e-02 8.03663135e-01 4.36761677e-01 4.64762002e-01 -2.11026222e-01 -5.45807362e-01 7.36827970e-01 -6.19698346e-01 -3.96648526e-01 3.16822588e-01 2.35177837e-02 -3.10857683e-01 -5.38508952e-01 1.39204144e+00 3.60934317e-01 -1.48920730e-01 6.98712707e-01 -1.38840175e+00 -5.41978300e-01 1.06361485e+00 -6.62727833e-01 1.46694362e-01 4.76415396e-01 -6.41131341e-01 1.57266903e+00 -5.76470435e-01 5.48764169e-01 7.88114965e-01 5.08936644e-01 -7.88369216e-03 -1.31005502e+00 -5.66296816e-01 -9.16815698e-02 -5.45608364e-02 -1.71869481e+00 -3.14764738e-01 5.18975735e-01 -3.66067857e-01 6.65991455e-02 5.09039700e-01 5.49824238e-01 4.67679054e-01 7.18947276e-02 4.93755162e-01 7.59413421e-01 -6.01168275e-01 3.16430837e-01 3.10951978e-01 3.17849517e-01 7.85420299e-01 1.13736427e+00 5.46727739e-02 -4.78548050e-01 -4.36127722e-01 5.96280396e-01 7.01031089e-02 -5.97183108e-01 -8.69249284e-01 -9.79522049e-01 7.53612041e-01 2.28883564e-01 3.89897615e-01 -3.73492122e-01 -9.25235916e-04 -9.53558907e-02 7.98295677e-01 7.47618526e-02 2.32110083e-01 -1.30132496e-01 2.66219109e-01 -5.97285926e-01 -2.44294360e-01 1.33184052e+00 1.33048368e+00 6.45768762e-01 -3.68471563e-01 -3.30789834e-02 3.96330595e-01 2.74843663e-01 7.86991656e-01 -3.61465991e-01 -7.65481770e-01 5.87618887e-01 4.24432755e-01 4.11164224e-01 -1.09549999e+00 -3.47407848e-01 -1.05537498e+00 -1.05485678e+00 -3.86887163e-01 7.15266585e-01 -3.38406801e-01 4.33691889e-02 2.09374356e+00 2.17505679e-01 -2.72423685e-01 1.05374910e-01 8.23694050e-01 1.48466364e-01 3.62691015e-01 -7.74168074e-01 -6.40632868e-01 1.19549084e+00 -2.23789185e-01 -6.59403503e-01 1.06304020e-01 9.49469149e-01 -3.08480650e-01 5.88264942e-01 4.28342015e-01 -1.08119953e+00 1.06511109e-01 -1.28717756e+00 4.66485560e-01 2.47977540e-01 7.26204216e-02 4.54172283e-01 1.20062923e+00 -1.05150199e+00 5.75052738e-01 -3.96084726e-01 -4.77954298e-01 -9.98979583e-02 4.29412067e-01 -2.28937507e-01 -1.55842498e-01 -9.80110288e-01 2.34652355e-01 8.99862200e-02 8.15632641e-02 -5.58316469e-01 -5.41325331e-01 -4.26312327e-01 4.03654784e-01 6.20709896e-01 -6.50548816e-01 9.35374737e-01 -7.44538724e-01 -1.14128971e+00 3.06126446e-01 2.15651870e-01 -3.95323426e-01 3.92107785e-01 1.58993125e-01 -1.59372851e-01 4.10144091e-01 3.74507811e-03 -4.21756804e-01 4.62532729e-01 -1.06068051e+00 -5.44694006e-01 -5.82222402e-01 5.68918824e-01 1.54298678e-01 -7.36716866e-01 -3.65148336e-01 -4.02356774e-01 -1.59126088e-01 3.48644614e-01 -1.20314181e+00 -2.34347031e-01 -1.01132363e-01 -5.49876511e-01 1.36829481e-01 -3.92072424e-02 -2.73142546e-01 1.36280203e+00 -2.23107004e+00 2.15994477e-01 8.25019956e-01 5.32434285e-01 -3.73524189e-01 -1.18411623e-01 1.08169651e+00 1.50753081e-01 1.66322574e-01 1.32987499e-01 -1.08300811e-02 7.36223236e-02 -7.35559314e-02 -1.07973829e-01 8.80855858e-01 -7.04210818e-01 4.72046405e-01 -8.88545156e-01 4.55712080e-02 -2.18989730e-01 -1.71869118e-02 -8.78948689e-01 -1.43737867e-01 4.46001053e-01 1.01602025e-01 -6.12659633e-01 1.09916113e-01 8.00741017e-01 -7.88153887e-01 1.03149188e+00 -1.80951908e-01 1.89424217e-01 2.54128397e-01 -1.56084335e+00 1.15864635e+00 -5.54490268e-01 3.29526901e-01 6.24590218e-01 -9.48985517e-01 2.61858225e-01 3.40938509e-01 2.91289449e-01 -9.01447609e-02 2.53725946e-01 1.85260013e-01 1.62344992e-01 -1.33655295e-01 4.42085594e-01 6.48593307e-02 -2.62536019e-01 8.91818762e-01 -7.54976571e-02 6.42436385e-01 3.01067591e-01 1.03126955e+00 1.22894847e+00 -6.40988767e-01 6.63138688e-01 -6.14968479e-01 4.77089107e-01 -5.84326029e-01 3.05665702e-01 1.23718536e+00 2.30330780e-01 3.10421903e-02 9.02895451e-01 5.00685751e-01 -1.07072258e+00 -1.03461862e+00 6.25680492e-04 8.68471265e-01 4.04983759e-01 -7.50765324e-01 -6.51371419e-01 -3.78396153e-01 2.61334270e-01 3.28131437e-01 -8.20166647e-01 -1.29267812e-01 2.30145350e-01 -7.65505791e-01 1.34392679e-01 -1.28465686e-02 2.31823772e-01 1.45713001e-01 3.83803360e-02 -6.83719665e-02 -6.99697137e-02 -1.20367277e+00 -5.52791297e-01 -1.11200862e-01 -7.40685403e-01 -1.26362765e+00 -4.34804171e-01 -2.56821901e-01 9.38267529e-01 9.16898072e-01 5.74949324e-01 2.13590175e-01 4.45409149e-01 7.87072837e-01 -4.04792458e-01 1.36540234e-01 -4.77895051e-01 1.42039791e-01 2.49794453e-01 3.21463615e-01 -1.83940217e-01 -8.46705079e-01 -7.50328422e-01 4.03032690e-01 -7.67601788e-01 5.44549106e-03 6.77204370e-01 4.74664688e-01 2.82674730e-02 9.31038037e-02 6.79339886e-01 -8.30174267e-01 7.24426270e-01 -7.34490871e-01 -9.10225987e-01 4.10569400e-01 -6.44412696e-01 2.87580431e-01 7.21669257e-01 -7.05538839e-02 -5.56498826e-01 -1.00184336e-01 3.57540160e-01 2.91354626e-01 5.77684939e-01 6.79531038e-01 -3.66061658e-01 -2.30608985e-01 5.16422153e-01 1.65812463e-01 -1.76577712e-03 -4.40725029e-01 5.67265272e-01 8.03814232e-01 1.94672108e-01 -3.55985820e-01 1.05632472e+00 7.74485111e-01 5.68163574e-01 -9.85778689e-01 -8.51275504e-01 -5.27358890e-01 -4.37893301e-01 -2.66553730e-01 1.97628364e-01 -1.05994892e+00 -1.27476001e+00 -1.92611456e-01 -5.88503778e-01 1.31513402e-01 -1.29378796e-01 6.87894225e-01 -5.26587009e-01 8.63152444e-01 -6.28636599e-01 -1.39279127e+00 1.65762398e-02 -5.70415497e-01 3.82754654e-01 -2.00657621e-01 2.03823477e-01 -9.49393988e-01 -1.85660005e-01 1.05791472e-01 1.62370682e-01 -6.48612529e-02 6.22779489e-01 -7.11297214e-01 -9.05312598e-01 -4.76577312e-01 -2.82939672e-01 -1.04039304e-01 5.12000248e-02 -5.34298062e-01 -3.32247347e-01 -5.65749109e-01 -6.77841529e-02 2.63183147e-01 4.15080696e-01 1.62669405e-01 4.72029865e-01 -6.50354624e-01 -3.61843258e-01 2.62044460e-01 1.37454522e+00 -3.16509902e-01 1.41860723e-01 -3.46351564e-01 4.62245077e-01 4.55124527e-01 2.38963634e-01 8.88804853e-01 4.17768538e-01 7.34221578e-01 2.32900396e-01 4.53242868e-01 3.52039993e-01 -5.24357080e-01 2.84467667e-01 1.12309921e+00 2.91664451e-02 -5.13147831e-01 -2.59670258e-01 4.54083025e-01 -1.92570317e+00 -7.63736069e-01 -4.37267065e-01 2.93711710e+00 6.51626348e-01 -1.09160252e-01 4.38968569e-01 3.28473538e-01 8.21086764e-01 -2.83100009e-01 -3.62161934e-01 9.85244736e-02 -2.81325966e-01 -3.23195815e-01 1.14615083e+00 6.50415897e-01 -5.41580081e-01 1.68525457e-01 5.97417927e+00 7.33028352e-01 -1.82176635e-01 3.11838388e-01 1.12910211e-01 -3.06575537e-01 -5.81393182e-01 2.02103138e-01 -5.91881871e-01 3.66320103e-01 1.08974254e+00 -7.87514210e-01 7.65879691e-01 6.35253668e-01 1.09348528e-01 -2.83023953e-01 -1.19513190e+00 7.45588779e-01 -2.57336739e-02 -1.18650687e+00 -2.67301112e-01 8.41438830e-01 8.05471182e-01 -2.87011623e-01 -1.27427161e-01 -2.41840869e-01 6.55351996e-01 -1.74967676e-01 4.13485825e-01 2.02690080e-01 7.21846104e-01 -7.72999048e-01 4.23419774e-01 5.30596912e-01 -1.18850851e+00 -5.23787022e-01 -4.24738675e-01 -2.67378092e-01 2.44423047e-01 9.70408738e-01 -4.71242398e-01 9.39603329e-01 9.76612419e-02 4.55583215e-01 -2.83903360e-01 1.13035583e+00 -2.21200883e-01 6.49972379e-01 -6.82528973e-01 -1.25460967e-01 -1.82356328e-01 -6.30053878e-01 7.11677074e-01 7.82544434e-01 6.33876801e-01 3.44453245e-01 -1.25624657e-01 3.66959125e-01 -3.66560459e-01 4.21134800e-01 -6.62797332e-01 -8.91557487e-04 8.05401862e-01 1.19782758e+00 -7.45678425e-01 -2.38766953e-01 -5.11198223e-01 9.22775149e-01 2.86732167e-01 4.66574490e-01 -3.26974541e-01 -3.93474817e-01 5.69040418e-01 2.62176245e-01 5.98106802e-01 -2.25303531e-01 -2.09901109e-01 -1.22741890e+00 3.32719609e-02 -5.18833339e-01 4.46867883e-01 -2.68930197e-01 -1.10212946e+00 -1.56344935e-01 -1.08452380e-01 -1.14996195e+00 -7.86914527e-02 -1.80159211e-01 -2.62231112e-01 3.80500019e-01 -8.81394386e-01 -5.04164457e-01 1.45972639e-01 5.02882659e-01 -5.67049563e-01 9.21387374e-02 4.36878443e-01 5.32978594e-01 -5.39865434e-01 9.43459690e-01 8.75761926e-01 -6.23830408e-02 2.10587710e-01 -1.01859188e+00 -1.70820668e-01 1.07759476e+00 1.39565632e-01 8.20598662e-01 8.95227790e-01 -4.62080002e-01 -1.93321657e+00 -6.71449184e-01 8.95576060e-01 -2.01793462e-02 1.07907510e+00 -8.85265589e-01 -1.99547932e-01 6.10727727e-01 -4.40300517e-02 -1.28968582e-01 9.49226737e-01 5.06138325e-01 -4.40406114e-01 -1.89208642e-01 -1.04527438e+00 6.58860981e-01 1.48491597e+00 -6.04704738e-01 3.94812115e-02 6.43857479e-01 4.41632271e-01 1.01478748e-01 -1.01442719e+00 9.50682163e-02 7.41454661e-01 -1.03911471e+00 7.35599697e-01 -4.52797055e-01 9.02869478e-02 -3.53974104e-01 -5.99991500e-01 -9.77611661e-01 -5.32095492e-01 -1.09192181e+00 -2.62216300e-01 9.52446878e-01 5.24477482e-01 -7.62202621e-01 7.83918023e-01 4.54730093e-01 5.24117887e-01 -4.65827435e-01 -8.95769715e-01 -1.09762847e+00 -2.02580079e-01 -1.36984184e-01 2.61370480e-01 7.08115637e-01 6.98382199e-01 7.45244265e-01 -7.53219485e-01 4.69927311e-01 9.05921757e-01 3.84703636e-01 8.16173971e-01 -1.31016541e+00 -7.89829612e-01 -6.09740093e-02 -3.62187773e-01 -1.53781080e+00 -1.70676380e-01 -9.51100528e-01 -5.17600775e-01 -1.26665854e+00 5.85118592e-01 -5.14332950e-01 -2.13274866e-01 -2.18410701e-01 1.52822763e-01 2.69006550e-01 4.36813146e-01 2.68610269e-01 -9.76757407e-01 3.40529948e-01 9.41009879e-01 3.61418903e-01 -1.67290494e-01 4.98256445e-01 -1.28286636e+00 3.34911466e-01 6.49682522e-01 -4.71581787e-01 -6.25942826e-01 2.02594876e-01 1.09340560e+00 6.86240613e-01 9.16106701e-02 -5.63090503e-01 3.46418828e-01 2.25786179e-01 -3.15366983e-01 -4.00784761e-01 1.59381524e-01 -8.36734116e-01 3.16878766e-01 5.65872550e-01 -6.25438571e-01 -3.84109229e-01 -5.67476690e-01 1.23269558e+00 5.68949997e-01 -4.49790448e-01 3.80266696e-01 3.91262084e-01 2.95768470e-01 2.08541706e-01 -7.04620361e-01 -2.33103987e-02 1.06314111e+00 -1.62253305e-01 -1.17152184e-01 -1.21794212e+00 -8.02154064e-01 2.09114075e-01 4.14496750e-01 -1.11675918e-01 1.98513985e-01 -1.27504873e+00 -6.74997389e-01 2.27764361e-02 2.34238908e-01 -9.43961859e-01 2.93870181e-01 1.35028744e+00 2.61882782e-01 4.09453809e-01 2.57819980e-01 -3.24921936e-01 -1.29477906e+00 6.92542791e-01 1.06742121e-01 -1.94727629e-01 -2.57283330e-01 3.83086443e-01 4.20458913e-01 3.52648273e-02 1.17324874e-01 -6.00204617e-02 1.61072224e-01 -7.18178675e-02 3.75931859e-01 4.50105250e-01 -8.29148665e-02 -4.00000572e-01 -1.37979150e-01 2.07912982e-01 -8.39690343e-02 -3.22413325e-01 1.09628963e+00 -9.01613474e-01 -4.35600877e-02 1.63878843e-01 1.09677732e+00 6.85212195e-01 -8.90780389e-01 -6.62292957e-01 -4.98875603e-02 -5.56295037e-01 -9.68477279e-02 -2.35496297e-01 -1.12083948e+00 4.60846543e-01 2.22150713e-01 8.35090995e-01 9.30791616e-01 1.19234771e-01 2.93363303e-01 7.74755597e-01 7.60121822e-01 -7.75776803e-01 -3.86380941e-01 1.71778426e-01 5.43956041e-01 -6.23677135e-01 2.99438208e-01 -8.69270861e-01 -1.96755618e-01 7.66488791e-01 -2.73352712e-01 -9.07215476e-02 1.09043014e+00 -1.85657013e-02 -7.99858570e-01 -9.43374336e-02 -7.93758333e-01 -3.39437991e-01 3.82694714e-02 5.01893759e-01 3.00140113e-01 3.06882590e-01 -7.45842874e-01 8.77262294e-01 -2.80278265e-01 -2.88945913e-01 1.11974633e+00 5.02011001e-01 -5.03356397e-01 -1.14459836e+00 -1.28825262e-01 6.37346625e-01 -6.35047317e-01 -2.71704823e-01 -4.06589776e-01 5.50760150e-01 -6.02499127e-01 1.35715294e+00 -1.73514679e-01 -4.24628943e-01 5.55817224e-02 -5.55526257e-01 7.95211911e-01 -4.08746481e-01 -1.09821871e-01 1.00076184e-01 3.46278310e-01 -2.19136998e-01 -2.17352003e-01 -5.62291622e-01 -6.99362695e-01 -6.78126633e-01 -7.15616167e-01 5.72181940e-01 3.38397950e-01 8.51253271e-01 6.04704261e-01 -1.03262940e-03 1.12962437e+00 -2.46124551e-01 -8.99443150e-01 -6.48497105e-01 -1.20384824e+00 9.83131975e-02 2.22031400e-01 -3.77664089e-01 -7.62522280e-01 -4.56893444e-01]
[6.879227161407471, 4.9992356300354]
94a5d300-0bef-45bd-a0ee-ba1a08aa1ae8
online-photometric-calibration-of-automatic
2012.14292
null
https://arxiv.org/abs/2012.14292v2
https://arxiv.org/pdf/2012.14292v2.pdf
Online Photometric Calibration of Automatic Gain Thermal Infrared Cameras
Thermal infrared cameras are increasingly being used in various applications such as robot vision, industrial inspection and medical imaging, thanks to their improved resolution and portability. However, the performance of traditional computer vision techniques developed for electro-optical imagery does not directly translate to the thermal domain due to two major reasons: these algorithms require photometric assumptions to hold, and methods for photometric calibration of RGB cameras cannot be applied to thermal-infrared cameras due to difference in data acquisition and sensor phenomenology. In this paper, we take a step in this direction, and introduce a novel algorithm for online photometric calibration of thermal-infrared cameras. Our proposed method does not require any specific driver/hardware support and hence can be applied to any commercial off-the-shelf thermal IR camera. We present this in the context of visual odometry and SLAM algorithms, and demonstrate the efficacy of our proposed system through extensive experiments for both standard benchmark datasets, and real-world field tests with a thermal-infrared camera in natural outdoor environments.
['Shreyansh Daftry', 'Larry Matthies', 'Manash Pratim Das']
2020-12-07
null
null
null
null
['camera-auto-calibration', 'thermal-image-denoising']
['computer-vision', 'computer-vision']
[ 5.27082145e-01 -4.06901956e-01 2.66925514e-01 -3.63490820e-01 -1.38427734e-01 -6.63308978e-01 2.34430298e-01 -2.67932802e-01 -6.50212884e-01 3.65013629e-01 -5.62299907e-01 -3.26302141e-01 -1.19863480e-01 -4.20060605e-01 -4.32744950e-01 -7.91761279e-01 4.68937665e-01 3.28383058e-01 2.70625710e-01 -1.77983761e-01 2.53525913e-01 4.97280061e-01 -1.32558310e+00 -5.53220809e-01 6.28677189e-01 8.65967035e-01 2.99472481e-01 5.57051778e-01 3.10396314e-01 4.78995115e-01 -2.52640426e-01 -9.34107751e-02 6.99995875e-01 -2.38536134e-01 -5.96708298e-01 5.59542775e-01 3.19520324e-01 -4.20528948e-01 -2.35623360e-01 1.13612652e+00 2.25659236e-01 9.72605571e-02 2.54151642e-01 -1.30153477e+00 -5.07312536e-01 -3.66964847e-01 -8.61019552e-01 -1.25552893e-01 3.50138128e-01 1.55640423e-01 4.75123346e-01 -4.56570834e-01 3.29557121e-01 8.33634794e-01 6.93547487e-01 2.79033273e-01 -9.93926227e-01 -5.23433745e-01 -1.62828982e-01 1.04780123e-01 -1.15333986e+00 -2.13364482e-01 9.03983712e-01 -4.51837420e-01 7.13770390e-01 -1.68251440e-01 4.55589980e-01 8.62091243e-01 3.83709580e-01 9.35318395e-02 1.43736792e+00 -5.95712066e-01 2.68406093e-01 3.11120629e-01 1.17561452e-01 5.15436947e-01 6.94523573e-01 3.24264705e-01 -3.92386019e-01 3.31981219e-02 7.89827943e-01 1.70506760e-01 -2.92593509e-01 -8.12846780e-01 -1.03331637e+00 5.76087594e-01 4.03911471e-01 4.93320450e-02 -2.62133747e-01 1.95584655e-01 2.51181841e-01 3.33870620e-01 2.57705957e-01 2.22413480e-01 -2.91475058e-01 -1.42594412e-01 -4.68603551e-01 -4.63042825e-01 4.38622475e-01 1.13222814e+00 1.06634033e+00 -3.36007513e-02 9.38769281e-01 6.51246965e-01 4.74983215e-01 7.23782837e-01 3.91878396e-01 -1.00900364e+00 2.79213130e-01 4.67830002e-01 1.52451307e-01 -7.64469504e-01 -3.23206007e-01 8.87009781e-03 -6.99279189e-01 6.35580659e-01 1.05359435e-01 -2.24122286e-01 -6.07850790e-01 1.01204085e+00 4.38661486e-01 1.57111794e-01 4.21381563e-01 1.05986822e+00 2.53180802e-01 2.99582332e-01 -5.11388361e-01 -1.57601297e-01 1.33014798e+00 -6.72457039e-01 -4.19509888e-01 -7.32571185e-01 4.26777005e-01 -1.01966369e+00 7.68999338e-01 6.68766141e-01 -4.28857565e-01 -6.92004681e-01 -1.28023005e+00 -4.19110768e-02 -1.35859758e-01 3.10850978e-01 4.42274570e-01 8.51014197e-01 -1.00654089e+00 2.88047701e-01 -9.66675282e-01 -1.11270380e+00 -2.52759218e-01 4.23242748e-01 -5.47488093e-01 -1.84685990e-01 -7.50958323e-01 1.05577707e+00 4.54932749e-01 5.26963055e-01 -3.80629241e-01 -1.65759578e-01 -8.10973048e-01 -5.34749210e-01 4.28944111e-01 -5.56500971e-01 1.09229338e+00 -1.01317728e+00 -1.87640560e+00 8.41874778e-01 4.21217419e-02 -2.65482932e-01 2.08492815e-01 -5.31614602e-01 -1.91683948e-01 2.74436474e-01 -1.31280437e-01 2.31842875e-01 7.20001817e-01 -1.22365928e+00 -6.46916926e-01 -6.38624191e-01 2.41708905e-01 1.45012259e-01 -3.58143300e-01 -1.20147459e-01 -4.95284736e-01 8.25969428e-02 4.19105351e-01 -1.57413101e+00 -1.97142765e-01 1.27535269e-01 -1.24519326e-01 4.17769402e-01 1.10529888e+00 -1.84642255e-01 2.82272428e-01 -1.99572635e+00 -9.35101211e-02 5.82225770e-02 -4.18990076e-01 1.94373548e-01 2.73689717e-01 4.17944759e-01 -6.33486686e-03 -6.45605206e-01 -2.12376937e-01 -2.81719178e-01 -3.40126455e-01 2.75674671e-01 -2.71710724e-01 9.47258592e-01 -1.85312659e-01 2.85812110e-01 -8.28999221e-01 -3.91012937e-01 9.53387201e-01 5.09063244e-01 -6.26325831e-02 2.76105106e-02 1.77482560e-01 6.37452662e-01 -4.75433797e-01 3.83591920e-01 8.18256736e-01 2.51209706e-01 2.85010010e-01 -3.75260413e-01 -3.98189425e-01 -1.13822162e-01 -1.19907403e+00 1.74613166e+00 -7.50925064e-01 7.01406062e-01 1.66342437e-01 -8.62746418e-01 1.09073365e+00 1.81403130e-01 6.20690584e-01 -7.19587505e-01 2.43461102e-01 2.37766355e-01 -3.78502548e-01 -5.09557843e-01 7.70110250e-01 -1.82038903e-01 -2.20213551e-02 4.93582994e-01 -2.72263944e-01 -4.81907010e-01 -4.07839045e-02 -2.54188687e-01 6.82501793e-01 7.51225412e-01 3.48038852e-01 -1.18520088e-01 5.56370676e-01 4.27076310e-01 5.55669487e-01 2.58083165e-01 -3.04032937e-02 5.58289170e-01 -2.02979088e-01 -3.35682005e-01 -1.07726538e+00 -8.30098331e-01 -5.43545298e-02 6.86820090e-01 7.24215209e-01 1.53183728e-01 -4.71623600e-01 -5.56312203e-02 -1.42144516e-01 4.09829527e-01 -2.31257126e-01 -2.05251738e-01 -3.49940211e-01 -6.71346188e-01 3.47655863e-01 5.27269661e-01 7.04598069e-01 -5.81299126e-01 -1.35703301e+00 3.13340425e-02 1.62286580e-01 -1.59840906e+00 -4.17094119e-02 1.92815349e-01 -1.25087428e+00 -1.35802901e+00 -3.09962332e-01 -6.72617912e-01 7.05024958e-01 1.09975612e+00 6.29018903e-01 -1.13198049e-01 -4.78124291e-01 7.71476924e-01 -4.13286269e-01 -4.23756242e-01 -3.42082679e-01 -1.50271833e-01 2.21033007e-01 7.12075159e-02 4.76731718e-01 -3.99315596e-01 -7.57823348e-01 5.84357619e-01 -9.91092503e-01 -1.12376757e-01 8.06300879e-01 5.72702825e-01 2.14802086e-01 1.70757443e-01 -8.42055455e-02 -8.13042581e-01 6.37357756e-02 -3.77171062e-04 -1.22433901e+00 1.38918355e-01 -7.53282845e-01 8.11890662e-02 5.28691530e-01 -8.92330036e-02 -1.22190928e+00 7.04789102e-01 4.14571613e-01 -4.49091136e-01 -2.48448312e-01 3.26769143e-01 1.13177150e-01 -4.04205650e-01 6.85868502e-01 1.17627919e-01 1.58756495e-01 -3.32718760e-01 1.76722854e-01 9.25113440e-01 9.15671706e-01 -5.11003852e-01 1.32585633e+00 1.05699027e+00 3.00828397e-01 -1.16552770e+00 -5.27151763e-01 -8.77204776e-01 -9.66600537e-01 -1.76513046e-01 8.38361442e-01 -9.50874150e-01 -9.21569049e-01 5.23854196e-01 -1.01296985e+00 1.15846127e-01 3.50588530e-01 8.30795944e-01 -5.93534410e-01 9.79970515e-01 -3.07223976e-01 -9.54731345e-01 -3.13835591e-01 -1.30747747e+00 1.16982555e+00 3.22038293e-01 6.49819821e-02 -1.11194122e+00 1.94524184e-01 5.38197875e-01 2.50155270e-01 1.77629381e-01 2.73748815e-01 3.76100540e-01 -6.65409207e-01 -3.65423709e-01 -1.61322832e-01 5.32297134e-01 3.69349539e-01 1.54832184e-01 -1.24003029e+00 -5.45440972e-01 2.24267393e-01 -3.64983559e-01 5.06697834e-01 1.74009129e-01 4.89470512e-01 3.98737520e-01 -2.87001640e-01 7.23033071e-01 1.86034596e+00 2.86740482e-01 6.13062680e-01 8.11543226e-01 7.46480286e-01 7.27778733e-01 1.12546623e+00 2.83835590e-01 1.70513779e-01 8.68672371e-01 7.65316725e-01 -2.33701512e-01 5.08200884e-01 1.63124725e-01 5.38840353e-01 5.03640831e-01 -2.27020591e-01 9.46124643e-02 -8.02386999e-01 4.16762918e-01 -1.73569441e+00 -6.78760111e-01 -3.98375362e-01 2.76084352e+00 2.91777134e-01 -2.34897416e-02 -2.14451402e-01 1.07588738e-01 6.89604223e-01 -1.78184167e-01 -3.51509422e-01 -4.14990962e-01 3.56956899e-01 -2.25481726e-02 1.15417075e+00 3.22761655e-01 -1.03890908e+00 6.57716334e-01 5.62241888e+00 -1.43640786e-01 -1.39412844e+00 1.26147538e-01 -1.10933296e-02 3.05017799e-01 3.53020132e-01 3.86469632e-01 -5.20357847e-01 4.55148630e-02 6.76649570e-01 8.23257416e-02 4.35469002e-01 9.87294793e-01 2.80666143e-01 -5.96868694e-01 -1.15440238e+00 1.21980035e+00 9.67471674e-03 -6.32435262e-01 -6.10359371e-01 2.88770199e-01 5.53393185e-01 1.04153857e-01 9.12258253e-02 -3.31550598e-01 9.50913802e-02 -5.14341831e-01 5.01689792e-01 1.07352033e-01 5.15239358e-01 -5.23300767e-01 8.52644265e-01 3.27583492e-01 -1.05625594e+00 -5.18010696e-03 -8.05912018e-01 -1.60784245e-01 6.59278557e-02 4.78611499e-01 -1.00850046e+00 8.70313525e-01 8.45066607e-01 6.15484476e-01 -4.73475903e-01 7.73670435e-01 6.54997155e-02 1.13968581e-01 -4.28676486e-01 1.90165162e-01 1.95720479e-01 -7.32446611e-01 9.32829157e-02 8.23818624e-01 4.87993568e-01 -1.62940145e-01 6.47708327e-02 5.55057764e-01 5.10866761e-01 -2.57701159e-01 -8.70785236e-01 2.70131230e-01 1.84552833e-01 1.51620317e+00 -6.58317983e-01 4.86083655e-03 -6.82544172e-01 1.21016562e+00 -2.57352710e-01 3.28829437e-01 -9.18918848e-01 -4.03523833e-01 7.15856433e-01 -1.34397313e-01 1.75418377e-01 -7.07017124e-01 -8.21875259e-02 -1.11185205e+00 6.15704134e-02 -4.69806343e-01 5.11287302e-02 -1.02279723e+00 -9.46745336e-01 5.54638147e-01 1.85235545e-01 -1.70576859e+00 -2.02807814e-01 -9.92837608e-01 -3.47263634e-01 7.48743296e-01 -1.73818624e+00 -1.22242081e+00 -9.50906694e-01 5.40144563e-01 4.09892589e-01 2.08327338e-01 5.10213017e-01 7.36470148e-02 -3.62960219e-01 1.33549273e-02 4.69149500e-01 -7.05004334e-02 1.00392604e+00 -1.03195882e+00 3.95216674e-01 1.15375376e+00 -2.67805122e-02 7.63579190e-01 9.23117876e-01 -2.77262211e-01 -2.01648283e+00 -9.61173117e-01 1.99197397e-01 -3.61557424e-01 6.31287992e-01 -7.31639490e-02 -4.79093105e-01 7.35907018e-01 3.10175657e-01 6.41404614e-02 2.03416780e-01 -1.97128728e-01 -1.82934821e-01 -6.56085253e-01 -1.08029771e+00 1.52309716e-01 6.91840231e-01 -6.10270917e-01 -5.92025280e-01 4.44040626e-01 2.85580575e-01 -4.63803381e-01 -7.73417771e-01 2.14943081e-01 6.42356694e-01 -1.19757736e+00 9.32830811e-01 3.21267128e-01 -1.28711626e-01 -7.93716669e-01 -2.81867739e-02 -1.15737748e+00 1.64773539e-02 -6.24438226e-01 8.65515232e-01 1.11717093e+00 -2.21279651e-01 -1.02069807e+00 8.02790523e-01 5.08538187e-01 -2.64706016e-01 1.14983924e-01 -8.36006343e-01 -7.80570567e-01 -4.28963929e-01 -4.40300405e-01 1.76790804e-01 7.67489731e-01 -1.63159534e-01 2.60888696e-01 -4.93984520e-01 6.18957937e-01 1.04990995e+00 1.71984702e-01 1.07900381e+00 -1.20859528e+00 -5.00714540e-01 2.33305901e-01 -8.83860469e-01 -8.54088843e-01 -1.14276409e-01 -1.38990760e-01 4.06988829e-01 -1.26428163e+00 1.59159452e-01 -3.51001561e-01 7.56748840e-02 1.32151589e-01 1.62921205e-01 7.07778096e-01 -1.44023681e-02 3.83932322e-01 -2.04750478e-01 3.00507516e-01 6.94491029e-01 -3.74953151e-02 1.06827226e-02 3.97089645e-02 -2.82300740e-01 7.23445892e-01 6.51544929e-01 -2.76324332e-01 -7.40237176e-01 -6.21117353e-01 2.47344628e-01 -1.86470851e-01 4.92198884e-01 -1.29859626e+00 3.27087134e-01 -2.57843405e-01 1.46458805e-01 -3.38140637e-01 5.12025535e-01 -1.46010125e+00 4.01794046e-01 6.29242480e-01 3.24487716e-01 2.09291935e-01 -8.97513609e-03 6.24701560e-01 -5.53070456e-02 -2.96259046e-01 1.03689766e+00 -5.19367419e-02 -1.12815261e+00 -2.14525536e-01 -2.66478300e-01 -5.22152483e-01 1.35439014e+00 -7.91699946e-01 -5.91561377e-01 -2.47536197e-01 4.24065292e-02 -3.64084579e-02 1.25042462e+00 4.10813361e-01 5.20005763e-01 -9.04737711e-01 -9.83900353e-02 2.62914896e-01 5.34061134e-01 6.72477186e-02 9.90159586e-02 1.00590336e+00 -1.28147209e+00 5.92358768e-01 -3.49185169e-01 -9.80083883e-01 -1.41992152e+00 7.71092951e-01 3.54262173e-01 4.07103330e-01 -6.00785732e-01 5.60378022e-02 2.37086013e-01 -4.48907226e-01 -3.15580927e-02 -5.73559344e-01 3.10988873e-01 -6.30158305e-01 8.59250575e-02 3.58036250e-01 2.57017255e-01 -8.45199049e-01 -5.85217595e-01 1.29793811e+00 3.71965528e-01 -2.25756437e-01 1.03437638e+00 -5.59619784e-01 -2.12162867e-01 5.02495527e-01 1.08506286e+00 -3.56023580e-01 -1.07243896e+00 -1.83613569e-01 4.23535369e-02 -5.32495856e-01 3.15654457e-01 -1.03072949e-01 -1.02267492e+00 8.89334440e-01 8.16859603e-01 -9.42006856e-02 1.39188445e+00 -2.86505163e-01 6.31013691e-01 8.34034920e-01 9.31178331e-01 -1.16796613e+00 -2.28846818e-02 2.57243961e-01 1.21300675e-01 -1.26916814e+00 4.54695702e-01 -4.77163434e-01 -7.14286864e-01 1.44094527e+00 4.48422015e-01 -1.22494541e-01 3.16937029e-01 2.85245359e-01 6.34806931e-01 -7.13597937e-03 -5.08474171e-01 -1.72603279e-01 -2.51505613e-01 7.85302937e-01 5.50202310e-01 -4.15491879e-01 8.29864517e-02 -7.66292572e-01 2.81736195e-01 1.98431939e-01 9.52187598e-01 1.30036521e+00 -5.46133757e-01 -1.13170850e+00 -9.02333856e-01 -1.70099705e-01 -1.58797830e-01 4.43702936e-01 -2.11188734e-01 9.02790666e-01 -3.08959447e-02 1.23072016e+00 3.82088162e-02 -4.64338392e-01 2.22279489e-01 -1.73524171e-01 5.50134540e-01 -4.79076445e-01 -3.28945398e-01 1.50526240e-01 -2.27264285e-01 -4.89462227e-01 -9.77657437e-01 -7.51019537e-01 -1.04522252e+00 -2.80845672e-01 -6.09422803e-01 2.95068678e-02 1.31454587e+00 8.14832449e-01 1.45986825e-01 1.24806039e-01 8.60538900e-01 -8.85845661e-01 -5.17584741e-01 -8.39226723e-01 -6.57541275e-01 2.89255798e-01 1.37597710e-01 -6.44290745e-01 -2.85526782e-01 2.29826853e-01]
[8.081276893615723, -2.2626638412475586]
a2c7afc6-840f-4543-8a6c-2d1f90647b6b
sparsegnv-generating-novel-views-of-indoor
2305.07024
null
https://arxiv.org/abs/2305.07024v1
https://arxiv.org/pdf/2305.07024v1.pdf
SparseGNV: Generating Novel Views of Indoor Scenes with Sparse Input Views
We study to generate novel views of indoor scenes given sparse input views. The challenge is to achieve both photorealism and view consistency. We present SparseGNV: a learning framework that incorporates 3D structures and image generative models to generate novel views with three modules. The first module builds a neural point cloud as underlying geometry, providing contextual information and guidance for the target novel view. The second module utilizes a transformer-based network to map the scene context and the guidance into a shared latent space and autoregressively decodes the target view in the form of discrete image tokens. The third module reconstructs the tokens into the image of the target view. SparseGNV is trained across a large indoor scene dataset to learn generalizable priors. Once trained, it can efficiently generate novel views of an unseen indoor scene in a feed-forward manner. We evaluate SparseGNV on both real-world and synthetic indoor scenes and demonstrate that it outperforms state-of-the-art methods based on either neural radiance fields or conditional image generation.
['Ying Shan', 'Yan-Pei Cao', 'Weihao Cheng']
2023-05-11
null
null
null
null
['conditional-image-generation']
['computer-vision']
[ 5.34765363e-01 4.35220569e-01 5.16956866e-01 -6.05028749e-01 -7.45199323e-01 -7.77316034e-01 8.48750532e-01 -6.55864835e-01 3.12996536e-01 6.35715902e-01 4.24137712e-01 -2.16013249e-02 2.05862314e-01 -1.16224766e+00 -1.35499394e+00 -6.40811920e-01 4.07696694e-01 4.44858462e-01 -1.17906921e-01 3.67016206e-03 1.30957857e-01 5.51151752e-01 -1.65512133e+00 6.01337969e-01 6.79289520e-01 7.02256620e-01 6.42917037e-01 1.02176821e+00 1.26882538e-01 8.29323649e-01 -3.25138450e-01 -1.12933174e-01 6.96848691e-01 -3.41523528e-01 -4.51714665e-01 5.80806851e-01 8.11723351e-01 -5.95362127e-01 -5.28175592e-01 6.23890460e-01 2.81475544e-01 2.10788786e-01 6.41185284e-01 -1.00128973e+00 -1.07790124e+00 1.75056234e-01 -2.81253844e-01 -2.34588936e-01 5.93915403e-01 2.85339534e-01 7.13962257e-01 -1.32453299e+00 1.00571108e+00 1.19586766e+00 2.92753607e-01 5.57037294e-01 -1.44701767e+00 -2.89126754e-01 4.92563069e-01 -3.96337569e-01 -1.12590313e+00 -6.74076617e-01 9.29460645e-01 -5.60015500e-01 9.60140467e-01 2.35259682e-01 7.66298532e-01 1.42160678e+00 3.10117543e-01 4.59668577e-01 1.04597986e+00 -2.47336790e-01 5.17329395e-01 1.46271184e-01 -7.39977539e-01 7.25346923e-01 -1.56899780e-01 4.09779310e-01 -6.05760574e-01 -1.65102724e-02 1.22325790e+00 3.41904849e-01 -3.11758965e-01 -7.83336103e-01 -1.39091647e+00 7.48340249e-01 7.95772672e-01 -2.66242981e-01 -5.56565166e-01 4.72923666e-01 -4.34485197e-01 -1.18405193e-01 6.04469419e-01 4.51294631e-01 -3.39819252e-01 3.59766811e-01 -7.67918587e-01 4.62115318e-01 5.31870365e-01 1.26354134e+00 7.98954070e-01 3.60315174e-01 -1.38952378e-02 6.02213383e-01 6.01711392e-01 9.28654075e-01 -1.85034394e-01 -1.60539091e+00 5.72584867e-01 1.83387041e-01 1.18826963e-01 -8.09736609e-01 2.83123046e-01 -4.38639075e-01 -8.07514250e-01 4.62036073e-01 -2.36527443e-01 -1.15095235e-01 -1.28073037e+00 1.68531787e+00 5.02336323e-01 1.74368456e-01 1.45688280e-01 6.71785653e-01 9.91044044e-01 1.16466188e+00 -4.96719897e-01 1.53627589e-01 7.21693993e-01 -1.01278079e+00 -1.68737397e-01 -6.11358523e-01 -9.13004056e-02 -8.27821732e-01 7.77966201e-01 2.69877553e-01 -1.32623947e+00 -8.31275046e-01 -8.64298046e-01 -3.02427053e-01 -1.25167444e-01 1.16217487e-01 5.23675561e-01 2.15263695e-01 -1.40034068e+00 1.24654852e-01 -7.93732584e-01 -2.87928313e-01 3.86981875e-01 -3.92029174e-02 -5.07370472e-01 -5.20853817e-01 -3.58141363e-01 3.46567541e-01 1.47413224e-01 2.59361044e-02 -1.57036889e+00 -9.65532601e-01 -1.40995347e+00 -4.38247584e-02 2.60277893e-02 -1.60516822e+00 1.09728515e+00 -6.95232630e-01 -1.53578532e+00 7.33698428e-01 -3.11472118e-01 -6.86274767e-02 3.43856037e-01 -5.87771833e-02 9.32070985e-02 -1.28281862e-02 5.28076172e-01 1.10437024e+00 9.28818464e-01 -2.10706306e+00 -5.11197567e-01 -2.42370754e-01 3.58758539e-01 5.15617430e-01 7.64052987e-01 -7.77202547e-01 -4.40102488e-01 -6.66786373e-01 6.13592803e-01 -1.01165271e+00 -4.90029395e-01 5.56067340e-02 -6.44831121e-01 5.20137608e-01 7.74071515e-01 -5.04131854e-01 1.06610529e-01 -2.10273623e+00 5.21195829e-01 2.49279112e-01 1.75915271e-01 -5.39403975e-01 -1.03548169e-01 5.72334766e-01 -1.71218634e-01 -1.60129815e-01 -1.85725987e-01 -7.94556081e-01 -1.58278450e-01 3.02276164e-01 -8.06464911e-01 2.49209434e-01 2.68265307e-02 9.53756809e-01 -9.31526780e-01 7.18323290e-02 7.49865532e-01 7.58449554e-01 -9.84868586e-01 6.44500673e-01 -5.59347630e-01 9.44107413e-01 -1.53166518e-01 3.89382094e-01 7.84707010e-01 -3.04974884e-01 8.58181491e-02 -2.62025923e-01 -1.15776904e-01 2.25953028e-01 -1.01432061e+00 2.24469280e+00 -8.86737525e-01 4.74835008e-01 -3.11885700e-02 -5.11556447e-01 8.92817736e-01 3.93432416e-02 1.90684527e-01 -4.09058958e-01 -2.45330021e-01 -1.57991797e-01 -8.26232851e-01 -4.82088149e-01 5.99156916e-01 -2.14539319e-01 -3.01443227e-03 3.91581416e-01 2.72765607e-01 -7.67109394e-01 -3.72720391e-01 5.57783186e-01 8.74119937e-01 7.94552863e-01 6.20394833e-02 2.34584138e-02 1.44112051e-01 -3.62554729e-01 2.34125420e-01 6.14670753e-01 7.61492729e-01 1.22408485e+00 -1.29892007e-01 -5.89720428e-01 -1.39490008e+00 -1.87012649e+00 3.45964581e-01 6.09283745e-01 5.27468324e-02 -2.54398614e-01 -3.81905675e-01 -5.10884643e-01 -1.30971417e-01 1.13368833e+00 -7.46390283e-01 8.19649920e-02 -3.32001597e-01 -1.85871631e-01 -2.77635783e-01 5.43396175e-01 4.41127270e-01 -9.84305799e-01 -5.98002911e-01 -1.32306786e-02 -3.79034877e-01 -1.24923551e+00 -2.33594388e-01 1.09243765e-01 -6.62929177e-01 -7.76243091e-01 -5.83428800e-01 -8.27611506e-01 1.28345597e+00 6.25118136e-01 1.37563550e+00 -3.35081190e-01 -2.31984064e-01 8.81314158e-01 -9.32123214e-02 -2.48629481e-01 -5.33010066e-01 -3.33581448e-01 -1.24585472e-01 1.89226605e-02 -4.51802224e-01 -1.02200794e+00 -6.82223380e-01 4.56494875e-02 -8.44105542e-01 7.86218941e-01 3.55908126e-01 6.96238518e-01 9.49352264e-01 -3.61767054e-01 -9.80612561e-02 -8.21263790e-01 -1.21437363e-01 -4.65501040e-01 -8.16347539e-01 1.71973258e-01 3.88016552e-03 1.19061120e-01 6.42583251e-01 1.35825515e-01 -1.55688798e+00 3.45247597e-01 -1.31119147e-01 -5.78499854e-01 -4.31005001e-01 4.69886847e-02 -3.21712613e-01 9.28378329e-02 7.04277098e-01 5.39464414e-01 -4.90265578e-01 -2.69525200e-01 8.68013382e-01 6.37330338e-02 8.43881190e-01 -7.96078563e-01 1.20287025e+00 8.98410797e-01 -1.23415599e-02 -7.90641963e-01 -1.00854814e+00 -4.71239351e-02 -7.95538187e-01 -2.02889651e-01 1.11930633e+00 -1.28962350e+00 -2.86773145e-01 2.13490099e-01 -1.35156775e+00 -5.98186910e-01 -7.82684624e-01 2.34305382e-01 -1.02157986e+00 -9.82155129e-02 -2.89341986e-01 -6.22033536e-01 2.36384645e-02 -1.02231824e+00 1.74699557e+00 1.91918328e-01 1.26003474e-01 -8.88333857e-01 2.29693949e-01 4.68227774e-01 1.44195005e-01 6.07728183e-01 6.64604247e-01 3.10513139e-01 -1.59603107e+00 2.97941685e-01 -5.30498335e-03 3.32266003e-01 2.98070133e-01 5.30900322e-02 -1.33189797e+00 -4.00931776e-01 1.94393978e-01 -3.23635072e-01 7.42395580e-01 6.71695411e-01 1.18257749e+00 -3.69934857e-01 -3.26747179e-01 1.29214919e+00 1.70753407e+00 3.43392938e-01 7.24225223e-01 -5.79369739e-02 9.54602718e-01 4.62980777e-01 2.09125862e-01 2.73923725e-01 6.72429264e-01 4.21834737e-01 7.29637086e-01 -2.48757690e-01 -2.12316409e-01 -8.81132364e-01 1.60937935e-01 6.50783241e-01 -4.47896644e-02 -3.94545376e-01 -5.68370163e-01 5.06782293e-01 -1.59462929e+00 -1.14950764e+00 3.45132649e-01 1.99557495e+00 2.48479083e-01 -9.60325152e-02 -5.89827597e-01 -3.47689956e-01 2.17917487e-01 4.98491287e-01 -5.17595887e-01 -2.23991901e-01 -8.95195529e-02 1.04359597e-01 4.64852862e-02 7.81881809e-01 -7.89012432e-01 9.39489186e-01 6.43978834e+00 5.37322462e-03 -9.25902009e-01 -1.04175515e-01 8.25572789e-01 -2.19848529e-01 -9.63242173e-01 8.30824003e-02 -6.13041282e-01 -1.84152368e-02 4.78797376e-01 9.08636898e-02 7.64333606e-01 9.53645945e-01 1.78827584e-01 8.50484595e-02 -1.32609117e+00 1.25283754e+00 4.71412987e-01 -1.88953161e+00 3.82865101e-01 1.97031409e-01 1.47955120e+00 2.04450950e-01 3.26761752e-01 4.71530072e-02 8.05181980e-01 -1.01017308e+00 1.01426482e+00 9.10061896e-01 9.83285129e-01 -5.27461946e-01 1.03427857e-01 3.50003153e-01 -1.03693688e+00 1.55813515e-01 -5.22379577e-01 4.47621234e-02 4.73053426e-01 4.97178137e-01 -1.00810909e+00 6.98134005e-01 7.51941502e-01 9.33557391e-01 -4.38725024e-01 6.63990140e-01 -5.73158503e-01 1.74080074e-01 -2.07026452e-01 6.95442319e-01 7.40107372e-02 -3.18411767e-01 5.50688744e-01 6.09244406e-01 7.01156676e-01 1.95534512e-01 2.23644570e-01 1.55264390e+00 -1.46370098e-01 -5.05712092e-01 -1.34279275e+00 4.23351735e-01 2.09418699e-01 1.25711274e+00 -5.09121239e-01 -4.15112406e-01 -1.63505644e-01 1.37606680e+00 1.49338335e-01 8.00837755e-01 -6.76585972e-01 1.25741646e-01 5.12679040e-01 1.23177595e-01 6.73648357e-01 -3.94605130e-01 -2.01742351e-01 -1.49812829e+00 1.73090305e-02 -6.95583880e-01 -1.52019233e-01 -1.68120277e+00 -1.13158059e+00 7.41455972e-01 8.54599178e-02 -1.15716743e+00 -7.12346733e-01 -3.08144212e-01 -5.84543169e-01 1.01060057e+00 -1.14257026e+00 -1.68226278e+00 -8.44933093e-01 5.27535915e-01 9.62774336e-01 -1.34836704e-01 1.02178240e+00 -3.27066749e-01 8.32543075e-02 -1.10801972e-01 7.80527219e-02 -1.79006711e-01 2.91758209e-01 -1.35847640e+00 1.11185288e+00 1.10943484e+00 4.90352064e-01 5.84667385e-01 6.01524174e-01 -5.81167281e-01 -1.24779046e+00 -1.51504970e+00 4.40718681e-01 -9.12740171e-01 -2.80172944e-01 -7.97410309e-01 -2.83260792e-01 1.20277739e+00 3.60044599e-01 1.75576240e-01 5.67552805e-01 -3.28077048e-01 -4.81280506e-01 -3.39686237e-02 -1.08973944e+00 6.32701755e-01 1.29492080e+00 -6.50896132e-01 -4.54830766e-01 3.23792607e-01 1.05808854e+00 -8.86985362e-01 -6.06448770e-01 2.40864113e-01 4.03646976e-01 -1.40936160e+00 1.43572652e+00 -3.19489300e-01 8.99828315e-01 -5.83534181e-01 -8.00038040e-01 -1.71670687e+00 -5.80632150e-01 -3.79382163e-01 -7.56632462e-02 9.73356426e-01 3.75025153e-01 -3.78736496e-01 8.11795294e-01 3.96081179e-01 -4.31755096e-01 -6.79349244e-01 -4.73622590e-01 -2.88620532e-01 -2.31398538e-01 -4.70469236e-01 8.20074081e-01 6.47081852e-01 -8.89830232e-01 5.76321363e-01 -3.58369231e-01 5.07578552e-01 1.00411308e+00 4.86209601e-01 1.35585570e+00 -6.99916005e-01 -6.61615610e-01 3.36807400e-01 -2.71291554e-01 -1.38489568e+00 -3.72250974e-02 -8.83955359e-01 2.14266658e-01 -2.24323773e+00 1.02019332e-01 -2.93056786e-01 4.23246235e-01 1.66841418e-01 8.39168206e-02 3.03391039e-01 2.21387610e-01 -1.12594023e-01 -3.83627921e-01 9.54504550e-01 1.29329705e+00 -1.51308581e-01 -2.26174861e-01 -2.82352827e-02 -7.73683012e-01 6.91541076e-01 4.32635069e-01 -7.73360059e-02 -1.00143552e+00 -8.79753530e-01 4.57008690e-01 1.04152240e-01 8.15068901e-01 -1.15229976e+00 -9.11112353e-02 -3.60924691e-01 1.11420190e+00 -9.42741632e-01 8.41735244e-01 -7.46676385e-01 7.45788157e-01 5.16877801e-04 -2.11286634e-01 1.16267160e-01 1.27368141e-02 7.44048834e-01 1.60903186e-01 4.42883492e-01 5.95442355e-01 -5.62755466e-01 -4.83199865e-01 4.61988509e-01 1.84381697e-02 -9.60521623e-02 9.25646722e-01 -2.99460202e-01 -3.09899032e-01 -6.61512613e-01 -7.96448052e-01 -5.46620823e-02 9.31038082e-01 6.44608617e-01 1.17383420e+00 -1.53510225e+00 -6.88863993e-01 7.81483769e-01 2.16050029e-01 8.36171865e-01 6.20763302e-01 -3.19342166e-02 -6.92546189e-01 2.68967927e-01 -8.12512115e-02 -1.01500881e+00 -9.54575360e-01 6.85827076e-01 4.34788495e-01 1.49368718e-01 -8.76740694e-01 1.11406112e+00 1.12553680e+00 -9.02329445e-01 -2.52673268e-01 -4.92746025e-01 2.94293761e-01 -7.62150228e-01 3.22825462e-01 -9.00383294e-02 -2.54024059e-01 -7.65252948e-01 -2.38905493e-02 6.45060122e-01 2.82966971e-01 -4.96262193e-01 1.61075282e+00 -2.95582771e-01 -8.21637958e-02 5.24369299e-01 1.13764119e+00 2.31886432e-01 -1.75178969e+00 -1.36468410e-01 -9.49984491e-01 -8.89190674e-01 -1.84112474e-01 -7.97406197e-01 -9.80527222e-01 7.22978294e-01 2.95904219e-01 -2.86566824e-01 9.22637224e-01 2.74134547e-01 5.07490456e-01 2.72250831e-01 8.23808968e-01 -4.84667778e-01 3.61009449e-01 3.98483783e-01 1.32333803e+00 -1.02339196e+00 3.78206894e-02 -6.01333022e-01 -4.65347737e-01 8.78208637e-01 6.94931805e-01 -3.19261760e-01 6.98657215e-01 1.38235733e-01 8.71581733e-02 -5.45991123e-01 -9.12217796e-01 3.25447410e-01 4.48040575e-01 9.74145174e-01 9.38941166e-02 7.36363679e-02 8.91120732e-01 7.74022043e-02 -7.81870306e-01 -1.16967328e-01 5.30301809e-01 7.40121484e-01 -7.49367774e-02 -6.71632409e-01 -4.45819914e-01 2.65547186e-01 1.39675468e-01 -1.34333864e-01 -1.19666815e-01 3.67721260e-01 3.78383011e-01 7.46610522e-01 1.11573070e-01 -4.39233959e-01 3.31225336e-01 -3.02160412e-01 7.87897885e-01 -1.10958064e+00 -3.34400195e-03 7.48554692e-02 -1.21480875e-01 -8.85067999e-01 -4.08435792e-01 -5.25694251e-01 -7.99932718e-01 -9.56993103e-02 1.57929674e-01 -1.06858887e-01 6.93613589e-01 7.57665753e-01 5.58495045e-01 8.11744511e-01 8.86663973e-01 -1.50325680e+00 -8.33350234e-03 -4.42190945e-01 -2.82571137e-01 2.98676342e-01 4.68253702e-01 -3.84497344e-01 -3.67140055e-01 5.59694588e-01]
[9.174022674560547, -3.14700984954834]
87820789-52e5-449b-a853-5899ae0da28b
end-to-end-learning-of-keypoint
2106.07995
null
https://arxiv.org/abs/2106.07995v3
https://arxiv.org/pdf/2106.07995v3.pdf
Learning of feature points without additional supervision improves reinforcement learning from images
In many control problems that include vision, optimal controls can be inferred from the location of the objects in the scene. This information can be represented using feature points, which is a list of spatial locations in learned feature maps of an input image. Previous works show that feature points learned using unsupervised pre-training or human supervision can provide good features for control tasks. In this paper, we show that it is possible to learn efficient feature point representations end-to-end, without the need for unsupervised pre-training, decoders, or additional losses. Our proposed architecture consists of a differentiable feature point extractor that feeds the coordinates of the estimated feature points directly to a soft actor-critic agent. The proposed algorithm yields performance competitive to the state-of-the art on DeepMind Control Suite tasks.
['Juho Kannala', 'Alexander Ilin', 'Rinu Boney']
2021-06-15
null
null
null
null
['unsupervised-pre-training']
['methodology']
[-6.98061660e-02 2.57795453e-01 -2.42050946e-01 -4.62476015e-01 -6.76156521e-01 -3.55136573e-01 7.82303631e-01 1.36149213e-01 -7.94101298e-01 5.01346648e-01 -3.48352492e-02 1.68809712e-01 -1.47336766e-01 -3.91631663e-01 -1.22190034e+00 -7.18808889e-01 -3.27173099e-02 5.79929471e-01 1.60963655e-01 9.63832811e-02 5.33233821e-01 6.44611776e-01 -1.27100384e+00 -6.62577450e-02 4.39662427e-01 1.26934791e+00 4.86386150e-01 8.58896732e-01 3.21688473e-01 6.75084352e-01 -4.25223470e-01 1.24299154e-01 4.88179058e-01 -1.79185867e-01 -4.57748413e-01 4.05378282e-01 5.32311022e-01 -5.16232312e-01 -3.80702823e-01 8.30831647e-01 2.91647404e-01 2.35119671e-01 8.02779377e-01 -1.18842399e+00 -5.76916754e-01 -1.43202916e-01 -4.91668701e-01 -1.01239935e-01 -5.34353368e-02 6.13342941e-01 9.63385701e-01 -8.52981031e-01 7.06533670e-01 1.32361031e+00 2.84179002e-01 4.56287712e-01 -1.36863256e+00 -1.94909960e-01 2.16271266e-01 2.84252111e-02 -8.95935059e-01 -6.54346704e-01 6.78389072e-01 -6.39503777e-01 1.27354658e+00 -4.06411529e-01 8.48897099e-01 7.88682342e-01 3.93226951e-01 7.11220622e-01 5.92786431e-01 -4.59365577e-01 4.59087700e-01 2.22760681e-02 -3.79500210e-01 1.04647791e+00 -1.01355031e-01 5.38915753e-01 -4.28306758e-01 7.19160587e-02 1.15595949e+00 1.64908290e-01 -1.64734453e-01 -1.01912165e+00 -1.11978567e+00 9.69260633e-01 9.10221696e-01 -1.35616466e-01 -5.81286132e-01 8.01531553e-01 1.34670556e-01 3.26007187e-01 3.21542352e-01 7.02602744e-01 -5.63827634e-01 -5.37183844e-02 -5.36549509e-01 3.32515419e-01 3.91495705e-01 1.00218225e+00 7.65482843e-01 -2.11688248e-03 -2.67908692e-01 4.58146542e-01 5.25823653e-01 4.28818703e-01 3.34913313e-01 -1.43015540e+00 5.39641976e-01 4.49984193e-01 4.06746745e-01 -8.75158250e-01 -2.64546126e-01 -2.61605591e-01 -4.88933414e-01 8.92506242e-01 3.70895594e-01 -4.11067635e-01 -1.12040508e+00 1.73859429e+00 3.62134874e-01 2.99993962e-01 -7.89009035e-02 1.13566327e+00 -1.85947329e-01 7.13829696e-01 -1.55754939e-01 4.96435165e-02 6.59298837e-01 -9.53658521e-01 -4.61951584e-01 -5.86218536e-01 2.19408348e-01 -3.86640042e-01 1.01750636e+00 2.52679557e-01 -1.06205344e+00 -5.08048296e-01 -9.14638579e-01 -1.26033977e-01 -2.49008507e-01 3.70259523e-01 4.15591300e-01 -1.26595497e-01 -1.00744522e+00 7.62855828e-01 -1.23185110e+00 -1.38066709e-01 6.80769920e-01 7.03431785e-01 -3.70849788e-01 3.11128825e-01 -4.37118053e-01 1.24714184e+00 1.62647262e-01 1.22473024e-01 -1.52758038e+00 -7.14379370e-01 -1.08325934e+00 1.39845997e-01 6.01742864e-01 -8.36952388e-01 1.45279050e+00 -1.06577885e+00 -1.68814933e+00 5.63532531e-01 -4.73933853e-02 -6.37350678e-01 5.00252187e-01 -6.25682414e-01 5.43349683e-01 2.24342018e-01 5.90010397e-02 9.94765460e-01 1.34845829e+00 -1.10581458e+00 -7.51412213e-01 -3.76131684e-01 1.03158176e-01 1.88753411e-01 -6.84056804e-03 -3.62501323e-01 -4.80306089e-01 -3.14273447e-01 -1.99922279e-01 -1.05382216e+00 -6.45887136e-01 7.39815474e-01 -6.37677848e-01 -1.73517480e-01 8.26363862e-01 -2.94397950e-01 4.04800475e-01 -2.15967059e+00 4.87172484e-01 2.27607816e-01 1.30056590e-01 1.69807523e-01 -1.76551685e-01 1.51771381e-01 3.21110748e-02 -2.71892726e-01 -7.46711940e-02 -4.23479050e-01 1.57892957e-01 2.07617953e-01 -3.26190442e-01 9.18654382e-01 6.33310854e-01 9.51440156e-01 -8.17533135e-01 -1.99640960e-01 6.90707326e-01 5.00294447e-01 -7.29988337e-01 4.58729416e-01 -8.11313987e-01 5.65881550e-01 -6.38174534e-01 6.77395687e-02 5.77817578e-03 -1.52370647e-01 -2.27526069e-01 1.20938301e-01 -1.76618055e-01 2.14273036e-01 -9.33183074e-01 1.95279229e+00 -5.66065073e-01 8.03641617e-01 -7.62242079e-03 -1.20663774e+00 8.35795462e-01 1.65045023e-01 3.70645016e-01 -4.16430861e-01 2.27207094e-01 -1.09072082e-01 -5.96441105e-02 -3.53127688e-01 -8.89965370e-02 1.33126587e-01 -9.80815366e-02 -7.92560652e-02 3.65144879e-01 -3.51785570e-01 1.01751231e-01 -1.47671118e-01 1.07996607e+00 2.81096876e-01 2.34010622e-01 2.58177333e-02 4.08050388e-01 1.08883202e-01 3.97565961e-01 5.85261822e-01 1.10402472e-01 5.70253134e-01 8.60774875e-01 -3.83028060e-01 -1.30448306e+00 -9.01933372e-01 2.45708987e-01 8.06997776e-01 -1.71104252e-01 -2.51303047e-01 -5.04378915e-01 -5.66852808e-01 2.34978497e-01 5.62892973e-01 -6.56912506e-01 -3.01818669e-01 -5.47161222e-01 2.93409795e-01 3.07940468e-02 6.66797519e-01 2.82695204e-01 -1.06007016e+00 -1.14086401e+00 1.82863399e-01 7.22358704e-01 -9.70775664e-01 -7.51790345e-01 4.92434561e-01 -9.42584574e-01 -1.13376844e+00 -6.58502281e-01 -7.82196760e-01 1.00392580e+00 -2.64027148e-01 7.42555559e-01 -2.13495791e-01 -4.77417648e-01 4.68869984e-01 1.97863773e-01 -5.06568253e-01 7.06977174e-02 6.09416775e-02 -1.18319049e-01 -1.15619779e-01 -5.76440319e-02 -4.29211736e-01 -6.15874469e-01 8.91479850e-03 -3.14971417e-01 4.62417975e-02 7.23482788e-01 8.80236030e-01 9.58289623e-01 -3.47602338e-01 2.80935198e-01 -6.67107522e-01 3.79134864e-01 -4.29820940e-02 -1.40268052e+00 1.61086079e-02 -5.04037559e-01 6.50902152e-01 8.84292901e-01 -3.31748694e-01 -6.03667021e-01 9.21238720e-01 2.96473056e-01 -9.66322184e-01 -2.28912428e-01 3.19247782e-01 8.80107582e-02 -1.47640318e-01 4.56953973e-01 1.45587688e-02 3.55101258e-01 -4.67004329e-01 5.17578781e-01 2.04593420e-01 5.41420937e-01 -3.76357704e-01 6.57664835e-01 2.28182197e-01 2.96247751e-01 -6.92537010e-01 -8.45014095e-01 -3.56892616e-01 -8.53816271e-01 -1.29948020e-01 8.87368739e-01 -8.06669235e-01 -9.86754537e-01 1.09283768e-01 -1.20593512e+00 -6.30978882e-01 -5.50858021e-01 5.06696939e-01 -1.20964217e+00 -3.40900421e-01 -2.82603562e-01 -6.52649581e-01 -8.93765762e-02 -1.17106569e+00 1.22448003e+00 3.03431839e-01 4.05799523e-02 -9.48637426e-01 9.49676558e-02 -1.92668602e-01 3.10124606e-01 4.42120582e-01 8.49284053e-01 -2.83643186e-01 -8.44388425e-01 -3.90458763e-01 1.35293275e-01 2.95748502e-01 -1.66634563e-02 1.98857617e-02 -8.11186075e-01 -2.81516075e-01 -1.79909185e-01 -5.83589494e-01 8.97139728e-01 8.31426144e-01 1.28681970e+00 -5.43643057e-01 -4.15580362e-01 6.91720009e-01 1.48955655e+00 2.46798575e-01 3.65200669e-01 2.20942885e-01 6.58049524e-01 3.50516945e-01 5.40051579e-01 5.62850177e-01 1.14467055e-01 6.98192716e-01 8.54963124e-01 -7.21405670e-02 9.68496874e-02 -3.42698455e-01 3.99694651e-01 -3.47008407e-02 1.34244233e-01 -5.03931530e-02 -7.94787169e-01 7.18545794e-01 -2.16959810e+00 -6.94554508e-01 5.22473395e-01 2.09540987e+00 6.52981579e-01 1.18430480e-01 -1.48434583e-02 -1.35480061e-01 5.75699210e-01 1.87557265e-01 -1.14168298e+00 -3.00639033e-01 5.26974380e-01 1.23813890e-01 6.17359340e-01 7.08083808e-01 -1.25129390e+00 9.50453758e-01 6.11178303e+00 1.36780649e-01 -1.22018266e+00 -2.52319872e-01 4.83425707e-01 -5.31279981e-01 4.57140803e-01 -3.58410999e-02 -5.62488794e-01 1.93595499e-01 7.34613597e-01 -1.72094256e-01 6.20520592e-01 1.20968044e+00 3.86179000e-01 -3.47391888e-02 -1.58089602e+00 8.78358364e-01 -1.13759525e-01 -1.34753036e+00 -3.00655365e-01 1.41639501e-01 5.92627347e-01 2.45904729e-01 3.28528792e-01 1.31867766e-01 4.87717062e-01 -9.28748786e-01 6.67827487e-01 7.47594178e-01 5.98101974e-01 -8.05697620e-01 3.02295536e-01 5.02631009e-01 -6.17681503e-01 -3.50875109e-01 -5.67463875e-01 1.33149782e-02 9.96840000e-02 2.00289339e-01 -1.06417680e+00 -2.32038185e-01 3.37364078e-01 1.01033282e+00 -3.59473348e-01 1.30095196e+00 -5.40485740e-01 2.39864767e-01 -4.13315654e-01 -1.26257792e-01 6.14827335e-01 2.39871684e-02 4.25703168e-01 7.79693782e-01 1.51383549e-01 -1.77810192e-01 3.08133692e-01 9.71562803e-01 5.81214810e-03 -4.83764172e-01 -8.58209670e-01 -4.55852970e-02 -4.17245179e-02 1.02533460e+00 -3.37285548e-01 -1.60567254e-01 -2.92222828e-01 9.14323330e-01 7.29553521e-01 2.82681644e-01 -5.01926184e-01 -4.78711694e-01 8.73587310e-01 1.21407971e-01 7.31758833e-01 -5.78628778e-01 -7.64201730e-02 -8.65687072e-01 -9.59414169e-02 -4.57076848e-01 -1.06611708e-02 -8.26712310e-01 -9.75793362e-01 1.16537802e-01 -1.84867322e-01 -9.96629298e-01 -7.63632953e-01 -8.78412426e-01 -6.29372239e-01 6.97733939e-01 -1.32018590e+00 -6.21262193e-01 -1.15046114e-01 6.24964654e-01 6.67714834e-01 -2.73143768e-01 7.00577915e-01 -2.31989503e-01 -4.17129099e-01 2.48263747e-01 2.17578694e-01 2.10846201e-01 4.13963497e-01 -1.58065403e+00 3.52325380e-01 5.93907893e-01 2.98181683e-01 3.19483995e-01 7.03736961e-01 -2.97534883e-01 -1.66231823e+00 -1.02573299e+00 5.33784866e-01 -3.49734992e-01 4.68391329e-01 -3.29693407e-01 -4.17207330e-01 8.54427695e-01 3.49721789e-01 4.38747644e-01 -1.66248962e-01 -6.75998777e-02 7.71520436e-02 -3.43166769e-01 -1.02939177e+00 5.48089325e-01 7.03044951e-01 -3.63868952e-01 -4.48166937e-01 4.69672590e-01 6.63356245e-01 -6.38045728e-01 -4.72986013e-01 -1.30311772e-01 3.56023073e-01 -4.88733739e-01 9.82503235e-01 -8.95232022e-01 6.34762049e-01 -2.57341653e-01 7.40902275e-02 -1.75227249e+00 -3.54886830e-01 -5.60451269e-01 -2.90223449e-01 6.05871439e-01 5.32504261e-01 -2.79593915e-01 7.57058918e-01 8.52811337e-01 -2.18474403e-01 -8.35920811e-01 -1.10372245e+00 -3.92248154e-01 -1.95180282e-01 -5.55504337e-02 1.53450653e-01 2.93117136e-01 -7.59638473e-02 5.49264014e-01 -3.08616996e-01 3.08455706e-01 6.46859229e-01 1.13237768e-01 6.75383508e-01 -1.04037511e+00 -2.31595710e-01 -3.54472160e-01 -5.85882962e-01 -1.11773705e+00 5.18357873e-01 -7.25690126e-01 4.44829822e-01 -1.57575691e+00 -3.02601784e-01 -1.96546555e-01 -1.76142156e-01 7.88650036e-01 2.73270905e-01 -3.68263304e-01 4.59639192e-01 -1.16429351e-01 -7.51980245e-01 8.63459826e-01 1.47561586e+00 -2.94345707e-01 -2.74076104e-01 3.61350812e-02 -3.17118078e-01 7.39633143e-01 7.62068272e-01 -5.80340147e-01 -4.63891864e-01 -8.60165179e-01 -9.78490934e-02 3.85185331e-01 6.11963749e-01 -1.02495396e+00 4.74093825e-01 -3.69975984e-01 7.36147642e-01 -2.98395336e-01 7.11820900e-01 -1.21165025e+00 -4.32683587e-01 5.97336233e-01 -7.38802195e-01 5.77220321e-02 -1.84929352e-02 7.13372290e-01 -3.21136117e-02 -2.94615716e-01 7.88904846e-01 -1.98872656e-01 -6.07340038e-01 4.38213855e-01 -2.86105067e-01 -3.21022645e-02 1.31470609e+00 2.45851770e-01 -3.21346484e-02 -4.07955915e-01 -6.37174428e-01 4.84764040e-01 3.32188845e-01 3.52730542e-01 8.04843128e-01 -1.13570881e+00 -3.32853585e-01 3.07691872e-01 -2.05395311e-01 3.69363844e-01 -4.47681814e-01 4.72971231e-01 -3.72291327e-01 4.97717112e-01 -3.00673664e-01 -6.94806993e-01 -6.97628796e-01 7.66718507e-01 5.86094975e-01 3.76077406e-02 -7.91141570e-01 5.99522650e-01 5.94870411e-02 -1.46101207e-01 4.70761061e-01 -6.08742177e-01 -6.23743981e-02 -2.70057350e-01 3.40156823e-01 6.10983819e-02 -2.80852050e-01 -2.80722171e-01 -1.02587640e-01 7.38782167e-01 -9.97259095e-02 -3.52053225e-01 1.68274415e+00 1.35026440e-01 3.25563461e-01 5.06656766e-01 1.28778756e+00 -6.78596199e-01 -2.17387366e+00 -1.39613107e-01 7.60229453e-02 -6.46362185e-01 4.18246090e-01 -6.78733289e-01 -1.39403379e+00 9.45170820e-01 7.01015711e-01 -2.74630040e-01 8.72088790e-01 -9.33618564e-03 3.15543622e-01 7.75256932e-01 2.87494034e-01 -1.07213902e+00 3.80308390e-01 5.52927494e-01 1.15285480e+00 -1.27050233e+00 -2.54245490e-01 6.94304258e-02 -8.43665540e-01 9.61252093e-01 8.15066755e-01 -9.77440894e-01 7.46423781e-01 3.44468147e-01 -1.21580914e-01 -2.07449317e-01 -1.30518436e+00 -2.69000381e-01 3.92553926e-01 7.58774579e-01 9.59441811e-02 -3.63758028e-01 2.08387688e-01 1.57522455e-01 1.36743560e-02 -5.93618257e-03 2.38943353e-01 9.54920828e-01 -7.36466229e-01 -7.65263975e-01 2.18052976e-02 5.45168161e-01 -1.74155727e-01 2.23444521e-01 -3.22355479e-01 5.74375868e-01 -1.14901505e-01 6.03272855e-01 3.94122124e-01 3.64565328e-02 5.15278220e-01 -6.83004037e-02 6.74841881e-01 -7.84511507e-01 -3.40900779e-01 1.23197697e-01 -2.90858865e-01 -9.77701366e-01 -1.34228453e-01 -5.67927420e-01 -1.36781204e+00 2.08347082e-01 -2.23273620e-01 -1.83738455e-01 8.24539959e-01 9.53918815e-01 4.65980858e-01 5.44579327e-01 6.27483130e-01 -1.32768416e+00 -8.17249358e-01 -7.43627012e-01 -3.48750979e-01 7.95491040e-02 8.80166531e-01 -7.77464926e-01 -1.45957947e-01 2.82839447e-01]
[4.731792449951172, 0.8028873801231384]
1d4d19a1-1fe5-4329-9a00-927d34de8efd
subsampling-for-knowledge-graph-embedding
2209.12801
null
https://arxiv.org/abs/2209.12801v1
https://arxiv.org/pdf/2209.12801v1.pdf
Subsampling for Knowledge Graph Embedding Explained
In this article, we explain the recent advance of subsampling methods in knowledge graph embedding (KGE) starting from the original one used in word2vec.
['Katsuhiko Hayashi', 'Hidetaka Kamigaito']
2022-09-13
null
null
null
null
['knowledge-graph-embedding']
['graphs']
[-3.33686829e-01 2.88354099e-01 -4.15475160e-01 1.32089198e-01 2.45503336e-01 -3.00254136e-01 7.17889905e-01 9.04959962e-02 -5.67006826e-01 8.46697211e-01 8.29421759e-01 -5.51961064e-01 -3.72150183e-01 -1.29128671e+00 -2.73617804e-01 -3.09201777e-01 -2.27595121e-01 2.09428251e-01 1.84443593e-01 -7.81724274e-01 1.14602171e-01 4.24087167e-01 -1.20502496e+00 -1.77046489e-02 2.89550185e-01 3.43658417e-01 -2.91849196e-01 8.74399543e-01 -8.44711363e-01 8.55397820e-01 -6.53636336e-01 -1.02060103e+00 -2.48127803e-01 -2.23583341e-01 -1.02172720e+00 -2.98752785e-01 2.06398547e-01 4.82153557e-02 -1.32577944e+00 9.80347455e-01 5.61689734e-01 4.71460909e-01 8.95041406e-01 -1.23501992e+00 -1.38513219e+00 9.12343681e-01 -9.92049426e-02 3.53285730e-01 3.53318036e-01 -5.47304153e-01 1.26915193e+00 -8.82475138e-01 1.02544153e+00 1.36274469e+00 9.14554119e-01 5.16107440e-01 -9.11906719e-01 -2.22979054e-01 2.97309328e-02 1.10500240e+00 -1.73727751e+00 1.55832514e-01 1.02865982e+00 -6.05271757e-01 1.42497277e+00 5.70881516e-02 1.12865710e+00 1.29835260e+00 -1.16153397e-01 7.09304333e-01 7.51600087e-01 -9.09963489e-01 9.78395194e-02 3.78996730e-01 6.75507843e-01 8.92527103e-01 1.11706448e+00 3.60976756e-02 -5.21872640e-01 -5.53574622e-01 7.81855583e-01 -3.74466449e-01 -4.41656828e-01 -6.90274298e-01 -1.05777204e+00 1.48205674e+00 3.26240689e-01 8.48309994e-01 -2.95365572e-01 5.85437059e-01 6.70812309e-01 6.61686242e-01 5.60037315e-01 5.45898199e-01 -3.12504947e-01 1.89344734e-02 -5.86717546e-01 6.94971234e-02 1.09333014e+00 9.83902514e-01 6.40187442e-01 6.58489585e-01 -3.01823467e-01 5.09270549e-01 2.66965628e-01 3.39807957e-01 5.46252191e-01 -1.04703642e-01 -2.49409959e-01 3.04009587e-01 -1.72516391e-01 -1.40629494e+00 8.50014389e-02 -3.92834902e-01 -4.81852591e-01 -2.79226899e-01 -2.03490689e-01 1.73613191e-01 -6.46092057e-01 1.03743339e+00 2.05093294e-01 4.40716863e-01 1.86650649e-01 4.54711765e-01 1.16253352e+00 3.85723710e-01 1.40648201e-01 1.43736959e-01 1.37642121e+00 -1.09125364e+00 -1.30767524e+00 3.47934544e-01 8.02976191e-01 -3.34422052e-01 5.66049099e-01 1.79255709e-01 -3.96173477e-01 -4.17115778e-01 -1.13014293e+00 -5.79258613e-02 -1.56714463e+00 -3.39736044e-01 1.31531966e+00 1.11782968e+00 -1.31351757e+00 5.78847885e-01 -4.60434377e-01 -6.30976200e-01 3.13257307e-01 2.83436030e-02 -6.14466488e-01 -1.91984072e-01 -1.98749518e+00 1.23394036e+00 9.94387448e-01 -3.23128641e-01 -3.00532669e-01 -4.18828845e-01 -1.13596582e+00 -1.87638372e-01 4.95296359e-01 -1.02999520e+00 4.52727586e-01 -2.05716103e-01 -1.63465595e+00 5.52631557e-01 2.64580734e-02 -5.05877674e-01 -6.08055033e-02 -2.97237158e-01 -8.95153701e-01 -4.50991616e-02 -6.27679110e-01 1.16777979e-01 8.48525882e-01 -1.09874547e+00 4.08191793e-02 -2.52821892e-01 9.75187942e-02 -2.64170527e-01 -8.17301691e-01 -2.38926530e-01 -5.44980645e-01 -7.46802866e-01 -5.95474184e-01 -4.27464396e-01 -1.52438551e-01 -4.25986022e-01 -3.51392031e-01 -5.19140184e-01 6.48205161e-01 -6.96930707e-01 1.84948730e+00 -1.87658978e+00 6.88351333e-01 3.68115842e-01 5.88455617e-01 5.21103859e-01 -2.13883042e-01 1.18309486e+00 -2.43723258e-01 2.99133390e-01 1.92381024e-01 -3.01938444e-01 4.66273695e-01 8.58228922e-01 -4.00353819e-01 3.97922158e-01 -3.22923779e-01 1.61764371e+00 -1.28892434e+00 -4.73817378e-01 2.79410005e-01 9.70573068e-01 -2.92745978e-01 -2.32924134e-01 2.09105372e-01 -4.37709063e-01 -5.54562509e-01 4.47565198e-01 4.70064431e-01 -1.38652340e-01 3.17945153e-01 -4.06961054e-01 4.27024104e-02 -3.16453986e-02 -1.21315420e+00 1.61832368e+00 -2.88175255e-01 7.70567477e-01 -3.28577816e-01 -9.16012347e-01 6.96135402e-01 4.04113829e-01 3.43001872e-01 -7.77098015e-02 2.33956143e-01 -5.44016063e-02 -3.16717863e-01 -4.79364961e-01 1.04632330e+00 -1.18072607e-01 1.90463230e-01 3.42021435e-01 7.84982622e-01 -1.14301130e-01 9.66398120e-02 5.42339981e-01 9.66962695e-01 -1.71497345e-01 7.85024285e-01 -3.90571594e-01 4.09616649e-01 -1.79166961e-02 -2.83752263e-01 8.38678122e-01 -1.06732823e-01 4.64200266e-02 2.35967219e-01 -4.87754703e-01 -9.70478296e-01 -9.36812818e-01 -1.94281582e-02 6.77014709e-01 -3.23774904e-01 -1.08838892e+00 -1.50586128e-01 -1.23645711e+00 6.36049747e-01 1.02270961e+00 -1.03135884e+00 -4.22552735e-01 -1.48224846e-01 -3.96290392e-01 8.65784883e-01 5.71677268e-01 -2.55584389e-01 -1.01836896e+00 1.96929321e-01 1.63893342e-01 3.95564526e-01 -6.68473125e-01 -1.19210102e-01 -2.58298479e-02 -6.96741045e-01 -9.56084073e-01 -6.16627097e-01 -5.96817195e-01 3.70302975e-01 2.99687386e-01 1.36848116e+00 6.46052957e-02 -4.10088509e-01 1.03560531e+00 -1.05654299e+00 -5.87403774e-01 -4.59472835e-02 1.32393718e-01 2.19803900e-01 -3.04654300e-01 9.51556027e-01 -5.51910222e-01 -8.13105926e-02 -5.44290662e-01 -1.13413012e+00 -5.86844265e-01 2.45482132e-01 9.04561520e-01 3.56291652e-01 4.24392194e-01 4.83366698e-01 -1.11446738e+00 1.14423203e+00 -6.36362970e-01 -2.08652407e-01 5.28585970e-01 -1.20533848e+00 3.12959820e-01 3.22269559e-01 -3.37151229e-01 -1.67667747e-01 -7.65531540e-01 -3.27497751e-01 -7.62127399e-01 2.58429796e-01 1.06986785e+00 4.47738886e-01 -6.93732619e-01 2.90705919e-01 3.11588824e-01 -1.89809680e-01 -5.97640395e-01 1.36025333e+00 3.82354081e-01 -8.68476331e-02 -1.56373400e-02 9.77674603e-01 4.13428634e-01 -2.26012722e-01 -1.12498569e+00 -2.00131670e-01 -5.40706754e-01 -6.73053145e-01 -1.02070570e-01 6.97139263e-01 -3.64169359e-01 -5.76146722e-01 9.06779489e-04 -1.12427938e+00 2.04269439e-01 -8.23075533e-01 9.02067721e-01 -9.82706100e-02 6.81097627e-01 -1.81881234e-01 -7.53763139e-01 -2.97833979e-01 -5.06939173e-01 5.83439350e-01 -2.41492391e-01 1.23438768e-01 -1.96719456e+00 7.93971896e-01 -1.96450353e-02 7.90468335e-01 1.06770314e-01 9.41946268e-01 -7.97692716e-01 1.31334782e-01 -3.88007939e-01 -2.63106339e-02 4.68580604e-01 2.51664668e-01 -4.83361185e-02 -7.26372480e-01 -3.33315581e-01 -3.14832121e-01 1.86517403e-01 1.28128815e+00 1.86576724e-01 6.46065474e-01 -2.03943476e-01 -6.03491187e-01 6.56361699e-01 2.13419747e+00 -3.84600967e-01 5.12402356e-01 4.93005306e-01 1.07874191e+00 3.89569968e-01 2.21676938e-02 1.54666483e-01 4.87417936e-01 5.40858626e-01 -5.61865568e-02 1.25300497e-01 -3.41883481e-01 -5.00050843e-01 2.09939674e-01 1.72643375e+00 -3.27512652e-01 -2.44996458e-01 -7.53688455e-01 1.05414653e+00 -1.62993109e+00 -8.50641131e-01 -1.58404067e-01 1.67581058e+00 6.97828054e-01 -4.13253784e-01 5.21674380e-03 1.20680161e-01 6.23734057e-01 6.90099597e-01 1.31814286e-01 -8.20549905e-01 -2.73016989e-01 7.43983865e-01 9.60097134e-01 9.59649563e-01 -6.54942155e-01 1.34087086e+00 7.61493206e+00 1.07078183e+00 -4.72074628e-01 3.76602173e-01 -5.99169910e-01 3.84469151e-01 -9.23715532e-01 1.43718719e-01 -6.75652206e-01 -1.04558468e-01 8.73847842e-01 -6.02714360e-01 5.31936467e-01 8.86216044e-01 -6.55951619e-01 5.62818050e-01 -7.12912083e-01 1.09104455e+00 5.19794643e-01 -1.40710902e+00 5.15164435e-01 -4.08125930e-02 6.20072305e-01 -1.18325632e-02 -2.56335884e-01 5.92938840e-01 4.59673196e-01 -1.02810359e+00 -2.08254218e-01 4.70720291e-01 6.68825865e-01 -8.19602966e-01 8.69531155e-01 -6.26065552e-01 -1.36051309e+00 4.42433804e-01 -7.71104515e-01 9.07633454e-02 4.36507344e-01 7.39257038e-01 -9.20349002e-01 1.30761266e+00 2.27619261e-01 6.77490115e-01 -5.44032574e-01 4.79457855e-01 -5.43277979e-01 8.00158501e-01 5.44629693e-02 -5.25595725e-01 1.91017717e-01 -3.62472385e-01 9.23372746e-01 1.60766542e+00 8.56237561e-02 -1.67610615e-01 -4.55009490e-01 5.11167169e-01 1.60447434e-01 3.72905135e-01 -1.32691979e+00 -7.15622306e-01 2.66937613e-01 1.02660704e+00 -3.74978930e-01 -4.81846184e-01 -7.65052676e-01 1.07887900e+00 5.50216794e-01 4.98484880e-01 -6.44675791e-01 -9.39553082e-01 6.64390564e-01 -2.96681702e-01 1.04186547e+00 -4.78490502e-01 4.35751140e-01 -1.39147866e+00 -3.07286441e-01 -4.86290991e-01 5.32872140e-01 -4.22745317e-01 -1.88495147e+00 6.82996333e-01 1.38012469e-01 -5.63531101e-01 8.92344788e-02 -8.83857071e-01 -2.51059324e-01 7.85220563e-01 -1.89357936e+00 -1.13272023e+00 1.95181832e-01 8.30236733e-01 -1.69807568e-01 -2.31585890e-01 1.33772457e+00 5.23368597e-01 -1.77239493e-01 7.58872390e-01 6.58400238e-01 3.18485767e-01 3.66748959e-01 -1.54648113e+00 8.63328874e-01 3.26229155e-01 8.22104931e-01 8.57433796e-01 8.14990520e-01 -8.66116226e-01 -1.91992009e+00 -1.15508568e+00 1.38630462e+00 -4.95723784e-01 1.24452198e+00 -3.11618716e-01 -7.65818119e-01 9.91826773e-01 4.17806238e-01 6.55823201e-02 8.87548804e-01 5.76483428e-01 -7.20344067e-01 3.13346982e-01 -8.50269735e-01 7.91080713e-01 1.03789103e+00 -8.40138018e-01 -1.24961686e+00 2.07842186e-01 1.01783645e+00 3.88658643e-01 -1.48976624e+00 3.17111135e-01 6.35291159e-01 -2.23645389e-01 1.53774953e+00 -1.29065776e+00 -3.27584893e-01 -1.15144245e-01 -3.52474600e-01 -1.71263981e+00 -7.35437214e-01 -5.36041379e-01 -9.00928259e-01 7.29796827e-01 6.17725924e-02 -1.13725090e+00 7.23890960e-01 -2.10107356e-01 1.26398101e-01 -8.90153050e-01 -9.66576636e-01 -1.13328528e+00 1.03086643e-01 -5.05302608e-01 5.30913413e-01 1.51204967e+00 4.92231607e-01 3.37363183e-01 -4.46344614e-01 -2.02662081e-01 7.25292861e-01 -2.12051675e-01 4.89947498e-01 -1.23153067e+00 -4.19419736e-01 -4.68909681e-01 -1.34750521e+00 -7.03539252e-01 4.07650769e-01 -1.26006210e+00 -8.47769022e-01 -1.95252860e+00 -9.16353315e-02 3.61227334e-01 -7.44798541e-01 2.37368166e-01 -5.10442376e-01 1.78635478e-01 -1.73765659e-01 -3.31632853e-01 -4.86772209e-01 9.58552420e-01 9.32410359e-01 -3.80271733e-01 1.57012880e-01 -7.46398330e-01 -6.08148932e-01 1.92881450e-01 7.00158417e-01 -4.61635351e-01 -6.39970958e-01 -4.06064063e-01 5.76089025e-01 -4.35102522e-01 1.76298231e-01 -4.66727227e-01 1.59388170e-01 2.53847629e-01 -9.27541852e-02 -1.17734611e-01 2.10404038e-01 -7.49300480e-01 9.28569138e-02 4.86477911e-01 6.00296482e-02 1.30503774e-01 2.85641134e-01 1.06834686e+00 -5.03471553e-01 -3.93369943e-01 -3.67064252e-02 -1.58622697e-01 -1.19245148e+00 3.71424377e-01 -4.37237114e-01 6.88879983e-03 6.14163041e-01 -3.06383580e-01 -4.02830392e-01 -2.57821232e-01 -9.11658883e-01 -1.71891898e-01 1.81550711e-01 5.92030227e-01 1.07793558e+00 -1.79191864e+00 -7.01380730e-01 -2.71629691e-01 6.47076666e-01 -1.22568095e+00 2.18664065e-01 8.05461764e-01 -3.46646100e-01 8.04564953e-01 9.54187736e-02 4.82760370e-01 -9.24119890e-01 1.21346068e+00 -1.13266625e-01 -2.17919737e-01 -8.37247372e-01 1.12293208e+00 -4.91504788e-01 -3.52898091e-01 1.23357661e-01 1.10711336e-01 -9.82236505e-01 1.62862837e-01 3.91281545e-01 7.10095227e-01 -1.14268333e-01 -4.72037822e-01 -4.73023474e-01 6.00829065e-01 1.48499772e-01 -1.01091199e-01 1.39509237e+00 2.38783777e-01 -6.07879102e-01 1.13400877e+00 1.44798505e+00 3.57680857e-01 1.43203139e-01 -3.30746025e-01 -2.06402421e-01 -4.35273379e-01 3.81914437e-01 -2.36874610e-01 -9.70407367e-01 5.92147887e-01 3.66014093e-01 5.33470035e-01 5.37852407e-01 1.59706715e-02 7.66393483e-01 6.30436182e-01 6.52907014e-01 -1.07382631e+00 -7.05928087e-01 6.35544121e-01 5.59469759e-01 -7.08619833e-01 3.75682592e-01 -3.18902791e-01 -4.63214755e-01 1.24190140e+00 -1.18968964e-01 -3.18601608e-01 1.59030163e+00 -3.07091624e-01 -2.48565838e-01 -9.33375835e-01 -4.99493867e-01 -6.91584766e-01 4.56222653e-01 1.05418777e+00 3.99767607e-01 2.47262612e-01 -1.01603591e+00 5.18648922e-01 -1.48891166e-01 -9.89090353e-02 2.72128522e-01 8.65346134e-01 -1.39579907e-01 -1.38724339e+00 8.58009309e-02 4.10649747e-01 -1.38128415e-01 -6.48741245e-01 -6.97999716e-01 1.20754039e+00 -1.39232039e-01 7.41225183e-01 -4.75779206e-01 -8.61479282e-01 4.48185503e-01 4.09676403e-01 7.49386311e-01 -5.55064619e-01 -2.94208229e-01 -8.45186472e-01 4.02362585e-01 -4.07600164e-01 -5.98063767e-01 -1.93786159e-01 -5.32805622e-01 -4.70892400e-01 -8.25527132e-01 8.14853907e-01 8.06965411e-01 5.30639172e-01 4.26958323e-01 8.29503238e-01 2.75782108e-01 -3.51727962e-01 -3.05662394e-01 -1.30969954e+00 -1.29451537e+00 2.57576883e-01 -1.66308023e-02 -1.00257385e+00 -6.93299115e-01 -2.38325939e-01]
[8.806171417236328, 7.881773471832275]
8a4ca13c-67be-4996-a7dd-b9ee2b7dd246
improving-graph-based-sentence-ordering-with
2110.06446
null
https://arxiv.org/abs/2110.06446v1
https://arxiv.org/pdf/2110.06446v1.pdf
Improving Graph-based Sentence Ordering with Iteratively Predicted Pairwise Orderings
Dominant sentence ordering models can be classified into pairwise ordering models and set-to-sequence models. However, there is little attempt to combine these two types of models, which inituitively possess complementary advantages. In this paper, we propose a novel sentence ordering framework which introduces two classifiers to make better use of pairwise orderings for graph-based sentence ordering. Specially, given an initial sentence-entity graph, we first introduce a graph-based classifier to predict pairwise orderings between linked sentences. Then, in an iterative manner, based on the graph updated by previously predicted high-confident pairwise orderings, another classifier is used to predict the remaining uncertain pairwise orderings. At last, we adapt a GRN-based sentence ordering model on the basis of final graph. Experiments on five commonly-used datasets demonstrate the effectiveness and generality of our model. Particularly, when equipped with BERT and FHDecoder, our model achieves state-of-the-art performance.
['Jinsong Su', 'Degen Huang', 'Junfeng Yao', 'Jiali Zeng', 'Yubin Ge', 'Jie zhou', 'Fandong Meng', 'Ante Wang', 'Shaopeng Lai']
2021-10-13
null
https://aclanthology.org/2021.emnlp-main.186
https://aclanthology.org/2021.emnlp-main.186.pdf
emnlp-2021-11
['sentence-ordering']
['natural-language-processing']
[ 1.90620422e-01 2.14915007e-01 -1.52468726e-01 -8.10178816e-01 -5.02300680e-01 -5.36839843e-01 3.29656482e-01 5.42417467e-01 -1.25339836e-01 6.78758919e-01 2.58075804e-01 -2.89358944e-01 -4.11494911e-01 -8.73234868e-01 -4.93704438e-01 -4.39114392e-01 -2.30587214e-01 5.45398653e-01 3.98090005e-01 -3.38285208e-01 3.84451330e-01 6.64009526e-02 -1.24749863e+00 3.45955014e-01 1.12010789e+00 8.91385138e-01 1.93728969e-01 5.70430517e-01 -2.46053651e-01 9.10915136e-01 -3.62084955e-01 -6.42825663e-01 1.04379751e-01 -6.39270306e-01 -8.01858187e-01 1.14579007e-01 7.64506906e-02 -3.11529875e-01 -3.86933297e-01 1.07065117e+00 2.03374878e-01 -4.04450074e-02 5.76837242e-01 -9.22578752e-01 -5.20262897e-01 1.13703370e+00 -3.07699919e-01 1.07368696e-02 6.86439097e-01 -3.33613038e-01 1.54624176e+00 -5.56189060e-01 5.69593728e-01 1.07948041e+00 5.13906300e-01 1.33436307e-01 -8.65112901e-01 -3.22712362e-01 5.46692371e-01 4.76951063e-01 -1.28306854e+00 -1.95224047e-01 1.11926007e+00 -7.18116537e-02 8.20295036e-01 4.43332940e-01 5.90350628e-01 5.78752995e-01 3.84130359e-01 8.74386549e-01 9.23150063e-01 -5.15087306e-01 2.08400220e-01 -1.28789827e-01 5.34263849e-01 8.31028759e-01 1.41339347e-01 -4.30873394e-01 -5.74087143e-01 -3.04555476e-01 5.59751829e-03 -1.53231248e-01 -3.86576355e-01 -3.44207138e-01 -1.04213119e+00 5.38249969e-01 4.71876413e-01 4.49587584e-01 -1.92005873e-01 -4.19681251e-01 3.28726858e-01 2.73943752e-01 6.98409617e-01 2.55266845e-01 -4.37846631e-01 1.32145360e-01 -9.34827089e-01 -1.43625066e-02 9.77506340e-01 9.96052325e-01 7.20653474e-01 -6.41072392e-01 -2.59192169e-01 8.45331311e-01 7.10526586e-01 -2.53216475e-02 2.49273255e-01 -3.51481467e-01 6.29399419e-01 7.30101585e-01 -3.04039210e-01 -1.47187614e+00 -3.84284228e-01 -3.65091294e-01 -1.00815463e+00 -8.18341911e-01 -3.59236114e-02 1.24529719e-01 -5.82051337e-01 1.45488167e+00 3.60377133e-01 3.72741044e-01 6.32573571e-03 7.39654839e-01 8.78754258e-01 4.63486254e-01 -2.70143807e-01 -4.93048966e-01 1.12410533e+00 -7.93424308e-01 -6.03588641e-01 -3.39419581e-02 6.27846181e-01 -4.04535562e-01 5.16897678e-01 3.13124239e-01 -8.27085555e-01 -4.85741287e-01 -1.17560124e+00 8.71974882e-03 -7.20366687e-02 3.24668586e-02 5.12046874e-01 5.27932823e-01 -1.14577794e+00 8.13630164e-01 -6.79867327e-01 -2.70724565e-01 -2.70964839e-02 4.25239414e-01 -1.51369676e-01 -4.84446995e-02 -1.65637803e+00 6.42954469e-01 7.80728579e-01 4.56438184e-01 -3.39463443e-01 -2.13593245e-01 -9.00313377e-01 8.97511393e-02 3.65361929e-01 -8.50676239e-01 9.18876112e-01 -3.50372612e-01 -1.38144076e+00 3.73333514e-01 -3.83979380e-01 -6.45338535e-01 4.83701348e-01 6.42491877e-02 -4.29790884e-01 2.06389606e-01 1.81342721e-01 2.55085796e-01 4.49085325e-01 -1.20602155e+00 -5.45037270e-01 -3.43245983e-01 4.05459970e-01 6.42009318e-01 -6.49128199e-01 -1.66592479e-01 -6.16862297e-01 -2.75024742e-01 6.09128773e-01 -6.85777366e-01 -2.95692563e-01 -4.27451968e-01 -8.61901760e-01 -4.80631769e-01 4.17546630e-01 -5.89077175e-01 1.58136714e+00 -2.01566219e+00 3.16756845e-01 3.40826631e-01 3.93748939e-01 -1.24249615e-01 1.26975114e-02 7.12176085e-01 1.52574167e-01 4.22945172e-01 -4.26167518e-01 -5.26922584e-01 6.27090782e-02 3.79048139e-01 -1.08058445e-01 1.63259029e-01 8.79872292e-02 4.34638590e-01 -1.09467554e+00 -7.50223637e-01 7.35628828e-02 4.77968492e-02 -4.25425947e-01 2.85163641e-01 -1.98204666e-01 3.48826349e-01 -6.06670797e-01 3.33204061e-01 6.30511642e-01 -2.72778481e-01 9.16544318e-01 -2.55407631e-01 9.64518562e-02 4.48728085e-01 -1.03931165e+00 1.58992207e+00 -1.21241391e-01 1.69335410e-01 -4.15723592e-01 -1.10142934e+00 1.13324821e+00 8.04110765e-02 4.24761921e-01 -2.57753760e-01 6.82865977e-02 2.10454613e-01 7.65266046e-02 -4.12022352e-01 6.52644992e-01 7.91219771e-02 -1.75422236e-01 1.72729492e-01 5.81693389e-02 -2.86311172e-02 5.58830261e-01 5.99851787e-01 1.14654577e+00 -2.40485687e-02 5.43700814e-01 -5.17336093e-02 8.35238218e-01 -3.56850475e-01 6.96049213e-01 8.35774660e-01 6.11004606e-03 7.59187818e-01 7.82963395e-01 -4.60120231e-01 -5.87530732e-01 -8.27085137e-01 -1.07416719e-01 7.45771587e-01 4.05339211e-01 -8.26224685e-01 -5.86634696e-01 -1.18893147e+00 -1.32603824e-01 6.47389412e-01 -4.08984005e-01 -2.04595942e-02 -4.25390273e-01 -7.60339260e-01 2.36736923e-01 2.47097924e-01 7.42279232e-01 -7.26495564e-01 2.19588488e-01 2.54457742e-01 -4.52523589e-01 -1.11745036e+00 -5.24057806e-01 4.36928645e-02 -7.77971923e-01 -1.12542617e+00 -2.80781925e-01 -8.42423439e-01 7.29305983e-01 2.51172394e-01 1.03993058e+00 4.55655932e-01 4.43476081e-01 7.23899305e-02 -8.26464415e-01 8.31650421e-02 -3.38522851e-01 2.45109603e-01 1.18178643e-01 2.85941422e-01 2.55602181e-01 -5.99248290e-01 -3.64444017e-01 9.56108123e-02 -7.08417714e-01 -1.44415246e-02 6.54183149e-01 8.67942989e-01 4.39883322e-01 2.63695627e-01 5.42134047e-01 -1.22520244e+00 8.50019813e-01 -4.73555326e-01 -2.76291162e-01 6.69254482e-01 -6.63980663e-01 1.19008005e-01 8.95738661e-01 1.40503556e-01 -1.08382452e+00 5.16756997e-02 -2.31273025e-01 -1.33995429e-01 -5.63957058e-02 1.08217168e+00 -3.72095317e-01 1.38666138e-01 1.42760083e-01 3.81623209e-01 -2.82497078e-01 -2.74217010e-01 3.24920714e-01 9.26434994e-01 3.92892540e-01 -4.11995351e-01 6.58715010e-01 1.37599200e-01 -6.13429844e-02 -5.95727324e-01 -1.01534557e+00 -5.39209068e-01 -1.02766812e+00 -2.44766161e-01 7.05314815e-01 -6.25284791e-01 -6.51579738e-01 3.34937036e-01 -1.42155540e+00 3.02446544e-01 2.47350693e-01 3.80254537e-01 -2.25833938e-01 1.15714502e+00 -7.20969856e-01 -9.31858242e-01 -2.61875927e-01 -9.17882025e-01 9.72298265e-01 1.64409265e-01 -1.07553266e-01 -1.01614571e+00 8.64865854e-02 3.87008011e-01 -1.12820551e-01 1.09180408e-02 9.55786824e-01 -1.18462026e+00 -5.45030951e-01 -3.04328591e-01 -1.15922779e-01 3.83283138e-01 2.14326710e-01 3.95879075e-02 -5.29552639e-01 -1.51150569e-01 5.16530536e-02 3.51889208e-02 9.02136505e-01 1.76459059e-01 1.11012435e+00 -3.16950589e-01 -5.25835633e-01 3.10249716e-01 1.19182408e+00 1.63499147e-01 4.73753333e-01 1.24637276e-01 7.15615749e-01 5.28140843e-01 7.41410851e-01 5.45513511e-01 8.39025021e-01 4.42543030e-01 4.24137235e-01 3.91678095e-01 2.88267225e-01 -4.49274600e-01 1.57056704e-01 1.56798410e+00 3.94992493e-02 -6.26888454e-01 -6.57005012e-01 3.13199639e-01 -2.09626794e+00 -8.74698877e-01 -3.59080017e-01 2.01358604e+00 6.89524174e-01 4.61217076e-01 -1.87157929e-01 2.04508454e-01 9.37155128e-01 4.45298254e-01 -2.49321043e-01 -1.69798285e-01 -4.54395898e-02 -3.75280797e-01 1.72140479e-01 5.78641474e-01 -1.19816494e+00 6.42969608e-01 6.11666393e+00 6.11008108e-01 -4.84993786e-01 -3.10474575e-01 6.97078943e-01 4.59572762e-01 -7.24534631e-01 2.38483697e-01 -9.21271563e-01 7.43697345e-01 6.70316637e-01 -1.77294627e-01 2.22590804e-01 6.06163204e-01 2.32219249e-02 1.26221389e-01 -1.05145264e+00 6.13180101e-01 4.58455533e-01 -1.14492118e+00 1.08615771e-01 -2.49727011e-01 4.35489237e-01 -2.17995182e-01 -4.21749890e-01 3.39034528e-01 3.14797699e-01 -3.84042144e-01 4.48527515e-01 5.92596710e-01 2.37675488e-01 -7.73798883e-01 1.08138394e+00 5.92024148e-01 -1.42705941e+00 -1.18333260e-02 -4.56787735e-01 -2.17987657e-01 3.61797810e-01 1.14959466e+00 -9.79676127e-01 1.56172574e+00 5.34876347e-01 1.19878793e+00 -6.33837402e-01 1.01336527e+00 -5.28004110e-01 4.83364880e-01 -1.35114968e-01 -5.20714045e-01 1.94594357e-02 -2.95089453e-01 5.39625466e-01 1.09356582e+00 1.42523587e-01 3.50739390e-01 6.14045441e-01 2.92523742e-01 -1.84052721e-01 1.51532799e-01 -6.12033129e-01 2.99832314e-01 7.71772325e-01 1.28069520e+00 -6.82913959e-01 -3.05239499e-01 -3.09648037e-01 1.07968175e+00 6.64081991e-01 2.37587467e-02 -8.53833437e-01 -4.24686015e-01 -2.59547867e-02 -3.41660440e-01 1.96833476e-01 -3.20645243e-01 -1.71361074e-01 -1.45331776e+00 3.21772337e-01 -5.02486169e-01 4.90538776e-01 -6.24131262e-01 -1.61686516e+00 8.33983123e-01 9.53385308e-02 -1.42293346e+00 -3.44669431e-01 -3.31310809e-01 -5.41904807e-01 4.90713984e-01 -1.40705514e+00 -9.64145839e-01 -1.22458771e-01 1.21380612e-01 3.19993943e-01 6.05511367e-02 6.99184656e-01 1.14837863e-01 -5.02053499e-01 4.18345094e-01 1.06784850e-01 3.36620957e-01 5.64720571e-01 -1.44037592e+00 3.44633102e-01 9.27814066e-01 3.05183172e-01 6.33668721e-01 5.32037973e-01 -7.96947539e-01 -1.10043657e+00 -1.06093180e+00 1.38658392e+00 -2.50766098e-01 6.99225187e-01 -3.77471000e-01 -8.95894587e-01 6.31129503e-01 3.73747528e-01 -1.87839299e-01 7.46075869e-01 5.47000945e-01 -1.54278576e-01 -2.94098496e-01 -8.37620795e-01 5.85386932e-01 1.37958169e+00 -3.94534439e-01 -7.97172368e-01 6.96611106e-01 8.09356272e-01 -5.80598116e-01 -1.00712729e+00 5.94252884e-01 3.58798593e-01 -1.10725152e+00 6.81821465e-01 -3.42832297e-01 3.76164228e-01 -3.90474737e-01 -1.61856830e-01 -1.38267481e+00 -4.35678452e-01 -4.23744053e-01 1.26049863e-02 1.47631526e+00 5.43641925e-01 -8.21929097e-01 7.06098258e-01 2.80798435e-01 -4.05966431e-01 -9.64470685e-01 -8.04609179e-01 -7.68120766e-01 -4.87586886e-01 -2.09699944e-01 7.01839864e-01 9.22612786e-01 4.17036414e-01 6.27757251e-01 -3.31356049e-01 3.59439999e-01 5.78802049e-01 3.59894961e-01 4.03774977e-01 -1.14550781e+00 -5.47924757e-01 -9.36469659e-02 -5.14295399e-01 -1.46013033e+00 3.94415170e-01 -1.06263149e+00 3.73815417e-01 -1.72443044e+00 4.97401267e-01 -3.12706292e-01 -5.05308032e-01 3.34200621e-01 -5.02789080e-01 -1.53314233e-01 2.25352556e-01 3.03146631e-01 -1.14170778e+00 7.26564229e-01 1.02769828e+00 -3.43968630e-01 -7.12217763e-02 -1.96239147e-02 -7.54512787e-01 6.11221075e-01 7.30418324e-01 -4.04840708e-01 -5.90831995e-01 -5.06867170e-01 1.20122440e-01 2.26901039e-01 5.82748791e-03 -7.52809942e-01 7.17557728e-01 1.08538354e-02 1.02335677e-01 -8.95390093e-01 2.69170050e-02 -5.77339590e-01 6.50378466e-02 3.70511502e-01 -4.94317740e-01 -2.58043744e-02 -4.62643594e-01 9.13676381e-01 -4.62595075e-01 -4.49060708e-01 2.60006219e-01 -7.97630847e-02 -6.05354607e-01 2.82213718e-01 1.39310891e-02 -1.45221651e-01 8.05626512e-01 -7.97903985e-02 -2.43865341e-01 -3.90041262e-01 -7.61753023e-01 4.81733918e-01 1.68543056e-01 3.18297803e-01 7.09314167e-01 -1.18084145e+00 -7.41228938e-01 2.11958550e-02 9.38337669e-02 1.66264027e-02 3.38712305e-01 6.60650611e-01 -2.84887940e-01 3.62205267e-01 4.33184028e-01 -7.84265876e-01 -1.32368922e+00 4.60504174e-01 1.23733643e-03 -6.74296021e-01 -2.59433329e-01 7.42226422e-01 -1.15164861e-01 -6.46226048e-01 -4.78275344e-02 -2.85751969e-01 -4.94181931e-01 -1.09726479e-02 2.36210257e-01 2.17564404e-02 2.66624987e-01 -4.36748296e-01 -6.13575459e-01 4.01230127e-01 -3.26697409e-01 1.43440068e-01 1.15494716e+00 -2.69115716e-01 -6.46692395e-01 4.34035867e-01 1.06607902e+00 -9.31397174e-03 -8.84990990e-01 -4.74464223e-02 2.84386873e-01 -4.78425860e-01 -3.45598966e-01 -4.79669988e-01 -7.54474640e-01 4.66278344e-01 -1.74383804e-01 8.16191852e-01 1.19296038e+00 -1.09194383e-01 5.50915360e-01 3.20582151e-01 6.39726579e-01 -6.90519214e-01 -1.75565124e-01 7.15063751e-01 6.62238896e-01 -1.04254854e+00 1.68948680e-01 -9.43055034e-01 -5.95788360e-01 1.17752314e+00 5.00018120e-01 -3.75230685e-02 7.54660606e-01 -4.53275479e-02 -3.66413772e-01 5.53830601e-02 -8.41249287e-01 -1.22315697e-01 4.30832684e-01 3.16741705e-01 5.05602241e-01 1.48443982e-01 -7.80301213e-01 5.58392048e-01 -2.56048650e-01 -1.61494195e-01 4.55920517e-01 6.45992339e-01 -6.08297646e-01 -1.36826205e+00 8.28501806e-02 6.31773651e-01 -1.18428700e-01 -3.12480062e-01 -5.22738099e-01 4.86702502e-01 -1.43396882e-02 1.31276858e+00 -5.42053059e-02 -9.01903629e-01 2.32158974e-01 -1.69041261e-01 3.78065765e-01 -7.47401416e-01 -4.60800260e-01 -1.10411160e-01 5.19513667e-01 -1.65315881e-01 -4.73187506e-01 -8.06381762e-01 -1.09645236e+00 -2.45983765e-01 -6.20597661e-01 4.03636128e-01 2.34019846e-01 1.23052061e+00 4.72883850e-01 5.71719289e-01 1.04068649e+00 -4.45853382e-01 -5.83004773e-01 -9.23952579e-01 -6.91036642e-01 2.48666883e-01 -3.18182781e-02 -3.26210737e-01 -4.41200733e-01 -2.76617348e-01]
[10.982259750366211, 8.827964782714844]
60ef7382-bb06-404c-b03b-a918cf82d7f1
factorbase-sql-for-learning-a-multi
1508.02428
null
http://arxiv.org/abs/1508.02428v1
http://arxiv.org/pdf/1508.02428v1.pdf
FactorBase: SQL for Learning A Multi-Relational Graphical Model
We describe FactorBase, a new SQL-based framework that leverages a relational database management system to support multi-relational model discovery. A multi-relational statistical model provides an integrated analysis of the heterogeneous and interdependent data resources in the database. We adopt the BayesStore design philosophy: statistical models are stored and managed as first-class citizens inside a database. Whereas previous systems like BayesStore support multi-relational inference, FactorBase supports multi-relational learning. A case study on six benchmark databases evaluates how our system supports a challenging machine learning application, namely learning a first-order Bayesian network model for an entire database. Model learning in this setting has to examine a large number of potential statistical associations across data tables. Our implementation shows how the SQL constructs in FactorBase facilitate the fast, modular, and reliable development of highly scalable model learning systems.
['Zhensong Qian', 'Oliver Schulte']
2015-08-10
null
null
null
null
['model-discovery']
['miscellaneous']
[-7.43630946e-01 1.17110044e-01 -8.53565872e-01 -7.99015164e-01 -1.02168357e+00 -2.66111076e-01 5.00610709e-01 4.10100430e-01 6.80362880e-02 4.03948694e-01 5.46890944e-02 -6.83134556e-01 -6.39339328e-01 -1.29541957e+00 -9.48259473e-01 -1.43400982e-01 -3.88505578e-01 9.73890603e-01 7.19326794e-01 1.02169760e-01 -8.89714882e-02 6.61276281e-01 -1.86421216e+00 8.60615671e-01 5.72237372e-02 1.09562767e+00 -3.77646118e-01 6.34287894e-01 -1.74891502e-01 1.36620831e+00 -3.20767581e-01 -5.34795880e-01 6.81083873e-02 5.68336964e-01 -5.73990047e-01 -7.48770356e-01 6.26899660e-01 -2.60395825e-01 -2.55720496e-01 5.43951452e-01 1.97377384e-01 -1.85208932e-01 3.16964626e-01 -1.73795676e+00 3.04645300e-02 9.23545778e-01 -1.50341257e-01 2.89756805e-01 3.16971093e-01 -1.29576996e-01 1.10996413e+00 -8.48828971e-01 6.32534325e-01 1.63483810e+00 9.12643015e-01 -2.24656135e-01 -1.55900693e+00 -6.73112929e-01 -2.59758443e-01 2.90395081e-01 -1.76160502e+00 -7.08103001e-01 1.48491804e-02 -4.36918736e-01 1.50911641e+00 6.90074086e-01 3.70843589e-01 4.78126615e-01 5.77931821e-01 5.98512232e-01 8.59273016e-01 -3.07042629e-01 2.69268632e-01 5.57100773e-01 7.11996615e-01 7.73912728e-01 5.18028021e-01 5.21322452e-02 -1.20767653e+00 -8.64670396e-01 2.36754477e-01 7.49248918e-03 7.26858914e-01 -5.62077463e-01 -7.48289049e-01 5.40346444e-01 -2.60688104e-02 -1.82713851e-01 -4.72303659e-01 2.73241848e-01 4.72839355e-01 2.74533004e-01 1.31423578e-01 7.24852905e-02 -9.21789348e-01 -1.88516423e-01 -7.08453238e-01 4.74269450e-01 1.38721526e+00 1.37375164e+00 9.48799074e-01 -3.76504272e-01 3.14491242e-01 7.43969858e-01 1.00584614e+00 3.37430328e-01 -2.22327724e-01 -9.59632635e-01 4.40009952e-01 6.98243201e-01 -2.56381065e-01 -9.33169007e-01 -3.41931581e-01 8.48141238e-02 -2.06632286e-01 -4.15252224e-02 1.38073862e-01 4.01612222e-01 -2.80278295e-01 1.10059559e+00 8.03026676e-01 -1.80396140e-01 1.78891838e-01 3.58682945e-02 9.72584665e-01 3.52772087e-01 3.64346236e-01 -5.52415699e-02 1.45780206e+00 -3.39580506e-01 -8.10862422e-01 5.53353801e-02 7.83805966e-01 -5.29675484e-01 7.65356481e-01 7.97332406e-01 -9.80833471e-01 -3.25746000e-01 -1.07209253e+00 -1.14775270e-01 -8.77691627e-01 -4.09689039e-01 1.00605202e+00 8.24779153e-01 -7.73465335e-01 1.35035172e-01 -1.30489743e+00 -2.59318024e-01 3.16241026e-01 3.25435042e-01 -6.23250246e-01 -3.06569606e-01 -9.32550788e-01 1.04752564e+00 6.94632053e-01 -3.11441481e-01 -7.00889468e-01 -1.22154880e+00 -5.53005219e-01 1.30400300e-01 7.06869125e-01 -5.38498998e-01 1.16029596e+00 2.81307429e-01 -6.78621411e-01 6.65148854e-01 -1.42053932e-01 -1.35383725e-01 1.07572734e-01 -2.36701787e-01 -7.97527730e-01 -3.50767225e-01 5.11099701e-04 -2.30068117e-01 -2.94986796e-02 -1.32401693e+00 -7.99687624e-01 -8.20595026e-01 -2.78686345e-01 -4.68171686e-01 -1.26911432e-01 4.63445425e-01 -7.99184680e-01 -1.77868739e-01 3.06412399e-01 -5.68267822e-01 -8.96868706e-02 -2.69956112e-01 -6.27137363e-01 -2.99229920e-01 5.86034298e-01 -5.08846641e-01 1.69027865e+00 -2.03252864e+00 -3.26257139e-01 8.97792459e-01 2.25494608e-01 -4.72701043e-01 4.47867244e-01 6.10294402e-01 2.68666428e-02 5.17235816e-01 3.95602971e-01 1.22751184e-01 7.48634040e-02 4.80591804e-01 -2.48785630e-01 2.61241123e-02 7.47762173e-02 4.61501539e-01 -1.88207194e-01 -1.03525448e+00 -1.39962152e-01 1.92349792e-01 -7.41756618e-01 2.25416079e-01 -5.51012337e-01 -7.25918055e-01 -4.71254975e-01 1.05954635e+00 7.98417449e-01 -3.42715263e-01 9.66976941e-01 -4.54293907e-01 -7.50111532e-04 4.44124669e-01 -1.85438514e+00 1.10232449e+00 -1.40535161e-01 -7.04266205e-02 1.21688016e-01 -5.78325391e-01 9.58658159e-01 1.00308873e-01 6.17224991e-01 -3.85056555e-01 -6.93299770e-01 9.12503600e-02 -2.60138124e-01 -6.36221409e-01 4.15965110e-01 -1.60566680e-02 -1.72728494e-01 6.69821143e-01 2.29880214e-01 5.52490875e-02 3.46666873e-01 3.90858650e-01 1.48394525e+00 1.67186067e-01 6.07652187e-01 -3.22223008e-01 7.77356774e-02 2.42823213e-01 7.51585960e-01 8.95430803e-01 4.88808692e-01 -1.80040374e-01 7.91386843e-01 -7.83858120e-01 -5.77038050e-01 -1.55098546e+00 -7.27326810e-01 1.53768706e+00 -4.93449748e-01 -1.40350235e+00 5.52337654e-02 -4.38471556e-01 8.47154319e-01 7.70911932e-01 -2.92639315e-01 6.23121634e-02 -1.90709919e-01 -1.09659386e+00 6.11995161e-01 5.89913607e-01 -2.13211954e-01 -7.45880723e-01 -2.79062092e-01 3.13298941e-01 3.13005716e-01 -8.59698892e-01 5.69153786e-01 5.59562564e-01 -8.99689674e-01 -1.51776552e+00 1.17299330e+00 9.80792046e-02 -1.95662364e-01 -2.47716114e-01 1.70398021e+00 4.82708961e-02 -4.51743245e-01 3.77461493e-01 2.38095358e-01 -7.91514993e-01 -6.64714158e-01 8.16528052e-02 -1.89425182e-02 -4.24992800e-01 1.13239276e+00 -4.78014946e-01 3.03328365e-01 4.65759099e-01 -8.91336024e-01 -2.70366549e-01 3.69719535e-01 4.02525157e-01 8.56926918e-01 4.52359021e-01 3.26437891e-01 -1.22899950e+00 1.82709798e-01 -1.00853765e+00 -9.30660307e-01 8.92927229e-01 -1.08580792e+00 1.59971565e-01 -5.77313267e-02 1.35880992e-01 -8.84050846e-01 -1.38145387e-02 2.91262835e-01 -6.14906736e-02 -1.16778433e-01 1.22957003e+00 -7.28054047e-01 6.26719356e-01 6.99430883e-01 -2.53505558e-01 4.10927981e-01 -6.71348870e-01 3.89079571e-01 6.68358982e-01 2.94493318e-01 -1.11781991e+00 3.42370093e-01 2.09382325e-01 3.42246473e-01 -7.40278661e-01 -4.93212730e-01 -4.39836383e-01 -8.16306233e-01 -1.32820576e-01 3.75189126e-01 -1.07242966e+00 -1.37557077e+00 2.25108936e-02 -8.65516961e-01 -2.74921805e-01 -3.50201465e-02 4.05688643e-01 -5.05982757e-01 -3.79824549e-01 -3.44876647e-01 -8.53680253e-01 1.87159136e-01 -9.28654552e-01 6.72981262e-01 -3.90277088e-01 -4.51087147e-01 -7.97736108e-01 2.45308727e-01 5.79649568e-01 3.04908216e-01 6.92331567e-02 1.51572406e+00 -1.22112525e+00 -1.09583211e+00 -3.48453224e-01 -9.02435705e-02 -1.48736984e-01 -2.32473806e-01 7.83902168e-01 -9.14778888e-01 1.49292320e-01 -4.89742517e-01 -4.45621610e-01 3.46772045e-01 -1.56026138e-02 1.17911923e+00 -2.25422576e-01 -7.15307653e-01 3.72710258e-01 1.62211812e+00 1.00291267e-01 5.10908961e-01 4.62270677e-01 6.22245729e-01 7.47173190e-01 6.82873130e-01 6.40496552e-01 9.87207353e-01 6.88974917e-01 2.06866220e-01 2.55076408e-01 4.97568935e-01 -1.88126475e-01 1.14845783e-02 4.97864932e-01 2.40720198e-01 2.86504120e-01 -1.73173404e+00 4.68577333e-02 -1.92027915e+00 -1.09420609e+00 -6.08156681e-01 2.17921233e+00 1.34157479e+00 2.70196557e-01 2.10768223e-01 -9.82297286e-02 2.50623435e-01 -2.95485735e-01 -6.37792945e-01 -2.46701285e-01 3.75448610e-03 1.21807128e-01 4.23859894e-01 3.68362457e-01 -9.95529413e-01 6.94430590e-01 7.74828529e+00 5.49562573e-01 -5.23547053e-01 6.97055086e-02 4.41560626e-01 -5.14945686e-01 -3.95470411e-01 1.31201029e-01 -1.42251611e+00 1.05828382e-01 1.80721498e+00 -2.41865590e-01 1.98552623e-01 1.30040967e+00 -7.78023824e-02 -3.55119228e-01 -1.41700876e+00 5.44404566e-01 -3.31721157e-01 -1.78993416e+00 6.17592223e-02 3.83630842e-01 -7.82676861e-02 4.69559997e-01 -3.29372078e-01 3.64942580e-01 9.73848403e-01 -1.13459575e+00 6.85693502e-01 1.32127845e+00 4.35621023e-01 -8.84707332e-01 4.23798949e-01 2.07227781e-01 -1.00254691e+00 -2.84617513e-01 -1.82600752e-01 2.87139595e-01 -2.18659088e-01 8.69662821e-01 -7.39262283e-01 5.68621576e-01 1.42831051e+00 6.81851208e-01 -1.03997433e+00 6.32519484e-01 5.41884184e-01 4.01092976e-01 -6.06259525e-01 2.81389534e-01 -7.43120849e-01 -1.00774188e-02 -1.61201298e-01 1.16314042e+00 -1.15503833e-01 4.64286879e-02 3.74034554e-01 6.48901165e-01 2.67418742e-01 1.16409592e-01 -7.50914216e-01 1.48520246e-01 1.06256437e+00 1.03216958e+00 -2.19179854e-01 -5.46180189e-01 -7.59005904e-01 -8.86232972e-01 4.78679985e-01 3.05116996e-02 -5.11812091e-01 -2.69435421e-02 8.72385323e-01 4.25149798e-01 2.85200924e-01 -2.32947301e-02 -4.99161601e-01 -8.79274786e-01 1.03401363e-01 -1.48278308e+00 1.00034118e+00 -6.12718165e-01 -1.56040478e+00 1.07006125e-01 6.02752447e-01 -3.95770609e-01 -3.93459916e-01 -7.12498426e-01 7.55787641e-02 7.84661472e-01 -7.49735534e-01 -1.31691039e+00 4.37732264e-02 6.19630337e-01 -4.68150169e-01 -6.67575896e-01 1.12762201e+00 3.22513998e-01 -8.86353612e-01 3.26998144e-01 7.95504451e-02 -1.18115194e-01 8.17775905e-01 -1.29771292e+00 9.23759341e-02 3.37503105e-01 4.11947407e-02 1.39515305e+00 3.80208254e-01 -9.74799991e-01 -2.00654602e+00 -8.73142898e-01 8.75394046e-01 -1.08787417e+00 9.01977897e-01 -5.71070373e-01 -1.29670870e+00 1.16844797e+00 -1.91140041e-01 2.71670252e-01 1.42830634e+00 1.12822318e+00 -1.06079149e+00 -1.01756275e+00 -1.08990061e+00 1.83665659e-02 5.28234005e-01 -8.25649858e-01 -3.69787425e-01 2.77235359e-01 3.58788401e-01 1.09161045e-02 -2.02696848e+00 4.35993850e-01 9.51196134e-01 -1.09002066e+00 1.13250136e+00 -1.29925978e+00 2.51824349e-01 -2.83096135e-01 -9.36127603e-01 -3.32546622e-01 -5.26988395e-02 -3.77277762e-01 -7.89535522e-01 1.30197906e+00 6.11277819e-01 -6.46009862e-01 7.61109829e-01 1.42419648e+00 2.32867613e-01 -5.34973145e-01 -7.78233230e-01 -5.67349076e-01 -1.35796577e-01 -9.69617963e-01 1.08944440e+00 9.54838991e-01 1.68882105e-02 9.43749622e-02 1.72374398e-01 5.77953577e-01 7.48364389e-01 -2.20195781e-02 1.36964285e+00 -1.66701376e+00 -4.90987241e-01 2.16071028e-02 -4.65935588e-01 -1.65876195e-01 8.23981315e-02 -9.06449497e-01 -3.27000171e-01 -1.16116083e+00 3.92477840e-01 -1.19129002e+00 -3.97356659e-01 8.54816079e-01 1.02701545e-01 -4.56252217e-01 -2.01178879e-01 2.25783497e-01 -6.18160605e-01 -1.31715029e-01 1.32347569e-01 -1.45662557e-02 6.31527230e-02 1.03530169e-01 -6.93580031e-01 4.63148385e-01 2.88725972e-01 -9.30023253e-01 -4.13608581e-01 -3.63370441e-02 8.25676560e-01 2.90113181e-01 3.96192849e-01 -8.99144769e-01 7.33416080e-01 -5.13129890e-01 3.31364810e-01 -1.20604277e+00 2.32186049e-01 -9.29180026e-01 1.04648852e+00 1.99321836e-01 -2.83241063e-01 3.95001322e-01 4.49863225e-01 5.40837228e-01 -3.60938340e-01 1.26946628e-01 5.18042624e-01 -1.39534235e-01 -4.31705773e-01 1.49493366e-01 -2.15753093e-01 -1.21007618e-02 1.09586298e+00 3.27708811e-01 -6.03018582e-01 3.11849356e-01 -9.04911339e-01 1.87146902e-01 2.06400752e-01 4.18357551e-01 4.68672365e-01 -1.26216614e+00 -4.24450517e-01 5.67859292e-01 4.91476685e-01 5.46374060e-02 -9.60707664e-02 5.37392139e-01 -4.02007848e-01 4.88104314e-01 -4.16106582e-02 -8.29209983e-01 -1.44325459e+00 7.16048419e-01 3.06972325e-01 -1.99175447e-01 -8.58191326e-02 4.09417301e-01 -6.49948597e-01 -9.07948971e-01 2.94718146e-01 -2.33268067e-01 2.95297623e-01 4.03289348e-01 6.48887157e-01 5.08556664e-01 5.33596098e-01 -4.44944911e-02 -6.48590505e-01 -1.09513670e-01 -2.79137641e-01 -1.15362868e-01 1.76083815e+00 1.74708143e-01 -9.15815234e-01 1.26021266e+00 7.49033093e-01 -2.53595561e-02 -5.78218877e-01 -3.51022720e-01 9.16872501e-01 -4.51273412e-01 8.25429335e-02 -9.78405595e-01 -6.42853618e-01 3.12698245e-01 2.40580603e-01 3.52541119e-01 5.22459507e-01 3.75666112e-01 -4.07722503e-01 9.56452668e-01 5.06888628e-01 -1.06859016e+00 -2.98808575e-01 2.13245407e-01 8.10741305e-01 -1.15439785e+00 7.43947148e-01 -3.80222082e-01 -3.69049281e-01 1.21477044e+00 8.13171566e-01 5.99629998e-01 1.61442304e+00 9.43035543e-01 1.72035903e-01 -7.28640378e-01 -1.71122050e+00 3.62329304e-01 5.46213388e-02 4.74578977e-01 5.66556096e-01 1.66196600e-02 5.30432642e-01 8.16421270e-01 -2.13185802e-01 1.66710421e-01 2.69998014e-01 1.31769562e+00 -1.42069057e-01 -1.40511715e+00 -5.87338269e-01 9.65514004e-01 -4.92482424e-01 3.07383835e-02 2.28955578e-02 1.25164533e+00 -9.30061042e-02 8.09297085e-01 3.78106415e-01 -2.77701110e-01 6.71256065e-01 5.59384644e-01 1.76355213e-01 -7.89530277e-01 -6.92196786e-01 -5.70191778e-02 5.37867010e-01 -1.03358841e+00 -1.73589110e-01 -1.01079309e+00 -1.03685081e+00 -6.73868358e-01 -1.07745184e-02 -3.74379680e-02 1.00838685e+00 6.46444499e-01 6.30930781e-01 2.85152674e-01 2.28733256e-01 2.24522948e-01 -7.47956812e-01 -6.88554466e-01 -8.46503377e-01 -5.69754429e-02 1.59638468e-02 -7.10059166e-01 9.64171663e-02 1.01342008e-01]
[9.204015731811523, 7.591470241546631]
bbb9f914-cbf8-4ca7-8c39-14ffe02aceef
bodies-at-rest-3d-human-pose-and-shape
2004.01166
null
https://arxiv.org/abs/2004.01166v1
https://arxiv.org/pdf/2004.01166v1.pdf
Bodies at Rest: 3D Human Pose and Shape Estimation from a Pressure Image using Synthetic Data
People spend a substantial part of their lives at rest in bed. 3D human pose and shape estimation for this activity would have numerous beneficial applications, yet line-of-sight perception is complicated by occlusion from bedding. Pressure sensing mats are a promising alternative, but training data is challenging to collect at scale. We describe a physics-based method that simulates human bodies at rest in a bed with a pressure sensing mat, and present PressurePose, a synthetic dataset with 206K pressure images with 3D human poses and shapes. We also present PressureNet, a deep learning model that estimates human pose and shape given a pressure image and gender. PressureNet incorporates a pressure map reconstruction (PMR) network that models pressure image generation to promote consistency between estimated 3D body models and pressure image input. In our evaluations, PressureNet performed well with real data from participants in diverse poses, even though it had only been trained with synthetic data. When we ablated the PMR network, performance dropped substantially.
['Ariel Kapusta', 'Henry M. Clever', 'C. Karen Liu', 'Zackory Erickson', 'Charles C. Kemp', 'Greg Turk']
2020-04-02
bodies-at-rest-3d-human-pose-and-shape-1
http://openaccess.thecvf.com/content_CVPR_2020/html/Clever_Bodies_at_Rest_3D_Human_Pose_and_Shape_Estimation_From_CVPR_2020_paper.html
http://openaccess.thecvf.com/content_CVPR_2020/papers/Clever_Bodies_at_Rest_3D_Human_Pose_and_Shape_Estimation_From_CVPR_2020_paper.pdf
cvpr-2020-6
['3d-human-pose-and-shape-estimation']
['computer-vision']
[-1.32986484e-02 5.63740253e-01 3.00928831e-01 -3.48630160e-01 -2.16743320e-01 -1.46802759e-03 1.42622814e-01 -2.68348932e-01 -1.77606881e-01 5.10650635e-01 3.69807035e-01 3.69823724e-01 5.83590150e-01 -8.74674261e-01 -1.05678689e+00 -1.29856635e-02 -6.04330860e-02 1.06233084e+00 -4.11891332e-03 -4.41510946e-01 -3.73975426e-01 2.79292613e-01 -1.53716862e+00 1.28989801e-01 5.78104615e-01 8.63488615e-01 -2.21654668e-01 7.64565647e-01 5.16175926e-01 -2.96866186e-02 -6.40587807e-01 -7.61801183e-01 7.02204764e-01 -1.07331842e-01 -3.78275365e-01 2.93082833e-01 9.92546618e-01 -6.84270442e-01 -3.80226731e-01 2.58803844e-01 1.04634786e+00 3.70578724e-03 5.00119269e-01 -1.05718267e+00 -1.12480544e-01 7.94486701e-02 -8.09650302e-01 -3.65913540e-01 1.07385421e+00 4.44496542e-01 4.49777365e-01 -9.06431079e-01 4.46265787e-01 1.60460472e+00 1.43599331e+00 8.32744241e-01 -1.40803754e+00 -4.90563899e-01 -2.22201869e-01 -5.82832456e-01 -1.30476880e+00 -2.88988173e-01 6.33930922e-01 -5.64246655e-01 7.73231447e-01 4.27624434e-01 1.69359565e+00 1.43082023e+00 4.57967788e-01 8.50475371e-01 1.02561677e+00 -1.79578006e-01 4.57324609e-02 3.78485806e-02 -5.36641240e-01 7.77827322e-01 3.03060651e-01 -6.67133555e-02 -9.46242630e-01 -2.98631728e-01 1.18906236e+00 -8.34796801e-02 -1.07635617e-01 -5.67048192e-01 -1.06969941e+00 2.20879167e-01 6.39048755e-01 -5.04089177e-01 -3.96622956e-01 4.02562737e-01 4.40136790e-02 -3.60768318e-01 5.20564854e-01 3.79515767e-01 -4.94055718e-01 -1.63733780e-01 -8.82706881e-01 1.11617947e+00 9.51365173e-01 7.59316444e-01 2.81800896e-01 -1.60299927e-01 -2.63864219e-01 8.05631936e-01 4.22437727e-01 1.25201094e+00 5.35535067e-02 -1.06658494e+00 6.09955966e-01 6.50204420e-01 4.06130075e-01 -1.30085993e+00 -7.95999050e-01 -1.01552680e-01 -6.89334989e-01 1.97320655e-01 6.86199903e-01 -4.29059178e-01 -1.04502356e+00 1.49730206e+00 6.42542601e-01 -2.66319841e-01 -5.16921103e-01 1.25760996e+00 9.41691458e-01 4.36474323e-01 7.63898492e-02 5.13886571e-01 1.36611784e+00 -6.31417811e-01 -1.63342461e-01 -8.36734653e-01 -3.27433906e-02 -3.92783821e-01 1.38094568e+00 5.00157893e-01 -1.65148747e+00 -7.48155177e-01 -9.08170521e-01 -1.60924822e-01 1.57153770e-01 -3.42197828e-02 5.50275266e-01 6.80954874e-01 -1.01786602e+00 9.21668053e-01 -1.16230011e+00 -6.83202744e-01 1.76718965e-01 5.42836785e-01 -3.43271464e-01 4.08379078e-01 -1.19171917e+00 9.34833765e-01 -2.18393460e-01 3.19075644e-01 -4.21164662e-01 -7.33710706e-01 -1.21012604e+00 -3.92515361e-01 -1.82229280e-01 -1.29374981e+00 1.24368095e+00 -1.38994873e-01 -1.44044948e+00 1.24335992e+00 1.09907985e-01 -5.30629814e-01 1.24181831e+00 -1.03083134e+00 5.30341975e-02 1.32385015e-01 -5.53926900e-02 1.28480768e+00 8.69175017e-01 -1.07406461e+00 4.08465475e-01 -6.85076892e-01 -3.82991225e-01 7.46182740e-01 -3.10711339e-02 -3.01236928e-01 -5.71711719e-01 -3.60236585e-01 2.46924669e-01 -1.21989048e+00 -3.46075952e-01 6.82267666e-01 -6.29086316e-01 4.24658686e-01 4.11446601e-01 -1.00908113e+00 4.92437065e-01 -1.53738308e+00 7.64142722e-02 2.71412343e-01 1.54348239e-01 1.61622781e-02 4.14746732e-01 1.35835201e-01 3.70109856e-01 -3.96421328e-02 -1.18126504e-01 -8.83491278e-01 1.18622698e-01 3.56073752e-02 -9.11211148e-02 4.84448582e-01 -4.83415881e-03 1.02006364e+00 -5.50626576e-01 -7.28710830e-01 4.66942817e-01 7.71577418e-01 -8.55266750e-01 3.76906097e-01 -2.71668047e-01 8.08362603e-01 -5.09116501e-02 9.98773694e-01 7.44290054e-01 -1.59369737e-01 -4.23083827e-02 -2.69025475e-01 4.00771648e-01 9.38347634e-03 -9.97934759e-01 1.85165620e+00 -1.64755583e-01 1.58909261e-01 2.34434903e-01 -2.93652326e-01 9.71422851e-01 2.04875320e-01 7.09533393e-01 -5.34893572e-01 1.32569790e-01 5.44278771e-02 -2.10639402e-01 -5.67753136e-01 5.45361757e-01 -5.07649541e-01 -3.73753846e-01 1.79471567e-01 -5.97913384e-01 -6.79850519e-01 -3.25947762e-01 -1.30868435e-01 7.06599116e-01 6.77632987e-01 -4.84319739e-02 -1.48066431e-01 -2.50576496e-01 -1.27585962e-01 2.61783123e-01 6.22535467e-01 -2.52367854e-01 1.32136953e+00 -5.29535376e-02 -9.63471055e-01 -1.29483938e+00 -1.59246337e+00 -2.94151455e-01 9.18121338e-01 2.35965803e-01 -3.20131600e-01 -8.47850144e-01 8.64107907e-03 6.03760660e-01 1.35043025e-01 -8.75458837e-01 3.16740870e-02 -7.31495142e-01 -7.14454591e-01 6.28123701e-01 8.69627833e-01 8.13195169e-01 -1.18755066e+00 -1.12144411e+00 7.29262382e-02 -2.98550546e-01 -8.85627508e-01 -4.52623963e-01 -4.47643816e-01 -7.18314350e-01 -8.38863254e-01 -1.20467305e+00 -2.42621869e-01 7.24485219e-01 -4.07065451e-01 1.55564129e+00 -2.10258484e-01 -6.91856086e-01 4.83400375e-01 1.07556239e-01 -8.88794124e-01 -1.31086633e-01 -1.39471278e-01 5.76582968e-01 -4.98424292e-01 1.84153497e-01 -6.19405866e-01 -1.18860817e+00 6.08205497e-01 -6.96207881e-02 4.68464166e-01 2.05902144e-01 4.36189026e-01 6.89973414e-01 -6.63823664e-01 -2.06220046e-01 -4.99037296e-01 4.59391892e-01 1.26445189e-01 8.40950198e-03 -2.44318098e-01 2.34389186e-01 -4.40392643e-01 -2.19057072e-02 -4.12211269e-01 -8.69494319e-01 1.98174536e-01 -4.07842606e-01 -3.25816572e-01 -3.60847972e-02 -1.45404458e-01 -2.55574659e-02 2.31076002e-01 1.14850342e+00 -1.55752227e-01 5.25271177e-01 -4.65342015e-01 2.79209912e-02 3.84820074e-01 7.47392952e-01 -9.55649257e-01 6.96052015e-01 6.45149171e-01 -2.04186141e-02 -8.70316148e-01 -9.07767296e-01 1.66577213e-02 -8.93704057e-01 -5.22774339e-01 8.30360651e-01 -1.07076585e+00 -1.23477924e+00 8.32626581e-01 -8.01726639e-01 -7.89992750e-01 -4.94584918e-01 3.70411098e-01 -7.55742610e-01 -3.27128842e-02 -8.66427422e-01 -8.72747600e-01 -7.61400282e-01 -6.16209924e-01 1.73958206e+00 3.68096918e-01 -1.03018916e+00 -4.84491318e-01 6.99383765e-02 7.97884524e-01 3.76025170e-01 8.25622380e-01 -2.67310794e-02 1.98611647e-01 2.14732569e-02 -3.57494175e-01 1.39720663e-01 -1.41722225e-02 -9.75283906e-02 -2.64992118e-01 -1.03269660e+00 -3.82128209e-01 -2.55866110e-01 -9.41890538e-01 3.36353421e-01 7.26030529e-01 1.21879578e+00 -2.99700588e-01 -4.90299344e-01 6.58414960e-01 6.70944273e-01 -6.16623521e-01 6.93397045e-01 4.37811716e-03 9.92888510e-01 7.20367491e-01 4.65677291e-01 7.32845187e-01 5.07732093e-01 8.10477674e-01 1.94801569e-01 -5.06565750e-01 -2.06236467e-01 -7.86442459e-01 2.66411573e-01 3.22462320e-01 -5.02372444e-01 1.46118894e-01 -1.19489360e+00 5.53224757e-02 -1.31434929e+00 -6.84653401e-01 8.32575336e-02 2.26827359e+00 9.03842568e-01 4.69435066e-01 5.36302328e-01 -4.42171767e-02 4.41564500e-01 -6.40669912e-02 -9.42980409e-01 -2.85503328e-01 1.87885001e-01 2.41400495e-01 4.49226499e-01 2.61825949e-01 -1.11973810e+00 4.62368339e-01 7.20937967e+00 -8.78549144e-02 -1.03006208e+00 -2.11188629e-01 8.55319500e-01 -5.06994188e-01 2.37458013e-02 -6.59698546e-01 -7.65606880e-01 3.30694556e-01 5.63487113e-01 2.66554207e-01 -1.88361052e-02 1.02044249e+00 2.57193506e-01 -2.99450636e-01 -1.21044028e+00 1.17367995e+00 2.06236929e-01 -6.38315380e-01 -2.65429050e-01 1.44056752e-01 5.32785714e-01 6.26207069e-02 1.60736814e-01 2.37274662e-01 1.72088698e-01 -1.28706288e+00 1.01718259e+00 7.31923461e-01 1.17251921e+00 -2.63796240e-01 2.52826244e-01 5.20018220e-01 -9.52601969e-01 3.72679085e-01 -3.48127544e-01 -5.25516987e-01 4.73197937e-01 5.21816432e-01 -1.04961741e+00 -5.48406020e-02 9.93528366e-01 4.16385680e-01 -6.18634224e-01 8.36851239e-01 -1.53922334e-01 2.83453673e-01 -8.40623319e-01 2.53509358e-02 -7.51523376e-01 7.92609230e-02 4.51549381e-01 1.04859531e+00 2.86930948e-01 -4.34566103e-02 4.74216431e-01 1.07346690e+00 -3.03692799e-02 -9.60945264e-02 -6.15202963e-01 4.57546830e-01 1.16385661e-01 1.00764024e+00 -5.02990186e-01 -2.14325383e-01 1.57298714e-01 1.08254766e+00 4.13201824e-02 -2.68618278e-02 -9.01954174e-01 8.99709389e-02 5.77932179e-01 1.14618015e+00 -3.42497826e-01 -2.88600951e-01 -5.74943066e-01 -1.12064314e+00 2.84769028e-01 -4.73709643e-01 3.66770998e-02 -1.33957648e+00 -1.19623768e+00 2.17688724e-01 3.42107326e-01 -1.17084420e+00 -2.59841233e-01 -4.69089478e-01 -2.49517500e-01 1.00104332e+00 -4.56349850e-01 -1.23263943e+00 -7.35050619e-01 2.21850231e-01 3.03180188e-01 7.21385539e-01 8.98221850e-01 8.04999694e-02 -2.54586011e-01 7.60485649e-01 -8.10330212e-01 1.02804147e-01 7.72106946e-01 -1.36361301e+00 1.08351541e+00 2.25634128e-01 -3.93669426e-01 6.36017025e-01 1.19003558e+00 -1.15414846e+00 -1.50756824e+00 -8.90818298e-01 3.93576473e-01 -1.01263344e+00 -1.66000873e-01 -8.18045914e-01 -5.97161770e-01 7.70822048e-01 -2.16342419e-01 2.15381861e-01 4.85273153e-01 1.32397264e-01 4.41287197e-02 -3.18820551e-02 -1.53663528e+00 7.88553953e-01 1.41109443e+00 -8.73988867e-02 -6.39535725e-01 3.79036665e-01 5.00843287e-01 -1.51080191e+00 -9.26354289e-01 7.78060019e-01 1.31499720e+00 -9.40771699e-01 1.40079761e+00 -3.31521891e-02 6.04582846e-01 -1.25089005e-01 1.93366945e-01 -1.25000787e+00 -3.52975316e-02 -4.68444824e-01 -3.68739218e-01 5.00242949e-01 1.82758749e-01 -3.84130567e-01 1.46802652e+00 1.43949962e+00 1.74231187e-01 -9.85769689e-01 -4.90114659e-01 -5.62658787e-01 -1.92433444e-03 -3.48735332e-01 4.87690687e-01 2.19233900e-01 7.05742538e-02 1.38229489e-01 -6.52274549e-01 -1.19692154e-01 6.76055431e-01 -1.99795906e-02 1.41314876e+00 -1.28193855e+00 -6.21139884e-01 1.69022322e-01 -3.23360860e-01 -1.08668292e+00 -4.01887149e-01 -2.92032629e-01 2.96034247e-01 -1.71980393e+00 1.93663076e-01 -4.31522787e-01 5.09871364e-01 5.71867228e-01 -1.32398978e-01 7.11760700e-01 2.21552342e-01 2.43325666e-01 -2.24619493e-01 4.76679295e-01 1.70576823e+00 -3.15816561e-03 -4.80880171e-01 2.32614994e-01 -4.45848286e-01 1.07605970e+00 6.19433701e-01 6.87309057e-02 -3.64735365e-01 -3.99735063e-01 3.75508666e-01 1.80732310e-01 7.09134996e-01 -1.53408265e+00 -2.29233205e-01 -1.40245855e-02 1.46495521e+00 -8.62896740e-01 1.11715508e+00 -3.34113032e-01 6.36208713e-01 8.46739590e-01 -7.74915963e-02 -8.25135484e-02 4.35721040e-01 1.91437900e-01 6.55744195e-01 5.69729745e-01 6.31621361e-01 -6.59212470e-01 1.07691117e-01 3.74217123e-01 5.25280982e-02 6.03979491e-02 5.66224515e-01 -6.93535388e-01 3.19057889e-02 -4.15440172e-01 -1.10862434e+00 3.11321020e-01 1.00164604e+00 2.58265436e-01 6.65758789e-01 -1.31807232e+00 -7.15797365e-01 5.21481693e-01 -1.36319131e-01 7.74463058e-01 1.85102269e-01 5.62954366e-01 -9.77899134e-01 -5.96954226e-02 -3.01642865e-01 -9.27283585e-01 -1.10674584e+00 -1.70645900e-02 5.87893069e-01 -1.60350993e-01 -1.06206548e+00 1.06280267e+00 9.08576846e-02 -8.82058918e-01 2.04041079e-01 -4.33676392e-01 3.67507696e-01 -4.78415608e-01 5.53199947e-01 2.69084394e-01 1.48897404e-02 -7.56385803e-01 -3.16384971e-01 7.68514633e-01 3.64251047e-01 -3.82651746e-01 1.03890073e+00 4.82943878e-02 3.54923069e-01 3.80439311e-01 4.89283144e-01 1.82400540e-01 -1.73824525e+00 9.22033191e-02 -8.68655920e-01 -5.54345787e-01 -8.67055595e-01 -7.74683893e-01 -8.08169067e-01 8.27278793e-01 4.44605082e-01 -2.11984769e-01 6.44686878e-01 4.80648279e-02 1.08040524e+00 2.44875520e-01 7.72840798e-01 -1.14127564e+00 4.52626705e-01 2.03424945e-01 1.30830836e+00 -1.07128417e+00 3.85222137e-01 -7.06565082e-01 -4.95375216e-01 6.08309805e-01 1.23596549e+00 -2.26359338e-01 4.53651190e-01 3.14761847e-01 -9.06572267e-02 -4.24255073e-01 -1.59808248e-01 3.24947685e-01 3.84385049e-01 8.09667706e-01 6.21969640e-01 5.46447814e-01 -1.39332160e-01 3.86335701e-01 -1.21073234e+00 1.76871493e-01 7.06769750e-02 1.09769392e+00 -3.49820971e-01 -7.37778485e-01 -8.06844652e-01 6.75310314e-01 -3.55189413e-01 4.21137214e-01 -3.95244628e-01 6.51398599e-01 2.15203792e-01 2.64159143e-01 2.81433612e-01 -4.97067302e-01 6.74698412e-01 -1.83144175e-02 9.75144744e-01 -8.01474571e-01 -4.59716499e-01 -1.23756766e-01 1.67645007e-01 -7.37516284e-01 -1.47630870e-01 -7.21279204e-01 -1.18214822e+00 -4.10122335e-01 1.82786450e-01 -5.38784802e-01 6.51682198e-01 4.58277345e-01 3.82940084e-01 1.52076736e-01 6.28602803e-02 -1.77963591e+00 -3.90382171e-01 -1.11817706e+00 -5.82787037e-01 8.84751737e-01 -1.90899268e-01 -7.73369431e-01 5.04172676e-05 2.17510343e-01]
[7.019432067871094, -1.1348596811294556]
ed22fa76-77da-45fd-93f8-1e87caf8081b
orbits-online-recovery-of-missing-blocks-in
null
null
http://www.vldb.org/pvldb/vol14/p294-khayati.pdf
http://www.vldb.org/pvldb/vol14/p294-khayati.pdf
ORBITS: Online Recovery of Missing Blocks in Multiple Time Series Streams
With the emergence of the Internet of Things (IoT), time series streams have become ubiquitous in our daily life. Recording such data is rarely a perfect process, as sensor failures frequently occur, yielding occasional blocks of data that go missing in multiple time series. These missing blocks do not only affect real-time monitoring but also compromise the quality of online data analyses. Effective streaming recovery (imputation) techniques either have a quadratic runtime complexity, which is infeasible for any moderately sized data or cannot recover more than one time series at a time. In this paper, we introduce a new online recovery technique to recover multiple time series streams in linear time. Our recovery technique implements a novel incremental version of the Centroid Decomposition technique and reduces its complexity from quadratic to linear. Using this incremental technique, missing blocks are efficiently recovered in a continuous manner based on previous recoveries. We formally prove the correctness of our new incremental computation, which yields an accurate recovery. Our experimental results on real-world time series show that our recovery technique is, on average, 30% more accurate than the state of the art while being vastly more efficient.
