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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
40abf085-36aa-4d5c-b440-c6ea989bfa1a | an-in-depth-analysis-of-the-effect-of-lexical | null | null | https://aclanthology.org/D19-5515 | https://aclanthology.org/D19-5515.pdf | An In-depth Analysis of the Effect of Lexical Normalization on the Dependency Parsing of Social Media | Existing natural language processing systems have often been designed with standard texts in mind. However, when these tools are used on the substantially different texts from social media, their performance drops dramatically. One solution is to translate social media data to standard language before processing, this is also called normalization. It is well-known that this improves performance for many natural language processing tasks on social media data. However, little is known about which types of normalization replacements have the most effect. Furthermore, it is unknown what the weaknesses of existing lexical normalization systems are in an extrinsic setting. In this paper, we analyze the effect of manual as well as automatic lexical normalization for dependency parsing. After our analysis, we conclude that for most categories, automatic normalization scores close to manually annotated normalization and that small annotation differences are important to take into consideration when exploiting normalization in a pipeline setup. | ['Rob van der Goot'] | 2019-11-01 | null | null | null | ws-2019-11 | ['lexical-normalization'] | ['natural-language-processing'] | [ 2.31376782e-01 2.16346696e-01 -8.38779956e-02 -5.91334105e-01
-6.05377614e-01 -7.78031170e-01 6.00226581e-01 8.59153330e-01
-9.98087525e-01 6.43221319e-01 6.12212956e-01 -2.21498489e-01
1.35751531e-01 -6.23145640e-01 -3.49765569e-01 -3.64401639e-01
4.86773640e-01 5.01279235e-01 3.62047672e-01 -5.38128853e-01
2.19655618e-01 2.74842381e-01 -1.22627997e+00 1.25518218e-01
6.72160387e-01 6.29303530e-02 3.32008190e-02 6.01016760e-01
-7.51977265e-01 6.96433842e-01 -8.62189174e-01 -8.18359017e-01
2.75419205e-01 -3.96717966e-01 -1.06541407e+00 -3.36071402e-01
2.24321634e-01 -9.50894225e-03 3.77235293e-01 1.29301536e+00
4.61961180e-01 1.48702553e-02 3.38616669e-01 -7.58708894e-01
-2.42162392e-01 1.46169972e+00 -4.33341354e-01 3.82138282e-01
5.66690326e-01 -1.15857393e-01 1.14573097e+00 -5.03213584e-01
9.69306707e-01 1.21804368e+00 7.36918449e-01 4.04824823e-01
-1.34446561e+00 -3.92232984e-01 5.65158539e-02 -3.06243807e-01
-1.13411760e+00 -5.92801094e-01 2.55945116e-01 -4.50557798e-01
1.28468227e+00 2.01645836e-01 1.14190102e-01 8.34322929e-01
1.25487059e-01 3.64990562e-01 9.30697858e-01 -1.00747454e+00
-9.91697833e-02 2.51713723e-01 6.81129217e-01 2.12013021e-01
5.28583944e-01 -6.15317523e-01 -3.68941784e-01 -1.54858053e-01
7.64088556e-02 -4.05643553e-01 -4.10708040e-03 1.20983943e-01
-1.20777702e+00 8.50276530e-01 -1.40243232e-01 8.46021235e-01
-2.39838690e-01 -2.57795542e-01 8.09016585e-01 5.16369581e-01
6.16419971e-01 6.83840275e-01 -6.96863115e-01 -1.94317445e-01
-8.10942292e-01 2.20552132e-01 1.09446502e+00 9.78995740e-01
8.69316697e-01 -4.19365108e-01 -1.04172133e-01 9.47358966e-01
1.71110645e-01 3.63169193e-01 5.77295125e-01 -5.46902120e-01
5.67471504e-01 8.01822722e-01 -9.74570960e-02 -8.40645790e-01
-8.28859210e-01 -1.39250055e-01 -4.21536833e-01 -8.52777734e-02
1.02111483e+00 -3.77107978e-01 -5.00006557e-01 1.83660424e+00
3.21748853e-01 -4.88450021e-01 2.59345800e-01 5.24212301e-01
8.14349592e-01 2.03452438e-01 4.70851451e-01 -3.80507797e-01
1.57641244e+00 -5.63007891e-01 -9.92456675e-01 -3.90877128e-01
1.17706394e+00 -1.48784363e+00 1.32583022e+00 9.88311991e-02
-9.16943848e-01 -2.36239940e-01 -8.87595594e-01 -2.63867080e-01
-5.32855928e-01 -3.27968508e-01 5.63146770e-01 9.98062134e-01
-6.79973841e-01 7.71278203e-01 -8.62755716e-01 -1.09454811e+00
-2.07818076e-01 3.68879348e-01 -4.45205510e-01 1.13923870e-01
-1.31473815e+00 1.12091017e+00 4.10405993e-01 -3.69050145e-01
2.66227752e-01 -4.94616568e-01 -7.98228860e-01 -1.20598860e-01
5.35114765e-01 -3.58691156e-01 1.50712764e+00 -1.06995535e+00
-1.39986241e+00 1.14275551e+00 -1.88668013e-01 -3.38414073e-01
5.84919035e-01 -3.54171187e-01 -3.30219835e-01 -2.75797069e-01
3.41491431e-01 2.36659855e-01 3.13420147e-01 -7.72238612e-01
-7.14422286e-01 -2.66698450e-01 2.66671509e-01 1.79541260e-01
-5.14599621e-01 9.32083905e-01 -4.86369371e-01 -6.31219447e-01
-5.21294810e-02 -9.32565928e-01 -3.05526644e-01 -6.58959210e-01
-3.47722709e-01 -1.89718649e-01 2.44130030e-01 -7.59586692e-01
1.41860974e+00 -2.14182186e+00 1.96409151e-02 2.58672684e-01
6.20834380e-02 2.66893297e-01 -7.91390017e-02 6.07744813e-01
-1.02817915e-01 6.34294748e-01 -2.45450929e-01 -4.57757980e-01
8.11195280e-03 4.42966729e-01 9.49997641e-03 3.26869935e-01
1.52605519e-01 6.62945509e-01 -9.15179670e-01 -5.70873678e-01
-1.04141377e-01 3.40940237e-01 -4.44683433e-01 6.25179410e-02
-8.47156793e-02 4.48959649e-01 -1.77044019e-01 2.46560872e-01
5.44466972e-01 3.08192044e-01 4.12223250e-01 9.46268812e-03
-4.58564103e-01 6.61277831e-01 -1.24004829e+00 1.58618116e+00
-1.10248514e-01 6.07528448e-01 4.43442985e-02 -6.28287017e-01
6.90766156e-01 3.22767377e-01 3.63977432e-01 -5.59765935e-01
3.63230199e-01 1.60009578e-01 3.95874918e-01 -5.01564980e-01
7.14547515e-01 -2.94176042e-01 -3.93682092e-01 7.49928296e-01
1.04655541e-01 2.45798994e-02 8.38348150e-01 3.74817461e-01
1.17472303e+00 -1.55282930e-01 7.09064126e-01 -4.19183522e-01
5.20685554e-01 2.80939609e-01 7.14053094e-01 7.03054488e-01
-5.71485236e-02 6.86304033e-01 8.44730139e-01 -8.59649703e-02
-1.04606986e+00 -4.30959731e-01 -1.15232076e-02 1.40239632e+00
-3.76380354e-01 -9.31365073e-01 -1.26030254e+00 -6.25903487e-01
-3.66530985e-01 7.77168810e-01 -3.49517941e-01 6.66234922e-03
-7.76537538e-01 -9.81915116e-01 9.16019440e-01 2.76443154e-01
-5.62472306e-02 -1.11094844e+00 -3.59028518e-01 3.50416094e-01
-2.25014791e-01 -1.38870037e+00 -3.03548694e-01 2.90754497e-01
-6.04440749e-01 -1.15748250e+00 -2.33051658e-01 -5.89309871e-01
3.85922104e-01 2.07561269e-01 1.23873615e+00 4.02141750e-01
1.69208899e-01 2.31563121e-01 -7.64776886e-01 -7.05320239e-01
-1.00548804e+00 7.77029335e-01 3.49293463e-02 -2.25451186e-01
8.91863644e-01 -2.86930054e-01 7.12180063e-02 1.08692646e-01
-1.07405555e+00 -2.22253218e-01 4.61623788e-01 3.98818225e-01
2.55385280e-01 -1.59012794e-01 2.91427970e-01 -1.76219320e+00
9.44488704e-01 -3.12026858e-01 -2.08283141e-01 2.60632604e-01
-5.41403115e-01 2.38545135e-01 7.46028125e-01 -2.66934931e-01
-1.07034302e+00 1.21652626e-01 -5.24184585e-01 3.73481929e-01
-5.03894150e-01 7.53965080e-01 -2.26146296e-01 2.55610794e-01
9.24901128e-01 -5.44583619e-01 -1.74840078e-01 -5.91329396e-01
3.07994097e-01 6.65546358e-01 2.31041625e-01 -5.84595084e-01
7.76371956e-01 1.06190950e-01 -4.65249807e-01 -1.07587516e+00
-1.02634382e+00 -6.44162059e-01 -1.06742561e+00 1.79366112e-01
9.74652827e-01 -6.20558858e-01 -1.02774024e-01 3.15233976e-01
-1.21286786e+00 -3.32824916e-01 -3.54917228e-01 4.83108699e-01
-8.95371288e-02 5.10438323e-01 -5.55242419e-01 -4.72602129e-01
-2.99208194e-01 -9.48621273e-01 4.78684813e-01 2.11668864e-01
-9.08277214e-01 -1.02117074e+00 2.42929474e-01 3.27267051e-01
3.73980224e-01 9.66317728e-02 9.26293850e-01 -1.12021101e+00
1.99078277e-01 -3.70765418e-01 -9.00446475e-02 3.30673784e-01
2.33983099e-01 4.22064096e-01 -9.02745247e-01 2.75615323e-02
-1.72791272e-01 3.70183238e-03 4.91958171e-01 1.01447202e-01
2.10838467e-01 1.47138471e-02 2.62852330e-02 1.97827801e-01
1.10094190e+00 -2.52630800e-01 5.03638983e-01 5.04957139e-01
7.14175463e-01 8.33161652e-01 6.94991112e-01 1.13772437e-01
2.96900898e-01 4.58827615e-01 -1.06052324e-01 -8.06643441e-02
2.34619025e-02 -4.36499529e-03 5.57986557e-01 1.09764826e+00
5.64699508e-02 -2.96017438e-01 -1.17749965e+00 3.29978555e-01
-1.81274986e+00 -7.90471733e-01 -5.39411247e-01 2.03459764e+00
9.86019194e-01 6.41371906e-01 2.23181859e-01 2.60704786e-01
8.30915570e-01 -2.41549406e-02 1.44192232e-02 -7.58005559e-01
-3.59756172e-01 2.99361348e-01 6.76728606e-01 6.90679610e-01
-1.07236016e+00 1.19115806e+00 6.11937189e+00 3.20711136e-01
-1.06186080e+00 4.33173954e-01 1.29231259e-01 3.14890817e-02
-1.11394770e-01 3.66278827e-01 -1.10343552e+00 8.88355076e-02
1.18184948e+00 -2.93200463e-01 2.34612569e-01 5.67608774e-01
3.15055132e-01 -1.04641691e-01 -1.05728257e+00 5.54845452e-01
4.11362434e-03 -7.44369090e-01 -3.33002508e-01 -2.19137907e-01
2.90980041e-01 3.19617927e-01 -7.45958030e-01 3.59127790e-01
5.00555992e-01 -6.36093199e-01 7.43096352e-01 9.58847031e-02
3.58105928e-01 -6.23505235e-01 1.11661136e+00 4.04945672e-01
-8.38546038e-01 3.36531848e-01 -4.69966501e-01 -3.91729742e-01
4.68687862e-01 7.60752499e-01 -8.21922481e-01 4.25074548e-01
4.84690636e-01 4.34333771e-01 -9.83732700e-01 7.23667324e-01
-5.49116731e-01 8.26053381e-01 -5.24395525e-01 -4.63513732e-02
2.79063322e-02 -5.43375015e-02 4.73559588e-01 1.70320737e+00
-1.08818956e-01 -3.21087390e-02 1.45763069e-01 1.95272043e-01
-4.32663448e-02 8.49145055e-01 -4.32872683e-01 -2.78336078e-01
1.83959946e-01 1.54467785e+00 -1.32712746e+00 -3.03918839e-01
-7.54837036e-01 6.01074159e-01 5.63360453e-01 1.74054578e-02
-6.14579141e-01 -3.81457925e-01 6.30631149e-01 4.38933969e-01
-2.03658611e-01 -2.61582196e-01 -3.75002801e-01 -1.31023121e+00
-4.42456938e-02 -9.14145529e-01 6.94538713e-01 -3.12900424e-01
-1.30004263e+00 5.95022619e-01 1.02307945e-02 -7.26161242e-01
-1.03775531e-01 -6.19119883e-01 -5.56679130e-01 7.04446912e-01
-1.30882251e+00 -7.55391121e-01 -2.64795832e-02 3.24494839e-01
2.46739149e-01 1.40691757e-01 9.62326288e-01 6.06556058e-01
-7.52547145e-01 7.47391462e-01 -2.18034342e-01 3.74016076e-01
1.33645511e+00 -1.30837333e+00 6.09681487e-01 1.09265769e+00
2.19731048e-01 9.28349316e-01 1.06124651e+00 -6.77292824e-01
-9.63920236e-01 -8.98249865e-01 1.54473591e+00 -5.20125926e-01
9.26191568e-01 -4.23864245e-01 -1.25406229e+00 7.29550064e-01
5.50977588e-01 -3.70634437e-01 1.07084572e+00 5.35339713e-01
-3.62093180e-01 2.00605527e-01 -9.85965788e-01 6.05061471e-01
8.73853922e-01 -3.29924554e-01 -8.25527012e-01 4.67667341e-01
7.65616119e-01 -2.74933606e-01 -9.29881155e-01 -4.42950018e-02
3.68835747e-01 -6.93748713e-01 4.71190870e-01 -7.19594300e-01
2.61053056e-01 -2.70360649e-01 -9.45044905e-02 -1.20556486e+00
-2.03021690e-01 -5.98295808e-01 7.86522806e-01 1.85428810e+00
7.62890279e-01 -7.24857748e-01 4.68677402e-01 9.89479244e-01
-7.38647720e-03 6.48886338e-02 -4.50933248e-01 -4.71091032e-01
2.76829213e-01 -4.62025583e-01 4.14859951e-01 1.42362607e+00
2.65379310e-01 9.28825021e-01 2.44701579e-02 4.03930098e-02
2.29616702e-01 -4.10650343e-01 1.09817111e+00 -1.49174249e+00
-9.47995111e-02 -3.83784980e-01 -3.48861724e-01 -2.84452438e-01
3.01870674e-01 -8.88688564e-01 -4.95078675e-02 -1.37424862e+00
1.97644997e-02 -3.27231109e-01 -6.65952563e-02 5.64371169e-01
-2.80906051e-01 3.55435163e-01 2.69716561e-01 1.16960213e-01
-5.76397896e-01 -1.62317172e-01 6.41897798e-01 1.76939726e-01
-5.31995118e-01 -2.47857243e-01 -9.12707865e-01 1.00086653e+00
1.02465987e+00 -1.01442814e+00 9.13264081e-02 -6.58475459e-01
7.72883236e-01 -5.86697459e-01 -3.74258786e-01 -7.29246080e-01
2.92305768e-01 -2.44757459e-01 -1.65540561e-01 -1.52612776e-01
-4.34321553e-01 -5.07768393e-01 -4.60358476e-03 2.01585621e-01
-3.39263916e-01 2.19368085e-01 2.36459464e-01 -2.68926732e-02
-1.00259170e-01 -7.60312796e-01 8.55943620e-01 -2.05662355e-01
-5.17853200e-01 -3.05062056e-01 -7.16034889e-01 3.67350727e-01
8.10011208e-01 6.69884831e-02 -2.01439589e-01 -1.68790802e-01
-7.52833903e-01 -1.54471956e-02 7.57196307e-01 5.08502066e-01
-3.33656520e-01 -8.40162754e-01 -7.38998175e-01 3.28954719e-02
2.13324726e-01 -5.61241210e-02 -2.66171217e-01 8.09665501e-01
-7.29325354e-01 1.31437451e-01 -9.84348953e-02 -3.02925497e-01
-1.53564203e+00 2.97112465e-01 -9.57561284e-02 -4.35142696e-01
-4.75209028e-01 5.25423706e-01 -4.04542804e-01 -7.91577697e-01
1.37114320e-02 -4.07683223e-01 -5.36971211e-01 5.41831374e-01
6.77772462e-01 1.67613000e-01 6.40233517e-01 -1.00229907e+00
-4.25730824e-01 4.03012693e-01 -3.69278967e-01 -1.20190665e-01
1.35672677e+00 -3.00187200e-01 -5.03480315e-01 6.03277683e-01
8.15968871e-01 5.93061686e-01 -4.03182626e-01 -3.35099906e-01
6.42378747e-01 -1.47587150e-01 -2.90067226e-01 -4.09725219e-01
-7.88323760e-01 5.98287582e-01 1.60336584e-01 6.72197402e-01
8.67343307e-01 -5.70258358e-03 4.86552656e-01 7.09628999e-01
6.26393557e-02 -1.43504655e+00 -6.77698374e-01 1.05355537e+00
5.01110137e-01 -1.31617665e+00 2.22078323e-01 -7.77140617e-01
-5.61589658e-01 1.12991667e+00 3.49782735e-01 2.13311221e-02
5.91349244e-01 6.05119944e-01 5.49024343e-01 -7.39152133e-02
-4.20602441e-01 -4.36440945e-01 -2.18785331e-01 2.92877823e-01
1.08605587e+00 -5.64007796e-02 -8.89762163e-01 8.75662684e-01
-6.47006989e-01 -3.35271418e-01 8.41665030e-01 1.11128390e+00
-3.20439935e-01 -1.75277376e+00 -5.27334154e-01 3.88841420e-01
-1.04704893e+00 -1.59876913e-01 -7.85851061e-01 8.47713232e-01
-4.14548349e-03 1.06081784e+00 -1.52008617e-02 -1.82102814e-01
6.78593516e-01 4.81044471e-01 2.01174140e-01 -1.34277356e+00
-1.25177574e+00 2.14961618e-02 4.05827582e-01 -3.25577438e-01
-6.22475445e-01 -9.90017176e-01 -1.53303707e+00 -4.36724275e-01
-5.32518864e-01 2.24502891e-01 6.57609046e-01 1.34723330e+00
3.56834158e-02 3.45132113e-01 -1.02867782e-01 -5.82504034e-01
-3.39470416e-01 -1.13594043e+00 -1.98926702e-01 7.26387262e-01
-1.63022846e-01 -3.68429631e-01 -3.36688519e-01 3.78339022e-01] | [10.108417510986328, 9.939483642578125] |
b063a9df-9590-4ee5-98ed-b159cea73229 | linking-generative-semi-supervised-learning | 2303.11702 | null | https://arxiv.org/abs/2303.11702v3 | https://arxiv.org/pdf/2303.11702v3.pdf | On the link between generative semi-supervised learning and generative open-set recognition | This study investigates the relationship between semi-supervised learning (SSL) and open-set recognition (OSR) under the context of generative adversarial networks (GANs). Although no previous study has formally linked SSL and OSR, their respective methods share striking similarities. Specifically, SSL-GANs and OSR-GANs require their generators to produce samples in the complementary space, which are then used to regularise their respective classifier networks. In turn, classifiers trained under SSL and OSR generalise the open space by tightening classification boundaries around the labelled categories. In other words, a classifier trained using an SSL-GAN intrinsically achieves OSR and vice-versa. To prove this SSL-OSR link, we theoretically and experimentally compare the state-of-the-art SSL-GAN with the state-of-the-art OSR-GAN. Our results find that all SSL-GANs and OSR-GANs work towards the same goal, and that the SSL-optimised Margin-GANs set the new state-of-the-art for the combined task of SSL-OSR. Future studies could further explore the theoretical similarities between SSL-GANs and OSR-GANs, as well as extend SSL-OSR to other learning policies. | ['Johan du Preez', 'Emile Reyn Engelbrecht'] | 2023-03-21 | null | null | null | null | ['open-set-learning'] | ['miscellaneous'] | [ 6.39315367e-01 7.86561549e-01 -3.71799678e-01 -3.05161923e-01
-7.79791355e-01 -9.07852113e-01 8.85447800e-01 -5.19399762e-01
1.63815081e-01 9.37950492e-01 1.41938537e-01 -4.59445328e-01
1.56245321e-01 -9.38264191e-01 -9.14171576e-01 -7.19622135e-01
3.58763039e-01 4.50694114e-01 -2.13698104e-01 -3.37296098e-01
-3.84609073e-01 5.04706681e-01 -1.22571170e+00 1.17677070e-01
8.66160393e-01 9.37541425e-01 -3.85354102e-01 7.96571732e-01
2.93953538e-01 8.06489408e-01 -6.99064851e-01 -7.14530528e-01
6.79066598e-01 -8.12257528e-01 -7.58917272e-01 -7.38430396e-02
3.98452312e-01 2.29777545e-01 -6.74619615e-01 9.73508716e-01
3.48722637e-01 2.55886227e-01 1.01329088e+00 -1.64746964e+00
-1.54642844e+00 5.37600219e-01 -3.56095523e-01 -1.11657031e-01
3.94251734e-01 4.23610002e-01 9.89747107e-01 -2.52514333e-01
4.77800071e-01 1.25646520e+00 8.79221916e-01 1.31786168e+00
-1.51500225e+00 -6.54143155e-01 1.01268716e-01 -5.08770943e-01
-1.15060806e+00 -2.11559176e-01 6.95654035e-01 -2.59597778e-01
6.74956441e-01 5.74349582e-01 6.22754991e-01 1.58953094e+00
6.72874600e-02 8.52272928e-01 1.66113508e+00 -7.15141892e-01
2.52032220e-01 3.07387471e-01 -7.95905143e-02 4.88949955e-01
2.12357610e-01 8.33537459e-01 -1.83718085e-01 9.94738042e-02
1.20602310e+00 1.05501479e-02 -2.29403913e-01 -1.03647120e-01
-9.10005748e-01 9.87077773e-01 6.86034024e-01 1.31370232e-01
-1.47072405e-01 3.16132605e-01 2.42248878e-01 6.41443789e-01
6.78581417e-01 7.89202332e-01 -3.97498548e-01 7.09387287e-02
-9.05488670e-01 -1.14151381e-01 7.44102061e-01 1.09657979e+00
5.81770480e-01 3.81424636e-01 -2.97441870e-01 6.75604880e-01
3.12281549e-01 3.54712188e-01 7.16164827e-01 -7.96299815e-01
2.76936829e-01 3.13712806e-01 -5.17740250e-01 -1.97681308e-01
9.58168730e-02 -6.68118060e-01 -1.12969995e+00 2.54564285e-01
4.34287041e-01 -2.00110108e-01 -1.26691520e+00 1.92400205e+00
-1.46752447e-01 3.92278969e-01 6.09694958e-01 5.63718081e-01
6.36361361e-01 7.89307237e-01 -1.43100619e-01 -8.55844840e-02
8.93920124e-01 -1.16468811e+00 -4.44386482e-01 -5.33562958e-01
5.87453127e-01 -2.55192697e-01 1.22235131e+00 2.08569001e-02
-8.15443754e-01 -6.86747611e-01 -1.09608567e+00 2.50482410e-01
-6.05771124e-01 -9.27929655e-02 5.72800875e-01 1.04774916e+00
-1.21949184e+00 4.95859623e-01 -6.54981315e-01 -3.81239325e-01
8.52265298e-01 1.26581132e-01 -2.79193133e-01 -3.19263414e-02
-1.42402720e+00 9.59403813e-01 2.86962390e-01 5.38445860e-02
-1.22116864e+00 -7.78025031e-01 -9.27513957e-01 -2.37081558e-01
2.18744248e-01 -8.14826965e-01 1.20461082e+00 -1.35425687e+00
-1.81543350e+00 1.17368770e+00 1.34394377e-01 -7.72018492e-01
6.67077899e-01 -1.46257207e-01 -4.42780316e-01 -4.01243508e-01
-6.54582772e-03 6.59894645e-01 9.98339653e-01 -1.44264793e+00
-1.92489848e-01 -2.11685583e-01 7.10456073e-02 1.16502210e-01
-3.60980332e-02 -1.40839100e-01 1.20901801e-01 -1.12923431e+00
-2.88513899e-01 -1.13405263e+00 -3.14000875e-01 -3.92306596e-01
-6.76979065e-01 -2.93245524e-01 7.37001657e-01 -3.12343091e-01
1.09936535e+00 -2.05897117e+00 1.64632052e-01 1.54034957e-01
7.97270611e-02 5.40844142e-01 -3.36889684e-01 1.41084105e-01
-5.08454740e-01 4.18525547e-01 -3.87774289e-01 -2.91196555e-01
1.73239768e-01 5.52913368e-01 -8.27975690e-01 5.12467444e-01
3.23462456e-01 1.50656331e+00 -9.00066912e-01 -8.12422261e-02
7.90124387e-02 3.13757837e-01 -3.50158572e-01 4.47050512e-01
-3.93406063e-01 6.39894485e-01 -9.98546276e-03 5.22061884e-01
2.08156511e-01 1.72672421e-02 -1.86875880e-01 3.08408350e-01
3.96784902e-01 1.88132942e-01 -7.40689337e-01 1.40766001e+00
-8.14911127e-01 9.15368915e-01 -4.60216075e-01 -9.99612868e-01
1.21111965e+00 3.35790515e-01 -2.09045947e-01 -4.51745182e-01
-3.25716287e-02 3.73162061e-01 -2.03723356e-01 1.56439647e-01
1.73410326e-01 -5.28140008e-01 -2.72502273e-01 5.47224224e-01
3.33780378e-01 -3.05800885e-01 -1.88036859e-01 -8.30499008e-02
9.31126654e-01 4.88019794e-01 5.08481622e-01 -4.98201437e-02
4.05862361e-01 -5.38957298e-01 3.71822089e-01 1.04007888e+00
-1.13239497e-01 7.47824192e-01 3.49736631e-01 -2.30309099e-01
-1.02831960e+00 -1.42504430e+00 -2.19789237e-01 9.81393874e-01
-2.78027445e-01 6.29981002e-03 -1.02791095e+00 -1.20458519e+00
-4.32183556e-02 1.16709900e+00 -1.09038746e+00 -5.14927089e-01
-2.86009550e-01 -5.49455822e-01 9.85333562e-01 8.73196006e-01
4.50496644e-01 -1.14294529e+00 -2.22178131e-01 -1.18033074e-01
3.41385990e-01 -8.88126254e-01 -5.00666916e-01 4.29216683e-01
-6.81635916e-01 -7.98718870e-01 -9.10170972e-01 -7.69999683e-01
9.03214335e-01 -3.34508270e-01 1.22184646e+00 -2.57726550e-01
1.69271588e-01 3.86782527e-01 -5.14494061e-01 -5.48753679e-01
-1.11675954e+00 3.82687002e-01 6.14444204e-02 1.73354983e-01
1.89395279e-01 -6.77470922e-01 -8.04611593e-02 6.10480428e-01
-9.26199973e-01 3.73007841e-02 5.64714313e-01 1.09475827e+00
4.92258817e-01 9.67839360e-03 9.07005727e-01 -1.26088440e+00
5.21837354e-01 -2.98774302e-01 -4.76066828e-01 3.93847078e-01
-6.60375118e-01 2.57292986e-01 1.01549959e+00 -6.77988172e-01
-9.58795011e-01 -2.38277614e-01 -3.01199764e-01 -8.27071249e-01
-5.35059750e-01 2.59067193e-02 -4.56219584e-01 -7.22415652e-03
9.28382874e-01 3.71886551e-01 -7.91910570e-03 -2.06559873e-03
9.21088517e-01 1.00299716e+00 7.53593564e-01 -5.76121211e-01
1.26338995e+00 3.32322389e-01 3.99388000e-02 -6.57422721e-01
-1.36745965e+00 8.86658505e-02 -4.83102441e-01 -2.81730518e-02
9.18324351e-01 -8.19462299e-01 -1.58263817e-01 6.76218748e-01
-8.06852102e-01 -8.49265337e-01 -1.25245357e+00 -6.28229380e-02
-1.03072882e+00 -1.95315853e-01 -4.92337972e-01 -9.20426011e-01
-2.30936348e-01 -1.02458000e+00 1.08572865e+00 3.96124661e-01
-3.59742492e-01 -1.55409515e+00 1.63640976e-01 3.83371532e-01
4.00302887e-01 4.63226140e-01 6.77796781e-01 -1.08687520e+00
-2.15886012e-01 -4.10448372e-01 1.03728816e-01 8.70740831e-01
1.61708027e-01 -3.39965492e-01 -1.27580798e+00 -4.26388681e-01
-1.53421815e-02 -6.11279964e-01 8.69305193e-01 3.56142938e-01
1.13618994e+00 -5.34638047e-01 -1.59223124e-01 9.76640761e-01
1.23517549e+00 2.61554867e-01 1.06702673e+00 2.85246432e-01
9.02829111e-01 2.55261213e-01 3.78421724e-01 -3.10028940e-01
-8.98871645e-02 4.47681725e-01 2.58156925e-01 -4.41364199e-01
-4.06253219e-01 -8.58864367e-01 6.57265246e-01 5.05647063e-01
-8.82990845e-03 -3.18418741e-01 -5.31551480e-01 3.09847534e-01
-1.52809775e+00 -8.38535309e-01 3.38329107e-01 2.13267875e+00
1.02406955e+00 2.39112988e-01 1.02355845e-01 2.77521968e-01
8.09126496e-01 5.27670920e-01 -4.86240834e-01 -6.68376744e-01
-2.93765217e-01 7.05133915e-01 6.32939935e-01 4.71553713e-01
-1.07209682e+00 9.70229805e-01 6.44437456e+00 1.11406910e+00
-1.00604892e+00 2.71781236e-01 8.97127509e-01 9.48161483e-02
-4.05376643e-01 1.01660945e-01 -8.63756239e-01 3.64696383e-01
1.07818913e+00 1.59016967e-01 8.77797544e-01 9.07490015e-01
-4.41760063e-01 3.38797718e-01 -1.55347347e+00 8.59053969e-01
2.22798914e-01 -1.38209522e+00 1.19861573e-01 1.18147209e-01
1.22262788e+00 -1.20357320e-01 4.06071424e-01 6.72858238e-01
8.00180316e-01 -1.48228061e+00 7.66993880e-01 3.61426383e-01
1.50861049e+00 -6.38598144e-01 6.96576715e-01 3.36030871e-01
-7.55452335e-01 -1.73599586e-01 -5.78973480e-02 7.70751461e-02
-3.94822843e-02 4.32931632e-01 -5.28434992e-01 6.35926545e-01
2.19506070e-01 9.28672314e-01 -6.68662488e-01 7.43266121e-02
-8.49023402e-01 9.04564679e-01 -1.30017370e-01 2.11143941e-01
2.36861616e-01 -1.95677921e-01 5.17773330e-01 9.86977816e-01
-6.86375573e-02 -2.00776264e-01 -4.35411604e-03 1.19040060e+00
-3.31245214e-01 -4.99053389e-01 -8.70117188e-01 -4.21728224e-01
3.10943633e-01 7.62929559e-01 -4.31731790e-01 -2.75878519e-01
-3.01765591e-01 9.37865973e-01 3.77048016e-01 3.41802090e-01
-8.65312338e-01 -1.32209942e-01 6.05326474e-01 2.17347145e-01
1.19231112e-01 2.57463604e-01 -5.00525832e-01 -1.21285486e+00
-3.09765667e-01 -1.14695501e+00 2.89071321e-01 -8.65707219e-01
-1.85267746e+00 7.14501143e-01 -2.20276758e-01 -1.12357116e+00
-2.86531776e-01 -6.34961843e-01 -6.60947025e-01 8.34699690e-01
-1.46792865e+00 -1.47872663e+00 3.59835662e-02 4.64803904e-01
6.17963016e-01 -5.09477973e-01 1.11701393e+00 -3.28380376e-01
-7.57857800e-01 1.05397391e+00 3.54944706e-01 4.02671874e-01
4.12061542e-01 -1.39260185e+00 8.77805054e-01 1.07678449e+00
5.25904536e-01 4.71254051e-01 3.71342093e-01 -6.17380440e-01
-1.02616954e+00 -1.55627429e+00 2.58194029e-01 -9.01253998e-01
5.22057950e-01 -5.04775703e-01 -7.02712655e-01 1.27414453e+00
2.61509538e-01 4.68640089e-01 7.61733830e-01 2.66874563e-02
-6.07328057e-01 1.44210430e-02 -1.19613957e+00 5.22030354e-01
1.15455461e+00 -9.30880129e-01 -8.48249972e-01 1.58373594e-01
8.82668793e-01 -5.51773012e-01 -9.52468276e-01 3.14242691e-01
3.47599447e-01 -9.07914519e-01 8.78454030e-01 -9.02745247e-01
7.61444747e-01 5.70183713e-03 -3.22664768e-01 -1.88876104e+00
-1.45368442e-01 -1.09525824e+00 -2.73523033e-01 1.50790906e+00
4.33748364e-01 -1.28207159e+00 5.79401910e-01 1.28738895e-01
-2.58529991e-01 -1.04350269e+00 -9.06922877e-01 -1.24019909e+00
4.21585262e-01 -4.71936077e-01 6.60629749e-01 9.77363765e-01
-4.95264620e-01 3.08006138e-01 -2.65425354e-01 8.09863023e-03
5.73820889e-01 -9.76075232e-02 8.66484821e-01 -1.04771554e+00
-6.19660795e-01 -3.59863669e-01 -4.73737776e-01 -8.08733165e-01
6.90147996e-01 -1.31398451e+00 -4.10879515e-02 -1.22098589e+00
-1.98097154e-01 -9.11045015e-01 -1.77758455e-01 6.22569025e-01
-1.93425640e-01 4.31784898e-01 3.50271761e-01 1.05310999e-01
-3.00344020e-01 4.91628677e-01 1.30570436e+00 -1.80990659e-02
-2.81065762e-01 1.90451548e-01 -1.08278513e+00 7.08629131e-01
7.18775511e-01 -4.57759708e-01 -5.43599308e-01 7.53865093e-02
2.71847248e-02 -2.52927721e-01 5.91505408e-01 -8.68226588e-01
-2.90996790e-01 -1.23729199e-01 4.58573222e-01 9.13873911e-02
6.23470917e-02 -6.71603918e-01 2.39715457e-01 3.83550406e-01
-7.40271926e-01 -5.28858304e-01 -9.26829316e-03 6.64725780e-01
-3.28950770e-02 -2.03497216e-01 1.05007768e+00 -9.03162733e-02
-2.37283424e-01 2.10923582e-01 -2.20836028e-01 7.33098805e-01
1.32467031e+00 -3.84177864e-01 -4.42401379e-01 -4.10245180e-01
-6.46454453e-01 -2.01799855e-01 7.65170932e-01 4.64182913e-01
3.99097890e-01 -1.45097554e+00 -6.16445899e-01 7.29905546e-01
4.24888358e-02 4.79940504e-01 -1.97412491e-01 3.52287441e-01
-1.53900936e-01 4.91662115e-01 -1.17902808e-01 -4.19723809e-01
-6.68605030e-01 7.08736122e-01 5.41963518e-01 -7.39426434e-01
-4.70042914e-01 1.11357212e+00 3.48733813e-01 -6.57333612e-01
9.44326594e-02 2.98578553e-02 2.83751935e-01 -2.73556322e-01
1.19916111e-01 2.56184906e-01 -2.32632056e-01 -4.32170331e-01
-3.06591727e-02 3.33118618e-01 1.90394163e-01 5.15920855e-02
1.23630559e+00 2.27775887e-01 1.01309150e-01 5.48566103e-01
1.16920114e+00 -1.19720124e-01 -1.36030650e+00 -1.45860523e-01
-4.86516267e-01 -2.95584798e-01 -1.69967845e-01 -8.06407690e-01
-1.12497342e+00 7.34926879e-01 3.25732678e-01 3.73074055e-01
1.26537895e+00 3.73838991e-01 6.33019626e-01 -2.68680215e-01
2.87627429e-01 -6.82745993e-01 1.23673797e-01 3.54398608e-01
9.57917631e-01 -1.03081810e+00 -2.49437273e-01 -4.78113592e-01
-8.44914794e-01 7.13166654e-01 6.61371052e-01 -4.63721991e-01
3.89230758e-01 3.44547719e-01 2.25853305e-02 9.25711915e-02
-6.45081580e-01 -5.59977628e-02 3.41224879e-01 1.09866953e+00
1.40166283e-01 2.95576602e-01 4.04113472e-01 6.68013930e-01
-6.69142008e-01 -4.74887341e-02 4.39799488e-01 8.25061977e-01
3.53001624e-01 -1.09954929e+00 -4.96172100e-01 6.40961170e-01
-2.20609263e-01 -5.83155453e-02 -4.70991075e-01 6.48959875e-01
1.03913717e-01 7.22037375e-01 2.44346648e-01 -4.37741548e-01
2.27548599e-01 2.84689546e-01 6.90994978e-01 -8.25179517e-01
-7.44622350e-01 -4.35236275e-01 1.70172309e-03 -2.29000390e-01
-3.67460012e-01 -6.04304075e-01 -7.97206700e-01 9.44169536e-02
-5.99507391e-01 5.23691848e-02 3.74238461e-01 9.69369650e-01
1.22556500e-01 8.62767398e-01 8.99532676e-01 -7.02545464e-01
-9.87989485e-01 -9.94255543e-01 -5.86689532e-01 5.12563169e-01
3.30423027e-01 -4.67020750e-01 -7.51919925e-01 -8.85504261e-02] | [11.609551429748535, -0.12981708347797394] |
52955f74-f74b-4a97-8e90-1529bfc8b1ef | smm4h-shared-task-2020-a-hybrid-pipeline-for | null | null | https://aclanthology.org/2020.smm4h-1.9 | https://aclanthology.org/2020.smm4h-1.9.pdf | SMM4H Shared Task 2020 - A Hybrid Pipeline for Identifying Prescription Drug Abuse from Twitter: Machine Learning, Deep Learning, and Post-Processing | This paper presents our approach to multi-class text categorization of tweets mentioning prescription medications as being indicative of potential abuse/misuse (A), consumption/non-abuse (C), mention-only (M), or an unrelated reference (U) using natural language processing techniques. Data augmentation increased our training and validation corpora from 13,172 tweets to 28,094 tweets. We also created word-embeddings on domain-specific social media and medical corpora. Our hybrid pipeline of an attention-based CNN with post-processing was the best performing system in task 4 of SMM4H 2020, with an F1 score of 0.51 for class A. | ['Yindalon Aphinyanaphongs', 'William McMahon', 'Mark T. Rutledge', 'Rajat S. Chandra', 'Whitley M. Yi', 'Allison Black', 'Emir Y. Haskovic', 'Isabel Metzger'] | null | null | null | null | smm4h-coling-2020-12 | ['text-categorization'] | ['natural-language-processing'] | [ 2.65708059e-01 2.58699596e-01 -7.00383544e-01 -3.93217921e-01
-8.70734453e-01 -4.02838886e-01 9.75666583e-01 1.41257632e+00
-8.42902780e-01 8.31603229e-01 6.34593964e-01 -8.41084838e-01
9.13863778e-02 -7.13732064e-01 -3.24146807e-01 -2.46552154e-01
-1.69250876e-01 4.44557309e-01 -4.27756548e-01 -3.40876691e-02
2.91990042e-01 3.49877417e-01 -8.40301275e-01 4.56638217e-01
5.88501990e-01 1.04097950e+00 -5.78392506e-01 7.37864375e-01
-3.19561839e-01 8.84636879e-01 -7.04076290e-01 -6.86891317e-01
-9.89808608e-03 -1.43818006e-01 -8.83199632e-01 -1.72241792e-01
5.02472937e-01 -2.40256563e-01 -3.72149050e-01 9.25625443e-01
6.05791330e-01 1.11265620e-02 8.11678350e-01 -9.76520419e-01
-9.49112594e-01 8.67520809e-01 -6.15659952e-01 6.52508080e-01
4.61259842e-01 2.65190750e-01 6.95221722e-01 -6.79408848e-01
7.19378948e-01 1.13954687e+00 7.67961025e-01 5.91318071e-01
-1.12179410e+00 -1.14345586e+00 -2.53953516e-01 -7.35413954e-02
-1.38410234e+00 -4.11152899e-01 -6.43735155e-02 -7.96018064e-01
1.57662976e+00 3.59596312e-01 3.75053465e-01 1.43210220e+00
4.52372015e-01 3.12049329e-01 7.96744406e-01 3.38219777e-02
-4.24572686e-03 2.53871500e-01 6.68070316e-01 2.67610669e-01
4.39341366e-01 -4.20657516e-01 -6.20626472e-02 -8.81160557e-01
1.99556842e-01 9.28781852e-02 6.51870370e-02 7.54574716e-01
-1.00851011e+00 1.20641506e+00 3.22154403e-01 4.17082220e-01
-9.45570111e-01 -1.31324574e-01 7.54919469e-01 3.48469801e-02
8.97437871e-01 8.92129064e-01 -6.53399527e-01 2.43823826e-02
-8.20448458e-01 2.31998160e-01 5.84715307e-01 7.21782923e-01
2.46443391e-01 -1.29568994e-01 -4.61689919e-01 9.38538909e-01
1.33874044e-01 2.17201367e-01 1.06831086e+00 -2.88407862e-01
5.53008914e-01 6.77626252e-01 4.93137687e-02 -9.55782950e-01
-9.81354654e-01 -5.49375236e-01 -8.51143956e-01 -5.60664117e-01
6.83068261e-02 -6.28518999e-01 -1.08933914e+00 1.39117718e+00
5.62140644e-02 1.91438869e-01 1.86310843e-01 2.77616739e-01
1.30561042e+00 6.22987092e-01 9.75603223e-01 -2.69314289e-01
1.68645072e+00 -6.20540619e-01 -8.76774609e-01 -1.24767847e-01
9.04601812e-01 -8.44228923e-01 3.18730623e-01 2.51864418e-02
-8.82658601e-01 -2.44688824e-01 -7.89085329e-01 -1.40138790e-01
-8.86303246e-01 -4.81755823e-01 3.68101180e-01 7.46627152e-01
-7.50659168e-01 6.30200922e-01 -2.29308009e-01 -4.55342770e-01
9.34793055e-01 4.82661009e-01 -5.48448563e-01 1.76653951e-01
-1.46098363e+00 1.11984539e+00 5.12795746e-01 -5.89308023e-01
-3.86939168e-01 -1.12501144e+00 -9.63244021e-01 -1.27948761e-01
-3.22001100e-01 -6.66532934e-01 1.25659001e+00 -6.65670693e-01
-7.67612219e-01 1.06668317e+00 9.45119932e-03 -9.62137818e-01
2.57371008e-01 -6.47700727e-02 -9.75150645e-01 9.29189697e-02
3.26924503e-01 6.60802186e-01 3.78550380e-01 -5.81745684e-01
-8.15407217e-01 -3.76592696e-01 -1.10240571e-01 -1.42132714e-01
-6.23265564e-01 3.77979487e-01 2.42594421e-01 -5.82899570e-01
-5.04550040e-01 -8.41098309e-01 -4.48958576e-01 -3.61797422e-01
-5.78504205e-01 -6.58821166e-01 4.28426683e-01 -7.79304981e-01
1.59127855e+00 -1.97057664e+00 -6.20289147e-01 1.71113446e-01
4.61469084e-01 5.15054703e-01 -1.71786919e-01 5.54124773e-01
-5.63443244e-01 7.64366210e-01 -2.01274231e-01 -3.19778830e-01
-4.21309710e-01 3.54615152e-02 -1.31662190e-01 7.21634924e-01
5.01658201e-01 9.00556326e-01 -1.32966423e+00 -4.27841693e-01
2.93027610e-01 5.64464688e-01 -3.73042852e-01 -4.45149019e-02
1.28232595e-02 2.65090436e-01 -4.25316960e-01 5.50746620e-01
5.77035844e-01 -2.52062827e-01 -3.28467274e-03 -7.37177208e-02
-3.14477503e-01 5.18082976e-01 -4.00924802e-01 1.19522536e+00
-6.76317573e-01 8.95028055e-01 -2.06916273e-01 -4.46339399e-01
5.75549304e-01 5.42595029e-01 8.18505704e-01 -5.13641357e-01
5.97100556e-01 2.11298212e-01 3.52017075e-01 -7.75453568e-01
4.97806340e-01 -1.62761271e-01 3.81275713e-02 1.77053943e-01
-6.84903413e-02 4.27100927e-01 6.47360235e-02 1.69913277e-01
1.38052404e+00 -6.73538983e-01 7.16677308e-01 -4.20667380e-01
3.62760067e-01 1.68917608e-02 2.49123842e-01 6.53943479e-01
-1.80182829e-01 5.24962902e-01 4.24603313e-01 -4.01923537e-01
-8.93534839e-01 -4.25817728e-01 -9.16487098e-01 8.52244139e-01
-5.73810995e-01 -5.79600155e-01 -4.29145724e-01 -5.15887022e-01
3.34516168e-01 1.19779062e+00 -1.04650199e+00 -1.03578590e-01
-3.33350062e-01 -1.16365731e+00 7.74116158e-01 3.50229919e-01
1.01873964e-01 -1.01783824e+00 -2.32978001e-01 5.70104003e-01
-6.68319464e-02 -1.01774037e+00 -5.21473825e-01 3.16779256e-01
-5.25513470e-01 -1.45770311e+00 -7.75608242e-01 -5.03298879e-01
3.33492696e-01 -2.20825195e-01 1.03921616e+00 -1.29866078e-02
-2.19732553e-01 1.37943119e-01 -2.29210988e-01 -7.66114831e-01
-4.22758222e-01 1.31551772e-01 1.42313808e-01 1.02010235e-01
8.99211287e-01 -8.13096911e-02 -6.55229747e-01 -3.75401974e-01
-9.61420894e-01 -3.97247374e-01 1.81576505e-01 3.83986950e-01
4.60312627e-02 -5.53732038e-01 7.64180660e-01 -1.32263267e+00
9.21444952e-01 -1.46119165e+00 3.39702442e-02 -3.54321271e-01
-7.31391490e-01 -5.85428715e-01 5.58131337e-01 -6.52378082e-01
-3.57660115e-01 9.87268016e-02 -7.76237786e-01 -3.04689914e-01
-3.71355474e-01 7.51458585e-01 5.19742966e-01 3.46559048e-01
1.03819823e+00 -1.37930661e-01 -3.28414649e-01 -4.77639079e-01
1.79045051e-01 1.43803191e+00 2.47042611e-01 2.03512862e-01
2.97944069e-01 5.31828701e-02 -2.67749220e-01 -1.17640364e+00
-9.63913321e-01 -9.43885505e-01 -4.17006671e-01 4.32382166e-01
1.33027387e+00 -8.51189137e-01 -8.58208716e-01 3.73173118e-01
-1.26015794e+00 -5.58900572e-02 -4.66541573e-02 5.79305887e-01
2.71717280e-01 1.82185173e-01 -7.58941710e-01 -7.56332934e-01
-7.46539950e-01 -8.84784460e-01 6.31072640e-01 -1.73248760e-02
-1.08852041e+00 -1.05193424e+00 1.82486102e-01 3.23539168e-01
5.51293373e-01 5.67490935e-01 1.21247423e+00 -1.67668653e+00
5.78284323e-01 -5.60909510e-01 -3.78731847e-01 -2.05135029e-02
4.78881061e-01 -2.26420108e-02 -1.00744712e+00 2.73871627e-02
-1.61258250e-01 -1.50349170e-01 7.95081258e-01 4.59140182e-01
1.45071197e+00 -7.57208169e-01 -4.06295210e-01 1.40845925e-01
1.37509644e+00 5.85521460e-01 7.54857421e-01 4.58539367e-01
6.78202391e-01 3.99279743e-01 1.18080392e-01 6.37204409e-01
1.98660359e-01 4.27381516e-01 5.26330352e-01 -1.48741439e-01
2.68663883e-01 -4.27335314e-02 -5.02732582e-02 3.23800057e-01
3.25887382e-01 -4.22403008e-01 -1.28726625e+00 8.72304857e-01
-1.37382638e+00 -7.92128325e-01 -7.58430660e-01 1.96738541e+00
1.10100842e+00 9.13635045e-02 1.34058595e-01 -2.32626632e-01
8.40190530e-01 3.03456448e-02 -4.10292089e-01 -7.50465214e-01
4.08592261e-03 6.62308156e-01 9.11821067e-01 3.12464893e-01
-1.54440439e+00 4.09550428e-01 6.41925812e+00 7.53393173e-01
-1.11695397e+00 2.38006532e-01 9.50446367e-01 -6.17486313e-02
-1.59333110e-01 -7.08140016e-01 -7.41016150e-01 7.57099569e-01
1.73561072e+00 -2.92328507e-01 -1.20400503e-01 6.91504717e-01
8.05064142e-02 2.50351787e-01 -9.97921467e-01 9.87633228e-01
1.21214665e-01 -1.63051379e+00 2.35520437e-01 1.76420316e-01
6.51253819e-01 5.10186195e-01 6.14976659e-02 3.83541465e-01
2.10265845e-01 -1.22447038e+00 3.09989035e-01 1.94500595e-01
9.12368357e-01 -5.79501629e-01 1.08791447e+00 1.17374212e-01
-7.36223936e-01 -1.82281047e-01 -1.47098094e-01 -5.51645383e-02
5.01273423e-02 4.65978205e-01 -1.05599332e+00 3.16805303e-01
4.47256327e-01 1.05787289e+00 -5.66637039e-01 1.02228940e+00
2.58589089e-01 6.30235255e-01 -1.37463644e-01 -2.34578848e-01
7.04718292e-01 2.02247992e-01 3.59968394e-01 1.86918652e+00
1.36384023e-02 4.26456928e-01 1.39198927e-02 2.48401448e-01
-5.60608327e-01 4.87559408e-01 -6.91758275e-01 -5.26917517e-01
2.15835512e-01 1.28611231e+00 -5.33992767e-01 -9.17441010e-01
-4.85027999e-01 4.85264570e-01 -1.47685781e-01 -1.98802594e-02
-7.72818685e-01 -6.25456870e-01 8.24852407e-01 4.26743299e-01
-2.73644507e-01 2.24767715e-01 -3.51422429e-01 -7.74054885e-01
-6.86432242e-01 -7.30690420e-01 7.22259283e-01 -4.89470869e-01
-1.74522471e+00 7.55717456e-01 -3.08680415e-01 -1.10278773e+00
-3.63373123e-02 -5.34883320e-01 -4.05028194e-01 1.02868474e+00
-1.36606145e+00 -6.70786679e-01 9.62834731e-02 1.32336900e-01
7.22633600e-01 -2.88388371e-01 1.28919351e+00 9.65337574e-01
-5.77402353e-01 6.05767488e-01 -1.71117727e-02 5.12653053e-01
7.24672735e-01 -1.19547820e+00 4.77244020e-01 4.33825888e-02
-3.56925189e-01 9.95375574e-01 5.59281290e-01 -7.13462114e-01
-6.93753004e-01 -1.74176133e+00 1.57591307e+00 -5.18119872e-01
1.11880505e+00 -4.89843376e-02 -1.05703592e+00 5.92859745e-01
6.54353261e-01 -5.62746346e-01 1.41763330e+00 1.04242481e-01
-3.17382187e-01 5.77466071e-01 -1.43893147e+00 5.91192365e-01
5.18819869e-01 -4.55302149e-01 -5.74757814e-01 1.12063086e+00
7.86222816e-01 -3.53525132e-01 -1.39272177e+00 1.14805244e-01
4.93123025e-01 -1.40153825e-01 9.15778875e-01 -1.42846060e+00
9.70994353e-01 2.75399446e-01 4.48491499e-02 -1.02556419e+00
-4.55799788e-01 -5.20610332e-01 1.61196321e-01 1.01181757e+00
8.81991327e-01 -9.09253657e-01 2.52133787e-01 9.96519446e-01
-2.30224445e-01 -9.00449336e-01 -8.47500741e-01 -3.43171567e-01
5.28551102e-01 -4.89337385e-01 5.79573214e-01 1.75314736e+00
2.32585296e-01 5.92905998e-01 -2.16130927e-01 -1.11393906e-01
-1.75994746e-02 -3.43677849e-01 1.93365917e-01 -1.19469750e+00
2.43992046e-01 -6.69948399e-01 -4.73538041e-01 -4.66359377e-01
2.72028625e-01 -1.19307196e+00 -4.08632666e-01 -1.91507757e+00
4.09578323e-01 -1.76356450e-01 -5.66234112e-01 8.84724200e-01
-7.29174092e-02 5.08852661e-01 -2.57341474e-01 -7.83502404e-03
-1.57148749e-01 -5.75057939e-02 5.85858107e-01 -5.00675738e-01
-2.09661469e-01 1.65285990e-01 -9.45703447e-01 6.14611983e-01
9.03159022e-01 -7.96496451e-01 1.86943755e-01 -1.73821300e-01
1.08194135e-01 -8.14796537e-02 -4.65908460e-02 -3.96139711e-01
-9.35249031e-02 -2.57218312e-02 4.11220044e-01 -5.36961377e-01
2.00400680e-01 -7.56639481e-01 1.83337666e-02 9.22002792e-01
-8.87919903e-01 1.68718144e-01 6.72895789e-01 2.72641897e-01
2.08289608e-01 -3.89211893e-01 7.06610203e-01 -1.14862852e-01
-1.75403744e-01 3.77923012e-01 -8.97966623e-01 1.00356929e-01
6.74724758e-01 1.00896344e-01 -5.44465184e-01 -3.15604061e-01
-8.15379739e-01 3.39625984e-01 -2.07794204e-01 6.37535512e-01
3.51336837e-01 -1.16596210e+00 -9.46694672e-01 -2.19548106e-01
4.21746373e-01 -5.66546798e-01 6.80794269e-02 7.32646227e-01
-4.76288944e-01 6.58814847e-01 -5.85247390e-02 -1.56587601e-01
-1.34054255e+00 5.24457037e-01 1.19335704e-01 -1.10414967e-01
-3.62200648e-01 6.89804852e-01 -2.53967762e-01 -3.38396937e-01
9.78047997e-02 -3.38444352e-01 -7.37966299e-01 6.18987620e-01
7.23699212e-01 3.59632105e-01 1.87544614e-01 -1.03422236e+00
-6.04324520e-01 2.05680564e-01 -9.70900357e-02 3.09426606e-01
1.37046349e+00 3.97946566e-01 -1.96241960e-01 4.05091166e-01
2.01085401e+00 -2.26727411e-01 -2.33148932e-02 -4.66119237e-02
1.41347334e-01 -2.45178968e-01 3.72002900e-01 -7.79525101e-01
-8.07650268e-01 6.18781328e-01 6.68855548e-01 4.93800819e-01
5.65047085e-01 -1.54555991e-01 9.98210490e-01 2.63004780e-01
-3.01569253e-01 -7.91202068e-01 -5.05530715e-01 5.32768607e-01
8.35701287e-01 -1.33133042e+00 9.94908661e-02 -7.71827772e-02
-6.75972044e-01 1.06834173e+00 1.30840719e-01 5.08525707e-02
1.01564527e+00 -2.11273938e-01 2.12726280e-01 -4.58379984e-01
-5.33064187e-01 -2.04654083e-01 4.57618058e-01 2.81103551e-01
8.97859871e-01 4.19438407e-02 -6.55231118e-01 6.32939398e-01
-1.78985149e-01 -3.43849584e-02 6.28737509e-01 6.63003743e-01
-8.17125142e-02 -8.25144053e-01 1.38005421e-01 1.28290021e+00
-1.26859498e+00 -5.39373815e-01 -4.49298441e-01 8.20272386e-01
1.24316081e-01 1.18841553e+00 4.50814128e-01 -3.51682127e-01
4.23669130e-01 3.31017375e-01 -1.81423470e-01 -1.13550472e+00
-1.42500055e+00 6.34441823e-02 5.73056281e-01 -7.68120214e-02
-4.09308344e-01 -6.44559324e-01 -1.25104535e+00 -3.48967344e-01
-2.08002135e-01 2.22844109e-02 7.37273753e-01 8.09001863e-01
4.42227274e-01 6.06122553e-01 4.75030988e-01 -2.55721480e-01
-2.06652820e-01 -1.25082016e+00 -2.10663289e-01 6.25161469e-01
6.58830881e-01 -2.64272422e-01 -3.66672665e-01 -4.90332134e-02] | [8.42078971862793, 8.92742919921875] |
8701b988-7f88-4103-aab7-43df88404d9d | predicting-pulmonary-hypertension-by | 2304.12447 | null | https://arxiv.org/abs/2304.12447v1 | https://arxiv.org/pdf/2304.12447v1.pdf | Predicting Pulmonary Hypertension by Electrocardiograms Using Machine Learning | Pulmonary hypertension (PH) is a condition of high blood pressure that affects the arteries in the lungs and the right side of the heart (Mayo Clinic, 2017). A mean pulmonary artery pressure greater than 25 mmHg is defined as Pulmonary hypertension. The estimated 5-year survival rate from the time of diagnosis of pulmonary hypertension is only 57% without therapy and patients with right heart failure only survive for approximately 1 year without treatment (Benza et al., 2012). Given the indolent nature of the disease, early detection of PH remains a challenge leading to delays in therapy. Echocardiography is currently used as a screening tool for diagnosing PH. However, electrocardiography (ECG), a more accessible, simple to use, and cost-effective tool compared to echocardiography, is less studied and explored for screening at-risk patients for PH. The goal of this project is to create a neural network model which can process an ECG signal and detect the presence of PH with a confidence probability. I created a dense neural network (DNN) model that has an accuracy of 98% over the available training sample. For future steps, the current model will be updated with a model suited for time-series data. To balance the dataset with proper training samples, I will generate additional data using data augmentation techniques. Through early and accurate detection of conditions such as PH, we widen the spectrum of innovation in detecting chronic life-threatening health conditions and reduce associated mortality and morbidity. | ['Praveen Kumar Pandian Shanmuganathan', 'Eashan Kosaraju'] | 2023-04-24 | null | null | null | null | ['electrocardiography-ecg'] | ['methodology'] | [ 1.35171711e-01 1.55338496e-01 -8.97856653e-02 -3.29921097e-02
-1.28220931e-01 -2.72341251e-01 -1.91473857e-01 1.92606553e-01
-2.25334242e-01 7.34374583e-01 1.66381374e-01 -8.21654916e-01
-3.66408825e-01 -8.96834731e-01 -6.63380921e-02 -4.35212463e-01
-4.49505776e-01 8.24191391e-01 1.45108914e-02 4.69155103e-01
-1.64617166e-01 8.10073912e-01 -1.04558790e+00 -1.85421985e-02
8.16980302e-01 7.59166539e-01 1.20873414e-01 1.00791323e+00
2.68869132e-01 8.46509695e-01 -5.91657698e-01 2.14466006e-01
2.46735290e-01 -1.01107478e+00 -5.64203858e-01 -2.26393774e-01
3.44546646e-01 -6.09269500e-01 -3.57050061e-01 2.95132697e-01
1.15816748e+00 -1.56246871e-01 5.75458407e-01 -7.26387441e-01
-2.43323997e-01 3.90569508e-01 -4.79924344e-02 6.70308888e-01
-6.63072765e-02 3.12926859e-01 5.81903160e-01 -7.25993633e-01
2.28941843e-01 6.16655827e-01 9.46360886e-01 7.59466112e-01
-1.20521617e+00 -4.99676198e-01 -5.94806135e-01 8.86904299e-02
-1.02431810e+00 -7.57293701e-02 2.51541227e-01 -4.21575010e-01
8.25041473e-01 4.21593994e-01 1.32086635e+00 7.44744122e-01
3.78762811e-01 9.50021222e-02 1.07750762e+00 -2.85722166e-01
2.58979145e-02 1.00560822e-01 -2.76423454e-01 5.25467873e-01
3.17846626e-01 3.07316512e-01 -2.24180266e-01 -2.69719422e-01
1.16199088e+00 3.94992799e-01 -4.48067665e-01 -9.23894048e-02
-9.96198416e-01 7.03731537e-01 -9.71403811e-03 5.80767572e-01
-6.18845582e-01 -9.22232047e-02 3.28682542e-01 4.08543110e-01
-1.92419868e-02 7.25413978e-01 -5.41077912e-01 -4.33169544e-01
-1.17714536e+00 6.35911971e-02 1.05482209e+00 8.30474943e-02
-1.65874258e-01 3.21977884e-01 -4.43492770e-01 9.98139560e-01
2.90122628e-01 5.02412677e-01 6.17577076e-01 -1.31976044e+00
-1.05143346e-01 5.52430630e-01 -9.08185393e-02 -8.50693703e-01
-5.53240955e-01 -9.61847842e-01 -1.07836211e+00 3.37920785e-01
5.71487308e-01 -5.19542933e-01 -9.81077254e-01 1.48923171e+00
6.40350580e-02 1.27742752e-01 -1.16395749e-01 8.30031395e-01
6.21777117e-01 4.32141870e-01 1.54537752e-01 -3.81395906e-01
1.19988823e+00 -4.99594212e-01 -4.58233237e-01 -8.55958462e-02
4.60000515e-01 -3.32202524e-01 6.88430548e-01 4.48845088e-01
-1.15789211e+00 -3.58132601e-01 -9.01162386e-01 2.32345134e-01
-1.56628750e-02 6.06560335e-02 5.44296563e-01 9.24467325e-01
-9.70405161e-01 1.04049253e+00 -9.27134871e-01 -5.17550528e-01
5.52484334e-01 3.39650333e-01 -2.06175342e-01 1.25873715e-01
-1.42117894e+00 1.09846091e+00 8.99282023e-02 1.28355669e-03
-7.56243765e-01 -1.26625586e+00 -4.43060935e-01 3.51540297e-01
6.75221235e-02 -1.18954587e+00 9.42032099e-01 -4.22210187e-01
-1.09532416e+00 7.22511113e-01 -4.82725613e-02 -7.66157448e-01
5.27425587e-01 -7.27392957e-02 -3.88339132e-01 4.35686439e-01
-9.45357382e-02 2.72968888e-01 7.27380872e-01 -4.58365440e-01
-8.15935075e-01 -2.91027755e-01 -5.97201049e-01 1.74694806e-01
-8.20294842e-02 1.07989244e-01 5.21112839e-03 -3.73399884e-01
3.85354310e-02 -8.20890367e-01 -4.01494384e-01 1.18427210e-01
9.54799447e-03 7.94406310e-02 7.84485698e-01 -1.12057924e+00
1.17764223e+00 -1.95203376e+00 -3.55142593e-01 2.34319732e-01
8.27576041e-01 5.65976381e-01 5.55982471e-01 3.31905782e-01
-3.45989823e-01 4.59350288e-01 -2.67955452e-01 2.21362248e-01
-2.46254697e-01 2.89634943e-01 2.13146299e-01 3.78517359e-01
1.76231027e-01 8.08231890e-01 -5.62978923e-01 -3.44101906e-01
3.95666569e-01 6.82630122e-01 -3.13662320e-01 3.43343019e-01
3.79674375e-01 6.90095127e-01 1.62984744e-01 5.30148149e-01
2.35360920e-01 -5.36799967e-01 3.07925582e-01 3.09985816e-01
-3.98021303e-02 1.41534522e-01 -9.86090004e-01 6.54601872e-01
-1.02946475e-01 8.53667676e-01 -1.35356531e-01 -8.84039402e-01
9.46546257e-01 8.24816227e-01 6.50796056e-01 -4.73241150e-01
7.59210140e-02 2.94541806e-01 7.68211424e-01 -7.39299178e-01
-4.38848346e-01 -5.27374268e-01 6.32096708e-01 3.04709852e-01
-4.05454993e-01 1.00174561e-01 9.54579115e-02 -1.16109610e-01
1.47695804e+00 -2.43437201e-01 3.76057178e-01 -1.16375210e-02
3.64958644e-01 -2.36495301e-01 8.32955956e-01 8.86780083e-01
-5.49013615e-01 7.89054751e-01 6.46580875e-01 -6.20176792e-01
-1.22957289e+00 -1.07230973e+00 -6.09244943e-01 3.31379533e-01
-6.84390545e-01 -4.37367745e-02 -2.22434089e-01 -2.19836161e-01
4.31727022e-01 3.55247796e-01 -4.97635365e-01 -7.67806694e-02
-6.05580986e-01 -7.75101364e-01 6.49502814e-01 7.97346592e-01
5.64125597e-01 -1.00830150e+00 -1.00404215e+00 5.47948897e-01
4.33079787e-02 -5.85595608e-01 1.98623091e-01 1.75599843e-01
-1.22317219e+00 -1.26184225e+00 -1.27092350e+00 -7.61205256e-01
3.48139822e-01 -6.09090924e-01 1.01762068e+00 1.52644411e-01
-9.19776440e-01 2.95547783e-01 2.37100095e-01 -4.27725047e-01
-6.09015167e-01 -2.54566908e-01 1.01037107e-01 -4.52535957e-01
1.94868252e-01 -8.33052576e-01 -1.05837822e+00 8.56052618e-03
-2.64841676e-01 -1.46483734e-01 8.63386452e-01 6.73003554e-01
6.88532054e-01 8.42529535e-02 8.95252407e-01 -9.16738212e-01
8.52131665e-01 -5.90796590e-01 -1.45256639e-01 -3.09680998e-01
-1.17170143e+00 -6.03639364e-01 6.10233366e-01 -4.48742360e-01
-5.56083322e-01 -2.59826481e-01 -9.33531523e-02 -6.21332586e-01
-3.28225404e-01 6.84162259e-01 4.62304473e-01 1.25644445e-01
7.17782319e-01 9.73314866e-02 4.19461846e-01 -3.83301258e-01
-3.19204897e-01 5.45540094e-01 6.17722750e-01 -1.36905491e-01
4.53906983e-01 1.19198352e-01 6.66404188e-01 -9.59117413e-01
-5.29070199e-01 -3.01646233e-01 -3.36415291e-01 -2.68825710e-01
7.66385317e-01 -7.01217771e-01 -4.83279467e-01 2.51998991e-01
-4.46743459e-01 -2.93968499e-01 -6.09161198e-01 8.05849433e-01
-2.31673881e-01 4.66680199e-01 -7.54696310e-01 -8.25785279e-01
-7.68498003e-01 -5.27847886e-01 7.68892318e-02 2.67916560e-01
-6.95406556e-01 -1.19988608e+00 3.29021871e-01 3.56822968e-01
7.89828718e-01 6.20382786e-01 1.26838028e+00 -8.12286794e-01
-2.92147040e-01 -4.19400036e-01 -7.79697299e-02 9.16149676e-01
1.52145639e-01 1.28341362e-01 -4.83624279e-01 -1.49014816e-01
1.85274258e-01 1.77375332e-01 4.65091318e-01 8.12623620e-01
1.07870412e+00 -9.08053070e-02 -1.01685673e-01 5.06058991e-01
1.16669250e+00 6.51881516e-01 6.41384840e-01 1.33412600e-01
3.87860715e-01 4.11152303e-01 1.24889888e-01 3.96562219e-01
3.55502628e-02 -1.42961398e-01 5.09300865e-02 -3.05019468e-01
-2.32185706e-01 6.53970316e-02 -2.79279679e-01 5.50725162e-01
-4.24819589e-01 2.58800574e-02 -1.31615984e+00 6.62301660e-01
-1.22387731e+00 -1.12328196e+00 -5.34969926e-01 2.38219976e+00
8.54946971e-01 1.29659951e-01 2.69170165e-01 4.04867709e-01
6.07773840e-01 -4.05280977e-01 -5.50908864e-01 -4.32229012e-01
1.14946194e-01 6.78804636e-01 2.82884032e-01 2.79766470e-01
-8.39243472e-01 5.19553833e-02 7.31362486e+00 -2.52668679e-01
-1.40079951e+00 -1.56716377e-01 7.61926889e-01 -3.68467234e-02
2.22005814e-01 8.60949978e-03 -4.84426677e-01 5.76195240e-01
1.29453504e+00 -3.28944027e-01 8.24426264e-02 7.28781223e-01
5.70109665e-01 -1.42269969e-01 -9.71594036e-01 8.75871956e-01
-1.16325356e-01 -1.29840815e+00 -6.10291302e-01 5.88216111e-02
3.00606072e-01 1.67972416e-01 1.65617950e-02 4.03839260e-01
-3.96864355e-01 -1.22281694e+00 -2.17266515e-01 6.34380817e-01
1.08585954e+00 -4.38013405e-01 1.02788508e+00 2.49030650e-01
-6.60745919e-01 -1.75037310e-01 -3.99137437e-02 -3.79679859e-01
1.58125505e-01 7.41776288e-01 -1.22647166e+00 -1.26802802e-01
9.21719432e-01 3.74760211e-01 -4.44290429e-01 1.45864379e+00
-2.20510915e-01 1.08694804e+00 -4.80543971e-01 9.29406881e-02
-3.87789369e-01 3.21861766e-02 6.39152110e-01 7.09654748e-01
4.72613126e-01 -2.83441897e-02 6.24154955e-02 9.48522151e-01
1.11421525e-01 1.63421601e-01 -6.77576602e-01 -5.30182794e-02
4.76100951e-01 1.10515690e+00 -5.18164873e-01 -4.43599433e-01
-3.15086395e-01 5.42877614e-01 -1.72794685e-01 2.74572730e-01
-4.19241577e-01 -3.94936353e-01 -3.56347673e-03 7.31336176e-01
-3.95532325e-02 3.97563308e-01 -6.60171449e-01 -5.26734769e-01
-1.40395656e-01 -9.78835583e-01 4.53851163e-01 -5.31036258e-01
-8.89728487e-01 1.59525231e-01 -2.96723038e-01 -9.03953135e-01
-4.67879206e-01 -3.97304684e-01 -7.65419781e-01 1.50699079e+00
-1.18183839e+00 -5.64413786e-01 -3.97784591e-01 6.82781264e-02
-3.65788527e-02 -1.44320697e-01 1.20416224e+00 4.57038850e-01
-5.02453923e-01 1.17463708e-01 -2.17239350e-01 2.76879251e-01
7.51998901e-01 -1.72292686e+00 -1.92490309e-01 6.52360141e-01
-8.70731592e-01 8.41466129e-01 6.30755067e-01 -9.34512198e-01
-8.37896407e-01 -1.04304874e+00 1.24523866e+00 -3.26647431e-01
2.22279176e-01 3.85027558e-01 -9.56868649e-01 3.79473984e-01
8.65683332e-02 2.37750858e-02 1.22799277e+00 -2.92608682e-02
2.73263484e-01 -1.27131775e-01 -1.33698213e+00 4.15399581e-01
3.71255785e-01 -3.33121806e-01 -6.39474154e-01 1.36285305e-01
5.37308827e-02 -4.79204386e-01 -1.63675642e+00 7.95345426e-01
8.30029011e-01 -9.89724994e-01 7.64709473e-01 -4.27263647e-01
4.83214080e-01 -3.01917586e-02 4.69888657e-01 -1.01519227e+00
-6.44525945e-01 -5.58328688e-01 -4.14380848e-01 7.35137641e-01
5.00628531e-01 -7.83689260e-01 1.03247023e+00 8.63254249e-01
-2.89703626e-02 -1.06810725e+00 -7.06503212e-01 -4.93009627e-01
1.55766860e-01 -1.08379908e-01 6.62946701e-02 9.09847140e-01
-1.15220606e-01 3.65514159e-01 -3.18510652e-01 -7.46610463e-02
4.34694618e-01 -8.86017308e-02 4.09648538e-01 -1.68316698e+00
-5.20478249e-01 -3.76685560e-01 -4.47494656e-01 -2.69761473e-01
-6.24681234e-01 -6.86389327e-01 -4.49044138e-01 -2.04769397e+00
1.14161447e-01 -2.40566760e-01 -3.95580322e-01 4.47324812e-01
-6.96287155e-02 8.65420997e-02 -3.07520851e-03 7.81671703e-02
3.78742993e-01 -5.30547723e-02 1.35693145e+00 3.45736682e-01
-4.57127035e-01 3.42144161e-01 -7.61266828e-01 5.64986587e-01
1.25382578e+00 -5.64550579e-01 -5.51922262e-01 3.59779239e-01
9.45725515e-02 5.45970559e-01 2.89051175e-01 -1.25463438e+00
-5.92002608e-02 2.27017235e-03 1.02377868e+00 -6.66284800e-01
1.54133245e-01 -8.45713496e-01 3.78438622e-01 9.78147864e-01
-1.93785727e-01 -5.73756333e-05 9.17936787e-02 1.93208888e-01
-1.26972552e-02 -1.31373331e-01 1.19132495e+00 -4.65884000e-01
-2.39448130e-01 3.12852353e-01 -9.50242877e-01 3.03992331e-01
1.08404422e+00 -2.45136589e-01 -8.22205693e-02 -6.32456481e-01
-1.28265011e+00 2.00178266e-01 7.56271780e-02 2.67016888e-03
6.59819007e-01 -8.41481507e-01 -8.21277678e-01 2.65616506e-01
-4.13244098e-01 9.10104215e-02 3.66766036e-01 1.36640060e+00
-1.02398539e+00 3.53494138e-01 -1.37014672e-01 -6.05717838e-01
-1.37485683e+00 1.53732315e-01 9.55722928e-01 -2.63666481e-01
-1.30838358e+00 6.33575320e-01 -3.98051441e-01 -4.45045643e-02
5.39449811e-01 -1.27528787e-01 -3.72405291e-01 -1.97076961e-01
5.22478104e-01 6.10114038e-01 -2.65884072e-01 1.05634876e-01
-1.50608003e-01 3.30405720e-02 1.37728751e-01 8.55930075e-02
1.20269608e+00 1.10603191e-01 -3.28273863e-01 5.26447892e-01
7.10290670e-01 -1.38600871e-01 -5.47816098e-01 2.03498930e-01
-2.62263179e-01 -2.61241287e-01 1.09095223e-01 -1.16746950e+00
-1.00671756e+00 8.20037901e-01 1.10028410e+00 3.98317605e-01
1.05859649e+00 -2.12214068e-01 5.34419596e-01 1.90204769e-01
-4.64734882e-01 -9.78968740e-01 -1.77607268e-01 8.97548199e-02
6.50659025e-01 -8.11722398e-01 1.76707823e-02 -4.64242063e-02
-4.38470215e-01 1.17496240e+00 6.19818628e-01 -1.60553679e-01
8.07981849e-01 2.69585401e-01 4.32071656e-01 -1.68883920e-01
-7.78200805e-01 1.04295902e-01 -1.40754908e-01 7.82456160e-01
6.70363307e-01 1.43570170e-01 -6.14753604e-01 4.83330578e-01
-1.24721617e-01 4.64008093e-01 4.88526165e-01 9.19591486e-01
-5.98087132e-01 -9.81546044e-01 -1.81506425e-01 1.24066520e+00
-1.08135962e+00 -1.54604644e-01 -2.78886080e-01 7.16248810e-01
1.39410838e-01 6.99073076e-01 -7.60997832e-02 7.08719790e-02
4.23764765e-01 6.02042317e-01 2.24687383e-01 -6.74461424e-01
-4.81074959e-01 2.19714642e-01 4.08183217e-01 -8.72723237e-02
-2.20325321e-01 -7.16716111e-01 -1.27210236e+00 -2.69781500e-01
1.28691331e-01 5.97900003e-02 3.56326550e-01 7.03388989e-01
3.78434211e-01 9.08924997e-01 3.96913409e-01 -7.52255088e-04
-6.41835690e-01 -1.15536606e+00 -1.01646304e+00 7.94831011e-03
4.37990457e-01 -2.63461769e-01 -7.36613452e-01 -1.82711750e-01] | [14.285680770874023, 3.2185983657836914] |
5a50f288-a54a-4573-9ee3-5eaf07685383 | facial-expression-recognition-using-vanilla | 2207.11081 | null | https://arxiv.org/abs/2207.11081v3 | https://arxiv.org/pdf/2207.11081v3.pdf | Emotion Separation and Recognition from a Facial Expression by Generating the Poker Face with Vision Transformers | Representation learning and feature disentanglement have recently attracted much research interests in facial expression recognition. The ubiquitous ambiguity of emotion labels is detrimental to those methods based on conventional supervised representation learning. Meanwhile, directly learning the mapping from a facial expression image to an emotion label lacks explicit supervision signals of facial details. In this paper, we propose a novel FER model, called Poker Face Vision Transformer or PF-ViT, to separate and recognize the disturbance-agnostic emotion from a static facial image via generating its corresponding poker face without the need for paired images. Here, we regard an expressive face as the comprehensive result of a set of facial muscle movements on one's poker face (i.e., emotionless face), inspired by Facial Action Coding System. The proposed PF-ViT leverages vanilla Vision Transformers, and are firstly pre-trained as Masked Autoencoders on a large facial expression dataset without emotion labels, obtaining excellent representations. It mainly consists of five components: 1) an encoder mapping the facial expression to a complete representation, 2) a separator decomposing the representation into an emotional component and an orthogonal residue, 3) a generator that can reconstruct the expressive face and synthesize the poker face, 4) a discriminator distinguishing the fake face produced by the generator, trained adversarially with the encoder and generator, 5) a classification head recognizing the emotion. Quantitative and qualitative results demonstrate the effectiveness of our method, which trumps the state-of-the-art methods on four popular FER testing sets. | ['Meng Wang', 'Richang Hong', 'Dan Guo', 'Jiantao Nie', 'Jia Li'] | 2022-07-22 | null | null | null | null | ['facial-expression-recognition'] | ['computer-vision'] | [ 4.59881574e-01 4.59855407e-01 6.28560930e-02 -3.78091484e-01
-5.78367651e-01 -4.58890855e-01 4.95816112e-01 -1.16587162e+00
1.81108639e-01 6.32386208e-01 1.53816804e-01 1.49588957e-01
4.04140234e-01 -4.20892268e-01 -8.45961571e-01 -1.12518501e+00
1.62494689e-01 -9.83873196e-03 -6.95903361e-01 -3.76271278e-01
-3.63423854e-01 4.94353086e-01 -1.74427700e+00 4.54875946e-01
5.51658392e-01 1.74301815e+00 -4.32613671e-01 3.83461177e-01
7.20317960e-02 1.26930821e+00 -7.42707849e-01 -6.54702187e-01
3.36186171e-01 -9.07548785e-01 -4.86138999e-01 3.98272932e-01
4.05808657e-01 -4.79260951e-01 -5.56727409e-01 1.24043024e+00
3.63665462e-01 -2.19433516e-01 8.86260390e-01 -1.67361140e+00
-1.18696332e+00 7.57126808e-02 -7.95238316e-01 -4.66819674e-01
3.94543171e-01 1.78207904e-01 6.04482234e-01 -1.01542854e+00
6.70133173e-01 1.34658515e+00 6.03133559e-01 1.01494563e+00
-1.25330198e+00 -1.20927370e+00 -1.59015149e-01 -1.19195938e-01
-1.45786846e+00 -9.63891864e-01 1.13723743e+00 -6.31214261e-01
4.16314751e-01 1.62545249e-01 6.58411682e-01 1.61844385e+00
3.96516100e-02 5.78022957e-01 1.36533689e+00 -1.49701312e-01
7.92092457e-02 1.73248678e-01 -4.41606134e-01 9.29735422e-01
-3.91132444e-01 5.95322907e-01 -3.44314575e-01 -2.60751754e-01
6.50240004e-01 1.53105095e-01 -5.94814479e-01 -3.21240202e-02
-4.19090807e-01 7.63368487e-01 3.99306446e-01 -7.03212321e-02
-6.49332583e-01 2.19261482e-01 3.24152470e-01 5.38458824e-01
4.94159788e-01 1.39986426e-01 -1.34342551e-01 1.96450278e-01
-8.80072474e-01 1.50059664e-03 6.94810748e-01 6.70135617e-01
8.32986951e-01 6.89862251e-01 -1.87895611e-01 6.65022910e-01
3.81408483e-01 6.59863114e-01 6.05922103e-01 -9.46365237e-01
-1.49014741e-01 4.85599965e-01 -1.40623704e-01 -1.14916444e+00
1.13836825e-01 9.01575238e-02 -1.01382113e+00 6.41533315e-01
3.27271298e-02 -3.69881719e-01 -8.43821406e-01 2.23372984e+00
1.76296681e-01 6.28979981e-01 3.97258013e-01 1.05063045e+00
9.60316956e-01 7.49450445e-01 -8.78566504e-02 -4.19310212e-01
1.42595840e+00 -8.78784299e-01 -8.88799429e-01 -6.38741329e-02
1.65452793e-01 -5.70135772e-01 6.20990455e-01 3.72115582e-01
-9.63306904e-01 -6.25975072e-01 -1.09188080e+00 -3.47773843e-02
-1.40483961e-01 5.47301590e-01 6.32549465e-01 4.22986448e-01
-9.80857551e-01 4.36760426e-01 -3.54556113e-01 2.43762866e-01
8.39606047e-01 3.79190505e-01 -9.66805875e-01 8.80961493e-02
-1.25587153e+00 6.56286061e-01 -5.29269390e-02 3.55976105e-01
-1.26546836e+00 -5.23920834e-01 -1.10105455e+00 1.63421869e-01
1.02255195e-01 -4.92785394e-01 1.03439307e+00 -2.20891547e+00
-2.02167892e+00 1.20723128e+00 -1.20363347e-01 -7.32648233e-03
3.89348984e-01 1.12982079e-01 -7.73173869e-01 2.30249509e-01
-3.68658118e-02 5.90826094e-01 1.71945059e+00 -1.37387478e+00
-2.81574093e-02 -5.17181396e-01 -2.96392977e-01 -1.59939751e-01
-1.06721230e-01 8.61031413e-02 -2.59829666e-02 -7.85519302e-01
-1.78996667e-01 -8.41077685e-01 3.13593447e-01 4.36822563e-01
-4.14857805e-01 8.11727569e-02 1.19033897e+00 -7.71651924e-01
7.39145398e-01 -2.48881912e+00 2.57761776e-01 4.72275056e-02
4.04983580e-01 2.71577030e-01 -4.92336363e-01 7.08449334e-02
-7.97980011e-01 -9.62591097e-02 -1.44212335e-01 -3.21546704e-01
-5.31955361e-02 1.01035886e-01 -8.64507139e-01 7.50872612e-01
7.65639186e-01 1.16757214e+00 -7.01717794e-01 -1.55539513e-01
-2.79918760e-01 8.23273659e-01 -4.04130518e-01 6.17823362e-01
-3.47636901e-02 6.02273703e-01 -4.19716090e-01 8.95392299e-01
8.51827860e-01 1.75889898e-02 4.53027375e-02 -4.01493728e-01
2.86360979e-01 -3.07974964e-01 -6.67156696e-01 1.44187558e+00
-2.82862991e-01 5.47561228e-01 4.54372406e-01 -1.13737428e+00
1.26940894e+00 6.82468534e-01 3.60936522e-01 -6.11511767e-01
5.84468961e-01 1.71592787e-01 -1.76708251e-01 -6.89424276e-01
-1.73752513e-02 -7.56983459e-01 2.08107140e-02 5.02691150e-01
5.58878779e-01 5.85005544e-02 -4.44870740e-01 -2.45288819e-01
9.55433190e-01 3.40516508e-01 1.30114973e-01 1.21966422e-01
6.30235493e-01 -6.33433342e-01 6.39023244e-01 -4.61233407e-02
-2.44638562e-01 4.58286285e-01 7.17760503e-01 -2.86659747e-01
-6.04841650e-01 -8.92180145e-01 7.06813857e-02 8.08443725e-01
-6.70872554e-02 -7.39936829e-02 -8.86653304e-01 -7.73848295e-01
6.84796721e-02 2.95759827e-01 -1.01002562e+00 -7.08920419e-01
-1.56174928e-01 -1.81644365e-01 1.03583956e+00 3.20190907e-01
6.31870925e-01 -1.27249491e+00 -3.45101535e-01 -2.50447243e-01
-1.08212128e-01 -9.84428525e-01 -4.07657444e-01 9.96662006e-02
-2.65165389e-01 -1.13754606e+00 -7.52215803e-01 -7.98724353e-01
7.56054878e-01 -6.60950318e-02 8.41420710e-01 -1.91056412e-02
-2.12679833e-01 1.71415329e-01 -1.73024178e-01 -3.81967932e-01
-5.86516082e-01 -8.96869004e-01 2.04797432e-01 8.92855346e-01
2.70277470e-01 -7.91667819e-01 -4.99922037e-01 1.28101230e-01
-9.32748437e-01 -9.39447209e-02 6.44839168e-01 1.22664976e+00
5.60389400e-01 -1.14902057e-01 6.22942805e-01 -7.84224510e-01
4.33733046e-01 -6.01802170e-01 -2.14446336e-01 8.84214565e-02
-2.02228814e-01 3.15763652e-02 8.16179097e-01 -8.11010778e-01
-1.15017760e+00 1.73656702e-01 -1.41025782e-01 -1.29962969e+00
-4.00210358e-02 8.97937790e-02 -5.90775847e-01 -2.96888024e-01
6.45667672e-01 4.08364177e-01 5.86604059e-01 -8.81872103e-02
5.92715085e-01 6.99255645e-01 8.73278916e-01 -5.42545974e-01
8.57931256e-01 6.31106853e-01 -7.66482204e-02 -6.34729803e-01
-6.50740862e-01 1.65063709e-01 -2.38478273e-01 -3.21815133e-01
8.01455021e-01 -1.08950067e+00 -8.51913333e-01 6.70525074e-01
-1.31552482e+00 -6.65636286e-02 -3.73223722e-01 7.28506073e-02
-7.74198890e-01 3.96181680e-02 -6.94850206e-01 -7.65201569e-01
-2.68444568e-01 -1.06665063e+00 1.27623129e+00 3.24975044e-01
-6.94287568e-02 -5.63876510e-01 5.00968434e-02 2.72679597e-01
2.65548080e-01 7.61068165e-01 6.90315127e-01 -3.51236284e-01
-2.46499181e-01 -2.76865691e-01 -1.31839946e-01 9.38673854e-01
2.14869604e-01 9.67594013e-02 -1.55879033e+00 -1.81627035e-01
3.77986014e-01 -9.16477680e-01 7.28762805e-01 -6.05296530e-02
1.14384675e+00 -7.74179339e-01 -1.00801855e-01 1.08081269e+00
1.15811324e+00 1.99294671e-01 9.15036500e-01 -3.63420486e-01
5.96667230e-01 6.80992961e-01 2.11608037e-01 2.77429521e-01
-1.32072330e-01 5.28853238e-01 5.92869520e-01 -3.79092872e-01
-2.03899026e-01 -5.24450660e-01 8.29736173e-01 4.66415316e-01
-2.03947216e-01 -2.65996121e-02 -7.82223269e-02 -1.09815389e-01
-1.53098357e+00 -1.10540330e+00 5.86681366e-01 1.62765419e+00
7.42472351e-01 -5.50952137e-01 -2.16724351e-01 1.64779693e-01
6.24278069e-01 2.91621178e-01 -8.44312608e-01 -5.33188641e-01
-1.95245326e-01 5.89294016e-01 -2.20114484e-01 9.38183963e-02
-9.59476590e-01 9.89990830e-01 5.31800175e+00 6.69178963e-01
-1.72741425e+00 1.28286153e-01 7.53458798e-01 9.25669149e-02
-8.29567686e-02 -1.49998084e-01 -1.87763900e-01 5.51184654e-01
7.54348159e-01 -7.46073127e-02 6.04579926e-01 1.06563699e+00
2.90343389e-02 5.73737204e-01 -1.18952334e+00 1.33550072e+00
5.02768815e-01 -1.02437377e+00 2.95525432e-01 1.07973561e-01
5.30125916e-01 -5.64879417e-01 4.03666705e-01 5.59757948e-01
1.78147666e-02 -1.57005072e+00 8.42453599e-01 8.85428786e-01
1.58209908e+00 -5.22671640e-01 4.94349808e-01 3.26915756e-02
-8.24146748e-01 -5.86206615e-02 -1.18948475e-01 -1.61767513e-01
-8.18354785e-02 1.38390571e-01 -1.98599026e-01 4.53387529e-01
4.51435775e-01 8.15142632e-01 1.84475798e-02 -6.21010326e-02
-4.24382538e-01 5.47249913e-01 9.16356072e-02 5.33090472e-01
-2.77400911e-02 -4.95069176e-01 3.66437882e-01 8.43554974e-01
3.69324267e-01 2.42295995e-01 -1.95897952e-01 1.32977378e+00
-5.96783876e-01 -1.71709135e-01 -9.68376756e-01 -3.85394603e-01
1.75600737e-01 1.56443870e+00 -3.91756147e-02 -2.26073653e-01
-4.53296453e-01 1.40562499e+00 3.44557792e-01 6.85377955e-01
-1.00015473e+00 -2.37143934e-01 1.10495138e+00 -1.19332537e-01
2.93696523e-01 5.83294451e-01 3.33679855e-01 -1.20645630e+00
9.55902487e-02 -1.13876164e+00 -7.79858604e-02 -1.17877769e+00
-1.35736489e+00 1.00055122e+00 -6.15303755e-01 -1.27118289e+00
-4.17211473e-01 -8.20903122e-01 -7.83849299e-01 1.00222504e+00
-1.49251914e+00 -1.46992183e+00 -4.99774545e-01 9.11631465e-01
-8.22917651e-03 -4.88557070e-01 1.24948430e+00 2.46099502e-01
-6.03271186e-01 9.25200224e-01 -2.81459183e-01 4.07480717e-01
6.13769174e-01 -6.43591166e-01 -4.08054352e-01 5.43852270e-01
9.25714821e-02 3.38051617e-01 3.11801612e-01 -1.28931239e-01
-1.50659621e+00 -1.19886541e+00 4.30943221e-01 -1.24168172e-01
5.75777173e-01 -4.17124957e-01 -8.61346126e-01 7.99905658e-01
-1.10555878e-02 5.58783889e-01 8.55840027e-01 -5.40533602e-01
-8.96187603e-01 -2.42440149e-01 -1.33998120e+00 4.71486837e-01
8.86809647e-01 -9.08696473e-01 -4.87656474e-01 6.89211413e-02
4.38166499e-01 -2.70623535e-01 -6.51792467e-01 3.68964106e-01
8.61841559e-01 -9.69808698e-01 7.40452707e-01 -8.94575059e-01
1.00840449e+00 -2.31106177e-01 -2.43662998e-01 -1.29478765e+00
-1.22847043e-01 -8.63845050e-01 -2.30227694e-01 1.24303710e+00
-1.25985146e-02 -6.54640794e-01 6.46882594e-01 3.88468206e-01
-6.56607896e-02 -9.36970830e-01 -9.86926377e-01 -5.61780810e-01
-4.98806238e-02 -1.06276758e-03 7.49504089e-01 1.25107396e+00
-1.55322120e-01 5.49966753e-01 -6.64713860e-01 -4.16400991e-02
1.98763773e-01 1.76220462e-01 7.78860211e-01 -1.02606273e+00
-4.55612481e-01 -4.94574130e-01 -6.31987870e-01 -7.94498742e-01
1.02518952e+00 -1.22874165e+00 1.70390848e-02 -5.50531924e-01
1.33965030e-01 2.12015864e-02 -8.52428526e-02 8.17436337e-01
1.00656040e-01 5.37133455e-01 8.56948718e-02 2.09023699e-01
-1.13376724e-02 1.18811643e+00 1.38456357e+00 -2.83854455e-01
2.59833902e-01 -3.00724328e-01 -1.11240411e+00 8.50104511e-01
3.38795036e-01 -3.95250231e-01 -3.83763731e-01 -4.86573540e-02
-9.74389687e-02 3.80410254e-01 6.56011999e-01 -5.80242276e-01
-2.28200078e-01 2.19540726e-02 6.31027341e-01 3.62352282e-01
6.52402163e-01 -8.91749680e-01 3.34494472e-01 1.65969461e-01
-1.71823606e-01 -1.88211098e-01 1.70402363e-01 5.39351463e-01
-5.72799265e-01 3.03478800e-02 9.38556194e-01 5.20410128e-02
-5.93966305e-01 5.58357239e-01 -7.38051832e-02 -1.43673509e-01
1.10163629e+00 -2.21922487e-01 -3.12914312e-01 -5.96740365e-01
-7.62801528e-01 -2.00996011e-01 3.59089971e-01 3.91920298e-01
7.97035813e-01 -1.68084311e+00 -6.71161473e-01 6.92129850e-01
1.35311753e-01 -3.51820767e-01 2.33601868e-01 5.95884979e-01
-9.68528613e-02 -1.97829217e-01 -5.30904531e-01 -2.31905133e-01
-1.17251229e+00 6.27182901e-01 6.45228088e-01 1.56395346e-01
-4.00763065e-01 8.63532424e-01 6.60064042e-01 -3.28091741e-01
-1.13842167e-01 2.31723934e-01 -8.62327442e-02 8.30249637e-02
6.18274808e-01 -1.50589153e-01 -2.66405344e-01 -1.29737723e+00
-2.17066303e-01 5.17600477e-01 2.38171726e-01 4.41432483e-02
1.05250299e+00 3.37666452e-01 -3.80823523e-01 1.58353597e-01
1.73633814e+00 -6.88134432e-02 -1.46869528e+00 -2.42389962e-02
-5.88876843e-01 -3.64452124e-01 -2.67168134e-02 -6.86224520e-01
-1.56557465e+00 8.42635691e-01 5.39409637e-01 -2.43270725e-01
1.52757013e+00 -1.91096980e-02 7.83529818e-01 -3.57618295e-02
8.75540823e-02 -4.84516621e-01 4.53639209e-01 1.67780578e-01
1.28301024e+00 -1.03645909e+00 -5.11231780e-01 -3.28608364e-01
-8.28700900e-01 1.13819146e+00 5.07038534e-01 -3.60785723e-01
8.20420563e-01 3.68705899e-01 3.28956217e-01 -4.12930012e-01
-7.35474527e-01 3.82005237e-02 2.55965650e-01 7.06741989e-01
1.15313016e-01 5.93972690e-02 3.03306311e-01 1.24659848e+00
-4.20648634e-01 4.53273982e-01 2.01991513e-01 4.05543029e-01
1.77982360e-01 -6.88995183e-01 -1.97389856e-01 2.37075850e-01
-4.88239080e-01 1.17440172e-01 -6.16715133e-01 7.23582208e-01
4.59441841e-01 6.50794327e-01 2.03928456e-01 -7.32334971e-01
3.00571412e-01 3.27633560e-01 6.07266784e-01 -3.47708881e-01
-3.52768302e-01 -9.66438204e-02 -2.66085416e-01 -7.61228442e-01
-3.75191897e-01 -2.34885588e-01 -1.09104419e+00 -1.76415592e-01
-1.38620749e-01 1.15203992e-01 3.15073460e-01 8.33624542e-01
5.30445039e-01 3.33632350e-01 1.08117032e+00 -1.03542960e+00
-6.45623624e-01 -9.01677668e-01 -8.81547570e-01 7.87331820e-01
6.39902353e-01 -9.57718670e-01 -7.34215975e-01 3.95586908e-01] | [12.956404685974121, 0.2759247124195099] |
39295310-ee79-4753-9beb-af15363e3f0a | vision-language-pre-training-with-object | 2305.10714 | null | https://arxiv.org/abs/2305.10714v1 | https://arxiv.org/pdf/2305.10714v1.pdf | Vision-Language Pre-training with Object Contrastive Learning for 3D Scene Understanding | In recent years, vision language pre-training frameworks have made significant progress in natural language processing and computer vision, achieving remarkable performance improvement on various downstream tasks. However, when extended to point cloud data, existing works mainly focus on building task-specific models, and fail to extract universal 3D vision-language embedding that generalize well. We carefully investigate three common tasks in semantic 3D scene understanding, and derive key insights into the development of a pre-training model. Motivated by these observations, we propose a vision-language pre-training framework 3DVLP (3D vision-language pre-training with object contrastive learning), which transfers flexibly on 3D vision-language downstream tasks. 3DVLP takes visual grounding as the proxy task and introduces Object-level IoU-guided Detection (OID) loss to obtain high-quality proposals in the scene. Moreover, we design Object-level Cross-Contrastive alignment (OCC) task and Object-level Self-Contrastive learning (OSC) task to align the objects with descriptions and distinguish different objects in the scene, respectively. Extensive experiments verify the excellent performance of 3DVLP on three 3D vision-language tasks, reflecting its superiority in semantic 3D scene understanding. | ['Shu-Tao Xia', 'Zhi Wang', 'Bin Chen', 'Dai Tao', 'Sunan He', 'Taolin Zhang'] | 2023-05-18 | null | null | null | null | ['visual-grounding', 'scene-understanding'] | ['computer-vision', 'computer-vision'] | [ 2.45765075e-02 -1.41310379e-01 -1.32310614e-01 -5.80376804e-01
-5.27577460e-01 -5.41212380e-01 9.70397770e-01 2.73048095e-02
-2.65565932e-01 -1.91342250e-01 2.53471918e-02 -4.57349062e-01
2.62326926e-01 -6.08310282e-01 -6.91995144e-01 -4.49586719e-01
4.62712169e-01 4.91230428e-01 3.78379256e-01 -1.52552590e-01
2.93050200e-01 6.79382622e-01 -1.45703804e+00 2.78960496e-01
6.32630408e-01 9.56521630e-01 7.65048683e-01 5.01314223e-01
-5.74114621e-01 5.78677058e-01 -2.66359448e-01 -2.37502694e-01
5.04866004e-01 -4.02484424e-02 -8.07649493e-01 4.60863411e-01
7.72166014e-01 -3.63652140e-01 -4.00413603e-01 1.27794540e+00
3.38986903e-01 -1.08286127e-01 8.02218258e-01 -1.40212798e+00
-1.30566287e+00 -2.15693992e-02 -6.40746713e-01 2.50172883e-01
1.88014776e-01 4.64771152e-01 1.12049985e+00 -1.46568012e+00
3.76361430e-01 1.72199833e+00 3.91123354e-01 7.38943279e-01
-1.15036595e+00 -2.93266475e-01 5.85366488e-01 2.61293679e-01
-1.30597353e+00 -2.40499437e-01 9.97363985e-01 -5.73481679e-01
1.23521185e+00 -2.83097494e-02 5.76516509e-01 1.01590204e+00
-1.63408611e-02 1.14018965e+00 1.18169653e+00 -3.08132917e-01
1.19228242e-02 1.27395794e-01 2.92038947e-01 7.52114058e-01
2.43565142e-01 2.90689111e-01 -3.51308942e-01 3.75505328e-01
9.91815627e-01 1.71956345e-01 -1.40107363e-01 -6.26264691e-01
-1.33196318e+00 6.16250277e-01 8.81271303e-01 3.30484137e-02
-2.29548886e-01 1.12872444e-01 3.29751283e-01 1.95486665e-01
5.26904881e-01 1.14805311e-01 -4.33729559e-01 5.83986342e-01
-4.67947870e-01 2.16410190e-01 2.15574428e-01 1.38926613e+00
8.93142939e-01 9.51841474e-02 -3.34805578e-01 8.30181062e-01
6.85282171e-01 8.61724854e-01 -4.51998711e-02 -9.53588843e-01
4.49341118e-01 7.32578874e-01 -1.47400528e-01 -9.02356684e-01
-2.72949845e-01 -5.26118994e-01 -9.46778119e-01 3.56101125e-01
8.84263664e-02 5.53872347e-01 -1.16110039e+00 1.63683355e+00
2.51141071e-01 1.43157840e-01 3.65190834e-01 1.34822500e+00
1.22924709e+00 7.23032951e-01 2.55026072e-01 9.60169882e-02
1.35843158e+00 -1.16852677e+00 -1.09831020e-01 -5.44907629e-01
4.43882585e-01 -7.08638728e-01 1.39115012e+00 -9.38967019e-02
-1.18905592e+00 -1.12768710e+00 -9.28152919e-01 -6.46471560e-01
-3.10083658e-01 -3.99600714e-03 6.22013628e-01 1.99787438e-01
-1.26069558e+00 -2.26402104e-01 -5.67087889e-01 -4.56155002e-01
6.46526158e-01 -8.42112303e-02 -3.26421678e-01 -4.14431334e-01
-8.32075417e-01 1.04858840e+00 4.90685046e-01 8.22983086e-02
-1.33769178e+00 -7.43337333e-01 -1.02660525e+00 -2.10430503e-01
4.32877123e-01 -1.42556798e+00 1.13855100e+00 -7.99759209e-01
-1.26494420e+00 1.74727356e+00 -1.73577145e-01 -3.35457683e-01
4.18234795e-01 -2.53559481e-02 -4.63685505e-02 2.79364865e-02
4.28958416e-01 1.20577419e+00 1.03239739e+00 -1.76188993e+00
-6.61049724e-01 -5.46193898e-01 4.85428125e-01 4.89975274e-01
1.20860912e-01 -1.52115479e-01 -9.42730725e-01 -4.64804560e-01
4.61655051e-01 -7.07320392e-01 -2.62417495e-01 4.05027032e-01
-3.21912110e-01 -6.12929583e-01 7.03294337e-01 -2.11126536e-01
3.15872848e-01 -2.13320041e+00 2.49509484e-01 -3.52054596e-01
3.88351858e-01 3.10535997e-01 -5.98710418e-01 1.65008418e-02
9.84826759e-02 -1.66736260e-01 -2.97649026e-01 -5.70596755e-01
1.53757304e-01 4.60957229e-01 -6.20200038e-01 2.86533177e-01
7.25399911e-01 1.42036498e+00 -1.06385148e+00 -5.60565293e-01
6.72200024e-01 2.52018988e-01 -6.63304925e-01 4.28497344e-01
-5.23805857e-01 4.35517877e-01 -7.44735599e-01 8.52308750e-01
8.16451132e-01 -4.71403569e-01 -4.60548460e-01 -4.01122391e-01
-2.32960746e-01 1.68899804e-01 -6.14384353e-01 2.06297469e+00
-6.26334429e-01 5.33521712e-01 -9.67379287e-02 -1.30279589e+00
1.09919405e+00 -3.16664189e-01 1.17720790e-01 -8.60427260e-01
2.63968799e-02 5.73233664e-02 -2.75541335e-01 -6.18021190e-01
3.15733999e-01 -9.28726420e-03 -1.22999072e-01 1.01413429e-01
2.25887254e-01 -9.06842232e-01 -3.57571840e-01 1.81748852e-01
4.33037549e-01 1.77295640e-01 2.54748702e-01 -2.79752642e-01
9.16054487e-01 1.35957524e-01 2.12177783e-01 8.21855307e-01
-5.27053475e-01 8.69329154e-01 1.33585885e-01 -3.44700038e-01
-9.78650153e-01 -1.53724134e+00 -1.00698538e-01 9.31555390e-01
8.73582721e-01 -1.91328391e-01 -3.06607813e-01 -8.51595521e-01
1.80603057e-01 8.56531799e-01 -2.74828434e-01 -2.29857415e-01
-3.03980529e-01 -3.69996816e-01 3.60133618e-01 6.01177812e-01
7.72332668e-01 -9.13689375e-01 -3.46516132e-01 -1.67642012e-01
2.57466510e-02 -1.55704391e+00 -4.47849303e-01 6.29441217e-02
-7.79857039e-01 -8.98724735e-01 -7.06836462e-01 -1.32806265e+00
7.35746682e-01 1.20454895e+00 1.27010429e+00 -1.01131231e-01
-1.64893538e-01 9.58635509e-01 -1.46673650e-01 -5.97197890e-01
-4.65071410e-01 -2.77876645e-01 1.12323426e-01 -1.80505723e-01
6.25814915e-01 -2.55519390e-01 -4.22519624e-01 1.46198407e-01
-7.69200683e-01 3.10266137e-01 7.88484931e-01 7.23553002e-01
8.91370296e-01 -4.05485392e-01 3.29037994e-01 -2.04445556e-01
3.00310493e-01 -1.85017079e-01 -6.85967803e-01 3.94662410e-01
-5.30538023e-01 -1.65397927e-04 4.49997127e-01 -1.33728668e-01
-8.18728149e-01 -3.43348421e-02 -2.86957532e-01 -1.00359702e+00
-4.52703446e-01 1.38751090e-01 -4.29210961e-01 -1.24817625e-01
3.50417048e-01 6.84696019e-01 7.28938207e-02 -3.42445940e-01
8.92776608e-01 5.60385346e-01 7.66158283e-01 -5.48455417e-01
1.05660164e+00 7.97607243e-01 -2.01468617e-02 -1.01622522e+00
-1.44137156e+00 -6.97977483e-01 -8.26515853e-01 -1.08644381e-01
1.18818462e+00 -1.34216130e+00 -5.75157344e-01 5.13694644e-01
-1.53300810e+00 -1.80342391e-01 -2.58362085e-01 3.03458989e-01
-7.36153424e-01 5.43342113e-01 -1.34682059e-01 -5.44259548e-01
-3.12617153e-01 -1.18570101e+00 1.53682232e+00 1.37200400e-01
3.51559132e-01 -9.75197315e-01 -3.35792184e-01 4.17384416e-01
4.17661332e-02 -2.34655768e-01 1.15534055e+00 -3.07387829e-01
-9.44727719e-01 3.41045260e-01 -1.02463508e+00 7.94373393e-01
-1.30527988e-01 -3.87088716e-01 -1.00965285e+00 -1.80895761e-01
2.16870815e-01 -5.24520338e-01 1.09039700e+00 4.70308959e-01
1.31737721e+00 2.03321651e-01 -1.83082610e-01 8.79299223e-01
1.40273213e+00 -2.90157735e-01 2.51690775e-01 1.94983944e-01
1.07675457e+00 6.80176020e-01 6.53095186e-01 -6.65567815e-02
8.48042309e-01 7.34082282e-01 8.16206098e-01 -3.66846800e-01
-6.12407446e-01 -4.89078760e-01 3.87069166e-01 7.85359681e-01
7.43985325e-02 -1.46995336e-01 -9.22644079e-01 5.20256639e-01
-1.70568967e+00 -7.03411162e-01 -2.09882468e-01 1.72392726e+00
4.38137054e-01 3.17131281e-01 -1.26767918e-01 -3.77979279e-01
4.94569957e-01 5.21673679e-01 -8.47422779e-01 -9.31314677e-02
-6.08176112e-01 -1.81741998e-01 3.66870403e-01 5.22679746e-01
-1.16336823e+00 1.47834516e+00 5.34714699e+00 5.91057420e-01
-1.16974962e+00 1.85550317e-01 4.66941804e-01 2.59507000e-01
-4.75508988e-01 -2.33359952e-02 -9.03599858e-01 -4.22787517e-02
-4.86562327e-02 -5.77308349e-02 1.80707984e-02 9.40421402e-01
3.17580283e-01 3.03749323e-01 -1.32674241e+00 1.65116668e+00
3.54799300e-01 -1.48368359e+00 7.82500803e-01 -1.51679635e-01
7.56344378e-01 4.45860118e-01 3.00870389e-01 3.84855002e-01
1.78739298e-02 -7.86855638e-01 9.82477725e-01 4.98084337e-01
7.08251595e-01 -3.28455329e-01 4.50562626e-01 4.24445808e-01
-1.12699270e+00 1.54309198e-02 -7.06009746e-01 -1.61421075e-02
2.79263020e-01 3.07276398e-01 -6.68471277e-01 5.96448600e-01
8.58345151e-01 1.23939323e+00 -5.30277550e-01 7.37825811e-01
-3.76956165e-01 1.87907651e-01 -5.25687896e-02 2.70106792e-02
7.50809312e-01 -2.54793614e-01 9.16396976e-01 1.12360930e+00
-1.67056452e-02 8.16554874e-02 3.91901225e-01 1.40985739e+00
1.77222565e-02 -2.88371801e-01 -6.78717911e-01 1.27327636e-01
2.39506945e-01 9.23198283e-01 -4.28581864e-01 -2.66774148e-01
-6.92004263e-01 1.21131682e+00 2.57911623e-01 4.49996680e-01
-6.50191307e-01 1.81535736e-01 1.13845992e+00 -1.07989885e-01
3.12019676e-01 -7.83370078e-01 -4.32501167e-01 -1.32961428e+00
2.28868909e-02 -4.63128328e-01 1.31415680e-01 -1.06014204e+00
-1.80737376e+00 3.32439244e-01 1.32398102e-02 -1.30800426e+00
1.17818885e-01 -9.60423231e-01 -4.96394396e-01 8.58758986e-01
-2.05615616e+00 -1.57576513e+00 -4.47799802e-01 7.39134789e-01
1.07554960e+00 -2.25456804e-01 4.42029655e-01 1.72643438e-02
-1.67848527e-01 3.16177964e-01 -3.38721722e-01 1.99295674e-02
4.32985187e-01 -1.16417003e+00 6.65691018e-01 8.49317968e-01
4.06492859e-01 4.42187846e-01 3.23144078e-01 -2.53435522e-01
-1.71118987e+00 -1.43036127e+00 8.76539469e-01 -8.98055017e-01
5.68930149e-01 -6.30402327e-01 -9.25360382e-01 3.67234379e-01
-1.47025764e-01 2.05584690e-01 2.53134198e-03 -2.19071642e-01
-6.34030163e-01 -7.04002276e-04 -7.04351902e-01 7.57878482e-01
1.68479502e+00 -1.02087879e+00 -1.08163810e+00 4.45283264e-01
1.25121677e+00 -3.31733704e-01 -3.71153384e-01 6.13512516e-01
1.37140527e-01 -8.83285522e-01 1.66632760e+00 -7.13439047e-01
3.79313529e-01 -6.27426028e-01 -5.09296417e-01 -8.75161409e-01
-3.56201202e-01 -9.60135385e-02 1.21277705e-01 1.02361035e+00
-2.72062905e-02 -3.68758917e-01 4.48204100e-01 5.26669957e-02
-6.14546657e-01 -5.33507466e-01 -7.26056993e-01 -8.91116202e-01
1.70279682e-01 -8.80996525e-01 2.52596319e-01 8.72937441e-01
-8.56619835e-01 7.23016024e-01 -5.59035912e-02 6.62357450e-01
8.99144113e-01 5.91438591e-01 1.02211285e+00 -1.12924373e+00
2.38256827e-02 -7.63499260e-01 -6.31360710e-01 -1.99488640e+00
5.14412224e-01 -1.47060204e+00 5.68037108e-02 -1.79755330e+00
3.03411752e-01 -5.16065836e-01 -3.25856835e-01 3.33755076e-01
-1.23050071e-01 1.93465859e-01 4.25025314e-01 3.78954858e-01
-8.70419145e-01 9.26721334e-01 1.52100277e+00 -6.40609264e-01
-1.79057404e-01 -6.80902749e-02 -6.67210877e-01 8.14056516e-01
3.03868681e-01 -5.46116345e-02 -5.91662347e-01 -1.01536810e+00
-1.14916317e-01 -3.41318458e-01 1.10484207e+00 -6.09193444e-01
1.06285900e-01 -2.11248070e-01 1.92603767e-01 -1.04938924e+00
3.01079273e-01 -7.28015482e-01 -5.60754478e-01 3.37287784e-01
-2.88160712e-01 -2.33979598e-01 1.91579610e-01 6.96686327e-01
-3.03480983e-01 1.01996049e-01 8.74922216e-01 -2.27210328e-01
-1.50347793e+00 7.80546427e-01 1.29590601e-01 1.96661234e-01
8.51606607e-01 -4.27137583e-01 -2.27520719e-01 7.79564604e-02
-5.72095513e-01 4.76407439e-01 5.72012067e-01 9.13354039e-01
1.09582257e+00 -1.17645621e+00 -9.25308943e-01 4.59830642e-01
8.50945592e-01 4.80784327e-01 3.27877849e-01 5.66920578e-01
-4.33415383e-01 6.37864292e-01 -1.84576094e-01 -1.32094133e+00
-8.98767233e-01 7.75182426e-01 4.75361466e-01 4.01188433e-01
-9.04699266e-01 1.14985120e+00 1.01812983e+00 -5.57343364e-01
2.29065791e-01 -5.16805232e-01 2.20386535e-02 -3.65590602e-01
3.56576562e-01 -1.98730022e-01 -1.43545792e-01 -7.68083394e-01
-5.71765363e-01 1.07058322e+00 -1.14591345e-02 1.28428057e-01
1.17420185e+00 -5.62741935e-01 -4.75000702e-02 6.05008423e-01
1.37250185e+00 -6.06933355e-01 -1.41912484e+00 -7.70841837e-01
-9.77296382e-02 -4.38203901e-01 2.51744211e-01 -4.34433967e-01
-8.15137088e-01 1.31555355e+00 5.72647929e-01 8.86863843e-03
1.02208364e+00 6.31691277e-01 5.53356230e-01 5.42570889e-01
4.52075064e-01 -5.86189687e-01 4.42939728e-01 9.58243012e-01
1.10981011e+00 -1.72769713e+00 -2.67722338e-01 -3.91699702e-01
-6.78643048e-01 9.27326381e-01 8.32401097e-01 -2.10880250e-01
6.30772054e-01 -2.96759874e-01 1.75704032e-01 -3.41283828e-01
-5.85893810e-01 -8.05731654e-01 5.56938529e-01 1.06455374e+00
-1.97171614e-01 -6.75543100e-02 2.83263743e-01 3.91625583e-01
2.70587057e-02 -4.12676573e-01 8.70229602e-02 3.98235172e-01
-6.34929299e-01 -6.45417809e-01 -6.35420457e-02 -7.34656155e-02
2.76425004e-01 -2.41993412e-01 -5.76591074e-01 7.51978219e-01
1.96179762e-01 7.68604338e-01 2.57335693e-01 -1.75982028e-01
4.64384109e-01 -3.76420259e-01 7.56350219e-01 -7.46917009e-01
1.90013889e-02 -1.45846277e-01 -3.14716816e-01 -5.73474228e-01
-6.50748789e-01 -4.08693790e-01 -1.33383930e+00 1.38664529e-01
-5.97439483e-02 -3.74782443e-01 5.85125566e-01 1.05638278e+00
3.62560749e-01 3.12693566e-01 6.80513799e-01 -1.01791370e+00
-4.97659475e-01 -4.64974552e-01 -3.71784389e-01 3.92093301e-01
5.56780994e-01 -8.98754954e-01 -3.01500320e-01 -1.90583861e-03] | [8.149785995483398, -3.345304250717163] |
0a89680a-7736-4e16-b234-4f0d9f4dc392 | coqar-question-rewriting-on-coqa | 2207.03240 | null | https://arxiv.org/abs/2207.03240v1 | https://arxiv.org/pdf/2207.03240v1.pdf | CoQAR: Question Rewriting on CoQA | Questions asked by humans during a conversation often contain contextual dependencies, i.e., explicit or implicit references to previous dialogue turns. These dependencies take the form of coreferences (e.g., via pronoun use) or ellipses, and can make the understanding difficult for automated systems. One way to facilitate the understanding and subsequent treatments of a question is to rewrite it into an out-of-context form, i.e., a form that can be understood without the conversational context. We propose CoQAR, a corpus containing $4.5$K conversations from the Conversational Question-Answering dataset CoQA, for a total of $53$K follow-up question-answer pairs. Each original question was manually annotated with at least 2 at most 3 out-of-context rewritings. CoQAR can be used in the supervised learning of three tasks: question paraphrasing, question rewriting and conversational question answering. In order to assess the quality of CoQAR's rewritings, we conduct several experiments consisting in training and evaluating models for these three tasks. Our results support the idea that question rewriting can be used as a preprocessing step for question answering models, thereby increasing their performances. | ['Lina M. Rojas-Barahona', 'Gwenole Lecorve', 'Quentin Brabant'] | 2022-07-07 | null | https://aclanthology.org/2022.lrec-1.13 | https://aclanthology.org/2022.lrec-1.13.pdf | lrec-2022-6 | ['question-rewriting'] | ['natural-language-processing'] | [ 2.15321794e-01 6.82122171e-01 2.08199099e-01 -8.10427606e-01
-8.79631281e-01 -1.13800335e+00 7.86074877e-01 2.85463274e-01
-3.13569397e-01 8.72609198e-01 5.35019815e-01 -8.12772334e-01
8.86589512e-02 -8.12352955e-01 -5.05716801e-01 -2.57517658e-02
3.80353421e-01 7.09104121e-01 3.17677230e-01 -8.70075345e-01
2.63701826e-01 -1.90540089e-03 -1.13075364e+00 7.76382685e-01
1.06124139e+00 5.63938618e-01 2.06088066e-01 8.60840559e-01
-7.58111775e-01 1.15868604e+00 -9.31194782e-01 -1.01111138e+00
-2.46014267e-01 -7.13365197e-01 -1.77629089e+00 7.23124444e-02
2.52194643e-01 -2.42465854e-01 -1.76252246e-01 6.58233702e-01
-7.24592656e-02 3.45664352e-01 4.93357152e-01 -1.07500410e+00
-6.09351456e-01 6.34656429e-01 2.21499920e-01 3.32187235e-01
1.22053981e+00 3.28345150e-02 1.31494153e+00 -6.26530409e-01
5.01488626e-01 1.51390731e+00 2.09358141e-01 8.50055039e-01
-1.14309633e+00 -3.38184536e-01 1.77575693e-01 4.11095351e-01
-7.73116410e-01 -5.07855177e-01 6.13693178e-01 -2.30375081e-01
1.32879126e+00 6.67896152e-01 3.67628008e-01 7.59703159e-01
1.91644147e-01 6.25912070e-01 9.78604972e-01 -6.34730041e-01
-2.79826415e-03 3.21255088e-01 6.90570712e-01 5.38057268e-01
-3.68738085e-01 -4.30454373e-01 -2.97812253e-01 -5.73246777e-01
2.83625275e-01 -3.54009271e-01 -5.02221763e-01 2.66683370e-01
-8.48482907e-01 1.03014171e+00 2.30331823e-01 3.71276855e-01
-8.15999508e-03 -2.40563452e-01 3.58049959e-01 8.98919821e-01
1.87806651e-01 1.05036700e+00 -6.66214526e-01 -4.44792777e-01
3.18387784e-02 5.66197693e-01 1.51536620e+00 1.03550601e+00
8.28995228e-01 -8.38755310e-01 -1.76987395e-01 1.06382871e+00
9.38333198e-02 4.04465586e-01 3.70916039e-01 -1.35823941e+00
7.02984095e-01 9.34789598e-01 4.93011534e-01 -8.39454830e-01
-2.30880126e-01 3.68104577e-01 -3.62191886e-01 -6.73176348e-01
7.28203297e-01 -2.23672926e-01 -2.93812811e-01 1.77844834e+00
3.47847909e-01 -2.81141281e-01 5.08913875e-01 4.95685875e-01
1.24680257e+00 8.31447780e-01 -8.86150822e-02 -3.91698718e-01
1.84587514e+00 -1.16090453e+00 -1.01655936e+00 -3.12204450e-01
9.36583579e-01 -1.05606008e+00 1.33275652e+00 5.75422309e-02
-1.07547593e+00 -4.27723974e-01 -4.07030493e-01 -4.84477133e-01
-1.28988743e-01 -3.25359076e-01 3.50700319e-01 3.62607151e-01
-8.23460758e-01 2.33671769e-01 -2.55168319e-01 -4.20967489e-01
-3.97339374e-01 1.98464677e-01 -2.50663757e-01 -2.02625662e-01
-1.51365042e+00 1.08260548e+00 -1.35659337e-01 1.71097323e-01
-3.74603122e-01 -3.11418593e-01 -9.30329025e-01 2.44172998e-02
6.37625515e-01 -6.97946727e-01 1.91958117e+00 -9.61246610e-01
-1.59142625e+00 1.00205600e+00 -7.42256761e-01 -3.31499130e-01
7.14304578e-03 -4.16898847e-01 -3.18206608e-01 1.88669145e-01
1.40688196e-01 2.36697257e-01 4.87535596e-01 -9.37027216e-01
-6.45469546e-01 -2.30605632e-01 1.11237979e+00 4.28652495e-01
1.93799138e-01 2.03868553e-01 -2.85133660e-01 -2.13864416e-01
1.09950139e-03 -1.14224517e+00 -9.39886495e-02 -5.65111697e-01
-1.88473269e-01 -8.86501193e-01 5.45679271e-01 -9.34039712e-01
1.29735565e+00 -1.82383847e+00 2.80179471e-01 -1.45968243e-01
1.98648095e-01 2.84990758e-01 -3.02856714e-01 7.86770999e-01
-2.83520063e-03 2.92198807e-01 -1.64482847e-01 -1.53902367e-01
-8.52595419e-02 6.10032618e-01 -5.70447147e-01 -3.19184273e-01
2.49114007e-01 1.18086827e+00 -9.13933396e-01 -2.75001973e-01
-1.30668119e-01 -2.56407946e-01 -5.44908822e-01 1.01397574e+00
-6.41563594e-01 4.34339106e-01 -4.99347448e-01 1.61371604e-01
1.66963995e-01 -2.32206285e-01 3.11184824e-01 7.72219151e-02
2.03791007e-01 1.14790142e+00 -5.69832087e-01 1.32997668e+00
-7.99194574e-01 5.80620468e-01 1.58820689e-01 -7.36658335e-01
8.73909593e-01 4.61007446e-01 -3.62278163e-01 -6.63844705e-01
-4.16857824e-02 1.45473272e-01 2.29301974e-01 -8.78691614e-01
5.89507937e-01 -2.34554753e-01 -3.43935370e-01 8.57100606e-01
-1.51341394e-01 -5.83465159e-01 2.89717108e-01 5.42336702e-01
1.11958039e+00 -5.65370381e-01 5.37747622e-01 -1.87096164e-01
1.06516373e+00 9.52722952e-02 3.12910199e-01 7.64287531e-01
-1.82543606e-01 1.47282079e-01 8.81560385e-01 -2.62288004e-01
-6.41119540e-01 -6.97121501e-01 1.79643139e-01 1.32617939e+00
1.90692440e-01 -6.72586441e-01 -8.35436761e-01 -9.90837514e-01
-1.94497645e-01 1.01109731e+00 -5.09006202e-01 -6.39583990e-02
-1.13662302e+00 7.38044381e-02 5.64405382e-01 2.49209031e-01
4.50076878e-01 -1.25595534e+00 -2.59762853e-01 4.10700619e-01
-1.00309682e+00 -1.19978476e+00 -6.97036862e-01 -6.56897873e-02
-6.80445373e-01 -1.49002099e+00 -1.75788507e-01 -8.85660827e-01
4.24681455e-01 4.45962727e-01 1.61158752e+00 6.39634848e-01
4.61439997e-01 7.33794928e-01 -8.32620203e-01 -2.04755858e-01
-9.12677169e-01 3.03097129e-01 -3.74249250e-01 -2.93293744e-01
7.12253213e-01 -5.25141597e-01 -3.78023803e-01 6.81331635e-01
-8.63594472e-01 -6.75992444e-02 4.07605655e-02 9.90500271e-01
-8.34149495e-03 -6.95772827e-01 8.94598007e-01 -1.27786183e+00
1.39194751e+00 -3.05850863e-01 -3.18808138e-01 7.58439600e-01
-1.11043662e-01 1.71025455e-01 8.10367942e-01 -1.83647156e-01
-1.20721591e+00 -4.46323037e-01 -4.68371630e-01 2.98294276e-01
-2.66118735e-01 5.82905412e-01 -2.68255293e-01 2.04307377e-01
7.22934484e-01 1.56713605e-01 4.38483022e-02 -2.64145762e-01
5.52950442e-01 9.97409821e-01 2.44993120e-01 -8.59551251e-01
5.64580023e-01 -1.86705768e-01 -6.60908222e-01 -1.00625980e+00
-1.10866559e+00 -7.39814103e-01 -3.54475260e-01 -1.20954603e-01
8.44836950e-01 -5.09777844e-01 -9.41507041e-01 1.78191707e-01
-1.55813968e+00 -3.80111933e-01 -1.58187568e-01 -1.70168392e-02
-4.68363166e-01 7.19385207e-01 -7.38024652e-01 -7.46479392e-01
-4.18985248e-01 -1.02646983e+00 6.57270372e-01 3.54741871e-01
-8.53431582e-01 -1.07556188e+00 1.30434170e-01 1.07382035e+00
2.24963248e-01 -3.73135418e-01 1.44136071e+00 -1.05628276e+00
-3.05719078e-01 1.47062585e-01 4.77639697e-02 5.16267061e-01
4.11086470e-01 -3.59074175e-01 -7.35521019e-01 -4.50490154e-02
3.37430507e-01 -6.66466236e-01 2.60688484e-01 -4.10363853e-01
8.54421318e-01 -7.02953517e-01 -2.79933084e-02 -2.06780612e-01
6.55424833e-01 4.38294262e-01 7.79876709e-01 1.49071333e-03
1.88940823e-01 1.05713809e+00 8.35622013e-01 -8.78958106e-02
8.67842138e-01 6.31929934e-01 1.02241151e-01 5.05930007e-01
1.70472458e-01 -1.90219447e-01 2.11303458e-01 1.20253026e+00
3.19390029e-01 -3.19198787e-01 -9.73862231e-01 6.52327001e-01
-1.78599644e+00 -8.16939652e-01 -5.28754354e-01 1.73874044e+00
1.28843617e+00 -7.09380209e-02 -1.94831640e-01 -1.47836939e-01
5.88219821e-01 1.94081619e-01 -2.82868564e-01 -9.77688670e-01
2.18770206e-01 3.89262974e-01 -4.50036913e-01 1.17862427e+00
-5.72095752e-01 1.02000630e+00 6.10760593e+00 3.48004997e-01
-4.72539365e-01 -1.02018295e-02 3.58671308e-01 4.78263855e-01
-6.25924051e-01 3.27092767e-01 -5.76715767e-01 1.19835839e-01
8.66670430e-01 -3.50301623e-01 6.04329765e-01 5.43232024e-01
6.94204168e-03 -4.10179883e-01 -1.52818370e+00 6.57590449e-01
1.38900325e-01 -1.15299737e+00 2.54255593e-01 -3.96216035e-01
4.29724604e-01 -5.46128213e-01 -5.35050333e-01 8.26014221e-01
2.86302507e-01 -1.04627621e+00 9.43483040e-03 3.63361597e-01
3.39819640e-01 -6.46417558e-01 1.04424405e+00 8.22031617e-01
-9.81619418e-01 -3.25680226e-02 -2.39457354e-01 -4.39578801e-01
2.50433207e-01 1.19606592e-01 -1.05347562e+00 5.53891778e-01
4.43268508e-01 5.37217744e-02 -3.69809985e-01 3.92493486e-01
-5.74080706e-01 7.80478716e-01 -2.90709883e-01 -5.23421586e-01
1.43809438e-01 -3.58303219e-01 4.92992520e-01 1.07358587e+00
-1.99726462e-01 1.00546336e+00 3.39816250e-02 4.76440191e-01
-2.77417630e-01 2.96502501e-01 -3.74979883e-01 1.85357742e-02
5.95420957e-01 9.91527617e-01 7.17355346e-04 -5.01100242e-01
-4.28076714e-01 9.49760675e-01 5.29774308e-01 3.40495527e-01
-5.18029511e-01 -5.68729997e-01 6.00435853e-01 1.51845858e-01
1.04508577e-02 -1.78982183e-01 1.51121184e-01 -1.35036612e+00
3.44999194e-01 -1.49906778e+00 4.10230637e-01 -8.79861593e-01
-1.48241329e+00 7.46652246e-01 -2.30608396e-02 -5.76420546e-01
-6.69546902e-01 -4.46073800e-01 -9.61530805e-01 1.12247241e+00
-1.54529130e+00 -8.11313450e-01 -3.65856916e-01 5.83481848e-01
7.71925747e-01 2.01871932e-01 1.12093234e+00 5.66493981e-02
-2.18396619e-01 5.00754952e-01 -4.30304646e-01 2.55658388e-01
9.58346069e-01 -1.19852412e+00 3.99707019e-01 3.43634218e-01
5.37451233e-05 9.85478282e-01 8.71984184e-01 -3.63475859e-01
-1.20729876e+00 -8.03835213e-01 1.62337065e+00 -9.22966838e-01
6.94618404e-01 -3.74780148e-01 -1.46942365e+00 9.27782476e-01
6.30924642e-01 -8.07527483e-01 1.13417327e+00 4.56732780e-01
-2.95444131e-01 1.51806287e-02 -9.33373868e-01 7.19009995e-01
7.95782328e-01 -9.93289232e-01 -1.80292606e+00 4.18766707e-01
1.19704020e+00 -4.28688854e-01 -7.62709975e-01 1.78706989e-01
1.40959650e-01 -7.94859707e-01 4.71821189e-01 -1.03352833e+00
5.34015417e-01 -4.13530320e-02 -8.37692767e-02 -1.30136013e+00
6.44516051e-02 -7.55947292e-01 -3.31591628e-02 1.38036692e+00
6.72295630e-01 -6.86430871e-01 3.79219353e-01 1.12980568e+00
-1.00989483e-01 -7.57844150e-01 -1.01419938e+00 -2.61552364e-01
3.00504625e-01 -2.22927347e-01 6.64549053e-01 9.45707202e-01
6.66304529e-01 1.07150781e+00 6.51075542e-02 -1.01425163e-01
-1.32333845e-01 4.55302835e-01 1.11704898e+00 -9.75873888e-01
-3.63530874e-01 -8.26548785e-02 2.50522971e-01 -1.81543851e+00
5.11716604e-01 -7.29723155e-01 1.18502863e-01 -1.60022748e+00
-1.32036433e-01 -2.99898505e-01 6.06672585e-01 1.91525981e-01
-4.62092429e-01 -4.46609408e-01 1.52157381e-01 2.02066943e-01
-5.46054661e-01 6.35470092e-01 1.41533399e+00 -3.49727161e-02
-2.51479656e-01 2.31466562e-01 -8.12996745e-01 8.00940216e-01
7.73275554e-01 -2.86313951e-01 -3.91823441e-01 -4.45372701e-01
3.32516223e-01 6.87777102e-01 8.65599960e-02 -1.90505669e-01
2.40508571e-01 -3.71396840e-01 -5.67227900e-01 -3.29426765e-01
3.37186217e-01 -6.18289709e-01 -3.72094601e-01 1.67701617e-01
-7.15431154e-01 3.28844428e-01 1.00604579e-01 3.81445467e-01
-5.87489247e-01 -6.98039830e-01 3.79160851e-01 -3.56893659e-01
-3.69285673e-01 -3.64987165e-01 -7.43228316e-01 6.22396231e-01
7.66882658e-01 1.00512274e-01 -4.77146596e-01 -1.03678131e+00
-7.76039004e-01 8.12237024e-01 1.37676731e-01 5.22998989e-01
5.44773996e-01 -9.07002211e-01 -7.02776134e-01 -1.81787774e-01
3.36740106e-01 1.46266475e-01 1.33858815e-01 3.40641320e-01
-3.99183333e-01 6.80690825e-01 2.72519320e-01 -4.76334900e-01
-1.56030619e+00 2.45228872e-01 6.28188550e-01 -6.57158256e-01
-9.13131163e-02 8.06918919e-01 3.53824258e-01 -9.18017507e-01
9.12465602e-02 -6.05156839e-01 -6.39184058e-01 -1.41348811e-02
6.79821491e-01 -3.79592068e-02 2.04890780e-02 -4.04590905e-01
-2.40500763e-01 3.94590259e-01 -3.86710972e-01 -1.78995416e-01
8.17065954e-01 -5.07289469e-01 -5.93196332e-01 5.16979694e-01
1.09856582e+00 1.40639216e-01 -3.38448882e-01 -5.54613650e-01
1.66700184e-01 -3.83255512e-01 -8.49783719e-01 -9.06143665e-01
-2.40861773e-01 7.94664085e-01 -2.95997679e-01 7.95534849e-01
7.46132493e-01 5.11297166e-01 9.14948046e-01 1.28644705e+00
3.70002180e-01 -7.77523875e-01 3.02667320e-01 1.17537796e+00
1.48796332e+00 -1.16880238e+00 -4.41714197e-01 -7.78083444e-01
-6.73374712e-01 1.12003553e+00 8.89356613e-01 1.13035910e-01
3.02434713e-01 -2.47078866e-01 4.56289470e-01 -2.40863144e-01
-1.22795880e+00 3.72892134e-02 1.57193705e-01 3.10186684e-01
6.95223033e-01 -1.37637276e-02 -5.78435838e-01 8.74750912e-01
-6.34235144e-01 -3.58837426e-01 8.18633854e-01 8.47512662e-01
-4.76297677e-01 -1.44036329e+00 -2.80780286e-01 4.19385910e-01
-3.43965352e-01 -7.67824575e-02 -1.06029916e+00 5.68122387e-01
-4.07300889e-01 1.77858198e+00 -6.93433732e-02 -1.85523123e-01
8.33455861e-01 4.47125643e-01 2.73818761e-01 -1.01291275e+00
-1.25578654e+00 -6.06050193e-01 8.52328002e-01 -4.65024322e-01
-4.05768245e-01 -4.28287238e-01 -1.23868442e+00 -3.54117990e-01
-7.25520253e-01 8.66115928e-01 8.18838403e-02 1.41935778e+00
2.17189819e-01 1.18972979e-01 6.75987720e-01 1.57815963e-01
-7.75311232e-01 -1.16136050e+00 1.92874782e-02 7.63574719e-01
3.52070719e-01 -1.48248121e-01 -5.97579360e-01 -5.45248948e-03] | [11.948186874389648, 8.01573371887207] |
bde7d2e1-dae3-4a57-b3b7-2a3f786da28b | scene-graph-expansion-for-semantics-guided | 2205.02958 | null | https://arxiv.org/abs/2205.02958v1 | https://arxiv.org/pdf/2205.02958v1.pdf | Scene Graph Expansion for Semantics-Guided Image Outpainting | In this paper, we address the task of semantics-guided image outpainting, which is to complete an image by generating semantically practical content. Different from most existing image outpainting works, we approach the above task by understanding and completing image semantics at the scene graph level. In particular, we propose a novel network of Scene Graph Transformer (SGT), which is designed to take node and edge features as inputs for modeling the associated structural information. To better understand and process graph-based inputs, our SGT uniquely performs feature attention at both node and edge levels. While the former views edges as relationship regularization, the latter observes the co-occurrence of nodes for guiding the attention process. We demonstrate that, given a partial input image with its layout and scene graph, our SGT can be applied for scene graph expansion and its conversion to a complete layout. Following state-of-the-art layout-to-image conversions works, the task of image outpainting can be completed with sufficient and practical semantics introduced. Extensive experiments are conducted on the datasets of MS-COCO and Visual Genome, which quantitatively and qualitatively confirm the effectiveness of our proposed SGT and outpainting frameworks. | ['Yu-Chiang Frank Wang', 'Meng-Lin Wu', 'Cheng-Fu Yang', 'Wan-Cyuan Fan', 'Cheng-Yo Tan', 'Chiao-An Yang'] | 2022-05-05 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Yang_Scene_Graph_Expansion_for_Semantics-Guided_Image_Outpainting_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Yang_Scene_Graph_Expansion_for_Semantics-Guided_Image_Outpainting_CVPR_2022_paper.pdf | cvpr-2022-1 | ['image-outpainting'] | ['computer-vision'] | [ 6.82713628e-01 4.54811364e-01 -7.96133280e-02 -2.84727007e-01
-2.15841070e-01 -3.87104958e-01 4.88256454e-01 1.85785741e-01
6.28513619e-02 2.96706617e-01 2.97620863e-01 -2.29218572e-01
2.11714476e-01 -9.09263372e-01 -1.16853738e+00 -2.12929085e-01
4.26824003e-01 2.64248520e-01 -4.53575365e-02 -2.20834866e-01
2.37349585e-01 3.72401506e-01 -1.27043593e+00 2.69500077e-01
1.01683438e+00 7.42245138e-01 6.97854340e-01 5.58195651e-01
-4.95016724e-01 1.14594007e+00 -3.72710437e-01 -3.74922901e-01
1.60640761e-01 -5.96457958e-01 -8.32757533e-01 7.89364159e-01
5.76793253e-01 -4.72626120e-01 -6.63220227e-01 1.23963487e+00
7.60087222e-02 9.03903693e-02 3.85945886e-01 -1.51406002e+00
-1.26970065e+00 7.48023212e-01 -8.32543075e-01 -3.87273729e-01
5.17651260e-01 2.17749208e-01 9.18253601e-01 -8.99347425e-01
9.34968650e-01 1.48354936e+00 1.85460567e-01 4.01001245e-01
-1.28485548e+00 -4.48760659e-01 4.89073694e-01 8.12868997e-02
-1.27594793e+00 -8.49555731e-02 1.31471705e+00 -3.68129671e-01
7.55729139e-01 9.72544700e-02 9.24268603e-01 7.38826036e-01
-8.41637254e-02 8.87272239e-01 8.59110594e-01 -5.27576327e-01
1.51292682e-01 -1.64595366e-01 -1.78159863e-01 8.47598553e-01
3.26012857e-02 -6.38836250e-02 -5.23676515e-01 1.79747596e-01
1.07120562e+00 2.71627575e-01 -3.27831745e-01 -5.13164163e-01
-1.13214064e+00 6.47761822e-01 7.35941172e-01 2.12140717e-02
-3.87941152e-01 4.60701823e-01 2.64726937e-01 1.58296272e-01
5.60394526e-01 3.73197138e-01 -9.19394661e-03 3.62291515e-01
-8.67561698e-01 1.87605634e-01 5.22792697e-01 1.65361369e+00
9.60968912e-01 8.32931846e-02 -4.05624837e-01 7.22690403e-01
6.68732310e-03 3.68447751e-01 -6.39778748e-02 -1.13513935e+00
5.14326632e-01 8.98880541e-01 3.52414735e-02 -1.26237774e+00
4.87859510e-02 -2.48096868e-01 -1.03099585e+00 -2.30519429e-01
-1.41373783e-01 2.25585505e-01 -9.95780826e-01 1.70936334e+00
1.68711439e-01 3.69687349e-01 -1.21930666e-01 8.53356004e-01
9.81391013e-01 8.10412049e-01 2.02212334e-01 -1.01794958e-01
1.30314982e+00 -1.45749545e+00 -8.71736109e-01 -5.47480762e-01
3.71642500e-01 -6.91039264e-01 1.42450500e+00 3.75284143e-02
-1.27189076e+00 -7.53156006e-01 -9.32911634e-01 -6.92754269e-01
-1.92415193e-01 6.98483437e-02 6.66482449e-01 -5.51860444e-02
-1.18767691e+00 4.16369170e-01 -4.10215765e-01 -3.93638760e-01
5.96785963e-01 -1.21041261e-01 -3.63858134e-01 -5.53764403e-01
-8.28854680e-01 4.81593698e-01 4.69324052e-01 1.44589871e-01
-9.44245398e-01 -8.14950466e-01 -1.20843184e+00 3.41023207e-01
6.58956230e-01 -9.85182703e-01 1.01521385e+00 -1.04213023e+00
-1.02213216e+00 8.20216954e-01 -2.16951385e-01 -2.86756426e-01
3.28321278e-01 -1.99058816e-01 -2.57235318e-02 2.36618072e-01
3.98064166e-01 1.13867915e+00 9.35866535e-01 -1.58430064e+00
-4.70029593e-01 -1.44416466e-01 2.73755759e-01 3.73088598e-01
-3.61896813e-01 -2.57863581e-01 -1.13758886e+00 -9.43400800e-01
1.09861173e-01 -5.81427932e-01 -2.83930093e-01 3.31546336e-01
-7.47323692e-01 5.86059801e-02 1.04286230e+00 -9.15083170e-01
1.16928041e+00 -2.31471586e+00 4.60275650e-01 4.36232686e-02
5.31543493e-01 -1.00225858e-01 -5.57499766e-01 6.35349929e-01
-1.42203972e-01 2.01105878e-01 -5.50146461e-01 -6.90663338e-01
-1.51349127e-01 3.66394579e-01 -5.85853696e-01 1.18921816e-01
3.22059900e-01 1.32429361e+00 -1.10399485e+00 -5.09394944e-01
5.09323001e-01 4.55489069e-01 -8.22922945e-01 5.05290389e-01
-4.57181126e-01 4.59688842e-01 -4.80536729e-01 5.27905226e-01
8.32109272e-01 -3.58777702e-01 1.10237338e-01 -5.94138801e-01
2.21083388e-01 -1.57628000e-01 -7.30326176e-01 2.16961384e+00
-6.79743052e-01 6.65742695e-01 -1.83372200e-02 -9.67099547e-01
9.11967218e-01 -3.07093799e-01 2.76879460e-01 -8.55750680e-01
-2.04393063e-02 -1.76243931e-01 -5.47647595e-01 -4.66999590e-01
8.22512090e-01 1.57743469e-01 -3.71751189e-02 3.42761129e-01
3.58702280e-02 -4.58206505e-01 1.75726593e-01 7.56717861e-01
7.86326587e-01 4.35142964e-01 1.13353714e-01 -1.61622465e-01
3.85059446e-01 5.43829203e-02 2.56393552e-01 5.86247385e-01
2.26333812e-01 8.34651947e-01 6.18197620e-01 -2.86343902e-01
-1.25043070e+00 -9.39843714e-01 5.54068685e-01 7.54090786e-01
5.89677751e-01 -7.15445161e-01 -1.09623921e+00 -5.46479881e-01
-1.00133210e-01 9.38562095e-01 -5.79533100e-01 -3.19431573e-01
-4.72163439e-01 2.11882461e-02 1.14916109e-01 5.42943478e-01
7.65205443e-01 -1.24210215e+00 -3.81842017e-01 2.42189303e-01
-5.27637541e-01 -1.52263057e+00 -1.13527298e+00 -3.06606978e-01
-7.69858718e-01 -1.07307017e+00 -6.55895710e-01 -1.22226453e+00
1.26623213e+00 6.77973330e-01 1.25213683e+00 4.94022429e-01
-3.73776823e-01 4.69427317e-01 -3.57481062e-01 4.76077246e-03
-5.79549670e-01 -1.06057502e-01 -6.39851034e-01 9.97945890e-02
-2.55884111e-01 -8.18702221e-01 -6.16381884e-01 -1.00351393e-01
-1.43313658e+00 9.32853580e-01 6.27421618e-01 7.72183597e-01
8.31964791e-01 1.68852046e-01 2.40083605e-01 -1.31091309e+00
6.00957036e-01 -2.44924292e-01 -4.76889342e-01 4.58248883e-01
-4.75562662e-01 1.85929760e-01 8.25920045e-01 -1.37993813e-01
-1.07637537e+00 1.39786273e-01 6.56796917e-02 -8.46545339e-01
-2.21955068e-02 5.34103811e-01 -5.47108293e-01 5.41805886e-02
1.90030292e-01 5.50821543e-01 -7.50076249e-02 -3.74764472e-01
9.31433558e-01 2.55195886e-01 8.49349797e-01 -6.06332600e-01
8.62949312e-01 5.61411679e-01 -7.85488542e-03 -7.01871097e-01
-8.34976494e-01 -2.10996225e-01 -6.12748563e-01 -2.30709970e-01
8.80375922e-01 -9.35643196e-01 -3.61925662e-01 5.04323483e-01
-1.36942601e+00 -4.86188293e-01 -5.58325410e-01 -3.89557838e-01
-7.01013625e-01 7.33183920e-01 -5.20729542e-01 -4.20324475e-01
-2.14604020e-01 -1.13239455e+00 1.48577106e+00 1.22112565e-01
1.33223340e-01 -1.11454821e+00 -4.19848204e-01 2.93500602e-01
2.02657223e-01 3.95955265e-01 1.22076368e+00 1.42882034e-01
-9.27727759e-01 1.75462589e-01 -6.52655005e-01 2.83320397e-01
2.60511369e-01 -9.70233679e-02 -5.98773658e-01 -2.69687831e-01
-1.33974269e-01 -2.63439804e-01 8.98745179e-01 3.61187100e-01
1.51214302e+00 -4.61065918e-01 -2.75318801e-01 9.24645960e-01
1.58555377e+00 6.85774460e-02 8.38984728e-01 9.00601670e-02
1.27188599e+00 7.91178584e-01 6.73879981e-01 3.75971496e-01
5.87001979e-01 5.42232454e-01 6.62843406e-01 -6.18044853e-01
-5.65747976e-01 -1.11671543e+00 9.42936391e-02 7.30568111e-01
4.36870009e-01 -3.98907363e-01 -4.83356297e-01 5.65930903e-01
-1.97431588e+00 -6.89134419e-01 -4.88194264e-02 1.68522072e+00
6.56853557e-01 -1.87425092e-01 -2.62087077e-01 -2.72574425e-01
9.13311839e-01 4.29264575e-01 -6.26371026e-01 -1.69586375e-01
-2.01869607e-01 -5.93598280e-03 4.42357421e-01 6.82416379e-01
-8.26702893e-01 1.40368378e+00 5.58894539e+00 1.01671350e+00
-8.16366971e-01 -2.69480646e-02 1.00516725e+00 2.65193254e-01
-7.62513220e-01 4.00473893e-01 -2.36350179e-01 3.44247460e-01
9.67114419e-02 -1.35021597e-01 9.58583117e-01 7.93296278e-01
2.34845191e-01 -7.72428215e-02 -1.07538164e+00 1.15638316e+00
2.11470619e-01 -1.62129951e+00 7.79199898e-01 -1.65626273e-01
9.06106830e-01 -6.25202000e-01 -8.74860063e-02 8.52854028e-02
2.32653618e-01 -1.03727734e+00 1.15336335e+00 5.12806296e-01
1.12480307e+00 -8.17967415e-01 1.91934928e-01 2.24624887e-01
-1.43614161e+00 6.48164228e-02 -3.27109247e-01 4.26504351e-02
3.90883058e-01 6.08290315e-01 -3.69939864e-01 7.57429183e-01
4.46693599e-01 1.17613876e+00 -7.48339891e-01 6.11888647e-01
-6.11357212e-01 2.30877802e-01 -3.86402160e-02 3.62300783e-01
1.96614116e-01 -4.59984809e-01 3.84867787e-01 8.92411113e-01
2.14335725e-01 7.68721625e-02 3.35995764e-01 1.54880095e+00
-3.68103385e-01 3.52650844e-02 -7.94898629e-01 -1.60635233e-01
5.71181536e-01 1.27567208e+00 -9.24707234e-01 -4.67470825e-01
-2.76576847e-01 1.52175963e+00 5.02025843e-01 5.93071699e-01
-9.85492408e-01 -3.29392254e-01 2.41581231e-01 2.46216252e-01
2.33346879e-01 -1.64940849e-01 -3.58962506e-01 -1.17892742e+00
1.89556971e-01 -6.37461960e-01 4.33164835e-02 -1.36449695e+00
-1.04610670e+00 5.83505094e-01 1.60328805e-01 -9.63422239e-01
1.50850117e-01 -1.99148834e-01 -7.04804718e-01 6.62137270e-01
-1.51461840e+00 -1.62707055e+00 -6.80529833e-01 5.87382078e-01
8.08701217e-01 1.85531542e-01 3.11798632e-01 1.73080057e-01
-3.33800197e-01 3.86670887e-01 -3.49896312e-01 1.19153067e-01
3.03689778e-01 -1.00571048e+00 8.75066102e-01 1.13332093e+00
2.23438367e-01 5.56277156e-01 5.73950887e-01 -9.03920889e-01
-1.61694324e+00 -1.62237799e+00 5.87305129e-01 -5.91135956e-02
3.36474389e-01 -5.87748587e-01 -6.77516937e-01 7.64390290e-01
3.95994574e-01 4.85181715e-03 -3.85008603e-02 -5.24986088e-01
-3.35295230e-01 1.62910998e-01 -9.48250830e-01 8.92343700e-01
1.64777219e+00 -6.46034598e-01 -2.58895338e-01 4.41141605e-01
1.32452071e+00 -4.15849090e-01 -5.82472205e-01 1.95435002e-01
1.04814082e-01 -8.63384366e-01 1.05341446e+00 -4.41288024e-01
9.27474260e-01 -4.31023180e-01 -1.57510579e-01 -1.35094333e+00
-3.71921510e-01 -8.06396723e-01 9.08692628e-02 1.46864545e+00
-5.59556647e-04 -2.29873061e-01 6.02763057e-01 3.52707684e-01
-3.85600656e-01 -6.33697093e-01 -4.39646274e-01 -5.22049367e-01
-3.41763586e-01 -3.73457223e-01 7.29618013e-01 8.35861504e-01
-3.21532458e-01 4.02855664e-01 -6.22421324e-01 -5.15493564e-02
7.73865283e-01 5.26406765e-01 8.28248978e-01 -6.53581560e-01
-4.18320209e-01 -3.12298119e-01 -1.40442982e-01 -1.45949948e+00
2.16943935e-01 -9.42688346e-01 1.98878925e-02 -1.90623283e+00
4.56797779e-01 -1.43857598e-01 2.49899089e-01 4.94974971e-01
-2.29719505e-01 1.77231967e-01 5.32438457e-01 4.94371541e-02
-5.72074294e-01 8.47970068e-01 1.79338634e+00 -4.51111257e-01
-7.80718923e-02 -6.91091001e-01 -1.14063525e+00 4.94986773e-01
4.32622939e-01 -2.82057077e-01 -8.88498306e-01 -6.69267654e-01
3.00014049e-01 3.10184032e-01 6.44762814e-01 -5.88737726e-01
1.94352791e-01 -3.43981713e-01 3.02365385e-02 -6.02285445e-01
8.21400359e-02 -8.03151190e-01 2.82961428e-01 2.90960073e-01
-4.67265636e-01 1.04085475e-01 1.71316162e-01 7.97464013e-01
-3.28633279e-01 4.45101038e-02 5.46678960e-01 -2.15169623e-01
-9.96734679e-01 6.28697813e-01 4.75721173e-02 1.94875374e-01
9.58989918e-01 -3.12631369e-01 -3.79588455e-01 -6.55446529e-01
-6.79616272e-01 3.32783133e-01 8.58904481e-01 5.75859964e-01
1.04008687e+00 -1.53870940e+00 -5.15430212e-01 4.25366461e-01
3.60662252e-01 4.58981186e-01 4.97685879e-01 4.70137417e-01
-7.60591149e-01 1.15235113e-01 -2.57120907e-01 -5.18383324e-01
-9.97114956e-01 9.38140929e-01 8.51170868e-02 -1.38345942e-01
-7.83423543e-01 6.22300565e-01 9.86295104e-01 -1.99868813e-01
1.35198578e-01 -5.59664905e-01 2.04316542e-01 -4.39584941e-01
2.07043469e-01 -4.34115604e-02 -3.38176608e-01 -4.89484608e-01
5.63324504e-02 5.15024245e-01 -1.37843624e-01 -2.37279776e-02
1.26355028e+00 -3.84012163e-01 -4.53069419e-01 -1.08295128e-01
1.22377169e+00 -2.41082534e-01 -1.58289516e+00 -1.41761065e-01
-3.34769547e-01 -6.31364882e-01 -8.96937959e-03 -5.13833523e-01
-1.41570818e+00 9.20237362e-01 6.03840314e-02 1.95996147e-02
1.40210581e+00 1.21066324e-01 1.02929389e+00 3.79510149e-02
3.50984484e-01 -8.58956754e-01 3.98226917e-01 1.94179401e-01
1.25411153e+00 -8.68124366e-01 -8.43422413e-02 -1.02145994e+00
-6.20018840e-01 8.68862092e-01 7.57343650e-01 -4.12323862e-01
4.50071156e-01 2.60929793e-01 -5.22551656e-01 -2.82678485e-01
-5.10730982e-01 -7.87850842e-02 2.15207547e-01 8.53618383e-01
7.18510300e-02 3.61379273e-02 -7.12811798e-02 4.10595238e-01
-3.07320505e-02 -2.71578319e-02 6.19153976e-01 8.67479146e-01
-1.88871011e-01 -1.15914810e+00 4.38911747e-03 2.68303186e-01
2.01200873e-01 -3.84750158e-01 -5.60817897e-01 7.48715818e-01
-1.10939760e-02 7.52048194e-01 3.87680456e-02 -3.10930908e-01
3.87582779e-01 -3.43637347e-01 5.34613311e-01 -8.22556734e-01
-1.71104655e-01 1.87109560e-01 -9.41327065e-02 -8.27290118e-01
-4.12253290e-01 -2.02780321e-01 -1.29774177e+00 -2.48013198e-01
5.90938283e-03 -2.16632932e-01 3.30810100e-01 6.90828800e-01
6.30645275e-01 9.04089928e-01 5.35600662e-01 -8.47590387e-01
-6.03322312e-02 -6.19198501e-01 -7.36175239e-01 8.92511070e-01
1.38795227e-01 -4.25701559e-01 3.85557674e-03 5.23485601e-01] | [11.408853530883789, -0.48772168159484863] |
5b397907-a5f8-49e2-bf82-bec0c7b77e01 | unpaired-photo-to-caricature-translation-on | 1711.10735 | null | http://arxiv.org/abs/1711.10735v2 | http://arxiv.org/pdf/1711.10735v2.pdf | Unpaired Photo-to-Caricature Translation on Faces in the Wild | Recently, image-to-image translation has been made much progress owing to the
success of conditional Generative Adversarial Networks (cGANs). And some
unpaired methods based on cycle consistency loss such as DualGAN, CycleGAN and
DiscoGAN are really popular. However, it's still very challenging for
translation tasks with the requirement of high-level visual information
conversion, such as photo-to-caricature translation that requires satire,
exaggeration, lifelikeness and artistry. We present an approach for learning to
translate faces in the wild from the source photo domain to the target
caricature domain with different styles, which can also be used for other
high-level image-to-image translation tasks. In order to capture global
structure with local statistics while translation, we design a dual pathway
model with one coarse discriminator and one fine discriminator. For generator,
we provide one extra perceptual loss in association with adversarial loss and
cycle consistency loss to achieve representation learning for two different
domains. Also the style can be learned by the auxiliary noise input.