['Philippe Cudré-Mauroux', 'Zakhar Tymchenko', 'Ines Arous', 'Mourad Khayati']
2020-11-01
null
null
null
proceedings-of-the-vldb-endowment-pvldb-2020
['multivariate-time-series-imputation', 'time-series-streams']
['time-series', 'time-series']
[ 3.34568292e-01 -3.18796158e-01 -2.96637695e-02 -4.61441837e-02 -1.09403121e+00 -6.99095726e-01 1.05200954e-01 6.59718394e-01 -1.62419915e-01 5.10725200e-01 1.05959825e-01 -1.76294535e-01 -3.90209943e-01 -8.93651187e-01 -7.94995189e-01 -5.54198861e-01 -5.49115956e-01 4.34047908e-01 4.45471674e-01 7.86739588e-02 4.94305119e-02 4.07894522e-01 -1.55752647e+00 2.13028982e-01 6.42245710e-01 1.33370984e+00 -2.37826005e-01 8.79363716e-01 -6.23104982e-02 7.79403627e-01 -7.53289640e-01 1.71788838e-02 3.54940176e-01 -1.42107680e-01 -4.45644289e-01 2.80530062e-02 -2.04215601e-01 -6.80893123e-01 -2.41229877e-01 7.20013261e-01 2.70407051e-01 -6.33493587e-02 1.45491231e-02 -1.53192711e+00 1.86626986e-01 9.79244232e-01 -1.00275731e+00 2.46305943e-01 5.11055887e-01 -4.18797910e-01 6.27630651e-01 -3.82669181e-01 2.59015411e-01 6.53122067e-01 1.02837050e+00 -9.36924741e-02 -1.20263803e+00 -7.52744496e-01 -5.62849715e-02 2.10758090e-01 -1.35666192e+00 -6.48350298e-01 8.34951282e-01 5.34649901e-02 9.33553874e-01 5.39873481e-01 7.39062488e-01 3.93610686e-01 8.42193589e-02 6.58365369e-01 6.17527127e-01 -1.49067447e-01 4.48731422e-01 -6.87153935e-01 -6.39085770e-02 1.28523575e-03 4.57949907e-01 -1.03239901e-01 -6.26327515e-01 -6.76798999e-01 6.00826681e-01 8.58952641e-01 -2.11455792e-01 -2.54095439e-02 -1.48388207e+00 3.53094429e-01 -3.80207971e-02 3.00490230e-01 -5.89887083e-01 3.51611614e-01 6.60377920e-01 6.12186015e-01 6.44107640e-01 -4.24925119e-01 -4.65589941e-01 -6.76594079e-01 -1.24742138e+00 1.78786844e-01 1.10045028e+00 9.51032281e-01 5.16551852e-01 1.43710924e-02 3.87028933e-01 4.08063650e-01 -2.37867624e-01 7.68166125e-01 3.67241681e-01 -1.03520298e+00 7.33650029e-01 5.70862949e-01 3.20467323e-01 -1.13672543e+00 -4.62231904e-01 -1.51854500e-01 -1.33006930e+00 -2.08089590e-01 5.84999621e-01 2.10441016e-02 -2.60285228e-01 1.43374407e+00 5.05581379e-01 6.55176997e-01 -1.80342823e-01 3.93137872e-01 -1.38456538e-01 8.19804668e-01 -5.86732864e-01 -9.34060276e-01 9.66483593e-01 -2.58838415e-01 -8.32821310e-01 2.18074203e-01 3.32561493e-01 -7.22630441e-01 5.09087026e-01 7.53426731e-01 -1.27480018e+00 -2.60393359e-02 -9.98946071e-01 3.43497276e-01 1.95182189e-01 -4.26778793e-01 5.76269507e-01 5.53639770e-01 -6.63678110e-01 8.86678040e-01 -1.45044053e+00 -5.26995435e-02 2.95318753e-01 2.36292556e-01 -3.48268569e-01 -2.12604493e-01 -4.59386766e-01 1.56171188e-01 6.99959099e-02 -5.09292297e-02 -5.93433380e-01 -9.33757126e-01 -4.07562792e-01 1.06439866e-01 5.06862640e-01 -3.09840262e-01 1.28359807e+00 -5.96701622e-01 -1.11655033e+00 1.49168015e-01 -5.89024365e-01 -7.58791327e-01 5.63520610e-01 -1.07072040e-01 -6.42792583e-01 3.19419295e-01 -4.24817316e-02 -4.74955976e-01 9.65599358e-01 -7.33756900e-01 -6.68846309e-01 -6.32222176e-01 -2.07413077e-01 -4.43028033e-01 -3.05622369e-01 -3.27720046e-02 7.30598792e-02 -7.64718354e-01 5.15673757e-01 -8.24049473e-01 -3.78143936e-01 1.23700909e-01 1.24124296e-01 -1.63694859e-01 7.56195068e-01 -6.94492757e-01 1.46676278e+00 -2.17695355e+00 -1.99742600e-01 3.41847092e-01 4.40339774e-01 -1.25196785e-01 2.19999582e-01 1.06811213e+00 -1.56672932e-02 -3.82497907e-02 -6.22770190e-01 -4.70332205e-01 -1.31030232e-01 2.97238648e-01 -9.37451363e-01 7.74346828e-01 -3.63877445e-01 5.10088742e-01 -1.08027399e+00 -6.04617782e-02 2.12816238e-01 3.21282566e-01 -4.15413052e-01 1.14341155e-01 -7.46101066e-02 2.45399848e-01 -1.02562115e-01 6.82454228e-01 8.71680796e-01 -4.24868733e-01 2.63502777e-01 -1.08970463e-01 -1.43987164e-01 1.86090946e-01 -1.47360361e+00 1.45478380e+00 -5.58030844e-01 3.60034496e-01 1.55720234e-01 -9.97468054e-01 7.16479123e-01 5.50936580e-01 1.08506131e+00 -5.77224076e-01 -2.05833942e-01 4.67368662e-01 -4.77942795e-01 -2.33165517e-01 5.18978536e-01 -3.61608602e-02 -1.49634629e-01 1.09622884e+00 -6.05961382e-01 1.16859749e-01 1.58218637e-01 6.63238615e-02 1.78511524e+00 -2.54916936e-01 4.76416498e-01 3.50496560e-01 -4.37909327e-02 -1.62379026e-01 6.80514216e-01 6.19656026e-01 -1.11314893e-01 5.80756068e-01 3.26621085e-01 -5.16579926e-01 -1.15535951e+00 -9.63888526e-01 3.07357997e-01 6.82943225e-01 -8.48224610e-02 -7.27053702e-01 -2.69050270e-01 -6.57497495e-02 8.82807374e-03 3.66823137e-01 -3.23763669e-01 4.55193408e-02 -7.64875591e-01 -6.84922993e-01 4.35133249e-01 6.23293042e-01 1.68323278e-01 -6.79669440e-01 -1.15743721e+00 7.93582082e-01 -4.44056243e-01 -1.11673152e+00 -3.78060669e-01 -4.51411940e-02 -1.46941602e+00 -1.06485569e+00 -3.93973678e-01 -3.90403345e-02 4.94268000e-01 8.19231987e-01 1.05926323e+00 2.57758588e-01 -1.79403409e-01 2.97912121e-01 -6.34895444e-01 -5.43545902e-01 -1.55431107e-01 -2.65808702e-01 -1.39991089e-03 1.96110144e-01 3.29197586e-01 -1.33054090e+00 -7.09014237e-01 1.51599914e-01 -1.40736330e+00 -3.56929213e-01 9.04790163e-02 4.95862216e-01 7.64909863e-01 7.23749042e-01 8.24552357e-01 -4.00049448e-01 3.69030625e-01 -9.30823147e-01 -8.28301847e-01 -1.57121867e-02 -4.84576851e-01 -2.49447912e-01 1.02441895e+00 -5.56099594e-01 -3.81784111e-01 1.10962033e-01 4.21121120e-02 -5.55609822e-01 1.83872193e-01 6.52294040e-01 2.49511033e-01 3.02407175e-01 3.64169806e-01 4.57726717e-01 -8.93850103e-02 -7.04896450e-01 1.16096213e-01 7.92302251e-01 6.75294757e-01 -4.62727875e-01 7.34635890e-01 1.23220313e+00 3.99102531e-02 -9.24867213e-01 -5.11856079e-01 -8.72620463e-01 -3.76393408e-01 -1.74330473e-01 -1.14944883e-01 -7.80586898e-01 -9.68421042e-01 5.90415478e-01 -1.02598429e+00 -1.47082835e-01 -6.07980192e-01 2.87476301e-01 -3.66293579e-01 5.07538617e-01 -4.64538485e-01 -1.09603477e+00 -3.88711929e-01 -4.39725012e-01 1.26158988e+00 -2.81363100e-01 -1.78445846e-01 -6.03823721e-01 1.89403459e-01 -1.10619459e-02 5.03724217e-01 7.71654487e-01 3.27996463e-01 -3.84442598e-01 -6.48146331e-01 -6.56315923e-01 -1.69881344e-01 -1.38665155e-01 2.37510577e-01 -9.78732109e-02 -6.65456653e-01 -5.50478458e-01 4.93611008e-01 2.57927477e-01 4.93945181e-01 9.92012843e-02 1.39349008e+00 -6.57220721e-01 -2.55329549e-01 5.66734135e-01 1.55065477e+00 6.79869205e-03 7.59323299e-01 2.28268672e-02 4.20238018e-01 3.08059752e-01 5.90411663e-01 1.24022496e+00 4.34762686e-01 5.11639297e-01 5.39859235e-01 4.84576374e-01 3.25671732e-01 -1.02796867e-01 4.60765868e-01 1.39488637e+00 -3.40286523e-01 -1.65565498e-02 -7.58604407e-01 1.13032198e+00 -2.02749205e+00 -1.29662514e+00 -2.89252043e-01 2.92185593e+00 8.83215547e-01 -1.08929528e-02 5.27281404e-01 1.09014165e+00 4.74679738e-01 -4.78758588e-02 -7.28759229e-01 2.24080402e-02 2.05997273e-01 4.31793839e-01 8.97422850e-01 4.70278263e-02 -7.11664259e-01 -1.48030624e-01 6.18970203e+00 5.07708013e-01 -9.54627991e-01 3.35513115e-01 4.17258590e-03 -4.00461048e-01 -3.23549002e-01 7.96624199e-02 -2.16821745e-01 7.46641040e-01 1.55667901e+00 -7.85126507e-01 8.27482879e-01 6.75000310e-01 4.98901159e-01 -4.23188880e-02 -1.02676618e+00 1.35998094e+00 -2.03549758e-01 -1.24165666e+00 -5.12923419e-01 1.94133490e-01 4.86318916e-01 1.55477524e-01 -4.65627521e-01 -2.17054859e-01 1.50555447e-01 -7.17648625e-01 7.52864420e-01 4.39714640e-01 5.95362782e-01 -1.06239605e+00 6.02881968e-01 6.67186141e-01 -1.64634347e+00 -1.93669960e-01 -3.51190090e-01 -3.23269367e-01 6.96978152e-01 1.52468169e+00 -4.42494452e-01 8.12944949e-01 9.12126839e-01 7.50924766e-01 -3.30887809e-02 1.23481810e+00 3.43159854e-01 9.19805050e-01 -1.09590268e+00 2.18120173e-01 -4.18160200e-01 6.67812005e-02 8.24869931e-01 7.19204843e-01 8.14987957e-01 2.48477429e-01 1.89490870e-01 3.13122511e-01 -1.12648740e-01 -2.31368989e-01 -6.03950500e-01 1.84718929e-02 9.12754178e-01 7.70598412e-01 -7.68723488e-01 -3.27596217e-01 -5.02183437e-01 8.43190312e-01 -1.58580676e-01 2.34068967e-02 -8.95354807e-01 -4.63026524e-01 4.43702012e-01 3.28819990e-01 5.39640903e-01 -6.05406880e-01 -1.72756940e-01 -1.27657735e+00 7.36800075e-01 -8.97712171e-01 4.52634007e-01 -3.28890800e-01 -1.55423641e+00 1.92739233e-01 -2.24838350e-02 -1.74056911e+00 -4.53402609e-01 2.53916055e-01 -2.75993943e-01 2.85640478e-01 -1.11873174e+00 -6.39646530e-01 -3.79801840e-01 7.71479249e-01 2.33390197e-01 5.53434789e-01 8.56728017e-01 6.05470002e-01 -1.44419119e-01 3.25512230e-01 5.30369759e-01 -3.46502542e-01 3.88085991e-01 -9.50188518e-01 5.22498727e-01 9.83301580e-01 1.11319441e-02 5.46260893e-01 8.42885554e-01 -6.28199577e-01 -2.00449538e+00 -1.07828593e+00 7.78819025e-01 -8.82254317e-02 9.39583778e-01 -2.03606889e-01 -9.76772606e-01 6.94398224e-01 -2.67381161e-01 1.23080790e-01 7.10826755e-01 -4.34587859e-02 -4.97260153e-01 -7.18186021e-01 -1.31813896e+00 1.45001918e-01 9.74075913e-01 -6.06176555e-01 -4.48129922e-01 2.98274457e-01 7.92330325e-01 -1.65585726e-01 -1.19037056e+00 2.78486282e-01 6.33109868e-01 -1.05131078e+00 9.61221039e-01 6.12654425e-02 1.73073001e-02 -6.07476234e-01 -3.61826986e-01 -7.77034640e-01 2.13657275e-01 -1.14108157e+00 -1.02923179e+00 1.14651680e+00 -8.68386030e-02 -8.97673488e-01 4.83653963e-01 4.03715312e-01 2.95090139e-01 -3.43772143e-01 -1.23624229e+00 -1.12913823e+00 -5.69767296e-01 -9.15525913e-01 1.11971462e+00 9.26762521e-01 2.54525989e-01 -1.68958291e-01 -5.29030740e-01 2.15886623e-01 1.07711804e+00 4.65970725e-01 8.49735141e-01 -1.49052429e+00 -4.92371947e-01 3.55126970e-02 -3.88744146e-01 -1.11084092e+00 -5.90346515e-01 -4.93887335e-01 -1.90091401e-01 -1.21865499e+00 3.82518955e-02 -5.49211442e-01 -2.36978874e-01 4.74580973e-01 3.12539160e-01 2.50551432e-01 -3.61994505e-02 5.82710266e-01 -5.37538826e-01 3.20836693e-01 3.84460658e-01 8.18392560e-02 -2.92760909e-01 4.48233664e-01 -4.18512583e-01 6.04275107e-01 7.77112782e-01 -7.64385521e-01 -4.48815376e-01 -3.70093763e-01 7.13765264e-01 7.36658096e-01 4.38813686e-01 -1.18239057e+00 4.13975716e-01 9.33630578e-03 1.82974245e-02 -1.07570136e+00 1.41464084e-01 -1.34635460e+00 7.35774159e-01 3.65056098e-01 2.29188517e-01 4.53777641e-01 1.42652795e-01 9.78763521e-01 -1.46755233e-01 1.59374341e-01 3.55160505e-01 1.70738623e-01 1.01371473e-02 4.73532528e-01 -3.12304366e-02 -1.54967681e-01 9.92567003e-01 -2.56928325e-01 -1.61088139e-01 -6.72746241e-01 -6.01925015e-01 2.77464855e-02 6.72902405e-01 1.67072579e-01 5.54477811e-01 -1.30921197e+00 -6.91709876e-01 -3.27358283e-02 -1.66255981e-01 2.39217281e-01 3.75550807e-01 1.11148608e+00 -4.62673098e-01 -4.46631014e-03 1.97558373e-01 -7.55207539e-01 -1.16323471e+00 7.77877152e-01 -2.09588170e-01 -4.01955664e-01 -1.12483406e+00 7.91369081e-02 -6.43863082e-01 5.55013604e-02 2.62131959e-01 -6.70829475e-01 4.75539148e-01 3.02214045e-02 1.26336586e+00 9.28189933e-01 4.44628388e-01 -1.45567313e-01 -3.15361768e-01 4.96191263e-01 2.63016909e-01 -2.46330053e-01 1.92123580e+00 -4.26921517e-01 -5.00759721e-01 8.36899221e-01 1.08097374e+00 3.19217116e-01 -1.07859695e+00 -1.71502009e-01 7.92026296e-02 -6.75954819e-01 -3.14904392e-01 -1.86674222e-01 -1.01066911e+00 4.64606792e-01 2.79762745e-01 1.00576329e+00 1.69166780e+00 -2.68023908e-01 1.53035903e+00 1.68186516e-01 1.04767072e+00 -7.14251637e-01 -4.18673933e-01 1.32703991e-03 5.51443815e-01 -7.29324996e-01 1.50020525e-01 -2.96494424e-01 1.81348965e-01 1.04759943e+00 -3.78707677e-01 -2.51336306e-01 8.29556465e-01 6.26186669e-01 -5.10392308e-01 1.47025332e-01 -9.95515287e-01 2.37111673e-01 -4.47025836e-01 4.18109566e-01 7.12697655e-02 2.52504528e-01 -2.18389109e-01 5.86511612e-01 -1.03990994e-01 3.41248333e-01 7.83336341e-01 1.34253442e+00 -3.57092828e-01 -1.15977705e+00 -6.29625618e-01 5.84689319e-01 -7.43233681e-01 2.80069858e-01 2.22157791e-01 2.07228020e-01 -4.28700358e-01 1.44774497e+00 1.74503878e-01 -1.66430950e-01 2.99413890e-01 -1.78272530e-01 4.00757253e-01 -5.23358658e-02 -3.33838046e-01 1.24490492e-01 -2.39915475e-01 -1.03933883e+00 -5.56132436e-01 -9.31873262e-01 -1.32585824e+00 -1.03261566e+00 -2.17336908e-01 1.37720648e-02 7.60905921e-01 8.98809731e-01 5.35670877e-01 4.06359494e-01 9.73569989e-01 -8.37087154e-01 -6.72467589e-01 -7.21168518e-01 -4.91366684e-01 9.79272276e-02 6.21155918e-01 -1.54089451e-01 -4.83509392e-01 2.94190079e-01]
[7.303322792053223, 3.0315585136413574]
8c09505a-af9a-4066-b6c4-4aece09ddafc
query-tracking-for-e-commerce-conversational
1810.03274
null
http://arxiv.org/abs/1810.03274v1
http://arxiv.org/pdf/1810.03274v1.pdf
Query Tracking for E-commerce Conversational Search: A Machine Comprehension Perspective
With the development of dialog techniques, conversational search has attracted more and more attention as it enables users to interact with the search engine in a natural and efficient manner. However, comparing with the natural language understanding in traditional task-oriented dialog which focuses on slot filling and tracking, the query understanding in E-commerce conversational search is quite different and more challenging due to more diverse user expressions and complex intentions. In this work, we define the real-world problem of query tracking in E-commerce conversational search, in which the goal is to update the internal query after each round of interaction. We also propose a self attention based neural network to handle the task in a machine comprehension perspective. Further more we build a novel E-commerce query tracking dataset from an operational E-commerce Search Engine, and experimental results on this dataset suggest that our proposed model outperforms several baseline methods by a substantial gain for Exact Match accuracy and F1 score, showing the potential of machine comprehension like model for this task.
['Yunlun Yang', 'Yu Gong', 'Xi Chen']
2018-10-08
null
null
null
null
['conversational-search']
['natural-language-processing']
[ 5.59041984e-02 3.26011419e-01 -4.68155026e-01 -4.56645578e-01 -4.83867586e-01 -5.70836365e-01 7.47380197e-01 1.67771339e-01 -5.29356062e-01 4.14602846e-01 4.44780201e-01 -4.50000316e-01 -5.02285399e-02 -5.50430536e-01 -7.58418441e-02 9.10897180e-03 3.57552290e-01 9.95453775e-01 4.04210001e-01 -8.27213407e-01 2.89462715e-01 -1.22334078e-01 -1.13851202e+00 1.81507900e-01 1.14363217e+00 1.04659665e+00 4.73156750e-01 4.78193223e-01 -5.94190955e-01 7.89224803e-01 -4.22435045e-01 -6.12587154e-01 -2.95907445e-02 -4.01106864e-01 -1.66227126e+00 4.42998338e-04 8.38742554e-02 -3.77541453e-01 -1.97627947e-01 9.46736038e-01 3.59237492e-01 4.61096227e-01 2.25720912e-01 -1.14765191e+00 -6.94236934e-01 5.25116324e-01 -3.24821509e-02 1.28451526e-01 8.82016599e-01 -2.16625053e-02 1.49946606e+00 -5.64652503e-01 6.78200006e-01 1.35019875e+00 2.55632520e-01 5.86306393e-01 -1.01728058e+00 -4.49850082e-01 3.37170959e-01 2.61702210e-01 -8.71320188e-01 -2.36424729e-01 7.18208671e-01 -1.61735192e-01 1.06337774e+00 5.20419776e-01 4.71827447e-01 9.86302137e-01 -1.75875577e-03 1.21740472e+00 7.07427442e-01 -4.92587984e-01 8.98045227e-02 7.10374057e-01 6.65168107e-01 4.70163763e-01 -4.66948628e-01 -1.89396650e-01 -5.07599413e-01 -2.34085351e-01 4.02045399e-01 2.18432829e-01 -1.69429794e-01 -3.90340596e-01 -1.06569314e+00 1.16769135e+00 4.62765604e-01 4.98854995e-01 -5.52334785e-01 -2.83285648e-01 4.42394495e-01 4.82002646e-01 4.30675894e-01 7.43069649e-01 -6.46351695e-01 -3.03467184e-01 -3.82815540e-01 2.82833695e-01 1.63663173e+00 1.11471760e+00 4.88280267e-01 -6.84317112e-01 -5.54057539e-01 1.20205736e+00 3.74325275e-01 1.81120381e-01 5.06006479e-01 -9.63406384e-01 4.20079708e-01 1.02340174e+00 3.18974346e-01 -7.68715262e-01 -3.28678936e-01 -2.60501206e-01 -6.59299076e-01 -4.97763634e-01 3.26274753e-01 -4.43488508e-02 -3.62038672e-01 1.72825301e+00 2.67525226e-01 -4.85431343e-01 8.47001225e-02 8.45183253e-01 9.96863365e-01 3.92808139e-01 4.74506348e-01 -1.60396665e-01 1.87510943e+00 -1.44726849e+00 -1.17521131e+00 -3.35014433e-01 3.87997210e-01 -7.67226577e-01 1.23002720e+00 -7.36796558e-02 -8.11653256e-01 -4.12340015e-01 -6.14847898e-01 -4.70943153e-01 -6.30151272e-01 -1.92347765e-01 9.96912539e-01 5.75275838e-01 -8.87289703e-01 -5.18070236e-02 -4.23087329e-01 -8.31023932e-01 -1.10920938e-02 4.50967818e-01 4.20899093e-02 3.06046218e-01 -1.58015513e+00 7.92282462e-01 2.84929633e-01 2.26754881e-02 -2.32265770e-01 -4.41257924e-01 -7.07187355e-01 5.83774626e-01 9.24078643e-01 -8.31162274e-01 2.15374303e+00 -5.52331924e-01 -1.70011723e+00 7.32129753e-01 -5.03723323e-01 -3.15739274e-01 3.24738979e-01 -4.00704294e-01 -1.72779724e-01 -1.58624172e-01 2.25283906e-01 7.99242258e-01 2.82026082e-01 -1.10069954e+00 -8.17938507e-01 -4.05480444e-01 5.61444581e-01 5.99645793e-01 -1.15759686e-01 8.23708028e-02 -8.10532749e-01 -2.12231919e-01 -1.52276810e-02 -8.34883034e-01 -2.53184170e-01 -2.14735389e-01 -3.21466208e-01 -9.47270513e-01 8.81613076e-01 -5.82863927e-01 1.41737843e+00 -1.68766713e+00 -2.94719785e-02 -1.99299470e-01 3.76621068e-01 3.09660792e-01 1.26655146e-01 9.07138586e-01 3.57951313e-01 1.35757536e-01 1.37323231e-01 -1.90011993e-01 3.90411198e-01 4.09480408e-02 -4.22470272e-01 -4.66457456e-01 -1.21275850e-01 1.38645017e+00 -9.35753942e-01 -5.83597600e-01 -4.65066284e-02 -7.89019391e-02 -6.58802092e-01 6.08930767e-01 -7.14419365e-01 3.91186237e-01 -9.66435373e-01 6.53020084e-01 2.56545007e-01 -8.01500440e-01 1.75541893e-01 1.08714148e-01 1.09257258e-01 3.94097775e-01 -5.76608717e-01 1.78375363e+00 -7.05595076e-01 4.44439471e-01 4.35876369e-01 -8.13415051e-01 6.59735560e-01 3.56295109e-01 4.02250409e-01 -1.22396171e+00 6.51616752e-02 -9.09379795e-02 -8.27937946e-02 -6.28254592e-01 9.11004066e-01 2.81801522e-01 -3.37588221e-01 7.96478868e-01 -7.82067701e-02 -7.38717765e-02 4.42475453e-02 4.18363690e-01 8.57994139e-01 -3.47240895e-01 4.26084995e-01 -2.18716785e-01 8.82710934e-01 3.81965905e-01 1.53991714e-01 1.13762057e+00 -4.44718689e-01 -1.84650213e-01 2.41403565e-01 -3.17217261e-01 -5.12629628e-01 -5.96008182e-01 9.37911272e-02 1.86724687e+00 6.95513844e-01 -1.74232855e-01 -7.92663276e-01 -8.42129469e-01 -7.40400851e-02 7.59034872e-01 -1.66198254e-01 7.72641646e-03 -4.98160630e-01 -4.03135002e-01 1.09046079e-01 1.61120832e-01 1.19754374e+00 -1.50813365e+00 -5.40242493e-01 2.90322006e-01 -5.14050603e-01 -1.22857976e+00 -9.68854606e-01 9.71558690e-02 -6.79554105e-01 -1.02459443e+00 -6.20967746e-01 -1.22706890e+00 3.44560929e-02 4.65928286e-01 1.54264092e+00 3.36376369e-01 -2.73410883e-02 8.25207293e-01 -4.04134810e-01 -4.20710921e-01 -3.78191829e-01 6.00410283e-01 -4.28812772e-01 -3.09238315e-01 9.97000754e-01 -5.16425893e-02 -9.20717776e-01 6.98920429e-01 -8.20522010e-01 8.10330510e-02 4.21881765e-01 1.12529182e+00 -1.87841430e-01 -2.05374762e-01 8.04809749e-01 -1.03033972e+00 1.61235023e+00 -5.30754387e-01 -4.44582522e-01 6.36807859e-01 -1.05643797e+00 5.84340207e-02 2.47918099e-01 -3.12350363e-01 -1.53096497e+00 -1.11230098e-01 -2.93144792e-01 3.08214039e-01 -1.00703910e-01 5.44922948e-01 6.61848485e-02 1.32292032e-01 3.85195613e-01 4.88114983e-01 4.59318794e-02 -5.09441853e-01 4.40824181e-01 9.13439989e-01 3.29373419e-01 -3.34043592e-01 3.88792187e-01 -5.59626482e-02 -7.14039803e-01 -8.12249541e-01 -6.94802403e-01 -1.10770667e+00 -3.88812304e-01 -1.57390460e-02 1.07950068e+00 -5.62546432e-01 -1.72353792e+00 3.05777371e-01 -1.30220854e+00 -1.89568475e-01 9.50738713e-02 7.82804936e-02 -4.21597332e-01 4.37411487e-01 -7.85795271e-01 -1.21900904e+00 -6.45610929e-01 -1.24873269e+00 1.34390056e+00 4.33827996e-01 -4.65643764e-01 -1.22175050e+00 1.66678920e-01 1.00178802e+00 8.39424610e-01 -7.70518422e-01 1.17700374e+00 -1.36516047e+00 -8.75378191e-01 -1.83420211e-01 -3.72303516e-01 -2.48437911e-01 1.17844418e-01 -1.00514305e+00 -8.02151203e-01 5.52359484e-02 3.63315135e-01 -3.97774905e-01 4.70881909e-01 -1.10777065e-01 7.59636581e-01 -4.76669729e-01 -4.72175568e-01 -9.01725665e-02 9.67537582e-01 6.80388093e-01 2.72010267e-01 4.34389591e-01 5.08095603e-03 9.71384406e-01 6.66934788e-01 1.22700430e-01 8.22740853e-01 1.14115202e+00 1.67582437e-01 -1.70159698e-01 3.16927582e-01 -4.49489832e-01 -1.69433221e-01 5.40694594e-01 4.87254888e-01 -5.70406795e-01 -7.76222110e-01 3.12492609e-01 -2.16077256e+00 -8.46258044e-01 4.49045062e-01 1.72688818e+00 9.42200541e-01 -2.09634811e-01 1.29024342e-01 -5.59923589e-01 5.83245218e-01 2.69992054e-01 -8.43043387e-01 -4.04159665e-01 4.05272037e-01 -2.22272072e-02 -5.35392985e-02 7.44162917e-01 -1.09669316e+00 1.35070527e+00 6.38013887e+00 6.05858147e-01 -7.07905352e-01 7.06202090e-02 7.70253479e-01 6.15633488e-01 -1.44338340e-01 -1.66519448e-01 -7.44233072e-01 3.38265061e-01 5.68897784e-01 -3.37058991e-01 6.08968079e-01 9.53475118e-01 9.35254898e-03 -3.82299781e-01 -1.30067468e+00 9.30506229e-01 -1.98323071e-01 -1.16252387e+00 -9.37229022e-02 1.76320206e-02 2.23579481e-01 -2.71039218e-01 -1.63646683e-01 1.01925814e+00 4.36634392e-01 -8.47217262e-01 -4.05762978e-02 3.20538431e-01 1.42023623e-01 -7.91033879e-02 8.70083809e-01 9.69796658e-01 -1.02882266e+00 -1.18540481e-01 1.68523118e-01 -6.45589605e-02 3.92250359e-01 -3.15597087e-01 -1.25960970e+00 2.02287495e-01 6.54048026e-01 1.24523163e-01 -2.91513056e-01 6.77367568e-01 2.47546390e-01 2.31413394e-01 -1.45415217e-01 -8.23360145e-01 3.70221883e-01 -5.50485134e-01 3.93821627e-01 1.10404873e+00 -1.79925531e-01 3.45693827e-01 4.07707632e-01 1.07174945e+00 -1.99544191e-01 3.68587166e-01 -4.13169563e-01 -1.93939745e-01 3.34349066e-01 1.14814866e+00 -4.94842559e-01 -3.72615308e-01 -6.10688746e-01 1.28044534e+00 2.18751177e-01 5.42576671e-01 -4.62916970e-01 -2.74689585e-01 5.20231485e-01 -1.64280340e-01 1.21913038e-01 -3.64349484e-02 -1.22476956e-02 -1.15069461e+00 1.25219926e-01 -1.00471902e+00 4.55050558e-01 -5.35579026e-01 -1.18536389e+00 7.78865218e-01 -2.29856037e-02 -6.59609735e-01 -9.10303056e-01 -3.27049822e-01 -6.36171758e-01 9.16704476e-01 -1.38558507e+00 -9.80351686e-01 -3.81252825e-01 3.63821179e-01 1.26938760e+00 -1.90105215e-01 9.95195091e-01 1.66914716e-01 -3.16958219e-01 5.26421666e-01 -1.52552322e-01 1.78581044e-01 5.00815928e-01 -1.24877501e+00 4.15104568e-01 1.06139585e-01 1.17068425e-01 9.74790633e-01 6.45159721e-01 -5.45084894e-01 -1.67103899e+00 -3.80541950e-01 1.28809774e+00 -6.20431066e-01 6.94697618e-01 -5.67497194e-01 -9.04575944e-01 6.65202200e-01 8.18438232e-01 -8.15267205e-01 6.07672095e-01 4.90799785e-01 2.60498319e-02 1.98598415e-01 -1.06463826e+00 6.54333770e-01 1.03878331e+00 -7.76042283e-01 -9.47063863e-01 6.59588993e-01 1.10166669e+00 -3.81325573e-01 -7.34743953e-01 2.65788168e-01 5.54560721e-01 -8.77768576e-01 1.13802671e+00 -7.57947505e-01 -1.11865349e-01 2.77101040e-01 1.07803956e-01 -9.50538278e-01 -3.28911655e-02 -8.84997904e-01 7.51443356e-02 1.05686700e+00 5.34572601e-01 -7.40599751e-01 9.84209836e-01 1.24708605e+00 1.90706819e-01 -7.39908934e-01 -8.04718494e-01 -2.42697224e-01 -1.86450824e-01 -8.76201019e-02 5.33954561e-01 6.74572706e-01 6.19632244e-01 1.19149899e+00 -2.80933887e-01 -2.96989828e-01 1.07235581e-01 4.83969301e-01 7.33431637e-01 -1.44630694e+00 -2.90502131e-01 -6.45363092e-01 3.40368003e-01 -2.22093630e+00 2.94465274e-01 -7.75786877e-01 1.74939722e-01 -1.41699255e+00 4.31878984e-01 -3.18647712e-01 1.58169284e-01 2.97507904e-02 -3.49719942e-01 -5.74122965e-01 1.77668974e-01 3.85627002e-01 -1.20026398e+00 8.64598393e-01 1.37448025e+00 -3.83894444e-01 -6.56957507e-01 5.06174743e-01 -8.95738661e-01 3.20782781e-01 3.83368284e-01 6.49263412e-02 -5.08990824e-01 -2.11143702e-01 9.84567851e-02 6.86200023e-01 -3.81586924e-02 -2.99663514e-01 8.44931960e-01 -2.95621362e-02 -2.80069381e-01 -5.93367875e-01 5.00003874e-01 -1.15212893e+00 -2.74624407e-01 4.08058137e-01 -9.41828787e-01 1.22595906e-01 -1.27510726e-01 6.57047510e-01 -5.52050710e-01 -3.14853489e-01 3.59841287e-02 -2.62047142e-01 -9.11648512e-01 2.71173149e-01 -4.36208189e-01 9.61458161e-02 6.73774660e-01 -7.99267218e-02 -4.14944619e-01 -1.06519544e+00 -6.54480875e-01 8.78796458e-01 -9.91447791e-02 8.71218264e-01 2.29331553e-01 -9.82025146e-01 -2.20011964e-01 1.92111835e-01 3.33327591e-01 -1.14182584e-01 9.08254646e-03 5.08710265e-01 -1.04077309e-01 1.25963998e+00 2.59982347e-01 -6.69649065e-01 -1.07417727e+00 4.46797848e-01 2.47863919e-01 -7.23079920e-01 -2.60584682e-01 5.97883105e-01 4.42226171e-01 -7.97965825e-01 7.85580814e-01 -2.32448742e-01 -5.89409351e-01 -7.12097585e-02 3.58906031e-01 7.64573067e-02 -2.00561598e-01 -3.44868034e-01 -3.92182320e-02 1.31075978e-01 -3.77317816e-01 -2.07775742e-01 8.75550807e-01 -5.57348847e-01 -1.35860786e-01 1.06834017e-01 9.88394916e-01 -2.37352476e-01 -4.89308566e-01 -7.51176238e-01 7.29987204e-01 -2.26797014e-01 -1.67969242e-01 -1.14021373e+00 -5.91821671e-01 4.75490957e-01 5.51799893e-01 8.66113782e-01 9.28644121e-01 4.19475257e-01 1.16345680e+00 1.18383729e+00 3.31285268e-01 -1.05902719e+00 2.28191286e-01 9.08338547e-01 9.51801538e-01 -1.81811953e+00 -5.08681178e-01 -5.90439916e-01 -7.99566269e-01 8.86666059e-01 8.50473404e-01 4.78925824e-01 6.05617642e-01 -3.43880206e-01 3.66499096e-01 -5.60454130e-01 -8.59225571e-01 -2.89148718e-01 3.15582603e-01 2.38281816e-01 5.78852057e-01 -2.86726147e-01 -4.13825482e-01 6.10301614e-01 -2.14018404e-01 4.23363373e-02 -2.12821588e-01 7.71629751e-01 -5.17917633e-01 -1.29231381e+00 1.09228089e-01 3.64932269e-01 -3.18380326e-01 -1.72497809e-01 -7.42370903e-01 8.05009067e-01 -7.62423337e-01 1.41367090e+00 -5.09291589e-02 -2.43798897e-01 5.28278649e-01 5.26186943e-01 -2.80073196e-01 -6.31845474e-01 -1.05447745e+00 3.02218348e-02 4.13139641e-01 -7.10334003e-01 -4.21229154e-01 -2.71667898e-01 -9.87807870e-01 9.91999656e-02 -6.27583742e-01 6.77278519e-01 5.22920430e-01 8.85642767e-01 5.69509685e-01 1.87053531e-01 5.13054073e-01 -3.24924529e-01 -9.42220330e-01 -1.23560119e+00 -2.34252289e-01 6.86203539e-01 2.23688796e-01 -5.23149312e-01 -1.20365597e-01 -3.15445930e-01]
[12.217796325683594, 7.814226150512695]
75a837e1-853c-48f6-ab06-af809e42e0e7
yh-technologies-at-activitynet-challenge-2018
1807.00686
null
http://arxiv.org/abs/1807.00686v1
http://arxiv.org/pdf/1807.00686v1.pdf
YH Technologies at ActivityNet Challenge 2018
This notebook paper presents an overview and comparative analysis of our systems designed for the following five tasks in ActivityNet Challenge 2018: temporal action proposals, temporal action localization, dense-captioning events in videos, trimmed action recognition, and spatio-temporal action localization.
['Xue Li', 'Ting Yao']
2018-06-29
null
null
null
null
['dense-captioning', 'spatio-temporal-action-localization']
['computer-vision', 'computer-vision']
[ 4.94295806e-01 -4.54721181e-03 -7.18498349e-01 -2.56364614e-01 -7.19275594e-01 -4.53731269e-01 9.07117963e-01 -2.65459061e-01 -6.54588044e-01 8.24561596e-01 1.30938840e+00 4.36879188e-01 -5.76585494e-02 2.26094171e-01 -5.67003548e-01 -6.09258175e-01 -9.26068664e-01 8.52042250e-03 6.44712031e-01 3.66157323e-01 4.00752276e-01 5.07647157e-01 -1.19784653e+00 8.48104954e-01 1.31472975e-01 1.09685206e+00 -2.35851645e-01 9.54138100e-01 3.69304508e-01 1.86576450e+00 -7.19917834e-01 9.92900576e-04 -9.12683643e-03 -6.92644060e-01 -1.08282542e+00 2.97756612e-01 5.91572583e-01 -6.20053828e-01 -1.05565012e+00 5.68453789e-01 2.41978481e-01 7.48909950e-01 4.31351602e-01 -1.63137472e+00 -3.14314663e-01 7.54446566e-01 -2.07286343e-01 9.74012375e-01 1.08831739e+00 5.20012796e-01 7.93289602e-01 -8.25477004e-01 1.06883192e+00 1.12851810e+00 5.97772419e-01 7.64766157e-01 -5.75154483e-01 -1.88464001e-01 4.55958426e-01 9.62803185e-01 -1.00338244e+00 -7.24345863e-01 4.60531443e-01 -5.50008416e-01 1.76144779e+00 2.81038079e-02 1.15835476e+00 2.02398157e+00 3.52046430e-01 1.63406432e+00 3.76954794e-01 1.63029462e-01 3.34671646e-01 -9.05198753e-01 -3.23538870e-01 3.54400396e-01 -5.10443747e-01 8.22156761e-03 -1.10674894e+00 -5.32508194e-02 8.09035361e-01 -1.21506499e-02 -9.42580700e-02 -2.79835314e-01 -1.89676821e+00 1.99873820e-01 1.90573320e-01 5.78907669e-01 -8.66533637e-01 8.04890931e-01 1.11387336e+00 1.75467897e-02 4.15844411e-01 4.20634598e-01 -4.99271631e-01 -1.06946719e+00 -1.00785851e+00 3.01116884e-01 3.87610406e-01 8.50818694e-01 -1.50597766e-01 2.18552142e-01 -1.10931170e+00 2.18269750e-01 -5.73682450e-02 2.12651998e-01 6.16016865e-01 -1.38620174e+00 8.42686594e-01 4.32742089e-01 5.15396476e-01 -5.14630258e-01 -4.51625347e-01 4.66495007e-01 -2.67913163e-01 -3.24439108e-01 2.30657130e-01 -6.91520795e-02 -9.31264281e-01 1.22222507e+00 2.25469738e-01 8.28508735e-01 2.09782217e-02 7.83143878e-01 7.61647403e-01 5.58951676e-01 8.08204353e-01 -3.61070722e-01 1.12798369e+00 -1.71432555e+00 -1.13955486e+00 -4.09910649e-01 7.66646862e-01 -3.68787557e-01 3.78437847e-01 -6.27763793e-02 -1.25930130e+00 -5.47875881e-01 -6.87958896e-01 -1.80703271e-02 -2.26703689e-01 6.16126060e-01 8.13345730e-01 -8.94189402e-02 -8.74462187e-01 6.35694385e-01 -1.50530326e+00 -7.53148854e-01 6.75729096e-01 6.67133406e-02 -9.09746468e-01 2.44210377e-01 -9.03383911e-01 1.14011550e+00 8.57064784e-01 2.92673241e-02 -1.46925950e+00 -3.50188434e-01 -1.04234815e+00 -2.46488065e-01 4.62879151e-01 -1.44473433e-01 1.63219357e+00 -1.03990805e+00 -1.37595499e+00 8.42673838e-01 -8.45720097e-02 -1.02909303e+00 5.53915143e-01 -5.67896247e-01 -7.14923024e-01 5.87326527e-01 7.13803023e-02 8.98794591e-01 5.18523693e-01 8.40052590e-02 -8.86072218e-01 5.74249402e-02 9.44184437e-02 3.34794372e-01 9.18543413e-02 6.25327885e-01 -3.74318928e-01 -9.62290287e-01 -1.22148924e-01 -7.05591977e-01 -3.65723193e-01 -2.23215353e-02 -8.98827314e-02 -7.94069588e-01 8.36721599e-01 -7.78892696e-01 1.18081200e+00 -2.38061357e+00 2.51559585e-01 -6.60092890e-01 -1.69903427e-01 1.28472254e-01 -3.84605706e-01 5.61402023e-01 -2.66142040e-01 -4.28138524e-01 9.13590863e-02 -2.54348814e-01 2.11145028e-01 2.95393884e-01 -3.48442465e-01 6.31638944e-01 1.60854533e-01 1.46510243e+00 -1.34295690e+00 -9.09801781e-01 8.01365793e-01 2.81866729e-01 -6.46133274e-02 2.47495204e-01 -3.21448892e-01 7.87606239e-01 -5.62515974e-01 1.05895948e+00 -5.23464978e-02 7.42465779e-02 2.18166813e-01 -3.36841911e-01 -2.91836649e-01 4.88729030e-01 -1.06097436e+00 2.11859441e+00 6.97113797e-02 9.42158639e-01 -3.25823456e-01 -6.23319924e-01 7.66054615e-02 7.54382133e-01 1.51477051e+00 -7.92711079e-01 1.46687431e-02 -2.43796259e-01 -4.75132018e-01 -9.18723047e-01 3.34459126e-01 4.85155761e-01 -1.00557648e-01 4.34035301e-01 6.81892812e-01 5.24049282e-01 7.65844047e-01 1.19799770e-01 1.80558181e+00 1.01665258e+00 8.12521994e-01 3.65649194e-01 5.54444432e-01 2.16203108e-01 3.67737561e-01 6.83009982e-01 -1.22731614e+00 4.28430825e-01 4.33721781e-01 -9.89976883e-01 -7.04561889e-01 -9.73296165e-01 4.48151261e-01 1.69900930e+00 1.76057275e-02 -6.71082556e-01 -4.33391601e-01 -1.16645515e+00 -5.51818311e-01 4.97940838e-01 -1.08032870e+00 -6.61448911e-02 -1.17479706e+00 -5.32559715e-02 6.67152286e-01 1.25232160e+00 8.16784799e-01 -2.07940745e+00 -9.48695302e-01 3.29034656e-01 -4.44276541e-01 -1.44246733e+00 -7.61354148e-01 8.65272358e-02 -6.46415234e-01 -1.46894205e+00 -7.93879211e-01 -8.43906999e-01 1.33913904e-01 1.43867088e-02 1.04434562e+00 -7.22915769e-01 -2.20539480e-01 9.20823872e-01 -6.38829887e-01 -4.47366312e-02 2.40690149e-02 -3.28242481e-01 2.21463740e-02 6.19268492e-02 7.72749066e-01 -5.68955064e-01 -8.21721911e-01 4.85681474e-01 -6.32580578e-01 -1.99093923e-01 5.55094659e-01 6.03557490e-02 5.12398183e-01 -4.98135507e-01 -1.44479752e-01 1.89886779e-01 2.20030751e-02 -3.50844204e-01 -1.95037127e-02 7.37511218e-01 4.25009251e-01 -4.08287078e-01 -1.63022846e-01 -8.96259785e-01 -9.88424599e-01 7.17914701e-01 -1.16958089e-01 -5.36961436e-01 -3.71975839e-01 -5.90569563e-02 7.01129511e-02 -3.87625769e-02 6.51896298e-01 6.40019357e-01 -4.63654518e-01 -2.58099526e-01 4.84572500e-01 1.43316966e-02 7.43261397e-01 -8.38661566e-03 1.40406951e-01 7.59611249e-01 -3.30831558e-01 -4.94379610e-01 -1.02205479e+00 -6.95022047e-01 -1.14171994e+00 -7.51799047e-01 1.62893081e+00 -1.01782298e+00 -7.63749897e-01 4.98181850e-01 -1.08084857e+00 -5.64614236e-01 -7.31219113e-01 9.90762353e-01 -1.32524431e+00 2.58310884e-01 -6.10194266e-01 -5.28423846e-01 -1.63772151e-01 -7.55168557e-01 1.28810310e+00 -6.20223545e-02 -7.16614306e-01 -8.98003519e-01 5.09195745e-01 3.04072291e-01 4.54663821e-02 6.34406388e-01 -1.87113091e-01 -5.89672923e-01 -5.28885782e-01 -5.27359545e-01 1.36774525e-01 1.75377339e-01 1.46322638e-01 -2.48374939e-02 -5.15154481e-01 7.78333768e-02 -3.18682581e-01 -7.26580501e-01 1.01439214e+00 8.44643712e-01 1.13644898e+00 -3.80784482e-01 -6.43165886e-01 4.97588217e-01 5.35218120e-01 4.37963009e-01 8.06481481e-01 1.32020682e-01 5.79815209e-01 8.39963332e-02 1.15327430e+00 6.73846245e-01 -4.40710746e-02 8.08296859e-01 4.27410871e-01 2.62901932e-01 -2.15015098e-01 -2.23626986e-01 6.91105187e-01 -1.44483969e-01 -4.59837854e-01 -3.86659890e-01 -6.44561648e-01 7.87759185e-01 -2.36648941e+00 -1.69911277e+00 2.04051226e-01 1.34287155e+00 3.10254365e-01 -3.26876268e-02 5.59098601e-01 -2.47618198e-01 5.06737053e-01 8.92086923e-01 -6.78737760e-01 6.18913844e-02 3.80479870e-03 -9.49184224e-02 4.99578387e-01 -1.08304448e-01 -1.97938430e+00 1.02448249e+00 8.18144417e+00 5.81176221e-01 -4.99730855e-01 4.89570469e-01 -4.45236266e-02 -4.52752441e-01 7.54672229e-01 -2.15136781e-01 -3.06221962e-01 2.91381419e-01 8.81494641e-01 -1.09361939e-01 -8.42498019e-02 1.11970592e+00 3.95749658e-01 -3.63175571e-01 -1.37949944e+00 1.09129071e+00 3.88078988e-01 -1.44181252e+00 -1.74358159e-01 -3.80991399e-01 7.01149762e-01 3.84064525e-01 -4.39367801e-01 5.84720969e-01 2.34498620e-01 -7.53452659e-01 8.95022810e-01 7.53897965e-01 3.46654385e-01 -1.32135645e-01 4.33826536e-01 -7.09170625e-02 -1.51631451e+00 -4.34783190e-01 2.99147815e-01 -1.76287927e-02 7.77463138e-01 -3.18261236e-01 -4.45474595e-01 1.88363671e-01 7.19393730e-01 1.71320093e+00 -4.92425650e-01 1.30604160e+00 -4.55728501e-01 6.26437247e-01 -8.84430110e-03 1.20921314e-01 7.58459449e-01 2.92240351e-01 7.86927521e-01 1.38037324e+00 7.30987564e-02 3.20767850e-01 4.18907434e-01 3.35052088e-02 2.66581714e-01 -2.41074935e-01 -5.55546582e-01 -7.15777695e-01 -1.74563393e-01 9.59570944e-01 -7.68484414e-01 -6.31086648e-01 -4.36603516e-01 1.28898203e+00 4.46150899e-02 2.94140309e-01 -1.42671680e+00 4.31743935e-02 7.79385924e-01 -4.38243560e-02 3.87830913e-01 -4.67138022e-01 4.14440691e-01 -1.25284982e+00 7.99582154e-02 -4.58402485e-01 9.48393941e-01 -1.06488347e+00 -7.58328021e-01 1.24738082e-01 4.35058117e-01 -1.61953962e+00 -4.00570124e-01 -3.89984101e-01 -6.34124756e-01 -1.28837258e-01 -5.09286106e-01 -1.12136507e+00 -2.57724941e-01 7.15049744e-01 1.13056076e+00 -8.32082778e-02 4.37032491e-01 4.29529816e-01 -4.52754170e-01 4.04892787e-02 -4.03563023e-01 3.18175405e-01 5.40444195e-01 -1.12530017e+00 6.79647923e-01 8.56928229e-01 4.61124629e-01 -5.70030473e-02 5.98772049e-01 -8.21153045e-01 -1.04027760e+00 -1.24915183e+00 9.41856086e-01 -9.72494662e-01 1.14127183e+00 -2.57721126e-01 -4.54342991e-01 1.30938160e+00 2.52086580e-01 2.15236574e-01 3.29317570e-01 -3.57087910e-01 -1.25079844e-02 3.12956393e-01 -8.30773532e-01 6.16535485e-01 1.59318864e+00 -3.57894212e-01 -7.78998673e-01 1.27234054e+00 5.76426387e-01 -4.77343976e-01 -7.08271503e-01 3.79731625e-01 6.84010029e-01 -6.15315437e-01 1.11785638e+00 -1.13827813e+00 3.45557421e-01 -2.28285044e-01 3.43570635e-02 -4.98179436e-01 -4.48920876e-01 -8.41849625e-01 -8.69916558e-01 5.56460738e-01 -1.48143023e-01 2.51623660e-01 9.06357348e-01 2.29156494e-01 -3.93653899e-01 -2.57408082e-01 -1.41733670e+00 -9.37832296e-01 -7.64862001e-01 -5.11923194e-01 3.64586599e-02 7.81491339e-01 3.01685274e-01 -1.91761151e-01 -6.83289886e-01 -2.88887113e-01 1.85811877e-01 -3.20259154e-01 5.26756167e-01 -2.58890003e-01 3.12151127e-02 -4.33684975e-01 -9.43777621e-01 -1.19841373e+00 4.20536995e-01 -2.50494868e-01 2.56970584e-01 -1.80094588e+00 2.62629598e-01 1.01541018e+00 -5.09425581e-01 9.60614741e-01 3.94349366e-01 5.01420915e-01 -2.30026208e-02 2.00639162e-02 -1.84735215e+00 7.61592984e-01 1.08300436e+00 -2.43464515e-01 -1.72563791e-01 1.69824108e-01 3.72391015e-01 8.11249077e-01 3.98348421e-01 -4.18491304e-01 -1.63601130e-01 -1.14455469e-01 -2.02164009e-01 2.42053106e-01 6.74703121e-01 -1.50611961e+00 2.00268656e-01 -6.43779933e-01 4.28426355e-01 -1.03253436e+00 7.09690750e-01 -6.34755313e-01 -3.45467404e-02 6.23360097e-01 -6.07611656e-01 9.60122421e-02 9.62825865e-03 7.11133838e-01 -3.01432997e-01 3.28102648e-01 6.05094910e-01 -3.65292490e-01 -1.65107894e+00 2.92683065e-01 -1.02632654e+00 -5.73109761e-02 1.72169995e+00 -6.20640874e-01 -2.63384163e-01 -3.55823696e-01 -1.46235085e+00 3.72670352e-01 -3.36381286e-01 8.97798955e-01 5.67000508e-01 -1.58509791e+00 -4.90329295e-01 -4.76787478e-01 3.48000050e-01 -9.35385048e-01 6.88222468e-01 1.51873493e+00 -4.29748565e-01 8.14903259e-01 -6.43269897e-01 -4.80935633e-01 -1.43340051e+00 6.51723504e-01 5.49839079e-01 -2.29619429e-01 -8.94045174e-01 9.51298416e-01 -1.26406893e-01 2.30042011e-01 7.81578541e-01 -4.85969484e-01 -5.78112781e-01 1.83833800e-02 8.35800171e-01 7.53408968e-01 -4.43642706e-01 -7.81038344e-01 -8.22562158e-01 1.97761171e-02 2.81925112e-01 -1.11042216e-01 1.01994634e+00 4.17743586e-02 9.52071697e-02 5.83440125e-01 7.53748953e-01 -8.46830845e-01 -1.78607118e+00 -1.09720945e-01 3.57424051e-01 -4.18363869e-01 -2.60423571e-01 -8.89275432e-01 -7.12557852e-01 4.96938407e-01 5.23468494e-01 -1.76381916e-01 1.10263038e+00 3.91403139e-01 7.69397259e-01 6.81769431e-01 2.20712259e-01 -1.46889508e+00 7.80540764e-01 6.62757695e-01 1.07840526e+00 -1.09474313e+00 1.83547288e-01 4.39450797e-03 -1.10563540e+00 8.75542343e-01 9.70605791e-01 -1.55299515e-01 6.66995049e-01 -1.61329001e-01 -2.82960624e-01 -4.71545696e-01 -9.81254697e-01 -5.61039746e-01 4.60343987e-01 5.39696157e-01 1.54344663e-01 -1.86872557e-01 -1.86381906e-01 9.47225094e-03 5.96347809e-01 3.48909907e-02 1.45398259e-01 1.40268040e+00 -3.61988872e-01 -3.74827951e-01 1.77708596e-01 1.52326390e-01 -4.68810856e-01 2.14749649e-01 -8.52157593e-01 8.18608582e-01 1.10733226e-01 7.73693442e-01 3.89658511e-01 -1.35319859e-01 4.50093806e-01 2.81609714e-01 8.51979077e-01 -5.24597168e-01 -6.12091660e-01 -1.33678302e-01 2.88480669e-01 -1.61384189e+00 -1.19588816e+00 -1.12250507e+00 -9.83265758e-01 3.09902817e-01 4.25904006e-01 -2.47952901e-02 3.19472790e-01 1.25126159e+00 2.90577382e-01 4.64522511e-01 -1.01790112e-02 -1.12228608e+00 -3.77800204e-02 -1.47138822e+00 -4.19025153e-01 3.88197631e-01 2.71961421e-01 -6.93865716e-01 -3.47535372e-01 3.76653492e-01]
[8.262192726135254, 0.4519444406032562]
78357169-0af5-4ec3-9134-77a3d6f55ade
comparing-acoustic-based-approaches-for
2106.01555
null
https://arxiv.org/abs/2106.01555v2
https://arxiv.org/pdf/2106.01555v2.pdf
Comparing Acoustic-based Approaches for Alzheimer's Disease Detection
Robust strategies for Alzheimer's disease (AD) detection are important, given the high prevalence of AD. In this paper, we study the performance and generalizability of three approaches for AD detection from speech on the recent ADReSSo challenge dataset: 1) using conventional acoustic features 2) using novel pre-trained acoustic embeddings 3) combining acoustic features and embeddings. We find that while feature-based approaches have a higher precision, classification approaches relying on pre-trained embeddings prove to have a higher, and more balanced cross-validated performance across multiple metrics of performance. Further, embedding-only approaches are more generalizable. Our best model outperforms the acoustic baseline in the challenge by 2.8%.
['Jekaterina Novikova', 'Aparna Balagopalan']
2021-06-03
null
null
null
null
['alzheimer-s-disease-detection']
['medical']
[ 6.30979910e-02 5.64685091e-04 3.19619365e-02 -4.58453685e-01 -1.61805713e+00 -1.88552096e-01 6.49959505e-01 3.15827817e-01 -7.35032499e-01 3.91456097e-01 7.71145463e-01 2.54488271e-02 -1.04148082e-01 -4.85263973e-01 -5.29997945e-02 -3.80442828e-01 -3.99106383e-01 4.47313845e-01 3.79751444e-01 -2.14381330e-02 -3.30280215e-01 2.45575786e-01 -1.30435526e+00 3.40260297e-01 3.89834911e-01 1.19610131e+00 -2.40082089e-02 7.91362226e-01 1.42692789e-01 2.52171397e-01 -5.75080156e-01 -3.50108385e-01 -9.23822671e-02 5.34200929e-02 -7.46311903e-01 -3.63074005e-01 6.89012647e-01 -7.29279995e-01 -5.85675597e-01 7.98254430e-01 9.29930329e-01 -2.19357789e-01 6.93651974e-01 -9.46868122e-01 -7.55870581e-01 3.20968479e-01 2.04831194e-02 7.83802927e-01 5.71603835e-01 8.68653208e-02 1.28993607e+00 -7.94831932e-01 3.88145208e-01 1.40979779e+00 9.92075086e-01 7.69864559e-01 -1.40365386e+00 -4.25007939e-01 1.92134231e-01 3.43582302e-01 -1.08197379e+00 -8.17170739e-01 3.97470832e-01 -5.79054236e-01 1.07382464e+00 8.37821513e-02 7.54783928e-01 1.69156468e+00 1.06366582e-01 4.10651028e-01 1.03874564e+00 -2.53484815e-01 2.76579618e-01 1.53156042e-01 5.71523011e-01 4.78360713e-01 2.50248909e-01 1.04887731e-01 -2.23133773e-01 -9.91698563e-01 1.95061892e-01 2.26842668e-02 -1.20027594e-01 1.57816038e-01 -1.22677553e+00 1.07393289e+00 6.46082982e-02 4.21675175e-01 -4.53432232e-01 3.17634135e-01 6.64238870e-01 2.01143250e-01 5.96620142e-01 3.22218001e-01 -4.66399819e-01 -3.11728954e-01 -9.04178977e-01 5.56238592e-01 5.81277490e-01 3.41373891e-01 -5.18416762e-02 -1.97683293e-02 -7.59932622e-02 1.33211458e+00 7.45220363e-01 5.82059741e-01 9.61298645e-01 -9.38269973e-01 3.59924257e-01 2.58715421e-01 -1.25853390e-01 -6.84631467e-01 -7.12125182e-01 -2.88309693e-01 -2.86786437e-01 -4.46620770e-02 4.01269644e-01 -3.09975803e-01 -9.51264203e-01 1.77053368e+00 5.31031899e-02 -1.75446063e-01 1.25941172e-01 6.90050304e-01 6.66281879e-01 5.16763330e-01 4.98610914e-01 1.43554449e-01 1.87977660e+00 -5.52167118e-01 -6.27256989e-01 -5.42563915e-01 6.85017884e-01 -6.68791771e-01 8.74310434e-01 3.04288328e-01 -6.17967665e-01 -2.58306563e-01 -1.22550023e+00 -8.98795202e-02 -3.56503308e-01 1.19699597e-01 3.41459572e-01 1.15353072e+00 -1.23825800e+00 2.70864487e-01 -1.02131999e+00 -6.86676979e-01 5.12658834e-01 1.42590448e-01 -5.07846713e-01 -1.12270854e-01 -1.16514409e+00 1.00218105e+00 -8.65741372e-02 -2.94549346e-01 -7.29882658e-01 -8.86780918e-01 -6.41396701e-01 -2.85899937e-01 -3.51632625e-01 -7.31668890e-01 1.48706961e+00 -4.75898802e-01 -1.08681595e+00 7.90543377e-01 -1.37588799e-01 -8.46691251e-01 3.42261940e-01 -6.33316576e-01 -8.43323290e-01 5.92860937e-01 1.53010994e-01 6.46966040e-01 5.55366635e-01 -6.86045647e-01 -6.54323518e-01 -6.54520035e-01 -3.22830856e-01 -4.65287775e-01 -6.03010774e-01 4.02733296e-01 3.57552290e-01 -8.32962394e-01 1.32472850e-02 -1.00602067e+00 -1.95341617e-01 4.15772140e-01 -1.75943717e-01 -4.20379162e-01 7.77155638e-01 -1.08015430e+00 1.09918010e+00 -2.19588208e+00 -4.94016618e-01 -1.18920438e-01 6.42802000e-01 3.43286335e-01 -3.79595235e-02 1.49619952e-01 -1.16535798e-01 2.95182705e-01 -1.97611138e-01 -3.27308714e-01 1.41973376e-01 8.65878239e-02 -1.37739107e-01 4.07757640e-01 6.40783489e-01 6.01728082e-01 -6.84589207e-01 -4.59518760e-01 6.98593482e-02 8.06489587e-01 -8.23754251e-01 -3.28234136e-02 2.05779955e-01 -2.10688382e-01 -5.13837516e-01 5.92050672e-01 2.20044717e-01 -9.40934047e-02 8.39150771e-02 -3.12154770e-01 2.24830598e-01 7.50855863e-01 -8.49235475e-01 1.12065995e+00 -3.68383139e-01 7.14721918e-01 -9.11755562e-02 -9.78634238e-01 8.29943955e-01 5.84985197e-01 6.02592170e-01 -5.55251539e-01 -1.37862399e-01 3.83870900e-01 2.63184279e-01 -6.58650041e-01 -1.48889750e-01 -1.56983271e-01 1.32376462e-01 2.62214601e-01 1.89335763e-01 5.76968133e-01 -1.94659323e-01 1.04020327e-01 1.72257054e+00 -4.70205098e-01 2.88067311e-01 -3.03355157e-01 2.45046467e-01 -2.00809613e-01 4.35654283e-01 6.40887320e-01 -8.09544325e-01 8.40066016e-01 3.52773190e-01 -2.91294873e-01 -1.00474048e+00 -1.33044887e+00 -7.09429920e-01 8.97168458e-01 -6.56505287e-01 -8.17084908e-01 -4.88956749e-01 -7.54858732e-01 2.31059194e-01 5.23499727e-01 -5.91015875e-01 -2.35376731e-01 -5.42869508e-01 -1.25135100e+00 1.03219688e+00 9.08582568e-01 1.99667498e-01 -4.29759979e-01 -4.42472160e-01 4.02695358e-01 -1.14228591e-01 -1.28992736e+00 -3.35262865e-01 2.72061408e-01 -9.43713546e-01 -7.96332538e-01 -1.14759827e+00 -8.10604572e-01 6.51481450e-02 -7.23785460e-02 8.69233966e-01 -3.81406486e-01 -4.38375413e-01 9.56793606e-01 -3.80641371e-01 -3.10067385e-01 -4.49209005e-01 -9.14285108e-02 3.50412488e-01 -1.66162878e-01 7.89639950e-01 -6.56379223e-01 -7.12574542e-01 1.54009208e-01 -4.12185997e-01 -1.11237431e+00 7.20158219e-01 6.57107055e-01 5.05952477e-01 -4.97155756e-01 1.08757007e+00 -2.32792199e-01 8.20032656e-01 -5.52211702e-01 2.20007077e-01 -1.86543867e-01 -8.51088941e-01 -4.83634472e-02 2.48905241e-01 -5.87861121e-01 -4.55912173e-01 2.56904103e-02 -6.72073781e-01 -2.61323769e-02 -4.90753174e-01 1.08287632e-01 -1.24818515e-02 2.48344317e-01 8.79204512e-01 -2.70800628e-02 3.39286119e-01 -7.44385421e-01 9.54812840e-02 1.02407229e+00 2.87106544e-01 -4.26092446e-01 4.43670869e-01 4.13036972e-01 -6.26219988e-01 -1.11797464e+00 -4.97256488e-01 -5.57806134e-01 -3.32583427e-01 1.65135056e-01 1.27963126e+00 -1.06480336e+00 5.68458810e-02 2.45364681e-01 -1.21517754e+00 -2.62727998e-02 -9.71333012e-02 9.26936567e-01 -2.58624762e-01 5.05193949e-01 -6.25145912e-01 -8.88746321e-01 -4.20557648e-01 -1.24055874e+00 1.28859842e+00 -4.25897509e-01 -9.15587604e-01 -9.12495494e-01 6.42028034e-01 4.65746522e-01 5.01152635e-01 1.05728261e-01 1.02554750e+00 -1.39032745e+00 1.81670472e-01 -4.47633982e-01 -4.35135424e-01 6.85916901e-01 3.02726686e-01 -2.36164153e-01 -1.28597152e+00 -6.96952045e-02 -5.63154779e-02 -1.12455204e-01 1.04898179e+00 3.99865359e-01 5.50557375e-01 -2.43299335e-01 -5.42404592e-01 -8.90234411e-02 1.05898643e+00 2.91387677e-01 6.37587488e-01 6.11697137e-01 1.29034996e-01 3.27903301e-01 -8.82048011e-02 1.77036256e-01 4.56128657e-01 8.74795556e-01 3.75348292e-02 2.82638520e-01 -4.78799164e-01 2.19997138e-01 9.10200953e-01 7.08808601e-01 1.52401626e-01 -8.39627348e-03 -1.00471878e+00 7.84009159e-01 -1.37761366e+00 -9.68236506e-01 -1.49856240e-01 1.84660566e+00 6.52811229e-01 2.26452872e-01 7.65443683e-01 3.08218300e-01 6.40815377e-01 2.71206349e-01 -7.86326230e-02 -3.04010630e-01 9.94757004e-03 3.62826288e-01 2.48424664e-01 1.81310460e-01 -1.34569812e+00 6.49261773e-02 7.87733221e+00 2.80227393e-01 -8.18502665e-01 5.89918971e-01 3.24344188e-01 -2.66878277e-01 -2.42257826e-02 -6.79534733e-01 -9.87889290e-01 7.85595298e-01 1.66379344e+00 1.59661725e-01 -1.41065046e-01 9.17755008e-01 4.88391556e-02 2.19083741e-01 -1.30948031e+00 9.49255884e-01 1.48375645e-01 -9.47416067e-01 1.67322949e-01 3.26230466e-01 -1.37276649e-02 4.83954996e-01 1.46458372e-01 1.99586779e-01 -1.15254506e-01 -7.96480417e-01 5.32455444e-01 3.83440435e-01 4.89522845e-01 -3.51988822e-01 9.11995292e-01 -5.27063549e-01 -9.68592644e-01 -2.56844759e-01 -2.18154415e-01 4.18896884e-01 3.13351631e-01 8.55366707e-01 -8.31866622e-01 1.62862707e-02 9.14187372e-01 5.36634922e-01 -7.78510988e-01 1.16698885e+00 2.43686512e-01 1.06907237e+00 -5.57683229e-01 -6.36827126e-02 1.68523088e-01 3.46052468e-01 6.91265106e-01 1.60743093e+00 3.50656062e-01 -2.01951489e-01 4.92936783e-02 6.16320789e-01 3.51271838e-01 8.68423805e-02 -5.85427225e-01 -4.24720824e-01 2.68127978e-01 8.98838878e-01 -1.63922653e-01 -2.37902194e-01 -7.83255041e-01 6.51872218e-01 -5.88134453e-02 9.87388194e-03 -7.28797019e-01 -4.29879904e-01 1.07893670e+00 3.38652343e-01 2.13486359e-01 -4.07919586e-01 1.38235509e-01 -6.98532820e-01 3.23311508e-01 -8.64640474e-01 6.15338504e-01 -3.74101937e-01 -1.59983397e+00 6.73896194e-01 -1.47654563e-01 -8.92147362e-01 -2.32644692e-01 -1.02117801e+00 -4.83555466e-01 5.66595852e-01 -1.25364518e+00 -8.83057117e-01 -7.58992285e-02 1.52456895e-01 5.11319757e-01 -4.09323335e-01 1.21653342e+00 8.36445451e-01 -5.55199087e-01 6.75439060e-01 3.71451266e-02 2.57322311e-01 9.30023432e-01 -1.32007444e+00 3.71972322e-01 2.54688889e-01 1.50984168e-01 6.20638013e-01 6.02478802e-01 -5.18923998e-01 -1.06540501e+00 -1.03411329e+00 8.80197167e-01 -6.89564288e-01 1.04199827e+00 -3.51862371e-01 -9.22420979e-01 7.62421489e-01 -1.12465650e-01 -1.22385606e-01 9.93980050e-01 3.46974492e-01 -6.08205795e-01 -2.15659037e-01 -1.24761581e+00 2.75294751e-01 9.58698690e-01 -8.68961632e-01 -1.10942304e+00 5.71124852e-01 7.52332509e-01 4.40762579e-01 -1.25648332e+00 2.48851851e-01 1.01300609e+00 -4.67755973e-01 1.34628189e+00 -7.29685009e-01 3.74291807e-01 1.96238548e-01 -8.06374431e-01 -1.16520858e+00 -4.14548010e-01 -2.88310915e-01 7.86157846e-02 1.30110419e+00 6.98818803e-01 -1.17667401e+00 5.04980624e-01 4.19053465e-01 5.24613867e-03 -8.63980353e-01 -1.27341306e+00 -1.10583448e+00 3.15260649e-01 -7.46713042e-01 7.28361681e-02 4.49231923e-01 1.05158545e-01 2.22113863e-01 3.98582853e-02 3.65795583e-01 3.99456799e-01 -8.93929005e-01 6.94293156e-02 -1.58603430e+00 -1.64447010e-01 -4.42987293e-01 -1.21852422e+00 -5.27308524e-01 8.59708991e-03 -9.72897589e-01 -3.01969975e-01 -1.50230026e+00 2.78270930e-01 -3.38845819e-01 -5.54706514e-01 4.13219631e-01 -5.71820959e-02 3.17926466e-01 -1.16014004e-01 8.01562071e-02 -3.47716063e-01 4.74954695e-01 3.21267486e-01 -4.93953735e-01 -1.59209147e-01 -1.69583723e-01 -7.76190221e-01 5.52488148e-01 8.92532527e-01 -5.67874849e-01 -1.94613278e-01 -3.24443668e-01 -2.84066916e-01 -6.82158589e-01 8.58116686e-01 -1.23394895e+00 -2.13433951e-01 5.20051539e-01 4.75397974e-01 -1.35086372e-01 7.55682588e-01 -3.99354011e-01 -3.15137774e-01 6.17322624e-01 -4.70991313e-01 2.62899131e-01 2.60080606e-01 8.56643975e-01 7.44149238e-02 -3.43699157e-01 8.20488691e-01 9.14441571e-02 -3.06361526e-01 2.69686282e-02 -8.53340924e-01 2.80002981e-01 6.06872261e-01 -3.82302664e-02 -4.31554586e-01 -2.44437173e-01 -1.19240189e+00 -2.75278121e-01 -1.93380758e-01 5.23178339e-01 5.06983757e-01 -1.50755489e+00 -7.46030867e-01 -9.79461521e-02 3.58850032e-01 -8.59546542e-01 -1.06841050e-01 1.13980341e+00 -1.75091892e-01 5.51327229e-01 1.25027493e-01 -6.48010850e-01 -1.25382233e+00 1.60230264e-01 6.38947427e-01 8.46947059e-02 -1.03827083e+00 5.45085251e-01 3.59052829e-02 -2.56246209e-01 3.23453873e-01 -4.83162999e-01 -1.12262778e-02 4.57588196e-01 1.04984474e+00 6.89089358e-01 3.45805824e-01 -4.99603063e-01 -8.65791142e-01 3.04972649e-01 -3.57651502e-01 -3.45886320e-01 1.40700769e+00 2.32438385e-01 3.08367342e-01 5.44134259e-01 1.52967501e+00 -5.39286882e-02 -5.62523723e-01 -1.16462419e-02 3.56140494e-01 -3.15043069e-02 3.71294439e-01 -8.10492396e-01 -8.86357725e-01 6.67704046e-01 1.50561619e+00 4.49140549e-01 4.16509241e-01 4.35044944e-01 1.21479738e+00 3.70417953e-01 2.78826263e-02 -9.87712801e-01 7.72283152e-02 -1.00751854e-01 8.79035175e-01 -9.35842514e-01 -8.63886923e-02 -7.32655227e-02 -2.64817894e-01 1.31164491e+00 1.20128602e-01 -2.80749947e-01 1.10871577e+00 2.84443945e-01 2.11475402e-01 -2.10062712e-01 -5.91589510e-01 -2.50666052e-01 1.61376446e-01 1.09769320e+00 4.71508861e-01 3.24247144e-02 -2.74141192e-01 1.00593126e+00 -2.43626371e-01 -2.83543855e-01 1.49538681e-01 7.82205343e-01 -7.01414704e-01 -1.13270426e+00 -2.86857456e-01 9.24697995e-01 -7.66451240e-01 6.65393695e-02 -5.67219973e-01 6.61224484e-01 -3.12327426e-02 1.10504472e+00 8.85130987e-02 -2.88967729e-01 5.27480364e-01 6.81864500e-01 3.12122703e-01 -6.00926638e-01 -1.95039675e-01 8.74332637e-02 7.19325900e-01 -6.56549752e-01 -4.09968495e-01 -1.14145815e+00 -9.33637500e-01 2.47428194e-01 -1.23406671e-01 -2.50006706e-01 6.07870042e-01 8.98612976e-01 6.44035041e-01 5.01630068e-01 2.51557320e-01 -4.67983663e-01 -8.95065665e-01 -9.98193502e-01 -4.60808814e-01 -2.70476751e-03 7.11495697e-01 -7.79646635e-01 -6.90618336e-01 -2.02199202e-02]
[13.883306503295898, 5.372596740722656]
8e4b5034-f916-4095-90b9-792827b803c3
look-back-again-dual-parallel-attention
null
null
https://dl.acm.org/doi/10.1145/3460426.3463674
https://dl.acm.org/doi/pdf/10.1145/3460426.3463674
Look Back Again: Dual Parallel Attention Network for Accurate and Robust Scene Text Recognition
Nowadays, it is a trend that using a parallel-decoupled encoderdecoder (PDED) framework in scene text recognition for its flexibility and efficiency. However, due to the inconsistent information content between queries and keys in the parallel positional attention module (PPAM) used in this kind of framework(queries: position information, keys: context and position information), visual misalignment tends to appear when confronting hard samples(e.g., blurred texts, irregular texts, or low-quality images). To tackle this issue, in this paper, we propose a dual parallel attention network (DPAN), in which a newly designed parallel context attention module (PCAM) is cascaded with the original PPAM, using linguistic contextual information to compensate for the information inconsistency between queries and keys. Specifically, in PCAM, we take the visual features from PPAM as inputs and present a bidirectional language model to enhance them with linguistic contexts to produce queries. In this way, we make the information content of the queries and keys consistent in PCAM, which helps to generate more precise visual glimpses to improve the entire PDED framework’s accuracy and robustness. Experimental results verify the effectiveness of the proposed PCAM, showing the necessity of keeping the information consistency between queries and keys in the attention mechanism. On six benchmarks, including regular text and irregular text, the performance of DPAN surpasses the existing leading methods by large margins, achieving new state-of-the-art performance. The code is available on https://github.com/Jackandrome/DPAN.
['Junbo Guo', 'Hongtao Xie', 'Guoqing Jin', 'Zilong Fu']
2021-08-01
null
null
null
icmr-2021-8
['scene-text-recognition']
['computer-vision']
[-3.29063125e-02 -4.63940948e-01 -1.34680672e-02 -9.56006050e-02 -3.94671351e-01 -4.32868928e-01 9.23254013e-01 -2.12981656e-01 -4.09790993e-01 1.62185967e-01 4.03034478e-01 -2.08170921e-01 9.90086943e-02 -4.56350565e-01 -7.49728084e-01 -7.08189249e-01 6.07805669e-01 1.19910436e-02 1.83658987e-01 -8.86145383e-02 3.83043081e-01 2.71638930e-01 -1.54598582e+00 4.82268631e-01 1.02331614e+00 9.45695698e-01 8.25381994e-01 4.02792037e-01 -4.59336609e-01 8.30008566e-01 -5.16710281e-01 -6.05540872e-01 2.39707947e-01 -3.49294513e-01 -4.65900868e-01 1.39147982e-01 4.09064889e-01 -5.17332852e-01 -7.43104398e-01 1.13084495e+00 6.04700685e-01 6.02121800e-02 4.10996079e-01 -1.06251454e+00 -1.32458758e+00 5.48528850e-01 -8.72557104e-01 2.10214451e-01 3.65953803e-01 3.03598315e-01 1.08770120e+00 -1.27538514e+00 3.88775885e-01 1.34597218e+00 4.31242198e-01 2.92271793e-01 -9.62417960e-01 -5.35993993e-01 5.09288073e-01 6.45710588e-01 -1.54952824e+00 -4.50378418e-01 8.73263597e-01 -3.56842130e-01 7.83567905e-01 3.46225530e-01 5.90196908e-01 1.38836288e+00 3.05171102e-01 1.10166144e+00 8.21782589e-01 -4.13024306e-01 -6.59730732e-02 2.55101562e-01 9.73212719e-02 3.83320630e-01 7.93490037e-02 -1.89922839e-01 -5.54755628e-01 3.09606850e-01 6.10227644e-01 2.04835325e-01 -6.56564713e-01 -2.17616275e-01 -1.17665136e+00 4.28955823e-01 4.75095391e-01 5.11407733e-01 -3.25106382e-01 -2.10009396e-01 2.38837749e-01 9.14631784e-02 1.42912686e-01 1.63624510e-01 -9.37853530e-02 -5.32539375e-02 -6.92770243e-01 -2.48584747e-02 3.39578480e-01 1.03733158e+00 4.59854931e-01 4.96599041e-02 -5.01193881e-01 9.02068794e-01 4.26896095e-01 5.80290735e-01 8.57315838e-01 -3.45725745e-01 8.87620032e-01 8.27010393e-01 -1.60327945e-02 -1.30920446e+00 -2.16546252e-01 -4.14849579e-01 -1.05862761e+00 -8.79787132e-02 1.97249755e-01 1.86159417e-01 -7.60728002e-01 1.54428303e+00 1.16511039e-01 5.39618321e-02 1.66718379e-01 1.12640083e+00 8.15948486e-01 8.17020655e-01 -8.15438703e-02 -2.75949035e-02 1.49039519e+00 -1.05585051e+00 -9.76942778e-01 -3.86323184e-01 3.85063380e-01 -1.06841648e+00 1.54183984e+00 2.63595492e-01 -7.88748801e-01 -8.85772049e-01 -1.09551275e+00 -4.49803799e-01 -4.00521517e-01 6.86097980e-01 4.12238650e-02 2.66690224e-01 -8.46780360e-01 2.87261724e-01 -5.29665828e-01 -3.19765210e-01 2.28629857e-01 6.10116124e-02 -3.29789132e-01 -2.46069714e-01 -1.18316448e+00 6.43551767e-01 3.92119348e-01 5.05707920e-01 -4.52170312e-01 -4.20938760e-01 -8.62903893e-01 1.63350180e-01 4.24035281e-01 -4.74379450e-01 1.01632440e+00 -1.13950205e+00 -1.44658208e+00 5.08998871e-01 -1.45508513e-01 -1.96596548e-01 6.99664891e-01 -5.00609040e-01 -4.17888552e-01 9.36835930e-02 1.05919823e-01 5.30627429e-01 1.04626918e+00 -1.17821312e+00 -4.36365485e-01 -3.06034595e-01 6.60823882e-02 5.98813593e-01 -4.50431138e-01 -2.48197429e-02 -1.05355883e+00 -8.94659042e-01 -1.88215002e-02 -7.08596230e-01 3.61667305e-01 4.13031578e-02 -5.29335439e-01 -1.80116653e-01 1.24030888e+00 -9.77868557e-01 1.35855711e+00 -2.48362684e+00 3.20560902e-01 -2.37983584e-01 2.30871383e-02 4.58312422e-01 -1.65724084e-01 3.81273657e-01 -7.85558596e-02 2.73603313e-02 -1.88748032e-01 -6.50742590e-01 1.25111997e-01 2.15641573e-01 -3.53120834e-01 2.54001617e-01 4.14654166e-01 1.04129767e+00 -6.26302421e-01 -4.69564766e-01 4.52254742e-01 6.96653664e-01 -2.79450357e-01 2.90023983e-01 -1.31952375e-01 3.76544982e-01 -4.07321393e-01 6.41087532e-01 8.57684433e-01 -3.08719248e-01 -1.66158266e-02 -5.00943303e-01 -2.06265077e-01 1.78818807e-01 -1.26462877e+00 1.60776973e+00 -4.84196216e-01 6.56424165e-01 7.57490983e-03 -7.59645879e-01 8.96156311e-01 2.63991117e-01 8.83275568e-02 -1.10963798e+00 2.63118386e-01 2.37137172e-02 -9.90093201e-02 -7.29863584e-01 5.72466075e-01 4.34339494e-01 2.05376357e-01 1.59036189e-01 -1.72394753e-01 1.74625307e-01 1.56262904e-01 1.89724267e-01 6.43813908e-01 1.69356599e-01 7.24870190e-02 -5.52107655e-02 9.76379395e-01 -5.02457201e-01 4.82335687e-01 6.30156517e-01 -1.33920982e-01 8.88639033e-01 5.49648106e-01 -3.30926895e-01 -1.08106315e+00 -6.93072677e-01 5.85762598e-03 9.05550659e-01 3.59086573e-01 -5.69529533e-01 -5.43422878e-01 -6.62248611e-01 -2.24728957e-01 6.42782271e-01 -5.66854239e-01 -7.61338621e-02 -5.61010182e-01 -6.33836687e-01 4.08667743e-01 5.87155461e-01 9.85886037e-01 -9.99161541e-01 -3.46964598e-01 5.12285307e-02 -2.52346516e-01 -1.29853082e+00 -8.27246964e-01 -8.16225931e-02 -3.87771666e-01 -8.94032657e-01 -9.04819727e-01 -8.01234126e-01 5.64231455e-01 4.68789190e-01 7.67700016e-01 -8.07974895e-04 1.29999563e-01 1.91800401e-01 -5.38036346e-01 -1.30491843e-02 -1.95442721e-01 -1.19486116e-01 -2.56576449e-01 5.14851868e-01 1.87617183e-01 -3.79098564e-01 -6.17957771e-01 3.31315100e-01 -1.21693802e+00 5.05385816e-01 8.46814990e-01 1.01674712e+00 5.29386997e-01 -3.20066549e-02 1.53918579e-01 -4.99245286e-01 5.93413532e-01 -4.51972336e-01 -6.62635148e-01 3.15335780e-01 -4.36389357e-01 -1.05486251e-02 9.70323861e-01 -4.91208494e-01 -1.14976490e+00 2.85988878e-02 -1.69374228e-01 -7.14980006e-01 -3.62235568e-02 4.64368790e-01 -6.07167125e-01 1.89374954e-01 1.45184591e-01 6.69665039e-01 -1.52750745e-01 -7.27096856e-01 2.93050885e-01 8.29763353e-01 4.86320466e-01 -3.10141802e-01 6.78421736e-01 2.65786678e-01 -3.45830172e-01 -7.19651520e-01 -5.10384738e-01 -1.62267476e-01 -3.91725302e-01 -2.55109984e-02 8.33668113e-01 -8.86312187e-01 -4.34417546e-01 7.91172087e-01 -1.34973562e+00 -1.40946567e-01 1.16981931e-01 4.70740825e-01 -1.76157907e-01 7.48426795e-01 -6.43819273e-01 -5.39539635e-01 -2.69601017e-01 -1.52790642e+00 1.24822426e+00 3.51592839e-01 2.02454925e-01 -7.25215256e-01 -3.20959687e-01 3.38294357e-01 3.31366897e-01 -2.69594580e-01 8.36157620e-01 -5.60191453e-01 -7.36678183e-01 1.19853150e-02 -6.00992620e-01 3.82975101e-01 1.93443298e-01 1.25553399e-01 -1.01769817e+00 -1.86779037e-01 9.87440944e-02 -5.16140386e-02 7.50367045e-01 7.44072869e-02 1.20547867e+00 -5.08043110e-01 -8.94316807e-02 6.28391027e-01 1.27823257e+00 2.79748827e-01 7.32259691e-01 4.20818806e-01 9.63238835e-01 4.32265520e-01 4.13361073e-01 3.08151454e-01 4.64373887e-01 8.49132359e-01 4.13183391e-01 1.47890672e-03 -3.67172182e-01 -2.91769058e-01 4.43098426e-01 8.70745659e-01 2.29965344e-01 -5.43796778e-01 -7.85655260e-01 3.53407264e-01 -1.84905577e+00 -7.35400796e-01 -5.91408201e-02 2.13016081e+00 7.98959017e-01 7.13140070e-02 -3.49153966e-01 2.37374738e-01 8.97296786e-01 4.13512528e-01 -4.36890483e-01 -2.21137092e-01 -5.21620512e-01 -3.13226074e-01 1.19953007e-01 4.32009667e-01 -1.04207110e+00 8.21023166e-01 4.47876501e+00 1.09250808e+00 -1.34129584e+00 -2.02685092e-02 5.28528333e-01 6.24786913e-02 -3.50348890e-01 -1.78996772e-01 -7.45371401e-01 8.80103946e-01 3.57493371e-01 4.34656739e-02 5.26932359e-01 6.31029546e-01 3.38660687e-01 1.11115547e-02 -9.07921076e-01 1.19122326e+00 2.81502962e-01 -1.05999291e+00 2.67311811e-01 -1.59910455e-01 4.87849206e-01 -1.46291628e-01 2.69681245e-01 2.77861267e-01 -2.51444936e-01 -7.28153944e-01 9.31720078e-01 4.89171475e-01 6.45363450e-01 -5.48915207e-01 7.71057487e-01 2.74759501e-01 -1.34612048e+00 -1.32563382e-01 -3.31090659e-01 1.15616828e-01 -9.54591110e-02 4.93648827e-01 -4.39428806e-01 8.57040524e-01 7.87215769e-01 9.74804997e-01 -8.94530714e-01 8.81796658e-01 -3.81616890e-01 2.21909836e-01 -2.44314462e-01 8.45883414e-02 3.55553031e-01 -2.38139153e-01 6.36015534e-01 1.19146657e+00 2.46884763e-01 -2.90735722e-01 -7.82202557e-02 9.64302242e-01 -1.78884730e-01 3.48465830e-01 -5.44191599e-01 9.66058895e-02 4.75280672e-01 1.21730828e+00 -4.56345677e-01 -3.57251555e-01 -7.43899465e-01 1.25360715e+00 2.69215196e-01 5.42474568e-01 -9.31084454e-01 -3.96831989e-01 3.77856076e-01 -1.88430712e-01 5.29841244e-01 -5.90337440e-02 -1.96804121e-01 -1.37586927e+00 6.50226653e-01 -1.08770525e+00 2.71777540e-01 -1.13473749e+00 -1.17131305e+00 6.74019992e-01 -3.24707687e-01 -1.38390911e+00 1.76164821e-01 -5.85353374e-01 -5.96526802e-01 9.59594727e-01 -1.60825276e+00 -1.31093323e+00 -4.55114752e-01 6.73340440e-01 9.19942796e-01 3.86228077e-02 2.81958431e-01 4.74249363e-01 -9.46462870e-01 8.04116488e-01 2.66141772e-01 3.35389376e-01 8.68113637e-01 -1.05951273e+00 2.93105245e-01 1.11450624e+00 2.00697869e-01 5.45981348e-01 4.07892823e-01 -5.57819426e-01 -1.66827166e+00 -1.10611868e+00 7.92046070e-01 -1.81293294e-01 6.37334228e-01 -4.95171279e-01 -1.28441608e+00 5.55678725e-01 4.40821707e-01 -4.12316658e-02 2.02358559e-01 -4.79617178e-01 -5.49511313e-01 -2.43052959e-01 -6.50043666e-01 9.22022045e-01 7.69959986e-01 -6.35945737e-01 -7.50420928e-01 1.65657446e-01 8.41846168e-01 -5.29920220e-01 -4.46920812e-01 2.96812922e-01 5.37108183e-01 -1.07886434e+00 7.08265185e-01 3.58028412e-02 5.49229741e-01 -6.37370884e-01 -1.83049023e-01 -1.01223660e+00 -3.22700113e-01 -4.41235125e-01 -9.01897550e-02 1.47566688e+00 1.13966361e-01 -6.93273485e-01 3.14518988e-01 3.17517906e-01 -3.17792028e-01 -7.93355644e-01 -1.06084478e+00 -6.61691129e-01 -1.20904773e-01 -2.79349983e-01 6.02994680e-01 8.45750809e-01 -2.41899982e-01 5.15590250e-01 -5.21249712e-01 3.38912487e-01 1.44467965e-01 2.37413570e-01 5.37865043e-01 -6.43452942e-01 -3.72975618e-01 -6.77003622e-01 -2.81262547e-01 -1.47648382e+00 6.76255822e-02 -6.24608874e-01 -1.57538906e-01 -1.41083217e+00 1.40133858e-01 -1.73703626e-01 -1.81593359e-01 4.32822019e-01 -4.92157698e-01 5.62480055e-02 5.18193722e-01 3.67928565e-01 -6.17964983e-01 9.20097649e-01 1.27167964e+00 -3.88401479e-01 -1.55048549e-01 -3.63331586e-01 -6.30018115e-01 5.71573853e-01 6.71387553e-01 -7.22419992e-02 -3.93352181e-01 -8.55278671e-01 1.12800747e-01 -1.76133603e-01 5.46364069e-01 -9.79207277e-01 3.92099738e-01 1.02028012e-01 6.28798723e-01 -7.66784489e-01 2.65622616e-01 -1.02379358e+00 3.51089612e-02 2.85283327e-01 -2.15122446e-01 3.25187892e-01 3.17812681e-01 5.15075803e-01 -4.35684472e-01 -1.44654885e-01 5.74918270e-01 1.91940948e-01 -7.84117103e-01 1.43715352e-01 -1.96179569e-01 -2.30238326e-02 7.40548790e-01 -1.62965462e-01 -5.29612958e-01 -2.68271565e-01 -2.04178795e-01 3.89055341e-01 5.68096340e-01 7.51349092e-01 7.09690392e-01 -1.33904409e+00 -6.22774124e-01 4.52555716e-01 2.31182724e-01 6.33123219e-02 5.63710630e-01 1.00819397e+00 -3.64933670e-01 4.33445394e-01 4.12722714e-02 -6.74887002e-01 -1.06080580e+00 8.49434137e-01 3.91946465e-01 -1.97958142e-01 -7.11240590e-01 5.37546992e-01 5.52334905e-01 -1.09414816e-01 3.96359950e-01 -4.58532155e-01 -3.85749906e-01 4.33238223e-02 6.63218737e-01 3.87398759e-03 1.24929927e-01 -8.13995242e-01 -2.71373153e-01 9.43980038e-01 -2.35332102e-01 9.68255028e-02 9.35367882e-01 -4.92442966e-01 9.56970975e-02 3.83611649e-01 1.13542223e+00 2.81485498e-01 -1.48463309e+00 -4.14839894e-01 -2.51568288e-01 -6.37101233e-01 3.51230102e-03 -6.35989010e-01 -1.23267639e+00 9.98520970e-01 5.88497341e-01 1.41902238e-01 1.35665917e+00 -2.08242044e-01 6.18313253e-01 2.34448031e-01 -1.69932276e-01 -8.86304736e-01 1.92946106e-01 4.94394064e-01 1.21759498e+00 -1.23887002e+00 -1.53640062e-01 -1.54227123e-01 -8.84839833e-01 1.05681169e+00 7.46727288e-01 1.63839415e-01 2.94317961e-01 7.29276538e-02 1.15805358e-01 1.64290801e-01 -6.55491590e-01 -9.74687561e-02 4.44397151e-01 3.50840539e-01 3.04622114e-01 -3.49308848e-01 -2.20796555e-01 7.60802925e-01 8.98664296e-02 -3.34820271e-01 3.16921294e-01 7.97045171e-01 -1.25508964e-01 -1.04386103e+00 -4.99925256e-01 6.80461451e-02 -4.02350992e-01 -4.37903315e-01 -3.75060052e-01 7.86330462e-01 1.98785543e-01 8.46119463e-01 9.02424976e-02 -4.06601667e-01 4.35734361e-01 6.96261302e-02 3.61581072e-02 -2.13417187e-01 -4.47496772e-01 3.21539402e-01 -2.94139355e-01 -4.66845900e-01 -2.42977619e-01 -4.23172891e-01 -1.02748179e+00 -8.14047232e-02 -3.65430623e-01 -1.38731942e-01 4.69548047e-01 7.84016311e-01 6.21761262e-01 5.89648306e-01 7.09945977e-01 -6.31787837e-01 -4.16666865e-01 -9.98462498e-01 -2.75110990e-01 5.34914255e-01 3.17090094e-01 -6.40949786e-01 -3.42087209e-01 9.55886990e-02]
[11.792877197265625, 2.0484097003936768]
9746fc63-aa05-4479-82fb-e5330c320441
collaborative-intelligence-challenges-and
2102.06841
null
https://arxiv.org/abs/2102.06841v1
https://arxiv.org/pdf/2102.06841v1.pdf
Collaborative Intelligence: Challenges and Opportunities
This paper presents an overview of the emerging area of collaborative intelligence (CI). Our goal is to raise awareness in the signal processing community of the challenges and opportunities in this area of growing importance, where key developments are expected to come from signal processing and related disciplines. The paper surveys the current state of the art in CI, with special emphasis on signal processing-related challenges in feature compression, error resilience, privacy, and system-level design.
['Yonghong Tian', 'Weisi Lin', 'Ivan V. Bajić']
2021-02-13
null
null
null
null
['feature-compression']
['computer-vision']
[ 7.38484800e-01 -1.41478553e-01 3.21519822e-01 -3.90011758e-01 -4.81843203e-01 -2.53812969e-01 1.76344797e-01 1.18791468e-01 -3.74080479e-01 5.11663318e-01 4.07636613e-01 8.78016651e-02 -5.39851665e-01 -3.83846164e-01 1.10665634e-01 -5.75389624e-01 -9.52014625e-01 -3.81512403e-01 -2.49685839e-01 -1.48203388e-01 2.84418494e-01 5.43520987e-01 -1.30593741e+00 2.26813376e-01 6.46934867e-01 1.55574191e+00 -1.83975264e-01 1.03555989e+00 5.44270575e-01 9.20399070e-01 -1.13079655e+00 -5.53695261e-01 4.18029040e-01 -5.96280634e-01 -2.98539370e-01 -1.08745672e-01 6.42837677e-03 3.06363753e-03 -4.14637715e-01 1.20161450e+00 9.26227272e-01 5.65909073e-02 4.37621325e-01 -1.27224100e+00 -1.40762076e-01 7.13905752e-01 -4.46773082e-01 8.49135876e-01 1.54467925e-01 -1.78822219e-01 7.89268911e-01 -9.48193371e-01 2.79756308e-01 7.32417941e-01 9.72337842e-01 -2.30869308e-01 -9.73034680e-01 -8.16771984e-01 7.35689700e-02 5.91501176e-01 -1.88189459e+00 -8.46372306e-01 1.03470039e+00 -6.62587956e-02 9.95227933e-01 4.91316199e-01 5.64692974e-01 5.73604465e-01 1.35584503e-01 9.53707457e-01 7.65154302e-01 -8.25651586e-01 5.55476606e-01 -1.42303467e-01 2.12174430e-01 3.50196064e-02 4.10700381e-01 4.39071685e-01 -1.09388494e+00 -4.73905206e-01 6.20947361e-01 -2.52464533e-01 -5.28109372e-01 -5.01224399e-01 -1.12627482e+00 7.93630660e-01 2.54271775e-01 7.64535189e-01 -5.82654119e-01 2.64118195e-01 5.63903391e-01 8.63363266e-01 6.55090809e-01 3.87434334e-01 -2.81652302e-01 -2.42291853e-01 -1.30051351e+00 9.85563472e-02 1.04948556e+00 8.98792744e-01 3.42108384e-02 6.16668642e-01 -1.56190932e-01 8.36122513e-01 3.76640618e-01 6.01022601e-01 1.30539566e-01 -7.95151412e-01 4.17953581e-01 -2.50168949e-01 -1.54092520e-01 -1.62047172e+00 -3.59663486e-01 -9.71447706e-01 -1.11478579e+00 7.37384036e-02 -2.40998700e-01 -7.19384909e-01 -9.00919959e-02 1.27241504e+00 -1.70919269e-01 4.23542708e-01 3.41551632e-01 7.09692001e-01 6.55725539e-01 4.95841980e-01 -3.22595567e-01 -7.24576592e-01 1.11412132e+00 -4.48296607e-01 -1.18342507e+00 -4.69500124e-01 -2.67477781e-01 -8.92804146e-01 -3.28410178e-01 6.03480935e-01 -1.34719265e+00 -4.39202785e-01 -1.37420619e+00 6.62975967e-01 4.83762994e-02 4.49086092e-02 9.54046488e-01 1.26110399e+00 -1.03318167e+00 3.11773688e-01 -5.66707134e-01 -3.02767336e-01 6.85934126e-01 4.10042495e-01 1.56750649e-01 -1.25817597e-01 -1.30101144e+00 6.07668519e-01 1.54697243e-02 9.56724286e-02 -2.68040866e-01 -1.06518757e+00 -5.38991988e-01 1.05370298e-01 2.36205548e-01 -5.31432629e-01 1.22222674e+00 -7.17696488e-01 -1.55903244e+00 3.54494363e-01 2.51269396e-02 -1.02708602e+00 3.32231492e-01 -1.57703698e-01 -1.20738053e+00 2.32565194e-01 -5.06051898e-01 -2.76805777e-02 1.18776059e+00 -6.46574557e-01 -8.90964210e-01 -4.05221552e-01 -7.50005424e-01 8.87817740e-02 -2.53837913e-01 5.87081492e-01 -2.35449359e-01 -1.22338820e+00 9.73130539e-02 -4.78138864e-01 -5.16860723e-01 5.37429340e-02 -1.13099337e-01 4.68438536e-01 6.82411849e-01 -3.09328973e-01 1.53783774e+00 -2.41745090e+00 -1.03643470e-01 6.65410995e-01 2.16847852e-01 3.34398180e-01 3.36758852e-01 8.56247127e-01 1.43051222e-01 -2.52606034e-01 -1.01116389e-01 -2.03875527e-01 -2.86880702e-01 -2.70456493e-01 -4.52225327e-01 6.09371960e-01 -7.24763423e-02 6.21375680e-01 -6.68792129e-01 -3.14537645e-03 2.64405757e-01 3.65708977e-01 -4.21741515e-01 7.25743268e-03 5.15283465e-01 3.91917139e-01 -2.96586335e-01 6.55430615e-01 7.06269383e-01 7.31867924e-02 4.09446299e-01 -4.35284287e-01 -3.09601903e-01 -6.92320168e-02 -1.48893857e+00 1.23923600e+00 -2.12419733e-01 1.07010210e+00 6.01315618e-01 -1.41938245e+00 9.05100882e-01 6.68952405e-01 7.15214491e-01 -3.72946769e-01 3.59064847e-01 7.92875588e-02 3.69815439e-01 1.40177995e-01 2.96333551e-01 1.58047736e-01 -9.78503898e-02 5.48909009e-01 1.54507697e-01 -1.96765393e-01 -1.98097810e-01 1.22100823e-01 1.35193396e+00 -8.71267319e-01 8.94606173e-01 -3.96451175e-01 5.53966761e-01 -3.67471725e-01 4.94635791e-01 9.86493647e-01 -7.23247766e-01 2.93221533e-01 -1.90192923e-01 -1.88860595e-01 -4.98617500e-01 -1.04387188e+00 -3.90822351e-01 7.72903621e-01 2.79895943e-02 -6.64663434e-01 -4.44874346e-01 -3.77959050e-02 -7.89136291e-02 6.31329790e-02 -3.15887034e-01 -1.51178822e-01 -3.49151701e-01 -9.98711407e-01 5.50102174e-01 5.10287583e-01 8.05878699e-01 -8.04446697e-01 -9.02003288e-01 5.74518323e-01 -1.00582168e-01 -1.13029861e+00 -2.70287424e-01 1.84283793e-01 -1.06138289e+00 -7.31581628e-01 -5.97563803e-01 -6.34897232e-01 1.95590109e-01 4.23463494e-01 1.05409682e+00 -3.76659840e-01 -6.89772248e-01 7.89709210e-01 -5.60632288e-01 -9.29326713e-01 1.12042306e-02 -4.47104156e-01 2.61279315e-01 3.71674120e-01 3.69596303e-01 -7.92688668e-01 -7.05771148e-01 4.33610529e-01 -5.98712265e-01 -6.88752770e-01 4.68440890e-01 6.33655667e-01 3.77532959e-01 4.48402703e-01 9.96005893e-01 -7.26944625e-01 9.83884037e-01 -3.01442832e-01 -4.20965582e-01 1.55539498e-01 -5.77702701e-01 -4.72363323e-01 4.54338074e-01 -3.39922875e-01 -9.43292618e-01 -3.69489901e-02 -8.86635929e-02 -1.43605694e-02 2.81357706e-01 7.51516819e-01 -5.31351790e-02 -7.58486927e-01 8.11694324e-01 1.26570359e-01 -2.24493500e-02 -2.38890111e-01 4.35042739e-01 1.04550409e+00 7.99637437e-01 -1.28256008e-01 5.77675045e-01 5.14422596e-01 -6.93188533e-02 -1.20376348e+00 -3.55740547e-01 -5.75165451e-01 -2.10630611e-01 -2.81535715e-01 2.13497013e-01 -9.03553247e-01 -4.03432786e-01 4.82127011e-01 -8.36716056e-01 4.06972528e-01 -5.24102271e-01 7.19861686e-01 -6.25477076e-01 3.82875085e-01 -5.13204813e-01 -9.64113832e-01 -1.00251305e+00 -7.95359373e-01 2.04985812e-01 7.98115060e-02 -3.87529731e-01 -7.31903732e-01 3.45057584e-02 2.61638239e-02 1.05303335e+00 1.37714949e-02 2.94623733e-01 -7.79848218e-01 -3.36999416e-01 -4.67594266e-01 2.17809513e-01 4.50799197e-01 1.94814596e-02 -3.17359090e-01 -1.20483387e+00 -3.16418231e-01 2.21139684e-01 1.37338594e-01 5.11096358e-01 8.10905874e-01 7.40954280e-01 -1.27229840e-01 -5.51698148e-01 5.00873864e-01 1.19197011e+00 5.38739800e-01 5.95745742e-01 -2.75820673e-01 -3.70982587e-01 2.13100478e-01 5.44102788e-01 1.01119447e+00 -3.94564271e-01 6.60936713e-01 -1.59609988e-01 4.24942300e-02 -7.24427551e-02 1.25502199e-01 1.10069066e-01 1.02029991e+00 3.75894643e-02 -2.40304694e-01 -6.23936236e-01 3.33525300e-01 -1.82057881e+00 -1.42013931e+00 2.05002084e-01 2.27914119e+00 4.35478538e-01 -1.37822092e-01 6.83717430e-03 9.73047137e-01 7.17195809e-01 1.67896792e-01 -4.92877752e-01 -2.95719624e-01 -7.63636306e-02 3.39004815e-01 5.86837828e-01 1.41621113e-01 -1.36045504e+00 3.73297632e-01 8.47760868e+00 8.30561399e-01 -9.58847463e-01 -8.67839083e-02 3.74989092e-01 3.03581320e-02 2.76568711e-01 -3.81309837e-01 -5.31622112e-01 9.73644853e-02 1.13503528e+00 -6.49648190e-01 4.42257881e-01 7.67399609e-01 1.17742918e-01 3.24862674e-02 -8.01424444e-01 1.34869826e+00 5.36212683e-01 -1.42626715e+00 -7.06302106e-01 -1.40111953e-01 6.70559704e-01 3.29193711e-01 3.62353683e-01 -2.76064500e-02 -1.12725213e-01 -5.24526060e-01 6.66934848e-01 6.32679522e-01 7.08531141e-01 -1.00720882e+00 7.83290148e-01 3.83055992e-02 -1.37916958e+00 -4.12713975e-01 -3.37686360e-01 -4.08937484e-01 4.40411240e-01 1.17275071e+00 -6.31408453e-01 7.98774362e-01 9.53149557e-01 7.45783567e-01 -8.23223218e-02 1.63981187e+00 2.48412751e-02 6.27786994e-01 -5.60354352e-01 -1.79382116e-01 -5.35763681e-01 -1.31829843e-01 8.41373920e-01 1.40821922e+00 2.82100797e-01 4.16091681e-01 2.79267579e-01 8.17429647e-02 1.63176149e-01 5.13076112e-02 -7.48455524e-01 -7.97363818e-02 1.05548310e+00 9.08329666e-01 -5.12479603e-01 -2.23296151e-01 -4.53342021e-01 8.70032370e-01 -3.47860187e-01 3.78154404e-02 -4.05447453e-01 -7.30012655e-01 8.49365413e-01 -6.30924344e-01 4.75058585e-01 -4.71850812e-01 -6.66119099e-01 -9.02273893e-01 -1.91496924e-01 -9.97236311e-01 6.24520123e-01 -2.35761091e-01 -1.28000057e+00 6.94378078e-01 -1.17624095e-02 -1.63579488e+00 -4.42336023e-01 -6.28811941e-02 -4.56910938e-01 7.58640766e-01 -1.05563414e+00 -6.08058691e-01 8.73871222e-02 5.74004471e-01 2.79282570e-01 -4.39901948e-01 9.76821125e-01 3.85321975e-01 -6.03425689e-02 1.04819810e+00 3.90086830e-01 -1.34237021e-01 5.33588350e-01 -5.18673420e-01 7.18077600e-01 1.15745354e+00 5.83642244e-01 6.85743988e-01 8.20913792e-01 -2.56501913e-01 -1.70823133e+00 -8.26354742e-01 9.00708735e-01 -1.82639450e-01 7.01956451e-01 -3.56298834e-01 -2.79518515e-01 2.53755748e-01 2.50699133e-01 3.54560852e-01 1.24205232e+00 2.46331766e-01 -1.08064875e-01 -2.80379355e-01 -1.45566094e+00 4.48914945e-01 9.91569877e-01 -4.52290714e-01 -3.99120957e-01 1.15121379e-01 2.11997479e-01 -9.57627445e-02 -8.85874212e-01 3.22759807e-01 6.32407248e-01 -8.49445820e-01 1.28791678e+00 -8.55994150e-02 -5.99166214e-01 -3.27041000e-01 -4.44040447e-01 -1.33439553e+00 -4.11882013e-01 -1.14572465e+00 -2.21973836e-01 1.12032795e+00 1.99112505e-01 -6.63509071e-01 7.81697869e-01 2.11237177e-01 -5.19677587e-02 -3.86211514e-01 -1.21843457e+00 -8.49679708e-01 -4.92144912e-01 -1.00700474e+00 2.49686778e-01 7.15846300e-01 7.41018534e-01 4.06740874e-01 -6.81374669e-01 1.99609563e-01 8.08545768e-01 8.33650231e-02 4.38031614e-01 -1.34487152e+00 -3.81056815e-01 -2.36884773e-01 -1.01171958e+00 -1.06101918e+00 -5.67730069e-01 -7.05870509e-01 -1.35110617e-01 -1.02204454e+00 -4.48048323e-01 -1.52797595e-01 -3.42568606e-01 -2.31494948e-01 2.45173857e-01 6.00947082e-01 6.02284133e-01 8.70810002e-02 -6.49122655e-01 2.29049772e-01 4.69045162e-01 -1.92643598e-01 -3.30857992e-01 6.31499052e-01 -9.81624961e-01 5.54092348e-01 1.02407885e+00 -1.23024009e-01 -4.57090020e-01 -5.14399186e-02 6.69781715e-02 2.26376131e-01 -1.44191578e-01 -1.52788949e+00 9.18982804e-01 2.85809487e-01 4.36359525e-01 -5.97809017e-01 6.06116176e-01 -1.22784948e+00 4.21851426e-01 5.01808167e-01 -2.98625201e-01 8.00434351e-02 -2.03883890e-02 7.72159696e-01 -4.97937500e-01 -2.71173250e-02 7.57121086e-01 1.48751259e-01 -7.36299276e-01 4.60581817e-02 -8.75962913e-01 -9.52270534e-03 1.03909552e+00 -5.33462316e-02 1.03162706e-01 -7.03274429e-01 -7.84495592e-01 -1.75062343e-02 -3.41178000e-01 1.31736398e-01 8.36755633e-01 -9.77131128e-01 -8.34573030e-01 5.44853687e-01 1.89056486e-01 -8.00709307e-01 4.06404108e-01 9.41434264e-01 -6.41593337e-02 6.50966704e-01 -7.41325468e-02 -3.81193697e-01 -1.32058907e+00 4.08216298e-01 7.66020641e-02 -1.27938077e-01 -8.23960006e-01 1.02234137e+00 -4.50097293e-01 3.63595784e-01 6.03251338e-01 -2.70740483e-02 -2.69472182e-01 -1.35514200e-01 1.30229974e+00 6.47340298e-01 4.36342359e-01 -2.89075613e-01 -5.68137348e-01 3.44177932e-01 -1.32447314e-02 -2.51939595e-01 9.39843476e-01 -3.24270487e-01 1.37389973e-01 2.72219032e-01 1.09909475e+00 6.63716421e-02 -7.51700759e-01 -4.36400652e-01 1.18049368e-01 -6.34437501e-01 5.68601012e-01 -1.05733764e+00 -1.13503623e+00 6.87785089e-01 1.15446925e+00 9.62247923e-02 1.60305882e+00 -2.77574599e-01 5.37005246e-01 3.81316751e-01 1.11267614e+00 -1.17219841e+00 -3.68013054e-01 5.32318890e-01 1.01462746e+00 -6.76594675e-01 4.16236877e-01 -5.77109337e-01 -6.55053079e-01 9.91144359e-01 -5.01466751e-01 -1.85730442e-01 1.47854984e+00 8.14166903e-01 -2.13099383e-02 -5.74420057e-02 -7.59126127e-01 -6.96389377e-03 1.09367549e-01 1.34027886e+00 5.01647830e-01 1.00182192e-02 -4.77583915e-01 7.26849139e-01 -1.54401019e-01 3.34560364e-01 3.42008770e-01 1.22653294e+00 -5.73167562e-01 -1.24653280e+00 -6.64698899e-01 7.79558718e-01 -7.25456655e-01 -2.15790257e-01 -2.90530145e-01 2.14099083e-02 -3.31312776e-01 1.67132938e+00 -1.45035058e-01 -6.29748285e-01 2.70183742e-01 -4.05269593e-01 3.35114241e-01 -2.64766634e-01 -8.35040390e-01 1.26897246e-01 3.60592842e-01 -4.49854046e-01 -4.15095657e-01 -1.07791984e+00 -6.17397845e-01 -1.18881099e-01 -3.14229935e-01 2.66949594e-01 9.01713431e-01 5.13125420e-01 7.67952442e-01 5.82153916e-01 8.62569869e-01 -6.52890027e-01 -6.38801038e-01 -8.00454199e-01 -8.84235084e-01 -4.54907417e-01 1.93507850e-01 -3.65123421e-01 -5.05815923e-01 -1.47030866e-02]
[15.40478229522705, 5.575064659118652]
9b5c944e-e45d-4010-b616-038f15856f25
unsupervised-full-constituency-parsing-with-1
null
null
https://openreview.net/forum?id=R73K-lxO9eU
https://openreview.net/pdf?id=R73K-lxO9eU
Unsupervised Full Constituency Parsing with Neighboring Distribution Divergence
Unsupervised constituency parsing has been explored much but is still far from being solved as currently mainstream unsupervised constituency parser only captures the unlabeled structure of sentences. Properties in the substitution of constituents make it possible to detect constituents in a particular label. We propose an unsupervised and training-free labeling procedure by leveraging a newly introduced metric, Neighboring Distribution Divergence (NDD), which evaluates semantic changes caused by editions. We develop NDD into Dual POS-NDD (DP-NDD) and build templates called "molds" to extract labeled constituents from sentences. We show that DP-NDD labels constituents precisely and inducts more accurate unlabeled constituency trees than all previous unsupervised methods. Following two frameworks for labeled constituency trees inference, we set the new state-of-the-art for unlabeled F1 and labeled F1. Further studies show our approach can be scaled to other span labeling problems, i.e., named entity recognition.
['Anonymous']
2022-01-16
null
null
null
acl-arr-january-2022-1
['constituency-parsing']
['natural-language-processing']
[ 3.56493711e-01 6.47794604e-01 -5.37902594e-01 -1.01722944e+00 -1.02577090e+00 -1.23151672e+00 4.50592667e-01 3.44251841e-01 -2.25637525e-01 8.99684787e-01 5.42930663e-01 -3.85330528e-01 3.99146795e-01 -8.53534281e-01 -6.98923290e-01 -4.78527606e-01 1.65750772e-01 5.32659590e-01 3.62780780e-01 -8.60827118e-02 1.18754461e-01 4.59626943e-01 -1.43195558e+00 3.49330932e-01 5.53984404e-01 5.53339183e-01 9.92340893e-02 5.04831254e-01 -6.56962037e-01 8.35572481e-01 -5.00482559e-01 -4.69694227e-01 1.91052988e-01 -4.84921306e-01 -1.23919690e+00 5.33893034e-02 6.14408910e-01 7.95264840e-02 2.53123581e-01 1.18656933e+00 1.58161387e-01 -9.75382477e-02 6.63822174e-01 -7.94344068e-01 -6.03224277e-01 1.26550770e+00 -3.17658156e-01 4.42552954e-01 1.91722408e-01 -5.38590312e-01 1.52597523e+00 -8.52179646e-01 1.06123090e+00 1.44183898e+00 7.04504907e-01 7.24915683e-01 -1.30197394e+00 -4.39876050e-01 5.61770022e-01 -4.23501164e-01 -5.82748592e-01 -4.45164472e-01 7.10834861e-01 -5.16866982e-01 9.58695114e-01 2.56808490e-01 6.77841529e-02 9.02653217e-01 -4.95217368e-02 9.69873905e-01 1.49012303e+00 -9.61927712e-01 3.74659985e-01 -4.99959802e-03 1.12804198e+00 6.63783550e-01 3.16680193e-01 2.37418115e-01 -2.66277015e-01 1.12124011e-01 3.07578236e-01 -5.56066513e-01 3.20235312e-01 -1.69196039e-01 -1.08152175e+00 1.06666911e+00 -1.01733394e-01 6.93387449e-01 -6.21168502e-02 -1.12061940e-01 4.02902544e-01 2.40155578e-01 7.47128844e-01 4.22646433e-01 -9.17208552e-01 -3.93603090e-03 -8.10951054e-01 1.60715908e-01 9.03499305e-01 1.08377254e+00 1.00172698e+00 -3.67885321e-01 -4.06922460e-01 6.94641352e-01 1.65851533e-01 2.65840739e-01 2.56575912e-01 -9.67153430e-01 4.68356401e-01 8.09709132e-01 -3.42092812e-01 -3.88639301e-01 -5.57060361e-01 -1.28032640e-01 -1.36487082e-01 -7.53528848e-02 4.97711748e-01 -3.53668958e-01 -1.08428943e+00 2.20790148e+00 6.34561241e-01 -2.21182425e-02 2.18926981e-01 3.80654693e-01 9.61493194e-01 5.45434892e-01 4.57271993e-01 -4.61483061e-01 1.50004959e+00 -8.84554684e-01 -7.82258928e-01 -3.90555412e-01 1.01480329e+00 -6.11725688e-01 7.88159490e-01 2.33977716e-02 -9.32207346e-01 -4.05514508e-01 -8.25309336e-01 -1.83744401e-01 -5.65427423e-01 -5.83918989e-02 1.04494441e+00 1.21925139e+00 -9.59818780e-01 5.31743586e-01 -1.01919019e+00 -2.62024909e-01 3.42165768e-01 5.62980413e-01 -5.16873360e-01 1.08543254e-01 -9.18362498e-01 9.11140263e-01 5.74891806e-01 -2.21373975e-01 -6.02734029e-01 -5.28475165e-01 -1.11798203e+00 -2.66755342e-01 3.15084398e-01 -2.95318812e-01 1.49038565e+00 -7.23252594e-01 -1.47161233e+00 1.47484517e+00 -3.41261148e-01 -1.69284716e-01 -1.05303235e-01 -5.72159141e-02 -3.79091978e-01 -2.69493222e-01 6.07288837e-01 7.03523219e-01 2.93149829e-01 -1.33288252e+00 -9.75274980e-01 -5.38522422e-01 1.69840544e-01 1.51578663e-02 -1.65671989e-01 4.32734072e-01 7.39863962e-02 -5.85370719e-01 5.55653393e-01 -8.92904818e-01 -2.47432694e-01 -8.20237398e-01 -7.93888330e-01 -6.63214087e-01 6.14015102e-01 -3.58640969e-01 1.04494989e+00 -1.92752397e+00 -1.26930296e-01 -3.34761553e-02 2.58339047e-01 2.57246107e-01 -2.59249628e-01 2.37487987e-01 -2.89838433e-01 3.04907948e-01 -4.57247734e-01 -4.91291821e-01 3.10443789e-01 6.06995881e-01 -1.85764804e-01 3.26056808e-01 5.56794107e-01 8.35107505e-01 -1.01027822e+00 -7.00278640e-01 -2.74959765e-02 -1.45476371e-01 -5.96036911e-01 6.63376153e-02 -4.09359396e-01 5.41493893e-01 -5.40046155e-01 7.24384367e-01 8.41927230e-01 5.09938933e-02 6.39416814e-01 -4.06207331e-03 -5.45051277e-01 8.84806156e-01 -1.07871068e+00 1.72480297e+00 -4.19959515e-01 4.07937795e-01 -1.08900711e-01 -1.11957955e+00 9.80575860e-01 1.65608555e-01 8.29122204e-04 -3.50380212e-01 4.27113265e-01 2.67040163e-01 -1.78692769e-02 -2.91260093e-01 5.99109411e-01 -4.79743630e-01 -8.90496612e-01 3.21294427e-01 5.41595995e-01 -8.32114145e-02 5.96730947e-01 -9.69512984e-02 1.35850418e+00 2.07576856e-01 4.87023145e-01 -6.22536540e-01 4.33405340e-01 2.26533383e-01 9.24915195e-01 6.74058259e-01 -3.01013559e-01 2.85233974e-01 5.02785981e-01 -2.16404796e-01 -9.19501603e-01 -1.29751742e+00 -5.50875902e-01 1.65675199e+00 1.12035982e-01 -1.89033002e-01 -1.02240610e+00 -1.38246560e+00 -3.25751662e-01 9.56523895e-01 -5.41501164e-01 3.39012831e-01 -1.02319574e+00 -8.81250858e-01 5.48773170e-01 6.65063739e-01 4.26867306e-02 -1.35913360e+00 -2.29749337e-01 3.57714951e-01 -4.55732457e-02 -1.32876694e+00 -2.26443380e-01 9.89568710e-01 -1.00941432e+00 -8.83075833e-01 -1.79274887e-01 -1.57473135e+00 5.92207432e-01 -3.67999762e-01 1.55547166e+00 -4.38945264e-01 3.62584740e-02 6.14921600e-02 -4.86526012e-01 -2.31297627e-01 -9.34119582e-01 3.78797948e-01 -4.51548770e-02 -4.93196845e-01 8.57787788e-01 -6.42502904e-01 4.90472168e-02 2.99929976e-02 -6.87170446e-01 -2.71749914e-01 5.69154441e-01 6.43355191e-01 9.33873177e-01 -9.42410454e-02 6.62517190e-01 -1.84457445e+00 3.07811886e-01 -3.35298091e-01 -6.22257531e-01 4.00969267e-01 -4.53739464e-01 4.96288091e-01 5.34358799e-01 -1.74910873e-01 -1.31712759e+00 4.33066189e-01 -4.46244150e-01 3.42541069e-01 -7.24260688e-01 1.34678483e-01 -5.37194073e-01 3.91631246e-01 4.63980794e-01 -3.43128622e-01 -6.86330914e-01 -5.76546729e-01 8.63171041e-01 4.99160498e-01 7.51175702e-01 -8.88028800e-01 5.80675066e-01 1.73187181e-01 -7.52439424e-02 -5.30948639e-01 -1.42667365e+00 -5.71012318e-01 -9.64479744e-01 2.52295792e-01 1.05781150e+00 -7.38171339e-01 -3.34881455e-01 6.26314655e-02 -1.29023767e+00 -3.95319052e-02 -6.29970849e-01 3.78514826e-01 -4.38337266e-01 3.91909927e-01 -8.91103327e-01 -7.31996655e-01 -1.56886637e-01 -8.35241079e-01 1.15917146e+00 1.94862142e-01 -3.09831321e-01 -1.16482997e+00 7.94811964e-01 2.36147001e-01 -2.53705233e-01 2.80307710e-01 1.18860507e+00 -1.19244611e+00 -1.66415125e-01 -7.17683882e-02 -3.71844135e-02 4.94362742e-01 2.40810171e-01 -1.31051242e-01 -1.15735304e+00 1.02746718e-01 1.36545524e-01 -3.42210323e-01 8.63634050e-01 4.23576117e-01 4.49446946e-01 -1.33434385e-01 -4.49592352e-01 2.19967574e-01 1.24892020e+00 1.98635399e-01 3.91240716e-01 2.49302685e-01 5.78565955e-01 9.28173482e-01 7.70487905e-01 -2.80780010e-02 4.19946551e-01 1.84054717e-01 1.17910594e-01 -8.96770209e-02 -2.65694052e-01 -3.50232065e-01 3.77781689e-01 1.10851955e+00 2.45010883e-01 -4.12838072e-01 -8.07511747e-01 7.81616449e-01 -1.50382590e+00 -6.10406876e-01 -3.65934342e-01 1.63230300e+00 1.09836411e+00 3.47922295e-01 -8.89854059e-02 2.82270312e-01 1.27542269e+00 8.72086436e-02 -2.57759899e-01 -5.87323248e-01 -3.19996595e-01 8.19805086e-01 5.02558887e-01 6.39367998e-01 -1.48069024e+00 1.41224074e+00 6.78096294e+00 5.97672641e-01 -6.55842245e-01 3.47674131e-01 7.50843406e-01 5.97298026e-01 -5.89165568e-01 3.82826924e-01 -1.48333490e+00 7.01130405e-02 9.54585433e-01 3.76290977e-01 -2.67853349e-01 1.07916415e+00 -1.98022872e-01 2.24225391e-02 -1.37871993e+00 4.44308907e-01 -1.54989764e-01 -1.08013129e+00 -3.92489523e-01 -1.02849595e-01 9.39381361e-01 3.68474307e-03 -2.61695862e-01 5.13444245e-01 8.52066576e-01 -6.17751300e-01 7.11017251e-01 -1.50727078e-01 7.50740170e-01 -5.26544392e-01 6.83289707e-01 3.81267995e-01 -1.24622738e+00 2.01097593e-01 -3.12301099e-01 -5.29539287e-02 3.73359323e-01 8.71525288e-01 -8.65657330e-01 3.00863951e-01 2.15290412e-01 5.01339853e-01 -4.07490462e-01 3.34074587e-01 -6.31224811e-01 1.02881551e+00 -4.41129625e-01 -1.27716646e-01 3.32666904e-01 -8.32717493e-02 4.40703750e-01 1.44281840e+00 5.09458296e-02 2.29911342e-01 3.99129689e-01 7.19723403e-01 -3.11860859e-01 3.38378817e-01 -4.09125179e-01 -1.64477140e-01 6.07036710e-01 1.31395960e+00 -1.23352063e+00 -4.18798506e-01 -2.97838390e-01 8.11018705e-01 6.18436813e-01 -3.08989853e-01 -5.28343081e-01 -1.72647890e-02 4.04372901e-01 -2.33673319e-01 4.85819012e-01 8.40358064e-02 -3.03369701e-01 -1.07721817e+00 -2.25062203e-02 -6.31246626e-01 4.94251341e-01 3.68135050e-02 -1.66543055e+00 8.94545615e-01 1.25436977e-01 -7.72452712e-01 -2.48240352e-01 -9.08210814e-01 -5.35375416e-01 4.41567957e-01 -1.58842742e+00 -1.21980381e+00 2.94996381e-01 2.39072531e-01 6.02150023e-01 -2.25001648e-02 1.18398607e+00 2.02512994e-01 -6.81131780e-01 5.17091334e-01 -1.10626161e-01 3.05010885e-01 4.06435877e-01 -1.67073965e+00 8.38763297e-01 1.24285734e+00 5.85395515e-01 5.05022645e-01 5.50597727e-01 -7.29471445e-01 -8.22802126e-01 -1.12790024e+00 1.61234891e+00 -7.73638070e-01 5.92732728e-01 -7.08855748e-01 -4.68627363e-01 9.85909998e-01 -2.77154632e-02 1.44358814e-01 9.47296441e-01 5.06451666e-01 -5.11179924e-01 3.03357631e-01 -1.14950228e+00 1.98805243e-01 1.41638219e+00 -4.85007495e-01 -1.14771926e+00 4.73600388e-01 9.44869995e-01 -2.57436156e-01 -6.96160674e-01 6.24173284e-01 3.72038968e-02 -1.03493059e+00 6.33343816e-01 -6.99085712e-01 2.63285547e-01 -9.50273275e-02 -5.44432819e-01 -8.61840427e-01 -3.63609314e-01 -4.50176179e-01 1.60572454e-01 1.93022943e+00 9.16814625e-01 -4.91405040e-01 1.16880465e+00 5.77924192e-01 -5.71295798e-01 -2.34234840e-01 -8.63347590e-01 -7.58155286e-01 4.60738569e-01 -5.92150688e-01 4.86020625e-01 1.02309036e+00 8.52494091e-02 8.30975711e-01 1.54271543e-01 3.15614641e-01 5.27013540e-01 5.01344800e-01 1.63878828e-01 -1.34151578e+00 -2.82690316e-01 -1.38239577e-01 -3.72755647e-01 -1.16317761e+00 7.62138486e-01 -1.26313627e+00 4.64702070e-01 -1.34692836e+00 1.69027910e-01 -8.54376912e-01 -1.92398474e-01 7.54509807e-01 -2.66418993e-01 2.07266554e-01 -4.98838387e-02 -1.31309301e-01 -6.17139876e-01 8.82555842e-02 9.69880104e-01 -6.29393011e-02 -2.09785476e-01 -1.41664862e-03 -6.61152363e-01 9.85160530e-01 7.79740751e-01 -1.04779601e+00 -1.74445152e-01 -2.15916038e-01 6.59015998e-02 -1.96504854e-02 -4.08135474e-01 -7.65629768e-01 -1.36613116e-01 4.80991229e-02 2.24946793e-02 -8.33229184e-01 -7.31379241e-02 -5.67907214e-01 -2.95027196e-01 1.11041225e-01 -5.25823832e-01 1.10907778e-01 -3.05419024e-02 3.33156854e-01 -3.20417255e-01 -7.24548817e-01 6.69376731e-01 -3.08769852e-01 -8.73341203e-01 1.26266643e-01 -4.07566994e-01 6.36775732e-01 9.21603501e-01 -2.41385773e-02 -3.78990650e-01 3.63392800e-01 -9.93463457e-01 -1.58606410e-01 2.62889206e-01 2.30417848e-01 -2.06569303e-02 -1.05959678e+00 -7.18854904e-01 7.82508329e-02 1.21789835e-01 3.47643048e-02 -7.43345842e-02 2.66463757e-01 -4.39865440e-01 3.12833786e-01 1.38910219e-01 -5.85211813e-01 -1.18247712e+00 5.71615100e-01 -1.83721617e-01 -6.83107436e-01 -3.57592136e-01 1.22031212e+00 4.20463443e-01 -1.11566389e+00 -1.26441242e-02 -7.08051324e-01 -4.18411613e-01 1.06310628e-01 1.45831361e-01 3.24820764e-02 2.32084751e-01 -8.61329615e-01 -5.38764298e-01 5.04578590e-01 -2.45049864e-01 4.00428586e-02 1.33655322e+00 -2.96660531e-02 -3.14402401e-01 4.71994549e-01 1.26599491e+00 3.56207728e-01 -8.61999989e-01 -9.52135101e-02 7.23578811e-01 5.04172929e-02 -1.78086042e-01 -7.19434381e-01 -7.51034021e-01 5.62037110e-01 5.29794395e-01 3.21349889e-01 9.51961040e-01 4.66197342e-01 8.34218740e-01 3.94273102e-01 3.03424567e-01 -1.04893529e+00 -3.87508661e-01 8.95063162e-01 -3.63487098e-03 -1.19709957e+00 -3.71585190e-01 -1.05775952e+00 -3.62652779e-01 9.57725167e-01 3.18972915e-01 -2.41282567e-01 6.89276040e-01 7.86710262e-01 3.34492356e-01 -1.76285580e-01 -4.67828155e-01 -4.81631666e-01 3.78574356e-02 5.28395414e-01 7.69689322e-01 3.03073823e-01 -6.00018382e-01 8.51498783e-01 -5.27240157e-01 -4.29831386e-01 2.40342632e-01 1.15355539e+00 -4.84551191e-01 -1.94001400e+00 -1.15517095e-01 1.79181069e-01 -9.29987609e-01 -3.63363802e-01 -4.21320587e-01 7.44599819e-01 6.80308759e-01 1.07177675e+00 1.32726610e-01 -2.52996743e-01 4.01095301e-01 3.47253799e-01 6.09538794e-01 -1.38223636e+00 -8.04784119e-01 -6.39567077e-02 6.44621193e-01 -1.01308614e-01 -1.05390155e+00 -7.22184718e-01 -1.46144867e+00 6.27299696e-02 -7.91423082e-01 4.37654883e-01 6.70044243e-01 1.14222205e+00 -3.56813967e-02 3.34592462e-01 7.25811958e-01 -5.25429785e-01 -2.95686781e-01 -8.97949040e-01 -6.34389639e-01 4.41370994e-01 -5.29848300e-02 -5.24661601e-01 -3.60482395e-01 2.37696990e-01]
[10.367127418518066, 9.667684555053711]
9999c650-3900-400a-9dc0-eeec9dd03fe9
global-spectral-filter-memory-network-for
2210.05567
null
https://arxiv.org/abs/2210.05567v2
https://arxiv.org/pdf/2210.05567v2.pdf
Global Spectral Filter Memory Network for Video Object Segmentation
This paper studies semi-supervised video object segmentation through boosting intra-frame interaction. Recent memory network-based methods focus on exploiting inter-frame temporal reference while paying little attention to intra-frame spatial dependency. Specifically, these segmentation model tends to be susceptible to interference from unrelated nontarget objects in a certain frame. To this end, we propose Global Spectral Filter Memory network (GSFM), which improves intra-frame interaction through learning long-term spatial dependencies in the spectral domain. The key components of GSFM is 2D (inverse) discrete Fourier transform for spatial information mixing. Besides, we empirically find low frequency feature should be enhanced in encoder (backbone) while high frequency for decoder (segmentation head). We attribute this to semantic information extracting role for encoder and fine-grained details highlighting role for decoder. Thus, Low (High) Frequency Module is proposed to fit this circumstance. Extensive experiments on the popular DAVIS and YouTube-VOS benchmarks demonstrate that GSFM noticeably outperforms the baseline method and achieves state-of-the-art performance. Besides, extensive analysis shows that the proposed modules are reasonable and of great generalization ability. Our source code is available at https://github.com/workforai/GSFM.
['Yujiu Yang', 'Yansong Tang', 'Yitong Wang', 'Xinyuan Zhao', 'Jiahao Wang', 'Ran Yu', 'Yong liu']
2022-10-11
null
null
null
null
['semi-supervised-video-object-segmentation', 'video-object-segmentation']
['computer-vision', 'computer-vision']
[ 1.01095930e-01 -1.55822635e-01 -5.44147074e-01 -3.50608140e-01 -5.52654326e-01 -2.75893629e-01 4.24886227e-01 -2.74803996e-01 -3.43735099e-01 5.77192664e-01 3.55026960e-01 1.75522909e-01 1.47263687e-02 -6.11416578e-01 -9.30792093e-01 -8.31803381e-01 -1.80796355e-01 -2.88340241e-01 7.65961885e-01 -1.66559853e-02 2.82667696e-01 1.24096163e-01 -1.25885618e+00 6.22582912e-01 6.82291031e-01 1.00123572e+00 6.97399080e-01 4.77494895e-01 2.33484227e-02 9.81291652e-01 -4.46847051e-01 7.01827109e-02 1.49626821e-01 -4.44717228e-01 -8.47218275e-01 3.25196773e-01 2.81974643e-01 -5.05891263e-01 -6.52955234e-01 1.21957386e+00 4.86879885e-01 2.83538580e-01 4.32919800e-01 -1.14084148e+00 -5.45235455e-01 6.40490770e-01 -8.71614754e-01 7.27150500e-01 1.97456926e-01 3.32248449e-01 7.55434394e-01 -9.97216284e-01 5.39912283e-01 1.25054812e+00 4.13874716e-01 4.50437188e-01 -9.01040852e-01 -7.39053905e-01 4.94429231e-01 4.54713821e-01 -1.29735637e+00 -5.26090980e-01 9.92457390e-01 -3.09142768e-01 8.26177955e-01 1.00789912e-01 4.58799839e-01 1.09929287e+00 3.36566120e-01 1.19536805e+00 9.87941206e-01 -1.12920418e-01 -1.25954628e-01 -8.88992175e-02 2.37688869e-01 5.97595870e-01 -3.36850077e-01 3.19820523e-01 -8.34516585e-01 2.70781070e-01 1.13361812e+00 6.93845842e-03 -6.50726199e-01 -9.43238065e-02 -1.23241961e+00 6.67225420e-01 5.22489309e-01 4.28494453e-01 -2.00875089e-01 2.76278615e-01 5.06111562e-01 1.43179089e-01 4.34777230e-01 -3.13513517e-01 -5.50267518e-01 -5.60957827e-02 -9.66861546e-01 -1.35613531e-01 2.99266279e-01 9.85597968e-01 7.53954887e-01 1.17460676e-01 -3.91277730e-01 8.27290952e-01 5.70815623e-01 2.13903069e-01 6.24202430e-01 -1.09676743e+00 3.86711329e-01 1.99748635e-01 -1.33284554e-01 -1.07593369e+00 -3.76054138e-01 -5.89548588e-01 -9.04252589e-01 -1.50838554e-01 2.65438616e-01 -4.86820266e-02 -6.74500465e-01 1.76577008e+00 4.57700014e-01 7.16985345e-01 -1.62253350e-01 1.13712096e+00 1.12965333e+00 8.01459908e-01 2.80122876e-01 -5.42027295e-01 1.19065535e+00 -1.32538903e+00 -8.70246291e-01 -1.83065519e-01 3.88143390e-01 -9.03933108e-01 8.52518260e-01 2.22066954e-01 -1.18975115e+00 -1.24893630e+00 -7.97538221e-01 5.43389209e-02 -1.86638292e-02 9.88429114e-02 5.42832077e-01 3.47756624e-01 -8.93440425e-01 6.29166543e-01 -8.38764429e-01 -2.12577268e-01 5.54386973e-01 4.05111343e-01 -7.59892985e-02 7.25502968e-02 -1.35185647e+00 3.25094908e-01 6.47531092e-01 7.28095695e-02 -8.98310661e-01 -6.07937634e-01 -6.55700564e-01 -2.54118919e-01 4.44403678e-01 -4.72844064e-01 1.05780506e+00 -1.31108189e+00 -1.32476532e+00 6.12366080e-01 -2.64022768e-01 -5.49999893e-01 1.98543519e-01 -3.43638659e-01 -4.73997563e-01 6.19703948e-01 1.42962605e-01 1.07595575e+00 1.13076150e+00 -1.14733219e+00 -7.20511436e-01 -2.90472716e-01 1.20638154e-01 3.28807324e-01 -4.98617411e-01 1.55262761e-02 -7.91860163e-01 -1.12985599e+00 1.30172029e-01 -5.95756114e-01 1.29740447e-01 -1.99119687e-01 -3.52804393e-01 -2.59993345e-01 1.09928453e+00 -6.77422523e-01 1.58697689e+00 -2.20501518e+00 -2.93830223e-02 -1.54900193e-01 1.91239685e-01 4.26867366e-01 -2.18243703e-01 7.73849264e-02 -2.20489934e-01 -8.89916271e-02 4.70509641e-02 4.74942885e-02 -3.53060871e-01 -8.00117925e-02 -1.63757280e-01 5.74548006e-01 1.01665005e-01 9.92766500e-01 -7.51135707e-01 -7.07852662e-01 4.43696022e-01 6.81429207e-01 -5.32259285e-01 8.99417326e-02 -7.99012259e-02 6.80233955e-01 -5.70994675e-01 5.84272087e-01 8.64167809e-01 -4.66879457e-01 -1.53384492e-01 -8.01506519e-01 -6.89838007e-02 1.02228045e-01 -1.05471170e+00 2.05989957e+00 -3.66795284e-04 5.26594460e-01 4.80373316e-02 -1.22156215e+00 5.73570728e-01 1.41446069e-01 6.01380110e-01 -8.68315339e-01 3.11885566e-01 -7.06440508e-02 -6.06806874e-02 -6.11295223e-01 9.42872167e-02 1.49365067e-01 4.62659508e-01 8.43198895e-02 2.55324900e-01 5.40723622e-01 2.49011129e-01 2.36482531e-01 8.14535379e-01 3.78595620e-01 1.55375496e-01 -5.02332449e-01 7.91549623e-01 -4.06248510e-01 6.63544297e-01 5.45807719e-01 -5.45090973e-01 5.94919264e-01 2.30509434e-02 -1.96754068e-01 -5.32770276e-01 -1.05696809e+00 -9.51692760e-02 1.24140203e+00 7.55211234e-01 -4.41712081e-01 -9.66793358e-01 -6.31808758e-01 -3.60550761e-01 3.43868047e-01 -5.49430966e-01 -1.98193744e-01 -7.95001328e-01 -6.38381541e-01 3.83882672e-01 6.41355395e-01 1.16104960e+00 -1.03679991e+00 -6.32314801e-01 3.08162928e-01 -5.92153132e-01 -1.15548635e+00 -9.27596688e-01 1.48409302e-03 -9.85835552e-01 -1.11439991e+00 -8.87418270e-01 -9.01579380e-01 2.35220134e-01 6.88486993e-01 1.04212856e+00 3.83008420e-02 -2.11413994e-01 4.81015503e-01 -4.13981438e-01 1.92594916e-01 1.52657419e-01 -5.00862487e-02 -2.54752785e-01 1.12517692e-01 5.09912252e-01 -5.69990277e-01 -1.01816547e+00 6.40935898e-01 -8.80044639e-01 2.47759625e-01 5.53802371e-01 7.84299612e-01 6.37554646e-01 1.79404467e-01 5.67533672e-01 -5.74570954e-01 1.14581138e-01 -5.61434090e-01 -2.33322397e-01 2.52653845e-02 -1.09437279e-01 -2.95197010e-01 3.49305123e-01 -5.48404872e-01 -1.20837593e+00 3.73620540e-02 -1.13253236e-01 -5.90488851e-01 -3.81505460e-01 1.45889595e-01 -3.37799907e-01 -8.25619847e-02 4.11518842e-01 4.58401412e-01 -2.38630697e-01 -2.42200598e-01 6.11161329e-02 4.71722782e-01 3.55315268e-01 -5.75492144e-01 4.67301995e-01 7.06859529e-01 -2.20658287e-01 -8.38391364e-01 -9.42590177e-01 -5.88114679e-01 -5.82738161e-01 -5.56628406e-01 1.06676531e+00 -1.20822465e+00 -7.22430408e-01 4.80104327e-01 -1.10059905e+00 -4.91962135e-01 -5.87678421e-03 4.65390503e-01 -6.19012177e-01 5.18657506e-01 -8.72323453e-01 -5.34609556e-01 -5.86594157e-02 -1.21501052e+00 1.09948421e+00 4.05939072e-01 -3.32144536e-02 -1.06512547e+00 -3.10226917e-01 4.23829377e-01 3.00266534e-01 3.92171890e-02 3.45928609e-01 -2.83110172e-01 -8.60077679e-01 5.14866710e-01 -4.48540747e-01 3.55232060e-01 2.54093390e-02 -1.94844082e-01 -1.00670195e+00 -2.10813671e-01 4.20988828e-01 -1.68266207e-01 1.22067499e+00 9.54337418e-01 1.23569357e+00 -9.86646339e-02 -3.60805601e-01 7.02670693e-01 1.24449706e+00 3.99946094e-01 6.92981303e-01 2.82546967e-01 8.77754629e-01 5.40186882e-01 9.32232201e-01 3.82516891e-01 3.46065134e-01 7.22091675e-01 3.05316538e-01 -1.92859992e-01 -4.46546406e-01 -4.73460704e-02 5.63459218e-01 7.23522365e-01 5.18025598e-04 -4.10135120e-01 -5.98355651e-01 3.72042239e-01 -2.14053369e+00 -1.07507920e+00 -3.27446431e-01 1.81541860e+00 7.43055165e-01 2.90457278e-01 1.91201627e-01 6.97633922e-02 9.34312642e-01 4.20226276e-01 -5.78900874e-01 2.27118671e-01 -4.02056396e-01 -4.65739891e-02 5.07278979e-01 3.64919007e-01 -1.50830185e+00 1.02775311e+00 5.12910891e+00 1.52223957e+00 -9.77392614e-01 4.20845479e-01 1.13983655e+00 -1.08954623e-01 7.87264854e-02 -1.79744467e-01 -8.09791982e-01 7.21031547e-01 7.55187213e-01 2.18147323e-01 2.08314970e-01 4.57415044e-01 5.20883381e-01 -3.27966124e-01 -7.89800882e-01 1.19176257e+00 -3.54290269e-02 -1.40753961e+00 1.18578551e-02 -1.00138709e-01 6.95247710e-01 2.73683202e-02 5.35264760e-02 5.75620905e-02 -2.80900180e-01 -9.27105546e-01 7.89290488e-01 5.72049439e-01 5.11763513e-01 -8.35898817e-01 4.12616074e-01 2.56856769e-01 -1.66278982e+00 -7.25990860e-03 -2.83622831e-01 7.31492639e-02 2.53634036e-01 6.05908692e-01 1.02445774e-01 3.37281078e-01 9.41581845e-01 1.07755864e+00 -3.99453133e-01 7.94492424e-01 -4.50269133e-02 8.20551276e-01 -1.33816615e-01 4.70934808e-01 3.62266511e-01 5.55290515e-03 5.33073246e-01 1.43775141e+00 1.10633709e-01 3.51167619e-01 4.77538764e-01 5.94021738e-01 1.01413086e-01 3.60524654e-02 -2.41980910e-01 9.55267549e-02 2.76237518e-01 1.07006931e+00 -1.02136612e+00 -4.50335145e-01 -6.56701386e-01 1.16395974e+00 -1.30742818e-01 6.58172131e-01 -1.27702856e+00 -5.79456836e-02 5.20044684e-01 2.04155594e-01 4.86442715e-01 -1.44751579e-01 -1.45792678e-01 -1.05143869e+00 -1.35208011e-01 -7.81124055e-01 4.21861768e-01 -6.71178997e-01 -9.74824548e-01 4.13549513e-01 -1.45431414e-01 -1.28499520e+00 1.84290424e-01 -3.15948308e-01 -3.90466154e-01 5.05514026e-01 -1.60691082e+00 -1.13515639e+00 -3.39222878e-01 1.05913711e+00 1.13933492e+00 -1.12711797e-02 2.21840695e-01 6.63851976e-01 -4.79732126e-01 5.27261257e-01 -1.92826033e-01 2.19859466e-01 8.32867503e-01 -7.22188234e-01 8.13858658e-02 8.82630348e-01 1.15168087e-01 5.16635954e-01 5.30859470e-01 -8.29366744e-01 -1.25924921e+00 -1.30551422e+00 4.80172276e-01 -7.12690428e-02 4.61517245e-01 5.31580150e-02 -9.02680337e-01 5.15087187e-01 4.10470575e-01 2.21426800e-01 4.87854213e-01 -4.53181475e-01 -9.95218009e-02 4.34691124e-02 -9.28975284e-01 3.06350768e-01 1.30987763e+00 -4.71392602e-01 -2.39803657e-01 3.92695457e-01 7.59411633e-01 -4.61647928e-01 -7.36422241e-01 5.72588384e-01 3.60794783e-01 -1.39687061e+00 1.12345982e+00 -9.86855775e-02 5.94494164e-01 -6.08232439e-01 -2.77013749e-01 -7.60826826e-01 -5.21344721e-01 -5.76753974e-01 -4.22204226e-01 1.29505658e+00 -8.27552378e-02 -1.69907272e-01 7.09334791e-01 1.30538970e-01 -2.01767862e-01 -8.00168931e-01 -7.17634737e-01 -7.72009730e-01 -2.44455040e-01 -5.49501061e-01 9.24869254e-02 8.99383545e-01 -2.06506908e-01 3.72109652e-01 -5.41065872e-01 2.07273021e-01 7.30969608e-01 -1.00126639e-02 5.01749039e-01 -7.32259274e-01 -5.69101155e-01 -3.65907609e-01 -3.55033129e-01 -1.67672908e+00 1.22080460e-01 -6.45265341e-01 -1.17444523e-01 -1.15065944e+00 4.33996409e-01 -2.64564209e-04 -5.16248465e-01 9.94878709e-02 -2.75090665e-01 5.79863429e-01 1.83260933e-01 2.92020440e-01 -9.96824622e-01 5.80188215e-01 1.52682555e+00 -7.27668032e-02 -1.77095234e-01 -1.10987842e-01 -5.13739049e-01 9.30913210e-01 8.71170640e-01 -2.30561301e-01 -5.14258862e-01 -4.49084342e-01 -3.66077155e-01 9.86543000e-02 5.77238321e-01 -1.07109141e+00 3.50955188e-01 -1.27662942e-01 5.83887756e-01 -7.53986776e-01 3.22040647e-01 -6.86612725e-01 1.02568187e-01 3.90979171e-01 -2.83368528e-01 -2.30343848e-01 2.94865698e-01 6.62071466e-01 -5.74295342e-01 3.29554603e-02 1.01509655e+00 -2.48933375e-01 -1.20189881e+00 3.87653053e-01 -3.08314413e-01 1.53669313e-01 9.71190751e-01 -4.72903252e-01 -8.13447461e-02 -3.70989442e-01 -8.34691107e-01 1.49606168e-01 1.31832600e-01 3.97707820e-01 6.65286660e-01 -1.39043450e+00 -5.18497825e-01 1.30950391e-01 -2.39777625e-01 -1.94849387e-01 8.56422961e-01 1.09259534e+00 -7.65937194e-02 4.00100887e-01 -8.82975385e-02 -9.47273791e-01 -1.26247227e+00 4.65151429e-01 1.81131467e-01 2.87815090e-02 -5.49996436e-01 1.32005155e+00 9.40776646e-01 2.05704674e-01 4.17148620e-01 -2.00802013e-01 -2.32907400e-01 2.01612756e-01 7.15918124e-01 4.33227360e-01 -3.84733558e-01 -1.00931275e+00 -3.37063491e-01 7.60141075e-01 -8.77577215e-02 1.58655092e-01 1.15590203e+00 -6.18541181e-01 1.50769427e-01 3.54210734e-01 1.41286838e+00 -2.76643157e-01 -1.74179840e+00 -3.40576530e-01 -2.28444234e-01 -6.04616463e-01 1.60891399e-01 -5.17957509e-01 -1.45455277e+00 8.94624531e-01 9.21355247e-01 1.61445886e-02 1.56143200e+00 -7.83956200e-02 1.01936221e+00 -4.24668863e-02 1.71848208e-01 -1.24573755e+00 4.31644082e-01 4.22110140e-01 6.72519267e-01 -1.33989334e+00 -1.41239718e-01 -6.73415363e-01 -5.63278496e-01 1.15463555e+00 8.08317423e-01 -3.21149439e-01 9.39442217e-01 3.00737888e-01 -1.13243777e-02 -7.69650936e-02 -6.16701722e-01 -2.46602997e-01 4.13712978e-01 4.98060763e-01 6.88033283e-01 -2.56400377e-01 -2.26221904e-01 4.97650892e-01 2.31614649e-01 2.69046370e-02 4.99930009e-02 7.68643975e-01 -6.47292972e-01 -8.31112146e-01 -3.23475033e-01 2.15994507e-01 -6.16087675e-01 -2.52732009e-01 8.02015215e-02 6.93582952e-01 4.29762512e-01 1.08632481e+00 -4.25104313e-02 -3.70603263e-01 -1.27432346e-01 -2.05440447e-01 4.52293038e-01 -2.44276240e-01 -5.79885066e-01 9.91532505e-01 -2.16220841e-01 -1.03613734e+00 -1.03026783e+00 -5.98109126e-01 -1.51361704e+00 -9.80001241e-02 -3.21591198e-01 -1.58386439e-01 1.20691180e-01 9.39850271e-01 3.39853942e-01 8.47401917e-01 4.22127217e-01 -1.06520772e+00 6.47748113e-02 -8.87475491e-01 -5.56796551e-01 4.07992601e-01 3.24932694e-01 -7.99455881e-01 -1.54259846e-01 4.75926965e-01]
[9.241135597229004, -0.03387390822172165]
93a68f6c-a8d1-4319-91a0-eaf95a6ed743
learning-word-embeddings-for-data-sparse-and
null
null
https://aclanthology.org/N18-4007
https://aclanthology.org/N18-4007.pdf
Learning Word Embeddings for Data Sparse and Sentiment Rich Data Sets
This research proposal describes two algorithms that are aimed at learning word embeddings for data sparse and sentiment rich data sets. The goal is to use word embeddings adapted for domain specific data sets in downstream applications such as sentiment classification. The first approach learns word embeddings in a supervised fashion via SWESA (Supervised Word Embeddings for Sentiment Analysis), an algorithm for sentiment analysis on data sets that are of modest size. SWESA leverages document labels to jointly learn polarity-aware word embeddings and a classifier to classify unseen documents. In the second approach domain adapted (DA) word embeddings are learned by exploiting the specificity of domain specific data sets and the breadth of generic word embeddings. The new embeddings are formed by aligning corresponding word vectors using Canonical Correlation Analysis (CCA) or the related nonlinear Kernel CCA. Experimental results on binary sentiment classification tasks using both approaches for standard data sets are presented.
['Prathusha Kameswara Sarma']
2018-06-01
null
null
null
naacl-2018-6
['learning-word-embeddings']
['methodology']
[ 6.93806857e-02 1.55034224e-02 -5.97170115e-01 -7.09644735e-01 -4.74962413e-01 -7.29258955e-01 6.25943840e-01 4.90508050e-01 -6.24571204e-01 2.52562970e-01 7.41322458e-01 -1.25684187e-01 -2.10571945e-01 -6.42316401e-01 2.32727528e-01 -8.23174715e-01 -1.61125306e-02 5.01651704e-01 -2.57079691e-01 -6.50826871e-01 4.94056791e-01 3.89416516e-01 -1.28737497e+00 1.68224871e-01 1.46320224e-01 1.03225505e+00 -2.35846475e-01 8.53503942e-01 -8.57935846e-01 2.32577100e-01 -3.83564860e-01 -3.16642314e-01 3.87083411e-01 1.16158482e-02 -6.77274346e-01 -1.07866793e-03 2.36126512e-01 2.47436583e-01 6.08385466e-02 7.81514943e-01 4.92605299e-01 2.10421428e-01 9.98198450e-01 -1.34852207e+00 -1.25353456e+00 3.68041396e-01 -4.27236527e-01 3.77430350e-01 1.49739683e-01 -2.70368308e-01 1.58168685e+00 -1.30603540e+00 5.39329708e-01 1.17974615e+00 7.08729982e-01 6.97677553e-01 -1.11936939e+00 -4.61485088e-01 3.58756959e-01 -1.83103859e-01 -1.00227427e+00 -5.39314300e-02 9.71359015e-01 -6.38907433e-01 1.20945096e+00 -3.13809067e-02 5.64139009e-01 1.24895656e+00 1.86076552e-01 4.95721132e-01 9.50413704e-01 -5.99215567e-01 5.80929697e-01 8.39301050e-01 9.35295880e-01 1.47124708e-01 4.43715870e-01 -2.86788911e-01 -6.98954344e-01 -3.58067870e-01 -8.49183649e-02 3.41407508e-01 1.70520127e-01 -7.80472815e-01 -1.02633464e+00 1.44028771e+00 1.93565026e-01 5.77045262e-01 -1.80906191e-01 -5.83589934e-02 8.24068248e-01 7.52096236e-01 9.94292080e-01 9.60249841e-01 -1.31064487e+00 1.32441735e-02 -5.53211093e-01 1.77597076e-01 7.44558513e-01 7.00283289e-01 8.50347877e-01 1.90733373e-01 -1.00616058e-02 9.64663446e-01 6.62461519e-01 3.82760555e-01 1.38619125e+00 7.57451653e-02 1.87307984e-01 8.74287605e-01 -5.91333136e-02 -1.00138140e+00 -6.89571500e-01 -1.16354510e-01 -1.56095520e-01 -5.59950843e-02 1.67013705e-01 -4.20752048e-01 -1.04997170e+00 1.44747114e+00 2.65315533e-01 -7.71707594e-02 6.83030844e-01 6.18995726e-01 9.20690715e-01 6.43776894e-01 2.39866003e-01 6.85615465e-02 1.73179209e+00 -8.43600690e-01 -7.30593264e-01 -6.03103399e-01 1.19118333e+00 -5.72911322e-01 1.35452604e+00 3.08264524e-01 -2.69445717e-01 -3.59409600e-01 -1.27111876e+00 -3.13067019e-01 -1.53604698e+00 -2.71830261e-01 8.81448984e-01 1.12755191e+00 -8.20150614e-01 -7.24863112e-02 -2.57127732e-01 -4.74259853e-01 5.47892928e-01 4.64226812e-01 -5.64592063e-01 -8.35927501e-02 -1.24974787e+00 8.34095359e-01 3.02147985e-01 -4.53820914e-01 -3.68117303e-01 -8.89651537e-01 -1.46734691e+00 7.45659173e-02 -4.37087625e-01 -1.49647191e-01 7.89542794e-01 -1.09605086e+00 -1.25123346e+00 1.09228098e+00 -1.47221193e-01 -1.72825754e-01 -4.95770186e-01 -1.83555081e-01 -8.19395781e-01 -1.44814745e-01 2.43161052e-01 3.69807988e-01 1.12941551e+00 -1.11746824e+00 -4.19086933e-01 -7.10959792e-01 -2.26020724e-01 4.63105403e-02 -1.56529224e+00 -3.33909430e-02 1.75372586e-01 -7.73862123e-01 -7.50031918e-02 -7.23932683e-01 -5.31564772e-01 -2.39652976e-01 7.42916614e-02 -4.85234767e-01 1.26259363e+00 -1.17450885e-01 1.24881637e+00 -2.40537405e+00 1.02286622e-01 3.45159352e-01 1.24741197e-01 2.26590633e-01 -6.52064621e-01 6.36587501e-01 -6.76355243e-01 1.30919054e-01 -6.95740059e-02 -3.76520932e-01 1.29392341e-01 3.43950540e-01 -5.74615896e-01 3.64460677e-01 7.08791614e-01 9.26397383e-01 -1.08490443e+00 -9.35380682e-02 1.36991031e-02 4.32287693e-01 -6.53228998e-01 1.51517630e-01 -4.29573692e-02 -2.53021181e-01 -6.90660477e-01 6.61165476e-01 5.53722322e-01 1.35168120e-01 4.19484168e-01 -1.59792349e-01 6.99371994e-02 1.84523016e-01 -1.24528360e+00 1.56512976e+00 -8.41623724e-01 8.34573567e-01 -4.51072097e-01 -1.26456368e+00 1.28904390e+00 2.58613199e-01 2.91700155e-01 -3.35591912e-01 4.48626250e-01 2.71571930e-02 -2.58026659e-01 -5.39899111e-01 8.11046362e-01 -6.10250175e-01 -7.36372173e-01 5.24369121e-01 7.64582276e-01 -4.79219943e-01 1.12465478e-01 2.65176564e-01 8.49762022e-01 -4.65402961e-01 4.92749035e-01 -5.64357042e-01 6.38363302e-01 2.58576423e-01 1.73255920e-01 2.00817168e-01 -1.11942388e-01 5.03583789e-01 8.14018786e-01 -7.55673826e-01 -8.82214189e-01 -9.42977726e-01 -3.89665484e-01 1.62253952e+00 -1.42211959e-01 -7.97509134e-01 -1.10069744e-01 -9.51526582e-01 3.02315533e-01 5.85611761e-01 -1.11729038e+00 -4.20009732e-01 -5.66659942e-02 -8.73884201e-01 4.34861369e-02 9.23740983e-01 -5.69212675e-01 -9.18004990e-01 -1.13912411e-01 9.86692905e-02 8.92728090e-01 -9.11630094e-01 -2.84266740e-01 9.46096838e-01 -8.63990963e-01 -1.05635345e+00 -4.54147905e-01 -1.23284674e+00 8.14032972e-01 3.70749027e-01 9.28490102e-01 -4.57985371e-01 -3.11455816e-01 6.59549177e-01 -8.12199771e-01 -7.17714250e-01 2.13251412e-01 1.39061630e-01 6.68243170e-01 2.32007414e-01 1.41399968e+00 -3.22477698e-01 -2.29608104e-01 -8.88940990e-02 -1.27267754e+00 -1.01755559e+00 2.10995764e-01 1.17465734e+00 3.62928599e-01 -2.49023244e-01 8.72936249e-01 -1.15933025e+00 1.35360873e+00 -9.22522962e-01 -2.30914846e-01 -2.09102407e-01 -1.00709772e+00 1.88115314e-01 8.06895614e-01 -7.24170983e-01 -6.55816853e-01 -4.44633216e-02 1.13783069e-02 5.66527247e-03 -1.31768271e-01 5.39239645e-01 1.92078024e-01 2.38922551e-01 7.97899067e-01 6.42283708e-02 -8.70808437e-02 -3.59381348e-01 8.22304308e-01 1.20504045e+00 -6.36971148e-04 -3.13285172e-01 7.46542990e-01 4.71925914e-01 -5.28009176e-01 -9.44288671e-01 -1.10336041e+00 -1.12491238e+00 -8.24200690e-01 2.80079395e-01 1.09728658e+00 -9.44540381e-01 2.17007883e-02 1.45904173e-03 -7.23384440e-01 1.37087405e-01 -8.06979239e-01 6.60931766e-01 -1.78107455e-01 3.11609227e-02 -1.46550983e-01 -5.90206325e-01 -2.97743440e-01 -8.21482480e-01 1.10567772e+00 2.29999155e-01 -8.00009668e-01 -1.74382699e+00 7.65892565e-01 8.85271206e-02 3.47683042e-01 -3.35827731e-02 9.74805653e-01 -1.65379977e+00 7.08355367e-01 -8.11332703e-01 -3.95418443e-02 7.81916082e-01 4.04568940e-01 -3.07497770e-01 -1.20820415e+00 -2.61020482e-01 -1.67020693e-01 -4.98102784e-01 8.56545508e-01 1.17084503e-01 8.80235434e-01 2.61205379e-02 -2.61163145e-01 5.90998054e-01 1.60717463e+00 2.38963794e-02 2.84252852e-01 6.11081660e-01 5.93609691e-01 8.04926932e-01 5.90103626e-01 4.45904374e-01 2.17042983e-01 7.77157620e-02 1.36766210e-01 2.44328566e-02 3.15335870e-01 -7.58568645e-02 5.54123521e-01 1.02734268e+00 6.80799067e-01 -1.04988627e-01 -9.16514158e-01 1.30300128e+00 -1.38279736e+00 -4.59789842e-01 -2.67109454e-01 1.47888613e+00 6.23010457e-01 1.43299446e-01 -2.10874781e-01 3.73302013e-01 1.90983206e-01 7.40657628e-01 -2.86337048e-01 -1.49020720e+00 -1.45885170e-01 8.58476639e-01 5.37688673e-01 3.38031918e-01 -1.11496341e+00 1.12095594e+00 6.52978802e+00 3.88230920e-01 -9.66168404e-01 3.06211293e-01 2.04492897e-01 -7.08524883e-02 -5.89599967e-01 -1.32638976e-01 -8.50040615e-01 2.24632725e-01 1.07912171e+00 -2.21278906e-01 -3.36825103e-01 1.27074456e+00 -1.04779072e-01 2.60206491e-01 -1.02072716e+00 8.30806971e-01 5.39358079e-01 -1.30188644e+00 1.29407451e-01 -2.31902733e-01 1.06303227e+00 1.86077338e-02 5.72341383e-01 5.19230783e-01 5.29375911e-01 -1.09403026e+00 9.80725884e-03 8.61811936e-02 7.58597553e-01 -7.02069104e-01 1.11446226e+00 -3.57536525e-01 -9.55686390e-01 -3.66755128e-01 -5.11634648e-01 -3.03007662e-01 -1.20596051e-01 5.21451354e-01 -7.77313054e-01 3.93148288e-02 8.76370966e-01 1.28554440e+00 -5.08596659e-01 2.09227949e-01 -1.49287850e-01 5.58593035e-01 6.08175434e-02 -3.13633680e-01 7.30074525e-01 -1.86984167e-01 2.86993057e-01 1.55970180e+00 8.82638444e-04 -1.92216784e-01 -2.35246599e-01 2.54678637e-01 -1.01678804e-01 3.78130466e-01 -8.71105134e-01 -4.04857814e-01 1.34183869e-01 1.66842163e+00 -4.25625712e-01 -2.84061164e-01 -7.27502644e-01 9.15798187e-01 3.99622917e-02 2.57235169e-01 -5.28868549e-02 -6.66865289e-01 1.45388806e+00 -2.92812914e-01 6.24675393e-01 -3.14293623e-01 -5.21732271e-01 -1.22983444e+00 -3.17772776e-01 -5.63239396e-01 5.96288919e-01 -5.52040637e-01 -1.89504004e+00 5.09988248e-01 -2.48441786e-01 -1.21913099e+00 -4.41904403e-02 -1.44640565e+00 -8.18211317e-01 7.68528521e-01 -2.02635360e+00 -1.01742113e+00 -6.54412881e-02 6.35276616e-01 7.13581741e-01 -7.18726873e-01 1.39295769e+00 -1.09252505e-01 -3.44568998e-01 5.28231800e-01 6.36069715e-01 2.80016631e-01 1.03502548e+00 -1.55138361e+00 1.76012859e-01 2.75423437e-01 2.96632648e-01 8.58372092e-01 6.12718046e-01 -1.54965430e-01 -1.42530549e+00 -9.81298983e-01 1.23843658e+00 -8.84668112e-01 1.30025947e+00 -7.39827752e-01 -6.07757747e-01 7.23773241e-01 1.60068512e-01 3.91796559e-01 1.79892659e+00 4.19209033e-01 -8.62225473e-01 -1.70874938e-01 -1.12048066e+00 3.45389336e-01 3.36346179e-01 -6.82474017e-01 -1.10294616e+00 4.59109306e-01 8.52604806e-01 2.31496289e-01 -9.78772223e-01 -2.94540614e-01 5.63051939e-01 -2.70366549e-01 9.40527856e-01 -1.54162180e+00 5.60462534e-01 -1.31257668e-01 -5.49807131e-01 -1.77388048e+00 -1.73615277e-01 -1.07410505e-01 2.81466190e-02 1.08618116e+00 6.07348025e-01 -8.71679664e-01 6.95473850e-01 2.83907682e-01 2.42882684e-01 -6.29132509e-01 -6.61094606e-01 -6.81069791e-01 4.97519970e-01 -6.18933737e-01 6.05057299e-01 1.43419266e+00 4.04411763e-01 5.78325033e-01 -3.30785066e-02 1.36990964e-01 1.03421479e-01 6.28009886e-02 6.46093190e-01 -1.35978460e+00 2.47497067e-01 -2.12307498e-01 -8.98717821e-01 -7.04292119e-01 4.69487101e-01 -1.02792966e+00 -4.07803148e-01 -1.28093421e+00 -3.14478904e-01 -3.02089125e-01 -8.04235816e-01 4.19219047e-01 5.84603921e-02 4.39867526e-01 -1.43205911e-01 -2.79886335e-01 -3.68997127e-01 5.65023005e-01 4.59610343e-01 -3.76367867e-01 -3.01678777e-01 -4.39192593e-01 -1.39238179e+00 7.54314005e-01 8.82353723e-01 -8.04761827e-01 -4.64482307e-01 -3.07206362e-01 6.00404501e-01 -8.52106810e-01 -1.67245582e-01 -2.86077112e-01 -8.14318582e-02 -9.23593044e-02 2.71662921e-01 -1.76068977e-01 2.42783874e-01 -1.12178195e+00 -9.16866362e-01 1.07014887e-01 -5.77315748e-01 2.16777042e-01 2.63169318e-01 7.19867945e-01 -5.40492594e-01 -4.90283608e-01 5.05973756e-01 2.23718941e-01 -1.12857199e+00 1.80204913e-01 -5.30464172e-01 2.23727882e-01 1.18285501e+00 -3.79101068e-01 1.08934857e-01 -4.38262969e-02 -8.41493070e-01 2.76902378e-01 1.93346113e-01 1.04628301e+00 6.96578562e-01 -1.44683003e+00 -4.88157183e-01 5.98767042e-01 1.00282979e+00 -3.15586358e-01 -2.40177974e-01 3.49280357e-01 -1.79887891e-01 4.06585902e-01 -4.99811769e-02 -1.16644926e-01 -1.10112906e+00 6.60949767e-01 -1.51627669e-02 -2.76199520e-01 -8.29765499e-02 1.17933571e+00 5.82786649e-02 -9.31763113e-01 -2.95153588e-01 -3.69276315e-01 -8.78559470e-01 9.13887739e-01 6.98264420e-01 -1.33746013e-01 8.83608162e-02 -6.09927714e-01 -5.89530766e-01 8.45003664e-01 -3.89498800e-01 1.91583410e-01 1.90586054e+00 -5.52430972e-02 1.24774668e-02 6.04060292e-01 1.82001507e+00 2.04816341e-01 -6.94138587e-01 -3.32037538e-01 3.46910506e-01 -5.14440894e-01 3.31517130e-01 -4.46050555e-01 -1.00672543e+00 9.57678199e-01 7.10466743e-01 4.05676484e-01 8.21940184e-01 1.08715363e-01 6.38880670e-01 4.12495703e-01 -3.70529741e-01 -1.40692472e+00 3.32655907e-01 7.04423308e-01 4.78231609e-01 -1.43953311e+00 -4.53999601e-02 1.54773518e-01 -1.02665377e+00 1.56376791e+00 3.76514077e-01 -5.18461764e-01 1.14737427e+00 2.38158792e-01 3.85008305e-01 -6.61111057e-01 -6.81621492e-01 -3.18788975e-01 2.23040879e-01 1.03579688e+00 4.24865454e-01 -1.18420934e-02 -3.94381553e-01 1.22662318e+00 -1.62274316e-01 -4.76109415e-01 6.49732649e-01 9.01881039e-01 -3.55975777e-01 -1.36238372e+00 -2.11476862e-01 5.82950115e-01 -4.35824305e-01 -4.69940186e-01 -6.01210356e-01 5.76882720e-01 1.11449286e-01 9.70342457e-01 3.34548831e-01 -3.44143510e-01 4.92440790e-01 4.63394523e-01 -2.17505500e-01 -1.22425842e+00 -6.71641946e-01 -5.63906491e-01 -8.63636434e-02 -2.37540826e-01 -4.25041556e-01 -4.99952048e-01 -1.10775030e+00 3.55507672e-01 -4.10710901e-01 2.09299862e-01 8.87909114e-01 7.97575414e-01 3.22841763e-01 4.36661363e-01 9.36777353e-01 -5.90992153e-01 -4.05574054e-01 -7.91951537e-01 -9.38626051e-01 6.83927476e-01 5.63275278e-01 -6.22393966e-01 -7.62201130e-01 8.03956091e-02]
[10.429464340209961, 8.601731300354004]
77b8429a-bacd-428f-b03f-81d7015c2e7a
mtcue-learning-zero-shot-control-of-extra
2305.15904
null
https://arxiv.org/abs/2305.15904v1
https://arxiv.org/pdf/2305.15904v1.pdf
MTCue: Learning Zero-Shot Control of Extra-Textual Attributes by Leveraging Unstructured Context in Neural Machine Translation
Efficient utilisation of both intra- and extra-textual context remains one of the critical gaps between machine and human translation. Existing research has primarily focused on providing individual, well-defined types of context in translation, such as the surrounding text or discrete external variables like the speaker's gender. This work introduces MTCue, a novel neural machine translation (NMT) framework that interprets all context (including discrete variables) as text. MTCue learns an abstract representation of context, enabling transferability across different data settings and leveraging similar attributes in low-resource scenarios. With a focus on a dialogue domain with access to document and metadata context, we extensively evaluate MTCue in four language pairs in both translation directions. Our framework demonstrates significant improvements in translation quality over a parameter-matched non-contextual baseline, as measured by BLEU (+0.88) and Comet (+1.58). Moreover, MTCue significantly outperforms a "tagging" baseline at translating English text. Analysis reveals that the context encoder of MTCue learns a representation space that organises context based on specific attributes, such as formality, enabling effective zero-shot control. Pre-training on context embeddings also improves MTCue's few-shot performance compared to the "tagging" baseline. Finally, an ablation study conducted on model components and contextual variables further supports the robustness of MTCue for context-based NMT.
['Carolina Scarton', 'Robert Flynn', 'Sebastian Vincent']
2023-05-25
null
null
null
null
['nmt']
['computer-code']
[ 4.50119078e-01 6.58939872e-03 -6.42017245e-01 -3.85169625e-01 -1.19325626e+00 -7.95037270e-01 1.13407505e+00 1.95665643e-01 -6.21080577e-01 8.92180562e-01 7.63093829e-01 -7.68437326e-01 4.05609548e-01 -5.88141918e-01 -5.71078777e-01 -3.00604522e-01 4.46296334e-01 6.18381023e-01 -4.33939695e-01 -5.50252438e-01 6.37090579e-02 -2.79065073e-01 -1.00897670e+00 3.22541326e-01 9.79779899e-01 4.83878672e-01 2.95586854e-01 5.97157001e-01 -3.93069178e-01 1.70017272e-01 -7.63986945e-01 -7.19392896e-01 1.24621533e-01 -5.65266490e-01 -7.45945752e-01 -3.40264142e-01 5.95457852e-01 7.61057884e-02 -2.00302020e-01 6.01251721e-01 7.55268574e-01 3.74977626e-02 6.03146672e-01 -7.78353274e-01 -1.36041689e+00 8.41388345e-01 -4.31643240e-02 3.40205818e-01 3.94107163e-01 1.53425097e-01 1.25444973e+00 -1.28217208e+00 8.68588448e-01 1.11618590e+00 6.76913023e-01 8.36336076e-01 -1.58670974e+00 -4.64717776e-01 -1.39170915e-01 -7.48608857e-02 -1.06018591e+00 -6.63652718e-01 2.60011852e-01 -2.81785309e-01 1.38964677e+00 4.69632477e-01 3.12422127e-01 1.87289011e+00 4.56489295e-01 5.46272993e-01 1.11900294e+00 -1.05346572e+00 2.43871436e-01 1.61413521e-01 2.82553080e-02 2.41939262e-01 -6.90709949e-02 1.11397721e-01 -6.94584191e-01 -1.95602998e-01 4.70660597e-01 -4.04561520e-01 -9.81244817e-02 2.14733467e-01 -1.65190709e+00 8.62634182e-01 -1.91269115e-01 3.84209692e-01 -4.19452451e-02 9.02388990e-02 6.72744811e-01 5.16741931e-01 8.27207446e-01 7.05228686e-01 -6.61992788e-01 -6.37451112e-01 -8.34015727e-01 -3.64283659e-02 9.27487314e-01 1.32260287e+00 7.43407667e-01 -1.82878792e-01 -8.69248688e-01 1.13135910e+00 -1.12569228e-01 8.64544392e-01 7.20004082e-01 -6.65249288e-01 9.81746674e-01 4.81132656e-01 6.66592568e-02 -4.79775161e-01 1.22572035e-01 -4.93206054e-01 -4.96067554e-01 -5.13597250e-01 1.53866678e-01 -4.71233457e-01 -8.79948556e-01 2.01456165e+00 1.02245808e-01 -2.63945647e-02 3.30661654e-01 7.28723526e-01 6.45368576e-01 7.39420474e-01 2.14473888e-01 -1.13659263e-01 1.40423155e+00 -9.90170479e-01 -9.53573942e-01 -6.61245465e-01 7.39572704e-01 -1.07022524e+00 1.69451118e+00 -3.04446071e-01 -8.17012608e-01 -3.99126410e-01 -9.35838580e-01 -4.26622838e-01 -6.07023776e-01 3.24136466e-01 5.63635468e-01 6.43679917e-01 -1.12193871e+00 4.56255794e-01 -5.46490192e-01 -6.61134124e-01 8.19242448e-02 3.00523639e-01 -4.33049262e-01 -1.58320084e-01 -1.57051849e+00 1.18938684e+00 -1.15464786e-02 -1.99908823e-01 -4.44935352e-01 -7.07382858e-01 -1.00658059e+00 -6.28632633e-03 2.35366508e-01 -8.63830745e-01 1.42045867e+00 -1.09322047e+00 -1.58462477e+00 6.82889760e-01 -6.04515970e-01 -3.28746289e-01 2.98508495e-01 -3.98498654e-01 -4.51223224e-01 -2.03225359e-01 3.12887460e-01 6.98723912e-01 5.53341687e-01 -8.22141945e-01 -4.53746766e-01 -6.23557009e-02 -6.32211715e-02 4.31891114e-01 -5.76083779e-01 4.32208896e-01 -5.46910644e-01 -8.54864180e-01 -4.35764462e-01 -1.06123352e+00 2.97109764e-02 -4.81027007e-01 -6.89836144e-02 -3.01692277e-01 7.61612654e-01 -8.80658567e-01 1.63925910e+00 -1.79087520e+00 3.34687114e-01 -3.24184448e-01 -3.37563783e-01 3.01538914e-01 -4.40494537e-01 7.74297476e-01 2.48836935e-01 5.02396345e-01 -4.15610313e-01 -7.10127711e-01 1.73502102e-01 5.99870145e-01 -2.92498708e-01 9.40604806e-02 5.60967147e-01 1.41476810e+00 -1.06818998e+00 -3.93786073e-01 1.85505655e-02 6.70921147e-01 -4.10033166e-01 1.21815056e-01 -3.26969683e-01 3.13719541e-01 3.39969322e-02 7.06106246e-01 4.74104136e-02 8.87376815e-02 4.77511168e-01 2.46438861e-01 1.32249072e-02 7.51714051e-01 -4.21660334e-01 2.00623798e+00 -1.00707591e+00 9.49380398e-01 -1.56267107e-01 -3.92795652e-01 9.55324888e-01 6.56845748e-01 -6.99535757e-02 -8.87353539e-01 1.22739017e-01 3.83978218e-01 -1.11251146e-01 -3.80490482e-01 1.02660918e+00 -2.09722653e-01 -4.83987421e-01 6.61393881e-01 3.63483816e-01 2.03611646e-02 4.50751046e-03 1.66165143e-01 1.03610277e+00 3.51552963e-01 2.59609640e-01 -4.29199100e-01 1.48736283e-01 -1.46247093e-02 7.08297610e-01 5.53106308e-01 -1.26774892e-01 5.97445786e-01 2.70855546e-01 -1.22183226e-01 -1.22302580e+00 -7.08646953e-01 -1.05838798e-01 1.50637412e+00 -1.82746142e-01 -7.13714898e-01 -9.46373761e-01 -9.35278952e-01 3.77403125e-02 1.01087487e+00 -9.18011189e-01 -1.29993603e-01 -9.14209843e-01 -6.50689304e-01 5.19420147e-01 4.88026381e-01 3.04727480e-02 -8.82646680e-01 -2.44781658e-01 4.42429751e-01 -6.63701653e-01 -1.06200337e+00 -8.69627953e-01 2.92854369e-01 -8.15228939e-01 -5.11539936e-01 -3.90762210e-01 -7.46880352e-01 2.34240517e-01 1.67467371e-01 1.59385669e+00 -6.87323809e-02 3.81456316e-02 2.99555451e-01 -6.09802246e-01 -3.69262755e-01 -7.99486041e-01 6.18162930e-01 1.52169824e-01 -3.02056879e-01 9.10635710e-01 -5.19857705e-01 -3.62557024e-01 3.81852448e-01 -6.52772427e-01 -9.51353461e-03 5.40638268e-01 1.23891640e+00 4.89677787e-01 -8.81540298e-01 7.66852200e-01 -8.13376248e-01 9.61994231e-01 -5.72535455e-01 -9.51179042e-02 5.25577188e-01 -7.73287773e-01 3.86288874e-02 4.93976474e-01 -5.31977177e-01 -1.00224662e+00 -4.78264004e-01 2.82159358e-01 -1.27860606e-01 7.55877718e-02 5.57814717e-01 -2.79345244e-01 3.63743901e-01 8.12036157e-01 7.99095556e-02 -1.21609941e-01 -4.15857077e-01 6.17412508e-01 1.10720575e+00 5.91000676e-01 -1.10204411e+00 4.71889824e-01 -2.60134518e-01 -3.80385101e-01 -5.19353092e-01 -3.75052035e-01 -2.76345342e-01 -6.93074644e-01 1.66464135e-01 8.32762539e-01 -9.31964517e-01 -7.15315342e-02 -2.24854961e-01 -1.15807760e+00 -5.71933389e-01 -2.48889104e-01 5.25619209e-01 -4.37771976e-01 1.95152491e-01 -7.93012261e-01 -5.18887460e-01 -6.56155467e-01 -1.15296793e+00 1.28835285e+00 -9.70423892e-02 -7.11778760e-01 -1.19491696e+00 2.01496720e-01 4.40072954e-01 6.15959466e-01 2.31711585e-02 1.22250342e+00 -7.65239000e-01 -3.20520669e-01 6.67703003e-02 -2.78223753e-02 2.12827876e-01 2.23009095e-01 -1.61735252e-01 -9.78344917e-01 -2.97449321e-01 -2.80120850e-01 -2.73334324e-01 5.08156061e-01 -4.52240482e-02 3.60545933e-01 -4.82539803e-01 -1.30367488e-01 3.81455302e-01 1.27556598e+00 -3.98578960e-03 3.45646501e-01 4.65251118e-01 5.00068069e-01 5.68844855e-01 6.50625944e-01 7.76174590e-02 3.95130336e-01 9.19580996e-01 1.67175338e-01 3.87841575e-02 -3.63625616e-01 -3.76962662e-01 6.52618706e-01 1.30065060e+00 1.14390418e-01 -3.57632965e-01 -1.04170370e+00 7.35058188e-01 -1.82091355e+00 -7.93573678e-01 1.00481786e-01 2.07950282e+00 1.37094176e+00 -1.14383437e-01 -3.37847650e-01 -5.31827748e-01 8.54467332e-01 -3.15136127e-02 -3.50569516e-01 -1.05581295e+00 -2.64057279e-01 5.23014903e-01 4.30602640e-01 7.23166823e-01 -9.60101187e-01 1.37158155e+00 6.37453270e+00 7.52785504e-01 -1.09314799e+00 5.53090096e-01 3.85002494e-01 -3.26691240e-01 -6.71287179e-01 1.86942101e-01 -8.09128344e-01 6.80998206e-01 1.47581518e+00 -3.44073534e-01 6.69777393e-01 6.02692723e-01 1.56079188e-01 2.29751557e-01 -1.47489607e+00 6.24498308e-01 2.20339507e-01 -1.37137008e+00 4.89406772e-02 3.25296700e-01 8.65957975e-01 2.64378071e-01 1.25675499e-01 8.66988897e-01 4.08922881e-01 -1.10389328e+00 7.16295362e-01 4.08821888e-02 1.45006299e+00 -7.23061085e-01 7.42555499e-01 1.25685468e-01 -7.16090739e-01 3.89538216e-03 -3.29493880e-01 -1.00853667e-01 6.53334409e-02 3.37017655e-01 -1.29339731e+00 7.46245384e-01 1.54682741e-01 5.43642998e-01 -5.58862805e-01 1.29313603e-01 -4.52368259e-01 9.31915045e-01 -2.81379074e-02 -4.63520586e-02 4.41312402e-01 -1.99744880e-01 6.24312401e-01 2.00538445e+00 3.70382309e-01 -1.64828643e-01 1.18678048e-01 6.27472162e-01 -4.30006087e-01 3.27307910e-01 -5.35973728e-01 -1.61757454e-01 9.94813263e-01 1.03083646e+00 -1.35211781e-01 -4.71929252e-01 -5.84588766e-01 1.15727341e+00 3.52236420e-01 4.63444680e-01 -6.17054462e-01 -3.01852942e-01 1.07485247e+00 -1.61100745e-01 2.29578063e-01 -4.21646565e-01 -5.61985433e-01 -1.40536749e+00 1.66464016e-01 -1.10106933e+00 -7.51001835e-02 -4.68987405e-01 -1.05560362e+00 7.64934897e-01 -2.11547196e-01 -1.12011373e+00 -5.46616137e-01 -4.98971343e-01 -5.38186014e-01 1.46430099e+00 -1.37947130e+00 -1.47803581e+00 3.66908431e-01 2.24211171e-01 9.08781052e-01 -4.56938118e-01 1.29556513e+00 2.54794300e-01 -6.15281284e-01 1.18003833e+00 2.33192414e-01 2.86649019e-01 1.25293016e+00 -1.34137619e+00 8.62995505e-01 1.03658390e+00 2.34982669e-01 1.15301132e+00 6.50529504e-01 -8.26994956e-01 -1.61304331e+00 -1.28632104e+00 1.73088825e+00 -1.03514016e+00 6.78462148e-01 -9.37407792e-01 -7.69174099e-01 7.31823325e-01 7.45713472e-01 -5.09197898e-02 1.10386884e+00 6.29615784e-01 -4.56553131e-01 2.10096031e-01 -9.21769321e-01 9.31195796e-01 1.12576008e+00 -1.16372275e+00 -1.05551648e+00 1.05997115e-01 1.32795036e+00 -3.37129593e-01 -1.00781631e+00 1.02098770e-01 5.71492016e-01 -3.72645527e-01 6.14459991e-01 -6.89685822e-01 7.98905790e-01 7.47349337e-02 -6.67466164e-01 -1.55542469e+00 -2.10252777e-01 -8.59122336e-01 -2.26640299e-01 1.43613851e+00 7.05236375e-01 -3.96725982e-01 3.89703780e-01 6.29011333e-01 -3.50590885e-01 -9.79682863e-01 -1.16678262e+00 -8.13535869e-01 5.68120420e-01 -4.00633842e-01 8.74989808e-01 1.36646259e+00 1.10971086e-01 9.34943557e-01 -2.63099492e-01 -2.18223959e-01 3.24369292e-04 -1.04606241e-01 6.69347703e-01 -8.64356697e-01 -3.76257211e-01 -3.47187191e-01 -1.82405755e-01 -6.34276450e-01 3.54740828e-01 -1.33856761e+00 -2.27151774e-02 -1.52118027e+00 2.27441356e-01 -4.44767326e-01 -1.84991673e-01 6.58112764e-01 -5.26108921e-01 1.54485181e-01 7.92121589e-02 1.75819650e-01 -2.48673722e-01 6.31902874e-01 8.88109505e-01 -1.15396857e-01 -1.10950284e-01 -5.14400959e-01 -7.30555773e-01 -7.68731087e-02 8.53057861e-01 -3.94184887e-01 -3.10939461e-01 -9.79777992e-01 1.46861300e-01 -2.52738018e-02 -2.59789377e-02 -4.34773326e-01 -3.50020751e-02 -3.03574979e-01 1.56195700e-01 -7.64787197e-02 1.75241560e-01 -5.50916135e-01 -2.18032762e-01 -2.43738130e-01 -5.91883063e-01 4.70237166e-01 2.97111988e-01 3.67257625e-01 -1.22624457e-01 -1.56750828e-01 1.22910820e-01 -1.06967978e-01 -4.86086935e-01 -6.40386492e-02 -3.87661606e-01 3.74018908e-01 4.20609981e-01 -1.28477395e-01 -4.38454479e-01 -1.23674378e-01 -4.43693399e-01 2.62247827e-02 6.14292979e-01 8.60601783e-01 9.84772518e-02 -1.63259506e+00 -9.07706976e-01 2.42571488e-01 4.66087818e-01 -3.92778069e-01 -3.09549719e-01 5.84303141e-01 1.52046889e-01 5.33524811e-01 9.36071798e-02 -5.62599778e-01 -1.28515327e+00 3.55756640e-01 5.45772212e-03 -2.49864459e-01 -4.33448911e-01 5.92986107e-01 -2.86490142e-01 -8.64736557e-01 6.56155348e-02 -3.60802442e-01 3.23767811e-01 1.46848008e-01 5.01707852e-01 1.65633515e-01 3.82345140e-01 -7.04632044e-01 -2.03057021e-01 2.28939742e-01 -1.37668312e-01 -7.13798344e-01 1.07825136e+00 -4.31329817e-01 -1.12710319e-01 4.91682798e-01 1.33430386e+00 2.37867951e-01 -1.13334608e+00 -5.08122861e-01 2.91107595e-01 -4.19043720e-01 1.78417731e-02 -1.52264261e+00 -2.71165609e-01 1.06770122e+00 3.89459819e-01 -3.80706340e-01 7.34853089e-01 -2.32993022e-01 8.89013767e-01 4.81594205e-01 4.57158595e-01 -1.34713531e+00 -2.11468533e-01 1.12092817e+00 8.39455485e-01 -1.40630257e+00 -3.82843614e-01 7.03018438e-03 -7.23374069e-01 8.95851493e-01 4.27475393e-01 4.80796039e-01 -5.32304810e-04 1.46579847e-01 4.92528468e-01 3.17428440e-01 -1.03925145e+00 5.21662235e-02 3.18195999e-01 5.97322643e-01 1.07372820e+00 4.24856633e-01 -5.31120598e-01 7.33386755e-01 -4.46418405e-01 -1.58407882e-01 3.69046867e-01 9.70623255e-01 -9.15795937e-02 -1.74636674e+00 -2.03805372e-01 1.55029610e-01 -6.69360995e-01 -6.81046069e-01 -5.68899214e-01 7.43109643e-01 1.92420334e-01 1.13539791e+00 1.80179417e-01 -5.18803179e-01 2.29298264e-01 6.78156376e-01 3.34174782e-01 -1.07773244e+00 -1.10972250e+00 2.88240891e-02 5.13187587e-01 -4.20355171e-01 -8.17667469e-02 -6.66397095e-01 -7.99002826e-01 -2.95970529e-01 -2.51819223e-01 2.91710913e-01 9.18982446e-01 8.18472803e-01 6.93726599e-01 2.62653977e-01 4.72793698e-01 -5.75358629e-01 -5.89080453e-01 -1.26244080e+00 1.99289262e-01 3.58844250e-01 2.70637095e-01 -4.07418430e-01 -1.34904847e-01 5.48387878e-02]
[11.602646827697754, 10.275280952453613]
b39ebd4c-2cc8-43c4-81a0-e39f58be3163
hybrid-quantum-classical-generative
2212.11614
null
https://arxiv.org/abs/2212.11614v2
https://arxiv.org/pdf/2212.11614v2.pdf
Hybrid Quantum-Classical Generative Adversarial Network for High Resolution Image Generation
Quantum machine learning (QML) has received increasing attention due to its potential to outperform classical machine learning methods in problems pertaining classification and identification tasks. A subclass of QML methods is quantum generative adversarial networks (QGANs) which have been studied as a quantum counterpart of classical GANs widely used in image manipulation and generation tasks. The existing work on QGANs is still limited to small-scale proof-of-concept examples based on images with significant downscaling. Here we integrate classical and quantum techniques to propose a new hybrid quantum-classical GAN framework. We demonstrate its superior learning capabilities by generating $28 \times 28$ pixels grey-scale images without dimensionality reduction or classical pre/post-processing on multiple classes of the standard MNIST and Fashion MNIST datasets, which achieves comparable results to classical frameworks with three orders of magnitude less trainable generator parameters. To gain further insight into the working of our hybrid approach, we systematically explore the impact of its parameter space by varying the number of qubits, the size of image patches, the number of layers in the generator, the shape of the patches and the choice of prior distribution. Our results show that increasing the quantum generator size generally improves the learning capability of the network. The developed framework provides a foundation for future design of QGANs with optimal parameter set tailored for complex image generation tasks.
['Muhammad Usman', 'Sarah M. Erfani', 'Maxwell T. West', 'Shu Lok Tsang']
2022-12-22
null
null
null
null
['image-manipulation']
['computer-vision']
[ 6.98263824e-01 3.43337834e-01 1.88751131e-01 9.28875208e-02 -1.10033882e+00 -7.02739656e-01 9.88494992e-01 -2.64963567e-01 -4.56697792e-01 9.09271777e-01 -3.69570643e-01 -3.25664371e-01 7.77795017e-02 -1.18234229e+00 -7.42474556e-01 -1.24119413e+00 2.27063254e-01 3.97525162e-01 -3.97610925e-02 -5.89608490e-01 1.97687000e-01 2.97517717e-01 -1.33444953e+00 6.20514601e-02 7.39922464e-01 8.92508209e-01 -1.37664631e-01 7.56484568e-01 3.94609869e-01 4.81366694e-01 -7.97784150e-01 -6.74440622e-01 7.80787528e-01 -1.14955819e+00 -6.35355234e-01 -7.61808455e-02 2.66913205e-01 -6.83107451e-02 -5.05507171e-01 1.32629573e+00 8.87620807e-01 6.12435862e-02 6.31420255e-01 -1.08409131e+00 -8.02802920e-01 5.38052917e-01 4.70087789e-02 8.67107585e-02 -1.53564746e-02 6.71939313e-01 9.28065419e-01 -9.22343433e-02 6.65876210e-01 1.03965211e+00 3.29268545e-01 8.09283197e-01 -1.53795588e+00 -7.10463583e-01 -8.92364502e-01 3.28008264e-01 -1.37198389e+00 -2.82794029e-01 6.21105194e-01 -2.68162817e-01 7.76588917e-01 3.20323855e-02 4.47381675e-01 1.15793073e+00 2.86952704e-01 7.64499754e-02 1.66750956e+00 -7.77539551e-01 5.01015306e-01 2.49390379e-01 -6.82883263e-01 6.72156334e-01 4.23068665e-02 5.96874297e-01 -3.67295712e-01 -8.54557455e-02 9.87424314e-01 -4.51819360e-01 1.34339139e-01 -2.65102655e-01 -1.12122798e+00 1.14400685e+00 6.34831190e-01 2.07160562e-01 -3.14996034e-01 6.80682778e-01 1.61750630e-01 5.61146319e-01 1.10721253e-01 7.18902171e-01 8.91274065e-02 -1.95398647e-03 -6.25866950e-01 2.97375798e-01 5.90381086e-01 9.21327114e-01 1.12553132e+00 3.14828306e-01 -3.87266994e-01 2.47765064e-01 -2.72474259e-01 6.56448901e-01 4.09551054e-01 -9.71548498e-01 1.73854932e-01 2.96800941e-01 -1.80567726e-02 -4.40008014e-01 -5.79115450e-02 -2.61180013e-01 -8.96903932e-01 1.99962899e-01 4.16935623e-01 -3.65358412e-01 -8.89041722e-01 1.73046529e+00 1.13958173e-01 6.84162229e-02 1.87896222e-01 6.26598656e-01 4.88598168e-01 5.46037614e-01 -1.29643440e-01 1.16245359e-01 1.42100275e+00 -5.75580657e-01 -3.58751833e-01 2.45708544e-02 5.20393550e-01 -7.12483346e-01 8.80718946e-01 3.03710997e-01 -7.99067497e-01 -5.21337688e-01 -1.18756270e+00 2.88068689e-02 -5.36068439e-01 4.52015772e-02 7.67538786e-01 1.31439662e+00 -9.85621214e-01 8.56989264e-01 -5.22876143e-01 -1.85809731e-01 5.39209306e-01 6.97264791e-01 -3.84797364e-01 3.32706171e-04 -1.45776033e+00 7.60044098e-01 7.06391752e-01 -1.30158409e-01 -1.14228153e+00 -2.55079955e-01 -5.95409214e-01 -9.10149217e-02 1.39339894e-01 -1.14020598e+00 9.30053174e-01 -9.38131511e-01 -2.08873796e+00 9.69187021e-01 4.66118306e-01 -8.30295861e-01 3.93691123e-01 4.50119287e-01 -1.95769072e-01 2.61543095e-01 -1.13833882e-01 9.73613083e-01 1.33365381e+00 -6.65069759e-01 -1.95105866e-01 -2.98972070e-01 2.22722545e-01 -1.48264393e-01 -2.90738828e-02 -2.42077515e-01 1.35985821e-01 -5.32357335e-01 -2.47609273e-01 -1.38175368e+00 -4.55917627e-01 -6.28998935e-01 -4.94767785e-01 6.87637776e-02 2.99238473e-01 -1.45730212e-01 4.99805599e-01 -2.07840776e+00 3.56278569e-01 1.36264652e-01 -1.58456981e-01 2.88433969e-01 -3.08118403e-01 7.08422482e-01 -7.72915781e-03 4.70134579e-02 -2.63478011e-01 -7.25914165e-02 2.01211289e-01 2.01105356e-01 -2.16506481e-01 5.39804161e-01 3.86516005e-01 1.33251095e+00 -7.46469915e-01 -7.13834316e-02 2.16221556e-01 5.08115530e-01 -4.50752825e-01 1.29275188e-01 -2.78408647e-01 9.59309399e-01 -3.48533154e-01 1.85981959e-01 4.60491091e-01 -1.92081273e-01 7.00447783e-02 -9.99379009e-02 1.59850210e-01 2.80481368e-01 -9.68675852e-01 1.69848382e+00 -3.47214878e-01 3.78183186e-01 -3.43632936e-01 -1.13240397e+00 8.67160738e-01 2.66306281e-01 1.31745294e-01 -1.02070498e+00 3.20407838e-01 1.99110895e-01 2.95779496e-01 -1.67971611e-01 2.19794482e-01 -7.14651823e-01 -3.10735047e-01 4.95487690e-01 5.06635487e-01 -7.59480417e-01 2.55364835e-01 4.79488559e-02 1.14606833e+00 1.02153413e-01 2.75817662e-01 -1.12495162e-01 5.11916220e-01 -1.69090033e-01 1.03057571e-01 9.67484057e-01 2.12103948e-02 5.74062586e-01 7.15649784e-01 -2.20882282e-01 -1.45278418e+00 -8.85025561e-01 -5.72237326e-03 8.13420653e-01 4.49656434e-02 -3.92403811e-01 -1.15290487e+00 -4.69997555e-01 -4.11392868e-01 5.87428033e-01 -6.35862827e-01 -4.27723616e-01 -4.63042766e-01 -1.28390217e+00 8.93568158e-01 9.78111327e-02 7.00561106e-01 -1.32752562e+00 -5.01031101e-01 2.27887705e-02 2.88621694e-01 -1.36757660e+00 1.18960485e-01 2.48046935e-01 -7.34251738e-01 -9.41214442e-01 -5.15429378e-01 -4.10739422e-01 5.99560678e-01 -3.33060771e-01 9.50712800e-01 -4.44548994e-01 -3.17478865e-01 2.88873047e-01 -4.32143062e-01 -2.98596442e-01 -1.10358751e+00 3.70499432e-01 -1.83071285e-01 -1.98633764e-02 4.48738784e-02 -6.65399790e-01 -7.40098178e-01 1.85012948e-02 -1.29052782e+00 8.34902003e-02 9.59392548e-01 1.18026388e+00 5.08834600e-01 2.81038553e-01 5.21379411e-01 -1.08378768e+00 4.72033381e-01 -1.13555819e-01 -9.61365402e-01 -2.73409225e-02 -5.89255869e-01 4.71000165e-01 9.23247695e-01 -1.52302235e-01 -6.81039393e-01 -4.44441885e-02 -2.64329135e-01 9.73169804e-02 -2.52830684e-01 1.50532043e-02 -1.03651807e-01 -6.50701880e-01 8.73255849e-01 3.71848285e-01 -8.32844339e-03 2.01748647e-02 7.50580966e-01 3.45596373e-01 3.95357162e-01 -4.32675958e-01 1.16733873e+00 6.11212432e-01 5.64099133e-01 -7.97211468e-01 -4.35048789e-01 1.34551376e-01 -5.08987010e-01 2.77489156e-01 1.01341593e+00 -7.10479379e-01 -7.16660917e-01 5.47643840e-01 -7.89104164e-01 -4.61970776e-01 -6.28043950e-01 3.35664392e-01 -9.54033852e-01 5.19540682e-02 -6.88175380e-01 -6.27636969e-01 -5.08217573e-01 -1.41193581e+00 1.08168089e+00 2.37306356e-01 3.44041526e-01 -8.86663973e-01 1.67197380e-02 4.66950148e-01 6.44212246e-01 4.31223571e-01 9.68202233e-01 -3.33045125e-01 -1.06436133e+00 -1.63616776e-01 2.99905590e-03 6.67445362e-01 -1.13800213e-01 -6.07758880e-01 -1.03203857e+00 -4.82600093e-01 -6.89902622e-03 -4.16627586e-01 9.03867483e-01 8.81921966e-03 8.06147933e-01 -2.58673638e-01 1.89347520e-01 7.76375592e-01 1.71684241e+00 2.33276621e-01 1.07105386e+00 2.80132174e-01 6.76738560e-01 2.99189568e-01 4.78490107e-02 3.03953171e-01 -3.83833051e-02 7.83648610e-01 5.91678500e-01 8.56481865e-03 -1.21141501e-01 -2.07173720e-01 5.60941875e-01 4.95366454e-01 -2.40314066e-01 -7.60136768e-02 -5.38032234e-01 6.05781227e-02 -1.35307741e+00 -1.01528621e+00 3.38468283e-01 2.37275076e+00 7.32603729e-01 3.44515592e-02 1.36993453e-01 1.55190095e-01 5.73403239e-01 -4.04563081e-03 -4.69406962e-01 -4.07516479e-01 -3.06417853e-01 1.11876273e+00 9.04901385e-01 1.42953888e-01 -1.10494018e+00 1.04243338e+00 5.98653030e+00 1.15741539e+00 -1.22416830e+00 4.34251994e-01 5.34278095e-01 1.88304096e-01 -2.28277773e-01 3.40185672e-01 -6.43466771e-01 3.44242185e-01 1.21593797e+00 8.78153443e-02 8.83751750e-01 5.01513183e-01 -1.82829008e-01 -3.59426672e-03 -1.10171425e+00 1.10450590e+00 9.74904373e-03 -1.42400277e+00 2.70815101e-02 3.12405795e-01 1.06257021e+00 2.45456383e-01 4.13359165e-01 2.24203333e-01 1.63719788e-01 -1.37572038e+00 5.18280387e-01 2.10445985e-01 1.05034912e+00 -7.35819042e-01 8.43843579e-01 2.15459123e-01 -4.72142965e-01 -2.22470924e-01 -3.30738127e-01 -1.38523206e-01 2.29038458e-04 2.30579868e-01 -8.13839138e-01 6.66743279e-01 3.17767233e-01 1.88490480e-01 -8.45083058e-01 4.07607675e-01 -5.15937626e-01 5.55573285e-01 -2.95131564e-01 -4.25996520e-02 4.87992346e-01 -6.22994304e-01 3.89338642e-01 7.40242302e-01 3.89141947e-01 -6.36507198e-02 -2.51945883e-01 1.22878695e+00 -1.69138864e-01 -1.33325681e-01 -7.11722314e-01 -3.68127227e-01 1.59031495e-01 1.27710176e+00 -7.67066121e-01 -1.88117579e-01 -6.93581030e-02 1.08265340e+00 1.03640147e-02 2.40974039e-01 -7.30088174e-01 -3.93043905e-01 9.42934006e-02 9.44644213e-02 5.71491659e-01 -2.39137381e-01 6.05016015e-03 -1.07103217e+00 -2.36088663e-01 -1.04403329e+00 -4.30999883e-03 -3.72602373e-01 -1.22540796e+00 5.64982712e-01 -6.51491657e-02 -1.02749622e+00 -4.97165769e-01 -6.85946345e-01 -4.30765539e-01 8.55680048e-01 -1.25732052e+00 -1.30634880e+00 -4.53835465e-02 4.31905746e-01 -2.46977895e-01 -4.53779161e-01 1.16512752e+00 7.96896219e-02 -5.01633584e-01 6.80505395e-01 4.38876152e-01 1.70935124e-01 4.04885411e-01 -1.25734472e+00 6.01817489e-01 8.86073291e-01 3.67495894e-01 4.91882503e-01 7.55582631e-01 -9.98163149e-02 -1.64751601e+00 -9.16377962e-01 1.29349843e-01 -4.11769599e-01 5.81206203e-01 -7.07229257e-01 -2.68888652e-01 4.64560211e-01 4.42062944e-01 2.44100764e-02 6.31029546e-01 -4.89723265e-01 -2.11671680e-01 -2.29441568e-01 -1.20041597e+00 3.86483908e-01 7.29476035e-01 -8.87184739e-01 -3.73344980e-02 4.42577064e-01 4.15337414e-01 -1.75461560e-01 -1.00441313e+00 2.90393740e-01 3.17292452e-01 -1.29945302e+00 8.14725637e-01 -2.55741090e-01 2.65770346e-01 -2.47121602e-01 -7.38770217e-02 -1.40531111e+00 -7.55291954e-02 -1.13346946e+00 1.87688947e-01 8.63949060e-01 1.30587861e-01 -8.94293487e-01 9.59109664e-01 1.07589923e-01 1.35276258e-01 -5.82879722e-01 -1.19696975e+00 -6.72270596e-01 3.50403488e-01 -1.57786846e-01 5.88237643e-01 4.61625278e-01 -4.20841902e-01 6.36185706e-01 -3.75164777e-01 -1.16496034e-01 6.16321146e-01 6.26212880e-02 1.07933855e+00 -7.11991966e-01 -7.54710555e-01 -3.91255021e-01 -8.38110268e-01 -5.82211852e-01 4.93590087e-02 -8.73771966e-01 -2.32644856e-01 -8.05273473e-01 6.79101348e-02 -4.73113954e-01 -1.71219826e-01 1.00845464e-01 8.37547064e-04 8.55734825e-01 3.02425683e-01 -4.93095405e-02 -3.41700524e-01 6.42276227e-01 1.27300715e+00 -1.48576856e-01 -1.52128807e-03 -1.26348376e-01 -4.42617089e-01 1.23014338e-01 7.92739689e-01 -5.12897909e-01 -4.19154912e-01 -3.07606049e-02 5.78427851e-01 -7.99710304e-02 6.55782521e-01 -1.19533050e+00 -9.50434878e-02 1.42463967e-01 1.38833717e-01 1.40101299e-01 4.11050081e-01 -2.82667458e-01 4.90907669e-01 4.91020262e-01 -1.11404009e-01 -6.68744668e-02 -1.10843047e-01 4.65467244e-01 -1.51709735e-01 -3.44900548e-01 1.04043603e+00 -3.84383947e-01 -4.48847532e-01 2.76268572e-01 -2.07940713e-01 6.07438944e-03 1.08579063e+00 -1.92362834e-02 -2.74013221e-01 -2.95129031e-01 -4.97973144e-01 -4.57765698e-01 6.77606940e-01 1.44649178e-01 2.05937326e-01 -1.21879327e+00 -6.25501752e-01 4.68970537e-01 1.37839392e-01 -1.48277670e-01 2.67409891e-01 7.13588238e-01 -7.68143475e-01 6.53642714e-01 -5.01421273e-01 -4.04361099e-01 -6.68393552e-01 6.11461103e-01 4.92064863e-01 -3.60174805e-01 -3.00058573e-01 7.04031527e-01 1.46342278e-01 -2.86650658e-01 -4.02850896e-01 8.56209248e-02 1.27076253e-01 -2.31403559e-01 1.79195598e-01 3.38051289e-01 2.51583338e-01 -7.50036657e-01 -2.53708195e-02 3.88122439e-01 2.17686787e-01 -2.98329830e-01 1.10809612e+00 2.63701826e-01 -2.48424426e-01 1.89938337e-01 1.02118850e+00 -2.77997553e-01 -1.11431456e+00 -1.36822471e-02 -2.60738224e-01 -3.62970890e-03 -7.29907230e-02 -4.27804857e-01 -9.19887960e-01 9.93799388e-01 9.52598751e-01 3.19457829e-01 1.15210676e+00 5.99998906e-02 8.19120109e-01 3.77231896e-01 7.79779792e-01 -9.86177087e-01 1.33779913e-01 1.93830311e-01 4.14463788e-01 -1.23060989e+00 -7.35593354e-03 -1.20483331e-01 -3.46183956e-01 9.45919931e-01 1.45737112e-01 -4.72655684e-01 1.69439852e-01 -1.69809163e-01 -1.06955945e-01 -2.03683406e-01 -3.83389443e-01 -3.80747318e-01 1.29347160e-01 5.83375454e-01 1.74001113e-01 3.26504380e-01 -2.26018921e-01 7.52001554e-02 -5.95931411e-01 -2.42236838e-01 5.34587920e-01 7.09401906e-01 1.46267653e-01 -1.70646656e+00 -2.75752455e-01 2.27391228e-01 -7.73389876e-01 -2.42015675e-01 -2.00068280e-01 6.25744343e-01 5.01879156e-01 6.23221338e-01 -1.61711141e-01 -3.13701004e-01 -9.60609410e-03 1.29108146e-01 1.15197086e+00 -6.02376819e-01 -7.34560609e-01 -2.22388998e-01 -2.83506125e-01 -3.51990461e-01 -6.04585171e-01 -5.11672616e-01 -9.01905715e-01 -3.40306878e-01 -4.97500896e-01 4.54540849e-02 7.37197936e-01 1.04953432e+00 4.97102737e-01 4.01998580e-01 5.69198132e-01 -8.72525394e-01 -9.13032353e-01 -1.07989967e+00 -5.44660270e-01 5.90209305e-01 1.26983086e-02 -3.93064648e-01 -1.42601550e-01 -2.44669244e-03]
[5.626674175262451, 4.96585750579834]
f6bd75a6-c654-45e3-9282-34a2fbb44766
sn-computer-science-towards-offensive
2108.10939
null
https://arxiv.org/abs/2108.10939v2
https://arxiv.org/pdf/2108.10939v2.pdf
Towards Offensive Language Identification for Tamil Code-Mixed YouTube Comments and Posts
Offensive Language detection in social media platforms has been an active field of research over the past years. In non-native English spoken countries, social media users mostly use a code-mixed form of text in their posts/comments. This poses several challenges in the offensive content identification tasks, and considering the low resources available for Tamil, the task becomes much harder. The current study presents extensive experiments using multiple deep learning, and transfer learning models to detect offensive content on YouTube. We propose a novel and flexible approach of selective translation and transliteration techniques to reap better results from fine-tuning and ensembling multilingual transformer networks like BERT, Distil- BERT, and XLM-RoBERTa. The experimental results showed that ULMFiT is the best model for this task. The best performing models were ULMFiT and mBERTBiLSTM for this Tamil code-mix dataset instead of more popular transfer learning models such as Distil- BERT and XLM-RoBERTa and hybrid deep learning models. The proposed model ULMFiT and mBERTBiLSTM yielded good results and are promising for effective offensive speech identification in low-resourced languages.
['Uthayasanker Thayasivam', 'Charangan Vasantharajan']
2021-08-24
null
null
null
null
['transliteration']
['natural-language-processing']
[-4.24527407e-01 -2.59528667e-01 -4.17752624e-01 9.74904895e-02 -1.18080485e+00 -6.35950744e-01 6.87301874e-01 2.85423417e-02 -6.13151371e-01 4.35531288e-01 3.65374982e-01 -5.18654764e-01 2.90105969e-01 -4.10790682e-01 -3.82625669e-01 -3.14225048e-01 8.56236443e-02 3.79660368e-01 -1.07429080e-01 -7.15859890e-01 4.18968320e-01 1.68068781e-02 -8.79836679e-01 7.15316236e-01 1.22503126e+00 8.14438820e-01 1.81850493e-01 6.43909872e-01 -1.93226501e-01 1.22320616e+00 -6.32461727e-01 -9.62857604e-01 1.15676813e-01 -3.06612402e-01 -8.67017448e-01 -6.01727426e-01 4.90406632e-01 -4.29111242e-01 -4.27436262e-01 1.41846812e+00 8.61494303e-01 -2.67757654e-01 5.55603981e-01 -1.08643532e+00 -7.52954662e-01 1.25847566e+00 -7.54585445e-01 5.23074508e-01 4.14258689e-01 2.18622922e-03 9.26223755e-01 -8.27005208e-01 2.72737145e-01 1.45449650e+00 8.71462166e-01 3.45681906e-01 -5.65929651e-01 -1.31904614e+00 -1.11802667e-01 2.84730256e-01 -1.30439413e+00 -4.04016048e-01 7.77619541e-01 -7.28760839e-01 9.55766320e-01 1.82782546e-01 3.33104819e-01 1.90191686e+00 2.13960677e-01 1.17456710e+00 1.14962316e+00 -3.33513528e-01 -4.66630697e-01 5.37943065e-01 2.77428385e-02 6.88329339e-01 -1.92892969e-01 -2.36428365e-01 -5.77531755e-01 -4.21325803e-01 1.03228614e-01 -1.41123027e-01 -2.01739576e-02 4.71545070e-01 -1.19481921e+00 1.38034189e+00 9.93872285e-02 6.87601805e-01 4.89418767e-02 -2.25655943e-01 1.11820018e+00 6.99387252e-01 9.79573846e-01 7.31055379e-01 -3.10258508e-01 -7.45754778e-01 -9.62905228e-01 -1.23656206e-01 7.50575840e-01 8.09051752e-01 3.04812908e-01 1.61685750e-01 1.09398335e-01 1.43376422e+00 7.02175498e-02 6.64927006e-01 1.16695344e+00 -3.39818567e-01 1.09344447e+00 3.75545651e-01 -3.44105750e-01 -1.14288914e+00 -2.55247295e-01 -5.90597868e-01 -5.58308065e-01 -1.76328540e-01 1.52449146e-01 -5.83458245e-01 -4.88868535e-01 1.31843722e+00 -2.15585619e-01 -1.80759504e-01 -1.45361304e-01 5.19580603e-01 7.60378480e-01 8.82674158e-01 1.93775725e-02 7.53494278e-02 1.03863502e+00 -1.20211101e+00 -6.15155160e-01 -1.55758485e-01 1.06427550e+00 -1.20216537e+00 1.43296468e+00 5.94198048e-01 -7.96431363e-01 -5.88024735e-01 -9.79410112e-01 -7.05238655e-02 -5.24830699e-01 2.76584178e-01 2.81170368e-01 1.02269137e+00 -8.21782410e-01 3.95265281e-01 -4.60820824e-01 -4.39824373e-01 3.83674294e-01 1.20959185e-01 -3.12902153e-01 3.68221313e-01 -1.41571951e+00 9.93026972e-01 3.24826449e-01 -2.10689023e-01 -1.05672610e+00 -3.86572808e-01 -5.67864180e-01 -1.91322327e-01 1.56703115e-01 3.16689819e-01 1.35104954e+00 -1.33376491e+00 -1.59722126e+00 1.13320577e+00 5.94915926e-01 -5.60403883e-01 7.99149215e-01 -6.84067428e-01 -7.74266601e-01 1.97672062e-02 2.66521096e-01 9.84521210e-02 9.84639764e-01 -6.09069049e-01 -6.07208908e-01 -9.72157940e-02 5.18662743e-02 1.42200857e-01 -1.05518126e+00 8.72659266e-01 1.31630674e-01 -7.85517335e-01 -5.25650084e-01 -8.54985058e-01 4.98891443e-01 -6.35140836e-01 -7.44086266e-01 -2.45726317e-01 1.08228064e+00 -1.21312118e+00 1.61277974e+00 -2.28732944e+00 3.35837044e-02 7.24173384e-04 3.17916781e-01 8.35524976e-01 -1.12920046e-01 7.50478983e-01 -5.22188805e-02 4.28890616e-01 2.51782417e-01 -3.06170732e-01 -3.43214646e-02 -4.30956483e-01 -5.32696962e-01 6.17498577e-01 -2.47714624e-01 6.29599452e-01 -9.95231450e-01 -5.25388598e-01 8.01557675e-02 2.16203973e-01 -5.79964399e-01 3.13636720e-01 1.92051500e-01 1.61615014e-01 -3.90151024e-01 7.47045100e-01 4.50800925e-01 3.18694413e-01 -1.83705449e-01 2.50764728e-01 -4.52532679e-01 5.52699506e-01 -3.73995811e-01 1.29824495e+00 -9.04839396e-01 1.08959758e+00 2.63768673e-01 -7.61186302e-01 9.20672953e-01 2.11323380e-01 2.69505531e-01 -8.94316316e-01 6.24471605e-01 6.65364087e-01 3.83284688e-01 -8.29534352e-01 4.45614129e-01 -1.06430747e-01 -4.47909623e-01 4.40615624e-01 6.46395087e-02 6.52162954e-02 8.87390226e-03 2.68354446e-01 8.68526042e-01 -2.66615719e-01 2.61385411e-01 -3.20973992e-01 7.10097134e-01 -3.07743460e-01 1.41087085e-01 5.46664119e-01 -5.71676433e-01 3.82395059e-01 5.43152750e-01 -1.86228320e-01 -8.94824505e-01 -5.68934679e-01 -5.53883389e-02 1.79558027e+00 -2.50177354e-01 -6.03933752e-01 -9.30036664e-01 -8.39636683e-01 -1.55292273e-01 7.21370399e-01 -4.64467645e-01 -3.38673353e-01 -5.19924700e-01 -6.37923300e-01 1.30479634e+00 1.35511920e-01 8.21162462e-01 -1.10962355e+00 -3.58011462e-02 1.41959995e-01 -7.28180289e-01 -9.57573533e-01 -8.31743538e-01 7.38707557e-02 -2.38519177e-01 -7.78373420e-01 -8.29150379e-01 -8.23435307e-01 -5.30036390e-02 1.26210734e-01 6.59978807e-01 -1.22802347e-01 -5.79358172e-03 -3.01523805e-01 -7.44957149e-01 -5.59527934e-01 -9.70069706e-01 6.04843676e-01 1.27019301e-01 2.78315842e-01 4.74853486e-01 -1.83405101e-01 -1.71432327e-02 2.28066355e-01 -4.85737205e-01 -1.20684087e-01 4.87971902e-01 8.73624682e-01 -4.22993839e-01 -3.10613364e-01 5.89849949e-01 -9.68987048e-01 1.08575928e+00 -8.20079923e-01 -1.59671172e-01 -4.83502187e-02 -1.28511176e-01 -3.01898897e-01 1.00719857e+00 -8.16027582e-01 -7.92503536e-01 -5.41908503e-01 -7.21915662e-01 -4.41474468e-01 1.51329800e-01 5.56825399e-01 3.02693900e-02 -1.07880183e-01 9.14805591e-01 1.08515665e-01 1.35915093e-02 -7.40127504e-01 -7.80763254e-02 1.54390299e+00 5.96479997e-02 -2.76155382e-01 7.07065701e-01 -2.47392617e-02 -1.17509985e+00 -1.12395298e+00 -8.51589561e-01 -7.50534594e-01 -4.87479717e-01 -1.50534913e-01 9.51423824e-01 -1.08514953e+00 -6.25362694e-01 1.00943387e+00 -1.10669506e+00 -3.13297451e-01 4.75748867e-01 5.28710425e-01 -2.02643707e-01 6.58418298e-01 -9.97381628e-01 -7.18297005e-01 -6.86286747e-01 -1.33541286e+00 8.88712406e-01 -4.97811735e-01 -5.33358634e-01 -9.93651986e-01 2.74643332e-01 7.35312402e-01 5.76404870e-01 1.64197851e-02 9.75775301e-01 -1.16047287e+00 3.80561352e-01 -3.74021977e-01 -3.01139355e-01 7.61926353e-01 6.00361191e-02 2.56422430e-01 -9.77031231e-01 -4.87807274e-01 -3.93798091e-02 -7.83595979e-01 8.34095716e-01 -1.16323769e-01 1.06584811e+00 -6.52903557e-01 -1.43685892e-01 5.55788040e-01 8.75976562e-01 1.32758334e-01 3.84713739e-01 6.56257689e-01 9.44046378e-01 2.80445814e-01 4.74102497e-01 5.42289078e-01 2.61962920e-01 6.36712313e-01 4.79164630e-01 -1.43461689e-01 -1.08157452e-02 -4.36363250e-01 1.08480680e+00 1.44873738e+00 4.49038344e-05 -7.65086487e-02 -1.15704381e+00 4.35474992e-01 -1.51522708e+00 -1.22255361e+00 -1.08444631e-01 2.02952743e+00 9.45990562e-01 1.92270160e-01 6.32489026e-01 1.48604482e-01 8.60098720e-01 2.15341046e-01 -1.63472518e-01 -9.48023856e-01 -1.62047192e-01 -8.32753163e-03 4.57962424e-01 4.86434281e-01 -1.49589705e+00 1.08433437e+00 5.56330299e+00 1.27862835e+00 -1.58859515e+00 7.08060980e-01 5.90407431e-01 -2.05462754e-01 1.19569466e-01 -6.16390049e-01 -9.52593505e-01 9.71932173e-01 1.33359754e+00 -9.20746773e-02 4.38620239e-01 1.09495008e+00 1.03185125e-01 5.03055155e-01 -7.20950782e-01 1.20909619e+00 4.35404718e-01 -8.80951941e-01 -1.26903236e-01 1.11183092e-01 7.86703289e-01 7.54406333e-01 4.57537413e-01 9.99613285e-01 4.31056589e-01 -9.38996375e-01 1.06223035e+00 -3.43642682e-01 7.71662891e-01 -9.31991816e-01 8.59058917e-01 6.66600704e-01 -5.46454787e-01 -6.52792156e-01 -4.00901228e-01 -2.69661527e-02 -3.20964009e-01 3.60124886e-01 -8.61451864e-01 7.23958537e-02 7.36742020e-01 8.99288535e-01 -6.25248551e-01 7.99257696e-01 -9.73315686e-02 9.92762208e-01 -1.10182948e-01 -4.78667647e-01 6.74551964e-01 1.04150899e-01 5.69150567e-01 1.76889944e+00 3.86141390e-01 -7.36496747e-01 2.99219519e-01 4.30073142e-01 -3.85267019e-01 7.65959561e-01 -6.79558098e-01 -5.28451860e-01 1.55866057e-01 1.35600352e+00 -2.45353505e-01 -2.50753134e-01 -5.19741356e-01 8.75248551e-01 6.47672772e-01 -1.27855778e-01 -1.24563861e+00 -4.76772994e-01 5.39177001e-01 3.40119451e-01 -2.10687786e-01 -1.50896579e-01 9.72443745e-02 -1.44339502e+00 -3.14106077e-01 -1.53709090e+00 4.14724141e-01 -4.06429887e-01 -1.29629505e+00 9.71847773e-01 -2.10033581e-01 -1.53566551e+00 -2.61580527e-01 -7.20720410e-01 -6.07124627e-01 6.96426988e-01 -1.28277469e+00 -1.30579412e+00 4.05198289e-03 6.99145019e-01 9.69591022e-01 -8.56255293e-01 5.77227831e-01 7.40538418e-01 -8.69523108e-01 9.96763587e-01 4.60620344e-01 6.58519924e-01 9.61983204e-01 -9.89005506e-01 2.26879790e-01 8.48111331e-01 -5.62323555e-02 6.11085892e-01 4.56563860e-01 -6.79279506e-01 -1.01654756e+00 -1.03895843e+00 7.53435314e-01 -2.92845279e-01 1.29859209e+00 -6.91958606e-01 -6.01277828e-01 4.89039063e-01 6.79736912e-01 -6.52048230e-01 8.74534667e-01 2.08543852e-01 -6.72796249e-01 7.59774074e-02 -9.34800982e-01 4.69542682e-01 7.03675747e-01 -9.69287455e-01 -3.58255595e-01 5.79483569e-01 4.48108584e-01 -1.78166240e-01 -5.40931821e-01 -7.35082999e-02 4.78707016e-01 -8.94996941e-01 5.03639400e-01 -5.52457511e-01 8.64736855e-01 3.72387439e-01 -1.68868437e-01 -1.57469845e+00 -2.55253106e-01 -8.97914469e-01 1.34393618e-01 1.33320320e+00 6.80521846e-01 -6.25361204e-01 2.86212176e-01 -4.46584038e-02 -2.62643605e-01 -3.43538195e-01 -9.37585771e-01 -6.75796747e-01 5.98107398e-01 -3.27127427e-01 -1.36500346e-02 1.45669270e+00 4.70743328e-01 4.50621724e-01 -7.54935205e-01 -4.23934698e-01 2.56445795e-01 -1.89934835e-01 6.63683712e-01 -8.84434700e-01 -2.93434024e-01 -7.21443057e-01 -3.10353488e-01 -7.92883635e-01 5.37353992e-01 -1.31150019e+00 -3.54286581e-01 -9.29766357e-01 3.22569966e-01 -1.66661844e-01 -5.59709743e-02 4.84189451e-01 5.00391722e-02 4.14553761e-01 2.43481353e-01 2.03194290e-01 -4.40713257e-01 4.31053013e-01 9.71741676e-01 -5.62844396e-01 -4.99832071e-02 1.78820968e-01 -8.13600302e-01 8.09538722e-01 9.24991846e-01 -3.80729854e-01 -2.67750293e-01 -3.92610252e-01 3.37047845e-01 -2.39617214e-01 -1.78882524e-01 -8.13821614e-01 -1.31102473e-01 2.17981577e-01 4.34735082e-02 -3.87416154e-01 1.86033085e-01 -4.65285927e-01 -5.42345524e-01 5.66597939e-01 -5.04295826e-01 2.41741732e-01 1.17829397e-01 -1.36025310e-01 -3.65999043e-01 -2.95885623e-01 1.04724717e+00 -2.38979414e-01 -4.94166911e-01 8.97112042e-02 -8.12495768e-01 3.59991610e-01 8.19923639e-01 -6.63564950e-02 -4.76792157e-01 -5.97521603e-01 -5.15611351e-01 -1.19312540e-01 3.42315920e-02 1.01243186e+00 4.71588522e-01 -1.08917367e+00 -1.13068843e+00 1.40650600e-01 2.31298268e-01 -9.06136036e-01 4.20093052e-02 1.05145335e+00 -7.81360745e-01 3.77286047e-01 -3.88751417e-01 -2.14142010e-01 -1.35081804e+00 4.32492346e-01 5.12636006e-01 -2.67680675e-01 -2.21222132e-01 8.73847663e-01 1.14361942e-01 -8.35822761e-01 1.42616361e-01 -3.16581950e-02 -3.39356571e-01 4.45922464e-01 5.53826451e-01 6.75105512e-01 2.14407980e-01 -1.12703109e+00 -6.26238212e-02 1.66175678e-01 -5.31824529e-01 5.80402724e-02 1.11765337e+00 -3.28581110e-02 -2.70737708e-01 4.73825961e-01 1.76580262e+00 4.71448839e-01 -4.03666258e-01 -8.88197422e-02 -1.70519605e-01 -4.88502324e-01 2.26328939e-01 -8.15315545e-01 -9.79856730e-01 1.27511430e+00 3.88535321e-01 3.67057800e-01 6.59181118e-01 -2.53049195e-01 1.02733266e+00 2.14547247e-01 3.81841451e-01 -1.45666242e+00 3.60710114e-01 9.19406950e-01 1.13125086e+00 -1.28298819e+00 -4.76679236e-01 -1.69808581e-01 -7.16892898e-01 1.07218230e+00 7.82424927e-01 8.93465057e-02 7.22075641e-01 2.85299063e-01 4.16595370e-01 1.34119496e-01 -4.15581733e-01 1.02851212e-01 9.61645618e-02 2.54187733e-01 8.53499889e-01 1.00371383e-01 -2.52676964e-01 5.04477799e-01 -6.35661542e-01 -6.22630477e-01 6.00815594e-01 4.89829242e-01 -4.38822716e-01 -9.35129881e-01 -3.44098866e-01 6.61053181e-01 -1.18247104e+00 -4.39887017e-01 -7.73508549e-01 7.45736480e-01 -3.36097963e-02 1.07372212e+00 -2.87244320e-01 -9.19640958e-01 4.22048904e-02 1.00284663e-03 -8.38602856e-02 -6.93204224e-01 -1.34663177e+00 2.57435352e-01 4.01315004e-01 -1.50686041e-01 7.29331896e-02 -7.16186643e-01 -7.76230633e-01 -6.44359350e-01 -5.10409951e-01 2.30010793e-01 6.41157568e-01 7.93681920e-01 7.48736709e-02 1.85043722e-01 9.71554160e-01 -6.54823959e-01 -8.29230785e-01 -1.53636956e+00 -4.35852945e-01 5.22939384e-01 3.86723697e-01 -5.39816737e-01 -5.25219917e-01 -3.47877830e-01]
[9.004748344421387, 10.618927001953125]
c88f7955-eb02-4010-886a-5e4eec1900d0
improved-chord-recognition-by-combining
1808.05335
null
http://arxiv.org/abs/1808.05335v1
http://arxiv.org/pdf/1808.05335v1.pdf
Improved Chord Recognition by Combining Duration and Harmonic Language Models
Chord recognition systems typically comprise an acoustic model that predicts chords for each audio frame, and a temporal model that casts these predictions into labelled chord segments. However, temporal models have been shown to only smooth predictions, without being able to incorporate musical information about chord progressions. Recent research discovered that it might be the low hierarchical level such models have been applied to (directly on audio frames) which prevents learning musical relationships, even for expressive models such as recurrent neural networks (RNNs). However, if applied on the level of chord sequences, RNNs indeed can become powerful chord predictors. In this paper, we disentangle temporal models into a harmonic language model---to be applied on chord sequences---and a chord duration model that connects the chord-level predictions of the language model to the frame-level predictions of the acoustic model. In our experiments, we explore the impact of each model on the chord recognition score, and show that using harmonic language and duration models improves the results.
['Filip Korzeniowski', 'Gerhard Widmer']
2018-08-16
null
null
null
null
['chord-recognition']
['audio']
[ 3.81616622e-01 3.37049067e-01 -1.37791991e-01 -1.59709886e-01 -7.26137578e-01 -7.88121700e-01 4.31248397e-01 9.20853913e-02 -2.33706459e-01 3.09833109e-01 6.75829113e-01 -3.92042726e-01 -9.17569771e-02 -6.47325039e-01 -4.45109308e-01 -3.68053973e-01 -3.96065474e-01 1.72132850e-01 5.84760725e-01 -4.03808445e-01 2.78248549e-01 2.64618158e-01 -1.52877593e+00 5.16747296e-01 1.06223516e-01 9.34930742e-01 1.15912631e-01 1.12814343e+00 5.28098308e-02 1.47202551e+00 -5.65677583e-01 -5.95927462e-02 2.53987253e-01 -7.37716377e-01 -7.34451294e-01 -5.05392849e-01 1.28964663e-01 -2.20378160e-01 -3.42905223e-01 5.03891170e-01 1.34663939e-01 3.46232295e-01 5.55932760e-01 -8.67854834e-01 -2.52967983e-01 1.21683943e+00 -1.86364159e-01 -5.95017560e-02 3.83590788e-01 -1.31865561e-01 1.44607389e+00 -5.82873821e-01 6.55084431e-01 1.05941069e+00 1.17781305e+00 4.94797319e-01 -1.33945990e+00 -7.18696594e-01 -2.97448542e-02 3.62092704e-01 -1.10875344e+00 -5.36417782e-01 1.06655693e+00 -5.21012783e-01 1.20348406e+00 1.28223643e-01 8.41859579e-01 8.89601707e-01 1.81657270e-01 8.24835241e-01 6.98695600e-01 -4.77209479e-01 -1.82205830e-02 -4.02873009e-01 -8.25503469e-02 2.86377877e-01 -8.89311194e-01 5.74298322e-01 -8.56910527e-01 -1.06929407e-01 9.32608485e-01 -5.01906395e-01 -5.81002571e-02 5.16154291e-03 -9.09975827e-01 7.62174129e-01 2.74588853e-01 3.63393545e-01 -3.40170175e-01 3.53449106e-01 8.28007400e-01 4.42970395e-01 2.99467947e-02 7.22517431e-01 -5.42797923e-01 -5.63666224e-01 -1.40091610e+00 3.83086175e-01 9.55385864e-01 4.35308576e-01 2.97408074e-01 3.34494054e-01 -2.29286045e-01 8.97459149e-01 4.37384583e-02 -2.19119906e-01 6.09764636e-01 -1.33141041e+00 -1.94228411e-01 8.81507248e-02 -9.66148600e-02 -8.52182448e-01 -4.76086348e-01 -4.92579639e-01 -5.26473045e-01 1.47113681e-01 7.55770206e-01 4.84446675e-04 -5.62216938e-01 2.07015705e+00 -1.35633603e-01 5.73618591e-01 -4.25694659e-02 6.98406875e-01 3.54763418e-01 8.45777392e-01 4.18214202e-02 -4.76327419e-01 1.12205684e+00 -8.88247311e-01 -4.82963860e-01 7.82609209e-02 7.45879531e-01 -9.33567286e-01 1.24428809e+00 8.05253267e-01 -1.43989015e+00 -1.01020348e+00 -1.16980553e+00 -2.79026359e-01 1.36543989e-01 -6.92387670e-02 1.86183855e-01 2.14833573e-01 -1.01366818e+00 1.22299898e+00 -8.18794549e-01 -9.08412561e-02 -3.29411000e-01 3.03076476e-01 1.50704399e-01 5.87809324e-01 -1.48386192e+00 8.95175159e-01 4.38943714e-01 -1.83661237e-01 -9.42555666e-01 -7.63248980e-01 -4.89817530e-01 2.47279018e-01 9.87005606e-02 -4.07711655e-01 1.78190958e+00 -1.01538503e+00 -1.71369874e+00 4.40269321e-01 -1.04159251e-01 -7.41482913e-01 1.28801182e-01 -1.62770644e-01 -1.27716944e-01 6.62463382e-02 -3.61918569e-01 8.39056492e-01 5.34631968e-01 -8.87840450e-01 -6.66830003e-01 -9.94859934e-02 1.60876483e-01 1.27869412e-01 -1.42888114e-01 -1.17710363e-02 -1.63397580e-01 -8.52477849e-01 9.78187397e-02 -1.20615757e+00 -2.70800531e-01 -3.51949215e-01 -2.80075252e-01 -4.78332430e-01 5.13626635e-01 -9.20114160e-01 1.66491795e+00 -2.36241341e+00 2.37953037e-01 1.54063582e-01 -3.06362659e-01 -2.21666954e-02 -1.12065390e-01 5.22046924e-01 -2.96098650e-01 -1.13254517e-01 7.44016469e-02 -3.94823194e-01 2.91521400e-02 1.86544731e-01 -6.79978669e-01 -2.63330415e-02 -1.00585192e-01 6.77122355e-01 -5.93171239e-01 -1.21611819e-01 -2.17638910e-02 6.08081877e-01 -7.59551704e-01 2.09116653e-01 -5.91261268e-01 3.73879671e-01 1.08547471e-01 7.24636689e-02 -1.59657791e-01 1.80109590e-01 3.36825877e-01 -5.15604950e-02 -2.90526897e-01 1.09730983e+00 -9.33818579e-01 1.76722205e+00 -4.48505312e-01 6.74689949e-01 -2.20060617e-01 -7.58175552e-01 9.71404135e-01 8.36110294e-01 4.47864801e-01 -6.49008811e-01 -2.57107109e-01 2.10794464e-01 4.52304929e-01 8.13292898e-03 8.42184544e-01 -7.05791831e-01 -3.77194196e-01 4.64174598e-01 -7.79041369e-03 -1.77168638e-01 8.76488015e-02 -1.10068500e-01 1.08374631e+00 6.01133823e-01 1.71398953e-01 9.97759700e-02 3.10259908e-01 4.79061939e-02 6.12542689e-01 6.58132195e-01 9.69079360e-02 7.99519718e-01 5.80933809e-01 -4.17039335e-01 -1.27707624e+00 -9.73151982e-01 9.40503851e-02 1.70538819e+00 -3.45293641e-01 -1.01348078e+00 -7.62884378e-01 4.63436246e-02 -2.94956505e-01 8.29503357e-01 -5.01389623e-01 -2.69003898e-01 -9.05165553e-01 -1.29647791e-01 1.14236331e+00 8.63466918e-01 -3.61517102e-01 -1.42901158e+00 -5.69827378e-01 7.14044750e-01 -8.41420144e-02 -8.12031627e-01 -4.59547907e-01 5.68877816e-01 -8.60879779e-01 -7.35760689e-01 -2.79592097e-01 -7.65524328e-01 -4.91805583e-01 -4.40184206e-01 1.13517058e+00 -5.43435961e-02 -1.50825121e-02 1.00905590e-01 -3.35695505e-01 -5.15422761e-01 -6.04557097e-01 2.84075350e-01 2.25661397e-01 -3.97925884e-01 6.64999783e-02 -1.14909005e+00 -3.41672778e-01 1.71200916e-01 -7.07087934e-01 2.63130635e-01 2.55190820e-01 6.07049823e-01 6.08734727e-01 -1.48730308e-01 8.89187515e-01 -7.55278766e-01 7.13537157e-01 -7.08305240e-02 -2.18553454e-01 6.34046569e-02 -3.09469700e-01 1.70437545e-02 9.01795924e-01 -6.72668397e-01 -6.55187428e-01 1.48123786e-01 -4.90794152e-01 -6.14663422e-01 -1.22590512e-01 7.71241009e-01 2.62366772e-01 3.29306185e-01 7.53048599e-01 3.31059694e-01 -2.48789713e-01 -6.01729691e-01 4.67479169e-01 2.82415152e-01 1.03396547e+00 -9.33407962e-01 5.84676683e-01 -1.02118045e-01 6.23716973e-03 -8.56306493e-01 -8.72691274e-01 -3.21845323e-01 -6.78807855e-01 -3.00230145e-01 7.73554504e-01 -7.64947057e-01 -6.68700099e-01 9.61107295e-03 -1.09208012e+00 -4.61776882e-01 -7.51906097e-01 5.06692350e-01 -1.16430449e+00 6.64546192e-02 -1.42182410e+00 -1.06649566e+00 -1.14229910e-01 -8.91516507e-01 5.55961192e-01 1.45287588e-01 -9.52476978e-01 -8.73999715e-01 4.27662790e-01 -1.09189436e-01 3.91722739e-01 1.43307343e-01 1.33946121e+00 -7.42836773e-01 -3.89694363e-01 5.25883883e-02 3.00346196e-01 2.25603238e-01 -1.20353565e-01 1.21134549e-01 -1.17492056e+00 3.13643403e-02 -1.89808868e-02 -4.49517190e-01 8.41268659e-01 4.96859163e-01 1.07757509e+00 -2.80047119e-01 1.75846323e-01 4.62930858e-01 9.86241639e-01 5.19859850e-01 6.56904042e-01 1.72035173e-01 5.23847878e-01 7.46642590e-01 4.82428521e-01 3.49970043e-01 1.80676505e-01 8.87746632e-01 1.45644084e-01 2.69760668e-01 -3.08440924e-01 -6.63408339e-01 6.25409305e-01 1.53146470e+00 -4.33003813e-01 2.74814218e-02 -7.91181207e-01 6.81750000e-01 -1.93290043e+00 -1.17473352e+00 1.09199695e-01 2.05432987e+00 1.01201308e+00 5.39098084e-01 5.72290659e-01 6.76081538e-01 2.32382700e-01 2.08136171e-01 -3.89520645e-01 -8.42618287e-01 1.01494692e-01 5.39754391e-01 -1.39265224e-01 5.73081613e-01 -7.79368162e-01 1.00254941e+00 6.80270910e+00 9.57136393e-01 -1.02159214e+00 -1.17011391e-01 2.68517852e-01 -2.63318628e-01 -2.42068350e-01 2.83022016e-01 -6.11679435e-01 2.35631387e-03 1.42227447e+00 -1.75811648e-01 4.91349012e-01 7.57923961e-01 3.58591527e-01 4.06690717e-01 -1.52307928e+00 5.13037801e-01 -3.33198786e-01 -1.32673347e+00 2.03746647e-01 -6.21965080e-02 4.02218699e-01 -1.62940234e-01 1.16904914e-01 6.47895515e-01 1.56407893e-01 -1.34158623e+00 1.05504513e+00 6.84384227e-01 6.08374059e-01 -1.01560903e+00 2.52900362e-01 4.98901904e-01 -1.50279713e+00 -8.98051485e-02 -1.82285383e-01 -4.42135543e-01 3.08249444e-01 2.42799334e-02 -1.00079453e+00 2.98507035e-01 3.35920900e-01 7.25075543e-01 -2.52584308e-01 9.66981769e-01 -2.17383102e-01 1.06383252e+00 -1.66316971e-01 2.82265186e-01 2.86074787e-01 -6.13931194e-02 4.54156905e-01 1.32479048e+00 4.20145750e-01 2.45316625e-01 5.03184974e-01 5.31390369e-01 3.08271945e-01 1.40127018e-01 -4.26730543e-01 -1.98407918e-01 4.32655334e-01 8.19482863e-01 -6.08258545e-01 -1.31536469e-01 -2.66279608e-01 6.17764473e-01 2.85053074e-01 1.76183015e-01 -5.61631978e-01 -2.27991223e-01 6.59923613e-01 3.38443041e-01 3.53259504e-01 -3.59621137e-01 -2.99787968e-01 -6.85724616e-01 -3.54040772e-01 -8.80109608e-01 4.27793950e-01 -1.00961149e+00 -1.04749405e+00 4.79388922e-01 -3.90000880e-01 -9.94418919e-01 -1.04324925e+00 -4.59887117e-01 -6.46548450e-01 7.92993963e-01 -9.48463261e-01 -9.25195456e-01 5.88723302e-01 5.27382493e-01 6.64577544e-01 1.13946997e-01 1.14167178e+00 -4.80921827e-02 1.33989692e-01 4.04579014e-01 -1.80126265e-01 3.32095116e-01 6.03827357e-01 -1.28894222e+00 3.83310497e-01 4.20496017e-01 7.96393156e-01 6.15908146e-01 9.21270907e-01 -3.32673311e-01 -7.34652817e-01 -5.82234383e-01 1.13078153e+00 -3.67151678e-01 8.02414417e-01 -1.70250833e-01 -1.27902985e+00 6.89566791e-01 8.53676796e-02 -5.02081275e-01 1.01636410e+00 6.55075729e-01 -5.47220290e-01 1.07616350e-01 -1.57725826e-01 6.10152721e-01 8.87325048e-01 -1.24719787e+00 -9.47229862e-01 -1.69986501e-01 9.18041170e-01 -3.50496262e-01 -1.04897821e+00 6.33321941e-01 1.16568446e+00 -1.23389232e+00 9.78201568e-01 -6.72329485e-01 5.63478410e-01 -3.32513958e-01 -2.53886282e-01 -1.10258281e+00 -4.66845930e-01 -5.87228060e-01 -2.83990294e-01 1.12621069e+00 5.55073082e-01 4.20639306e-01 8.32833111e-01 1.73309311e-01 -3.35692465e-01 -7.93969572e-01 -8.66087675e-01 -7.57449210e-01 2.83402711e-01 -9.33835030e-01 2.10389227e-01 8.84730637e-01 3.92784297e-01 5.73083699e-01 -6.33420765e-01 -1.31265849e-01 -1.80988051e-02 2.67607450e-01 4.78313982e-01 -1.42219114e+00 -1.00429630e+00 -7.62490988e-01 -3.47440392e-01 -1.15003109e+00 1.11533985e-01 -7.64516950e-01 2.19447508e-01 -9.33535755e-01 -1.29762948e-01 -2.29744345e-01 -7.78772235e-01 5.11259556e-01 1.41154155e-01 3.17856848e-01 5.53689480e-01 2.03359649e-01 -2.82165617e-01 2.72804856e-01 9.15721953e-01 1.60662711e-01 -5.86487055e-01 1.35324702e-01 -3.42744857e-01 1.02752137e+00 7.39875495e-01 -5.08156180e-01 -5.30603647e-01 3.87264825e-02 4.88729954e-01 8.47124040e-01 6.85141608e-02 -1.35410321e+00 5.62123597e-01 2.97837630e-02 3.52799743e-01 -5.48587620e-01 6.89972401e-01 -1.40829518e-01 2.46229291e-01 4.16473120e-01 -9.15306270e-01 6.56908154e-02 3.10316265e-01 3.60345960e-01 -4.19120491e-01 -1.29922032e-01 5.18139064e-01 -1.03892960e-01 -4.42831337e-01 -5.31230457e-02 -5.89130938e-01 -9.73396525e-02 3.15653235e-01 -3.14210176e-01 2.99710810e-01 -7.35178173e-01 -1.36285293e+00 -2.68378049e-01 3.66460770e-01 4.68411416e-01 3.48181695e-01 -1.27166235e+00 -5.60791910e-01 8.43172297e-02 -1.62950233e-01 -3.25361222e-01 6.49128854e-02 6.51068568e-01 -8.96951109e-02 3.36893857e-01 -7.13180378e-02 -4.47977334e-01 -1.29487002e+00 4.28358644e-01 2.66036063e-01 -5.60055494e-01 -5.55335045e-01 9.22960758e-01 3.72083694e-01 -4.39589262e-01 5.48276603e-01 -4.62881953e-01 -3.87624711e-01 2.94967860e-01 4.23566580e-01 9.24061760e-02 -2.56325543e-01 -7.28755116e-01 -7.29950815e-02 6.29950106e-01 8.00100043e-02 -5.43358088e-01 1.34000456e+00 5.16915023e-02 6.63670227e-02 1.41363716e+00 7.62918591e-01 1.49869382e-01 -1.26365960e+00 -1.13082379e-01 4.34552550e-01 3.49309504e-01 -2.89173245e-01 -8.16031158e-01 -3.36992294e-01 1.12865889e+00 1.17403390e-02 4.49065417e-01 1.15557313e+00 -2.00587064e-01 1.00290132e+00 1.85687900e-01 2.16943964e-01 -9.29484487e-01 6.88650738e-03 9.54185307e-01 6.41372383e-01 -3.07319224e-01 -5.22874892e-01 -5.00864834e-02 -4.07548606e-01 1.47253287e+00 2.80408472e-01 -4.12308931e-01 6.79205954e-01 4.65920687e-01 -5.30070513e-02 2.01288864e-01 -1.26417446e+00 -2.03997064e-02 4.68523055e-01 2.34237388e-01 9.37427163e-01 5.23317754e-02 -2.70786826e-02 1.03804672e+00 -1.06900597e+00 8.65524858e-02 5.29774189e-01 4.88433212e-01 -5.85418344e-01 -1.38789940e+00 -3.07042927e-01 9.90447626e-02 -6.74001157e-01 -2.86788046e-01 -5.05634367e-01 5.26096642e-01 3.25309217e-01 8.86194587e-01 1.81857392e-01 -8.92767668e-01 3.17916960e-01 6.05686843e-01 5.32960713e-01 -6.22330248e-01 -1.02380550e+00 4.36337978e-01 2.33375356e-01 -6.36577070e-01 -3.81103605e-01 -6.58378065e-01 -1.35027540e+00 -5.62600568e-02 -7.19140768e-02 5.30771196e-01 3.74881744e-01 9.89033997e-01 -1.62987143e-01 7.02601612e-01 3.75754178e-01 -1.01051641e+00 -4.30927664e-01 -9.48959172e-01 -7.70580053e-01 1.46859989e-01 3.75527352e-01 -2.34349016e-02 -2.81468600e-01 2.80813068e-01]
[15.891481399536133, 5.317929267883301]
9b624ed2-19ed-4ae8-9481-a269c7090ac4
towards-fairness-aware-multi-objective
2207.12138
null
https://arxiv.org/abs/2207.12138v1
https://arxiv.org/pdf/2207.12138v1.pdf
Towards Fairness-Aware Multi-Objective Optimization
Recent years have seen the rapid development of fairness-aware machine learning in mitigating unfairness or discrimination in decision-making in a wide range of applications. However, much less attention has been paid to the fairness-aware multi-objective optimization, which is indeed commonly seen in real life, such as fair resource allocation problems and data driven multi-objective optimization problems. This paper aims to illuminate and broaden our understanding of multi-objective optimization from the perspective of fairness. To this end, we start with a discussion of user preferences in multi-objective optimization and then explore its relationship to fairness in machine learning and multi-objective optimization. Following the above discussions, representative cases of fairness-aware multiobjective optimization are presented, further elaborating the importance of fairness in traditional multi-objective optimization, data-driven optimization and federated optimization. Finally, challenges and opportunities in fairness-aware multi-objective optimization are addressed. We hope that this article makes a small step forward towards understanding fairness in the context of optimization and promote research interest in fairness-aware multi-objective optimization.
['Yaochu Jin', 'Wenli Du', 'Wei Du', 'Lianbo Ma', 'Guo Yu']
2022-07-22
null
null
null
null
['multiobjective-optimization']
['methodology']
[-4.85689081e-02 -2.23164022e-01 -8.32105577e-01 -8.26732695e-01 -4.46126789e-01 -3.48492652e-01 2.24804699e-01 6.26984477e-01 -8.76518667e-01 1.05865633e+00 4.08209383e-01 -4.27648008e-01 -6.71806455e-01 -5.87837279e-01 1.69570401e-01 -4.18929279e-01 -6.56968355e-02 3.41375083e-01 -9.49121177e-01 -2.99337268e-01 6.51704609e-01 4.44088578e-01 -1.38540053e+00 -6.83407811e-03 1.36334479e+00 1.08584654e+00 -7.28377998e-01 7.77624846e-01 -8.62324312e-02 8.64507258e-01 -7.89241135e-01 -1.24359834e+00 3.23996872e-01 -3.25409979e-01 -1.19759631e+00 -1.31730467e-01 2.59328306e-01 -2.84631312e-01 1.73041791e-01 1.01315439e+00 7.64772356e-01 6.62347794e-01 3.92006934e-01 -1.81805921e+00 -6.70752764e-01 4.59339887e-01 -4.60080624e-01 2.84535795e-01 2.13003829e-02 3.23755383e-01 1.54711080e+00 -1.66506901e-01 -2.95628048e-02 1.50294101e+00 9.96453762e-02 6.46793485e-01 -9.52790618e-01 -3.53875667e-01 2.51370966e-01 4.10341173e-01 -7.39848852e-01 -6.40730143e-01 4.68719274e-01 -2.19354957e-01 6.64487481e-01 9.76487875e-01 4.72682536e-01 2.41823807e-01 9.18296650e-02 8.05700123e-01 1.14204895e+00 -4.56876487e-01 2.95682847e-01 1.18316934e-01 9.80096757e-02 3.72086197e-01 3.96851778e-01 5.87240815e-01 -4.66206938e-01 -2.76852369e-01 1.13770112e-01 -1.54884964e-01 1.12349115e-01 -3.98917884e-01 -9.26954329e-01 1.27525795e+00 3.41628969e-01 -9.94514376e-02 -3.49525988e-01 -1.62019785e-02 7.52636850e-01 4.24193084e-01 5.97881615e-01 7.58987069e-01 -3.17708075e-01 -9.17350277e-02 -1.17289340e+00 4.05431569e-01 8.95681620e-01 4.80227679e-01 5.47992826e-01 3.84503990e-01 -6.82060897e-01 7.98734426e-01 2.68538684e-01 3.14617902e-01 -6.11158498e-02 -1.53480482e+00 4.68564004e-01 3.03693444e-01 3.25862288e-01 -8.84146869e-01 -4.21028554e-01 -5.69453537e-01 -7.75454819e-01 7.87412882e-01 5.37497222e-01 -2.59327292e-01 -1.34267941e-01 1.56805110e+00 2.80889928e-01 -8.08322847e-01 -1.81223541e-01 1.38951504e+00 4.83495444e-01 1.18112653e-01 6.47145450e-01 -5.06198823e-01 1.13076961e+00 -1.12231076e+00 -7.21560955e-01 5.92825636e-02 1.50504589e-01 -9.27018046e-01 7.94987619e-01 1.25448287e-01 -1.23037612e+00 -9.72575769e-02 -5.74381351e-01 -1.11023366e-01 -4.05641198e-01 -4.97847170e-01 9.70567822e-01 1.29311717e+00 -4.90392327e-01 6.09392941e-01 -2.57787798e-02 -1.89501539e-01 9.23556685e-01 4.87073213e-01 -3.96794043e-02 4.78716940e-02 -1.26486957e+00 1.33169377e+00 1.50158346e-01 -1.58966184e-01 -7.54563212e-01 -9.55730975e-01 -6.81555927e-01 3.00715268e-01 6.70370758e-01 -9.00709510e-01 1.07080138e+00 -1.21574318e+00 -1.44287336e+00 1.09769142e+00 -1.21529445e-01 -3.45868140e-01 7.74406672e-01 -1.28665522e-01 -5.68786502e-01 -4.62406904e-01 -8.53300467e-03 1.57330886e-01 5.72261572e-01 -9.66645122e-01 -9.86712933e-01 -3.08259636e-01 4.90555078e-01 5.87685585e-01 -2.37567589e-01 7.63436019e-01 3.80716562e-01 -4.50881511e-01 -1.13417208e+00 -3.48389655e-01 -5.57741404e-01 -3.36392806e-03 -4.39377457e-01 3.78973559e-02 2.51492977e-01 -2.90330768e-01 1.63247406e+00 -1.59809697e+00 9.51538086e-02 2.57642388e-01 5.33398151e-01 4.88644212e-01 -2.03133807e-01 1.75156921e-01 -5.54157048e-03 6.28476322e-01 -1.86496928e-01 -3.61701161e-01 5.96900046e-01 1.51259303e-01 1.67668372e-01 7.39829242e-01 -6.56761080e-02 1.02637792e+00 -9.78827894e-01 -4.54662919e-01 5.60204804e-01 -2.79185027e-01 -7.82319367e-01 4.49300349e-01 -1.70009732e-02 6.00861847e-01 -4.32016224e-01 1.11261308e+00 6.75786316e-01 1.25729784e-01 -9.01344866e-02 -6.80036545e-02 -5.36842406e-01 7.79512618e-03 -8.28061163e-01 1.13840044e+00 -5.89448392e-01 2.99562216e-01 4.58892345e-01 -1.35821676e+00 3.89487028e-01 1.54694036e-01 7.56419778e-01 -9.05900717e-01 4.23595548e-01 2.47151256e-01 3.90776396e-01 -2.98797995e-01 9.54087257e-01 -7.45641470e-01 -2.87642926e-01 8.34800482e-01 -2.13125423e-01 -8.17648694e-02 3.05957556e-01 -1.67510405e-01 1.77123591e-01 -4.81647044e-01 9.72980797e-01 -4.43154991e-01 8.67166579e-01 -1.44673586e-01 5.52656770e-01 7.41620541e-01 -1.16159987e+00 1.09111339e-01 4.65648323e-01 -6.58465445e-01 -8.16014290e-01 -8.04835618e-01 -4.18360159e-02 1.83098257e+00 1.43856272e-01 -8.72322172e-02 -3.03819835e-01 -8.39641750e-01 6.17899597e-01 8.49489272e-01 -5.34877598e-01 -1.48305953e-01 -1.42641082e-01 -1.38798749e+00 4.29152250e-01 -1.98926449e-01 3.98433149e-01 -7.77464330e-01 -6.68728173e-01 4.29481007e-02 -2.67886013e-01 -7.35960662e-01 -6.18976116e-01 -1.33781090e-01 -7.07171261e-01 -1.19210875e+00 -6.79920673e-01 9.17435884e-02 1.66469693e-01 8.01584050e-02 1.55844307e+00 1.49818957e-01 -5.76201677e-01 4.48824465e-01 -7.27383494e-02 -6.47935450e-01 -2.74203688e-01 -1.61874473e-01 1.39600381e-01 1.56911969e-01 4.43468422e-01 -2.16579050e-01 -7.63776839e-01 5.85558176e-01 -7.22781956e-01 -6.00585520e-01 1.51987553e-01 6.95798457e-01 2.73244590e-01 -4.23085541e-02 8.33519042e-01 -8.55090797e-01 1.18546033e+00 -7.71862864e-01 -5.04825532e-01 5.27955949e-01 -1.24918640e+00 -1.62471443e-01 8.11767578e-01 1.08429603e-01 -1.07586837e+00 -8.51949036e-01 7.63334930e-02 -5.62979467e-02 3.85649130e-02 3.62003654e-01 -2.41869643e-01 -3.53202283e-01 4.79444444e-01 -3.81849229e-01 2.06383124e-01 -3.26576591e-01 7.69290566e-01 6.16743386e-01 1.67916223e-01 -9.37862098e-01 4.07307863e-01 2.71384984e-01 2.10329458e-01 -2.79755622e-01 -9.66330826e-01 -3.10900271e-01 -1.19805001e-01 -4.15346265e-01 6.91885173e-01 -3.55649531e-01 -1.31739783e+00 2.53335685e-01 -6.66837931e-01 -9.22277272e-02 -5.77432275e-01 4.22275215e-01 -7.55590022e-01 3.88542950e-01 -1.66573986e-01 -1.30444384e+00 -7.21621573e-01 -1.13614607e+00 2.92889088e-01 6.33717835e-01 -3.61206114e-01 -1.42510712e+00 -1.40510067e-01 8.93737376e-01 1.00380802e+00 2.31553748e-01 9.61027145e-01 -6.49447024e-01 -2.59574264e-01 -1.17096148e-01 -2.60569841e-01 3.32298487e-01 1.55403316e-01 -1.04538746e-01 -8.30324590e-01 -3.22843790e-01 -1.47036776e-01 -5.55990219e-01 4.37274098e-01 5.58347225e-01 1.32001066e+00 -6.31857634e-01 5.08417368e-01 6.64295018e-01 1.48449659e+00 -7.25974888e-02 1.98194623e-01 5.24561346e-01 3.68746161e-01 9.55328822e-01 9.16237950e-01 1.13094354e+00 5.56805253e-01 4.77860063e-01 1.02739120e+00 -3.62443775e-01 2.46419534e-01 3.44079643e-01 -1.73526362e-01 1.25950992e-01 -6.03151917e-01 -4.76745486e-01 -5.90378642e-01 2.84240067e-01 -1.89596498e+00 -1.21686625e+00 1.13451093e-01 2.52012038e+00 5.82019508e-01 -4.44067270e-01 6.00422204e-01 -1.15486677e-03 9.83737290e-01 4.64558065e-01 -7.40987122e-01 -1.40653539e+00 -2.46281996e-01 9.23191383e-02 6.04644120e-01 8.66693199e-01 -1.22523153e+00 8.05812716e-01 6.96427107e+00 8.09461534e-01 -1.00302315e+00 3.76996323e-02 1.13033819e+00 -6.99078262e-01 -4.70358759e-01 -1.88124985e-01 -1.65654495e-01 4.07911807e-01 6.83518410e-01 -9.31917965e-01 9.10758972e-01 6.55812383e-01 7.61737108e-01 -1.92370206e-01 -8.87691855e-01 9.04740453e-01 -2.65819162e-01 -1.03837359e+00 -3.39868590e-02 1.92220762e-01 8.11131597e-01 -2.82641798e-01 3.21652889e-01 1.93211794e-01 5.51891446e-01 -1.65114069e+00 5.30165255e-01 4.48051423e-01 6.97823048e-01 -1.11342156e+00 7.62823403e-01 1.22218482e-01 -5.93529224e-01 -2.80723542e-01 -3.54156286e-01 -4.50512737e-01 4.04969871e-01 1.05330110e+00 1.99430063e-01 8.03123116e-01 4.70366865e-01 3.64133894e-01 1.63769081e-01 1.29693520e+00 5.71476184e-02 1.64595753e-01 2.58676946e-01 2.41944063e-02 2.76494771e-01 -3.00505131e-01 6.94961011e-01 1.20989847e+00 -1.09308243e-01 8.24847445e-02 1.29344448e-01 9.07958984e-01 -2.49449298e-01 4.92777765e-01 4.21652533e-02 -2.17524543e-01 1.66397586e-01 1.39503109e+00 4.77935113e-02 1.09133683e-02 -5.00972271e-01 6.20238006e-01 3.24790418e-01 2.45127410e-01 -8.87780130e-01 -2.41144001e-01 1.42380357e+00 -4.16658670e-01 -5.30318558e-01 2.33532578e-01 -7.96613216e-01 -1.20237958e+00 -4.22937363e-01 -1.11136889e+00 9.72311735e-01 5.59346303e-02 -1.61355758e+00 4.06748801e-02 -2.24481359e-01 -8.92832339e-01 1.49895519e-01 -4.23286289e-01 -8.57951045e-01 1.42516994e+00 -2.19604921e+00 -7.87437499e-01 3.69559950e-03 4.69455838e-01 1.03376731e-01 -3.79971981e-01 9.77162361e-01 5.33347726e-01 -7.25255728e-01 8.01605344e-01 1.55454233e-01 -3.02065790e-01 1.03691196e+00 -1.15635693e+00 -2.42598444e-01 7.70894408e-01 -5.07634282e-01 5.71401179e-01 7.68672705e-01 -7.57824406e-02 -1.14505458e+00 -9.96466815e-01 1.06646812e+00 -2.48446584e-01 3.45882386e-01 3.53659302e-01 -2.94864416e-01 4.51289713e-02 2.65115291e-01 1.83447778e-01 1.45723593e+00 3.49562228e-01 -9.55073014e-02 -3.03573787e-01 -1.75989866e+00 5.38140357e-01 7.45021582e-01 -3.04513693e-01 -9.66395810e-02 1.38470933e-01 9.99172479e-02 -2.18500927e-01 -1.16758680e+00 1.44056544e-01 8.23920012e-01 -1.19614577e+00 1.20724177e+00 -1.38324845e+00 6.71362638e-01 1.70639947e-01 -3.19076896e-01 -1.52516580e+00 -5.72737277e-01 -1.01628006e+00 -1.15826510e-01 1.04611909e+00 1.87571138e-01 -6.51970804e-01 7.54422367e-01 1.34567964e+00 2.28048548e-01 -7.70830512e-01 -9.97128665e-01 -6.69101536e-01 5.93879461e-01 -1.71402052e-01 9.64369774e-01 1.15135968e+00 2.73937732e-01 -1.30923748e-01 -9.97037053e-01 -3.08473349e-01 1.04760528e+00 5.43791771e-01 5.69315493e-01 -9.16252494e-01 -1.80814601e-02 -1.17855692e+00 -1.25435770e-01 -3.69978040e-01 2.35950381e-01 -9.59134161e-01 -4.23387945e-01 -1.30217361e+00 2.95349747e-01 -5.39695263e-01 -7.25657165e-01 3.41967106e-01 -6.81824267e-01 1.30900666e-01 6.76084399e-01 -4.03057225e-03 -1.05339587e+00 6.78423107e-01 1.44993579e+00 -1.63830832e-01 8.29763338e-02 4.85073537e-01 -1.54105496e+00 2.67452240e-01 1.16879606e+00 -4.43057060e-01 -9.91484597e-02 -2.46264741e-01 -2.38500796e-02 1.53371349e-01 8.12419727e-02 -2.47971281e-01 -1.29399374e-01 -1.27420783e+00 6.03220500e-02 2.57293493e-01 -1.80667281e-01 -8.90257955e-01 -3.16246986e-01 4.23741043e-01 -5.84928870e-01 8.57746899e-02 1.60826761e-02 2.07820013e-01 -2.16471910e-01 -1.44760847e-01 1.26033473e+00 -4.36660916e-01 -7.60138810e-01 7.03982711e-01 -2.34579355e-01 6.62254274e-01 1.05646324e+00 8.53395015e-02 -5.33293843e-01 -5.48081338e-01 -7.31561899e-01 1.00551975e+00 1.83862016e-01 3.13913673e-01 2.44679227e-01 -1.16184330e+00 -1.07170177e+00 -3.07298690e-01 9.23521817e-02 -7.77133763e-01 2.08894297e-01 9.15652156e-01 -3.30896854e-01 6.23325467e-01 -5.35299122e-01 7.48443231e-02 -1.30675316e+00 4.60826546e-01 7.33503997e-01 -5.14525235e-01 5.35567045e-01 7.52187073e-01 -1.26822621e-01 -6.47776306e-01 2.92240471e-01 5.75646698e-01 -8.58451277e-02 1.04820915e-01 4.76149172e-01 1.20209014e+00 -1.44860655e-01 -7.08684027e-01 -5.81514955e-01 3.31350386e-01 3.57998490e-01 5.27093336e-02 1.02041852e+00 -5.10124266e-01 -5.46459675e-01 6.31921273e-03 1.12615621e+00 1.16063774e-01 -7.57557988e-01 -4.03595902e-02 -6.73541352e-02 -1.07246256e+00 6.77060261e-02 -1.42156732e+00 -1.25370693e+00 8.72813225e-01 3.61470908e-01 1.86087444e-01 1.23638928e+00 -6.47604704e-01 4.60190088e-01 -1.09784275e-01 2.48595491e-01 -1.43040550e+00 -3.46308440e-01 3.71684372e-01 6.32640600e-01 -1.71923745e+00 4.30781871e-01 -4.05366533e-02 -8.92712593e-01 1.18565559e+00 4.81532305e-01 1.60981387e-01 5.39595306e-01 -2.42811322e-01 1.42457187e-01 1.91054702e-01 -4.37235653e-01 -4.26310837e-01 5.07408798e-01 6.03462696e-01 6.18900836e-01 5.83721697e-01 -7.81058073e-01 5.49590528e-01 -1.44921720e-01 -1.16588501e-02 2.55609155e-01 7.39689112e-01 -4.39093173e-01 -1.59518576e+00 -4.55958724e-01 7.43640423e-01 -1.05801857e+00 -2.67644733e-01 -3.89189988e-01 3.69354308e-01 1.58975467e-01 1.51319981e+00 -2.56465316e-01 5.32041341e-02 2.49601737e-01 -1.60631865e-01 2.52994359e-01 -2.78416574e-01 -1.19676995e+00 -4.78511572e-01 3.95435363e-01 -9.26945567e-01 -6.35383904e-01 -3.99410635e-01 -5.76697052e-01 -1.28767073e+00 -4.33801152e-02 4.12061930e-01 5.94012499e-01 8.38385761e-01 1.41764164e-01 5.03574729e-01 8.12943637e-01 -4.24929559e-01 -7.60200620e-01 -1.43205747e-01 -5.65233469e-01 3.61691147e-01 6.28806353e-01 -3.41144443e-01 -4.63764757e-01 -6.91582918e-01]
[8.994924545288086, 5.305873870849609]
b93eec45-1741-4c62-8681-3c35e9e262eb
opal-offline-primitive-discovery-for-1
2010.13611
null
https://arxiv.org/abs/2010.13611v3
https://arxiv.org/pdf/2010.13611v3.pdf
OPAL: Offline Primitive Discovery for Accelerating Offline Reinforcement Learning
Reinforcement learning (RL) has achieved impressive performance in a variety of online settings in which an agent's ability to query the environment for transitions and rewards is effectively unlimited. However, in many practical applications, the situation is reversed: an agent may have access to large amounts of undirected offline experience data, while access to the online environment is severely limited. In this work, we focus on this offline setting. Our main insight is that, when presented with offline data composed of a variety of behaviors, an effective way to leverage this data is to extract a continuous space of recurring and temporally extended primitive behaviors before using these primitives for downstream task learning. Primitives extracted in this way serve two purposes: they delineate the behaviors that are supported by the data from those that are not, making them useful for avoiding distributional shift in offline RL; and they provide a degree of temporal abstraction, which reduces the effective horizon yielding better learning in theory, and improved offline RL in practice. In addition to benefiting offline policy optimization, we show that performing offline primitive learning in this way can also be leveraged for improving few-shot imitation learning as well as exploration and transfer in online RL on a variety of benchmark domains. Visualizations are available at https://sites.google.com/view/opal-iclr
['Ofir Nachum', 'Sergey Levine', 'Pulkit Agrawal', 'Aviral Kumar', 'Anurag Ajay']
2020-10-26
opal-offline-primitive-discovery-for
https://openreview.net/forum?id=V69LGwJ0lIN
https://openreview.net/pdf?id=V69LGwJ0lIN
iclr-2021-1
['few-shot-imitation-learning']
['methodology']
[-1.35168597e-01 -7.33508542e-02 -4.00744468e-01 3.17921564e-02 -6.06179118e-01 -9.25861597e-01 6.47867858e-01 3.25988859e-01 -6.99577093e-01 8.53054404e-01 2.31307149e-01 -4.59290653e-01 -8.33289474e-02 -6.96844757e-01 -6.12022042e-01 -7.59901345e-01 -5.69002628e-01 4.53741729e-01 1.33526132e-01 -3.15075666e-01 9.23990682e-02 6.59585357e-01 -1.43812895e+00 -2.80828804e-01 7.21627593e-01 7.81225979e-01 4.89263505e-01 5.52212536e-01 1.04391918e-01 7.76257932e-01 -3.61025095e-01 1.43574804e-01 5.28966725e-01 -4.43390906e-01 -5.45481503e-01 1.13198377e-01 -3.03356677e-01 -7.31541932e-01 -6.12166584e-01 8.90755355e-01 3.56989264e-01 8.32707644e-01 3.52015793e-01 -1.16962850e+00 -1.18603610e-01 3.81477892e-01 -2.89618731e-01 1.60279691e-01 4.17838424e-01 7.36624122e-01 1.02021813e+00 -3.72513801e-01 7.23780036e-01 9.74357784e-01 7.27639273e-02 5.32701075e-01 -1.30599594e+00 -5.14595151e-01 4.61110622e-01 1.49870202e-01 -8.46440196e-01 -4.06653911e-01 6.35330021e-01 -2.72031069e-01 1.00284755e+00 8.16481113e-02 9.78569925e-01 1.06136382e+00 1.50111150e-02 1.05227256e+00 1.04233170e+00 -1.58515647e-01 6.73450768e-01 -6.40704483e-02 -1.86600715e-01 6.59610748e-01 -2.24085495e-01 5.93335986e-01 -5.76961756e-01 -1.32338390e-01 9.21007514e-01 2.92916924e-01 -2.21639648e-01 -5.72408915e-01 -1.14867318e+00 8.29390287e-01 4.66876358e-01 3.71318422e-02 -4.48585421e-01 1.49852917e-01 4.73489046e-01 8.59887064e-01 1.20264038e-01 8.22403789e-01 -3.26831609e-01 -7.13778436e-01 -3.90791625e-01 3.12796175e-01 1.08527720e+00 8.53466094e-01 8.71608138e-01 5.63501567e-02 -1.24862209e-01 5.28788507e-01 -2.55413562e-01 2.95372456e-01 5.79688251e-01 -1.26710427e+00 5.97276568e-01 3.60500336e-01 5.87194383e-01 -5.73548853e-01 -3.74325842e-01 -1.63002476e-01 -3.57530773e-01 6.19769514e-01 6.70753837e-01 -3.39892298e-01 -7.97578514e-01 2.02351165e+00 4.07063574e-01 1.20565452e-01 1.21003017e-01 8.66255581e-01 -1.15418687e-01 5.98200858e-01 -2.02982798e-01 -3.67270321e-01 8.69862795e-01 -1.04784441e+00 -4.66796577e-01 -3.79370421e-01 8.24235082e-01 -2.15403587e-01 1.50790489e+00 4.44462776e-01 -9.38348651e-01 -1.80084839e-01 -8.46590221e-01 6.05308786e-02 -3.59108359e-01 -2.97988415e-01 7.65477061e-01 -4.21394370e-02 -9.59351420e-01 9.95457649e-01 -1.21750677e+00 -3.49083543e-01 4.64977086e-01 3.12194258e-01 -3.12845141e-01 7.87902437e-03 -1.01532543e+00 8.50712240e-01 4.21810836e-01 -1.17080227e-01 -1.31428647e+00 -2.81093031e-01 -7.37316191e-01 1.96257427e-01 1.08035254e+00 -2.57151127e-01 1.53219330e+00 -9.30260241e-01 -1.73584998e+00 2.27175847e-01 6.07295409e-02 -5.69954157e-01 8.01565409e-01 -2.85931021e-01 3.66338454e-02 2.34098464e-01 -4.30175066e-02 4.57978517e-01 7.60065258e-01 -7.99716055e-01 -6.83471084e-01 -3.91784370e-01 5.02468050e-01 7.95876205e-01 -4.09790814e-01 -4.10526276e-01 -5.12253881e-01 -4.43774313e-01 -3.20447266e-01 -1.22236311e+00 -4.94761407e-01 7.75410086e-02 1.89628899e-02 -2.65952051e-01 8.12219977e-01 -5.12442231e-01 9.45105910e-01 -2.22182226e+00 3.32574278e-01 2.70056546e-01 1.59664199e-01 9.98429656e-02 -1.52584344e-01 9.89995658e-01 3.18529338e-01 -8.90947357e-02 -1.64314598e-01 -3.70070308e-01 3.50296237e-02 3.38929176e-01 -6.20643556e-01 4.66390282e-01 -2.19972894e-01 9.24915433e-01 -1.32763541e+00 -2.48507097e-01 2.72581458e-01 -1.61403194e-01 -6.25770688e-01 3.64985466e-01 -5.22317767e-01 8.62643957e-01 -8.40597391e-01 3.67815584e-01 -7.34229758e-02 -1.61295310e-01 2.82494962e-01 6.09895349e-01 -1.61299527e-01 3.42388690e-01 -1.20626318e+00 1.79592812e+00 -6.71210587e-01 5.49113333e-01 1.21861897e-01 -8.77338648e-01 4.74799573e-01 1.66203097e-01 5.92506886e-01 -8.31628501e-01 -1.09175324e-01 1.62866972e-02 3.63011919e-02 -5.14688551e-01 4.52932328e-01 7.40901902e-02 -1.71353102e-01 8.68240774e-01 -1.13863289e-01 -1.41929001e-01 3.32910001e-01 2.55677670e-01 1.30221438e+00 3.91989350e-01 4.02741998e-01 9.54195634e-02 -1.32525876e-01 1.91572279e-01 5.83082736e-01 9.05964315e-01 -2.63152957e-01 -1.75240219e-01 6.84846461e-01 -9.07120705e-02 -1.02347338e+00 -1.16591489e+00 3.92181933e-01 1.17337990e+00 2.94279397e-01 -4.61464882e-01 -2.82743663e-01 -5.42824864e-01 9.79567543e-02 6.90352738e-01 -4.67572629e-01 -1.85223460e-01 -6.82557881e-01 -1.67205781e-02 2.17393875e-01 5.73640823e-01 3.60725641e-01 -1.32234895e+00 -1.27003205e+00 2.45490074e-01 -5.25812507e-02 -7.48923421e-01 -6.46865427e-01 4.21687365e-01 -1.00789464e+00 -1.05632532e+00 -5.48961103e-01 -3.76972198e-01 5.70296228e-01 4.13572758e-01 7.48824120e-01 -9.77896675e-02 -2.69282728e-01 7.83164740e-01 -3.95000935e-01 -2.78762817e-01 -2.04892561e-01 -7.69574791e-02 2.66845077e-01 -2.58290499e-01 1.66907068e-02 -8.55835795e-01 -6.61831379e-01 4.52081673e-02 -8.60264122e-01 4.19317707e-02 5.04351497e-01 1.03761327e+00 4.60616618e-01 -4.04571109e-02 5.04864037e-01 -6.33974254e-01 7.05027878e-01 -6.52922451e-01 -9.07998323e-01 -2.84513216e-02 -6.08255029e-01 3.06975633e-01 1.08526051e+00 -7.14801252e-01 -8.85208189e-01 -8.20226967e-02 2.37010852e-01 -7.43679285e-01 -1.05383955e-01 5.60219526e-01 2.05739617e-01 3.01774055e-01 4.90547568e-01 3.44776183e-01 3.80619049e-01 -4.64511633e-01 3.84289771e-01 3.81938964e-01 1.49334878e-01 -8.30557287e-01 6.60123050e-01 4.28338170e-01 -8.08815733e-02 -8.14698279e-01 -4.22693461e-01 -5.21401048e-01 -3.55663031e-01 -2.52098352e-01 5.64757705e-01 -6.47262394e-01 -9.06086981e-01 2.42550701e-01 -4.36065257e-01 -1.16265130e+00 -6.91839159e-01 5.15187085e-01 -1.01128459e+00 3.37880313e-01 -5.73574364e-01 -8.70212018e-01 1.68720737e-01 -1.32910812e+00 5.36804736e-01 3.15136224e-01 -3.48881483e-02 -1.00259829e+00 2.87369955e-02 -1.60897091e-01 2.42386550e-01 8.29424709e-02 7.91373551e-01 -8.05080593e-01 -6.92393839e-01 1.44646913e-01 3.33879024e-01 1.00423500e-01 2.96697170e-01 -4.65543568e-01 -6.34220958e-01 -8.04778695e-01 -8.97237211e-02 -6.69251978e-01 6.39375687e-01 1.47328109e-01 1.30425584e+00 -5.29445350e-01 -2.99250364e-01 4.96817350e-01 1.31125367e+00 5.36968589e-01 1.99681938e-01 3.27068835e-01 3.64164591e-01 6.07527912e-01 9.64227021e-01 6.75633967e-01 2.19773024e-01 8.00092518e-01 5.26007712e-01 1.41334444e-01 3.76877874e-01 -6.11086786e-01 5.84463835e-01 4.21090424e-01 2.69286055e-02 2.27309968e-02 -6.96850538e-01 4.62730527e-01 -2.29948854e+00 -1.03116477e+00 8.05926561e-01 2.58466172e+00 9.51439440e-01 5.54968379e-02 5.42851150e-01 -3.74819100e-01 3.34748089e-01 1.32546902e-01 -1.35589409e+00 -2.89761335e-01 3.79550576e-01 1.41314998e-01 4.87571120e-01 5.17588556e-01 -8.28405678e-01 9.53248203e-01 5.43304920e+00 7.72457838e-01 -1.20607233e+00 -1.25158895e-02 4.99955684e-01 -5.34277141e-01 -2.31230576e-02 2.41187558e-01 -3.99078310e-01 5.02459526e-01 7.66286314e-01 -4.13558125e-01 1.19172990e+00 1.01838601e+00 4.90161508e-01 -4.24722016e-01 -1.48588490e+00 9.30214703e-01 -5.57711720e-01 -1.15811992e+00 -4.68195856e-01 4.13962364e-01 4.93405998e-01 1.92304760e-01 1.87647287e-02 5.93259037e-01 6.32170975e-01 -9.66208994e-01 6.37668967e-01 2.97323793e-01 7.90212333e-01 -8.06888342e-01 8.72733369e-02 8.76631141e-01 -1.12012231e+00 -5.28584003e-01 -1.09414943e-01 -3.91840369e-01 4.78621223e-04 -4.00479324e-02 -8.61149430e-01 3.40638340e-01 5.84722638e-01 6.39107585e-01 -1.91891640e-01 8.85949910e-01 -3.53656679e-01 4.62348789e-01 -4.86345410e-01 -3.24233472e-01 5.94542980e-01 -6.62035286e-01 7.42322385e-01 8.55328918e-01 1.83580175e-01 1.37878835e-01 5.67629397e-01 6.83960199e-01 5.68259582e-02 -8.15217122e-02 -1.18387103e+00 -2.79607922e-01 8.68230045e-01 1.03613400e+00 -7.20374644e-01 -2.40793869e-01 -2.62814641e-01 8.98633063e-01 5.99089861e-01 6.85821891e-01 -7.80250669e-01 -3.36473942e-01 7.86736548e-01 -2.14195117e-01 2.55753011e-01 -6.99199557e-01 1.95855662e-01 -1.25626802e+00 1.07891709e-02 -1.04318631e+00 5.06730616e-01 -3.27950060e-01 -8.69718313e-01 3.07576433e-02 1.27512693e-01 -1.41025841e+00 -6.94414020e-01 -3.29087973e-01 -5.88804424e-01 6.60181522e-01 -1.16931617e+00 -5.55554807e-01 -5.22530265e-02 7.26443708e-01 8.69920731e-01 -6.05716929e-02 7.11831212e-01 -1.15860857e-01 -4.33379948e-01 3.54346007e-01 4.33596224e-01 -7.48333707e-02 5.67365110e-01 -1.32329178e+00 1.19037926e-01 7.11682260e-01 3.47822338e-01 6.25766873e-01 7.45448470e-01 -5.28761208e-01 -1.87031960e+00 -8.35048795e-01 -1.01867922e-01 -1.11481175e-01 8.10330570e-01 -3.59772265e-01 -6.93026900e-01 8.96781981e-01 -9.35123786e-02 -1.68810636e-01 3.70620757e-01 1.25179589e-01 -1.46240043e-02 -1.17672766e-02 -9.00427222e-01 1.20402789e+00 1.13374174e+00 -5.38939297e-01 -4.06011045e-01 3.51824969e-01 6.64041996e-01 -4.93373156e-01 -6.45408392e-01 -9.94171426e-02 3.96517992e-01 -8.24179232e-01 7.48172104e-01 -8.17166746e-01 2.25239307e-01 -1.03059284e-01 2.81014647e-02 -1.66673970e+00 1.67169183e-01 -1.18169653e+00 -5.05872726e-01 6.58864677e-01 1.39584526e-01 -1.11098015e+00 5.73869646e-01 5.02163470e-01 -7.39847496e-02 -1.02769625e+00 -9.52540278e-01 -1.09038651e+00 -2.25347537e-03 -2.98700243e-01 4.25143391e-01 6.67486489e-01 2.32159555e-01 1.58125937e-01 -2.47918159e-01 -9.83476937e-02 5.42585492e-01 3.59730363e-01 8.14086795e-01 -7.29509175e-01 -7.44498968e-01 -4.64538515e-01 1.27085015e-01 -1.60302269e+00 2.32867464e-01 -7.75288403e-01 1.15792565e-01 -1.25290179e+00 -5.60019314e-02 -7.94444442e-01 -3.61611187e-01 5.80609500e-01 1.44004852e-01 -4.29670691e-01 5.72971702e-01 5.08890390e-01 -5.55466294e-01 7.76575804e-01 1.40916216e+00 2.32950658e-01 -6.99002028e-01 1.44188330e-01 -3.90269279e-01 6.29379213e-01 9.59112823e-01 -2.48844624e-01 -8.15511227e-01 -1.80083781e-01 4.91488725e-02 5.09918928e-01 2.79010832e-01 -8.04014385e-01 2.26587683e-01 -6.30834401e-01 5.19925691e-02 -2.67606497e-01 6.15588069e-01 -7.09019423e-01 -1.89495072e-01 6.04455352e-01 -5.66750646e-01 1.18669361e-01 5.94788454e-02 9.63850915e-01 8.01673308e-02 -2.24924639e-01 6.42318308e-01 -4.48837757e-01 -9.27814007e-01 5.43552041e-01 -3.15404266e-01 2.71525979e-01 1.26224399e+00 -1.69651166e-01 -1.42398506e-01 -7.65852988e-01 -7.99348354e-01 7.16666102e-01 7.93943584e-01 3.25038910e-01 4.77805555e-01 -1.07918143e+00 -9.50972885e-02 7.77339712e-02 -1.72510408e-02 1.62281580e-02 6.65844381e-02 9.32043612e-01 -1.35022059e-01 2.02967420e-01 -3.65362078e-01 -3.14520746e-01 -9.16052997e-01 6.41377449e-01 2.48781323e-01 -4.34535980e-01 -1.20673752e+00 4.91502643e-01 3.36819440e-01 -1.86691895e-01 5.28376460e-01 -2.84754932e-01 -1.05314314e-01 5.70343621e-02 3.57004553e-01 4.49249774e-01 -2.52523839e-01 9.98093709e-02 -2.09812578e-02 -2.55450383e-02 -6.62036091e-02 -5.97994685e-01 1.39734912e+00 -1.37244180e-01 3.40505481e-01 7.98268139e-01 8.72408330e-01 -1.45530524e-02 -2.05016303e+00 -2.80045897e-01 2.61481944e-02 -5.79140902e-01 -1.78937092e-01 -5.96604705e-01 -7.24295735e-01 7.30902493e-01 3.41172725e-01 3.54076147e-01 1.08617580e+00 -3.29386331e-02 6.77138209e-01 8.51225555e-01 8.50107729e-01 -1.42386985e+00 4.50867593e-01 6.25477493e-01 7.58901715e-01 -1.15602255e+00 -2.48934090e-01 3.07105362e-01 -8.84910762e-01 1.09771073e+00 5.01134872e-01 -2.22729728e-01 2.83582330e-01 2.97442406e-01 -3.79204184e-01 -3.62263876e-03 -1.09868479e+00 -3.92454594e-01 -3.91275316e-01 5.08212149e-01 -1.59112975e-01 1.43293083e-01 3.34530286e-02 2.42086962e-01 -1.13095507e-01 -1.05252504e-01 6.95250392e-01 1.36679959e+00 -4.41652179e-01 -1.05716014e+00 3.89463939e-02 6.65862918e-01 -2.10037738e-01 1.61972567e-01 -1.10179760e-01 8.93793821e-01 -5.08715153e-01 8.11134875e-01 6.27444312e-02 -8.95213634e-02 6.79750741e-02 3.36843939e-03 5.95882595e-01 -8.48643184e-01 -2.73863941e-01 2.66836267e-02 2.61139035e-01 -1.00972915e+00 2.10518062e-01 -6.79828167e-01 -1.39118135e+00 -1.41640052e-01 -8.10919981e-03 1.31488010e-01 4.89015400e-01 9.78680253e-01 3.74904215e-01 4.23765004e-01 8.34365487e-01 -1.06281543e+00 -1.19441140e+00 -6.93789184e-01 -6.36682034e-01 2.55340219e-01 6.99241817e-01 -8.73014033e-01 -4.68181074e-01 -2.21664608e-01]
[4.15925931930542, 1.8207002878189087]
acbb3218-dcee-4848-8d17-e8240dc1c839
model-based-validation-as-probabilistic
2305.09930
null
https://arxiv.org/abs/2305.09930v1
https://arxiv.org/pdf/2305.09930v1.pdf
Model-based Validation as Probabilistic Inference
Estimating the distribution over failures is a key step in validating autonomous systems. Existing approaches focus on finding failures for a small range of initial conditions or make restrictive assumptions about the properties of the system under test. We frame estimating the distribution over failure trajectories for sequential systems as Bayesian inference. Our model-based approach represents the distribution over failure trajectories using rollouts of system dynamics and computes trajectory gradients using automatic differentiation. Our approach is demonstrated in an inverted pendulum control system, an autonomous vehicle driving scenario, and a partially observable lunar lander. Sampling is performed using an off-the-shelf implementation of Hamiltonian Monte Carlo with multiple chains to capture multimodality and gradient smoothing for safe trajectories. In all experiments, we observed improvements in sample efficiency and parameter space coverage compared to black-box baseline approaches. This work is open sourced.
['Mykel J. Kochenderfer', 'Anthony Corso', 'Harrison Delecki']
2023-05-17
null
null
null
null
['bayesian-inference']
['methodology']
[-3.18722755e-01 -8.26577321e-02 -3.13936383e-01 -8.45354721e-02 -9.38164353e-01 -5.90086758e-01 7.43833899e-01 -6.92155585e-02 -2.32880235e-01 9.61022735e-01 -5.07939100e-01 -9.18347359e-01 1.95793621e-02 -6.31744683e-01 -9.86686826e-01 -4.21834886e-01 -5.11902213e-01 9.72088397e-01 5.49414277e-01 -2.14050248e-01 1.82809457e-01 3.91822219e-01 -1.48841083e+00 -5.51878035e-01 6.99990571e-01 3.54557842e-01 -1.99625343e-01 1.05099237e+00 9.08742845e-01 4.66365576e-01 -4.81749743e-01 2.29319409e-01 9.85079780e-02 -3.13722789e-01 -4.42360848e-01 -1.18730694e-01 -1.29452690e-01 -4.98131067e-01 -5.20756602e-01 8.75237107e-01 2.21459985e-01 2.91263331e-02 6.86480522e-01 -1.80533516e+00 4.72800106e-01 6.24609649e-01 -1.82856560e-01 3.26024115e-01 1.98508143e-01 8.58522952e-01 5.44178188e-01 -5.25457144e-01 2.60584474e-01 1.20550215e+00 7.54710376e-01 2.46196836e-01 -1.66603935e+00 -6.17059410e-01 -2.46548757e-01 2.31269717e-01 -1.78864992e+00 -4.79773968e-01 4.59069535e-02 -5.64380825e-01 1.41686320e+00 2.33055782e-02 4.53596115e-01 1.05190837e+00 9.52224076e-01 3.67195040e-01 7.97945738e-01 -3.26542631e-02 8.80499184e-01 -2.35151738e-01 3.20377558e-01 5.78859270e-01 5.72295487e-01 7.82226741e-01 -8.85078758e-02 -7.70595729e-01 1.92661077e-01 -1.52030632e-01 1.87362254e-01 -2.96045601e-01 -1.09976804e+00 5.57968020e-01 -3.51466805e-01 -5.59732199e-01 -8.04055631e-02 9.27822471e-01 3.01601052e-01 3.87825966e-01 -1.23043790e-01 1.14858285e-01 -3.41512889e-01 -5.15847027e-01 -9.92053807e-01 8.98202837e-01 1.30525732e+00 1.16362405e+00 8.52763116e-01 2.65470594e-01 -4.12693657e-02 -1.40352100e-01 3.87988240e-01 1.05388582e+00 -1.53713182e-01 -1.22943521e+00 2.03149304e-01 9.17917043e-02 7.53560483e-01 -7.34218061e-02 -4.41100538e-01 -1.25337034e-01 -5.46930060e-02 5.89145660e-01 3.38613123e-01 -5.11112750e-01 -1.08861232e+00 1.51834619e+00 3.69477123e-01 4.83692110e-01 7.26601779e-02 5.95399022e-01 -3.54384750e-01 5.75720787e-01 -2.33119994e-01 -2.64474064e-01 1.01214397e+00 -4.85519141e-01 -4.14074093e-01 -2.60290802e-01 7.28136659e-01 -5.27281821e-01 7.46151984e-01 4.43984449e-01 -1.03699780e+00 -1.20053150e-01 -1.41338992e+00 5.36032379e-01 1.95506915e-01 -7.19785541e-02 -5.93573079e-02 8.25772583e-01 -9.78367448e-01 5.98055005e-01 -1.72140956e+00 -3.09405446e-01 -7.96700791e-02 6.11920238e-01 1.91657424e-01 -1.01856552e-01 -9.78140831e-01 1.15798783e+00 2.44569331e-01 -1.49185225e-01 -1.89621997e+00 -3.66010457e-01 -8.89441013e-01 -5.60242310e-02 7.61966050e-01 -5.13854504e-01 1.68876278e+00 1.36036888e-01 -1.42392373e+00 -8.97604302e-02 -1.47937164e-01 -8.82471621e-01 5.35209179e-01 -1.47957921e-01 -2.59517729e-01 -3.89261752e-01 9.83835477e-03 1.35739133e-01 7.28805244e-01 -1.01205575e+00 -4.92953241e-01 7.51137435e-02 -1.21487506e-01 -2.58696526e-02 6.11177385e-01 -3.01383078e-01 -3.09899569e-01 2.35357225e-01 -1.89721659e-01 -1.55740190e+00 -4.18931514e-01 -7.44232357e-01 -4.81474727e-01 -1.42362058e-01 8.64229679e-01 -4.03343141e-01 1.22317886e+00 -1.87042570e+00 4.50550281e-02 6.75001323e-01 -3.26570809e-01 -4.53618944e-01 3.14036876e-01 1.11135495e+00 4.46693569e-01 -1.55204674e-02 -4.06680912e-01 -1.46848187e-01 4.26187485e-01 2.65072256e-01 -5.66083848e-01 9.17213202e-01 1.57172397e-01 4.94272470e-01 -7.85523176e-01 -3.84801388e-01 3.31057876e-01 -5.06182462e-02 -4.69074637e-01 -7.99900666e-02 -5.17839253e-01 2.77573586e-01 -4.30935889e-01 5.20539820e-01 5.02455950e-01 5.20611666e-02 3.94049823e-01 4.13135231e-01 -2.47076303e-01 1.72022521e-01 -1.32707644e+00 1.12037194e+00 -4.01925951e-01 3.75613719e-01 -1.27146050e-01 -3.63845915e-01 6.10258996e-01 7.31377304e-02 1.37898177e-01 -3.70664060e-01 2.99206078e-01 1.88170969e-01 1.82783127e-01 -3.77251446e-01 5.93401372e-01 -9.63884890e-02 -6.12281322e-01 6.58477664e-01 -2.76189804e-01 -7.37569630e-01 3.23630691e-01 2.78964609e-01 1.56273139e+00 4.43259507e-01 2.21440136e-01 -5.29042125e-01 -5.81639335e-02 5.75541615e-01 4.40636456e-01 9.25552249e-01 -1.52107596e-01 1.25333279e-01 7.78580189e-01 -5.49391285e-02 -1.38782871e+00 -1.10088515e+00 -4.57517534e-01 2.17822254e-01 5.16519427e-01 -6.50184929e-01 -9.30847645e-01 -2.62243032e-01 3.91441286e-01 1.15759683e+00 -4.28427339e-01 -2.55378395e-01 -6.23122633e-01 -6.87110603e-01 7.14991450e-01 3.32493037e-01 -4.30671312e-02 -5.79374194e-01 -8.32874238e-01 3.16942275e-01 2.64632910e-01 -7.77007878e-01 -3.71732533e-01 1.26861811e-01 -6.47810698e-01 -1.27324045e+00 1.10394426e-01 -1.62426472e-01 5.90497315e-01 -1.46675333e-01 9.58969235e-01 1.80057734e-01 -2.11979002e-01 2.70758778e-01 2.08456650e-01 -1.80050239e-01 -7.76651442e-01 -1.58742636e-01 5.62734663e-01 -7.25022018e-01 1.50397345e-01 -2.15797514e-01 -4.27471191e-01 8.45062137e-01 -4.95333016e-01 -3.17972749e-01 2.44431362e-01 9.18818772e-01 5.50067306e-01 3.66991252e-01 2.65734971e-01 -4.25397664e-01 6.75542057e-01 -9.17346954e-01 -1.28560042e+00 2.44116448e-02 -8.17946553e-01 4.31693524e-01 4.99672979e-01 -2.93839127e-01 -7.08997190e-01 2.35992804e-01 7.34146088e-02 -2.08614230e-01 -4.09454852e-02 5.43181539e-01 2.19031349e-01 2.19155952e-01 6.95148289e-01 -3.79932038e-02 3.05131704e-01 1.17376544e-01 1.08462296e-01 4.88287747e-01 3.83897394e-01 -1.02872646e+00 6.54311240e-01 4.22355205e-01 2.58769363e-01 -6.44200563e-01 7.50761330e-02 1.01927184e-02 3.97656821e-02 -3.07900816e-01 2.04068869e-01 -9.42694962e-01 -1.49220645e+00 3.88695002e-01 -5.50502658e-01 -1.02976620e+00 -1.32536694e-01 5.45883060e-01 -7.89883554e-01 2.82381654e-01 -4.17029113e-01 -1.40569949e+00 1.70578852e-01 -1.52575624e+00 1.25614274e+00 2.00366795e-01 -3.61698657e-01 -5.80879152e-01 5.11168599e-01 -2.56200969e-01 2.11835667e-01 1.23510167e-01 5.74273348e-01 -3.16304594e-01 -9.51947212e-01 -5.84978461e-01 3.88348520e-01 -1.83608845e-01 -5.25403023e-01 3.66913468e-01 -4.40299928e-01 -8.52353215e-01 -1.76428348e-01 -2.08501756e-01 5.82851410e-01 4.96317804e-01 4.66944039e-01 -1.90115824e-01 -8.02982509e-01 6.94239791e-03 1.33528912e+00 1.39233798e-01 5.62937200e-01 2.65123844e-01 9.77069288e-02 2.52021730e-01 9.22322690e-01 7.08761334e-01 5.93374670e-01 7.62883961e-01 3.93234193e-01 7.00071573e-01 3.98190171e-01 -2.68659294e-01 9.48319852e-01 1.01508707e-01 4.24640596e-01 -8.25442001e-02 -1.43225265e+00 8.79006684e-01 -1.92124343e+00 -8.59858155e-01 -2.97118038e-01 2.57610202e+00 5.74323833e-01 6.79307580e-01 4.31025565e-01 -1.27470583e-01 5.56075513e-01 -6.33799136e-01 -9.83524144e-01 -5.26935495e-02 5.49855649e-01 -1.05942421e-01 1.00294089e+00 1.00337195e+00 -6.47921562e-01 7.01938391e-01 7.45750046e+00 6.97051823e-01 -7.53142476e-01 3.73570733e-02 3.16442966e-01 -2.52291322e-01 -9.08140838e-02 7.84599423e-01 -1.19568980e+00 6.48998857e-01 1.77126229e+00 -4.87426311e-01 5.10275483e-01 6.86458886e-01 5.02563953e-01 -9.71067488e-01 -1.23870420e+00 2.71746129e-01 -4.54492092e-01 -9.44130540e-01 -9.78691757e-01 3.90971214e-01 8.14909637e-01 6.63378239e-01 -9.85275283e-02 4.62939650e-01 1.08745897e+00 -8.41441631e-01 1.01487672e+00 6.50782585e-01 6.49454892e-01 -9.53934908e-01 8.13780069e-01 6.11872256e-01 -8.44941676e-01 -1.41756251e-01 3.11467182e-02 -2.61795670e-01 5.78613222e-01 5.48983872e-01 -1.29854178e+00 3.07418138e-01 6.52795136e-01 7.16085956e-02 -2.97943175e-01 1.16119206e+00 -1.28560513e-01 1.24921179e+00 -1.06868041e+00 -1.77131325e-01 1.45265004e-02 -5.57412356e-02 7.00212121e-01 7.67905772e-01 3.99961412e-01 -4.03168082e-01 4.56289679e-01 8.97289515e-01 6.43319070e-01 -8.02035511e-01 -7.10101247e-01 1.78454936e-01 7.85290539e-01 8.68767917e-01 -8.12390327e-01 -2.73860633e-01 9.60044190e-03 2.49055281e-01 -2.38115966e-01 5.66030383e-01 -1.26283038e+00 -4.02817011e-01 5.01702130e-01 4.33672071e-02 2.24599913e-01 -8.62285793e-01 -4.28391457e-01 -8.50457728e-01 -1.07524805e-01 -7.18943179e-01 6.15878515e-02 -3.38530004e-01 -7.04558671e-01 1.90969571e-01 6.31166756e-01 -1.14082825e+00 -8.72448564e-01 -1.93720922e-01 -7.12152958e-01 8.13831925e-01 -9.02421594e-01 -5.53610563e-01 1.03063524e-01 -6.37635216e-02 4.48711455e-01 6.60131499e-02 3.73354554e-01 -1.67504355e-01 -9.53046620e-01 3.16923946e-01 5.18373013e-01 -8.48013222e-01 3.89912575e-01 -9.85333443e-01 9.67515051e-01 1.00806355e+00 -6.22668386e-01 6.65603757e-01 1.61659706e+00 -1.24030387e+00 -1.98397660e+00 -8.16433847e-01 2.34438896e-01 -7.86489546e-01 9.58197892e-01 -3.80671382e-01 -7.14065194e-01 7.47875810e-01 2.19399273e-01 -2.25679234e-01 -1.09615758e-01 6.22062236e-02 3.37437809e-01 1.70239121e-01 -1.04595423e+00 6.43604398e-01 5.50444305e-01 -3.10204059e-01 -1.17860712e-01 3.92746240e-01 2.78275996e-01 -5.81519842e-01 -7.44634867e-01 2.44369417e-01 5.27332783e-01 -6.01116538e-01 4.47826236e-01 -1.64692357e-01 -2.09821284e-01 -9.85200882e-01 -6.95793629e-02 -1.03967655e+00 2.17362180e-01 -1.14516401e+00 -4.12656218e-01 7.09414363e-01 5.79709291e-01 -6.86754525e-01 5.52102149e-01 7.18949914e-01 -3.79700959e-01 -5.39425015e-01 -1.12309313e+00 -1.01192403e+00 4.47328575e-02 -6.57572746e-01 5.67933798e-01 5.78985400e-02 4.15073544e-01 2.10196152e-02 -1.88480496e-01 7.46169150e-01 9.80472147e-01 -2.26798028e-01 1.02510703e+00 -6.83616638e-01 -4.76995826e-01 -1.22760842e-02 -2.11268172e-01 -7.36140609e-01 3.38877946e-01 -4.19347078e-01 6.41991973e-01 -9.54210937e-01 2.39452660e-01 -4.21612322e-01 1.95889890e-01 3.48347574e-01 1.71143450e-02 -2.43684966e-02 -4.00672138e-01 3.47656995e-01 -6.88121796e-01 5.62028825e-01 3.54093313e-01 -1.74064450e-02 -1.68604217e-02 1.98078081e-01 1.83989674e-01 3.24388295e-01 8.02903712e-01 -6.92829549e-01 -4.98262703e-01 1.48537368e-01 4.32888925e-01 5.83374321e-01 6.22693956e-01 -1.29839849e+00 3.54679465e-01 -5.17661154e-01 -4.10950661e-01 -1.05147767e+00 1.71374217e-01 -5.02691925e-01 7.28476465e-01 9.23709095e-01 1.48815721e-01 3.37355465e-01 6.15885258e-01 8.22614670e-01 1.57530650e-01 -2.72948563e-01 5.12703478e-01 3.85915756e-01 -5.85970938e-01 -8.62217098e-02 -8.84962320e-01 -2.17571124e-01 1.45500922e+00 4.73184772e-02 -4.98239338e-01 -4.11644757e-01 -5.67378223e-01 9.74089086e-01 1.00631917e+00 -1.32786846e-02 3.87916714e-01 -9.71101701e-01 -5.94597816e-01 1.98601931e-01 6.34975582e-02 -3.05204004e-01 2.38137782e-01 9.80796576e-01 -6.99698925e-01 3.93853426e-01 1.28141597e-01 -8.37886512e-01 -8.77516329e-01 3.26202571e-01 5.48829556e-01 6.64367899e-02 -6.27304494e-01 1.68478191e-01 -3.56125146e-01 -4.10542428e-01 -1.27553314e-01 -4.23954934e-01 6.04718447e-01 -7.38099694e-01 1.80717409e-01 7.89623618e-01 5.91649823e-02 -3.60523015e-01 -5.56294918e-01 1.80776179e-01 1.00349702e-01 -7.30870783e-01 6.77801073e-01 -1.30954340e-01 1.46580413e-01 6.09088540e-01 5.16261101e-01 -3.05961967e-01 -1.73291862e+00 3.87296438e-01 9.64348763e-02 -5.55549702e-03 2.17312664e-01 -5.96886754e-01 -1.71421707e-01 3.02308768e-01 5.29903471e-01 3.37252349e-01 4.21297222e-01 -3.03534232e-02 5.18551052e-01 6.92131937e-01 8.99212062e-01 -1.10845578e+00 -5.65899134e-01 6.33557558e-01 3.96562874e-01 -1.01354754e+00 2.41008848e-01 -4.47303131e-02 -5.29613197e-01 7.72590399e-01 6.24375701e-01 -4.39924031e-01 6.16415203e-01 8.23434949e-01 -4.37109679e-01 -1.11653335e-01 -1.23597586e+00 -5.25175147e-02 -3.47127140e-01 1.13635801e-01 -1.66211918e-01 2.86806494e-01 -3.52653742e-01 1.77624464e-01 -1.14573978e-01 1.33364558e-01 1.03655457e+00 1.34054828e+00 -5.36251068e-01 -1.02719784e+00 -5.36027849e-01 4.10058111e-01 -1.10129774e-01 1.74519643e-01 2.03130409e-01 9.52483594e-01 -3.75741363e-01 1.13811386e+00 1.45450816e-01 -5.02670407e-01 3.54902834e-01 7.32317343e-02 4.72742856e-01 -3.85276943e-01 -1.23515785e-01 -5.73561378e-02 4.88329023e-01 -5.91581464e-01 3.86895269e-01 -1.27755284e+00 -1.56999266e+00 -6.16125047e-01 -7.38996148e-01 2.85891742e-01 8.43533456e-01 1.07076097e+00 5.99892735e-01 4.60949093e-01 5.33519745e-01 -9.33937192e-01 -1.11907303e+00 -8.80342960e-01 -5.20464301e-01 -1.80845976e-01 4.10485089e-01 -9.55817103e-01 -3.83046329e-01 -7.72029310e-02]
[4.83386754989624, 2.1098368167877197]
93644853-4413-4522-b438-5c48fdaaffbe
distributional-reinforcement-learning-for
1905.06125
null
https://arxiv.org/abs/1905.06125v1
https://arxiv.org/pdf/1905.06125v1.pdf
Distributional Reinforcement Learning for Efficient Exploration
In distributional reinforcement learning (RL), the estimated distribution of value function models both the parametric and intrinsic uncertainties. We propose a novel and efficient exploration method for deep RL that has two components. The first is a decaying schedule to suppress the intrinsic uncertainty. The second is an exploration bonus calculated from the upper quantiles of the learned distribution. In Atari 2600 games, our method outperforms QR-DQN in 12 out of 14 hard games (achieving 483 \% average gain across 49 games in cumulative rewards over QR-DQN with a big win in Venture). We also compared our algorithm with QR-DQN in a challenging 3D driving simulator (CARLA). Results show that our algorithm achieves near-optimal safety rewards twice faster than QRDQN.
['Yao-Liang Yu', 'Kaiwen Wu', 'Linglong Kong', 'Shangtong Zhang', 'Borislav Mavrin', 'Hengshuai Yao']
2019-05-13
null
null
null
null
['distributional-reinforcement-learning']
['methodology']
[-7.58714795e-01 4.85696435e-01 -1.59727246e-01 6.48300350e-02 -1.14411700e+00 -5.46083272e-01 3.82055223e-01 -1.73587799e-01 -9.48972166e-01 1.39184606e+00 1.51427642e-01 -4.55403924e-01 -4.90689814e-01 -7.70788312e-01 -1.01875234e+00 -8.38737607e-01 -3.67518157e-01 5.72413385e-01 2.80421674e-01 -6.73549533e-01 1.67544380e-01 1.73685491e-01 -1.37939823e+00 -4.10351634e-01 9.32117522e-01 1.11587858e+00 1.43388316e-01 5.61919212e-01 1.60990164e-01 9.86746073e-01 -8.38366270e-01 -6.79444000e-02 6.50522470e-01 -3.61714870e-01 -3.31964940e-01 -5.69994569e-01 -4.35945302e-01 -7.12504745e-01 -6.63027406e-01 9.97539937e-01 8.25309753e-01 5.18011570e-01 5.90977550e-01 -1.40262461e+00 -1.02683462e-01 1.08608985e+00 -9.80062842e-01 2.14152426e-01 -1.71844959e-02 3.94060463e-01 7.32451618e-01 -4.03467596e-01 1.53532490e-01 1.32127655e+00 2.93579340e-01 6.33286476e-01 -1.13749671e+00 -9.78017092e-01 -1.18299693e-01 -2.36830726e-01 -1.10811627e+00 -9.57895443e-02 3.15182477e-01 -2.41428182e-01 8.04727793e-01 -3.24119627e-01 7.42522717e-01 8.41430485e-01 6.87351048e-01 8.45020175e-01 1.35441935e+00 2.98774578e-02 1.01996183e+00 -2.38500431e-01 -3.34624231e-01 2.85550028e-01 1.22308709e-01 1.16025496e+00 -5.06385922e-01 -2.34967396e-01 8.60089839e-01 -5.21242678e-01 3.61119300e-01 -3.80969822e-01 -5.98285258e-01 1.00027061e+00 2.10805744e-01 -5.39465308e-01 -4.83559191e-01 9.21489835e-01 2.88766414e-01 4.53890920e-01 2.38700911e-01 5.12423694e-01 -3.51899922e-01 -8.96458089e-01 -5.21673262e-01 9.22890306e-01 5.67983389e-01 9.28407192e-01 6.35087132e-01 5.93903959e-01 -4.64297235e-01 5.36855698e-01 2.38583311e-01 9.71336067e-01 4.11870271e-01 -1.45400345e+00 4.51512218e-01 -1.37485966e-01 7.14001417e-01 -3.49209696e-01 -4.60706115e-01 -4.88060862e-01 -3.06860536e-01 1.07598412e+00 5.64050317e-01 -9.53423142e-01 -8.66354465e-01 1.90987468e+00 3.87559235e-02 -1.92354806e-02 1.40387952e-01 8.47421885e-01 2.32548237e-01 4.25071716e-01 -2.93871053e-02 -3.98031808e-02 6.94951832e-01 -6.34442270e-01 -6.47783041e-01 -1.74040377e-01 2.81728387e-01 -9.47327763e-02 1.08973157e+00 7.22133636e-01 -1.28255892e+00 -3.42856705e-01 -1.07563055e+00 4.90950078e-01 1.86906233e-01 -4.02974248e-01 5.46835601e-01 6.57643259e-01 -9.90842044e-01 7.52318323e-01 -5.87465823e-01 4.96777982e-01 4.87076372e-01 3.55069995e-01 2.49781817e-01 3.50193590e-01 -1.42942369e+00 8.89415324e-01 5.58384299e-01 -2.99251109e-01 -1.90503299e+00 -8.07971954e-01 -7.41680324e-01 3.67012806e-03 8.35494995e-01 -9.14662983e-03 1.68669415e+00 -3.07423085e-01 -2.07052279e+00 1.91012084e-01 5.53904295e-01 -1.08528519e+00 8.12217534e-01 -5.51999450e-01 1.02475852e-01 -1.01439998e-01 -1.94117147e-02 8.68328631e-01 8.16531479e-01 -1.10952580e+00 -6.54024959e-01 -1.14618696e-01 2.26218596e-01 5.63689828e-01 1.82880312e-01 -5.03700852e-01 1.40655279e-01 -4.66065526e-01 -6.49922311e-01 -9.76048172e-01 -6.27880931e-01 -4.27639723e-01 -4.55336235e-02 -3.54913235e-01 2.48947173e-01 -2.75422066e-01 9.99520421e-01 -2.07755280e+00 5.31150252e-02 2.87428170e-01 2.00614363e-01 -2.16015317e-02 -1.87386855e-01 4.06425923e-01 2.65805960e-01 -2.17900291e-01 -1.06738366e-01 6.82608970e-03 4.65148211e-01 4.93943989e-01 -8.05775344e-01 3.44544291e-01 -1.44338906e-01 1.05092645e+00 -1.22356319e+00 -2.89638378e-02 1.80475071e-01 -1.44278780e-01 -4.86715376e-01 2.88125157e-01 -5.26182950e-01 3.13242912e-01 -3.57274562e-01 3.10101002e-01 7.00845718e-01 6.25216901e-01 -7.68107548e-02 8.50622654e-01 -4.91341874e-02 -3.40304747e-02 -1.06494188e+00 1.55931425e+00 -2.93513477e-01 2.12662473e-01 3.44674326e-02 -5.72458267e-01 1.13794696e+00 -1.77626386e-01 5.96087515e-01 -1.13319278e+00 3.16300690e-01 1.05214834e-01 7.67205795e-03 -8.37999210e-02 8.38536918e-01 -3.44057381e-01 -5.78865945e-01 6.55460238e-01 -6.62301779e-02 -6.01428747e-01 1.17737532e-01 1.85231879e-01 1.20143855e+00 4.48224515e-01 -5.40510863e-02 -6.82117045e-01 -1.89481080e-01 -1.38124198e-01 7.23799706e-01 1.07669330e+00 -6.15814567e-01 5.98798133e-03 1.31123888e+00 -3.10755432e-01 -9.01925206e-01 -1.63560510e+00 1.67670622e-01 1.18989336e+00 3.10031056e-01 -1.64595440e-01 -7.22158015e-01 -7.38006830e-01 6.24417841e-01 1.24583709e+00 -8.97085845e-01 -5.04733980e-01 -1.96356088e-01 -4.86951232e-01 8.38386536e-01 6.81055069e-01 3.78499269e-01 -1.04219735e+00 -9.41453576e-01 8.11281279e-02 2.18905270e-01 -4.90688950e-01 -5.48829973e-01 6.42544687e-01 -5.95966578e-01 -7.77280688e-01 -5.79314053e-01 -8.14898163e-02 2.26339046e-02 -1.86607629e-01 1.02641356e+00 -5.87636232e-01 8.12829807e-02 2.81862766e-01 3.34147215e-02 -9.04852152e-01 -1.57272816e-01 -2.33419150e-01 3.38976622e-01 -6.94627047e-01 3.24719220e-01 -3.86207342e-01 -4.87691283e-01 2.60084480e-01 -5.88410616e-01 -5.61072171e-01 4.51494396e-01 8.59732032e-01 6.46530032e-01 3.19046706e-01 1.06207657e+00 -2.96773642e-01 1.27266824e+00 -6.26880407e-01 -1.28762639e+00 -3.60463917e-01 -6.84471428e-01 5.32898307e-01 5.05970120e-01 -3.90856385e-01 -1.04821467e+00 -1.22246124e-01 -1.22999094e-01 -5.97561717e-01 2.87637591e-01 2.49517217e-01 2.08404347e-01 1.52910143e-01 8.53378892e-01 4.10610363e-02 3.76862973e-01 -1.83316004e-02 6.12958133e-01 3.11569989e-01 3.79200041e-01 -1.09853148e+00 7.82723367e-01 7.41305798e-02 6.37760982e-02 -4.87214476e-01 -7.31929183e-01 1.26639411e-01 1.96245074e-01 -4.86022472e-01 6.32299483e-01 -9.99300003e-01 -1.31586885e+00 4.01811004e-01 -4.51217562e-01 -1.10392761e+00 -1.08468378e+00 6.31283164e-01 -1.33366060e+00 -5.35136238e-02 -5.59540272e-01 -1.23867726e+00 -4.74492610e-02 -1.07540643e+00 7.08872557e-01 6.05315089e-01 2.44575933e-01 -6.29502714e-01 5.43688476e-01 -7.57338181e-02 4.25910890e-01 3.34129661e-01 5.51970541e-01 -2.52218664e-01 -2.64232725e-01 3.59293669e-01 1.62554439e-02 4.03786659e-01 -2.23985791e-01 -4.42239106e-01 -6.64335668e-01 -5.46068132e-01 -1.01862460e-01 -1.03576684e+00 7.74420261e-01 7.38721132e-01 1.25868964e+00 1.11041397e-01 1.47314608e-01 3.82796168e-01 1.21763945e+00 5.75802982e-01 8.45571101e-01 4.35150445e-01 2.89983023e-02 1.62351221e-01 1.19494963e+00 1.17021441e+00 2.97966361e-01 2.99337059e-01 9.73799527e-01 2.73825586e-01 3.68945569e-01 -7.33015060e-01 7.26868927e-01 1.45705789e-01 -2.54627187e-02 5.13210520e-02 -7.86792517e-01 5.49381673e-01 -2.04657125e+00 -9.30738211e-01 4.05493855e-01 2.50457621e+00 1.01521766e+00 7.36589909e-01 6.91168666e-01 -3.09372932e-01 2.96992481e-01 -1.17648788e-01 -1.25601041e+00 -8.93840671e-01 7.00555518e-02 4.40685421e-01 1.00146985e+00 6.75051510e-01 -6.22075737e-01 9.23617780e-01 7.51690769e+00 1.24177182e+00 -4.34615254e-01 -2.42191538e-01 6.94480002e-01 -5.92644393e-01 -5.52359402e-01 -2.06001908e-01 -8.05245042e-01 5.42467952e-01 1.07160735e+00 -7.24755764e-01 9.37020421e-01 1.03448200e+00 2.57602096e-01 -6.30205691e-01 -7.39974618e-01 7.81327844e-01 -4.61731046e-01 -1.02556658e+00 -5.59715271e-01 4.55494463e-01 8.75866294e-01 3.10630471e-01 4.11111951e-01 1.11102808e+00 1.43563163e+00 -1.42014956e+00 9.84775424e-01 6.33333445e-01 7.65499473e-01 -1.76846337e+00 7.14945912e-01 6.04501545e-01 -7.12252378e-01 -4.33016628e-01 -6.93124592e-01 -3.26036960e-01 -6.25318661e-02 3.96986514e-01 -7.48887837e-01 2.53719956e-01 7.85012364e-01 2.52021551e-01 2.64526233e-02 8.26517701e-01 -5.62895417e-01 6.11656845e-01 -2.96397895e-01 -2.50148118e-01 6.57391906e-01 -3.35649699e-01 5.63215017e-01 5.98326027e-01 2.01624393e-01 -1.48707451e-02 1.57268614e-01 1.03746104e+00 6.36739731e-02 -4.00885850e-01 -5.66680133e-01 -6.74553290e-02 5.68910837e-01 1.00748241e+00 -2.61186957e-01 1.48560420e-01 2.94281393e-01 1.78795338e-01 3.95083338e-01 2.98150808e-01 -1.31477451e+00 -7.46751606e-01 9.17723060e-01 -1.57576278e-01 3.97629350e-01 -3.48149955e-01 -2.71924585e-01 -4.70705062e-01 -3.74110222e-01 -7.97604799e-01 2.72942126e-01 -5.81263483e-01 -1.04501164e+00 3.90216142e-01 2.76405185e-01 -1.10306978e+00 -5.96556246e-01 -4.09918427e-01 -3.58208954e-01 8.48875701e-01 -1.54987359e+00 -2.74180591e-01 1.14473306e-01 5.39107203e-01 2.90397733e-01 -4.83377099e-01 4.28062528e-01 -1.44526958e-01 -3.24244618e-01 5.65082908e-01 4.56790149e-01 -1.90994367e-01 6.55270815e-01 -1.76225460e+00 3.58680218e-01 3.76577318e-01 -5.61459959e-01 -4.31126989e-02 8.91985536e-01 -7.09150314e-01 -1.32492220e+00 -8.34213912e-01 -3.30307782e-01 -3.40344936e-01 8.54330301e-01 -3.57884288e-01 -4.12118852e-01 4.65822726e-01 4.66725230e-01 -2.13398129e-01 3.22537810e-01 -9.41963047e-02 -7.16711208e-02 -1.87018484e-01 -1.30900848e+00 6.64765537e-01 7.37575293e-01 -1.40390366e-01 -5.21478951e-01 -1.97732970e-01 9.30466056e-01 -8.56215656e-01 -6.84995770e-01 8.26952457e-02 4.28635329e-01 -1.00593591e+00 5.47360659e-01 -5.55136859e-01 3.16239178e-01 -1.56466797e-01 -1.76268265e-01 -1.92206836e+00 -2.56511979e-02 -1.09958410e+00 -3.37104499e-01 5.29538512e-01 1.94489196e-01 -6.20745122e-01 9.89329159e-01 4.68646169e-01 -1.95063174e-01 -1.03854394e+00 -1.15643716e+00 -1.22912979e+00 8.11191499e-01 -5.21254301e-01 6.70270503e-01 6.30752072e-02 1.32996067e-01 -1.23340435e-01 -5.65511942e-01 -2.29215711e-01 1.12114251e+00 -2.04808563e-01 7.42719710e-01 -6.33167028e-01 -5.80936551e-01 -3.67667377e-01 3.06352396e-02 -1.04831719e+00 2.63842463e-01 -4.47453171e-01 4.81603980e-01 -1.13019204e+00 4.81941961e-02 -5.05103588e-01 -6.20666385e-01 4.29004908e-01 1.52107149e-01 -3.36366594e-01 1.25131428e-01 -3.67703259e-01 -7.96263993e-01 1.26091135e+00 1.36991811e+00 2.21027024e-02 -4.39251900e-01 -4.45457771e-02 -8.31726670e-01 5.99620581e-01 1.01048970e+00 -5.75490952e-01 -9.16152000e-01 -1.72160372e-01 4.27421331e-01 6.07640147e-01 8.58562440e-02 -9.11149442e-01 -3.20269726e-02 -7.61864126e-01 1.26661748e-01 -7.88396955e-01 3.09273541e-01 -3.41695458e-01 -2.22203583e-01 7.14210272e-01 -4.82093573e-01 1.11305878e-01 6.53567374e-01 9.24858987e-01 4.35531251e-02 4.14749943e-02 8.45119715e-01 6.20018207e-02 -6.01033092e-01 2.48704135e-01 -7.61129618e-01 7.59654582e-01 1.39266324e+00 1.64387196e-01 -5.40181756e-01 -8.12743247e-01 -5.16642392e-01 9.49938238e-01 1.47216544e-01 3.76841038e-01 8.60427976e-01 -1.44735849e+00 -6.68777704e-01 -6.42543659e-02 -2.28163078e-01 4.64443527e-02 1.84699312e-01 4.22897428e-01 -3.06746960e-01 9.51022580e-02 -5.12598813e-01 -3.04050595e-01 -4.06839699e-01 4.63627964e-01 6.35587871e-01 -4.19915795e-01 -3.63711387e-01 9.35077012e-01 1.92311481e-01 -5.24202943e-01 4.83611912e-01 -2.22379774e-01 1.41281053e-01 -1.29482180e-01 5.39154530e-01 7.17865050e-01 -3.57274801e-01 5.61116077e-02 -2.24297315e-01 6.23516850e-02 5.00926860e-02 -6.85579181e-01 1.18286479e+00 5.67686595e-02 4.13093746e-01 4.33314949e-01 4.02048767e-01 -5.97466938e-02 -2.18951154e+00 8.03216100e-02 -7.27601871e-02 -4.76027578e-01 2.40079850e-01 -1.05318666e+00 -9.37798142e-01 5.16590476e-01 5.85798502e-01 4.89488468e-02 8.73615921e-01 -2.63834357e-01 7.93483436e-01 3.38398844e-01 9.12104189e-01 -1.68472922e+00 4.15428251e-01 8.19051206e-01 9.64529872e-01 -9.54967618e-01 -1.51425958e-01 6.46452725e-01 -1.24799669e+00 7.03937173e-01 8.63238156e-01 -6.08015299e-01 6.31414294e-01 7.16353595e-01 2.17868853e-03 9.50767379e-03 -1.07411599e+00 -3.10533583e-01 -3.84141326e-01 7.48959303e-01 -3.75827968e-01 2.80330837e-01 -1.74553350e-01 8.09006929e-01 -4.06089842e-01 1.10403681e-02 8.00472021e-01 8.96424711e-01 -7.42409408e-01 -9.20811355e-01 -1.70343667e-01 4.12371337e-01 -2.62494564e-01 2.68915534e-01 1.89696819e-01 8.72125924e-01 -1.69250160e-01 9.70984757e-01 5.48597649e-02 -5.85469961e-01 4.11861867e-01 -4.86809313e-01 5.29929519e-01 -2.97572374e-01 -4.45670694e-01 2.98241545e-02 2.82376744e-02 -9.50186551e-01 2.98871219e-01 -6.36545300e-01 -1.70055771e+00 -5.07345200e-01 -2.34536771e-02 4.47874725e-01 5.40901542e-01 7.11792767e-01 1.47050709e-01 6.65507138e-01 8.76438200e-01 -4.97228056e-01 -1.47793055e+00 -5.83413363e-01 -1.19400215e+00 -2.07438812e-01 2.94987559e-01 -1.09463179e+00 -4.55636948e-01 -9.57337558e-01]
[4.080782890319824, 2.4855148792266846]
24801ebd-f032-4e44-917a-49a5fab0d996
global-table-extractor-gte-a-framework-for
2005.00589
null
https://arxiv.org/abs/2005.00589v2
https://arxiv.org/pdf/2005.00589v2.pdf
Global Table Extractor (GTE): A Framework for Joint Table Identification and Cell Structure Recognition Using Visual Context
Documents are often used for knowledge sharing and preservation in business and science, within which are tables that capture most of the critical data. Unfortunately, most documents are stored and distributed as PDF or scanned images, which fail to preserve logical table structure. Recent vision-based deep learning approaches have been proposed to address this gap, but most still cannot achieve state-of-the-art results. We present Global Table Extractor (GTE), a vision-guided systematic framework for joint table detection and cell structured recognition, which could be built on top of any object detection model. With GTE-Table, we invent a new penalty based on the natural cell containment constraint of tables to train our table network aided by cell location predictions. GTE-Cell is a new hierarchical cell detection network that leverages table styles. Further, we design a method to automatically label table and cell structure in existing documents to cheaply create a large corpus of training and test data. We use this to enhance PubTabNet with cell labels and create FinTabNet, real-world and complex scientific and financial datasets with detailed table structure annotations to help train and test structure recognition. Our framework surpasses previous state-of-the-art results on the ICDAR 2013 and ICDAR 2019 table competition in both table detection and cell structure recognition with a significant 5.8% improvement in the full table extraction system. Further experiments demonstrate a greater than 45% improvement in cell structure recognition when compared to a vanilla RetinaNet object detection model in our new out-of-domain FinTabNet.