Experiments on photo-to-caricature translation of faces in the wild show
considerable performance gain of our proposed method over state-of-the-art
translation methods as well as its potential real applications. | ['Haiyong Zheng', 'Wang Chao', 'Nan Wang', 'Ziqiang Zheng', 'Zhibin Yu', 'Bing Zheng'] | 2017-11-29 | null | null | null | null | ['photo-to-caricature-translation', 'caricature'] | ['computer-vision', 'computer-vision'] | [ 5.01240671e-01 7.58468509e-02 1.12226591e-01 -3.07505786e-01
-6.23300970e-01 -6.09103143e-01 7.40823090e-01 -8.05881262e-01
4.77976389e-02 9.69290793e-01 4.15274454e-03 7.16830269e-02
4.44742292e-01 -9.28425491e-01 -1.00393093e+00 -9.12400663e-01
4.47921634e-01 2.80490607e-01 -1.23593494e-01 -3.48496228e-01
-1.10534780e-01 4.13031906e-01 -1.17090917e+00 3.77789319e-01
7.53334045e-01 1.01899338e+00 -6.93142563e-02 4.87396777e-01
-8.38865116e-02 5.52039802e-01 -5.70521951e-01 -9.09466743e-01
5.41562080e-01 -1.09096444e+00 -4.40316796e-01 5.43171875e-02
6.12753212e-01 -2.52065033e-01 -2.77930826e-01 1.25738478e+00
7.25669324e-01 -3.01511288e-01 8.21314096e-01 -1.55269921e+00
-1.34715295e+00 1.97988570e-01 -9.00764644e-01 -3.61937523e-01
3.73108387e-01 4.46223438e-01 4.32621568e-01 -7.49268651e-01
7.64173806e-01 1.84595585e+00 5.73527277e-01 8.47448230e-01
-1.37413645e+00 -1.06227779e+00 -2.67962754e-01 7.15155229e-02
-1.23402131e+00 -4.06954169e-01 9.90861654e-01 -2.29484499e-01
1.88039884e-01 1.96002975e-01 6.30622387e-01 1.56272757e+00
4.50353295e-01 5.17584324e-01 1.68757558e+00 -3.23789179e-01
-1.18157201e-01 2.73873180e-01 -9.88602459e-01 7.76872396e-01
3.60774808e-02 4.04080242e-01 -2.33496815e-01 1.66104212e-01
1.16107249e+00 2.08196184e-03 -1.50374025e-01 -1.39182955e-01
-1.11091030e+00 7.47122586e-01 7.23333597e-01 7.38779902e-02
-1.26042306e-01 3.55777264e-01 1.86832741e-01 6.04045331e-01
3.37794423e-01 2.56122947e-01 5.76830916e-02 2.69176513e-01
-7.69129276e-01 2.16396675e-01 5.13802350e-01 1.13072336e+00
7.45862782e-01 4.48241830e-01 -4.58389401e-01 7.70392776e-01
2.28687242e-01 9.34428394e-01 4.35069293e-01 -8.02270830e-01
3.61127228e-01 3.14937770e-01 -4.30255160e-02 -1.00767946e+00
3.16278636e-01 -3.09913427e-01 -1.39777565e+00 6.63339198e-01
6.23238571e-02 -2.89601833e-01 -1.13126230e+00 2.02650928e+00
3.02621752e-01 1.51224121e-01 9.97242033e-02 8.68628025e-01
6.80487216e-01 8.98171723e-01 9.65732783e-02 -2.38862395e-01
1.34296846e+00 -9.30535257e-01 -8.01594019e-01 -1.50085881e-01
-2.89100967e-02 -1.15920746e+00 1.08413625e+00 1.17913976e-01
-1.20074368e+00 -8.39082241e-01 -1.11660182e+00 -1.69298366e-01
-3.13025296e-01 1.89241871e-01 3.68832052e-01 6.63230240e-01
-1.00079846e+00 4.12494749e-01 -3.57375085e-01 -2.68808246e-01
6.25190735e-01 2.37634674e-01 -6.00224853e-01 -1.27787039e-01
-1.29962480e+00 8.74908328e-01 2.29389951e-01 8.91800448e-02
-1.14593530e+00 -5.97028017e-01 -7.51455128e-01 -1.86017737e-01
-3.19487341e-02 -1.04968119e+00 7.20748127e-01 -1.63928604e+00
-1.75712395e+00 1.08247936e+00 3.20115328e-01 -8.47393945e-02
8.96709383e-01 2.02257261e-01 -5.43148041e-01 -7.63376430e-02
4.59712446e-02 1.02760851e+00 1.38771796e+00 -1.36788404e+00
-3.57177436e-01 -3.91417265e-01 -8.66746996e-03 2.22629115e-01
-2.26943567e-01 -6.40254060e-04 -3.12176585e-01 -9.53660011e-01
-2.55691290e-01 -1.04767191e+00 2.34041676e-01 4.88256425e-01
-3.44733953e-01 2.20771521e-01 1.17862570e+00 -7.82508910e-01
6.52049005e-01 -2.21954966e+00 3.65133852e-01 -2.24421471e-01
-8.32812861e-02 3.69458795e-01 -4.89151388e-01 3.55953902e-01
-1.43586248e-01 1.63836271e-01 -3.32201868e-01 -3.52569163e-01
-1.28607616e-01 1.99230134e-01 -4.63208109e-01 3.99931788e-01
5.72170675e-01 1.24802470e+00 -7.65764534e-01 -6.20366812e-01
2.64926273e-02 6.93550766e-01 -5.31829834e-01 5.09482086e-01
-1.12714924e-01 8.90755951e-01 -1.76355675e-01 6.31917655e-01
1.08018517e+00 2.44642124e-01 -2.33177990e-01 -3.04919302e-01
2.47371495e-01 -4.68799591e-01 -8.11833382e-01 1.58083737e+00
-7.01272368e-01 5.57361424e-01 2.06449330e-01 -8.13263297e-01
1.14070523e+00 2.65764475e-01 1.24632241e-02 -7.22616971e-01
1.88790739e-01 2.44799167e-01 -4.82760258e-02 -3.55182827e-01
8.82762671e-02 -7.40014672e-01 -1.54968843e-01 1.53081432e-01
4.43331338e-02 -5.58213949e-01 -3.04597497e-01 -1.98025599e-01
6.34412348e-01 5.39058924e-01 1.07052429e-02 -6.55788928e-02
7.08678067e-01 -2.04275802e-01 6.40697002e-01 -4.87595215e-04
2.55253091e-02 9.65640306e-01 4.65170830e-01 -2.48051390e-01
-1.49455035e+00 -1.11259401e+00 1.27205089e-01 6.29417777e-01
1.87184513e-01 3.35056454e-01 -9.67150986e-01 -5.24820030e-01
-1.87724262e-01 3.78198832e-01 -8.13074648e-01 -5.60000300e-01
-6.10306561e-01 -4.17716324e-01 8.65537941e-01 3.09109151e-01
1.23312116e+00 -1.18177557e+00 3.86444107e-02 -4.20391187e-02
-1.30660668e-01 -1.19759715e+00 -9.15137291e-01 -3.73380065e-01
-6.15987360e-01 -6.01158619e-01 -1.21391284e+00 -1.09161842e+00
8.51436734e-01 4.79659177e-02 9.55682814e-01 -2.00912029e-01
-2.40947664e-01 8.99129780e-04 -3.06419998e-01 -3.32199991e-01
-7.34811485e-01 -2.44969711e-01 -2.46011019e-02 4.09481406e-01
-1.78859904e-01 -9.16551054e-01 -8.48161101e-01 4.94492799e-01
-1.22674215e+00 2.95533448e-01 8.66207719e-01 1.02437782e+00
5.77597618e-01 -2.06586927e-01 5.06431818e-01 -9.40630734e-01
6.64913654e-01 -2.72085488e-01 -4.49618131e-01 2.87959546e-01
-4.82458979e-01 7.24323466e-02 9.10726488e-01 -7.36266732e-01
-1.17643142e+00 -1.29671416e-05 -7.01620951e-02 -8.86698663e-01
8.42063203e-02 -6.16564639e-02 -5.69177032e-01 -3.53253841e-01
6.95713699e-01 4.21370059e-01 2.82319397e-01 -1.03731655e-01
5.96116841e-01 5.97486317e-01 7.04330266e-01 -5.44540584e-01
1.35203302e+00 4.09217298e-01 2.73297906e-01 -4.88589704e-01
-4.18654442e-01 4.86891985e-01 -3.74260634e-01 -8.90882686e-02
1.11793649e+00 -1.04016244e+00 -5.87884545e-01 6.11529291e-01
-1.31881964e+00 -2.12233841e-01 -3.10127407e-01 7.77246654e-02
-8.00886452e-01 1.91881880e-01 -4.15717155e-01 -4.26986039e-01
-5.44417918e-01 -1.09893262e+00 1.14968741e+00 3.55511606e-01
3.54263723e-01 -8.44727337e-01 3.94686218e-03 2.70407617e-01
5.85963726e-01 8.13089311e-01 8.99486840e-01 1.73018038e-01
-6.92444026e-01 -4.98970710e-02 -3.68162662e-01 7.94900060e-01
2.02812478e-01 -3.38250548e-02 -7.38193393e-01 -4.43299562e-01
3.48456175e-04 -5.82601666e-01 6.84417427e-01 -1.86159275e-02
9.76604700e-01 -6.51040971e-01 -1.68450058e-01 1.02876091e+00
1.62622118e+00 1.21191695e-01 1.18673599e+00 -1.93915859e-01
8.83899450e-01 4.11304981e-01 5.05749643e-01 -4.05752622e-02
-4.62150387e-02 7.99741983e-01 4.41338211e-01 -3.95394355e-01
-7.86605000e-01 -7.63853073e-01 7.19888926e-01 4.54659849e-01
-1.26518473e-01 -2.51086980e-01 -3.07494700e-01 3.96082491e-01
-1.43844581e+00 -1.03413594e+00 2.20646575e-01 2.01940799e+00
9.35283124e-01 -1.76669791e-01 -5.47288842e-02 -1.11903742e-01
9.06392753e-01 1.50925860e-01 -5.57891488e-01 -6.57575905e-01
-3.58203113e-01 2.65997142e-01 3.72041583e-01 3.63310158e-01
-7.68579483e-01 1.12662923e+00 5.72534466e+00 1.23672199e+00
-1.40268075e+00 3.43760163e-01 9.90198970e-01 3.05562705e-01
-5.04833043e-01 1.49373617e-02 -5.65655291e-01 6.58881664e-01
4.69025701e-01 1.07512868e-03 6.33570790e-01 7.64929652e-01
4.56579626e-02 3.04966658e-01 -9.73468840e-01 1.28370810e+00
2.22950056e-01 -1.17210472e+00 4.42116410e-01 1.77645776e-02
1.08855247e+00 -6.37640595e-01 4.61034477e-01 2.61501342e-01
1.96292862e-01 -1.30780435e+00 7.08354115e-01 4.45202321e-01
1.62300014e+00 -7.38783777e-01 5.71772695e-01 9.58382413e-02
-1.00632679e+00 1.26065418e-01 -4.86973703e-01 1.58160776e-01
7.75304884e-02 2.98652381e-01 -4.45857644e-01 6.01741195e-01
2.79791385e-01 5.68561018e-01 -4.48863536e-01 5.23607314e-01
-4.18791413e-01 2.45764449e-01 -1.08259264e-02 1.61320642e-01
2.09591180e-01 -5.01653075e-01 4.26624954e-01 9.18601036e-01
6.79846942e-01 -2.32996959e-02 -1.12843320e-01 1.21788812e+00
-4.28110927e-01 6.21407591e-02 -1.04954493e+00 -2.01594401e-02
3.20655495e-01 1.28220487e+00 -2.80618548e-01 -2.61962771e-01
-4.65614162e-02 1.57425320e+00 -1.70137249e-02 3.03268611e-01
-1.12212157e+00 -4.20894146e-01 5.99274755e-01 2.90457875e-01
1.75026760e-01 -3.31905554e-03 -1.89369410e-01 -1.03292692e+00
4.46706675e-02 -1.05268526e+00 -2.17606366e-01 -1.00684130e+00
-1.42296100e+00 8.87142301e-01 -2.20832735e-01 -1.48412120e+00
-1.37086064e-01 -3.24454933e-01 -8.77083302e-01 1.09557843e+00
-1.38663304e+00 -1.93641448e+00 -5.27322054e-01 9.14223850e-01
3.09584677e-01 -4.04212475e-01 6.76306069e-01 4.27958608e-01
-2.66358823e-01 1.04534352e+00 4.99995090e-02 2.48879910e-01
1.06603789e+00 -7.92354941e-01 3.25560927e-01 7.55698025e-01
-7.88216889e-02 1.17811196e-01 6.26504242e-01 -4.59642053e-01
-1.43239415e+00 -1.40458536e+00 4.79175597e-01 -3.04074258e-01
1.76438585e-01 -3.69334906e-01 -5.23098350e-01 5.38293481e-01
6.33132517e-01 6.25182092e-02 2.30820611e-01 -6.45986378e-01
-4.48636770e-01 -5.06187558e-01 -1.53178847e+00 7.56504893e-01
1.12604666e+00 -4.56974059e-01 -3.09783518e-01 2.55281389e-01
8.20751369e-01 -3.55535269e-01 -7.29808688e-01 3.56857836e-01
5.74790478e-01 -9.30465817e-01 1.03779757e+00 -3.90228570e-01
9.05953825e-01 -4.32568520e-01 -1.52045637e-01 -1.39662302e+00
-4.20973778e-01 -8.73886585e-01 4.68522549e-01 1.56195474e+00
1.14509404e-01 -5.57184696e-01 5.99940777e-01 8.29423666e-02
7.60851279e-02 -5.89685559e-01 -8.85205746e-01 -8.90477538e-01
4.00365382e-01 3.23920906e-01 7.08817899e-01 7.99434543e-01
-6.18896842e-01 6.20398164e-01 -9.54667330e-01 -1.43914953e-01
6.46057546e-01 2.09187910e-01 1.07858527e+00 -8.11620712e-01
-2.85184920e-01 -2.74430931e-01 -5.64164937e-01 -7.94679880e-01
2.14750573e-01 -9.86508727e-01 -8.14243481e-02 -1.14666474e+00
9.27622020e-02 -3.37345481e-01 1.72882318e-01 3.07399213e-01
2.92563140e-02 8.75237107e-01 3.97621512e-01 1.81933656e-01
5.27376793e-02 8.42468023e-01 2.09949398e+00 -3.40643376e-01
1.39426053e-01 -1.50264382e-01 -7.40845799e-01 3.27866554e-01
6.37399971e-01 -4.35590655e-01 -5.11187851e-01 -4.26643580e-01
3.99629623e-02 1.91477105e-01 4.97820675e-01 -1.06171155e+00
-1.86405808e-01 -2.91451156e-01 6.57902956e-01 6.30875081e-02
5.83532333e-01 -7.30608761e-01 6.57377839e-01 5.30688226e-01
-1.52336895e-01 1.67608168e-02 2.77143214e-02 6.15148604e-01
-5.11522353e-01 2.10188612e-01 1.40963066e+00 -2.18403712e-01
-3.82564336e-01 6.36570215e-01 3.98125380e-01 4.30088490e-02
1.14559853e+00 -2.39285111e-01 -2.57072300e-01 -5.32874525e-01
-5.22626519e-01 -1.86865702e-01 6.84434235e-01 5.75526893e-01
7.12929070e-01 -1.98328698e+00 -1.08965063e+00 4.85583603e-01
9.65375528e-02 -2.66229659e-01 2.09626347e-01 5.08976161e-01
-7.69626558e-01 1.01688847e-01 -1.05135202e+00 -4.18379068e-01
-1.05362809e+00 7.14206457e-01 2.23416612e-01 -1.27822325e-01
-3.55952084e-01 6.52065158e-01 5.54065049e-01 -2.53609627e-01
-1.48743957e-01 1.47998959e-01 1.85212880e-01 -2.69457400e-01
3.70755017e-01 1.00262888e-01 -3.41453582e-01 -7.59282589e-01
-8.44045430e-02 9.69890714e-01 2.00214788e-01 -2.11181536e-01
1.01147377e+00 -6.38654605e-02 -2.84393996e-01 -4.94372398e-02
1.32535946e+00 -1.67903483e-01 -1.38995004e+00 -6.92586899e-02
-9.15844619e-01 -7.05631137e-01 -1.91501826e-01 -7.62162924e-01
-1.38752317e+00 1.15215611e+00 9.86897886e-01 -2.14180052e-01
1.31963360e+00 -3.15282315e-01 8.89894366e-01 -1.20266967e-01
4.68676329e-01 -7.80403316e-01 4.04502422e-01 3.99669036e-02
1.46205270e+00 -1.30729842e+00 -3.01430017e-01 -2.52571851e-01
-7.93270886e-01 7.99109340e-01 6.47949994e-01 -5.29589415e-01
5.49099624e-01 9.68032405e-02 7.49049038e-02 3.22739005e-01
-4.83041972e-01 3.36554646e-02 1.93589836e-01 9.65251565e-01
2.52418607e-01 1.55172169e-01 -4.74273175e-01 1.45857275e-01
-2.38936484e-01 4.79656681e-02 4.00195986e-01 2.87186891e-01
-5.40357195e-02 -1.47435927e+00 -5.59594274e-01 2.20584683e-02
-4.07947183e-01 -1.07643558e-02 -3.90479177e-01 6.86428666e-01
5.29781818e-01 7.36232758e-01 -8.81105959e-02 -5.10111034e-01
1.79215461e-01 -1.25295699e-01 8.00039768e-01 -4.01745647e-01
-4.87888783e-01 1.00074580e-03 -2.97286808e-01 -4.51587677e-01
-2.76181519e-01 -1.87494889e-01 -7.60289252e-01 -7.04805315e-01
7.61882514e-02 -1.30677447e-01 5.84186554e-01 4.59596306e-01
4.31682080e-01 4.23397064e-01 8.78104091e-01 -8.56205940e-01
-5.27642250e-01 -1.12587380e+00 -5.46622932e-01 7.16081381e-01
9.97968912e-02 -5.10707378e-01 -1.54978856e-01 3.43118936e-01] | [12.0967435836792, -0.2946988642215729] |
9bfcefc7-bbe6-4f66-a49d-31aafa262bca | source-free-domain-adaptation-of-a-dnn-for | 2305.17403 | null | https://arxiv.org/abs/2305.17403v1 | https://arxiv.org/pdf/2305.17403v1.pdf | Source Free Domain Adaptation of a DNN for SSVEP-based Brain-Computer Interfaces | This paper presents a source free domain adaptation method for steady-state visually evoked potential (SSVEP) based brain-computer interface (BCI) spellers. SSVEP-based BCI spellers help individuals experiencing speech difficulties, enabling them to communicate at a fast rate. However, achieving a high information transfer rate (ITR) in the current methods requires an extensive calibration period before using the system, leading to discomfort for new users. We address this issue by proposing a method that adapts the deep neural network (DNN) pre-trained on data from source domains (participants of previous experiments conducted for labeled data collection), using only the unlabeled data of the new user (target domain). This adaptation is achieved by minimizing our proposed custom loss function composed of self-adaptation and local-regularity loss terms. The self-adaptation term uses the pseudo-label strategy, while the novel local-regularity term exploits the data structure and forces the DNN to assign the same labels to adjacent instances. Our method achieves striking 201.15 bits/min and 145.02 bits/min ITRs on the benchmark and BETA datasets, respectively, and outperforms the state-of-the-art alternative techniques. Our approach alleviates user discomfort and shows excellent identification performance, so it would potentially contribute to the broader application of SSVEP-based BCI systems in everyday life. | ['Huseyin Ozkan', 'Deniz Kucukahmetler', 'Osman Berke Guney'] | 2023-05-27 | null | null | null | null | ['source-free-domain-adaptation', 'pseudo-label'] | ['computer-vision', 'miscellaneous'] | [ 3.14437747e-01 -2.20102862e-01 4.32189517e-02 -3.00034612e-01
-8.72242570e-01 -3.47618014e-01 2.44298846e-01 -1.65573046e-01
-7.29750693e-01 1.26590705e+00 8.03610831e-02 -8.38252530e-02
-7.89301023e-02 -2.34872431e-01 -4.46872264e-01 -8.67729783e-01
-2.92449705e-02 7.44759142e-02 8.43782946e-02 -4.67861183e-02
1.75095081e-01 4.64071572e-01 -1.40180159e+00 1.68562979e-01
1.20036697e+00 1.39051270e+00 2.88561672e-01 4.07070339e-01
1.85051456e-01 1.10348679e-01 -1.13673723e+00 -2.51960993e-01
8.15181469e-04 -4.30271655e-01 -6.00958288e-01 -4.76786792e-01
1.20893300e-01 -1.41392425e-01 -5.72605908e-01 1.03294730e+00
1.41276228e+00 1.42594978e-01 6.76726282e-01 -1.24701774e+00
-5.40396512e-01 5.01631387e-02 -4.63581592e-01 4.03228968e-01
2.75240004e-01 9.28805992e-02 2.84395248e-01 -6.58772230e-01
2.03025177e-01 6.59866989e-01 5.84153593e-01 1.11523116e+00
-1.32342494e+00 -1.29749441e+00 -1.85607314e-01 7.07934916e-01
-1.77891374e+00 -7.78276563e-01 8.98049295e-01 -2.65644252e-01
1.04164541e+00 1.25148997e-01 8.05622160e-01 1.66150510e+00
-1.03092231e-01 8.38088870e-01 1.14740074e+00 -2.99881905e-01
5.62451780e-01 3.90556931e-01 2.18666866e-01 -1.83609515e-01
1.97369829e-02 2.09463850e-01 -1.05801260e+00 -6.65176660e-03
6.69180155e-01 -4.79972631e-01 -8.33438277e-01 -2.45596275e-01
-7.34813631e-01 5.05376339e-01 1.86622977e-01 4.94214654e-01
-6.75978899e-01 -2.00094968e-01 3.58784735e-01 2.90009379e-01
3.23342472e-01 2.81988949e-01 -3.96780312e-01 -6.55438364e-01
-9.00860667e-01 -3.84084210e-02 7.20035672e-01 8.83123338e-01
7.22551439e-03 5.07760830e-02 -5.72547376e-01 1.33385944e+00
-1.90821782e-01 6.89929128e-01 8.33006382e-01 -4.55944240e-01
4.89180535e-01 3.07939872e-02 1.20809808e-01 -4.96233523e-01
-4.11127925e-01 -5.09163439e-01 -8.07297707e-01 1.02351397e-01
2.93444455e-01 -4.44639117e-01 -8.43013585e-01 2.02302337e+00
-9.78892297e-02 2.12084353e-01 -2.56832913e-02 1.00982881e+00
6.45164192e-01 3.42461228e-01 1.69427693e-01 -3.78644735e-01
1.11388361e+00 -4.33990687e-01 -1.05959392e+00 -2.13520586e-01
7.42536336e-02 -1.20543487e-01 1.04451632e+00 8.00434947e-01
-1.06848752e+00 -4.53659862e-01 -1.13795435e+00 3.77656162e-01
-2.21219689e-01 -1.51005061e-02 3.57420802e-01 1.09903646e+00
-1.27694321e+00 3.34169686e-01 -4.57938820e-01 -3.58755887e-01
5.87803364e-01 8.26978862e-01 -2.86035389e-01 4.42812413e-01
-1.38131857e+00 9.33578074e-01 3.05953413e-01 -1.79685980e-01
-7.25056648e-01 -6.53143704e-01 -3.49786431e-01 1.55505776e-01
-6.14028424e-02 -3.51599514e-01 1.02350938e+00 -9.04335320e-01
-1.92190099e+00 7.97380328e-01 -3.38676929e-01 -5.38642585e-01
2.70180464e-01 -9.53164101e-02 -8.78248513e-01 1.77930713e-01
-2.79218882e-01 8.55666578e-01 1.04252863e+00 -9.48478222e-01
-6.87353015e-01 -5.17298043e-01 -5.59678018e-01 3.04052353e-01
-8.43808413e-01 -1.44755915e-01 -2.72680849e-01 -6.53707504e-01
-3.55643719e-01 -5.91524363e-01 5.91126025e-01 -1.33872122e-01
-2.72097528e-01 -3.48935723e-01 5.79892218e-01 -9.70495343e-01
1.38346100e+00 -2.46091104e+00 5.45362988e-03 3.58732969e-01
6.60113841e-02 7.76981890e-01 -3.09694290e-01 -9.48266089e-02
-3.15193325e-01 -1.87804952e-01 -2.34658092e-01 -1.19046539e-01
-2.93811083e-01 -2.19931468e-01 2.82103010e-02 3.56971473e-01
-1.03217535e-01 7.76998997e-01 -6.54422402e-01 1.78077370e-02
2.06066087e-01 8.10765564e-01 -3.54006648e-01 2.18926728e-01
6.43098414e-01 7.50796497e-01 1.42051458e-01 3.21504533e-01
8.20243657e-01 2.56586373e-01 -1.62467480e-01 -8.93806666e-02
1.46442011e-01 5.38120985e-01 -8.77720714e-01 1.84080517e+00
-5.66447794e-01 9.19595361e-01 8.22836161e-03 -1.27822638e+00
1.02176201e+00 6.72165692e-01 5.02025008e-01 -1.25687134e+00
3.84746939e-01 3.31531465e-01 1.56458884e-01 -5.14903605e-01
-1.79874927e-01 -1.33137470e-02 2.53796309e-01 -1.24896709e-02
1.54599696e-01 2.66696177e-02 -2.63062686e-01 -7.30017424e-02
9.94572222e-01 -3.47924680e-01 2.85380691e-01 -2.43859127e-01
5.29542863e-01 -6.68094635e-01 3.21668714e-01 7.39622831e-01
-4.94349748e-01 3.86209637e-01 6.04113452e-02 -2.38007233e-02
-6.55099094e-01 -1.23804116e+00 -5.79909205e-01 6.64733827e-01
1.36729285e-01 2.35973597e-02 -1.03658366e+00 -3.90143692e-01
-7.85722956e-02 1.04192710e+00 -3.30235004e-01 -7.64156997e-01
-8.29547551e-03 -4.87541586e-01 6.43464148e-01 6.37218952e-01
8.19672048e-01 -1.24387765e+00 -5.22989810e-01 3.44311804e-01
-6.66486084e-01 -8.87236774e-01 -5.16479731e-01 3.17438990e-01
-7.56343782e-01 -6.35683298e-01 -1.33593965e+00 -8.14995587e-01
2.80769885e-01 -7.60425255e-02 5.16500354e-01 -6.19802594e-01
-1.23183824e-01 4.75492060e-01 -2.33138338e-01 -6.72846377e-01
9.07312930e-02 -2.98496950e-02 4.49366659e-01 1.95606068e-01
1.00511050e+00 -9.81885016e-01 -9.13195252e-01 1.96413457e-01
-3.06146055e-01 -2.58868724e-01 3.69043827e-01 1.02114868e+00
3.51219386e-01 -1.85964599e-01 9.56329465e-01 -1.28522322e-01
1.28975916e+00 -2.71419495e-01 -1.38420403e-01 2.24046633e-01
-5.42014897e-01 -1.12840272e-01 3.65673035e-01 -1.00898695e+00
-1.05301094e+00 -1.53273135e-01 -2.24920884e-01 -6.15785494e-02
-3.30107987e-01 3.54111046e-02 -2.86696076e-01 -3.53440344e-01
9.53334510e-01 5.28728187e-01 -2.68120050e-01 -4.57738996e-01
-7.41343498e-02 1.68895066e+00 9.59189773e-01 -2.40747273e-01
1.36052877e-01 5.64583093e-02 -5.75512826e-01 -1.12152505e+00
-1.73537850e-01 -5.22667944e-01 -3.09829384e-01 -1.36097029e-01
5.73483944e-01 -8.40363860e-01 -1.01325548e+00 9.22402263e-01
-1.38506138e+00 -3.25676650e-01 -1.52364597e-01 7.30985522e-01
-6.17968619e-01 1.89267583e-02 -2.57686645e-01 -1.15000165e+00
-8.11860681e-01 -8.08439672e-01 5.85787952e-01 3.48480403e-01
-4.62968826e-01 -3.80772114e-01 -7.08277896e-02 1.34810969e-01
6.88178301e-01 -1.10075556e-01 6.89668357e-01 -8.18700373e-01
3.17588329e-01 -2.43782416e-01 -2.25932360e-01 6.99178934e-01
2.78097272e-01 -1.09794962e+00 -1.43915558e+00 -4.50725585e-01
3.55939776e-01 -3.45563352e-01 4.62097049e-01 6.70303762e-01
1.53091466e+00 -1.62985981e-01 -3.38758260e-01 6.92813754e-01
9.26916659e-01 8.63478661e-01 1.02691603e+00 7.26752430e-02
2.92955369e-01 2.73091018e-01 -2.12655172e-01 5.04612267e-01
1.62384570e-01 8.92421722e-01 -1.59382492e-01 6.49361759e-02
-1.89925298e-01 -1.26680329e-01 1.89502656e-01 4.55423772e-01
-2.06322640e-01 -2.79059827e-01 -7.10146129e-01 6.76274717e-01
-1.43719852e+00 -6.90773368e-01 2.34332234e-01 2.72042489e+00
1.12082529e+00 3.90334660e-03 1.84064940e-01 5.97730160e-01
7.37901509e-01 -5.15571535e-01 -1.04325926e+00 -4.92707282e-01
-1.24343589e-01 9.11965430e-01 4.09895957e-01 2.47878224e-01
-8.74011993e-01 5.57788253e-01 5.93630123e+00 1.16883993e+00
-1.42325592e+00 3.94788921e-01 4.42711890e-01 -4.87512261e-01
2.55260199e-01 -8.46683145e-01 -4.73167568e-01 1.00884461e+00
1.34356213e+00 -2.08686486e-01 9.37259853e-01 5.57838380e-01
2.53764868e-01 -2.48458534e-01 -1.17270482e+00 1.96774161e+00
1.67145073e-01 -8.20357978e-01 -3.07378262e-01 2.34086793e-02
2.90603250e-01 -3.18814255e-02 1.71745896e-01 1.58936411e-01
-4.32372451e-01 -9.17196155e-01 6.11712813e-01 5.98747849e-01
1.57822061e+00 -9.21946943e-01 4.99932259e-01 2.38812879e-01
-8.30698192e-01 -2.56397396e-01 2.82945437e-03 1.41588092e-01
6.07816875e-03 3.19871306e-01 -5.41128218e-01 5.84249832e-02
8.88817191e-01 3.32592010e-01 -1.89637467e-01 1.39640617e+00
-3.95093471e-01 7.48115063e-01 -4.63095278e-01 -4.15565044e-01
-2.78053224e-01 1.63031638e-01 6.87705934e-01 1.16859388e+00
3.75151992e-01 3.97138238e-01 -6.39714479e-01 6.87759161e-01
-7.47064352e-02 -4.24722433e-02 -5.37192166e-01 2.63156086e-01
8.83994043e-01 7.29846478e-01 -4.36126702e-02 -1.51315168e-01
-1.02314495e-01 1.69641912e+00 2.89694697e-01 7.35415876e-01
-6.34272754e-01 -1.06662893e+00 4.94524688e-01 -2.22374499e-02
-1.96809396e-02 -1.27396407e-02 -7.10837841e-01 -8.79159033e-01
2.77371615e-01 -9.85122383e-01 1.63866431e-01 -7.23846436e-01
-1.22743845e+00 6.39358521e-01 -1.64657682e-02 -1.20947218e+00
-2.40916640e-01 -5.51695645e-01 -4.06922281e-01 1.18509221e+00
-1.36389613e+00 -4.05829638e-01 -2.49990880e-01 1.13407302e+00
3.05240065e-01 -2.97702998e-01 9.74965394e-01 5.95121443e-01
-3.93372983e-01 1.34842932e+00 2.13488504e-01 -6.82665035e-02
9.14201021e-01 -8.27486753e-01 1.10177554e-01 6.10760629e-01
-4.52664420e-02 5.44669807e-01 3.56174946e-01 -2.84178972e-01
-7.02246726e-01 -4.95119631e-01 8.77080500e-01 -8.61655399e-02
6.70879707e-02 -5.15106201e-01 -8.12004268e-01 -4.32772748e-02
2.62520969e-01 -4.34006482e-01 7.43517816e-01 -1.08411156e-01
6.86106086e-02 -5.05363047e-01 -1.48467743e+00 5.00432312e-01
1.31242645e+00 -7.29762495e-01 -4.89453256e-01 7.96279758e-02
1.81832567e-01 -1.32281289e-01 -4.79627848e-01 2.40278631e-01
7.37394273e-01 -6.01313531e-01 8.60021830e-01 -2.89060116e-01
-3.54974240e-01 2.78784305e-01 1.51123732e-01 -1.85412550e+00
-2.99042404e-01 -9.69557106e-01 -2.67716110e-01 1.05311978e+00
3.88609558e-01 -9.75894153e-01 4.15356934e-01 1.01373446e+00
1.09656557e-01 -5.57341814e-01 -1.29062068e+00 -1.01234961e+00
-1.36305122e-02 -5.22295654e-01 4.89612013e-01 4.46404427e-01
6.91660881e-01 2.47352377e-01 -4.66139883e-01 -1.22605167e-01
4.82519329e-01 -8.26469779e-01 2.38626435e-01 -1.33483887e+00
-1.78549979e-02 -5.45437992e-01 -5.02500534e-01 -1.21200371e+00
-7.31121376e-02 -6.93907678e-01 1.80923089e-01 -1.18340266e+00
2.77569704e-02 -2.14697003e-01 -7.92838514e-01 5.61798215e-01
-1.03386007e-01 2.94740617e-01 -1.33127138e-01 -7.72848204e-02
-8.71503726e-02 7.68655777e-01 6.79242194e-01 -3.91608000e-01
-7.41352081e-01 1.01920895e-01 -7.19357252e-01 4.75219280e-01
9.54123318e-01 -4.10617679e-01 -5.41462481e-01 -2.39331722e-01
-4.74407226e-01 -7.45897219e-02 3.35690141e-01 -1.48515761e+00
5.72103202e-01 3.05152416e-01 6.10679984e-01 -2.03584850e-01
5.73856354e-01 -6.85975730e-01 -2.12018356e-01 3.85144502e-01
-3.41607004e-01 -6.12818897e-01 5.32848179e-01 4.50502485e-01
-1.58834755e-02 -1.24829851e-01 1.09169757e+00 4.02815908e-01
-6.26980424e-01 1.88668892e-01 -4.92951453e-01 1.44201368e-01
9.04017031e-01 -4.85697001e-01 8.36336762e-02 -6.52913630e-01
-7.20528960e-01 -3.84232625e-02 -1.98065311e-01 4.68945563e-01
8.30677330e-01 -1.35249257e+00 -4.80826527e-01 6.70761883e-01
1.99486747e-01 -5.01326025e-01 2.34529182e-01 7.82472670e-01
1.71268702e-01 5.36210954e-01 -5.90015590e-01 -4.39905763e-01
-1.02942586e+00 2.30008200e-01 5.04554033e-01 3.41707855e-01
-6.56630397e-01 1.23521328e+00 5.42574888e-03 1.21785671e-01
9.64871883e-01 -3.58779542e-03 -3.78512859e-01 -3.16486545e-02
8.20666671e-01 5.10666490e-01 3.83871138e-01 -2.96699554e-01
-6.11146510e-01 1.99346691e-01 -5.97392060e-02 -6.12423480e-01
1.23460722e+00 -1.57505527e-01 5.67464948e-01 1.57556787e-01
1.13821542e+00 -3.56359214e-01 -1.15061772e+00 -2.69081771e-01
-2.14254916e-01 -3.25167328e-01 2.76313186e-01 -1.45882213e+00
-1.01166391e+00 1.25920522e+00 1.46563137e+00 -3.36324811e-01
1.53902435e+00 -2.63605446e-01 1.02674961e+00 3.78575385e-01
5.90010524e-01 -1.52658772e+00 2.44457088e-02 5.31569533e-02
9.31658387e-01 -9.00203288e-01 -6.76130056e-01 1.82245329e-01
-8.46870542e-01 7.17488050e-01 4.96501297e-01 1.49227932e-01
6.18083477e-01 2.08692044e-01 -1.36968940e-01 1.99882999e-01
-1.79454103e-01 -5.38330413e-02 4.24610704e-01 1.31293952e+00
1.16065599e-01 2.53002226e-01 -8.01127195e-01 1.20846856e+00
-1.22602068e-01 4.29814994e-01 1.39238872e-02 4.92391825e-01
-1.52572557e-01 -9.27059948e-01 -2.13982582e-01 5.92553616e-01
-2.93631375e-01 -3.03235024e-01 -3.35516721e-01 4.40979004e-01
1.66069105e-01 1.40954769e+00 8.30815062e-02 -5.53772032e-01
6.75989270e-01 4.31675732e-01 4.55644161e-01 -1.61520585e-01
-3.12979907e-01 -6.36118203e-02 -2.10411817e-01 -4.67871487e-01
-1.02165632e-01 -4.72600579e-01 -1.02210164e+00 -1.59219876e-02
-2.25896940e-01 4.46968339e-02 8.68203878e-01 7.93574154e-01
6.83485150e-01 4.23054934e-01 6.02344036e-01 -7.49416709e-01
-4.61856753e-01 -1.35921240e+00 -8.04436207e-01 5.95736504e-01
3.58713031e-01 -7.71500707e-01 -2.60334611e-01 -1.05818145e-01] | [13.139204978942871, 3.4308183193206787] |
9a651c0c-3dcb-4a4b-a293-b32594319ddf | a-multi-task-selected-learning-approach-for | 1804.06896 | null | http://arxiv.org/abs/1804.06896v3 | http://arxiv.org/pdf/1804.06896v3.pdf | A Multi-task Selected Learning Approach for Solving 3D Flexible Bin Packing Problem | A 3D flexible bin packing problem (3D-FBPP) arises from the process of
warehouse packing in e-commerce. An online customer's order usually contains
several items and needs to be packed as a whole before shipping. In particular,
5% of tens of millions of packages are using plastic wrapping as outer
packaging every day, which brings pressure on the plastic surface minimization
to save traditional logistics costs. Because of the huge practical
significance, we focus on the issue of packing cuboid-shaped items orthogonally
into a least-surface-area bin. The existing heuristic methods for classic 3D
bin packing don't work well for this particular NP-hard problem and designing a
good problem-specific heuristic is non-trivial. In this paper, rather than
designing heuristics, we propose a novel multi-task framework based on Selected
Learning to learn a heuristic-like policy that generates the sequence and
orientations of items to be packed simultaneously. Through comprehensive
experiments on a large scale real-world transaction order dataset and online AB
tests, we show: 1) our selected learning method trades off the imbalance and
correlation among the tasks and significantly outperforms the single task
Pointer Network and the multi-task network without selected learning; 2) our
method obtains an average 5.47% cost reduction than the well-designed greedy
algorithm which is previously used in our online production system. | ['Jiangwen Wei', 'Yu Qian', 'Yinghui Xu', 'Haoyuan Hu', 'Yu Gong', 'Xiaodong Zhang', 'Lu Duan'] | 2018-04-17 | null | null | null | null | ['3d-bin-packing'] | ['miscellaneous'] | [-2.99641848e-01 6.61458969e-02 -3.89413238e-01 -2.97664374e-01
-6.31606698e-01 -5.96344888e-01 -4.54927504e-01 3.82642806e-01
-2.59241402e-01 7.50316441e-01 -2.06669748e-01 -6.24955952e-01
-4.99875128e-01 -8.85095358e-01 -1.29877019e+00 -7.49032676e-01
-6.39810622e-01 1.33067775e+00 2.78208584e-01 -3.23368609e-01
3.24951887e-01 4.22467142e-01 -9.41235244e-01 5.49348176e-01
9.35429096e-01 1.44316971e+00 7.40215003e-01 2.18271896e-01
-2.37638265e-01 -1.03793673e-01 -5.11409938e-01 -5.58077455e-01
8.44093323e-01 -4.94200364e-02 -6.29187644e-01 4.84471083e-01
-5.59072793e-02 -1.99019924e-01 1.49412766e-01 7.68730760e-01
4.67863441e-01 2.19957650e-01 3.04320008e-01 -1.45199645e+00
-7.06176996e-01 7.92282045e-01 -7.64097273e-01 2.00781927e-01
1.21901333e-01 3.54858823e-02 1.07592118e+00 -2.37138823e-01
2.65640259e-01 9.95796859e-01 3.87629807e-01 8.87309983e-02
-1.28931105e+00 -2.83016264e-01 4.16482747e-01 3.88830975e-02
-8.34616125e-01 1.75308526e-01 7.72916138e-01 -3.68780009e-02
1.10983205e+00 1.77911669e-01 9.03910398e-01 9.18488324e-01
7.69514441e-01 1.05392909e+00 1.02806580e+00 -2.27332219e-01
3.92019510e-01 -2.21283957e-02 1.37876287e-01 4.58274662e-01
7.73056448e-01 -1.19545668e-01 -3.57258230e-01 5.85023500e-02
6.76322162e-01 1.85182154e-01 1.49033610e-02 -9.21733022e-01
-9.67549264e-01 1.01793051e+00 4.81009185e-01 -1.19372457e-01
-4.01640832e-01 -1.64391100e-02 5.01742423e-01 5.72086155e-01
5.52884102e-01 7.26505876e-01 -9.51451957e-01 7.64074549e-03
-5.23249865e-01 7.13798523e-01 1.38138711e+00 1.76085472e+00
4.70111400e-01 -3.14242095e-01 -3.01398616e-02 5.54761887e-01
3.58567983e-02 3.99354070e-01 3.61441635e-02 -5.37333548e-01
9.82848465e-01 4.95655477e-01 4.55704063e-01 -8.96757483e-01
-8.77531171e-01 -4.14978683e-01 -7.52716362e-01 -4.13354367e-01
6.55944765e-01 -9.91152376e-02 -9.44016218e-01 1.10086405e+00
2.50933290e-01 -6.07246637e-01 -3.46050262e-01 1.15126431e+00
2.31904924e-01 7.32427299e-01 -2.63616830e-01 -3.98566186e-01
1.57338536e+00 -1.38019085e+00 -6.55168414e-01 -3.76066327e-01
4.64235485e-01 -7.70299256e-01 7.78862357e-01 8.21863055e-01
-1.36070275e+00 -3.74514639e-01 -1.10872471e+00 4.69659381e-02
-6.59036994e-01 -2.79422224e-01 1.11816430e+00 5.49151123e-01
-5.24636745e-01 6.48211777e-01 -6.65599644e-01 -1.07479617e-01
2.30925411e-01 6.43731177e-01 -2.64938086e-01 -4.53895599e-01
-9.98393476e-01 1.01841879e+00 6.07533395e-01 4.01843339e-01
-5.16975045e-01 -4.12047774e-01 -7.35906124e-01 1.78262502e-01
1.12831199e+00 -4.83949840e-01 1.19175613e+00 -2.55085588e-01
-1.26370776e+00 3.52984607e-01 1.49622336e-01 -5.15490353e-01
4.28282529e-01 -2.91414499e-01 9.30518843e-03 -2.36399129e-01
-7.37466738e-02 3.30757588e-01 4.74564046e-01 -1.30771887e+00
-9.36879098e-01 -4.67443913e-01 3.03672433e-01 2.01141775e-01
-1.35181961e-03 -3.39018852e-01 -5.83374381e-01 -4.14071262e-01
4.05145437e-01 -1.07287753e+00 -4.43560541e-01 -5.73999822e-01
-4.82598811e-01 -3.71971369e-01 3.93533260e-01 -3.94238472e-01
8.91640425e-01 -1.88915753e+00 2.10866749e-01 3.01340640e-01
4.66636419e-02 -1.16816275e-01 3.82163338e-02 5.05346477e-01
2.54449338e-01 -1.59749505e-03 3.55258077e-01 -1.05179757e-01
5.24937212e-01 5.24710655e-01 -2.30836719e-01 2.88860738e-01
-1.13096967e-01 9.16600525e-01 -8.93106759e-01 -1.85168102e-01
-1.09775968e-01 -4.10924107e-01 -7.61006057e-01 2.73598850e-01
-5.13102233e-01 -2.95577824e-01 -5.70370138e-01 1.09554362e+00
1.03870368e+00 -2.54251927e-01 4.10868675e-01 -2.34427989e-01
-1.91483274e-02 2.60099053e-01 -1.13624477e+00 1.52349341e+00
-2.89905906e-01 -2.94166327e-01 1.55323431e-01 -1.37787259e+00
1.02830172e+00 -1.76930249e-01 4.50212091e-01 -9.78631973e-01
5.30011505e-02 3.68717909e-01 7.67284334e-02 -4.82761919e-01
6.81231976e-01 2.02721134e-02 -6.40077174e-01 3.61574590e-01
-1.82890043e-01 -1.26051858e-01 5.08849800e-01 -3.38439763e-01
9.33384776e-01 2.15385437e-01 1.74037457e-01 -5.07482350e-01
-2.84009397e-01 4.16360766e-01 7.46417165e-01 8.25223565e-01
-1.50424242e-02 2.18971461e-01 9.20184135e-01 -1.01962090e+00
-9.78776157e-01 -9.96188819e-01 3.35060842e-02 1.32196748e+00
4.77969170e-01 3.87326926e-02 -4.32432503e-01 -8.21364224e-01
7.90299892e-01 6.91304564e-01 -5.22498071e-01 1.64177626e-01
-7.62911677e-01 -1.01432991e+00 -4.23509061e-01 5.67592502e-01
2.33805686e-01 -9.21668291e-01 -5.57698071e-01 4.96149987e-01
7.45649962e-03 -1.15566933e+00 -8.27235401e-01 1.22216928e+00
-9.58484650e-01 -9.40090120e-01 -3.84004593e-01 -9.08312857e-01
8.71769249e-01 8.26114357e-01 1.06723380e+00 -2.89257497e-01
-5.32904863e-02 -4.00639683e-01 -6.13646567e-01 -5.56419015e-01
1.91006929e-01 5.94485760e-01 9.81209725e-02 -5.40497184e-01
5.17247379e-01 -2.44684979e-01 -4.11120057e-01 7.05914974e-01
-4.74401027e-01 2.32539296e-01 9.20910239e-01 9.18401718e-01
6.64389729e-01 5.74579239e-01 5.89609325e-01 -1.05219531e+00
7.09031641e-01 -5.58990240e-01 -8.66907418e-01 3.50582629e-01
-7.15165555e-01 4.76934575e-02 8.49043012e-01 -4.35265809e-01
-5.47806442e-01 -6.74618632e-02 3.38535845e-01 -2.23325133e-01
2.90898263e-01 5.00471830e-01 -1.95890188e-01 2.25844115e-01
-2.24361539e-01 3.85367461e-02 -6.97272941e-02 -4.90591884e-01
-1.35668367e-01 2.98153102e-01 -3.69487293e-02 -8.84461105e-01
6.04574203e-01 -1.02147385e-01 9.26511437e-02 -1.69988975e-01
-1.12772262e+00 -4.22333777e-01 -5.70802867e-01 1.10648856e-01
7.02005029e-01 -5.61216950e-01 -1.32592499e+00 1.79909244e-01
-1.06822062e+00 -6.42644703e-01 4.26633321e-02 2.68232644e-01
-7.91175902e-01 -4.82494794e-02 -7.37604797e-01 -6.19385004e-01
-1.39266580e-01 -1.29241574e+00 8.90065551e-01 1.04311742e-01
1.57874852e-01 -6.64600492e-01 -3.64362597e-01 6.54313982e-01
4.15044725e-01 5.38849086e-02 1.32421303e+00 -1.01877236e+00
-8.47039104e-01 -6.83282241e-02 -3.40768486e-01 -2.83969268e-02
1.39858261e-01 -6.95275366e-01 -1.03749884e-02 -4.59031075e-01
1.07250623e-02 -3.20586830e-01 5.78313529e-01 6.42413557e-01
1.30245411e+00 -3.18262160e-01 -4.76340503e-01 6.76910758e-01
1.44310868e+00 7.46632755e-01 2.72929817e-01 7.35871255e-01
4.56487566e-01 6.92080319e-01 1.09711480e+00 4.57807600e-01
5.29764235e-01 6.66120410e-01 6.04979157e-01 1.72340140e-01
6.21095836e-01 -3.64664078e-01 -6.69147223e-02 9.79359329e-01
-1.90020993e-01 -7.93998957e-01 -6.46861076e-01 3.24439824e-01
-2.09556198e+00 -7.32758999e-01 1.02784619e-01 2.15175176e+00
7.24217057e-01 8.52011025e-01 1.54798135e-01 -1.80446714e-01
6.94325268e-01 -1.37238324e-01 -6.81616724e-01 -7.88862884e-01
9.64614227e-02 -6.42150342e-02 1.14910817e+00 1.76498935e-01
-1.29802191e+00 6.82009518e-01 5.95351219e+00 6.87981844e-01
-6.58302963e-01 -6.17104061e-02 7.19319999e-01 -3.24208021e-01
-1.08926997e-01 -2.56641835e-01 -1.25826776e+00 8.61212969e-01
6.46225452e-01 5.16543835e-02 7.54209340e-01 1.15061605e+00
7.83659741e-02 -2.76181549e-01 -1.29066181e+00 6.77855670e-01
-7.13342950e-02 -1.25214994e+00 -1.16611354e-01 4.81989771e-01
6.73610568e-01 -4.70049590e-01 -1.41317949e-01 4.20461327e-01
4.60822284e-01 -7.05421448e-01 9.25848246e-01 5.22726364e-02
1.59477890e-01 -1.05428421e+00 1.23665071e+00 3.61481041e-01
-9.70686138e-01 -4.26286221e-01 -7.22015798e-01 -7.16127902e-02
5.26919603e-01 7.49434352e-01 -7.44265974e-01 6.11501217e-01
9.54788148e-01 -3.14128064e-02 3.60156633e-02 1.34256768e+00
1.06683597e-01 4.23510596e-02 -3.80626410e-01 -5.04368007e-01
5.62182665e-01 -5.58792055e-01 1.45420849e-01 1.04336894e+00
3.59897882e-01 5.21204919e-02 7.22437203e-01 4.98516470e-01
-2.06281036e-01 -1.41280040e-01 -4.95035887e-01 1.14294067e-02
3.79070163e-01 1.16270471e+00 -1.31531882e+00 1.60997465e-01
-2.51089782e-01 7.59330511e-01 3.87653112e-01 7.17706047e-03
-1.13394725e+00 -4.48183358e-01 2.80495495e-01 1.36643052e-01
1.01158392e+00 -5.23637533e-01 -4.01513934e-01 -5.63929319e-01
2.28558734e-01 -7.29093611e-01 1.75358355e-01 -3.24085444e-01
-1.60289848e+00 3.90957981e-01 1.44731790e-01 -1.03148365e+00
2.47988075e-01 -9.75253761e-01 -2.26689741e-01 4.80719239e-01
-1.58068800e+00 -7.44936526e-01 -1.69874113e-02 7.28648901e-02
7.21904337e-01 -3.98288891e-02 3.69743764e-01 2.25071043e-01
-5.66452324e-01 5.87206900e-01 1.42376348e-01 -2.85038739e-01
6.67571664e-01 -1.49709535e+00 4.87625629e-01 2.03468710e-01
-5.44036210e-01 4.99207377e-01 8.39456320e-01 -7.38992989e-01
-2.41060662e+00 -8.78818333e-01 8.60667884e-01 -5.04691601e-02
6.38214469e-01 -9.19295669e-01 -5.17741323e-01 6.40336037e-01
7.94691145e-02 -1.67035922e-01 5.78909338e-01 4.15390998e-01
-1.29596808e-03 -5.97260594e-01 -1.11514354e+00 2.88265258e-01
1.32365286e+00 4.48164284e-01 -7.03725338e-01 9.17181373e-01
1.00933218e+00 -9.05530572e-01 -8.52415681e-01 2.33514562e-01
4.66627926e-01 -6.90633059e-01 7.46730983e-01 -7.75264680e-01
4.06362683e-01 1.47306219e-01 -3.12367022e-01 -1.46844327e+00
-5.60462713e-01 -8.53606820e-01 -1.69631779e-01 8.11645389e-01
6.23806298e-01 -7.69914925e-01 1.00551605e+00 7.46421158e-01
-5.81308722e-01 -1.38177419e+00 -9.61554110e-01 -1.07561576e+00
-2.69616246e-01 -3.20789143e-02 7.63834059e-01 6.35832667e-01
9.98599231e-02 1.04265183e-01 -3.67247999e-01 1.03242755e-01
6.61422908e-01 7.86132455e-01 6.77679539e-01 -1.01809120e+00
-4.00255561e-01 -3.06582391e-01 1.03980742e-01 -1.40288246e+00
-3.13492209e-01 -7.25753009e-01 3.94533664e-01 -1.64911246e+00
2.96354771e-01 -9.46793079e-01 -3.32774401e-01 6.22897625e-01
2.24220082e-01 -2.81145662e-01 3.34396958e-01 -4.28553261e-02
-9.89974141e-01 1.42820522e-01 1.70647538e+00 -1.12143151e-01
-3.78763288e-01 3.03882241e-01 -1.04175580e+00 2.71213144e-01
7.41970479e-01 -4.99698281e-01 -2.91102052e-01 -6.12072229e-01
5.80970943e-01 4.01166230e-01 -2.40944326e-01 -1.95747018e-01
3.77644092e-01 -4.85502601e-01 4.68289763e-01 -8.15385342e-01
1.34852543e-01 -1.04703271e+00 -6.68547601e-02 5.39138258e-01
4.18933704e-02 5.93667567e-01 9.51322764e-02 6.04212105e-01
2.18340993e-01 -3.94009858e-01 2.04252332e-01 -3.15664411e-01
-4.82870370e-01 3.64784002e-01 -2.12611899e-01 -4.94015552e-02
1.37148333e+00 -2.42159903e-01 -6.94131374e-01 2.33341590e-01
-5.67345679e-01 8.58457267e-01 2.33709484e-01 4.41240281e-01
3.60671103e-01 -1.18278861e+00 -2.74522036e-01 2.39354178e-01
-3.47271681e-01 4.90879387e-01 3.24916720e-01 7.45630682e-01
-6.83896601e-01 8.65831614e-01 -5.41258633e-01 -4.65374231e-01
-5.07329583e-01 1.08859086e+00 -2.54417479e-01 -7.86157668e-01
-5.59184849e-01 7.94755518e-01 -1.97697565e-01 -3.41793269e-01
2.95496970e-01 -6.79893613e-01 2.83150136e-01 3.96691114e-01
8.01750738e-03 3.00762415e-01 4.01544392e-01 3.63930762e-01
-1.58408836e-01 1.86611637e-01 -6.07289732e-01 5.86803019e-01
1.98176706e+00 2.78730273e-01 -1.28798857e-01 2.35956132e-01
9.14989352e-01 -2.87677050e-01 -1.32143307e+00 5.51783778e-02
1.86752841e-01 -6.40047252e-01 -2.18632177e-01 -7.44751215e-01
-1.16453958e+00 3.49589139e-01 -2.62445305e-02 9.37788665e-01
9.80712235e-01 3.09753623e-02 1.13635588e+00 6.90471530e-01
1.13949108e+00 -1.48018909e+00 9.32665393e-02 5.54713070e-01
8.02655935e-01 -1.12229288e+00 2.74286896e-01 -5.22744060e-01
-6.83070481e-01 1.11957479e+00 7.79914618e-01 -4.65557665e-01
4.91632134e-01 5.54404557e-01 -5.79702020e-01 -2.35864028e-01
-9.36503828e-01 9.57136080e-02 -9.49011445e-02 4.11524862e-01
9.14104730e-02 3.15387458e-01 -4.35861140e-01 9.73345399e-01
-3.57998252e-01 -3.02419186e-01 2.92555928e-01 1.23467457e+00
-6.75337791e-01 -1.06699717e+00 -2.89661676e-01 7.82744348e-01
-2.15249956e-01 2.90332168e-01 2.86480099e-01 1.03989279e+00
2.43511871e-01 6.31572962e-01 4.74582046e-01 -2.85909504e-01
6.62051857e-01 -3.67884375e-02 7.61049807e-01 -6.99860513e-01
-5.99715948e-01 1.79660559e-01 2.40743265e-01 -5.64639151e-01
-1.04037352e-01 -5.80267131e-01 -9.71040606e-01 -4.83668566e-01
-5.26138783e-01 2.34373450e-01 7.52642334e-01 9.18162525e-01
2.24952310e-01 4.86437142e-01 8.05149913e-01 -1.22030699e+00
-9.31796610e-01 -5.65593958e-01 -8.94003749e-01 1.26417190e-01
-5.08640036e-02 -1.11287975e+00 -2.22279519e-01 -4.22207803e-01] | [4.978424549102783, 2.697413444519043] |
29c388cb-d027-4d81-b09f-04c0ef91d952 | short-term-aggregated-residential-load | 2302.05033 | null | https://arxiv.org/abs/2302.05033v1 | https://arxiv.org/pdf/2302.05033v1.pdf | Short-Term Aggregated Residential Load Forecasting using BiLSTM and CNN-BiLSTM | Higher penetration of renewable and smart home technologies at the residential level challenges grid stability as utility-customer interactions add complexity to power system operations. In response, short-term residential load forecasting has become an increasing area of focus. However, forecasting at the residential level is challenging due to the higher uncertainties involved. Recently deep neural networks have been leveraged to address this issue. This paper investigates the capabilities of a bidirectional long short-term memory (BiLSTM) and a convolutional neural network-based BiLSTM (CNN-BiLSTM) to provide a day ahead (24 hr.) forecasting at an hourly resolution while minimizing the root mean squared error (RMSE) between the actual and predicted load demand. Using a publicly available dataset consisting of 38 homes, the BiLSTM and CNN-BiLSTM models are trained to forecast the aggregated active power demand for each hour within a 24 hr. span, given the previous 24 hr. load data. The BiLSTM model achieved the lowest RMSE of 1.4842 for the overall daily forecast. In addition, standard LSTM and CNN-LSTM models are trained and compared with the BiLSTM architecture. The RMSE of BiLSTM is 5.60%, 2.85% and 2.60% lower than the LSTM, CNN-LSTM and CNN-BiLSTM models respectively. The source code of this work is available at https://github.com/Varat7v2/STLF-BiLSTM-CNNBiLSTM.git. | ['Xingpeng Li', 'Vysali Gollapudi', 'Raymond I. Fernandez', 'Bharat Bohara'] | 2023-02-10 | null | null | null | null | ['load-forecasting'] | ['miscellaneous'] | [-6.23423338e-01 -2.16463238e-01 1.51971415e-01 -4.40485388e-01
-6.54212534e-01 -2.21050516e-01 3.90096575e-01 -1.97147846e-01
3.20224017e-01 1.19074893e+00 4.21971470e-01 -5.10392487e-01
6.45145448e-03 -1.07875669e+00 -2.84463674e-01 -9.30191457e-01
-3.82535785e-01 8.05160329e-02 -5.71301818e-01 -3.21156114e-01
-2.26742223e-01 2.66146600e-01 -1.18186450e+00 1.32182434e-01
1.16763377e+00 1.42517781e+00 2.71696299e-01 5.12670219e-01
4.49691206e-01 9.26188707e-01 -6.25824869e-01 2.27364361e-01
4.43509445e-02 -2.74040997e-01 -5.47306657e-01 -5.06294072e-01
-2.32579783e-01 -6.61526442e-01 -2.51569360e-01 6.81473076e-01
1.07404447e+00 2.67750353e-01 3.45492661e-01 -1.22278392e+00
-8.23430717e-01 6.13451421e-01 -3.47189248e-01 8.25458825e-01
2.80267954e-01 1.51979953e-01 7.18841434e-01 -6.44882083e-01
-4.26764399e-01 6.62119925e-01 9.30121481e-01 4.12573069e-02
-1.15145576e+00 -6.36282742e-01 4.15778309e-02 6.18867993e-01
-1.24624372e+00 -2.91941345e-01 5.68432331e-01 -4.36635405e-01
1.98172474e+00 2.18907312e-01 8.73395205e-01 1.00630331e+00
7.46442497e-01 4.09303397e-01 1.12577355e+00 -8.40527415e-02
2.17348650e-01 -2.11858451e-01 7.76346698e-02 6.69111982e-02
-4.19050939e-02 3.95426899e-01 -3.82275507e-02 1.55097410e-01
4.12051976e-01 6.54691681e-02 -1.87538773e-01 6.89632833e-01
-7.99256504e-01 6.88465655e-01 7.61597812e-01 9.11553144e-01
-7.79777884e-01 6.77458569e-03 6.38453543e-01 3.56239229e-01
1.02460957e+00 -1.61119565e-01 -8.33551109e-01 -3.34992647e-01
-1.19337344e+00 -1.29571363e-01 9.69072461e-01 6.63878679e-01
2.88671046e-01 1.14376366e+00 -1.56930879e-01 7.65211046e-01
2.06475541e-01 7.25252569e-01 1.00286663e+00 -3.77386332e-01
3.26581359e-01 1.47270545e-01 -2.43859440e-02 -8.28730762e-01
-7.42940545e-01 -9.21915174e-01 -1.42900944e+00 6.51210323e-02
-1.67247519e-01 -7.13380337e-01 -5.97423434e-01 1.33942008e+00
-1.81658313e-01 1.40565619e-01 -2.15377435e-02 3.11847299e-01
6.37599766e-01 1.14725399e+00 1.12832285e-01 -2.51292467e-01
1.01312029e+00 -9.83904600e-01 -1.06766677e+00 -5.57192750e-02
7.70765901e-01 -6.74581289e-01 4.44334328e-01 1.23794228e-01
-1.02367139e+00 -6.15190506e-01 -1.11645055e+00 1.59041852e-01
-7.67516673e-01 2.58458078e-01 -1.14011340e-01 4.43072170e-01
-1.22975695e+00 7.94480085e-01 -8.35844815e-01 -5.28302252e-01
2.40937084e-01 1.66011199e-01 2.21886918e-01 5.35131454e-01
-1.60412097e+00 1.43446767e+00 5.25795102e-01 6.58845961e-01
-7.02881753e-01 -9.49436069e-01 -8.21120381e-01 3.36118311e-01
-4.53726172e-01 -5.55962741e-01 1.49007702e+00 -7.79074371e-01
-1.53625929e+00 -1.01268344e-01 -1.27588168e-01 -1.13139904e+00
3.88076097e-01 -1.22212157e-01 -9.47787046e-01 -6.52629137e-01
2.43285559e-02 8.25654343e-02 4.53671217e-01 -7.90838718e-01
-9.28635418e-01 -3.18446279e-01 -5.31055272e-01 9.44236964e-02
-3.00532263e-02 -2.68582553e-01 9.02499735e-01 -7.34716117e-01
-2.35935137e-01 -5.33752263e-01 4.77373488e-02 -9.84562695e-01
-4.20473754e-01 -4.83508676e-01 1.22693539e+00 -1.49450445e+00
1.30541968e+00 -1.63175142e+00 -6.96702182e-01 -8.17570761e-02
-4.23002094e-01 2.74489164e-01 3.59098494e-01 7.49203026e-01
-5.89041293e-01 -4.51857410e-02 -1.88750587e-02 -3.10051560e-01
3.40175122e-01 3.43225569e-01 -5.23161709e-01 3.57500494e-01
-1.37955084e-01 1.41326058e+00 -5.97369134e-01 3.44844013e-01
7.43449092e-01 7.46220946e-01 3.82341295e-01 1.46061823e-01
-2.06327122e-02 3.92859608e-01 -3.84175666e-02 5.31674922e-01
6.42384291e-01 -2.13569522e-01 -4.77562435e-02 -3.32485080e-01
-5.45241594e-01 5.33907115e-01 -7.30606496e-01 1.31708324e+00
-8.88919413e-01 8.61003458e-01 -4.75284368e-01 -1.28247821e+00
9.00089860e-01 8.90004635e-01 5.94026864e-01 -1.38079488e+00
6.20503165e-02 4.07410145e-01 -1.14632227e-01 -3.32890987e-01
1.54634923e-01 -2.43766740e-01 2.38271281e-01 4.49080229e-01
-2.41330601e-02 3.76564741e-01 1.17836073e-01 -5.16136169e-01
6.88905358e-01 1.01375900e-01 2.79277533e-01 -5.99622548e-01
3.39510828e-01 -3.80506366e-01 6.43509448e-01 1.32807150e-01
-1.24560997e-01 1.49878427e-01 -2.38305867e-01 -7.66388655e-01
-1.10945940e+00 -7.93114901e-01 -3.85156095e-01 9.00702715e-01
-7.75383174e-01 2.17989042e-01 -3.75797987e-01 -1.74606353e-01
6.26031607e-02 1.50723183e+00 -5.51605642e-01 1.13655746e-01
-5.09915650e-01 -1.16520345e+00 3.54984164e-01 8.75770390e-01
9.35777545e-01 -1.11024714e+00 -7.63345540e-01 7.16659248e-01
-2.27478132e-01 -8.07450294e-01 -1.77467957e-01 4.13006872e-01
-7.33052433e-01 -5.54363728e-01 -8.05753946e-01 -6.84156716e-01
1.88340545e-01 -2.63140976e-01 1.39694381e+00 -3.39888275e-01
4.55028638e-02 -5.36152758e-02 -1.15334675e-01 -4.13311452e-01
1.49602011e-01 4.17435288e-01 9.08506215e-02 -5.07349491e-01
4.46988344e-01 -1.22472537e+00 -7.53194332e-01 9.74219963e-02
-3.10070246e-01 1.77569285e-01 1.52305424e-01 9.02948499e-01
1.87748700e-01 3.79096031e-01 1.42065692e+00 -1.27754256e-01
6.26481891e-01 -9.42800224e-01 -6.99371636e-01 -6.09160820e-03
-1.22732687e+00 -4.82390910e-01 1.00263596e+00 4.13726009e-02
-1.09606528e+00 -4.50281531e-01 -4.68055815e-01 -3.52373607e-02
-2.02953443e-01 7.24167645e-01 1.55467629e-01 3.19737226e-01
-6.61020502e-02 4.17186916e-01 -5.58198512e-01 -6.78318501e-01
-3.99474204e-02 6.89959407e-01 2.09794089e-01 1.27173411e-02
5.66162169e-01 6.62461594e-02 -3.39920044e-01 -6.90840602e-01
-8.45414996e-01 2.15391368e-02 -5.68992138e-01 -2.61189491e-01
9.20620143e-01 -1.27568996e+00 -7.83186972e-01 7.58422494e-01
-1.01868033e+00 -8.06537151e-01 -4.87258017e-01 2.62481302e-01
-4.07144070e-01 -1.17446072e-01 -7.80117750e-01 -8.92827213e-01
-1.17018330e+00 -8.81676614e-01 4.72307503e-01 3.73598337e-01
-1.74913451e-01 -1.56942868e+00 -1.20738581e-01 -7.98385590e-03
1.13022971e+00 6.89214408e-01 9.07712936e-01 -6.10661447e-01
7.37546384e-02 -3.32501233e-01 -2.05100179e-01 7.63352513e-01
5.31402946e-01 -2.56338775e-01 -1.25977230e+00 -7.01184094e-01
1.85783207e-01 1.19929753e-01 1.91565886e-01 8.53309810e-01
9.26813424e-01 -7.63716936e-01 7.43467584e-02 4.65348810e-01
1.74745822e+00 7.92017817e-01 7.04340100e-01 6.18991852e-01
4.73851502e-01 1.84795577e-02 -1.29089907e-01 7.08639205e-01
1.00139868e+00 2.62641340e-01 3.52059633e-01 -2.91905016e-01
5.93037903e-02 -9.36036557e-02 4.37128156e-01 1.22742021e+00
8.39566812e-02 -2.48165682e-01 -9.53048527e-01 7.52821147e-01
-1.85593414e+00 -1.16823351e+00 -2.55622029e-01 1.91550922e+00
4.56113309e-01 5.96407726e-02 2.35803574e-02 5.25159538e-01
4.06978488e-01 2.40807652e-01 -6.90742612e-01 -7.69363582e-01
-2.90822417e-01 1.80522159e-01 6.06087446e-01 4.30948615e-01
-8.97980928e-01 1.62191346e-01 5.43982601e+00 6.46104097e-01
-1.14957881e+00 3.85233015e-01 1.13550544e+00 8.79727900e-02
1.63849354e-01 -3.44471484e-01 -6.60214901e-01 1.01383018e+00
1.71800232e+00 -7.66602039e-01 4.29251224e-01 7.07024515e-01
1.04471743e+00 -3.75346690e-01 -9.92626071e-01 7.96619892e-01
-2.44960830e-01 -9.92758512e-01 -2.70735621e-01 2.79670209e-01
1.22371495e+00 5.96676528e-01 3.41599733e-02 5.23178577e-01
4.25047129e-01 -1.24356866e+00 4.56544220e-01 8.71298730e-01
3.58998567e-01 -9.33801532e-01 1.35430753e+00 3.61671776e-01
-1.52295160e+00 -4.90250647e-01 -1.12249635e-01 -4.96482044e-01
5.47826886e-01 1.06421053e+00 -5.76103568e-01 9.30421352e-01
1.34742498e+00 1.03613901e+00 -3.23156804e-01 7.76404738e-01
-1.51728364e-02 7.51686335e-01 -3.93857211e-01 4.00854856e-01
2.83997446e-01 -2.32263893e-01 1.57510087e-01 9.89566803e-01
8.03234100e-01 -9.92789865e-02 1.41433403e-01 6.40586138e-01
1.29402727e-01 -3.04927528e-01 -4.95232195e-01 3.28723729e-01
4.19871897e-01 1.21850812e+00 1.83131453e-02 -5.37020326e-01
-3.31878543e-01 7.57500470e-01 -2.18558118e-01 5.55965185e-01
-8.05141091e-01 -3.19571704e-01 5.39938331e-01 -1.34318635e-01
3.06616694e-01 -1.53978407e-01 -6.85006440e-01 -9.62202787e-01
-4.11093757e-02 -4.42086816e-01 4.23198700e-01 -1.05510950e+00
-1.69998777e+00 7.10140407e-01 -1.59854591e-01 -1.16687620e+00
-8.64972293e-01 -7.18955621e-02 -1.23786807e+00 1.54628253e+00
-1.98663449e+00 -1.15673292e+00 -4.07691151e-01 3.50946277e-01
9.06057537e-01 -6.21992089e-02 1.11755443e+00 5.98943353e-01
-9.79374111e-01 1.91556260e-01 8.27807426e-01 7.08296672e-02
-5.72790354e-02 -1.51198018e+00 6.48163557e-01 8.35086942e-01
-7.01900959e-01 -6.02548830e-02 7.39790082e-01 -3.99763167e-01
-9.63955343e-01 -1.39034677e+00 1.08907294e+00 9.72630307e-02
7.99014747e-01 4.88929786e-02 -8.77638578e-01 1.10275114e+00
1.09126759e+00 -1.63807362e-01 6.81217551e-01 -2.92360395e-01
3.21184397e-01 -5.65630734e-01 -1.32198679e+00 -2.03411188e-02
2.27645323e-01 -5.28825581e-01 -3.62945408e-01 2.13530824e-01
2.58759052e-01 -1.55458614e-01 -1.66609144e+00 5.93900800e-01
5.82686365e-01 -8.98528874e-01 5.87013721e-01 1.13746427e-01
-8.67965980e-04 -1.62045166e-01 -3.62387449e-01 -1.95776331e+00
-5.46315253e-01 -3.10682535e-01 -5.41214883e-01 1.28255284e+00
4.79781508e-01 -1.30108583e+00 2.28703648e-01 5.18329620e-01
-3.17365289e-01 -8.89571369e-01 -1.30074167e+00 -8.36192489e-01
2.82607585e-01 -2.90986478e-01 1.07434893e+00 9.62252557e-01
-1.31003493e-02 2.80295968e-01 -2.77114809e-01 3.54235113e-01
6.91721797e-01 1.11229874e-01 6.11334220e-02 -1.10268748e+00
3.80234867e-01 -4.25565869e-01 -4.32159007e-02 -5.94615817e-01
6.73378929e-02 -6.74702942e-01 -2.50413597e-01 -2.28444028e+00
-3.66502494e-01 -5.99594042e-02 -6.48643792e-01 8.67900431e-01
4.36579615e-01 6.21741302e-02 1.32926345e-01 -6.71867356e-02
2.88813949e-01 1.03012621e+00 5.89744747e-01 -2.30847806e-01
-3.87393497e-02 1.56634524e-01 -1.81461304e-01 4.70822245e-01
1.74297941e+00 1.44955352e-01 -1.51394725e-01 -6.05332732e-01
-7.01818615e-02 6.92443475e-02 2.50556022e-01 -1.34976935e+00
1.88123584e-01 3.57818365e-01 1.11114848e+00 -1.18015087e+00
8.34817365e-02 -8.41424823e-01 6.42127752e-01 7.39394724e-01
8.39420035e-02 8.18539083e-01 4.92906928e-01 3.83670703e-02
-1.93927158e-02 3.24034482e-01 5.74495792e-01 -8.36836360e-03
-4.95761275e-01 2.43580312e-01 -7.75454223e-01 -6.00973785e-01
8.85309994e-01 -1.42972425e-01 -5.50778151e-01 -5.01458883e-01
-9.70882237e-01 6.71400011e-01 -2.50933886e-01 5.05502880e-01
3.46536994e-01 -1.59498191e+00 -7.48594582e-01 2.72105008e-01
-5.78210831e-01 -2.47128189e-01 5.44579268e-01 9.55021024e-01
-8.05767477e-02 1.02480817e+00 -1.07903130e-01 -3.58134329e-01
-6.77841008e-01 1.02373540e-01 1.01010203e+00 -3.88974905e-01
-5.55410981e-01 2.62128264e-01 -2.72899508e-01 -3.09072942e-01
6.44088984e-02 -8.10505390e-01 -3.98951739e-01 4.71706837e-01
3.76127332e-01 9.61266994e-01 6.72166407e-01 -7.71735966e-01
-1.00555770e-01 2.05875218e-01 3.14336121e-01 3.50773811e-01
1.73378229e+00 -4.74663436e-01 -3.41292322e-01 9.93509412e-01
1.23243475e+00 -1.04652870e+00 -8.64494145e-01 4.65866476e-02
2.98914999e-01 8.85735527e-02 5.90560734e-01 -1.27321911e+00
-1.30341101e+00 7.22611248e-01 1.23867381e+00 9.68368828e-01
1.40845072e+00 -7.03898013e-01 1.33897150e+00 -3.63836400e-02
2.54699469e-01 -1.18206501e+00 -6.80961430e-01 7.54564345e-01
9.58668947e-01 -1.08107400e+00 -2.89499342e-01 6.07913077e-01
-1.24880768e-01 1.16789317e+00 4.90565360e-01 -5.08500412e-02
1.18246686e+00 2.76308298e-01 1.32303806e-02 1.64998770e-01
-1.10850120e+00 -1.02203125e-02 4.75233532e-02 3.62879097e-01
6.82256877e-01 4.38285261e-01 -1.17071895e-02 4.74118799e-01
-6.10230625e-01 2.54657805e-01 2.00230852e-01 7.30681956e-01
-6.84433207e-02 -4.98814970e-01 -9.63364691e-02 8.51765096e-01
-7.13306963e-01 -2.57076859e-01 7.52272129e-01 2.56108880e-01
1.49089113e-01 1.38073123e+00 3.49890620e-01 -1.54838279e-01
5.02190650e-01 1.47673011e-01 -9.59074870e-02 -1.73472077e-01
-8.38884294e-01 1.12556554e-01 -2.93304473e-02 -2.63285577e-01
-3.08790863e-01 -5.40876389e-01 -1.09442449e+00 -7.50086725e-01
-2.71617025e-01 2.23724872e-01 7.58731544e-01 1.08860803e+00
2.50837982e-01 1.01909435e+00 1.06414449e+00 -1.23933280e+00
-5.07841408e-01 -1.54013181e+00 -7.53802121e-01 -2.88769424e-01
5.50782561e-01 -9.65678096e-02 -4.23816383e-01 -4.46338095e-02] | [6.177830219268799, 2.8071658611297607] |
92adc197-c823-4ac2-a29a-2f9bdce670f6 | automated-audio-captioning-using-transfer | 2108.04692 | null | https://arxiv.org/abs/2108.04692v1 | https://arxiv.org/pdf/2108.04692v1.pdf | Automated Audio Captioning using Transfer Learning and Reconstruction Latent Space Similarity Regularization | In this paper, we examine the use of Transfer Learning using Pretrained Audio Neural Networks (PANNs), and propose an architecture that is able to better leverage the acoustic features provided by PANNs for the Automated Audio Captioning Task. We also introduce a novel self-supervised objective, Reconstruction Latent Space Similarity Regularization (RLSSR). The RLSSR module supplements the training of the model by minimizing the similarity between the encoder and decoder embedding. The combination of both methods allows us to surpass state of the art results by a significant margin on the Clotho dataset across several metrics and benchmarks. | ['Eng Siong Chng', 'Fuzhao Xue', 'Andrew Koh'] | 2021-08-10 | null | null | null | null | ['audio-captioning'] | ['audio'] | [ 5.65801501e-01 2.55063146e-01 1.53723925e-01 -3.70870203e-01
-1.18159699e+00 -3.75909507e-01 6.33004904e-01 -1.00292824e-01
-2.57018715e-01 3.56373042e-01 6.64381325e-01 1.66280016e-01
-2.62573478e-03 -4.16216046e-01 -9.17127788e-01 -3.81800711e-01
-1.70843259e-01 2.05972970e-01 -7.26355910e-02 -2.48910546e-01
-1.34643599e-01 1.51992127e-01 -1.51054502e+00 4.27772939e-01
4.74163383e-01 1.38641417e+00 1.47721827e-01 6.46541476e-01
3.39586735e-01 8.63252580e-01 -3.70878756e-01 -1.67000189e-01
1.73707262e-01 -4.12454814e-01 -7.99091756e-01 -1.50082245e-01
6.17405832e-01 -3.23562384e-01 -4.33032453e-01 8.10233116e-01
5.65426350e-01 3.00535291e-01 5.51443636e-01 -1.19379616e+00
-6.73124969e-01 1.05019343e+00 -6.24754056e-02 3.75681072e-02
5.05912840e-01 8.27763379e-02 1.61735773e+00 -8.61654997e-01
4.24428731e-01 1.28914583e+00 8.31906497e-01 5.37507474e-01
-1.35722601e+00 -7.60038197e-01 -1.09712668e-01 2.67800033e-01
-1.37502635e+00 -8.11614096e-01 1.07299805e+00 -3.95608485e-01
9.54735875e-01 1.69407442e-01 3.75191420e-01 1.51011491e+00
-1.70580611e-01 9.01317418e-01 9.03438866e-01 -4.96753693e-01
2.20490277e-01 1.63221374e-01 -1.56069562e-01 3.28124315e-01
-4.79671270e-01 2.90480494e-01 -9.39502656e-01 -1.82073578e-01
4.69352961e-01 -5.63692391e-01 -2.34909654e-01 -4.87044215e-01
-1.16395402e+00 8.67681324e-01 3.50996703e-01 3.31420600e-01
-2.57881731e-01 4.92382079e-01 5.68840504e-01 3.62475246e-01
5.28986096e-01 6.93491399e-01 -4.40230072e-01 -4.22780484e-01
-1.03910077e+00 8.30284506e-02 6.92752779e-01 7.18170822e-01
5.32330275e-01 4.23081189e-01 -2.19661370e-01 1.11104500e+00
5.10427952e-01 3.09916764e-01 5.43934524e-01 -1.19209456e+00
4.28169072e-01 7.00437799e-02 -1.59385040e-01 -7.58095205e-01
-1.04349308e-01 -4.51343447e-01 -6.03845000e-01 -1.27873868e-01
2.21844949e-02 -1.94509238e-01 -7.68983424e-01 2.03525925e+00
-9.69813243e-02 4.23070610e-01 1.88658297e-01 8.28282595e-01
7.10619986e-01 1.02665186e+00 6.88335449e-02 2.27114886e-01
9.02903259e-01 -1.41690361e+00 -8.41011345e-01 -3.27021360e-01
3.93181890e-01 -7.08818078e-01 1.32973373e+00 3.49553168e-01
-1.16611910e+00 -6.70894921e-01 -1.40388978e+00 -1.78241450e-02
-7.80767426e-02 2.67729670e-01 4.37491328e-01 2.60940611e-01
-1.20817578e+00 8.74417543e-01 -9.48348522e-01 -3.24757278e-01
1.53877556e-01 3.81066114e-01 -2.83717602e-01 2.71521837e-01
-1.36610734e+00 6.12497449e-01 2.44857356e-01 5.23857214e-02
-1.29417682e+00 -8.34742367e-01 -1.06875336e+00 3.40502679e-01
2.48006195e-01 -4.92659092e-01 1.52230227e+00 -1.01695073e+00
-1.94630349e+00 3.90618831e-01 2.50066727e-01 -7.90835321e-01
3.35267782e-01 -5.93784750e-01 -4.43578929e-01 2.33838931e-01
-1.21279299e-01 1.03591871e+00 1.07728803e+00 -1.01783657e+00
-1.21221356e-01 1.81973681e-01 -1.56168193e-01 1.25374198e-01
-5.66752076e-01 1.68076679e-02 -2.16757342e-01 -9.79331255e-01
-3.69012028e-01 -1.11481774e+00 5.36456192e-03 -5.47171161e-02
-3.88830841e-01 -2.49716282e-01 7.94927061e-01 -7.93752789e-01
1.12110114e+00 -2.50611186e+00 5.01633286e-01 1.04728088e-01
-1.00605473e-01 2.60637254e-01 -5.86580336e-01 5.60081840e-01
-2.43662089e-01 -1.67291746e-01 -5.11984110e-01 -7.43034244e-01
2.63486177e-01 1.04821555e-01 -7.08519518e-01 9.87510234e-02
2.51035601e-01 7.02141464e-01 -6.63279593e-01 -1.39289513e-01
3.19191925e-02 6.74901545e-01 -7.22795010e-01 5.62555432e-01
-3.46690506e-01 3.50542158e-01 1.32633641e-01 3.32034349e-01
2.10529506e-01 -1.88609794e-01 4.57129162e-03 -1.34806290e-01
9.78827775e-02 8.05036426e-01 -1.04971838e+00 2.13372707e+00
-6.51678681e-01 9.78346109e-01 3.81692648e-02 -8.80092740e-01
9.69399929e-01 6.50895357e-01 5.31359792e-01 -3.72264087e-01
1.48298300e-03 2.26199448e-01 -1.85658574e-01 -3.16706717e-01
1.50529727e-01 -1.81411326e-01 3.21339667e-02 4.68379378e-01
6.59327805e-01 -6.40960559e-02 -2.32767031e-01 1.40181437e-01
1.10798657e+00 2.39082649e-01 2.02275753e-01 -8.32101982e-03
6.08832181e-01 -5.82176566e-01 2.12428793e-01 6.50059521e-01
-8.19595978e-02 7.64774501e-01 3.44575703e-01 -6.23910949e-02
-1.14895928e+00 -1.13643575e+00 9.14115682e-02 1.36303782e+00
-4.24330294e-01 -7.89341867e-01 -7.48618126e-01 -5.63037753e-01
-1.29939970e-02 8.27565908e-01 -6.55183852e-01 -3.77800047e-01
-4.59187984e-01 -2.80063540e-01 9.59860921e-01 8.22279990e-01
1.39785215e-01 -1.14751232e+00 -2.48821676e-01 2.66485542e-01
-3.39349687e-01 -1.34295928e+00 -6.49804771e-01 3.94637346e-01
-6.39340103e-01 -3.42833430e-01 -6.90167546e-01 -7.18029857e-01
-9.38156769e-02 -1.96524486e-01 8.29008937e-01 -4.28682595e-01
-1.29030235e-02 5.88261664e-01 -4.39696461e-01 -3.59458596e-01
-5.92553914e-01 4.17986810e-01 1.63107619e-01 2.49311477e-01
1.48030311e-01 -1.03228045e+00 -1.73626497e-01 1.34560406e-01
-8.37093174e-01 -1.12369008e-01 5.23663521e-01 6.67257488e-01
4.92098361e-01 -5.86986661e-01 7.38675475e-01 -3.13918024e-01
6.13805652e-01 -4.37513262e-01 -4.95307744e-01 4.62494530e-02
-5.93594313e-01 3.11357766e-01 6.05391502e-01 -5.74844003e-01
-7.79933095e-01 1.45642266e-01 -4.86290842e-01 -6.69428289e-01
-1.69012919e-01 3.94800812e-01 -1.27119943e-01 9.53618363e-02
4.09966886e-01 5.50468117e-02 6.43702075e-02 -9.08690810e-01
4.97586191e-01 8.70369792e-01 6.60036564e-01 -2.69681334e-01
8.08348954e-01 3.31173152e-01 -2.67503500e-01 -9.06763494e-01
-1.10507178e+00 -3.29735458e-01 -5.07519364e-01 -1.21515781e-01
9.24177885e-01 -1.00461411e+00 -3.94844055e-01 2.79746652e-01
-1.19585085e+00 -4.81439352e-01 -6.09461665e-01 5.60686707e-01
-9.35975075e-01 2.36272842e-01 -7.12567031e-01 -6.70035839e-01
-4.05267298e-01 -1.09189856e+00 1.15382326e+00 -2.89622813e-01
-4.24980760e-01 -8.44742298e-01 5.00094295e-01 2.68173397e-01
6.14326537e-01 3.79792526e-02 8.44039321e-01 -1.02613413e+00
-4.74749655e-01 -8.05266481e-03 3.89558971e-02 9.40643787e-01
-2.93184258e-02 -2.59214967e-01 -1.48197842e+00 -2.73767591e-01
8.77449661e-02 -6.60988569e-01 1.14787829e+00 2.79970109e-01
1.17700171e+00 -3.69393677e-01 1.91540599e-01 7.46351719e-01
1.04254031e+00 -9.39593241e-02 4.34207588e-01 3.45656663e-01
6.28529966e-01 5.94951987e-01 2.23282158e-01 2.92500317e-01
3.24146956e-01 1.05803752e+00 3.99666339e-01 5.96756162e-03
-4.66847330e-01 -6.25217080e-01 8.23675990e-01 1.32870710e+00
1.24376312e-01 -1.61546767e-02 -8.53794396e-01 5.23271978e-01
-1.85141194e+00 -5.70679009e-01 3.70598257e-01 1.81726742e+00
9.25301731e-01 5.61483540e-02 6.43628240e-02 3.59286934e-01
3.60517085e-01 4.62813526e-01 -3.08264017e-01 -4.96048421e-01
1.13662547e-02 2.63703495e-01 1.25082716e-01 5.62484264e-01
-1.28833389e+00 8.24475408e-01 7.38759899e+00 7.58261502e-01
-9.96782899e-01 2.38493890e-01 9.81441438e-02 -8.49463716e-02
-1.64873853e-01 -1.40011013e-01 -4.61002320e-01 2.75223941e-01
1.57451189e+00 1.23294346e-01 8.35669041e-01 7.83308685e-01
-8.29544589e-02 3.98454040e-01 -1.57615948e+00 9.71944571e-01
3.53657424e-01 -1.18133068e+00 6.50825649e-02 -2.35115420e-02
5.53152859e-01 2.13741452e-01 4.89202976e-01 3.64056528e-01
3.79025447e-03 -9.49925959e-01 8.72765899e-01 4.51605231e-01
6.41724765e-01 -5.60519516e-01 7.57334828e-01 -1.15627035e-01
-1.10052264e+00 -1.70857251e-01 -8.70898142e-02 -7.09104165e-02
2.86374658e-01 1.47552803e-01 -9.55095351e-01 4.31367755e-01
7.24424481e-01 8.61220181e-01 -5.57039082e-01 9.43550467e-01
-2.49394327e-01 1.14419031e+00 -3.70199740e-01 2.80737400e-01
4.54317003e-01 4.95617427e-02 9.83269632e-01 1.15303934e+00
1.79807439e-01 -6.54242456e-01 -1.05616882e-01 9.58806157e-01
-3.20817828e-01 2.13737592e-01 -5.85285485e-01 -2.72784531e-01
2.51865327e-01 9.17522013e-01 -1.45296752e-01 -1.25297025e-01
-2.89296478e-01 9.37749982e-01 4.27345842e-01 1.91488340e-01
-7.89200425e-01 -5.11106074e-01 5.61740279e-01 -4.20827605e-02
5.12220383e-01 -2.91968852e-01 -3.69515829e-02 -1.03393614e+00
-3.94220240e-02 -9.49420094e-01 2.41204709e-01 -8.26161563e-01
-1.36530876e+00 7.63767898e-01 4.65094149e-02 -1.10225916e+00
-5.99446714e-01 -3.58170003e-01 -4.39441532e-01 5.87509871e-01
-1.62608588e+00 -1.24066734e+00 -7.98583254e-02 3.47372860e-01
5.32216549e-01 -3.63174707e-01 1.13289285e+00 4.90444213e-01
-3.53847355e-01 7.31938064e-01 1.40791222e-01 1.12753429e-01
9.71255660e-01 -1.39592636e+00 5.51815689e-01 4.29068267e-01
5.94004989e-01 3.11859816e-01 8.76985550e-01 -1.22668892e-01
-1.02290356e+00 -1.06554019e+00 8.24531615e-01 -4.58183408e-01
9.59012747e-01 -7.76846528e-01 -9.47537243e-01 9.08369064e-01
4.81312960e-01 -1.56516105e-01 1.04968929e+00 2.50693500e-01
-6.45460308e-01 -3.26694310e-01 -5.62204897e-01 3.22994649e-01
7.06437588e-01 -1.08900785e+00 -8.49735141e-01 6.71018660e-02
1.17081022e+00 -1.37223318e-01 -8.08835149e-01 4.34922546e-01
5.60506821e-01 -5.83179533e-01 9.64726865e-01 -6.68611050e-01
5.86284995e-01 3.17953415e-02 -5.23542941e-01 -1.35601866e+00
-4.58026230e-02 -1.05574989e+00 -4.45939839e-01 1.41236746e+00
5.82494080e-01 -2.02498496e-01 4.22319263e-01 1.01656273e-01
-3.67053390e-01 -5.51660657e-01 -1.15854001e+00 -1.00387943e+00
-1.69392154e-02 -7.12019742e-01 5.19058526e-01 6.53166771e-01
6.64010346e-02 6.08751357e-01 -8.58012438e-01 1.06687695e-02
5.02008617e-01 -2.48665884e-01 5.36781371e-01 -1.25412536e+00
-4.46164370e-01 -6.05810769e-02 -4.61235702e-01 -1.03314424e+00
4.64499325e-01 -1.09450924e+00 2.85414785e-01 -1.26231790e+00
-1.08986102e-01 -1.83055438e-02 -6.27151847e-01 5.83130360e-01
2.14944571e-01 5.06462872e-01 4.96252537e-01 2.03779459e-01
-8.34226131e-01 1.01465499e+00 6.20121241e-01 -1.83350235e-01
-4.38419282e-01 -8.97083506e-02 -5.32145381e-01 8.45103979e-01
7.53632486e-01 -5.27559996e-01 -1.64645940e-01 -6.31141961e-01
7.14149922e-02 -1.28326997e-01 4.09042954e-01 -1.40512359e+00
1.73785016e-01 4.75919366e-01 -9.58528668e-02 -3.61215293e-01
7.79671788e-01 -8.19155455e-01 -1.08568743e-01 1.28947154e-01
-9.96217668e-01 -1.42794907e-01 3.69352639e-01 6.58226550e-01
-4.88469660e-01 -2.30099231e-01 7.95450091e-01 3.29347014e-01
-2.12323472e-01 -1.23245396e-01 -4.81282949e-01 -3.98591012e-02
5.15525281e-01 2.06590638e-01 1.00437477e-01 -7.93029487e-01
-8.89850378e-01 8.46752315e-04 -1.38001600e-02 7.23466992e-01
5.61098814e-01 -1.69334590e+00 -6.65896297e-01 2.42857620e-01
1.79853901e-01 -4.52938795e-01 2.42655948e-02 7.51229346e-01
-5.09936735e-02 7.89801002e-01 -1.87229291e-01 -5.96598327e-01
-1.12048078e+00 3.19787383e-01 3.28332573e-01 -2.45073438e-01
-5.87674379e-01 8.25268865e-01 -1.00814980e-02 -5.34165144e-01
7.03440726e-01 -3.41296375e-01 -1.06794514e-01 9.73736495e-02
4.13911343e-01 4.15296465e-01 9.51797441e-02 -7.11021125e-01
-1.73799589e-01 1.52544439e-01 -1.04398746e-03 -4.89909559e-01
1.70183349e+00 -8.64971578e-02 1.63912341e-01 8.64690304e-01
1.44730234e+00 2.74879765e-02 -1.42647183e+00 -3.34805071e-01
1.69179127e-01 -2.84197889e-02 1.56865776e-01 -8.95236135e-01
-6.61451876e-01 1.16530728e+00 7.22542405e-01 1.78500786e-01
1.03376555e+00 2.28003800e-01 9.35398579e-01 4.42669511e-01
-1.25960007e-01 -1.13724947e+00 5.11146605e-01 6.52252257e-01
1.09873748e+00 -9.38389301e-01 -3.73123109e-01 -1.78095162e-01
-7.71536171e-01 1.03160512e+00 2.89220214e-01 -2.93507606e-01
6.61976635e-01 3.22852165e-01 -1.04153603e-02 6.58855364e-02
-9.93264318e-01 -4.52163890e-02 7.42040992e-01 5.85899115e-01
3.59240651e-01 -1.29623964e-01 1.83872536e-01 9.52496886e-01
-3.99937004e-01 -1.20293066e-01 3.46794099e-01 6.26828909e-01
-2.09786296e-01 -1.16464233e+00 -8.33499581e-02 5.84851503e-02
-5.04163504e-01 -1.10002503e-01 -5.91817856e-01 4.31841522e-01
-1.77883402e-01 8.96995127e-01 4.04708311e-02 -6.96652412e-01
3.47220391e-01 3.49750400e-01 2.28426591e-01 -5.75345457e-01
-6.30663753e-01 3.76414269e-01 7.73392245e-02 -6.52228117e-01
-5.32185614e-01 -5.65056980e-01 -7.50400782e-01 4.18492198e-01
-3.92401546e-01 2.66479433e-01 7.47456253e-01 9.05185103e-01
2.39125371e-01 7.76498616e-01 6.84334993e-01 -9.65533733e-01
-9.65248585e-01 -1.20346045e+00 -4.57657844e-01 2.23979056e-01
7.49066532e-01 -5.33361316e-01 -5.67977250e-01 1.23635791e-01] | [15.252867698669434, 5.146806716918945] |
03dc1807-cabe-4786-91f9-23013f00e9a1 | a-unifying-framework-for-spectrum-preserving | null | null | http://papers.nips.cc/paper/8989-a-unifying-framework-for-spectrum-preserving-graph-sparsification-and-coarsening | http://papers.nips.cc/paper/8989-a-unifying-framework-for-spectrum-preserving-graph-sparsification-and-coarsening.pdf | A Unifying Framework for Spectrum-Preserving Graph Sparsification and Coarsening | How might one ``reduce'' a graph?
That is, generate a smaller graph that preserves the global structure at the expense of discarding local details?
There has been extensive work on both graph sparsification (removing edges) and graph coarsening (merging nodes, often by edge contraction); however, these operations are currently treated separately.
Interestingly, for a planar graph, edge deletion corresponds to edge contraction in its planar dual (and more generally, for a graphical matroid and its dual).
Moreover, with respect to the dynamics induced by the graph Laplacian (e.g., diffusion), deletion and contraction are physical manifestations of two reciprocal limits: edge weights of $0$ and $\infty$, respectively.
In this work, we provide a unifying framework that captures both of these operations, allowing one to simultaneously sparsify and coarsen a graph while preserving its large-scale structure.
The limit of infinite edge weight is rarely considered, as many classical notions of graph similarity diverge. However, its algebraic, geometric, and physical interpretations are reflected in the Laplacian pseudoinverse $\mat{L}^\dagger$, which remains finite in this limit.
Motivated by this insight, we provide a probabilistic algorithm that reduces graphs while preserving $\mat{L}^\dagger$, using an unbiased procedure that minimizes its variance.
We compare our algorithm with several existing sparsification and coarsening algorithms using real-world datasets, and demonstrate that it more accurately preserves the large-scale structure. | ['Gecia Bravo Hermsdorff', 'Lee Gunderson'] | 2019-12-01 | null | null | null | neurips-2019-12 | ['graph-similarity'] | ['graphs'] | [ 4.85768735e-01 5.89982033e-01 -2.17769686e-02 1.48792714e-01
-1.95849568e-01 -8.73323381e-01 3.62710118e-01 3.30362618e-01
-4.05292213e-02 6.41633928e-01 5.08775301e-02 -1.39643431e-01
-4.29356396e-01 -1.20973563e+00 -7.63426065e-01 -1.08029306e+00
-4.40496236e-01 4.04824048e-01 2.15009451e-01 -1.72322601e-01
3.46304253e-02 4.37324315e-01 -1.15104187e+00 -3.86861444e-01
8.81911635e-01 4.32391793e-01 1.81403123e-02 5.05918384e-01
8.09643418e-02 3.32854003e-01 -1.38743967e-01 -6.77429497e-01
3.34044933e-01 -5.81074834e-01 -8.76159251e-01 3.33572596e-01
4.63670105e-01 3.97385567e-01 -7.02500641e-01 1.48482478e+00
1.48811385e-01 1.33889660e-01 5.14228523e-01 -1.09055042e+00
-7.31387019e-01 9.70610976e-01 -9.91260350e-01 1.04616201e-02
1.56824157e-01 1.36995250e-02 1.30943739e+00 -3.10715467e-01
8.37255061e-01 9.93053317e-01 8.09851587e-01 8.23192224e-02
-1.92057407e+00 -3.54262471e-01 1.65722802e-01 -3.13816816e-01
-1.80623817e+00 -4.02156785e-02 1.00545096e+00 -3.76413405e-01
4.38385785e-01 4.58405703e-01 7.19828308e-01 6.41004205e-01
3.39160383e-01 3.42811018e-01 1.08650970e+00 -5.12111962e-01
2.89124519e-01 -1.63945377e-01 2.71953493e-01 9.73411739e-01
9.34074879e-01 -1.18884873e-02 -3.53976995e-01 -2.45365649e-01
6.98519349e-01 -1.00248396e-01 -4.28163916e-01 -8.56537044e-01
-1.12016881e+00 6.94978595e-01 3.71389955e-01 4.71858978e-01
-2.90562809e-01 3.01754534e-01 7.78292790e-02 2.15974048e-01
1.39156759e-01 3.21215481e-01 1.14085421e-01 2.44236290e-01
-8.71667445e-01 1.06734432e-01 9.82948422e-01 1.05694771e+00
1.24146187e+00 -1.85607951e-02 1.93612695e-01 3.49100381e-01
2.41979025e-02 6.47004962e-01 -2.41887063e-01 -8.26304615e-01
2.03673467e-01 7.06987917e-01 -2.70028979e-01 -1.39151180e+00
-4.59775984e-01 -6.26445353e-01 -1.47547054e+00 -5.65544106e-02
4.83367860e-01 1.07243560e-01 -7.52390563e-01 2.30464768e+00
1.57614544e-01 3.10183495e-01 -3.54460657e-01 5.26582778e-01
3.51308584e-01 2.64662623e-01 -7.91207179e-02 -4.58054334e-01
1.12178385e+00 -3.73878181e-01 -5.22910953e-01 -1.56572878e-01
4.95308936e-01 -5.09880185e-01 9.58046734e-01 2.07856640e-01
-1.26688015e+00 -1.91793889e-01 -9.96857762e-01 1.46862939e-01
-6.36053309e-02 -4.57041472e-01 5.18088341e-01 8.16871643e-01
-1.35444605e+00 8.70413363e-01 -8.14878225e-01 -4.62056667e-01
1.22103557e-01 3.74227077e-01 -4.87175196e-01 -1.22083917e-01
-1.01203668e+00 5.74622273e-01 1.76414326e-01 -1.97928667e-01
-5.04978776e-01 -5.66170037e-01 -8.15284491e-01 1.40935525e-01
6.53948128e-01 -8.29652727e-01 4.96610790e-01 -5.54298520e-01
-1.10716391e+00 8.89347970e-01 -1.12665385e-01 -4.43154514e-01
2.27440253e-01 3.44194114e-01 -2.86967486e-01 1.69488564e-01
1.31746223e-02 2.21473321e-01 8.99018884e-01 -1.34936500e+00
-5.63422106e-02 -6.69229925e-01 7.84070119e-02 8.71692672e-02
-2.60756791e-01 -5.08083165e-01 -5.36457658e-01 -8.26111019e-01
4.93046284e-01 -1.35331488e+00 -3.60635191e-01 -3.61700803e-01
-6.90710902e-01 2.90256679e-01 5.98166108e-01 -4.29910719e-01
1.65263641e+00 -2.14082170e+00 6.24777257e-01 8.88171852e-01
9.04973567e-01 -1.84766963e-01 -1.45512804e-01 7.60023832e-01
-2.00790495e-01 4.23011780e-01 -7.07600236e-01 -2.21215226e-02
-5.58473915e-03 2.71576256e-01 -8.50533471e-02 9.48471725e-01
-2.59346932e-01 9.12474215e-01 -1.00283504e+00 -2.80231953e-01
5.85852042e-02 4.52894151e-01 -6.84525490e-01 -4.32281733e-01
5.94508573e-02 2.56021023e-01 -3.70932013e-01 3.36427003e-01
7.98305631e-01 -4.20754552e-01 8.19853783e-01 -3.98772269e-01
-5.43019585e-02 -1.50178194e-01 -1.57020485e+00 1.48904824e+00
-7.16749057e-02 2.19427004e-01 4.96353745e-01 -1.16453528e+00
7.93670237e-01 -6.76642656e-02 6.54964387e-01 -3.17927688e-01
5.77344149e-02 5.12827598e-02 2.27684453e-02 8.64089876e-02
3.81558478e-01 -2.78340667e-01 -1.91611007e-01 8.22004676e-01
-1.71438321e-01 -3.08278978e-01 3.78023416e-01 7.14673698e-01
1.58221769e+00 -2.93944478e-01 4.65387404e-01 -7.94690430e-01
3.29183340e-01 -3.84792209e-01 3.22969735e-01 8.76598775e-01
-3.64140235e-02 4.85988379e-01 6.61398470e-01 -1.29951581e-01
-1.04584682e+00 -1.48951471e+00 1.20491823e-02 6.45534515e-01
6.57675743e-01 -6.72266722e-01 -1.10168207e+00 -5.09698153e-01
1.54760003e-01 4.98189777e-01 -7.23361492e-01 -6.60119414e-01
-5.83616734e-01 -9.68032837e-01 3.69904727e-01 9.42340866e-02
2.42506474e-01 -6.20283544e-01 -2.06467703e-01 4.16372828e-02
-3.67335752e-02 -8.60891759e-01 -8.41888726e-01 1.67181697e-02
-8.84421885e-01 -1.12375307e+00 -2.38717586e-01 -5.86248875e-01
1.00309002e+00 5.35696983e-01 1.22188485e+00 3.53004247e-01
-3.04273158e-01 3.75658393e-01 -9.77156982e-02 2.59088039e-01
-2.75059015e-01 1.74311653e-01 1.85298905e-01 1.81443140e-01
-1.95369184e-01 -1.12945306e+00 -4.31349039e-01 2.48606101e-01
-1.28174293e+00 5.77975018e-03 5.60082734e-01 5.33410907e-01
8.16947818e-01 5.90252876e-01 6.19025603e-02 -1.07092464e+00
6.28517807e-01 -3.85089189e-01 -5.65198958e-01 2.90819228e-01
-6.55897677e-01 5.25138199e-01 5.86082697e-01 -1.24395005e-01
-7.42012620e-01 -5.58220781e-02 2.84404427e-01 -1.30830392e-01
3.20801109e-01 5.70165217e-01 -3.00951928e-01 -4.23827142e-01
4.85081196e-01 2.91973293e-01 1.05400700e-02 -4.44732189e-01
8.06343317e-01 2.48065982e-02 5.91751695e-01 -6.22643173e-01
1.24690270e+00 9.41730976e-01 5.42889535e-01 -9.11771834e-01
-6.46687210e-01 -9.08827633e-02 -7.16979861e-01 -1.37323025e-03
5.23314953e-01 -3.67960215e-01 -8.83384526e-01 2.58105814e-01
-6.38229609e-01 -7.21590519e-02 -6.18803680e-01 3.03211451e-01
-5.46576500e-01 6.99098289e-01 -4.96381521e-01 -5.23579597e-01
-6.70537427e-02 -9.37355161e-01 7.74519920e-01 -1.20247878e-01
-2.77450949e-01 -1.21175170e+00 2.67515451e-01 -7.58454129e-02
3.75632226e-01 4.63677049e-01 1.17577684e+00 -2.91970223e-01
-6.11404836e-01 -2.04325438e-01 -2.68262237e-01 8.07523727e-02
1.69092819e-01 1.75412409e-02 -3.05978954e-01 -5.82118273e-01
2.94092949e-02 2.91208982e-01 8.50261986e-01 4.39377278e-01
8.94674420e-01 -4.49702471e-01 -4.16281760e-01 6.33491039e-01
1.48438227e+00 -2.87058353e-01 6.17973149e-01 -3.09295177e-01
9.44547713e-01 3.73973370e-01 -2.06614941e-01 3.22227150e-01
2.54428953e-01 5.74177742e-01 3.12613010e-01 9.98630561e-03
-2.84730554e-01 -4.52169180e-01 2.64562339e-01 1.18002772e+00
-3.83556515e-01 -2.75353879e-01 -7.82485068e-01 3.06885302e-01
-1.55790389e+00 -9.68342423e-01 -1.75392002e-01 2.52370048e+00
6.56039417e-01 2.13987842e-01 7.63115510e-02 8.92100483e-02
1.17218041e+00 5.53174913e-01 -4.58688527e-01 -8.82830843e-02
-4.43783700e-01 5.55203140e-01 8.75851750e-01 8.49692225e-01
-8.05304527e-01 8.04237247e-01 6.22063732e+00 8.89471710e-01
-9.27930892e-01 -1.22077232e-02 4.51646417e-01 8.76853317e-02
-7.85048664e-01 4.76925790e-01 -3.64492089e-01 3.14452320e-01
6.33324087e-01 -5.07781029e-01 8.42433929e-01 5.39508522e-01
-6.97418451e-02 -9.36370045e-02 -8.87263000e-01 8.38292718e-01
6.27785698e-02 -1.40477967e+00 2.31372640e-01 3.30451399e-01
9.29338098e-01 -2.06554711e-01 -5.49846962e-02 -1.55604765e-01
6.88719511e-01 -8.80814254e-01 5.47906637e-01 4.25152808e-01
9.11736608e-01 -7.46667743e-01 2.53378570e-01 7.79699758e-02
-1.63378656e+00 2.74180412e-01 -4.64017987e-02 -1.47829270e-02
3.16271424e-01 1.12292242e+00 -1.85822800e-01 8.40789616e-01
4.51650172e-01 5.53202987e-01 -3.80162984e-01 6.17446125e-01
-1.34157896e-01 4.53992903e-01 -4.35140759e-01 4.27244425e-01
-4.72776517e-02 -9.74434555e-01 1.02146506e+00 8.01316977e-01
2.89499611e-01 2.68054038e-01 2.91954964e-01 9.91049349e-01
-4.33150709e-01 -5.98913208e-02 -7.83231020e-01 -2.06445932e-01
5.14060676e-01 1.21925020e+00 -1.14440703e+00 -1.72554165e-01
-2.96323240e-01 8.84476304e-01 4.77079779e-01 3.95664304e-01
-7.82049835e-01 -4.24575180e-01 5.58552384e-01 4.77985293e-01
2.35582307e-01 -5.02283692e-01 -3.98358911e-01 -1.21405208e+00
1.14378586e-01 -6.37191832e-01 4.55410302e-01 -2.29380116e-01
-1.13927329e+00 4.36775208e-01 -4.48478274e-02 -7.58506656e-01
2.40610957e-01 -9.43768099e-02 -5.02714694e-01 5.37784517e-01
-9.55738246e-01 -8.94244373e-01 -2.35483617e-01 5.37546098e-01
-3.70036870e-01 4.75084543e-01 5.32150567e-01 2.52938896e-01
-5.41087985e-01 5.34684420e-01 3.76748204e-01 -1.41048670e-01
3.49994183e-01 -1.23118472e+00 3.66101444e-01 1.00437462e+00
3.05479884e-01 8.31013918e-01 8.53086352e-01 -9.30164814e-01
-1.95768547e+00 -1.08746707e+00 7.22859502e-01 -1.74183249e-01
1.05464363e+00 -4.46962416e-01 -9.25534189e-01 8.09584379e-01
2.75609009e-02 6.30197674e-02 2.71321326e-01 5.33512309e-02
-4.27704692e-01 -1.62015140e-01 -1.22111964e+00 8.88041556e-01
1.58479524e+00 -5.53400576e-01 -1.63137317e-01 2.34116718e-01
6.44743800e-01 -5.44969477e-02 -9.54315364e-01 4.43614066e-01
3.51379871e-01 -9.93289948e-01 1.01029265e+00 -3.43621194e-01
-1.83432132e-01 -3.88045639e-01 -1.32309362e-01 -1.29685390e+00
-6.00756109e-01 -1.29467821e+00 3.56191695e-02 1.01281500e+00
1.91664577e-01 -6.77285016e-01 7.79185057e-01 3.24373722e-01
5.48910834e-02 -5.31214058e-01 -9.78798032e-01 -9.93669569e-01
-5.39869517e-02 -2.09680557e-01 4.57161427e-01 9.75234866e-01
2.19309494e-01 3.18739504e-01 -2.87441343e-01 1.76517457e-01
9.49164927e-01 2.96208024e-01 5.69943309e-01 -1.20917976e+00
-3.28068167e-01 -6.05547786e-01 -5.23552477e-01 -8.65931273e-01
7.51437768e-02 -1.17557919e+00 -2.25512579e-01 -1.20129085e+00
4.99081939e-01 -4.09572303e-01 -3.68528925e-02 1.35008022e-01
-3.05129252e-02 4.07773554e-01 9.10413936e-02 1.74491853e-01
-5.65548599e-01 4.52369541e-01 1.28328156e+00 -3.87160182e-02
-3.11004668e-01 -2.67358452e-01 -9.11267281e-01 7.98358321e-01
5.23771524e-01 -4.04635042e-01 -3.21689934e-01 -1.86032936e-01
5.59795499e-01 7.05585480e-02 3.39789510e-01 -9.39417899e-01
1.10705681e-01 -1.35744289e-01 -4.24545199e-01 -3.35293240e-03
3.95126035e-03 -5.59979200e-01 7.46333420e-01 5.61078787e-01
-2.07129031e-01 6.22647777e-02 -1.19193844e-01 7.95887947e-01
-1.14098936e-02 4.18159738e-03 1.16435003e+00 -5.98210432e-02
-2.36822277e-01 5.79735219e-01 -1.46168977e-01 3.90952915e-01
1.11410165e+00 -2.40680382e-01 -3.39862674e-01 -5.62502027e-01
-9.30311918e-01 -1.53760463e-01 9.13936317e-01 -6.51561767e-02
2.56921917e-01 -1.36761451e+00 -6.03153050e-01 1.19221956e-01
-8.44948143e-02 -9.98943895e-02 4.35926884e-01 1.13790381e+00
-4.37134087e-01 1.61054969e-01 1.66512072e-01 -3.43425691e-01
-1.02453709e+00 7.06163824e-01 1.14570618e-01 -4.71442968e-01
-6.78273797e-01 6.70059621e-01 6.39772654e-01 -1.11742072e-01
-1.91431150e-01 -2.52923816e-01 3.87697637e-01 -1.52136192e-01
1.71585336e-01 4.77060765e-01 1.49531821e-02 -7.34511256e-01
-2.56218582e-01 7.64332891e-01 4.25994322e-02 1.17058069e-01
1.28246355e+00 -4.71104562e-01 -6.86077476e-01 5.29897809e-02
1.03020477e+00 6.75611734e-01 -9.93022442e-01 -3.66691709e-01
4.02729884e-02 -4.07035470e-01 -1.34123698e-01 -1.08858511e-01
-1.28565609e+00 4.79331762e-01 -8.22445657e-03 8.95816922e-01
9.82461393e-01 3.97799283e-01 6.91059589e-01 1.03802934e-01
4.38434511e-01 -9.26789403e-01 -9.11312476e-02 3.98384035e-01
7.39360273e-01 -6.40068531e-01 1.43031135e-01 -8.94604146e-01
-2.40845412e-01 6.66868746e-01 1.48380965e-01 -3.69974345e-01
8.25085163e-01 3.43844444e-01 -6.57166779e-01 -4.92198169e-01
-2.74231017e-01 -1.39339417e-01 9.66744795e-02 4.43167478e-01
1.58009425e-01 4.10316229e-01 -3.61347377e-01 4.20702368e-01
-2.84281850e-01 -3.99143159e-01 6.37968183e-01 7.47973800e-01
-4.89787161e-01 -1.19191313e+00 -3.18743080e-01 5.28381288e-01
-8.80006328e-02 -2.81850219e-01 -5.52666724e-01 9.44077730e-01
7.19792396e-03 5.77137768e-01 -1.14658527e-01 -5.10062933e-01
1.20617501e-01 -2.57374942e-01 6.30877018e-01 -6.48388028e-01
-4.36014757e-02 -4.97165509e-02 -1.13146491e-01 -6.07790828e-01
-2.92733371e-01 -7.19237924e-01 -1.27009857e+00 -8.57766390e-01
-2.70257860e-01 2.67225116e-01 1.30021721e-01 8.76347244e-01
4.93441373e-01 3.64921123e-01 4.89462644e-01 -6.07523620e-01
-3.94776583e-01 -4.03839320e-01 -1.24827933e+00 6.68107629e-01
-1.28052915e-02 -6.86601520e-01 -6.51703596e-01 1.62855182e-02] | [7.083838939666748, 5.128429412841797] |
4ccbe705-61ec-4b7d-830a-54f0c3dc7aed | noise-estimation-using-density-estimation-for | 2003.03186 | null | https://arxiv.org/abs/2003.03186v3 | https://arxiv.org/pdf/2003.03186v3.pdf | Noise Estimation Using Density Estimation for Self-Supervised Multimodal Learning | One of the key factors of enabling machine learning models to comprehend and solve real-world tasks is to leverage multimodal data. Unfortunately, annotation of multimodal data is challenging and expensive. Recently, self-supervised multimodal methods that combine vision and language were proposed to learn multimodal representations without annotation. However, these methods often choose to ignore the presence of high levels of noise and thus yield sub-optimal results. In this work, we show that the problem of noise estimation for multimodal data can be reduced to a multimodal density estimation task. Using multimodal density estimation, we propose a noise estimation building block for multimodal representation learning that is based strictly on the inherent correlation between different modalities. We demonstrate how our noise estimation can be broadly integrated and achieves comparable results to state-of-the-art performance on five different benchmark datasets for two challenging multimodal tasks: Video Question Answering and Text-To-Video Retrieval. Furthermore, we provide a theoretical probabilistic error bound substantiating our empirical results and analyze failure cases. Code: https://github.com/elad-amrani/ssml. | ['Rami Ben-Ari', 'Daniel Rotman', 'Alex Bronstein', 'Elad Amrani'] | 2020-03-06 | null | null | null | null | ['noise-estimation'] | ['medical'] | [-4.40271720e-02 -1.70791730e-01 -1.24193318e-01 -4.29087311e-01
-1.56337190e+00 -7.02889323e-01 6.58587992e-01 1.97092474e-01
-4.62611765e-01 6.01859212e-01 1.79788738e-01 -1.41148522e-01
-6.16898350e-02 -2.57905513e-01 -7.59177864e-01 -7.01920271e-01
2.64745444e-01 5.11433601e-01 -1.05099224e-01 1.13008671e-01
-4.17825095e-02 4.26452942e-02 -1.64467847e+00 4.66663241e-01
7.21405566e-01 1.02867723e+00 1.26103520e-01 1.00889206e+00
-2.52396584e-01 9.63125348e-01 -3.42329830e-01 -7.45241582e-01
-2.83999979e-01 -3.34718674e-01 -8.74696195e-01 3.67959328e-02
5.67217588e-01 -2.72819936e-01 -4.56694603e-01 1.14584637e+00
5.77453494e-01 8.61837640e-02 9.71674144e-01 -1.45667660e+00
-4.64243889e-01 4.01538700e-01 -5.78719735e-01 -8.87144580e-02
6.01439297e-01 -7.95229748e-02 9.79625463e-01 -1.04222798e+00
3.66179407e-01 1.44868040e+00 5.23540735e-01 7.01944768e-01
-1.33733332e+00 -2.45076731e-01 5.64769804e-02 3.60876828e-01
-1.48070300e+00 -7.24562407e-01 5.32483816e-01 -4.34037566e-01
5.34295201e-01 3.17937404e-01 2.13543251e-02 1.46700728e+00
-2.63726175e-01 1.30748796e+00 8.40362549e-01 -4.99848932e-01
2.51351535e-01 1.60391033e-01 1.39828607e-01 7.64005244e-01
4.03033420e-02 -4.55477268e-01 -7.23978579e-01 -3.92707407e-01
2.61429101e-01 -1.50287613e-01 -4.18714911e-01 -3.98929983e-01
-1.07147586e+00 8.34872365e-01 -7.87521675e-02 1.15763813e-01
-1.98567808e-01 4.68772322e-01 4.31771427e-01 1.06918722e-01
3.36196035e-01 -4.93148603e-02 -3.08198065e-01 -6.02640688e-01
-8.18412006e-01 -4.46378700e-02 1.02577233e+00 8.38976085e-01
6.47691011e-01 -2.33652830e-01 -5.30322678e-02 9.91180122e-01
5.64000547e-01 7.96965957e-01 1.39558449e-01 -1.34541285e+00
4.68030304e-01 3.32082748e-01 9.18791294e-02 -8.57462764e-01
-2.74761230e-01 1.19302534e-01 -7.95534730e-01 -1.71496153e-01
7.49128640e-01 -3.14358592e-01 -6.86929226e-01 1.88354456e+00
1.68482229e-01 -1.77918319e-02 1.25028729e-01 9.50626075e-01
9.20227170e-01 5.52002013e-01 2.39460513e-01 -4.34987340e-03
1.26200914e+00 -7.39077449e-01 -8.16966772e-01 -6.98622093e-02
6.20484531e-01 -7.72636592e-01 1.09090579e+00 4.84526902e-01
-1.01180351e+00 -2.55551666e-01 -6.39726579e-01 -1.49458051e-01
-1.91464931e-01 3.72510672e-01 5.02191961e-01 6.97489977e-01
-9.39387143e-01 2.39426702e-01 -9.48441684e-01 -5.76732993e-01
3.39852303e-01 1.69096246e-01 -5.77581108e-01 -4.06920224e-01
-8.22663307e-01 7.48114765e-01 1.78630084e-01 1.67309090e-01
-1.02635610e+00 -4.43790168e-01 -1.00433850e+00 -2.06857491e-02
6.54750884e-01 -7.93438375e-01 1.45822835e+00 -9.21275675e-01
-1.26482534e+00 7.02517748e-01 -5.85561693e-01 -7.69387707e-02
4.41120535e-01 -5.27269304e-01 -5.46952449e-02 5.16005993e-01
-1.20538376e-01 7.95913637e-01 9.09248888e-01 -1.66394758e+00
-3.62284601e-01 -2.66309768e-01 6.64758012e-02 3.15253556e-01
-4.38679546e-01 -1.23561382e-01 -8.06390107e-01 -1.86320826e-01
-1.39820382e-01 -8.09945405e-01 1.04031958e-01 -2.83780005e-02
-3.42471093e-01 -2.93645650e-01 7.24419713e-01 -6.67756617e-01
1.07222986e+00 -2.17593336e+00 5.12203157e-01 1.01558812e-01
3.65454167e-01 -1.52631151e-03 -3.88790935e-01 5.51302135e-01
3.32822144e-01 1.61147371e-01 -2.19406888e-01 -9.20249999e-01
3.85170996e-01 3.48200738e-01 -1.94588169e-01 4.22836930e-01
3.05873543e-01 9.97906625e-01 -7.78757930e-01 -6.31799459e-01
2.19808966e-01 7.71373510e-01 -3.68616849e-01 3.38692516e-01
-3.03595573e-01 3.75191242e-01 -1.81927294e-01 9.86154616e-01
5.48578024e-01 -4.43115860e-01 3.16405535e-01 -5.58436453e-01
3.69103819e-01 -1.51858479e-01 -1.17075384e+00 1.98263288e+00
-2.91562110e-01 7.90734589e-01 4.67204064e-01 -9.72276032e-01
3.72094214e-01 3.43333870e-01 3.12072009e-01 -3.77995163e-01
2.98413247e-01 8.36342797e-02 -3.95440161e-01 -8.73873770e-01
5.22323251e-01 -1.81370303e-02 -1.43523261e-01 3.24597538e-01
5.14854610e-01 -1.44286200e-01 3.20060492e-01 5.89273572e-01
9.97192919e-01 1.47425994e-01 -1.03022661e-02 9.46724266e-02
4.04682070e-01 -3.38868290e-01 7.75446966e-02 1.00520039e+00
-2.11669117e-01 8.48411024e-01 8.43952000e-01 1.44965366e-01
-8.44917536e-01 -1.16822672e+00 6.74586371e-02 1.33404613e+00
-2.21244376e-02 -6.08877599e-01 -8.50544572e-01 -7.63704181e-01
-1.34757876e-01 5.08993089e-01 -4.90072042e-01 1.41921910e-02
-2.30116174e-02 -8.27655971e-01 7.97271013e-01 4.69531298e-01
5.56258038e-02 -3.95522237e-01 -2.34909013e-01 -3.11601073e-01
-5.99762857e-01 -1.47740102e+00 -2.73628086e-01 4.15930264e-02
-5.71920931e-01 -1.06990933e+00 -8.79438877e-01 -4.77078676e-01
5.97623408e-01 2.36958608e-01 1.20231354e+00 6.52066385e-03
-2.03192368e-01 1.49102461e+00 -3.53998870e-01 -1.42370760e-01
-4.04997826e-01 1.00247242e-01 2.69612460e-03 1.09328501e-01
4.91784841e-01 -3.43880415e-01 -4.69775230e-01 2.61371315e-01
-1.16400981e+00 -1.87197655e-01 5.47896266e-01 9.21710193e-01
4.70247984e-01 -3.35271567e-01 4.87034321e-01 -3.93220007e-01
6.17309272e-01 -6.62954211e-01 -3.95994395e-01 5.27761042e-01
-6.92809969e-02 2.03991473e-01 1.25536159e-01 -6.74536884e-01
-8.84244680e-01 2.05066606e-01 -5.28482907e-02 -5.75101972e-01
-4.29245561e-01 7.06012249e-01 -3.94174486e-01 1.45447245e-02
4.03281599e-01 1.44694418e-01 1.90342143e-01 -4.66483861e-01
7.09250212e-01 6.46117985e-01 4.26813394e-01 -9.96154904e-01
4.89759028e-01 5.17444432e-01 -7.21856356e-02 -9.64779079e-01
-8.36132765e-01 -5.21761775e-01 -4.14923251e-01 -4.64575797e-01
8.06693435e-01 -1.07167625e+00 -9.86141682e-01 2.93612033e-01
-1.17171645e+00 -2.36794561e-01 -6.22460432e-03 4.99761909e-01
-5.62091589e-01 8.44742596e-01 -5.98784387e-01 -1.26829374e+00
-4.42524403e-02 -1.11305249e+00 1.36645222e+00 2.85718828e-01
-1.37827918e-01 -1.07769883e+00 6.52140379e-02 6.62603498e-01
4.04041767e-01 -1.48235753e-01 8.44964862e-01 -6.77834272e-01
-5.18227398e-01 -1.27959222e-01 -4.79968041e-01 4.77632940e-01
-2.22793564e-01 -2.97027342e-02 -1.30822647e+00 -1.47579998e-01
-2.11400211e-01 -9.02521193e-01 1.08650601e+00 2.34438762e-01
1.07055569e+00 -6.65276349e-02 -8.55402201e-02 2.54774183e-01
1.25480855e+00 -4.33090895e-01 6.21838331e-01 -5.28375283e-02
8.26528192e-01 6.61536336e-01 2.41089001e-01 4.70413268e-01
6.99241698e-01 3.95014703e-01 5.11923075e-01 1.33475393e-01
2.35962104e-02 -1.44017518e-01 3.56816232e-01 8.32446754e-01
1.02991357e-01 -5.41327059e-01 -1.08820796e+00 5.66358685e-01
-2.17044139e+00 -8.82935584e-01 -9.74448398e-02 2.10164857e+00
7.44287491e-01 -3.88862759e-01 1.31541133e-01 -1.80200502e-01
4.82382596e-01 -6.15936294e-02 -2.73758024e-01 1.38452128e-01
-2.99216777e-01 -2.30114013e-01 1.04782537e-01 5.56435108e-01
-1.32577753e+00 6.87445998e-01 6.07842922e+00 9.81204391e-01
-6.02688432e-01 2.52916604e-01 6.12179875e-01 -1.78907081e-01
-3.88806999e-01 -2.47701377e-01 -6.52980864e-01 2.93498099e-01
1.11931360e+00 2.49250114e-01 4.96205956e-01 5.06280422e-01
-1.23198047e-01 -6.20147645e-01 -1.32874560e+00 1.42625070e+00
5.02173185e-01 -9.25235331e-01 1.03157081e-01 -1.87466443e-01
6.30019546e-01 3.19392756e-02 2.32271060e-01 2.93481499e-01
2.21691560e-02 -1.22499681e+00 4.70713824e-01 9.43479836e-01
5.26567757e-01 -5.89667499e-01 8.97045791e-01 3.73711735e-01
-9.31333125e-01 -2.30694003e-02 -2.30415106e-01 2.67694414e-01
2.16962844e-01 5.53917110e-01 -4.20552313e-01 6.38876081e-01
6.49195194e-01 4.05484855e-01 -6.34834111e-01 1.02536809e+00
7.33761489e-02 6.10477030e-01 -4.57292259e-01 -1.48876105e-02
8.74877514e-05 -3.14634629e-02 4.70596880e-01 1.47486103e+00
3.85533005e-01 -1.75184503e-01 -1.23199798e-01 7.29521990e-01
-3.42360377e-01 9.12608579e-02 -6.11464918e-01 -3.66294831e-01
2.89900154e-01 1.27431655e+00 -3.32116097e-01 -1.90302894e-01
-6.88998282e-01 1.03349113e+00 5.50732732e-01 6.97994292e-01
-8.73608112e-01 -2.85575502e-02 5.04358649e-01 -2.14538425e-01
2.94754177e-01 -4.72296387e-01 -3.18603739e-02 -1.48172593e+00
8.94788280e-02 -9.82197046e-01 4.57165182e-01 -7.54521549e-01
-1.76979375e+00 3.08503449e-01 1.40265107e-01 -8.42901289e-01
-3.25702578e-01 -7.97685146e-01 -9.93183553e-02 6.25250638e-01
-1.37720406e+00 -1.22420335e+00 -4.15938854e-01 6.23378158e-01
3.13231051e-01 -1.98744893e-01 9.19922829e-01 6.04622006e-01
-4.98816103e-01 6.12346232e-01 2.82641143e-01 1.83154806e-01
8.61055255e-01 -1.19552016e+00 -4.05220032e-01 5.64652741e-01
3.07239383e-01 4.00533050e-01 4.92992222e-01 -3.57564062e-01
-1.87389064e+00 -6.85759604e-01 6.43397927e-01 -7.91868925e-01
8.94807756e-01 -4.91487682e-01 -9.26223338e-01 4.53912050e-01
3.96111548e-01 -1.14407174e-01 9.71395314e-01 1.70170829e-01
-5.92753708e-01 1.06993325e-01 -8.24407995e-01 4.88084406e-01
6.03852451e-01 -9.36269104e-01 -2.88183898e-01 2.15612829e-01
4.39314485e-01 -2.44903877e-01 -7.65899956e-01 3.96364927e-01
6.05073631e-01 -8.27996850e-01 8.21589470e-01 -7.29616702e-01
4.13566828e-01 -2.84586787e-01 -7.86685586e-01 -9.18733358e-01
3.53107423e-01 -5.24182260e-01 -5.06847978e-01 1.39457870e+00
5.01541674e-01 -1.47623003e-01 5.41363120e-01 9.62932765e-01
2.61128753e-01 -4.01050210e-01 -1.11162961e+00 -4.65783715e-01
3.16929854e-02 -7.51665890e-01 -5.37392460e-02 7.66255677e-01
1.85310766e-02 4.77689058e-01 -5.88097095e-01 2.31488645e-01
7.99405396e-01 -4.25615191e-01 8.16498756e-01 -1.02149034e+00
-2.68643975e-01 -3.80270630e-01 -4.63721424e-01 -1.14826059e+00
5.23270130e-01 -5.90214491e-01 1.50076866e-01 -1.54832137e+00
6.83404207e-01 1.34210885e-01 -9.52767581e-02 2.65819430e-01
-2.21699700e-01 3.50843549e-01 2.38390312e-01 9.67282578e-02
-1.32171440e+00 6.48036897e-01 7.36663282e-01 -3.05867583e-01
2.10938334e-01 -2.69169062e-01 -5.58770478e-01 7.58798182e-01
7.41707206e-01 -2.34982178e-01 -3.41602027e-01 -6.24272704e-01
4.99081850e-01 -3.08915358e-02 6.68641627e-01 -7.89626539e-01
3.84464860e-01 4.64410409e-02 1.95675686e-01 -5.16094506e-01
7.99245954e-01 -8.27917457e-01 -2.20395848e-01 -1.95278674e-01
-4.21570390e-01 -1.05447456e-01 3.73790264e-01 8.59851182e-01
-3.49611372e-01 -4.09988046e-01 3.55813831e-01 1.02131359e-01
-6.40768945e-01 -6.16765879e-02 -5.10575294e-01 2.32748210e-01
6.74559593e-01 5.11224091e-01 -5.79317093e-01 -9.43110049e-01
-8.48341405e-01 3.72608453e-01 2.24418715e-01 2.59030521e-01
7.55904853e-01 -1.29289448e+00 -6.96403980e-01 -3.19969773e-01
2.13475466e-01 -2.06889898e-01 4.34761763e-01 1.20449269e+00
-2.17706084e-01 2.91756749e-01 3.81682009e-01 -1.03784299e+00
-1.32038736e+00 3.12814176e-01 3.05985838e-01 8.88912939e-04
-3.06147467e-02 7.37449169e-01 3.39773782e-02 -4.36196715e-01
7.48030901e-01 -1.16475318e-02 2.66566798e-02 2.42260620e-01
6.21057928e-01 2.84625679e-01 -1.02890946e-01 -6.38594747e-01
-3.48864794e-01 3.66630644e-01 6.05554283e-02 -4.68724251e-01
9.70762312e-01 -5.50498307e-01 -1.50968745e-01 7.55177796e-01
1.34122241e+00 -1.79767579e-01 -1.14274883e+00 -2.82314599e-01
4.13529240e-02 -3.00065190e-01 -1.71367601e-02 -7.78000653e-01
-7.09364653e-01 1.13556135e+00 6.94927633e-01 2.16799572e-01
1.01734984e+00 4.74476546e-01 4.81106639e-01 8.35826516e-01
-1.69211775e-02 -1.06766975e+00 3.70156407e-01 5.91793776e-01
6.35427833e-01 -1.87505245e+00 -2.45735928e-01 -2.01550722e-01
-9.21768129e-01 1.12799919e+00 3.65378290e-01 3.85608971e-01
4.91360635e-01 2.99968779e-01 1.80270344e-01 -3.22682620e-03
-8.71471882e-01 -4.71278965e-01 6.65077448e-01 6.72506809e-01
5.51509202e-01 -1.41607091e-01 1.64425552e-01 7.39135802e-01
4.27637666e-01 -2.97917500e-02 3.19996834e-01 9.12027299e-01
-2.74588794e-01 -1.02854812e+00 -4.75388497e-01 2.26894289e-01
-4.77672994e-01 -1.92546189e-01 -3.92178744e-01 6.65449858e-01
-3.37074697e-01 1.22243559e+00 -5.94719164e-02 -3.09941024e-01
-1.42736854e-02 5.57053268e-01 7.13893414e-01 -1.82977945e-01
4.36953567e-02 3.44887674e-01 2.01964095e-01 -5.54829299e-01
-8.02690685e-01 -6.00413620e-01 -9.53967750e-01 -3.46435815e-01
-1.08131036e-01 2.15904310e-01 8.06913972e-01 1.03858697e+00
4.23686296e-01 2.13784695e-01 1.37842759e-01 -8.64731789e-01
-5.61006308e-01 -9.17721570e-01 -2.77503759e-01 3.88840616e-01
4.15673852e-01 -5.82581937e-01 -5.14112711e-01 2.82620251e-01] | [10.686805725097656, 1.5791171789169312] |
656c1993-6c19-4f1a-b98a-02d04012a8f1 | collfren-rich-bilingual-english-french | null | null | https://aclanthology.org/2020.mwe-1.1 | https://aclanthology.org/2020.mwe-1.1.pdf | CollFrEn: Rich Bilingual English–French Collocation Resource | Collocations in the sense of idiosyncratic lexical co-occurrences of two syntactically bound words traditionally pose a challenge to language learners and many Natural Language Processing (NLP) applications alike. Reliable ground truth (i.e., ideally manually compiled) resources are thus of high value. We present a manually compiled bilingual English–French collocation resource with 7,480 collocations in English and 6,733 in French. Each collocation is enriched with information that facilitates its downstream exploitation in NLP tasks such as machine translation, word sense disambiguation, natural language generation, relation classification, and so forth. Our proposed enrichment covers: the semantic category of the collocation (its lexical function), its vector space representation (for each individual word as well as their joint collocation embedding), a subcategorization pattern of both its elements, as well as their corresponding BabelNet id, and finally, indices of their occurrences in large scale reference corpora. | ['Leo Wanner', 'Joan Codina-Filbá', 'Luis Espinosa Anke', 'Beatriz Fisas'] | null | null | null | null | coling-mwe-2020-12 | ['relation-classification'] | ['natural-language-processing'] | [-3.04480225e-01 -3.98796760e-02 -4.60888326e-01 -5.74509725e-02
-8.14784467e-01 -1.12034667e+00 7.08927393e-01 1.03503573e+00
-5.77242672e-01 1.29978788e+00 5.29478967e-01 -5.15794098e-01
1.25017300e-01 -8.58062923e-01 -4.89079833e-01 -5.98528266e-01
7.45262429e-02 7.60428309e-01 -1.06083639e-01 -5.02052903e-01
8.25120881e-02 -1.61659066e-02 -1.31368268e+00 1.23168863e-01
1.35035634e+00 6.45632803e-01 4.15305883e-01 -2.36428291e-01
-6.05527103e-01 3.00417811e-01 -7.40956306e-01 -9.73987401e-01
-3.98663104e-01 -2.33707726e-01 -1.01867950e+00 -2.95850515e-01
2.06935868e-01 6.64475381e-01 3.47529203e-01 1.07164347e+00
2.18650132e-01 3.05541366e-01 9.60518003e-01 -7.83451319e-01
-9.06904101e-01 1.09643567e+00 -3.02150100e-01 3.38551462e-01
6.40474916e-01 -2.02970311e-01 1.71693933e+00 -1.03676236e+00
1.00921440e+00 8.85709465e-01 4.63218093e-01 2.84908324e-01
-1.50137091e+00 -5.68558156e-01 5.84485345e-02 1.15968011e-01
-1.53893995e+00 1.75218374e-01 3.18816245e-01 -7.94875681e-01
1.21401572e+00 -8.52337554e-02 6.71867490e-01 1.39345253e+00
1.88166127e-02 2.67293096e-01 1.10818923e+00 -9.45162892e-01
2.13404402e-01 5.31280041e-01 4.04445618e-01 3.42269987e-01
5.24493456e-01 -2.37101257e-01 -5.82396090e-01 -2.67158508e-01
3.63545507e-01 -6.39553666e-01 -2.95113117e-01 -1.21056259e-01
-1.43595994e+00 9.86722827e-01 3.34140927e-01 1.00407183e+00
-5.10919988e-01 -2.99696065e-02 5.07715702e-01 1.66020781e-01
3.65445703e-01 7.23774672e-01 -8.19362283e-01 1.17991474e-02
-3.87057513e-01 3.05689394e-01 8.50378931e-01 1.20964170e+00
9.60026383e-01 -2.71111459e-01 -8.71377289e-02 1.19187808e+00
3.60911526e-02 2.07193911e-01 8.91345501e-01 -7.60877356e-02
5.64968288e-01 6.73289239e-01 1.80512935e-01 -9.70887005e-01
-1.91437915e-01 -1.99466214e-01 -5.00257552e-01 -7.32055187e-01
4.44702834e-01 -2.78668076e-01 -8.00907537e-02 2.13081956e+00
4.60593492e-01 -5.03642410e-02 3.90001237e-01 6.12668991e-01
9.82918859e-01 7.48973131e-01 8.84935439e-01 -2.92862922e-01
1.94924021e+00 -3.58403862e-01 -5.76500833e-01 -3.08391064e-01
6.85609162e-01 -9.75348413e-01 1.27688599e+00 6.08245656e-02
-6.67425931e-01 -3.18540066e-01 -7.75651932e-01 3.18199024e-02
-9.21325386e-01 3.59731555e-01 8.64952445e-01 5.41789591e-01
-5.23706675e-01 2.79751718e-01 -1.45993531e-01 -4.44878548e-01
-1.86755210e-01 2.18742803e-01 -7.47569680e-01 6.66185617e-02
-1.83089340e+00 1.27902699e+00 9.14755702e-01 -3.34198266e-01
7.64933899e-02 -7.92670906e-01 -1.32040882e+00 -1.57272480e-02
2.08420083e-01 -4.20533836e-01 8.21707785e-01 -9.37792838e-01
-9.28619385e-01 1.32620800e+00 -7.87796974e-02 -3.19574833e-01
-2.77786463e-01 -8.52954462e-02 -8.08297753e-01 -3.15977573e-01
7.88055897e-01 4.58008528e-01 2.23556712e-01 -9.24213290e-01
-9.18295979e-01 -1.49188250e-01 -3.16303223e-02 3.35456133e-01
-3.48264754e-01 1.16407663e-01 8.25015157e-02 -9.25510228e-01
2.88967014e-04 -8.95817280e-01 7.28369206e-02 -7.79217601e-01
-3.60345840e-01 -1.02833927e+00 -1.63410619e-01 -6.28493071e-01
1.37352252e+00 -2.01625538e+00 2.69054025e-01 1.88874096e-01
-1.30717352e-01 2.14719456e-02 1.12617858e-01 7.36526072e-01
-4.17925656e-01 2.41339371e-01 -1.87142566e-02 1.69331983e-01
2.54587919e-01 1.40546754e-01 -3.65863711e-01 4.36964124e-01
5.38609922e-01 8.64861369e-01 -1.20754862e+00 -3.26496542e-01
1.15871683e-01 2.00617492e-01 -3.29423457e-01 -8.34712014e-02
-2.97136605e-01 2.64436066e-01 -3.06569129e-01 3.69565457e-01
9.27498937e-02 1.20741643e-01 7.31015384e-01 -3.32847446e-01
-2.38854110e-01 9.79004323e-01 -1.07186830e+00 1.36991727e+00
-1.10887885e+00 2.77231485e-01 -3.86871308e-01 -6.03415370e-01
9.00337160e-01 6.80113733e-01 1.89222962e-01 -1.81113288e-01
5.84855638e-02 6.65942252e-01 2.51521736e-01 -1.96016505e-01
6.99115813e-01 -4.34897691e-01 -7.16499150e-01 2.52575696e-01
5.63830495e-01 -1.27628982e-01 4.69867289e-01 2.09295694e-02
6.69654727e-01 5.22192046e-02 1.21209514e+00 -8.57606947e-01
6.73148036e-01 4.22859162e-01 4.40285981e-01 3.33021842e-02
1.94483429e-01 3.40837240e-02 3.45321447e-01 -2.50854015e-01
-7.03994274e-01 -1.23638177e+00 -6.96146667e-01 1.17160070e+00
9.03653353e-02 -4.84523118e-01 -5.16207993e-01 -4.74068195e-01
1.58259556e-01 1.12368298e+00 -5.09550273e-01 7.69150928e-02
-5.74232638e-01 -5.75892270e-01 3.43737245e-01 4.68098849e-01
-8.91447291e-02 -1.50842202e+00 4.97812033e-02 5.80471873e-01
-3.79281580e-01 -1.12174129e+00 -7.10515797e-01 4.55579937e-01
-2.88191438e-01 -1.05790496e+00 -2.21467897e-01 -1.25162804e+00
2.33295903e-01 -2.17243806e-01 1.67998922e+00 -4.14248221e-02
-1.22935876e-01 -2.74802223e-02 -5.63282847e-01 -2.51876324e-01
-4.77036387e-01 1.58559099e-01 4.97152805e-01 -1.75363690e-01
1.00195944e+00 -4.79531348e-01 -1.95730366e-02 -1.23994857e-01
-6.38707995e-01 -5.02164781e-01 2.24925652e-01 1.12439549e+00
6.35565281e-01 -2.49531046e-01 7.96648860e-01 -1.10872257e+00
7.30423987e-01 -9.93777871e-01 -5.67615747e-01 2.87918657e-01
-3.39699030e-01 -3.73837277e-02 5.96686602e-01 -5.34604013e-01
-1.05136251e+00 -6.87877685e-02 -2.65393257e-01 5.29213428e-01
-4.28648770e-01 8.32148969e-01 -4.80048001e-01 5.10467827e-01
9.49193537e-01 9.00812298e-02 -7.00005352e-01 -3.96697670e-01
7.50218630e-01 5.92977405e-01 5.77820718e-01 -1.15945613e+00
4.42002356e-01 -3.91046703e-01 -4.40664828e-01 -9.73550022e-01
-1.05265164e+00 -1.02534842e+00 -8.89888644e-01 3.75599474e-01
1.11570919e+00 -1.02878976e+00 -5.18223941e-01 -2.72762775e-01
-1.49764836e+00 2.70861894e-01 -1.82801455e-01 7.30107188e-01
-2.55539715e-01 -7.55248293e-02 -4.74897951e-01 -3.12165916e-01
-2.43716940e-01 -8.98225307e-01 9.49713767e-01 9.29629654e-02
-9.70817029e-01 -1.29559767e+00 2.60228634e-01 1.18769966e-01
-1.49905190e-01 4.74906787e-02 1.67185652e+00 -1.09728265e+00
-4.83543165e-02 4.35692072e-02 -9.15138647e-02 9.10217986e-02
2.68836439e-01 -2.22884223e-01 -4.58162367e-01 5.56834340e-02
-4.19985801e-01 -3.44289958e-01 4.98371512e-01 1.09103546e-01
2.91116267e-01 -4.93697792e-01 -1.94979519e-01 1.16192259e-01
1.64982975e+00 3.23453136e-02 1.69573531e-01 9.10844132e-02
5.78360975e-01 9.06824112e-01 6.31367207e-01 -1.20122701e-01
2.73702949e-01 7.74025738e-01 -2.30073214e-01 4.92743462e-01
2.78396923e-02 -3.67822051e-01 3.99867803e-01 1.16589785e+00
3.02811954e-02 -1.77731603e-01 -1.02276576e+00 1.15179157e+00
-1.31199133e+00 -7.49416053e-01 -3.98442566e-01 2.15294909e+00
1.61983085e+00 -1.56438082e-01 -8.38995054e-02 -9.64265019e-02
1.01869774e+00 -1.47093073e-01 1.75782084e-01 -4.54656452e-01
-2.99289972e-01 6.79224849e-01 6.41102269e-02 6.15415573e-01
-1.03454220e+00 1.51140690e+00 4.79136658e+00 1.09913182e+00
-7.67751575e-01 3.12713891e-01 1.11590251e-01 5.91671646e-01
-6.54066265e-01 -2.20484752e-02 -1.13900256e+00 4.24177468e-01
7.50023246e-01 -3.57224315e-01 2.12642491e-01 9.20266449e-01
-4.59381729e-01 -1.45935848e-01 -1.22778618e+00 9.25866663e-01
-6.98871389e-02 -1.49075258e+00 2.60700077e-01 -1.63695365e-01
9.12888467e-01 -1.40829295e-01 -3.29662234e-01 4.58962083e-01
4.37252998e-01 -8.07179987e-01 8.19779634e-01 -2.69977063e-01
1.23418868e+00 -8.51512909e-01 9.54496861e-01 5.48632219e-02
-1.29151917e+00 4.05637354e-01 -3.94151211e-01 1.53668985e-01
-4.91756313e-02 5.86336553e-01 -7.35374331e-01 5.57350039e-01
2.28709251e-01 3.96214962e-01 -2.22468898e-01 5.14756143e-01
-8.72806311e-01 9.71314490e-01 -1.34945020e-01 -5.78432560e-01
3.52680147e-01 -4.29410785e-01 6.14207089e-01 1.55314112e+00
3.48280638e-01 1.97940141e-01 2.82644123e-01 8.41567278e-01
-1.33280426e-01 1.04394388e+00 -5.18303990e-01 -2.03241929e-01
9.93658185e-01 1.23124075e+00 -7.28707016e-01 -3.42959851e-01
-4.89383668e-01 8.38002086e-01 7.21744359e-01 2.33448371e-01
-4.46375757e-01 -4.92170811e-01 1.20469546e+00 -3.26727889e-02
1.57121584e-01 -1.19391702e-01 -1.45270079e-01 -1.17008555e+00
1.79552662e-04 -6.29133761e-01 4.90240693e-01 -3.15551430e-01
-1.97280610e+00 9.63467777e-01 -1.72849908e-01 -9.64658260e-01
-4.74696606e-01 -7.54878402e-01 -5.17872810e-01 1.46101987e+00
-1.13530588e+00 -1.12964499e+00 3.87159586e-01 4.31086808e-01
4.49086040e-01 -3.83595526e-01 1.32094705e+00 3.30562085e-01
-4.03406024e-01 6.36557162e-01 -2.09622771e-01 2.59755611e-01
6.74919546e-01 -1.48673308e+00 1.92242473e-01 3.88153672e-01
7.84276843e-01 9.90020692e-01 6.90994918e-01 -8.14815938e-01
-7.09842801e-01 -1.11646414e+00 1.82451665e+00 -5.87360024e-01
1.33964074e+00 -4.67170745e-01 -9.07263458e-01 6.79012299e-01
2.49310285e-01 -2.13797748e-01 1.07047939e+00 6.43399596e-01
-3.83503288e-01 1.42959073e-01 -8.08430076e-01 7.58224249e-01
8.24830294e-01 -6.21846199e-01 -9.98576105e-01 6.36652112e-01
9.80203986e-01 -2.83108801e-01 -7.89507449e-01 -3.11769783e-01
8.13071355e-02 -4.69220340e-01 8.26790750e-01 -9.41663444e-01
4.79478389e-01 -2.27859810e-01 -3.15489799e-01 -1.72997665e+00
-3.08358341e-01 -5.24057031e-01 4.08871979e-01 1.47953391e+00
8.87810767e-01 -4.68786925e-01 2.55204201e-01 7.70922080e-02
-1.71087608e-01 -5.71803331e-01 -1.08742118e+00 -6.74284697e-01
4.20518875e-01 -4.19007719e-01 6.26911640e-01 1.32232761e+00
6.28236055e-01 9.97065663e-01 2.10475549e-01 1.34201022e-02
1.39016837e-01 4.75454271e-01 7.63906166e-02 -1.37675858e+00
-1.67609423e-01 -3.71392846e-01 -4.74960417e-01 -4.90924180e-01
6.28342330e-01 -1.43062675e+00 1.40250310e-01 -1.11071575e+00
-1.56686038e-01 -8.23745787e-01 9.71671566e-02 4.00169879e-01
-4.71275449e-01 1.92031786e-01 -2.49716058e-01 -9.40048173e-02
-6.83329627e-02 4.58044261e-01 6.51050389e-01 1.46953642e-01
-1.82702482e-01 -2.51986593e-01 -7.10727811e-01 7.24380255e-01
5.13998806e-01 -6.33251607e-01 1.04037844e-01 -2.14987010e-01
5.54596841e-01 -8.64008069e-02 9.16661099e-02 -4.36111212e-01
-8.21931064e-02 -4.77526426e-01 3.42751592e-02 -9.04271603e-02
7.32059544e-03 -7.19811141e-01 -1.67669244e-02 2.82672793e-01
-2.66498774e-01 2.25583568e-01 1.69789091e-01 2.26764023e-01
-5.90328813e-01 -6.18182421e-01 6.32844031e-01 -1.04865491e-01
-6.87012374e-01 4.13535871e-02 -5.25068521e-01 5.10545790e-01
1.16779983e+00 2.78053641e-01 -5.63975871e-02 1.21824451e-01
-7.81294584e-01 -1.68547317e-01 2.34072506e-01 5.51615477e-01
1.90870881e-01 -1.55489051e+00 -8.89734626e-01 1.45398453e-01
5.91113150e-01 -4.40890253e-01 -9.97712314e-02 3.36797833e-01
-3.79692912e-01 5.68112850e-01 -1.21723287e-01 5.41374125e-02
-7.95867443e-01 6.53037131e-01 -1.57667607e-01 -2.62934774e-01
-2.68285304e-01 1.05324519e+00 1.78393692e-01 -5.40915430e-01
-6.02980331e-03 -2.96681345e-01 -4.93766189e-01 5.55807769e-01
3.21128160e-01 -1.16216063e-01 2.89254189e-01 -1.16359198e+00
-5.50829291e-01 3.59678328e-01 2.02137560e-01 -1.14333794e-01
9.92910743e-01 -9.12441462e-02 -5.65896928e-01 6.99500024e-01
9.18193042e-01 8.26626003e-01 -1.01360567e-01 -3.99113774e-01
4.33595330e-01 -2.72019533e-05 -4.24345195e-01 -8.23893070e-01
-5.41089296e-01 6.13052905e-01 -2.81981260e-01 1.27880141e-01
4.66869771e-01 4.78713840e-01 8.37417126e-01 3.58037889e-01
5.77346742e-01 -8.91358376e-01 -5.08649528e-01 9.01970208e-01
9.49048281e-01 -1.04665053e+00 -3.83380860e-01 -9.23791289e-01
-7.67211616e-01 8.36275876e-01 5.03662527e-01 -1.41368762e-01
7.09794283e-01 -6.31869435e-02 1.03568159e-01 -1.44538954e-01
-6.12904072e-01 -3.94591421e-01 5.66120207e-01 7.25749910e-01
1.04844201e+00 4.30473685e-01 -1.21378064e+00 1.30250430e+00
-7.09595978e-01 -9.51728940e-01 2.78117657e-01 2.45823920e-01
-1.06850877e-01 -1.43432891e+00 -1.99630901e-01 3.73678207e-01
-4.85220492e-01 -8.85519922e-01 -3.90002906e-01 7.36325741e-01
6.24897122e-01 7.11795151e-01 1.56540260e-01 -2.40631010e-02
2.50521988e-01 2.27750868e-01 3.68539214e-01 -1.40484130e+00
-9.74251091e-01 -3.21966529e-01 4.97873008e-01 -1.26861572e-01
-2.18748197e-01 -9.11854029e-01 -1.34900284e+00 -5.07730953e-02
-5.06454170e-01 6.91891551e-01 4.47818726e-01 9.84864652e-01
-2.06715286e-01 1.41394094e-01 1.27446502e-01 -3.38028580e-01
-2.97190934e-01 -1.13895714e+00 -8.97466481e-01 6.74058020e-01
-2.12907538e-01 -7.67965078e-01 -2.31836393e-01 2.68866569e-01] | [10.41144847869873, 9.503273963928223] |
a8cb6d46-7e72-40c9-a74e-d7c45e87b7a9 | 3d-molecular-geometry-analysis-with-2d-graphs | 2305.13315 | null | https://arxiv.org/abs/2305.13315v1 | https://arxiv.org/pdf/2305.13315v1.pdf | 3D Molecular Geometry Analysis with 2D Graphs | Ground-state 3D geometries of molecules are essential for many molecular analysis tasks. Modern quantum mechanical methods can compute accurate 3D geometries but are computationally prohibitive. Currently, an efficient alternative to computing ground-state 3D molecular geometries from 2D graphs is lacking. Here, we propose a novel deep learning framework to predict 3D geometries from molecular graphs. To this end, we develop an equilibrium message passing neural network (EMPNN) to better capture ground-state geometries from molecular graphs. To provide a testbed for 3D molecular geometry analysis, we develop a benchmark that includes a large-scale molecular geometry dataset, data splits, and evaluation protocols. Experimental results show that EMPNN can efficiently predict more accurate ground-state 3D geometries than RDKit and other deep learning methods. Results also show that the proposed framework outperforms self-supervised learning methods on property prediction tasks. | ['Shuiwang Ji', 'Maho Nakata', 'Cheng Deng', 'Kaleb Dickerson', 'Meng Liu', 'Xinyi Xu', 'Xuan Zhang', 'Youzhi Luo', 'Yaochen Xie', 'Zhao Xu'] | 2023-05-01 | null | null | null | null | ['property-prediction'] | ['medical'] | [-1.00897104e-01 -4.42922205e-01 -4.43567961e-01 -5.32642305e-01
-8.37408960e-01 -4.77969766e-01 1.77746773e-01 8.13549936e-01
-1.53679579e-01 1.06663275e+00 -1.72023565e-01 -9.49471891e-01
6.02213889e-02 -1.16671133e+00 -1.00769877e+00 -7.24045098e-01
-6.12885416e-01 3.61742765e-01 -4.77147758e-01 -6.98364079e-02
4.62146938e-01 1.08428788e+00 -8.71332705e-01 2.23086178e-01
1.10137784e+00 8.14137220e-01 -2.70189166e-01 6.55409873e-01
1.12863682e-01 7.41523325e-01 -3.23378235e-01 -3.05011302e-01
2.42084205e-01 -4.30884928e-01 -1.07334065e+00 -4.80460167e-01
7.15417147e-01 -5.43321431e-01 -6.91057920e-01 7.81840086e-01
8.86218429e-01 3.86323839e-01 9.59152579e-01 -9.00026858e-01
-6.72653735e-01 4.28663522e-01 -1.50246575e-01 1.60116225e-03
7.21236289e-01 5.23429096e-01 9.96091783e-01 -6.93215251e-01
6.50250494e-01 9.57905710e-01 8.17930937e-01 2.79354066e-01
-1.58241963e+00 -6.45783305e-01 -2.61214465e-01 3.98607761e-01
-1.64880872e+00 -3.12023193e-01 9.09446895e-01 -1.54379487e-01
1.74240315e+00 -1.12047128e-01 5.92065036e-01 1.16284895e+00
7.62009859e-01 2.72591740e-01 8.73007715e-01 -8.79344493e-02
5.62698603e-01 -6.38465345e-01 1.79184958e-01 8.88610125e-01
3.50550920e-01 1.28669739e-01 -4.54463601e-01 -4.92628217e-01
4.72801089e-01 2.38485083e-01 -9.77841839e-02 -4.81781662e-01
-7.32583582e-01 8.98219168e-01 9.29096997e-01 -2.56139308e-01
-6.15799904e-01 3.08945090e-01 2.24471658e-01 4.61505383e-01
3.44338179e-01 6.89585388e-01 -5.58892965e-01 -9.88063887e-02
-5.79239905e-01 4.83005345e-01 1.00368345e+00 7.31595397e-01
9.97593820e-01 -1.00721784e-01 2.14123428e-01 1.60462797e-01
2.94249296e-01 1.58253953e-01 4.24577557e-02 -4.35326606e-01
3.76449972e-01 7.44991481e-01 -1.25705063e-01 -8.25187445e-01
-7.59794593e-01 -1.71493530e-01 -1.04978037e+00 -6.69072270e-02
2.03729853e-01 -1.33278549e-01 -8.66053164e-01 1.35811388e+00
6.17041588e-01 -2.45797361e-04 3.57599109e-01 6.76028550e-01
1.39520824e+00 1.02577865e+00 2.92342931e-01 -2.40491167e-01
5.79961538e-01 -5.59765279e-01 -2.11749777e-01 5.27120352e-01
1.32143068e+00 -5.39134264e-01 4.09486592e-01 3.60282511e-01
-9.51667786e-01 -5.60960233e-01 -1.22525477e+00 -4.41423327e-01
-7.14794934e-01 -7.43150571e-03 1.35952652e+00 7.77900159e-01
-6.08145773e-01 1.47672415e+00 -9.91922617e-01 6.87564388e-02
6.94094121e-01 8.04840684e-01 -7.04292715e-01 -1.57342881e-01
-1.11650395e+00 7.69870937e-01 7.32744277e-01 3.40975896e-02
-1.13939178e+00 -9.96340930e-01 -1.14932489e+00 -1.59257829e-01
1.45414963e-01 -6.16022468e-01 1.02524841e+00 -1.40617475e-01
-1.65903437e+00 6.07316077e-01 -2.00383052e-01 -5.54708242e-01
1.31337224e-02 1.11811906e-01 -2.35710725e-01 1.18555807e-01
-3.04864615e-01 6.40843809e-01 3.24368507e-01 -7.09123850e-01
3.15382659e-01 -5.11385798e-01 8.29727501e-02 2.71764807e-02
1.14274152e-01 -5.66159606e-01 3.19506288e-01 1.57575980e-02
3.29863280e-02 -7.38539159e-01 -6.32822812e-01 -3.38146001e-01
-9.01711404e-01 -4.02960330e-01 2.84386128e-01 -2.05588952e-01
8.22219074e-01 -1.51743555e+00 1.88948527e-01 3.94886285e-01
5.85731983e-01 3.23897243e-01 -2.22709179e-01 9.52842176e-01
-5.78533351e-01 1.89584732e-01 -6.39851019e-02 -1.02321297e-01
-1.38735384e-01 -5.08710369e-02 -1.71049774e-01 5.85094869e-01
3.21797997e-01 1.06713712e+00 -8.87884319e-01 -2.14521304e-01
3.82082850e-01 6.90332949e-01 -9.11835074e-01 2.43530571e-01
-4.70309645e-01 3.51462722e-01 -5.03894627e-01 7.84604847e-01
1.01463771e+00 -5.85629761e-01 4.39159662e-01 -4.24292743e-01
-1.59793422e-01 7.24939466e-01 -7.03372180e-01 2.04468131e+00
-2.88757794e-02 -5.53384051e-02 -7.64974415e-01 -1.02716184e+00
1.06556451e+00 3.99339795e-02 6.27526760e-01 -7.64346063e-01
1.42660841e-01 2.47894987e-01 2.30590090e-01 -2.92076111e-01
4.56781834e-01 -2.20746905e-01 1.35598958e-01 3.47448558e-01
3.60256702e-01 -4.52762961e-01 2.24781618e-01 2.87540704e-01
9.66577888e-01 3.77794713e-01 4.72905517e-01 4.50983495e-02
5.59022844e-01 1.24660924e-01 3.20660323e-01 6.34618759e-01
-9.25835148e-02 1.91350460e-01 5.89995325e-01 -1.07649541e+00
-1.04236317e+00 -9.29590821e-01 -4.23477829e-01 6.28665030e-01
1.23185113e-01 -9.20822144e-01 -5.37958384e-01 -5.43626428e-01
2.43847668e-01 3.45996976e-01 -3.66367251e-01 -5.35561323e-01
-2.14974016e-01 -1.19427156e+00 3.55491638e-01 2.38589421e-01
3.08107555e-01 -7.42081285e-01 2.00227037e-01 4.97153729e-01
2.72391438e-01 -8.18447471e-01 -2.92809308e-01 5.08767605e-01
-1.09438276e+00 -1.36291397e+00 -8.27887580e-02 -5.23985445e-01
4.30118293e-01 2.08117023e-01 1.10657358e+00 3.93108698e-03
-6.19724691e-01 -3.67007911e-01 -8.83499607e-02 -3.01052064e-01
-6.29122317e-01 5.57643831e-01 2.12626413e-01 -3.06367993e-01
7.16771543e-01 -8.18543375e-01 -7.35021770e-01 -1.81932271e-01
-7.05147505e-01 4.39328067e-02 5.62051833e-01 4.89923745e-01
9.45128262e-01 4.47592139e-03 5.42095900e-01 -7.20789731e-01
6.52166605e-01 -4.51303691e-01 -6.97852194e-01 1.20039940e-01
-4.36631024e-01 4.97389406e-01 1.16388178e+00 5.29681668e-02
-5.92991233e-01 3.05002898e-01 -5.40512264e-01 -1.11691616e-01
-4.57900524e-01 8.24146986e-01 -1.34328350e-01 -4.93744671e-01
8.78773868e-01 2.82754987e-01 -1.57319516e-01 -4.74993616e-01
3.66844356e-01 6.36527956e-01 1.76214501e-01 -9.58201230e-01
3.77264619e-01 1.50817037e-01 8.78515601e-01 -9.44031358e-01
-8.60315740e-01 -2.16225132e-01 -1.05875289e+00 3.75585645e-01
6.16489232e-01 -9.36069071e-01 -1.42054856e+00 3.96721005e-01
-1.28798628e+00 -4.60238189e-01 1.37847707e-01 6.05269849e-01
-6.30027592e-01 7.04855084e-01 -8.72580945e-01 -5.29545367e-01
-6.50769234e-01 -1.36602402e+00 1.14873016e+00 1.25687853e-01
-7.29196370e-02 -1.03939748e+00 4.46176648e-01 3.24209780e-01
-3.78624015e-02 7.92961538e-01 1.27463686e+00 -7.07969606e-01
-7.28530943e-01 -3.71153831e-01 -6.93687052e-02 -9.03106555e-02
2.47224033e-01 2.64827937e-01 -8.14444959e-01 -3.29638481e-01
-7.24237859e-01 -8.91956329e-01 9.37972307e-01 4.47524577e-01
1.38333750e+00 -1.82615533e-01 -3.82854283e-01 1.06788290e+00
1.24825394e+00 2.19530806e-01 5.92959583e-01 -2.78959066e-01
9.78437781e-01 1.56909227e-01 6.64180964e-02 5.04688263e-01
4.96587664e-01 4.00312334e-01 5.49969733e-01 -1.83376878e-01
2.32839718e-01 -6.45403266e-01 3.13673049e-01 5.81173956e-01
-3.37425828e-01 -3.71313751e-01 -7.87421823e-01 -2.74950117e-01
-1.46316409e+00 -1.01399684e+00 -4.08102810e-01 2.13344884e+00
1.21257365e+00 -7.67562538e-02 1.61792502e-01 1.10384278e-01
2.21865952e-01 1.30911767e-01 -1.05622566e+00 -5.73868394e-01
4.51337248e-02 6.34143293e-01 2.73336142e-01 4.72033232e-01
-1.15675235e+00 1.08211648e+00 6.71473837e+00 6.70455039e-01
-1.30897117e+00 -5.38234234e-01 8.93857777e-01 -3.62439007e-02
-1.27523616e-01 -4.43349443e-02 -8.71342897e-01 2.89583802e-01
1.16726816e+00 -7.03077167e-02 3.24340314e-01 8.27102482e-01
2.34538466e-01 2.75056183e-01 -1.61745977e+00 1.23874187e+00
-2.21843019e-01 -1.95310712e+00 4.24527079e-01 4.58748817e-01
8.11440051e-01 2.05019817e-01 -9.40758213e-02 1.66734681e-01
3.35629165e-01 -1.48315060e+00 4.35652062e-02 4.21826839e-01
9.25587893e-01 -1.13759851e+00 4.74151611e-01 7.40139410e-02
-1.15294337e+00 5.18965721e-01 -7.54608691e-01 -3.52516174e-01
-1.98938593e-01 5.56165755e-01 -8.52984488e-01 9.76423800e-01
5.15084416e-02 1.13081086e+00 -4.78813857e-01 1.01890135e+00
-2.29263585e-02 3.34505469e-01 -2.11720467e-01 -3.23000699e-01
3.53886902e-01 -5.77127814e-01 -1.31489664e-01 7.86991656e-01
1.01245090e-01 2.78736204e-01 2.59381950e-01 1.16572881e+00
-7.04861403e-01 2.75060177e-01 -9.45289791e-01 -4.72959608e-01
3.92337799e-01 1.15005016e+00 -4.13179964e-01 -1.41369760e-01
-2.70520180e-01 8.16632271e-01 5.54603219e-01 2.52108485e-01
-7.47652769e-01 -7.08814800e-01 9.36212361e-01 -1.21642284e-01
-1.19664542e-01 -7.72443235e-01 2.07007319e-01 -1.25302327e+00
-3.91264409e-01 -8.08519721e-01 2.29482472e-01 -6.75171793e-01
-1.47017395e+00 3.17299217e-01 -3.12402546e-01 -6.37666941e-01
1.87212974e-02 -1.18034637e+00 -7.11959779e-01 7.12363124e-01
-1.68121040e+00 -7.04698503e-01 -1.90513939e-01 5.83404005e-01
-2.68228441e-01 -3.99644636e-02 1.10751808e+00 1.20392174e-01
-1.06995058e+00 5.45263886e-01 2.32659772e-01 -1.12993652e-02
6.07996643e-01 -1.30971372e+00 7.59664714e-01 3.44977617e-01
2.48068616e-01 9.56346154e-01 3.21200937e-01 -6.85593247e-01
-2.26009965e+00 -1.28996921e+00 6.68749034e-01 -4.08573508e-01
4.23018634e-01 -2.78589934e-01 -9.09906566e-01 2.08423868e-01
-2.63370901e-01 2.42809147e-01 1.38576901e+00 1.82812780e-01
-4.16390032e-01 -1.03495687e-01 -1.08764374e+00 4.56303507e-01
1.34286284e+00 -6.84610188e-01 -1.15651637e-01 8.85914564e-01
7.00927198e-01 -5.60200214e-01 -1.39501452e+00 3.68443280e-01
2.73448348e-01 -8.57868254e-01 1.22212136e+00 -1.33783615e+00
5.18796146e-01 -9.86634567e-02 -8.10319732e-04 -1.11913490e+00
-6.91340864e-02 -1.03525305e+00 -4.19145286e-01 3.10636789e-01
4.85243022e-01 -5.05793154e-01 1.23120916e+00 7.19030559e-01
-2.81759858e-01 -8.17384720e-01 -8.70452166e-01 -7.43148983e-01
5.47860205e-01 -4.69284236e-01 9.18621302e-01 8.39909971e-01
6.68654293e-02 6.34729505e-01 -2.24976435e-01 1.58507839e-01
7.05394149e-01 7.48885393e-01 8.91513884e-01 -1.22468424e+00
-5.03831021e-02 -2.14704737e-01 -6.79225564e-01 -1.25391984e+00
4.75457937e-01 -1.34119046e+00 -5.95719695e-01 -1.36837411e+00
3.55288059e-01 -1.26727358e-01 -2.81842887e-01 5.35207868e-01
1.30907774e-01 1.61678255e-01 -4.30153400e-01 -1.52714059e-01
-8.19349647e-01 9.53997970e-01 1.33386087e+00 -2.81329840e-01
-5.25677323e-01 -2.46934503e-01 -4.47522432e-01 2.85241812e-01
9.59819615e-01 -3.50675941e-01 -1.44078180e-01 2.57553458e-02
5.85910261e-01 1.61022365e-01 4.20899212e-01 -9.14030254e-01
2.01376930e-01 -3.72874737e-01 7.61102855e-01 -9.91454840e-01
3.39584947e-01 -2.56467462e-01 -1.04303055e-01 5.60581923e-01
-2.36500606e-01 -2.75137722e-01 1.69079587e-01 6.16762698e-01
2.44639292e-02 1.03469849e-01 7.45323420e-01 -4.76785414e-02
-1.73641309e-01 1.15959632e+00 -3.25973779e-01 -2.75738537e-01
8.52778614e-01 -2.57012665e-01 -1.71040803e-01 -8.74108821e-02
-7.15642571e-01 1.08862989e-01 4.65418458e-01 -1.36341304e-01
8.69071186e-01 -1.36854208e+00 -3.23468089e-01 3.52946788e-01
1.70891985e-01 1.81591079e-01 2.96405494e-01 5.07893622e-01
-1.13545382e+00 8.42691243e-01 -1.74859151e-01 -4.30883139e-01
-1.07135022e+00 6.58296168e-01 6.90639079e-01 -7.80272558e-02
-1.42639875e-01 5.95967531e-01 -2.23777249e-01 -8.55130255e-01
-1.03599159e-02 -5.65257788e-01 3.08853447e-01 -3.51213664e-01
2.92360097e-01 2.08777130e-01 3.86756688e-01 -5.26506007e-01
-3.54381323e-01 4.02322412e-01 -4.02111351e-01 8.10522676e-01
1.68525076e+00 5.07804275e-01 -1.18569829e-01 -7.46965930e-02
1.63509595e+00 -4.67608064e-01 -1.34531963e+00 -3.57353948e-02
-3.29614162e-01 -2.14980990e-01 1.90674126e-01 -4.27298278e-01
-6.13656819e-01 1.17340279e+00 3.00774068e-01 5.77564817e-03
5.30979276e-01 -3.36401433e-01 9.48625922e-01 1.53921092e+00
4.79535699e-01 -7.36702323e-01 1.49472460e-01 6.05657816e-01
4.34157580e-01 -1.50286222e+00 4.02543426e-01 -3.06419879e-01
3.35045941e-02 1.70090854e+00 4.77629840e-01 -9.00888443e-03
5.46382368e-01 -2.44721800e-01 -5.26642382e-01 -5.61693907e-01
-7.32760966e-01 -6.01972975e-02 3.66886497e-01 5.99729300e-01
9.69779611e-01 -5.84838986e-02 2.62474030e-01 3.12508017e-01
-3.05782229e-01 1.13569610e-02 5.55253208e-01 8.63807499e-01
-2.52161235e-01 -1.53938746e+00 1.30565986e-01 2.84542859e-01
-3.27575386e-01 -3.60798240e-01 -8.70988071e-01 5.47320664e-01
-1.25040486e-01 9.28406656e-01 -2.37531245e-01 -3.52647781e-01
2.23044455e-02 1.21793099e-01 1.10318899e+00 -6.52929544e-01
-2.64643162e-01 -5.16916394e-01 -8.90298039e-02 -7.49239683e-01
-3.10080945e-01 -6.65041506e-02 -1.39136302e+00 -1.10156739e+00
-2.61558086e-01 3.96190763e-01 5.24373591e-01 7.64095485e-01
8.34003985e-01 1.23695664e-01 8.85202765e-01 -1.00766373e+00
-5.77848673e-01 -4.65951055e-01 -4.79029655e-01 2.47722626e-01
2.97801822e-01 -4.08504218e-01 5.89245707e-02 -3.00122321e-01] | [5.147913932800293, 5.763134002685547] |
4d7f4898-8a63-49a8-816c-8b66eda88440 | face-alignment-assisted-by-head-pose | 1507.03148 | null | http://arxiv.org/abs/1507.03148v2 | http://arxiv.org/pdf/1507.03148v2.pdf | Face Alignment Assisted by Head Pose Estimation | In this paper we propose a supervised initialization scheme for cascaded face
alignment based on explicit head pose estimation. We first investigate the
failure cases of most state of the art face alignment approaches and observe
that these failures often share one common global property, i.e. the head pose
variation is usually large. Inspired by this, we propose a deep convolutional
network model for reliable and accurate head pose estimation. Instead of using
a mean face shape, or randomly selected shapes for cascaded face alignment
initialisation, we propose two schemes for generating initialisation: the first
one relies on projecting a mean 3D face shape (represented by 3D facial
landmarks) onto 2D image under the estimated head pose; the second one searches
nearest neighbour shapes from the training set according to head pose distance.