['Xu Zhong', 'Nancy Xin Ru Wang', 'Lucian Popa', 'Xinyi Zheng', 'Doug Burdick']
2020-05-01
null
null
null
null
['table-recognition', 'cell-detection', 'table-detection', 'table-extraction']
['computer-vision', 'computer-vision', 'miscellaneous', 'miscellaneous']
[-1.21977299e-01 1.52550675e-02 -2.25358665e-01 -2.08152607e-02 -1.00010514e+00 -8.52976382e-01 7.04630315e-01 5.43721259e-01 -2.19338179e-01 7.68727422e-01 2.28292167e-01 -2.65043825e-01 2.69903898e-01 -1.01978350e+00 -1.00301492e+00 -4.34032798e-01 1.98564753e-01 1.06284642e+00 1.82181358e-01 6.03594407e-02 3.27322483e-01 6.17895961e-01 -1.26952624e+00 1.06868804e+00 5.83213806e-01 1.62387025e+00 -9.61904824e-02 6.35461628e-01 -5.59577167e-01 1.32950830e+00 -6.09378040e-01 -5.92776895e-01 3.14121217e-01 -3.80588584e-02 -7.43103147e-01 1.02122566e-02 9.16719854e-01 -4.08011049e-01 -5.36451399e-01 6.39303982e-01 4.14487630e-01 -4.91852343e-01 7.08624303e-01 -9.42821503e-01 -8.94049287e-01 6.74883783e-01 -5.28951466e-01 2.93377161e-01 1.95058241e-01 1.78209960e-01 1.24058509e+00 -1.04849207e+00 1.03180218e+00 1.14654839e+00 8.34340036e-01 2.76828825e-01 -1.20554280e+00 -7.14640617e-01 1.65733919e-01 2.12331980e-01 -1.47580218e+00 -6.59176588e-01 2.11455524e-01 -5.35349131e-01 1.44034004e+00 2.35886395e-01 7.15769231e-01 8.76492679e-01 4.94631112e-01 1.02490652e+00 7.55929887e-01 -1.66560605e-01 2.26248875e-01 2.70338535e-01 3.31384540e-02 1.11289215e+00 8.42853069e-01 -4.86855745e-01 -1.01741719e+00 2.15344027e-01 8.04786682e-01 -1.44865319e-01 -8.39454159e-02 -5.34538865e-01 -1.38836038e+00 5.55970609e-01 2.94848472e-01 -4.52606119e-02 9.03923158e-03 1.43388063e-01 5.60439944e-01 7.19519472e-03 2.57128537e-01 6.45189881e-01 -5.57076216e-01 2.24921271e-01 -9.12979782e-01 3.14217418e-01 1.16735888e+00 1.15472829e+00 4.02735502e-01 -1.82002947e-01 -6.51276290e-01 6.88343644e-01 1.10558681e-01 4.20560092e-01 4.76208190e-03 -7.94265866e-01 7.37083375e-01 1.13477850e+00 -1.13716125e-01 -8.54008973e-01 -4.38844919e-01 -7.42294431e-01 -9.49025214e-01 7.69639537e-02 7.34286547e-01 5.21475434e-01 -1.24050415e+00 9.68486309e-01 1.74869016e-01 -2.89925665e-01 -6.20680749e-02 4.05771226e-01 1.11803722e+00 4.30877686e-01 -3.91847908e-01 2.08235443e-01 1.64468920e+00 -8.99925590e-01 -7.09280968e-01 -8.56153667e-02 7.69301057e-01 -4.87016201e-01 9.69287217e-01 8.18135977e-01 -1.06704307e+00 -1.32111654e-01 -1.13085032e+00 -6.37386978e-01 -9.73783672e-01 1.43728048e-01 8.09344292e-01 6.76150620e-01 -9.51901197e-01 7.14346692e-02 -7.34800875e-01 -3.09714824e-01 1.35087705e+00 4.92360830e-01 -4.81595457e-01 -1.60862118e-01 -5.66198289e-01 9.24838006e-01 2.19058037e-01 1.30412057e-01 -8.72396290e-01 -9.19938862e-01 -7.65575111e-01 2.84850925e-01 7.85754681e-01 -8.46489310e-01 1.01313043e+00 1.36647403e-01 -7.22537756e-01 1.27435803e+00 -1.49354294e-01 -9.10492897e-01 5.18907368e-01 -7.23239407e-02 -5.72324917e-02 2.35048868e-02 3.05489004e-01 8.84708881e-01 1.88112050e-01 -1.22795129e+00 -9.05322969e-01 -3.56311798e-01 -2.68591821e-01 -2.08680164e-02 -1.74539700e-01 -3.04340869e-01 -1.08692217e+00 -5.64603031e-01 5.01715466e-02 -3.77104610e-01 2.05232248e-01 3.91997665e-01 -7.84203172e-01 -8.08350816e-02 6.02627993e-01 -6.96483493e-01 1.09889519e+00 -1.69644189e+00 -4.04998839e-01 1.99216038e-01 6.91443980e-01 -8.11431371e-03 1.92638576e-01 7.42939934e-02 4.56644505e-01 2.65348047e-01 -1.61108106e-01 -4.78561401e-01 4.92973179e-02 -4.52771559e-02 -5.67422271e-01 2.58005738e-01 2.82408386e-01 1.20821226e+00 -3.97038609e-01 -6.89494371e-01 -7.30201527e-02 4.33037758e-01 -7.58571684e-01 -4.48460609e-01 -5.15839577e-01 -7.76291043e-02 -2.04309478e-01 1.49329436e+00 5.42504311e-01 -6.40988410e-01 2.58480132e-01 -2.15772480e-01 1.43165678e-01 4.58052725e-01 -1.20107090e+00 1.37221813e+00 8.17575380e-02 7.10337520e-01 9.90785807e-02 -7.39968181e-01 8.35377634e-01 -2.17389926e-01 4.59504038e-01 -1.04745901e+00 9.18149501e-02 3.41454983e-01 -1.19250774e-01 8.71090069e-02 5.12265861e-01 3.83966178e-01 -6.61738142e-02 -4.00864407e-02 -1.30776372e-02 -3.89095619e-02 5.31413317e-01 5.00577033e-01 1.38598621e+00 -1.04094401e-01 2.30700508e-01 -3.46956074e-01 5.34106791e-01 3.52963477e-01 4.82642889e-01 1.06098521e+00 -5.42091280e-02 8.25755477e-01 8.37307513e-01 -7.64261901e-01 -9.76699531e-01 -9.77817178e-01 -3.27547342e-01 7.46259749e-01 -9.70512554e-02 -5.52941442e-01 -6.90969884e-01 -6.80680275e-01 5.66511989e-01 3.29906881e-01 -7.88803816e-01 2.57025242e-01 -5.08799613e-01 -6.69838071e-01 7.37295091e-01 8.56111705e-01 7.34029114e-01 -1.18111145e+00 -3.40287447e-01 3.05744618e-01 2.00889669e-02 -1.46297133e+00 -6.14713967e-01 5.84826469e-01 -5.44847310e-01 -1.20444918e+00 -2.73360163e-01 -8.91519427e-01 4.65510339e-01 -9.34985727e-02 1.33846796e+00 1.48243502e-01 -6.01609409e-01 2.02651203e-01 1.90770015e-01 -7.92770803e-01 -1.07600734e-01 2.89094567e-01 -1.02381870e-01 -1.10293984e-01 5.01795709e-01 -2.75079384e-02 -3.13456297e-01 7.23256618e-02 -5.90866864e-01 1.40085280e-01 5.80607653e-01 9.74799395e-01 1.07407153e+00 -9.52973031e-03 3.05254042e-01 -1.25123668e+00 2.82669395e-01 2.84158830e-02 -9.57226455e-01 3.94908965e-01 -1.06414044e+00 2.50612289e-01 4.99966085e-01 6.90198541e-02 -6.33460164e-01 1.62870094e-01 1.87596634e-01 -5.05196452e-01 1.28416479e-01 5.52415788e-01 -3.93902034e-01 5.74992448e-02 4.33532625e-01 3.71926844e-01 -1.17467269e-01 -4.96706516e-01 2.58250207e-01 4.79520261e-01 9.95909274e-01 -4.50310707e-01 5.66083193e-01 9.01871383e-01 1.09034888e-01 -1.67970076e-01 -9.59658146e-01 -4.18068647e-01 -6.91519976e-01 3.21017534e-01 8.89429748e-01 -1.16222334e+00 -1.03923547e+00 6.25938177e-01 -1.01381099e+00 -3.65897179e-01 -1.89821601e-01 -7.34707937e-02 -4.79300827e-01 -1.11982107e-01 -7.79192924e-01 -7.34476924e-01 -5.19951880e-01 -8.92480254e-01 1.26002252e+00 -3.34212035e-02 2.21464876e-03 -6.95354998e-01 -2.91929454e-01 6.91956222e-01 2.38603622e-01 1.31635234e-01 1.15715492e+00 -8.71693134e-01 -1.45461893e+00 -1.43017516e-01 -5.39497793e-01 -2.32001826e-01 -1.07696019e-01 -5.77429347e-02 -1.10266995e+00 -4.10242006e-02 -6.12717271e-01 -3.86588484e-01 1.62669981e+00 3.64396065e-01 1.36378026e+00 -3.89100045e-01 -7.74545729e-01 9.82187510e-01 1.35589027e+00 4.74875480e-01 8.98882091e-01 6.79846227e-01 1.09765148e+00 5.01007557e-01 2.01775610e-01 4.44347411e-01 6.20184362e-01 4.40512896e-01 2.53610283e-01 -1.53407037e-01 -3.59851062e-01 -3.33110541e-01 6.20938204e-02 2.17022017e-01 5.06797373e-01 -6.07347906e-01 -1.21676004e+00 6.59751952e-01 -1.64125299e+00 -7.90368617e-01 -8.94649774e-02 1.89224494e+00 1.02439857e+00 7.13427365e-01 -7.65217096e-02 1.84361726e-01 3.50679129e-01 -1.47864506e-01 -7.45675921e-01 -1.30396187e-01 -5.27871251e-01 1.76938206e-01 9.21543539e-01 1.92280471e-01 -1.30677772e+00 1.05624616e+00 6.28002834e+00 9.94295776e-01 -8.33529890e-01 -3.87829214e-01 1.30673838e+00 -2.97824830e-01 -2.03415498e-01 -3.97721410e-01 -1.52446342e+00 1.88635632e-01 6.92306757e-01 2.25289777e-01 4.60846573e-01 7.48315871e-01 -2.87077069e-01 -2.20824257e-01 -1.42756462e+00 1.37627530e+00 1.25444204e-01 -2.43571615e+00 4.35404509e-01 5.00031114e-01 4.97156441e-01 -1.10141262e-01 1.96128532e-01 4.21674758e-01 2.78202116e-01 -1.44303274e+00 9.30397272e-01 5.13013661e-01 1.20997477e+00 -6.57251537e-01 7.84748673e-01 -8.13943669e-02 -1.29919755e+00 -1.50668425e-02 -2.90817112e-01 2.73624212e-01 -3.86798233e-01 5.72684050e-01 -1.01353419e+00 1.78591475e-01 1.03146958e+00 7.96472907e-01 -9.99519944e-01 1.02914560e+00 3.99327725e-01 5.32243371e-01 -4.59882140e-01 2.96798255e-02 -6.41629845e-02 2.06921890e-01 1.97487339e-01 1.34838879e+00 -1.91910844e-02 -9.53652635e-02 4.56684455e-02 1.26535368e+00 -7.54184544e-01 -1.90910280e-01 -6.68402553e-01 -3.20268869e-01 6.74452603e-01 1.33045280e+00 -1.36036420e+00 -5.71417809e-01 -5.02153993e-01 5.72844326e-01 3.34569693e-01 -1.02640083e-02 -5.99708438e-01 -4.11051571e-01 4.98916626e-01 2.80447662e-01 9.96327996e-01 1.21073164e-01 -1.33806717e+00 -1.09611285e+00 1.23198792e-01 -1.02401567e+00 5.20811319e-01 -6.46578729e-01 -1.26968193e+00 4.54247147e-01 -3.96604329e-01 -8.73245895e-01 -4.40861098e-02 -1.16162837e+00 -6.91026002e-02 6.38062954e-01 -1.26854432e+00 -1.22680819e+00 -3.41947228e-01 5.09200752e-01 3.47825766e-01 -6.75179899e-01 6.54284418e-01 2.68396944e-01 -6.72314346e-01 9.79130030e-01 2.64876068e-01 6.89390779e-01 7.21277416e-01 -1.53093708e+00 7.64856517e-01 5.84672391e-01 4.75173950e-01 6.51870430e-01 2.47400343e-01 -9.41399634e-01 -1.75535119e+00 -1.17369270e+00 9.07302499e-01 -1.31681550e+00 4.76562113e-01 -1.25755167e+00 -8.81944776e-01 6.25557780e-01 4.61562611e-02 3.03173274e-01 2.56299645e-01 1.24918066e-01 -8.00287306e-01 -5.63379228e-01 -1.20134151e+00 4.88949090e-01 1.17934871e+00 -6.70224845e-01 -1.23591200e-01 2.66066581e-01 5.84362209e-01 -6.51636839e-01 -7.08404362e-01 3.71566594e-01 8.35659981e-01 -7.93941915e-01 9.70399678e-01 -1.53976575e-01 5.19407272e-01 -4.94875729e-01 -2.45003060e-01 -4.27724600e-01 -3.38831574e-01 -4.38815147e-01 -6.57451093e-01 1.22481263e+00 6.78703785e-01 -3.54898751e-01 1.38127887e+00 4.26559210e-01 -1.78770989e-01 -1.08068037e+00 -8.16563070e-01 -7.38347948e-01 5.06341010e-02 -1.68977097e-01 5.65585375e-01 7.71334410e-01 -3.03924799e-01 5.35288572e-01 2.29580268e-01 -6.27070218e-02 4.40167606e-01 3.16335931e-02 8.08290899e-01 -1.30538654e+00 -1.00139268e-01 -6.95836842e-01 -3.60646039e-01 -8.84378970e-01 -2.74060071e-01 -1.11696756e+00 -8.97651091e-02 -2.11003113e+00 5.57022393e-01 -1.96525186e-01 -4.67507601e-01 6.98439956e-01 1.58368453e-01 4.59247172e-01 2.93311507e-01 4.35386121e-01 -9.02097702e-01 1.61189884e-01 1.13537502e+00 -6.86547041e-01 -6.38156757e-02 -7.52662420e-01 -1.13483644e+00 2.67651677e-01 3.08280259e-01 -4.00593102e-01 -9.80587602e-02 -2.31398031e-01 2.29051501e-01 -2.11705551e-01 3.23239267e-01 -1.19435096e+00 5.32786906e-01 1.36871457e-01 1.18899751e+00 -1.36622989e+00 5.73374778e-02 -4.24467444e-01 -2.19100714e-01 3.53194177e-01 -2.95215547e-01 -3.78739312e-02 5.20007610e-01 6.63069546e-01 -2.75097191e-02 4.66251105e-01 4.95480418e-01 -2.35802770e-01 -4.55317318e-01 2.95763999e-01 -4.84244108e-01 6.42835274e-02 5.95218301e-01 -5.95125079e-01 -8.67443144e-01 -1.94191933e-02 -2.37312838e-01 4.69533384e-01 4.14618164e-01 1.33666113e-01 5.36675692e-01 -1.07623911e+00 -6.43482685e-01 3.03952187e-01 4.02990669e-01 5.57667017e-01 -2.74316907e-01 7.09244251e-01 -8.86828542e-01 9.88234818e-01 -1.89120233e-01 -6.66152179e-01 -9.74154413e-01 7.11877882e-01 4.02601421e-01 -8.32049966e-01 -7.09833980e-01 1.11747587e+00 5.21607041e-01 -3.69187564e-01 6.39882803e-01 -9.95474398e-01 -1.98316053e-01 2.21154839e-01 4.76314902e-01 4.00538407e-02 5.99023759e-01 1.18016023e-02 -6.88606918e-01 1.74748406e-01 -7.11064458e-01 -4.67309393e-02 1.13462818e+00 1.53052494e-01 -2.91733705e-02 2.89234638e-01 7.30337083e-01 2.97810912e-01 -1.22076166e+00 -1.23960804e-02 2.52697110e-01 -1.00002810e-01 -1.24077927e-02 -1.42190683e+00 -1.08409095e+00 6.19221270e-01 3.57765883e-01 -7.58780017e-02 8.15011859e-01 2.76608169e-02 8.65147829e-01 8.65908623e-01 3.58074129e-01 -1.08358634e+00 9.53870118e-02 7.24300504e-01 8.50577712e-01 -1.14394295e+00 1.82564065e-01 -5.00009239e-01 -5.70710540e-01 1.02308726e+00 8.33794534e-01 9.92007256e-02 3.93517315e-01 8.42464805e-01 -7.50335753e-02 -4.97119397e-01 -1.18619323e+00 -1.95799127e-01 4.23664898e-01 6.75405324e-01 3.38330150e-01 -2.57081747e-01 2.61694431e-01 8.43768418e-01 -3.04440618e-01 9.18274075e-02 4.92717147e-01 8.84956241e-01 -3.84232372e-01 -7.38716245e-01 -4.30931866e-01 1.06522167e+00 -5.65798700e-01 -4.98271823e-01 -8.38947415e-01 8.81060898e-01 1.93371892e-01 5.79521060e-01 4.52124536e-01 -3.91287208e-01 1.95892647e-01 8.01705569e-02 4.86205369e-01 -6.27391696e-01 -6.67815328e-01 2.19828665e-01 2.78789222e-01 -6.35794222e-01 2.48847723e-01 -6.45529091e-01 -1.53515196e+00 -5.31588018e-01 -1.02874555e-01 -3.42401147e-01 2.30980754e-01 8.27880859e-01 6.41640723e-01 6.39776051e-01 -2.01814562e-01 -2.02680916e-01 -1.37788951e-01 -6.67238176e-01 -7.79754341e-01 -1.51806732e-03 4.10057753e-01 -6.73214912e-01 1.60964951e-01 3.05313200e-01]
[11.694283485412598, 3.013561725616455]
e11ac91d-7296-437e-8dfc-ec78de42a1f4
observational-and-interventional-causal
2212.02435
null
https://arxiv.org/abs/2212.02435v1
https://arxiv.org/pdf/2212.02435v1.pdf
Observational and Interventional Causal Learning for Regret-Minimizing Control
We explore how observational and interventional causal discovery methods can be combined. A state-of-the-art observational causal discovery algorithm for time series capable of handling latent confounders and contemporaneous effects, called LPCMCI, is extended to profit from casual constraints found through randomized control trials. Numerical results show that, given perfect interventional constraints, the reconstructed structural causal models (SCMs) of the extended LPCMCI allow 84.6% of the time for the optimal prediction of the target variable. The implementation of interventional and observational causal discovery is modular, allowing causal constraints from other sources. The second part of this thesis investigates the question of regret minimizing control by simultaneously learning a causal model and planning actions through the causal model. The idea is that an agent to optimize a measured variable first learns the system's mechanics through observational causal discovery. The agent then intervenes on the most promising variable with randomized values allowing for the exploitation and generation of new interventional data. The agent then uses the interventional data to enhance the causal model further, allowing improved actions the next time. The extended LPCMCI can be favorable compared to the original LPCMCI algorithm. The numerical results show that detecting and using interventional constraints leads to reconstructed SCMs that allow 60.9% of the time for the optimal prediction of the target variable in contrast to the baseline of 53.6% when using the original LPCMCI algorithm. Furthermore, the induced average regret decreases from 1.2 when using the original LPCMCI algorithm to 1.0 when using the extended LPCMCI algorithm with interventional discovery.
['Christian Reiser']
2022-12-05
null
null
null
null
['causal-discovery']
['knowledge-base']
[ 4.21435475e-01 7.25445151e-01 -8.93732488e-01 1.34925935e-02 -4.23953384e-01 -3.98011595e-01 6.04294121e-01 2.53688842e-01 -3.81910533e-01 1.32134545e+00 4.82780367e-01 -7.95912921e-01 -9.80400562e-01 -7.62064934e-01 -9.08173859e-01 -8.40019703e-01 -7.97203898e-01 6.56104803e-01 -4.11674500e-01 4.01729941e-01 8.40320736e-02 3.37925375e-01 -1.40938985e+00 -5.60164861e-02 1.00059783e+00 4.98427868e-01 6.69130608e-02 5.30709445e-01 3.58560771e-01 7.88589001e-01 -2.70085245e-01 4.65495996e-02 3.08510393e-01 -7.22093701e-01 -3.89132142e-01 -5.61213255e-01 1.01757087e-01 -2.62320727e-01 -1.01027250e-01 6.03999197e-01 5.85522890e-01 2.14275971e-01 4.06511515e-01 -1.36259151e+00 -1.49777561e-01 1.03645420e+00 -5.47101736e-01 7.14647919e-02 4.61569637e-01 1.27938777e-01 9.56288695e-01 -2.73025006e-01 8.08914065e-01 1.49212778e+00 2.65111476e-01 5.26661634e-01 -1.69178402e+00 -1.12361050e+00 5.68596601e-01 1.20567463e-01 -7.36784458e-01 -1.92655236e-01 3.09017241e-01 -4.85840529e-01 5.75425029e-01 6.82078958e-01 7.10434139e-01 1.23391604e+00 1.96603701e-01 4.09460425e-01 1.37631977e+00 -5.96254289e-01 5.98810911e-01 -4.51861918e-02 1.41239986e-01 6.09510660e-01 5.22953689e-01 1.36217117e+00 -6.78069353e-01 -5.73989689e-01 7.34082460e-01 9.51204263e-03 -1.66703954e-01 -3.20126742e-01 -1.22455931e+00 1.07745087e+00 2.58449614e-01 -3.50332968e-02 -7.40502357e-01 3.14373314e-01 2.06939474e-01 3.78901780e-01 2.11417869e-01 8.36890876e-01 -7.09842384e-01 8.89109448e-02 -7.81645656e-01 6.19228005e-01 6.97538137e-01 6.15611792e-01 1.19446188e-01 -3.02211285e-01 -4.07032102e-01 5.72396182e-02 2.34899402e-01 7.34572887e-01 7.57983280e-03 -1.33219111e+00 3.17219824e-01 5.61092436e-01 3.62069726e-01 -4.82332259e-01 -6.86360538e-01 -6.52734220e-01 -7.39705622e-01 3.71702880e-01 4.02796775e-01 -7.48254776e-01 -7.60960996e-01 2.05706835e+00 5.37338138e-01 2.37846404e-01 3.93997952e-02 7.58842111e-01 1.41973689e-01 5.87392807e-01 4.30262029e-01 -1.17432666e+00 9.83129680e-01 -3.37556392e-01 -8.15230072e-01 4.41001952e-02 6.21533096e-01 -2.80230224e-01 4.85246688e-01 3.65257949e-01 -1.11244500e+00 -1.64726883e-01 -7.25021660e-01 6.27675712e-01 -7.16968346e-03 -1.17435344e-01 9.49633837e-01 5.83482385e-01 -6.21289194e-01 4.98939812e-01 -8.35951030e-01 -7.31435269e-02 2.29716912e-01 6.23969674e-01 -1.15694977e-01 9.57648177e-03 -1.31060457e+00 5.91882348e-01 4.82312828e-01 -2.29669645e-01 -1.24503767e+00 -1.72587681e+00 -3.84602189e-01 3.56188148e-01 7.81233668e-01 -1.02062821e+00 9.91683781e-01 -9.16231155e-01 -1.11294937e+00 1.56924590e-01 -2.60583311e-01 -6.12043619e-01 7.95837104e-01 -1.13519341e-01 -1.03778608e-01 -4.92667370e-02 4.62004781e-01 3.78000528e-01 4.16892231e-01 -1.23160625e+00 -9.98038948e-01 -4.09525245e-01 7.20980093e-02 9.20540765e-02 1.52242079e-01 -4.12009507e-02 1.67486668e-01 -5.61626494e-01 -2.46544704e-01 -1.15091407e+00 -5.57856858e-01 -3.84214997e-01 -4.64609951e-01 -1.84070930e-01 3.95626307e-01 -3.57434899e-01 1.37179554e+00 -1.86887372e+00 4.13006902e-01 3.64979029e-01 1.45179659e-01 -2.50423759e-01 1.07584678e-01 4.90215242e-01 -7.33640850e-01 3.69928390e-01 -2.44637772e-01 -5.24935313e-02 -1.99235067e-01 2.15223487e-02 -2.92395711e-01 5.07959247e-01 -1.94934666e-01 6.92914724e-01 -1.00711334e+00 -1.71368390e-01 2.09175542e-01 -2.46588334e-01 -7.24358797e-01 1.74503922e-01 -3.96667898e-01 8.58978391e-01 -5.64381301e-01 2.80924439e-01 4.39968050e-01 -4.52391319e-02 6.31386876e-01 4.35804784e-01 -5.13591349e-01 7.87962899e-02 -1.16930521e+00 1.22815621e+00 -2.92250901e-01 3.20013225e-01 -1.02147490e-01 -1.16801941e+00 4.77598786e-01 6.57399952e-01 8.89687657e-01 -7.93533266e-01 -1.67434618e-01 1.45122871e-01 2.12770879e-01 -6.00946307e-01 -3.64328206e-01 -4.46791440e-01 -5.84312752e-02 4.50719625e-01 -3.46433550e-01 7.84232557e-01 5.48857078e-02 3.54989409e-01 1.30601990e+00 1.91286998e-03 2.97288567e-01 -5.39284647e-01 2.74061918e-01 1.83953702e-01 9.04491842e-01 1.11216176e+00 2.89893091e-01 -1.76443011e-01 8.48926604e-01 -3.38470906e-01 -7.75946259e-01 -1.18673718e+00 -1.71831563e-01 7.78142154e-01 -2.55550295e-01 9.55586284e-02 -2.90881068e-01 -7.87860215e-01 4.09068376e-01 1.29508173e+00 -1.10977781e+00 -2.57818282e-01 -6.45226002e-01 -1.05453420e+00 1.08532697e-01 1.97940663e-01 2.23899424e-01 -8.72863650e-01 -8.51812422e-01 3.83356422e-01 -1.29075035e-01 -3.09307337e-01 -2.29718219e-02 2.90842205e-01 -1.05859482e+00 -1.24443948e+00 -2.92605281e-01 9.85787734e-02 6.42868340e-01 -1.96114317e-01 6.97311401e-01 -1.22221716e-01 -2.50671715e-01 2.70205647e-01 -2.63354868e-01 -7.31669366e-01 -3.81809860e-01 -2.41287157e-01 1.49883047e-01 -1.46183938e-01 3.35191786e-02 -6.26580060e-01 -6.08919263e-01 -7.74923014e-04 -4.41233009e-01 -1.48030715e-02 3.99985433e-01 8.91079068e-01 3.01649630e-01 3.83407027e-02 1.02067995e+00 -9.57690835e-01 2.43772045e-01 -9.11029398e-01 -1.10507762e+00 1.51342243e-01 -1.27878857e+00 2.48881429e-01 1.99146673e-01 -6.79227173e-01 -1.51529074e+00 -2.78221350e-03 5.84789336e-01 -2.93792397e-01 -9.11274273e-03 7.84348428e-01 -1.20064095e-01 5.63966751e-01 6.20378315e-01 -4.99555171e-01 -1.22308947e-01 -4.63615328e-01 3.38059157e-01 -1.59869827e-02 2.08097652e-01 -6.86537564e-01 4.18493837e-01 5.18653750e-01 6.25529051e-01 -1.61334753e-01 -7.80663013e-01 -1.89966932e-01 -2.67548621e-01 -1.36777848e-01 8.80333602e-01 -7.56614327e-01 -1.30369818e+00 -3.24857056e-01 -9.18593228e-01 -4.40029770e-01 -6.27581894e-01 1.12695909e+00 -7.49938667e-01 -3.93405348e-01 -4.85816598e-02 -1.29984641e+00 -9.78672691e-03 -9.17479336e-01 6.44539893e-01 3.67564261e-02 -3.62802744e-01 -1.00611913e+00 2.78925568e-01 1.33972332e-01 -2.08258763e-01 6.48350477e-01 1.22415352e+00 -4.00836080e-01 -5.97448528e-01 -1.34923786e-01 6.19517267e-02 -5.25294065e-01 1.41845718e-02 -1.45221189e-01 -6.39554977e-01 -2.97361195e-01 -1.97192896e-02 3.74811381e-01 6.35375440e-01 1.17713213e+00 8.81808698e-01 -6.03956103e-01 -8.43906164e-01 4.08847481e-01 1.41466749e+00 7.15319753e-01 3.96119177e-01 4.93844807e-01 3.29833537e-01 9.25397456e-01 8.76202166e-01 5.70380986e-01 1.07916147e-01 6.30690455e-01 6.50997519e-01 -3.70002836e-01 3.36857051e-01 -3.97907317e-01 2.56671965e-01 -2.28147343e-01 -3.58160406e-01 -1.20429918e-01 -6.01875782e-01 5.46223879e-01 -2.21730375e+00 -1.21312475e+00 -6.25124753e-01 2.74059868e+00 8.63603413e-01 -9.66970921e-02 2.88046122e-01 -1.22733831e-01 5.53062737e-01 -6.00126266e-01 -5.81744313e-01 -5.03475070e-01 6.87124953e-02 2.74444103e-01 1.07914782e+00 7.06698477e-01 -7.59395838e-01 3.30916166e-01 6.15436268e+00 3.60111177e-01 -5.42119265e-01 1.68992147e-01 6.26580417e-01 -6.08954787e-01 -2.43661791e-01 4.65517402e-01 -7.04791665e-01 4.16548818e-01 1.53766680e+00 -6.29985213e-01 3.74144554e-01 2.10466653e-01 1.17908025e+00 -4.91891742e-01 -1.34696949e+00 3.16197097e-01 -6.78518414e-01 -1.42635155e+00 -2.47313440e-01 5.66057682e-01 1.14466119e+00 -4.26378220e-01 -1.23391189e-01 8.86112601e-02 8.95763278e-01 -9.37206507e-01 7.36896932e-01 8.94899189e-01 7.99587250e-01 -8.58891368e-01 7.08181322e-01 4.89758015e-01 -6.05531096e-01 -6.07609689e-01 3.40887308e-02 -3.62925410e-01 2.31873184e-01 7.56965399e-01 -7.31893599e-01 8.51905346e-01 6.55351937e-01 3.14945161e-01 2.88810842e-02 9.70549941e-01 -3.86193097e-01 9.68283713e-01 -1.62724525e-01 2.61405885e-01 7.57443532e-02 -1.51717305e-01 9.08841491e-01 7.93594956e-01 2.21540764e-01 3.21896613e-01 1.78598672e-01 8.89100492e-01 1.58147156e-01 -1.05761737e-01 -5.84328055e-01 1.46645114e-01 5.56868017e-01 5.01811504e-01 -1.79357097e-01 -5.24179220e-01 -5.79979829e-02 3.90105247e-01 -2.41932347e-02 5.12218416e-01 -9.57316935e-01 4.50111926e-01 3.98837358e-01 1.72957554e-01 -1.68178350e-01 3.37403804e-01 -6.34501755e-01 -6.46633446e-01 -3.27720731e-01 -9.36371744e-01 1.01059163e+00 -4.26459879e-01 -9.31787252e-01 -3.88119817e-01 7.05796659e-01 -7.45740533e-01 -3.38074535e-01 1.19185187e-02 -5.06488562e-01 1.15168035e+00 -1.04995203e+00 -6.59051657e-01 2.09970787e-01 3.42173845e-01 4.38456774e-01 9.72530618e-02 8.63646507e-01 3.91286165e-02 -6.21855676e-01 3.34773332e-01 1.65034980e-01 -6.18766487e-01 5.88704169e-01 -1.39244580e+00 -3.84499848e-01 6.25045776e-01 -5.21143913e-01 7.88722873e-01 9.73916829e-01 -1.17326128e+00 -1.30562425e+00 -1.04087579e+00 9.46299493e-01 -1.75471529e-01 7.07772553e-01 -4.77119721e-02 -5.41043937e-01 6.90148652e-01 6.10267892e-02 -5.32783091e-01 4.36174124e-01 5.49539208e-01 8.26139599e-02 -9.04075056e-02 -1.03378069e+00 6.98239267e-01 1.02088737e+00 1.55678019e-01 -4.54112560e-01 5.76219440e-01 9.72110212e-01 3.43858860e-02 -8.70599866e-01 5.00180602e-01 7.19090164e-01 -6.39722586e-01 9.31626916e-01 -1.04440618e+00 3.68074894e-01 -4.17243689e-02 1.97351545e-01 -1.33733046e+00 -4.04668331e-01 -8.44530523e-01 -3.75889763e-02 1.00043476e+00 7.12272346e-01 -8.44230056e-01 5.81874371e-01 7.60777414e-01 1.94981292e-01 -3.11891913e-01 -1.13720429e+00 -8.24532449e-01 3.01922679e-01 -2.64658540e-01 3.58962595e-01 1.12652361e+00 1.78829227e-02 2.33165011e-01 -3.78538579e-01 4.96144444e-01 1.12109303e+00 3.62400860e-01 4.78610605e-01 -1.40270722e+00 -5.49028158e-01 -3.90753627e-01 2.48548582e-01 -2.41860360e-01 8.15712884e-02 -7.70434499e-01 -3.24095845e-01 -1.29781306e+00 5.09119868e-01 -8.04003358e-01 -2.22723007e-01 5.05361438e-01 -4.58496392e-01 -7.78299570e-01 1.22167394e-01 4.54594679e-02 3.01007390e-01 3.33771706e-01 1.03937936e+00 -4.91439179e-02 -7.58434832e-01 3.01672578e-01 -5.44545591e-01 4.77987468e-01 8.88543904e-01 -9.05777752e-01 -6.88331664e-01 2.78866857e-01 3.33938599e-01 8.90246451e-01 8.33287954e-01 -1.72862336e-01 1.01416335e-01 -7.52401590e-01 2.64606029e-01 -3.73320758e-01 -2.71017194e-01 -9.28399384e-01 8.87275219e-01 1.09341860e+00 -7.41171896e-01 -8.48793685e-02 3.20137799e-01 9.85100448e-01 3.61168146e-01 -1.84004813e-01 3.70374024e-01 -1.15939878e-01 -5.68433851e-02 -1.26289323e-01 -4.96372551e-01 -2.58839488e-01 1.25161469e+00 3.18782330e-01 -3.64246398e-01 -2.82078058e-01 -1.26422369e+00 5.44275224e-01 -8.75372291e-02 2.27607459e-01 1.63243920e-01 -1.05117011e+00 -8.93628657e-01 -2.58351862e-01 -3.27820927e-01 -3.74054939e-01 2.45984957e-01 1.25319934e+00 2.40045741e-01 6.90679014e-01 1.32611141e-01 -2.21229941e-01 -1.23435640e+00 1.24587023e+00 4.12519902e-01 -5.63814878e-01 -6.41396821e-01 3.84348184e-01 6.31459415e-01 -1.29444852e-01 1.83112398e-01 4.02630083e-02 3.12921740e-02 2.82350898e-01 2.80879110e-01 9.38946128e-01 -3.23668748e-01 2.29822308e-01 -3.09021056e-01 2.96276505e-03 2.62622207e-01 -4.69435960e-01 1.35248744e+00 -1.46697521e-01 -2.92131245e-01 3.88765424e-01 8.81659031e-01 8.39100108e-02 -1.11517835e+00 1.86635792e-01 1.48760483e-01 -3.08479875e-01 2.70486236e-01 -1.44423473e+00 -7.56449044e-01 3.02353859e-01 8.15771759e-01 1.20573819e-01 1.28056335e+00 -1.94077075e-01 -3.01001489e-01 -2.37684369e-01 4.07197773e-01 -6.01294398e-01 -5.15955746e-01 -3.42354745e-01 1.01905549e+00 -9.85875487e-01 3.99574667e-01 -4.20432746e-01 -3.30490887e-01 6.61691129e-01 1.68691739e-01 -4.69498262e-02 4.77422684e-01 1.51487827e-01 -4.89709407e-01 -4.41739142e-01 -1.13935006e+00 -5.14875911e-02 7.79078379e-02 3.56306434e-01 1.49382129e-01 6.81453466e-01 -9.79070127e-01 3.28969628e-01 9.28262845e-02 3.67659852e-02 4.71200377e-01 6.59477353e-01 3.53496731e-03 -1.23615229e+00 -6.85606003e-01 7.07581222e-01 -4.96356875e-01 -8.93427506e-02 -2.27018207e-01 1.02847600e+00 2.33560294e-01 1.28032053e+00 8.42453018e-02 2.49768317e-01 5.05574405e-01 -5.76129146e-02 2.14697868e-01 -3.74657243e-01 -7.02259898e-01 4.35827821e-01 3.95885587e-01 -7.64838517e-01 -5.53009689e-01 -1.35199201e+00 -1.31146026e+00 -3.04345161e-01 -3.52752477e-01 3.88024241e-01 4.97525275e-01 8.87592554e-01 3.38720739e-01 8.40258718e-01 8.04799736e-01 -2.52346903e-01 -4.51396406e-01 -7.64161229e-01 -3.53394240e-01 -1.19786829e-01 4.90303189e-01 -1.00798869e+00 -5.46106577e-01 6.63736463e-02]
[7.827875137329102, 5.257971286773682]
10a69b52-f55b-4102-9d3d-c19dc4629c9d
optimized-high-resolution-3d-dense-u-net
null
null
https://www.mdpi.com/2076-3417/9/3/404/htm
https://www.mdpi.com/2076-3417/9/3/404/pdf
Optimized High Resolution 3D Dense-U-Net Network for Brain and Spine Segmentation
The 3D image segmentation is the process of partitioning a digital 3D volumes into multiple segments. This paper presents a fully automatic method for high resolution 3D volumetric segmentation of medical image data using modern supervised deep learning approach. We introduce 3D Dense-U-Net neural network architecture implementing densely connected layers. It has been optimized for graphic process unit accelerated high resolution image processing on currently available hardware (Nvidia GTX 1080ti). The method has been evaluated on MRI brain 3D volumetric dataset and CT thoracic scan dataset for spine segmentation. In contrast with many previous methods, our approach is capable of precise segmentation of the input image data in the original resolution, without any pre-processing of the input image. It can process image data in 3D and has achieved accuracy of 99.72% on MRI brain dataset, which outperformed results achieved by human expert. On lumbar and thoracic vertebrae CT dataset it has achieved the accuracy of 99.80%. The architecture proposed in this paper can also be easily applied to any task already using U-Net network as a segmentation algorithm to enhance its results.
['Malay Kishore Dutta', 'Kamil Říha', 'Václav Uher', 'Radim Burget', 'Martin Kolařík']
2019-01-25
null
null
null
applied-sciences-2019-1
['unet-segmentation']
['computer-vision']
[ 2.36271873e-01 5.40571988e-01 3.30577604e-02 -5.41177869e-01 -3.53293717e-01 6.95978478e-02 1.77704692e-01 3.22622955e-01 -9.30622637e-01 4.89937246e-01 -2.75034219e-01 -6.19857788e-01 9.95967984e-02 -1.02365386e+00 -4.62796420e-01 -2.28448182e-01 -2.40221769e-01 1.11565149e+00 5.74994743e-01 9.08426121e-02 1.41205549e-01 9.66201127e-01 -1.19169044e+00 2.16803879e-01 3.02867919e-01 1.13050151e+00 2.86445111e-01 7.24032819e-01 -2.97498405e-01 7.23630548e-01 -5.81357539e-01 4.28343296e-01 4.42496926e-01 4.04633284e-02 -1.20828903e+00 1.55744076e-01 4.72619861e-01 -6.16061211e-01 1.06444106e-01 9.10819173e-01 7.82412529e-01 -8.49481598e-02 7.93687403e-01 -5.77004135e-01 -1.63625926e-01 5.29414773e-01 -9.65198100e-01 8.97904396e-01 -4.60502915e-02 -3.94885957e-01 6.44570440e-02 -4.91923630e-01 6.65232539e-01 1.07556331e+00 6.76458776e-01 5.60221612e-01 -1.00929570e+00 -3.75187248e-01 -7.38342583e-01 -2.30588749e-01 -1.15101409e+00 4.35385227e-01 3.93648952e-01 -7.90321648e-01 1.39231217e+00 9.53539684e-02 9.32457268e-01 1.65682316e-01 5.88984370e-01 6.53441727e-01 1.33819330e+00 -3.11869621e-01 3.53601754e-01 -3.67569387e-01 7.37052500e-01 8.93782198e-01 3.53698432e-01 -1.66860580e-01 1.37475893e-01 1.14465140e-01 1.62183404e+00 -5.68483062e-02 5.67434840e-02 -1.09836385e-01 -9.90451813e-01 9.05223787e-01 7.52106667e-01 8.31572235e-01 -5.55925786e-01 1.29863814e-01 8.05772126e-01 -1.88330542e-02 5.20382822e-01 1.36224538e-01 -4.76479471e-01 1.09446786e-01 -1.34083545e+00 -9.91127118e-02 3.27562541e-01 6.23959601e-01 3.09926033e-01 3.74712408e-01 9.36401915e-03 5.83589077e-01 6.34402707e-02 -3.59237194e-02 1.09449291e+00 -7.16730416e-01 -1.01228720e-02 6.77560210e-01 -4.69615936e-01 -4.65378910e-01 -1.15894520e+00 -1.69691920e-01 -1.07370365e+00 7.19402254e-01 3.01117629e-01 -2.75031239e-01 -1.76095188e+00 6.65522158e-01 5.17161965e-01 -2.88057178e-01 -1.89647943e-01 9.96466219e-01 1.35322630e+00 4.40211296e-01 2.54779123e-02 4.22954075e-02 1.53612399e+00 -7.86744654e-01 -6.35286808e-01 7.86637068e-02 5.41765571e-01 -5.37815750e-01 7.40256011e-01 4.67315823e-01 -1.31842422e+00 -7.43613124e-01 -1.26636589e+00 -3.26809138e-01 -3.12714845e-01 1.53628802e-02 6.54549181e-01 1.02068460e+00 -1.47730064e+00 6.71166122e-01 -1.35560882e+00 -2.74997085e-01 1.00755048e+00 9.88514960e-01 -3.01400930e-01 4.43424731e-01 -7.35345960e-01 9.95225608e-01 1.10877490e+00 -3.44393313e-01 -5.72431266e-01 -6.17418826e-01 -6.33324742e-01 -1.98813602e-01 -1.14112377e-01 -5.34920812e-01 1.33104324e+00 -7.48675048e-01 -1.42721021e+00 1.37390518e+00 4.16608006e-01 -9.82526720e-01 6.28357351e-01 1.57932296e-01 1.38456926e-01 7.69119561e-01 -7.85503313e-02 1.27720773e+00 5.68312705e-01 -7.33921468e-01 -5.25723159e-01 -7.80068040e-01 -3.45014781e-01 1.52265623e-01 3.29517871e-01 -1.35108650e-01 -2.51476556e-01 -3.23643357e-01 5.75120866e-01 -4.81262743e-01 -4.91212100e-01 -2.04457507e-01 -1.14784308e-01 -2.25360915e-01 1.04013824e+00 -7.51544416e-01 4.14907396e-01 -1.54439223e+00 -4.03004497e-01 3.61035824e-01 5.26879251e-01 4.58734989e-01 5.97552299e-01 -5.81942677e-01 -3.63270015e-01 -4.33338694e-02 -4.86134052e-01 -6.03632443e-02 -6.66757286e-01 3.72301310e-01 4.79025930e-01 6.17678285e-01 -2.42672473e-01 1.00934970e+00 -4.10302788e-01 -9.00443733e-01 6.24869943e-01 7.89705157e-01 -3.63844216e-01 -5.19643240e-02 1.67543620e-01 5.06318927e-01 -3.83239001e-01 4.25138712e-01 7.66857982e-01 -9.84991416e-02 -5.22683337e-02 -2.22676620e-01 -6.25299960e-02 -3.36371899e-01 -8.15750957e-01 1.92955375e+00 -2.32973099e-01 6.73707545e-01 1.40589982e-01 -1.35888863e+00 1.11704373e+00 5.42171419e-01 1.00355852e+00 -9.09342647e-01 9.61759746e-01 2.93393224e-01 8.68889019e-02 -4.42582399e-01 3.74489278e-01 -4.46302742e-01 2.55954862e-01 5.92095852e-01 2.90520698e-01 -4.20373291e-01 8.92709009e-03 -8.13361164e-03 7.81458676e-01 -1.32687027e-02 4.30485487e-01 -8.59913826e-01 5.01245797e-01 4.24911320e-01 -1.39987752e-01 6.24021053e-01 -5.17732978e-01 7.04122603e-01 2.53853530e-01 -9.11088049e-01 -1.34502721e+00 -1.05837202e+00 -6.85726941e-01 4.14503008e-01 -1.10835075e-01 2.72468418e-01 -1.30196679e+00 -2.55501032e-01 -4.91258085e-01 1.48761839e-01 -7.72909999e-01 6.02759242e-01 -8.12089443e-01 -1.00573468e+00 5.21692038e-01 6.63820386e-01 8.40565741e-01 -1.27827466e+00 -1.58891428e+00 3.51777554e-01 3.31304938e-01 -1.07846534e+00 1.53617278e-01 5.05741298e-01 -1.62726080e+00 -1.09246433e+00 -9.57976401e-01 -1.12716174e+00 7.68940806e-01 -1.82781801e-01 1.15825522e+00 1.10462189e-01 -7.87866712e-01 2.76330501e-01 1.50957089e-02 -4.69065130e-01 -1.97684005e-01 6.87135905e-02 -1.78229481e-01 -9.15898204e-01 5.12327552e-01 -4.38711196e-01 -8.01667809e-01 -3.55166256e-01 -1.08963621e+00 1.85297489e-01 3.60608429e-01 6.10448062e-01 1.08651030e+00 1.86906546e-01 1.91001654e-01 -1.05986226e+00 6.63940310e-01 -1.40227601e-01 -4.58482206e-01 -3.71418208e-01 -5.42566001e-01 -1.80836514e-01 3.67808759e-01 -1.09509520e-01 -7.41817772e-01 3.03038538e-01 -7.75484085e-01 -3.14676136e-01 -5.48237503e-01 1.74782664e-01 5.28715789e-01 -3.33847851e-01 7.04024613e-01 -5.30153476e-02 3.68175834e-01 -4.02034223e-01 -5.21407090e-02 6.42195821e-01 8.47473562e-01 -2.03767821e-01 -1.52043356e-02 8.33422005e-01 3.02364886e-01 -9.77006733e-01 -3.60344559e-01 -4.90186781e-01 -1.36512399e+00 -3.72036397e-01 1.57861769e+00 -6.47539735e-01 -4.51815307e-01 4.69141930e-01 -9.70539689e-01 -4.91842538e-01 -3.54660720e-01 5.41462243e-01 -6.78269506e-01 2.72190988e-01 -1.07948923e+00 -2.14413568e-01 -1.08713889e+00 -1.61447012e+00 9.64118958e-01 3.27416718e-01 -1.18134990e-01 -1.17764056e+00 -1.54926226e-01 3.95181388e-01 3.13702434e-01 6.74340546e-01 8.80980611e-01 -6.71109736e-01 1.19924713e-02 -2.85719156e-01 -4.30626333e-01 2.86677033e-01 -5.20216599e-02 -3.39320153e-01 -8.57214630e-01 -1.04262389e-01 4.38852251e-01 -3.61866534e-01 6.31465077e-01 1.04410863e+00 1.55442274e+00 2.81855881e-01 -1.70661345e-01 4.80639458e-01 1.75423872e+00 3.93825799e-01 5.24749756e-01 5.57591140e-01 8.45563829e-01 1.87435865e-01 6.28522784e-02 4.12281156e-02 3.32794897e-02 1.16645694e-01 3.99334013e-01 -6.68447196e-01 -2.07909584e-01 5.95124900e-01 -5.11960864e-01 6.43369555e-01 -2.45402023e-01 4.34891433e-01 -1.37248480e+00 3.88851643e-01 -1.29183257e+00 -5.12241840e-01 -5.47838211e-01 1.76351678e+00 6.25689566e-01 4.39823657e-01 3.18795204e-01 4.48113889e-01 7.06983924e-01 -2.09679097e-01 -4.07180518e-01 -1.03197944e+00 4.12302822e-01 1.03738403e+00 7.22982705e-01 5.46185851e-01 -1.13177419e+00 7.82423437e-01 6.47524405e+00 8.86184990e-01 -1.50672817e+00 5.63976467e-01 1.10356700e+00 -1.64741725e-01 4.35527295e-01 -8.50055814e-01 -3.89815867e-01 1.02118798e-01 9.42878067e-01 2.57065266e-01 -2.85766661e-01 8.60511243e-01 1.70166686e-01 -6.20170057e-01 -7.31835783e-01 1.08935273e+00 -8.58722106e-02 -1.46148980e+00 -3.31213959e-02 4.53104563e-02 5.99406123e-01 4.79499876e-01 -1.86053514e-01 1.14389934e-01 3.57965901e-02 -1.50296617e+00 3.49990368e-01 -7.74091482e-02 9.20738518e-01 -1.02973580e+00 8.01053703e-01 4.68010455e-01 -8.33794296e-01 4.11183685e-01 -5.27948737e-01 -1.36898756e-01 1.37823716e-01 4.38111991e-01 -1.31425107e+00 2.53933340e-01 9.22653258e-01 -5.21201566e-02 -3.34405839e-01 9.09849048e-01 1.89930201e-01 3.73492718e-01 -6.32664740e-01 2.26503059e-01 9.15467739e-01 -1.83579549e-01 9.88972336e-02 1.22798765e+00 8.50502402e-02 3.87225211e-01 1.65388182e-01 3.43733966e-01 7.92841911e-02 4.80722278e-01 -4.81945038e-01 5.77240109e-01 -1.75096318e-01 1.17406249e+00 -1.82681823e+00 -7.80035317e-01 -2.13971049e-01 8.72261703e-01 1.50718942e-01 -2.75440603e-01 -4.90341932e-01 -3.36640999e-02 -2.08533123e-01 2.87194818e-01 2.67693669e-01 -2.90878147e-01 -9.52039123e-01 -4.51689214e-01 -5.19931674e-01 -2.18916073e-01 5.35636783e-01 -8.07303429e-01 -5.64978421e-01 1.08742440e+00 1.24130450e-01 -7.09269106e-01 -1.27064303e-01 -9.84948218e-01 -5.56764901e-01 7.49769509e-01 -1.12432516e+00 -8.96211505e-01 -2.02842981e-01 6.66108072e-01 6.09598696e-01 -3.84405628e-02 5.87918162e-01 2.52536029e-01 -2.52933979e-01 -1.22974195e-01 4.75777388e-02 4.66286719e-01 -3.71490382e-02 -1.48533940e+00 2.99910456e-01 6.79165483e-01 -1.93439394e-01 1.19023986e-01 3.51993829e-01 -8.65166187e-01 -7.37982333e-01 -1.11123300e+00 2.82140970e-01 9.33783203e-02 2.97590405e-01 2.01007389e-02 -8.60410154e-01 7.14107633e-01 4.91884708e-01 1.67252988e-01 7.14849055e-01 -6.21699691e-01 4.25055563e-01 5.32048523e-01 -1.83607185e+00 1.02665856e-01 4.43412095e-01 6.46140147e-03 -8.42541635e-01 5.02243400e-01 3.56393516e-01 -1.10013735e+00 -1.28037560e+00 4.53301847e-01 2.63321877e-01 -1.01195991e+00 1.06667662e+00 -2.31225133e-01 5.43679655e-01 -4.15406749e-02 2.19447866e-01 -8.38718414e-01 -9.40814540e-02 2.19540477e-01 -6.67882711e-03 1.92986518e-01 8.54789838e-02 -2.18847036e-01 1.12354875e+00 4.86050755e-01 -2.57301122e-01 -1.08455420e+00 -1.31426990e+00 -3.17079127e-01 4.37723100e-01 -4.73556668e-01 3.55178833e-01 7.37800002e-01 -4.66551572e-01 2.07751438e-01 2.24500284e-01 -1.08574562e-01 9.00270462e-01 2.66376641e-02 1.42940611e-01 -1.22410417e+00 2.00962335e-01 -6.59239709e-01 -7.95865059e-01 -5.73877037e-01 -2.19772175e-01 -1.00648010e+00 -3.62850726e-01 -2.06001592e+00 -3.26459780e-02 -2.82254279e-01 1.27192840e-01 4.10574138e-01 2.81491697e-01 9.96924937e-01 -1.34424880e-01 3.00185621e-01 3.63689549e-02 -4.35704328e-02 1.76321757e+00 6.53579552e-03 -3.01532209e-01 -2.00328112e-01 -5.48364259e-02 9.37978387e-01 1.23628414e+00 -2.66050190e-01 -4.60563868e-01 -4.86388296e-01 -5.77524185e-01 1.92346498e-01 1.06255956e-01 -1.42899227e+00 1.19600378e-01 5.79897881e-01 1.04776895e+00 -1.29534113e+00 1.29426047e-01 -9.37593460e-01 -1.39139175e-01 1.01328218e+00 4.36649658e-02 1.62987590e-01 3.90629619e-01 -1.68177605e-01 -2.87118368e-02 -4.75156724e-01 1.31938744e+00 -8.50572109e-01 -8.56155992e-01 5.99583328e-01 -7.83103526e-01 -3.26785147e-01 1.30340862e+00 -7.68117785e-01 2.00023755e-01 2.60267228e-01 -1.18315208e+00 1.43834576e-02 2.74836063e-01 -6.40046373e-02 6.83028758e-01 -1.04181266e+00 -4.44359839e-01 1.80668324e-01 -6.58134878e-01 5.87472379e-01 3.06051165e-01 7.86060214e-01 -1.60355127e+00 6.97279274e-01 -9.13483918e-01 -1.08481061e+00 -1.37173867e+00 3.21001917e-01 7.86265194e-01 -2.23143250e-01 -1.32805765e+00 7.49010205e-01 -1.62957877e-01 -3.67663383e-01 5.88882267e-02 -6.04243934e-01 -8.42283905e-01 -3.23924832e-02 5.13613462e-01 2.02406675e-01 5.40557563e-01 -7.64717400e-01 -2.57871866e-01 4.73657608e-01 -8.99605528e-02 -7.15619251e-02 1.51262689e+00 1.15411356e-01 -1.90093592e-01 3.27790171e-01 1.45044732e+00 -8.08135986e-01 -8.57218206e-01 1.36088297e-01 -4.66432981e-02 6.54974431e-02 7.74398386e-01 -6.17927551e-01 -1.63251173e+00 1.10572839e+00 1.31695294e+00 1.09507389e-01 1.01129258e+00 -5.91110252e-02 9.18676257e-01 1.45012349e-01 3.86612892e-01 -1.15663588e+00 -2.05739617e-01 5.11533558e-01 3.30388874e-01 -1.26376033e+00 2.44976953e-01 -2.70272374e-01 -3.03050071e-01 1.52508271e+00 7.60537028e-01 -6.67841494e-01 8.71230125e-01 5.92112124e-01 2.71317065e-01 -7.92533100e-01 1.96341276e-01 -1.00416183e-01 1.80664569e-01 7.52552092e-01 6.79510236e-01 2.82597274e-01 -6.89523578e-01 1.03258498e-01 -3.97099435e-01 2.29939535e-01 7.82645404e-01 1.17558396e+00 -6.63722456e-01 -6.50975645e-01 -6.32440269e-01 9.56799805e-01 -9.63575363e-01 2.64845472e-02 5.73501401e-02 1.04444993e+00 2.10480526e-01 2.01692984e-01 7.14510560e-01 1.64361030e-01 -4.01623324e-02 -8.36197808e-02 8.32324445e-01 -5.88438869e-01 -9.03445363e-01 2.33822957e-01 -2.27206245e-01 -4.17896420e-01 -4.60874975e-01 -3.64944607e-01 -1.99104905e+00 -1.40045926e-01 2.93545425e-01 -1.96892601e-02 9.95683551e-01 9.56356227e-01 -1.89339638e-01 8.48253667e-01 -4.50790823e-02 -1.28341210e+00 2.55359918e-01 -1.00182045e+00 -6.91102624e-01 1.68139219e-01 8.81404206e-02 -5.74621975e-01 3.93052310e-01 7.62290433e-02]
[14.36684513092041, -2.489110231399536]
21f5685f-329b-4352-8e2e-7b698dc5325f
mortality-prediction-with-adaptive-feature
2301.07107
null
https://arxiv.org/abs/2301.07107v2
https://arxiv.org/pdf/2301.07107v2.pdf
Mortality Prediction with Adaptive Feature Importance Recalibration for Peritoneal Dialysis Patients: a deep-learning-based study on a real-world longitudinal follow-up dataset
Objective: Peritoneal Dialysis (PD) is one of the most widely used life-supporting therapies for patients with End-Stage Renal Disease (ESRD). Predicting mortality risk and identifying modifiable risk factors based on the Electronic Medical Records (EMR) collected along with the follow-up visits are of great importance for personalized medicine and early intervention. Here, our objective is to develop a deep learning model for a real-time, individualized, and interpretable mortality prediction model - AICare. Method and Materials: Our proposed model consists of a multi-channel feature extraction module and an adaptive feature importance recalibration module. AICare explicitly identifies the key features that strongly indicate the outcome prediction for each patient to build the health status embedding individually. This study has collected 13,091 clinical follow-up visits and demographic data of 656 PD patients. To verify the application universality, this study has also collected 4,789 visits of 1,363 hemodialysis dialysis (HD) as an additional experiment dataset to test the prediction performance, which will be discussed in the Appendix. Results: 1) Experiment results show that AICare achieves 81.6%/74.3% AUROC and 47.2%/32.5% AUPRC for the 1-year mortality prediction task on PD/HD dataset respectively, which outperforms the state-of-the-art comparative deep learning models. 2) This study first provides a comprehensive elucidation of the relationship between the causes of mortality in patients with PD and clinical features based on an end-to-end deep learning model. 3) This study first reveals the pattern of variation in the importance of each feature in the mortality prediction based on built-in interpretability. 4) We develop a practical AI-Doctor interaction system to visualize the trajectory of patients' health status and risk indicators.
['Tao Wang', 'Wenjie Ruan', 'Xinju Zhao', 'Wen Tang', 'Yasha Wang', 'Xinyu Ma', 'Zhihao Yu', 'Xianfeng Jiao', 'Junyi Gao', 'Chaohe Zhang', 'Liantao Ma']
2023-01-17
null
null
null
null
['mortality-prediction']
['medical']
[-3.50768059e-01 -1.25712365e-01 -9.04838145e-02 -4.43556279e-01 -4.26399767e-01 8.68082121e-02 5.48272058e-02 4.94993895e-01 -1.68152526e-01 1.07315874e+00 5.62659621e-01 -2.93323994e-01 -6.17102802e-01 -8.92320573e-01 -1.91575646e-01 -6.90797508e-01 -7.84423590e-01 7.98013628e-01 -6.74632192e-01 6.07297681e-02 -2.98870821e-02 4.91492301e-01 -1.09615529e+00 2.75132090e-01 1.21718776e+00 9.42340434e-01 6.61109909e-02 4.93735194e-01 3.19661088e-02 7.74190307e-01 -3.94284904e-01 1.49435863e-01 -6.37742206e-02 -4.34178412e-01 -6.85474753e-01 -3.79827946e-01 -9.10390317e-02 -6.19304955e-01 -5.14482915e-01 3.94167662e-01 1.08051813e+00 -3.59550357e-01 9.36015010e-01 -9.89929199e-01 -7.95461476e-01 6.61422551e-01 -1.47249073e-01 1.84678555e-01 5.17246015e-02 2.93575644e-01 6.03903472e-01 -6.60729289e-01 1.73166618e-01 1.10945892e+00 9.81505096e-01 6.23364508e-01 -1.16359949e+00 -4.46117669e-01 -3.07455182e-01 3.09373587e-01 -1.20238066e+00 1.14834858e-02 3.88390362e-01 -7.87907898e-01 6.14335954e-01 3.87237847e-01 6.78621888e-01 1.03156233e+00 5.30832052e-01 4.89453077e-01 9.35272217e-01 -1.54442759e-02 -2.42954064e-02 -3.88124585e-02 4.39475149e-01 8.94603193e-01 2.91124433e-01 4.96194243e-01 -1.11478113e-01 -5.65805554e-01 9.76200879e-01 5.41341841e-01 -4.83401299e-01 -9.08247158e-02 -1.46670210e+00 7.53666162e-01 5.14527261e-01 1.86015233e-01 -8.43646705e-01 -6.21066168e-02 5.15150130e-01 4.92859542e-01 9.81953666e-02 6.71450347e-02 -9.10125732e-01 6.82793856e-02 -3.42100859e-01 5.29471003e-02 8.29170585e-01 5.56874752e-01 3.30529958e-01 -1.31449148e-01 -7.66286910e-01 9.23033059e-01 4.44026351e-01 6.90269768e-01 6.16871059e-01 -7.66691327e-01 8.98486525e-02 9.59776103e-01 2.16809049e-01 -6.82399094e-01 -9.87977326e-01 -5.81651270e-01 -1.56940150e+00 8.64049122e-02 5.53382821e-02 -4.88595366e-01 -9.07281697e-01 1.49606109e+00 1.09848574e-01 -2.44880207e-02 2.03923985e-01 7.03968704e-01 9.86657202e-01 3.49865675e-01 6.15202606e-01 -3.89508575e-01 1.54263878e+00 -5.02446115e-01 -6.81484520e-01 3.36514682e-01 8.87991846e-01 -5.12872338e-02 9.24572408e-01 7.48916641e-02 -6.40906096e-01 -3.52548122e-01 -6.50147796e-01 1.06066532e-01 -1.83977425e-01 6.71999216e-01 6.16050422e-01 2.61743546e-01 -6.23364151e-01 1.02426326e+00 -8.64824355e-01 -6.50273919e-01 6.15588248e-01 5.39714336e-01 -4.17006999e-01 1.20282583e-01 -1.50942981e+00 1.11490476e+00 2.86390692e-01 3.89441196e-03 -9.59793329e-01 -1.13576078e+00 -4.46510762e-01 2.32052356e-01 -5.33192217e-01 -1.49270582e+00 6.72146142e-01 -5.56059897e-01 -1.14318085e+00 7.58963525e-01 2.94085755e-03 -4.78043288e-01 5.73861480e-01 -5.42636395e-01 -4.12391633e-01 -3.37084234e-02 -2.54098549e-02 1.47144854e-01 2.17400327e-01 -8.77991974e-01 -6.31633997e-01 -8.26201797e-01 -6.39043212e-01 1.96007594e-01 -3.84594589e-01 -2.59518176e-01 5.41418344e-02 -6.54156506e-01 -1.56387180e-01 -6.81409717e-01 -3.61705750e-01 6.87550604e-02 -3.05971086e-01 -3.38187814e-01 6.00257695e-01 -9.91559684e-01 1.53838003e+00 -2.00217533e+00 1.22684479e-01 2.00627483e-02 6.61957085e-01 1.81196302e-01 3.59711558e-01 4.64395642e-01 -2.88491368e-01 5.32896668e-02 -3.22903037e-01 -5.46977036e-02 -9.68924910e-02 4.21894416e-02 1.04062058e-01 4.38931465e-01 1.66858688e-01 1.04547799e+00 -6.49968505e-01 -4.37422276e-01 6.38216555e-01 6.60379767e-01 -3.15549821e-01 5.64996123e-01 2.73987561e-01 8.14941585e-01 -6.79909766e-01 5.21728814e-01 4.93025213e-01 -6.75568283e-01 2.75462568e-01 -3.03586125e-01 1.11487314e-01 -5.70492446e-02 -7.11019933e-01 1.30607092e+00 -1.80404678e-01 1.75369337e-01 -5.19751728e-01 -8.63804877e-01 1.05006468e+00 4.81840700e-01 9.75389183e-01 -5.18846750e-01 2.11221114e-01 8.58516768e-02 -1.52711660e-01 -8.39794159e-01 -4.94236410e-01 -3.75389636e-01 -1.56829774e-01 1.53419256e-01 -2.23818198e-01 9.06694055e-01 -4.55931872e-01 -1.54095322e-01 1.38159168e+00 -2.55831063e-01 5.94510317e-01 -5.05162835e-01 6.88655078e-01 -2.01601893e-01 7.11877525e-01 7.19754636e-01 -5.72833121e-01 5.78607142e-01 3.49457562e-01 -1.08514822e+00 -7.44738758e-01 -1.24041784e+00 -7.76831210e-01 4.93177801e-01 -1.36578232e-01 -2.33406238e-02 -1.27685755e-01 -7.02327788e-01 7.59818852e-01 4.82256800e-01 -9.30749059e-01 -3.14028680e-01 -3.26236278e-01 -1.28053868e+00 7.04713702e-01 7.22939491e-01 6.42242432e-01 -1.07042480e+00 -4.46513325e-01 4.56051171e-01 -1.84067875e-01 -1.42258838e-01 1.47278786e-01 2.95484930e-01 -1.31868351e+00 -1.34260035e+00 -8.17737818e-01 -7.55554378e-01 4.37510639e-01 -4.97693598e-01 1.17020261e+00 2.23247021e-01 -3.17081034e-01 2.58327723e-01 3.95913757e-02 -2.21114904e-01 -3.06539863e-01 -7.71086738e-02 3.15836281e-01 -3.48108888e-01 7.81551123e-01 -6.04388654e-01 -1.20852697e+00 1.82049230e-01 -4.15115505e-01 -1.36733735e-02 8.14373672e-01 9.64542925e-01 6.20923221e-01 -1.52822345e-01 1.13885331e+00 -9.40143764e-01 5.68971157e-01 -9.60471928e-01 -1.91162135e-02 2.75828332e-01 -1.28375554e+00 1.21700622e-01 5.94086945e-01 -1.11619316e-01 -7.74056971e-01 6.79381564e-02 -1.60495564e-02 -2.92600513e-01 -3.92030984e-01 6.25501752e-01 -6.04626583e-03 3.74271005e-01 6.80166543e-01 1.89283744e-01 9.32909325e-02 -8.01429033e-01 -2.02047899e-02 1.02318776e+00 4.36977446e-01 -5.09162009e-01 2.80434459e-01 8.27520043e-02 9.12721381e-02 -5.50266504e-01 -3.98025721e-01 -2.70052552e-01 -5.05310059e-01 1.16484083e-01 8.60252202e-01 -1.23597646e+00 -1.14935267e+00 6.51567817e-01 -8.12515855e-01 -2.50457764e-01 -7.74138570e-02 6.77660704e-01 -5.14365792e-01 4.76194263e-01 -9.29511845e-01 -6.49392903e-01 -1.16174722e+00 -7.03802109e-01 7.65313804e-01 2.10410520e-01 -3.81666780e-01 -1.13896382e+00 1.53380275e-01 9.81159732e-02 6.65431261e-01 6.44776285e-01 1.56785560e+00 -6.48286521e-01 -1.67359188e-01 -1.41561264e-02 -4.25235271e-01 2.31261313e-01 5.44498146e-01 -3.88246715e-01 -7.27959275e-01 -2.17676044e-01 -3.30799781e-02 2.48015188e-02 7.96266139e-01 6.76141441e-01 1.30775905e+00 -2.28077397e-01 -5.72913826e-01 5.93592763e-01 1.59190524e+00 3.32696080e-01 8.90504062e-01 3.33256543e-01 7.62645066e-01 2.28141606e-01 2.80591935e-01 9.54564154e-01 7.38515913e-01 2.12649956e-01 3.27123255e-01 -1.55065462e-01 2.01837420e-01 -4.40459326e-02 8.37734267e-02 6.53135478e-01 -1.17539309e-01 6.77865595e-02 -1.20296931e+00 5.54032326e-01 -1.92909479e+00 -7.33518302e-01 -6.29577994e-01 2.35486555e+00 8.62553060e-01 -4.15552467e-01 -6.79217726e-02 -1.74413875e-01 6.20685637e-01 -3.97125661e-01 -7.47789979e-01 8.82037804e-02 -5.59770018e-02 8.73966590e-02 4.34731066e-01 2.37285152e-01 -1.24929428e+00 1.28381729e-01 5.86175585e+00 -1.12262920e-01 -9.50316429e-01 -2.97198743e-01 9.35777545e-01 8.22066963e-02 -1.63297374e-02 -3.92473102e-01 -4.29628581e-01 8.84119749e-01 1.10874081e+00 -2.35653475e-01 1.11342855e-01 6.23790383e-01 5.53379476e-01 3.01393002e-01 -1.44865346e+00 1.06910741e+00 -3.78209561e-01 -1.10750318e+00 1.90171301e-01 -1.44281521e-01 3.87808353e-01 2.25360975e-01 -2.97248602e-01 5.90076745e-01 1.66959032e-01 -1.31405449e+00 -2.09596604e-01 1.18514645e+00 1.01039743e+00 -5.17270029e-01 1.11512482e+00 1.62832960e-01 -8.24834466e-01 -4.88291115e-01 -4.95115370e-02 1.27819385e-02 1.27749294e-02 7.92213619e-01 -6.55944824e-01 7.00841308e-01 1.02619457e+00 9.47035849e-01 -3.90846670e-01 1.04868758e+00 1.44355163e-01 6.13712132e-01 -1.27136394e-01 3.57029796e-01 -3.68545294e-01 -5.04024886e-03 1.58539951e-01 1.05197883e+00 4.39815909e-01 4.69413787e-01 -1.39605194e-01 9.02791739e-01 -6.46127090e-02 9.11295861e-02 -5.34439504e-01 3.69852073e-02 3.14104706e-01 8.68357301e-01 1.39944986e-01 -4.13223714e-01 -2.95919329e-01 9.64183927e-01 2.14763179e-01 2.29661703e-01 -8.02300453e-01 -2.63246328e-01 8.49062979e-01 9.09668952e-02 -4.89295982e-02 1.90042228e-01 -7.52845287e-01 -1.17984724e+00 -2.61484206e-01 -6.79930747e-01 8.74635935e-01 -4.69860137e-01 -1.97562969e+00 2.96431541e-01 -4.34142530e-01 -1.12912047e+00 2.21950576e-01 -4.18759853e-01 -6.89926863e-01 1.41311109e+00 -1.59046197e+00 -8.46529543e-01 -7.45056629e-01 5.45765936e-01 1.32483065e-01 -1.84048817e-01 1.37742007e+00 5.41940451e-01 -9.00860727e-01 4.96775657e-01 5.44785380e-01 2.49753743e-01 8.48255515e-01 -1.31456780e+00 -1.99451908e-01 -2.43791286e-02 -1.08738875e+00 8.29849660e-01 3.08894396e-01 -9.20213580e-01 -1.39858603e+00 -1.36135256e+00 1.19381428e+00 -7.57357419e-01 1.52071550e-01 4.45905894e-01 -1.04970825e+00 6.70255125e-01 -2.05122024e-01 -1.29802912e-01 9.72723961e-01 2.01224461e-01 3.45522702e-01 -1.65099144e-01 -1.34917414e+00 2.56677747e-01 7.74707258e-01 -2.35782892e-01 -7.62264132e-01 1.71573803e-01 4.58160192e-01 -1.57927543e-01 -1.73766077e+00 9.79081750e-01 7.45204151e-01 -6.81856036e-01 1.08862484e+00 -9.47859585e-01 8.41782987e-01 -3.03809285e-01 -4.61432710e-02 -1.01067412e+00 -9.23684537e-01 -9.38476622e-02 -5.10536969e-01 8.91870916e-01 1.65243372e-01 -7.20074236e-01 5.08797228e-01 9.58367825e-01 -2.77073700e-02 -1.10130847e+00 -6.62900984e-01 -1.67813897e-01 2.65680552e-01 2.26525828e-01 6.60688162e-01 1.35791075e+00 -4.19808365e-02 3.93721789e-01 -3.94200951e-01 4.12696868e-01 6.92409754e-01 1.60154998e-01 4.96311873e-01 -1.77383292e+00 -5.24624134e-04 -2.36736029e-01 -4.32361066e-01 -4.78469998e-01 -3.88437808e-01 -8.83695543e-01 -5.23494184e-01 -2.03183198e+00 4.88433033e-01 -7.33422756e-01 -1.11276507e+00 6.55433774e-01 -5.48565686e-01 -5.29283404e-01 -3.49236935e-01 5.21023989e-01 -1.70724526e-01 9.98511314e-01 1.00037491e+00 -5.89095242e-02 -4.38625097e-01 -5.81940031e-03 -7.20556676e-01 2.46887326e-01 9.05083656e-01 -3.30960035e-01 -1.93007082e-01 -2.50082403e-01 -2.57454723e-01 4.54886854e-01 5.18454194e-01 -9.84550178e-01 -3.50288749e-01 -1.95522308e-01 1.03540158e+00 -4.50929701e-01 -2.63987094e-01 -1.01698124e+00 4.58174258e-01 9.48077440e-01 -1.90403372e-01 -1.23181000e-01 1.76064461e-01 7.19259441e-01 1.49829134e-01 4.21591192e-01 7.76174068e-01 -4.67246324e-02 -5.86764574e-01 5.76122105e-01 -2.18646988e-01 -2.28853807e-01 1.07466316e+00 8.71162638e-02 -4.09822911e-01 2.17892900e-02 -9.74334419e-01 6.84893429e-01 2.12202817e-01 3.67257714e-01 6.40249789e-01 -1.58177567e+00 -1.01163530e+00 3.45377266e-01 2.21276388e-01 -7.66477287e-02 6.47087932e-01 1.00731397e+00 -7.29001701e-01 4.07826781e-01 -5.06205678e-01 -6.03054941e-01 -9.49452758e-01 4.33655888e-01 7.55580485e-01 -3.59639496e-01 -1.06885445e+00 3.51422220e-01 3.15619797e-01 -3.46607476e-01 3.86307061e-01 -1.91463158e-01 -5.36957145e-01 -1.32321164e-01 4.09431040e-01 5.47078729e-01 -1.04500517e-01 -5.38459830e-02 -4.97342080e-01 2.29846224e-01 3.95897776e-03 7.56702960e-01 1.61919904e+00 -3.22400451e-01 -1.64454103e-01 5.11245668e-01 1.02934170e+00 -6.86805129e-01 -9.74751174e-01 -1.62819803e-01 -6.39917478e-02 -6.73231408e-02 1.08160833e-02 -1.42369950e+00 -1.04016793e+00 7.76684701e-01 1.48116457e+00 -2.11804211e-01 1.24315369e+00 -1.55816302e-01 9.71048474e-01 2.31312975e-01 2.67131060e-01 -6.45234168e-01 -4.31146264e-01 3.88416916e-01 8.33289802e-01 -1.44743824e+00 -1.24626510e-01 2.27874026e-01 -6.36886418e-01 1.14118862e+00 5.85685611e-01 5.86503781e-02 9.71816123e-01 -2.65917242e-01 3.04025948e-01 -4.37169522e-01 -7.00516045e-01 2.75060624e-01 -2.29557436e-02 4.38111156e-01 5.71044803e-01 5.75430632e-01 -6.55564964e-01 1.27519429e+00 2.49443755e-01 5.68435609e-01 -1.85358375e-02 7.57189751e-01 -4.63490844e-01 -8.73282015e-01 -3.89505476e-02 1.09775674e+00 -4.74149317e-01 -1.15223013e-01 -1.58095673e-01 3.34983289e-01 2.05455776e-02 6.03675902e-01 -3.10831875e-01 -5.45216799e-01 4.49937493e-01 5.25482774e-01 4.13443521e-02 -2.28558347e-01 -6.48638070e-01 -5.60155153e-01 -8.62072036e-02 -4.04023349e-01 -5.10821119e-02 -4.00288492e-01 -1.61964273e+00 -4.63202566e-01 1.13186963e-01 1.14079723e-02 3.45327377e-01 7.28383899e-01 9.65418577e-01 8.04372847e-01 7.62592256e-01 -2.45711133e-01 -7.33322740e-01 -1.15429056e+00 -7.91731656e-01 6.39006376e-01 5.06318510e-01 -6.43067122e-01 -4.00417626e-01 1.03316065e-02]
[7.952764511108398, 6.031208038330078]
4608ffb3-8dd1-4983-82de-58ba2947ce16
adversarial-self-supervised-scene-flow
2011.00551
null
https://arxiv.org/abs/2011.00551v1
https://arxiv.org/pdf/2011.00551v1.pdf
Adversarial Self-Supervised Scene Flow Estimation
This work proposes a metric learning approach for self-supervised scene flow estimation. Scene flow estimation is the task of estimating 3D flow vectors for consecutive 3D point clouds. Such flow vectors are fruitful, \eg for recognizing actions, or avoiding collisions. Training a neural network via supervised learning for scene flow is impractical, as this requires manual annotations for each 3D point at each new timestamp for each scene. To that end, we seek for a self-supervised approach, where a network learns a latent metric to distinguish between points translated by flow estimations and the target point cloud. Our adversarial metric learning includes a multi-scale triplet loss on sequences of two-point clouds as well as a cycle consistency loss. Furthermore, we outline a benchmark for self-supervised scene flow estimation: the Scene Flow Sandbox. The benchmark consists of five datasets designed to study individual aspects of flow estimation in progressive order of complexity, from a moving object to real-world scenes. Experimental evaluation on the benchmark shows that our approach obtains state-of-the-art self-supervised scene flow results, outperforming recent neighbor-based approaches. We use our proposed benchmark to expose shortcomings and draw insights on various training setups. We find that our setup captures motion coherence and preserves local geometries. Dealing with occlusions, on the other hand, is still an open challenge.
['Pascal Mettes', 'Olaf Booij', 'Joris van Vugt', 'Victor Zuanazzi']
2020-11-01
null
null
null
null
['scene-flow-estimation']
['computer-vision']
[ 1.61434039e-01 -3.35450321e-01 -2.56951541e-01 -2.37711310e-01 -6.90754712e-01 -6.96052730e-01 5.39527476e-01 -8.86083916e-02 -3.82986397e-01 5.92621565e-01 3.67500447e-02 -9.96819884e-02 4.12986390e-02 -7.61123896e-01 -7.96044350e-01 -6.05881035e-01 -4.32679474e-01 5.55979133e-01 4.39772993e-01 -5.89731373e-02 4.39530849e-01 9.45246756e-01 -1.29474795e+00 -1.39057692e-02 7.23685324e-01 9.08370912e-01 -4.39052470e-02 1.21011245e+00 -7.29822963e-02 1.27949417e+00 -5.51224291e-01 -3.44354808e-01 7.21137404e-01 -2.39923730e-01 -1.10124230e+00 1.13383994e-01 1.33287871e+00 -5.94468415e-01 -7.84407556e-01 7.80250132e-01 2.08681241e-01 3.99187833e-01 7.46531367e-01 -1.62476528e+00 -5.45216026e-03 -8.40626508e-02 -4.37373936e-01 4.48681325e-01 4.35347944e-01 5.15407741e-01 8.71820927e-01 -7.28771687e-01 9.48586881e-01 1.18461692e+00 7.97196448e-01 3.58572185e-01 -1.20519733e+00 -4.39301491e-01 8.45854953e-02 2.75751710e-01 -1.16277289e+00 -4.66566026e-01 1.16814435e+00 -7.90166855e-01 8.60663831e-01 1.36095196e-01 5.97775578e-01 9.31953847e-01 1.62573308e-01 7.36184478e-01 5.63491166e-01 -6.17438704e-02 2.55216658e-01 -3.98415700e-02 -2.08561838e-01 9.15848792e-01 7.99360275e-02 3.44443053e-01 -6.36641085e-01 -4.94687520e-02 9.29961145e-01 -1.66559011e-01 -3.24771345e-01 -9.47331607e-01 -1.47801149e+00 7.52010167e-01 6.40098572e-01 -9.24018119e-03 1.15685929e-02 5.89744091e-01 5.50212145e-01 3.05299282e-01 6.98949754e-01 3.67765754e-01 -3.66122633e-01 -2.97586828e-01 -1.00720525e+00 2.79286802e-01 1.02293921e+00 1.06318188e+00 1.09555447e+00 2.81309217e-01 -1.72644982e-03 3.12797688e-02 3.69629785e-02 5.66308737e-01 1.09113501e-02 -1.65210247e+00 7.40054548e-01 3.49260896e-01 2.28518516e-01 -1.42079294e+00 -2.47660190e-01 -8.75102729e-02 -9.80032206e-01 5.95988870e-01 8.18231344e-01 2.54646502e-02 -4.52672631e-01 1.64561141e+00 5.70030808e-01 8.96398187e-01 1.00742979e-02 8.81253541e-01 5.38949430e-01 6.81952894e-01 -2.42061108e-01 -1.69100255e-01 6.03163660e-01 -1.14134395e+00 -2.46149406e-01 -1.12727515e-01 8.46331537e-01 -7.57963419e-01 9.39563692e-01 1.21346250e-01 -1.17317915e+00 -7.02112436e-01 -8.91601980e-01 -1.76769897e-01 -2.48242542e-01 -1.81856990e-01 6.39907718e-01 5.63836813e-01 -9.81972396e-01 1.01553440e+00 -1.08856475e+00 -3.00006241e-01 6.22811019e-01 8.89357328e-02 -4.81559873e-01 1.78220160e-02 -8.05231214e-01 7.67401278e-01 1.29072666e-01 -7.16229901e-02 -8.44638288e-01 -1.27108383e+00 -1.22710717e+00 -2.44007900e-01 5.84952533e-02 -1.07036602e+00 8.85006368e-01 -6.64175808e-01 -1.30819750e+00 1.00435305e+00 -2.58110225e-01 -6.02559149e-01 1.09162092e+00 -3.88391435e-01 6.79053664e-02 5.25712788e-01 3.79372358e-01 8.06350350e-01 9.44189548e-01 -1.24864042e+00 -7.94506192e-01 -5.84227182e-02 2.20686272e-01 1.85399488e-01 -4.80855210e-03 -2.78775334e-01 -2.57323802e-01 -5.65852344e-01 -1.54235229e-01 -9.77739453e-01 -3.68086427e-01 5.79774857e-01 -4.76936221e-01 1.19148999e-01 1.26565158e+00 -1.94911093e-01 7.26802826e-01 -1.96092272e+00 2.29757950e-02 1.12105645e-02 2.88125575e-01 7.66752139e-02 5.10728806e-02 7.94250518e-02 -1.62991390e-01 -1.19745843e-01 -5.18982351e-01 -8.14721704e-01 -2.35846937e-01 2.28854045e-01 -5.89908719e-01 1.00327992e+00 3.61129105e-01 9.21569884e-01 -1.23491001e+00 -7.32718289e-01 9.18425143e-01 3.41966510e-01 -7.51340926e-01 4.23269987e-01 -7.64241070e-02 8.22882116e-01 -1.72670975e-01 4.12073076e-01 7.98485100e-01 -2.99824893e-01 -4.79167163e-01 -3.46598744e-01 -1.23325557e-01 1.16475582e-01 -1.36891377e+00 2.28835177e+00 -5.34892380e-01 1.04163837e+00 -4.17656124e-01 -9.35658395e-01 7.26393819e-01 -7.05384389e-02 1.05008793e+00 -2.59521931e-01 -1.54215237e-02 7.86362886e-02 -4.55122948e-01 -5.05098403e-01 5.36791861e-01 2.08426923e-01 6.87897578e-02 4.45017576e-01 1.79973282e-02 -5.82628787e-01 3.50873888e-01 3.36773992e-01 1.31719363e+00 4.48776066e-01 1.03568450e-01 -1.89889312e-01 6.97272241e-01 3.73587221e-01 5.19769192e-01 6.85207248e-01 -6.56540036e-01 8.70029628e-01 3.14624727e-01 -9.18855190e-01 -1.10742521e+00 -1.18839622e+00 2.62424275e-02 4.84333724e-01 5.56502640e-01 -2.18672529e-01 -3.90046746e-01 -8.99868548e-01 6.32002205e-02 4.16264296e-01 -5.61934173e-01 -6.71060979e-02 -1.11824822e+00 -1.79236755e-01 4.85150844e-01 5.22727668e-01 5.95007420e-01 -8.42981994e-01 -1.09152722e+00 1.58064485e-01 -2.94681370e-01 -1.60451579e+00 -8.30591440e-01 -9.51249674e-02 -9.34956849e-01 -1.40531421e+00 -5.81209421e-01 -6.52351737e-01 5.86780071e-01 4.88918632e-01 1.55327189e+00 -3.52402292e-02 -3.77582103e-01 7.73538530e-01 -1.39980182e-01 3.48965637e-02 -5.24928033e-01 6.45838231e-02 9.62188616e-02 1.15855195e-01 1.42168775e-02 -8.29148352e-01 -9.23874021e-01 4.44084436e-01 -6.81845725e-01 -1.63545609e-01 -6.29653549e-03 4.51102316e-01 4.94581550e-01 -1.74938917e-01 -3.11215632e-02 -7.24225700e-01 -1.37925878e-01 -3.05185497e-01 -6.71655595e-01 -5.61275193e-03 -1.54204845e-01 -6.62984252e-02 6.91362143e-01 -2.73515910e-01 -9.47296619e-01 4.88620520e-01 1.32463813e-01 -1.07994628e+00 -2.34203458e-01 -3.62294376e-01 -4.42955717e-02 -5.84148824e-01 8.54460895e-01 3.81133668e-02 -5.68964556e-02 6.32081032e-02 6.49924099e-01 -1.79737266e-02 9.80370224e-01 -5.68710864e-01 1.37653923e+00 1.13116622e+00 5.50404847e-01 -7.64642060e-01 -8.99026215e-01 -8.77331853e-01 -1.10976076e+00 -5.10157883e-01 9.81810153e-01 -7.90363431e-01 -8.09989631e-01 3.84856343e-01 -1.56188107e+00 -5.11697292e-01 -7.77808905e-01 5.37999094e-01 -1.21464634e+00 5.39582729e-01 -5.02615809e-01 -5.26335299e-01 -9.20161605e-02 -1.22574162e+00 1.25484252e+00 -1.37131691e-01 -2.31921107e-01 -1.49990869e+00 4.59209859e-01 2.03645840e-01 2.02529207e-01 7.45968223e-01 2.62445062e-01 -8.08826163e-02 -1.03826988e+00 9.32197571e-02 -1.84178919e-01 3.30442607e-01 2.56744504e-01 1.58177018e-01 -1.06356657e+00 -3.00857544e-01 1.39187619e-01 -3.97449911e-01 7.13799775e-01 4.76594478e-01 1.13668847e+00 -2.90219814e-01 -2.30962217e-01 1.30138242e+00 1.39133263e+00 -1.27477124e-01 6.34575009e-01 2.67797559e-01 1.28533423e+00 5.38642526e-01 8.20202291e-01 3.31859022e-01 3.81604195e-01 5.86399853e-01 6.69695377e-01 -1.89850718e-01 -3.85094464e-01 -3.37938249e-01 1.76223218e-01 5.14469504e-01 -2.18110569e-02 -2.20502168e-01 -8.82029414e-01 5.37089169e-01 -1.76169586e+00 -1.04342628e+00 -3.33755255e-01 2.18878889e+00 3.23600322e-01 1.98166817e-01 5.51656224e-02 2.67713219e-01 3.77761215e-01 5.95645607e-01 -7.66718388e-01 -6.54513687e-02 -1.12559251e-01 6.83019459e-02 7.34608352e-01 7.58132517e-01 -1.58082855e+00 1.10821819e+00 6.02363110e+00 4.79359508e-01 -1.12327349e+00 -3.04219313e-03 6.69643044e-01 -5.24240546e-02 5.49042672e-02 1.45795792e-01 -5.33942163e-01 2.85364777e-01 4.67268378e-01 -1.36638805e-01 9.66198444e-02 8.10332179e-01 2.74178416e-01 -1.90147106e-02 -1.42798913e+00 1.23968947e+00 1.38271466e-01 -1.76950431e+00 6.78956956e-02 -1.56790331e-01 9.63512182e-01 1.98009968e-01 -1.35367334e-01 -1.87251791e-01 3.27213526e-01 -7.24721909e-01 5.31322479e-01 4.89836633e-01 8.32903147e-01 -6.06562793e-01 2.22652271e-01 1.26800209e-01 -1.46417367e+00 3.70029926e-01 -1.93200707e-01 -5.71433827e-02 5.52470088e-01 4.84421849e-01 -5.51416218e-01 6.91793561e-01 6.95711315e-01 1.37858498e+00 -4.16945338e-01 1.18523598e+00 6.26191869e-02 3.67734581e-01 -3.93158436e-01 4.93080765e-01 3.39099795e-01 -3.22339386e-01 9.73482668e-01 1.25286984e+00 1.23011068e-01 -2.35932231e-01 3.04858476e-01 8.17804694e-01 7.49053955e-02 -8.69282484e-02 -1.06300318e+00 6.94905758e-01 1.24233261e-01 1.05395091e+00 -7.94263184e-01 -2.58911103e-01 -1.52902767e-01 1.17234218e+00 2.64327437e-01 4.01448816e-01 -8.02938998e-01 -2.19194129e-01 1.05324912e+00 5.22325598e-02 7.84270465e-02 -5.65256238e-01 -2.98824638e-01 -1.38607764e+00 8.24178830e-02 -2.40129635e-01 3.81724089e-01 -6.21214151e-01 -1.17672420e+00 3.96116436e-01 4.01336737e-02 -1.93553066e+00 -4.97165442e-01 -3.81568134e-01 -7.97118783e-01 4.43991780e-01 -1.71117198e+00 -8.94613922e-01 -7.25985467e-01 7.65407443e-01 6.97749615e-01 -2.77935304e-02 3.56307954e-01 3.61236781e-01 -2.19470173e-01 3.14490497e-01 -2.05738157e-01 3.15089405e-01 7.83772111e-01 -1.38568449e+00 7.46022165e-01 1.12712586e+00 5.05439878e-01 6.71855435e-02 6.65986538e-01 -3.95449638e-01 -1.33642352e+00 -1.44132364e+00 7.31770456e-01 -1.02448177e+00 6.84359252e-01 -2.33466059e-01 -8.04297149e-01 7.34551847e-01 -1.39854640e-01 6.92035198e-01 1.70580074e-01 -6.80103958e-01 -2.39901796e-01 -2.33414486e-01 -1.09555936e+00 5.41627526e-01 1.58920360e+00 -5.28385699e-01 -1.96986139e-01 5.96711755e-01 9.83130515e-01 -7.41972327e-01 -8.30788612e-01 3.70711654e-01 2.06481799e-01 -1.25982201e+00 1.37080288e+00 -5.87936521e-01 6.35506272e-01 -4.97140437e-01 -6.54177070e-02 -1.15427375e+00 7.01593757e-02 -1.08063114e+00 -4.91290480e-01 8.63415956e-01 -1.96340457e-01 -3.58484596e-01 1.35549462e+00 5.10637760e-01 -9.70396549e-02 -4.55647856e-01 -1.05355763e+00 -9.47924912e-01 7.02515244e-02 -7.72623420e-01 2.72695065e-01 1.14947665e+00 -5.92507064e-01 1.60483226e-01 -3.95369470e-01 2.53901035e-01 1.15573776e+00 1.46740183e-01 1.42191780e+00 -8.71928394e-01 2.35418882e-02 -5.25518477e-01 -9.28956747e-01 -1.52867615e+00 5.58385015e-01 -7.64355540e-01 -1.09854877e-01 -1.18420768e+00 -3.54212195e-01 -5.59592068e-01 2.29702190e-01 -4.03629839e-02 2.91919541e-02 3.77729535e-01 3.44818622e-01 3.30456287e-01 -6.58823848e-01 5.76431274e-01 1.36317337e+00 -4.08638239e-01 -4.03718829e-01 1.64623290e-01 5.07425703e-02 8.62847686e-01 6.34327173e-01 -5.37297428e-01 -6.41258121e-01 -5.30663311e-01 -1.11288130e-01 2.29963526e-01 7.53733337e-01 -1.35495579e+00 5.96573412e-01 -3.84594411e-01 1.18421726e-01 -8.20238352e-01 4.63975847e-01 -9.63338733e-01 -8.14077929e-02 5.76122880e-01 -2.06120819e-01 1.20293558e-01 6.74674958e-02 6.61135495e-01 -1.77608460e-01 6.75278679e-02 7.51408279e-01 -1.29298091e-01 -8.85324299e-01 8.49001825e-01 1.87863261e-01 5.10685980e-01 1.10645521e+00 -5.19504309e-01 -1.80114597e-01 -4.99919444e-01 -3.69054019e-01 1.81593329e-01 4.94413614e-01 4.79957908e-01 6.63942814e-01 -1.22101748e+00 -7.43513346e-01 1.32138088e-01 2.39813462e-01 6.16972625e-01 1.29513040e-01 5.17528653e-01 -1.10381317e+00 1.83317304e-01 -1.24627657e-01 -1.20755136e+00 -9.45682287e-01 4.59734380e-01 5.41084528e-01 -2.69788623e-01 -8.31509531e-01 6.92824960e-01 3.39176118e-01 -4.44770187e-01 3.11437845e-01 -4.86165434e-01 1.07733719e-01 -2.44577214e-01 3.24926257e-01 7.67666638e-01 -1.49741739e-01 -9.50719714e-01 -3.44218761e-01 9.92245376e-01 3.52713108e-01 2.27093250e-02 8.96438122e-01 -1.26230240e-01 1.89678729e-01 5.02202451e-01 1.84044218e+00 -1.82143018e-01 -1.78279400e+00 -1.47774100e-01 -2.30651930e-01 -9.59559381e-01 -1.77203178e-01 1.07457034e-01 -1.40640104e+00 8.25345755e-01 4.67998773e-01 -6.97657093e-02 8.30414057e-01 -1.58844590e-01 9.30419922e-01 5.06501675e-01 3.97415102e-01 -5.49041629e-01 3.84826779e-01 5.60302079e-01 6.20058537e-01 -1.48874879e+00 4.74194363e-02 -5.63628018e-01 -3.64100248e-01 1.08054280e+00 7.15294361e-01 -5.34094393e-01 7.70242929e-01 1.17455035e-01 1.73761338e-01 -2.56541260e-02 -3.87855917e-01 -1.31136224e-01 2.75698096e-01 6.90349698e-01 -1.80094555e-01 -3.02043140e-01 4.40126926e-01 -7.69157231e-01 -4.13964391e-01 -1.53393694e-03 5.21927178e-01 9.53322887e-01 2.70411056e-02 -8.13064098e-01 -2.95852184e-01 1.41687110e-01 5.52800903e-03 2.81266719e-01 -8.45480412e-02 8.42172146e-01 -2.36801151e-02 7.82026172e-01 4.32090372e-01 -2.46971771e-01 4.46325660e-01 -4.96937186e-01 4.76364255e-01 -1.66175932e-01 -3.52413535e-01 -5.32886386e-01 -2.10982263e-01 -1.13127148e+00 -9.82423306e-01 -6.98182821e-01 -9.53267813e-01 -4.75007236e-01 1.77815720e-01 -1.81887269e-01 4.24319655e-01 7.22933292e-01 2.43664011e-01 2.31903359e-01 1.11223090e+00 -1.32677078e+00 -2.00189039e-01 -3.22446078e-01 -1.39381438e-01 8.80520344e-01 7.90663660e-01 -6.31507516e-01 -7.55381167e-01 4.56908911e-01]
[8.563060760498047, -1.9957183599472046]
b733c4de-9c4d-4ab9-af1e-b086699d6cee
vp-slam-a-monocular-real-time-visual-slam
2210.12756
null
https://arxiv.org/abs/2210.12756v2
https://arxiv.org/pdf/2210.12756v2.pdf
VP-SLAM: A Monocular Real-time Visual SLAM with Points, Lines and Vanishing Points
Traditional monocular Visual Simultaneous Localization and Mapping (vSLAM) systems can be divided into three categories: those that use features, those that rely on the image itself, and hybrid models. In the case of feature-based methods, new research has evolved to incorporate more information from their environment using geometric primitives beyond points, such as lines and planes. This is because in many environments, which are man-made environments, characterized as Manhattan world, geometric primitives such as lines and planes occupy most of the space in the environment. The exploitation of these schemes can lead to the introduction of algorithms capable of optimizing the trajectory of a Visual SLAM system and also helping to construct an exuberant map. Thus, we present a real-time monocular Visual SLAM system that incorporates real-time methods for line and VP extraction, as well as two strategies that exploit vanishing points to estimate the robot's translation and improve its rotation.Particularly, we build on ORB-SLAM2, which is considered the current state-of-the-art solution in terms of both accuracy and efficiency, and extend its formulation to handle lines and VPs to create two strategies the first optimize the rotation and the second refine the translation part from the known rotation. First, we extract VPs using a real-time method and use them for a global rotation optimization strategy. Second, we present a translation estimation method that takes advantage of last-stage rotation optimization to model a linear system. Finally, we evaluate our system on the TUM RGB-D benchmark and demonstrate that the proposed system achieves state-of-the-art results and runs in real time, and its performance remains close to the original ORB-SLAM2 system
['Petros Maragos', 'Panagiotis Mermigkas', 'Andreas Georgis']
2022-10-23
null
null
null
null
['simultaneous-localization-and-mapping']
['computer-vision']
[-1.71150580e-01 -2.86981285e-01 -5.93780167e-02 -1.90108970e-01 -2.43232265e-01 -5.44720769e-01 8.23444307e-01 -7.36309737e-02 -6.07361734e-01 5.64222097e-01 -4.14534330e-01 -3.28919113e-01 -8.46707001e-02 -7.74103224e-01 -6.23992622e-01 -5.34773827e-01 4.95841764e-02 8.36008608e-01 4.71289515e-01 -4.62215960e-01 4.99419272e-01 1.02654195e+00 -1.57162702e+00 -5.72678566e-01 8.51330400e-01 8.10203314e-01 4.03686881e-01 3.40046406e-01 -2.24245086e-01 4.01381075e-01 -1.49315342e-01 6.93852529e-02 4.74715620e-01 -2.80789971e-01 -5.57958424e-01 1.13335393e-01 2.03245193e-01 -6.43158183e-02 -3.34480226e-01 9.96047556e-01 2.59809643e-01 -5.25442585e-02 4.35134590e-01 -1.28164506e+00 6.35671094e-02 -2.03272607e-02 -7.70681739e-01 -5.17015338e-01 6.19161487e-01 1.16544943e-02 4.96611267e-01 -1.02861214e+00 1.02501452e+00 1.05333042e+00 6.38238192e-01 -5.39895929e-02 -1.07720172e+00 -3.97550195e-01 -3.09964884e-02 4.04717088e-01 -1.75406158e+00 -4.10143614e-01 8.06579828e-01 -5.13366461e-01 8.83573472e-01 2.81778365e-01 9.55867767e-01 6.74757063e-01 4.25510049e-01 3.52780163e-01 1.11192596e+00 -6.58935905e-01 2.06330985e-01 1.65457502e-01 -1.29539251e-01 8.74977291e-01 3.50018650e-01 2.02378556e-01 -4.04700518e-01 -1.50223700e-02 8.77128959e-01 5.35825780e-03 -4.67196435e-01 -1.31595337e+00 -1.35960102e+00 7.76606262e-01 7.19990194e-01 2.89664507e-01 -2.46995255e-01 6.75352812e-02 -1.03123441e-01 1.10625535e-01 1.34320959e-01 5.33595264e-01 -4.92816307e-02 -1.90162718e-01 -9.23701465e-01 2.02505186e-01 7.02682972e-01 1.16632617e+00 1.38052440e+00 -2.04740569e-01 5.20153105e-01 4.81108695e-01 5.10654628e-01 8.18307340e-01 6.12621427e-01 -7.34111130e-01 4.11977500e-01 9.20836151e-01 3.28538656e-01 -1.24067044e+00 -8.31404269e-01 -3.02889913e-01 -6.12877131e-01 3.82484913e-01 1.34929925e-01 2.16093928e-01 -8.83741558e-01 1.30156636e+00 4.41924065e-01 -2.53594100e-01 3.14005464e-02 1.03434837e+00 3.13902885e-01 3.43373209e-01 -7.98154175e-01 -1.03537582e-01 1.13564861e+00 -9.66981649e-01 -5.87501347e-01 -4.43119556e-01 7.26656079e-01 -8.51385832e-01 6.22989655e-01 4.08815414e-01 -5.11689425e-01 -4.39593285e-01 -1.33507168e+00 -1.05701216e-01 -4.71517414e-01 3.43602747e-01 5.29266000e-01 3.45740527e-01 -1.10046279e+00 4.57592189e-01 -1.02655149e+00 -7.28379071e-01 -2.19443768e-01 3.08605790e-01 -7.57485926e-01 -5.55207729e-02 -8.22330654e-01 1.34390736e+00 5.12301862e-01 2.44268134e-01 -4.55943495e-01 -7.85292014e-02 -9.82440472e-01 -2.26503551e-01 6.11051083e-01 -8.43822718e-01 8.20204675e-01 -7.56590128e-01 -1.70145285e+00 7.90718794e-01 -3.44657302e-01 -3.82399052e-01 8.74570727e-01 -2.31054857e-01 7.70713538e-02 1.30397826e-01 5.74740693e-02 5.40926754e-01 5.33229411e-01 -1.52993906e+00 -5.84704518e-01 -5.77052772e-01 -4.36257571e-02 4.95007902e-01 2.50124753e-01 -4.83594269e-01 -6.69138074e-01 7.18776658e-02 1.06836712e+00 -1.34798253e+00 -4.05413538e-01 5.77093437e-02 -4.07411247e-01 2.58302331e-01 8.59345019e-01 -4.42495078e-01 7.30937421e-01 -2.07129717e+00 3.86209071e-01 4.62144852e-01 6.37986735e-02 1.47125404e-03 1.97937548e-01 6.26194775e-01 2.70140260e-01 -2.48103902e-01 -4.60453145e-02 -5.97730219e-01 -1.50541991e-01 3.63618255e-01 -1.54194117e-01 9.43252504e-01 -2.34668374e-01 6.31522000e-01 -7.69194484e-01 -4.06160474e-01 6.78238630e-01 3.94944996e-01 -1.67678595e-01 2.95031201e-02 3.65143642e-02 5.28698623e-01 -3.51854086e-01 5.62901795e-01 8.47058177e-01 1.62285432e-01 2.22857445e-01 1.34818964e-02 -8.06288898e-01 6.42371103e-02 -1.54380429e+00 2.06532884e+00 -4.44709361e-01 5.82039714e-01 8.53032619e-03 -6.12813950e-01 1.34799612e+00 -1.19370289e-01 4.46532398e-01 -6.10557020e-01 2.71447569e-01 6.98332131e-01 -3.31150919e-01 -1.11114323e-01 9.13577020e-01 3.51452261e-01 1.79368168e-01 -1.48719206e-01 -1.52412564e-01 -6.47069693e-01 -6.50744364e-02 -8.78562182e-02 9.04605925e-01 7.07640350e-01 8.22358131e-01 -2.11651504e-01 8.05402040e-01 3.39794755e-01 3.69621962e-01 4.94726658e-01 1.96683824e-01 5.20469487e-01 3.07481766e-01 -4.45552796e-01 -1.12549555e+00 -8.22904825e-01 -3.45847048e-02 1.12383544e-01 8.71656954e-01 -3.20281804e-01 -3.41717333e-01 -2.36153245e-01 1.32188499e-01 2.71728516e-01 -2.31952816e-01 5.80476634e-02 -5.18846393e-01 -3.91936898e-01 3.81881036e-02 5.70485890e-02 6.07838750e-01 -6.00205898e-01 -1.17734516e+00 1.32259816e-01 -1.91210628e-01 -1.15519440e+00 1.60562962e-01 3.18773240e-01 -8.46237063e-01 -1.19700253e+00 -4.95432228e-01 -4.35031801e-01 6.39138401e-01 6.47935092e-01 5.85041583e-01 -7.66433924e-02 -1.37243092e-01 2.39758894e-01 -6.24226809e-01 -2.36143157e-01 -1.87040269e-01 1.56048447e-01 1.58720255e-01 8.49617422e-02 4.52703759e-02 -5.39765716e-01 -3.15309703e-01 5.78501999e-01 -4.62852061e-01 4.12016839e-01 7.60413647e-01 5.74468791e-01 7.09761322e-01 -4.22929496e-01 -1.79567710e-01 -3.73800755e-01 -2.51828674e-02 -1.70683205e-01 -1.15711904e+00 1.18990717e-02 -7.46843576e-01 1.80550635e-01 4.26862359e-01 -1.71064124e-01 -6.98934019e-01 6.60220742e-01 -1.86633896e-02 -4.82453197e-01 4.97116409e-02 5.23108542e-01 -6.82021603e-02 -6.90612853e-01 5.63248456e-01 4.31450605e-01 1.69699565e-01 -5.02367616e-01 4.45388973e-01 6.81064487e-01 5.41067302e-01 -3.01193774e-01 1.08062208e+00 8.13694537e-01 4.77968395e-01 -1.07104588e+00 -8.36692676e-02 -9.06694591e-01 -1.04349852e+00 -2.18054771e-01 6.86113060e-01 -7.41245270e-01 -6.65512323e-01 5.06608129e-01 -1.33961380e+00 -5.98556362e-02 -3.39825116e-02 8.65452528e-01 -9.19419527e-01 6.17932022e-01 -3.07013728e-02 -7.79692352e-01 -8.62877890e-02 -1.61932421e+00 1.18985724e+00 1.53185666e-01 1.80608496e-01 -5.93075931e-01 3.71218652e-01 -1.74979210e-01 2.49839544e-01 4.03305173e-01 5.26568949e-01 -2.58726627e-01 -1.08280635e+00 -1.56809792e-01 -1.77888513e-01 -1.91247389e-01 -7.42069930e-02 -3.69410440e-02 -6.48072064e-01 -4.94367927e-01 1.30669791e-02 2.85401195e-02 6.20444775e-01 -1.54799214e-02 3.42751324e-01 1.69336706e-01 -7.72772193e-01 1.16453934e+00 1.86199868e+00 3.01605344e-01 7.28086770e-01 1.01714873e+00 7.03671515e-01 3.97735775e-01 9.64852750e-01 4.26079392e-01 5.72739422e-01 1.32423556e+00 1.07376707e+00 -5.76053634e-02 1.75284266e-01 -3.15576047e-01 2.36075848e-01 7.09503114e-01 -3.39751482e-01 2.81683356e-01 -9.95760262e-01 3.58448476e-01 -2.15641046e+00 -4.11886454e-01 -2.97946930e-01 2.53000998e+00 2.12045997e-01 -1.00414641e-01 -2.90410548e-01 1.21327609e-01 3.30112666e-01 1.64341465e-01 -2.46039972e-01 -1.72818497e-01 -8.99641588e-02 -1.34565994e-01 9.11670804e-01 6.70548439e-01 -9.41227734e-01 1.17888415e+00 5.06982613e+00 4.45670515e-01 -1.52866828e+00 -1.48082659e-01 -4.35641348e-01 4.31848049e-01 -3.80922332e-02 5.72502911e-01 -8.66467655e-01 7.91397411e-03 5.62828362e-01 4.89746369e-02 5.01747131e-01 1.08070815e+00 4.88389395e-02 -5.00689328e-01 -9.37450826e-01 1.17424035e+00 2.14359269e-01 -1.00282288e+00 -3.24770927e-01 3.98817241e-01 5.41000307e-01 2.50955433e-01 -4.53566581e-01 8.45547020e-02 -1.80484354e-01 -5.86148143e-01 1.02112150e+00 5.43707490e-01 5.54163635e-01 -5.81560194e-01 9.40429509e-01 7.36308575e-01 -1.36552560e+00 6.78298250e-02 -5.10994196e-01 -1.25811875e-01 2.12625697e-01 4.26902801e-01 -1.08256161e+00 1.30477703e+00 4.03221995e-01 5.67010105e-01 -7.58396208e-01 1.38507962e+00 -3.24340641e-01 -3.17662746e-01 -5.44170439e-01 -1.36793777e-01 2.86168098e-01 -5.39545655e-01 6.88878596e-01 7.90366888e-01 7.18564272e-01 -3.82761806e-01 3.50923061e-01 7.61876762e-01 4.69991326e-01 4.20763731e-01 -8.90167177e-01 4.88392472e-01 3.99295896e-01 1.27200258e+00 -8.11193824e-01 -9.01407227e-02 -2.87818611e-01 1.05174983e+00 3.06694984e-01 2.16141835e-01 -7.99186110e-01 -4.53228980e-01 5.39052784e-01 1.22829936e-01 1.33735508e-01 -9.64989364e-01 -1.72148794e-01 -1.27496743e+00 1.93808183e-01 -6.04468822e-01 -4.00174439e-01 -9.87223387e-01 -3.38110089e-01 8.43483567e-01 -9.05673578e-03 -1.59135950e+00 -5.36592722e-01 -6.91241264e-01 7.40996227e-02 9.31638479e-01 -1.73684502e+00 -1.34052205e+00 -7.74697721e-01 5.56792140e-01 2.52246171e-01 5.63121475e-02 6.63604617e-01 3.19926292e-02 2.29477555e-01 3.67303863e-02 1.76404178e-01 -2.16388255e-01 7.94760704e-01 -1.03784680e+00 3.24858218e-01 1.01463675e+00 4.24534947e-01 5.15198171e-01 7.04252899e-01 -5.84267616e-01 -1.81037593e+00 -6.33141637e-01 8.77999306e-01 -3.47831249e-01 5.52972794e-01 -5.01381814e-01 -5.76801717e-01 9.39035892e-01 -2.69715220e-01 -1.05120149e-02 -2.71165341e-01 -1.38751641e-01 -9.35004950e-02 -1.49337068e-01 -9.84861255e-01 5.37357926e-01 1.04680252e+00 -3.03648710e-01 -5.75859427e-01 1.64385423e-01 5.94530463e-01 -9.56165612e-01 -3.89713883e-01 5.21181226e-01 7.24742830e-01 -1.24951649e+00 8.73433053e-01 3.03943455e-01 -2.78908134e-01 -7.26991713e-01 -2.28835553e-01 -1.35642374e+00 -7.03776777e-02 -5.28711617e-01 1.36540115e-01 8.91921401e-01 -3.70453894e-02 -1.10918522e+00 6.46566093e-01 -2.50069678e-01 -2.36660793e-01 -4.99223202e-01 -1.14215922e+00 -9.50407267e-01 -6.35836840e-01 -2.59854913e-01 6.61158979e-01 7.74328947e-01 -1.80944234e-01 3.32005210e-02 -3.11898708e-01 3.79933089e-01 5.11924207e-01 3.32306951e-01 1.49699867e+00 -1.32192802e+00 -6.06073812e-02 -2.81103373e-01 -1.03263962e+00 -1.30379963e+00 4.48713563e-02 -5.82577050e-01 3.25351566e-01 -1.52386010e+00 -2.52998441e-01 -6.27384484e-01 3.39002669e-01 2.89874405e-01 3.16145957e-01 1.46261036e-01 4.44386721e-01 5.86076796e-01 -2.88696468e-01 6.20685041e-01 8.41455162e-01 1.41830668e-01 -4.28139687e-01 -2.84153037e-02 4.29602899e-02 8.52417529e-01 3.10099870e-01 -2.55723417e-01 -1.09343790e-01 -2.97101945e-01 3.21998000e-01 2.36146569e-01 3.22331429e-01 -1.32026875e+00 3.43975872e-01 -2.82014281e-01 8.31657201e-02 -9.44580615e-01 7.17654824e-01 -1.16877425e+00 4.60879385e-01 7.26734221e-01 4.55556840e-01 1.76254496e-01 -1.10364027e-01 3.50843787e-01 -3.58068883e-01 -2.90545523e-01 6.33372426e-01 -1.62787840e-01 -9.90134001e-01 1.20898865e-01 -8.50516837e-03 -7.32293606e-01 1.14448750e+00 -4.77357119e-01 -4.76447940e-01 -2.79183775e-01 -4.28779483e-01 1.36509791e-01 1.20351934e+00 4.07776862e-01 6.27615035e-01 -1.28493142e+00 -3.68831128e-01 4.15629774e-01 4.29395854e-01 1.91927150e-01 -1.00974858e-01 1.14402187e+00 -1.21806574e+00 6.74483716e-01 -3.34123582e-01 -1.14812362e+00 -1.15261030e+00 6.75566256e-01 2.89522231e-01 -2.19365612e-01 -5.91906369e-01 1.50451109e-01 1.10928945e-01 -5.80911636e-01 -1.52573332e-01 -3.92420530e-01 -1.92598954e-01 -1.54495835e-01 1.24368154e-01 4.00602669e-01 1.44866526e-01 -9.85451758e-01 -6.35718107e-01 1.19342506e+00 6.53813541e-01 -2.86553025e-01 1.07210386e+00 -4.59518433e-01 -3.25374365e-01 5.52084684e-01 1.02465498e+00 3.10792774e-01 -8.56259525e-01 -8.78805965e-02 5.12514040e-02 -6.80772483e-01 -2.04386130e-01 -3.47048074e-01 -5.13973355e-01 7.65371680e-01 6.93113387e-01 7.66401086e-03 9.10210013e-01 -2.43883476e-01 1.83273003e-01 4.86834377e-01 1.26063073e+00 -7.09204316e-01 -3.48958790e-01 8.00728381e-01 9.01366234e-01 -1.17954862e+00 2.41259187e-01 -6.28187120e-01 -2.51668364e-01 1.43999672e+00 3.44610125e-01 -1.84403300e-01 2.24943876e-01 4.92906943e-02 1.64016396e-01 1.04734749e-01 -1.45837680e-01 -3.50437552e-01 1.74421176e-01 4.68314439e-01 -8.63581374e-02 1.55696258e-01 -5.92855692e-01 -7.43618980e-02 -3.85266155e-01 7.80159095e-03 5.97827077e-01 1.02684128e+00 -8.06084216e-01 -1.16644466e+00 -7.47950375e-01 -3.11320484e-01 3.53523880e-01 2.33295605e-01 -3.03565174e-01 1.30013466e+00 1.75735876e-01 6.18190944e-01 -7.88233802e-02 -5.61497033e-01 3.85316044e-01 6.52659908e-02 7.04316139e-01 -3.39680910e-01 -1.47373915e-01 8.13210234e-02 -7.75108039e-02 -9.88291502e-01 -4.00278062e-01 -6.23162270e-01 -1.22512364e+00 -8.54966044e-02 -5.12797177e-01 9.00545940e-02 1.56535888e+00 1.02161348e+00 2.49573812e-01 1.45921528e-01 7.19411731e-01 -1.30194521e+00 -5.45682728e-01 -8.82677495e-01 -6.15444899e-01 1.66023895e-02 4.10324574e-01 -9.47989285e-01 -2.39531383e-01 -4.23119277e-01]
[7.413369655609131, -2.142005443572998]
7bb18f43-3281-46af-92d6-b87cbddc5a63
lstm-knowledge-transfer-for-hrv-based-sleep
1809.06221
null
http://arxiv.org/abs/1809.06221v1
http://arxiv.org/pdf/1809.06221v1.pdf
LSTM knowledge transfer for HRV-based sleep staging
Automated sleep stage classification using heart-rate variability is an active field of research. In this work limitations of the current state-of-the-art are addressed through the use of deep learning techniques and their efficacy is demonstrated. First, a temporal model is proposed for the inference of sleep stages from electrocardiography using a deep long- and short-term (LSTM) classifier and it is shown that this model outperforms previous approaches which were often limited to non-temporal or Markovian classifiers on a comprehensive benchmark data set (292 participants, 541214 samples) comprising a wide range of ages and pathological profiles, achieving a Cohen's $\kappa$ of $0.61\pm0.16$ and accuracy of $76.30\pm10.17$ annotated according to the Rechtschaffen & Kales annotation standard. Subsequently, it is demonstrated how knowledge learned on this large benchmark data set can be re-used through transfer learning for the classification of photoplethysmography (PPG) data. This is done using a smaller data set (60 participants, 91479 samples) that is annotated with the more recent American Association of Sleep Medicine annotation standard, achieving a Cohen's $\kappa$ of $0.63\pm0.13$ and accuracy of $74.65\pm8.63$ for wrist-mounted PPG-based sleep stage classification, higher than any previously reported performance using this sensor modality. This demonstrates the feasibility of knowledge transfer in sleep staging to adapt models for new sensor modalities as well as different annotation strategies.
[]
2018-09-12
null
null
null
null
['photoplethysmography-ppg', 'heart-rate-variability', 'sleep-staging']
['medical', 'medical', 'medical']
[ 1.98455080e-01 9.15179178e-02 -1.66089892e-01 -6.40166044e-01 -6.26104295e-01 -6.46343604e-02 -6.82030842e-02 -6.20714948e-02 -9.10778761e-01 1.08625174e+00 -1.93931639e-01 -2.62721866e-01 -1.57862291e-01 -3.60198677e-01 -2.08901241e-01 -6.79901958e-01 -4.05034721e-01 2.59787858e-01 -2.24575460e-01 2.33169913e-01 3.71347070e-02 4.65671904e-02 -1.41881406e+00 1.01397887e-01 5.74948549e-01 1.44498324e+00 -3.87470424e-01 7.30209291e-01 4.11187112e-01 2.65509963e-01 -9.23864603e-01 -1.31221220e-01 5.08314595e-02 -5.29964089e-01 -7.31515527e-01 -5.83640218e-01 5.25431573e-01 -1.69286028e-01 5.16982749e-02 5.33580601e-01 1.06911731e+00 2.87947595e-01 3.32129180e-01 -8.73021960e-01 -3.39554787e-01 1.91777363e-01 -7.97296539e-02 8.60000551e-01 3.26234072e-01 1.69722155e-01 7.30152845e-01 -3.07827115e-01 1.90876633e-01 2.30188847e-01 1.26811028e+00 9.96903360e-01 -1.28088260e+00 -1.20585918e+00 -4.80879158e-01 4.14895415e-01 -1.46663976e+00 -4.85823810e-01 5.06366014e-01 -3.67465436e-01 1.44251990e+00 2.03781888e-01 1.25929821e+00 1.09234524e+00 6.34546101e-01 9.36280340e-02 1.52318406e+00 -3.31466466e-01 3.78071457e-01 1.12397596e-01 2.77935475e-01 7.48633265e-01 1.99719816e-01 2.30732575e-01 -7.22966671e-01 -8.82386044e-02 3.53073418e-01 -4.77686599e-02 3.42975520e-02 2.46901184e-01 -9.02333677e-01 5.59239149e-01 2.07787201e-01 5.57299435e-01 -3.19882363e-01 1.85829103e-01 6.71700537e-01 1.64456084e-01 6.28564775e-01 3.81972283e-01 -6.11514807e-01 -6.61680818e-01 -1.55669808e+00 -1.84159204e-02 9.14841115e-01 2.76374757e-01 4.60746020e-01 6.38757944e-02 -1.24977775e-01 6.79351151e-01 3.79122257e-01 5.22951961e-01 7.09864736e-01 -1.24579513e+00 -1.14371836e-01 5.37790179e-01 6.99425489e-02 -5.01858711e-01 -1.00900340e+00 -4.15332079e-01 -9.82011676e-01 1.45132765e-01 4.44705188e-01 -2.68420249e-01 -7.27690339e-01 1.70470929e+00 -1.42929226e-01 3.00469160e-01 -6.09353520e-02 5.38792431e-01 1.00811327e+00 2.46960744e-02 3.43151808e-01 -3.78677011e-01 1.55946350e+00 -2.99094200e-01 -6.91771626e-01 -2.89086312e-01 3.33447427e-01 -1.49094924e-01 9.15365160e-01 6.76580787e-01 -1.14224935e+00 -7.89788544e-01 -1.16062319e+00 -4.29735221e-02 -2.23308906e-01 6.04523048e-02 3.70518863e-01 1.22390580e+00 -1.41199529e+00 8.71018171e-01 -1.17264593e+00 -6.24208808e-01 7.18893409e-01 8.53864551e-01 -1.48960687e-02 3.49927664e-01 -1.38403475e+00 1.01670313e+00 2.80470271e-02 2.97999769e-01 -7.13138878e-01 -8.22466612e-01 -6.43909156e-01 -6.09499216e-02 -3.59050870e-01 -9.90450740e-01 9.96064067e-01 -6.33271217e-01 -1.62585223e+00 1.20113719e+00 -2.38885522e-01 -7.86484122e-01 2.65960991e-01 -3.59229669e-02 -6.89267457e-01 3.30640525e-01 4.87069450e-02 5.72250664e-01 6.25472724e-01 -3.77026469e-01 -3.94483387e-01 -4.19570148e-01 -2.26100579e-01 -1.14895090e-01 -2.73521781e-01 -3.00964992e-02 1.80437267e-01 -1.67590678e-01 -3.61382514e-01 -1.07707143e+00 -5.10189719e-02 -3.98614928e-02 7.75049329e-02 -2.52438635e-01 1.60326123e-01 -8.81973386e-01 1.40174210e+00 -1.92591739e+00 -4.41763818e-01 1.49486512e-01 5.82376897e-01 4.69685346e-01 5.27701497e-01 1.45912766e-01 -2.71321051e-02 1.22431569e-01 -2.90306240e-01 -6.88236535e-01 8.31140764e-03 3.45479757e-01 4.43565875e-01 7.11468756e-01 -4.25513715e-01 9.92200792e-01 -6.60073340e-01 -3.70226413e-01 4.55192983e-01 5.26885808e-01 -1.92877635e-01 -2.04996727e-02 5.92713714e-01 4.24515992e-01 7.50492215e-02 5.05406022e-01 1.95828974e-01 -2.67529488e-01 5.94390603e-03 -1.13478117e-01 -1.65114060e-01 2.45360419e-01 -5.18630683e-01 2.06929517e+00 -5.86285949e-01 8.06514263e-01 -2.28951350e-01 -1.16003394e+00 1.17598665e+00 5.40035188e-01 6.69556439e-01 -6.82186902e-01 4.33579385e-01 1.51900768e-01 3.78275722e-01 -5.83778858e-01 -7.70566091e-02 -8.23680818e-01 -1.38250858e-01 5.28881609e-01 2.34268993e-01 -8.63850303e-03 2.34671403e-02 -3.49474102e-01 1.44666564e+00 -6.47196770e-02 3.65029842e-01 -6.11273646e-01 4.90777314e-01 -3.91249597e-01 7.49031544e-01 9.01774645e-01 -8.90827894e-01 3.70240182e-01 7.15511441e-02 -6.27180278e-01 -6.85346842e-01 -1.08543074e+00 -6.08045101e-01 5.42367995e-01 -2.97398835e-01 -5.12194812e-01 -8.00927460e-01 -3.92588496e-01 -2.01414689e-01 5.79909682e-01 -9.58497822e-01 -5.11443794e-01 -2.02127859e-01 -9.50176299e-01 1.11172378e+00 8.06231320e-01 6.79257631e-01 -1.50136983e+00 -1.29486144e+00 1.55910343e-01 -1.10842176e-01 -9.83649731e-01 1.03464257e-02 3.44174206e-01 -1.07905233e+00 -1.09529769e+00 -6.89792752e-01 -4.24689084e-01 2.22867429e-01 -6.39370918e-01 1.20793462e+00 -8.47301781e-02 -6.75073326e-01 6.23214841e-01 -8.31395015e-02 -6.14168465e-01 -1.15389146e-01 2.14596242e-01 4.87726331e-01 -2.19527587e-01 9.71122205e-01 -9.39896703e-01 -1.19069028e+00 3.44591439e-02 -1.87228307e-01 -2.66840845e-01 5.94822288e-01 8.01833868e-01 4.84929770e-01 -3.60875160e-01 7.45309532e-01 -4.62310016e-01 5.89009881e-01 -2.18839273e-01 -1.83772340e-01 -1.20874755e-01 -1.35942709e+00 -3.86486053e-01 5.74807346e-01 -2.51996696e-01 -7.97542036e-01 -2.69624382e-01 -2.78783917e-01 -3.97879452e-01 -4.59367990e-01 2.65814155e-01 4.03852522e-01 3.93841974e-02 7.57546902e-01 4.38298255e-01 2.97604471e-01 -3.27357680e-01 -3.13970894e-01 6.73906446e-01 6.05706036e-01 -2.99189389e-01 2.46265829e-01 5.55212319e-01 1.95244804e-01 -9.35511887e-01 -7.34613419e-01 -4.22587067e-01 -7.67374933e-01 -1.68035984e-01 1.21015179e+00 -7.97820568e-01 -1.17619348e+00 4.99718070e-01 -3.47376704e-01 -8.94456029e-01 -4.94911134e-01 6.36824369e-01 -5.93971968e-01 2.26855934e-01 -5.37192822e-01 -8.70368242e-01 -1.19040215e+00 -5.95542014e-01 7.13896394e-01 3.49546641e-01 -8.44444692e-01 -1.16255212e+00 4.18093801e-01 5.55799663e-01 6.36948466e-01 3.44471425e-01 5.42669535e-01 -6.22719765e-01 2.82499015e-01 -3.97471189e-01 6.39513060e-02 7.19208717e-01 2.65831947e-01 -3.80577207e-01 -1.10146916e+00 -3.65564972e-01 2.58611262e-01 -2.45766044e-01 4.75184351e-01 6.81165159e-01 1.03464949e+00 1.28160864e-01 -7.89386034e-02 5.81356466e-01 1.11157858e+00 4.60483730e-01 8.29825401e-01 3.87278527e-01 3.26137930e-01 8.74138996e-02 1.15995936e-01 3.76667857e-01 5.57548285e-01 3.04423094e-01 -3.23703401e-02 -1.21615864e-01 2.02336945e-02 3.34217310e-01 3.47245067e-01 4.66041386e-01 -3.16793501e-01 4.09456760e-01 -8.69575262e-01 4.25089419e-01 -1.35162783e+00 -8.66600335e-01 5.11574447e-02 2.18249989e+00 7.81579256e-01 3.79101396e-01 4.11747396e-01 3.11455071e-01 3.78812909e-01 1.45020792e-02 -6.70837522e-01 -9.41854239e-01 4.56669331e-01 1.12159348e+00 2.35427573e-01 -3.78092453e-02 -8.22847188e-01 3.60705763e-01 6.45258808e+00 3.25715095e-01 -1.17463148e+00 3.46826583e-01 5.13129890e-01 -5.08171499e-01 5.54687321e-01 -3.04588825e-01 -6.24748468e-01 9.13067102e-01 2.03158069e+00 -1.80643737e-01 5.36108911e-01 4.74655330e-01 5.42048156e-01 -6.15777016e-01 -1.01921165e+00 1.31145477e+00 3.37664813e-01 -1.12555480e+00 -1.10334599e+00 5.51499799e-02 3.62861127e-01 2.94134825e-01 -1.28140360e-01 4.67710465e-01 -4.26253647e-01 -1.08243656e+00 2.03736633e-01 8.30425322e-01 1.23727107e+00 -4.39373553e-01 8.17099810e-01 4.39155698e-02 -1.02583539e+00 -5.88570945e-02 -1.84525661e-02 -5.05008698e-01 2.77559757e-02 4.86159861e-01 -7.55381525e-01 2.74889708e-01 1.45187378e+00 9.76145685e-01 -6.15199208e-01 8.12620163e-01 -1.85949758e-01 8.88522446e-01 -4.73435551e-01 -1.34639472e-01 -1.64142132e-01 3.00795734e-02 5.45248464e-02 1.02370059e+00 2.78679699e-01 4.32879962e-02 -2.63093978e-01 9.60244536e-01 6.90195411e-02 -3.01201075e-01 -3.09454650e-01 1.48313105e-01 9.51991901e-02 1.37915933e+00 -6.13327205e-01 -4.20279443e-01 -4.51413214e-01 6.75290465e-01 -9.66031030e-02 5.86954989e-02 -8.13242972e-01 -5.02643764e-01 3.95811260e-01 1.17700033e-01 7.04607368e-02 6.06190637e-02 -6.25922084e-01 -9.24031019e-01 -8.02496821e-02 -5.17855763e-01 6.83634818e-01 -7.47347534e-01 -1.26006365e+00 4.94145840e-01 2.22885579e-01 -1.06738055e+00 -1.86818838e-01 -3.68578196e-01 -7.73335874e-01 1.06367505e+00 -1.19257355e+00 -9.70582664e-01 -4.30880338e-01 5.29834092e-01 2.34388947e-01 -1.66643560e-02 1.13652158e+00 4.30010349e-01 -8.22760940e-01 7.63545990e-01 -1.32780418e-01 -8.52086395e-02 7.03085899e-01 -1.40870523e+00 1.14358790e-01 5.21851897e-01 -3.15167308e-01 8.22600245e-01 3.57138753e-01 -3.65661532e-01 -8.98085237e-01 -8.61197293e-01 9.51018691e-01 -6.03292108e-01 4.58981037e-01 -2.25133196e-01 -6.96195245e-01 5.67358315e-01 1.65912181e-01 1.88167810e-01 1.33643150e+00 2.70159602e-01 2.88028121e-01 -6.20929003e-01 -1.42833960e+00 3.61110941e-02 8.09866071e-01 -6.22404039e-01 -7.50642836e-01 -1.91469669e-01 -2.49546468e-01 -3.68021399e-01 -1.48596120e+00 5.20408809e-01 1.08175099e+00 -1.19095039e+00 7.07089901e-01 -1.41129538e-01 -3.08590885e-02 1.01503998e-01 3.48047167e-01 -8.95590305e-01 -8.65990818e-02 -7.07522750e-01 -2.96184272e-01 1.02837682e+00 1.82981312e-01 -7.88906693e-01 9.33686376e-01 1.07447410e+00 -4.03265655e-01 -9.76568758e-01 -1.35639167e+00 -4.63871598e-01 1.48775235e-01 -5.45532525e-01 1.04934387e-01 6.04136348e-01 2.03568548e-01 2.93178529e-01 -2.74927735e-01 -3.32945108e-01 4.70285952e-01 -1.26183406e-01 4.00717348e-01 -1.41237879e+00 -7.01644039e-03 -1.65176496e-01 -4.84364599e-01 -1.58420980e-01 1.92643657e-01 -8.75886083e-01 -3.37584987e-02 -1.54827392e+00 2.26716697e-01 -3.82588387e-01 -9.93430257e-01 6.85161293e-01 8.17438290e-02 8.65628362e-01 -2.94750899e-01 1.94959752e-02 -6.05458677e-01 1.60190165e-01 6.61677897e-01 2.21399114e-01 -4.26272392e-01 2.00915843e-01 -6.12041414e-01 6.75190985e-01 1.18948710e+00 -4.89373237e-01 -3.92246991e-01 2.53882021e-01 2.55738407e-01 4.52115126e-02 4.68328089e-01 -1.58893085e+00 3.54474969e-02 5.47901988e-01 8.08638692e-01 -3.03619206e-01 5.56244493e-01 -4.96986657e-01 2.74279058e-01 8.94791007e-01 -1.67965412e-01 1.18946299e-01 4.70247060e-01 3.86872143e-01 3.16316754e-01 4.61005196e-02 7.83122063e-01 -2.77022213e-01 -5.36734343e-01 1.75463080e-01 -6.69778168e-01 1.93779603e-01 8.08332503e-01 -6.75415099e-01 -1.15781814e-01 -2.34693199e-01 -1.20973766e+00 -6.80933893e-02 2.22956195e-01 1.75904945e-01 3.94625664e-01 -8.52638304e-01 -2.26269737e-01 4.31909621e-01 7.32218549e-02 -3.58331829e-01 6.11650229e-01 1.50437891e+00 -2.74094820e-01 4.23675239e-01 -5.32759786e-01 -6.99462533e-01 -1.27653170e+00 1.74102738e-01 7.23679602e-01 -2.17855155e-01 -7.17706442e-01 6.88702881e-01 -5.15543222e-01 6.43684193e-02 -7.03965314e-03 -5.68832219e-01 -1.97141618e-01 1.04235010e-02 3.10781658e-01 7.31718063e-01 3.68239075e-01 -3.09056044e-01 -6.58320606e-01 4.88237530e-01 2.75721908e-01 2.19918955e-02 1.23977828e+00 -2.53645301e-01 -1.26930967e-01 8.59360158e-01 1.09943783e+00 -4.82385606e-01 -8.93249869e-01 2.64074922e-01 -2.16292843e-01 6.45899996e-02 -4.51897122e-02 -1.20647573e+00 -8.37500751e-01 1.09189665e+00 1.51955783e+00 3.97587940e-02 1.22097242e+00 -2.40337923e-01 9.71620739e-01 2.42760539e-01 1.69404730e-01 -1.11208987e+00 -1.91444233e-01 -2.60110125e-02 1.47033125e-01 -1.07619417e+00 1.72609806e-01 4.56178576e-01 -4.84449774e-01 1.04330266e+00 4.76993918e-01 -1.87238753e-01 7.96752453e-01 -8.57832190e-03 2.81454742e-01 -4.61592585e-01 -4.96130347e-01 -7.20389560e-02 3.68228197e-01 6.96741521e-01 4.47734743e-01 -5.07502332e-02 -7.11987376e-01 9.02542830e-01 -3.17806214e-01 6.83815897e-01 3.63675267e-01 8.38775933e-01 1.02768596e-02 -1.06372714e+00 1.60186931e-01 9.77642477e-01 -1.08450103e+00 -9.48772877e-02 3.82296890e-02 7.34691560e-01 4.84123290e-01 1.12578344e+00 2.21767128e-01 -3.40641022e-01 8.81710202e-02 5.99422753e-01 4.70495164e-01 -8.19479287e-01 -8.44840825e-01 -1.70266584e-01 8.00487995e-02 -5.55287421e-01 -8.91026556e-01 -7.44255543e-01 -1.07997704e+00 5.93079478e-02 4.47635874e-02 2.03241229e-01 5.10803819e-01 1.04353726e+00 6.01285875e-01 5.85065424e-01 9.89322737e-02 -8.54774177e-01 -7.05909356e-03 -1.13914549e+00 -7.15109468e-01 2.42377788e-01 3.27269912e-01 -6.32586181e-01 -3.67292523e-01 7.59279355e-02]
[13.536994934082031, 3.4653046131134033]
1ca9f837-e881-4bec-b63b-4a653a23fef1
scattering-spectra-models-for-physics
2306.17210
null
https://arxiv.org/abs/2306.17210v1
https://arxiv.org/pdf/2306.17210v1.pdf
Scattering Spectra Models for Physics
Physicists routinely need probabilistic models for a number of tasks such as parameter inference or the generation of new realizations of a field. Establishing such models for highly non-Gaussian fields is a challenge, especially when the number of samples is limited. In this paper, we introduce scattering spectra models for stationary fields and we show that they provide accurate and robust statistical descriptions of a wide range of fields encountered in physics. These models are based on covariances of scattering coefficients, i.e. wavelet decomposition of a field coupled with a point-wise modulus. After introducing useful dimension reductions taking advantage of the regularity of a field under rotation and scaling, we validate these models on various multi-scale physical fields and demonstrate that they reproduce standard statistics, including spatial moments up to 4th order. These scattering spectra provide us with a low-dimensional structured representation that captures key properties encountered in a wide range of physical fields. These generic models can be used for data exploration, classification, parameter inference, symmetry detection, and component separation.
['Stéphane Mallat', 'Brice Ménard', 'Erwan Allys', 'Rudy Morel', 'Sihao Cheng']
2023-06-29
null
null
null
null
['symmetry-detection']
['computer-vision']
[ 4.15031314e-01 -5.79855323e-01 4.74530496e-02 -3.45077336e-01 -7.13255763e-01 -6.49387956e-01 7.90694594e-01 3.60953569e-01 -1.16667002e-01 7.22331822e-01 5.42294532e-02 -2.05949828e-01 -9.51230347e-01 -7.70655453e-01 -4.04191792e-01 -1.24735582e+00 -3.88293296e-01 8.06934357e-01 2.78024793e-01 -1.29465625e-01 4.01957482e-01 1.03045559e+00 -1.49554479e+00 -2.24956274e-01 4.78299141e-01 1.08691680e+00 3.65285665e-01 6.32732272e-01 2.05105811e-01 1.71382889e-01 -2.12057412e-01 1.12718932e-01 6.19682632e-02 7.43297860e-02 -7.43660927e-01 3.48974578e-02 2.44345263e-01 1.55596346e-01 -4.57999617e-01 1.28688657e+00 3.83140981e-01 3.69973660e-01 1.00513244e+00 -8.83729577e-01 -2.49900550e-01 3.02426487e-01 -3.85359526e-01 3.77050400e-01 1.96482435e-01 -2.86189288e-01 9.37269688e-01 -5.39565265e-01 6.68388784e-01 1.33105874e+00 6.95031226e-01 1.43001437e-01 -1.69464159e+00 -2.80200303e-01 -5.56497574e-01 2.87929147e-01 -1.25467026e+00 -3.96725327e-01 8.68551493e-01 -7.02509284e-01 3.61960500e-01 3.26296121e-01 1.95145965e-01 9.05790448e-01 4.00272191e-01 1.73477799e-01 1.29550958e+00 -7.01044023e-01 4.09839958e-01 -4.42530438e-02 4.19061720e-01 3.18659753e-01 4.66690809e-01 1.17078595e-01 -5.66159904e-01 -5.98187804e-01 6.81939065e-01 -1.64032787e-01 -3.65851939e-01 -5.57693005e-01 -1.32288051e+00 8.71369541e-01 -1.73111744e-02 3.53828162e-01 -4.26649511e-01 9.02656093e-02 -8.42015892e-02 -1.06838979e-01 3.48390847e-01 5.34874916e-01 -4.29516643e-01 5.74883707e-02 -8.40322137e-01 5.23479879e-01 6.77375674e-01 6.58580542e-01 1.00879967e+00 -2.87749887e-01 -6.32378012e-02 6.30785048e-01 2.56706774e-01 1.26971078e+00 -4.89367135e-02 -9.68148172e-01 -4.70196530e-02 -1.04138605e-01 3.02316755e-01 -9.59665656e-01 -7.01463044e-01 -4.55493540e-01 -1.08317268e+00 1.34688675e-01 5.36745429e-01 1.85945213e-01 -7.95282662e-01 1.67280209e+00 3.87522995e-01 1.61971107e-01 5.93346432e-02 6.22321665e-01 5.52676618e-01 5.84659398e-01 -2.73110926e-01 -1.85689270e-01 1.45488536e+00 2.48911589e-01 -4.93186533e-01 -9.05216392e-03 1.05463862e-01 -9.73316491e-01 3.47287267e-01 4.64991122e-01 -9.47838843e-01 -3.87782127e-01 -8.52126539e-01 3.65289122e-01 -3.93239446e-02 -5.24492860e-02 7.82510817e-01 6.01231754e-01 -7.12145507e-01 9.42560375e-01 -8.90743136e-01 -1.33272871e-01 2.59430319e-01 6.70218468e-02 -3.88318032e-01 -5.54964356e-02 -9.26308095e-01 6.03091478e-01 7.53345201e-03 1.09108515e-01 -5.38687110e-01 -7.70460308e-01 -6.97113097e-01 -1.24880066e-02 7.51205385e-02 -3.99935365e-01 1.00621367e+00 2.30373755e-01 -9.69398856e-01 4.59386140e-01 -4.98033673e-01 -3.19739401e-01 2.20271945e-02 4.22030576e-02 -6.86546445e-01 3.90734375e-01 2.16110423e-01 -1.50727600e-01 9.39458370e-01 -9.88162100e-01 -8.74876007e-02 -4.44941103e-01 -2.21945331e-01 -2.40406528e-01 7.36187175e-02 -3.10454648e-02 2.43427634e-01 -5.27275324e-01 6.80015981e-01 -1.05279243e+00 -5.41637123e-01 -3.34795773e-01 -6.23866796e-01 1.95571244e-01 6.65596724e-01 -2.58888870e-01 6.95739269e-01 -2.02056837e+00 3.09366912e-01 8.84631097e-01 3.27252924e-01 -1.82515979e-01 3.02660707e-02 5.11585116e-01 -3.09287488e-01 -6.66776076e-02 -4.08759117e-01 -1.76412966e-02 3.54080298e-03 1.19531238e-02 -4.72000986e-01 1.01101851e+00 3.78392160e-01 3.57860446e-01 -6.90859258e-01 -1.12081677e-01 3.27006757e-01 4.97148335e-01 -3.56380641e-01 -1.00184321e-01 -2.09174559e-01 7.49475956e-01 -8.33563507e-01 3.65122080e-01 1.18598437e+00 -5.15564919e-01 -3.53371724e-02 -4.44522351e-01 -1.90216899e-02 2.05548704e-01 -1.73797011e+00 1.33277881e+00 -2.70295203e-01 6.59416378e-01 2.98314005e-01 -1.20532954e+00 7.04312623e-01 4.39090766e-02 7.37339139e-01 -2.51902908e-01 5.89703955e-02 1.42151132e-01 1.37651963e-02 -3.77749652e-01 3.78642708e-01 -4.05710787e-01 1.61594152e-02 4.19872612e-01 1.13654055e-01 -4.10470515e-01 4.09399033e-01 2.74078697e-01 1.24986625e+00 -4.43860412e-01 6.03599250e-02 -7.96478093e-01 4.91596460e-01 -3.19704503e-01 2.55511492e-01 9.22586501e-01 3.07382494e-01 5.69762647e-01 2.76604325e-01 -1.52316228e-01 -9.18711841e-01 -1.26771045e+00 -9.63731050e-01 4.37250674e-01 1.33104458e-01 -4.19261098e-01 -2.76267827e-01 7.21667856e-02 1.22356929e-01 4.79914010e-01 -4.84132588e-01 -1.20482534e-01 -3.13830197e-01 -1.39333880e+00 6.02297932e-02 5.12189083e-02 1.74204409e-01 -6.98832631e-01 -4.71492082e-01 7.93385133e-02 1.33755393e-02 -1.57004464e+00 8.85764435e-02 2.70721614e-01 -7.54447222e-01 -1.19227123e+00 -4.39968914e-01 -1.90737933e-01 4.43178445e-01 2.91731387e-01 1.03629017e+00 -3.36662173e-01 -8.25281441e-01 5.59924483e-01 -2.06932843e-01 -2.48695463e-01 -5.96144497e-01 -3.78982455e-01 4.87533569e-01 1.02864005e-01 8.44247919e-03 -8.36947560e-01 -2.39403501e-01 2.99376994e-01 -9.48189437e-01 -4.09550667e-01 5.46825767e-01 6.95587635e-01 5.63455880e-01 6.41274691e-01 4.35828678e-02 -5.98048091e-01 5.28460264e-01 -4.24254775e-01 -1.05727065e+00 2.29576066e-01 -2.11312830e-01 6.83750093e-01 5.01042008e-01 -2.99207568e-01 -9.07516122e-01 -2.05350190e-01 1.43004879e-01 -1.39556646e-01 -3.74935806e-01 4.58864987e-01 1.03592627e-01 -5.59457660e-01 7.33316958e-01 2.05094442e-01 -2.55472541e-01 -7.71453559e-01 1.40772045e-01 3.41327548e-01 5.57304263e-01 -9.78838921e-01 1.07288456e+00 8.15348029e-01 1.09768033e+00 -1.43584728e+00 -9.36993062e-01 -8.08094561e-01 -6.84291661e-01 6.92340508e-02 6.90845966e-01 -5.41703641e-01 -8.39427352e-01 3.27659845e-01 -1.12208796e+00 8.96982476e-02 -2.27234080e-01 1.05993378e+00 -5.06108046e-01 5.48500717e-01 -2.07830340e-01 -1.08770537e+00 2.73975134e-01 -1.06336129e+00 1.28166974e+00 1.13063060e-01 4.76154312e-02 -1.23774254e+00 3.62341732e-01 5.18051870e-02 4.38526541e-01 6.76697120e-02 1.21016288e+00 -3.45456868e-01 -5.58600068e-01 -1.02148749e-01 -3.45979035e-02 1.45725235e-01 -1.17381550e-01 1.99162029e-02 -8.60599339e-01 -1.95978492e-01 3.80357951e-01 -1.07978150e-01 9.04977143e-01 8.62608910e-01 1.30847120e+00 2.97531337e-01 -4.69038755e-01 8.59493852e-01 1.25886548e+00 -2.33978838e-01 4.40511912e-01 -5.71098514e-02 3.95988345e-01 6.23839438e-01 1.90567508e-01 6.35341644e-01 -3.06309313e-01 6.38810337e-01 8.50398988e-02 2.82170027e-01 8.39757845e-02 2.92166352e-01 -7.30359703e-02 5.00716090e-01 -2.83911765e-01 -3.79041508e-02 -9.03593898e-01 2.41035774e-01 -1.43116486e+00 -1.24596024e+00 -4.51516509e-01 2.45831609e+00 4.68209714e-01 1.30609766e-01 -1.87387660e-01 2.45583102e-01 8.15898597e-01 9.21335295e-02 -3.46103072e-01 2.69428790e-01 -4.41608548e-01 8.32831681e-01 8.42470229e-01 6.55035555e-01 -1.11062980e+00 2.59070456e-01 7.29233599e+00 9.43085015e-01 -1.11508918e+00 -2.44849220e-01 1.46513969e-01 4.13272411e-01 -3.87985528e-01 1.90181538e-01 -9.08542097e-01 4.28218752e-01 8.79801333e-01 -1.36349753e-01 3.44389260e-01 5.11500120e-01 8.62105638e-02 -3.40877056e-01 -8.55908990e-01 1.08689797e+00 -3.79872769e-01 -1.52383173e+00 -1.38869733e-01 2.68173248e-01 6.18608832e-01 2.44890749e-01 8.12623054e-02 -3.97666454e-01 1.78681552e-01 -1.15015781e+00 4.14254934e-01 7.94537783e-01 6.98982000e-01 -4.87133920e-01 4.60693181e-01 3.04168999e-01 -9.98223424e-01 3.31892937e-01 -5.57869911e-01 1.65990785e-01 5.05740523e-01 1.51232862e+00 -5.09387136e-01 7.60801852e-01 5.62405765e-01 6.44396305e-01 -4.97930229e-01 1.08108902e+00 1.80435523e-01 7.14664340e-01 -8.48290145e-01 9.16310474e-02 -4.10518348e-02 -5.62254190e-01 9.63889599e-01 1.09819746e+00 7.18686283e-01 2.23080531e-01 1.17831621e-02 8.99555802e-01 3.22460711e-01 -3.01759571e-01 -3.22190613e-01 -1.25272125e-01 3.47869188e-01 1.48096013e+00 -1.01426864e+00 8.38255286e-02 -2.32084356e-02 2.44272351e-01 -3.42268467e-01 3.81106645e-01 -3.79028708e-01 -3.62462848e-01 7.26633310e-01 2.51599491e-01 4.16692913e-01 -6.75294697e-01 -4.63943742e-03 -1.25179923e+00 4.81214710e-02 -4.43390161e-01 -7.93943629e-02 -6.60093427e-01 -1.59756160e+00 3.35102856e-01 7.13277817e-01 -9.76101279e-01 -2.76684850e-01 -1.06964612e+00 -6.22631967e-01 9.96253788e-01 -1.36638522e+00 -6.02644980e-01 -1.30345687e-01 6.83795512e-01 -3.48836303e-01 -1.03214338e-01 9.00079370e-01 3.99938785e-02 -1.02756068e-01 -2.99556881e-01 6.61933362e-01 -1.84710801e-01 4.55364764e-01 -1.31779659e+00 3.35024238e-01 6.44061446e-01 2.57375866e-01 7.83133745e-01 1.27352953e+00 -6.18992448e-01 -1.63261151e+00 -6.00117922e-01 4.51931834e-01 -3.15429449e-01 9.15145934e-01 -5.16492665e-01 -8.18861127e-01 1.72258645e-01 -3.95658135e-01 4.85891283e-01 7.10779369e-01 1.72988966e-01 -2.39232957e-01 2.02511195e-02 -1.19561100e+00 1.04992285e-01 7.95746326e-01 -5.22300184e-01 -2.92774588e-01 7.93794930e-01 9.54971910e-02 -1.63567230e-01 -8.06529224e-01 5.55628002e-01 4.03638184e-01 -9.81740236e-01 1.30286729e+00 -5.92397213e-01 -6.90060854e-02 -3.15049171e-01 -5.46213627e-01 -1.33528626e+00 -5.39858997e-01 -9.38856959e-01 6.26420155e-02 6.86534464e-01 1.18456513e-01 -6.30081892e-01 6.18595958e-01 2.39704683e-01 2.48627856e-01 -3.06261003e-01 -1.04707956e+00 -9.64980423e-01 -4.80757169e-02 -7.65933394e-01 2.76287109e-01 8.22061479e-01 -3.75146568e-01 3.43794316e-01 -1.17997795e-01 7.18286216e-01 1.31748748e+00 4.11430627e-01 4.22664970e-01 -1.83328652e+00 -4.92804885e-01 -2.48747095e-01 -8.90737295e-01 -9.59386230e-01 2.28900895e-01 -8.34546626e-01 -4.01681140e-02 -1.14843643e+00 2.07851350e-01 -7.84158051e-01 1.92724288e-01 -2.62643099e-01 3.41875166e-01 6.97211847e-02 -2.09316984e-01 1.64091855e-01 -2.04159513e-01 3.93031299e-01 1.00414097e+00 4.86763306e-02 2.88092822e-01 3.96691948e-01 -3.98344129e-01 9.16452050e-01 5.92840552e-01 -4.51937109e-01 -1.55372605e-01 9.35835987e-02 4.99542147e-01 -1.25690987e-02 7.36111224e-01 -1.32005000e+00 1.54507503e-01 -4.54050928e-01 5.22973955e-01 -5.69826126e-01 6.37313247e-01 -6.18053555e-01 2.48147860e-01 1.01217091e-01 -9.08082873e-02 -2.07750216e-01 7.87845403e-02 6.67763948e-01 -1.09676644e-01 -6.25322461e-01 9.03831303e-01 8.87093917e-02 -4.96229440e-01 3.45392674e-01 -3.28013390e-01 1.79006711e-01 6.66174591e-01 4.04059380e-01 -2.63641834e-01 -5.81384003e-01 -8.90644968e-01 -2.52960861e-01 2.36380711e-01 -2.51216829e-01 3.54082763e-01 -1.23451507e+00 -8.34946036e-01 3.85956645e-01 -5.68730123e-02 -4.85941619e-02 4.33591008e-01 8.13993990e-01 -4.13725168e-01 4.53193754e-01 -3.77315492e-03 -8.95935833e-01 -1.12172532e+00 2.57826000e-01 1.18350811e-01 -2.55476862e-01 -4.41830188e-01 6.25958562e-01 1.55741483e-01 -2.69795775e-01 -3.35229516e-01 -5.11014879e-01 1.07210666e-01 -1.78520471e-01 6.80831730e-01 4.32520986e-01 2.48366326e-01 -9.69394982e-01 -5.11358798e-01 1.08898056e+00 2.45601580e-01 -2.17741191e-01 1.58042002e+00 7.02157021e-02 -3.74060988e-01 4.28502053e-01 1.26327586e+00 5.61643481e-01 -9.86734152e-01 -3.84454668e-01 -1.28597811e-01 -4.72417742e-01 2.37728462e-01 -3.16971838e-01 -6.47645116e-01 9.57007945e-01 3.89907837e-01 6.88894808e-01 8.05409014e-01 4.72642630e-01 3.77430648e-01 7.19628036e-01 3.72525364e-01 -6.72360718e-01 -3.53129596e-01 5.48856080e-01 6.68488324e-01 -1.02263284e+00 4.04367834e-01 -6.50504887e-01 -1.22275893e-02 1.39275849e+00 -4.08730686e-01 -1.19150698e-01 1.26658106e+00 4.17060226e-01 -3.87166142e-01 -3.15164953e-01 -4.20999527e-01 -3.08155835e-01 6.57980025e-01 7.12086797e-01 1.83426350e-01 1.42729789e-01 1.46633700e-01 1.39838830e-01 -4.36900735e-01 -5.37929535e-01 6.24119699e-01 7.85976827e-01 -6.07041299e-01 -1.29984915e+00 -7.36327052e-01 5.13216197e-01 -4.57995713e-01 -3.52311581e-02 2.16057122e-01 3.01201642e-01 -1.68164209e-01 9.02940094e-01 -7.63944164e-02 -7.32789040e-02 -3.62124573e-03 -3.04759499e-02 9.49301302e-01 -4.60012168e-01 4.68152910e-01 1.68329477e-01 -1.52121216e-01 -3.01141739e-01 -7.51881540e-01 -1.01776767e+00 -8.89690816e-01 -3.29278141e-01 -3.21335793e-01 4.35748279e-01 9.95186031e-01 1.08018613e+00 6.64224178e-02 2.82265753e-01 5.55272698e-01 -1.09164870e+00 -9.49079752e-01 -9.09621060e-01 -1.23094177e+00 3.67129683e-01 6.84789360e-01 -1.04268944e+00 -7.22567379e-01 -2.85027865e-02]
[7.141607761383057, 3.9522688388824463]
05b1e514-973b-4ef0-a34f-66dd69ae59a3
simcgnn-simple-contrastive-graph-neural
2302.03997
null
https://arxiv.org/abs/2302.03997v1
https://arxiv.org/pdf/2302.03997v1.pdf
SimCGNN: Simple Contrastive Graph Neural Network for Session-based Recommendation
Session-based recommendation (SBR) problem, which focuses on next-item prediction for anonymous users, has received increasingly more attention from researchers. Existing graph-based SBR methods all lack the ability to differentiate between sessions with the same last item, and suffer from severe popularity bias. Inspired by nowadays emerging contrastive learning methods, this paper presents a Simple Contrastive Graph Neural Network for Session-based Recommendation (SimCGNN). In SimCGNN, we first obtain normalized session embeddings on constructed session graphs. We next construct positive and negative samples of the sessions by two forward propagation and a novel negative sample selection strategy, and then calculate the constructive loss. Finally, session embeddings are used to give prediction. Extensive experiments conducted on two real-word datasets show our SimCGNN achieves a significant improvement over state-of-the-art methods.