By doing so, the initialisation gets closer to the actual shape, which enhances
the possibility of convergence and in turn improves the face alignment
performance. We demonstrate the proposed method on the benchmark 300W dataset
and show very competitive performance in both head pose estimation and face
alignment. | ['Hatice Gunes', 'Yichi Zhang', 'Wenxuan Mou', 'Heng Yang', 'Ioannis Patras', 'Peter Robinson'] | 2015-07-11 | null | null | null | null | ['head-pose-estimation'] | ['computer-vision'] | [-9.88391340e-02 3.80016178e-01 9.47301984e-02 -6.18893445e-01
-5.33508003e-01 -3.45980316e-01 5.84642649e-01 -2.45557889e-01
-3.71633589e-01 3.63122076e-01 2.75220066e-01 2.45446667e-01
-4.70879041e-02 -4.52520698e-01 -8.39915335e-01 -8.95605445e-01
1.87241629e-01 9.34007406e-01 -1.57628432e-01 -3.18226755e-01
1.91534713e-01 8.28538775e-01 -1.57835734e+00 -5.02961576e-01
3.98976982e-01 8.20697665e-01 -2.82945901e-01 2.33279929e-01
1.00141495e-01 -1.17142588e-01 -5.65679193e-01 -6.52663708e-01
4.48366344e-01 -5.11572063e-01 -7.42900610e-01 4.33148563e-01
7.68216848e-01 -2.78523296e-01 2.50613600e-01 9.54828262e-01
1.00489366e+00 -3.65186818e-02 7.24208951e-01 -1.17226720e+00
-1.76994354e-02 2.61246294e-01 -8.86152923e-01 -2.46831313e-01
5.02851486e-01 -1.93040535e-01 6.65199339e-01 -1.11511278e+00
6.96685076e-01 1.35077536e+00 9.36442614e-01 8.31504762e-01
-1.05774724e+00 -6.42108917e-01 1.66575040e-03 -9.09135938e-02
-1.69803476e+00 -1.02243090e+00 9.97439861e-01 -2.61903048e-01
3.89521420e-01 1.58996601e-02 6.42850161e-01 7.73132384e-01
-2.04405755e-01 5.55382073e-01 7.97351718e-01 -6.58325315e-01
8.13532546e-02 -1.84401587e-01 -2.55814314e-01 6.87844932e-01
7.21977428e-02 -8.59521702e-02 -3.83862883e-01 -2.39773974e-01
8.22902322e-01 -3.09430420e-01 -3.64772201e-01 -7.53226757e-01
-8.73883426e-01 6.73913360e-01 5.02935827e-01 3.79998446e-01
-3.85278523e-01 -8.34275708e-02 1.94281489e-01 -6.86024278e-02
4.25691158e-01 7.07354099e-02 -1.78370386e-01 3.25016171e-01
-1.10563171e+00 3.67931336e-01 6.76086903e-01 7.06350982e-01
7.79677868e-01 -9.44427550e-02 -4.30063717e-03 8.57360542e-01
5.96578598e-01 5.25760770e-01 3.89813215e-01 -6.28797174e-01
2.65653819e-01 4.64447469e-01 6.08820431e-02 -9.97215807e-01
-6.36145711e-01 -4.01005834e-01 -7.44547248e-01 1.69005916e-01
7.52163410e-01 -1.68390065e-01 -9.59300756e-01 1.84914625e+00
8.08940589e-01 2.82845199e-01 -1.18038252e-01 8.46894264e-01
7.86568880e-01 1.01303281e-02 -9.47391018e-02 -2.63373971e-01
1.32173312e+00 -8.00911725e-01 -6.08058989e-01 -1.01282060e-01
4.43438560e-01 -9.93070781e-01 4.89455819e-01 9.17512253e-02
-1.16740227e+00 -5.77385545e-01 -8.22549403e-01 1.58764079e-01
6.23403601e-02 3.20834696e-01 1.18520655e-01 7.87826061e-01
-1.37814355e+00 4.23482269e-01 -6.49568260e-01 -4.80571300e-01
4.46413666e-01 8.24659586e-01 -7.52764523e-01 1.89927906e-01
-6.82279468e-01 7.92606950e-01 2.24390164e-01 4.31384295e-01
-5.40291607e-01 -3.99573952e-01 -7.86753356e-01 -1.04691707e-01
1.66427925e-01 -6.73736453e-01 1.17148161e+00 -1.07332921e+00
-1.85691071e+00 1.27876627e+00 -4.49369848e-01 -2.06215128e-01
7.75422513e-01 -3.76062363e-01 6.61462396e-02 -1.72316819e-01
-1.59525827e-01 8.03518474e-01 1.33422506e+00 -1.27712083e+00
-2.03792915e-01 -8.42890203e-01 -4.45037544e-01 4.38370317e-01
-2.96117038e-01 2.62956262e-01 -7.45810688e-01 -2.39192039e-01
3.20928335e-01 -1.06153226e+00 -9.83757079e-02 -5.34928683e-03
-4.26183343e-01 -3.55561316e-01 7.38372028e-01 -6.24279141e-01
8.55029225e-01 -1.85501003e+00 4.79472876e-01 6.34196758e-01
1.71131730e-01 3.18350911e-01 -7.49581158e-02 3.67886876e-03
-4.46627706e-01 -2.81866342e-01 -7.48022050e-02 -1.01018953e+00
2.31777318e-02 -2.51657609e-02 -9.11215842e-02 8.44147503e-01
7.48729780e-02 9.57756758e-01 -5.79512894e-01 -4.59186792e-01
3.02442163e-02 7.29387462e-01 -5.96357942e-01 2.78437078e-01
1.35238126e-01 7.53854454e-01 -1.32145345e-01 3.85982782e-01
9.31837618e-01 2.78023750e-01 3.55541795e-01 -3.22141021e-01
2.67005283e-02 -3.97466160e-02 -1.37994182e+00 1.46435738e+00
-2.94789016e-01 5.73498160e-02 1.79276243e-01 -8.83021355e-01
1.26559472e+00 4.80036616e-01 5.75220823e-01 -3.16338062e-01
4.35321331e-01 2.40304977e-01 -8.50137994e-02 -2.20210180e-02
-4.31256182e-02 -2.18061656e-02 3.04680347e-01 5.64279199e-01
2.63179243e-01 -1.97044179e-01 -1.43362820e-01 -3.94535929e-01
2.76553661e-01 3.31791401e-01 5.35061002e-01 -3.72135699e-01
9.41504180e-01 -8.45342100e-01 3.72377366e-01 1.56442836e-01
-6.61706999e-02 9.23068345e-01 4.22113270e-01 -7.16662526e-01
-1.06416214e+00 -6.82891190e-01 -2.22841367e-01 9.16498363e-01
-1.58261791e-01 -2.95816511e-01 -1.25799561e+00 -8.72813463e-01
-1.37210757e-01 4.52134684e-02 -6.97920024e-01 1.99618340e-02
-1.00566900e+00 -7.87617207e-01 3.65916759e-01 4.87725556e-01
3.70352298e-01 -9.99621332e-01 -4.07335848e-01 -5.65059632e-02
-8.87818113e-02 -8.78746450e-01 -6.36792779e-01 -2.37463921e-01
-7.55615890e-01 -8.33269000e-01 -1.07363880e+00 -9.62618291e-01
1.11799324e+00 -1.46980405e-01 1.14597273e+00 4.91833210e-01
5.87234534e-02 1.90233335e-01 -9.96270403e-02 -4.46194172e-01
-3.32213879e-01 2.37807781e-01 4.40765828e-01 3.93317848e-01
3.61327887e-01 -5.71298897e-01 -6.72667027e-01 4.97715086e-01
-4.68952835e-01 -2.56436974e-01 4.29486990e-01 6.16743982e-01
4.76769447e-01 -2.18400314e-01 3.73170525e-01 -5.39744973e-01
3.61920863e-01 -5.90576977e-02 -7.45588899e-01 2.19745189e-01
-2.24901721e-01 1.05476938e-01 3.28958809e-01 -2.38094211e-01
-8.20468426e-01 6.38723314e-01 -7.45349467e-01 -3.68048787e-01
-3.96127552e-01 -7.12026581e-02 -4.15392011e-01 -3.97873878e-01
5.53628445e-01 1.61993340e-01 3.29430968e-01 -5.32060146e-01
2.49869362e-01 4.95760739e-01 6.36800230e-01 -5.86371779e-01
1.09580052e+00 4.97226238e-01 1.55031577e-01 -7.99186230e-01
-5.14986515e-01 -3.24294716e-01 -1.24306428e+00 -3.32135856e-01
7.83596456e-01 -6.89024448e-01 -9.12576258e-01 7.64747918e-01
-1.27180290e+00 -9.97636542e-02 1.09933898e-01 3.17915797e-01
-5.75384438e-01 4.26362813e-01 -1.34654686e-01 -7.77870059e-01
-6.65267467e-01 -1.24724996e+00 1.68429065e+00 1.86924979e-01
-3.18633169e-01 -9.33213711e-01 1.32113114e-01 6.15900150e-03
2.33865723e-01 1.89185500e-01 5.37799001e-01 -5.94946623e-01
-5.25231846e-02 -4.82239246e-01 1.17181443e-01 9.53552406e-03
1.84597924e-01 -3.25121395e-02 -1.02248073e+00 -5.67046225e-01
1.41117498e-01 -3.88413928e-02 2.71243751e-01 4.88946646e-01
7.83307433e-01 -1.08346611e-01 -4.19983357e-01 6.27125263e-01
1.06883180e+00 -1.06535986e-01 6.96563125e-01 1.24764502e-01
6.74770534e-01 7.82194912e-01 3.67456943e-01 4.73778456e-01
3.06753665e-01 1.26257980e+00 4.28492546e-01 -2.39380747e-01
-9.24042538e-02 -2.08686799e-01 1.87735826e-01 5.34109950e-01
-4.73205924e-01 1.24172047e-01 -9.09961998e-01 2.75336653e-01
-1.60152137e+00 -6.06696784e-01 1.74696639e-01 2.67124939e+00
6.71275914e-01 -8.78988728e-02 6.28347099e-01 2.18565181e-01
9.06693161e-01 -1.72069311e-01 -2.26146430e-01 -5.58453165e-02
2.03042999e-01 3.79225463e-01 1.03460409e-01 6.31119072e-01
-1.08075857e+00 9.06525791e-01 5.77970743e+00 4.72181469e-01
-1.27134585e+00 -3.08471750e-02 6.09263062e-01 8.15281272e-02
9.33709517e-02 -3.82105738e-01 -1.31132758e+00 2.83322603e-01
4.15517479e-01 1.19571887e-01 1.76466450e-01 6.67626381e-01
-1.07983269e-01 2.16792122e-01 -1.13971519e+00 1.11745358e+00
3.14213365e-01 -7.93078661e-01 -2.61264239e-02 2.69753039e-01
6.26035333e-01 -3.82126808e-01 3.74831483e-02 -2.92371791e-02
-5.85647076e-02 -1.11286986e+00 8.03121746e-01 3.80877048e-01
7.38595128e-01 -9.48535502e-01 9.25431252e-01 3.32861125e-01
-1.28376234e+00 1.62248984e-01 -2.94710904e-01 2.75375158e-01
2.30949804e-01 3.48524481e-01 -9.19122934e-01 5.01804471e-01
5.40824473e-01 2.92945564e-01 -5.94371438e-01 1.17913425e+00
-3.25006783e-01 1.25627965e-01 -6.71800613e-01 3.51773590e-01
-4.37666066e-02 -1.83943182e-01 6.10569835e-01 8.37278485e-01
4.25476462e-01 -3.07568520e-01 5.30045619e-03 4.27403748e-01
-1.24721199e-01 4.15649772e-01 -4.93293881e-01 6.21231854e-01
4.22966659e-01 1.42770994e+00 -8.61742139e-01 7.60055855e-02
-2.06078693e-01 8.81973743e-01 4.09554899e-01 4.90560308e-02
-7.40642250e-01 1.17760822e-01 4.12792623e-01 9.20038894e-02
3.80638272e-01 -7.07731545e-02 1.11129291e-01 -8.16574812e-01
2.13615328e-01 -8.92384470e-01 2.08550736e-01 -3.44840854e-01
-9.54833627e-01 9.25881326e-01 -4.25415747e-02 -1.04809833e+00
-6.12590849e-01 -3.37954193e-01 -7.46230662e-01 9.18959260e-01
-1.29177642e+00 -1.39429295e+00 -3.92299503e-01 5.91525257e-01
2.63077438e-01 1.94528867e-02 8.77336860e-01 3.37759554e-01
-5.82383096e-01 1.10096383e+00 -3.64909142e-01 2.73967296e-01
8.75900924e-01 -1.04039013e+00 7.36046970e-01 4.88213629e-01
1.96674407e-01 6.89030945e-01 7.18933463e-01 -5.01177073e-01
-1.11424255e+00 -8.91280174e-01 1.06269395e+00 -6.58032477e-01
5.67290857e-02 -4.68141317e-01 -7.88213551e-01 7.48896420e-01
4.38346006e-02 1.00776471e-01 3.87426525e-01 8.94252732e-02
-2.35874489e-01 -1.88866511e-01 -1.22285759e+00 6.46766067e-01
1.12005198e+00 -1.51777044e-01 -4.54770356e-01 3.08407098e-01
6.05995208e-02 -6.12274170e-01 -7.24654913e-01 5.62283218e-01
7.29698002e-01 -8.60255837e-01 1.08296907e+00 -4.38248187e-01
-5.70800491e-02 -3.27998817e-01 2.26761848e-01 -1.21937716e+00
-2.03552008e-01 -7.52899826e-01 7.82889128e-02 1.36962974e+00
7.85866603e-02 -4.78826225e-01 9.80893672e-01 4.60091472e-01
1.97717607e-01 -8.18148792e-01 -1.11875272e+00 -4.84295666e-01
1.31690815e-01 1.35339752e-01 9.66750026e-01 6.01867914e-01
-4.73808229e-01 2.62303531e-01 -3.54256868e-01 2.43623257e-01
7.32834339e-01 3.95234935e-02 1.19575953e+00 -1.42199016e+00
6.99164420e-02 -5.51392496e-01 -6.62932575e-01 -1.00253999e+00
4.86109883e-01 -6.12066627e-01 1.84115931e-01 -8.89999151e-01
5.87947257e-02 -3.45095307e-01 1.25265762e-01 5.40615976e-01
-1.09104715e-01 6.56413734e-01 -1.33883571e-02 5.87080084e-02
-2.45617941e-01 5.07760227e-01 1.03452921e+00 3.51255566e-01
-3.51580083e-01 3.55003536e-01 -4.54889894e-01 9.74186063e-01
6.65864170e-01 -2.61911303e-01 -1.13875061e-01 -3.26480091e-01
-1.43394992e-01 -1.42874643e-01 2.12669283e-01 -9.87789869e-01
1.64949358e-01 3.84481430e-01 6.27683818e-01 -4.37375695e-01
3.36652756e-01 -8.93590987e-01 1.78392977e-01 3.69245470e-01
3.54122333e-02 1.86227858e-01 1.54395580e-01 1.36404604e-01
5.58156706e-02 -3.49489123e-01 1.11381698e+00 2.31178571e-02
-3.04522723e-01 4.11989480e-01 2.19610259e-01 -2.65684694e-01
9.43486631e-01 -2.76385456e-01 2.62855381e-01 -4.93427217e-01
-7.87148952e-01 -6.31468371e-02 5.48631847e-01 4.41402704e-01
4.75729376e-01 -1.47387350e+00 -9.51357722e-01 6.93334699e-01
-1.14053287e-01 6.42917380e-02 -1.31106585e-01 1.11990130e+00
-3.10683459e-01 2.24902913e-01 -1.73651144e-01 -8.62036407e-01
-1.68783414e+00 3.70244384e-01 6.65484965e-01 6.25526905e-02
-3.93907964e-01 1.08323514e+00 2.61602432e-01 -6.56139791e-01
3.33148241e-01 1.79317638e-01 -5.41209579e-01 2.61252731e-01
6.07104421e-01 4.52189408e-02 3.91394377e-01 -1.43492651e+00
-5.21237910e-01 1.38874626e+00 -4.72027250e-02 9.07875001e-02
1.39524662e+00 -1.12582386e-01 -3.22260857e-01 -1.92962751e-01
1.24706399e+00 1.25850111e-01 -1.10142100e+00 -2.17384785e-01
-8.24783836e-03 -4.60272431e-01 -2.56941587e-01 -2.13563859e-01
-1.41369498e+00 7.52169073e-01 7.95476913e-01 -1.81645676e-01
1.07456136e+00 6.07098974e-02 6.06282890e-01 1.78315505e-01
3.91011715e-01 -7.63660550e-01 -6.27028570e-02 3.84110749e-01
1.07893157e+00 -1.00231254e+00 2.17475444e-02 -4.29479539e-01
-3.79667550e-01 1.25990510e+00 6.31840408e-01 -1.23779535e-01
6.28669977e-01 2.40811870e-01 2.25050107e-01 -3.08754146e-01
-3.95558774e-03 -2.39637077e-01 5.62170088e-01 5.40660143e-01
4.48477656e-01 -2.82694787e-01 -2.81896979e-01 1.62307739e-01
-4.78058100e-01 -1.45270824e-01 -1.01060227e-01 6.06599033e-01
-2.28225693e-01 -1.28519976e+00 -7.72870779e-01 -3.83830513e-03
-4.41096097e-01 1.89025149e-01 -5.44342399e-01 6.77414417e-01
1.20176531e-01 5.07283747e-01 1.58035129e-01 -5.77873588e-02
4.22409505e-01 3.41424525e-01 8.48088443e-01 -4.32871699e-01
-3.97269875e-01 3.58521014e-01 -2.92032838e-01 -4.60129648e-01
-4.78532553e-01 -6.40572786e-01 -9.72881496e-01 -2.26162046e-01
-5.49654782e-01 1.70654908e-01 6.92316115e-01 1.02069747e+00
2.22081736e-01 -1.38659835e-01 7.17437267e-01 -1.59145522e+00
-5.97266674e-01 -8.49649608e-01 -3.04979384e-01 5.34742355e-01
3.25852841e-01 -9.35806632e-01 -3.57911825e-01 3.34033892e-02] | [13.477272033691406, 0.3072497546672821] |
6afae744-4229-4769-9f45-7fa407f9ca38 | scaffold-induced-molecular-graph-simg | 2109.05012 | null | https://arxiv.org/abs/2109.05012v1 | https://arxiv.org/pdf/2109.05012v1.pdf | Scaffold-Induced Molecular Graph (SIMG): Effective Graph Sampling Methods for High-Throughput Computational Drug Discovery | Scaffold based drug discovery (SBDD) is a technique for drug discovery which pins chemical scaffolds as the framework of design. Scaffolds, or molecular frameworks, organize the design of compounds into local neighborhoods. We formalize scaffold based drug discovery into a network design. Utilizing docking data from SARS-CoV-2 virtual screening studies and JAK2 kinase assay data, we showcase how a scaffold based conception of chemical space is intuitive for design. Lastly, we highlight the utility of scaffold based networks for chemical space as a potential solution to the intractable enumeration problem of chemical space by working inductively on local neighborhoods. | ['Rick Stevens', 'Arvind Ramanathan', 'Max Zvyagin', 'Ashka Shah', 'Austin Clyde'] | 2021-09-10 | null | null | null | null | ['graph-sampling'] | ['graphs'] | [ 3.30563039e-01 2.27162614e-01 -9.97106373e-01 1.05536133e-01
-1.73629999e-01 -1.03359592e+00 4.19659734e-01 4.42621201e-01
7.87985623e-02 1.45983732e+00 4.59794760e-01 -1.23269928e+00
-5.78575969e-01 -9.12947237e-01 -8.18037450e-01 -4.18250412e-01
-3.44269156e-01 1.55536890e-01 1.15527004e-01 -2.74528384e-01
4.86495763e-01 1.12106407e+00 -5.92197478e-01 -3.05385981e-02
7.10461199e-01 4.31129247e-01 -3.47819068e-02 1.79289609e-01
-1.31361902e-01 5.78735948e-01 -5.29310346e-01 1.79015666e-01
1.41087398e-01 -5.61889172e-01 -8.69898081e-01 -4.91725057e-01
1.17163271e-01 3.01390707e-01 -2.20007524e-02 7.63333082e-01
5.52404881e-01 1.14115894e-01 6.99702859e-01 -6.24058187e-01
-8.52077246e-01 2.86861032e-01 2.71803856e-01 3.14988941e-01
6.75941885e-01 -5.48763759e-02 9.73461032e-01 -1.35586786e+00
1.12990546e+00 1.09041274e+00 6.34053826e-01 5.01538694e-01
-1.74487400e+00 -5.52936018e-01 1.63095742e-01 -1.26865283e-01
-1.57477045e+00 -4.54429686e-01 5.22347450e-01 -7.03677714e-01
1.18901682e+00 4.11149710e-01 8.97716045e-01 7.33198166e-01
5.10358393e-01 1.75842240e-01 6.53196096e-01 -1.15690663e-01
8.93819153e-01 -4.25964028e-01 -8.40615854e-02 4.72879827e-01
6.99158847e-01 1.92639932e-01 -8.86280775e-01 -9.30916846e-01
1.12844288e+00 1.31823584e-01 -3.99186641e-01 -6.70080900e-01
-8.64551365e-01 1.06638908e+00 6.28780901e-01 1.58919752e-01
-3.71232241e-01 1.75389394e-01 1.07184216e-01 -2.51516588e-02
1.57522470e-01 1.21146643e+00 -8.70610952e-01 6.18346035e-01
-5.15690327e-01 1.32119834e-01 7.20669627e-01 6.52908802e-01
5.96948147e-01 -5.17504811e-02 3.21490020e-01 1.20483125e-02
6.54881239e-01 -2.09829450e-01 -2.78000087e-01 -3.63935649e-01
-6.85787350e-02 2.23078638e-01 4.28453475e-01 -7.83022285e-01
-7.38969922e-01 -6.75186276e-01 -2.14657813e-01 3.48326527e-02
2.82736659e-01 -1.52269423e-01 -7.45693743e-01 1.79382813e+00
5.87562978e-01 2.95065135e-01 2.61534929e-01 4.62383419e-01
7.44318545e-01 2.80696452e-01 6.16970956e-01 -5.93185127e-01
9.41148102e-01 -4.92894769e-01 -4.58700031e-01 6.72956109e-01
8.00758362e-01 -4.98387396e-01 4.43554759e-01 6.35495186e-01
-8.55363846e-01 1.28506035e-01 -1.22417939e+00 4.17404562e-01
-6.54145062e-01 -4.65124398e-01 1.09547544e+00 7.24478483e-01
-1.29739761e+00 1.21017337e+00 -6.02945626e-01 -2.49664530e-01
8.58461142e-01 9.01602805e-01 -3.23365033e-01 2.78944373e-01
-1.13626730e+00 9.73489046e-01 4.82596606e-01 1.32018002e-02
-1.19811141e+00 -1.03622448e+00 -5.76187432e-01 -2.47367159e-01
2.70033717e-01 -1.06622028e+00 5.78258753e-01 -7.91374668e-02
-1.40577579e+00 4.28800993e-02 -1.63206056e-01 -4.39490855e-01
-2.87032962e-01 3.55351359e-01 -5.25514603e-01 1.76736280e-01
2.52815425e-01 3.48357916e-01 5.53859435e-02 -1.03154695e+00
4.09002714e-02 -2.35425517e-01 -7.23871663e-02 7.20878020e-02
2.10095853e-01 1.32423967e-01 4.83614117e-01 -4.79585558e-01
2.09431008e-01 -6.31686866e-01 -1.05324185e+00 -2.28256240e-01
-1.07583034e+00 -3.38917255e-01 2.70494670e-01 -3.03707570e-02
1.23669302e+00 -1.53442049e+00 1.33928508e-01 1.00592351e+00
7.88691759e-01 -8.16512704e-02 -3.19178969e-01 1.06932211e+00
-9.24958527e-01 8.40245008e-01 4.26057160e-01 7.17978716e-01
-4.44029838e-01 -3.03419381e-01 -3.24672222e-01 8.40052426e-01
1.24356627e-01 1.06257868e+00 -1.07098293e+00 2.79177818e-02
-1.97497442e-01 5.07599711e-01 -5.88174999e-01 -6.15095317e-01
-6.43605947e-01 8.19308221e-01 -1.19731331e+00 1.23277009e+00
4.50411588e-01 -4.78348404e-01 6.65776253e-01 -6.74839318e-02
-4.11086708e-01 6.07764423e-01 -6.29002571e-01 1.76072180e+00
5.36601543e-01 1.31809758e-02 -4.70222592e-01 -5.58674872e-01
7.83858955e-01 5.64563036e-01 7.44130731e-01 -3.47434551e-01
-1.19169936e-01 4.93007958e-01 1.69041567e-02 1.76190004e-01
-2.77893096e-01 -4.96195197e-01 3.72795165e-01 1.59181446e-01
-1.96048766e-01 3.07735085e-01 -1.07447878e-02 1.40026093e-01
1.49541521e+00 7.66166374e-02 4.33194995e-01 -8.73648226e-01
3.27196240e-01 5.24713755e-01 5.31224668e-01 3.60615015e-01
1.28604040e-01 -6.33900762e-02 5.30089438e-01 -7.51024663e-01
-1.00926244e+00 -1.21177185e+00 -5.43803334e-01 8.12882841e-01
8.79827980e-03 -8.83733690e-01 -3.77281606e-01 -4.53387290e-01
-3.73695977e-02 1.64907724e-01 -8.52813125e-01 1.11589409e-01
-3.60906959e-01 -4.26488280e-01 5.48291147e-01 8.83608237e-02
-5.79797268e-01 -6.67992234e-01 1.43185750e-01 7.06361651e-01
6.71974480e-01 -2.98557371e-01 -3.49265933e-01 7.77535856e-01
-9.40647423e-01 -1.53603160e+00 -5.32585740e-01 -8.24293017e-01
7.27141857e-01 9.40068588e-02 8.17968249e-01 2.00708932e-03
-4.01404172e-01 -9.89839807e-02 -9.54778343e-02 -5.03307343e-01
-2.90530562e-01 -3.18150401e-01 5.84323704e-01 -7.83381581e-01
5.09111509e-02 -1.06106949e+00 -1.01587331e+00 3.67690027e-01
-5.34513116e-01 -4.22379792e-01 2.22658500e-01 5.92936516e-01
1.13759756e+00 -1.94272920e-01 1.14561379e+00 -8.08164477e-01
9.78640735e-01 -8.65853429e-01 -5.37217438e-01 2.12231338e-01
-5.73176980e-01 3.20393853e-02 4.80479330e-01 -3.01344424e-01
-1.69347242e-01 3.37527543e-01 5.67873158e-02 7.31982887e-02
2.55378366e-01 9.72407103e-01 -4.49941069e-01 -6.73960328e-01
1.23786664e+00 7.25611001e-02 2.84690738e-01 -6.07201457e-01
3.96372318e-01 -2.40347087e-01 -3.15128081e-02 -7.70090878e-01
5.65522432e-01 3.13979983e-01 8.39199722e-01 -6.66144907e-01
-9.95780528e-02 -2.82672405e-01 -3.47993076e-01 3.77798110e-01
9.71390128e-01 -8.87123346e-01 -1.41539645e+00 -7.37354457e-01
-1.28011727e+00 -1.28584310e-01 -2.32776105e-01 3.94062668e-01
-6.84511960e-01 4.89202961e-02 -6.58772811e-02 -2.54619002e-01
-1.46589056e-01 -1.11448228e+00 3.93234372e-01 -1.46483317e-01
-7.08980978e-01 -1.19679034e+00 6.35408640e-01 -2.18645006e-01
2.07771748e-01 1.02564418e+00 1.37213993e+00 -1.16306925e+00
-1.01627886e+00 1.33860976e-01 9.59451273e-02 -4.44242120e-01
5.41775405e-01 -2.20815331e-01 -4.04652238e-01 4.26060893e-02
-4.17315602e-01 6.40980750e-02 2.80930579e-01 7.04046667e-01
7.93364584e-01 -5.37317395e-01 -8.00136745e-01 3.61496598e-01
1.61894405e+00 9.32807028e-01 7.34042346e-01 3.81038785e-01
6.53503895e-01 3.28813672e-01 3.90351295e-01 5.75595796e-01
-2.97012150e-01 3.87871802e-01 4.03393745e-01 -4.44097966e-01
4.56091538e-02 -5.99386275e-01 -1.99108645e-01 -1.72756791e-01
-1.70114517e-01 -1.50455132e-01 -1.01920652e+00 1.83433726e-01
-1.56959593e+00 -7.97042727e-01 -2.06829011e-01 2.11993361e+00
1.42641246e+00 -3.21153998e-01 6.09946609e-01 -3.62276971e-01
5.31602800e-01 -5.32937407e-01 -9.16355848e-01 -2.97899753e-01
-3.69428068e-01 8.33241999e-01 7.39545047e-01 6.77918732e-01
-7.71317363e-01 1.16457081e+00 8.73237133e+00 1.05955064e+00
-7.85471439e-01 -2.90918022e-01 6.82355404e-01 7.73201212e-02
-7.83795595e-01 5.00872433e-01 -6.67441666e-01 9.82800126e-02
9.53648329e-01 -5.12672484e-01 3.37220818e-01 6.15047514e-01
7.46721148e-01 1.38266191e-01 -1.54763663e+00 6.13061011e-01
-7.83790946e-01 -2.80067468e+00 4.51591074e-01 4.87763405e-01
8.98657680e-01 -7.26948259e-03 -2.09033098e-02 -7.73322999e-01
5.95029712e-01 -1.76674187e+00 4.60407227e-01 3.21785927e-01
9.29154694e-01 -7.55891621e-01 -1.96831286e-01 -1.27960443e-01
-9.97727275e-01 4.58738618e-02 -1.53626561e-01 6.06369935e-02
1.20151183e-02 1.86777011e-01 -1.23511553e+00 3.84975344e-01
-4.65840753e-03 6.98557198e-01 -3.03011864e-01 1.42340982e+00
-1.41880959e-01 2.22201392e-01 -1.69087753e-01 -2.39014193e-01
3.29302192e-01 -1.51983276e-01 5.40830135e-01 6.83222115e-01
-2.11710870e-01 4.19424891e-01 5.10346174e-01 1.01370752e+00
1.71368364e-02 3.55240703e-01 -9.25452769e-01 -3.42048138e-01
7.14090288e-01 3.83086205e-01 -9.34041440e-01 1.06564043e-02
9.32216197e-02 2.69068390e-01 -3.28199893e-01 7.66925097e-01
-3.98565680e-01 -4.18119460e-01 1.18119144e+00 3.52217436e-01
2.07203507e-01 -3.25959086e-01 -5.51102877e-01 -4.42586422e-01
-7.23399401e-01 -7.42732525e-01 7.55498409e-02 -4.39568251e-01
-1.05516529e+00 2.37997636e-01 -2.23021999e-01 -1.06501317e+00
4.19902146e-01 -5.18213511e-01 -5.90137303e-01 1.03953838e+00
-9.29778039e-01 -6.94134355e-01 7.15080082e-01 6.41879499e-01
1.06716082e-01 -4.94333245e-02 9.97904003e-01 3.13230492e-02
-3.52917045e-01 1.46809682e-01 2.94701129e-01 -6.19812667e-01
3.58238161e-01 -8.58830154e-01 3.73879224e-01 3.10649455e-01
-2.29282424e-01 1.48642242e+00 9.04116571e-01 -1.19805205e+00
-1.37458885e+00 -1.00386381e+00 5.23692906e-01 -8.53009939e-01
1.04243398e+00 -1.45830184e-01 -3.15185338e-01 2.00173020e-01
-1.54730082e-01 -2.84384906e-01 1.65833306e+00 1.41468883e-01
-3.47950459e-01 6.60330236e-01 -1.28228962e+00 7.87673175e-01
1.33548474e+00 -5.83870888e-01 1.47697479e-02 1.03271496e+00
1.26439095e+00 -1.88092455e-01 -1.32634568e+00 2.87568212e-01
3.64153594e-01 -3.49126071e-01 1.42252743e+00 -1.37151849e+00
-4.74103503e-02 -9.37612176e-01 -1.05810106e-01 -9.31670189e-01
-7.40510702e-01 -1.30424285e+00 2.05175012e-01 2.83819914e-01
1.07852483e+00 -6.63471341e-01 1.18624139e+00 4.91443038e-01
-2.75674433e-01 -1.13559210e+00 -9.97271299e-01 -9.97644067e-01
4.06862646e-01 7.07537755e-02 8.17404687e-01 1.11712182e+00
6.35356247e-01 7.05015182e-01 2.54714221e-01 -1.45422593e-01
3.45336914e-01 -3.19602758e-01 3.34601730e-01 -1.43185174e+00
-4.15728182e-01 -5.34860253e-01 -5.71800470e-01 -7.56938696e-01
-1.10789016e-01 -8.79541278e-01 -5.77620208e-01 -1.65380108e+00
1.40106708e-01 -6.73629880e-01 -3.34430188e-01 5.89282811e-01
6.59158468e-01 -5.00801578e-02 -2.77412593e-01 4.49746698e-01
-5.47916889e-01 2.08888561e-01 1.35315406e+00 -1.24291830e-01
-9.78951693e-01 -1.85816333e-01 -1.46327245e+00 3.30688834e-01
8.75586569e-01 -4.09137130e-01 -7.87875175e-01 4.17637497e-01
8.07127118e-01 2.53526330e-01 4.55622151e-02 -3.51249188e-01
2.76186705e-01 -9.04222965e-01 5.43188155e-01 -6.01695895e-01
1.07299656e-01 -6.21202230e-01 7.74549961e-01 6.29913628e-01
-1.95213482e-01 -4.22188550e-01 2.68664360e-01 1.05678368e+00
1.57525733e-01 2.41774455e-01 4.02112246e-01 -6.49958774e-02
-1.44521117e-01 5.09472013e-01 -6.77013934e-01 -4.48731571e-01
9.23957109e-01 -1.02352047e+00 -4.34922367e-01 2.37254500e-01
-1.50895059e+00 -5.76501638e-02 5.19846976e-01 -8.77358764e-02
7.70725191e-01 -1.33931160e+00 -1.23959139e-01 -2.74846405e-01
8.87963995e-02 -1.26495570e-01 -2.85864592e-01 7.45945811e-01
-8.63123119e-01 9.41990137e-01 4.85077314e-02 -2.26190284e-01
-1.11219800e+00 8.74659359e-01 4.45524186e-01 2.77407616e-01
-7.12908432e-02 1.04916465e+00 4.93503481e-01 8.03724453e-02
8.48352090e-02 -1.92235768e-01 -1.29838392e-01 -9.91218165e-02
5.42778075e-01 2.45441481e-01 -1.30454987e-01 -2.04172879e-01
-1.01830518e+00 4.45528150e-01 1.62385758e-02 2.81005025e-01
1.57995939e+00 2.69824684e-01 -4.91933197e-01 -2.71375537e-01
1.04813838e+00 8.95966217e-02 -7.42450774e-01 1.16057687e-01
2.48365253e-01 -1.23918779e-01 -2.50743032e-01 -9.28388417e-01
-3.53338867e-02 -7.62372464e-02 2.48417452e-01 -1.51316017e-01
4.42829609e-01 3.45196754e-01 6.15747869e-02 6.92845583e-01
5.53654432e-01 -7.57359862e-01 2.31790245e-01 3.43730152e-01
9.25427616e-01 -3.21697950e-01 4.21428710e-01 -9.24187541e-01
1.63988605e-01 1.18256736e+00 3.26546165e-03 -7.86171407e-02
1.03586876e+00 -2.66350120e-01 -6.28873348e-01 -9.14414227e-01
-4.52444226e-01 -2.39778217e-02 1.27139747e-01 8.49122882e-01
4.98493135e-01 5.58727756e-02 -6.70006216e-01 5.72106242e-01
8.62738267e-02 -3.23154852e-02 4.22208816e-01 1.29178119e+00
-8.81064117e-01 -1.77792740e+00 -5.28354347e-01 -8.28443766e-02
-4.18910801e-01 -6.94400370e-01 -1.13692117e+00 4.85142976e-01
1.92249015e-01 1.05614340e+00 -4.24293756e-01 -3.35279226e-01
2.37503156e-01 -1.24381095e-01 5.77416182e-01 -1.11779177e+00
-4.36007023e-01 5.24691343e-01 4.70633000e-01 -4.64276701e-01
-3.29782963e-01 -2.03400195e-01 -1.60319829e+00 -3.41071695e-01
-6.97901845e-01 7.16548681e-01 7.18459904e-01 7.39019334e-01
9.27683055e-01 3.53115737e-01 7.68783808e-01 -4.72371578e-01
1.11295424e-01 -5.27864099e-02 -7.80894220e-01 -7.95487881e-01
2.64931619e-01 -6.51859939e-01 9.31003988e-02 9.87775549e-02] | [5.007588863372803, 5.773229598999023] |
c3e67ea2-b108-4525-8274-26c854c2e113 | tab-vcr-tags-and-attributes-based-vcr | 1910.14671 | null | https://arxiv.org/abs/1910.14671v2 | https://arxiv.org/pdf/1910.14671v2.pdf | TAB-VCR: Tags and Attributes based Visual Commonsense Reasoning Baselines | Reasoning is an important ability that we learn from a very early age. Yet, reasoning is extremely hard for algorithms. Despite impressive recent progress that has been reported on tasks that necessitate reasoning, such as visual question answering and visual dialog, models often exploit biases in datasets. To develop models with better reasoning abilities, recently, the new visual commonsense reasoning (VCR) task has been introduced. Not only do models have to answer questions, but also do they have to provide a reason for the given answer. The proposed baseline achieved compelling results, leveraging a meticulously designed model composed of LSTM modules and attention nets. Here we show that a much simpler model obtained by ablating and pruning the existing intricate baseline can perform better with half the number of trainable parameters. By associating visual features with attribute information and better text to image grounding, we obtain further improvements for our simpler & effective baseline, TAB-VCR. We show that this approach results in a 5.3%, 4.4% and 6.5% absolute improvement over the previous state-of-the-art on question answering, answer justification and holistic VCR. | ['Alexander G. Schwing', 'Jingxiang Lin', 'Unnat Jain'] | 2019-10-31 | null | null | null | neurips-2019-12 | ['visual-commonsense-reasoning'] | ['reasoning'] | [ 2.55042404e-01 5.70514619e-01 4.01375480e-02 -3.23064655e-01
-6.51669443e-01 -5.28735518e-01 8.32081318e-01 1.86093435e-01
-5.45496523e-01 6.35741293e-01 4.13421243e-01 -6.84411108e-01
1.20939091e-01 -7.44874060e-01 -8.93153310e-01 -1.17183745e-01
5.72074354e-01 6.85584784e-01 2.87981272e-01 -3.07294458e-01
4.73048240e-01 2.36700445e-01 -1.38980401e+00 8.99132192e-01
8.40394020e-01 9.64584231e-01 2.17200257e-02 8.60651135e-01
-6.31942332e-01 1.77756441e+00 -7.92009175e-01 -8.21894526e-01
-2.00747713e-01 -5.66607416e-01 -1.32373405e+00 -2.75242120e-01
9.59211051e-01 -5.04823923e-01 -3.15449119e-01 7.63280213e-01
3.70942920e-01 1.19077533e-01 7.60012269e-01 -1.07836950e+00
-1.59820545e+00 7.41442919e-01 -5.03957093e-01 2.50083148e-01
6.53705299e-01 4.43195850e-01 1.15703738e+00 -8.99805129e-01
6.70548856e-01 1.53273487e+00 3.99418533e-01 1.07596588e+00
-1.32952130e+00 -4.02663976e-01 2.97527194e-01 7.55585253e-01
-8.11397612e-01 -2.89569646e-01 6.53623164e-01 -3.25231820e-01
1.23763216e+00 4.24029708e-01 5.41264892e-01 1.20679569e+00
-2.91033626e-01 1.00874615e+00 1.35320425e+00 -5.45331538e-01
6.14836030e-02 2.54109859e-01 2.15974241e-01 9.15219426e-01
2.63180807e-02 -3.64533007e-01 -5.27329445e-01 3.31641376e-01
5.41926682e-01 -3.78640220e-02 -2.93863565e-01 -4.26038265e-01
-1.16333985e+00 9.68395710e-01 1.04846632e+00 2.43097737e-01
-1.51072130e-01 6.34385407e-01 3.72614443e-01 3.14229608e-01
7.12656975e-02 6.74792171e-01 -6.42080456e-02 7.61780515e-02
-7.52396822e-01 2.63848841e-01 8.64999175e-01 6.95000231e-01
4.32829827e-01 1.66856602e-01 -8.25101495e-01 7.28719950e-01
2.77967393e-01 4.40678775e-01 2.07123935e-01 -1.19884515e+00
7.31363177e-01 1.09266257e+00 2.62918096e-04 -9.42899644e-01
-3.62962693e-01 -2.51013547e-01 -6.80577815e-01 3.91316891e-01
7.34261155e-01 1.16259381e-01 -1.20065391e+00 1.72934043e+00
2.41926715e-01 -4.28661674e-01 1.61703527e-01 9.71748173e-01
1.47205377e+00 4.21611726e-01 3.62960756e-01 3.98505270e-01
1.63312125e+00 -1.27268386e+00 -7.98194945e-01 -5.02150059e-01
4.53849584e-01 -4.67169702e-01 1.71431720e+00 1.85395554e-01
-1.45090675e+00 -3.88083756e-01 -1.09325206e+00 -8.62002313e-01
-6.38686359e-01 -1.95940547e-02 8.01960170e-01 4.10620660e-01
-1.25348055e+00 4.12257761e-01 -9.16396156e-02 -3.53126407e-01
9.16611016e-01 4.74045239e-02 -3.45496982e-01 -2.34609306e-01
-1.09937429e+00 1.45212436e+00 1.53889120e-01 8.10701922e-02
-8.89981806e-01 -7.22946048e-01 -7.84874201e-01 3.54962289e-01
7.03042567e-01 -1.12644804e+00 1.37642765e+00 -9.23189700e-01
-1.22568882e+00 1.16402280e+00 -1.16456196e-01 -6.80298030e-01
7.73242652e-01 -5.00099838e-01 8.29124153e-02 4.23298955e-01
-4.31646667e-02 8.55172157e-01 7.40732491e-01 -1.33833218e+00
-1.10941336e-01 -2.63932317e-01 7.07582355e-01 2.75340788e-02
-6.16271682e-02 -6.78338483e-02 -3.68153244e-01 -3.84517610e-01
-1.22814357e-01 -5.90608239e-01 1.29133373e-01 4.72695857e-01
-3.70813280e-01 -5.14471829e-01 5.69645047e-01 -9.41980660e-01
8.37152362e-01 -1.79600787e+00 4.07157600e-01 -2.81651407e-01
5.95828831e-01 4.16419148e-01 -6.72294199e-02 3.95442724e-01
2.38893125e-02 3.81286740e-01 -2.87487477e-01 -2.88479090e-01
4.04604346e-01 1.62672639e-01 -5.52691519e-01 7.44708255e-02
5.03823519e-01 1.46290433e+00 -7.02748060e-01 -6.65523350e-01
1.82396904e-01 4.43390012e-01 -6.65022016e-01 2.66890436e-01
-6.99767649e-01 2.13875607e-01 3.53928143e-03 5.93969166e-01
4.87913907e-01 -7.36755848e-01 -2.34339554e-02 -1.35382801e-01
3.19891304e-01 2.04348490e-01 -6.61439359e-01 1.63379192e+00
-5.36561668e-01 1.07874322e+00 -1.31018683e-01 -8.81634355e-01
7.94816494e-01 1.99310958e-01 -4.43799108e-01 -1.24430072e+00
5.06012142e-02 -2.27920581e-02 1.54570505e-01 -7.35961199e-01
2.35433340e-01 -1.70134604e-01 6.13405667e-02 2.57264227e-01
1.09660231e-01 -1.92232326e-01 2.10247174e-01 6.94686353e-01
1.07645631e+00 3.86935145e-01 2.41490826e-01 1.48627013e-02
8.58818650e-01 1.98055595e-01 -6.97534531e-02 9.01353002e-01
-2.92773396e-01 5.09412408e-01 7.90860057e-01 -5.14093161e-01
-1.14719760e+00 -9.78654563e-01 2.77981967e-01 1.07917464e+00
-3.58231403e-02 -1.92861810e-01 -7.17252374e-01 -7.41541743e-01
1.07747480e-01 1.12883878e+00 -1.02326536e+00 -4.80126515e-02
-8.03452611e-01 -1.37426525e-01 6.05489433e-01 9.49623764e-01
9.10210669e-01 -1.40748835e+00 -8.43366683e-01 -9.32440236e-02
-3.96201853e-03 -1.22533011e+00 7.40298405e-02 -6.10427409e-02
-5.91837823e-01 -1.05462277e+00 -8.70423019e-01 -6.55803084e-01
4.24704880e-01 8.04878399e-02 1.69774663e+00 6.62617624e-01
-2.16949761e-01 4.47903991e-01 -2.59608209e-01 -5.32240570e-01
-2.02201471e-01 -1.07621402e-01 -8.56730700e-01 -2.94195384e-01
2.65550375e-01 -3.65083307e-01 -7.40193129e-01 -1.84638679e-01
-8.01571846e-01 4.70908582e-01 5.93221307e-01 9.00737226e-01
2.02457443e-01 -8.96149218e-01 5.85995197e-01 -1.00000918e+00
5.74712813e-01 -2.09236994e-01 -2.34445766e-01 6.19994342e-01
-4.44281340e-01 5.24514735e-01 6.31845891e-01 -2.27265880e-01
-1.34563053e+00 -3.47598523e-01 -2.29390517e-01 -4.05193776e-01
-1.32931471e-01 2.15058312e-01 7.80742094e-02 1.82790935e-01
7.26182282e-01 -6.58067092e-02 -1.96550816e-01 -2.36800849e-01
1.00300765e+00 2.57013977e-01 6.58802748e-01 -5.56499064e-01
7.24277675e-01 6.55574322e-01 6.87664822e-02 -4.06174093e-01
-1.24850130e+00 -1.36494949e-01 -3.50663781e-01 -2.28571057e-01
1.37733638e+00 -7.32659876e-01 -1.33170199e+00 -6.69092461e-02
-1.46471572e+00 -5.43351948e-01 -2.38599181e-01 -1.49223626e-01
-4.94169384e-01 2.89477319e-01 -5.46968639e-01 -9.74833488e-01
-5.24261475e-01 -7.61608064e-01 8.12457204e-01 2.36172631e-01
-4.20929730e-01 -8.70691180e-01 -1.96569055e-01 8.87448728e-01
7.58864939e-01 3.81189972e-01 1.41826332e+00 -6.43577814e-01
-8.66771102e-01 2.93991745e-01 -9.71607268e-01 1.44838929e-01
-4.79319543e-01 -9.47836563e-02 -1.19906211e+00 1.36223644e-01
-1.09560333e-01 -8.40115964e-01 1.33724320e+00 -1.44647554e-01
1.26901686e+00 -3.30850750e-01 -9.33074430e-02 3.94586056e-01
1.37162817e+00 1.34712756e-01 8.23289514e-01 4.46436048e-01
1.02217162e+00 9.38864231e-01 1.84545517e-01 -1.29344106e-01
8.53406370e-01 5.21124005e-01 7.95623958e-01 -3.63852680e-01
-7.59541214e-01 -2.63862401e-01 9.32956487e-02 1.63783059e-01
-1.72971383e-01 -1.71720609e-01 -1.20218444e+00 6.08935237e-01
-2.06998467e+00 -1.21335590e+00 -2.56706148e-01 1.67749488e+00
8.90106201e-01 6.66066334e-02 -6.77323192e-02 1.55136988e-01
3.03582132e-01 1.56162262e-01 -6.70423746e-01 -9.21850026e-01
-1.79757819e-01 3.84322733e-01 1.02596655e-02 5.88306725e-01
-8.33856463e-01 1.07650137e+00 5.82593775e+00 5.37236452e-01
-7.92961121e-01 7.85798877e-02 5.56555331e-01 -1.04298465e-01
-6.11328363e-01 -7.69031495e-02 -4.97351378e-01 -7.70168304e-02
7.93419421e-01 2.54039407e-01 5.86489022e-01 6.92345977e-01
-3.53235781e-01 -2.85896838e-01 -1.39227653e+00 1.04287457e+00
5.82172573e-01 -1.59863043e+00 4.10815299e-01 -4.72322047e-01
5.07807493e-01 -4.21948254e-01 2.33278915e-01 6.97036743e-01
3.86524141e-01 -1.61181164e+00 9.90796745e-01 6.65437877e-01
5.44658065e-01 -6.18255138e-01 6.10737443e-01 1.85577586e-01
-8.47255409e-01 -2.04759330e-01 -1.79646477e-01 -2.79784173e-01
-6.81525767e-02 1.45322785e-01 -7.80077219e-01 3.02827686e-01
7.72318840e-01 3.91943276e-01 -1.12451148e+00 7.75479436e-01
-8.21014822e-01 3.25423092e-01 2.26264611e-01 -4.59060669e-01
2.47869119e-01 2.67477512e-01 1.84528410e-01 1.07389617e+00
-1.08085826e-01 3.50137390e-02 -3.67445350e-01 1.30600071e+00
-3.80812377e-01 -1.00371324e-01 -5.14736772e-01 -1.14863500e-01
1.17899068e-01 1.16342986e+00 -5.26795626e-01 -6.85893953e-01
-5.02714336e-01 1.05887866e+00 1.00194860e+00 4.65590656e-01
-1.12591755e+00 -3.12892199e-01 1.75223008e-01 -8.92658532e-02
4.33923244e-01 -8.66049994e-03 -5.42813599e-01 -1.10765135e+00
1.14953056e-01 -1.08285868e+00 6.09925568e-01 -1.34061754e+00
-1.28289270e+00 5.54849684e-01 -1.96351647e-01 -4.06964958e-01
-3.18346769e-01 -1.10239160e+00 -5.71660578e-01 8.99483383e-01
-1.68333483e+00 -1.38763893e+00 -5.61707318e-01 5.85935175e-01
5.92868149e-01 2.08573908e-01 7.87671864e-01 -3.85681540e-02
-3.39373499e-01 4.26218063e-01 -6.11083567e-01 2.71622151e-01
5.82998693e-01 -1.67337883e+00 3.71950477e-01 4.83252883e-01
3.57656121e-01 6.16646409e-01 7.74689972e-01 -3.13590586e-01
-1.48305082e+00 -5.55735350e-01 9.84034002e-01 -1.05978405e+00
6.57073855e-01 -6.05476618e-01 -1.14849937e+00 8.42865109e-01
8.52507293e-01 -8.55544508e-02 4.43888128e-01 2.08274052e-01
-9.53282773e-01 6.57047406e-02 -1.13111877e+00 1.04741406e+00
1.05256641e+00 -8.17384362e-01 -1.17291343e+00 2.52680749e-01
9.49297249e-01 -4.20922220e-01 -3.55005562e-01 1.27562195e-01
4.95211691e-01 -1.15435016e+00 1.17568040e+00 -8.84464920e-01
8.90652180e-01 -2.28000835e-01 -5.63366786e-02 -8.18674743e-01
-1.15117155e-01 -2.98930287e-01 -3.98158610e-01 1.26013291e+00
6.01650715e-01 -3.78847212e-01 6.46901131e-01 9.22349870e-01
6.13256320e-02 -7.92877257e-01 -6.85710430e-01 -4.80423957e-01
3.09364378e-01 -2.29180396e-01 4.05738056e-01 6.69562757e-01
3.90026532e-02 9.43777800e-01 -2.44404361e-01 -2.40754142e-01
4.05933827e-01 2.77584612e-01 8.09770882e-01 -1.11686373e+00
-3.42735380e-01 -5.50551414e-01 8.72443765e-02 -8.46055627e-01
2.75547087e-01 -1.00339019e+00 -2.52155811e-01 -2.40491128e+00
4.99577731e-01 2.15342119e-01 -7.01733157e-02 5.17406166e-01
-3.65964353e-01 2.45878607e-01 6.14269793e-01 -8.58432129e-02
-8.47671509e-01 3.34916592e-01 1.50064290e+00 -4.14549172e-01
2.75548428e-01 -6.45319939e-01 -8.02755594e-01 8.73480439e-01
7.26937771e-01 -1.29957616e-01 -3.99803460e-01 -7.84229457e-01
7.09523082e-01 2.86335461e-02 1.02832305e+00 -7.89166570e-01
2.71942198e-01 9.25277621e-02 8.03599238e-01 -5.80640376e-01
6.47702992e-01 -5.95444679e-01 -3.30456764e-01 4.93723810e-01
-7.05907762e-01 3.03425252e-01 4.06404793e-01 4.16911215e-01
-1.19048044e-01 -2.76600718e-01 5.88996589e-01 -4.90597546e-01
-8.18106949e-01 -3.07388246e-01 -2.37243269e-02 3.47554475e-01
8.71201038e-01 -8.95231888e-02 -1.01661694e+00 -6.09850347e-01
-6.43111348e-01 3.51990849e-01 1.58196703e-01 2.12066785e-01
8.08859706e-01 -1.15707266e+00 -7.35576570e-01 -4.77182060e-01
2.01556802e-01 -2.43031561e-01 4.57957566e-01 6.15866840e-01
-6.34046793e-01 5.81417382e-01 -2.90142506e-01 -4.14028138e-01
-1.23961234e+00 9.83864725e-01 1.83439970e-01 -3.10359448e-01
-6.36585057e-01 1.12078154e+00 1.54952362e-01 -3.08016747e-01
2.83661664e-01 -4.56209928e-01 -5.64768374e-01 1.78802237e-01
5.41430593e-01 3.21367800e-01 -2.17204496e-01 -2.92665243e-01
-4.77603108e-01 5.56681275e-01 3.67488861e-02 -1.43684328e-01
1.14543879e+00 8.66730288e-02 -1.12978280e-01 3.70020539e-01
7.85740972e-01 -9.86515284e-02 -1.15640593e+00 -6.73588074e-05
1.02313034e-01 -3.79758179e-01 -3.22437376e-01 -1.37178278e+00
-7.82255292e-01 1.59650612e+00 2.43819550e-01 3.35726142e-01
8.20581853e-01 3.26856703e-01 3.42703104e-01 5.55199265e-01
-6.00033924e-02 -6.71316087e-01 6.32311523e-01 5.26583672e-01
1.50173652e+00 -1.49109852e+00 3.58477384e-02 -1.31422684e-01
-8.07444155e-01 1.03435683e+00 9.13627803e-01 -1.70281366e-01
-1.94129914e-01 -6.62140623e-02 9.11240578e-02 -4.92112905e-01
-1.00376523e+00 -4.25319940e-01 5.51769733e-01 5.33931375e-01
6.30865335e-01 -2.21304476e-01 -5.83721660e-02 3.25334549e-01
-1.06171988e-01 -2.10087717e-01 2.72536457e-01 5.82840204e-01
-3.72158945e-01 -7.93058872e-01 -2.81109869e-01 9.92240235e-02
-4.74885613e-01 -4.09193724e-01 -8.29674363e-01 1.15922964e+00
-1.86032727e-01 9.46571589e-01 4.97622229e-03 6.07084334e-02
4.62371528e-01 3.58836859e-01 7.39525855e-01 -4.65600431e-01
-8.13510418e-01 -7.34350622e-01 4.53468263e-01 -7.22965956e-01
-4.68351990e-01 -1.36608705e-01 -1.44576359e+00 -3.79116952e-01
6.60140663e-02 -3.39167058e-01 3.71265411e-01 9.94736671e-01
2.15340614e-01 8.85284185e-01 -3.96292001e-01 -4.48204458e-01
-6.14765286e-01 -8.29139471e-01 1.39972001e-01 5.77284992e-01
4.20474112e-01 -5.55959523e-01 -1.88852265e-01 -9.77908894e-02] | [10.842796325683594, 1.921000599861145] |
b310aa06-709e-454a-bb3b-33e5e9287f6c | hide-and-seek-forcing-a-network-to-be | 1704.04232 | null | http://arxiv.org/abs/1704.04232v2 | http://arxiv.org/pdf/1704.04232v2.pdf | Hide-and-Seek: Forcing a Network to be Meticulous for Weakly-supervised Object and Action Localization | We propose `Hide-and-Seek', a weakly-supervised framework that aims to
improve object localization in images and action localization in videos. Most
existing weakly-supervised methods localize only the most discriminative parts
of an object rather than all relevant parts, which leads to suboptimal
performance. Our key idea is to hide patches in a training image randomly,
forcing the network to seek other relevant parts when the most discriminative
part is hidden. Our approach only needs to modify the input image and can work
with any network designed for object localization. During testing, we do not
need to hide any patches. Our Hide-and-Seek approach obtains superior
performance compared to previous methods for weakly-supervised object
localization on the ILSVRC dataset. We also demonstrate that our framework can
be easily extended to weakly-supervised action localization. | ['Yong Jae Lee', 'Krishna Kumar Singh'] | 2017-04-13 | hide-and-seek-forcing-a-network-to-be-1 | http://openaccess.thecvf.com/content_iccv_2017/html/Singh_Hide-And-Seek_Forcing_a_ICCV_2017_paper.html | http://openaccess.thecvf.com/content_ICCV_2017/papers/Singh_Hide-And-Seek_Forcing_a_ICCV_2017_paper.pdf | iccv-2017-10 | ['weakly-supervised-action-localization'] | ['computer-vision'] | [ 1.80253953e-01 1.62928522e-01 -6.74088478e-01 -2.14916304e-01
-9.32208300e-01 -7.62497127e-01 1.24204099e-01 -1.22549385e-01
-6.08232558e-01 5.65251350e-01 2.31432065e-01 6.49546310e-02
1.28106505e-01 -2.89627373e-01 -1.09153426e+00 -7.89207876e-01
-3.07057232e-01 3.59600902e-01 7.89468884e-01 1.57773092e-01
-1.09889232e-01 4.79070783e-01 -1.25445592e+00 5.50613105e-01
4.01296228e-01 6.49794877e-01 3.42863172e-01 5.49461663e-01
3.88512403e-01 1.10514271e+00 -3.69382054e-01 2.61365306e-02
2.80367047e-01 -5.14488995e-01 -1.12241566e+00 4.88539517e-01
7.90398538e-01 -4.18094575e-01 -2.30585247e-01 1.09263730e+00
2.84245312e-01 4.78744768e-02 4.69226539e-01 -1.05466723e+00
-5.87418616e-01 5.58974266e-01 -6.46478117e-01 2.94140995e-01
3.25321823e-01 1.34036481e-01 9.28461909e-01 -1.02177429e+00
6.83131218e-01 1.07414317e+00 5.30173957e-01 6.99842513e-01
-1.44182587e+00 -2.15184480e-01 6.39853656e-01 1.84808224e-01
-1.62181985e+00 -5.55136859e-01 8.04183960e-01 -2.50163615e-01
7.20778525e-01 1.20520540e-01 3.70434910e-01 8.50442708e-01
-1.71378314e-01 1.40568161e+00 9.81503725e-01 -4.20860678e-01
2.19793186e-01 2.91403472e-01 -1.10328525e-01 1.14662230e+00
-5.55730127e-02 -1.21774815e-01 -4.56336170e-01 -9.18032378e-02
8.56312394e-01 9.40109715e-02 -2.19126463e-01 -8.82405281e-01
-1.49424458e+00 5.76211095e-01 5.92045069e-01 4.25806850e-01
-2.94100374e-01 5.23198485e-01 2.00533390e-01 1.36029020e-01
4.24804598e-01 4.54018772e-01 -7.53277481e-01 -6.17643483e-02
-1.04581571e+00 5.27996244e-03 4.48973298e-01 9.99415874e-01
9.07967508e-01 -2.29351997e-01 -2.74118423e-01 6.00888491e-01
7.28748217e-02 2.27358252e-01 1.43737063e-01 -1.21814919e+00
4.31385636e-01 6.19336009e-01 2.32143164e-01 -7.35314846e-01
-1.67490989e-01 -3.69671971e-01 -3.83199960e-01 2.49714389e-01
4.45460647e-01 -6.22900091e-02 -9.25866425e-01 1.64886105e+00
4.45135325e-01 7.39871562e-02 3.81101258e-02 9.89525378e-01
6.33964479e-01 2.86566377e-01 -1.78215932e-02 1.20098375e-01
9.31283474e-01 -1.64232278e+00 -4.23369139e-01 -4.67645317e-01
8.60317111e-01 -6.12180114e-01 1.21767902e+00 1.85837418e-01
-1.11245215e+00 -6.94926739e-01 -6.51998520e-01 2.81609558e-02
-2.70124406e-01 6.77986026e-01 6.65286541e-01 1.61514372e-01
-1.09811556e+00 4.96541500e-01 -7.94656396e-01 -3.33695829e-01
7.33589888e-01 3.42057496e-01 -7.65423536e-01 -1.71867102e-01
-5.04528463e-01 6.41016841e-01 6.15680933e-01 9.98873338e-02
-1.29876471e+00 -2.09089175e-01 -9.51086879e-01 -1.17459416e-01
9.18844581e-01 -2.82453209e-01 1.14740229e+00 -1.48307550e+00
-1.14106965e+00 7.81607211e-01 -4.00131404e-01 -3.91571522e-01
4.50086951e-01 -3.04735094e-01 -2.62219198e-02 5.60923219e-01
3.27329993e-01 1.13376307e+00 8.91747296e-01 -1.35132802e+00
-7.17942953e-01 -9.12956968e-02 3.69382739e-01 1.26774907e-01
-3.01631719e-01 4.89757024e-02 -1.01962614e+00 -7.53462553e-01
2.26559952e-01 -1.09554160e+00 -3.24848622e-01 3.53418380e-01
-3.90670776e-01 -3.22241157e-01 1.08491731e+00 -4.09068197e-01
7.63082743e-01 -2.31415009e+00 1.57158226e-01 -7.81102553e-02
1.40704051e-01 2.01828524e-01 -6.01551890e-01 1.45244554e-01
-1.95738420e-01 -1.76145043e-02 1.04799144e-01 -5.68415523e-01
-1.12835936e-01 4.14296389e-01 3.70857865e-02 7.36657619e-01
3.85188192e-01 1.29787362e+00 -1.07824659e+00 -8.03839922e-01
2.77644336e-01 2.06600398e-01 -7.85597086e-01 3.00854929e-02
-3.38806719e-01 4.37021822e-01 -4.76387948e-01 1.14701939e+00
4.51328665e-01 -5.13864696e-01 -2.79386584e-02 -1.67504549e-01
9.25515741e-02 4.31455970e-02 -1.14385509e+00 1.82247376e+00
-1.76225290e-01 5.84828496e-01 1.16882198e-01 -1.19886971e+00
3.80852669e-01 9.70965624e-02 6.12777054e-01 -4.77366865e-01
-5.60406595e-02 5.66876726e-03 -1.40062019e-01 -6.35210454e-01
9.15463418e-02 9.72196311e-02 -6.01400323e-02 4.72453952e-01
1.90037966e-01 3.46369892e-01 2.26011619e-01 2.56718099e-01
1.20016325e+00 5.29433668e-01 1.07522100e-01 -1.14724472e-01
5.46898782e-01 8.37297887e-02 5.12904465e-01 8.70152652e-01
-4.91842270e-01 6.19487047e-01 4.79895800e-01 -3.55115354e-01
-6.31981492e-01 -9.76621449e-01 1.50956288e-01 1.53955793e+00
3.25726420e-01 -3.17308664e-01 -7.02526987e-01 -1.63991916e+00
-3.76560539e-03 1.46583587e-01 -8.76074314e-01 -1.62459016e-01
-4.34420258e-01 7.18610808e-02 3.50933284e-01 8.64003360e-01
4.23291057e-01 -1.24651217e+00 -3.70565444e-01 -6.02789111e-02
-2.67197818e-01 -1.19674969e+00 -6.89484954e-01 6.07044101e-01
-9.38187003e-01 -1.16616225e+00 -7.51699984e-01 -1.18125665e+00
1.38927960e+00 5.44351399e-01 8.74009609e-01 1.94060266e-01
8.45542701e-04 5.51403701e-01 -5.39103150e-01 1.93816245e-01
-8.95400345e-02 6.24930263e-02 8.47057924e-02 1.31809369e-01
3.86454403e-01 -1.60697922e-01 -4.86716896e-01 6.23887897e-01
-8.23215485e-01 -2.67731100e-01 8.72687399e-01 7.84396946e-01
7.58937180e-01 3.00385267e-01 9.16492343e-02 -7.66929090e-01
-1.17965698e-01 -9.83058214e-02 -2.98448265e-01 3.14668477e-01
-2.93226130e-02 5.33380844e-02 5.62516868e-01 -7.80858397e-01
-5.33007264e-01 7.55704522e-01 1.86190233e-01 -6.72484875e-01
-5.73380411e-01 3.03278416e-01 -3.05136383e-01 -4.68752801e-01
4.78654921e-01 2.74130434e-01 -3.06941718e-01 -5.41841328e-01
3.09751719e-01 3.43952268e-01 3.73776495e-01 -5.24398327e-01
9.82724607e-01 7.38904774e-01 -2.66775727e-01 -5.83154619e-01
-1.17353392e+00 -7.75046527e-01 -1.23841417e+00 -2.75006622e-01
7.32419670e-01 -1.12727475e+00 -5.98864138e-01 1.01783350e-02
-9.88394916e-01 -6.80181324e-01 -6.05511606e-01 5.76790810e-01
-5.16321361e-01 4.18448359e-01 -3.24276388e-01 -6.14937544e-01
3.28818589e-01 -1.12420571e+00 1.43002141e+00 -1.13211535e-01
-1.91842407e-01 -9.59740222e-01 -1.15134887e-01 4.53502297e-01
2.30797529e-02 1.03704976e-02 4.09003645e-01 -5.67933083e-01
-7.66998410e-01 -4.46943492e-01 -1.79660335e-01 5.81373513e-01
3.49311501e-01 -4.23079699e-01 -8.07728052e-01 -4.87080306e-01
-3.97352159e-01 -9.11489308e-01 1.19487810e+00 2.42166057e-01
1.24670470e+00 -4.79504555e-01 -5.36963165e-01 5.22998035e-01
1.10429108e+00 -9.82957259e-02 4.97451365e-01 3.26257259e-01
9.14936066e-01 5.45692563e-01 8.43407929e-01 -4.43956405e-02
9.59038809e-02 6.34979546e-01 3.90360922e-01 -5.69401681e-01
-1.10221393e-01 -4.39239681e-01 6.97947979e-01 1.86275572e-01
-6.13059774e-02 -5.67506775e-02 -4.94690508e-01 7.65280664e-01
-1.92754996e+00 -8.82107496e-01 2.29238987e-01 1.89705300e+00
8.57224166e-01 1.73680291e-01 3.32106292e-01 6.56596734e-04
5.11352122e-01 2.95665026e-01 -4.31255490e-01 1.41664207e-01
4.31640306e-03 7.28430226e-02 6.08867407e-01 4.40017790e-01
-1.49455166e+00 1.38896990e+00 6.68109369e+00 7.51842201e-01
-9.29350495e-01 2.96488523e-01 4.79911566e-01 -3.59061807e-01
4.20727804e-02 2.19082996e-01 -8.63564134e-01 2.23451108e-01
2.66849488e-01 6.43619657e-01 1.56332731e-01 1.28081846e+00
3.00632119e-01 -4.51013565e-01 -1.46836662e+00 6.99923217e-01
3.85995865e-01 -9.73616362e-01 -8.86201560e-02 -2.56007840e-03
9.87674057e-01 -1.65714622e-01 4.45656627e-02 2.52105385e-01
1.81299731e-01 -9.23126757e-01 7.86976635e-01 2.73282707e-01
4.76792127e-01 -6.92125738e-01 7.28153646e-01 5.61864257e-01
-1.15765142e+00 -2.00172961e-02 -4.49972332e-01 -1.79112516e-02
-4.83716503e-02 3.30337375e-01 -7.63103724e-01 -8.65978748e-02
8.67393553e-01 8.27170730e-01 -8.80703628e-01 8.84979606e-01
-5.28540373e-01 6.18968546e-01 -2.12044790e-01 -3.92491184e-03
4.82176602e-01 1.84365019e-01 3.88064921e-01 1.17848754e+00
-2.00585529e-01 -1.05946988e-01 8.83153617e-01 7.76051223e-01
-3.51593830e-03 6.38697743e-02 -5.86004138e-01 -1.44308612e-01
-1.89807601e-02 1.00106180e+00 -9.76171792e-01 -4.57312167e-01
-4.15400326e-01 1.60687447e+00 5.85018814e-01 6.36972785e-01
-9.23567116e-01 -1.61290720e-01 1.63987637e-01 2.44171664e-01
9.91401076e-01 -3.26051146e-01 2.19064251e-01 -1.15471458e+00
2.25985631e-01 -9.71260130e-01 3.13558072e-01 -8.02631974e-01
-8.73735964e-01 2.98059523e-01 -1.01114355e-01 -1.35325241e+00
1.02008097e-01 -5.71254492e-01 -2.00413197e-01 4.18275237e-01
-1.40773690e+00 -1.51995718e+00 -1.75428420e-01 8.75646770e-01
6.92274153e-01 5.13409413e-02 6.09433115e-01 1.82121441e-01
-3.81218135e-01 4.99163896e-01 -1.03920452e-01 3.91592383e-01
7.87159741e-01 -1.23241103e+00 -1.06171392e-01 9.11013484e-01
6.06949449e-01 6.97378635e-01 4.59630966e-01 -6.22239828e-01
-1.12325752e+00 -1.24864542e+00 8.25329542e-01 -5.82793891e-01
6.00273490e-01 -5.77158809e-01 -7.23564684e-01 9.54263151e-01
5.53148650e-02 6.81926608e-01 3.71362329e-01 4.43310402e-02
-3.51973683e-01 -1.44578233e-01 -8.45479310e-01 6.35998189e-01
1.16055632e+00 -7.51623869e-01 -5.16009867e-01 7.08827317e-01
5.61685205e-01 -2.59762734e-01 -3.85677189e-01 4.30037051e-01
3.90503973e-01 -8.61626744e-01 9.81858313e-01 -7.74189889e-01
1.79284856e-01 -6.71062350e-01 -1.00729361e-01 -9.15710270e-01
-4.71731544e-01 -5.52769363e-01 -3.08919936e-01 9.91092920e-01
6.28829598e-01 -2.06430912e-01 1.24164140e+00 1.90288246e-01
1.02557518e-01 -5.64390242e-01 -8.07544708e-01 -9.61193442e-01
-3.08560669e-01 -3.13738972e-01 -2.29449719e-02 8.95528972e-01
-5.05728386e-02 7.28888214e-02 -5.12308478e-01 4.59298104e-01
5.65027416e-01 1.35726884e-01 9.74200368e-01 -6.79882705e-01
-6.43369734e-01 -1.52780926e-02 -4.01999176e-01 -1.58490324e+00
2.93354630e-01 -7.00497150e-01 5.36370516e-01 -1.41566539e+00
5.00331283e-01 -3.35422873e-01 -4.80658114e-01 1.12766683e+00
-1.48953348e-01 9.04554844e-01 6.58199713e-02 2.07447544e-01
-1.35184562e+00 3.81981403e-01 1.37355876e+00 -3.51529598e-01
-2.22548619e-01 4.91400063e-02 -5.58211863e-01 9.21461999e-01
6.50192261e-01 -7.17885435e-01 -2.95251399e-01 -3.65726769e-01
-1.59112006e-01 -3.34171355e-01 8.55649590e-01 -9.58912790e-01
1.77316591e-01 -1.67103007e-01 7.73165882e-01 -7.64143407e-01
2.08098009e-01 -1.08647203e+00 -3.64438266e-01 4.47192341e-01
-5.00362992e-01 -2.38644794e-01 1.36200652e-01 5.63108742e-01
-2.45539233e-01 -3.41053486e-01 8.25852454e-01 -2.98692346e-01
-9.66227710e-01 2.00288698e-01 -4.27190661e-01 -2.12625518e-01
1.36390412e+00 -2.38714650e-01 -2.50869151e-02 -3.03441852e-01
-1.17410064e+00 1.28770307e-01 5.80489337e-01 3.96695375e-01
6.85418189e-01 -1.40289879e+00 -2.98546493e-01 1.41426161e-01
3.29339236e-01 -5.11414371e-03 -4.12640302e-03 1.04804480e+00
-2.71216869e-01 4.82784957e-01 2.26326123e-01 -8.56159806e-01
-1.58037484e+00 9.19336498e-01 3.18725973e-01 -3.88902053e-02
-6.09521985e-01 1.08832514e+00 3.76857907e-01 -2.84177601e-01
6.73333108e-01 -4.28594947e-01 -9.81829781e-03 -1.46066099e-01
4.68879461e-01 9.26090032e-02 -3.94072324e-01 -8.59164774e-01
-7.00422227e-01 6.77336752e-01 -2.14722529e-01 8.71606730e-03
1.26126587e+00 -6.28816560e-02 -1.12214677e-01 3.82657588e-01
1.41640091e+00 2.49872133e-01 -1.54625440e+00 -4.41699177e-01
-8.81243795e-02 -5.81564486e-01 1.11211106e-01 -7.26900339e-01
-1.23181605e+00 6.04627430e-01 5.56285858e-01 -1.23518765e-01
1.00255096e+00 6.28438771e-01 3.34690303e-01 7.12906182e-01
5.46829939e-01 -1.18929195e+00 7.34457672e-01 3.25580120e-01
7.03956485e-01 -1.46359432e+00 1.55644014e-01 -3.09953898e-01
-8.18120837e-01 8.13804567e-01 8.98210764e-01 -1.52729854e-01
4.74303156e-01 2.56450206e-01 5.93058765e-02 -3.29683155e-01
-5.00667572e-01 -6.81742132e-01 6.57388449e-01 6.14099324e-01
2.22513676e-01 -3.53509694e-01 6.73696622e-02 1.62130877e-01
5.57169974e-01 4.82860804e-02 4.00821529e-02 1.37671793e+00
-5.67782998e-01 -9.87554073e-01 -3.18584859e-01 -3.94369941e-03
-3.73343587e-01 -4.91356179e-02 -8.78826559e-01 8.89000356e-01
6.70314789e-01 8.02256286e-01 -8.61746296e-02 -2.68521965e-01
3.24881285e-01 -1.11901246e-01 5.20728946e-01 -8.84659588e-01
-1.71892300e-01 5.65946341e-01 -3.94390933e-02 -9.97229815e-01
-9.08035755e-01 -6.87908053e-01 -1.20421076e+00 2.06175745e-01
-5.85609674e-01 2.08428606e-01 1.70530885e-01 9.21680689e-01
2.44308382e-01 4.01561141e-01 6.08220935e-01 -8.96745861e-01
-1.40376464e-01 -8.77806664e-01 -5.54055154e-01 2.13931262e-01
6.59934759e-01 -6.29132509e-01 -3.88438076e-01 4.37890589e-01] | [8.502265930175781, 0.5204090476036072] |
323c4b98-f8fe-4a3a-b1b4-91239cb281cb | self-supervised-character-to-character | 2211.00288 | null | https://arxiv.org/abs/2211.00288v3 | https://arxiv.org/pdf/2211.00288v3.pdf | Self-supervised Character-to-Character Distillation for Text Recognition | When handling complicated text images (e.g., irregular structures, low resolution, heavy occlusion, and uneven illumination), existing supervised text recognition methods are data-hungry. Although these methods employ large-scale synthetic text images to reduce the dependence on annotated real images, the domain gap still limits the recognition performance. Therefore, exploring the robust text feature representations on unlabeled real images by self-supervised learning is a good solution. However, existing self-supervised text recognition methods conduct sequence-to-sequence representation learning by roughly splitting the visual features along the horizontal axis, which limits the flexibility of the augmentations, as large geometric-based augmentations may lead to sequence-to-sequence feature inconsistency. Motivated by this, we propose a novel self-supervised Character-to-Character Distillation method, CCD, which enables versatile augmentations to facilitate general text representation learning. Specifically, we delineate the character structures of unlabeled real images by designing a self-supervised character segmentation module. Following this, CCD easily enriches the diversity of local characters while keeping their pairwise alignment under flexible augmentations, using the transformation matrix between two augmented views from images. Experiments demonstrate that CCD achieves state-of-the-art results, with average performance gains of 1.38% in text recognition, 1.7% in text segmentation, 0.24 dB (PSNR) and 0.0321 (SSIM) in text super-resolution. Code will be released soon. | ['Zekun Jiang', 'Qi Feng', 'Xue Yang', 'Wei Shen', 'Tongkun Guan'] | 2022-11-01 | null | null | null | null | ['self-learning'] | ['natural-language-processing'] | [ 7.98504353e-01 -2.94555038e-01 -2.52287418e-01 -4.83055592e-01
-6.90825045e-01 -6.57754183e-01 3.64823371e-01 -3.84409577e-01
-2.28155524e-01 5.75267494e-01 4.63857055e-02 -2.80357040e-02
4.20685738e-01 -5.12027204e-01 -5.88446558e-01 -8.60854268e-01
7.63231516e-01 3.32151771e-01 3.15568238e-01 -5.29488400e-02
5.28190851e-01 3.47990245e-01 -1.39492083e+00 4.53555495e-01
1.37695396e+00 7.15744853e-01 6.27030194e-01 5.77015698e-01
-7.39044189e-01 4.15963173e-01 -5.96642792e-01 -3.30216646e-01
6.81545734e-02 -5.43271363e-01 -5.49447894e-01 9.57433462e-01
5.42865634e-01 -4.66741025e-01 -3.86618525e-01 1.18535054e+00
3.43660027e-01 -1.12179331e-01 6.89177096e-01 -8.69642377e-01
-8.61194551e-01 4.80182201e-01 -9.67837572e-01 -1.22125506e-01
3.48959863e-01 2.30803668e-01 9.55707014e-01 -1.22060537e+00
6.33630872e-01 1.06111848e+00 4.31900382e-01 5.03659725e-01
-1.36955023e+00 -5.94211638e-01 2.92592615e-01 7.61583969e-02
-1.37154448e+00 -4.27647918e-01 7.58530676e-01 -3.56369793e-01
6.43977284e-01 3.55445266e-01 3.96139324e-01 1.17346442e+00
-2.52052963e-01 1.25649393e+00 1.03170240e+00 -5.22960722e-01
1.94451064e-02 1.61111370e-01 -3.92639413e-02 7.75097668e-01
2.34101266e-01 -4.00666505e-01 -5.29505134e-01 1.67233199e-01
1.02123427e+00 1.43932663e-02 -2.32234657e-01 -5.12459636e-01
-1.36146653e+00 6.23164177e-01 3.53430696e-02 2.07269579e-01
1.83511347e-01 -4.12808061e-01 3.56227458e-01 9.54364985e-02
2.50226200e-01 2.50486851e-01 -2.63762295e-01 -1.54767931e-01
-1.08137023e+00 -2.25353181e-01 4.99755025e-01 1.26885986e+00
7.76154578e-01 5.22982001e-01 8.96548480e-02 1.25728571e+00
2.03826144e-01 7.36058354e-01 7.12113440e-01 -5.48338830e-01
7.47564077e-01 7.99342990e-01 -2.39182159e-01 -8.77789795e-01
-1.03364050e-01 -1.34920806e-01 -1.16950679e+00 -1.17036209e-01
3.56422603e-01 -5.68068237e-04 -1.06850088e+00 1.20942283e+00
4.96033244e-02 -2.12683842e-01 8.32199380e-02 8.83248210e-01
6.12902403e-01 7.45885968e-01 -3.68266851e-01 -4.52340335e-01
1.15604854e+00 -1.24787796e+00 -8.68639648e-01 -3.43922496e-01
5.84178984e-01 -1.08481801e+00 1.35878217e+00 3.06814283e-01
-8.86542857e-01 -6.14018261e-01 -1.13149631e+00 2.90597100e-02
-1.43361285e-01 6.51670098e-01 2.61966497e-01 8.27724576e-01
-6.08208477e-01 2.37380460e-01 -8.52902234e-01 -3.80715668e-01
3.94827902e-01 3.56951684e-01 -2.25197375e-01 -2.70703703e-01
-7.51290619e-01 2.98735678e-01 3.49165142e-01 2.76598074e-02
-3.16084355e-01 -3.44814807e-01 -7.79306829e-01 -2.79454757e-02
5.58872521e-01 -3.14016670e-01 8.36410046e-01 -1.22209811e+00
-1.68430030e+00 8.03575099e-01 -2.85963923e-01 6.79789782e-02
7.20915020e-01 -4.22560237e-02 -5.80492854e-01 2.65261799e-01
1.88819319e-01 5.99846959e-01 1.27885783e+00 -1.32865894e+00
-5.62246442e-01 -4.04843301e-01 -5.99006355e-01 4.80195433e-01
-5.91484964e-01 5.17518036e-02 -9.97997224e-01 -1.12426317e+00
5.97506583e-01 -9.05340493e-01 -1.94592163e-01 2.23988757e-01
-5.81550598e-01 3.13708544e-01 1.24330163e+00 -5.50465524e-01
1.04035795e+00 -2.22576714e+00 2.53313314e-02 6.11421354e-02
-1.64386649e-02 3.98102760e-01 -3.01689625e-01 2.80583024e-01
6.56230375e-02 1.59561798e-01 -5.11605084e-01 -3.74815762e-01
-2.21097037e-01 3.01063448e-01 -2.35585198e-01 3.66481453e-01
2.64454365e-01 9.06190813e-01 -4.55375969e-01 -7.78926253e-01
2.96162099e-01 2.03134879e-01 -3.27810258e-01 1.69692263e-01
-2.62936622e-01 5.00792503e-01 -6.99399292e-01 1.00644946e+00
9.40037906e-01 -4.10994709e-01 3.00396532e-01 -2.67565876e-01
-1.15699068e-01 -1.69096217e-01 -1.38389659e+00 1.67244339e+00
-7.07463175e-02 7.20818818e-01 3.46409716e-02 -1.05971146e+00
1.23378634e+00 -7.48122260e-02 4.36775178e-01 -6.43109560e-01
-5.95983602e-02 1.98050842e-01 -2.61786193e-01 -5.56885719e-01
7.01666474e-01 2.18309537e-01 1.24350362e-01 4.87894297e-01
-1.97931677e-01 -1.42382264e-01 2.26127520e-01 1.19342729e-01
5.41907847e-01 2.11631462e-01 -3.11177857e-02 -1.05186835e-01
5.87198138e-01 1.38501078e-01 5.14873803e-01 4.99646217e-01
3.41264680e-02 1.11059928e+00 3.93862695e-01 -2.82487184e-01
-1.44822049e+00 -7.85712123e-01 -2.78318495e-01 9.73449707e-01
3.60805869e-01 -1.89862460e-01 -8.67884457e-01 -5.59493959e-01
-2.73381650e-01 1.91102907e-01 -2.92742610e-01 2.44098619e-01
-6.73254371e-01 -8.34154189e-01 6.36909604e-01 7.40634680e-01
9.09339905e-01 -7.66487241e-01 -5.84668964e-02 5.77174835e-02
-2.28086933e-01 -1.55359077e+00 -9.33807075e-01 3.61560360e-02
-1.00582957e+00 -7.18295395e-01 -1.04951584e+00 -1.19660890e+00
1.07583046e+00 5.32288432e-01 4.89784628e-01 -5.12887277e-02
-3.94831747e-01 2.58275330e-01 -4.88580227e-01 3.38019311e-01
-3.38133603e-01 4.41276468e-02 2.81629525e-02 1.88684374e-01
5.38000725e-02 -3.13043356e-01 -3.77273977e-01 7.60830820e-01
-1.04452896e+00 3.59949887e-01 7.34689355e-01 1.33134890e+00
8.52509558e-01 2.82252370e-03 2.42921993e-01 -8.91570330e-01
2.83348471e-01 1.90318510e-01 -6.50057971e-01 3.28816235e-01
-6.15226328e-01 3.57072614e-02 8.45782042e-01 -7.53303170e-01
-1.31773293e+00 5.36997974e-01 1.04561999e-01 -3.76203746e-01
-2.45665580e-01 2.95237124e-01 -5.97267151e-01 -1.12002373e-01
4.81995255e-01 9.28084850e-01 2.21922070e-01 -2.74621904e-01
3.42973471e-01 9.80167508e-01 5.58526635e-01 -6.32289886e-01
9.34519708e-01 4.25096840e-01 -3.95550698e-01 -1.21776986e+00
-6.20246589e-01 -5.25967121e-01 -8.78890753e-01 1.54042542e-01
7.71797955e-01 -9.25075710e-01 -3.02309513e-01 8.33442032e-01
-7.48377383e-01 -2.61213511e-01 -7.86866173e-02 3.75082642e-01
-6.08759761e-01 1.13016486e+00 -6.30098820e-01 -6.23374283e-01
-2.59299934e-01 -1.27996171e+00 1.20505905e+00 2.53223240e-01
1.21132262e-01 -7.39470840e-01 -3.86740863e-01 7.62084246e-01
1.93673342e-01 -1.70220047e-01 8.44102561e-01 -5.56615472e-01
-7.40836084e-01 -4.55074757e-02 -6.23931885e-01 4.84982669e-01
3.85544181e-01 1.25917047e-01 -7.09544957e-01 -2.80008078e-01
-2.19928026e-01 -4.59354103e-01 6.94755971e-01 1.39546752e-01
1.31835663e+00 -3.01639944e-01 -2.20355719e-01 8.91630232e-01
1.26028574e+00 2.62306958e-01 7.54831374e-01 3.43293935e-01
1.18076086e+00 5.56855083e-01 6.32558763e-01 4.99096513e-01
5.25231324e-02 5.94946682e-01 5.47513142e-02 -3.05650383e-01
-2.48434749e-02 -2.91851401e-01 3.06918293e-01 8.86462808e-01
2.02123195e-01 -3.95457953e-01 -7.53787100e-01 2.58563340e-01
-1.72226536e+00 -7.16480255e-01 -1.83057338e-01 1.96657181e+00
8.57191682e-01 7.58695677e-02 4.42659780e-02 1.48738101e-01
1.16911578e+00 2.33722419e-01 -9.59349334e-01 -6.04685321e-02
-7.02294707e-01 -2.79169649e-01 5.16663551e-01 1.34676039e-01
-1.04033875e+00 1.29325569e+00 5.85904217e+00 1.17689049e+00
-1.14670706e+00 -4.86785740e-01 8.33254337e-01 3.12198430e-01
-1.49517089e-01 -1.45819098e-01 -9.42502618e-01 5.01973212e-01
3.52223337e-01 1.02484420e-01 3.37439030e-01 7.33004689e-01
1.24761134e-01 7.89480656e-02 -7.87493169e-01 1.17281580e+00
4.05895472e-01 -1.30955112e+00 4.42410588e-01 3.16305459e-02
1.02132487e+00 -3.17609847e-01 1.90264046e-01 -1.39094532e-01
-7.47045353e-02 -9.84291315e-01 3.59112144e-01 9.68503505e-02
1.35694194e+00 -5.82960844e-01 3.73606265e-01 3.70551765e-01
-1.25076127e+00 1.75481483e-01 -4.93152022e-01 4.78309721e-01
4.71492447e-02 3.30125749e-01 -6.55559123e-01 4.28689688e-01
4.52858508e-01 8.67201686e-01 -5.62521100e-01 6.77733243e-01
7.48984292e-02 5.34358442e-01 -2.21226141e-01 -1.99705318e-01
2.15405583e-01 -5.81824660e-01 3.97412568e-01 1.35318255e+00
2.17809692e-01 1.18843645e-01 3.50332916e-01 8.84102702e-01
-3.49630564e-02 3.54628474e-01 -3.93607110e-01 -3.07361901e-01
3.21814030e-01 1.08364880e+00 -9.91801441e-01 -3.58868569e-01
-4.55267131e-01 1.51166284e+00 -8.50083306e-02 5.79341650e-01
-6.43159628e-01 -4.94800121e-01 2.11385101e-01 3.04367058e-02
5.98026276e-01 -2.69555658e-01 -6.25036120e-01 -1.48763394e+00
3.20215970e-01 -1.24536431e+00 1.47451505e-01 -7.57029593e-01
-1.25902104e+00 5.01560450e-01 -4.83774990e-01 -1.46168029e+00
-3.80399935e-02 -6.76382780e-01 -3.12199205e-01 6.95482731e-01
-1.29865360e+00 -1.35545540e+00 -3.52700710e-01 6.29624009e-01
1.17820919e+00 -4.44012851e-01 6.10789299e-01 3.57608162e-02
-7.65284061e-01 1.01840878e+00 6.44365966e-01 3.97699296e-01
8.24968159e-01 -1.00155914e+00 4.91721928e-01 8.23176622e-01
2.07318947e-01 3.98426831e-01 2.44260311e-01 -6.86520636e-01
-1.76524758e+00 -9.50725615e-01 2.84912407e-01 -1.05478190e-01
6.93525076e-01 -5.16354501e-01 -1.28336418e+00 4.03030038e-01
-8.12557936e-02 6.39467984e-02 6.24652445e-01 -3.46200824e-01
-4.53583896e-01 1.54492632e-01 -1.07313561e+00 8.97280395e-01
1.07468987e+00 -5.13504148e-01 -2.54102141e-01 3.07150751e-01
5.31384766e-01 -4.85850811e-01 -7.06060827e-01 3.23924065e-01
6.31135821e-01 -8.77854586e-01 8.21016073e-01 -2.48455659e-01
4.23236251e-01 -4.34763461e-01 -2.45168492e-01 -7.23477542e-01
-2.17519566e-01 -7.37027347e-01 1.26137301e-01 1.34191382e+00
4.49114174e-01 -5.33325911e-01 1.30607748e+00 3.44249636e-01
-4.64702733e-02 -6.20355725e-01 -7.34751999e-01 -7.99947858e-01
-7.00533539e-02 -3.23478356e-02 2.94265747e-01 1.23054016e+00
-1.19585143e-02 4.16202277e-01 -5.00320256e-01 9.56012160e-02
5.99157631e-01 2.83586800e-01 8.09799850e-01 -9.17945266e-01
-1.87627524e-01 -4.85279977e-01 -3.50028187e-01 -1.85789490e+00
7.28342012e-02 -4.89084423e-01 7.51495287e-02 -1.29395032e+00
4.49412197e-01 -3.91436726e-01 2.96600819e-01 2.83988655e-01
-3.14760566e-01 2.26568267e-01 9.83275324e-02 5.31763852e-01
-5.73593497e-01 7.55075872e-01 1.65590060e+00 -4.16996211e-01
-2.37723082e-01 -4.79786322e-02 -5.24731755e-01 7.44019985e-01
8.71687770e-01 -1.56090120e-02 -3.27085644e-01 -6.38588607e-01
-2.39983693e-01 1.13692202e-01 -2.49059081e-01 -6.97970748e-01
3.50965917e-01 -2.85456419e-01 6.67394102e-01 -9.04193521e-01
2.91171640e-01 -6.40715301e-01 -2.74788767e-01 2.03617930e-01
-3.82990539e-01 -1.34241626e-01 1.34050474e-01 7.24964499e-01
-3.71415943e-01 -4.49381620e-01 8.33965480e-01 -3.14367153e-02
-6.92268789e-01 2.73013473e-01 -5.29241145e-01 2.99772751e-02
6.77830637e-01 -9.61043656e-01 -2.38352582e-01 -3.13507229e-01
-4.59999114e-01 1.96306825e-01 7.74063587e-01 2.71102816e-01
8.86167765e-01 -9.79623973e-01 -7.04245150e-01 5.45954406e-01
2.95188844e-01 2.27668226e-01 3.52505893e-01 5.49591184e-01
-5.87951064e-01 3.71274740e-01 -1.72753632e-01 -8.89008939e-01
-1.41906488e+00 2.82618314e-01 6.06690571e-02 -1.50665611e-01
-7.76613355e-01 5.79814315e-01 4.41531420e-01 -4.27358836e-01
2.55651295e-01 -1.23150185e-01 -8.56517106e-02 -5.03703542e-02
4.49692041e-01 1.81178495e-01 -1.21033631e-01 -6.70068979e-01
4.85197082e-02 1.02710748e+00 -5.54562509e-01 -2.68684477e-02
9.83334303e-01 -4.13093776e-01 1.72088981e-01 2.94621676e-01
1.07048881e+00 1.81390554e-01 -1.58446598e+00 -5.25103271e-01
-1.05030112e-01 -6.66880786e-01 -2.58451432e-01 -5.89841545e-01
-1.08345270e+00 8.80505562e-01 5.15784442e-01 -7.26101398e-02
1.20178759e+00 -3.59348863e-01 8.43900025e-01 7.05461264e-01
3.36868167e-02 -1.32487273e+00 5.38850188e-01 6.14547670e-01
6.37066960e-01 -1.40857100e+00 1.81232050e-01 -8.09115529e-01
-1.08752012e+00 1.30143452e+00 7.75631964e-01 1.82214841e-01
3.50517221e-02 4.38788205e-01 1.50349438e-01 2.36129969e-01
-3.55826855e-01 -1.57065064e-01 1.15799390e-01 6.85043871e-01
3.74375761e-01 -1.71228901e-01 -6.93823770e-02 2.78399229e-01
5.05803451e-02 -4.96689528e-01 6.24044299e-01 8.63360047e-01
-4.92718518e-01 -1.21783531e+00 -5.31709552e-01 4.20120269e-01
-3.50020647e-01 -2.62734741e-01 -4.80390549e-01 6.12826288e-01
-3.10478538e-01 6.81899369e-01 1.66141212e-01 -3.01890850e-01
2.53885210e-01 -1.11786082e-01 3.22464108e-01 -6.10200703e-01
-9.57150310e-02 6.86739564e-01 -9.91726965e-02 -9.00868606e-03
-3.27377826e-01 -7.11140931e-01 -1.41308641e+00 -2.70410161e-02
-6.72041059e-01 -1.84086978e-01 6.73208952e-01 7.53674269e-01
2.83133477e-01 3.72239292e-01 9.54335690e-01 -6.36169255e-01
-5.26803911e-01 -9.17877913e-01 -5.87164640e-01 2.66470104e-01
4.03150320e-02 -2.90072620e-01 -3.04326445e-01 6.35398448e-01] | [11.805229187011719, 2.0041048526763916] |
c6e48d52-dc90-4a88-8d99-17ae2110cfc2 | open-surgery-tool-classification-and-hand | 2111.06098 | null | https://arxiv.org/abs/2111.06098v1 | https://arxiv.org/pdf/2111.06098v1.pdf | Open surgery tool classification and hand utilization using a multi-camera system | Purpose: The goal of this work is to use multi-camera video to classify open surgery tools as well as identify which tool is held in each hand. Multi-camera systems help prevent occlusions in open surgery video data. Furthermore, combining multiple views such as a Top-view camera covering the full operative field and a Close-up camera focusing on hand motion and anatomy, may provide a more comprehensive view of the surgical workflow. However, multi-camera data fusion poses a new challenge: a tool may be visible in one camera and not the other. Thus, we defined the global ground truth as the tools being used regardless their visibility. Therefore, tools that are out of the image should be remembered for extensive periods of time while the system responds quickly to changes visible in the video. Methods: Participants (n=48) performed a simulated open bowel repair. A Top-view and a Close-up cameras were used. YOLOv5 was used for tool and hand detection. A high frequency LSTM with a 1 second window at 30 frames per second (fps) and a low frequency LSTM with a 40 second window at 3 fps were used for spatial, temporal, and multi-camera integration. Results: The accuracy and F1 of the six systems were: Top-view (0.88/0.88), Close-up (0.81,0.83), both cameras (0.9/0.9), high fps LSTM (0.92/0.93), low fps LSTM (0.9/0.91), and our final architecture the Multi-camera classifier(0.93/0.94). Conclusion: By combining a system with a high fps and a low fps from the multiple camera array we improved the classification abilities of the global ground truth. | ['Shlomi Laufer', 'Carla M Pugh', 'Adam Goldbraikh', 'Kristina Basiev'] | 2021-11-11 | null | null | null | null | ['hand-detection'] | ['computer-vision'] | [-1.73859578e-02 -2.41288394e-01 -2.55533457e-01 2.24777222e-01
-8.38415146e-01 -7.52811074e-01 -2.88690231e-03 2.45499402e-01
-6.68683589e-01 1.46245345e-01 -1.43050015e-01 -1.42444327e-01
-6.25956878e-02 -4.63463694e-01 -6.68932617e-01 -6.63345098e-01
-5.37517853e-03 -6.48473203e-02 4.70421016e-01 7.35568479e-02
-7.59553015e-02 6.04098260e-01 -1.56927240e+00 6.31876469e-01
2.18428150e-01 1.01563597e+00 7.24871337e-01 1.07599413e+00
2.82280177e-01 8.92076552e-01 -6.03594124e-01 1.22852117e-01
4.15499687e-01 -1.34370536e-01 -4.45483536e-01 -7.96975046e-02
6.27642214e-01 -8.49927604e-01 -3.25956583e-01 8.20601583e-01
6.01332963e-01 1.20285079e-01 -1.73896481e-03 -7.79321492e-01
2.17991471e-01 1.43995592e-02 -6.49655998e-01 4.26898658e-01
6.69561327e-01 1.95516691e-01 1.55085206e-01 -4.68245476e-01
9.23081517e-01 8.13708425e-01 6.41273558e-01 4.14134115e-01
-8.21309507e-01 -7.19475925e-01 -4.48811539e-02 -1.77506730e-02
-1.04710829e+00 -7.18992576e-02 4.58068848e-01 -6.78766370e-01
9.32548106e-01 2.59652495e-01 1.19436038e+00 1.24830973e+00
1.13102067e+00 3.32202226e-01 9.08777714e-01 -5.10748804e-01
-3.16347301e-01 2.85840482e-01 -1.25268355e-01 1.06894016e+00
3.23436022e-01 1.80085808e-01 -4.89341170e-01 -2.13256627e-02
1.37092817e+00 8.02340627e-01 -5.75865388e-01 -3.05472314e-01
-1.63964462e+00 6.12245679e-01 3.38349879e-01 7.43576527e-01
-6.15143836e-01 1.85466651e-02 5.09613454e-01 1.56717077e-01
-9.92641896e-02 5.84578574e-01 -1.01048417e-01 -2.41110653e-01
-7.73711145e-01 -4.48003858e-01 5.91067910e-01 8.56691122e-01
2.95317173e-01 -1.41243294e-01 -1.80521548e-01 5.94304621e-01
-1.15392499e-01 3.45307887e-01 7.90999889e-01 -1.16138530e+00
7.97028616e-02 5.26821971e-01 1.27052590e-02 -1.09910882e+00
-8.39896023e-01 -4.15379435e-01 -5.39494932e-01 4.90603536e-01
4.32394296e-01 -2.63042182e-01 -9.70228136e-01 1.25626671e+00
-2.31202934e-02 -2.15023458e-01 -1.99260607e-01 1.05781293e+00
7.90266275e-01 3.12864006e-01 -6.25047237e-02 -3.49885434e-01
1.83169508e+00 -1.11606896e+00 -9.58606541e-01 -3.58721584e-01
8.87133300e-01 -1.04122996e+00 8.62767875e-01 6.74972951e-01
-1.02210677e+00 -7.72912979e-01 -1.15846944e+00 1.64137855e-01
-1.12696052e-01 7.21995533e-01 5.57279348e-01 4.64824557e-01
-1.02400911e+00 5.73239505e-01 -1.36366045e+00 -4.59820330e-01
-1.68237105e-01 6.62058413e-01 -8.26835036e-01 -1.00329503e-01
-7.23416388e-01 1.19830155e+00 1.10685952e-01 1.86743259e-01
-7.30752528e-01 -4.06898171e-01 -8.74647379e-01 -2.07724333e-01
4.05523807e-01 -7.83930480e-01 1.27758121e+00 -1.25012577e+00
-1.43237472e+00 1.11404645e+00 5.11504747e-02 -1.59411937e-01
7.10959077e-01 -5.48679829e-01 -2.86258191e-01 6.97492719e-01
-4.96317968e-02 3.59415948e-01 5.00379741e-01 -1.03676105e+00
-9.18772519e-01 -5.14413297e-01 3.09046537e-01 3.73993576e-01
-2.54208222e-02 8.55378285e-02 -4.24028367e-01 -2.88887918e-01
2.63023794e-01 -1.20714712e+00 1.04785889e-01 2.94963270e-01
-2.54883375e-02 4.18741137e-01 6.64387524e-01 -8.99592340e-01
1.06359220e+00 -2.10220408e+00 -1.08995952e-01 -3.33789080e-01
3.50572973e-01 2.66722947e-01 3.19797486e-01 1.40762120e-01
-2.66295493e-01 -2.70531535e-01 5.02471089e-01 -2.37005681e-01
-1.11851418e+00 9.93970111e-02 3.44505876e-01 8.43385339e-01
-7.52483726e-01 3.43814999e-01 -9.48781848e-01 -6.19165540e-01
9.51340854e-01 5.50788820e-01 -1.74048394e-01 3.91823649e-01
6.25082374e-01 4.33520913e-01 1.32026985e-01 8.09346855e-01
2.58063763e-01 -7.94655532e-02 7.83680007e-02 -4.76856858e-01
-2.87065148e-01 -4.97778356e-01 -1.23236740e+00 2.25988770e+00
-8.50909472e-01 7.66633570e-01 2.79196739e-01 -3.81272137e-01
5.59167862e-01 7.45160520e-01 7.75743723e-01 -6.06752992e-01
3.52242172e-01 4.00761306e-01 2.43787095e-01 -1.01338863e+00
2.35443935e-01 -1.54525712e-01 2.98110396e-01 -1.21106073e-01
2.57361710e-01 2.76006281e-01 9.45847109e-02 -1.85136069e-02
1.03572071e+00 8.03630501e-02 4.91527051e-01 2.97694225e-02
6.29887506e-02 1.30386636e-01 1.62077129e-01 8.37977409e-01
-3.85272354e-01 8.38819087e-01 4.30490255e-01 -6.71893895e-01
-4.66048211e-01 -8.21584642e-01 2.16471583e-01 7.02522933e-01
3.36944133e-01 -2.27662891e-01 -4.61849511e-01 -5.55466950e-01
-3.69920760e-01 2.46113241e-01 -7.16153443e-01 -4.09098297e-01
-7.58846641e-01 -2.24022884e-02 9.26249325e-02 5.93292356e-01
8.50382000e-02 -7.79864788e-01 -1.51783371e+00 1.15716957e-01
-3.54463309e-01 -1.06306708e+00 -6.14794910e-01 3.04497778e-01
-9.59330916e-01 -1.41721356e+00 -1.11599755e+00 -8.25825989e-01
6.55793488e-01 6.24572992e-01 7.25401342e-01 -2.70892799e-01
-5.80020964e-01 6.48663163e-01 -2.41340682e-01 -1.43800989e-01
-3.48726660e-01 -2.93147892e-01 1.01417921e-01 -1.93603098e-01
-2.16592804e-01 -1.33884966e-01 -7.38364100e-01 2.43372083e-01
-4.80163813e-01 3.73412907e-01 6.78420842e-01 7.44885683e-01
4.79937822e-01 -6.72958374e-01 -3.80655199e-01 -5.34024775e-01
2.28474468e-01 -1.54117510e-01 -5.73935866e-01 1.89440966e-01
-2.19273269e-01 -4.50686008e-01 5.89724183e-01 -8.72357786e-01
-7.26724923e-01 4.37246084e-01 2.82920450e-01 -1.23617387e+00
-1.58952534e-01 1.29385576e-01 3.99519235e-01 -2.42042825e-01
6.16422653e-01 -1.69424430e-01 3.81292284e-01 -1.32619748e-02
-3.33864510e-01 5.17874599e-01 6.74967170e-01 4.53808196e-02
-1.67692423e-01 4.67029274e-01 -1.24500349e-01 -7.75980532e-01
-6.44516647e-01 -8.43594313e-01 -6.74106419e-01 -8.45481813e-01
1.08118272e+00 -1.01168263e+00 -9.02986526e-01 2.52114624e-01
-1.28424716e+00 6.98360652e-02 -4.77587022e-02 1.36110532e+00
-3.74745429e-01 2.05906227e-01 -6.42450452e-01 -7.17099071e-01
-5.24939775e-01 -1.60133934e+00 1.13373446e+00 4.03629720e-01
-4.61269259e-01 -9.04249609e-01 -1.97769031e-02 2.44202152e-01
2.29674354e-01 8.02948475e-01 2.29813039e-01 -3.99284810e-01
-2.23247096e-01 -7.97591150e-01 9.84704196e-02 2.83461303e-01
5.70165873e-01 1.09410278e-01 -9.71071541e-01 -3.36952776e-01
4.21097904e-01 2.56146461e-01 5.54096997e-01 1.03080714e+00
8.73856246e-01 1.12056732e-01 -6.88822269e-01 7.65064418e-01
1.45778775e+00 8.27643216e-01 2.87695825e-01 4.21603292e-01
6.25374854e-01 2.96717107e-01 6.50732279e-01 5.30629940e-02
-4.13464978e-02 6.82908893e-01 4.50450718e-01 -4.51260895e-01
-1.12777568e-01 5.05364425e-02 4.75826025e-01 5.22640824e-01
-3.29400092e-01 2.95453686e-02 -1.00834703e+00 4.13198173e-01
-1.46506786e+00 -6.78117394e-01 -2.46738493e-01 2.54306936e+00
2.11883754e-01 3.87214832e-02 8.03860277e-02 2.55451892e-02
8.96090448e-01 -6.37851283e-02 -2.74724603e-01 -3.80236238e-01
2.27746412e-01 -1.78892851e-01 8.60364497e-01 2.77242243e-01
-1.19982028e+00 2.80800134e-01 5.82847023e+00 3.49567443e-01
-1.60061860e+00 1.68457031e-01 2.08432928e-01 -7.35713184e-01
6.23254240e-01 -3.38977426e-01 -4.45707500e-01 6.83794081e-01
8.43779266e-01 1.64306119e-01 -7.52021745e-02 9.76992965e-01
1.96913317e-01 -8.31467986e-01 -1.17486560e+00 1.37106860e+00
4.43428218e-01 -1.14374709e+00 -2.23162830e-01 1.06048606e-01
1.84741765e-01 -5.06616291e-03 -3.05766314e-01 -5.73314447e-03
-3.80276829e-01 -7.75213242e-01 5.38425565e-01 6.28827751e-01
1.11752069e+00 -2.71636873e-01 9.23963010e-01 5.16332388e-01
-1.19151390e+00 -1.34897456e-01 2.30340257e-01 1.89155508e-02
1.57510430e-01 9.85326469e-02 -8.51236820e-01 3.99440110e-01
8.09508622e-01 5.30999005e-01 -1.85409173e-01 1.03393459e+00
3.41384523e-02 -2.63138920e-01 -3.01146120e-01 1.77130029e-01
9.35704112e-02 3.11093569e-01 5.26572824e-01 1.30055273e+00
4.41479415e-01 6.76906928e-02 3.32940012e-01 9.97175723e-02
3.32074374e-01 -2.64092356e-01 -7.40541279e-01 5.58081686e-01
1.34928226e-01 1.35120642e+00 -8.76544476e-01 -2.58189678e-01
-5.92077315e-01 1.06781173e+00 -2.47523174e-01 8.94839615e-02
-8.02170455e-01 -4.83862400e-01 1.54894188e-01 2.22995266e-01
-3.84449214e-01 -9.91510898e-02 1.49172945e-02 -1.05693269e+00
2.84192294e-01 -6.00250602e-01 6.72625482e-01 -1.15619016e+00
-3.21936011e-01 6.48037553e-01 -1.32419005e-01 -1.90958130e+00
-2.33598322e-01 -8.47702205e-01 -2.18287066e-01 6.60456002e-01
-9.52900112e-01 -9.78737473e-01 -9.25963521e-01 5.15645564e-01
8.24168563e-01 1.21262200e-01 1.08622670e+00 1.88728556e-01
-4.23726469e-01 2.36988455e-01 -1.74869582e-01 1.92203477e-01
9.03389215e-01 -1.12152171e+00 -6.07944727e-01 7.30186224e-01
-2.20893085e-01 7.34033227e-01 3.55279893e-01 -5.59675813e-01
-1.68031538e+00 -4.97110993e-01 1.68238133e-01 -5.91410398e-01
1.66850209e-01 4.94413674e-02 -4.33387011e-01 9.08187211e-01
1.02100722e-01 2.58682430e-01 6.86418295e-01 -2.95726806e-01
1.14662327e-01 -1.85587078e-01 -1.30469882e+00 1.72053188e-01
4.39213425e-01 -3.05106759e-01 -5.46167910e-01 3.78081053e-01
3.83434653e-01 -1.45994079e+00 -9.45654154e-01 2.96059817e-01
1.24658918e+00 -1.16460562e+00 8.28113258e-01 7.12698027e-02
5.97092628e-01 2.11179137e-01 9.08179730e-02 -1.05147624e+00
-1.96362689e-01 -3.25087070e-01 9.81375650e-02 2.82890618e-01
-7.47310231e-03 -6.75290823e-01 4.34633285e-01 2.73663223e-01
-3.48508447e-01 -8.10206592e-01 -1.08789766e+00 -6.17193878e-01
-4.34656292e-01 -2.38509804e-01 -5.08524477e-01 6.55679464e-01
2.31690928e-01 -2.18306497e-01 -1.84417680e-01 2.82384127e-01
1.95557252e-01 -6.36135563e-02 4.93974626e-01 -9.70282018e-01
-1.69840589e-01 -2.92537779e-01 -5.31268895e-01 -4.08774883e-01
-5.93975246e-01 -5.63818216e-01 -3.68148267e-01 -1.80029690e+00
3.04190159e-01 3.07347298e-01 -1.99315518e-01 3.96003813e-01
2.09756359e-01 -2.11788177e-01 1.64548621e-01 4.44779664e-01
-2.38831937e-01 -4.34862196e-01 1.27349138e+00 2.96195537e-01
-2.94639796e-01 1.12750143e-01 -9.59378183e-02 9.09979463e-01
6.01017356e-01 -5.63678920e-01 5.43777905e-02 -5.92526495e-01
-2.01049596e-01 7.74844944e-01 4.47704315e-01 -1.26100564e+00
5.40136933e-01 2.20969155e-01 6.23448670e-01 -4.27752763e-01
8.21297586e-01 -1.02750993e+00 4.91819084e-01 1.25881815e+00
2.77206302e-02 3.10990065e-01 4.61062163e-01 2.39255100e-01
-3.12451363e-01 -1.07786968e-01 1.09354675e+00 -8.76511455e-01
-4.77586120e-01 -9.57140326e-02 -5.88452876e-01 -3.96376222e-01
1.26035094e+00 -8.44765365e-01 -1.88126370e-01 -9.17881280e-02
-9.81077969e-01 3.88234667e-03 4.99655902e-01 6.38822019e-01
7.10196793e-01 -7.65262485e-01 -1.62830323e-01 4.39966440e-01
-2.44196504e-02 -4.21512611e-02 7.80693233e-01 1.46607316e+00
-8.68368268e-01 5.41444540e-01 -3.30284238e-01 -9.68309164e-01
-1.70912921e+00 4.17794287e-01 9.78094697e-01 -1.63986221e-01
-9.37584162e-01 7.59714127e-01 2.22268388e-01 1.45844787e-01
4.25606430e-01 -7.55509079e-01 -2.17355013e-01 3.50142866e-01
5.15733361e-01 3.80347878e-01 3.56244892e-01 -3.24805886e-01
-4.70454752e-01 9.97804642e-01 -9.01867915e-03 1.38042733e-01
9.08744991e-01 1.79993697e-02 1.64605260e-01 9.14446414e-01
1.13479090e+00 3.78952064e-02 -7.66852081e-01 1.66654974e-01
-6.66329861e-01 -6.66640222e-01 3.26403618e-01 -9.46155012e-01
-1.05812216e+00 8.00052762e-01 1.23154259e+00 8.44928399e-02
1.23831606e+00 -1.77715585e-01 4.69235808e-01 -4.72653769e-02
4.75836754e-01 -9.67815280e-01 -8.02932009e-02 7.46995583e-02
9.12030756e-01 -1.22421527e+00 -1.10885315e-01 -2.59090185e-01
-6.66084528e-01 1.60092163e+00 7.57737875e-01 -1.25448341e-02
2.93890357e-01 5.72277188e-01 1.95304960e-01 -2.33408034e-01
-5.02716482e-01 1.03015244e-01 5.80305196e-02 1.62711516e-01
5.75928807e-01 1.07255660e-01 -3.18162553e-02 4.96259302e-01
3.55410054e-02 3.85261863e-01 9.19301331e-01 1.26206720e+00
-2.81911671e-01 -1.18086867e-01 -4.58871156e-01 6.29418552e-01
-7.37220526e-01 4.01015073e-01 1.56000316e-01 1.14141107e+00
2.94672936e-01 9.39777076e-01 8.83553401e-02 -3.08470488e-01
8.76495838e-01 -2.46495932e-01 6.17017567e-01 -6.72226548e-01
-1.01590443e+00 5.30187964e-01 -7.73894861e-02 -1.03195167e+00
-1.93756104e-01 -6.25166118e-01 -1.12548137e+00 3.95937189e-02
-4.69218910e-01 -1.46165967e-01 7.66850293e-01 4.15979624e-01
1.87842175e-01 1.10267675e+00 2.98961341e-01 -1.07539070e+00
-4.93373051e-02 -8.66458178e-01 -6.58877730e-01 -1.27349719e-01
9.06058669e-01 -7.82050729e-01 -5.49914718e-01 1.03320010e-01] | [14.036246299743652, -3.348606824874878] |
9de860eb-58f1-44d4-abb3-2bd18042aeb0 | calls-japanese-empathetic-dialogue-speech | 2305.13713 | null | https://arxiv.org/abs/2305.13713v1 | https://arxiv.org/pdf/2305.13713v1.pdf | CALLS: Japanese Empathetic Dialogue Speech Corpus of Complaint Handling and Attentive Listening in Customer Center | We present CALLS, a Japanese speech corpus that considers phone calls in a customer center as a new domain of empathetic spoken dialogue. The existing STUDIES corpus covers only empathetic dialogue between a teacher and student in a school. To extend the application range of empathetic dialogue speech synthesis (EDSS), we designed our corpus to include the same female speaker as the STUDIES teacher, acting as an operator in simulated phone calls. We describe a corpus construction methodology and analyze the recorded speech. We also conduct EDSS experiments using the CALLS and STUDIES corpora to investigate the effect of domain differences. The results show that mixing the two corpora during training causes biased improvements in the quality of synthetic speech due to the different degrees of expressiveness. Our project page of the corpus is http://sython.org/Corpus/STUDIES-2. | ['Hiroshi Saruwatari', 'Kentaro Tachibana', 'Shinnosuke Takamichi', 'Eiji Iimori', 'Yuki Saito'] | 2023-05-23 | null | null | null | null | ['speech-synthesis'] | ['speech'] | [-4.73948717e-01 6.87366605e-01 1.79879472e-01 -5.88765979e-01
-8.15553188e-01 -4.61049676e-01 5.78686059e-01 -3.83951873e-01
-1.63080260e-01 8.13586831e-01 6.55694544e-01 -9.89853889e-02
2.68374771e-01 -4.47942734e-01 -4.04001474e-02 -6.37082338e-01
6.08476758e-01 8.59488249e-01 1.13801636e-01 -9.31759953e-01
1.14269890e-01 3.50372374e-01 -9.95873034e-01 6.79221869e-01
7.74357915e-01 -8.45379569e-03 3.77656817e-01 6.59257591e-01
-3.48724216e-01 1.05101371e+00 -1.48734725e+00 -5.66653192e-01
-1.81054875e-01 -1.04467177e+00 -1.24141932e+00 8.00401717e-02
-3.65669519e-01 -3.38900357e-01 -2.23650619e-01 8.04950833e-01
1.11457276e+00 4.56176221e-01 3.52121115e-01 -1.37904644e+00
-2.53653884e-01 1.47860658e+00 3.32052931e-02 -1.38396183e-02
6.87394619e-01 -4.03798297e-02 6.47818387e-01 -4.67560261e-01
6.97018027e-01 1.41976452e+00 4.61233288e-01 9.55366671e-01
-9.92357910e-01 -7.68445969e-01 -6.11093044e-01 1.75007612e-01
-9.32077885e-01 -7.01259673e-01 1.05414844e+00 -2.09399343e-01
8.77611279e-01 2.75477558e-01 6.60607219e-01 1.67921734e+00
-4.47988778e-01 8.11307609e-01 9.54289377e-01 -6.55615687e-01
1.63903743e-01 6.61510527e-01 1.89516380e-01 -7.03863055e-02
-8.19351554e-01 -2.15930209e-01 -6.07517421e-01 -2.62470186e-01
4.54929441e-01 -9.51475620e-01 -3.79182518e-01 3.27109605e-01
-1.19509089e+00 1.22776139e+00 -2.88908422e-01 6.90789640e-01
-1.90377980e-01 -4.16068226e-01 8.21668744e-01 6.66357696e-01
1.62400648e-01 7.75299966e-01 -2.65419066e-01 -8.09591472e-01
-3.29046637e-01 2.25913554e-01 1.45086527e+00 9.93323565e-01
-6.37196451e-02 2.96172321e-01 9.18301716e-02 1.59899139e+00
-1.42547593e-01 2.99966335e-01 7.69662559e-01 -1.51763976e+00
4.01183665e-01 1.54154032e-01 1.21203355e-01 -5.61784625e-01
-4.63715047e-01 4.35424745e-02 -1.44890144e-01 -2.11470991e-01
5.35833657e-01 -5.05096674e-01 1.52218550e-01 1.71936059e+00
5.39219201e-01 -3.51278841e-01 8.57270122e-01 7.10779369e-01
1.40420222e+00 1.05429995e+00 -1.13904491e-01 -5.39250791e-01
1.31967092e+00 -1.49100828e+00 -1.20770621e+00 1.41658470e-01
1.09892952e+00 -1.33891702e+00 1.57222915e+00 5.34624815e-01
-1.43830574e+00 -3.01550090e-01 -7.22615778e-01 -4.11977880e-02
-1.20565236e-01 1.87522247e-01 2.16594428e-01 6.45263553e-01
-6.81510746e-01 1.71842828e-01 -3.65778446e-01 -5.66716075e-01
-5.60274363e-01 9.40978527e-02 -3.07072729e-01 5.59085190e-01
-1.44835532e+00 1.01195931e+00 1.24014817e-01 -4.18391049e-01
-2.28900909e-01 -3.98849905e-01 -6.00454390e-01 2.23079380e-02
-1.12615246e-02 -1.40817821e-01 2.11240959e+00 -9.47705925e-01
-2.41827512e+00 8.84914696e-01 3.64268482e-01 -2.31603742e-01
4.05184895e-01 3.18377465e-02 -5.86456776e-01 4.09697622e-01
3.08044702e-02 5.86453676e-01 1.34985760e-01 -1.14556646e+00
-4.30543810e-01 -2.72786897e-02 -3.78444605e-02 2.92601526e-01
-2.66395696e-02 6.51222408e-01 9.06307325e-02 -7.18159199e-01
-1.94982171e-01 -1.03324533e+00 1.68859139e-01 -6.87624753e-01
-2.84432828e-01 -3.11499506e-01 7.10784376e-01 -6.30120158e-01
1.07531059e+00 -2.30809736e+00 -1.10067777e-01 -2.05968082e-01
-3.88218701e-01 1.43969491e-01 -1.17697276e-01 1.03699052e+00
-2.33004794e-01 -2.83421099e-01 1.19159352e-02 -5.23441613e-01
1.32612050e-01 6.09435499e-01 -5.19547522e-01 1.83141679e-01
-3.81934702e-01 3.15926373e-01 -7.69924045e-01 -5.83238959e-01
-4.20138240e-03 2.56353408e-01 -3.95149082e-01 5.95929801e-01
5.90138659e-02 8.33942413e-01 -3.26349676e-01 1.18263178e-01
2.68424720e-01 3.93861324e-01 1.65524617e-01 3.13864529e-01
-2.15145037e-01 9.12965059e-01 -8.30530822e-01 1.76514530e+00
-6.89252019e-01 6.17167175e-01 3.83985251e-01 -9.25595939e-01
1.17577672e+00 1.01612890e+00 2.24328443e-01 -5.92798829e-01
3.12683135e-01 3.96717608e-01 4.45285976e-01 -7.25487709e-01
7.55934775e-01 -3.19096208e-01 -4.21336323e-01 7.53118813e-01
9.01505053e-02 -7.43638813e-01 1.28478184e-01 1.72285736e-01
7.91754067e-01 -3.42258126e-01 2.60507196e-01 -1.39451027e-01
6.74808681e-01 2.53052950e-01 6.25771999e-01 3.69540155e-01
-4.56346184e-01 3.87963057e-01 8.57198358e-01 5.55857597e-03
-8.16399813e-01 -8.45665336e-01 -1.27575845e-01 1.29929841e+00
-2.52324104e-01 -1.76950186e-01 -9.10042703e-01 -4.96311963e-01
-6.66618526e-01 1.43920755e+00 -4.09224816e-02 -8.25186372e-02
-8.09191823e-01 -2.89318800e-01 1.20522761e+00 5.27049117e-02
4.50147003e-01 -1.66255438e+00 -4.07767713e-01 2.68434793e-01
-6.92784190e-01 -1.14505601e+00 -5.95429659e-01 -1.02483376e-03
-2.93560356e-01 -7.30766833e-01 -5.69807708e-01 -1.04135513e+00
1.01187527e-01 -1.52660891e-01 8.19367468e-01 -2.94673592e-01
2.22432658e-01 1.68225497e-01 -7.11192906e-01 -3.43230873e-01
-1.64002192e+00 1.52796865e-01 2.66136210e-02 -5.56980968e-01
1.18850492e-01 -8.05857718e-01 -1.16381180e-02 6.02997303e-01
-3.87129843e-01 1.25874922e-01 -6.19668290e-02 1.11505234e+00
-3.51237923e-01 -1.94281965e-01 1.07471061e+00 -7.76562989e-01
1.22550094e+00 -3.64804506e-01 -2.11374730e-01 2.00621574e-03
7.30085820e-02 -1.50058672e-01 7.47731090e-01 -7.78566658e-01
-1.48052454e+00 -1.67923301e-01 -7.65719295e-01 2.26765182e-02
-5.75723052e-01 1.20998837e-01 -3.33469450e-01 3.72372359e-01
7.40531385e-01 9.99078527e-02 2.90398896e-01 -5.29031932e-01
1.19595781e-01 1.26934338e+00 7.50263512e-01 -8.74727726e-01
5.79626560e-02 -2.87032187e-01 -6.56397581e-01 -1.13091135e+00
-4.01401252e-01 -3.66866469e-01 -2.09654257e-01 -3.13078821e-01
6.22602344e-01 -8.37567449e-01 -1.04931319e+00 3.65266651e-01
-1.55720699e+00 -7.33504176e-01 -4.30987775e-01 9.01152670e-01
-9.54053342e-01 2.74927944e-01 -1.07223797e+00 -6.48525774e-01
-3.86941910e-01 -1.34264207e+00 6.89334869e-01 2.45077491e-01
-8.95071149e-01 -9.15145516e-01 4.56713438e-01 8.81870806e-01
8.28040242e-02 -1.52258947e-01 1.04974425e+00 -1.31333745e+00
6.67437255e-01 1.92382038e-01 5.68905532e-01 4.28159714e-01
-1.18166491e-01 2.55033135e-01 -1.01256990e+00 2.45654926e-01
3.31319124e-01 -7.62986481e-01 -3.12318563e-01 -9.72805470e-02
3.56594741e-01 -6.44633770e-01 2.74709851e-01 -5.51964762e-03
3.80946487e-01 7.84636259e-01 6.16545200e-01 2.25824237e-01
-1.88160371e-02 1.29427695e+00 7.18068957e-01 6.14542902e-01
3.69046628e-01 8.86621833e-01 -3.47393662e-01 1.74220383e-01
5.36450669e-02 -1.59128129e-01 8.03237021e-01 1.71762216e+00
2.83822328e-01 -1.89316779e-01 -9.19133484e-01 6.48105681e-01
-1.70342147e+00 -1.21085608e+00 -3.35838228e-01 1.72389638e+00
1.45431161e+00 -2.75470227e-01 5.46153963e-01 1.86291531e-01
8.43702674e-01 6.17867783e-02 3.35068792e-01 -1.35142016e+00
-5.91051392e-02 1.77128822e-01 -3.08584273e-01 8.86268079e-01
-3.38832915e-01 1.13833523e+00 5.97858477e+00 1.01774025e+00
-1.06203163e+00 1.32191926e-01 1.77989587e-01 -1.79134011e-01
-4.70540859e-02 -1.32326828e-02 -4.41356748e-01 4.99095559e-01
1.23352802e+00 -3.28242153e-01 2.28539214e-01 9.20352817e-01
6.09830737e-01 -7.63644129e-02 -9.99340057e-01 7.62782872e-01
-1.44232735e-01 -9.03526425e-01 -3.10421407e-01 -1.28247336e-01
3.26830089e-01 -2.52972126e-01 -2.96170145e-01 6.25479817e-01
3.74516219e-01 -6.52540684e-01 6.63406491e-01 5.30417971e-02
2.86274374e-01 -1.03712595e+00 7.47290015e-01 7.08379030e-01
-2.52127111e-01 1.60191506e-01 -9.27508175e-02 -1.50118083e-01
4.61451471e-01 -2.74299290e-02 -1.30727720e+00 9.51651707e-02
5.28316379e-01 1.17136799e-01 9.82329696e-02 4.49055076e-01
-4.11094248e-01 1.01892257e+00 -2.03188986e-01 -1.19525611e-01
1.64814088e-02 -5.19500315e-01 8.79778087e-01 1.10993326e+00
1.93862960e-01 4.01121229e-01 1.53201744e-02 8.42519760e-01
2.15567678e-01 5.09570420e-01 -4.81488466e-01 -9.47767124e-02
8.29865038e-01 1.18485367e+00 -3.72023433e-01 -2.00358614e-01
-1.48930430e-01 9.99698281e-01 -8.96338448e-02 8.88289809e-02
-9.18943524e-01 -6.69888377e-01 5.70821881e-01 -9.76009592e-02
-2.12498009e-01 1.97315682e-02 -1.62568748e-01 -5.34329116e-01
-3.92188847e-01 -1.27569950e+00 8.75267088e-02 -1.07487524e+00
-1.08061016e+00 8.18028927e-01 1.37962162e-01 -9.61945295e-01
-7.64975250e-01 5.69790527e-02 -8.89533162e-01 7.38405466e-01
-5.65480828e-01 -8.63477588e-01 4.21430878e-02 5.20786107e-01
9.06271279e-01 -4.35808510e-01 1.13008058e+00 3.93325895e-01
-7.22682297e-01 7.28468478e-01 -1.43126756e-01 1.98749900e-01
1.11480558e+00 -9.31256413e-01 8.81014839e-02 2.83530559e-02
-1.63463935e-01 3.50385606e-01 1.16992509e+00 -2.31466725e-01
-4.56241876e-01 -3.50104779e-01 1.31021011e+00 -4.18315232e-02
8.03834617e-01 -3.56115699e-01 -8.86591971e-01 4.11086082e-01
8.89324009e-01 -6.98751271e-01 1.00779545e+00 -2.61961132e-01
9.69004780e-02 2.11824447e-01 -1.38030827e+00 7.12745845e-01
5.71900070e-01 -3.98908257e-01 -1.11161888e+00 3.62398952e-01
1.08490014e+00 -5.28714597e-01 -1.07850742e+00 -1.34664416e-01
3.02568316e-01 -9.95468140e-01 4.72245812e-01 -3.86990041e-01
5.05485535e-01 2.12700665e-01 -3.85527499e-02 -1.58758867e+00
4.71861899e-01 -1.16650534e+00 5.35191178e-01 1.59297729e+00
4.78669256e-01 -8.85863721e-01 4.83896881e-01 4.65871632e-01
-6.27759933e-01 -2.92126924e-01 -9.10073042e-01 -5.53950191e-01
3.83287072e-01 -1.38861373e-01 5.64529896e-01 1.34150004e+00
9.12446141e-01 8.57073367e-01 -1.46254927e-01 -1.23573475e-01
-5.54297157e-02 -6.17457833e-03 9.83364940e-01 -9.03983891e-01
-5.34336448e-01 -4.10796762e-01 2.47235224e-01 -7.54971206e-01
5.76779604e-01 -4.57416713e-01 1.85815424e-01 -1.15074742e+00
-6.07154012e-01 -3.02536786e-01 9.21261549e-01 5.38016438e-01
3.70142877e-01 -2.43503347e-01 2.07013041e-01 6.75661564e-02
1.07239708e-01 8.95436704e-01 1.26966798e+00 3.88638079e-01
-7.53586769e-01 3.84012222e-01 -2.81968832e-01 7.95280337e-01
1.21592200e+00 -6.40228033e-01 -6.23244166e-01 4.76292782e-02
-2.82821417e-01 8.22180629e-01 -1.58688456e-01 -6.40089273e-01
1.41519666e-01 -2.89718300e-01 -6.14531636e-01 -4.08216804e-01
6.82480514e-01 -5.47303021e-01 7.34350905e-02 2.76984572e-01
-6.65386736e-01 4.62454557e-03 1.75053790e-01 -4.35412914e-01
-5.59417605e-01 -5.48228145e-01 1.01931036e+00 -6.20491952e-02
-8.13116133e-02 -7.72501826e-01 -8.97794604e-01 2.75060982e-01
1.00094271e+00 -2.14070663e-01 -5.80007732e-01 -9.68158364e-01
-7.25711703e-01 1.31335661e-01 2.45214775e-01 2.57110089e-01
2.37907916e-01 -1.04102647e+00 -6.72588885e-01 -1.05116367e-01
-7.03400653e-03 -2.05184385e-01 2.69477397e-01 9.88719761e-01
-7.39648819e-01 1.73414305e-01 -3.40507001e-01 -8.63514468e-02
-1.62911749e+00 -3.15774493e-02 3.88266325e-01 4.96274419e-02
-6.40413702e-01 6.50930941e-01 -1.52008105e-02 -1.11618137e+00
4.27473158e-01 -6.65807202e-02 -3.36594701e-01 1.15203924e-01
2.11878911e-01 6.62929177e-01 -2.38532975e-01 -8.89133811e-01
3.66067812e-02 -2.27551058e-01 1.12247854e-01 -7.54821062e-01
1.11768162e+00 -1.43590778e-01 -5.68763204e-02 7.16489315e-01
1.08372867e+00 5.43967783e-01 -4.86043334e-01 1.13668501e-01
-1.24699228e-01 -2.78246198e-02 -4.54394192e-01 -7.51645327e-01
-6.21769965e-01 7.04342186e-01 -6.43858463e-02 1.70621082e-01
7.88147151e-01 8.23137611e-02 1.09028339e+00 5.94786346e-01
-2.34868620e-02 -1.68312347e+00 1.59162790e-01 9.13137913e-01
1.18432748e+00 -8.79777372e-01 -5.10631204e-01 -5.24057686e-01
-1.45368755e+00 1.05492961e+00 7.45549679e-01 2.76498377e-01
2.64353693e-01 4.57842886e-01 7.17208207e-01 -6.66802153e-02
-8.38323712e-01 2.08930790e-01 -4.75657046e-01 6.24622047e-01
6.83658123e-01 7.01882392e-02 -7.94329882e-01 1.11621165e+00
-1.24747717e+00 -4.07971323e-01 1.22424865e+00 8.25749874e-01
-3.26426446e-01 -1.28171134e+00 -5.96904874e-01 -5.34409404e-01
-3.78536165e-01 8.35529715e-02 -9.94865060e-01 1.00528896e+00
-1.87175959e-01 1.26622009e+00 1.37948662e-01 -3.20981473e-01
6.34076238e-01 3.84479463e-01 1.66607693e-01 -7.25213528e-01
-1.15798402e+00 1.63388118e-01 8.57943773e-01 -9.78075638e-02
-3.53069514e-01 -5.00546932e-01 -1.87109137e+00 -5.63960731e-01
-2.27232054e-01 9.20628309e-01 7.35249281e-01 6.95113420e-01
-2.92221084e-02 2.87868351e-01 9.51842785e-01 -5.35862625e-01
-8.04052353e-01 -1.37218332e+00 -8.40481699e-01 1.68107569e-01
-1.45376101e-01 -3.18655282e-01 -6.09429300e-01 -1.79051384e-01] | [13.09902286529541, 7.687444686889648] |
fb8ac24e-9050-4e61-a432-0e6d2d939f7a | exact-recovery-for-the-non-uniform-hypergraph | 2304.13139 | null | https://arxiv.org/abs/2304.13139v1 | https://arxiv.org/pdf/2304.13139v1.pdf | Exact recovery for the non-uniform Hypergraph Stochastic Block Model | Consider the community detection problem in random hypergraphs under the non-uniform hypergraph stochastic block model (HSBM), where each hyperedge appears independently with some given probability depending only on the labels of its vertices. We establish, for the first time in the literature, a sharp threshold for exact recovery under this non-uniform case, subject to minor constraints; in particular, we consider the model with $K$ classes as well as the symmetric binary model ($K=2$). One crucial point here is that by aggregating information from all the uniform layers, we may obtain exact recovery even in cases when this may appear impossible if each layer were considered alone. Two efficient algorithms that successfully achieve exact recovery above the threshold are provided. The theoretical analysis of our algorithms relies on the concentration and regularization of the adjacency matrix for non-uniform random hypergraphs, which could be of independent interest. We also address some open problems regarding parameter knowledge and estimation. | ['Haixiao Wang', 'Ioana Dumitriu'] | 2023-04-25 | null | null | null | null | ['stochastic-block-model', 'community-detection'] | ['graphs', 'graphs'] | [ 4.71468896e-01 4.13977742e-01 -2.73026645e-01 3.95429015e-01
-6.13149226e-01 -8.00762236e-01 1.23859599e-01 3.87707114e-01
-2.26486221e-01 8.75810623e-01 -2.40681425e-01 -2.13791400e-01
-5.36142230e-01 -1.11350822e+00 -8.06255996e-01 -1.30162835e+00
-3.61611038e-01 1.06621313e+00 3.92482877e-01 1.54122859e-01
1.01046972e-01 5.81799567e-01 -1.10545146e+00 -2.31880069e-01
7.64036655e-01 5.00110567e-01 -1.32783711e-01 9.43953037e-01
8.87315795e-02 5.13174295e-01 -3.32604706e-01 -7.00019777e-01
6.10240579e-01 -2.86090583e-01 -9.61419046e-01 4.87615287e-01
3.06589782e-01 1.14558131e-01 -6.19237185e-01 1.53041553e+00
2.39109948e-01 -2.50109434e-01 8.69600534e-01 -1.12242317e+00
-2.75436670e-01 1.16947234e+00 -1.32130539e+00 1.60713241e-01
2.65380055e-01 -1.27658471e-01 1.30443192e+00 -1.58385411e-01
6.88011885e-01 8.63234937e-01 5.40388823e-01 1.58678845e-01
-1.63850749e+00 -7.19656587e-01 -8.56584385e-02 3.37711066e-01
-1.85716009e+00 -9.94744226e-02 5.79421759e-01 -5.58832705e-01
4.08325285e-01 3.34171325e-01 4.45247024e-01 5.11809111e-01
-2.64344692e-01 5.69192410e-01 1.10408604e+00 -6.91062689e-01
2.52633337e-02 1.11553170e-01 4.97078270e-01 8.71883452e-01
1.29440296e+00 -7.82932267e-02 -5.26167676e-02 -4.86131728e-01
5.33344865e-01 -1.85596332e-01 -6.47289574e-01 -7.42735922e-01
-9.01603043e-01 7.72847295e-01 3.02229196e-01 5.68714201e-01
-2.27610350e-01 1.59484327e-01 1.53219894e-01 3.72865647e-01
2.10102066e-01 -7.79276341e-02 -2.96253413e-02 4.67387408e-01
-1.00802433e+00 -8.68404210e-02 1.02389956e+00 1.26376498e+00
1.03336501e+00 -2.60149837e-01 1.05815344e-01 4.05400842e-01
-7.20683634e-02 7.05337524e-01 -6.38451278e-01 -6.18238330e-01
6.69770777e-01 3.36222142e-01 1.58259749e-01 -1.07773650e+00
-2.87118495e-01 -7.76812971e-01 -1.21134341e+00 -7.75422677e-02
7.39342272e-01 8.74969363e-02 -6.18647456e-01 1.95523024e+00
2.23046064e-01 3.27239245e-01 -9.54566151e-02 5.43160617e-01
4.00682509e-01 3.49638134e-01 -3.45336765e-01 -6.63086951e-01
1.01208889e+00 -5.36977291e-01 -4.74367052e-01 1.28828585e-01
5.27323246e-01 -4.77885842e-01 2.93408185e-01 4.82670337e-01
-1.12772894e+00 2.40981892e-01 -9.12801981e-01 6.14869356e-01
6.79394230e-02 -1.19410425e-01 3.40486735e-01 9.44516838e-01
-1.16869414e+00 4.26024646e-01 -6.22491419e-01 -4.59994435e-01
6.23052008e-02 4.00934368e-01 -3.85449439e-01 -5.40088356e-01
-8.43393207e-01 4.77308691e-01 2.43302450e-01 8.93049836e-02
-4.99848843e-01 -1.43787891e-01 -5.74683189e-01 3.37922215e-01
7.29498327e-01 -5.54720402e-01 6.75247848e-01 -7.21530735e-01
-5.80871642e-01 8.83536875e-01 -1.92328930e-01 -4.39025849e-01
5.04605889e-01 6.88415229e-01 3.39616388e-02 6.84295416e-01
8.51990581e-02 -2.75098085e-01 4.83641475e-01 -1.35560524e+00
-2.76398838e-01 -5.43812394e-01 2.61940241e-01 -1.46785662e-01
-6.17756695e-02 -1.98666647e-01 -5.70476055e-01 -1.92636549e-01
2.85108507e-01 -1.30767429e+00 -4.25390959e-01 -3.38581324e-01
-8.42467368e-01 1.07421011e-01 1.00794956e-01 -4.75436270e-01
1.26652539e+00 -1.93480468e+00 5.18405020e-01 9.39037681e-01
5.75699031e-01 -5.00014499e-02 2.96891443e-02 8.12270820e-01
-1.54729217e-01 2.63794720e-01 -5.71818352e-01 -4.49279845e-02
-3.12817134e-02 3.51184420e-02 7.26744831e-02 1.10835171e+00
-3.13604027e-01 2.94730067e-01 -7.70485640e-01 -5.31976521e-01
-1.49484098e-01 5.24039418e-02 -6.54784501e-01 -2.49684542e-01
9.51813161e-02 1.35698281e-02 -3.85996670e-01 4.75718945e-01
1.21743190e+00 -7.19955921e-01 7.60386050e-01 -2.74037160e-02
1.05828218e-01 -3.68195593e-01 -1.83066392e+00 8.54055822e-01
-6.11573085e-02 4.40920293e-01 7.84460962e-01 -1.21534050e+00
1.31935135e-01 4.24570352e-01 5.88541985e-01 8.11873451e-02
-5.29228710e-03 2.17289537e-01 1.27734393e-01 -2.06298470e-01
2.51223445e-01 -3.27581495e-01 -3.81769583e-04 7.28105664e-01
-1.82070509e-01 4.08373445e-01 5.44021368e-01 8.20296347e-01
1.62544489e+00 -7.58873761e-01 2.74360001e-01 -4.76752490e-01
3.63531381e-01 -1.52384162e-01 3.32870722e-01 1.07016885e+00
4.68990533e-03 5.18433571e-01 9.39849794e-01 4.11959708e-01
-8.36041808e-01 -8.88450444e-01 -2.64313728e-01 4.15616393e-01
4.21429306e-01 -4.23550367e-01 -7.08260477e-01 -4.96970266e-01
6.91961721e-02 2.41294697e-01 -8.05592418e-01 -6.23963308e-03
-2.12900057e-01 -1.34484899e+00 1.16176270e-01 8.86857063e-02
2.37212285e-01 -5.50881207e-01 1.38681486e-01 8.22698772e-02
-2.90291935e-01 -1.23551440e+00 -3.53477925e-01 1.95320129e-01
-7.68221200e-01 -1.75017369e+00 -7.58796453e-01 -6.73754036e-01
1.02555645e+00 7.14805663e-01 9.61050630e-01 4.88480836e-01
-2.25416541e-01 6.13417387e-01 -2.57549495e-01 3.86207104e-01
-3.52092087e-01 2.56328464e-01 -2.96327025e-01 1.08662553e-01
9.69549194e-02 -5.26249528e-01 -5.10933697e-01 1.98361740e-01
-9.69719946e-01 -2.66003102e-01 5.03707469e-01 6.22649670e-01
3.67326558e-01 6.66994691e-01 1.37484837e-02 -1.39943433e+00
1.38436422e-01 -7.71256387e-01 -7.66565561e-01 2.85156041e-01
-4.91683185e-01 1.56565905e-01 3.63930732e-01 -2.39009857e-02
-5.69498539e-01 2.56793369e-02 3.01209718e-01 -2.47076415e-02
2.26525635e-01 5.10727406e-01 -2.58850485e-01 -4.60022748e-01
2.80755609e-01 6.02035858e-02 -3.02560955e-01 -3.91878784e-01
1.96314946e-01 4.59188014e-01 1.50530636e-01 -4.65396047e-01
1.12810576e+00 7.24518597e-01 5.93329847e-01 -1.05992460e+00
-4.89599764e-01 -7.12203920e-01 -4.96558130e-01 -1.65473506e-01
3.92536044e-01 -7.48185754e-01 -9.14253175e-01 3.34794313e-01
-9.71844852e-01 -1.80976912e-01 7.65889138e-02 2.77310014e-01
-3.85685742e-01 9.14407134e-01 -9.22858775e-01 -1.10847664e+00
1.36947095e-01 -1.00756574e+00 5.95369875e-01 -1.19936422e-01
3.16011250e-01 -1.01550508e+00 1.02695823e-01 1.62859261e-01
2.02824548e-02 1.85488477e-01 1.17681456e+00 -4.91923302e-01
-1.09743273e+00 -4.27262485e-01 -5.51738262e-01 -1.95089832e-01
-2.20127299e-01 1.88113302e-02 -3.90144885e-01 -4.90613401e-01
-2.76979119e-01 6.88265488e-02 1.22421551e+00 5.03320992e-01
8.86428833e-01 -4.04609412e-01 -7.19199240e-01 4.14400548e-01
1.81188583e+00 -4.25157607e-01 5.42588174e-01 3.30874533e-03
5.97990036e-01 5.42263031e-01 -1.57854762e-02 4.08005476e-01
2.08954841e-01 5.54572403e-01 4.19408530e-01 -3.63325924e-02
-1.54723767e-02 1.49898767e-01 -1.61461845e-01 8.96601558e-01
-1.34323433e-01 -7.16147661e-01 -7.23276675e-01 7.22430229e-01
-1.64739168e+00 -1.14308202e+00 -7.10593224e-01 2.69659615e+00
7.57888734e-01 2.07797617e-01 2.01844007e-01 3.98224264e-01
1.30089533e+00 8.14555511e-02 -1.56021729e-01 1.07797563e-01
-4.05636489e-01 2.17235759e-01 1.09910452e+00 8.75100374e-01
-7.21133590e-01 4.55957592e-01 6.41333151e+00 8.63754928e-01
-4.14789587e-01 1.71775550e-01 3.41242105e-01 5.88834397e-02
-5.02876997e-01 3.30631226e-01 -7.53811419e-01 4.54854071e-01
6.70354486e-01 -3.49550515e-01 3.69284689e-01 4.41174686e-01
-3.18720043e-01 -4.38648015e-01 -6.38285160e-01 6.51107788e-01
3.85292843e-02 -9.39866066e-01 -9.13809389e-02 6.44529700e-01
1.09577417e+00 -2.83003449e-01 -2.03672349e-01 -1.66450918e-01
7.31086016e-01 -6.41890824e-01 6.25942126e-02 2.72863060e-01
8.58098030e-01 -9.03615892e-01 7.21735835e-01 4.28880483e-01
-1.32900119e+00 1.86037570e-02 -4.05255437e-01 2.50982940e-01
1.01735905e-01 9.37894225e-01 -5.41980624e-01 7.95925736e-01
2.05349967e-01 3.66417080e-01 -1.77244723e-01 1.47815859e+00
8.07059258e-02 8.59169722e-01 -7.15376079e-01 1.48033276e-01
-7.61741772e-02 -5.16941845e-01 8.98606539e-01 1.14264297e+00
4.26929444e-01 3.69428813e-01 1.50787339e-01 4.01791632e-01
-1.87485933e-01 2.46837690e-01 -5.47928512e-01 3.94903868e-02
3.03251654e-01 1.04851007e+00 -1.18934679e+00 -1.95153773e-01
-4.75831926e-01 7.59994507e-01 4.16185439e-01 5.70995390e-01
-4.83533770e-01 -4.19383377e-01 2.33801007e-01 5.36597133e-01
7.09875345e-01 -1.69454515e-01 -7.15906918e-02 -1.31260943e+00
-1.36083374e-02 -5.69224000e-01 6.53728783e-01 -1.62862524e-01
-1.29538643e+00 4.43333179e-01 1.45855069e-01 -8.11963081e-01
-2.95061618e-02 -5.20501852e-01 -3.35461408e-01 6.87029004e-01
-1.33735156e+00 -6.82498336e-01 1.34653542e-02 4.46097791e-01
-3.78846705e-01 3.04511160e-01 5.21816313e-01 3.20133656e-01
-7.09484637e-01 5.77481627e-01 4.95427638e-01 -4.34549861e-02
3.27496111e-01 -1.33842611e+00 -1.73773855e-01 1.11844981e+00
1.19455069e-01 4.60050285e-01 1.04027641e+00 -7.06355453e-01
-1.38819206e+00 -7.51893759e-01 8.72126997e-01 5.33198044e-02
7.45364249e-01 -4.85804409e-01 -8.23687196e-01 6.27529085e-01
4.10587862e-02 -3.07268854e-02 6.93163991e-01 1.76389679e-01
-3.32442135e-01 -2.93351505e-02 -1.25365961e+00 3.71358424e-01
1.13196111e+00 -2.55944282e-01 1.34900227e-01 7.01901793e-01
2.58260638e-01 -2.84681637e-02 -5.41218221e-01 3.28354597e-01
1.93327948e-01 -9.24838126e-01 8.41075003e-01 -2.69190043e-01
7.10341185e-02 -2.97576249e-01 -2.29194582e-01 -1.01424170e+00
-4.29693013e-01 -7.43022323e-01 -8.17356929e-02 9.68528450e-01
4.38008934e-01 -6.57668889e-01 9.85402584e-01 3.37796152e-01
6.37089372e-01 -3.01516354e-01 -1.05114162e+00 -6.89933717e-01
-5.98579720e-02 -1.34787604e-01 3.42731208e-01 8.22662354e-01
1.93033651e-01 2.47759059e-01 -4.34322834e-01 5.83163261e-01
1.25236750e+00 4.94036406e-01 5.39507389e-01 -1.44947076e+00
-8.14974904e-01 -4.33476925e-01 -4.88030344e-01 -1.01076722e+00
2.41365701e-01 -1.08561003e+00 -2.66621932e-02 -1.50092185e+00
9.15214300e-01 -6.42850876e-01 -1.30438879e-01 -7.05841333e-02
-1.67297602e-01 4.25828636e-01 4.23130691e-02 1.40246794e-01
-7.38020957e-01 1.21747345e-01 1.07578397e+00 -1.03426136e-01
1.60830498e-01 5.58979332e-01 -8.12479138e-01 4.27352816e-01
4.73858714e-01 -6.49917543e-01 -1.05044335e-01 -1.34064719e-01
4.38855201e-01 5.69008827e-01 2.59111941e-01 -8.45657885e-01
4.04709309e-01 1.13700368e-02 -1.59904435e-01 -4.93899405e-01
9.91298556e-02 -9.55165803e-01 5.30569911e-01 5.02052188e-01
-2.43364006e-01 -2.84019858e-01 -1.42024279e-01 1.12203407e+00
1.44557044e-01 -7.77425289e-01 8.76651406e-01 -4.89962175e-02
1.24405831e-01 5.00364006e-01 -4.15466517e-01 -6.05881214e-04
1.10636222e+00 -1.41539171e-01 -4.35911596e-01 -8.03750753e-01
-9.18278635e-01 2.14519665e-01 6.84038281e-01 -6.34748697e-01
6.57552034e-02 -1.09040749e+00 -8.90191138e-01 4.51752320e-02
4.25363258e-02 -2.80948132e-01 6.15415215e-01 1.08448946e+00
-3.48616213e-01 3.41144830e-01 3.65728348e-01 -4.98694658e-01
-1.36182642e+00 7.98561990e-01 2.08089411e-01 -5.60481369e-01
-4.28413659e-01 5.95761776e-01 3.35929364e-01 2.60154933e-01
1.57627255e-01 2.36862421e-01 6.79050088e-02 -9.54856500e-02
4.03936327e-01 5.68492353e-01 6.64994791e-02 -7.87538648e-01
-1.44376099e-01 7.23136902e-01 -1.95608750e-01 -1.90473333e-01
1.07662940e+00 -3.49670261e-01 -4.93335992e-01 1.75251231e-01
9.78108287e-01 6.15711987e-01 -8.42333794e-01 -5.75680971e-01
-1.16697013e-01 -5.51918685e-01 -1.00129709e-01 -3.09956431e-01
-1.36186862e+00 6.96663141e-01 -4.23226156e-04 7.22053289e-01
1.05352771e+00 3.89876574e-01 4.23273921e-01 1.90447152e-01
6.89996064e-01 -5.12609422e-01 -4.93902296e-01 2.33366862e-01
4.95114505e-01 -7.28123605e-01 2.12020084e-01 -9.57175732e-01
-1.18684769e-03 9.54866946e-01 1.85662732e-01 -3.48619491e-01
8.00958514e-01 3.46061498e-01 -7.50250340e-01 -1.46162823e-01
-6.72095716e-01 -6.76315010e-01 -5.87306917e-02 4.69944745e-01
7.47299567e-02 2.25348011e-01 -5.64331472e-01 2.69786984e-01
2.30997518e-01 -5.21997929e-01 9.53961849e-01 5.63990116e-01
-5.23639262e-01 -1.26607323e+00 -4.37441587e-01 6.60469353e-01
-6.38895094e-01 -2.21276909e-01 -4.31709141e-01 7.61091590e-01
-1.82507589e-01 8.90371978e-01 -2.05369964e-01 -1.39524534e-01
8.91645402e-02 -3.24157685e-01 8.50810587e-01 -6.68048203e-01
-2.67223209e-01 3.17071497e-01 -3.40277888e-03 -6.96725920e-02
-3.98425579e-01 -8.07792902e-01 -7.59286463e-01 -7.85762250e-01
-7.10673332e-01 5.72498381e-01 1.76626220e-01 7.02936292e-01
-8.55170116e-02 -2.08258405e-02 7.01595366e-01 -4.00762886e-01
-5.51528215e-01 -8.12458336e-01 -1.34046531e+00 1.05658829e-01
3.28287899e-01 -5.25467455e-01 -1.03370404e+00 -2.34709829e-01] | [6.810599327087402, 5.037517070770264] |
4ba4716d-29c9-4ddd-8264-af714ae0aafa | multi-source-morphosyntactic-tagging-for | null | null | https://aclanthology.org/W17-1210 | https://aclanthology.org/W17-1210.pdf | Multi-source morphosyntactic tagging for spoken Rusyn | This paper deals with the development of morphosyntactic taggers for spoken varieties of the Slavic minority language Rusyn. As neither annotated corpora nor parallel corpora are electronically available for Rusyn, we propose to combine existing resources from the etymologically close Slavic languages Russian, Ukrainian, Slovak, and Polish and adapt them to Rusyn. Using MarMoT as tagging toolkit, we show that a tagger trained on a balanced set of the four source languages outperforms single language taggers by about 9{\%}, and that additional automatically induced morphosyntactic lexicons lead to further improvements. The best observed accuracies for Rusyn are 82.4{\%} for part-of-speech tagging and 75.5{\%} for full morphological tagging. | ['Yves Scherrer', 'Achim Rabus'] | 2017-04-01 | null | null | null | ws-2017-4 | ['morphological-tagging'] | ['natural-language-processing'] | [-3.87747735e-01 2.20671475e-01 -5.35786152e-01 -2.16474652e-01
-7.87770748e-01 -9.59320247e-01 5.35950720e-01 2.27636456e-01
-7.24039853e-01 1.12868261e+00 1.40804097e-01 -6.58503830e-01
-9.35838223e-02 -6.26272976e-01 2.29162380e-01 -3.03508162e-01
2.46060312e-01 1.07757366e+00 4.11347479e-01 -5.07799625e-01
-7.41177574e-02 5.72181404e-01 -8.62204850e-01 -2.18250349e-01
1.03137636e+00 2.75193732e-02 2.10764855e-01 4.46748197e-01
-4.24549013e-01 3.65209997e-01 -7.51359284e-01 -6.00635290e-01
2.94273466e-01 -4.48026389e-01 -1.22385645e+00 2.38451343e-02
5.22511363e-01 2.88173616e-01 -7.96327293e-02 1.27004313e+00
2.95374066e-01 1.20463572e-01 6.97310507e-01 -5.00518680e-01
-6.47345066e-01 1.28146458e+00 -2.28611529e-01 1.63055733e-01
1.88343897e-01 -1.92346469e-01 1.28966188e+00 -7.64982164e-01
1.10708928e+00 1.01291418e+00 5.37250638e-01 7.66625702e-01
-1.06050146e+00 -7.84730554e-01 1.09737501e-01 -1.43967897e-01
-1.70359385e+00 -3.85567218e-01 5.86154044e-01 -6.42508388e-01
1.16496098e+00 8.16103220e-02 7.15218544e-01 3.12628418e-01
-2.31353343e-01 4.00790364e-01 1.12757444e+00 -9.33277488e-01
1.20054092e-02 3.32455158e-01 1.66953385e-01 7.91839957e-01
5.21867394e-01 -1.37649057e-02 -3.26731026e-01 -2.00531855e-01
6.87693834e-01 -5.34626126e-01 2.77131438e-01 1.27616867e-01
-7.57214844e-01 7.67084301e-01 -3.95266175e-01 9.11867321e-01
2.28856102e-01 -2.16459394e-01 7.13073254e-01 4.25473064e-01
6.57990158e-01 6.02472425e-01 -9.05598402e-01 -2.36532569e-01
-9.36975658e-01 7.70358220e-02 6.16652310e-01 1.15428615e+00
7.26265550e-01 6.98091030e-01 5.74986696e-01 1.45856190e+00
1.12099953e-01 8.05048704e-01 6.34608448e-01 -5.46063185e-01
4.01083767e-01 7.13561296e-01 -7.27159604e-02 -1.11158803e-01
-5.42473614e-01 -6.30526338e-03 -1.92447323e-02 -3.46728079e-02
6.48048818e-01 -5.09717107e-01 -1.19312239e+00 1.64557731e+00
1.80632621e-01 -4.31357354e-01 4.75102991e-01 4.75189775e-01
9.76233602e-01 4.79018897e-01 6.11347795e-01 -3.12464684e-01
1.51169848e+00 -5.07998407e-01 -4.47476327e-01 -2.29898244e-01
9.89448428e-01 -1.15057075e+00 7.50478387e-01 5.36341593e-02
-1.28873348e+00 -3.70165825e-01 -8.00045073e-01 2.97532856e-01
-3.22877914e-01 3.76022726e-01 8.23848784e-01 1.17349803e+00
-1.14203930e+00 3.38705778e-01 -7.59247899e-01 -6.60494983e-01
-2.19317704e-01 3.96354854e-01 -5.76484799e-01 4.46277857e-01
-1.09524739e+00 6.39120221e-01 6.16820812e-01 -4.26344216e-01
-1.64666086e-01 -3.70529681e-01 -1.16455626e+00 -4.03501540e-01
8.58820416e-03 1.51751041e-01 1.26103640e+00 -1.11908197e+00
-1.71272731e+00 1.76905990e+00 1.97806790e-01 -1.13908783e-01
4.44212332e-02 1.80299640e-01 -8.92289996e-01 1.90656289e-01
3.13820034e-01 3.14349383e-01 1.02861926e-01 -7.72453785e-01
-9.72633719e-01 -2.89274752e-01 -2.59723544e-01 1.29651591e-01
-6.92395195e-02 9.86111104e-01 -1.53983459e-01 -7.11504400e-01
3.10682833e-01 -1.07690704e+00 -2.58721679e-01 -1.04542446e+00
-1.14931934e-01 -4.80668336e-01 1.27702653e-01 -1.13291955e+00
9.89160240e-01 -1.93768609e+00 -3.06025326e-01 3.12727690e-01
-1.01264328e-01 6.48403227e-01 -7.11484402e-02 6.21460974e-01
-2.15571538e-01 2.38252714e-01 7.75294602e-02 5.61934821e-02
-1.06107414e-01 4.15297806e-01 -9.28168073e-02 5.30760288e-01
7.57105695e-03 6.71990573e-01 -1.00133765e+00 -5.14057040e-01
2.38818467e-01 -2.46170804e-01 -3.06236267e-01 -3.93492967e-01
1.14242367e-01 2.79294729e-01 -2.13081613e-01 7.57846236e-01
5.67180634e-01 5.82542837e-01 8.67186248e-01 9.14024472e-01
-4.62595612e-01 9.50934410e-01 -7.87116528e-01 1.43421996e+00
-5.88572681e-01 4.74310130e-01 4.98608612e-02 -6.91594183e-01
1.34010994e+00 5.32481432e-01 4.16156590e-01 -3.52043867e-01
1.25644088e-01 9.13670182e-01 4.64265853e-01 -4.43972051e-02
8.68092656e-01 -4.94334757e-01 -7.80648708e-01 3.57691906e-02
5.00848532e-01 -3.26179892e-01 2.78847218e-01 -3.74168158e-02
8.99287462e-01 1.82438016e-01 8.57693315e-01 -9.44136262e-01
6.50562882e-01 4.82963145e-01 1.16669178e+00 2.43621305e-01
-2.99972296e-01 2.12679103e-01 2.43262559e-01 -2.84370422e-01
-1.05468130e+00 -1.28784287e+00 -3.70206416e-01 1.32671344e+00
-1.28448069e-01 -6.29610896e-01 -5.68088055e-01 -7.15835631e-01
-3.65939438e-01 5.40801346e-01 -7.81056210e-02 3.19045126e-01
-1.02091825e+00 -4.09809619e-01 1.19562805e+00 4.83188570e-01
1.71850845e-01 -1.33984506e+00 -9.48214903e-02 3.45704526e-01
1.96794540e-01 -1.25880325e+00 -1.86007008e-01 2.45344192e-01
-7.45382607e-01 -9.17653561e-01 -4.46144015e-01 -1.51389325e+00
3.09994549e-01 -2.20058069e-01 1.07406545e+00 2.79637966e-02
-1.18244961e-01 -8.56993273e-02 -4.21916783e-01 -3.50622892e-01
-7.72965372e-01 2.96023011e-01 2.49078915e-01 -8.91203701e-01
7.72690177e-01 -4.00264889e-01 3.06786358e-01 8.69909599e-02
-2.82257259e-01 -3.34787369e-01 3.93626615e-02 6.06155396e-01
2.90228367e-01 -4.44330841e-01 5.60332894e-01 -1.60345566e+00
-1.76038638e-01 -4.36977744e-02 -8.83561134e-01 -1.69435292e-01
-4.00130153e-01 -2.85122186e-01 8.49945247e-01 -2.75849789e-01
-1.21583200e+00 1.40430525e-01 -6.69157624e-01 2.45259374e-01
-5.57579219e-01 3.50621015e-01 -4.75488037e-01 8.05500671e-02
5.88691413e-01 3.65571454e-02 -7.39067733e-01 -5.96434176e-01
4.10573244e-01 8.06156695e-01 7.15583265e-01 -6.31484687e-01
8.12249005e-01 1.12246975e-01 -2.70524591e-01 -1.17620170e+00
-3.44089389e-01 -9.49005008e-01 -9.23316956e-01 1.50480360e-01
8.45292091e-01 -1.00316942e+00 -2.47130364e-01 3.82952452e-01
-1.04002059e+00 -5.58787882e-01 -6.57721698e-01 7.40680754e-01
-6.44225955e-01 3.98120254e-01 -9.36004162e-01 -7.20136046e-01
-3.83073777e-01 -5.93075633e-01 7.88759708e-01 2.39751503e-01
-5.16984582e-01 -1.42953479e+00 5.21490335e-01 5.84931672e-02
-4.04721826e-01 -2.81965792e-01 1.25748742e+00 -1.22885692e+00
2.79595047e-01 -1.11595802e-01 5.62805198e-02 2.18235672e-01
2.87672073e-01 -1.02292076e-01 -6.82681441e-01 -3.14452238e-02
-6.02810681e-01 -1.95220798e-01 6.44206762e-01 1.25779631e-03
-1.89256191e-01 -2.16428161e-01 -1.18169628e-01 3.57026249e-01
1.32202435e+00 6.93791330e-01 3.23572278e-01 2.59483784e-01
5.22513151e-01 6.76890314e-01 5.90914488e-01 1.24495082e-01
3.63147467e-01 6.25962913e-01 -4.44965780e-01 2.95370221e-01
-4.33618486e-01 -3.76892269e-01 7.35365629e-01 1.49338603e+00
-1.97510630e-01 2.42684171e-01 -1.30783248e+00 1.15542245e+00
-1.59984338e+00 -8.08073997e-01 -2.47790352e-01 2.12758255e+00
1.01962078e+00 -1.19549647e-01 2.19542444e-01 -4.35246415e-02
9.45353031e-01 7.24560618e-02 3.93254042e-01 -8.17934394e-01
-3.46329302e-01 9.87201452e-01 8.82420480e-01 1.00669694e+00
-9.92073715e-01 2.27055979e+00 6.36983967e+00 1.05541241e+00
-9.80989993e-01 2.67076999e-01 -4.89759445e-02 3.42073619e-01
-4.07704294e-01 4.45121109e-01 -1.15952194e+00 -7.06208870e-02
1.14710796e+00 -2.19705135e-01 7.27700368e-02 8.57714236e-01
1.06499694e-01 -3.20434161e-02 -4.50899750e-01 8.51334572e-01
4.81813587e-02 -1.00625181e+00 -2.73949593e-01 1.13573767e-01
1.15437126e+00 3.01197946e-01 -3.49768460e-01 4.11699176e-01
1.20082057e+00 -6.25723779e-01 9.20695186e-01 -2.21162423e-01
1.12593019e+00 -8.93796384e-01 8.49650562e-01 5.87957948e-02
-1.40027511e+00 5.50769448e-01 -8.99795711e-01 -2.20193774e-01
1.17212765e-01 3.44520956e-01 -1.09752095e+00 6.83851421e-01
1.14245355e-01 2.74213314e-01 -5.29742122e-01 9.25334036e-01
-8.59695435e-01 1.19646776e+00 -2.88959116e-01 -1.84119269e-01
4.88539338e-01 -4.12341654e-01 6.07726932e-01 1.42451322e+00
2.77136236e-01 -7.54314363e-02 6.49587631e-01 -1.73158254e-02
1.29891476e-02 1.01560640e+00 -4.89013582e-01 -8.50986131e-03
5.01923144e-01 1.22979987e+00 -1.07488108e+00 -3.93776268e-01
-4.10026133e-01 9.03321445e-01 4.22859341e-01 9.42797810e-02
-1.18109576e-01 -5.75027645e-01 9.25039947e-01 1.80804893e-01
1.27243936e-01 -5.24698138e-01 -4.42609966e-01 -1.01330256e+00
-4.23082530e-01 -4.98495311e-01 8.04657936e-01 -1.00770853e-01
-1.06065631e+00 8.37169528e-01 -4.28550728e-02 -7.89949119e-01
-7.31061399e-01 -9.48162794e-01 -5.33434391e-01 1.06314540e+00
-1.09278858e+00 -1.41812551e+00 4.96571720e-01 2.88842827e-01
4.82358098e-01 -7.98081398e-01 1.07990158e+00 -3.60283367e-02
-1.98468238e-01 6.56398356e-01 6.48874864e-02 6.24703467e-01
7.57811725e-01 -1.42103171e+00 6.86435103e-01 9.21692252e-01
5.38949728e-01 3.25039148e-01 6.36469424e-01 -9.54731703e-01
-4.08035845e-01 -1.06861258e+00 1.85317385e+00 -1.88271016e-01
1.20721829e+00 -3.80100816e-01 -4.63353187e-01 8.51527274e-01
3.12954217e-01 -2.35841349e-01 9.40253377e-01 5.79730451e-01
-4.84416485e-01 3.47218066e-01 -1.05404174e+00 7.74880886e-01
1.38778079e+00 -6.47730589e-01 -7.21564174e-01 3.52490127e-01
4.39766318e-01 -6.58996701e-02 -8.41260672e-01 -7.29106069e-02
1.25023246e-01 -3.84246469e-01 2.72718638e-01 -9.29548264e-01
-2.03767985e-01 2.72550471e-02 -1.50965557e-01 -1.54350388e+00
-3.35516185e-01 -9.33262408e-01 1.02851439e+00 1.67026138e+00
6.48817062e-01 -7.30681419e-01 1.02306128e+00 1.31538481e-01
-6.68129802e-01 2.67775357e-01 -1.19394898e+00 -1.17128992e+00
7.03958094e-01 -6.47669554e-01 2.76342005e-01 1.15458965e+00
9.09558415e-01 4.59208935e-01 -1.55030966e-01 -5.86295761e-02
5.26924133e-02 1.08487673e-01 4.65002686e-01 -1.57149124e+00
-3.32938313e-01 -3.91721368e-01 -9.58742857e-01 -3.56965482e-01
9.24895108e-01 -1.32561338e+00 1.61436126e-02 -1.34753716e+00
-4.45609726e-02 -7.89489746e-01 1.04679741e-01 8.03081870e-01
4.19220701e-02 6.74591064e-01 2.32921317e-01 8.14535692e-02
-2.99099892e-01 1.86509684e-01 6.92197323e-01 2.11910933e-01
-6.33494079e-01 -3.70995104e-02 -5.40606916e-01 1.29665720e+00
1.04461062e+00 -4.51326817e-01 9.42650363e-02 -1.98675632e-01
1.93078578e-01 -2.67182197e-02 -3.57289881e-01 -7.55998790e-01
-2.20954567e-01 -6.07106686e-01 -1.20374247e-01 -1.90122247e-01
7.95583352e-02 -3.12620699e-01 -1.13818794e-01 4.59256023e-01
2.27673113e-01 2.48090580e-01 4.09490615e-01 -2.63708472e-01
-1.79401994e-01 -8.30304563e-01 1.16578305e+00 -2.46619374e-01
-9.76909399e-01 1.01572886e-01 -9.60662842e-01 5.83439767e-01
1.01686752e+00 -2.39604264e-02 -2.17489213e-01 -1.27801955e-01
-8.19619358e-01 -3.18031400e-01 6.60312116e-01 1.10625811e-01
6.76816031e-02 -1.11737859e+00 -6.88764811e-01 1.62251949e-01
4.59767759e-01 -7.62682676e-01 -9.87970643e-03 2.05873176e-01
-9.23756361e-01 6.00421011e-01 -3.84125710e-01 2.18777314e-01
-1.44327855e+00 1.63225740e-01 -3.75634022e-02 -4.77524757e-01
-5.34834743e-01 7.07220912e-01 -1.28190545e-02 -1.07637835e+00
-4.06782269e-01 -1.92344002e-02 -3.18695962e-01 -3.24438065e-02
2.51795650e-02 1.31194487e-01 1.96072280e-01 -1.24962282e+00
-6.14813685e-01 4.62086678e-01 -1.68664139e-02 -6.04011655e-01
9.24177468e-01 -1.23364367e-02 -7.15141445e-02 5.25475383e-01
4.31628555e-01 1.14409757e+00 -4.83924955e-01 -2.35146701e-01
3.34625810e-01 -1.11474752e-01 -4.21233654e-01 -5.43178558e-01
-6.36018157e-01 5.04125834e-01 -7.30894133e-02 -1.48094997e-01
6.28006339e-01 3.14491421e-01 6.09975219e-01 1.67790771e-01
6.56454384e-01 -1.42693079e+00 -7.51409471e-01 1.17107284e+00
3.68957594e-02 -4.36526179e-01 -3.82916212e-01 -9.25089598e-01
-7.37512290e-01 6.94386303e-01 3.00248921e-01 -5.51167309e-01
6.31120861e-01 3.34860384e-01 6.07233703e-01 7.56259114e-02
-3.79226416e-01 -9.83831882e-01 1.01297177e-01 9.04680133e-01
7.99925923e-01 5.77945292e-01 -1.24856162e+00 8.13807905e-01
-9.24163282e-01 -6.54721260e-01 8.09041917e-01 5.91205478e-01
-6.01219237e-01 -1.72962034e+00 -3.29258859e-01 2.10946947e-01
-9.51302588e-01 -5.19939065e-01 -6.86168253e-01 1.22800922e+00
3.98478836e-01 8.53165209e-01 2.02469483e-01 -2.86941528e-01
1.05800159e-01 3.84479791e-01 5.52146375e-01 -1.35410154e+00
-9.06087041e-01 4.94740248e-01 8.55315208e-01 2.69244164e-01
-1.98791668e-01 -8.96856248e-01 -1.62282836e+00 -2.56221294e-01
-7.14446530e-02 7.55146205e-01 2.86853820e-01 1.06598651e+00
-5.84384680e-01 -5.14194816e-02 1.48100838e-01 -2.74836212e-01
-2.82547832e-01 -9.71787989e-01 -1.26280344e+00 1.23618700e-01
-5.95945537e-01 -4.14076209e-01 1.27397016e-01 2.76613124e-02] | [10.367143630981445, 10.113484382629395] |
a433c11e-1959-4987-824c-028b84ab145b | quantifying-the-knowledge-in-a-dnn-to-explain | 2208.08741 | null | https://arxiv.org/abs/2208.08741v1 | https://arxiv.org/pdf/2208.08741v1.pdf | Quantifying the Knowledge in a DNN to Explain Knowledge Distillation for Classification | Compared to traditional learning from scratch, knowledge distillation sometimes makes the DNN achieve superior performance. This paper provides a new perspective to explain the success of knowledge distillation, i.e., quantifying knowledge points encoded in intermediate layers of a DNN for classification, based on the information theory. To this end, we consider the signal processing in a DNN as the layer-wise information discarding. A knowledge point is referred to as an input unit, whose information is much less discarded than other input units. Thus, we propose three hypotheses for knowledge distillation based on the quantification of knowledge points. 1. The DNN learning from knowledge distillation encodes more knowledge points than the DNN learning from scratch. 2. Knowledge distillation makes the DNN more likely to learn different knowledge points simultaneously. In comparison, the DNN learning from scratch tends to encode various knowledge points sequentially. 3. The DNN learning from knowledge distillation is often optimized more stably than the DNN learning from scratch. In order to verify the above hypotheses, we design three types of metrics with annotations of foreground objects to analyze feature representations of the DNN, \textit{i.e.} the quantity and the quality of knowledge points, the learning speed of different knowledge points, and the stability of optimization directions. In experiments, we diagnosed various DNNs for different classification tasks, i.e., image classification, 3D point cloud classification, binary sentiment classification, and question answering, which verified above hypotheses. | ['Zhefan Rao', 'Yilan Chen', 'Xu Cheng', 'Quanshi Zhang'] | 2022-08-18 | null | null | null | null | ['3d-point-cloud-classification', 'point-cloud-classification'] | ['computer-vision', 'computer-vision'] | [-2.09098175e-01 1.07850201e-01 -1.46266118e-01 -2.63871193e-01
-8.50318447e-02 -6.01871371e-01 6.14930801e-02 7.79938176e-02
-4.92747337e-01 6.62132680e-01 -1.89388826e-01 -2.05848530e-01
-5.27415335e-01 -9.81747925e-01 -8.65394711e-01 -8.57524037e-01
2.36495182e-01 1.78055137e-01 4.06884819e-01 -7.24364538e-03
1.90957144e-01 5.50643027e-01 -1.41759324e+00 3.11127096e-01
7.97997892e-01 1.44977713e+00 3.49828511e-01 3.62838507e-01
-6.22636974e-01 7.82208562e-01 -7.25236833e-01 -4.69079703e-01
3.12605768e-01 -1.89649194e-01 -7.76499033e-01 -1.59074321e-01
3.03804219e-01 -1.82974488e-01 -3.46184283e-01 1.44476724e+00
3.13665390e-01 1.22936018e-01 6.00095987e-01 -1.16447759e+00
-9.88033772e-01 9.00948107e-01 -1.94621339e-01 5.00699818e-01
-2.02634528e-01 9.24990773e-02 9.73838151e-01 -9.16049659e-01
3.38800758e-01 1.10933053e+00 6.39987707e-01 4.19481844e-01
-7.33073771e-01 -7.37935424e-01 4.05927002e-01 4.45721477e-01
-1.65847242e+00 6.32110692e-04 8.81964326e-01 -4.24007893e-01
4.36696261e-01 6.78589288e-03 8.68600965e-01 7.35946774e-01
-1.78913832e-01 1.37859261e+00 7.43156791e-01 -1.52331561e-01
3.31071973e-01 4.84230489e-01 5.30213237e-01 6.59718275e-01
5.29045284e-01 1.12230942e-01 -6.01839721e-01 3.51076424e-01
8.37980986e-01 2.98770189e-01 -3.63880932e-01 -2.74954766e-01
-1.04690564e+00 5.75206876e-01 8.91457856e-01 4.26210672e-01
-2.66285479e-01 -3.16758193e-02 2.86074728e-01 4.65895385e-01
1.12234078e-01 4.24330175e-01 -9.43229616e-01 1.28243968e-03
-7.67558157e-01 -9.66849700e-02 6.66512251e-01 1.13680732e+00
1.22789347e+00 1.81593344e-01 -2.32481748e-01 5.32908916e-01
1.99242488e-01 5.45969069e-01 4.74768758e-01 -6.34558976e-01
4.72742498e-01 1.13605094e+00 -2.88216710e-01 -9.26596045e-01
-3.29701573e-01 -5.71559787e-01 -1.00932336e+00 6.74367994e-02
3.11222792e-01 -5.38124084e-01 -9.12829220e-01 1.52091241e+00
1.29669935e-01 2.67582178e-01 4.87003148e-01 8.41627479e-01
1.20458841e+00 6.49408162e-01 -3.63175511e-01 -2.45156154e-01
1.03905213e+00 -7.32634306e-01 -7.61735082e-01 1.90029740e-01
5.28158486e-01 -3.48020345e-01 9.00283635e-01 7.60364175e-01
-8.72856021e-01 -9.92457628e-01 -9.79254127e-01 3.33901234e-02
-8.01019251e-01 1.69154033e-01 6.42277896e-01 4.73806173e-01
-7.73204863e-01 7.26682246e-01 -5.86597621e-01 -1.09083481e-01
7.65425146e-01 2.28378609e-01 -8.58501643e-02 -2.72782817e-02
-1.23337841e+00 7.66677022e-01 1.06734180e+00 2.57485956e-01
-7.09658146e-01 -8.09522986e-01 -5.23910940e-01 2.75733113e-01
5.41058123e-01 -5.42371988e-01 1.01333392e+00 -1.27838635e+00
-1.34629512e+00 4.07433897e-01 1.74373135e-01 -3.41372967e-01
1.36972070e-01 -4.02279258e-01 -3.79922956e-01 1.15902491e-01
-2.33430311e-01 6.04555786e-01 7.47909725e-01 -1.29558361e+00
-9.03757036e-01 -3.32776070e-01 1.42628506e-01 2.75215119e-01
-4.62087035e-01 -6.32915080e-01 -5.15739560e-01 -5.72117150e-01
2.68802196e-01 -4.15278912e-01 4.81740758e-02 6.83939755e-02
-3.33810359e-01 -3.91624540e-01 9.57368433e-01 -3.08193237e-01
1.05779207e+00 -2.28572965e+00 2.19317794e-01 3.44637334e-01
3.43512595e-01 5.46776533e-01 -3.66805270e-02 -3.04321408e-01
-4.25016880e-02 1.13745488e-01 4.44394685e-02 2.67254889e-01
-6.11539781e-02 6.40152395e-01 -2.67646253e-01 -5.02352268e-02
1.89126462e-01 1.16377091e+00 -9.69123304e-01 -3.80553693e-01
1.53927997e-01 1.06562853e-01 -5.06401777e-01 -2.15597391e-01
-3.88812065e-01 1.18815862e-01 -5.67755103e-01 7.39786863e-01
6.84419274e-01 -1.75239533e-01 -3.46767426e-01 -6.09037042e-01
-6.14911057e-02 -2.66774982e-01 -1.33678019e+00 1.60249245e+00
-7.83362240e-02 7.11856246e-01 -1.53156877e-01 -1.14830482e+00
1.09302938e+00 1.59104273e-01 2.59055048e-01 -6.61583662e-01
2.59533048e-01 1.92630097e-01 1.15010418e-01 -4.82998550e-01
2.46075585e-01 -1.35773331e-01 1.15124784e-01 -4.04777117e-02
4.67445433e-01 -8.05304348e-02 -1.30546000e-02 -5.36660217e-02
9.68383491e-01 -1.36748031e-01 8.23955536e-02 -3.19736004e-01
4.48646337e-01 -5.62118478e-02 6.43912852e-01 8.14466417e-01
-2.12067664e-01 1.96203172e-01 3.99213195e-01 -4.62371439e-01
-5.51206768e-01 -1.42727864e+00 -2.06138507e-01 7.63774872e-01
5.27857125e-01 -1.52675435e-01 -5.02291083e-01 -1.00680900e+00
1.29739791e-01 6.29410207e-01 -4.51164335e-01 -5.65881550e-01
-1.11185737e-01 -3.29806119e-01 6.44856095e-01 5.99212348e-01
1.12687552e+00 -9.58685696e-01 -2.45415077e-01 -5.64120524e-02
6.30123913e-02 -9.53727007e-01 -8.31396356e-02 2.77365655e-01
-1.25249946e+00 -1.21691811e+00 -7.50007570e-01 -9.22513306e-01
7.65124440e-01 1.82148233e-01 9.86119211e-01 -2.01934157e-03
-9.40229092e-03 3.96914780e-01 -5.63233137e-01 -7.79325485e-01
1.97319150e-01 1.70065671e-01 1.87361717e-01 -1.22969691e-02
7.27722406e-01 -4.44481134e-01 -4.43765074e-01 4.02029514e-01
-1.18280685e+00 -9.78678912e-02 1.08078969e+00 6.36699855e-01
8.98598135e-01 7.87495852e-01 4.74363416e-01 -5.14743567e-01
6.66780055e-01 -3.04316431e-01 -5.26655734e-01 3.48416775e-01
-3.27782303e-01 4.14284110e-01 7.63866961e-01 -6.29454136e-01
-8.19823325e-01 -1.19661562e-01 -8.58826265e-02 -8.72544706e-01
-2.27512702e-01 6.70661390e-01 -3.72370601e-01 -1.65309131e-01
7.36522079e-01 4.79625642e-01 -2.43428364e-01 -6.12569153e-01
5.66770196e-01 3.52573276e-01 4.04734075e-01 -5.54878116e-01
8.56294990e-01 2.79092014e-01 -1.45782083e-01 -8.69800687e-01
-1.12466156e+00 -2.94500083e-01 -7.20546126e-01 -2.05500871e-01
8.49367976e-01 -5.85748792e-01 -1.02628458e+00 5.62618315e-01
-1.16844308e+00 -1.29317269e-01 -6.95816696e-01 5.86207449e-01
-1.24374293e-01 1.49862081e-01 -2.01730758e-01 -6.01444006e-01
-2.95510858e-01 -7.66343176e-01 6.39238834e-01 5.72959006e-01
3.73099357e-01 -1.06261897e+00 -2.57094294e-01 1.69509631e-02
1.03202142e-01 4.58466187e-02 9.23194826e-01 -7.54929960e-01
-6.57095015e-01 -3.23379561e-02 -4.47091788e-01 8.85013580e-01
2.02796564e-01 -1.25971884e-01 -9.85051811e-01 1.22192271e-01
2.60523051e-01 -2.10952073e-01 1.15123820e+00 2.42797256e-01
1.41555059e+00 -6.63522661e-01 -2.23401964e-01 6.13401711e-01
1.39861977e+00 3.35601300e-01 6.83433115e-01 4.74403165e-02
8.22727203e-01 2.41020456e-01 3.94569397e-01 1.86493680e-01
1.56378478e-01 1.50846988e-01 3.98838460e-01 1.19722135e-01
-7.58050159e-02 -2.20773101e-01 2.10421264e-01 1.16251421e+00
-1.08061060e-01 7.95039907e-02 -9.59136188e-01 4.39321697e-01
-1.86428344e+00 -8.06124568e-01 1.46432489e-01 1.94779682e+00
8.56185615e-01 5.01351535e-01 -3.02213877e-01 3.07706565e-01
5.62545896e-01 -2.56747484e-01 -9.03019071e-01 -1.09356835e-01
-2.17597842e-01 7.05494657e-02 5.12102783e-01 2.78487533e-01
-7.75892675e-01 9.91458833e-01 5.45371294e+00 1.13420975e+00
-1.11644363e+00 -3.95261832e-02 4.04258639e-01 -1.05861358e-01
-1.73497438e-01 -1.90756500e-01 -1.03008783e+00 6.17381990e-01
2.91059226e-01 -8.60240459e-02 4.25972193e-01 1.05475938e+00
-1.70576662e-01 -1.07995495e-01 -1.27457237e+00 1.30942166e+00
-1.29744992e-01 -1.50402308e+00 6.12079084e-01 -2.48945668e-01
9.08191681e-01 6.92917360e-03 9.21249166e-02 7.57854044e-01
1.99310690e-01 -8.28568876e-01 7.38499820e-01 1.12135386e+00
3.11609954e-01 -8.40219617e-01 9.33212698e-01 6.83629632e-01
-1.18766773e+00 -1.52849823e-01 -5.28975129e-01 -1.12007409e-01
-3.07408452e-01 9.26859319e-01 -5.43190122e-01 7.98680186e-01
8.09696496e-01 7.69922912e-01 -4.95990932e-01 1.16149509e+00
-4.71343696e-01 5.97437561e-01 -4.32143122e-01 -2.89073616e-01
2.79349744e-01 -5.82882538e-02 3.33361864e-01 8.94083738e-01
2.29613096e-01 4.15195048e-01 -2.69431509e-02 1.15691173e+00
-2.39138499e-01 -1.67691931e-01 -4.46840882e-01 -5.17133176e-02
4.79572713e-01 9.91637707e-01 -6.89837515e-01 -3.82213473e-01
-1.97363630e-01 8.39635611e-01 2.50636101e-01 6.25594676e-01
-6.97317183e-01 -7.44535744e-01 6.27592146e-01 -2.19716251e-01
5.02596259e-01 -2.47130901e-01 -5.43057561e-01 -1.18423223e+00
1.02295853e-01 -3.59705478e-01 2.87922770e-01 -8.56123388e-01
-1.36415184e+00 3.60391825e-01 -1.29540041e-02 -1.11123097e+00
4.26460803e-01 -1.05869353e+00 -6.50182009e-01 7.26916730e-01
-1.77857375e+00 -5.30903280e-01 -5.65493703e-01 8.62737536e-01
3.29472542e-01 -2.18932956e-01 3.87599230e-01 3.94438714e-01
-6.30989134e-01 6.47810340e-01 1.59364462e-01 4.33129042e-01
2.27315590e-01 -1.09557748e+00 -2.12869540e-01 5.69339335e-01
3.47352713e-01 6.10170424e-01 1.63456455e-01 -4.85272706e-01
-1.37145424e+00 -1.17049336e+00 3.95036131e-01 -4.04943287e-01
5.73851705e-01 -1.33448079e-01 -1.00681543e+00 4.53618407e-01
-2.63494879e-01 -1.63089596e-02 5.46008766e-01 1.78949699e-01
-2.02379957e-01 -6.43545866e-01 -1.02135205e+00 4.53955084e-01
8.46481204e-01 -4.60145265e-01 -1.03463519e+00 1.56145334e-01
9.20838177e-01 -1.24472268e-01 -8.07747841e-01 5.14055371e-01
2.15213597e-01 -7.00433731e-01 8.97133887e-01 -4.62766796e-01
2.16431946e-01 -5.25111020e-01 -1.62966639e-01 -1.21151149e+00
-4.00017083e-01 -1.58810228e-01 -5.54928780e-01 1.14684856e+00
3.61647487e-01 -5.15242159e-01 9.30408418e-01 2.61807203e-01
-2.30514526e-01 -8.79961252e-01 -9.79080081e-01 -1.00806475e+00
1.86838537e-01 -5.74888885e-01 6.02652669e-01 9.91529226e-01
-2.58527249e-01 4.86594260e-01 1.50422707e-01 3.69937569e-01
3.90559554e-01 -4.74421382e-02 4.48041081e-01 -1.48454678e+00
-5.59227727e-02 -7.68188894e-01 -6.77531600e-01 -1.60870790e+00
-1.63380265e-01 -1.03503513e+00 -1.87284291e-01 -1.73782432e+00
6.97063208e-02 -4.72529113e-01 -5.82360864e-01 7.42532969e-01
-1.66873112e-01 -2.13179559e-01 1.76015303e-01 3.75593752e-01
-7.75761545e-01 5.33399820e-01 1.69408250e+00 -3.10194731e-01
-3.17550719e-01 1.01911351e-01 -8.55205297e-01 9.79649305e-01
7.14284301e-01 -1.97927579e-01 -6.40360296e-01 -5.81175208e-01
5.14085531e-01 -3.56378287e-01 4.36358750e-01 -1.19374025e+00
5.66837907e-01 -1.60177629e-02 7.18388319e-01 -9.17340457e-01
1.39977440e-01 -1.07389021e+00 -3.26506555e-01 3.48944336e-01
-1.49182305e-02 -5.11136055e-01 4.27037150e-01 4.85725552e-01
-3.65550041e-01 -4.35168862e-01 6.41778827e-01 -2.83185005e-01
-1.09071505e+00 3.40368271e-01 6.27613533e-03 2.63400108e-01
8.73565674e-01 -6.08772993e-01 -1.26723066e-01 6.07978441e-02
-8.08078885e-01 3.62675786e-01 -1.13836966e-01 4.23618644e-01
9.14620101e-01 -1.50238287e+00 -2.13568062e-01 3.77023607e-01
-1.42763942e-01 5.46959162e-01 5.78566045e-02 6.43927753e-01
-3.69393826e-01 3.41901958e-01 -6.65410385e-02 -6.81642175e-01
-7.50795007e-01 6.57461226e-01 4.49732244e-01 7.22994655e-02
-2.91565567e-01 1.04899347e+00 3.55867952e-01 -4.82629210e-01
5.78797877e-01 -1.00381386e+00 -2.81919271e-01 2.38130704e-01
3.81258816e-01 4.08589125e-01 -5.81138767e-03 4.81470190e-02
-2.14629039e-01 6.73990726e-01 -9.52210799e-02 2.90470630e-01
9.22565877e-01 2.94248220e-02 3.21716629e-02 6.34433210e-01
1.15753496e+00 -4.45473760e-01 -1.17159641e+00 -6.07044101e-01
2.28227768e-02 -1.09973587e-01 2.10999563e-01 -9.49067950e-01
-1.31921542e+00 8.92344415e-01 5.50307333e-01 1.82585970e-01
1.18110681e+00 8.61224756e-02 6.27247691e-01 1.03510964e+00
3.42529744e-01 -1.30201185e+00 3.86732846e-01 9.93988037e-01
6.87411129e-01 -1.06805217e+00 -9.57600549e-02 -5.19071519e-02
-5.73171675e-01 1.33027458e+00 7.76183367e-01 8.65293965e-02
1.07623267e+00 -8.61944705e-02 -1.59872368e-01 -3.39827985e-01
-4.09782976e-01 -2.72225618e-01 4.14497584e-01 3.33037436e-01
-3.11142981e-01 -1.02417454e-01 1.56091034e-01 1.17598939e+00
-2.89253384e-01 3.51808995e-01 -1.01237640e-01 7.42632449e-01
-8.69789183e-01 -3.72983068e-01 -2.01417059e-01 4.73283350e-01
-2.41953842e-02 -2.32350409e-01 -4.47120219e-01 6.74314797e-01
8.35463107e-01 6.89652264e-01 8.59666616e-03 -7.16287613e-01
5.79268336e-01 1.26853250e-02 3.71087909e-01 -5.80406606e-01
-4.12526101e-01 -5.21890044e-01 -4.77786750e-01 -2.73171544e-01
-2.56896585e-01 -2.26937741e-01 -1.55407810e+00 -1.89636618e-01
-6.97000086e-01 3.54606718e-01 6.00720823e-01 1.13105059e+00
2.48914823e-01 7.72835493e-01 4.44776386e-01 -3.41698289e-01
-3.45865637e-01 -6.82462335e-01 -6.86767340e-01 1.49905324e-01
1.54108316e-01 -7.18163848e-01 -4.99977648e-01 -1.01402192e-03] | [9.528505325317383, 3.2108795642852783] |
6ed1a259-09d8-4d22-840e-6fb8e96b4c43 | few-shot-audio-visual-learning-of-environment | 2206.04006 | null | https://arxiv.org/abs/2206.04006v2 | https://arxiv.org/pdf/2206.04006v2.pdf | Few-Shot Audio-Visual Learning of Environment Acoustics | Room impulse response (RIR) functions capture how the surrounding physical environment transforms the sounds heard by a listener, with implications for various applications in AR, VR, and robotics. Whereas traditional methods to estimate RIRs assume dense geometry and/or sound measurements throughout the environment, we explore how to infer RIRs based on a sparse set of images and echoes observed in the space. Towards that goal, we introduce a transformer-based method that uses self-attention to build a rich acoustic context, then predicts RIRs of arbitrary query source-receiver locations through cross-attention. Additionally, we design a novel training objective that improves the match in the acoustic signature between the RIR predictions and the targets. In experiments using a state-of-the-art audio-visual simulator for 3D environments, we demonstrate that our method successfully generates arbitrary RIRs, outperforming state-of-the-art methods and -- in a major departure from traditional methods -- generalizing to novel environments in a few-shot manner. Project: http://vision.cs.utexas.edu/projects/fs_rir. | ['Kristen Grauman', 'Ziad Al-Halah', 'Changan Chen', 'Sagnik Majumder'] | 2022-06-08 | null | null | null | null | ['room-impulse-response'] | ['audio'] | [ 3.49571407e-01 -1.64700404e-01 1.02245283e+00 -3.31593603e-01
-1.38222933e+00 -6.05718791e-01 4.94642943e-01 -1.13602266e-01
-4.31724302e-02 1.39636979e-01 6.14189565e-01 -1.21609271e-01
-4.71745022e-02 -5.60089767e-01 -9.25332844e-01 -3.75294685e-01
-3.85981858e-01 3.82646054e-01 1.04301900e-01 -4.04823184e-01
1.57888904e-01 4.67768431e-01 -1.71776843e+00 1.91109344e-01
2.86540568e-01 8.09983313e-01 5.54466903e-01 1.39035892e+00
2.55473822e-01 8.61122429e-01 -6.10623240e-01 2.02464551e-01
3.30523461e-01 -2.51791030e-01 -3.17243963e-01 -3.41749877e-01
5.31655431e-01 -2.45737508e-01 -8.21600556e-01 6.40381992e-01
8.31119835e-01 7.12849379e-01 5.76843023e-01 -8.75219584e-01
-8.67470741e-01 3.03985119e-01 -2.06309780e-01 2.98692524e-01
8.90472174e-01 3.58787342e-03 9.47577655e-01 -1.13511336e+00
2.19016328e-01 1.24249995e+00 7.66580105e-01 4.98277217e-01
-1.10654163e+00 -7.16343880e-01 1.54368848e-01 9.67134386e-02
-1.47406614e+00 -9.31280255e-01 9.13925052e-01 -3.70637774e-01
9.67985630e-01 4.18579817e-01 3.67055148e-01 1.31728792e+00
-1.81025252e-01 5.28922737e-01 7.19480157e-01 -4.18374121e-01
3.52990955e-01 3.70648429e-02 -2.80046493e-01 4.88349617e-01
-5.63186884e-01 3.87695849e-01 -7.33134747e-01 -2.13509500e-01
8.31814706e-01 -2.90342987e-01 -5.88024199e-01 -2.80172139e-01
-1.23213875e+00 5.67117333e-01 5.53114295e-01 2.48386651e-01
-1.84669137e-01 5.69132507e-01 -1.10271014e-01 1.06150262e-01
4.13216442e-01 5.24453223e-01 -1.34880483e-01 -2.06157073e-01
-5.60430050e-01 1.11841753e-01 8.13024998e-01 8.08149219e-01
4.70406681e-01 4.82454628e-01 1.08354017e-01 9.66824591e-01
5.56135237e-01 8.56425107e-01 1.71263590e-01 -1.22844636e+00
1.72449604e-01 -6.29210353e-01 2.60773599e-01 -1.00795305e+00
-1.66717231e-01 -5.92702627e-01 -3.16262931e-01 2.39611372e-01
-1.08812310e-01 4.47889753e-02 -1.07349610e+00 1.76765060e+00
4.49405044e-01 1.00967884e+00 7.30742067e-02 1.05099380e+00
9.29237902e-01 9.39472437e-01 -3.83517683e-01 1.19422868e-01
7.89671421e-01 -8.57843280e-01 -4.78974193e-01 -5.35340667e-01
9.32282954e-02 -7.41823912e-01 1.20835710e+00 4.95076418e-01
-9.74657714e-01 -9.25134957e-01 -8.81309986e-01 1.96871191e-01
3.30412462e-02 -3.01776201e-01 3.07792485e-01 6.13254845e-01
-1.32514608e+00 1.19050778e-01 -7.97761738e-01 -1.94358528e-01
-1.43577278e-01 -3.30682658e-02 -2.07627848e-01 -2.05668166e-01
-8.39456677e-01 6.61864281e-01 -5.41131794e-01 3.06514263e-01
-1.47952259e+00 -1.02081025e+00 -1.05850112e+00 -4.98804636e-02
2.27942377e-01 -6.18002176e-01 1.63540900e+00 -5.21796942e-01
-1.64884877e+00 2.12375820e-01 -2.92396545e-01 -1.20936789e-01
5.87504730e-02 -5.27306020e-01 -6.40270531e-01 8.43173787e-02
1.21713430e-01 4.54480112e-01 6.99425638e-01 -1.96036065e+00
-3.72869521e-01 -2.65997142e-01 -5.45055121e-02 5.59472799e-01
2.05324814e-01 -6.85070232e-02 -3.01651180e-01 -5.33293843e-01
1.67455658e-01 -7.67188787e-01 -5.34607947e-01 -1.17686771e-01
-2.99023986e-01 3.51786107e-01 5.34391582e-01 -4.17921185e-01
5.74147224e-01 -2.38798022e+00 -3.28926146e-02 3.69964540e-01
2.76228845e-01 -7.60291740e-02 -4.81762648e-01 5.85599184e-01
1.65406093e-02 -2.81090200e-01 -2.40346361e-02 -7.81063378e-01
9.05248970e-02 -7.51040205e-02 -9.04501379e-01 5.27475953e-01
-1.69759572e-01 3.83361429e-01 -1.12259614e+00 -1.35412559e-01
5.07281542e-01 1.02965248e+00 -7.07308173e-01 5.59906781e-01
5.72672114e-04 8.59455645e-01 -5.18426657e-01 2.82182544e-01
4.35197622e-01 1.86230615e-01 -3.78687739e-01 -2.83565768e-03
-1.25111595e-01 4.30162698e-01 -1.25339973e+00 1.94355762e+00
-1.06257975e+00 8.04756701e-01 4.48892862e-01 -8.06453288e-01
1.12991822e+00 2.92635113e-01 3.48469794e-01 -6.56172693e-01
-4.87956256e-02 1.14181086e-01 -2.35388920e-01 -5.50379455e-01
6.34158671e-01 -8.57098177e-02 -5.49527109e-02 3.39340448e-01
-4.89068255e-02 -5.42077124e-01 -6.60658777e-01 2.36954361e-01
1.37688339e+00 1.56909704e-01 -6.24379478e-02 3.89841586e-01
1.94680944e-01 -3.93482745e-01 3.23638618e-01 1.21606302e+00
7.65112787e-02 9.91196871e-01 -4.41777140e-01 -1.09157681e-01
-7.74209976e-01 -1.67376554e+00 2.10282952e-02 1.30192554e+00
2.08191618e-01 -1.70518294e-01 -3.73247355e-01 -2.61217654e-02
-1.08722538e-01 1.25402331e+00 -6.66812122e-01 -1.30365640e-01
-4.92665321e-01 -8.80161598e-02 7.11413383e-01 6.79595411e-01
-3.50127667e-02 -1.17624879e+00 -7.94722140e-01 3.58682871e-01
-3.04392576e-01 -1.36313343e+00 -3.61083657e-01 1.50346890e-01
-2.62178123e-01 -3.75493526e-01 -4.81762648e-01 -5.77538013e-01
2.19388172e-01 6.65271044e-01 1.13215566e+00 -2.53449917e-01
-6.34620726e-01 1.37807715e+00 -4.66938496e-01 -6.08980894e-01
-4.08693135e-01 -7.11453378e-01 2.08109736e-01 2.03032717e-02
-2.36049101e-01 -8.70580494e-01 -5.14213383e-01 2.04878047e-01
-4.18788224e-01 -2.63319910e-01 2.10094109e-01 3.32106233e-01
5.67617714e-01 -3.97442579e-01 5.12468755e-01 -4.04578269e-01
3.97310674e-01 -4.41525161e-01 -2.73391217e-01 -1.93377119e-02
1.73721448e-01 -1.55531377e-01 6.43445969e-01 -5.52183747e-01
-1.17611361e+00 1.11084618e-01 -4.16829884e-01 -7.18231857e-01
-4.93320137e-01 7.04552308e-02 -8.79554451e-02 -1.21245511e-01
8.25921595e-01 3.74001175e-01 -3.80908310e-01 -3.63132030e-01
5.60668826e-01 7.20933318e-01 9.02575672e-01 -5.73833466e-01
1.04186058e+00 5.59252799e-01 -1.49213955e-01 -1.34271491e+00
-7.71147728e-01 -6.15444779e-01 -1.27245024e-01 -3.65152478e-01
6.68428183e-01 -9.77414727e-01 -6.48872733e-01 8.86293575e-02
-1.07536304e+00 -7.16036558e-01 -4.15883720e-01 8.70085895e-01
-7.87660897e-01 4.01504375e-02 -3.98692548e-01 -1.33119702e+00
2.71379054e-02 -1.04115915e+00 1.45224333e+00 -5.01098707e-02
-1.38673648e-01 -8.58033180e-01 6.88633084e-01 2.80320704e-01
5.44353247e-01 -5.70649691e-02 3.35312724e-01 -4.46228713e-01
-6.75911665e-01 -1.13533316e-02 1.83229357e-01 9.39633921e-02
5.66900149e-02 -2.12747440e-01 -1.62664402e+00 -1.16049491e-01
5.62991062e-03 -2.84941822e-01 6.51287377e-01 6.55775607e-01
1.16501141e+00 3.26571008e-03 -3.11618149e-01 7.12307870e-01
1.22097373e+00 3.94020170e-01 4.69236493e-01 -5.02257608e-02
8.00115705e-01 4.31166351e-01 5.61918437e-01 5.81633866e-01
4.58359063e-01 8.12826633e-01 6.51608527e-01 -1.86934635e-01
-9.06853676e-02 -4.40007001e-01 3.57329458e-01 7.70444751e-01
1.52605429e-01 -4.86905992e-01 -6.56962335e-01 6.85686886e-01
-1.41718018e+00 -9.89256859e-01 2.54314661e-01 2.26170707e+00
2.89070725e-01 -2.80503720e-01 -2.84707546e-01 -1.57864079e-01
3.74604732e-01 4.01872128e-01 -4.81599033e-01 -4.46780115e-01
3.63507628e-01 4.66752529e-01 1.23681881e-01 9.92466271e-01
-7.52778530e-01 6.87772334e-01 6.25109816e+00 2.41011590e-01
-9.84034956e-01 3.54593247e-02 1.27744123e-01 -1.81716740e-01
-7.71807373e-01 -2.84764737e-01 -4.27555889e-01 -2.33052418e-01
9.96436059e-01 1.62223667e-01 9.06156838e-01 6.48765266e-01
3.62825543e-01 2.03157082e-01 -1.28523183e+00 9.71035063e-01
4.56659228e-01 -1.09778476e+00 -5.62606573e-01 1.33901858e-03
1.90275177e-01 3.05892378e-01 4.10773546e-01 3.10933053e-01
6.76988900e-01 -1.12382793e+00 1.14970005e+00 5.97472429e-01
7.20889390e-01 -4.31764871e-01 6.82834163e-02 2.71613121e-01
-1.50463414e+00 -1.12784907e-01 -3.24666083e-01 4.59714867e-02
4.18544799e-01 2.55006373e-01 -1.24571240e+00 2.81533301e-01
1.10789335e+00 3.04377705e-01 -1.16880588e-01 1.14302230e+00
-1.01673104e-01 7.22893000e-01 -5.30876338e-01 5.18729072e-03
7.20091686e-02 1.41945645e-01 1.02011323e+00 1.28067660e+00
6.91996098e-01 3.45070124e-01 3.24458688e-01 8.39616239e-01
1.30914047e-01 -3.83319780e-02 -1.11289895e+00 4.20271218e-01
8.13244641e-01 1.08012986e+00 -2.61912972e-01 1.08523570e-01
-1.59655154e-01 7.69470334e-01 9.09839943e-02 7.89382696e-01
-9.23512638e-01 -3.96305025e-01 8.43907416e-01 -1.93981603e-02
5.34460723e-01 -5.10363400e-01 1.65093228e-01 -6.99468970e-01
-1.75488532e-01 -6.39156044e-01 -8.12674165e-02 -1.42767859e+00
-1.08629668e+00 8.19082916e-01 -1.03102401e-01 -1.27955782e+00
-3.23099703e-01 -2.53666759e-01 -7.50487506e-01 7.78655052e-01
-1.28269613e+00 -1.03082049e+00 -5.30477285e-01 5.18465042e-01
7.29553342e-01 1.29099473e-01 1.05107319e+00 1.91657588e-01
-1.96656082e-02 3.57613176e-01 7.63695687e-02 2.91474834e-02
6.15385950e-01 -1.21243179e+00 7.63525546e-01 7.71895885e-01
7.83266842e-01 5.35886824e-01 1.23559117e+00 -2.12080717e-01
-1.51284695e+00 -9.29806054e-01 2.06513077e-01 -5.74586987e-01
6.03437901e-01 -6.93290293e-01 -8.91268194e-01 8.74977469e-01
-2.43647899e-02 3.63136470e-01 8.66108835e-01 1.23329401e-01
-6.09606445e-01 -2.09219530e-01 -9.51724887e-01 6.40867949e-01
1.21583378e+00 -6.77781761e-01 -5.83974421e-01 2.09726304e-01
1.08652616e+00 -5.66982806e-01 -4.78900582e-01 1.41130581e-01
6.33385360e-01 -9.37901616e-01 1.55183351e+00 -1.10478237e-01
-1.15662394e-02 -3.98075968e-01 -8.24201167e-01 -1.69635832e+00
-4.30234641e-01 -5.81515074e-01 2.43839826e-02 1.04158318e+00
2.60359764e-01 -5.87255418e-01 3.94734830e-01 2.53793538e-01
-7.28377879e-01 -2.03807458e-01 -9.16432083e-01 -6.27293825e-01
-2.78183103e-01 -1.17128265e+00 4.28564429e-01 4.37176496e-01
-3.83211464e-01 3.71489555e-01 -4.27622944e-01 9.18362319e-01
9.09385264e-01 -6.26492575e-02 9.71571624e-01 -9.66405272e-01
-7.99675763e-01 -6.98014861e-03 -3.73966604e-01 -1.38136280e+00
1.95453078e-01 -4.02197480e-01 9.35272038e-01 -1.74483502e+00
-4.83138770e-01 -7.90231824e-01 -3.17777008e-01 9.37186182e-02
1.79150641e-01 2.38645613e-01 2.19142109e-01 -1.98938742e-01
-5.61470449e-01 8.83117259e-01 9.26514864e-01 6.20668679e-02
-4.70737100e-01 2.45088622e-01 -6.41174436e-01 6.13234758e-01
4.67530698e-01 -3.86439294e-01 -6.71026468e-01 -8.06463540e-01
7.04174936e-02 4.51278746e-01 6.81422055e-01 -1.52769053e+00
3.59386712e-01 -1.60591267e-02 3.07440102e-01 -5.96028090e-01
1.12456918e+00 -7.07044184e-01 1.60039097e-01 -1.16027139e-01
-6.50966644e-01 -1.63234964e-01 2.93722570e-01 9.41182077e-01
1.28673598e-01 -1.54791251e-02 4.36633050e-01 -2.04646811e-01
-6.06171727e-01 1.68496028e-01 -5.44112325e-01 2.28035405e-01
5.35557210e-01 -1.32162377e-01 -2.00977415e-01 -9.33689237e-01
-6.73795283e-01 -3.07072029e-02 1.15050144e-01 5.35128057e-01
1.08590126e+00 -1.03633046e+00 -7.64216125e-01 3.13107312e-01
1.59004912e-01 1.28404289e-01 5.07309914e-01 3.05338740e-01
-2.09604397e-01 1.30047113e-01 2.70192116e-01 -7.79715002e-01
-1.18061280e+00 3.05895478e-01 6.01822615e-01 3.19992065e-01
-1.02277493e+00 1.19517207e+00 8.17202330e-01 -7.93057263e-01
4.16271985e-01 -2.51219749e-01 -1.09521493e-01 -6.97367609e-01
6.71488822e-01 1.50292873e-01 -7.90345967e-02 -8.19869161e-01
-3.72526914e-01 7.21668839e-01 3.91371638e-01 -8.31942797e-01
1.39676547e+00 -3.57939363e-01 5.43123782e-01 9.38816071e-01
1.22071648e+00 6.04426682e-01 -1.37309515e+00 -3.94015819e-01
-6.69744074e-01 -7.74982035e-01 3.59698802e-01 -7.25041151e-01
-6.43119931e-01 1.06723094e+00 6.49118900e-01 5.87771758e-02
1.05671525e+00 3.30430984e-01 7.03071892e-01 5.36214590e-01
6.17511153e-01 -5.97966790e-01 4.53663677e-01 5.33273578e-01
1.03960979e+00 -7.95762360e-01 -4.69045281e-01 -2.44789615e-01
-7.49313831e-01 7.82216072e-01 4.25365776e-01 -1.14149652e-01
8.00987422e-01 6.17427528e-01 4.02044713e-01 -4.60718609e-02
-7.51335263e-01 -1.63805068e-01 1.61482900e-01 1.09391677e+00
1.77642792e-01 -2.60643922e-02 1.14525676e+00 3.83437991e-01
-5.45344234e-01 -3.18026394e-01 6.82117939e-01 7.30764091e-01
-6.58556402e-01 -4.20549661e-01 -6.66117609e-01 2.46840939e-02
-2.30326444e-01 -2.30510831e-01 -1.99430433e-05 3.60758007e-01
-3.32186788e-01 1.33160293e+00 1.12176798e-01 -6.30796731e-01
5.76127172e-01 -2.55588979e-01 6.30052865e-01 -7.60951698e-01
-3.12467098e-01 4.23489034e-01 5.97032085e-02 -7.35395312e-01
-1.78511381e-01 -7.02876508e-01 -1.35887241e+00 1.43534154e-01
-1.95975021e-01 2.89420396e-01 9.04419601e-01 7.32602417e-01
3.55862975e-01 1.03404629e+00 1.07139802e+00 -1.36403155e+00
-4.81049508e-01 -7.78727353e-01 -5.64770579e-01 1.22861844e-03
8.50918472e-01 -6.09360337e-01 -5.55018544e-01 -1.02836259e-01] | [15.032825469970703, 5.496488571166992] |
39b33544-11d1-49ac-9e2d-4d52f0468fe2 | jointly-visual-and-semantic-aware-graph | 2303.01046 | null | https://arxiv.org/abs/2303.01046v2 | https://arxiv.org/pdf/2303.01046v2.pdf | Jointly Visual- and Semantic-Aware Graph Memory Networks for Temporal Sentence Localization in Videos | Temporal sentence localization in videos (TSLV) aims to retrieve the most interested segment in an untrimmed video according to a given sentence query. However, almost of existing TSLV approaches suffer from the same limitations: (1) They only focus on either frame-level or object-level visual representation learning and corresponding correlation reasoning, but fail to integrate them both; (2) They neglect to leverage the rich semantic contexts to further benefit the query reasoning. To address these issues, in this paper, we propose a novel Hierarchical Visual- and Semantic-Aware Reasoning Network (HVSARN), which enables both visual- and semantic-aware query reasoning from object-level to frame-level. Specifically, we present a new graph memory mechanism to perform visual-semantic query reasoning: For visual reasoning, we design a visual graph memory to leverage visual information of video; For semantic reasoning, a semantic graph memory is also introduced to explicitly leverage semantic knowledge contained in the classes and attributes of video objects, and perform correlation reasoning in the semantic space. Experiments on three datasets demonstrate that our HVSARN achieves a new state-of-the-art performance. | ['Pan Zhou', 'Daizong Liu'] | 2023-03-02 | null | null | null | null | ['visual-reasoning', 'visual-reasoning'] | ['computer-vision', 'reasoning'] | [-1.93751693e-01 -2.49950871e-01 -3.79452795e-01 -3.90348792e-01
-6.64977133e-01 -3.19465369e-01 4.22832221e-01 1.93567544e-01
-1.38955876e-01 1.98675275e-01 4.42897916e-01 -5.36152394e-03
-5.59673160e-02 -7.61863470e-01 -7.12783635e-01 -4.78089631e-01
2.07338840e-01 8.75101089e-02 8.77712369e-01 -1.14709064e-01
2.59648442e-01 2.76740730e-01 -1.57795846e+00 7.24692702e-01
4.93043065e-01 1.27103245e+00 3.95962894e-01 2.85742074e-01
-2.77659029e-01 1.61707902e+00 -4.31785494e-01 -2.24170819e-01
-2.89857656e-01 -5.87196589e-01 -8.38470876e-01 1.26571298e-01
3.45793307e-01 -3.22668731e-01 -6.32823050e-01 1.20021820e+00
2.63280541e-01 3.73102605e-01 3.98055434e-01 -1.43793499e+00
-8.75929952e-01 3.84484798e-01 -5.10563970e-01 3.73709321e-01
7.54404247e-01 1.15427569e-01 1.03017652e+00 -1.00436950e+00
7.92043746e-01 1.48772967e+00 1.65296525e-01 3.11621487e-01
-6.91876292e-01 -4.29589480e-01 5.27876019e-01 8.21474493e-01
-1.51693118e+00 -2.89801598e-01 1.14936912e+00 -3.58119845e-01
9.39828813e-01 1.66392371e-01 8.67497027e-01 9.35490429e-01
-6.30019754e-02 1.26381481e+00 9.29479122e-01 -9.56404433e-02
2.33547091e-01 -1.16190193e-02 8.57756287e-02 9.70749319e-01
-2.47378320e-01 -1.03795201e-01 -8.38201046e-01 1.48846999e-01
7.37446010e-01 2.94683337e-01 -2.97056913e-01 -6.75529063e-01
-1.25228906e+00 7.60841250e-01 8.10012698e-01 4.60095912e-01
-3.60335320e-01 3.03775966e-01 7.19254792e-01 6.69467226e-02
2.29159772e-01 -3.21296081e-02 -1.66177556e-01 1.17790587e-01
-8.58188868e-01 4.67629768e-02 5.09468734e-01 1.28430915e+00
9.07913446e-01 -1.13079092e-02 -5.99546373e-01 5.37023127e-01
5.03916502e-01 5.80382764e-01 3.68518949e-01 -1.00386739e+00
5.93992054e-01 9.79684949e-01 -2.42663950e-01 -1.41774905e+00
-2.89431006e-01 -1.32251397e-01 -5.90786815e-01 -1.29788101e-01
-1.13918304e-01 4.44071829e-01 -8.37622523e-01 1.50724792e+00
4.11073565e-01 2.67492503e-01 1.03202455e-01 1.27630925e+00
1.45135653e+00 5.93526781e-01 3.38635355e-01 -3.22971165e-01
1.77633429e+00 -1.21061099e+00 -8.29624593e-01 -2.67060578e-01
5.17397285e-01 -5.68066239e-01 1.16012824e+00 -2.63501741e-02
-1.17120337e+00 -7.70908952e-01 -9.05257404e-01 -3.11421961e-01
-5.43549299e-01 -2.29711886e-02 2.97474980e-01 1.12989180e-01
-1.01840806e+00 -7.07068592e-02 -5.48711419e-01 -5.07186532e-01
4.36829031e-01 -1.46372631e-01 -3.23827505e-01 -3.04705590e-01
-1.36433494e+00 6.85825884e-01 5.62631011e-01 1.87859163e-02
-1.09342039e+00 -4.30206031e-01 -1.04280388e+00 1.28578946e-01
9.20611143e-01 -9.13572311e-01 9.29695010e-01 -9.84205902e-01
-9.28800344e-01 7.60780990e-01 -3.37794989e-01 -2.57549495e-01
2.90126830e-01 -1.76190242e-01 -4.72586989e-01 8.55597913e-01
3.17727745e-01 6.07802331e-01 7.67024398e-01 -1.38974392e+00
-5.13235688e-01 -3.97833228e-01 4.39422518e-01 3.83235872e-01
-3.90066028e-01 7.29887560e-02 -1.17852533e+00 -7.90800512e-01
1.36490434e-01 -4.37048137e-01 4.90288399e-02 1.20554395e-01
-2.68488437e-01 -4.23117995e-01 1.34289467e+00 -7.31081486e-01
1.32729495e+00 -2.26647091e+00 2.89884955e-01 -1.42711811e-02
3.19909662e-01 1.69746727e-01 -4.82764393e-02 4.92549479e-01
2.36449748e-01 -1.19413488e-01 1.04362465e-01 -1.82235017e-01
-3.69301923e-02 3.15451741e-01 -3.51221949e-01 2.45835274e-01
5.78781776e-03 1.23019159e+00 -1.21232080e+00 -1.15930629e+00
4.26379412e-01 5.10390401e-01 -4.71409708e-01 3.21131617e-01
-4.47802842e-01 1.62853584e-01 -8.99715960e-01 9.81403947e-01
3.64549905e-01 -4.67384994e-01 1.04692310e-01 -8.40522110e-01
1.97938532e-01 -2.81917572e-01 -8.82417798e-01 2.00859356e+00
-2.13827699e-01 4.27418202e-01 -1.22381017e-01 -1.00500512e+00
8.19163084e-01 1.51172906e-01 5.40254116e-01 -1.02359533e+00
4.32219990e-02 -1.55830547e-01 -5.73327422e-01 -9.20092225e-01
3.59832704e-01 7.54535273e-02 -1.01974517e-01 1.61834523e-01
1.05005883e-01 2.13319793e-01 1.45956323e-01 5.96839070e-01
9.23798025e-01 3.57658118e-01 2.37180859e-01 -5.77187277e-02
9.38062727e-01 6.82075042e-03 5.20876408e-01 5.58822811e-01
-4.12025779e-01 4.84848320e-01 5.64961374e-01 -4.14794803e-01
-6.30085647e-01 -1.17493176e+00 4.49044406e-01 1.31694114e+00
8.29709888e-01 -6.48017466e-01 -6.09046876e-01 -8.78706396e-01
-1.80004016e-01 5.66477895e-01 -5.73203623e-01 -3.62719327e-01
-4.20894563e-01 -1.02109268e-01 1.55227304e-01 8.97615075e-01
9.73348975e-01 -1.20506561e+00 -6.90629125e-01 -1.51235268e-01
-5.26058793e-01 -1.49806499e+00 -6.90778613e-01 -4.39903677e-01
-6.21100485e-01 -1.33481944e+00 -4.75748390e-01 -9.04381573e-01
6.45027697e-01 6.71961427e-01 1.09358764e+00 3.28422189e-01
-2.05243498e-01 8.03545356e-01 -8.02713513e-01 9.75215212e-02
-6.06627055e-02 -3.39225501e-01 -3.18014801e-01 2.70700276e-01
4.61234331e-01 -2.25213185e-01 -8.35072458e-01 3.44976097e-01
-9.04505193e-01 1.72693938e-01 5.38767219e-01 6.09412372e-01
8.01429212e-01 1.45201772e-01 3.46305072e-01 -6.74009264e-01
3.34802955e-01 -3.87837291e-01 -2.88444251e-01 7.08542347e-01
-4.12091613e-01 -9.75707173e-02 4.47636485e-01 -1.77442521e-01
-9.79032159e-01 -1.25678983e-02 1.87077880e-01 -1.16701460e+00
-6.19859667e-03 5.13096392e-01 -3.97872716e-01 1.46187171e-01
8.43351930e-02 6.44069135e-01 -1.77121356e-01 -2.75856256e-01
6.19871020e-01 2.64177084e-01 6.33954942e-01 -5.30341685e-01
6.00395322e-01 6.14036381e-01 1.58342663e-02 -4.94072229e-01
-1.06759870e+00 -6.80068314e-01 -6.71215236e-01 -6.20039105e-01
1.46413791e+00 -1.14212394e+00 -9.47927833e-01 -1.01346467e-02
-1.12883472e+00 -2.34899595e-02 -1.81203470e-01 3.66063833e-01
-8.78396153e-01 6.29270315e-01 -4.17127818e-01 -6.15215421e-01
-2.96295822e-01 -1.28765392e+00 1.34469843e+00 1.39426693e-01
1.89475030e-01 -9.42324936e-01 -4.45798844e-01 6.79986238e-01
1.34954572e-01 2.72389442e-01 9.16793287e-01 -4.96267349e-01
-9.18328524e-01 7.29708523e-02 -7.89827704e-01 1.37793794e-01
-1.79016799e-01 -2.63664156e-01 -8.14418435e-01 -2.90986240e-01
-4.11370881e-02 -3.10288519e-01 8.95018220e-01 1.71083555e-01
1.41185343e+00 -2.54157394e-01 -4.32224005e-01 5.18198907e-01
1.44650102e+00 1.53665096e-01 4.95519519e-01 2.56875783e-01
9.93816733e-01 5.49204230e-01 9.31164920e-01 3.24227601e-01
7.35185862e-01 8.02466452e-01 6.40422821e-01 -1.55526116e-01
-4.78806525e-01 -5.98155439e-01 4.22214836e-01 8.81427169e-01
-8.55508372e-02 -9.03432593e-02 -6.72282636e-01 4.62955594e-01
-2.40786481e+00 -1.12860692e+00 4.67934236e-02 1.69822824e+00
3.58969986e-01 -3.15761417e-02 2.49970317e-01 -2.24452555e-01
7.28413165e-01 5.01537502e-01 -6.00551546e-01 1.12292528e-01
-1.66190058e-01 -5.46078920e-01 1.69964194e-01 1.44520909e-01
-1.02670848e+00 1.13362026e+00 5.45863628e+00 9.21147466e-01
-7.57758915e-01 3.72096151e-01 3.29510957e-01 -6.26104027e-02
-2.69673318e-01 8.56504515e-02 -4.44830507e-01 2.69786894e-01
4.17442113e-01 -1.19754791e-01 2.42037281e-01 9.62554395e-01
5.30831255e-02 -1.98349450e-03 -1.01603580e+00 1.39995754e+00
5.81093073e-01 -1.40139055e+00 4.87963378e-01 -3.44616503e-01
3.09675753e-01 -3.77834260e-01 -1.20771326e-01 4.77750361e-01
-2.98432149e-02 -7.15524197e-01 9.30779278e-01 9.81701493e-01
5.88248789e-01 -8.37149024e-01 6.34133220e-01 1.34360969e-01
-1.77440512e+00 -1.27003089e-01 -3.04474264e-01 3.57311785e-01
2.15875164e-01 1.43054992e-01 -4.73956168e-01 9.18643057e-01
1.01948571e+00 1.21125293e+00 -8.21507931e-01 7.32803404e-01
-4.05590177e-01 1.73422322e-01 3.33947688e-01 -1.09248921e-01
2.74861455e-01 1.11249879e-01 3.74910623e-01 1.18163848e+00
1.48247018e-01 1.37724772e-01 5.92249930e-01 8.65898788e-01
5.25494441e-02 2.03308463e-01 -5.01924515e-01 1.12514608e-01
5.18170536e-01 1.05625129e+00 -8.99417043e-01 -5.21252096e-01
-7.90224910e-01 1.12301087e+00 3.85437608e-01 6.91763937e-01
-1.03659725e+00 -3.10178816e-01 4.51465160e-01 6.18073791e-02
5.86421072e-01 -1.76232025e-01 1.67571500e-01 -1.33889318e+00
1.46469817e-01 -6.33798361e-01 7.37267315e-01 -1.32353818e+00
-1.26210165e+00 4.98995036e-01 2.32097596e-01 -1.32186246e+00
5.49593382e-03 -4.41926360e-01 -2.50024647e-01 4.52969193e-01
-1.58915734e+00 -1.54747009e+00 -5.84042490e-01 1.19280636e+00
8.27449620e-01 -2.24317566e-01 4.34560388e-01 2.61856526e-01
-3.05383950e-01 2.99083173e-01 -4.79874581e-01 4.92155135e-01
4.84785974e-01 -8.78513873e-01 -1.71368405e-01 7.94766843e-01
3.19908410e-01 5.47375917e-01 4.94476527e-01 -6.26637697e-01
-1.84680808e+00 -1.27630234e+00 6.46477938e-01 -4.28893656e-01
6.70018971e-01 -2.41761386e-01 -8.98407638e-01 6.62487149e-01
1.39810830e-01 3.43307674e-01 5.01475573e-01 -1.12769701e-01
-6.49313092e-01 -2.73860008e-01 -7.87109852e-01 5.71795106e-01
1.21183169e+00 -1.05420351e+00 -7.82468915e-01 3.47193182e-01
1.16605401e+00 -3.19601893e-01 -9.18878734e-01 4.12629783e-01
4.05186027e-01 -1.05613768e+00 1.38315570e+00 -5.31155109e-01
3.52716148e-01 -7.22369075e-01 -4.96216476e-01 -6.85948491e-01
-3.02892178e-01 -9.70083922e-02 -4.82862264e-01 1.36354268e+00
-1.50944844e-01 -2.10264012e-01 5.04121304e-01 1.24678761e-01
-1.38930693e-01 -8.81197095e-01 -8.44425857e-01 -7.11915374e-01
-6.29088581e-01 -5.83458543e-01 6.45493865e-01 8.68623674e-01
-2.25490704e-02 4.37025845e-01 -4.49863225e-01 3.38168263e-01
5.11639357e-01 5.29601157e-01 6.53319716e-01 -8.30629528e-01
-1.11399405e-01 -4.80051875e-01 -8.30070019e-01 -1.07270801e+00
2.44070306e-01 -8.38315725e-01 -9.57300887e-02 -1.92024577e+00
5.72006762e-01 1.04904838e-01 -5.23008585e-01 5.16383111e-01
-1.76838785e-01 2.80296385e-01 6.40934229e-01 3.25083107e-01
-1.56930637e+00 6.31094635e-01 1.47667933e+00 -3.15029144e-01
1.66653782e-01 -4.73422676e-01 -6.35233283e-01 6.47282660e-01
1.79693550e-01 -1.82919085e-01 -8.53601635e-01 -5.05578876e-01
1.95893437e-01 4.60564435e-01 8.48020971e-01 -1.08168042e+00
5.85702717e-01 -2.41382986e-01 4.50061560e-01 -9.21710253e-01
4.00480390e-01 -9.41295743e-01 -1.46439344e-01 3.81727517e-01
-3.59398276e-01 6.94747716e-02 -1.29035100e-01 9.97153044e-01
-6.29802704e-01 1.48573354e-01 4.84400868e-01 -1.63682982e-01
-1.43695199e+00 5.37319362e-01 -6.61873519e-02 1.79247126e-01
1.28998721e+00 -1.92890167e-01 -5.14529407e-01 -3.38902354e-01
-7.24519908e-01 5.13249993e-01 4.71166551e-01 7.00483203e-01
1.03584838e+00 -1.60541451e+00 -3.14316720e-01 7.95153901e-03
6.14300013e-01 -1.47741020e-01 5.94249725e-01 9.15292501e-01
-2.91873276e-01 4.28676546e-01 -4.89668399e-02 -7.39346445e-01
-1.27427626e+00 1.31309032e+00 1.53103232e-01 1.93623584e-02
-7.59582400e-01 7.01985300e-01 6.30745173e-01 1.62182134e-02
2.95537204e-01 -1.93640098e-01 -4.16657627e-01 1.35955170e-01
5.95053911e-01 1.40392736e-01 -2.19464347e-01 -1.08243299e+00
-6.42492652e-01 8.46588135e-01 2.02969968e-01 2.00697601e-01
9.03598011e-01 -3.73614818e-01 -1.39523998e-01 5.36629438e-01
1.33897448e+00 -3.18081975e-01 -1.14246035e+00 -5.64742386e-01
-1.10505469e-01 -6.16144419e-01 6.62157610e-02 -5.89587688e-01
-1.29640317e+00 9.58120227e-01 3.76208693e-01 1.03721119e-01
1.51031125e+00 3.97586465e-01 7.99706638e-01 2.65133828e-01
3.79706919e-01 -1.07827556e+00 6.02226317e-01 2.71187097e-01
8.29592764e-01 -1.11243868e+00 2.05938280e-01 -4.95811343e-01
-9.07113314e-01 1.18695521e+00 7.73846865e-01 9.74547416e-02
5.53334475e-01 -1.65358245e-01 -6.15861565e-02 -6.19032145e-01
-7.68311620e-01 -5.61547518e-01 5.86305737e-01 5.51921070e-01
1.96380734e-01 -1.51049390e-01 -6.37255535e-02 5.83770633e-01
4.10853714e-01 -3.60124223e-02 -2.63777841e-02 9.18126822e-01
-6.23326361e-01 -6.70133650e-01 -1.88981354e-01 1.07811451e-01
9.33956169e-03 -2.60291304e-02 -2.74267375e-01 7.56657839e-01
1.11967459e-01 1.01310623e+00 1.48200560e-02 -4.75161612e-01
4.22255546e-01 -1.09783806e-01 2.34913468e-01 -4.59376991e-01
-3.62712294e-01 1.76706463e-01 -1.35418743e-01 -1.03654337e+00
-9.33322012e-01 -4.25084084e-01 -1.55299461e+00 7.26193264e-02
4.58680764e-02 1.04561523e-01 2.43971303e-01 1.08318675e+00
3.77283633e-01 7.56510317e-01 3.86423796e-01 -4.24709857e-01
-2.11306475e-03 -5.60797751e-01 -5.79705060e-01 7.09737718e-01
2.39172876e-01 -7.77439177e-01 -8.29367936e-02 1.62102193e-01] | [10.130155563354492, 0.9351569414138794] |
f8348a40-d232-489f-8f86-61a9daca5359 | a-digital-score-of-tumour-associated-stroma | 2104.12862 | null | https://arxiv.org/abs/2104.12862v1 | https://arxiv.org/pdf/2104.12862v1.pdf | A digital score of tumour-associated stroma infiltrating lymphocytes predicts survival in head and neck squamous cell carcinoma | The infiltration of T-lymphocytes in the stroma and tumour is an indication of an effective immune response against the tumour, resulting in better survival. In this study, our aim is to explore the prognostic significance of tumour-associated stroma infiltrating lymphocytes (TASILs) in head and neck squamous cell carcinoma (HNSCC) through an AI based automated method. A deep learning based automated method was employed to segment tumour, stroma and lymphocytes in digitally scanned whole slide images of HNSCC tissue slides. The spatial patterns of lymphocytes and tumour-associated stroma were digitally quantified to compute the TASIL-score. Finally, prognostic significance of the TASIL-score for disease-specific and disease-free survival was investigated with the Cox proportional hazard analysis. Three different cohorts of Haematoxylin & Eosin (H&E) stained tissue slides of HNSCC cases (n=537 in total) were studied, including publicly available TCGA head and neck cancer cases. The TASIL-score carries prognostic significance (p=0.002) for disease-specific survival of HNSCC patients. The TASIL-score also shows a better separation between low- and high-risk patients as compared to the manual TIL scoring by pathologists for both disease-specific and disease-free survival. A positive correlation of TASIL-score with molecular estimates of CD8+ T cells was also found, which is in line with existing findings. To the best of our knowledge, this is the first study to automate the quantification of TASIL from routine H&E slides of head and neck cancer. Our TASIL-score based findings are aligned with the clinical knowledge with the added advantages of objectivity, reproducibility and strong prognostic value. A comprehensive evaluation on large multicentric cohorts is required before the proposed digital score can be adopted in clinical practice. | ['Nasir Rajpoot', 'Syed Ali Khurram', 'Hisham Mehanna', 'Max Robinson', 'Neil Sharma', 'Paul Nankivell', 'Jill Brooks', 'Nikolaos Batis', 'Asif Loya', 'Sajid Mushtaq', 'Arif Jamshed', 'Mariam Hassan', 'Shan E Ahmed Raza', 'Muhammad Shaban'] | 2021-04-16 | null | null | null | null | ['clinical-knowledge'] | ['miscellaneous'] | [ 4.30173241e-02 -9.29456502e-02 -5.51844537e-01 2.48552665e-01
-1.21938622e+00 -3.71011704e-01 4.16221708e-01 7.08364844e-01
-8.81544888e-01 6.42447948e-01 4.47106421e-01 -4.29766178e-01
-2.40631312e-01 -8.84572327e-01 1.52095512e-01 -1.52854574e+00
-7.10571604e-03 9.65211272e-01 1.06451772e-01 -1.37325019e-01
-1.32756904e-01 8.28409076e-01 -1.04250538e+00 3.80323768e-01
4.75450814e-01 8.83171499e-01 3.47084552e-01 9.16460752e-01
-2.58448780e-01 3.80087644e-01 -1.61848947e-01 6.50442764e-02
-3.01888525e-01 -2.67631084e-01 -5.03401458e-01 -2.72483915e-01
1.47416815e-01 9.03755501e-02 -6.24102401e-03 5.05245745e-01
7.03529119e-01 -5.16345680e-01 9.87931371e-01 -9.12344337e-01
2.45524332e-01 7.50024337e-03 -4.90704030e-01 1.11571424e-01
4.98932041e-02 3.91170412e-01 8.76347363e-01 -7.33407319e-01
9.34434891e-01 5.57813883e-01 8.40041041e-01 1.69389516e-01
-1.24623883e+00 -4.04039025e-01 -5.68340242e-01 2.50488818e-01
-1.50022113e+00 1.37394622e-01 2.23592117e-01 -6.49083912e-01
7.38947213e-01 5.86247742e-01 1.09477532e+00 4.53182101e-01
7.98953474e-01 6.23339713e-01 1.38803446e+00 -5.24660766e-01
4.87954289e-01 2.22038612e-01 1.67807877e-01 6.82101786e-01
2.41425857e-01 8.53768289e-02 -3.06577623e-01 -1.92297578e-01
3.96037102e-01 3.25291455e-01 -2.52909213e-01 1.41602149e-02
-9.66812491e-01 9.49986637e-01 5.81158102e-01 9.73223448e-01
-3.20423692e-01 -6.75506238e-03 9.59245741e-01 1.47551909e-01
3.90538812e-01 -4.43042666e-01 -7.53583014e-02 1.59131959e-01
-9.72078621e-01 -3.46904725e-01 4.34067130e-01 -1.32208988e-01
2.21098587e-01 -5.43219805e-01 -3.71800573e-03 6.33113146e-01
3.77901018e-01 6.06419623e-01 1.03337359e+00 -2.34086081e-01
-5.11773944e-01 8.60231936e-01 -5.14208734e-01 -5.41880250e-01
-9.02657866e-01 -4.84840214e-01 -1.16114759e+00 2.89714843e-01
7.80221403e-01 4.14788634e-01 -6.01417303e-01 1.16489315e+00
3.25279117e-01 -3.07948768e-01 -7.80111551e-02 8.84619772e-01
5.75188756e-01 1.87824517e-01 2.55227029e-01 -2.03456700e-01
1.90579045e+00 -4.73005474e-01 -6.64734900e-01 2.05901444e-01
1.48149776e+00 -6.75430477e-01 1.06336725e+00 1.18177332e-01
-5.53478479e-01 3.82570480e-03 -3.82988304e-01 -7.27996882e-03
-5.18217266e-01 2.87198305e-01 7.93388546e-01 7.21765399e-01
-1.26640737e+00 3.95582709e-03 -8.17563117e-01 -1.01174331e+00
6.19393885e-01 5.97461104e-01 -8.84307206e-01 1.66421473e-01
-8.59650373e-01 8.87423217e-01 1.29602820e-01 -1.74690723e-01
-4.86991882e-01 -7.48498738e-01 -4.57375586e-01 -3.67347956e-01
-1.76903546e-01 -8.08782279e-01 9.35026288e-01 -8.92600417e-01
-1.43384397e+00 1.41049230e+00 -5.86469829e-01 -1.93721011e-01
1.57426745e-01 8.23588610e-01 -2.08869979e-01 3.25148642e-01
-1.28812939e-01 2.28153884e-01 -1.16490608e-03 -8.71970057e-01
-8.30822945e-01 -7.85582364e-01 -8.62698555e-01 4.78877034e-03
-2.40718722e-01 -2.53953367e-01 -1.49544179e-01 -4.45796341e-01
-1.26243770e-01 -1.11058319e+00 -3.43865484e-01 4.87996906e-01
-9.59501415e-03 -2.99415797e-01 6.08999372e-01 -6.73577487e-01
9.85289693e-01 -1.94140983e+00 8.75964910e-02 5.86940527e-01
2.02972248e-01 2.76677012e-01 -1.16824284e-02 2.57477194e-01
1.92319438e-01 3.52106035e-01 3.11813746e-02 -2.09369570e-01
-3.78902033e-02 1.62004188e-01 4.65380967e-01 1.13022578e+00
-2.51790166e-01 1.23660004e+00 -6.05625689e-01 -9.17853236e-01
2.71668255e-01 6.41180277e-01 -1.25604644e-01 -1.47061929e-01
-7.22067952e-02 2.15519071e-01 -1.69281065e-01 7.97458768e-01
5.33327818e-01 -8.95786509e-02 4.74263012e-01 -1.30055085e-01
-7.78211281e-02 -9.65136886e-02 -2.99953431e-01 1.17562246e+00
-5.06251991e-01 1.02333581e+00 3.79636794e-01 -1.81667075e-01
5.21080375e-01 3.91698271e-01 3.06902826e-01 -9.23147082e-01
6.22246444e-01 5.15680313e-01 5.30191548e-02 -3.33147317e-01
-6.86447620e-02 -1.00640917e+00 -4.76094820e-02 5.32063246e-01
-2.35488996e-01 -1.78756341e-01 6.82272241e-02 -4.36478388e-03
1.05120242e+00 -6.24707520e-01 7.93759763e-01 -4.30385590e-01
9.01718140e-01 2.69427389e-01 1.26192927e-01 1.86112225e-01
-3.83793980e-01 2.63246357e-01 7.68709004e-01 -2.98206031e-01
-4.82059687e-01 -9.20833647e-01 -5.37351191e-01 6.89168155e-01
-4.71972227e-01 1.90187559e-01 -3.61421615e-01 -5.18844903e-01
9.80758071e-02 4.68406230e-01 -1.00682998e+00 3.35057586e-01
-3.57154846e-01 -1.02248490e+00 6.72842145e-01 3.39585483e-01
-6.13579974e-02 -7.51375616e-01 -2.77403831e-01 1.88233361e-01
-5.59201390e-02 -4.73355532e-01 -1.19884782e-01 4.48480010e-01
-9.15048242e-01 -1.25783920e+00 -9.51750696e-01 -1.01665115e+00
8.99779916e-01 -3.31279114e-02 7.22858250e-01 4.35924411e-01
-6.08250022e-01 8.55849385e-02 -1.02253288e-01 -4.29684430e-01
-7.65943766e-01 3.68838608e-02 -3.20455372e-01 -2.58529097e-01
6.62203372e-01 -1.45470157e-01 -7.55569637e-01 3.85677040e-01
-8.52253258e-01 -5.87292425e-02 1.20333922e+00 8.44698668e-01
1.15890431e+00 1.93310857e-01 2.28269473e-01 -7.49522388e-01
1.48407787e-01 -4.31138247e-01 -2.74035633e-01 -9.20908451e-02
-3.97511691e-01 -4.39041376e-01 6.68950737e-01 -2.61599626e-02
-7.90670991e-01 -1.41825840e-01 -3.77525002e-01 6.33266628e-01
-3.76526743e-01 7.23962784e-01 1.35888845e-01 -2.75897980e-01
4.24424708e-01 1.53336182e-01 6.81823790e-01 5.55252582e-02
-3.13655168e-01 7.60979295e-01 2.25765064e-01 1.57311857e-01
6.28766060e-01 1.05237031e+00 6.36577845e-01 -1.10541046e+00
-4.61945742e-01 -1.01950955e+00 -5.37964284e-01 -2.27214962e-01
7.65041053e-01 -8.11141074e-01 -1.06685042e+00 6.85903668e-01
-4.86033499e-01 -8.46257091e-01 -1.28075153e-01 5.14963925e-01
-5.04799604e-01 1.82870060e-01 -1.01951849e+00 -5.34472406e-01
-8.23190510e-01 -8.75395417e-01 1.22012103e+00 1.82064429e-01
-6.26453876e-01 -1.56299603e+00 9.30391550e-01 3.10385585e-01
7.59777904e-01 4.71600264e-01 1.23115575e+00 -6.61433458e-01
-4.15411405e-02 -8.22177410e-01 -2.37721965e-01 -2.72118777e-01
-1.37924380e-03 2.40310237e-01 -8.00127089e-01 -2.72585422e-01
-3.10248256e-01 -1.82981938e-02 8.33257675e-01 4.00544554e-01
5.50134361e-01 -1.31459758e-01 -9.24518108e-01 3.50194305e-01
1.91209459e+00 8.81934389e-02 8.34499538e-01 7.87455380e-01
2.84358203e-01 8.21912944e-01 5.21768451e-01 2.59084284e-01
2.12262467e-01 5.64671397e-01 3.04870963e-01 -4.65770364e-01
-2.89257735e-01 6.85973186e-03 2.43923515e-01 5.32744527e-01
3.20139825e-02 -5.58426797e-01 -1.15011382e+00 6.28800273e-01
-9.00633872e-01 -6.87659681e-01 -5.62197506e-01 1.97170365e+00
8.02533865e-01 1.66720554e-01 1.04555590e-02 4.75282162e-01
4.17202115e-01 -1.25853226e-01 -1.92425549e-01 -3.20988566e-01
-2.46241868e-01 4.07362044e-01 6.36242092e-01 5.86987495e-01
-5.02697468e-01 3.74782026e-01 5.84975815e+00 1.12929630e+00
-1.53293800e+00 1.24907464e-01 8.35716009e-01 -3.65306497e-01
-2.91319966e-01 -1.58999078e-02 -5.99295676e-01 2.31742963e-01
1.03954077e+00 -5.94588965e-02 -3.57009858e-01 1.96839854e-01
5.10097146e-01 -6.66330040e-01 -6.88962400e-01 4.70929384e-01
3.18545587e-02 -1.45535481e+00 -3.00995111e-01 8.09105933e-01
3.56315434e-01 9.73679125e-02 1.39737830e-01 1.78122371e-01
-1.05949841e-01 -9.71015871e-01 8.84786025e-02 7.13767350e-01
1.40229499e+00 -5.62111795e-01 1.74841642e+00 1.57413512e-01
-9.48037446e-01 3.01633209e-01 1.02969550e-01 2.99785703e-01
1.18399411e-01 7.51201928e-01 -1.62583089e+00 1.39725609e-02
1.77541837e-01 3.02048802e-01 -6.16510510e-01 7.57935882e-01
1.80467352e-01 6.95177615e-01 -4.80309218e-01 -4.97132659e-01
1.58406436e-01 3.93864363e-01 1.48827344e-01 1.38543999e+00
2.82112211e-01 1.15312621e-01 -6.25883520e-01 1.75117001e-01
5.38027108e-01 4.24292207e-01 1.16783697e-02 -9.73559767e-02
9.62190181e-02 1.71191108e+00 -1.13313520e+00 -4.54174057e-02
-2.53043592e-01 5.68826258e-01 -9.16260332e-02 -4.00305763e-02
-3.52317065e-01 -1.17194146e-01 9.05078799e-02 8.96535397e-01
-1.16875738e-01 2.13745698e-01 -4.91883814e-01 -2.50178039e-01
-7.78253853e-01 -5.48487484e-01 7.30805278e-01 -3.31481725e-01
-1.21542668e+00 4.21291709e-01 -4.76275712e-01 -1.09765720e+00
1.41636908e-01 -8.91324103e-01 -1.00868559e+00 7.30669916e-01
-1.58118451e+00 -1.21495330e+00 -4.97862428e-01 1.59813523e-01
-2.51447074e-02 7.62506649e-02 1.33061290e+00 -4.65424806e-01
-6.22085333e-01 7.14811862e-01 2.95580655e-01 -1.21852748e-01
6.46555901e-01 -1.46546054e+00 -5.90692639e-01 -1.71500206e-01
-9.97780919e-01 3.01465601e-01 5.82957685e-01 -5.26097655e-01
-1.35637021e+00 -1.02070761e+00 1.02607000e+00 -1.67767763e-01
9.79826152e-01 -5.91175631e-02 -5.72394848e-01 2.89054036e-01
9.99944508e-02 -5.97070977e-02 1.80380642e+00 -2.93975890e-01
1.48661822e-01 -4.56562191e-02 -1.46331573e+00 6.92324996e-01
1.16426304e-01 -5.06506264e-01 -6.63084490e-03 3.58752578e-01
-1.10108346e-01 -1.39909893e-01 -1.45780158e+00 2.65891820e-01
8.98505807e-01 -1.08619082e+00 6.60202861e-01 3.25215608e-01
-1.13567308e-01 -3.95214379e-01 7.69084506e-03 -9.38934743e-01
-4.58806962e-01 9.51195359e-02 5.92055440e-01 7.38983095e-01
3.94005060e-01 -9.25817907e-01 1.15828788e+00 1.25258848e-01
-8.96546543e-02 -1.01633644e+00 -1.10506439e+00 -3.22387785e-01
4.81766403e-01 5.04182503e-02 1.74410969e-01 3.94010901e-01
3.07298303e-01 -1.87646315e-01 8.15401077e-01 -2.56609529e-01
4.60236490e-01 -1.15964405e-01 7.18367875e-01 -1.06003022e+00
-3.17648724e-02 -1.07717693e+00 -7.38267958e-01 5.87245636e-02
7.43748248e-02 -1.18601263e+00 -4.37455207e-01 -1.87552059e+00
6.75856948e-01 -4.58594203e-01 -2.68640608e-01 4.21976268e-01
1.03873082e-01 4.77263391e-01 -2.16323182e-01 4.11405593e-01
-2.91181803e-01 1.62779555e-01 1.21491468e+00 -1.57011509e-01
-1.65061310e-01 -1.25068769e-01 -5.01832008e-01 6.59112990e-01
8.81718278e-01 -4.10011858e-01 1.75138682e-01 5.39461851e-01
5.26433140e-02 3.01692992e-01 4.86440599e-01 -7.28082657e-01
2.85595953e-01 -2.82497436e-01 5.40813744e-01 -1.04224181e+00
8.49863961e-02 -9.51462686e-01 4.24031913e-01 1.12047482e+00
-1.44237339e-01 -4.87530708e-01 3.83540481e-01 3.17491174e-01
-2.46424720e-01 -6.69272691e-02 8.76513660e-01 -7.62100145e-02
-2.31143042e-01 8.70304853e-02 -1.20190287e+00 -7.43483365e-01
1.23811734e+00 -7.57496357e-01 -5.97009420e-01 -5.88609837e-02
-5.56060970e-01 -1.07685655e-01 8.65818143e-01 -5.32093406e-01
2.75131196e-01 -1.11496329e+00 -8.04049492e-01 5.23572117e-02
4.55743760e-01 -1.05985619e-01 7.08053768e-01 1.73079920e+00
-9.40972090e-01 8.52850378e-01 -1.75248533e-02 -6.07949138e-01
-1.38543558e+00 2.22090244e-01 5.57193696e-01 -8.86661351e-01
-2.56135255e-01 8.64542246e-01 2.09103897e-01 -2.07570851e-01
1.19165480e-01 -5.45128956e-02 -2.45550141e-01 3.88100237e-01
5.37244678e-01 4.27717388e-01 1.09129474e-01 -9.78630722e-01
-3.54295224e-01 7.96838820e-01 -3.65507185e-01 3.94445397e-02
9.75982010e-01 1.11445487e-02 -6.70398593e-01 5.27219653e-01
1.65465558e+00 2.23222956e-01 -4.47657138e-01 -9.13516805e-02
1.69248991e-02 9.58881807e-03 3.52970332e-01 -9.12906706e-01
-1.05538189e+00 4.76913124e-01 1.06887436e+00 -3.05352837e-01
1.32353377e+00 2.20397279e-01 9.38741148e-01 -4.41565454e-01
3.33129585e-01 -7.55272508e-01 -6.72503635e-02 1.41119406e-01
6.41828418e-01 -1.11423457e+00 -4.57463302e-02 -3.37290078e-01
-4.04009283e-01 1.20688581e+00 4.59456705e-02 7.50886127e-02
5.86337864e-01 6.20343924e-01 3.46143812e-01 -2.94824332e-01
-9.46855128e-01 -3.70823264e-01 -3.63575760e-03 5.78851700e-01
7.36962497e-01 5.73854089e-01 -6.49407268e-01 6.65162146e-01
-3.87817055e-01 1.72764927e-01 2.51933634e-01 7.12359011e-01
-6.68610275e-01 -1.16137540e+00 -5.19716144e-01 5.13498068e-01
-4.49068964e-01 1.70331866e-01 -6.78135514e-01 1.26864409e+00
-9.38348845e-03 4.93741184e-01 1.62129685e-01 2.73535941e-02
7.79875368e-02 -7.46424273e-02 1.10965811e-01 -1.79858178e-01
-8.25214982e-01 3.95555437e-01 3.86728682e-02 -4.62441221e-02
-3.61628801e-01 -7.55093634e-01 -1.62504959e+00 -5.04668832e-01
-2.65475035e-01 -8.35882407e-03 7.71931410e-01 6.50928557e-01
-2.17820644e-01 3.74971330e-01 3.40120018e-01 -2.88451672e-01
2.41014138e-01 -9.13754761e-01 -9.45575595e-01 -8.17393437e-02
4.04591292e-01 -2.97676265e-01 -1.18334711e+00 -3.24845374e-01] | [15.149307250976562, -3.1091551780700684] |
06d7f318-27b5-4f9e-b8b6-778d70b0206a | onednn-graph-compiler-a-hybrid-approach-for | 2301.01333 | null | https://arxiv.org/abs/2301.01333v1 | https://arxiv.org/pdf/2301.01333v1.pdf | oneDNN Graph Compiler: A Hybrid Approach for High-Performance Deep Learning Compilation | With the rapid development of deep learning models and hardware support for dense computing, the deep learning (DL) workload characteristics changed significantly from a few hot spots on compute-intensive operations to a broad range of operations scattered across the models. Accelerating a few compute-intensive operations using the expert-tuned implementation of primitives does not fully exploit the performance potential of AI hardware. Various efforts are made to compile a full deep neural network (DNN) graph. One of the biggest challenges is to achieve end-to-end compilation by generating expert-level performance code for the dense compute-intensive operations and applying compilation optimization at the scope of DNN computation graph across multiple compute-intensive operations. We present oneDNN Graph Compiler, a tensor compiler that employs a hybrid approach of using techniques from both compiler optimization and expert-tuned kernels for high-performance code generation of the deep neural network graph. oneDNN Graph Compiler addresses unique optimization challenges in the deep learning domain, such as low-precision computation, aggressive fusion, optimization for static tensor shapes and memory layout, constant weight optimization, and memory buffer reuse. Experimental results demonstrate up to 2x performance gains over primitives-based optimization for performance-critical DNN computation graph patterns on Intel Xeon Scalable Processors. | ['Dan Lavery', 'Eric Lin', 'Jason Ye', 'Baihui Jin', 'Xianhang Cheng', 'Longsheng Du', 'Yifei Zhang', 'Ciyong Chen', 'Yunfei Song', 'Jingze Cui', 'Yijie Mei', 'Zhennan Qin', 'Jianhui Li'] | 2023-01-03 | null | null | null | null | ['compiler-optimization'] | ['computer-code'] | [-4.44770694e-01 -1.71223626e-01 -9.40221623e-02 -5.05230725e-01
-2.60394812e-01 -4.16613042e-01 2.68565953e-01 3.25291216e-01
-5.35522997e-01 -1.25702292e-01 2.94265747e-01 -7.96550810e-01
2.35913917e-02 -1.12593877e+00 -7.29328215e-01 -5.74426711e-01
-4.79577005e-01 6.98766172e-01 1.85986787e-01 -4.21224624e-01
-5.42113325e-03 9.20124471e-01 -1.68767142e+00 5.72204232e-01
2.98230112e-01 1.23936355e+00 -1.58841327e-01 9.75865543e-01
-4.82512087e-01 1.06181502e+00 -4.48678732e-01 -5.06604433e-01
6.32937372e-01 3.21209401e-01 -5.62341273e-01 -4.34749961e-01
1.10161126e+00 -5.19335568e-01 -3.47037911e-01 1.06390142e+00
6.10232651e-01 -2.28049308e-02 1.99234977e-01 -1.31568718e+00
-2.04472110e-01 1.06806064e+00 -5.22147954e-01 5.14414251e-01
-3.98572654e-01 3.05334270e-01 1.14712512e+00 -8.87449741e-01
4.04048622e-01 1.21685421e+00 9.53308403e-01 1.75509438e-01
-1.23494816e+00 -7.38844097e-01 -1.85061917e-01 -1.39049860e-02
-1.44826734e+00 -2.25580722e-01 3.49777550e-01 -5.47918081e-01
1.73333001e+00 -9.70785022e-02 1.13281882e+00 5.62913418e-01
7.16308773e-01 5.16390562e-01 5.06859660e-01 -1.59890965e-01
4.29051071e-01 -6.58184588e-01 6.53632402e-01 1.17974007e+00
4.87033784e-01 3.49582553e-01 -5.84221423e-01 -1.99342877e-01
6.62639260e-01 -1.42565906e-01 1.57611564e-01 -6.75949827e-02
-1.15808058e+00 8.10049295e-01 7.73612022e-01 2.52789166e-02
-5.15785038e-01 7.37338245e-01 1.21002972e+00 3.61668169e-01
2.64422476e-01 3.55283529e-01 -6.56706333e-01 -4.42486018e-01
-9.91111994e-01 6.37526095e-01 1.12038982e+00 1.09848511e+00
9.66760695e-01 8.31700027e-01 -1.34316221e-01 3.22018206e-01
3.07606477e-02 3.13226968e-01 4.23463494e-01 -8.48081410e-01
4.72394437e-01 8.37631941e-01 -1.09134197e+00 -9.83054042e-01
-8.97937894e-01 -7.97773838e-01 -1.02509391e+00 4.46607351e-01
1.80391103e-01 -4.95806873e-01 -1.04161608e+00 1.17962897e+00
4.32314903e-01 -1.89947799e-01 9.15766433e-02 9.15106058e-01
1.07600665e+00 6.90522432e-01 2.85397857e-01 5.56945324e-01
1.30227709e+00 -1.15927231e+00 -1.05021976e-01 -2.14381590e-01
1.41213071e+00 -6.39149487e-01 9.39434707e-01 3.44576716e-01
-9.95794117e-01 -5.55532753e-01 -1.16825962e+00 -8.17629993e-01
-3.19357812e-01 -3.64429057e-02 1.28476453e+00 4.52028662e-01
-1.39876676e+00 4.79925841e-01 -1.13970411e+00 2.26555675e-01
3.40729922e-01 7.08481193e-01 -1.29588053e-01 -2.56016962e-02
-5.97331464e-01 4.31539923e-01 8.92119765e-01 -3.76484431e-02
-8.47350478e-01 -1.48046672e+00 -7.20279455e-01 3.47759843e-01
6.83779418e-02 -8.24961066e-01 1.16442716e+00 -1.00682557e+00
-1.08149278e+00 6.66584373e-01 5.25292754e-01 -7.24618196e-01
-1.16511881e-01 -7.06821457e-02 -1.75241694e-01 -2.10575059e-01
-2.97626376e-01 7.42081523e-01 6.24327779e-01 -3.61178577e-01
-5.62838972e-01 -4.42309439e-01 -3.38743664e-02 1.73927605e-01
-3.47587347e-01 -1.58788443e-01 -2.98448652e-01 -6.80844605e-01
-1.11835375e-01 -1.10791671e+00 -2.29550198e-01 -4.02644575e-02
-2.10214436e-01 -5.83242849e-02 8.69813561e-01 -2.42946923e-01
9.35678780e-01 -2.16143322e+00 1.49857551e-01 3.51712734e-01
8.25752854e-01 2.01030016e-01 -1.45191833e-01 9.71710607e-02
-1.02582544e-01 -3.77850920e-01 2.84192592e-01 -9.08115879e-02
1.58963010e-01 5.75181365e-01 -4.65594113e-01 1.86641455e-01
-1.42786905e-01 8.73860180e-01 -6.65747046e-01 -3.16805869e-01
-2.65272886e-01 5.16791224e-01 -1.23769295e+00 8.92028883e-02
-5.09536266e-01 -4.34871405e-01 -2.35523432e-01 8.05046022e-01
5.74927807e-01 -2.88393885e-01 1.04023241e-01 -9.12767470e-01
-2.44719058e-01 2.11262167e-01 -8.11076581e-01 1.65947545e+00
-4.33821380e-01 6.39910638e-01 2.73301393e-01 -7.29908884e-01
6.81721210e-01 -2.11240426e-01 3.49663824e-01 -6.91247821e-01
2.95723855e-01 3.53102058e-01 4.41055834e-01 1.98264681e-02
1.01612198e+00 4.84340824e-02 -6.07002452e-02 4.92304087e-01
4.19996649e-01 -3.35944027e-01 4.11883324e-01 3.76914680e-01
1.43750918e+00 -2.66990662e-01 -2.19247937e-01 -7.31343091e-01
-1.53469324e-01 6.14404023e-01 3.05452675e-01 3.90435368e-01
3.44173312e-02 -3.43262143e-02 8.20455492e-01 -1.10268056e+00
-1.36279130e+00 -9.21880543e-01 -4.15212959e-02 1.97846699e+00
-7.10540891e-01 -8.84657860e-01 -9.88877594e-01 -1.64939597e-01
1.13050282e-01 5.15201211e-01 -3.80940259e-01 -1.33718133e-01
-8.61244261e-01 -1.05807221e+00 9.59593177e-01 8.27339828e-01
6.78656340e-01 -4.94809955e-01 -8.74969482e-01 2.66198486e-01
8.33604455e-01 -1.29934883e+00 -4.02909607e-01 6.08963549e-01
-1.18846762e+00 -6.12684369e-01 -1.52750239e-01 -7.17869818e-01
4.51454282e-01 -4.72354777e-02 1.68452513e+00 2.61017144e-01
-4.54058021e-01 4.76789363e-02 1.06706925e-01 -4.50279146e-01
-4.55317259e-01 4.04231250e-01 -7.92921036e-02 -5.10684788e-01
4.13607389e-01 -7.01690376e-01 -3.55459034e-01 -3.01667154e-01
-8.02479684e-01 4.64059025e-01 6.86160624e-01 8.44957113e-01
7.25863099e-01 5.24161849e-03 -3.57639194e-01 -8.25233340e-01
7.78173685e-01 -1.40253529e-01 -9.99937177e-01 -9.34758708e-02
-6.67604208e-01 6.40680075e-01 8.89475882e-01 -4.51984644e-01
-5.17253220e-01 1.74168602e-01 -3.68061304e-01 -7.15858757e-01
4.42160934e-01 7.12820411e-01 2.50663847e-01 -4.34556752e-01
1.09375131e+00 -8.42986628e-02 7.93132260e-02 -1.49330482e-01
4.97708023e-01 -1.50962889e-01 6.42174542e-01 -1.21795607e+00
4.46173251e-01 7.27719367e-02 5.14151216e-01 -7.41501212e-01
-7.71882772e-01 7.03158304e-02 -4.27373141e-01 -5.93182854e-02
8.42336774e-01 -1.07588685e+00 -9.35523748e-01 5.64506531e-01
-1.08658493e+00 -8.89272928e-01 -4.00730610e-01 3.04706752e-01
2.24165842e-02 -1.53095379e-01 -1.03266037e+00 9.47552025e-02
-9.89820838e-01 -1.69635880e+00 1.13948596e+00 2.29624044e-02
-5.00716306e-02 -1.09926903e+00 -5.70358383e-03 4.22906801e-02
7.75353849e-01 2.17655212e-01 1.26958919e+00 -6.05575323e-01
-9.95899975e-01 2.39188634e-02 -7.19063759e-01 2.64039844e-01
-7.21247733e-01 1.97652578e-01 -7.12616622e-01 -2.55346864e-01
-2.41465479e-01 -3.90610158e-01 7.66082227e-01 1.91479892e-01
1.15829253e+00 -4.00526136e-01 5.59439212e-02 1.40842712e+00
1.53120375e+00 -2.39924997e-01 1.80876836e-01 3.75061668e-02
1.52829051e+00 -9.22995806e-02 -1.54258013e-01 3.91144454e-01
5.31424403e-01 4.57136780e-01 6.50751412e-01 5.67169338e-02
-2.48572096e-01 1.89338744e-01 2.99153537e-01 1.29237926e+00
-1.71763927e-01 3.48922700e-01 -1.52226126e+00 2.33612776e-01
-1.37499034e+00 -4.04285103e-01 -3.14541310e-01 1.62460613e+00
8.88488889e-01 4.06935543e-01 -1.84362009e-02 -9.17784404e-03
-1.22865714e-01 1.64367259e-01 -4.56279755e-01 -1.08630013e+00
2.39392266e-01 6.89677656e-01 1.23795795e+00 3.60129267e-01
-7.17876256e-01 1.14142060e+00 6.44088602e+00 8.38328600e-01
-1.58711970e+00 2.20978215e-01 4.65787560e-01 -3.57187718e-01
-1.86932608e-01 -2.09922493e-01 -1.41094744e+00 -2.50994228e-02
1.34226120e+00 -1.72294781e-01 6.21286809e-01 1.51855659e+00
-3.67776960e-01 1.99022681e-01 -1.27085590e+00 1.06023335e+00
-6.83735609e-02 -1.87737703e+00 3.80144149e-01 1.75533950e-01
7.30931699e-01 7.94829547e-01 -2.18521412e-02 6.21428847e-01
7.46195138e-01 -9.61092651e-01 8.65730643e-01 1.52844205e-01
7.50585675e-01 -1.08623183e+00 5.36253870e-01 -6.53599352e-02
-1.28570604e+00 -3.91633101e-02 -4.75828230e-01 -2.00621948e-01
-2.90392846e-01 9.56288695e-01 -9.68131065e-01 -9.19431522e-02
8.81635904e-01 3.67749602e-01 -7.49905944e-01 3.89723092e-01
3.29092294e-01 4.51131195e-01 -3.69253963e-01 -2.09058553e-01
6.16200864e-01 -4.89081107e-02 1.07381880e-01 1.56329513e+00
9.76507887e-02 1.20103890e-02 2.28622019e-01 7.95018017e-01
-3.02415937e-01 -7.73469219e-03 -4.29591596e-01 -2.82392412e-01
1.56555787e-01 1.41202688e+00 -7.10489929e-01 -5.53119600e-01
-3.75908434e-01 5.68643630e-01 7.97016501e-01 5.02566695e-02
-8.66439223e-01 -3.67344916e-01 1.11466467e+00 8.31071883e-02
1.68471962e-01 -8.23387265e-01 -9.03817177e-01 -9.30294216e-01
-3.21363896e-01 -1.02435017e+00 3.62966508e-01 -5.47422111e-01
-7.21599698e-01 7.78916955e-01 5.54243810e-02 -3.79241228e-01
-2.87780643e-01 -1.07840729e+00 -3.55358869e-01 7.01574802e-01
-9.18751299e-01 -1.06143308e+00 -4.44355011e-01 5.95707834e-01
1.62888601e-01 -6.12564683e-01 7.28618622e-01 6.14298224e-01
-5.68544865e-01 8.07357430e-01 -1.49975151e-01 3.67158294e-01
-1.05458751e-01 -1.13973987e+00 8.76635373e-01 8.55908930e-01
-1.91485230e-02 6.07086897e-01 4.49266583e-01 -5.62620103e-01
-2.26854587e+00 -1.30058217e+00 4.97939020e-01 1.06817499e-01
1.03545713e+00 -4.95528489e-01 -8.59349072e-01 9.65297163e-01
-3.23283561e-02 4.44559634e-01 5.11276960e-01 2.48656362e-01
-6.73779607e-01 -5.68987489e-01 -6.51629388e-01 7.74357975e-01
1.01970780e+00 -5.18466890e-01 6.28639460e-02 6.23867989e-01
9.07979131e-01 -1.25014293e+00 -1.20797360e+00 3.01417798e-01
4.40933466e-01 -8.64892364e-01 1.00394952e+00 -4.91734266e-01
5.04180372e-01 7.14737549e-02 -4.59487468e-01 -1.04730201e+00
-3.18933368e-01 -2.78743088e-01 -3.04482043e-01 5.05304933e-01
1.46881595e-01 -4.40059364e-01 9.72039223e-01 7.22484112e-01
-7.82102942e-01 -8.90888870e-01 -5.85720837e-01 -4.60692257e-01
3.00458018e-02 -8.60388339e-01 7.68161178e-01 7.89006770e-01
-4.91823077e-01 4.44065988e-01 4.30041790e-01 -1.15437247e-02
6.82715535e-01 7.99885951e-03 1.01435542e+00 -8.20552766e-01
-6.09244585e-01 -8.24017644e-01 -7.80194938e-01 -8.81668985e-01
1.35090873e-01 -1.29295599e+00 -3.80480707e-01 -1.05821860e+00
-1.43488765e-01 -5.73248029e-01 2.49675885e-01 8.10954213e-01
4.88391221e-01 2.15711713e-01 2.03185156e-01 -2.08437331e-02
-1.50941029e-01 1.48024902e-01 1.13265717e+00 -1.43099830e-01
1.51881725e-01 -6.47860587e-01 -5.88407457e-01 8.09989274e-01
4.73808914e-01 -2.95298994e-01 -5.37826538e-01 -1.21653652e+00
6.70139551e-01 -2.63594717e-01 4.52885389e-01 -1.41224337e+00
5.80344081e-01 1.44684300e-01 2.20203161e-01 -7.21287072e-01
2.12424472e-01 -5.80154419e-01 2.13300437e-01 5.74420154e-01
-2.42788807e-01 9.37262058e-01 6.95716977e-01 -6.08420819e-02
-2.10188739e-02 1.04242206e-01 9.70589042e-01 -1.85165957e-01
-1.06177843e+00 6.74954832e-01 -8.76571834e-02 2.63292879e-01
6.70931458e-01 2.71022879e-02 -5.95960081e-01 2.36629158e-01
-5.52377462e-01 -1.63337544e-01 1.67889327e-01 1.03948511e-01
5.26345134e-01 -1.22526884e+00 -6.69934511e-01 5.06743908e-01
-2.05533564e-01 4.50835109e-01 6.41711205e-02 6.45768762e-01
-1.58074152e+00 4.52693045e-01 -5.22896349e-01 -7.12799370e-01
-1.15921092e+00 5.07391214e-01 5.22677958e-01 -4.53858525e-01
-8.51353288e-01 1.16207623e+00 1.44446284e-01 -3.37260425e-01
3.52043569e-01 -9.95087445e-01 5.33767819e-01 -1.58734009e-01
5.50601304e-01 4.24615115e-01 6.96331084e-01 -4.53661591e-01
-2.32666656e-01 2.89761126e-01 -9.65518802e-02 2.79073328e-01
1.25717330e+00 7.23100960e-01 -5.48293829e-01 7.41843954e-02
1.66998458e+00 -4.06112939e-01 -8.65171194e-01 -1.13918185e-01
-1.54735595e-01 2.11443990e-01 8.51234794e-01 -3.95240068e-01
-1.84613574e+00 8.80615592e-01 5.89247584e-01 -1.84234515e-01
1.14619803e+00 -3.42720211e-01 1.02564442e+00 7.97055125e-01
2.67555714e-01 -1.11370611e+00 -6.54211687e-03 1.06177604e+00
6.60728037e-01 -7.56573558e-01 4.05854851e-01 -1.20444432e-01
-1.35809034e-01 1.47103620e+00 8.43659341e-01 -4.29946065e-01
1.08561027e+00 1.24410832e+00 -1.42749637e-01 -6.69379115e-01
-9.84320641e-01 2.33502477e-01 3.69207948e-01 1.94620714e-01
4.91671354e-01 2.77485549e-01 1.18559688e-01 2.73082266e-03
-8.31332445e-01 -4.98599634e-02 2.42053330e-01 8.40512693e-01
-2.34612331e-01 -7.35303164e-01 -1.39045671e-01 7.93780565e-01
-2.64476836e-01 -5.91950417e-01 2.20607862e-01 8.49650800e-01
7.38692656e-02 -2.39435080e-02 6.25472307e-01 -7.48454332e-01
1.93389967e-01 -1.86582133e-01 4.77457166e-01 -5.12985528e-01
-1.23134828e+00 -2.62806207e-01 3.18298161e-01 -8.32856834e-01
1.62423730e-01 -3.30223739e-01 -1.63672507e+00 -1.10161400e+00
2.48301640e-01 -3.90357971e-01 1.21874738e+00 7.14624584e-01
6.59871638e-01 8.52370858e-01 -1.83934733e-01 -1.12911868e+00
-5.74105740e-01 -2.91223615e-01 -4.26169902e-01 -8.00112903e-04
8.95337239e-02 -2.07328275e-01 -3.00726900e-03 -1.73049480e-01] | [8.296557426452637, 3.0453877449035645] |
2dc0d686-930b-46b3-81f8-4dfb0412b926 | sketch-bert-learning-sketch-bidirectional | 2005.09159 | null | https://arxiv.org/abs/2005.09159v1 | https://arxiv.org/pdf/2005.09159v1.pdf | Sketch-BERT: Learning Sketch Bidirectional Encoder Representation from Transformers by Self-supervised Learning of Sketch Gestalt | Previous researches of sketches often considered sketches in pixel format and leveraged CNN based models in the sketch understanding. Fundamentally, a sketch is stored as a sequence of data points, a vector format representation, rather than the photo-realistic image of pixels. SketchRNN studied a generative neural representation for sketches of vector format by Long Short Term Memory networks (LSTM). Unfortunately, the representation learned by SketchRNN is primarily for the generation tasks, rather than the other tasks of recognition and retrieval of sketches. To this end and inspired by the recent BERT model, we present a model of learning Sketch Bidirectional Encoder Representation from Transformer (Sketch-BERT). We generalize BERT to sketch domain, with the novel proposed components and pre-training algorithms, including the newly designed sketch embedding networks, and the self-supervised learning of sketch gestalt. Particularly, towards the pre-training task, we present a novel Sketch Gestalt Model (SGM) to help train the Sketch-BERT. Experimentally, we show that the learned representation of Sketch-BERT can help and improve the performance of the downstream tasks of sketch recognition, sketch retrieval, and sketch gestalt. | ['xiangyang xue', 'Yu-Gang Jiang', 'Hangyu Lin', 'Yanwei Fu'] | 2020-05-19 | sketch-bert-learning-sketch-bidirectional-1 | http://openaccess.thecvf.com/content_CVPR_2020/html/Lin_Sketch-BERT_Learning_Sketch_Bidirectional_Encoder_Representation_From_Transformers_by_Self-Supervised_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Lin_Sketch-BERT_Learning_Sketch_Bidirectional_Encoder_Representation_From_Transformers_by_Self-Supervised_CVPR_2020_paper.pdf | cvpr-2020-6 | ['sketch-recognition'] | ['computer-vision'] | [ 1.79115593e-01 -8.98969769e-02 -1.08586185e-01 -4.39288795e-01
-3.19752455e-01 -4.79626417e-01 1.12150860e+00 -6.19069755e-01
1.30681574e-01 3.54499698e-01 2.67604560e-01 -3.59153718e-01
-1.06674306e-01 -1.28913665e+00 -9.55770910e-01 -5.60041368e-01
2.21003741e-01 3.08415622e-01 -4.18048650e-01 -1.40190154e-01
3.69817138e-01 7.84819603e-01 -1.15610135e+00 5.43104768e-01
1.00779988e-01 1.07005954e+00 3.28723073e-01 8.62882912e-01
-7.11195469e-01 8.23845863e-01 -5.49185514e-01 -6.51247084e-01
2.20120728e-01 -3.73248518e-01 -3.99671853e-01 -3.58121358e-02
7.43932605e-01 -9.69784200e-01 -1.07239485e+00 6.25830591e-01
3.04820597e-01 3.32145318e-02 9.93253827e-01 -1.50306153e+00
-1.52416599e+00 6.37691975e-01 -4.22600359e-02 -5.18501520e-01
9.00310874e-02 3.30795556e-01 1.12550223e+00 -1.35292673e+00
8.35193813e-01 1.38841999e+00 4.74670291e-01 8.86887789e-01
-9.04679954e-01 -8.46943438e-01 1.79762274e-01 -1.25146866e-01
-1.35850775e+00 -3.97776067e-01 1.01261699e+00 -2.47509420e-01
6.70021713e-01 3.18644121e-02 9.46997106e-01 1.52310884e+00
-1.22426547e-01 1.25735319e+00 5.02136827e-01 -1.84956282e-01
-6.12744922e-03 1.01646714e-01 -2.33885080e-01 9.54719067e-01
-1.78062886e-01 2.73808718e-01 -5.87989628e-01 -1.41627025e-02
1.87551129e+00 7.78428555e-01 -3.93729936e-03 -4.02081162e-01
-1.14487350e+00 7.92722404e-01 7.81687677e-01 3.75752598e-01
-3.39223415e-01 9.48692083e-01 1.88972503e-01 5.20609260e-01
8.99654627e-02 1.17265314e-01 9.50621292e-02 5.54082729e-02
-1.09585404e+00 2.58801669e-01 6.07382238e-01 1.38106620e+00
5.42300880e-01 3.58360678e-01 -3.32816750e-01 8.21840107e-01
4.21562791e-01 6.76992536e-01 4.24037546e-01 -4.49725658e-01
5.03073215e-01 5.93829632e-01 -2.12546393e-01 -8.75020683e-01
4.96655524e-01 3.36939543e-02 -1.22505677e+00 1.86762348e-01
1.39710560e-01 2.95131564e-01 -1.00637817e+00 1.61793256e+00
-4.34899867e-01 2.92932898e-01 4.68291901e-02 9.25392389e-01
1.19213557e+00 1.03395987e+00 5.75618669e-02 4.79130447e-01
1.04003656e+00 -1.01426888e+00 -6.13499522e-01 1.69446036e-01
-1.09401979e-01 -5.81015110e-01 1.26258957e+00 2.02335417e-01
-1.19446301e+00 -9.96494889e-01 -1.19906747e+00 -3.97147387e-01
-7.45959938e-01 5.16487896e-01 6.70060754e-01 3.14396799e-01
-1.07305944e+00 8.13870132e-01 -2.50180930e-01 -4.46334124e-01
7.60312021e-01 -6.92876205e-02 -4.03122663e-01 -1.61748394e-01
-9.85343397e-01 4.52160865e-01 -9.43751931e-02 4.19127047e-01
-1.09730685e+00 -6.02034211e-01 -7.97470152e-01 4.89051908e-01
-9.69843641e-02 -1.00564897e+00 8.48526835e-01 -1.05311310e+00
-1.65060663e+00 6.68333650e-01 4.80608195e-02 -3.24258864e-01
4.85824257e-01 -1.30323112e-01 -1.34872958e-01 4.91960831e-02
-4.35729772e-01 9.68118489e-01 1.47583246e+00 -1.36118793e+00
-9.04924143e-03 -8.93589929e-02 1.31157383e-01 -1.51828051e-01
-2.94754922e-01 -5.11040866e-01 -3.65496397e-01 -8.75966549e-01
-9.24942642e-02 -5.06403208e-01 9.25392061e-02 7.15168595e-01
-1.46009654e-01 -4.01400417e-01 1.02226055e+00 -4.59697247e-01
9.15586054e-01 -2.33446169e+00 2.05854811e-02 2.84581631e-01
3.52151603e-01 4.05819714e-01 -6.81981385e-01 9.80636477e-01
-1.24549687e-01 2.33061731e-01 -6.65871203e-02 -4.03474271e-01
4.68102098e-01 5.66130400e-01 -1.31270123e+00 -5.15200309e-02
5.54882824e-01 1.67689204e+00 -8.75959814e-01 -2.42798239e-01
4.65169042e-01 5.81138492e-01 -2.22733900e-01 7.05666244e-01
-4.44789052e-01 -1.30983889e-01 -3.48040879e-01 6.29065216e-01
7.51212597e-01 -1.57381743e-01 1.25780255e-01 -4.27185833e-01
2.26625651e-01 3.04522198e-02 -9.18005109e-01 2.15702653e+00
-8.23675871e-01 7.59359062e-01 -6.20398708e-02 -8.99253726e-01
1.33857763e+00 4.35794532e-01 3.65461297e-02 -6.50897563e-01
-2.35692292e-01 5.12350202e-02 -6.15546405e-01 -3.80794257e-01
4.83044833e-01 -3.41785550e-01 8.18654746e-02 8.22162747e-01
3.57596338e-01 -3.93191576e-01 -3.09094578e-01 4.79393125e-01
7.86694348e-01 5.17514527e-01 -1.37106270e-01 2.89716661e-01
2.63795793e-01 -6.39314532e-01 -4.10653830e-01 8.60151231e-01
3.90452325e-01 7.28971601e-01 4.23811644e-01 -8.33597362e-01
-1.23856318e+00 -1.53212917e+00 2.01525509e-01 9.53051031e-01
-8.98642838e-03 -5.45282483e-01 -3.65812778e-01 -6.90122306e-01
4.15318251e-01 5.04031479e-01 -6.34020269e-01 -1.96290731e-01
-6.35787189e-01 2.05919906e-01 9.12315786e-01 7.50199795e-01
4.20519799e-01 -1.68446827e+00 -3.32162619e-01 1.52984962e-01
5.13223231e-01 -7.76917517e-01 -5.53879559e-01 -2.24543005e-01
-7.79327154e-01 -8.70341182e-01 -1.21861625e+00 -9.02938068e-01
8.66041541e-01 2.02905849e-01 1.07747090e+00 4.98345792e-01
-4.11642879e-01 6.76344454e-01 -1.39532089e-01 -1.49891287e-01
-1.80366144e-01 -1.95413992e-01 -3.63971740e-01 1.22323602e-01
-1.26696527e-01 -8.81156623e-01 -6.26768589e-01 -5.80194853e-02
-1.27259946e+00 4.39338297e-01 1.01081038e+00 1.20449102e+00
1.75413266e-01 -5.95617056e-01 7.07041204e-01 -6.09648943e-01
8.99045408e-01 -2.54364163e-01 -1.95124254e-01 6.31511211e-01
-4.50979263e-01 3.88056964e-01 7.95150578e-01 -6.41693771e-01
-9.05609846e-01 -1.77062422e-01 -3.83505613e-01 -9.45909262e-01
-1.05394214e-01 8.91424567e-02 -2.04219237e-01 5.24714366e-02
1.69000208e-01 7.18693972e-01 6.49855984e-03 -5.83029807e-01
1.01041210e+00 5.92885435e-01 3.93492103e-01 -9.16413546e-01
8.50877464e-01 3.54401559e-01 -8.21080133e-02 -8.84714246e-01
-2.09104836e-01 1.43376037e-01 -5.57487190e-01 -2.24950075e-01
4.96894419e-01 -6.16245508e-01 -6.75678968e-01 2.47747630e-01
-1.46408904e+00 -4.33700889e-01 -5.89322686e-01 -2.23589554e-01
-7.43884146e-01 2.03547552e-01 -7.13942051e-01 -1.03059566e+00
-7.25663006e-01 -8.55004907e-01 1.49899614e+00 1.36549577e-01
1.15089528e-01 -9.62590039e-01 -1.11096039e-01 -2.40165740e-01
6.70640588e-01 -4.97392118e-02 1.26743412e+00 -1.10114805e-01
-1.17362690e+00 -1.92373291e-01 -7.96626508e-01 3.66292179e-01
-5.90745583e-02 -4.51795012e-02 -1.09646463e+00 -9.68814418e-02
-4.67696071e-01 -4.83838946e-01 1.08724308e+00 1.62338708e-02
1.39431846e+00 -4.42148954e-01 -1.69001758e-01 8.50377619e-01
1.51244819e+00 9.12162140e-02 8.72310340e-01 -3.95874441e-01
7.28584170e-01 1.88058272e-01 2.38274131e-02 3.14828783e-01
1.32324040e-01 3.14666599e-01 2.24721491e-01 -2.53824204e-01
-4.80335444e-01 -1.21791518e+00 3.19476992e-01 7.61975229e-01
-8.36517438e-02 -2.25557059e-01 -3.62053126e-01 4.59120393e-01
-1.92087805e+00 -1.05073726e+00 5.14268219e-01 1.82814670e+00
4.82708246e-01 -1.36997864e-01 -1.83722258e-01 -1.61936268e-01
3.48934889e-01 4.85431433e-01 -5.36313951e-01 -5.32081485e-01
-1.93603337e-02 6.30823195e-01 -7.00853169e-02 2.46217877e-01
-6.16670549e-01 1.27451456e+00 6.05265284e+00 8.28042507e-01
-1.37724614e+00 -2.40450114e-01 2.95426995e-01 7.57406503e-02
-6.92745805e-01 -7.23681971e-02 -2.74384230e-01 2.63477653e-01
5.55320859e-01 9.71665904e-02 8.53092492e-01 8.26031268e-01
-4.44316328e-01 5.29215157e-01 -1.58981264e+00 1.49433744e+00
1.07179224e-01 -1.81708431e+00 1.00208497e+00 -1.21973760e-01
2.67719507e-01 -5.94012737e-01 3.18266600e-01 7.06407964e-01
1.15340210e-01 -1.40123963e+00 7.59465337e-01 1.07488418e+00
1.23824012e+00 -5.00858366e-01 2.31295839e-01 2.47132644e-01
-1.36155415e+00 1.23493411e-02 -6.87934756e-01 -1.72653627e-02
2.18863692e-02 7.07874000e-02 -8.16801488e-01 4.87894744e-01
7.24922195e-02 1.10993910e+00 -3.67910266e-01 5.18296480e-01
-3.22326064e-01 2.83042222e-01 6.49832487e-02 -2.74349302e-01
5.26205480e-01 -3.99205863e-01 1.49507165e-01 1.39645052e+00
3.73418093e-01 1.81167841e-01 -1.94082037e-01 1.62850654e+00
-4.30191875e-01 -1.60541475e-01 -1.18650460e+00 -8.08934331e-01
3.53536636e-01 1.27494669e+00 -2.45146140e-01 -6.76201880e-01
-1.48616359e-01 1.27472103e+00 4.66549128e-01 7.22298443e-01
-6.85046434e-01 -5.19594610e-01 6.49564326e-01 1.19341575e-01
4.57918644e-01 -4.37487662e-01 -1.93984538e-01 -1.05553675e+00
1.14958815e-01 -4.39210385e-01 -1.87464789e-01 -1.14781392e+00
-1.62921691e+00 4.67644721e-01 -2.34902903e-01 -9.71160889e-01
-1.83219537e-01 -8.78337383e-01 -1.01693892e+00 1.02615857e+00
-1.42720509e+00 -1.87532687e+00 -3.33833307e-01 6.41434133e-01
7.42230535e-01 -4.17070746e-01 1.03160214e+00 3.33490998e-01
-1.50899187e-01 6.36556983e-01 -1.11032389e-02 6.25865459e-01
6.94719732e-01 -1.17629480e+00 1.01684654e+00 2.89258510e-01
7.84881473e-01 1.12628591e+00 -9.08981338e-02 -4.96352881e-01
-2.00424767e+00 -1.04516244e+00 6.23299956e-01 -2.78237939e-01
5.89736700e-01 -9.21502054e-01 -6.44317508e-01 7.41106212e-01
1.58056676e-01 2.78135967e-02 3.25390667e-01 -2.11883113e-01
-5.70169508e-01 -1.39131516e-01 -8.58652651e-01 7.68919468e-01
1.46533287e+00 -1.14855015e+00 -6.48370624e-01 1.03407465e-01
4.32125866e-01 -1.16435431e-01 -7.92943418e-01 -1.72311142e-01
1.33701682e+00 -5.77082634e-01 1.39205801e+00 -9.26653266e-01
7.49509871e-01 2.07725354e-02 -1.92977667e-01 -1.03406668e+00
-2.74155498e-01 -4.29106295e-01 -2.81880558e-01 1.18296444e+00
3.56924012e-02 -1.45528898e-01 8.81050348e-01 2.85107553e-01
1.26744077e-01 -8.73948395e-01 -5.40034652e-01 -5.73989391e-01
3.39577235e-02 -3.20773482e-01 1.02122188e+00 6.56574965e-01
-3.49987000e-01 3.89188677e-01 -7.13054717e-01 -3.37620199e-01
4.33572888e-01 7.01940238e-01 1.08771110e+00 -1.00759804e+00
-1.76246956e-01 -5.90510845e-01 -3.51505488e-01 -1.72372282e+00
2.59294003e-01 -1.04626620e+00 -1.28705278e-01 -1.74206781e+00
2.23422213e-03 -5.75786471e-01 -2.09661782e-01 3.70791525e-01
1.21379130e-01 1.04251809e-01 6.44205749e-01 1.74409240e-01
-3.70892853e-01 8.59241784e-01 1.58863103e+00 -5.85669458e-01
2.90940017e-01 -2.15517566e-01 -3.99647176e-01 3.14161062e-01
2.50295490e-01 -1.30587369e-01 -5.97516119e-01 -6.53417110e-01
8.84436369e-02 3.81472409e-01 8.92528713e-01 -6.89242601e-01
2.98043817e-01 1.21577003e-03 8.19787085e-01 -7.96979249e-01
7.28180945e-01 -9.49967802e-01 1.74634740e-01 2.68107772e-01
-6.10773921e-01 3.82242561e-03 -1.85716555e-01 5.80914021e-01
-1.58011675e-01 -2.33360067e-01 2.98294067e-01 -3.73454213e-01
-6.06819570e-01 6.87389195e-01 -8.55707601e-02 -5.96274555e-01
4.68538731e-01 -2.05531627e-01 -2.08226919e-01 -5.88577151e-01
-7.05199182e-01 -2.66778708e-01 2.67988682e-01 6.06391370e-01
1.31201923e+00 -1.82500267e+00 -3.80791157e-01 7.27603316e-01
1.02882884e-01 -1.81379303e-01 4.83938977e-02 1.90050974e-02
-3.55226129e-01 5.12018144e-01 -5.28802216e-01 -2.89845169e-01
-7.79789567e-01 7.60816693e-01 1.71577752e-01 -1.43989222e-02
-9.95260477e-01 8.35383415e-01 5.04771650e-01 -5.33279717e-01
4.33988094e-01 -6.41260386e-01 2.49854073e-01 -2.34062314e-01
5.22629857e-01 -7.26038367e-02 -3.81542414e-01 7.20352028e-03
1.57455415e-01 5.66184342e-01 1.91408321e-01 -1.84155017e-01
1.41604650e+00 4.11514491e-01 -2.66758539e-02 5.62274516e-01
1.22957742e+00 -3.72869521e-01 -1.28569829e+00 -1.90115422e-01
-2.24270403e-01 -4.71929878e-01 -2.34839946e-01 -7.65980303e-01
-1.01800883e+00 1.44484699e+00 4.00171131e-01 -8.77479315e-02
6.31489396e-01 -2.10974425e-01 1.15412021e+00 7.82679141e-01
4.19329941e-01 -6.59653306e-01 4.73634839e-01 4.02447045e-01
1.50481021e+00 -8.12905371e-01 -1.90734476e-01 2.08808318e-01
-4.24329519e-01 1.65516925e+00 3.55663955e-01 -6.36703074e-01
6.77134097e-01 2.73691297e-01 -1.58151790e-01 -3.57034445e-01
-8.35072160e-01 1.36839941e-01 4.26906258e-01 4.89240676e-01
3.01491052e-01 1.06964886e-01 3.51897746e-01 7.83958435e-01
-2.07463596e-02 2.90679961e-01 -1.33421361e-01 7.44289577e-01
-1.52041867e-01 -1.29456627e+00 -8.16160217e-02 3.94295573e-01
4.63700622e-01 -2.74034709e-01 -7.35752761e-01 7.15223432e-01
1.44156113e-01 1.31971031e-01 1.90065250e-01 -3.15764785e-01
2.80120730e-01 3.70348513e-01 8.92732441e-01 -3.39955419e-01
-4.91201550e-01 -3.28815192e-01 -3.09545219e-01 -4.53940988e-01
-7.10424781e-02 5.20387739e-02 -8.43205333e-01 -3.40729177e-01
-6.49629347e-03 -1.68843195e-02 8.33546638e-01 7.00670600e-01
2.38533899e-01 4.32407677e-01 4.96186972e-01 -1.07992005e+00
-5.68212748e-01 -9.90114391e-01 -6.18242979e-01 6.20612979e-01
4.16785955e-01 -5.12898803e-01 1.01322882e-01 1.96060285e-01] | [11.773115158081055, 0.387342244386673] |
5eaf581c-c85f-4461-aff2-fc4c836a8f62 | generating-a-temporally-coherent-image | null | null | https://openreview.net/forum?id=hyQFVwTynuA | https://openreview.net/pdf?id=hyQFVwTynuA | Generating a Temporally Coherent Image Sequence for a Story by Multimodal Recurrent Transformers | Story visualization is a challenging text-to-image generation task for the difficulty of rendering visual details from abstract text descriptions.