['Jinpeng Chen', 'Yongheng Wang', 'Xiongnan Jin', 'Josiah Poon', 'Feifei Kou', 'Fan Zhang', 'Xudong Zhang', 'Yuan Cao']
2023-02-08
null
null
null
null
['session-based-recommendations']
['miscellaneous']
[-5.57650533e-03 -2.61184335e-01 -6.22706473e-01 -5.42203665e-01 -2.31642947e-01 -4.11814123e-01 4.21375096e-01 4.18912411e-01 -3.77184987e-01 5.35130858e-01 3.69887829e-01 -5.91348171e-01 -5.00234187e-01 -9.64134753e-01 -3.73661011e-01 -3.77953827e-01 -7.50978053e-01 3.01324487e-01 1.23619981e-01 -2.61422724e-01 3.62680405e-01 1.75454289e-01 -1.04117775e+00 1.47391960e-01 1.06242907e+00 9.62285578e-01 -8.11221078e-02 4.37481552e-01 -3.02409589e-01 4.82004970e-01 -2.61526704e-01 -1.09620202e+00 2.15576440e-01 -4.62671548e-01 -6.41453385e-01 -1.19995981e-01 3.22358876e-01 -3.85043710e-01 -7.84045756e-01 8.87085676e-01 5.01962245e-01 7.90085375e-01 8.44756126e-01 -1.42935920e+00 -1.54744470e+00 1.07667768e+00 -6.33839726e-01 1.86840788e-01 5.30903816e-01 -4.04107124e-01 1.81664050e+00 -8.40244055e-01 4.60624665e-01 1.11979055e+00 6.37042224e-01 9.18736279e-01 -1.26692665e+00 -6.42348826e-01 5.90221643e-01 5.28886855e-01 -1.08190513e+00 2.02829897e-01 9.80044961e-01 -3.57895792e-02 6.99123323e-01 4.55196381e-01 8.36238086e-01 1.25795317e+00 -2.52261639e-01 1.06503344e+00 6.26336038e-01 -9.26491525e-03 1.87800959e-01 3.03999871e-01 4.27000046e-01 3.63994420e-01 4.64087069e-01 -9.80374515e-02 -5.72864771e-01 -3.33782375e-01 5.75528324e-01 3.34081233e-01 -4.66553271e-01 -5.41307211e-01 -5.50145864e-01 1.20624280e+00 7.66450763e-01 5.76912016e-02 -1.07906796e-01 -1.97663039e-01 3.75523686e-01 4.62839216e-01 7.20663905e-01 3.42964619e-01 -2.91432559e-01 3.43959421e-01 -6.26196504e-01 1.27958387e-01 9.79009867e-01 6.61958992e-01 5.29922545e-01 -3.17276120e-02 -1.54002905e-01 1.21424091e+00 5.71061671e-01 7.18465373e-02 5.09523392e-01 -3.54842752e-01 3.16361129e-01 5.56389093e-01 -5.40725030e-02 -1.22290075e+00 -1.95367306e-01 -6.94680035e-01 -1.01478374e+00 -4.01648104e-01 3.71958882e-01 5.25330156e-02 -6.50738299e-01 1.38635480e+00 8.64131972e-02 5.42748988e-01 -8.20019767e-02 9.97665763e-01 1.18861091e+00 7.59132624e-01 1.68796644e-01 -2.98820704e-01 6.48127079e-01 -1.18861902e+00 -4.21239823e-01 -1.14034519e-01 6.53658986e-01 -2.93724418e-01 1.09423351e+00 3.18385094e-01 -9.23312187e-01 -2.53508627e-01 -9.34685111e-01 4.99823084e-03 -5.28442800e-01 -1.69782177e-01 1.00860322e+00 8.69841456e-01 -1.18588793e+00 9.63367403e-01 -1.09292455e-01 -4.89111483e-01 5.44064105e-01 4.04369801e-01 -3.94015074e-01 -3.10567439e-01 -1.24295998e+00 3.51469338e-01 5.48590720e-02 8.33768472e-02 -2.08069295e-01 -9.27230835e-01 -8.23311090e-01 2.78190941e-01 1.72192290e-01 -5.40076494e-01 9.06854868e-01 -8.84606361e-01 -1.48998797e+00 6.90598607e-01 -1.62855864e-01 -6.00051701e-01 3.81079465e-01 5.67584001e-02 -9.47343707e-01 -3.03042889e-01 -3.69057775e-01 7.73987621e-02 7.97655344e-01 -1.30338514e+00 -5.78154325e-01 -5.02156377e-01 2.66958952e-01 3.13047975e-01 -1.06901586e+00 -2.56463856e-01 -6.20214522e-01 -6.74956143e-01 4.43511307e-02 -7.39486396e-01 -3.54752660e-01 -4.42018174e-02 -3.33033800e-01 -6.82417572e-01 4.56501395e-01 -6.24320805e-01 1.79995871e+00 -1.96464527e+00 9.25240442e-02 5.88869035e-01 4.55125809e-01 4.60014522e-01 -5.67699432e-01 6.52568221e-01 -5.33233918e-02 2.51647830e-01 1.19598143e-01 -4.42746371e-01 1.24038763e-01 -8.78752694e-02 -3.03118467e-01 4.11281198e-01 -4.50693786e-01 8.90060067e-01 -1.04083276e+00 -1.07992359e-01 7.62451068e-02 3.46501142e-01 -8.30372930e-01 4.99921143e-01 2.78820749e-02 1.00873932e-02 -2.83983111e-01 2.75134951e-01 8.41485858e-01 -5.63656628e-01 7.22876966e-01 -1.53751612e-01 4.78312016e-01 4.79733944e-01 -9.87323999e-01 1.54209363e+00 -4.98512655e-01 2.14145124e-01 -4.26440090e-01 -1.06767917e+00 1.10474539e+00 -1.07725829e-01 3.51217419e-01 -8.73729765e-01 6.52522296e-02 4.98857722e-02 2.75989249e-02 -1.69365421e-01 8.62603724e-01 2.07152739e-01 1.50212198e-01 6.56785905e-01 1.55499831e-01 5.37714303e-01 1.41949847e-01 8.20956230e-01 8.53724122e-01 -1.72983184e-01 2.16861770e-01 9.83583108e-02 7.59440243e-01 -6.76057637e-01 4.10865307e-01 7.79986024e-01 -2.67399639e-01 5.38115263e-01 3.44235450e-01 -2.39009008e-01 -7.28872120e-01 -1.30987823e+00 3.76768500e-01 1.46822584e+00 4.39982891e-01 -5.84738731e-01 -2.10210115e-01 -1.22314739e+00 3.83815140e-01 7.10883737e-01 -8.40453565e-01 -2.46180341e-01 -4.05949533e-01 -7.91910112e-01 -1.69894025e-01 5.29850960e-01 -6.22598417e-02 -1.00069630e+00 8.42550039e-01 1.94736034e-01 1.91613026e-02 -6.57449663e-01 -9.57104504e-01 -3.93191934e-01 -1.01285815e+00 -9.31797981e-01 -9.21502829e-01 -1.15284455e+00 6.54267132e-01 8.56026590e-01 1.24451125e+00 3.39846760e-01 7.86625296e-02 3.77956927e-01 -6.02820516e-01 2.30112031e-01 9.13163796e-02 1.82432130e-01 1.92107871e-01 3.32664013e-01 6.37144983e-01 -9.63735163e-01 -9.68565226e-01 2.88998365e-01 -5.56161404e-01 -4.49633539e-01 2.45037749e-01 7.86088288e-01 3.02612841e-01 -3.43859434e-01 1.21164894e+00 -1.59497356e+00 1.16758549e+00 -7.93918431e-01 -2.69981414e-01 3.85279626e-01 -9.32097673e-01 -4.39723074e-01 1.01609647e+00 -5.44307053e-01 -9.47003901e-01 -6.74813032e-01 -2.44999290e-01 -2.56626099e-01 2.66174227e-01 6.91371202e-01 -5.56306988e-02 -2.17229705e-02 5.40917039e-01 3.34447652e-01 -2.21620232e-01 -4.91368085e-01 7.46279538e-01 9.06556606e-01 4.15990315e-02 -1.04847290e-01 8.89737844e-01 2.35812083e-01 -5.10803521e-01 -7.16917813e-01 -8.31221104e-01 -8.53490591e-01 -1.75698653e-01 -2.84075022e-01 3.78347129e-01 -7.12133408e-01 -1.00097859e+00 3.59405220e-01 -6.16409898e-01 -2.28740871e-01 -3.40687335e-02 5.91705382e-01 -1.06122561e-01 6.15683019e-01 -8.37225974e-01 -9.59222376e-01 -7.56428182e-01 -2.89913893e-01 1.45921648e-01 4.29974496e-01 7.80167282e-02 -1.47655141e+00 2.84361094e-02 3.89801294e-01 4.55389231e-01 -6.33081853e-01 9.24525082e-01 -1.25210512e+00 -4.05415326e-01 -4.33874518e-01 -5.20111561e-01 3.76680851e-01 2.57774711e-01 -3.38511854e-01 -5.99322617e-01 -5.81793427e-01 -6.75015152e-01 -4.64080498e-02 9.05868113e-01 3.36315125e-01 1.58993208e+00 -3.63103986e-01 -3.30908090e-01 4.97910440e-01 1.41219521e+00 -1.00043137e-03 6.54165566e-01 -1.45645319e-02 9.71872747e-01 5.05980492e-01 4.64091450e-01 3.94450784e-01 4.49283212e-01 3.59740168e-01 4.01104987e-01 3.00162673e-01 4.85104509e-02 -6.87079310e-01 2.56193101e-01 1.21096373e+00 -8.35919678e-02 -8.93684566e-01 -1.08104430e-01 6.55604541e-01 -2.00119328e+00 -9.56446648e-01 -2.14181051e-01 2.55296493e+00 3.13937962e-01 4.58033048e-02 5.98431408e-01 2.38693319e-02 9.38804984e-01 5.73061407e-01 -6.49476826e-01 -5.79339921e-01 1.71826348e-01 1.21113226e-01 5.10498762e-01 1.95450455e-01 -9.60615814e-01 8.18581522e-01 5.80957556e+00 9.65959489e-01 -8.80886137e-01 -1.00652620e-01 3.86996478e-01 -4.44830209e-02 -7.76189387e-01 -4.28921908e-01 -6.06733322e-01 5.31450868e-01 9.48474169e-01 -5.73650658e-01 7.90021420e-01 8.72072339e-01 2.08469585e-01 6.18062377e-01 -9.58053946e-01 1.15178037e+00 5.03283381e-01 -1.36479962e+00 2.98246771e-01 -1.78241599e-02 8.32286060e-01 -1.19016871e-01 3.80269170e-01 4.98840213e-01 5.19365966e-01 -7.61756897e-01 -1.22127898e-01 2.92782128e-01 5.04428446e-01 -8.81541133e-01 6.50489271e-01 -7.07162768e-02 -1.30399048e+00 -3.82082105e-01 -6.95982099e-01 7.99906775e-02 3.04686755e-01 4.21600431e-01 -5.48281670e-01 8.12699974e-01 6.32177532e-01 1.42970943e+00 -6.48220122e-01 1.58607233e+00 -1.74834132e-01 1.01621580e+00 1.11652426e-01 -7.48854160e-01 1.23108970e-02 -6.86589241e-01 3.73145401e-01 9.42456245e-01 3.43718588e-01 -8.85774568e-02 4.97267814e-03 4.66110677e-01 -7.40858674e-01 7.03818202e-01 -4.38470542e-01 -2.23249704e-01 4.75169063e-01 1.52874601e+00 -4.25648123e-01 -1.04961060e-01 -7.42876053e-01 1.16521525e+00 8.51288378e-01 4.97174323e-01 -6.43208444e-01 -4.54399318e-01 6.64388597e-01 2.73721993e-01 4.48807746e-01 1.59436651e-02 -1.50573626e-01 -1.47678232e+00 -9.99277458e-02 -4.04708833e-01 7.84066200e-01 -4.87308294e-01 -2.28379488e+00 2.78742015e-01 -5.83615482e-01 -1.43891501e+00 2.04422310e-01 -5.26661873e-01 -1.01929069e+00 7.76871800e-01 -1.51905215e+00 -1.16978502e+00 9.29816291e-02 5.68006217e-01 3.59686524e-01 -3.11541051e-01 8.52624476e-01 7.79931545e-01 -5.70879638e-01 1.26179314e+00 5.38405418e-01 2.20203117e-01 6.59284830e-01 -1.43403184e+00 6.30423427e-01 4.89958286e-01 2.61577755e-01 9.75107968e-01 4.05954748e-01 -5.46962321e-01 -1.38487136e+00 -1.44093680e+00 9.72744763e-01 -8.73256400e-02 8.90727043e-01 -3.66851538e-01 -1.12789381e+00 7.27627575e-01 3.12335771e-02 3.58076096e-02 1.20059896e+00 8.85280550e-01 -7.62651682e-01 -4.21035856e-01 -1.12818468e+00 8.25609148e-01 1.53575432e+00 -6.31936967e-01 -3.43333632e-01 3.51304650e-01 7.67470360e-01 2.40910023e-01 -7.99920261e-01 -4.15435657e-02 7.82117605e-01 -7.59510100e-01 1.09025097e+00 -1.10170281e+00 3.29757810e-01 1.70638576e-01 1.44686431e-01 -1.62762022e+00 -7.12895393e-01 -4.89059180e-01 -4.78239983e-01 1.24981201e+00 6.27374351e-01 -8.21629882e-01 1.31467319e+00 3.73272568e-01 5.64485863e-02 -8.31986964e-01 -3.92370701e-01 -9.23009753e-01 1.58681974e-01 -2.32413098e-01 5.04561484e-01 1.04813206e+00 3.58388543e-01 6.29773855e-01 -7.88327217e-01 4.81644124e-02 8.29173326e-01 2.93900967e-01 5.75022340e-01 -1.61385202e+00 -4.32488024e-01 -3.48779857e-01 -2.86872387e-01 -1.33333635e+00 2.54168510e-01 -1.22290111e+00 -3.49042654e-01 -2.02746463e+00 2.34271646e-01 -4.13416296e-01 -8.15177679e-01 -1.06733121e-01 -2.95281023e-01 5.48229992e-01 2.58681830e-02 -1.09249026e-01 -8.92834604e-01 6.15901172e-01 1.14137244e+00 -8.01689774e-02 -4.97594625e-01 4.77562249e-01 -9.39356685e-01 4.89247739e-01 8.64090681e-01 -2.39678845e-01 -6.92239821e-01 -1.52834266e-01 3.90805542e-01 -2.54048314e-03 -9.53601599e-02 -5.45758426e-01 1.63677633e-01 -2.57863812e-02 1.16106734e-01 -4.64016497e-01 2.80668527e-01 -6.13736153e-01 -3.86817217e-01 1.16597660e-01 -7.94648468e-01 -2.36399546e-01 -5.16920924e-01 1.31266904e+00 2.59170622e-01 -1.66843131e-01 5.54440498e-01 2.16880605e-01 -7.76545763e-01 1.07225120e+00 -1.90671697e-01 -9.33520049e-02 6.43632233e-01 -5.23935147e-02 -2.63415664e-01 -6.56096816e-01 -9.25090611e-01 2.78667420e-01 2.55858283e-02 8.41296613e-01 9.02079582e-01 -1.54946530e+00 -5.62539756e-01 -9.79123861e-02 4.11803812e-01 -8.46369624e-01 6.39923811e-01 4.08051699e-01 -2.43670959e-03 1.80372354e-02 1.85492173e-01 5.24307415e-02 -1.45541608e+00 7.56912291e-01 -2.10639406e-02 -2.63875306e-01 -3.46163452e-01 1.03707016e+00 -1.39406204e-01 -8.12443733e-01 3.71890545e-01 3.38892370e-01 -8.18782389e-01 2.88436413e-01 6.22030616e-01 5.32700241e-01 -1.02146879e-01 -3.99982452e-01 -1.21580012e-01 2.44126450e-02 -6.37756407e-01 2.75404721e-01 1.48626590e+00 -2.86679655e-01 -3.41370665e-02 3.69600505e-01 1.61732566e+00 6.01595156e-02 -5.69988966e-01 -6.01293385e-01 -2.47018620e-01 -8.59459639e-01 -9.21801105e-02 -6.51383340e-01 -1.30200171e+00 8.76007617e-01 4.75173771e-01 8.91781390e-01 7.17129111e-01 -3.05029005e-01 1.28078640e+00 2.85002559e-01 2.57936835e-01 -1.03257859e+00 1.65435866e-01 2.62199193e-01 4.44997162e-01 -1.32481158e+00 1.77331921e-02 -5.16516864e-01 -5.58419228e-01 8.80631447e-01 7.62813330e-01 -4.67559814e-01 1.11790657e+00 -8.43908608e-01 -1.69269651e-01 2.50954062e-01 -4.36241627e-01 -2.18653053e-01 5.84990323e-01 8.83958876e-01 5.32198608e-01 2.15642378e-01 -8.20253611e-01 1.01416516e+00 1.02356270e-01 -6.79415464e-02 5.89007437e-01 2.90515155e-01 -2.77646899e-01 -1.30414629e+00 5.99056244e-01 1.19303274e+00 -4.10792410e-01 -2.87869424e-01 -2.19892308e-01 3.09439600e-01 -5.16078949e-01 1.09478331e+00 -4.43065614e-02 -8.70402515e-01 2.13386998e-01 -2.68453985e-01 2.88045406e-01 -7.14331985e-01 -6.24999464e-01 -3.69341552e-01 1.60493508e-01 -3.60702366e-01 -1.87522262e-01 -3.10406417e-01 -7.90609241e-01 -7.61349797e-01 -6.02241576e-01 5.43726623e-01 3.47781569e-01 5.77124059e-01 3.43356043e-01 3.86319935e-01 1.11516750e+00 -6.82742119e-01 -5.00164151e-01 -8.11945736e-01 -1.18847597e+00 7.28653431e-01 -5.87962270e-02 -2.25007251e-01 -7.24268019e-01 -5.77100992e-01]
[10.181093215942383, 5.617000102996826]
56202457-29b7-46f5-82f1-6fbc377b48e9
end-to-end-active-speaker-detection
2203.14250
null
https://arxiv.org/abs/2203.14250v2
https://arxiv.org/pdf/2203.14250v2.pdf
End-to-End Active Speaker Detection
Recent advances in the Active Speaker Detection (ASD) problem build upon a two-stage process: feature extraction and spatio-temporal context aggregation. In this paper, we propose an end-to-end ASD workflow where feature learning and contextual predictions are jointly learned. Our end-to-end trainable network simultaneously learns multi-modal embeddings and aggregates spatio-temporal context. This results in more suitable feature representations and improved performance in the ASD task. We also introduce interleaved graph neural network (iGNN) blocks, which split the message passing according to the main sources of context in the ASD problem. Experiments show that the aggregated features from the iGNN blocks are more suitable for ASD, resulting in state-of-the art performance. Finally, we design a weakly-supervised strategy, which demonstrates that the ASD problem can also be approached by utilizing audiovisual data but relying exclusively on audio annotations. We achieve this by modelling the direct relationship between the audio signal and the possible sound sources (speakers), as well as introducing a contrastive loss. All the resources of this project will be made available at: https://github.com/fuankarion/end-to-end-asd.
['Bernard Ghanem', 'Chen Zhao', 'Moritz Cordes', 'Juan Leon Alcazar']
2022-03-27
null
null
null
null
['audio-visual-active-speaker-detection']
['computer-vision']
[ 8.55351686e-02 1.50754884e-01 1.75482780e-01 -4.55709696e-01 -1.10075819e+00 -3.98546904e-01 7.86666453e-01 3.69829327e-01 -4.37101483e-01 1.58992112e-01 5.93290865e-01 2.94497423e-02 -3.67872208e-01 -5.06055832e-01 -5.09289265e-01 -5.71663737e-01 -6.29462004e-01 1.07885897e-01 3.04708987e-01 -2.44876929e-02 4.67639640e-02 3.18813324e-01 -1.84654117e+00 6.07047617e-01 3.84871244e-01 1.08520532e+00 2.61622310e-01 1.03866363e+00 -1.28594473e-01 8.34913969e-01 -3.22079718e-01 7.24315364e-03 -5.86226117e-03 -4.16727811e-01 -8.68666410e-01 -1.85467992e-02 3.98349673e-01 4.78325635e-02 -3.62597704e-01 6.18166029e-01 7.77037203e-01 3.49534690e-01 2.06726357e-01 -1.40910578e+00 -3.97302121e-01 8.05643618e-01 -2.52012193e-01 2.95885772e-01 5.50804317e-01 1.20240405e-01 1.44394171e+00 -1.30174971e+00 3.34520042e-01 1.24050999e+00 6.07180417e-01 3.93400431e-01 -1.19802201e+00 -4.89314705e-01 6.36364162e-01 6.00324810e-01 -1.35630584e+00 -7.90711880e-01 1.12043452e+00 -3.69790167e-01 9.83390450e-01 3.13371867e-01 6.88566625e-01 1.17864430e+00 -2.84936041e-01 9.91549611e-01 6.61849856e-01 -6.43031538e-01 3.32519650e-01 5.59737943e-02 -2.31449045e-02 5.40313423e-01 -5.32587469e-01 1.53009549e-01 -1.19200027e+00 -2.02161953e-01 1.88495249e-01 3.09030134e-02 -2.01556429e-01 -3.63643497e-01 -1.13969564e+00 7.22791255e-01 5.41602910e-01 4.90089476e-01 -3.88635516e-01 1.22994639e-01 3.57774228e-01 3.42753142e-01 9.08061922e-01 -3.24819150e-04 -3.59452337e-01 -1.46379933e-01 -1.02904701e+00 1.70102194e-01 6.86651349e-01 5.83660841e-01 7.50188649e-01 -5.94392084e-02 -1.36780083e-01 1.03378952e+00 7.01980352e-01 9.43883732e-02 2.37973303e-01 -7.23451078e-01 5.05609274e-01 5.84854543e-01 -2.60279268e-01 -7.76326239e-01 -5.72859466e-01 -5.50352931e-01 -4.62344617e-01 2.17676580e-01 9.14489776e-02 -1.06732763e-01 -6.73671663e-01 1.79808736e+00 7.08850980e-01 7.49787450e-01 -1.41682401e-01 8.92262220e-01 9.23960507e-01 6.55176640e-01 1.08277395e-01 -5.46092205e-02 1.29786170e+00 -1.19068420e+00 -6.47853315e-01 -2.84535021e-01 6.00826085e-01 -7.25168467e-01 1.11184442e+00 3.14867288e-01 -1.06755352e+00 -5.20296752e-01 -8.89085233e-01 -1.07678480e-01 -3.77323598e-01 -5.37394956e-02 2.96677679e-01 1.12933092e-01 -1.43921506e+00 4.39745307e-01 -9.72705305e-01 -2.97402233e-01 4.09425884e-01 2.62832493e-01 -2.60446519e-01 1.08087614e-01 -1.17994475e+00 4.85669494e-01 6.94379359e-02 2.73595124e-01 -1.06731284e+00 -8.47577989e-01 -9.54780638e-01 1.51802329e-02 5.12597382e-01 -5.73610246e-01 1.38913202e+00 -9.31414723e-01 -1.47266984e+00 5.98286152e-01 -2.56098986e-01 -4.30540383e-01 4.97602522e-01 -3.53105068e-01 -4.53435689e-01 3.17954093e-01 -5.06525971e-02 4.51618969e-01 8.46917987e-01 -1.17939329e+00 -8.48156691e-01 -4.29330617e-01 1.78887434e-02 2.65678853e-01 -5.40161252e-01 3.08441967e-01 -4.55539048e-01 -5.70077419e-01 -1.03398480e-01 -9.25017238e-01 -7.51758814e-02 2.01625675e-01 -3.24942559e-01 -5.42408764e-01 9.06439662e-01 -6.48428559e-01 1.40869892e+00 -2.50573468e+00 3.57587934e-01 1.57022730e-01 3.99363309e-01 8.89028385e-02 -3.71922433e-01 8.53109717e-01 -1.32796124e-01 -1.59191623e-01 -2.59961367e-01 -1.14361262e+00 2.15894997e-01 -8.18512142e-02 -1.90307721e-01 4.03659374e-01 3.54487896e-01 5.60377300e-01 -9.51980531e-01 -3.24852437e-01 2.20855251e-01 8.11453700e-01 -5.74904382e-01 4.39101130e-01 -2.37690136e-01 5.68277597e-01 -1.50708541e-01 3.78434211e-01 3.58763009e-01 -7.94674829e-02 -3.12066879e-02 2.20037907e-01 -2.31626064e-01 6.14348233e-01 -1.29560196e+00 2.01057720e+00 -8.05920124e-01 7.20347106e-01 3.91273141e-01 -8.75340343e-01 7.34350324e-01 5.52908301e-01 4.69195485e-01 -5.43023169e-01 -2.92696059e-01 4.33252752e-02 -2.62195498e-01 -6.35370672e-01 2.82013535e-01 1.78959265e-01 1.42373017e-03 4.53509748e-01 3.73423308e-01 2.87550807e-01 7.89384767e-02 3.80442113e-01 1.15151966e+00 1.29402325e-01 1.14607431e-01 -6.91748932e-02 5.70789993e-01 -3.99885476e-01 4.49239492e-01 5.49954176e-01 -1.63905874e-01 5.58816612e-01 4.88775313e-01 -2.87348419e-01 -5.43931186e-01 -9.87900376e-01 1.47304282e-01 1.62875617e+00 -1.68051869e-01 -7.12277174e-01 -4.76488858e-01 -7.78260171e-01 -1.10240869e-01 6.80665433e-01 -7.79147863e-01 -9.66215506e-02 -6.85594380e-01 -1.80531621e-01 4.24483001e-01 5.38640738e-01 -2.84862914e-03 -1.04148757e+00 -3.40013742e-01 2.58281082e-01 -2.03611046e-01 -1.02022970e+00 -3.45034391e-01 2.84929514e-01 -4.69089687e-01 -7.60231614e-01 -5.12660503e-01 -7.50122190e-01 2.14323267e-01 2.37185463e-01 1.05667603e+00 4.92217913e-02 -3.11482280e-01 7.26221919e-01 -4.90504622e-01 -3.93208325e-01 -2.76574165e-01 1.29448816e-01 -4.03020009e-02 4.15231615e-01 1.90348715e-01 -9.17678535e-01 -8.49914551e-01 5.34543544e-02 -7.68300474e-01 -2.72089001e-02 2.79048115e-01 6.26726210e-01 5.68188071e-01 -3.58937770e-01 7.37140536e-01 -4.92528617e-01 4.64235812e-01 -7.08290517e-01 -3.35608691e-01 -1.61830615e-02 -3.94526660e-01 -1.57937139e-01 5.39885819e-01 -4.21532780e-01 -9.37781870e-01 3.25752139e-01 -3.14012378e-01 -5.14790773e-01 -4.75761086e-01 5.77023208e-01 -2.06209853e-01 2.88851172e-01 6.43433094e-01 4.54995520e-02 -1.68473204e-03 -6.58438683e-01 4.67509955e-01 7.46565580e-01 1.75209016e-01 -2.72408783e-01 5.45323730e-01 4.26059455e-01 -2.43365154e-01 -1.00610185e+00 -8.16115975e-01 -7.57537246e-01 -5.68044901e-01 -4.39666301e-01 7.30497539e-01 -1.14590812e+00 -4.33804452e-01 2.61215150e-01 -1.01004195e+00 -5.10185122e-01 -4.60229337e-01 5.42150974e-01 -5.19648612e-01 2.05209419e-01 -4.30334479e-01 -1.08407450e+00 -1.86225086e-01 -8.44020665e-01 1.31086564e+00 -8.43449309e-02 -3.33338737e-01 -1.18136358e+00 3.29699367e-01 2.99065202e-01 4.13411707e-01 5.96093200e-02 5.64523458e-01 -9.49069142e-01 -4.80450898e-01 -6.87095220e-04 8.81033167e-02 1.72431275e-01 -2.32057795e-02 -1.03485063e-01 -1.61923218e+00 -2.89485782e-01 -1.88825637e-01 -2.16781035e-01 1.21391714e+00 1.38027295e-01 9.38467085e-01 -3.82542193e-01 -1.35615155e-01 3.36140901e-01 1.12825286e+00 -1.64001033e-01 1.42167225e-01 1.54699594e-01 7.62072206e-01 9.61505353e-01 4.15355027e-01 6.17329717e-01 7.08981633e-01 9.83429492e-01 6.19228482e-01 -1.25809178e-01 -5.09083748e-01 -2.07512274e-01 6.19374454e-01 8.91331255e-01 2.71676004e-01 -2.43699268e-01 -1.20900691e+00 9.48625147e-01 -2.02656174e+00 -9.03583646e-01 -1.13530315e-01 2.03391767e+00 7.42987454e-01 -5.14685065e-02 5.10986865e-01 5.56131184e-01 6.37981594e-01 4.02302384e-01 -2.83708513e-01 -3.01961720e-01 1.78853780e-01 2.52019558e-02 -3.60012531e-01 8.59213233e-01 -1.23146617e+00 7.09228337e-01 5.05156946e+00 7.29341269e-01 -1.21789694e+00 3.50873381e-01 1.48117810e-01 -6.11488938e-01 -2.96801567e-01 -6.32499754e-02 -6.37668490e-01 3.90817165e-01 1.20576167e+00 6.68135583e-02 3.79002780e-01 6.52078390e-01 3.99026901e-01 1.15694053e-01 -1.34109163e+00 8.85821581e-01 8.38067308e-02 -1.09213257e+00 -2.15574414e-01 -4.34555225e-02 2.41998628e-01 3.42150569e-01 7.37292245e-02 2.43887827e-01 4.47775647e-02 -5.77322125e-01 1.04331589e+00 4.49078828e-01 4.90029633e-01 -6.58654690e-01 3.76463234e-01 2.83055574e-01 -1.60983348e+00 -3.86925846e-01 1.78274632e-01 5.39878793e-02 2.06790239e-01 8.16231430e-01 -9.47125375e-01 7.27704048e-01 8.32633018e-01 7.09003747e-01 -5.75076699e-01 1.18215346e+00 -2.87545234e-01 8.50190103e-01 -5.13909519e-01 3.54967900e-02 2.38382012e-01 2.75787920e-01 8.60484064e-01 1.60254896e+00 2.41645992e-01 -3.39524388e-01 2.65443087e-01 6.55795455e-01 -1.95207018e-02 2.05827445e-01 -6.32672548e-01 1.13010414e-01 5.07511616e-01 1.18990111e+00 -4.47465241e-01 2.50322055e-02 -6.12762511e-01 7.93019354e-01 4.80565280e-01 3.54918957e-01 -6.87449157e-01 -3.94489467e-01 6.03988767e-01 1.29570559e-01 4.90433425e-01 -9.91299003e-02 1.46769704e-02 -9.64876771e-01 2.77960002e-01 -5.64909339e-01 6.05689108e-01 -4.40671325e-01 -1.30323255e+00 7.41598606e-01 -4.48552109e-02 -1.36987376e+00 -4.32703048e-01 -2.50940561e-01 -8.31726909e-01 6.84233904e-01 -1.64519036e+00 -1.45973563e+00 -2.06043020e-01 7.65032470e-01 6.50894523e-01 1.08256591e-02 1.00867677e+00 5.00900328e-01 -5.39894402e-01 5.47516525e-01 -2.37804934e-01 4.49971743e-02 6.64638102e-01 -1.42256272e+00 3.74787629e-01 8.45785141e-01 5.40447593e-01 2.60298491e-01 6.16009176e-01 -1.90871835e-01 -1.17225552e+00 -1.22407317e+00 1.18217301e+00 -3.07009846e-01 9.40197885e-01 -8.38860750e-01 -8.28066170e-01 5.54004848e-01 4.32351142e-01 3.46214712e-01 9.11221623e-01 5.31415343e-01 -5.97984135e-01 -2.78563023e-01 -7.82876313e-01 3.66915077e-01 1.14383721e+00 -9.19515848e-01 -4.70146805e-01 2.55763859e-01 8.39415789e-01 -9.22715664e-02 -5.82377017e-01 1.07138224e-01 3.52913260e-01 -9.40164447e-01 8.33219469e-01 -4.15480882e-01 3.71384591e-01 -2.42045060e-01 -1.82732955e-01 -1.39079034e+00 -2.54690289e-01 -7.05064356e-01 -4.14495230e-01 1.53415740e+00 7.06104398e-01 -4.60767627e-01 4.21896368e-01 1.41564906e-01 -3.02409470e-01 -9.02912259e-01 -1.30031824e+00 -6.51095271e-01 -2.53140181e-01 -8.80253792e-01 3.55506331e-01 9.18306410e-01 1.14599012e-01 4.79288369e-01 -3.10145915e-01 3.66427004e-01 5.48018754e-01 5.63151129e-02 5.04978597e-01 -1.33937109e+00 -4.92445648e-01 -3.65380764e-01 -5.36741555e-01 -8.25875282e-01 2.19344854e-01 -1.00282168e+00 1.40428618e-01 -1.63713431e+00 -2.35304739e-02 -2.93935508e-01 -7.12753057e-01 6.34492159e-01 -8.61954913e-02 1.70398250e-01 3.15911561e-01 3.88087071e-02 -9.79301333e-01 6.57707334e-01 7.19497919e-01 -5.22633232e-02 -4.44021314e-01 9.50875953e-02 -5.20094156e-01 5.91100097e-01 9.08395588e-01 -6.68186307e-01 -4.20533150e-01 -5.09831131e-01 1.85104966e-01 -8.45339056e-03 6.19329333e-01 -8.85901809e-01 5.07975578e-01 2.31157973e-01 9.49443951e-02 -5.53029895e-01 6.98071241e-01 -8.05385709e-01 -1.95105538e-01 6.21501096e-02 -7.21215844e-01 -1.55362859e-01 2.68557847e-01 8.20969760e-01 -3.78955364e-01 1.74151942e-01 5.00358403e-01 2.06597403e-01 -5.03327012e-01 1.86766520e-01 -3.49601567e-01 -9.34351759e-04 8.05001140e-01 9.62274000e-02 -8.44275132e-02 -5.46525776e-01 -1.23081601e+00 3.05690140e-01 -1.60165325e-01 7.35062361e-01 6.29166007e-01 -1.37057197e+00 -1.01233590e+00 2.45845094e-01 3.11276287e-01 -5.89631386e-02 5.28163373e-01 1.10654628e+00 5.98872527e-02 8.62609521e-02 3.57146174e-01 -9.33234036e-01 -1.48152053e+00 3.98210138e-01 2.49570772e-01 -1.95762351e-01 -7.30037272e-01 1.16020191e+00 1.29323348e-01 -4.65188712e-01 7.18888223e-01 -2.50838131e-01 -3.44773382e-01 5.52094102e-01 7.31304646e-01 3.26070398e-01 2.78495550e-01 -6.54416621e-01 -5.66375256e-01 3.41066241e-01 -4.19643186e-02 -3.23643386e-01 1.75964725e+00 -4.59557235e-01 7.16722161e-02 7.21446753e-01 1.31413352e+00 2.23851025e-01 -1.32718682e+00 -6.69410646e-01 1.62281692e-01 -1.69676661e-01 3.12258869e-01 -8.39221537e-01 -1.10608506e+00 1.11019778e+00 8.77997279e-01 6.32188857e-01 1.22267711e+00 4.09085721e-01 5.20829499e-01 1.21791244e-01 6.79767802e-02 -9.05866027e-01 1.70743898e-01 3.72356653e-01 1.16716480e+00 -9.89421666e-01 -2.56526351e-01 -2.48200759e-01 -6.26650155e-01 1.01006603e+00 2.96805024e-01 -2.33931858e-02 9.41603601e-01 4.03796047e-01 2.62312472e-01 -1.33842364e-01 -1.16811883e+00 -3.94505292e-01 4.54551429e-01 4.73663449e-01 4.35779989e-01 -4.44232598e-02 2.45619088e-01 7.88322747e-01 -5.31207509e-02 -1.96816146e-01 1.54947579e-01 9.67561901e-01 -2.87788451e-01 -1.11101067e+00 -2.30565205e-01 -9.71546173e-02 -4.01626438e-01 -1.47985697e-01 -5.85891366e-01 4.96700317e-01 1.30232751e-01 1.27606201e+00 1.61033664e-02 -4.92138624e-01 5.41418314e-01 2.44187117e-01 1.58698216e-01 -6.83182120e-01 -7.75507748e-01 3.62380296e-01 2.48676404e-01 -9.07070994e-01 -3.72868389e-01 -6.84620142e-01 -1.18013895e+00 1.01783223e-01 -2.32837796e-01 1.77802697e-01 6.70847237e-01 7.49464631e-01 7.57313728e-01 7.49784589e-01 9.09200847e-01 -1.10610294e+00 -2.70944506e-01 -1.05084550e+00 -3.62476587e-01 1.60663903e-01 8.24494779e-01 -4.57508773e-01 -7.19743907e-01 4.79862094e-02]
[14.731039047241211, 5.002188682556152]
19b103b2-df44-4e5a-860a-5e866bf0ad30
discoscene-spatially-disentangled-generative
2212.11984
null
https://arxiv.org/abs/2212.11984v1
https://arxiv.org/pdf/2212.11984v1.pdf
DisCoScene: Spatially Disentangled Generative Radiance Fields for Controllable 3D-aware Scene Synthesis
Existing 3D-aware image synthesis approaches mainly focus on generating a single canonical object and show limited capacity in composing a complex scene containing a variety of objects. This work presents DisCoScene: a 3Daware generative model for high-quality and controllable scene synthesis. The key ingredient of our method is a very abstract object-level representation (i.e., 3D bounding boxes without semantic annotation) as the scene layout prior, which is simple to obtain, general to describe various scene contents, and yet informative to disentangle objects and background. Moreover, it serves as an intuitive user control for scene editing. Based on such a prior, the proposed model spatially disentangles the whole scene into object-centric generative radiance fields by learning on only 2D images with the global-local discrimination. Our model obtains the generation fidelity and editing flexibility of individual objects while being able to efficiently compose objects and the background into a complete scene. We demonstrate state-of-the-art performance on many scene datasets, including the challenging Waymo outdoor dataset. Project page: https://snap-research.github.io/discoscene/
['Sergey Tulyakov', 'Bolei Zhou', 'Hsin-Ying Lee', 'Yujun Shen', 'Ceyuan Yang', 'Aliaksandr Siarohin', 'Ivan Skorokhodov', 'Sida Peng', 'Zifan Shi', 'Menglei Chai', 'Yinghao Xu']
2022-12-22
null
http://openaccess.thecvf.com//content/CVPR2023/html/Xu_DisCoScene_Spatially_Disentangled_Generative_Radiance_Fields_for_Controllable_3D-Aware_Scene_CVPR_2023_paper.html
http://openaccess.thecvf.com//content/CVPR2023/papers/Xu_DisCoScene_Spatially_Disentangled_Generative_Radiance_Fields_for_Controllable_3D-Aware_Scene_CVPR_2023_paper.pdf
cvpr-2023-1
['3d-aware-image-synthesis']
['computer-vision']
[ 0.2918554 -0.21979433 0.21060976 -0.25862485 -0.4506362 -0.8852473 0.84491 -0.20331359 0.2203015 0.36710334 0.15685663 0.03209176 0.0265527 -0.99941045 -0.94175243 -0.9253097 0.40254888 0.42543778 0.3214671 -0.35112146 -0.09792456 0.808605 -1.7639511 0.08029699 1.0034003 0.8432254 0.7427745 0.7733631 -0.03938012 0.7695727 -0.50991005 -0.1723429 0.55770785 -0.44235533 -0.3472899 0.615375 0.83370656 -0.31491166 -0.51671374 0.9144905 0.47036558 0.08262369 0.55561215 -1.240971 -0.93776435 0.22389516 -0.32835296 -0.29041383 0.27506062 0.5181334 0.85353285 -0.8885166 0.6731334 1.1670622 0.11991276 0.29631808 -1.7117486 -0.4382696 0.30903465 -0.16702987 -1.4675634 -0.46253294 1.0724946 -0.602359 0.54710585 0.6680744 0.8601187 1.0712532 0.00617382 0.6276998 1.2368621 -0.30491233 0.16595158 0.15642461 -0.16894448 0.6121985 0.20638806 0.01998032 -0.46723405 0.1917327 1.0871072 0.05180259 -0.47902897 -0.8439013 -1.4581319 0.44620618 0.60988 -0.05874157 -0.13419685 0.24220598 -0.1450203 -0.18467933 0.4224182 0.54979753 -0.21639903 0.39910164 -0.5947501 0.59675425 0.4322983 1.5600429 0.916008 0.25413424 -0.65940887 0.81605303 0.08206478 0.95798385 -0.26368338 -1.0309823 0.25298396 0.52325994 0.33271977 -0.92212564 -0.22053081 -0.6048974 -0.9948912 0.34401864 0.160254 0.14539069 -1.086672 1.69723 0.5650094 0.18031298 -0.35029596 1.0000457 1.047752 0.7810141 -0.10684825 0.18830365 1.413842 -1.1981089 -0.5350716 -0.29734054 0.09125073 -0.88267463 1.255459 0.39756757 -1.2970009 -0.78235084 -0.92702836 -0.41095093 -0.35050878 0.22440042 0.93741554 0.53016216 -0.88311857 0.117553 -0.50262594 -0.06566308 0.47378728 -0.08289797 -0.25120237 -0.28228158 -0.7396748 0.61509544 0.42948788 -0.06449637 -1.1031755 -1.1593907 -1.0150833 0.05039383 0.55414575 -1.2229502 0.9843201 -0.5700222 -1.6954962 0.93889296 0.0256865 0.08225752 0.59102994 -0.09362007 -0.25283882 0.06103597 0.03277162 0.6603666 0.9135682 -1.7425234 -0.34707788 -0.10252414 0.2290705 0.4535077 0.05565747 -0.30876872 -0.72081554 -0.8539257 0.09961976 -0.8583138 -0.25262788 0.32084808 -0.67098343 0.38927504 0.62080264 -0.31303906 0.9438236 -2.2011998 0.4587634 -0.06853385 0.37315404 -0.02340225 -0.14089662 0.40337828 -0.05721039 -0.02004162 -0.3256989 -0.4788133 0.26317355 0.1353072 -0.35836676 0.37607557 0.26728916 0.97525585 -1.0369327 -0.34592232 0.68224764 0.65808165 -0.79875165 0.43428662 -0.594504 0.8630899 -0.39238623 0.6345458 1.0864589 -0.28021422 0.02533626 -0.5003032 -0.16432224 0.08664415 -1.3662863 2.0922692 -0.5685185 0.41675434 0.14836212 -0.46379936 0.89168596 -0.07652905 0.36425683 -0.6368635 0.11456897 0.02429516 -0.43777567 -0.2079005 0.5044478 0.07668602 -0.286667 0.13882934 0.1603325 -0.96132076 0.1879677 0.18845925 0.6353529 0.40200442 0.20079473 -0.4578565 0.25841403 -0.21390893 0.31768802 0.7734397 0.36701003 1.0649 0.14493781 -0.19059826 -1.1227921 -1.5443547 -0.23678307 0.8271038 0.5746765 -0.35287476 -0.6119439 -0.193953 -0.1327598 0.95241904 -0.4562176 -0.04714829 -0.39559138 -0.60766524 0.06624067 0.10585249 0.63802093 -0.7041653 -0.5474582 -0.03927052 -0.2729707 -1.3514507 -0.68372315 -0.07809216 -0.43359682 -0.85956436 -0.5174595 -0.55884 0.813763 0.7261953 1.4551471 -0.06780601 -0.57349586 0.24985498 -0.1901855 -0.2240274 -0.41070032 -0.22834677 -0.2623527 0.34068054 -0.47553176 -0.8775677 -0.7434323 0.3267242 -1.140586 0.9183569 0.291008 0.57715803 0.8165308 0.13374087 -0.06848601 -0.85527617 0.06020677 -0.268532 -0.8770607 0.2569181 0.00941953 -0.03180612 0.6150431 -0.4286096 -1.3692925 0.07617872 0.13372247 -0.43747872 -0.43915278 -0.24877372 -0.7173718 -0.08253659 0.738807 0.43677822 -0.5405979 -0.6045089 0.87431073 0.11230807 0.69002926 -0.9052184 1.1748511 0.6743538 0.228544 -0.8520543 -0.91404223 -0.2976865 -0.75553197 -0.24113669 0.9267589 -1.2184454 -0.43520868 0.66399205 -1.0908204 -0.64974385 -0.5364941 0.08500612 -0.71242297 0.01803477 -0.24923553 -0.49028307 0.09261755 -1.2166512 1.5606135 0.18717963 0.14022385 -0.77057576 -0.09095416 0.2690423 0.32544163 0.7202967 0.8893394 0.32114133 -1.3541275 0.12099709 -0.34719685 0.10297654 0.31676382 0.22528368 -1.1450825 -0.03944282 -0.26994327 -0.0134778 0.78221 0.16327077 1.4748479 -0.22729138 -0.10073239 1.0883201 1.6224358 -0.11043137 0.7192529 -0.06447737 1.1400033 0.48496068 0.3552255 0.57919335 0.41758803 1.036585 0.59304374 -0.3062503 -0.79380363 -0.523823 0.03521201 0.5859279 -0.03714253 -0.58453554 -0.5721097 0.40209064 -1.6127009 -0.87141323 -0.3214782 2.0850382 0.79563874 -0.12914559 -0.07474495 -0.15017222 0.5603534 0.41028062 -0.49377024 0.14145973 -0.46789795 0.19983894 0.41308665 0.61947006 -1.0647284 1.0617815 5.691693 1.0254234 -1.0033395 0.03730805 0.45764744 -0.28335512 -0.8221556 -0.01412988 -0.71716297 0.34932017 0.18136963 -0.11833525 0.56708676 0.7703545 0.2517022 -0.05751626 -0.9729535 1.0731013 -0.03718593 -1.6401463 0.41473737 -0.07475955 1.0791693 -0.32097438 0.17602336 -0.09445681 0.370084 -0.8002045 1.3332441 0.7113714 1.0564929 -0.43845084 -0.08525874 0.286697 -1.2093282 0.13459425 -0.21440022 0.13944434 0.2870248 0.7268863 -0.45131755 0.72404164 0.6332002 0.59185445 -0.61731243 1.0409049 -0.42004377 0.17338927 -0.32065448 0.24833077 -0.09221923 -0.43634567 0.69591075 1.250556 0.38213965 0.15174372 0.34027517 1.4211446 -0.10553873 -0.09220459 -0.5910966 0.2078631 0.39897612 1.3589652 -0.71625924 -0.24440542 -0.1552322 1.1867204 0.24293472 0.5246781 -1.1172979 -0.21476941 0.8175945 0.30290908 0.37938094 -0.4646462 -0.41558892 -1.4627978 0.08836763 -0.6496727 -0.23545209 -1.3793308 -1.1246866 0.4979686 0.24966148 -1.3183199 0.2744863 -0.64920235 -0.41183344 0.9422543 -1.4325969 -1.623571 -0.90188456 0.7126312 0.55236584 0.25544253 0.6686975 0.37974897 -0.51700026 0.24006596 -0.12516947 -0.20526572 0.57141525 -1.3017765 0.39031437 1.009114 0.19920583 0.3828418 0.7035654 -0.41347292 -1.6278948 -1.3148887 0.39827013 -0.7350917 0.30180797 -1.0420825 -0.56835926 0.39695027 0.13841239 0.13492876 0.34554964 -0.36637756 -0.5237926 -0.26132295 -0.9058572 1.0429869 1.5753646 -0.45170152 -0.05201123 0.50566417 1.2263789 -0.80844766 -0.7762047 0.32163778 0.37252882 -1.1954728 1.3882798 -0.18567374 0.5390728 -0.7475165 -0.38195983 -1.240736 -0.65878576 -0.7889541 -0.10698483 1.1462836 0.16760251 -0.39368597 0.26531613 0.42938054 -0.46059313 -0.34844175 -0.2905316 -0.6509559 -0.28509527 -0.5935929 0.97541445 0.8278874 -0.8335734 0.24004647 -0.48661688 0.49371856 0.9010457 0.58118814 1.2086905 -0.9759999 -0.51798046 -0.5576324 -0.18438432 -1.2837609 -0.12480412 -0.87254614 0.05080578 -1.5681537 0.16350222 -0.5881948 0.13251524 0.19727796 -0.11716812 0.46636766 0.4940711 0.01271505 -0.46001723 0.7983694 1.9221699 -0.10699307 -0.18672854 -0.0916407 -0.8578931 0.48898342 0.6577561 -0.15493521 -0.62694454 -0.67156637 0.12080643 -0.13066863 0.8521132 -0.7631545 -0.15259737 -0.5515687 0.34702057 -0.5382528 0.5900266 -0.8150236 0.72342545 -0.07363793 -0.09162544 -0.5783633 0.308005 0.5138709 0.06282586 0.26088482 0.87326765 -0.22418581 -0.7662097 0.6107207 0.1342798 -0.00832468 0.9424832 -0.19110006 -0.46731314 -0.15920071 -0.5844691 -0.05074968 0.82459503 0.50111413 0.4117104 -1.5490134 -0.71878684 0.44346818 0.44650438 0.4967658 0.67028385 0.32343856 -0.83437085 0.23680384 -0.2219241 -0.8406131 -0.9643851 0.63636875 0.3958961 0.01381419 -0.72467273 0.8543764 1.0591533 -0.56485045 -0.06340134 -0.51246923 0.2848161 -0.3149761 0.46453196 0.07794303 -0.09083851 -0.59084 -0.10128202 0.8063703 0.5282939 0.04223723 1.2564555 -0.30442864 -0.14853263 0.38426265 1.0478294 0.248794 -1.7132348 -0.31385797 -0.7546473 -0.8886264 0.09834963 -0.7767619 -0.9518898 0.7725137 0.28161928 0.12681462 1.3360246 0.144001 0.3806291 0.05282168 0.6245047 -0.5565379 0.19446121 0.22702909 1.2797918 -1.046682 0.11910101 -0.8746679 -0.5369502 0.7665338 0.53285885 -0.13579684 0.59694076 0.26505193 -0.25961632 -0.17421448 -0.5314912 -0.38023594 0.66120946 0.7292662 0.1089993 0.36615872 0.45371523 0.2621173 -0.35066184 -0.37489134 0.39721826 0.5849066 -0.2070999 -0.86690384 -0.18935478 -0.06632924 0.10110931 -0.17479308 -0.09510859 0.737256 0.43979144 0.68616045 0.06463064 -0.07791269 0.57823706 -0.37855712 0.9080503 -0.79472244 -0.16433449 0.3440128 -0.03600975 -0.75163966 -0.43288672 -0.4641555 -0.6823998 -0.25910884 -0.10468124 -0.4564639 0.66428304 0.46488583 0.4850073 0.80524266 0.7248789 -1.236636 0.1117066 -0.5160973 -0.72483873 0.50363904 0.40397176 -0.79745036 -0.16578652 0.39537176]
[9.281058311462402, -3.138479471206665]
f7ed5d38-dd73-4453-8874-b972aaa88cfe
learning-to-guide-a-saturation-based-theorem
2106.03906
null
https://arxiv.org/abs/2106.03906v1
https://arxiv.org/pdf/2106.03906v1.pdf
Learning to Guide a Saturation-Based Theorem Prover
Traditional automated theorem provers have relied on manually tuned heuristics to guide how they perform proof search. Recently, however, there has been a surge of interest in the design of learning mechanisms that can be integrated into theorem provers to improve their performance automatically. In this work, we introduce TRAIL, a deep learning-based approach to theorem proving that characterizes core elements of saturation-based theorem proving within a neural framework. TRAIL leverages (a) an effective graph neural network for representing logical formulas, (b) a novel neural representation of the state of a saturation-based theorem prover in terms of processed clauses and available actions, and (c) a novel representation of the inference selection process as an attention-based action policy. We show through a systematic analysis that these components allow TRAIL to significantly outperform previous reinforcement learning-based theorem provers on two standard benchmark datasets (up to 36% more theorems proved). In addition, to the best of our knowledge, TRAIL is the first reinforcement learning-based approach to exceed the performance of a state-of-the-art traditional theorem prover on a standard theorem proving benchmark (solving up to 17% more problems).
['Achille Fokoue', 'Michael Witbrock', 'Kavitha Srinivas', 'Ndivhuwo Makondo', 'Pavan Kapanipathi', 'Shajith Ikbal', 'Cristina Cornelio', 'Vernon Austil', 'Bassem Makni', 'Maxwell Crouse', 'Ibrahim Abdelaziz']
2021-06-07
null
null
null
null
['automated-theorem-proving', 'automated-theorem-proving']
['miscellaneous', 'reasoning']
[ 2.49995068e-01 5.59688091e-01 -3.95725161e-01 -1.06677867e-01 -7.85428822e-01 -7.17797041e-01 6.04364574e-01 3.75274330e-01 2.17229174e-03 7.35984862e-01 -1.30665690e-01 -1.26885962e+00 -2.40658432e-01 -1.17407286e+00 -1.34173119e+00 -5.91273233e-02 -3.88724893e-01 6.10730290e-01 4.53462213e-01 -2.71762103e-01 1.98797598e-01 3.55409026e-01 -1.44802403e+00 5.04027843e-01 7.39355326e-01 8.57947528e-01 -2.78637737e-01 8.33152831e-01 -1.44636324e-02 1.52891231e+00 -4.39975321e-01 -1.24316402e-01 1.71076685e-01 -5.23682475e-01 -1.34830940e+00 -6.34841502e-01 7.09064186e-01 -6.92830026e-01 -5.59573174e-01 1.10769439e+00 -6.33444786e-02 -7.09226429e-02 1.73665866e-01 -1.49374235e+00 -4.90769029e-01 1.27839935e+00 -1.59685642e-01 4.31409657e-01 5.05654752e-01 5.09555936e-01 1.36272550e+00 1.37992367e-01 8.43166888e-01 1.45531213e+00 3.86569232e-01 6.69945359e-01 -1.44664741e+00 -6.49606347e-01 1.83607638e-01 6.28505349e-01 -9.34325337e-01 -4.89176661e-01 7.76338518e-01 -2.25959539e-01 1.56446385e+00 2.23751068e-01 8.53696108e-01 8.95265639e-01 2.84700453e-01 9.69332516e-01 9.02203262e-01 -5.63234806e-01 2.99359709e-01 -8.69305357e-02 1.33430406e-01 1.24637222e+00 3.27436775e-01 4.21847343e-01 -2.60848433e-01 -1.55909598e-01 3.98606092e-01 -5.48208356e-01 2.11387011e-03 -7.79262245e-01 -1.09402704e+00 9.20576811e-01 4.73240644e-01 2.45735168e-01 -2.89001185e-02 8.61988842e-01 7.70752728e-01 4.49041426e-01 -1.29363224e-01 1.20718241e+00 -6.63796961e-01 -6.63680881e-02 -8.23638082e-01 7.04881728e-01 1.07150650e+00 7.24655926e-01 5.93194664e-01 1.98976815e-01 -4.31876183e-01 -1.26351744e-01 2.32721180e-01 4.40715224e-01 -1.72370166e-01 -1.22692740e+00 5.07322729e-01 8.96927297e-01 -7.73830339e-02 -6.53932154e-01 -3.30881357e-01 -2.60077685e-01 -1.29113704e-01 3.85093927e-01 4.26587284e-01 -2.07556993e-01 -5.24279714e-01 1.91902208e+00 2.32872218e-01 2.17342675e-01 1.12600960e-01 5.27978837e-01 5.78849733e-01 6.38989806e-01 -2.85993427e-01 5.30344881e-02 1.02450299e+00 -6.24910116e-01 -3.15773368e-01 -4.12154235e-02 7.66718328e-01 3.46917920e-02 6.81109846e-01 5.24400055e-01 -1.31055892e+00 -3.65443230e-02 -1.37993658e+00 6.20467067e-02 -4.36753839e-01 -5.17614245e-01 1.19107211e+00 3.55212152e-01 -1.18716013e+00 7.42977917e-01 -8.73107374e-01 -3.77840921e-02 7.88205504e-01 6.11395299e-01 -2.89036781e-02 -3.56249094e-01 -1.25757575e+00 9.86357570e-01 6.58875763e-01 -6.14944063e-02 -1.32615960e+00 -1.01197302e+00 -1.17874801e+00 3.90124500e-01 1.01564801e+00 -8.25369000e-01 1.76696467e+00 -8.72217536e-01 -1.31975472e+00 3.44651610e-01 6.31354898e-02 -9.25335169e-01 2.93941408e-01 1.30251318e-01 -7.99699947e-02 2.43130744e-01 2.76218187e-02 7.28132308e-01 6.13456666e-01 -1.05990171e+00 -7.39697337e-01 -9.68772620e-02 9.04614449e-01 -3.63886416e-01 2.23456517e-01 -1.01194270e-01 4.16036434e-02 2.25791276e-01 -4.87214178e-01 -6.34074390e-01 1.90490521e-02 -1.25549421e-01 -4.52566385e-01 -7.17413366e-01 7.65742183e-01 -4.76002783e-01 1.06137669e+00 -1.82471085e+00 2.42951602e-01 2.29355261e-01 5.89367211e-01 3.82312626e-01 -1.95438057e-01 3.30763131e-01 -3.17115663e-03 2.89183468e-01 1.62019059e-01 3.57301772e-01 5.40601730e-01 5.02061509e-02 -2.38672405e-01 3.95838410e-01 5.70512176e-01 1.33501649e+00 -1.45040762e+00 -5.16974807e-01 1.88930348e-01 -9.15396065e-02 -8.62258852e-01 5.65823466e-02 -1.06529784e+00 -1.58086464e-01 -5.71140349e-01 6.94701314e-01 3.53527308e-01 -2.59028941e-01 5.15845835e-01 7.29857236e-02 -2.58226749e-02 8.09344471e-01 -6.87636673e-01 1.59083259e+00 -2.64239192e-01 8.12834501e-01 1.41966537e-01 -1.20684862e+00 3.07446986e-01 3.39515746e-01 1.78792998e-01 -9.11499560e-01 3.46806228e-01 1.32780671e-01 5.22234976e-01 -3.74528408e-01 2.26986334e-01 -9.20377299e-03 2.46413369e-02 5.96760809e-01 1.39849409e-01 -3.48456681e-01 7.60868371e-01 4.45583045e-01 1.66122544e+00 3.34386975e-01 1.09751988e-02 -1.58475727e-01 3.66152912e-01 2.99837142e-01 4.00131643e-01 1.16805482e+00 -8.91913846e-02 -2.66154677e-01 1.19916940e+00 -5.99482596e-01 -1.15860522e+00 -6.33685768e-01 1.90680295e-01 9.23065960e-01 -2.54346907e-01 -5.91239214e-01 -7.76937902e-01 -9.05251980e-01 3.79677534e-01 9.81024504e-01 -7.26227582e-01 -6.04058683e-01 -5.09788692e-01 1.11595452e-01 9.15390551e-01 5.74524581e-01 4.33459252e-01 -1.46032715e+00 -8.08364868e-01 1.94566354e-01 2.19777822e-01 -8.96537066e-01 6.86619803e-02 4.68932629e-01 -6.91096306e-01 -1.75728667e+00 2.41366357e-01 -6.01224840e-01 3.98256391e-01 -1.19976759e-01 1.27323198e+00 4.24836785e-01 -1.69942379e-01 3.03828955e-01 -1.35653943e-01 -3.43778580e-01 -8.63253534e-01 1.48234144e-01 -2.05819607e-01 -7.90607154e-01 2.17653424e-01 -4.59095806e-01 -4.84669618e-02 -3.68662685e-01 -7.55704701e-01 1.53167233e-01 5.17988026e-01 8.19602430e-01 -1.26710720e-02 3.45864326e-01 1.93144843e-01 -1.01670420e+00 7.49681771e-01 -3.31007510e-01 -1.42632008e+00 3.24866474e-01 -7.13484943e-01 5.74153781e-01 9.55032468e-01 -1.52681768e-01 -5.23376822e-01 -2.78549492e-01 3.25827956e-01 -4.76744622e-01 -1.92912128e-02 7.24397719e-01 1.75575897e-01 -1.41680047e-01 7.63635218e-01 9.47007537e-02 -4.10789438e-02 3.26131761e-01 3.40329677e-01 6.53885603e-02 4.23026353e-01 -1.19713724e+00 1.16065311e+00 -4.99751046e-02 4.76916254e-01 2.01120856e-03 -9.34177339e-01 4.44912389e-02 -2.01348171e-01 1.97826445e-01 4.37898308e-01 -3.98035973e-01 -1.60692501e+00 -7.63632655e-02 -1.17937541e+00 -1.03762496e+00 -2.55861342e-01 3.79887596e-02 -7.99180150e-01 3.42574418e-02 -6.35646701e-01 -8.45395625e-01 -4.19192523e-01 -1.19699526e+00 8.20009649e-01 1.49270343e-02 -2.62012929e-01 -8.00516009e-01 1.15481079e-01 3.06310475e-01 3.36964756e-01 3.89217466e-01 1.40601695e+00 -6.48145080e-01 -8.82813632e-01 -8.99951980e-02 -5.10717154e-01 2.95881689e-01 -1.69484407e-01 6.28318787e-02 -8.73171568e-01 -2.55662799e-01 -6.66450202e-01 -6.24389768e-01 7.45058358e-01 2.94838488e-01 1.17484248e+00 -5.85603416e-01 -2.89440125e-01 4.24167603e-01 1.19417739e+00 2.10460305e-01 5.42018056e-01 7.06187069e-01 3.20922196e-01 -1.04456536e-01 2.44697630e-01 5.64615615e-02 2.65084624e-01 2.81563580e-01 8.11482072e-01 4.73964289e-02 6.37397077e-03 -4.81547266e-01 2.91392684e-01 -2.29527920e-01 1.04005843e-01 6.28955364e-02 -1.18099225e+00 4.91939455e-01 -1.92741358e+00 -1.32664788e+00 3.38718563e-01 1.82105803e+00 1.15888286e+00 6.37280524e-01 1.20486379e-01 3.92466933e-01 2.82512695e-01 -1.59424786e-02 -8.55608165e-01 -9.06567931e-01 4.25212741e-01 5.82927465e-01 3.99097055e-01 6.98867261e-01 -1.10313165e+00 1.16191399e+00 6.72524548e+00 3.71865392e-01 -9.63898122e-01 -4.45739299e-01 6.68828189e-02 -3.27531211e-02 -3.05991828e-01 1.97595805e-01 -4.11725581e-01 -1.27413586e-01 1.11759090e+00 -2.34243065e-01 1.14575195e+00 9.37243462e-01 -9.27860215e-02 -7.49373212e-02 -1.72815454e+00 2.15113059e-01 -1.04142129e-01 -1.81930399e+00 3.38076949e-02 2.85807960e-02 5.30820787e-01 -8.43191519e-02 -6.02283068e-02 9.08620417e-01 7.24444151e-01 -1.25449538e+00 8.23540926e-01 2.29251459e-01 6.52188301e-01 -9.91571188e-01 8.50713491e-01 1.50114343e-01 -7.38513410e-01 -3.44902694e-01 7.51236361e-03 -4.10597235e-01 -6.14064872e-01 3.27987261e-02 -1.36451900e+00 4.74777520e-01 4.15987134e-01 7.11514294e-01 -6.23892784e-01 8.70330811e-01 -6.20253265e-01 8.57100725e-01 -1.83366567e-01 -4.48504388e-01 5.93607068e-01 4.90738750e-01 4.60496485e-01 1.06517267e+00 -4.02803123e-01 -8.73559788e-02 1.86452284e-01 1.49537528e+00 -2.14301690e-01 -6.76404536e-01 -6.40959501e-01 -5.91439247e-01 3.73412073e-01 1.02737105e+00 -4.29052889e-01 -4.14240688e-01 -2.16420621e-01 2.25220963e-01 6.31421089e-01 3.95727903e-01 -1.01694334e+00 -5.02915680e-01 3.86602253e-01 1.04286596e-01 5.74058473e-01 -1.55139044e-01 -1.48014945e-03 -8.97454858e-01 -3.63951661e-02 -1.42536664e+00 1.80061355e-01 -8.29452395e-01 -6.69186175e-01 -8.84463638e-03 1.54407546e-01 -2.51372099e-01 -3.64204675e-01 -7.56176293e-01 -5.55492938e-01 7.07883239e-01 -1.76797092e+00 -9.30118918e-01 1.07112467e-01 4.28638548e-01 5.78212477e-02 -2.33971491e-01 9.12719071e-01 -1.48529708e-01 -6.11615121e-01 5.63965261e-01 -3.23652744e-01 3.96210253e-01 5.41620590e-02 -1.53608024e+00 3.74125332e-01 8.99105728e-01 3.40068713e-02 7.77090490e-01 8.30068290e-01 -2.77655125e-01 -2.40269446e+00 -8.26594412e-01 4.53536034e-01 -3.23236436e-01 1.18361306e+00 -3.49950343e-01 -6.47862077e-01 9.91755545e-01 2.46402338e-01 -3.36792618e-02 -6.42586406e-03 6.30017161e-01 -6.89717770e-01 -2.86857486e-01 -1.01413155e+00 6.27764702e-01 7.17553318e-01 -6.42380834e-01 -7.90007770e-01 4.83642489e-01 8.75785112e-01 -7.74557233e-01 -5.65843523e-01 2.28635639e-01 4.37044382e-01 -8.04090202e-01 6.13814414e-01 -1.05868518e+00 8.38670671e-01 -3.85660082e-01 3.45791057e-02 -8.81288648e-01 -4.18040931e-01 -7.79199779e-01 -8.32688272e-01 7.97007501e-01 4.66680020e-01 -3.81533861e-01 7.30197370e-01 4.89953071e-01 -1.83420718e-01 -1.05381727e+00 -8.24874341e-01 -5.88067055e-01 2.43083119e-01 -5.42730570e-01 6.03756964e-01 7.89209247e-01 6.41929090e-01 5.00557780e-01 3.34417850e-01 1.85752064e-01 3.42958629e-01 3.59342784e-01 6.35433793e-01 -1.13212943e+00 -4.27409381e-01 -1.00794685e+00 -3.82344604e-01 -5.55651665e-01 7.64829457e-01 -1.28443050e+00 1.13435730e-01 -1.72316957e+00 3.52133125e-01 -2.54340112e-01 -4.83817935e-01 1.07773483e+00 1.96741402e-01 -4.25365001e-01 1.68628301e-02 -6.61133051e-01 -1.04508686e+00 1.34557679e-01 1.16284132e+00 -6.98868871e-01 9.12039205e-02 -2.05755353e-01 -9.71426129e-01 3.23558927e-01 8.02527606e-01 -3.36911380e-01 -3.49122852e-01 -3.87259871e-01 7.10990310e-01 2.77472466e-01 5.65977156e-01 -9.83658552e-01 3.06219935e-01 -3.95289719e-01 7.88274780e-02 -1.46331429e-01 -1.26840398e-01 -5.95296204e-01 -5.39070189e-01 7.16952920e-01 -5.49181163e-01 -1.54768720e-01 7.29336023e-01 2.74376214e-01 1.29664272e-01 -4.27952968e-02 4.05397564e-01 -1.16648555e-01 -6.60390973e-01 1.88657075e-01 -4.34942991e-01 1.48080185e-01 8.15774858e-01 1.63727298e-01 -4.88486439e-01 -1.61668807e-01 -1.05637886e-01 6.02654457e-01 2.91914731e-01 1.82154074e-01 4.05147463e-01 -9.00532782e-01 -5.53287268e-01 -6.61981478e-02 -8.43208507e-02 2.46809661e-01 -4.11236286e-01 7.78382182e-01 -5.15255094e-01 8.61719787e-01 -1.56859532e-01 -1.88843101e-01 -1.08270633e+00 9.40338254e-01 7.63598919e-01 -5.54412723e-01 -6.61330283e-01 5.58114409e-01 -3.14431012e-01 -3.68822485e-01 3.77552390e-01 -8.24116230e-01 2.90558994e-01 -5.39133132e-01 4.51759398e-01 1.34070948e-01 5.01514487e-02 3.07312548e-01 -5.62035739e-01 -1.99086770e-01 -1.43493325e-01 1.75136537e-03 1.49164355e+00 8.73549998e-01 -3.07567984e-01 2.75933146e-01 6.16169333e-01 -2.36563906e-01 -9.83320653e-01 -1.00868747e-01 1.80456825e-02 4.50051129e-02 3.50363284e-01 -1.22098505e+00 -7.19440699e-01 5.69071293e-01 -2.29382738e-01 4.34917569e-01 8.32744658e-01 2.16478873e-02 4.95608270e-01 1.08345151e+00 4.19338316e-01 -8.87186289e-01 -8.71766880e-02 8.45541775e-01 5.53292572e-01 -1.05667174e+00 2.56388694e-01 1.59048006e-01 5.06458171e-02 1.30587173e+00 5.58572650e-01 -3.46014172e-01 -2.07200907e-02 5.27226925e-01 -5.40934384e-01 -2.75025666e-01 -1.19617331e+00 8.57775360e-02 1.62000865e-01 5.30739129e-01 4.06180292e-01 -1.25803620e-01 3.14517349e-01 3.87937874e-01 -2.31160924e-01 3.38808686e-01 4.58821118e-01 1.12927294e+00 -4.65636253e-01 -9.27150309e-01 -2.10935608e-01 4.99913603e-01 -2.42817640e-01 -1.87881082e-01 -4.84277457e-01 1.17215073e+00 9.95415542e-03 1.04091406e+00 -2.23875776e-01 -1.70687914e-01 1.37664333e-01 2.13816062e-01 1.21903741e+00 -6.98497713e-01 -6.27925217e-01 -6.65579319e-01 3.54344159e-01 -9.27901804e-01 -3.71501669e-02 -5.76968312e-01 -1.42676508e+00 -8.11895251e-01 -3.84469241e-01 2.43482381e-01 3.59464854e-01 1.11621475e+00 1.21629395e-01 1.04515779e+00 1.60154223e-01 -5.42344987e-01 -7.04178751e-01 -6.66219592e-01 -3.56198698e-01 -4.35331613e-02 6.49758160e-01 -5.07707477e-01 -2.03480929e-01 -4.65633541e-01]
[8.895305633544922, 7.1019744873046875]
888f03f2-56ad-4e62-88e7-f2cd51e18008
understanding-the-importance-of-heart-sound
2005.10480
null
https://arxiv.org/abs/2005.10480v2
https://arxiv.org/pdf/2005.10480v2.pdf
A Robust Interpretable Deep Learning Classifier for Heart Anomaly Detection Without Segmentation
Traditionally, abnormal heart sound classification is framed as a three-stage process. The first stage involves segmenting the phonocardiogram to detect fundamental heart sounds; after which features are extracted and classification is performed. Some researchers in the field argue the segmentation step is an unwanted computational burden, whereas others embrace it as a prior step to feature extraction. When comparing accuracies achieved by studies that have segmented heart sounds before analysis with those who have overlooked that step, the question of whether to segment heart sounds before feature extraction is still open. In this study, we explicitly examine the importance of heart sound segmentation as a prior step for heart sound classification, and then seek to apply the obtained insights to propose a robust classifier for abnormal heart sound detection. Furthermore, recognizing the pressing need for explainable Artificial Intelligence (AI) models in the medical domain, we also unveil hidden representations learned by the classifier using model interpretation techniques. Experimental results demonstrate that the segmentation plays an essential role in abnormal heart sound classification. Our new classifier is also shown to be robust, stable and most importantly, explainable, with an accuracy of almost 100% on the widely used PhysioNet dataset.
['Houman Ghaemmaghami', 'Sridha Sridharan', 'Tharindu Fernando', 'Theekshana Dissanayake', 'Clinton Fookes', 'Simon Denman']
2020-05-21
null
null
null
null
['sound-classification']
['audio']
[ 7.64463782e-01 5.31025767e-01 -5.97904697e-02 -3.13697577e-01 -3.82376760e-01 -3.38918895e-01 1.86523274e-01 3.16283107e-01 -1.05578840e-01 4.41386431e-01 2.53499858e-02 -6.01495206e-01 -2.85796732e-01 -5.63355267e-01 -4.57368940e-02 -7.13089705e-01 4.58565503e-02 5.16083598e-01 2.48831790e-02 1.36570945e-01 3.23002458e-01 4.86194432e-01 -1.56167924e+00 2.47233026e-02 7.52130270e-01 8.60117137e-01 -2.73338616e-01 1.15086913e+00 3.63032408e-02 9.70764160e-01 -7.23758340e-01 -1.34246305e-01 -1.47136033e-01 -1.08830786e+00 -1.22598279e+00 1.55060410e-01 -5.68167344e-02 -3.29613835e-02 3.47847939e-01 7.02867031e-01 5.64612746e-01 -2.25192472e-01 7.01737344e-01 -1.02049637e+00 -2.19421312e-01 7.19114840e-01 1.10417567e-01 4.32350785e-01 3.68576646e-01 -4.71778102e-02 1.19592702e+00 -5.66973686e-01 3.20757419e-01 6.13609135e-01 7.95163810e-01 6.75236881e-01 -1.20921850e+00 -4.07279223e-01 -3.65048319e-01 1.10455595e-01 -1.21313930e+00 -5.11423647e-01 9.70058203e-01 -7.30380714e-01 7.30106235e-01 8.69321585e-01 9.46309686e-01 6.47963047e-01 1.75546482e-01 3.45252693e-01 1.06470490e+00 -8.64047587e-01 2.02982962e-01 3.08636725e-01 5.74049652e-01 7.91881621e-01 5.29472709e-01 4.21896614e-02 -2.41241336e-01 -1.97250143e-01 4.72543001e-01 -1.39422640e-01 -3.56635481e-01 2.02716589e-01 -1.08625090e+00 7.56426990e-01 -1.62433654e-01 8.42840612e-01 -5.53786755e-01 -8.27639550e-02 3.17485064e-01 2.36415714e-01 3.87923747e-01 7.23891914e-01 -3.40060025e-01 -2.24432215e-01 -1.26865792e+00 -3.36766765e-02 1.03266203e+00 8.09445307e-02 3.72098446e-01 6.55102804e-02 7.33570904e-02 5.60407639e-01 5.08221984e-01 2.58622348e-01 4.84310359e-01 -9.24985111e-01 -1.21509247e-01 6.10016406e-01 -2.66873419e-01 -1.23157549e+00 -6.29201233e-01 -5.94910443e-01 -8.55658114e-01 -2.85477545e-02 4.70887721e-01 1.31495884e-02 -6.20968461e-01 1.33445418e+00 2.22812876e-01 1.32927895e-01 1.39462411e-01 7.23921001e-01 8.97257686e-01 1.53288335e-01 1.81280643e-01 -4.42394316e-01 1.44651544e+00 -5.21445096e-01 -7.77455330e-01 8.06801319e-02 5.04856706e-01 -8.20354760e-01 6.85128629e-01 6.85580492e-01 -8.80209565e-01 -6.62205040e-01 -9.98465300e-01 4.38605845e-02 -1.06384128e-01 1.09731294e-01 5.64574957e-01 1.01719022e+00 -6.91130817e-01 8.46466482e-01 -8.55395496e-01 -3.61016363e-01 2.27461517e-01 3.15643996e-01 -9.38971788e-02 6.24906957e-01 -1.14008904e+00 8.15765858e-01 1.48884252e-01 3.78937125e-01 -3.59585106e-01 -4.41416502e-01 -4.78757292e-01 5.03309704e-02 -7.85823092e-02 -7.63268113e-01 1.00095677e+00 -8.71033967e-01 -1.37544966e+00 9.60915685e-01 -1.70266271e-01 -5.81889033e-01 4.09525335e-01 2.59645116e-02 -5.67521274e-01 5.00969350e-01 -8.96198526e-02 1.54392585e-01 1.10946512e+00 -1.15802181e+00 -5.65711558e-01 -1.77091032e-01 -2.92169839e-01 -3.71072441e-01 -1.65361181e-01 -1.47166729e-01 1.44156948e-01 -6.03245497e-01 5.75977981e-01 -1.00165689e+00 -4.94317897e-03 -4.12702918e-01 -6.06119871e-01 -2.54317194e-01 5.31633317e-01 -9.66029108e-01 1.63244188e+00 -2.08820987e+00 -7.81050846e-02 5.04980087e-01 6.39094293e-01 2.26019844e-01 4.86506581e-01 2.13974968e-01 -3.65516186e-01 4.17138875e-01 -4.75295514e-01 -1.93160117e-01 -1.01844825e-01 2.82257587e-01 -3.94559950e-01 4.18581426e-01 2.44847253e-01 6.59862459e-01 -5.11178613e-01 -7.92394876e-01 3.10409665e-01 5.07064581e-01 -6.33627594e-01 1.83284923e-01 3.54424536e-01 6.93509877e-01 -5.84114909e-01 5.66608965e-01 1.57028973e-01 -3.29597205e-01 1.63651064e-01 -5.87627925e-02 -1.51829749e-01 3.29620928e-01 -9.99234736e-01 1.02390587e+00 -1.50584340e-01 6.76596820e-01 -2.90278167e-01 -1.42917442e+00 1.14262629e+00 8.42292011e-01 6.91232085e-01 -1.87119856e-01 2.36688018e-01 4.27029848e-01 3.82704526e-01 -9.28744197e-01 1.04288734e-01 -4.68066961e-01 3.73680592e-02 4.97614682e-01 -1.41045853e-01 -2.96197832e-01 -9.56023335e-02 -2.04383746e-01 8.83894026e-01 -9.59996954e-02 6.01276934e-01 -3.40340346e-01 7.89932847e-01 6.88314438e-03 4.86755878e-01 6.85302913e-01 -4.07057226e-01 8.23970735e-01 4.89187747e-01 -7.43271172e-01 -5.81369877e-01 -5.59930384e-01 -4.41880435e-01 6.50951684e-01 -2.80906767e-01 -1.49129391e-01 -1.02037311e+00 -6.35980308e-01 -2.21636534e-01 5.28103232e-01 -7.24458814e-01 -2.91812271e-01 -5.66927195e-01 -8.80575895e-01 8.47208858e-01 5.16612351e-01 2.00265199e-01 -1.23247302e+00 -1.43577611e+00 3.46678972e-01 -2.60070503e-01 -7.64876306e-01 1.58630669e-01 3.35350245e-01 -1.01909149e+00 -1.34979200e+00 -5.03598869e-01 -5.50452828e-01 5.25560498e-01 -3.91106427e-01 1.15353799e+00 7.45414138e-01 -7.57078111e-01 4.04661745e-01 -4.90332007e-01 -5.80663443e-01 -7.30092883e-01 1.59823582e-01 -2.15839639e-01 2.24480391e-01 3.15220267e-01 -5.32506704e-01 -5.06164730e-01 -2.62776688e-02 -6.52802289e-01 -5.96458763e-02 5.32791972e-01 6.93315804e-01 3.83808970e-01 1.41938388e-01 8.03194404e-01 -1.19094646e+00 4.71179634e-01 -2.85427511e-01 1.46836322e-02 1.36367410e-01 -9.27526712e-01 -2.49587119e-01 4.84362930e-01 -9.66948494e-02 -7.07474589e-01 1.50325105e-01 -5.11785090e-01 -7.31358724e-03 -5.82710147e-01 3.79551202e-01 1.03642650e-01 1.16348982e-01 6.64198101e-01 3.08941424e-01 2.35415846e-01 -5.16027033e-01 -1.79299608e-01 7.17982113e-01 4.78354603e-01 -2.40248695e-01 6.62568033e-01 3.83908540e-01 2.76019722e-01 -1.27689958e+00 -7.15817094e-01 -4.36136514e-01 -8.61814797e-01 -3.53292912e-01 1.28162611e+00 -2.31681705e-01 -6.52343929e-01 1.92765325e-01 -9.80624676e-01 1.03423379e-01 -5.73855281e-01 6.53633535e-01 -5.03975511e-01 5.29985130e-01 -4.37088788e-01 -1.08364964e+00 -6.07218385e-01 -8.04807246e-01 8.17383707e-01 3.31008621e-02 -9.52882230e-01 -1.18313742e+00 1.04593918e-01 7.43347704e-01 2.97493041e-01 3.94358397e-01 1.21818173e+00 -1.01968801e+00 -9.29701626e-02 -3.88627261e-01 1.93136886e-01 4.58252698e-01 3.40667725e-01 1.45786956e-01 -1.41556454e+00 1.10521689e-01 3.27260584e-01 7.93272778e-02 8.59447360e-01 4.38129455e-01 9.92483735e-01 -1.91513270e-01 -4.78891004e-03 3.91854793e-01 9.59933162e-01 2.65903413e-01 5.04901528e-01 -5.15866131e-02 4.45784360e-01 7.66977549e-01 3.72441858e-01 2.43600056e-01 1.83403060e-01 1.67411521e-01 1.72505349e-01 -4.32259768e-01 -2.11092070e-01 -1.16431704e-02 -1.27872244e-01 1.33698702e+00 -5.29212475e-01 1.73218235e-01 -1.00324035e+00 3.44390035e-01 -1.39735520e+00 -1.00786066e+00 -5.46575904e-01 1.89034033e+00 6.09692752e-01 3.16478968e-01 2.62725085e-01 1.29769135e+00 5.74278116e-01 -8.00294951e-02 5.39457332e-03 -5.17783105e-01 2.36762300e-01 5.61343372e-01 -2.39444584e-01 6.57930195e-01 -1.15541840e+00 4.53143299e-01 6.91421413e+00 1.11493051e-01 -1.35834885e+00 -1.32475689e-01 6.98108435e-01 5.29654264e-01 -1.62226126e-01 8.42160080e-03 -2.97139645e-01 3.18627208e-01 9.51525927e-01 1.22741520e-01 6.63350821e-02 6.60998166e-01 2.95667857e-01 -1.39869437e-01 -1.09863651e+00 7.05251157e-01 7.93284401e-02 -9.72320080e-01 -1.10085040e-01 -3.02894064e-03 1.27722353e-01 -7.12631702e-01 -1.23478249e-01 2.39153225e-02 -7.48458385e-01 -1.05183733e+00 5.52234769e-01 8.54930580e-01 4.51977849e-01 -5.08394659e-01 8.93358350e-01 3.21416825e-01 -1.06176913e+00 -4.38170582e-02 -7.17140287e-02 -3.32071781e-01 -1.01095900e-01 7.37701833e-01 -1.22727323e+00 4.43058103e-01 4.01896447e-01 4.39172804e-01 -6.96783185e-01 8.92679214e-01 -1.77504450e-01 1.20480978e+00 -1.02888666e-01 5.38904816e-02 -3.00491482e-01 1.19678758e-01 5.59870839e-01 1.27602410e+00 2.48821080e-01 1.41648188e-01 -2.13474315e-02 9.57082689e-01 5.32921851e-01 2.82097638e-01 -5.97558618e-01 -3.04905802e-01 1.17787801e-01 1.11632419e+00 -1.28412175e+00 -3.60282868e-01 -4.32383232e-02 7.44432271e-01 -4.36629087e-01 6.42616823e-02 -6.77503765e-01 -4.75335598e-01 1.56607524e-01 1.76600367e-01 1.17374428e-01 3.47574294e-01 -7.01357305e-01 -8.56647015e-01 -1.97489768e-01 -7.63768673e-01 4.75588888e-01 -3.58360052e-01 -7.34487772e-01 7.05946267e-01 -1.70613229e-01 -1.13745236e+00 -6.06218576e-01 -3.68780464e-01 -6.87343359e-01 7.98465371e-01 -1.44926870e+00 -9.85818565e-01 -1.31828874e-01 3.43678355e-01 4.79581892e-01 5.37786670e-02 1.29448938e+00 3.26714367e-01 -5.68742931e-01 2.57547379e-01 -6.42638743e-01 2.64496624e-01 8.13334733e-02 -1.31259656e+00 -9.97324660e-02 6.49393559e-01 3.39263260e-01 7.75053501e-01 8.42947662e-01 -5.30409038e-01 -9.84913826e-01 -6.56300783e-01 1.55261028e+00 -5.73020399e-01 1.83560014e-01 7.39676058e-02 -9.84593034e-01 3.52862358e-01 -2.26478949e-02 1.51406020e-01 1.09670019e+00 -4.73782867e-02 1.18859515e-01 3.44520956e-02 -1.05589128e+00 -6.90940693e-02 4.79951799e-01 -5.16012430e-01 -1.16707683e+00 -1.33594707e-01 2.53485054e-01 1.03396587e-01 -9.26429272e-01 4.20868695e-01 9.37576890e-01 -1.17140126e+00 7.84606516e-01 -5.42347133e-01 3.21358144e-01 -3.48395035e-02 1.41681850e-01 -7.82697380e-01 -2.02529520e-01 -6.36790037e-01 1.49660064e-02 1.27709138e+00 5.91741920e-01 -7.07285404e-01 6.90497100e-01 4.34270978e-01 2.31861845e-02 -7.56322145e-01 -7.60052860e-01 -3.20355147e-01 -1.29687473e-01 -6.73099756e-01 3.25316995e-01 1.14248312e+00 1.36189073e-01 2.56148785e-01 -2.86244482e-01 1.20161563e-01 5.66492677e-01 4.39265311e-01 4.23103809e-01 -1.88442135e+00 -4.08516794e-01 -3.73178273e-01 -4.82076406e-01 -1.97705358e-01 -7.47328624e-02 -8.17135453e-01 4.83029075e-02 -1.32667840e+00 3.90199199e-02 -3.64457250e-01 -4.85737622e-01 3.53969634e-01 -2.50537008e-01 5.15421748e-01 4.13540266e-02 3.14803660e-01 -8.22191462e-02 -2.33141348e-01 1.13018930e+00 2.34826297e-01 -2.48799965e-01 4.62450147e-01 -6.64642811e-01 1.15629876e+00 8.05833519e-01 -6.21355832e-01 -3.88012528e-01 2.42963225e-01 1.36641324e-01 2.40790233e-01 5.41999936e-01 -1.20184803e+00 2.20430247e-03 1.10542610e-01 3.51361662e-01 -3.12621832e-01 1.83520988e-01 -8.88134897e-01 9.56000462e-02 9.02173638e-01 -5.16619802e-01 -3.15191656e-01 -1.76112071e-01 1.85310364e-01 -1.61739424e-01 -4.71753687e-01 8.03387821e-01 -8.47737715e-02 -2.07858637e-01 -2.24352881e-01 -6.91558003e-01 1.92293599e-01 8.20535362e-01 -3.74400109e-01 2.66082793e-01 -2.90536284e-01 -1.36337984e+00 -4.78704870e-01 -1.26563802e-01 1.84408249e-03 6.05148792e-01 -7.24338293e-01 -5.93304396e-01 5.86390674e-01 -1.98751077e-01 -3.57336551e-01 1.31421611e-01 1.23437774e+00 -8.05869699e-01 5.15298188e-01 1.25391381e-02 -7.33891308e-01 -1.60804129e+00 2.95003623e-01 3.35396111e-01 -1.27686381e-01 -1.03248906e+00 7.18629956e-01 -2.12961212e-01 -3.79921533e-02 2.41807893e-01 -7.72524595e-01 -6.14198804e-01 3.66748631e-01 3.13584477e-01 4.86729920e-01 9.88066867e-02 -5.99357545e-01 -3.57601076e-01 7.15474725e-01 5.22611797e-01 5.08303083e-02 1.24728274e+00 -2.67566424e-02 -1.91671953e-01 7.73319900e-01 8.51060987e-01 1.39070461e-02 -3.65219325e-01 3.42908949e-01 3.63251001e-01 -1.96742475e-01 7.11781234e-02 -7.16556370e-01 -9.47946608e-01 1.22661710e+00 6.54565215e-01 1.04871964e+00 1.21100760e+00 -4.57870821e-03 7.03533828e-01 3.20519596e-01 -7.40993321e-02 -9.37628448e-01 -1.22152090e-01 4.22327667e-02 7.73531020e-01 -9.43042994e-01 -2.34172046e-02 -5.58228672e-01 -4.93976384e-01 1.30716395e+00 -1.14449844e-01 -8.07687417e-02 8.84874642e-01 1.76735908e-01 3.02840441e-01 -2.80349046e-01 -5.56636393e-01 -1.68136701e-01 4.67764288e-01 3.65813106e-01 6.92208648e-01 6.51345775e-02 -6.04825497e-01 9.00046468e-01 -5.24705350e-01 2.56347507e-01 4.07772899e-01 9.35790718e-01 -6.92836106e-01 -9.25261199e-01 -4.04038131e-01 4.89732772e-01 -9.72430050e-01 9.87412632e-02 -6.65208280e-01 7.41291344e-01 5.69300473e-01 1.13694870e+00 -2.03805149e-01 -3.39812964e-01 1.51983052e-01 6.80880725e-01 3.17503035e-01 -6.41028941e-01 -7.52841711e-01 2.11591676e-01 1.25801608e-01 -1.86963156e-01 -5.97402811e-01 -6.82217479e-01 -1.38575768e+00 2.49752343e-01 -3.64349633e-01 5.84744990e-01 5.01992464e-01 1.18484557e+00 -5.06338961e-02 8.26881766e-01 6.03369594e-01 -4.77626264e-01 -4.89470512e-01 -8.82896841e-01 -5.67234516e-01 4.46375042e-01 7.55230188e-01 -3.03405821e-01 -6.95412815e-01 4.97390091e-01]
[14.27788257598877, 3.2880403995513916]
649337f7-5be2-4488-831e-d4a2e97a6245
contact-aware-retargeting-of-skinned-motion
2109.07431
null
https://arxiv.org/abs/2109.07431v1
https://arxiv.org/pdf/2109.07431v1.pdf
Contact-Aware Retargeting of Skinned Motion
This paper introduces a motion retargeting method that preserves self-contacts and prevents interpenetration. Self-contacts, such as when hands touch each other or the torso or the head, are important attributes of human body language and dynamics, yet existing methods do not model or preserve these contacts. Likewise, interpenetration, such as a hand passing into the torso, are a typical artifact of motion estimation methods. The input to our method is a human motion sequence and a target skeleton and character geometry. The method identifies self-contacts and ground contacts in the input motion, and optimizes the motion to apply to the output skeleton, while preserving these contacts and reducing interpenetration. We introduce a novel geometry-conditioned recurrent network with an encoder-space optimization strategy that achieves efficient retargeting while satisfying contact constraints. In experiments, our results quantitatively outperform previous methods and we conduct a user study where our retargeted motions are rated as higher-quality than those produced by recent works. We also show our method generalizes to motion estimated from human videos where we improve over previous works that produce noticeable interpenetration.
['Jun Saito', 'Jimei Yang', 'Aaron Hertzmann', 'Duygu Ceylan', 'Ruben Villegas']
2021-09-15
null
http://openaccess.thecvf.com//content/ICCV2021/html/Villegas_Contact-Aware_Retargeting_of_Skinned_Motion_ICCV_2021_paper.html
http://openaccess.thecvf.com//content/ICCV2021/papers/Villegas_Contact-Aware_Retargeting_of_Skinned_Motion_ICCV_2021_paper.pdf
iccv-2021-1
['motion-retargeting']
['computer-vision']
[ 7.65315965e-02 9.65425465e-03 -3.86490941e-01 7.71175250e-02 -4.44168091e-01 -5.16543210e-01 4.47132677e-01 -3.79032075e-01 -4.32986617e-01 3.52430165e-01 8.72842610e-01 2.26577088e-01 1.66937813e-01 -3.91370416e-01 -8.49396169e-01 -3.39455813e-01 -1.78274691e-01 3.52546483e-01 5.85461676e-01 -3.71105731e-01 4.50835377e-02 3.15217346e-01 -1.21201456e+00 3.37663561e-01 3.97318512e-01 2.09275782e-01 2.42038071e-01 1.15069926e+00 4.85786498e-01 6.73400819e-01 -5.19908130e-01 -3.78006846e-01 1.92943275e-01 -4.89195615e-01 -1.00291967e+00 9.03812647e-02 7.37290382e-01 -5.87749720e-01 -7.10577786e-01 7.24844575e-01 7.29087710e-01 4.42690611e-01 6.35270894e-01 -9.02172148e-01 -5.78244865e-01 6.90317810e-01 -6.83314264e-01 1.70437917e-01 6.66373968e-01 1.99464291e-01 9.28251326e-01 -8.56541455e-01 1.04879129e+00 1.41685379e+00 8.62065852e-01 9.04263258e-01 -1.17749441e+00 -3.28925431e-01 2.04460263e-01 -1.03158699e-02 -1.38057446e+00 -6.46658003e-01 7.53330171e-01 -6.15697801e-01 9.98398066e-01 3.78073007e-01 9.56560075e-01 1.59515345e+00 5.62975287e-01 1.04825652e+00 -4.80875708e-02 -2.32841656e-01 -2.95503028e-02 -5.75181901e-01 -2.02100694e-01 7.69570291e-01 -1.00904383e-01 -6.08858513e-03 -7.44159222e-01 -5.60137630e-02 1.38985884e+00 -2.30164863e-02 -4.06045914e-01 -5.27016580e-01 -1.83387494e+00 3.76997560e-01 2.12004766e-01 2.01689273e-01 -2.31096625e-01 5.22133410e-01 3.85721713e-01 -7.32762516e-02 9.86801982e-02 3.00661832e-01 -1.42466992e-01 -6.00168586e-01 -1.15455472e+00 5.97256124e-01 8.87965560e-01 1.19559729e+00 1.27056658e-01 7.17931241e-02 -4.14260417e-01 6.82870030e-01 1.42217398e-01 3.34988803e-01 6.50011718e-01 -1.43459344e+00 6.64572895e-01 1.83258623e-01 3.73037785e-01 -1.31366932e+00 -5.96794367e-01 -1.64234862e-01 -8.71251881e-01 -4.07046340e-02 4.72805738e-01 -1.04053453e-01 -7.38446712e-01 1.90810382e+00 3.09795558e-01 2.47318000e-01 -5.08537553e-02 1.17192173e+00 7.46539831e-01 5.04501402e-01 4.47852053e-02 6.50909394e-02 1.15404177e+00 -1.36654150e+00 -7.79995203e-01 -1.03934675e-01 8.16652715e-01 -9.00804102e-01 1.37101293e+00 1.97909102e-01 -1.40599895e+00 -6.08646929e-01 -1.02263331e+00 -2.58715808e-01 4.89718825e-01 3.42936069e-01 2.24586189e-01 2.23778740e-01 -9.66868877e-01 1.20742011e+00 -1.31357050e+00 -5.01558542e-01 -2.25311652e-01 4.27161366e-01 -4.65490878e-01 5.54665565e-01 -9.95546818e-01 7.49194384e-01 3.23182307e-02 1.32915840e-01 -6.96596920e-01 -4.37860459e-01 -1.07138646e+00 -3.38528603e-01 4.28097099e-01 -1.12309885e+00 1.46954429e+00 -9.36386883e-01 -1.91605830e+00 6.80320740e-01 -9.92799550e-02 -2.53009886e-01 8.39011371e-01 -8.76201570e-01 -2.97774136e-01 1.55391335e-01 1.60467163e-01 8.09845865e-01 8.97957027e-01 -9.62581456e-01 -3.64692181e-01 3.39576695e-03 -3.02368961e-02 5.67244709e-01 -1.69574142e-01 -1.29309744e-01 -9.70317483e-01 -1.31852603e+00 1.60526231e-01 -1.26720929e+00 -2.38759726e-01 1.81042403e-01 -8.72949004e-01 1.82934046e-01 7.62612879e-01 -8.46580625e-01 1.55244684e+00 -2.01954007e+00 9.91308928e-01 1.28854752e-01 2.88162977e-01 -5.77764586e-02 -1.76652446e-02 3.70952547e-01 -1.72495954e-02 9.22749341e-02 -1.37720615e-01 -6.21844828e-01 -6.43776953e-02 2.14070916e-01 -3.35231982e-02 6.81944907e-01 -2.83526570e-01 9.36666906e-01 -8.20732117e-01 -5.82674801e-01 1.80227876e-01 5.41249514e-01 -9.08614576e-01 2.46277899e-01 -9.35298428e-02 6.33119106e-01 -2.87499547e-01 5.31205773e-01 -6.51017129e-02 -2.58999974e-01 2.47716568e-02 -4.71089482e-01 2.30777748e-02 1.35773271e-01 -1.29041064e+00 2.20729327e+00 -1.38121843e-01 3.18285555e-01 -6.11224733e-02 -2.93469489e-01 3.53584170e-01 3.92100155e-01 6.30789995e-01 -2.27501899e-01 9.80408117e-02 -2.06539825e-01 -4.95264046e-02 -7.27711916e-01 8.56976271e-01 2.84244627e-01 -8.18317235e-02 4.00379121e-01 -2.61866361e-01 -7.37987384e-02 -6.88238293e-02 3.10837120e-01 9.29465055e-01 7.06876278e-01 3.23090106e-01 -3.58575076e-01 1.42227679e-01 -1.36135012e-01 6.39830291e-01 4.97425556e-01 -2.28596538e-01 1.08935750e+00 2.35904887e-01 -3.22342128e-01 -1.23904347e+00 -1.14133346e+00 6.46542311e-01 1.20252895e+00 3.48346442e-01 -7.11204648e-01 -9.69345331e-01 -4.22165573e-01 -2.65906543e-01 2.38389432e-01 -6.21292233e-01 -2.75817513e-01 -1.12666476e+00 -1.27859697e-01 8.23783815e-01 1.02417552e+00 2.71196574e-01 -8.61052692e-01 -8.73334527e-01 3.29826742e-01 -6.64960146e-01 -1.03878140e+00 -1.49718702e+00 -4.24016982e-01 -9.71161783e-01 -8.03789794e-01 -1.24879563e+00 -8.08616877e-01 4.83845055e-01 9.64494236e-03 1.01548374e+00 8.25576931e-02 -3.24805319e-01 3.74974370e-01 -3.11359465e-01 3.00453216e-01 -3.11123699e-01 8.07663575e-02 4.66727316e-01 -7.51135424e-02 -3.59058857e-01 -5.99956632e-01 -8.74605834e-01 6.24327958e-01 -6.09786570e-01 4.43119913e-01 1.68687299e-01 7.41878569e-01 5.57449996e-01 -4.21534091e-01 3.15136127e-02 -5.40042758e-01 3.37782264e-01 -1.65032968e-01 7.79576972e-02 -2.34150048e-03 5.12881875e-02 1.89762309e-01 4.14626330e-01 -8.73452425e-01 -9.26506639e-01 4.06506181e-01 -2.11093932e-01 -7.15064585e-01 2.11364046e-01 -3.27439569e-02 -2.33520836e-01 2.09378257e-01 7.47816324e-01 -6.15502857e-02 -1.17443874e-01 -4.81106341e-01 5.48480272e-01 1.03623070e-01 1.14324629e+00 -7.36831725e-01 7.29318440e-01 6.42904162e-01 -2.85132766e-01 -8.84900689e-01 -2.98047811e-01 -3.92348856e-01 -1.14080119e+00 -3.04827720e-01 1.04209042e+00 -7.71878421e-01 -7.17039227e-01 3.40073556e-01 -1.24954152e+00 -5.13949573e-01 -3.99519026e-01 7.04675078e-01 -8.85828257e-01 9.06230390e-01 -1.01368463e+00 -4.81657803e-01 -2.82327712e-01 -1.25507247e+00 1.35254431e+00 5.64644672e-02 -1.16876030e+00 -8.83255661e-01 2.72180319e-01 4.73229922e-02 -1.30791133e-02 3.07292700e-01 6.56897187e-01 -1.76268518e-01 -1.49524555e-01 -1.46802207e-02 5.14641404e-01 -1.95804268e-01 3.21902692e-01 2.67508477e-01 -3.54205757e-01 -4.43169683e-01 -5.02424181e-01 -6.08208440e-02 5.74600756e-01 4.77183700e-01 7.96577096e-01 -6.31646216e-01 -5.28251469e-01 7.94354975e-01 5.94710588e-01 -2.10035890e-02 7.25899160e-01 2.08107293e-01 1.08675623e+00 7.00751901e-01 4.16767687e-01 5.79796612e-01 4.40361768e-01 1.13592613e+00 1.55603036e-01 -2.00684443e-02 -3.57327431e-01 -5.71074843e-01 6.75746977e-01 1.09029746e+00 -6.79890215e-01 -2.56817728e-01 -6.95673764e-01 6.58917427e-01 -2.13575602e+00 -9.57693815e-01 1.83162559e-03 2.22863412e+00 8.53156507e-01 6.17658645e-02 5.63191354e-01 -3.95056978e-02 6.50699198e-01 1.84934869e-01 -5.75914621e-01 -6.58589005e-02 -8.94208625e-02 -5.92034683e-02 3.57549846e-01 7.61988580e-01 -1.16766727e+00 1.30743062e+00 7.02721739e+00 4.10528630e-01 -8.82036686e-01 2.04442758e-02 4.70713004e-02 -5.10458052e-01 -2.04437569e-01 -1.42206565e-01 -6.61163509e-01 2.14107111e-01 4.07086849e-01 1.30772814e-01 2.22320288e-01 6.76568806e-01 4.67489690e-01 2.18966350e-01 -1.51750481e+00 1.18859315e+00 1.80534899e-01 -1.13001692e+00 2.48657212e-01 -1.92040354e-01 5.72751164e-01 -5.21189153e-01 2.79443972e-02 -1.56438202e-01 3.72674346e-01 -1.17443871e+00 1.18303776e+00 7.67707407e-01 7.90847600e-01 -6.19947255e-01 1.54911712e-01 1.92693979e-01 -1.55740392e+00 2.95711249e-01 -1.44717861e-02 -5.51244318e-02 4.53262657e-01 5.12128100e-02 -4.33447242e-01 3.33845168e-01 7.15705931e-01 1.02885830e+00 -2.88358510e-01 7.62993395e-01 -2.76662469e-01 4.50698644e-01 -2.48894140e-01 1.40735090e-01 -6.56558275e-02 8.77190828e-02 1.17362535e+00 1.26990330e+00 2.08845720e-01 1.14594243e-01 2.57173479e-01 7.33902514e-01 1.39646426e-01 2.22401023e-01 -6.30290270e-01 4.78626713e-02 2.17603013e-01 8.54230285e-01 -6.55168772e-01 -3.52842927e-01 -1.13381535e-01 1.85158777e+00 7.05384761e-02 6.05736494e-01 -9.11230922e-01 -2.07990870e-01 7.81560004e-01 3.77567321e-01 2.70919595e-02 -5.50904155e-01 -4.98540141e-02 -1.50960577e+00 1.59668401e-01 -8.68191719e-01 2.99208939e-01 -5.87059140e-01 -8.53503466e-01 4.77768809e-01 7.93338045e-02 -1.34534001e+00 -6.64083004e-01 -1.66703910e-01 -4.92834836e-01 4.59704876e-01 -2.35335127e-01 -1.19075751e+00 -9.03846696e-02 7.00949550e-01 8.55068326e-01 1.23751499e-01 5.92809558e-01 3.41016054e-01 -5.58675587e-01 7.77719796e-01 -4.74996656e-01 3.10622394e-01 7.92051256e-01 -9.82897937e-01 8.63259554e-01 9.82918918e-01 2.11349443e-01 8.18748236e-01 8.69295537e-01 -1.12787378e+00 -1.34745228e+00 -1.11000240e+00 5.81349552e-01 -6.23999894e-01 3.90519321e-01 -4.24088061e-01 -7.10813701e-01 1.16521454e+00 -6.23887703e-02 -1.37630284e-01 3.79764944e-01 -1.94975361e-01 -2.47706354e-01 5.01330435e-01 -5.73891938e-01 1.39103913e+00 1.70005190e+00 -3.86714041e-01 -9.00714278e-01 1.76765859e-01 7.98473179e-01 -9.72824752e-01 -7.28158236e-01 4.56370234e-01 1.08671939e+00 -6.79839015e-01 1.19859242e+00 -7.41452932e-01 4.64988649e-01 -3.34999621e-01 -7.71752372e-02 -1.23250222e+00 -5.77474475e-01 -1.18250060e+00 -4.22368437e-01 7.07255840e-01 2.64757991e-01 1.43017918e-01 1.00021577e+00 5.62550306e-01 -1.09045126e-01 -5.56009352e-01 -8.40484500e-01 -8.19998026e-01 -3.40764225e-02 -3.03347081e-01 1.69104263e-01 9.10418570e-01 1.47093728e-01 2.62713939e-01 -1.07742500e+00 3.72549035e-02 6.12192392e-01 -2.19022751e-01 1.01367617e+00 -6.20178103e-01 -7.37616301e-01 -5.30968308e-01 -3.75765920e-01 -1.66611600e+00 7.29924217e-02 -5.96803069e-01 1.77393153e-01 -1.38605380e+00 1.98382467e-01 1.01208456e-01 4.22598451e-01 5.51081836e-01 -4.57026064e-02 2.41884083e-01 3.85168731e-01 5.21895349e-01 -3.88234824e-01 5.26197314e-01 1.57270658e+00 -9.12894905e-02 -7.29740202e-01 1.37342475e-02 -1.91238701e-01 1.20125544e+00 3.35194379e-01 -2.73806989e-01 -4.11884904e-01 -4.72844779e-01 1.15320750e-01 3.73801857e-01 4.38156456e-01 -1.02576709e+00 4.20499563e-01 -9.85518470e-02 4.72127318e-01 -5.08739233e-01 5.73575854e-01 -4.22115982e-01 4.92961764e-01 7.15690732e-01 -3.97073388e-01 4.58384097e-01 6.79140612e-02 6.05184734e-01 2.07069233e-01 1.94827065e-01 6.43780470e-01 -1.94894895e-01 -5.88796198e-01 2.48258352e-01 -6.98737681e-01 1.69863313e-01 9.05326307e-01 -5.43254912e-01 2.93594927e-01 -7.56438255e-01 -1.37033165e+00 7.18746930e-02 5.62301695e-01 6.97942257e-01 8.03199351e-01 -1.51667547e+00 -5.70235848e-01 -6.01140857e-02 -1.52665988e-01 1.19195608e-02 1.75774977e-01 7.88200796e-01 -7.78071463e-01 -1.45672560e-02 -8.31847265e-02 -7.98479855e-01 -1.58468711e+00 2.05799520e-01 5.32457352e-01 1.48096591e-01 -1.34552157e+00 8.63958716e-01 3.89129728e-01 -2.51610547e-01 5.09737134e-01 -7.57283807e-01 -1.51502907e-01 -3.28576684e-01 3.88283104e-01 6.12437844e-01 -4.11806226e-01 -9.68898833e-01 -2.96231449e-01 1.05162239e+00 6.51248731e-03 -5.22811890e-01 9.06179786e-01 -3.01662803e-01 3.01662564e-01 7.16829717e-01 9.94564056e-01 2.70919859e-01 -1.50825143e+00 -1.18891507e-01 -2.38011345e-01 -3.91687691e-01 -6.66338801e-01 -3.74286711e-01 -8.85751963e-01 6.64277017e-01 2.31798828e-01 -5.88139057e-01 7.20623612e-01 1.78962890e-02 1.20954490e+00 3.69637907e-01 3.79574358e-01 -1.23778427e+00 6.20502293e-01 5.96332192e-01 1.14774871e+00 -5.64784944e-01 -1.51012555e-01 -5.66176176e-01 -9.38162267e-01 9.52148438e-01 8.10711503e-01 -2.59614438e-01 5.04549265e-01 4.53647733e-01 -1.02687813e-01 -7.55569711e-02 -4.89652812e-01 2.58306772e-01 5.27769208e-01 3.73253137e-01 6.26544893e-01 9.13670361e-02 -1.16352424e-01 6.88237369e-01 -5.39213181e-01 -2.37067178e-01 4.86893415e-01 1.10917032e+00 -1.71604484e-01 -9.31746781e-01 -4.39477563e-01 3.35728563e-02 -6.09382749e-01 -8.40358660e-02 -6.26591742e-01 8.84215772e-01 -9.68437195e-02 6.05461359e-01 -3.47474366e-02 -7.77612388e-01 6.13639235e-01 -2.29267448e-01 7.03912675e-01 -7.15423465e-01 -6.27655447e-01 5.03581226e-01 1.57746673e-01 -1.14830577e+00 -4.50523585e-01 -5.39366961e-01 -1.62492692e+00 -2.60371715e-01 -1.09084621e-01 -1.87565625e-01 -4.52487245e-02 8.34472537e-01 3.45599025e-01 6.06238723e-01 4.30666097e-02 -1.28122795e+00 -3.26135993e-01 -9.35320020e-01 -4.54906195e-01 7.40687907e-01 4.30655718e-01 -7.69781709e-01 -1.25501165e-02 6.71701372e-01]
[7.361909866333008, -0.41498738527297974]
a996da2b-f283-481d-9a3a-6032f68b4f58
a-kinematic-chain-space-for-monocular-motion
1702.00186
null
http://arxiv.org/abs/1702.00186v1
http://arxiv.org/pdf/1702.00186v1.pdf
A Kinematic Chain Space for Monocular Motion Capture
This paper deals with motion capture of kinematic chains (e.g. human skeletons) from monocular image sequences taken by uncalibrated cameras. We present a method based on projecting an observation into a kinematic chain space (KCS). An optimization of the nuclear norm is proposed that implicitly enforces structural properties of the kinematic chain. Unlike other approaches our method does not require specific camera or object motion and is not relying on training data or previously determined constraints such as particular body lengths. The proposed algorithm is able to reconstruct scenes with limited camera motion and previously unseen motions. It is not only applicable to human skeletons but also to other kinematic chains for instance animals or industrial robots. We achieve state-of-the-art results on different benchmark data bases and real world scenes.
['Bastian Wandt', 'Hanno Ackermann', 'Bodo Rosenhahn']
2017-02-01
null
null
null
null
['industrial-robots']
['robots']
[ 3.60887259e-01 -9.80957299e-02 -1.64868996e-01 5.03793266e-03 9.60319936e-02 -6.26294911e-01 5.67864418e-01 -4.63859469e-01 -6.68138206e-01 6.86386466e-01 -1.90125033e-01 1.13893244e-02 -1.21065918e-02 -3.49456817e-01 -8.11852872e-01 -8.39297473e-01 1.19821325e-01 7.23810494e-01 5.39732575e-01 7.91200176e-02 5.62765822e-02 8.55993807e-01 -1.35497320e+00 -2.07252249e-01 1.38061970e-01 3.45880777e-01 5.46221912e-01 1.06105852e+00 5.21076739e-01 6.33431435e-01 -1.99921608e-01 -3.26191545e-01 3.56733859e-01 -4.63347822e-01 -7.94688940e-01 7.25172043e-01 6.10458255e-01 -1.88396096e-01 -4.72900480e-01 1.07557404e+00 1.43083543e-01 2.07573816e-01 6.87791884e-01 -9.32621181e-01 -6.86410442e-02 1.75158426e-01 -4.60611820e-01 6.69080997e-04 6.27216220e-01 1.61167756e-01 7.62132704e-01 -7.71243334e-01 1.34345782e+00 1.20872688e+00 6.32837236e-01 5.66685081e-01 -1.40065622e+00 -1.87353760e-01 3.34106460e-02 3.28995407e-01 -1.12668645e+00 -3.30037385e-01 6.81427479e-01 -6.18693829e-01 7.37740993e-01 6.97038695e-02 8.26828301e-01 1.26426709e+00 3.26218694e-01 7.24600196e-01 8.38184416e-01 -5.36234438e-01 2.79165804e-01 -4.97649670e-01 -5.91916405e-02 7.14893758e-01 5.30909956e-01 6.13982156e-02 -3.51438999e-01 8.85003712e-03 1.09477806e+00 -2.44689167e-01 -3.70097876e-01 -1.19091988e+00 -1.91910529e+00 5.97161293e-01 -1.21273242e-01 7.13673010e-02 -2.39167348e-01 4.64454502e-01 2.64754057e-01 -1.39270099e-02 -2.23558918e-01 1.90116867e-01 -3.56097728e-01 -1.05660029e-01 -8.34797382e-01 3.53244007e-01 7.85787642e-01 1.24365067e+00 6.04493499e-01 9.57308337e-02 6.16926670e-01 5.48056006e-01 2.57371455e-01 7.51853347e-01 3.09100419e-01 -1.42586410e+00 5.20320117e-01 2.02295765e-01 2.60408491e-01 -6.58431530e-01 -6.77069783e-01 -1.77191585e-01 -7.38852859e-01 3.08075458e-01 8.15993607e-01 -8.58765692e-02 -8.67832065e-01 1.54532444e+00 4.97479707e-01 1.42108068e-01 3.39538418e-02 1.15994751e+00 2.54313439e-01 3.22436035e-01 -3.57486457e-01 -4.21891689e-01 1.30524635e+00 -1.02767885e+00 -5.70512056e-01 -1.37940347e-01 3.45688283e-01 -8.47293735e-01 6.26407027e-01 9.58169699e-01 -1.15083385e+00 -6.14455700e-01 -9.31141734e-01 -6.37034178e-02 1.34336725e-01 3.65802526e-01 3.17197412e-01 5.92281282e-01 -6.39785707e-01 7.34933674e-01 -1.32614851e+00 -7.40878582e-01 -3.23475242e-01 5.73128581e-01 -8.28366578e-01 -3.41069661e-02 -5.00584245e-01 7.62110531e-01 4.38031346e-01 3.51586044e-01 -9.93087113e-01 -1.60698090e-02 -9.28897262e-01 -5.15492737e-01 6.36464775e-01 -1.10263932e+00 1.22568750e+00 -7.31865346e-01 -1.81811392e+00 8.95793736e-01 -2.61531118e-02 -4.32313770e-01 9.01030838e-01 -5.79374909e-01 -1.15227893e-01 4.89767790e-01 -2.42949113e-01 5.36762238e-01 1.02580953e+00 -1.24891198e+00 -3.62214178e-01 -3.25072438e-01 -3.27125154e-02 2.44096488e-01 2.14456186e-01 -5.07251024e-02 -7.51089275e-01 -6.79493666e-01 4.98602062e-01 -1.72597373e+00 -4.37274814e-01 1.12737432e-01 -6.35523081e-01 3.06933105e-01 8.86928797e-01 -4.50347245e-01 6.68767691e-01 -1.63050795e+00 1.08661509e+00 1.03793278e-01 -2.40090817e-01 1.33427873e-01 3.09627037e-03 3.84070605e-01 7.68575370e-02 -5.79043269e-01 -4.06544030e-01 -4.46479380e-01 -2.16872454e-01 6.27411664e-01 1.56254411e-01 1.08728373e+00 -2.43008494e-01 3.62720490e-01 -9.97224987e-01 -6.15065098e-01 5.35238802e-01 3.21838558e-01 -5.12082279e-01 8.59514922e-02 -2.00835943e-01 9.60667312e-01 -2.43951783e-01 5.50850749e-01 5.05069017e-01 6.44722627e-03 3.27336699e-01 -2.22485662e-01 -1.67053953e-01 -2.80447155e-01 -1.61842740e+00 2.06577754e+00 -2.83587962e-01 3.99696767e-01 2.84814626e-01 -7.79536784e-01 6.94239378e-01 2.69109249e-01 5.82046628e-01 1.29552990e-01 1.15125671e-01 1.71540990e-01 4.47780639e-02 -6.92065477e-01 3.78446579e-01 -3.62591833e-01 8.13502297e-02 1.67344213e-01 2.32654408e-01 -1.67097062e-01 4.85691905e-01 -1.45324171e-01 1.08702171e+00 8.08735549e-01 5.57339966e-01 -2.92364180e-01 8.30259144e-01 1.73166022e-01 6.63219810e-01 1.96272716e-01 8.70614946e-02 8.62926245e-01 1.43423513e-01 -4.68630761e-01 -1.35255992e+00 -1.24527621e+00 -7.06072971e-02 5.72251856e-01 3.98299873e-01 -2.09508106e-01 -6.34403348e-01 -4.24602240e-01 -1.20448604e-01 2.30544969e-01 -4.76186514e-01 1.23761691e-01 -9.79713559e-01 -5.86714864e-01 4.01589423e-01 5.35647213e-01 2.82006741e-01 -9.35611904e-01 -1.17385328e+00 2.88114309e-01 -2.24429205e-01 -1.71349096e+00 -3.65522265e-01 3.64361331e-02 -1.20160747e+00 -1.35739529e+00 -8.60222697e-01 -6.50695205e-01 7.16190577e-01 2.37176642e-01 9.80165839e-01 -3.46138895e-01 -2.89004177e-01 7.21406639e-01 -2.47170776e-01 4.58697975e-02 -4.81522828e-01 4.11789343e-02 3.83390039e-01 1.37873441e-01 -5.28695881e-02 -6.41095102e-01 -6.22081339e-01 7.33525515e-01 -7.90562809e-01 3.65988202e-02 5.82088590e-01 7.11228013e-01 7.19989479e-01 -2.15908781e-01 -2.05704823e-01 -7.36341059e-01 -1.83582112e-01 -1.08367272e-01 -8.96314740e-01 -6.66111633e-02 -1.16841502e-01 1.34776056e-01 6.93759382e-01 -6.08270168e-01 -1.16395020e+00 1.10508239e+00 2.60419399e-01 -7.60352373e-01 -5.43044329e-01 1.02203064e-01 -1.91483483e-01 -3.22127789e-02 4.80070889e-01 9.04783607e-02 -1.03559770e-01 -7.43105471e-01 4.14184570e-01 3.09530161e-02 1.14041519e+00 -6.08789563e-01 9.35202062e-01 9.92728293e-01 5.73896825e-01 -1.32019663e+00 -8.38182122e-02 -8.44364703e-01 -1.42910171e+00 -4.22864169e-01 1.13526857e+00 -7.77832389e-01 -6.69034183e-01 4.97675925e-01 -1.34216237e+00 -1.84422612e-01 -1.34993494e-01 1.16653049e+00 -1.20523274e+00 1.02807903e+00 -8.32051933e-01 -6.14930987e-01 9.39756483e-02 -1.05438113e+00 1.17691898e+00 -2.06806019e-01 -3.12595516e-01 -1.02450967e+00 4.83066529e-01 4.42865700e-01 -5.19241333e-01 6.08834088e-01 3.78638715e-01 4.75885794e-02 -7.99833953e-01 -1.09042153e-01 3.80305737e-01 3.46524864e-01 3.19899083e-03 2.89330482e-01 -6.12120092e-01 -3.56975883e-01 2.53854748e-02 -7.51652196e-02 7.44395792e-01 3.14559996e-01 4.04133379e-01 -1.21684887e-01 -3.93354148e-01 7.04976737e-01 1.39883626e+00 1.33658692e-01 6.45699203e-01 4.49971825e-01 1.02968764e+00 8.56531203e-01 7.35516071e-01 3.48064452e-01 -1.61353126e-02 1.08599567e+00 6.28458321e-01 3.55173677e-01 -1.48482397e-01 -7.44348392e-02 6.52073145e-01 8.28356981e-01 -6.65339589e-01 -1.42263547e-01 -8.50722313e-01 6.40347838e-01 -2.01195240e+00 -9.23998177e-01 -7.55019009e-01 2.19388700e+00 3.50313216e-01 -8.69955868e-03 3.05497587e-01 2.66777903e-01 7.79348552e-01 -1.09724954e-01 -6.15693569e-01 -1.69437528e-01 -1.26295999e-01 -9.53079090e-02 7.75379121e-01 5.21737278e-01 -1.12823784e+00 7.23681331e-01 6.42077732e+00 2.05078840e-01 -7.94230103e-01 -3.54451239e-02 -3.93500865e-01 -6.55756220e-02 3.35069180e-01 1.26030803e-01 -7.46019065e-01 1.68900043e-01 5.22981226e-01 2.51131624e-01 2.74046630e-01 8.90029252e-01 1.22140177e-01 -3.51058304e-01 -1.26298702e+00 9.32331324e-01 -1.86660129e-03 -8.92225564e-01 -1.06276803e-01 2.99097180e-01 7.36443579e-01 -6.63304180e-02 -2.38548279e-01 -5.85177422e-01 2.53390849e-01 -6.95239604e-01 9.84190464e-01 5.33454597e-01 4.42400455e-01 -4.76642102e-01 5.83816648e-01 5.44145465e-01 -1.16673195e+00 8.64752084e-02 -5.71538627e-01 -5.79998642e-02 5.32843173e-01 4.36110646e-01 -6.50524557e-01 6.50537252e-01 4.10287887e-01 8.64202738e-01 -3.30572128e-01 1.13887870e+00 -3.06307673e-01 4.41925228e-01 -5.70450425e-01 3.67292762e-01 1.84921771e-01 -5.79004347e-01 8.86645019e-01 1.36376429e+00 3.65590394e-01 9.09502581e-02 7.11248890e-02 2.48436138e-01 3.24165791e-01 -2.78522400e-03 -8.07724655e-01 4.13163930e-01 -2.51528919e-01 1.18414986e+00 -1.03718293e+00 -2.31464237e-01 -5.32066882e-01 1.25465822e+00 -2.42626891e-02 3.35958362e-01 -6.50001407e-01 1.68460563e-01 5.20572543e-01 1.84571192e-01 6.71009064e-01 -8.67639959e-01 1.71090201e-01 -1.50647092e+00 -1.31793441e-02 -5.78594148e-01 2.79584467e-01 -6.96721077e-01 -7.56985962e-01 2.54542053e-01 4.11818326e-01 -1.47361803e+00 -4.92840677e-01 -9.69283938e-01 -1.49537995e-01 2.79561341e-01 -1.01939201e+00 -1.10768628e+00 -3.23747426e-01 8.41060281e-01 6.67100251e-01 3.32773291e-02 4.67396528e-01 7.70131722e-02 -2.94772536e-01 -2.43886724e-01 1.74887940e-01 -4.28836197e-02 6.67314887e-01 -1.30629718e+00 2.24000961e-01 9.27639425e-01 3.66065323e-01 4.86403674e-01 1.17676425e+00 -6.43509328e-01 -1.59258318e+00 -8.52261186e-01 4.27302480e-01 -7.02661335e-01 6.22462869e-01 -2.57302314e-01 -5.28548896e-01 9.32419479e-01 4.51405905e-03 1.92187980e-01 3.08541775e-01 -3.50982130e-01 -1.31018147e-01 1.93652317e-01 -7.53885031e-01 4.09715742e-01 1.48836899e+00 -9.35935080e-02 -6.78932786e-01 3.61384183e-01 1.97942838e-01 -6.53715670e-01 -8.17973077e-01 4.35664862e-01 8.27577472e-01 -9.56097603e-01 1.27542508e+00 -6.82697892e-01 2.34292001e-01 -6.94131613e-01 -2.03913152e-01 -8.95280719e-01 -2.36587241e-01 -6.87389851e-01 -2.57993132e-01 5.37258148e-01 -3.95439491e-02 -9.03547188e-05 1.06611490e+00 1.22125790e-01 -1.22796856e-01 -2.53434420e-01 -1.03495264e+00 -1.34747839e+00 -1.28491089e-01 -3.95969957e-01 -1.20591611e-01 7.14301765e-01 -2.54543126e-01 3.72523159e-01 -7.67626226e-01 3.99842530e-01 1.06788766e+00 2.02792391e-01 1.12888014e+00 -1.18379259e+00 -5.11867285e-01 2.95227710e-02 -8.99163544e-01 -1.04993260e+00 3.13317806e-01 -4.98256236e-01 7.02624321e-02 -1.26490831e+00 8.19472149e-02 5.84190339e-02 3.11863929e-01 1.39656486e-02 3.29182267e-01 5.13720095e-01 2.91930914e-01 3.86635035e-01 -6.68827593e-01 4.29122627e-01 1.38006473e+00 2.04939842e-01 3.31268087e-02 2.47577548e-01 4.61225569e-01 1.33605766e+00 5.56271851e-01 -4.83269632e-01 -3.97198707e-01 -3.47524792e-01 1.94980815e-01 3.46902728e-01 6.68093503e-01 -1.32254720e+00 4.58986461e-01 -3.70797962e-01 1.91502184e-01 -7.11901724e-01 5.74147105e-01 -1.08669198e+00 8.97628486e-01 8.06253374e-01 9.68612060e-02 2.14428261e-01 -1.65233910e-01 1.03376317e+00 2.03701332e-02 -5.96356571e-01 8.98380160e-01 -5.04105628e-01 -7.49594092e-01 -6.56673759e-02 -5.30997872e-01 -2.62883812e-01 1.14224434e+00 -5.25893807e-01 5.51613048e-02 -2.77494192e-01 -1.29098833e+00 -1.13434099e-01 1.05010927e+00 2.63699293e-01 5.09316564e-01 -1.22389472e+00 -4.54456210e-01 -5.59471436e-02 1.27900183e-01 1.22405246e-01 -5.50193563e-02 8.29677820e-01 -1.19374239e+00 6.44717574e-01 -4.82078314e-01 -9.35217917e-01 -1.56613648e+00 9.12634313e-01 2.20951632e-01 -1.57979071e-01 -8.37485194e-01 2.94525772e-01 1.42394289e-01 -4.22052234e-01 9.31122247e-03 -6.49670959e-01 3.87326106e-02 -1.78316742e-01 2.07758788e-02 7.61248469e-01 -1.44344300e-01 -1.19829154e+00 -4.75566506e-01 1.24159634e+00 4.27050948e-01 -3.90801758e-01 1.10837460e+00 -3.27062488e-01 8.24364573e-02 5.03908873e-01 1.10520542e+00 1.11154132e-01 -1.50210834e+00 -3.02795861e-02 1.54907435e-01 -4.51936513e-01 -4.40255404e-01 -1.02280170e-01 -9.92786407e-01 8.36374402e-01 2.31758520e-01 -3.49484235e-01 8.77852559e-01 2.51086298e-02 6.39849186e-01 7.12728202e-01 8.23565722e-01 -1.04635906e+00 2.09569901e-01 3.06811422e-01 7.20028996e-01 -8.22816670e-01 4.14917946e-01 -7.45738328e-01 -4.50920194e-01 1.50867140e+00 2.87124932e-01 -3.57952476e-01 4.32141662e-01 2.44926453e-01 -4.31569181e-02 -1.75697226e-02 -6.03250086e-01 -2.98760563e-01 9.30543244e-02 6.86990321e-01 2.79738829e-02 -6.15430763e-03 -4.54333663e-01 -1.88669547e-01 -1.53878108e-01 5.69773354e-02 8.54997814e-01 1.05088413e+00 -1.82383776e-01 -1.22654331e+00 -8.09704721e-01 -2.01119542e-01 -3.89408946e-01 5.96673250e-01 -3.67659867e-01 1.09450531e+00 3.39325607e-01 6.56481802e-01 -1.32945001e-01 -2.07245201e-01 5.46503067e-01 -8.63930508e-02 1.00845110e+00 -6.44381046e-01 -1.78706974e-01 4.26208586e-01 2.08980069e-01 -7.69430161e-01 -1.11464989e+00 -1.37320220e+00 -1.10776722e+00 -8.21023583e-02 -3.40710551e-01 -2.03088298e-01 6.97455943e-01 7.01682031e-01 -2.41189614e-01 9.14059728e-02 7.28109628e-02 -1.18059170e+00 -2.86934614e-01 -5.96178055e-01 -7.40824521e-01 5.06278574e-01 4.69766945e-01 -8.49445581e-01 -1.29148334e-01 8.26128542e-01]
[7.442643642425537, -1.4216597080230713]
6cb13d94-52c8-4d9e-a507-1526c8201a07
potential-based-credit-assignment-for
2305.18380
null
https://arxiv.org/abs/2305.18380v1
https://arxiv.org/pdf/2305.18380v1.pdf
Potential-based Credit Assignment for Cooperative RL-based Testing of Autonomous Vehicles
While autonomous vehicles (AVs) may perform remarkably well in generic real-life cases, their irrational action in some unforeseen cases leads to critical safety concerns. This paper introduces the concept of collaborative reinforcement learning (RL) to generate challenging test cases for AV planning and decision-making module. One of the critical challenges for collaborative RL is the credit assignment problem, where a proper assignment of rewards to multiple agents interacting in the traffic scenario, considering all parameters and timing, turns out to be non-trivial. In order to address this challenge, we propose a novel potential-based reward-shaping approach inspired by counterfactual analysis for solving the credit-assignment problem. The evaluation in a simulated environment demonstrates the superiority of our proposed approach against other methods using local and global rewards.
['Hao Shen', 'Chih-Hong Cheng', 'Utku Ayvaz']
2023-05-28
null
null
null
null
['autonomous-vehicles']
['computer-vision']
[ 5.82397506e-02 4.46842551e-01 1.09645603e-02 -2.75547475e-01 -4.98772532e-01 -5.08015156e-01 8.72562051e-01 3.25889409e-01 -5.94559550e-01 1.38281131e+00 -4.73858677e-02 -4.86514211e-01 -4.96153921e-01 -8.42794955e-01 -5.18221855e-01 -7.17056751e-01 -3.67079765e-01 6.61328971e-01 3.58041674e-01 -4.75149781e-01 4.94714618e-01 5.00327766e-01 -1.72742212e+00 -2.91143119e-01 1.06302905e+00 8.15784872e-01 3.14470738e-01 5.57506621e-01 1.29338726e-01 8.72284114e-01 -7.12910235e-01 -1.35343269e-01 4.28975910e-01 -2.74543583e-01 -7.64125228e-01 -1.66410059e-01 -3.11256528e-01 -5.17920434e-01 2.58238107e-01 9.10918891e-01 4.03443813e-01 5.73263466e-01 7.58983016e-01 -1.93973148e+00 -1.06338710e-02 4.22545910e-01 -5.17267585e-01 1.08251318e-01 3.67086917e-01 6.08609736e-01 8.91822755e-01 -4.18531429e-03 4.17783380e-01 1.10351396e+00 1.12612300e-01 5.27188838e-01 -9.86569941e-01 -5.41757643e-01 3.00235331e-01 4.96867388e-01 -9.73227084e-01 -1.30911052e-01 6.18211329e-01 -1.97622582e-01 7.97221661e-01 5.36156669e-02 6.40431046e-01 9.91177917e-01 4.36559618e-01 6.38151586e-01 1.41625082e+00 -1.60490349e-01 6.93171203e-01 1.79067165e-01 -4.55250770e-01 1.65663034e-01 3.99571031e-01 6.92524433e-01 6.81220591e-02 -4.05601799e-01 3.14605892e-01 -2.21921116e-01 3.35074097e-01 -5.54209113e-01 -9.30930674e-01 9.48159575e-01 6.64175898e-02 3.79263125e-02 -8.90351295e-01 2.74470121e-01 4.85390723e-01 4.65212494e-01 -9.75882262e-02 4.27479953e-01 -3.13192993e-01 -2.64095098e-01 -5.51874876e-01 8.36240590e-01 6.82305276e-01 6.92097604e-01 4.16010052e-01 3.80916178e-01 -3.59434336e-01 2.90502965e-01 5.45105755e-01 3.51732552e-01 1.36698589e-01 -9.58537281e-01 4.82336074e-01 2.38794118e-01 8.63880634e-01 -8.68675351e-01 -4.76968825e-01 -3.65847796e-01 -3.10736120e-01 1.08636510e+00 5.53463697e-01 -5.55063248e-01 -3.91920358e-01 1.81659007e+00 5.32365441e-01 2.62838870e-01 5.25788963e-01 8.43508184e-01 1.65311292e-01 7.17566729e-01 4.42664772e-01 -6.66231692e-01 8.42115104e-01 -6.15208685e-01 -6.99715853e-01 1.34321824e-01 3.76726270e-01 -5.42683363e-01 6.96999133e-01 2.10925460e-01 -1.00185812e+00 -2.03428254e-01 -1.07998276e+00 9.30164814e-01 -4.78540882e-02 -5.04590690e-01 4.92597997e-01 6.55817986e-01 -8.25113356e-01 5.54228306e-01 -3.25617433e-01 -3.00879747e-01 2.54038632e-01 4.24662948e-01 1.43846953e-02 2.23772630e-01 -1.26061904e+00 1.04474509e+00 4.15879309e-01 8.66231415e-03 -1.49045181e+00 -2.40126118e-01 -3.49271476e-01 8.41721818e-02 9.24103379e-01 -3.92282158e-01 1.41666949e+00 -9.58422005e-01 -1.78836691e+00 1.16262063e-01 4.22148228e-01 -8.51536274e-01 9.40562129e-01 1.28486603e-01 -1.67799532e-01 -1.23691052e-01 1.17511965e-01 5.38548887e-01 7.92371154e-01 -1.42363858e+00 -7.65659153e-01 1.72781162e-02 5.80233514e-01 5.25001168e-01 2.19809189e-01 -2.73972582e-02 7.31462300e-01 -3.84953678e-01 -5.80787122e-01 -9.23185706e-01 -6.99184477e-01 -6.41223669e-01 -1.30111784e-01 -4.61194217e-01 6.23650789e-01 8.99897665e-02 7.68242240e-01 -1.83241630e+00 -2.62575775e-01 2.36764774e-01 -1.54852882e-01 1.89750552e-01 -8.86535794e-02 7.64166117e-01 1.31659761e-01 -8.12587887e-02 -1.06167920e-01 1.06177494e-01 2.57208914e-01 2.87816375e-01 -2.06805795e-01 3.17124695e-01 2.19552100e-01 5.27973413e-01 -1.10222912e+00 -5.56745589e-01 2.53768116e-01 -1.82811424e-01 -4.28929299e-01 5.18001616e-01 -5.06303370e-01 6.65503860e-01 -7.87464857e-01 4.58603084e-01 5.73144674e-01 5.19228876e-01 2.86901817e-02 4.06259298e-01 -3.43240947e-01 -2.63345391e-01 -1.48446500e+00 8.45014632e-01 -2.67260641e-01 2.17038959e-01 -1.25011742e-01 -1.08413804e+00 8.44644010e-01 4.20991898e-01 6.89631402e-01 -8.00396144e-01 4.68715966e-01 2.28289917e-01 1.66733772e-01 -6.01522744e-01 6.54648662e-01 -4.29673433e-01 -2.25856513e-01 6.84534550e-01 -2.04885036e-01 -6.43357337e-02 2.59085130e-02 7.21601471e-02 1.16770566e+00 3.08582544e-01 5.24211287e-01 -1.96417749e-01 6.86384976e-01 3.28294516e-01 7.42158830e-01 7.83267558e-01 -7.92445719e-01 -1.28792211e-01 8.56178522e-01 -2.98351854e-01 -9.49212015e-01 -8.73711228e-01 4.02211785e-01 6.55189633e-01 3.55551660e-01 2.72429407e-01 -5.24941206e-01 -8.87776315e-01 1.84570886e-02 1.17109096e+00 -4.93133068e-01 -2.35797048e-01 -3.19970846e-01 -6.37685835e-01 2.86202461e-01 1.00618832e-01 4.31309044e-01 -1.29505110e+00 -1.30681956e+00 5.22387385e-01 2.08950549e-01 -9.58081424e-01 1.16602276e-02 7.10666329e-02 -4.90883201e-01 -1.01615489e+00 -5.54717362e-01 -2.50525355e-01 4.09149468e-01 2.51019239e-01 7.57514238e-01 6.99641928e-02 1.66653693e-02 2.82173812e-01 -6.37510955e-01 -7.25671649e-01 -7.40253329e-01 -3.20577860e-01 1.44684136e-01 1.32466257e-01 1.08994439e-01 -5.07110775e-01 -7.22433507e-01 5.01821280e-01 -8.14252555e-01 -2.38561377e-01 5.82903564e-01 6.84444904e-01 2.73327976e-01 3.53438765e-01 1.28979266e+00 -6.25350177e-01 1.12333465e+00 -7.40162015e-01 -1.02341163e+00 2.24353328e-01 -7.13359296e-01 1.57353625e-01 8.19200039e-01 -4.76132363e-01 -1.15404665e+00 1.00733250e-01 1.03813998e-01 4.48718034e-02 -3.22344810e-01 1.53078467e-01 -1.25251248e-01 -1.07151680e-02 5.16610682e-01 -1.20503195e-01 1.19626030e-01 1.85450733e-01 2.06498727e-01 4.75709915e-01 1.90229155e-02 -7.34470844e-01 8.64896297e-01 8.62009227e-02 5.12996495e-01 -3.64428878e-01 -9.40065533e-02 -2.37055011e-02 -3.20234001e-02 -7.36122489e-01 6.99337065e-01 -4.69718397e-01 -1.39201474e+00 2.25165367e-01 -1.04476142e+00 -3.39857101e-01 -3.87327611e-01 4.46476340e-01 -1.10808146e+00 2.98728228e-01 1.83774978e-01 -1.46488833e+00 9.24766809e-02 -1.45378363e+00 2.77077645e-01 5.14434397e-01 4.49977443e-02 -5.88171840e-01 2.27613673e-01 2.97240973e-01 6.53584003e-01 7.34592080e-01 7.07868814e-01 -8.23660433e-01 -6.73728526e-01 -9.25640687e-02 -6.29229052e-03 8.52382928e-02 -2.40513355e-01 2.01319847e-02 -6.64451003e-01 -1.02334075e-01 -1.69518590e-01 -5.51370800e-01 5.87430932e-02 1.89883783e-01 3.92726123e-01 -4.19285893e-01 -7.08497083e-03 -4.45975244e-01 1.61583269e+00 8.18150342e-01 6.13440812e-01 6.04083478e-01 -2.43287925e-02 8.97781849e-01 1.45004582e+00 9.60597992e-01 5.32348216e-01 7.32417345e-01 1.09935832e+00 2.83410788e-01 5.13316333e-01 -1.18676044e-01 3.77250105e-01 -2.03633249e-01 -1.96810260e-01 -2.46733367e-01 -8.44435573e-01 5.24125576e-01 -2.18080664e+00 -1.21556711e+00 -1.64557964e-01 2.47239947e+00 4.23572272e-01 4.47846085e-01 6.24095619e-01 5.90497255e-02 6.11205757e-01 -8.37286860e-02 -4.29231554e-01 -9.37914073e-01 6.19536228e-02 -1.92996144e-01 7.24387467e-01 6.28429770e-01 -6.90326810e-01 7.87573099e-01 6.19580221e+00 8.44500780e-01 -7.50294626e-01 1.33957729e-01 5.82416415e-01 1.98253039e-02 -3.09033990e-01 4.08244491e-01 -5.17596185e-01 4.00868088e-01 1.12954116e+00 -5.60128093e-01 4.12376404e-01 7.55118072e-01 8.81696880e-01 -7.22142875e-01 -8.41131628e-01 4.13916439e-01 -5.60734510e-01 -1.05255687e+00 -2.94158995e-01 1.54229831e-02 6.98043942e-01 -2.08226085e-01 -1.44469425e-01 4.04100627e-01 7.54671931e-01 -1.02633035e+00 8.42195690e-01 5.79421878e-01 1.32481337e-01 -1.32389319e+00 1.09093535e+00 4.90606666e-01 -7.70577610e-01 -4.84508395e-01 -2.25172073e-01 -2.83428252e-01 2.43225813e-01 4.27027931e-04 -1.19309747e+00 5.76566815e-01 2.65382767e-01 -1.38189495e-01 -1.73194841e-01 1.20150173e+00 -6.65333942e-02 3.47428024e-01 -1.91639587e-01 -7.03754663e-01 6.20806813e-01 -2.16215819e-01 7.90762544e-01 7.03984022e-01 1.67752728e-01 7.96666369e-02 2.61066761e-02 8.35643828e-01 5.66426158e-01 2.15892524e-01 -9.69402790e-01 1.51148930e-01 5.04113913e-01 1.17882001e+00 -8.03959012e-01 -1.35584399e-02 -7.27348700e-02 4.82782334e-01 6.36778697e-02 3.52038115e-01 -1.12680125e+00 -1.61938176e-01 4.21635747e-01 -3.00517045e-02 -7.09846849e-03 -2.40542814e-01 -1.88751921e-01 -3.43777150e-01 -1.05701208e-01 -7.53593802e-01 1.75650567e-01 -5.73505163e-01 -9.28502858e-01 5.89696229e-01 4.49670941e-01 -1.51665592e+00 -5.81206620e-01 -1.40825287e-01 -8.72642279e-01 5.29080272e-01 -1.67521489e+00 -8.68025005e-01 -2.03999672e-02 5.13121128e-01 4.72937733e-01 -5.18653035e-01 3.79439712e-01 1.05105182e-02 -3.78309667e-01 3.17918539e-01 -3.47334109e-02 -7.62991369e-01 3.87117356e-01 -1.06998158e+00 -1.56395778e-01 6.06520832e-01 -5.16818285e-01 -1.31158575e-01 1.41281545e+00 -5.67378104e-01 -1.34325361e+00 -7.01145470e-01 5.27291059e-01 1.07851746e-02 7.71535099e-01 -3.16528454e-02 -4.27609801e-01 1.24519303e-01 4.63094264e-01 -1.62042677e-01 2.58242756e-01 -3.02370340e-01 3.26253265e-01 -2.30311453e-01 -1.55151331e+00 7.97385335e-01 6.60988569e-01 2.29428515e-01 -5.30022860e-01 2.48508483e-01 5.43782651e-01 2.45956518e-02 -5.23849130e-01 2.98520356e-01 3.87251467e-01 -1.11478472e+00 5.04819691e-01 -6.96496427e-01 1.32421777e-01 -4.15481478e-01 -2.00795293e-01 -1.51853824e+00 5.53132705e-02 -9.13348138e-01 3.61180961e-01 1.20464385e+00 4.34046388e-01 -7.30617523e-01 7.41459608e-01 7.63092041e-01 -1.81304440e-02 -6.19213939e-01 -1.33136892e+00 -9.04870093e-01 5.03644049e-02 -4.33853954e-01 7.36166537e-01 5.46362460e-01 1.41935411e-03 -5.54006919e-02 -5.20926595e-01 1.86363891e-01 8.20970237e-01 -1.13947608e-01 9.41545188e-01 -9.58007336e-01 -3.61398876e-01 -3.25342923e-01 -4.59627837e-01 -2.57134527e-01 1.95381120e-01 -2.48210132e-01 2.61590213e-01 -1.43322051e+00 -1.86743855e-01 -8.16119969e-01 -2.53269434e-01 1.98879614e-01 1.57740936e-01 -3.46412271e-01 4.32258070e-01 -8.04143846e-02 -7.78747737e-01 7.97027409e-01 1.16888762e+00 -8.00942257e-02 -2.02330723e-01 3.65669489e-01 -3.98095310e-01 4.36335295e-01 1.16309118e+00 -5.87160945e-01 -7.93366611e-01 1.87632486e-01 3.50814104e-01 7.47080863e-01 1.88486874e-01 -1.04427636e+00 2.20691174e-01 -1.13686812e+00 -4.92056310e-01 -5.67227721e-01 3.56509611e-02 -1.16323352e+00 2.62248874e-01 6.31925404e-01 -2.09850460e-01 2.81205028e-01 1.42242283e-01 8.84340286e-01 -1.59984753e-01 -5.33187151e-01 7.31423676e-01 -9.06376075e-03 -6.67853296e-01 5.01014665e-02 -8.88678789e-01 -2.95522287e-02 1.91544712e+00 -2.90637434e-01 -1.85546413e-01 -6.44704640e-01 -3.90825689e-01 6.71552360e-01 1.43525183e-01 3.74821842e-01 5.45615971e-01 -1.16547477e+00 -7.69677818e-01 -2.45963857e-01 4.29695211e-02 -2.98819155e-01 3.24100524e-01 8.33937228e-01 -4.03108746e-01 2.00624019e-01 -7.66833544e-01 -1.67677298e-01 -1.01284230e+00 6.26000285e-01 3.81122023e-01 -3.26648086e-01 -2.76942909e-01 1.60035104e-01 -1.52778938e-01 -7.93275163e-02 2.38533884e-01 1.06034450e-01 -6.39745891e-01 6.54217750e-02 2.32493609e-01 6.05333030e-01 -2.01829538e-01 -5.17536163e-01 -2.06772983e-01 2.33135615e-02 1.20691985e-01 -4.50030208e-01 1.22628033e+00 -2.28622124e-01 4.20622766e-01 2.39051029e-01 4.40172106e-01 -2.38955989e-01 -1.40912652e+00 2.02965602e-01 2.62986392e-01 -4.41616029e-01 -1.21741958e-01 -8.50805223e-01 -6.56996846e-01 3.75934094e-01 6.41127586e-01 5.07509232e-01 9.73105013e-01 -4.81223375e-01 5.18773973e-01 4.18020338e-01 7.40013778e-01 -1.53557837e+00 3.29116322e-02 4.10858333e-01 8.18456709e-01 -1.43595958e+00 -2.40191430e-01 1.11960762e-04 -1.13884425e+00 9.86327529e-01 6.52618766e-01 -3.41431051e-01 3.41528714e-01 9.69925374e-02 -4.94245775e-02 6.53681010e-02 -1.05316389e+00 -2.32137337e-01 -4.92544532e-01 8.76085937e-01 -1.26903012e-01 3.15329731e-01 -7.85751581e-01 3.15825641e-01 1.89005628e-01 8.32608640e-02 9.99462008e-01 1.08202219e+00 -6.11615956e-01 -1.28143346e+00 -5.75220108e-01 2.39176869e-01 -2.56536961e-01 5.85287631e-01 -2.07355469e-01 8.41817319e-01 2.16272414e-01 1.36826110e+00 -3.20279628e-01 -3.23966265e-01 3.63841474e-01 -1.70685887e-01 1.50250778e-01 -2.45918244e-01 -7.47378230e-01 -1.34302884e-01 3.69325042e-01 -4.61884022e-01 -6.14352345e-01 -9.96104300e-01 -1.51190662e+00 -3.14531535e-01 -8.09170306e-02 3.80058914e-01 7.39128292e-01 1.09885991e+00 1.15229219e-01 5.28695107e-01 1.15188253e+00 -5.76232910e-01 -1.08848548e+00 -5.93690336e-01 -5.49500346e-01 1.81322843e-01 2.73825526e-01 -1.09611189e+00 -3.29854816e-01 -7.09532320e-01]
[4.533801078796387, 1.9693312644958496]
19292532-82fe-4f43-92ba-906937de66c6
target-driven-structured-transformer-planner
2207.11201
null
https://arxiv.org/abs/2207.11201v1
https://arxiv.org/pdf/2207.11201v1.pdf
Target-Driven Structured Transformer Planner for Vision-Language Navigation
Vision-language navigation is the task of directing an embodied agent to navigate in 3D scenes with natural language instructions. For the agent, inferring the long-term navigation target from visual-linguistic clues is crucial for reliable path planning, which, however, has rarely been studied before in literature. In this article, we propose a Target-Driven Structured Transformer Planner (TD-STP) for long-horizon goal-guided and room layout-aware navigation. Specifically, we devise an Imaginary Scene Tokenization mechanism for explicit estimation of the long-term target (even located in unexplored environments). In addition, we design a Structured Transformer Planner which elegantly incorporates the explored room layout into a neural attention architecture for structured and global planning. Experimental results demonstrate that our TD-STP substantially improves previous best methods' success rate by 2% and 5% on the test set of R2R and REVERIE benchmarks, respectively. Our code is available at https://github.com/YushengZhao/TD-STP .
['Si Liu', 'Huaxia Xia', 'Haibing Ren', 'Lirong Yang', 'Wenguan Wang', 'Chen Gao', 'Jinyu Chen', 'Yusheng Zhao']
2022-07-19
null
null
null
null
['vision-language-navigation']
['computer-vision']
[ 3.09691802e-02 1.63002953e-01 1.84872255e-01 -4.19165403e-01 -8.43731940e-01 -6.11998022e-01 5.98087430e-01 -2.20231131e-01 -3.77575606e-01 5.84059596e-01 5.43700039e-01 -7.16753960e-01 -1.92087833e-02 -6.52779400e-01 -8.48028421e-01 -5.79930604e-01 -1.88235775e-01 3.00050884e-01 9.15643200e-02 -4.58253235e-01 3.79675329e-01 2.96001196e-01 -1.10352492e+00 -1.57498047e-01 8.95441055e-01 8.19230855e-01 9.34125364e-01 6.77723110e-01 7.79798701e-02 1.09493935e+00 1.10039078e-01 3.73899311e-01 2.63666272e-01 -4.12368625e-01 -1.00142741e+00 -9.37892348e-02 2.79808845e-02 -3.38784605e-01 -5.45491040e-01 8.36694598e-01 4.39396679e-01 6.65397406e-01 5.81406295e-01 -9.78348613e-01 -7.71664321e-01 5.13053417e-01 -4.66659695e-01 8.88831392e-02 7.32096136e-01 6.60696089e-01 9.55789089e-01 -9.50412333e-01 5.40481210e-01 1.37809801e+00 1.41651556e-01 5.94167054e-01 -9.50972438e-01 -1.77180529e-01 7.65472710e-01 2.65056491e-01 -1.38419950e+00 -6.43316686e-01 7.86147058e-01 -3.39755863e-01 1.15931845e+00 -3.76803731e-03 5.04944623e-01 1.21259391e+00 1.94647312e-01 8.62347841e-01 9.68759537e-01 -2.97709614e-01 3.64497900e-01 -4.46062177e-01 -8.12525451e-02 1.24733937e+00 -1.64284602e-01 3.55522156e-01 -4.19509590e-01 4.27357107e-01 9.36330616e-01 -2.73212105e-01 -6.13391459e-01 -6.11478567e-01 -1.56931353e+00 6.39630914e-01 9.97893214e-01 4.39472720e-02 -5.00700712e-01 3.20154399e-01 9.40201953e-02 -3.17953825e-02 1.30200878e-01 6.11602366e-01 -1.47685662e-01 -2.59539813e-01 -1.18108198e-01 2.37604246e-01 5.15997350e-01 1.34088850e+00 6.01367176e-01 8.06140751e-02 -3.72455269e-01 4.79256809e-01 4.37747627e-01 6.14676774e-01 1.90038010e-01 -1.25555623e+00 7.41691589e-01 4.55925375e-01 5.83799303e-01 -9.37018991e-01 -7.31064975e-01 -1.87819228e-01 -5.46173036e-01 2.62460113e-01 5.11852145e-01 -8.18668008e-02 -1.06559193e+00 1.76582205e+00 3.93434554e-01 -7.44213238e-02 2.35114589e-01 1.30338264e+00 9.37186241e-01 7.77857065e-01 -1.47632673e-01 2.83088535e-01 1.20325959e+00 -1.84595442e+00 -4.73189741e-01 -8.26727509e-01 5.17416060e-01 -2.20171884e-01 1.61203289e+00 1.61625117e-01 -8.52079928e-01 -4.13526922e-01 -7.99358785e-01 -4.56842542e-01 -9.49614719e-02 1.04892537e-01 5.67958117e-01 7.46547431e-02 -1.33574843e+00 -3.18963863e-02 -1.01063144e+00 -5.97023129e-01 4.48247671e-01 -4.13408205e-02 -5.28443694e-01 -2.79561907e-01 -7.84423113e-01 9.19441462e-01 2.27004930e-01 4.06811357e-01 -1.46097231e+00 -1.83075055e-01 -1.21570337e+00 -8.40869620e-02 7.25280702e-01 -8.12409759e-01 1.55955243e+00 -2.75605649e-01 -1.81390834e+00 5.88275492e-01 -5.19093931e-01 -2.81293392e-01 3.92122507e-01 -3.68987560e-01 1.13822103e-01 3.47192436e-02 4.59762931e-01 8.00621510e-01 3.24412465e-01 -1.39839387e+00 -5.08192897e-01 -4.02140975e-01 4.54592586e-01 8.18941414e-01 4.68536317e-01 -5.20796359e-01 -6.41229808e-01 -1.40534937e-01 3.85419101e-01 -8.60400915e-01 -8.60175610e-01 1.03361540e-01 -7.67691493e-01 -1.59261867e-01 6.45951182e-02 -5.66158712e-01 6.94101274e-01 -1.90029538e+00 2.98648775e-01 -1.23926289e-01 1.41859621e-01 -4.05510873e-01 -3.08868825e-01 4.10949111e-01 3.96832198e-01 -2.08947390e-01 -2.71803498e-01 -5.69896996e-01 1.16144285e-01 1.53024465e-01 -5.27266681e-01 3.88839364e-01 -1.85274392e-01 1.23234761e+00 -1.18490613e+00 -1.09319851e-01 4.20652360e-01 4.26537335e-01 -5.52169383e-01 3.16898346e-01 -3.91112715e-01 9.42781329e-01 -8.62745047e-01 6.34086549e-01 2.86862820e-01 -2.90220141e-01 -8.89309943e-02 3.13166261e-01 -6.10940337e-01 7.04047680e-01 -6.16180837e-01 2.37741470e+00 -7.11896658e-01 5.96178830e-01 -1.98509209e-02 -6.46101475e-01 7.72734046e-01 -1.88348293e-01 -2.48484001e-01 -1.28740394e+00 9.99547318e-02 9.44198892e-02 -3.53026032e-01 -4.53479230e-01 6.47220373e-01 3.60778093e-01 -1.94457829e-01 2.10403264e-01 -5.12984216e-01 -6.84645623e-02 -1.67243168e-01 1.81598794e-02 1.20988476e+00 6.83876514e-01 5.20284295e-01 -2.36994848e-01 4.51681346e-01 4.29199278e-01 2.81397164e-01 1.08725643e+00 -3.85174960e-01 5.88066697e-01 3.09169739e-01 -4.94033873e-01 -6.76407099e-01 -1.12850642e+00 5.54281771e-01 1.21728015e+00 7.38479733e-01 -2.00373501e-01 -5.84394872e-01 -5.17926216e-01 -5.52471697e-01 1.14991450e+00 -6.72771692e-01 6.68160394e-02 -6.74987793e-01 -9.93696004e-02 1.35792196e-01 5.81269741e-01 6.13231480e-01 -1.35140455e+00 -1.23963487e+00 1.50514856e-01 -6.04338944e-01 -1.17277038e+00 -7.42748797e-01 2.84001350e-01 -4.82763261e-01 -9.24047172e-01 -5.50300360e-01 -9.62414443e-01 8.75818014e-01 8.43844950e-01 9.55915034e-01 -1.79472432e-01 1.50541337e-02 5.07277071e-01 -4.75173801e-01 1.45210885e-02 1.81695640e-01 -2.70703691e-03 -1.30483508e-01 -3.96218032e-01 7.96678960e-02 -4.80258465e-01 -7.95156658e-01 3.57733160e-01 7.78352469e-02 6.29925668e-01 4.92635220e-01 7.53287673e-01 6.46012843e-01 -2.61842906e-01 1.52451813e-01 -3.22096825e-01 5.75942636e-01 -3.22715968e-01 -8.87727261e-01 5.49721085e-02 -3.90599936e-01 3.04368645e-01 5.75698137e-01 -4.65011857e-02 -1.21048188e+00 3.46565843e-02 -2.12278172e-01 -4.03994806e-02 -5.08334875e-01 5.30709326e-01 -3.16167593e-01 -1.18693113e-02 5.34734964e-01 5.71367085e-01 -3.72973800e-01 -2.48257861e-01 6.26891971e-01 9.21827629e-02 6.52883947e-01 -7.69852638e-01 6.53842688e-01 7.07745373e-01 -7.95679390e-02 -6.17247105e-01 -8.47783446e-01 -3.74456167e-01 -5.50759375e-01 -1.34542435e-01 9.91624773e-01 -9.79134560e-01 -7.80234694e-01 1.77601919e-01 -1.26605105e+00 -1.11597574e+00 -5.60179055e-02 3.56228203e-01 -9.96807992e-01 4.13460918e-02 -7.33903795e-02 -9.00386274e-01 -1.81777164e-01 -1.51665437e+00 1.07575715e+00 3.56605887e-01 -2.24775150e-02 -8.28238249e-01 1.31821975e-01 2.58517981e-01 3.86799634e-01 1.83728650e-01 6.12375855e-01 -2.92366177e-01 -1.16044641e+00 2.69535899e-01 -3.75717372e-01 -5.63481152e-01 1.33420587e-01 -7.47543275e-01 -8.46533775e-01 -1.84822790e-02 -2.35695302e-01 -2.41945535e-01 8.85745108e-01 5.91758251e-01 9.63428259e-01 -3.17258060e-01 -5.29912531e-01 8.70887697e-01 1.31302023e+00 6.15729809e-01 5.98074675e-01 8.40402365e-01 8.96210015e-01 6.83615863e-01 8.34917426e-01 2.92109549e-01 1.07443070e+00 6.90844119e-01 9.18017507e-01 2.04053998e-01 7.00904578e-02 -6.57305896e-01 4.12660211e-01 2.24003583e-01 -6.50428161e-02 -4.92253423e-01 -1.04995787e+00 7.10586965e-01 -2.01401305e+00 -8.39835167e-01 1.32525548e-01 1.96619594e+00 3.62023622e-01 1.39805853e-01 -2.30445683e-01 -4.96708244e-01 2.29360774e-01 4.31291699e-01 -7.83304513e-01 -2.15862200e-01 1.74577147e-01 -3.65735799e-01 4.06910002e-01 1.13310432e+00 -9.48210537e-01 1.41439617e+00 4.56988382e+00 4.26566333e-01 -7.59828866e-01 2.96121109e-02 4.10101175e-01 -1.04406802e-02 -3.10245156e-01 -6.10835152e-03 -8.08628857e-01 -4.90988828e-02 5.11381865e-01 -2.62123849e-02 8.57539296e-01 1.07202053e+00 6.55079484e-01 -3.72037977e-01 -1.11858523e+00 9.78721142e-01 -1.03260249e-01 -1.07131398e+00 -3.19219798e-01 2.19836328e-02 4.34507877e-01 3.25493515e-01 2.40289956e-01 5.12327433e-01 5.85419536e-01 -1.29618394e+00 9.84532297e-01 4.86168444e-01 4.97138768e-01 -6.75497890e-01 2.41811559e-01 5.74788809e-01 -1.40077055e+00 -1.52036428e-01 -2.04182953e-01 -2.70759881e-01 4.67982858e-01 -5.86449802e-02 -9.29702461e-01 4.98028576e-01 6.53489172e-01 6.58820152e-01 -3.63382131e-01 9.86205935e-01 -7.02863753e-01 1.82754248e-01 -1.80764794e-01 -3.25607806e-01 8.11432004e-01 -3.21621269e-01 7.47192085e-01 7.04252422e-01 3.77738386e-01 3.22666317e-01 3.75568062e-01 1.00846469e+00 3.39332521e-01 -1.74259976e-01 -9.32897687e-01 1.75098956e-01 3.95438790e-01 9.47797954e-01 -7.86320865e-01 2.37898454e-01 -8.52042437e-02 1.09749234e+00 7.03692734e-01 6.81328475e-01 -9.07917619e-01 -2.15090171e-01 6.12538338e-01 -8.34785029e-02 4.16476697e-01 -6.37949407e-01 -2.95334637e-01 -9.07034814e-01 7.67167881e-02 -5.13155222e-01 -1.72208533e-01 -1.22122693e+00 -6.01262510e-01 9.78460431e-01 -2.26873890e-01 -1.01831698e+00 -1.59086421e-01 -5.77582300e-01 -5.83603978e-01 9.48136508e-01 -1.78559232e+00 -1.23957407e+00 -7.34567583e-01 4.67653066e-01 9.37718928e-01 4.84327152e-02 9.70230401e-01 -3.00839782e-01 -4.37150002e-01 2.34438390e-01 -1.57409281e-01 2.96437535e-02 3.07133466e-01 -1.26011300e+00 9.36946571e-01 9.95665312e-01 -1.04755275e-01 6.98397815e-01 8.22621644e-01 -4.50789571e-01 -1.55460560e+00 -1.05678713e+00 5.81510603e-01 -6.27045274e-01 4.00354683e-01 -4.21621442e-01 -5.69564700e-01 1.01614642e+00 3.28303218e-01 -1.34850934e-01 1.56601608e-01 -2.15003155e-02 -3.78045410e-01 2.78018057e-01 -8.69856477e-01 1.31601894e+00 1.59777570e+00 -2.95532167e-01 -4.82180119e-01 3.07339042e-01 1.14654732e+00 -8.56734514e-01 2.81881448e-02 2.46073604e-02 4.58142012e-01 -1.06970811e+00 1.06842124e+00 -2.33639404e-01 3.72588038e-01 -5.64572573e-01 -4.99602735e-01 -1.36105883e+00 -4.21331704e-01 -6.87731326e-01 3.34789567e-02 6.23509228e-01 5.27274966e-01 -5.78985274e-01 5.65075397e-01 4.18013245e-01 -5.36732733e-01 -8.54290783e-01 -9.06746149e-01 -5.91881394e-01 -3.49120021e-01 -5.69356978e-01 4.60311413e-01 3.09127092e-01 -2.92265490e-02 4.37469661e-01 -3.79650235e-01 6.16671741e-01 5.92642009e-01 2.77676135e-01 8.95406544e-01 -6.45656347e-01 5.30995801e-02 -5.27590573e-01 1.08077824e-01 -1.97604370e+00 3.41000199e-01 -7.16638148e-01 7.95530915e-01 -2.35112166e+00 -1.96934283e-01 -4.96836782e-01 3.66375931e-02 5.03355622e-01 7.70386308e-02 -3.48406136e-01 8.13203827e-02 -1.28204271e-01 -8.62250328e-01 1.05348504e+00 1.61138999e+00 -1.34283274e-01 -5.57468951e-01 -1.09451562e-01 -9.18659091e-01 7.30894566e-01 9.07796264e-01 -1.39133215e-01 -7.67928958e-01 -8.87986422e-01 1.40685886e-01 2.46668160e-01 5.98997474e-01 -9.77528095e-01 5.59647024e-01 -5.62730014e-01 -2.80150790e-02 -8.20868254e-01 6.73414588e-01 -6.62925065e-01 -4.05781031e-01 4.52505857e-01 -4.72537309e-01 2.41709560e-01 1.55672371e-01 8.65930378e-01 3.45093757e-01 8.25457200e-02 3.76379341e-01 -4.36446071e-01 -1.34490049e+00 3.69353831e-01 -4.67190474e-01 1.74642771e-01 8.89582455e-01 -2.07468182e-01 -4.92779374e-01 -5.03792226e-01 -5.34716964e-01 8.41891229e-01 4.60294425e-01 5.06402254e-01 1.17895591e+00 -1.19419265e+00 -3.85498375e-01 7.97413811e-02 3.16263795e-01 5.91216564e-01 3.25221062e-01 8.62397373e-01 -7.04915404e-01 9.03420806e-01 -1.08343050e-01 -5.26636720e-01 -7.81075835e-01 6.40812576e-01 4.76160556e-01 -7.78552294e-02 -1.07151675e+00 1.19369447e+00 7.81257153e-01 -7.67401695e-01 3.28275234e-01 -5.91467083e-01 -3.73996854e-01 -6.16157293e-01 5.81997693e-01 2.91179735e-02 -5.40622652e-01 -6.08983815e-01 -5.19335926e-01 6.46630704e-01 1.22247159e-01 -3.66686463e-01 1.21262026e+00 -6.78118050e-01 4.37141433e-02 4.94443089e-01 6.74721539e-01 -7.99767002e-02 -1.62609351e+00 -1.42731279e-01 -8.56139436e-02 -3.33714843e-01 2.75249574e-02 -8.91505599e-01 -5.78923821e-01 8.56429398e-01 2.57343590e-01 -2.72648752e-01 8.94372940e-01 7.95433968e-02 5.01373351e-01 9.00351167e-01 8.70770633e-01 -3.91690195e-01 2.86499709e-01 1.01175141e+00 1.27028334e+00 -1.42837811e+00 -5.03544927e-01 -1.03465848e-01 -8.66930366e-01 7.80148506e-01 1.04765534e+00 1.00331664e-01 2.70326287e-01 -1.14749141e-01 5.72256893e-02 -2.92614311e-01 -7.54640520e-01 -4.66466844e-01 1.84164107e-01 9.09228683e-01 -5.59359491e-02 9.07887965e-02 4.00432020e-01 5.50224841e-01 -3.48979533e-01 -3.51171196e-01 4.86817747e-01 8.48835409e-01 -8.00201297e-01 -3.59963357e-01 -1.87694803e-01 -2.16679007e-01 3.48274022e-01 -3.07357430e-01 -1.33431211e-01 6.56732738e-01 -2.70508051e-01 1.02166557e+00 -1.46736667e-01 -2.30562210e-01 3.99495065e-01 -2.24780113e-01 4.94568825e-01 -6.53369129e-01 2.80449539e-02 -6.83440268e-02 2.69866943e-01 -1.04741716e+00 -2.40684792e-01 -4.83790815e-01 -1.72316790e+00 2.30798479e-02 1.39990345e-01 5.49262203e-02 6.05592728e-01 9.36492801e-01 4.30379391e-01 7.87335098e-01 3.91770691e-01 -1.27760100e+00 -1.52130201e-01 -6.45183980e-01 1.07864449e-02 -3.67799252e-01 7.47830808e-01 -8.49165797e-01 -2.45570987e-01 -2.47417524e-01]
[4.508424758911133, 0.5073919892311096]
e752bd1a-912d-4e28-946c-beff6c4297d5
bmad-benchmarks-for-medical-anomaly-detection
2306.11876
null
https://arxiv.org/abs/2306.11876v2
https://arxiv.org/pdf/2306.11876v2.pdf
BMAD: Benchmarks for Medical Anomaly Detection
Anomaly detection (AD) is a fundamental research problem in machine learning and computer vision, with practical applications in industrial inspection, video surveillance, and medical diagnosis. In medical imaging, AD is especially vital for detecting and diagnosing anomalies that may indicate rare diseases or conditions. However, there is a lack of a universal and fair benchmark for evaluating AD methods on medical images, which hinders the development of more generalized and robust AD methods in this specific domain. To bridge this gap, we introduce a comprehensive evaluation benchmark for assessing anomaly detection methods on medical images. This benchmark encompasses six reorganized datasets from five medical domains (i.e. brain MRI, liver CT, retinal OCT, chest X-ray, and digital histopathology) and three key evaluation metrics, and includes a total of fourteen state-of-the-art AD algorithms. This standardized and well-curated medical benchmark with the well-structured codebase enables comprehensive comparisons among recently proposed anomaly detection methods. It will facilitate the community to conduct a fair comparison and advance the field of AD on medical imaging. More information on BMAD is available in our GitHub repository: https://github.com/DorisBao/BMAD
['Xingyu Li', 'Zhaoxiang Zhang', 'Yinsheng He', 'Hanqiu Deng', 'Hanshi Sun', 'Jinan Bao']
2023-06-20
null
null
null
null
['medical-diagnosis', 'anomaly-detection']
['medical', 'methodology']
[ 7.25588277e-02 -2.01372519e-01 2.64726188e-02 -1.42729729e-01 -5.88731945e-01 -1.94629416e-01 1.40158966e-01 6.84032381e-01 -5.10525852e-02 2.94762135e-01 -2.42685780e-01 -3.82359117e-01 -1.64081350e-01 -4.85240817e-01 -1.17546998e-01 -8.01202834e-01 -2.39696845e-01 3.02353024e-01 3.02397192e-01 1.63038686e-01 2.42729962e-01 7.75754690e-01 -1.49026990e+00 2.55091578e-01 9.63336706e-01 1.25916982e+00 -3.25106196e-02 4.19140548e-01 -3.00093181e-02 6.52160466e-01 -5.54653227e-01 -3.74525279e-01 1.03025377e-01 -6.36236489e-01 -9.09081757e-01 2.65742332e-01 4.26996320e-01 -3.92450184e-01 -3.47501159e-01 1.26263928e+00 6.86560333e-01 -2.12938979e-01 7.40247130e-01 -1.34599531e+00 -7.64533758e-01 -4.64491993e-02 -6.90386653e-01 8.65159810e-01 3.01513076e-01 4.35757309e-01 8.14460039e-01 -7.90860891e-01 3.73259693e-01 6.81968689e-01 4.49631751e-01 7.29972959e-01 -6.63405359e-01 -3.57034862e-01 5.89051545e-02 7.13065743e-01 -1.01215482e+00 -7.56621435e-02 6.40320778e-01 -6.63770854e-01 6.44189537e-01 3.58556360e-01 6.10203862e-01 8.09251666e-01 5.53821743e-01 1.02608371e+00 8.05605233e-01 -2.54290164e-01 2.40697831e-01 -3.92361671e-01 1.52728051e-01 9.82698381e-01 5.59866488e-01 -1.30910173e-01 -2.75624812e-01 -3.81590247e-01 5.45014858e-01 3.76730174e-01 -2.36482203e-01 -4.03816044e-01 -1.25327957e+00 7.11409152e-01 3.50421220e-01 3.79005551e-01 -7.07706034e-01 -2.82600880e-01 8.10754061e-01 4.04128015e-01 4.33830857e-01 4.65855390e-01 -3.96488577e-01 8.97899568e-02 -5.44617653e-01 1.02879256e-01 5.02380431e-01 5.11910081e-01 3.33233118e-01 2.61839032e-01 -1.20566331e-01 1.04173756e+00 3.99417937e-01 3.07921231e-01 6.88526690e-01 -8.19803178e-01 -8.25112239e-02 6.84826434e-01 -4.66064900e-01 -9.72296774e-01 -2.35623017e-01 -2.55065501e-01 -1.03742468e+00 3.16614032e-01 3.52228105e-01 3.10095251e-01 -9.50766921e-01 9.78154898e-01 5.11679232e-01 3.57185543e-01 -2.63322920e-01 9.06693101e-01 1.06983721e+00 7.96391740e-02 -1.30291268e-01 -2.37291321e-01 1.46724868e+00 -8.54169011e-01 -8.10863137e-01 -1.75728843e-01 7.12892890e-01 -7.64757752e-01 9.91814852e-01 8.15611899e-01 -9.37927127e-01 -1.34322777e-01 -9.89648521e-01 1.28063053e-01 -3.98019522e-01 3.00442688e-02 4.74347055e-01 3.86902750e-01 -5.82121670e-01 2.95840204e-01 -1.21031213e+00 -5.97521544e-01 8.09828281e-01 -2.62628905e-02 -2.84841508e-01 -3.88179630e-01 -9.11682725e-01 9.87172365e-01 -1.04317605e-01 1.41183585e-01 -1.15595543e+00 -7.82294631e-01 -9.56276059e-01 -6.15405619e-01 3.37859482e-01 -5.77262521e-01 1.22533154e+00 -4.39965218e-01 -9.33416069e-01 1.28027320e+00 4.34907898e-02 -4.57383007e-01 2.95526266e-01 -3.23145449e-01 -7.75955260e-01 4.02500033e-01 3.18780363e-01 1.41583517e-01 7.04536080e-01 -9.48373377e-01 -5.31654894e-01 -6.67777181e-01 -2.31251419e-01 -2.74941266e-01 -3.48391056e-01 1.04823001e-01 -1.42952591e-01 -8.12179804e-01 2.90868163e-01 -7.20940709e-01 -3.20323646e-01 4.14660335e-01 -6.32185996e-01 -2.41943225e-01 9.08514798e-01 -8.37139487e-01 1.31036448e+00 -2.37354469e+00 -9.58590358e-02 1.11878410e-01 4.39377457e-01 4.51734394e-01 5.61259240e-02 9.37840901e-03 -2.97897279e-01 -1.46537066e-01 -7.38046050e-01 1.85193587e-03 -2.98638403e-01 2.07718879e-01 1.14911422e-01 8.76454651e-01 3.22347552e-01 6.06214166e-01 -9.15267467e-01 -4.91934717e-01 4.92585480e-01 9.22685936e-02 -3.30113322e-01 2.74700165e-01 -9.17874090e-03 8.73838723e-01 -6.38337016e-01 1.29553795e+00 4.15278018e-01 -3.16273302e-01 -3.24584097e-01 -1.71275035e-01 2.77177304e-01 2.04702839e-02 -1.04348135e+00 1.68882477e+00 1.23190179e-01 4.16716188e-01 -8.88303109e-03 -1.27272856e+00 7.58978844e-01 3.31781745e-01 1.09207737e+00 -7.01844752e-01 1.11747473e-01 3.72073025e-01 1.42414212e-01 -9.10407782e-01 -1.93006843e-01 2.46687055e-01 2.29233637e-01 4.85235095e-01 -1.05836140e-02 -1.03916995e-01 5.13934731e-01 2.13056114e-02 1.34177732e+00 -4.09492195e-01 6.16197288e-01 -2.07360566e-01 8.06493163e-01 9.94254351e-02 5.65244555e-01 4.14989859e-01 -6.19360328e-01 8.24066758e-01 3.68826479e-01 -7.64185786e-01 -7.02367961e-01 -1.22222579e+00 -5.88529468e-01 6.32317901e-01 8.38800967e-02 -2.51971841e-01 -6.56923592e-01 -8.82748485e-01 6.70056939e-02 4.04294074e-01 -4.90202338e-01 -3.47640604e-01 -4.19750214e-01 -1.14523220e+00 5.90130925e-01 4.34065938e-01 5.53906679e-01 -1.11958349e+00 -5.30763865e-01 -6.02220260e-02 -3.17817122e-01 -9.05641198e-01 -2.47265965e-01 -2.82042891e-01 -1.10698056e+00 -1.71087432e+00 -7.86126554e-01 -6.72336042e-01 9.01336074e-01 2.00893497e-03 1.23713660e+00 5.37185669e-01 -1.10028803e+00 7.23147035e-01 -3.35112333e-01 -8.26652884e-01 -5.32466352e-01 -3.60541433e-01 1.98239401e-01 -1.24371246e-01 5.78457713e-01 -3.42378318e-01 -9.30539072e-01 4.25702631e-01 -1.16853487e+00 -6.06916368e-01 7.43417323e-01 7.53894091e-01 9.34043586e-01 -1.03761829e-01 3.90459538e-01 -7.68565416e-01 5.44181228e-01 -6.71998203e-01 -4.37667847e-01 7.65858293e-02 -1.01908433e+00 -3.91353965e-01 1.21722445e-01 -5.48935086e-02 -7.29587257e-01 -1.73991606e-01 -4.90929574e-01 -2.89525598e-01 -6.53067291e-01 5.83758771e-01 1.19030923e-01 3.70358713e-02 9.01149929e-01 4.61426564e-02 2.55067408e-01 -5.66938519e-01 -6.32979199e-02 5.29017448e-01 6.79980695e-01 -3.68392378e-01 4.55797970e-01 4.46773261e-01 1.62374169e-01 -9.15682077e-01 -7.23017812e-01 -7.05320001e-01 -6.39549673e-01 -1.86864972e-01 8.92776012e-01 -6.44316196e-01 -1.01988139e-02 8.11401248e-01 -8.80806863e-01 -6.01872541e-02 -3.37089241e-01 4.61812705e-01 -2.98275113e-01 7.77369618e-01 -8.74857605e-01 -3.02604020e-01 -5.16939163e-01 -1.22268701e+00 8.39221001e-01 -1.20692283e-01 -3.31048250e-01 -9.84229565e-01 1.91856936e-01 3.72117519e-01 3.36664021e-01 6.01297557e-01 1.14620602e+00 -8.96585405e-01 -3.88228685e-01 -4.66594785e-01 5.34280874e-02 7.22013116e-01 4.60259348e-01 1.85134396e-01 -8.25590551e-01 -1.96913540e-01 1.90860480e-02 -7.49422684e-02 7.69229412e-01 5.20500064e-01 1.65403247e+00 -6.04608795e-03 -3.58176142e-01 3.70968163e-01 9.80442166e-01 3.80969286e-01 7.26121187e-01 5.14115810e-01 6.44612312e-01 3.90621245e-01 8.73546422e-01 4.28506076e-01 1.44315362e-01 3.73471946e-01 8.48540246e-01 -3.21349472e-01 -1.13027796e-01 6.61528647e-01 7.22428486e-02 7.51131594e-01 -2.05067679e-01 -9.07250568e-02 -1.25154209e+00 7.60888875e-01 -1.53569710e+00 -7.13082552e-01 -4.25769538e-01 2.12255764e+00 6.30648553e-01 -1.33107334e-01 1.69222161e-01 4.89047974e-01 7.60428011e-01 -2.13185415e-01 -8.29728544e-01 -1.07919686e-01 1.60638869e-01 1.17609531e-01 -1.07525900e-01 -7.88764730e-02 -1.32760942e+00 1.20888442e-01 6.55932999e+00 4.69041556e-01 -1.22618079e+00 1.97949797e-01 7.41894543e-01 2.33838856e-02 1.79084376e-01 -7.92220712e-01 -1.70024022e-01 5.22035182e-01 7.58109808e-01 -1.37438878e-01 -1.12071149e-01 1.02526486e+00 -2.03443039e-02 -2.67315954e-02 -1.21269023e+00 8.84028077e-01 1.17954858e-01 -1.06440639e+00 -6.57131299e-02 7.32188448e-02 5.29027104e-01 2.87616491e-01 2.61624187e-01 -1.81518033e-01 -1.18843094e-01 -7.92545736e-01 1.84570551e-02 3.08176875e-01 6.71076417e-01 -4.61464167e-01 9.47591364e-01 -1.71958774e-01 -8.31606805e-01 -5.42068444e-02 -1.56229302e-01 3.29997331e-01 7.17374086e-02 1.05147815e+00 -5.09966493e-01 6.92945182e-01 1.12979054e+00 1.01009178e+00 -6.47401214e-01 1.50695753e+00 -1.76710784e-01 6.46979392e-01 2.66032875e-01 6.13846898e-01 -1.05815262e-01 -5.21714129e-02 8.06606114e-01 1.08170879e+00 4.13706005e-01 -9.92984623e-02 1.79216504e-01 6.26384020e-01 1.93162680e-01 1.43750951e-01 -6.62135839e-01 -3.72161269e-02 9.70267802e-02 1.35061800e+00 -7.59831190e-01 -1.25034213e-01 -7.03085840e-01 9.68222201e-01 -4.07933414e-01 6.14327230e-02 -7.88375854e-01 -2.00307786e-01 9.73292768e-01 2.76721239e-01 -1.66804329e-01 -1.12874005e-02 -2.33752787e-01 -1.04547167e+00 1.29190935e-02 -1.18446076e+00 9.91144776e-01 -4.63744909e-01 -1.87686253e+00 6.70116246e-01 1.82368040e-01 -1.60759068e+00 -2.68752933e-01 -9.25918818e-01 -7.30837107e-01 3.87812883e-01 -1.39381671e+00 -8.33868206e-01 -7.26104856e-01 7.03290641e-01 4.57336485e-01 -5.91068387e-01 9.64486420e-01 7.42749333e-01 -9.52249467e-01 3.59141320e-01 -1.57880902e-01 3.12531382e-01 8.17946553e-01 -1.39022207e+00 3.43572766e-01 1.03781617e+00 -1.08476125e-01 3.50930572e-01 5.77118814e-01 -5.05374074e-01 -9.40783381e-01 -1.30497897e+00 1.66392714e-01 -7.60135472e-01 5.71498275e-01 3.65995169e-01 -1.29701209e+00 6.86726928e-01 5.07732518e-02 6.10339940e-01 9.51607287e-01 -4.05076325e-01 -5.02961949e-02 -1.42235488e-01 -1.41001749e+00 2.08662838e-01 8.30967009e-01 -2.63680875e-01 -4.93358046e-01 6.52434170e-01 1.81066513e-01 -6.03274584e-01 -1.11287725e+00 6.71438158e-01 2.11740151e-01 -9.29673493e-01 1.04086804e+00 -5.70751011e-01 5.01562059e-01 -4.81589079e-01 1.19169652e-01 -1.16452098e+00 -2.01866716e-01 -7.32739344e-02 -3.88970077e-01 9.86483514e-01 1.54287770e-01 -9.93897438e-01 4.67723459e-01 2.67015368e-01 -6.82063997e-01 -1.11752391e+00 -7.79982924e-01 -7.34707415e-01 3.15859392e-02 -4.37884480e-01 4.75528449e-01 1.08266258e+00 -2.49513298e-01 -2.49044940e-01 1.96272030e-01 3.39858651e-01 6.46008074e-01 -1.52311280e-01 4.03862000e-01 -1.44592941e+00 1.59729555e-01 -5.35032034e-01 -9.35445070e-01 -4.14548457e-01 -1.88354224e-01 -9.07798767e-01 -3.34364735e-02 -1.70689189e+00 1.54282808e-01 -2.76224524e-01 -6.89136744e-01 5.60011208e-01 -1.46878660e-01 5.63937366e-01 -5.01944542e-01 3.44769388e-01 -5.29081762e-01 2.63408542e-01 1.30037749e+00 -3.05716604e-01 2.29318559e-01 1.05884463e-01 -5.89787066e-01 9.64189112e-01 9.02890563e-01 -4.19914931e-01 -2.16770485e-01 -2.80611098e-01 -1.95751131e-01 -5.11416972e-01 5.26373208e-01 -1.15678108e+00 1.13364749e-01 -1.06142655e-01 3.60315979e-01 -4.94743764e-01 -1.37232125e-01 -7.42303193e-01 -2.10251108e-01 8.81305993e-01 4.80010435e-02 5.20456672e-01 2.05576077e-01 4.53559995e-01 -5.68672657e-01 -1.79368064e-01 9.50626552e-01 -3.54891539e-01 -1.09550130e+00 7.81871021e-01 -4.60405380e-01 5.85311987e-02 1.32990253e+00 -1.90114945e-01 -3.60782802e-01 -7.11422935e-02 -7.55142272e-01 2.20216691e-01 2.65436321e-01 3.52915466e-01 8.46882463e-01 -1.34059072e+00 -8.15104127e-01 2.97350198e-01 5.61837494e-01 3.56691599e-01 3.12856704e-01 1.36010945e+00 -7.46474266e-01 1.87432721e-01 -6.05650604e-01 -9.64542925e-01 -1.36823225e+00 4.52311218e-01 5.65232992e-01 -9.81553793e-02 -7.37454355e-01 5.46723545e-01 1.84344560e-01 -2.21061945e-01 2.41650760e-01 -4.34353948e-01 -2.03120515e-01 -2.33819306e-01 6.69931054e-01 7.60967016e-01 6.36791110e-01 -4.60007429e-01 -6.06793880e-01 3.01795274e-01 -1.37521207e-01 6.22184515e-01 1.27236497e+00 -9.31273177e-02 -6.05200946e-01 4.63021249e-01 6.71206713e-01 -2.43769571e-01 -5.89354336e-01 -1.20601811e-01 1.35239244e-01 -4.88286793e-01 -1.88211109e-02 -6.51559174e-01 -1.47012699e+00 8.58408689e-01 1.03166258e+00 4.86009806e-01 1.42147052e+00 3.15192610e-01 9.29747701e-01 3.38997692e-01 8.62649828e-02 -8.39844763e-01 7.34522104e-01 2.46978372e-01 1.01637840e+00 -1.38351417e+00 2.94540167e-01 -5.04216909e-01 -7.24665523e-01 1.08890545e+00 9.37839150e-01 7.38810375e-02 8.55065167e-01 1.80801928e-01 4.86050159e-01 -7.25715339e-01 -2.30020955e-01 -5.54813221e-02 5.65157652e-01 6.85173631e-01 6.67355418e-01 -1.60821080e-01 -2.03273654e-01 3.84256572e-01 3.41200113e-01 -1.75930887e-01 3.40446532e-01 1.07778716e+00 -3.85904253e-01 -1.07037950e+00 -3.16747636e-01 1.03123403e+00 -1.00904012e+00 2.08802834e-01 -1.95381612e-01 6.15137458e-01 1.39953077e-01 7.73148596e-01 9.90135521e-02 -1.03074489e-02 5.07725179e-01 -2.01234780e-02 3.44323635e-01 -6.86438799e-01 -1.51927546e-01 -2.43574634e-01 -9.64366123e-02 -8.30098271e-01 -5.47399580e-01 -8.10154796e-01 -1.24491847e+00 -1.69935346e-01 -2.29307637e-01 -1.30546749e-01 5.12247026e-01 9.35640037e-01 3.98259103e-01 7.81889319e-01 3.01750243e-01 -2.90943444e-01 -2.74795532e-01 -9.08991456e-01 -6.98516250e-01 6.35081410e-01 4.82838184e-01 -8.76467943e-01 -3.29597116e-01 2.04149440e-01]
[7.627323627471924, 2.054788112640381]
2d21d0e6-5e31-4eab-8b80-13879404cf8d
cross-lingual-wolastoqey-english-definition
null
null
https://aclanthology.org/2021.ranlp-main.17
https://aclanthology.org/2021.ranlp-main.17.pdf
Cross-Lingual Wolastoqey-English Definition Modelling
Definition modelling is the task of automatically generating a dictionary-style definition given a target word. In this paper, we consider cross-lingual definition generation. Specifically, we generate English definitions for Wolastoqey (Malecite-Passamaquoddy) words. Wolastoqey is an endangered, low-resource polysynthetic language. We hypothesize that sub-word representations based on byte pair encoding (Sennrich et al., 2016) can be leveraged to represent morphologically-complex Wolastoqey words and overcome the challenge of not having large corpora available for training. Our experimental results demonstrate that this approach outperforms baseline methods in terms of BLEU score. 
['Paul Cook', 'Diego Bear']
null
null
https://aclanthology.org/2021.ranlp-1.17
https://aclanthology.org/2021.ranlp-1.17.pdf
ranlp-2021-9
['definition-modelling']
['natural-language-processing']
[ 2.90284604e-01 2.00738922e-01 -2.48658001e-01 -2.03867808e-01 -8.29267502e-01 -1.07028723e+00 7.97103584e-01 2.45156348e-01 -7.00450540e-01 1.20840228e+00 4.15860921e-01 -6.08904302e-01 1.27230957e-01 -9.38408613e-01 -5.10189116e-01 -8.40675682e-02 5.22886038e-01 5.54833710e-01 -4.73321646e-01 -6.31173313e-01 3.06220055e-01 1.49604991e-01 -1.07619393e+00 1.85512304e-01 1.02418137e+00 1.67865321e-01 4.60089147e-01 5.08736014e-01 1.31430458e-02 1.86137527e-01 -7.76530027e-01 -9.46707785e-01 1.35646403e-01 -5.20414472e-01 -8.39572906e-01 -2.10672662e-01 4.59719896e-01 1.11973889e-01 -4.46743183e-02 9.95441794e-01 4.28693265e-01 2.73506761e-01 1.03098106e+00 -5.55599809e-01 -9.59559381e-01 1.35848439e+00 -4.75890003e-02 2.25275025e-01 4.73176807e-01 2.91440785e-01 1.54465711e+00 -1.35671306e+00 8.74251902e-01 1.28557825e+00 5.95307767e-01 8.46514106e-01 -1.36063457e+00 -7.44897306e-01 -9.04704183e-02 5.35839573e-02 -1.54732430e+00 -3.98134351e-01 4.76978540e-01 -1.90098256e-01 1.35317314e+00 3.12016189e-01 7.33972251e-01 1.30464172e+00 1.23438172e-01 6.69068992e-01 9.23730373e-01 -7.79403031e-01 -4.06573825e-02 -2.52643704e-01 -4.05682176e-02 5.10026813e-01 6.00807786e-01 1.15518793e-01 -6.65033698e-01 1.69682160e-01 6.07741892e-01 -5.80848694e-01 -1.72760457e-01 4.11900759e-01 -1.41612864e+00 1.12917328e+00 2.07495894e-02 4.30710256e-01 -2.51271486e-01 2.94376016e-01 5.88698208e-01 1.51053563e-01 3.19435507e-01 1.32504117e+00 -5.71766853e-01 -3.93232673e-01 -8.39606166e-01 3.75552505e-01 5.52831948e-01 1.36311293e+00 5.63495576e-01 3.22975785e-01 -1.04177423e-01 9.79893386e-01 -5.45580797e-02 5.83462775e-01 7.07265377e-01 -4.47887808e-01 4.84143019e-01 1.77328169e-01 3.01883370e-02 -4.20119762e-01 -1.28206342e-01 -1.94900721e-01 -4.06547695e-01 -5.09259760e-01 2.03603089e-01 -3.57625932e-01 -1.09555316e+00 1.74869621e+00 -1.82374865e-01 -3.81665677e-01 3.67445618e-01 5.92870831e-01 9.33904052e-01 9.19532180e-01 3.58350754e-01 -7.08364993e-02 1.38728261e+00 -9.70536530e-01 -5.64103067e-01 -4.63542819e-01 6.55230165e-01 -8.37554514e-01 1.57110894e+00 3.40467125e-01 -1.11902988e+00 -6.47344649e-01 -1.15755713e+00 -1.67296141e-01 -6.28268659e-01 3.54200870e-01 7.48003006e-01 8.78057778e-01 -5.71655691e-01 3.08486819e-01 -4.56863552e-01 -3.23918879e-01 -7.81267434e-02 -1.30695745e-01 -2.42361590e-01 -3.11820991e-02 -1.50997150e+00 1.18147612e+00 1.07609892e+00 -4.17933352e-02 -1.03006768e+00 -7.72719562e-01 -1.27880275e+00 -1.43558279e-01 2.18075156e-01 -5.19416630e-01 1.16591454e+00 -7.87256598e-01 -1.17234135e+00 9.99149024e-01 9.37630534e-02 -3.99767011e-01 3.01518869e-02 -2.43960053e-01 -4.46955085e-01 -8.30148682e-02 2.78363794e-01 8.03787351e-01 4.78200734e-01 -1.05296528e+00 -4.91514713e-01 2.00981393e-01 2.02475652e-01 4.31342870e-01 -1.32669464e-01 2.18118504e-01 -1.07263364e-01 -1.29857194e+00 -3.06274623e-01 -8.12575519e-01 -1.39871433e-01 -5.67526817e-01 -4.26190138e-01 -5.58860719e-01 -5.15661500e-02 -7.17466354e-01 1.44044709e+00 -1.78240609e+00 1.75519660e-01 1.47507221e-01 -1.45415783e-01 5.44599831e-01 -3.64452600e-01 8.60924542e-01 1.13789914e-02 7.49368072e-01 -3.75045329e-01 -1.47846028e-01 3.68881613e-01 4.78596091e-01 -2.79535800e-01 -1.07974827e-01 5.56798220e-01 1.12960410e+00 -1.32617593e+00 -1.64941445e-01 1.75935552e-01 2.88790643e-01 -3.41445267e-01 2.33250648e-01 -4.56693679e-01 2.24162146e-01 -4.29216959e-03 5.38291872e-01 2.36771300e-01 2.96067536e-01 5.20020902e-01 3.01562119e-02 -1.20771579e-01 1.05193007e+00 -7.85125434e-01 1.87243652e+00 -1.14113677e+00 4.72788364e-01 -7.27959454e-01 -4.51992691e-01 1.00593185e+00 1.77003592e-01 -3.04958791e-01 -4.76622373e-01 1.37111902e-01 7.31308401e-01 2.50752062e-01 -7.57181942e-02 1.01509547e+00 -5.28548956e-01 -5.63767254e-01 4.75653172e-01 1.47612467e-01 -1.04383290e+00 7.97328830e-01 3.73409316e-02 8.14785063e-01 3.04223806e-01 8.30051124e-01 -5.75837672e-01 4.36313987e-01 2.55193692e-02 5.89503884e-01 5.69275737e-01 1.86990559e-01 6.82187915e-01 1.74832985e-01 -2.05657616e-01 -1.29342222e+00 -1.51795506e+00 -3.53057653e-01 1.06886446e+00 -9.48532298e-02 -9.70702708e-01 -5.58001518e-01 -6.21942282e-01 -1.25074878e-01 1.32690012e+00 -5.44339418e-01 -9.56119299e-02 -9.55227435e-01 -5.42408884e-01 1.05859625e+00 3.72516245e-01 6.10653386e-02 -1.32618666e+00 -6.36436462e-01 5.17847240e-01 -8.18223208e-02 -8.89856875e-01 -9.29465771e-01 1.83361501e-01 -1.75750121e-01 -7.09866405e-01 -6.63139403e-01 -9.93646979e-01 4.51800168e-01 -3.43206942e-01 1.75811791e+00 8.66878852e-02 -2.73580164e-01 1.02241226e-01 -7.30453491e-01 -6.83633029e-01 -7.08380401e-01 1.71853974e-01 2.03986406e-01 -6.01245165e-01 6.27679884e-01 -7.58831874e-02 -2.45309263e-01 -3.29610944e-01 -1.13008916e+00 6.28018156e-02 3.12680691e-01 1.08410203e+00 6.97991550e-01 -4.42059010e-01 7.99283862e-01 -1.05523360e+00 7.77019799e-01 -4.59786862e-01 -2.87238002e-01 3.43553960e-01 -6.34200275e-01 1.95950463e-01 8.31598878e-01 -5.52532732e-01 -8.71297777e-01 -3.72318149e-01 -4.05415088e-01 2.15179279e-01 8.88155922e-02 6.65881693e-01 -2.90039450e-01 5.10030091e-01 6.15140975e-01 2.39832625e-01 -6.48407936e-01 -2.99386680e-01 7.93157697e-01 6.85877323e-01 4.81067419e-01 -9.84418750e-01 6.92743361e-01 -1.51183188e-01 -4.89269018e-01 -8.54555011e-01 -8.60999346e-01 -9.57190245e-02 -6.01227582e-01 -1.30421971e-03 9.23534453e-01 -8.82721364e-01 1.46692097e-01 7.81371370e-02 -1.36027408e+00 -3.00209463e-01 -4.95393068e-01 3.61455858e-01 -4.79597092e-01 3.00321877e-01 -5.33552825e-01 -5.76199532e-01 -6.45413756e-01 -8.82083416e-01 1.02145159e+00 2.29768023e-01 -8.14856172e-01 -1.16109514e+00 1.72963679e-01 2.72667687e-02 1.85795054e-01 2.52234876e-01 1.15735316e+00 -6.12556815e-01 -1.23407803e-01 3.80238891e-01 -9.97729823e-02 4.83989745e-01 2.71001279e-01 -7.12795258e-02 -3.92773151e-01 -1.93503395e-01 -3.23613107e-01 -6.57797694e-01 6.34986818e-01 -1.28979623e-01 7.48948872e-01 -3.79403949e-01 2.88241774e-01 6.02222025e-01 1.47956729e+00 3.42866443e-02 4.62851584e-01 3.42835337e-01 5.26028216e-01 5.28389335e-01 7.48329937e-01 3.74490380e-01 6.76790595e-01 4.10141379e-01 -1.55023277e-01 3.00842933e-02 -7.01297104e-01 -6.06915236e-01 5.04820287e-01 1.09626317e+00 3.53990465e-01 -4.95983154e-01 -1.27531111e+00 1.20771313e+00 -1.29991949e+00 -5.06190240e-01 -5.77974841e-02 1.83832979e+00 1.51321054e+00 8.42069909e-02 -2.90478528e-01 -1.59448653e-01 5.71378589e-01 2.63015151e-01 -1.53637663e-01 -1.06456673e+00 -5.33074558e-01 1.16843951e+00 2.84891933e-01 4.84918594e-01 -7.71864951e-01 1.58844423e+00 6.15725803e+00 1.06706262e+00 -7.85871208e-01 1.46862432e-01 3.43807459e-01 1.13564871e-01 -9.73834276e-01 1.34828180e-01 -8.91593397e-01 5.70535839e-01 1.04568255e+00 -5.03789544e-01 2.99088269e-01 3.35597456e-01 -1.09208234e-01 9.46468040e-02 -1.15171957e+00 1.09347689e+00 4.48829591e-01 -1.18812275e+00 4.63708401e-01 -3.57155889e-01 8.63156378e-01 8.09993967e-03 3.20375264e-02 5.93568742e-01 6.09217763e-01 -1.36748672e+00 9.85754907e-01 3.26497525e-01 1.25225627e+00 -1.20050836e+00 5.82163751e-01 2.80056037e-02 -1.17149734e+00 4.13118213e-01 -6.02804244e-01 -1.38824269e-01 3.63630176e-01 4.29040283e-01 -8.78994167e-01 4.33511466e-01 6.38312176e-02 6.82731628e-01 -5.88016510e-01 5.45193553e-01 -9.29281473e-01 9.24611986e-01 -2.32509553e-01 -5.05322516e-01 4.22213227e-01 -3.22406322e-01 4.91706789e-01 1.79935348e+00 4.59557742e-01 1.21088721e-01 2.39401430e-01 1.04954123e+00 -2.21112177e-01 5.31804740e-01 -7.28553712e-01 -4.68526334e-01 6.05795383e-01 9.47281539e-01 -5.73871315e-01 -3.71268123e-01 -3.88088562e-02 1.10293782e+00 5.03756642e-01 6.92867339e-02 -6.92507744e-01 -6.86317325e-01 7.71851361e-01 -3.68236721e-01 4.43935722e-01 -6.00844502e-01 -5.86606085e-01 -1.39421964e+00 -1.46440968e-01 -1.15080369e+00 1.95350483e-01 -6.26986802e-01 -1.53766561e+00 7.37069547e-01 2.95652747e-02 -8.73924911e-01 -3.92483890e-01 -7.48013437e-01 -5.07229388e-01 1.05282700e+00 -1.42890477e+00 -1.55973768e+00 3.84142697e-01 3.18959705e-03 9.18464899e-01 -2.93107361e-01 1.19957066e+00 -2.89454218e-02 -4.19983327e-01 8.48411798e-01 -1.24020986e-01 1.88572839e-01 5.82972467e-01 -1.59409082e+00 9.41865206e-01 1.15992129e+00 7.06830621e-01 1.03102946e+00 7.62259007e-01 -7.75983930e-01 -1.17115855e+00 -1.14966154e+00 1.38559067e+00 -6.82325900e-01 8.49285126e-01 -7.09217310e-01 -4.69093800e-01 5.47574520e-01 6.11368358e-01 -6.26438379e-01 1.05376542e+00 8.73181298e-02 -6.29437745e-01 3.77771884e-01 -8.52305651e-01 1.02793336e+00 1.22414291e+00 -7.33250678e-01 -1.12230217e+00 3.00599903e-01 8.37803304e-01 -3.27707112e-01 -1.05635321e+00 2.51965731e-01 4.33571875e-01 -3.93756747e-01 5.34293950e-01 -7.75852144e-01 4.62483704e-01 -2.00777218e-01 -2.95165211e-01 -1.82549453e+00 2.48943511e-02 -8.27196538e-01 2.23186702e-01 1.31752920e+00 9.69108939e-01 -1.39054596e-01 1.92496613e-01 1.01058803e-01 -3.01216334e-01 -5.30586720e-01 -8.90843868e-01 -1.00514269e+00 7.96469152e-01 -6.28928185e-01 7.88646877e-01 8.96724463e-01 9.07286778e-02 8.63609076e-01 -2.18389362e-01 -3.34808230e-01 1.91033900e-01 -9.41065252e-02 5.00893772e-01 -5.83551347e-01 -4.24284905e-01 -3.80529255e-01 -2.51134425e-01 -9.93882000e-01 6.56269014e-01 -1.37672317e+00 2.83440113e-01 -1.53217208e+00 -2.21345313e-02 -5.66644549e-01 -1.63511753e-01 4.33183104e-01 -4.86352295e-01 6.19663894e-01 4.82256472e-01 -4.15503114e-01 -2.92010605e-01 7.32170045e-01 1.00849211e+00 -9.98474360e-02 1.62835896e-01 -5.98188639e-01 -7.83971429e-01 4.10075575e-01 8.95339727e-01 -4.40883517e-01 -4.09770995e-01 -9.99964595e-01 5.84803343e-01 -4.94826049e-01 -3.32894802e-01 -6.49597585e-01 -3.87444764e-01 -3.53652537e-01 1.29832715e-01 -2.87150949e-01 2.80558467e-01 -1.62024632e-01 -4.69094329e-02 4.77185011e-01 -3.62440914e-01 7.06867337e-01 1.81919187e-01 -4.01714537e-03 -1.25175133e-01 -6.76479936e-01 6.28983915e-01 -3.17578822e-01 -7.10174620e-01 1.13679282e-01 -6.13837183e-01 5.82199216e-01 8.72336268e-01 -1.21914633e-01 -1.92462996e-01 -1.90389067e-01 -2.43471205e-01 2.94942521e-02 6.79221511e-01 5.87496877e-01 8.48906636e-01 -1.36685431e+00 -1.29970443e+00 2.62791604e-01 5.81612110e-01 -1.64053321e-01 -4.88289565e-01 -1.09687060e-01 -7.55866945e-01 3.31112534e-01 -1.66596502e-01 4.86459620e-02 -7.30973244e-01 3.27604204e-01 1.69197738e-01 -2.72556394e-01 -4.66853380e-01 1.13175464e+00 -3.08174524e-03 -9.34509277e-01 -2.86102235e-01 -2.02574074e-01 -6.77908137e-02 7.11021200e-02 5.20227373e-01 9.68080983e-02 -1.10641032e-01 -7.49639750e-01 -2.61377573e-01 9.40181166e-02 -2.45067328e-01 -3.84084731e-01 1.16676509e+00 -1.63735151e-02 -2.47894078e-01 4.34793770e-01 1.04501653e+00 5.31685114e-01 -4.62044448e-01 1.00899123e-01 4.33007210e-01 -3.06865513e-01 -3.78242314e-01 -9.86537397e-01 -2.94236630e-01 7.05657423e-01 5.32587059e-02 -2.20930263e-01 8.02742183e-01 6.00083079e-03 1.23651755e+00 4.33048606e-01 5.23967385e-01 -1.41732991e+00 -7.24005103e-02 1.25282061e+00 1.08554530e+00 -9.19264674e-01 -2.21669927e-01 -2.13741973e-01 -7.89773703e-01 1.25668335e+00 6.41982257e-01 -2.60238796e-01 4.37202245e-01 -2.89360397e-02 1.22216500e-01 -7.50371888e-02 -8.64817441e-01 -5.80183208e-01 2.49648839e-01 7.72327125e-01 8.84644747e-01 5.42301953e-01 -1.32362378e+00 6.32148266e-01 -9.92928863e-01 -7.31330097e-01 8.35037827e-01 5.95785618e-01 -8.54472816e-02 -1.56568968e+00 -5.61583713e-02 5.39369464e-01 -4.31955725e-01 -1.09456587e+00 -6.41098380e-01 7.03840911e-01 5.33042490e-01 1.06759882e+00 8.49004388e-02 -3.11978776e-02 2.75487393e-01 -1.44879771e-02 5.67253888e-01 -1.40717518e+00 -8.38186860e-01 -1.10421397e-01 7.01598346e-01 4.32670377e-02 -1.08747326e-01 -6.27418399e-01 -1.28276849e+00 -1.16076380e-01 -1.99823424e-01 4.47378904e-01 5.28751910e-01 7.48388886e-01 -1.79661602e-01 3.17323983e-01 4.47802454e-01 -4.23604846e-01 -5.58346391e-01 -1.23035693e+00 -6.08344197e-01 4.38137561e-01 -1.05439812e-01 -4.84876633e-01 -2.68897116e-01 -8.10515136e-02]
[10.948023796081543, 9.689929962158203]
9a48574f-579d-4c74-9e24-e49ecf98001a
multimodal-attention-fusion-for-target
2102.01326
null
https://arxiv.org/abs/2102.01326v1
https://arxiv.org/pdf/2102.01326v1.pdf
Multimodal Attention Fusion for Target Speaker Extraction
Target speaker extraction, which aims at extracting a target speaker's voice from a mixture of voices using audio, visual or locational clues, has received much interest. Recently an audio-visual target speaker extraction has been proposed that extracts target speech by using complementary audio and visual clues. Although audio-visual target speaker extraction offers a more stable performance than single modality methods for simulated data, its adaptation towards realistic situations has not been fully explored as well as evaluations on real recorded mixtures. One of the major issues to handle realistic situations is how to make the system robust to clue corruption because in real recordings both clues may not be equally reliable, e.g. visual clues may be affected by occlusions. In this work, we propose a novel attention mechanism for multi-modal fusion and its training methods that enable to effectively capture the reliability of the clues and weight the more reliable ones. Our proposals improve signal to distortion ratio (SDR) by 1.0 dB over conventional fusion mechanisms on simulated data. Moreover, we also record an audio-visual dataset of simultaneous speech with realistic visual clue corruption and show that audio-visual target speaker extraction with our proposals successfully work on real data.
['Shoko Araki', 'Tomohiro Nakatani', 'Marc Delcroix', 'Keisuke Kinoshita', 'Tsubasa Ochiai', 'Hiroshi Sato']
2021-02-02
null
null
null
null
['target-speaker-extraction']
['audio']
[ 1.58830658e-02 -1.55296922e-01 1.48826852e-01 -5.24963364e-02 -1.48885691e+00 -4.45562631e-01 5.48064113e-01 2.30428621e-01 -2.22664773e-01 7.19365060e-01 3.82390797e-01 6.31418079e-02 -1.24171667e-01 -6.21380173e-02 -3.47073197e-01 -8.78987849e-01 5.55601493e-02 3.02714646e-01 5.02621651e-01 -3.13409567e-02 2.72666156e-01 5.61025202e-01 -2.11584759e+00 4.06642616e-01 5.76110959e-01 9.57146525e-01 4.52466071e-01 1.03584778e+00 4.12616283e-02 4.52467889e-01 -1.08922029e+00 -7.74654001e-03 7.84855634e-02 -3.66802365e-01 -2.25226521e-01 1.90990478e-01 6.70243084e-01 -1.09645970e-01 1.83741469e-02 1.00657260e+00 1.18416643e+00 -1.02733850e-01 7.91631401e-01 -1.43193686e+00 -1.85166597e-01 5.28624475e-01 -7.35113919e-01 3.05637628e-01 7.87698150e-01 2.81053483e-01 7.45363235e-01 -1.02514207e+00 2.20829859e-01 1.30282426e+00 3.22192729e-01 4.78471458e-01 -1.16548884e+00 -6.23635650e-01 -1.05128452e-01 6.24297559e-01 -1.61248648e+00 -1.08862197e+00 9.08536196e-01 -2.22911835e-01 6.50940895e-01 5.85197031e-01 4.61239040e-01 1.10875642e+00 -1.40898541e-01 8.77614379e-01 9.75672007e-01 -6.67609513e-01 5.31968363e-02 7.16187537e-01 -1.87689930e-01 1.53475642e-01 -4.91090864e-02 2.20558584e-01 -7.89563835e-01 -2.47547105e-01 1.52314544e-01 -4.64069545e-01 -7.99722433e-01 -1.53766334e-01 -1.26954067e+00 4.39228147e-01 4.90859374e-02 6.96608901e-01 -3.94095629e-01 -5.83915636e-02 1.68951914e-01 2.25792959e-01 2.07057327e-01 1.54594898e-01 4.07159030e-02 2.81733423e-02 -1.25151539e+00 7.80513734e-02 4.94460940e-01 8.01451385e-01 2.28787571e-01 4.95924950e-01 -3.71321172e-01 1.07463229e+00 6.84686005e-01 8.76065433e-01 4.37410295e-01 -7.68148005e-01 3.39895755e-01 3.35432589e-02 1.80622369e-01 -8.92209470e-01 -2.67744511e-01 -5.07302284e-01 -6.61827385e-01 5.87606847e-01 3.87518436e-01 1.60599574e-01 -4.99007016e-01 1.67970383e+00 3.04129571e-01 2.29765475e-01 1.57147050e-01 1.07334483e+00 1.02270222e+00 6.45494163e-01 -2.93610632e-01 -6.59802914e-01 1.50997376e+00 -4.97685581e-01 -1.10933030e+00 -3.55069409e-03 -1.74601406e-01 -1.30702651e+00 7.22226202e-01 7.88070679e-01 -1.27577186e+00 -8.16562831e-01 -1.16248310e+00 4.11757022e-01 4.20605019e-02 3.15282732e-01 -8.98447931e-02 9.53879774e-01 -1.18130064e+00 -7.37412125e-02 -3.57230127e-01 -2.57448405e-01 -1.16658822e-01 3.31928343e-01 -4.12219524e-01 4.94030118e-02 -1.15409470e+00 8.66210699e-01 9.26471129e-02 1.43508330e-01 -1.09693635e+00 -2.42547736e-01 -6.80377901e-01 1.10118896e-01 4.47742939e-01 -4.84318763e-01 1.22788429e+00 -8.55776370e-01 -1.53084242e+00 5.69068611e-01 -4.40167248e-01 -4.02566195e-01 4.35281068e-01 1.18414368e-02 -9.07433808e-01 2.93084800e-01 -1.29013419e-01 5.81922233e-01 1.46990740e+00 -1.68886602e+00 -5.81474066e-01 -2.00704366e-01 -6.07735038e-01 1.67591467e-01 -2.57767230e-01 2.82594740e-01 -3.34329456e-01 -6.01069808e-01 4.09110561e-02 -4.72841740e-01 3.79067570e-01 -1.37347355e-01 -3.71117353e-01 -1.66236460e-01 8.96120846e-01 -6.54634893e-01 1.09684741e+00 -2.29404521e+00 -5.17915078e-02 1.70181885e-01 2.78659731e-01 5.79721630e-01 -2.26425663e-01 3.42685014e-01 -2.03069355e-02 -2.88842171e-01 1.61865294e-01 -6.25884056e-01 -3.58211137e-02 -2.46979281e-01 -2.27300599e-01 6.12763584e-01 3.73342298e-02 2.71137476e-01 -4.63240206e-01 -7.86730886e-01 3.01912040e-01 9.16628301e-01 -3.34628373e-01 2.93935239e-01 2.24857226e-01 2.88371056e-01 1.24011368e-01 8.41932595e-01 7.12028384e-01 3.50395650e-01 -9.55044106e-02 -3.95730764e-01 -1.21702373e-01 5.67889437e-02 -1.58577418e+00 1.21482944e+00 -3.60781908e-01 1.04784596e+00 5.78777254e-01 -5.82171977e-01 1.16630769e+00 8.79709780e-01 1.88497588e-01 -6.20459139e-01 1.73833072e-02 4.44972128e-01 2.43030354e-01 -5.74849129e-01 5.54334044e-01 -1.72959179e-01 2.59337574e-01 1.45967320e-01 6.13892637e-02 -2.22920299e-01 -8.44194368e-02 2.13193983e-01 6.88932419e-01 -3.67554963e-01 2.61282831e-01 1.20403491e-01 9.04695690e-01 -5.15587747e-01 3.86365473e-01 5.58164597e-01 -3.71506304e-01 8.41324806e-01 2.61350840e-01 3.76577020e-01 -7.04136670e-01 -1.17302060e+00 5.27207851e-02 8.55717123e-01 8.94072428e-02 -3.71642470e-01 -5.34724593e-01 -2.23460183e-01 -2.06258640e-01 8.00507545e-01 -1.81119144e-01 2.52012536e-03 -2.50477225e-01 -3.73430133e-01 6.29327416e-01 2.37884849e-01 9.32378098e-02 -9.50802088e-01 -4.23447609e-01 2.78231084e-01 -5.05903661e-01 -1.12718582e+00 -4.81275737e-01 9.40416846e-03 -2.86411047e-01 -9.23314869e-01 -1.02949846e+00 -5.39392591e-01 2.37588793e-01 5.10939717e-01 8.16637635e-01 -3.08308363e-01 -2.37266451e-01 6.87566757e-01 -1.87851906e-01 -5.51197171e-01 -8.80572319e-01 -4.46825087e-01 2.65354842e-01 5.06902039e-01 1.05159692e-01 -5.74049056e-01 -4.28120822e-01 5.61422348e-01 -6.49405181e-01 -2.50078231e-01 5.50317526e-01 5.88347733e-01 2.83073545e-01 7.62115568e-02 8.29023421e-01 2.75397263e-02 7.28926301e-01 -1.70414031e-01 -4.33325887e-01 4.01606075e-02 -2.60209233e-01 -1.62309319e-01 3.08041871e-01 -7.92315125e-01 -1.01731884e+00 -2.03794893e-02 -1.81942761e-01 -6.80060983e-01 -4.41392422e-01 7.37347156e-02 -5.13530970e-01 1.37485012e-01 8.13878894e-01 2.80692846e-01 -5.97156957e-02 -5.20886004e-01 1.22998416e-01 1.18256414e+00 6.04800045e-01 -2.60073155e-01 4.81958091e-01 3.20098549e-01 -1.28633574e-01 -1.24473333e+00 -2.90535688e-02 -8.13461423e-01 -2.76866138e-01 -4.65408206e-01 5.78812718e-01 -9.30331290e-01 -1.00720346e+00 3.67517501e-01 -1.37280679e+00 2.56987572e-01 -1.33103542e-02 8.92315924e-01 -5.01020908e-01 6.04137123e-01 -9.11756232e-02 -1.41371036e+00 -1.29702672e-01 -1.46620190e+00 1.14102352e+00 -2.90347785e-02 -2.66666114e-01 -4.56512719e-01 -1.80392768e-02 4.54932094e-01 5.75593889e-01 -2.49523774e-01 4.25756007e-01 -7.32023120e-01 -4.48548764e-01 -1.49552286e-01 -1.87862776e-02 4.03946310e-01 2.68701583e-01 1.79876357e-01 -1.55129886e+00 -3.11215907e-01 2.42754012e-01 -5.50043806e-02 8.49401474e-01 5.26491582e-01 4.32060450e-01 -2.32111752e-01 -3.44177693e-01 4.24681120e-02 9.70308602e-01 4.12992209e-01 5.54375768e-01 -2.87461072e-01 3.42801660e-01 8.30033362e-01 4.96549100e-01 3.26239496e-01 4.80644666e-02 1.16701829e+00 5.62621534e-01 -9.44730788e-02 -6.80546820e-01 1.15358196e-01 6.08536839e-01 8.66739333e-01 8.04217458e-02 -6.07314527e-01 -7.16128290e-01 6.85008228e-01 -1.48642385e+00 -1.13406384e+00 -2.73902148e-01 2.47205400e+00 5.88050485e-01 1.97742522e-01 3.77985716e-01 7.89219439e-01 1.05956507e+00 3.40745077e-02 -2.41268545e-01 -1.04385965e-01 -4.76602972e-01 -1.16082035e-01 -3.55647579e-02 8.01841795e-01 -7.08939314e-01 2.07530484e-01 5.74511719e+00 1.06623411e+00 -1.46893799e+00 2.95027375e-01 1.80145279e-01 -3.96666437e-01 -3.05568725e-01 -3.56972218e-01 -7.67519534e-01 4.10521895e-01 8.93259406e-01 1.80465113e-02 1.39076104e-02 3.64338905e-01 4.35979217e-01 -4.04635638e-01 -1.03538764e+00 1.41934454e+00 5.69787204e-01 -7.48387396e-01 -1.00501344e-01 1.39366224e-01 1.01111151e-01 -3.16362709e-01 2.14060441e-01 -2.28380505e-02 -2.70214140e-01 -9.41919625e-01 8.34693313e-01 4.63581860e-01 6.89782798e-01 -7.61615872e-01 5.77053964e-01 3.41077268e-01 -1.22475600e+00 -1.99838076e-02 -3.33665982e-02 4.63739187e-01 2.05009431e-01 5.47465444e-01 -1.33338678e+00 5.32632887e-01 4.66329128e-01 9.12665352e-02 -5.48456848e-01 1.41617167e+00 -1.38474151e-01 5.75320601e-01 -5.11212826e-01 -6.31612213e-03 -2.86071897e-01 5.13353765e-01 1.16039610e+00 1.29712152e+00 6.09876156e-01 -4.01803493e-01 -7.76000097e-02 5.59015274e-01 4.56189483e-01 3.83486062e-01 -7.19769776e-01 2.78763086e-01 5.58499277e-01 1.15912104e+00 -4.98867601e-01 -1.33967906e-01 -2.29307517e-01 6.02159321e-01 -3.30433220e-01 3.60138714e-01 -8.11393321e-01 -2.75379837e-01 4.46552694e-01 2.74968922e-01 2.64473349e-01 1.79139141e-03 8.50946605e-02 -8.21903348e-01 3.72518823e-02 -1.00793481e+00 1.95990309e-01 -1.15127599e+00 -9.20358062e-01 9.42594349e-01 -2.31670886e-02 -1.69976509e+00 -4.78309095e-01 -2.48046890e-01 -5.59213698e-01 1.00927138e+00 -1.23052084e+00 -1.03307223e+00 -1.01839058e-01 8.04211795e-01 7.27804124e-01 -4.78266686e-01 6.16238892e-01 4.86726701e-01 -2.38672674e-01 7.50683606e-01 -1.24110706e-01 -3.73843908e-01 1.13152790e+00 -1.02558565e+00 -2.50391126e-01 8.04438055e-01 4.16543245e-01 2.06054851e-01 1.13189769e+00 -3.20452034e-01 -1.09245872e+00 -4.12021369e-01 9.65449750e-01 -9.49801952e-02 3.55523527e-01 -3.94095272e-01 -1.03531384e+00 1.02205969e-01 5.48368454e-01 -1.40425503e-01 7.29580700e-01 -2.84638166e-01 -4.08162594e-01 -4.36962634e-01 -1.04104984e+00 4.23171401e-01 3.87356877e-01 -5.10616720e-01 -6.44072950e-01 -4.97617535e-02 3.92116159e-01 1.91777330e-02 -3.70519549e-01 3.51860642e-01 5.44913232e-01 -1.28540373e+00 1.05566335e+00 7.77332038e-02 -3.22048128e-01 -7.30699122e-01 -3.79797846e-01 -1.43675315e+00 6.21542670e-02 -7.67154574e-01 -8.71065855e-02 1.65229213e+00 4.47921038e-01 -5.38068533e-01 1.12970516e-01 -1.39056966e-01 2.15933155e-02 -1.16652079e-01 -1.25273395e+00 -6.70276582e-01 -6.49544179e-01 -7.77214170e-01 3.12699527e-01 6.76324964e-01 4.63419128e-03 6.12825155e-01 -7.78655767e-01 3.01142216e-01 6.88760698e-01 -1.11229770e-01 8.33297908e-01 -1.34063387e+00 -5.26806772e-01 -6.03316128e-01 -5.66719830e-01 -7.40935624e-01 -4.46139723e-02 -4.94676411e-01 9.91098434e-02 -1.50397551e+00 -2.44157374e-01 6.06942875e-03 -2.70501375e-01 1.31747201e-01 1.62651986e-02 3.15136373e-01 2.83622116e-01 1.21532872e-01 -3.97217423e-01 5.82745910e-01 1.01298428e+00 -4.12076026e-01 -3.44090492e-01 4.23333079e-01 -6.26401782e-01 6.49733067e-01 4.49783325e-01 -4.50367242e-01 -5.10430932e-01 7.46134371e-02 -1.62738245e-02 6.95721745e-01 3.82585973e-01 -1.15724754e+00 4.54639137e-01 2.38230243e-01 2.77164489e-01 -8.83612037e-01 1.01054466e+00 -9.55025852e-01 1.58267170e-01 2.13416174e-01 -1.73905030e-01 -2.16846451e-01 4.40664262e-01 6.24749005e-01 -5.56207240e-01 -1.88650474e-01 8.18153203e-01 1.69664040e-01 -2.44815752e-01 -2.81247348e-01 -6.40188694e-01 -3.75629783e-01 9.37938154e-01 -3.56962711e-01 -1.92472249e-01 -8.56214702e-01 -9.81734216e-01 -5.10583669e-02 -3.73999029e-02 3.88489664e-01 8.86643946e-01 -1.33643365e+00 -1.15009618e+00 2.60138601e-01 1.21843062e-01 -5.02280533e-01 4.17159706e-01 1.12230146e+00 5.93143841e-03 3.19199175e-01 -1.16006106e-01 -1.01448977e+00 -2.06018710e+00 7.53055573e-01 3.53715450e-01 2.67254025e-01 -2.72892088e-01 7.18727410e-01 2.46170804e-01 2.89564967e-01 7.57144928e-01 -2.02512339e-01 -5.64341426e-01 4.54606324e-01 7.97452807e-01 5.12134194e-01 1.92230612e-01 -1.03802931e+00 -5.89373410e-01 6.33400500e-01 1.68863311e-01 -5.79777718e-01 9.28532362e-01 -4.41028506e-01 2.67390698e-01 7.15365589e-01 9.70682204e-01 7.82846272e-01 -8.06158841e-01 -3.55175942e-01 -2.99042851e-01 -5.84282339e-01 1.54299110e-01 -7.30411649e-01 -9.55405653e-01 1.25274456e+00 8.75687480e-01 6.43694222e-01 1.35377944e+00 1.35479927e-01 2.48463139e-01 9.13927555e-02 7.54644871e-02 -6.68046951e-01 3.58024150e-01 -6.83492720e-02 1.31886709e+00 -1.17455339e+00 -1.82797477e-01 -4.33436841e-01 -6.74425602e-01 1.14510322e+00 3.28460872e-01 4.93110269e-01 5.62041104e-01 4.96180326e-01 3.55053574e-01 4.81741503e-02 -8.33481908e-01 -5.72636724e-01 6.88849211e-01 9.24111187e-01 2.93284595e-01 -1.14111111e-01 1.25559032e-01 4.50780869e-01 3.08576822e-02 -4.62320834e-01 4.50210899e-01 3.50007385e-01 -6.65831983e-01 -8.57881606e-01 -1.28316152e+00 -1.15809210e-01 -5.34234166e-01 8.67040008e-02 -4.26597327e-01 5.39008737e-01 -1.50607899e-02 1.55836034e+00 -1.01503558e-01 -2.98850507e-01 4.66391653e-01 2.02663213e-01 5.87374866e-01 -3.12223375e-01 -5.78059614e-01 1.00930941e+00 -9.03217369e-05 -9.98243243e-02 -5.71956635e-01 -6.17404342e-01 -8.55390668e-01 -5.72666898e-02 -5.63346207e-01 2.81552792e-01 8.39709640e-01 6.93358302e-01 1.67721450e-01 7.91205227e-01 5.71669638e-01 -9.93206739e-01 -2.58203149e-01 -1.26064742e+00 -6.69273615e-01 2.21591532e-01 7.47110546e-01 -7.28205919e-01 -6.80699646e-01 -9.06747133e-02]
[14.494242668151855, 5.295816421508789]
4d9fe829-3cfb-48f8-a79b-58d9198e3f60
classification-of-household-materials-via
1805.04051
null
http://arxiv.org/abs/1805.04051v3
http://arxiv.org/pdf/1805.04051v3.pdf
Classification of Household Materials via Spectroscopy
Recognizing an object's material can inform a robot on the object's fragility or appropriate use. To estimate an object's material during manipulation, many prior works have explored the use of haptic sensing. In this paper, we explore a technique for robots to estimate the materials of objects using spectroscopy. We demonstrate that spectrometers provide several benefits for material recognition, including fast response times and accurate measurements with low noise. Furthermore, spectrometers do not require direct contact with an object. To explore this, we collected a dataset of spectral measurements from two commercially available spectrometers during which a robotic platform interacted with 50 flat material objects, and we show that a neural network model can accurately analyze these measurements. Due to the similarity between consecutive spectral measurements, our model achieved a material classification accuracy of 94.6% when given only one spectral sample per object. Similar to prior works with haptic sensors, we found that generalizing material recognition to new objects posed a greater challenge, for which we achieved an accuracy of 79.1% via leave-one-object-out cross-validation. Finally, we demonstrate how a PR2 robot can leverage spectrometers to estimate the materials of everyday objects found in the home. From this work, we find that spectroscopy poses a promising approach for material classification during robotic manipulation.
['Sonia Chernova', 'Nathan Luskey', 'Zackory Erickson', 'Charles C. Kemp']
2018-05-10
null
null
null
null
['material-classification', 'material-recognition']
['computer-vision', 'computer-vision']
[ 5.09700298e-01 -3.55413742e-02 -1.82792749e-02 -2.52389044e-01 -5.98969162e-01 -5.01895905e-01 -1.71024472e-01 3.07985634e-01 -2.12723643e-01 5.26856422e-01 -4.44671363e-01 3.82981971e-02 -1.33939907e-01 -9.44673121e-01 -1.02661347e+00 -3.17904115e-01 -2.59283651e-02 4.29200143e-01 2.18576252e-01 -5.28523251e-02 4.29945856e-01 9.43521738e-01 -1.88860929e+00 4.40482259e-01 5.45928538e-01 1.61633813e+00 4.50550914e-01 3.67983729e-01 9.81005058e-02 5.02470553e-01 -5.91511369e-01 3.14016134e-01 4.81462091e-01 3.22212666e-01 -7.50424325e-01 -1.71693712e-01 2.74379343e-01 -8.20324719e-01 -1.26613081e-01 7.72535384e-01 3.48480076e-01 1.24982677e-01 1.07089043e+00 -1.23295951e+00 -6.57734811e-01 6.12542868e-01 -3.85593742e-01 -6.08764887e-01 7.58247733e-01 1.18334107e-01 6.22592032e-01 -8.23699534e-01 4.53487396e-01 1.07893634e+00 9.85818386e-01 5.67942023e-01 -1.40687788e+00 -7.33712912e-01 -1.92398012e-01 5.53511418e-02 -1.18145394e+00 -5.24147630e-01 7.77401626e-01 -8.82886052e-01 9.87376690e-01 2.52901018e-01 8.36960435e-01 9.10757840e-01 3.97477388e-01 4.81116712e-01 1.20293915e+00 -3.08628976e-01 1.91776827e-01 3.82526845e-01 -1.32803947e-01 6.78819716e-01 3.34759653e-01 6.07868880e-02 -5.66465437e-01 -2.66282171e-01 6.80607677e-01 3.84601615e-02 -1.97862208e-01 -5.89253664e-01 -1.37122560e+00 2.34983832e-01 5.37180066e-01 -2.15185806e-01 -3.47445816e-01 2.18187943e-01 2.44179696e-01 3.09358031e-01 2.37465873e-01 1.15984428e+00 -4.95772898e-01 -7.33649433e-02 -1.59191564e-01 1.68349177e-01 1.15498340e+00 1.11568868e+00 7.13522315e-01 -1.90930232e-01 2.46867523e-01 9.81084704e-01 3.67140204e-01 8.54870796e-01 2.07115598e-02 -1.02045619e+00 2.39545956e-01 2.90875107e-01 7.62719393e-01 -8.68525624e-01 -7.27607310e-01 2.38097832e-01 -1.27881065e-01 5.81972659e-01 4.12973821e-01 -7.40237087e-02 -6.11879468e-01 1.34213471e+00 4.18202817e-01 -6.20710373e-01 -6.39701169e-03 1.18937504e+00 7.05498040e-01 3.42814326e-01 -9.38557163e-02 4.87989895e-02 1.02481699e+00 -6.30080163e-01 -2.90741503e-01 -7.63560310e-02 3.60133469e-01 -9.92834210e-01 1.23862457e+00 8.77395153e-01 -9.05761480e-01 -5.68307757e-01 -1.54860234e+00 -3.57921310e-02 -5.50617456e-01 2.63693184e-01 9.25259173e-01 5.88245571e-01 -6.08902454e-01 1.22110474e+00 -8.48425567e-01 -5.11034071e-01 7.23502189e-02 6.11929595e-01 -1.66469201e-01 1.75785363e-01 -6.61646366e-01 9.83331800e-01 3.19925308e-01 1.37803210e-02 -4.88700569e-01 -8.61708879e-01 -4.91442621e-01 -4.23078746e-01 2.84557015e-01 -2.94912010e-01 1.46014130e+00 -3.09400767e-01 -2.01135612e+00 7.90105581e-01 4.66232598e-01 -1.03434369e-01 3.92660528e-01 -3.69247466e-01 -3.50247055e-01 4.38563675e-01 5.36800176e-02 7.64674604e-01 8.11738551e-01 -1.24187005e+00 -1.44825131e-02 -1.82237878e-01 5.16816266e-02 -5.24826720e-02 -4.46678787e-01 -1.82381928e-01 3.69850934e-01 -1.99492976e-01 6.67740285e-01 -1.11074829e+00 3.13289851e-01 7.77422488e-01 -5.19805670e-01 -5.45188319e-03 6.59419596e-01 -5.35573721e-01 1.54991671e-01 -2.04456496e+00 -1.91373512e-01 4.61338758e-01 9.28700119e-02 -3.61510456e-01 -3.47163454e-02 7.52148449e-01 1.38551712e-01 -1.19034238e-01 2.32658729e-01 1.89040646e-01 1.17524721e-01 -5.41379154e-01 -2.04394042e-01 5.12137055e-01 3.60255122e-01 3.81699026e-01 -8.16458285e-01 -5.66478558e-02 1.74035832e-01 2.98939377e-01 -4.07307297e-01 8.12444761e-02 -1.81097984e-01 2.74431825e-01 -4.05789167e-01 1.16345489e+00 7.01388121e-01 -2.09734216e-01 1.74830928e-01 -8.69989753e-01 -1.31703377e-01 4.58318889e-01 -1.16577220e+00 1.68226266e+00 -3.86496276e-01 4.51262802e-01 5.31025052e-01 -7.57057428e-01 1.23708177e+00 -1.24916807e-01 1.06212473e+00 -6.25901103e-01 1.33916810e-01 6.45940304e-01 -9.65641141e-02 -8.11023831e-01 6.23967707e-01 -1.47686273e-01 1.32919252e-01 5.15292108e-01 -4.43704903e-01 -9.52962756e-01 -1.95672706e-01 -3.44874769e-01 1.13427472e+00 5.44029832e-01 -3.30309987e-01 -3.83183479e-01 -2.50141859e-01 2.11966157e-01 6.59772486e-04 7.31272459e-01 -1.02912508e-01 4.23442006e-01 -3.33244145e-01 -8.32680687e-02 -1.02518535e+00 -1.43057299e+00 -3.41033340e-01 9.50451314e-01 5.93471348e-01 -4.24379744e-02 -3.86495113e-01 9.14712399e-02 8.61175060e-01 2.82527715e-01 -2.49009252e-01 -3.15612167e-01 -1.03279725e-01 -1.34334102e-01 1.18908115e-01 6.88897133e-01 2.75002211e-01 -9.24498677e-01 -9.08276737e-01 3.72524917e-01 -7.35707805e-02 -1.12978196e+00 6.61281589e-03 2.89516866e-01 -8.68281662e-01 -1.24741423e+00 -2.90108621e-01 -7.27108896e-01 5.03728747e-01 5.62739491e-01 5.77085555e-01 -1.58696637e-01 -7.08359599e-01 9.03644919e-01 -2.79286474e-01 -9.10831392e-01 -4.94229615e-01 3.23495716e-02 5.77760279e-01 -5.51907480e-01 1.93008676e-01 -6.60351932e-01 -6.46812081e-01 5.60972035e-01 -2.15980411e-01 1.01222165e-01 6.10012233e-01 3.47971201e-01 4.75677729e-01 -2.59902887e-02 5.03177345e-01 -3.34029287e-01 6.43177748e-01 -3.72273505e-01 -4.31402922e-01 8.96649063e-02 -5.49227297e-01 -5.22365831e-02 3.24659169e-01 -7.93029249e-01 -7.57838845e-01 2.59194165e-01 3.45111459e-01 -1.94227025e-01 8.87976959e-02 5.39191723e-01 2.86823004e-01 -5.30672967e-01 9.44930434e-01 -4.69981432e-01 4.97652411e-01 -4.61683840e-01 -1.75803989e-01 1.25912511e+00 4.18621629e-01 -1.20711720e+00 4.71916586e-01 3.60876769e-01 1.02282353e-01 -8.93748581e-01 -4.40939128e-01 -4.98112440e-01 -5.71518302e-01 -5.45327246e-01 5.22591174e-01 -9.27785456e-01 -1.63682377e+00 6.50271595e-01 -8.31759572e-01 -6.72009706e-01 -1.21898502e-01 9.14843976e-01 -8.42216372e-01 1.42368063e-01 -8.46998513e-01 -1.26155710e+00 -3.81375611e-01 -1.15563095e+00 1.32827556e+00 -1.48046494e-01 -6.24896407e-01 -2.42053032e-01 -7.49090314e-01 5.73782682e-01 4.09303844e-01 3.83935243e-01 6.71128750e-01 6.62493929e-02 -5.12944818e-01 -7.00716004e-02 -4.99187678e-01 -3.27237137e-03 6.86373115e-01 -2.55204886e-02 -1.19264257e+00 -4.41867530e-01 2.35314132e-03 -7.80955732e-01 5.53307235e-01 -5.16434982e-02 1.16349328e+00 9.26142633e-02 -4.03898805e-01 1.20862074e-01 1.01458025e+00 4.33836073e-01 3.78919870e-01 3.51146042e-01 6.11002326e-01 8.12714994e-01 1.10163033e+00 3.56478810e-01 -2.35452220e-01 6.50885820e-01 3.23328495e-01 2.91020900e-01 -3.72980721e-02 -5.05266562e-02 4.45910394e-01 4.35143977e-01 -1.62041023e-01 1.64655939e-01 -7.24180877e-01 4.58675949e-03 -1.35692120e+00 -5.82872808e-01 -6.30107149e-02 2.18460584e+00 8.63080323e-01 2.55344480e-01 2.19499230e-01 2.49759704e-01 5.44977367e-01 -5.98986149e-01 -1.27814603e+00 -1.35497227e-01 4.10296559e-01 5.98589852e-02 8.29334199e-01 9.90099609e-02 -9.61712539e-01 3.71911705e-01 7.16690636e+00 1.50428072e-01 -1.28467894e+00 -4.10441190e-01 -1.43431248e-02 -2.33439818e-01 9.69843864e-02 -5.34887791e-01 -3.66311580e-01 3.24208200e-01 6.63033307e-01 3.65838319e-01 9.08020854e-01 1.01579082e+00 2.05203164e-02 -6.02568209e-01 -1.48457909e+00 1.01416814e+00 -1.29729182e-01 -9.24864113e-01 -4.32098925e-01 -1.64984003e-01 2.22951591e-01 3.59938107e-02 4.40036803e-02 -1.15878873e-01 -1.40875727e-01 -6.13237321e-01 1.29040670e+00 7.79563725e-01 1.02393484e+00 -1.51630670e-01 9.87311304e-02 1.07127763e-01 -1.09921455e+00 -2.54947394e-01 -3.61937732e-01 -2.87896365e-01 -2.01596439e-01 7.50077665e-01 -1.50250947e+00 1.04831256e-01 1.04698896e+00 3.53538811e-01 -1.24923050e-01 8.49335909e-01 3.30074519e-01 1.75843105e-01 -7.51402736e-01 -5.21879256e-01 -6.18444741e-01 -1.80946931e-01 3.11117947e-01 7.72525251e-01 5.09659410e-01 -1.47201747e-01 5.13650000e-01 1.18742812e+00 6.72156364e-02 4.82264347e-02 -6.45719826e-01 -4.70424443e-01 5.76949120e-01 9.94783819e-01 -8.48377466e-01 -3.22616696e-02 -1.63446039e-01 9.04925406e-01 1.99113458e-01 2.45046109e-01 -4.68033671e-01 -5.78910768e-01 4.73007828e-01 3.18528980e-01 -6.19107671e-02 -7.96716750e-01 -4.06318247e-01 -6.86138570e-01 5.39806724e-01 -5.27703762e-01 -4.01547790e-01 -1.49211347e+00 -1.54761696e+00 -2.81038284e-01 -2.24557221e-02 -1.39040887e+00 2.20586032e-01 -1.28299093e+00 3.04693162e-01 7.92023897e-01 -1.04020119e+00 -1.03451681e+00 -6.36613965e-01 -2.82756798e-02 1.39618501e-01 6.08103015e-02 1.00925314e+00 6.72713965e-02 8.57897624e-02 1.32725000e-01 1.58915937e-01 -2.02512458e-01 9.67406869e-01 -8.99575055e-01 2.09505156e-01 -1.75418317e-01 -5.10609329e-01 8.07374895e-01 7.59700418e-01 -1.17555785e+00 -2.23981452e+00 -6.08862519e-01 -2.22870246e-01 -2.97906399e-01 8.42747927e-01 -5.10608912e-01 -7.41054714e-01 3.46999824e-01 -4.70483869e-01 -3.37822765e-01 5.93012691e-01 3.26189876e-01 -3.34021986e-01 -7.31124207e-02 -1.63628805e+00 4.12351847e-01 1.44906199e+00 -9.41672504e-01 -1.87910616e-01 7.30980814e-01 6.27086222e-01 -7.83732653e-01 -1.40043843e+00 5.47152579e-01 1.42337704e+00 -7.12434232e-01 1.00983655e+00 -1.05966553e-01 2.01609522e-01 -1.58358514e-01 -2.84490794e-01 -9.90658820e-01 -3.05194199e-01 -2.40679696e-01 -1.10528246e-01 8.12775493e-01 3.77421945e-01 -9.49564040e-01 9.16476190e-01 1.11070824e+00 -4.00591940e-01 -5.78358769e-01 -3.32695186e-01 -1.23959386e+00 -3.09254646e-01 -3.72497857e-01 5.99878550e-01 6.92655504e-01 4.05714393e-01 -1.26016647e-01 -4.31780629e-02 3.22306752e-01 5.09888172e-01 2.17715368e-01 5.90755522e-01 -1.67379630e+00 -3.98485035e-01 -8.98793638e-02 -3.75922561e-01 -7.06356108e-01 1.27386987e-01 -8.69756222e-01 4.57929075e-01 -1.46268976e+00 2.12762028e-01 -9.33794439e-01 -2.03357860e-01 4.13272113e-01 3.91225517e-01 2.63768256e-01 4.92464229e-02 3.57295245e-01 -8.43580663e-02 2.44604498e-01 1.18794930e+00 -6.56038225e-01 -4.68999207e-01 -1.29052281e-01 -3.99220735e-01 4.87169236e-01 9.78066266e-01 -6.09536283e-02 -3.08917493e-01 -3.27515185e-01 4.20971066e-01 -1.43557742e-01 4.94480580e-01 -1.48505521e+00 6.94497302e-02 -3.66478443e-01 6.77313983e-01 -4.58887458e-01 9.66606021e-01 -1.14172482e+00 3.78867745e-01 6.21676862e-01 -1.54974669e-01 -4.78215486e-01 5.03216803e-01 3.91364753e-01 5.57528317e-01 -3.62588614e-01 4.79724735e-01 -4.39054593e-02 -5.37778556e-01 -2.56281108e-01 -4.35914308e-01 -7.13517487e-01 8.61084461e-01 -3.79812956e-01 -4.30291444e-01 -2.84166709e-02 -6.84169352e-01 -1.45320311e-01 7.80144632e-01 2.94887125e-01 6.76724970e-01 -1.31989372e+00 4.37989474e-05 2.70040214e-01 2.49357909e-01 -2.36759484e-01 -1.06108440e-02 6.92196131e-01 -5.93215883e-01 -1.12420328e-01 -5.27760863e-01 -7.37565875e-01 -9.79956448e-01 5.28918087e-01 2.65814722e-01 8.12178433e-01 -4.14170712e-01 6.17756546e-01 -4.50157404e-01 -6.36314631e-01 1.53847262e-01 -9.22477543e-01 4.61367428e-01 -5.26585989e-02 1.14968799e-01 6.37801468e-01 3.04177195e-01 -9.94565058e-03 -2.28928566e-01 7.53060937e-01 9.76250395e-02 -1.44168466e-01 1.32510519e+00 1.45353496e-01 5.79369813e-02 1.06800199e+00 1.21210957e+00 1.28384858e-01 -1.30941486e+00 1.02224275e-01 -2.92205602e-01 -3.00601125e-01 -2.49982327e-01 -9.73004103e-01 -5.65853417e-01 4.51717585e-01 7.16521204e-01 4.48071688e-01 7.77862191e-01 -3.26735415e-02 8.02230120e-01 1.29689252e+00 9.63978350e-01 -1.63040006e+00 3.53775144e-01 1.91250071e-01 1.27572560e+00 -1.39392793e+00 2.85500675e-01 -1.10043609e+00 -2.65957750e-02 1.39214897e+00 5.84802032e-01 2.54026115e-01 5.70784688e-01 4.04477000e-01 -1.56314537e-01 -3.11801821e-01 -1.89663023e-02 2.10456967e-01 1.91912651e-01 8.53281260e-01 3.61320734e-01 3.75692695e-01 4.00578499e-01 1.94072902e-01 -6.05991483e-01 1.84585690e-01 3.44892442e-01 1.48429346e+00 -8.41897130e-01 -7.30323255e-01 -5.70698977e-01 9.12396967e-01 1.19722351e-01 4.00630623e-01 -5.14826059e-01 4.59583223e-01 -1.93604708e-01 1.12286937e+00 -1.94737434e-01 -7.88688481e-01 5.93503416e-01 4.45263050e-02 1.13655412e+00 -7.42184579e-01 -1.25541344e-01 -3.80586058e-01 4.60434616e-01 -8.13552022e-01 -5.56978047e-01 -7.80197501e-01 -1.44158041e+00 -6.24823608e-02 -5.33268571e-01 -1.76349729e-01 1.34820187e+00 5.21648109e-01 3.39592040e-01 4.31876183e-01 5.40996611e-01 -1.39335477e+00 -7.07045674e-01 -9.72515523e-01 -1.01528442e+00 3.68452579e-01 1.39275551e-01 -1.32770348e+00 -1.92558259e-01 8.57069995e-03]
[5.835573673248291, -0.8115377426147461]
7cd6fb67-7ac6-4546-a0b5-87dd33829616
twice-mixing-a-rank-learning-based-quality
2102.00670
null
https://arxiv.org/abs/2102.00670v1
https://arxiv.org/pdf/2102.00670v1.pdf
Twice Mixing: A Rank Learning based Quality Assessment Approach for Underwater Image Enhancement
To improve the quality of underwater images, various kinds of underwater image enhancement (UIE) operators have been proposed during the past few years. However, the lack of effective objective evaluation methods limits the further development of UIE techniques. In this paper, we propose a novel rank learning guided no-reference quality assessment method for UIE. Our approach, termed Twice Mixing, is motivated by the observation that a mid-quality image can be generated by mixing a high-quality image with its low-quality version. Typical mixup algorithms linearly interpolate a given pair of input data. However, the human visual system is non-uniformity and non-linear in processing images. Therefore, instead of directly training a deep neural network based on the mixed images and their absolute scores calculated by linear combinations, we propose to train a Siamese Network to learn their quality rankings. Twice Mixing is trained based on an elaborately formulated self-supervision mechanism. Specifically, before each iteration, we randomly generate two mixing ratios which will be employed for both generating virtual images and guiding the network training. In the test phase, a single branch of the network is extracted to predict the quality rankings of different UIE outputs. We conduct extensive experiments on both synthetic and real-world datasets. Experimental results demonstrate that our approach outperforms the previous methods significantly.
['Xinghao Ding', 'Yue Huang', 'Xueyang Fu', 'Zhenqi Fu']
2021-02-01
null
null
null
null
['uie']
['computer-vision']
[ 3.86076927e-01 -1.05757855e-01 4.49408829e-01 -5.45760274e-01 -8.26919854e-01 -2.07037315e-01 3.99661154e-01 -3.96936871e-02 -6.07503176e-01 7.27736235e-01 3.14027481e-02 1.31807998e-01 -2.18217522e-01 -9.50534225e-01 -7.58219898e-01 -8.78456712e-01 -1.10778064e-01 1.21380672e-01 1.54769093e-01 -3.83243829e-01 3.14572841e-01 1.68732062e-01 -1.69661331e+00 1.89291418e-01 1.49014783e+00 1.15921688e+00 4.02052283e-01 7.00547814e-01 1.59287304e-02 6.02812469e-01 -7.74204671e-01 -5.66878319e-01 6.48507297e-01 -8.87909710e-01 -3.73515368e-01 2.90018231e-01 5.33083022e-01 -4.42380726e-01 -2.32383773e-01 1.45210838e+00 6.43876791e-01 2.79651642e-01 4.92634445e-01 -1.18576467e+00 -6.63180411e-01 5.78952909e-01 -4.62037176e-01 1.64347906e-02 2.13283375e-01 1.83176875e-01 1.04169476e+00 -1.06286299e+00 2.96839058e-01 1.09716189e+00 4.93607551e-01 5.24024963e-01 -1.00895298e+00 -6.98395491e-01 -1.47336766e-01 2.31485784e-01 -1.07172871e+00 -3.91484857e-01 9.94004250e-01 -1.39184147e-01 3.80609185e-02 1.15660280e-01 7.94947445e-01 4.36632335e-01 -1.76585894e-02 8.18197548e-01 1.49767900e+00 -2.93733656e-01 2.28566825e-01 1.22078285e-01 -3.84607375e-01 7.33715534e-01 6.50416389e-02 1.61568344e-01 -5.31838059e-01 1.57170966e-01 6.69733107e-01 -8.37813318e-02 -6.45034969e-01 -4.40604120e-01 -9.34722483e-01 5.34841061e-01 6.90825582e-01 9.55950171e-02 -4.24503088e-01 -1.66854858e-01 1.46414638e-01 5.44905007e-01 2.23806813e-01 5.03041685e-01 -6.56992495e-02 1.91571712e-01 -1.13145912e+00 1.03289753e-01 4.95915592e-01 5.04757047e-01 9.77412403e-01 -9.38190222e-02 -7.12237582e-02 1.04983270e+00 4.52410996e-01 4.56806451e-01 5.16812265e-01 -1.07149351e+00 5.97751677e-01 5.74538887e-01 4.23551559e-01 -1.02942538e+00 -1.05758563e-01 -3.74999017e-01 -9.69514668e-01 6.65820539e-01 4.32999760e-01 -1.59633890e-01 -9.33415115e-01 1.48058164e+00 1.23773105e-01 1.89544469e-01 5.44592977e-01 1.15548229e+00 7.80108392e-01 8.57962549e-01 -2.23645166e-01 -2.12478235e-01 7.60996342e-01 -9.81330872e-01 -6.52773857e-01 -5.29109240e-02 1.43744990e-01 -6.29268765e-01 1.11682677e+00 5.79877496e-01 -1.32303751e+00 -7.87112117e-01 -1.43289053e+00 1.93870366e-01 -1.11775450e-01 2.57147551e-01 1.80990085e-01 4.29915071e-01 -1.09421802e+00 9.46705699e-01 -6.62940621e-01 1.35495499e-01 3.56417179e-01 2.11188257e-01 -3.72417033e-01 -3.04137737e-01 -1.28727722e+00 7.76668668e-01 4.92899179e-01 6.05856180e-01 -1.06026518e+00 -2.73238152e-01 -9.06529129e-01 8.58560950e-03 9.66251716e-02 -3.35012615e-01 9.45794880e-01 -1.18868756e+00 -1.77681732e+00 5.34181595e-01 2.86622375e-01 -2.30631843e-01 7.46075928e-01 -1.44165680e-01 -4.03823286e-01 2.71895915e-01 1.18517600e-01 6.27596855e-01 7.26333618e-01 -1.74109304e+00 -9.29062247e-01 -1.51887015e-01 2.62365669e-01 5.50027311e-01 -4.16343361e-01 -1.91925257e-01 -6.62474215e-01 -4.40185964e-01 3.78993720e-01 -6.70883000e-01 -3.26880217e-01 3.52249026e-01 -2.15090543e-01 1.89092666e-01 4.43862468e-01 -5.71631670e-01 9.71217573e-01 -2.19019437e+00 2.38234758e-01 2.99077600e-01 5.48855662e-02 3.94541681e-01 -4.33726251e-01 1.84520394e-01 2.25064874e-01 -8.11512470e-02 -6.83791161e-01 -4.67179000e-01 -8.55565295e-02 3.53934705e-01 -5.02414927e-02 4.27977949e-01 3.42781484e-01 4.35285479e-01 -1.39628470e+00 -6.82114720e-01 1.93193510e-01 3.58758152e-01 -4.58740652e-01 7.25442588e-01 1.46973208e-01 5.34425378e-01 -1.90357998e-01 5.59969425e-01 1.03771925e+00 -3.70961726e-02 4.88617942e-02 -4.56169844e-01 -2.40820363e-01 -4.91966940e-02 -1.35240138e+00 1.62759614e+00 -5.59121668e-01 5.07951438e-01 1.34129196e-01 -9.60385621e-01 1.07279968e+00 1.05426878e-01 2.99320430e-01 -8.18468332e-01 1.06749177e-01 5.14806330e-01 4.31429595e-02 -5.74450910e-01 5.65345168e-01 -3.11528444e-01 1.96174398e-01 2.31353447e-01 -1.12747196e-02 -2.57122904e-01 5.77799916e-01 4.34622355e-02 6.39219761e-01 2.78652906e-01 1.02977809e-02 7.98835233e-03 6.48838222e-01 -2.57970035e-01 7.28002250e-01 5.98362625e-01 -3.26979101e-01 8.52748811e-01 2.49541551e-01 -2.82622546e-01 -9.65122581e-01 -1.09627891e+00 -2.20636293e-01 8.92059147e-01 8.54118109e-01 5.96992970e-02 -7.15956211e-01 -4.71126378e-01 -2.82299727e-01 2.52288133e-01 -6.80249810e-01 -1.79855257e-01 -5.01151383e-01 -9.32401299e-01 4.01855886e-01 3.40776563e-01 9.92862105e-01 -1.28467894e+00 -5.96919954e-01 3.41687173e-01 -3.46125633e-01 -8.34748209e-01 -4.16041791e-01 1.53694361e-01 -9.15658712e-01 -8.57699096e-01 -1.10422635e+00 -8.54750931e-01 8.31874073e-01 4.45967227e-01 9.38204348e-01 3.44656706e-01 2.70462781e-01 -2.86927789e-01 -4.88393188e-01 -6.37709126e-02 -5.39026558e-01 -4.30955738e-01 -7.61640146e-02 3.61941129e-01 -1.91615984e-01 -5.06269038e-01 -9.10184085e-01 4.71300274e-01 -1.32272029e+00 8.12286362e-02 1.08132112e+00 1.20208395e+00 4.63763714e-01 9.85131487e-02 4.94349420e-01 -4.98775840e-01 5.94817221e-01 -3.06719214e-01 -6.49881959e-01 3.95313531e-01 -6.41374707e-01 3.15035224e-01 6.22449100e-01 -4.37497526e-01 -1.10777128e+00 -5.16829342e-02 -2.22217560e-01 -4.00967419e-01 2.12774962e-01 6.64723456e-01 -3.16876918e-01 -1.99594885e-01 3.97940695e-01 3.62367064e-01 1.31167307e-01 -1.64109543e-01 2.19941154e-01 7.21728325e-01 9.04079437e-01 -4.52543855e-01 1.07640219e+00 4.24113899e-01 -3.03412557e-01 -5.03396869e-01 -8.10584903e-01 -3.02151322e-01 -3.08405459e-01 -5.25224924e-01 6.96295381e-01 -8.93932641e-01 -3.75943542e-01 7.74996102e-01 -8.80038798e-01 -4.12681460e-01 -2.71620490e-02 4.99529570e-01 -3.40513766e-01 5.53729296e-01 -5.70907176e-01 -7.42014885e-01 -2.94556797e-01 -1.37922347e+00 9.54492509e-01 6.98443830e-01 5.77168643e-01 -6.96371734e-01 9.52313095e-02 2.82792687e-01 3.93831164e-01 2.36796513e-01 2.82072812e-01 -8.27999711e-02 -5.38923979e-01 -5.65920025e-02 -5.23294032e-01 7.35291004e-01 2.20364496e-01 2.53104623e-02 -8.22058976e-01 -3.09027612e-01 -1.49622470e-01 -5.79374850e-01 8.91005278e-01 1.25805974e-01 8.89845431e-01 -1.06902227e-01 1.69444919e-01 8.24294865e-01 1.49302340e+00 2.44798735e-01 7.74271250e-01 5.51916122e-01 4.78391737e-01 7.73364902e-01 8.67622077e-01 4.13383335e-01 3.75632048e-01 4.32563901e-01 7.02985168e-01 -3.65252823e-01 -1.73371693e-04 -2.03281716e-01 4.65790629e-01 8.43298018e-01 -3.19355816e-01 -2.52762705e-01 -4.28203136e-01 6.03327096e-01 -1.48423982e+00 -8.88572752e-01 1.53033301e-01 2.24861383e+00 1.00503528e+00 1.88370302e-01 -2.35402882e-01 3.32048863e-01 7.74204671e-01 7.80113786e-02 -5.58876097e-01 -1.03307003e-02 -3.72563481e-01 -2.39454489e-03 4.38368201e-01 3.57392639e-01 -1.16022444e+00 6.54895782e-01 5.32233238e+00 6.02079332e-01 -1.09239936e+00 -2.09324896e-01 6.80191875e-01 3.08880955e-01 -4.02473539e-01 -1.63841188e-01 -3.60449910e-01 5.08179069e-01 4.66619015e-01 -2.82572564e-02 3.33978713e-01 5.49920261e-01 2.75638819e-01 -3.94712128e-02 -7.45585203e-01 9.12752986e-01 1.69185683e-01 -9.41809535e-01 1.29639819e-01 -2.18578503e-01 1.06107414e+00 -1.88943043e-01 6.68683425e-02 1.78497344e-01 3.57226998e-01 -7.11501539e-01 9.24744010e-01 6.69680595e-01 7.71336436e-01 -6.77910507e-01 1.06266487e+00 1.66775063e-01 -1.09456670e+00 -2.44824097e-01 -5.61403394e-01 1.06496811e-01 1.91249803e-01 4.08250391e-01 -2.54445940e-01 6.94260538e-01 9.34296548e-01 6.22200847e-01 -5.21172583e-01 1.39676213e+00 -3.35108966e-01 3.90273005e-01 -2.61030316e-01 9.90987122e-02 3.17424744e-01 -6.42754257e-01 3.21774423e-01 9.56927001e-01 5.72099447e-01 1.73289716e-01 6.48686066e-02 6.29542649e-01 -1.55612320e-01 1.75106466e-01 -3.07031870e-01 3.11017543e-01 2.39631921e-01 1.27340853e+00 -6.04323566e-01 -4.69111592e-01 -3.53749394e-01 1.09689832e+00 1.29971772e-01 2.28490308e-01 -7.20767379e-01 -7.43598819e-01 3.56932163e-01 -2.68526286e-01 1.95628211e-01 6.17899634e-02 -1.11847751e-01 -1.13941467e+00 6.83408082e-02 -8.89160156e-01 1.43109247e-01 -8.81228328e-01 -1.34391630e+00 9.96773422e-01 -1.94482073e-01 -1.91131079e+00 -5.76748016e-06 -3.37525368e-01 -7.74788320e-01 8.51838410e-01 -2.01881313e+00 -7.63374329e-01 -8.58685732e-01 3.21803421e-01 4.54587549e-01 1.01142235e-01 3.29998344e-01 5.20501137e-01 -4.12115484e-01 6.20243490e-01 1.95210993e-01 3.34884495e-01 7.34509528e-01 -1.44273448e+00 -5.40818982e-02 1.06882918e+00 -9.41002145e-02 3.39165986e-01 8.87091994e-01 -3.92936707e-01 -1.03526354e+00 -9.93881047e-01 4.99341637e-01 2.22093031e-01 5.39379835e-01 1.35459945e-01 -1.14220786e+00 -2.05858946e-02 2.58219093e-01 2.16979176e-01 4.02724266e-01 -7.06427634e-01 2.13876041e-03 -4.68854338e-01 -1.03738606e+00 5.83244860e-01 8.49491596e-01 -1.97744265e-01 -5.39493084e-01 -7.49087185e-02 4.33123440e-01 -4.01188135e-01 -9.62056398e-01 6.51459575e-01 5.81200898e-01 -1.24608505e+00 8.19087744e-01 1.32390307e-02 8.26817513e-01 -8.17998290e-01 -1.87667772e-01 -1.54893363e+00 1.24560058e-01 -3.42286527e-01 2.99554616e-01 1.05636930e+00 4.20215189e-01 -3.67932498e-01 6.74686849e-01 3.56321484e-01 -4.08665389e-01 -6.59157276e-01 -6.55394614e-01 -5.57542682e-01 -2.22579420e-01 -7.90858269e-02 6.08630836e-01 6.96200073e-01 -1.97548106e-01 8.49609822e-02 -6.52584672e-01 5.82326531e-01 9.70070183e-01 3.89824927e-01 8.24166656e-01 -1.08920240e+00 -3.88826549e-01 -5.48026919e-01 -3.69882941e-01 -1.31112802e+00 -3.03389192e-01 -4.97260332e-01 7.01686263e-01 -1.68437088e+00 1.63351938e-01 -5.71502447e-01 -5.67529738e-01 3.10485989e-01 -6.22459531e-01 7.76574254e-01 1.76559344e-01 2.58857757e-01 -6.81095898e-01 1.04380047e+00 1.66833615e+00 -3.19161654e-01 -2.75255948e-01 4.81027849e-02 -5.47696590e-01 5.58126330e-01 7.29749084e-01 -3.28504145e-01 -2.95512319e-01 -4.92063820e-01 1.90429062e-01 3.27765703e-01 7.20408484e-02 -1.35622621e+00 2.37931237e-01 -1.50770724e-01 3.43761355e-01 -3.67225677e-01 1.50913075e-01 -6.58958495e-01 -1.01951644e-01 3.94536257e-01 -4.02060568e-01 -1.09827153e-01 -2.97808558e-01 4.44972843e-01 -7.59362817e-01 -4.64744389e-01 1.08292389e+00 -1.05108976e-01 -8.16643775e-01 2.60184228e-01 3.78191546e-02 -1.15857147e-01 6.80478573e-01 -4.26229239e-01 -5.09976819e-02 -4.52004224e-01 -5.01731634e-01 4.54004556e-01 5.76140642e-01 1.44015446e-01 1.03745902e+00 -1.22190046e+00 -9.13535416e-01 7.58579001e-02 3.17932606e-01 1.54815137e-01 3.44548553e-01 5.89048207e-01 -7.16084778e-01 -5.42066753e-01 -5.28873801e-01 -4.25184786e-01 -9.08380568e-01 1.68756112e-01 5.65339386e-01 -2.15536997e-01 -4.54050243e-01 7.06071019e-01 9.39324275e-02 -4.62738395e-01 1.32644221e-01 -2.09589690e-01 -6.67044818e-01 -1.00259528e-01 6.74548566e-01 2.65422791e-01 -1.63619414e-01 -7.04454720e-01 7.66315013e-02 7.11410165e-01 2.28861704e-01 -2.24129900e-01 1.50455332e+00 -2.53837585e-01 -1.12888403e-01 2.89937049e-01 1.19431937e+00 -8.85884836e-02 -1.79810536e+00 -3.14438164e-01 -1.41599000e-01 -6.26989484e-01 8.07229280e-02 -6.08082950e-01 -1.51596165e+00 7.83776820e-01 8.71808112e-01 1.43801913e-01 1.57663131e+00 -3.27617615e-01 8.17808628e-01 3.09036762e-01 2.41020814e-01 -1.04918969e+00 3.47648531e-01 1.04925580e-01 9.18225229e-01 -1.60372758e+00 -1.07007049e-01 3.07777449e-02 -7.03762412e-01 1.03614128e+00 7.38212168e-01 -1.45813242e-01 2.82429516e-01 1.18738011e-01 6.71645463e-01 -4.08223160e-02 -3.01990032e-01 -3.14876944e-01 1.93929911e-01 3.72234643e-01 1.35116085e-01 -1.60240516e-01 -3.55532765e-01 2.76754528e-01 -1.31417364e-01 1.82048194e-02 8.09432089e-01 7.51879990e-01 -6.11861765e-01 -1.00399995e+00 -4.80915695e-01 3.53856832e-01 -3.59683752e-01 -6.12188242e-02 2.00621426e-01 4.61701334e-01 2.76663780e-01 1.11322176e+00 -6.18989915e-02 -5.30361950e-01 3.90626848e-01 -5.31432271e-01 1.92526504e-01 -2.52284497e-01 -2.50489503e-01 3.05573810e-02 -2.92011738e-01 -3.25425684e-01 -8.66264284e-01 -5.23390889e-01 -1.33113837e+00 2.31508493e-01 -3.77668798e-01 6.18082166e-01 6.87016308e-01 7.25927711e-01 -2.09712625e-01 4.83617574e-01 1.06716478e+00 -1.26196885e+00 -6.28934145e-01 -1.12079751e+00 -6.14280164e-01 7.25716889e-01 4.37588573e-01 -7.87341118e-01 -6.67753279e-01 1.67920381e-01]
[10.716172218322754, -3.510483980178833]
52b1c309-8cee-45ac-9a4f-8f077ae1c479
bert-proof-syntactic-structures-investigating
null
null
https://aclanthology.org/2021.findings-acl.288
https://aclanthology.org/2021.findings-acl.288.pdf
BERT-Proof Syntactic Structures: Investigating Errors in Discontinuous Constituency Parsing
null
['Maximin Coavoux']
null
null
null
null
findings-acl-2021-8
['constituency-parsing']
['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]
[-7.213083744049072, 3.8001513481140137]
9c4794cd-18db-4fcc-b2bb-7a07f8cc1e12
energy-analysis-of-bursting-hindmarsh-rose
2203.11252
null
https://arxiv.org/abs/2203.11252v1
https://arxiv.org/pdf/2203.11252v1.pdf
Energy analysis of bursting Hindmarsh-Rose neurons with time-delayed coupling
Mathematical modeling is an important tool to study the role of delay in neural systems and to evaluate its effects on the signaling activity of coupled neurons. Models for delayed neurons are often used to represent the dynamics of real neurons, but rarely to assess the energy required to maintain these dynamics. In this work, we address these questions from an energy perspective by considering a pair of Hindmarsh-Rose burst neurons coupled by reciprocal time-delayed coupling with electrical and chemical synapses. We examine the average energy consumption required to maintain cooperative behavior and quantify the contribution of synapses to total energy consumption. We show that unlike electrical coupling, where the time delay appears to reduce the instantaneous average relative weight of the synaptic contribution, in chemical coupling this average synaptic contribution appears to be much higher in delayed coupling than in instantaneous coupling, except at certain values of coupling strength where the instantaneous synaptic contribution is more important.
['Fernando Vadillo', 'Abdelmalik Moujahid']
2022-03-21
null
null
null
null
['total-energy']
['miscellaneous']
[ 1.85636044e-01 -1.63065270e-01 1.54945537e-01 3.84556472e-01 1.71838880e-01 -7.60088861e-01 4.99763936e-01 5.30222893e-01 -9.04728830e-01 8.83103132e-01 -3.72418582e-01 -4.79115397e-02 -1.10969141e-01 -6.12201154e-01 -6.36603951e-01 -1.27834809e+00 -2.10233405e-01 -4.79876772e-02 5.38215995e-01 -5.22423387e-01 2.93231964e-01 6.22259498e-01 -1.28221357e+00 -3.23062718e-01 3.34360987e-01 5.58424413e-01 4.05652910e-01 6.22896254e-01 7.09970295e-02 5.38872778e-01 -7.18106985e-01 2.08712921e-01 -1.14069767e-01 -6.08873129e-01 -3.54583472e-01 -5.03622711e-01 -5.45681119e-01 9.80050117e-02 -3.17838311e-01 6.62908494e-01 7.06159353e-01 1.87932625e-01 7.81392694e-01 -1.23866880e+00 6.43522516e-02 8.35988581e-01 -3.19130570e-01 6.92029417e-01 -2.25878134e-01 -1.60721466e-01 6.60926342e-01 -1.93108648e-01 4.31878775e-01 6.73613310e-01 5.28438747e-01 7.89748967e-01 -1.75549281e+00 -7.20887363e-01 2.02786215e-02 -1.63158312e-01 -1.30445218e+00 -4.00692284e-01 6.33404016e-01 -1.93383545e-01 1.10647702e+00 1.15128676e-03 1.11059415e+00 6.22083426e-01 7.87591875e-01 -4.40127850e-02 8.96387458e-01 -1.89096153e-01 6.32122993e-01 -3.17936003e-01 3.96020144e-01 1.96170628e-01 5.52349389e-01 1.01081081e-01 -5.85387945e-01 -2.94832230e-01 6.56096041e-01 -1.76532507e-01 -3.61957580e-01 2.74992615e-01 -9.49536324e-01 2.62979329e-01 1.63560174e-02 6.81280673e-01 -2.67045051e-01 1.09344423e+00 -1.48344394e-02 3.60893726e-01 1.87014535e-01 2.71358699e-01 -3.57530832e-01 -2.75949687e-01 -7.12585807e-01 4.15730566e-01 7.60534883e-01 2.83729851e-01 4.08216834e-01 -2.24079356e-01 2.01788172e-01 6.26513660e-01 6.55123293e-02 4.29236233e-01 8.81940424e-02 -1.27938855e+00 -9.87672284e-02 5.33648551e-01 3.18343043e-02 -4.67463493e-01 -6.68875456e-01 -2.81029195e-01 -7.73497581e-01 2.63476640e-01 8.34103942e-01 -3.65358233e-01 -7.38828257e-02 1.91716564e+00 -2.68028080e-01 4.05898616e-02 1.09027602e-01 3.71594787e-01 1.65010408e-01 8.75491321e-01 1.21955752e-01 -8.23596060e-01 8.80044997e-01 -8.80741253e-02 -6.10563457e-01 1.01061948e-01 6.20578945e-01 -5.08978903e-01 2.89403796e-01 -2.51425728e-02 -1.37803113e+00 6.67824447e-02 -1.09490561e+00 3.97493392e-01 -3.79437804e-01 -4.71205115e-01 -3.33760493e-02 2.51901180e-01 -1.24808812e+00 9.42047298e-01 -1.05560446e+00 -4.36099559e-01 -2.03098282e-02 6.50290489e-01 3.56760137e-02 4.50727552e-01 -9.40177977e-01 9.34579492e-01 -9.99447107e-02 -1.54823750e-01 -5.14809847e-01 -7.22938955e-01 6.12748638e-02 3.75551492e-01 -1.29237756e-01 -5.21924675e-01 1.06377482e+00 -3.78192127e-01 -1.27981818e+00 5.29876828e-01 -1.19957058e-02 -5.29083610e-01 8.17534775e-02 6.29147768e-01 2.10670292e-01 1.59628019e-01 -3.67482781e-01 7.40568161e-01 -2.23998353e-01 -1.11221206e+00 -9.92446393e-02 -3.16911519e-01 3.77029739e-02 2.04985708e-01 -2.26585686e-01 2.81186290e-02 1.49017677e-01 -3.81118387e-01 -6.04510196e-02 -1.26525915e+00 -3.11197072e-01 1.50483176e-01 1.18149512e-01 -2.58147538e-01 5.33771694e-01 3.05729091e-01 9.56468940e-01 -2.15303278e+00 4.54975009e-01 1.28583327e-01 3.25445622e-01 -1.20476969e-01 9.05178487e-02 8.94093275e-01 1.47088513e-01 3.25768776e-02 -2.99438834e-01 -4.68952209e-02 -5.38719714e-01 3.15588087e-01 7.07530454e-02 4.69011039e-01 1.73011124e-01 4.62015241e-01 -7.04068899e-01 -1.42727911e-01 -2.80461699e-01 9.16022539e-01 -2.52471983e-01 -1.61913976e-01 8.85782316e-02 1.89630494e-01 -3.85761350e-01 1.90435827e-01 1.78130880e-01 8.67586881e-02 2.84089059e-01 2.08021775e-02 -6.96429729e-01 -7.04930276e-02 -7.54733682e-01 8.11502457e-01 -2.91594893e-01 9.23172534e-01 2.35852972e-01 -1.06852996e+00 9.39120650e-01 4.74844664e-01 8.60248148e-01 -8.68357420e-01 5.27772009e-01 3.31329525e-01 5.73905408e-01 7.74098113e-02 2.37635057e-02 -2.53298640e-01 5.84377050e-02 7.28594720e-01 -8.71446431e-02 -1.47300184e-01 6.52569532e-01 3.12517345e-01 1.27109611e+00 -3.47599804e-01 -9.57798287e-02 -7.68829584e-01 9.68515426e-02 -2.36669451e-01 4.44358915e-01 1.03586629e-01 -2.34734446e-01 2.14545786e-01 9.15476799e-01 8.17511976e-02 -1.06266367e+00 -7.51507223e-01 -9.21533406e-02 5.62835217e-01 7.66527414e-01 -1.39936984e-01 -9.88041162e-01 4.71679956e-01 -1.50930151e-01 4.46111679e-01 -6.83476508e-01 -6.17005587e-01 -3.71197969e-01 -1.15166163e+00 5.19519567e-01 3.52977246e-01 1.49409816e-01 -9.32160318e-01 -1.03924096e+00 5.58966696e-01 -1.19267561e-01 -8.61181438e-01 -5.41049540e-01 8.67839813e-01 -1.01974964e+00 -9.16073799e-01 -7.10203767e-01 -3.61955851e-01 8.03484201e-01 2.46816501e-01 7.10132718e-01 5.20954788e-01 -4.15824026e-01 2.24441573e-01 6.04994297e-02 -4.23206240e-01 -2.19117641e-01 -1.31750122e-01 1.90272659e-01 -4.83670682e-01 -1.26094192e-01 -1.02406502e+00 -6.48862004e-01 4.19527054e-01 -9.88296568e-01 -1.98811233e-01 1.17091291e-01 4.30582494e-01 6.54673636e-01 8.54877830e-02 5.21616399e-01 -6.54388368e-02 7.66066313e-01 -3.45661670e-01 -5.84111512e-01 -1.70635581e-02 -4.94679928e-01 4.01940793e-01 6.01814985e-01 -7.03635752e-01 -5.00627697e-01 2.58206446e-02 1.58785403e-01 2.78041631e-01 4.45642203e-01 1.39421895e-01 2.13278830e-01 -3.42375398e-01 3.13286901e-01 4.63385671e-01 1.44641504e-01 -2.01395124e-01 -3.58084887e-01 -2.36044638e-02 2.74570305e-02 -5.46559513e-01 2.53659070e-01 4.22401339e-01 6.07221603e-01 -7.92262793e-01 1.30443692e-01 1.36122510e-01 -2.24689052e-01 -5.02376318e-01 6.18793666e-01 -3.80396634e-01 -1.25844109e+00 6.49399400e-01 -1.12528276e+00 -7.81411588e-01 -1.75206438e-01 4.39585984e-01 -7.23631799e-01 -2.31165797e-01 -6.91106260e-01 -1.07259989e+00 -2.06275418e-01 -9.44275618e-01 4.29036438e-01 4.41287696e-01 -3.37982550e-02 -9.57766593e-01 5.15482724e-01 -4.18850064e-01 4.52670783e-01 1.70326114e-01 1.09620368e+00 -9.02785957e-02 -3.41618329e-01 9.52264108e-03 1.54511273e-01 -1.62925646e-01 -7.52001330e-02 6.60850883e-01 -3.67908120e-01 -4.85839210e-02 -3.42268556e-01 9.61480364e-02 7.90424883e-01 8.47691715e-01 3.83999318e-01 7.93742388e-02 -7.47416437e-01 4.64034081e-03 1.61274636e+00 7.06366837e-01 4.05494779e-01 6.50336817e-02 -1.10573962e-01 7.26688623e-01 -6.83978274e-02 2.80906975e-01 3.13789211e-02 6.46440446e-01 3.72731507e-01 8.67082104e-02 1.50396405e-02 3.13364208e-01 4.59503531e-01 8.62390518e-01 -4.91485953e-01 -5.47515869e-01 -6.61570191e-01 4.32537049e-01 -1.65638006e+00 -8.59483302e-01 -1.85952708e-01 2.21925092e+00 9.17660475e-01 3.66721421e-01 3.82758975e-01 3.67421150e-01 7.77589977e-01 -5.07961988e-01 -2.90784359e-01 -4.23702985e-01 -3.43113154e-01 -1.67398706e-01 7.03345478e-01 6.22322261e-01 2.03420743e-01 1.74473584e-01 7.62105227e+00 4.30092424e-01 -1.17666626e+00 -9.51582491e-02 4.35857087e-01 -7.50254273e-01 -3.72516721e-01 1.55014664e-01 -7.27438152e-01 9.06618655e-01 1.24357557e+00 -6.18963718e-01 5.73053598e-01 -1.72022492e-01 5.90978980e-01 -6.05974615e-01 -9.54101562e-01 5.63166976e-01 -8.63346994e-01 -1.28025842e+00 -3.22047055e-01 5.63299596e-01 3.02015424e-01 5.49426954e-03 -3.48645568e-01 -2.98563659e-01 7.41944611e-02 -6.04115427e-01 5.91699719e-01 7.61623144e-01 1.78636909e-01 -8.72713268e-01 5.57356119e-01 5.85300684e-01 -1.14006746e+00 -4.24599908e-02 -3.34517360e-01 -3.19589496e-01 2.91155130e-01 7.80607045e-01 -6.97526187e-02 -3.70421916e-01 4.66675311e-01 1.43981948e-01 -2.43221242e-02 8.50635171e-01 5.69967091e-01 4.08764720e-01 -8.75743866e-01 -5.05127907e-01 -3.51938568e-02 -3.61707360e-01 4.29408938e-01 6.46031678e-01 4.40673858e-01 4.95895505e-01 -6.68343902e-01 9.62857842e-01 -8.01089257e-02 -2.34414518e-01 -5.88678896e-01 -3.81840974e-01 8.01887095e-01 1.11244249e+00 -1.43913150e+00 1.57012616e-03 -1.04793236e-01 4.66362208e-01 8.02666694e-02 2.80184418e-01 -7.60461390e-01 -4.08817291e-01 7.55399168e-01 3.74589086e-01 -7.41147473e-02 -6.91152334e-01 3.21706794e-02 -4.68115717e-01 -3.09134185e-01 3.61377031e-01 -3.81146461e-01 -4.45742995e-01 -5.09836912e-01 4.06883746e-01 1.63641840e-01 -7.71161020e-01 1.10807307e-01 -1.54845417e-01 -8.75750363e-01 4.82952118e-01 -1.11781073e+00 -2.31908843e-01 7.78902844e-02 3.85010540e-01 3.54555883e-02 2.80456752e-01 5.34172595e-01 2.60222733e-01 -8.02815378e-01 1.66545153e-01 4.29232091e-01 -4.15753931e-01 1.55482545e-01 -8.36938858e-01 4.49158698e-02 3.45860958e-01 -1.06529064e-01 6.23296976e-01 1.08030725e+00 -3.63823533e-01 -1.28701365e+00 -5.34712732e-01 9.39817607e-01 -1.06589653e-01 4.26056504e-01 -3.46153647e-01 -5.96774518e-01 -3.85038257e-02 3.52988571e-01 8.10425580e-02 8.81191313e-01 -6.75049245e-01 1.81073248e-01 -2.33005092e-01 -9.12000418e-01 9.08876181e-01 9.45035219e-01 -1.22858025e-01 -8.54169112e-03 -1.12330556e-01 4.56985652e-01 4.79829967e-01 -8.45094144e-01 4.30083089e-03 9.34212863e-01 -7.76892900e-01 5.46185434e-01 1.72481045e-01 1.00989100e-02 -2.93435454e-01 -6.23518489e-02 -1.29456723e+00 -1.93506241e-01 -5.13104200e-01 1.49741456e-01 1.16262019e+00 6.31504297e-01 -7.18767583e-01 3.85635585e-01 4.11206752e-01 1.57128438e-01 -9.11335051e-01 -1.10474539e+00 -9.06354725e-01 1.81209490e-01 2.85026401e-01 -1.37496412e-01 4.09374058e-01 4.95824218e-01 2.57433027e-01 2.90947765e-01 -3.69577050e-01 6.00700974e-01 -3.44811767e-01 -2.18181863e-01 -1.30306518e+00 -8.57939720e-02 -8.01550269e-01 -3.31216574e-01 -4.14340049e-01 -8.78850892e-02 -4.99929160e-01 1.00405090e-01 -1.37625301e+00 3.77085000e-01 -1.48748815e-01 -4.65142071e-01 1.14267521e-01 2.21286193e-01 5.52242994e-01 6.61607087e-02 2.26163089e-01 -7.15214908e-02 1.45140037e-01 9.41950560e-01 2.08267108e-01 -3.21256369e-01 -6.56621233e-02 -4.23292160e-01 3.47296685e-01 7.09011436e-01 -9.50996935e-01 -4.81759578e-01 -5.24733402e-02 6.01120234e-01 1.74040467e-01 8.30797628e-02 -1.03574669e+00 4.82078910e-01 -2.88185120e-01 -4.48833704e-02 -2.36224040e-01 4.50125903e-01 -8.37006629e-01 4.09530371e-01 9.71226931e-01 -5.92835307e-01 1.52484760e-01 2.41873130e-01 5.28671503e-01 7.96848983e-02 -1.53123155e-01 9.65691328e-01 1.92315523e-02 1.11849904e-01 -3.08495816e-02 -1.47914457e+00 -2.34043777e-01 1.25862324e+00 -4.28948373e-01 -3.32119286e-01 -2.38788083e-01 -8.20499837e-01 8.46420750e-02 6.22484565e-01 -1.96352527e-01 1.91586271e-01 -1.06286049e+00 -1.98579565e-01 -2.71906376e-01 -1.92981333e-01 -7.51894176e-01 1.71627983e-01 1.14624524e+00 -4.33940202e-01 2.56790936e-01 -6.18998647e-01 -3.46698612e-01 -1.39980114e+00 2.93983936e-01 6.85088396e-01 8.85068178e-02 -1.47505319e-02 8.84428263e-01 3.71018192e-03 5.50412178e-01 4.18473817e-02 -4.09705549e-01 -9.72196832e-02 1.98753357e-01 2.43127823e-01 6.90922797e-01 -8.77026394e-02 -2.34552860e-01 -5.69593787e-01 9.97053146e-01 2.85559297e-01 -3.22610855e-01 1.17556751e+00 -2.83287555e-01 -3.33556801e-01 7.81698644e-01 8.90658975e-01 -3.69677037e-01 -1.42617309e+00 3.16253930e-01 -2.38919780e-01 2.93335885e-01 1.84346542e-01 -3.71498287e-01 -1.15605521e+00 1.03976429e+00 4.42016244e-01 7.04903126e-01 9.92448866e-01 1.40853450e-01 6.31067038e-01 4.33700740e-01 2.78144479e-01 -1.11342919e+00 1.71259239e-01 1.45337582e-01 4.24050719e-01 -3.35296452e-01 -2.00018153e-01 -4.29350853e-01 -1.86824556e-02 1.22135150e+00 3.58414054e-01 -4.60526109e-01 9.06896055e-01 1.09924972e+00 -3.67594808e-01 1.15665235e-02 -1.47728086e+00 -1.97011363e-02 -4.84198213e-01 2.88144082e-01 6.99329913e-01 -1.78835049e-01 -8.88953030e-01 2.80230820e-01 2.59350091e-01 1.75690800e-01 9.20140445e-01 1.00870264e+00 -7.02696025e-01 -1.20455003e+00 1.42277926e-01 2.68991888e-01 -6.34185433e-01 -4.61849682e-02 -7.59825945e-01 5.41341960e-01 8.43959078e-02 8.77602935e-01 5.83380759e-01 -1.94811761e-01 2.07342952e-01 4.58735228e-02 9.22345221e-01 -2.44759738e-01 -5.38341761e-01 3.40664625e-01 -1.80809557e-01 -1.86927572e-01 -6.07187212e-01 -5.30413032e-01 -1.89983547e+00 -6.51931584e-01 -4.65600222e-01 3.49210441e-01 8.76532614e-01 1.07170439e+00 3.01670462e-01 6.12498343e-01 4.87741739e-01 -6.97257936e-01 1.99900139e-02 -5.20747244e-01 -1.08537555e+00 -3.12124521e-01 4.51123685e-01 -6.50467634e-01 -6.44454479e-01 9.15936455e-02]
[8.014143943786621, 2.8544862270355225]
9df0c625-5867-4269-bf76-b8d1cb377c01
300-sparsans-at-semeval-2018-task-9-hypernymy
null
null
https://aclanthology.org/S18-1152
https://aclanthology.org/S18-1152.pdf
300-sparsans at SemEval-2018 Task 9: Hypernymy as interaction of sparse attributes
This paper describes 300-sparsians{'}s participation in SemEval-2018 Task 9: Hypernym Discovery, with a system based on sparse coding and a formal concept hierarchy obtained from word embeddings. Our system took first place in subtasks (1B) Italian (all and entities), (1C) Spanish entities, and (2B) music entities.
['P{\\\'e}ter F{\\"o}ldi{\\\'a}k', "M{\\'a}rton Makrai", "G{\\'a}bor Berend"]
2018-06-01
null
null
null
semeval-2018-6
['hypernym-discovery']
['natural-language-processing']
[-2.25452960e-01 7.98272848e-01 -1.20884247e-01 -2.03238860e-01 -2.15619057e-01 -5.14917254e-01 7.26497769e-01 7.64797926e-01 -9.41670120e-01 9.88794446e-01 6.03920639e-01 -3.07595789e-01 -3.40130240e-01 -7.70510674e-01 -4.82032537e-01 -1.21008888e-01 -4.98413384e-01 1.08935928e+00 6.69920594e-02 -5.13868868e-01 -8.18144158e-02 6.82050735e-02 -1.40610111e+00 3.95884424e-01 7.26094544e-01 8.93175423e-01 -1.90342382e-01 2.62098819e-01 -2.67727286e-01 8.73939216e-01 -5.94919205e-01 -1.07460988e+00 1.51989341e-01 1.44769549e-01 -1.33245325e+00 -3.66277635e-01 6.87729716e-01 2.93135107e-01 -5.40290534e-01 1.23875952e+00 2.61407375e-01 5.79756320e-01 9.00211275e-01 -1.36688304e+00 -9.02974784e-01 1.35388732e+00 -2.92677522e-01 2.63584286e-01 5.78160942e-01 -2.46507719e-01 1.93916094e+00 -1.29294777e+00 1.22611701e+00 1.19665432e+00 7.43527532e-01 3.44729424e-01 -1.09304261e+00 -6.85331523e-01 -1.24770299e-01 6.35156572e-01 -1.89976716e+00 -8.37113615e-03 8.46441761e-02 -3.05269003e-01 1.93198979e+00 1.53728113e-01 9.22734618e-01 1.27449369e+00 -8.51557791e-01 4.96116161e-01 7.36706138e-01 -6.27028704e-01 2.60220826e-01 7.70470127e-02 4.76264477e-01 6.84029937e-01 1.02287638e+00 -2.44734392e-01 -5.90177298e-01 -3.97843510e-01 4.18227226e-01 -4.20749217e-01 4.39050421e-02 -2.93893874e-01 -1.17118633e+00 1.03232038e+00 4.77837950e-01 7.55725443e-01 -4.57793176e-01 -6.14946932e-02 3.99106294e-01 7.07249194e-02 3.43010388e-02 1.26118219e+00 -6.68829501e-01 -2.01413199e-01 -9.50448811e-01 4.36683178e-01 1.28954983e+00 1.30417788e+00 8.30697775e-01 7.08891973e-02 7.11229295e-02 9.59304154e-01 1.27612993e-01 2.82893836e-01 6.41891599e-01 -6.43490911e-01 4.95260268e-01 6.57700837e-01 -6.44998923e-02 -9.80626285e-01 -5.97847342e-01 -1.96492463e-01 -6.61482394e-01 -4.94710326e-01 -8.29688460e-02 -2.43417382e-01 -9.02859569e-01 1.41464400e+00 1.90368980e-01 4.01198864e-01 1.85158059e-01 7.19851673e-01 1.61885905e+00 2.44434074e-01 5.15929282e-01 1.15314677e-01 1.70048869e+00 -7.31130421e-01 -5.29377043e-01 1.04559369e-01 6.96014524e-01 -4.30201083e-01 6.65293992e-01 1.94698364e-01 -9.08472598e-01 -2.13857219e-01 -8.71174276e-01 -3.18678111e-01 -1.06980002e+00 -3.47521231e-02 8.52546692e-01 3.44393551e-01 -1.02227557e+00 4.09189731e-01 -1.48768857e-01 -1.00409079e+00 5.59472263e-01 4.12027687e-01 -7.87238419e-01 3.11767235e-02 -1.89905000e+00 1.23532450e+00 1.18473172e+00 -1.27239302e-01 -5.24948001e-01 -9.17407274e-01 -1.44887614e+00 2.71541685e-01 3.85526836e-01 -8.44265699e-01 7.07651734e-01 -1.80291221e-01 -7.98183799e-01 1.40982985e+00 1.43834174e-01 -1.02008748e+00 -4.65572804e-01 -2.90994465e-01 -8.88604879e-01 6.99398667e-02 4.31180149e-01 9.12081540e-01 4.58768725e-01 -9.92103100e-01 -5.85810184e-01 -1.08579444e-02 3.37489724e-01 1.71567366e-01 -6.82408094e-01 1.63860664e-01 -7.51794502e-02 -7.49432623e-01 -8.99157599e-02 -6.84868038e-01 -2.99197704e-01 -6.41765177e-01 -9.42916691e-01 -7.75281072e-01 1.71765238e-01 -5.44376612e-01 1.57761776e+00 -1.93795502e+00 1.83385178e-01 5.24743676e-01 6.33656919e-01 2.96757668e-01 -2.63678223e-01 5.98265231e-01 -6.78886592e-01 3.85139465e-01 -9.65534896e-02 -3.52413267e-01 2.66545236e-01 3.94314736e-01 -2.93241769e-01 1.75575227e-01 1.71984747e-01 1.03532422e+00 -1.10676765e+00 -5.70139468e-01 -8.80754068e-02 7.98245966e-02 -6.22996628e-01 -7.91582540e-02 -4.75578830e-02 -4.58320796e-01 -1.70368463e-01 7.30059743e-01 2.20558122e-01 -4.68654007e-01 5.47354460e-01 -5.00116110e-01 -2.41311267e-02 5.42419076e-01 -1.35873294e+00 1.80364954e+00 -1.35922521e-01 5.50260603e-01 -1.91025034e-01 -8.62362802e-01 6.22605622e-01 4.09015894e-01 8.12707126e-01 -3.36537182e-01 6.31371737e-02 1.92551002e-01 -1.13603370e-02 -4.71946388e-01 7.15617239e-01 -1.09732062e-01 -2.79764235e-01 1.65656418e-01 9.63575065e-01 -9.36935395e-02 7.15484321e-01 9.08212185e-01 1.41467977e+00 -8.48011822e-02 8.55500042e-01 -4.76232409e-01 1.88185006e-01 7.29799420e-02 4.20184374e-01 5.03403366e-01 1.14475898e-01 5.00327528e-01 4.09418523e-01 -7.28562057e-01 -1.06546140e+00 -1.00123751e+00 -2.46026754e-01 1.15649438e+00 -5.37447818e-02 -1.32681894e+00 9.17075127e-02 -7.05028653e-01 6.14184439e-01 9.10691559e-01 -9.37340081e-01 2.24589869e-01 -4.77027744e-01 -2.30992481e-01 1.01394725e+00 4.30278063e-01 3.62123698e-02 -1.36827004e+00 -3.37941080e-01 1.06171938e-02 -3.18654567e-01 -1.54902601e+00 -2.02232167e-01 5.65873086e-01 -2.14462981e-01 -1.39188838e+00 -2.98861533e-01 -1.05125141e+00 3.86970252e-01 -4.58146453e-01 1.70143032e+00 -4.84121079e-03 -7.01702476e-01 4.37538236e-01 -5.94226897e-01 -4.79194820e-01 4.52803940e-01 5.75369179e-01 3.38806719e-01 -4.08668041e-01 8.56523931e-01 -7.70005167e-01 -3.88231203e-02 -3.55779469e-01 -7.05823660e-01 -4.81607258e-01 2.79220819e-01 9.63538170e-01 4.99450773e-01 -8.16644579e-02 3.73385906e-01 -1.38551092e+00 5.64683735e-01 -9.00193036e-01 -3.32832932e-01 2.96139985e-01 -8.09752524e-01 -1.78975120e-01 2.46385768e-01 -2.21514925e-01 -7.36528814e-01 7.98530281e-02 -1.16537236e-01 -4.90110070e-01 -9.55351368e-02 8.13344419e-01 -1.72373861e-01 2.41060182e-01 1.17670894e+00 -1.50931433e-01 -7.52968311e-01 -4.89104837e-01 1.02694070e+00 4.60932612e-01 7.74882734e-01 -9.24351454e-01 1.11234462e+00 1.17756635e-01 1.25313736e-02 -1.07324767e+00 -1.10401940e+00 -1.03043103e+00 -7.72000730e-01 4.36907887e-01 9.97039676e-01 -1.32011139e+00 -7.52645254e-01 -1.81361437e-01 -1.29818034e+00 2.79705644e-01 -1.03164208e+00 5.69509268e-01 5.85949086e-02 1.81845829e-01 -5.69216311e-01 -4.30540830e-01 -3.61527711e-01 -1.08444570e-02 9.93197501e-01 2.11426452e-01 -7.41986156e-01 -9.67374802e-01 3.97938162e-01 2.21969411e-01 2.02247247e-01 2.48428416e-02 1.27156138e+00 -1.49190283e+00 -1.43721998e-01 -1.33309931e-01 -3.07743520e-01 9.47223976e-02 -1.98687688e-01 -2.23820627e-01 -7.00562894e-01 -1.99904293e-03 -7.56143212e-01 -8.27074349e-01 1.06737864e+00 -5.88241592e-02 9.99992251e-01 -4.82711077e-01 -3.89534950e-01 6.05865359e-01 1.32601619e+00 -3.79326820e-01 6.37804568e-01 4.36620653e-01 9.91273403e-01 3.86577845e-01 2.30029643e-01 6.95551813e-01 6.19326532e-01 5.22275209e-01 4.32831973e-01 1.45836115e-01 -1.47497147e-01 -5.74776590e-01 -5.40025644e-02 8.89688313e-01 -2.05160528e-01 -8.17162469e-02 -1.02173126e+00 1.19043207e+00 -1.71035731e+00 -1.13309348e+00 -2.50976622e-01 1.64953423e+00 1.02531493e+00 -2.44347870e-01 1.66740879e-01 -1.76199943e-01 5.97663641e-01 1.41036108e-01 -1.62836581e-01 -2.99171954e-01 -7.04150617e-01 1.12889874e+00 3.19494456e-01 3.54834437e-01 -1.32941258e+00 1.55273318e+00 6.44342184e+00 9.17454600e-01 -1.18379354e-01 2.17083454e-01 -2.67877337e-02 -1.69114679e-01 -9.45629239e-01 4.62078154e-01 -9.88526344e-01 1.99510425e-01 5.87389529e-01 -3.25952798e-01 3.37350905e-01 7.24058211e-01 -1.00654411e+00 2.56728619e-01 -9.91292417e-01 1.09652960e+00 4.10266548e-01 -1.54228199e+00 2.98852831e-01 -2.17527971e-01 1.02034879e+00 2.01814309e-01 -3.83696675e-01 9.91085291e-01 8.67108285e-01 -1.57580495e+00 3.69582087e-01 2.65895158e-01 1.21971464e+00 -4.36108708e-01 7.13514030e-01 -2.04078868e-01 -1.45931268e+00 -1.00492872e-01 -4.25460577e-01 9.13625434e-02 4.64675017e-02 7.50359714e-01 -9.20803726e-01 5.86122632e-01 8.10803771e-01 9.10025001e-01 -5.63340127e-01 1.01337194e+00 -1.03000414e+00 8.54869187e-01 -3.63223344e-01 -2.07273319e-01 2.05310583e-01 5.80215231e-02 8.82284641e-01 1.59575975e+00 -1.32629931e-01 3.32978249e-01 1.47519171e-01 7.70745397e-01 -7.85236478e-01 2.74695396e-01 -6.19148493e-01 -3.99966747e-01 1.03258717e+00 1.38531816e+00 -5.15919447e-01 -4.68918145e-01 -3.44805002e-01 7.74239540e-01 7.57267475e-01 5.02064645e-01 -6.11236095e-01 -9.75556314e-01 7.19167650e-01 7.93636516e-02 5.79766810e-01 -1.09330006e-01 -2.00888306e-01 -1.46949637e+00 -2.30449721e-01 -6.52627587e-01 1.09569120e+00 -7.06623197e-01 -1.80893326e+00 6.78757071e-01 -8.89644176e-02 -5.13035536e-01 -2.95226872e-01 -5.96094728e-01 -1.08872972e-01 6.31754100e-01 -1.10860598e+00 -1.37403107e+00 4.70890217e-02 7.81077206e-01 7.58554637e-02 -5.58944702e-01 1.33823442e+00 6.26428843e-01 -1.71481431e-01 7.03738153e-01 -3.77019942e-01 4.00629789e-01 6.93387270e-01 -1.66835368e+00 5.31805754e-01 3.97646606e-01 1.06246877e+00 9.06364322e-01 3.46338719e-01 -7.89416790e-01 -7.39351928e-01 -8.09888601e-01 1.95435607e+00 -7.39254415e-01 1.21220672e+00 -4.45020467e-01 -5.79763174e-01 7.49713123e-01 4.49641734e-01 3.00073028e-01 9.80875731e-01 9.89202201e-01 -8.62921596e-01 3.60835701e-01 -9.11640048e-01 2.42739037e-01 1.44750011e+00 -8.63346040e-01 -1.20689583e+00 4.93926525e-01 9.84638095e-01 -3.67057443e-01 -1.16238999e+00 3.69051903e-01 3.42198491e-01 8.75112116e-02 1.10733593e+00 -1.52191937e+00 4.67548519e-01 -1.49562091e-01 -2.83673793e-01 -1.15820062e+00 -4.64438051e-01 -7.05796242e-01 -5.74354529e-01 8.72959554e-01 7.24506319e-01 -2.45906442e-01 6.63062155e-01 2.89728165e-01 1.31752491e-01 -6.53324008e-01 -1.17593348e+00 -8.59781206e-01 6.91208337e-03 -3.04897517e-01 5.56302488e-01 1.64851654e+00 4.11594540e-01 8.12140346e-01 -5.26914485e-02 -3.35354246e-02 3.54498059e-01 6.17291071e-02 2.57113874e-01 -1.74552894e+00 -1.79957673e-01 -4.24269766e-01 -7.73301065e-01 -6.37004733e-01 6.32347822e-01 -1.35154605e+00 -4.03472751e-01 -1.59126651e+00 4.02092814e-01 -2.52501279e-01 -4.99378055e-01 1.12542462e+00 -1.98375031e-01 2.66705006e-01 1.34778664e-01 -6.43590465e-02 -1.18589687e+00 7.31311142e-01 3.58305544e-01 -1.90760523e-01 3.32625002e-01 -5.44055820e-01 -8.64376426e-01 5.73978364e-01 2.09341317e-01 -6.36620581e-01 -1.25894815e-01 -5.50087750e-01 9.06684101e-01 -6.03607833e-01 3.46564978e-01 -5.12035489e-01 6.09931707e-01 2.51287103e-01 1.20184898e-01 -2.11681053e-01 2.45301217e-01 -7.63442338e-01 -2.25720942e-01 1.45994984e-02 -3.71329546e-01 -4.76029143e-02 1.65292159e-01 4.40538824e-01 -4.71536696e-01 -4.63702857e-01 2.96963781e-01 -1.97978511e-01 -9.59667921e-01 3.02334696e-01 -7.80686736e-02 8.47059906e-01 7.01637089e-01 4.10726927e-02 -3.59834164e-01 -7.49830306e-02 -1.24978220e+00 4.40582454e-01 -7.08810892e-03 3.05106550e-01 5.37413776e-01 -1.65447974e+00 -8.86481881e-01 -1.74643502e-01 5.41895032e-01 -2.57411189e-02 -4.18216083e-03 4.12816197e-01 -2.85723180e-01 6.18799865e-01 -2.43378561e-02 1.99232385e-01 -1.04250550e+00 3.52414668e-01 -1.30426452e-01 -6.42365158e-01 -5.59839427e-01 1.53002954e+00 -5.12280107e-01 -5.06986320e-01 4.10011858e-01 -1.16991932e-02 -7.51495004e-01 5.64442217e-01 1.74928308e-01 4.36639816e-01 1.31341085e-01 -6.59203172e-01 -8.02914262e-01 3.01963501e-02 4.48194593e-02 -1.46015599e-01 1.55087256e+00 4.11917686e-01 -4.24945116e-01 2.71777630e-01 9.46460962e-01 -9.57453717e-03 -1.22602005e-02 -3.66767645e-01 6.13060176e-01 -6.56153783e-02 -4.38937068e-01 -9.52220976e-01 -6.64417326e-01 4.42778856e-01 1.25700250e-01 1.02190450e-01 4.89491999e-01 2.63203830e-01 8.54134858e-01 9.70001876e-01 5.78581393e-01 -1.29330266e+00 -1.77027881e-01 1.14543223e+00 6.56287134e-01 -1.00295699e+00 2.63035417e-01 -4.08989817e-01 -8.16506565e-01 9.75355506e-01 6.38066232e-01 -1.14111647e-01 1.12346482e+00 3.45761552e-02 -5.97489834e-01 -8.94027591e-01 -7.46442020e-01 -9.84855533e-01 5.64400673e-01 6.11557961e-01 4.70983952e-01 3.42699498e-01 -6.02322042e-01 1.41131568e+00 -3.69554698e-01 -3.93478096e-01 1.39030263e-01 6.48710728e-01 -1.30772546e-01 -1.05883574e+00 4.38676566e-01 5.21028161e-01 -2.21456051e-01 -9.29823458e-01 -8.09704006e-01 8.57556701e-01 7.61585236e-01 6.03037834e-01 -4.69540469e-02 -4.26034093e-01 3.88113886e-01 1.22458674e-01 1.58058330e-01 -1.38977861e+00 -7.85960913e-01 -7.21299887e-01 6.01134002e-01 -7.76836753e-01 -2.58550704e-01 -6.68832242e-01 -1.49325955e+00 1.01946846e-01 -3.41630459e-01 5.32174528e-01 2.88517267e-01 1.00569582e+00 3.84222627e-01 2.89162189e-01 7.54682347e-02 1.89050898e-01 -2.41697311e-01 -1.23157227e+00 -1.08989644e+00 7.96323299e-01 -5.71873367e-01 -7.18112826e-01 -4.41820741e-01 -2.14897886e-01]
[9.80933952331543, 8.708108901977539]
10b8c8d3-4918-4f3a-89b2-831b98d8fc32
fence-gan-towards-better-anomaly-detection
1904.01209
null
http://arxiv.org/abs/1904.01209v1
http://arxiv.org/pdf/1904.01209v1.pdf
Fence GAN: Towards Better Anomaly Detection
Anomaly detection is a classical problem where the aim is to detect anomalous data that do not belong to the normal data distribution. Current state-of-the-art methods for anomaly detection on complex high-dimensional data are based on the generative adversarial network (GAN). However, the traditional GAN loss is not directly aligned with the anomaly detection objective: it encourages the distribution of the generated samples to overlap with the real data and so the resulting discriminator has been found to be ineffective as an anomaly detector. In this paper, we propose simple modifications to the GAN loss such that the generated samples lie at the boundary of the real data distribution. With our modified GAN loss, our anomaly detection method, called Fence GAN (FGAN), directly uses the discriminator score as an anomaly threshold. Our experimental results using the MNIST, CIFAR10 and KDD99 datasets show that Fence GAN yields the best anomaly classification accuracy compared to state-of-the-art methods.
['Farhan Akram', 'Connie Kou Khor Li', 'Sojeong Park', 'Amadeus Aristo Winarto', 'Hwee Kuan Lee', 'Cuong Phuc Ngo']
2019-04-02
null
null
null
null
['anomaly-classification']
['computer-vision']
[ 1.58117294e-01 -2.27148551e-02 3.54059905e-01 -3.18532050e-01 -4.22170520e-01 -4.09242064e-01 4.72397149e-01 1.63058028e-01 -2.35320017e-01 7.21798182e-01 -3.75622392e-01 -2.00007379e-01 7.72835612e-02 -9.56271231e-01 -6.38434172e-01 -8.72882485e-01 -6.31813854e-02 6.01021409e-01 2.29297891e-01 -1.44563958e-01 2.11408913e-01 5.00670791e-01 -1.37230587e+00 8.05416703e-02 1.22443545e+00 1.21947932e+00 -1.00753748e+00 5.91848731e-01 -2.80174404e-01 5.37040293e-01 -1.08217525e+00 -3.87747735e-01 4.86980736e-01 -9.73439097e-01 -3.33639860e-01 -1.99089468e-01 5.41701376e-01 -3.36600363e-01 -1.46145001e-01 1.23580956e+00 3.77421081e-01 1.47186331e-02 7.56973803e-01 -1.78066516e+00 -5.26649952e-01 1.61273167e-01 -6.08574033e-01 5.06527483e-01 1.37431204e-01 -5.53721711e-02 8.44249785e-01 -6.37458503e-01 6.77432120e-02 9.94225919e-01 6.50401056e-01 7.22730696e-01 -1.37405229e+00 -7.63093174e-01 1.54098436e-01 9.60236937e-02 -1.19033217e+00 4.46329452e-02 9.60081756e-01 -3.84291679e-01 4.34983760e-01 2.99223959e-01 5.17143309e-01 1.32287633e+00 2.90486485e-01 6.62389576e-01 8.22657883e-01 -3.18759888e-01 5.81668615e-01 -2.95633525e-01 -1.10574864e-01 4.66646284e-01 3.60373110e-01 2.27853164e-01 -1.63420901e-01 -5.07508993e-01 5.46849549e-01 2.29886040e-01 -1.74147949e-01 -4.48914200e-01 -6.31975234e-01 1.12709141e+00 3.21234554e-01 4.00051355e-01 -4.27064508e-01 2.80694887e-02 5.46113610e-01 6.00509584e-01 7.44281769e-01 4.75226521e-01 -5.63835874e-02 -1.93352014e-01 -7.27039278e-01 4.11006689e-01 5.27522981e-01 3.31202835e-01 4.30768341e-01 5.35220325e-01 -2.98806190e-01 5.42410135e-01 2.60706007e-01 3.66418272e-01 6.58126295e-01 -4.43744838e-01 2.23276034e-01 9.65001643e-01 -1.72723398e-01 -8.48427951e-01 -4.58264165e-02 -5.72889090e-01 -9.73463297e-01 6.51745379e-01 6.52602792e-01 -2.71376729e-01 -1.12377357e+00 1.70805967e+00 2.75131792e-01 3.84863764e-01 2.48285457e-01 6.43690467e-01 2.30910093e-01 3.68882596e-01 -7.11318329e-02 1.79061711e-01 5.07417500e-01 -4.51085091e-01 -6.11141980e-01 -2.59885699e-01 6.48049831e-01 -3.89786988e-01 1.22679019e+00 4.01734859e-01 -7.35271573e-01 -2.85057694e-01 -1.20891631e+00 5.63422740e-01 -4.36380804e-01 -3.52779448e-01 4.06209737e-01 7.49839604e-01 -6.01383209e-01 7.11951852e-01 -9.56687748e-01 -1.74328029e-01 6.70845091e-01 1.04411036e-01 -3.83295417e-01 1.22229747e-01 -9.61534500e-01 5.65588236e-01 4.13243830e-01 -1.69385076e-01 -6.92354083e-01 -6.91317081e-01 -6.89436853e-01 -1.90789476e-02 2.10319474e-01 -2.79098779e-01 9.58836079e-01 -1.37961173e+00 -1.21974599e+00 7.74236262e-01 3.29824567e-01 -8.76116097e-01 9.11139965e-01 -3.59587789e-01 -6.84907436e-01 -1.52313650e-01 -8.77012312e-02 1.27517506e-01 1.09286451e+00 -1.05297649e+00 -5.50940335e-01 -5.23163736e-01 -5.04572988e-01 -2.30730817e-01 -9.63943303e-02 -3.40325296e-01 1.81654334e-01 -1.01452065e+00 2.25105226e-01 -7.72312403e-01 -1.76537074e-02 -1.09179020e-01 -6.63703620e-01 -1.16093814e-01 1.39696419e+00 -4.15661365e-01 1.18527246e+00 -2.44348288e+00 -3.57605278e-01 8.03751826e-01 1.04212165e-01 5.44726789e-01 1.53570101e-01 1.38825521e-01 -2.81886607e-01 1.72994584e-02 -7.93190658e-01 -2.12735012e-01 -6.99765980e-02 4.40794051e-01 -6.30622268e-01 7.39131272e-01 3.26785892e-01 5.71379185e-01 -8.17925930e-01 6.97979182e-02 4.78141308e-02 2.63539165e-01 -6.57967448e-01 5.77803850e-01 -1.93463504e-01 7.15025783e-01 -3.83673042e-01 5.08301139e-01 7.42925465e-01 2.12547794e-01 -4.03037071e-01 4.47601497e-01 3.33024532e-01 -4.68464456e-02 -9.10752654e-01 1.30999815e+00 1.49589360e-01 4.86443251e-01 -3.97118986e-01 -1.14453065e+00 1.26633513e+00 1.33963853e-01 5.56646228e-01 -7.44145989e-01 -5.37488125e-02 4.24462318e-01 4.28196639e-01 -7.47116059e-02 1.21985316e-01 -8.88565853e-02 -2.75070015e-02 3.91662717e-01 -3.59262377e-02 2.65298635e-01 8.76472965e-02 -6.37355670e-02 1.29524088e+00 -1.59769475e-01 9.81216356e-02 -1.49602041e-01 5.68963766e-01 -3.04096371e-01 7.58938611e-01 7.94521451e-01 -1.79926008e-01 8.31790566e-01 1.03309703e+00 -5.42530000e-01 -1.14701605e+00 -1.47233415e+00 -1.34406418e-01 6.17042184e-01 -3.12805593e-01 -1.25381202e-01 -1.01098430e+00 -1.35268450e+00 1.63035735e-01 1.03126144e+00 -7.28407204e-01 -5.62522948e-01 -5.00072122e-01 -9.28440034e-01 8.59675109e-01 5.87917387e-01 7.17865050e-01 -1.07827580e+00 -5.65186083e-01 1.32441863e-01 1.30060092e-01 -8.43158662e-01 -2.82671005e-01 4.63444591e-02 -8.19005311e-01 -1.23910487e+00 -3.92693937e-01 -1.43742353e-01 9.70529199e-01 -6.55156255e-01 1.01518750e+00 1.05752550e-01 -2.35790521e-01 1.50649950e-01 -4.90221679e-01 -7.94771731e-01 -7.22017169e-01 -2.23594174e-01 5.06803133e-02 4.52008128e-01 6.11880600e-01 -7.11167455e-01 -5.29883504e-01 3.60587031e-01 -1.18018985e+00 -6.95112228e-01 2.56434023e-01 9.12791193e-01 6.98495984e-01 1.85588107e-01 9.32455540e-01 -1.20574796e+00 6.67345822e-01 -5.74760377e-01 -6.40386581e-01 -1.81985632e-01 -9.17015970e-01 6.08595945e-02 9.27068830e-01 -4.77994919e-01 -6.58724844e-01 -2.67454058e-01 -4.83134478e-01 -6.56786263e-01 -5.24588287e-01 5.96095510e-02 -2.84167349e-01 5.60031198e-02 7.75568187e-01 2.36232981e-01 2.01614439e-01 -5.27721643e-01 -1.56345904e-01 3.67821544e-01 6.66324615e-01 -2.35079169e-01 8.78519416e-01 4.81781781e-01 1.33893400e-01 -6.65901065e-01 -6.15236282e-01 -1.69782594e-01 -4.35164154e-01 6.21494129e-02 6.73496544e-01 -3.61235887e-01 -2.25315690e-01 8.65196943e-01 -7.51672387e-01 -2.37797320e-01 -9.03599441e-01 3.13408464e-01 -4.66382980e-01 3.07962686e-01 -2.03642145e-01 -8.94801378e-01 -5.44107199e-01 -7.58727551e-01 7.16794252e-01 5.76626584e-02 -2.11467132e-01 -9.37875032e-01 3.21974725e-01 -2.36615524e-01 6.16115153e-01 1.03292596e+00 1.19196165e+00 -1.44243526e+00 3.65257263e-03 -8.14006329e-01 1.62260011e-01 9.48238790e-01 2.69352287e-01 6.49200231e-02 -1.02640498e+00 -3.80458862e-01 1.06601305e-01 -5.92891639e-03 6.80030107e-01 1.94180623e-01 1.70758235e+00 -4.32444751e-01 8.23535025e-02 7.74720371e-01 1.28216052e+00 4.49982256e-01 1.01483393e+00 2.42322028e-01 6.90315604e-01 1.70360953e-01 3.55460733e-01 3.74432981e-01 -2.89835483e-01 4.10309464e-01 7.97343016e-01 -1.58315644e-01 4.49518025e-01 -3.73144418e-01 4.36752945e-01 1.64259329e-01 1.11633331e-01 -4.66686755e-01 -9.57423866e-01 3.43162566e-01 -1.88413835e+00 -1.05021846e+00 -1.73997821e-03 2.62508583e+00 2.86782533e-01 5.92376530e-01 3.92759174e-01 6.25534832e-01 5.73744178e-01 -2.56320890e-02 -8.89517784e-01 -5.58236599e-01 -1.47330195e-01 3.68310183e-01 3.14463645e-01 1.31128179e-02 -1.25795209e+00 5.47992349e-01 6.13211918e+00 5.80421865e-01 -1.12523627e+00 -1.28201902e-01 7.33645320e-01 3.24416906e-02 -1.08927131e-01 -3.19479942e-01 -3.37993592e-01 8.79464209e-01 8.45519960e-01 1.00686848e-02 8.12398270e-02 8.92331660e-01 -1.96712941e-01 1.26245484e-01 -1.31393969e+00 8.10970306e-01 1.50723726e-01 -6.32148147e-01 1.38699472e-01 2.88740069e-01 6.19307876e-01 -1.05775900e-01 2.14999393e-01 3.69761586e-01 1.10756882e-01 -1.18259108e+00 3.07824433e-01 5.33437908e-01 5.21340668e-01 -1.13875008e+00 1.11776567e+00 2.20391721e-01 -5.56054413e-01 3.35910101e-03 -2.40291104e-01 2.36408427e-01 -8.50678235e-02 8.36232781e-01 -8.22155952e-01 1.70956612e-01 6.54909670e-01 4.31355774e-01 -6.14479601e-01 9.97975111e-01 -8.21363553e-02 8.23328972e-01 -3.02066088e-01 5.26095927e-01 3.62820268e-01 -3.48789454e-01 1.07254422e+00 7.11378753e-01 5.12981772e-01 -4.33885843e-01 1.30480260e-01 1.02777267e+00 -1.61220193e-01 1.28462523e-01 -8.83795500e-01 -1.16656050e-01 8.98233354e-02 6.79874301e-01 -4.76033032e-01 -6.02624286e-03 -3.31915230e-01 1.11025321e+00 6.62412941e-02 2.74680972e-01 -7.08415270e-01 -5.24637282e-01 9.41205978e-01 2.65442878e-01 2.20846429e-01 2.35008508e-01 -2.17148364e-01 -1.02415955e+00 3.12203646e-01 -1.05963027e+00 8.21465135e-01 -1.83639109e-01 -1.82900882e+00 6.18991256e-01 -2.46716037e-01 -1.44853914e+00 -6.41299009e-01 -5.79079628e-01 -1.16897237e+00 7.94657409e-01 -1.03293085e+00 -7.19700933e-01 -3.33561718e-01 7.76758254e-01 2.28507414e-01 -6.19882405e-01 9.49104488e-01 2.91979760e-01 -5.13702691e-01 9.56066430e-01 2.45668843e-01 6.04267359e-01 4.56867933e-01 -1.62469697e+00 6.47283971e-01 1.19954205e+00 1.38508096e-01 1.51305841e-02 7.08340108e-01 -7.70480812e-01 -5.97754419e-01 -1.29858840e+00 1.41165495e-01 -3.60630900e-01 5.18366694e-01 -3.00275534e-01 -1.46614945e+00 7.89490104e-01 -7.23612979e-02 4.57552969e-01 7.90205479e-01 -2.32240394e-01 -3.20582241e-01 -2.85421908e-01 -1.79663181e+00 3.74130875e-01 6.61729813e-01 -2.08345503e-01 -4.05425638e-01 1.40653374e-02 2.77104825e-01 -3.82634491e-01 -4.78812665e-01 6.76045418e-01 1.49293751e-01 -1.12454307e+00 7.38674164e-01 -9.50204968e-01 2.05670938e-01 -2.86673039e-01 -1.30196497e-01 -1.65738630e+00 5.18894680e-02 -2.80018538e-01 -5.78724980e-01 1.22875643e+00 1.97515413e-01 -1.07370031e+00 8.32723975e-01 2.54993320e-01 7.64354132e-03 -8.16033959e-01 -1.13989055e+00 -8.74923408e-01 1.57792479e-01 -3.32490981e-01 8.33904684e-01 8.88883829e-01 -5.52522659e-01 -3.14881615e-02 -2.69160092e-01 1.57966480e-01 8.05187583e-01 -1.99887529e-01 7.74981916e-01 -1.56869829e+00 -2.19102889e-01 -5.51875234e-01 -1.08619082e+00 -2.46922359e-01 1.64329544e-01 -6.29411638e-01 -1.01390734e-01 -8.74580562e-01 -4.49286610e-01 -5.53206325e-01 -5.62838137e-01 4.09322798e-01 -9.52055156e-02 3.53723377e-01 -2.10685655e-01 5.91323478e-03 -2.04772294e-01 6.87237442e-01 7.31922090e-01 -5.60523272e-02 -2.98603773e-01 5.00570297e-01 -4.98709321e-01 8.83214891e-01 1.00934517e+00 -6.12819791e-01 -3.63372445e-01 2.09451746e-02 -2.65252274e-02 -5.87336659e-01 4.91319805e-01 -1.43815756e+00 -2.65661538e-01 7.07866699e-02 6.52504861e-01 -5.64800680e-01 -7.07194814e-03 -9.60634649e-01 -7.50870407e-02 5.39566219e-01 -2.78819919e-01 5.87603986e-01 9.96683016e-02 7.11092055e-01 -3.75104040e-01 -4.12525274e-02 1.03408551e+00 2.02003017e-01 -2.78909326e-01 4.48933333e-01 -1.58198580e-01 5.59468329e-01 1.25302911e+00 3.34615335e-02 -3.25223595e-01 -3.93022478e-01 -4.42406505e-01 9.08149853e-02 5.32283306e-01 4.19023663e-01 6.94679141e-01 -1.44887805e+00 -8.41366589e-01 7.56583869e-01 2.56539285e-01 1.89212918e-01 1.81654729e-02 5.61883569e-01 -5.88996291e-01 -1.97348207e-01 -4.05246109e-01 -7.80459583e-01 -9.54116106e-01 4.62894142e-01 7.08226204e-01 -2.85684258e-01 -8.82988155e-01 5.15519977e-01 1.09479301e-01 -3.45630169e-01 2.31766701e-01 -1.74246579e-01 -4.08255309e-02 -3.05677652e-01 4.64508891e-01 5.48632085e-01 2.52968192e-01 -4.37681109e-01 -2.62916446e-01 9.28262025e-02 -7.95952752e-02 9.41552669e-02 1.20830703e+00 4.38310862e-01 8.18679109e-03 5.83743632e-01 1.03708863e+00 -4.12590466e-02 -1.08443987e+00 -1.80757806e-01 4.72745858e-02 -7.20605612e-01 -1.45535171e-01 -7.21533179e-01 -1.34262872e+00 7.35711157e-01 1.11309135e+00 6.43743932e-01 1.30487096e+00 -3.14682424e-01 7.99165964e-01 1.14204518e-01 -7.18576983e-02 -1.08239782e+00 1.81853488e-01 4.74840105e-01 7.29818046e-01 -1.23542774e+00 -4.34470773e-01 2.07908284e-02 -6.63206756e-01 8.90310764e-01 1.07899535e+00 -5.14898002e-01 4.97351646e-01 1.89487100e-01 1.85732365e-01 -1.32981896e-01 -2.47833312e-01 1.69206649e-01 3.69087100e-01 5.51726162e-01 1.09130614e-01 3.01172305e-02 -1.12311535e-01 3.84837896e-01 -2.37153500e-01 -4.74525601e-01 4.24007446e-01 8.44037712e-01 -9.86935869e-02 -1.07831669e+00 -3.92127961e-01 9.68423605e-01 -9.66881156e-01 1.64650366e-01 -4.77497816e-01 6.44793689e-01 1.10649511e-01 5.16098917e-01 7.14116395e-01 -1.95704222e-01 5.07760823e-01 5.66261888e-01 -1.57741681e-02 -2.78501809e-01 -3.99265617e-01 -3.26320350e-01 -5.05863965e-01 -6.42360806e-01 2.86696218e-02 -4.76830900e-01 -1.31160712e+00 -3.03291619e-01 -3.23787592e-02 3.25086355e-01 4.79324102e-01 1.00888014e+00 4.20336783e-01 6.55100048e-01 7.34656572e-01 -1.59600303e-01 -6.40302300e-01 -9.51217413e-01 -6.89273775e-01 8.90605211e-01 6.57993317e-01 -6.20408416e-01 -6.87102914e-01 -4.66648400e-01]
[7.59529447555542, 2.361971139907837]
6ba71002-ff32-45ed-9eb1-912f65621356
study-on-the-concept-and-development-of-a
2208.09697
null
https://arxiv.org/abs/2208.09697v1
https://arxiv.org/pdf/2208.09697v1.pdf
Study on the Concept and Development of a Mobile Incubator
Creating the best possible conditions is essential for proper cell growth. Incubators, a type of biotechnological instrument, are used to simulate this condition and maintain the cells within them. The processes involved in creating a mobile incubator, which are essential for monitoring a cell culture's physiological parameters, are outlined in this article. The goal is to keep image-taking during cell development from compromising data accuracy. The cell culture is prone to contamination once it has been removed from the incubation environment for further monitoring. The proposed approach allows for on-the-go monitoring of the cell culture. Moreover, it enables constant monitoring.