Besides the difficulty of image generation, the generator also need to conform to the narrative of a multi-sentence story input.
While prior arts in this domain has focused on improving semantic relevance between generated images and input text, controlling the generated images to be temporally consistent still remains as a challenge. To generate a semantically coherent image sequence, we propose an explicit memory controller which can augment the temporal coherence of images in the multi-modal autoregressive transformer, and call Story visualization by MultimodAl Recurrent Transformers or SMART for short. Our method generates high resolution high quality images, outperforming prior works by a significant margin across multiple evaluation metrics on PororoSV dataset. | ['Anonymous'] | 2021-11-16 | null | null | null | acl-arr-november-2021-11 | ['story-visualization'] | ['computer-vision'] | [ 5.12616515e-01 2.49189705e-01 2.59059012e-01 -2.57422596e-01
-7.72150338e-01 -6.12635732e-01 1.13610983e+00 -2.73869604e-01
2.56328970e-01 7.69479752e-01 7.17723429e-01 -4.97758426e-02
2.04703644e-01 -5.10366321e-01 -6.63993359e-01 -4.22514468e-01
3.97220224e-01 3.90353084e-01 8.59589055e-02 -1.44216552e-01
2.08822221e-01 1.28144786e-01 -1.54015732e+00 8.97676647e-01
5.62482178e-01 6.88909650e-01 5.18433392e-01 9.73499537e-01
-3.62550259e-01 1.17834878e+00 -8.34480703e-01 -3.13866973e-01
-5.83954528e-02 -9.32509482e-01 -7.77499497e-01 2.57490039e-01
6.48182929e-01 -3.79609406e-01 -2.06672862e-01 7.37308323e-01
4.72219437e-01 2.74175137e-01 9.22328830e-01 -1.47055554e+00
-9.61914480e-01 7.20927238e-01 -6.86685324e-01 1.33407339e-01
6.49944007e-01 2.32275933e-01 7.46358812e-01 -8.76029253e-01
1.21287394e+00 1.37401104e+00 7.75244012e-02 6.02746785e-01
-1.57158959e+00 -4.95832622e-01 3.56869280e-01 9.61173624e-02
-9.75883603e-01 -4.91246670e-01 6.87727928e-01 -7.90177882e-01
9.35665786e-01 4.58224595e-01 7.72212863e-01 1.45605767e+00
1.12364456e-01 6.68173850e-01 1.14901042e+00 -2.03349665e-01
4.42913081e-03 1.36970252e-01 -3.55545849e-01 4.61384118e-01
-2.54879355e-01 -5.69495037e-02 -9.61608469e-01 2.38486946e-01
9.45520461e-01 -2.49893382e-01 -2.36801296e-01 -2.07502708e-01
-1.45580375e+00 7.82427609e-01 1.39975712e-01 1.59711450e-01
-4.86971468e-01 3.87741625e-01 2.98795253e-01 2.91087721e-02
4.62939382e-01 5.32913864e-01 2.14341402e-01 -2.52703488e-01
-1.08593845e+00 4.90037799e-01 3.64208877e-01 1.05791068e+00
2.71779358e-01 4.92631584e-01 -7.86490679e-01 4.74635839e-01
9.43445116e-02 4.29573745e-01 2.63988286e-01 -8.99394274e-01
5.55773139e-01 4.34247851e-01 1.29811257e-01 -1.03537309e+00
-1.86681747e-01 -1.81186974e-01 -8.94999325e-01 4.14528400e-01
3.24569255e-01 -4.64296155e-02 -8.65143239e-01 1.70517874e+00
1.75884351e-01 1.00389540e-01 1.27552569e-01 1.00082552e+00
1.16467226e+00 1.17949545e+00 2.65411645e-01 -2.64432818e-01
1.49250078e+00 -1.03515840e+00 -1.05062377e+00 -2.58481920e-01
3.01160384e-02 -8.77489984e-01 1.48066437e+00 1.86178401e-01
-1.49771178e+00 -4.17908818e-01 -9.36569393e-01 -2.82736599e-01
-1.05095878e-01 5.50914295e-02 3.33127081e-01 -1.44727845e-02
-1.13036489e+00 1.82581976e-01 -4.01572555e-01 -2.43134737e-01
2.73475617e-01 -3.58538806e-01 -3.62550676e-01 1.69406980e-01
-1.08364785e+00 9.14195895e-01 4.54228401e-01 -3.10922414e-01
-8.42911959e-01 -1.13139224e+00 -8.61057997e-01 -1.52529217e-02
2.12718230e-02 -1.11120689e+00 1.34124804e+00 -1.07920003e+00
-1.59402514e+00 8.03291500e-01 -2.00563654e-01 -3.10106009e-01
7.72924542e-01 -8.36697519e-02 -2.91459054e-01 3.63527834e-01
5.15712500e-02 1.15332770e+00 1.03941226e+00 -1.58969414e+00
-4.45777982e-01 1.40924126e-01 9.92833748e-02 5.56486368e-01
-1.60751596e-01 1.25293121e-01 -3.65655482e-01 -1.06517887e+00
-2.58353621e-01 -7.57113099e-01 -1.75525218e-01 -5.53552508e-02
-5.99211037e-01 8.49699900e-02 1.30051076e+00 -7.78845131e-01
1.04011559e+00 -2.05034614e+00 2.98423350e-01 -2.66698182e-01
1.90849096e-01 -2.52772927e-01 -1.86340809e-01 5.50093591e-01
-3.63876283e-01 1.99369788e-01 -1.39147848e-01 -5.78728318e-01
-9.34611354e-03 -2.23217115e-01 -9.60152149e-01 2.78635211e-02
2.62291402e-01 1.13970041e+00 -8.48145723e-01 -7.56011367e-01
2.13779151e-01 7.83776581e-01 -3.73305738e-01 4.58555549e-01
-5.65663338e-01 7.45696664e-01 -1.39629617e-01 3.18692252e-02
2.88518548e-01 -6.30152047e-01 -6.16245531e-02 -4.21925426e-01
-1.28876343e-01 6.72551543e-02 -9.49329734e-01 1.89143455e+00
-5.06367445e-01 1.15651500e+00 -3.25320691e-01 -4.21543479e-01
6.84307158e-01 4.72732335e-01 2.15476468e-01 -1.02578163e+00
-2.18385428e-01 -4.75743026e-01 -3.70772421e-01 -6.49926186e-01
1.05251634e+00 -1.55334726e-01 -7.14927986e-02 8.41045618e-01
-4.55152333e-01 -5.09923637e-01 3.17694336e-01 6.06821120e-01
5.70391953e-01 5.82144976e-01 -1.00092359e-01 7.60471150e-02
-5.57083823e-02 2.51762986e-01 4.72707190e-02 5.55979669e-01
5.10910213e-01 1.26837289e+00 6.21183217e-01 -1.07983813e-01
-1.38747799e+00 -1.28823781e+00 2.74925321e-01 9.51472521e-01
2.23089769e-01 -5.17945409e-01 -6.60984695e-01 -2.69135952e-01
-4.68843669e-01 1.24804115e+00 -6.09098732e-01 1.01128407e-01
-6.57868803e-01 -4.96443599e-01 3.12715948e-01 5.26036084e-01
3.52651387e-01 -1.26493812e+00 -1.09091282e+00 4.81927805e-02
-6.47929847e-01 -1.25795066e+00 -7.68969774e-01 -3.56500804e-01
-4.99036670e-01 -7.78241098e-01 -8.91512692e-01 -6.08843207e-01
6.89422786e-01 2.14882150e-01 1.39093530e+00 -3.24610233e-01
-5.77727199e-01 3.87803078e-01 -2.78690189e-01 -2.26692542e-01
-5.82619369e-01 -2.93871790e-01 -5.30912459e-01 2.69169398e-02
-7.67075598e-01 -5.60939133e-01 -7.78983355e-01 2.88324766e-02
-1.24545848e+00 1.42230761e+00 2.59546787e-01 7.46135771e-01
6.06386840e-01 -2.74288833e-01 3.69299799e-01 -7.56131709e-01
8.87089312e-01 -3.68842334e-01 -2.00145066e-01 2.56594926e-01
-2.83975005e-01 1.68301716e-01 4.83055025e-01 -6.85191154e-01
-1.41283560e+00 8.60586092e-02 4.62825119e-01 -6.37763739e-01
1.79963097e-01 3.13859463e-01 5.72043322e-02 7.02091277e-01
4.59410191e-01 2.29775757e-01 -1.77054971e-01 4.70625088e-02
8.56642485e-01 -3.36608887e-02 7.27273345e-01 -4.78791714e-01
7.37742782e-01 5.00265062e-01 -9.10652801e-02 -6.20010257e-01
-4.78386819e-01 2.59258926e-01 -2.67971754e-01 -6.42640173e-01
1.30066693e+00 -9.03533936e-01 -5.84284425e-01 2.76654661e-02
-1.50139534e+00 -5.85759640e-01 -4.03762728e-01 2.98407450e-02
-8.78321469e-01 3.60941961e-02 -4.79707509e-01 -6.75023735e-01
-4.11350995e-01 -1.16927290e+00 1.27736652e+00 1.91881821e-01
-7.38283515e-01 -7.50925243e-01 -7.31570721e-02 2.42800280e-01
5.04484475e-01 8.37564945e-01 1.12430596e+00 1.03083998e-01
-8.53616178e-01 1.19186893e-01 -4.31083888e-01 -3.77009243e-01
-1.12423472e-01 2.69330204e-01 -9.56871212e-01 2.17603855e-02
-2.28093386e-01 -5.10857344e-01 7.50567675e-01 3.65891010e-01
1.08557785e+00 -5.76325953e-01 -2.03707218e-01 2.86665767e-01
1.16638517e+00 1.83358371e-01 7.16234148e-01 8.23922604e-02
7.96504855e-01 8.85460198e-01 5.20348907e-01 4.94964391e-01
5.49495459e-01 9.04328585e-01 2.52801538e-01 -3.82517844e-01
-5.97305775e-01 -6.58165097e-01 4.07537013e-01 4.58580524e-01
2.05937102e-01 -5.69887459e-01 -8.01443696e-01 4.76112574e-01
-2.00823808e+00 -1.42894840e+00 -2.67702430e-01 1.68424845e+00
6.84696853e-01 -2.15628639e-01 1.62774287e-02 -1.46022171e-01
6.02891445e-01 4.53402072e-01 -4.61530656e-01 -4.39137548e-01
-4.72063541e-01 -2.48023883e-01 -1.33398905e-01 5.16461909e-01
-7.79281318e-01 8.28831255e-01 6.53951836e+00 8.39234173e-01
-1.18086660e+00 9.61312801e-02 9.61157262e-01 -5.89978635e-01
-7.81645238e-01 7.85875022e-02 -3.62144232e-01 4.10402119e-01
6.67159319e-01 -4.30613667e-01 2.20238224e-01 4.45316553e-01
7.09957361e-01 -3.64386350e-01 -1.10090661e+00 1.21088278e+00
3.10450822e-01 -1.77179468e+00 6.19329154e-01 -2.14317217e-01
8.89615715e-01 -7.12759793e-01 5.23716629e-01 -1.26401573e-01
4.28093463e-01 -1.39899826e+00 1.35490131e+00 8.58175278e-01
1.14577436e+00 -6.12760246e-01 -1.10270366e-01 1.35316655e-01
-1.27074623e+00 2.82469958e-01 1.88716948e-01 1.09745152e-01
7.63648808e-01 2.86988437e-01 -8.48684013e-01 2.45417908e-01
6.63117170e-01 6.48889601e-01 -7.29397655e-01 5.47122240e-01
-1.20031454e-01 2.43955836e-01 1.64785713e-01 1.36154994e-01
5.31574488e-02 -6.28183782e-02 6.04420722e-01 1.39081311e+00
4.62564439e-01 1.95426777e-01 1.68798417e-02 1.33229852e+00
8.91066343e-02 -1.17313012e-03 -8.29791367e-01 -3.68371218e-01
2.18997911e-01 1.27649021e+00 -7.74833798e-01 -5.23264527e-01
3.64150740e-02 1.32736802e+00 9.09503847e-02 6.46291971e-01
-9.53004301e-01 -3.86028849e-02 4.09320533e-01 3.20898622e-01
-3.70834880e-02 -2.53530174e-01 -5.64332247e-01 -9.75374877e-01
1.87577233e-01 -7.41445959e-01 1.98682442e-01 -1.68274713e+00
-8.65114868e-01 9.44233656e-01 2.64888942e-01 -1.34865010e+00
-6.76337779e-01 1.41898870e-01 -8.11073303e-01 7.87674665e-01
-1.12723279e+00 -1.46626151e+00 -5.25535285e-01 5.22998273e-01
8.95675778e-01 6.91567585e-02 6.48675323e-01 -4.84250411e-02
-2.87596434e-01 2.43891895e-01 -3.85730654e-01 -3.18046272e-01
7.13166177e-01 -1.15358543e+00 5.02172351e-01 8.37776303e-01
1.92700356e-01 1.65964380e-01 1.14662707e+00 -7.62879431e-01
-1.31544304e+00 -1.06913173e+00 6.50921583e-01 -4.99436021e-01
3.40193421e-01 -4.91581023e-01 -8.31058681e-01 4.63420928e-01
9.95370567e-01 -4.29803133e-01 3.15720290e-01 -5.92933774e-01
-4.68117267e-01 3.48456383e-01 -8.23662519e-01 1.23035800e+00
1.13470924e+00 -5.43808103e-01 -2.15717271e-01 2.91718394e-01
9.42059994e-01 -6.40265882e-01 -6.63946569e-01 1.03711255e-01
4.60234791e-01 -8.96630943e-01 1.02787030e+00 -5.62198520e-01
1.15191841e+00 -5.39264560e-01 5.90379015e-02 -1.23307860e+00
-1.17897652e-01 -7.77728498e-01 9.58024990e-03 1.49873388e+00
4.35914636e-01 1.02098286e-01 5.66505253e-01 7.79932082e-01
-1.79358590e-02 -3.00162584e-01 -6.07112944e-01 -2.95115143e-01
-2.50328153e-01 -2.95103580e-01 4.63247418e-01 8.90558243e-01
7.63239637e-02 7.10318327e-01 -7.14698613e-01 -5.41133955e-02
4.05850500e-01 3.97161901e-01 8.94298136e-01 -6.90113723e-01
-3.04446369e-01 -7.60118783e-01 3.55843417e-02 -7.77058065e-01
1.37481198e-01 -8.72880638e-01 1.27018645e-01 -1.96134830e+00
4.83978778e-01 -7.55695533e-03 4.77657169e-01 2.41232023e-01
-1.50242791e-01 4.12135333e-01 7.53144205e-01 1.49671823e-01
-6.67942107e-01 7.72386014e-01 1.56578946e+00 -2.33410820e-01
-1.39437288e-01 -6.24064267e-01 -5.83920121e-01 4.53510821e-01
3.88538808e-01 -1.80397406e-01 -7.95733809e-01 -5.49897552e-01
4.53877449e-01 6.78297579e-01 6.55115426e-01 -7.82114029e-01
2.00732633e-01 -4.07447696e-01 6.70390427e-01 -8.04604053e-01
7.03865469e-01 -4.66004908e-01 9.01506722e-01 1.99371457e-01
-7.99734831e-01 7.22446322e-01 3.25685024e-01 5.29051065e-01
-1.82340965e-01 2.73194402e-01 7.86092997e-01 5.64423390e-03
-5.49972713e-01 1.21262237e-01 -5.44374943e-01 2.76525378e-01
9.81661022e-01 -3.87778103e-01 -5.53146482e-01 -8.42385054e-01
-6.90410852e-01 1.77648574e-01 5.17348230e-01 9.03384149e-01
9.84330118e-01 -1.58796632e+00 -9.51721787e-01 -8.86803865e-02
1.12909034e-01 -1.05707541e-01 6.28543496e-01 4.58642632e-01
-2.87997812e-01 1.90634787e-01 -2.52169460e-01 -6.73709273e-01
-1.43380475e+00 3.42997104e-01 1.40681103e-01 -2.20464081e-01
-1.09326208e+00 5.45536101e-01 7.62151539e-01 2.97474921e-01
8.88027996e-02 -1.85318459e-02 -3.31628323e-01 2.51400083e-01
7.94711947e-01 9.21877921e-02 -3.61817598e-01 -6.92679584e-01
1.29338115e-01 4.58428711e-01 1.90309901e-02 -6.45161092e-01
1.20139694e+00 -3.78477424e-01 2.12735459e-01 7.50623941e-01
8.64435256e-01 -2.96088517e-01 -1.52002954e+00 7.12084845e-02
-1.50293514e-01 -3.62474054e-01 -2.19238579e-01 -9.75045562e-01
-1.04935884e+00 8.19275558e-01 5.74190199e-01 1.05467193e-01
9.61708188e-01 1.26467958e-01 6.37605190e-01 -9.64502841e-02
-1.67739600e-01 -8.70786786e-01 8.27838838e-01 2.59424269e-01
1.73322845e+00 -9.83014584e-01 -7.28170946e-02 -2.60266334e-01
-1.27696741e+00 9.56278443e-01 7.93056726e-01 1.42072082e-01
1.73627749e-01 4.07541633e-01 8.55691358e-02 -3.13031644e-01
-1.15033531e+00 1.41676545e-01 3.38108450e-01 5.79840004e-01
5.80075622e-01 -1.52843237e-01 4.02115770e-02 3.06712568e-01
-5.10856807e-01 -1.65637732e-01 7.31575727e-01 4.58690047e-01
2.87847389e-02 -6.93767965e-01 -3.96231413e-01 3.18827301e-01
-1.43391833e-01 -1.86280042e-01 -2.90043920e-01 5.23671806e-01
-3.18092138e-01 9.00885940e-01 3.78257960e-01 -1.27203763e-01
3.12188834e-01 -7.21939728e-02 3.88931811e-01 -4.41871136e-01
-6.43342495e-01 1.59813479e-01 3.06585103e-01 -4.88875836e-01
-4.04020578e-01 -6.04606032e-01 -1.34595752e+00 -1.88225776e-01
3.40504944e-01 -1.83160216e-01 6.69756234e-01 7.62793362e-01
4.69881088e-01 9.43117797e-01 2.40625605e-01 -1.06568730e+00
1.85664266e-01 -7.63745964e-01 -1.74545839e-01 7.03565001e-01
2.14314431e-01 -4.48660403e-01 1.15919732e-01 6.67284310e-01] | [11.178227424621582, 0.6317145824432373] |
57115ef3-4f07-44bb-9f7b-826e2238e2d6 | nominal-metaphor-generation-with-multitask | 2206.05195 | null | https://arxiv.org/abs/2206.05195v3 | https://arxiv.org/pdf/2206.05195v3.pdf | Nominal Metaphor Generation with Multitask Learning | Metaphor generation is a challenging task which can impact many downstream tasks such as improving user satisfaction with dialogue systems and story generation. This paper tackles the problem of Chinese nominal metaphor generation by introducing a multitask metaphor generation framework with self-training and metaphor identification mechanisms. Self-training addresses the data scarcity issue of metaphor datasets. That is, instead of solely relying on labelled metaphor datasets which are usually small in size, self-training helps identify potential metaphors from a large-scale unlabelled corpus for metaphor generation. The metaphor weighting mechanism enables our model to focus on the metaphor-related parts of the input (e.g., the comparison of the metaphor and comparator) during model learning and thus improves the metaphoricity of the generated metaphors. Our model is trained on an annotated corpus consisting of 6.3k sentences that contain diverse metaphorical expressions. Experimental results show that our model is able to generate metaphors with better readability and creativity compared to the baseline models, even in the situation where training data is insufficient. | ['Frank Geurin', 'Chenghua Lin', 'Yucheng Li'] | 2022-06-10 | null | null | null | null | ['story-generation'] | ['natural-language-processing'] | [ 8.73006359e-02 4.17366356e-01 -2.64991764e-02 -1.90622747e-01
-3.22458982e-01 -7.53132105e-01 8.72259617e-01 7.96064213e-02
-3.99372190e-01 6.83974385e-01 6.45114899e-01 -3.43810856e-01
2.01893315e-01 -9.07733440e-01 -1.86356679e-01 -3.84940445e-01
5.16228497e-01 8.46844196e-01 -4.88135695e-01 -8.88521135e-01
4.98006254e-01 -2.06203073e-01 -1.25825608e+00 5.44124365e-01
1.29796684e+00 6.32954776e-01 7.02581048e-01 4.23446000e-01
-7.51583815e-01 4.20606971e-01 -9.93045688e-01 -2.43485019e-01
-1.94968328e-01 -7.06394136e-01 -1.16635990e+00 -3.85981530e-01
-2.60937572e-01 -2.12643832e-01 3.87131929e-01 7.25812256e-01
5.31826973e-01 3.94357860e-01 7.57650495e-01 -8.01960886e-01
-1.04532528e+00 9.43718433e-01 -1.44559354e-01 -8.89771283e-02
4.59368378e-01 -2.83854012e-03 1.10564733e+00 -9.67835367e-01
6.80347085e-01 1.65328348e+00 3.48271012e-01 1.04948664e+00
-1.42546487e+00 -5.77609003e-01 -1.85170799e-01 1.86582908e-01
-9.62337315e-01 -2.29519814e-01 1.18615186e+00 -3.14130157e-01
1.19017911e+00 3.53448391e-01 7.52044022e-01 1.67164469e+00
-3.69122207e-01 4.96514887e-01 1.27120697e+00 -8.77035320e-01
4.02850121e-01 4.80030894e-01 -1.13961071e-01 3.71670187e-01
-4.68916714e-01 1.10937026e-03 -6.00284398e-01 -4.24450338e-01
8.89280438e-01 -5.30032396e-01 -1.75628886e-01 1.37127995e-01
-1.37632573e+00 1.42064369e+00 5.21254301e-01 6.74899042e-01
-2.57385194e-01 -2.80195564e-01 6.56707644e-01 3.79128337e-01
4.99871224e-01 1.45260763e+00 -3.06522816e-01 -4.01805222e-01
-5.40570438e-01 2.54060715e-01 8.04476976e-01 1.00319552e+00
3.89307797e-01 -5.20094149e-02 -2.95327902e-01 1.53635013e+00
-2.56231576e-01 3.14947218e-01 1.20654666e+00 -7.00034618e-01
2.48762935e-01 7.80113101e-01 1.48111939e-01 -7.82933176e-01
-2.56721795e-01 1.04234584e-01 -6.52915061e-01 -4.29192334e-01
2.31124356e-01 -2.37232521e-01 -5.06266475e-01 2.05711222e+00
1.41278192e-01 -2.46295437e-01 2.30201945e-01 9.90262091e-01
1.29364324e+00 6.05435371e-01 1.42756239e-01 -1.86827749e-01
1.69831550e+00 -1.12604034e+00 -6.60408080e-01 -5.15342176e-01
6.99934423e-01 -7.00183511e-01 2.14009809e+00 5.37829287e-02
-8.98069620e-01 -6.26568615e-01 -7.69989669e-01 -3.85307431e-01
-7.35355973e-01 1.18767746e-01 7.11263537e-01 4.10973847e-01
-8.57592940e-01 1.73454538e-01 2.98998728e-02 -6.40803695e-01
5.73457591e-02 -2.11317629e-01 5.25851771e-02 1.85628295e-01
-1.66908228e+00 1.35018623e+00 6.73885822e-01 -2.78561383e-01
-3.64757657e-01 -6.27952516e-01 -7.53825545e-01 1.14691600e-01
2.16296613e-01 -1.04369199e+00 1.12205052e+00 -1.31431687e+00
-1.57321739e+00 8.86569977e-01 -1.45321846e-01 -3.02366257e-01
2.74421841e-01 -1.60463467e-01 -9.82400179e-02 1.97059158e-02
3.22693467e-01 1.13184428e+00 6.75139487e-01 -1.11786389e+00
-1.62514463e-01 1.62386581e-01 3.75481755e-01 5.68588614e-01
-8.92057359e-01 -1.20180294e-01 5.49936518e-02 -8.87259245e-01
-3.57468486e-01 -8.01912963e-01 -8.70860294e-02 -4.64297980e-01
-3.10616821e-01 -5.63379943e-01 5.30940294e-01 -4.35609818e-01
1.01616740e+00 -1.96131909e+00 3.35566312e-01 -1.06617287e-01
-3.10528744e-02 -1.96232814e-02 -2.89256483e-01 6.66761339e-01
8.69917050e-02 2.25406364e-01 -6.15513250e-02 -2.65520036e-01
-7.50955865e-02 3.39906931e-01 -7.00273573e-01 -6.66411459e-01
1.74949050e-01 1.13277638e+00 -1.30258167e+00 -7.18477547e-01
2.75790989e-01 1.98297143e-01 -4.09700930e-01 5.55254459e-01
-5.28491974e-01 7.02582002e-01 -3.59982044e-01 4.40196484e-01
1.97906598e-01 -8.91754776e-02 1.34992003e-01 1.16092093e-01
7.54305050e-02 5.63829124e-01 -2.56336808e-01 2.22387457e+00
-1.28158391e+00 6.72998667e-01 -6.20161057e-01 -9.23661649e-01
1.13634312e+00 2.54212707e-01 5.36458800e-03 -9.21763182e-01
7.05362558e-02 2.07036465e-01 1.72119930e-01 -5.68166733e-01
7.17668176e-01 -7.02437997e-01 -3.65658760e-01 6.26600325e-01
-7.17265904e-02 -4.14660543e-01 1.69021860e-01 5.70929199e-02
8.41519952e-01 1.03023812e-01 3.53694975e-01 -5.36490262e-01
3.33777785e-01 3.39941680e-01 3.54887135e-02 5.41541517e-01
3.96674335e-01 1.36627525e-01 4.06031132e-01 -3.40576917e-01
-8.30245912e-01 -9.81288970e-01 -8.08513835e-02 1.30060172e+00
8.11288059e-02 -6.35682046e-01 -7.94940352e-01 -2.91872084e-01
-2.83100456e-01 1.47533619e+00 -8.52007747e-01 -4.36007082e-01
-5.20547271e-01 -5.19877791e-01 6.82774782e-01 3.91883522e-01
5.47360420e-01 -1.65762103e+00 -1.22329009e+00 3.02910358e-01
-8.59905124e-01 -8.81303012e-01 -5.29447012e-02 1.61790438e-02
-8.85710299e-01 -6.78782940e-01 -5.86269021e-01 -8.65614951e-01
5.75465381e-01 4.55789357e-01 1.59362257e+00 2.23231480e-01
-1.56517282e-01 -6.45011738e-02 -6.01322234e-01 -6.19149745e-01
-6.30242467e-01 -3.73068675e-02 -2.39420056e-01 -6.53869212e-01
5.09161711e-01 -6.95802748e-01 -5.03555715e-01 9.13059190e-02
-7.92914927e-01 7.54318118e-01 2.66349763e-01 1.51627374e+00
2.55544707e-02 -2.33207077e-01 7.83416152e-01 -7.65676916e-01
1.59691119e+00 -6.46767557e-01 -1.49003381e-03 3.94848257e-01
-6.01515472e-01 1.76258571e-02 8.72580767e-01 -8.78714025e-01
-1.21619320e+00 -4.71262157e-01 2.55573124e-01 -2.87097543e-01
5.88052385e-02 6.00805283e-01 3.94191928e-02 3.81563485e-01
9.64105785e-01 1.13084160e-01 1.15649708e-01 -4.31528986e-01
8.51707101e-01 6.10125542e-01 2.93357283e-01 -1.03542233e+00
3.60486895e-01 5.32407537e-02 -3.59769225e-01 -9.47918594e-01
-5.21962583e-01 -1.17294230e-01 -3.06398571e-01 -9.14958492e-02
6.43838465e-01 -6.43909097e-01 -5.08225083e-01 -2.13364497e-01
-1.47144878e+00 -3.83958369e-01 -3.77662539e-01 -8.41312483e-02
-6.17381215e-01 -3.46911512e-02 -2.95239687e-01 -9.25782442e-01
-6.44516647e-01 -7.56421685e-01 1.06416750e+00 3.30867380e-01
-1.06910205e+00 -1.36423552e+00 1.34078488e-01 2.67623991e-01
6.52655423e-01 3.54597121e-01 1.76610684e+00 -7.09131241e-01
2.40136623e-01 2.30117828e-01 -1.78387627e-01 -3.26600894e-02
4.02546972e-01 -5.13120174e-01 -1.04311109e+00 1.91636533e-01
3.11674029e-01 -8.59967589e-01 6.04425192e-01 1.16365083e-01
1.16094959e+00 -5.04377544e-01 -2.73429066e-01 1.90592557e-01
1.11569691e+00 2.06406057e-01 2.74959475e-01 4.36190993e-01
4.10936773e-01 1.07923603e+00 8.27287018e-01 4.26455647e-01
5.80661595e-01 8.09627652e-01 4.25707409e-03 -2.78564662e-01
1.52795821e-01 -4.31678802e-01 2.68451758e-02 5.21071434e-01
1.67044416e-01 -1.88673250e-02 -1.08471429e+00 7.60116816e-01
-1.86558819e+00 -6.86178148e-01 2.18063921e-01 1.89556861e+00
1.32576084e+00 -2.65874900e-02 2.14764744e-01 -3.14119190e-01
4.18559611e-01 5.99956419e-03 -6.15711808e-01 -1.03827882e+00
-2.72892475e-01 6.10116541e-01 -3.12797815e-01 4.82693374e-01
-5.55242300e-01 1.53873909e+00 5.61504841e+00 7.46480584e-01
-9.78394270e-01 1.74184620e-01 6.35087252e-01 -2.55553752e-01
-6.68663442e-01 1.52311310e-01 3.22187096e-02 5.48763871e-01
8.33680749e-01 -3.60729814e-01 6.78494811e-01 7.36922026e-01
3.14230949e-01 -1.94842681e-01 -1.14539647e+00 9.27573264e-01
2.70629209e-02 -1.24362135e+00 4.04344022e-01 -1.82347640e-01
4.17096555e-01 -6.59499228e-01 -7.51860067e-02 4.89351988e-01
2.04401389e-01 -1.44107294e+00 3.77830803e-01 2.77750552e-01
8.64485681e-01 -6.65580273e-01 5.59459209e-01 7.31948018e-01
-7.06218779e-01 -1.94452465e-01 -5.61956346e-01 -3.86995852e-01
1.79765418e-01 2.19637290e-01 -1.25543106e+00 3.40152949e-01
4.62049782e-01 2.52250075e-01 -6.02286279e-01 1.00994349e-01
-3.20676804e-01 4.41525310e-01 -5.32601923e-02 -5.21813750e-01
2.24854589e-01 -3.44182432e-01 3.88240904e-01 1.19857848e+00
2.73953855e-01 1.89040735e-01 7.97607675e-02 1.42568517e+00
1.55999601e-01 4.91679579e-01 -8.34654212e-01 -3.63336533e-01
7.91155517e-01 1.30427468e+00 -6.27303720e-01 -5.06528318e-01
-7.67945126e-02 1.07939017e+00 4.78899300e-01 1.45569131e-01
-6.72474146e-01 -1.93770438e-01 4.20039326e-01 -3.35550249e-01
-2.62981325e-01 4.77392003e-02 -8.93923163e-01 -8.80301952e-01
-1.17088445e-01 -8.70822847e-01 3.23364049e-01 -1.08992136e+00
-1.42907894e+00 6.92306161e-01 3.14137489e-01 -9.81008768e-01
-7.70042658e-01 -4.31883514e-01 -1.07666934e+00 1.22006214e+00
-1.11968601e+00 -1.49609482e+00 -3.09584141e-01 3.17794144e-01
9.42464948e-01 -3.43571901e-01 1.61046684e+00 -5.49714506e-01
-4.41557020e-02 4.93239999e-01 -2.94727594e-01 -1.02720998e-01
7.59162068e-01 -1.44210029e+00 6.76530600e-01 2.50408530e-01
7.69103765e-02 1.24043930e+00 9.67490256e-01 -5.72939098e-01
-9.55291748e-01 -4.57868427e-01 1.06812394e+00 -3.92336220e-01
7.52822816e-01 -5.75117111e-01 -8.25687766e-01 -1.31766796e-01
6.06755555e-01 -8.79180789e-01 1.16263616e+00 5.30348837e-01
-4.84275401e-01 6.29652143e-01 -1.02279675e+00 1.02393544e+00
1.28086865e+00 -5.61087549e-01 -1.19517493e+00 5.86282432e-01
8.42880607e-01 -3.98878336e-01 -4.47649449e-01 -1.25626609e-01
5.88973403e-01 -7.10865021e-01 1.01583445e+00 -1.01916993e+00
9.95589375e-01 2.97473371e-01 4.72871065e-02 -1.83004713e+00
-1.01527505e-01 -6.33853436e-01 -6.26637489e-02 1.33381271e+00
3.20465118e-01 -4.12532926e-01 1.82647720e-01 6.29239678e-01
9.42834020e-02 -7.76852131e-01 -9.65962887e-01 -5.92014313e-01
4.76952881e-01 1.51003480e-01 7.94241071e-01 1.16214824e+00
6.22425318e-01 1.13946116e+00 -1.13263227e-01 -8.36270332e-01
-3.35606374e-02 6.09426260e-01 6.68565214e-01 -1.23930991e+00
-3.12908053e-01 -7.65751779e-01 3.81228983e-01 -7.78027356e-01
6.69820249e-01 -9.10804927e-01 -3.13609689e-02 -1.47914088e+00
6.27538040e-02 -3.32067221e-01 -5.53879887e-02 4.33410704e-01
-3.83877605e-01 -2.13474125e-01 1.22447267e-01 1.22697592e-01
6.26982972e-02 7.19782591e-01 1.31875741e+00 -3.14324014e-02
-4.95085776e-01 -4.52888161e-01 -1.18110180e+00 6.22512102e-01
1.01053774e+00 -6.69499114e-02 -9.71844435e-01 -4.50788498e-01
1.09797314e-01 8.62479955e-02 4.73235101e-01 -2.44082168e-01
-2.15955362e-01 -5.22894144e-01 3.13236594e-01 1.29299145e-02
5.14714122e-01 -4.34411228e-01 -2.23682344e-01 3.10949177e-01
-7.39330351e-01 5.01186311e-01 3.81996334e-01 2.53826559e-01
3.85781452e-02 -3.29998285e-01 5.05122662e-01 -6.86730921e-01
-5.53066134e-01 -7.35549688e-01 -1.53419197e-01 3.63513887e-01
6.63853168e-01 -3.63330632e-01 -4.09442484e-01 -4.61419284e-01
-1.48629189e-01 3.04247111e-01 6.63772523e-01 1.02181184e+00
7.20461369e-01 -1.56931984e+00 -5.42904973e-01 -2.65595466e-02
4.97811943e-01 5.12020774e-02 5.65367974e-02 2.69059807e-01
3.91797982e-02 3.77424002e-01 -5.38786173e-01 -3.33007306e-01
-1.07681203e+00 5.82491934e-01 6.58547357e-02 -2.42588773e-01
-5.60627341e-01 1.05830550e+00 3.56001079e-01 -3.69663447e-01
-1.74757063e-01 -9.07062739e-02 -6.06350958e-01 2.58701295e-01
3.15175891e-01 1.08858332e-01 -4.53693211e-01 -4.50271457e-01
-1.62861601e-01 1.37025878e-01 -1.74971782e-02 -6.74706519e-01
1.04098487e+00 8.46377760e-02 -3.89562845e-01 8.38487089e-01
8.06630850e-01 -3.11929017e-01 -6.23077393e-01 -1.54283419e-01
2.11548150e-01 -5.51321626e-01 -2.91206539e-01 -1.16659975e+00
-2.08504945e-01 1.01427174e+00 2.24012733e-01 2.48835206e-01
1.13051069e+00 4.06210916e-03 9.44805801e-01 4.60275501e-01
2.62453198e-01 -1.49599791e+00 5.29289961e-01 7.13169754e-01
1.33204222e+00 -1.16011500e+00 -4.00890321e-01 -7.73956180e-02
-1.04622841e+00 1.25317121e+00 9.73746359e-01 6.74893633e-02
-4.02190089e-02 -2.36798674e-01 2.41260231e-01 -1.82195574e-01
-1.02380121e+00 1.79526564e-02 3.05574208e-01 7.07411706e-01
8.67833734e-01 3.13141525e-01 -8.02406788e-01 8.00548553e-01
-9.98835146e-01 -4.31598634e-01 2.52046674e-01 5.85709512e-01
-1.17624447e-01 -1.24640822e+00 -1.55919731e-01 3.76509964e-01
3.85867506e-01 -6.22904480e-01 -1.22843611e+00 5.15775621e-01
-1.37856109e-02 1.13860595e+00 -3.78519781e-02 -2.93282241e-01
1.62279844e-01 4.83675063e-01 2.99859911e-01 -7.67391801e-01
-7.88400114e-01 -3.87748480e-02 3.80411714e-01 -2.37289414e-01
-2.81221360e-01 -3.64889592e-01 -1.29562044e+00 -9.73116457e-02
-1.86212063e-01 3.73367578e-01 4.48878437e-01 1.12402463e+00
3.32836539e-01 4.09985274e-01 6.25560820e-01 -6.21075094e-01
-6.46083236e-01 -1.46728325e+00 5.18108085e-02 6.27557755e-01
-6.56505525e-02 -9.43261325e-01 6.07570671e-02 -1.18706338e-01] | [11.247831344604492, 9.077905654907227] |
648d9ccc-7caa-4132-a7cf-d08bbf66aca5 | non-convex-optimizations-for-machine-learning | 2306.16557 | null | https://arxiv.org/abs/2306.16557v1 | https://arxiv.org/pdf/2306.16557v1.pdf | Non-Convex Optimizations for Machine Learning with Theoretical Guarantee: Robust Matrix Completion and Neural Network Learning | Despite the recent development in machine learning, most learning systems are still under the concept of "black box", where the performance cannot be understood and derived. With the rise of safety and privacy concerns in public, designing an explainable learning system has become a new trend in machine learning. In general, many machine learning problems are formulated as minimizing (or maximizing) some loss function. Since real data are most likely generated from non-linear models, the loss function is non-convex in general. Unlike the convex optimization problem, gradient descent algorithms will be trapped in spurious local minima in solving non-convex optimization. Therefore, it is challenging to provide explainable algorithms when studying non-convex optimization problems. In this thesis, two popular non-convex problems are studied: (1) low-rank matrix completion and (2) neural network learning. | ['Shuai Zhang'] | 2023-06-28 | null | null | null | null | ['low-rank-matrix-completion', 'matrix-completion'] | ['methodology', 'methodology'] | [ 8.40975493e-02 3.39727193e-01 -2.41115674e-01 -6.79777026e-01
-7.52081692e-01 -4.35538769e-01 1.25454351e-01 -7.71397725e-03
-7.94738382e-02 1.03808308e+00 1.91167314e-02 -2.74233282e-01
-2.92369813e-01 -4.23781604e-01 -8.75035405e-01 -7.02909827e-01
5.50700948e-02 4.28942323e-01 -7.30194092e-01 1.06309280e-01
5.32757156e-02 2.27353260e-01 -1.09608340e+00 2.99193114e-01
9.72690701e-01 7.60577738e-01 -1.71000883e-02 3.16789210e-01
3.82730626e-02 5.08940339e-01 -1.90982908e-01 -5.23682177e-01
3.98229629e-01 -3.98176551e-01 -6.21136844e-01 1.06882840e-01
5.34869373e-01 5.09707518e-02 -3.54046583e-01 1.39559579e+00
2.11946771e-01 6.40228689e-02 4.97334898e-01 -1.62778080e+00
-7.43779421e-01 3.72829527e-01 -4.94858414e-01 -4.71881360e-01
2.94453174e-01 -3.85262072e-01 1.20952916e+00 -1.19622886e+00
3.42167228e-01 1.33703017e+00 5.38893044e-01 5.28121531e-01
-1.30571401e+00 -6.54565990e-01 1.54169291e-01 2.61730909e-01
-1.51614189e+00 -2.79080212e-01 4.48986977e-01 -4.17376250e-01
3.50483537e-01 6.13026321e-01 3.60360473e-01 8.48826647e-01
2.32351199e-01 9.25672233e-01 1.03819013e+00 -1.93606481e-01
1.80954576e-01 5.90682209e-01 1.04668826e-01 6.94370031e-01
5.48149705e-01 1.99562646e-02 -6.74412251e-01 -2.42963850e-01
4.23947603e-01 9.75790173e-02 -4.09183621e-01 -6.29580140e-01
-9.92662191e-01 9.81711030e-01 3.36110920e-01 -5.69315106e-02
-2.89927036e-01 1.34719357e-01 3.26188914e-02 3.78669947e-01
3.74674559e-01 4.60665792e-01 -4.28627938e-01 -5.75287901e-02
-9.91196275e-01 2.05436260e-01 1.10272479e+00 7.47373521e-01
8.95411372e-01 2.19264030e-01 1.63400024e-01 5.48758328e-01
5.57967961e-01 3.48942548e-01 1.74117535e-01 -6.00792706e-01
6.23758256e-01 6.91350043e-01 1.68072939e-01 -1.33762956e+00
-3.92689258e-01 -7.75343895e-01 -1.14586031e+00 1.32337064e-01
5.31000316e-01 -2.31243774e-01 -5.51433384e-01 1.60910356e+00
3.21839720e-01 1.56898990e-01 -8.15235171e-03 1.21857178e+00
5.84914446e-01 8.43289018e-01 -1.50330096e-01 -4.60280657e-01
7.66372442e-01 -8.27801526e-01 -9.86872613e-01 -2.84346431e-01
6.71522439e-01 -6.12410188e-01 1.02644658e+00 6.55928969e-01
-8.48965824e-01 -8.17004740e-02 -1.05375957e+00 -1.60825908e-01
-5.65674603e-02 1.12717219e-01 9.55094099e-01 7.07376659e-01
-6.28259182e-01 4.72223461e-01 -6.18441582e-01 4.34309095e-02
3.14141691e-01 6.81316257e-01 -6.32534742e-01 -3.67001951e-01
-9.48140800e-01 8.34430337e-01 2.83383071e-01 5.42446315e-01
-1.00828612e+00 -6.78425729e-01 -5.07091284e-01 7.57769719e-02
5.00144720e-01 -6.54015899e-01 6.51184797e-01 -1.10109198e+00
-1.12296927e+00 8.59774649e-01 -4.35592175e-01 -2.71743238e-01
6.77204967e-01 -6.44661784e-01 -1.96494609e-01 -6.00215077e-01
-1.89705119e-01 2.88301483e-02 9.87409115e-01 -1.42512333e+00
-3.06833237e-01 -5.05108535e-01 -5.17027788e-02 3.07559907e-01
-3.05371881e-01 -9.30525884e-02 -2.71800786e-01 -4.21783000e-01
3.86731595e-01 -1.23317897e+00 -4.30721968e-01 2.31170744e-01
-8.29719424e-01 3.20145413e-02 1.04727221e+00 -6.44560456e-01
1.22677588e+00 -2.01967359e+00 2.57619798e-01 4.45614994e-01
4.28822607e-01 9.76564586e-02 -1.21400647e-01 2.05270305e-01
-3.53034765e-01 2.51766980e-01 -3.15709382e-01 -2.95366645e-01
-2.12338995e-02 3.16098094e-01 -3.04855555e-01 8.44790995e-01
-1.21908888e-01 5.35866082e-01 -7.43514478e-01 -2.01497063e-01
6.76624030e-02 4.21570361e-01 -5.66983044e-01 1.69083714e-01
6.49168193e-02 4.85816658e-01 -8.00301552e-01 6.32852495e-01
7.00968146e-01 -3.92950714e-01 8.82328972e-02 -1.71695039e-01
1.40195101e-01 -3.20825547e-01 -1.59528482e+00 1.54302907e+00
-2.81992346e-01 9.02687132e-01 4.80988711e-01 -1.49095738e+00
8.73082757e-01 4.71974611e-01 5.87056816e-01 -1.22012623e-01
-1.68417804e-02 3.22036892e-02 -9.69930738e-02 -6.30100250e-01
3.85729700e-01 -1.83896631e-01 3.94961804e-01 4.47395056e-01
-5.99895537e-01 5.65357581e-02 -2.29630858e-01 9.59065557e-02
7.88932025e-01 -3.12328011e-01 1.85144901e-01 -3.63222927e-01
6.38028741e-01 7.94378370e-02 8.66550207e-01 4.56284821e-01
1.25403374e-01 8.36369276e-01 4.71201897e-01 -6.03896022e-01
-7.90521383e-01 -8.95364344e-01 -2.77742773e-01 9.90288079e-01
1.73830897e-01 -3.94993573e-01 -7.92717993e-01 -5.61815917e-01
1.23449281e-01 6.02713883e-01 -2.37772852e-01 -1.18939161e-01
-3.96484017e-01 -9.86046314e-01 2.87668258e-01 -5.63312396e-02
3.48183423e-01 -2.96476513e-01 1.77196875e-01 -1.43819870e-02
-2.49414235e-01 -9.87376511e-01 -6.43666863e-01 1.29121088e-03
-8.70755553e-01 -1.10148811e+00 -3.90656769e-01 -5.19984305e-01
1.27103961e+00 4.91809905e-01 1.02940440e+00 3.23352158e-01
-4.35354948e-01 3.89597267e-01 -1.14246227e-01 -7.80162454e-01
2.41342224e-02 3.66567373e-02 4.22530651e-01 4.99248475e-01
2.56026000e-01 -1.56897902e-01 -4.48842287e-01 4.39785510e-01
-9.33039069e-01 2.05072731e-01 6.09194696e-01 9.21903670e-01
9.33769524e-01 3.21220219e-01 5.05804837e-01 -1.12554276e+00
8.46828640e-01 -6.95036769e-01 -6.12293184e-01 6.51989579e-01
-8.60398114e-01 3.02784950e-01 6.01918519e-01 -4.47502702e-01
-1.01408231e+00 5.57050407e-01 5.24389446e-01 -2.86054850e-01
1.83086276e-01 8.30638766e-01 -5.42512059e-01 -1.02433614e-01
7.89278805e-01 6.49998412e-02 7.53137022e-02 -4.87296045e-01
3.94952685e-01 6.68422580e-01 1.18697092e-01 -6.05103433e-01
1.20109248e+00 4.47912514e-01 3.58253747e-01 -1.03014350e+00
-1.18543601e+00 -2.52170056e-01 -4.38535094e-01 -3.14087361e-01
7.74664819e-01 -8.78560662e-01 -6.00072861e-01 -2.29298603e-02
-1.06851757e+00 2.17815071e-01 -1.89319059e-01 9.56295729e-01
-3.02697837e-01 2.72999972e-01 -1.05448820e-01 -9.25267458e-01
-1.72490656e-01 -1.04848886e+00 6.85641170e-01 3.29635173e-01
-2.96675086e-01 -1.17089891e+00 6.09826893e-02 8.31689239e-01
2.80009538e-01 3.03695112e-01 9.03643250e-01 -6.23927414e-01
-8.66020501e-01 -6.45311236e-01 -2.58369237e-01 4.07970041e-01
2.89196782e-02 -1.11569278e-01 -1.03556657e+00 -6.00928605e-01
2.05086738e-01 -3.98434013e-01 4.18909073e-01 4.66191351e-01
1.35344458e+00 -7.21731246e-01 -1.30563498e-01 7.57516086e-01
1.30219805e+00 -2.31170118e-01 3.88890147e-01 -1.23937065e-02
8.05929065e-01 7.98772871e-01 5.91447234e-01 5.37256718e-01
2.13658422e-01 3.92681330e-01 7.70749450e-01 -2.20720172e-01
5.29022098e-01 -4.44445103e-01 3.35995317e-01 6.33711636e-01
-1.43438905e-01 -2.89870426e-02 -8.08443367e-01 2.15026364e-01
-2.26153207e+00 -9.57058489e-01 -6.21421576e-01 2.51669478e+00
7.45790482e-01 -4.33907092e-01 -3.55048537e-01 7.06683844e-03
8.65121067e-01 -1.75198376e-01 -8.58824551e-01 -3.82761985e-01
-2.57435173e-01 -1.99761957e-01 5.06950796e-01 4.77717876e-01
-9.95952606e-01 7.13373661e-01 6.66312313e+00 6.83115482e-01
-1.13187516e+00 -1.12162024e-01 7.89128006e-01 -1.61279604e-01
-4.56445783e-01 2.96602190e-01 -6.53816640e-01 1.33947879e-01
4.95639205e-01 -4.34807032e-01 7.62293577e-01 1.16594946e+00
6.24305904e-01 1.32082596e-01 -1.49683166e+00 1.39250088e+00
-1.21528804e-01 -9.99783456e-01 -2.39739493e-02 2.89261699e-01
1.08533514e+00 -2.22001240e-01 3.13663691e-01 9.95966941e-02
1.46356255e-01 -1.77564228e+00 4.58546788e-01 4.90937114e-01
4.83379424e-01 -5.83069444e-01 4.48231220e-01 7.35976934e-01
-7.94555426e-01 -4.06226888e-02 -7.20628381e-01 -7.71731585e-02
3.47535610e-02 1.12655377e+00 -5.03242075e-01 3.58827740e-01
3.62245560e-01 6.93529725e-01 -1.22284316e-01 1.22508121e+00
-3.52429092e-01 7.24967718e-01 -2.77982175e-01 -1.25285061e-02
1.37403607e-01 -7.20607996e-01 5.92737913e-01 8.52424681e-01
3.27204138e-01 9.54417437e-02 2.19490856e-01 1.01353037e+00
-3.37996334e-01 2.98818409e-01 -8.22637796e-01 -1.83418185e-01
2.02083036e-01 1.34157717e+00 -3.43222827e-01 1.87020898e-01
-6.62764549e-01 7.70147204e-01 1.93973303e-01 5.75544238e-01
-6.98857367e-01 6.17574267e-02 8.45220029e-01 1.01172574e-01
-5.61307430e-01 -3.05820227e-01 -5.82691908e-01 -1.31657529e+00
1.62350968e-01 -9.76300776e-01 3.27596366e-01 -4.16769743e-01
-1.35285544e+00 2.02312335e-01 -2.73978829e-01 -1.09428346e+00
6.04689792e-02 -5.22787094e-01 -6.25936925e-01 7.69383669e-01
-1.18423426e+00 -7.35681832e-01 -2.67935127e-01 8.70869696e-01
2.82793105e-01 -2.44818926e-01 7.36759543e-01 4.62307930e-01
-8.61525893e-01 5.78164279e-01 5.88861227e-01 -2.31879503e-01
6.02183700e-01 -1.17881858e+00 -5.42123079e-01 6.39921844e-01
3.53837550e-01 7.50190318e-01 9.42252815e-01 -5.16865075e-01
-1.91735768e+00 -1.07603979e+00 9.29242313e-01 -2.98404932e-01
5.14014781e-01 -3.39150995e-01 -1.00105333e+00 5.87089479e-01
-7.60954320e-02 1.11670457e-01 8.78680110e-01 4.20979023e-01
-2.01550394e-01 -3.07237953e-01 -1.23527980e+00 3.47461730e-01
6.83104992e-01 -6.78271472e-01 -6.85795024e-02 9.37695801e-01
1.57296896e-01 -3.24536473e-01 -6.39494896e-01 9.57338884e-02
5.46437025e-01 -5.94332874e-01 7.88326681e-01 -1.21144342e+00
2.30761528e-01 -3.58993798e-01 -2.69825011e-01 -1.24745226e+00
-3.66224721e-02 -9.68115985e-01 -1.87469348e-02 1.00795567e+00
6.99308276e-01 -4.22937661e-01 1.40239263e+00 1.57139432e+00
1.18680187e-01 -8.05519402e-01 -9.66542184e-01 -7.89162338e-01
-6.79087937e-02 -4.47280884e-01 2.15729654e-01 1.25940645e+00
-1.51800234e-02 3.88935119e-01 -9.73187387e-01 2.94373989e-01
9.74181592e-01 4.81885225e-02 8.94550681e-01 -1.21585655e+00
-1.89464882e-01 -4.55305204e-02 -3.12115908e-01 -1.12889111e+00
2.99093783e-01 -1.08123529e+00 -2.73601040e-02 -1.36196983e+00
4.76077527e-01 -4.85263586e-01 -1.03246845e-01 4.83594179e-01
-1.80776641e-01 -3.32963377e-01 1.75821513e-01 3.38919938e-01
-4.60432529e-01 7.09587455e-01 1.33349395e+00 -2.14917362e-01
-1.58367857e-01 3.28349829e-01 -7.71330357e-01 9.20003414e-01
7.27995157e-01 -7.82600343e-01 -6.61302447e-01 -6.26087725e-01
6.37092590e-01 2.64596283e-01 1.10866994e-01 -6.90810561e-01
4.16930407e-01 -5.16529977e-01 2.41476530e-03 -3.10697794e-01
3.52236003e-01 -1.20612645e+00 5.01000226e-01 4.07744020e-01
-4.08975244e-01 3.48788803e-03 -3.09055567e-01 7.08365262e-01
-4.36516792e-01 -2.97216475e-01 6.82198882e-01 -1.38404621e-02
-2.49483839e-01 6.49630785e-01 -1.86691523e-01 2.13711917e-01
1.07245874e+00 -2.07730114e-01 -1.48029074e-01 -7.73586273e-01
-7.05117404e-01 4.95373547e-01 2.37274896e-02 3.55715483e-01
7.13566899e-01 -1.25404346e+00 -9.71115828e-01 -1.08642623e-01
-2.58277684e-01 2.03106731e-01 1.50455147e-01 6.29925072e-01
-4.47527826e-01 5.55424213e-01 2.25513190e-01 -3.82842392e-01
-1.35350406e+00 2.99640447e-01 1.86037868e-01 -2.41740867e-01
-1.52227193e-01 7.88254023e-01 4.26833719e-01 -4.40104574e-01
5.43512642e-01 1.89042389e-01 1.43876433e-01 -2.21976742e-01
3.79955053e-01 6.23836279e-01 -2.51389533e-01 -6.25825405e-01
-4.01578397e-02 2.92736083e-01 -3.40229012e-02 2.98925459e-01
1.47767448e+00 -1.79596394e-02 -3.69775325e-01 2.51110852e-01
1.40779531e+00 -1.28995910e-01 -9.41478908e-01 -3.38805139e-01
2.38006949e-01 -6.15976155e-01 1.19398326e-01 -3.44364375e-01
-1.27560878e+00 8.48139644e-01 5.35042048e-01 1.94178551e-01
8.17639112e-01 -3.33682865e-01 5.42518795e-01 9.01822090e-01
3.02102566e-01 -1.22615349e+00 -7.42295310e-02 3.96622777e-01
9.80524719e-01 -1.63316011e+00 4.16767776e-01 -5.03215253e-01
-6.43610716e-01 1.01281619e+00 5.31516910e-01 7.39784166e-02
9.51853156e-01 8.11349377e-02 -1.07851988e-02 -1.66392371e-01
-6.10724330e-01 3.97438347e-01 5.51206291e-01 3.77413750e-01
5.57925224e-01 1.68296382e-01 -4.39579397e-01 7.85406351e-01
-3.99934590e-01 -1.38956770e-01 4.18484956e-01 4.94677454e-01
-3.61878723e-01 -1.24017656e+00 -5.59129119e-01 5.58816314e-01
-6.38214290e-01 1.16717055e-01 -5.96252382e-01 6.09448195e-01
-6.56378418e-02 1.17650509e+00 -5.16669989e-01 -3.69363487e-01
1.74972728e-01 -2.87345409e-01 -4.05835994e-02 -5.56423426e-01
-2.26727307e-01 -4.53121960e-01 -1.93273611e-02 -6.82314396e-01
-2.22288236e-01 -5.70916355e-01 -1.35791540e+00 -5.76903224e-01
-5.98792255e-01 3.01115364e-01 1.12519205e+00 9.99577761e-01
1.46801278e-01 3.82458940e-02 7.52308249e-01 -3.58100593e-01
-9.32741404e-01 -3.82753521e-01 -8.24022591e-01 5.12429595e-01
2.20624447e-01 -4.32503849e-01 -3.50906491e-01 -3.34649235e-02] | [7.27926778793335, 4.387484073638916] |
1059fa8e-8a26-4f60-9c3e-c3fe7a475ca9 | verifiably-safe-exploration-for-end-to-end | 2007.01223 | null | https://arxiv.org/abs/2007.01223v1 | https://arxiv.org/pdf/2007.01223v1.pdf | Verifiably Safe Exploration for End-to-End Reinforcement Learning | Deploying deep reinforcement learning in safety-critical settings requires developing algorithms that obey hard constraints during exploration. This paper contributes a first approach toward enforcing formal safety constraints on end-to-end policies with visual inputs. Our approach draws on recent advances in object detection and automated reasoning for hybrid dynamical systems. The approach is evaluated on a novel benchmark that emphasizes the challenge of safely exploring in the presence of hard constraints. Our benchmark draws from several proposed problem sets for safe learning and includes problems that emphasize challenges such as reward signals that are not aligned with safety constraints. On each of these benchmark problems, our algorithm completely avoids unsafe behavior while remaining competitive at optimizing for as much reward as is safe. We also prove that our method of enforcing the safety constraints preserves all safe policies from the original environment. | ['Armando Solar-Lezama', 'Nghia Hoang', 'Nathan Fulton', 'Sara Magliacane', 'Subhro Das', 'Nathan Hunt'] | 2020-07-02 | null | null | null | null | ['safe-exploration'] | ['robots'] | [ 2.44318414e-02 5.28330564e-01 -3.60689193e-01 -9.06883851e-02
-7.30075538e-01 -9.61444795e-01 7.90229261e-01 3.67282592e-02
-5.23250222e-01 1.10923660e+00 -1.14583366e-01 -5.51681399e-01
-2.80367076e-01 -5.13842463e-01 -1.01814950e+00 -6.42111659e-01
-6.24666095e-01 2.60785341e-01 4.96066988e-01 -3.84094447e-01
1.41095877e-01 5.52260816e-01 -1.68489337e+00 3.03797871e-02
5.77483356e-01 8.72333169e-01 -3.19021612e-01 8.44194591e-01
4.72931087e-01 1.28990245e+00 -5.53430736e-01 1.38431592e-02
6.89356565e-01 -2.10457176e-01 -8.43281507e-01 -1.77438661e-01
6.18323684e-01 -6.17799997e-01 -1.32804617e-01 1.18525493e+00
1.53861955e-01 3.28113049e-01 2.81482279e-01 -2.11075521e+00
-2.52887905e-01 6.82581365e-01 -2.28392184e-01 -6.72973366e-03
1.32544622e-01 9.73427296e-01 9.44629908e-01 -4.83934581e-02
6.42575741e-01 1.25903940e+00 2.02292308e-01 9.34764504e-01
-1.26352513e+00 -4.51988488e-01 6.17574930e-01 2.96603888e-02
-9.99013424e-01 -4.53521162e-01 3.34693998e-01 -5.55863202e-01
1.46615934e+00 4.97709811e-01 7.58876860e-01 1.38879287e+00
4.23189670e-01 5.81736267e-01 1.17466140e+00 -2.49808624e-01
7.96061575e-01 4.26491573e-02 -1.58719823e-01 7.57730842e-01
3.51586282e-01 1.16511178e+00 -3.72720987e-01 -1.91473857e-01
5.30428767e-01 -5.98807156e-01 1.98817790e-01 -7.26613998e-01
-9.10663068e-01 7.12156951e-01 1.22724913e-01 -2.78620034e-01
1.27699867e-01 8.69488537e-01 4.69110191e-01 3.10859203e-01
-2.26763025e-01 8.05616379e-01 -2.36381248e-01 -2.36100510e-01
-5.45536458e-01 6.73999310e-01 9.01447415e-01 1.05403793e+00
2.16907293e-01 6.13163114e-01 -4.38477755e-01 -4.60108221e-02
2.81248003e-01 4.88931835e-01 -2.75593936e-01 -1.51234233e+00
1.50121555e-01 1.08205445e-01 5.87901413e-01 -4.69659448e-01
-3.54968965e-01 -2.51331270e-01 -3.70384045e-02 1.38953245e+00
2.55400658e-01 -5.84782302e-01 -9.03860211e-01 2.17984843e+00
4.87260044e-01 1.62491024e-01 1.52505502e-01 8.14479291e-01
6.40794039e-02 5.94597816e-01 3.61745000e-01 -3.16405743e-01
7.48163760e-01 -7.60134757e-01 -4.92137074e-01 -2.66667783e-01
3.52207243e-01 -8.69968906e-02 1.07294095e+00 6.71088159e-01
-1.29709411e+00 -8.77347663e-02 -1.21292949e+00 2.23694697e-01
-1.44301683e-01 -4.45242047e-01 5.17109931e-01 5.19132733e-01
-1.30278862e+00 5.52420616e-01 -1.04234695e+00 5.45262992e-02
4.00634915e-01 5.48926771e-01 -7.45602921e-02 7.20632672e-01
-9.35342133e-01 1.38124275e+00 4.98655349e-01 -2.26517871e-01
-1.99166155e+00 -6.56443357e-01 -9.29659128e-01 1.13744125e-01
1.14332151e+00 -2.27289081e-01 1.80381489e+00 -8.67302418e-01
-1.66781235e+00 4.47150379e-01 6.49443388e-01 -8.55171323e-01
1.08128393e+00 -4.92806554e-01 -4.87411320e-02 5.31660467e-02
1.76910181e-02 8.37486386e-01 9.49867547e-01 -1.47359383e+00
-9.12747324e-01 6.93868101e-02 6.55290723e-01 1.34670943e-01
-8.18760693e-02 -1.28328890e-01 9.13295150e-02 -3.29255581e-01
-9.85814691e-01 -1.18599451e+00 -6.02709353e-01 3.21185231e-01
-5.43617189e-01 -1.10800929e-01 8.88091922e-01 9.57804620e-02
8.44006360e-01 -1.89956105e+00 2.14090720e-01 3.74606878e-01
1.98805392e-01 3.65041830e-02 -1.82248771e-01 2.66886055e-01
1.59246519e-01 1.35171086e-01 -3.65540206e-01 7.10608065e-02
5.40481269e-01 4.53749865e-01 -9.98824894e-01 5.63653708e-01
2.49428034e-01 7.50371695e-01 -9.83220041e-01 -5.28264999e-01
2.16504887e-01 -1.41159534e-01 -7.48145700e-01 4.26162094e-01
-8.92944336e-01 2.88349330e-01 -2.60669202e-01 4.51368451e-01
2.46846780e-01 3.41914922e-01 3.45581532e-01 5.84102750e-01
-3.40059072e-01 -1.85801074e-01 -1.26242650e+00 1.10316503e+00
-9.07286927e-02 3.50048453e-01 3.24592113e-01 -6.86394691e-01
4.76308525e-01 -3.48763652e-02 5.19398630e-01 -7.60209382e-01
1.57303393e-01 -1.20229274e-01 3.06188241e-02 -4.39075053e-01
3.82885963e-01 -1.71053633e-01 -4.29959297e-01 4.31172162e-01
2.96453107e-02 -4.26318377e-01 2.59219646e-01 7.62778372e-02
1.28556871e+00 7.58448303e-01 3.52709770e-01 -6.43881798e-01
1.71205759e-01 2.70399958e-01 7.02185154e-01 1.30057311e+00
-6.05707049e-01 -2.56314158e-01 1.11217976e+00 -4.91743863e-01
-1.16612089e+00 -1.26062930e+00 1.12491675e-01 1.18194413e+00
3.31329972e-01 -3.90990466e-01 -7.88873196e-01 -1.07010829e+00
2.41714343e-01 9.93684113e-01 -8.77229095e-01 -4.54437524e-01
-5.39921165e-01 -3.10004205e-02 8.08147788e-01 7.21130013e-01
7.90250301e-02 -1.19676030e+00 -1.76732969e+00 -3.57081480e-02
5.54980397e-01 -8.85896444e-01 -2.54554331e-01 5.84871948e-01
-3.13516676e-01 -1.32567060e+00 4.34097052e-02 -3.96194041e-01
4.86491680e-01 -2.63800502e-01 9.71795082e-01 2.07593560e-01
-3.91307652e-01 6.87747598e-01 -1.09298937e-02 -6.28295839e-01
-5.79969823e-01 -3.18750858e-01 1.97606400e-01 -6.38116062e-01
-2.02990964e-01 -1.03319168e-01 -2.52004545e-02 1.00647986e-01
-9.61291850e-01 3.27776978e-03 -6.56777844e-02 7.07907856e-01
6.59793377e-01 -8.37624632e-03 2.69316018e-01 -4.70372200e-01
7.61120379e-01 -3.95777762e-01 -1.56656218e+00 3.65898818e-01
-6.14643216e-01 5.68521261e-01 8.07639539e-01 -6.05172455e-01
-9.43086982e-01 3.58157128e-01 1.96780458e-01 -5.26606917e-01
-1.83690891e-01 -1.77898198e-01 6.00275584e-02 -2.35693023e-01
8.83436799e-01 -2.29834661e-01 -2.26122513e-02 2.36633167e-01
3.39848131e-01 -1.50124699e-01 5.15512466e-01 -1.31956196e+00
9.52973723e-01 3.63219410e-01 4.27971452e-01 -3.60902369e-01
-5.17597675e-01 3.76442015e-01 -6.29091710e-02 -4.58712637e-01
6.97745562e-01 -4.78220254e-01 -1.46761668e+00 6.21599853e-02
-6.55507326e-01 -1.06895590e+00 -8.04017782e-01 4.25105281e-02
-1.29769158e+00 1.37403429e-01 -3.43032748e-01 -1.28379488e+00
-5.63041791e-02 -1.40970349e+00 8.97097945e-01 1.70756489e-01
-3.44910055e-01 -5.96587181e-01 5.19948483e-01 -5.04574656e-01
3.23893487e-01 7.27251589e-01 8.51063311e-01 -4.13139582e-01
-6.69854522e-01 4.33812708e-01 3.76145601e-01 1.13575473e-01
-4.97630835e-01 3.85918617e-01 -7.83776939e-01 -5.09384930e-01
-1.49570704e-01 -8.08591127e-01 4.91888106e-01 2.67807811e-01
1.28655457e+00 -7.82519698e-01 -9.05516148e-02 3.83264273e-01
1.46041358e+00 5.38823187e-01 3.01423967e-01 6.55268371e-01
3.06263328e-01 6.10112607e-01 9.43978369e-01 7.41205275e-01
9.97902453e-02 5.98676503e-01 1.21845675e+00 1.24945357e-01
3.88264835e-01 -2.06925377e-01 7.03103185e-01 -7.26413667e-01
1.49260700e-01 -1.33910045e-01 -1.00103867e+00 5.36845446e-01
-2.24803829e+00 -1.16423047e+00 3.90180409e-01 2.13315678e+00
1.10294724e+00 5.33413291e-01 5.62763095e-01 -2.27102682e-01
2.23157465e-01 1.76053680e-02 -9.99618232e-01 -9.99679267e-01
3.29515636e-01 3.25381666e-01 5.74285269e-01 1.04476941e+00
-1.33945858e+00 1.13445592e+00 7.12439394e+00 5.04331350e-01
-8.99870336e-01 -2.42692694e-01 3.07905585e-01 -7.72543907e-01
-3.11288536e-01 2.10421205e-01 -7.87241995e-01 1.60907596e-01
8.06318521e-01 -3.90042990e-01 8.31376612e-01 1.34219897e+00
5.82445860e-02 -3.11924934e-01 -1.47437155e+00 1.98556736e-01
-2.86903381e-01 -1.25745213e+00 -4.10192221e-01 -7.35866427e-02
6.84687436e-01 -1.22046858e-01 4.56763089e-01 6.58367872e-01
1.30882061e+00 -1.29983699e+00 1.48259497e+00 2.37719640e-01
6.95731223e-01 -1.19184065e+00 3.22049931e-02 3.83958817e-01
-8.13337445e-01 -4.80490446e-01 1.29049107e-01 -1.76913485e-01
-1.23903230e-01 -2.38509044e-01 -8.35841537e-01 -7.62484744e-02
9.27946687e-01 3.83373499e-01 -4.02197391e-01 9.31411207e-01
-6.28166735e-01 2.68636167e-01 -4.28206563e-01 -9.43769217e-02
7.11935520e-01 3.12284768e-01 7.71779835e-01 1.03371775e+00
-7.31570050e-02 -9.28902552e-02 6.05066478e-01 1.07843351e+00
4.89608109e-01 -5.41295111e-01 -1.19017816e+00 6.85503185e-02
2.80504405e-01 1.12866366e+00 -5.56693673e-01 -2.52297282e-01
1.10115558e-01 1.89328730e-01 4.83291388e-01 3.98184776e-01
-1.32643628e+00 -3.82180847e-02 1.08741868e+00 -2.42278859e-01
1.85394526e-01 -3.24798375e-01 -4.16257590e-01 -6.85252547e-01
-7.87815899e-02 -1.20446014e+00 6.32994354e-01 -5.21129370e-01
-8.57496262e-01 4.36481982e-01 4.60421532e-01 -1.11338437e+00
-4.15866226e-01 -5.91531992e-01 -5.35549283e-01 2.99210340e-01
-1.27475095e+00 -8.42365205e-01 3.81698608e-02 6.80843234e-01
3.59404236e-01 -5.32533266e-02 6.78533852e-01 -3.19458425e-01
-6.49919927e-01 5.05289972e-01 -4.34182346e-01 -5.08534670e-01
4.85195428e-01 -1.54480588e+00 1.36338413e-01 1.31040967e+00
-2.72079974e-01 4.92069185e-01 1.02172196e+00 -7.74114311e-01
-1.41583729e+00 -1.11353445e+00 -2.68756300e-01 -6.68705761e-01
7.50112772e-01 -3.69257033e-01 -5.23856878e-01 9.59837556e-01
5.27827621e-01 1.23852000e-01 1.55067965e-01 -2.02875540e-01
-3.79256487e-01 1.09398492e-01 -1.30745804e+00 1.06291175e+00
1.12123322e+00 -3.01147252e-01 -4.85802174e-01 1.80265680e-01
8.46790254e-01 -8.26454043e-01 -4.52957332e-01 4.36493009e-01
4.76100981e-01 -7.55701125e-01 7.62998819e-01 -1.30632043e+00
4.64972556e-01 -5.41203797e-01 -9.79359224e-02 -1.16963172e+00
-1.79413706e-01 -1.12001145e+00 -6.42184317e-01 5.56335032e-01
8.18568841e-02 -2.30828702e-01 7.02299893e-01 9.43504632e-01
-3.52707118e-01 -6.04831040e-01 -1.11269832e+00 -1.19685793e+00
3.39431584e-01 -3.41184974e-01 4.63907689e-01 6.39419436e-01
3.09770763e-01 -2.16948330e-01 -5.63248694e-01 2.38639295e-01
8.39557052e-01 2.95022386e-03 6.34598136e-01 -6.33810282e-01
-4.96480256e-01 -6.31886601e-01 2.76952144e-02 -3.46010327e-02
6.48960412e-01 -6.62617922e-01 6.12022519e-01 -1.05267859e+00
1.29251005e-02 -3.03880394e-01 -2.38502175e-01 1.20890522e+00
2.48782739e-01 -2.58768857e-01 4.69262362e-01 -3.34904253e-01
-1.04187930e+00 5.32530129e-01 1.18713117e+00 -3.13140482e-01
-2.71427810e-01 -3.42149436e-01 -6.98293269e-01 7.56054282e-01
9.09508288e-01 -4.34384018e-01 -7.63766468e-01 -2.59180337e-01
4.37877387e-01 1.41984761e-01 7.39168346e-01 -1.00629282e+00
7.17590451e-02 -1.20635462e+00 5.29146902e-02 -1.34094909e-01
1.55663148e-01 -1.09390831e+00 8.13610330e-02 9.61751401e-01
-9.21505213e-01 2.22612739e-01 7.58994520e-01 5.04230559e-01
3.11032057e-01 -2.30039079e-02 1.06028187e+00 -5.66233834e-03
-1.07195520e+00 1.00484945e-01 -6.67011917e-01 3.90572876e-01
2.03867173e+00 1.85032338e-01 -6.87586546e-01 -2.31464863e-01
-7.44418979e-01 7.83663213e-01 5.71816027e-01 6.06611311e-01
5.10725796e-01 -1.04626727e+00 -3.22602600e-01 5.06232269e-02
5.55028617e-02 -2.43594795e-01 -1.36603236e-01 4.16213155e-01
-3.76041025e-01 2.09101364e-01 -8.91610980e-01 -3.33911896e-01
-1.23146796e+00 9.12054539e-01 7.02902198e-01 -2.27916479e-01
-5.93641698e-01 6.46000445e-01 1.61508754e-01 -1.64377078e-01
7.79761970e-01 -4.03694242e-01 3.90240587e-02 -5.04574180e-01
3.67339015e-01 3.17779303e-01 -2.31202826e-01 -2.37170197e-02
-5.81215918e-01 9.76624414e-02 -1.62880421e-02 -6.89925373e-01
1.29344618e+00 4.55073416e-01 2.49887392e-01 1.78557187e-02
4.41164792e-01 -2.16965362e-01 -2.02558684e+00 4.01816577e-01
2.69164871e-02 -4.12818193e-01 -7.46623054e-02 -1.15149450e+00
-6.77678466e-01 6.97172165e-01 5.71178317e-01 2.50185728e-01
9.24219191e-01 -2.37258792e-01 1.44444421e-01 8.15534830e-01
5.81333935e-01 -1.47846937e+00 2.38360539e-01 6.85648441e-01
1.09380269e+00 -8.96505296e-01 -6.68152869e-02 2.43466645e-01
-9.76846874e-01 8.47190738e-01 1.44232333e+00 -3.58080357e-01
9.84476060e-02 1.06739676e+00 -2.75784522e-01 -1.93940774e-02
-1.15702164e+00 -2.60829538e-01 -1.69203043e-01 6.74611211e-01
-4.22756135e-01 2.01116875e-01 -2.79248059e-02 8.55084062e-02
-4.93603535e-02 -9.26035121e-02 6.81968212e-01 1.43364632e+00
-6.01577818e-01 -1.02679467e+00 -3.38117748e-01 9.65602249e-02
-2.59530663e-01 2.77916044e-01 -5.43401480e-01 9.88701999e-01
5.59373908e-02 7.35952795e-01 -2.56082624e-01 -3.07870477e-01
1.71415627e-01 -2.68936396e-01 7.73792803e-01 -5.83182514e-01
-8.28620076e-01 -1.80264980e-01 3.55423242e-01 -1.19606781e+00
3.19757462e-02 -4.96044219e-01 -1.63286471e+00 -4.36294265e-02
2.49783829e-01 -1.28400996e-01 1.94789559e-01 7.78589845e-01
1.73165455e-01 6.35304034e-01 4.88973737e-01 -6.32469058e-01
-1.29636359e+00 5.64639531e-02 -2.37009913e-01 1.09283105e-01
9.26004589e-01 -9.40964520e-01 -1.17762603e-01 -1.67010054e-01] | [4.466373443603516, 2.09000301361084] |
8870134a-d112-41f0-a9ba-d843631c3c5c | 6d-vnet-end-to-end-6dof-vehicle-pose | null | null | http://openaccess.thecvf.com/content_CVPRW_2019/html/WAD/Wu_6D-VNet_End-To-End_6-DoF_Vehicle_Pose_Estimation_From_Monocular_RGB_Images_CVPRW_2019_paper.html | http://openaccess.thecvf.com/content_CVPRW_2019/papers/WAD/Wu_6D-VNet_End-To-End_6-DoF_Vehicle_Pose_Estimation_From_Monocular_RGB_Images_CVPRW_2019_paper.pdf | 6D-VNet: End-to-end 6DoF Vehicle Pose Estimation from Monocular RGB Images | We present a conceptually simple framework for 6DoF object pose estimation, especially for autonomous driving scenario. Our approach efficiently detects traffic partic- ipants in a monocular RGB image while simultaneously regressing their 3D translation and rotation vectors. The method, called 6D-VNet, extends Mask R-CNN by adding customised heads for predicting vehicle s finer class, ro- tation and translation. The proposed 6D-VNet is trained end-to-end compared to previous methods. Furthermore, we show that the inclusion of translational regression in the joint losses is crucial for the 6DoF pose estimation task, where object translation distance along longitudinal axis varies significantly, e.g., in autonomous driving sce- narios. Additionally, we incorporate the mutual informa- tion between traffic participants via a modified non-local block. As opposed to the original non-local block imple- mentation, the proposed weighting modification takes the spatial neighbouring information into consideration whilst counteracting the effect of extreme gradient values. Our 6D-VNet reaches the 1 st place in ApolloScape challenge 3D Car Instance task1 [21]. Code has been made available at: https://github.com/stevenwudi/6DVNET. | ['Wenbin Zou and Xia Li', 'Di Wu', 'Zhaoyong Zhuang', 'Canqun Xiang'] | 2019-06-15 | null | null | null | the-ieee-conference-on-computer-vision-and-1 | ['vehicle-pose-estimation'] | ['computer-vision'] | [-3.52922648e-01 1.30257010e-01 -2.18887553e-01 -5.34450948e-01
-5.29008746e-01 -3.37057799e-01 7.19322920e-01 -4.39014912e-01
-6.14774525e-01 3.65049452e-01 1.05612189e-01 -3.73847902e-01
1.16029061e-01 -5.31764925e-01 -1.18333721e+00 -7.53762662e-01
1.97447807e-01 5.80192804e-01 3.06155741e-01 -2.39898443e-01
1.11318238e-01 9.39576030e-01 -1.61341858e+00 -1.76064894e-01
5.56556821e-01 1.06699777e+00 1.56109378e-01 7.35471070e-01
3.82192522e-01 6.83386743e-01 -1.84010878e-01 -4.48108554e-01
6.72608018e-01 2.41481751e-01 -2.09695339e-01 8.12615752e-02
1.12073290e+00 -4.43945378e-01 -8.11592579e-01 7.25541472e-01
5.25207460e-01 2.43074268e-01 6.98261321e-01 -1.52793741e+00
-1.75063699e-01 1.08070843e-01 -6.58506870e-01 3.79851371e-01
-1.09499335e-01 5.95930815e-01 9.02024984e-01 -9.88699555e-01
6.90883160e-01 1.54162621e+00 6.51650906e-01 4.96431798e-01
-9.71543372e-01 -8.97903383e-01 3.40625286e-01 5.68906128e-01
-1.37400460e+00 -4.63305861e-01 9.79912877e-01 -5.05023420e-01
1.07607353e+00 1.51813477e-01 4.30140734e-01 1.21676171e+00
2.51598179e-01 1.05837166e+00 7.99995542e-01 4.98357415e-02
-2.53778070e-01 1.56683344e-02 8.66906345e-02 6.05118155e-01
1.53667808e-01 4.31591570e-01 -4.14662570e-01 3.35397393e-01
4.60986912e-01 3.46333496e-02 2.47383907e-01 -8.97031128e-01
-1.12775385e+00 5.78344941e-01 7.40246117e-01 -1.68114558e-01
-1.84348613e-01 5.89454114e-01 2.91878581e-01 1.54439390e-01
4.88523215e-01 -1.51982214e-02 -4.40196186e-01 -1.35005921e-01
-5.68749309e-01 6.08825803e-01 1.25582874e-01 1.21290982e+00
7.42901146e-01 5.93232624e-02 -2.25235462e-01 5.96097291e-01
4.37294215e-01 8.05250823e-01 4.64338139e-02 -9.69914138e-01
9.20856059e-01 3.14953566e-01 1.88771561e-01 -8.50761175e-01
-6.28924549e-01 -5.75846851e-01 -4.27922100e-01 6.36022031e-01
4.56551552e-01 -2.85357181e-02 -1.02779746e+00 1.60583353e+00
5.93825936e-01 4.02552456e-01 -2.36351639e-01 1.17241967e+00
9.41814005e-01 2.72047490e-01 -7.02616721e-02 3.71388197e-01
1.24429440e+00 -1.02605331e+00 -3.67106557e-01 -4.59557682e-01
7.31592715e-01 -6.57721460e-01 6.59428000e-01 1.51225567e-01
-8.80821347e-01 -7.72431493e-01 -9.18249369e-01 -4.62772727e-01
-4.42435473e-01 3.02092850e-01 3.60011339e-01 5.28179288e-01
-8.50381613e-01 1.10561341e-01 -8.28671873e-01 -1.85127288e-01
7.35547066e-01 4.18052912e-01 -4.94697392e-01 -9.03494135e-02
-9.16162550e-01 1.30529416e+00 -1.53220873e-02 3.59025002e-01
-8.33989263e-01 -6.89877748e-01 -9.31106627e-01 -3.97680312e-01
5.02369821e-01 -5.72616279e-01 1.19422424e+00 -3.67491186e-01
-1.32848763e+00 9.31622326e-01 -4.90142643e-01 -7.53574550e-01
1.06641424e+00 -3.51099730e-01 -2.73872375e-01 -1.74331516e-01
4.09550816e-02 1.16502488e+00 9.63350832e-01 -1.17782593e+00
-8.20531845e-01 -5.28571725e-01 1.09560780e-01 4.49313223e-01
3.35090935e-01 -1.87467217e-01 -4.30822104e-01 -3.94763172e-01
4.84370962e-02 -1.19512284e+00 -1.86150163e-01 3.30907524e-01
-3.70115101e-01 -3.92080218e-01 1.21026993e+00 -3.84900421e-01
5.25358260e-01 -2.33135223e+00 1.20147839e-01 1.07283764e-01
5.55107355e-01 2.38795727e-01 -1.42571166e-01 -3.93652022e-02
-9.18376148e-02 -3.70762765e-01 2.10694391e-02 -6.14250064e-01
2.99499273e-01 9.28585604e-02 -4.95955832e-02 9.11032319e-01
4.88805115e-01 1.02912688e+00 -7.25033462e-01 -2.11472914e-01
6.29169047e-01 7.92943597e-01 -4.12699163e-01 -1.13175258e-01
-3.05175707e-02 4.63149250e-01 -4.25986528e-01 6.11508667e-01
1.05475461e+00 3.35741580e-01 -6.90809250e-01 -2.61508763e-01
-3.48363161e-01 3.72418165e-01 -1.14894080e+00 1.27122986e+00
-5.68577409e-01 9.92264509e-01 1.31133780e-01 -8.40641081e-01
9.05277908e-01 -4.65518013e-02 4.06573296e-01 -1.01553035e+00
3.92140031e-01 1.48325607e-01 1.66357353e-01 -5.04475772e-01
5.45830846e-01 1.34053126e-01 7.44269565e-02 -1.72193140e-01
-1.50732428e-01 -9.56411511e-02 1.17341556e-01 5.83116838e-04
7.81849444e-01 3.33635062e-01 -1.78983789e-02 -1.67966075e-02
5.69471180e-01 -1.68952476e-02 4.99295384e-01 6.79676652e-01
-6.63646281e-01 5.63604176e-01 4.47580844e-01 -4.64536130e-01
-1.05790651e+00 -1.01944661e+00 -3.59343648e-01 8.72604370e-01
4.26718622e-01 -6.67853653e-02 -3.84098470e-01 -7.52789319e-01
5.77574909e-01 7.05601990e-01 -7.60209203e-01 -1.45587280e-01
-9.98651028e-01 -3.14707249e-01 6.20581508e-01 6.82397068e-01
3.65499079e-01 -7.09739685e-01 -7.39401460e-01 6.19391091e-02
-2.75530349e-02 -1.35702455e+00 -4.83701140e-01 4.03726757e-01
-6.06799126e-01 -7.36711800e-01 -4.56811816e-01 -4.41935152e-01
4.05901492e-01 6.40865505e-01 7.64201522e-01 -3.20323050e-01
-3.09445411e-01 1.97069690e-01 -8.84732306e-02 -6.56551421e-01
-3.00184935e-02 2.07142606e-01 1.09099634e-01 5.62357977e-02
7.90857077e-01 -2.47792020e-01 -8.69586766e-01 6.17643178e-01
-1.33787379e-01 -3.94002758e-02 6.13752663e-01 4.54254746e-01
5.93961179e-01 -3.35738868e-01 2.98697770e-01 -4.53629941e-01
-8.97793844e-02 -4.97009873e-01 -7.02443421e-01 -4.90025192e-01
-2.34898865e-01 -1.58865437e-01 2.23621070e-01 -3.34576249e-01
-8.22135508e-01 2.66130596e-01 -4.40437824e-01 -7.89479077e-01
-3.64109993e-01 -2.91911811e-01 -2.47133106e-01 -2.33872071e-01
5.94808221e-01 -2.72685587e-01 1.60052359e-01 -2.81713158e-01
5.05924463e-01 5.63930213e-01 4.02456909e-01 -1.23889577e-02
1.14902258e+00 7.30643988e-01 3.06352884e-01 -7.68900573e-01
-5.21611631e-01 -7.32922435e-01 -7.98733175e-01 -4.59475070e-01
9.13639843e-01 -1.23002982e+00 -9.12825942e-01 4.97950971e-01
-1.02057314e+00 -3.30167383e-01 -1.33271426e-01 7.39734352e-01
-6.11127019e-01 6.21640943e-02 -1.40926525e-01 -7.13391244e-01
1.05249614e-01 -1.38410568e+00 1.40342557e+00 -1.09542072e-01
-9.05307382e-03 -6.99430168e-01 -2.88479447e-01 6.98479772e-01
2.72425860e-01 2.82962233e-01 3.45379859e-01 -3.21003973e-01
-9.00326252e-01 -4.03656870e-01 -4.35550690e-01 2.64273435e-01
-2.56986350e-01 -1.81090429e-01 -1.08533585e+00 -1.40391737e-01
-2.42749244e-01 -8.10458064e-02 1.20477688e+00 5.78925729e-01
8.08211803e-01 8.90493095e-02 -3.01081330e-01 6.66168869e-01
9.51097548e-01 -7.14265257e-02 5.35961211e-01 3.89057428e-01
1.29535234e+00 6.53376877e-01 8.75768423e-01 4.37575839e-02
6.88807547e-01 1.08179176e+00 8.76921833e-01 -8.32736716e-02
-3.29410374e-01 -1.34832457e-01 4.10062820e-01 2.49569729e-01
-1.62053391e-01 -1.85113594e-01 -7.88016438e-01 5.53469241e-01
-1.79754174e+00 -8.75031233e-01 -6.47120059e-01 1.93809760e+00
1.79639935e-01 5.31192243e-01 4.35332060e-01 3.44121493e-02
4.03273553e-01 4.33653682e-01 -7.98691809e-01 -3.43898565e-01
-1.90583035e-01 -1.67093188e-01 1.23304451e+00 7.81873405e-01
-1.22341168e+00 1.07394600e+00 4.82376242e+00 7.84322441e-01
-1.24887574e+00 3.95172954e-01 2.98384160e-01 -4.69267637e-01
-1.26497820e-02 -2.27868855e-01 -1.33321881e+00 3.08598727e-01
7.27805316e-01 4.18483794e-01 1.96155176e-01 8.79713416e-01
5.54720104e-01 -1.14421740e-01 -9.62140203e-01 9.58935261e-01
1.63144648e-01 -9.61828887e-01 -3.79884392e-01 2.79956937e-01
3.85047644e-01 7.41441071e-01 2.76209205e-01 3.45200151e-01
8.25143233e-02 -9.65863049e-01 1.22245216e+00 4.00622815e-01
6.35704279e-01 -8.64107013e-01 6.84805155e-01 3.53954077e-01
-1.15866637e+00 -2.25663409e-01 -2.93832511e-01 7.63592720e-02
2.86849737e-01 4.47217166e-01 -1.06137812e+00 2.98788637e-01
6.88062310e-01 8.61713469e-01 -6.94064319e-01 9.35674012e-01
-9.74738449e-02 4.14161980e-01 -4.70940560e-01 3.64496652e-03
5.04833400e-01 -1.17472671e-02 9.49529409e-01 1.25395262e+00
2.35929817e-01 -3.84274572e-01 -1.63579106e-01 6.69717073e-01
-1.79466198e-03 -2.58746266e-01 -7.04194069e-01 5.03682554e-01
1.92895055e-01 1.13164365e+00 -4.18389320e-01 -2.65357625e-02
-4.50590700e-01 5.43854833e-01 2.33055979e-01 4.32114124e-01
-1.14720094e+00 -2.70867348e-01 1.08454311e+00 4.83483136e-01
8.18220377e-01 -3.78861248e-01 -4.11903083e-01 -8.58386099e-01
2.46735767e-01 -5.39308846e-01 -1.50511697e-01 -7.29273975e-01
-9.14889634e-01 3.82632941e-01 2.53320187e-01 -1.44198000e+00
-5.50400279e-02 -8.66365910e-01 -3.11550915e-01 8.19857836e-01
-1.76495993e+00 -1.38639033e+00 -3.26224864e-01 4.71727908e-01
6.67825103e-01 -9.12121758e-02 -1.10424854e-01 5.26677132e-01
-5.13508320e-01 6.99876070e-01 -1.96864992e-01 -1.00444384e-01
7.10956991e-01 -1.14993882e+00 7.74191797e-01 6.07452929e-01
-1.87420845e-01 2.74308234e-01 1.00407195e+00 -5.25196552e-01
-1.39075339e+00 -1.39600599e+00 1.05115747e+00 -1.01129901e+00
5.11809587e-01 -8.06075454e-01 -6.37546301e-01 8.25266242e-01
-1.05921820e-01 3.25811476e-01 -8.73271078e-02 -1.59297094e-01
-4.06726062e-01 -5.09447217e-01 -1.07876170e+00 6.57354295e-01
1.29412913e+00 -4.03037667e-01 -3.15414548e-01 3.56140196e-01
5.38949668e-01 -8.93306553e-01 -4.13860470e-01 4.70272750e-01
6.39177620e-01 -7.06503510e-01 1.20299995e+00 -2.56435245e-01
5.25355823e-02 -7.02591300e-01 -7.81480372e-02 -9.58248913e-01
-3.20745200e-01 -4.26209718e-01 -2.63293445e-01 6.50964260e-01
3.24938834e-01 -6.32030308e-01 1.00342190e+00 1.40134439e-01
-3.87263387e-01 -7.18015611e-01 -1.41561890e+00 -7.89770365e-01
5.44367284e-02 -8.44250679e-01 4.69344795e-01 4.23391789e-01
-6.69386327e-01 2.90251732e-01 -5.98808885e-01 2.92380393e-01
6.93630993e-01 -4.14961874e-01 1.25622976e+00 -1.13935876e+00
2.75892131e-02 -5.69888294e-01 -9.28712428e-01 -1.48730421e+00
2.74753749e-01 -9.10881996e-01 9.18795243e-02 -1.15888011e+00
-3.63023520e-01 -2.74355024e-01 -1.02342367e-01 2.10331038e-01
3.78826335e-02 4.94596392e-01 2.45848045e-01 -6.42905459e-02
-4.61930811e-01 6.23680472e-01 1.47372782e+00 -2.07004651e-01
6.74256939e-04 2.86194593e-01 -3.85726810e-01 5.85047424e-01
8.26704204e-01 -5.57644665e-01 -3.28638524e-01 -3.62417728e-01
2.14401886e-01 -3.38136435e-01 8.53346586e-01 -9.37581480e-01
2.14467615e-01 -1.16890334e-01 3.41440916e-01 -1.26769662e+00
6.46715224e-01 -9.66844738e-01 -2.87919007e-02 3.18922073e-01
-4.93493453e-02 2.93800384e-02 2.02290609e-01 6.27916813e-01
-1.65644344e-02 7.87304640e-02 7.95073926e-01 9.03747752e-02
-7.53759503e-01 4.43312794e-01 -3.89885932e-01 -1.28577709e-01
1.18603981e+00 -6.71572864e-01 -2.21280321e-01 -2.03995749e-01
-4.85140711e-01 4.87445831e-01 2.30087727e-01 8.17409933e-01
4.24308360e-01 -1.33335876e+00 -7.44917631e-01 2.36774459e-01
4.29640114e-01 1.24497853e-01 4.04579401e-01 1.25273716e+00
-5.66942632e-01 6.60569131e-01 -1.68288842e-01 -9.13775265e-01
-1.25130677e+00 3.62815589e-01 5.31063139e-01 1.24271184e-01
-7.86443055e-01 9.55906689e-01 4.15841103e-01 -7.10069537e-01
4.17006850e-01 -6.82902515e-01 -1.76695287e-01 3.00555751e-02
2.13862315e-01 5.24887502e-01 3.52692336e-01 -1.30438912e+00
-5.44721842e-01 8.12436044e-01 -3.10607582e-01 8.10500085e-02
1.20655310e+00 -3.61998528e-01 4.61547792e-01 3.11973333e-01
1.44785345e+00 -6.05210289e-02 -1.76256275e+00 -8.26322362e-02
-4.15150017e-01 -4.96986866e-01 2.87505567e-01 -4.41456616e-01
-1.09031463e+00 9.90205646e-01 8.53164434e-01 -4.67503250e-01
5.02497792e-01 1.38953313e-01 6.90398753e-01 4.45124179e-01
1.40640691e-01 -1.01726592e+00 -1.85970843e-01 7.16355979e-01
8.80913615e-01 -1.54076111e+00 -4.20805551e-02 -3.59311819e-01
-6.62550688e-01 8.01524997e-01 8.35660160e-01 -4.13707942e-01
6.77307248e-01 8.21301639e-02 2.13314056e-01 -2.95289606e-01
-5.95402181e-01 -4.47407007e-01 4.16541159e-01 7.31847346e-01
-1.38533441e-02 1.23576924e-01 -1.29509056e-02 -1.77198723e-02
-3.83304536e-01 -2.48010159e-01 1.83240399e-01 7.73706317e-01
-2.86957592e-01 -6.38093829e-01 -4.11943465e-01 3.07437927e-01
-1.63795948e-01 2.82194108e-01 -3.02380800e-01 9.75881457e-01
6.23559177e-01 7.28369057e-01 4.01775509e-01 -4.12319332e-01
7.80017436e-01 -3.03707480e-01 4.07079220e-01 -2.52581507e-01
-4.50187147e-01 -1.09987101e-02 1.54255480e-01 -7.28851199e-01
-3.27645868e-01 -1.01399851e+00 -1.04940295e+00 -3.83066207e-01
-2.61066586e-01 -4.77805257e-01 9.61686671e-01 9.17237997e-01
3.91627073e-01 4.95889097e-01 6.05048001e-01 -1.28601372e+00
-2.99319178e-01 -8.32392693e-01 -2.19031736e-01 1.29355356e-01
9.59732056e-01 -1.06564140e+00 -5.51620007e-01 -3.30573618e-01] | [8.025219917297363, -2.116116762161255] |
c776702f-9ae5-42bf-89a2-a792153580ca | cross-x-learning-for-fine-grained-visual | 1909.04412 | null | https://arxiv.org/abs/1909.04412v1 | https://arxiv.org/pdf/1909.04412v1.pdf | Cross-X Learning for Fine-Grained Visual Categorization | Recognizing objects from subcategories with very subtle differences remains a challenging task due to the large intra-class and small inter-class variation. Recent work tackles this problem in a weakly-supervised manner: object parts are first detected and the corresponding part-specific features are extracted for fine-grained classification. However, these methods typically treat the part-specific features of each image in isolation while neglecting their relationships between different images. In this paper, we propose Cross-X learning, a simple yet effective approach that exploits the relationships between different images and between different network layers for robust multi-scale feature learning. Our approach involves two novel components: (i) a cross-category cross-semantic regularizer that guides the extracted features to represent semantic parts and, (ii) a cross-layer regularizer that improves the robustness of multi-scale features by matching the prediction distribution across multiple layers. Our approach can be easily trained end-to-end and is scalable to large datasets like NABirds. We empirically analyze the contributions of different components of our approach and demonstrate its robustness, effectiveness and state-of-the-art performance on five benchmark datasets. Code is available at \url{https://github.com/cswluo/CrossX}. | ['Ser-Nam Lim', 'Jian Yang', 'Xianjie Mo', 'Jun Li', 'Wei Luo', 'Larry S. Davis', 'Yuheng Lu', 'Xitong Yang'] | 2019-09-10 | cross-x-learning-for-fine-grained-visual-1 | http://openaccess.thecvf.com/content_ICCV_2019/html/Luo_Cross-X_Learning_for_Fine-Grained_Visual_Categorization_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/Luo_Cross-X_Learning_for_Fine-Grained_Visual_Categorization_ICCV_2019_paper.pdf | iccv-2019-10 | ['fine-grained-visual-categorization'] | ['computer-vision'] | [ 1.51490927e-01 -2.71574736e-01 -3.80558997e-01 -6.19191468e-01
-9.78748620e-01 -6.72367871e-01 5.44501305e-01 1.28834739e-01
-3.28633815e-01 2.47830659e-01 8.06703120e-02 2.47557461e-01
-2.69997329e-01 -5.58698177e-01 -9.43525195e-01 -5.60466528e-01
1.19016871e-01 2.95316905e-01 6.46728754e-01 1.09152608e-01
1.70833588e-01 6.10018909e-01 -1.73606551e+00 6.72009766e-01
6.30235732e-01 1.53734207e+00 2.29336843e-01 2.88785696e-01
-2.47323096e-01 6.58838630e-01 -2.17128396e-01 -2.34871477e-01
3.96778733e-01 -1.07831769e-01 -9.40795958e-01 3.71418327e-01
8.23799133e-01 4.72841822e-02 -2.59717494e-01 1.05036783e+00
3.36023360e-01 -7.95711800e-02 5.94963908e-01 -1.27446818e+00
-5.17563164e-01 4.06396657e-01 -8.55105639e-01 4.75835316e-02
-1.08355872e-01 5.82455583e-02 1.15822351e+00 -9.00664210e-01
3.13584477e-01 1.27345872e+00 8.00204396e-01 4.80571061e-01
-1.41072965e+00 -8.21260095e-01 4.18292403e-01 2.95592546e-01
-1.62344062e+00 -3.49565446e-01 8.94271970e-01 -4.71208513e-01
8.23388815e-01 1.45597577e-01 3.19137096e-01 8.18610907e-01
2.65465472e-02 1.01606667e+00 1.33399916e+00 -1.78337663e-01
2.20018905e-02 1.68860674e-01 2.56654114e-01 7.94870734e-01
2.59582363e-02 -9.78339314e-02 -4.95043069e-01 -1.66729555e-01
5.13324261e-01 4.53154355e-01 -1.43293738e-01 -8.00911844e-01
-1.17025244e+00 7.01298237e-01 9.24363971e-01 4.32304293e-01
-3.64442825e-01 1.18055053e-01 4.66159582e-01 1.81760445e-01
4.96438324e-01 1.29843548e-01 -9.80640829e-01 2.43643671e-01
-8.99629593e-01 8.30169916e-02 4.98533547e-01 8.69672537e-01
9.79949653e-01 -4.66425598e-01 -2.66103059e-01 1.18748367e+00
2.93919861e-01 2.05941185e-01 5.40995240e-01 -7.18983829e-01
4.39181417e-01 8.25694859e-01 -1.92138478e-01 -7.93715060e-01
-4.32553798e-01 -5.40716529e-01 -7.17481673e-01 1.11481898e-01
4.21601325e-01 2.83922255e-01 -9.49986994e-01 1.85924029e+00
5.17960072e-01 2.08185986e-01 -2.51826078e-01 7.41480589e-01
8.97497475e-01 3.81448120e-01 2.26521552e-01 3.04011911e-01
1.54983103e+00 -1.30274630e+00 -1.14468530e-01 -4.45817113e-01
4.16722596e-01 -7.50470221e-01 1.14432168e+00 4.69427276e-03
-8.96623194e-01 -7.14080572e-01 -8.10688317e-01 -1.20698594e-01
-6.36285901e-01 4.69666302e-01 4.85062510e-01 2.10571855e-01
-9.02481496e-01 6.45296872e-01 -7.44184673e-01 -3.52288842e-01
9.65795040e-01 4.24833179e-01 -4.33422804e-01 -3.39956135e-01
-7.48567641e-01 5.67717552e-01 4.85009164e-01 -1.82757508e-02
-6.54797316e-01 -7.72761464e-01 -8.84048641e-01 1.96195170e-01
5.00244915e-01 -5.10586441e-01 1.09011984e+00 -1.32475150e+00
-1.08081365e+00 1.08294702e+00 -1.24925740e-01 -6.58450881e-03
3.57304424e-01 -3.15242410e-01 -1.32380977e-01 2.27200031e-01
4.52958524e-01 8.01689088e-01 8.40745568e-01 -1.41074359e+00
-8.25609744e-01 -6.58549249e-01 -4.52605681e-03 1.03873182e-02
-2.80548692e-01 9.52762514e-02 -8.65702331e-01 -7.88343430e-01
1.47181898e-01 -1.08618855e+00 -1.00957938e-01 3.66294771e-01
-3.62698972e-01 -5.26455879e-01 7.28832543e-01 -3.54592353e-01
8.24994266e-01 -2.39824176e+00 -2.41008047e-02 6.81557059e-02
1.40028968e-01 3.16059798e-01 -3.83706361e-01 2.31760517e-01
-6.32867217e-02 -2.11726263e-01 -2.56030560e-01 -3.11769187e-01
1.58069506e-01 -5.29237539e-02 -1.39073655e-01 5.10231912e-01
5.51429570e-01 1.06465364e+00 -6.99653506e-01 -4.17891383e-01
2.28884935e-01 5.44373274e-01 -2.93634981e-01 2.63749391e-01
-1.11063123e-02 1.75055325e-01 -4.77354586e-01 8.16594958e-01
7.60687113e-01 -5.53725064e-01 -7.98877422e-03 -5.85210264e-01
1.11814700e-01 2.04078108e-01 -1.36875916e+00 1.51228189e+00
-5.25106609e-01 2.55755663e-01 1.44612834e-01 -1.09661150e+00
6.63216412e-01 1.55130483e-03 3.76037598e-01 -6.54748082e-01
1.23906121e-01 3.46065581e-01 -1.87115952e-01 -2.84771621e-01
1.07328348e-01 -1.41366214e-01 -2.03742176e-01 3.38232279e-01
3.78155202e-01 1.48861349e-01 2.80899644e-01 8.53833482e-02
9.52432334e-01 1.77858204e-01 4.84105676e-01 -3.57661098e-01
5.28639555e-01 -3.06884646e-01 5.96069098e-01 7.21224785e-01
-2.53384590e-01 7.33772278e-01 2.47367114e-01 -3.96868050e-01
-6.84905052e-01 -1.06454360e+00 -3.02763462e-01 1.35331726e+00
3.88574690e-01 -4.15928513e-01 -5.46030045e-01 -1.20658958e+00
4.15513217e-01 2.24242866e-01 -8.81197333e-01 -3.14297497e-01
-4.07161683e-01 -5.36587477e-01 2.65433520e-01 9.54120398e-01
4.84785169e-01 -1.13054049e+00 -2.67691046e-01 4.73801903e-02
-7.16838166e-02 -1.22326207e+00 -6.33695066e-01 3.80745441e-01
-8.42038155e-01 -1.14484680e+00 -5.37312031e-01 -8.99658501e-01
7.01146007e-01 6.99257374e-01 1.26537108e+00 1.14822567e-01
-4.32375222e-01 3.45151216e-01 -2.89933920e-01 -2.69332141e-01
5.57608865e-02 1.51533410e-02 -2.62772322e-01 3.02907795e-01
4.86035705e-01 -4.44568515e-01 -6.60591900e-01 5.70777476e-01
-8.80601645e-01 -1.67650923e-01 9.85418797e-01 7.31868744e-01
9.19291675e-01 1.73094254e-02 4.24316555e-01 -8.92346859e-01
2.62561161e-02 -4.73815233e-01 -4.02506292e-01 3.59688699e-01
-2.87757725e-01 -1.90695282e-02 6.97424710e-01 -4.00786370e-01
-7.68232167e-01 3.14369142e-01 -5.22765284e-03 -3.90035659e-01
-4.77751374e-01 6.66569993e-02 -4.47381675e-01 -1.86852574e-01
2.88166195e-01 2.61954814e-01 -2.14261040e-01 -7.36265898e-01
3.91906023e-01 7.28873789e-01 3.08070332e-01 -5.87251604e-01
8.45593870e-01 6.40439630e-01 -2.65407890e-01 -6.68774545e-01
-1.46784127e+00 -9.48233128e-01 -9.30348992e-01 1.02756443e-02
7.10684597e-01 -1.29364300e+00 -3.32085580e-01 6.29792690e-01
-7.97928274e-01 -4.07337248e-01 -3.89433384e-01 2.96183497e-01
-4.79052901e-01 2.06489831e-01 -5.57825267e-01 -1.90279871e-01
-3.09336364e-01 -1.06450510e+00 1.55541575e+00 2.96408802e-01
7.69951791e-02 -7.79962838e-01 -1.69650599e-01 5.12280881e-01
2.45091036e-01 -2.14184029e-03 6.22896016e-01 -5.86230278e-01
-5.75746834e-01 -2.43308961e-01 -7.78762162e-01 5.19776046e-01
2.99094528e-01 -1.95342675e-01 -9.55528915e-01 -4.82995361e-01
-3.03872615e-01 -6.84113204e-01 1.22671986e+00 1.58508629e-01
1.54003274e+00 -2.52407372e-01 -5.09812295e-01 6.75554752e-01
1.59436226e+00 -4.60078984e-01 3.12219858e-01 3.52362037e-01
9.99070942e-01 6.20402217e-01 6.60928905e-01 2.78805196e-01
5.32813072e-01 8.00453603e-01 2.66229689e-01 -4.17449288e-02
-4.08051968e-01 1.04189692e-02 2.31811777e-01 5.42792022e-01
1.14291258e-01 1.21915989e-01 -6.07851267e-01 7.44162023e-01
-1.86178076e+00 -8.82653415e-01 1.00042887e-01 1.99504161e+00
8.71342301e-01 6.21193200e-02 2.75758475e-01 -8.07493106e-02
7.55749226e-01 1.65412843e-01 -5.72565496e-01 4.00968157e-02
-5.35725988e-02 1.50358230e-01 3.72356325e-01 1.56871408e-01
-1.52283061e+00 9.43451285e-01 4.89604855e+00 1.09125340e+00
-1.17605293e+00 3.06931198e-01 6.29193902e-01 -1.42880961e-01
1.15719162e-01 -2.15543285e-01 -9.62759197e-01 4.89750236e-01
4.63881344e-01 4.25947726e-01 1.21330656e-01 1.05843985e+00
-2.08235472e-01 3.38501744e-02 -1.06605780e+00 9.25897956e-01
2.03933448e-01 -1.21384943e+00 -9.65128615e-02 -1.03869110e-01
7.90063500e-01 3.88081640e-01 -1.43422887e-01 3.42552930e-01
7.76895136e-02 -7.41862893e-01 9.64176595e-01 3.60074788e-01
6.47019207e-01 -6.75093591e-01 6.57355428e-01 1.80785418e-01
-1.64537191e+00 -2.95625985e-01 -4.29402739e-01 2.30395570e-01
-2.33720437e-01 6.52465820e-01 -3.73533428e-01 5.31785071e-01
1.08685565e+00 8.51262450e-01 -9.58588958e-01 1.10555053e+00
-2.19218448e-01 5.40838003e-01 -2.05368221e-01 2.89574027e-01
3.47532570e-01 1.26933470e-01 1.49878830e-01 1.53013897e+00
8.82690996e-02 -2.35002115e-01 4.95964736e-01 6.68093860e-01
-1.90681458e-01 1.20181870e-02 -1.97999656e-01 2.37963066e-01
2.64255524e-01 1.55538654e+00 -1.01291835e+00 -3.22379112e-01
-8.36961627e-01 1.08388329e+00 7.79613018e-01 8.07386488e-02
-7.98796058e-01 -3.77279907e-01 7.55571902e-01 8.84309039e-02
1.06156731e+00 3.51477211e-05 -2.06049874e-01 -1.18987763e+00
3.45900953e-01 -8.21124434e-01 6.59075379e-01 -3.68166089e-01
-1.78570640e+00 6.03416681e-01 -2.08247021e-01 -1.25939083e+00
2.49617398e-01 -8.08508158e-01 -4.27164972e-01 6.43937290e-01
-1.87482846e+00 -1.63140655e+00 -3.98988605e-01 7.43067145e-01
5.61255813e-01 3.54664237e-03 5.80214918e-01 4.01272863e-01
-5.51227391e-01 7.41011798e-01 2.99236298e-01 2.63432294e-01
7.59213448e-01 -1.30850935e+00 1.99522674e-01 7.00380445e-01
2.68144041e-01 5.08512199e-01 1.90424040e-01 -2.86157638e-01
-1.11810923e+00 -1.55338657e+00 8.56326163e-01 -4.35957611e-01
6.75522029e-01 -5.75415969e-01 -1.13845885e+00 5.73731303e-01
-9.56747457e-02 7.73519099e-01 7.45139241e-01 2.08579212e-01
-9.16635692e-01 -5.05447626e-01 -9.78772879e-01 1.30230978e-01
1.01674139e+00 -6.46079719e-01 -4.66196150e-01 3.31932276e-01
4.68065441e-01 -7.69742057e-02 -7.67104685e-01 5.28940618e-01
6.01239443e-01 -1.01093578e+00 1.00825810e+00 -5.56485295e-01
2.56803513e-01 -4.23377067e-01 -3.32356423e-01 -1.09068537e+00
-5.83013535e-01 -1.73224807e-01 -1.84227735e-01 1.39743078e+00
3.10100913e-01 -5.12284696e-01 5.96168160e-01 3.08983713e-01
-2.91753784e-02 -1.01519358e+00 -6.55041456e-01 -1.00493598e+00
1.21791032e-03 -3.91907305e-01 4.32890177e-01 8.30732882e-01
-4.24092948e-01 3.77479464e-01 -1.83667466e-01 2.14264572e-01
6.79139197e-01 8.35360408e-01 6.30236924e-01 -1.25612938e+00
-4.06215370e-01 -5.54196954e-01 -6.15436733e-01 -9.25511122e-01
2.57107794e-01 -1.02787292e+00 2.52216786e-01 -1.33465576e+00
7.55898416e-01 -5.56446373e-01 -6.26881301e-01 7.41810083e-01
-4.76993173e-01 7.88889945e-01 4.38587844e-01 2.34369189e-01
-1.10060811e+00 5.84382653e-01 1.05289710e+00 -1.19403504e-01
3.98458093e-02 1.00746550e-01 -9.12541747e-01 8.65207791e-01
7.34489024e-01 -6.12199366e-01 -1.54429719e-01 -3.22831631e-01
-2.61016428e-01 -6.23896122e-01 6.78953052e-01 -1.12831378e+00
-6.75169304e-02 3.81414331e-02 6.15135074e-01 -5.07291615e-01
1.51470974e-01 -8.33458543e-01 -2.49951437e-01 3.09256762e-01
-3.23297739e-01 -2.50831693e-01 3.28429878e-01 6.31765783e-01
-2.99450099e-01 -3.70526314e-02 1.16512215e+00 -1.66655391e-01
-8.32949877e-01 5.53487599e-01 2.17528224e-01 2.32659996e-01
1.06323445e+00 -5.58069274e-02 -1.34433642e-01 1.33216843e-01
-4.66229498e-01 9.09012929e-02 4.94844735e-01 7.34311938e-01
3.63082767e-01 -1.43980050e+00 -6.20240450e-01 7.44266137e-02
5.39981306e-01 -3.20938975e-02 4.61735278e-01 8.50651503e-01
3.26257572e-02 3.44814360e-01 -2.02870801e-01 -8.01361084e-01
-1.46832550e+00 7.05472767e-01 2.60467827e-01 -2.82689393e-01
-4.76367593e-01 1.07899404e+00 7.34556735e-01 -7.61429965e-01
2.68436700e-01 -1.77143991e-01 -4.54798639e-02 1.59503594e-01
6.63849115e-01 4.31386493e-02 1.27281815e-01 -1.10417557e+00
-7.00800180e-01 1.00680447e+00 -4.35394883e-01 5.21662712e-01
1.54261613e+00 -1.85821056e-01 -1.53772697e-01 4.57221478e-01
1.67012799e+00 -9.30971801e-02 -1.62524104e+00 -8.08441997e-01
-1.24355908e-02 -6.17555141e-01 1.83045436e-02 -9.29306269e-01
-1.30260122e+00 7.37545848e-01 6.87580109e-01 5.68567999e-02
1.26642752e+00 7.24197686e-01 7.81078398e-01 1.81895196e-01
1.53742403e-01 -1.01484513e+00 3.07163447e-01 4.19617116e-01
8.69925082e-01 -1.52246749e+00 -6.68421807e-03 -5.04752219e-01
-5.72673678e-01 9.72766459e-01 7.26146519e-01 -2.48133168e-01
7.89209366e-01 9.96098891e-02 1.03999361e-01 -2.90164709e-01
-6.25963271e-01 -5.72344244e-01 7.03361988e-01 4.39534873e-01
4.49404240e-01 3.59148532e-02 1.66590530e-02 6.86655521e-01
2.45621532e-01 -7.65191168e-02 -1.17385454e-01 9.97659028e-01
-3.92380536e-01 -1.06210959e+00 -2.20479533e-01 5.25768757e-01
-5.73575199e-01 -1.42236101e-02 -4.75604892e-01 7.69219160e-01
4.46040779e-01 6.43879235e-01 9.91046652e-02 -2.00770959e-01
4.25768703e-01 -5.02583683e-02 5.89635909e-01 -6.74576342e-01
-5.74834347e-01 7.46643320e-02 -2.59860694e-01 -1.04121351e+00
-6.30296826e-01 -9.37588930e-01 -1.07460642e+00 1.21945925e-01
-2.25206152e-01 -1.51005760e-01 4.84572351e-01 1.01329470e+00
7.43890762e-01 4.71950203e-01 8.07901978e-01 -1.00843871e+00
-6.32185757e-01 -8.51155102e-01 -6.16544485e-01 7.21039295e-01
4.29423243e-01 -7.81856656e-01 -2.79343367e-01 -8.46862718e-02] | [9.598058700561523, 1.9769309759140015] |
fd66338a-ac7e-43d9-b724-e6a9635f327d | multi-level-temporal-channel-speaker | 2305.07204 | null | https://arxiv.org/abs/2305.07204v1 | https://arxiv.org/pdf/2305.07204v1.pdf | Multi-level Temporal-channel Speaker Retrieval for Robust Zero-shot Voice Conversion | Zero-shot voice conversion (VC) converts source speech into the voice of any desired speaker using only one utterance of the speaker without requiring additional model updates. Typical methods use a speaker representation from a pre-trained speaker verification (SV) model or learn speaker representation during VC training to achieve zero-shot VC. However, existing speaker modeling methods overlook the variation of speaker information richness in temporal and frequency channel dimensions of speech. This insufficient speaker modeling hampers the ability of the VC model to accurately represent unseen speakers who are not in the training dataset. In this study, we present a robust zero-shot VC model with multi-level temporal-channel retrieval, referred to as MTCR-VC. Specifically, to flexibly adapt to the dynamic-variant speaker characteristic in the temporal and channel axis of the speech, we propose a novel fine-grained speaker modeling method, called temporal-channel retrieval (TCR), to find out when and where speaker information appears in speech. It retrieves variable-length speaker representation from both temporal and channel dimensions under the guidance of a pre-trained SV model. Besides, inspired by the hierarchical process of human speech production, the MTCR speaker module stacks several TCR blocks to extract speaker representations from multi-granularity levels. Furthermore, to achieve better speech disentanglement and reconstruction, we introduce a cycle-based training strategy to simulate zero-shot inference recurrently. We adopt perpetual constraints on three aspects, including content, style, and speaker, to drive this process. Experiments demonstrate that MTCR-VC is superior to the previous zero-shot VC methods in modeling speaker timbre while maintaining good speech naturalness. | ['Yuping Wang', 'Qiao Tian', 'Yuanzhe Chen', 'Lei Xie', 'Qiuqiang Kong', 'Liumeng Xue', 'Zhichao Wang'] | 2023-05-12 | null | null | null | null | ['voice-conversion', 'voice-conversion', 'speaker-verification'] | ['audio', 'speech', 'speech'] | [ 1.58770680e-01 -2.41318911e-01 -2.66346604e-01 -2.31336817e-01
-1.02527022e+00 -5.26966035e-01 6.01569831e-01 -3.19229901e-01
1.33198738e-01 2.36789137e-01 4.93020236e-01 -2.35517681e-01
6.75198361e-02 -5.07557213e-01 -3.41287673e-01 -6.87149584e-01
2.11188272e-01 1.14377499e-01 1.26714930e-02 -3.23844910e-01
-4.65444773e-02 3.86948526e-01 -1.88231075e+00 2.89151996e-01
6.29383445e-01 7.23576367e-01 4.70804721e-01 8.97785664e-01
-5.41734874e-01 4.62362528e-01 -8.10942233e-01 9.08252373e-02
-1.48691162e-01 -8.78774107e-01 -4.87456858e-01 2.07840011e-01
-1.68071181e-01 -1.42471641e-01 -1.03003435e-01 7.77164996e-01
7.54658461e-01 4.02237743e-01 6.96884453e-01 -8.52392495e-01
-5.52089632e-01 8.59069645e-01 -2.35825315e-01 3.04767281e-01
5.88450491e-01 5.05302623e-02 8.76846850e-01 -9.23254848e-01
4.47979391e-01 1.52504551e+00 3.43922526e-01 9.01722848e-01
-1.24482667e+00 -8.73873770e-01 4.24237937e-01 1.06379829e-01
-1.52435064e+00 -1.09337366e+00 1.14136612e+00 -4.51080412e-01
8.35614324e-01 4.99858409e-01 6.20598376e-01 1.41433644e+00
-1.93232089e-01 5.53750694e-01 6.98818922e-01 -6.59855902e-01
3.71815711e-01 2.88306415e-01 2.98828661e-01 3.22855681e-01
-5.80639541e-01 4.78038609e-01 -9.58977818e-01 -3.42364311e-01
5.30458331e-01 -1.88990325e-01 -4.33183223e-01 -8.16670898e-03
-9.49883580e-01 7.31568217e-01 -1.22711286e-01 6.37009561e-01
-2.88864255e-01 -1.22366704e-01 3.90955925e-01 4.68451351e-01
3.66716892e-01 5.99671081e-02 -2.31964707e-01 -3.05011332e-01
-1.09555221e+00 -5.09762317e-02 8.24260175e-01 1.00401747e+00
6.31493270e-01 7.25658774e-01 -4.51430649e-01 1.09147954e+00
5.09263396e-01 5.21914601e-01 1.08634925e+00 -5.63115001e-01
1.11333713e-01 3.80153321e-02 -1.83758467e-01 -4.46344823e-01
9.27906334e-02 -6.03695273e-01 -6.58887208e-01 -2.86110163e-01
-2.02502295e-01 4.20426801e-02 -1.00489521e+00 1.94354451e+00
3.95421535e-01 4.70760435e-01 3.96781325e-01 8.04513454e-01
9.15795803e-01 9.46157575e-01 -1.19796954e-01 -7.79857159e-01
1.41380334e+00 -7.34458804e-01 -1.04822183e+00 -1.27395228e-01
1.60323918e-01 -6.54884934e-01 1.27741146e+00 2.30109826e-01
-7.89207101e-01 -7.77944565e-01 -1.14112830e+00 3.24883968e-01
-2.17185721e-01 -1.27838224e-01 1.42431468e-01 1.05465913e+00
-7.66287029e-01 1.50716469e-01 -6.11786008e-01 -1.86237879e-02
-2.40354091e-01 2.92414632e-02 1.39880255e-01 3.05903167e-01
-1.53183305e+00 4.14704829e-01 1.99493486e-02 -3.23551111e-02
-1.32069731e+00 -6.67532444e-01 -8.64220917e-01 1.99486822e-01
2.40579560e-01 -2.54489124e-01 1.49105096e+00 -1.00294089e+00
-2.23042059e+00 3.32729846e-01 -6.75669611e-01 -2.96540707e-01
1.41663343e-01 2.97506750e-01 -9.23291445e-01 1.20645352e-02
-1.64851367e-01 1.79366395e-01 1.55630636e+00 -1.30719781e+00
-2.09244817e-01 -2.23389849e-01 -4.76681352e-01 9.14107934e-02
-4.10565019e-01 2.38322258e-01 -4.41511959e-01 -7.83785343e-01
2.14561462e-01 -6.11572146e-01 1.97075412e-01 -4.10456747e-01
-3.57814461e-01 -1.44434378e-01 9.86430228e-01 -5.46780109e-01
1.54526305e+00 -2.66725850e+00 1.56526178e-01 1.45263270e-01
-1.75476700e-01 1.78290054e-01 -2.01922491e-01 5.01515269e-01
4.83962474e-03 -4.34634183e-03 -1.85334951e-01 -6.98190093e-01
7.47626051e-02 1.54799432e-01 -7.09754765e-01 2.07230702e-01
-6.04238510e-02 5.03004730e-01 -7.10236192e-01 -4.82986301e-01
1.96995616e-01 8.55565846e-01 -4.42276627e-01 3.79268825e-01
-1.09810717e-01 6.34103119e-01 -2.52207994e-01 5.36011219e-01
4.12376583e-01 1.66859552e-01 1.20247975e-01 1.40205950e-01
-2.15709001e-01 4.59432781e-01 -1.17148256e+00 1.61237156e+00
-6.21885121e-01 4.71987903e-01 3.22278559e-01 -6.31294191e-01
1.19194758e+00 9.89321113e-01 2.28374183e-01 -5.76801956e-01
8.22837204e-02 9.73178074e-02 6.25333097e-03 -3.15615088e-01
4.44231749e-01 -5.58094382e-01 -3.25713605e-02 4.55490798e-01
2.38079473e-01 -1.71653464e-01 -2.90853471e-01 -5.14152050e-02
4.66234118e-01 -3.98455948e-01 2.36439154e-01 5.47502041e-02
6.72973692e-01 -6.10785127e-01 6.61790133e-01 5.44859767e-01
-4.36163664e-01 5.87418616e-01 4.98510413e-02 2.71136612e-01
-6.23164654e-01 -1.16821885e+00 -2.01196849e-01 1.48443604e+00
-2.87256353e-02 -4.25912052e-01 -8.35269749e-01 1.42114654e-01
-2.86796242e-01 1.00781763e+00 -3.09858263e-01 -3.74017864e-01
-4.10296351e-01 -3.22099775e-01 7.61673331e-01 1.77286416e-01
1.06999032e-01 -9.43040848e-01 -2.40278885e-01 4.81380224e-01
-3.07690650e-01 -7.77254581e-01 -7.95825958e-01 1.26735464e-01
-5.54524779e-01 -3.89856786e-01 -7.02657402e-01 -7.62568533e-01
2.60804519e-02 3.40694070e-01 5.17367780e-01 -2.88411289e-01
-1.28294632e-01 4.26966161e-01 -4.62169111e-01 -3.96727800e-01
-9.32432652e-01 2.79479250e-02 3.62958193e-01 4.61057514e-01
1.70521915e-01 -6.71529472e-01 -1.40905738e-01 2.07597047e-01
-6.66161835e-01 -1.66059628e-01 1.86849847e-01 8.37796867e-01
4.46593136e-01 1.23887621e-01 1.05833495e+00 -3.34508508e-01
9.30851698e-01 -4.10575628e-01 -3.30349505e-01 2.03101367e-01
-5.86396575e-01 9.47906598e-02 5.96965134e-01 -9.71951604e-01
-1.23595524e+00 7.53455097e-03 -2.16459453e-01 -8.39734674e-01
-3.29566710e-02 3.48713458e-01 -3.80025506e-01 3.89223635e-01
6.33164406e-01 8.35268497e-01 2.55943596e-01 -5.49611688e-01
5.02672017e-01 1.14023101e+00 4.66388017e-01 -4.36184525e-01
6.04732096e-01 1.18521467e-01 -7.96383619e-01 -1.24577320e+00
-3.55768263e-01 -5.83091676e-01 -4.16589648e-01 -2.49433354e-01
6.59803629e-01 -1.02699792e+00 -4.66823846e-01 3.42645824e-01
-1.18443263e+00 -1.05749264e-01 -2.31649175e-01 6.13740623e-01
-5.83446503e-01 2.10802808e-01 -5.75322151e-01 -1.41070175e+00
-4.45554286e-01 -1.10679948e+00 1.09234166e+00 -1.08009562e-01
-1.89092278e-01 -7.33766675e-01 1.55644506e-01 1.63950354e-01
6.22500479e-01 -3.97311985e-01 8.31466138e-01 -5.72017789e-01
-4.16807234e-01 -1.63438206e-03 4.45230991e-01 3.50761354e-01
4.19113606e-01 6.69067800e-02 -1.46668351e+00 -3.59669834e-01
3.99872839e-01 1.08567610e-01 6.80774987e-01 3.85129869e-01
8.95229638e-01 -3.94748420e-01 -1.77215457e-01 4.52064246e-01
9.52693582e-01 4.21882898e-01 2.68145233e-01 -3.60296667e-01
4.22387540e-01 6.37603641e-01 2.69004762e-01 5.34971356e-01
1.28633529e-01 7.46324658e-01 -8.91855508e-02 3.01352829e-01
-2.59393364e-01 -3.95898968e-01 7.63500333e-01 1.33709908e+00
1.58121288e-01 -2.17127353e-01 -6.33067548e-01 4.70969707e-01
-1.30043197e+00 -1.15693521e+00 5.33287942e-01 2.22376251e+00
1.01349926e+00 1.15182117e-01 1.16381101e-01 3.36923152e-01
9.22207892e-01 4.10046816e-01 -6.01486385e-01 -5.43787360e-01
5.49920164e-02 -5.89918196e-02 -1.34987757e-01 7.42859006e-01
-5.70559978e-01 9.96061504e-01 6.50486660e+00 9.61976349e-01
-1.52702987e+00 3.84502321e-01 1.65514633e-01 -2.35238597e-01
-5.73969960e-01 -1.66011080e-01 -9.58834529e-01 4.01622593e-01
1.43715894e+00 -5.72738349e-01 7.88843036e-01 5.98012567e-01
4.49989229e-01 6.21766388e-01 -1.16263580e+00 1.10980022e+00
2.39642650e-01 -1.00920963e+00 1.28563657e-01 3.12025491e-02
2.58216321e-01 -1.91403523e-01 1.54199317e-01 7.02647328e-01
-9.61701497e-02 -9.16459382e-01 1.02798104e+00 5.15364766e-01
1.04421341e+00 -6.16058469e-01 7.75411129e-02 7.31897712e-01
-1.58928430e+00 -1.80153966e-01 -9.53277424e-02 2.84443289e-01
1.73665404e-01 1.90738946e-01 -1.00042367e+00 4.02421355e-01
2.60580480e-01 2.87209272e-01 8.95028841e-03 4.17499036e-01
7.90836960e-02 8.99971187e-01 -5.74902743e-02 -5.73463459e-03
-9.17114541e-02 2.01977253e-01 8.61777604e-01 1.17292500e+00
4.24441487e-01 8.83812010e-02 1.45331085e-01 1.00260413e+00
1.84747770e-01 1.31970003e-01 -4.37195420e-01 -2.87937254e-01
1.07686293e+00 7.55499423e-01 -1.50266245e-01 -3.31812322e-01
-2.12027997e-01 8.56818616e-01 -1.07909337e-01 6.14572227e-01
-6.43315077e-01 -2.17648730e-01 7.55899131e-01 8.52370262e-02
4.40500438e-01 -1.89570397e-01 1.45360634e-01 -1.07992280e+00
-2.78055578e-01 -9.38984990e-01 2.25871429e-02 -4.86701161e-01
-1.11742949e+00 8.76171589e-01 -6.65377304e-02 -1.41462493e+00
-8.66279960e-01 -1.67939439e-02 -7.54358947e-01 1.11973715e+00
-1.35959554e+00 -1.12610602e+00 8.60725120e-02 8.79947543e-01
1.05425441e+00 -3.80981803e-01 1.21773100e+00 1.06241725e-01
-5.82183480e-01 7.80488610e-01 5.07705286e-02 -1.19674549e-01
5.41747212e-01 -8.47413898e-01 2.77203679e-01 6.29980743e-01
1.06926046e-01 7.91681409e-01 7.27890730e-01 -3.96082371e-01
-1.46783042e+00 -1.06214142e+00 8.12074959e-01 -2.09740866e-02
4.07340556e-01 -6.41282499e-01 -1.14477682e+00 4.91743714e-01
-1.17950654e-02 -1.36922956e-01 9.36026812e-01 1.85966477e-01
-5.87216914e-01 -3.06256920e-01 -8.69319022e-01 4.53479439e-01
7.06037641e-01 -1.21511316e+00 -8.26681256e-01 -1.25603706e-01
1.30978632e+00 -9.28900316e-02 -5.80537498e-01 9.23437998e-03
5.88690460e-01 -5.62811732e-01 9.89228725e-01 -2.69274771e-01
-3.43588144e-01 -1.98136926e-01 -4.76333350e-01 -1.34389365e+00
-2.95311779e-01 -8.73301268e-01 -3.30238968e-01 1.52347529e+00
3.81472260e-01 -6.48640990e-01 2.10804209e-01 1.57892540e-01
-2.57005453e-01 -3.55337143e-01 -1.22325993e+00 -9.93606746e-01
-1.66016985e-02 -7.23244011e-01 1.06015801e+00 9.06102121e-01
1.53966397e-01 5.53644717e-01 -4.96151775e-01 3.31480473e-01
5.06901145e-01 1.47923112e-01 5.80473661e-01 -1.22370064e+00
-5.54111123e-01 -4.17941898e-01 1.33588742e-02 -9.90698159e-01
2.72966713e-01 -7.99727857e-01 3.99180323e-01 -1.06897497e+00
-1.08044945e-01 -1.33909643e-01 -4.92330909e-01 8.19830745e-02
1.17645249e-01 -4.07801688e-01 1.36471108e-01 4.57245201e-01
4.13428992e-02 8.92834365e-01 1.23731148e+00 -2.31619865e-01
-6.98575974e-01 2.88339376e-01 -6.10308647e-01 2.65438884e-01
5.40903986e-01 -3.11393142e-01 -7.88427234e-01 -3.10493540e-02
-5.31700790e-01 7.02589691e-01 1.15803219e-01 -8.49495590e-01
2.91132122e-01 -1.60039544e-01 -8.98074545e-03 -6.24661565e-01
8.20555925e-01 -4.20432597e-01 2.73113489e-01 3.23882818e-01
-5.35056710e-01 -5.00054657e-01 8.70772079e-02 7.76788414e-01
-3.88368964e-01 -1.28347620e-01 8.35791290e-01 -7.08837584e-02
-4.96453345e-01 2.83271372e-01 -8.21261108e-01 -3.45399648e-01
6.63706303e-01 -2.51273572e-01 -2.94572394e-02 -4.78119791e-01
-9.01638031e-01 -1.68950602e-01 5.21666743e-03 6.80553257e-01
7.35940933e-01 -1.27468181e+00 -6.05136633e-01 7.50759602e-01
3.02798033e-01 -3.08074832e-01 4.66574997e-01 4.46689427e-01
3.23068976e-01 5.12292862e-01 2.80007601e-01 -7.32276499e-01
-1.23280609e+00 6.72834039e-01 5.01490116e-01 2.15095922e-01
-6.31318212e-01 7.66715646e-01 1.82991624e-01 -5.48763454e-01
4.22841787e-01 -3.00054997e-01 -2.44787216e-01 1.59004331e-01
8.10508370e-01 -2.97268629e-02 1.10417735e-02 -8.77051592e-01
-3.44312489e-01 3.06534201e-01 9.80531145e-03 -7.50298798e-01
8.43451619e-01 -4.72761154e-01 2.19813570e-01 1.14215851e+00
1.15721202e+00 2.00633913e-01 -1.11318278e+00 -4.57922995e-01
-2.36301377e-01 -1.99612573e-01 3.22344333e-01 -4.81557786e-01
-6.29920959e-01 9.30661321e-01 7.41891205e-01 2.40200073e-01
1.04136789e+00 5.30939102e-02 7.59360254e-01 8.95163417e-02
3.40659648e-01 -1.15525365e+00 1.56465024e-01 5.09058833e-01
9.47014868e-01 -9.52937782e-01 -6.17755771e-01 -3.44570398e-01
-5.85976183e-01 9.23959434e-01 2.98053265e-01 4.12243903e-01
9.41912055e-01 1.63301304e-01 2.05039471e-01 8.49982798e-02
-1.04746699e+00 -1.43943280e-01 2.72857815e-01 5.35101116e-01
3.82101238e-01 3.76316518e-01 1.22419290e-01 9.26165104e-01
-5.82804739e-01 -3.80817592e-01 2.44656384e-01 7.52462149e-01
-8.29132438e-01 -1.01507401e+00 -6.13588095e-01 -7.38199204e-02
-1.20100409e-01 -1.55430168e-01 -2.73581237e-01 1.93275198e-01
-1.87239885e-01 1.34944761e+00 9.93644297e-02 -6.21696651e-01
2.23069489e-01 5.31221688e-01 1.55081972e-01 -8.33723366e-01
-4.45268124e-01 6.38887644e-01 -9.73015279e-02 -1.71216279e-01
-1.68297172e-01 -6.34268343e-01 -1.33650589e+00 -1.27266169e-01
-6.01305842e-01 3.19486499e-01 9.69288886e-01 9.25760388e-01
2.79615611e-01 8.60158384e-01 1.15400493e+00 -6.99911773e-01
-8.36115360e-01 -1.27159393e+00 -7.97514081e-01 -3.60697806e-02
8.31765294e-01 -4.87146527e-01 -6.61654890e-01 1.08215921e-01] | [14.959050178527832, 6.511525630950928] |
9c88691f-d1bf-4ffd-9848-377518a17a95 | cross-language-learning-with-adversarial-1 | 1706.06749 | null | http://arxiv.org/abs/1706.06749v1 | http://arxiv.org/pdf/1706.06749v1.pdf | Cross-language Learning with Adversarial Neural Networks: Application to Community Question Answering | We address the problem of cross-language adaptation for question-question
similarity reranking in community question answering, with the objective to
port a system trained on one input language to another input language given
labeled training data for the first language and only unlabeled data for the
second language. In particular, we propose to use adversarial training of
neural networks to learn high-level features that are discriminative for the
main learning task, and at the same time are invariant across the input
languages. The evaluation results show sizable improvements for our
cross-language adversarial neural network (CLANN) model over a strong
non-adversarial system. | ['Lluís Màrquez', 'Preslav Nakov', 'Israa Jaradat', 'Shafiq Joty'] | 2017-06-21 | null | null | null | null | ['question-similarity'] | ['natural-language-processing'] | [ 2.77895570e-01 1.91533178e-01 2.18656778e-01 -4.36876327e-01
-1.33171487e+00 -8.60007286e-01 7.12461710e-01 2.68528789e-01
-7.39540815e-01 4.53711450e-01 3.75353187e-01 -5.26295424e-01
2.31227770e-01 -7.49990940e-01 -5.94134450e-01 -1.47838384e-01
1.54947877e-01 8.34268332e-01 5.00905395e-01 -5.04260898e-01
-3.33523154e-01 2.76304901e-01 -8.79975736e-01 6.87609315e-01
8.22145641e-01 8.02486777e-01 -2.00767070e-01 1.18574333e+00
-2.77172983e-01 1.26503444e+00 -6.60868585e-01 -7.23720551e-01
1.63931087e-01 -5.28084874e-01 -1.55237460e+00 -2.44679332e-01
1.18027246e+00 -1.13763280e-01 -8.00879240e-01 8.94835591e-01
3.12148273e-01 1.54930919e-01 7.09892809e-01 -1.16569316e+00
-1.20313513e+00 2.50641167e-01 1.01711236e-01 4.79591966e-01
5.28645813e-01 1.12390473e-01 1.56170309e+00 -6.92681551e-01
3.66173625e-01 1.43078208e+00 3.34922105e-01 9.74492073e-01
-1.28725505e+00 -4.05119956e-01 1.86112836e-01 2.84810662e-01
-1.09167647e+00 -2.90045142e-01 6.76322818e-01 -4.44417417e-01
6.76887631e-01 4.43305254e-01 -3.06512296e-01 1.05361044e+00
-1.44619405e-01 7.53472388e-01 9.08488631e-01 -3.99730027e-01
6.36884868e-02 2.01096907e-01 5.57538629e-01 6.55397654e-01
-1.21171415e-01 -3.45107794e-01 -3.98295149e-02 -7.77639151e-01
-7.16337711e-02 -1.03101693e-01 -2.55184114e-01 -2.41736129e-01
-8.82798910e-01 1.03162801e+00 6.11150265e-01 7.19712973e-01
4.57589813e-02 8.07262808e-02 4.46605057e-01 8.26592863e-01
3.96001667e-01 8.73765349e-01 -6.87364459e-01 5.47931492e-01
-5.65207481e-01 4.39205766e-01 1.03380978e+00 5.08476198e-01
6.42551184e-01 -1.99863583e-01 -7.27254927e-01 9.89305615e-01
9.73053575e-02 6.94574952e-01 7.40471363e-01 -7.99086452e-01
7.59548068e-01 5.87869525e-01 -2.57149518e-01 -6.93129838e-01
-1.13864437e-01 -1.75309032e-01 -8.64793301e-01 -1.91073954e-01
6.79893196e-01 -1.01091817e-01 -5.29168546e-01 2.00023937e+00
-9.72556230e-03 3.98373827e-02 4.35856044e-01 5.48911214e-01
1.08041167e+00 5.85597157e-01 3.59777629e-01 4.04693455e-01
1.44164145e+00 -1.36778653e+00 -2.98658192e-01 -5.60731769e-01
5.46163559e-01 -6.06999040e-01 1.41521943e+00 -3.38860571e-01
-8.38654399e-01 -7.78479695e-01 -7.78264999e-01 -3.91130656e-01
-5.02563119e-01 -2.16414899e-01 9.61818993e-02 5.65701485e-01
-1.16861916e+00 2.53308773e-01 -2.40784064e-01 -5.21125555e-01
1.23454146e-01 1.55034572e-01 -3.22753817e-01 -3.06117296e-01
-1.46846807e+00 8.67411494e-01 3.73027593e-01 -5.24513125e-01
-9.15870786e-01 -6.82706714e-01 -7.90905118e-01 1.76461801e-01
5.73552214e-03 -7.45416164e-01 1.57097220e+00 -1.34918952e+00
-1.21344924e+00 1.12549603e+00 6.62609935e-02 -4.96614456e-01
2.47143313e-01 -5.03806630e-03 -4.93017703e-01 4.74901080e-01
3.08744550e-01 4.31322396e-01 8.03973079e-01 -1.09072435e+00
-2.77669966e-01 -2.57303953e-01 4.36190277e-01 -1.44142611e-02
-6.69958591e-01 1.34632289e-01 -3.58128309e-01 -7.14053094e-01
-3.92157346e-01 -6.92529440e-01 -1.98938221e-01 7.66972005e-02
-4.65290844e-02 -6.44301116e-01 7.51145661e-01 -1.19209790e+00
8.45677018e-01 -1.80183887e+00 4.95356321e-02 4.80564870e-02
3.67432982e-01 4.93975043e-01 -8.84371459e-01 2.72152603e-01
-3.40183109e-01 7.99020231e-02 -2.97696918e-01 -1.58937454e-01
-1.93450570e-01 2.67862201e-01 -4.75603074e-01 2.20285982e-01
5.80807686e-01 1.13710284e+00 -1.25990772e+00 -5.38098693e-01
-3.61730903e-01 -1.19496115e-01 -7.05401480e-01 1.01855147e+00
-4.32972550e-01 3.46111298e-01 -5.43095410e-01 1.20039813e-01
3.75150234e-01 -4.98145580e-01 1.19625963e-01 2.66394615e-02
8.55023623e-01 5.32930911e-01 -6.18596792e-01 1.29504490e+00
-8.27038765e-01 5.30995965e-01 1.07209638e-01 -9.96319294e-01
7.71717429e-01 4.73883718e-01 1.33008435e-01 -6.44965351e-01
-1.75089866e-01 2.57342458e-01 -8.90026987e-02 -6.02193356e-01
3.58499348e-01 -3.02172840e-01 -3.20635408e-01 6.73176289e-01
4.41084296e-01 -2.02080488e-01 -1.25409380e-01 5.39452434e-01
1.32452488e+00 -4.68021512e-01 3.87609154e-02 -1.98920295e-01
1.21433616e+00 -2.99601972e-01 -7.12951720e-02 8.49280238e-01
-2.79529393e-01 7.43976116e-01 2.97742963e-01 -2.18751371e-01
-1.01050818e+00 -1.12905097e+00 2.76066482e-01 1.78157127e+00
-1.53279722e-01 -1.10462874e-01 -6.93822443e-01 -1.51529455e+00
2.18869761e-01 7.80290782e-01 -7.49489605e-01 -4.92559791e-01
-8.73074889e-01 1.29080620e-02 1.13506854e+00 3.62133116e-01
3.67861837e-01 -9.14340317e-01 2.05688775e-01 -1.94529481e-02
-3.74189973e-01 -1.35393012e+00 -1.05108869e+00 -5.46454415e-02
-5.03157973e-01 -1.40695083e+00 -9.48677421e-01 -1.26713753e+00
6.57266855e-01 6.99492171e-02 1.65740895e+00 4.13000941e-01
2.12161198e-01 9.45419550e-01 -2.06632078e-01 1.73419178e-01
-9.67333198e-01 4.85286951e-01 -5.61291650e-02 2.63376176e-01
3.65336061e-01 -4.70221281e-01 -1.97945327e-01 2.08126843e-01
-1.21594214e+00 -4.72242892e-01 5.42408042e-02 1.02176630e+00
2.55245864e-01 -6.50361657e-01 8.26588631e-01 -9.30185735e-01
1.10207367e+00 -6.63816392e-01 -3.44017684e-01 1.00376749e+00
-2.21813247e-01 5.13983727e-01 1.17717481e+00 -6.42538428e-01
-9.06732321e-01 -3.65681857e-01 -4.80208606e-01 -2.98530757e-01
-6.29034340e-02 1.15044400e-01 -3.47175270e-01 -5.71719445e-02
1.03611755e+00 7.93789774e-02 -2.49285847e-01 -6.78450584e-01
7.48773217e-01 7.92625129e-01 8.15303087e-01 -7.81026483e-01
1.20793414e+00 2.28023455e-01 -4.68482643e-01 -5.57600915e-01
-1.10090303e+00 -7.07622409e-01 -6.85742617e-01 1.93919390e-01
1.08412099e+00 -7.39912808e-01 -6.28957152e-01 2.49524415e-01
-1.44736755e+00 -2.71586537e-01 -6.03512228e-01 -6.21259362e-02
-2.29746208e-01 6.09635890e-01 -5.47792852e-01 -3.46764326e-01
-6.15880132e-01 -7.63086915e-01 7.98021913e-01 7.16419220e-02
-1.75215542e-01 -1.53634751e+00 5.39644599e-01 9.94704127e-01
4.77000684e-01 -2.59732395e-01 1.29522896e+00 -1.82656515e+00
-2.91696012e-01 -3.26505601e-01 -1.83780953e-01 9.16156292e-01
2.25183651e-01 -6.71587884e-01 -9.58118200e-01 -5.30638814e-01
1.50858998e-01 -8.51009965e-01 6.75214767e-01 -6.12286925e-01
1.17392874e+00 -5.41764021e-01 1.60430282e-01 1.64474428e-01
9.92911220e-01 -5.82645953e-01 2.96617329e-01 1.70117050e-01
8.19817662e-01 6.18887186e-01 -7.06371590e-02 -5.31684458e-01
5.82340300e-01 6.98713720e-01 1.05227947e-01 -2.31738895e-01
-4.09336478e-01 -5.55930734e-01 3.03696990e-01 1.10501623e+00
7.01010823e-01 -3.30751002e-01 -1.02989984e+00 7.75841713e-01
-1.39420938e+00 -7.87023485e-01 8.50652605e-02 2.01441908e+00
1.01237857e+00 -2.61336118e-01 -8.61576200e-02 -3.85442436e-01
6.85232401e-01 4.38334912e-01 -4.61321294e-01 -4.07175571e-01
-2.24049479e-01 6.41629994e-01 2.22153038e-01 8.00235391e-01
-1.21308279e+00 7.25288332e-01 6.50149822e+00 7.35076368e-01
-7.70543277e-01 6.68663800e-01 4.76148546e-01 4.95395422e-01
-7.38356292e-01 -5.55914454e-03 -2.37989336e-01 1.63724214e-01
1.05326140e+00 -3.82772863e-01 3.39739650e-01 7.60489404e-01
-5.50055981e-01 4.93279070e-01 -1.39922738e+00 5.40388882e-01
4.54896390e-01 -1.08146906e+00 6.04426026e-01 -4.59979862e-01
8.57484579e-01 2.39522725e-01 -1.27927423e-01 7.09918201e-01
5.83299220e-01 -1.10909772e+00 1.92774206e-01 4.31959122e-01
6.34901047e-01 -3.47375691e-01 8.19852591e-01 4.31095958e-01
-9.40425515e-01 1.96864173e-01 -1.47950321e-01 8.59725177e-02
-1.72700733e-03 -8.34232867e-02 -8.22394609e-01 5.67467213e-01
4.57026482e-01 1.68153375e-01 -1.09755123e+00 5.76963544e-01
-2.26813674e-01 8.94955814e-01 -1.58504229e-02 -6.24832623e-02
3.01576972e-01 1.25097215e-01 4.50928032e-01 1.16785264e+00
-3.56460243e-01 -1.66402236e-01 4.45659339e-01 6.89063311e-01
-8.59418392e-01 4.12587285e-01 -5.47147155e-01 -4.65205610e-02
-4.89060283e-02 8.69681120e-01 1.82958305e-01 -5.81529558e-01
-5.49955010e-01 1.32327223e+00 7.77963758e-01 4.98666704e-01
-3.51132810e-01 -3.30206096e-01 6.21744752e-01 3.04009169e-02
2.93539409e-02 2.45297477e-02 1.18131451e-01 -1.52609968e+00
1.02581143e-01 -1.37826967e+00 9.56340551e-01 -5.82111776e-01
-2.19718862e+00 8.12582850e-01 -3.25752378e-01 -8.40884924e-01
-4.50609207e-01 -6.04627013e-01 -7.49331295e-01 1.28826213e+00
-1.59026408e+00 -1.33051372e+00 5.13305888e-02 9.50692892e-01
5.68057358e-01 -6.99407935e-01 1.14911771e+00 4.55312073e-01
2.44600102e-02 1.13457441e+00 4.24940258e-01 6.83097720e-01
8.90032649e-01 -1.34755039e+00 7.78470337e-01 8.70453835e-01
6.14401877e-01 4.79612172e-01 3.42942208e-01 -1.86741650e-01
-9.65794444e-01 -1.37958550e+00 1.39914250e+00 -9.17773247e-01
1.06606269e+00 -3.51673305e-01 -1.42893267e+00 6.77431345e-01
4.22821581e-01 2.39797816e-01 7.65519857e-01 5.52473329e-02
-1.09920025e+00 -2.68813297e-02 -1.16703391e+00 4.48505342e-01
2.76868433e-01 -1.52024162e+00 -1.24157023e+00 6.47934616e-01
1.12872601e+00 -1.28626190e-02 -8.05829644e-01 1.28175825e-01
1.56430781e-01 -3.58627915e-01 1.21670365e+00 -1.63735795e+00
2.31479004e-01 -6.37276471e-02 -2.96971262e-01 -8.42111170e-01
-3.43846083e-01 -3.30917358e-01 1.15096718e-01 1.36857677e+00
4.94135857e-01 -6.55500412e-01 5.41220903e-01 4.89372522e-01
4.26539630e-01 -3.55142921e-01 -1.17968845e+00 -6.80557013e-01
8.46211016e-01 -2.25735769e-01 3.92955542e-01 1.03798866e+00
-3.59443128e-01 1.12377095e+00 -2.01967880e-01 3.83338511e-01
3.20836037e-01 6.07976690e-02 6.49714291e-01 -1.00561881e+00
-5.37528217e-01 -1.67632788e-01 -1.52562633e-01 -1.26155376e+00
8.73507738e-01 -1.48989475e+00 -4.39271182e-02 -1.39628780e+00
3.42651546e-01 -2.33154774e-01 -4.72154319e-01 2.68399537e-01
-8.39382052e-01 1.22127622e-01 3.31644386e-01 1.76414609e-01
-1.00618184e+00 7.19233036e-01 1.05391777e+00 -7.49176502e-01
3.26167285e-01 3.51108819e-01 -7.00053096e-01 5.70191562e-01
5.96618056e-01 -6.91917896e-01 -4.00232673e-01 -7.03838766e-01
-1.00123778e-01 -5.44360802e-02 4.98943686e-01 -6.35823071e-01
1.72731951e-01 5.71148694e-02 -2.07447708e-01 -1.60270631e-02
2.24239156e-02 -8.58295918e-01 -6.73447311e-01 5.33567190e-01
-1.10859287e+00 1.15693673e-01 9.80392098e-02 6.50126338e-01
-3.12139392e-01 -6.54409885e-01 8.47677290e-01 3.01182494e-02
-2.63633013e-01 3.51405799e-01 -1.28328741e-01 1.14007831e+00
3.91788006e-01 6.10480070e-01 -4.04062808e-01 -7.61871457e-01
-5.89922905e-01 5.17097116e-01 1.96303576e-01 8.93440843e-01
2.22813591e-01 -1.45686781e+00 -1.22825694e+00 -1.68813124e-01
4.87198770e-01 -5.09637117e-01 1.63648594e-02 -5.84362075e-02
-4.21359897e-01 4.67843533e-01 1.65796876e-01 -3.17324132e-01
-1.26020503e+00 7.24097013e-01 9.18969393e-01 -1.02857757e+00
9.86214355e-02 9.85152900e-01 2.66235828e-01 -1.33147895e+00
1.13612533e-01 -2.40287427e-02 -2.72793978e-01 -2.93092966e-01
3.94195855e-01 1.14398852e-01 1.74019948e-01 -7.71157026e-01
-3.57070565e-01 2.59196758e-01 -1.32205710e-01 -8.18398502e-03
9.46740508e-01 1.04318768e-01 -3.40948880e-01 3.01447511e-01
1.76835132e+00 2.27621406e-01 -4.64297652e-01 -9.05581713e-01
3.28068793e-01 -5.09772152e-02 -4.91469920e-01 -7.05327392e-01
-8.18406343e-01 9.16087449e-01 7.18071222e-01 2.35101417e-01
7.27942467e-01 4.89304900e-01 8.65817785e-01 9.64123964e-01
-1.56840131e-01 -6.34112120e-01 5.88503420e-01 9.33077395e-01
1.14405036e+00 -1.25117159e+00 -3.74635398e-01 -1.81148201e-03
-4.59874719e-01 8.51299047e-01 6.99724972e-01 -2.73639739e-01
5.50156951e-01 -3.55999678e-01 3.23616326e-01 -8.70881453e-02
-6.79476976e-01 -2.00324625e-01 7.76183009e-01 6.72753692e-01
3.99061441e-01 -5.27352691e-02 5.45393527e-02 5.64106464e-01
-2.78488472e-02 -4.76117611e-01 -2.20645145e-02 4.40467805e-01
-5.20414785e-02 -1.48221040e+00 -3.35919797e-01 3.41063380e-01
-7.15039611e-01 -2.50667959e-01 -5.84192157e-01 4.78782594e-01
-2.79569030e-02 1.11602807e+00 -6.12035096e-02 -3.27102751e-01
6.16005182e-01 4.18670833e-01 1.50295466e-01 -8.34945500e-01
-1.26570582e+00 -8.74095500e-01 2.71513224e-01 -1.69435531e-01
-2.42469549e-01 -3.29567730e-01 -7.40696430e-01 2.59337634e-01
-1.53377146e-01 6.81579173e-01 5.73471561e-03 1.19381952e+00
1.73113257e-01 4.21516627e-01 8.77678990e-01 2.09998414e-02
-1.05843675e+00 -1.09118426e+00 -4.18416299e-02 1.08825397e+00
6.50364876e-01 9.01485160e-02 -4.78451371e-01 2.08915740e-01] | [11.326654434204102, 8.202881813049316] |
cc2c78a0-d8ca-4318-894f-c5548bf75951 | position-bias-mitigation-a-knowledge-aware | 2106.03518 | null | https://arxiv.org/abs/2106.03518v2 | https://arxiv.org/pdf/2106.03518v2.pdf | Position Bias Mitigation: A Knowledge-Aware Graph Model for Emotion Cause Extraction | The Emotion Cause Extraction (ECE)} task aims to identify clauses which contain emotion-evoking information for a particular emotion expressed in text. We observe that a widely-used ECE dataset exhibits a bias that the majority of annotated cause clauses are either directly before their associated emotion clauses or are the emotion clauses themselves. Existing models for ECE tend to explore such relative position information and suffer from the dataset bias. To investigate the degree of reliance of existing ECE models on clause relative positions, we propose a novel strategy to generate adversarial examples in which the relative position information is no longer the indicative feature of cause clauses. We test the performance of existing models on such adversarial examples and observe a significant performance drop. To address the dataset bias, we propose a novel graph-based method to explicitly model the emotion triggering paths by leveraging the commonsense knowledge to enhance the semantic dependencies between a candidate clause and an emotion clause. Experimental results show that our proposed approach performs on par with the existing state-of-the-art methods on the original ECE dataset, and is more robust against adversarial attacks compared to existing models. | ['Yulan He', 'Gabriele Pergola', 'Lin Gui', 'Hanqi Yan'] | 2021-06-07 | null | https://aclanthology.org/2021.acl-long.261 | https://aclanthology.org/2021.acl-long.261.pdf | acl-2021-5 | ['emotion-cause-extraction'] | ['natural-language-processing'] | [ 5.51586747e-01 6.20831609e-01 -1.25623912e-01 -5.57600975e-01
-7.44722366e-01 -9.27044988e-01 7.86182880e-01 3.42940956e-01
3.85094956e-02 7.26212740e-01 5.18897831e-01 -1.21294148e-01
5.68845756e-02 -9.91441905e-01 -8.05683315e-01 -4.01982814e-01
-5.71098253e-02 2.73319542e-01 -7.77291879e-02 -6.02769434e-01
1.42804906e-01 3.45555723e-01 -1.21968400e+00 6.78405166e-01
8.86261106e-01 8.24542701e-01 -8.19924653e-01 3.24177504e-01
-2.53855407e-01 1.40146923e+00 -1.15043950e+00 -1.00635982e+00
2.09010206e-02 -6.74912095e-01 -1.08640754e+00 -1.19862400e-01
2.27433264e-01 1.27856031e-01 -3.06857258e-01 1.43186474e+00
4.14771080e-01 -1.63069651e-01 6.48014784e-01 -1.90523565e+00
-7.51788020e-01 1.15449607e+00 -3.14629197e-01 1.31934747e-01
6.53461277e-01 -1.41388074e-01 1.36871672e+00 -5.59386671e-01
1.03833389e+00 1.39128435e+00 5.10352850e-01 9.02939796e-01
-9.40529525e-01 -8.79654646e-01 4.02282625e-01 3.72515678e-01
-1.11845672e+00 -2.87209809e-01 1.49400473e+00 2.10018437e-02
9.66517150e-01 6.12290323e-01 4.33963567e-01 1.60735202e+00
3.39640260e-01 8.42753649e-01 1.30891967e+00 -2.47522295e-01
3.36844921e-01 6.87035620e-02 9.16827470e-02 4.69912618e-01
-3.77847031e-02 1.05677411e-01 -6.05437815e-01 -4.88565177e-01
-6.91725016e-02 -4.26200807e-01 -3.45027477e-01 -1.35807022e-01
-8.49085331e-01 1.13523018e+00 6.40876353e-01 3.37279797e-01
-2.54801601e-01 1.90381274e-01 7.77972341e-01 2.64628768e-01
6.21389389e-01 8.65968704e-01 -6.39781594e-01 1.01645710e-02
-5.36289930e-01 5.46839833e-01 9.94461954e-01 1.01602173e+00
4.79939103e-01 -1.03938237e-01 -1.49987340e-01 4.31904078e-01
2.19027568e-02 1.10985145e-01 2.21536055e-01 -6.10763907e-01
5.89091063e-01 9.63396966e-01 -1.33030847e-01 -1.43509996e+00
-1.51021436e-01 -3.72120768e-01 -7.03809500e-01 -1.35540575e-01
5.89486659e-02 -2.47368485e-01 -6.34469926e-01 2.04196477e+00
4.77990925e-01 2.33286306e-01 5.13897240e-01 8.08505535e-01
1.03008258e+00 5.79908371e-01 3.29088300e-01 -2.04947025e-01
1.16870582e+00 -6.25763476e-01 -1.09666526e+00 -5.13176858e-01
7.67393172e-01 -6.23283803e-01 1.03626382e+00 2.58791596e-01
-6.37926042e-01 1.47934422e-01 -1.14190090e+00 1.79474294e-01
-5.99576414e-01 -4.04209912e-01 9.35273349e-01 5.20067155e-01
-5.06531239e-01 5.69806755e-01 -1.09625272e-01 2.69117821e-02
3.14908087e-01 2.08228871e-01 -3.53098035e-01 -1.28858507e-01
-1.94061303e+00 7.87835181e-01 6.32972836e-01 6.80816621e-02
-5.89888871e-01 -7.58362710e-01 -1.22854710e+00 -2.20253449e-02
7.48369992e-01 -3.23762208e-01 8.35762978e-01 -1.46165752e+00
-1.09038579e+00 9.64894831e-01 -2.55537443e-02 -4.40224379e-01
2.53375530e-01 -2.95554161e-01 -8.90202940e-01 1.61639377e-01
7.03699514e-02 5.69681466e-01 7.10149884e-01 -1.57066667e+00
-1.95458829e-01 -2.52359957e-01 5.43529689e-01 -1.81164108e-02
-3.39674503e-01 3.40642333e-01 6.61270916e-02 -9.71815228e-01
-2.58655995e-01 -9.64659095e-01 -8.14096183e-02 -3.88353497e-01
-1.03808117e+00 -4.22371656e-01 1.08796215e+00 -2.73661494e-01
1.48264480e+00 -2.08667445e+00 6.62910566e-02 4.10187632e-01
2.65277237e-01 5.00658825e-02 1.08472742e-02 4.97415334e-01
-6.27067685e-01 6.12305760e-01 -3.46441597e-01 1.74966633e-01
2.86116064e-01 4.42958713e-01 -7.90661812e-01 2.16663882e-01
5.76305687e-01 8.62472415e-01 -1.05489755e+00 -6.52754128e-01
-1.63486257e-01 2.15246856e-01 -6.74073875e-01 3.34022701e-01
-3.94544184e-01 -9.73249413e-03 -4.96022731e-01 7.32167244e-01
5.82996368e-01 3.77743132e-02 2.66867071e-01 -4.61405128e-01
6.04192734e-01 5.09832263e-01 -9.56429422e-01 1.27061057e+00
-2.01695338e-01 4.93256360e-01 -1.74355879e-02 -7.26132572e-01
9.48930919e-01 4.68417972e-01 1.19355127e-01 -2.87013233e-01
2.82230020e-01 4.36748341e-02 2.13396460e-01 -4.80164349e-01
2.85189122e-01 -6.35537207e-01 -6.67288005e-01 2.47685298e-01
-1.34840518e-01 -3.39149565e-01 4.81717065e-02 6.85826838e-01
1.53373432e+00 -3.99177037e-02 5.26971936e-01 -1.85833812e-01
4.70471293e-01 2.55236268e-01 9.24001873e-01 3.29942852e-01
-3.60205293e-01 7.38448426e-02 1.05314791e+00 -3.29392731e-01
-5.84009230e-01 -7.08927870e-01 1.71794102e-01 8.36951315e-01
2.10839927e-01 -7.80009866e-01 -7.45649159e-01 -1.35845363e+00
-2.45538637e-01 1.35041893e+00 -1.06035888e+00 -5.79758286e-01
-4.42078948e-01 -7.08269119e-01 9.65958893e-01 4.56650525e-01
4.98907954e-01 -1.48543262e+00 -2.25773707e-01 1.66364983e-01
-4.78960663e-01 -1.17745256e+00 -2.12995738e-01 2.57146090e-01
-3.69230837e-01 -1.15205586e+00 2.16109931e-01 -4.50490564e-01
7.21209407e-01 -6.14592433e-01 1.53278196e+00 8.34300965e-02
-1.43571496e-01 2.62672782e-01 -6.50365949e-01 -6.80547297e-01
-6.98270380e-01 -1.72497481e-01 -2.69524664e-01 3.00047342e-02
6.57285810e-01 -5.63600659e-01 -3.17117959e-01 -6.98236302e-02
-1.29923797e+00 -2.47426167e-01 3.41252834e-01 7.17809260e-01
4.33011532e-01 3.49573404e-01 9.05162275e-01 -1.59072888e+00
8.17971289e-01 -7.78460264e-01 -3.68993776e-03 2.35785186e-01
-6.59112871e-01 4.60210405e-02 8.51148725e-01 -5.24983764e-01
-1.03627563e+00 -4.46607247e-02 -2.01629609e-01 -5.00230670e-01
-3.31648558e-01 6.02783322e-01 -6.92381084e-01 2.02419147e-01
5.40676951e-01 -1.25209317e-01 -6.65303171e-01 2.21374512e-01
4.96116966e-01 1.50914147e-01 5.61245561e-01 -8.28858674e-01
1.02420652e+00 3.41927975e-01 1.68834236e-02 -6.27621114e-02
-1.22795272e+00 -9.86231863e-02 -2.14131296e-01 -2.70574898e-01
7.67825186e-01 -5.99453390e-01 -4.13087100e-01 1.01062186e-01
-1.39207470e+00 -4.84825969e-02 -2.50898242e-01 -6.61979392e-02
-3.10558081e-01 1.24862701e-01 -8.00736725e-01 -6.78021312e-01
-4.31276470e-01 -7.75266171e-01 1.00058830e+00 -1.44146249e-01
-7.59404123e-01 -1.07255757e+00 1.71169862e-02 2.26476833e-01
-7.16767646e-03 1.11407828e+00 1.41202164e+00 -1.23682141e+00
-5.18040247e-02 -3.91815126e-01 7.47524053e-02 1.01604156e-01
1.39354303e-01 2.00520352e-01 -9.28763330e-01 3.76066491e-02
1.54144317e-01 -6.70262575e-01 4.13866401e-01 -2.91638613e-01
1.13961518e+00 -8.04058373e-01 -2.09879518e-01 1.80202931e-01
1.58344889e+00 1.08305857e-01 8.56970906e-01 2.92278588e-01
7.07779348e-01 8.91596079e-01 9.71245766e-01 3.19059879e-01
1.61134839e-01 3.39454949e-01 9.61186290e-01 -3.10060263e-01
2.69367307e-01 -4.04758692e-01 5.31306386e-01 4.57796395e-01
2.97452450e-01 -6.27600372e-01 -7.33791530e-01 7.11440146e-01
-1.65325320e+00 -1.21251392e+00 -2.98703462e-01 1.42913795e+00
1.18986225e+00 3.66039902e-01 -3.51160377e-01 4.72798705e-01
7.94239819e-01 3.62667888e-01 -3.63734841e-01 -9.10839438e-01
-2.85356373e-01 4.33396041e-01 9.93359163e-02 4.18162256e-01
-1.18644679e+00 1.17480969e+00 6.22810316e+00 9.28173423e-01
-8.80331099e-01 -6.43243715e-02 5.35780489e-01 -4.64902893e-02
-8.36457789e-01 3.01734023e-02 -2.71804720e-01 1.28741249e-01
6.30118132e-01 -1.71680301e-01 2.44848028e-01 1.04169977e+00
-1.73136085e-01 4.59245235e-01 -1.36788249e+00 4.88303632e-01
3.50324839e-01 -1.08537388e+00 2.30518386e-01 -6.51960000e-02
6.30765796e-01 -6.23907626e-01 -1.78765878e-02 5.21946192e-01
3.77863914e-01 -1.14437640e+00 8.03087115e-01 5.65102473e-02
5.21877944e-01 -1.20381367e+00 1.06077647e+00 4.65631187e-02
-1.05911171e+00 4.51904610e-02 -5.21537699e-02 -1.12108812e-01
-9.04023275e-02 6.59370720e-01 -8.16809535e-01 5.24961412e-01
4.49070692e-01 5.66466749e-01 -6.96426272e-01 1.36140347e-01
-9.62525725e-01 9.25966740e-01 2.15855911e-02 -2.29308382e-01
3.14142108e-01 2.55735487e-01 8.25685382e-01 1.34728062e+00
-8.90624523e-02 3.40116143e-01 -6.64514154e-02 1.23284388e+00
-5.05192399e-01 1.23662122e-01 -9.21472251e-01 -1.48601621e-01
5.69751441e-01 1.42350185e+00 -7.28097737e-01 -3.62449735e-01
-2.05520377e-01 1.27870226e+00 3.96456182e-01 2.25766778e-01
-1.30382442e+00 -5.90803206e-01 5.08587062e-01 -1.85846031e-01
1.88659728e-01 5.18800020e-01 -9.83581394e-02 -9.59130764e-01
1.65713117e-01 -1.28500247e+00 5.62194943e-01 -8.67795467e-01
-1.71367943e+00 7.45963812e-01 -7.77111724e-02 -8.53745520e-01
-3.98588896e-01 -4.41921711e-01 -1.03426075e+00 2.80222625e-01
-1.33391213e+00 -1.26691604e+00 5.39793149e-02 9.46825862e-01
3.25224698e-01 1.90398470e-01 1.17579055e+00 8.68195221e-02
-5.54255664e-01 7.08128273e-01 -8.61364067e-01 4.36041623e-01
8.25154126e-01 -1.50240648e+00 1.16675749e-01 1.18566489e+00
2.44593725e-01 7.37250447e-01 1.12653005e+00 -8.86808276e-01
-1.22625065e+00 -1.28162062e+00 1.22742140e+00 -6.33879721e-01
9.02657151e-01 -5.50213933e-01 -9.42070425e-01 9.47319090e-01
5.28728306e-01 7.11111128e-02 1.04707265e+00 1.87830880e-01
-7.78553009e-01 6.01249188e-02 -1.43327475e+00 6.82877541e-01
1.07297993e+00 -6.58466876e-01 -1.22213852e+00 2.29308814e-01
1.05861485e+00 -2.54420698e-01 -7.42506742e-01 6.68494225e-01
-5.22596650e-02 -9.14321244e-01 6.47707403e-01 -1.05521691e+00
1.09706461e+00 -2.61735618e-01 -2.04231367e-01 -1.41781867e+00
8.07844382e-03 -8.49852741e-01 -1.76079035e-01 1.90965998e+00
6.91732943e-01 -3.75791728e-01 5.46418488e-01 1.01246834e+00
-1.47964388e-01 -8.06397557e-01 -8.76243651e-01 -4.59295452e-01
5.81580140e-02 -6.15854859e-01 7.18277872e-01 1.50347686e+00
3.89410764e-01 6.03060365e-01 -1.91100091e-01 3.97356123e-01
4.40580338e-01 2.99778044e-01 4.87546682e-01 -8.43421638e-01
-2.85746813e-01 -2.25889683e-01 -3.82643342e-01 -1.54741496e-01
8.80512893e-01 -9.84515727e-01 1.22273214e-01 -1.32282400e+00
2.09720373e-01 -2.55872965e-01 -3.09573054e-01 7.01782763e-01
-7.08073854e-01 1.24689013e-01 1.22046627e-01 -2.96677768e-01
-6.13963187e-01 5.69402218e-01 1.00225234e+00 -2.38897473e-01
2.01092184e-01 -4.11548674e-01 -1.02180791e+00 1.04220366e+00
6.82714760e-01 -8.31295431e-01 -5.37084341e-01 1.17009170e-01
7.57716000e-01 -2.08931863e-01 5.30895829e-01 -5.68657637e-01
4.52142507e-02 -2.94061899e-01 1.13186046e-01 -2.74946302e-01
2.90522613e-02 -1.00338423e+00 -2.76280157e-02 2.78458804e-01
-6.66273117e-01 8.76903757e-02 2.71660507e-01 6.09097421e-01
-5.60156286e-01 -1.96352169e-01 4.10055608e-01 4.12318148e-02
-6.38694465e-01 3.94251458e-02 -3.53646785e-01 5.25582433e-01
1.08271456e+00 2.54086643e-01 -4.55201089e-01 -5.12744069e-01
-4.79127228e-01 -8.34596604e-02 1.98438138e-01 4.68903244e-01
7.82041192e-01 -1.50973308e+00 -7.26153731e-01 -3.54404271e-01
3.49071890e-01 -2.05917552e-01 -4.14605998e-02 3.12815905e-01
4.07130048e-02 -1.16785988e-01 1.14207014e-01 1.50441408e-01
-1.41483080e+00 1.02559185e+00 3.77286345e-01 -6.68235660e-01
-3.41702998e-01 9.89383638e-01 1.96011543e-01 -4.27631438e-01
-6.85432851e-02 -1.90924913e-01 -1.93380833e-01 -1.89415202e-01
1.94695935e-01 3.04694828e-02 -1.31618395e-01 -6.24762774e-01
-7.07026243e-01 1.11539885e-01 8.78838748e-02 1.35247409e-02
1.16751587e+00 3.25726032e-01 -4.68299568e-01 2.43753120e-01
1.18855071e+00 6.26454055e-01 -5.55661798e-01 6.94030151e-02
8.25108886e-02 -2.98373818e-01 -2.07431093e-01 -1.17646468e+00
-1.17756033e+00 5.21526158e-01 -4.90593836e-02 2.87393957e-01
1.38162458e+00 2.41743267e-01 8.22083235e-01 1.95783842e-02
1.89739659e-01 -9.70820129e-01 1.85932547e-01 3.89572680e-01
1.27797127e+00 -9.98225510e-01 -9.97995138e-02 -1.13800347e+00
-8.99722397e-01 9.79266822e-01 9.15844381e-01 -3.81046683e-01
2.57823050e-01 6.48583710e-01 2.13566676e-01 -7.32531846e-01
-8.90071571e-01 1.19997688e-01 4.04260159e-01 5.08846760e-01
5.42145371e-01 7.90835544e-02 -3.39693934e-01 1.15746045e+00
-4.78976727e-01 -5.04213572e-01 3.94858450e-01 9.00066435e-01
2.06465423e-01 -1.18298638e+00 -2.78982550e-01 1.01476878e-01
-1.00210690e+00 -4.55157340e-01 -1.52341366e+00 1.09008145e+00
3.55484813e-01 1.20740688e+00 -3.78155410e-01 -6.14990354e-01
3.97994369e-01 2.76155829e-01 2.59021074e-01 -5.19081891e-01
-1.13830328e+00 -1.97794557e-01 6.60041749e-01 -7.79598117e-01
-6.68057144e-01 -3.32210124e-01 -1.62288749e+00 -3.30535531e-01
-4.18340147e-01 1.96540684e-01 1.57849073e-01 1.07442105e+00
2.19209433e-01 9.00082409e-01 7.06834972e-01 -1.84678555e-01
-2.97176242e-01 -6.92964852e-01 -4.28396821e-01 1.11662054e+00
-3.65633145e-02 -4.33290362e-01 -7.19107151e-01 -2.55456921e-02] | [12.588302612304688, 6.197442054748535] |
b7a44fd7-6c05-4e58-a7be-8e04055c0abd | scribbleseg-scribble-based-interactive-image | 2303.11320 | null | https://arxiv.org/abs/2303.11320v1 | https://arxiv.org/pdf/2303.11320v1.pdf | ScribbleSeg: Scribble-based Interactive Image Segmentation | Interactive segmentation enables users to extract masks by providing simple annotations to indicate the target, such as boxes, clicks, or scribbles. Among these interaction formats, scribbles are the most flexible as they can be of arbitrary shapes and sizes. This enables scribbles to provide more indications of the target object. However, previous works mainly focus on click-based configuration, and the scribble-based setting is rarely explored. In this work, we attempt to formulate a standard protocol for scribble-based interactive segmentation. Basically, we design diversified strategies to simulate scribbles for training, propose a deterministic scribble generator for evaluation, and construct a challenging benchmark. Besides, we build a strong framework ScribbleSeg, consisting of a Prototype Adaption Module(PAM) and a Corrective Refine Module (CRM), for the task. Extensive experiments show that ScribbleSeg performs notably better than previous click-based methods. We hope this could serve as a more powerful and general solution for interactive segmentation. Our code will be made available. | ['Hengshuang Zhao', 'Ser-Nam Lim', 'Yau Shing Jonathan Cheung', 'Xi Chen'] | 2023-03-20 | null | null | null | null | ['interactive-segmentation'] | ['computer-vision'] | [ 9.60147530e-02 -4.50040959e-02 -4.86619212e-02 -4.18361813e-01
-6.78375363e-01 -9.59041417e-01 5.21424353e-01 -2.32924800e-02
-2.42742509e-01 4.76336181e-01 -2.76677996e-01 -5.48589587e-01
1.71168000e-01 -6.29155099e-01 -6.03054404e-01 -5.36670268e-01
1.61388367e-01 6.89989686e-01 8.94389331e-01 -2.51356781e-01
4.14522529e-01 4.30198342e-01 -1.47560441e+00 3.56972903e-01
1.26020718e+00 5.81803620e-01 5.98267317e-01 5.75056374e-01
-1.98661700e-01 1.35076523e-01 -9.20894384e-01 -5.67106068e-01
2.62225300e-01 -2.15753764e-01 -6.86134577e-01 1.05699085e-01
3.32726866e-01 -2.86082178e-01 4.81637925e-01 7.74946094e-01
4.74163622e-01 6.70431927e-02 5.51714718e-01 -1.08546853e+00
-3.78978163e-01 9.59911585e-01 -6.48997068e-01 -8.66279975e-02
5.45180798e-01 5.37795901e-01 1.11294448e+00 -1.01088464e+00
7.10954309e-01 1.16231382e+00 5.54918110e-01 7.04174995e-01
-1.31307292e+00 -6.10502899e-01 3.42235923e-01 -9.03914869e-02
-1.19219708e+00 -2.46548325e-01 6.31216943e-01 -5.29558122e-01
4.39860493e-01 8.69629025e-01 7.35102892e-01 1.04806983e+00
-1.64180741e-01 1.28017116e+00 1.09193385e+00 -5.30852437e-01
1.97669968e-01 3.59737754e-01 2.27789566e-01 6.89073384e-01
-6.57435134e-02 -5.22453748e-02 -2.49000445e-01 -4.15543281e-03
9.12741184e-01 -6.72326609e-02 -4.36873049e-01 -5.86782396e-01
-1.18319058e+00 7.19832301e-01 4.28494006e-01 6.43193424e-02
3.03991120e-02 1.44366426e-02 1.24999650e-01 1.21916793e-01
3.77777427e-01 7.12874293e-01 -4.19053078e-01 -5.09159788e-02
-9.51374829e-01 6.87423587e-01 9.11833584e-01 1.03750908e+00
6.91641331e-01 -3.47918332e-01 -7.60595798e-01 9.80484903e-01
1.99182034e-01 1.10464454e-01 1.13143101e-01 -8.79478514e-01
3.74197781e-01 5.21234691e-01 3.43317121e-01 -5.92077672e-01
-2.57262796e-01 -3.25165629e-01 -3.36209655e-01 5.84032834e-01
5.33229649e-01 -2.88830817e-01 -1.12733686e+00 1.32247007e+00
4.31520164e-01 2.65313033e-02 -6.47386968e-01 1.04380131e+00
9.40880835e-01 4.72418517e-01 -5.51979244e-03 2.91349113e-01
1.24970281e+00 -1.31514549e+00 -5.31394243e-01 -1.32331833e-01
5.20877838e-01 -7.91571975e-01 1.75817108e+00 5.32149792e-01
-1.13566577e+00 -4.87712324e-01 -8.84380102e-01 1.24353096e-01
-4.44658399e-01 1.84410870e-01 8.27042401e-01 7.89291322e-01
-1.03720355e+00 6.95951819e-01 -9.03993964e-01 -1.61671892e-01
3.18555713e-01 1.50283426e-01 2.22166732e-01 2.78959960e-01
-8.48621547e-01 5.18224716e-01 3.13355535e-01 -3.93373258e-02
-6.96963251e-01 -5.45286655e-01 -6.91285014e-01 -3.73307057e-02
6.59669220e-01 -7.33651161e-01 1.47622192e+00 -8.01677644e-01
-1.55958152e+00 8.50926220e-01 9.14112199e-03 -3.42494845e-01
8.55248630e-01 -2.26597100e-01 2.15502977e-01 -5.61948679e-02
-5.19615635e-02 9.72207785e-01 8.65801275e-01 -1.62708557e+00
-4.63356555e-01 -1.44377146e-02 3.36129248e-01 1.24930926e-01
-7.03328103e-02 2.72053272e-01 -1.02613962e+00 -8.21459949e-01
-1.46784574e-01 -9.05702531e-01 -5.15407503e-01 -7.13923052e-02
-9.75379348e-01 -2.51306087e-01 7.22411096e-01 -3.19370419e-01
1.62988281e+00 -2.09559894e+00 8.34069506e-05 2.16206670e-01
1.76096708e-01 3.82919312e-01 -2.06058174e-02 5.00284970e-01
-4.05208096e-02 5.36884844e-01 -3.76170099e-01 -6.21953011e-01
2.12690607e-01 3.02360766e-02 -1.89519808e-01 -1.74029231e-01
8.76621678e-02 1.05525088e+00 -9.10175323e-01 -4.96762246e-01
2.26954028e-01 7.08375871e-02 -7.95585871e-01 2.05723301e-01
-5.78612983e-01 5.54356813e-01 -5.17357171e-01 8.07705343e-01
7.36075819e-01 -2.91294605e-01 -1.52714327e-01 1.83685590e-02
-2.26784840e-01 2.76558757e-01 -1.11285174e+00 1.70197487e+00
-3.24619114e-01 4.84402031e-01 1.98126167e-01 -3.40595990e-01
8.23261023e-01 -1.21174954e-01 6.10058196e-02 -1.40688851e-01
3.81990373e-02 1.37577616e-02 -4.33732793e-02 -3.95828515e-01
6.47194862e-01 4.33723778e-01 -7.88585320e-02 4.49333370e-01
-2.83058107e-01 -5.48423529e-01 6.07237518e-01 3.69023651e-01
1.09962559e+00 4.74981010e-01 2.05621645e-01 -7.71969408e-02
1.38674244e-01 -1.36601314e-01 3.33337098e-01 1.10738373e+00
6.55024797e-02 1.06118917e+00 6.38117373e-01 -1.80136964e-01
-7.71225870e-01 -1.14161587e+00 -8.34002793e-02 1.27776706e+00
3.46314192e-01 -6.74830258e-01 -1.07713902e+00 -9.54869688e-01
-1.14711992e-01 8.55355740e-01 -5.02815127e-01 3.59019756e-01
-5.97474158e-01 -6.10240996e-01 1.19189709e-01 4.57256764e-01
3.96888971e-01 -1.39916039e+00 -6.62429094e-01 -4.48744111e-02
5.93411066e-02 -6.75241232e-01 -8.40805829e-01 9.03586298e-02
-7.84928262e-01 -1.00632071e+00 -7.55528212e-01 -6.78380728e-01
8.49953532e-01 1.39685974e-01 1.27733362e+00 5.10317981e-01
-2.73908257e-01 1.45569950e-01 -7.51738966e-01 -4.77504253e-01
-3.93135041e-01 3.13150495e-01 -6.32799983e-01 -1.76230326e-01
-4.00291122e-02 -5.77883780e-01 -9.04089451e-01 7.17386842e-01
-1.01724410e+00 4.73312825e-01 5.84410965e-01 6.31946683e-01
5.40230691e-01 -3.25633377e-01 2.77744174e-01 -1.59104478e+00
8.52969527e-01 -1.61565334e-01 -6.00469351e-01 8.15068111e-02
-5.71882963e-01 -1.90209895e-01 5.76632082e-01 -5.82949579e-01
-8.67581785e-01 2.88412452e-01 -4.56940204e-01 -2.36751035e-01
-4.85988110e-01 3.98213744e-01 -3.51739079e-01 1.01625379e-02
7.42143691e-01 -1.65218138e-03 -4.63263690e-01 -7.40735173e-01
6.29997015e-01 5.10803282e-01 4.71867919e-01 -7.61324644e-01
8.52013886e-01 1.95129603e-01 -7.73935080e-01 -2.22473547e-01
-5.65664589e-01 -4.00639415e-01 -6.08291626e-01 -1.58762887e-01
6.59814835e-01 -3.57063740e-01 -6.33253753e-01 3.77072930e-01
-1.11152697e+00 -8.37331891e-01 -3.04909170e-01 -2.38743827e-01
-2.99634874e-01 2.71185488e-01 -6.21988595e-01 -6.66658998e-01
-1.79734081e-01 -1.39594448e+00 1.30430651e+00 4.15418804e-01
-3.90285313e-01 -7.83413589e-01 -2.27029264e-01 2.66995609e-01
2.99972385e-01 3.67348224e-01 6.28654242e-01 -7.63603091e-01
-9.65226412e-01 -2.48609632e-01 -1.59478366e-01 4.75047268e-02
3.80406454e-02 3.39358568e-01 -7.64315784e-01 -1.24391593e-01
-4.39851016e-01 -1.67427734e-01 9.60427999e-01 2.35353097e-01
1.80990446e+00 -3.21935147e-01 -7.16965079e-01 8.03123236e-01
1.03481841e+00 3.66455585e-01 7.17242956e-01 4.33823049e-01
6.27821684e-01 3.39286149e-01 6.61972404e-01 4.82095897e-01
3.76390517e-01 9.14929211e-01 5.14278293e-01 -4.43035752e-01
-5.76578714e-02 -2.51232296e-01 -3.36566009e-02 4.31460470e-01
-6.69718161e-02 -4.79795396e-01 -9.05025482e-01 3.88766944e-01
-1.79981041e+00 -6.86313629e-01 -2.40442321e-01 2.13006258e+00
1.10568702e+00 4.32170033e-01 2.72027314e-01 1.01776809e-01
4.95197475e-01 9.93662253e-02 -4.37291920e-01 -2.78123021e-01
1.59453556e-01 3.75500858e-01 2.98830658e-01 3.99802744e-01
-1.18250036e+00 1.29259419e+00 6.50025797e+00 1.11357927e+00
-9.82692599e-01 -4.56325598e-02 7.34613776e-01 1.51048571e-01
-5.74788153e-01 1.64619699e-01 -1.02072310e+00 6.24517202e-01
1.93149790e-01 2.59103060e-01 3.63256097e-01 8.39639008e-01
4.90945131e-01 -3.39951217e-01 -1.22582245e+00 6.81423903e-01
-2.28387848e-01 -1.21078074e+00 2.18075654e-03 -3.35896790e-01
6.05700254e-01 -3.49343300e-01 -1.12885348e-02 3.26800287e-01
7.05912352e-01 -9.88602102e-01 7.94392228e-01 3.80319118e-01
7.40279675e-01 -4.02395010e-01 2.21264675e-01 3.72945577e-01
-9.65362310e-01 1.85690239e-01 2.40961723e-02 9.76106003e-02
3.90286297e-01 5.30642152e-01 -1.11393774e+00 4.32707965e-01
6.53228879e-01 3.96089196e-01 -8.48112047e-01 1.71708262e+00
-6.29617035e-01 9.67488348e-01 -1.65081680e-01 -3.09176683e-01
-5.12255728e-02 -2.12590411e-01 6.48074925e-01 1.53016627e+00
7.76689053e-02 -1.00814827e-01 5.87894678e-01 1.11098266e+00
6.45104721e-02 2.44726658e-01 -2.08236620e-01 1.78090736e-01
6.04457855e-01 1.55061781e+00 -1.16937375e+00 -3.50995600e-01
-1.39654115e-01 1.16667616e+00 2.12259084e-01 5.72266877e-01
-1.07231283e+00 -6.32457376e-01 2.46798426e-01 3.22511345e-01
4.59358037e-01 -3.34965020e-01 -5.89935184e-01 -1.05535913e+00
-9.88376439e-02 -1.04559088e+00 2.66306013e-01 -9.98995185e-01
-1.01151919e+00 7.33156800e-01 2.17389852e-01 -1.00737441e+00
-5.92439696e-02 -4.29693907e-01 -9.37604308e-01 8.39045525e-01
-1.02859569e+00 -9.99695182e-01 -5.19410074e-01 2.26119474e-01
9.37882245e-01 3.85516822e-01 5.17162621e-01 1.78581700e-01
-6.46030664e-01 6.63240373e-01 -2.61736035e-01 2.28096321e-01
6.87641203e-01 -1.72132838e+00 7.65402913e-01 7.34358013e-01
3.21916848e-01 7.61993170e-01 9.07349408e-01 -6.50290489e-01
-8.67370963e-01 -8.77945721e-01 1.91885114e-01 -5.89388072e-01
5.66500425e-01 -8.87122095e-01 -9.11282897e-01 6.66203856e-01
2.71408677e-01 -2.68259913e-01 3.93628180e-01 2.27875069e-01
-7.11467341e-02 5.67110032e-02 -8.10749948e-01 1.05960011e+00
1.16947925e+00 7.30859414e-02 -3.07907552e-01 4.47520107e-01
9.51644421e-01 -8.85754824e-01 -3.92350614e-01 3.43254447e-01
4.51311588e-01 -1.26472366e+00 9.09634054e-01 -2.10010514e-01
3.86723697e-01 -5.00498950e-01 5.20219624e-01 -1.53961718e+00
-4.51982617e-02 -1.00777256e+00 1.56100929e-01 1.46610343e+00
8.80991757e-01 -5.70028007e-01 9.57073271e-01 5.45465231e-01
-4.33482826e-01 -1.02494240e+00 -2.89356291e-01 -6.85806930e-01
-2.30791405e-01 -4.22206193e-01 7.66874254e-01 6.25931025e-01
-1.73987046e-01 1.55570298e-01 -1.43490627e-01 -8.04911107e-02
1.90129340e-01 3.26341629e-01 1.13810277e+00 -1.03421974e+00
-6.66861534e-01 -7.97275782e-01 1.60686791e-01 -1.66873610e+00
-2.94963509e-01 -7.19838619e-01 3.49127114e-01 -1.60554135e+00
1.39179930e-01 -1.00350988e+00 1.23765416e-01 7.03965545e-01
-5.78320086e-01 2.13260874e-01 3.02103460e-01 8.88498724e-02
-6.81593537e-01 2.67520219e-01 1.63771152e+00 -4.02950943e-02
-7.01499104e-01 4.31355536e-01 -7.49629557e-01 7.42787600e-01
6.92357481e-01 -1.98673278e-01 -3.30645651e-01 -3.15783381e-01
2.30907295e-02 -4.56103571e-02 3.57789099e-01 -7.50170708e-01
-2.98249684e-02 -3.38147491e-01 8.46916288e-02 -7.37663329e-01
2.18723282e-01 -4.73566502e-01 6.99931830e-02 1.82240345e-02
-4.23478156e-01 -6.03074804e-02 6.78877532e-02 1.96320206e-01
7.82400777e-04 -5.27968049e-01 4.78230000e-01 -2.36840844e-01
-4.73557085e-01 3.69657636e-01 -1.69203281e-01 -7.06005543e-02
9.35738206e-01 -4.50602949e-01 -2.26001024e-01 -3.02646548e-01
-8.58446240e-01 4.75710630e-01 7.29316771e-01 2.59353638e-01
5.69489837e-01 -9.24170196e-01 -4.14997220e-01 5.49217798e-02
2.91371737e-02 3.96392554e-01 -2.88141787e-01 7.00175226e-01
-6.01888359e-01 -6.56883642e-02 2.36283973e-01 -5.80499768e-01
-1.34493804e+00 4.71069694e-01 2.06057921e-01 -3.61860096e-01
-6.42456472e-01 1.21866786e+00 5.55992246e-01 -5.00985682e-01
3.74975592e-01 -5.83478332e-01 -2.15195268e-01 -1.47786573e-01
4.38761234e-01 1.19197458e-01 -7.84753188e-02 1.02933750e-01
-1.38136864e-01 1.96109578e-01 -1.33735776e-01 -2.07345143e-01
1.09153771e+00 -7.16023892e-02 1.08577058e-01 3.56046855e-01
3.69878173e-01 3.78119409e-01 -1.56434488e+00 2.31759474e-01
7.75245428e-02 -5.58339000e-01 -5.90781033e-01 -1.25276828e+00
-9.07845914e-01 6.82479799e-01 1.78539008e-01 5.70177913e-01
1.20681524e+00 1.92537293e-01 6.32649601e-01 3.02413553e-02
4.49185371e-01 -8.09761405e-01 2.01759964e-01 3.58116537e-01
1.07855737e+00 -9.91812050e-01 -2.07587883e-01 -9.27394629e-01
-5.69502652e-01 9.23811257e-01 9.31076407e-01 -3.10348254e-03
4.07532990e-01 4.33223665e-01 1.89250365e-01 -9.20288861e-02
-6.52538836e-01 -3.35660458e-01 4.16156530e-01 5.99602044e-01
5.66290915e-01 6.13287911e-02 -5.49204409e-01 6.78927362e-01
-5.10450244e-01 -1.75039068e-01 4.13501859e-01 7.33171046e-01
-5.97909749e-01 -1.48244607e+00 -4.92905915e-01 4.74931031e-01
-4.07894880e-01 -3.35882723e-01 -5.28816164e-01 7.80597091e-01
1.85477197e-01 8.62530112e-01 -1.98262393e-01 -3.91044259e-01
3.43512118e-01 -1.44656703e-01 2.64144242e-01 -9.76735055e-01
-7.84685254e-01 2.65859038e-01 1.27005100e-01 -5.38065553e-01
-6.60746470e-02 -5.81671596e-01 -1.17157626e+00 1.54298529e-01
-5.95845520e-01 2.56589085e-01 5.04759669e-01 7.69813061e-01
2.06116587e-01 5.31413853e-01 4.12607759e-01 -1.05781400e+00
-1.93195865e-01 -9.20996010e-01 -2.30379820e-01 5.66891313e-01
-9.14300159e-02 -6.52516544e-01 -1.09379485e-01 2.37475768e-01] | [9.466835021972656, -0.09707864373922348] |
277e241f-90d7-4a77-ad02-ebba764a20f5 | dynamic-spatiotemporal-graph-convolutional | 2109.08357 | null | https://arxiv.org/abs/2109.08357v1 | https://arxiv.org/pdf/2109.08357v1.pdf | Dynamic Spatiotemporal Graph Convolutional Neural Networks for Traffic Data Imputation with Complex Missing Patterns | Missing data is an inevitable and ubiquitous problem for traffic data collection in intelligent transportation systems. Despite extensive research regarding traffic data imputation, there still exist two limitations to be addressed: first, existing approaches fail to capture the complex spatiotemporal dependencies in traffic data, especially the dynamic spatial dependencies evolving with time; second, prior studies mainly focus on randomly missing patterns while other more complex missing scenarios are less discussed. To fill these research gaps, we propose a novel deep learning framework called Dynamic Spatiotemporal Graph Convolutional Neural Networks (DSTGCN) to impute missing traffic data. The model combines the recurrent architecture with graph-based convolutions to model the spatiotemporal dependencies. Moreover, we introduce a graph structure estimation technique to model the dynamic spatial dependencies from real-time traffic information and road network structure. Extensive experiments based on two public traffic speed datasets are conducted to compare our proposed model with state-of-the-art deep learning approaches in four types of missing patterns. The results show that our proposed model outperforms existing deep learning models in all kinds of missing scenarios and the graph structure estimation technique contributes to the model performance. We further compare our proposed model with a tensor factorization model and find distinct behaviors across different model families under different training schemes and data availability. | ['Lijun Sun', 'Zhan Zhao', 'Yuebing Liang'] | 2021-09-17 | null | null | null | null | ['traffic-data-imputation'] | ['time-series'] | [-1.47945330e-01 -4.82645035e-01 -6.33255422e-01 -5.33562422e-01
-1.04915507e-01 -9.62573977e-05 2.96268314e-01 -4.37186748e-01
1.12056553e-01 8.08146060e-01 5.78285456e-01 -8.50834191e-01
-5.01231611e-01 -9.29208755e-01 -9.69783425e-01 -3.75929683e-01
-5.64173013e-02 5.06492138e-01 2.10262924e-01 -6.49257183e-01
-5.71019724e-02 5.64075589e-01 -1.57887888e+00 3.71946663e-01
1.14091969e+00 8.09179664e-01 -1.33002475e-01 4.55379099e-01
-4.39765841e-01 1.28776395e+00 -2.79332817e-01 -5.71844399e-01
2.77798474e-01 4.52219248e-02 -4.98818845e-01 -1.64050534e-02
4.96871799e-01 -4.84923214e-01 -1.32256019e+00 6.50869906e-01
3.19137692e-01 2.79255122e-01 2.43623152e-01 -1.90364206e+00
-9.34440911e-01 5.79262912e-01 -5.79689324e-01 6.12018645e-01
-1.73764259e-01 3.21804821e-01 5.88132560e-01 -5.08704424e-01
2.28390813e-01 1.29870391e+00 1.02171624e+00 3.22799087e-01
-9.41330492e-01 -9.46181774e-01 6.83760107e-01 7.65188813e-01
-1.23462880e+00 -5.59862554e-01 1.19608843e+00 -4.20754433e-01
1.05698931e+00 1.00175321e-01 3.75576168e-01 1.43699062e+00
4.16337550e-01 8.88808489e-01 6.71465278e-01 2.98403740e-01
-2.38948122e-01 -4.08226073e-01 5.48638225e-01 5.15891731e-01
3.52769673e-01 1.33545756e-01 -2.81524837e-01 2.88929157e-02
4.89486516e-01 5.54360569e-01 3.55252475e-01 -6.25039488e-02
-1.02393961e+00 4.94913340e-01 4.53699946e-01 4.35361313e-03
-4.25708801e-01 5.51195920e-01 5.32904923e-01 4.42617059e-01
7.45901883e-01 -6.16947055e-01 -3.00953865e-01 -1.82483748e-01
-6.90869510e-01 2.87428647e-01 4.86105174e-01 1.21808338e+00
8.37333322e-01 6.09012008e-01 -2.46075451e-01 7.90646195e-01
1.05279855e-01 5.82428873e-01 7.13922754e-02 -6.86916053e-01
1.32334554e+00 8.15501869e-01 -9.87823308e-03 -1.52418315e+00
-7.17660010e-01 -5.12178123e-01 -1.36568868e+00 -4.23743010e-01
4.25314516e-01 -4.36421305e-01 -1.00019908e+00 1.75190425e+00
1.63949803e-01 9.62476254e-01 -2.49253869e-01 9.02493596e-01
1.01516640e+00 5.32996893e-01 1.63967624e-01 2.13285416e-01
8.12260211e-01 -1.05935168e+00 -1.15576851e+00 -1.20757267e-01
7.44112611e-01 -3.90177965e-01 6.82239652e-01 -1.04226591e-02
-6.89445257e-01 -8.84007096e-01 -5.22208273e-01 -1.07301705e-01
-7.24551380e-01 1.88998178e-01 1.13974142e+00 6.38295889e-01
-9.49375331e-01 2.35507458e-01 -8.63374352e-01 -2.84852475e-01
5.62075853e-01 4.37975109e-01 -2.83079326e-01 -3.49943221e-01
-1.51528049e+00 6.51073754e-01 -4.97870296e-02 8.86348009e-01
-7.06842124e-01 -8.56261075e-01 -7.83231676e-01 -6.22460507e-02
4.88143474e-01 -8.85522127e-01 6.74454391e-01 -5.27189016e-01
-8.43996763e-01 6.20408580e-02 -6.18678033e-01 -5.99523067e-01
5.26625395e-01 -4.98981960e-02 -1.19985998e+00 -5.88873386e-01
1.97872087e-01 1.20382443e-01 5.54347754e-01 -9.88342762e-01
-5.16653240e-01 -3.53908718e-01 9.32886377e-02 -1.74663529e-01
-1.60473809e-01 -1.95360050e-01 -4.23918158e-01 -7.49569893e-01
-8.24254379e-02 -8.50740433e-01 -3.52238506e-01 -3.96294534e-01
-6.11855984e-01 -4.16651726e-01 1.31824744e+00 -7.67108977e-01
1.58516335e+00 -1.72137105e+00 -1.98470116e-01 5.22876754e-02
5.40377975e-01 2.62611926e-01 -3.52776736e-01 6.87269449e-01
-2.95915127e-01 4.30754609e-02 -6.56612739e-02 -4.89065319e-01
7.27288797e-02 7.20359623e-01 -5.21048486e-01 2.75710583e-01
1.88415512e-01 1.30793893e+00 -6.84089541e-01 -2.11470813e-01
5.51637471e-01 6.43151641e-01 -3.50332767e-01 -4.58630919e-02
-6.65657893e-02 6.27546310e-01 -5.05297363e-01 6.68175697e-01
1.25237346e+00 -1.59114420e-01 -1.43512130e-01 -3.83868307e-01
-5.88275790e-02 2.92597115e-01 -1.14454889e+00 1.50270879e+00
-3.83792907e-01 5.53921878e-01 -1.21891081e-01 -1.46800113e+00
8.65580559e-01 5.95141165e-02 8.25303078e-01 -1.22345352e+00
-2.86085960e-02 -2.45262161e-01 4.10444066e-02 -9.48599696e-01
6.06527448e-01 2.52445251e-01 -5.09696947e-05 4.12481964e-01
-1.59619808e-01 9.98001397e-01 2.39676565e-01 3.21141034e-01
1.27243280e+00 -2.46848371e-02 -7.67792761e-01 1.92365617e-01
3.66293252e-01 -2.77228449e-02 8.06760073e-01 7.64413416e-01
-4.04700458e-01 3.54545206e-01 6.24670804e-01 -1.11424446e+00
-7.86748409e-01 -9.00509298e-01 1.49892241e-01 1.14546788e+00
2.02068567e-01 -1.81606412e-01 -2.53913373e-01 -8.32547843e-01
2.94376343e-01 5.98213553e-01 -7.81438172e-01 -1.92717493e-01
-8.99082839e-01 -1.07788789e+00 6.71481609e-01 6.57589376e-01
7.10846841e-01 -8.28353524e-01 4.80508357e-01 2.86177605e-01
-6.65093541e-01 -1.53772318e+00 -3.10794413e-01 -4.89897758e-01
-7.98123598e-01 -1.12586784e+00 -2.46606562e-02 -4.35752720e-01
4.00062323e-01 8.31498861e-01 1.21176302e+00 2.48496234e-01
5.83414845e-02 1.97867796e-01 -2.56596535e-01 -1.64116338e-01
-1.00091070e-01 3.62251937e-01 -1.03569799e-03 5.33987880e-01
8.04254353e-01 -9.77376759e-01 -6.48607671e-01 5.19651234e-01
-7.36368418e-01 2.48716086e-01 5.54356515e-01 6.04076743e-01
3.19892228e-01 4.18308794e-01 8.85725319e-01 -9.05170143e-01
6.83554232e-01 -1.35524309e+00 -2.87359059e-01 1.24174386e-01
-3.74759227e-01 -7.83093274e-02 7.76536584e-01 -2.83930957e-01
-1.12418878e+00 -1.04653582e-01 -2.74672657e-01 -6.98322654e-01
-3.67027879e-01 5.63817918e-01 -3.24629933e-01 1.02329828e-01
1.06255874e-01 3.04860145e-01 -6.49303547e-04 -5.25025725e-01
3.78020436e-01 5.09082317e-01 4.56992149e-01 -5.94280303e-01
7.88930655e-01 7.11911619e-01 1.41312584e-01 -6.90266967e-01
-7.40696251e-01 -4.04480457e-01 -8.69971395e-01 -4.12960649e-01
5.16285539e-01 -9.56340671e-01 -8.75284016e-01 5.54701447e-01
-1.23854184e+00 -2.41525725e-01 6.80226367e-03 4.35833216e-01
-3.26595634e-01 4.81608719e-01 -5.23734450e-01 -8.65164220e-01
-9.68273729e-02 -1.07265997e+00 9.48402703e-01 -2.23796576e-01
4.84039307e-01 -1.23453343e+00 -8.02741721e-02 7.45695770e-01
7.90244162e-01 4.77081269e-01 8.60412240e-01 -2.88957179e-01
-9.02821183e-01 -2.21842602e-01 -5.73712230e-01 4.62111384e-02
3.60590130e-01 -8.04095995e-03 -5.95834255e-01 1.05869852e-01
-5.08188069e-01 2.88368553e-01 1.18055165e+00 5.85251391e-01
1.48473155e+00 -5.56591630e-01 -5.37416279e-01 7.32602894e-01
1.04090822e+00 -1.49889991e-01 9.27165270e-01 -1.10788278e-01
1.37724483e+00 7.18841791e-01 1.54616401e-01 2.11595699e-01
1.17454505e+00 6.00007594e-01 7.32554078e-01 -2.64314115e-01
-4.57162529e-01 -5.88909566e-01 2.29687408e-01 1.05299330e+00
-1.25738248e-01 -5.59331954e-01 -9.26457226e-01 8.96536648e-01
-2.43480587e+00 -1.37651515e+00 -9.57094491e-01 1.71583271e+00
-6.75661787e-02 1.94920003e-01 5.56328297e-01 3.49656790e-02
8.10338318e-01 3.05667818e-01 -7.80005574e-01 -2.39299074e-01
-4.37409431e-01 -1.60719082e-01 7.83447266e-01 3.89828324e-01
-1.08726239e+00 1.00965643e+00 6.35417509e+00 6.79099202e-01
-9.71233189e-01 2.00897217e-01 5.12339354e-01 -2.65039299e-02
-3.92285943e-01 -4.39033546e-02 -7.35343635e-01 9.37675297e-01
1.35514772e+00 1.68187007e-01 5.58293521e-01 3.92461807e-01
7.16363013e-01 5.31161547e-01 -6.62466943e-01 9.70936894e-01
-1.62842140e-01 -1.66998982e+00 4.31339294e-01 1.63549915e-01
7.21571207e-01 6.27523363e-01 1.79274470e-01 5.49076557e-01
7.32221663e-01 -1.19126236e+00 1.25828758e-01 1.12465143e+00
3.62337291e-01 -9.63299930e-01 7.45802402e-01 3.86922479e-01
-1.46828341e+00 -3.38361353e-01 -4.99935746e-01 -4.70619410e-01
5.21910906e-01 6.85549736e-01 -2.82464176e-01 9.47522044e-01
6.74536407e-01 1.31349099e+00 -8.09375465e-01 9.40450191e-01
2.99193591e-01 1.05616927e+00 -2.31185541e-01 4.02336895e-01
3.47825259e-01 -4.60921377e-01 4.07326877e-01 1.21706915e+00
2.17719972e-01 -2.63542086e-01 1.25566021e-01 9.14400399e-01
-1.28570005e-01 -2.79381186e-01 -1.09826112e+00 2.78531402e-01
4.01911914e-01 9.65829670e-01 -8.28199834e-02 -1.66313797e-01
-8.41751158e-01 5.29714704e-01 3.77121121e-01 9.08231914e-01
-1.22141051e+00 -1.42758876e-01 9.85951960e-01 4.57212865e-01
2.76559621e-01 -6.50312066e-01 -3.04554135e-01 -1.32441294e+00
2.69206464e-01 -4.51827437e-01 6.09058321e-01 -6.58299744e-01
-1.74343789e+00 4.35029477e-01 3.45499292e-02 -1.30682635e+00
3.52061875e-02 -3.24307948e-01 -8.07016015e-01 6.95114493e-01
-1.82259655e+00 -1.89282954e+00 -5.19008160e-01 1.10468364e+00
4.32101667e-01 -3.82154703e-01 1.45761609e-01 9.87601817e-01
-1.14040470e+00 5.78892112e-01 1.61953703e-01 4.75866675e-01
2.69783974e-01 -6.84862256e-01 9.61287320e-01 1.04400933e+00
-4.47575338e-02 5.38188875e-01 3.48835260e-01 -8.60276937e-01
-1.70882678e+00 -1.81493819e+00 9.70864713e-01 -7.58568108e-01
8.96151781e-01 -4.99489516e-01 -9.43627298e-01 1.14172888e+00
9.87291858e-02 5.06839216e-01 5.20565987e-01 3.35836917e-01
-4.70998079e-01 -6.76395595e-01 -7.46771038e-01 4.76206005e-01
1.56015599e+00 -3.33705693e-01 -1.67576805e-01 4.10935402e-01
9.33596015e-01 -2.39853278e-01 -6.75374448e-01 5.04604459e-01
4.63170707e-01 -7.74701238e-01 9.22387123e-01 -1.18767929e+00
7.86528513e-02 -4.90232140e-01 -9.57345143e-02 -1.04086268e+00
-5.97893596e-01 -5.24658859e-01 -5.49950957e-01 1.26077223e+00
3.32033813e-01 -8.10587883e-01 1.02087092e+00 7.79745102e-01
-3.64705563e-01 -4.35764015e-01 -1.25238132e+00 -6.23742402e-01
2.03995667e-02 -1.06956089e+00 1.30746329e+00 1.05923307e+00
-6.02101564e-01 4.87935871e-01 -1.15101647e+00 2.61642426e-01
7.31646061e-01 6.37708455e-02 1.21443141e+00 -1.30444932e+00
3.12130570e-01 -2.78430462e-01 -6.36718512e-01 -1.24989104e+00
5.32202959e-01 -8.92750859e-01 -6.98745608e-01 -1.71228790e+00
-2.10436527e-02 -6.32904410e-01 -5.94343662e-01 5.42142391e-01
5.84389195e-02 -6.55532777e-02 -2.85051912e-01 7.44282976e-02
-9.04939830e-01 8.39231968e-01 1.19105387e+00 -5.03148854e-01
-5.37356324e-02 3.53739321e-01 -8.31838191e-01 2.19815522e-01
8.19662213e-01 -2.90941954e-01 -7.34828711e-01 -1.03956687e+00
2.80249178e-01 2.31421143e-01 6.88020945e-01 -9.19572413e-01
5.25541902e-01 -4.59753275e-01 1.60865411e-01 -1.28249395e+00
1.30245149e-01 -1.14745152e+00 3.82536501e-01 -5.40648289e-02
2.64855530e-02 5.26954830e-01 3.68517160e-01 1.10264516e+00
-6.25823289e-02 8.34774852e-01 -8.64586979e-02 2.63549596e-01
-8.99967790e-01 9.61379647e-01 -3.82245272e-01 -1.06707059e-01
8.05721998e-01 -4.20835793e-01 -6.83815658e-01 -6.27180457e-01
-7.12274849e-01 6.44160271e-01 -2.01888919e-01 1.08743620e+00
6.55202210e-01 -1.79837573e+00 -8.51673841e-01 4.51197773e-01
1.17781766e-01 -2.00866878e-01 9.27535236e-01 1.09594417e+00
-4.82036546e-02 5.30534863e-01 -1.74638972e-01 -6.23165727e-01
-5.46725750e-01 7.93033481e-01 4.05783415e-01 -2.10416630e-01
-7.46793628e-01 5.52937090e-02 -1.44653499e-01 -9.95564818e-01
9.96009856e-02 -4.94232625e-01 -2.17658743e-01 -2.60652035e-01
2.20545217e-01 7.91166961e-01 3.35799992e-01 -1.13957882e+00
-3.01771581e-01 3.39125425e-01 4.54037264e-02 7.11294353e-01
1.41411006e+00 -5.48991442e-01 -3.25020924e-02 3.52590412e-01
1.07366693e+00 -3.24143291e-01 -1.19333541e+00 -4.17590559e-01
-1.88382506e-01 -5.13748348e-01 -7.03749806e-02 -6.26877427e-01
-1.72486961e+00 1.01255322e+00 4.52129245e-01 2.03503132e-01
9.44830000e-01 -4.87633348e-01 1.39651251e+00 2.92695582e-01
2.33958602e-01 -9.16639149e-01 -3.33263427e-01 6.94029510e-01
6.24232411e-01 -1.45603836e+00 -3.45561475e-01 -3.92542511e-01
-5.55476308e-01 8.17489088e-01 9.29269314e-01 -2.08960101e-01
1.12994862e+00 6.13328665e-02 -2.09074602e-01 -3.49759609e-01
-1.07259631e+00 -5.01416147e-01 2.78075486e-01 8.83055687e-01
1.72813073e-01 1.15735605e-01 1.43274730e-02 6.12886727e-01
-1.19766213e-01 2.96248376e-01 4.04940546e-01 3.86894554e-01
6.86324090e-02 -1.03190601e+00 -1.43954474e-02 6.81691051e-01
-1.61275685e-01 5.23258336e-02 -1.32045358e-01 8.45314860e-01
2.46173188e-01 1.39949954e+00 2.50798464e-01 -8.17522407e-01
5.87184012e-01 -2.51220196e-01 -5.23045808e-02 1.66782625e-02
-2.40489513e-01 -5.04155695e-01 1.99097231e-01 -8.35231781e-01
-4.49075371e-01 -3.60578746e-01 -7.57320106e-01 -1.08752692e+00
2.47656099e-05 8.70068520e-02 3.85894209e-01 1.02644944e+00
9.56793547e-01 8.17888498e-01 5.86383402e-01 -4.47707385e-01
2.07967103e-01 -8.58178735e-01 -3.96268189e-01 6.84480786e-01
6.95583403e-01 -1.00867593e+00 -1.21389873e-01 -2.74759144e-01] | [6.512206554412842, 2.0647900104522705] |
ae76514d-f7a9-475d-a1ea-f6950bd360f8 | glass-segmentation-with-rgb-thermal-image | 2204.05453 | null | https://arxiv.org/abs/2204.05453v4 | https://arxiv.org/pdf/2204.05453v4.pdf | Glass Segmentation with RGB-Thermal Image Pairs | This paper proposes a new glass segmentation method utilizing paired RGB and thermal images. Due to the large difference between the transmission property of visible light and that of the thermal energy through the glass where most glass is transparent to the visible light but opaque to thermal energy, glass regions of a scene are made more distinguishable with a pair of RGB and thermal images than solely with an RGB image. To exploit such a unique property, we propose a neural network architecture that effectively combines an RGB-thermal image pair with a new multi-modal fusion module based on attention, and integrate CNN and transformer to extract local features and non-local dependencies, respectively. As well, we have collected a new dataset containing 5551 RGB-thermal image pairs with ground-truth segmentation annotations. The qualitative and quantitative evaluations demonstrate the effectiveness of the proposed approach on fusing RGB and thermal data for glass segmentation. Our code and data are available at https://github.com/Dong-Huo/RGB-T-Glass-Segmentation. | ['Yee-Hong Yang', 'Yiming Qian', 'Jian Wang', 'Dong Huo'] | 2022-04-12 | null | null | null | null | ['thermal-image-segmentation'] | ['computer-vision'] | [ 2.46830449e-01 1.34234846e-01 1.33829147e-01 -5.34823477e-01
-7.00290859e-01 -5.51395535e-01 7.58216381e-02 -4.11834776e-01
-2.03435421e-01 1.28334031e-01 2.31351843e-03 -1.09989077e-01
2.81355351e-01 -9.54699934e-01 -6.41707182e-01 -1.16592169e+00
6.16276562e-01 -1.08378232e-01 2.52002388e-01 -2.01159455e-02
-5.12524098e-02 1.02754824e-01 -1.61880541e+00 4.67987210e-01
7.56717980e-01 1.41382325e+00 2.69876629e-01 6.59896314e-01
-6.32741302e-02 4.48066741e-01 -1.61748782e-01 -1.01694025e-01
5.91253042e-01 -5.00982106e-01 -8.89046788e-01 3.17725360e-01
5.98076522e-01 -3.80236447e-01 -3.85622650e-01 1.02005267e+00
2.84064651e-01 1.63972914e-01 2.64508098e-01 -1.16317618e+00
-1.09358239e+00 3.94428134e-01 -6.09028578e-01 -3.45643908e-02
5.16974330e-01 5.98159611e-01 9.10708666e-01 -4.97817129e-01
9.15696323e-02 9.51058447e-01 5.62043369e-01 4.79335874e-01
-7.12995291e-01 -4.55414772e-01 1.75923601e-01 -4.74274419e-02
-1.10826993e+00 -2.78309494e-01 9.35719013e-01 -1.18890494e-01
8.84268999e-01 4.46806967e-01 1.13191569e+00 8.43191743e-01
2.16268882e-01 8.25624466e-01 1.37322164e+00 -3.93790573e-01
-6.36881590e-02 1.40681982e-01 5.82069121e-02 9.55987573e-01
-7.06051216e-02 3.19741160e-01 -3.36763263e-01 4.70307201e-01
5.77860177e-01 4.49686140e-01 -3.98039669e-01 2.07956344e-01
-9.04459357e-01 2.25483671e-01 1.09155142e+00 5.42844117e-01
-1.09723896e-01 3.79589587e-01 -8.85488838e-02 -3.04236431e-02
4.62248534e-01 1.50201190e-02 -1.89479068e-01 2.17467740e-01
-8.19939733e-01 -5.02038002e-01 3.51026237e-01 7.67538726e-01
1.27404523e+00 -1.40477106e-01 -2.53790640e-03 4.69635159e-01
8.79555464e-01 9.21390057e-01 4.99122649e-01 -9.22655940e-01
2.94154048e-01 9.00793135e-01 -2.31431529e-01 -6.08506322e-01
-4.02046174e-01 5.23210391e-02 -5.83471835e-01 1.11076981e-01
3.34605068e-01 4.54191715e-02 -1.32824266e+00 1.34580791e+00
5.25597751e-01 1.12734333e-01 -1.40139654e-01 1.07903564e+00
1.20602727e+00 4.70820516e-01 -1.45508587e-01 4.55875322e-02
1.37324786e+00 -1.18932140e+00 -6.43519640e-01 -4.64634806e-01
2.50903606e-01 -7.56396949e-01 1.32394481e+00 1.05795644e-01
-9.94678855e-01 -6.19897366e-01 -1.02183187e+00 -5.00306129e-01
-6.95984006e-01 1.20792404e-01 7.85480499e-01 8.45108569e-01
-1.23455858e+00 3.19546819e-01 -1.13695598e+00 -4.95093793e-01
4.15167600e-01 3.74055892e-01 -2.23704904e-01 -1.66669935e-01
-9.30008173e-01 4.51417655e-01 3.94077361e-01 8.39539587e-01
-4.82801795e-01 -3.20933610e-01 -6.91205382e-01 -2.75874436e-01
3.41778487e-01 -6.82335258e-01 9.50639546e-01 -9.58112776e-01
-1.74147677e+00 8.65095317e-01 -1.80012107e-01 3.39716643e-01
3.60365778e-01 -3.17114979e-01 -2.07956493e-01 4.05622184e-01
-6.57533109e-02 4.14199352e-01 7.38015234e-01 -1.45532966e+00
-5.36931098e-01 -6.46144509e-01 1.17021494e-01 2.40094066e-01
-2.27924049e-01 -2.03731224e-01 -8.08033645e-01 2.34142300e-02
5.50241172e-01 -9.63195145e-01 9.97081026e-03 -3.86925370e-01
-9.40602005e-01 -3.06298099e-02 9.45387661e-01 -6.15379393e-01
8.00709665e-01 -2.13649631e+00 -1.98384300e-01 1.46575749e-01
3.89143735e-01 5.21508083e-02 1.61702186e-01 2.47672498e-01
-2.78495401e-02 2.37530082e-01 -4.59357917e-01 -6.80828929e-01
-6.12971894e-02 7.56675377e-02 2.57296655e-02 5.42724550e-01
-1.19996771e-01 1.22165561e+00 -7.95083046e-01 -6.41057611e-01
6.75945580e-01 5.35249650e-01 9.84789133e-02 3.21724385e-01
-1.33116931e-01 7.00415730e-01 -6.55618072e-01 1.29547775e+00
8.36861968e-01 -2.56004542e-01 -1.21511661e-01 -6.30553663e-01
-9.38747302e-02 -3.96957994e-02 -6.67250633e-01 1.82107019e+00
-5.49719691e-01 4.09876019e-01 9.45826294e-04 -3.93838972e-01
6.88731253e-01 1.98321179e-01 6.23192132e-01 -8.87775540e-01
4.66857761e-01 1.59941629e-01 -4.74492371e-01 -6.67626202e-01
5.79404294e-01 -8.53068307e-02 -3.01661193e-02 6.63667679e-01
-8.39029849e-02 -6.73173428e-01 -7.09038824e-02 -1.80982351e-01
9.18012500e-01 4.69559640e-01 -2.98310518e-01 3.68530601e-01
2.70414978e-01 -2.59636492e-01 6.28831208e-01 5.72932541e-01
-4.99500096e-01 8.95197749e-01 -7.73185343e-02 -4.74346936e-01
-7.03795791e-01 -1.06869483e+00 1.66357517e-01 8.83283496e-01
6.67114377e-01 -1.53389424e-01 -7.34796166e-01 -5.47629535e-01
-9.26046520e-02 3.60761166e-01 -8.22845876e-01 -2.54411846e-01
-4.19974655e-01 -6.42769277e-01 3.41119319e-01 5.35925150e-01
1.14949143e+00 -1.10902667e+00 -6.61801755e-01 -2.99378157e-01
-3.27691376e-01 -1.30349565e+00 -4.37520623e-01 1.84684783e-01
-7.97304451e-01 -1.17266715e+00 -3.11874121e-01 -4.22067225e-01
5.61288238e-01 3.57817948e-01 1.06125247e+00 2.67179996e-01
-4.38884765e-01 5.60926497e-01 -4.19431180e-01 2.05664542e-02
-2.05233321e-01 -8.46472010e-02 -3.64335299e-01 2.49228805e-01
4.14589494e-01 -4.88710254e-01 -9.77055073e-01 2.27089182e-01
-1.07386565e+00 3.67793173e-01 8.48088920e-01 2.79443443e-01
5.89120448e-01 -3.22660878e-02 -3.19786578e-01 -5.85104585e-01
1.15647197e-01 -1.78561583e-01 -4.22728747e-01 4.98957425e-01
-3.37287724e-01 -3.39142025e-01 2.38348708e-01 1.00654066e-01
-1.37566018e+00 3.90269965e-01 6.32448867e-02 -4.58714187e-01
-5.54119408e-01 1.02969140e-01 -3.51254642e-01 -2.27953285e-01
1.70539230e-01 2.28567153e-01 -3.14925671e-01 -2.01193854e-01
6.50664449e-01 8.11691403e-01 4.71213639e-01 -2.89584875e-01
7.59474397e-01 9.23676848e-01 -1.89985767e-01 -7.14192629e-01
-8.47883999e-01 -5.25893450e-01 -8.20636213e-01 -6.26048446e-01
1.47375226e+00 -7.59122550e-01 -1.10386574e+00 9.09175456e-01
-9.08989906e-01 -5.80238819e-01 -1.32646695e-01 3.98370594e-01
-4.58182126e-01 4.98713791e-01 -7.86267698e-01 -8.94590855e-01
-4.09138292e-01 -1.30971491e+00 1.44830298e+00 7.18602180e-01
3.99760693e-01 -9.33068335e-01 -1.83375226e-03 9.44049537e-01
3.31408679e-01 3.28222215e-01 4.28367674e-01 1.37477756e-01
-1.02301002e+00 -1.58629075e-01 -4.71664876e-01 4.73404437e-01
5.75774312e-01 3.03214848e-01 -1.53231108e+00 -3.82273458e-02
1.64879143e-01 -4.40087736e-01 1.22162127e+00 5.68204224e-01
1.02476180e+00 9.13029537e-02 -2.93439478e-01 9.90188539e-01
1.72720134e+00 2.04617947e-01 6.55343652e-01 2.22976938e-01
1.36195111e+00 3.94230872e-01 4.04898793e-01 -8.00141506e-03
5.59606314e-01 3.53012383e-01 7.53804445e-01 -5.98632336e-01
-1.91983789e-01 9.42017511e-02 3.33253503e-01 8.44543755e-01
-3.20154339e-01 -4.13975775e-01 -8.74748647e-01 4.10703748e-01
-1.70627618e+00 -6.09147608e-01 -2.06033379e-01 1.96710002e+00
5.76247931e-01 -2.16626883e-01 -1.21238969e-01 -8.59412029e-02
6.98627949e-01 1.25166193e-01 -6.74542248e-01 -2.51195669e-01
-2.05950111e-01 -4.32851817e-03 7.91127145e-01 4.21686232e-01
-1.21218610e+00 1.00057817e+00 6.06450653e+00 4.70455945e-01
-1.29137766e+00 1.37855589e-01 8.69305789e-01 -1.70900539e-01
-3.65568101e-01 -3.18361409e-02 -1.98781699e-01 5.94031632e-01
6.93636417e-01 5.70044279e-01 4.49281186e-01 3.68716180e-01
1.27359048e-01 -5.98597765e-01 -9.07253087e-01 9.43416417e-01
-1.11852653e-01 -8.22804391e-01 -4.21814412e-01 1.69103056e-01
7.62675047e-01 2.83406526e-01 4.34095055e-01 -2.85275757e-01
4.16202217e-01 -8.61919701e-01 6.65785611e-01 8.01332772e-01
5.19406199e-01 -2.51780212e-01 7.30651259e-01 -2.47993395e-01
-1.57342899e+00 -1.55838296e-01 -6.08727976e-04 2.72666067e-01
4.56728227e-02 7.85552084e-01 -4.24993306e-01 1.01594234e+00
1.20886123e+00 9.42288339e-01 -6.91999614e-01 6.32088244e-01
-5.19085169e-01 4.31311995e-01 -5.68361700e-01 2.06730783e-01
2.14073554e-01 -7.04139590e-01 2.99713556e-02 8.32266927e-01
3.06377828e-01 4.66901176e-02 1.12150207e-01 1.45039856e+00
9.70698446e-02 -4.17120367e-01 -6.09898925e-01 -1.53717339e-01
2.14813754e-01 1.62008774e+00 -1.16573930e+00 -3.90678912e-01
-5.94738603e-01 1.12904990e+00 1.99909024e-02 6.97389781e-01
-8.81327331e-01 -2.00955287e-01 2.68929452e-01 -3.31274599e-01
2.34260961e-01 -1.83473766e-01 -5.68885386e-01 -1.09219050e+00
2.07266718e-01 -2.22881898e-01 9.89458337e-02 -1.27494287e+00
-1.17599511e+00 8.10047150e-01 -1.68749735e-01 -1.15939856e+00
3.57341796e-01 -5.65559745e-01 -6.31559372e-01 8.48694980e-01
-1.42764747e+00 -1.78286076e+00 -8.06243062e-01 6.72961414e-01
-1.41733354e-02 4.70591784e-01 4.97077376e-01 1.74253806e-01
-8.60761106e-01 2.09493652e-01 4.36741114e-02 1.35286570e-01
5.85878432e-01 -1.44319534e+00 2.02182457e-01 1.03119922e+00
-9.22147259e-02 4.73652542e-01 3.46884131e-01 -5.10667086e-01
-1.56796968e+00 -8.93601298e-01 1.00319579e-01 -6.05289340e-01
3.61559510e-01 -4.04611737e-01 -6.02930665e-01 7.55880356e-01
5.00585973e-01 2.08348379e-01 5.63563526e-01 -2.28706837e-01
-3.09765726e-01 -2.86869377e-01 -1.20979595e+00 2.32671753e-01
7.51594961e-01 -8.46218646e-01 -2.67742097e-01 3.66296172e-01
1.01697183e+00 -5.49000144e-01 -9.41037834e-01 2.03866825e-01
6.50096595e-01 -1.58723438e+00 7.99905896e-01 2.82853454e-01
4.03330088e-01 -4.48251277e-01 -2.63088703e-01 -1.00671792e+00
1.14466563e-01 -3.51262510e-01 3.31435502e-01 1.19396853e+00
3.13723385e-01 -8.46079886e-01 9.20174897e-01 1.06392133e+00
-5.63548863e-01 -6.24370694e-01 -9.49535370e-01 -3.11841041e-01
-2.34770969e-01 -5.97750604e-01 6.69226468e-01 6.82953477e-01
-2.92377472e-01 -1.29769653e-01 -7.63288960e-02 4.47381079e-01
4.97198969e-01 6.43660486e-01 4.20993984e-01 -6.17675602e-01
-2.80562282e-01 -3.58777046e-01 -2.52389312e-01 -7.14947224e-01
-1.48816511e-01 -6.30238771e-01 3.82548720e-01 -1.90557539e+00
2.30987251e-01 -3.05025548e-01 -5.16131461e-01 8.65213633e-01
-2.50028968e-01 6.47551358e-01 2.16090277e-01 1.51501313e-01
-6.73549771e-01 6.30709767e-01 1.54361129e+00 -3.12657058e-01
-3.56617242e-01 -1.75189063e-01 -5.36076784e-01 5.99906564e-01
7.35148370e-01 -2.06198201e-01 -1.51733831e-01 -6.29711688e-01
2.33557582e-01 -1.29952177e-01 3.97260487e-01 -9.76217270e-01
1.09524854e-01 -1.29035667e-01 3.49721164e-01 -5.01593709e-01
6.07172728e-01 -9.70233023e-01 4.13647443e-02 2.81495661e-01
1.96610346e-01 -2.68859833e-01 1.53863326e-01 5.59553802e-01
-2.20323324e-01 2.48006791e-01 5.36676526e-01 -4.91006792e-01
-7.61712790e-01 3.81981462e-01 -7.04883784e-02 -4.48691905e-01
9.58287656e-01 -6.12805545e-01 -6.62710190e-01 -1.77978337e-01
-7.25327194e-01 9.11291912e-02 7.60175169e-01 1.83382124e-01
7.28231609e-01 -1.17732048e+00 5.69178397e-03 2.68033445e-01
9.82243046e-02 2.67725885e-01 6.54267848e-01 1.08390951e+00
-6.00192964e-01 2.66513288e-01 -1.97107747e-01 -8.05595338e-01
-1.21836042e+00 2.56642312e-01 6.69126570e-01 1.05214544e-01
-5.17554581e-01 1.08803260e+00 2.78694808e-01 -4.89055842e-01
-1.97298393e-01 -9.83649075e-01 2.64284462e-01 -3.39051366e-01
1.16616927e-01 1.25668928e-01 1.43032655e-01 -7.86460757e-01
-4.35808510e-01 9.48777437e-01 3.26812774e-01 -1.00920275e-01
1.13454878e+00 -6.64222598e-01 -4.90318924e-01 7.30266988e-01
1.30672371e+00 -2.48448998e-01 -1.28120708e+00 -1.93602011e-01
-5.59166968e-01 -6.81070328e-01 3.34682554e-01 -7.12042093e-01
-1.78480637e+00 9.01843309e-01 9.96666312e-01 5.56228101e-01
1.58239400e+00 2.10143641e-01 1.07492542e+00 3.58321853e-02
7.53376484e-02 -1.00746000e+00 2.27722064e-01 1.44352213e-01
3.58408570e-01 -1.60269392e+00 4.11250331e-02 -5.39052963e-01
-4.67851132e-01 9.87562656e-01 7.32086837e-01 1.65484510e-02
5.66361487e-01 4.88109216e-02 4.65049863e-01 -4.28580821e-01
-2.86262244e-01 -5.74891865e-01 2.65007496e-01 6.17470324e-01
4.49909538e-01 5.21620959e-02 5.13799369e-01 9.77373272e-02
-1.45192206e-01 -1.33422986e-01 3.90229315e-01 9.30943489e-01
-1.71732873e-01 -8.10571909e-01 -4.96684015e-01 2.37519354e-01
-1.16546795e-01 -8.07243511e-02 -5.74729562e-01 6.81097388e-01
4.15139556e-01 1.38994718e+00 2.23425299e-01 -7.83461988e-01
-1.21961378e-01 -6.69325590e-02 5.24211228e-01 -4.31829959e-01
-6.05050087e-01 3.31234008e-01 -3.45521867e-01 -1.09411085e+00
-9.25707281e-01 -2.68932790e-01 -1.28967214e+00 -1.67771682e-01
-4.67519104e-01 -2.49521896e-01 8.04965794e-01 1.10013461e+00
1.76844746e-01 8.44862044e-01 7.13561654e-01 -1.08245599e+00
4.16156292e-01 -9.85003889e-01 -8.64251733e-01 4.22547877e-01
5.70493698e-01 -4.30949301e-01 -5.60101926e-01 9.15730372e-02] | [9.436257362365723, -1.214186191558838] |
421fef95-ffc6-4a37-ad07-d131039d3a9e | stg2seq-spatial-temporal-graph-to-sequence | 1905.10069 | null | https://arxiv.org/abs/1905.10069v1 | https://arxiv.org/pdf/1905.10069v1.pdf | STG2Seq: Spatial-temporal Graph to Sequence Model for Multi-step Passenger Demand Forecasting | Multi-step passenger demand forecasting is a crucial task in on-demand vehicle sharing services. However, predicting passenger demand over multiple time horizons is generally challenging due to the nonlinear and dynamic spatial-temporal dependencies. In this work, we propose to model multi-step citywide passenger demand prediction based on a graph and use a hierarchical graph convolutional structure to capture both spatial and temporal correlations simultaneously. Our model consists of three parts: 1) a long-term encoder to encode historical passenger demands; 2) a short-term encoder to derive the next-step prediction for generating multi-step prediction; 3) an attention-based output module to model the dynamic temporal and channel-wise information. Experiments on three real-world datasets show that our model consistently outperforms many baseline methods and state-of-the-art models. | ['Salil S. Kanhere', 'Quan Z. Sheng', 'Lina Yao', 'Xianzhi Wang', 'Lei Bai'] | 2019-05-24 | null | null | null | null | ['graph-to-sequence'] | ['natural-language-processing'] | [-4.28786069e-01 -9.51754972e-02 -4.95319009e-01 -7.95129299e-01
-6.56540871e-01 -2.13656113e-01 6.32132649e-01 -8.24953914e-02
7.73200616e-02 6.80356503e-01 7.28662193e-01 -7.32167304e-01
-1.24945842e-01 -1.08752310e+00 -7.13100731e-01 -1.74980789e-01
-4.26541984e-01 7.55013287e-01 4.63083029e-01 -7.04048514e-01
-7.99606889e-02 5.41784644e-01 -1.28668070e+00 3.79798412e-01
7.69052863e-01 1.24214399e+00 1.88296393e-01 7.62046576e-01
-2.49334991e-01 1.24407506e+00 -1.25437692e-01 -5.90588033e-01
1.49576783e-01 -1.38693035e-01 -5.91138422e-01 4.63234261e-02
-2.44648397e-01 -5.42163312e-01 -1.04027951e+00 3.55855912e-01
3.72060984e-01 4.46808130e-01 4.75519270e-01 -1.53301048e+00
-8.55941892e-01 5.64672410e-01 -2.40184486e-01 6.76423609e-01
9.99457985e-02 3.05874228e-01 9.69601989e-01 -5.37388027e-01
2.42030188e-01 9.87919688e-01 5.02402365e-01 4.17552650e-01
-9.59418178e-01 -6.21912599e-01 9.61442411e-01 5.16754687e-01
-1.47446620e+00 -2.98956096e-01 1.02058887e+00 -5.21387994e-01
1.43176436e+00 1.16289429e-01 6.48031294e-01 8.95848870e-01
2.51186013e-01 9.70664382e-01 2.55223364e-01 3.99808675e-01
-1.34643927e-01 -1.07246064e-01 -2.48088688e-02 4.01152700e-01
-4.17092294e-01 2.84308761e-01 -2.57806748e-01 2.44859725e-01
3.32739085e-01 2.56368965e-01 2.63604790e-01 4.26149666e-01
-4.75943178e-01 8.16200197e-01 6.01441205e-01 7.77596757e-02
-8.07515085e-01 3.10478121e-01 3.44552606e-01 1.15971468e-01
1.08377540e+00 -3.31287652e-01 -5.16533792e-01 -3.58122528e-01
-9.95242774e-01 3.77822846e-01 7.68052697e-01 1.33265579e+00
6.40604377e-01 3.90485823e-01 -3.65231752e-01 6.23684943e-01
4.40995008e-01 2.35480204e-01 4.04885083e-01 -1.28994375e-01
9.78045106e-01 3.73561203e-01 7.86971953e-03 -9.43468690e-01
-8.36895943e-01 -5.16786516e-01 -7.30310917e-01 -5.52391946e-01
-1.72757193e-01 -5.01019239e-01 -9.77391839e-01 1.59856081e+00
7.67004397e-03 7.91470051e-01 -4.78175730e-01 9.41697359e-01
8.76943529e-01 1.07184064e+00 2.96116859e-01 -2.51014885e-02
5.61140776e-01 -1.50239134e+00 -7.44858146e-01 -2.04086468e-01
9.27865684e-01 -1.73159048e-01 4.17092144e-01 -1.07921205e-01
-1.25548303e+00 -7.04798818e-01 -3.87324721e-01 -2.79288858e-01
-7.42948055e-01 2.47797221e-02 5.22321939e-01 2.15938032e-01
-1.08376932e+00 1.54376909e-01 -5.58166146e-01 1.73851669e-01
3.00909758e-01 5.34317255e-01 1.76525325e-01 -9.39333625e-03
-1.50914800e+00 6.38523340e-01 1.78202525e-01 2.78976887e-01
-8.61253202e-01 -1.00131106e+00 -9.89951551e-01 3.13652635e-01
1.07735498e-02 -6.52705550e-01 1.15842724e+00 -7.23771095e-01
-1.43770587e+00 3.03541988e-01 -5.54892540e-01 -4.82817769e-01
4.44892883e-01 2.13220671e-01 -1.42168629e+00 -7.37127841e-01
8.44271779e-02 3.46671462e-01 5.80558240e-01 -9.35291111e-01
-1.10802913e+00 -3.48284036e-01 1.06651723e-01 2.32241035e-01
1.77438799e-02 -1.84393004e-01 -9.23340976e-01 -5.68278015e-01
-5.35904825e-01 -9.80018198e-01 -4.58727419e-01 -9.23752487e-01
-4.55622643e-01 -5.27691603e-01 7.83797920e-01 -9.65124965e-01
1.89191854e+00 -1.98681176e+00 -2.23085538e-01 2.60615855e-01
1.10633813e-01 1.36856630e-01 -2.80986071e-01 9.22963440e-01
-1.58302128e-01 -6.54075146e-02 3.59347761e-01 -7.72925019e-01
1.84281573e-01 2.32509896e-01 -5.74204206e-01 1.81144953e-01
2.18192846e-01 1.63772047e+00 -8.59618008e-01 -6.67221174e-02
3.86987597e-01 5.80190480e-01 -4.79868770e-01 4.64017659e-01
-2.70360440e-01 4.92288381e-01 -4.29672331e-01 4.35435712e-01
7.72660434e-01 -2.82812834e-01 -4.59790090e-03 5.99773303e-02
-1.49025917e-01 8.23878825e-01 -6.44202054e-01 1.30143070e+00
-9.25198078e-01 5.71301639e-01 -2.60938287e-01 -9.96764839e-01
9.26149189e-01 1.42109588e-01 4.52068597e-01 -1.30743420e+00
-1.87042728e-02 1.15240522e-01 -2.24888608e-01 -4.95530188e-01
9.51557875e-01 5.94412722e-02 -3.97957325e-01 2.28137851e-01
-1.35420278e-01 2.64517397e-01 2.74674535e-01 2.54736453e-01
9.61762965e-01 -3.19978982e-01 -5.87892413e-01 8.03178176e-02
6.83824122e-01 -2.69019812e-01 6.64486408e-01 1.18093200e-01
-6.70838133e-02 4.38525021e-01 4.35351670e-01 -9.29837227e-01
-7.71273911e-01 -8.06232035e-01 4.17876720e-01 1.26242220e+00
1.43331096e-01 -2.21858144e-01 2.82845069e-02 -5.53747833e-01
4.74860162e-01 1.16766167e+00 -6.90559089e-01 -1.41286656e-01
-6.70729876e-01 -5.88557482e-01 2.05333740e-01 1.01352668e+00
-1.49267297e-02 -8.28718543e-01 4.29643214e-01 5.21411598e-01
-4.13094386e-02 -1.24301040e+00 -1.02300549e+00 -3.64104033e-01
-4.49721456e-01 -4.23631847e-01 -6.52767599e-01 -6.90745890e-01
4.86670524e-01 3.89334142e-01 1.42229521e+00 1.07527539e-01
2.54495531e-01 6.35891333e-02 -1.13251589e-01 -4.60485667e-01
-1.17433406e-01 6.18211329e-01 -2.38583922e-01 4.28559095e-01
5.61364353e-01 -7.27754653e-01 -8.59956563e-01 3.01835090e-01
-7.31447160e-01 3.85678530e-01 1.21498376e-01 5.33854842e-01
5.64097106e-01 1.08883530e-01 1.14252639e+00 -9.75696266e-01
9.79043305e-01 -1.25288415e+00 -6.14768624e-01 -1.49875609e-02
-6.66006923e-01 -3.87674659e-01 1.08286321e+00 -4.84407181e-03
-1.00003290e+00 -1.08995892e-01 -7.70778596e-01 -3.50198030e-01
-5.67005910e-02 7.36837745e-01 -3.16949608e-03 5.37124991e-01
-2.55281538e-01 4.46869552e-01 -4.84534889e-01 -4.61372614e-01
4.62938488e-01 7.41221607e-01 2.71606386e-01 -4.10725921e-02
5.56521714e-01 2.13872254e-01 -8.75966102e-02 -3.06957841e-01
-6.75874591e-01 -5.92020214e-01 -5.67577660e-01 -3.50806922e-01
7.11909890e-01 -1.30118942e+00 -8.84714365e-01 3.47295403e-01
-1.06796432e+00 -8.13133836e-01 -2.62895167e-01 3.21206123e-01
-5.59028864e-01 -2.52023339e-01 -8.68848741e-01 -1.03272390e+00
-1.45084918e-01 -1.11971653e+00 1.16648102e+00 1.25560254e-01
2.79161632e-01 -1.60603321e+00 -5.57807572e-02 3.74261647e-01
7.16166437e-01 -1.41804338e-01 8.12849998e-01 -5.38374543e-01
-5.82810581e-01 -3.92654330e-01 -4.37625736e-01 -2.29699910e-02
5.60084432e-02 -2.47489929e-01 -7.75360286e-01 -1.54386073e-01
-9.41092193e-01 1.30585536e-01 8.36087763e-01 4.65937585e-01
1.64176381e+00 -3.74635279e-01 -5.79274297e-01 6.01926684e-01
1.17781711e+00 3.60571414e-01 5.69977283e-01 -2.85125881e-01
1.02244830e+00 6.71361864e-01 3.24744880e-01 5.22509217e-01
1.65617490e+00 6.72747135e-01 6.80324554e-01 -1.77122608e-01
-2.22986817e-01 -8.07281137e-01 8.27624872e-02 1.31301939e+00
4.54194285e-02 -9.06565547e-01 -9.21686709e-01 1.13250470e+00
-2.26696038e+00 -9.68682766e-01 -4.24630374e-01 1.78640211e+00
1.05773509e-01 2.61801958e-01 6.68725371e-01 -2.72792637e-01
2.88932234e-01 4.49943036e-01 -7.39734650e-01 -8.14711213e-01
8.77236426e-02 -4.34929222e-01 8.52615237e-01 6.06010377e-01
-6.89160645e-01 8.75138164e-01 6.42249489e+00 6.71504080e-01
-1.11856520e+00 8.43136609e-02 1.11079085e+00 -2.41434380e-01
-9.21553195e-01 -4.06526864e-01 -9.37265933e-01 1.12763417e+00
1.45295894e+00 -4.76662874e-01 7.95990348e-01 6.19529843e-01
5.17838717e-01 4.91166085e-01 -8.63373458e-01 8.66825640e-01
6.54840749e-03 -1.46162844e+00 1.67894140e-01 2.29765639e-01
1.08343375e+00 5.89681447e-01 1.48688972e-01 7.29453743e-01
5.65164387e-01 -1.13645828e+00 6.76515400e-01 1.14929676e+00
6.58369839e-01 -1.13068938e+00 6.14956439e-01 6.70684993e-01
-1.74394536e+00 -3.90652090e-01 6.13983572e-02 -1.80454165e-01
1.09771729e+00 5.97860694e-01 -5.76883972e-01 7.12245882e-01
2.86637545e-01 1.18813062e+00 -2.65887082e-01 7.88902521e-01
1.50339141e-01 6.88480616e-01 -4.09519039e-02 1.60329908e-01
5.08123517e-01 -7.82141462e-02 2.98867542e-02 1.50895584e+00
4.93879467e-01 8.92032012e-02 1.86236843e-01 5.80070019e-01
-2.75096536e-01 -1.48198441e-01 -6.75428510e-01 -8.28626230e-02
1.90453485e-01 1.03688896e+00 -3.89446765e-02 -3.77388269e-01
-1.10363603e+00 9.48456287e-01 3.23312968e-01 6.79729640e-01
-1.16405344e+00 -1.43887565e-01 9.79411364e-01 4.65014637e-01
6.06466413e-01 -3.85902643e-01 -1.05131000e-01 -8.98622990e-01
-1.70530751e-01 -5.91080301e-02 6.01959705e-01 -8.16667616e-01
-1.52637935e+00 7.10064292e-01 -3.59889686e-01 -1.20614254e+00
-6.80007279e-01 -2.64168739e-01 -8.92321527e-01 1.38594532e+00
-2.18804193e+00 -1.49127221e+00 -1.96528882e-01 7.95451045e-01
8.87297571e-01 -1.21264279e-01 4.06681508e-01 9.25773859e-01
-7.56200314e-01 5.34474492e-01 -4.40057442e-02 6.54152930e-02
2.23816056e-02 -9.04515505e-01 1.19324744e+00 6.81287885e-01
-2.77193069e-01 -2.06176385e-01 1.88482791e-01 -6.80373847e-01
-1.37260628e+00 -1.63966286e+00 1.42092896e+00 -3.27322304e-01
8.47237527e-01 -3.46707791e-01 -7.72668302e-01 1.18350184e+00
1.34844422e-01 2.04572693e-01 7.49279916e-01 2.13595167e-01
2.89899372e-02 -4.30747330e-01 -6.42806828e-01 3.06934834e-01
1.15252423e+00 -6.13875926e-01 1.54612005e-01 3.79327267e-01
9.88072276e-01 -4.90755618e-01 -7.46598005e-01 1.14712007e-01
4.14452016e-01 -6.99864328e-01 6.52766824e-01 -8.47545087e-01
3.00425708e-01 6.23359270e-02 1.96600035e-02 -1.68417060e+00
-8.63543332e-01 -5.49422562e-01 -4.85559732e-01 1.08209789e+00
9.13412035e-01 -7.95828640e-01 7.06397593e-01 1.20651221e+00
-6.11957967e-01 -1.01684284e+00 -8.87319326e-01 -5.31424880e-01
-5.72644882e-02 -1.11105084e+00 1.45788264e+00 6.78085506e-01
-1.93333626e-01 3.42193753e-01 -7.85715163e-01 2.18735173e-01
-7.75928944e-02 2.90640205e-01 7.29016006e-01 -9.60410953e-01
-8.18360895e-02 -6.39931679e-01 -3.68652314e-01 -1.79679525e+00
4.30875003e-01 -9.23553288e-01 -1.97337478e-01 -1.82481074e+00
-2.81435192e-01 -4.43145007e-01 -5.97018003e-01 2.24657040e-02
9.58386362e-02 1.34609221e-02 2.39334982e-02 -1.32348105e-01
-9.12355244e-01 7.20471561e-01 1.36849356e+00 -1.48678809e-01
-4.41783637e-01 2.94283628e-01 -7.66114056e-01 1.33622050e-01
7.66893327e-01 -7.89418519e-02 -7.85261095e-01 -7.00113893e-01
6.24195635e-01 5.06382585e-01 3.64485644e-02 -4.67215210e-01
4.85824078e-01 -4.28608328e-01 5.48387803e-02 -1.17004788e+00
5.59551656e-01 -8.01860452e-01 2.14121670e-01 -1.44118946e-02
-1.96276709e-01 6.53272510e-01 4.04221863e-01 8.12116027e-01
-2.99470216e-01 5.80319464e-01 -1.65004790e-01 3.09664845e-01
-1.13111460e+00 1.24238098e+00 -4.54115242e-01 -2.23948553e-01
9.20063019e-01 1.81425691e-01 -2.65568972e-01 -1.02807391e+00
-7.27029860e-01 1.00204170e+00 -3.00174147e-01 1.06772971e+00
4.66949373e-01 -1.74399900e+00 -7.37115085e-01 5.85070193e-01
8.52400213e-02 -2.16407061e-01 7.65450835e-01 8.36160183e-01
-1.69500440e-01 8.83671045e-01 2.12441832e-01 -1.11085974e-01
-4.89744931e-01 7.76866198e-01 2.98210531e-01 -6.25227511e-01
-3.28021437e-01 8.72597873e-01 1.81524709e-01 -4.04888451e-01
-2.89807115e-02 -3.54211837e-01 -2.37610832e-01 4.24465127e-02
5.27834713e-01 5.91068506e-01 1.37613267e-01 -1.29478347e+00
-3.39049429e-01 1.78988665e-01 8.93566385e-02 1.97894752e-01
1.38532925e+00 -7.13216782e-01 4.01594311e-01 5.09498537e-01
1.46888745e+00 -3.62760603e-01 -1.31035089e+00 -2.87076026e-01
-3.32066804e-01 -6.09021485e-01 3.22424233e-01 -7.27529764e-01
-1.55802464e+00 9.49995220e-01 2.12925356e-02 9.35237467e-01
1.34129214e+00 -1.99441165e-02 1.74348068e+00 -3.75387639e-01
2.23609641e-01 -1.23366249e+00 -5.12909055e-01 7.04107523e-01
8.76387715e-01 -1.34739196e+00 -6.29108369e-01 -2.98871547e-01
-8.38219941e-01 7.80699372e-01 5.39004147e-01 -1.75788954e-01
1.25613058e+00 1.31864727e-01 -1.11123957e-01 -8.49203914e-02
-1.40385091e+00 -6.73446536e-01 7.87400782e-01 4.85443026e-01
3.59140694e-01 5.14274657e-01 -1.53334513e-01 1.08589935e+00
-1.91979066e-01 3.01225334e-01 4.62299511e-02 2.33447358e-01
2.18265325e-01 -9.52552259e-01 5.52787244e-01 8.03589046e-01
-1.91069052e-01 -5.00762351e-02 1.29185095e-01 4.88996476e-01
2.56289188e-02 1.15756357e+00 5.31141043e-01 -9.53654826e-01
6.29208744e-01 -1.02769144e-01 -2.04988122e-01 -3.01566094e-01
-5.87270737e-01 -2.67041564e-01 3.51129621e-01 -7.96710014e-01
9.39818621e-02 -4.54344779e-01 -1.17534590e+00 -8.55467021e-01
1.56101659e-01 -2.12862656e-01 7.28388190e-01 8.75623107e-01
9.71372664e-01 9.36029434e-01 1.10705757e+00 -8.13511312e-01
2.12483764e-01 -7.46646285e-01 -8.59739900e-01 5.89663208e-01
7.13258624e-01 -4.82146025e-01 3.28050740e-02 -2.62450755e-01] | [6.531729698181152, 2.136101245880127] |
0801b56b-0cef-41bf-a177-82d9d10eca59 | haa4d-few-shot-human-atomic-action | 2202.07308 | null | https://arxiv.org/abs/2202.07308v1 | https://arxiv.org/pdf/2202.07308v1.pdf | HAA4D: Few-Shot Human Atomic Action Recognition via 3D Spatio-Temporal Skeletal Alignment | Human actions involve complex pose variations and their 2D projections can be highly ambiguous. Thus 3D spatio-temporal or 4D (i.e., 3D+T) human skeletons, which are photometric and viewpoint invariant, are an excellent alternative to 2D+T skeletons/pixels to improve action recognition accuracy. This paper proposes a new 4D dataset HAA4D which consists of more than 3,300 RGB videos in 300 human atomic action classes. HAA4D is clean, diverse, class-balanced where each class is viewpoint-balanced with the use of 4D skeletons, in which as few as one 4D skeleton per class is sufficient for training a deep recognition model. Further, the choice of atomic actions makes annotation even easier, because each video clip lasts for only a few seconds. All training and testing 3D skeletons in HAA4D are globally aligned, using a deep alignment model to the same global space, making each skeleton face the negative z-direction. Such alignment makes matching skeletons more stable by reducing intraclass variations and thus with fewer training samples per class needed for action recognition. Given the high diversity and skeletal alignment in HAA4D, we construct the first baseline few-shot 4D human atomic action recognition network without bells and whistles, which produces comparable or higher performance than relevant state-of-the-art techniques relying on embedded space encoding without explicit skeletal alignment, using the same small number of training samples of unseen classes. | ['Yu-Wing Tai', 'Chi-Keung Tang', 'Abhishek Gupta', 'Mu-Ruei Tseng'] | 2022-02-15 | null | null | null | null | ['atomic-action-recognition'] | ['computer-vision'] | [ 2.44425878e-01 9.49211568e-02 -4.73028600e-01 -1.20215580e-01
-4.74179149e-01 -2.69300938e-01 5.43526232e-01 -3.43935907e-01
-3.95607024e-01 4.30934668e-01 2.77117580e-01 4.12990570e-01
2.04578951e-01 -5.47198713e-01 -6.51590884e-01 -7.38332152e-01
7.01778978e-02 5.51212728e-01 6.10145450e-01 -3.06129046e-02
-2.83527002e-02 7.44036913e-01 -1.83746302e+00 2.49919847e-01
1.28218681e-01 1.19289958e+00 -2.51413435e-01 5.51545441e-01
8.11178908e-02 6.02957845e-01 -6.24295890e-01 -3.21238965e-01
8.23204756e-01 -6.77212417e-01 -6.44414902e-01 6.25438154e-01
9.01268125e-01 -6.32275641e-01 -6.47380471e-01 6.31117702e-01
5.29965699e-01 4.09097731e-01 6.00073397e-01 -1.09389651e+00
-4.47071314e-01 -3.04348096e-02 -8.85076761e-01 -5.33395000e-02
5.97088516e-01 4.45639163e-01 8.51250112e-01 -8.08062792e-01
7.63382435e-01 1.38234031e+00 5.94086766e-01 9.38282967e-01
-1.08074927e+00 -4.81704801e-01 1.11330107e-01 3.99277717e-01
-1.24790204e+00 -5.09952486e-01 6.83765531e-01 -3.33529472e-01
1.27992690e+00 2.01408327e-01 1.21131086e+00 1.58816850e+00
1.96638584e-01 1.06892312e+00 7.63984799e-01 -4.48247552e-01
2.99734443e-01 -6.42602444e-01 -1.41051214e-03 8.09007168e-01
2.78490782e-01 -2.16151416e-01 -9.01241362e-01 1.90096498e-01
1.18334258e+00 3.02250713e-01 -1.53719127e-01 -9.52940881e-01
-1.46154118e+00 5.05472243e-01 2.17263252e-01 1.23223938e-01
-3.95450532e-01 3.57702523e-01 5.32061398e-01 1.47823483e-01
1.68563686e-02 1.12085059e-01 -3.36115956e-01 -3.96492869e-01
-7.73110926e-01 2.14701548e-01 2.13330060e-01 9.03474987e-01
5.90662956e-01 1.92552462e-01 -1.50257364e-01 7.99736142e-01
-3.83408507e-03 7.73181796e-01 7.69226253e-01 -1.46315324e+00
5.50091624e-01 7.37017453e-01 4.44946345e-04 -9.21269596e-01
-4.05045599e-01 -6.04294315e-02 -8.78583431e-01 5.39521337e-01
6.73159719e-01 3.32585216e-01 -1.13992155e+00 1.65952158e+00
5.62262893e-01 -6.72999537e-03 -9.28717628e-02 9.23608780e-01
3.85013223e-01 1.88616350e-01 -2.87100017e-01 5.46706319e-02
1.27207077e+00 -1.16562331e+00 -4.45949614e-01 -3.33444238e-01
5.46757996e-01 -4.27839100e-01 1.16568887e+00 4.83353555e-01
-1.04668093e+00 -8.34746420e-01 -9.89785552e-01 -2.73778766e-01
-1.46193758e-01 1.73575968e-01 5.59571683e-01 7.18126655e-01
-6.91522419e-01 5.80523372e-01 -1.26495636e+00 -5.01738489e-01
5.53021014e-01 2.78558254e-01 -8.75569344e-01 -8.23850557e-02
-8.45281839e-01 8.40265155e-01 2.36185342e-01 -6.26244769e-02
-8.82696748e-01 -2.20729783e-01 -1.12096870e+00 -3.09337556e-01
4.79397357e-01 -6.72939837e-01 9.73409772e-01 -8.54224145e-01
-1.71947527e+00 1.18702316e+00 -6.33961940e-03 -5.00224531e-01
6.79559052e-01 -4.71796989e-01 -1.48416623e-01 4.25721109e-01
2.56486565e-01 8.97873700e-01 1.01094401e+00 -6.64977312e-01
-4.43389952e-01 -7.53007889e-01 1.19329365e-02 2.70096272e-01
-2.56602287e-01 -3.11232686e-01 -7.75386810e-01 -8.53574276e-01
3.45038086e-01 -1.10047889e+00 -9.30707529e-02 4.24050570e-01
-2.63405055e-01 -5.83093427e-02 7.28222311e-01 -5.58655500e-01
7.55425692e-01 -2.19565058e+00 4.93104309e-01 5.33046536e-02
-1.64482780e-02 1.80997133e-01 -3.56253207e-01 8.68665799e-02
-1.60371631e-01 -3.64872485e-01 -2.98790336e-01 -3.79765689e-01
2.24971622e-02 5.60101628e-01 1.00104725e-02 7.51980186e-01
7.09423721e-02 8.45754623e-01 -7.71589875e-01 -4.73195761e-01
5.63097835e-01 4.03391719e-01 -4.12952870e-01 -1.31094130e-02
7.22366646e-02 5.49364686e-01 -4.33333397e-01 9.24666762e-01
4.08052087e-01 -1.13242656e-01 -1.66190267e-02 -9.35032964e-02
2.35329241e-01 1.00830846e-01 -1.28828788e+00 2.27106285e+00
-2.46074557e-01 4.69087273e-01 -3.65131438e-01 -1.04522943e+00
9.54264641e-01 1.95197865e-01 9.58320796e-01 -9.37040925e-01
1.18364014e-01 1.67587698e-01 -2.94512302e-01 -3.68375808e-01
3.56916040e-02 3.32960635e-02 -7.26949126e-02 5.14877200e-01
6.53646365e-02 -9.98778269e-02 2.24643290e-01 -2.18020454e-01
1.28261340e+00 5.43665588e-01 3.78135741e-01 3.11514556e-01
3.59398186e-01 -1.72753811e-01 8.54170263e-01 5.67778230e-01
-5.51267445e-01 9.90180850e-01 3.45567942e-01 -6.61607146e-01
-1.07177877e+00 -1.07170963e+00 1.35588727e-03 6.07668281e-01
2.44858995e-01 -3.01573455e-01 -8.16102743e-01 -7.67055631e-01
-7.39636347e-02 4.38688636e-01 -7.57251501e-01 -1.95889726e-01
-8.43329906e-01 -2.71534145e-01 5.70802331e-01 8.97088408e-01
8.74446511e-01 -1.03743017e+00 -1.25173199e+00 -3.77831049e-03
-9.95393917e-02 -1.09476840e+00 -3.32106411e-01 1.87367469e-01
-1.08951020e+00 -1.40991008e+00 -1.17512071e+00 -3.83928865e-01
5.53893626e-01 4.30574298e-01 9.41968143e-01 -2.55795866e-01
-5.43713570e-01 7.96182930e-01 -6.01812780e-01 7.97937587e-02
9.19855163e-02 -3.81729722e-01 4.12108421e-01 1.46835029e-01
6.30325794e-01 -6.89192772e-01 -6.80288017e-01 5.35767078e-01
-6.05264246e-01 8.13099965e-02 6.42344713e-01 8.75617862e-01
9.00032759e-01 -1.55964568e-01 -6.80169538e-02 -2.79349655e-01
-3.49214286e-01 1.20093405e-01 -1.83685347e-01 1.72629222e-01
-7.57192299e-02 -2.15060890e-01 3.38055998e-01 -4.09564734e-01
-7.97365785e-01 5.02656341e-01 1.28982902e-01 -9.11046982e-01
-3.44037622e-01 -1.93528652e-01 -2.92551875e-01 6.32555038e-02
8.94374907e-01 1.71389177e-01 3.22120517e-01 -6.28649712e-01
3.02424639e-01 2.99171150e-01 6.69457436e-01 -3.23349088e-01
6.81523800e-01 7.83029675e-01 2.59753793e-01 -9.90350485e-01
-6.45915151e-01 -6.08192325e-01 -1.34087586e+00 -3.58636707e-01
1.30645037e+00 -8.82775843e-01 -3.68800163e-01 8.16918492e-01
-9.82219040e-01 -5.59046686e-01 -7.87767351e-01 7.10777164e-01
-9.78724718e-01 7.44214416e-01 -5.35861254e-01 -5.80087602e-01
-6.60189763e-02 -1.09912026e+00 1.58268905e+00 -1.93982005e-01
-5.32930374e-01 -4.28631067e-01 1.27600729e-01 6.75919175e-01
-3.06873679e-01 6.22841656e-01 5.52905023e-01 -2.88277775e-01
-4.87802148e-01 -4.60439742e-01 1.92754999e-01 6.52402461e-01
3.61159593e-01 -2.26015687e-01 -8.62147272e-01 -1.74319923e-01
9.32272896e-02 -5.39358079e-01 8.12247455e-01 2.65209258e-01
1.14339042e+00 6.67880923e-02 -9.92159843e-02 5.16233027e-01
6.96382284e-01 8.50721076e-02 8.49758804e-01 5.34837067e-01
8.72244477e-01 5.51057220e-01 8.19214582e-01 5.54067135e-01
-4.32712473e-02 1.12051415e+00 5.10035813e-01 5.61115041e-04
-5.37234962e-01 -1.04035027e-01 7.63844788e-01 5.78013659e-01
-5.25414109e-01 6.81218356e-02 -7.81782508e-01 3.02824646e-01
-1.77422702e+00 -1.23159158e+00 -3.45158987e-02 2.36929679e+00
6.44693375e-01 9.86894965e-03 3.96279305e-01 5.32333910e-01
4.96792674e-01 4.06481266e-01 -7.28966594e-01 8.84907544e-02
-2.74945349e-01 4.42064643e-01 4.76379484e-01 1.62371382e-01
-1.18025899e+00 6.64173365e-01 5.68057394e+00 8.72586548e-01
-8.18401694e-01 -1.19511731e-01 3.77148479e-01 -4.62334484e-01
2.19568431e-01 -3.14961612e-01 -6.62681997e-01 2.64572680e-01
4.84027117e-01 5.95092237e-01 1.01968482e-01 1.10946834e+00
1.02529585e-01 -1.84044063e-01 -1.29764867e+00 1.52284992e+00
2.73567498e-01 -1.13135922e+00 1.90141961e-01 1.01996377e-01
7.04421878e-01 -1.01966545e-01 -1.03530899e-01 8.98682252e-02
-1.07947260e-01 -8.50271642e-01 8.38110685e-01 4.15561527e-01
9.69728529e-01 -5.41328847e-01 3.99702787e-01 3.06264430e-01
-1.05938530e+00 -3.47952209e-02 -5.67607522e-01 6.31671622e-02
2.41233125e-01 2.13224113e-01 -1.39027014e-01 3.74929160e-01
9.78967071e-01 1.20475769e+00 -6.10969722e-01 6.74779296e-01
-2.08110586e-01 1.09977067e-01 -3.73201787e-01 4.01239127e-01
2.47270405e-01 -2.33590305e-01 4.83821630e-01 9.16995347e-01
4.02990520e-01 2.05706596e-01 1.14142366e-01 2.06212088e-01
1.96070999e-01 -1.25499427e-01 -6.90149367e-01 1.90739766e-01
-4.77257147e-02 6.96020424e-01 -8.57546926e-01 -4.30986375e-01
-5.80140352e-01 1.58990502e+00 -6.76118433e-02 2.00830385e-01
-8.51058662e-01 -1.31853223e-01 9.38005328e-01 1.43443912e-01
5.38187861e-01 -4.38673139e-01 -1.48542643e-01 -1.40233934e+00
3.11304033e-01 -1.16823387e+00 5.15407026e-01 -6.41481221e-01
-1.12119615e+00 2.86579877e-01 6.72987700e-02 -1.74409068e+00
-3.48344684e-01 -9.86311078e-01 -2.17401654e-01 3.03548008e-01
-1.02887225e+00 -1.09683323e+00 -5.63253999e-01 9.57366228e-01
8.75278413e-01 -3.07113856e-01 9.83016074e-01 2.47215912e-01
-4.44120049e-01 5.53776503e-01 -8.41149017e-02 2.25543782e-01
6.97201192e-01 -1.10919976e+00 5.89686096e-01 7.35370278e-01
5.09690166e-01 2.94741839e-01 1.80472165e-01 -5.61469316e-01
-1.54392326e+00 -8.66751015e-01 5.84559262e-01 -9.33480322e-01
3.31149101e-01 -3.50212157e-01 -7.66688943e-01 6.23061419e-01
-2.93150336e-01 2.67497420e-01 6.73360348e-01 -1.55316189e-01
-5.04405439e-01 -1.97784409e-01 -9.52826142e-01 7.08656728e-01
1.68068230e+00 -5.17101705e-01 -6.48840606e-01 4.77437913e-01
2.54516840e-01 -5.36191344e-01 -8.37356567e-01 4.54149187e-01
7.91633487e-01 -1.42720532e+00 1.28905940e+00 -4.17241186e-01
2.33399674e-01 -4.66141343e-01 -4.12462294e-01 -6.26790702e-01
-1.14474423e-01 -3.04355532e-01 -4.09568489e-01 5.39174438e-01
-1.51270881e-01 -2.73199350e-01 1.19263804e+00 4.25128937e-01
-1.39575586e-01 -7.52734303e-01 -1.37392223e+00 -1.23832822e+00
-3.29081148e-01 -5.40517449e-01 3.21414262e-01 7.42147744e-01
-3.46202731e-01 6.23281784e-02 -5.61430693e-01 -3.18822503e-01
6.37895763e-01 1.51309565e-01 1.27203691e+00 -1.01135480e+00
-5.63477755e-01 -4.21712220e-01 -1.14602757e+00 -1.50122344e+00
1.32594958e-01 -6.13808274e-01 -2.16884956e-01 -1.13274705e+00
4.98681143e-02 -1.50413662e-01 -8.57293755e-02 8.30335915e-01
2.40317553e-01 7.16391802e-01 1.33215204e-01 3.50275397e-01
-4.50069636e-01 8.19753349e-01 1.40518332e+00 -3.45863044e-01
-1.19914584e-01 -9.13233124e-03 1.24765530e-01 8.67498994e-01
4.23069686e-01 -2.02605903e-01 -4.58331168e-01 -5.25365770e-01
-3.92828614e-01 -9.36090574e-02 5.69014490e-01 -1.37043524e+00
-1.42646194e-01 -1.92104876e-01 7.81185806e-01 -6.06920540e-01
9.70234752e-01 -9.08733487e-01 2.55921990e-01 5.77637017e-01
-1.24289237e-01 -1.52349442e-01 -2.89556235e-02 7.32184947e-01
-2.37317264e-01 -1.13409214e-01 8.20236504e-01 -4.09834027e-01
-9.17820334e-01 4.80322242e-01 -1.83311135e-01 -1.08358145e-01
1.25806284e+00 -9.89266515e-01 2.91305780e-01 -2.53174037e-01
-7.80941904e-01 -2.57424355e-01 7.66576052e-01 5.00844717e-01
6.54506266e-01 -1.56540322e+00 -3.79564792e-01 3.08486104e-01
1.97724432e-01 1.06507823e-01 4.60909158e-01 7.66739070e-01
-6.23539507e-01 3.55546862e-01 -8.06576192e-01 -8.90627563e-01
-1.34169424e+00 4.43840116e-01 3.33557725e-01 9.57118571e-02
-1.17049539e+00 7.84795165e-01 2.84631401e-01 -2.60116190e-01
4.53308791e-01 -2.40343541e-01 4.83741574e-02 6.11580946e-02
5.66225052e-01 6.38513267e-01 -1.03745066e-01 -9.22886252e-01
-5.30004799e-01 1.10287440e+00 3.51844668e-01 -1.23135455e-01
1.25347221e+00 9.21176523e-02 1.56171903e-01 5.13597369e-01
1.10420394e+00 -1.53349265e-01 -1.59062278e+00 -7.20428750e-02
-1.72567934e-01 -8.81798506e-01 -2.80097455e-01 -3.55007231e-01
-1.18464136e+00 1.10342860e+00 6.51490331e-01 -2.31931314e-01
1.09396124e+00 -1.12281203e-01 9.74779785e-01 5.45182824e-01
6.02560759e-01 -1.29181957e+00 6.73682094e-01 3.53242069e-01
9.50493336e-01 -1.18789077e+00 1.39796391e-01 -1.64518878e-01
-7.49285638e-01 1.27439010e+00 6.85471296e-01 -8.71278495e-02
1.31366879e-01 -3.04373950e-01 1.48089960e-01 -2.19248638e-01
-1.51424959e-01 -1.97863862e-01 2.07399562e-01 9.14461195e-01
5.72877862e-02 -9.58713964e-02 6.55617639e-02 1.07192166e-01
4.10224870e-02 -3.57461751e-01 7.49241486e-02 1.03111899e+00
-2.56143361e-01 -1.25655222e+00 -3.65931928e-01 7.05674440e-02
-4.53587212e-02 4.95073050e-01 -4.20090407e-01 1.04107022e+00
1.73234940e-01 5.34287870e-01 9.66829360e-02 -3.64178687e-01
5.76062620e-01 1.79835305e-01 8.85404885e-01 -6.11194134e-01
-7.71995857e-02 3.89937423e-02 -2.07559634e-02 -1.24452245e+00
-6.87088132e-01 -9.38807487e-01 -1.38202894e+00 -9.72247794e-02
5.05823456e-02 -4.64973360e-01 3.13064873e-01 8.06539536e-01
3.14922124e-01 6.14596903e-02 4.21017677e-01 -1.24166822e+00
-5.80596685e-01 -6.97577834e-01 -7.10078835e-01 6.02213800e-01
1.59881607e-01 -1.15908229e+00 -2.20627204e-01 2.92001367e-01] | [7.8349761962890625, 0.3679516017436981] |
2818762d-ec50-4fd3-8fa1-a3a36b37875e | multinet-real-time-joint-semantic-reasoning | 1612.07695 | null | http://arxiv.org/abs/1612.07695v2 | http://arxiv.org/pdf/1612.07695v2.pdf | MultiNet: Real-time Joint Semantic Reasoning for Autonomous Driving | While most approaches to semantic reasoning have focused on improving
performance, in this paper we argue that computational times are very important
in order to enable real time applications such as autonomous driving. Towards
this goal, we present an approach to joint classification, detection and
semantic segmentation via a unified architecture where the encoder is shared
amongst the three tasks. Our approach is very simple, can be trained end-to-end
and performs extremely well in the challenging KITTI dataset, outperforming the
state-of-the-art in the road segmentation task. Our approach is also very
efficient, taking less than 100 ms to perform all tasks. | ['Raquel Urtasun', 'Michael Weber', 'Marvin Teichmann', 'Marius Zoellner', 'Roberto Cipolla'] | 2016-12-22 | null | null | null | null | ['road-segementation'] | ['computer-vision'] | [ 2.81558812e-01 4.15667921e-01 -1.07398391e-01 -5.76102316e-01
-9.60460305e-01 -5.35771132e-01 7.00045764e-01 -7.11387619e-02
-7.01231897e-01 5.13904274e-01 -1.71859369e-01 -6.61141217e-01
1.14346422e-01 -6.76088631e-01 -7.75050938e-01 -1.77265391e-01
1.13897629e-01 8.57086539e-01 9.19817686e-01 -2.87798166e-01
5.13497412e-01 1.99060842e-01 -1.74593472e+00 1.53223142e-01
6.53603911e-01 9.26221192e-01 1.71262994e-01 1.11644673e+00
-2.16730371e-01 9.41269279e-01 -3.07399213e-01 -5.16628146e-01
2.81387448e-01 -1.71301365e-01 -1.23719442e+00 -5.89082502e-02
6.27261579e-01 -2.33526066e-01 -1.64721921e-01 9.18023586e-01
3.53424013e-01 1.19707920e-01 4.21595067e-01 -1.06249511e+00
-2.19644010e-01 4.37146783e-01 -4.55795407e-01 2.65730739e-01
1.27211392e-01 -9.30249915e-02 1.15666032e+00 -6.65573120e-01
5.63703656e-01 1.20105267e+00 5.74137092e-01 5.00695050e-01
-1.02220130e+00 -3.72657597e-01 2.91106015e-01 2.31524661e-01
-1.04280615e+00 -5.05979061e-01 3.29726189e-01 -3.49023879e-01
1.27688491e+00 1.97727438e-02 3.99167329e-01 5.07258356e-01
-7.68973902e-02 1.22420657e+00 1.07855594e+00 -2.36235604e-01
1.79613635e-01 -6.76819086e-02 2.07853764e-01 8.51710439e-01
1.33106828e-01 -6.56491220e-02 -3.40759456e-01 2.99521714e-01
4.41392094e-01 -3.87034148e-01 3.74401331e-01 -4.03049558e-01
-1.19961238e+00 8.69227409e-01 4.88434374e-01 2.07086027e-01
-6.71457648e-02 6.06984437e-01 6.26801908e-01 2.10579440e-01
5.05812407e-01 1.71076566e-01 -5.14353573e-01 -5.63869298e-01
-1.03701270e+00 5.19982159e-01 8.28541875e-01 8.52889597e-01
5.88409245e-01 -1.45318076e-01 2.76161015e-01 6.82087779e-01
1.44634739e-01 2.01202571e-01 1.40858248e-01 -1.30608201e+00
4.63446885e-01 3.32935244e-01 1.17486879e-01 -3.91824216e-01
-6.07468426e-01 -1.95250422e-01 -2.27070928e-01 6.34894311e-01
4.82215732e-01 -1.43520817e-01 -1.29147995e+00 1.37080216e+00
2.35870793e-01 2.79999286e-01 3.76565605e-01 8.75095487e-01
6.75055742e-01 3.74105006e-01 3.99213880e-01 5.73142111e-01
1.65837383e+00 -1.45433486e+00 -4.05136704e-01 -9.46092904e-01
8.50624621e-01 -8.43014181e-01 6.64519429e-01 3.96330208e-01
-1.15610862e+00 -5.85189342e-01 -1.21541512e+00 -4.71701860e-01
-7.02258110e-01 1.48389772e-01 9.14162219e-01 5.45027316e-01
-1.18480361e+00 5.59637129e-01 -9.76709008e-01 -6.27195060e-01
7.05179513e-01 3.12378317e-01 -3.71193260e-01 1.86168067e-02
-9.85160053e-01 1.29519069e+00 8.00130010e-01 -5.33550270e-02
-6.04817390e-01 -2.58078933e-01 -8.29544961e-01 -1.38873398e-01
5.43840349e-01 -7.94157445e-01 1.89443874e+00 -7.91926503e-01
-1.60276783e+00 1.21010280e+00 -1.78824827e-01 -1.00490642e+00
6.13537848e-01 -5.10408938e-01 -1.47170082e-01 1.88060358e-01
2.52243221e-01 1.11673141e+00 4.92536694e-01 -1.02591479e+00
-1.38188970e+00 -2.84352660e-01 2.99453735e-01 2.26578966e-01
2.93860227e-01 1.52541593e-01 -5.70354223e-01 -9.11522582e-02
8.57038945e-02 -1.08527267e+00 -4.08035457e-01 -1.59618761e-02
-2.05565214e-01 -3.40056956e-01 1.04615545e+00 -6.32137239e-01
5.92130482e-01 -2.01081896e+00 7.79072940e-03 -2.20253721e-01
5.40757831e-03 5.33296645e-01 2.02400550e-01 5.69830425e-02
2.53797114e-01 -1.28922649e-02 -6.21736407e-01 -4.46353257e-01
2.09872440e-01 5.39046526e-01 -2.58238673e-01 3.59809756e-01
2.81791449e-01 1.08361161e+00 -1.03180528e+00 -4.79772687e-01
6.03295028e-01 2.91588008e-01 -2.43935451e-01 1.56714004e-02
-3.55761230e-01 2.80997127e-01 -4.68741149e-01 3.97924930e-01
3.57201666e-01 -8.45617205e-02 -2.83952430e-02 2.58797973e-01
-1.86316803e-01 4.24754977e-01 -9.94204640e-01 1.88813686e+00
-5.60447276e-01 9.54733729e-01 2.90266603e-01 -1.35041869e+00
7.30981112e-01 2.45466888e-01 1.53058752e-01 -9.57866311e-01
1.26229763e-01 5.95572531e-01 -3.12610269e-02 -3.19089115e-01
5.88619590e-01 -8.76105279e-02 -2.54882634e-01 2.12118626e-01
1.55822942e-02 -3.84482861e-01 4.70916063e-01 1.47751555e-01
1.04288328e+00 5.41657329e-01 1.95625484e-01 -2.53655434e-01
6.00300908e-01 4.86002505e-01 2.33691409e-01 6.01361930e-01
-3.81241620e-01 4.66954052e-01 5.51502347e-01 -5.69995701e-01
-1.11946905e+00 -9.11098897e-01 9.10127163e-02 1.17030418e+00
4.08565134e-01 -2.20050082e-01 -1.05884910e+00 -8.17176342e-01
-5.22646345e-02 9.05408561e-01 -5.74004114e-01 1.80763260e-01
-5.55386484e-01 -4.23403263e-01 7.28432715e-01 9.81400549e-01
9.53032255e-01 -9.10005569e-01 -1.10848415e+00 4.06685680e-01
3.95251289e-02 -1.68764949e+00 1.50898620e-01 1.86258495e-01
-7.41145194e-01 -1.10204244e+00 -4.94887561e-01 -6.97708845e-01
2.22216427e-01 3.86974990e-01 1.24846745e+00 -1.01247020e-01
-4.15449589e-01 1.58860520e-01 -3.02771747e-01 -6.10488415e-01
-3.62979740e-01 3.61665785e-01 -6.90054953e-01 -3.71817678e-01
4.44205284e-01 -9.58888084e-02 -5.98670959e-01 2.08383769e-01
-6.46973193e-01 2.30285302e-01 6.22313678e-01 4.48414922e-01
5.54390848e-01 -4.57828306e-02 3.83860826e-01 -1.20960140e+00
2.67422616e-01 -2.35863999e-01 -6.78233743e-01 -3.97488289e-02
-4.67240483e-01 9.98448394e-03 4.29486871e-01 3.90735537e-01
-9.63649154e-01 4.79177117e-01 -6.66070759e-01 1.06470957e-01
-5.63246310e-01 2.41214469e-01 2.98520446e-01 -1.56767115e-01
4.32046324e-01 -1.54863447e-01 9.60077122e-02 -4.17917401e-01
6.87618315e-01 6.93683863e-01 8.26371968e-01 -2.31821895e-01
6.46293223e-01 8.14652860e-01 2.24197805e-01 -8.39564085e-01
-1.16008472e+00 -7.39071965e-01 -7.42671728e-01 -2.08981819e-02
1.25637734e+00 -1.09193599e+00 -4.38158005e-01 4.11376119e-01
-1.09898961e+00 -5.66690385e-01 4.62487526e-02 1.63242742e-01
-9.98433709e-01 1.76341861e-01 -4.59925950e-01 -6.46028936e-01
-2.37524763e-01 -1.25326490e+00 1.43463993e+00 2.22078457e-01
-1.59843966e-01 -1.16792285e+00 -1.98520839e-01 7.51934171e-01
5.14461756e-01 2.65403509e-01 3.79245669e-01 -6.79798067e-01
-6.28936946e-01 -4.42603707e-01 -7.18340814e-01 1.81769252e-01
-2.52453685e-01 -1.40848413e-01 -1.27105582e+00 5.12393489e-02
-4.99656439e-01 -3.55333030e-01 1.33744192e+00 8.71582031e-02
1.14655662e+00 3.83826464e-01 -4.85831916e-01 2.62379408e-01
1.35959530e+00 -6.31664693e-02 6.72945857e-01 6.28066182e-01
7.68122911e-01 8.14850152e-01 9.24651623e-01 -2.16294959e-01
8.49115729e-01 6.83968067e-01 5.58674693e-01 -3.52301270e-01
-3.08040619e-01 8.28681961e-02 2.23698113e-02 1.43468320e-01
3.74280550e-02 -1.35867357e-01 -1.22204804e+00 9.55763638e-01
-2.38920379e+00 -8.08854103e-01 -5.27626812e-01 1.95639873e+00
1.25090584e-01 6.58182323e-01 2.38362923e-01 1.73012868e-01
4.07188237e-01 3.07148755e-01 -3.67567658e-01 -8.83397758e-01
3.27990919e-01 3.16978544e-01 9.38179553e-01 7.95643747e-01
-1.46903789e+00 1.52973926e+00 7.03502941e+00 6.36326075e-01
-8.89840961e-01 3.17156434e-01 5.73738158e-01 1.60294726e-01
2.43708998e-01 2.26487011e-01 -6.69914901e-01 1.92139775e-01
1.22275186e+00 2.62763023e-01 8.21491331e-02 1.04156280e+00
-3.38935554e-02 -5.99749207e-01 -7.86950827e-01 7.40163863e-01
4.23728349e-03 -1.10297120e+00 -4.38581169e-01 -1.82368681e-01
4.78546172e-01 5.09613812e-01 -3.10621709e-01 3.17345202e-01
6.30769789e-01 -1.08767319e+00 8.29892516e-01 1.16275921e-01
4.85918581e-01 -8.83509219e-01 8.40516090e-01 4.24953341e-01
-1.22966993e+00 4.56834547e-02 -2.43637994e-01 -2.01638937e-01
3.85120928e-01 4.91009742e-01 -6.74586892e-01 5.51268637e-01
6.43341005e-01 6.21745944e-01 -5.75312018e-01 1.00427210e+00
-5.15981972e-01 4.94954556e-01 -4.57694829e-01 1.42932221e-01
8.71121168e-01 1.20662153e-01 2.50418574e-01 1.47518826e+00
2.87200473e-02 -1.74887881e-01 3.76539081e-01 3.96424025e-01
2.55580038e-01 -2.14028075e-01 -6.67255998e-01 -4.98117087e-03
1.05997324e-01 1.26735866e+00 -1.31091177e+00 -6.54203057e-01
-5.72486758e-01 1.17865670e+00 3.99246216e-01 4.36664000e-02
-8.77757192e-01 -6.07154548e-01 6.53833330e-01 -1.94298461e-01
7.54487097e-01 -5.03877282e-01 -6.72890365e-01 -7.42715478e-01
-2.46474314e-02 -4.10967708e-01 2.47622102e-01 -6.15509987e-01
-6.94345832e-01 4.49262798e-01 -2.06687197e-01 -6.97344780e-01
-3.65090400e-01 -9.75072920e-01 -5.43234944e-01 6.58024073e-01
-1.98148882e+00 -1.26633537e+00 -3.88594061e-01 1.82385266e-01
7.28661716e-01 2.14543179e-01 7.58266330e-01 3.55644763e-01
-3.65084171e-01 1.42931104e-01 -2.22700834e-01 9.76076350e-02
5.42346597e-01 -1.58459115e+00 1.14579570e+00 1.01791501e+00
3.70507568e-01 2.65290849e-02 8.59629095e-01 -3.28513235e-01
-1.08076322e+00 -1.01310456e+00 1.04194570e+00 -5.74681282e-01
6.26222610e-01 -3.28527033e-01 -7.34841704e-01 7.52883971e-01
2.70009041e-01 1.40929967e-01 3.02068442e-01 1.77450284e-01
-3.08975697e-01 1.18934572e-01 -8.84150147e-01 3.91276926e-01
1.19247937e+00 -4.93456990e-01 -6.17613375e-01 5.56547582e-01
4.28939492e-01 -8.44212413e-01 -5.40664196e-01 3.52604002e-01
4.43434179e-01 -1.08067727e+00 9.32328403e-01 -2.48381466e-01
4.11879212e-01 -4.27261472e-01 -7.51975030e-02 -9.69207525e-01
-6.32793009e-02 -4.03200567e-01 -1.43969849e-01 7.92501688e-01
6.29885256e-01 -7.58184195e-01 9.39078629e-01 3.40627879e-01
-3.71747941e-01 -8.01580548e-01 -9.31193888e-01 -6.58496082e-01
3.57432961e-02 -6.98541284e-01 2.72130787e-01 3.82145494e-01
-3.12885135e-01 5.49523175e-01 -2.35917300e-01 1.89566195e-01
5.24570823e-01 2.96905816e-01 8.41095924e-01 -1.38205397e+00
-6.11946620e-02 -6.49132669e-01 -7.52631247e-01 -1.06530762e+00
3.56908888e-01 -6.90333605e-01 4.30156738e-01 -2.08611679e+00
-2.29361475e-01 -2.41472304e-01 -1.08375445e-01 6.97515130e-01
-1.82826966e-01 7.16346264e-01 1.57163873e-01 -6.59447089e-02
-9.98180568e-01 2.49383356e-02 1.04548979e+00 1.08359061e-01
2.84838974e-01 -6.97218180e-02 -8.29124391e-01 8.79553556e-01
9.57340717e-01 -2.47243926e-01 -4.42480654e-01 -6.29933953e-01
1.09590970e-01 -2.12734565e-01 5.69211125e-01 -1.36746454e+00
3.15522432e-01 2.75916994e-01 2.14932710e-02 -7.06282437e-01
4.14848119e-01 -7.18937159e-01 -4.35803145e-01 4.28979009e-01
-2.02605978e-01 -1.65331885e-01 3.47179204e-01 4.35794324e-01
-4.62497920e-01 -2.80810118e-01 9.14616942e-01 -2.58682966e-01
-1.42839026e+00 1.08538046e-01 -3.81578743e-01 1.78947553e-01
1.37265873e+00 -4.07613635e-01 -3.37506771e-01 -1.38070866e-01
-6.63220584e-01 6.22441053e-01 4.59231079e-01 5.53689063e-01
2.67548740e-01 -7.60481954e-01 -6.01390719e-01 -5.20614050e-02
1.94350317e-01 3.96345884e-01 -1.40389558e-02 7.29611278e-01
-9.96914566e-01 8.09217274e-01 -1.26047194e-01 -5.92493773e-01
-1.18522179e+00 2.77253836e-01 3.51974159e-01 -1.26325995e-01
-8.38348329e-01 9.71196771e-01 -1.91744700e-01 -5.01394749e-01
1.16826110e-01 -1.29676163e-01 -1.78533360e-01 2.14533899e-02
3.86504531e-01 3.66200775e-01 2.26460904e-01 -6.59545422e-01
-4.55418199e-01 7.18158007e-01 -1.98611334e-01 -2.77180016e-01
1.07708156e+00 -1.09145984e-01 1.69153810e-01 4.34951752e-01
1.02779090e+00 -4.22093004e-01 -1.52563715e+00 1.22783510e-02
3.48536044e-01 -2.85654992e-01 3.60076040e-01 -8.82966220e-01
-7.79929042e-01 1.04654729e+00 5.03946066e-01 3.17883193e-01
7.73924232e-01 1.65336102e-01 1.17900538e+00 5.04227757e-01
4.90944654e-01 -1.40292966e+00 -3.67320836e-01 7.96607673e-01
2.31417850e-01 -1.51969171e+00 7.36588910e-02 -6.35533452e-01
-7.75568366e-01 9.67288375e-01 4.06408608e-01 -4.87988532e-01
3.43773693e-01 2.71819592e-01 3.62184018e-01 -3.66964281e-01
-6.76037014e-01 -9.14099753e-01 6.29365146e-02 4.65470642e-01
4.00307238e-01 1.06870160e-01 -2.10312352e-01 -2.30346933e-01
-1.60900980e-01 4.16648090e-02 2.47589171e-01 1.16698706e+00
-8.26576412e-01 -1.12444592e+00 1.03206135e-01 3.57612252e-01
-6.51738465e-01 1.79827616e-01 -5.40522158e-01 9.59452510e-01
8.74438956e-02 1.00310433e+00 1.55520886e-01 -8.76640379e-02
4.20688838e-01 2.62706846e-01 4.03313100e-01 -7.27945805e-01
-4.16791767e-01 -3.24697614e-01 7.44616151e-01 -9.09616351e-01
-6.09520257e-01 -6.61268711e-01 -1.64173830e+00 -1.32291391e-01
-1.48693666e-01 -1.08978972e-01 9.92930055e-01 1.21977603e+00
3.14304262e-01 6.83890164e-01 1.59710333e-01 -9.38237548e-01
-1.57393798e-01 -6.47503138e-01 -2.44494513e-01 1.61987677e-01
3.00784141e-01 -7.00386226e-01 -9.67043936e-02 -1.28777429e-01] | [9.405842781066895, 0.12609069049358368] |
a9c24209-fcb4-4bc6-ae53-7cf4d39829d1 | multi-view-3d-reconstruction-with-transformer | 2103.12957 | null | https://arxiv.org/abs/2103.12957v1 | https://arxiv.org/pdf/2103.12957v1.pdf | Multi-view 3D Reconstruction with Transformer | Deep CNN-based methods have so far achieved the state of the art results in multi-view 3D object reconstruction. Despite the considerable progress, the two core modules of these methods - multi-view feature extraction and fusion, are usually investigated separately, and the object relations in different views are rarely explored. In this paper, inspired by the recent great success in self-attention-based Transformer models, we reformulate the multi-view 3D reconstruction as a sequence-to-sequence prediction problem and propose a new framework named 3D Volume Transformer (VolT) for such a task. Unlike previous CNN-based methods using a separate design, we unify the feature extraction and view fusion in a single Transformer network. A natural advantage of our design lies in the exploration of view-to-view relationships using self-attention among multiple unordered inputs. On ShapeNet - a large-scale 3D reconstruction benchmark dataset, our method achieves a new state-of-the-art accuracy in multi-view reconstruction with fewer parameters ($70\%$ less) than other CNN-based methods. Experimental results also suggest the strong scaling capability of our method. Our code will be made publicly available. | ['Rabab Ward', 'Z. Jane Wang', 'Septimiu Salcudean', 'Tianyang Shi', 'Zhengxia Zou', 'Xun Chen', 'Xinrui Cui', 'Dan Wang'] | 2021-03-24 | null | null | null | null | ['3d-object-reconstruction'] | ['computer-vision'] | [-1.56255871e-01 -1.49737805e-01 1.77240763e-02 -5.20772040e-01
-9.05216098e-01 -4.09934551e-01 5.81711888e-01 -4.61653531e-01
2.82089770e-01 2.81495005e-01 5.54636240e-01 -2.42520019e-01
-7.24119991e-02 -8.49389315e-01 -9.57727671e-01 -4.49048698e-01
2.80778110e-01 6.32923603e-01 2.99499750e-01 -3.65507305e-01
1.56464148e-02 7.17108250e-01 -1.57510912e+00 6.66470826e-01
4.13975865e-01 1.35844648e+00 2.94118464e-01 3.58610034e-01
-1.17745616e-01 6.15609050e-01 7.09394217e-02 -6.58384740e-01
3.56504112e-01 -2.72569537e-01 -8.15657675e-01 4.40742105e-01
7.31195867e-01 -6.30884945e-01 -5.22802532e-01 7.86100805e-01
8.41841817e-01 -3.36796761e-01 4.26724970e-01 -9.61277127e-01
-8.91922653e-01 3.84044111e-01 -5.87064385e-01 1.47653222e-01
3.57315481e-01 -9.58978608e-02 1.13119256e+00 -1.42713082e+00
7.62767851e-01 1.26875222e+00 7.67100573e-01 3.31126213e-01
-1.10090292e+00 -5.67220926e-01 3.18273634e-01 3.33697945e-01
-1.13680470e+00 -4.76240754e-01 1.03747678e+00 -3.90285283e-01
1.42973948e+00 -4.56090346e-02 1.06461382e+00 9.96102333e-01
3.47852439e-01 1.10875332e+00 1.04521334e+00 1.75574096e-03
-2.96882629e-01 -2.03428701e-01 -4.31793511e-01 1.03980243e+00
-1.45635024e-01 1.88553393e-01 -5.67908823e-01 1.28467798e-01
1.13601911e+00 3.32756877e-01 -3.32343608e-01 -9.91792917e-01
-1.51938152e+00 8.60292912e-01 6.29971445e-01 1.14534236e-01
-2.72814721e-01 1.07630745e-01 5.84542274e-01 2.78863102e-01
8.89128983e-01 9.78168622e-02 -5.84802270e-01 9.23741385e-02
-6.13841832e-01 3.04589629e-01 5.39975941e-01 1.24217153e+00
4.53001916e-01 9.82638299e-02 1.52342901e-01 7.60439277e-01
2.13965893e-01 2.13128626e-01 1.59535795e-01 -6.36029959e-01
6.29034698e-01 8.25341344e-01 -2.00464070e-01 -8.82123351e-01
-4.45833564e-01 -6.47194624e-01 -1.08361685e+00 7.59389400e-02
-9.79055613e-02 5.30924737e-01 -8.47571373e-01 1.30821860e+00
4.45491135e-01 -2.30253935e-01 -2.26842239e-01 9.15470958e-01
1.27543175e+00 3.84823173e-01 -4.58893716e-01 -1.25390753e-01
1.22479248e+00 -1.23439777e+00 -4.59056437e-01 1.05059266e-01
1.41025424e-01 -8.00595880e-01 7.07923889e-01 4.57692116e-01
-1.41683304e+00 -7.17138231e-01 -9.91086245e-01 -4.23263431e-01
-2.30100781e-01 -2.19393596e-01 7.49249160e-01 8.54390040e-02
-1.07152867e+00 6.53613269e-01 -7.49591291e-01 -3.61735135e-01
7.05412686e-01 2.85113722e-01 -6.64459586e-01 -2.72219956e-01
-7.85311639e-01 8.34740818e-01 7.19216652e-03 8.35826322e-02
-9.19188559e-01 -8.22217643e-01 -9.09666300e-01 1.78258382e-02
5.28872013e-01 -1.20914102e+00 1.15852857e+00 -4.62129295e-01
-1.29422295e+00 1.12624323e+00 -1.08249441e-01 -6.25860542e-02
5.64822853e-01 -2.05690652e-01 -2.12760195e-01 3.37966941e-02
1.19631819e-01 5.10980666e-01 7.72083342e-01 -1.57053626e+00
-5.03413856e-01 -6.91158891e-01 1.92034081e-01 3.02986026e-01
-2.75779907e-02 7.99480826e-03 -7.75242865e-01 -5.48734784e-01
5.57952642e-01 -6.69589341e-01 -1.44675165e-01 3.00992459e-01
-2.78352290e-01 -2.11083502e-01 7.89297879e-01 -4.27752405e-01
7.20599473e-01 -1.79274654e+00 6.44777596e-01 -3.66220593e-01
7.27695704e-01 -3.03213019e-02 3.91453989e-02 3.65059823e-01
-2.97403216e-01 -1.94899086e-02 -8.80470872e-02 -6.03683591e-01
-1.86656639e-01 2.10809469e-01 -3.11558127e-01 5.95479012e-01
-2.42452584e-02 1.29329884e+00 -8.13059509e-01 -4.63218302e-01
5.48031271e-01 5.81960618e-01 -7.07917690e-01 4.23097104e-01
-9.45459381e-02 5.28738678e-01 -3.86691689e-01 9.84893322e-01
6.80252135e-01 -7.94380605e-01 3.95316035e-02 -7.66063929e-01
3.89792910e-03 2.19491661e-01 -7.00302899e-01 2.23399639e+00
-5.98111808e-01 3.78267974e-01 -9.58624780e-02 -1.13506556e+00
9.38701987e-01 3.66323352e-01 8.82729590e-01 -6.46033585e-01
1.96528628e-01 -1.63142271e-02 -2.42952019e-01 -4.31735039e-01
5.38683891e-01 -3.82397473e-01 1.68496147e-01 4.25682962e-01
4.22644526e-01 -5.26107132e-01 -2.63213217e-01 6.40427172e-02
6.24815345e-01 6.89396560e-01 5.41619718e-01 4.13428247e-02
2.84216851e-01 -2.73595631e-01 4.07556683e-01 1.54150188e-01
-6.16626926e-02 1.05798054e+00 2.78980047e-01 -8.86607766e-01
-1.37150812e+00 -1.07296848e+00 -1.45424888e-01 6.81515157e-01
1.52008668e-01 -5.52366138e-01 -1.40845254e-01 -9.63258862e-01
1.41007736e-01 2.24175572e-01 -7.07291126e-01 2.94327766e-01
-7.01084733e-01 -4.08565879e-01 9.20288847e-04 9.01304126e-01
4.86671120e-01 -8.44431877e-01 -5.07160366e-01 2.30365977e-01
-2.25616336e-01 -1.46816158e+00 -4.46846336e-01 1.17896475e-01
-1.20165622e+00 -1.00528979e+00 -8.33806813e-01 -6.52503014e-01
5.32002926e-01 6.23916805e-01 1.48395956e+00 -1.26468450e-01
-1.20954670e-01 5.53164840e-01 -4.50323254e-01 -2.88694650e-01
-1.25462115e-01 -9.51977670e-02 5.19081689e-02 -1.26750752e-01
-9.52501129e-03 -1.02934289e+00 -6.87191665e-01 5.46953499e-01
-6.73672020e-01 4.06179875e-01 5.08439064e-01 1.07982612e+00
1.01341069e+00 -3.33445162e-01 2.83991992e-01 -7.48872757e-01
4.40985858e-02 -3.29027236e-01 -4.37773079e-01 3.37843806e-01
-5.35496116e-01 -1.45692065e-01 6.52495980e-01 -1.23040617e-01
-8.39759529e-01 1.71313211e-01 -3.62811267e-01 -1.18386197e+00
-1.19887225e-01 3.41356754e-01 -3.79007816e-01 -1.45316884e-01
1.39285237e-01 6.89135134e-01 5.89084141e-02 -6.91429496e-01
4.88166690e-01 2.00007811e-01 1.48722157e-01 -2.45492980e-01
6.15639329e-01 7.48120368e-01 2.38314774e-02 -5.32524765e-01
-1.11851466e+00 -2.92627573e-01 -9.08298612e-01 -3.65731597e-01
9.96480525e-01 -1.19048083e+00 -8.54735970e-01 3.99081886e-01
-1.21480870e+00 1.99782073e-01 -1.70863688e-01 2.90093541e-01
-1.06081653e+00 4.67331052e-01 -3.92877728e-01 -3.77010584e-01
-5.20165026e-01 -1.48844719e+00 1.47288764e+00 -2.59315550e-01
2.65789002e-01 -8.60370040e-01 -1.98061004e-01 6.84300125e-01
3.48716527e-01 2.48073190e-01 9.05557692e-01 -3.30694795e-01
-1.02541971e+00 7.98673853e-02 -5.10040104e-01 2.17229888e-01
1.43844336e-01 -3.87327075e-01 -9.41755056e-01 -4.15013343e-01
2.59431601e-01 -4.29192424e-01 9.19415772e-01 3.98189157e-01
1.54535592e+00 1.13390476e-01 -3.06757599e-01 1.09518886e+00
1.51389277e+00 3.52091342e-02 5.80195665e-01 7.30607063e-02
1.11822176e+00 2.38912046e-01 2.61921614e-01 3.61356854e-01
6.92568898e-01 8.43615294e-01 9.36400235e-01 -2.31858552e-01
-2.63226897e-01 -5.10541439e-01 3.33020114e-03 1.29105926e+00
-3.32909375e-01 -3.91432762e-01 -6.35414362e-01 5.90442955e-01
-1.76627970e+00 -9.87495899e-01 1.16637520e-01 1.77273285e+00
2.10452795e-01 8.95608142e-02 1.21532343e-01 -1.27132609e-01
3.02767724e-01 5.35188198e-01 -6.95976675e-01 -9.35848951e-02
-9.76585820e-02 8.46103653e-02 2.84575313e-01 1.46299571e-01
-1.08095944e+00 7.48144805e-01 6.28192997e+00 6.88532054e-01
-1.14909422e+00 1.86970100e-01 6.68311357e-01 -2.22696662e-01
-5.71651876e-01 -2.54074544e-01 -8.28542888e-01 -1.06324531e-01
1.85366288e-01 2.10925236e-01 2.38566220e-01 8.03435624e-01
-2.34797269e-01 2.76961893e-01 -1.30781436e+00 1.32559431e+00
5.05226195e-01 -1.79465199e+00 4.37582821e-01 2.53500760e-01
6.58121228e-01 3.72319430e-01 6.80280896e-03 7.28360862e-02
1.39714018e-01 -9.64103281e-01 9.04159784e-01 4.75422472e-01
1.01752400e+00 -7.31090605e-01 4.20140654e-01 2.05392480e-01
-1.67368639e+00 -3.47694270e-02 -3.87590230e-01 2.83063263e-01
5.83787203e-01 5.57944357e-01 -4.91119802e-01 1.17386675e+00
9.95246828e-01 1.37450516e+00 -2.47935772e-01 7.46138871e-01
1.49933964e-01 -7.67460559e-03 -2.13554092e-02 1.99847668e-01
1.66533992e-01 -2.40712836e-02 7.39825666e-01 6.65368676e-01
3.80527139e-01 1.54775769e-01 1.35511786e-01 9.63815272e-01
-9.38897654e-02 -3.69455032e-02 -9.67284441e-01 1.87649563e-01
-1.32212058e-01 1.16191721e+00 -7.37884283e-01 -3.04309160e-01
-8.55112135e-01 9.55829561e-01 6.79138243e-01 1.17685245e-02
-8.47814262e-01 2.50224978e-01 6.38254344e-01 3.78494799e-01
1.02491963e+00 -3.13393533e-01 -3.86268288e-01 -1.43775725e+00
1.94568008e-01 -7.90567338e-01 2.13218853e-01 -1.07544899e+00
-1.60739267e+00 9.96050119e-01 -4.67432896e-03 -1.81536126e+00
-3.27244289e-02 -6.47967100e-01 -1.77087143e-01 6.96227908e-01
-1.44496512e+00 -1.66335785e+00 -1.96117431e-01 6.55286610e-01
9.59655106e-01 -2.50870168e-01 7.89475262e-01 4.16569382e-01
-1.05189092e-01 4.17441487e-01 -1.27821967e-01 4.81891781e-02
4.13066208e-01 -1.04322517e+00 6.93518400e-01 4.65527058e-01
2.43022978e-01 3.24821591e-01 1.08619235e-01 -5.89722395e-01
-1.87124252e+00 -9.88977015e-01 8.15229237e-01 -6.22876763e-01
4.65346783e-01 -5.11946142e-01 -6.20833576e-01 8.29842389e-01
3.31713319e-01 3.41517478e-01 5.68399310e-01 1.87915370e-01
-5.89491010e-01 -1.16917759e-01 -9.14090216e-01 3.29517156e-01
1.71192765e+00 -6.70384169e-01 -5.63375711e-01 2.35004038e-01
1.01585758e+00 -8.49926174e-01 -1.26242566e+00 7.10351110e-01
7.75097549e-01 -1.45736897e+00 1.45554674e+00 -6.78658605e-01
9.81187344e-01 -4.73434776e-02 -5.86173058e-01 -1.25834572e+00
-6.64372206e-01 -1.90054283e-01 -2.97515035e-01 7.64062643e-01
1.27579466e-01 -4.44941431e-01 6.07790232e-01 9.52833146e-02
-5.54211199e-01 -1.55115902e+00 -1.09151590e+00 -4.85988826e-01
-5.26927132e-03 -4.32605237e-01 8.15754294e-01 8.13938558e-01
-3.82259965e-01 5.16149402e-01 -6.35508537e-01 -1.34409359e-02
6.69637144e-01 8.84303093e-01 8.84381592e-01 -1.05008912e+00
-3.57108146e-01 -4.38075304e-01 -4.68422264e-01 -1.45940876e+00
-8.66969898e-02 -1.06582630e+00 -2.51534730e-01 -1.73037076e+00
4.99823779e-01 -1.46202058e-01 -3.23498696e-02 2.51748532e-01
1.63173407e-01 3.56662124e-01 3.73713791e-01 6.24954365e-02
-6.77351117e-01 1.04015720e+00 1.91092801e+00 -3.49506736e-02
1.76700071e-01 -2.17412934e-02 -6.99318647e-01 6.78794026e-01
1.66030124e-01 -1.88992903e-01 -2.98546374e-01 -7.61769831e-01
2.58364975e-01 5.95474601e-01 4.42600101e-01 -6.62013173e-01
5.39499447e-02 9.56972167e-02 6.43071592e-01 -1.42156601e+00
7.08429277e-01 -9.76023614e-01 2.57495970e-01 1.58593476e-01
3.07544321e-02 5.61088800e-01 4.93320040e-02 7.01487541e-01
-2.85305440e-01 5.22829056e-01 6.94153786e-01 -7.38091350e-01
-6.82784677e-01 8.83496165e-01 1.50376618e-01 -1.07212722e-01
7.42003202e-01 -4.25806671e-01 -7.82076120e-02 -4.65856820e-01
-8.46522689e-01 7.54518211e-02 4.66669858e-01 6.40867174e-01
1.10546291e+00 -1.52058172e+00 -7.38309383e-01 4.26214099e-01
3.25634658e-01 3.27375025e-01 5.86965203e-01 8.27856779e-01
-1.80653900e-01 4.25404519e-01 -3.30066025e-01 -9.74674702e-01
-1.17952383e+00 8.21101308e-01 4.30833966e-01 -3.99484307e-01
-1.08442056e+00 8.81910443e-01 5.06795645e-01 -6.15190268e-01
6.72304705e-02 -2.73682147e-01 -9.49936062e-02 -6.57340232e-03
2.09798351e-01 8.63956362e-02 1.91926658e-01 -7.88535893e-01
-2.91493118e-01 1.03316247e+00 -1.18767612e-01 2.53868669e-01
1.93757927e+00 -1.49859294e-01 -1.39411002e-01 4.73997325e-01
1.47745156e+00 -3.43758076e-01 -1.09208524e+00 -4.38645422e-01
-8.46322298e-01 -6.62424445e-01 1.26310587e-01 -6.34037733e-01
-1.54960227e+00 1.02011311e+00 4.78416890e-01 1.11034915e-01
1.24702275e+00 3.60727489e-01 1.06949151e+00 6.18766434e-02
5.85012138e-01 -4.57360923e-01 1.77364185e-01 7.08264410e-01
1.43310153e+00 -1.44884014e+00 3.71766329e-01 -4.46379870e-01
-5.46730876e-01 1.02932847e+00 8.24703395e-01 -1.59042388e-01
1.00157535e+00 1.49130538e-01 -2.45553359e-01 -6.89345002e-01
-7.84028769e-01 -1.29786104e-01 5.31820118e-01 4.89783496e-01
5.17861545e-01 -1.04641311e-01 1.46446005e-01 4.53203380e-01
-1.28024712e-01 -7.28778392e-02 1.39253184e-01 7.19258904e-01
-2.28738673e-02 -9.53042626e-01 -6.51918119e-03 5.55186689e-01
-3.40659857e-01 -7.13823810e-02 -2.74991959e-01 7.87645459e-01
-8.69704783e-02 3.82492989e-01 1.76320955e-01 -7.14224219e-01
5.99476635e-01 -1.89151436e-01 8.39823723e-01 -3.78007561e-01
-7.45047688e-01 3.08549732e-01 -1.40116923e-03 -9.68927801e-01
-7.93188155e-01 -5.55244207e-01 -7.05521107e-01 -3.98739219e-01
-2.59028226e-01 -5.54184616e-01 5.18222570e-01 8.92255783e-01
6.03185356e-01 6.31694674e-01 7.91153014e-01 -1.27513587e+00
-4.55502212e-01 -6.74813032e-01 -3.72276127e-01 2.78232872e-01
3.42019439e-01 -1.01235151e+00 1.16348885e-01 -2.85227478e-01] | [8.273987770080566, -3.551143169403076] |
0e7b6ed9-f112-4c3e-9007-f898751dbd3a | pseudocell-hard-negative-mining-as-pseudo | 2307.03211 | null | https://arxiv.org/abs/2307.03211v1 | https://arxiv.org/pdf/2307.03211v1.pdf | PseudoCell: Hard Negative Mining as Pseudo Labeling for Deep Learning-Based Centroblast Cell Detection | Patch classification models based on deep learning have been utilized in whole-slide images (WSI) of H&E-stained tissue samples to assist pathologists in grading follicular lymphoma patients. However, these approaches still require pathologists to manually identify centroblast cells and provide refined labels for optimal performance. To address this, we propose PseudoCell, an object detection framework to automate centroblast detection in WSI (source code is available at https://github.com/IoBT-VISTEC/PseudoCell.git). This framework incorporates centroblast labels from pathologists and combines them with pseudo-negative labels obtained from undersampled false-positive predictions using the cell's morphological features. By employing PseudoCell, pathologists' workload can be reduced as it accurately narrows down the areas requiring their attention during examining tissue. Depending on the confidence threshold, PseudoCell can eliminate 58.18-99.35% of non-centroblasts tissue areas on WSI. This study presents a practical centroblast prescreening method that does not require pathologists' refined labels for improvement. Detailed guidance on the practical implementation of PseudoCell is provided in the discussion section. | ['Theerawit Wilaiprasitporn', 'Chanitra Thuwajit', 'Sumeth Yuenyong', 'Narit Hnoohom', 'Napat Angkathunyakul', 'Ananya Pongpaibul', 'Komgrid Charngkaew', 'Phoomraphee Luenam', 'Kanyakorn Veerakanjana', 'Supasan Sripodok', 'Paisarn Boonsakan', 'Peti Thuwajit', 'Phattarapong Sawangjai', 'Thapanun Sudhawiyangkul', 'Piyalitt Ittichaiwong', 'Narongrid Seesawad'] | 2023-07-06 | null | null | null | null | ['whole-slide-images', 'object-detection', 'cell-detection'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 2.77789440e-02 1.69634536e-01 -1.98102817e-01 -1.41220316e-01
-1.24668550e+00 -6.35039568e-01 -3.27449664e-02 6.22781694e-01
-3.95330131e-01 6.52428627e-01 -2.39318192e-01 -3.75049919e-01
1.00325033e-01 -9.43375289e-01 -1.84196427e-01 -1.26064074e+00
5.85751295e-01 8.68849456e-01 3.21623087e-01 7.63739049e-02
4.12367344e-01 8.20102870e-01 -9.87359464e-01 6.75170660e-01
8.53766918e-01 6.43575072e-01 2.93424249e-01 9.93168473e-01
-2.77819365e-01 3.56541306e-01 -6.70595229e-01 -1.34389639e-01
2.46797339e-03 -1.58044085e-01 -6.94216490e-01 -1.75072160e-02
4.07556802e-01 -2.02751622e-01 -1.17613450e-01 1.00199711e+00
5.51978052e-01 -4.55603957e-01 1.00636530e+00 -1.08340669e+00
-1.60936832e-01 5.09964526e-01 -8.20575893e-01 5.07529676e-01
-1.83299437e-01 4.57922459e-01 9.40656841e-01 -8.76338780e-01
7.93228626e-01 6.82732940e-01 8.42214286e-01 3.92969698e-01
-1.25207150e+00 -7.97538817e-01 -6.62476003e-01 3.38815212e-01
-1.90009868e+00 -1.92766666e-01 2.91873813e-01 -5.63456416e-01
8.76616597e-01 4.85059857e-01 7.42812753e-01 1.91788107e-01
4.38669741e-01 9.33811069e-01 8.68644536e-01 -5.76311350e-01
7.69779682e-02 4.54502016e-01 5.12027204e-01 6.09870970e-01
3.83171946e-01 -2.97874629e-01 -1.11741386e-01 -2.31429320e-02
6.13904536e-01 3.78640890e-02 -2.79256463e-01 5.47763258e-02
-7.60514557e-01 6.86834216e-01 5.46731651e-01 7.33263612e-01
-7.08984062e-02 -9.02414471e-02 4.97236103e-01 -9.17645097e-02
4.12093014e-01 2.98862934e-01 8.59281607e-03 3.55632693e-01
-1.13819075e+00 -1.26975641e-01 1.57218307e-01 5.10002136e-01
6.84708834e-01 -4.48287100e-01 -2.71470517e-01 1.02407455e+00
-8.60564504e-03 3.27327073e-01 4.41509962e-01 -5.95125258e-01
-4.79584843e-01 1.12217796e+00 -1.99986324e-01 -4.64535773e-01
-7.71688342e-01 -6.53941691e-01 -9.63561535e-01 3.36271763e-01
7.42982507e-01 2.79511273e-01 -9.94520128e-01 1.03726745e+00
4.64062840e-01 8.48385245e-02 -5.13944983e-01 1.01435459e+00
8.00683141e-01 2.90460408e-01 5.41539155e-02 1.05593614e-01
1.64583170e+00 -6.39640987e-01 -5.62778115e-01 2.44183943e-01
1.10520816e+00 -6.62586510e-01 1.15230477e+00 5.36013365e-01
-9.12325680e-01 -2.55539954e-01 -1.01672018e+00 -2.00228631e-01
-6.10281944e-01 6.43439829e-01 5.43744445e-01 6.74531281e-01
-1.40031981e+00 2.24149376e-01 -9.68761563e-01 -6.20860994e-01
8.54993165e-01 7.07257390e-01 -3.91385019e-01 1.90915987e-01
-6.61451638e-01 7.26120591e-01 4.07047302e-01 -2.58457381e-03
-6.31659269e-01 -9.98704553e-01 -4.85961556e-01 7.69795254e-02
1.60638958e-01 -5.00061095e-01 1.16435027e+00 -6.59772575e-01
-1.14373732e+00 1.34881651e+00 -2.92313099e-01 -1.25973254e-01
2.98656315e-01 5.43584645e-01 -7.99728334e-02 6.01854265e-01
-3.22854035e-02 8.77762556e-01 2.45539159e-01 -1.01813376e+00
-1.11084032e+00 -4.78335053e-01 -2.96425045e-01 -7.10135931e-03
-1.20301768e-01 -4.53186363e-01 -6.96862936e-01 -3.87880653e-01
4.79555801e-02 -9.39926863e-01 9.66227334e-03 3.25985491e-01
-5.80141068e-01 -4.82466400e-01 6.64390385e-01 -5.23621202e-01
1.12330222e+00 -2.14304447e+00 -4.86193717e-01 4.71304446e-01
4.21451420e-01 5.73836565e-01 2.90275551e-02 -1.13857649e-01
2.50019412e-02 2.43893951e-01 1.02510378e-01 -2.02167466e-01
-9.86276641e-02 -2.15347230e-01 3.07786882e-01 7.96640456e-01
2.25284263e-01 9.36917186e-01 -7.24899411e-01 -8.39416444e-01
4.56069440e-01 4.69610691e-01 -4.31049645e-01 3.89296524e-02
1.23671636e-01 1.35402828e-01 1.03242815e-01 1.06198895e+00
1.01934278e+00 -6.29918218e-01 2.88724571e-01 -2.97878921e-01
3.46319884e-01 1.83584057e-02 -8.65936756e-01 1.16634142e+00
-1.59700766e-01 8.15265059e-01 3.86117339e-01 -6.69479549e-01
5.62593639e-01 2.13694394e-01 3.62681478e-01 -1.98856071e-01
3.02610576e-01 3.48861992e-01 1.51555007e-02 -4.49488521e-01
3.38725269e-01 -3.90135348e-01 3.23826134e-01 5.00163257e-01
-5.49504943e-02 -7.51465261e-02 4.44252253e-01 2.27738827e-01
1.16277158e+00 -3.86128545e-01 4.76986349e-01 -3.61793846e-01
9.44632351e-01 3.85749727e-01 7.35639632e-01 6.07001424e-01
-4.94624823e-01 6.69892073e-01 4.80054975e-01 -3.85484040e-01
-8.48534703e-01 -1.03360546e+00 -5.01663804e-01 1.00400317e+00
1.95387751e-02 1.27660036e-01 -6.71530664e-01 -8.92262518e-01
2.04239547e-01 5.15620232e-01 -8.47868741e-01 1.22348137e-01
-4.82980222e-01 -1.06701672e+00 6.36943281e-01 4.91172731e-01
-7.12621957e-02 -9.75580812e-01 -2.77330279e-01 2.55335905e-02
-1.97069779e-01 -3.97663534e-01 -3.70722443e-01 4.26618993e-01
-5.85682809e-01 -1.40560615e+00 -6.79079354e-01 -8.11486244e-01
1.38698971e+00 4.28517878e-01 8.44527364e-01 5.93373239e-01
-1.01520860e+00 -6.18574210e-02 -2.75770545e-01 -3.42366010e-01
-6.80349648e-01 3.14877301e-01 -3.85357827e-01 -2.93245137e-01
1.10876262e+00 -1.05519164e-02 -8.17706764e-01 3.36345434e-01
-1.05635643e+00 2.73472607e-01 8.86509120e-01 1.13625789e+00
9.71886337e-01 1.22282170e-01 6.64003372e-01 -1.20334303e+00
7.94439986e-02 -3.59613746e-01 -5.47103524e-01 3.31476867e-01
-3.85420144e-01 -3.43248755e-01 5.23196518e-01 -2.58829445e-01
-1.03753984e+00 -2.26278320e-01 -4.12284166e-01 -3.00775338e-02
-5.27545631e-01 1.79801628e-01 2.38104135e-01 -1.99468940e-01
6.27117753e-01 1.18371777e-01 1.83857590e-01 -4.18814458e-02
-3.00044268e-01 9.70006824e-01 4.49263275e-01 7.36017227e-02
2.69075900e-01 9.49107349e-01 -4.33514677e-02 -5.96981168e-01
-5.13433993e-01 -9.84155595e-01 -4.83866990e-01 -3.66586417e-01
4.50091422e-01 -5.15600085e-01 -1.02132034e+00 4.18303132e-01
-5.97058773e-01 -5.21295607e-01 -1.42443374e-01 4.91342247e-01
-2.01275513e-01 2.32357025e-01 -1.26030231e+00 -4.62807417e-01
-7.73684382e-01 -1.11049569e+00 1.21053064e+00 6.32011175e-01
-5.09977698e-01 -1.07452095e+00 1.18182823e-01 5.28924525e-01
3.90070528e-01 -5.84299080e-02 1.12782550e+00 -8.52513731e-01
-3.98688614e-01 -4.66287255e-01 -3.38329852e-01 -1.88011657e-02
2.74459094e-01 3.82305622e-01 -1.10730636e+00 -4.73878562e-01
-4.50976372e-01 -3.01901460e-01 1.03443766e+00 7.40094066e-01
1.18137908e+00 1.67922288e-01 -1.00239527e+00 9.27000165e-01
1.58489048e+00 1.80943131e-01 4.00308818e-01 4.04597074e-01
2.52954036e-01 6.83931828e-01 6.60680115e-01 3.52180034e-01
2.07344934e-01 1.77752912e-01 3.16528797e-01 -5.73428988e-01
-3.67420673e-01 8.24372619e-02 -2.73881316e-01 3.04779410e-01
4.00078177e-01 -3.76193404e-01 -1.17811525e+00 8.95118415e-01
-1.21553123e+00 -5.89391947e-01 -2.09044293e-01 1.81066966e+00
1.00826740e+00 9.26820189e-02 -4.77831401e-02 4.04955864e-01
1.06490135e+00 -5.76412559e-01 -6.52236938e-01 -7.13844001e-02
2.43027862e-02 2.20552683e-01 4.00596142e-01 7.02778995e-01
-8.59065235e-01 7.96527267e-01 5.67945242e+00 1.35949886e+00
-1.47649407e+00 -4.17329408e-02 1.28142333e+00 -3.07057947e-01
-7.34784007e-02 -3.56306314e-01 -1.19603002e+00 3.12543035e-01
2.39401191e-01 -4.40260880e-02 -1.90870360e-01 4.80648756e-01
1.71709329e-01 -7.18698382e-01 -9.96722698e-01 7.69472003e-01
-1.85115393e-02 -1.67947662e+00 -4.92270887e-02 1.80896819e-01
4.23076659e-01 -1.65127963e-01 -5.67660592e-02 3.56066138e-01
1.83698907e-01 -8.58020544e-01 3.22454534e-02 4.56296891e-01
1.16884780e+00 -6.57405972e-01 1.42133152e+00 1.29790336e-01
-9.05413210e-01 2.69895971e-01 -5.11319518e-01 5.53905964e-01
-2.25298181e-01 8.19820464e-01 -2.01052308e+00 1.44524455e-01
4.84709680e-01 2.52006590e-01 -7.56602407e-01 1.19637740e+00
-7.95328915e-02 7.86510527e-01 -5.10577917e-01 -1.20572262e-01
-1.83992535e-02 4.88197543e-02 2.07143277e-01 1.65291989e+00
3.79492730e-01 9.60112438e-02 -4.33743775e-01 8.17445934e-01
1.87382460e-01 2.91571766e-01 1.94848225e-01 1.57300442e-01
6.35183811e-01 1.84405506e+00 -1.40472305e+00 -4.03138489e-01
-1.24380708e-01 3.66541058e-01 2.90198177e-01 2.44949892e-01
-6.33876324e-01 -5.48992693e-01 2.46976510e-01 4.54797387e-01
3.34527604e-02 6.87424183e-01 -7.27918983e-01 -6.14978075e-01
-6.39855921e-01 -5.83750784e-01 8.72801602e-01 -6.63683116e-01
-1.26492381e+00 4.21835363e-01 -4.66925740e-01 -1.18052995e+00
2.04403892e-01 -8.07263732e-01 -8.11949074e-01 6.81740344e-01
-1.43652225e+00 -1.33765090e+00 -5.45940995e-01 1.49143994e-01
4.14698541e-01 9.94474217e-02 6.44564509e-01 -8.93149711e-03
-6.30031288e-01 9.81985211e-01 1.46133050e-01 1.08214758e-01
1.03347135e+00 -1.57033515e+00 -2.29375824e-01 2.72979736e-01
-6.11158192e-01 4.91263539e-01 5.05330384e-01 -5.01586497e-01
-9.26363885e-01 -1.03898358e+00 6.84795141e-01 -5.23943342e-02
5.29295325e-01 -1.35061219e-01 -1.00101304e+00 5.26446164e-01
2.28582054e-01 6.12560436e-02 1.54090977e+00 -1.47845566e-01
2.43666276e-01 -2.71209002e-01 -1.47708178e+00 5.68250835e-01
1.24544516e-01 -2.05408186e-01 -1.69034794e-01 3.86601537e-01
-1.47318259e-01 -5.10769963e-01 -9.07866240e-01 2.53757149e-01
4.33551490e-01 -8.20010245e-01 5.11051714e-01 1.50885239e-01
-1.05037935e-01 -4.93988037e-01 5.37697256e-01 -1.07149589e+00
-6.62442565e-01 -8.16387311e-02 3.52067649e-01 9.84636128e-01
5.13118625e-01 -6.86016500e-01 1.37876058e+00 1.85365066e-01
-3.48124295e-01 -1.04230154e+00 -8.52204621e-01 -1.54074758e-01
2.95886159e-01 -5.72437271e-02 4.44474190e-01 7.59321988e-01
4.82155204e-01 -2.56110221e-01 4.67829317e-01 2.86706295e-02
4.18105394e-01 -3.31591927e-02 7.13783026e-01 -9.49392855e-01
-2.39950567e-01 -9.27241862e-01 -3.89203995e-01 -5.38491845e-01
-2.10428372e-01 -1.23675239e+00 -9.34344158e-02 -1.50794280e+00
6.50795043e-01 -6.26627862e-01 -4.18243587e-01 7.95501709e-01
-5.46059251e-01 7.84552872e-01 -1.35935828e-01 3.22796822e-01
-5.38636923e-01 -1.83003575e-01 1.29633808e+00 -3.89173001e-01
-3.31061669e-02 -1.00734323e-01 -8.64915907e-01 6.56595230e-01
1.04205728e+00 -4.15705174e-01 6.78906264e-03 1.08382933e-01
7.97614902e-02 -2.98116338e-02 1.03007875e-01 -9.98140454e-01
4.54169422e-01 -1.87315315e-01 8.80788028e-01 -1.38894999e+00
1.49672017e-01 -4.24354047e-01 1.72757685e-01 6.97514117e-01
-2.79804766e-01 -6.22168660e-01 4.22108531e-01 5.58302552e-02
-1.49065778e-01 -4.69876885e-01 1.11817670e+00 -2.63748951e-02
-1.39157668e-01 1.17998697e-01 -8.36490154e-01 -5.57319283e-01
1.33438671e+00 -5.75534999e-01 -8.11219573e-01 -6.42930791e-02
-8.24877799e-01 4.59090412e-01 7.66785622e-01 -4.78158981e-01
4.13374782e-01 -9.78039861e-01 -8.26059043e-01 2.73861468e-01
2.36622378e-01 2.35497385e-01 8.19110036e-01 1.19345391e+00
-1.24843872e+00 7.72628188e-01 -2.23603562e-01 -8.76474619e-01
-1.52639127e+00 3.46724033e-01 5.87508500e-01 -6.82079971e-01
-2.96818435e-01 1.35571182e+00 6.13375545e-01 -3.93881261e-01
1.32506326e-01 -3.71270090e-01 -2.24840567e-01 1.49776237e-02
6.79456413e-01 4.45278615e-01 3.69787872e-01 -4.03580576e-01
-4.09622520e-01 2.32990384e-01 -7.56634593e-01 2.93411285e-01
9.38123047e-01 -5.09341657e-02 -2.47821897e-01 2.13774040e-01
9.81738210e-01 1.26133963e-01 -1.10726237e+00 -1.95850611e-01
-1.65906355e-01 -4.06032413e-01 1.99270189e-01 -9.49744165e-01
-1.17965913e+00 8.09805751e-01 7.96116114e-01 4.22160700e-02
1.33507967e+00 1.29893497e-01 6.47524059e-01 -6.74812198e-02
1.84486419e-01 -1.18892050e+00 -5.68905920e-02 1.23404421e-01
4.08269852e-01 -1.30152297e+00 3.00032124e-02 -5.53446651e-01
-3.19791049e-01 1.30326927e+00 9.01544988e-01 -4.13943790e-02
4.39326614e-01 6.78337693e-01 4.07539248e-01 -1.54091612e-01
-9.30939376e-01 -5.17631099e-02 -1.12071581e-01 7.04147279e-01
6.69412911e-01 2.69514680e-01 -4.97739494e-01 6.41808450e-01
3.13375182e-02 1.09076604e-01 6.49009407e-01 8.32405388e-01
-6.78996980e-01 -8.35905015e-01 -8.07666957e-01 8.37028086e-01
-5.67345023e-01 1.41811520e-01 -4.21514630e-01 8.24836552e-01
2.92298853e-01 6.27557516e-01 4.80052263e-01 -5.40987663e-02
-3.12492639e-01 -2.20519900e-01 2.41272137e-01 -6.70677662e-01
-7.72906601e-01 5.80537498e-01 -3.12926680e-01 3.10732592e-02
2.69804318e-02 -4.33115929e-01 -1.50432265e+00 -5.07946551e-01
-6.83687270e-01 2.38688305e-01 2.85664797e-01 5.97344398e-01
1.83171868e-01 6.04396522e-01 3.53117406e-01 -5.40890694e-01
-7.27646723e-02 -1.26625419e+00 -8.97560179e-01 2.30088130e-01
2.86211461e-01 -4.91853297e-01 -5.63481510e-01 1.56804204e-01] | [15.125106811523438, -3.119310140609741] |
a88c27eb-9c13-45eb-93d3-7e92185866d7 | align-mlm-word-embedding-alignment-is-crucial | 2211.08547 | null | https://arxiv.org/abs/2211.08547v1 | https://arxiv.org/pdf/2211.08547v1.pdf | ALIGN-MLM: Word Embedding Alignment is Crucial for Multilingual Pre-training | Multilingual pre-trained models exhibit zero-shot cross-lingual transfer, where a model fine-tuned on a source language achieves surprisingly good performance on a target language. While studies have attempted to understand transfer, they focus only on MLM, and the large number of differences between natural languages makes it hard to disentangle the importance of different properties. In this work, we specifically highlight the importance of word embedding alignment by proposing a pre-training objective (ALIGN-MLM) whose auxiliary loss guides similar words in different languages to have similar word embeddings. ALIGN-MLM either outperforms or matches three widely adopted objectives (MLM, XLM, DICT-MLM) when we evaluate transfer between pairs of natural languages and their counterparts created by systematically modifying specific properties like the script. In particular, ALIGN-MLM outperforms XLM and MLM by 35 and 30 F1 points on POS-tagging for transfer between languages that differ both in their script and word order (left-to-right v.s. right-to-left). We also show a strong correlation between alignment and transfer for all objectives (e.g., rho=0.727 for XNLI), which together with ALIGN-MLM's strong performance calls for explicitly aligning word embeddings for multilingual models. | ['Karthik Narasimhan', 'Ameet Deshpande', 'Henry Tang'] | 2022-11-15 | null | null | null | null | ['zero-shot-cross-lingual-transfer', 'cross-lingual-transfer'] | ['natural-language-processing', 'natural-language-processing'] | [-3.54966475e-03 -2.95682345e-02 -4.31867480e-01 -4.09323037e-01
-9.68883932e-01 -9.14492846e-01 1.00420666e+00 3.52684200e-01
-9.09526646e-01 6.25912249e-01 5.35773933e-01 -7.14109182e-01
7.28461817e-02 -5.45600235e-01 -8.75670135e-01 -3.50224674e-01
6.09937087e-02 3.97058189e-01 -3.64107601e-02 -3.90523314e-01
7.03807771e-02 1.89535514e-01 -9.13596570e-01 9.63080116e-03
1.05695689e+00 1.37530401e-01 4.56630200e-01 4.53198493e-01
-2.50614345e-01 1.55935913e-01 -2.27012381e-01 -7.72012830e-01
1.19902313e-01 -5.82813442e-01 -6.23907506e-01 -5.96135139e-01
9.90093768e-01 -6.36787042e-02 -2.11432308e-01 9.13954794e-01
5.58191538e-01 -1.05477674e-02 9.61419046e-01 -9.07726943e-01
-1.08937585e+00 8.25426400e-01 -4.00421292e-01 3.37630898e-01
1.64467812e-01 2.67573506e-01 1.64832664e+00 -8.67441893e-01
7.40783334e-01 1.39145577e+00 8.74739945e-01 5.68241477e-01
-1.71738231e+00 -4.97540087e-01 2.36307621e-01 -8.93380120e-02
-1.21597695e+00 -4.84026164e-01 2.27060333e-01 -6.47290349e-01
1.12192702e+00 -5.19482195e-02 3.41154575e-01 1.25251293e+00
4.38966095e-01 6.89680278e-01 1.28219724e+00 -7.05922127e-01
-1.88680425e-01 4.50030059e-01 1.88925534e-01 4.70854789e-01
4.56821144e-01 3.35134387e-01 -6.87778115e-01 2.04113126e-01
4.58894104e-01 -4.84827071e-01 -2.99025416e-01 -2.87722737e-01
-1.37698412e+00 8.67841065e-01 1.32158190e-01 7.40590453e-01
-4.82795760e-03 1.99147444e-02 4.70616788e-01 6.41248584e-01
2.72373438e-01 8.35681915e-01 -6.22752607e-01 -2.37216637e-01
-7.06866086e-01 5.37516102e-02 5.21560788e-01 9.51763690e-01
9.23647881e-01 1.17709428e-01 -1.70515999e-01 9.73827422e-01
2.11305782e-01 4.85105604e-01 8.77695739e-01 -3.37849051e-01
6.56785190e-01 2.00421721e-01 -2.59923160e-01 -6.04968190e-01
-9.64917168e-02 -3.85945857e-01 -3.43349755e-01 -6.34921119e-02
5.08541405e-01 -1.34747654e-01 -5.60135901e-01 2.21070957e+00
-7.33764991e-02 -2.44135067e-01 2.60471165e-01 4.70039427e-01
4.64986920e-01 7.33538389e-01 5.25951028e-01 2.24103585e-01
1.34085381e+00 -8.83749485e-01 -4.02790278e-01 -6.96394265e-01
1.09099960e+00 -9.30607677e-01 1.81608081e+00 -1.89918548e-01
-1.01506555e+00 -6.64174199e-01 -1.10951078e+00 -1.74651697e-01
-5.38802087e-01 -5.53063639e-02 3.42101127e-01 6.24599993e-01
-1.14999986e+00 7.09330797e-01 -5.81855774e-01 -7.84844041e-01
-9.20242630e-03 2.08050489e-01 -7.58735299e-01 -2.31318802e-01
-1.38568342e+00 1.59610617e+00 3.55450988e-01 -5.00934005e-01
-7.81268239e-01 -1.06696999e+00 -1.10659063e+00 -4.81391884e-02
-4.50108647e-01 -3.57113093e-01 1.17289615e+00 -8.18511546e-01
-1.47496223e+00 1.24467516e+00 -5.58067411e-02 -2.27819949e-01
2.39430353e-01 -3.88570875e-01 -4.39449877e-01 -4.25998688e-01
3.34607303e-01 8.62566710e-01 3.85291457e-01 -1.04840434e+00
-6.12910807e-01 -1.90523520e-01 -1.72865987e-02 4.06348139e-01
-7.89737642e-01 2.42978334e-01 -1.93715587e-01 -7.17684627e-01
-4.41760153e-01 -8.32142293e-01 1.37857236e-02 -3.41458440e-01
-6.23746552e-02 -4.57868665e-01 3.27742040e-01 -9.48791504e-01
1.17170644e+00 -2.07641935e+00 1.92058817e-01 -1.67486534e-01
-3.05514812e-01 3.20200115e-01 -7.89299369e-01 7.67573953e-01
-1.94603130e-01 3.09892237e-01 -1.33349895e-01 -3.87421638e-01
3.36199969e-01 2.69764751e-01 -2.55943447e-01 5.04538476e-01
4.70387638e-01 1.05932248e+00 -9.35920238e-01 -3.94450516e-01
6.85158223e-02 4.72569287e-01 -8.08854163e-01 2.73233831e-01
8.35885033e-02 4.00329888e-01 2.81577617e-01 2.51520276e-01
2.58472949e-01 2.18111694e-01 4.14529681e-01 -5.36570624e-02
-3.07144731e-01 8.31671119e-01 -5.05838573e-01 1.72065890e+00
-1.03476655e+00 7.91499376e-01 -1.59637868e-01 -7.30494618e-01
7.13737547e-01 2.50187844e-01 -8.89593735e-02 -8.64211380e-01
-8.97766352e-02 5.93106627e-01 6.39171183e-01 -2.29724959e-01
5.32570064e-01 -5.19140422e-01 -3.32853526e-01 6.05751812e-01
5.80833912e-01 -2.52602369e-01 1.37789547e-01 1.64282080e-02
8.61524820e-01 1.92418799e-01 4.08028573e-01 -8.11038315e-01
2.64639050e-01 -1.71571478e-01 3.84778023e-01 5.91239989e-01
-1.54300749e-01 2.03971311e-01 4.03583437e-01 -6.26792088e-02
-1.18646193e+00 -1.44321299e+00 -3.81283522e-01 1.59071732e+00
-1.50149018e-01 -3.03847313e-01 -6.51688814e-01 -7.73331702e-01
2.35569388e-01 1.29967368e+00 -6.45205736e-01 -3.55984539e-01
-8.31972718e-01 -4.88190562e-01 6.69486642e-01 4.64665949e-01
-2.90631354e-01 -9.99510825e-01 -5.27339540e-02 1.04849182e-01
7.13324174e-02 -1.01328158e+00 -8.94845903e-01 2.40156084e-01
-6.75285459e-01 -5.78042805e-01 -6.10429108e-01 -1.01640415e+00
5.62265396e-01 2.37564333e-02 1.42869949e+00 -1.66368291e-01
9.82740335e-03 2.85840988e-01 -2.17558160e-01 -3.08247417e-01
-6.24605179e-01 5.04498124e-01 4.02154475e-01 -2.30054125e-01
5.42892933e-01 -5.54296494e-01 -1.48501411e-01 2.58086234e-01
-7.64831722e-01 -2.40939870e-01 6.68608248e-01 8.14172208e-01
2.40799353e-01 -7.74436355e-01 5.02402902e-01 -7.32543170e-01
6.82264328e-01 -5.09840429e-01 -4.45068598e-01 3.01133126e-01
-7.09636867e-01 2.42195353e-01 6.23718500e-01 -6.33355081e-01
-6.62371933e-01 -3.39024097e-01 -1.92321107e-01 3.89490984e-02
-8.06987286e-02 3.95750642e-01 -2.92320430e-01 2.14316890e-01
5.24785519e-01 2.07677901e-01 -2.30630830e-01 -5.23017466e-01
6.81636810e-01 5.87413073e-01 3.93499523e-01 -9.51029301e-01
1.00995171e+00 -6.48752749e-02 -6.39257431e-01 -1.00079024e+00
-7.17984080e-01 -2.66751975e-01 -7.99730659e-01 2.49121681e-01
1.10414505e+00 -1.01269281e+00 -1.21198654e-01 2.02681959e-01
-1.28762710e+00 -6.30027413e-01 -3.02015513e-01 8.52698743e-01
-3.93204659e-01 8.58543292e-02 -6.24846518e-01 -2.16937721e-01
-2.14998275e-01 -1.10344934e+00 8.59041810e-01 -2.69228488e-01
-8.65783930e-01 -1.59543169e+00 5.22960246e-01 -5.75443618e-02
5.52348912e-01 -2.55392373e-01 1.48506713e+00 -1.04277754e+00
-1.84913859e-01 -2.80985832e-02 -1.54560775e-01 5.54911137e-01
3.76698107e-01 -1.53436854e-01 -7.70973206e-01 -5.54086924e-01
-3.49067301e-01 -3.14020008e-01 7.31991172e-01 1.69325382e-01
1.37629092e-01 -2.57418156e-01 -1.38835087e-01 6.29581869e-01
1.56470203e+00 3.99590433e-02 4.68286365e-01 3.92311603e-01
7.18490243e-01 7.24678516e-01 3.40343475e-01 -2.51314282e-01
4.61734146e-01 7.57346451e-01 -7.19119832e-02 -1.01001844e-01
-3.32330048e-01 -6.64885104e-01 1.06774640e+00 1.56237209e+00
3.20746183e-01 -7.97602683e-02 -9.21287894e-01 1.09397137e+00
-1.28608775e+00 -5.72526515e-01 4.52875458e-02 2.43811679e+00
1.29705107e+00 7.19219521e-02 -7.24678859e-02 -3.71398211e-01
6.60876453e-01 2.98580199e-01 -1.51973665e-01 -8.14770937e-01
-2.24287465e-01 6.62772834e-01 3.75850052e-01 1.02811885e+00
-7.75779307e-01 1.41893876e+00 6.19331360e+00 8.57706666e-01
-1.04552317e+00 3.46469760e-01 2.05923229e-01 8.20040330e-02
-7.24829972e-01 9.34239402e-02 -1.19685161e+00 3.99033606e-01
1.20000756e+00 -2.96217263e-01 3.12887341e-01 5.79290152e-01
-1.19438812e-01 2.79526770e-01 -1.63792825e+00 5.05949616e-01
1.20144002e-01 -8.81421685e-01 3.03201944e-01 3.45930830e-02
9.42240477e-01 4.21888202e-01 1.51380315e-01 6.39354706e-01
6.95733130e-01 -1.11976671e+00 8.16731751e-01 -3.09789889e-02
9.79802489e-01 -6.62815809e-01 4.87355649e-01 1.85733482e-01
-1.03668642e+00 5.49156010e-01 -3.11092496e-01 -1.41813219e-01
1.78521916e-01 2.84079552e-01 -7.73739278e-01 2.83673525e-01
2.91098714e-01 6.06630445e-01 -5.59890270e-01 4.38036621e-01
-6.55176640e-01 7.14055538e-01 5.19011766e-02 -1.59082383e-01
4.30522174e-01 -2.57238060e-01 4.34050053e-01 1.65682530e+00
4.72565681e-01 -7.00852513e-01 -2.34048255e-02 7.00000107e-01
-1.20807692e-01 6.43163681e-01 -9.14586186e-01 -3.20829511e-01
4.93375510e-01 1.05937576e+00 -1.10474534e-01 -1.46297708e-01
-7.73596704e-01 1.08693826e+00 7.89450228e-01 1.91720739e-01
-7.99793661e-01 -5.64341784e-01 1.20616376e+00 2.03452662e-01
2.40005776e-01 -5.28659523e-01 -1.71959952e-01 -1.05514395e+00
-2.02331856e-01 -8.91269743e-01 7.89753348e-02 -4.14850056e-01
-1.56963825e+00 7.94748187e-01 -1.86926872e-01 -9.87218499e-01
-1.95586905e-01 -8.93135548e-01 -5.68799675e-01 1.29288995e+00
-1.60001755e+00 -1.13055611e+00 4.06915307e-01 4.06003296e-01
4.31950063e-01 -3.59679371e-01 1.02347124e+00 4.97725755e-01
-4.63362128e-01 1.09287798e+00 1.32935315e-01 3.19228262e-01
1.28593099e+00 -1.30455172e+00 8.69401038e-01 9.29668009e-01
4.91124332e-01 9.67980802e-01 5.44377565e-01 -4.12520796e-01
-1.10135913e+00 -1.20527077e+00 1.62442935e+00 -7.55969822e-01
1.08670580e+00 -5.01118362e-01 -8.69181514e-01 1.02891695e+00
8.31471086e-01 -3.41789454e-01 7.97065020e-01 5.29145718e-01
-7.77636766e-01 1.75408758e-02 -6.36215568e-01 9.92417037e-01
9.85671520e-01 -9.32967663e-01 -8.61823201e-01 2.04992816e-01
9.71736193e-01 2.76791573e-01 -9.87951040e-01 1.76990762e-01
4.65563804e-01 -7.75098860e-01 7.24010527e-01 -1.05377448e+00
6.41225219e-01 6.95618764e-02 -4.61779624e-01 -1.93765664e+00
-4.38432783e-01 -4.62081701e-01 4.81190920e-01 1.49293339e+00
6.82772279e-01 -8.06039453e-01 8.58000517e-02 6.76040491e-03
-1.97920889e-01 -6.15540564e-01 -8.09858322e-01 -1.28683114e+00
8.62670243e-01 -3.99760664e-01 3.05461138e-01 1.39826560e+00
4.47159000e-02 8.19406629e-01 -9.18297619e-02 9.36359633e-03
3.28035682e-01 -1.40292212e-01 7.19576597e-01 -9.90120411e-01
-3.71724129e-01 -7.32287407e-01 -2.96229511e-01 -9.41556036e-01
5.99706411e-01 -1.55448878e+00 6.56715631e-02 -1.46068156e+00
1.38356313e-01 -5.45822740e-01 -4.15520728e-01 4.25113559e-01
-3.86743873e-01 1.48993164e-01 3.14559251e-01 4.12839018e-02
-1.30353242e-01 5.40558815e-01 9.13465679e-01 3.87457199e-03
-8.97497758e-02 -4.14186925e-01 -8.61344039e-01 7.52959073e-01
8.55659366e-01 -5.29834211e-01 -1.47980809e-01 -9.36774254e-01
-4.37287427e-02 -3.95008147e-01 -1.27170548e-01 -8.61747563e-01
-2.94281781e-01 -1.19790465e-01 -1.16090521e-01 3.07174660e-02
1.35671452e-01 -4.73082364e-01 -3.04096788e-01 5.69540560e-01
-3.86955827e-01 4.90361691e-01 4.07890558e-01 3.25585133e-03
-2.75437534e-01 -4.45274264e-01 9.06045735e-01 1.08927727e-01
-6.11993015e-01 1.35461807e-01 -3.38964641e-01 7.30527759e-01
7.18220413e-01 -1.00429900e-01 -3.06065053e-01 -9.45772603e-02
-1.82368338e-01 -1.32845163e-01 5.93634129e-01 8.11257184e-01
-3.69685399e-03 -1.66640735e+00 -1.03932393e+00 2.63596088e-01
4.72906202e-01 -7.28990793e-01 -1.29737914e-01 8.03802192e-01
-2.97137886e-01 6.03842795e-01 -3.37151617e-01 -4.32877839e-01
-1.11866283e+00 2.65634179e-01 1.05731890e-01 -4.63753819e-01
-3.65389464e-03 9.99534965e-01 5.20412266e-01 -1.10083449e+00
-9.98390839e-02 -3.75387669e-01 3.46695781e-01 2.23876327e-01
2.03266338e-01 1.27048269e-01 3.69080938e-02 -7.66776323e-01
-3.74938011e-01 7.62851357e-01 -2.39539981e-01 -3.59498143e-01
1.32219207e+00 1.01071559e-01 -1.58806220e-01 7.10919499e-01
1.56154740e+00 6.28115177e-01 -1.06659901e+00 -4.29695994e-01
1.96284130e-01 -1.10985301e-01 -3.17509830e-01 -6.41553402e-01
-6.21402204e-01 1.25688064e+00 3.16958457e-01 -4.11616832e-01
4.63653952e-01 1.17728025e-01 8.12732697e-01 9.56792012e-02
2.44234934e-01 -9.92527962e-01 1.60419364e-02 9.62840736e-01
8.53595018e-01 -1.04948831e+00 -2.41328657e-01 6.96773827e-02
-6.10555351e-01 6.02334082e-01 7.32983351e-01 -1.21768810e-01
4.33120728e-01 7.21291304e-02 2.70927459e-01 2.91338623e-01
-5.66930711e-01 -2.20599651e-01 4.02765244e-01 5.93437374e-01
9.47621047e-01 2.32364133e-01 -4.67631489e-01 4.62913245e-01
-5.99332094e-01 -6.29307210e-01 7.43605047e-02 5.79754412e-01
-3.51367474e-01 -1.52026665e+00 -9.41596627e-02 1.91204980e-01
-3.40645432e-01 -7.13164806e-01 -2.76119053e-01 1.16240799e+00
1.41290843e-01 5.09224772e-01 3.37457359e-01 -3.80256355e-01
4.08283681e-01 3.93887401e-01 7.10039079e-01 -9.26119983e-01
-7.14723527e-01 -3.15328985e-01 1.39672503e-01 -1.13479875e-01
-5.83354533e-02 -7.47644842e-01 -8.32734764e-01 -2.73548335e-01
-1.78018287e-01 4.60041948e-02 5.72963417e-01 8.19993854e-01
1.63634792e-01 3.50422174e-01 2.95504600e-01 -6.40812218e-01
-8.05997014e-01 -1.13304448e+00 -4.75814551e-01 5.68232954e-01
1.47509620e-01 -3.86928082e-01 -3.51536632e-01 1.10702217e-02] | [10.912108421325684, 9.998140335083008] |
bafb69a2-0a1e-401e-a769-49dbc8d80b76 | an-improved-analysis-of-variance-reduced-1 | 2211.07937 | null | https://arxiv.org/abs/2211.07937v2 | https://arxiv.org/pdf/2211.07937v2.pdf | An Improved Analysis of (Variance-Reduced) Policy Gradient and Natural Policy Gradient Methods | In this paper, we revisit and improve the convergence of policy gradient (PG), natural PG (NPG) methods, and their variance-reduced variants, under general smooth policy parametrizations. More specifically, with the Fisher information matrix of the policy being positive definite: i) we show that a state-of-the-art variance-reduced PG method, which has only been shown to converge to stationary points, converges to the globally optimal value up to some inherent function approximation error due to policy parametrization; ii) we show that NPG enjoys a lower sample complexity; iii) we propose SRVR-NPG, which incorporates variance-reduction into the NPG update. Our improvements follow from an observation that the convergence of (variance-reduced) PG and NPG methods can improve each other: the stationary convergence analysis of PG can be applied to NPG as well, and the global convergence analysis of NPG can help to establish the global convergence of (variance-reduced) PG methods. Our analysis carefully integrates the advantages of these two lines of works. Thanks to this improvement, we have also made variance-reduction for NPG possible, with both global convergence and an efficient finite-sample complexity. | ['Wotao Yin', 'Tamer Başar', 'Kaiqing Zhang', 'Yanli Liu'] | 2022-11-15 | an-improved-analysis-of-variance-reduced | http://proceedings.neurips.cc/paper/2020/hash/56577889b3c1cd083b6d7b32d32f99d5-Abstract.html | http://proceedings.neurips.cc/paper/2020/file/56577889b3c1cd083b6d7b32d32f99d5-Paper.pdf | neurips-2020-12 | ['policy-gradient-methods'] | ['methodology'] | [-1.72936529e-01 8.50892067e-02 -4.12708580e-01 3.09209585e-01