['Huseyin Uvet', 'Abdurrahim Yilmaz', 'Ufuk Gorkem Kirabali', 'Atasangu Yilmaz', 'Rahmetullah Varol', 'Nesim Bilici', 'Fehmi Can Ay']
2022-08-20
null
null
null
null
['culture']
['speech']
[ 3.84896100e-02 -3.13310832e-01 6.47955984e-02 4.72449452e-01 2.98644602e-01 -4.65629369e-01 9.34817195e-02 7.88392961e-01 -5.20814776e-01 8.84060681e-01 -5.24104357e-01 -2.85460770e-01 5.94199955e-01 -7.08206773e-01 -4.27709848e-01 -1.12197721e+00 2.45244727e-01 8.22508708e-02 2.80019641e-01 1.81003228e-01 4.83193427e-01 1.06987476e+00 -1.48398852e+00 -3.41621161e-01 3.58885378e-01 6.61666274e-01 7.11255133e-01 9.92304504e-01 -8.79007950e-02 6.44876659e-01 -6.36274159e-01 2.27254689e-01 -8.32251012e-02 -4.34504449e-01 -2.01329634e-01 3.01735401e-01 -6.39330387e-01 -1.81164682e-01 3.28839958e-01 6.14828587e-01 3.41707408e-01 -1.16066515e-01 8.18388283e-01 -7.75829792e-01 -7.47366808e-03 -3.89141470e-01 -2.07222059e-01 1.61721960e-01 3.48287731e-01 2.32877776e-01 -2.38548070e-01 -9.07060146e-01 7.11441159e-01 4.38832819e-01 2.42351681e-01 2.89272457e-01 -1.10567272e+00 -2.68978447e-01 -3.36118966e-01 -8.90576318e-02 -1.54878128e+00 -7.74037778e-01 5.46101034e-01 -6.48159921e-01 5.71652710e-01 2.02554196e-01 1.04937673e+00 6.00530505e-01 1.02845156e+00 -2.08529711e-01 1.07304406e+00 -8.09182882e-01 5.54910421e-01 6.12346590e-01 -1.61711544e-01 4.82515395e-01 8.08698833e-01 -3.11381593e-02 -2.72614956e-01 1.98897302e-01 1.06699681e+00 -1.89660490e-02 -5.25969505e-01 -3.26254994e-01 -7.08115220e-01 -3.12445499e-02 -2.86751151e-01 9.94298875e-01 -5.98461926e-01 -6.32097423e-02 2.29258716e-01 9.12491009e-02 9.99662355e-02 5.58837712e-01 -3.58408630e-01 -2.28955001e-01 -6.64595485e-01 -1.94553524e-01 9.05065119e-01 7.71916449e-01 2.48971850e-01 -2.42749974e-01 1.04127534e-01 4.64640349e-01 2.40005150e-01 2.48122960e-01 4.48965818e-01 -6.71477675e-01 -3.38027954e-01 6.99713767e-01 4.32092518e-01 -9.02946174e-01 -2.79266655e-01 -5.32010913e-01 -5.99469066e-01 3.43045503e-01 8.03781092e-01 -7.91097432e-02 -6.96814299e-01 1.08673275e+00 7.23111928e-01 3.96378249e-01 2.85698920e-01 4.61292684e-01 6.57264292e-01 6.02265477e-01 1.09005244e-02 -1.02292311e+00 1.35904050e+00 -2.12962598e-01 -1.27558494e+00 4.83024180e-01 7.63522029e-01 -8.31933916e-01 7.29322851e-01 4.27388191e-01 -1.04675257e+00 -4.32961524e-01 -1.17387056e+00 3.09185743e-01 -7.28842318e-01 -8.56972337e-02 1.80741981e-01 6.55562639e-01 -8.09662580e-01 4.80941176e-01 -8.17795157e-01 -6.44688964e-01 2.13759113e-02 2.14415997e-01 -4.55377638e-01 1.84875712e-01 -6.61469817e-01 1.04570568e+00 3.03492039e-01 3.91116023e-01 -7.42165029e-01 -1.04527518e-01 -8.54681790e-01 -1.72053706e-02 -1.65361702e-01 -5.20531297e-01 7.77591825e-01 -1.70916900e-01 -2.15544462e+00 1.00378096e+00 -6.27733171e-01 9.86483470e-02 4.43508327e-01 2.99721938e-02 -2.21435174e-01 3.03721011e-01 -1.63928226e-01 -1.04119748e-01 3.45552117e-01 -1.34600663e+00 -3.65842968e-01 -1.95742041e-01 -4.42104578e-01 1.40651807e-01 9.60879251e-02 2.72488147e-01 -6.09748662e-01 -2.01884985e-01 1.45476133e-01 -7.99494982e-01 -2.69955575e-01 -4.04528260e-01 -1.52995691e-01 2.48974442e-01 9.10079300e-01 -4.89095926e-01 8.34139526e-01 -1.98047352e+00 -3.84223580e-01 2.13528201e-01 -1.61927789e-01 3.86480957e-01 4.78099555e-01 6.76560283e-01 1.09550945e-01 2.76523773e-02 3.07272613e-01 -2.91695386e-01 -6.69030011e-01 -1.70016110e-01 7.12337732e-01 9.03121889e-01 -1.85863957e-01 5.07564068e-01 -5.66301286e-01 -6.10086799e-01 4.85121667e-01 5.10093212e-01 1.70548230e-01 4.53505516e-01 1.84840977e-01 1.02408028e+00 -1.18410356e-01 7.95859158e-01 5.35774529e-01 4.27153334e-02 1.94344878e-01 1.58069152e-02 -6.32133186e-01 -2.59579539e-01 -9.66795444e-01 1.06771469e+00 -2.09006920e-01 4.15496558e-01 4.91192847e-01 -5.41284204e-01 1.13687873e+00 7.68585861e-01 7.60805011e-01 -5.08586407e-01 7.44439602e-01 5.95965683e-01 -2.71218628e-01 -9.40887451e-01 2.57720470e-01 -3.76337469e-01 3.60443354e-01 1.37865618e-01 -5.61815917e-01 -2.36819282e-01 5.20453572e-01 -1.25398442e-01 4.92256314e-01 1.88556053e-02 6.47963524e-01 -5.40986001e-01 9.32827652e-01 -2.04860494e-01 5.53724349e-01 1.28723204e-01 -3.92819166e-01 2.45323822e-01 3.52152348e-01 -2.02043146e-01 -9.06143665e-01 -2.05577254e-01 -3.27705562e-01 2.01676607e-01 5.42472899e-01 1.47485569e-01 -9.45466757e-01 2.38499269e-01 -1.70527771e-01 5.45064211e-01 -5.28688014e-01 1.91619292e-01 -3.73057634e-01 -4.41041499e-01 4.67149429e-02 7.24156797e-02 9.88130644e-02 -7.56824195e-01 -9.69832361e-01 6.68682992e-01 -9.20733251e-03 -8.47075820e-01 7.43278477e-04 1.13519669e-01 -8.49802732e-01 -1.42036974e+00 -7.38897741e-01 -7.07055867e-01 8.07796836e-01 3.08754534e-01 5.84409416e-01 4.86114085e-01 -8.82587209e-02 2.26963133e-01 -2.77486533e-01 -7.79516995e-01 -8.07544768e-01 -2.96718895e-01 -4.37848158e-02 -5.15697971e-02 4.24884409e-01 -5.39920390e-01 -7.62304246e-01 5.95859230e-01 -7.19619930e-01 -5.68297505e-02 9.79992375e-02 3.22959721e-01 9.68229890e-01 1.20268464e-01 6.42359614e-01 -7.23431528e-01 4.05014336e-01 -2.88177371e-01 -7.09576249e-01 1.34074986e-01 -4.62585092e-01 -5.91419101e-01 6.75593376e-01 -2.13604689e-01 -7.94574022e-01 -1.43507197e-01 4.71793711e-02 -4.54666140e-03 -6.79998696e-01 4.26854789e-01 -3.28398526e-01 -1.53667852e-01 3.12887013e-01 2.92182624e-01 5.97342432e-01 -3.62887442e-01 -6.08204544e-01 7.01727509e-01 4.39791620e-01 -3.74489836e-02 3.29829454e-01 4.54889417e-01 5.41667581e-01 -9.36804831e-01 -1.70876384e-01 -8.24757397e-01 -5.09286880e-01 -7.56841540e-01 7.57599294e-01 -4.74950850e-01 -8.74271750e-01 9.19441521e-01 -1.06975460e+00 -3.08017701e-01 3.89669044e-03 8.26665342e-01 -3.42246532e-01 2.60182947e-01 -7.83588529e-01 -1.26392412e+00 -1.37881830e-01 -1.19062674e+00 4.22810674e-01 5.80962956e-01 -8.38523805e-02 -1.11039448e+00 2.89145440e-01 -9.50489342e-02 5.47509730e-01 7.26721585e-01 3.66802484e-01 -1.99876279e-01 -3.95819515e-01 -5.82190871e-01 4.79784369e-01 -1.48648128e-01 6.14708245e-01 6.58395588e-01 -1.11194289e+00 -2.48379812e-01 5.71612179e-01 3.44412774e-01 2.75034964e-01 8.85657847e-01 7.96366572e-01 -2.67566498e-02 -8.42666388e-01 6.88124001e-01 1.64899254e+00 9.33639348e-01 8.83915842e-01 5.41316867e-01 3.90541315e-01 7.08850563e-01 9.75656927e-01 2.73672968e-01 -1.08806863e-01 3.61844242e-01 7.03966439e-01 -4.40118641e-01 2.34442681e-01 7.38180727e-02 1.16618024e-02 6.65486455e-01 -2.32558414e-01 -4.34418648e-01 -5.94550133e-01 3.74247044e-01 -1.06302631e+00 -6.84887588e-01 -7.14006364e-01 2.44223332e+00 7.27847695e-01 -1.22507185e-01 1.34403437e-01 3.58107924e-01 9.15586174e-01 -8.07048917e-01 -2.08293393e-01 -6.03497326e-01 6.22589774e-02 -8.25594142e-02 5.99402606e-01 6.50925040e-01 -7.54447401e-01 4.52385187e-01 7.19897842e+00 -6.79849647e-03 -1.61425269e+00 -3.11665714e-01 6.21752918e-01 -3.69759500e-02 3.95758599e-02 5.73160946e-02 -6.28598869e-01 7.10681558e-01 9.11079586e-01 -1.03064373e-01 -1.24319740e-01 3.99063647e-01 9.72731948e-01 -9.49800134e-01 -1.01898599e+00 6.57180667e-01 -3.25309068e-01 -1.23643148e+00 -5.15790761e-01 5.09718657e-01 2.12298408e-01 -8.06046247e-01 -5.39346099e-01 -3.35457653e-01 -4.08613712e-01 -7.67265141e-01 5.86582661e-01 6.37453794e-01 9.73224044e-01 -8.98745477e-01 1.21887541e+00 4.80664551e-01 -9.62432683e-01 3.46279711e-01 -1.28666028e-01 -2.02691942e-01 5.29210508e-01 8.97326648e-01 -1.28155756e+00 4.08040822e-01 2.29733393e-01 2.53678292e-01 -2.51390159e-01 1.29998028e+00 3.04646879e-01 3.04856092e-01 -3.80310923e-01 -9.93739143e-02 -2.75785834e-01 -6.25340760e-01 3.83473128e-01 1.11926365e+00 5.94973624e-01 1.85977057e-01 -2.03664333e-01 5.36236823e-01 3.77499670e-01 5.39089143e-01 -9.79066312e-01 3.36401388e-02 6.30998373e-01 7.96164334e-01 -1.24873090e+00 -3.04909855e-01 1.99516062e-02 5.56016684e-01 -1.82690755e-01 3.21539700e-01 -6.06191278e-01 -4.02311236e-01 3.58853191e-01 5.02731144e-01 -2.87451725e-02 -3.18929762e-01 -3.57478261e-01 -5.59840798e-01 -1.22276001e-01 -4.09969985e-01 2.23170906e-01 -3.87858123e-01 -2.34245732e-01 -6.33441731e-02 -1.66478053e-01 -1.25632417e+00 1.45498604e-01 -6.30009770e-01 -6.10432506e-01 1.02414060e+00 -1.42548728e+00 -6.85396850e-01 -5.31983018e-01 9.78836566e-02 3.57540578e-01 1.89565748e-01 9.68649745e-01 1.22695677e-01 -9.34878170e-01 9.58148018e-02 1.45824715e-01 -5.14383674e-01 4.49192554e-01 -8.40782106e-01 -5.75923681e-01 1.02874148e+00 -8.36690903e-01 6.71879947e-01 1.09181619e+00 -8.20490777e-01 -1.25917280e+00 -5.28642237e-01 8.53916526e-01 -4.87784371e-02 -1.09549411e-01 -2.61628628e-01 -1.02646387e+00 1.42332017e-01 1.40119731e-01 -4.83498752e-01 1.06494379e+00 -2.97281742e-01 5.85724354e-01 -2.26157177e-02 -1.31762719e+00 4.99146938e-01 3.16375375e-01 -1.84434682e-01 1.63288757e-01 1.91714317e-01 1.73648983e-01 -6.46104872e-01 -1.07111156e+00 1.97615758e-01 8.90525281e-01 -8.99301767e-01 3.39994282e-01 -1.75182924e-01 1.82101242e-02 -6.81504011e-01 3.84741664e-01 -9.44396436e-01 -3.47639531e-01 -8.12113881e-01 5.31194136e-02 1.20707357e+00 3.67522389e-01 -1.00067091e+00 6.24218047e-01 6.55102730e-01 3.72830518e-02 -7.16552734e-01 -5.82596004e-01 -8.97685647e-01 -2.36617893e-01 -3.38991545e-02 5.61968863e-01 7.91138351e-01 4.19746310e-01 -6.74625188e-02 -3.31319273e-02 2.39232793e-01 2.05231190e-01 -5.07683933e-01 1.00719464e+00 -1.03929770e+00 6.66438267e-02 -3.61867934e-01 -4.22140777e-01 -3.42556030e-01 -5.02793431e-01 -5.02216332e-02 1.55961424e-01 -1.51743484e+00 -1.07672855e-01 -3.67846608e-01 -2.68888623e-01 -4.06264722e-01 -2.46369645e-01 -1.02235585e-01 4.47162017e-02 3.84686887e-01 9.50357243e-02 -1.71014816e-02 1.59783959e+00 4.47201341e-01 -5.20292997e-01 -3.51664685e-02 -3.47875983e-01 3.95742625e-01 1.09575784e+00 -4.58908498e-01 -4.43414539e-01 1.99612245e-01 -1.20201811e-01 2.58460268e-02 -1.11930527e-01 -1.08357251e+00 3.49355847e-01 -3.59085500e-01 2.68750906e-01 -7.49578238e-01 1.89282730e-01 -1.30633688e+00 1.03526640e+00 8.95952940e-01 9.21017602e-02 1.75085645e-02 9.26718488e-03 3.15829337e-01 -9.84245464e-02 -5.35698831e-01 1.26210451e+00 -1.64899215e-01 -7.68375322e-02 -1.76625088e-01 -9.03995097e-01 -6.94815874e-01 1.75818825e+00 -1.14960325e+00 -2.29079589e-01 -8.66461173e-02 -8.29164684e-01 7.75825977e-02 1.37315667e+00 -5.99288404e-01 2.72514760e-01 -9.22768533e-01 -3.61751944e-01 2.27811396e-01 -1.25364169e-01 2.82938808e-01 3.36881995e-01 1.32659578e+00 -1.44927227e+00 3.85972708e-01 -2.51171052e-01 -5.88545024e-01 -1.60834324e+00 8.18091273e-01 5.72571933e-01 -3.51770282e-01 -5.83047867e-01 8.04290295e-01 1.07022859e-01 5.02079189e-01 -2.17708740e-02 -4.35683131e-01 -5.97526312e-01 -7.19063208e-02 7.09265172e-01 5.16566753e-01 1.19059078e-01 -3.13498706e-01 -3.06582719e-01 5.06362498e-01 2.49490857e-01 -1.58827364e-01 1.04867041e+00 -5.41244149e-01 -2.02775195e-01 9.22797978e-01 8.98773909e-01 3.63010138e-01 -1.21137488e+00 3.26382190e-01 -2.98863977e-01 -5.90228856e-01 -1.26913205e-01 -3.55224073e-01 -6.26890600e-01 4.81674284e-01 4.14022654e-01 5.71379244e-01 1.24264944e+00 -4.02422130e-01 3.90661448e-01 -1.54581517e-02 5.13173103e-01 -1.45322204e+00 -3.48000973e-01 3.33413929e-01 8.17196906e-01 -6.97356462e-01 -6.78913444e-02 -8.06829929e-01 -8.37812945e-02 1.18739092e+00 3.48495185e-01 1.60066485e-01 7.13449478e-01 6.16046011e-01 2.29274154e-01 -3.45331907e-01 -7.39948094e-01 5.97231537e-02 -5.78033090e-01 6.89771712e-01 7.64375269e-01 -5.87441325e-02 -6.43478096e-01 -6.15341263e-03 2.34340802e-01 5.59210181e-01 8.75595689e-01 1.27190816e+00 -7.34953761e-01 -7.93992698e-01 -8.63621235e-01 1.29298165e-01 -7.52472281e-01 6.23774290e-01 -2.43222803e-01 7.81957030e-01 2.98883859e-02 1.17995286e+00 -1.56610221e-01 1.89580917e-01 3.92155170e-01 2.76699007e-01 4.54606384e-01 -3.79435986e-01 -3.64151806e-01 4.88014996e-01 7.70502761e-02 -3.26750726e-01 -4.86978322e-01 -7.07494974e-01 -1.41176581e+00 -3.05724770e-01 -5.57803929e-01 3.61057281e-01 1.05911624e+00 8.07282507e-01 2.60712594e-01 5.59841633e-01 7.83604503e-01 -7.43010044e-01 3.92029166e-01 -9.09682095e-01 -1.04823017e+00 7.77376667e-02 3.44415337e-01 -7.48289168e-01 -4.58712965e-01 5.88625610e-01]
[13.87893295288086, -3.0372865200042725]
964716f1-790b-4a77-a02e-6db57b7a3f07
look-further-to-recognize-better-learning
1907.12924
null
https://arxiv.org/abs/1907.12924v1
https://arxiv.org/pdf/1907.12924v1.pdf
Look Further to Recognize Better: Learning Shared Topics and Category-Specific Dictionaries for Open-Ended 3D Object Recognition
Service robots are expected to operate effectively in human-centric environments for long periods of time. In such realistic scenarios, fine-grained object categorization is as important as basic-level object categorization. We tackle this problem by proposing an open-ended object recognition approach which concurrently learns both the object categories and the local features for encoding objects. In this work, each object is represented using a set of general latent visual topics and category-specific dictionaries. The general topics encode the common patterns of all categories, while the category-specific dictionary describes the content of each category in details. The proposed approach discovers both sets of general and specific representations in an unsupervised fashion and updates them incrementally using new object views. Experimental results show that our approach yields significant improvements over the previous state-of-the-art approaches concerning scalability and object classification performance. Moreover, our approach demonstrates the capability of learning from very few training examples in a real-world setting. Regarding computation time, the best result was obtained with a Bag-of-Words method followed by a variant of the Latent Dirichlet Allocation approach.
['S. Hamidreza Kasaei']
2019-07-26
null
null
null
null
['3d-object-recognition', 'object-categorization']
['computer-vision', 'computer-vision']
[-8.12638551e-02 1.16061959e-02 -2.17949778e-01 -5.92070580e-01 -3.63063633e-01 -3.47393364e-01 1.01984251e+00 4.77002472e-01 -4.27195340e-01 4.29031819e-01 -5.13785258e-02 2.69438535e-01 -2.45710894e-01 -7.66937315e-01 -5.52084684e-01 -9.50322986e-01 -2.64944166e-01 1.15529013e+00 4.82894897e-01 2.70202719e-02 5.27388990e-01 6.89778745e-01 -2.33765626e+00 3.25421333e-01 3.89738709e-01 1.18076992e+00 6.49657309e-01 3.29128504e-01 -4.27803755e-01 5.52503169e-01 -5.19042432e-01 2.26465408e-02 1.92554191e-01 1.49337202e-01 -7.60295212e-01 7.40495920e-01 3.09664249e-01 -2.70754755e-01 -7.55696073e-02 9.62891459e-01 1.51758045e-01 3.44152868e-01 9.07914102e-01 -1.50376844e+00 -5.15691280e-01 1.86058283e-01 -2.73076475e-01 1.19628251e-01 5.89987040e-02 6.67945221e-02 9.46134269e-01 -1.05157268e+00 5.86807311e-01 1.45338869e+00 2.25033417e-01 3.24696541e-01 -1.22281563e+00 -3.01890701e-01 6.71979487e-01 6.46031976e-01 -1.42280078e+00 -3.42904538e-01 5.50373912e-01 -6.98019385e-01 1.07722700e+00 -3.22796583e-01 5.99438548e-01 6.91233933e-01 2.14788616e-01 6.58508599e-01 9.14460599e-01 -4.92276281e-01 7.72501409e-01 4.37573940e-01 3.37109447e-01 5.75914919e-01 5.02204061e-01 -3.35200787e-01 -4.07355815e-01 -3.04325163e-01 5.77940762e-01 5.25159657e-01 1.59452617e-01 -1.22336495e+00 -1.26035607e+00 9.92799103e-01 2.72143006e-01 4.41314310e-01 -7.58282959e-01 8.21521580e-02 5.32080531e-01 1.85567111e-01 3.89188260e-01 6.21959791e-02 -3.89188111e-01 4.79510203e-02 -4.40558136e-01 1.59672320e-01 9.94487762e-01 1.28453910e+00 1.11406612e+00 -2.57090479e-01 7.95915499e-02 9.75811839e-01 4.58852351e-01 2.00014070e-01 7.18982100e-01 -7.89298773e-01 5.45228645e-02 6.34483695e-01 2.76130766e-01 -8.81329834e-01 -2.88367242e-01 -2.82189965e-01 -4.15185362e-01 1.59058064e-01 1.32071003e-01 4.82817233e-01 -1.10410571e+00 1.43755198e+00 5.47388136e-01 -2.47388169e-01 1.99967995e-01 7.25778162e-01 5.73477745e-01 7.00944006e-01 3.08793068e-01 -1.42179191e-01 1.61377883e+00 -1.12235665e+00 -6.56579494e-01 -3.65635514e-01 2.09655583e-01 -5.10167837e-01 8.39697599e-01 3.40710640e-01 -6.64370656e-01 -6.83070242e-01 -8.08108389e-01 1.37825042e-01 -6.08541608e-01 1.08678617e-01 7.25829661e-01 4.42256451e-01 -1.00876617e+00 1.06995940e-01 -8.86260211e-01 -1.00460005e+00 4.24820870e-01 3.42455566e-01 -5.14992833e-01 -4.02148902e-01 -5.25908947e-01 8.03816974e-01 7.60533750e-01 -3.44804436e-01 -1.44159269e+00 -2.57329643e-01 -8.28421593e-01 1.49737000e-01 3.62535506e-01 -3.65389854e-01 1.37153828e+00 -6.04568839e-01 -1.38246381e+00 9.01477814e-01 -2.03143924e-01 -3.50586027e-01 7.43129104e-02 2.04601996e-02 -7.69790187e-02 3.53005111e-01 2.33955428e-01 8.46585214e-01 9.67831790e-01 -1.67975485e+00 -1.08684373e+00 -5.69702685e-01 2.73361057e-01 3.75869393e-01 -5.94962597e-01 -1.72872782e-01 -5.73659480e-01 -3.49561185e-01 5.41116416e-01 -9.02624369e-01 -2.47160077e-01 2.19314694e-01 1.74117163e-01 -7.07955182e-01 8.98335397e-01 -1.40269592e-01 5.52768230e-01 -2.21982288e+00 1.80786937e-01 1.01683959e-01 1.52473733e-01 -2.37781644e-01 1.20436363e-01 6.35187685e-01 1.81151450e-01 -4.97776508e-01 -1.66876223e-02 -5.19213140e-01 2.62423337e-01 6.69756830e-01 -2.04802886e-01 6.63487196e-01 -1.17755041e-01 4.67330098e-01 -1.00866020e+00 -5.02127290e-01 2.73969203e-01 3.01925838e-01 -5.66830337e-01 3.60363632e-01 -2.83127397e-01 3.02056640e-01 -3.72754157e-01 6.29214585e-01 4.01202917e-01 -2.74833858e-01 4.38773841e-01 2.95086429e-02 -7.73475990e-02 4.80329506e-02 -1.28555787e+00 1.78345668e+00 -4.54668105e-01 3.49300236e-01 -3.06232423e-02 -1.42789114e+00 1.22248650e+00 3.11582476e-01 4.95696396e-01 -4.13843036e-01 -1.52215343e-02 1.59893826e-01 -3.09919059e-01 -4.68527675e-01 5.08630455e-01 -2.83814222e-01 -3.63068640e-01 4.83806223e-01 4.75940973e-01 -4.57405187e-02 3.06549817e-01 9.09571350e-02 8.70015562e-01 -1.67249829e-01 7.21616983e-01 -4.12534535e-01 2.93123066e-01 1.99102521e-01 2.83826292e-01 9.42091644e-01 -2.50076592e-01 2.66800523e-01 9.55300480e-02 -7.43686616e-01 -1.09247112e+00 -1.11122775e+00 -1.12635128e-01 1.45656371e+00 4.05944407e-01 -1.79677501e-01 -5.02442002e-01 -6.71428680e-01 1.61917329e-01 8.05426955e-01 -5.69994688e-01 -1.71377257e-01 -2.19598457e-01 -3.82538915e-01 -1.34438455e-01 5.89777470e-01 3.17614108e-01 -1.27288413e+00 -9.73427832e-01 3.51421773e-01 -6.41737580e-02 -1.04635119e+00 8.83722082e-02 4.70448732e-01 -8.97809684e-01 -8.91282976e-01 -6.91769004e-01 -1.07355988e+00 9.14264083e-01 6.95704699e-01 9.37873781e-01 -2.38254130e-01 -4.53908235e-01 7.88050234e-01 -6.59925163e-01 -4.43889648e-01 -3.22183102e-01 -9.00117382e-02 3.48751575e-01 3.56071979e-01 7.09386230e-01 -5.23784995e-01 -4.11943406e-01 4.48387057e-01 -9.37647521e-01 -3.41975778e-01 7.76823699e-01 7.75930345e-01 5.41216254e-01 3.37742776e-01 5.33429146e-01 -6.84630632e-01 2.86028564e-01 -7.68839955e-01 -5.73623717e-01 1.80783451e-01 -6.11347854e-01 -1.98070668e-02 3.02744865e-01 -5.93114018e-01 -1.02119541e+00 1.60455346e-01 3.67489576e-01 -3.86700183e-01 -7.68386364e-01 2.73976505e-01 -2.10796654e-01 1.53106257e-01 4.27048326e-01 4.90434200e-01 -1.08307660e-01 -7.22904325e-01 4.34933394e-01 1.01071596e+00 4.07765180e-01 -6.79262996e-01 5.71087182e-01 7.86148667e-01 -4.98183370e-01 -8.20956349e-01 -5.77590883e-01 -1.17260420e+00 -1.04192424e+00 -3.49802732e-01 6.72412574e-01 -1.06621516e+00 -5.48020184e-01 4.53984469e-01 -1.16059756e+00 -7.01370761e-02 -6.16638541e-01 4.94239300e-01 -8.60926092e-01 2.46011034e-01 -1.93033352e-01 -7.46642232e-01 1.65000170e-01 -1.07068217e+00 1.36864042e+00 6.81990618e-03 -7.85093103e-03 -7.29971349e-01 4.72655781e-02 7.01000988e-02 2.31773794e-01 -2.40666837e-01 9.98144388e-01 -9.24582481e-01 -6.34047449e-01 -3.04767102e-01 -2.03355372e-01 1.42268494e-01 3.56111020e-01 -6.72833741e-01 -8.85341287e-01 -6.16045415e-01 7.50036985e-02 -3.28426510e-01 7.36427009e-01 3.30703482e-02 1.01198959e+00 -2.48418018e-01 -5.24029434e-01 2.14548353e-02 1.55121458e+00 4.63994563e-01 2.96115905e-01 4.93071675e-01 4.41687524e-01 7.40489483e-01 7.94485450e-01 9.26738620e-01 6.78043842e-01 8.83947134e-01 5.67970395e-01 2.41431490e-01 1.07773952e-01 8.96743834e-02 1.37229234e-01 6.36953652e-01 1.22771494e-01 -2.18437865e-01 -1.04967999e+00 1.05786490e+00 -1.90422261e+00 -8.71298611e-01 3.37146133e-01 2.07200551e+00 5.47142386e-01 2.07728092e-02 1.45871386e-01 1.94278404e-01 8.82000804e-01 -1.65400565e-01 -5.49398005e-01 -2.97074407e-01 3.70193541e-01 -3.62307459e-01 2.45027348e-01 4.76541556e-02 -1.25266385e+00 9.32521105e-01 6.18473196e+00 6.13891602e-01 -8.34301651e-01 4.37339783e-01 1.34480774e-01 2.41134062e-01 1.77074417e-01 -1.50179908e-01 -1.08822143e+00 1.50595173e-01 7.67220497e-01 -1.20128557e-01 7.14090094e-02 1.50633216e+00 -1.89847037e-01 -1.93696007e-01 -1.20883667e+00 8.89948666e-01 4.90680099e-01 -9.48418379e-01 2.14416742e-01 2.00600952e-01 6.72720015e-01 -8.76558125e-02 -1.71108678e-01 4.21223223e-01 4.31787521e-01 -4.27601546e-01 1.22864020e+00 5.38605750e-01 5.12790322e-01 -5.01891375e-01 6.98440075e-01 5.29452384e-01 -1.23690999e+00 -5.10564268e-01 -7.24262655e-01 -3.47783826e-02 -9.58053097e-02 3.61825854e-01 -1.02610278e+00 2.68576980e-01 9.89825308e-01 5.24590909e-01 -3.42138439e-01 1.30268085e+00 -4.17847522e-02 2.84566045e-01 -3.28033954e-01 -6.28474131e-02 3.71006817e-01 1.47869468e-01 4.29755062e-01 1.34144628e+00 2.46514857e-01 1.56480327e-01 6.84582949e-01 3.91231358e-01 2.93430418e-01 1.96546018e-02 -5.25327742e-01 7.46831745e-02 4.60560143e-01 1.18556428e+00 -1.23032737e+00 -7.07807004e-01 -4.00515765e-01 8.36869895e-01 3.96542937e-01 1.59930304e-01 -4.64616597e-01 -2.40319192e-01 5.52072644e-01 -2.55504288e-02 9.72252548e-01 -4.88720924e-01 8.21257681e-02 -9.77338135e-01 -5.36295101e-02 -5.96453786e-01 5.43682277e-01 -5.16553283e-01 -1.44362998e+00 6.84638917e-01 4.58205283e-01 -1.44706070e+00 -2.14835539e-01 -6.55631900e-01 -1.21609876e-02 2.55509943e-01 -1.60289598e+00 -1.12665880e+00 -5.93761325e-01 6.50981188e-01 1.08746004e+00 -3.75626504e-01 1.09271955e+00 5.48758805e-02 -2.67075631e-03 2.23850921e-01 5.37247539e-01 -3.12661707e-01 5.45923769e-01 -1.14432955e+00 7.25832880e-02 3.54014605e-01 1.68303549e-01 6.34234488e-01 7.60220349e-01 -5.55201828e-01 -1.28717422e+00 -1.05594301e+00 7.78610587e-01 -4.96939756e-02 5.87534845e-01 -5.59472620e-01 -9.07761455e-01 6.96312010e-01 -4.38200645e-02 1.27589434e-01 6.81288421e-01 1.12953477e-01 -5.11160612e-01 -2.09216401e-01 -1.23143327e+00 1.04801543e-01 9.23944116e-01 -3.19691718e-01 -9.43024099e-01 5.82853973e-01 4.37840879e-01 1.29477186e-02 -7.13867545e-01 8.64511728e-02 4.99734521e-01 -7.01941609e-01 7.78595924e-01 -3.97156507e-01 -4.67332006e-02 -4.27235484e-01 -5.87284684e-01 -1.20256495e+00 -5.58349550e-01 3.11421137e-02 -1.66495219e-01 1.03199267e+00 -6.61021024e-02 -6.63866818e-01 6.25217915e-01 2.71170497e-01 -2.33028337e-01 -2.94672906e-01 -7.69506454e-01 -1.00441551e+00 -4.29845542e-01 -2.36900285e-01 4.96413440e-01 6.47341192e-01 -5.72853200e-02 1.12016179e-01 -6.77760318e-02 3.43196690e-01 9.35384750e-01 5.21115363e-01 7.77986407e-01 -1.60636401e+00 -9.37284604e-02 -9.33272243e-02 -1.10222149e+00 -1.07904494e+00 2.41080537e-01 -7.19463170e-01 6.17335200e-01 -1.77030647e+00 4.94279891e-01 -6.72065914e-01 -4.04586792e-01 6.62021279e-01 2.36637726e-01 2.33504459e-01 2.46173844e-01 6.69663846e-01 -1.01765800e+00 6.19040787e-01 5.59114277e-01 -3.39815378e-01 4.07547168e-02 2.71339770e-02 -4.42675114e-01 8.13130736e-01 5.78673601e-01 -6.29531801e-01 -4.37335700e-01 -4.44484442e-01 -4.40842569e-01 -4.19864386e-01 3.49708050e-01 -1.19274580e+00 5.88874102e-01 -2.65274137e-01 9.37821642e-02 -7.29133546e-01 6.46613359e-01 -1.10322165e+00 3.64552438e-02 4.00791347e-01 -2.06834614e-01 -2.82085776e-01 4.51409854e-02 9.18792427e-01 -2.64737636e-01 -3.88834000e-01 9.30004299e-01 -3.92228216e-01 -1.43852484e+00 1.71189532e-01 -7.37181008e-01 -5.64186692e-01 1.45478892e+00 -4.42083597e-01 5.22543304e-03 -2.37148792e-01 -8.96575391e-01 2.35815141e-02 5.53495944e-01 5.36253393e-01 5.85937679e-01 -1.20160890e+00 -4.42982942e-01 3.66821229e-01 7.55300105e-01 -1.03769809e-01 1.24729365e-01 2.68411398e-01 -2.18243331e-01 7.37778306e-01 -5.89984596e-01 -1.00154734e+00 -1.03763664e+00 8.12674999e-01 -1.19931512e-01 -2.92327069e-03 -8.29752386e-01 7.48160362e-01 6.29335105e-01 -4.41629857e-01 6.35789156e-01 -1.55177666e-02 -6.02007031e-01 3.76759529e-01 5.16840637e-01 2.44355381e-01 1.53083146e-01 -9.10408318e-01 -4.62917626e-01 5.77649236e-01 -3.63030553e-01 4.17235754e-02 1.61994731e+00 -4.48899686e-01 -2.55861819e-01 9.13736999e-01 1.10058463e+00 -6.22398674e-01 -1.27334845e+00 -5.79077780e-01 2.56787509e-01 -5.15071511e-01 -4.67140712e-02 -5.46068490e-01 -6.98860049e-01 7.08340883e-01 8.48705471e-01 3.02451074e-01 1.06550848e+00 5.64254522e-01 1.37627497e-01 7.85541296e-01 1.05792081e+00 -8.97242069e-01 3.21209282e-01 5.27809203e-01 8.04588258e-01 -1.18881381e+00 -1.36616221e-02 -3.89551044e-01 -5.18217623e-01 1.04064524e+00 3.80195230e-01 -1.29259869e-01 6.43242598e-01 -7.42135048e-02 -9.81480703e-02 -4.40255880e-01 -9.15102839e-01 -3.35696608e-01 1.39131546e-01 6.75009906e-01 -1.68284357e-01 1.13220923e-01 -7.51018971e-02 2.83023417e-01 1.27979919e-01 -2.24841565e-01 3.08187068e-01 1.34345233e+00 -1.01978564e+00 -9.10611391e-01 -3.68654907e-01 2.53589451e-01 -1.97894238e-02 3.47523868e-01 8.79075378e-02 6.35546327e-01 1.97511062e-01 8.91785085e-01 3.81725997e-01 -2.86635812e-02 3.68186563e-01 2.16018543e-01 3.55539352e-01 -1.06336391e+00 -5.78347035e-02 -3.49158980e-02 -1.90812200e-01 -5.17055273e-01 -5.90482056e-01 -9.35895979e-01 -1.26202023e+00 2.04857558e-01 -2.32582882e-01 3.07527214e-01 1.24550021e+00 1.02690840e+00 3.28642428e-01 3.00355315e-01 6.41693532e-01 -1.38980019e+00 -6.27410054e-01 -1.05839646e+00 -7.58343041e-01 3.52022439e-01 1.60929948e-01 -1.22340500e+00 -2.92017996e-01 5.29928565e-01]
[7.617640972137451, -1.19450044631958]
18348197-7d0c-497b-aca7-1f9218d3d195
few-shot-class-incremental-learning-for-named
null
null
https://aclanthology.org/2022.acl-long.43
https://aclanthology.org/2022.acl-long.43.pdf
Few-Shot Class-Incremental Learning for Named Entity Recognition
Previous work of class-incremental learning for Named Entity Recognition (NER) relies on the assumption that there exists abundance of labeled data for the training of new classes. In this work, we study a more challenging but practical problem, i.e., few-shot class-incremental learning for NER, where an NER model is trained with only few labeled samples of the new classes, without forgetting knowledge of the old ones. To alleviate the problem of catastrophic forgetting in few-shot class-incremental learning, we reconstruct synthetic training data of the old classes using the trained NER model, augmenting the training of new classes. We further develop a framework that distills from the existing model with both synthetic data, and real data from the current training set. Experimental results show that our approach achieves significant improvements over existing baselines.
['Ricardo Henao', 'Ruiyi Zhang', 'Subrata Mitra', 'Sungchul Kim', 'Handong Zhao', 'Tong Yu', 'Rui Wang']
null
null
null
null
acl-2022-5
['few-shot-class-incremental-learning']
['methodology']
[ 2.03897461e-01 4.22720641e-01 -1.27344012e-01 -4.03828681e-01 -8.89492750e-01 -4.64430571e-01 2.72037029e-01 9.25497413e-02 -8.01322162e-01 1.08761322e+00 3.75594586e-01 2.27471832e-02 4.92457598e-01 -9.94077146e-01 -7.00434446e-01 -4.19008166e-01 1.81045219e-01 6.58229828e-01 4.45768803e-01 -2.05750287e-01 -2.29157940e-01 3.82440984e-01 -1.63571143e+00 3.54330152e-01 8.03971231e-01 4.48243409e-01 9.00863409e-02 6.54525757e-01 -5.88833570e-01 1.16963565e+00 -5.05611777e-01 -5.60650229e-01 1.47145256e-01 -5.08932769e-01 -9.72927094e-01 5.10333665e-03 1.31722435e-01 -3.73891056e-01 -6.95357800e-01 8.17845345e-01 6.23569131e-01 5.59950829e-01 3.58187467e-01 -9.37921405e-01 -8.54572773e-01 9.66006696e-01 -3.39938924e-02 1.55595243e-01 -1.16540499e-01 -3.74218374e-02 6.50108814e-01 -1.51726866e+00 8.94384682e-01 9.33707595e-01 1.10915053e+00 1.21297336e+00 -1.03458405e+00 -5.86312950e-01 1.45401254e-01 1.67922854e-01 -1.32493687e+00 -7.70938873e-01 4.83491182e-01 2.35182345e-02 8.41715097e-01 -2.45975077e-01 4.16770995e-01 1.23318219e+00 -2.91088939e-01 1.01762879e+00 7.82399416e-01 -7.49891341e-01 6.85782731e-01 1.89990699e-01 4.97879446e-01 5.42660534e-01 4.17187184e-01 5.87560572e-02 -2.54664391e-01 -2.49872223e-01 1.98790774e-01 5.37044108e-01 -6.65379167e-02 -4.12976325e-01 -8.00882578e-01 5.33566475e-01 3.08094233e-01 4.02513593e-01 -4.34936672e-01 -2.08622187e-01 3.32943171e-01 3.04677308e-01 5.66336513e-01 1.84465140e-01 -1.06641054e+00 7.74737622e-04 -1.01822662e+00 -2.09202871e-01 9.39500630e-01 1.28456104e+00 1.02905512e+00 3.21683139e-01 -2.77496785e-01 8.22492123e-01 -1.62262782e-01 3.10288578e-01 8.56155276e-01 -7.15904832e-01 4.05085951e-01 5.45361757e-01 4.19166625e-01 4.33161967e-02 -1.52131960e-01 -1.29603997e-01 -6.56748235e-01 -1.70962095e-01 1.63519964e-01 -4.83142972e-01 -1.58225417e+00 1.59024382e+00 6.23693705e-01 7.58309305e-01 6.68579996e-01 -3.16232257e-02 7.63818502e-01 7.45960653e-01 4.41642910e-01 -4.24347281e-01 7.84031212e-01 -1.25913811e+00 -5.74792802e-01 -3.35071802e-01 7.68112183e-01 -1.37065843e-01 1.08586323e+00 8.02604854e-02 -6.57925308e-01 -7.11754322e-01 -7.65784085e-01 -7.66817704e-02 -7.91212499e-01 1.45998970e-01 4.48385000e-01 6.73894465e-01 -7.21882641e-01 8.74742687e-01 -8.53186429e-01 -4.93872732e-01 6.92530096e-01 -9.91744548e-02 -3.19770634e-01 -6.66137576e-01 -1.16869664e+00 5.72876692e-01 9.07249331e-01 -1.82329372e-01 -1.15057814e+00 -9.31118906e-01 -9.64865327e-01 2.66134322e-01 5.37477612e-01 -4.52306181e-01 1.68952751e+00 -6.42790854e-01 -1.15548956e+00 4.36167270e-01 -1.48369789e-01 -4.39818799e-01 2.04992861e-01 -2.98127741e-01 -7.14604795e-01 -2.63630211e-01 8.52324888e-02 5.38005412e-01 6.48449719e-01 -1.46250129e+00 -7.45336890e-01 -1.61235601e-01 -8.75657797e-02 -8.89217705e-02 -6.44831419e-01 -4.08671945e-01 -5.01994230e-02 -5.48266828e-01 -6.61885813e-02 -7.24696517e-01 -4.28624272e-01 -2.30162546e-01 -3.05825502e-01 -2.00864822e-01 8.27446282e-01 -3.77575159e-01 1.18293941e+00 -2.10506344e+00 -3.44932377e-01 -4.71013784e-01 1.68853834e-01 7.03864157e-01 -4.68706876e-01 4.20198232e-01 -2.45953977e-01 1.79486170e-01 -3.84837717e-01 -6.36151552e-01 -2.58660585e-01 5.75777590e-01 -5.91463387e-01 -1.32339403e-01 2.05027506e-01 8.94516349e-01 -1.45026767e+00 -2.87141889e-01 -5.25646210e-02 4.88566190e-01 -3.99353772e-01 3.46186340e-01 -1.48172706e-01 3.09630811e-01 -1.27882734e-01 6.36400580e-01 7.27609217e-01 -6.06264137e-02 9.90897864e-02 -1.18038535e-01 3.05184927e-02 1.01740852e-01 -1.33637476e+00 1.77540600e+00 -4.28126991e-01 7.02015236e-02 -7.16056168e-01 -7.42272317e-01 7.63526440e-01 6.60788059e-01 1.27753705e-01 -2.60221034e-01 -3.59839238e-02 9.18103307e-02 -3.94816607e-01 -4.23272729e-01 4.87796456e-01 -7.96356738e-01 -1.80602387e-01 5.46335459e-01 8.74111533e-01 1.37361988e-01 3.73996586e-01 2.64329433e-01 1.24067330e+00 1.51833430e-01 4.80271995e-01 5.97042799e-01 -3.89749035e-02 1.12659544e-01 1.10941696e+00 1.03850436e+00 -3.87531519e-01 6.30401969e-01 -1.09519593e-01 -7.44109929e-01 -1.40381980e+00 -1.24871624e+00 4.86918800e-02 1.26087153e+00 -3.07379454e-01 -3.12152416e-01 -7.08782554e-01 -1.41545045e+00 -4.42482620e-01 1.42897558e+00 -7.22154379e-01 -4.49771553e-01 -4.69487548e-01 -7.18079150e-01 7.18272567e-01 8.04790676e-01 4.95957434e-01 -1.24754214e+00 -4.62138593e-01 3.96795869e-01 -1.51501209e-01 -8.88612330e-01 -4.57490146e-01 4.03284311e-01 -1.24808896e+00 -1.00310361e+00 -7.10599065e-01 -1.12945616e+00 8.80133510e-01 4.61672395e-01 1.10661268e+00 -1.51207581e-01 -2.82078326e-01 5.83692610e-01 -5.90876698e-01 -5.16818762e-01 -5.41904569e-01 1.75321028e-01 1.71273395e-01 -1.53001249e-01 5.63106775e-01 -5.66670239e-01 -7.20140897e-03 -2.28583008e-01 -1.15409029e+00 -2.00078860e-01 7.56336629e-01 1.27946591e+00 7.57788897e-01 3.74716252e-01 8.10186505e-01 -1.65640604e+00 9.49142575e-02 -8.51311862e-01 1.72214136e-01 6.65737152e-01 -5.94583631e-01 2.39038453e-01 9.58504140e-01 -8.35856915e-01 -1.70572269e+00 2.81583071e-01 -8.21023509e-02 -6.77807570e-01 -3.61597538e-01 4.15869802e-01 -2.45153189e-01 3.76684695e-01 9.15689409e-01 4.31124657e-01 -8.17061365e-01 -8.04122090e-01 8.30771625e-01 6.43409789e-01 7.36438751e-01 -4.95755315e-01 9.00081992e-01 4.65246886e-01 -6.16495848e-01 -7.62896538e-01 -1.61030519e+00 -5.19932985e-01 -1.06800437e+00 5.99070117e-02 1.92166626e-01 -1.05999792e+00 4.50626999e-01 4.56422627e-01 -1.19992709e+00 -3.64334106e-01 -1.36753285e+00 3.92609477e-01 -3.13940465e-01 2.37606257e-01 -7.59796798e-01 -8.45206439e-01 -3.60802144e-01 -2.19189540e-01 8.93287003e-01 5.00354409e-01 1.97344974e-01 -1.04911172e+00 6.62217617e-01 -4.00446691e-02 1.93760529e-01 6.20329147e-03 8.75962794e-01 -1.19776380e+00 -1.11421198e-01 -6.61718547e-01 2.07356110e-01 3.82985562e-01 2.35549971e-01 -5.14097691e-01 -1.33261871e+00 -4.64481473e-01 1.95413753e-01 -7.96596110e-01 1.03172076e+00 -2.75695652e-01 8.46136391e-01 -4.14301813e-01 -5.03580213e-01 3.36933672e-01 1.46971846e+00 1.83489338e-01 7.53904521e-01 -6.65349588e-02 7.33864248e-01 3.06235164e-01 6.78123713e-01 6.82111561e-01 4.52098191e-01 -2.07031369e-01 -1.13581546e-01 8.53234157e-02 -4.31138396e-01 -8.75556707e-01 1.92876875e-01 1.12112999e+00 1.51764646e-01 -2.60226369e-01 -1.05081987e+00 9.56136584e-01 -1.68759429e+00 -1.11428928e+00 5.40633321e-01 2.19293165e+00 1.13132226e+00 1.37901962e-01 -2.55608380e-01 3.14017013e-02 1.05886626e+00 -1.56333908e-01 -9.93685782e-01 2.59973049e-01 4.22521159e-02 6.34300828e-01 1.73563153e-01 2.68236428e-01 -1.22771633e+00 1.26568031e+00 6.64355755e+00 6.14060521e-01 -7.26670563e-01 6.72483027e-01 4.45724756e-01 -1.29515715e-02 -2.24083260e-01 3.06333542e-01 -1.15800965e+00 1.49867401e-01 1.49312758e+00 -3.10662925e-01 2.56926656e-01 1.06088746e+00 -3.25268716e-01 2.93398380e-01 -1.04497337e+00 6.52313769e-01 3.20413947e-01 -1.15105391e+00 2.22386733e-01 -4.62445319e-01 1.13352597e+00 2.71126419e-01 -3.98924768e-01 1.13382244e+00 8.96185875e-01 -4.02893305e-01 4.98152047e-01 8.15150023e-01 1.02215266e+00 -7.12990105e-01 8.92381668e-01 7.72241652e-01 -1.19547784e+00 -3.31092566e-01 -6.68445826e-01 1.37268141e-01 1.67821780e-01 6.29262567e-01 -9.43684459e-01 3.17655832e-01 5.37415445e-01 7.37740993e-01 -5.13405204e-01 1.27288640e+00 -3.66388619e-01 8.91215622e-01 -1.33114502e-01 3.19235831e-01 -1.79845035e-01 3.16289485e-01 1.98981375e-01 1.15371776e+00 4.61731106e-01 4.33949322e-01 1.81863502e-01 5.28475642e-01 -5.91982245e-01 -2.00666692e-02 -8.63072276e-01 -2.89128721e-01 9.06095326e-01 1.19045722e+00 -7.51171708e-01 -8.22911441e-01 -5.53681672e-01 1.28825712e+00 7.25147247e-01 3.56763631e-01 -6.74400091e-01 -8.44726443e-01 7.13752210e-02 -2.39030302e-01 8.28696549e-01 3.70039628e-03 9.15088654e-02 -1.74082828e+00 -2.14048967e-01 -3.73972774e-01 7.12031543e-01 -7.15972364e-01 -1.52476203e+00 7.75875866e-01 -4.08089370e-01 -1.25175881e+00 -1.16350062e-01 -2.34165579e-01 -8.06183219e-01 3.76470059e-01 -1.57545519e+00 -1.25961673e+00 -1.97242737e-01 4.57175374e-01 9.91309822e-01 -1.49720922e-01 1.22468925e+00 2.94322759e-01 -5.22221744e-01 7.21673906e-01 4.52171177e-01 3.89140338e-01 6.94433272e-01 -1.11867297e+00 6.18927538e-01 9.59567487e-01 4.39715594e-01 5.24848700e-01 3.00115496e-01 -9.09231246e-01 -9.29583073e-01 -1.68861604e+00 1.36686218e+00 -8.00690889e-01 4.58634555e-01 -4.17578697e-01 -1.05248260e+00 1.13570249e+00 -1.40473396e-01 4.38140690e-01 1.09977055e+00 -6.43382072e-02 -6.42571628e-01 4.27854285e-02 -1.39516211e+00 4.67495859e-01 1.16868675e+00 -4.72541809e-01 -1.20774114e+00 2.64365882e-01 1.17184341e+00 -4.71546203e-02 -8.23168278e-01 1.72124937e-01 1.75466210e-01 -4.44532096e-01 9.45550144e-01 -1.19678068e+00 -6.49510473e-02 -2.51641423e-01 -1.05428524e-01 -1.59421706e+00 -2.76034385e-01 -1.98771521e-01 -6.05919778e-01 1.51380789e+00 4.11873102e-01 -1.67804539e-01 9.41667020e-01 7.81658232e-01 -2.83469290e-01 -2.34133959e-01 -9.17566240e-01 -9.47401941e-01 2.28054345e-01 -2.76123196e-01 5.46794772e-01 1.32791436e+00 -7.37537742e-02 7.90031552e-01 -5.49668431e-01 1.10952549e-01 6.48313046e-01 -2.05384627e-01 4.61522132e-01 -1.50870192e+00 -1.20149627e-01 6.14961982e-01 -1.10207506e-01 -7.15679407e-01 1.07226387e-01 -8.75356138e-01 4.71762568e-01 -1.52107108e+00 5.62202811e-01 -4.66879487e-01 -7.06243098e-01 1.14465368e+00 -3.78372282e-01 -1.28755311e-03 1.07634924e-01 2.70167589e-01 -1.14355791e+00 8.77406299e-01 7.79676616e-01 -1.16366982e-01 -3.36346418e-01 -9.29545984e-02 -7.68980086e-01 7.44591296e-01 5.83568573e-01 -1.15554583e+00 -5.44261754e-01 -3.10108960e-01 -1.32855758e-01 -2.17301995e-01 -2.08050504e-01 -1.30840206e+00 6.16587639e-01 -1.21291585e-01 6.24275148e-01 -5.54692864e-01 1.49143040e-01 -7.54563630e-01 -6.59032688e-02 4.25659448e-01 -3.32114458e-01 -3.09898585e-01 5.39379865e-02 1.07001686e+00 1.83976851e-02 -6.99669957e-01 8.27692926e-01 -5.05788684e-01 -1.17095816e+00 6.48136497e-01 -5.35104871e-02 5.88161945e-01 9.15911734e-01 -1.18929828e-02 -2.50958890e-01 3.24231852e-03 -1.27937090e+00 -1.13495300e-02 3.85971516e-01 3.13612759e-01 7.35683024e-01 -1.43089962e+00 -5.96556902e-01 1.28123388e-01 4.74368989e-01 1.52241439e-01 3.32249910e-01 9.04510692e-02 2.42583565e-02 -8.71756971e-02 4.62825298e-02 1.68298751e-01 -6.14714146e-01 1.10588002e+00 1.37540936e-01 -3.55827689e-01 -5.57884514e-01 6.90659463e-01 -2.69568861e-01 -1.15247035e+00 2.45069861e-01 -9.24751386e-02 -2.87484825e-01 1.83065251e-01 9.37606931e-01 5.91303766e-01 3.53179947e-02 -2.56757945e-01 1.03632644e-01 -5.30702472e-02 -5.09753227e-01 3.76667604e-02 1.81618512e+00 -1.58192411e-01 2.63323486e-01 1.12641501e+00 9.05888557e-01 -1.09386824e-01 -1.35519004e+00 -6.98000908e-01 2.73954093e-01 -1.85492396e-01 -2.72898406e-01 -8.75926137e-01 -7.16918051e-01 9.01811540e-01 7.64142036e-01 -3.36797714e-01 8.28004956e-01 -6.04393370e-02 1.14973605e+00 1.09807587e+00 7.44478762e-01 -1.39280725e+00 1.42558664e-01 9.16256368e-01 1.30232098e-02 -1.22480345e+00 -3.66347760e-01 -4.32143621e-02 -5.54216385e-01 1.08744347e+00 7.47711420e-01 2.86149178e-02 8.83103073e-01 1.13964058e-01 -1.01587012e-01 1.71359554e-01 -8.85647893e-01 -4.27266926e-01 -3.86139095e-01 8.05587053e-01 2.22638115e-01 -1.29428208e-01 1.87898595e-02 1.15039241e+00 3.62620234e-01 5.05949676e-01 6.54658616e-01 1.37371457e+00 -8.28104079e-01 -1.18110657e+00 7.62308687e-02 4.93575960e-01 -7.82741755e-02 -5.19029737e-01 -2.08774686e-01 5.51442623e-01 5.13433099e-01 7.03728855e-01 -9.33409762e-03 -4.30085450e-01 6.54031754e-01 9.47017491e-01 1.22763604e-01 -1.35983217e+00 -1.35266379e-01 -4.51977313e-01 7.30032176e-02 4.05920036e-02 -2.24739537e-01 -6.44394636e-01 -1.41137040e+00 -1.95269827e-02 -4.67319280e-01 5.94679296e-01 2.64979333e-01 1.06024194e+00 4.67611998e-01 3.84114563e-01 8.03863823e-01 -6.61698937e-01 -8.32807004e-01 -1.05331099e+00 -7.89556801e-01 4.19635534e-01 2.44395494e-01 -5.31306922e-01 -4.74253029e-01 3.67098749e-01]
[9.693087577819824, 9.303074836730957]
9022178c-8d55-4bf3-8f49-bfb5ecd5bab3
ssmd-semi-supervised-medical-image-detection
2106.01544
null
https://arxiv.org/abs/2106.01544v1
https://arxiv.org/pdf/2106.01544v1.pdf
SSMD: Semi-Supervised Medical Image Detection with Adaptive Consistency and Heterogeneous Perturbation
Semi-Supervised classification and segmentation methods have been widely investigated in medical image analysis. Both approaches can improve the performance of fully-supervised methods with additional unlabeled data. However, as a fundamental task, semi-supervised object detection has not gained enough attention in the field of medical image analysis. In this paper, we propose a novel Semi-Supervised Medical image Detector (SSMD). The motivation behind SSMD is to provide free yet effective supervision for unlabeled data, by regularizing the predictions at each position to be consistent. To achieve the above idea, we develop a novel adaptive consistency cost function to regularize different components in the predictions. Moreover, we introduce heterogeneous perturbation strategies that work in both feature space and image space, so that the proposed detector is promising to produce powerful image representations and robust predictions. Extensive experimental results show that the proposed SSMD achieves the state-of-the-art performance at a wide range of settings. We also demonstrate the strength of each proposed module with comprehensive ablation studies.
['Yizhou Yu', 'Weimin Li', 'Shu Zhang', 'Gang Wang', 'Haofeng Li', 'Chengdi Wang', 'Hong-Yu Zhou']
2021-06-03
null
null
null
null
['semi-supervised-object-detection', 'medical-image-detection']
['computer-vision', 'computer-vision']
[ 4.42642182e-01 3.38564813e-01 -4.53766763e-01 -5.47879815e-01 -9.80836272e-01 -3.95821817e-02 2.78949708e-01 1.27252281e-01 -3.49000275e-01 4.42691982e-01 6.11795904e-03 -1.64483823e-02 -3.14529217e-03 -3.01607937e-01 -4.75963712e-01 -9.62366223e-01 2.24896312e-01 3.24256212e-01 3.72971803e-01 1.49279267e-01 -3.34487073e-02 2.37560004e-01 -1.27030528e+00 2.51884218e-02 1.15374362e+00 9.95172024e-01 4.01266426e-01 1.63681477e-01 1.54164657e-01 7.20804811e-01 -2.11143836e-01 -1.42449319e-01 2.38790870e-01 -5.67053854e-01 -9.11931396e-01 7.90116429e-01 1.53912693e-01 -2.32904162e-02 -1.07163966e-01 1.35930896e+00 5.80285490e-01 -8.16104338e-02 7.46396184e-01 -9.38856304e-01 -4.48635221e-01 6.34746909e-01 -8.79153013e-01 1.76706314e-01 -7.46961683e-02 -1.01483226e-01 8.61298561e-01 -8.01818073e-01 5.70861042e-01 8.22042763e-01 4.93882447e-01 7.03719854e-01 -1.12396049e+00 -4.99813825e-01 2.26258993e-01 9.58518311e-02 -1.38792515e+00 -2.51459152e-01 9.27841365e-01 -2.89428055e-01 2.53282040e-01 9.17807072e-02 3.20056319e-01 7.85514772e-01 8.24281853e-03 1.30750084e+00 1.13475502e+00 -6.25212014e-01 3.01258057e-01 4.08715993e-01 2.49957323e-01 1.00392354e+00 3.07580173e-01 -2.47372597e-01 -2.58609593e-01 -1.21197373e-01 6.19303286e-01 2.27873057e-01 -3.95604432e-01 -6.53582454e-01 -1.14091027e+00 7.62745917e-01 4.82276827e-01 4.10441190e-01 -2.73863018e-01 -2.49028116e-01 2.32509688e-01 -1.67696849e-01 7.08808362e-01 1.85456797e-01 -1.60987973e-01 3.24142903e-01 -1.05444801e+00 -7.91569203e-02 3.68365079e-01 8.93692017e-01 5.40759623e-01 -2.27061242e-01 -3.34089220e-01 9.80699241e-01 4.02234077e-01 3.52399468e-01 7.65178740e-01 -6.84648097e-01 2.61661351e-01 8.97844076e-01 -8.18072557e-02 -8.56161594e-01 -6.17404282e-01 -6.61543727e-01 -1.15909731e+00 -7.06237927e-02 2.91215301e-01 5.69514297e-02 -1.06728232e+00 1.48963857e+00 5.59510887e-01 3.40289116e-01 -3.79319154e-02 9.75571573e-01 7.12594151e-01 3.18688661e-01 3.64669934e-02 -5.39758384e-01 1.14401937e+00 -1.23029292e+00 -7.76419640e-01 -3.16109955e-01 8.13083887e-01 -5.54747939e-01 9.55242276e-01 1.71892688e-01 -8.85084450e-01 -4.84969437e-01 -1.07944798e+00 3.22803468e-01 2.18458340e-01 4.19577837e-01 5.01603067e-01 5.68517983e-01 -6.55232489e-01 4.51834887e-01 -1.24551618e+00 -2.08042368e-01 7.72615314e-01 2.90555924e-01 -3.44232470e-01 -2.24940956e-01 -7.67986536e-01 6.52065039e-01 3.27430010e-01 1.10994704e-01 -5.46682239e-01 -3.86620998e-01 -8.43460679e-01 -2.04393372e-01 4.64912593e-01 -4.48426187e-01 1.17174613e+00 -9.17381644e-01 -1.26005483e+00 1.16182363e+00 -2.12699041e-01 -5.14345169e-01 6.18631840e-01 -5.68802655e-02 -1.99416056e-01 4.35332417e-01 3.37507427e-01 6.58983588e-01 9.43417549e-01 -1.30951786e+00 -6.09201789e-01 -5.78423440e-01 -6.28000438e-01 2.95156866e-01 -5.72278142e-01 -9.63058025e-02 -8.45539570e-01 -7.68402100e-01 5.79941750e-01 -1.16193783e+00 -7.60918736e-01 1.80832461e-01 -7.70416737e-01 -5.80029897e-02 5.24361253e-01 -4.42474276e-01 1.12120891e+00 -2.24220014e+00 5.06954864e-02 2.82453507e-01 1.85435459e-01 3.71509671e-01 1.55395851e-01 -1.67868316e-01 1.13732725e-01 -2.56947298e-02 -7.63275743e-01 -6.71315551e-01 -3.58417928e-01 3.26720744e-01 -1.67189073e-02 8.22239399e-01 3.38940501e-01 9.07876372e-01 -8.29682171e-01 -1.02903032e+00 2.31214687e-01 3.17698181e-01 -5.54311037e-01 1.56661615e-01 -3.46955918e-02 7.83919811e-01 -8.67736578e-01 7.02627659e-01 7.84709990e-01 -8.69833291e-01 2.37356916e-01 -5.22884205e-02 3.99379551e-01 -2.05723658e-01 -1.25263679e+00 1.78111279e+00 2.50280071e-02 6.18353076e-02 -7.33156502e-03 -1.52778816e+00 6.97412193e-01 2.39789054e-01 8.21924627e-01 -4.66936290e-01 5.44548780e-02 2.00591207e-01 -2.24237084e-01 -4.55335706e-01 8.85601267e-02 -1.77054301e-01 1.44963175e-01 2.67865270e-01 -5.53108007e-03 6.40273988e-02 7.41105378e-02 2.24717617e-01 9.29606915e-01 3.40430327e-02 6.16581857e-01 -1.96952894e-01 6.29338264e-01 6.97243959e-02 8.75435770e-01 6.91570461e-01 -5.77013373e-01 8.47221911e-01 1.48958161e-01 -1.75719172e-01 -5.47330499e-01 -7.28640854e-01 -4.72509772e-01 6.34870470e-01 6.24133706e-01 -1.17625967e-01 -8.97583842e-01 -1.18879640e+00 -1.84367344e-01 2.08612099e-01 -6.55892730e-01 -2.03636602e-01 -4.07885104e-01 -1.07251465e+00 3.39278042e-01 5.89480281e-01 6.42357171e-01 -8.80954981e-01 -4.99548256e-01 1.15001634e-01 -3.36322546e-01 -1.35870421e+00 -5.17803252e-01 3.13469201e-01 -1.08405340e+00 -1.04997897e+00 -9.58235323e-01 -1.06036663e+00 1.15115488e+00 4.58183825e-01 7.50049710e-01 2.14934513e-01 -4.52421576e-01 1.32447332e-01 -5.78399658e-01 -1.90722853e-01 -4.77999270e-01 6.58651739e-02 4.65040617e-02 2.58587360e-01 3.24015319e-02 -2.57841974e-01 -6.90979898e-01 5.60291529e-01 -1.02117002e+00 6.72201216e-02 7.27413595e-01 1.05010974e+00 1.11523485e+00 3.38758938e-02 5.42716801e-01 -1.24950886e+00 1.05669901e-01 -2.80533165e-01 -3.29944134e-01 3.74700367e-01 -8.93982172e-01 1.75111338e-01 5.79492509e-01 -2.39188567e-01 -1.11412001e+00 5.85768998e-01 -1.00982778e-01 -4.66026455e-01 -2.60579884e-01 5.19363344e-01 -6.31230474e-02 -1.08887188e-01 6.69449747e-01 4.42343771e-01 1.73118442e-01 -5.15591681e-01 3.16655159e-01 6.72529161e-01 4.91969496e-01 -1.95769057e-01 7.19797015e-01 9.15424585e-01 1.49709508e-02 -5.98121822e-01 -1.25641978e+00 -1.00573051e+00 -6.81195259e-01 -6.16799900e-03 7.81177163e-01 -9.30175960e-01 -1.29683048e-01 3.51431787e-01 -6.88955605e-01 8.12414009e-03 -2.73070693e-01 5.60388386e-01 -5.01035869e-01 7.84151733e-01 -5.66440046e-01 -7.04590738e-01 -4.08823401e-01 -1.46682799e+00 1.19531858e+00 3.25125515e-01 1.72023345e-02 -9.26981091e-01 -6.71747420e-03 5.43409646e-01 -7.69147947e-02 8.30919072e-02 4.84149277e-01 -9.07954276e-01 -3.14914256e-01 -2.68932670e-01 -1.77964464e-01 5.64360321e-01 3.32793146e-01 -2.97648817e-01 -9.08286810e-01 -3.05437356e-01 2.54102379e-01 -4.92273718e-01 1.15357256e+00 6.30968392e-01 1.35564220e+00 1.95436720e-02 -6.90921664e-01 6.43177629e-01 1.17812967e+00 -9.35786441e-02 4.17617202e-01 2.65817195e-01 6.21892393e-01 4.92864728e-01 9.67207551e-01 5.19426823e-01 1.29145071e-01 6.99168921e-01 3.37108880e-01 -4.29804951e-01 -1.03789039e-01 -6.49372861e-02 4.79319915e-02 7.44781435e-01 2.06953555e-01 3.47420238e-02 -8.11856508e-01 4.92690384e-01 -2.15701318e+00 -6.98884249e-01 4.19761054e-02 2.07803822e+00 9.96213198e-01 1.90000191e-01 -2.02899873e-02 2.78116196e-01 8.13259482e-01 5.27144643e-04 -5.57258964e-01 4.45782959e-01 -3.23253758e-02 -8.04924127e-03 4.49897468e-01 1.13583729e-01 -1.51523638e+00 8.71604860e-01 5.83151865e+00 1.03316736e+00 -1.06845939e+00 1.05496936e-01 9.10998344e-01 2.49547184e-01 2.45154239e-02 -3.33585232e-01 -9.06487644e-01 4.05528069e-01 3.45708787e-01 1.43240184e-01 -1.71693727e-01 1.17421591e+00 1.97040483e-01 -1.44072697e-01 -9.35182452e-01 1.01648092e+00 2.18450472e-01 -1.26689124e+00 5.37710916e-03 -1.19122021e-01 9.05178845e-01 -1.38326019e-01 7.43195713e-02 -9.70592443e-03 4.23311256e-03 -7.51884162e-01 3.97159994e-01 5.87269515e-02 5.89626670e-01 -4.29402411e-01 7.53860474e-01 6.15346849e-01 -9.78603125e-01 6.30819127e-02 -3.15880895e-01 4.15233195e-01 1.99040040e-01 8.22123468e-01 -8.05105090e-01 6.68723583e-01 5.52272618e-01 9.53963637e-01 -6.44235671e-01 1.11444533e+00 -2.23060414e-01 8.67992520e-01 -2.64722735e-01 1.54594302e-01 1.75014377e-01 -3.90440188e-02 3.30731392e-01 1.10735226e+00 -4.15052101e-02 1.61314562e-01 5.66577852e-01 5.70972443e-01 1.19567011e-02 3.45263213e-01 -2.25935638e-01 1.53310090e-01 2.34215960e-01 1.28026712e+00 -1.14844930e+00 -2.60708600e-01 -3.46446723e-01 1.03204429e+00 3.35933715e-01 1.45609915e-01 -7.63844132e-01 4.29796763e-02 1.48713246e-01 6.25739545e-02 2.68067956e-01 1.47801518e-01 -4.19892043e-01 -1.32798958e+00 2.12341577e-01 -6.94822252e-01 6.64686322e-01 -2.65190810e-01 -1.25224543e+00 4.93060797e-01 -8.43935683e-02 -1.54778755e+00 -8.81186724e-02 -4.22907680e-01 -3.50102156e-01 3.62230629e-01 -1.76358032e+00 -1.12107110e+00 -3.25395882e-01 7.02393532e-01 5.89412570e-01 -8.76921117e-02 7.46698678e-01 2.78261125e-01 -9.03961003e-01 7.08505929e-01 2.58673131e-01 1.82629883e-01 8.01787734e-01 -1.17073691e+00 -7.56161436e-02 9.32158530e-01 3.35207641e-01 4.69909936e-01 4.61219311e-01 -5.39983332e-01 -1.05753422e+00 -1.27744758e+00 3.82723182e-01 -1.30197247e-02 2.29227677e-01 3.64144295e-02 -9.41582918e-01 5.11750579e-01 -2.65731990e-01 6.46247089e-01 7.82733858e-01 -2.17999056e-01 -1.32455379e-01 -1.27973691e-01 -1.18167019e+00 3.75022441e-01 8.63403797e-01 -1.58479810e-01 -3.95731747e-01 6.77338958e-01 5.13616562e-01 -4.81882542e-01 -6.39499485e-01 7.36143053e-01 6.92383498e-02 -8.63588691e-01 8.03197980e-01 -3.06888878e-01 3.42966884e-01 -3.14486235e-01 4.93392944e-02 -1.10539699e+00 -3.09484154e-01 -4.99451995e-01 -1.37942061e-01 9.26638901e-01 4.29246664e-01 -3.91191661e-01 1.12996852e+00 4.40904289e-01 -2.59852260e-01 -1.10420167e+00 -9.43172932e-01 -6.89751863e-01 -1.27014518e-01 -3.81503820e-01 -3.00678760e-02 9.03419316e-01 2.26664215e-01 2.01263547e-01 -4.27772045e-01 3.57651800e-01 8.46239150e-01 2.97284156e-01 3.92634183e-01 -9.93462443e-01 -4.89355475e-01 -2.03528598e-01 -5.15374124e-01 -1.16967452e+00 1.36339247e-01 -1.01282358e+00 3.45163226e-01 -1.52828968e+00 6.97190821e-01 -6.03059292e-01 -4.24724132e-01 5.07926464e-01 -7.62502253e-01 4.22239125e-01 -9.18263718e-02 5.87486446e-01 -9.46426392e-01 6.08787477e-01 1.32315958e+00 -1.88208669e-01 -1.33103296e-01 2.38731042e-01 -7.95836747e-01 8.75169337e-01 7.23596454e-01 -5.58465242e-01 -4.39612299e-01 -1.25889152e-01 -3.52843106e-01 -1.42444119e-01 1.13452129e-01 -9.51015949e-01 2.72894770e-01 -2.41453238e-02 2.79902309e-01 -4.74121332e-01 3.72007117e-02 -7.77648389e-01 -3.75256896e-01 6.23723090e-01 -4.08248305e-01 -5.23213446e-01 -1.57837570e-01 9.44173396e-01 -4.76704806e-01 -1.66434631e-01 1.27427232e+00 -3.62029262e-02 -6.84434175e-01 4.13804680e-01 8.18224624e-02 7.28730708e-02 1.31942022e+00 -1.29049882e-01 7.66664452e-04 -2.26623759e-01 -7.76174247e-01 4.73917186e-01 3.13120335e-01 1.02562897e-01 7.42076278e-01 -1.03550780e+00 -5.73307455e-01 1.63537741e-01 3.73980045e-01 2.38304988e-01 2.86317259e-01 1.05067658e+00 -3.07696909e-01 2.85977155e-01 1.54755160e-01 -1.05697215e+00 -1.24173129e+00 6.95643127e-01 2.72983044e-01 -5.31323254e-01 -8.13058376e-01 9.26958144e-01 3.41596246e-01 -3.44372660e-01 4.34165388e-01 -3.03938895e-01 -2.81810373e-01 -2.34877780e-01 5.56959689e-01 -8.58531159e-04 1.45726472e-01 -6.70881093e-01 -4.58820462e-01 5.63609362e-01 -3.55038226e-01 2.18116209e-01 1.12109649e+00 -2.19497412e-01 2.10041627e-01 1.92505866e-01 1.04283965e+00 -2.28553250e-01 -1.27956188e+00 -5.80124438e-01 9.41344723e-02 -2.23480999e-01 1.66952938e-01 -4.15424019e-01 -1.19463420e+00 7.18146503e-01 7.17820108e-01 1.00253159e-02 1.21703935e+00 2.54158676e-01 7.56678283e-01 3.12088698e-01 2.72342771e-01 -1.05795896e+00 2.50072420e-01 -3.05739958e-02 3.11034858e-01 -1.76749396e+00 1.59620970e-01 -8.95736635e-01 -9.81492162e-01 7.58506417e-01 5.44518948e-01 4.44738232e-02 7.21337557e-01 1.37147769e-01 2.27800354e-01 -1.31365061e-01 -3.59295160e-01 -3.82226497e-01 4.51831937e-01 3.75833869e-01 4.27090645e-01 -1.93156209e-02 -4.10878181e-01 6.88514352e-01 4.86358106e-01 8.02632347e-02 2.12141842e-01 1.02699518e+00 -4.73885268e-01 -1.09471750e+00 -2.70296156e-01 5.30846655e-01 -7.24482894e-01 1.21087573e-01 -9.95578542e-02 5.35230219e-01 -1.10699058e-01 9.32095528e-01 -3.94561768e-01 -1.98482364e-01 1.24381870e-01 -1.53048784e-01 2.91067660e-01 -7.99226940e-01 -2.31897429e-01 3.86684239e-01 -3.72491956e-01 -5.15651464e-01 -7.86287308e-01 -6.51953816e-01 -1.56527781e+00 5.67684114e-01 -7.76643872e-01 1.61473364e-01 3.68651211e-01 1.10068071e+00 3.05327773e-01 2.98285842e-01 9.62081313e-01 -6.02567255e-01 -8.79601061e-01 -8.23205948e-01 -7.78396547e-01 7.01560438e-01 1.59535021e-01 -6.67909205e-01 -1.93811432e-01 2.34742314e-01]
[14.759422302246094, -2.0814921855926514]
41f56950-39d8-45a0-9422-f3a3b36d87ce
a-survey-and-approach-to-chart-classification
2307.04147
null
https://arxiv.org/abs/2307.04147v1
https://arxiv.org/pdf/2307.04147v1.pdf
A Survey and Approach to Chart Classification
Charts represent an essential source of visual information in documents and facilitate a deep understanding and interpretation of information typically conveyed numerically. In the scientific literature, there are many charts, each with its stylistic differences. Recently the document understanding community has begun to address the problem of automatic chart understanding, which begins with chart classification. In this paper, we present a survey of the current state-of-the-art techniques for chart classification and discuss the available datasets and their supported chart types. We broadly classify these contributions as traditional approaches based on ML, CNN, and Transformers. Furthermore, we carry out an extensive comparative performance analysis of CNN-based and transformer-based approaches on the recently published CHARTINFO UB-UNITECH PMC dataset for the CHART-Infographics competition at ICPR 2022. The data set includes 15 different chart categories, including 22,923 training images and 13,260 test images. We have implemented a vision-based transformer model that produces state-of-the-art results in chart classification.
['David S Doermann', 'Mohammed Javed', 'Anurag Dhote']
2023-07-09
null
null
null
null
['classification-1']
['methodology']
[-1.12625413e-01 -3.88699055e-01 -4.51558352e-01 -1.89766377e-01 -5.73409081e-01 -8.07977498e-01 8.96966338e-01 5.21510065e-01 3.44548523e-01 3.40704232e-01 6.45213604e-01 -7.60679066e-01 2.03270555e-01 -5.66106081e-01 -5.40157139e-01 -2.55630672e-01 -2.72608757e-01 4.47512507e-01 -3.77964526e-01 -1.08794849e-02 6.03748679e-01 6.77063823e-01 -1.34627461e+00 9.05663669e-01 8.02998066e-01 1.65001059e+00 -3.22634935e-01 1.08091986e+00 -9.48345304e-01 1.39101624e+00 -1.08496296e+00 -8.84923100e-01 -4.12844330e-01 -4.22942519e-01 -7.48687148e-01 5.56846023e-01 9.79007006e-01 -2.56469250e-01 -3.23931485e-01 8.97437751e-01 3.81253958e-01 -2.37575531e-01 1.11763072e+00 -1.39810622e+00 -1.63542914e+00 3.71721447e-01 -8.82419169e-01 1.60004139e-01 8.41344714e-01 -3.11823457e-01 1.28349924e+00 -1.02615273e+00 1.03661990e+00 1.35218501e+00 7.16031790e-01 1.09723061e-01 -1.14572930e+00 -3.10855538e-01 3.17525625e-01 5.41928291e-01 -1.05064166e+00 -1.53959021e-01 8.97064149e-01 -9.33552742e-01 1.05616629e+00 3.23284477e-01 1.07965183e+00 1.05493975e+00 3.64113063e-01 1.36814988e+00 8.94591689e-01 -7.35886633e-01 3.58668417e-01 -1.00806266e-01 1.81843877e-01 1.03084302e+00 2.36514077e-01 -5.55764139e-01 -7.91799963e-01 5.08930907e-02 8.86733592e-01 -3.61377895e-01 -4.00831640e-01 -7.16833293e-01 -1.28203368e+00 7.59759009e-01 6.00039065e-01 3.38198960e-01 2.11125277e-02 -1.63779221e-02 9.38914180e-01 3.03529710e-01 5.47308683e-01 3.80682200e-01 -1.90706268e-01 -3.26522350e-01 -9.78827477e-01 4.29148436e-01 9.81122673e-01 1.53177750e+00 3.79773587e-01 2.11615145e-01 -6.47275388e-01 8.84149373e-01 2.82681018e-01 3.17199588e-01 2.80498341e-02 -6.07249439e-01 8.59787881e-01 9.57058787e-01 -1.83134019e-01 -1.18673193e+00 -3.57019067e-01 -1.62869051e-01 -1.03659046e+00 1.83580980e-01 3.04724365e-01 1.83072031e-01 -1.57313287e+00 5.03697872e-01 -3.86739343e-01 -7.69529045e-01 -3.41298163e-01 4.76621151e-01 1.35563767e+00 7.33231604e-01 -9.25189182e-02 1.31973371e-01 1.53514385e+00 -1.06405354e+00 -1.30125237e+00 2.72827327e-01 6.01939082e-01 -1.09194314e+00 1.24674153e+00 5.82192063e-01 -8.14886212e-01 -4.14827913e-01 -1.20151162e+00 -4.73471016e-01 -8.80950272e-01 6.49602056e-01 6.90454304e-01 9.30782735e-01 -1.12301743e+00 1.57660395e-01 -3.30429584e-01 -6.92031980e-01 1.04783082e+00 -3.81792694e-01 -3.44845116e-01 -1.10969305e-01 -5.49241364e-01 8.85942161e-01 1.95061490e-01 8.37576017e-02 -4.39654499e-01 -8.66001189e-01 -9.88824368e-01 2.34208316e-01 1.89424008e-01 -2.89538831e-01 1.58926857e+00 -4.32531595e-01 -1.16618419e+00 8.32992554e-01 -4.91810858e-01 -4.53633547e-01 9.54983294e-01 -3.08503747e-01 -5.31695068e-01 2.29009554e-01 8.72442573e-02 5.27939975e-01 5.79072356e-01 -1.53065932e+00 -3.42224330e-01 -1.91305697e-01 -5.69681898e-02 -5.90782166e-02 -1.88608333e-01 5.47983944e-02 -7.92932153e-01 -8.37713838e-01 -1.09849229e-01 -4.48506624e-01 3.66570681e-01 5.20541608e-01 -8.57723951e-01 -1.44254714e-01 1.14284921e+00 -8.48495185e-01 1.24257791e+00 -1.75186932e+00 -1.82323143e-01 1.89230591e-01 4.06353861e-01 -7.69739449e-02 2.56588548e-01 7.45154798e-01 -1.20724909e-01 4.01003122e-01 -3.04084748e-01 -5.09620547e-01 8.47545564e-02 -5.85146360e-02 -6.05523050e-01 2.35785201e-01 3.90647464e-02 1.12929046e+00 -4.38546926e-01 -7.80497193e-01 6.17833257e-01 5.47200322e-01 -1.30136479e-02 -1.95369627e-02 -2.91849971e-01 1.07329100e-01 -2.58140981e-01 1.34407890e+00 6.77995384e-01 -5.01759112e-01 -2.10249443e-02 -3.79960895e-01 -2.80824542e-01 7.04381317e-02 -8.18386316e-01 1.85857677e+00 -2.68350154e-01 1.74268234e+00 -2.67935455e-01 -5.60722470e-01 1.17359197e+00 3.39308590e-01 2.23015383e-01 -9.39978302e-01 2.36102983e-01 3.34699601e-02 -4.33007866e-01 -2.42331073e-01 9.54164386e-01 3.56764376e-01 2.50923395e-01 3.77576835e-02 1.58547640e-01 -4.72538650e-01 7.40089774e-01 3.20540011e-01 8.96418512e-01 1.05387196e-01 5.59385002e-01 -2.69536644e-01 6.63514674e-01 5.47323585e-01 -3.45301628e-01 6.76558137e-01 -1.13675497e-01 1.06707335e+00 9.80060756e-01 -8.25916648e-01 -1.26735735e+00 -1.01400423e+00 -3.44791621e-01 8.34732652e-01 -2.99330682e-01 -8.47636640e-01 -5.15646458e-01 -5.02878845e-01 2.76759028e-01 7.57995069e-01 -1.19607854e+00 7.06173718e-01 -3.93512011e-01 -3.60587478e-01 5.39123952e-01 1.01138258e+00 8.28239799e-01 -1.12566078e+00 -4.36627477e-01 -8.88307989e-02 1.80108115e-01 -1.13988972e+00 -3.17513019e-01 2.23587811e-01 -8.19183052e-01 -1.18947494e+00 -1.08560860e+00 -6.15635216e-01 7.47851789e-01 1.58402860e-01 1.44316351e+00 -4.14666533e-02 -2.83830136e-01 5.15819013e-01 -5.70156455e-01 -8.54617059e-01 -2.50511706e-01 2.63097525e-01 -8.11972022e-01 -2.68963337e-01 3.63155156e-01 5.09076342e-02 7.06556626e-03 -1.11230493e-01 -6.27591550e-01 1.40383035e-01 4.44375634e-01 6.09591663e-01 3.50147545e-01 -3.49088728e-01 -5.43932877e-02 -9.98405993e-01 1.11431158e+00 1.20932475e-01 -5.87333739e-01 5.41621804e-01 -5.67935765e-01 1.10808372e-01 4.65123773e-01 -3.19784544e-02 -1.28393793e+00 -8.39813948e-02 1.81011334e-01 -3.28742832e-01 -1.61153242e-01 7.69355834e-01 -1.38501361e-01 -3.77887972e-02 5.55154383e-01 1.41216174e-01 -3.91447783e-01 -5.86255908e-01 9.21353817e-01 5.56551754e-01 7.39967883e-01 -5.13468981e-01 7.50072300e-01 7.33235240e-01 -1.64453715e-01 -1.03984571e+00 -5.86525500e-01 -4.82372314e-01 -9.60936010e-01 -7.94804394e-01 1.06617880e+00 -6.49734557e-01 -8.08651507e-01 4.60827768e-01 -1.49025774e+00 -1.50082316e-02 -1.62947759e-01 -4.63244207e-02 -6.17105782e-01 9.52321216e-02 -5.85833490e-01 -8.07655275e-01 -3.27312976e-01 -1.18242228e+00 1.21757758e+00 2.30172351e-01 -4.15799886e-01 -1.29836857e+00 1.19224198e-01 3.33649516e-02 3.26897264e-01 7.69453943e-01 1.26515090e+00 -9.04092938e-02 -5.28681457e-01 -3.53448123e-01 -8.71618927e-01 -2.45476812e-02 1.42343804e-01 6.60214603e-01 -1.21386743e+00 1.03587955e-01 -6.81229889e-01 -4.32633400e-01 1.11440468e+00 6.41745269e-01 1.47398865e+00 -2.96170022e-02 -3.85696262e-01 6.58541918e-01 1.49710429e+00 7.22992003e-01 8.70339751e-01 7.76847601e-01 1.26022017e+00 4.99380857e-01 1.34141862e-01 4.10961211e-01 3.48284096e-01 1.26571119e-01 2.90965229e-01 -3.09175491e-01 -1.02541730e-01 -6.09788537e-01 -1.58000931e-01 9.18558657e-01 -4.02054548e-01 -5.43063462e-01 -1.25802982e+00 5.12586176e-01 -1.71540582e+00 -7.03092158e-01 -3.73580575e-01 1.33379829e+00 2.79481232e-01 3.35447997e-01 -8.36341605e-02 5.00856519e-01 7.61024058e-01 7.51069427e-01 -2.11247608e-01 -8.68191600e-01 -3.17480505e-01 -7.55960420e-02 6.78754807e-01 -1.25166908e-01 -1.32454336e+00 5.57699859e-01 7.40802670e+00 4.80951786e-01 -8.77624631e-01 -5.63784063e-01 8.06245685e-01 3.92389089e-01 -2.65567958e-01 -3.38414520e-01 -3.37959796e-01 -2.11585566e-01 6.30015850e-01 -4.00944680e-01 -4.39397246e-02 1.20195687e+00 -2.18308002e-01 -2.53877163e-01 -1.36349368e+00 1.36490822e+00 5.16336858e-01 -2.33499694e+00 7.03681529e-01 -1.78615391e-01 9.12266552e-01 -5.11996388e-01 2.53854215e-01 5.81126520e-03 1.27782792e-01 -1.46259344e+00 1.34630013e+00 3.15102667e-01 1.11495471e+00 -7.32431710e-01 5.73408306e-01 -5.96879840e-01 -1.69855332e+00 1.51790932e-01 -2.20710695e-01 3.38115729e-02 -5.06197549e-02 4.01282161e-01 -4.92710680e-01 9.43198919e-01 9.30689096e-01 1.38324571e+00 -1.14510012e+00 1.35744095e+00 -7.74311200e-02 4.43285018e-01 5.32361627e-01 -3.81858826e-01 4.52958345e-01 -1.66906357e-01 2.15508476e-01 1.71268487e+00 1.58107460e-01 -4.44686145e-01 -1.85772568e-01 1.03244865e+00 -4.18293029e-01 2.14807257e-01 -7.79094934e-01 -3.46894741e-01 1.42966256e-01 9.48257446e-01 -1.21393728e+00 -6.94307446e-01 -8.38446081e-01 5.09216666e-01 1.77338973e-01 6.33886755e-01 -4.80829746e-01 -5.73568940e-01 3.01773250e-01 -2.22916812e-01 4.89075005e-01 -3.45927566e-01 -1.00633597e+00 -1.17323530e+00 2.50243276e-01 -8.36254656e-01 2.68582880e-01 -1.24862492e+00 -1.22973883e+00 6.16542280e-01 1.21979095e-01 -1.58405221e+00 1.65126994e-02 -1.25609624e+00 -5.93566537e-01 6.31177723e-01 -1.32827330e+00 -1.15093720e+00 -7.01852322e-01 2.43198365e-01 8.00156176e-01 -2.92654842e-01 7.81584203e-01 1.81536421e-01 -2.24136442e-01 1.71213225e-01 3.77150744e-01 8.10021698e-01 7.24928796e-01 -1.74836612e+00 1.03962028e+00 4.96693909e-01 4.38204825e-01 2.51471877e-01 8.32506418e-01 -4.40797746e-01 -1.44640124e+00 -8.74154747e-01 5.27106524e-01 -7.33483315e-01 8.79419744e-01 -7.36198664e-01 -1.06685317e+00 8.99394810e-01 6.55011892e-01 -3.09873968e-01 3.43634307e-01 9.74435136e-02 -7.71375060e-01 1.05357751e-01 -6.37836576e-01 7.32400894e-01 7.63745070e-01 -8.21289778e-01 -6.76764488e-01 1.12386666e-01 3.97922546e-01 -8.65569055e-01 -7.69563556e-01 -9.61749405e-02 8.36483240e-01 -1.01430035e+00 6.45284534e-01 -2.63058335e-01 9.60212052e-01 -2.00059146e-01 8.93071219e-02 -1.29921532e+00 3.08846496e-02 -3.94761413e-01 -2.45029166e-01 1.18634641e+00 5.87979734e-01 -5.56256659e-02 8.76141548e-01 4.64756876e-01 -4.20057505e-01 -6.82545125e-01 -6.52873814e-01 -6.65157199e-01 8.54400247e-02 -7.82348514e-01 6.66388571e-01 9.13041949e-01 1.35752752e-01 1.26892701e-01 -1.42712384e-01 -6.23596191e-01 4.43072557e-01 8.77839774e-02 9.43403840e-01 -1.20387959e+00 7.12575436e-01 -8.58228147e-01 -6.24666810e-01 -9.01899457e-01 -1.73790187e-01 -7.58230388e-01 -3.24659824e-01 -2.51408434e+00 6.51035607e-02 4.42089647e-01 1.42878711e-01 1.17256798e-01 3.33195478e-01 -1.13142498e-01 6.11524582e-01 3.25694233e-01 -4.60300028e-01 5.35708308e-01 1.66368234e+00 -7.79367268e-01 -1.52420849e-01 -4.67876256e-01 -3.98897827e-01 5.54682076e-01 5.03154039e-01 5.87981604e-02 -3.66728783e-01 -3.99755657e-01 1.98405072e-01 6.11638054e-02 1.45526394e-01 -1.02114892e+00 1.41801924e-01 2.21135929e-01 1.24277627e+00 -1.62796569e+00 -1.05389562e-02 -6.31687939e-01 -4.13005054e-01 2.75558174e-01 -6.70723319e-01 6.32586777e-01 5.33956289e-01 6.11466646e-01 -6.51406467e-01 1.38051644e-01 4.32495415e-01 -1.13221548e-01 -8.38694036e-01 1.34524098e-03 -6.03233993e-01 -5.21491701e-03 5.69416702e-01 -4.21739995e-01 -9.74655867e-01 -6.42319739e-01 -4.47738677e-01 2.13933036e-01 3.74409795e-01 7.74489760e-01 8.36633325e-01 -1.58828306e+00 -5.09864748e-01 -1.47732208e-02 8.61403108e-01 -2.11528778e-01 -1.22353911e-01 3.84409636e-01 -1.27874196e+00 1.01243722e+00 -5.61470032e-01 -6.58637404e-01 -1.15361071e+00 5.36040545e-01 8.35846066e-02 -2.09248707e-01 -8.37839127e-01 5.67231715e-01 4.44561504e-02 -2.42670216e-02 4.89449888e-01 -1.11840582e+00 -6.30394518e-01 5.23483932e-01 7.12845981e-01 5.37094474e-01 3.41672927e-01 -4.71488148e-01 -3.68675798e-01 6.04667246e-01 -2.93337762e-01 -2.64581144e-01 8.97037745e-01 2.33299077e-01 -3.78273427e-02 9.41320181e-01 1.38421714e+00 -3.20277601e-01 -9.23213124e-01 1.71775445e-01 3.45070481e-01 -6.74617112e-01 -4.37127873e-02 -7.82069325e-01 -9.56108987e-01 1.14937758e+00 3.04634064e-01 6.62408650e-01 9.01032984e-01 -1.22339539e-01 1.92750648e-01 5.96507609e-01 -5.26719801e-02 -1.12781060e+00 1.71198428e-01 6.47360563e-01 1.51111209e+00 -1.11144137e+00 4.70005959e-01 -4.59806055e-01 -9.95861292e-01 1.73779094e+00 4.92699325e-01 -1.10805139e-01 5.32707274e-01 1.68210506e-01 2.81373650e-01 -5.45207381e-01 -6.52748108e-01 -7.10828975e-02 8.93178821e-01 1.01313353e+00 1.11388803e+00 1.92066878e-02 1.62853643e-01 -2.94841021e-01 -6.20104969e-01 -3.24236751e-02 7.95757830e-01 1.10465217e+00 -1.48447484e-01 -5.21717310e-01 -7.39718258e-01 9.26313162e-01 -8.53589401e-02 -1.23922965e-02 -8.75229597e-01 1.35428655e+00 -6.38471007e-01 6.52023613e-01 4.27781850e-01 -9.92261246e-02 4.54389453e-01 1.33753508e-01 4.31116670e-01 -1.50338396e-01 -2.62144148e-01 -2.97973827e-02 3.51289988e-01 -4.11977738e-01 -7.11784005e-01 -6.51277542e-01 -9.69922841e-01 -5.78324080e-01 1.11857496e-01 -1.43275395e-01 7.39844918e-01 5.74835598e-01 -1.22240365e-01 9.56546843e-01 -2.00272724e-01 -7.37506211e-01 2.90183663e-01 -9.84749198e-01 -6.76571429e-01 1.33686125e-01 6.56046867e-01 -7.54495502e-01 -4.23218310e-02 5.72892427e-01]
[11.33210277557373, 2.197071075439453]
7b915e0b-7af0-4cf2-bd1f-b6639f0f4138
detector-free-weakly-supervised-group
2204.02139
null
https://arxiv.org/abs/2204.02139v1
https://arxiv.org/pdf/2204.02139v1.pdf
Detector-Free Weakly Supervised Group Activity Recognition
Group activity recognition is the task of understanding the activity conducted by a group of people as a whole in a multi-person video. Existing models for this task are often impractical in that they demand ground-truth bounding box labels of actors even in testing or rely on off-the-shelf object detectors. Motivated by this, we propose a novel model for group activity recognition that depends neither on bounding box labels nor on object detector. Our model based on Transformer localizes and encodes partial contexts of a group activity by leveraging the attention mechanism, and represents a video clip as a set of partial context embeddings. The embedding vectors are then aggregated to form a single group representation that reflects the entire context of an activity while capturing temporal evolution of each partial context. Our method achieves outstanding performance on two benchmarks, Volleyball and NBA datasets, surpassing not only the state of the art trained with the same level of supervision, but also some of existing models relying on stronger supervision.
['Suha Kwak', 'Minsu Cho', 'Jinsung Lee', 'Dongkeun Kim']
2022-04-05
null
http://openaccess.thecvf.com//content/CVPR2022/html/Kim_Detector-Free_Weakly_Supervised_Group_Activity_Recognition_CVPR_2022_paper.html
http://openaccess.thecvf.com//content/CVPR2022/papers/Kim_Detector-Free_Weakly_Supervised_Group_Activity_Recognition_CVPR_2022_paper.pdf
cvpr-2022-1
['group-activity-recognition']
['computer-vision']
[ 2.48542905e-01 -2.72515923e-01 -4.14650857e-01 -3.15720677e-01 -4.68132019e-01 -6.55146897e-01 8.25869262e-01 3.05150300e-01 -4.83906984e-01 3.93809319e-01 5.25220513e-01 1.91102296e-01 1.12847701e-01 -4.86019462e-01 -7.24728823e-01 -7.43298531e-01 -2.17491806e-01 1.41270518e-01 4.20791626e-01 1.46167114e-01 7.31806457e-02 1.60942212e-01 -1.38526320e+00 4.71383601e-01 4.18544561e-01 1.21166301e+00 -7.74196610e-02 5.89650333e-01 2.47884214e-01 1.20793569e+00 -6.86509490e-01 -2.23564401e-01 1.36875406e-01 -5.69524944e-01 -4.14924592e-01 5.15200913e-01 6.54180288e-01 -3.72148305e-01 -7.17730343e-01 6.23262286e-01 1.52562872e-01 2.63297737e-01 5.27133763e-01 -1.28198755e+00 -6.20277166e-01 2.97712415e-01 -4.00861710e-01 7.15890944e-01 6.96595073e-01 2.02498883e-01 1.22023702e+00 -8.97238970e-01 6.15545571e-01 8.48299921e-01 5.00437140e-01 4.75539029e-01 -1.09914184e+00 -5.21964610e-01 7.81529903e-01 5.51106930e-01 -1.41419911e+00 -2.07104519e-01 8.50616992e-01 -6.71176672e-01 9.99144435e-01 -4.01637480e-02 1.17445934e+00 1.50925601e+00 -4.38074172e-02 9.53932881e-01 6.51453614e-01 -8.74775946e-02 3.89952987e-01 -2.07509965e-01 2.10492015e-01 8.64798009e-01 2.71635234e-01 -4.59047318e-01 -1.02980947e+00 -2.22302571e-01 7.63034701e-01 5.04654348e-01 -3.54918122e-01 -5.89952886e-01 -1.47947145e+00 7.24624038e-01 2.62642950e-01 3.30171704e-01 -4.39746648e-01 3.99903983e-01 3.04556966e-01 2.83027086e-02 4.66071308e-01 1.31140068e-01 -1.28848568e-01 -2.88399160e-01 -9.83804464e-01 4.16624904e-01 6.71642065e-01 6.59051478e-01 7.12496579e-01 -3.76283824e-01 -3.83537173e-01 4.53517169e-01 1.62087068e-01 -8.94855037e-02 3.27756435e-01 -7.59284019e-01 7.31348515e-01 1.20890355e+00 2.08759069e-01 -9.19865549e-01 -1.78031981e-01 -3.12146127e-01 -4.24629778e-01 -1.11632615e-01 6.19765520e-01 5.31086810e-02 -6.98830366e-01 1.78729618e+00 2.79748321e-01 7.44303465e-01 -4.30694401e-01 9.36569512e-01 4.74894553e-01 7.28204012e-01 -7.08131539e-03 -8.01028535e-02 1.43035269e+00 -1.46122909e+00 -6.47304714e-01 -4.29085195e-01 5.84115565e-01 7.01177642e-02 9.68922257e-01 4.51656044e-01 -9.02071655e-01 -6.46771252e-01 -1.21511054e+00 -9.60621908e-02 -2.83256292e-01 5.68517298e-02 6.11166537e-01 5.70798874e-01 -6.75081491e-01 5.09015024e-01 -9.94917810e-01 -4.40187246e-01 7.75864482e-01 1.56695530e-01 -7.13455677e-01 -2.15585396e-01 -6.69183552e-01 4.51907009e-01 4.77807112e-02 1.45792086e-02 -1.33320391e+00 -5.40447652e-01 -9.74422395e-01 1.25171423e-01 7.01493442e-01 -5.37947893e-01 9.20416534e-01 -9.61292267e-01 -1.12204492e+00 7.15057194e-01 -1.66039765e-01 -3.95022005e-01 4.88691777e-01 -5.12993813e-01 -4.10877168e-01 3.40228200e-01 6.58088876e-03 2.77993590e-01 7.43846536e-01 -8.32512319e-01 -7.31220543e-01 -4.66133356e-01 3.40987593e-01 2.18099385e-01 -6.72232389e-01 2.64711261e-01 -6.46160543e-01 -6.91832721e-01 -1.73633337e-01 -8.59011471e-01 -1.19949050e-01 2.30314955e-01 -3.18409428e-02 -5.72408438e-01 9.89971042e-01 -6.63869500e-01 1.54523599e+00 -2.12199163e+00 3.42670649e-01 -1.05466008e-01 1.90132931e-01 6.02893829e-02 -9.34424996e-02 5.90785086e-01 2.86983494e-02 1.29980324e-02 -1.06711820e-01 -5.34011364e-01 -1.64891966e-02 2.63614506e-01 -2.51758009e-01 8.62878919e-01 2.89243639e-01 8.03025365e-01 -1.11472726e+00 -5.01141489e-01 7.22492039e-02 3.54533076e-01 -7.02533543e-01 3.68187457e-01 -1.86014637e-01 5.55730104e-01 -4.02342290e-01 7.90938973e-01 4.15485501e-02 -4.45848674e-01 3.90762866e-01 -1.20292529e-02 -5.06687835e-02 2.37818807e-01 -1.22837818e+00 2.09408164e+00 2.72576585e-02 6.07548356e-01 -2.10226730e-01 -1.35889912e+00 5.34751415e-01 4.97897625e-01 7.83578932e-01 -2.68455207e-01 -8.30054730e-02 -9.13515314e-02 -2.53125690e-02 -6.40455842e-01 2.42034525e-01 2.92118698e-01 -4.29716617e-01 7.06107020e-01 4.42328602e-01 7.10948169e-01 4.68163699e-01 1.67660147e-01 1.57349753e+00 3.12796056e-01 5.14381647e-01 8.00646022e-02 4.72999752e-01 -3.13444197e-01 8.26726615e-01 9.28604364e-01 -5.30166805e-01 5.93396366e-01 7.41845071e-01 -7.30865955e-01 -8.77976060e-01 -9.23250556e-01 4.51006442e-01 1.45501769e+00 2.64669806e-01 -7.91043222e-01 -6.11415327e-01 -1.06342530e+00 -7.27240145e-02 -2.64178589e-02 -1.07912803e+00 -6.43042773e-02 -9.33901250e-01 -6.44480348e-01 2.65964538e-01 1.00039446e+00 4.38682556e-01 -9.83539581e-01 -6.99215710e-01 2.08290726e-01 -2.67528236e-01 -1.38818574e+00 -7.83272684e-01 -1.38975844e-01 -7.13420093e-01 -1.34437215e+00 -5.13375580e-01 -6.59421980e-01 6.35649621e-01 2.90270537e-01 1.02767789e+00 2.01047093e-01 -3.05010617e-01 6.43615365e-01 -4.16051835e-01 -3.58340025e-01 3.20130467e-01 -1.27854913e-01 1.03838302e-01 7.87484109e-01 4.57462519e-01 -6.88710451e-01 -8.37106943e-01 5.46128452e-01 -6.69216633e-01 -1.21480837e-01 4.32778835e-01 6.72978640e-01 6.02398813e-01 -1.87481046e-01 4.13870215e-01 -4.75998402e-01 2.97003180e-01 -5.79164207e-01 -2.12481081e-01 3.77575725e-01 -9.33272690e-02 -3.78110617e-01 5.24433792e-01 -7.40839005e-01 -7.85808921e-01 9.31030437e-02 5.22782624e-01 -5.91198742e-01 -1.34885445e-01 2.71017283e-01 -3.69741827e-01 2.59034246e-01 4.36955333e-01 1.33870915e-01 -3.54154646e-01 -6.26701653e-01 1.16234526e-01 3.68728369e-01 5.79775870e-01 -3.86226118e-01 5.78724027e-01 9.09642637e-01 -1.20016053e-01 -6.46166801e-01 -1.05281448e+00 -8.51092160e-01 -8.60076189e-01 -5.38420081e-01 1.11016500e+00 -1.09692657e+00 -5.94295204e-01 3.17977488e-01 -9.86308932e-01 -2.93866456e-01 -2.01903775e-01 5.61404645e-01 -5.37630796e-01 4.75430846e-01 -4.64163095e-01 -8.13421547e-01 1.70571774e-01 -7.93730736e-01 1.10431540e+00 1.67531848e-01 -2.72827268e-01 -7.77322888e-01 2.00291857e-01 7.14127541e-01 -1.85545251e-01 4.69075471e-01 4.11429405e-01 -7.53080189e-01 -8.61879945e-01 -6.30875468e-01 1.04173943e-01 2.72890866e-01 8.20562392e-02 -1.39790371e-01 -8.59544694e-01 -2.73753405e-01 -1.28107533e-01 -2.30873242e-01 9.01776671e-01 1.54209539e-01 1.40704799e+00 -3.28310728e-01 -4.70924675e-01 4.57011342e-01 1.09365261e+00 1.10867426e-01 4.77475613e-01 1.83321178e-01 7.73424864e-01 4.50734973e-01 5.01686037e-01 5.86037874e-01 3.72180134e-01 9.50836658e-01 5.21678030e-01 2.16394857e-01 1.47380844e-01 -4.50614184e-01 5.95506668e-01 4.43060637e-01 -5.68711579e-01 -3.54763389e-01 -7.07202554e-01 8.44432056e-01 -2.26674867e+00 -1.45738339e+00 2.62863457e-01 2.10006189e+00 3.27954322e-01 3.59204710e-02 5.98711312e-01 2.51052111e-01 6.14223123e-01 7.82086372e-01 -4.66625929e-01 1.81138571e-02 1.67751625e-01 -5.82898818e-02 9.05936286e-02 -7.66157806e-02 -1.46134663e+00 5.08020878e-01 6.25011110e+00 4.70116466e-01 -5.96052468e-01 3.08057010e-01 2.27136925e-01 -6.39116168e-01 2.20178396e-01 1.62352771e-02 -7.24612236e-01 6.91991448e-01 8.86493146e-01 5.56405075e-02 3.55027705e-01 8.35470736e-01 1.97940752e-01 -8.52363706e-02 -1.58162713e+00 1.02059305e+00 6.42226636e-01 -1.29447246e+00 -1.42096713e-01 3.17632467e-01 6.78665996e-01 -2.89399445e-01 -2.14442372e-01 4.05019075e-01 2.19220091e-02 -9.47182417e-01 9.63640988e-01 5.88550150e-01 2.42261380e-01 -3.79799426e-01 5.57373226e-01 5.39244473e-01 -1.34544313e+00 -6.15760148e-01 -7.80978724e-02 -4.55014795e-01 4.51527119e-01 1.78140238e-01 -4.63789970e-01 3.66167635e-01 7.92010307e-01 1.11186910e+00 -6.11256301e-01 9.55433309e-01 -2.91272461e-01 9.09442544e-01 -2.04767600e-01 1.40132338e-01 2.18595743e-01 -1.97903112e-01 4.85444427e-01 1.20378041e+00 2.58250594e-01 1.12624668e-01 5.65418124e-01 5.76745629e-01 -2.85220414e-01 9.29395296e-03 -5.41006625e-01 -3.80510062e-01 3.06541860e-01 1.13700175e+00 -6.03843987e-01 -6.04000211e-01 -9.29598093e-01 1.00611913e+00 5.49651325e-01 2.55789071e-01 -1.13348126e+00 5.77984937e-02 7.65761256e-01 3.70549887e-01 5.67569613e-01 -3.75640869e-01 2.82090828e-02 -1.44613194e+00 4.23225820e-01 -7.22220719e-01 7.63013959e-01 -5.65727174e-01 -1.06491566e+00 2.94850677e-01 3.78073491e-02 -1.41179895e+00 -1.03737891e-01 -5.88613689e-01 -8.95137429e-01 3.50867808e-01 -1.05032718e+00 -1.24049187e+00 -7.47679532e-01 6.75581396e-01 6.35833383e-01 -2.10658446e-01 6.30835533e-01 3.61984938e-01 -1.00565183e+00 3.31337720e-01 -4.92447406e-01 4.81947213e-01 5.95801532e-01 -1.02554369e+00 2.46032596e-01 1.07660055e+00 5.30340552e-01 5.99860013e-01 4.79086876e-01 -5.13312638e-01 -1.20904469e+00 -9.61716056e-01 9.70844924e-01 -9.61620331e-01 7.70596564e-01 -8.45576882e-01 -8.84780049e-01 1.05588830e+00 3.13326567e-02 6.58219099e-01 1.08761394e+00 2.51786321e-01 -5.51788151e-01 -1.53727531e-01 -6.99393094e-01 5.00886738e-01 1.61577559e+00 -6.86751604e-01 -9.08917606e-01 2.80900002e-01 3.62619728e-01 -7.82820359e-02 -7.90519416e-01 8.04019794e-02 7.08937883e-01 -1.04634309e+00 1.02704442e+00 -1.01587081e+00 4.67194915e-01 -4.58004624e-01 -1.06107540e-01 -8.33856285e-01 -4.03771400e-01 -4.57641870e-01 -7.17559636e-01 9.89214420e-01 -1.31954670e-01 -2.19238743e-01 9.91674900e-01 4.91057724e-01 -2.01111346e-01 -7.17669010e-01 -9.69451606e-01 -8.89600575e-01 -6.52343988e-01 -3.60841155e-01 5.46506047e-01 8.37093592e-01 2.49456972e-01 1.12416387e-01 -6.20645642e-01 1.57652527e-01 6.37594938e-01 7.63416663e-02 9.50794756e-01 -1.07378554e+00 -4.63388205e-01 -2.18682259e-01 -8.59718144e-01 -1.11626768e+00 1.01366401e-01 -6.14757359e-01 -1.71806857e-01 -1.43532610e+00 5.15674174e-01 1.54424012e-01 -6.35650277e-01 5.65567255e-01 -2.22381011e-01 4.95260805e-01 2.18155622e-01 2.68400759e-01 -1.30869114e+00 4.05439883e-01 8.31793487e-01 -2.03320608e-01 -1.19872764e-01 -1.52131140e-01 -6.10302687e-01 8.87079954e-01 4.29570228e-01 -4.60644841e-01 -4.33491886e-01 -3.34479809e-01 6.15548268e-02 -1.46189779e-01 7.72802234e-01 -1.33263028e+00 2.03172535e-01 -2.94538200e-01 3.49540710e-01 -4.68437612e-01 5.51844358e-01 -6.88520610e-01 5.24657443e-02 1.20787613e-01 -3.37859184e-01 -2.15228945e-02 -3.37353379e-01 1.19798279e+00 -3.32069427e-01 8.21129978e-02 4.21262413e-01 -1.53210923e-01 -9.40856338e-01 6.37434304e-01 -4.03403491e-01 -5.42293303e-02 1.28390193e+00 -4.25771385e-01 -2.59880602e-01 -2.93867618e-01 -8.32852364e-01 1.32573768e-01 4.46942568e-01 5.06636262e-01 3.46340120e-01 -1.53163159e+00 -4.59740579e-01 1.22344136e-01 5.13467371e-01 -2.01402575e-01 4.64305431e-01 9.34182405e-01 -2.23901391e-01 3.11793178e-01 -2.34710306e-01 -5.06459057e-01 -1.34860945e+00 6.62797749e-01 2.10838765e-01 -4.95183259e-01 -7.77939975e-01 7.90276766e-01 4.83181566e-01 3.08407933e-01 4.01815504e-01 -4.90973234e-01 -3.54469568e-01 2.07956105e-01 9.80041683e-01 3.58708739e-01 -2.30074465e-01 -7.53616691e-01 -6.27664387e-01 5.12708485e-01 7.37144127e-02 4.29936945e-02 1.37869895e+00 9.43097323e-02 2.56201506e-01 6.43833816e-01 9.99208927e-01 -8.22291970e-02 -1.86410046e+00 -2.88381964e-01 5.66163473e-03 -6.91700876e-01 -2.81591207e-01 -4.76693839e-01 -8.62552881e-01 8.92640650e-01 2.45367840e-01 1.87455937e-02 8.83108974e-01 2.34993458e-01 6.82293713e-01 2.59364784e-01 4.70928520e-01 -1.32709467e+00 7.76955247e-01 4.68236953e-02 8.33156228e-01 -1.05587077e+00 1.96381256e-01 -2.10700333e-01 -6.97139084e-01 7.82441676e-01 4.17559355e-01 -4.67221260e-01 4.54693377e-01 -6.96155131e-02 -3.21997434e-01 -2.45668188e-01 -1.02431726e+00 -1.90844521e-01 4.19796467e-01 6.13145828e-01 1.77971944e-01 -1.14070304e-01 -1.80161640e-01 1.07830536e+00 4.69118237e-01 3.01019341e-01 3.37623835e-01 1.16749096e+00 -4.06617463e-01 -8.79345298e-01 -2.44190812e-01 3.99604619e-01 -5.78869998e-01 4.69147801e-01 -3.78226876e-01 7.00323880e-01 5.38362861e-01 8.82939517e-01 2.86008239e-01 -4.46761698e-01 4.38671827e-01 2.23328754e-01 5.88717103e-01 -8.18807781e-01 -5.96725285e-01 -1.30398050e-01 1.16731577e-01 -9.09501374e-01 -6.54641926e-01 -1.02352083e+00 -7.19337046e-01 5.31211495e-02 1.86045736e-01 -7.81542510e-02 2.24005893e-01 9.41465557e-01 2.32470945e-01 2.86928117e-01 3.61449540e-01 -1.11989200e+00 -1.85985327e-01 -9.95147347e-01 -7.66542792e-01 9.12263453e-01 3.67990315e-01 -9.97774839e-01 -3.67453635e-01 3.66326421e-01]
[8.238384246826172, 0.6250066757202148]
27630196-3e82-4ea5-9111-4d10e8515b91
consistent-jumpy-predictions-for-videos-and
1807.02033
null
http://arxiv.org/abs/1807.02033v3
http://arxiv.org/pdf/1807.02033v3.pdf
Consistent Generative Query Networks
Stochastic video prediction models take in a sequence of image frames, and generate a sequence of consecutive future image frames. These models typically generate future frames in an autoregressive fashion, which is slow and requires the input and output frames to be consecutive. We introduce a model that overcomes these drawbacks by generating a latent representation from an arbitrary set of frames that can then be used to simultaneously and efficiently sample temporally consistent frames at arbitrary time-points. For example, our model can "jump" and directly sample frames at the end of the video, without sampling intermediate frames. Synthetic video evaluations confirm substantial gains in speed and functionality without loss in fidelity. We also apply our framework to a 3D scene reconstruction dataset. Here, our model is conditioned on camera location and can sample consistent sets of images for what an occluded region of a 3D scene might look like, even if there are multiple possibilities for what that region might contain. Reconstructions and videos are available at https://bit.ly/2O4Pc4R.
['S. M. Ali Eslami', 'Edward Lockhart', 'Fabio Viola', 'Murray Shanahan', 'Marta Garnelo', 'Danilo J. Rezende', 'Ananya Kumar']
2018-07-05
null
null
null
iclr-2019-5
['3d-scene-reconstruction']
['computer-vision']
[ 3.38265687e-01 2.10385188e-03 -8.04910213e-02 -1.94180548e-01 -7.60949910e-01 -6.98695064e-01 6.51397407e-01 -4.55538690e-01 1.59947313e-02 5.40130258e-01 3.27868789e-01 -2.04203710e-01 4.02395308e-01 -6.49441004e-01 -1.06503105e+00 -4.90921170e-01 -9.02469233e-02 1.12414867e-01 4.65449959e-01 3.17629367e-01 1.24628715e-01 4.86893654e-01 -1.68599129e+00 6.75962448e-01 1.63319036e-01 7.21796155e-01 6.64631546e-01 1.27581847e+00 2.10241020e-01 1.23004401e+00 -2.55613595e-01 -3.45134437e-02 4.64683473e-01 -5.65004885e-01 -6.93105042e-01 8.08219671e-01 6.06100321e-01 -1.02973902e+00 -6.04764640e-01 8.04155052e-01 1.09124027e-01 1.81789219e-01 3.13596487e-01 -1.01085043e+00 -2.58983433e-01 1.23119548e-01 -3.34742785e-01 1.55916229e-01 9.23011363e-01 5.86267471e-01 7.64094055e-01 -9.29739773e-01 1.00654829e+00 1.23716998e+00 4.07563269e-01 5.79481840e-01 -1.52219987e+00 -3.92412275e-01 3.46586883e-01 1.99961722e-01 -1.22852683e+00 -8.87115419e-01 7.39400387e-01 -4.78323489e-01 7.76361287e-01 2.09897295e-01 9.68766809e-01 1.26318669e+00 1.94297537e-01 8.35893333e-01 6.77825212e-01 -3.15528512e-01 3.08014095e-01 -2.36284569e-01 -3.44550639e-01 6.14588261e-01 -3.20454806e-01 2.11710885e-01 -6.95383608e-01 -1.61489099e-01 1.24747300e+00 3.05949301e-01 -6.31453812e-01 -3.58948767e-01 -1.40308034e+00 4.42404360e-01 -2.27629617e-01 -1.90633491e-01 -6.25121057e-01 4.12470549e-01 -4.37992513e-02 2.98268765e-01 3.72594744e-01 -4.05511037e-02 -2.36854255e-01 -2.91955113e-01 -1.23186100e+00 4.22941327e-01 5.78018486e-01 1.18935168e+00 8.35909545e-01 1.98927239e-01 -1.00617699e-01 4.20556068e-01 1.71875924e-01 5.21775305e-01 1.12249449e-01 -1.91107774e+00 2.63693064e-01 -1.62248462e-01 4.35391247e-01 -7.78529644e-01 1.79213509e-01 3.16249132e-01 -4.39799041e-01 4.22261983e-01 4.79524225e-01 -2.92131692e-01 -7.99445450e-01 1.55743349e+00 3.78648996e-01 8.47684264e-01 -5.44053093e-02 9.10364151e-01 3.09227198e-01 1.13965905e+00 -2.87152201e-01 -3.96297187e-01 8.86049867e-01 -8.27222228e-01 -3.36831510e-01 -1.99742839e-01 2.33804315e-01 -9.23269928e-01 9.21881020e-01 4.05304462e-01 -1.61361516e+00 -7.72892535e-01 -6.64574623e-01 -1.32501557e-01 4.73194540e-01 -4.45081182e-02 3.17894250e-01 1.28837347e-01 -1.33536720e+00 6.24027610e-01 -1.01626289e+00 -1.25177205e-01 8.26271549e-02 7.73421377e-02 -3.27190191e-01 -1.96437076e-01 -6.67590678e-01 3.62271547e-01 1.25557363e-01 -3.50383408e-02 -1.49045491e+00 -5.41857183e-01 -7.91826129e-01 2.58244500e-02 3.00662279e-01 -8.54747057e-01 1.55718124e+00 -1.27662849e+00 -1.59906876e+00 5.81663787e-01 -8.37055326e-01 -5.71577847e-01 6.80992723e-01 -3.08050841e-01 -1.58784196e-01 6.33694768e-01 -4.89015016e-04 9.72880602e-01 1.15923083e+00 -1.23844123e+00 -7.66776025e-01 8.20191279e-02 3.75421852e-01 3.29166532e-01 3.59381825e-01 7.56089166e-02 -8.85582030e-01 -5.66263080e-01 1.68791324e-01 -1.12062931e+00 -3.84791404e-01 3.38861197e-01 -2.87848651e-01 3.17821026e-01 7.84349263e-01 -5.35661340e-01 8.38211954e-01 -2.08199120e+00 1.89579397e-01 -6.43923283e-02 -1.11357905e-01 -1.83088541e-01 -1.08154856e-01 3.28352600e-01 -1.42785057e-01 1.08595595e-01 4.26114500e-02 -5.12278199e-01 -4.45877075e-01 2.26122573e-01 -7.50677586e-01 4.35304880e-01 1.93711922e-01 5.56236327e-01 -1.00099969e+00 -4.28161025e-01 6.40505612e-01 6.56747222e-01 -8.88407886e-01 3.56648117e-01 -6.46673858e-01 8.45903516e-01 -2.87585169e-01 4.07819003e-01 4.31283236e-01 -3.67564887e-01 2.34937459e-01 -9.43098366e-02 -2.31699303e-01 1.98463917e-01 -1.21389508e+00 1.71490002e+00 -3.78729433e-01 1.03201675e+00 -1.39354527e-01 -5.78638494e-01 5.34347773e-01 6.64120138e-01 6.59175634e-01 -2.18399197e-01 -1.42742947e-01 -1.95866421e-01 -3.76914233e-01 -5.76396823e-01 6.77793264e-01 1.07563056e-01 3.95337701e-01 5.17828763e-01 -1.92915171e-01 -3.30377460e-01 2.05097869e-01 2.41800740e-01 1.04690802e+00 6.19845390e-01 -9.02467221e-03 1.10665552e-01 2.57158399e-01 -4.37849686e-02 6.07891202e-01 8.17910135e-01 -1.02421656e-01 1.09919631e+00 5.80470622e-01 -5.96271455e-01 -1.48875475e+00 -1.23843896e+00 1.83754444e-01 5.30367076e-01 3.10526311e-01 -5.40585995e-01 -5.03472686e-01 -2.15704471e-01 -4.33971554e-01 8.17061841e-01 -3.04564774e-01 2.60702521e-01 -7.03449428e-01 1.47429675e-01 -1.10529281e-01 2.00102761e-01 4.85357530e-02 -1.06077385e+00 -9.83675778e-01 3.04391593e-01 -3.92203391e-01 -1.25053823e+00 -5.60948670e-01 -2.65297562e-01 -1.10561621e+00 -1.05002081e+00 -5.92630684e-01 -6.77519321e-01 8.50859284e-01 6.12021387e-01 1.26891232e+00 7.93479457e-02 -5.83403818e-02 6.14082098e-01 -1.72524020e-01 1.12648323e-01 -5.12662470e-01 -5.47841370e-01 4.66728657e-02 1.39271975e-01 -9.39009935e-02 -6.41885996e-01 -8.03299785e-01 2.48850107e-01 -8.78578603e-01 6.37633979e-01 2.04837248e-02 7.10497200e-01 9.56542552e-01 6.89847693e-02 -6.19581826e-02 -7.63430536e-01 -3.33504542e-03 -6.55821741e-01 -7.17467010e-01 -1.73931401e-02 6.65872693e-02 -1.23035692e-01 7.44723439e-01 -4.64213103e-01 -1.11177135e+00 4.11798358e-01 7.62193203e-02 -1.15404356e+00 -3.27128083e-01 1.88617408e-02 2.24743560e-01 4.68595922e-01 3.65689874e-01 5.04207671e-01 -1.37133047e-03 -2.71157265e-01 3.23242128e-01 1.83655038e-01 5.91052234e-01 -5.89313626e-01 5.21514237e-01 8.08791220e-01 -1.92366809e-01 -8.02507281e-01 -4.44852203e-01 -7.61257261e-02 -5.29531956e-01 -5.32681465e-01 6.87243938e-01 -1.29900491e+00 -5.20504713e-01 3.04825783e-01 -1.31250358e+00 -7.64378548e-01 -3.68271023e-01 5.00083208e-01 -9.82319534e-01 3.39172870e-01 -7.31081009e-01 -7.75948167e-01 1.70553520e-01 -1.40674746e+00 1.21282005e+00 9.95370373e-02 -4.02186751e-01 -8.48990142e-01 -2.86042452e-01 -3.47962640e-02 -4.79715317e-03 1.33680567e-01 4.36061591e-01 2.36157298e-01 -1.19262576e+00 -3.30467485e-02 2.16951758e-01 1.25297248e-01 1.53893992e-01 5.62489092e-01 -7.74722695e-01 -3.14949930e-01 8.42594132e-02 -3.60491276e-02 4.43262428e-01 7.00322270e-01 1.33386302e+00 -4.47200209e-01 -1.96813881e-01 6.13724351e-01 1.35825539e+00 3.70921165e-01 7.46955335e-01 -8.46855994e-03 4.30455059e-01 4.63822216e-01 5.35899758e-01 5.88218629e-01 3.58141929e-01 7.58859873e-01 2.83444136e-01 2.07484439e-01 -1.52761713e-01 -5.96835732e-01 6.75638020e-01 5.02946079e-01 -1.18660927e-01 -3.61533523e-01 -6.81184947e-01 7.00752676e-01 -1.76803112e+00 -1.50013566e+00 1.57314669e-02 2.27377462e+00 5.97994268e-01 5.12501262e-02 7.24793077e-02 -1.95188135e-01 7.18056560e-01 3.30184668e-01 -6.65013134e-01 -7.85627216e-02 -4.42775972e-02 -5.93748800e-02 4.29127753e-01 8.46455574e-01 -7.74997175e-01 8.26771319e-01 6.93572569e+00 4.23728943e-01 -1.09377646e+00 -3.13572347e-01 9.56731975e-01 -6.21941805e-01 -5.83342195e-01 4.60311234e-01 -6.53991401e-01 7.35065103e-01 1.07048774e+00 -2.19516933e-01 5.52855670e-01 6.64258182e-01 7.28309691e-01 -4.00351971e-01 -1.30270863e+00 9.86118257e-01 -1.86009035e-01 -1.75614560e+00 1.93744451e-01 -4.40205447e-02 8.40579152e-01 1.95889752e-02 2.21537054e-02 -1.44881949e-01 5.11429250e-01 -7.38313377e-01 1.20868254e+00 8.04078102e-01 7.60775805e-01 -4.67229307e-01 -2.58540045e-02 4.80687410e-01 -1.05215442e+00 -4.07058150e-02 -3.59969795e-01 -2.19231650e-01 6.94873095e-01 3.33190978e-01 -7.51949728e-01 1.40659958e-01 8.30398142e-01 8.55352759e-01 -1.74159229e-01 7.87343442e-01 -3.22691232e-01 7.02421069e-01 -3.93684030e-01 4.39027965e-01 3.07474043e-02 -3.31552982e-01 6.16041541e-01 8.85958910e-01 7.81882703e-01 3.27870697e-01 2.99012661e-01 7.51889884e-01 1.48839459e-01 -4.06962484e-01 -9.91550982e-01 3.44962180e-01 7.27150261e-01 7.34455466e-01 -5.25515676e-01 -6.43753171e-01 -6.69123113e-01 1.25966048e+00 -2.27442682e-02 7.56145597e-01 -9.66009915e-01 2.59489745e-01 7.61245728e-01 2.87274748e-01 4.64047045e-01 -5.62462509e-01 1.13447465e-01 -1.52950215e+00 3.89121436e-02 -8.10310006e-01 3.70785967e-02 -1.35681820e+00 -7.15460896e-01 3.49266350e-01 5.26685733e-03 -1.61014712e+00 -8.16779733e-01 -9.15616304e-02 -5.30060172e-01 6.58582985e-01 -1.24784160e+00 -6.37120247e-01 -2.01019272e-01 7.84401834e-01 1.17764473e+00 1.34727344e-01 6.29135013e-01 1.52738951e-02 -1.87365517e-01 -6.23830482e-02 2.33304482e-02 -1.18433572e-01 4.27476466e-01 -8.63708794e-01 6.54418886e-01 1.12063396e+00 3.34430426e-01 3.53421777e-01 8.21573675e-01 -5.58222592e-01 -1.49643803e+00 -9.66754258e-01 6.74281120e-01 -4.60260004e-01 4.15015519e-01 -1.49096087e-01 -6.33073330e-01 1.10563421e+00 1.40604898e-01 1.70494795e-01 2.99896896e-01 -4.90721226e-01 -4.35573608e-02 1.46619692e-01 -9.49811161e-01 1.03667152e+00 1.10659409e+00 -6.68914676e-01 -1.35819495e-01 3.00344229e-01 7.14674592e-01 -7.30097950e-01 -6.34456277e-01 -1.09046614e-02 6.56140685e-01 -1.59876323e+00 1.11363304e+00 -3.26000184e-01 7.90307581e-01 -5.80194652e-01 -3.90459597e-01 -9.79377687e-01 -1.84947133e-01 -9.57701027e-01 -4.25288856e-01 8.68725657e-01 1.05082966e-01 -2.82700330e-01 9.80200291e-01 9.71566498e-01 4.39142436e-02 -5.49050093e-01 -7.96444356e-01 -5.78410745e-01 -2.47434080e-01 -7.82762766e-01 4.85608965e-01 4.19904917e-01 -2.64337510e-01 3.33473608e-02 -8.15934837e-01 4.98333544e-01 6.03880584e-01 3.64269048e-01 1.06234229e+00 -5.14987230e-01 -5.24602413e-01 -5.89908250e-02 -2.72714674e-01 -1.72038817e+00 5.57166152e-02 -3.33131373e-01 1.36842662e-02 -1.18364918e+00 -1.47584872e-02 -2.89199650e-01 2.30404079e-01 1.95922226e-01 4.24685469e-03 1.49569929e-01 4.91652340e-01 5.92550099e-01 -5.23053348e-01 3.35444123e-01 1.18527126e+00 2.30716079e-01 -1.99139535e-01 -4.97544035e-02 -8.32950473e-02 8.38295698e-01 6.07251525e-01 -2.07662433e-01 -7.22337723e-01 -8.09470177e-01 -1.34293303e-01 9.19414282e-01 6.23094857e-01 -1.02190542e+00 1.21097080e-01 -4.35643584e-01 7.26005316e-01 -5.78194976e-01 8.15006733e-01 -9.01923776e-01 9.62441385e-01 2.37994999e-01 -3.77210259e-01 2.19799712e-01 -4.99939360e-03 6.89487994e-01 -1.21663563e-01 -1.88916698e-01 7.13846803e-01 -5.14802873e-01 -8.68378401e-01 5.09593904e-01 -6.88266873e-01 -3.44920039e-01 1.18197453e+00 -5.70428014e-01 1.05781503e-01 -7.91370749e-01 -9.01860535e-01 1.42547458e-01 1.15819633e+00 3.48583281e-01 8.34157825e-01 -1.22710311e+00 -5.36723673e-01 4.06298637e-01 -2.75301695e-01 2.58967757e-01 4.98956174e-01 3.68652195e-01 -7.60530472e-01 3.97959910e-02 -3.43934894e-02 -1.02230668e+00 -1.21268117e+00 6.19565308e-01 2.59599179e-01 1.17110543e-01 -9.64598775e-01 5.74190140e-01 3.04870695e-01 2.98440039e-01 8.49222485e-03 -3.09617281e-01 1.89347208e-01 -2.92987376e-01 7.44537652e-01 7.77807087e-02 -5.87560534e-01 -7.30397761e-01 9.50499102e-02 5.14937997e-01 6.84984550e-02 -4.02337551e-01 1.26232862e+00 -5.64996421e-01 2.33829528e-01 6.31159902e-01 1.21227658e+00 9.58531573e-02 -2.24905372e+00 -1.16970107e-01 -3.90989214e-01 -9.34179366e-01 -2.04068929e-01 -1.12429157e-01 -9.70610559e-01 6.52436137e-01 3.22093666e-01 1.30560711e-01 1.13453233e+00 -9.78465099e-03 7.92846143e-01 5.46266027e-02 6.67408347e-01 -6.68923378e-01 -2.46235151e-02 3.96407783e-01 6.54299557e-01 -1.06192148e+00 -6.01144731e-02 -2.65550345e-01 -5.83022118e-01 1.36788464e+00 3.03180814e-01 -3.22977334e-01 5.00795364e-01 2.74262875e-01 -4.12939861e-02 5.89799061e-02 -1.35773206e+00 1.27647191e-01 -1.38327658e-01 5.70063412e-01 3.81381869e-01 -1.40771925e-01 2.62006581e-01 -1.53011382e-01 -1.16925657e-01 2.77167916e-01 9.22815442e-01 8.28716218e-01 -2.73622394e-01 -9.64098752e-01 -4.76148218e-01 1.57765627e-01 -3.67104322e-01 -1.71168130e-02 3.17140460e-01 3.66532773e-01 -1.22590557e-01 9.02340949e-01 3.54662359e-01 -1.48392007e-01 -3.40331979e-02 -4.65554260e-02 5.89045525e-01 -5.71718037e-01 9.48497802e-02 3.45168263e-01 -2.38629784e-02 -9.24313366e-01 -6.45072699e-01 -1.10566366e+00 -1.07404172e+00 -3.30793440e-01 9.27160233e-02 -5.86626120e-02 2.37319723e-01 5.10107696e-01 6.09249353e-01 1.40167207e-01 9.26052034e-01 -1.51797438e+00 6.30618185e-02 -4.43751216e-01 -2.43389875e-01 5.04435360e-01 7.74793386e-01 -2.00349748e-01 -2.69806057e-01 9.48194921e-01]
[9.637214660644531, -2.103463888168335]
29357dcf-c3e3-433c-8af1-2ad2a9df18a0
balanced-training-of-energy-based-models-with
2306.00684
null
https://arxiv.org/abs/2306.00684v3
https://arxiv.org/pdf/2306.00684v3.pdf
Balanced Training of Energy-Based Models with Adaptive Flow Sampling
Energy-based models (EBMs) are versatile density estimation models that directly parameterize an unnormalized log density. Although very flexible, EBMs lack a specified normalization constant of the model, making the likelihood of the model computationally intractable. Several approximate samplers and variational inference techniques have been proposed to estimate the likelihood gradients for training. These techniques have shown promising results in generating samples, but little attention has been paid to the statistical accuracy of the estimated density, such as determining the relative importance of different classes in a dataset. In this work, we propose a new maximum likelihood training algorithm for EBMs that uses a different type of generative model, normalizing flows (NF), which have recently been proposed to facilitate sampling. Our method fits an NF to an EBM during training so that an NF-assisted sampling scheme provides an accurate gradient for the EBMs at all times, ultimately leading to a fast sampler for generating new data.
['Marylou Gabrié', 'Éric Moulines', 'Louis Grenioux']
2023-06-01
null
null
null
null
['density-estimation']
['methodology']
[ 1.59215033e-01 -1.85475918e-03 -4.71721381e-01 -4.44513917e-01 -5.94709277e-01 -2.64267117e-01 7.56438732e-01 -5.66690415e-02 -4.10949737e-01 1.12983072e+00 7.62818158e-02 -3.50621015e-01 3.99117172e-02 -1.22775578e+00 -7.79858947e-01 -7.75532603e-01 2.78197765e-01 6.08172834e-01 2.18355581e-01 3.71444613e-01 2.53819764e-01 5.62739432e-01 -1.65324056e+00 -2.10016057e-01 1.15076542e+00 6.69108868e-01 4.59002316e-01 7.81397581e-01 -4.64535952e-01 6.22237623e-01 -7.95005441e-01 -5.20107090e-01 -2.09756896e-01 -7.94039965e-01 -4.06350166e-01 -2.48842418e-01 3.30300719e-01 -4.97732073e-01 -9.81239453e-02 1.04003549e+00 4.19854105e-01 3.84733558e-01 1.36166668e+00 -1.18230796e+00 -2.16937587e-01 5.86676061e-01 -3.46908599e-01 1.76141098e-01 4.24188524e-02 -1.61363721e-01 7.59382367e-01 -8.10817599e-01 4.29289877e-01 1.15833616e+00 4.60051566e-01 6.61154628e-01 -1.54833949e+00 -7.73689389e-01 -1.09050691e-01 1.95064858e-01 -1.56762922e+00 -4.16519403e-01 6.62056267e-01 -3.67420614e-01 6.20854974e-01 2.78226405e-01 9.30019915e-01 1.09248233e+00 1.32420370e-02 8.44215453e-01 9.21122730e-01 -6.69175267e-01 8.34451914e-01 6.54449403e-01 -2.66525358e-01 5.67133725e-01 4.48826879e-01 -1.29750386e-01 -7.21389532e-01 -4.33113217e-01 1.02025867e+00 -3.16562027e-01 -1.76639706e-01 -4.48310882e-01 -5.10885656e-01 1.19045675e+00 2.68764138e-01 2.82200456e-01 -2.86707968e-01 4.25445139e-01 1.42666534e-01 -5.51773250e-01 4.82883543e-01 -5.15282489e-02 4.49997839e-03 -3.37340117e-01 -1.39928079e+00 4.47746694e-01 9.42561686e-01 7.57396698e-01 1.04049432e+00 2.20037594e-01 -3.61638427e-01 8.62836063e-01 6.07471228e-01 7.13690281e-01 3.09324622e-01 -1.04985738e+00 2.79876798e-01 2.97362179e-01 1.68764979e-01 -5.63679636e-01 1.93317518e-01 -4.28949207e-01 -8.86256337e-01 2.64977157e-01 5.57594955e-01 -5.02434885e-03 -7.80046105e-01 2.08486295e+00 5.82300663e-01 3.47829819e-01 -1.93172842e-01 4.72084463e-01 2.05203861e-01 9.07798707e-01 3.87694687e-01 -2.24522293e-01 1.03091991e+00 -4.61474240e-01 -7.55944669e-01 -4.13994603e-02 2.11860999e-01 -4.23053682e-01 1.13587582e+00 4.01005149e-01 -1.03695571e+00 -4.00517911e-01 -1.02774048e+00 4.31315377e-02 -1.80412576e-01 3.42516936e-02 6.53945684e-01 1.22526538e+00 -1.00547624e+00 8.54744434e-01 -9.77677524e-01 -1.83198005e-01 5.28886616e-01 2.07371160e-01 3.06006849e-01 1.41594023e-01 -1.08045483e+00 7.80947089e-01 3.86598438e-01 -1.43667385e-01 -1.00361955e+00 -7.23094702e-01 -7.54448056e-01 2.29787737e-01 -1.06233105e-01 -7.46981561e-01 1.31655407e+00 -7.13621259e-01 -1.84206438e+00 2.18171820e-01 -6.76240563e-01 -5.80187082e-01 5.32751083e-01 -9.30410475e-02 -4.61086296e-02 1.13119990e-01 -8.36900473e-02 9.44087088e-01 1.05930710e+00 -1.12028956e+00 -3.09425473e-01 1.35176286e-01 -1.83319628e-01 1.60617024e-01 -4.63977069e-01 -3.72920722e-01 -2.23358154e-01 -5.43840349e-01 -4.17568713e-01 -4.81365681e-01 -9.19858590e-02 1.87494367e-01 -2.82654434e-01 -2.23821878e-01 6.28676772e-01 -3.61859977e-01 1.51684713e+00 -1.90411997e+00 -2.40684226e-02 3.63579661e-01 -5.85178323e-02 3.96907300e-01 2.04885513e-01 2.86100537e-01 4.54581112e-01 1.91636726e-01 -5.03171384e-01 -5.58692098e-01 1.49764061e-01 5.60065866e-01 -4.20905411e-01 2.96148121e-01 2.85032779e-01 6.76722586e-01 -1.03347719e+00 -7.56216764e-01 5.44423759e-01 9.78101790e-01 -8.90059531e-01 3.76466423e-01 -2.79178202e-01 1.75927222e-01 -3.14024538e-01 2.31765345e-01 8.46926332e-01 -2.64708936e-01 2.43009597e-01 -2.31541559e-01 -5.28951026e-02 3.88087839e-01 -1.28976953e+00 1.27358031e+00 -5.55416763e-01 4.45283234e-01 -1.03089787e-01 -7.56047070e-01 8.43189895e-01 2.37261504e-01 3.27757210e-01 -1.96570560e-01 1.18712790e-01 3.09517205e-01 -2.33025864e-01 4.98156920e-02 6.55602455e-01 -3.93704325e-01 3.73850167e-01 5.28453827e-01 3.24603796e-01 -3.26200426e-01 5.57114124e-01 1.85582578e-01 4.74341244e-01 4.29802954e-01 4.42465842e-01 -3.47464979e-01 4.55280900e-01 -3.39507908e-01 3.19084406e-01 8.34433317e-01 8.59054849e-02 6.96946919e-01 2.87886620e-01 2.91684140e-02 -1.20430899e+00 -1.43850100e+00 -4.24240559e-01 6.48981273e-01 -6.60714507e-02 -4.93806392e-01 -1.16647387e+00 -4.71684307e-01 -1.87899262e-01 1.05477631e+00 -3.42918426e-01 -2.75050879e-01 -2.91160345e-01 -1.09757340e+00 5.07583737e-01 4.64858353e-01 4.71298635e-01 -8.31761897e-01 -5.09307146e-01 3.60183835e-01 -3.00545931e-01 -7.21623361e-01 -5.25198340e-01 2.67270058e-01 -1.24529004e+00 -9.37122226e-01 -8.80520523e-01 -2.03644365e-01 7.87027955e-01 -3.92477423e-01 9.80649590e-01 -3.16073477e-01 -2.13412449e-01 1.92988306e-01 1.08684395e-02 -3.89453590e-01 -8.29729199e-01 2.28516117e-01 1.00962371e-01 -2.40922091e-03 1.84710786e-01 -7.74526834e-01 -6.06197357e-01 2.10077330e-01 -9.87767458e-01 7.40126446e-02 5.72799325e-01 5.61650574e-01 6.92017734e-01 -2.21159056e-01 6.73626065e-01 -8.66593361e-01 7.00058639e-01 -4.31287229e-01 -6.82315767e-01 1.34763911e-01 -8.73815954e-01 5.06786883e-01 7.01052010e-01 -7.20358968e-01 -1.25523221e+00 -7.50885755e-02 -5.11353791e-01 -5.40214658e-01 -6.06425889e-02 1.27786949e-01 -4.88490462e-02 7.81860724e-02 7.39063144e-01 4.04714823e-01 3.53069194e-02 -4.72552121e-01 3.63070637e-01 5.87921739e-01 3.07203799e-01 -8.18923056e-01 6.10777199e-01 2.61219412e-01 9.98134911e-02 -8.51750970e-01 -6.82780623e-01 -2.21667618e-01 -4.23329353e-01 -3.17475945e-01 7.42316663e-01 -6.75122857e-01 -4.81547296e-01 3.94715548e-01 -9.24191177e-01 -4.98235852e-01 -6.94900692e-01 5.61926484e-01 -5.36660731e-01 3.36891443e-01 -4.40404803e-01 -1.35963094e+00 -3.85389596e-01 -9.65695083e-01 9.17063832e-01 5.67836761e-01 -2.58295655e-01 -1.26032460e+00 3.71369153e-01 8.46232288e-03 7.92372406e-01 -8.45963061e-02 8.35971117e-01 -1.25808865e-01 -6.18305624e-01 -2.11635530e-02 -4.12909985e-02 5.60506582e-01 2.67088264e-01 3.14738601e-01 -1.18819928e+00 -2.05249399e-01 -1.66714713e-01 -1.73298508e-01 6.60812020e-01 8.11583102e-01 1.11234641e+00 -4.15710479e-01 -4.13759291e-01 4.09366310e-01 1.40959442e+00 3.32558714e-02 8.43179941e-01 -1.35914311e-01 3.67266029e-01 -6.49997592e-02 2.09537268e-01 6.57525063e-01 3.99531648e-02 4.40073609e-01 1.97860673e-01 9.32516381e-02 -9.33923647e-02 -6.26861632e-01 5.13772130e-01 7.62351274e-01 -4.04793248e-02 -4.77975518e-01 -3.36647838e-01 4.25739884e-01 -1.49080491e+00 -9.96810853e-01 -6.50772378e-02 2.50647855e+00 1.24695408e+00 1.19475566e-01 1.23368949e-01 8.12603682e-02 7.56365299e-01 1.41182765e-01 -6.98218465e-01 -3.27959955e-01 3.25480610e-01 6.21760011e-01 3.59273911e-01 6.39428139e-01 -6.99533284e-01 5.88365674e-01 7.58118200e+00 1.30318081e+00 -8.77136171e-01 2.54885852e-02 6.86556399e-01 -7.93839842e-02 -5.64065158e-01 9.01941210e-02 -1.13291836e+00 6.45448208e-01 1.24936712e+00 -3.30805480e-01 4.63041097e-01 8.95794809e-01 2.72991508e-01 -6.52670085e-01 -9.13822830e-01 7.67208636e-01 -2.08928704e-01 -1.15263033e+00 3.73082608e-01 2.04625219e-01 5.81930220e-01 -3.00371289e-01 -5.52077368e-02 2.45258108e-01 2.11569041e-01 -7.69339561e-01 7.60028183e-01 7.53578961e-01 8.85231316e-01 -9.20155585e-01 4.05784905e-01 5.86161017e-01 -1.15903604e+00 3.88531536e-01 -5.10591149e-01 1.78903237e-01 4.24605131e-01 1.03005266e+00 -1.07523787e+00 1.08655699e-01 3.16924036e-01 1.63300112e-01 -2.99929500e-01 1.05431414e+00 -2.44063929e-01 9.89539027e-01 -7.04537928e-01 -3.77354056e-01 -9.01387036e-02 -5.50564826e-01 5.14532804e-01 1.10246134e+00 5.59030771e-01 -4.78073359e-01 -3.28116059e-01 1.61275220e+00 -1.38591021e-01 6.18690848e-02 -2.03929588e-01 -2.40021676e-01 6.90402448e-01 1.34476316e+00 -7.67267585e-01 -4.01098162e-01 -7.77845159e-02 7.28240013e-01 4.13456768e-01 3.64373922e-01 -9.74284589e-01 -2.32209519e-01 6.15943193e-01 3.95391047e-01 2.84775257e-01 -1.59980685e-01 9.41246077e-02 -1.07309055e+00 -4.09819335e-01 -3.80874217e-01 1.81008115e-01 -6.48021460e-01 -1.22094846e+00 2.30039299e-01 4.64807361e-01 -9.89653885e-01 -6.52075350e-01 -5.20765722e-01 -5.77326417e-01 1.08638358e+00 -1.39722884e+00 -4.72445726e-01 -1.33039847e-01 2.50881284e-01 3.57334316e-01 1.02915309e-01 8.37935567e-01 3.95557582e-01 -5.52751899e-01 4.94064182e-01 2.10734934e-01 -2.13922620e-01 5.52418411e-01 -1.46743333e+00 1.90927744e-01 7.26999044e-01 3.08704376e-01 7.50111461e-01 8.08908224e-01 -6.57659769e-01 -8.75839233e-01 -9.21416640e-01 6.71374798e-01 -2.12233394e-01 2.42581338e-01 -4.28772628e-01 -9.47604597e-01 2.61915326e-01 -1.28867023e-03 -2.41235703e-01 7.92351842e-01 -6.91132843e-02 1.31609663e-01 -4.73527536e-02 -1.26073384e+00 4.19968963e-01 7.15165555e-01 -4.79538709e-01 -1.43131211e-01 -5.59397694e-03 1.95841700e-01 -3.19757104e-01 -5.63997686e-01 1.36879757e-01 4.87618148e-01 -1.13281465e+00 8.90006065e-01 -7.85033405e-02 9.73249972e-02 -4.14347678e-01 4.37031798e-02 -1.20862627e+00 -1.99605763e-01 -2.70994186e-01 -7.08590925e-01 1.40106428e+00 3.30179542e-01 -6.46461427e-01 9.99758780e-01 5.19616246e-01 2.63137370e-01 -6.69211268e-01 -9.36103702e-01 -6.00063086e-01 -1.71809103e-02 -5.93732655e-01 5.41183889e-01 4.02703881e-01 -4.06019151e-01 1.35274842e-01 -4.01129663e-01 -1.17735386e-01 8.13981652e-01 -2.26457328e-01 6.36456430e-01 -1.25763011e+00 -3.05839658e-01 -4.79208469e-01 -1.10179536e-01 -1.27069128e+00 1.16453037e-01 -8.81662786e-01 2.46318374e-02 -1.63609028e+00 4.62937534e-01 -3.60407054e-01 5.95595092e-02 1.25805855e-01 -2.85289854e-01 1.84927344e-01 -3.09753746e-01 8.35456923e-02 -1.96593076e-01 8.27223837e-01 1.01547253e+00 1.40839040e-01 -1.19252354e-01 1.99917018e-01 -4.56943274e-01 4.56586450e-01 7.33934760e-01 -6.36339307e-01 -8.30747545e-01 1.01721972e-01 1.81416005e-01 -3.56792688e-01 3.71931344e-01 -1.09989166e+00 1.39577523e-01 -9.20451134e-02 6.11408293e-01 -5.94542980e-01 3.17968845e-01 -5.08005559e-01 5.52883387e-01 3.23414326e-01 -2.01494560e-01 -4.85853791e-01 1.23024739e-01 6.74166739e-01 -1.54226124e-01 -8.39601338e-01 9.49026823e-01 6.16118684e-02 -2.36627892e-01 3.08368206e-01 -6.54236674e-01 1.86055183e-01 6.14399135e-01 -1.83866039e-01 1.78498596e-01 -5.73971331e-01 -4.03393745e-01 -1.92704573e-01 3.78545225e-01 -1.57373458e-01 4.41868663e-01 -1.41508234e+00 -2.76390404e-01 1.87945992e-01 -3.76805604e-01 3.35368901e-01 7.27967620e-02 5.33041656e-01 -5.57336032e-01 1.03489466e-01 7.04681650e-02 -6.87431216e-01 -8.00033152e-01 1.48782849e-01 5.93342721e-01 -4.62035507e-01 -3.26505810e-01 5.54125965e-01 -8.72129351e-02 -3.42054367e-01 1.61666602e-01 -3.19268256e-01 -5.95994852e-02 9.13270786e-02 5.61070263e-01 5.56002319e-01 -1.34576604e-01 -2.78118759e-01 -8.32600072e-02 2.94677913e-01 1.38925642e-01 -4.45500910e-01 9.99461114e-01 6.35327622e-02 1.88462779e-01 6.24616802e-01 9.75815058e-01 -1.53050777e-02 -1.58899903e+00 -1.36727905e-02 -2.54239976e-01 -5.21496236e-01 2.72108316e-01 -3.87076646e-01 -8.35996807e-01 9.57494020e-01 5.32226503e-01 1.97699070e-01 8.86135995e-01 -2.27430552e-01 6.57817185e-01 -3.12657394e-02 4.03200775e-01 -1.17998111e+00 -1.78486276e-02 2.72611827e-01 3.11701268e-01 -8.53134871e-01 1.69330642e-01 -3.92396927e-01 -3.89369905e-01 1.10855687e+00 3.31665576e-01 7.22738132e-02 5.75549126e-01 4.52863961e-01 -4.15226161e-01 2.72251397e-01 -4.70347196e-01 -3.95730287e-02 4.50927645e-01 6.43479705e-01 4.44335312e-01 -1.76097989e-01 -3.03267241e-01 3.11144322e-01 -1.53589159e-01 3.36272418e-01 3.01858544e-01 7.13412583e-01 -4.92097616e-01 -1.29892743e+00 -2.12368339e-01 5.32539248e-01 -2.92075276e-01 -2.11152852e-01 5.48567809e-02 6.60902202e-01 -7.55607337e-02 5.77179253e-01 3.10392529e-01 1.18728697e-01 -2.65815370e-02 5.07467330e-01 7.72180080e-01 -3.50382030e-01 -8.06409679e-03 6.21314086e-02 -8.93916339e-02 -2.51878738e-01 -4.61562485e-01 -7.51390934e-01 -1.22876298e+00 -3.70559990e-01 -6.42130613e-01 5.02061069e-01 7.11612165e-01 8.21009159e-01 1.47123933e-01 4.24162835e-01 3.61710519e-01 -1.05150509e+00 -4.50261474e-01 -9.97523010e-01 -8.32614422e-01 2.12130278e-01 -3.87545414e-02 -7.68653691e-01 -5.71033597e-01 -9.06969234e-03]
[7.041081428527832, 3.8995015621185303]
32ef5f03-8f83-4f47-85bf-09fe04c7d6db
generative-models-improve-radiomics-1
2109.02252
null
https://arxiv.org/abs/2109.02252v1
https://arxiv.org/pdf/2109.02252v1.pdf
Generative Models Improve Radiomics Performance in Different Tasks and Different Datasets: An Experimental Study
Radiomics is an active area of research focusing on high throughput feature extraction from medical images with a wide array of applications in clinical practice, such as clinical decision support in oncology. However, noise in low dose computed tomography (CT) scans can impair the accurate extraction of radiomic features. In this article, we investigate the possibility of using deep learning generative models to improve the performance of radiomics from low dose CTs. We used two datasets of low dose CT scans -NSCLC Radiogenomics and LIDC-IDRI - as test datasets for two tasks - pre-treatment survival prediction and lung cancer diagnosis. We used encoder-decoder networks and conditional generative adversarial networks (CGANs) trained in a previous study as generative models to transform low dose CT images into full dose CT images. Radiomic features extracted from the original and improved CT scans were used to build two classifiers - a support vector machine (SVM) and a deep attention based multiple instance learning model - for survival prediction and lung cancer diagnosis respectively. Finally, we compared the performance of the models derived from the original and improved CT scans. Encoder-decoder networks and CGANs improved the area under the curve (AUC) of survival prediction from 0.52 to 0.57 (p-value<0.01). On the other hand, Encoder-decoder network and CGAN can improve the AUC of lung cancer diagnosis from 0.84 to 0.88 and 0.89 respectively (p-value<0.01). Moreover, there are no statistically significant differences in improving AUC by using encoder-decoder network and CGAN (p-value=0.34) when networks trained at 75 and 100 epochs. Generative models can improve the performance of low dose CT-based radiomics in different tasks. Hence, denoising using generative models seems to be a necessary pre-processing step for calculating radiomic features from low dose CTs.
['Leonard Wee', 'Andre Dekker', 'Inigo Bermejo', 'Junhua Chen']
2021-09-06
null
null
null
null
['deep-attention', 'lung-cancer-diagnosis', 'deep-attention']
['computer-vision', 'medical', 'natural-language-processing']
[ 3.08031261e-01 2.33533323e-01 -9.64532793e-02 -3.85074198e-01 -1.21976566e+00 -1.60395294e-01 5.05118966e-01 9.85710844e-02 -6.41086102e-01 8.39086771e-01 3.87378722e-01 -4.76762027e-01 -2.21367136e-01 -1.12892818e+00 -6.76893115e-01 -1.10980308e+00 1.51509121e-01 7.39546180e-01 8.36697593e-02 -3.95787843e-02 -2.76851505e-01 5.33036232e-01 -1.01907575e+00 8.80953550e-01 5.48642993e-01 7.33695030e-01 4.99415845e-01 1.08484101e+00 8.53672773e-02 9.33965802e-01 -4.10947561e-01 -1.22333728e-01 2.46181991e-02 -7.43174195e-01 -6.70186639e-01 -3.87959361e-01 -7.26717636e-02 -2.95858830e-01 -5.98717093e-01 6.90264046e-01 9.12601054e-01 -3.58228654e-01 1.05909133e+00 -9.05223191e-01 -5.54089367e-01 6.34416521e-01 -1.06094442e-01 3.98596436e-01 7.93533400e-02 1.38626426e-01 3.43673587e-01 -5.24092495e-01 2.73501575e-01 7.75225580e-01 9.88745093e-01 8.11057091e-01 -8.43400180e-01 -5.69600523e-01 -7.69310117e-01 -5.57733327e-02 -1.03243673e+00 7.12166820e-03 2.68972605e-01 -4.70884323e-01 8.87439609e-01 5.04654288e-01 6.10305011e-01 1.41680253e+00 1.07040393e+00 3.06348979e-01 1.24411857e+00 -2.05242887e-01 -1.32461429e-01 1.83584034e-01 -3.08408767e-01 8.58733356e-01 2.15080008e-01 6.24735415e-01 5.24139814e-02 -2.29689702e-02 9.47486401e-01 4.04003501e-01 -3.47207695e-01 3.52969944e-01 -1.15674996e+00 9.62483764e-01 9.58890975e-01 7.63427317e-01 -5.09625137e-01 3.73194546e-01 4.96765167e-01 3.03672999e-01 3.37428063e-01 2.38524109e-01 -2.88523734e-01 3.49153906e-01 -7.75372624e-01 -2.14984640e-01 4.41730440e-01 4.32546616e-01 -9.18184370e-02 -2.99954806e-02 -5.59326768e-01 6.46717310e-01 2.84614027e-01 4.99100804e-01 1.19978940e+00 -2.55195564e-03 2.54084378e-01 2.87929118e-01 -5.58074176e-01 -2.98247606e-01 -9.53126848e-01 -8.61784697e-01 -1.34465015e+00 -1.50145516e-01 2.83446252e-01 -1.67553440e-01 -1.23543704e+00 1.53887665e+00 -2.23800465e-02 3.14131975e-01 2.68303603e-01 8.06966901e-01 1.28501487e+00 5.31292796e-01 6.01517141e-01 -1.64000764e-01 1.55122542e+00 -7.15135038e-01 -4.62172657e-01 3.35324034e-02 1.12104475e+00 -6.88896537e-01 8.41753602e-01 -1.18332259e-01 -8.37440848e-01 -4.42042798e-01 -8.76906276e-01 -9.57787037e-03 -1.83680236e-01 2.77401209e-01 4.74490225e-01 8.64904702e-01 -8.50965321e-01 7.55464792e-01 -1.09034979e+00 -3.23123515e-01 7.57064521e-01 6.87132180e-01 -3.59686762e-01 -3.35355222e-01 -1.22962189e+00 9.06855464e-01 1.54655814e-01 -2.73264498e-01 -1.28203773e+00 -1.04133248e+00 -4.41506922e-01 9.54781547e-02 -1.67089328e-01 -1.33733988e+00 1.18342257e+00 -8.50535750e-01 -1.28887939e+00 8.38784933e-01 3.43276471e-01 -5.56013405e-01 4.91107047e-01 7.68866539e-02 -3.36432874e-01 2.63606645e-02 1.92548692e-01 5.13034046e-01 4.70961601e-01 -8.07145655e-01 -4.68514889e-01 -5.87286115e-01 -5.12777269e-01 2.02018976e-01 3.31471488e-02 -2.48582914e-01 3.09584457e-02 -7.82710671e-01 1.37502626e-01 -1.02471805e+00 -3.41799736e-01 -2.65340626e-01 -3.54319692e-01 1.54331163e-01 4.90229368e-01 -8.91583979e-01 6.91206336e-01 -1.82050657e+00 -6.62661493e-02 1.71366304e-01 1.82116941e-01 2.06711004e-03 4.15073931e-02 -5.54124415e-02 -4.92406368e-01 4.60310847e-01 -2.87111044e-01 1.09438531e-01 -5.17200172e-01 1.72244534e-01 3.99721205e-01 5.73811471e-01 1.13400146e-01 1.41913116e+00 -5.64711511e-01 -4.58889306e-01 2.58854806e-01 6.66460335e-01 -2.46179760e-01 2.93032229e-01 4.32275608e-02 7.67619967e-01 -5.11211693e-01 4.57637399e-01 4.27633166e-01 -1.04722805e-01 -1.63571775e-01 -2.66274989e-01 3.92873228e-01 1.46879163e-02 -1.49963319e-01 1.56702411e+00 -7.60597944e-01 4.92588878e-01 -5.23632169e-01 -8.99173796e-01 5.68809152e-01 7.62280583e-01 8.51677597e-01 -1.08443046e+00 4.96475190e-01 1.57632947e-01 3.89548093e-01 -6.71025753e-01 -2.65692830e-01 -1.11931360e+00 -4.16057929e-02 2.68175721e-01 1.31131500e-01 -2.51310378e-01 -4.15100902e-01 -1.23606078e-01 1.39711380e+00 -4.31077570e-01 1.72038868e-01 -6.93927780e-02 5.77304304e-01 -3.15746553e-02 2.61740647e-02 5.72980940e-01 -2.80801468e-02 9.15552318e-01 4.61668581e-01 -1.39567703e-01 -1.06949341e+00 -1.01698756e+00 -5.95592320e-01 6.45759463e-01 -5.51363766e-01 1.56812549e-01 -6.80846393e-01 -8.56161296e-01 -1.83853760e-01 8.28247368e-01 -9.68583703e-01 -4.56004977e-01 -4.22938883e-01 -1.52638447e+00 8.30498695e-01 7.86631584e-01 4.21918243e-01 -9.50223207e-01 -3.18914115e-01 2.20096305e-01 -1.13401271e-01 -9.64519262e-01 -1.48153290e-01 7.74029613e-01 -1.26890707e+00 -1.04823697e+00 -1.05590391e+00 -6.08511388e-01 7.39345312e-01 -1.93177372e-01 1.05163968e+00 2.40543738e-01 -5.17039895e-01 2.30493903e-01 -2.85623610e-01 -5.20872891e-01 -1.03783107e+00 1.69139549e-01 -3.18166226e-01 -4.84915048e-01 1.10463053e-01 -3.65467489e-01 -5.26293099e-01 2.57107466e-01 -1.06678963e+00 1.08528011e-01 1.19925964e+00 1.04322231e+00 7.95737028e-01 8.75747502e-02 6.78377330e-01 -1.28032112e+00 5.07838786e-01 -7.87992716e-01 -3.49021778e-02 -4.78734896e-02 -5.24875760e-01 2.52154946e-01 8.43377590e-01 -1.72963142e-01 -8.37657273e-01 4.34127413e-02 -5.99323928e-01 -2.35041693e-01 -1.47859812e-01 4.62351441e-01 1.27121344e-01 -2.68610921e-02 1.12421191e+00 1.33752197e-01 -3.85643356e-02 -1.15476079e-01 -2.45897487e-01 5.60972631e-01 4.43121701e-01 -2.17282828e-02 6.64149284e-01 1.93886846e-01 6.28041029e-01 -4.69916493e-01 -9.01672125e-01 -3.67415160e-01 -5.53753436e-01 1.05237015e-01 1.28006470e+00 -9.51011300e-01 -3.50218058e-01 2.17025116e-01 -7.02065229e-01 -2.62731135e-01 -2.80054063e-01 8.35184336e-01 -4.99089748e-01 6.83648931e-03 -7.77861357e-01 -1.17180355e-01 -7.90059209e-01 -1.24107730e+00 8.79171789e-01 2.21721917e-01 3.54404338e-02 -1.11809158e+00 1.82984531e-01 3.75865340e-01 5.75484455e-01 6.22245967e-01 1.14619422e+00 -7.94222534e-01 -2.27713555e-01 -6.17805243e-01 -1.88636646e-01 2.88268298e-01 1.72583446e-01 -3.69508654e-01 -1.21742678e+00 -2.05527663e-01 2.71571487e-01 -2.24204287e-01 9.69769835e-01 8.89816642e-01 1.42080188e+00 1.70374159e-02 -5.39960504e-01 7.39875436e-01 1.63543212e+00 4.00464505e-01 1.00661969e+00 1.36809275e-01 6.48102820e-01 3.14499289e-02 1.27974495e-01 1.65355459e-01 -1.79180473e-01 3.99723321e-01 5.39216995e-01 -2.79470474e-01 -4.60031837e-01 -1.79813012e-01 2.29903072e-01 4.00974035e-01 -1.59187689e-01 -5.07353008e-01 -9.77065504e-01 2.11906090e-01 -1.16999722e+00 -9.68386471e-01 -6.83611333e-01 2.10427356e+00 5.75384498e-01 1.79992273e-01 -1.98675796e-01 3.15697134e-01 5.12319803e-01 -3.64301890e-01 -3.53513867e-01 -2.93369889e-01 1.92786157e-02 7.40403235e-01 7.13620663e-01 1.07803285e-01 -9.11640525e-01 3.11126441e-01 5.28703451e+00 7.54559636e-01 -1.30997634e+00 5.09468079e-01 1.05552995e+00 -1.67061463e-01 -3.03074419e-02 -3.13636512e-01 -3.80722880e-01 4.77403998e-01 1.53313577e+00 5.73826171e-02 -1.34249821e-01 6.68292224e-01 1.38040543e-01 -3.36084336e-01 -1.11374497e+00 8.50353062e-01 6.01586588e-02 -1.26424325e+00 -4.05272208e-02 7.25759715e-02 6.59122050e-01 3.08102220e-01 2.41375878e-01 3.82998228e-01 5.95369525e-02 -1.67046642e+00 3.45220380e-02 7.01068103e-01 1.12529075e+00 -8.54455411e-01 1.52389729e+00 4.34526414e-01 -7.76320398e-01 7.79941157e-02 -3.28094989e-01 5.31513453e-01 -1.84898998e-03 6.54202819e-01 -1.48611474e+00 7.54181683e-01 5.10340333e-01 2.78159589e-01 -7.75397182e-01 8.30483973e-01 -4.84427176e-02 7.75152206e-01 -2.30391189e-01 1.36563759e-02 1.75322384e-01 3.30124766e-01 -3.35827060e-02 1.08228230e+00 5.65178037e-01 3.65914077e-01 -2.92922854e-01 5.93721211e-01 -1.27665490e-01 3.44284736e-02 -5.24042487e-01 5.43975085e-02 -1.13928795e-01 1.19191802e+00 -9.20189321e-01 -2.95956671e-01 -3.31773072e-01 8.92276108e-01 -2.81669259e-01 -1.93816438e-01 -1.31777096e+00 -5.34579530e-03 -2.35936612e-01 6.60173595e-01 -1.65949866e-01 4.24142033e-01 -6.13111436e-01 -7.18177199e-01 -6.42370880e-01 -5.93846381e-01 5.29109120e-01 -9.29977715e-01 -1.09468043e+00 9.66697335e-01 -1.65583774e-01 -1.20273519e+00 -5.56822643e-02 -4.61309493e-01 -7.80625224e-01 9.29015934e-01 -1.18707883e+00 -1.06706274e+00 -7.11514533e-01 4.77418274e-01 3.80404621e-01 -1.75029919e-01 9.81538355e-01 2.63319403e-01 -4.63655353e-01 5.65151691e-01 2.34682158e-01 1.34108588e-01 5.86358726e-01 -1.14741123e+00 -5.57807237e-02 3.40205431e-01 7.40224670e-04 -1.96689159e-01 2.74480700e-01 -6.25737488e-01 -9.80084300e-01 -1.54728973e+00 6.60655677e-01 -6.00305021e-01 -1.05881663e-02 1.09131195e-01 -7.38856435e-01 5.14320433e-01 -9.77375507e-02 2.18296796e-02 9.60001111e-01 -6.29975140e-01 3.91480893e-01 2.19228446e-01 -1.51169991e+00 9.57036987e-02 6.38941169e-01 -2.33571470e-01 -3.26525867e-01 6.08332157e-01 4.62098956e-01 -4.76843268e-01 -1.31721616e+00 6.56414568e-01 3.72096479e-01 -8.65098357e-01 1.24218357e+00 -4.55517590e-01 7.82227814e-01 1.94848374e-01 -9.75033343e-02 -1.31720853e+00 -5.36685765e-01 3.23505998e-01 4.98326063e-01 6.01677477e-01 7.22455323e-01 -5.24084330e-01 8.78125846e-01 3.64141524e-01 -4.57501352e-01 -1.09120297e+00 -9.65976954e-01 -3.74965608e-01 4.04460192e-01 -4.90418464e-01 3.09666038e-01 7.48538196e-01 -6.03148997e-01 2.50324726e-01 1.88296095e-01 7.80377761e-02 2.70641208e-01 -4.28592980e-01 2.44477540e-01 -1.00653946e+00 -3.99303377e-01 -2.58543283e-01 -3.80734801e-01 -2.11709157e-01 -1.09143861e-01 -1.52967525e+00 -3.47164571e-01 -1.88865352e+00 5.73993027e-01 -4.33497280e-01 -4.38837737e-01 5.17140627e-01 -6.03212118e-02 2.86116004e-01 -1.49762005e-01 1.40196875e-01 2.20903069e-01 5.36699951e-01 1.63902330e+00 -3.03269532e-02 4.57467102e-02 4.34917420e-01 -5.96264780e-01 4.22997892e-01 9.92418468e-01 -1.17866337e+00 -3.29014242e-01 -1.15550924e-02 -1.20116025e-01 6.15039647e-01 7.12305188e-01 -1.25717247e+00 -8.50267261e-02 1.01880722e-01 1.09335816e+00 -4.51514333e-01 3.37887965e-02 -8.78398776e-01 5.47316253e-01 1.16083622e+00 -1.38354212e-01 2.46020593e-02 3.09551120e-01 5.00492990e-01 -6.77343756e-02 -2.99564868e-01 1.07966864e+00 -3.56472820e-01 -3.48203033e-02 3.99446040e-01 -5.42979062e-01 -2.55330652e-01 1.13501835e+00 -2.54090190e-01 -6.72344640e-02 -1.22699887e-01 -1.00267529e+00 -3.49582911e-01 -4.50080484e-02 1.03452936e-01 5.86093843e-01 -1.33745039e+00 -8.93663287e-01 2.21907303e-01 -1.75873473e-01 2.03844011e-01 5.26819468e-01 1.24698484e+00 -8.03852797e-01 6.82671070e-01 -3.05047989e-01 -6.75723076e-01 -1.23840749e+00 3.57485205e-01 8.38995457e-01 -7.32993484e-01 -4.46140468e-01 9.50350523e-01 3.58660161e-01 -2.81800121e-01 -2.97086179e-01 -6.77160800e-01 -1.00940861e-01 -3.14423263e-01 -2.67711468e-02 2.11515710e-01 5.20155370e-01 -4.54486012e-01 -2.46306002e-01 5.36742270e-01 -7.79654831e-02 9.20646191e-02 1.58879840e+00 4.00554776e-01 1.90272972e-01 9.51303467e-02 1.43158424e+00 -4.63783890e-01 -7.23167181e-01 1.80302769e-01 -2.37826020e-01 1.80831347e-02 5.13972402e-01 -1.18010330e+00 -1.29441142e+00 7.72462606e-01 1.20453906e+00 -2.23137513e-01 1.28628707e+00 1.94305852e-01 6.78222179e-01 9.70259830e-02 -1.23591751e-01 -4.72626656e-01 2.48149902e-01 1.89772666e-01 1.00079632e+00 -1.30816221e+00 2.69078966e-02 -2.81729847e-01 -6.60172284e-01 1.15872109e+00 5.16210675e-01 -1.25420228e-01 7.55575657e-01 5.23144066e-01 -5.74115152e-03 -3.30311596e-01 -7.42909849e-01 -3.89910606e-03 3.12761843e-01 5.14015555e-01 9.24706221e-01 2.34639823e-01 -1.91667631e-01 1.06632805e+00 -3.52037400e-01 2.17543423e-01 4.15759265e-01 7.77388275e-01 -3.92902732e-01 -9.92583692e-01 -5.87458134e-01 1.00309837e+00 -7.28542566e-01 -8.41620117e-02 -1.51820853e-01 1.03391314e+00 1.36542544e-01 5.15678883e-01 -1.89357787e-01 -4.13385689e-01 2.35965386e-01 2.02300146e-01 5.82159579e-01 -7.86554039e-01 -9.95299280e-01 -1.19140958e-02 -1.03745736e-01 -1.50270283e-01 -2.78858662e-01 -4.04807538e-01 -1.44609821e+00 -2.46716544e-01 -5.25588036e-01 -9.72887278e-02 8.52104545e-01 8.16043973e-01 -8.34734663e-02 1.37759960e+00 5.85468531e-01 -3.02501440e-01 -4.57366616e-01 -1.27669346e+00 -3.73813868e-01 2.35359982e-01 -6.55970499e-02 -4.45237607e-01 -2.88156420e-01 2.16168761e-02]
[15.241116523742676, -2.243168592453003]
83a54b07-034d-4032-b661-fb0b5d704f27
text-guided-high-definition-consistency
2305.05901
null
https://arxiv.org/abs/2305.05901v1
https://arxiv.org/pdf/2305.05901v1.pdf
Text-guided High-definition Consistency Texture Model
With the advent of depth-to-image diffusion models, text-guided generation, editing, and transfer of realistic textures are no longer difficult. However, due to the limitations of pre-trained diffusion models, they can only create low-resolution, inconsistent textures. To address this issue, we present the High-definition Consistency Texture Model (HCTM), a novel method that can generate high-definition and consistent textures for 3D meshes according to the text prompts. We achieve this by leveraging a pre-trained depth-to-image diffusion model to generate single viewpoint results based on the text prompt and a depth map. We fine-tune the diffusion model with Parameter-Efficient Fine-Tuning to quickly learn the style of the generated result, and apply the multi-diffusion strategy to produce high-resolution and consistent results from different viewpoints. Furthermore, we propose a strategy that prevents the appearance of noise on the textures caused by backpropagation. Our proposed approach has demonstrated promising results in generating high-definition and consistent textures for 3D meshes, as demonstrated through a series of experiments.