-8.48626912e-01 -7.75382936e-01 6.09588623e-01 -8.05774480e-02
-3.94466430e-01 1.09593523e+00 1.33139074e-01 -8.16029251e-01
-5.41141868e-01 -4.51138824e-01 -8.08570027e-01 -8.63327503e-01
-3.14882100e-01 2.89139092e-01 3.19621116e-01 -3.80411714e-01
4.20304805e-01 5.00690043e-01 -1.26313114e+00 -7.84280360e-01
1.08214712e+00 1.19078183e+00 4.43016216e-02 7.80414522e-01
2.24342227e-01 5.00378251e-01 -1.83249906e-01 1.98240858e-02
6.23683929e-01 -4.97106522e-01 -7.17073381e-01 -1.70565814e-01
2.58035362e-01 -2.67911762e-01 -1.43555641e-01 1.41228592e+00
5.48643053e-01 5.86944461e-01 5.11558473e-01 -1.08142304e+00
-4.97883350e-01 3.13284665e-01 -5.89994192e-01 1.40235066e-01
-2.29842171e-01 3.74288522e-02 8.66079926e-01 -4.60670978e-01
4.95799929e-01 1.47410977e+00 7.74930596e-01 7.02228785e-01
-1.30031192e+00 -3.74987006e-01 3.94884735e-01 -4.06152941e-02
-8.09655011e-01 -2.37557992e-01 4.12538558e-01 -4.80322212e-01
5.83344340e-01 2.05017641e-01 5.99927247e-01 8.25894415e-01
1.24308608e-01 8.99739206e-01 1.39714587e+00 -4.56324220e-01
5.21520197e-01 7.45178089e-02 3.83381136e-02 7.74408877e-01
4.66044068e-01 4.94551450e-01 9.86164138e-02 -2.47456953e-01
8.88419867e-01 -3.45503002e-01 -6.60526752e-01 -6.23921037e-01
-8.99070323e-01 9.78576422e-01 1.82477280e-01 2.78055131e-01
-4.25619006e-01 3.05903256e-01 3.60952079e-01 3.48304451e-01
7.42748857e-01 4.79803175e-01 -3.75715077e-01 -6.10153198e-01
-7.42951870e-01 5.42698979e-01 1.06301975e+00 8.47302139e-01
4.83969718e-01 2.18581557e-01 -5.51217139e-01 5.34049630e-01
3.36761653e-01 9.71107006e-01 5.36981404e-01 -1.33066356e+00
5.04373372e-01 -1.22903615e-01 7.47765839e-01 -9.09518063e-01
-3.65827799e-01 -7.57094741e-01 -6.24379337e-01 3.24584365e-01
7.40812659e-01 -4.35159534e-01 -4.99963015e-01 2.18062949e+00
2.80809551e-01 4.37702984e-02 4.87331264e-02 7.68636942e-01
-1.90941170e-01 6.93242252e-01 -3.49163055e-01 -7.19017446e-01
8.92928958e-01 -7.90263414e-01 -7.50346780e-01 -7.47449920e-02
4.97757882e-01 -3.58274788e-01 1.24328387e+00 2.95187056e-01
-1.02478802e+00 -2.32290104e-01 -9.54129100e-01 3.60712290e-01
1.19824275e-01 1.54824227e-01 3.98569196e-01 8.36099803e-01
-1.29414880e+00 1.10146117e+00 -9.80253398e-01 -4.15544629e-01
1.94700852e-01 1.32695556e-01 6.54311627e-02 5.06386399e-01
-9.05365884e-01 1.02240527e+00 2.36538455e-01 -4.61014621e-02
-7.95080602e-01 -7.09568441e-01 -5.73926270e-01 1.15514919e-01
9.82115448e-01 -6.96029365e-01 1.69173598e+00 -8.10623884e-01
-2.30201221e+00 -1.23322587e-02 -3.92320871e-01 -5.86584985e-01
1.01894927e+00 -4.31986719e-01 2.35350177e-01 2.59661078e-01
5.13556376e-02 -1.72461718e-01 7.98493445e-01 -1.09139895e+00
-8.13586831e-01 -5.01603365e-01 -1.21903755e-01 2.20684648e-01
-2.35727772e-01 -1.82603270e-01 -9.23308283e-02 -4.65613037e-01
-8.76624435e-02 -9.36399996e-01 -4.14134115e-01 -2.01951399e-01
-1.95317149e-01 -2.66086459e-01 5.59309244e-01 -5.79134881e-01
1.20065475e+00 -1.81741703e+00 2.14687198e-01 3.15626204e-01
-5.56900650e-02 4.32921499e-01 -1.58883091e-02 3.57818782e-01
1.58256829e-01 3.26829046e-01 -1.71440423e-01 -2.91350216e-01
2.44308427e-01 1.80199653e-01 -6.87510848e-01 8.70044410e-01
-1.06827706e-01 7.89494932e-01 -1.27828431e+00 4.08399291e-02
2.65119433e-01 1.44645810e-01 -5.37222207e-01 -2.22606957e-01
-1.12991810e-01 4.94907320e-01 -6.78897262e-01 -4.72046807e-02
4.98922735e-01 -2.05834329e-01 2.23514661e-01 1.08140819e-01
-3.82740140e-01 2.39670694e-01 -1.10007823e+00 1.06825352e+00
-5.17338574e-01 4.01727319e-01 4.46723253e-01 -1.26336443e+00
5.69995463e-01 1.50403380e-01 6.29381657e-01 -3.10673684e-01
1.09233595e-01 5.24312675e-01 -3.49592835e-01 -1.40239596e-01
3.87624562e-01 -2.98154712e-01 3.93721849e-01 2.88944542e-01
-7.26784319e-02 6.92829192e-02 3.38272750e-01 2.69726664e-02
6.52886569e-01 2.37524137e-01 4.18183655e-01 -7.95144022e-01
6.42063141e-01 -3.94541025e-01 4.47460055e-01 1.14899600e+00
-4.36646849e-01 -4.19891924e-02 8.82116735e-01 2.55261034e-01
-9.73514676e-01 -8.82417381e-01 -9.22269821e-02 9.60074127e-01
4.18220460e-02 -3.16494614e-01 -7.24315166e-01 -6.92658007e-01
2.80534208e-01 7.25920022e-01 -7.06047714e-01 1.13593955e-02
-5.32333195e-01 -7.10416377e-01 2.58506954e-01 4.04103637e-01
5.67228377e-01 -2.81397104e-01 -4.64360148e-01 4.38783944e-01
1.92984581e-01 -8.26913118e-01 -5.42985976e-01 -6.06475770e-02
-9.84651625e-01 -1.10755849e+00 -1.00417042e+00 -1.87315449e-01
3.82307202e-01 8.82285461e-02 5.84571302e-01 -3.98394495e-01
5.80150604e-01 7.66267002e-01 -2.24212989e-01 -4.03958678e-01
-5.59614897e-01 -4.51508947e-02 5.02004504e-01 6.70008063e-02
-4.50745612e-01 -6.76535189e-01 -4.69550252e-01 5.02392352e-01
-5.09058475e-01 -2.22445786e-01 2.88556486e-01 8.76893222e-01
5.83545208e-01 -5.43195680e-02 5.64029515e-01 -7.04774976e-01
1.05058062e+00 -2.24059016e-01 -1.42080688e+00 2.17718661e-01
-1.19906986e+00 4.94753391e-01 9.60264504e-01 -3.47721070e-01
-9.19986367e-01 -3.56350929e-01 -1.11918114e-01 -5.83206534e-01
5.51251590e-01 2.51476407e-01 2.02968284e-01 -2.03748569e-01
5.11885762e-01 3.27841312e-01 5.73842704e-01 -6.50826693e-01
7.06683934e-01 4.97377396e-01 5.94172835e-01 -9.09551382e-01
7.86807656e-01 6.32546902e-01 3.71969163e-01 -6.77499592e-01
-8.10493469e-01 -5.02218485e-01 -1.42211048e-02 -1.59282964e-02
5.05497754e-01 -4.65163857e-01 -1.24736238e+00 3.13183039e-01
-7.76498020e-01 -7.67242312e-01 -6.14948213e-01 6.69996262e-01
-1.06377017e+00 7.94973910e-01 -4.47460324e-01 -1.58736253e+00
-2.60554045e-01 -1.05283427e+00 7.41152406e-01 2.80059636e-01
4.12508577e-01 -1.24226832e+00 2.05697179e-01 -3.87650132e-01
5.72729409e-01 1.52863160e-01 5.03429592e-01 -4.49880540e-01
-3.16564925e-02 1.32677212e-01 -1.61587462e-01 5.61895907e-01
9.39060003e-02 -6.94224983e-02 -6.08382404e-01 -7.88817048e-01
4.56812710e-01 1.44828424e-01 8.49940777e-01 8.53385806e-01
9.39783275e-01 -8.14270675e-01 -3.82804751e-01 6.20625377e-01
1.49832058e+00 1.93110496e-01 2.08371833e-01 4.70091105e-01
3.46984088e-01 4.72429633e-01 6.60787761e-01 5.47572613e-01
7.28747919e-02 6.78145885e-01 3.08946341e-01 2.64604509e-01
1.73954263e-01 -4.17067945e-01 6.18926525e-01 5.85317731e-01
-4.07137454e-01 -3.19042578e-02 -3.85280490e-01 3.47768784e-01
-2.29690552e+00 -9.49008107e-01 -1.03157096e-01 2.67583823e+00
8.10862005e-01 -2.26511344e-01 5.63110590e-01 -7.02756569e-02
7.26097465e-01 1.48421958e-01 -6.55853868e-01 -2.91441232e-01
-2.51871217e-02 2.75738746e-01 1.15057302e+00 9.17053342e-01
-1.02634108e+00 7.74804831e-01 7.02880478e+00 1.14987648e+00
-1.11836064e+00 3.37610304e-01 3.46615463e-01 -5.90143260e-03
-4.66581024e-02 8.27923566e-02 -9.85590041e-01 6.37657762e-01
9.72754717e-01 -6.34445667e-01 7.80505121e-01 1.21809828e+00
5.79378664e-01 -9.60074272e-03 -6.36699200e-01 9.03856993e-01
-4.57230538e-01 -1.14393270e+00 -3.62541944e-01 4.39791352e-01
7.97214270e-01 -5.78213036e-02 2.05153525e-01 5.30873477e-01
4.50333238e-01 -5.54427743e-01 6.62272871e-01 2.60423958e-01
5.33298969e-01 -8.64180207e-01 6.03901684e-01 3.47359747e-01
-9.79619086e-01 -4.62857962e-01 -5.23091853e-01 -1.37066841e-01
2.68730462e-01 5.66223979e-01 -4.44059968e-01 9.63398159e-01
2.84910709e-01 5.42281091e-01 -1.12934448e-01 1.01724315e+00
-4.64537293e-01 8.22378337e-01 -5.87355673e-01 -2.56053537e-01
6.18727922e-01 -6.86540604e-01 1.14812016e+00 7.95223594e-01
3.79302621e-01 -2.79378384e-01 2.32733518e-01 8.57387900e-01
3.45961362e-01 2.02767909e-01 -2.03543663e-01 -8.09013024e-02
2.52315432e-01 9.06270087e-01 -4.51425314e-01 -4.18435901e-01
-5.13099916e-02 6.18379831e-01 1.93424031e-01 5.26674807e-01
-6.30186141e-01 -4.23225343e-01 8.59235346e-01 -9.39906016e-02
5.65118968e-01 -3.45091432e-01 1.97523963e-02 -1.20137799e+00
1.83580190e-01 -6.97998285e-01 1.84093028e-01 -1.85380965e-01
-1.12648737e+00 2.47049123e-01 2.49138236e-01 -1.11508584e+00
-6.66661680e-01 -7.00457931e-01 -3.00634146e-01 9.37610745e-01
-1.56608963e+00 -3.20784479e-01 4.09485519e-01 3.34221810e-01
2.85960823e-01 -7.23763257e-02 4.63166237e-01 9.58752912e-03
-4.98477876e-01 5.39486289e-01 9.76442277e-01 -1.63956985e-01
3.49494547e-01 -1.62825918e+00 1.70449600e-01 1.18155205e+00
-3.40032369e-01 7.47213900e-01 9.75337863e-01 -5.88803530e-01
-1.42134953e+00 -1.05458593e+00 2.77675658e-01 -1.66601151e-01
1.07025123e+00 -8.79630633e-03 -5.56981385e-01 7.72262573e-01
-1.94449335e-01 -8.31992328e-02 -3.93460281e-02 1.85081318e-01
1.51596040e-01 -6.04727343e-02 -9.22040522e-01 7.17752159e-01
9.51681077e-01 -2.34385639e-01 -4.91608590e-01 4.09384638e-01
5.89518249e-01 -4.73187894e-01 -7.48649120e-01 3.03692430e-01
5.61646163e-01 -8.85229766e-01 6.71552956e-01 -8.65471482e-01
-1.99941650e-01 -4.02882218e-01 -1.20222932e-02 -1.53569329e+00
-2.21972957e-01 -1.47703159e+00 -4.13618386e-01 9.04860139e-01
1.32499665e-01 -1.41889679e+00 3.81630927e-01 3.27234060e-01
-1.06369942e-01 -9.24220204e-01 -1.16213346e+00 -1.63722086e+00
4.38677877e-01 -5.49186349e-01 1.90040395e-01 4.78722364e-01
1.60577312e-01 1.79438010e-01 -5.73985338e-01 5.60390912e-02
5.78923285e-01 2.00587418e-03 6.85591400e-01 -9.65716600e-01
-7.66132832e-01 -7.91387618e-01 -1.71715003e-02 -1.56111860e+00
1.09970659e-01 -5.59050679e-01 5.00899553e-02 -1.42610514e+00
-1.25312537e-01 -4.59892452e-01 -1.42854109e-01 2.43112102e-01
-4.15095299e-01 -1.95822299e-01 4.67907995e-01 3.49991798e-01
-3.35025281e-01 9.08138335e-01 1.44791365e+00 2.67827481e-01
-5.40304840e-01 4.85907644e-01 -6.57123744e-01 6.22383058e-01
7.55546451e-01 -2.98480153e-01 -6.41496956e-01 1.96103349e-01
1.35919541e-01 2.08767559e-02 1.84329763e-01 -8.79374087e-01
-1.93751544e-01 -1.49440646e-01 -2.22731337e-01 -2.09471211e-01
5.05295768e-03 -3.18057179e-01 -2.79424459e-01 8.65478039e-01
-3.25611770e-01 -2.50183314e-01 1.15877584e-01 8.80051136e-01
2.59229839e-01 -4.54671949e-01 9.30765569e-01 1.14030414e-03
-2.64833510e-01 3.40049505e-01 -5.50262272e-01 2.06891581e-01
6.59598708e-01 5.66844046e-02 -3.23552221e-01 -8.50761175e-01
-5.86298347e-01 3.30058843e-01 3.00540745e-01 -1.31438032e-01
1.29443025e-02 -1.24562860e+00 -3.85321468e-01 -2.09721938e-01
-6.04654253e-01 -3.76498371e-01 -9.53600630e-02 1.40617096e+00
-1.57771140e-01 6.77014172e-01 2.14634091e-01 -5.50681055e-01
-8.23916376e-01 7.02351630e-01 4.85562831e-01 -5.86518049e-01
-6.56882763e-01 4.93902326e-01 7.79731348e-02 -2.64098614e-01
9.86158177e-02 -6.72926188e-01 -1.52785005e-02 -1.92081019e-01
6.16884828e-01 7.39598691e-01 -1.98574245e-01 -1.97707817e-01
-1.46409929e-01 6.99834764e-01 2.65014082e-01 -6.09825253e-01
1.08637631e+00 -2.71845609e-01 2.26150364e-01 3.30031842e-01
1.11376810e+00 1.87299788e-01 -1.56080341e+00 2.10331706e-03
-1.62110943e-02 -4.30889726e-01 1.82387903e-01 -4.55992252e-01
-9.47823942e-01 6.15169525e-01 5.46178579e-01 4.01411623e-01
1.00324929e+00 -4.65042681e-01 5.33199072e-01 3.60699475e-01
5.53786397e-01 -1.09624553e+00 -3.87825131e-01 7.29047477e-01
6.34783685e-01 -8.17415357e-01 2.59321090e-02 -2.45780692e-01
-2.87331671e-01 1.07308400e+00 1.61448552e-04 -3.72979075e-01
5.81489623e-01 -1.86678290e-01 -3.77677888e-01 3.01640064e-01
-5.54411173e-01 -4.77146894e-01 1.67021796e-01 6.08548939e-01
-2.60861758e-02 3.07769999e-02 -9.99596357e-01 2.99797118e-01
-1.44751623e-01 5.97613379e-02 3.70311141e-01 6.78809106e-01
-4.72858131e-01 -1.09537137e+00 -2.22899929e-01 2.39201427e-01
-5.69711208e-01 2.82942858e-02 2.18330231e-02 1.04437232e+00
-7.40174353e-01 9.88152683e-01 -1.93638429e-01 9.31206495e-02
2.73182988e-01 -1.07354812e-01 6.86525822e-01 -2.43370742e-01
-1.94818646e-01 2.56607294e-01 3.82661410e-02 -9.58768785e-01
-4.21548039e-01 -5.55306256e-01 -9.70534682e-01 -2.73808688e-01
-3.43419433e-01 5.61722279e-01 7.18301535e-01 1.00941479e+00
4.93650019e-01 2.23389670e-01 7.60025501e-01 -7.57973552e-01
-1.65409601e+00 -1.02162540e+00 -8.48080575e-01 1.08608879e-01
3.95725131e-01 -8.85162234e-01 -7.52903819e-01 -5.94309926e-01] | [4.2951860427856445, 2.61540150642395] |
750cdc60-e108-4e77-89cc-586474f373e1 | quick-tune-quickly-learning-which-pretrained | 2306.03828 | null | https://arxiv.org/abs/2306.03828v3 | https://arxiv.org/pdf/2306.03828v3.pdf | Quick-Tune: Quickly Learning Which Pretrained Model to Finetune and How | With the ever-increasing number of pretrained models, machine learning practitioners are continuously faced with which pretrained model to use, and how to finetune it for a new dataset. In this paper, we propose a methodology that jointly searches for the optimal pretrained model and the hyperparameters for finetuning it. Our method transfers knowledge about the performance of many pretrained models with multiple hyperparameter configurations on a series of datasets. To this aim, we evaluated over 20k hyperparameter configurations for finetuning 24 pretrained image classification models on 87 datasets to generate a large-scale meta-dataset. We meta-learn a multi-fidelity performance predictor on the learning curves of this meta-dataset and use it for fast hyperparameter optimization on new datasets. We empirically demonstrate that our resulting approach can quickly select an accurate pretrained model for a new dataset together with its optimal hyperparameters. | ['Josif Grabocka', 'Frank Hutter', 'Arlind Kadra', 'Fabio Ferreira', 'Sebastian Pineda Arango'] | 2023-06-06 | null | null | null | null | ['hyperparameter-optimization'] | ['methodology'] | [ 3.22138518e-01 -2.09742531e-01 -3.05095345e-01 -4.30957526e-01
-1.34841812e+00 -6.64325356e-01 1.66088909e-01 -9.31996405e-02
-6.65002167e-01 6.58629119e-01 -9.27845109e-03 6.32665411e-04
-5.42172074e-01 -4.70063001e-01 -7.89065003e-01 -9.27970111e-01
2.38358006e-01 8.83265793e-01 1.08793363e-01 1.75038353e-01
3.60429376e-01 5.66765964e-01 -1.38433981e+00 2.16304481e-01
1.06186831e+00 8.36391866e-01 2.62465447e-01 9.71073627e-01
2.74229109e-01 -1.33164212e-01 -5.84798276e-01 -4.65580761e-01
3.95784080e-01 -2.21509606e-01 -8.75808477e-01 1.37793675e-01
7.88471520e-01 -5.17842509e-02 -2.07308769e-01 8.14982951e-01
3.66433531e-01 2.76759863e-01 8.45842719e-01 -9.76535618e-01
-2.68611580e-01 7.08622277e-01 -3.92581254e-01 3.58396411e-01
-3.68528783e-01 5.36130965e-01 8.18936229e-01 -5.53440571e-01
5.77178717e-01 8.38464200e-01 7.25866735e-01 4.21548545e-01
-1.62222171e+00 -7.18365431e-01 -1.51283249e-01 1.12031840e-01
-1.52658606e+00 -3.77866119e-01 6.64602399e-01 -4.10220325e-01
7.53140271e-01 -2.94337142e-02 6.36459529e-01 1.19817638e+00
3.40908527e-01 1.59218073e-01 9.20270741e-01 -5.05218863e-01
1.90594271e-01 3.78262222e-01 1.63331643e-01 7.83358574e-01
2.87558079e-01 2.42197305e-01 -3.89011741e-01 -3.79475504e-01
7.34276474e-01 -3.73584479e-01 -9.60227177e-02 -4.95152622e-01
-1.19865787e+00 5.98884940e-01 3.85467917e-01 2.43159905e-01
-3.65829498e-01 3.34830791e-01 3.76457930e-01 1.59632534e-01
2.20581964e-01 1.12483883e+00 -8.24710667e-01 -2.02928469e-01
-7.57532895e-01 2.01512873e-01 4.71960753e-01 7.08974242e-01
1.16114056e+00 -2.00462252e-01 -2.15939343e-01 1.17489660e+00
-2.86514126e-03 4.02088135e-01 5.52003026e-01 -1.11687183e+00
5.75996876e-01 4.71097916e-01 -1.92081928e-02 -6.41651988e-01
-4.71359760e-01 -6.07957184e-01 -7.93218136e-01 -3.10418993e-01
2.67764628e-01 -2.98026383e-01 -1.19892061e+00 1.66113687e+00
4.76475745e-01 5.68044484e-01 -1.38261259e-01 6.11998856e-01
3.48951131e-01 6.87849879e-01 1.67055726e-01 -1.75828502e-01
8.43151927e-01 -8.40237975e-01 -4.09184396e-02 -2.66825646e-01
7.55518854e-01 -6.87619984e-01 1.42083502e+00 6.52591765e-01
-8.69791090e-01 -6.31435990e-01 -1.10408461e+00 1.86709493e-01
4.50644549e-03 2.80901372e-01 3.23932797e-01 5.60623765e-01
-9.23831522e-01 8.89073670e-01 -6.58460498e-01 -1.64239973e-01
4.55477625e-01 7.52035558e-01 -2.17309982e-01 -8.41068476e-02
-8.42458248e-01 9.18350637e-01 1.05142307e+00 1.11263758e-03
-9.90457416e-01 -1.25979018e+00 -2.46921241e-01 -9.11454782e-02
4.84455138e-01 -9.42177474e-01 1.28901112e+00 -1.01279378e+00
-1.61358273e+00 7.06381619e-01 3.07518750e-01 -2.74368286e-01
2.24806726e-01 1.20707667e-02 -3.14344078e-01 -2.30515599e-02
-4.62891757e-01 8.16844344e-01 1.35305488e+00 -1.38667226e+00
-4.98880208e-01 -2.70258069e-01 -4.25967515e-01 2.84695834e-01
-5.74487031e-01 -4.42669541e-01 -6.96323812e-01 -5.75751960e-01
-2.35872269e-02 -1.27101231e+00 -3.37993771e-01 -4.63757426e-01
-6.09224379e-01 1.64481491e-01 5.21021366e-01 -1.76691577e-01
1.36935997e+00 -1.98945951e+00 6.11127079e-01 6.17322505e-01
1.65042043e-01 4.64394629e-01 -5.11252403e-01 4.93642129e-02
-1.04070961e-01 3.03915232e-01 -2.74395257e-01 -2.27309763e-01
-3.38068902e-01 5.03942728e-01 -1.45491987e-01 2.99736410e-01
-3.43271904e-02 8.05060983e-01 -5.18306136e-01 -5.05678236e-01
1.17150873e-01 7.31830597e-01 -8.60880315e-01 3.30841631e-01
-2.36545965e-01 4.84750748e-01 -6.41627252e-01 2.22003743e-01
4.43293542e-01 -4.71782535e-01 2.09715024e-01 -6.37413263e-01
3.00765097e-01 -3.06613356e-01 -1.10362232e+00 1.51047277e+00
-6.83728456e-01 3.24777633e-01 -3.13428700e-01 -8.16653371e-01
9.99962151e-01 -9.44968611e-02 5.73610246e-01 -3.92445385e-01
3.32389623e-01 2.30717212e-01 -6.02850579e-02 -5.44523060e-01
4.88491684e-01 -3.06758910e-01 -2.71349028e-02 4.84771907e-01
3.51234615e-01 -2.20929474e-01 -7.00438097e-02 -3.00108522e-01
9.01035726e-01 -4.39511351e-02 8.92739072e-02 -3.80492419e-01
3.18469375e-01 9.40910429e-02 2.25156337e-01 8.51270139e-01
8.15993696e-02 6.23782158e-01 9.70712975e-02 -7.20624745e-01
-1.42195189e+00 -7.46068299e-01 -3.45251679e-01 1.02515733e+00
-2.68063873e-01 -3.48763406e-01 -1.02111614e+00 -6.22091413e-01
4.53478750e-03 4.99716073e-01 -7.89754272e-01 -7.20391572e-01
-8.04789186e-01 -9.94606733e-01 5.31961620e-01 2.09438309e-01
5.42616844e-01 -7.61745214e-01 -5.95279336e-01 1.22096963e-01
-1.32193461e-01 -1.05567884e+00 -8.21921229e-01 4.69003499e-01
-1.26284456e+00 -1.12964952e+00 -4.41397786e-01 -5.76040089e-01
7.34363794e-01 -1.43103555e-01 1.18069589e+00 3.17019016e-01
-3.38995308e-01 2.10484535e-01 -1.57953389e-02 -4.76816297e-02
-8.06762457e-01 7.46448755e-01 8.40709060e-02 1.03406496e-01
-9.94187742e-02 -3.89063299e-01 -4.08973277e-01 5.92997193e-01
-1.09762073e+00 1.07257012e-02 6.03229821e-01 9.03225303e-01
1.11536825e+00 3.44008803e-01 4.18345362e-01 -1.19836617e+00
4.99527603e-01 -3.57170701e-01 -8.56759787e-01 6.82092130e-01
-1.00114989e+00 6.62919104e-01 7.62192249e-01 -8.47819865e-01
-9.82745588e-01 1.69104278e-01 4.65513505e-02 -7.36869097e-01
6.20162599e-02 4.77302879e-01 -4.56300564e-02 -3.87615532e-01
1.09660757e+00 -2.31647745e-01 -3.76900345e-01 -5.84760189e-01
5.39293408e-01 3.81503433e-01 7.74307609e-01 -1.01456714e+00
8.05347681e-01 -4.80256928e-03 4.71802652e-02 -6.56386018e-01
-9.98161674e-01 -2.42665246e-01 -9.24219429e-01 -9.23334509e-02
5.64282298e-01 -4.76026505e-01 -3.58156919e-01 5.87986648e-01
-6.95096493e-01 -7.64515758e-01 -3.62772614e-01 3.28810096e-01
-6.04958296e-01 7.78923333e-02 6.34976178e-02 -6.51240498e-02
-3.00098330e-01 -1.45686936e+00 8.58443022e-01 1.30852595e-01
1.80412494e-02 -1.19153821e+00 4.88929152e-01 5.01636088e-01
3.86562377e-01 7.11255223e-02 1.03825521e+00 -3.97545695e-01
-5.25934279e-01 -1.67549759e-01 -5.11133187e-02 2.94446081e-01
3.97633435e-03 3.27866286e-01 -9.65568841e-01 -5.21421432e-01
-1.88151866e-01 -2.99092084e-01 9.37775016e-01 6.45944774e-01
1.68273103e+00 -5.57530284e-01 -3.50831389e-01 1.39894295e+00
1.50696337e+00 4.66867816e-03 7.12384462e-01 4.62282807e-01
8.03059638e-01 1.72066875e-02 3.57595295e-01 3.43505114e-01
1.04740441e-01 9.57762003e-01 6.84414729e-02 4.52797115e-01
1.37973443e-01 -2.68407583e-01 9.70373079e-02 5.70469558e-01
4.22610808e-03 -3.13987732e-01 -1.03397977e+00 2.43493482e-01
-1.48148763e+00 -4.32170898e-01 5.16148508e-01 2.32836080e+00
1.08509493e+00 9.48163792e-02 1.44337758e-01 -1.06509887e-01
6.20991170e-01 -2.00808570e-01 -1.12833798e+00 -1.35685995e-01
1.66374549e-01 3.97554904e-01 7.44173586e-01 3.99424851e-01
-1.09825385e+00 9.95401978e-01 7.01471806e+00 8.51375520e-01
-1.33199823e+00 -2.32879534e-01 8.67326796e-01 -1.28920242e-01
-2.72218496e-01 6.39573392e-03 -1.16478312e+00 1.80622458e-01
1.36845303e+00 -2.61976838e-01 6.59760356e-01 6.46564484e-01
-1.70668233e-02 8.43646750e-02 -1.00881875e+00 1.11762321e+00
-1.20551931e-02 -1.55531573e+00 3.05942565e-01 7.15694502e-02
1.22770476e+00 8.71690065e-02 2.48981014e-01 2.01524049e-01
2.15045691e-01 -1.11574829e+00 2.51983345e-01 5.56734383e-01
1.03764069e+00 -8.55948448e-01 4.56299067e-01 1.92692503e-01
-7.81593561e-01 -2.53704607e-01 -3.83419663e-01 8.06552529e-01
-3.10052723e-01 4.58560318e-01 -1.17260170e+00 6.10098615e-02
5.19462943e-01 3.50178242e-01 -9.47996855e-01 1.43876374e+00
6.55381307e-02 7.85927594e-01 -5.67399085e-01 2.26969972e-01
3.88600454e-02 1.56579688e-02 3.18503946e-01 9.67902899e-01
4.85556334e-01 1.38798460e-01 -6.08073547e-03 5.83782732e-01
-2.71823049e-01 -2.68562175e-02 -1.40684843e-01 -6.31028935e-02
6.19495392e-01 1.21226442e+00 -5.35150945e-01 -1.49133652e-01
3.41609180e-01 6.74411714e-01 4.70505565e-01 3.97124738e-01
-8.39108288e-01 4.79301577e-03 4.92893547e-01 7.38780126e-02
2.45429948e-01 -1.78437099e-01 -3.23403805e-01 -1.17638481e+00
-2.48356014e-01 -1.07566392e+00 7.86374271e-01 -6.91649318e-01
-1.14352894e+00 8.34427297e-01 5.33285260e-01 -1.02330768e+00
-2.79811889e-01 -5.92051029e-01 -3.16395968e-01 5.62353253e-01
-1.28485274e+00 -9.80012655e-01 -3.96823853e-01 3.97650182e-01
2.62076616e-01 -2.47729227e-01 6.93545699e-01 5.26737496e-02
-8.27676117e-01 8.79612267e-01 2.68914431e-01 -3.72294813e-01
8.51583064e-01 -8.52542937e-01 1.70863762e-01 4.97354418e-01
2.05731034e-01 3.91815811e-01 7.58711278e-01 -4.21632469e-01
-1.41826284e+00 -1.15678537e+00 3.54168206e-01 -4.40840840e-01
5.46634674e-01 1.65813908e-01 -1.25419772e+00 6.00008965e-01
-2.20269114e-01 -1.29261211e-01 5.54651678e-01 1.52953222e-01
-3.84216309e-01 -3.90197664e-01 -1.18313062e+00 3.12034488e-01
8.12377274e-01 -2.86080837e-01 -1.93650037e-01 2.37573564e-01
5.79176605e-01 -7.52036572e-01 -1.20805597e+00 4.26496804e-01
4.95913208e-01 -6.10834002e-01 1.04554844e+00 -7.53488839e-01
1.23323344e-01 6.20954111e-02 -6.77335188e-02 -1.87822354e+00
-3.40962201e-01 -7.38824010e-01 -9.80579332e-02 8.80953968e-01
6.66899562e-01 -4.49674249e-01 8.41411471e-01 8.91409338e-01
-7.05029890e-02 -1.00206745e+00 -7.66823471e-01 -5.01464546e-01
2.92331040e-01 -3.67760509e-01 9.17393029e-01 5.80305457e-01
-6.34374559e-01 3.56554478e-01 -3.60159785e-01 1.90788612e-01
8.06674659e-01 -1.05606668e-01 9.32581723e-01 -1.23658156e+00
-4.69443053e-01 -3.85015666e-01 -1.68405741e-01 -7.77370214e-01
8.86202678e-02 -7.43690610e-01 -5.35879172e-02 -8.34531486e-01
5.37775517e-01 -1.00007212e+00 -2.44146422e-01 5.61790943e-01
-2.58625716e-01 6.47838414e-02 -3.55563760e-02 2.61454374e-01
-3.08076084e-01 4.87821966e-01 1.28394580e+00 -6.24051020e-02
-6.24511898e-01 -2.22488083e-02 -7.41674602e-01 4.03987497e-01
9.81922805e-01 -6.57265365e-01 -5.61717272e-01 -6.05561078e-01
4.11023945e-01 -5.57834618e-02 2.24872857e-01 -1.05099583e+00
1.08780205e-01 -5.43825626e-01 4.34898347e-01 -2.63974071e-01
1.57910988e-01 -7.08708227e-01 5.26593149e-01 3.42179537e-02
-6.48851275e-01 6.57545701e-02 6.22577786e-01 5.10276437e-01
1.93908498e-01 -4.61737186e-01 1.35526884e+00 2.05196932e-01
-6.93188310e-01 5.91743708e-01 1.59105584e-01 3.87191296e-01
8.50617051e-01 -9.82467830e-02 -2.65807778e-01 9.16031078e-02
-6.99010551e-01 1.90930367e-01 6.13775671e-01 4.31678057e-01
6.19539022e-01 -1.07428551e+00 -6.20300651e-01 4.31424052e-01
1.17994815e-01 1.26673579e-01 4.57102835e-01 3.79775017e-01
-3.92411113e-01 1.52711883e-01 -2.07071289e-01 -6.67284906e-01
-1.20604801e+00 4.85185236e-01 8.17475855e-01 -4.62151289e-01
-5.07605553e-01 8.25666070e-01 -1.65714428e-01 -4.00534272e-01
1.77324161e-01 -5.37253261e-01 1.05217464e-01 -1.90795302e-01
3.25989246e-01 2.04942688e-01 2.74295747e-01 -5.38766623e-01
-2.28016973e-02 9.65049684e-01 -1.59073725e-01 -1.15876263e-02
1.49363112e+00 8.95357952e-02 2.70982414e-01 2.77915031e-01
1.72815621e+00 -5.43048143e-01 -1.60372841e+00 -3.20826650e-01
-5.67619726e-02 -4.88611966e-01 5.04375458e-01 -7.78972149e-01
-1.29244554e+00 3.34761560e-01 6.62582338e-01 -3.59370917e-01
1.34012508e+00 -1.65681157e-03 7.09752738e-01 8.93932343e-01
2.54439384e-01 -1.31379342e+00 4.18005198e-01 3.75589430e-01
6.87366843e-01 -1.12420094e+00 2.31365696e-01 4.53771986e-02
-8.16255748e-01 1.29526210e+00 6.26138985e-01 -4.55680229e-02
8.41769159e-01 1.06543854e-01 -2.20959703e-03 -3.24685901e-01
-9.02477384e-01 3.68606746e-01 6.74822986e-01 4.84308392e-01
-7.02250898e-02 2.17639469e-02 4.82639998e-01 1.17689990e-01
-4.04013038e-01 1.84079856e-01 4.47276831e-01 3.88930351e-01
-4.28667575e-01 -1.27959311e+00 -2.51827925e-01 8.21036577e-01
-3.97012681e-02 1.39949888e-01 -1.50484160e-01 5.34499526e-01
5.34478724e-02 4.39650655e-01 -1.35724500e-01 -7.02166677e-01
5.16181588e-01 4.30489816e-02 7.60763764e-01 -6.32675529e-01
-7.77208865e-01 -1.57759205e-01 -1.46160290e-01 -3.38570476e-01
-2.09414631e-01 -6.17174447e-01 -1.03793383e+00 -2.24297583e-01
-5.39680362e-01 9.86258388e-02 6.59121096e-01 9.61812675e-01
4.90994155e-01 3.34641486e-01 8.05835307e-01 -8.65604997e-01
-9.26808059e-01 -6.62988126e-01 -1.62512049e-01 2.73952365e-01
2.48862416e-01 -7.36120760e-01 -3.15092534e-01 1.96364354e-02] | [9.050699234008789, 3.298794746398926] |
bb36fd84-6429-4e9e-9478-d8c43a45901a | explaining-face-presentation-attack-detection | 2111.04862 | null | https://arxiv.org/abs/2111.04862v1 | https://arxiv.org/pdf/2111.04862v1.pdf | Explaining Face Presentation Attack Detection Using Natural Language | A large number of deep neural network based techniques have been developed to address the challenging problem of face presentation attack detection (PAD). Whereas such techniques' focus has been on improving PAD performance in terms of classification accuracy and robustness against unseen attacks and environmental conditions, there exists little attention on the explainability of PAD predictions. In this paper, we tackle the problem of explaining PAD predictions through natural language. Our approach passes feature representations of a deep layer of the PAD model to a language model to generate text describing the reasoning behind the PAD prediction. Due to the limited amount of annotated data in our study, we apply a light-weight LSTM network as our natural language generation model. We investigate how the quality of the generated explanations is affected by different loss functions, including the commonly used word-wise cross entropy loss, a sentence discriminative loss, and a sentence semantic loss. We perform our experiments using face images from a dataset consisting of 1,105 bona-fide and 924 presentation attack samples. Our quantitative and qualitative results show the effectiveness of our model for generating proper PAD explanations through text as well as the power of the sentence-wise losses. To the best of our knowledge, this is the first introduction of a joint biometrics-NLP task. Our dataset can be obtained through our GitHub page. | ['Wael Abd-Almageed', 'Jonathan May', 'Leonidas Spinoulas', 'Mohamed E. Hussein', 'Hengameh Mirzaalian'] | 2021-11-08 | null | null | null | null | ['face-presentation-attack-detection'] | ['computer-vision'] | [ 4.88497913e-01 4.34336573e-01 -2.73773214e-03 -5.38581610e-01
-9.23796237e-01 -2.04352364e-01 6.63185596e-01 1.46659285e-01
-7.56355887e-03 6.00685060e-01 3.41288269e-01 -1.80728242e-01
5.35969697e-02 -6.19832218e-01 -8.15985441e-01 -3.13567191e-01
-1.43095404e-01 2.57073045e-01 -2.76305795e-01 -1.51877433e-01
1.59932747e-01 5.00424325e-01 -1.38354266e+00 7.58469522e-01
6.23200953e-01 1.26072872e+00 -3.28137010e-01 5.79481304e-01
-8.37925002e-02 5.83149850e-01 -7.51708984e-01 -1.33913267e+00
1.93506889e-02 -6.30805552e-01 -7.81102717e-01 -5.29456139e-02
6.86294734e-01 -4.00477886e-01 -3.16835612e-01 8.52364123e-01
8.09850633e-01 -4.01583493e-01 5.27936757e-01 -1.40980804e+00
-8.20905507e-01 6.74526036e-01 -4.42730606e-01 5.60108060e-03
6.23189747e-01 1.85548186e-01 1.07776570e+00 -8.01664174e-01
5.08299530e-01 1.32618558e+00 8.33028853e-01 1.05201137e+00
-1.24161935e+00 -9.01338100e-01 5.40143624e-03 2.66440779e-01
-1.25917900e+00 -5.56239903e-01 8.72893572e-01 -2.19045609e-01
8.73207569e-01 8.94242823e-02 1.75860107e-01 1.70425832e+00
4.34735984e-01 6.47647500e-01 8.80808711e-01 -5.03895581e-01
8.27895254e-02 5.09270489e-01 1.42386882e-02 7.48288870e-01
1.85116976e-01 2.57143620e-02 -7.98192978e-01 -3.04842234e-01
2.53644139e-01 -3.52009147e-01 -2.17424288e-01 1.23444118e-01
-3.36892843e-01 1.02169442e+00 3.87749732e-01 2.16457069e-01
-2.97826082e-01 9.18825045e-02 3.86624277e-01 8.17576423e-02
6.73213899e-01 3.32127243e-01 -3.79872143e-01 4.01492417e-02
-8.61272931e-01 2.67785341e-01 9.82762277e-01 4.88122255e-01
1.65079817e-01 7.01215044e-02 -3.08596134e-01 7.33679593e-01
2.63000876e-01 3.81352156e-01 4.47596461e-01 -4.22020495e-01
4.50583756e-01 3.30799103e-01 -3.39559019e-01 -1.39437258e+00
-1.90984398e-01 -2.22869456e-01 -7.11049020e-01 -7.88144618e-02
3.30251276e-01 -1.86219543e-01 -6.76753581e-01 2.15623140e+00
-1.77584335e-01 1.50989905e-01 2.21139926e-04 7.56714344e-01
9.60125864e-01 4.75992560e-01 3.62558633e-01 5.70483855e-04
1.54148495e+00 -3.36135566e-01 -6.52441502e-01 -3.73526126e-01
3.44262511e-01 -8.48241210e-01 9.63973045e-01 4.63888317e-01
-1.19992256e+00 -4.07698274e-01 -9.02295053e-01 7.82601461e-02
-2.19654381e-01 2.13428602e-01 4.49577123e-01 8.97251129e-01
-1.08545339e+00 5.75485826e-01 -3.93264920e-01 -5.51887512e-01
7.90917099e-01 4.71680641e-01 -4.81391490e-01 -5.41147999e-02
-1.53503096e+00 1.03099740e+00 2.31182147e-02 1.28999397e-01
-5.81052423e-01 -6.76978886e-01 -8.71927023e-01 3.25284064e-01
1.62581936e-01 -5.75408041e-01 1.07292402e+00 -1.05528414e+00
-1.27257681e+00 8.86865973e-01 -1.92912668e-01 -8.45472813e-01
3.48300219e-01 -2.39024520e-01 -4.02812719e-01 1.18407145e-01
-2.48099744e-01 9.00150120e-01 8.15703630e-01 -1.07172406e+00
3.32748331e-02 -3.57879639e-01 7.02986552e-04 -3.86479944e-01
-6.00495100e-01 7.87562281e-02 -1.99367970e-01 -8.19623947e-01
-4.30125266e-01 -8.43389869e-01 1.42356902e-01 1.55108362e-01
-5.34347594e-01 -1.41870022e-01 6.11681759e-01 -1.09022665e+00
1.09957755e+00 -2.16408658e+00 -2.53880769e-01 8.72805249e-03
1.06181195e-02 4.54339713e-01 -4.54135060e-01 5.49434841e-01
-3.61933708e-01 4.87621874e-01 -2.10594654e-01 -6.69550061e-01
1.42325670e-01 2.58635376e-02 -1.04649365e+00 1.99184254e-01
8.81099701e-01 7.96904564e-01 -4.49800014e-01 -3.52302372e-01
-1.25095412e-01 1.07439244e+00 -7.80827761e-01 2.04917073e-01
-2.84364223e-01 1.47577107e-01 -2.55565524e-01 3.27197105e-01
6.51763976e-01 -4.10375372e-02 8.34916160e-03 -2.23967984e-01
5.35806894e-01 3.65323514e-01 -6.13136411e-01 1.46461594e+00
-6.08021796e-01 8.11803460e-01 -3.58579010e-01 -7.80666411e-01
9.91912246e-01 4.28616166e-01 2.69745946e-01 -7.37716019e-01
2.18528867e-01 8.75814259e-02 -6.05556034e-02 -3.20982844e-01
3.42937678e-01 -4.02714342e-01 -1.98173504e-02 5.26515007e-01
1.18211009e-01 1.99740633e-01 -9.82403606e-02 2.80582309e-01
9.90783453e-01 -4.22104113e-02 6.12300076e-02 1.01529449e-01
6.04590356e-01 -3.25861424e-01 3.52232426e-01 6.82873130e-01
-1.14745416e-01 8.34963560e-01 7.78301656e-01 -4.42871928e-01
-7.90825486e-01 -9.98068929e-01 -1.70191750e-01 6.50936842e-01
-3.33488196e-01 -5.74675798e-01 -1.13446224e+00 -6.90329134e-01
-4.07623649e-02 8.63945007e-01 -7.39705265e-01 -6.07364953e-01
-5.23123980e-01 -6.83818877e-01 9.33020949e-01 3.54981899e-01
4.84204233e-01 -1.34216821e+00 -2.98116177e-01 -9.07534361e-02
-3.67620260e-01 -1.43773794e+00 -4.47096586e-01 -3.75105828e-01
-4.64188874e-01 -9.35579836e-01 -3.13838840e-01 -4.10374850e-01
5.88697016e-01 -3.38212401e-01 1.09613109e+00 2.78644085e-01
-6.79641366e-01 4.14032906e-01 -2.21864030e-01 -6.47513747e-01
-5.79607785e-01 -2.99088713e-02 7.52344057e-02 2.68633664e-01
5.32311380e-01 -2.97530085e-01 -5.33758163e-01 -2.42084134e-02
-9.56383884e-01 -1.91887468e-02 6.08715534e-01 7.62007356e-01
2.00120762e-01 -1.21454373e-01 5.91812789e-01 -6.43120170e-01
9.76198018e-01 -4.04336065e-01 -2.74385005e-01 2.35066846e-01
-5.40845752e-01 2.26071879e-01 5.74505985e-01 -5.00442743e-01
-1.02754140e+00 -5.35406657e-02 -4.96116370e-01 -3.07710409e-01
-2.20880881e-01 2.71359354e-01 -2.77455419e-01 -6.40522409e-03
6.42534018e-01 2.53501415e-01 1.61094844e-01 -2.40477368e-01
1.92841798e-01 6.42795503e-01 3.88585329e-01 -5.26167989e-01
7.37670124e-01 2.48647943e-01 1.64432880e-02 -7.31693923e-01
-9.44939196e-01 3.64544988e-01 -1.74630702e-01 -1.03233300e-01
7.80132532e-01 -5.34397423e-01 -9.07143831e-01 2.26911202e-01
-1.43624651e+00 1.07787073e-01 -1.57209277e-01 3.17949317e-02
-5.20413876e-01 5.07273138e-01 -7.13472366e-01 -9.84742820e-01
-7.31015921e-01 -1.09202957e+00 1.22591352e+00 4.53017885e-03
-5.58004558e-01 -9.85827506e-01 -2.43183061e-01 4.80356097e-01
5.29336751e-01 3.97982866e-01 9.75977063e-01 -8.50959837e-01
-2.32291609e-01 -5.79462707e-01 -2.40401357e-01 4.70538437e-01
-2.57900003e-02 -1.72153413e-01 -1.31861401e+00 -1.29985303e-01
4.72792648e-02 -4.82738942e-01 9.36050653e-01 3.28177452e-01
1.47786546e+00 -7.41304219e-01 -5.30862622e-02 2.61303276e-01
1.17666757e+00 -1.07623033e-01 8.63319993e-01 -1.17798395e-01
2.88255215e-01 8.96670461e-01 3.50593239e-01 4.24834579e-01
1.99492246e-01 8.28500152e-01 4.03335094e-01 -6.25486150e-02
-2.34936342e-01 -5.81586540e-01 5.94084680e-01 9.01548713e-02
4.67572987e-01 -6.09233260e-01 -9.54088390e-01 2.27948621e-01
-1.45963526e+00 -1.08547091e+00 1.37035578e-01 2.10155439e+00
7.42296338e-01 2.08980486e-01 5.43288840e-03 1.58722118e-01
6.79453850e-01 5.79562783e-02 -2.91157454e-01 -8.98574948e-01
-5.30794486e-02 3.96050483e-01 -7.82129392e-02 5.13018966e-01
-1.00680459e+00 8.45063388e-01 5.72263002e+00 8.18038881e-01
-1.33254814e+00 -3.02361343e-02 1.02878797e+00 -1.24994002e-01
-2.00607896e-01 -3.35487574e-01 -8.64951372e-01 5.18480241e-01
1.25067270e+00 -3.85464430e-02 2.33984306e-01 6.12522304e-01
1.58550844e-01 2.65146434e-01 -1.25962937e+00 9.00038540e-01
5.53342640e-01 -1.17240214e+00 4.54141974e-01 2.45252967e-01
2.16809973e-01 -4.95197415e-01 6.04492903e-01 5.49942628e-02
-1.46934941e-01 -1.32178628e+00 5.72939515e-01 4.92092282e-01
6.62504017e-01 -6.95816278e-01 7.95734286e-01 8.29994306e-02
-6.95528269e-01 -1.93655357e-01 -8.30788091e-02 2.72593275e-02
2.56133210e-02 5.39087296e-01 -1.13417220e+00 2.62116462e-01
4.15473610e-01 2.71489441e-01 -5.69865704e-01 6.10158563e-01
-3.27764004e-01 6.32419646e-01 -9.88872051e-02 -1.29817026e-02
-2.07523303e-03 3.75211895e-01 3.65598202e-01 1.17947328e+00
4.20874149e-01 1.04318239e-01 -2.60682374e-01 1.11887217e+00
-2.98619598e-01 2.11876094e-01 -5.93143582e-01 -3.90040189e-01
1.61658466e-01 1.11469483e+00 -2.53945827e-01 1.80893596e-02
-1.28665984e-01 1.00274920e+00 2.60199904e-01 6.27491251e-02
-9.13253069e-01 -2.27309301e-01 7.04563558e-01 3.44258338e-01
5.79927824e-02 1.96637735e-01 -4.72474337e-01 -1.02825880e+00
2.60862172e-01 -9.20736074e-01 5.29819667e-01 -8.21036696e-01
-1.62419260e+00 1.06598222e+00 -1.19758867e-01 -9.02193308e-01
-5.09975433e-01 -7.31982410e-01 -7.09854305e-01 1.02126920e+00
-1.36484754e+00 -1.24493182e+00 -6.02724589e-02 2.93758661e-01
5.24477541e-01 -3.11608374e-01 1.08566797e+00 3.39200854e-01
-5.87665677e-01 1.08558095e+00 -6.39367461e-01 2.61225730e-01
6.91611469e-01 -8.05066407e-01 6.08566105e-01 7.74782121e-01
1.65369272e-01 7.31321514e-01 8.75272036e-01 -5.48832595e-01
-1.05956507e+00 -9.76246357e-01 1.35434675e+00 -4.94035244e-01
5.27289808e-01 -5.90886056e-01 -9.58903432e-01 3.95803928e-01
2.18977004e-01 -2.14616075e-01 9.99891639e-01 -6.40556142e-02
-5.17637908e-01 -1.65267289e-01 -1.44388568e+00 6.50041044e-01
8.68379653e-01 -5.26753902e-01 -3.01101089e-01 4.33084548e-01
5.07618725e-01 -1.35811970e-01 -7.35064566e-01 4.47756022e-01
6.68229699e-01 -8.78136337e-01 9.52832997e-01 -8.67870808e-01
9.55511689e-01 2.96075642e-01 -4.10436280e-02 -1.09401107e+00
1.17911182e-01 -3.54104877e-01 9.80107337e-02 1.46309936e+00
4.99409765e-01 -5.29924035e-01 8.88977468e-01 9.73461032e-01
2.22988889e-01 -9.27453339e-01 -8.53250563e-01 -4.10345137e-01
1.84753597e-01 -6.43857598e-01 6.70606434e-01 6.76454365e-01
-2.77632158e-02 2.20107049e-01 -6.14217997e-01 1.18757807e-01
6.09133780e-01 -1.23219736e-01 5.99225223e-01 -1.17028177e+00
-3.44694316e-01 -4.30115074e-01 -4.91385013e-01 -3.80710572e-01
6.44383669e-01 -8.98427367e-01 -6.48830980e-02 -1.24693775e+00
3.05532575e-01 -6.59257174e-03 -7.05835968e-02 7.56079435e-01
2.41702143e-03 4.06710267e-01 3.93787920e-01 -1.31350622e-01
-6.19715229e-02 5.52874386e-01 7.68839180e-01 -6.85883984e-02
1.32519439e-01 -1.59756675e-01 -1.04297173e+00 6.52845919e-01
1.05904722e+00 -5.09415507e-01 -4.02707815e-01 -5.38745463e-01
-1.29528958e-02 1.38799459e-01 6.51316881e-01 -7.60796964e-01
4.97024157e-04 1.12395465e-01 4.95305717e-01 -2.47939583e-02
6.54036999e-01 -5.81842899e-01 -6.65453589e-03 6.16204441e-01
-6.63086653e-01 -8.91632736e-02 6.04875803e-01 3.04112405e-01
-1.50173008e-01 2.21751239e-02 6.61248624e-01 1.58916995e-01
-2.26696998e-01 3.46543550e-01 -9.79669392e-02 -5.40759563e-02
7.48219907e-01 -2.07796976e-01 -5.13000965e-01 -7.76661277e-01
-6.79589093e-01 -1.97827950e-01 9.02621076e-02 8.33050132e-01
9.85021710e-01 -1.26736259e+00 -9.96881247e-01 4.82819736e-01
-4.48083039e-03 -8.19650888e-01 2.43771985e-01 5.59105515e-01
-9.48089957e-02 6.64263666e-01 -2.43726864e-01 -2.85896927e-01
-1.39808524e+00 3.30442071e-01 3.77194256e-01 -1.51956916e-01
-4.79598373e-01 8.38752568e-01 2.76493728e-01 -1.85994253e-01
2.75926113e-01 1.79262593e-01 -9.70280915e-02 -1.04850441e-01
7.24407613e-01 -1.61028415e-01 8.31993893e-02 -7.44335830e-01
-5.01783073e-01 2.83127218e-01 -2.28868276e-01 -9.37049910e-02
1.26277554e+00 2.35663503e-01 -6.67549893e-02 -1.99320108e-01
1.18069315e+00 -1.83885634e-01 -8.22809696e-01 -3.11111379e-02
-8.36877748e-02 -5.06189883e-01 -2.09128305e-01 -1.07506454e+00
-1.11431634e+00 1.25442088e+00 7.51448274e-01 6.28641294e-03
1.33975267e+00 1.69995166e-02 9.41890478e-01 7.39111751e-02
1.33696645e-01 -5.49362004e-01 2.82573044e-01 2.37900376e-01
1.18316579e+00 -1.17465520e+00 -3.93518716e-01 -4.18911219e-01
-7.55454600e-01 9.57336843e-01 7.06806004e-01 6.75844327e-02
4.33305502e-01 2.80640155e-01 -1.04104407e-01 -3.27423401e-02
-8.55981708e-01 1.02293305e-01 4.46592420e-01 5.80198109e-01
7.12319195e-01 -2.15806961e-01 -3.57474238e-01 1.02253711e+00
-5.27256191e-01 4.98230979e-02 4.64166015e-01 3.46923321e-01
-5.23711322e-03 -1.23697150e+00 -2.42750362e-01 3.06194693e-01
-8.55002761e-01 -1.43471241e-01 -9.60227013e-01 5.01934707e-01
-7.23147579e-03 1.00608826e+00 -5.29452339e-02 -5.70531785e-01
2.69603163e-01 3.21799695e-01 3.86794597e-01 -3.88540149e-01
-5.76991975e-01 -5.20914972e-01 2.17694446e-01 -4.56304908e-01
2.97406372e-02 -6.59516156e-01 -1.19630051e+00 -5.40206373e-01
-5.51835261e-02 8.71497691e-02 8.18175018e-01 9.93824959e-01
6.29956424e-01 2.77314812e-01 4.98643160e-01 -4.95542496e-01
-6.38107061e-01 -7.91157782e-01 -2.47213647e-01 5.33712566e-01
1.68772802e-01 -4.31888074e-01 -2.72324055e-01 1.08741038e-02] | [12.91537094116211, 1.066261887550354] |
879e4d11-bb8b-43da-b0ec-c20e1f3191e8 | direct-output-connection-for-a-high-rank | 1808.10143 | null | http://arxiv.org/abs/1808.10143v2 | http://arxiv.org/pdf/1808.10143v2.pdf | Direct Output Connection for a High-Rank Language Model | This paper proposes a state-of-the-art recurrent neural network (RNN)
language model that combines probability distributions computed not only from a
final RNN layer but also from middle layers. Our proposed method raises the
expressive power of a language model based on the matrix factorization
interpretation of language modeling introduced by Yang et al. (2018). The
proposed method improves the current state-of-the-art language model and
achieves the best score on the Penn Treebank and WikiText-2, which are the
standard benchmark datasets. Moreover, we indicate our proposed method
contributes to two application tasks: machine translation and headline
generation. Our code is publicly available at:
https://github.com/nttcslab-nlp/doc_lm. | ['Jun Suzuki', 'Sho Takase', 'Masaaki Nagata'] | 2018-08-30 | direct-output-connection-for-a-high-rank-1 | https://aclanthology.org/D18-1489 | https://aclanthology.org/D18-1489.pdf | emnlp-2018-10 | ['headline-generation'] | ['natural-language-processing'] | [-1.80124238e-01 1.56491637e-01 -5.89245617e-01 -1.17172375e-01
-1.07227528e+00 -5.12391984e-01 6.88117683e-01 -3.53497356e-01
-5.56905329e-01 1.05138397e+00 5.87582350e-01 -9.79057252e-01
3.94036144e-01 -5.96014082e-01 -8.52598369e-01 -4.38003868e-01
4.26196307e-01 6.10311329e-01 -2.30852097e-01 -3.25202554e-01
-1.82952434e-01 2.50430703e-01 -8.18275511e-01 4.51176494e-01
8.13443720e-01 4.80422258e-01 4.68367428e-01 5.77190638e-01
-4.93940771e-01 9.47914183e-01 -2.22589508e-01 -6.12696469e-01
-4.93062697e-02 -3.74900639e-01 -7.93071687e-01 -4.37807918e-01
9.95997414e-02 -1.49805397e-01 -6.31080925e-01 1.06167233e+00
4.95711714e-01 1.55455485e-01 4.80294555e-01 -8.22203457e-01
-1.27325583e+00 1.54809988e+00 -4.20710832e-01 1.34219617e-01
-2.79208459e-02 -1.00130104e-01 1.19655740e+00 -1.06944990e+00
6.07471049e-01 1.32918119e+00 3.59856695e-01 9.48336244e-01
-1.01108801e+00 -8.16761136e-01 3.91380310e-01 8.09779763e-02
-1.25168014e+00 -5.25441229e-01 7.89766967e-01 -1.98784813e-01
1.25097883e+00 -1.40361637e-01 2.18307883e-01 1.55109060e+00
1.39064401e-01 1.42475963e+00 9.28759098e-01 -6.24290586e-01
-2.02881098e-01 6.57153428e-02 3.15859556e-01 6.07780814e-01
-5.04928865e-02 -1.52495578e-01 -4.78202194e-01 -1.38504758e-01
8.11341107e-01 -2.38156930e-01 -2.00706601e-01 1.81162491e-01
-1.23603582e+00 8.37704062e-01 -4.11181332e-04 6.36368930e-01
-4.49745804e-01 3.14992219e-01 4.36103731e-01 1.52097836e-01
8.11284959e-01 2.10832179e-01 -8.48980665e-01 -4.05075461e-01
-1.00505245e+00 -2.83762198e-02 7.27294326e-01 8.84566188e-01
5.13886094e-01 3.24172825e-01 -2.44886428e-01 1.03799760e+00
4.83456403e-01 5.01136422e-01 8.05128574e-01 -5.88550389e-01
8.14601600e-01 3.84212285e-01 -7.87987486e-02 -3.13899070e-01
-2.57541627e-01 -6.04241014e-01 -9.64769185e-01 -6.93619430e-01
2.54409105e-01 -5.18512309e-01 -7.51871645e-01 1.98505712e+00
-5.77571802e-02 3.13925594e-01 3.77517849e-01 4.95108932e-01
9.83772635e-01 1.24590623e+00 6.68355962e-03 -2.33707979e-01
1.17074418e+00 -1.39421248e+00 -8.99729609e-01 -1.29699320e-01
9.35004354e-01 -9.59823251e-01 1.17606163e+00 3.61129582e-01
-1.27428925e+00 -5.28354168e-01 -5.74378729e-01 -3.45439792e-01
-2.66646147e-01 7.58835912e-01 7.35677660e-01 3.39860678e-01
-1.23046494e+00 3.62136066e-01 -1.00341415e+00 -2.66701281e-01
-6.15921877e-02 7.54682943e-02 -3.39514524e-01 1.38482243e-01
-1.65328217e+00 9.99022067e-01 4.55721974e-01 4.15373415e-01
-6.74841821e-01 -5.60054421e-01 -8.19406748e-01 2.02482827e-02
9.28707272e-02 -7.95374334e-01 1.61180925e+00 -7.71823704e-01
-2.09995651e+00 5.35472453e-01 -4.89123374e-01 -7.30754852e-01
4.54365194e-01 -9.10310566e-01 -3.90603423e-01 -2.36447245e-01
-1.47323221e-01 5.10321558e-01 2.14161277e-01 -8.69702876e-01
-4.45049793e-01 -2.10383609e-02 -1.72674119e-01 2.02598516e-03
-3.70224983e-01 3.87717783e-01 -5.46200335e-01 -7.24365592e-01
-2.62113869e-01 -8.78427744e-01 -3.84575725e-01 -5.90134382e-01
-6.38426661e-01 -6.00596964e-01 3.09901744e-01 -9.55090821e-01
1.61483550e+00 -1.69892049e+00 3.46892625e-01 -2.31483027e-01
-2.91809767e-01 5.16515017e-01 -4.79382455e-01 7.73244977e-01
-2.16365382e-01 2.59566367e-01 -1.18384548e-01 -7.11402357e-01
9.49396193e-02 8.87829363e-02 -7.77978420e-01 1.94254890e-01
1.08200073e-01 1.18783951e+00 -7.72244334e-01 -1.47430331e-01
6.95880055e-02 8.07449460e-01 -1.60921678e-01 1.81466833e-01
-5.12484014e-01 3.15745890e-01 -1.66780949e-01 2.14598745e-01
4.49663848e-01 -8.19030032e-02 2.42911577e-01 1.43591151e-01
-2.70613074e-01 9.85339403e-01 -7.42858291e-01 2.02404761e+00
-7.13355958e-01 4.59879428e-01 -3.16442668e-01 -7.24881470e-01
9.53659892e-01 6.14195228e-01 1.59946859e-01 -3.01566333e-01
2.25482240e-01 2.11498335e-01 1.58165619e-01 -1.77467659e-01
7.40744233e-01 1.08556941e-01 -5.35105690e-02 7.74810553e-01
4.73024338e-01 1.51366159e-01 3.14183354e-01 3.56293350e-01
7.31953621e-01 4.83330935e-01 2.80609548e-01 -3.13354015e-01
6.34514213e-01 -5.15116096e-01 5.78775883e-01 7.09191382e-01
1.74163237e-01 3.43596607e-01 4.98054266e-01 -1.31603241e-01
-1.05311465e+00 -1.13866866e+00 1.26272421e-02 1.34864104e+00
-6.82525992e-01 -5.87334871e-01 -6.96348488e-01 -6.32127404e-01
-3.43040615e-01 1.10482466e+00 -5.45362175e-01 1.79037005e-01
-7.48783350e-01 -7.73544908e-01 1.17303467e+00 4.53716218e-01
2.16517687e-01 -1.20290196e+00 3.26630533e-01 3.26761097e-01
-5.96692204e-01 -1.19737899e+00 -4.81034875e-01 6.24391101e-02
-8.78829122e-01 -4.29538935e-01 -9.55665052e-01 -8.50368261e-01
4.24321026e-01 -9.04398263e-02 1.13912880e+00 -3.39618325e-01
4.64686364e-01 4.76807170e-03 -4.28180754e-01 -4.97269541e-01
-7.29678869e-01 6.59623265e-01 2.27864608e-01 -5.15579954e-02
4.36079204e-01 -6.03526831e-01 -1.17699854e-01 -1.37910008e-01
-7.67494261e-01 4.29465055e-01 5.81817746e-01 7.28889108e-01
6.26894593e-01 -4.51846153e-01 7.23442614e-01 -1.08552778e+00
9.78862464e-01 -5.56455612e-01 -7.46163011e-01 5.22561014e-01
-2.52003759e-01 5.00159502e-01 8.97569954e-01 -4.72799927e-01
-1.27254951e+00 -1.21341467e-01 -4.64233935e-01 -2.92292833e-01
-1.03763916e-01 7.73282707e-01 -1.44508928e-01 5.69432855e-01
2.06192404e-01 6.42692327e-01 -3.91203463e-01 -8.09866846e-01
8.76328647e-01 7.14106560e-01 4.11860824e-01 -8.30957949e-01
6.19885981e-01 7.67294094e-02 -4.43231106e-01 -5.96794009e-01
-7.79466808e-01 -4.14490998e-01 -8.27634871e-01 1.57662868e-01
5.12469232e-01 -1.16911685e+00 -4.78834033e-01 5.27455449e-01
-1.72205055e+00 -5.14366388e-01 7.21529871e-02 6.62212610e-01
-4.70221221e-01 3.36255670e-01 -1.20701361e+00 -9.52384531e-01
-7.97461689e-01 -9.73119020e-01 9.21656489e-01 1.16265401e-01
-1.13312475e-01 -1.18915963e+00 3.35369021e-01 1.88050747e-01
4.34140652e-01 -3.55415910e-01 9.48566556e-01 -7.58841991e-01
-3.05075794e-01 1.96900934e-01 -4.19621803e-02 5.03567100e-01
3.01836804e-02 1.34197652e-01 -9.49860394e-01 -2.04203147e-02
-2.12811783e-01 -3.51123780e-01 1.24983239e+00 4.35844839e-01
9.51013505e-01 -4.26645428e-01 -3.78176905e-02 4.77897406e-01
1.14770484e+00 -8.85505080e-02 6.93714440e-01 2.14934528e-01
7.36556649e-01 5.30715168e-01 1.35612056e-01 9.37174708e-02
5.59907317e-01 3.60833406e-01 -1.54161960e-01 -5.30355424e-02
-1.52496397e-01 -6.36884332e-01 1.01447272e+00 1.82574368e+00
5.59904277e-02 -5.09722829e-01 -1.03285086e+00 6.19673908e-01
-2.29290891e+00 -7.92118013e-01 -4.85216193e-02 1.78156960e+00
1.20462334e+00 7.13460520e-02 -6.44565597e-02 -3.86995673e-01
6.14668250e-01 2.39534482e-01 -1.83559045e-01 -7.03600526e-01
-2.81506807e-01 4.12808537e-01 4.61099297e-01 9.22473431e-01
-8.84902656e-01 1.60834813e+00 5.62893677e+00 1.14037526e+00
-1.20693409e+00 3.13240767e-01 6.08818471e-01 -1.52000366e-02
-5.19825876e-01 -4.23472524e-02 -1.30038822e+00 3.83156419e-01
1.43362808e+00 -2.55937397e-01 3.36430401e-01 7.52757311e-01
3.65364611e-01 2.39864737e-01 -9.48358476e-01 7.63716221e-01
1.73539147e-01 -1.46911073e+00 4.19563383e-01 -5.38051762e-02
5.84143996e-01 7.70090342e-01 2.22330377e-01 6.01088881e-01
7.02663243e-01 -1.05836093e+00 4.93036032e-01 5.31335771e-01
6.48172319e-01 -8.52049947e-01 7.51954257e-01 6.20997012e-01
-1.02260149e+00 2.98466146e-01 -4.99725044e-01 -1.27116218e-01
5.02088428e-01 7.55471289e-01 -9.51486647e-01 6.43768907e-01
1.10576339e-01 7.73132503e-01 -3.10566008e-01 5.55195987e-01
-8.42799127e-01 1.36763048e+00 -2.54392177e-01 -1.97010025e-01
5.24055421e-01 -2.49977157e-01 5.12695134e-01 1.57375526e+00
2.90608913e-01 -1.93069607e-01 5.40391281e-02 9.69542980e-01
-5.32920480e-01 4.43489254e-01 -5.90486526e-01 -4.13213491e-01
4.04304832e-01 1.28002799e+00 -3.62726212e-01 -3.69029433e-01
-4.79636222e-01 8.14544857e-01 7.65816092e-01 5.65053999e-01
-6.81271136e-01 -2.23084435e-01 5.63824415e-01 -2.79544741e-01
1.21907659e-01 -5.72897971e-01 -9.45468247e-02 -1.58607042e+00
-1.16161801e-01 -7.66221344e-01 9.43355858e-02 -7.02545464e-01
-1.36483252e+00 9.28407490e-01 -1.22045837e-01 -8.74323249e-01
-6.94183707e-01 -7.79516459e-01 -5.07500410e-01 1.15006590e+00
-1.57198894e+00 -1.49621868e+00 7.29851007e-01 3.66084397e-01
6.58898771e-01 -4.94233578e-01 1.10990894e+00 3.96156251e-01
-7.54662097e-01 7.90843487e-01 4.22883332e-01 5.92492819e-01
6.61149263e-01 -9.84005928e-01 1.09105790e+00 1.14633238e+00
5.04986346e-01 9.98868942e-01 4.48464543e-01 -5.40858746e-01
-1.29835832e+00 -1.18235576e+00 1.65660655e+00 -5.19302487e-01
1.18077135e+00 -7.43144751e-01 -8.29933524e-01 1.13925958e+00
7.13149190e-01 -3.54403943e-01 7.89707422e-01 3.12084347e-01
-3.82104158e-01 2.93887287e-01 -5.10984957e-01 8.78550231e-01
8.17010581e-01 -7.12970495e-01 -6.48248434e-01 3.21356803e-01
1.10551906e+00 -3.27962667e-01 -5.96960425e-01 3.31194252e-01
5.79087138e-01 -4.00797099e-01 5.83726764e-01 -8.18059683e-01
7.22020209e-01 -1.27570122e-01 -1.43131465e-01 -1.51583385e+00
-4.13065404e-01 -8.99586022e-01 -3.98831964e-01 1.38047397e+00
1.01153910e+00 -6.62464857e-01 5.00783920e-01 1.24027945e-01
-1.01561256e-01 -9.91733551e-01 -8.60972345e-01 -7.00000226e-01
6.45020187e-01 -8.11523139e-01 4.40635860e-01 8.58751774e-01
3.69656943e-02 6.21521592e-01 -7.06363022e-01 -5.58171421e-02
1.94139943e-01 -2.02767998e-01 7.01534271e-01 -7.72527516e-01
-6.34406984e-01 -5.11448443e-01 2.11773977e-01 -1.39020884e+00
8.13317657e-01 -1.30070853e+00 -1.55488655e-01 -1.75794625e+00
7.34884739e-02 -4.17353958e-02 -5.83846927e-01 6.21567607e-01
-1.29040673e-01 8.26526526e-03 4.95452583e-01 -2.19937842e-02
-4.44746971e-01 7.10220277e-01 9.31549728e-01 3.69086005e-02
-2.16091365e-01 -4.46193218e-02 -7.61467218e-01 6.62227154e-01
1.18336785e+00 -4.57534492e-01 -2.50066996e-01 -8.68186116e-01
4.33237433e-01 6.55355006e-02 -1.07111178e-01 -4.86552596e-01
2.49086753e-01 7.87350070e-03 1.62920490e-01 -7.68928111e-01
3.47410917e-01 -2.53317773e-01 -1.61684960e-01 3.55602145e-01
-7.12226927e-01 4.01166409e-01 4.28613573e-01 5.37946373e-02
-2.01867878e-01 -2.36133367e-01 4.28858578e-01 -2.62467027e-01
-3.24337363e-01 2.54790872e-01 -5.55233002e-01 3.09002511e-02
2.60749996e-01 5.49686968e-01 -5.04868805e-01 -5.91995358e-01
-4.30728555e-01 1.55427232e-01 -1.78461030e-01 7.66695917e-01
4.27498788e-01 -1.31648874e+00 -1.17902946e+00 3.44123580e-02
-2.15656251e-01 -2.23451272e-01 6.67978972e-02 7.48834491e-01
-1.65102065e-01 9.15493846e-01 1.76770866e-01 -1.14127509e-01
-8.82131875e-01 2.03079656e-01 1.54071867e-01 -8.32876146e-01
-3.87436926e-01 1.07121420e+00 2.22238824e-01 -9.42406237e-01
2.74814188e-01 -4.63327378e-01 -2.34966978e-01 -2.26581171e-01
5.94549239e-01 1.68759510e-01 -1.72454983e-01 -7.68882632e-01
-2.70566076e-01 2.51452982e-01 -5.60802817e-01 -4.85365540e-01
1.27650714e+00 -5.54004051e-02 -4.66735661e-01 9.78828013e-01
1.11670089e+00 1.66869536e-01 -6.13144755e-01 -5.43294847e-01
9.79556069e-02 3.49548429e-01 -3.17857713e-02 -9.53139961e-01
-8.26320887e-01 1.16085005e+00 -2.35139877e-02 -4.09311593e-01
7.21484661e-01 -9.93040577e-02 1.18192828e+00 6.64816678e-01
2.07168177e-01 -1.14923191e+00 -4.80120063e-01 1.27433586e+00
8.10066819e-01 -9.21301901e-01 -2.13486865e-01 -7.59916613e-03
-5.87778330e-01 1.19657290e+00 5.30341029e-01 -2.39785686e-01
8.16455364e-01 4.53115642e-01 3.23410898e-01 4.47215438e-01
-1.30526137e+00 -1.80722307e-02 3.90208155e-01 1.55345380e-01
1.21044815e+00 3.17517012e-01 -7.00216115e-01 1.01211870e+00
-5.52623570e-01 2.38923460e-01 5.26745260e-01 3.74804765e-01
-1.50313556e-01 -1.49091876e+00 -6.18876554e-02 1.74114376e-01
-7.62756944e-01 -8.49191129e-01 -4.69522715e-01 4.61279899e-01
-2.61777639e-01 7.72815645e-01 -1.77751690e-01 -3.79578441e-01
-1.50215119e-01 4.66532588e-01 2.33299732e-01 -6.87037289e-01
-5.71542501e-01 2.67166197e-01 2.60163069e-01 -3.10055643e-01
-3.75859857e-01 -3.92407984e-01 -1.33928418e+00 -1.19907521e-01
-2.66181648e-01 4.92669344e-01 7.80758679e-01 9.38445449e-01
4.66488093e-01 5.39907217e-01 2.65835077e-01 -6.25127852e-01
-4.77044255e-01 -1.46966994e+00 -2.62564719e-01 -2.61656910e-01
2.39731789e-01 -1.43047184e-01 -1.98807284e-01 1.22197559e-02] | [11.662351608276367, 9.304156303405762] |
6b656e19-39f9-4151-b4b8-0da8699c657a | leveraging-bert-for-extractive-text | 1906.04165 | null | https://arxiv.org/abs/1906.04165v1 | https://arxiv.org/pdf/1906.04165v1.pdf | Leveraging BERT for Extractive Text Summarization on Lectures | In the last two decades, automatic extractive text summarization on lectures has demonstrated to be a useful tool for collecting key phrases and sentences that best represent the content. However, many current approaches utilize dated approaches, producing sub-par outputs or requiring several hours of manual tuning to produce meaningful results. Recently, new machine learning architectures have provided mechanisms for extractive summarization through the clustering of output embeddings from deep learning models. This paper reports on the project called Lecture Summarization Service, a python based RESTful service that utilizes the BERT model for text embeddings and KMeans clustering to identify sentences closes to the centroid for summary selection. The purpose of the service was to provide students a utility that could summarize lecture content, based on their desired number of sentences. On top of the summary work, the service also includes lecture and summary management, storing content on the cloud which can be used for collaboration. While the results of utilizing BERT for extractive summarization were promising, there were still areas where the model struggled, providing feature research opportunities for further improvement. | ['Derek Miller'] | 2019-06-07 | null | null | null | null | ['extractive-document-summarization'] | ['natural-language-processing'] | [-2.97371864e-01 1.77259609e-01 -6.16083555e-02 -4.99337971e-01
-1.24330854e+00 -6.53391719e-01 3.03124398e-01 8.80589545e-01
-9.32305530e-02 5.35300553e-01 8.79050910e-01 -2.20669419e-01
-3.10122129e-02 -5.63522220e-01 -3.92172873e-01 -5.98912418e-01
2.47910574e-01 4.43535715e-01 6.98852837e-02 -2.63243347e-01
8.03759754e-01 5.47887802e-01 -1.70232630e+00 6.27465546e-01
7.86870897e-01 4.52371567e-01 1.79551810e-01 1.37674701e+00
-7.19981074e-01 6.67722344e-01 -1.24950397e+00 -2.32273012e-01
-1.63580075e-01 -4.89839137e-01 -9.50131297e-01 -2.92844713e-01
8.66277218e-01 -2.54595757e-01 -1.52247712e-01 6.55206859e-01
1.03008544e+00 4.05093640e-01 4.17332351e-01 -1.05127859e+00
-4.02124882e-01 8.82816374e-01 1.20630816e-01 3.42775166e-01
6.16766691e-01 -2.16228783e-01 1.22877407e+00 -6.25908971e-01
4.07170177e-01 8.06775212e-01 5.99163651e-01 5.82991958e-01
-7.96020091e-01 -4.00677919e-01 -2.62594581e-01 2.64804810e-01
-8.56387258e-01 -6.10105038e-01 8.30089033e-01 -1.11927249e-01
1.43883181e+00 8.06662917e-01 8.25858235e-01 7.65564084e-01
2.62865484e-01 1.07426465e+00 3.59870464e-01 -5.84363759e-01
4.43910748e-01 6.86771810e-01 7.36096203e-01 5.72224021e-01
3.13808709e-01 -1.11165154e+00 -8.90313983e-01 -2.51797259e-01
-1.22436047e-01 5.44727519e-02 8.60077853e-04 -1.20158726e-02
-8.60761166e-01 7.87152946e-01 8.19696486e-02 3.77884448e-01
-4.25137341e-01 -1.16128936e-01 7.24803507e-01 2.21842483e-01
6.92178547e-01 8.63366961e-01 -5.20794630e-01 -8.85138571e-01
-1.52765465e+00 4.65885669e-01 1.36599290e+00 1.02011931e+00
5.27861297e-01 4.06724364e-02 -3.70839983e-01 8.01113307e-01
2.63632666e-02 1.42477881e-02 8.51992130e-01 -8.46746385e-01
6.28043652e-01 9.28437710e-01 -1.48182645e-01 -8.18008065e-01
-2.42346704e-01 -4.79274035e-01 -3.10607970e-01 -2.85445482e-01
2.51675472e-02 -4.31693166e-01 -3.53342801e-01 9.43210661e-01
1.45913810e-01 -2.90803332e-02 2.11066812e-01 4.48466450e-01
1.43119526e+00 9.61912930e-01 -1.87530547e-01 5.69323860e-02
1.19285524e+00 -1.12742603e+00 -7.23701060e-01 1.98656932e-01
1.01774204e+00 -1.19875550e+00 1.08679354e+00 3.58180046e-01
-1.40712786e+00 -3.16933781e-01 -1.03316951e+00 -2.11392224e-01
-4.58103806e-01 2.84017064e-03 2.40899056e-01 7.06771135e-01
-1.45406067e+00 9.20785010e-01 -8.50412786e-01 -7.41176605e-01
3.12734574e-01 2.56945044e-01 -1.03700928e-01 8.04353431e-02
-4.70080167e-01 7.46757627e-01 9.82320383e-02 -3.24683100e-01
-3.35608125e-01 -1.10098541e+00 -7.53894627e-01 6.46081805e-01
-3.03960919e-01 -7.19020963e-01 1.59109080e+00 -3.50324124e-01
-1.56089473e+00 4.19898212e-01 -1.83780506e-01 -4.26969320e-01
1.86216652e-01 -4.60590959e-01 1.50083154e-01 3.79974931e-01
1.13939606e-01 4.93122339e-01 1.14784554e-01 -7.98430622e-01
-6.03030324e-01 -2.55828351e-01 -3.42047274e-01 6.52677596e-01
-8.93097758e-01 1.30634159e-01 -1.86747283e-01 -2.20921233e-01
-9.42965373e-02 -6.29661977e-01 -9.59091038e-02 -9.20727968e-01
-6.56857491e-01 -7.39735305e-01 1.00989342e+00 -9.24004376e-01
1.33496547e+00 -1.71644795e+00 -4.91363518e-02 -1.77343532e-01
4.48712446e-02 4.66403157e-01 -1.72541495e-02 1.21461582e+00
1.02344126e-01 3.86481136e-01 3.47980648e-01 -7.21671581e-01
1.27012745e-01 -2.53174216e-01 -4.64598715e-01 -6.11748174e-02
9.47848633e-02 5.80343246e-01 -8.41273665e-01 -7.37666130e-01
2.02829197e-01 3.28319490e-01 -4.81080294e-01 4.10234094e-01
-9.27022547e-02 -5.50066046e-02 -5.17833889e-01 3.32574755e-01
3.39931816e-01 1.52160376e-01 -1.19632065e-01 2.43975282e-01
-4.63932276e-01 8.53161812e-01 -9.81797040e-01 1.65704691e+00
-4.50693101e-01 1.21596909e+00 6.74071535e-02 -9.29260135e-01
9.37898993e-01 4.44345087e-01 6.66307330e-01 -1.73483603e-02
7.13131279e-02 2.96859425e-02 -4.38952535e-01 -8.51503730e-01
1.21744764e+00 3.52208912e-01 -1.12116940e-01 7.90532172e-01
2.49534607e-01 -6.22511327e-01 4.43558365e-01 6.61713123e-01
1.46427679e+00 1.45294573e-02 1.16894282e-01 -1.09955400e-01
2.59928197e-01 3.34984869e-01 4.46013696e-02 7.29358912e-01
-1.33448377e-01 7.90924311e-01 4.93538350e-01 -3.33840460e-01
-1.05612934e+00 -7.63382554e-01 1.49907276e-01 1.28214014e+00
-5.19030869e-01 -8.41183245e-01 -1.02050829e+00 -5.93625367e-01
-3.55898529e-01 1.12418163e+00 3.18898968e-02 -1.92058906e-01
-4.45017517e-01 -2.37531543e-01 5.22404313e-01 4.55279768e-01
1.79753825e-01 -1.24975669e+00 -6.63230300e-01 3.29557031e-01
-2.36777619e-01 -7.04040527e-01 -4.64209408e-01 1.91568449e-01
-9.76360261e-01 -5.20854354e-01 -6.71002686e-01 -1.00928628e+00
5.56952834e-01 6.02098346e-01 9.44373012e-01 2.60936245e-02
-4.61688608e-01 8.19591701e-01 -3.73852432e-01 -9.17183816e-01
-4.87776220e-01 6.44898176e-01 -1.50345325e-01 -7.22020924e-01
5.53099215e-01 -6.80870116e-01 -5.91533959e-01 -5.35503149e-01
-9.65304077e-01 2.18203962e-02 4.18867767e-01 3.02086830e-01
1.60728291e-01 -4.62231226e-02 8.72814238e-01 -7.26951420e-01
1.30528545e+00 -4.00688976e-01 -3.88598479e-02 9.56822261e-02
-4.51039493e-01 3.44140362e-03 8.45044613e-01 -2.81025413e-02
-7.28140354e-01 -2.08371505e-01 -4.35232043e-01 -8.04985464e-02
-2.63559103e-01 5.38612068e-01 1.03080742e-01 2.48392358e-01
5.01881599e-01 4.30450588e-01 -7.81027675e-02 -4.97405380e-01
2.08362296e-01 1.27744579e+00 -8.88709724e-02 -2.29125455e-01
3.00827950e-01 -1.06595099e-01 -4.50806290e-01 -1.30410385e+00
-1.03409469e+00 -9.12129104e-01 -6.30724132e-01 -2.14137316e-01
6.70930386e-01 -7.26846516e-01 -4.88963991e-01 -1.23218104e-01
-1.19467187e+00 -8.66632983e-02 -6.02708101e-01 4.66570735e-01
-3.25647086e-01 3.27554256e-01 -7.25625992e-01 -7.16948807e-01
-1.12743819e+00 -8.83729577e-01 8.94975722e-01 8.57464194e-01
-6.91520512e-01 -1.07193792e+00 3.36000770e-01 8.34263861e-01
5.42108238e-01 -1.21675350e-01 8.08611631e-01 -1.45162845e+00
-3.91320586e-01 -8.36713731e-01 3.07926774e-01 5.63835979e-01
-1.31523386e-02 5.02719164e-01 -1.02371812e+00 -1.17998712e-01
-9.77335051e-02 -7.68259028e-03 5.72454214e-01 4.07092631e-01
1.10319662e+00 -6.89684749e-01 -7.27138668e-02 1.88424215e-01
1.06688702e+00 -1.24091253e-01 2.75902569e-01 4.97039646e-01
6.13880694e-01 6.58178091e-01 3.55626345e-01 4.72421974e-01
5.65928817e-01 1.88262165e-01 1.03173554e-01 2.59689659e-01
-1.40820503e-01 -1.73298895e-01 6.42113805e-01 1.69358647e+00
4.13890243e-01 -1.11154035e-01 -8.71416330e-01 8.52194846e-01
-1.68759668e+00 -1.16885662e+00 -2.70287901e-01 1.99451303e+00
8.48098397e-01 8.41316730e-02 2.10540548e-01 -5.18459864e-02
3.70105475e-01 8.91271457e-02 -1.26359865e-01 -1.44514799e+00
5.00050306e-01 4.02352750e-01 3.03197592e-01 2.46713564e-01
-7.26311564e-01 7.77568817e-01 5.86577988e+00 6.26575530e-01
-1.07749391e+00 -1.42252088e-01 3.25355500e-01 -4.43664968e-01
-3.21252078e-01 -7.27016181e-02 -1.18017077e+00 4.43169326e-01
1.49195898e+00 -7.50643969e-01 -1.07873473e-02 1.12319589e+00
6.07583106e-01 7.04818126e-03 -9.35537815e-01 5.51221490e-01
5.27756095e-01 -1.76331389e+00 1.50249247e-02 -1.40970930e-01
9.69678283e-01 6.30054995e-02 -1.63071945e-01 6.21165693e-01
3.30757350e-01 -5.87924600e-01 2.17550203e-01 4.04443741e-01
-6.79110736e-02 -1.03120887e+00 8.97820532e-01 6.21922672e-01
-6.97104573e-01 1.01978995e-01 -4.80231911e-01 -1.99116483e-01
-1.98881775e-01 4.65738177e-01 -1.64352882e+00 4.72832471e-01
6.05031788e-01 5.18417656e-01 -8.33211243e-01 1.35060239e+00
-4.94450592e-02 8.46820593e-01 -1.16496488e-01 -8.32964420e-01
3.74035031e-01 -1.85112640e-01 8.15329909e-01 1.73965585e+00
5.55819690e-01 -2.44018704e-01 8.20580497e-02 3.65073532e-01
-2.27775037e-01 6.00648403e-01 -6.95605993e-01 -1.79284349e-01
6.22464001e-01 1.79723454e+00 -8.05947125e-01 -4.58674908e-01
5.76980077e-02 7.21706092e-01 9.25344005e-02 1.18067190e-01
-3.89764398e-01 -1.09168541e+00 5.87384105e-01 1.75773706e-02
2.75445193e-01 -2.96223074e-01 -4.64119554e-01 -9.15032983e-01
-2.16774810e-02 -8.26625943e-01 1.66160017e-01 -7.72728384e-01
-8.02387536e-01 3.69649351e-01 -1.63864031e-01 -9.42951858e-01
-3.13914984e-01 -1.30869508e-01 -1.44040430e+00 6.85076952e-01
-1.07245600e+00 -8.88973653e-01 -3.22642624e-01 -4.37814184e-02
1.11102724e+00 -2.68889338e-01 1.16269195e+00 3.25548500e-02
-8.33688796e-01 4.78592694e-01 5.51195860e-01 -1.08088292e-01
8.96499693e-01 -1.76141131e+00 1.30084768e-01 7.38223672e-01
3.17577690e-01 8.79595518e-01 1.13739359e+00 -3.57355654e-01
-1.52911556e+00 -1.06935763e+00 1.33502197e+00 -5.16433418e-01
5.67458034e-01 -2.45006442e-01 -7.41495132e-01 5.08061886e-01
9.10485327e-01 -7.76251018e-01 1.24409235e+00 2.54081041e-01
3.36394191e-01 -2.75525570e-01 -9.06791389e-01 7.81518996e-01
2.25722551e-01 -2.19869390e-01 -8.15123856e-01 7.70323873e-01
9.15651500e-01 -4.11197126e-01 -9.04521883e-01 -4.45100725e-01
2.07488462e-01 -7.88804710e-01 4.93463069e-01 -7.19549417e-01
8.49116385e-01 1.25100598e-01 3.10160756e-01 -1.66999555e+00
9.74434763e-02 -8.07403266e-01 -1.77107945e-01 1.91305983e+00
3.52276623e-01 -1.43583193e-01 1.22445917e+00 8.41325223e-01
-6.69647098e-01 -7.09487021e-01 -6.65416002e-01 -3.06829631e-01
1.27260163e-01 -2.27623299e-01 4.57869709e-01 6.63535655e-01
3.79933745e-01 7.21293151e-01 3.14724296e-01 -2.15922862e-01
3.24571133e-01 2.04038218e-01 1.03547835e+00 -9.63275015e-01
2.84018278e-01 -7.11065531e-01 -4.83227313e-01 -7.55776465e-01
1.71486646e-01 -1.16380262e+00 6.51300475e-02 -2.43020058e+00
3.29357058e-01 9.45481956e-02 -2.45469641e-02 3.11596453e-01
-1.42523766e-01 -2.58336395e-01 2.57366210e-01 -2.01387946e-02
-9.84479904e-01 3.10227871e-01 8.17366064e-01 -1.45848706e-01
-6.10243499e-01 5.45569181e-01 -1.19350660e+00 5.48395038e-01
1.08937991e+00 -5.86988091e-01 -2.70300031e-01 -3.54995370e-01
6.95309862e-02 -1.23857312e-01 -2.12311193e-01 -1.26787210e+00
8.31556857e-01 1.54373169e-01 3.61183167e-01 -1.20841885e+00
1.50864750e-01 -5.15663922e-01 -5.36153972e-01 1.03961192e-01
-9.15452421e-01 2.73786634e-01 4.24673855e-01 1.06872663e-01
-2.63740867e-01 -7.91690648e-01 1.49560943e-01 -3.72475795e-02
-1.38929620e-01 -2.45804861e-01 -7.85844326e-01 -6.35825656e-03
9.93034065e-01 -2.08876744e-01 -2.23322496e-01 -6.93389893e-01
-3.95542860e-01 1.98369756e-01 1.93226084e-01 3.71318430e-01
7.46660590e-01 -8.53949130e-01 -8.31944823e-01 -5.80540597e-02
-2.20318854e-01 2.49847636e-01 3.19534689e-01 5.38695812e-01
-1.00977802e+00 6.67566180e-01 1.31320190e-02 -4.90631670e-01
-1.93750346e+00 -7.58664012e-02 -1.98230803e-01 -3.20358068e-01
-8.00023556e-01 8.30103755e-01 -7.67498493e-01 -5.43366432e-01
5.29679835e-01 -3.31885129e-01 -4.54657257e-01 4.80770499e-01
8.73032331e-01 6.76282585e-01 6.44487262e-01 1.94463285e-03
-3.11326161e-02 -1.11587264e-01 -2.83391237e-01 -1.34071633e-01
1.94498634e+00 -7.39614069e-02 -1.63599312e-01 6.39737368e-01
1.41033459e+00 4.31146324e-01 -7.40741789e-01 1.91121519e-01
1.66583076e-01 -1.41264601e-02 6.40804842e-02 -5.86553931e-01
-4.87863570e-01 1.10084152e+00 1.72074318e-01 7.24249244e-01
7.39824772e-01 -2.77281046e-01 8.57246995e-01 8.12660992e-01
-4.90323156e-01 -1.37100530e+00 1.54429287e-01 6.78219318e-01
6.63199544e-01 -9.35173869e-01 3.07957172e-01 2.89836764e-01
-7.06275880e-01 1.58637643e+00 4.12790865e-01 -1.37220442e-01
2.56993562e-01 1.74827263e-01 7.82584120e-03 -1.43426001e-01
-1.08390641e+00 3.75069439e-01 2.28453875e-01 3.43484879e-01
9.41561699e-01 9.10571963e-02 -1.78592384e-01 5.51096082e-01
-9.24319267e-01 -2.95275658e-01 1.11928105e+00 1.06299210e+00
-9.55258846e-01 -1.04303432e+00 -3.55113536e-01 7.52861798e-01
-8.52593541e-01 -2.64217675e-01 -8.16713631e-01 4.65080082e-01
-6.29428506e-01 1.20446002e+00 1.64348766e-01 -1.65296257e-01
4.78046805e-01 4.92915809e-01 7.44067878e-02 -1.36516190e+00
-1.26672518e+00 -3.05032283e-01 2.68945664e-01 -7.25859851e-02
2.59163797e-01 -7.46376693e-01 -1.23528290e+00 -5.84744334e-01
-3.11804771e-01 8.01845670e-01 1.36752224e+00 6.89998865e-01
6.83205545e-01 7.83327103e-01 6.61794543e-01 -9.37912345e-01
-8.39307427e-01 -1.13187063e+00 -2.32010320e-01 -1.84701178e-02
1.47845685e-01 4.59044158e-01 -2.58345932e-01 1.31837025e-01] | [12.493980407714844, 9.486716270446777] |
ab523ed7-29e0-4e3e-b3ff-efef7366c759 | semi-supervised-object-detection-via-multi | null | null | http://openaccess.thecvf.com//content/CVPR2022/html/Li_Semi-Supervised_Object_Detection_via_Multi-Instance_Alignment_With_Global_Class_Prototypes_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Li_Semi-Supervised_Object_Detection_via_Multi-Instance_Alignment_With_Global_Class_Prototypes_CVPR_2022_paper.pdf | Semi-Supervised Object Detection via Multi-Instance Alignment With Global Class Prototypes | Semi-Supervised object detection (SSOD) aims to improve the generalization ability of object detectors with large-scale unlabeled images. Current pseudo-labeling-based SSOD methods individually learn from labeled data and unlabeled data, without considering the relation between them. To make full use of labeled data, we propose a Multi-instance Alignment model which enhances the prediction consistency based on Global Class Prototypes (MA-GCP). Specifically, we impose the consistency between pseudo ground-truths and their high-IoU candidates by minimizing the cross-entropy loss of their class distributions computed based on global class prototypes. These global class prototypes are estimated with the whole labeled dataset via the exponential moving average algorithm. To evaluate the proposed MA-GCP model, we integrate it into the state-of-the-art SSOD framework and experiments on two benchmark datasets demonstrate the effectiveness of our MA-GCP approach. | ['Zhenguo Li', 'Peng Yuan', 'Aoxue Li'] | 2022-01-01 | null | null | null | cvpr-2022-1 | ['semi-supervised-object-detection'] | ['computer-vision'] | [-3.36650610e-02 1.30840778e-01 -4.98685449e-01 -7.01158822e-01
-8.10854673e-01 -4.28615332e-01 5.36199987e-01 9.28864554e-02
-3.91198516e-01 5.56407392e-01 -4.18773681e-01 6.21778220e-02
3.59412991e-02 -5.65412343e-01 -7.14624763e-01 -7.41668522e-01
3.13822895e-01 6.66694462e-01 5.98826110e-01 4.23769027e-01
7.97451213e-02 3.05924565e-01 -1.56720865e+00 2.35518366e-01
9.04051721e-01 1.29433310e+00 1.86779946e-01 3.60233933e-02
-7.20404238e-02 6.54873669e-01 -1.57122567e-01 -3.99967581e-01
4.74391311e-01 -2.71040022e-01 -6.73528492e-01 5.10282040e-01
6.61762774e-01 -2.14849457e-01 -7.90150464e-02 1.44821167e+00
2.74186522e-01 4.66415733e-02 8.77405167e-01 -1.43845844e+00
-7.96771705e-01 4.06617969e-01 -5.46565235e-01 1.74339801e-01
-2.54382908e-01 1.88638523e-01 1.06557941e+00 -1.32068872e+00
8.01505387e-01 1.22393584e+00 5.65070927e-01 3.95501345e-01
-1.39166629e+00 -6.79558575e-01 4.09695566e-01 4.14993465e-01
-1.76578867e+00 -6.59290478e-02 6.68603122e-01 -5.52893758e-01
5.88296115e-01 -6.22221157e-02 3.75081748e-01 5.42975307e-01
-2.08955660e-01 1.12196517e+00 9.90429878e-01 -5.82105100e-01
3.79432768e-01 5.71908534e-01 7.56996095e-01 6.94087923e-01
5.21044433e-01 2.08020896e-01 -2.78234184e-01 -1.60557017e-01
2.95245111e-01 7.65343085e-02 -5.64493723e-02 -9.84051704e-01
-1.16283262e+00 6.09626412e-01 7.13120878e-01 4.01450787e-03
-2.36014813e-01 -3.35685492e-01 9.20890942e-02 -2.41290797e-02
3.97808284e-01 1.21848874e-01 -4.20822024e-01 6.77978039e-01
-7.72323191e-01 -2.46421620e-03 4.44446057e-01 1.23981237e+00
9.95723009e-01 -4.09385771e-01 -2.87846386e-01 7.66354084e-01
5.57817400e-01 5.06641686e-01 5.00569940e-01 -4.69190925e-01
2.88712531e-01 1.06453407e+00 2.41784856e-01 -7.23145425e-01
-3.11019063e-01 -5.97205639e-01 -4.11031157e-01 6.09992929e-02
4.04650629e-01 8.30574855e-02 -1.01249886e+00 1.69421256e+00
8.38345706e-01 4.33989376e-01 -1.65354293e-02 9.66208637e-01
6.42973781e-01 4.95953053e-01 3.20373952e-01 -4.71387088e-01
9.29539084e-01 -1.22622144e+00 -5.25818527e-01 -3.35245609e-01
7.54549980e-01 -3.74652892e-01 7.54744291e-01 1.53403506e-02
-5.51145077e-01 -8.82308125e-01 -1.18578410e+00 2.20718578e-01
-2.76499569e-01 7.20984638e-01 9.37162936e-02 5.10213375e-01
-4.38356161e-01 3.91234428e-01 -9.99664545e-01 -2.88478285e-01
6.87517166e-01 3.90541613e-01 -3.32078278e-01 -1.03024930e-01
-6.82103455e-01 8.39739263e-01 1.08663309e+00 1.40426621e-01
-8.95903051e-01 -5.04014552e-01 -6.10762537e-01 -7.64889270e-02
4.90845561e-01 -3.01330149e-01 1.10065401e+00 -1.08682787e+00
-1.04736495e+00 1.08322525e+00 -4.16237153e-02 -4.01908875e-01
4.21857148e-01 -1.36278450e-01 -4.09516513e-01 -2.15564389e-03
1.35143027e-01 9.56738472e-01 7.14391172e-01 -1.31915295e+00
-9.63674128e-01 -5.76796651e-01 -5.05402803e-01 2.42511913e-01
-4.50834543e-01 -3.19590211e-01 -5.04748344e-01 -2.84761339e-01
6.13272548e-01 -1.10765994e+00 -2.05419272e-01 3.24491471e-01
-6.66572750e-01 -4.48461056e-01 1.04790592e+00 -1.89671233e-01
1.09013855e+00 -2.35734534e+00 -1.15263805e-01 2.21566573e-01
3.15756388e-02 5.91568470e-01 -1.35875091e-01 -2.19796509e-01
4.35936861e-02 -3.37341726e-01 -2.54848599e-01 -2.47551590e-01
1.89903155e-02 -1.83071382e-02 -1.58732295e-01 5.27328134e-01
4.78141397e-01 8.88656378e-01 -8.89401793e-01 -8.07963550e-01
2.03198269e-01 5.77549897e-02 -1.69507757e-01 1.37386635e-01
-3.68232727e-01 2.28808105e-01 -6.37964845e-01 6.61545157e-01
8.88405442e-01 -6.40791774e-01 3.66177917e-01 -2.79027611e-01
1.83407024e-01 -1.85675949e-01 -1.43651736e+00 1.37879038e+00
1.50800049e-01 1.76119372e-01 -4.42370981e-01 -1.09575045e+00
1.13130832e+00 -9.73548293e-02 3.23744893e-01 -4.58893269e-01
1.07083172e-01 4.06079888e-01 -1.48077786e-01 -1.58219069e-01
1.84116617e-01 -6.56028986e-02 3.38274390e-01 3.14237922e-01
4.07808781e-01 3.46200615e-01 1.90320283e-01 2.39541993e-01
4.54463512e-01 1.98883608e-01 6.68087065e-01 -4.56532657e-01
5.80485225e-01 2.13674814e-01 7.80768573e-01 8.43423784e-01
-5.01620173e-01 6.73776090e-01 1.27143949e-01 -3.98372531e-01
-9.80278790e-01 -1.20334888e+00 -5.29244423e-01 7.63457596e-01
5.98820627e-01 7.77021572e-02 -7.60347307e-01 -1.28452682e+00
1.87989011e-01 7.60313928e-01 -6.62359238e-01 -2.63281971e-01
-6.80147335e-02 -8.60993922e-01 6.48844391e-02 5.65792024e-01
6.17239714e-01 -8.56623888e-01 -3.23014051e-01 3.37852091e-01
3.40404734e-02 -1.17131186e+00 -5.77764094e-01 2.44084537e-01
-8.47093880e-01 -1.17487240e+00 -4.15517867e-01 -1.08379161e+00
1.07768404e+00 4.21559483e-01 7.99686491e-01 2.88840774e-02
-4.39741999e-01 1.61941618e-01 -2.33399168e-01 -3.93588275e-01
-2.86924303e-01 -1.52858153e-01 1.24952167e-01 4.62532133e-01
9.03406322e-01 -1.80038977e-02 -4.59134668e-01 7.50021279e-01
-6.12797558e-01 -2.15185084e-03 5.50129175e-01 1.03262711e+00
1.24199688e+00 -1.84290156e-01 7.03831077e-01 -9.53943789e-01
-2.26682395e-01 -5.19059539e-01 -1.01932275e+00 7.59862363e-01
-1.07563019e+00 7.71800280e-02 1.61452264e-01 -6.74591362e-01
-1.19588363e+00 5.93378663e-01 4.25935984e-01 -7.22784162e-01
-1.43078631e-02 2.67478555e-01 -4.07694101e-01 -2.15986595e-02
6.88683152e-01 2.57181674e-01 -1.81826368e-01 -4.12836254e-01
4.90793169e-01 8.63323927e-01 7.04169154e-01 -3.93242776e-01
6.71119332e-01 5.57369828e-01 -2.84513645e-02 -3.29273850e-01
-1.35744953e+00 -8.77801895e-01 -9.55849707e-01 -3.02793328e-02
7.27524698e-01 -9.72795248e-01 -2.31734306e-01 5.56523442e-01
-7.64556050e-01 -7.58006647e-02 -3.05291742e-01 7.90786326e-01
-2.76478171e-01 4.53575850e-01 -2.87038743e-01 -8.52774084e-01
-2.45532021e-01 -1.05796742e+00 1.04270089e+00 4.55161303e-01
2.95726448e-01 -6.61303997e-01 -6.14100359e-02 2.83945948e-01
-2.53512740e-01 -7.44432062e-02 7.26413369e-01 -1.14700162e+00
-7.15956867e-01 -4.98006463e-01 -5.66294432e-01 5.56246579e-01
4.53152694e-02 -1.71858091e-02 -1.00735176e+00 -2.95647979e-01
-2.98711896e-01 -4.96831834e-01 8.97916675e-01 4.61430907e-01
9.71757889e-01 -8.27500150e-02 -7.37935901e-01 2.47919858e-01
1.54381919e+00 1.07618585e-01 3.15745980e-01 1.77623868e-01
6.48251534e-01 5.12978613e-01 1.20482230e+00 4.94631380e-01
1.65417299e-01 7.20518947e-01 3.92867118e-01 2.75652677e-01
-1.10618867e-01 -5.07044792e-01 2.99445875e-02 4.39858854e-01
5.04176140e-01 -2.54704982e-01 -7.81494319e-01 5.13354063e-01
-2.10428715e+00 -7.34268665e-01 -2.08255768e-01 2.47257137e+00
6.05054021e-01 3.41544300e-01 3.13417204e-02 -1.16831392e-01
1.32475531e+00 -3.66993487e-01 -8.49043310e-01 4.05816466e-01
-1.09357834e-01 -2.78237283e-01 3.91301334e-01 1.83137447e-01
-1.48006821e+00 8.28639090e-01 5.72341967e+00 1.07783091e+00
-9.47141886e-01 2.27491930e-01 8.11420739e-01 8.57049748e-02
2.03234285e-01 3.52361873e-02 -1.39753151e+00 4.24717396e-01
4.61174965e-01 -2.53067464e-01 6.71298429e-03 1.55364335e+00
-1.83304563e-01 -4.86368723e-02 -1.31942654e+00 8.55229795e-01
-4.63854615e-03 -1.19286978e+00 3.16484086e-03 -1.63233250e-01
1.14614248e+00 1.52979776e-01 2.41200402e-02 3.46384495e-01
1.80855811e-01 -2.92426229e-01 9.63032484e-01 1.55472457e-01
5.78492284e-01 -4.22968686e-01 7.05999374e-01 5.23976028e-01
-1.30337477e+00 -3.02671283e-01 -6.43244386e-01 4.10886288e-01
-1.07076079e-01 6.24947965e-01 -9.75433171e-01 4.52205241e-01
4.65274930e-01 8.21877539e-01 -7.99923420e-01 1.29998660e+00
-1.71700567e-01 6.11043632e-01 -4.72229332e-01 -1.18666850e-02
1.60929814e-01 -2.51604110e-01 3.84153754e-01 8.95868063e-01
1.71599127e-02 1.01554424e-01 5.26559472e-01 9.03131008e-01
-1.25398576e-01 2.36994863e-01 -1.21671632e-01 5.83342463e-02
7.34799564e-01 1.26425779e+00 -1.01319098e+00 -6.70484483e-01
-3.11124951e-01 6.10973239e-01 5.59465170e-01 2.11190134e-01
-7.74671793e-01 -1.67972576e-02 1.63455591e-01 -2.27195293e-01
4.89398718e-01 2.30201825e-01 -4.30700570e-01 -1.13960755e+00
1.52201876e-01 -4.67337459e-01 7.60830581e-01 -6.85224175e-01
-1.50738513e+00 4.98259336e-01 1.06651196e-02 -1.66902828e+00
1.73958749e-01 -5.70969284e-01 -4.38721716e-01 4.23967123e-01
-1.42771304e+00 -1.27974570e+00 -3.15727890e-01 4.18282151e-01
3.92899722e-01 -6.88577592e-02 5.33195019e-01 2.39081517e-01
-7.01885998e-01 6.95878208e-01 2.98564702e-01 1.27590567e-01
5.91994405e-01 -1.04169607e+00 9.79282036e-02 9.84219551e-01
3.54579479e-01 4.42028344e-01 3.20938200e-01 -7.39593029e-01
-9.66117799e-01 -1.58398306e+00 6.83364272e-01 -3.96451920e-01
4.23363745e-01 -2.40663271e-02 -1.13048959e+00 7.74290621e-01
-5.40616572e-01 8.34758818e-01 4.01181400e-01 -3.67974751e-02
-6.20180666e-01 -2.84424722e-01 -1.33206582e+00 2.22271889e-01
9.86419261e-01 -3.93519312e-01 -6.17024899e-01 5.63296080e-01
6.22057378e-01 -1.76461041e-01 -4.60577160e-01 6.88028455e-01
4.41384465e-01 -6.50079727e-01 8.64647806e-01 -6.86830938e-01
-3.66208851e-02 -7.41414189e-01 -1.47760510e-01 -9.67219353e-01
-3.91286403e-01 7.45115578e-02 -4.82776575e-02 1.19036269e+00
4.54742342e-01 -4.88405794e-01 8.85705113e-01 6.88582480e-01
-8.08368623e-02 -5.35178483e-01 -9.69757438e-01 -1.14455473e+00
-3.51897180e-01 -1.53779134e-01 4.28820729e-01 9.51493502e-01
-2.91463044e-02 2.61644095e-01 -5.30901849e-02 6.62243783e-01
9.92690265e-01 4.73900676e-01 6.09216452e-01 -1.28584945e+00
-4.49664950e-01 -1.48634225e-01 -7.54997492e-01 -9.61450577e-01
1.37888268e-01 -9.65615749e-01 3.01150143e-01 -1.15294731e+00
7.09049463e-01 -6.25288427e-01 -7.08376706e-01 5.75303912e-01
-6.54584944e-01 4.50712979e-01 1.76063031e-01 5.14424026e-01
-1.19108987e+00 6.76011920e-01 1.04631031e+00 -2.47832909e-01
-2.14621663e-01 -4.91606025e-03 -1.43759221e-01 5.62670827e-01
4.27101851e-01 -8.22610378e-01 -4.37040567e-01 -3.88313308e-02
-3.74953061e-01 -2.31769010e-01 3.72927248e-01 -1.07093728e+00
2.41286770e-01 -1.69375718e-01 2.87528187e-01 -8.02406073e-01
2.99304966e-02 -8.61118615e-01 1.92739174e-01 6.75577760e-01
-4.47806150e-01 -6.23641670e-01 -6.94425628e-02 1.02878845e+00
-2.95824647e-01 -2.90368289e-01 1.28516984e+00 2.57593960e-01
-9.71963406e-01 4.22656983e-01 2.31657386e-01 -3.52082029e-02
1.50781858e+00 -1.17900617e-01 -3.12339395e-01 3.11966509e-01
-7.92496562e-01 3.70231032e-01 2.83992469e-01 2.27244243e-01
3.64708543e-01 -1.48978472e+00 -5.47748685e-01 2.26336956e-01
7.71958232e-01 2.88728047e-02 2.08967358e-01 6.37686789e-01
-1.93652391e-01 4.15442795e-01 -7.91099444e-02 -9.40910816e-01
-1.19676197e+00 1.14074337e+00 4.42023307e-01 -2.02133626e-01
-3.31583351e-01 8.56940329e-01 4.75639880e-01 -6.13468707e-01
2.26298705e-01 -8.52595195e-02 -1.01579778e-01 -9.54510644e-02
5.15498638e-01 2.42048323e-01 1.97583418e-02 -4.90459651e-01
-3.97647411e-01 3.85173500e-01 -2.94547230e-01 2.07635224e-01
9.43512022e-01 -2.13000670e-01 1.52284533e-01 3.61162722e-01
1.14554083e+00 -6.13555133e-01 -1.65234494e+00 -7.83124089e-01
3.12165737e-01 -4.84310031e-01 5.59719540e-02 -6.63518012e-01
-1.00681734e+00 6.80513561e-01 1.07110691e+00 -3.07587981e-01
9.33114111e-01 1.45470351e-01 3.31161588e-01 5.77239752e-01
5.25188327e-01 -1.14790773e+00 2.07408547e-01 7.61823216e-03
2.81958252e-01 -1.58061218e+00 7.38583580e-02 -7.14922547e-01
-6.88338339e-01 1.01048744e+00 9.65497851e-01 -3.95722501e-02
7.99918115e-01 -3.27159107e-01 2.88974121e-02 7.29729310e-02
-6.18388474e-01 -3.13872755e-01 5.38093686e-01 4.48782533e-01
-1.71899170e-01 1.63863838e-01 -3.89311820e-01 5.40846407e-01
6.37884796e-01 3.19482267e-01 8.47556666e-02 8.81668866e-01
-6.66713715e-01 -9.13131237e-01 -2.39603445e-01 5.13181925e-01
5.75543456e-02 8.82263780e-02 -1.42119423e-01 5.95019341e-01
1.71346962e-01 7.62349784e-01 -1.00083696e-02 -3.44915122e-01
5.21982312e-02 1.79043785e-01 3.31789374e-01 -8.07318866e-01
-7.43919313e-02 2.86150239e-02 -3.41613770e-01 -1.49213105e-01
-4.84425038e-01 -8.11479568e-01 -1.33009017e+00 3.81888241e-01
-1.04259360e+00 3.72026041e-02 5.55745125e-01 9.98889267e-01
4.55502450e-01 -1.32390223e-02 8.83111835e-01 -5.32176912e-01
-1.09247577e+00 -9.42342401e-01 -7.98708558e-01 5.58775008e-01
-6.86964616e-02 -8.30195189e-01 -2.74031550e-01 1.05872385e-01] | [9.257606506347656, 1.3531970977783203] |
b91140fb-5b00-4254-8d65-c3a68a4729b7 | featfsda-towards-few-shot-domain-adaptation | 2305.08420 | null | https://arxiv.org/abs/2305.08420v1 | https://arxiv.org/pdf/2305.08420v1.pdf | FeatFSDA: Towards Few-shot Domain Adaptation for Video-based Activity Recognition | Domain adaptation is essential for activity recognition, as common spatiotemporal architectures risk overfitting due to increased parameters arising from the temporal dimension. Unsupervised domain adaptation methods have been extensively studied, yet, they require large-scale unlabeled data from the target domain. In this work, we address few-shot domain adaptation for video-based activity recognition (FSDA-AR), which leverages a very small amount of labeled target videos to achieve effective adaptation. This setting is attractive and promising for applications, as it requires recording and labeling only a few, or even a single example per class in the target domain, which often includes activities that are rare yet crucial to recognize. We construct FSDA-AR benchmarks using five established datasets: UCF101, HMDB51, EPIC-KITCHEN, Sims4Action, and Toyota Smart Home. Our results demonstrate that FSDA-AR performs comparably to unsupervised domain adaptation with significantly fewer (yet labeled) target examples. We further propose a novel approach, FeatFSDA, to better leverage the few labeled target domain samples as knowledge guidance. FeatFSDA incorporates a latent space semantic adjacency loss, a domain prototypical similarity loss, and a graph-attentive-network-based edge dropout technique. Our approach achieves state-of-the-art performance on all datasets within our FSDA-AR benchmark. To encourage future research of few-shot domain adaptation for video-based activity recognition, we will release our benchmarks and code at https://github.com/KPeng9510/FeatFSDA. | ['Alina Roitberg', 'Rainer Stiefelhagen', 'M. Saquib Sarfraz', 'Kailun Yang', 'Jiaming Zhang', 'David Schneider', 'Di Wen', 'Kunyu Peng'] | 2023-05-15 | null | null | null | null | ['unsupervised-domain-adaptation'] | ['methodology'] | [ 3.26095730e-01 -2.71474391e-01 -7.14676499e-01 -2.46276230e-01
-7.54773557e-01 -4.23815399e-01 5.18848300e-01 -3.11035454e-01
-4.23497200e-01 8.68486106e-01 4.43002611e-01 1.45887643e-01
-8.64812434e-02 -5.44175267e-01 -7.57123172e-01 -7.07154751e-01
-1.16073310e-01 4.09128666e-01 3.34218532e-01 1.66116640e-01
-2.75235116e-01 2.13617951e-01 -1.02466714e+00 1.48042977e-01
7.93137491e-01 1.00617361e+00 2.00187303e-02 5.04921257e-01
1.86656028e-01 1.09254622e+00 -2.98620701e-01 -2.23309509e-02
1.43406034e-01 -7.32859492e-01 -8.15042019e-01 2.88288355e-01
3.88259828e-01 -5.25965571e-01 -9.24697578e-01 5.71806371e-01
5.08227468e-01 7.14276075e-01 1.01064718e+00 -1.40483880e+00
-6.66426003e-01 1.94845885e-01 -4.13188934e-01 6.57929301e-01
3.29606712e-01 4.51756328e-01 9.04382348e-01 -8.77943635e-01
8.70949924e-01 7.18920231e-01 6.84536695e-01 7.23570645e-01
-1.29705083e+00 -8.94586205e-01 4.24481779e-01 4.93428409e-01
-1.39395487e+00 -5.58461308e-01 9.11680400e-01 -5.76344848e-01
1.18328762e+00 -2.43936747e-01 7.53089547e-01 2.07584333e+00
-2.96986997e-01 1.08845448e+00 7.32471168e-01 -2.81197987e-02
6.14178479e-01 -3.05982977e-01 2.76660621e-01 5.49581885e-01
9.26346630e-02 1.05621085e-01 -8.31778646e-01 -1.01195909e-01
1.00961185e+00 2.51519024e-01 -2.96306252e-01 -8.18684936e-01
-1.14181483e+00 7.56698132e-01 2.26002544e-01 1.78874329e-01
-3.18471968e-01 1.73862964e-01 5.71125507e-01 3.98352712e-01
5.02021968e-01 2.16033503e-01 -4.52393442e-01 -6.03666961e-01
-7.20630825e-01 -6.32753074e-02 6.84806466e-01 1.10543454e+00
4.55724090e-01 3.20633292e-01 -4.30197537e-01 9.70296979e-01
-1.16298720e-01 4.43642050e-01 5.98528683e-01 -1.06366611e+00
6.78076744e-01 4.29158211e-01 3.55828814e-02 -5.56400836e-01
-1.92996949e-01 -2.56184161e-01 -8.30799699e-01 -1.34363219e-01
5.23931503e-01 -1.86117381e-01 -1.14001131e+00 1.86842883e+00
1.62037373e-01 9.45740223e-01 1.01394370e-01 9.90404308e-01
6.63291752e-01 5.84981382e-01 2.68053442e-01 -9.45242569e-02
9.58744347e-01 -1.16158068e+00 -5.21968961e-01 -3.93812656e-01
8.55654538e-01 1.60715953e-01 1.58579302e+00 2.99049824e-01
-7.33414054e-01 -3.25823277e-01 -9.79731202e-01 9.39897746e-02
-2.50590712e-01 1.73427854e-02 6.04758918e-01 2.66491443e-01
-5.62903464e-01 4.93649542e-01 -1.28419459e+00 -7.23800063e-01
1.12777972e+00 2.75953025e-01 -5.24849057e-01 -3.08267713e-01
-1.07588208e+00 4.89039958e-01 3.83957326e-01 -4.20267045e-01
-1.40816295e+00 -8.35443854e-01 -9.98787522e-01 4.93388101e-02
7.47144222e-01 -7.61065423e-01 1.18435371e+00 -9.83700097e-01
-1.46141469e+00 7.95575261e-01 -2.81559285e-02 -7.56781578e-01
3.81407410e-01 -3.59216064e-01 -6.29723251e-01 3.83792192e-01
3.30362506e-02 4.87781614e-01 7.95468986e-01 -6.98488355e-01
-3.02922100e-01 -2.98608899e-01 -1.02732398e-01 3.66584957e-01
-7.02278912e-01 -2.24984214e-01 -5.14147103e-01 -1.01340485e+00
-5.66017807e-01 -8.77444744e-01 2.35995073e-02 3.54965474e-03
-1.78322166e-01 -2.27022290e-01 9.78452325e-01 -5.00400901e-01
1.19134498e+00 -2.33841276e+00 2.59045124e-01 1.52690634e-02
1.17382050e-01 3.84407610e-01 -5.01557708e-01 1.95795864e-01
-4.13691886e-02 -2.94989288e-01 -5.61503232e-01 -1.07385635e-01
-1.27528414e-01 3.89852047e-01 -7.79352561e-02 4.58780468e-01
2.91299015e-01 1.05205584e+00 -1.08563232e+00 -2.86968797e-01
1.59949273e-01 5.06130576e-01 -6.36974216e-01 2.12040901e-01
-2.54385650e-01 6.31541133e-01 -5.99724770e-01 7.37786472e-01
8.73858035e-02 -5.30151069e-01 4.82595973e-02 -6.32429793e-02
4.75554705e-01 1.66257843e-01 -9.47686970e-01 2.28478384e+00
-2.93068856e-01 6.48337066e-01 -3.93074095e-01 -1.29386163e+00
6.12573266e-01 2.38434553e-01 7.68063664e-01 -8.08763325e-01
4.88594100e-02 -9.53881517e-02 -2.26769879e-01 -3.97530496e-01
-4.74402122e-02 9.61087197e-02 -1.45069450e-01 3.47304493e-01
6.23678327e-01 3.32120001e-01 2.96550095e-01 4.64476973e-01
1.76582599e+00 3.42097789e-01 4.35271680e-01 1.20684721e-01
1.76871151e-01 1.71337649e-01 7.67048180e-01 7.33613372e-01
-5.52866578e-01 6.46570623e-01 3.28369945e-01 -2.26868466e-01
-9.25738811e-01 -1.31031203e+00 2.01838180e-01 1.28762615e+00
3.45249251e-02 -2.96720564e-01 -6.68366909e-01 -1.14182889e+00
-9.31886062e-02 6.50642991e-01 -6.53591990e-01 -6.48559213e-01
-4.83011335e-01 -4.47154343e-01 7.21693993e-01 1.10185003e+00
6.47633135e-01 -1.19816089e+00 -1.90146834e-01 2.58509338e-01
-1.84042484e-01 -1.35886264e+00 -9.16733444e-01 2.53790945e-01
-1.06329775e+00 -1.11389089e+00 -1.00830650e+00 -7.14952409e-01
5.03271580e-01 2.55029321e-01 1.17203009e+00 -4.50049222e-01
-1.51140064e-01 8.61572742e-01 -4.48494345e-01 -5.97359538e-02
3.71785872e-02 3.26992184e-01 2.93321192e-01 7.42817670e-02
7.83439755e-01 -9.80699539e-01 -6.33351207e-01 5.06344736e-01
-7.17606306e-01 -2.55010128e-01 5.48404694e-01 9.76169288e-01
7.87867844e-01 -3.63162130e-01 7.46595800e-01 -9.32144284e-01
4.34165865e-01 -8.10557663e-01 -2.97611415e-01 7.64107257e-02
-4.88478422e-01 -1.41410470e-01 7.27257788e-01 -8.56635749e-01
-1.05018020e+00 3.14030230e-01 2.16441780e-01 -1.10405004e+00
-4.32944536e-01 5.08166134e-01 -2.71366775e-01 9.39006954e-02
9.64944065e-01 4.28797632e-01 -7.64608309e-02 -4.87511247e-01
2.17346355e-01 4.12516028e-01 7.54423261e-01 -5.48199475e-01
7.11480498e-01 5.60433567e-01 -2.91249275e-01 -9.76601958e-01
-8.42494845e-01 -7.89077640e-01 -6.31242275e-01 -5.54853752e-02
8.35150898e-01 -1.30317140e+00 -2.84329951e-01 5.81942439e-01
-6.32792294e-01 -1.13145638e+00 -6.55178666e-01 6.80865586e-01
-8.96448910e-01 2.58843422e-01 -6.29403055e-01 -4.23395962e-01
-2.17777535e-01 -5.65580666e-01 7.82359838e-01 9.13250521e-02
-4.13169205e-01 -1.11165845e+00 3.64640534e-01 4.27211761e-01
3.17287967e-02 1.67693406e-01 4.97289181e-01 -9.63411987e-01
-4.52383906e-01 6.66456670e-02 3.02067213e-02 4.89438444e-01
3.33871335e-01 -5.21362901e-01 -7.72313893e-01 -3.96170378e-01
-3.08775008e-01 -7.40355849e-01 1.06822228e+00 4.37756598e-01
1.43920255e+00 -2.39189953e-01 -3.46110195e-01 7.94700146e-01
1.09427261e+00 2.71406412e-01 7.43789852e-01 2.00799510e-01
9.34503913e-01 -2.85045449e-02 7.25644112e-01 4.99447346e-01
3.46854657e-01 8.33125770e-01 1.41518414e-02 2.97487434e-02
-3.72742504e-01 -4.45719570e-01 7.37489223e-01 7.54738688e-01
-1.72682181e-01 -3.43178898e-01 -8.98421109e-01 7.51730382e-01
-1.99182737e+00 -1.22248590e+00 2.88784623e-01 2.16558266e+00
7.01799989e-01 1.25590637e-02 6.70101225e-01 -4.29766290e-02
4.28332478e-01 2.80635864e-01 -1.05940497e+00 1.39239386e-01
-1.48260787e-01 3.49056184e-01 4.10896391e-01 1.13647003e-02
-1.38821054e+00 9.24810052e-01 5.58486271e+00 1.03116512e+00
-8.12839687e-01 3.73453945e-01 4.72029269e-01 -5.92545033e-01
2.73901522e-01 -3.42934400e-01 -6.15189731e-01 6.81189358e-01
1.06876087e+00 -1.89038031e-02 5.32819390e-01 9.87900138e-01
3.47305864e-01 1.29529713e-02 -1.33712518e+00 1.21636677e+00
1.11102566e-01 -1.34594309e+00 -1.10559344e-01 -3.69892828e-02
7.71260262e-01 3.41775537e-01 -4.34425361e-02 6.17394924e-01
4.17418540e-01 -8.88937771e-01 1.69867888e-01 2.62993068e-01
9.95183587e-01 -5.87394595e-01 2.70531118e-01 1.59902573e-01
-1.27938056e+00 -1.82282776e-01 -1.40341535e-01 7.75456578e-02
1.25648171e-01 1.85936093e-01 -6.24263585e-01 2.51420289e-01
7.99785197e-01 1.39854741e+00 -3.46819907e-01 1.00108576e+00
-1.77863196e-01 1.10294271e+00 -3.72238904e-01 3.01243782e-01
2.96497971e-01 -1.38329014e-01 5.78537583e-01 1.18241847e+00
2.68721581e-01 3.32861960e-01 2.66293824e-01 6.01757169e-01
-4.52704489e-01 -1.38215274e-01 -8.88040185e-01 -4.33466822e-01
4.88980591e-01 8.53570819e-01 -4.73685652e-01 -4.52653110e-01
-7.82940745e-01 1.33557022e+00 4.83991235e-01 6.82521701e-01
-1.00002575e+00 -2.15866953e-01 7.81200588e-01 2.83477128e-01
4.75907624e-01 -2.22265959e-01 1.76391080e-02 -1.55665588e+00
4.23654430e-02 -9.05214190e-01 9.75524127e-01 -6.75326109e-01
-1.41370058e+00 1.23250440e-01 1.14579394e-01 -1.51249743e+00
-3.36043715e-01 -3.77536207e-01 -5.84823608e-01 2.43692309e-01
-1.30808389e+00 -1.00218201e+00 -5.08343697e-01 1.13721633e+00
8.87881935e-01 -4.49409336e-01 6.86915219e-01 6.46301150e-01
-8.41009498e-01 8.51498246e-01 1.21449813e-01 5.83040953e-01
7.63314724e-01 -9.73063827e-01 3.65369320e-01 8.05643082e-01
4.24649328e-01 2.09599108e-01 2.88858831e-01 -6.78268909e-01
-1.29466903e+00 -1.44920158e+00 2.24008754e-01 -6.00113094e-01
7.44668841e-01 -5.09486556e-01 -1.14909196e+00 1.03231037e+00
-4.29827869e-02 5.41104019e-01 7.76287854e-01 7.46559948e-02
-4.21061277e-01 -7.24133104e-02 -1.00023246e+00 5.62365294e-01
1.85304677e+00 -6.83214545e-01 -6.18808866e-01 3.24802220e-01
5.41474462e-01 -1.90742195e-01 -9.79207993e-01 3.24611038e-01
3.80624235e-01 -5.97451031e-01 1.04644179e+00 -9.49741840e-01
2.67880291e-01 -1.60206139e-01 3.65609787e-02 -1.45123529e+00
-4.15125281e-01 -5.28712809e-01 -9.21728134e-01 1.03605342e+00
2.66078979e-01 -3.87475640e-01 1.24683452e+00 3.63376349e-01
-1.89566657e-01 -4.33225393e-01 -8.78548265e-01 -1.48601210e+00
-5.25373742e-02 -4.11510646e-01 1.76717341e-01 1.25842059e+00
-2.27350947e-02 4.86929715e-01 -6.26711369e-01 -3.43510099e-02
6.43656254e-01 -3.05913001e-01 8.47345710e-01 -1.02376127e+00
-4.81218278e-01 -1.59582868e-01 -5.00471234e-01 -1.24113119e+00
2.96922594e-01 -8.40879142e-01 -1.08081177e-01 -1.37241805e+00
2.83344924e-01 -1.47690505e-01 -6.78888977e-01 6.59085572e-01
1.10407665e-01 2.26801664e-01 -1.23514093e-01 2.29738757e-01
-1.07561433e+00 8.96846771e-01 1.02439654e+00 -2.70898193e-01
-5.19073606e-01 -2.41537355e-02 -3.80787075e-01 6.73641682e-01
8.19161236e-01 -6.10815048e-01 -7.96053171e-01 -3.09634626e-01
-4.64790434e-01 -6.38408959e-02 5.57511926e-01 -1.28937960e+00
2.03040794e-01 -2.95605898e-01 4.46619123e-01 -3.15291405e-01
5.78676045e-01 -7.30329871e-01 8.53936821e-02 2.39549473e-01
-2.94881463e-01 -4.46950614e-01 1.08812645e-01 1.06241012e+00
-1.96615860e-01 2.43608236e-01 7.20915496e-01 2.66244505e-02
-1.36517072e+00 7.85843015e-01 -3.89209807e-01 6.38969302e-01
1.19208932e+00 -5.99088788e-01 -2.33236268e-01 -3.34013969e-01
-9.82284665e-01 2.50122488e-01 4.05933142e-01 4.85158741e-01
6.56443536e-01 -1.59574151e+00 -4.67006058e-01 5.06113656e-02
5.69299400e-01 -2.55931970e-02 4.64350998e-01 9.69668269e-01
-1.04355969e-01 2.21794412e-01 -3.18496883e-01 -5.34974515e-01
-1.04510748e+00 5.62538505e-01 2.61425883e-01 -2.38545567e-01
-8.71708095e-01 7.45179534e-01 4.31204408e-01 -2.08286569e-01
4.55933005e-01 -3.63803327e-01 1.78658627e-02 -8.78842473e-02
4.08904374e-01 5.74605942e-01 -1.65871188e-01 -3.97819698e-01
-5.15138447e-01 2.78272599e-01 -7.39022866e-02 1.26816511e-01
1.40643322e+00 8.43451992e-02 8.07494760e-01 4.00866032e-01
1.20332587e+00 -5.40893555e-01 -1.93239117e+00 -5.20946026e-01
3.41199897e-02 -4.20632452e-01 -8.96548629e-02 -8.53730202e-01
-1.12173200e+00 7.69532502e-01 6.27128243e-01 -4.27914530e-01
1.31725454e+00 1.58988059e-01 9.66409385e-01 5.93940794e-01
3.02100569e-01 -1.45300579e+00 6.94430947e-01 5.23801744e-01
6.29508317e-01 -1.38022041e+00 -2.19292402e-01 -1.91224143e-01
-9.72795308e-01 5.13963044e-01 8.46736133e-01 -2.48193353e-01
4.75069940e-01 -1.08230021e-02 -3.39710474e-01 -1.52421683e-01
-8.01096320e-01 -3.03576529e-01 2.09485620e-01 1.04180455e+00
-4.00962941e-02 -1.38236165e-01 2.85241306e-01 9.52706158e-01
6.20352209e-01 5.35119057e-01 2.06634805e-01 9.14219737e-01
-9.82458442e-02 -8.78826022e-01 1.03868805e-01 7.28418469e-01
-3.78787220e-01 -1.47055462e-02 -3.47422421e-01 8.15989494e-01
-3.56389105e-01 5.98580241e-01 5.63774295e-02 -2.97464192e-01
4.97763813e-01 2.72774547e-01 5.16180873e-01 -7.25132406e-01
-1.54149741e-01 -1.75627396e-01 1.90136865e-01 -8.30253124e-01
-4.64121163e-01 -7.94038534e-01 -1.29237342e+00 -1.47694990e-01
1.21849678e-01 -1.12353720e-01 -7.64109492e-02 8.77103209e-01
7.25508213e-01 4.66307968e-01 3.53410542e-01 -6.30676210e-01
-3.41513216e-01 -8.38373959e-01 -7.18847871e-01 6.70939445e-01
2.04938933e-01 -1.06136167e+00 -2.95060068e-01 2.09718317e-01] | [8.694497108459473, 0.8528569340705872] |
0d2798ec-cecc-4d4f-8d90-b2934ee1e8f1 | modeling-multi-action-policy-for-task | 1908.11546 | null | https://arxiv.org/abs/1908.11546v1 | https://arxiv.org/pdf/1908.11546v1.pdf | Modeling Multi-Action Policy for Task-Oriented Dialogues | Dialogue management (DM) plays a key role in the quality of the interaction with the user in a task-oriented dialogue system. In most existing approaches, the agent predicts only one DM policy action per turn. This significantly limits the expressive power of the conversational agent and introduces unwanted turns of interactions that may challenge users' patience. Longer conversations also lead to more errors and the system needs to be more robust to handle them. In this paper, we compare the performance of several models on the task of predicting multiple acts for each turn. A novel policy model is proposed based on a recurrent cell called gated Continue-Act-Slots (gCAS) that overcomes the limitations of the existing models. Experimental results show that gCAS outperforms other approaches. The code is available at https://leishu02.github.io/ | ['Piero Molino', 'Bing Liu', 'Hu Xu', 'Lei Shu'] | 2019-08-30 | modeling-multi-action-policy-for-task-1 | https://aclanthology.org/D19-1130 | https://aclanthology.org/D19-1130.pdf | ijcnlp-2019-11 | ['dialogue-management'] | ['natural-language-processing'] | [-2.15517238e-01 5.26680827e-01 -2.62022406e-01 -4.04688656e-01
-4.36131835e-01 -6.57143772e-01 8.26965511e-01 -1.05033360e-01
-3.16109866e-01 1.17243731e+00 5.31936049e-01 -4.86305207e-01
3.33677322e-01 -4.95004714e-01 2.38810509e-01 -3.29771847e-01
3.63730669e-01 6.58300102e-01 3.22447628e-01 -8.83787990e-01
3.49717706e-01 5.02566844e-02 -1.07551944e+00 4.05291289e-01
8.78358901e-01 3.71375024e-01 4.29692060e-01 8.66964757e-01
-5.14104843e-01 1.26673448e+00 -9.22874153e-01 -3.45795482e-01
2.87456643e-02 -6.87887907e-01 -1.31707180e+00 1.33075044e-01
-4.05417293e-01 -5.46785831e-01 -3.50896925e-01 6.64325535e-01
5.59266269e-01 5.26363194e-01 2.65487850e-01 -1.24489665e+00
2.16681004e-01 7.38792837e-01 8.28431249e-02 2.46037170e-01
8.34812462e-01 1.40320227e-01 8.58234346e-01 -4.57616121e-01
4.98942852e-01 1.47908163e+00 3.79339129e-01 1.07953322e+00
-9.58902478e-01 -1.52726665e-01 3.79754573e-01 8.72537121e-02
-7.89085150e-01 -7.26072907e-01 6.51351392e-01 -2.38913774e-01
1.15417051e+00 6.42206550e-01 6.44050002e-01 1.12001085e+00
1.38612598e-01 1.01525223e+00 1.08993483e+00 -4.75799471e-01
2.09125280e-01 3.35381210e-01 3.30436319e-01 5.91132879e-01
-5.13383508e-01 -5.32105446e-01 -5.58677971e-01 -4.85553443e-01
7.14650452e-01 -1.65694967e-01 -2.64564127e-01 1.43685490e-01
-7.99692452e-01 9.73959744e-01 -2.28784278e-01 5.69186389e-01
-3.58474731e-01 -5.01623034e-01 5.19255102e-01 5.33455253e-01
5.27553380e-01 6.79719150e-01 -3.91557574e-01 -1.13191307e+00
-1.76865295e-01 5.52526653e-01 1.54084432e+00 8.54658008e-01
3.66578102e-01 -2.52773792e-01 -3.89084756e-01 1.14276361e+00
1.59847766e-01 -3.44613828e-02 4.27844763e-01 -1.28098750e+00
4.17222053e-01 8.34067523e-01 4.72509384e-01 -7.19988048e-01
-6.80503666e-01 1.80531755e-01 -5.82044840e-01 -1.58978909e-01
6.10857606e-01 -4.74227816e-01 -2.11533532e-01 1.55863512e+00
4.36267793e-01 -3.76327515e-01 1.07462063e-01 6.91759050e-01
7.87433207e-01 6.96242154e-01 -5.17881811e-02 -6.17021441e-01
1.23042583e+00 -1.23077238e+00 -1.33289015e+00 -3.15679014e-01
7.74344802e-01 -1.02177823e+00 1.19548309e+00 3.09471607e-01
-1.26475084e+00 -1.73864380e-01 -6.30398214e-01 3.19092274e-02
-5.80121800e-02 -1.19341567e-01 6.14919722e-01 4.66471881e-01
-9.68175054e-01 3.02787125e-01 -5.85759223e-01 -4.02011365e-01
-3.50267738e-01 3.02854389e-01 8.60289186e-02 4.05458242e-01
-1.42175913e+00 1.31385040e+00 9.41734836e-02 2.01394245e-01
-2.29288027e-01 2.24715173e-02 -6.39661133e-01 -1.88626826e-01
5.05534470e-01 -2.92865664e-01 2.29966688e+00 -7.27206469e-01
-2.30855775e+00 5.41888952e-01 -3.80330861e-01 -2.70130068e-01
8.26455534e-01 -2.84695596e-01 -2.63533834e-02 -1.28961816e-01
-5.32838702e-02 2.43641481e-01 3.28778237e-01 -1.07543194e+00
-8.62958968e-01 -6.00054450e-02 7.15675414e-01 8.41682553e-01
-3.20166387e-02 1.13543123e-01 -5.18350244e-01 -2.00217485e-01
-2.04782099e-01 -1.16922605e+00 -3.84266496e-01 -7.81019628e-01
-3.91667187e-01 -7.95559883e-01 6.27637982e-01 -6.83209538e-01
1.78349113e+00 -1.88051212e+00 1.90144613e-01 -2.66573220e-01
2.46833086e-01 5.51085174e-01 1.97719127e-01 9.84500647e-01
4.45505828e-01 1.02841876e-01 8.29590186e-02 -5.44997573e-01
1.99616626e-02 4.16627228e-01 7.71815851e-02 8.44977796e-03
-3.41570467e-01 5.85922420e-01 -8.12041700e-01 -4.13840950e-01
3.05074036e-01 8.03003982e-02 -3.61554325e-01 8.66454363e-01
-4.84956026e-01 8.49538863e-01 -6.02598131e-01 1.78975806e-01
3.36911470e-01 -3.13305020e-01 6.14893436e-01 5.83296895e-01
-3.06164026e-01 1.08126330e+00 -7.90769458e-01 1.34383786e+00
-5.22179604e-01 3.99738163e-01 2.80461639e-01 -5.56922078e-01
6.01273358e-01 6.52531147e-01 3.35952193e-01 -5.68965435e-01
7.25877881e-02 -9.32817534e-02 3.43855262e-01 -5.94436407e-01
7.85078883e-01 9.72188562e-02 -2.19847515e-01 7.56802201e-01
-2.96841323e-01 -8.98846090e-02 3.62121910e-01 4.16288763e-01
8.92878234e-01 -1.99001595e-01 6.39712274e-01 -2.95861973e-03
6.77327812e-01 -4.15679738e-02 7.48647928e-01 9.01590824e-01
-5.47022164e-01 -3.53800058e-02 8.55667472e-01 -4.83578056e-01
-9.28697824e-01 -4.44032192e-01 1.66267022e-01 1.49575818e+00
4.64181080e-02 -7.75312126e-01 -9.47435498e-01 -6.40563726e-01
-3.77838582e-01 9.50293303e-01 -2.25655183e-01 1.22141443e-01
-8.20684433e-01 -5.52282810e-01 4.92524713e-01 7.52105638e-02
7.21089661e-01 -1.24357915e+00 -4.50296283e-01 5.20639360e-01
-1.01515603e+00 -1.18636572e+00 -5.87856114e-01 -1.68310851e-01
-6.22080863e-01 -1.00827706e+00 -2.01359153e-01 -4.86427307e-01
2.98612714e-01 3.57242465e-01 1.06072950e+00 2.72239059e-01
2.08693430e-01 3.54377568e-01 -5.67972183e-01 -4.73594487e-01
-9.61883008e-01 3.50546598e-01 1.48668319e-01 -2.75261939e-01
3.90416414e-01 -2.90254921e-01 -4.63054895e-01 5.84881783e-01
-3.13837826e-01 5.67234159e-01 6.01267889e-02 1.01131892e+00
-3.98324251e-01 -1.89942792e-01 7.12665200e-01 -1.07895648e+00
1.40357995e+00 -3.85266572e-01 -2.69152313e-01 9.78277773e-02
-5.83022237e-01 -1.13291599e-01 5.43723226e-01 -2.26406381e-01
-1.40955985e+00 -1.49641454e-01 -3.18254918e-01 4.98077214e-01
-2.27044627e-01 2.28590161e-01 7.02619031e-02 2.23728284e-01
3.56318831e-01 1.45071805e-01 3.14646751e-01 -5.36484241e-01
-3.54824737e-02 1.10981309e+00 -2.27337524e-01 -4.09677058e-01
1.54016078e-01 -1.05014734e-01 -6.71273053e-01 -1.16590035e+00
-7.76221931e-01 -7.25733280e-01 -3.93537372e-01 -5.76782644e-01
5.85446715e-01 -6.43159866e-01 -1.10117698e+00 4.34315950e-01
-1.28779626e+00 -6.84052110e-01 2.55621314e-01 2.44591087e-01
-5.89949727e-01 4.45343047e-01 -1.04191577e+00 -1.40421391e+00
-3.57169807e-01 -1.12056863e+00 4.69860643e-01 5.37304103e-01
-8.17694783e-01 -9.59905624e-01 1.74998604e-02 7.73536086e-01
4.53711301e-01 -3.49463701e-01 6.32345974e-01 -8.90901268e-01
-1.41275138e-01 -3.15246806e-02 4.06298459e-01 1.20221838e-01
2.38812312e-01 -1.58899948e-01 -8.09105039e-01 -1.48183450e-01
2.72731215e-01 -4.17459756e-01 1.35050774e-01 1.85193673e-01
5.68086505e-01 -7.83682466e-01 -2.37105682e-01 -3.48277748e-01
6.40623748e-01 6.95075035e-01 6.21705770e-01 3.91256094e-01
2.94265598e-01 7.64303923e-01 9.47986066e-01 7.54228652e-01
8.13434780e-01 8.56025219e-01 4.23453972e-02 1.15927503e-01
4.96973038e-01 -1.09669000e-01 5.85655391e-01 1.04882801e+00
-3.13134164e-01 -4.68611032e-01 -9.03752744e-01 1.88572705e-01
-2.40490675e+00 -1.04625463e+00 -5.41377887e-02 1.95157754e+00
1.16283977e+00 2.48550773e-01 4.97630328e-01 -2.50678323e-03
7.12214947e-01 3.23373437e-01 -2.73407549e-01 -9.59441066e-01
2.30928525e-01 -3.31270218e-01 -2.37116460e-02 1.15975225e+00
-9.25029278e-01 1.12185013e+00 6.48867750e+00 3.93189669e-01
-6.86789334e-01 1.59079507e-01 6.59406543e-01 -3.82507630e-02
1.54731855e-01 -7.62872398e-02 -9.63415742e-01 5.00352800e-01
1.02137625e+00 -3.78513366e-01 5.20380020e-01 7.53975749e-01
7.99256802e-01 -6.00747645e-01 -9.47786808e-01 7.69109726e-01
-8.37564990e-02 -9.86110210e-01 -3.31495047e-01 1.18946478e-01
2.22820505e-01 -2.49383047e-01 -4.18020606e-01 5.52453637e-01
5.61313272e-01 -8.52419496e-01 1.49394393e-01 4.74404544e-01
1.67308912e-01 -6.48454487e-01 8.73611510e-01 9.52849448e-01
-7.82311857e-01 -1.44463241e-01 -6.86198547e-02 -5.47030151e-01
3.23830515e-01 4.08245102e-02 -1.44110477e+00 2.09208503e-01
3.09833080e-01 1.66195020e-01 -5.25439568e-02 5.72550535e-01
-1.94877043e-01 5.57317197e-01 -1.03766724e-01 -4.43733245e-01
3.23400617e-01 -4.36799169e-01 7.97006190e-01 1.06203210e+00
-2.52837300e-01 5.82469940e-01 6.99361086e-01 1.95003003e-01
5.42817153e-02 1.96729898e-01 -6.42514229e-01 -7.41671920e-02
7.08261073e-01 1.15253735e+00 -2.88148731e-01 -4.34169114e-01
-4.23670888e-01 9.86481249e-01 3.24938864e-01 1.28560558e-01
-5.72769046e-01 7.35400468e-02 6.82671905e-01 3.91639061e-02
-3.20292950e-01 -3.53600204e-01 -7.82122612e-02 -1.07440865e+00
6.66049048e-02 -1.35532129e+00 2.83141077e-01 -2.34254509e-01
-9.65195060e-01 6.60047472e-01 -2.49856561e-01 -1.00440323e+00
-8.89855444e-01 -1.33245528e-01 -6.14923954e-01 8.43126774e-01
-1.03794336e+00 -7.54005134e-01 -1.05060555e-01 3.74555618e-01
1.14363515e+00 -9.77293998e-02 1.30129385e+00 4.79447208e-02
-7.24196076e-01 3.52275908e-01 -1.65662736e-01 1.06212050e-01
8.30713034e-01 -1.38361228e+00 2.54370660e-01 2.79508680e-01
-4.95397925e-01 6.28116608e-01 9.72552896e-01 -4.80909288e-01
-1.11819208e+00 -4.51550990e-01 1.30133498e+00 -3.52986872e-01
4.16923344e-01 -3.28253329e-01 -8.36891413e-01 7.63831258e-01
6.50275290e-01 -9.26999211e-01 9.64473784e-01 4.59870726e-01
3.20620179e-01 3.79049927e-01 -9.19237018e-01 8.34170878e-01
6.55184925e-01 -5.27871549e-01 -6.66993141e-01 5.18384814e-01
5.33126235e-01 -7.33542085e-01 -7.43047953e-01 7.07090925e-03
4.79153365e-01 -1.17884302e+00 4.29753989e-01 -5.85580766e-01
9.98107940e-02 1.33585617e-01 2.61623412e-01 -1.42113221e+00
-1.54626891e-01 -1.28042281e+00 -2.62489796e-01 9.68611717e-01
4.77279216e-01 -6.75614953e-01 5.39315820e-01 1.18945777e+00
2.94995494e-02 -7.81903028e-01 -7.02679336e-01 -4.30291325e-01
-7.98685178e-02 1.57459769e-02 4.00004238e-01 1.05680835e+00
8.81004632e-01 9.29495454e-01 -7.15311408e-01 -1.73842430e-01
3.40949036e-02 -1.00684069e-01 9.83521402e-01 -1.23429751e+00
-2.58163184e-01 -3.64942044e-01 2.16085196e-01 -1.46758425e+00
6.48752749e-02 -2.70805150e-01 1.35730013e-01 -1.51255572e+00
-4.21482977e-03 -4.16638017e-01 2.96990544e-01 5.10954320e-01
-2.31485650e-01 -4.70241606e-01 3.50079685e-01 3.00473899e-01
-8.67004037e-01 6.36853099e-01 1.20401752e+00 2.00874865e-01
-8.65355551e-01 6.06155157e-01 -6.31107211e-01 9.58954036e-01
1.48110116e+00 -6.38752505e-02 -3.93318743e-01 4.90899347e-02
-3.82364914e-02 6.70820117e-01 -3.60676110e-01 -5.29196560e-01
4.02543813e-01 -4.12623316e-01 -2.64199197e-01 -4.44804013e-01
5.86080313e-01 -3.31338048e-01 -2.60109454e-01 4.17330831e-01
-5.81327736e-01 2.26016939e-01 -1.09956684e-02 2.57860303e-01
-2.18849093e-01 -2.26853684e-01 6.73677027e-01 -4.40335810e-01
-4.11268234e-01 -2.00884283e-01 -1.14306498e+00 -9.75560844e-02
9.59826350e-01 5.12983464e-02 -3.00555080e-01 -1.15611947e+00
-9.43204463e-01 6.42963171e-01 2.51719683e-01 5.97971141e-01
3.41939151e-01 -8.96588326e-01 -5.45065641e-01 -2.76988387e-01
-8.90922099e-02 -1.01213276e-01 1.80722386e-01 7.41844952e-01
-4.17226881e-01 7.21632123e-01 -3.20958183e-03 -2.77216464e-01
-1.74009156e+00 -1.23450339e-01 5.42596757e-01 -6.73971236e-01
-5.62490702e-01 6.15477383e-01 -1.57381296e-01 -6.90473974e-01
5.70444763e-01 1.61315769e-01 -7.00943232e-01 1.21944025e-01
6.82624578e-01 4.45497394e-01 -1.72114566e-01 -5.69899797e-01
1.84978973e-02 -4.22576070e-01 -5.76829374e-01 -4.67454344e-01
1.09642220e+00 -5.47648311e-01 -3.30934018e-01 9.99109566e-01
4.67945069e-01 -9.59707573e-02 -1.00113785e+00 -4.20284510e-01
1.75733939e-01 -4.92638290e-01 -2.78587341e-01 -9.15145278e-01
-2.03958809e-01 4.37326580e-01 1.47208735e-01 8.02999496e-01
6.61268353e-01 -2.79467702e-01 8.33778143e-01 6.20564401e-01
4.75675881e-01 -1.64882267e+00 1.50274128e-01 1.08851385e+00
9.28766012e-01 -1.39394999e+00 -2.68169969e-01 -4.79076982e-01
-1.14456201e+00 1.04186642e+00 1.04658735e+00 5.21272182e-01
2.87845224e-01 9.50667486e-02 4.37959194e-01 -3.88172939e-02
-1.30203903e+00 -5.10947742e-02 -3.05167854e-01 1.64985314e-01
9.32986438e-01 3.13991696e-01 -8.58802736e-01 3.74988496e-01
-3.07080090e-01 -1.70215517e-01 9.35906947e-01 1.29206038e+00
-5.42053461e-01 -1.58508301e+00 -2.14342937e-01 3.32032621e-01
-4.73498911e-01 1.58521719e-02 -9.13473666e-01 4.84496772e-01
-4.29731309e-01 1.57648134e+00 -1.84641168e-01 -3.23244810e-01
4.60454077e-01 5.34012198e-01 7.54807983e-03 -7.17454135e-01
-9.23892081e-01 2.38004893e-01 9.11849380e-01 -5.66379905e-01
-4.05684054e-01 -6.37232840e-01 -1.36259472e+00 -8.24645817e-01
-2.86548287e-01 5.84616661e-01 2.81654328e-01 8.84947598e-01
2.77841210e-01 3.68152499e-01 1.02990675e+00 -5.81805766e-01
-7.05618322e-01 -1.44199097e+00 -3.58353853e-01 2.26536527e-01
2.47244447e-01 -5.24115920e-01 -2.16065347e-01 -3.29465061e-01] | [12.85023021697998, 7.975614070892334] |
94683713-2456-4ab4-83dc-1cf7ce255f55 | a-theory-of-unsupervised-translation | 2211.11081 | null | https://arxiv.org/abs/2211.11081v1 | https://arxiv.org/pdf/2211.11081v1.pdf | A Theory of Unsupervised Translation Motivated by Understanding Animal Communication | Recent years have seen breakthroughs in neural language models that capture nuances of language, culture, and knowledge. Neural networks are capable of translating between languages -- in some cases even between two languages where there is little or no access to parallel translations, in what is known as Unsupervised Machine Translation (UMT). Given this progress, it is intriguing to ask whether machine learning tools can ultimately enable understanding animal communication, particularly that of highly intelligent animals. Our work is motivated by an ambitious interdisciplinary initiative, Project CETI, which is collecting a large corpus of sperm whale communications for machine analysis. We propose a theoretical framework for analyzing UMT when no parallel data are available and when it cannot be assumed that the source and target corpora address related subject domains or posses similar linguistic structure. The framework requires access to a prior probability distribution that should assign non-zero probability to possible translations. We instantiate our framework with two models of language. Our analysis suggests that accuracy of translation depends on the complexity of the source language and the amount of ``common ground'' between the source language and target prior. We also prove upper bounds on the amount of data required from the source language in the unsupervised setting as a function of the amount of data required in a hypothetical supervised setting. Surprisingly, our bounds suggest that the amount of source data required for unsupervised translation is comparable to the supervised setting. For one of the language models which we analyze we also prove a nearly matching lower bound. Our analysis is purely information-theoretic and as such can inform how much source data needs to be collected, but does not yield a computationally efficient procedure. | ['Orr Paradise', 'Adam Tauman Kalai', 'David F. Gruber', 'Shafi Goldwasser'] | 2022-11-20 | null | null | null | null | ['unsupervised-machine-translation', 'culture'] | ['natural-language-processing', 'speech'] | [ 4.34470683e-01 3.16627562e-01 -2.92720199e-01 -2.81463683e-01
-7.54459620e-01 -8.46577466e-01 9.15970564e-01 3.39136839e-01
-8.24071050e-01 8.31077337e-01 3.07875276e-01 -6.48241043e-01
1.09019853e-01 -8.49331737e-01 -1.00108922e+00 -4.80309069e-01
4.89104316e-02 7.69140184e-01 -4.63441946e-02 -3.70133460e-01
-1.03149645e-01 2.05565289e-01 -1.05297196e+00 6.27853125e-02
7.62426198e-01 5.16318440e-01 3.11605960e-01 7.27080345e-01
-2.89032519e-01 4.22090858e-01 -3.08640361e-01 -6.74030900e-01
5.29887855e-01 -8.71727645e-01 -9.71444547e-01 -1.07686400e-01
1.66214462e-02 1.93211343e-02 -1.62614018e-01 1.15494990e+00
2.55307794e-01 -1.89824745e-01 7.55801141e-01 -8.96736801e-01
-6.55893803e-01 1.00710320e+00 -2.27534905e-01 3.54304910e-01
8.01088884e-02 4.43200720e-03 1.20540774e+00 -4.33931857e-01
7.27106273e-01 1.06938517e+00 5.51571190e-01 5.71382105e-01
-1.46705008e+00 -3.97644788e-01 -2.62610286e-01 -2.12334812e-01
-9.57830250e-01 -5.93696177e-01 5.73391557e-01 -5.43126643e-01
5.76857746e-01 1.26240015e-01 6.63072765e-01 1.07240057e+00
2.82878727e-01 5.94312668e-01 1.19468701e+00 -6.34139180e-01
2.53160924e-01 2.47091562e-01 2.75818836e-02 6.28964067e-01
3.60102087e-01 1.81829914e-01 -8.07852030e-01 -1.33446544e-01
6.63475394e-01 -4.03636098e-01 -1.59777001e-01 -8.37739035e-02
-1.50992203e+00 8.84010077e-01 2.59108320e-02 5.59373498e-01
-1.09389983e-01 1.19532719e-01 3.18117827e-01 7.35292554e-01
4.57609653e-01 4.93012875e-01 -5.80118120e-01 -1.41376749e-01
-8.59287441e-01 7.46441633e-02 1.16272712e+00 1.06205070e+00
7.83432782e-01 -3.38405162e-01 4.70436364e-01 6.48014784e-01
1.77173257e-01 6.82579279e-01 5.54207146e-01 -7.71550775e-01
6.15949452e-01 2.71234125e-01 1.24417104e-01 -7.65432417e-01
-3.42143148e-01 -3.17412317e-01 -6.37600899e-01 -3.60541612e-01
9.18043435e-01 -3.61107260e-01 -4.14839625e-01 2.25536585e+00
-5.73205389e-02 -2.94305176e-01 3.57944489e-01 7.50690699e-01
2.32688114e-01 6.40099049e-01 -1.91412985e-01 -3.28250825e-01
1.33624327e+00 -4.43936348e-01 -3.39215994e-01 -4.97094899e-01
8.23226988e-01 -8.25092554e-01 1.10853934e+00 -2.82933954e-02
-1.10385489e+00 -5.83158582e-02 -9.00660932e-01 -6.54659420e-02
-3.13524932e-01 -3.21778059e-02 9.10648704e-01 6.46332324e-01
-1.00361967e+00 6.26847088e-01 -9.17978227e-01 -7.88678646e-01
3.08859050e-01 4.28190291e-01 -4.42648053e-01 1.87705144e-01
-1.10443306e+00 9.85701919e-01 2.57872075e-01 1.65020097e-02
-6.95549428e-01 -2.36775458e-01 -7.09947944e-01 -1.00611120e-01
2.23778740e-01 -6.75428510e-01 1.27783382e+00 -1.51219809e+00
-1.38101411e+00 1.05029535e+00 -2.66399682e-01 -6.72965765e-01
4.90251720e-01 8.54788199e-02 -2.63698604e-02 -1.50827067e-02
2.88546324e-01 5.00979900e-01 3.65464777e-01 -9.44263279e-01
-4.99201357e-01 -4.53807294e-01 1.79715976e-01 9.44594815e-02
-2.16913477e-01 2.55528092e-01 -3.02182138e-01 -5.73149979e-01
8.57853442e-02 -1.18834186e+00 -6.45797625e-02 1.95424780e-01
-2.97468752e-01 -6.72303438e-02 -2.26995479e-02 -4.93368328e-01
5.70253253e-01 -1.96320152e+00 3.94193470e-01 1.49013802e-01
1.65265292e-01 -1.38632551e-01 -3.39949608e-01 5.65565467e-01
2.82911986e-01 3.18805873e-01 -5.80805719e-01 -1.48983553e-01
1.88569859e-01 6.38672411e-01 -4.88168210e-01 6.06324971e-01
1.89370379e-01 9.24359083e-01 -8.44774067e-01 -4.11392272e-01
-4.49166983e-01 1.12132624e-01 -6.11659944e-01 -1.90847889e-02
-2.36719728e-01 4.67046499e-01 -4.04733479e-01 2.62067348e-01
2.45448634e-01 -2.90324807e-01 4.46478844e-01 2.61788517e-01
-2.11910397e-01 6.11811280e-01 -6.04016542e-01 1.72977304e+00
-4.86407518e-01 8.65792811e-01 1.74901500e-01 -1.14615691e+00
6.96977258e-01 1.68556929e-01 1.88053042e-01 -4.77537394e-01
3.56506437e-01 4.18718696e-01 5.24800658e-01 -3.57220411e-01
2.99068421e-01 -6.66088760e-01 -2.63299972e-01 8.01199496e-01
1.66889817e-01 4.52692658e-02 3.94330949e-01 1.47286177e-01
1.07338333e+00 6.65751547e-02 1.50274977e-01 -6.78194523e-01
5.97373955e-02 2.44228214e-01 6.84904337e-01 7.34695494e-01
-1.78716090e-02 1.40810177e-01 6.81135297e-01 -2.83822089e-01
-1.34327531e+00 -1.02748907e+00 -1.44652143e-01 1.12481427e+00
1.38720289e-01 -3.01475257e-01 -8.53907466e-01 -3.15453172e-01
-2.90562779e-01 5.39840162e-01 -6.04055583e-01 2.76295617e-02
-5.14099061e-01 -9.55149472e-01 8.26755047e-01 3.55881900e-01
1.98367000e-01 -8.01345766e-01 -7.13368118e-01 3.21724713e-02
-1.34047076e-01 -1.06018209e+00 -3.60179961e-01 4.80958462e-01
-8.20677578e-01 -7.98502088e-01 -7.11463332e-01 -8.01302373e-01
6.47293448e-01 -1.23596862e-01 1.07842886e+00 4.24828455e-02
2.15363473e-01 2.83620447e-01 -1.67141721e-01 -5.41437864e-01
-8.61290634e-01 2.63756186e-01 3.24612886e-01 -1.46858454e-01
5.45885563e-01 -7.33122289e-01 -8.83517712e-02 1.42486170e-01
-1.01668751e+00 1.61194757e-01 6.70809388e-01 8.24506104e-01
2.43120924e-01 -3.56225759e-01 4.86388594e-01 -8.29934716e-01
5.74000776e-01 -6.61217988e-01 -7.80305803e-01 2.37340122e-01
-1.22864000e-01 5.12960017e-01 7.28923321e-01 -5.80875576e-01
-6.01260662e-01 -8.86271670e-02 -2.10685600e-02 2.85000801e-01
-1.16383471e-01 7.22580492e-01 -5.45211956e-02 1.19077750e-01
5.85681856e-01 2.83185124e-01 1.15025416e-01 -4.38557446e-01
2.66052246e-01 7.22119272e-01 4.55220014e-01 -1.02715921e+00
8.53605330e-01 4.09271419e-01 1.05769234e-02 -1.08303177e+00
-6.62622929e-01 1.17485700e-02 -8.11624169e-01 1.26001865e-01
8.14825654e-01 -7.44521499e-01 -5.46409845e-01 3.40746902e-02
-1.27479494e+00 -5.63788295e-01 -3.62383813e-01 6.62361145e-01
-7.81256318e-01 2.86249191e-01 -6.04538381e-01 -7.76995838e-01
-1.99295878e-02 -1.05349505e+00 7.40418494e-01 -6.22720532e-02
-3.89864653e-01 -1.11908603e+00 1.48804516e-01 2.74000198e-01
4.55534220e-01 -9.70007032e-02 1.22695649e+00 -1.07647634e+00
-5.34389615e-01 1.23782139e-02 1.25546847e-02 2.15922028e-01
3.30098085e-02 -4.25705433e-01 -5.44318914e-01 -9.43518728e-02
1.94585919e-01 -4.22216386e-01 6.41059637e-01 3.02069709e-02
3.92647147e-01 -5.91079235e-01 -1.21180721e-01 3.22929502e-01
1.13276708e+00 -1.08336687e-01 1.88503593e-01 1.94718525e-01
3.40947449e-01 9.62579608e-01 -2.04086211e-02 -8.47588480e-02
4.68078047e-01 5.83093882e-01 -2.53400385e-01 9.80229452e-02
2.10956678e-01 -3.55796158e-01 4.64065403e-01 1.29809153e+00
1.31746451e-03 -2.01142192e-01 -1.21154690e+00 8.52685630e-01
-1.62747240e+00 -8.53164434e-01 1.97460815e-01 2.30413723e+00
1.26170790e+00 1.98836401e-01 1.88441098e-01 -2.65426248e-01
4.89712566e-01 -3.30364138e-01 -4.22325701e-01 -4.39430386e-01
-3.51865917e-01 1.78235471e-01 7.22535193e-01 7.12055087e-01
-6.66071355e-01 8.44715595e-01 6.81088209e+00 4.71560866e-01
-9.96613920e-01 1.25124454e-01 5.37942231e-01 -6.20159954e-02
-6.04695320e-01 1.80655286e-01 -5.89238882e-01 4.64608371e-01
1.24183357e+00 -6.34736896e-01 7.05582500e-01 4.10727829e-01
5.11852931e-03 -9.04154703e-02 -1.48676467e+00 5.96151710e-01
2.46742740e-02 -1.21099222e+00 -9.89913717e-02 3.84545565e-01
4.78882909e-01 4.41145331e-01 -1.80588245e-01 -2.35535880e-03
6.39086664e-01 -8.38705420e-01 8.51710379e-01 3.44120681e-01
6.10426188e-01 -6.32749438e-01 6.70885563e-01 9.27181005e-01
-6.96897924e-01 1.49116442e-01 -5.36613464e-01 -3.12648416e-01
8.95323791e-03 3.19577157e-01 -6.26183331e-01 2.08479255e-01
2.95267850e-01 2.88129449e-01 -2.96975553e-01 5.55152476e-01
-1.66329220e-01 8.48230958e-01 -7.84254372e-01 -1.51415497e-01
2.40990594e-01 -3.71986657e-01 6.92620695e-01 1.13984704e+00
2.37652585e-01 1.30880684e-01 1.36490002e-01 9.13816512e-01
-3.82191479e-01 2.82350123e-01 -8.37391496e-01 -4.93108422e-01
3.87090713e-01 7.63207316e-01 -9.09251094e-01 -1.46967530e-01
-5.98044693e-01 8.37998807e-01 3.44230920e-01 1.64535731e-01
-4.26851034e-01 -6.17422089e-02 4.91127908e-01 5.49690761e-02
1.34676278e-01 -4.99195099e-01 -3.61874163e-01 -1.34646714e+00
-4.23921265e-05 -8.40828478e-01 -1.04510054e-01 -2.83082604e-01
-1.36766970e+00 5.66732705e-01 1.65620059e-01 -8.00015628e-01
-4.33756024e-01 -5.57173729e-01 -2.48447701e-01 9.31110799e-01
-1.18288231e+00 -1.04398942e+00 6.41168296e-01 2.16205150e-01
4.05345082e-01 -2.36184657e-01 7.97577858e-01 7.29083121e-02
-2.45056033e-01 6.33451819e-01 2.23478884e-01 3.64820540e-01
3.97751212e-01 -9.59022284e-01 4.36902881e-01 9.38602686e-01
5.41641831e-01 9.88813102e-01 8.84229124e-01 -5.53848147e-01
-1.67248392e+00 -8.05551291e-01 1.29412234e+00 -6.26548529e-01
1.01062810e+00 -6.91902995e-01 -6.78500116e-01 8.73905241e-01
2.69784212e-01 -3.03385586e-01 1.05139232e+00 2.09797978e-01
-3.55664343e-01 1.93820357e-01 -8.69485795e-01 6.98416591e-01
9.67959762e-01 -7.60834157e-01 -8.35499883e-01 3.54038388e-01
7.15954661e-01 1.54697495e-02 -7.76745617e-01 5.66989928e-02
5.43198884e-01 -5.13672352e-01 3.64035368e-01 -7.55545855e-01
6.59648478e-01 -2.84644276e-01 -6.06266856e-01 -1.07603490e+00
5.84411100e-02 -7.72035778e-01 4.16529268e-01 1.13745856e+00
8.59633267e-01 -5.75128615e-01 5.90396702e-01 7.24513531e-01
1.70883864e-01 -6.71662450e-01 -1.08951223e+00 -8.39840770e-01
6.43042088e-01 -5.01083136e-01 3.73859942e-01 9.86022234e-01
1.45385310e-01 7.27542579e-01 -2.05433711e-01 -3.02354037e-03
6.52976573e-01 2.04958186e-01 7.98965573e-01 -1.25441647e+00
-6.65327013e-01 -4.01785254e-01 -4.35788184e-01 -1.22655249e+00
4.06998783e-01 -1.24601710e+00 1.34206519e-01 -1.18308556e+00
3.21096748e-01 -3.92493993e-01 1.12450354e-01 5.37515819e-01
3.40609878e-01 2.82134265e-01 1.99708879e-01 4.73519176e-01
-3.16063553e-01 3.77806753e-01 8.20668161e-01 -5.98875172e-02
-4.84425873e-02 -6.90350085e-02 -9.27183509e-01 9.20375466e-01
8.12717974e-01 -6.12388492e-01 -2.32971013e-01 -8.19941103e-01
6.35802448e-01 6.91907927e-02 2.15231091e-01 -5.75317681e-01
2.85865843e-01 -2.92357057e-01 1.03071826e-02 -1.14850089e-01
8.52663349e-03 -7.49066114e-01 4.12014164e-02 2.80540854e-01
-7.66828179e-01 4.40523922e-02 1.08925089e-01 6.11728311e-01
6.65177107e-02 -2.73940265e-01 7.42129266e-01 -2.52037823e-01
-1.29129559e-01 -7.09231198e-03 -6.70966864e-01 2.88636655e-01
6.58396065e-01 1.30440861e-01 -1.54327691e-01 -4.19023275e-01
-5.98960876e-01 3.45803574e-02 8.35198700e-01 1.72446951e-01
1.61242083e-01 -1.17182934e+00 -8.01986814e-01 2.67291963e-01
3.15320566e-02 -3.73666495e-01 -4.11751091e-01 8.01442623e-01
-4.15125757e-01 3.86600286e-01 -1.29271284e-01 -4.04518396e-01
-9.12901044e-01 4.62893963e-01 2.06498995e-01 2.31838524e-02
-3.93250346e-01 6.20955884e-01 3.24162602e-01 -5.75341940e-01
-1.41240448e-01 -3.71579885e-01 2.80251741e-01 -1.70270130e-01
3.55699450e-01 -9.33352336e-02 -3.56038064e-01 -1.01601362e+00
-1.69645190e-01 3.90592724e-01 -1.08559839e-01 -4.95139360e-01
1.38659108e+00 -1.79292947e-01 -4.83424693e-01 8.59182179e-01
1.08162200e+00 2.51710445e-01 -8.71312857e-01 -4.38999504e-01
2.28145912e-01 -5.94214648e-02 -2.83021331e-01 -5.74095666e-01
-5.79719543e-01 9.75809276e-01 1.40927792e-01 1.58681184e-01
8.73960018e-01 4.49698418e-01 7.89003253e-01 7.97705173e-01
5.87091625e-01 -9.55748558e-01 -4.64660645e-01 7.02579975e-01
5.69691181e-01 -1.11083376e+00 -1.20865874e-01 -8.71173590e-02
-4.18151110e-01 8.62305999e-01 9.49341655e-02 -7.30768368e-02
3.94005775e-01 4.40717697e-01 -4.10338119e-02 4.32669185e-03
-8.30872893e-01 -2.40015283e-01 1.21088251e-01 5.30363679e-01
5.18664539e-01 7.04104081e-02 -5.63651800e-01 7.14181662e-01
-7.63491929e-01 -2.14345872e-01 5.38657069e-01 8.06764841e-01
-4.89562720e-01 -1.35189641e+00 -1.02462076e-01 3.47022891e-01
-7.42381811e-01 -3.66822541e-01 -7.62548268e-01 6.44489348e-01
9.78190675e-02 8.78786385e-01 9.59102139e-02 -1.55910015e-01
-1.31283358e-01 1.33810982e-01 6.95078254e-01 -7.23786354e-01
-3.77389938e-01 9.49635580e-02 1.47758693e-01 -1.03833981e-01
-6.30561531e-01 -6.52096391e-01 -1.15416896e+00 -4.51064497e-01
-3.18883240e-01 5.10599911e-01 5.94534159e-01 1.37803900e+00
6.54869825e-02 -1.97367236e-01 2.18958110e-01 -6.26281023e-01
-4.19031352e-01 -7.76412249e-01 -4.25759673e-01 1.05896637e-01
3.51505548e-01 -1.60747588e-01 -3.28160018e-01 3.15716267e-01] | [11.085549354553223, 9.936565399169922] |
d9f62205-0585-47ea-9b5e-e76d6735e282 | selective-inference-for-sparse-high-order | null | null | https://icml.cc/Conferences/2017/Schedule?showEvent=801 | http://proceedings.mlr.press/v70/suzumura17a/suzumura17a.pdf | Selective Inference for Sparse High-Order Interaction Models |
Finding statistically significant high-order interactions in predictive modeling is important but challenging task because the possible number of high-order interactions is extremely large (e.g., $> 10^{17}$). In this paper we study feature selection and statistical inference for sparse high-order interaction models. Our main contribution is to extend recently developed selective inference framework for linear models to high-order interaction models by developing a novel algorithm for efficiently characterizing the selection event for the selective inference of high-order interactions. We demonstrate the effectiveness of the proposed algorithm by applying it to an HIV drug response prediction problem.
| ['Kazuya Nakagawa', 'Ichiro Takeuchi', 'Yuta Umezu', 'Shinya Suzumura', 'Koji Tsuda'] | 2017-08-01 | null | null | null | icml-2017-8 | ['drug-response-prediction'] | ['medical'] | [ 6.68890595e-01 -1.63666204e-01 -6.92830503e-01 -4.84931707e-01
-6.05058610e-01 -1.22493699e-01 2.84008622e-01 3.85995507e-01
-1.21652313e-01 1.35682702e+00 1.00103393e-02 -4.77344930e-01
-6.08690798e-01 -6.37781680e-01 -9.63821530e-01 -6.69598937e-01
-7.06623137e-01 9.10596132e-01 -1.50670081e-01 2.33091861e-01
4.19986606e-01 5.64185441e-01 -1.32383657e+00 2.04542980e-01
1.35074866e+00 4.28240210e-01 9.02862847e-02 5.70157528e-01
3.54431897e-01 3.58328283e-01 -2.62407809e-01 -8.64798054e-02
-1.35985956e-01 -7.13157356e-01 -5.22929728e-01 -2.45307624e-01
3.34220797e-01 -1.42618224e-01 3.49589847e-02 8.04582119e-01
5.24884224e-01 1.09212533e-01 9.04469311e-01 -1.24306726e+00
5.29097728e-02 7.31126487e-01 -5.95418870e-01 3.38665664e-01
1.63458019e-01 -2.18570128e-01 1.32690060e+00 -1.03305089e+00
5.82220316e-01 1.29425788e+00 4.75517720e-01 1.22379474e-01
-1.57492161e+00 -9.35255647e-01 1.82353720e-01 1.02259822e-01
-1.58721554e+00 -5.04433215e-01 4.13518757e-01 -5.38871944e-01
1.28982341e+00 5.06330967e-01 2.61583537e-01 7.20282316e-01
5.13008118e-01 6.57650650e-01 9.43647265e-01 -4.28581685e-01
3.47380668e-01 2.91982642e-03 5.17835021e-01 7.51897335e-01
5.35064220e-01 2.03588888e-01 -6.50081217e-01 -1.22164130e+00
7.38089979e-01 2.49760777e-01 -9.27162394e-02 1.10572681e-01
-7.69732356e-01 1.23711824e+00 -1.11462891e-01 9.25642401e-02
-5.99977314e-01 1.25239342e-01 1.41851589e-01 8.25570151e-02
6.34982765e-01 3.92572224e-01 -7.85823405e-01 5.32334223e-02
-9.39152062e-01 5.74375629e-01 7.10434437e-01 8.29513788e-01
4.45386201e-01 -2.40202799e-01 -4.56980854e-01 9.55403507e-01
2.40271866e-01 3.27726781e-01 -2.51912445e-01 -4.00814027e-01
1.36527002e-01 3.53173077e-01 1.59029931e-01 -7.84591496e-01
-5.10247529e-01 -6.07148349e-01 -1.16677940e+00 -5.78628898e-01
6.91086128e-02 -2.03664824e-01 -5.33769190e-01 1.79897392e+00
4.07291114e-01 4.86752510e-01 -4.21552300e-01 2.73571163e-01
5.85435569e-01 5.19679546e-01 4.61528420e-01 -9.85944748e-01
1.22870266e+00 -3.88090789e-01 -7.13449895e-01 1.29267201e-01
5.67529321e-01 -4.59890187e-01 5.54158509e-01 2.29489058e-01
-1.04085934e+00 2.23742388e-02 -3.98822635e-01 2.47955993e-01
2.48542130e-01 2.27734581e-01 1.18844926e+00 1.25041470e-01
-6.97344959e-01 5.48534870e-01 -5.87694168e-01 -2.68580973e-01
5.41834533e-01 9.86900985e-01 -9.40275863e-02 -2.00694948e-01
-1.27408588e+00 4.82160866e-01 2.24750221e-01 -4.23111022e-02
-4.24572527e-01 -1.17310476e+00 -5.17035306e-01 7.20642358e-02
5.11228859e-01 -8.93749356e-01 6.87399507e-01 -3.53819072e-01
-9.01249528e-01 3.43059212e-01 -8.84470820e-01 -3.52513373e-01
2.46597886e-01 -1.17964279e-02 -9.66416821e-02 -1.39241591e-01
9.53229591e-02 9.41189826e-02 5.08865237e-01 -7.09712446e-01
-7.21142054e-01 -7.08042026e-01 -4.08557922e-01 7.73629546e-02
-1.35298848e-01 3.86303425e-01 -1.63227871e-01 -6.28427982e-01
2.83125211e-02 -9.89701033e-01 -7.83777237e-01 -6.11703992e-01
-7.50440001e-01 -4.26928878e-01 3.47961158e-01 -4.22379792e-01
1.46114373e+00 -1.79518855e+00 3.25333625e-01 7.11367726e-01
2.46530041e-01 -1.87803045e-01 5.21071218e-02 4.64751333e-01
-1.68580547e-01 1.19062483e-01 -4.01225209e-01 1.69506580e-01
-2.86215872e-01 5.23172691e-02 -2.22733572e-01 4.63901371e-01
1.60739467e-01 7.39709079e-01 -6.07981861e-01 -4.77960289e-01
-8.86971727e-02 6.00792170e-01 -1.01697898e+00 6.93754032e-02
-3.21488708e-01 5.76458275e-01 -9.12177145e-01 6.37446105e-01
8.18484187e-01 -6.41772389e-01 5.72695851e-01 -6.36274964e-02
-2.04895541e-01 8.26206058e-02 -1.03263676e+00 6.52198136e-01
-5.53032011e-02 2.99390163e-02 -3.09618890e-01 -9.37723100e-01
4.85307842e-01 1.50010452e-01 9.15302515e-01 -5.69329895e-02
2.75666378e-02 1.42059416e-01 2.33826920e-01 -1.45168185e-01
-4.33227904e-02 -3.80279958e-01 2.48853769e-02 1.31198531e-02
-2.27956876e-01 3.26981694e-01 1.03186302e-01 8.93282071e-02
1.26101112e+00 -7.15911508e-01 9.69112337e-01 -4.71089631e-01
3.16783816e-01 -2.45247558e-01 7.58834124e-01 1.21230233e+00
3.53652716e-01 3.89214717e-02 7.70750821e-01 -2.26069301e-01
-6.39033198e-01 -6.69110775e-01 -7.75398314e-01 9.77900743e-01
-4.37531501e-01 -4.79403555e-01 -2.15802163e-01 -4.82903749e-01
4.67482805e-01 7.07110882e-01 -6.65104747e-01 -6.45937622e-02
-3.45715523e-01 -1.45082414e+00 1.80812672e-01 3.57174069e-01
-5.39264455e-02 -7.21541703e-01 -2.90152011e-03 3.53789091e-01
2.78693456e-02 -6.87727630e-01 -5.15495241e-01 4.57184672e-01
-9.78754878e-01 -1.20686102e+00 -2.41982237e-01 -5.58813691e-01
9.24896002e-01 -4.06402722e-02 9.38340485e-01 5.20743169e-02
-4.45662111e-01 -1.75948158e-01 -5.66097423e-02 -3.74597877e-01
-1.49385454e-02 -2.34855279e-01 2.19545752e-01 -1.56510055e-01
5.80344975e-01 -4.50836241e-01 -4.74589258e-01 1.37893409e-01
-6.54619396e-01 -2.76985347e-01 5.86961210e-01 1.24100947e+00
9.61153984e-01 4.95218992e-01 7.28493452e-01 -1.59310091e+00
5.45222998e-01 -6.86491191e-01 -7.61369765e-01 2.43258804e-01
-5.85913241e-01 4.39745151e-02 5.35456955e-01 -3.37345898e-01
-7.46757329e-01 4.09530610e-01 -1.83024749e-01 3.85916866e-02
6.97573051e-02 9.73688722e-01 -2.89011568e-01 -1.75817221e-01
2.47153908e-01 1.84479415e-01 -3.74887347e-01 -4.23033357e-01
-1.22803152e-01 3.70241851e-01 -2.19656751e-01 -4.02914643e-01
3.14747125e-01 1.18712202e-01 6.37650132e-01 -1.29034460e+00
-3.67929846e-01 -3.76506329e-01 -3.99383485e-01 4.61620003e-01
3.65405053e-01 -8.39445651e-01 -1.23683524e+00 1.55653998e-01
-9.34522152e-01 -1.56289011e-01 1.94308698e-01 8.39661241e-01
-4.22081858e-01 3.74930710e-01 -5.85278749e-01 -1.01078188e+00
-1.82072952e-01 -1.12838566e+00 9.22255993e-01 -1.06961071e-01
-3.70140523e-01 -9.95140254e-01 3.27813208e-01 1.55749366e-01
-8.39019492e-02 3.76002267e-02 1.45485425e+00 -9.16281104e-01
-3.37666214e-01 -2.56287128e-01 -1.33368403e-01 -4.44154054e-01
1.29400820e-01 2.24126115e-01 -3.11763942e-01 -2.76062518e-01
-1.36802137e-01 -1.05463704e-02 1.10425484e+00 1.39849460e+00
1.42009008e+00 -5.41461110e-01 -9.33754265e-01 2.93334514e-01
1.06934512e+00 3.89621913e-01 4.29461658e-01 -4.96693075e-01
6.47938550e-01 4.20907497e-01 8.84002090e-01 1.04886031e+00
8.04924890e-02 1.09711289e+00 -2.43555665e-01 -1.45402372e-01
6.95265114e-01 -2.11951062e-01 -1.57411903e-01 1.45955205e-01
-1.38417473e-02 -2.66783297e-01 -7.69415140e-01 1.27510801e-01
-2.08986354e+00 -9.05298293e-01 -6.67309165e-01 2.54803300e+00
1.25027788e+00 -2.72094488e-01 2.61341125e-01 -2.79925525e-01
8.28545272e-01 -5.08313000e-01 -7.03614116e-01 -2.46669516e-01
-2.15808839e-01 4.59588319e-01 5.32374799e-01 8.67069542e-01
-1.15181792e+00 7.81446636e-01 7.52296877e+00 1.01183581e+00
-6.10639751e-01 -1.52490765e-01 9.39472258e-01 -6.44219294e-02
-3.97954464e-01 9.16346069e-03 -1.24743021e+00 3.93373072e-01
9.31791723e-01 -1.15605235e-01 6.42347187e-02 4.83653069e-01
5.81450701e-01 -2.68762529e-01 -1.25762963e+00 8.64150703e-01
-1.57158822e-01 -1.24879384e+00 2.57512659e-01 4.00356829e-01
8.04431081e-01 -3.82841349e-01 1.43609419e-01 1.83312281e-03
5.56685388e-01 -1.22026908e+00 -4.27175939e-01 2.99496621e-01
8.78457427e-01 -9.91167307e-01 5.10945380e-01 3.56787771e-01
-8.86265278e-01 -1.39727339e-01 -3.69285822e-01 -7.66756292e-03
2.92718075e-02 1.29516983e+00 -9.72448468e-01 2.96184629e-01
1.79671526e-01 6.89788818e-01 -1.08320840e-01 1.08897042e+00
6.46342188e-02 7.57999361e-01 -4.74205285e-01 -1.08590059e-01
-2.07932353e-01 -1.54296428e-01 4.31927592e-01 1.05566776e+00
3.35523456e-01 7.59047031e-01 2.58456528e-01 5.70909798e-01
4.66602528e-03 3.19705814e-01 -7.13164151e-01 -2.37335190e-01
5.58182001e-01 6.85840845e-01 -5.18418968e-01 -1.11900598e-01
-3.10494810e-01 6.96754038e-01 2.21147090e-01 3.47291082e-01
-8.09411287e-01 -3.19732688e-02 5.90110362e-01 1.47707462e-01
2.54176348e-01 1.13999486e-01 -3.64775121e-01 -1.09023821e+00
-4.24392790e-01 -8.21846068e-01 7.90847301e-01 1.06278710e-01
-1.49252987e+00 1.22595794e-01 1.41340569e-01 -7.35173047e-01
-3.98858756e-01 -3.01847816e-01 -2.25003421e-01 9.13045347e-01
-1.11120093e+00 -8.06942582e-01 4.55694854e-01 4.52200025e-01
3.78089994e-01 -3.16239186e-02 8.85001063e-01 2.97406822e-01
-7.14560151e-01 8.52880836e-01 3.79677564e-01 -6.06720090e-01
5.03872156e-01 -8.18734705e-01 -9.88807455e-02 3.77977252e-01
-3.12149882e-01 9.27104235e-01 9.72064376e-01 -1.11884511e+00
-1.62800920e+00 -9.90677357e-01 1.46805584e+00 -2.36900836e-01
4.75204110e-01 -4.41763550e-01 -6.84123695e-01 6.50044024e-01
-4.76866424e-01 -1.17137499e-01 1.24796820e+00 6.96935534e-01
-3.88874114e-02 1.54910982e-01 -1.29045701e+00 7.01078415e-01
1.17119730e+00 -9.45820585e-02 9.23810080e-02 7.93736935e-01
3.61243397e-01 -1.34741785e-02 -8.91008675e-01 7.96864569e-01
5.14935970e-01 -3.72101188e-01 1.09719872e+00 -9.58659232e-01
4.77248102e-01 -6.23003878e-02 -1.92997903e-01 -1.01453245e+00
-5.80656230e-01 -5.61824620e-01 -1.72610030e-01 7.08358645e-01
6.72939003e-01 -6.86459720e-01 7.48017192e-01 8.11568499e-01
3.68212998e-01 -8.36267531e-01 -1.11208630e+00 -4.00174975e-01
-6.94848597e-02 3.27022448e-02 4.84354943e-02 1.06749690e+00
3.04070085e-01 4.86213058e-01 -8.60648453e-01 2.82751560e-01
8.38370085e-01 2.75915295e-01 4.97950494e-01 -1.41768169e+00
-9.34213400e-01 -9.93222557e-03 -3.53331715e-01 -8.39195192e-01
3.97156447e-01 -5.99917054e-01 -4.80626449e-02 -9.68388736e-01
1.18160403e+00 -4.92179185e-01 -3.83315802e-01 4.83258665e-01
-6.09763145e-01 -6.39891177e-02 -5.21990955e-01 -6.92630606e-03
-4.48961556e-01 5.25625944e-01 8.24319601e-01 -3.27431522e-02
-5.05673409e-01 5.05326211e-01 -7.57809639e-01 4.07797962e-01
5.66906989e-01 -5.18576205e-01 -5.52845359e-01 3.66608560e-01
2.12414250e-01 5.45709729e-01 1.85846642e-01 -1.98839515e-01
1.78136248e-02 -7.00544894e-01 4.50708032e-01 -7.80507624e-01
1.36611834e-01 -6.31099522e-01 6.37184799e-01 6.81950271e-01
-6.24944031e-01 -3.27852190e-01 2.15484932e-01 6.72570765e-01
-1.55130848e-01 -2.04821415e-02 5.93234241e-01 1.27085343e-01
-6.99013174e-02 4.98171836e-01 -2.50533074e-01 -1.09923586e-01
9.14035201e-01 1.99555755e-01 -5.80697022e-02 -3.20039064e-01
-8.88988853e-01 1.98842019e-01 -6.81169704e-02 -1.34600505e-01
7.63067484e-01 -9.97660160e-01 -8.98641169e-01 1.90154314e-01
1.00368157e-01 -4.97063249e-01 3.67759585e-01 1.00261045e+00
1.38309300e-01 6.72395170e-01 2.39922032e-01 -5.61001539e-01
-1.86123478e+00 3.20597380e-01 -2.38367096e-01 -4.71441239e-01
-1.30550325e-01 1.01995909e+00 6.75201237e-01 -1.73216283e-01
7.53680095e-02 1.45156891e-03 -1.72366649e-01 -9.09321532e-02
6.19874895e-01 3.68911982e-01 -9.09854248e-02 -3.14387381e-01
-8.20981026e-01 3.28189760e-01 -5.19254684e-01 2.07276464e-01
1.58433938e+00 3.15047801e-01 -6.57750070e-01 3.50348741e-01
1.33265197e+00 5.16000763e-02 -8.24088991e-01 -4.65430170e-02
-9.54068750e-02 -6.26615465e-01 -4.45466004e-02 -7.32346892e-01
-5.98226130e-01 3.60125661e-01 3.40133965e-01 -3.01365316e-01
9.57389772e-01 1.70530647e-01 2.81900596e-02 4.41778928e-01
4.66126919e-01 -6.33075953e-01 -4.84762102e-01 5.19176722e-01
7.77144134e-01 -1.18389642e+00 3.72654647e-01 -1.07822871e+00
-2.99405843e-01 6.14771426e-01 3.14744473e-01 4.92565744e-02
9.42211688e-01 2.08142295e-01 -9.18885112e-01 -3.65866154e-01
-1.11501229e+00 4.65611219e-02 3.54531020e-01 3.10521543e-01
9.92212415e-01 3.33020121e-01 -1.13407266e+00 6.42626524e-01
2.70779699e-01 2.69021392e-01 1.15207985e-01 7.10903883e-01
-2.51103371e-01 -1.40264475e+00 -6.37921765e-02 1.19836175e+00
-7.09742010e-01 -6.14263415e-01 -5.70671916e-01 4.09338027e-01
-2.56491810e-01 1.10451233e+00 -4.59092446e-02 -1.42243639e-01
1.61639914e-01 -3.96082029e-02 4.91733670e-01 -5.75756192e-01
-4.39279765e-01 5.73117971e-01 2.12242246e-01 -3.86076629e-01
-1.81075364e-01 -9.64259267e-01 -1.22560835e+00 -4.05506402e-01
-4.19258565e-01 4.16296184e-01 1.00091770e-01 8.62755656e-01
7.63788998e-01 2.99792409e-01 8.58318925e-01 -2.86375463e-01
-4.43879902e-01 -7.11566329e-01 -8.14632833e-01 8.90775397e-02
2.13382289e-01 -8.91621888e-01 -2.47796655e-01 -2.67432868e-01] | [7.658432483673096, 4.959460258483887] |
f6386680-9220-4c70-9f0c-1e5203faf478 | a-neural-divide-and-conquer-reasoning | 2305.02265 | null | https://arxiv.org/abs/2305.02265v2 | https://arxiv.org/pdf/2305.02265v2.pdf | A Neural Divide-and-Conquer Reasoning Framework for Image Retrieval from Linguistically Complex Text | Pretrained Vision-Language Models (VLMs) have achieved remarkable performance in image retrieval from text. However, their performance drops drastically when confronted with linguistically complex texts that they struggle to comprehend. Inspired by the Divide-and-Conquer algorithm and dual-process theory, in this paper, we regard linguistically complex texts as compound proposition texts composed of multiple simple proposition sentences and propose an end-to-end Neural Divide-and-Conquer Reasoning framework, dubbed NDCR. It contains three main components: 1) Divide: a proposition generator divides the compound proposition text into simple proposition sentences and produces their corresponding representations, 2) Conquer: a pretrained VLMs-based visual-linguistic interactor achieves the interaction between decomposed proposition sentences and images, 3) Combine: a neural-symbolic reasoner combines the above reasoning states to obtain the final solution via a neural logic reasoning approach. According to the dual-process theory, the visual-linguistic interactor and neural-symbolic reasoner could be regarded as analogical reasoning System 1 and logical reasoning System 2. We conduct extensive experiments on a challenging image retrieval from contextual descriptions data set. Experimental results and analyses indicate NDCR significantly improves performance in the complex image-text reasoning problem. Code link: https://github.com/YunxinLi/NDCR. | ['Yuxin Ding', 'Min Zhang', 'Lin Ma', 'Baotian Hu', 'Yunxin Li'] | 2023-05-03 | null | null | null | null | ['logical-reasoning'] | ['reasoning'] | [ 1.29553765e-01 1.45159587e-01 4.83526774e-02 -3.71624112e-01
-6.20088398e-01 -4.51003909e-01 1.07521212e+00 -1.40012987e-02
-3.52236181e-01 2.28020921e-01 1.40424743e-01 -5.29645026e-01
-1.74490064e-01 -8.23256075e-01 -7.57703424e-01 -4.24125969e-01
5.97616315e-01 7.07430780e-01 1.46133989e-01 -1.94931656e-01
3.12370002e-01 9.69030187e-02 -1.45137286e+00 7.85424232e-01
8.73503625e-01 1.04173148e+00 4.30521607e-01 6.44138336e-01
-6.75190389e-01 1.57228196e+00 -3.47152472e-01 -7.59395123e-01
6.51020482e-02 -7.85013080e-01 -1.02725005e+00 2.20835537e-01
4.30130243e-01 -3.61424416e-01 -2.84090251e-01 1.54358459e+00
9.44138765e-02 -1.00051977e-01 8.19108188e-01 -1.25138307e+00
-1.51146829e+00 9.13419604e-01 -6.09421849e-01 9.86577347e-02
5.45186281e-01 1.77729502e-01 1.06038082e+00 -1.11506045e+00
5.11368394e-01 1.77878356e+00 3.88293415e-01 5.53209364e-01
-1.23278940e+00 -3.45042855e-01 3.58717144e-01 5.99408627e-01
-1.51557803e+00 -2.27955505e-01 7.55287945e-01 -4.17567223e-01
1.13603461e+00 9.46342424e-02 6.50469840e-01 8.04813266e-01
3.18845689e-01 1.34174550e+00 9.78036523e-01 -3.28926504e-01
1.67621579e-02 1.85402110e-01 2.47061729e-01 1.07068050e+00
2.16203649e-02 -2.85794288e-01 -2.32039973e-01 2.04579398e-01
5.79734504e-01 6.27785176e-02 -2.56902009e-01 -3.13600115e-02
-1.23927474e+00 7.48421133e-01 7.75684416e-01 3.83192241e-01
-6.87484562e-01 2.40848586e-01 3.90699089e-01 4.39614832e-01
-8.24147761e-02 9.99889299e-02 5.62212542e-02 5.74831784e-01
-1.11357856e+00 2.52208501e-01 6.68596327e-01 7.92027831e-01
5.50623536e-01 -2.42104866e-02 -2.46229067e-01 7.66504288e-01
7.00906813e-01 8.41870010e-01 4.97314692e-01 -1.01683569e+00
4.67060566e-01 6.53883874e-01 -4.04985070e-01 -1.23758805e+00
-1.66592732e-01 -7.01218769e-02 -9.04653907e-01 8.11912399e-03
1.04713619e-01 2.19718799e-01 -5.57081997e-01 1.36105943e+00
-1.67545855e-01 -2.42567495e-01 7.29867578e-01 1.12192154e+00
1.46509421e+00 8.64840209e-01 2.67702699e-01 -2.40161344e-01
1.86791742e+00 -1.28896332e+00 -8.20810318e-01 -3.92778486e-01
1.39844557e-02 -8.15301716e-01 1.22861874e+00 4.45921332e-01
-1.56444061e+00 -7.65140295e-01 -9.27313328e-01 -6.03428543e-01
-4.21935767e-01 4.06321079e-01 3.67857426e-01 -1.57403469e-01
-1.24388707e+00 1.08192256e-03 -3.52266014e-01 -2.04877794e-01
7.55780578e-01 -3.59551944e-02 -1.36952072e-01 -2.61872172e-01
-1.14194989e+00 9.23041403e-01 7.26740539e-01 4.02272493e-01
-7.89669216e-01 -1.26781046e-01 -7.64900208e-01 2.46999323e-01
8.26069832e-01 -1.27945483e+00 1.29351139e+00 -1.30776048e+00
-1.20524538e+00 1.18772888e+00 -2.24481627e-01 -5.43754995e-01
4.63299543e-01 -1.94966882e-01 -2.28929713e-01 6.86494946e-01
3.18726480e-01 9.16552246e-01 9.83502865e-01 -1.49906230e+00
-5.63868046e-01 -3.67780119e-01 2.52186149e-01 4.69430447e-01
2.67107725e-01 -1.24644060e-02 -8.00895512e-01 -4.16044772e-01
3.88425231e-01 -4.46276069e-01 7.24838749e-02 1.50854990e-01
-4.64668483e-01 -6.29136562e-01 6.24424756e-01 -6.15094662e-01
7.84101903e-01 -1.92220724e+00 4.25748169e-01 -1.76463649e-01
5.00561535e-01 1.16744384e-01 -2.63541102e-01 2.55728424e-01
1.63597316e-02 8.01168978e-02 8.18078145e-02 -1.36070490e-01
1.70360744e-01 3.74772131e-01 -8.97573769e-01 4.64279689e-02
3.99373084e-01 1.35523558e+00 -1.01927471e+00 -9.87384379e-01
1.40094653e-01 2.53042042e-01 -2.90649801e-01 -3.59149016e-02
-7.02484131e-01 -7.84651339e-02 -4.75247175e-01 7.05543578e-01
5.68297505e-01 -2.50654221e-01 8.62596110e-02 -4.97671783e-01
1.26154736e-01 -1.56634122e-01 -7.10892022e-01 1.57102752e+00
-1.06861092e-01 8.11729670e-01 -1.01413026e-01 -1.13124907e+00
8.82543385e-01 9.28772837e-02 -1.89067051e-01 -9.50669944e-01
4.14655089e-01 7.37096071e-02 -4.15441170e-02 -9.49430704e-01
2.97807992e-01 -5.40436268e-01 -8.13230276e-02 2.22983584e-01
-1.15753794e-02 -4.74209458e-01 3.82477134e-01 5.75922728e-01
6.25916421e-01 3.00501525e-01 3.52929503e-01 -7.95481130e-02
1.23613608e+00 2.39833459e-01 1.05802096e-01 7.85781801e-01
-2.54683435e-01 2.02336967e-01 9.15362895e-01 -3.81030172e-01
-7.38069355e-01 -1.45999491e+00 3.01443249e-01 8.84649217e-01
4.49992806e-01 -3.58124226e-01 -6.32029533e-01 -4.39061940e-01
-1.41959190e-01 1.05832505e+00 -2.69711822e-01 2.45351195e-02
-2.50806272e-01 -1.43454194e-01 7.53144324e-01 4.88322824e-01
1.15154505e+00 -1.39441550e+00 -6.78199649e-01 -1.15042791e-01
-4.39663231e-01 -1.41964471e+00 -8.04652125e-02 -2.55598515e-01
-4.47704524e-01 -1.13492835e+00 -4.00464177e-01 -1.07178211e+00
7.79006839e-01 3.11059922e-01 1.02439880e+00 2.74681747e-01
-2.14563549e-01 6.82758212e-01 -9.69621465e-02 -3.37652773e-01
-5.05898714e-01 -8.14485908e-01 -2.47787818e-01 -7.91874751e-02
4.92106229e-01 -1.85239479e-01 -3.37262511e-01 -1.82342246e-01
-9.93359029e-01 5.61439812e-01 8.19724441e-01 9.16324615e-01
8.86187077e-01 5.27107000e-01 1.07640862e-01 -3.06151211e-01
1.13978195e+00 -1.16800331e-01 -6.32205188e-01 7.31497645e-01
-3.44735980e-01 3.74955177e-01 8.55892301e-01 -4.96334970e-01
-1.32308364e+00 -2.75913775e-01 5.61956204e-02 -7.64189541e-01
-8.62229392e-02 6.63539588e-01 -2.20533963e-02 4.78760391e-01
4.45271820e-01 8.62436533e-01 1.42144248e-01 1.40061453e-01
7.42113590e-01 4.35417175e-01 9.86485541e-01 -7.79154897e-01
7.66975701e-01 6.14534557e-01 6.67253286e-02 -7.72114694e-01
-1.11531317e+00 -1.57325249e-02 -4.90623385e-01 -3.88580829e-01
1.35417092e+00 -1.06467676e+00 -1.21174598e+00 1.98225781e-01
-1.59111559e+00 -2.34111071e-01 -1.09678330e-02 2.02511519e-01
-7.52568066e-01 6.12006962e-01 -6.90148354e-01 -7.86459088e-01
-6.52133048e-01 -1.20168805e+00 9.47861195e-01 4.24119473e-01
9.84849967e-03 -8.68279576e-01 -4.99258965e-01 7.58915842e-01
1.37593493e-01 1.03471078e-01 1.20300150e+00 -4.94153321e-01
-6.62514091e-01 -1.02450214e-02 -7.61038840e-01 4.77602661e-01
-4.02041823e-01 -1.19413994e-02 -5.91916561e-01 2.01908767e-01
2.86259472e-01 -5.71340203e-01 9.66823280e-01 1.67742833e-01
1.05996072e+00 -4.11411732e-01 -4.96068075e-02 2.37322122e-01
1.58165944e+00 1.63636923e-01 6.25917196e-01 1.67247370e-01
6.55675769e-01 5.96245527e-01 4.39915687e-01 -1.98381424e-01
8.60755563e-01 1.70978203e-01 2.35445410e-01 8.05087611e-02
-3.85305792e-01 -3.59107077e-01 4.48489457e-01 7.68035531e-01
1.66763067e-02 -4.43130955e-02 -1.25919032e+00 2.12122425e-01
-2.21883678e+00 -1.47073305e+00 -3.02830458e-01 1.32527828e+00
8.06072772e-01 2.63236076e-01 -3.13697249e-01 -1.03609972e-01
4.02532488e-01 7.13054091e-02 -5.17166913e-01 -5.44754922e-01
-3.47992122e-01 -2.93136626e-01 -1.58401251e-01 5.88900626e-01
-8.44376683e-01 1.47534144e+00 4.76518059e+00 9.03719664e-01
-1.04065788e+00 5.64829707e-02 3.66249830e-01 1.00316428e-01
-2.38081768e-01 9.36628878e-02 -5.69965959e-01 -5.36346212e-02
4.05835599e-01 -2.82293379e-01 7.86277294e-01 7.21204400e-01
5.09651564e-02 -4.75973845e-01 -1.14909887e+00 1.43027842e+00
6.31519675e-01 -1.47486854e+00 7.96496928e-01 -4.20076698e-01
3.04196805e-01 -1.32847846e-01 2.32996400e-02 5.63929319e-01
1.34654984e-01 -9.30043399e-01 1.16066039e+00 9.58392560e-01
3.25939626e-01 -4.12188709e-01 4.47050273e-01 6.84256554e-01
-1.21768105e+00 -1.75364479e-01 -4.26180780e-01 -1.12877160e-01
8.79147723e-02 1.82046667e-01 -6.03494406e-01 6.57185912e-01
5.24940670e-01 5.81123650e-01 -6.88461840e-01 2.95567989e-01
-7.27469385e-01 -2.41033118e-02 1.47945032e-01 -2.85675377e-01
5.16023219e-01 -2.89445817e-01 4.50091064e-01 1.00914896e+00
-2.58034617e-01 2.97757536e-01 7.76430592e-02 1.62370503e+00
6.55306652e-02 -3.54033187e-02 -5.04713356e-01 -1.73562601e-01
2.41946876e-01 1.14703894e+00 -9.85875368e-01 -7.34222412e-01
-4.96361375e-01 1.16301155e+00 4.04572397e-01 6.03308022e-01
-9.40785050e-01 -2.69352823e-01 -6.78854715e-03 -3.94830495e-01
3.18746090e-01 -1.42369002e-01 -2.13011906e-01 -1.31687450e+00
9.20684263e-02 -9.89162445e-01 3.89398038e-01 -1.64057624e+00
-1.35936785e+00 7.34645069e-01 2.21702069e-01 -7.96819568e-01
-2.62616407e-02 -8.85713220e-01 -5.07848918e-01 7.73710132e-01
-1.61754060e+00 -1.37586927e+00 -3.28898132e-01 8.30843866e-01
9.63444769e-01 -2.46839836e-01 6.49944961e-01 -1.83743268e-01
-4.22690272e-01 -1.73838083e-02 -5.16479254e-01 3.56374353e-01
1.27617732e-01 -1.06385040e+00 -1.33975744e-01 8.48768532e-01
4.09932077e-01 8.10001671e-01 5.61883271e-01 -5.77081680e-01
-1.68049216e+00 -8.27396274e-01 8.23274076e-01 -2.90306926e-01
9.98400331e-01 -1.68096200e-01 -9.56136048e-01 6.67010725e-01
6.44525707e-01 -3.73734653e-01 3.87434572e-01 -4.19127017e-01
-6.37473881e-01 -2.46772487e-02 -8.94909203e-01 1.14146650e+00
7.88241982e-01 -9.36273575e-01 -1.41173911e+00 5.33766627e-01
6.12404704e-01 -2.69594193e-01 -4.01801229e-01 -1.93427852e-03
4.47803497e-01 -8.58519137e-01 1.05516946e+00 -4.06809777e-01
9.74441588e-01 -6.99539244e-01 -3.83918345e-01 -6.21668160e-01
-1.51970074e-01 -3.26424800e-02 8.63854587e-02 1.00808084e+00
1.70431510e-01 -4.42691535e-01 4.52055968e-02 3.97878408e-01
1.91597432e-01 -8.12653065e-01 -5.82370579e-01 -6.29964888e-01
6.22952916e-02 -6.01784229e-01 2.51472235e-01 6.84066176e-01
-1.68398563e-02 9.62088883e-01 1.17581964e-01 1.18105121e-01
7.08654165e-01 3.93428802e-01 4.77132380e-01 -9.81244564e-01
-3.39801550e-01 -1.09102046e+00 -1.49162680e-01 -1.13440013e+00
6.32475138e-01 -1.28542447e+00 1.98907390e-01 -2.16277695e+00
4.74621236e-01 3.46793383e-01 2.19912410e-01 6.97992980e-01
-1.08958356e-01 -6.93726465e-02 6.75403297e-01 4.82629567e-01
-9.67408359e-01 5.38036823e-01 1.43380034e+00 -5.38082004e-01
2.10613817e-01 -5.77834904e-01 -7.58919179e-01 1.01470220e+00
5.25703192e-01 -1.04948789e-01 -6.12876594e-01 -6.51114821e-01
5.67701399e-01 3.73232633e-01 8.86512756e-01 -7.35065401e-01
5.68839967e-01 -1.60529450e-01 4.39900875e-01 -9.43510592e-01
2.33435720e-01 -8.73642027e-01 -1.37053177e-01 7.03540862e-01
-4.99319851e-01 8.48560035e-02 1.68394431e-01 4.25788671e-01
-3.85821611e-01 -3.11370313e-01 5.26543617e-01 -4.35312122e-01
-1.01821232e+00 -1.61466822e-01 -3.69789928e-01 -1.70657970e-02
9.91168320e-01 -1.60059497e-01 -7.33749866e-01 -3.23566079e-01
-7.48775840e-01 6.74488664e-01 -8.37856010e-02 4.59351927e-01
1.07524097e+00 -1.28089511e+00 -6.72508538e-01 -1.00428179e-01
-2.65061278e-02 4.36774939e-02 3.23474795e-01 1.08101583e+00
-7.36024976e-01 5.57701051e-01 -1.38347164e-01 -7.14521050e-01
-1.37430322e+00 9.15052354e-01 4.81367171e-01 -2.06607115e-02
-6.75966918e-01 8.10220778e-01 3.36923510e-01 -1.32974595e-01
2.04693154e-01 -3.61374825e-01 -3.04939479e-01 1.95160702e-01
5.95828116e-01 -1.85040429e-01 -7.61326671e-01 -7.96564877e-01
-3.97244006e-01 7.53575027e-01 8.19756091e-02 -1.01252727e-01
1.05332577e+00 -2.23359987e-01 -7.18126714e-01 5.26109636e-01
1.03476703e+00 -5.27908683e-01 -7.38682330e-01 -5.03536880e-01
-8.96173716e-03 1.96734238e-02 6.12591812e-03 -6.98990166e-01
-7.88889885e-01 9.33932662e-01 1.18081577e-01 3.28987151e-01
1.30662858e+00 4.52989638e-01 5.26730061e-01 8.09472501e-01
-2.06990112e-02 -9.21098769e-01 5.16822815e-01 7.22119033e-01
1.32749045e+00 -1.20229506e+00 1.13720872e-01 -2.34098241e-01
-1.03583348e+00 1.42640781e+00 5.38568676e-01 -2.57792026e-01
3.34680259e-01 -6.33864105e-02 -1.05887223e-02 -6.32652879e-01
-1.01118731e+00 -3.64087164e-01 4.92965758e-01 3.10983837e-01
1.13407895e-01 6.07310012e-02 -3.78551215e-01 8.93173158e-01
-2.13579923e-01 6.94893375e-02 2.65870720e-01 6.26862049e-01
-3.87119114e-01 -4.50974077e-01 -5.35322368e-01 3.20163816e-02
5.57753816e-02 -3.41193736e-01 -4.86252010e-01 7.09384799e-01
2.63490617e-01 9.41814721e-01 5.50490543e-02 -1.08311035e-01
2.75086582e-01 2.53189594e-01 6.88446581e-01 -2.50273526e-01
-3.44480783e-01 1.13075688e-01 -8.70905295e-02 -4.79135185e-01
-6.40890956e-01 -4.53634679e-01 -1.96507025e+00 -2.11451247e-01
6.53843731e-02 -1.58553019e-01 5.64447284e-01 1.33215642e+00
-1.26022659e-02 7.75901794e-01 2.07823236e-02 -5.23031950e-01
-5.76977432e-01 -5.05065978e-01 -5.33831157e-02 3.92430961e-01
2.21513867e-01 -3.36770982e-01 -1.49129808e-01 3.94451410e-01] | [10.730905532836914, 1.7294589281082153] |
abe00e8c-04bf-4c2d-8edd-ffac05b7218f | fra-rir-fast-random-approximation-of-the | 2208.04101 | null | https://arxiv.org/abs/2208.04101v1 | https://arxiv.org/pdf/2208.04101v1.pdf | FRA-RIR: Fast Random Approximation of the Image-source Method | The training of modern speech processing systems often requires a large amount of simulated room impulse response (RIR) data in order to allow the systems to generalize well in real-world, reverberant environments. However, simulating realistic RIR data typically requires accurate physical modeling, and the acceleration of such simulation process typically requires certain computational platforms such as a graphics processing unit (GPU). In this paper, we propose FRA-RIR, a fast random approximation method of the widely-used image-source method (ISM), to efficiently generate realistic RIR data without specific computational devices. FRA-RIR replaces the physical simulation in the standard ISM by a series of random approximations, which significantly speeds up the simulation process and enables its application in on-the-fly data generation pipelines. Experiments show that FRA-RIR can not only be significantly faster than other existing ISM-based RIR simulation tools on standard computational platforms, but also improves the performance of speech denoising systems evaluated on real-world RIR when trained with simulated RIR. A Python implementation of FRA-RIR is available online\footnote{\url{https://github.com/yluo42/FRA-RIR}}. | ['Jianwei Yu', 'Yi Luo'] | 2022-08-08 | null | null | null | null | ['room-impulse-response', 'speech-denoising'] | ['audio', 'speech'] | [-2.16453131e-02 -6.10933602e-01 1.05368745e+00 -3.37942392e-01
-1.16132045e+00 -3.90557677e-01 3.46692026e-01 -3.03227305e-01
-3.12564403e-01 2.92850107e-01 2.29941070e-01 -7.41910517e-01
3.51578027e-01 -8.12262416e-01 -7.58265078e-01 -6.46285295e-01
6.42556027e-02 1.63603753e-01 1.42108887e-01 -4.27359134e-01
-1.16483420e-01 3.93959045e-01 -1.83901250e+00 2.74350107e-01
7.90833950e-01 6.55059457e-01 7.25338101e-01 1.28589797e+00
2.13137686e-01 6.60402060e-01 -9.81670797e-01 2.44670376e-01
2.24238336e-01 -4.97542411e-01 -1.53142363e-01 -3.88213217e-01
1.78197697e-01 -4.92615014e-01 -3.86402577e-01 8.84699702e-01
1.10729790e+00 6.72766507e-01 1.30964324e-01 -6.57818377e-01
-3.78386766e-01 2.70306110e-01 -2.87297547e-01 2.59233087e-01
5.53216934e-01 3.74688834e-01 3.31783146e-01 -1.00795913e+00
2.37580180e-01 1.24531269e+00 7.94920802e-01 7.26966381e-01
-1.05358028e+00 -8.89663458e-01 -3.27968866e-01 -1.87925890e-01
-1.52029634e+00 -7.33987570e-01 6.28408551e-01 -5.63246273e-02
1.11739719e+00 5.95317125e-01 5.67881942e-01 1.25726807e+00
-1.54929295e-01 4.67816204e-01 1.19099581e+00 -4.31574404e-01
5.16039312e-01 -3.17772448e-01 8.96069631e-02 4.86379147e-01
-1.44964263e-01 3.48593503e-01 -5.72777569e-01 -2.26651579e-01
9.98311639e-01 -3.56563687e-01 -6.87302589e-01 7.02686131e-01
-1.10412860e+00 4.01317924e-01 3.77083927e-01 1.69831052e-01
-1.74348086e-01 3.15185726e-01 2.91001379e-01 3.92023385e-01
5.50737143e-01 1.17270827e-01 -2.28539303e-01 -3.37475091e-01
-1.07275879e+00 3.32231760e-01 7.46418059e-01 9.00061607e-01
5.47193348e-01 6.04348540e-01 -3.29418182e-02 1.23127854e+00
5.65292478e-01 9.33971345e-01 4.86795962e-01 -1.11923122e+00
3.76555294e-01 -3.07878464e-01 8.29987004e-02 -5.19546270e-01
-2.15445951e-01 -4.68018919e-01 -1.05254197e+00 3.96079242e-01
2.35168412e-01 -2.22428575e-01 -1.04555643e+00 1.31344032e+00
4.04609740e-01 6.35786295e-01 2.53589243e-01 1.05166698e+00
1.21655250e+00 1.38642311e+00 -2.63496697e-01 -8.50016847e-02
1.08851635e+00 -8.36781800e-01 -5.87092996e-01 -3.17278266e-01
4.59100395e-01 -1.26511264e+00 1.49175417e+00 5.37859321e-01
-1.12768853e+00 -7.47187853e-01 -1.01040542e+00 -1.13683313e-01
2.51407474e-01 -8.35230947e-03 4.09587890e-01 8.18459451e-01
-1.44455218e+00 4.21286970e-01 -9.36316490e-01 3.97370718e-02
-1.81797240e-02 5.65630905e-02 -9.53769833e-02 -3.78023803e-01
-1.11540651e+00 3.72532040e-01 -3.45808774e-01 5.45232058e-01
-1.06997812e+00 -9.94084418e-01 -9.68642175e-01 -2.42320914e-02
-3.56137380e-02 -7.14028895e-01 1.72810566e+00 -7.07571805e-01
-1.98461568e+00 2.88237542e-01 -3.81296009e-01 -1.82234839e-01
3.73590499e-01 -3.80041987e-01 -5.88015378e-01 -1.18367463e-01
-4.23853278e-01 2.51088142e-01 9.51434672e-01 -1.56668115e+00
1.06040604e-01 6.27514198e-02 -2.25785360e-01 2.17905149e-01
7.89288208e-02 2.74385810e-01 -4.75008070e-01 -8.04948986e-01
-7.46806897e-03 -8.02920341e-01 -5.01090407e-01 -3.91112238e-01
-1.82654858e-01 1.68297574e-01 6.38226092e-01 -8.44787478e-01
1.05620992e+00 -2.44105077e+00 -6.03017330e-01 2.49152005e-01
-9.63892788e-02 6.87956274e-01 -1.83218077e-01 4.13545281e-01
-1.36138409e-01 -1.85745358e-01 -2.73444146e-01 -6.37468040e-01
-9.54465568e-02 9.49216336e-02 -6.35961056e-01 3.91028196e-01
-3.48837703e-01 3.30042571e-01 -7.14696527e-01 -3.13428640e-02
5.12428641e-01 1.00005102e+00 -7.14662135e-01 5.11654258e-01
-8.68013576e-02 7.73207247e-01 -2.66034782e-01 2.00488046e-01
9.85635459e-01 1.54030755e-01 -2.28521988e-01 -1.89944759e-01
-3.98351490e-01 6.92304552e-01 -1.27124906e+00 1.59475529e+00
-1.20359087e+00 7.08184958e-01 5.84726632e-01 -5.20923734e-01
1.12499332e+00 4.87059802e-01 -9.09273773e-02 -6.92195296e-01
1.75143644e-01 3.73880416e-01 -1.08271874e-01 -1.92046478e-01
5.31623781e-01 1.47928400e-02 2.08316833e-01 5.80803931e-01
-4.52963322e-01 -5.68293333e-01 -2.80530099e-02 1.11160174e-01
1.31962430e+00 6.44956976e-02 -1.53831184e-01 -9.24190581e-02
4.76377904e-01 -1.77377105e-01 4.52247977e-01 9.84773099e-01
1.07466504e-01 1.12925172e+00 -3.53452533e-01 -1.98924169e-01
-1.11606979e+00 -1.32882500e+00 -1.55833378e-01 8.87825191e-01
-1.09147020e-01 -4.41535443e-01 -9.22845304e-01 1.36784986e-01
-5.18185079e-01 1.08911800e+00 4.46180627e-02 1.52567595e-01
-8.22838843e-01 -8.37597072e-01 6.93874598e-01 4.10875559e-01
5.63343763e-01 -1.28174245e+00 -4.71240908e-01 2.96186268e-01
-2.57748663e-01 -1.02231038e+00 -4.35707062e-01 -3.05946148e-03
-6.27164900e-01 -2.65826583e-01 -6.42746329e-01 -5.69038749e-01
6.00010455e-01 7.41496563e-01 1.15040410e+00 5.31461716e-01
-4.54330325e-01 5.08847535e-01 -5.48044145e-01 -3.22717965e-01
-8.98760796e-01 -5.75449824e-01 7.09846541e-02 -3.59339625e-01
-1.00837752e-01 -7.39639640e-01 -8.05496395e-01 3.47154170e-01
-9.77436662e-01 2.76228875e-01 2.91905515e-02 6.31944478e-01
6.44373715e-01 1.16851777e-01 4.38516051e-01 -4.83975261e-01
6.38983548e-01 -7.27676228e-02 -8.00109386e-01 -3.46476659e-02
-8.46725237e-03 -9.75945592e-02 9.98962939e-01 -4.08109605e-01
-1.48159766e+00 -1.88989252e-01 -1.10143244e+00 -4.32265133e-01
-2.67738074e-01 3.11937511e-01 -1.28196314e-01 1.30809605e-01
8.22770357e-01 3.71870786e-01 -1.84592500e-01 -6.02557003e-01
1.73091739e-01 1.01142669e+00 5.88680029e-01 -5.99049807e-01
9.66545761e-01 3.21914673e-01 -4.55041260e-01 -1.25669134e+00
-6.22822165e-01 -4.73921984e-01 1.72026902e-01 -1.57144830e-01
5.57917953e-01 -1.23059607e+00 -5.07580221e-01 8.23061168e-01
-1.26456416e+00 -1.10700607e+00 -9.08276066e-02 7.38301873e-01
-4.05386299e-01 1.79152772e-01 -7.65429556e-01 -8.64686608e-01
-7.29679942e-01 -1.31504679e+00 1.18774676e+00 1.57087713e-01
-3.27716544e-02 -6.17556930e-01 2.06656948e-01 4.53763932e-01
7.54643917e-01 -3.41058552e-01 4.04707223e-01 -4.92889769e-02
-5.90309441e-01 -1.66121006e-01 7.91996941e-02 6.18318856e-01
-7.28606507e-02 2.22637549e-01 -1.37470365e+00 -3.81639093e-01
3.94364357e-01 -4.65063378e-02 5.99014938e-01 5.91358840e-01
1.36979270e+00 -3.73599567e-02 1.68399334e-01 7.85062253e-01
1.21582162e+00 1.23378359e-01 1.07572877e+00 8.90480205e-02
5.37427843e-01 1.19815938e-01 5.54703474e-01 6.83105230e-01
7.07681403e-02 6.80370867e-01 1.34228736e-01 -3.52255315e-01
-4.67956841e-01 -2.46748012e-02 7.29984403e-01 1.45845604e+00
-1.12546906e-01 -3.45508635e-01 -9.01613653e-01 1.15940638e-01
-1.24438620e+00 -1.09148216e+00 -3.60931069e-01 2.48040342e+00
9.33868408e-01 -2.31792971e-01 -3.61083180e-01 2.48376220e-01
6.43958390e-01 1.39758185e-01 -1.84040695e-01 -7.16873527e-01
1.18190505e-01 6.58091307e-01 2.86053538e-01 8.00262213e-01
-6.87606573e-01 7.08617806e-01 5.84997368e+00 7.21444249e-01
-1.16295016e+00 2.68941551e-01 6.12899661e-01 -1.52551502e-01
-4.21842337e-01 -2.66682565e-01 -7.63099968e-01 2.04414710e-01
1.38813102e+00 -9.34302658e-02 9.11016226e-01 8.75559568e-01
9.36821878e-01 -9.89018232e-02 -6.17327511e-01 1.11613011e+00
-6.68334439e-02 -1.14932275e+00 -3.17036510e-01 -2.12223992e-01
4.23292041e-01 3.25422496e-01 1.97362512e-01 2.62447536e-01
6.20219111e-01 -1.05344808e+00 6.15307331e-01 1.83881849e-01
8.26551139e-01 -7.00963378e-01 4.20182586e-01 3.39291036e-01
-1.28850102e+00 3.12011480e-01 -6.11675680e-01 1.20815719e-02
3.95221323e-01 9.35910881e-01 -9.31381702e-01 3.51701647e-01
9.90482807e-01 1.42411053e-01 -2.81980395e-01 9.94263530e-01
-3.16365898e-01 1.18030024e+00 -5.51140010e-01 2.42267996e-01
-8.62908512e-02 -1.37823999e-01 5.96864641e-01 1.31919014e+00
7.33515799e-01 3.34301382e-01 -1.35279477e-01 5.34555435e-01
2.76256874e-02 3.93618010e-02 -3.80183399e-01 5.19086182e-01
5.57564974e-01 1.35040402e+00 -4.67001289e-01 -2.20184788e-01
-2.53258437e-01 9.55871344e-01 -1.15612544e-01 4.84341353e-01
-1.23846972e+00 -4.26433086e-01 7.50431657e-01 2.06794173e-01
2.33333752e-01 -6.67491734e-01 -3.01565137e-02 -8.69947493e-01
-4.01883423e-02 -1.23546422e+00 -2.11580664e-01 -1.28518093e+00
-9.66260612e-01 1.00565553e+00 -3.23138475e-01 -1.33702028e+00
-1.40782505e-01 -4.08194840e-01 -8.26641619e-01 1.24880576e+00
-1.28641009e+00 -7.53198266e-01 -7.81487286e-01 8.00401151e-01
7.08526909e-01 1.63948223e-01 1.08742559e+00 5.13533533e-01
-6.26206934e-01 4.89005446e-01 3.64366263e-01 -7.15572983e-02
7.43291736e-01 -9.72917855e-01 1.06092334e+00 1.02489746e+00
9.60915238e-02 8.38155448e-01 1.11141622e+00 -4.51862425e-01
-1.58413208e+00 -1.20241559e+00 2.95857787e-01 6.13085590e-02
3.91372830e-01 -6.10559165e-01 -1.19470429e+00 3.11469436e-01
1.76294580e-01 2.52430499e-01 6.94540799e-01 -1.58348799e-01
-4.73149329e-01 -2.81345874e-01 -9.81665671e-01 7.81421781e-01
9.16349947e-01 -4.26535755e-01 -2.57816344e-01 4.25216764e-01
8.18360150e-01 -7.08734393e-01 -6.77948415e-01 1.20128222e-01
2.02366784e-01 -1.06199884e+00 1.38238871e+00 3.35752547e-01
2.53442854e-01 -6.59794211e-01 -2.83279032e-01 -1.51187277e+00
-9.09073558e-03 -8.94915521e-01 9.35556367e-02 1.18269730e+00
2.49265134e-01 -6.25605583e-01 2.74046838e-01 5.40879607e-01
-5.37669659e-01 -2.61410028e-01 -9.65027213e-01 -8.69610727e-01
-2.24354878e-01 -1.12618434e+00 6.34243548e-01 3.59089732e-01
-6.76241338e-01 -1.11877881e-02 -3.04861516e-01 6.90634251e-01
6.60679698e-01 -1.14810802e-01 1.03017521e+00 -5.61574101e-01
-7.77043223e-01 1.26192629e-01 -1.25384063e-01 -1.28906715e+00
-8.39132816e-02 -6.01814687e-01 6.17734432e-01 -1.83279598e+00
-4.94018108e-01 -8.47257972e-01 1.20350324e-01 1.43077850e-01
-1.73782870e-01 3.72585386e-01 8.36488679e-02 1.02652065e-01
-1.61699325e-01 6.27321959e-01 1.23587406e+00 2.24225864e-01
-4.77930307e-01 2.53605336e-01 -2.99219251e-01 7.29264259e-01
8.64675760e-01 -4.69793558e-01 -4.78331357e-01 -7.73175538e-01
-2.18490250e-02 6.72487617e-02 6.39157712e-01 -1.47134638e+00
9.26014110e-02 2.17951149e-01 2.84372807e-01 -5.58948338e-01
6.51789963e-01 -4.86282349e-01 3.10272068e-01 2.13978007e-01
1.06056649e-02 1.97925031e-01 3.54178995e-01 2.59097904e-01
-1.57735288e-01 -2.13323757e-01 8.86290967e-01 -1.84184343e-01
-4.14520442e-01 2.11421214e-02 -7.47948468e-01 1.28454464e-02
4.00784850e-01 3.99993062e-02 -4.37878877e-01 -5.52786410e-01
-4.44839239e-01 -4.35015976e-01 3.77754122e-01 1.61621362e-01
8.39319348e-01 -8.89448941e-01 -7.35623896e-01 2.96004921e-01
-3.83769602e-01 3.83725584e-01 7.20466971e-01 5.29089689e-01
-1.00783479e+00 -1.95880756e-01 4.49748248e-01 -6.12538517e-01
-1.48273969e+00 3.83496463e-01 4.42309439e-01 -4.50019352e-02
-1.07477510e+00 9.18541968e-01 3.43909323e-01 -5.94323635e-01
1.26916645e-02 -2.69417107e-01 3.24905157e-01 -7.06595242e-01
1.01086473e+00 2.58550048e-01 5.06637871e-01 -5.46882927e-01
-1.62479103e-01 2.35283941e-01 1.83670223e-01 -5.07481277e-01
1.38215947e+00 -7.10815340e-02 3.59588712e-02 3.06688458e-01
1.10818684e+00 3.09085399e-01 -1.23996019e+00 -9.77618154e-03
-6.93202853e-01 -5.85986197e-01 4.20927525e-01 -6.85500145e-01
-9.29761350e-01 9.00023043e-01 7.22133160e-01 -1.72239378e-01
1.50824165e+00 -2.38148674e-01 1.06157446e+00 4.26108003e-01
5.61933339e-01 -7.23484814e-01 -1.35334432e-02 6.85740232e-01
1.12088668e+00 -7.06249654e-01 -6.01433590e-02 -5.19614875e-01
-5.28381228e-01 9.37980115e-01 4.27429438e-01 -8.27867389e-02
7.72679150e-01 1.02746093e+00 6.23794675e-01 2.94949263e-01
-5.90120256e-01 1.16399042e-01 -1.61306992e-01 7.04447508e-01
6.84216678e-01 -1.07970104e-01 1.02727309e-01 3.69966865e-01
-6.37928545e-01 -9.09265131e-02 6.37907207e-01 8.89108658e-01
-4.04607862e-01 -1.17291582e+00 -9.07390952e-01 1.23099305e-01
-3.36456299e-01 -6.30040884e-01 5.62347889e-01 2.35315159e-01
-2.87720710e-01 1.29458821e+00 4.07278761e-02 -2.49449119e-01
2.95624316e-01 -3.93977970e-01 4.04584289e-01 -5.66768289e-01
-9.79113936e-01 2.94246823e-01 7.38071576e-02 -4.43374574e-01
-2.06325166e-02 -5.10975897e-01 -1.50605583e+00 -5.68901002e-01
-1.35143042e-01 3.26085240e-02 1.02824235e+00 5.61788797e-01
4.25892323e-01 8.64422023e-01 7.87241101e-01 -1.19811690e+00
-4.40260530e-01 -8.99261892e-01 -2.64670879e-01 1.17032669e-01
1.73838302e-01 -2.48841047e-01 -5.36658645e-01 7.97856152e-02] | [15.191695213317871, 5.7816243171691895] |
22148177-5cf3-415b-bb5e-07c3eeae9fb5 | human-pose-estimation-on-privacy-preserving | 2007.08340 | null | https://arxiv.org/abs/2007.08340v2 | https://arxiv.org/pdf/2007.08340v2.pdf | Human Pose Estimation on Privacy-Preserving Low-Resolution Depth Images | Human pose estimation (HPE) is a key building block for developing AI-based context-aware systems inside the operating room (OR). The 24/7 use of images coming from cameras mounted on the OR ceiling can however raise concerns for privacy, even in the case of depth images captured by RGB-D sensors. Being able to solely use low-resolution privacy-preserving images would address these concerns and help scale up the computer-assisted approaches that rely on such data to a larger number of ORs. In this paper, we introduce the problem of HPE on low-resolution depth images and propose an end-to-end solution that integrates a multi-scale super-resolution network with a 2D human pose estimation network. By exploiting intermediate feature-maps generated at different super-resolution, our approach achieves body pose results on low-resolution images (of size 64x48) that are on par with those of an approach trained and tested on full resolution images (of size 640x480). | ['Nicolas Padoy', 'Vinkle Srivastav', 'Afshin Gangi'] | 2020-07-16 | null | null | null | null | ['2d-human-pose-estimation'] | ['computer-vision'] | [ 6.20218337e-01 3.86341900e-01 1.26600757e-01 -5.96656740e-01
-7.97223032e-01 -2.94702291e-01 1.48030251e-01 -2.12664783e-01
-9.55696642e-01 5.85732639e-01 5.18596828e-01 4.00051445e-01
-3.91382724e-03 -6.27100408e-01 -5.71149886e-01 -2.22286761e-01
-1.27687141e-01 3.32641095e-01 2.74812698e-01 -2.78262705e-01
-1.03025861e-01 6.66028142e-01 -1.99800456e+00 5.02453148e-01
3.63656022e-02 1.06828988e+00 -1.35353521e-01 9.09503460e-01
6.21889353e-01 2.90992051e-01 -3.84441257e-01 -3.39578927e-01
6.48683429e-01 -3.96007039e-02 -7.72551298e-01 1.68284088e-01
1.07183051e+00 -7.14939237e-01 -1.76822677e-01 8.07755351e-01
7.09411740e-01 8.92587081e-02 -1.84430301e-01 -1.08262110e+00
2.63939016e-02 -2.64620900e-01 -5.92853069e-01 8.04247335e-02
9.52949941e-01 -5.55477105e-02 2.71761090e-01 -4.17968482e-01
8.90255690e-01 1.36200476e+00 7.63534606e-01 9.82107401e-01
-1.25231671e+00 -5.48262835e-01 -2.54266094e-02 -1.17033310e-01
-1.61053658e+00 -4.66497511e-01 4.63856250e-01 -8.38060528e-02
6.98064744e-01 4.63596553e-01 6.73253775e-01 1.34536040e+00
2.22432166e-01 4.53245968e-01 1.43177903e+00 -2.36765370e-01
7.93890432e-02 2.38560036e-01 -2.69840211e-01 7.36074030e-01
5.60144484e-01 1.78843841e-01 -1.00002325e+00 -1.77226961e-01
1.17451823e+00 2.16732919e-01 -5.31067193e-01 -1.01328957e+00
-1.14334774e+00 5.86929142e-01 4.74746227e-01 7.83800632e-02
-3.76467943e-01 -5.13685942e-02 3.71632487e-01 5.56929410e-02
1.65981799e-01 3.87585670e-01 -6.05673373e-01 3.07858773e-02
-8.47892404e-01 3.79046381e-01 5.48071742e-01 9.79061246e-01
4.79368746e-01 -5.66694736e-01 -1.65301189e-02 3.37596953e-01
2.50501633e-01 3.69505376e-01 3.68104368e-01 -1.12715781e+00
5.66304207e-01 5.06239772e-01 3.10454696e-01 -1.01944709e+00
-7.70010889e-01 -1.05564836e-02 -7.39403129e-01 5.48286080e-01
5.24406254e-01 -1.30391717e-01 -7.55973399e-01 1.52921486e+00
7.78527379e-01 -3.03239882e-01 -5.93235940e-02 1.37226081e+00
5.50874531e-01 1.01543637e-02 -4.82407538e-03 5.99584803e-02
1.84482539e+00 -6.58138812e-01 -7.37307787e-01 -4.87042576e-01
1.64222017e-01 -2.32893899e-01 9.00756180e-01 6.53108120e-01
-7.97815621e-01 -6.45491660e-01 -1.25494766e+00 -5.53542137e-01
-5.22773802e-01 -1.79432899e-01 3.72871131e-01 1.07461989e+00
-1.12489259e+00 6.16162956e-01 -9.67998683e-01 -6.04712784e-01
4.14555609e-01 8.03297520e-01 -1.12108183e+00 -3.18097621e-01
-9.93725896e-01 1.07433879e+00 2.59291291e-01 2.08295628e-01
-5.08556664e-01 -2.53916055e-01 -1.09598827e+00 -4.81505185e-01
6.12876713e-01 -9.05581951e-01 8.15167308e-01 -6.55314744e-01
-1.44323194e+00 1.22341156e+00 -5.47184907e-02 -4.94432002e-01
9.26953256e-01 -8.85097623e-01 -2.06082642e-01 4.99843657e-01
-5.68043515e-02 7.85244405e-01 8.87323439e-01 -1.05033469e+00
-7.52672374e-01 -1.10775435e+00 4.23108488e-02 5.31125844e-01
-3.61065380e-02 9.54150502e-03 -4.17890668e-01 -1.47505790e-01
3.73871401e-02 -1.18919790e+00 -3.65225852e-01 4.11152899e-01
-1.86805263e-01 6.63456380e-01 8.77575099e-01 -9.31766570e-01
7.23125339e-01 -1.96575272e+00 1.73382133e-01 2.55195767e-01
1.62741944e-01 2.82706201e-01 3.51038903e-01 -2.33852372e-01
-5.44736907e-03 -2.44181991e-01 -1.24612011e-01 -5.70769429e-01
-3.35839719e-01 1.91689208e-01 2.54823029e-01 7.74957120e-01
-1.17764607e-01 6.33022606e-01 -5.95462978e-01 -6.04347587e-01
4.91405904e-01 8.74376237e-01 -4.52410579e-01 1.49945453e-01
4.19702828e-01 8.85694265e-01 -2.71041065e-01 4.83896166e-01
7.46175051e-01 8.85419920e-02 2.22309291e-01 -1.29117876e-01
-3.82779688e-02 -1.01845622e-01 -1.41758752e+00 2.38176394e+00
-3.13457966e-01 3.22319001e-01 4.75759447e-01 -1.21636368e-01
6.46362185e-01 3.99908304e-01 2.95474976e-01 -6.07000828e-01
1.84581757e-01 -1.83190525e-01 -4.02720302e-01 -3.69970232e-01
6.98020577e-01 -2.86923170e-01 -4.92469907e-01 4.00496274e-02
-1.11399487e-01 1.43833533e-01 -3.61932367e-01 -3.06845129e-01
1.05125165e+00 4.90368128e-01 3.89826953e-01 1.00921147e-01
6.23718560e-01 -2.92873174e-01 4.02684659e-01 5.08368254e-01
-5.22884846e-01 1.05871499e+00 -6.55819476e-02 -8.40791941e-01
-1.04451799e+00 -9.84487176e-01 -2.09925398e-01 1.05651176e+00
9.32621062e-02 -5.10411203e-01 -8.57542336e-01 -6.50471151e-01
-4.73397039e-02 2.71227539e-01 -9.64626491e-01 -1.20382167e-01
-7.54842162e-01 -3.82591218e-01 4.84679282e-01 5.95826745e-01
8.47228706e-01 -7.63788521e-01 -1.73620105e+00 -3.22006010e-02
-3.07656944e-01 -1.42350781e+00 -2.51724303e-01 9.24844444e-02
-9.07512844e-01 -9.39175129e-01 -8.65348697e-01 -2.72248775e-01
6.65004015e-01 -1.13766082e-01 9.27494764e-01 -5.11298418e-01
-6.43569410e-01 8.32527280e-01 -2.34709486e-01 -3.93405497e-01
1.93748698e-01 2.13033948e-02 2.81413108e-01 -2.82460824e-02
4.51145142e-01 -5.20940781e-01 -8.37193549e-01 2.83931375e-01
-6.08496845e-01 -3.57489400e-02 5.21380305e-01 3.85736376e-01
8.00008178e-01 -2.38214314e-01 -1.43691629e-01 -7.81764209e-01
2.75516748e-01 1.73645467e-01 -5.68042636e-01 6.46072999e-02
-4.58550721e-01 7.41158277e-02 1.84613392e-01 -2.58328438e-01
-9.98142898e-01 7.67838776e-01 -1.79810207e-02 -4.80445296e-01
-5.67021132e-01 -4.19328004e-01 -3.16435069e-01 -1.19342737e-01
6.38744771e-01 -1.00847259e-01 2.20264867e-01 -4.45308626e-01
2.50997961e-01 5.97781241e-01 9.13389027e-01 -2.01227993e-01
6.60400033e-01 8.14150453e-01 3.14383447e-01 -6.66763544e-01
-7.42023945e-01 -5.26566207e-01 -1.10188866e+00 -9.67621878e-02
1.32694161e+00 -1.34937894e+00 -6.71996355e-01 1.20134085e-01
-6.54502153e-01 2.55310327e-01 -3.36260021e-01 4.82855082e-01
-7.53437519e-01 6.48299977e-02 -2.78143078e-01 -9.74946797e-01
-2.93487012e-01 -7.78593123e-01 1.67381203e+00 2.59221077e-01
-5.47619402e-01 -3.61742616e-01 1.47019431e-01 9.15164351e-01
2.75803119e-01 9.22607422e-01 -5.41451797e-02 -1.73618495e-01
-4.15561885e-01 -5.32526731e-01 2.88578738e-02 8.13992172e-02
-4.24723960e-02 -8.82862270e-01 -1.29232979e+00 -4.23314810e-01
1.06319517e-01 -3.84724528e-01 2.73572773e-01 3.52494776e-01
8.91475976e-01 -3.64228100e-01 -3.27787757e-01 6.32887006e-01
1.38749194e+00 -3.24235409e-01 8.32664549e-01 5.30491352e-01
8.31454456e-01 8.42203736e-01 8.15336347e-01 4.80571866e-01
5.36769569e-01 8.16223025e-01 5.41722655e-01 -1.05739214e-01
1.36764243e-01 -2.51844704e-01 2.48696938e-01 -1.98858395e-01
-5.19461632e-01 3.98527503e-01 -6.74890757e-01 3.61269534e-01
-1.69304097e+00 -7.60807991e-01 4.96599153e-02 2.69901395e+00
5.68826973e-01 -5.31171858e-02 4.53039020e-01 2.06807241e-01
6.17163062e-01 4.70666364e-02 -4.38929796e-01 -5.25416672e-01
2.21620247e-01 3.53293598e-01 9.03730989e-01 2.55107522e-01
-1.43799901e+00 4.97333497e-01 5.90882969e+00 -5.28774336e-02
-8.95915687e-01 1.49709672e-01 4.23249334e-01 -7.05589056e-01
4.33063567e-01 -7.14418054e-01 -8.61213267e-01 1.49143517e-01
1.16830873e+00 3.92229110e-01 3.36840034e-01 1.13755548e+00
-1.94896489e-01 -4.48220402e-01 -1.35983145e+00 1.27675986e+00
3.16307843e-01 -7.31269658e-01 -4.40674096e-01 3.60906601e-01
4.06747520e-01 -7.40926117e-02 1.66388988e-01 1.85649805e-02
-2.18163878e-02 -1.13035119e+00 4.86346573e-01 3.84640574e-01
9.84542012e-01 -9.90636170e-01 9.02074575e-01 3.96982044e-01
-1.16818953e+00 -1.75621644e-01 -3.16457599e-01 -5.46382591e-02
6.44212514e-02 -6.19386602e-03 -8.03160012e-01 4.75725114e-01
1.35819793e+00 1.54182449e-01 -7.35499918e-01 5.81782818e-01
-2.82338914e-02 -4.76075977e-01 -7.44491816e-01 3.85930717e-01
-3.68745685e-01 2.98302501e-01 3.95422667e-01 1.08717668e+00
2.36090317e-01 3.36310118e-01 -1.24836937e-01 3.31349134e-01
7.68224373e-02 -1.03049211e-01 -8.57950807e-01 6.20236933e-01
-3.68009061e-02 1.17145681e+00 -5.48515856e-01 -2.06927788e-02
-2.60361373e-01 1.56531239e+00 1.35238484e-01 -2.49861524e-01
-6.21828496e-01 -1.07119620e-01 7.38432944e-01 5.92507720e-01
5.73688328e-01 -2.31666028e-01 -1.72825098e-01 -1.09158432e+00
3.09876859e-01 -1.04916859e+00 6.27171218e-01 -7.39374995e-01
-7.50354469e-01 6.80218935e-01 3.38047981e-01 -1.28091490e+00
-5.34501672e-01 -6.52168512e-01 2.83520252e-01 6.72425807e-01
-1.31257713e+00 -1.30846155e+00 -6.20454133e-01 9.96345580e-01
1.38606265e-01 3.83404464e-01 1.28026211e+00 1.32553190e-01
-1.84964672e-01 6.17616951e-01 -4.57139552e-01 1.68955252e-01
7.42680371e-01 -1.11174500e+00 1.41083151e-01 7.91950941e-01
6.76958188e-02 6.65316522e-01 7.43478775e-01 -5.15715957e-01
-1.39307952e+00 -9.41144764e-01 7.03279436e-01 -1.20413589e+00
-4.05614763e-01 -6.81996882e-01 -4.54271704e-01 8.24016094e-01
-1.48590624e-01 5.08637309e-01 6.74864411e-01 8.86121616e-02
-2.41811335e-01 -1.22156508e-01 -1.88835001e+00 4.15269941e-01
1.23309040e+00 -4.90958244e-01 -7.73424387e-01 -1.02093168e-01
3.89731497e-01 -9.63646114e-01 -1.08710563e+00 4.20366883e-01
9.61236954e-01 -1.36550045e+00 1.31432378e+00 -1.99391976e-01
1.34115979e-01 -4.14213747e-01 -2.88464010e-01 -6.91793978e-01
-9.14139152e-02 -5.11325359e-01 7.37039968e-02 6.83681130e-01
1.01489089e-02 -4.93102252e-01 1.05582058e+00 1.29195547e+00
5.93356669e-01 -3.18481594e-01 -1.28908598e+00 -5.03871739e-01
-4.71333236e-01 -2.78669685e-01 4.31338161e-01 6.02543473e-01
-8.76214877e-02 6.64444342e-02 -6.18152142e-01 5.63926160e-01
8.62491488e-01 -1.02669887e-01 1.09061873e+00 -1.25530720e+00
-5.59648674e-04 4.48245019e-01 -8.86285245e-01 -5.37372112e-01
-3.65099937e-01 -1.31180614e-01 1.07735664e-01 -1.08913100e+00
-5.99381514e-02 6.55301809e-02 -5.61033078e-02 5.48226118e-01
1.66699335e-01 8.26204300e-01 1.85526669e-01 -9.92544219e-02
-6.32352114e-01 1.57186896e-01 9.75204527e-01 5.31979024e-01
-6.35074899e-02 -1.13932438e-01 -6.51757121e-01 1.02979064e+00
5.92125773e-01 -5.56027472e-01 -2.49562293e-01 -1.25230581e-01
1.38026595e-01 7.84825012e-02 6.81870341e-01 -1.56823194e+00
1.61567509e-01 2.19296053e-01 1.06795657e+00 -6.52139306e-01
9.19128537e-01 -1.45394051e+00 4.01825249e-01 4.94983166e-01
-4.30940062e-01 1.16282716e-01 1.93477884e-01 6.62899852e-01
1.66147705e-02 3.15012604e-01 7.15292215e-01 -6.60685897e-01
-5.67663550e-01 2.91794509e-01 -6.65649623e-02 -1.46337673e-01
1.14851153e+00 -6.06169760e-01 1.11176878e-01 -3.62694651e-01
-8.98102701e-01 -6.68113828e-02 8.70071471e-01 6.64267361e-01
8.55014861e-01 -1.13088965e+00 -4.50457424e-01 6.02991998e-01
2.77796865e-01 4.11973238e-01 2.15815514e-01 5.46637058e-01
-4.79398489e-01 4.25489932e-01 -7.09244609e-01 -5.71750224e-01
-1.76877058e+00 5.32874882e-01 2.42175326e-01 -4.35157806e-01
-9.41098332e-01 8.33845139e-01 1.98557109e-01 -5.36597371e-01
3.01118433e-01 -2.95661986e-01 -1.97637677e-01 -7.35235885e-02
8.92374158e-01 2.96143025e-01 5.06688915e-02 -9.78137732e-01
-6.64362192e-01 6.91681206e-01 -4.62286174e-02 -4.29553151e-01
1.33165729e+00 -5.15710413e-01 3.40451121e-01 2.80420840e-01
1.20469928e+00 4.14876752e-02 -1.47763956e+00 -3.17901373e-02
-2.58849382e-01 -9.27088022e-01 5.84775861e-03 -8.01133752e-01
-9.18612301e-01 7.40785182e-01 1.36380553e+00 -4.38146234e-01
1.19811666e+00 -1.56127706e-01 7.36541450e-01 2.63959885e-01
1.02585816e+00 -1.33871865e+00 -1.65429652e-01 -7.53354207e-02
9.80967879e-01 -1.45983744e+00 5.88216305e-01 -3.46992284e-01
-8.21807742e-01 9.34887350e-01 6.28964663e-01 -2.26956248e-01
3.19636583e-01 3.04830998e-01 1.22503810e-01 -2.21136957e-01
-2.96244889e-01 -1.93898976e-01 3.88767749e-01 1.12349021e+00
1.40146643e-01 -7.57680275e-04 2.48185083e-01 3.36829573e-01
-4.59200829e-01 3.14938545e-01 2.87856728e-01 1.43634927e+00
-7.89668784e-02 -1.00830960e+00 -8.58874857e-01 8.30179155e-02
-6.93316638e-01 2.91754037e-01 -4.61090744e-01 9.41944182e-01
4.26404417e-01 8.19052577e-01 9.53834355e-02 -4.31820095e-01
6.99445307e-01 1.03833899e-01 7.48890638e-01 -5.28234065e-01
-8.52537930e-01 -1.10477448e-01 2.78405964e-01 -1.30307972e+00
-6.16621315e-01 -9.48570788e-01 -8.96279216e-01 -1.32067934e-01
8.84112939e-02 -5.41299045e-01 7.32996345e-01 5.21569610e-01
5.53408146e-01 3.40633333e-01 1.59524709e-01 -1.20994651e+00
-1.98090419e-01 -6.20693386e-01 -5.76678932e-01 4.21024352e-01
7.07752705e-01 -4.93957460e-01 8.90962407e-03 9.39637125e-02] | [7.018146991729736, -0.8923249840736389] |
cc583f04-772a-440e-b0fc-d5cdc44f626d | conversational-machine-reading-comprehension | 2105.01542 | null | https://arxiv.org/abs/2105.01542v6 | https://arxiv.org/pdf/2105.01542v6.pdf | Conversational Machine Reading Comprehension for Vietnamese Healthcare Texts | Machine reading comprehension (MRC) is a sub-field in natural language processing that aims to assist computers understand unstructured texts and then answer questions related to them. In practice, the conversation is an essential way to communicate and transfer information. To help machines understand conversation texts, we present UIT-ViCoQA, a new corpus for conversational machine reading comprehension in the Vietnamese language. This corpus consists of 10,000 questions with answers over 2,000 conversations about health news articles. Then, we evaluate several baseline approaches for conversational machine comprehension on the UIT-ViCoQA corpus. The best model obtains an F1 score of 45.27%, which is 30.91 points behind human performance (76.18%), indicating that there is ample room for improvement. Our dataset is available at our website: http://nlp.uit.edu.vn/datasets/ for research purposes. | ['Ngan Luu-Thuy Nguyen', 'Kiet Van Nguyen', 'Khiem Vinh Tran', 'Loi Duc Nguyen', 'Mao Nguyen Bui', 'Son T. Luu'] | 2021-05-04 | null | null | null | null | ['vietnamese-datasets'] | ['natural-language-processing'] | [ 4.21377867e-01 7.44244695e-01 -1.19454851e-02 -3.88153315e-01
-1.17025054e+00 -6.70752347e-01 5.84693313e-01 6.75312519e-01
-4.30266529e-01 9.04178143e-01 1.00209689e+00 -8.15791607e-01
2.86915690e-01 -6.89480066e-01 -4.76388574e-01 -3.57827246e-01
1.81472883e-01 8.88976574e-01 1.07170999e-01 -7.81164467e-01
3.90465587e-01 -3.61186534e-01 -8.90846729e-01 1.02034760e+00
1.36361885e+00 4.58378375e-01 4.38897222e-01 1.18261981e+00
-3.04976732e-01 1.29029381e+00 -9.18105245e-01 -7.15665460e-01
-6.62246764e-01 -9.22932506e-01 -1.84733486e+00 -2.92109430e-01
1.33994371e-01 -3.49864483e-01 -1.04579486e-01 7.05598235e-01
2.68675864e-01 1.28680831e-02 6.57939315e-01 -8.98062825e-01
-4.28327531e-01 8.56032610e-01 -1.06891185e-01 5.18973410e-01
1.23055732e+00 -9.46499035e-02 1.11530137e+00 -5.45195639e-01
6.51118636e-01 1.43982863e+00 1.96209803e-01 8.44309926e-01
-6.00020766e-01 -2.07859024e-01 -2.20867842e-01 4.75486189e-01
-5.80861926e-01 -3.68558347e-01 3.73464286e-01 -3.43810380e-01
1.10602021e+00 6.43510878e-01 3.59285682e-01 1.36608565e+00
5.68489015e-01 1.10077929e+00 1.04326308e+00 -5.88558912e-01
-1.63218435e-02 2.38907393e-02 7.61171401e-01 5.20855367e-01
-2.21892864e-01 -5.60290933e-01 -5.69349527e-01 -2.14065179e-01
1.27471074e-01 -4.67640847e-01 -6.50875390e-01 5.12584925e-01
-1.54506552e+00 1.16258514e+00 2.20443681e-01 3.25413972e-01
-3.79130274e-01 -4.58015740e-01 5.88033915e-01 8.36098611e-01
4.43867952e-01 7.49722481e-01 -4.89514381e-01 -7.12112129e-01
3.31185795e-02 1.50415257e-01 1.51595354e+00 8.50724339e-01
2.36491635e-01 -6.70830607e-01 -4.03877914e-01 1.03093648e+00
3.21677811e-02 6.89764440e-01 3.20219576e-01 -1.08057094e+00
1.02557099e+00 5.04163682e-01 5.69804572e-02 -1.01943445e+00
-3.33048582e-01 -9.38142389e-02 -8.40076447e-01 -8.75455320e-01
7.28486240e-01 -4.20895189e-01 -2.75653988e-01 1.50946283e+00
6.27824217e-02 -6.34530962e-01 5.86298406e-01 6.26909316e-01
1.38599050e+00 1.13611746e+00 8.87518451e-02 -3.87600720e-01
1.91942132e+00 -1.30806482e+00 -1.10671723e+00 -3.96445245e-01
6.88781083e-01 -9.23840463e-01 1.23653173e+00 4.25469756e-01
-1.20294690e+00 -2.88175493e-01 -6.18961632e-01 -5.36790788e-01
-1.21765502e-01 -2.54048109e-01 2.74802715e-01 5.04244268e-01
-9.30268943e-01 -1.77377537e-01 -4.39049304e-01 -5.82519591e-01
2.35862240e-01 -1.42026201e-01 -4.38744612e-02 -4.85096395e-01
-1.58962560e+00 1.02349758e+00 2.45791778e-01 -5.49690379e-03
-8.07451427e-01 -3.43273908e-01 -7.44708419e-01 4.79415990e-02
5.78929782e-01 -6.84480727e-01 2.07884884e+00 -7.91612327e-01
-1.40376902e+00 8.13272893e-01 -7.80835569e-01 -4.26632404e-01
4.19449657e-01 -4.36888903e-01 -3.72930974e-01 6.10422373e-01
1.86301738e-01 4.76073354e-01 3.20475489e-01 -8.02310646e-01
-5.63656688e-01 -1.70280248e-01 3.55483443e-01 3.70609879e-01
6.87957853e-02 1.28554240e-01 -1.94481552e-01 -3.47761691e-01
-2.46812865e-01 -6.03112340e-01 -7.25722080e-03 -6.53195322e-01
-3.85808647e-01 -7.84160376e-01 3.72169077e-01 -1.36680436e+00
1.21324658e+00 -1.59751427e+00 2.50622690e-01 -1.11308351e-01
4.79840875e-01 9.42940786e-02 -3.09043050e-01 1.05350983e+00
2.37564027e-01 9.54008177e-02 -3.73353273e-01 1.06146060e-01
-1.27126336e-01 2.26920202e-01 -2.54124641e-01 -1.38136178e-01
1.87518165e-01 1.21148622e+00 -1.29055250e+00 -2.95407027e-01
-1.61522463e-01 7.74590522e-02 -3.52255166e-01 6.00515664e-01
-6.56578660e-01 7.36790180e-01 -7.49726653e-01 3.28220367e-01
3.68514597e-01 -3.79262567e-01 2.20131710e-01 4.09942091e-01
1.51227742e-01 7.63016164e-01 -9.74502042e-02 1.55078292e+00
-2.98582822e-01 8.59958887e-01 -3.41918617e-02 -8.56664062e-01
7.34045565e-01 5.76492786e-01 -1.40760630e-01 -9.55561221e-01
4.28672373e-01 -1.01955989e-02 2.61719912e-01 -9.76080060e-01
4.73312527e-01 3.22807878e-02 -3.06078821e-01 7.44230866e-01
-2.96508104e-01 -2.39189431e-01 2.56693870e-01 6.45510733e-01
1.19953775e+00 -6.92561686e-01 6.65313363e-01 -4.27646875e-01
8.40556204e-01 4.49745059e-01 8.28165859e-02 8.68129432e-01
-1.04893692e-01 2.87392586e-01 1.03819680e+00 -1.39672443e-01
-6.96092665e-01 -8.02407801e-01 1.49474293e-01 1.23911726e+00
-1.87887490e-01 -4.31008160e-01 -1.28668559e+00 -5.11410952e-01
-4.61325467e-01 1.19556820e+00 -6.10701919e-01 1.31891876e-01
-7.15448737e-01 -3.42411041e-01 5.03993154e-01 4.22253549e-01
7.72532463e-01 -1.37398565e+00 -4.92661059e-01 2.44229555e-01
-1.34278691e+00 -1.21025968e+00 -7.16261625e-01 -3.11222821e-01
-7.00480461e-01 -1.37503183e+00 -7.79080153e-01 -1.10578156e+00
4.37366396e-01 4.46407109e-01 1.60625339e+00 4.67865586e-01
2.37222269e-01 5.35632372e-01 -1.00715220e+00 -7.16356456e-01
-1.16720843e+00 4.25180286e-01 -5.47043145e-01 -4.64518189e-01
6.85634851e-01 2.68973298e-02 -3.88281912e-01 4.01229441e-01
-6.83007240e-01 4.58086938e-01 3.56173486e-01 9.99418199e-01
-1.74971864e-01 -4.76522416e-01 9.63536620e-01 -1.15983212e+00
1.39606106e+00 -7.56169200e-01 -2.70741992e-02 6.08952463e-01
-5.24359569e-03 -2.21895456e-01 5.38706422e-01 -1.23525262e-01
-1.36999118e+00 -6.87805831e-01 -6.14443541e-01 8.43998790e-01
-3.46000880e-01 7.59470403e-01 -5.86973354e-02 6.48985267e-01
7.48026609e-01 1.59505635e-01 2.35186636e-01 -2.25369632e-01
2.72644579e-01 1.25217545e+00 4.94471192e-01 -5.32395065e-01
1.36263743e-01 -1.51039705e-01 -8.98517370e-01 -1.28043818e+00
-1.20903647e+00 -6.43503010e-01 -3.60013872e-01 -3.18179280e-01
1.23011684e+00 -7.31338322e-01 -1.13073730e+00 4.24249291e-01
-1.45450962e+00 -5.61822057e-01 2.35032856e-01 3.22815746e-01
-5.06187797e-01 5.05501747e-01 -1.00484729e+00 -5.99606454e-01
-7.26100028e-01 -9.59264874e-01 8.57669413e-01 2.25974604e-01
-8.33955824e-01 -1.13174272e+00 3.44397649e-02 1.33395886e+00
2.76572466e-01 -1.05206817e-02 1.16855323e+00 -8.63737524e-01
-7.78032169e-02 1.67382181e-01 -6.24096133e-02 2.78920680e-01
5.62455542e-02 -6.21079564e-01 -7.85260618e-01 -4.15416181e-01
3.95799994e-01 -7.96740294e-01 5.30719697e-01 1.22689949e-02
1.02383029e+00 -4.76161718e-01 2.84666941e-02 -4.61444438e-01
6.60047352e-01 4.68230873e-01 8.99026930e-01 4.71794009e-02
2.84389406e-01 1.03345704e+00 6.66759372e-01 5.19206151e-02
9.29462373e-01 1.07783057e-01 -2.81210779e-03 1.81848601e-01
2.15082094e-01 -4.40640956e-01 3.67916852e-01 1.67124641e+00
-2.66799778e-02 -7.71939516e-01 -1.46096277e+00 5.97670019e-01
-1.68755519e+00 -8.35560322e-01 -4.82533306e-01 1.42659259e+00
9.63556886e-01 -3.38467583e-02 -3.14404517e-02 -2.00033765e-02
4.45737511e-01 4.48932089e-02 -3.28376949e-01 -8.74233484e-01
4.98531424e-02 8.50065798e-02 -2.41933614e-01 9.16512489e-01
-5.39177120e-01 8.21737409e-01 5.80403996e+00 5.92713356e-01
-4.58010614e-01 1.31536141e-01 8.06487143e-01 7.20620036e-01
-3.29054236e-01 -3.45470399e-01 -3.03778440e-01 1.68192104e-01
1.28066528e+00 -4.71717089e-01 3.18484485e-01 1.84113488e-01
3.63349169e-01 -4.82441843e-01 -1.09489048e+00 5.97329378e-01
2.86133260e-01 -1.14126003e+00 1.98008806e-01 -2.45329082e-01
4.91478652e-01 -1.32133916e-01 -3.27700227e-01 5.32063603e-01
1.18922599e-01 -1.22327864e+00 4.99384850e-02 4.47795004e-01
3.57698888e-01 -7.15565622e-01 1.23282385e+00 1.05271375e+00
-5.29882252e-01 6.61099795e-03 -3.53732139e-01 -3.73665750e-01
4.18338686e-01 1.24419741e-01 -1.32044280e+00 4.84996051e-01
5.85008025e-01 5.25562942e-01 -4.86089349e-01 5.71712673e-01
-5.75817168e-01 1.07816374e+00 2.41881445e-01 -7.63443053e-01
3.47656250e-01 -1.44204855e-01 4.19802040e-01 1.33709753e+00
-2.09255546e-01 8.66971552e-01 -3.38428058e-02 2.04455599e-01
-2.45855853e-01 4.01537985e-01 -3.38586330e-01 -2.05682099e-01
2.56088734e-01 6.61400259e-01 -3.76452863e-01 -5.75013816e-01
-4.71246064e-01 8.81127894e-01 4.45381314e-01 2.76297718e-01
-5.73386550e-01 -4.12136972e-01 1.01303548e-01 -1.72290146e-01
-2.54710674e-01 -9.74866152e-02 -4.77986634e-02 -1.19595325e+00
4.59193662e-02 -1.83770001e+00 5.53410172e-01 -8.61079574e-01
-1.37647057e+00 7.74274230e-01 8.31932127e-02 -6.28368318e-01
-5.28690040e-01 -5.25335431e-01 -6.70909643e-01 7.36937821e-01
-1.09658861e+00 -7.06072927e-01 -4.91256297e-01 4.31723863e-01
1.28069293e+00 -8.11598152e-02 1.01815593e+00 -6.65104166e-02
-2.23707929e-01 4.60560888e-01 -5.30770095e-03 2.97717988e-01
6.26644731e-01 -1.20805752e+00 5.44385076e-01 3.14470589e-01
-3.49205881e-01 6.57293439e-01 8.27533424e-01 -6.03963971e-01
-1.36644423e+00 -6.02076828e-01 1.65125573e+00 -8.36329043e-01
6.70705140e-01 -3.53509247e-01 -1.35636735e+00 7.59726286e-01
1.13318634e+00 -1.47641635e+00 1.01766407e+00 1.22020520e-01
-3.53182927e-02 2.73957103e-01 -1.12236094e+00 7.62158513e-01
8.40516984e-01 -5.87539256e-01 -1.46294677e+00 7.00505853e-01
1.14237428e+00 -4.22805578e-01 -1.03631651e+00 4.55509871e-02
2.73592412e-01 -6.87992036e-01 6.56120539e-01 -7.96266735e-01
7.26928651e-01 3.07106018e-01 -2.93438416e-02 -1.47049534e+00
1.06867254e-01 -7.94396043e-01 1.81619197e-01 7.94739902e-01
7.82400191e-01 -9.52268958e-01 3.85481566e-01 5.85167885e-01
-2.51008213e-01 -6.87931955e-01 -7.49216259e-01 2.91283987e-02
6.01791620e-01 -2.98952043e-01 4.73320812e-01 9.95972097e-01
6.68398082e-01 1.01693356e+00 -1.72140047e-01 -1.45991042e-01
2.69614518e-01 -1.26133114e-01 7.39672244e-01 -9.13210630e-01
-1.15496546e-01 -1.67870224e-01 4.16956812e-01 -1.63223851e+00
1.65141165e-01 -6.07403219e-01 1.63605124e-01 -1.91504836e+00
4.42356765e-01 2.91499794e-01 5.53138912e-01 2.38306448e-01
-4.29171354e-01 -3.27043742e-01 7.32534740e-04 -7.21610989e-03
-6.14628792e-01 5.12504280e-01 1.90042019e+00 -3.12549829e-01
-7.68151209e-02 1.78917423e-01 -9.82760191e-01 4.38176811e-01
1.52222753e+00 -1.92355588e-01 -5.97505212e-01 -6.90462410e-01
1.83739290e-02 7.66070604e-01 -4.93258424e-02 -3.24643552e-01
3.71486455e-01 -2.20256507e-01 -7.37583861e-02 -7.19210029e-01
1.78176701e-01 -1.80906579e-01 -3.46523494e-01 7.63349831e-01
-8.69912922e-01 5.08231044e-01 7.98623189e-02 3.61986846e-01
-4.37757194e-01 -3.33130211e-01 4.09492582e-01 -2.73655236e-01
-4.03450578e-01 -3.80911469e-01 -1.10291576e+00 6.32738590e-01
7.11741984e-01 2.85615623e-01 -8.94081950e-01 -1.18027258e+00
-5.55036366e-01 1.00598884e+00 -1.82126418e-01 7.68076360e-01
7.81069338e-01 -6.32311821e-01 -1.34466207e+00 -1.99859440e-01
2.76844889e-01 -2.92422418e-02 3.61582905e-01 6.42562628e-01
-8.34771693e-01 9.22268510e-01 -9.86414328e-02 -4.75315869e-01
-1.55698037e+00 2.04138041e-01 2.00504690e-01 -2.77615607e-01
-5.07744968e-01 5.63122928e-01 1.21182278e-01 -5.79733908e-01
1.88144356e-01 -4.24420953e-01 -8.22697639e-01 3.42032798e-02
1.01390314e+00 4.89012778e-01 -4.06717248e-02 -4.54990149e-01
1.02051785e-02 1.53989330e-01 -3.11900079e-01 -1.83211058e-01
8.87793124e-01 -3.72004479e-01 -5.03242075e-01 4.93740052e-01
1.23537767e+00 -1.26638368e-01 -2.34173551e-01 -2.29971305e-01
2.52155662e-01 -1.06854461e-01 -5.33628345e-01 -1.19608545e+00
-1.98034942e-01 1.01267755e+00 -6.31532967e-02 4.12969947e-01
9.28827465e-01 4.00064170e-01 1.13692915e+00 1.02333736e+00
2.23072335e-01 -9.01398003e-01 3.50078344e-01 1.04240918e+00
1.42304945e+00 -1.34972727e+00 -3.24599266e-01 -6.51037037e-01
-1.05577636e+00 1.06586230e+00 6.66859090e-01 4.64195937e-01
3.66625279e-01 -3.77853096e-01 6.36251032e-01 -3.24239761e-01
-1.06750536e+00 1.39768973e-01 2.62513369e-01 4.68135357e-01
7.17934787e-01 2.63086140e-01 -6.38342738e-01 4.52881515e-01
-7.00054109e-01 -3.26402158e-01 7.40995526e-01 7.91786015e-01
-6.13484561e-01 -9.75113392e-01 -3.22096854e-01 4.65455890e-01
-3.98401260e-01 -7.84203112e-02 -8.15772593e-01 7.12171614e-01
-7.27898717e-01 1.93352509e+00 -2.56132726e-02 -1.05004981e-01
5.13147652e-01 1.07946983e-02 1.45166233e-01 -5.72843432e-01
-8.90850663e-01 -1.96463734e-01 8.33217680e-01 -2.45868251e-01
-4.46414143e-01 -6.13289833e-01 -1.16264558e+00 -6.06056154e-01
-3.29769328e-02 9.56240714e-01 5.43511026e-02 9.17097509e-01
-6.59522489e-02 4.21979785e-01 4.38207835e-01 1.51667267e-01
-4.35932308e-01 -1.28588212e+00 1.86581463e-01 2.98232466e-01
4.17716771e-01 5.65784276e-02 -3.38845223e-01 -1.17178611e-01] | [11.756671905517578, 8.083256721496582] |
152f1f9c-0c18-4e2e-83cb-913c1eaba9b3 | selecting-computations-theory-and | 1408.2048 | null | http://arxiv.org/abs/1408.2048v1 | http://arxiv.org/pdf/1408.2048v1.pdf | Selecting Computations: Theory and Applications | Sequential decision problems are often approximately solvable by simulating
possible future action sequences. Metalevel decision procedures have been
developed for selecting which action sequences to simulate, based on estimating
the expected improvement in decision quality that would result from any
particular simulation; an example is the recent work on using bandit algorithms
to control Monte Carlo tree search in the game of Go. In this paper we develop
a theoretical basis for metalevel decisions in the statistical framework of
Bayesian selection problems, arguing (as others have done) that this is more
appropriate than the bandit framework. We derive a number of basic results
applicable to Monte Carlo selection problems, including the first finite
sampling bounds for optimal policies in certain cases; we also provide a simple
counterexample to the intuitive conjecture that an optimal policy will
necessarily reach a decision in all cases. We then derive heuristic
approximations in both Bayesian and distribution-free settings and demonstrate
their superiority to bandit-based heuristics in one-shot decision problems and
in Go. | ['David Tolpin', 'Solomon Eyal Shimony', 'Stuart Russell', 'Nicholas Hay'] | 2014-08-09 | null | null | null | null | ['game-of-go'] | ['playing-games'] | [ 3.24327022e-01 2.92057216e-01 -7.76862264e-01 -2.77248532e-01
-1.02482128e+00 -5.34426212e-01 6.21656597e-01 -1.36218682e-01
-7.48993218e-01 1.35744822e+00 4.03248578e-01 -9.03262079e-01
-8.18514645e-01 -8.33557367e-01 -5.18626511e-01 -6.84535623e-01
9.61064454e-03 9.00863111e-01 1.83543116e-01 1.62391260e-01
4.76660877e-01 3.14619601e-01 -1.33693874e+00 -6.69386536e-02
8.92747045e-01 9.52120483e-01 3.14432234e-02 9.69424188e-01
3.06285508e-02 7.81589746e-01 -5.26889920e-01 -5.45260191e-01
4.24830914e-01 -8.41401517e-01 -9.02698398e-01 1.73864096e-01
-2.27049544e-01 -7.24978685e-01 -2.29730442e-01 1.07752335e+00
4.65612203e-01 5.53460658e-01 9.14435089e-01 -1.05471957e+00
-3.10524832e-02 1.01421022e+00 -5.54020524e-01 2.94492811e-01
5.12956321e-01 3.97589117e-01 1.21884537e+00 1.97352245e-01
4.00992304e-01 1.57542479e+00 3.33477080e-01 4.61705089e-01
-1.43654478e+00 -3.74170452e-01 3.04265529e-01 2.77228534e-01
-9.75774944e-01 -3.16528708e-01 2.67665893e-01 -1.55328020e-01
6.77720666e-01 3.90011191e-01 1.08449066e+00 9.59031284e-01
1.54491931e-01 1.14503562e+00 1.44628799e+00 -8.36840451e-01
7.15174794e-01 -2.17965469e-01 6.92685768e-02 2.28804201e-01
7.53499448e-01 7.34320343e-01 -3.60731632e-01 -5.90678573e-01
8.35758030e-01 -7.81092718e-02 -1.09881617e-01 -3.73300314e-01
-7.96572745e-01 1.20315254e+00 -1.84648097e-01 -9.81919765e-02
-7.46913373e-01 5.93056738e-01 2.66167879e-01 2.83737063e-01
2.54793286e-01 3.24357808e-01 -2.65979946e-01 -3.38204086e-01
-9.88436997e-01 9.17890191e-01 1.08747363e+00 7.02408791e-01
3.83974612e-01 -1.09032288e-01 -6.78995132e-01 5.45986831e-01
3.83074254e-01 5.09543896e-01 6.52024150e-02 -1.45551944e+00
5.01579463e-01 -4.16992664e-01 1.08554614e+00 -2.57802039e-01
-1.94891125e-01 -3.33285481e-01 -2.92087018e-01 3.15158784e-01
7.04206228e-01 -5.53013206e-01 -7.37928450e-01 1.53042293e+00
2.43337139e-01 -1.49480477e-01 -2.61233479e-01 4.77883905e-01
-4.09090757e-01 5.60031652e-01 1.01143911e-01 -8.21708083e-01
9.06896055e-01 -3.59283984e-01 -7.98301101e-01 -3.02624047e-01
4.78459597e-01 -4.41850930e-01 6.18265629e-01 6.94529414e-01
-1.09345913e+00 1.14620917e-01 -8.58027637e-01 7.04428375e-01
3.28193784e-01 -4.00356382e-01 1.06431031e+00 1.11746705e+00
-7.62307107e-01 7.57796526e-01 -7.90052712e-01 -4.55145478e-01
4.88198996e-01 1.85370967e-01 6.05806768e-01 -2.03755185e-01
-1.02872694e+00 1.09398019e+00 5.43563366e-01 -1.03128828e-01
-1.17528200e+00 -7.83180520e-02 -2.75387168e-01 1.67489797e-01
1.11984575e+00 -7.73467243e-01 1.94632161e+00 -8.15785825e-01
-1.61075246e+00 3.94844860e-01 -1.08347610e-01 -9.06304657e-01
9.28221703e-01 -5.90668693e-02 1.33837387e-01 -1.22201301e-01
1.73747301e-01 2.98293948e-01 5.20798087e-01 -8.31454694e-01
-1.18572319e+00 -3.19500655e-01 4.78450835e-01 5.08930624e-01
4.01805520e-01 1.87155887e-01 9.22060907e-02 -2.81673521e-01
-1.65023848e-01 -9.64315176e-01 -8.72423768e-01 -5.75136006e-01
-4.52858537e-01 -3.74940902e-01 -2.77350098e-01 -7.75733739e-02
1.38269758e+00 -1.47316158e+00 -3.94274652e-01 3.59632552e-01
-3.67947489e-01 -2.23483518e-01 2.44675502e-01 5.16457200e-01
2.80818105e-01 1.88411281e-01 2.29010656e-02 3.38577360e-01
5.34376681e-01 1.55789182e-01 -3.86219770e-01 6.01263881e-01
-4.70538676e-01 6.50252163e-01 -7.75736392e-01 -3.85517806e-01
9.88079607e-02 -6.55258417e-01 -7.04704344e-01 -1.95265487e-01
-4.76602852e-01 4.95927557e-02 -9.02176440e-01 3.17136258e-01
4.15172398e-01 -6.96837008e-02 4.97096926e-01 6.59810305e-01
-1.04924239e-01 5.52266538e-01 -1.28082538e+00 9.77811992e-01
-2.97781885e-01 3.15854669e-01 1.47527516e-01 -1.15193510e+00
2.13667706e-01 2.35027567e-01 2.59352654e-01 -2.47806028e-01
3.95454139e-01 1.14340827e-01 9.38377753e-02 -3.13793182e-01
3.61114264e-01 -7.36882567e-01 -3.73560905e-01 8.71854067e-01
-3.56333375e-01 -4.94140327e-01 3.52333814e-01 -4.56095673e-02
1.01306212e+00 1.90812781e-01 8.18602204e-01 -2.60436594e-01
-2.15766057e-01 3.33076894e-01 5.49915910e-01 1.92811573e+00
-3.74779254e-01 -7.93050416e-03 7.53637791e-01 -2.51402617e-01
-8.61856699e-01 -8.46679509e-01 -1.79664820e-01 1.04017043e+00
3.90708521e-02 3.91663834e-02 -6.85462117e-01 -4.65975225e-01
1.53185353e-01 1.41659653e+00 -5.60533404e-01 1.15373902e-01
-9.23286304e-02 -1.10850406e+00 1.00306280e-01 3.41786027e-01
3.14505190e-01 -8.01800132e-01 -1.00693226e+00 5.28869271e-01
-5.96739128e-02 -6.40160382e-01 -2.13037953e-01 3.70555580e-01
-9.98358965e-01 -1.02372932e+00 -6.07828677e-01 2.78576851e-01
3.33556049e-02 1.39325112e-01 8.88412237e-01 -4.24534917e-01
6.53241426e-02 5.50087094e-01 -2.57444054e-01 -6.74710333e-01
-3.99680823e-01 -1.34332538e-01 7.02708513e-02 -4.54166651e-01
5.68894982e-01 -2.25637868e-01 -5.84372818e-01 2.89495438e-01
-4.89065140e-01 -6.87588453e-02 4.44735378e-01 8.40517044e-01
2.80234754e-01 2.55121946e-01 4.48701590e-01 -9.92425263e-01
1.03762615e+00 -4.57986236e-01 -1.06565034e+00 2.98251182e-01
-5.96243382e-01 1.29485384e-01 1.29870474e-01 -2.85370618e-01
-1.35490668e+00 -8.70792493e-02 1.30677238e-01 1.31972045e-01
-1.20569721e-01 5.98681986e-01 7.22697154e-02 3.17471027e-01
7.67268896e-01 3.60905230e-02 -2.69399911e-01 -1.91845089e-01
3.32009703e-01 6.17699742e-01 4.20868322e-02 -1.01075184e+00
2.54231423e-01 3.67327273e-01 1.34490982e-01 -3.56880993e-01
-1.10338664e+00 2.75086928e-02 -1.20814480e-01 -1.64614722e-01
6.33689821e-01 -5.35389721e-01 -1.00884473e+00 2.64965259e-02
-8.01224768e-01 -5.41761577e-01 -4.51833218e-01 8.52346838e-01
-1.48885453e+00 2.13463649e-01 -2.27953807e-01 -1.78092217e+00
4.08185244e-01 -9.63589907e-01 4.31302756e-01 3.59455585e-01
-3.10311884e-01 -7.63508260e-01 5.18532880e-02 3.32381725e-01
1.02602713e-01 2.29637045e-02 7.74941564e-01 -5.80461621e-01
-8.21320415e-01 -2.73720741e-01 1.49010211e-01 -1.65445477e-01
-2.75216609e-01 -8.35048258e-02 -4.93267506e-01 -2.47171938e-01
8.08768496e-02 -1.56799331e-01 9.34618950e-01 1.24427414e+00
8.61148596e-01 -5.89655519e-01 -6.44514441e-01 2.39771232e-02
1.37558746e+00 8.34388852e-01 5.06405354e-01 4.99882042e-01
-2.42423236e-01 4.47681457e-01 1.14486468e+00 8.79523396e-01
3.36639620e-02 7.95244873e-01 2.04073429e-01 5.58209538e-01
5.81950068e-01 -3.65786165e-01 2.08173797e-01 -3.59233886e-01
-1.18021116e-01 -4.75753307e-01 -6.29695535e-01 7.47301579e-01
-2.15213656e+00 -1.43201721e+00 1.57768041e-01 2.58923984e+00
7.89690733e-01 5.23668766e-01 6.81075692e-01 2.31412724e-02
9.04697180e-01 -1.08893840e-02 -6.80868447e-01 -5.86595654e-01
2.31568232e-01 1.39069363e-01 1.08770871e+00 5.93068719e-01
-9.58478808e-01 8.36941242e-01 8.15006256e+00 1.19883668e+00
-2.47810349e-01 1.70398772e-01 9.84893560e-01 -4.75729376e-01
-3.20747018e-01 2.82371253e-01 -1.01429558e+00 4.08327430e-01
1.14402163e+00 -6.80336297e-01 4.58486527e-01 8.29813838e-01
6.79254651e-01 -9.84637737e-01 -1.03144467e+00 6.03519022e-01
-6.95832312e-01 -1.28707230e+00 -3.38242263e-01 2.82107443e-01
1.00250983e+00 -2.37536788e-01 -2.36474067e-01 3.28263521e-01
1.63294220e+00 -9.68901336e-01 8.38928401e-01 4.62054193e-01
5.93727529e-01 -1.04699385e+00 5.51235318e-01 6.98321939e-01
-3.21885854e-01 -5.72259784e-01 -4.35172081e-01 -4.83414620e-01
3.32746744e-01 5.77596664e-01 -9.08020258e-01 3.77316773e-01
4.50765043e-01 -4.73767966e-02 2.09310845e-01 1.64192474e+00
-3.53077471e-01 9.22921598e-01 -5.80408633e-01 -5.14141440e-01
6.16677940e-01 -3.87761414e-01 5.12599826e-01 8.99841607e-01
4.31763321e-01 2.85051137e-01 1.89096436e-01 4.94794428e-01
5.47489762e-01 -2.53661901e-01 -5.84294021e-01 -4.32705432e-02
8.14772069e-01 3.82765561e-01 -8.10316384e-01 -4.21074033e-01
-1.47367522e-01 5.07342160e-01 -1.46122828e-01 3.92096788e-01
-7.12428272e-01 -3.02685857e-01 5.52261472e-01 -1.15917340e-01
6.59146369e-01 1.26064286e-01 -5.57490706e-01 -8.68219972e-01
-2.66382396e-01 -8.48254740e-01 6.11272335e-01 -5.77237904e-01
-9.87274170e-01 -2.38854349e-01 7.49290109e-01 -8.30561280e-01
-9.41585541e-01 -3.81311566e-01 -6.40422940e-01 8.10578465e-01
-9.17757928e-01 -2.82942027e-01 5.67202985e-01 1.06445141e-01
5.04419923e-01 2.56095231e-01 4.51637894e-01 -4.93790984e-01
-3.48006070e-01 1.97185919e-01 6.70059741e-01 -3.18000495e-01
1.02232508e-01 -1.25350296e+00 6.86316267e-02 7.76373565e-01
-1.36692047e-01 3.35462064e-01 1.26677465e+00 -6.47931933e-01
-1.08324456e+00 -4.78904694e-01 3.27565372e-01 -1.38100266e-01
6.85201287e-01 1.61824971e-01 -9.00448412e-02 8.05248737e-01
-1.47066778e-02 -6.61491275e-01 4.20922458e-01 3.98087800e-01
2.61776596e-01 6.43589795e-02 -1.28874171e+00 9.01527226e-01
1.16801250e+00 -1.01593234e-01 -6.06867015e-01 4.49428976e-01
2.45019868e-01 -1.64955258e-01 -3.45189214e-01 9.14318040e-02
7.98707306e-01 -1.23470700e+00 6.71565950e-01 -7.51894176e-01
8.27630609e-02 -5.19747548e-02 -1.72331065e-01 -1.46731579e+00
-3.78657579e-01 -1.18096864e+00 3.04611504e-01 7.24496305e-01
3.77936929e-01 -6.12282276e-01 9.54271913e-01 7.30170548e-01
3.69192094e-01 -6.80073023e-01 -1.29603291e+00 -9.64943588e-01
3.31050038e-01 -7.76293755e-01 5.79621315e-01 2.97912508e-01
5.86096346e-02 1.18692983e-02 -5.35782814e-01 -2.83173352e-01
9.03771877e-01 2.09902033e-01 6.16748929e-01 -9.49522436e-01
-8.27148736e-01 -6.70451343e-01 -7.34457150e-02 -1.50672328e+00
1.88200120e-02 -2.21908525e-01 2.37190261e-01 -1.62942767e+00
4.73483562e-01 -3.18735570e-01 -1.39557153e-01 9.18673426e-02
-2.07778037e-01 -4.76485819e-01 2.89831519e-01 -7.12128282e-02
-7.29543328e-01 3.71099919e-01 1.24273276e+00 4.72176410e-02
-6.03985377e-02 8.02166641e-01 -1.09921145e+00 7.05344737e-01
5.78884482e-01 -6.24701977e-01 -4.74689901e-01 1.36451945e-01
3.61524642e-01 9.51028287e-01 7.16539398e-02 -6.67679071e-01
4.27068993e-02 -1.04797232e+00 1.63693190e-01 -8.02597880e-01
1.11690909e-01 -5.37320137e-01 2.87029594e-01 6.54694736e-01
-6.91024244e-01 -4.75534558e-01 -7.29236007e-02 1.13844824e+00
2.94598758e-01 -9.33868647e-01 8.67469490e-01 -4.67771351e-01
-3.66065800e-01 5.01089729e-02 -1.07442844e+00 2.23388895e-02
1.05647862e+00 -3.76030922e-01 -1.59174427e-02 -1.13824332e+00
-1.01762867e+00 4.56806749e-01 3.74024838e-01 -4.72663075e-01
8.52253810e-02 -1.00651336e+00 -6.45503342e-01 -5.22379279e-01
-2.40213349e-01 -4.97332722e-01 2.41942257e-02 7.86571741e-01
-2.21673205e-01 6.90562010e-01 6.77350089e-02 -9.26115289e-02
-1.00916958e+00 3.64121646e-01 4.82060850e-01 -6.98190749e-01
-1.34472832e-01 7.33943880e-01 6.09023087e-02 1.57388180e-01
2.64768034e-01 -1.79869950e-01 1.58841476e-01 4.48294319e-02
6.33537352e-01 6.91483855e-01 -4.38171387e-01 2.36224025e-01
-3.62895653e-02 3.15871574e-02 -1.17740847e-01 -9.07402158e-01
1.24235797e+00 -4.54372048e-01 3.25918138e-01 4.22706693e-01
2.85183758e-01 -3.28588814e-01 -1.60376132e+00 -3.11203212e-01
2.77936757e-01 -7.82816410e-01 2.73846984e-01 -9.19194639e-01
-2.92469949e-01 5.00518143e-01 3.12192351e-01 7.03610361e-01
9.04124200e-01 -1.04729116e-01 9.60025713e-02 5.42086303e-01
9.28075194e-01 -1.38578296e+00 -6.94982409e-01 3.50158304e-01
4.89222378e-01 -8.66027832e-01 4.44213778e-01 -1.13190241e-01
-7.01476455e-01 8.92027795e-01 -6.77076951e-02 -1.79517180e-01
4.24005926e-01 -7.35441968e-02 -6.13101840e-01 6.24346659e-02
-1.14018631e+00 -5.99277139e-01 -3.15951347e-01 7.07652986e-01
8.00382271e-02 5.32970786e-01 -7.30128348e-01 4.28880692e-01
-2.62166321e-01 4.64655608e-01 7.96824992e-01 9.28548515e-01
-8.49157870e-01 -9.51964736e-01 -6.81680799e-01 9.91418064e-01
-7.34007955e-01 2.06175167e-02 -2.79902190e-01 8.03131282e-01
-3.54046822e-01 1.28269863e+00 1.53188780e-01 1.91117108e-01
-1.42822117e-02 1.41521767e-01 9.99333322e-01 -5.05500913e-01
-1.58671081e-01 4.74311024e-01 7.06794858e-01 -4.09828067e-01
-3.97401094e-01 -1.15695810e+00 -5.80676377e-01 -5.68365037e-01
-4.83793825e-01 4.14761037e-01 3.47300112e-01 1.20575571e+00
-1.69272259e-01 1.70757204e-01 5.20023227e-01 -7.62845755e-01
-1.37666345e+00 -9.12382662e-01 -1.05428302e+00 -2.94617265e-01
1.17683699e-02 -7.88205743e-01 -4.73314553e-01 -5.87528110e-01] | [4.486444473266602, 3.1057615280151367] |
f80af314-7499-4b16-9f33-2df0ef5efdeb | a-study-of-face-obfuscation-in-imagenet | 2103.06191 | null | https://arxiv.org/abs/2103.06191v3 | https://arxiv.org/pdf/2103.06191v3.pdf | A Study of Face Obfuscation in ImageNet | Face obfuscation (blurring, mosaicing, etc.) has been shown to be effective for privacy protection; nevertheless, object recognition research typically assumes access to complete, unobfuscated images. In this paper, we explore the effects of face obfuscation on the popular ImageNet challenge visual recognition benchmark. Most categories in the ImageNet challenge are not people categories; however, many incidental people appear in the images, and their privacy is a concern. We first annotate faces in the dataset. Then we demonstrate that face obfuscation has minimal impact on the accuracy of recognition models. Concretely, we benchmark multiple deep neural networks on obfuscated images and observe that the overall recognition accuracy drops only slightly (<= 1.0%). Further, we experiment with transfer learning to 4 downstream tasks (object recognition, scene recognition, face attribute classification, and object detection) and show that features learned on obfuscated images are equally transferable. Our work demonstrates the feasibility of privacy-aware visual recognition, improves the highly-used ImageNet challenge benchmark, and suggests an important path for future visual datasets. Data and code are available at https://github.com/princetonvisualai/imagenet-face-obfuscation. | ['Olga Russakovsky', 'Jia Deng', 'Li Fei-Fei', 'Jacqueline Yau', 'Kaiyu Yang'] | 2021-03-10 | a-study-of-face-obfuscation-in-imagenet-1 | https://openreview.net/forum?id=KVYq2Ea90PC | https://openreview.net/pdf?id=KVYq2Ea90PC | null | ['scene-recognition'] | ['computer-vision'] | [ 2.82132179e-01 1.08757108e-01 -1.64211974e-01 -5.07982433e-01
-4.57229435e-01 -8.88562739e-01 7.46934950e-01 -2.81525254e-01
-4.27813143e-01 6.31011844e-01 7.33328685e-02 -4.37097818e-01
3.35592955e-01 -4.28017706e-01 -1.00184464e+00 -7.00302660e-01
-2.82139570e-01 -7.91101623e-03 -5.58338761e-01 3.51942450e-01
3.19228880e-02 8.58353853e-01 -1.28488350e+00 4.91781980e-01
2.11826518e-01 1.06481564e+00 -6.10004544e-01 8.23380232e-01
4.14593220e-01 5.85044563e-01 -6.41684532e-01 -9.73019004e-01
6.77140236e-01 -1.40683994e-01 -8.32415819e-01 1.87692329e-01
1.36230087e+00 -9.56801414e-01 -8.96291435e-01 1.21489871e+00
2.10751668e-01 -2.24645153e-01 6.05396628e-01 -1.71111870e+00
-1.25655317e+00 1.18899412e-01 -5.20580649e-01 3.08221281e-01
1.45358695e-02 6.43708229e-01 6.86577678e-01 -7.39190876e-01
4.24446344e-01 1.40129435e+00 6.53504789e-01 8.32781672e-01
-1.36303544e+00 -1.03755999e+00 -2.40673646e-01 2.81257331e-01
-1.52206993e+00 -1.06061685e+00 3.58742446e-01 -4.48477149e-01
5.05084753e-01 6.08861625e-01 3.26273322e-01 1.37233698e+00
-8.31389576e-02 7.05709279e-01 1.08158684e+00 -9.35431719e-02
-9.08561572e-02 2.34893486e-01 1.57157287e-01 8.25187564e-01
5.23151159e-01 5.10550201e-01 -3.32077205e-01 -4.38424557e-01
5.42798579e-01 3.23857844e-01 -5.15400946e-01 -2.76651531e-01
-6.49260640e-01 7.43449628e-01 6.41613841e-01 -6.77824318e-02
1.51917920e-01 4.02103901e-01 4.55315322e-01 5.55204213e-01
4.54497576e-01 3.57935309e-01 -2.60429382e-01 3.29080641e-01
-6.83658719e-01 3.56150150e-01 8.01528156e-01 7.72420585e-01
7.33525693e-01 -2.23871158e-03 -2.84538776e-01 7.25452662e-01
1.63303316e-01 6.88279152e-01 1.77852035e-01 -1.07079959e+00
2.80764788e-01 1.69277415e-01 7.42677748e-02 -1.17305005e+00
1.72775388e-01 6.32046461e-02 -8.56163859e-01 4.30196464e-01
7.58584321e-01 -2.58763641e-01 -1.22963583e+00 1.67494965e+00
1.60926342e-01 2.48334423e-01 -1.56502441e-01 9.69278634e-01
7.67867029e-01 3.80178630e-01 3.04069340e-01 1.98633835e-01
1.61287391e+00 -8.09870541e-01 -5.65710723e-01 -2.16727018e-01
4.27623272e-01 -5.29066026e-01 7.04381227e-01 1.64088666e-01
-6.04582787e-01 -1.57825068e-01 -9.06650245e-01 -3.18136781e-01
-5.11509955e-01 1.76489979e-01 8.78107250e-01 1.24949026e+00
-1.15698338e+00 4.90938693e-01 -6.29032314e-01 -2.76969314e-01
1.51460087e+00 6.36710584e-01 -1.00409067e+00 -4.50784504e-01
-7.33477294e-01 8.28417182e-01 -1.46563184e-02 -3.90032418e-02
-1.44120038e+00 -8.19730639e-01 -8.68613720e-01 3.01137120e-01
2.08494559e-01 -3.56218755e-01 1.11851716e+00 -1.40066338e+00
-8.49319160e-01 1.25479639e+00 -4.98930290e-02 -6.35232091e-01
5.46520710e-01 -8.28621015e-02 -5.52851796e-01 2.42257297e-01
-3.39735866e-01 9.32951629e-01 1.39008248e+00 -1.17692077e+00
-2.57858336e-01 -6.01080298e-01 -3.30143347e-02 -2.29923978e-01
-5.90475380e-01 2.14482158e-01 -6.05554022e-02 -6.04892612e-01
-5.64967871e-01 -9.04170632e-01 1.72943890e-01 6.32389128e-01
-3.74291956e-01 1.21457651e-01 1.17339325e+00 -1.00217807e+00
6.09778881e-01 -2.56324530e+00 -4.07935917e-01 2.84410808e-02
5.05339265e-01 6.76151693e-01 -4.60447758e-01 -4.89510186e-02
-3.45427662e-01 5.68869591e-01 -3.51618558e-01 -6.11115634e-01
-7.04775378e-02 6.17780238e-02 -5.51991224e-01 9.85795915e-01
2.69443989e-01 1.14009690e+00 -3.95722717e-01 -5.12784384e-02
5.44494204e-02 6.62354887e-01 -6.08570576e-01 8.33545104e-02
-4.36112657e-02 2.83748448e-01 3.25152390e-02 9.63053763e-01
1.18318057e+00 -5.41086309e-03 1.16832472e-01 -1.48892757e-02
4.06156898e-01 -9.40584391e-03 -4.96256977e-01 9.67430413e-01
-1.15759619e-01 1.19745505e+00 4.28243876e-01 -7.89113343e-01
4.59494323e-01 2.70684958e-01 2.85193045e-02 -4.31942225e-01
2.54660070e-01 -8.10666233e-02 7.64328241e-02 -5.25306582e-01
2.61189699e-01 5.88273928e-02 3.22933823e-01 4.35228527e-01
2.34851658e-01 3.76503289e-01 -4.75191891e-01 6.44787848e-02
1.13184559e+00 -4.25017923e-01 8.78691077e-02 -2.08577827e-01
2.21001983e-01 -3.73922586e-01 1.59809992e-01 7.78884292e-01
-6.93196774e-01 7.28929937e-01 4.88328993e-01 -7.27912843e-01
-1.11554766e+00 -9.24363911e-01 -2.68680394e-01 1.05792260e+00
-3.06199603e-02 -2.71477222e-01 -9.86737788e-01 -1.27493310e+00
6.88370109e-01 4.09383655e-01 -9.65491891e-01 -2.95015961e-01
-3.71955425e-01 -5.57981730e-01 1.13977635e+00 2.17888102e-01
6.54069364e-01 -9.70290661e-01 -7.80407488e-02 -6.59720242e-01
1.46938145e-01 -9.58526015e-01 -7.92952776e-01 -5.06270587e-01
-4.93015409e-01 -1.20692456e+00 -6.66336119e-01 -4.81525272e-01
9.16893601e-01 5.44859767e-01 6.90298617e-01 5.57496071e-01
-6.77378237e-01 5.67891777e-01 3.53711657e-02 -3.79047841e-01
-4.19919968e-01 -1.99243531e-01 1.30895711e-02 5.62193692e-01
5.47062933e-01 -5.39556861e-01 -5.81281066e-01 1.10752150e-01
-9.45017457e-01 -4.47426230e-01 2.88768202e-01 8.00887048e-01
-6.19858541e-02 -2.34772995e-01 1.60654932e-01 -9.04629588e-01
3.18160146e-01 -4.74251717e-01 -6.99118733e-01 1.88731313e-01
-3.80661815e-01 -1.50874972e-01 3.53710830e-01 -5.64929843e-01
-7.87556410e-01 9.64185670e-02 -4.60122228e-02 -8.74408484e-01
-6.25376821e-01 -2.84596413e-01 -2.86662817e-01 -6.80217028e-01
6.19325042e-01 1.50526479e-01 4.18423444e-01 -5.60056150e-01
3.28773856e-01 8.32663536e-01 5.09938180e-01 -2.92029589e-01
9.60953057e-01 8.43130648e-01 -4.06997614e-02 -1.01617718e+00
-4.37911928e-01 -1.09030336e-01 -2.52828032e-01 2.37749126e-02
7.14652658e-01 -9.38709736e-01 -1.15146446e+00 4.81227010e-01
-1.19809198e+00 -2.38846689e-01 2.99604833e-02 1.67161793e-01
-2.08767146e-01 5.13389647e-01 -6.36979640e-01 -7.62284577e-01
-2.93124855e-01 -9.44775999e-01 9.41208720e-01 1.25022784e-01
-9.65173170e-02 -6.91422820e-01 -4.07383591e-01 6.61505044e-01
4.48045313e-01 1.98807836e-01 5.96775532e-01 -7.68888593e-01
-9.28864777e-01 -2.27897614e-01 -5.52403212e-01 6.24809623e-01
2.10660771e-01 -1.22003354e-01 -1.42371774e+00 -5.88135183e-01
-2.52552070e-02 -5.29997766e-01 1.23899007e+00 8.15138817e-02
1.72016883e+00 -1.15061879e+00 -3.33279759e-01 1.08170259e+00
1.14169705e+00 -2.40584966e-02 8.76422405e-01 -2.25864649e-02
8.44339371e-01 7.19892561e-01 -6.16657957e-02 2.99007773e-01
2.00222037e-03 4.43170905e-01 6.69158697e-01 -4.36064787e-03
-2.98765957e-01 -3.53741109e-01 3.85294855e-01 -3.44791621e-01
1.09262392e-01 -2.72239566e-01 -7.52617359e-01 5.65216422e-01
-1.52535439e+00 -1.08562517e+00 1.88831449e-01 2.14408040e+00
5.67003667e-01 -5.61157048e-01 9.68868956e-02 -3.43050569e-01
7.44438171e-01 2.75866061e-01 -5.19007146e-01 -4.77985203e-01
-4.54453789e-02 1.50264502e-01 1.00228310e+00 6.23085856e-01
-1.48806345e+00 9.68236148e-01 6.32627583e+00 6.45805895e-01
-1.25770152e+00 3.91501188e-01 1.18070984e+00 -4.64010924e-01
-7.13278279e-02 -2.55043268e-01 -6.83829606e-01 5.80850363e-01
8.20456445e-01 -1.13147803e-01 1.03888655e+00 1.04244697e+00
-3.79626065e-01 2.81269878e-01 -1.40101373e+00 1.25186574e+00
3.46780837e-01 -1.46186495e+00 1.32818297e-01 5.99414945e-01
5.00377774e-01 2.55461205e-02 7.04855859e-01 1.25401795e-01
3.08236122e-01 -1.61822212e+00 5.31152606e-01 1.15701690e-01
1.12868226e+00 -6.73161685e-01 4.40089494e-01 -4.97760624e-03
-5.17165601e-01 -2.23811060e-01 -5.36206961e-01 -1.70047823e-02
-3.16410512e-01 1.95284769e-01 -8.91848922e-01 -9.08827931e-02
8.63105655e-01 4.74981874e-01 -8.45386624e-01 7.72529125e-01
-1.27703145e-01 7.37473011e-01 -3.89899284e-01 1.62115023e-01
-2.78520305e-02 1.40604004e-01 4.90197271e-01 1.18031478e+00
1.67757213e-01 1.55018166e-01 -2.84425497e-01 9.83310819e-01
-9.20268595e-01 -3.58944058e-01 -1.21227551e+00 -3.38082641e-01
3.98781866e-01 1.16507673e+00 -1.81253836e-01 -1.97016254e-01
-3.81276995e-01 1.13745904e+00 4.90689874e-01 5.24136782e-01
-6.97150648e-01 -8.91942233e-02 1.48185134e+00 1.30766630e-01
5.53206444e-01 -2.32133716e-02 -3.83287728e-01 -1.34826720e+00
-3.87806296e-02 -1.17471015e+00 4.29193139e-01 -4.35482591e-01
-1.37783635e+00 2.31412634e-01 -2.61316985e-01 -6.11807823e-01
5.72589599e-02 -7.16834068e-01 -3.23974192e-01 8.10417712e-01
-1.28815365e+00 -1.20839071e+00 -1.77595392e-01 6.00808561e-01
1.33528367e-01 -3.82537603e-01 7.74528444e-01 3.90730590e-01
-5.39002955e-01 1.17835152e+00 8.79025310e-02 7.13162482e-01
6.59175098e-01 -7.15944827e-01 7.28621423e-01 8.08559060e-01
3.61198723e-01 7.00454593e-01 3.66218358e-01 -4.47560340e-01
-1.64218688e+00 -1.34687829e+00 6.51815712e-01 -8.60583663e-01
3.38071942e-01 -7.15727091e-01 -1.02517331e+00 1.25527763e+00
3.15400511e-01 5.39631903e-01 6.28254712e-01 -1.10396348e-01
-1.12991869e+00 -2.37018824e-01 -1.76508427e+00 6.12824798e-01
1.09958351e+00 -8.23593378e-01 -1.98907614e-01 4.68685269e-01
5.93016982e-01 -7.97866359e-02 -4.32740927e-01 1.38870150e-01
8.09874654e-01 -8.53623152e-01 1.21269822e+00 -1.14235210e+00
3.67565453e-01 -1.94700733e-02 -2.34410778e-01 -9.62237656e-01
-3.35153878e-01 -6.42633796e-01 -2.94124544e-01 1.28935516e+00
2.43981302e-01 -8.92279088e-01 1.00523913e+00 1.01517320e+00
5.86591840e-01 -1.50452733e-01 -1.14400208e+00 -9.30737495e-01
8.99811089e-02 -1.10870287e-01 7.81611383e-01 1.31526017e+00
-3.40409249e-01 -1.01765819e-01 -7.66883075e-01 4.88744557e-01
9.91574347e-01 -2.00338364e-01 8.10139835e-01 -6.52245343e-01
-2.22727343e-01 -3.59109133e-01 -5.57542980e-01 -7.06450164e-01
4.95684654e-01 -1.02126551e+00 -3.75219345e-01 -5.24909019e-01
2.74483442e-01 -2.54533887e-01 -5.33650145e-02 8.76143754e-01
-2.64035761e-02 7.62599766e-01 4.30575311e-01 1.16435707e-01
-1.86475784e-01 3.38598877e-01 1.01695776e+00 -4.99320298e-01
4.06757802e-01 -7.93945640e-02 -1.02700019e+00 3.74170661e-01
9.32778955e-01 -6.31158292e-01 -3.08272000e-02 -7.22035468e-01
-6.02073431e-01 -3.33803922e-01 1.00008070e+00 -7.17585206e-01
-6.17025830e-02 -8.39050189e-02 6.40452206e-01 2.39674136e-01
5.29459774e-01 -9.75324452e-01 1.08867317e-01 5.66924274e-01
-4.72649217e-01 -2.59490162e-01 5.22961020e-01 5.82625151e-01
2.04778910e-01 -1.60191774e-01 1.05852652e+00 -2.47588810e-02
-5.13792992e-01 7.48666346e-01 -1.18721426e-01 -1.76702768e-01
9.98831093e-01 -3.79789472e-02 -6.87331200e-01 -3.71408761e-01
-5.77558100e-01 -1.24936886e-01 6.53819680e-01 3.81680250e-01
6.08356833e-01 -1.22041678e+00 -7.94440508e-01 5.68377733e-01
1.68823034e-01 -6.38612568e-01 2.29362920e-01 3.09653670e-01
-5.20739377e-01 3.49261343e-01 -3.95503521e-01 -3.45899940e-01
-1.80684888e+00 8.92988145e-01 4.92848426e-01 5.28504193e-01
-3.30117166e-01 1.03986549e+00 4.55138981e-01 -2.78097481e-01
3.31396669e-01 -2.77023222e-02 2.52333283e-01 -3.10988694e-01
9.06531751e-01 2.83806473e-01 -3.65801938e-02 -7.35740662e-01
-3.67664695e-01 1.52864670e-02 -4.35408682e-01 6.32928461e-02
1.23146200e+00 5.30840494e-02 -2.62688875e-01 -4.87719744e-01
1.76120675e+00 -1.49557844e-01 -1.37132716e+00 -5.14586680e-02
-2.64961421e-01 -1.16301954e+00 -1.09703161e-01 -7.22787023e-01
-1.31208384e+00 9.64115143e-01 7.37498581e-01 2.62740195e-01
9.89682496e-01 6.85818195e-02 5.24038136e-01 5.04500747e-01
1.72372788e-01 -3.50022465e-01 -3.45359266e-01 1.36714384e-01
1.09866774e+00 -1.44930959e+00 6.75080791e-02 -4.77472872e-01
-4.07436848e-01 6.74457610e-01 6.33414745e-01 -4.91469614e-02
7.11941242e-01 1.70550182e-01 -2.62205712e-02 -9.86264944e-02
-7.12870896e-01 2.55982250e-01 1.80015311e-01 8.89412045e-01
-5.18034473e-02 2.16949672e-01 2.69619912e-01 2.26012319e-01
-1.79629207e-01 -6.65191188e-02 3.97332132e-01 6.35878921e-01
-6.50327876e-02 -9.15605366e-01 -5.74816167e-01 5.32903969e-01
-7.52278805e-01 -8.41440409e-02 -8.02333593e-01 6.15933955e-01
1.38778612e-01 7.02717662e-01 2.53344208e-01 -2.77834535e-01
-1.71473071e-01 2.30344590e-02 6.92089319e-01 -4.69393462e-01
-5.96087456e-01 -7.20258951e-01 1.08740330e-01 -6.66639924e-01
1.12867892e-01 -6.59429729e-01 -3.87188852e-01 -1.04242408e+00
7.42305964e-02 -2.30451182e-01 7.53144145e-01 7.06968546e-01
8.01374495e-01 -2.58603185e-01 5.72066247e-01 -7.52593875e-01
-6.03337348e-01 -7.81704962e-01 -4.31696981e-01 6.10502183e-01
1.00181830e+00 -3.50283176e-01 -5.81309259e-01 2.92961746e-01] | [12.780167579650879, 0.8319613337516785] |
7693e987-06e4-46ce-9707-22a80e466b34 | bootstrapping-deep-music-separation-from | 1910.11133 | null | https://arxiv.org/abs/1910.11133v1 | https://arxiv.org/pdf/1910.11133v1.pdf | Bootstrapping deep music separation from primitive auditory grouping principles | Separating an audio scene such as a cocktail party into constituent, meaningful components is a core task in computer audition. Deep networks are the state-of-the-art approach. They are trained on synthetic mixtures of audio made from isolated sound source recordings so that ground truth for the separation is known. However, the vast majority of available audio is not isolated. The brain uses primitive cues that are independent of the characteristics of any particular sound source to perform an initial segmentation of the audio scene. We present a method for bootstrapping a deep model for music source separation without ground truth by using multiple primitive cues. We apply our method to train a network on a large set of unlabeled music recordings from YouTube to separate vocals from accompaniment without the need for ground truth isolated sources or artificial training mixtures. | ['Prem Seetharaman', 'Bryan Pardo', 'Jonathan Le Roux', 'Gordon Wichern'] | 2019-10-23 | null | null | null | null | ['music-source-separation'] | ['music'] | [ 4.12997007e-01 -5.77348433e-02 1.59077615e-01 -1.37877494e-01
-1.36355853e+00 -1.05276787e+00 1.60819367e-01 -3.18369046e-02
-2.31147319e-01 4.24897373e-01 2.84320831e-01 3.33432965e-02
2.04982623e-01 -2.04369888e-01 -8.42845559e-01 -5.49893677e-01
1.22564554e-01 2.35214591e-01 -6.21381216e-02 6.18483461e-02
-5.22696525e-02 2.36227930e-01 -1.50907671e+00 7.20141351e-01
4.84445542e-01 1.13965106e+00 8.32684152e-03 1.16288912e+00
-7.03137517e-02 7.08954632e-01 -1.06414974e+00 -2.68965304e-01
5.32134831e-01 -9.72583175e-01 -5.93292475e-01 1.72785774e-01
4.62542653e-01 -3.68809044e-01 -1.68205962e-01 1.13701999e+00
5.87405324e-01 4.75803986e-02 6.33867919e-01 -1.25103951e+00
2.96817943e-02 1.10473657e+00 -3.02911818e-01 1.10923439e-01
2.16259211e-01 5.96931949e-02 1.21582365e+00 -7.72493839e-01
1.77247643e-01 8.35126638e-01 7.32803702e-01 3.84245783e-01
-1.29249084e+00 -9.94332790e-01 -4.82527427e-02 9.46741551e-02
-1.24317896e+00 -9.41239893e-01 1.09335744e+00 -4.70542282e-01
5.60017586e-01 3.22682828e-01 7.44470358e-01 1.26599550e+00
-5.48655450e-01 1.02969444e+00 8.16176593e-01 -5.73207021e-01
4.40396249e-01 1.29967421e-01 5.19918874e-02 1.48564607e-01
-1.75784662e-01 -1.15658537e-01 -7.46067166e-01 -3.27303141e-01
8.33253384e-01 -3.03124756e-01 -4.91849720e-01 9.23290674e-04
-1.08871400e+00 4.39135045e-01 2.41058484e-01 4.16913301e-01
-3.70374084e-01 1.99489221e-01 3.13811839e-01 2.45520562e-01
2.57347599e-02 7.38716483e-01 -2.34411925e-01 -1.14371538e-01
-1.56848347e+00 1.55959427e-01 9.90849257e-01 4.77551460e-01
4.02186275e-01 6.13330781e-01 1.66161284e-01 7.51323164e-01
2.81564415e-01 2.14035124e-01 7.26572514e-01 -1.45519722e+00
2.87225544e-01 6.21039011e-02 1.22172378e-01 -5.18899143e-01
-2.75140833e-02 -8.09436381e-01 -5.13612390e-01 4.27343577e-01
8.53932500e-01 -4.48130727e-01 -9.77678537e-01 1.72037196e+00
1.41999945e-01 7.89281845e-01 3.25791202e-02 9.89800334e-01
8.55886936e-01 5.76423824e-01 -4.08193797e-01 -5.09478673e-02
1.03039134e+00 -7.77273476e-01 -5.17169833e-01 -3.99579793e-01
-1.15034506e-01 -9.34123456e-01 8.69812191e-01 1.01178598e+00
-9.77066696e-01 -7.62247622e-01 -1.17424202e+00 1.90452889e-01
1.29401192e-01 1.52408808e-01 2.69243687e-01 7.22635567e-01
-4.90683943e-01 7.62256384e-01 -7.56695151e-01 2.52234787e-01
2.57057697e-01 2.38622189e-01 -3.16079587e-01 3.65283191e-01
-9.26581502e-01 1.03536136e-01 2.56849855e-01 6.07287213e-02
-1.60716379e+00 -5.36044717e-01 -6.63614750e-01 2.42260709e-01
3.62478018e-01 -4.21895802e-01 1.85186732e+00 -1.65201092e+00
-1.73092127e+00 4.85043079e-01 1.65588893e-02 -6.01142108e-01
3.02008927e-01 -4.23317671e-01 -2.86662638e-01 3.56180102e-01
-7.59210363e-02 4.30987805e-01 1.42774963e+00 -1.48609698e+00
-5.20806491e-01 -1.00067584e-02 -4.57592793e-02 2.01232433e-01
1.22221131e-02 1.20118342e-01 -2.35061079e-01 -8.26291502e-01
1.18145026e-01 -9.29076850e-01 1.28309075e-02 -4.98863876e-01
-8.24321628e-01 4.28598195e-01 3.36678684e-01 -8.56269002e-01
8.42009842e-01 -2.42438245e+00 1.77220643e-01 1.47987843e-01
1.65466443e-01 8.08360055e-02 -4.51566428e-01 1.60714343e-01
-2.30883420e-01 -1.01241156e-01 -3.43073279e-01 -7.22858608e-01
1.08333565e-01 -2.86274016e-01 -8.16287577e-01 3.05806369e-01
-1.15568198e-01 3.19837242e-01 -9.05651331e-01 -9.10556912e-02
-1.59036502e-01 5.53659678e-01 -6.77777171e-01 5.48615456e-01
-3.95118982e-01 8.50170016e-01 1.89619660e-01 4.80490595e-01
2.67997921e-01 2.94480652e-01 1.54409602e-01 -9.73903984e-02
1.77070051e-01 8.50428820e-01 -1.62900651e+00 2.00544238e+00
-3.00416768e-01 9.13535237e-01 6.66510761e-01 -6.90639973e-01
6.48707569e-01 8.87014508e-01 5.03760278e-01 2.22303078e-01
4.97068375e-01 3.66727620e-01 4.24302131e-01 -1.35198191e-01
2.40607172e-01 -5.90653956e-01 -2.04990506e-01 7.15388536e-01
5.86552382e-01 -5.75571239e-01 5.57373539e-02 1.06541915e-02
1.05473268e+00 1.93629153e-02 4.62574326e-02 3.40162009e-01
6.51816512e-03 -2.28668913e-01 7.59770572e-01 6.00935876e-01
-1.12332910e-01 1.14083707e+00 2.46128008e-01 -1.00650769e-02
-7.65716374e-01 -1.19650245e+00 1.91992536e-01 1.04596734e+00
-3.77447188e-01 -5.30682862e-01 -1.02419353e+00 -3.92082393e-01
-3.22942764e-01 6.64532304e-01 -2.44174182e-01 6.17801361e-02
-4.25554156e-01 -5.98416150e-01 8.43564928e-01 4.53477174e-01
1.09989271e-01 -1.26455939e+00 -4.39420491e-01 3.42857867e-01
-4.43392426e-01 -1.17276061e+00 -5.06586313e-01 4.68518615e-01
-5.84715784e-01 -1.15585530e+00 -6.52744949e-01 -5.62257171e-01
2.32050553e-01 5.73576242e-02 1.06056368e+00 -4.39386636e-01
-1.82269037e-01 3.63865077e-01 -9.31009352e-02 -9.59843874e-01
-7.43788481e-01 -2.34311402e-01 3.15236270e-01 4.71067816e-01
1.47729784e-01 -1.01852798e+00 -3.71394724e-01 1.73434079e-01
-8.09798837e-01 -9.09269005e-02 3.97828549e-01 3.56163085e-01
5.93502164e-01 3.12456667e-01 5.88823974e-01 -4.63191658e-01
8.30140471e-01 -3.89409363e-01 -2.44642049e-01 -3.09008390e-01
4.13136512e-01 -3.77990395e-01 7.22825229e-01 -8.00679684e-01
-7.02706158e-01 3.70963931e-01 -1.39400035e-01 -8.24285984e-01
-6.80725217e-01 3.41238499e-01 -3.66061449e-01 3.78463686e-01
8.74437511e-01 -1.31134561e-03 -4.92862195e-01 -9.37424183e-01
4.46379304e-01 7.95126438e-01 1.13646948e+00 -4.36756283e-01
5.97302020e-01 3.83666486e-01 -4.56905037e-01 -8.95193577e-01
-9.72950518e-01 -6.44060373e-01 -8.39921951e-01 -2.34783087e-02
7.77139068e-01 -9.88222003e-01 -6.13698721e-01 4.93769705e-01
-1.19760609e+00 -5.97088635e-01 -6.47942543e-01 9.16401267e-01
-5.26565790e-01 2.43231058e-01 -6.41551316e-01 -9.68847752e-01
-3.13493222e-01 -9.60057735e-01 8.49814713e-01 4.36283685e-02
-6.60684824e-01 -3.40135902e-01 3.98620665e-01 3.97031337e-01
-2.10708484e-01 1.00739725e-01 4.53434408e-01 -1.11956370e+00
-5.11327744e-01 -4.40015346e-01 3.45277548e-01 7.91395307e-01
4.22792494e-01 1.31130412e-01 -1.76021719e+00 5.27780363e-03
5.65680683e-01 -3.78880680e-01 8.76708150e-01 3.87639970e-01
9.23908472e-01 -4.72159028e-01 2.30739594e-01 6.94482684e-01
8.87490153e-01 2.74166167e-01 2.66793370e-01 -3.75927091e-02
6.66590750e-01 4.38471466e-01 6.81505818e-03 2.80329138e-01
-5.04626222e-02 2.59258091e-01 4.37097728e-01 1.88580994e-02
-2.85449356e-01 -3.56285810e-01 5.56571007e-01 1.00089216e+00
7.54769295e-02 -1.12144493e-01 -7.79957652e-01 7.84690738e-01
-1.39572251e+00 -1.00803781e+00 2.16276526e-01 2.41845441e+00
1.06720972e+00 3.71569037e-01 5.21931410e-01 9.77144778e-01
6.99009478e-01 -1.00771748e-01 -5.53011000e-01 -4.64124046e-02
6.06141388e-02 3.96687061e-01 -6.71749339e-02 5.45360625e-01
-1.18020558e+00 5.38042068e-01 6.54145575e+00 6.85586393e-01
-1.27535832e+00 -2.18478125e-02 2.09016189e-01 -6.42799854e-01
-1.12525925e-01 -1.09849691e-01 -4.27732646e-01 2.70388514e-01
1.05062413e+00 1.95057854e-01 7.57858038e-01 7.06844747e-01
-2.87537370e-02 -4.50707041e-02 -1.50545788e+00 1.26278257e+00
1.28361657e-01 -1.03069723e+00 -1.71767667e-01 -2.57445067e-01
6.78094208e-01 1.51604354e-01 7.65611157e-02 1.30151495e-01
5.17620623e-01 -1.05120265e+00 1.25412464e+00 2.47679681e-01
5.73054731e-01 -5.34106791e-01 3.02873492e-01 5.79909921e-01
-9.71848488e-01 -1.06517062e-01 8.21211562e-02 -1.09300189e-01
7.97300935e-02 4.90478367e-01 -1.17675579e+00 2.21794322e-01
5.23193479e-01 2.17948347e-01 -6.53736293e-02 1.41717148e+00
-5.94416142e-01 1.22345448e+00 -6.65928483e-01 5.53812683e-01
-5.58482436e-03 1.45786315e-01 9.01754320e-01 1.21299028e+00
2.15626135e-01 -1.12727486e-01 1.02567412e-01 8.10035825e-01
-4.30303127e-01 3.68909389e-02 -4.25070822e-01 -2.70358413e-01
4.05442029e-01 1.16303742e+00 -8.05798173e-01 -3.37887645e-01
-7.95462504e-02 7.68984914e-01 -2.34046608e-01 5.37005961e-01
-5.75527847e-01 -3.68622452e-01 6.29281878e-01 2.57740486e-02
3.67684871e-01 -2.72973984e-01 -3.48768026e-01 -1.15048671e+00
-8.08590651e-02 -1.45539522e+00 2.27243170e-01 -8.25540602e-01
-1.22531247e+00 9.91745532e-01 -2.95879781e-01 -1.49642289e+00
-5.19073069e-01 -3.83723319e-01 -9.60699737e-01 9.66276407e-01
-1.14758277e+00 -6.41802013e-01 -9.61199328e-02 6.32798314e-01
5.82507551e-01 -3.31978738e-01 9.59064364e-01 2.24579588e-01
-4.04750019e-01 3.67478490e-01 -3.05239949e-02 6.62679851e-01
7.79625595e-01 -1.25559795e+00 4.16038573e-01 8.08423281e-01
1.12112463e+00 5.02617359e-01 8.94270420e-01 -3.84333849e-01
-7.98554420e-01 -6.91609204e-01 3.11432779e-01 -4.73793596e-01
6.90444231e-01 -5.34918666e-01 -7.98995674e-01 6.43126667e-01
2.41594404e-01 -1.34498104e-01 1.46272826e+00 2.73672584e-03
-4.97028679e-01 -2.14557201e-01 -7.12409079e-01 3.57481956e-01
4.81955022e-01 -1.00229895e+00 -1.01426709e+00 -1.05212636e-01
4.87603694e-01 -2.85009354e-01 -2.70599991e-01 1.66781709e-01
7.13714719e-01 -1.02408612e+00 8.69157374e-01 -5.89702547e-01
1.88746557e-01 -3.89231890e-01 -3.09101790e-01 -1.35374200e+00
1.48894802e-01 -1.11041343e+00 -1.71525836e-01 1.28833902e+00
2.79791802e-01 -4.28648405e-02 7.94008791e-01 4.65468973e-01
-1.74193710e-01 -4.04537767e-02 -6.95939600e-01 -6.60688519e-01
-2.21569344e-01 -1.14062405e+00 4.69452858e-01 9.08512831e-01
1.52252335e-02 5.99325001e-01 -3.17208797e-01 1.88957915e-01
7.97744632e-01 3.02645028e-01 1.06152046e+00 -1.34664583e+00
-9.05050397e-01 -4.32478726e-01 -1.95529237e-01 -8.98105025e-01
1.73861206e-01 -8.03131461e-01 3.10473204e-01 -1.31591213e+00
-2.81661332e-01 -1.51750147e-01 -5.06775260e-01 3.80154967e-01
3.13466638e-02 6.53929889e-01 4.92942333e-01 3.11075807e-01
-2.38881662e-01 4.64742303e-01 7.63218284e-01 -3.85078162e-01
-5.10549188e-01 5.84828138e-01 -7.03422189e-01 1.39908171e+00
7.96577930e-01 -8.27236950e-01 -4.77263570e-01 -3.71516913e-01
6.09025136e-02 1.54365897e-01 3.32604825e-01 -1.34605157e+00
1.97403476e-01 1.92937762e-01 3.03455532e-01 -2.86271691e-01
7.74819851e-01 -7.69008040e-01 2.51515061e-01 -8.78059715e-02
-5.02761781e-01 -7.62249887e-01 3.60841006e-01 4.67543155e-01
-5.65935314e-01 -5.52963078e-01 5.94606161e-01 -3.05509120e-01
-1.26204878e-01 -4.85475697e-02 -3.18528563e-01 1.20645121e-01
3.93603325e-01 -4.91981953e-02 2.97342271e-01 -7.77832031e-01
-1.18850279e+00 -5.22645831e-01 1.22395501e-01 1.63503617e-01
3.67432356e-01 -1.19616663e+00 -7.28718579e-01 3.05065811e-01
-4.95807886e-01 1.18218057e-01 5.44748008e-02 4.59197432e-01
-2.38714784e-01 1.36868224e-01 -1.66885853e-01 -3.91549110e-01
-1.39448249e+00 5.56202650e-01 4.89675164e-01 1.93958148e-01
-3.80260110e-01 1.11055553e+00 6.04033113e-01 -7.74999559e-02
6.40908897e-01 -6.66800022e-01 -7.22358227e-02 2.32993469e-01
6.53264403e-01 1.22258082e-01 6.29099235e-02 -5.92290401e-01
-1.04245231e-01 2.51612782e-01 5.08737087e-01 -7.65298843e-01
1.37493551e+00 2.33728528e-01 1.32199556e-01 1.05238795e+00
1.02631760e+00 6.93109870e-01 -1.26154304e+00 -2.28902563e-01
-3.12669128e-01 -3.58727485e-01 1.26132881e-02 -6.51551723e-01
-9.33440387e-01 1.21097004e+00 1.76915690e-01 3.77300769e-01
1.17644703e+00 -2.12504968e-01 8.28235686e-01 4.19986784e-01
9.02247578e-02 -8.08771014e-01 3.06411445e-01 1.96788758e-01
9.30349767e-01 -8.96160364e-01 -3.60854685e-01 -7.03368485e-02
-6.45100176e-01 1.13039482e+00 2.14571819e-01 -3.78209889e-01
7.50334322e-01 6.83746457e-01 5.38354456e-01 1.07998187e-02
-3.72395277e-01 -2.81400472e-01 7.25946963e-01 4.32596743e-01
3.94029796e-01 -1.08129315e-01 5.99228382e-01 1.35812926e+00
-6.46351397e-01 -2.00523779e-01 5.10247946e-01 6.99776173e-01
-3.65931809e-01 -9.06307638e-01 -8.14256668e-01 5.80398664e-02
-1.14831400e+00 -3.43131721e-01 -8.94990325e-01 1.04075253e-01
2.61002719e-01 1.09230924e+00 -1.55271366e-01 -5.07153511e-01
1.74563870e-01 5.84089041e-01 5.49393833e-01 -9.44655359e-01
-7.37261474e-01 8.98855984e-01 8.49130738e-04 -1.56310752e-01
-3.41799557e-01 -5.30261576e-01 -1.21869946e+00 4.07910943e-01
-2.01702043e-01 4.17110860e-01 8.19018126e-01 8.58568847e-01
-1.38115650e-02 7.43549228e-01 5.74077368e-01 -1.43866873e+00
-3.04227442e-01 -1.16986334e+00 -6.79562867e-01 2.87928522e-01
8.85795891e-01 -1.12365581e-01 -6.00505292e-01 6.32869184e-01] | [15.437000274658203, 5.545867919921875] |
db524a0d-9866-4223-83ed-d214563b5f11 | melting-pot-2-0 | 2211.13746 | null | https://arxiv.org/abs/2211.13746v5 | https://arxiv.org/pdf/2211.13746v5.pdf | Melting Pot 2.0 | Multi-agent artificial intelligence research promises a path to develop intelligent technologies that are more human-like and more human-compatible than those produced by "solipsistic" approaches, which do not consider interactions between agents. Melting Pot is a research tool developed to facilitate work on multi-agent artificial intelligence, and provides an evaluation protocol that measures generalization to novel social partners in a set of canonical test scenarios. Each scenario pairs a physical environment (a "substrate") with a reference set of co-players (a "background population"), to create a social situation with substantial interdependence between the individuals involved. For instance, some scenarios were inspired by institutional-economics-based accounts of natural resource management and public-good-provision dilemmas. Others were inspired by considerations from evolutionary biology, game theory, and artificial life. Melting Pot aims to cover a maximally diverse set of interdependencies and incentives. It includes the commonly-studied extreme cases of perfectly-competitive (zero-sum) motivations and perfectly-cooperative (shared-reward) motivations, but does not stop with them. As in real-life, a clear majority of scenarios in Melting Pot have mixed incentives. They are neither purely competitive nor purely cooperative and thus demand successful agents be able to navigate the resulting ambiguity. Here we describe Melting Pot 2.0, which revises and expands on Melting Pot. We also introduce support for scenarios with asymmetric roles, and explain how to integrate them into the evaluation protocol. This report also contains: (1) details of all substrates and scenarios; (2) a complete description of all baseline algorithms and results. Our intention is for it to serve as a reference for researchers using Melting Pot 2.0. | ['Joel Z. Leibo', 'Dean Mobbs', 'Igor Mordatch', 'Julia Haas', 'Sukhdeep Singh', 'Michael B. Johanson', 'DJ Strouse', 'Ramona Comanescu', 'Kavya Kopparapu', 'Udari Madhushani', 'Raphael Köster', 'Peter Sunehag', 'Yiran Mao', 'Jayd Matyas', 'Edgar A. Duéñez-Guzmán', 'Alexander Sasha Vezhnevets', 'John P. Agapiou'] | 2022-11-24 | null | null | null | null | ['artificial-life'] | ['miscellaneous'] | [-6.06327057e-02 4.89883035e-01 -1.65360302e-01 8.29722285e-02
-1.48991647e-03 -7.14851260e-01 9.26179588e-01 1.13925554e-01
-7.10507691e-01 1.17494857e+00 -2.28856364e-03 -3.26194167e-01
-7.33521402e-01 -8.13838005e-01 -2.29611218e-01 -9.15470064e-01
-4.12487507e-01 1.01432085e+00 -7.41743483e-03 -7.12719858e-01
3.01048070e-01 1.76146314e-01 -1.71948886e+00 -7.95305818e-02
7.80615628e-01 1.78546190e-01 3.05998147e-01 6.65309191e-01
2.51174480e-01 8.18564594e-01 -7.50695586e-01 -6.17226481e-01
4.39151019e-01 -5.81981480e-01 -9.29720342e-01 -1.08881593e-02
-9.24451709e-01 -1.20782489e-02 1.61045820e-01 7.16664672e-01
6.27648532e-01 -1.03500392e-02 8.57243896e-01 -1.95283186e+00
-6.07350230e-01 9.10511255e-01 -6.07519209e-01 -3.04251909e-02
4.49082255e-01 4.42466438e-01 1.12513328e+00 -1.28700927e-01
8.95431340e-01 1.29849446e+00 4.58483845e-01 8.24996591e-01
-1.09086168e+00 -2.46572137e-01 2.25912452e-01 -1.74630970e-01
-9.72943664e-01 -3.30413878e-01 5.97626865e-01 -3.25417489e-01
1.12672722e+00 6.15572751e-01 1.03919291e+00 1.02655184e+00
1.42611608e-01 5.96567571e-01 1.05516791e+00 -5.29732585e-01
5.66044807e-01 1.94072589e-01 -3.12315375e-01 -1.33929886e-02
6.31599367e-01 3.36825192e-01 -3.12591612e-01 -6.36604071e-01
7.94461191e-01 -3.08406860e-01 2.38206494e-03 -8.60762835e-01
-1.36935961e+00 9.77271616e-01 1.56778440e-01 6.72932804e-01
-7.22931862e-01 -2.82763373e-02 8.85404348e-02 5.55220127e-01
1.29970089e-01 1.13979387e+00 -4.55269575e-01 -2.05887347e-01
-1.57191172e-01 7.57547438e-01 1.16079926e+00 5.21486759e-01
6.46476030e-01 -1.02532841e-01 4.77912277e-01 5.85780740e-01
5.46561062e-01 8.36189538e-02 3.94638717e-01 -1.58768451e+00
1.11574054e-01 6.89599633e-01 6.35590494e-01 -8.92709494e-01
-8.88657689e-01 -3.91162425e-01 -5.46584666e-01 5.66045642e-01
5.49347878e-01 -4.99668926e-01 -5.67414276e-02 2.00897813e+00
3.65727782e-01 -2.13772580e-01 4.96198297e-01 8.88371646e-01
4.93516177e-01 3.83912235e-01 4.83773090e-02 -5.72384119e-01
1.13730085e+00 -8.25929165e-01 -3.48435581e-01 -3.27455431e-01
7.94657171e-01 -3.75374019e-01 7.54586577e-01 2.23568141e-01
-1.39038908e+00 3.24759603e-01 -8.95816207e-01 6.65182829e-01
-5.06459475e-01 -8.74796867e-01 1.20573342e+00 1.04730272e+00
-1.44745970e+00 4.22911137e-01 -4.74247426e-01 -6.64453089e-01
-6.36482537e-02 5.18723309e-01 -1.36671916e-01 5.63171983e-01
-1.25661099e+00 1.05226326e+00 1.70472041e-01 -2.35774055e-01
-5.84992051e-01 -3.14639956e-01 -4.92703199e-01 -1.48442686e-01
5.18189192e-01 -9.34328020e-01 1.17725348e+00 -1.50471067e+00
-1.48545587e+00 1.11827099e+00 4.45676059e-01 -3.12901974e-01
7.92015314e-01 5.72646022e-01 -3.87361981e-02 -1.12040639e-01
4.04223919e-01 7.98978209e-01 6.18586056e-02 -1.70026433e+00
-4.79793549e-01 -2.28246868e-01 4.57746029e-01 8.49991083e-01
-9.97209325e-02 5.30387938e-01 2.82717139e-01 -4.08814549e-01
-3.59497339e-01 -9.55359519e-01 -7.69062638e-01 -6.11765563e-01
-1.24797463e-01 -2.83596963e-01 2.09526733e-01 3.19305539e-01
8.18636179e-01 -1.68162060e+00 4.07088518e-01 1.26442024e-02
4.00300056e-01 -9.55900401e-02 -4.81421798e-01 9.14616764e-01
6.26070648e-02 5.47926188e-01 -1.97888955e-01 -2.09584370e-01
4.86570328e-01 2.41888106e-01 2.78547764e-01 3.09700578e-01
1.62970468e-01 8.39198112e-01 -1.12499344e+00 -3.27509135e-01
5.46880364e-02 -4.49791253e-02 -7.02662647e-01 -9.94057730e-02
-2.50592679e-01 2.88422287e-01 -5.20315230e-01 6.07421577e-01
4.20851409e-01 -2.25781858e-01 7.85923243e-01 8.42207611e-01
-4.28304493e-01 -5.39424270e-02 -1.28039038e+00 1.14387000e+00
9.11472738e-02 2.03479588e-01 4.34879333e-01 -8.53970468e-01
6.11657321e-01 2.62027323e-01 8.00768197e-01 -3.74331981e-01
3.08216721e-01 4.52468604e-01 6.20305240e-01 -3.30405951e-01
6.65095806e-01 -2.85949677e-01 -3.58982086e-01 1.11130667e+00
-1.89164445e-01 -5.51022470e-01 4.31742638e-01 3.16984534e-01
1.38791978e+00 1.51237592e-01 5.51430941e-01 -6.06112838e-01
2.05018908e-01 2.17326656e-01 7.89396226e-01 1.07877648e+00
-7.82491744e-01 1.65924087e-01 8.30205798e-01 -5.90181351e-01
-1.08646548e+00 -9.39943969e-01 3.00784737e-01 1.27938652e+00
6.42988622e-01 -1.32305399e-01 -7.25705504e-01 -2.28323475e-01
5.67871593e-02 7.04738796e-01 -6.98411047e-01 1.46829069e-01
-4.22026843e-01 -1.41501284e+00 3.81837577e-01 -9.15232673e-02
3.27290028e-01 -1.43592381e+00 -1.02772188e+00 2.67247438e-01
-2.03621596e-01 -6.36019707e-01 2.96926767e-01 1.83052525e-01
-3.62239867e-01 -1.07629800e+00 -6.86150670e-01 -6.17577016e-01
3.36005241e-01 3.42407018e-01 1.23621631e+00 4.49919403e-01
-1.00034811e-01 5.72522163e-01 -4.81605440e-01 -7.94964433e-01
-7.23760307e-01 -1.36735499e-01 5.42501926e-01 -2.61024356e-01
3.64263296e-01 -7.91478336e-01 -4.58676994e-01 6.24657631e-01
-7.62149870e-01 -9.92005132e-03 4.58943039e-01 7.57970691e-01
-3.74091834e-01 8.59668776e-02 1.04302704e+00 -2.62973756e-01
1.12338734e+00 -8.34582150e-01 -2.36389071e-01 2.78183162e-01
-4.38039392e-01 -3.47842366e-01 3.65744889e-01 -4.04779255e-01
-9.16816354e-01 -3.30418974e-01 4.98774230e-01 4.28482145e-01
-1.53298020e-01 3.75102550e-01 -1.94078505e-01 9.09840763e-02
8.02220106e-01 -2.40665033e-01 3.66014987e-01 -3.74904014e-02
2.79794425e-01 7.98970938e-01 7.60485278e-03 -9.98937726e-01
5.61509073e-01 2.31800571e-01 -1.63079843e-01 -7.00269222e-01
2.23558620e-01 8.76854435e-02 -3.03235531e-01 -4.72449630e-01
4.42567527e-01 -5.20352125e-01 -1.12142467e+00 7.05743194e-01
-1.02298152e+00 -6.63661540e-01 -5.19186795e-01 5.10668397e-01
-1.00695539e+00 1.78826123e-01 -4.12206948e-01 -1.21425200e+00
3.08534771e-01 -1.24535608e+00 4.05851156e-01 6.00285590e-01
-3.13940942e-01 -1.06362545e+00 3.67185503e-01 4.69499975e-01
5.27724087e-01 2.47296572e-01 6.60887778e-01 -8.20392013e-01
-4.68994141e-01 2.80268639e-01 1.76656395e-01 -4.22062904e-01
6.87560961e-02 2.41862237e-01 -6.08044982e-01 -1.88987181e-01
-4.24854867e-02 -3.97852600e-01 2.27141887e-01 3.94371748e-01
-1.94329768e-01 -2.92330861e-01 -5.07290959e-01 -1.97958171e-01
1.09129441e+00 7.00760901e-01 4.98104453e-01 9.50053275e-01
-2.02414230e-01 1.35855162e+00 4.73124385e-01 7.56203413e-01
8.91916394e-01 6.75925493e-01 6.75540209e-01 2.27965936e-02
6.13762498e-01 2.97684699e-01 3.18740755e-01 3.06726247e-01
-6.10228717e-01 -5.70059955e-01 -1.13675106e+00 6.68528914e-01
-2.20370054e+00 -1.24722779e+00 1.09703876e-01 1.91332400e+00
6.14562511e-01 8.22432637e-02 7.28996813e-01 -8.61241445e-02
9.28611994e-01 3.01110242e-02 -5.31637073e-01 -8.08561623e-01
-5.27271986e-01 -4.18112636e-01 1.92720547e-01 5.09794652e-01
-8.15915048e-01 8.91913295e-01 7.21923971e+00 5.10947406e-01
-3.22819680e-01 8.27211216e-02 7.53252804e-01 -1.99280828e-01
-4.63409990e-01 1.32234082e-01 -1.31122559e-01 2.63773471e-01
7.99953699e-01 -7.64583647e-01 6.95525169e-01 3.90396416e-01
4.58711743e-01 -4.88658160e-01 -8.82819235e-01 2.66698152e-01
-2.72664905e-01 -1.22292054e+00 -3.61608505e-01 3.87501955e-01
6.91149771e-01 4.67827581e-02 -6.66315034e-02 1.93812475e-01
1.24205005e+00 -9.85606253e-01 8.19539547e-01 1.03712223e-01
7.01162964e-02 -7.23529100e-01 7.24140406e-01 6.08794034e-01
-9.07472670e-01 -4.95571047e-01 -1.20959736e-01 -9.55331326e-01
1.11642994e-01 2.81611979e-02 -4.91504312e-01 5.68336427e-01
6.18158460e-01 2.13357300e-01 -1.10276587e-01 8.62589657e-01
1.81207061e-01 -1.32271186e-01 -3.98563683e-01 -5.82282424e-01
4.42866832e-01 -5.83371580e-01 8.53572071e-01 6.57266498e-01
5.51060624e-02 3.11518699e-01 2.92044356e-02 8.82251322e-01
2.42677331e-01 2.28337109e-01 -1.03762972e+00 -1.29860476e-01
7.22652018e-01 1.13630176e+00 -1.14445484e+00 3.68093029e-02
-2.70703703e-01 5.16021729e-01 4.86676544e-02 2.20342949e-01
-7.33531415e-01 -2.24360704e-01 1.00727308e+00 -1.25744402e-01
-3.62801045e-01 4.60759103e-02 -2.65113622e-01 -1.04293978e+00
-3.40322018e-01 -1.11715627e+00 3.15431476e-01 -7.71442354e-01
-1.26988208e+00 3.57739925e-01 2.49638617e-01 -9.58699405e-01
-5.52962542e-01 -2.48765096e-01 -6.82602704e-01 4.75046456e-01
-1.08443141e+00 -1.18584371e+00 1.71262890e-01 2.20364720e-01
1.60400108e-01 -4.18166488e-01 8.08453858e-01 -2.24388972e-01
-6.56357408e-01 1.96264625e-01 2.36016259e-01 -1.96527809e-01
3.15329432e-01 -1.16818464e+00 2.45942965e-01 5.45682311e-01
-3.31524014e-01 6.61760807e-01 9.03498650e-01 -6.84693694e-01
-1.29688454e+00 -3.71240973e-01 6.45361245e-01 -5.77339411e-01
9.30737853e-01 -1.85108602e-01 -1.64037600e-01 5.73848307e-01
6.07411027e-01 -5.74105918e-01 6.42804444e-01 1.59498945e-01
8.42688680e-02 3.07093382e-01 -1.39026940e+00 1.23813450e+00
1.28857028e+00 1.94513366e-01 -5.86906612e-01 3.31473827e-01
6.77045584e-01 1.25736520e-01 -6.47180557e-01 2.15546057e-01
5.34908950e-01 -1.40349293e+00 9.27195609e-01 -7.13091850e-01
2.77652562e-01 -6.16274551e-02 -1.10963389e-01 -1.45717776e+00
-5.13657987e-01 -1.29432595e+00 5.70458591e-01 1.02924013e+00
6.34019971e-01 -1.29533863e+00 8.31529140e-01 9.81307209e-01
1.04612857e-01 -5.30382454e-01 -1.03508222e+00 -1.03229856e+00
6.35398507e-01 5.13064191e-02 8.90836835e-01 1.25950527e+00
7.90528238e-01 8.54405835e-02 -1.82396159e-01 -3.06573033e-01
8.53369713e-01 -5.97423799e-02 8.23389471e-01 -1.39062071e+00
-4.27214831e-01 -1.07646656e+00 -4.26136732e-01 -2.41905347e-01
-4.71291766e-02 -4.24834013e-01 -1.49902329e-01 -1.40715396e+00
4.28642184e-01 -8.18216264e-01 -6.65702224e-02 3.26397270e-01
9.05284956e-02 8.86824578e-02 4.76928174e-01 3.94975960e-01
-6.79822743e-01 2.57112503e-01 9.49183464e-01 1.50841802e-01
-6.71483219e-01 -9.85989496e-02 -1.34332824e+00 9.33477521e-01
1.14423192e+00 -3.25011641e-01 -4.20886546e-01 2.40867548e-02
6.52488947e-01 1.99722648e-01 2.73797631e-01 -6.98153913e-01
1.77928686e-01 -1.10315859e+00 -2.09914148e-01 2.32833639e-01
3.14257443e-01 -5.99328220e-01 7.17296481e-01 6.65576696e-01
-2.41209120e-01 1.38898671e-01 -1.39725372e-01 2.00256705e-01
3.11695725e-01 -5.14571071e-01 5.45102358e-01 -6.57598555e-01
-3.73319060e-01 -3.61277014e-01 -9.02724147e-01 -4.30771261e-02
1.81334686e+00 -5.03315389e-01 -8.55725110e-01 -5.87398052e-01
-5.92592239e-01 5.28378189e-01 1.12926221e+00 2.80918926e-01
1.97608396e-01 -9.54296172e-01 -1.00871909e+00 -2.67764300e-01
-2.93286201e-02 -3.88789922e-01 1.05710149e-01 6.71433985e-01
-5.46632707e-01 2.31478006e-01 -6.76683009e-01 -1.01960422e-02
-1.02645719e+00 6.45792842e-01 6.65560782e-01 -3.53557795e-01
-6.92070648e-02 7.94540763e-01 4.10827428e-01 -7.66982317e-01
-1.60799712e-01 3.53570729e-01 -4.04895991e-01 2.08471209e-01
3.61767530e-01 3.07417750e-01 -5.08912385e-01 -5.96293271e-01
-5.94558954e-01 2.70271391e-01 2.18580559e-01 -6.63944542e-01
1.64901268e+00 -3.56104076e-01 -3.95570368e-01 2.96663821e-01
2.26899743e-01 -7.97031745e-02 -8.22456181e-01 2.27594018e-01
8.99634883e-02 -1.79869726e-01 -5.07681668e-01 -9.34316158e-01
-6.68321490e-01 1.14704333e-01 -7.74456635e-02 1.05217195e+00
9.15484488e-01 3.85179445e-02 -1.71443656e-01 3.80922407e-01
8.18739057e-01 -1.17482042e+00 1.65037096e-01 2.44197085e-01
9.50082839e-01 -9.64530766e-01 1.94289908e-02 -1.73852429e-01
-1.01690543e+00 6.86891079e-01 8.00993085e-01 1.29833892e-01
3.64035279e-01 4.40274090e-01 4.53588292e-02 -1.83862627e-01
-1.18069100e+00 -4.48392510e-01 -7.57907629e-01 1.29755139e+00
7.45093152e-02 3.63195717e-01 -6.65489852e-01 7.49258339e-01
-1.94131568e-01 -1.76078483e-01 1.27420354e+00 1.22531629e+00
-4.51169938e-01 -1.47929990e+00 -6.30249619e-01 3.03103209e-01
-1.86237007e-01 1.50206059e-01 -1.04331875e+00 1.03870618e+00
1.78388014e-01 1.58419180e+00 1.11272380e-01 -2.76545405e-01
8.19469709e-03 -3.13280761e-01 2.87148714e-01 -3.67533624e-01
-1.06361389e+00 -3.73619199e-02 3.41720015e-01 -2.07956225e-01
-9.45122123e-01 -9.75183725e-01 -1.08823073e+00 -8.64184201e-01
-2.88422376e-01 5.77195108e-01 4.59190816e-01 5.56620657e-01
3.66489649e-01 5.83407618e-02 7.31782854e-01 -1.19726038e+00
-3.46348524e-01 -7.20542431e-01 -8.54132175e-01 8.09583366e-02
-3.80327031e-02 -8.35514605e-01 -6.97699547e-01 -4.43278819e-01] | [3.882479667663574, 2.2166242599487305] |
280375d0-4ee7-42f5-bb5d-8c473473e117 | improving-offline-rl-by-blending-heuristics | 2306.00321 | null | https://arxiv.org/abs/2306.00321v1 | https://arxiv.org/pdf/2306.00321v1.pdf | Improving Offline RL by Blending Heuristics | We propose Heuristic Blending (HUBL), a simple performance-improving technique for a broad class of offline RL algorithms based on value bootstrapping. HUBL modifies Bellman operators used in these algorithms, partially replacing the bootstrapped values with Monte-Carlo returns as heuristics. For trajectories with higher returns, HUBL relies more on heuristics and less on bootstrapping; otherwise, it leans more heavily on bootstrapping. We show that this idea can be easily implemented by relabeling the offline datasets with adjusted rewards and discount factors, making HUBL readily usable by many existing offline RL implementations. We theoretically prove that HUBL reduces offline RL's complexity and thus improves its finite-sample performance. Furthermore, we empirically demonstrate that HUBL consistently improves the policy quality of four state-of-the-art bootstrapping-based offline RL algorithms (ATAC, CQL, TD3+BC, and IQL), by 9% on average over 27 datasets of the D4RL and Meta-World benchmarks. | ['Ching-An Cheng', 'Andrey Kolobov', 'Aldo Pacchiano', 'Sinong Geng'] | 2023-06-01 | null | null | null | null | ['offline-rl', 'd4rl'] | ['playing-games', 'robots'] | [-6.23047769e-01 1.77129149e-01 -8.85135531e-01 -4.30435874e-02
-1.09796858e+00 -1.26828015e+00 5.05195498e-01 1.60784140e-01
-6.21840835e-01 1.31920099e+00 2.75411546e-01 -8.44857872e-01
-2.71225780e-01 -5.70930600e-01 -8.32445383e-01 -6.10808194e-01
-5.74775457e-01 9.18142855e-01 3.95702571e-01 -1.48208559e-01
6.52554035e-02 3.86255592e-01 -1.25877583e+00 5.57496548e-02
8.77349854e-01 9.24395144e-01 -2.05313787e-01 7.31959224e-01
4.84783091e-02 1.06366730e+00 -4.01302457e-01 -4.49197888e-01
6.97677910e-01 -4.83189911e-01 -7.80639231e-01 -4.42073226e-01
-3.02128166e-01 -6.01543427e-01 -2.74775207e-01 6.59084439e-01
5.54821789e-01 6.43755913e-01 2.35410795e-01 -1.50602257e+00
-2.59839624e-01 1.27878034e+00 -6.43466234e-01 -1.74917020e-02
2.21991763e-01 5.42392850e-01 1.13495147e+00 -3.56925398e-01
6.75580680e-01 1.35538769e+00 6.32372141e-01 4.36748326e-01
-1.41820848e+00 -5.66997647e-01 3.53324682e-01 4.98835072e-02
-8.66755009e-01 -2.61114061e-01 4.11485851e-01 5.56386039e-02
8.68095636e-01 2.24536493e-01 8.50120962e-01 1.04807174e+00
-2.86197573e-01 1.17331171e+00 1.51299632e+00 -1.20944485e-01
7.85252273e-01 -6.23546168e-02 -4.36712429e-02 6.51323020e-01
3.28768313e-01 5.22781134e-01 -4.27512527e-01 -5.19677877e-01
4.28131402e-01 -3.00507337e-01 -1.44382976e-02 -5.14940917e-01
-1.11281800e+00 9.10265625e-01 1.10570952e-01 -2.26492420e-01
-3.65707189e-01 6.50138378e-01 5.88182926e-01 4.56406713e-01
4.09939975e-01 5.76809824e-01 -6.78575397e-01 -8.10448527e-01
-8.50694418e-01 7.05948114e-01 9.17575240e-01 1.24191737e+00
4.02968705e-01 -2.98705369e-01 -8.41594458e-01 3.00096542e-01
-1.94826767e-01 7.01009512e-01 2.40174741e-01 -1.78465033e+00
6.23387754e-01 1.72722235e-01 1.14684486e+00 -3.21784675e-01
-5.76076567e-01 -3.43717635e-01 -1.17500857e-01 3.34756039e-02
7.27319539e-01 -3.55396450e-01 -5.61325133e-01 1.95834339e+00
3.88001144e-01 6.44388348e-02 -9.30610076e-02 8.05782974e-01
-1.91549107e-01 5.14201939e-01 -8.20461512e-02 -5.64253390e-01
6.38412833e-01 -1.33046997e+00 -5.13329864e-01 1.45467654e-01
7.17146993e-01 -1.87491924e-01 1.45779252e+00 5.51681936e-01
-1.46381116e+00 -8.51292983e-02 -6.53828919e-01 -4.53045405e-03
-2.03786016e-01 2.07845215e-02 9.32540357e-01 5.92098475e-01
-1.10539830e+00 9.17107105e-01 -9.20939028e-01 5.70593029e-03
4.23644394e-01 1.35272384e-01 2.30110779e-01 -1.04221795e-02
-9.89277005e-01 7.74358630e-01 4.30124432e-01 -2.29097486e-01
-1.19281673e+00 -7.28248537e-01 -2.64613241e-01 8.51796009e-03
1.00648737e+00 -3.35265845e-01 1.86321771e+00 -4.97499466e-01
-2.06446910e+00 2.55556822e-01 5.86971045e-02 -1.15270722e+00
1.10881364e+00 -2.35569417e-01 -7.69412937e-03 9.42492113e-02
-1.86100323e-02 5.98518133e-01 6.76849961e-01 -1.12100613e+00
-9.13239419e-01 1.82106242e-01 4.27645564e-01 1.52901098e-01
9.23175737e-02 -2.80856878e-01 -3.42505574e-01 -2.97096312e-01
-5.59725702e-01 -1.16843641e+00 -4.80181575e-01 -3.02833050e-01
-1.53644085e-01 -4.13608730e-01 4.25476521e-01 -4.40580755e-01
1.30900145e+00 -1.91090691e+00 -7.96796195e-03 3.63974571e-01
6.79447055e-02 1.50304124e-01 -2.79874533e-01 5.21401286e-01
5.57152808e-01 3.42084140e-01 -8.78618956e-02 -1.67996496e-01
4.23493087e-01 5.00770092e-01 -6.91493094e-01 5.25691688e-01
-4.50763434e-01 1.07333791e+00 -1.56666768e+00 -2.99315691e-01
1.70600429e-01 -4.83284563e-01 -1.08059359e+00 1.09107502e-01
-8.89577806e-01 5.17657340e-01 -3.95950019e-01 5.19958198e-01
4.86043513e-01 -8.96290913e-02 5.76000214e-01 3.32993597e-01
-2.79195786e-01 2.64520973e-01 -7.70617425e-01 1.71422040e+00
-6.27633333e-01 2.36563966e-01 -1.78250775e-01 -6.37786448e-01
4.09342140e-01 3.38657247e-03 6.18846953e-01 -7.41059840e-01
1.31678924e-01 3.21880102e-01 -2.49771193e-01 -1.66749045e-01
6.74261570e-01 3.63340229e-01 -2.86170334e-01 7.03061581e-01
-3.41329604e-01 -8.46252814e-02 4.08891022e-01 3.37795377e-01
1.26529491e+00 6.76293612e-01 3.54000449e-01 -1.35719553e-01
9.29426104e-02 1.33876964e-01 6.73389077e-01 1.29434872e+00
-5.72742760e-01 -2.28510767e-01 9.11027312e-01 -2.41071671e-01
-9.90772724e-01 -1.18725884e+00 2.39644796e-01 1.41402411e+00
8.67643282e-02 -4.76977497e-01 -7.28443801e-01 -1.09407187e+00
5.57865798e-01 1.13914621e+00 -6.17904186e-01 -5.51607907e-02
-3.69866878e-01 -4.06071782e-01 6.02997065e-01 5.07897794e-01
5.93933940e-01 -8.82357657e-01 -8.33496094e-01 5.34849167e-01
-2.38870040e-01 -1.00174677e+00 -4.79828328e-01 1.79742530e-01
-8.46733809e-01 -1.00093114e+00 -4.61314201e-01 1.80378020e-01
4.06168848e-01 3.66217375e-01 1.12720203e+00 -1.89267740e-01
2.69265294e-01 4.98576105e-01 -6.27120078e-01 -1.68661177e-01
-2.70194501e-01 2.81524807e-01 1.44604713e-01 -3.62184972e-01
-1.23544835e-01 -4.05383050e-01 -6.45206571e-01 3.22029442e-01
-3.05488318e-01 -1.51050732e-01 3.03119719e-01 8.83531809e-01
6.85149252e-01 -1.52351633e-01 7.01860726e-01 -8.77150834e-01
8.65634203e-01 -5.75345755e-01 -1.18944561e+00 3.00893605e-01
-7.89323270e-01 4.60960984e-01 1.04309893e+00 -5.18614531e-01
-8.60597074e-01 -1.51092082e-01 2.50943571e-01 -6.75585806e-01
5.28756082e-01 1.43123493e-01 2.74986446e-01 8.41632858e-02
5.44080615e-01 -6.22119978e-02 -1.72602698e-01 -4.26541597e-01
1.09219790e+00 4.09935415e-01 6.30503178e-01 -1.39029443e+00
6.24324203e-01 4.93122458e-01 -3.54125611e-02 -8.15217197e-02
-1.17625248e+00 -1.90545529e-01 2.11894158e-02 -2.54964292e-01
4.53408509e-01 -6.13833785e-01 -1.45739686e+00 8.32357183e-02
-6.10843062e-01 -1.13693321e+00 -7.98018456e-01 2.83335984e-01
-1.18556905e+00 5.55376261e-02 -3.96474659e-01 -1.11784375e+00
4.87363990e-03 -1.08889091e+00 7.32906342e-01 1.84871852e-01
9.42371115e-02 -7.21551359e-01 2.32211381e-01 6.84983283e-02
4.06438559e-01 1.98827729e-01 7.93521702e-01 -6.12728953e-01
-6.05484009e-01 1.71040818e-01 -1.04748622e-01 6.77414238e-02
-3.39392066e-01 -1.24915965e-01 -7.61836529e-01 -6.91138566e-01
-4.94158745e-01 -7.09180057e-01 6.88020349e-01 3.62178057e-01
1.62078273e+00 -7.90094435e-01 -2.53122479e-01 5.63628316e-01
1.29632771e+00 2.31494412e-01 2.93010771e-01 5.78827798e-01
1.33886546e-01 1.58042058e-01 1.20721126e+00 1.01002240e+00
4.65862453e-01 5.99102795e-01 3.76134723e-01 2.25640848e-01
3.57646197e-01 -7.41144896e-01 5.96285820e-01 3.60940248e-01
-2.69120991e-01 -2.62046874e-01 -5.42191744e-01 5.62140584e-01
-2.45717812e+00 -1.06983161e+00 3.40352595e-01 2.53697324e+00
1.22883523e+00 2.09108055e-01 9.24844205e-01 -1.43088192e-01
3.63687038e-01 1.57476202e-01 -1.12603068e+00 -6.12949312e-01
1.12845287e-01 4.40206110e-01 1.25293362e+00 4.64655191e-01
-7.68811882e-01 1.35422170e+00 6.94197845e+00 9.37201440e-01
-5.84993839e-01 3.59169990e-01 5.71296155e-01 -6.04459524e-01
-3.36057663e-01 3.83561879e-01 -6.14097476e-01 7.39705086e-01
1.22991216e+00 -4.33634669e-01 1.40625644e+00 1.14052486e+00
4.59460795e-01 -2.97718465e-01 -1.12345803e+00 6.37416601e-01
-6.74779475e-01 -1.39309740e+00 -4.50683445e-01 -1.34039596e-02
1.10697198e+00 2.00401291e-01 4.34794724e-02 9.31995273e-01
1.38178372e+00 -6.62775278e-01 9.83900785e-01 4.14815158e-01
6.62766874e-01 -1.22171164e+00 4.24320608e-01 3.66294235e-01
-9.15022433e-01 -6.32934809e-01 -2.97803551e-01 5.02733104e-02
8.17595944e-02 3.82945597e-01 -7.62918532e-01 4.29056913e-01
5.50673485e-01 4.07286018e-01 -2.45276853e-01 9.01482403e-01
-5.64507961e-01 9.34234738e-01 -3.62423301e-01 -1.12863354e-01
6.26448750e-01 -3.33436430e-01 4.20722485e-01 9.34615195e-01
2.19002832e-02 -2.64183078e-02 5.84873796e-01 6.85864925e-01
-2.68409282e-01 -1.98925316e-01 -4.31821853e-01 -3.14296603e-01
1.04534960e+00 1.07240748e+00 -4.99952763e-01 -4.16532964e-01
-4.80697816e-03 6.36045873e-01 4.86199319e-01 4.45665956e-01
-1.49288607e+00 -3.47610533e-01 5.48597157e-01 -2.08667234e-01
5.43439567e-01 -4.37585443e-01 2.52779536e-02 -8.36936176e-01
-2.03491554e-01 -1.01087022e+00 3.95364523e-01 -7.57946789e-01
-1.01237702e+00 1.37607902e-02 1.88758880e-01 -1.24343681e+00
-5.40765882e-01 -2.18692392e-01 -1.68900996e-01 2.40327477e-01
-1.57522380e+00 -6.61522269e-01 7.01272935e-02 4.70361143e-01
1.88605547e-01 1.49247229e-01 3.44493747e-01 -1.17083803e-01
-3.92596006e-01 7.13326871e-01 6.30733311e-01 -3.78568500e-01
6.62247121e-01 -1.33816016e+00 4.83389974e-01 6.24149859e-01
-2.15324730e-01 4.59574461e-01 7.54887521e-01 -4.60862905e-01
-1.64272988e+00 -1.12441039e+00 1.49830878e-01 -3.68110210e-01
1.09968340e+00 -2.81436890e-01 -1.75995871e-01 8.09747696e-01
-1.51415467e-02 1.47568673e-01 5.60609885e-02 1.95532441e-01
-2.94597447e-01 -2.10228264e-01 -1.43886960e+00 8.99704635e-01
1.42838752e+00 -3.15041959e-01 -2.09235430e-01 5.51840842e-01
1.05335462e+00 -5.92438459e-01 -8.35310400e-01 3.27722467e-02
5.44997752e-01 -1.00965369e+00 7.71304488e-01 -9.48117614e-01
-6.29066229e-02 -1.82287246e-01 -1.18462436e-01 -1.42767537e+00
-4.52893861e-02 -1.40963995e+00 -6.76268160e-01 9.41826642e-01
2.95937359e-01 -1.00105190e+00 4.44038689e-01 3.73486727e-01
1.11115679e-01 -8.64528656e-01 -9.65790629e-01 -1.42208743e+00
1.41308039e-01 -4.45646942e-01 1.08382285e+00 4.94743288e-01
5.69594419e-03 -3.49207550e-01 -4.68219787e-01 -2.05821738e-01
7.48250127e-01 3.56382191e-01 9.52703714e-01 -7.07927585e-01
-7.57505834e-01 -3.96856934e-01 4.93067563e-01 -1.18954968e+00
4.47297394e-01 -9.21667278e-01 1.33168800e-02 -1.24683535e+00
8.85574147e-02 -8.82806897e-01 -3.39924991e-01 8.40021849e-01
2.52539307e-01 -2.35613436e-01 4.39279705e-01 2.38648415e-01
-1.19995034e+00 6.50987029e-01 1.28857529e+00 2.13534623e-01
-5.86468518e-01 3.92352492e-02 -6.24422371e-01 5.16099751e-01
9.73213255e-01 -5.23645163e-01 -6.63429976e-01 -2.95775640e-03
4.28157717e-01 4.42534596e-01 1.54696882e-01 -7.43605375e-01
-6.15040101e-02 -6.60284460e-01 -1.84917793e-01 -7.18601167e-01
-6.31175935e-02 -5.53971946e-01 -2.33916298e-01 4.95048374e-01
-6.78597689e-01 1.07469521e-01 1.55814901e-01 7.60340214e-01
5.49758494e-01 8.22628960e-02 7.58053124e-01 -1.56476185e-01
-3.50580931e-01 3.76335323e-01 -2.30822906e-01 7.27065504e-01
1.12054300e+00 2.20148817e-01 -6.46629453e-01 -4.74203318e-01
-5.14583290e-01 8.01975727e-01 6.05350435e-01 1.11239791e-01
8.30838457e-02 -1.20213747e+00 -1.52863771e-01 -2.23392650e-01
-1.72616690e-01 -2.92487860e-01 -2.17619076e-01 1.15466762e+00
-2.60316670e-01 4.56298053e-01 4.93560135e-02 -1.83263615e-01
-4.51852769e-01 8.57375860e-01 2.83624023e-01 -6.85141027e-01
-5.61970532e-01 5.20269871e-01 -5.35052896e-01 -5.10505736e-01
4.32678163e-01 -7.83285558e-01 4.73222286e-01 -1.38864636e-01
2.30098009e-01 9.18121040e-01 -3.32073957e-01 3.09862882e-01
-2.33079627e-01 7.88869038e-02 1.30060852e-01 -6.94779813e-01
1.15844131e+00 4.92916293e-02 2.51159847e-01 3.19177449e-01
6.40822232e-01 2.96045572e-01 -1.83407366e+00 -2.31289789e-02
1.91064954e-01 -4.65723962e-01 3.93052027e-03 -1.22047281e+00
-8.54841352e-01 2.65352339e-01 2.05025092e-01 1.26379222e-01
1.06085563e+00 -3.16454470e-01 9.39773500e-01 7.22657263e-01
1.31028032e+00 -1.59609532e+00 -2.27261275e-01 4.03370738e-01
5.27898550e-01 -8.38479459e-01 1.73816621e-01 4.20452297e-01
-9.25182700e-01 6.56147003e-01 1.84673920e-01 -2.82032877e-01
1.58455238e-01 1.89843521e-01 -6.32527590e-01 3.88930827e-01
-1.17304862e+00 -5.25585890e-01 -4.41843480e-01 4.31975126e-01
-2.24281639e-01 5.22435367e-01 -6.71667278e-01 7.33355582e-01
-1.40916646e-01 4.47836459e-01 5.37760139e-01 1.04742968e+00
-3.57754171e-01 -1.24280906e+00 -6.24621958e-02 2.61839181e-01
-1.59164533e-01 1.36264026e-01 -2.95964569e-01 1.01446259e+00
-2.17545658e-01 1.01729000e+00 -8.29677433e-02 -3.16831350e-01
-4.33827937e-02 -2.16187119e-01 7.12456107e-01 1.93777960e-02
-6.45000160e-01 -1.73771039e-01 4.60431099e-01 -1.08103216e+00
-1.48460448e-01 -6.61593378e-01 -1.56985986e+00 -7.08499312e-01
5.78433201e-02 3.74892980e-01 6.11010432e-01 7.99950957e-01
6.53020442e-01 1.84469864e-01 1.10182714e+00 -6.78063571e-01
-1.34442973e+00 -4.90964293e-01 -4.57515597e-01 4.94184643e-02
1.72139779e-01 -1.05852997e+00 -5.08795083e-01 -6.82745397e-01] | [3.9904587268829346, 2.282942056655884] |
88cbf7ae-d32e-48ba-b47f-b7280b0c235e | beliefppg-uncertainty-aware-heart-rate | 2306.07730 | null | https://arxiv.org/abs/2306.07730v2 | https://arxiv.org/pdf/2306.07730v2.pdf | BeliefPPG: Uncertainty-aware Heart Rate Estimation from PPG signals via Belief Propagation | We present a novel learning-based method that achieves state-of-the-art performance on several heart rate estimation benchmarks extracted from photoplethysmography signals (PPG). We consider the evolution of the heart rate in the context of a discrete-time stochastic process that we represent as a hidden Markov model. We derive a distribution over possible heart rate values for a given PPG signal window through a trained neural network. Using belief propagation, we incorporate the statistical distribution of heart rate changes to refine these estimates in a temporal context. From this, we obtain a quantized probability distribution over the range of possible heart rate values that captures a meaningful and well-calibrated estimate of the inherent predictive uncertainty. We show the robustness of our method on eight public datasets with three different cross-validation experiments. | ['Christian Holz', 'Berken Utku Demirel', 'Paul Streli', 'Valentin Bieri'] | 2023-06-13 | null | null | null | null | ['photoplethysmography-ppg-heart-rate', 'heart-rate-estimation', 'time-series-anomaly-detection', 'time-series'] | ['medical', 'medical', 'time-series', 'time-series'] | [ 2.69354671e-01 2.02237099e-01 -2.84656227e-01 -7.06576943e-01
-9.66473401e-01 -2.39121690e-01 3.69546652e-01 -5.28007559e-03
-2.44106218e-01 8.63523662e-01 2.14237005e-01 1.07511878e-01
7.57474378e-02 -3.70166451e-01 -4.29409057e-01 -8.42754841e-01
-3.32484633e-01 1.47980839e-01 -5.26659824e-02 5.49970329e-01
1.90795019e-01 2.63469279e-01 -8.30585361e-01 -1.39200330e-01
4.71612006e-01 1.26895094e+00 -5.89869797e-01 1.17162645e+00
4.81465399e-01 6.45736694e-01 -7.87414432e-01 -1.65769786e-01
2.10036680e-01 -6.52167201e-01 -3.78807694e-01 -8.33083168e-02
2.47030169e-01 -3.94758433e-01 -4.43179578e-01 7.20122695e-01
6.68186605e-01 -9.14630741e-02 5.52603662e-01 -9.82493401e-01
-8.66571367e-02 5.00152826e-01 -2.99474865e-01 3.46795917e-01
3.26863199e-01 5.67515194e-01 6.67237461e-01 -2.92916328e-01
2.78413773e-01 8.22659671e-01 1.10537720e+00 5.46501040e-01
-1.65723193e+00 -3.70966226e-01 -4.46583807e-01 -2.78233588e-02
-1.56372416e+00 -4.30639416e-01 8.45701993e-01 -3.77011567e-01
7.28030860e-01 1.58054397e-01 8.33368897e-01 1.02583241e+00
8.08960736e-01 4.13209945e-01 1.36538076e+00 -3.55923921e-01
4.15641367e-01 -4.58188877e-02 -4.34886552e-02 5.20881414e-01
1.48137435e-01 3.13264400e-01 -7.46484518e-01 -5.59432030e-01
8.57254326e-01 -1.27780989e-01 -4.07289505e-01 -1.16552792e-01
-1.01554275e+00 5.56593418e-01 -2.70154148e-01 -1.40885174e-01
-6.76464677e-01 6.43838823e-01 2.44822100e-01 -1.51390806e-01
3.85151654e-01 2.87569851e-01 -5.92603981e-01 -5.59401035e-01
-1.12844992e+00 2.46600322e-02 1.43058991e+00 3.47699910e-01
4.74490047e-01 -5.88532314e-02 -5.57934284e-01 3.74052703e-01
6.23411179e-01 7.00030982e-01 2.53612399e-01 -1.48091078e+00
3.78304981e-02 -7.92275593e-02 3.17491084e-01 -6.69447720e-01
-4.69631135e-01 -1.41747758e-01 -7.25028574e-01 5.70243448e-02
6.65062547e-01 -5.10309935e-01 -7.96123028e-01 1.70905960e+00
2.38602057e-01 9.58448112e-01 8.70242715e-02 7.36688197e-01
3.80517721e-01 4.84729260e-01 3.24228168e-01 -6.49883091e-01
1.20229936e+00 -2.81434894e-01 -7.77070224e-01 -5.34409471e-02
-1.00116625e-01 -1.81802511e-01 3.05635035e-01 6.14852607e-01
-9.68321443e-01 -4.62602049e-01 -9.70147014e-01 2.54786998e-01
2.06751302e-01 -1.64373085e-01 4.11549687e-01 1.09685218e+00
-7.55432308e-01 1.35779476e+00 -1.23937726e+00 2.43450534e-02
5.90096116e-02 -2.15663966e-02 1.39928788e-01 4.60960716e-01
-1.53481138e+00 1.03171909e+00 1.82082489e-01 4.23728615e-01
-9.68163133e-01 -9.08904672e-01 -8.59772861e-01 9.02123675e-02
9.39956903e-02 -5.95663786e-01 1.19887912e+00 -1.92107677e-01
-2.17204595e+00 7.46218503e-01 -1.65667653e-01 -8.52309406e-01
6.37203634e-01 -1.44086391e-01 -4.93552357e-01 4.91382390e-01
-5.87134659e-01 4.00038809e-01 1.32997346e+00 -7.67035961e-01
2.53865290e-02 1.36535451e-01 -7.36255229e-01 -1.73206208e-03
1.60800263e-01 -3.21212590e-01 -4.25243407e-01 -1.74723983e-01
2.52530538e-02 -1.14360905e+00 -2.29022384e-01 8.03048387e-02
-3.73378932e-01 -7.54187703e-02 1.60003752e-01 -7.81985879e-01
1.46777701e+00 -2.11572838e+00 -1.56310946e-01 3.48816603e-01
4.74523515e-01 -1.22775368e-01 5.60034871e-01 1.45382255e-01
5.47880493e-02 2.29487807e-01 -3.54087114e-01 -4.90216941e-01
1.10506058e-01 2.05253363e-01 -3.53921860e-01 8.16150069e-01
2.63607483e-02 8.30424607e-01 -8.28885674e-01 -5.67889392e-01
6.52729750e-01 7.76794076e-01 -6.54482096e-03 4.43000734e-01
3.66469137e-02 4.59534168e-01 -1.76711485e-01 4.98606622e-01
3.78173888e-01 -2.53178865e-01 3.25649887e-01 -3.31276178e-01
1.99269071e-01 -1.08927421e-01 -9.34582829e-01 1.64694858e+00
-2.37627864e-01 6.39748454e-01 -6.20203972e-01 -3.90045345e-01
1.20477736e+00 6.34716988e-01 7.57388711e-01 -4.95594330e-02
1.85553983e-01 -1.60116449e-01 -1.72582656e-01 -6.15130126e-01
-7.05168843e-02 -7.79525936e-01 -1.39516756e-01 5.14369428e-01
1.38881907e-01 -3.42818081e-01 -2.59691566e-01 -2.76066542e-01
1.14032364e+00 3.42265457e-01 5.51638305e-01 -9.47427005e-02
2.72324830e-01 -8.07783604e-01 7.79623747e-01 1.14585793e+00
-8.65575194e-01 9.53553021e-01 7.95374334e-01 -7.19387531e-01
-8.84436667e-01 -1.30648565e+00 -5.12375176e-01 1.48334309e-01
-2.10014090e-01 -2.69757509e-01 -3.88486207e-01 -3.26052368e-01
2.91344374e-01 7.33197153e-01 -9.84965086e-01 -2.71746993e-01
-2.95797229e-01 -1.02101552e+00 7.26650059e-01 4.43800181e-01
2.81838655e-01 -1.14288485e+00 -1.07810020e+00 3.96224588e-01
-2.07902819e-01 -1.26540589e+00 -3.50740939e-01 2.19542339e-01
-1.07672954e+00 -8.75731409e-01 -6.79360807e-01 1.87597483e-01
3.66434306e-02 -9.07376766e-01 1.39666629e+00 -5.11938035e-01
-6.28216505e-01 5.93505204e-01 2.69467413e-01 -5.79865098e-01
-4.92112070e-01 -4.61762369e-01 1.79249033e-01 1.38764486e-01
4.51426923e-01 -5.63555062e-01 -1.03993785e+00 5.38008735e-02
-2.83684999e-01 -2.24365667e-01 2.43507817e-01 4.59112614e-01
8.34315240e-01 -2.90237993e-01 4.77559000e-01 -6.37415648e-01
5.78891993e-01 -3.93642128e-01 -7.20137954e-01 3.19347560e-01
-8.47153008e-01 2.62815833e-01 2.06995398e-01 -5.73995829e-01
-9.10728097e-01 2.22408623e-01 2.78687090e-01 -8.19440186e-01
-2.11185694e-01 4.71342385e-01 5.16860127e-01 7.70001858e-02
8.56506169e-01 3.53827357e-01 1.61516041e-01 -1.21345900e-01
3.74860585e-01 5.22632837e-01 9.44664598e-01 -6.98093176e-01
2.72100687e-01 5.21765411e-01 5.38640440e-01 -5.76639116e-01
-7.17235565e-01 -2.71347225e-01 -6.23808324e-01 -4.12909120e-01
8.75344217e-01 -8.59091699e-01 -1.23387909e+00 5.69232345e-01
-7.88569391e-01 -5.63878596e-01 -4.98337090e-01 7.60575593e-01
-1.05883050e+00 6.90576017e-01 -6.97309852e-01 -1.31553352e+00
-7.04733253e-01 -5.72453678e-01 9.56065655e-01 4.89676386e-01
-5.39796293e-01 -1.30857503e+00 7.04934001e-01 7.63104251e-03
3.44210058e-01 8.90496314e-01 3.26101661e-01 -2.51174361e-01
-3.58002216e-01 -5.75627685e-01 1.03741109e-01 5.23814440e-01
1.19221374e-01 1.66429773e-01 -1.22687483e+00 -1.20372064e-01
3.18473190e-01 -6.97337016e-02 6.38527453e-01 1.03401244e+00
1.22194993e+00 -1.70388803e-01 -1.22444354e-01 8.06067348e-01
1.45378458e+00 6.78241625e-02 9.28246737e-01 -1.32981718e-01
2.95243979e-01 1.66130781e-01 1.20899417e-01 1.04354417e+00
2.08001837e-01 1.77067995e-01 5.82808070e-02 3.52350622e-02
2.53009707e-01 -1.50815487e-01 1.42532721e-01 3.27967733e-01
-2.36633748e-01 7.22297505e-02 -7.75931060e-01 3.50581855e-01
-1.55917966e+00 -9.03073430e-01 3.01231474e-01 2.55755472e+00
1.26405656e+00 1.64630532e-01 1.46258309e-01 -1.71619654e-01
7.34059453e-01 2.27624267e-01 -8.02747846e-01 -3.94380093e-01
3.57584357e-01 3.08953524e-01 7.20253408e-01 4.44336921e-01
-1.03430462e+00 2.40683705e-01 8.12041092e+00 -3.89685072e-02
-1.11780846e+00 -2.98094481e-01 9.77427900e-01 -1.26927271e-02
2.04784155e-01 -1.43889517e-01 -7.85603702e-01 7.33835757e-01
1.80988765e+00 -4.90376920e-01 3.34111303e-01 5.78207970e-01
4.25483644e-01 -3.49456459e-01 -1.23062360e+00 1.23475587e+00
9.27404463e-02 -1.22257054e+00 -7.20318258e-01 -2.20341220e-01
3.95195752e-01 1.62420809e-01 -1.16789140e-01 9.57190096e-02
1.17556065e-01 -9.44579720e-01 2.73212224e-01 1.20096231e+00
1.10108161e+00 -3.34559560e-01 6.08588040e-01 -2.53217737e-03
-6.90701485e-01 2.28142947e-01 -1.94933087e-01 6.17807880e-02
3.77095193e-01 1.06762111e+00 -1.09371436e+00 8.61335397e-02
4.62069482e-01 7.26523042e-01 -1.76903337e-01 1.22384274e+00
-5.24814308e-01 9.99079943e-01 -7.49291778e-01 3.50244753e-02
-4.98228222e-01 1.98601875e-02 4.37165350e-01 1.13914728e+00
2.18164667e-01 4.45149869e-01 -1.85390651e-01 1.08781695e+00
4.52467464e-02 -3.51929933e-01 -2.60038048e-01 2.80986521e-02
4.94991750e-01 1.22820497e+00 -6.07845545e-01 -4.63308245e-01
-2.89912730e-01 6.93204641e-01 -1.75713673e-01 3.98452282e-01