['Tiantong He', 'Zhibin Tang']
2023-05-10
null
null
null
null
['text-guided-generation']
['computer-vision']
[ 3.12535375e-01 -9.97610986e-02 4.12047356e-01 -3.29999149e-01 -5.50104678e-01 -4.53097105e-01 8.00876737e-01 -1.43084720e-01 7.09430426e-02 5.25436878e-01 1.99012637e-01 1.99503183e-01 -5.41936532e-02 -1.08633792e+00 -7.35476673e-01 -6.08207166e-01 2.95907676e-01 4.11126941e-01 4.43344772e-01 -2.82839268e-01 4.99831557e-01 6.38147831e-01 -1.70409632e+00 6.28545284e-01 8.18570793e-01 9.22479510e-01 5.40829778e-01 6.70940101e-01 -2.72386521e-01 7.59832084e-01 -5.64527333e-01 -2.36424834e-01 2.43303508e-01 -5.95899940e-01 -5.56140423e-01 4.06674683e-01 6.28909349e-01 -5.94783604e-01 -5.07741235e-02 8.55166972e-01 4.36622858e-01 -2.37993672e-02 7.39772737e-01 -6.33202851e-01 -1.03485286e+00 1.85456812e-01 -7.42766321e-01 -1.46537662e-01 4.84256119e-01 -8.10377486e-03 3.50325465e-01 -8.83430183e-01 1.33079052e+00 1.43280280e+00 4.65276897e-01 7.49812663e-01 -1.29616475e+00 -3.46401721e-01 2.60961741e-01 -2.41235211e-01 -1.25228906e+00 -3.62276584e-01 1.08294058e+00 -5.69502831e-01 7.15193868e-01 1.97696015e-01 8.22474718e-01 1.20914197e+00 5.86079955e-01 4.21531826e-01 1.38034737e+00 -5.97153902e-01 3.32087368e-01 1.04511090e-01 -6.52206779e-01 8.60718906e-01 -2.21706137e-01 9.02205706e-02 -7.07650065e-01 -8.54185503e-03 1.57865286e+00 -2.14212596e-01 -2.74694830e-01 -4.46577102e-01 -1.21206152e+00 5.43889523e-01 3.38107079e-01 2.66025186e-01 -6.39383018e-01 1.54866681e-01 2.05916017e-02 3.78519118e-01 1.05044556e+00 4.09059018e-01 -9.01728570e-02 9.52168778e-02 -9.03508842e-01 2.05296382e-01 5.30564308e-01 9.32161510e-01 7.11762428e-01 1.78015530e-01 -2.94821650e-01 1.01213920e+00 2.66498834e-01 5.37818730e-01 2.99883068e-01 -1.26695442e+00 2.96098650e-01 3.96561712e-01 2.91975170e-01 -1.21687222e+00 4.05514613e-02 -2.85319626e-01 -9.89188790e-01 6.75832450e-01 2.04261109e-01 -6.98176306e-03 -1.11584365e+00 1.52922344e+00 6.63397908e-01 -1.04715884e-01 -8.99626613e-02 9.61030722e-01 4.88618463e-01 7.55250156e-01 -2.24192053e-01 -6.00789338e-02 8.08621645e-01 -9.88596559e-01 -7.36064970e-01 1.04520582e-01 1.92801565e-01 -1.00441468e+00 1.13899577e+00 3.33420366e-01 -1.31276643e+00 -6.02463126e-01 -8.19574058e-01 -1.55670345e-01 -1.06321067e-01 -6.87729195e-02 3.12165976e-01 2.55113095e-01 -1.30409992e+00 7.45331168e-01 -7.09741235e-01 -3.81650895e-01 3.46106201e-01 1.50688076e-02 -3.64602983e-01 -1.57468274e-01 -7.12297618e-01 7.50320733e-01 -4.65130880e-02 1.15435712e-01 -9.55593407e-01 -5.04257023e-01 -5.82260907e-01 -2.25712180e-01 2.15369873e-02 -1.01459360e+00 9.39953744e-01 -1.22741055e+00 -2.15311933e+00 8.17180276e-01 -1.13386519e-01 6.87623844e-02 9.56595659e-01 -8.70676786e-02 1.11873876e-02 2.93915480e-01 1.67803109e-01 1.00298047e+00 1.22173250e+00 -1.73736179e+00 -4.62028503e-01 -1.14027187e-01 -1.03996605e-01 3.12061578e-01 -1.92459345e-01 -2.79517263e-01 -6.16615832e-01 -8.93311322e-01 2.42579237e-01 -6.15263939e-01 -3.12265247e-01 5.12436450e-01 -2.05988333e-01 2.89238811e-01 8.17098320e-01 -6.02092028e-01 8.56590271e-01 -2.24298954e+00 3.06135684e-01 2.44255841e-01 1.74455523e-01 -1.10175259e-01 -2.79465556e-01 4.16907072e-01 4.54397231e-01 -7.36957788e-02 -2.12695464e-01 -5.93886912e-01 -2.16268256e-01 1.85086802e-01 -2.66259640e-01 6.60862178e-02 2.48746902e-01 6.56888783e-01 -9.29858804e-01 -3.94299179e-01 4.72383022e-01 8.34359229e-01 -6.71023548e-01 3.62532735e-01 -5.84119916e-01 8.65163445e-01 -4.70195264e-01 4.11719382e-01 7.91990936e-01 -2.57927030e-01 6.24475181e-02 -1.86709523e-01 -3.82476866e-01 -2.20978379e-01 -1.12135243e+00 1.92521250e+00 -6.83428407e-01 4.68317062e-01 1.21912055e-01 -4.05528843e-01 1.20190203e+00 1.28624782e-01 3.00450474e-01 -1.00367081e+00 3.15516964e-02 3.50879401e-01 -4.94613647e-01 -4.24309671e-01 6.41402483e-01 -1.98347822e-01 3.01170826e-01 5.40160596e-01 -2.96642989e-01 -6.16046727e-01 6.03238717e-02 -1.34628853e-02 6.93159759e-01 4.23483729e-01 -4.03992951e-01 -1.56379938e-01 4.32839781e-01 -2.99200844e-02 1.41137362e-01 7.89778709e-01 3.86426061e-01 1.21078134e+00 2.67091811e-01 -6.03233755e-01 -1.43390369e+00 -9.91008043e-01 6.61925524e-02 4.90389377e-01 3.19808543e-01 -8.98326039e-02 -8.36303771e-01 -4.25156683e-01 -1.62311792e-01 4.64056253e-01 -8.80233467e-01 1.07024848e-01 -4.72143561e-01 -4.42776978e-01 5.85456677e-02 2.94555038e-01 7.08678246e-01 -1.07624495e+00 -4.41438377e-01 5.58137476e-01 -5.67492098e-02 -9.60482657e-01 -5.01235604e-01 -1.86228901e-01 -9.95601177e-01 -6.89149678e-01 -1.21016526e+00 -9.82649803e-01 1.08073926e+00 2.26860642e-01 9.59620535e-01 1.12670690e-01 -6.49814457e-02 3.16785723e-01 -1.80964142e-01 6.61136284e-02 -7.91977465e-01 -1.36221647e-01 -3.41528952e-01 3.39435220e-01 -5.56620955e-01 -5.66842675e-01 -8.13709199e-01 4.59992439e-01 -1.17761910e+00 6.28539622e-01 4.66241986e-01 8.07217181e-01 9.65738237e-01 1.23234950e-01 4.56802338e-01 -8.36805642e-01 8.68239224e-01 -3.27862725e-02 -6.17614329e-01 2.02643096e-01 -4.58881289e-01 2.13998146e-02 5.88225901e-01 -7.46791720e-01 -1.42914426e+00 -9.37770531e-02 -2.99289405e-01 -5.74420691e-01 -1.36535503e-02 2.68458933e-01 2.44338289e-01 -2.49597758e-01 7.27252424e-01 3.24561864e-01 1.34296836e-02 -4.58902061e-01 4.73028094e-01 4.57232803e-01 2.81102777e-01 -7.14788258e-01 5.15544713e-01 8.83472741e-01 -2.36731201e-01 -7.83389091e-01 -6.40330434e-01 2.02599838e-01 -6.66876078e-01 -4.34231669e-01 9.28328931e-01 -7.69458413e-01 -1.25525221e-01 1.00351238e+00 -1.24474764e+00 -9.17843640e-01 -4.05341744e-01 2.30214983e-01 -7.02106476e-01 2.47344881e-01 -8.21172357e-01 -5.12993038e-01 -3.02551895e-01 -1.06835210e+00 1.25077713e+00 1.67590678e-02 -2.16469303e-01 -1.01413631e+00 4.83322926e-02 2.81450190e-02 8.76956522e-01 3.67490500e-01 1.09841549e+00 5.22420287e-01 -9.75099683e-01 -3.38511239e-03 -2.81569839e-01 2.75178939e-01 2.93542236e-01 2.21828148e-01 -7.05548167e-01 -1.42616868e-01 3.33451591e-02 -1.97763339e-01 5.57380855e-01 4.96301681e-01 9.91619051e-01 -7.31000677e-02 -2.02139080e-01 6.29785001e-01 1.46011233e+00 1.00643679e-01 6.33602202e-01 3.83764923e-01 8.07831585e-01 5.97349942e-01 3.95155758e-01 3.80838037e-01 3.68131131e-01 7.50893056e-01 1.51885435e-01 -2.68860966e-01 -7.02091515e-01 -5.78942835e-01 9.39732045e-03 7.91584790e-01 -1.94575563e-01 -3.46539795e-01 -4.46648359e-01 4.40307528e-01 -1.59153378e+00 -8.59185517e-01 -1.12480365e-01 2.05015206e+00 8.38748872e-01 1.98778436e-01 -2.67283946e-01 -2.20654875e-01 6.96454287e-01 1.88671082e-01 -4.54751730e-01 -4.65808690e-01 -1.25603333e-01 5.21269813e-02 8.01310614e-02 8.37318063e-01 -5.98863006e-01 1.20120597e+00 6.47930002e+00 8.47264290e-01 -1.67111540e+00 8.00625011e-02 7.39126086e-01 7.64458925e-02 -7.83010542e-01 -2.77764171e-01 -4.76705164e-01 3.26435566e-01 2.99259335e-01 9.56531018e-02 5.04194915e-01 6.01408124e-01 4.56778079e-01 -1.98660046e-01 -7.93879628e-01 8.99488986e-01 8.18521008e-02 -1.61292553e+00 4.59651053e-01 -9.45327207e-02 1.35677433e+00 -2.83508182e-01 1.92333549e-01 -2.64410555e-01 2.92716980e-01 -6.74004078e-01 1.11424696e+00 7.14644015e-01 9.66376424e-01 -4.73307997e-01 2.38860250e-01 2.84150213e-01 -8.99172723e-01 2.38575608e-01 -4.87685263e-01 2.24784255e-01 3.28694016e-01 8.43227863e-01 -3.92640978e-01 4.37303126e-01 6.69142842e-01 6.97947025e-01 -2.23874077e-01 7.80067027e-01 -2.59385198e-01 -1.70393533e-03 -2.74924040e-01 7.68442228e-02 7.10364506e-02 -1.52170897e-01 2.50001729e-01 9.04949486e-01 7.04480648e-01 -1.12120090e-02 -5.59365936e-02 1.17733037e+00 1.85371377e-02 1.51903272e-01 -6.82114780e-01 1.89165503e-01 2.80226529e-01 9.95139539e-01 -8.34765196e-01 -3.99305761e-01 -9.00730491e-02 1.56083333e+00 4.78548855e-01 4.86029744e-01 -7.17063904e-01 -9.12675634e-02 2.39136472e-01 3.43004405e-01 4.63144124e-01 -4.58777398e-01 -3.93083781e-01 -1.11629987e+00 2.71011144e-02 -7.99434125e-01 -2.86938488e-01 -1.17675745e+00 -1.31503975e+00 1.01720095e+00 -1.68732032e-01 -1.13862240e+00 -1.13812834e-01 -1.76602051e-01 -3.83974671e-01 1.00513482e+00 -1.51576877e+00 -1.19411063e+00 -5.17396390e-01 6.10452831e-01 6.77169740e-01 1.18080363e-01 7.15764284e-01 3.22563440e-01 -1.14530474e-01 1.93828166e-01 1.97209790e-01 -4.43279177e-01 7.16627061e-01 -9.45753753e-01 5.89561999e-01 4.66557533e-01 -1.99286759e-01 3.17734599e-01 5.58549643e-01 -8.84131432e-01 -1.23164046e+00 -9.75337923e-01 7.43870974e-01 -2.59843409e-01 2.69847155e-01 -2.80344039e-01 -9.45773721e-01 2.26876557e-01 2.71345377e-01 -1.68118358e-01 6.29949104e-03 -3.70347112e-01 -5.91874197e-02 -4.53289151e-02 -1.21956873e+00 8.02868903e-01 1.13142252e+00 -5.46111643e-01 -1.62073836e-01 1.75362274e-01 3.57220531e-01 -8.90853345e-01 -7.61437416e-01 2.13817269e-01 6.13677859e-01 -1.23351777e+00 7.45513797e-01 7.69855008e-02 9.40184951e-01 -3.35775524e-01 1.26277143e-02 -1.42406869e+00 -4.82082933e-01 -5.76407850e-01 1.84698761e-01 1.10067892e+00 3.74743849e-01 -4.08479929e-01 7.70664752e-01 4.27826494e-01 -1.19076706e-01 -7.07694054e-01 -5.84698498e-01 -5.43362975e-01 -7.50052631e-02 -5.32810688e-02 6.12679303e-01 1.04293382e+00 -4.92522985e-01 1.33987293e-01 -5.37361562e-01 1.36762530e-01 5.69410801e-01 3.53601098e-01 7.82663584e-01 -8.43203664e-01 -3.67792398e-01 -3.30838561e-01 1.47550972e-03 -1.43023956e+00 -2.48145327e-01 -5.81193209e-01 1.94389224e-01 -1.83944678e+00 -8.58915821e-02 -8.22325945e-01 2.54753768e-01 5.12083359e-02 2.94810813e-02 4.68743205e-01 2.85477079e-02 3.41734201e-01 -1.76079705e-01 6.40499473e-01 2.04156995e+00 -3.19611728e-02 -3.04838836e-01 -3.32681060e-01 -3.10397863e-01 5.62927544e-01 5.80262363e-01 -4.09510344e-01 -6.43744826e-01 -1.09919071e+00 1.13171540e-01 2.68546253e-01 1.81894794e-01 -9.32439983e-01 -1.33382520e-02 -2.83767313e-01 7.09678352e-01 -4.64965045e-01 5.42528987e-01 -6.99707866e-01 4.44160312e-01 2.93191463e-01 -4.24404562e-01 1.36160105e-01 1.34079799e-01 6.11097693e-01 -2.24818632e-01 6.06591664e-02 8.67386997e-01 -3.65633816e-01 -5.10783613e-01 3.04777861e-01 -4.75993484e-01 -1.71825066e-01 9.06284273e-01 -4.52036113e-01 -2.05537871e-01 -4.53716278e-01 -7.58720696e-01 -1.61599413e-01 9.19695556e-01 4.37755972e-01 8.83616865e-01 -1.41153228e+00 -6.48868203e-01 7.01957881e-01 -1.91128701e-01 2.43302211e-01 5.31734347e-01 3.65090042e-01 -9.04496491e-01 -8.09342787e-02 -3.74782979e-01 -7.34648526e-01 -8.86981905e-01 2.77457416e-01 4.57409978e-01 -1.87106386e-01 -1.00176394e+00 8.34065378e-01 3.77194643e-01 -3.38305712e-01 7.34375715e-02 -2.68741786e-01 1.65335059e-01 -4.22646821e-01 4.45768803e-01 -4.03812528e-02 -1.92859862e-02 -2.82767773e-01 1.38625070e-01 1.07626152e+00 -1.79851562e-01 -5.64010680e-01 1.33974457e+00 -3.44830483e-01 -3.28112468e-02 2.40470752e-01 8.43791664e-01 2.03316152e-01 -1.75792718e+00 8.65585357e-03 -5.64614475e-01 -8.32426429e-01 1.19729057e-01 -8.38146687e-01 -1.18878186e+00 7.98610508e-01 5.55571616e-01 1.23970963e-01 9.72187519e-01 -3.20455164e-01 9.06387746e-01 -3.65060233e-02 4.69587088e-01 -1.07408321e+00 5.35679460e-01 5.25076032e-01 1.12219310e+00 -8.59302998e-01 -2.96518236e-01 -4.86053884e-01 -5.01869678e-01 1.20426238e+00 6.96465492e-01 -1.83969244e-01 5.24480760e-01 2.82140166e-01 5.10070741e-01 -3.37718904e-01 -7.82113314e-01 2.61917055e-01 5.55914156e-02 6.67947888e-01 3.08768153e-01 -3.06949407e-01 -1.19425148e-01 -3.55533570e-01 1.17426611e-01 2.42704779e-01 4.98475552e-01 8.63492489e-01 -3.48772019e-01 -1.18890786e+00 -3.94702435e-01 1.91951059e-02 -6.77261278e-02 2.89096981e-02 -3.36972952e-01 4.78502899e-01 -2.65569724e-02 7.38264740e-01 1.70150444e-01 -2.96130061e-01 4.72681642e-01 -2.89019018e-01 9.26270366e-01 -4.67531770e-01 -2.81830817e-01 2.63161987e-01 -4.29018885e-02 -4.10250098e-01 -3.93886983e-01 -2.83316761e-01 -9.01907325e-01 -4.80985880e-01 -1.17542230e-01 -1.38361007e-01 6.84877455e-01 6.89554095e-01 6.97428226e-01 4.84139204e-01 7.38023400e-01 -1.20574892e+00 -2.25734264e-01 -6.92229331e-01 -5.54162145e-01 6.95425153e-01 1.64104149e-01 -5.19674420e-01 -1.77095234e-01 2.02684432e-01]
[9.449670791625977, -3.0935611724853516]
66e10c18-abb6-406b-978c-8cd5186c003f
perpetual-humanoid-control-for-real-time
2305.06456
null
https://arxiv.org/abs/2305.06456v2
https://arxiv.org/pdf/2305.06456v2.pdf
Perpetual Humanoid Control for Real-time Simulated Avatars
We present a physics-based humanoid controller that achieves high-fidelity motion imitation and fault-tolerant behavior in the presence of noisy input (e.g. pose estimates from video or generated from language) and unexpected falls. Our controller scales up to learning ten thousand motion clips without using any external stabilizing forces and learns to naturally recover from fail-state. Given reference motion, our controller can perpetually control simulated avatars without requiring resets. At its core, we propose the progressive multiplicative control policy (PMCP), which dynamically allocates new network capacity to learn harder and harder motion sequences. PMCP allows efficient scaling for learning from large-scale motion databases and adding new tasks, such as fail-state recovery, without catastrophic forgetting. We demonstrate the effectiveness of our controller by using it to imitate noisy poses from video-based pose estimators and language-based motion generators in a live and real-time multi-person avatar use case.
['Weipeng Xu', 'Kris Kitani', 'Alexander Winkler', 'Jinkun Cao', 'Zhengyi Luo']
2023-05-10
null
null
null
null
['humanoid-control']
['robots']
[ 1.33805033e-02 2.20607594e-01 1.06320448e-01 4.16934907e-01 -8.03976178e-01 -4.81518775e-01 3.23756456e-01 -6.57623947e-01 -5.56302667e-01 9.39636230e-01 1.75393611e-01 2.30737895e-01 1.61662802e-01 -4.89627540e-01 -1.19704914e+00 -5.57956636e-01 -5.29578447e-01 8.63621116e-01 4.71530229e-01 -3.08725178e-01 -2.13117003e-01 4.15355265e-01 -1.74465036e+00 -1.62465483e-01 4.22393709e-01 4.47561800e-01 2.82965690e-01 1.25434482e+00 8.12436461e-01 1.19497836e+00 -7.29941189e-01 2.24971265e-01 4.52887356e-01 -3.26166630e-01 -4.86589104e-01 3.18311304e-01 5.34680665e-01 -8.42025757e-01 -8.10524821e-01 3.69961441e-01 9.11961377e-01 5.16127050e-01 2.78033853e-01 -1.62626874e+00 -1.42486766e-01 2.99167454e-01 -7.51533732e-02 -1.66033372e-01 7.92857349e-01 1.31758785e+00 3.36607844e-01 -4.49116588e-01 9.35896158e-01 1.49665308e+00 9.29398417e-01 1.25522935e+00 -1.21039879e+00 -5.45202971e-01 -5.00677563e-02 4.65115607e-02 -1.37180161e+00 -5.27218223e-01 1.98615506e-01 -3.34582061e-01 1.19713151e+00 5.80317117e-02 1.05589461e+00 1.85381520e+00 4.76902038e-01 8.50020051e-01 3.93585682e-01 7.10689798e-02 5.36803186e-01 -7.11833477e-01 -6.64437056e-01 9.21877980e-01 1.42249882e-01 3.43062997e-01 -8.38427484e-01 -5.63567698e-01 1.33064401e+00 -2.05247700e-01 -5.17106414e-01 -3.90741318e-01 -1.54058278e+00 2.41101235e-01 -1.00375619e-02 -5.97410023e-01 -6.48996174e-01 1.17448437e+00 3.54171991e-01 4.50584143e-01 -2.74084300e-01 4.27544892e-01 -5.85479021e-01 -7.66798258e-01 -5.58055460e-01 1.05648410e+00 1.03410447e+00 1.31189704e+00 2.15552643e-01 6.99449122e-01 -1.01779252e-01 2.05430061e-01 -1.02856681e-01 1.02583826e+00 5.90668023e-01 -2.03141379e+00 1.85221836e-01 -2.93679573e-02 4.85099912e-01 -7.72267759e-01 -5.11255562e-01 7.05375969e-02 -5.90742886e-01 5.08245826e-01 2.77364105e-01 -5.05090296e-01 -9.77794945e-01 2.13753200e+00 3.23859274e-01 7.35903203e-01 -1.62847601e-02 1.20426857e+00 1.00425132e-01 4.60577339e-01 -1.59266859e-01 -1.47924811e-01 8.50008845e-01 -7.84360886e-01 -3.40451956e-01 -3.77456129e-01 2.97347099e-01 5.46802999e-03 1.26321149e+00 6.69140637e-01 -1.19039524e+00 -4.70066398e-01 -1.05150056e+00 3.56095344e-01 5.58494806e-01 -3.10624987e-01 5.04669175e-02 1.38429314e-01 -1.12166071e+00 9.45349038e-01 -1.42759669e+00 -4.92648482e-01 -1.46979809e-01 6.07659459e-01 -3.69882405e-01 1.08928286e-01 -1.11404181e+00 8.80286753e-01 1.68739498e-01 -1.48224115e-01 -1.72975695e+00 -7.42976427e-01 -7.31405377e-01 -3.44182879e-01 7.61725187e-01 -1.45377636e+00 1.55811143e+00 -8.56166422e-01 -2.04675412e+00 1.98927417e-01 4.42750424e-01 -5.57755768e-01 9.98451829e-01 -7.63044715e-01 1.58303846e-02 4.41806436e-01 1.26381859e-01 9.66190577e-01 1.34198785e+00 -1.13011682e+00 -2.25415900e-01 4.86397408e-02 -2.15038553e-01 4.88232553e-01 -1.58185288e-01 -5.49088717e-01 -5.68066239e-01 -7.04288006e-01 -3.02409679e-01 -1.58542502e+00 -2.91184008e-01 5.55190980e-01 -1.13258809e-01 2.30591953e-01 9.38739240e-01 -5.06852865e-01 7.25603461e-01 -1.83773065e+00 8.37574959e-01 8.83217826e-02 2.34400466e-01 -7.80077577e-02 -3.50021511e-01 3.97450835e-01 4.71603543e-01 -3.06907386e-01 -8.71055722e-02 -1.60509661e-01 -8.26024935e-02 6.32228136e-01 -1.59145996e-01 4.85657066e-01 1.06931716e-01 7.92882383e-01 -1.09951615e+00 -3.90499175e-01 -5.53067140e-02 4.86317903e-01 -9.58532572e-01 3.05487692e-01 -5.94208181e-01 7.71362484e-01 -3.35652322e-01 5.64808130e-01 -1.22345664e-01 -9.33708400e-02 3.43510389e-01 2.11249292e-01 2.49450177e-01 -2.77624190e-01 -1.32716238e+00 1.93805897e+00 -2.07927510e-01 2.79165089e-01 4.93711114e-01 -2.08142892e-01 4.69009519e-01 4.55019921e-01 5.34573495e-01 -2.26511136e-01 2.82646835e-01 1.74150076e-02 -2.18593076e-01 -6.73879623e-01 6.22097254e-01 1.03107825e-01 -3.91113281e-01 4.18821335e-01 2.15231925e-01 -4.80067104e-01 -1.19114757e-01 1.64315775e-01 1.84206915e+00 7.53482759e-01 -4.14989814e-02 -1.15692290e-02 1.54410467e-01 2.44946361e-01 8.13812017e-01 9.47913229e-01 -3.54243129e-01 7.67784059e-01 1.47634923e-01 -3.61937702e-01 -1.55619669e+00 -1.20858800e+00 9.55552697e-01 1.10923886e+00 6.83442056e-02 -3.14411700e-01 -8.07631135e-01 5.95215298e-02 2.91949630e-01 3.41907769e-01 -2.41454184e-01 -6.33999169e-01 -1.22204340e+00 -1.28482059e-01 8.88611674e-01 7.20267117e-01 3.98776114e-01 -1.12894988e+00 -1.25056982e+00 4.46882516e-01 -2.91842073e-01 -9.81130958e-01 -9.09695506e-01 -3.66353869e-01 -8.44044328e-01 -9.82971489e-01 -6.81465268e-01 -6.52876079e-01 3.08344752e-01 -1.02934204e-02 9.41490054e-01 2.05036610e-01 -5.10099113e-01 9.63219583e-01 2.19051261e-02 2.61099547e-01 -7.62621105e-01 -8.99722353e-02 9.25277948e-01 -4.68436062e-01 -6.41775370e-01 -8.14886034e-01 -4.63053226e-01 2.11849093e-01 -6.96335852e-01 6.34011533e-03 3.28255951e-01 1.13615727e+00 3.50238532e-01 -3.65083516e-01 2.84853697e-01 -1.04361363e-01 8.17533970e-01 -1.99376136e-01 -5.14849067e-01 -5.90880364e-02 -1.07926972e-01 1.48980066e-01 6.79211557e-01 -1.19913113e+00 -7.30529785e-01 4.98848617e-01 2.61233747e-01 -1.06978261e+00 1.23780087e-01 -1.13610879e-01 1.31223336e-01 -2.41818037e-02 1.07513738e+00 3.66973281e-01 5.06725788e-01 1.52668729e-01 6.50457025e-01 2.73515373e-01 1.29116106e+00 -1.05173957e+00 8.12627375e-01 3.56974483e-01 3.87379192e-02 -7.96638787e-01 1.24482796e-01 -7.04932353e-03 -4.10509408e-01 -7.95422435e-01 6.73982501e-01 -1.21528232e+00 -1.61379063e+00 8.01846385e-01 -1.11520767e+00 -1.01997590e+00 -2.73179084e-01 2.87384033e-01 -1.40820205e+00 2.47582376e-01 -1.11600959e+00 -8.48888218e-01 -3.06321084e-01 -8.91205788e-01 1.25027013e+00 -3.95703688e-03 -7.05958903e-01 -3.40656430e-01 2.73016572e-01 4.97821085e-02 3.71899575e-01 5.86586714e-01 3.21426272e-01 2.53748775e-01 -8.05790961e-01 -1.00117639e-01 7.61116385e-01 -2.00966783e-02 -2.67017066e-01 -1.68387234e-01 -4.97326314e-01 -9.30026472e-01 -2.39331469e-01 -8.55120361e-01 3.62765819e-01 6.05056137e-02 5.46010435e-01 -7.45752394e-01 -2.45111480e-01 3.89431924e-01 1.08991230e+00 -1.76557124e-01 4.82830882e-01 3.15033942e-01 7.92140186e-01 2.54286230e-01 4.61690396e-01 8.62520218e-01 2.88704813e-01 6.73029482e-01 4.72539842e-01 6.82484329e-01 -1.46615937e-01 -4.82523561e-01 8.99458170e-01 5.38125992e-01 -3.28752637e-01 -2.11229980e-01 -8.91514957e-01 3.90269279e-01 -2.20164800e+00 -1.05815613e+00 5.54522455e-01 2.15604281e+00 8.58080447e-01 3.23116720e-01 4.86003965e-01 -1.52782485e-01 6.11259162e-01 -2.98828870e-01 -1.19346642e+00 2.45003685e-01 7.27963671e-02 1.77480467e-02 6.18651569e-01 5.18099964e-01 -7.09180474e-01 1.12214923e+00 6.47417259e+00 3.66156816e-01 -1.06247687e+00 -3.57735083e-02 -1.60795655e-02 -7.20712960e-01 1.29627809e-01 -2.13053301e-01 -1.90558374e-01 2.96808302e-01 9.88379538e-01 -1.25567496e-01 8.11297297e-01 9.85119998e-01 5.26945710e-01 -1.05902821e-01 -1.06586981e+00 8.97282541e-01 2.34389260e-01 -1.15332389e+00 -7.93298408e-02 -3.41153890e-01 6.41787350e-01 1.81572847e-02 -5.81700318e-02 3.84475559e-01 8.45330656e-01 -7.84430027e-01 1.09851384e+00 7.51716197e-01 8.67329240e-01 -7.29083717e-01 1.01699367e-01 7.81455636e-01 -9.59449828e-01 -3.77528042e-01 8.11715946e-02 -3.40642571e-01 4.69850302e-01 -3.04902256e-01 -7.57009208e-01 -3.00156651e-03 6.58197343e-01 4.56202805e-01 -1.32386580e-01 6.54628873e-01 -1.27429292e-01 4.04385090e-01 -6.10352218e-01 -1.38318494e-01 -1.17196165e-01 2.93680519e-01 1.09865272e+00 5.83108842e-01 2.20555708e-01 2.16615185e-01 6.82975471e-01 6.33286178e-01 3.08438033e-01 -7.02970386e-01 -5.59055150e-01 2.24481970e-01 6.86516464e-01 8.02506804e-01 -3.39729726e-01 -4.94634449e-01 3.45494658e-01 1.46540260e+00 2.41643995e-01 3.32658023e-01 -1.07015872e+00 1.18983872e-01 1.04082668e+00 2.91159362e-01 1.34278551e-01 -9.52048302e-01 2.81881779e-01 -1.25321555e+00 -3.67312436e-03 -1.15856385e+00 -2.25709081e-02 -1.04854047e+00 -1.06532621e+00 2.84551412e-01 1.74315386e-02 -1.37944555e+00 -9.26754236e-01 -1.13312587e-01 -3.14305305e-01 1.90179750e-01 -3.50672960e-01 -7.12674499e-01 -4.40002769e-01 6.75858378e-01 5.57001650e-01 -7.68109709e-02 7.51507461e-01 1.34505287e-01 -2.96104699e-01 4.10864651e-01 -3.42207730e-01 -1.02933638e-01 8.43587160e-01 -8.60412419e-01 7.68361092e-01 6.27393603e-01 -5.95102370e-01 3.75759572e-01 1.15487468e+00 -1.09212899e+00 -1.96693361e+00 -1.15475380e+00 4.93547879e-02 -6.71713471e-01 7.99995720e-01 -4.36573982e-01 -6.71622396e-01 6.22509778e-01 -2.12213308e-01 -3.86964418e-02 -2.84066975e-01 -7.84165859e-01 -1.58126593e-01 9.06610712e-02 -1.10642028e+00 1.13700426e+00 1.54642963e+00 -3.35216433e-01 -4.33297873e-01 2.99455255e-01 9.64817166e-01 -9.06833351e-01 -6.57084763e-01 2.41785601e-01 7.50694990e-01 -3.71196955e-01 1.01445782e+00 -6.69284582e-01 1.85818598e-01 -4.50089902e-01 1.00155197e-01 -1.22001636e+00 -3.98282617e-01 -1.22033203e+00 -5.11401296e-01 3.63376409e-01 -1.84394475e-02 -1.48088232e-01 9.56464946e-01 5.95306098e-01 2.88860593e-02 -2.90811151e-01 -1.15742528e+00 -1.14961278e+00 -2.26638809e-01 -3.39857131e-01 1.16799720e-01 5.29542506e-01 -3.21362168e-02 1.58223212e-01 -1.21251845e+00 3.14888507e-01 7.41120577e-01 -5.65168381e-01 1.31741679e+00 -5.89788377e-01 -9.28991079e-01 3.78753222e-03 -5.64745963e-01 -1.02768540e+00 2.46082410e-01 -2.60344088e-01 7.26620615e-01 -1.00315142e+00 6.45576511e-03 2.01347053e-01 5.18274784e-01 6.02151155e-01 -1.21304370e-01 1.32239029e-01 5.59921563e-01 5.28337896e-01 -9.29202378e-01 8.18412721e-01 1.19786525e+00 -2.17127334e-02 -4.22366560e-01 -1.35452524e-01 1.63370311e-01 5.98524213e-01 7.02151179e-01 -3.75735253e-01 -5.50463319e-01 -4.09557790e-01 3.22778523e-02 7.79765904e-01 8.15191984e-01 -1.72053218e+00 5.56106031e-01 -2.53435612e-01 2.20137656e-01 -6.50192872e-02 5.54386616e-01 -4.47182924e-01 6.67433977e-01 1.06618071e+00 -2.66116440e-01 5.84643722e-01 1.86206311e-01 1.06639814e+00 5.21679580e-01 6.08743429e-01 4.98819053e-01 -3.69844943e-01 -7.20332384e-01 2.45479539e-01 -9.68288779e-01 2.07964182e-01 1.07363641e+00 -1.67570189e-01 -1.21396258e-01 -8.29029262e-01 -1.07682979e+00 6.64053738e-01 6.98001504e-01 6.13318861e-01 8.35486829e-01 -1.27979565e+00 -7.16281354e-01 -4.72245961e-02 -2.05566272e-01 -6.73339814e-02 3.01378131e-01 3.72353882e-01 -9.53192532e-01 -3.34776193e-01 -5.17099023e-01 -7.76901305e-01 -1.13447475e+00 4.31918800e-01 4.88512546e-01 2.84396082e-01 -1.09565127e+00 4.78918701e-01 -4.37989771e-01 -4.66083318e-01 4.25429791e-01 -1.38449684e-01 4.88064885e-01 -8.41785789e-01 3.41290921e-01 6.10534251e-01 -4.61944044e-01 -3.44272166e-01 -2.61866778e-01 3.68194103e-01 3.54949325e-01 -7.48852789e-01 1.12953293e+00 1.12430891e-02 2.07431212e-01 4.27797824e-01 5.16300201e-01 -5.94866872e-01 -2.16877437e+00 1.68292984e-01 -2.43188992e-01 -1.33121878e-01 -5.89542389e-01 -4.04824913e-01 -7.68694222e-01 1.05499923e-01 5.21166980e-01 -6.66358232e-01 6.97241545e-01 -2.08110392e-01 1.14896464e+00 1.05839419e+00 1.02621400e+00 -1.44614315e+00 7.33433366e-01 7.40292370e-01 1.07194054e+00 -7.04879224e-01 -6.79697841e-02 -5.52609563e-02 -7.65933990e-01 9.67488766e-01 1.06511688e+00 -7.19693840e-01 2.46870428e-01 8.55579078e-01 6.14428297e-02 2.85325497e-01 -1.32347238e+00 1.98495254e-01 -3.89550000e-01 8.98731709e-01 -4.13747162e-01 4.60006818e-02 1.56394884e-01 3.04896146e-01 -3.67758811e-01 2.68955052e-01 8.80637705e-01 1.34679210e+00 -6.22181594e-01 -6.19629681e-01 -5.83460271e-01 2.33258326e-02 2.01325655e-01 4.07577604e-01 -2.51877606e-01 8.42152774e-01 -1.34340420e-01 5.90512633e-01 9.22301859e-02 -7.89470315e-01 4.17792141e-01 -2.09520869e-02 8.31565976e-01 -3.83537680e-01 -6.29821241e-01 1.55389272e-02 1.30904540e-01 -1.14809132e+00 -7.63785169e-02 -8.04022670e-01 -1.60638142e+00 -7.25805581e-01 1.69415593e-01 -4.39897835e-01 -3.70469429e-02 5.99593282e-01 5.30763268e-01 5.19787192e-01 1.34179115e-01 -1.46166098e+00 -9.67625916e-01 -8.31442356e-01 -2.96440069e-02 5.02126276e-01 4.77944821e-01 -6.35977983e-01 -1.55181706e-01 3.69878322e-01]
[5.01881742477417, 0.7476443648338318]
97dffa07-d48c-4e6a-b5a7-aa42fa5dc619
a-mid-level-video-representation-based-on
1605.03804
null
http://arxiv.org/abs/1605.03804v1
http://arxiv.org/pdf/1605.03804v1.pdf
A Mid-level Video Representation based on Binary Descriptors: A Case Study for Pornography Detection
With the growing amount of inappropriate content on the Internet, such as pornography, arises the need to detect and filter such material. The reason for this is given by the fact that such content is often prohibited in certain environments (e.g., schools and workplaces) or for certain publics (e.g., children). In recent years, many works have been mainly focused on detecting pornographic images and videos based on visual content, particularly on the detection of skin color. Although these approaches provide good results, they generally have the disadvantage of a high false positive rate since not all images with large areas of skin exposure are necessarily pornographic images, such as people wearing swimsuits or images related to sports. Local feature based approaches with Bag-of-Words models (BoW) have been successfully applied to visual recognition tasks in the context of pornography detection. Even though existing methods provide promising results, they use local feature descriptors that require a high computational processing time yielding high-dimensional vectors. In this work, we propose an approach for pornography detection based on local binary feature extraction and BossaNova image representation, a BoW model extension that preserves more richly the visual information. Moreover, we propose two approaches for video description based on the combination of mid-level representations namely BossaNova Video Descriptor (BNVD) and BoW Video Descriptor (BoW-VD). The proposed techniques are promising, achieving an accuracy of 92.40%, thus reducing the classification error by 16% over the current state-of-the-art local features approach on the Pornography dataset.
['Arnaldo de A. Araújo', 'Silvio Jamil F. Guimarães', 'Sandra Avila', 'Carlos Caetano', 'William Robson Schwartz']
2016-05-12
null
null
null
null
['video-description', 'pornography-detection']
['computer-vision', 'computer-vision']
[ 1.98425457e-01 -3.46940517e-01 -1.93089887e-01 1.16136946e-01 -5.80321431e-01 -3.62128109e-01 5.52523196e-01 6.82405710e-01 -2.16427132e-01 3.56269300e-01 1.13184281e-01 2.53519714e-01 -3.08054239e-01 -1.02533913e+00 -3.69493306e-01 -9.08556283e-01 1.91125423e-01 -7.82717690e-02 4.49421495e-01 -6.59779087e-02 5.37937164e-01 8.10148299e-01 -2.04770279e+00 2.86695331e-01 5.18230915e-01 1.30399466e+00 -7.13324547e-02 6.71583474e-01 -1.13237768e-01 5.34410954e-01 -5.56411028e-01 -6.54811621e-01 1.64260283e-01 -4.10080314e-01 -6.60725534e-02 2.86337167e-01 7.57595301e-01 -1.92712918e-01 -5.77631891e-01 1.18430054e+00 4.40944612e-01 1.47795290e-01 6.90653145e-01 -9.55809355e-01 -3.47923428e-01 -6.56517893e-02 -6.47035062e-01 2.78404713e-01 6.64164603e-01 -2.30188370e-01 6.60993278e-01 -6.85720325e-01 6.36781335e-01 9.77498293e-01 5.74952602e-01 1.37411788e-01 -1.11446011e+00 -5.04389048e-01 -3.80526185e-01 6.99452102e-01 -1.46146500e+00 -2.15975806e-01 9.67449486e-01 -6.59566343e-01 4.60618973e-01 6.42548919e-01 7.81491280e-01 1.08423114e+00 3.30884993e-01 5.47741354e-01 1.27720964e+00 -4.64760005e-01 6.88479319e-02 4.03579235e-01 4.15779203e-02 5.91054320e-01 5.94523549e-01 -1.48465246e-01 -6.75852954e-01 -1.39833376e-01 5.79649925e-01 2.36718521e-01 -1.56631067e-01 -6.55616403e-01 -5.22875488e-01 7.89765418e-01 7.82459825e-02 7.70680189e-01 -4.85090196e-01 -4.98629548e-02 8.24882865e-01 6.75936416e-02 4.59033161e-01 1.23819813e-01 2.22008824e-01 -4.12768751e-01 -1.10719597e+00 3.66616458e-01 6.64557397e-01 4.67818707e-01 5.62126577e-01 -1.84698984e-01 5.04627824e-03 1.04445052e+00 8.02816765e-04 3.04829538e-01 3.56821239e-01 -3.58460367e-01 3.61856818e-01 6.50068104e-01 -1.32799134e-01 -2.05922747e+00 -1.74287215e-01 -1.21341288e-01 -5.97082078e-01 1.66977480e-01 3.95756125e-01 4.44535881e-01 -5.67591906e-01 1.06864965e+00 3.40729564e-01 -8.88627768e-03 -1.51216641e-01 1.00414038e+00 8.75749350e-01 7.18852818e-01 7.67389014e-02 -2.44981423e-01 1.64070892e+00 -8.60880017e-01 -8.19341540e-01 8.87947455e-02 1.78778097e-01 -9.80521977e-01 5.62623262e-01 8.17192435e-01 -8.00655544e-01 -5.62796712e-01 -9.78851199e-01 2.20047697e-01 -6.47578955e-01 1.06793195e-01 3.97156596e-01 1.18901277e+00 -5.74001014e-01 7.42639303e-01 -4.95452791e-01 -9.22549903e-01 8.27867016e-02 3.99309918e-02 -6.26960039e-01 -2.17932954e-01 -7.76252747e-01 6.77318454e-01 2.81492382e-01 -6.46440871e-03 -6.58208549e-01 -3.37980449e-01 -8.70335817e-01 3.33195746e-01 4.46562976e-01 7.32403994e-02 4.11537021e-01 -9.62721109e-01 -1.27841353e+00 9.57658529e-01 1.29285112e-01 -2.48505428e-01 5.24421632e-01 -6.15301430e-02 -7.09420085e-01 8.04884017e-01 -3.09575081e-01 1.56186819e-01 1.10320318e+00 -1.14114940e+00 -3.14882129e-01 -4.49586451e-01 1.92090473e-03 -5.84248565e-02 -7.25750625e-01 3.80630970e-01 -5.25130391e-01 -6.88160241e-01 1.81050554e-01 -6.72113180e-01 2.10599616e-01 3.28139327e-02 -6.45584688e-02 -1.89174682e-01 8.01832795e-01 -1.09114850e+00 1.32540011e+00 -2.33019423e+00 -2.04954855e-02 4.12669480e-01 1.02481887e-01 6.51573062e-01 -4.19288455e-03 6.81299746e-01 2.91888416e-02 8.89366725e-04 1.82727352e-01 -1.74116530e-02 -1.47085324e-01 4.08511758e-02 1.22357018e-01 8.17915738e-01 -5.50227389e-02 1.62103623e-01 -7.08716989e-01 -7.45937645e-01 6.85001373e-01 6.98042035e-01 -3.10559660e-01 5.28983250e-02 4.41508263e-01 5.23166545e-02 -3.56492579e-01 8.20988536e-01 8.01074862e-01 3.28873426e-01 1.52520761e-01 -3.33588481e-01 -2.60078579e-01 -2.28615090e-01 -1.35439062e+00 1.29742432e+00 -3.79675567e-01 7.15664446e-01 1.86923862e-01 -1.15870643e+00 1.23350453e+00 4.78427231e-01 7.53742397e-01 -7.28612781e-01 3.41303587e-01 2.78509706e-01 -4.04077441e-01 -8.25447440e-01 6.12052262e-01 -1.20449558e-01 2.05259830e-01 -2.51677364e-01 2.81594899e-02 2.20695794e-01 4.61266786e-01 -7.77212009e-02 8.87857735e-01 1.24555729e-01 5.11617780e-01 -6.28399253e-02 1.00929117e+00 -2.64573634e-01 3.34458023e-01 4.49926794e-01 -2.55461872e-01 7.88435936e-01 5.98779619e-01 -3.24349433e-01 -1.04270768e+00 -6.39865756e-01 -2.87851661e-01 5.73913038e-01 2.73406565e-01 -7.19221413e-01 -7.37424552e-01 -5.40629685e-01 5.77364899e-02 1.03284687e-01 -4.85876232e-01 -1.45623446e-01 -4.58527923e-01 -2.13474140e-01 3.72699678e-01 2.33247012e-01 4.54805136e-01 -9.55750227e-01 -6.46724343e-01 1.14483505e-01 8.87804255e-02 -1.06865823e+00 -1.32505298e-01 -3.35341394e-01 -7.48134136e-01 -1.27887559e+00 -9.83895481e-01 -5.42015254e-01 5.41051865e-01 3.17768008e-01 5.78878939e-01 3.96538153e-03 -8.20611835e-01 7.98444867e-01 -7.30080068e-01 1.05608128e-01 -3.42200786e-01 -3.89113367e-01 -1.21880069e-01 6.25068545e-01 6.43301964e-01 -3.06805700e-01 -6.28460050e-01 3.80925298e-01 -1.04777157e+00 -4.29714113e-01 5.48489988e-01 6.88116014e-01 3.05117577e-01 1.97651953e-01 1.61880448e-01 -6.04016840e-01 4.52590853e-01 -4.35479701e-01 -5.36180556e-01 6.42314404e-02 -4.55316216e-01 -5.55257201e-01 6.23515964e-01 -3.34696472e-01 -9.12800193e-01 -2.66240716e-01 -1.06796779e-01 -6.45080864e-01 -5.41247964e-01 2.56933361e-01 1.12534292e-01 -5.98672330e-01 5.14095426e-01 2.99387485e-01 -1.57233328e-02 -7.91454494e-01 -1.15761012e-01 9.11651790e-01 2.21956387e-01 -3.06867957e-01 6.96050048e-01 3.65781188e-01 2.58189738e-01 -1.32216668e+00 -1.09225392e-01 -1.13603532e+00 -3.99734139e-01 -5.95064104e-01 1.01552141e+00 -5.77357769e-01 -1.01822591e+00 4.57005560e-01 -8.08484256e-01 5.94650984e-01 8.95934477e-02 5.32273829e-01 -4.43088055e-01 1.12302792e+00 -4.97037053e-01 -1.12774587e+00 -8.09262320e-02 -9.49546993e-01 7.34657764e-01 3.79561156e-01 2.71281768e-02 -8.13513875e-01 -8.94741621e-03 7.52969861e-01 4.07318056e-01 5.11846423e-01 7.59674788e-01 -4.14655626e-01 -3.00829649e-01 -7.05146670e-01 -3.79134536e-01 6.98573589e-01 -7.18971863e-02 -1.57844666e-02 -8.06950450e-01 -2.63343990e-01 -8.56860578e-02 -2.33040646e-01 6.52342439e-01 1.75153032e-01 1.24938571e+00 -1.22824177e-01 -1.80430591e-01 3.47781986e-01 1.79187453e+00 3.18793893e-01 8.00006866e-01 4.00828004e-01 3.99680853e-01 8.23382139e-01 7.96907365e-01 6.17779076e-01 -1.19927779e-01 9.64291453e-01 4.77325708e-01 1.49807781e-02 -4.28289652e-01 -3.08379620e-01 3.49167883e-01 7.52287209e-01 -3.60415995e-01 -5.49993441e-02 -6.00250244e-01 6.42852247e-01 -1.65847540e+00 -1.08870506e+00 -1.91014901e-01 2.50976014e+00 2.45605350e-01 -1.65786237e-01 2.71071076e-01 5.30714571e-01 8.99121344e-01 3.71046543e-01 2.07415506e-01 -8.67380917e-01 1.19041549e-02 1.92061469e-01 6.38208270e-01 6.33224547e-02 -1.26893711e+00 4.58849758e-01 5.11319208e+00 1.18697035e+00 -1.18256807e+00 1.92539752e-01 2.49689147e-01 2.20154271e-01 8.81433189e-02 -1.82666123e-01 -5.57878017e-01 5.72630763e-01 6.27897203e-01 2.96079755e-01 2.61955738e-01 9.26151454e-01 7.52986372e-02 -4.41338599e-01 -5.19535422e-01 1.61811793e+00 8.72497261e-01 -8.88951957e-01 -1.23856544e-01 1.75371524e-02 4.44797337e-01 -7.30807543e-01 -1.13603033e-01 5.40220328e-02 -1.01945686e+00 -7.70200074e-01 5.33827186e-01 4.98855084e-01 5.28689206e-01 -8.46349418e-01 8.43531847e-01 -2.40328796e-02 -1.10116804e+00 -1.64862126e-01 -5.61513484e-01 2.80924682e-02 -1.00670092e-01 5.49066663e-01 -3.66118729e-01 5.28209984e-01 7.30103910e-01 5.73116481e-01 -5.86089969e-01 1.69539368e+00 2.65315622e-02 4.04238641e-01 -1.37650281e-01 -2.96976238e-01 1.83602855e-01 -4.49809104e-01 8.44102442e-01 1.33043683e+00 5.15554488e-01 -1.11323401e-01 -4.10236567e-02 6.19666398e-01 3.29660177e-01 8.33635032e-01 -7.72850454e-01 -3.30700129e-01 -1.79328427e-01 1.57532442e+00 -8.43798220e-01 -1.63532287e-01 -6.82486534e-01 9.78787124e-01 -3.09870452e-01 -6.62567541e-02 -6.32831275e-01 -6.36378050e-01 4.65276301e-01 4.67829436e-01 4.18622464e-01 -1.75109044e-01 1.18163839e-01 -1.05065024e+00 9.97111872e-02 -8.79373014e-01 3.53599727e-01 -5.56719780e-01 -1.10562241e+00 3.03226233e-01 7.30285142e-03 -1.33529055e+00 1.71696752e-01 -8.61612558e-01 -2.70370543e-01 3.33395123e-01 -1.17432261e+00 -1.16204643e+00 -5.45696914e-01 4.84726608e-01 5.32618880e-01 -1.20915018e-01 6.06755912e-01 8.05660725e-01 -2.93261647e-01 4.30541813e-01 1.27794296e-01 1.48302391e-01 7.89858758e-01 -8.76346588e-01 -5.50590038e-01 7.52264738e-01 2.03181263e-02 4.31284010e-01 8.42027783e-01 -5.28905511e-01 -1.44834030e+00 -4.08681631e-01 9.11594391e-01 3.28512825e-02 5.52431583e-01 -3.62013698e-01 -5.62084973e-01 -9.15182754e-03 9.04352143e-02 3.97160165e-02 6.22487187e-01 -6.96852058e-03 -3.26754868e-01 -5.13289690e-01 -1.31428766e+00 2.90004998e-01 5.67458570e-01 -4.24350828e-01 -4.29679602e-01 2.80981630e-01 -2.64106899e-01 -1.06588900e-01 -9.81139004e-01 2.19945371e-01 6.47897720e-01 -1.27251410e+00 9.34711337e-01 -8.48724321e-02 3.82613182e-01 -2.33623371e-01 -1.04492620e-01 -8.22943509e-01 -9.15722176e-02 -2.17113003e-01 1.24225177e-01 1.33753777e+00 -2.85165727e-01 -4.03468251e-01 8.61321032e-01 3.04174572e-01 1.30134538e-01 -4.50305939e-01 -1.15195131e+00 -8.48576069e-01 -6.01729035e-01 -2.04955176e-01 5.71990125e-02 7.40511000e-01 1.47879109e-01 -2.13075891e-01 -6.28329277e-01 -1.65373981e-01 5.96538842e-01 -1.04474807e-02 7.07451403e-01 -1.05160892e+00 -3.48898798e-01 -4.94012266e-01 -1.42358172e+00 -4.63789821e-01 -2.30736852e-01 -5.32480359e-01 -2.71671981e-01 -1.29599655e+00 4.77791250e-01 -2.78110713e-01 -2.98939735e-01 1.98326603e-01 2.69388139e-01 8.68487835e-01 4.15797502e-01 -1.08659022e-01 -4.70789552e-01 1.71925500e-01 9.43839312e-01 -1.00996576e-01 8.14472791e-03 -7.80597925e-02 -1.39012843e-01 6.73856497e-01 5.20680189e-01 -3.54811937e-01 5.96869290e-02 2.77095169e-01 3.45964849e-01 7.16316402e-02 4.51276690e-01 -1.30931830e+00 3.35374810e-02 3.58736254e-02 2.82080948e-01 -4.70176786e-01 6.21950746e-01 -1.04773009e+00 1.07149377e-01 5.50035238e-01 6.89463988e-02 -1.80929244e-01 5.88139985e-03 6.02748036e-01 -6.71564639e-01 -5.97010195e-01 8.99118364e-01 -1.02605738e-01 -8.75368655e-01 -1.62122265e-01 -5.79621553e-01 -6.57246172e-01 1.40247738e+00 -6.30201101e-01 -2.67961383e-01 -3.86037588e-01 -6.87370598e-01 -5.00589252e-01 4.76324350e-01 5.72770476e-01 6.89793885e-01 -1.19407690e+00 -4.63413328e-01 2.69057006e-01 3.64942402e-01 -9.01035428e-01 6.93904042e-01 1.14502692e+00 -1.07635999e+00 5.68836510e-01 -5.92044830e-01 -3.43257844e-01 -1.84492171e+00 7.71969795e-01 -2.02617496e-02 1.02765625e-02 -7.32045591e-01 3.87355000e-01 1.96631774e-02 2.74582237e-01 1.95582092e-01 -5.98164871e-02 -7.27808714e-01 5.23800194e-01 4.89519060e-01 9.17399347e-01 2.32971191e-01 -9.06254828e-01 -4.30494398e-01 9.56817269e-01 2.88658530e-01 3.53404850e-01 1.10160255e+00 7.52528086e-02 -2.84108311e-01 1.35367215e-01 1.36022460e+00 5.42085886e-01 -6.69078827e-01 1.07269034e-01 -1.28633589e-01 -1.17174840e+00 5.66631481e-02 -5.19579768e-01 -1.07786953e+00 1.04319894e+00 8.94922554e-01 4.16562617e-01 1.12947440e+00 -9.61906612e-02 6.74845457e-01 -8.15433711e-02 4.36944455e-01 -1.23853195e+00 1.16347209e-01 -1.58862561e-01 8.18974733e-01 -9.47355032e-01 2.13330090e-01 -6.90406621e-01 -4.47155684e-01 1.52571785e+00 2.77989775e-01 -4.53626394e-01 3.66777360e-01 -1.40063629e-01 -1.12772010e-01 -1.25704437e-01 -4.07649018e-02 -3.21402818e-01 4.60868031e-01 4.96069103e-01 2.72209257e-01 1.42858654e-01 -1.16760612e+00 4.28405166e-01 3.54633689e-01 -1.46185040e-01 4.33916032e-01 6.85543358e-01 -4.31367546e-01 -1.11062407e+00 -8.91551316e-01 2.87066609e-01 -7.71375775e-01 2.81780243e-01 -3.03950012e-01 8.36501956e-01 4.38126713e-01 1.01077199e+00 -9.19937640e-02 -2.98438698e-01 1.87808350e-01 1.58961285e-02 6.21440411e-01 -3.08344960e-01 -6.95710421e-01 2.57110536e-01 2.81159669e-01 -5.66603661e-01 -4.94112253e-01 -7.88140535e-01 -5.73303938e-01 -3.56575429e-01 -2.92716891e-01 9.85704362e-02 1.06819379e+00 5.96822441e-01 -1.82739552e-02 1.94761634e-01 5.11209369e-01 -7.26443887e-01 -1.94656685e-01 -8.30817461e-01 -1.14060140e+00 8.68986905e-01 -3.85546237e-02 -8.58086288e-01 -3.38158876e-01 -2.06865892e-01]
[12.0521821975708, 0.49912768602371216]
d60bc799-5b05-43ab-923e-970d793b4741
cross-lingual-word-embeddings-beyond-zero
2011.01682
null
https://arxiv.org/abs/2011.01682v1
https://arxiv.org/pdf/2011.01682v1.pdf
Cross-lingual Word Embeddings beyond Zero-shot Machine Translation
We explore the transferability of a multilingual neural machine translation model to unseen languages when the transfer is grounded solely on the cross-lingual word embeddings. Our experimental results show that the translation knowledge can transfer weakly to other languages and that the degree of transferability depends on the languages' relatedness. We also discuss the limiting aspects of the multilingual architectures that cause weak translation transfer and suggest how to mitigate the limitations.
['Ali Basirat', 'Shifei Chen']
2020-11-03
null
null
null
null
['zero-shot-machine-translation']
['natural-language-processing']
[-3.96326900e-01 -3.57394628e-02 -7.46473372e-01 -3.33064735e-01 -8.30701411e-01 -9.67298210e-01 7.59818673e-01 -2.44307920e-01 -5.13429344e-01 1.05660105e+00 5.44593394e-01 -8.72811258e-01 3.68943393e-01 -7.27744520e-01 -1.03046012e+00 -2.42128730e-01 1.65810347e-01 5.73273242e-01 -1.18019015e-01 -6.12896144e-01 -2.85291642e-01 2.95924395e-01 -6.95687354e-01 2.80926973e-01 9.84166145e-01 6.29805997e-02 9.84475017e-02 2.90582657e-01 -3.73559952e-01 1.70103297e-01 -2.95389086e-01 -6.68635368e-01 4.53731447e-01 -5.12121201e-01 -1.00609314e+00 -4.27689850e-01 4.25982594e-01 -2.78744280e-01 -4.78785366e-01 1.15686440e+00 3.31536114e-01 -2.83763885e-01 6.60817027e-01 -9.11084294e-01 -1.57651186e+00 6.44237638e-01 -2.23041177e-01 5.54804981e-01 1.95936784e-01 -3.87724042e-02 1.20931578e+00 -1.04714489e+00 1.04622662e+00 1.38074422e+00 7.11794257e-01 5.53864658e-01 -1.09911740e+00 -5.80344558e-01 2.22596407e-01 6.17632717e-02 -1.31201589e+00 -4.22437131e-01 2.84753919e-01 -4.09387112e-01 1.27838552e+00 -2.73712993e-01 3.63215655e-01 1.34231102e+00 6.51910365e-01 4.23500627e-01 1.28345263e+00 -7.18588412e-01 -3.51847589e-01 3.98028851e-01 7.40910769e-02 6.94524825e-01 5.91212869e-01 4.44483638e-01 -5.58896482e-01 -4.56818985e-03 9.27213609e-01 -7.17648804e-01 -2.45479360e-01 -9.13675427e-02 -1.38529289e+00 9.79411125e-01 5.37639260e-01 6.98535144e-01 -2.49016508e-02 4.70494200e-03 4.35536146e-01 7.99078643e-01 7.05966890e-01 7.03091085e-01 -9.45578098e-01 2.10201487e-01 -4.82115805e-01 -3.94537449e-01 7.58702099e-01 1.14032733e+00 1.15418828e+00 1.15199134e-01 1.74289778e-01 8.61079574e-01 1.38315074e-02 7.85346627e-01 6.36159480e-01 -2.85234064e-01 6.93022370e-01 1.41102016e-01 1.84332170e-02 -4.27879155e-01 -3.64668109e-02 -5.21108508e-01 -1.01882271e-01 -2.60301769e-01 2.52803057e-01 -5.84816039e-01 -6.72538579e-01 2.15108895e+00 -5.19188046e-02 -3.96049321e-01 2.98177004e-01 8.35660160e-01 4.89849299e-01 6.55632138e-01 2.74056107e-01 -6.00197166e-02 9.58276451e-01 -1.05817378e+00 -7.99758911e-01 -5.66219151e-01 1.01841819e+00 -1.05154586e+00 1.14991748e+00 -4.80850875e-01 -8.62149596e-01 -4.45145398e-01 -1.01850522e+00 -2.72513747e-01 -6.05204284e-01 -7.42589906e-02 7.67214417e-01 5.45736849e-01 -1.36205590e+00 4.08507228e-01 -8.09600055e-01 -8.97859633e-01 -4.63084020e-02 4.34277028e-01 -6.09961152e-01 -1.91740170e-01 -1.73184097e+00 1.79554629e+00 4.95146811e-01 -2.76743788e-02 -6.62998140e-01 -3.17242444e-01 -8.39722753e-01 -2.49251783e-01 -4.55889285e-01 -8.18663597e-01 9.66860414e-01 -1.43973625e+00 -1.28888333e+00 9.75031495e-01 -3.40935022e-01 -3.39702959e-03 2.89842904e-01 -2.04384863e-01 -7.10271955e-01 -9.78410318e-02 3.91621262e-01 8.32248449e-01 1.97713524e-01 -1.00417411e+00 -6.12235606e-01 -1.65408164e-01 -6.17840253e-02 5.76645136e-01 -6.81970835e-01 3.13307106e-01 -3.30533415e-01 -5.71525753e-01 -2.54455984e-01 -1.22274494e+00 1.00742936e-01 -1.43689781e-01 4.41632196e-02 -2.47806281e-01 4.60189104e-01 -8.10623169e-01 6.72939658e-01 -1.96616483e+00 2.90567756e-01 -1.85713843e-01 -4.83273357e-01 1.01902574e-01 -7.40494668e-01 7.68219352e-01 -1.87251009e-02 2.77568102e-01 1.00485392e-01 1.73009470e-01 3.53677869e-02 6.22212231e-01 -4.96654063e-01 5.59133291e-01 4.42358583e-01 1.45522499e+00 -9.13951874e-01 -2.31298268e-01 -2.73576856e-01 5.79543114e-01 -3.88701230e-01 1.91552758e-01 4.84632440e-02 3.57984275e-01 -4.35754806e-01 4.48023856e-01 4.67550576e-01 6.77425563e-02 4.64970797e-01 -3.95199247e-02 -2.64889181e-01 9.79664803e-01 -1.64041072e-01 1.83782232e+00 -5.89537799e-01 8.46825898e-01 3.13953049e-02 -6.55158043e-01 7.65821159e-01 3.98725808e-01 -6.17344379e-02 -5.83281457e-01 2.15926208e-02 7.39172041e-01 4.42633599e-01 -2.90102869e-01 3.34829301e-01 -7.24798679e-01 -4.06940952e-02 8.52264047e-01 5.69404542e-01 6.17258484e-03 -8.62600058e-02 -1.28388599e-01 5.00414550e-01 4.73820031e-01 2.87526727e-01 -8.43394816e-01 6.32849559e-02 2.84197301e-01 5.69462836e-01 4.83683646e-01 -1.89606801e-01 -4.57595661e-03 1.35001421e-01 -4.71028537e-01 -1.09104455e+00 -1.23630774e+00 -4.40189540e-01 1.54170871e+00 1.18941583e-01 -7.58574307e-02 -5.34504533e-01 -8.41740549e-01 5.42964926e-03 7.75630891e-01 -5.36847830e-01 -3.46478522e-01 -7.20755935e-01 -8.88066292e-01 7.80619264e-01 6.38536692e-01 -6.61956593e-02 -8.96785676e-01 3.96391720e-01 6.69285432e-02 -2.73881972e-01 -1.29905617e+00 -5.94850659e-01 2.88647801e-01 -1.08403003e+00 -5.26469827e-01 -5.59768081e-01 -1.40648353e+00 7.91269720e-01 1.91449106e-01 1.13892055e+00 -3.69701535e-02 3.88637185e-01 3.01927567e-01 -1.98387191e-01 -2.53576934e-01 -6.46322966e-01 6.31529927e-01 5.23354709e-01 -5.58319449e-01 9.69825387e-01 -5.10892034e-01 -6.37988076e-02 3.34007680e-01 -6.08571053e-01 -2.40508363e-01 6.73195243e-01 8.28536630e-01 2.25019500e-01 -5.53569257e-01 7.61772335e-01 -6.39589846e-01 9.39996421e-01 -5.64700663e-01 -3.41368526e-01 4.70245808e-01 -6.40193582e-01 3.16151708e-01 3.49405795e-01 -5.70625663e-01 -9.49139059e-01 -3.58675122e-01 -2.09745858e-02 7.09832907e-02 7.77885392e-02 5.87734163e-01 -3.22819129e-02 -3.18833947e-01 6.84817433e-01 1.66007608e-01 -3.47214580e-01 -4.12130058e-01 7.81459212e-01 5.92487276e-01 1.03739403e-01 -1.09055686e+00 9.46033835e-01 1.11153357e-01 -5.04672766e-01 -6.87822640e-01 -6.35010183e-01 8.99616033e-02 -1.09196186e+00 2.74154544e-01 1.01227736e+00 -1.28390515e+00 3.13573122e-01 -2.24035397e-01 -1.56016874e+00 -3.29624325e-01 -5.80824912e-02 1.04831147e+00 -2.85424739e-01 1.48307076e-02 -1.16663694e+00 -2.47404072e-02 -2.53505796e-01 -1.18191206e+00 6.54949546e-01 -2.64235795e-01 -4.02589798e-01 -1.62308204e+00 4.90458518e-01 -1.32620446e-02 5.25884390e-01 -4.98304486e-01 1.39039707e+00 -7.93090999e-01 -5.25137961e-01 1.27765834e-01 -2.01615751e-01 4.01707679e-01 4.72015083e-01 -2.66109049e-01 -7.04916775e-01 -5.27911842e-01 -1.78939953e-01 -4.16215360e-01 7.78016031e-01 8.53705630e-02 -2.38243774e-01 -2.71208972e-01 -4.38559026e-01 7.49907374e-01 1.41315579e+00 -2.46245235e-01 2.06669405e-01 3.92345399e-01 7.61011720e-01 6.47271037e-01 1.66725740e-02 -7.32315004e-01 4.24688786e-01 3.81151915e-01 -9.74635407e-02 -3.22878182e-01 -2.61572421e-01 -5.59311450e-01 9.48646486e-01 1.99727190e+00 -7.53042176e-02 -4.82719438e-03 -8.85148764e-01 9.63452160e-01 -1.57339001e+00 -4.27602291e-01 1.13200443e-02 2.13889980e+00 1.18132997e+00 -4.65680212e-02 -2.96169698e-01 -8.94769788e-01 8.32033575e-01 7.45718703e-02 -2.05688789e-01 -1.09629560e+00 -5.13779581e-01 3.79446417e-01 6.83647573e-01 1.10787559e+00 -6.02569818e-01 1.71401465e+00 8.51862335e+00 3.72114807e-01 -1.23857296e+00 5.82456827e-01 1.13582581e-01 2.38558829e-01 -7.71910548e-01 2.71297574e-01 -7.46378899e-01 -8.32273364e-02 9.79439914e-01 -4.32413459e-01 4.64657754e-01 4.28258628e-01 -2.69852519e-01 6.02279902e-01 -1.48585498e+00 1.34060487e-01 1.74348965e-01 -9.26591754e-01 3.82682621e-01 2.00593233e-01 1.12916863e+00 8.67863417e-01 1.40600964e-01 4.88330036e-01 6.36980534e-01 -9.29364204e-01 2.51828432e-01 -1.66822389e-01 1.07706332e+00 -7.51091182e-01 6.00038409e-01 2.65800923e-01 -7.87821949e-01 3.67652684e-01 -6.86960816e-01 -3.02487075e-01 2.01376453e-01 1.40707061e-01 -9.77031648e-01 3.75944793e-01 2.07010373e-01 6.08802021e-01 -4.37317014e-01 2.72691369e-01 -7.51391292e-01 5.91281831e-01 -1.38674825e-01 9.80184227e-02 4.23856080e-01 -5.27836084e-01 4.27751571e-01 1.48689306e+00 5.61439157e-01 -3.74386311e-01 4.39848974e-02 8.17047060e-01 -4.22469378e-01 4.49127734e-01 -1.00194013e+00 -3.08461577e-01 2.39663810e-01 7.00020075e-01 -2.64755785e-01 -2.70943493e-01 -8.83722723e-01 1.26568782e+00 7.72687852e-01 7.92765260e-01 -4.44862336e-01 -7.74502829e-02 1.00231612e+00 -9.67927873e-02 2.75078982e-01 -5.20236373e-01 -2.63800383e-01 -1.66683078e+00 4.03414704e-02 -7.85000920e-01 2.82505644e-03 -6.33512557e-01 -1.64473534e+00 9.29766834e-01 -2.24549890e-01 -8.79149258e-01 -2.72387326e-01 -8.23447287e-01 -3.90400946e-01 1.37610698e+00 -1.67848384e+00 -1.47768009e+00 7.68334508e-01 5.73380768e-01 3.08285296e-01 -3.74931246e-01 1.27316308e+00 3.97819966e-01 -2.80418396e-01 9.05146062e-01 8.58070776e-02 4.08891469e-01 1.16188121e+00 -9.14082170e-01 7.44121611e-01 8.37475359e-01 4.05873179e-01 1.10279703e+00 6.33227527e-01 -7.10891664e-01 -1.26622558e+00 -1.10954833e+00 1.75140083e+00 -8.37257206e-01 1.07940483e+00 -5.26601434e-01 -9.77127075e-01 1.47980142e+00 9.50634778e-01 -1.77992415e-02 9.02168095e-01 7.00097740e-01 -9.13832009e-01 1.15823239e-01 -6.62670732e-01 7.02750444e-01 1.03684735e+00 -1.26904917e+00 -1.17120731e+00 4.78060246e-01 1.10555494e+00 1.06889680e-01 -8.99502575e-01 2.76051134e-01 5.33434570e-01 -2.84271866e-01 6.28963709e-01 -1.30742872e+00 3.78373653e-01 -7.74351582e-02 -3.57406318e-01 -1.83537543e+00 -5.65753996e-01 -3.41327429e-01 5.27135134e-01 9.38822985e-01 9.73831892e-01 -9.80386794e-01 4.99965213e-02 1.40080571e-01 1.54333776e-02 -2.00857073e-01 -1.11264467e+00 -1.04187226e+00 1.24141920e+00 -5.42933680e-02 4.76490915e-01 1.76858377e+00 3.30992907e-01 9.83378589e-01 -3.08061510e-01 2.87984967e-01 2.45309159e-01 1.57396093e-01 2.96415269e-01 -9.37802494e-01 -2.27911592e-01 -2.06525087e-01 -2.40236565e-01 -9.96598005e-01 7.89134979e-01 -1.50105858e+00 -8.72691497e-02 -1.41065335e+00 2.10455477e-01 -2.35258430e-01 -6.35923982e-01 4.32405353e-01 -3.61307114e-01 3.03149968e-01 1.64939374e-01 4.51602399e-01 -3.14259022e-01 4.20923889e-01 1.26314282e+00 8.48933961e-03 -8.50319788e-02 -5.30883431e-01 -8.34687173e-01 5.38322031e-01 8.27192605e-01 -6.93230808e-01 -2.77272165e-01 -1.48623443e+00 3.35783571e-01 -2.48845994e-01 -4.58333760e-01 -2.81533033e-01 5.93295088e-03 -3.61083835e-01 2.12014049e-01 1.04223587e-01 6.12975620e-02 -5.70207000e-01 -2.41714671e-01 3.80023062e-01 -5.22191584e-01 5.46618819e-01 5.57187021e-01 3.12312752e-01 -2.80179799e-01 7.63773248e-02 5.83906353e-01 -1.32973850e-01 -3.91066700e-01 1.32595956e-01 -5.31332493e-01 1.74384758e-01 4.39034492e-01 3.39725882e-01 -4.38846588e-01 -9.78253856e-02 -6.38030112e-01 -1.57381687e-02 7.15750396e-01 9.80983138e-01 -9.93567854e-02 -1.90319288e+00 -1.02207553e+00 1.65368259e-01 3.13948125e-01 -9.69395459e-01 -6.56694651e-01 5.99969983e-01 -3.70651901e-01 7.69795239e-01 -4.98954505e-01 -2.90281951e-01 -5.81328750e-01 5.92503786e-01 4.41374540e-01 -8.02407339e-02 -1.18322805e-01 7.28514135e-01 4.84340698e-01 -1.03027356e+00 -3.11040491e-01 -1.32702500e-01 2.21925244e-01 -2.29138955e-01 7.59709403e-02 -1.03426382e-01 9.07334015e-02 -1.08160245e+00 -5.22755265e-01 7.58700550e-01 -2.06025437e-01 -4.79017287e-01 1.13482440e+00 -1.77119151e-01 -5.11216879e-01 6.69765651e-01 1.36600327e+00 2.85686553e-01 -6.61881089e-01 -5.13808787e-01 3.79951708e-02 -1.68001235e-01 -1.44102812e-01 -8.09408128e-01 -7.44204521e-01 1.03356016e+00 3.57332706e-01 -4.59708631e-01 5.70446908e-01 1.40747786e-01 7.92569280e-01 4.80286926e-01 6.54119253e-01 -1.17978001e+00 -6.97985470e-01 9.79606450e-01 7.65648425e-01 -1.25373340e+00 -6.91299513e-02 -2.66931474e-01 -4.39704508e-01 1.06961584e+00 5.68991840e-01 -2.49794230e-01 5.41819751e-01 2.60330439e-01 7.04702318e-01 8.97291675e-02 -8.42878819e-01 -8.63242000e-02 2.91436881e-01 8.14932883e-01 9.12112474e-01 3.72884542e-01 -8.51718128e-01 2.40758762e-01 -3.74421000e-01 -2.27806360e-01 1.67544365e-01 5.64843714e-01 -2.50428736e-01 -1.62628710e+00 -2.08434090e-01 -8.93457830e-02 -5.71126878e-01 -5.80217957e-01 -8.47590268e-01 8.54039907e-01 2.80149162e-01 7.37069428e-01 -7.69757526e-03 -3.29437912e-01 1.64272994e-01 4.82508510e-01 7.30627120e-01 -7.89095342e-01 -6.37808084e-01 6.79545477e-02 2.34408155e-01 -2.18681782e-01 -2.48173311e-01 -3.22437316e-01 -9.02862549e-01 -2.81715661e-01 -3.88265163e-01 3.15387011e-01 5.56046247e-01 1.05232656e+00 3.50869179e-01 1.02168530e-01 3.57506245e-01 -3.36208522e-01 -5.42005956e-01 -9.63108778e-01 -1.11791722e-01 3.59697938e-01 3.87889594e-01 -2.31151089e-01 -4.31370944e-01 -1.05741248e-02]
[11.280980110168457, 10.135885238647461]
5fe15454-c9a6-40a1-ac00-e723b10cdc2d
assessing-gender-bias-in-predictive
2203.10264
null
https://arxiv.org/abs/2203.10264v1
https://arxiv.org/pdf/2203.10264v1.pdf
Assessing Gender Bias in Predictive Algorithms using eXplainable AI
Predictive algorithms have a powerful potential to offer benefits in areas as varied as medicine or education. However, these algorithms and the data they use are built by humans, consequently, they can inherit the bias and prejudices present in humans. The outcomes can systematically repeat errors that create unfair results, which can even lead to situations of discrimination (e.g. gender, social or racial). In order to illustrate how important is to count with a diverse training dataset to avoid bias, we manipulate a well-known facial expression recognition dataset to explore gender bias and discuss its implications.
['Silvia Ramis', 'Cristina Manresa-Yee']
2022-03-19
null
null
null
null
['facial-expression-recognition']
['computer-vision']
[ 4.09074932e-01 4.42705989e-01 -3.80765796e-01 -8.48651648e-01 7.63069279e-03 -3.12429100e-01 5.12718558e-01 2.59999465e-02 -5.90435922e-01 9.96431410e-01 -9.19719338e-02 -3.02566975e-01 7.31421774e-03 -7.79405236e-01 -5.76531112e-01 -6.31290257e-01 2.47876987e-01 6.55126721e-02 -5.77079654e-01 -3.90563086e-02 6.64402902e-01 5.32522142e-01 -1.79578614e+00 -9.96929407e-03 1.09503257e+00 7.66129315e-01 -6.62674665e-01 5.96865341e-02 -9.15478263e-03 7.98517764e-01 -8.50028098e-01 -9.80808437e-01 2.45061621e-01 -4.17395443e-01 -4.44253445e-01 -3.88008766e-02 6.06787384e-01 -3.05988908e-01 8.98151323e-02 1.32623935e+00 5.14184237e-01 2.55029172e-01 7.09011912e-01 -1.50976777e+00 -7.19677925e-01 2.50708997e-01 -6.70171201e-01 -1.05294280e-01 3.62616003e-01 3.92863840e-01 4.83634621e-01 -6.15100384e-01 7.95054793e-01 1.57304025e+00 5.85885584e-01 8.38110685e-01 -1.41507328e+00 -1.27579391e+00 5.69968335e-02 -1.25230446e-01 -1.31642544e+00 -8.10593665e-01 7.20140517e-01 -5.09536207e-01 1.79243386e-01 5.24887204e-01 8.18961799e-01 1.43631077e+00 3.72665018e-01 1.74661577e-01 1.62016940e+00 -3.27046722e-01 2.76967853e-01 2.47452036e-01 -2.05997704e-03 6.17341876e-01 7.16831684e-01 4.85344231e-01 -6.12307489e-01 -2.87498534e-01 5.00933349e-01 -7.44116008e-02 7.33617172e-02 -2.20615059e-01 -8.28063965e-01 9.65435624e-01 5.03160000e-01 4.81368862e-02 -1.85170323e-01 1.16610207e-01 2.39965141e-01 2.86510855e-01 1.58491820e-01 9.92423117e-01 -5.94847798e-02 -1.95737511e-01 -7.97978163e-01 2.99335092e-01 4.76685584e-01 4.30642784e-01 6.94653988e-01 1.75328493e-01 -2.44096220e-02 8.36112976e-01 1.29984841e-01 5.59625685e-01 5.92098951e-01 -1.17536771e+00 -4.43389453e-02 5.20686388e-01 8.32478926e-02 -1.60051489e+00 -3.67211908e-01 -5.23523875e-02 -6.56639338e-01 5.84882855e-01 5.35571456e-01 -1.54389784e-01 -9.67536688e-01 2.01880574e+00 3.85884494e-01 -3.14902991e-01 -3.41255337e-01 1.02259076e+00 7.39917278e-01 3.93652841e-02 7.00193763e-01 -2.04474077e-01 1.14933729e+00 -2.12363660e-01 -8.24653864e-01 -4.76778090e-01 6.59637213e-01 -5.83248615e-01 9.91523147e-01 3.49322677e-01 -9.34930682e-01 -1.29543960e-01 -8.58763874e-01 -2.97794100e-02 -4.87940043e-01 -2.79443502e-01 9.27704275e-01 1.12991726e+00 -4.80621994e-01 8.59589040e-01 -3.80973309e-01 -2.61306792e-01 9.17502761e-01 2.28243262e-01 -4.07297283e-01 5.88958859e-02 -1.11498666e+00 1.06981051e+00 1.22981302e-01 4.17052470e-02 -4.53714073e-01 -8.21857393e-01 -7.14854360e-01 -2.04991609e-01 2.34265491e-01 -4.44419414e-01 8.57853770e-01 -1.65742195e+00 -1.21730125e+00 1.54053283e+00 -1.21825881e-01 4.41958793e-02 7.89372563e-01 -7.96788037e-02 -4.47841883e-01 -2.91716605e-01 1.52543634e-01 7.06979275e-01 8.45748127e-01 -1.14010310e+00 -1.88716099e-01 -8.48311841e-01 -2.94185907e-01 1.94938504e-03 -9.73551646e-02 3.52664232e-01 6.52162671e-01 -7.37671435e-01 1.50518818e-02 -1.19018507e+00 -3.10642511e-01 1.88737333e-01 -4.20759290e-01 1.91571936e-01 5.03947735e-01 -5.35930037e-01 1.16032672e+00 -2.30895734e+00 -2.71942645e-01 4.71009403e-01 2.84153193e-01 2.89879054e-01 2.18120962e-01 -1.35963246e-01 -1.68784887e-01 8.24041486e-01 1.09704875e-03 1.67717129e-01 -2.71471739e-02 2.83551425e-01 -9.32383314e-02 5.49917936e-01 1.98996469e-01 8.25963438e-01 -7.38830566e-01 -6.54442072e-01 -9.93712395e-02 3.12813818e-01 -6.58484876e-01 3.91465127e-02 2.58066565e-01 4.93472457e-01 -3.79804581e-01 7.50641763e-01 8.45274091e-01 9.12231021e-03 2.83924818e-01 1.65305823e-01 -4.71318699e-02 8.60043243e-02 -5.18911004e-01 8.50526989e-01 -9.90613699e-02 6.31448388e-01 -8.83632991e-03 -9.28457737e-01 9.39693868e-01 -7.09937662e-02 1.03754990e-01 -6.79559469e-01 4.62253720e-01 2.24909604e-01 4.90062833e-01 -3.99119556e-01 4.61750627e-01 -7.27865458e-01 1.46730188e-02 4.03964013e-01 -3.16497028e-01 -3.22242111e-01 -2.69222617e-01 -1.16896652e-01 5.86745977e-01 -1.95181787e-01 3.74147624e-01 -4.06998336e-01 1.32617414e-01 -8.04497898e-02 1.01666439e+00 7.22331583e-01 -6.57773912e-01 3.37425023e-01 6.90441966e-01 -4.86664921e-01 -8.16828370e-01 -1.05596316e+00 -3.23546082e-01 1.24062324e+00 -1.12818621e-01 6.04289398e-02 -3.60739589e-01 -6.72854066e-01 4.54783946e-01 9.80003595e-01 -8.41398358e-01 -5.65035522e-01 1.17690647e-02 -7.50749111e-01 7.02067077e-01 2.60510653e-01 3.56416553e-01 -8.15871775e-01 -8.22410464e-01 -3.25353891e-01 1.49974748e-01 -7.95148551e-01 -4.93761934e-02 -1.54843569e-01 -6.83393955e-01 -1.08045375e+00 -4.64003682e-01 -1.63773268e-01 1.01248014e+00 -8.06200132e-02 1.18871343e+00 3.51910174e-01 -4.16008353e-01 2.60844678e-01 -7.70634636e-02 -1.04897845e+00 -5.47734320e-01 -3.68325055e-01 9.31757018e-02 -6.68693036e-02 8.08852732e-01 -6.04458332e-01 -5.28096557e-01 4.07519788e-01 -6.61337495e-01 -5.60123548e-02 3.52681756e-01 8.25463772e-01 2.24848222e-02 -3.56544793e-01 8.04582775e-01 -1.26453340e+00 6.45887852e-01 -4.39765334e-01 -4.71669793e-01 2.90049046e-01 -1.05743778e+00 -4.73848015e-01 2.26595506e-01 -6.21533692e-01 -1.07548821e+00 -3.25686306e-01 3.40956986e-01 -1.79551944e-01 -2.72184432e-01 4.34343427e-01 -5.90239055e-02 -3.42812359e-01 8.89536083e-01 -3.75896722e-01 3.26434255e-01 -6.68414682e-02 8.85846764e-02 7.97196090e-01 -2.98220403e-02 -7.74134696e-01 5.71132958e-01 3.37032586e-01 1.98792696e-01 -8.44807148e-01 -8.29405904e-01 4.90884542e-01 -4.19319868e-01 -6.54312611e-01 8.28942299e-01 -6.90294921e-01 -7.72753298e-01 3.73528957e-01 -6.83221400e-01 -6.27357066e-02 -2.18814850e-01 5.62525451e-01 -2.59935886e-01 -1.78037569e-01 -1.89368621e-01 -1.03839290e+00 1.94169834e-01 -1.00868905e+00 5.01453578e-01 5.07073104e-01 -9.01367843e-01 -6.12900555e-01 -3.31627548e-01 4.43079799e-01 5.87800443e-01 6.00047231e-01 8.80070806e-01 -6.99970305e-01 -1.75466031e-01 -1.29060056e-02 -2.19997078e-01 1.48249447e-01 2.67025352e-01 5.02473772e-01 -1.18820667e+00 2.41144262e-02 -2.23433048e-01 -7.13795543e-01 5.81441522e-01 1.08194888e-01 1.39983332e+00 -4.66925710e-01 -1.99205294e-01 4.26859140e-01 9.25755322e-01 5.28896749e-01 6.81557894e-01 1.98376328e-01 3.90807420e-01 9.95146692e-01 5.05358398e-01 3.86133432e-01 7.03571141e-02 3.52499694e-01 -1.24076791e-02 -1.19488731e-01 3.19355428e-01 -5.10216892e-01 5.53969368e-02 1.35760933e-01 -2.10416153e-01 3.26916188e-01 -9.69640553e-01 2.10890338e-01 -1.40346742e+00 -1.09584332e+00 1.47290826e-01 2.36968255e+00 1.04191482e+00 -2.08880767e-01 1.39190868e-01 9.80641767e-02 9.25303757e-01 6.31058887e-02 -6.07814312e-01 -1.14169955e+00 -1.38631269e-01 1.14116997e-01 3.31334233e-01 1.17593698e-01 -7.06319153e-01 8.98194253e-01 7.96508598e+00 5.44541061e-01 -1.45700121e+00 -2.86283910e-01 1.31867433e+00 -4.19630051e-01 -5.39542317e-01 -1.44565076e-01 -1.46262094e-01 5.21862984e-01 7.02706695e-01 -6.57344282e-01 3.03534448e-01 8.38520765e-01 2.38152936e-01 -2.38882422e-01 -1.03825188e+00 9.40910995e-01 -1.49998039e-01 -8.06232870e-01 -4.19038795e-02 6.38035834e-02 5.75179815e-01 -6.26769841e-01 4.59037244e-01 2.82773048e-01 4.84699517e-01 -1.71944106e+00 5.78487992e-01 2.36180127e-01 8.55933607e-01 -7.70698130e-01 4.33056980e-01 -1.79298371e-02 1.34204760e-01 -5.50353676e-02 -4.01261181e-01 -6.17116332e-01 -3.51454586e-01 8.31344843e-01 -4.98287886e-01 -5.36452755e-02 6.16884410e-01 3.67763877e-01 -6.03436351e-01 6.62037551e-01 -3.66142720e-01 5.64649105e-01 -1.18137293e-01 -2.41265416e-01 -2.15422362e-01 -2.19920337e-01 3.84103179e-01 7.51878560e-01 2.93345571e-01 1.60808101e-01 -3.03063899e-01 1.04461515e+00 -8.64709020e-02 3.21022332e-01 -9.23994541e-01 -5.19748926e-01 5.82670450e-01 1.03988254e+00 -5.20936787e-01 -2.30834827e-01 -2.46948764e-01 5.98994493e-01 2.62942642e-01 4.22515929e-01 -6.78377330e-01 5.78419156e-02 1.13622892e+00 2.02645566e-02 -4.66581404e-01 2.38924474e-01 -6.07301235e-01 -1.08925438e+00 -3.28331381e-01 -1.21254480e+00 3.08788270e-01 -5.30537844e-01 -1.43702245e+00 -1.63649797e-01 -2.03002542e-01 -8.04301143e-01 -2.78500989e-02 -4.60710227e-01 -4.04500306e-01 6.24093056e-01 -1.09055972e+00 -9.56770599e-01 -1.96860462e-01 2.45589107e-01 -3.06783646e-01 -1.88061744e-01 7.41826475e-01 2.43461445e-01 -7.12717354e-01 8.81042719e-01 -2.88561285e-01 1.84873402e-01 1.08618951e+00 -7.81282127e-01 -1.73068419e-01 4.10406590e-01 -2.34344259e-01 8.82578850e-01 7.45813012e-01 -6.21729851e-01 -1.00886345e+00 -6.94682240e-01 7.28578389e-01 -3.64769906e-01 3.32616568e-01 -1.59474358e-01 -6.87964618e-01 8.59367669e-01 -3.43553841e-01 -2.88831033e-02 1.35518956e+00 6.25479043e-01 -6.87514782e-01 -1.08284866e-02 -1.73472750e+00 9.26677823e-01 1.26209342e+00 -4.27173108e-01 -4.38461900e-01 -8.89494941e-02 2.60557950e-01 -3.96091908e-01 -7.09853590e-01 4.36556339e-01 1.16483867e+00 -1.42983770e+00 7.22298324e-01 -1.20998132e+00 8.98525059e-01 2.04279393e-01 -4.72618192e-02 -1.36253953e+00 -3.67375582e-01 -4.62831110e-01 4.74243879e-01 1.38253903e+00 5.08340001e-01 -1.09913087e+00 8.16749096e-01 1.68990266e+00 5.65226257e-01 -5.27615309e-01 -9.70007181e-01 -7.31214106e-01 4.94285852e-01 -2.61896670e-01 9.38033938e-01 1.53207159e+00 1.85042754e-01 -3.97813227e-03 -4.92328048e-01 -2.81126022e-01 6.94989502e-01 8.67607817e-02 9.23130751e-01 -1.43477952e+00 4.62596297e-01 -7.34060109e-01 -6.90021694e-01 -7.35538900e-02 4.67056334e-01 -6.92571998e-01 -2.54611045e-01 -4.83752042e-01 2.16261312e-01 -7.26703644e-01 -2.66025692e-01 3.98656547e-01 -4.58638459e-01 2.77643621e-01 3.91957402e-01 -9.79526341e-02 -2.92236805e-01 4.24226731e-01 1.29563177e+00 1.13895416e-01 -3.96572091e-02 -2.35419095e-01 -1.36730492e+00 9.77084935e-01 7.84552097e-01 -5.12003183e-01 -2.06001610e-01 -1.32006146e-02 5.13820887e-01 -1.33570492e-01 4.08287555e-01 -5.44067919e-01 -2.67254621e-01 -8.19486320e-01 8.03410828e-01 4.29121435e-01 2.30157077e-01 -8.42379212e-01 2.49824479e-01 5.61827779e-01 -6.29544616e-01 -2.62824804e-01 1.86546937e-01 1.84748545e-01 -1.27471149e-01 -2.94949740e-01 9.52398777e-01 -1.37144729e-01 -3.65131527e-01 -4.60777953e-02 -4.65384275e-01 3.22954059e-01 1.28652215e+00 -1.36343136e-01 -5.73327422e-01 -6.12176239e-01 -5.01513958e-01 2.32029930e-01 9.67331469e-01 3.68441641e-01 2.93434113e-01 -1.39709270e+00 -5.40508866e-01 2.17148900e-01 2.97982872e-01 -5.11660933e-01 1.62006378e-01 6.02362335e-01 -1.60794556e-01 -1.70530770e-02 -5.54943025e-01 -2.51228869e-01 -1.31981468e+00 4.57786947e-01 3.16527694e-01 4.19875622e-01 -7.86198452e-02 8.28297496e-01 2.17518106e-01 -5.03217757e-01 -1.85677081e-01 2.27313310e-01 -4.39960063e-02 2.34590948e-01 5.12873650e-01 5.60369849e-01 -3.55152011e-01 -6.50804579e-01 -5.86245894e-01 2.74339676e-01 1.81115285e-01 -1.84077043e-02 9.98904228e-01 2.44370494e-02 -3.51182073e-01 4.72759753e-01 7.45585024e-01 3.36507887e-01 -7.47188330e-01 3.58662635e-01 -6.75565898e-02 -1.33228731e+00 -3.61857004e-02 -9.96922016e-01 -9.88249898e-01 7.61061370e-01 5.61213911e-01 1.52075142e-01 9.34375823e-01 -4.54514056e-01 1.50208667e-01 1.33564159e-01 4.53792304e-01 -1.36943078e+00 -2.12821767e-01 -9.10493806e-02 7.09539771e-01 -1.65280199e+00 3.03604454e-01 -4.76254344e-01 -7.09597170e-01 7.51744270e-01 7.00503826e-01 1.52739212e-01 4.72740412e-01 2.59115994e-02 3.69684696e-01 -1.21133737e-01 -6.52990162e-01 6.21006563e-02 1.62401468e-01 9.10280883e-01 8.94600630e-01 3.75503957e-01 -9.39825535e-01 4.98547077e-01 -5.24026036e-01 3.03180248e-01 5.47521055e-01 8.31649303e-01 -1.05806865e-01 -1.03073466e+00 -6.66252017e-01 8.26657653e-01 -8.19398761e-01 2.40547910e-01 -8.70154321e-01 7.34306097e-01 1.90737605e-01 7.97730148e-01 1.19949996e-01 -3.82353455e-01 -4.58405018e-02 3.89397502e-01 5.22212267e-01 -2.93615878e-01 -5.57143390e-01 -5.36076009e-01 3.65092576e-01 -7.56586730e-01 -4.39509064e-01 -9.04300630e-01 -9.80685174e-01 -7.56460667e-01 -5.25965020e-02 -7.96210915e-02 5.81093729e-01 6.96128964e-01 4.53875989e-01 3.33290584e-02 4.64342207e-01 -4.16637391e-01 -3.88277799e-01 -5.98925710e-01 -7.46692836e-01 1.02415991e+00 7.70552754e-02 -8.40723991e-01 -6.52252078e-01 -3.82737309e-01]
[12.978278160095215, 1.406218409538269]
308a1b21-3a8f-41e6-be22-66ca949ae5aa
lpyolo-low-precision-yolo-for-face-detection
2207.10482
null
https://arxiv.org/abs/2207.10482v1
https://arxiv.org/pdf/2207.10482v1.pdf
LPYOLO: Low Precision YOLO for Face Detection on FPGA
In recent years, number of edge computing devices and artificial intelligence applications on them have advanced excessively. In edge computing, decision making processes and computations are moved from servers to edge devices. Hence, cheap and low power devices are required. FPGAs are very low power, inclined to do parallel operations and deeply suitable devices for running Convolutional Neural Networks (CNN) which are the fundamental unit of an artificial intelligence application. Face detection on surveillance systems is the most expected application on the security market. In this work, TinyYolov3 architecture is redesigned and deployed for face detection. It is a CNN based object detection method and developed for embedded systems. PYNQ-Z2 is selected as a target board which has low-end Xilinx Zynq 7020 System-on-Chip (SoC) on it. Redesigned TinyYolov3 model is defined in numerous bit width precisions with Brevitas library which brings fundamental CNN layers and activations in integer quantized form. Then, the model is trained in a quantized structure with WiderFace dataset. In order to decrease latency and power consumption, onchip memory of the FPGA is configured as a storage of whole network parameters and the last activation function is modified as rescaled HardTanh instead of Sigmoid. Also, high degree of parallelism is applied to logical resources of the FPGA. The model is converted to an HLS based application with using FINN framework and FINN-HLS library which includes the layer definitions in C++. Later, the model is synthesized and deployed. CPU of the SoC is employed with multithreading mechanism and responsible for preprocessing, postprocessing and TCP/IP streaming operations. Consequently, 2.4 Watt total board power consumption, 18 Frames-Per-Second (FPS) throughput and 0.757 mAP accuracy rate on Easy category of the WiderFace are achieved with 4 bits precision model.
['Hasan Şakir Bilge', 'Sefa Burak Okcu', 'Bestami Günay']
2022-07-21
null
null
null
null
['face-detection']
['computer-vision']
[-2.38191802e-03 -6.84903935e-02 -1.58993617e-01 -2.92819530e-01 5.20228922e-01 -2.79159158e-01 1.43268526e-01 -2.63961852e-01 -6.45884573e-01 4.07904953e-01 -6.94476128e-01 -6.29959464e-01 2.15994433e-01 -1.01618648e+00 -5.02464771e-01 -4.17250454e-01 1.38223782e-01 -2.36271560e-01 5.59072316e-01 1.88410487e-02 -1.94043010e-01 7.38212705e-01 -1.61819959e+00 3.17003667e-01 8.46826658e-02 1.68062854e+00 -1.30378893e-02 8.82392764e-01 2.13112652e-01 3.71582925e-01 -5.53428054e-01 -4.84373957e-01 5.78175843e-01 9.01379660e-02 2.55015437e-02 -3.65163088e-01 2.45141134e-01 -5.36865413e-01 -4.43299055e-01 1.14212728e+00 7.05515325e-01 -5.87378979e-01 4.54517037e-01 -1.47518659e+00 1.78177692e-02 2.79229015e-01 -4.86058503e-01 4.89594787e-01 -2.90378690e-01 3.45032394e-01 1.19014516e-01 -6.68129206e-01 2.98152208e-01 1.14389050e+00 7.76632607e-01 4.75394905e-01 -5.25789976e-01 -9.73148048e-01 -4.42327172e-01 3.07140648e-01 -1.32829583e+00 -4.39202458e-01 5.24835944e-01 -1.16868518e-01 1.38153183e+00 2.26376474e-01 8.20816934e-01 7.09480822e-01 7.46216714e-01 7.31427819e-02 7.13380337e-01 -1.83900073e-01 2.86363482e-01 3.68845016e-01 2.94667125e-01 8.65314484e-01 7.87007868e-01 7.00233504e-02 -5.38830519e-01 1.20924458e-01 1.07913721e+00 2.55603373e-01 -1.15216151e-01 4.95631307e-01 -6.78029299e-01 4.95595485e-01 5.52849054e-01 -5.46050631e-03 -3.82215619e-01 4.88498151e-01 8.82037938e-01 1.82104200e-01 -1.51609510e-01 -4.30954099e-01 -7.84986615e-01 -1.47255406e-01 -6.01731360e-01 -8.62250570e-03 6.83807611e-01 1.33056700e+00 4.31803972e-01 4.81045425e-01 2.27418557e-01 2.07120940e-01 5.38909972e-01 7.00615108e-01 6.20476127e-01 -4.12458926e-01 1.82710364e-01 9.52022374e-01 -3.67757082e-01 -8.63769472e-01 -6.49039686e-01 -4.82267678e-01 -1.03294897e+00 6.80264115e-01 -9.94906276e-02 -5.32922566e-01 -9.54152644e-01 1.07600081e+00 4.48477447e-01 2.32443243e-01 2.87779629e-01 8.77470255e-01 1.05041277e+00 8.58617008e-01 1.42174855e-01 2.88226940e-02 2.16546965e+00 -6.18803978e-01 -3.81112576e-01 -9.78228375e-02 3.42236698e-01 -8.89744401e-01 6.82296634e-01 4.81389165e-01 -8.14253509e-01 -9.88637686e-01 -1.69918799e+00 -1.67576164e-01 -6.13855302e-01 6.37121499e-01 6.60661519e-01 1.00425911e+00 -1.02146828e+00 3.98729205e-01 -8.40986490e-01 -3.93923372e-01 4.97555763e-01 8.26641500e-01 -2.32328385e-01 4.75967884e-01 -1.03773332e+00 3.43305826e-01 5.92251062e-01 4.44779694e-01 -5.94081938e-01 -5.61435103e-01 -6.19481921e-01 2.56608278e-01 -4.27913927e-02 -7.39920199e-01 1.02435148e+00 -1.31062782e+00 -1.78058147e+00 5.68078399e-01 4.40961957e-01 -6.24469399e-01 2.69068867e-01 3.09025973e-01 -8.16089034e-01 7.73341432e-02 -4.62058514e-01 7.17967451e-01 1.10200191e+00 4.84588407e-02 -9.37548459e-01 -3.73356551e-01 -1.07876584e-01 2.69912165e-02 -5.13432920e-01 9.32856873e-02 -1.99717268e-01 -2.45796219e-01 -1.37473375e-01 -6.51579261e-01 1.00291573e-01 5.00737965e-01 -4.69050482e-02 -1.17803127e-01 1.49512339e+00 -1.76484242e-01 1.06614411e+00 -2.21777558e+00 -7.86529243e-01 2.26235360e-01 5.14589846e-02 8.56143177e-01 5.16454995e-01 -2.36506745e-01 1.06087834e-01 -2.77035952e-01 3.87740821e-01 2.75628686e-01 -2.14484185e-01 -1.27551615e-01 1.68137792e-02 4.09787595e-01 1.86903551e-01 5.54632246e-01 -1.30599782e-01 -3.69718403e-01 2.09799737e-01 8.27033162e-01 -5.16127467e-01 -2.01876059e-01 1.77948013e-01 -1.94839835e-01 -4.57229406e-01 9.84232366e-01 9.73633587e-01 1.03771210e-01 -2.80752424e-02 -7.31677592e-01 -4.18404013e-01 -1.89047769e-01 -1.29049456e+00 1.11475599e+00 -4.66920376e-01 8.46166730e-01 4.29070860e-01 -6.56581700e-01 1.06235230e+00 4.80721772e-01 -1.22232109e-01 -5.06700635e-01 8.86667728e-01 1.03042826e-01 4.54743803e-02 -7.04224288e-01 2.30398506e-01 1.42407179e-01 2.87936002e-01 -2.18110919e-01 5.33643402e-02 2.93642759e-01 -2.15622485e-01 -1.44882262e-01 1.03180468e+00 1.53354360e-02 2.04232007e-01 -4.52203631e-01 6.74741745e-01 2.27443904e-01 6.48953736e-01 9.79699865e-02 -4.00146633e-01 -2.31242970e-01 4.92669374e-01 -7.13965952e-01 -1.21129072e+00 -9.64127839e-01 -5.40013075e-01 5.52582979e-01 -3.82590294e-02 -3.10087711e-01 -9.43061173e-01 -1.72755525e-01 -2.09703356e-01 -1.06349245e-01 -1.12333760e-01 -9.60922241e-02 -4.10641283e-01 -6.40823305e-01 8.66141260e-01 6.58739388e-01 1.14717853e+00 -1.15607071e+00 -1.64342844e+00 2.98883885e-01 1.19019032e+00 -1.30059385e+00 -6.93986192e-03 1.85382590e-01 -1.12661433e+00 -1.09524918e+00 -1.83311075e-01 -1.04728043e+00 5.74913681e-01 -3.39719728e-02 6.54168129e-01 2.88080927e-02 -9.23346758e-01 -1.17181921e-02 -1.54588550e-01 -6.72366321e-01 2.18723610e-01 -9.70380008e-02 1.76231042e-01 -7.35823736e-02 7.20201373e-01 -4.03349519e-01 -9.92100775e-01 -1.42024979e-01 -7.39273190e-01 6.86881989e-02 6.73609614e-01 7.46805191e-01 2.08831146e-01 1.46989480e-01 6.02205157e-01 -6.79742455e-01 2.24964797e-01 -1.33528769e-01 -1.23990929e+00 -1.87742814e-01 -4.24385220e-01 -3.29245657e-01 1.12051618e+00 -3.02340060e-01 -8.66086125e-01 4.10234064e-01 1.28782419e-02 -3.00424606e-01 1.90206319e-02 -5.20191640e-02 -4.31354165e-01 -2.65506268e-01 5.47287822e-01 -2.35270485e-01 6.37997016e-02 1.01086564e-01 -2.21402720e-01 1.30508161e+00 6.74727798e-01 9.30898637e-02 2.19373435e-01 4.17745888e-01 3.03083569e-01 -1.24528313e+00 2.37272292e-01 1.94752246e-01 -7.74037540e-02 -4.11166668e-01 1.01437211e+00 -1.17995811e+00 -1.37061369e+00 5.93762755e-01 -1.27862680e+00 -1.32422134e-01 4.41591352e-01 5.65551162e-01 3.80958393e-02 -2.89903045e-01 -8.39797437e-01 -6.73188567e-01 -1.09435260e+00 -1.33759677e+00 7.25417972e-01 1.15502012e+00 2.02603444e-01 -4.95030522e-01 -8.10115457e-01 -1.83079362e-01 5.16432762e-01 2.04059631e-01 5.35028279e-01 -1.12361787e-02 -6.66803598e-01 -4.80086833e-01 -5.54962873e-01 4.32850033e-01 -5.04117645e-02 4.04115945e-01 -1.04829967e+00 -2.62481540e-01 2.99403518e-01 1.55389547e-01 5.93400359e-01 5.18384695e-01 9.67718065e-01 -1.28820613e-01 -4.88215774e-01 1.00890243e+00 1.95387125e+00 6.72586679e-01 5.73303163e-01 8.33552480e-02 3.65245759e-01 -1.67963915e-02 2.70084798e-01 4.79134619e-01 3.17608640e-02 3.37315291e-01 6.41772747e-01 9.14222654e-03 4.15483601e-02 2.43144274e-01 6.19682312e-01 3.95427644e-01 1.38120875e-01 -1.76847100e-01 -5.85945904e-01 -4.46517728e-02 -1.30211973e+00 -6.12470031e-01 -3.03880394e-01 1.94860137e+00 3.46563935e-01 5.74987471e-01 -1.37837932e-01 3.62304926e-01 7.69795418e-01 -2.59709865e-01 -6.58597112e-01 -8.64490986e-01 1.11988395e-01 6.04653597e-01 9.95227396e-01 1.27911463e-01 -9.75411475e-01 8.43641698e-01 4.36119986e+00 7.00408995e-01 -1.87773013e+00 9.93721411e-02 6.36747360e-01 -1.05002984e-01 7.17304409e-01 -1.81872517e-01 -1.30023575e+00 6.66169286e-01 1.18230581e+00 3.39780033e-01 1.03383422e-01 1.20400381e+00 1.13638453e-01 -2.64229536e-01 -8.44053984e-01 1.37386394e+00 -3.75422686e-01 -1.25518155e+00 -1.93482414e-01 -1.81696489e-01 2.26382181e-01 -1.40640646e-01 1.18280746e-01 5.52396998e-02 -5.16728997e-01 -7.84258246e-01 5.67323327e-01 2.04522267e-01 1.27532113e+00 -1.11930311e+00 1.12176716e+00 1.24854902e-02 -1.42698395e+00 -2.88127899e-01 -8.80228221e-01 -3.62524331e-01 -9.29075927e-02 4.82957363e-01 -9.22812164e-01 -1.54431686e-02 1.02906501e+00 2.96893030e-01 -3.17544371e-01 5.85556984e-01 1.83440655e-01 5.06329000e-01 -6.75609827e-01 -4.66298670e-01 1.45940647e-01 -2.07620915e-02 1.04607530e-01 1.25657892e+00 4.28776622e-01 4.14370626e-01 -4.06879425e-01 3.54710490e-01 -1.22135229e-01 -8.94071534e-02 -6.29386544e-01 2.47226790e-01 6.56326234e-01 1.87553453e+00 -1.04250300e+00 -6.22454464e-01 -4.68928784e-01 8.10383320e-01 -2.94440001e-01 -1.38416514e-01 -1.03824818e+00 -1.01053107e+00 7.79084384e-01 2.72116482e-01 1.70217037e-01 -1.69330269e-01 -3.29831392e-01 -5.89922667e-01 -4.32279073e-02 -5.77036500e-01 1.28153339e-01 -6.58349395e-01 -5.28589904e-01 9.29365933e-01 -4.38808411e-01 -1.15747380e+00 1.78774953e-01 -1.35978401e+00 -7.30290830e-01 7.20069230e-01 -1.31484771e+00 -1.05576527e+00 -8.72873783e-01 6.94053829e-01 6.39455080e-01 -5.20017505e-01 5.24106979e-01 5.83961666e-01 -9.62812424e-01 8.19402277e-01 -2.43958205e-01 3.41647178e-01 4.46656108e-01 -5.34630179e-01 2.60487467e-01 9.58318412e-01 -5.65526783e-01 4.25275147e-01 2.69435942e-01 -5.91043770e-01 -1.86259639e+00 -1.17680597e+00 2.69790411e-01 3.51139694e-01 3.44458282e-01 -5.94128847e-01 -4.08697963e-01 6.78831697e-01 1.81241259e-01 5.36348581e-01 1.85068563e-01 -9.79705155e-01 1.12521067e-01 -6.67420149e-01 -1.48759997e+00 5.57574391e-01 5.56308031e-01 -2.07100883e-01 2.27740392e-01 1.07754223e-01 5.55091679e-01 -6.32354319e-01 -6.03987992e-01 1.25212416e-01 7.17157483e-01 -1.02955592e+00 5.68134606e-01 -2.04158798e-01 -5.71954735e-02 -5.24241865e-01 -1.87787428e-01 -4.22141612e-01 3.64225954e-02 -6.46949530e-01 8.61138999e-02 9.94298339e-01 2.57081628e-01 -9.39735234e-01 1.25284386e+00 2.98548460e-01 -3.39707971e-01 -7.90252447e-01 -1.11696470e+00 -4.42749590e-01 -6.63856804e-01 -2.71395773e-01 5.52349329e-01 3.22096050e-01 -3.80769335e-02 4.99341995e-01 1.99934617e-01 5.44708550e-01 4.94372308e-01 -6.57108068e-01 5.33661485e-01 -1.14023113e+00 -2.32525673e-02 -2.53021091e-01 -1.21355772e+00 -6.42958283e-01 -4.26912993e-01 -5.74234188e-01 -5.51472306e-01 -6.70264304e-01 -4.74793136e-01 -3.65977943e-01 7.45782405e-02 3.28377604e-01 4.42360640e-01 4.86230642e-01 7.38319336e-03 -3.56718212e-01 -5.87902358e-03 -1.09260999e-01 6.64604664e-01 2.89004669e-02 -9.89528596e-02 1.21520320e-03 6.60823584e-02 8.12305748e-01 9.03249502e-01 -1.22990169e-01 -5.37531555e-01 -3.44671935e-01 2.44157791e-01 1.55709207e-01 5.98459303e-01 -1.57191205e+00 7.63489246e-01 4.43219841e-01 1.00900590e+00 -4.67391849e-01 5.06593227e-01 -1.19993317e+00 3.06962848e-01 1.05616558e+00 5.59950709e-01 5.97134888e-01 4.24775213e-01 1.52027784e-02 4.96948138e-02 -1.69395640e-01 9.12812173e-01 1.25271320e-01 -8.30735743e-01 4.36284035e-01 -3.10996860e-01 -6.26327634e-01 1.41072690e+00 -6.66092277e-01 -3.12302768e-01 2.93761045e-01 -4.56647187e-01 -1.07283100e-01 1.51834488e-01 6.77218884e-02 7.83257902e-01 -1.11732066e+00 -4.68282610e-01 8.05662632e-01 -3.88028592e-01 9.28735211e-02 5.57990849e-01 6.50860310e-01 -1.35262430e+00 6.83609426e-01 -6.92997992e-01 -5.02554893e-01 -1.47957230e+00 3.69896233e-01 5.86301446e-01 4.21434551e-01 -6.03220642e-01 7.89458573e-01 -2.91739732e-01 3.11726898e-01 1.82343975e-01 -7.15592861e-01 -2.74547130e-01 -1.08446375e-01 7.85276055e-01 6.17939293e-01 2.43218809e-01 -2.90574402e-01 -5.97619951e-01 6.74404204e-01 1.79909691e-01 2.30275374e-02 1.09480035e+00 3.02136570e-01 -2.25909293e-01 -3.17947716e-01 1.21880400e+00 -5.00395715e-01 -1.35604608e+00 4.53156888e-01 -2.56354332e-01 7.37152472e-02 3.39417219e-01 -5.30441463e-01 -1.31445050e+00 8.12289178e-01 1.27123725e+00 -1.41000077e-01 1.39018655e+00 -6.90001667e-01 8.11684132e-01 3.27517420e-01 3.55801523e-01 -1.09907126e+00 -3.26261491e-01 2.90158898e-01 2.31687844e-01 -9.83517885e-01 1.06874168e-01 -3.89026463e-01 -1.94675356e-01 1.71965683e+00 1.13515043e+00 -3.86483848e-01 1.09484243e+00 9.70274270e-01 5.23931719e-03 -1.68808565e-01 -6.66950405e-01 3.20859849e-01 -4.10311997e-01 6.70165956e-01 1.78522870e-01 1.52555034e-01 -2.82663733e-01 8.25852513e-01 -3.20695102e-01 3.54443580e-01 6.79282546e-01 9.44806814e-01 -5.57286799e-01 -4.28433239e-01 -4.20125157e-01 5.28651416e-01 -6.71621323e-01 -5.09305373e-02 2.99601227e-01 8.33648622e-01 6.75776660e-01 7.65147626e-01 6.72633052e-01 -6.25541270e-01 1.15392923e-01 -2.94288188e-01 1.65030614e-01 -1.63944677e-01 -9.49215531e-01 -1.73777625e-01 -2.33009443e-01 -5.48165083e-01 2.82694131e-01 -8.25810954e-02 -1.50939751e+00 -5.35130382e-01 -1.92228839e-01 -1.06285959e-01 1.21145618e+00 3.83725613e-01 5.39728045e-01 8.25638711e-01 4.01216596e-01 -6.55589879e-01 -2.45189309e-01 -8.98670435e-01 -4.63712633e-01 -6.53438985e-01 1.79854006e-01 -3.67662489e-01 5.51115489e-03 -7.62741594e-03]
[8.254861831665039, 2.642193078994751]
92462d61-bca5-41f8-822d-0fd2469949a4
learning-how-to-infer-partial-mdps-for-in
2302.04250
null
https://arxiv.org/abs/2302.04250v2
https://arxiv.org/pdf/2302.04250v2.pdf
Learning How to Infer Partial MDPs for In-Context Adaptation and Exploration
To generalize across tasks, an agent should acquire knowledge from past tasks that facilitate adaptation and exploration in future tasks. We focus on the problem of in-context adaptation and exploration, where an agent only relies on context, i.e., history of states, actions and/or rewards, rather than gradient-based updates. Posterior sampling (extension of Thompson sampling) is a promising approach, but it requires Bayesian inference and dynamic programming, which often involve unknowns (e.g., a prior) and costly computations. To address these difficulties, we use a transformer to learn an inference process from training tasks and consider a hypothesis space of partial models, represented as small Markov decision processes that are cheap for dynamic programming. In our version of the Symbolic Alchemy benchmark, our method's adaptation speed and exploration-exploitation balance approach those of an exact posterior sampling oracle. We also show that even though partial models exclude relevant information from the environment, they can nevertheless lead to good policies.
['Hado van Hasselt', 'Nan Rosemary Ke', 'Chentian Jiang']
2023-02-08
null
null
null
null
['thompson-sampling']
['methodology']
[ 3.40414703e-01 2.76496112e-01 -3.71233612e-01 -2.59412020e-01 -8.05849552e-01 -6.04521453e-01 8.99015963e-01 1.11667790e-01 -8.74823272e-01 1.37678862e+00 -7.88520277e-02 -3.30892354e-01 -1.55602386e-02 -6.99152946e-01 -9.86785710e-01 -7.25071192e-01 -1.02604240e-01 9.52835143e-01 3.61041605e-01 -6.16661645e-02 3.93514752e-01 3.01545352e-01 -1.51179838e+00 -9.69523937e-02 9.15508270e-01 7.81490743e-01 5.31204224e-01 9.28366959e-01 -1.19442582e-01 5.83422959e-01 -4.63326961e-01 -1.39494091e-01 1.74175382e-01 -5.15257239e-01 -8.99603128e-01 -8.38105753e-02 -3.12907010e-01 -3.00713480e-01 9.62774977e-02 1.18766606e+00 2.49931753e-01 6.17223322e-01 6.03348196e-01 -9.85170662e-01 -2.34337002e-01 7.49979973e-01 -2.10142776e-01 1.90709516e-01 2.71462739e-01 5.67615509e-01 7.24842310e-01 -5.15744865e-01 7.55424321e-01 1.39608264e+00 3.11350614e-01 6.32555008e-01 -1.67907929e+00 -1.96409479e-01 8.01148891e-01 2.41322026e-01 -9.00059998e-01 -4.33194250e-01 4.05724138e-01 -2.21243858e-01 1.15734720e+00 8.37154314e-02 8.84868681e-01 1.48697960e+00 2.47802243e-01 9.76230264e-01 1.46935403e+00 -5.94047368e-01 9.80967641e-01 4.07493234e-01 -2.20290184e-01 5.23202240e-01 2.26481929e-01 4.24264252e-01 -6.16452396e-01 -4.49222624e-01 7.10799575e-01 -1.30980223e-01 -9.16188136e-02 -8.89756083e-01 -1.20555663e+00 7.69231081e-01 2.64688041e-02 -1.38020590e-01 -5.74153960e-01 3.42943728e-01 2.45230630e-01 5.07027388e-01 5.42353950e-02 8.46942127e-01 -6.11887276e-01 -4.52370197e-01 -5.72614193e-01 5.04383922e-01 1.20792282e+00 9.30416703e-01 9.68960047e-01 7.53808394e-02 -3.99287343e-02 3.38884145e-01 1.11863352e-01 5.62553465e-01 5.18645644e-01 -1.29732585e+00 5.03179848e-01 -1.56975016e-02 7.63107061e-01 -9.09498408e-02 -1.20988511e-01 -3.32363188e-01 -1.71019390e-01 5.40597796e-01 5.90128064e-01 -5.52820265e-01 -1.01684022e+00 2.05285287e+00 4.43654358e-01 -3.27823013e-02 1.57713920e-01 5.77990711e-01 -5.15932262e-01 4.75922793e-01 1.78319693e-01 -5.83554745e-01 7.44167447e-01 -8.19807529e-01 -6.62378311e-01 -7.29763091e-01 4.70346183e-01 -2.33927816e-01 1.05642319e+00 7.24788964e-01 -1.40283358e+00 -4.04044211e-01 -1.00377989e+00 3.34848613e-01 -3.14838856e-01 -2.33236834e-01 8.29994261e-01 5.21275818e-01 -1.05543375e+00 9.41638410e-01 -1.39820743e+00 -3.38654101e-01 2.83026904e-01 3.71835768e-01 6.58596307e-02 1.32000208e-01 -1.08237815e+00 1.34255517e+00 8.75041008e-01 -1.22294471e-01 -1.70150900e+00 -1.93005607e-01 -5.76935053e-01 4.85771298e-02 1.03860331e+00 -5.96148551e-01 1.61418974e+00 -1.01660407e+00 -2.07525611e+00 2.27421552e-01 -1.54779181e-01 -7.62140155e-01 6.26662314e-01 -3.72071832e-01 1.59057125e-01 -1.01463489e-01 -1.99557677e-01 5.25174260e-01 1.00695360e+00 -9.88457620e-01 -8.80424380e-01 -3.89781117e-01 3.46951574e-01 6.42431915e-01 2.83845770e-03 -1.55531749e-01 -1.31689146e-01 -9.00192931e-02 -1.17599675e-02 -1.20251453e+00 -6.49963975e-01 -1.67822570e-01 -2.05418989e-01 -5.79049587e-02 2.70304054e-01 -5.69282770e-01 8.62601280e-01 -1.75385559e+00 4.72296953e-01 3.61109585e-01 -3.47506255e-01 -1.68684348e-01 9.21053290e-02 2.19642043e-01 2.60980129e-01 3.84649709e-02 -3.95629644e-01 -4.04195666e-01 2.77664900e-01 5.87777138e-01 -4.60534722e-01 3.33429843e-01 6.11121878e-02 7.90467381e-01 -1.09340382e+00 -3.63644242e-01 -7.32596219e-02 1.39882620e-02 -5.85447431e-01 1.39645979e-01 -1.12248302e+00 6.31608009e-01 -8.63563180e-01 4.91272599e-01 1.01308979e-01 -7.55625591e-02 5.66691399e-01 7.35422730e-01 -1.30759969e-01 4.40229416e-01 -1.39563620e+00 1.97808588e+00 -7.23857462e-01 2.43106827e-01 6.12351149e-02 -1.00536859e+00 6.23072386e-01 2.36445710e-01 -3.82645950e-02 -3.39994490e-01 -1.43143043e-01 2.77457416e-01 -1.35175269e-02 -3.62350315e-01 2.85439372e-01 -1.28115907e-01 1.70904957e-02 7.32579529e-01 1.29482195e-01 -3.76672089e-01 2.18369126e-01 -1.37634739e-01 1.04016316e+00 9.24531043e-01 5.88037252e-01 -2.02981859e-01 1.31514773e-01 2.24843830e-01 6.68087125e-01 1.31284142e+00 2.10334975e-02 -3.67041631e-03 8.24178398e-01 -3.23013812e-01 -1.14976990e+00 -9.88760650e-01 -1.13790277e-02 1.34254098e+00 -1.65164426e-01 -9.96443182e-02 -6.90510750e-01 -7.43361533e-01 -5.48809282e-02 1.23099828e+00 -7.39842236e-01 -1.51296914e-01 -6.18980706e-01 -8.77323985e-01 1.01930415e-02 5.72293282e-01 1.80525839e-01 -1.18743360e+00 -1.17773855e+00 5.64590037e-01 2.10137889e-01 -5.90533435e-01 -1.42250061e-01 7.34625399e-01 -1.35041177e+00 -7.16427386e-01 -5.93057811e-01 -1.08801797e-01 6.26790404e-01 -4.81740594e-01 1.17734289e+00 -4.61847365e-01 2.05182415e-02 4.78043973e-01 1.00803167e-01 -6.39090359e-01 -3.99305731e-01 1.59540936e-01 2.13616669e-01 -2.85665542e-01 6.22131899e-02 -8.32956731e-01 -4.04344171e-01 1.19512990e-01 -4.19718266e-01 6.92880293e-03 7.65818477e-01 1.14376223e+00 7.93147564e-01 -3.28057818e-02 3.57093126e-01 -9.85055149e-01 6.73653662e-01 -4.61523086e-01 -1.19642544e+00 4.59110737e-01 -8.93310070e-01 6.09426498e-01 4.81289119e-01 -7.84035385e-01 -1.65135550e+00 1.58019260e-01 4.47004467e-01 -2.77102202e-01 -1.67012811e-01 4.91883218e-01 -1.98966354e-01 2.01589003e-01 9.44132745e-01 3.56394559e-01 -2.73069162e-02 -6.65383756e-01 4.04896498e-01 1.03248350e-01 3.35285246e-01 -1.25833917e+00 3.15057695e-01 2.32593387e-01 9.15294364e-02 -3.91928256e-01 -7.03306973e-01 1.82568282e-01 -5.67422748e-01 1.26419693e-01 5.47561049e-01 -6.79676592e-01 -6.83568299e-01 3.00156549e-02 -9.24878895e-01 -9.51849461e-01 -7.04325318e-01 6.64015830e-01 -1.09972286e+00 -4.79085464e-03 -3.80510390e-01 -1.33355725e+00 1.80249617e-01 -1.20919979e+00 7.07722723e-01 3.56695503e-01 -2.25299507e-01 -1.06896830e+00 1.91871107e-01 -2.16916338e-01 4.16453093e-01 1.39408603e-01 9.46897626e-01 -8.43958080e-01 -1.01439309e+00 -5.66119179e-02 4.49504226e-01 9.40515250e-02 -2.85013765e-01 -3.47607404e-01 -8.88119519e-01 -4.79755491e-01 3.19020182e-01 -5.94784856e-01 8.68829608e-01 2.47066021e-01 1.19038558e+00 -6.10975862e-01 -6.14832819e-01 4.48556244e-01 1.19054735e+00 4.55786616e-01 2.75337607e-01 4.58801985e-01 2.29321197e-01 5.62146127e-01 8.63895595e-01 6.88350141e-01 1.50382519e-01 6.19492531e-01 3.48249763e-01 7.71505356e-01 3.58926922e-01 -4.72972810e-01 5.80739915e-01 2.05374379e-02 -2.68533647e-01 1.22361407e-01 -8.83044779e-01 4.96707618e-01 -2.05980396e+00 -8.27656031e-01 7.36543655e-01 2.57461786e+00 1.32619715e+00 5.48944592e-01 1.58801883e-01 -3.54118526e-01 3.83466184e-01 -7.86110312e-02 -1.31745827e+00 -4.45803553e-01 2.73629308e-01 3.45581591e-01 4.33360457e-01 8.28235388e-01 -7.52348781e-01 8.59293878e-01 7.00009823e+00 6.26372993e-01 -6.11094117e-01 2.55677432e-01 4.84609187e-01 -4.69977170e-01 -4.33023185e-01 3.91155601e-01 -1.11366463e+00 4.53678608e-01 1.15141010e+00 -2.23738164e-01 1.01646125e+00 1.21079671e+00 -1.95365459e-01 -6.60942435e-01 -1.47486174e+00 3.15029979e-01 -3.42884392e-01 -8.49695802e-01 -4.53429878e-01 1.61846176e-01 1.03063488e+00 2.01696187e-01 9.26552936e-02 6.47304296e-01 1.15646565e+00 -7.99052417e-01 7.91551769e-01 6.79585695e-01 4.46264029e-01 -8.08979273e-01 4.97480541e-01 8.38672280e-01 -4.80691522e-01 -3.83223861e-01 -5.02482653e-01 -2.01557711e-01 6.94843158e-02 3.17295820e-01 -9.39167142e-01 1.45562336e-01 4.20296401e-01 2.18105063e-01 -2.86908746e-01 8.32543433e-01 -6.77469671e-01 5.32869577e-01 -6.47258043e-01 -4.21217799e-01 2.39207059e-01 -3.18864882e-01 5.68167746e-01 9.21712816e-01 3.79468560e-01 1.80468913e-02 4.08603311e-01 1.10069680e+00 3.78283620e-01 -2.90444970e-01 -6.04767084e-01 1.77913494e-02 4.36784148e-01 7.92291939e-01 -5.22579312e-01 -5.76538146e-01 -8.25531930e-02 7.01605618e-01 5.44763744e-01 6.49125040e-01 -5.86990058e-01 -4.18658443e-02 5.74673891e-01 -3.05336207e-01 4.73163664e-01 -3.32941562e-01 -2.63808072e-01 -1.14066672e+00 -7.14520812e-02 -1.00531209e+00 3.59711856e-01 -5.35794199e-01 -6.86566234e-01 2.45422617e-01 5.46893179e-01 -5.49888134e-01 -9.77606833e-01 -4.46353704e-01 -4.82505828e-01 1.15192115e+00 -1.45276201e+00 -4.92269963e-01 2.89772749e-01 3.91035080e-01 7.47903645e-01 -7.29409233e-02 8.72066796e-01 -6.58098102e-01 -4.63515073e-01 1.47957250e-01 3.00889969e-01 -6.47641420e-01 3.78515512e-01 -1.61885989e+00 5.09659767e-01 7.56535947e-01 4.49433289e-02 7.40650296e-01 1.01657116e+00 -8.06598783e-01 -1.50365102e+00 -6.53087378e-01 4.18659687e-01 -7.35108674e-01 7.91195452e-01 -3.52502972e-01 -9.55618083e-01 1.15204716e+00 1.02521650e-01 -1.86859578e-01 1.15990862e-01 3.94989848e-01 4.75882441e-02 1.18231557e-01 -9.68872249e-01 9.14986134e-01 1.15335906e+00 -4.15048659e-01 -7.77702451e-01 3.65746170e-01 5.14385641e-01 -7.26999581e-01 -5.76513529e-01 -1.00149997e-01 5.00464201e-01 -7.90449440e-01 7.96915889e-01 -9.20580089e-01 -9.30844396e-02 -1.35265023e-01 -8.12722594e-02 -1.55382717e+00 -1.89566351e-02 -1.07097590e+00 -6.14345551e-01 7.34201550e-01 6.50159657e-01 -8.62474620e-01 7.18324065e-01 1.02203095e+00 1.11311138e-01 -7.22409368e-01 -9.19865608e-01 -9.50781405e-01 1.14924483e-01 -3.68318945e-01 6.87529385e-01 4.57237601e-01 5.45682199e-02 1.13320470e-01 -2.45508149e-01 3.67780924e-02 8.95137489e-01 2.09012061e-01 6.44422531e-01 -1.11548531e+00 -9.82250273e-01 -2.71398276e-01 3.57022077e-01 -1.04256487e+00 3.05328727e-01 -5.03404319e-01 4.32155788e-01 -1.12527347e+00 1.56640545e-01 -4.27826077e-01 -2.99963355e-01 4.87427384e-01 -1.34461477e-01 -7.66064465e-01 1.14120215e-01 1.61321402e-01 -8.70756805e-01 5.94783783e-01 1.21006119e+00 1.04671113e-01 -4.93669480e-01 4.10292953e-01 -4.61093545e-01 9.50048864e-01 9.08760190e-01 -6.25556767e-01 -7.03527391e-01 -3.04574341e-01 5.40875554e-01 4.78878498e-01 2.29344264e-01 -7.37592936e-01 4.28289056e-01 -7.09881425e-01 5.12509227e-01 -2.03222349e-01 5.01589656e-01 -4.39550012e-01 2.28149191e-01 6.33367419e-01 -7.73476899e-01 -1.12412684e-03 -1.43495733e-02 1.02573216e+00 2.07616791e-01 -6.21279061e-01 5.57257771e-01 -6.87484980e-01 -6.35712385e-01 1.36000827e-01 -4.89459306e-01 1.13368087e-01 8.13402355e-01 -1.64338648e-01 -3.35519202e-02 -3.15563291e-01 -1.25737631e+00 4.39975679e-01 6.37858510e-01 -5.53874299e-02 3.21474552e-01 -1.00272250e+00 -4.22435820e-01 2.24517155e-02 -9.26939249e-02 2.54181594e-01 -1.97400555e-01 7.67828941e-01 1.56512007e-01 3.74765605e-01 -2.45744824e-01 -3.70808363e-01 -6.82628334e-01 7.77130902e-01 2.80659556e-01 -6.08690858e-01 -4.06707406e-01 7.73643434e-01 9.79930013e-02 -1.63818091e-01 4.42712188e-01 -4.74066496e-01 -3.14944386e-02 -3.95548157e-02 4.37960058e-01 3.85880053e-01 -3.48699987e-01 4.31874216e-01 -6.28545880e-02 6.66197389e-02 -2.85727292e-01 -8.92857313e-01 1.19907463e+00 -2.26027563e-01 1.13694511e-01 8.50555122e-01 4.46612537e-01 -4.66995776e-01 -2.01767302e+00 -5.83584130e-01 3.81455660e-01 -6.45146251e-01 -6.15593791e-02 -1.07177556e+00 -3.34845990e-01 8.53451371e-01 3.90671015e-01 1.97252169e-01 7.54989266e-01 -1.66483536e-01 1.09387375e-01 1.07162642e+00 7.45522022e-01 -1.61569905e+00 5.34600019e-02 5.17018199e-01 7.92955875e-01 -1.16832066e+00 1.23430043e-01 4.40975338e-01 -8.48713458e-01 8.67346525e-01 5.54394901e-01 8.30556527e-02 2.86311626e-01 1.31051123e-01 -5.08248925e-01 2.52643019e-01 -1.35173047e+00 -2.58130640e-01 -2.22687751e-01 8.29605997e-01 -1.64486989e-01 8.72954205e-02 8.46696794e-02 3.55276376e-01 -4.11660261e-02 1.80232152e-01 3.81245136e-01 1.20817852e+00 -6.56009674e-01 -1.19141769e+00 -3.61956567e-01 3.98917735e-01 -4.51710522e-01 4.50087041e-02 -4.56816331e-02 7.48627782e-01 -2.16477439e-01 4.81668204e-01 -1.92074984e-01 2.86079854e-01 1.38056189e-01 3.89949381e-01 7.58935690e-01 -8.00191462e-01 -1.00287952e-01 1.23750374e-01 2.81288892e-01 -8.57064843e-01 -4.19738889e-02 -1.14517272e+00 -9.03213441e-01 4.74108756e-02 -2.01235712e-01 2.87263066e-01 9.20442343e-01 1.00975716e+00 1.04664087e-01 3.85630548e-01 4.37268406e-01 -9.53897119e-01 -1.44337738e+00 -9.95225608e-01 -5.31114161e-01 -1.47380978e-01 4.46652204e-01 -6.96681142e-01 -3.83411616e-01 -3.72500978e-02]
[4.141273021697998, 1.9219046831130981]
2a5e6cb8-62a7-4257-9542-d0feb048d9dd
impact-of-redundancy-on-resilience-in
2211.08622
null
https://arxiv.org/abs/2211.08622v1
https://arxiv.org/pdf/2211.08622v1.pdf
Impact of Redundancy on Resilience in Distributed Optimization and Learning
This report considers the problem of resilient distributed optimization and stochastic learning in a server-based architecture. The system comprises a server and multiple agents, where each agent has its own local cost function. The agents collaborate with the server to find a minimum of the aggregate of the local cost functions. In the context of stochastic learning, the local cost of an agent is the loss function computed over the data at that agent. In this report, we consider this problem in a system wherein some of the agents may be Byzantine faulty and some of the agents may be slow (also called stragglers). In this setting, we investigate the conditions under which it is possible to obtain an "approximate" solution to the above problem. In particular, we introduce the notion of $(f, r; \epsilon)$-resilience to characterize how well the true solution is approximated in the presence of up to $f$ Byzantine faulty agents, and up to $r$ slow agents (or stragglers) -- smaller $\epsilon$ represents a better approximation. We also introduce a measure named $(f, r; \epsilon)$-redundancy to characterize the redundancy in the cost functions of the agents. Greater redundancy allows for a better approximation when solving the problem of aggregate cost minimization. In this report, we constructively show (both theoretically and empirically) that $(f, r; \mathcal{O}(\epsilon))$-resilience can indeed be achieved in practice, given that the local cost functions are sufficiently redundant.
['Nitin H. Vaidya', 'Nirupam Gupta', 'Shuo Liu']
2022-11-16
null
null
null
null
['distributed-optimization']
['methodology']
[-5.25255799e-01 -2.38075331e-02 2.30652109e-01 1.42428493e-02 -5.76718152e-01 -5.51717401e-01 -2.04215106e-02 4.10167485e-01 -6.08254015e-01 8.10505390e-01 -4.09996271e-01 -3.92646864e-02 -5.71735799e-01 -6.63939953e-01 -8.91182005e-01 -1.00795305e+00 -7.93523490e-01 5.72877765e-01 6.13592528e-02 -3.00450325e-01 1.57858416e-01 4.55753624e-01 -1.57466710e+00 -1.48722544e-01 7.57812381e-01 1.26253641e+00 1.62543640e-01 8.62559378e-01 2.35565752e-01 8.62361431e-01 -1.04620469e+00 -1.46646142e-01 5.09836316e-01 -3.62737596e-01 -7.77898610e-01 5.71321361e-02 -3.44265513e-02 3.22866137e-03 3.24210804e-03 1.24487364e+00 3.93232107e-01 2.50027657e-01 3.38892877e-01 -1.44517243e+00 2.91379411e-02 6.24338031e-01 -4.64254379e-01 2.88107038e-01 2.25491717e-01 3.02822262e-01 9.72947776e-01 -2.24105313e-01 2.96300650e-01 1.02940178e+00 4.63642508e-01 3.34552616e-01 -1.21487427e+00 -3.69215459e-01 3.62548262e-01 -1.11013778e-01 -1.52124310e+00 -2.98352540e-01 4.19510305e-01 -2.42917866e-01 1.04820144e+00 4.23879892e-01 2.02942640e-01 1.87088743e-01 3.84893090e-01 2.55861461e-01 8.61020863e-01 -3.30721527e-01 7.10335493e-01 2.92736650e-01 -6.55868948e-02 6.67700291e-01 3.47950131e-01 -7.15983361e-02 -5.43578565e-01 -5.14205933e-01 4.71683443e-01 6.31880611e-02 -1.90363884e-01 -1.97081611e-01 -7.05504239e-01 7.19168961e-01 4.12426054e-01 2.41782531e-01 -5.11677384e-01 4.41639870e-01 2.81753600e-01 9.77841437e-01 5.22255719e-01 6.02465570e-01 -4.54789191e-01 3.12643172e-03 -5.52987397e-01 1.89358592e-01 9.66622174e-01 7.81495273e-01 8.51897657e-01 3.20782065e-01 3.67125809e-01 5.82953513e-01 -1.08245149e-01 2.73389071e-01 2.87814945e-01 -1.33518028e+00 6.72828913e-01 4.43391383e-01 5.64702570e-01 -8.11755896e-01 -3.57617557e-01 -4.76072252e-01 -7.26467371e-01 5.32643437e-01 3.34046364e-01 -2.91870773e-01 1.01538278e-01 2.04220080e+00 2.95391679e-01 2.13300269e-02 1.15275271e-01 8.97515535e-01 -3.71089667e-01 6.35053039e-01 -4.31195021e-01 -8.85069370e-01 8.73545706e-01 -7.58486509e-01 -2.51891285e-01 9.72304195e-02 7.20894754e-01 -6.18312657e-01 1.05144262e+00 4.89263892e-01 -1.38564503e+00 5.01806103e-02 -8.62264931e-01 8.40040982e-01 1.00158919e-02 -3.64291877e-01 1.18519343e-01 5.95634580e-01 -1.28491759e+00 7.27879822e-01 -8.08588982e-01 -2.10851687e-03 -1.02570668e-01 6.07300460e-01 -5.25370352e-02 -2.79082488e-02 -6.65203035e-01 8.94894958e-01 3.11247800e-02 -6.68865517e-02 -1.22091961e+00 -5.02648652e-01 -1.83216915e-01 8.17744136e-02 6.39625132e-01 -5.96541226e-01 1.11987484e+00 -1.10598361e+00 -1.31493914e+00 2.57014960e-01 1.05682328e-01 -3.78469557e-01 7.88114369e-01 1.28753558e-01 1.02381140e-01 2.11458310e-01 9.87883136e-02 -2.68204451e-01 6.78792775e-01 -1.18738747e+00 -6.55821145e-01 -4.26953375e-01 4.54769403e-01 3.39675009e-01 -4.00158525e-01 2.34520331e-01 3.27394426e-01 -2.52292246e-01 -2.07737371e-01 -9.68068123e-01 -2.95046121e-01 -1.16403103e-01 -1.08396776e-01 -3.59176666e-01 4.42508757e-01 -9.86668169e-02 9.09915209e-01 -2.20184278e+00 9.40910056e-02 4.05236095e-01 3.95286471e-01 -1.19292252e-01 -2.03824371e-01 7.56167948e-01 1.71597719e-01 2.09674805e-01 -2.29579937e-02 -4.96608704e-01 -1.02702320e-01 2.40005732e-01 4.83220629e-02 9.70332205e-01 -2.92954445e-01 -3.99922132e-02 -7.77493834e-01 -9.01253670e-02 -2.95584589e-01 8.54301304e-02 -6.63819551e-01 2.82363415e-01 -3.94817978e-01 -5.13628609e-02 -5.76097667e-01 4.07371759e-01 3.61311615e-01 -2.59346098e-01 2.96804965e-01 4.79918599e-01 -2.18478054e-01 -9.95962322e-02 -1.70367765e+00 8.53292644e-01 -6.21925652e-01 2.58319825e-01 7.10432291e-01 -1.14473021e+00 6.79760516e-01 4.17540193e-01 8.24215591e-01 -2.74447650e-01 5.96062327e-03 4.28037584e-01 -1.87784761e-01 -2.58366764e-01 3.00221682e-01 -2.37807080e-01 -3.29949185e-02 9.82290089e-01 -4.23844308e-01 2.13404983e-01 2.29331523e-01 2.13528574e-01 1.58325946e+00 -7.37100005e-01 -1.88812852e-01 -3.90325069e-01 4.28539962e-01 -2.17159301e-01 5.49405217e-01 9.10856068e-01 -2.99468040e-01 -1.01462111e-01 8.81206393e-01 -3.96665603e-01 -1.11254096e+00 -9.70428288e-01 4.77822155e-01 1.27978814e+00 2.86829054e-01 -2.99783289e-01 -7.45373011e-01 -4.54397976e-01 1.68389276e-01 5.60626328e-01 -3.37899953e-01 -2.15563491e-01 -4.44623649e-01 -5.63533366e-01 4.20867920e-01 2.01544762e-01 2.69262671e-01 -7.63992012e-01 -8.08776081e-01 2.29959354e-01 4.54610214e-02 -5.73324859e-01 -8.44146073e-01 1.87554717e-01 -8.24599743e-01 -1.08912611e+00 -3.23731720e-01 -3.38793814e-01 9.12763357e-01 3.99622351e-01 8.61895442e-01 3.41593921e-01 -8.03231671e-02 5.11852443e-01 -2.35722318e-01 -1.28832161e-01 -4.32638913e-01 -2.39927992e-02 5.20490706e-01 3.88413072e-02 -4.83640552e-01 -5.17394543e-01 -5.95629930e-01 3.23727787e-01 -1.02382016e+00 -6.61283255e-01 1.15866974e-01 5.73175907e-01 5.33496439e-01 6.47302747e-01 6.50659740e-01 -4.76459116e-01 8.12976480e-01 -4.41909701e-01 -1.01498139e+00 2.24419326e-01 -7.62133896e-01 2.93579668e-01 1.32960641e+00 -5.56344151e-01 -4.73651499e-01 -1.69488013e-01 5.58461189e-01 -5.44883847e-01 3.80280405e-01 3.54996055e-01 1.05687261e-01 -2.52491713e-01 7.00588226e-01 2.54393607e-01 3.83896559e-01 -4.08932656e-01 1.76998675e-01 4.38805938e-01 -5.45745380e-02 -7.79012322e-01 4.57425535e-01 3.21917146e-01 2.45100051e-01 -6.26854002e-01 -2.00423166e-01 -6.57685772e-02 2.62009501e-01 -3.19004267e-01 -8.79977345e-02 -4.94551301e-01 -1.60196257e+00 2.76939780e-01 -9.19986308e-01 -3.82438391e-01 -5.54639876e-01 2.41031617e-01 -9.10393059e-01 1.40381098e-01 -4.78832215e-01 -1.37805057e+00 -1.81390733e-01 -1.16522241e+00 3.04612786e-01 1.02346517e-01 1.97477147e-01 -6.78537369e-01 3.96574289e-02 4.30597328e-02 6.82582438e-01 1.11173011e-01 8.83101463e-01 -8.22515666e-01 -6.86328709e-01 -3.68353903e-01 1.72231495e-01 4.39492315e-01 -1.78345084e-01 -1.06234038e-02 -1.90902680e-01 -8.83717179e-01 3.17679942e-01 -2.11302310e-01 3.10471565e-01 2.43099749e-01 8.77131164e-01 -1.21830177e+00 -5.99095423e-04 2.20608741e-01 1.80532157e+00 1.70726165e-01 9.02097076e-02 3.69303942e-01 9.95460376e-02 6.17764890e-01 2.98511565e-01 9.43078816e-01 3.13980699e-01 4.67381775e-01 8.00953031e-01 4.65136349e-01 3.82188886e-01 2.59771705e-01 6.85318351e-01 6.83043420e-01 -1.23626672e-01 -2.52399832e-01 -7.06997454e-01 4.95027333e-01 -1.72479343e+00 -8.44481468e-01 3.86074036e-01 2.72461557e+00 8.85030746e-01 7.01993927e-02 4.34850574e-01 -3.18378247e-02 8.39314461e-01 2.52894294e-02 -7.18191624e-01 -7.32776105e-01 -2.18892414e-02 -4.11889004e-03 6.40028059e-01 6.03451669e-01 -3.54831427e-01 3.83890390e-01 6.12335587e+00 7.38871276e-01 -1.01324439e+00 3.04426312e-01 6.28417790e-01 -7.83042371e-01 -9.26984474e-02 1.74238458e-02 -3.28765601e-01 8.82377326e-01 1.17644727e+00 -4.45292503e-01 1.03522027e+00 1.03056324e+00 4.01409179e-01 -3.04930925e-01 -1.04696369e+00 5.11289537e-01 -2.09184349e-01 -1.03412747e+00 -4.82190281e-01 2.21659377e-01 8.48863125e-01 9.78323668e-02 2.33322214e-02 -2.64146417e-01 6.81581676e-01 -7.79959917e-01 7.13848770e-01 5.41336238e-01 4.75753456e-01 -1.25932848e+00 6.77985847e-01 7.44515002e-01 -1.09402180e+00 -5.10771751e-01 -1.61933079e-01 -1.77334636e-01 -1.26768306e-01 6.81605637e-01 -4.52763289e-01 4.22423005e-01 7.64248788e-01 -3.10220689e-01 7.26518556e-02 9.09382999e-01 3.90450984e-01 1.18355438e-01 -7.10380614e-01 -2.62949973e-01 1.18875891e-01 -2.21639201e-01 6.35683000e-01 6.89319611e-01 2.72905856e-01 1.45781308e-01 6.17893636e-01 5.27637959e-01 -2.72235185e-01 1.63569465e-01 -2.49757916e-01 1.21984355e-01 7.97664881e-01 7.86988258e-01 -4.81173605e-01 -3.00228983e-01 -8.36534947e-02 6.99410200e-01 4.95011866e-01 4.44230169e-01 -5.86929023e-01 -5.84696651e-01 1.07177544e+00 1.73117369e-01 4.20170324e-03 -4.19568032e-01 -1.18066259e-01 -9.59747672e-01 2.49875620e-01 -8.62134337e-01 5.36204636e-01 -4.16463554e-01 -1.31471622e+00 6.16599083e-01 -3.51912111e-01 -9.41857636e-01 -3.97160381e-01 -5.93959615e-02 -5.53966522e-01 7.21908391e-01 -1.00385070e+00 -2.76774824e-01 2.09305272e-01 7.21258879e-01 2.51069486e-01 -2.89501905e-01 5.53795218e-01 1.99728776e-02 -8.49604428e-01 6.44562066e-01 6.81994021e-01 -5.39779425e-01 2.46046871e-01 -9.29380000e-01 -4.75369215e-01 7.56194890e-01 -3.25417191e-01 4.47092265e-01 1.07591629e+00 -3.38436723e-01 -1.48635721e+00 -9.25954223e-01 7.46312320e-01 7.13491812e-02 6.98111951e-01 1.05536468e-01 -5.11627853e-01 5.39375603e-01 -1.18143268e-01 1.01448901e-01 3.02748978e-01 -2.09382936e-01 -3.01038951e-01 -6.53630257e-01 -1.60865498e+00 4.89156276e-01 5.98660886e-01 -5.05361557e-01 1.40320286e-01 6.77224040e-01 6.61069274e-01 2.24058218e-02 -8.27876925e-01 -1.55113190e-01 1.34655029e-01 -9.46880877e-01 4.68009621e-01 -5.38303912e-01 6.44505629e-03 -3.21983069e-01 -3.29904348e-01 -1.35659575e+00 4.21079397e-02 -9.57351983e-01 -2.05493346e-01 6.90610051e-01 1.72773972e-01 -1.16630435e+00 7.46467412e-01 6.82494998e-01 1.72842946e-02 -8.24529827e-01 -1.63376415e+00 -1.24760246e+00 1.89636886e-01 -1.49350971e-01 3.77692252e-01 6.30829573e-01 2.22282782e-01 -3.50537479e-01 -6.63531050e-02 3.25822502e-01 7.84898460e-01 -1.50220573e-01 4.50022519e-01 -8.13199818e-01 -5.63896060e-01 -5.89234769e-01 -1.88141793e-01 -6.51441813e-01 1.64414585e-01 -5.33280909e-01 9.70550254e-02 -9.83827949e-01 2.14225322e-01 -8.32935154e-01 -3.77450079e-01 3.99589211e-01 1.97960526e-01 -3.15444708e-01 4.88579363e-01 6.29972100e-01 -8.51054728e-01 3.30569118e-01 6.38676345e-01 1.09989211e-01 -4.00436550e-01 3.34907442e-01 -4.84806836e-01 4.37046170e-01 8.58962059e-01 -8.80681694e-01 -2.84472495e-01 -4.39479440e-01 3.62826347e-01 4.63139266e-01 2.30709374e-01 -7.54443109e-01 4.91200536e-01 -5.28725266e-01 -4.76349443e-01 -2.96393558e-02 2.87539482e-01 -8.35754752e-01 3.58890235e-01 8.85809600e-01 -5.26349843e-01 4.51764017e-01 -4.34618801e-01 5.28704405e-01 1.60937905e-01 -4.41836625e-01 8.51129353e-01 -2.24455029e-01 2.79534876e-01 1.80440828e-01 -5.61621428e-01 -1.63893595e-01 1.27287614e+00 7.17070922e-02 -4.85165179e-01 -6.19447291e-01 -4.65421557e-01 5.02740562e-01 5.78500748e-01 -3.34269255e-01 3.24879467e-01 -7.69147038e-01 -6.25907481e-01 -4.34918106e-02 -3.70610386e-01 -1.33481562e-01 2.79041767e-01 9.23147202e-01 -5.02427638e-01 1.61806017e-01 1.05611019e-01 -2.82011867e-01 -1.02909935e+00 5.79733074e-01 7.49944985e-01 -2.87193567e-01 -1.25801459e-01 8.34529221e-01 -3.59759599e-01 7.46681355e-03 3.65832269e-01 -9.57506821e-02 6.25698090e-01 -3.73591743e-02 3.64961863e-01 9.61600721e-01 1.12723440e-01 -2.73194045e-01 -4.57139432e-01 1.78670824e-01 -1.03519633e-02 -4.77247566e-01 1.44414198e+00 -3.71855021e-01 -4.31316674e-01 2.03748271e-01 1.05967498e+00 -7.80461803e-02 -1.32487547e+00 -2.99082577e-01 -2.13959679e-01 -3.58462811e-01 -1.60339445e-01 -6.08917117e-01 -1.29940820e+00 3.56692314e-01 4.02497500e-01 8.02910149e-01 1.27814806e+00 1.05914623e-01 3.99432421e-01 4.20600951e-01 8.74386668e-01 -1.39368784e+00 3.47493738e-01 4.35891986e-01 6.58835530e-01 -6.28374338e-01 -8.65485966e-02 3.97294462e-02 -5.87200224e-01 9.34129238e-01 4.20223296e-01 -5.50422966e-01 4.29647982e-01 3.14918041e-01 -2.82895714e-01 4.96177189e-02 -1.18840241e+00 8.47498998e-02 -6.10506058e-01 3.03613454e-01 8.56828876e-03 2.91103065e-01 -3.71961206e-01 5.10087907e-01 -1.35434151e-01 -3.18818659e-01 9.06698227e-01 1.07265377e+00 -8.32431734e-01 -8.86128068e-01 -5.42055368e-01 4.11450535e-01 -4.63399023e-01 1.91568941e-01 -8.52255300e-02 3.55742484e-01 7.50640556e-02 1.24552596e+00 1.35452732e-01 -2.08415747e-01 3.25456113e-01 -8.55146572e-02 1.50474593e-01 -3.85258824e-01 -8.50304246e-01 2.43512467e-02 -2.51220483e-02 -5.75894713e-01 -1.38589159e-01 -5.19946814e-01 -1.36569417e+00 -9.34451163e-01 -2.83548594e-01 4.74692225e-01 8.35244179e-01 8.72162342e-01 4.47773933e-01 2.69838214e-01 1.49786150e+00 -4.57596481e-01 -1.22062278e+00 -5.77679574e-01 -9.93336976e-01 -1.04186215e-01 4.31910187e-01 -3.14966589e-01 -7.96234250e-01 -2.45545372e-01]
[6.108375072479248, 4.900088787078857]