-9.98845816e-01 -4.78218943e-01 3.65774930e-01 -2.30055228e-01
1.45267218e-01 -2.47109104e-02 -4.37597424e-01 -1.17960918e+00
-8.61732587e-02 -3.57030213e-01 6.86421156e-01 -8.68457019e-01
-1.26883185e+00 3.88382584e-01 3.34146500e-01 -1.20974314e+00
-6.49622738e-01 -2.37143517e-01 -7.02799499e-01 1.24672103e+00
-1.64948499e+00 -3.28333437e-01 -3.44049394e-01 2.83292890e-01
1.18503690e-01 3.45431775e-01 9.99540269e-01 -2.84696668e-01
-5.21101654e-01 2.65714586e-01 -3.06193888e-01 1.61462009e-01
8.52257192e-01 -1.51028323e+00 4.09720868e-01 7.54581213e-01
-2.04353258e-01 5.65382063e-01 9.12834227e-01 -6.25885606e-01
-1.32189238e+00 -7.88823307e-01 7.42841184e-01 -8.66864145e-01
6.47586465e-01 -2.59694993e-03 -9.54499304e-01 6.18059218e-01
-1.96276680e-02 6.51790738e-01 8.91086340e-01 1.61839984e-02
-1.60656586e-01 -1.24144971e-01 -1.32526398e+00 1.46727180e-02
1.36664048e-01 -7.28474200e-01 -8.44927192e-01 5.94828452e-04
2.00317591e-01 -9.06683326e-01 -1.54421139e+00 3.25202435e-01
1.04657650e+00 -7.46836603e-01 8.02180707e-01 -2.42340416e-01
1.92932978e-01 -1.93502083e-01 2.51881480e-01 -1.19913805e+00
-2.12281689e-01 -1.37943888e+00 -7.10822165e-01 6.52098179e-01
1.82215452e-01 -6.68321013e-01 8.47732723e-01 1.44434953e+00
4.07414228e-01 -5.83585620e-01 -1.09573686e+00 -6.18712425e-01
-1.24925978e-01 -5.61870217e-01 2.74345011e-01 6.65572286e-01
4.25864458e-01 -1.46715358e-01 -7.81320035e-01 3.16226333e-01
1.29601061e+00 1.11506589e-01 2.68158287e-01 -1.11775482e+00
-6.99483633e-01 1.74372062e-01 -4.38125670e-01 -5.87898374e-01
-1.30489841e-01 -2.46000081e-01 5.16742945e-01 -1.00842571e+00
3.31478864e-01 -5.97023219e-02 -8.84890139e-01 4.42472577e-01
-4.14141268e-01 3.67519438e-01 5.19628525e-02 7.19692186e-02
-5.12123942e-01 1.97419479e-01 7.57060766e-01 2.70635843e-01
-5.74830055e-01 1.39697790e-01 -3.28077614e-01 4.53183234e-01
7.97771573e-01 -6.89317286e-01 -1.81879893e-01 3.32209289e-01
1.25509679e-01 9.48620081e-01 2.96077996e-01 -1.14433372e+00
1.05502538e-01 -1.51261702e-01 7.51333475e-01 -4.40687567e-01
3.46586794e-01 -5.03796995e-01 4.20666993e-01 5.75050592e-01
-4.98196065e-01 -5.38742244e-01 2.20904261e-01 7.90169358e-01
5.30600920e-02 -3.15449107e-03 9.41616297e-01 -9.52223688e-02
-2.87186772e-01 3.03943574e-01 -2.47247085e-01 1.56310722e-01
6.18672788e-01 -6.19442165e-02 -5.28119616e-02 -6.48473322e-01
-1.07136977e+00 6.39947131e-02 -7.61613101e-02 1.55456528e-01
5.05533040e-01 -1.03375173e+00 -7.67854035e-01 2.42749900e-01
4.26230989e-02 -4.69664395e-01 3.87866348e-01 9.16875780e-01
-4.26485062e-01 2.22872332e-01 -1.78585410e-01 -9.10293758e-01
-8.77794087e-01 3.79905760e-01 9.53015625e-01 -8.84954184e-02
-8.07417512e-01 4.46564525e-01 -4.17069763e-01 2.57702023e-01
1.19506717e-01 -6.71809375e-01 5.06569929e-02 -3.45029801e-01
5.76537430e-01 4.06913906e-01 -3.24021995e-01 1.85900591e-02
-3.47935677e-01 4.85920548e-01 5.30689776e-01 -2.06635445e-01
9.05446410e-01 -4.53421801e-01 1.67503357e-01 7.91763008e-01
1.02449787e+00 -3.02580088e-01 -1.84321702e+00 -1.69596314e-01
-2.01090753e-01 -4.06021059e-01 1.16651215e-01 -1.04601562e+00
-8.10866058e-01 6.65612459e-01 8.97901714e-01 1.75480828e-01
1.10507619e+00 -2.89813608e-01 4.25581902e-01 2.34530389e-01
6.52455837e-02 -1.00035763e+00 -3.03923666e-01 -5.07216007e-02
4.39402044e-01 -1.22920597e+00 4.59935099e-01 -1.86070368e-01
-8.25654984e-01 1.41804302e+00 2.21613124e-02 -1.46723509e-01
9.63313520e-01 4.30054814e-01 4.68908995e-01 9.98385176e-02
-1.04354596e+00 2.89558083e-01 2.84255832e-01 5.55594087e-01
2.89250821e-01 2.70895779e-01 -1.20846964e-01 3.66163582e-01
1.81490898e-01 7.15311348e-01 6.97881103e-01 6.05160415e-01
-2.41042078e-01 -5.02052367e-01 -1.55557379e-01 3.36883008e-01
-8.10595572e-01 3.45985629e-02 3.43556374e-01 3.37128639e-01
-3.63039434e-01 6.92459583e-01 6.30273521e-02 -1.16108015e-01
4.90058884e-02 5.79236150e-01 5.02567172e-01 -4.04736549e-01
-1.24001712e-01 2.12789223e-01 -4.64944728e-02 -7.38608778e-01
-3.50897044e-01 -7.83542395e-01 -1.01408815e+00 7.33428672e-02
2.66620647e-02 -1.63907073e-02 7.37587214e-01 8.41867208e-01
2.09943220e-01 2.18819514e-01 7.59277642e-01 -6.69467151e-01
-7.11557984e-01 -9.57379460e-01 -7.56251276e-01 4.41016048e-01
5.76262891e-01 -1.68952912e-01 -8.46212268e-01 3.15275341e-01] | [13.928325653076172, 2.8725457191467285] |
fea84ef2-1388-4eed-9059-1b8ee0fbc291 | diffusion-models-already-have-a-semantic | 2210.10960 | null | https://arxiv.org/abs/2210.10960v2 | https://arxiv.org/pdf/2210.10960v2.pdf | Diffusion Models already have a Semantic Latent Space | Diffusion models achieve outstanding generative performance in various domains. Despite their great success, they lack semantic latent space which is essential for controlling the generative process. To address the problem, we propose asymmetric reverse process (Asyrp) which discovers the semantic latent space in frozen pretrained diffusion models. Our semantic latent space, named h-space, has nice properties for accommodating semantic image manipulation: homogeneity, linearity, robustness, and consistency across timesteps. In addition, we introduce a principled design of the generative process for versatile editing and quality boost ing by quantifiable measures: editing strength of an interval and quality deficiency at a timestep. Our method is applicable to various architectures (DDPM++, iD- DPM, and ADM) and datasets (CelebA-HQ, AFHQ-dog, LSUN-church, LSUN- bedroom, and METFACES). Project page: https://kwonminki.github.io/Asyrp/ | ['Youngjung Uh', 'Jaeseok Jeong', 'Mingi Kwon'] | 2022-10-20 | null | null | null | null | ['image-manipulation'] | ['computer-vision'] | [-2.59065658e-01 4.34423238e-02 7.79288486e-02 -3.14535856e-01
-4.02645141e-01 -6.19044006e-01 1.02092505e+00 -5.21981657e-01
-1.02556013e-01 6.03899717e-01 4.00153279e-01 -7.93739036e-03
-2.15842187e-01 -9.61296916e-01 -5.47579110e-01 -8.36594105e-01
2.98467815e-01 6.42449081e-01 5.50207086e-02 -1.43179342e-01
1.30267337e-01 2.19959632e-01 -9.75586414e-01 1.60694063e-01
1.09889090e+00 6.64823949e-01 3.75699878e-01 5.66875100e-01
1.75747238e-02 8.10605586e-01 -2.62549311e-01 -5.02876580e-01
3.01656455e-01 -6.49481595e-01 -6.58415616e-01 8.42773542e-02
-1.49122262e-02 -2.71606296e-01 -5.58184326e-01 1.08597910e+00
6.23959899e-01 1.12410840e-02 8.66039872e-01 -1.54776907e+00
-1.41381848e+00 5.82563579e-01 -5.19669056e-01 1.66465476e-01
-4.49569896e-02 4.54380602e-01 8.71093988e-01 -9.01423097e-01
9.30079401e-01 1.34187901e+00 5.48610806e-01 6.65550172e-01
-1.34136331e+00 -6.98970675e-01 1.74460351e-01 5.57783572e-03
-1.26557112e+00 -4.83347535e-01 6.17879987e-01 -7.64582276e-01
6.60893261e-01 -3.50996144e-02 7.26204455e-01 1.51231909e+00
3.66305053e-01 9.89838958e-01 1.12827742e+00 -2.76542492e-02
5.13962030e-01 -7.24371895e-02 -1.15219578e-02 6.46169782e-01
-8.28007907e-02 -1.97551437e-02 -7.47312367e-01 -1.14145994e-01
1.20828891e+00 1.52535483e-01 -2.37932816e-01 -4.95275885e-01
-1.29429698e+00 1.02149367e+00 2.74617285e-01 1.39207333e-01
-4.78924364e-01 3.40916157e-01 1.44869983e-01 4.30337369e-01
7.53369451e-01 3.28326583e-01 -2.92005509e-01 -3.75523716e-01
-1.18772280e+00 3.27840477e-01 5.90894103e-01 1.38338745e+00
4.96186286e-01 2.15093777e-01 -4.73762214e-01 8.59668374e-01
2.98431695e-01 4.68332350e-01 4.55636621e-01 -1.15555930e+00
1.81198835e-01 3.20651144e-01 -8.25098529e-02 -8.09452474e-01
-1.59851640e-01 -3.47723365e-01 -1.14111626e+00 -1.03236549e-01
-3.63264307e-02 -9.82055664e-02 -1.27254224e+00 1.88447165e+00
2.67939270e-01 -5.91853857e-02 -1.16523273e-01 7.06264973e-01
6.44486666e-01 7.52448678e-01 2.14252204e-01 -1.05484858e-01
1.15173066e+00 -1.25335205e+00 -7.33554184e-01 -1.63338482e-01
2.18928590e-01 -6.64515138e-01 1.13706601e+00 2.42037103e-01
-1.38138950e+00 -4.53977644e-01 -7.02326834e-01 -2.45683745e-01
-1.48032144e-01 7.48794675e-02 8.02547514e-01 3.13879877e-01
-1.39816403e+00 5.97390413e-01 -1.05879319e+00 -4.66565132e-01
5.39698184e-01 6.11045100e-02 -1.32601306e-01 -6.94175512e-02
-1.11015546e+00 4.28572804e-01 1.27045512e-01 -9.54654813e-03
-1.30619097e+00 -8.78202915e-01 -6.31269038e-01 -3.60346474e-02
7.65733719e-02 -1.19399083e+00 1.17244673e+00 -8.62678289e-01
-1.76737332e+00 8.11135292e-01 -1.69819016e-02 -3.89252990e-01
1.07463658e+00 -3.20573866e-01 -2.44695827e-01 7.81596601e-02
3.68432194e-01 1.02201056e+00 1.28423953e+00 -9.15013969e-01
-9.54426229e-02 -5.06733835e-01 -1.66830063e-01 3.25687796e-01
-2.98413306e-01 -8.86782855e-02 -7.18230844e-01 -1.03851092e+00
2.04050153e-01 -1.05442250e+00 -1.73170939e-01 4.22040485e-02
-5.42037725e-01 -9.01133642e-02 7.88499951e-01 -7.43450522e-01
1.14066803e+00 -2.42809868e+00 5.32391727e-01 7.23302411e-03
4.68712091e-01 3.05012707e-02 -3.10829818e-01 5.82966864e-01
1.68890476e-01 7.43637681e-02 -4.09421951e-01 -5.93296528e-01
2.57363051e-01 4.30842489e-01 -4.16307569e-01 3.55913669e-01
6.91310018e-02 1.27902913e+00 -8.91897082e-01 -3.02525014e-01
-4.05323543e-02 6.91751361e-01 -7.22648978e-01 1.68139726e-01
-3.19866061e-01 5.55348754e-01 -4.18084919e-01 5.88860631e-01
5.54259360e-01 -6.02828443e-01 7.10431337e-02 -3.75262238e-02
3.59217077e-03 6.23342842e-02 -1.05548501e+00 1.94926524e+00
-2.07764655e-01 5.31439960e-01 1.47669733e-01 -3.46084267e-01
8.22145820e-01 1.57085985e-01 3.49102974e-01 -8.99306595e-01
-1.86908454e-01 2.87580676e-02 -4.13406760e-01 -1.27847731e-01
7.62618780e-01 -4.00159955e-02 8.91839564e-02 5.78994513e-01
2.01317027e-01 -1.82671294e-01 2.82271594e-01 4.42823797e-01
9.85374808e-01 2.18173862e-01 -6.91239461e-02 -5.02040684e-01
-2.48012498e-01 -2.00131223e-01 5.36420465e-01 7.65653729e-01
-2.30799124e-01 7.51373470e-01 5.48389256e-01 -2.55586207e-01
-1.37895095e+00 -1.35045409e+00 -1.01607144e-01 1.20480049e+00
2.00831965e-02 -3.76943648e-01 -7.60521889e-01 -3.63412768e-01
6.00715913e-03 7.70815253e-01 -6.61896110e-01 -2.75119305e-01
-3.27235490e-01 -9.00844455e-01 5.30223370e-01 6.10928535e-01
7.42131472e-01 -1.06738389e+00 -2.42805481e-01 2.50066668e-01
-9.80449021e-02 -7.93929875e-01 -9.26709175e-01 -1.19767077e-01
-9.47130501e-01 -5.26095569e-01 -9.93400633e-01 -4.51814473e-01
6.83183849e-01 9.02764797e-02 1.15203166e+00 -3.81444126e-01
-1.76453814e-01 3.45693707e-01 -1.03933014e-01 -3.15210596e-02
-3.96717489e-01 2.17759728e-01 5.03482521e-02 -1.01663806e-01
6.25673905e-02 -6.49170518e-01 -8.35039258e-01 3.76500905e-01
-9.34797287e-01 4.78107184e-01 4.21818793e-01 7.69119799e-01
6.27609491e-01 -4.04418856e-02 3.38806748e-01 -9.56209362e-01
8.58593762e-01 -5.78277290e-01 -4.54417199e-01 9.94156376e-02
-8.75619292e-01 -5.58227114e-02 1.60293356e-01 -3.96468878e-01
-1.22367537e+00 -3.52713883e-01 -1.35726044e-02 -4.97827649e-01
5.13114743e-02 1.70415193e-01 -9.61713344e-02 3.83740872e-01
5.12597740e-01 4.37789679e-01 -1.58839533e-03 -4.99932706e-01
8.19868803e-01 3.19441199e-01 3.07571441e-01 -5.34935355e-01
6.73395216e-01 7.65916407e-01 -4.50661987e-01 -7.97720194e-01
-6.03064954e-01 -1.18994087e-01 -6.42922401e-01 2.21370131e-01
1.01889706e+00 -1.21481264e+00 -1.82886064e-01 8.47259820e-01
-1.01557755e+00 -6.94012046e-01 -4.96521533e-01 9.33567956e-02
-6.90413833e-01 9.39685926e-02 -1.05453193e+00 -4.62536126e-01
-5.79496861e-01 -1.13353825e+00 1.05077887e+00 1.50465056e-01
-4.08069223e-01 -1.21895957e+00 1.46533102e-01 2.48523340e-01
6.77305162e-01 -3.71127971e-03 7.84687936e-01 -2.13441193e-01
-7.92189181e-01 3.42503995e-01 -2.48660356e-01 3.38366032e-01
2.93283314e-02 2.47869134e-01 -6.87051773e-01 -4.35289562e-01
-5.40026575e-02 -1.01626180e-01 9.84013975e-01 7.05965638e-01
1.08041131e+00 -5.05049944e-01 -2.72103727e-01 1.00554085e+00
1.13926494e+00 1.50747880e-01 7.82219470e-01 3.28814596e-01
7.03017950e-01 1.30945578e-01 9.53513831e-02 4.74055439e-01
5.87624967e-01 5.87557852e-01 -4.22805808e-02 -1.32091179e-01
-3.73484701e-01 -4.64220166e-01 5.02726018e-01 1.18753350e+00
-5.70194051e-02 -3.02682370e-01 -7.93859959e-01 5.96304476e-01
-1.88919282e+00 -9.84352946e-01 3.01122572e-02 1.67325163e+00
9.12565947e-01 3.93374972e-02 -3.60525660e-02 -4.32867676e-01
6.01267993e-01 3.50065768e-01 -6.39421642e-01 -1.86521202e-01
-1.97222307e-01 -2.45492101e-01 2.99782664e-01 4.06181455e-01
-8.49721491e-01 1.03294015e+00 6.47121811e+00 9.71180677e-01
-9.02022123e-01 4.99430597e-01 7.43201315e-01 -2.84388244e-01
-6.86059773e-01 -5.47167473e-02 -8.37737799e-01 7.47612834e-01
6.90665364e-01 -9.49674174e-02 6.09896958e-01 8.72538328e-01
3.12680334e-01 2.20133826e-01 -1.03043854e+00 8.90287995e-01
-1.69590041e-01 -1.49096262e+00 1.87607184e-01 8.11469778e-02
1.08548236e+00 2.08621129e-01 6.26738131e-01 2.89462090e-01
8.07044208e-01 -8.64728928e-01 9.55620170e-01 6.43748045e-01
9.52709258e-01 -5.73062301e-01 1.38119355e-01 3.14438283e-01
-7.71776617e-01 7.73998499e-02 -3.64303827e-01 3.21825802e-01
3.49954128e-01 7.75392592e-01 -4.22414184e-01 3.38217735e-01
8.01988423e-01 9.47706938e-01 -4.05523777e-01 4.84803647e-01
-5.01119733e-01 5.51321566e-01 -1.17452130e-01 4.32993591e-01
3.11299950e-01 -5.28371096e-01 5.73171139e-01 1.11377990e+00
5.02979457e-01 4.00196435e-03 -1.46767810e-01 1.34989882e+00
-1.81749403e-01 -3.46798211e-01 -5.20052314e-01 -3.92038226e-01
5.01022935e-01 1.07452583e+00 -8.62311542e-01 -3.12028676e-01
1.79735036e-03 1.45082200e+00 2.54823357e-01 6.31837964e-01
-1.00350714e+00 9.49330553e-02 8.81990790e-01 2.25082412e-01
2.73949385e-01 -5.14059901e-01 -2.69579679e-01 -1.33128452e+00
-1.72807187e-01 -8.29463780e-01 2.09210664e-01 -9.46587145e-01
-1.49417293e+00 5.99086404e-01 -3.24172080e-02 -7.45856106e-01
-3.59524675e-02 -1.04996689e-01 -4.38929677e-01 9.03603494e-01
-1.24703419e+00 -1.46580279e+00 -3.72373819e-01 6.27112806e-01
7.72437513e-01 -1.72811463e-01 5.15449464e-01 4.17242408e-01
-6.42708302e-01 3.38301599e-01 2.89664775e-01 -3.72377746e-02
6.48835301e-01 -1.27105975e+00 8.79748821e-01 7.71355331e-01
1.06650956e-01 8.50057244e-01 6.63456976e-01 -8.40861320e-01
-1.22208548e+00 -1.27794349e+00 6.22111201e-01 -7.22381771e-01
7.02651918e-01 -4.29695040e-01 -8.35162342e-01 8.87543797e-01
3.20510775e-01 -3.58170152e-01 4.86431658e-01 -7.91941881e-02
-4.05716836e-01 1.61268815e-01 -8.51556182e-01 7.38162518e-01
1.37944818e+00 -6.72071338e-01 -3.18482488e-01 4.68148112e-01
9.63787138e-01 -4.51059401e-01 -9.28604126e-01 7.05021024e-02
3.44445944e-01 -9.25074816e-01 9.81164932e-01 -3.63225013e-01
6.89012647e-01 -5.90049587e-02 -4.80119921e-02 -1.52697206e+00
-7.70051241e-01 -8.57352853e-01 -2.79579401e-01 1.25622296e+00
1.50410503e-01 -6.76226199e-01 6.24108434e-01 6.77068889e-01
-1.73479319e-01 -6.16271734e-01 -5.75689852e-01 -7.85865247e-01
1.31722644e-01 -1.84601545e-01 7.74072170e-01 1.11087871e+00
-5.20500124e-01 1.91564560e-01 -4.36713487e-01 -8.91505927e-02
8.47424507e-01 6.96818829e-02 6.71130478e-01 -9.01797175e-01
-4.97743428e-01 -6.37456596e-01 -1.39826350e-02 -1.19757211e+00
-1.47703111e-01 -8.24384272e-01 -3.70530546e-01 -1.74587226e+00
3.08252275e-01 -5.55890024e-01 -8.05185586e-02 4.93010491e-01
-1.10731967e-01 9.50351581e-02 1.79805979e-01 6.99527919e-01
-5.20768583e-01 9.13443685e-01 1.62335134e+00 5.01249824e-03
-2.25004211e-01 -2.36378849e-01 -6.40682578e-01 5.29596150e-01
8.02459776e-01 -4.00324911e-01 -6.93351090e-01 -8.74025881e-01
3.20138395e-01 -1.19318217e-01 4.99417365e-01 -6.03176475e-01
3.00466359e-01 -2.11691707e-01 2.49282718e-01 -3.70040059e-01
3.01200986e-01 -3.37881774e-01 7.67888606e-01 3.43059510e-01
-4.46324199e-01 3.77326101e-01 -1.63156286e-01 6.31447434e-01
-1.50129452e-01 2.22589791e-01 9.70110059e-01 -2.73608506e-01
-9.00770307e-01 6.30249500e-01 -4.74375784e-02 2.49790624e-01
1.02363122e+00 -1.84487894e-01 -5.62088251e-01 -3.47445786e-01
-8.98813069e-01 2.81333804e-01 7.86709785e-01 5.84719718e-01
6.40455067e-01 -1.38967514e+00 -6.51612759e-01 4.79243577e-01
1.33047635e-02 3.94804552e-02 5.96725702e-01 8.87810409e-01
-5.33828557e-01 8.46283212e-02 -3.52859646e-01 -4.57257181e-01
-7.81478405e-01 4.10743415e-01 1.24243207e-01 -2.53646761e-01
-9.32051122e-01 1.09783375e+00 4.84822720e-01 -3.89792532e-01
2.78949682e-02 -1.53895512e-01 2.80244172e-01 1.00249641e-01
2.19161183e-01 3.08495432e-01 -1.46145478e-01 -3.65804285e-01
1.05466796e-02 1.70435563e-01 -1.12063549e-01 -1.48777694e-01
1.55882645e+00 -3.55451703e-01 -2.60032952e-01 4.38563317e-01
1.00195801e+00 -3.50534052e-01 -1.67524505e+00 -1.33718446e-01
-2.80932844e-01 -3.76463532e-01 8.42350125e-02 -6.56580269e-01
-1.04462016e+00 8.18464220e-01 4.89642680e-01 1.05860032e-01
9.74004567e-01 1.51478291e-01 8.06435943e-01 -1.93351969e-01
2.59135723e-01 -1.25843585e+00 3.62534791e-01 5.43929815e-01
1.01172817e+00 -9.55971718e-01 -2.61760890e-01 1.52217653e-02
-1.04026043e+00 5.54506719e-01 4.23054188e-01 8.98332372e-02
6.77830935e-01 1.78891197e-01 -1.06061444e-01 -4.81283277e-01
-8.93439293e-01 2.07925081e-01 3.66314016e-02 4.52187449e-01
2.75360793e-01 3.11684787e-01 -3.72575782e-02 3.15375179e-01
-2.65994012e-01 5.36331125e-02 3.59800965e-01 7.99391925e-01
-2.13031858e-01 -9.10899699e-01 1.82610340e-02 3.67471933e-01
-2.53816396e-01 -2.78805375e-01 -6.10950105e-02 6.51574790e-01
-7.94068575e-02 4.36172396e-01 1.93241268e-01 2.32877582e-03
1.44532666e-01 8.54203999e-02 3.78580689e-01 -5.61204314e-01
-3.11509341e-01 2.88433611e-01 -2.07437009e-01 -6.39180422e-01
-2.66305476e-01 -6.29030645e-01 -1.03954732e+00 -6.54641926e-01
5.89619987e-02 7.00256750e-02 5.57344556e-01 5.62185228e-01
6.54183865e-01 5.74501455e-01 2.77876645e-01 -5.07970095e-01
-6.21150792e-01 -9.21783805e-01 -9.06344295e-01 5.95486879e-01
5.25857992e-02 -6.39506102e-01 -1.77954122e-01 3.67527276e-01] | [11.348504066467285, -0.24705089628696442] |
127e8757-32f5-4541-88cb-23e4393e8bfd | learning-the-enigma-with-recurrent-neural | 1708.07576 | null | http://arxiv.org/abs/1708.07576v2 | http://arxiv.org/pdf/1708.07576v2.pdf | Learning the Enigma with Recurrent Neural Networks | Recurrent neural networks (RNNs) represent the state of the art in
translation, image captioning, and speech recognition. They are also capable of
learning algorithmic tasks such as long addition, copying, and sorting from a
set of training examples. We demonstrate that RNNs can learn decryption
algorithms -- the mappings from plaintext to ciphertext -- for three
polyalphabetic ciphers (Vigen\`ere, Autokey, and Enigma). Most notably, we
demonstrate that an RNN with a 3000-unit Long Short-Term Memory (LSTM) cell can
learn the decryption function of the Enigma machine. We argue that our model
learns efficient internal representations of these ciphers 1) by exploring
activations of individual memory neurons and 2) by comparing memory usage
across the three ciphers. To be clear, our work is not aimed at 'cracking' the
Enigma cipher. However, we do show that our model can perform elementary
cryptanalysis by running known-plaintext attacks on the Vigen\`ere and Autokey
ciphers. Our results indicate that RNNs can learn algorithmic representations
of black box polyalphabetic ciphers and that these representations are useful
for cryptanalysis. | ['Sam Greydanus'] | 2017-08-24 | null | null | null | null | ['cryptanalysis'] | ['miscellaneous'] | [ 5.24798691e-01 2.64804401e-02 5.08557633e-02 1.10804223e-01
-6.42010033e-01 -7.98962474e-01 9.64773238e-01 -1.58980519e-01
-7.78524458e-01 5.49307287e-01 2.97217853e-02 -1.23908913e+00
2.65046448e-01 -8.05988908e-01 -1.13566577e+00 -1.12826574e+00
-2.83660740e-01 4.95621920e-01 -6.08329415e-01 -5.93387127e-01
3.79119068e-01 5.09643376e-01 -1.19137990e+00 8.14466059e-01
2.89099723e-01 8.74760687e-01 1.76783592e-01 1.31737435e+00
-1.26328126e-01 8.43784928e-01 -9.12149727e-01 -7.05417991e-01
4.45095927e-01 -2.80045122e-01 -1.02612269e+00 -8.24721754e-01
4.96818237e-02 -1.11771517e-01 -8.95347834e-01 8.43850315e-01
2.91522294e-01 -3.10486495e-01 6.22950017e-01 -8.75405371e-01
-6.47469580e-01 6.61593676e-01 -1.39647856e-01 3.41253608e-01
-4.98933867e-02 3.93608481e-01 9.87262607e-01 -4.68161851e-01
6.84721112e-01 1.18207419e+00 6.60926700e-01 9.11186218e-01
-7.84878016e-01 -9.97901022e-01 -5.25155962e-02 3.27192158e-01
-1.05026639e+00 -2.83033222e-01 1.36601508e-01 -9.23280492e-02
1.35372162e+00 3.52345616e-01 7.10397482e-01 1.45637429e+00
8.87857795e-01 9.76247013e-01 1.14604855e+00 -2.71910846e-01
-1.85020581e-01 -1.60984769e-01 4.68486585e-02 7.79715836e-01
1.01702400e-01 5.62934816e-01 -3.85330915e-01 -2.44193655e-02
4.65669066e-01 5.56771755e-02 -2.26464212e-01 2.94998825e-01
-1.37202024e+00 7.79364288e-01 5.57820439e-01 2.78340071e-01
-1.28198907e-01 9.12126243e-01 6.92798734e-01 9.30376589e-01
1.37112364e-01 9.67663050e-01 -7.59504318e-01 -1.97097257e-01
-4.85435545e-01 -5.37856184e-02 1.05597389e+00 6.54654264e-01
6.50792360e-01 2.14365348e-01 4.15188581e-01 1.27139464e-01
-1.17034838e-01 6.27377391e-01 8.93072307e-01 -2.44867727e-01
7.41346836e-01 -1.74981076e-02 -3.84873688e-01 -4.11304593e-01
-5.12130499e-01 -4.92624432e-01 -1.08971477e+00 1.61138654e-01
3.76098633e-01 -2.33952403e-01 -1.00412512e+00 1.50744689e+00
-6.02588952e-01 2.94867754e-02 6.57320201e-01 5.83883345e-01
7.30817795e-01 1.14880216e+00 3.66374515e-02 3.77868079e-02
1.40923405e+00 -8.59324753e-01 -3.27943414e-01 -4.67171907e-01
1.11882341e+00 -6.56364739e-01 5.85287869e-01 2.65559793e-01
-1.13930213e+00 -4.65352267e-01 -1.32476199e+00 -2.32316449e-01
-8.47140551e-01 -2.74890751e-01 8.98077607e-01 7.89556563e-01
-1.06694865e+00 7.43223846e-01 -7.11045444e-01 -2.93079447e-02
4.21624571e-01 8.98843229e-01 -4.48613435e-01 2.10587621e-01
-1.39346075e+00 1.07324815e+00 4.64315325e-01 2.08836153e-01
-1.17221761e+00 -3.74647141e-01 -6.91688180e-01 2.57968694e-01
-1.37596130e-01 -9.82823133e-01 1.21949124e+00 -1.40596795e+00
-1.56247258e+00 1.12322915e+00 6.05695732e-02 -1.22375536e+00
-3.60891148e-02 -8.87186080e-02 -3.70090306e-01 4.30151783e-02
-8.22247624e-01 8.36917043e-01 9.73438263e-01 -7.65066206e-01
-3.86869550e-01 -3.00733507e-01 1.08906828e-01 1.83790207e-01
-1.34353042e-02 1.20105453e-01 1.16650671e-01 -8.05637479e-01
-1.11353034e-02 -1.22349226e+00 -9.87595916e-02 -4.36659575e-01
-3.25594246e-01 1.08049221e-01 6.69312537e-01 -6.63772106e-01
7.01344371e-01 -2.16508126e+00 1.62553445e-01 2.75911510e-01
3.07348758e-01 6.89769506e-01 -4.89779055e-01 5.48873007e-01
-4.45663393e-01 4.05754507e-01 -1.12864524e-01 1.90913022e-01
-1.18425302e-01 2.39000544e-01 -9.66639102e-01 7.81319022e-01
7.08464310e-02 1.57461524e+00 -7.49801457e-01 2.60332137e-01
-2.25058556e-01 5.76896548e-01 -2.28001192e-01 -4.08528782e-02
-5.14064252e-01 1.46513909e-01 -6.96853995e-02 3.26350778e-01
5.52465260e-01 -1.37178198e-01 2.03437135e-01 2.56337881e-01
1.04093216e-01 7.63101161e-01 -1.38110459e-01 1.66995287e+00
-8.55196297e-01 1.42356408e+00 -9.44757015e-02 -1.01615334e+00
6.77530229e-01 2.92281181e-01 -1.78532809e-01 -1.05667484e+00
3.67543817e-01 1.99500889e-01 5.73472381e-01 -8.31757337e-02
4.66724902e-01 -1.90531671e-01 -1.53596327e-01 9.45717573e-01
-7.32055977e-02 -1.41867429e-01 -2.46874258e-01 7.59952143e-02
1.07398081e+00 -2.42653385e-01 -5.72556928e-02 -6.08875304e-02
3.82619441e-01 -3.22851926e-01 -7.86613598e-02 8.85215104e-01
2.20302969e-01 2.91317493e-01 6.89319313e-01 -1.08851290e+00
-1.24849331e+00 -1.01052487e+00 2.91013718e-01 1.30935645e+00
-8.29041675e-02 -5.80089033e-01 -7.62871802e-01 -3.72844815e-01
-3.14141482e-01 6.59021378e-01 -6.43373549e-01 -2.89775759e-01
-1.04276919e+00 -1.07712185e+00 1.19717431e+00 3.25600028e-01
4.90277439e-01 -1.22189653e+00 -7.72383213e-01 -9.13317576e-02
2.52418876e-01 -8.12438428e-01 -1.49709731e-01 8.27600837e-01
-8.30300987e-01 -1.12168229e+00 -6.04679227e-01 -1.09996676e+00
8.54837954e-01 -1.38818389e-02 1.10127115e+00 3.70055050e-01
-1.93106502e-01 1.90350026e-01 2.14440033e-01 -5.03899634e-01
-9.74732518e-01 5.00415742e-01 -9.08783525e-02 -4.19518888e-01
1.66445449e-01 -4.56596226e-01 -6.35232270e-01 4.98064645e-02
-1.00645447e+00 5.11301041e-01 1.03323638e+00 9.80231583e-01
3.75144333e-01 -1.87482461e-01 8.91957134e-02 -1.24086046e+00
8.06969583e-01 -3.93348753e-01 -8.46417069e-01 2.59424895e-01
-4.46791142e-01 6.17680848e-01 1.28745914e+00 -3.73100996e-01
-5.80763400e-01 -1.92775875e-01 -3.42036754e-01 -1.34644508e-01
2.73533791e-01 2.22293764e-01 1.98754162e-01 -3.97306085e-01
6.70525312e-01 7.14813292e-01 -2.89512873e-02 -2.88394541e-02
4.41772223e-01 5.31454623e-01 8.12656045e-01 -4.42711234e-01
9.67316508e-01 4.92867738e-01 2.73118407e-01 -5.99849463e-01
-2.30692402e-01 5.44448034e-04 -2.91501254e-01 5.31627059e-01
6.14319921e-01 -8.22614491e-01 -1.29920137e+00 5.33117115e-01
-1.36572337e+00 -5.80298662e-01 -2.09623232e-01 9.96997133e-02
-6.91863477e-01 1.58092901e-01 -1.02476954e+00 -6.03899062e-01
-8.24497104e-01 -1.19358337e+00 9.42980468e-01 -1.69677325e-02
2.76248297e-03 -1.05572915e+00 -9.49448124e-02 -1.99191533e-02
4.71658796e-01 -6.84192926e-02 1.49648166e+00 -8.45873594e-01
-8.97728980e-01 -1.28991231e-01 -2.20091179e-01 3.05907309e-01
-4.83342916e-01 -3.34091753e-01 -1.12387228e+00 -5.06199419e-01
3.22682381e-01 -3.57320160e-01 1.37292647e+00 8.81386548e-03
1.57250381e+00 -7.67313778e-01 -2.71382421e-01 1.38189423e+00
9.84659135e-01 2.37664267e-01 1.04081821e+00 6.06941164e-01
4.04124230e-01 3.97373617e-01 -1.45463243e-01 -1.63848296e-01
1.39284626e-01 5.65119870e-02 6.70652807e-01 -6.33441880e-02
5.61479069e-02 -3.03315520e-01 7.22626567e-01 1.05616987e+00
7.99612105e-02 -2.99466223e-01 -1.00362790e+00 1.26499370e-01
-1.20330489e+00 -8.99260402e-01 2.74312645e-01 2.13336134e+00
5.37641227e-01 2.50607818e-01 -3.22004527e-01 9.89733785e-02
5.16024709e-01 3.09780329e-01 -4.28009778e-01 -1.30754137e+00
-4.99096900e-01 9.80439305e-01 1.17295432e+00 4.59547102e-01
-1.10628414e+00 1.08237326e+00 6.60230064e+00 8.20107341e-01
-1.35997629e+00 -1.01875722e-01 1.15129876e+00 -3.04077193e-02
-5.05092800e-01 1.23265320e-02 -5.29237092e-01 3.61608624e-01
1.77078450e+00 1.43429928e-03 9.94245589e-01 3.86626929e-01
-4.69161063e-01 2.72298753e-01 -1.24199009e+00 1.22548962e+00
3.96209583e-02 -1.86274350e+00 2.29194239e-01 1.32257655e-01
7.03007042e-01 3.87488455e-01 6.80670440e-01 4.80749577e-01
1.82923287e-01 -1.79752302e+00 3.08113605e-01 1.81969762e-01
1.01434815e+00 -9.31066096e-01 6.74795449e-01 2.68455952e-01
-6.38528645e-01 -3.31862152e-01 -4.69185472e-01 -4.92817499e-02
-4.95913833e-01 -3.61144394e-01 -9.94478464e-01 2.55468898e-02
9.88715664e-02 2.99370825e-01 -4.61641252e-01 6.35039091e-01
-4.33581084e-01 4.50148433e-01 -1.01058863e-01 -3.31775665e-01
7.93947995e-01 -3.10478434e-02 2.07652554e-01 1.41080880e+00
1.55610204e-01 1.07309401e-01 -5.91326952e-01 5.64689338e-01
-8.31566155e-01 -1.80699870e-01 -6.88007832e-01 -5.03032923e-01
-3.24140163e-03 9.44741130e-01 -6.99152827e-01 -1.78155869e-01
-4.61424552e-02 1.22512567e+00 2.61221547e-02 4.82584536e-01
-5.96663654e-01 -6.09461725e-01 8.33594263e-01 -1.60866752e-01
3.85100067e-01 -3.00028592e-01 -3.00405413e-01 -1.04888391e+00
-4.58723009e-01 -1.40188694e+00 1.93779424e-01 -7.30575025e-01
-6.84241951e-01 7.78277516e-01 -4.58642781e-01 -8.08590293e-01
-5.18607259e-01 -1.16752088e+00 -7.28929162e-01 9.20252264e-01
-1.47192061e+00 -1.12200904e+00 3.81123781e-01 4.28678423e-01
3.98639679e-01 -5.25873125e-01 1.17524374e+00 -1.51857838e-01
-3.61515284e-01 7.30164170e-01 5.87475955e-01 2.89324969e-01
1.48462236e-01 -1.00856602e+00 1.54001546e+00 4.23522979e-01
6.79458499e-01 1.01042008e+00 6.15005672e-01 -1.14360824e-01
-2.24249911e+00 -8.41298997e-01 7.90997148e-01 -3.35989267e-01
7.90564358e-01 -9.07269359e-01 -5.58974624e-01 9.14522707e-01
2.93031752e-01 -1.05074063e-01 6.18785441e-01 -1.63601965e-01
-7.79724956e-01 6.33740500e-02 -6.95979834e-01 6.94123924e-01
8.58608842e-01 -1.11490119e+00 -3.42217714e-01 8.13556194e-01
6.30595446e-01 -6.86831415e-01 -3.11625212e-01 2.09339019e-02
8.61215949e-01 -6.28576338e-01 1.10547411e+00 -1.08353746e+00
4.94473904e-01 1.67136267e-01 1.26943186e-01 -1.39673245e+00
3.17179471e-01 -1.40391958e+00 -2.92205572e-01 2.77622212e-02
4.82655883e-01 -8.01429331e-01 9.39137757e-01 -9.37594101e-02
-9.07179788e-02 -1.02648020e+00 -1.08134186e+00 -6.74408197e-01
6.28869116e-01 -5.59237421e-01 6.48254037e-01 4.40713227e-01
-1.82001755e-01 3.55786413e-01 -4.26070184e-01 -5.98555841e-02
-1.27547234e-01 2.50388086e-01 7.06818223e-01 -7.19373345e-01
-4.35270250e-01 -7.28704631e-01 -4.48848009e-01 -1.31364346e+00
6.19272768e-01 -1.40941894e+00 -3.64518195e-01 -9.89316285e-01
8.54287148e-02 -1.55154958e-01 -3.35563630e-01 2.60796607e-01
3.94130290e-01 3.09887230e-01 2.49277696e-01 2.63014704e-01
-2.99102455e-01 1.03146642e-01 1.06343794e+00 -5.16501129e-01
2.97546744e-01 2.17060968e-01 -8.05324316e-01 6.06461167e-01
9.43435431e-01 -4.78022248e-01 -1.62617281e-01 -8.24241161e-01
9.60988700e-01 -1.76788673e-01 2.31304973e-01 -7.54896283e-01
3.81371707e-01 3.71872127e-01 5.96347272e-01 -3.97010386e-01
4.16005731e-01 -4.57358450e-01 -1.80874437e-01 1.12009907e+00
-4.66238201e-01 7.28069186e-01 4.24594253e-01 4.20849353e-01
4.56788465e-02 -8.99279639e-02 6.31786585e-01 -4.78250593e-01
-6.77351534e-01 2.49778822e-01 -7.35586464e-01 8.76013469e-03
5.71612179e-01 -1.28020480e-01 -7.91761935e-01 -6.43449366e-01
-3.34596276e-01 9.74591705e-04 3.18360388e-01 2.81829596e-01
7.58278906e-01 -6.51858687e-01 -6.61931455e-01 5.54983556e-01
-2.18803719e-01 -4.89628047e-01 -2.15762898e-01 4.80690718e-01
-1.25594568e+00 1.01732516e+00 -4.39723551e-01 -1.70391515e-01
-1.10470998e+00 7.23036468e-01 5.37114322e-01 -5.73898911e-01
-4.73000824e-01 1.00965393e+00 5.78705549e-01 -3.63383323e-01
1.83009461e-01 -3.25712979e-01 1.53495938e-01 -2.85322577e-01
4.53355432e-01 6.49407133e-02 1.83930114e-01 -5.56954682e-01
-1.87464043e-01 3.44966590e-01 -4.76583153e-01 1.17928185e-01
1.21021676e+00 4.48401600e-01 -4.00207132e-01 2.26980317e-02
1.54012787e+00 -2.84605742e-01 -6.70600295e-01 1.79763615e-01
1.25317536e-02 6.58646375e-02 -3.66225392e-01 -8.46052825e-01
-9.39230442e-01 1.51435459e+00 6.43970490e-01 -1.66591123e-01
1.13378417e+00 -4.15748358e-01 1.36150265e+00 1.31018102e+00
1.40180796e-01 -7.85892725e-01 -1.58540532e-01 1.03254151e+00
5.79381108e-01 -8.27787817e-01 -2.16675803e-01 3.71939391e-01
-2.58830965e-01 1.64421082e+00 1.12585112e-01 -1.45062298e-01
1.92716449e-01 4.80671495e-01 -1.19397212e-02 -9.78076085e-02
-1.24273050e+00 9.05524045e-02 1.14385515e-01 3.87852550e-01
2.85696000e-01 4.17096652e-02 7.82906786e-02 2.63910562e-01
-6.90580785e-01 -6.73936486e-01 6.46986067e-01 6.51867747e-01
-1.53889000e-01 -1.26294959e+00 -4.30102795e-01 5.13255537e-01
-8.09857130e-01 -7.43529022e-01 -6.17441833e-01 9.22809958e-01
-1.25742303e-02 3.60138297e-01 1.52420357e-01 -6.14469230e-01
-4.01099414e-01 2.04609066e-01 7.66011417e-01 -4.37802881e-01
-1.29815114e+00 -4.46165025e-01 -5.95227443e-02 -3.65463912e-01
2.69253075e-01 -1.70882598e-01 -1.07331109e+00 -7.65446007e-01
-2.29272723e-01 1.24561712e-01 1.19645321e+00 8.72601390e-01
2.75957972e-01 3.94463390e-01 6.19471729e-01 -6.38666213e-01
-7.48329997e-01 -5.12365580e-01 -2.02943534e-01 1.80377617e-01
6.92765534e-01 3.45667988e-01 -2.72657990e-01 -3.06260198e-01] | [10.500907897949219, 7.154221057891846] |
159e28d8-6e5b-4dfc-b9b2-ecc3f82fc5ee | topicspam-a-topic-model-based-approach-for | null | null | https://aclanthology.org/P13-2039 | https://aclanthology.org/P13-2039.pdf | TopicSpam: a Topic-Model based approach for spam detection | null | ['Jiwei Li', 'Sujian Li', 'Claire Cardie'] | 2013-08-01 | null | null | null | acl-2013-8 | ['spam-detection'] | ['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] | [-7.192071437835693, 3.594228982925415] |
2b4476b2-c199-4eb9-a403-27e9759cd13e | poster-a-pyramid-cross-fusion-transformer | 2204.04083 | null | https://arxiv.org/abs/2204.04083v1 | https://arxiv.org/pdf/2204.04083v1.pdf | POSTER: A Pyramid Cross-Fusion Transformer Network for Facial Expression Recognition | Facial Expression Recognition (FER) has received increasing interest in the computer vision community. As a challenging task, there are three key issues especially prevalent in FER: inter-class similarity, intra-class discrepancy, and scale sensitivity. Existing methods typically address some of these issues, but do not tackle them all in a unified framework. Therefore, in this paper, we propose a two-stream Pyramid crOss-fuSion TransformER network (POSTER) that aims to holistically solve these issues. Specifically, we design a transformer-based cross-fusion paradigm that enables effective collaboration of facial landmark and direct image features to maximize proper attention to salient facial regions. Furthermore, POSTER employs a pyramid structure to promote scale invariance. Extensive experimental results demonstrate that our POSTER outperforms SOTA methods on RAF-DB with 92.05%, FERPlus with 91.62%, AffectNet (7 cls) with 67.31%, and AffectNet (8 cls) with 63.34%, respectively. | ['Chen Chen', 'Matias Mendieta', 'Ce Zheng'] | 2022-04-08 | null | null | null | null | ['facial-expression-recognition'] | ['computer-vision'] | [ 1.06001966e-01 -4.51689154e-01 5.76740019e-02 -5.31361341e-01
-5.06430149e-01 -3.39426771e-02 2.20735982e-01 -2.85110831e-01
-1.83491334e-01 5.58502257e-01 8.98430571e-02 3.18545312e-01
2.63422582e-04 -5.56082547e-01 -2.46372357e-01 -8.56794477e-01
1.78946599e-01 -4.18662667e-01 2.31895372e-01 -5.55279851e-01
2.57164799e-02 6.33413434e-01 -1.54082489e+00 4.06820238e-01
8.94728780e-01 1.51432455e+00 -2.64277816e-01 -1.79779574e-01
-8.51855651e-02 5.67949057e-01 -4.66316491e-01 -6.57934725e-01
1.54844716e-01 -2.58931190e-01 -3.02737415e-01 1.39460936e-01
5.72328866e-01 -2.48299599e-01 -1.10068299e-01 1.28172338e+00
7.85943270e-01 -4.68362123e-02 5.44779360e-01 -1.60065663e+00
-4.55687672e-01 1.55028760e-01 -1.46808434e+00 2.73431569e-01
1.35143846e-01 -5.76387346e-02 7.83557594e-01 -1.29650486e+00
1.02414407e-01 1.69857252e+00 6.21418238e-01 2.43853629e-01
-1.08918273e+00 -1.30668569e+00 3.03712785e-01 3.27847421e-01
-1.68432498e+00 -6.25557363e-01 9.02460754e-01 -1.48092359e-01
4.02994961e-01 1.64355069e-01 5.88037491e-01 7.63974607e-01
7.23900497e-02 8.21554840e-01 1.41881466e+00 -1.86981738e-01
-7.40759596e-02 -7.95417726e-02 -1.91143751e-01 7.20055878e-01
6.24438450e-02 -2.17826352e-01 -6.63743854e-01 -3.39353345e-02
8.63516808e-01 -4.29502502e-02 -3.26311439e-01 -1.95427492e-01
-7.03867078e-01 6.47951663e-01 5.38517118e-01 4.81011778e-01
-4.42747295e-01 -8.96110311e-02 4.60223258e-01 3.53497230e-02
6.00952804e-01 -1.58057913e-01 -2.72939324e-01 1.47135571e-01
-7.93777227e-01 1.45882964e-01 1.33756116e-01 7.74448931e-01
8.54782403e-01 3.23468566e-01 -2.69632369e-01 1.32126856e+00
4.31705117e-01 6.63870275e-01 2.02532932e-01 -8.28453183e-01
1.66601032e-01 6.79125428e-01 -2.59804517e-01 -1.63110781e+00
-3.61620992e-01 -4.83431727e-01 -1.12412107e+00 1.43821463e-01
3.41705815e-03 -1.35785714e-01 -5.93325734e-01 1.95333385e+00
4.23071206e-01 4.13561940e-01 -1.48723245e-01 9.44439709e-01
9.34247375e-01 4.31362361e-01 4.34328943e-01 -4.35205579e-01
1.55734885e+00 -7.83355415e-01 -8.12164307e-01 -4.87352721e-02
1.85640112e-01 -1.06865740e+00 7.88454235e-01 3.15630376e-01
-1.07970524e+00 -6.93723500e-01 -8.78810585e-01 1.61243930e-01
1.18749728e-02 5.35824597e-01 5.28944552e-01 6.03536606e-01
-1.13616991e+00 7.63006136e-02 -3.12644273e-01 -3.21830064e-01
5.74403822e-01 5.16935587e-01 -4.12669808e-01 -8.36468190e-02
-1.17698824e+00 6.14542902e-01 4.15114388e-02 2.95650929e-01
-3.16609651e-01 -5.73397040e-01 -7.19684362e-01 2.65344232e-02
1.31475702e-01 -3.97191644e-01 1.10155880e+00 -1.35586345e+00
-1.38329530e+00 8.48722577e-01 -4.37561721e-01 2.09652260e-02
1.89937413e-01 -1.41820297e-01 -7.47655034e-01 3.82830739e-01
1.13141410e-01 8.07513297e-01 8.96476090e-01 -1.28554213e+00
-7.63053298e-01 -7.92324483e-01 -1.16281047e-01 3.74517024e-01
-6.28640592e-01 4.50314671e-01 -5.44514656e-01 -6.95139527e-01
1.93104252e-01 -6.87098801e-01 7.21747503e-02 3.21442783e-01
-1.06508493e-01 -3.54931891e-01 1.06157124e+00 -5.71357906e-01
1.42295241e+00 -2.33542657e+00 -1.46741822e-01 1.35202199e-01
2.59204268e-01 6.21177435e-01 -1.46567747e-01 -9.74012762e-02
-1.92535743e-01 -1.33736148e-01 -4.32818532e-02 -2.89945871e-01
-2.67704070e-01 -5.30363694e-02 -9.41700488e-02 4.79311734e-01
4.41311449e-01 5.90919197e-01 -6.29490435e-01 -8.89378548e-01
1.42207846e-01 8.63552630e-01 -5.26601374e-01 -7.40957260e-02
4.21152592e-01 2.31578082e-01 -6.35333717e-01 1.13661385e+00
1.29544449e+00 -7.04662874e-02 5.50399572e-02 -7.66047776e-01
-8.41026083e-02 -5.81064284e-01 -1.25106061e+00 1.39889252e+00
-1.90686628e-01 3.02183062e-01 4.23796356e-01 -9.39838648e-01
1.25251901e+00 2.31448770e-01 7.77904391e-01 -9.09199476e-01
4.14948136e-01 2.88814574e-01 -1.37579396e-01 -3.75675380e-01
3.65069240e-01 -3.68645370e-01 2.56551743e-01 -1.73554316e-01
1.26160160e-01 3.92635792e-01 4.76866253e-02 -1.10331148e-01
6.49460137e-01 -1.32696986e-01 3.79118532e-01 -2.97392815e-01
1.05648708e+00 -6.92792952e-01 9.22019482e-01 1.35903778e-02
-6.46429002e-01 6.28419220e-01 4.44367498e-01 -2.90146172e-01
-4.60982233e-01 -9.49936628e-01 -2.58741051e-01 1.10725069e+00
4.34930682e-01 -4.11154121e-01 -8.17675471e-01 -5.56340039e-01
-2.36480795e-02 2.63754018e-02 -5.87850749e-01 -1.73987806e-01
-4.24016058e-01 -6.27439439e-01 5.39481640e-01 4.51231927e-01
1.09543252e+00 -8.55156362e-01 -3.17192286e-01 -3.97756249e-02
-4.21684623e-01 -1.10409164e+00 -5.82419038e-01 -5.46712995e-01
-5.17331123e-01 -9.50124919e-01 -9.30894375e-01 -9.51905549e-01
7.03282177e-01 6.88133955e-01 7.83299148e-01 -1.33637544e-02
-3.07080299e-01 2.09076732e-01 -3.14638197e-01 -3.34504277e-01
2.87798464e-01 -2.65417546e-01 6.71396479e-02 7.28368580e-01
5.77134311e-01 -7.19700694e-01 -8.19291174e-01 7.12154686e-01
-7.48492420e-01 -9.35724191e-03 8.10228050e-01 7.80039787e-01
6.24183536e-01 -1.71386331e-01 6.76633000e-01 -2.13902608e-01
5.26525557e-01 -4.67455268e-01 -2.98093915e-01 3.31377357e-01
-3.81012648e-01 -5.44909239e-01 4.23701286e-01 -3.21903527e-01
-1.25774097e+00 3.00577637e-02 -1.99775785e-01 -7.11798728e-01
-1.40500903e-01 1.73576012e-01 -5.77451527e-01 -4.66193110e-01
3.80079418e-01 2.46473789e-01 2.26785779e-01 -3.38522345e-01
-3.60706411e-02 7.49478519e-01 3.35803509e-01 -6.94406092e-01
5.89935482e-01 5.10002613e-01 1.45213738e-01 -8.39161038e-01
-8.04204226e-01 -4.46284086e-01 -3.83001745e-01 -4.74495351e-01
6.53895080e-01 -1.29923165e+00 -8.88401330e-01 6.28900170e-01
-1.00119936e+00 4.00406122e-01 1.26589596e-01 2.77378052e-01
-3.01249236e-01 3.63778651e-01 -4.35880959e-01 -7.49863029e-01
-5.31291366e-01 -1.25092554e+00 1.30518556e+00 7.44757414e-01
8.13262817e-03 -5.66125810e-01 -4.17131573e-01 3.46789181e-01
6.25244498e-01 3.45422566e-01 3.16429555e-01 -1.08569838e-01
-1.80224702e-01 1.78601563e-01 -8.33504438e-01 5.46742976e-01
4.42221493e-01 9.47246477e-02 -1.03831351e+00 -1.63067341e-01
-2.94451773e-01 -3.39682698e-01 6.11472607e-01 3.43748391e-01
1.28548431e+00 2.30833422e-02 -2.47028142e-01 6.18494391e-01
1.17067087e+00 3.77894640e-01 6.29843533e-01 4.02676761e-02
4.35184032e-01 6.34700537e-01 8.23123813e-01 6.64523184e-01
3.31455976e-01 8.43260646e-01 3.60933036e-01 -5.30785024e-01
-9.60507318e-02 -5.53696826e-02 3.34984869e-01 5.87121010e-01
-1.85434386e-01 1.61566973e-01 -5.99659681e-01 3.20789099e-01
-1.73128641e+00 -8.97126496e-01 8.20287019e-02 1.66218805e+00
8.18484187e-01 -3.17452729e-01 1.11932628e-01 7.40305930e-02
8.40493381e-01 3.04506749e-01 -3.70241493e-01 -9.25347060e-02
-4.72114027e-01 2.60046244e-01 4.73332331e-02 9.64798108e-02
-1.18360567e+00 8.83808851e-01 5.43381214e+00 1.29572392e+00
-1.39368165e+00 -3.32948230e-02 9.92329836e-01 6.28829598e-02
4.24461327e-02 -2.92084605e-01 -9.07251239e-01 5.81638277e-01
2.27667943e-01 -2.53688782e-01 1.25947654e-01 7.94030607e-01
3.12705576e-01 1.00077605e-02 -4.06780064e-01 1.46184731e+00
3.15083981e-01 -8.04707289e-01 2.09756568e-01 2.68692952e-02
5.84674299e-01 -4.72748160e-01 2.79776424e-01 1.88438356e-01
-1.86338842e-01 -1.01012480e+00 4.87303793e-01 4.56919223e-01
8.72089863e-01 -1.08854425e+00 7.25649953e-01 -1.56605691e-01
-1.53354669e+00 9.68220644e-03 -2.67464250e-01 1.81078732e-01
3.83285023e-02 6.08796954e-01 -7.84376115e-02 6.44726455e-01
1.00325727e+00 8.99113953e-01 -4.77966666e-01 8.86806488e-01
-1.11834385e-01 3.20337057e-01 -4.03061420e-01 1.32016584e-01
1.24813214e-01 -2.59597093e-01 3.22601020e-01 1.15728462e+00
4.52395260e-01 2.78004259e-01 1.71870336e-01 4.93538201e-01
-9.22034010e-02 5.15057147e-01 -2.97648013e-01 3.55363518e-01
3.74236256e-01 1.59875739e+00 -4.22026783e-01 -6.97013959e-02
-4.37722594e-01 8.29774618e-01 1.35257304e-01 3.46216559e-01
-1.06350458e+00 -3.32874388e-01 1.15859020e+00 4.07345593e-02
3.40760529e-01 1.33779824e-01 3.82968113e-02 -9.91564274e-01
1.48187160e-01 -1.04310536e+00 3.96153450e-01 -9.21660006e-01
-1.20642543e+00 6.05078518e-01 -1.07647516e-01 -1.25821126e+00
3.50594819e-01 -5.70895612e-01 -4.69858527e-01 7.94022501e-01
-2.01084208e+00 -1.40699756e+00 -6.54248714e-01 1.06113350e+00
3.38085234e-01 -1.39210865e-01 5.99930167e-01 7.48682618e-01
-7.73482144e-01 1.03402615e+00 -2.37585559e-01 1.37648672e-01
9.55556452e-01 -5.72083950e-01 -3.60236228e-01 6.12545788e-01
-2.69367129e-01 7.17944562e-01 4.14029211e-01 -1.90631583e-01
-1.13082159e+00 -1.08755851e+00 6.57096803e-01 2.53820956e-01
4.17823017e-01 6.03083745e-02 -8.70708764e-01 3.28919768e-01
7.19125643e-02 4.40928549e-01 6.19778216e-01 7.08232354e-03
-5.23932278e-01 -7.27185667e-01 -1.28044641e+00 6.73214555e-01
1.09062970e+00 -2.43431672e-01 4.12367657e-02 -2.09588557e-02
2.43608698e-01 -3.69637430e-01 -9.44363475e-01 7.49451816e-01
6.61562681e-01 -1.33223093e+00 8.14437509e-01 -1.06159613e-01
3.75573665e-01 -6.19103134e-01 -3.86469811e-01 -1.06820154e+00
-3.72397959e-01 -5.57163298e-01 2.03353539e-01 1.53221798e+00
-6.39942363e-02 -7.05411434e-01 5.47979474e-01 1.00269988e-01
4.80658486e-02 -1.03846467e+00 -9.13528681e-01 -5.24497926e-01
-2.38425672e-01 -6.57008961e-02 6.88572049e-01 9.39988971e-01
-1.72407255e-01 3.73955399e-01 -4.51845050e-01 3.99570838e-02
5.53986311e-01 9.16051418e-02 7.59470284e-01 -1.17702460e+00
2.07113162e-01 -6.16754532e-01 -6.30519331e-01 -9.41944957e-01
6.19385168e-02 -3.71112525e-01 -1.57126456e-01 -1.24449861e+00
3.72523248e-01 -3.31521720e-01 -5.79841137e-01 6.45962536e-01
-2.81100422e-01 4.85668689e-01 1.99128002e-01 6.18242659e-02
-7.29509532e-01 9.35138166e-01 1.38025022e+00 6.95719272e-02
1.79109231e-01 -3.25235337e-01 -1.04098868e+00 8.87690723e-01
8.55849385e-01 -2.41123345e-02 -4.67733324e-01 -2.63354093e-01
-3.33642483e-01 -7.45159611e-02 2.83259600e-01 -1.09894717e+00
1.30934596e-01 -3.22800994e-01 6.27986610e-01 -4.44173068e-01
5.77885091e-01 -8.11384976e-01 -1.44293830e-01 1.63097024e-01
1.32774800e-01 1.34316295e-01 3.95355999e-01 4.02875662e-01
-6.91365063e-01 5.91934741e-01 1.20164347e+00 2.69794434e-01
-8.42016637e-01 7.13935137e-01 -2.55977456e-03 -5.95113486e-02
1.29432058e+00 -3.09312493e-01 -2.10987806e-01 -4.41538393e-01
-3.88841033e-01 2.63951242e-01 8.95681903e-02 5.22288561e-01
7.12877870e-01 -1.55473876e+00 -6.90931916e-01 3.63505214e-01
2.35643625e-01 -2.90201992e-01 6.53416514e-01 1.11505079e+00
-9.24978480e-02 1.97974011e-01 -6.05691671e-01 -6.04503751e-01
-1.73239958e+00 5.66298179e-02 4.31443363e-01 -6.15906939e-02
-9.01845843e-02 9.91414189e-01 5.58424473e-01 -8.13818723e-02
1.55687258e-01 1.11261472e-01 -4.85711247e-01 2.43996084e-01
7.61087835e-01 2.20895693e-01 -1.04543522e-01 -1.22574055e+00
-4.15916711e-01 1.03006220e+00 -1.39628291e-01 2.80142367e-01
1.09015059e+00 -1.51944965e-01 -3.74704778e-01 4.92094504e-03
1.28018856e+00 3.99709046e-02 -1.24114406e+00 -5.13590753e-01
-4.04933065e-01 -6.94082975e-01 1.14400066e-01 -5.93278110e-01
-1.57577229e+00 8.90069067e-01 8.25021386e-01 -2.26277784e-01
1.89040971e+00 -3.35835665e-01 7.93578506e-01 -1.87628880e-01
3.61093253e-01 -9.70307946e-01 3.78758430e-01 3.41904074e-01
1.02198124e+00 -1.07457054e+00 -2.72528846e-02 -7.80333698e-01
-6.02180123e-01 1.01413321e+00 1.06432152e+00 2.24186648e-02
8.62413883e-01 1.40814170e-01 5.04046530e-02 -1.34223461e-01
-5.47429383e-01 -3.13056856e-01 3.15256178e-01 4.70602840e-01
5.57313144e-01 -1.40489221e-01 -3.76958072e-01 6.62273407e-01
8.68484080e-02 1.92483604e-01 -1.55071452e-01 8.69430482e-01
-6.26449585e-01 -8.73613954e-01 -4.31471884e-01 1.76559299e-01
-6.79387689e-01 3.10058072e-02 -2.60107100e-01 8.06960762e-01
4.93009329e-01 9.79393065e-01 -1.31169697e-02 -7.17113376e-01
4.87172663e-01 -2.96109110e-01 3.38867992e-01 -1.00590326e-01
-3.88035804e-01 4.55444187e-01 -1.70600206e-01 -6.36709571e-01
-6.82743728e-01 -6.70256317e-01 -1.09766924e+00 -4.48455930e-01
-2.59081662e-01 1.07113970e-02 4.68586922e-01 7.59581745e-01
6.97169304e-01 3.43425512e-01 8.78467798e-01 -5.78010201e-01
-3.66488457e-01 -8.40862095e-01 -7.29342520e-01 2.96732157e-01
1.44089356e-01 -1.00568855e+00 -2.49826267e-01 -6.67732283e-02] | [13.586246490478516, 1.5280145406723022] |
7882f20f-ce4e-4e81-b906-af5a16eea60f | a-flow-based-latent-state-generative-model-of | null | null | http://proceedings.neurips.cc/paper/2021/hash/84a529a92de322be42dd3365afd54f91-Abstract.html | http://proceedings.neurips.cc/paper/2021/file/84a529a92de322be42dd3365afd54f91-Paper.pdf | A flow-based latent state generative model of neural population responses to natural images | We present a joint deep neural system identification model for two major sources of neural variability: stimulus-driven and stimulus-conditioned fluctuations. To this end, we combine (1) state-of-the-art deep networks for stimulus-driven activity and (2) a flexible, normalizing flow-based generative model to capture the stimulus-conditioned variability including noise correlations. This allows us to train the model end-to-end without the need for sophisticated probabilistic approximations associated with many latent state models for stimulus-conditioned fluctuations. We train the model on the responses of thousands of neurons from multiple areas of the mouse visual cortex to natural images. We show that our model outperforms previous state-of-the-art models in predicting the distribution of neural population responses to novel stimuli, including shared stimulus-conditioned variability. Furthermore, it successfully learns known latent factors of the population responses that are related to behavioral variables such as pupil dilation, and other factors that vary systematically with brain area or retinotopic location. Overall, our model accurately accounts for two critical sources of neural variability while avoiding several complexities associated with many existing latent state models. It thus provides a useful tool for uncovering the interplay between different factors that contribute to variability in neural activity. | ['Fabian Sinz', 'Andreas Tolias', 'Zhuokun Ding', 'Zhiwei Ding', 'Taliah Muhammad', 'Akshay Jagadish', 'Konstantin-Klemens Lurz', 'Edgar Walker', 'Mohammad Bashiri'] | 2021-12-01 | null | https://openreview.net/forum?id=1yeYYtLqq7K | https://openreview.net/pdf?id=1yeYYtLqq7K | neurips-2021-12 | ['pupil-dilation'] | ['computer-vision'] | [ 2.17559382e-01 -6.76945686e-01 6.04464300e-02 -2.62504488e-01
-7.42101014e-01 -5.88797629e-01 8.34965706e-01 -1.24446698e-01
-5.57432115e-01 6.70900047e-01 2.83935845e-01 -7.44268298e-02
-1.06302693e-01 -3.49780291e-01 -9.53174651e-01 -1.07532775e+00
-1.93316907e-01 1.79992646e-01 1.87035292e-01 4.18833233e-02
2.89474100e-01 3.36725563e-01 -1.52695787e+00 1.50916204e-01
6.33031547e-01 9.79955852e-01 2.87446439e-01 7.57924080e-01
2.14070722e-01 6.87676072e-01 -5.52203238e-01 3.46293032e-01
-6.80340175e-03 -7.21267402e-01 -2.60127395e-01 -2.20545560e-01
5.88438690e-01 -3.43250223e-02 -3.83701712e-01 1.00182521e+00
5.46110272e-01 -3.57646495e-02 9.47452843e-01 -9.35309529e-01
-5.71698010e-01 4.85742718e-01 -4.59343344e-01 7.09981918e-01
-3.41741174e-01 6.67470098e-01 7.20105112e-01 -6.51960552e-01
4.61969733e-01 1.17839038e+00 3.52015793e-01 6.08214438e-01
-2.09228873e+00 -7.17945814e-01 2.42236704e-01 -7.02610314e-02
-1.17230546e+00 -7.07051635e-01 7.06568658e-01 -1.10687304e+00
9.79386926e-01 -2.46343598e-01 7.45377243e-01 1.61742568e+00
6.52059197e-01 4.40987825e-01 1.37150347e+00 8.28207284e-02
5.41863680e-01 -1.92912266e-01 3.53614539e-01 1.77757755e-01
4.93748151e-02 2.49140620e-01 -7.28787184e-01 -3.02071869e-01
1.24291873e+00 -2.68282056e-01 -2.80525684e-01 -2.21669957e-01
-1.19127929e+00 5.62036335e-01 1.45024821e-01 1.14926152e-01
-3.82580340e-01 7.69185543e-01 1.02903187e-01 -1.25910088e-01
2.13674948e-01 5.29125392e-01 -5.71540296e-01 -3.41247529e-01
-1.03771257e+00 3.08115780e-01 5.22548556e-01 4.26301718e-01
7.56912887e-01 5.21816432e-01 -2.15749457e-01 7.82062471e-01
4.81816232e-01 5.87199688e-01 7.20391572e-01 -1.17300951e+00
7.01553971e-02 1.83380902e-01 -1.24336429e-01 -4.18360800e-01
-5.78979731e-01 -8.20373535e-01 -7.13748693e-01 4.17699993e-01
6.74306870e-01 -1.13950379e-01 -1.08134282e+00 2.51821375e+00
-3.56245488e-01 2.88395226e-01 -3.13448280e-01 5.72001219e-01
4.36538994e-01 4.49304879e-01 2.90771425e-01 -3.83367658e-01
1.17250586e+00 -2.63488859e-01 -5.74054897e-01 -5.73772132e-01
4.71244790e-02 -4.04998720e-01 9.16377306e-01 1.59611285e-01
-1.22786891e+00 -5.43680191e-01 -7.99165070e-01 2.96647519e-01
-8.60080495e-02 -1.13145456e-01 7.18491256e-01 3.07233065e-01
-1.04898500e+00 5.52765608e-01 -1.34643757e+00 -2.47730583e-01
7.01434851e-01 4.48253125e-01 1.36028593e-02 5.90430737e-01
-8.57116044e-01 7.65790701e-01 -1.07725210e-01 -3.65819782e-02
-1.44570994e+00 -1.03923666e+00 -3.69693279e-01 2.90993780e-01
-1.65052384e-01 -1.00853801e+00 1.25635087e+00 -7.51316130e-01
-1.52873552e+00 6.11990452e-01 -5.81268072e-01 -2.81604826e-01
-3.04271951e-02 8.38585421e-02 -4.65107523e-03 1.72629429e-03
-1.16435632e-01 6.50279224e-01 7.42214322e-01 -1.11502504e+00
1.58447865e-02 -4.68355358e-01 -6.96487010e-01 2.05320325e-02
-1.21753793e-02 5.09235933e-02 -8.80986750e-02 -5.68867743e-01
4.53349426e-02 -8.84986699e-01 -1.64462939e-01 -1.11929839e-02
-4.38495204e-02 7.74964988e-02 1.21303976e-01 -1.80031478e-01
9.88136947e-01 -2.12534285e+00 3.04234922e-01 3.22479680e-02
4.60084379e-01 -4.90763411e-02 -4.55079168e-01 2.71585077e-01
-2.08205774e-01 1.25584722e-01 -2.70171583e-01 -3.08268219e-01
-2.21150294e-01 -2.00920627e-01 -5.63640893e-01 5.45092642e-01
5.11773527e-01 1.19129634e+00 -6.70914710e-01 1.56601474e-01
4.22581658e-02 6.71727598e-01 -5.53036571e-01 8.39202777e-02
-3.48707944e-01 7.83243537e-01 -4.40798886e-02 3.00970674e-01
3.75910491e-01 -3.88197690e-01 -2.23627090e-02 -2.12894697e-02
-2.59612560e-01 6.69943869e-01 -8.07871044e-01 1.63580179e+00
-3.10293257e-01 9.89297450e-01 -1.82306573e-01 -7.84507513e-01
7.20629454e-01 2.06493884e-01 6.27425253e-01 -5.66089094e-01
4.74849343e-01 1.86425790e-01 5.66242337e-01 -1.62807211e-01
-2.76790202e-01 -4.06160355e-02 1.44709751e-01 5.91906071e-01
8.00859749e-01 -3.48008946e-02 1.23370610e-01 6.65963963e-02
1.12265205e+00 2.18939036e-01 1.94832370e-01 -5.95772266e-01
-1.77411899e-01 -5.68918109e-01 5.97147703e-01 8.56195629e-01
-2.60816127e-01 5.01677513e-01 8.67408514e-01 -2.00628489e-01
-1.00171137e+00 -1.55363190e+00 -5.26829779e-01 8.34406197e-01
-2.12269932e-01 -1.27922073e-01 -6.10617340e-01 4.17379409e-01
-2.07020000e-01 3.90799403e-01 -9.26756084e-01 -5.31417906e-01
-1.94571048e-01 -1.29818034e+00 5.06754816e-01 6.19187295e-01
1.40797973e-01 -1.03883147e+00 -5.52857935e-01 4.21915680e-01
-8.57716277e-02 -9.82858181e-01 -2.13987753e-01 7.49396980e-01
-9.97502267e-01 -7.49088943e-01 -5.44640779e-01 -5.11466920e-01
5.90839684e-01 -1.53755099e-01 1.00861645e+00 -5.49563050e-01
-7.00872123e-01 2.60717809e-01 4.37997043e-01 -4.92222369e-01
-1.71008989e-01 -1.51321441e-01 3.89770508e-01 -6.86944574e-02
4.97595966e-01 -1.10274136e+00 -6.42827332e-01 2.96332061e-01
-8.38208616e-01 -1.46612823e-01 4.22430933e-01 8.67904305e-01
7.91750669e-01 -2.13566974e-01 5.92142105e-01 -6.46796107e-01
6.84540272e-01 -5.20853639e-01 -8.31569552e-01 -4.29914482e-02
-4.87330437e-01 4.86467123e-01 4.83628541e-01 -1.00136089e+00
-9.36649799e-01 1.40238896e-01 2.04915777e-01 -2.38365561e-01
-3.86358857e-01 1.74703985e-01 2.16533065e-01 -1.07076883e-01
1.12741673e+00 5.85975111e-01 -9.41685811e-02 -2.13984728e-01
1.59509405e-01 1.45304486e-01 7.23874807e-01 -5.85832477e-01
6.51292682e-01 3.84482205e-01 2.00489789e-01 -6.75235271e-01
-7.20678627e-01 -2.01370209e-01 -5.04082680e-01 -1.11796245e-01
9.10397649e-01 -1.06136060e+00 -9.37061906e-01 1.02852249e+00
-1.17667556e+00 -7.86715508e-01 -1.72116205e-01 8.11187685e-01
-1.10466123e+00 -2.82430410e-01 -8.36448908e-01 -1.06866479e+00
-7.73678198e-02 -1.33337152e+00 8.04296613e-01 4.11602467e-01
-2.66992450e-01 -8.87714028e-01 6.51150942e-01 -1.55879796e-01
7.27958381e-01 6.61765188e-02 1.00567949e+00 -3.08906764e-01
-7.14653134e-01 1.44615069e-01 3.61124799e-02 1.43063843e-01
-2.06835885e-02 3.41357142e-01 -1.24006677e+00 -1.19699672e-01
1.67199597e-01 -5.40084660e-01 1.09514678e+00 1.24943757e+00
1.08067095e+00 -1.48465335e-02 -1.34740680e-01 1.03033996e+00
1.38814604e+00 1.81859255e-01 6.68094277e-01 -4.12857458e-02
3.38549525e-01 3.53956461e-01 -4.95179325e-01 2.97358721e-01
1.76906392e-01 4.60708618e-01 2.98177570e-01 1.07144773e-01
4.38735001e-02 -2.47803062e-01 4.18896765e-01 5.79434037e-01
-8.85394588e-02 -7.07251802e-02 -7.93867648e-01 4.79056537e-01
-1.73057997e+00 -1.20196414e+00 -1.28431410e-01 2.34365654e+00
9.77130234e-01 2.82590568e-01 2.90562391e-01 -3.46509010e-01
4.40409213e-01 -7.24755647e-03 -1.05328965e+00 -1.44095585e-01
-3.79452705e-01 2.93018162e-01 3.62758487e-01 3.67689818e-01
-6.73335552e-01 8.07624757e-01 7.80704403e+00 3.33720714e-01
-1.24668348e+00 -1.98969394e-01 6.03591919e-01 -5.27706325e-01
-3.30990851e-01 -2.87681408e-02 -1.20880377e+00 5.72166264e-01
1.25605905e+00 -2.64301926e-01 9.94573116e-01 3.39860320e-01
3.64250362e-01 -3.46457839e-01 -1.14711428e+00 9.21746016e-01
-4.16459031e-02 -1.30688131e+00 -2.18422115e-01 3.48832041e-01
9.17563975e-01 5.69705248e-01 5.93587756e-01 1.57317251e-01
3.77048999e-01 -1.02495587e+00 6.30481064e-01 9.06895220e-01
6.53861105e-01 -2.31490582e-01 3.00409347e-01 4.91141081e-01
-8.14208090e-01 -2.47451905e-02 -4.60932076e-01 -1.55596761e-02
5.22937328e-02 4.75613743e-01 -3.01480055e-01 -5.65573275e-01
5.16269147e-01 4.59879339e-01 -6.69420123e-01 1.13088107e+00
-9.29753855e-02 8.70949686e-01 -4.28072006e-01 -1.67787001e-02
-1.39067203e-01 4.67528999e-02 5.83323121e-01 8.91325951e-01
2.60666937e-01 -8.60887468e-02 -5.20715415e-01 1.74092591e+00
5.86493015e-02 -3.79010022e-01 -4.89597350e-01 -3.71717930e-01
5.65598249e-01 1.13305068e+00 -5.59425890e-01 5.13126999e-02
-1.90413922e-01 5.74190199e-01 5.05690575e-01 8.50302160e-01
-6.07770324e-01 -1.16321504e-01 8.16809952e-01 4.28620167e-02
1.75952792e-01 -5.57996631e-01 -4.55029249e-01 -1.33018458e+00
-1.90748975e-01 -5.69320202e-01 -1.89347997e-01 -8.20196211e-01
-1.44029438e+00 7.65774488e-01 -3.68624404e-02 -8.22666943e-01
-4.93225098e-01 -7.42195189e-01 -6.81889117e-01 1.35411978e+00
-1.36283469e+00 -5.93214512e-01 -5.80989793e-02 6.96988881e-01
2.82909244e-01 -2.16079637e-01 8.99100959e-01 6.55357668e-04
-7.54745305e-01 4.19402689e-01 9.57206339e-02 -9.96552631e-02
8.06663632e-01 -1.15994644e+00 5.79242408e-01 9.06148791e-01
4.77193981e-01 1.10794699e+00 6.32901371e-01 -4.28400636e-01
-1.02010417e+00 -7.03002453e-01 4.00239408e-01 -8.15949619e-01
8.86866271e-01 -8.51171851e-01 -9.27380145e-01 5.91899157e-01
-1.86060429e-01 2.11121306e-01 9.61387396e-01 2.94812620e-01
-5.46240389e-01 -9.44096968e-02 -4.01293755e-01 6.83146060e-01
8.31318855e-01 -7.55518615e-01 -3.63667607e-01 -1.92600146e-01
3.84929717e-01 -8.42653066e-02 -5.10671020e-01 9.40565839e-02
7.96868324e-01 -7.33497024e-01 6.99173987e-01 -7.79055357e-01
2.57519960e-01 -2.04283640e-01 -1.24240011e-01 -1.64485276e+00
-9.10129607e-01 -7.55228877e-01 -2.36302853e-01 9.53022718e-01
6.24454737e-01 -6.30445302e-01 4.30010349e-01 8.16972971e-01
3.11052892e-02 -5.84729433e-01 -8.75944257e-01 -7.10355699e-01
3.75238031e-01 -4.11461920e-01 -7.44075328e-03 4.24880594e-01
9.39762406e-03 5.57025373e-01 -1.86363727e-01 -2.99862511e-02
7.65970826e-01 -1.99840277e-01 5.01586556e-01 -1.27130306e+00
-6.06984138e-01 -8.65097463e-01 -5.12085736e-01 -1.11047602e+00
6.03150539e-02 -6.48016155e-01 2.89795309e-01 -1.09573948e+00
6.54611826e-01 1.44287758e-02 -6.48334324e-01 2.68263370e-01
-2.82812774e-01 1.13148570e-01 -5.80282824e-04 2.74002373e-01
-1.88137785e-01 5.39661467e-01 1.02847075e+00 3.04888457e-01
-3.80642593e-01 -6.78320602e-02 -8.49121332e-01 5.51733494e-01
8.09146404e-01 -5.84459901e-01 -4.21389222e-01 -5.12075722e-01
3.26650888e-01 -1.10304825e-01 6.18362784e-01 -1.16794062e+00
2.18928620e-01 -3.34179163e-01 6.32675529e-01 -2.50274509e-01
3.39070052e-01 -4.71239477e-01 -3.49029414e-02 2.08781749e-01
-6.63305163e-01 -3.00835371e-01 3.88258398e-01 8.12292218e-01
7.12937862e-02 2.81090319e-01 1.10397184e+00 -8.51439312e-03
-1.97263896e-01 5.10213554e-01 -9.57326710e-01 2.50669241e-01
7.10936487e-01 1.47765409e-02 -8.86179328e-01 -3.39817017e-01
-6.78724527e-01 -9.83889699e-02 2.03067854e-01 3.41198564e-01
1.60840362e-01 -1.21874428e+00 -5.61294675e-01 5.77377021e-01
1.25373676e-01 -4.68268335e-01 3.49623173e-01 9.53814447e-01
6.11601286e-02 5.14355242e-01 -5.52460313e-01 -9.15802658e-01
-5.94617486e-01 3.37518573e-01 5.56436896e-01 8.38654935e-02
6.80607930e-02 1.09139395e+00 6.56014144e-01 1.16183996e-01
1.57778319e-02 -4.79215950e-01 -2.15905067e-02 -3.22632901e-02
5.26645482e-01 4.75620069e-02 -2.28496253e-01 -3.61795574e-01
-2.69348145e-01 5.71852624e-01 1.33583173e-01 -4.54012871e-01
1.33221567e+00 -8.64753947e-02 5.44049367e-02 1.29802012e+00
1.10852969e+00 -3.57118994e-01 -1.89837408e+00 -3.38948101e-01
-6.36810601e-01 -7.01386556e-02 2.39861399e-01 -9.80002582e-01
-8.77761841e-01 1.38552153e+00 7.07064629e-01 -2.32561156e-01
1.07418907e+00 4.41988744e-02 3.60004276e-01 1.84557348e-01
3.25848669e-01 -9.03793156e-01 2.90803224e-01 3.93558949e-01
5.61770022e-01 -1.09006834e+00 -2.10575148e-01 1.95675120e-01
-5.49272895e-01 9.91488099e-01 8.02305341e-01 -4.31619734e-01
7.63199031e-01 5.86424112e-01 8.26401412e-02 -1.05784163e-01
-1.43335736e+00 -1.60456106e-01 4.81085211e-01 6.30378783e-01
5.76337397e-01 -3.11859362e-02 -4.10370082e-02 6.62862301e-01
3.90031701e-03 1.76508315e-02 3.72974664e-01 4.93509203e-01
-5.20739913e-01 -7.72594690e-01 2.24028584e-02 7.01002240e-01
-4.97962087e-01 -4.21793222e-01 -1.49439856e-01 3.66149098e-01
-2.00895190e-01 5.14272988e-01 3.22631270e-01 -1.98632836e-01
2.28681322e-02 4.30879056e-01 7.79521227e-01 -7.54862309e-01
-3.47480297e-01 6.28189623e-01 -6.33501589e-01 -6.10516727e-01
-1.96302697e-01 -8.95807207e-01 -1.00677764e+00 1.22593015e-01
-1.64720342e-01 -5.47341526e-01 7.99765050e-01 8.42725992e-01
4.11456317e-01 6.92245722e-01 3.94060820e-01 -8.97411466e-01
-6.32865965e-01 -1.04079843e+00 -7.56494582e-01 2.71885067e-01
5.23393393e-01 -7.25644529e-01 -8.54694128e-01 3.67229044e-01] | [9.386775016784668, 2.6345813274383545] |
05c90258-b1d0-48af-b3f4-57101478505a | low-rank-covariance-completion-for-graph | 2209.08273 | null | https://arxiv.org/abs/2209.08273v1 | https://arxiv.org/pdf/2209.08273v1.pdf | Low-Rank Covariance Completion for Graph Quilting with Applications to Functional Connectivity | As a tool for estimating networks in high dimensions, graphical models are commonly applied to calcium imaging data to estimate functional neuronal connectivity, i.e. relationships between the activities of neurons. However, in many calcium imaging data sets, the full population of neurons is not recorded simultaneously, but instead in partially overlapping blocks. This leads to the Graph Quilting problem, as first introduced by (Vinci et.al. 2019), in which the goal is to infer the structure of the full graph when only subsets of features are jointly observed. In this paper, we study a novel two-step approach to Graph Quilting, which first imputes the complete covariance matrix using low-rank covariance completion techniques before estimating the graph structure. We introduce three approaches to solve this problem: block singular value decomposition, nuclear norm penalization, and non-convex low-rank factorization. While prior works have studied low-rank matrix completion, we address the challenges brought by the block-wise missingness and are the first to investigate the problem in the context of graph learning. We discuss theoretical properties of the two-step procedure, showing graph selection consistency of one proposed approach by proving novel L infinity-norm error bounds for matrix completion with block-missingness. We then investigate the empirical performance of the proposed methods on simulations and on real-world data examples, through which we show the efficacy of these methods for estimating functional connectivity from calcium imaging data. | ['Genevera I. Allen', 'Lili Zheng', 'Andersen Chang'] | 2022-09-17 | null | null | null | null | ['low-rank-matrix-completion', 'matrix-completion'] | ['methodology', 'methodology'] | [ 4.75681484e-01 1.29423544e-01 7.57285878e-02 -4.67364304e-02
-5.58898449e-01 -5.07427037e-01 2.28718400e-01 -9.21253711e-02
-5.23402929e-01 9.02292073e-01 7.69457966e-02 -1.31345645e-01
-6.01200283e-01 -1.88063055e-01 -8.94411683e-01 -1.05765402e+00
-4.09990788e-01 4.51464683e-01 -2.59668261e-01 3.43537956e-01
2.49416023e-01 4.63505238e-01 -1.01765680e+00 -1.13756947e-01
6.88995600e-01 5.45531392e-01 2.34558597e-01 7.02720702e-01
3.85484099e-01 5.91115177e-01 -1.09421022e-01 -9.16841254e-02
2.20418632e-01 -3.81055564e-01 -6.46114826e-01 4.77559090e-01
4.62151080e-01 -1.09026715e-01 -6.08101964e-01 1.29985416e+00
4.58751231e-01 -1.79328024e-01 7.89628029e-01 -1.52107251e+00
-1.30208984e-01 5.96517980e-01 -9.51607704e-01 4.43318486e-01
1.59115449e-01 -3.75607818e-01 1.02672434e+00 -9.06606555e-01
7.06851006e-01 9.97034848e-01 5.02744198e-01 3.36569190e-01
-2.01541996e+00 -6.43825710e-01 1.45081699e-01 1.01324901e-01
-1.52892792e+00 -4.55425352e-01 7.60752201e-01 -9.16455984e-01
4.84979540e-01 -7.80273695e-03 6.70145571e-01 8.76898527e-01
1.79926872e-01 5.18617272e-01 1.34645367e+00 -2.13664919e-01
2.60130852e-01 -4.74450439e-01 5.11757314e-01 5.98947346e-01
8.14377367e-01 -2.98045009e-01 -5.46342015e-01 -5.26000500e-01
9.72181797e-01 6.79003913e-03 -5.27360857e-01 -7.77359068e-01
-1.55645394e+00 7.11527288e-01 1.17191866e-01 3.58545929e-01
-4.64087099e-01 2.98114568e-01 1.81570783e-01 1.43747553e-01
3.17733407e-01 -2.50993297e-02 -1.01380371e-01 2.99657911e-01
-8.77435505e-01 2.45726675e-01 8.57674539e-01 9.02831376e-01
7.59553015e-01 -3.90021205e-02 7.11267069e-02 5.13576448e-01
5.81408083e-01 4.52178001e-01 3.23641598e-02 -1.05077493e+00
6.26161575e-01 4.18663949e-01 -1.67378142e-01 -1.05923426e+00
-6.72673464e-01 -4.61722821e-01 -1.35941064e+00 2.00543948e-03
8.22774231e-01 -3.59083444e-01 -7.08681703e-01 2.09077883e+00
1.73174009e-01 5.48822761e-01 -1.55038819e-01 6.62534118e-01
4.26264286e-01 1.81421027e-01 -1.89589456e-01 -6.87977791e-01
1.07059145e+00 -2.30495334e-01 -8.48672390e-01 1.38047770e-01
6.21472239e-01 -1.73129052e-01 3.36420000e-01 5.24945021e-01
-1.02202988e+00 4.35761660e-02 -9.93311822e-01 1.74924821e-01
1.05412088e-01 9.09261405e-02 6.13164961e-01 5.05549192e-01
-1.26352179e+00 4.76007879e-01 -1.01250958e+00 -2.91015983e-01
5.77797353e-01 6.19490921e-01 -7.51072705e-01 -2.42072567e-01
-6.70557618e-01 3.42516690e-01 -9.91927609e-02 3.37373227e-01
-9.02655303e-01 -6.04964316e-01 -5.47230422e-01 5.42195402e-02
3.81273568e-01 -7.46665001e-01 4.55572486e-01 -6.75557971e-01
-7.38632441e-01 6.52401865e-01 -4.02892709e-01 -3.57288331e-01
4.08372968e-01 1.79265544e-01 1.58588663e-01 3.14190537e-01
2.23906368e-01 2.86751866e-01 8.34290862e-01 -1.25925338e+00
-6.52614608e-02 -9.32027996e-01 -5.25274687e-02 9.98301804e-02
-2.11354673e-01 -4.17295843e-01 -4.52149421e-01 -5.17425954e-01
7.12923110e-01 -1.09873235e+00 -4.53026265e-01 2.89959591e-02
-5.56561649e-01 1.82593718e-01 4.16441768e-01 -5.75011969e-01
9.77684021e-01 -2.09596729e+00 6.86786652e-01 6.26865566e-01
8.75634670e-01 -2.08427161e-01 -2.61379093e-01 5.03785670e-01
-4.19907033e-01 -6.22675614e-03 -6.17726386e-01 -3.02218944e-01
-3.16716135e-01 3.22490424e-01 9.67314839e-03 1.00374413e+00
-9.75998193e-02 6.17245197e-01 -7.98076928e-01 -3.74646962e-01
-2.04051882e-01 6.94427609e-01 -5.71315467e-01 -2.40529716e-01
4.36530739e-01 6.43576801e-01 -4.62553889e-01 2.81640172e-01
6.69985533e-01 -5.72604358e-01 6.83711708e-01 -5.56910157e-01
2.39812806e-02 -3.12078565e-01 -1.48141849e+00 1.61446571e+00
8.21584836e-02 4.64914650e-01 5.50007522e-01 -1.58913898e+00
3.28730166e-01 4.18444753e-01 9.56915379e-01 -2.02689052e-01
1.62308633e-01 1.15640201e-02 2.29636595e-01 -2.19199464e-01
-2.73915797e-01 2.23011691e-02 2.25798249e-01 5.14724970e-01
6.98882621e-03 2.92515665e-01 5.90538561e-01 7.69364476e-01
1.61863446e+00 -9.71899033e-02 3.17688346e-01 -5.30263901e-01
3.88638318e-01 -3.01711142e-01 5.04378617e-01 6.90208435e-01
-8.77782255e-02 5.72533786e-01 1.10880125e+00 1.30085588e-01
-8.54503274e-01 -1.00553453e+00 -1.77204385e-01 4.04467463e-01
-1.59016714e-01 -2.82641739e-01 -9.62371290e-01 -3.94388497e-01
-6.14348613e-02 -3.02381162e-02 -1.03166425e+00 6.83396906e-02
-2.28619576e-01 -1.12085092e+00 3.32985729e-01 2.15352148e-01
4.56056744e-02 -3.50435644e-01 -1.95358604e-01 1.63360998e-01
-2.55430341e-01 -1.30226099e+00 -3.78820151e-01 3.56572360e-01
-1.22115350e+00 -1.36797619e+00 -7.51769364e-01 -6.72511280e-01
1.25648654e+00 4.40827519e-01 7.01720774e-01 1.77856967e-01
-4.02308345e-01 7.30168819e-01 -6.71636313e-02 2.45821342e-01
1.39364481e-01 -3.11039358e-01 2.52851248e-01 3.29280257e-01
-1.40694857e-01 -1.01373267e+00 -4.57539827e-01 2.09962204e-01
-1.07714868e+00 7.27598220e-02 7.66593575e-01 9.08982098e-01
9.28213000e-01 -1.42271724e-02 5.80572426e-01 -1.15197456e+00
5.52860796e-01 -6.04538977e-01 -6.03901863e-01 1.73775747e-01
-4.75316525e-01 3.08926165e-01 2.78154671e-01 -4.04467136e-01
-4.47607040e-01 5.13845563e-01 3.79714370e-01 -4.89197612e-01
2.85544515e-01 8.38795364e-01 -3.10114950e-01 -4.76761848e-01
2.91815430e-01 2.03122720e-01 2.35312238e-01 -3.35345179e-01
2.71400422e-01 8.36421773e-02 2.54836977e-01 -4.80185747e-01
7.39428580e-01 7.80143738e-01 6.34695649e-01 -1.06524825e+00
-5.69974601e-01 -6.41493738e-01 -1.00675285e+00 -2.43749142e-01
7.09287167e-01 -9.63121951e-01 -8.97119582e-01 3.42228323e-01
-8.61541092e-01 -1.30143821e-01 5.97143136e-02 6.95757747e-01
-7.52237141e-01 8.46294522e-01 -5.44292331e-01 -6.77993953e-01
-3.03164637e-03 -9.82148349e-01 7.92826653e-01 -4.30862963e-01
1.58079609e-01 -1.02421308e+00 4.98445809e-01 2.84632683e-01
-1.60423547e-01 3.41205150e-01 9.85704362e-01 -2.45044455e-01
-4.30423647e-01 -5.87476864e-02 -4.14919198e-01 7.25576654e-02
-1.21891215e-01 -3.49701464e-01 -5.81123471e-01 -4.90058362e-01
7.61147887e-02 6.13881694e-03 8.03728223e-01 8.29639912e-01
8.39800477e-01 -1.75189272e-01 -4.26829964e-01 4.14113402e-01
1.53448796e+00 -3.01322848e-01 5.80412805e-01 -3.02172840e-01
9.33054686e-01 6.21488571e-01 1.53226376e-01 6.10416591e-01
3.01404864e-01 5.96215308e-01 5.18103719e-01 1.11173153e-01
-4.56270166e-02 9.46868360e-02 1.72096431e-01 1.06899786e+00
-2.27032349e-01 -8.52912962e-02 -7.29666889e-01 5.09681284e-01
-2.16094375e+00 -8.57124329e-01 -6.47154331e-01 2.40082860e+00
5.04892528e-01 -2.39613041e-01 2.16733620e-01 2.44103655e-01
9.50500607e-01 -2.46688113e-01 -6.31521940e-01 4.83363092e-01
-2.21541047e-01 -3.21816243e-02 6.63038015e-01 6.21742070e-01
-7.99635708e-01 3.30989361e-01 6.58006954e+00 4.98547941e-01
-4.64335084e-01 9.75618511e-02 3.85706872e-01 -2.43728533e-02
-2.39122733e-01 2.29279190e-01 -5.16964734e-01 1.87394127e-01
6.58645988e-01 -6.50956482e-02 7.26168275e-01 1.54425859e-01
4.06523138e-01 -2.94184446e-01 -1.09623039e+00 1.28033149e+00
2.36686885e-01 -1.05037570e+00 -1.26489356e-01 6.97604597e-01
7.12930441e-01 5.32304682e-02 -2.30624065e-01 -1.65734589e-01
1.48186505e-01 -8.15577686e-01 2.38101348e-01 6.63133442e-01
6.56571686e-01 -5.62022090e-01 3.11303169e-01 3.08462411e-01
-1.18503368e+00 1.16768904e-01 -3.95280957e-01 -1.17354669e-01
1.91772878e-01 1.08775294e+00 -4.67784435e-01 4.76677746e-01
3.29504430e-01 9.87842262e-01 -4.01946396e-01 1.14644289e+00
4.72270995e-02 5.99173129e-01 -4.50031877e-01 2.95917958e-01
-1.69558838e-01 -6.79599643e-01 6.04175329e-01 6.48821831e-01
2.59091884e-01 3.13795447e-01 -3.66006279e-03 8.04790437e-01
-1.07708082e-01 6.88744262e-02 -7.48294830e-01 -2.30180517e-01
8.53572264e-02 1.53390384e+00 -9.89219427e-01 -2.66154427e-02
-5.80153763e-01 7.53105879e-01 6.02148116e-01 6.27091825e-01
-4.28495854e-01 -3.01625300e-02 4.24449772e-01 2.60726273e-01
2.81271905e-01 -5.03320873e-01 -1.93322375e-01 -1.32825947e+00
1.49339335e-02 -7.01700032e-01 2.03360900e-01 -4.62504894e-01
-1.34912741e+00 2.15871543e-01 6.45527765e-02 -9.45308328e-01
2.33990997e-02 -5.68316817e-01 -1.99655131e-01 6.00821912e-01
-1.09874058e+00 -8.83394659e-01 -1.16054723e-02 8.42622995e-01
7.26060523e-03 2.53974915e-01 3.93878520e-01 5.24585843e-01
-7.24568009e-01 1.18856736e-01 5.06679237e-01 5.82999140e-02
2.58938283e-01 -1.24814558e+00 -1.10885791e-01 9.42087293e-01
3.64342391e-01 7.11116731e-01 6.30559146e-01 -8.66216660e-01
-1.84099758e+00 -7.60371387e-01 5.57832539e-01 -1.71851397e-01
9.98703659e-01 -6.37148619e-01 -7.77743638e-01 7.74694681e-01
-2.37837166e-01 2.93862164e-01 6.75813735e-01 1.54204387e-02
-1.24196053e-01 6.61061183e-02 -9.11264658e-01 4.95569706e-01
1.13014925e+00 -4.51940209e-01 -3.30919921e-02 4.38978076e-01
3.96623015e-02 1.71055496e-01 -1.08291757e+00 2.69480914e-01
5.81935704e-01 -7.10914254e-01 7.86156476e-01 -5.22862971e-01
-8.15765634e-02 -4.07834113e-01 -2.71273732e-01 -1.08107185e+00
-3.41003627e-01 -5.39926648e-01 -7.27519765e-02 1.02027166e+00
3.58127147e-01 -4.62449968e-01 9.59052742e-01 4.12203252e-01
2.05973014e-01 -5.39524376e-01 -1.08013391e+00 -6.62198186e-01
-1.10724732e-01 -3.57138515e-01 -3.00456107e-01 9.96255279e-01
2.33515918e-01 6.56598270e-01 -4.98126268e-01 2.58226097e-01
1.23094106e+00 -1.79441378e-01 6.39603198e-01 -1.37382460e+00
-4.36412901e-01 -3.24748829e-02 -6.89050615e-01 -8.67581964e-01
3.01553488e-01 -1.06501722e+00 1.84955858e-02 -1.49620342e+00
8.03859830e-01 -1.22332409e-01 -1.56627625e-01 2.80873448e-01
-2.41863523e-02 2.58602530e-01 -3.96065824e-02 2.37148404e-01
-5.73720813e-01 3.49886656e-01 1.18491364e+00 -7.87511319e-02
7.31787086e-02 -1.18582286e-01 -6.11960351e-01 5.61012626e-01
4.79901701e-01 -6.53968394e-01 -6.18742287e-01 -2.48646900e-01
6.32015347e-01 3.86848569e-01 4.66411889e-01 -1.08420599e+00
3.41284782e-01 2.37972453e-01 2.77837068e-01 -4.40479279e-01
3.56414527e-01 -9.55496967e-01 4.63097841e-01 4.79777485e-01
-2.06058696e-01 -8.46004486e-02 -1.73097670e-01 1.12453818e+00
4.52382080e-02 -2.63182491e-01 5.73633075e-01 7.84806907e-02
-1.22144587e-01 5.63975751e-01 -5.70053935e-01 1.84861705e-01
8.97863209e-01 -1.56028762e-01 -8.30469131e-02 -4.14687991e-01
-1.29714847e+00 1.09998174e-01 1.67282030e-01 -2.87153065e-01
7.22830951e-01 -1.23770571e+00 -7.97201335e-01 4.16089855e-02
-1.10402502e-01 -3.46799016e-01 4.93572116e-01 1.54332721e+00
-8.92246515e-02 2.32166901e-01 -2.65620500e-01 -7.26123393e-01
-1.35667813e+00 5.46434641e-01 -1.88868977e-02 -2.85396963e-01
-4.15820986e-01 4.83764440e-01 7.30923891e-01 -8.38299915e-02
9.77283195e-02 -1.41317964e-01 -3.64479065e-01 1.21263482e-01
3.01761687e-01 4.96775895e-01 5.35623282e-02 -8.23029399e-01
-3.20544630e-01 7.45193720e-01 -2.23042555e-02 -3.78650129e-01
1.39478338e+00 -4.92200196e-01 -5.36771595e-01 5.13094068e-01
1.27539945e+00 -1.91732243e-01 -1.26966715e+00 -2.90402621e-01
9.50308070e-02 -1.55468419e-01 2.94203669e-01 -1.21866651e-01
-1.35595369e+00 8.29075158e-01 4.91369396e-01 1.41555130e-01
9.41611052e-01 2.68165991e-02 6.21269085e-02 5.24433911e-01
5.27776778e-01 -6.33956134e-01 -3.24701406e-02 2.32882366e-01
8.41651976e-01 -8.17606390e-01 2.93701321e-01 -6.77759469e-01
-3.18495721e-01 8.08382452e-01 1.49321854e-01 -3.51155132e-01
9.76464331e-01 2.15947703e-01 -5.29583514e-01 -5.46777010e-01
-8.14505577e-01 -2.93139219e-01 9.73453894e-02 6.83517516e-01
3.32100242e-01 -8.57406929e-02 -5.02286613e-01 3.16064715e-01
3.53903890e-01 -7.06391633e-02 7.95280039e-01 7.48922944e-01
-2.23988503e-01 -1.00170946e+00 -3.14587891e-01 8.77131641e-01
-5.18611968e-01 -2.15208620e-01 -5.39050341e-01 5.93002081e-01
-3.14551175e-01 8.80385160e-01 -2.79317439e-01 -4.17412370e-02
1.99377209e-01 -2.06293344e-01 8.88887823e-01 -5.80552697e-01
7.82605708e-02 5.42138934e-01 -1.32698163e-01 -4.86208767e-01
-8.27436864e-01 -1.02225304e+00 -1.07790816e+00 8.58065486e-02
-4.27320659e-01 2.43630216e-01 6.39000773e-01 9.17482376e-01
2.63778627e-01 4.82168496e-01 4.33987647e-01 -9.61526692e-01
-2.67239988e-01 -8.49495173e-01 -1.29428673e+00 2.77149081e-01
3.02318782e-01 -8.06448936e-01 -6.53177440e-01 2.33733371e-01] | [7.080098628997803, 4.90025520324707] |
d9c47fe9-e293-4e95-98b7-c8591d2ff0dd | planet-photo-geolocation-with-convolutional | 1602.05314 | null | http://arxiv.org/abs/1602.05314v1 | http://arxiv.org/pdf/1602.05314v1.pdf | PlaNet - Photo Geolocation with Convolutional Neural Networks | Is it possible to build a system to determine the location where a photo was
taken using just its pixels? In general, the problem seems exceptionally
difficult: it is trivial to construct situations where no location can be
inferred. Yet images often contain informative cues such as landmarks, weather
patterns, vegetation, road markings, and architectural details, which in
combination may allow one to determine an approximate location and occasionally
an exact location. Websites such as GeoGuessr and View from your Window suggest
that humans are relatively good at integrating these cues to geolocate images,
especially en-masse. In computer vision, the photo geolocation problem is
usually approached using image retrieval methods. In contrast, we pose the
problem as one of classification by subdividing the surface of the earth into
thousands of multi-scale geographic cells, and train a deep network using
millions of geotagged images. While previous approaches only recognize
landmarks or perform approximate matching using global image descriptors, our
model is able to use and integrate multiple visible cues. We show that the
resulting model, called PlaNet, outperforms previous approaches and even
attains superhuman levels of accuracy in some cases. Moreover, we extend our
model to photo albums by combining it with a long short-term memory (LSTM)
architecture. By learning to exploit temporal coherence to geolocate uncertain
photos, we demonstrate that this model achieves a 50% performance improvement
over the single-image model. | ['Ilya Kostrikov', 'Tobias Weyand', 'James Philbin'] | 2016-02-17 | null | null | null | null | ['photo-geolocation-estimation'] | ['computer-vision'] | [-5.53558543e-02 -5.72027788e-02 -1.01765320e-01 -3.88579369e-01
-1.01539159e+00 -8.27799559e-01 8.72031093e-01 2.40751624e-01
-6.65714800e-01 5.55492759e-01 1.19365796e-01 2.32915580e-02
1.80581674e-01 -1.06980360e+00 -1.02336848e+00 -4.65506822e-01
-8.92499164e-02 4.10868406e-01 7.91942328e-02 -1.07125519e-02
4.62002426e-01 6.49822831e-01 -1.62264371e+00 -1.33450162e-02
3.80117923e-01 1.31277514e+00 1.69487849e-01 6.25731766e-01
-1.01424910e-01 6.33025944e-01 -5.94252467e-01 -3.20401341e-01
3.30656201e-01 1.37555683e-02 -8.94342840e-01 -1.55096119e-02
1.12968421e+00 -5.15454888e-01 -4.06275123e-01 1.00451839e+00
4.20821279e-01 2.49070808e-01 5.12495399e-01 -1.12666643e+00
-1.05840850e+00 4.14223224e-01 -6.65481985e-01 2.19914585e-01
5.22316694e-01 -4.07279693e-02 8.72728586e-01 -8.86116922e-01
6.38828695e-01 9.87053931e-01 1.14010978e+00 -2.14095578e-01
-8.70695889e-01 -2.20955551e-01 2.73792237e-01 1.78458095e-01
-1.96712983e+00 -4.15328294e-01 3.75807256e-01 -3.67700905e-01
8.81753683e-01 1.21502027e-01 3.95921975e-01 8.53027225e-01
1.03126923e-02 5.10070920e-01 8.60292375e-01 -3.88027370e-01
9.62428078e-02 -7.32418150e-02 -3.93792391e-01 8.42993140e-01
1.66554675e-02 -1.69630170e-01 -5.72762787e-01 -1.72778815e-01
8.95057857e-01 3.26170683e-01 -3.72140020e-01 -9.22342688e-02
-1.71939278e+00 4.24569190e-01 1.15850985e+00 5.75497806e-01
-5.63239992e-01 5.09521246e-01 -1.62057325e-01 9.66543257e-02
5.60114264e-01 6.36993110e-01 1.16430588e-01 1.40667543e-01
-1.23957181e+00 1.01778898e-02 7.96778142e-01 8.39499176e-01
1.25999081e+00 -3.35219875e-02 3.12082618e-01 6.38008296e-01
-1.49620906e-01 7.12783039e-01 4.36633438e-01 -1.33734107e+00
2.92770147e-01 2.91673303e-01 6.47391498e-01 -1.62799203e+00
-3.16273391e-01 -4.19588163e-02 -8.52766156e-01 9.19952244e-02
6.57634199e-01 -1.27856568e-01 -1.06607842e+00 1.74476695e+00
2.38402516e-01 5.71140409e-01 -3.27544272e-01 1.03981578e+00
4.18273032e-01 7.73582399e-01 2.00792596e-01 4.96049315e-01
1.27906287e+00 -7.54388154e-01 -1.01246946e-01 -6.79603934e-01
4.47574526e-01 -5.62390864e-01 6.26084983e-01 4.33999039e-02
-8.50522101e-01 -5.32166362e-01 -9.85120416e-01 -2.09189415e-01
-1.07448304e+00 1.73896044e-01 5.39615452e-01 2.07446307e-01
-1.63832545e+00 6.95158184e-01 -5.07420778e-01 -8.86991620e-01
5.82900271e-02 2.31211543e-01 -7.53820777e-01 -2.18835205e-01
-1.13172328e+00 1.02308142e+00 2.20617309e-01 3.09203357e-01
-4.59121287e-01 -3.25587124e-01 -1.01947725e+00 1.20525308e-01
4.93810624e-02 -8.87281060e-01 1.03634346e+00 -1.31497550e+00
-8.48188221e-01 1.11998928e+00 -1.95148572e-01 -4.61027771e-01
3.88770878e-01 7.48004541e-02 -3.59807163e-01 3.27659965e-01
3.55864137e-01 1.11682773e+00 6.95282578e-01 -1.03523302e+00
-8.18770885e-01 -3.99355173e-01 4.53180164e-01 3.09125781e-01
-2.50287443e-01 -1.16902292e-01 -4.88286048e-01 -4.85601693e-01
2.94912636e-01 -1.00102711e+00 -2.29217723e-01 4.30747092e-01
-3.45532060e-01 2.08474536e-04 2.85308063e-01 -7.52578437e-01
8.44482481e-01 -2.06344414e+00 -4.30585057e-01 3.12087148e-01
1.75917372e-01 -1.57776386e-01 -3.31253260e-01 4.98850137e-01
2.48266950e-01 3.34132522e-01 -6.32278100e-02 -2.26711720e-01
1.31277785e-01 4.47601117e-02 -4.18803990e-01 6.82538867e-01
-1.81043193e-01 1.06648910e+00 -1.05157220e+00 -3.84174019e-01
3.13427448e-01 7.02673614e-01 9.56456363e-02 -2.64899999e-01
-6.09021075e-02 -1.50276879e-02 -4.06363159e-01 8.54620337e-01
5.94152451e-01 -5.75807452e-01 7.03680515e-02 -6.46278784e-02
-2.46949315e-01 -1.44070499e-02 -1.14727116e+00 1.71640062e+00
-6.83093548e-01 1.06782794e+00 -5.12731895e-02 -8.06924880e-01
7.50860572e-01 1.36989072e-01 4.23142165e-01 -8.95897090e-01
-1.69761211e-01 1.07795507e-01 -8.70583236e-01 -3.00420105e-01
9.66373265e-01 2.66856134e-01 -3.00049394e-01 5.24344504e-01
-3.08649033e-01 -4.23693992e-02 -2.57536709e-01 -2.98845321e-02
9.64977860e-01 2.48272438e-02 2.19971269e-01 6.59271553e-02
1.14068314e-01 2.38790497e-01 2.36766070e-01 9.83817041e-01
-5.33869229e-02 1.04795647e+00 7.88790211e-02 -8.99094045e-01
-1.01658559e+00 -9.65162396e-01 1.04217216e-01 1.05532944e+00
4.75746095e-01 1.63785443e-02 -7.36669123e-01 -3.42631370e-01
1.49047479e-01 4.79319066e-01 -6.96223438e-01 2.95826405e-01
-4.38859582e-01 -4.79297459e-01 6.42427862e-01 5.58818042e-01
7.91134179e-01 -8.18609476e-01 -4.90069360e-01 1.47899002e-01
-4.56482261e-01 -1.01388359e+00 -6.42381966e-01 -3.64522964e-01
-2.78203458e-01 -9.72255051e-01 -1.05425596e+00 -8.31305325e-01
7.80711472e-01 8.28132093e-01 1.28898394e+00 2.69741684e-01
-1.25566065e-01 7.33886600e-01 -4.53419462e-02 -4.70131040e-02
3.04953724e-01 1.34381548e-01 3.55689153e-02 1.32719234e-01
4.77607578e-01 -6.88228548e-01 -9.19601738e-01 4.48139220e-01
-8.29769135e-01 -3.26757193e-01 5.15425026e-01 5.22997320e-01
7.25855529e-01 -1.90732837e-01 2.90349364e-01 -6.04843140e-01
8.67888927e-02 -8.35759759e-01 -6.76909447e-01 5.58412373e-01
-3.01836938e-01 -6.81897923e-02 2.86532462e-01 -2.58191288e-01
-7.17429161e-01 3.97109717e-01 1.87828436e-01 -2.86867142e-01
-2.73027301e-01 6.24229193e-01 2.02414334e-01 -5.60479224e-01
7.99855828e-01 2.44859308e-01 -5.11013031e-01 -2.39499763e-01
6.16369188e-01 6.97406709e-01 1.01470566e+00 -4.03184026e-01
8.10365140e-01 7.91498482e-01 -2.68882364e-01 -8.88801098e-01
-8.36918950e-01 -5.94479859e-01 -8.03498507e-01 -2.82874465e-01
7.71625340e-01 -1.25872529e+00 -6.47576272e-01 4.45846021e-01
-1.19744051e+00 -2.63559461e-01 1.43739656e-01 3.67987186e-01
-3.87920827e-01 2.72637546e-01 -3.66235256e-01 -6.08611822e-01
1.50664985e-01 -6.14469647e-01 1.43094838e+00 3.30239326e-01
-5.40249981e-02 -1.04233718e+00 -5.25053404e-02 6.99912235e-02
6.39630854e-01 5.80049634e-01 3.14365119e-01 -3.59412760e-01
-1.10625815e+00 -5.77820361e-01 -4.90935326e-01 -2.37447038e-01
2.67392583e-02 -2.47304186e-01 -1.15679216e+00 2.46488769e-02
-5.03335476e-01 -2.01694548e-01 1.00903893e+00 4.17808324e-01
1.16025388e+00 -5.27223885e-01 -6.46699965e-01 7.59663105e-01
1.66428471e+00 -2.36658588e-01 7.49169111e-01 6.28077149e-01
7.28220820e-01 5.56452155e-01 4.36712295e-01 3.23901087e-01
9.69365776e-01 5.99995196e-01 5.79029202e-01 -1.98541984e-01
5.84009215e-02 -4.75550205e-01 6.71620667e-02 1.75642237e-01
-1.87619403e-01 -5.83104610e-01 -1.06517839e+00 9.60131586e-01
-1.75492191e+00 -1.25832236e+00 1.69190228e-01 2.41998291e+00
2.68776745e-01 -2.58412868e-01 -1.71381921e-01 -4.79133457e-01
8.89929175e-01 5.49332857e-01 -5.46504974e-01 2.56432798e-02
-3.48889381e-01 -4.45328712e-01 1.06351519e+00 5.35555243e-01
-1.40523911e+00 9.45535839e-01 7.05287027e+00 4.76199538e-01
-1.42768872e+00 -9.50818881e-02 9.78503227e-01 6.67271018e-02
-3.91834855e-01 -5.69937788e-02 -4.93964553e-01 4.90978271e-01
8.69128168e-01 4.24921587e-02 6.33598566e-01 8.50188613e-01
-2.42375210e-02 -5.52896857e-01 -9.46204782e-01 1.20248258e+00
4.76355076e-01 -1.60254407e+00 -1.81537181e-01 1.59050394e-02
8.41405869e-01 5.17227054e-01 2.21663848e-01 -1.44872680e-01
4.68467712e-01 -1.20751023e+00 8.67233515e-01 9.82206643e-01
8.89470339e-01 -4.20847982e-01 4.88746047e-01 2.88954645e-01
-1.32115531e+00 -7.23428652e-02 -5.09974539e-01 -1.28094936e-02
6.70118108e-02 3.74581277e-01 -8.91011417e-01 3.23931217e-01
8.15203130e-01 5.92209399e-01 -8.11700404e-01 1.41469526e+00
-1.33668914e-01 1.47538120e-02 -7.04817474e-01 1.33340597e-01
5.45475125e-01 -3.18030715e-02 1.45219043e-01 9.86234426e-01
1.08300650e+00 6.50792941e-02 2.15567976e-01 6.56844318e-01
-2.72802204e-01 -2.05271423e-01 -1.10449862e+00 8.12875554e-02
8.71372044e-01 1.25407732e+00 -7.83351123e-01 -3.73736829e-01
-3.59520197e-01 1.27175689e+00 4.34719205e-01 6.19825244e-01
-6.62633836e-01 -5.61598003e-01 5.28757393e-01 3.23325843e-01
2.56175935e-01 -5.16543329e-01 8.90385360e-03 -1.09631836e+00
1.00735329e-01 -3.92272145e-01 3.66832495e-01 -1.53000844e+00
-1.10112906e+00 7.82281995e-01 -3.48291367e-01 -1.29519081e+00
-5.31201363e-01 -3.83537591e-01 -5.90662360e-01 9.10871625e-01
-1.58794463e+00 -1.56442606e+00 -7.22328484e-01 3.67364734e-01
1.25121027e-01 2.48326123e-01 6.96811676e-01 2.95198262e-01
-1.57208279e-01 3.30143362e-01 4.92183864e-01 2.95463592e-01
9.79982793e-01 -1.32911396e+00 7.98050165e-01 9.39433932e-01
6.31276369e-01 4.55668271e-01 5.37077308e-01 -3.67879331e-01
-1.15429592e+00 -1.09549189e+00 1.34263563e+00 -6.83332860e-01
7.11673975e-01 -1.56855524e-01 -6.60292089e-01 9.83817995e-01
1.36738196e-01 2.96373010e-01 1.30686641e-01 -5.94103485e-02
-6.47182465e-01 -1.31066069e-01 -1.02460790e+00 5.55993438e-01
1.00632250e+00 -9.11322773e-01 -3.08419138e-01 5.22924364e-01
4.48136985e-01 -3.34226638e-01 -7.19920635e-01 -2.68166214e-01
4.72607911e-01 -1.02381730e+00 1.26722920e+00 5.26038669e-02
3.62504497e-02 -4.71427530e-01 -2.16556028e-01 -1.37775195e+00
-2.66303390e-01 -4.54087466e-01 5.41045368e-01 1.17526674e+00
4.04594153e-01 -8.21899176e-01 7.31742382e-01 1.02725518e+00
1.25777110e-01 -9.37250778e-02 -9.62141573e-01 -5.71542919e-01
-2.47684136e-01 -2.75496274e-01 8.93754840e-01 1.13069630e+00
-2.06546247e-01 -1.98165864e-01 -4.64320242e-01 5.38947821e-01
3.87115926e-01 3.95724595e-01 5.64847052e-01 -1.10283720e+00
1.98575288e-01 -3.63319993e-01 -6.02805316e-01 -1.23430419e+00
1.49043903e-01 -6.14325166e-01 2.63033926e-01 -1.72644615e+00
-2.30085731e-01 -6.77851200e-01 -2.87136704e-01 6.79489791e-01
8.08647126e-02 8.21758926e-01 -6.36219606e-02 3.81799549e-01
-8.84460449e-01 -2.91595701e-02 5.45821011e-01 -4.48197842e-01
1.25728667e-01 -2.23651931e-01 -6.35988116e-01 7.33132184e-01
6.03878796e-01 -3.53413671e-01 -1.97379570e-03 -9.96008575e-01
6.26275003e-01 -2.07314417e-02 9.03655350e-01 -1.25728095e+00
8.50133002e-01 -2.29092687e-01 6.72039986e-01 -4.59928989e-01
6.23165131e-01 -7.39880979e-01 3.15979838e-01 -1.23261996e-01
-2.11283058e-01 2.46564612e-01 -4.33896221e-02 8.86957288e-01
-4.34237897e-01 -5.78251667e-02 2.98422813e-01 -5.20929933e-01
-1.32445312e+00 5.71438253e-01 -3.66038978e-01 -2.45392427e-01
7.35219777e-01 -4.68787998e-01 -5.64069569e-01 -7.67064869e-01
-4.27965760e-01 2.94549346e-01 9.72454607e-01 3.37698460e-01
3.57553989e-01 -1.43899071e+00 -4.31429863e-01 -1.83746099e-01
2.73921102e-01 -2.50764161e-01 4.61934030e-01 7.44914114e-01
-7.89790809e-01 3.49966943e-01 -2.19867736e-01 -4.20953572e-01
-8.00733030e-01 5.75121820e-01 5.47493279e-01 3.42916757e-01
-4.69433665e-01 8.01592469e-01 2.38321900e-01 -2.56359488e-01
1.77814975e-01 -2.47614235e-01 -8.74643773e-02 2.66005516e-01
7.59600222e-01 9.40841809e-02 -6.98890314e-02 -1.00224483e+00
-4.17943031e-01 8.57555389e-01 4.97093529e-01 -1.92909181e-01
1.39137828e+00 -6.46864176e-01 -1.80151999e-01 2.94004321e-01
1.18289614e+00 -5.77850491e-02 -1.30906391e+00 -4.40467954e-01
1.45873308e-01 -6.14100277e-01 3.61248963e-02 -7.35178351e-01
-9.74167705e-01 8.65719676e-01 5.60265660e-01 3.99227381e-01
9.84710574e-01 -1.20691709e-01 9.28289652e-01 8.73167455e-01
7.04477191e-01 -1.09423077e+00 -1.79090455e-01 3.80948097e-01
6.97194993e-01 -1.51064777e+00 -1.42964154e-01 2.51504898e-01
-5.05683720e-01 1.12687230e+00 3.08189720e-01 -1.64409697e-01
5.89399517e-01 -6.25816137e-02 2.48994514e-01 -1.83076978e-01
-3.07969242e-01 -4.66338784e-01 3.56806993e-01 7.40778804e-01
-3.15204561e-02 9.05812532e-03 5.70603490e-01 -6.64865002e-02
-1.87001392e-01 -2.11938635e-01 5.35668135e-01 5.92209816e-01
-7.89539874e-01 -6.38870895e-01 -6.68274343e-01 2.08593041e-01
-2.97842681e-01 -1.80386811e-01 -4.92686331e-01 7.51622319e-01
1.67560995e-01 8.13140452e-01 4.03197378e-01 -2.72353232e-01
-1.43388793e-01 -7.22051859e-02 1.39369190e-01 -2.26512298e-01
-3.02211165e-01 -4.90926504e-01 1.34098873e-01 -7.74504542e-01
-5.62067807e-01 -4.28541660e-01 -7.64508665e-01 -6.20592296e-01
4.41271886e-02 -9.15799513e-02 8.67817044e-01 8.52905512e-01
6.16867423e-01 -2.07001418e-01 4.93878335e-01 -1.31656146e+00
-3.85323241e-02 -6.28168583e-01 -6.10299826e-01 1.59476370e-01
5.81063330e-01 -4.90846336e-01 -3.56427491e-01 2.32415497e-01] | [7.683464050292969, -1.84306800365448] |
5d11aeb3-0c45-4276-8291-dbbb8d4530fa | analyzing-the-impact-of-varied-window-hyper | 2209.05804 | null | https://arxiv.org/abs/2209.05804v1 | https://arxiv.org/pdf/2209.05804v1.pdf | Analyzing the Impact of Varied Window Hyper-parameters on Deep CNN for sEMG based Motion Intent Classification | The use of deep neural networks in electromyogram (EMG) based prostheses control provides a promising alternative to the hand-crafted features by automatically learning muscle activation patterns from the EMG signals. Meanwhile, the use of raw EMG signals as input to convolution neural networks (CNN) offers a simple, fast, and ideal scheme for effective control of prostheses. Therefore, this study investigates the relationship between window length and overlap, which may influence the generation of robust raw EMG 2-dimensional (2D) signals for application in CNN. And a rule of thumb for a proper combination of these parameters that could guarantee optimal network performance was derived. Moreover, we investigate the relationship between the CNN receptive window size and the raw EMG signal size. Experimental results show that the performance of the CNN increases with the increase in overlap within the generated signals, with the highest improvement of 9.49% accuracy and 23.33% F1-score realized when the overlap is 75% of the window length. Similarly, the network performance increases with the increase in receptive window (kernel) size. Findings from this study suggest that a combination of 75% overlap in 2D EMG signals and wider network kernels may provide ideal motor intents classification for adequate EMG-CNN based prostheses control scheme. | ['Guanglin Li', 'Olumide Olayinka Obe', 'Mojisola Grace Asogbon', 'Oluwarotimi Williams Samuel', 'Frank Kulwa'] | 2022-09-13 | null | null | null | null | ['intent-classification'] | ['natural-language-processing'] | [ 2.90322691e-01 -1.15496784e-01 -3.03196251e-01 1.90363020e-01
-1.64765656e-01 1.26426339e-01 2.64033347e-01 -5.65818191e-01
-9.69474733e-01 7.10862339e-01 7.12755620e-02 -1.20016150e-01
-4.23360229e-01 -6.53605759e-01 -4.47552145e-01 -8.94928575e-01
-2.86998332e-01 -2.85273641e-01 8.17453638e-02 -1.64144129e-01
2.72049576e-01 6.46702707e-01 -1.67218173e+00 2.65844584e-01
6.79594934e-01 1.26426196e+00 6.20697379e-01 3.47504467e-01
5.64575568e-02 7.75083303e-02 -9.77677286e-01 1.73276931e-01
4.18696612e-01 -2.46445492e-01 -2.89346606e-01 -1.76128626e-01
-1.50651827e-01 -1.24891803e-01 -4.22901869e-01 8.52387667e-01
7.90099025e-01 2.89176643e-01 5.77407241e-01 -8.15205574e-01
-4.25149471e-01 5.64851284e-01 -8.26140419e-02 4.95990157e-01
-5.87468557e-02 6.54809296e-01 3.90766978e-01 -5.11528373e-01
5.03552794e-01 6.73314095e-01 6.79421425e-01 6.44915998e-01
-1.04665256e+00 -6.75632894e-01 -3.30554605e-01 2.84417629e-01
-1.24994254e+00 -2.97193229e-03 7.31921792e-01 -4.28869724e-01
1.18046963e+00 1.95417941e-01 1.12011862e+00 1.17478585e+00
6.14447892e-01 4.55807209e-01 1.00053072e+00 -4.05602843e-01
1.82499345e-02 1.88838523e-02 -7.47572482e-02 -1.69414192e-01
2.80355752e-01 5.44400096e-01 -3.51958752e-01 2.04952687e-01
1.34848487e+00 -5.22463024e-02 -4.90568012e-01 4.10829782e-01
-1.11600482e+00 5.82449019e-01 4.82151806e-01 7.27855265e-01
-8.53021264e-01 1.28782958e-01 5.83947122e-01 3.02217722e-01
2.05724791e-01 9.42323983e-01 -2.78225958e-01 -5.95314741e-01
-8.18236053e-01 1.88083619e-01 3.64895612e-01 4.20834422e-01
-8.61613676e-02 6.56587958e-01 -3.44685286e-01 8.87712955e-01
-2.99434423e-01 1.71816081e-01 9.14540350e-01 -7.91062474e-01
2.65340388e-01 6.45027399e-01 -1.87103987e-01 -8.00843060e-01
-4.21622157e-01 -6.81761980e-01 -7.28110015e-01 6.27743125e-01
5.76223195e-01 -2.78619975e-01 -8.61823082e-01 1.39247704e+00
-1.01053804e-01 -1.90400869e-01 -1.11262470e-01 1.31434536e+00
3.16417456e-01 3.31496179e-01 1.87876403e-01 -1.33638144e-01
1.46176469e+00 -3.90291035e-01 -9.62892532e-01 -8.86948034e-02
3.57554287e-01 -5.42598963e-01 1.17279339e+00 4.92338836e-01
-9.15354848e-01 -8.08486462e-01 -1.11477399e+00 6.57763720e-01
-2.05502450e-01 6.29393578e-01 6.54924273e-01 3.92913759e-01
-5.87787449e-01 1.26573348e+00 -9.08997834e-01 -1.89850271e-01
2.18510509e-01 9.53362584e-01 -4.32295650e-01 5.82151771e-01
-1.22422838e+00 1.15419531e+00 7.32298136e-01 4.59232897e-01
-1.67937130e-01 -5.46711683e-01 -3.07008445e-01 -3.02539486e-02
-1.01027116e-01 -3.11470687e-01 6.37766004e-01 -1.06482053e+00
-1.59043479e+00 4.72465813e-01 4.83424783e-01 -5.94061971e-01
3.05650294e-01 -3.89116764e-01 -4.27615494e-01 5.16987815e-02
-4.64218527e-01 5.69178820e-01 9.09269452e-01 -5.01135409e-01
-3.87641519e-01 -4.10854191e-01 -2.21315682e-01 1.08626232e-01
-6.85767651e-01 -4.64235321e-02 4.16752324e-02 -8.38812411e-01
-1.93309128e-01 -8.94418001e-01 -9.20022279e-03 -6.45327419e-02
-3.16398777e-02 -5.10400832e-01 7.43311763e-01 -8.30957294e-01
1.36074090e+00 -2.39140368e+00 1.62196904e-01 1.91730261e-01
3.28043923e-02 6.72783375e-01 -1.62204057e-02 1.77271515e-01
-3.21477175e-01 -2.51278728e-01 1.91138312e-01 7.16155648e-01
-3.40166539e-01 1.01267084e-01 3.52988780e-01 4.22234118e-01
4.66046214e-01 7.80793369e-01 -4.94121581e-01 -9.79894996e-02
4.42911506e-01 6.89683676e-01 -3.16099614e-01 3.59968543e-01
1.64991230e-01 1.73418015e-01 -2.34539628e-01 3.54708970e-01
3.53759766e-01 2.63699949e-01 1.24431349e-01 -6.05388522e-01
-1.06333464e-01 2.67327428e-01 -1.05977964e+00 1.32303286e+00
-2.95701802e-01 9.14450884e-01 8.02708566e-02 -9.97582912e-01
1.38675046e+00 5.90620518e-01 7.32955337e-01 -7.33120561e-01
6.12773716e-01 2.91881680e-01 6.68804169e-01 -7.83424735e-01
5.93414269e-02 -1.58699244e-01 4.49406803e-01 1.80816241e-02
1.83708295e-02 3.64529312e-01 -9.31267440e-03 -7.18997061e-01
8.85306358e-01 2.51264572e-01 2.27137841e-02 -3.07154059e-01
2.86958575e-01 8.97727348e-03 3.11066747e-01 3.09122443e-01
-1.16693549e-01 3.69222552e-01 2.86626875e-01 -4.80277479e-01
-9.49658155e-01 -7.46912181e-01 -2.36135215e-01 5.50654650e-01
-2.55603999e-01 9.40527543e-02 -8.48503113e-01 2.18950026e-03
7.51490965e-02 2.40908638e-01 -4.37301636e-01 -4.71356630e-01
-8.74779403e-01 -4.48823899e-01 6.17586493e-01 9.71343696e-01
3.04087102e-01 -1.65943456e+00 -9.63614404e-01 4.16334391e-01
1.31590307e-01 -9.11169291e-01 -2.79808730e-01 4.98032093e-01
-1.34365630e+00 -9.90690708e-01 -8.22511137e-01 -8.49933863e-01
6.07989907e-01 -2.49123812e-01 2.80348003e-01 -2.51311772e-02
-3.97373021e-01 -7.61784911e-02 -3.10425282e-01 -3.88716400e-01
-1.30966559e-01 2.49866068e-01 1.34020075e-01 -3.88877898e-01
5.05942583e-01 -7.35370636e-01 -6.62017107e-01 5.10031991e-02
-8.42436552e-01 -5.92793748e-02 1.01694942e+00 1.02672625e+00
4.15302992e-01 -4.07347269e-02 6.69040561e-01 -1.40329763e-01
1.33947515e+00 -7.41494223e-02 -1.46714598e-01 -2.52090227e-02
-6.50461257e-01 3.87579352e-02 7.00700879e-01 -1.17509294e+00
-6.96020305e-01 -1.71223342e-01 -2.57213205e-01 -6.94456697e-01
-2.45689213e-01 3.80125850e-01 1.98677450e-01 -2.69947723e-02
7.96107113e-01 1.69443741e-01 6.71816587e-01 -3.81682873e-01
-4.87491935e-02 9.21864748e-01 5.04809320e-01 -2.36285359e-01
1.61900818e-01 -7.11685270e-02 -2.22388893e-01 -8.38576257e-01
2.34756947e-01 -2.55275577e-01 -5.55372596e-01 -5.97710252e-01
7.85186410e-01 -4.95831370e-01 -9.06905651e-01 5.63461661e-01
-1.00019097e+00 -2.89463252e-01 -2.46541232e-01 1.21228957e+00
-5.87142110e-01 7.74881467e-02 -6.83436692e-01 -9.48194742e-01
-8.76928031e-01 -1.01148093e+00 5.18866897e-01 2.31406406e-01
-6.93434775e-01 -5.33830523e-01 -4.78206038e-01 1.50371939e-01
5.87537408e-01 4.39318359e-01 9.82252181e-01 -5.58128119e-01
2.01871153e-02 -5.50801754e-01 -7.44079659e-03 7.55304694e-01
2.37148359e-01 -2.23921239e-01 -6.30089164e-01 -1.50462002e-01
2.29499236e-01 6.94992319e-02 4.07931328e-01 6.94192171e-01
9.84454215e-01 -2.95375705e-01 -1.20550655e-01 3.32413256e-01
1.38501716e+00 6.24393523e-01 9.63713884e-01 4.54242647e-01
3.79015803e-01 5.14879704e-01 5.39347351e-01 2.86718726e-01
-5.65327883e-01 8.35059702e-01 3.65326852e-01 -1.35342717e-01
-3.97096395e-01 1.62575319e-01 2.91543901e-01 7.95055747e-01
-7.48375535e-01 3.85910898e-01 -5.29786587e-01 5.18763065e-01
-1.44900441e+00 -8.60532224e-01 -3.34760323e-02 2.30808759e+00
9.44411635e-01 3.88356447e-01 3.31084907e-01 5.20340204e-01
7.93277740e-01 -2.89157271e-01 -5.31085730e-01 -5.75645566e-01
-2.36088969e-02 6.99034154e-01 4.82529849e-01 -1.47342294e-01
-6.54573321e-01 4.45628554e-01 6.01957989e+00 1.03834629e+00
-1.69624352e+00 -9.21371728e-02 -5.03737070e-02 -5.77417493e-01
2.13652551e-01 -6.13999605e-01 -5.36784172e-01 7.67393708e-01
9.58598495e-01 7.30604008e-02 5.00204444e-01 8.07807028e-01
5.62960267e-01 -5.56875765e-02 -7.41228938e-01 8.29144895e-01
-3.53358567e-01 -1.26878500e+00 -2.52152115e-01 2.97852576e-01
3.23644608e-01 -6.78563640e-02 -2.44019121e-01 1.37091547e-01
-6.74071789e-01 -9.97448444e-01 5.28171182e-01 5.29418170e-01
7.09296942e-01 -6.46652699e-01 1.01210022e+00 2.06297964e-01
-9.80755210e-01 -3.24962050e-01 -2.03577682e-01 -2.47967005e-01
-2.37767324e-02 3.76049787e-01 -8.50587487e-01 1.87892005e-01
5.71597278e-01 3.52157205e-01 2.19887756e-02 1.14197433e+00
6.98142648e-02 5.39365649e-01 -3.29420924e-01 -6.58734858e-01
9.54814926e-02 -1.31946579e-01 6.32141650e-01 1.07777214e+00
3.95344019e-01 -1.15768701e-01 -3.08668584e-01 1.04016066e+00
3.38674307e-01 4.31669205e-02 -6.45334363e-01 -4.12943721e-01
5.22341371e-01 9.16482806e-01 -4.41305339e-01 1.32854760e-01
-9.70752984e-02 5.62960744e-01 9.24360007e-02 3.31238031e-01
-5.57167768e-01 -7.71338820e-01 8.49724650e-01 4.45787966e-01
2.00590283e-01 -3.27613086e-01 -5.77215791e-01 -3.89232635e-01
5.22263467e-01 -8.40695977e-01 -2.54878942e-02 -5.45710921e-01
-1.11755621e+00 6.83414936e-01 -1.10368863e-01 -1.47156143e+00
-2.86635816e-01 -1.17500401e+00 -7.83908010e-01 1.05418730e+00
-8.92298102e-01 -5.81018209e-01 -6.98572174e-02 4.76755798e-01
6.75684750e-01 -9.89471748e-03 8.80811453e-01 2.46745229e-01
-4.61108863e-01 3.66917968e-01 -2.18609422e-02 2.20856369e-01
2.02242762e-01 -8.05480361e-01 -7.08389804e-02 5.65456629e-01
-3.74018431e-01 8.03702295e-01 5.90237260e-01 -5.89301109e-01
-1.46940708e+00 -6.41793311e-01 5.68222761e-01 2.54630446e-01
4.10429657e-01 4.45420407e-02 -8.65736425e-01 -7.20167756e-02
-7.16269948e-04 -3.96641850e-01 5.01277328e-01 -1.02396667e-01
2.44813517e-01 -3.23521852e-01 -9.83330786e-01 8.32782865e-01
7.41792202e-01 -2.68594384e-01 -7.19706297e-01 -1.20713159e-01
2.40054578e-01 -2.79493332e-01 -1.47365940e+00 7.02623606e-01
1.20151067e+00 -5.70101380e-01 8.53808939e-01 -4.27614093e-01
3.27738047e-01 3.10995467e-02 2.92709947e-01 -1.26918340e+00
-2.59769261e-01 -1.90115437e-01 -2.67699249e-02 5.84204793e-01
4.65701908e-01 -6.65556371e-01 7.67754257e-01 5.66909492e-01
-3.27480912e-01 -1.43040490e+00 -1.09675241e+00 -9.92480516e-01
-2.04257965e-01 -4.73117948e-01 2.35124722e-01 2.98782170e-01
4.07069236e-01 -2.49057144e-01 -2.91625679e-01 -3.83353829e-01
1.19192600e-01 -3.36081892e-01 3.41940999e-01 -1.07045734e+00
-9.72118154e-02 -7.26146162e-01 -7.07108498e-01 -6.67991042e-01
-2.49070227e-01 -5.91865301e-01 9.52076614e-02 -1.40813088e+00
-4.46196377e-01 -1.81391060e-01 -5.23556709e-01 4.45537657e-01
-8.32659975e-02 2.64022276e-02 2.35478222e-01 2.71543831e-01
4.21706140e-01 3.48270833e-01 1.56942499e+00 1.00558167e-02
-6.12735808e-01 4.55801487e-01 -1.86407477e-01 1.69632912e-01
1.03492224e+00 -3.34776133e-01 -2.62949377e-01 -1.69247240e-01
-4.92033601e-01 5.30564524e-02 2.29171276e-01 -1.22121763e+00
1.59260795e-01 -6.93822727e-02 6.10553265e-01 -3.81845564e-01
4.28959191e-01 -8.45402181e-01 4.31627333e-01 8.40266943e-01
-3.37776601e-01 -1.33020997e-01 3.80305409e-01 1.26144916e-01
-1.36014700e-01 -2.22554326e-01 4.15029228e-01 -9.26287174e-02
-6.49677873e-01 -1.11126110e-01 -5.97844243e-01 -5.46754718e-01
1.00004625e+00 -1.22186613e+00 1.89607918e-01 1.67886600e-01
-8.33682120e-01 -3.06059718e-01 -1.45740926e-01 8.46121073e-01
7.14586616e-01 -1.43263483e+00 -3.94546747e-01 5.36429763e-01
-2.54815817e-01 -6.31373882e-01 2.20935941e-01 1.11895692e+00
-3.11121732e-01 5.70512652e-01 -1.03796768e+00 -4.52554882e-01
-1.24856591e+00 -1.54888025e-02 4.25502926e-01 4.35320847e-03
-8.68004203e-01 5.82626343e-01 -5.14668167e-01 1.00109354e-01
5.65413535e-01 -4.58717644e-01 -4.46588129e-01 -1.96474910e-01
3.70313287e-01 4.80641484e-01 2.47828156e-01 -1.37116596e-01
-1.86487317e-01 2.99398720e-01 3.23580742e-01 1.83517694e-01
1.42910337e+00 4.75136220e-01 9.10954252e-02 2.65624315e-01
9.18666124e-01 -5.56051373e-01 -1.35632622e+00 1.94015786e-01
-5.99150434e-02 -3.24063182e-01 1.35851964e-01 -7.90701747e-01
-1.23009992e+00 7.22014129e-01 1.02390432e+00 4.99140099e-02
1.33349371e+00 -3.37626636e-01 8.74929070e-01 -7.41290450e-02
4.20822620e-01 -1.37393618e+00 -1.07669696e-01 6.20504431e-02
1.08124137e+00 -6.42219365e-01 -1.63822278e-01 -1.58315361e-01
-6.40034199e-01 1.58053720e+00 8.96065593e-01 -4.52564806e-01
5.26573002e-01 3.78452241e-01 2.55474895e-02 -2.61108756e-01
-3.32628787e-01 -1.62322134e-01 5.40751517e-01 7.86661744e-01
5.15983939e-01 3.06474507e-01 -1.05824351e+00 9.78532612e-01
-1.74504668e-01 4.89330083e-01 -1.73723057e-01 8.04798722e-01
-4.83451754e-01 -9.83696640e-01 -3.39657784e-01 9.34583545e-01
-5.52136302e-01 -1.87874101e-02 -3.68182808e-02 1.12911177e+00
4.03739005e-01 6.56901240e-01 2.57882565e-01 -7.87514627e-01
6.81115746e-01 1.41678780e-01 6.47654355e-01 -4.22567278e-01
-1.08208728e+00 1.75112307e-01 1.16871431e-01 -3.24830294e-01
-2.75761247e-01 -1.69600129e-01 -1.31828976e+00 -2.17028350e-01
-8.80033493e-01 1.31884277e-01 8.83197725e-01 1.03528631e+00
3.22188079e-01 6.98998034e-01 2.71753728e-01 -9.36115742e-01
-7.73602366e-01 -1.53848815e+00 -6.43832743e-01 2.34693959e-01
-1.72837824e-01 -7.42999673e-01 -3.24147642e-01 -1.03972629e-01] | [6.859349250793457, 0.21105888485908508] |
33e72783-43ce-4e0c-a6b3-87423fcb0bb5 | audio-dequantization-using-co-sparse-non | 2010.16386 | null | https://arxiv.org/abs/2010.16386v2 | https://arxiv.org/pdf/2010.16386v2.pdf | Audio Dequantization Using (Co)Sparse (Non)Convex Methods | The paper deals with the hitherto neglected topic of audio dequantization. It reviews the state-of-the-art sparsity-based approaches and proposes several new methods. Convex as well as non-convex approaches are included, and all the presented formulations come in both the synthesis and analysis variants. In the experiments the methods are evaluated using the signal-to-distortion ratio (SDR) and PEMO-Q, a perceptually motivated metric. | ['Ondřej Mokrý', 'Pavel Rajmic', 'Pavel Záviška'] | 2020-10-30 | null | null | null | null | ['audio-dequantization'] | ['audio'] | [ 2.67743379e-01 2.39277706e-02 -1.45498320e-01 -1.18349232e-02
-1.16692901e+00 -2.50747353e-01 1.87579423e-01 -1.04650140e-01
-2.37889513e-01 7.92108238e-01 5.64807415e-01 2.18601048e-01
-2.85482556e-01 -4.99698035e-02 -1.85822889e-01 -7.22135723e-01
-4.53910261e-01 -1.36453867e-01 1.47430748e-01 -2.12285474e-01
2.63219982e-01 3.94986689e-01 -1.65247262e+00 2.05380559e-01
6.04813933e-01 1.30132651e+00 -1.31220698e-01 7.97464073e-01
5.09669840e-01 3.75052482e-01 -8.14504564e-01 -4.86693323e-01
2.36294895e-01 -6.23294413e-01 -3.05399507e-01 3.14310163e-01
3.43435526e-01 9.50644091e-02 -1.83652088e-01 1.38967144e+00
9.90851760e-01 1.06575578e-01 5.33346117e-01 -1.15382397e+00
-3.30788076e-01 5.89446306e-01 -5.82832932e-01 5.74367642e-01
6.45200491e-01 -2.45327011e-01 1.21093452e+00 -1.35777295e+00
4.80761707e-01 1.13095939e+00 6.88256383e-01 -4.11782600e-03
-1.01592290e+00 -4.13674891e-01 -4.41789031e-02 5.20918608e-01
-1.62148428e+00 -6.74327374e-01 9.56529975e-01 -3.08942437e-01
6.51146173e-01 5.24649262e-01 5.83074331e-01 4.18849647e-01
1.82405099e-01 8.26171398e-01 1.30426872e+00 -7.01533794e-01
2.41025031e-01 1.61013171e-01 -9.97665823e-02 6.28460884e-01
7.42533728e-02 3.00332934e-01 -8.61820161e-01 -1.66270375e-01
4.63132322e-01 -9.33957279e-01 -5.81862509e-01 -3.26730050e-02
-1.06146288e+00 5.71649909e-01 -1.97112307e-01 4.59651411e-01
-3.07068914e-01 2.11022105e-02 4.67186093e-01 4.87647265e-01
8.66862178e-01 3.49282742e-01 -3.04397553e-01 -4.20906544e-01
-1.70620084e+00 3.01861346e-01 6.59805179e-01 9.25367236e-01
1.78008646e-01 9.50399578e-01 -2.79928058e-01 9.10927773e-01
2.46927217e-01 5.78717768e-01 1.79014325e-01 -9.41216230e-01
4.87837434e-01 -5.65641463e-01 8.14773887e-02 -1.17825377e+00
-3.13532799e-01 -7.67071187e-01 -6.92803919e-01 6.00002557e-02
-3.69587354e-02 -2.66688734e-01 -5.33955276e-01 1.32787788e+00
1.34118527e-01 2.60727346e-01 -1.44304052e-01 1.06097054e+00
9.53853011e-01 8.06305110e-01 -3.63275826e-01 -9.26849067e-01
1.04330528e+00 -1.02869856e+00 -1.41925395e+00 4.40250158e-01
-1.42438427e-01 -1.32467771e+00 6.12490296e-01 1.16778731e+00
-1.87133038e+00 -5.05679905e-01 -1.28400159e+00 1.14806965e-01
7.29102865e-02 3.41595381e-01 7.03332871e-02 1.05235255e+00
-1.04011214e+00 6.41491950e-01 -4.33489591e-01 2.40178823e-01
1.32347912e-01 3.30332667e-01 4.81642671e-02 1.45930484e-01
-1.15535593e+00 7.82306969e-01 1.57573307e-03 1.70291394e-01
-1.14529455e+00 -7.19638109e-01 -5.85321307e-01 1.20272309e-01
4.77864772e-01 -4.71405238e-01 1.27194178e+00 -6.83661580e-01
-2.11899447e+00 6.15458608e-01 -2.49427140e-01 -4.60859895e-01
3.77201408e-01 -4.35376525e-01 -9.27010536e-01 3.50186557e-01
-3.37502331e-01 1.12014942e-01 1.42885280e+00 -9.77250457e-01
-6.94797754e-01 -1.37569876e-02 -1.07410155e-01 2.31357679e-01
-6.58135563e-02 4.06626225e-01 -1.73544735e-01 -1.39892125e+00
9.05509889e-02 -5.93284369e-01 8.50231573e-02 -2.42320389e-01
-5.95826149e-01 5.18835522e-02 8.39355946e-01 -9.62746143e-01
1.77568543e+00 -2.25922632e+00 5.87550402e-01 1.63723066e-01
6.49894327e-02 3.58830482e-01 -1.85982972e-01 7.97120869e-01
-3.31945539e-01 -8.26060027e-02 -2.10511714e-01 -7.00690627e-01
2.03655094e-01 -2.44984731e-01 -3.87046248e-01 9.29138124e-01
1.96411628e-02 1.56109989e-01 -7.49435425e-01 -3.61093938e-01
3.49039942e-01 4.66622531e-01 -5.77533484e-01 1.38828352e-01
2.59120882e-01 2.12185472e-01 5.50847091e-02 8.16263318e-01
9.03767943e-01 5.63731730e-01 -1.16751775e-01 -4.13610250e-01
-4.63061243e-01 2.52047718e-01 -1.57649112e+00 1.63487160e+00
-3.64629269e-01 8.90335917e-01 5.77632546e-01 -1.31300068e+00
7.52217710e-01 7.97539592e-01 5.30517638e-01 -2.54660130e-01
6.47097230e-02 5.12055635e-01 -4.50096652e-02 -4.23985064e-01
6.64732933e-01 -3.76068860e-01 2.70411074e-01 1.43185213e-01
4.31082010e-01 -6.36547387e-01 4.11245108e-01 -6.82244897e-02
5.96600771e-01 -2.95636028e-01 7.46755242e-01 -6.67908490e-01
8.53864372e-01 -4.93176043e-01 4.48591650e-01 3.01313281e-01
-2.01946706e-01 7.10149229e-01 4.14430529e-01 1.56370014e-01
-1.16912913e+00 -1.15735102e+00 -4.13006067e-01 9.59923983e-01
-3.42175029e-02 -5.70878625e-01 -8.81030321e-01 5.03073866e-03
-2.37633646e-01 5.94879329e-01 -9.67133567e-02 1.91139057e-01
-2.93082029e-01 -4.94364887e-01 5.84679723e-01 6.78728074e-02
2.82890737e-01 -5.92666388e-01 -1.95183650e-01 3.62077057e-01
-2.50577956e-01 -1.28442883e+00 -5.89008808e-01 -3.09635680e-02
-9.04184520e-01 -6.65935218e-01 -1.09686422e+00 -5.89154959e-01
1.68003246e-01 1.17622852e-01 9.29438591e-01 -5.14480770e-01
-1.93586394e-01 5.56637108e-01 -5.75490236e-01 -6.79967940e-01
-1.72090411e-01 -2.51771003e-01 3.86912823e-01 3.57700258e-01
-1.02607764e-01 -7.56488740e-01 -3.65330547e-01 2.84673214e-01
-6.32115483e-01 -5.60008109e-01 2.85182953e-01 7.24949479e-01
8.31946492e-01 3.79982412e-01 7.01564550e-01 -4.64780778e-01
1.00865078e+00 -1.62943915e-01 -6.65291786e-01 -1.04454890e-01
-5.33963084e-01 -2.34076604e-01 6.29205644e-01 -5.20949066e-01
-8.27959418e-01 -1.11004770e-01 -3.02655667e-01 -4.58739012e-01
3.03487748e-01 5.96912861e-01 -2.68670350e-01 -3.72426867e-01
5.21109343e-01 2.58257151e-01 -4.00465220e-01 -5.91716588e-01
4.69829470e-01 7.21758246e-01 5.47717631e-01 -3.60992879e-01
8.95096958e-01 4.68060106e-01 1.20167568e-01 -1.51444948e+00
-7.83617735e-01 -7.04191566e-01 -2.50920683e-01 -4.04241681e-01
4.69081283e-01 -8.39729786e-01 -3.20598602e-01 3.98692787e-01
-1.13658404e+00 2.47967288e-01 -7.28324234e-01 8.71622682e-01
-8.85152698e-01 3.97841752e-01 -3.46873373e-01 -1.22363234e+00
-3.18301171e-01 -1.23585117e+00 8.87092590e-01 -1.17613569e-01
-8.93863514e-02 -9.06069815e-01 3.78424555e-01 -9.57661271e-02
4.38143373e-01 1.12781674e-01 5.59367716e-01 -1.39775708e-01
-2.24025205e-01 1.99867394e-02 6.01904355e-02 8.40176105e-01
-2.70478398e-01 -5.24429269e-02 -1.28775501e+00 -3.64498943e-01
3.26280951e-01 -2.83451355e-03 6.46370173e-01 7.48250365e-01
9.65034485e-01 -4.03405577e-01 2.00559989e-01 5.43071210e-01
1.37682843e+00 1.87381819e-01 7.76317000e-01 -1.88680723e-01
5.68815097e-02 3.96550566e-01 7.31179118e-01 9.72434044e-01
-3.46673548e-01 1.01926196e+00 3.67900968e-01 -1.42299861e-01
-2.93603241e-01 5.76779731e-02 5.05366147e-01 1.37225759e+00
-3.18395197e-01 -5.08150041e-01 -1.28575251e-01 7.63034403e-01
-1.48421478e+00 -1.10762393e+00 -1.17920801e-01 2.09599280e+00
5.72305620e-01 -8.90099332e-02 3.77139300e-01 1.13972151e+00
8.47080827e-01 5.18861592e-01 3.62422504e-02 -9.53917623e-01
-2.08356485e-01 6.89023793e-01 5.69432259e-01 6.78782761e-01
-1.17382348e+00 3.85633171e-01 7.92312145e+00 1.29526365e+00
-1.15549231e+00 4.14367110e-01 9.38359946e-02 -1.05316214e-01
-2.90844347e-02 -3.09483737e-01 -4.33929354e-01 1.88779727e-01
1.01168656e+00 -5.20692945e-01 5.83961070e-01 7.22209871e-01
6.05686665e-01 -7.47071207e-02 -8.47071230e-01 1.50559151e+00
2.81620741e-01 -1.04355228e+00 -3.61710012e-01 -6.66409880e-02
6.80821955e-01 -3.92049313e-01 5.39386809e-01 -1.17181666e-01
-6.69064164e-01 -8.87912035e-01 1.13350034e+00 5.17066061e-01
1.12665188e+00 -8.14113021e-01 6.51577294e-01 -6.05481640e-02
-1.40951264e+00 -5.62278144e-02 -3.00682306e-01 2.65452135e-02
5.42366028e-01 1.15448201e+00 -4.27884609e-01 8.34378779e-01
4.47409809e-01 8.28289449e-01 5.07992767e-02 1.42159331e+00
-4.46429551e-01 9.48277891e-01 -3.23595703e-01 4.70237136e-02
1.51126057e-01 -1.20465398e-01 1.25679111e+00 1.54747760e+00
6.64391637e-01 3.48888248e-01 -1.90873351e-03 5.78078747e-01
1.12958170e-01 4.08674449e-01 -3.10948431e-01 -6.39013574e-02
2.63468534e-01 9.98180985e-01 -2.36164421e-01 -1.61487803e-01
-2.82622933e-01 6.37618423e-01 -5.47493458e-01 2.93802738e-01
-7.12484956e-01 -8.50580096e-01 5.55780351e-01 3.78075898e-01
4.31277007e-01 -2.55969882e-01 -1.29465491e-01 -1.02911270e+00
5.51204123e-02 -1.01844823e+00 1.92308590e-01 -5.97955942e-01
-1.07376802e+00 5.01366377e-01 4.13707316e-01 -1.61288309e+00
-1.75740585e-01 -4.32747126e-01 -2.99137741e-01 8.26457262e-01
-1.67545462e+00 -3.84324014e-01 6.37727678e-02 5.23981929e-01
6.99395776e-01 -4.09986824e-01 6.96166992e-01 8.73281419e-01
-3.23081076e-01 7.88728774e-01 2.53468066e-01 -4.82854873e-01
5.40126681e-01 -1.06263947e+00 -3.45644280e-02 9.15639341e-01
3.11650962e-01 2.75484562e-01 1.21033657e+00 -1.02449367e-02
-1.14740217e+00 -6.12250209e-01 9.90655482e-01 3.36989254e-01
7.42857933e-01 -5.76021224e-02 -5.38665473e-01 3.66482837e-03
4.25227433e-01 8.76230467e-03 6.10675097e-01 -1.20438799e-01
-8.29065870e-03 -2.32976913e-01 -1.23011518e+00 2.86916286e-01
7.84188211e-01 -6.59563780e-01 -4.84119862e-01 3.26705575e-01
3.66272300e-01 -6.51141703e-01 -9.30804133e-01 2.32910559e-01
5.20938754e-01 -1.16012204e+00 1.01418531e+00 -2.80252367e-01
1.28712878e-01 -3.03759277e-01 -4.54913616e-01 -1.40478027e+00
-2.18753532e-01 -1.44402039e+00 -2.68957168e-01 9.65827286e-01
4.12900358e-01 -4.34908241e-01 4.78782982e-01 -5.07908165e-01
-5.63418210e-01 -6.89881504e-01 -1.60018504e+00 -1.18845880e+00
-3.57894152e-01 -6.26725972e-01 1.89062864e-01 5.41458309e-01
4.10937041e-01 3.52781415e-01 -8.40631068e-01 2.16067601e-02
6.59716368e-01 -3.31606388e-01 3.59386027e-01 -9.56080019e-01
-5.22225201e-01 -4.47911918e-01 -9.39657509e-01 -1.00063670e+00
-1.95559874e-01 -5.85292101e-01 -9.70664844e-02 -1.06728244e+00
-3.73492777e-01 4.48477007e-02 -4.01109248e-01 -2.17666775e-01
3.40320021e-01 4.28260595e-01 4.14892107e-01 -1.92864031e-01
-3.95456612e-01 5.94866216e-01 1.26964605e+00 -2.13377610e-01
-2.48641759e-01 3.44747901e-01 -4.23799872e-01 7.02330053e-01
6.04556203e-01 -5.96618414e-01 -5.77984035e-01 -2.21111789e-01
1.22785732e-01 2.90998220e-01 2.56312303e-02 -1.27812660e+00
-8.25563893e-02 3.20518285e-01 -3.49289417e-01 -8.60229790e-01
8.52060020e-01 -7.70660639e-01 8.13646168e-02 2.52181977e-01
-3.84900361e-01 8.94925967e-02 3.22678536e-02 5.32560408e-01
-8.67085874e-01 -3.90762419e-01 1.20147991e+00 2.52707958e-01
-1.16625905e-01 1.52040794e-01 -3.44103545e-01 3.27057004e-01
8.61971259e-01 -2.67671376e-01 8.15355256e-02 -9.15849090e-01
-1.07049632e+00 -5.18799901e-01 -3.13203454e-01 -6.96210340e-02
9.52357173e-01 -1.31621563e+00 -1.11625671e+00 -9.34075266e-02
-1.37036487e-01 -8.12439203e-01 2.67748982e-01 1.26571929e+00
-4.13077980e-01 7.00902760e-01 -1.18145697e-01 -4.63531971e-01
-1.48590124e+00 6.02162480e-01 1.68210343e-01 -3.00419539e-01
-4.38276589e-01 1.04512298e+00 -2.12680921e-01 4.10816103e-01
4.86954272e-01 -3.51796985e-01 -2.27578595e-01 1.17448261e-02
6.04895890e-01 1.09433913e+00 3.05012226e-01 -7.61491835e-01
-2.80813783e-01 6.91965461e-01 4.55092460e-01 -6.92358851e-01
1.19752276e+00 -5.53257614e-02 -1.12527564e-01 5.85977733e-01
1.36592948e+00 4.44312155e-01 -5.65657377e-01 -1.64014414e-01
-3.13885272e-01 -7.71268308e-01 4.62890059e-01 -5.99124372e-01
-1.02057612e+00 1.20113397e+00 8.79804909e-01 5.33509493e-01
1.63946569e+00 -4.73469466e-01 8.18661809e-01 -4.85682525e-02
4.13629830e-01 -1.42078745e+00 5.50537407e-02 3.01173300e-01
1.32920074e+00 -5.82027793e-01 7.42695928e-01 -8.51442873e-01
-4.11453038e-01 1.10878038e+00 -4.33578134e-01 -2.09220439e-01
1.10346818e+00 4.93746400e-01 -1.92719907e-01 2.68833011e-01
-4.24163043e-01 -1.84120417e-01 6.14587247e-01 1.03460550e+00
6.21276677e-01 -3.32915597e-03 -7.96490312e-01 4.41527098e-01
-3.84291708e-01 -6.72776178e-02 6.85236335e-01 5.38275599e-01
-4.28741038e-01 -1.15720415e+00 -6.24210954e-01 7.97424540e-02
-1.00140846e+00 -2.61404335e-01 -7.68061727e-02 7.15697169e-01
1.46005616e-01 1.48229718e+00 -5.06329238e-01 -4.07920748e-01
6.13194108e-01 -3.97331983e-01 5.37581801e-01 -4.80429232e-01
-6.81309462e-01 6.96896434e-01 3.98612291e-01 -4.52397376e-01
-8.06173623e-01 -8.09015214e-01 -7.18974352e-01 -2.81882823e-01
-7.43491292e-01 1.47049382e-01 7.28924513e-01 5.65597475e-01
3.54575482e-03 7.62746692e-01 9.29991066e-01 -9.42895830e-01
-5.57491601e-01 -9.95091200e-01 -1.17510760e+00 -8.71862769e-02
6.23350680e-01 -7.31928825e-01 -4.24135804e-01 1.22862905e-01] | [15.445869445800781, 5.622194766998291] |
f8a284fa-22df-408e-b686-141770949c7f | diffusion-based-speech-enhancement-with-joint | 2305.10734 | null | https://arxiv.org/abs/2305.10734v1 | https://arxiv.org/pdf/2305.10734v1.pdf | Diffusion-Based Speech Enhancement with Joint Generative and Predictive Decoders | Diffusion-based speech enhancement (SE) has been investigated recently, but its decoding is very time-consuming. One solution is to initialize the decoding process with the enhanced feature estimated by a predictive SE system. However, this two-stage method ignores the complementarity between predictive and diffusion SE. In this paper, we propose a unified system that integrates these two SE modules. The system encodes both generative and predictive information, and then applies both generative and predictive decoders, whose outputs are fused. Specifically, the two SE modules are fused in the first and final diffusion steps: the first step fusion initializes the diffusion process with the predictive SE for improving the convergence, and the final step fusion combines the two complementary SE outputs to improve the SE performance. Experiments on the Voice-Bank dataset show that the diffusion score estimation can benefit from the predictive information and speed up the decoding. | ['Yuki Mitsufuji', 'Tatsuya Kawahara', 'Shusuke Takahashi', 'Zhi Zhong', 'Yuichiro Koyama', 'Takashi Shibuya', 'Masato Hirano', 'Kazuki Shimada', 'Hao Shi'] | 2023-05-18 | null | null | null | null | ['speech-enhancement'] | ['speech'] | [ 2.36316636e-01 -1.25190735e-01 2.53643602e-01 -1.73605680e-01
-8.03174615e-01 -1.06395394e-01 5.71701825e-01 -2.09759489e-01
-5.45314968e-01 4.50325221e-01 4.55516040e-01 -9.61660668e-02
1.75549492e-01 -7.22665489e-01 -1.05304360e-01 -1.05304718e+00
3.79002750e-01 -5.89961559e-02 6.18123412e-01 -2.50880271e-01
1.34536490e-01 5.81337549e-02 -1.44111633e+00 4.47511166e-01
1.31805801e+00 1.05921221e+00 7.46528983e-01 8.31247330e-01
-4.11903888e-01 5.85760295e-01 -6.78895652e-01 -3.66665244e-01
-1.30233541e-01 -1.07642758e+00 -8.43640044e-02 8.92614201e-02
-6.31874144e-01 -3.54608536e-01 -6.41634390e-02 1.33300745e+00
9.39035475e-01 1.33748576e-01 4.41436589e-01 -8.80231440e-01
-7.99493074e-01 5.82745671e-01 -4.08110976e-01 2.27427185e-01
1.92089930e-01 -1.03461538e-02 6.54738724e-01 -9.86737072e-01
3.24783534e-01 1.23846769e+00 5.84694803e-01 5.48837721e-01
-6.68848097e-01 -6.75424695e-01 -2.28160545e-02 1.85326830e-01
-1.22479713e+00 -7.62571156e-01 9.94820535e-01 -3.01086575e-01
7.66317546e-01 2.62088180e-02 8.54711771e-01 8.45677912e-01
1.70934528e-01 1.04188299e+00 1.14200485e+00 -4.16175187e-01
3.20923835e-01 3.22490096e-01 7.92340934e-02 6.15267575e-01
-1.04719087e-01 3.68570507e-01 -5.25200307e-01 1.91611290e-01
6.63208544e-01 -1.86147153e-01 -4.31166202e-01 3.72357726e-01
-8.73382926e-01 8.03268671e-01 6.54750392e-02 6.66878283e-01
-7.25093067e-01 -3.64076614e-01 1.51970372e-01 2.82822043e-01
7.01082647e-01 -3.67382988e-02 -1.54069781e-01 -4.47349101e-01
-1.38383710e+00 -2.80530423e-01 7.86961198e-01 6.29042804e-01
7.32498705e-01 1.71481505e-01 -2.79608130e-01 1.10508430e+00
6.94628656e-01 7.23225355e-01 6.69183195e-01 -5.61756492e-01
4.58055764e-01 2.46816725e-01 -1.00783944e-01 -8.24785829e-01
-9.56041366e-02 -6.89076424e-01 -9.20158207e-01 4.57907915e-02
5.40948138e-02 -5.98060191e-01 -8.30732346e-01 1.53508079e+00
2.33649686e-01 1.69531301e-01 3.36509109e-01 9.51437354e-01
8.96415353e-01 1.18567264e+00 9.06444415e-02 -6.27243102e-01
1.05025148e+00 -1.30921221e+00 -1.38782609e+00 -2.00777277e-01
3.34398240e-01 -1.16281223e+00 5.89429319e-01 3.84397328e-01
-1.23769951e+00 -7.43839800e-01 -1.24689531e+00 2.61289235e-02
1.72033329e-02 5.14860868e-01 1.77667573e-01 8.50343943e-01
-1.24058032e+00 4.95859325e-01 -7.13210166e-01 -9.67312530e-02
3.06198280e-02 3.29218656e-01 2.05042344e-02 1.27142996e-01
-1.40718722e+00 7.67066956e-01 3.04239184e-01 3.83878797e-01
-5.42802453e-01 -8.82883146e-02 -6.11683130e-01 2.71770477e-01
1.41944557e-01 -5.56429923e-01 1.26206315e+00 -1.23314857e+00
-2.22000670e+00 5.79382889e-02 -4.71558124e-01 -7.30917230e-02
5.02521098e-01 -1.66872546e-01 -7.14227200e-01 5.85011877e-02
-2.37708315e-01 4.16321844e-01 1.14645278e+00 -1.09722412e+00
-8.73389602e-01 -1.19007587e-01 -4.08610940e-01 4.78788167e-01
-6.55273378e-01 9.14167464e-02 -9.07957256e-01 -6.79410577e-01
3.82084936e-01 -4.73889947e-01 -3.78003828e-02 -2.82093406e-01
-8.00730288e-02 3.77069116e-02 8.49056304e-01 -1.22556531e+00
1.66268539e+00 -2.41576743e+00 2.89861709e-01 1.58390909e-01
-4.09530737e-02 5.20150602e-01 -1.30044863e-01 2.86151499e-01
1.50129020e-01 -1.59881771e-01 -2.49293193e-01 -7.52489507e-01
-8.28867629e-02 1.45711288e-01 -9.50047299e-02 9.51215252e-02
2.71929383e-01 6.40962422e-01 -7.62375891e-01 -5.53086042e-01
1.58257902e-01 7.52046347e-01 -6.06307209e-01 4.43470001e-01
1.93715349e-01 5.42824328e-01 -4.79138732e-01 3.76089215e-01
8.56323957e-01 8.15726295e-02 2.96553344e-01 -1.27650216e-01
-3.52446020e-01 3.37495118e-01 -1.38739598e+00 1.44370031e+00
-4.33031380e-01 5.04134476e-01 2.89288938e-01 -6.88384056e-01
1.23759973e+00 5.10481834e-01 3.59427571e-01 -7.67022729e-01
2.84533083e-01 3.62741381e-01 7.60579556e-02 -4.80151266e-01
6.96641743e-01 -3.49886447e-01 3.63080710e-01 5.09644806e-01
1.72404349e-01 -1.15409225e-01 -5.72447442e-02 1.50972307e-01
6.67649269e-01 1.60926759e-01 2.52814353e-01 2.15769380e-01
8.74990702e-01 -3.14365834e-01 7.94858754e-01 2.89743155e-01
-1.76545709e-01 6.28004313e-01 4.22993660e-01 3.40493321e-01
-9.36280966e-01 -1.00078535e+00 2.61581928e-01 7.60717750e-01
4.71666843e-01 -4.94137287e-01 -1.02798533e+00 -6.70614779e-01
-5.37634134e-01 5.80799162e-01 -2.01155946e-01 -3.39163810e-01
-4.25592214e-01 -9.09811020e-01 3.95493329e-01 4.50731188e-01
8.08650196e-01 -7.95644701e-01 -8.86300877e-02 5.70448637e-01
-3.76241773e-01 -7.21573889e-01 -5.80013096e-01 6.86704218e-02
-8.53030980e-01 -3.64055455e-01 -1.12418342e+00 -6.39397442e-01
4.84856576e-01 2.34861895e-01 3.81216317e-01 1.71054691e-01
6.62684619e-01 -1.20388292e-01 -7.22443819e-01 -8.48083422e-02
-8.56022239e-01 2.39640363e-02 -2.03179017e-01 2.73046672e-01
1.72970369e-01 -3.59955758e-01 -4.85254407e-01 2.61982173e-01
-7.23931253e-01 2.89317250e-01 9.53516424e-01 8.90541136e-01
4.96939152e-01 1.84280097e-01 7.27476120e-01 -4.25563902e-01
1.00650167e+00 -3.35606039e-01 -4.74246770e-01 1.45267472e-01
-7.51254082e-01 2.29828969e-01 5.37038088e-01 -5.26812553e-01
-1.59631610e+00 5.55978157e-03 -1.02941144e+00 -1.57965422e-01
1.93795964e-01 6.49907351e-01 -5.29465079e-01 1.20027691e-01
1.77937493e-01 5.65204382e-01 3.50123569e-02 -7.64861166e-01
1.72551930e-01 1.16014409e+00 2.71093339e-01 -7.57589787e-02
5.84643185e-01 -1.26441002e-01 -5.51463127e-01 -7.25730002e-01
-2.81383783e-01 -2.94656396e-01 -3.83431077e-01 -4.52677816e-01
8.56180191e-01 -8.21712136e-01 -3.61818671e-01 7.91367948e-01
-1.37930596e+00 -2.43951064e-02 -2.36747012e-01 8.98318231e-01
-1.98762566e-01 6.26888335e-01 -7.75916159e-01 -9.30079281e-01
-4.83646482e-01 -1.57966495e+00 8.78068924e-01 5.22455394e-01
2.52651572e-01 -8.65106225e-01 1.12067632e-01 -1.28007829e-01
5.39185107e-01 -7.50261128e-01 4.60403085e-01 -6.12404287e-01
-4.13013965e-01 -1.06709041e-02 -2.35117391e-01 7.25085437e-01
7.05488026e-02 1.20119527e-01 -1.05964231e+00 -1.10049658e-01
4.16327477e-01 3.60946178e-01 8.84798944e-01 4.68562305e-01
4.89275247e-01 -7.47395828e-02 -1.12544894e-01 5.77390254e-01
1.15892839e+00 7.93709695e-01 7.20276594e-01 2.28698120e-01
1.75996840e-01 2.92049468e-01 7.80662000e-01 3.60878170e-01
3.14361125e-01 4.74249840e-01 -5.29485308e-02 -2.57632613e-01
-5.36085606e-01 -2.33773932e-01 8.55207801e-01 1.91347349e+00
-1.93928540e-01 -5.89737296e-01 -6.69079542e-01 3.27060133e-01
-1.70873761e+00 -1.03146219e+00 -1.00905016e-01 1.98964643e+00
8.68747532e-01 8.74358192e-02 -1.11165717e-01 3.34910095e-01
9.88434076e-01 2.87893087e-01 -1.97783425e-01 -4.10193264e-01
-1.60369113e-01 8.64644945e-02 7.33291134e-02 6.81613326e-01
-8.39877546e-01 8.01649809e-01 6.42774343e+00 1.03188598e+00
-1.27906692e+00 6.25779867e-01 4.57848340e-01 1.16613582e-01
-3.88666600e-01 -3.36426310e-02 -8.47465456e-01 6.79297984e-01
9.23215628e-01 -8.69224072e-02 4.24961537e-01 5.88495612e-01
2.95344889e-01 -2.63401210e-01 -4.40101534e-01 8.70538533e-01
3.82834375e-02 -9.56552327e-01 -1.88776150e-01 -8.03909265e-03
6.54908299e-01 -2.48737156e-01 1.08932473e-01 2.71166116e-01
1.27391472e-01 -3.55456918e-01 8.68107080e-01 8.86539698e-01
6.39870346e-01 -7.66464591e-01 1.04436123e+00 6.61973894e-01
-1.32938504e+00 -1.90163121e-01 -2.52716452e-01 1.43649846e-01
6.07224524e-01 9.16147768e-01 -4.74002928e-01 8.36017609e-01
3.63680393e-01 6.70871317e-01 -2.00157359e-01 1.04947007e+00
-6.55524015e-01 7.43668914e-01 -1.93210781e-01 -1.19046800e-01
1.29359022e-01 -6.26636267e-01 7.15123236e-01 1.27358031e+00
8.23426962e-01 1.63948596e-01 -1.69232503e-01 7.32923925e-01
2.03594595e-01 2.30362520e-01 -1.28994286e-01 -1.23622850e-01
3.25475425e-01 1.10206258e+00 -4.63430315e-01 -6.03805184e-01
-4.64319199e-01 1.22598350e+00 9.42490995e-02 4.30813670e-01
-9.46599364e-01 -5.26417255e-01 3.42433959e-01 -4.08407032e-01
3.97073865e-01 -3.57195497e-01 -5.21319449e-01 -1.13892186e+00
-1.72845125e-01 -7.50606954e-01 9.91449039e-03 -8.41823339e-01
-7.91421294e-01 8.96049857e-01 -3.88446778e-01 -1.16890264e+00
-3.84817868e-01 -1.34454921e-01 -4.92949992e-01 1.44386029e+00
-1.72450185e+00 -7.84481585e-01 -1.78755194e-01 4.03844923e-01
7.43741274e-01 -4.22360674e-02 4.04200763e-01 7.12334871e-01
-8.28183353e-01 5.12586057e-01 3.81272763e-01 -1.92955017e-01
7.09640563e-01 -8.97665918e-01 9.83992070e-02 1.16389883e+00
-1.30504861e-01 4.20419425e-01 5.18083155e-01 -8.50381851e-01
-9.46804106e-01 -8.51726711e-01 1.20579708e+00 3.99092227e-01
3.20438504e-01 9.72051080e-03 -1.05477154e+00 1.43363953e-01
3.48455876e-01 -5.18238008e-01 4.66904163e-01 -1.58133462e-01
2.83408880e-01 -1.13582037e-01 -9.68073606e-01 4.34035033e-01
7.55473375e-01 -5.38704693e-01 -7.53286898e-01 -2.44829223e-01
6.38758481e-01 -3.58095050e-01 -9.38708127e-01 1.46135673e-01
5.01503825e-01 -9.51119006e-01 6.41303599e-01 2.76470840e-01
3.72449756e-01 -4.84535694e-01 -6.81381375e-02 -1.57331920e+00
-4.20017689e-01 -5.67090154e-01 -5.23804538e-02 1.66152585e+00
5.93227923e-01 -7.33835995e-01 1.80495694e-01 1.40702248e-01
-3.72052401e-01 -5.59731662e-01 -8.17323506e-01 -5.82455099e-01
-2.99487382e-01 -5.18902004e-01 7.11025178e-01 5.74526846e-01
-5.51714934e-02 5.26706457e-01 -5.60365736e-01 2.67260522e-01
1.14532761e-01 -1.09927088e-01 4.07692999e-01 -9.15659487e-01
-4.86271352e-01 -5.14634907e-01 3.36970389e-02 -1.68005514e+00
-2.87504703e-01 -6.22496128e-01 4.97878194e-01 -1.71953166e+00
3.87844965e-02 -3.91735196e-01 -1.75016686e-01 2.18732834e-01
-6.80747271e-01 -9.03436840e-02 2.50242800e-01 3.76656801e-01
-2.29866803e-01 9.61143672e-01 1.39759088e+00 6.33993465e-03
-5.44669509e-01 6.08881712e-02 -5.10674179e-01 5.95109999e-01
6.51488602e-01 -5.06595254e-01 -4.82049346e-01 -4.76954192e-01
-1.72498330e-01 2.60973960e-01 -5.60092479e-02 -1.18221474e+00
6.22410655e-01 1.37257054e-01 3.03510994e-01 -7.52825320e-01
5.39493322e-01 -6.80849671e-01 1.30187511e-01 4.65728492e-01
-1.69442326e-03 -8.24904665e-02 6.68060407e-02 4.83603567e-01
-6.16854429e-01 -3.68377000e-01 8.36479008e-01 2.53536612e-01
-6.90918744e-01 1.56700820e-01 -5.82843781e-01 -4.85464364e-01
8.60772848e-01 -4.05330658e-01 -1.44529715e-01 -5.22281289e-01
-8.28145802e-01 -1.45026982e-01 1.29309118e-01 2.29738519e-01
7.12044299e-01 -1.26727533e+00 -6.53521895e-01 5.27311206e-01
-5.26759863e-01 -4.88676190e-01 3.96996528e-01 1.29006302e+00
-2.23089948e-01 -2.10935697e-02 2.67891795e-03 -5.00807166e-01
-1.18231010e+00 4.93200570e-01 2.86836058e-01 -3.29105884e-01
-3.34936619e-01 8.77184153e-01 1.53571457e-01 2.65342705e-02
4.49636728e-02 -9.71753597e-02 -5.33229113e-01 1.84066400e-01
5.59306681e-01 3.51298213e-01 -2.65966002e-02 -7.88093209e-01
-2.76991129e-02 6.55247331e-01 1.24026895e-01 -7.46826172e-01
1.24823797e+00 -5.89550614e-01 -3.79791968e-02 3.35714728e-01
1.08111393e+00 2.78711677e-01 -1.14742160e+00 -2.69892246e-01
-2.69154370e-01 -4.43687320e-01 4.64462668e-01 -8.54402125e-01
-1.24992204e+00 1.28554535e+00 6.52093947e-01 1.94303602e-01
1.77309000e+00 -3.95540595e-01 1.04045916e+00 -2.02869698e-01
-2.57318560e-02 -1.16092026e+00 -1.11063011e-01 6.17669284e-01
5.82043946e-01 -6.94747031e-01 -2.45567545e-01 -6.84323370e-01
-7.67083228e-01 1.16385603e+00 2.82899886e-01 3.71682674e-01
5.60821056e-01 5.21137536e-01 2.27289367e-02 2.22448155e-01
-7.47523367e-01 -4.46325034e-01 4.81135666e-01 5.14829755e-01
4.40147668e-01 -1.99219272e-01 -1.02598846e+00 1.10727680e+00
8.50351676e-02 1.66951671e-01 1.17065407e-01 7.91718662e-01
-7.03549206e-01 -1.49985993e+00 -5.33013761e-01 1.03710838e-01
-3.70998174e-01 -4.04643357e-01 -1.61781102e-01 2.88367867e-01
3.02215964e-01 1.32713091e+00 -9.90575105e-02 -8.12611997e-01
2.76937962e-01 7.49922842e-02 2.27535143e-02 -2.23246917e-01
-7.39259899e-01 7.84585416e-01 1.94048405e-01 -2.39394009e-01
-4.49644595e-01 -6.67857111e-01 -1.18189216e+00 -2.46692598e-01
-8.55537415e-01 3.61232460e-01 9.45304096e-01 1.00741208e+00
3.98366421e-01 7.88146555e-01 8.65609407e-01 -5.92082560e-01
-3.79142582e-01 -1.14347959e+00 -6.12784147e-01 1.74330417e-02
2.01018915e-01 -5.54827750e-01 -3.75371486e-01 2.24689469e-02] | [14.982165336608887, 5.985536098480225] |
e85bea85-c7ae-471f-86ae-65f001ea5daa | unsupervised-action-localization-crop-in | 2111.07426 | null | https://arxiv.org/abs/2111.07426v2 | https://arxiv.org/pdf/2111.07426v2.pdf | Unsupervised Action Localization Crop in Video Retargeting for 3D ConvNets | Untrimmed videos on social media or those captured by robots and surveillance cameras are of varied aspect ratios. However, 3D CNNs usually require as input a square-shaped video, whose spatial dimension is smaller than the original. Random- or center-cropping may leave out the video's subject altogether. To address this, we propose an unsupervised video cropping approach by shaping this as a retargeting and video-to-video synthesis problem. The synthesized video maintains a 1:1 aspect ratio, is smaller in size and is targeted at video-subject(s) throughout the entire duration. First, action localization is performed on each frame by identifying patches with homogeneous motion patterns. Thus, a single salient patch is pinpointed per frame. But to avoid viewpoint jitters and flickering, any inter-frame scale or position changes among the patches should be performed gradually over time. This issue is addressed with a polyBezier fitting in 3D space that passes through some chosen pivot timestamps and whose shape is influenced by the in-between control timestamps. To corroborate the effectiveness of the proposed method, we evaluate the video classification task by comparing our dynamic cropping technique with random cropping on three benchmark datasets, viz. UCF-101, HMDB-51 and ActivityNet v1.3. The clip and top-1 accuracy for video classification after our cropping, outperform 3D CNN performances for same-sized random-crop inputs, also surpassing some larger random-crop sizes. | ['Partha Pratim Mohanta', 'Swarnabja Bhaumik', 'Prithwish Jana'] | 2021-11-14 | null | null | null | null | ['video-to-video-synthesis'] | ['computer-vision'] | [ 2.39065588e-01 -1.60891742e-01 -1.98250085e-01 1.75768867e-01
-3.33960861e-01 -7.73003340e-01 3.95442277e-01 -1.70766458e-01
-5.00433028e-01 6.44804657e-01 -6.04372099e-03 4.84886914e-02
2.38939703e-01 -5.59439719e-01 -1.02601898e+00 -9.44815755e-01
-2.00397044e-01 -1.53940603e-01 6.35462999e-01 2.25300729e-01
6.25388026e-02 4.52640265e-01 -1.50197518e+00 2.79538214e-01
4.13078278e-01 1.21213233e+00 3.18494946e-01 6.61461711e-01
1.88880146e-01 8.88466716e-01 -9.84893501e-01 -1.72376052e-01
4.97524977e-01 -2.60693818e-01 -2.73185402e-01 6.77488267e-01
5.32103300e-01 -5.90702891e-01 -5.07898986e-01 1.08379126e+00
2.39300549e-01 1.18027151e-01 3.52788717e-01 -1.48569632e+00
-3.38907778e-01 3.40816706e-01 -8.46668720e-01 4.10524219e-01
5.68841994e-01 2.28159174e-01 5.26536286e-01 -8.44281852e-01
1.06436908e+00 1.09601986e+00 5.94704747e-01 3.17066044e-01
-9.78228807e-01 -6.64583087e-01 4.57204163e-01 2.07674921e-01
-1.56747055e+00 -2.22658366e-01 9.65864778e-01 -4.44042623e-01
7.19951689e-01 1.43017322e-01 9.33936894e-01 1.30782008e+00
3.16314489e-01 6.89531505e-01 5.31116962e-01 -1.57831326e-01
2.55141854e-01 -2.09409505e-01 -4.42792833e-01 4.83307362e-01
2.70023495e-01 -2.09282234e-01 -5.65007448e-01 1.68022200e-01
1.11916375e+00 3.33826721e-01 -6.92413926e-01 -5.21774888e-01
-1.75058758e+00 4.30688292e-01 2.06212476e-01 2.68859804e-01
-5.31983376e-01 6.98075220e-02 5.26489854e-01 3.40600371e-01
4.43466872e-01 1.87900513e-01 -4.97553110e-01 -2.14483768e-01
-1.10447752e+00 3.39644611e-01 3.46693873e-01 1.44323492e+00
5.98523378e-01 1.70706481e-01 -2.65286028e-01 4.15644318e-01
-2.72892892e-01 5.02323925e-01 4.60823119e-01 -9.83211100e-01
5.25177717e-01 5.42099833e-01 3.03941637e-01 -1.41117048e+00
-4.39546347e-01 -2.72077829e-01 -1.04296088e+00 9.57361385e-02
5.94938695e-01 -2.94107765e-01 -7.41604209e-01 1.61607623e+00
5.37092388e-01 4.73191530e-01 -6.89385384e-02 1.17872131e+00
7.56577432e-01 7.71311879e-01 -1.55429751e-01 -6.50996745e-01
1.39808095e+00 -9.50208843e-01 -7.32989609e-01 1.98849425e-01
4.22120422e-01 -5.80069423e-01 7.07119882e-01 5.96629739e-01
-1.03951097e+00 -8.81291628e-01 -1.01291895e+00 1.55922323e-01
-1.15505882e-01 2.17036948e-01 -5.64610697e-02 2.83810914e-01
-1.01910257e+00 3.59849930e-01 -5.85603774e-01 -3.47552687e-01
2.59715199e-01 1.33616120e-01 -6.58928454e-01 1.30319353e-02
-1.05314553e+00 3.88180912e-01 3.99016559e-01 -7.86904693e-02
-1.08676660e+00 -6.77949250e-01 -7.02319086e-01 -8.10817555e-02
6.72602654e-01 -5.43596148e-01 1.07328165e+00 -1.37100697e+00
-1.24905384e+00 6.95931673e-01 -2.73527224e-02 -5.26522100e-01
7.67899573e-01 -1.19517848e-01 -3.27910542e-01 5.29024839e-01
1.76965758e-01 8.21714938e-01 1.43154788e+00 -1.01412535e+00
-8.36534798e-01 -1.96836188e-01 2.92048573e-01 3.75049412e-01
-2.32026979e-01 1.99435558e-02 -9.47800338e-01 -1.03239393e+00
1.01226278e-01 -1.04398811e+00 4.96952385e-02 8.07336271e-02
-3.91229719e-01 1.08621724e-01 1.12159932e+00 -4.73060876e-01
1.30750859e+00 -2.35784984e+00 3.29834044e-01 -2.15899646e-01
2.99873978e-01 9.37594026e-02 -5.35518192e-02 8.56818557e-02
-1.23332992e-01 -6.09307252e-02 1.95841473e-02 -2.08748847e-01
-4.43148583e-01 -7.31642470e-02 -1.48642555e-01 6.41008496e-01
2.13335186e-01 6.23297811e-01 -1.01625919e+00 -5.20663500e-01
4.10588712e-01 4.22788650e-01 -6.42964482e-01 1.38628021e-01
-2.75202572e-01 5.55846691e-01 -2.96013832e-01 8.97724092e-01
7.95076728e-01 -1.71209931e-01 6.09067567e-02 -5.16259074e-01
-2.80512333e-01 -3.23174059e-01 -1.28659475e+00 1.77161622e+00
-6.39987960e-02 9.09477293e-01 1.21384207e-02 -8.88001680e-01
8.45654845e-01 4.24355268e-01 9.73648131e-01 -5.19840896e-01
2.40329266e-01 -5.14126271e-02 -3.64079505e-01 -7.24177957e-01
6.85390055e-01 5.52399397e-01 6.80463389e-02 9.31491181e-02
-1.68551710e-02 2.63949275e-01 2.96243429e-01 -1.58081148e-02
1.18033159e+00 3.30075681e-01 3.78914684e-01 -1.33803874e-01
4.30215329e-01 -7.11049438e-02 7.50892639e-01 5.34586549e-01
-4.66326147e-01 9.08534408e-01 7.73817718e-01 -5.47486961e-01
-1.14319265e+00 -6.66011214e-01 6.85797110e-02 8.18553150e-01
4.96285498e-01 -2.97174990e-01 -1.02327514e+00 -8.24669182e-01
-2.36010343e-01 1.73445806e-01 -7.63384938e-01 1.05064526e-01
-7.14646518e-01 -3.57894123e-01 3.90901327e-01 2.62505829e-01
6.68061852e-01 -1.17906916e+00 -1.07627201e+00 3.83535892e-01
-4.93590623e-01 -1.55268705e+00 -7.61230290e-01 3.22162211e-02
-7.29918480e-01 -1.29930985e+00 -1.15972209e+00 -7.30606258e-01
7.93234944e-01 7.86068559e-01 9.04715121e-01 -1.70936748e-01
7.71981180e-02 2.62273014e-01 -7.98763871e-01 -6.18371041e-03
-1.31650656e-01 -6.21382752e-03 1.96708038e-01 3.63280684e-01
2.26640955e-01 -5.45135081e-01 -8.32210243e-01 6.22295499e-01
-1.07109964e+00 2.13144526e-01 2.49492675e-01 5.88388443e-01
4.97668654e-01 3.38960648e-01 2.33481497e-01 -1.89008802e-01
7.83826336e-02 -4.99211043e-01 -5.12590408e-01 -2.31753923e-02
1.44498408e-01 -5.58345735e-01 7.16330290e-01 -1.07666719e+00
-5.14575839e-01 1.63732201e-01 4.67433184e-01 -1.23035920e+00
-1.78906798e-01 2.32548155e-02 -2.74421066e-01 1.01847373e-01
4.95777041e-01 3.30739975e-01 -1.97108164e-02 -9.54046100e-02
1.21793628e-01 4.06903416e-01 5.77449024e-01 -1.00952588e-01
7.69760251e-01 6.39868915e-01 -2.46328548e-01 -1.04297423e+00
-2.87992775e-01 -1.41456649e-01 -6.98718548e-01 -7.19961405e-01
1.15660715e+00 -1.06077707e+00 -6.66427255e-01 8.73938680e-01
-1.43395817e+00 -1.90448567e-01 -1.00002058e-01 4.89735246e-01
-4.24466252e-01 3.18636537e-01 -3.87383223e-01 -5.44685900e-01
-1.03896819e-02 -1.36226428e+00 1.10603082e+00 2.37625048e-01
-1.05622083e-01 -5.86799324e-01 -3.01858723e-01 9.73087922e-02
4.10928875e-02 4.85689014e-01 4.73398864e-01 -3.56256753e-01
-8.16504002e-01 -1.46671399e-01 -1.49855435e-01 1.88644022e-01
2.92186528e-01 2.91119903e-01 -8.08976769e-01 -3.43196839e-01
2.28596665e-03 1.18270703e-01 5.81772327e-01 6.68176889e-01
1.11433637e+00 -3.72905880e-01 -2.74582982e-01 6.64396346e-01
1.04226482e+00 6.67380273e-01 4.98821795e-01 5.42625666e-01
6.75157189e-01 3.28233153e-01 8.67980003e-01 6.11859441e-01
1.50258526e-01 9.18878973e-01 7.10288465e-01 -2.69566998e-02
-2.88893003e-02 -2.03078926e-01 6.29808605e-01 3.91083688e-01
-2.15981737e-01 -5.13383806e-01 -3.85145098e-01 4.75796372e-01
-1.74675083e+00 -1.03697848e+00 9.13778543e-02 2.32159328e+00
4.78158057e-01 3.35242033e-01 2.71462739e-01 2.50507414e-01
9.23264861e-01 5.14472961e-01 -6.32497907e-01 2.25797161e-01
-3.31163496e-01 -5.44888914e-01 7.44507968e-01 1.62594438e-01
-1.35021043e+00 8.58823717e-01 5.20094252e+00 9.11569476e-01
-1.30835497e+00 -8.30596015e-02 6.58005178e-01 -2.95149446e-01
1.76075235e-01 -3.73178542e-01 -7.85607219e-01 8.33961189e-01
3.41123223e-01 1.27988890e-01 4.13396031e-01 7.84811616e-01
6.59243822e-01 -3.48997325e-01 -1.07273757e+00 1.19234788e+00
1.99957445e-01 -1.39490223e+00 1.62828322e-02 -9.42278206e-02
8.39268148e-01 -1.42533630e-01 -5.16190305e-02 7.05209300e-02
-3.11322659e-01 -5.30439496e-01 1.12045038e+00 3.32263738e-01
8.30656290e-01 -6.68744981e-01 4.53171998e-01 2.43555963e-01
-1.39600885e+00 -9.77817029e-02 -2.95718133e-01 1.03252053e-01
2.47597158e-01 4.96045977e-01 -5.33317208e-01 4.73713189e-01
9.38750744e-01 9.57957983e-01 -3.61289501e-01 9.13235307e-01
1.05965592e-01 1.71246514e-01 -1.93671003e-01 7.22493753e-02
2.53105193e-01 -2.11355031e-01 9.18825030e-01 1.16679156e+00
6.61966980e-01 3.26287113e-02 8.74464884e-02 4.46023971e-01
-1.27890393e-01 -4.23892178e-02 -8.34577739e-01 3.04966658e-01
6.28172934e-01 1.04553163e+00 -1.03324497e+00 -4.71272469e-01
-6.20230377e-01 1.14404225e+00 -2.51257688e-01 5.88213563e-01
-1.44703460e+00 -1.71567932e-01 6.44967675e-01 3.04395020e-01
6.71884298e-01 -3.54890049e-01 2.93335795e-01 -1.28080332e+00
2.52811581e-01 -1.06686819e+00 1.60543546e-01 -9.92646337e-01
-5.68338215e-01 8.74559999e-01 1.89898103e-01 -1.97489119e+00
-8.32486078e-02 -3.49236935e-01 -3.52025181e-01 1.03012972e-01
-1.11274898e+00 -7.89006591e-01 -6.01055384e-01 8.37244749e-01
1.03712201e+00 -1.59038648e-01 2.06380397e-01 5.26704073e-01
-6.27894461e-01 5.24086595e-01 -9.71137136e-02 2.61025131e-01
6.87765062e-01 -6.64760292e-01 2.51897275e-01 9.66775596e-01
7.87303448e-02 3.47106904e-02 6.55109048e-01 -6.51203752e-01
-1.36296761e+00 -1.34974539e+00 6.25663757e-01 -3.48687112e-01
4.86665010e-01 -3.12726736e-01 -7.60915637e-01 6.27169967e-01
1.94388762e-01 1.46490961e-01 -1.54755060e-02 -7.03487754e-01
-1.48435801e-01 -2.03448623e-01 -1.07103634e+00 7.13966489e-01
1.20703578e+00 -1.44307807e-01 -3.23389292e-01 2.73311794e-01
1.08517444e+00 -6.64343238e-01 -7.01813221e-01 4.11109030e-01
5.32844961e-01 -1.19519830e+00 9.49535131e-01 -2.95682818e-01
4.94110018e-01 -5.33985257e-01 -2.14810148e-01 -1.08937490e+00
-1.36672139e-01 -8.86884987e-01 -1.69135600e-01 9.97211218e-01
-3.10933981e-02 -2.02328011e-01 8.29532266e-01 2.47909933e-01
-4.23650444e-02 -6.38224244e-01 -1.00819623e+00 -7.56911278e-01
-5.44470012e-01 -4.47074026e-01 4.60951835e-01 1.01897156e+00
-1.98462442e-01 -1.00861624e-01 -6.66651964e-01 2.45220959e-01
5.12627244e-01 -1.41808167e-01 1.03206408e+00 -7.57934809e-01
3.15843150e-02 -2.91312307e-01 -6.27793491e-01 -1.40006387e+00
-1.97974101e-01 -2.51192302e-01 -1.92579344e-01 -8.02025914e-01
-8.55846703e-02 -4.46071886e-02 -4.81957868e-02 2.49053463e-01
1.15729965e-01 4.53201503e-01 3.81715149e-01 2.98847914e-01
-6.85580134e-01 3.88081431e-01 1.50937605e+00 -1.55118912e-01
-4.71253723e-01 -6.89912662e-02 -2.91385930e-02 7.69115806e-01
5.45168102e-01 -4.02364820e-01 -5.49042344e-01 -3.90331417e-01
5.51805086e-02 4.92842436e-01 5.56211710e-01 -1.22483885e+00
1.95284232e-01 -1.66178301e-01 3.72786999e-01 -7.61991024e-01
3.15365165e-01 -1.01881909e+00 3.93622518e-01 2.56559640e-01
-1.51947483e-01 3.99977744e-01 5.77892661e-02 6.24138832e-01
-2.16244072e-01 4.07394394e-02 7.45035768e-01 -9.28047299e-02
-6.75303876e-01 4.48623300e-01 -6.53068841e-01 -5.59582002e-02
1.43612349e+00 -6.97064459e-01 -9.19730514e-02 -3.52119118e-01
-6.62061334e-01 -7.57691339e-02 6.62132800e-01 7.40785539e-01
5.61091959e-01 -1.31665778e+00 -4.55277473e-01 3.80824208e-01
1.38655398e-02 2.55620986e-01 5.18086910e-01 1.08240032e+00
-6.61428988e-01 3.88248533e-01 -2.70868063e-01 -9.13796782e-01
-1.33320117e+00 8.36619020e-01 2.71257371e-01 8.04131180e-02
-8.94228160e-01 5.76456308e-01 3.61289144e-01 2.76095510e-01
5.99421144e-01 -8.45011234e-01 -3.51794302e-01 4.03148144e-01
5.83543122e-01 2.89217532e-01 -1.53372541e-01 -6.99816465e-01
-2.39616707e-01 7.74386466e-01 1.63051024e-01 1.32856324e-01
9.61365700e-01 -3.80693525e-01 1.98337048e-01 3.26977938e-01
1.23994493e+00 -7.02562481e-02 -1.82635951e+00 -1.86226621e-01
-4.29362893e-01 -7.00652421e-01 -2.90041864e-01 -1.11593388e-01
-1.36026084e+00 4.23809201e-01 3.83755416e-01 4.45858568e-01
1.30642343e+00 -2.30251640e-01 6.84774220e-01 1.40434846e-01
4.72598314e-01 -8.92032802e-01 2.38694564e-01 3.31085026e-01
8.93910706e-01 -1.05988741e+00 -1.83270514e-01 -2.63462961e-01
-7.59513199e-01 1.08293319e+00 7.06769645e-01 -1.61604479e-01
5.00473440e-01 2.35149130e-01 -1.07037336e-01 1.05170175e-01
-6.63545847e-01 7.59645700e-02 -3.37930471e-02 5.87853134e-01
-8.95943195e-02 -2.47023061e-01 -3.43763195e-02 3.30669016e-01
2.64802799e-02 4.00883779e-02 6.18155539e-01 7.92546272e-01
-3.50515127e-01 -3.71942520e-01 -6.71726227e-01 -4.07602787e-02
-4.31168884e-01 1.74178898e-01 -8.59107226e-02 9.91575181e-01
4.05412167e-01 8.35573077e-01 3.96119982e-01 -6.40583634e-01
2.85191506e-01 -2.96117991e-01 2.63211191e-01 -9.34075564e-02
-3.25498432e-01 4.15750384e-01 -1.59137964e-01 -7.71261156e-01
-7.35997200e-01 -7.83353090e-01 -1.04376233e+00 -2.34193981e-01
-8.76354575e-02 -1.04713932e-01 3.17436039e-01 6.43108606e-01
3.97600889e-01 6.10603869e-01 8.58772516e-01 -1.32737124e+00
-4.51614968e-02 -8.81133139e-01 -4.73996311e-01 3.33887607e-01
6.39512897e-01 -8.68650079e-01 -5.05397081e-01 5.83261788e-01] | [9.17924976348877, -0.06889337301254272] |
2aa7c44b-4d7f-4310-ae34-6ebc24912fd9 | egnet-edge-guidance-network-for-salient | null | null | http://openaccess.thecvf.com/content_ICCV_2019/html/Zhao_EGNet_Edge_Guidance_Network_for_Salient_Object_Detection_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/Zhao_EGNet_Edge_Guidance_Network_for_Salient_Object_Detection_ICCV_2019_paper.pdf | EGNet: Edge Guidance Network for Salient Object Detection | Fully convolutional neural networks (FCNs) have shown their advantages in the salient object detection task. However, most existing FCNs-based methods still suffer from coarse object boundaries. In this paper, to solve this problem, we focus on the complementarity between salient edge information and salient object information. Accordingly, we present an edge guidance network (EGNet) for salient object detection with three steps to simultaneously model these two kinds of complementary information in a single network. In the first step, we extract the salient object features by a progressive fusion way. In the second step, we integrate the local edge information and global location information to obtain the salient edge features. Finally, to sufficiently leverage these complementary features, we couple the same salient edge features with salient object features at various resolutions. Benefiting from the rich edge information and location information in salient edge features, the fused features can help locate salient objects, especially their boundaries more accurately. Experimental results demonstrate that the proposed method performs favorably against the state-of-the-art methods on six widely used datasets without any pre-processing and post-processing. The source code is available at http: //mmcheng.net/egnet/.
| [' Ming-Ming Cheng', ' Jufeng Yang', ' Yang Cao', ' Deng-Ping Fan', ' Jiang-Jiang Liu', 'Jia-Xing Zhao'] | 2019-10-01 | null | null | null | iccv-2019-10 | ['co-saliency-detection', 'camouflaged-object-segmentation'] | ['computer-vision', 'computer-vision'] | [ 1.19896986e-01 -1.93124279e-01 -2.34059945e-01 -1.56893119e-01
-5.20383060e-01 -5.73986806e-02 3.67998421e-01 3.76203716e-01
-3.14855248e-01 5.43922961e-01 4.44801927e-01 2.47834131e-01
-1.98657691e-01 -6.68379605e-01 -5.59433818e-01 -8.16126406e-01
1.31090572e-02 -3.93057466e-01 8.16645622e-01 -2.28358656e-01
3.83592337e-01 4.46594685e-01 -1.71616936e+00 1.85602561e-01
1.06425774e+00 1.46142054e+00 6.35215819e-01 1.05867781e-01
-6.74063489e-02 7.56061375e-01 1.23393014e-01 2.73531526e-01
1.74610615e-01 -7.48459846e-02 -5.17803907e-01 7.11389910e-03
2.59893060e-01 -3.91322285e-01 -1.09839126e-01 1.43490994e+00
4.91629183e-01 1.61233038e-01 3.49972159e-01 -1.16909933e+00
-8.23219180e-01 4.48015749e-01 -9.72029686e-01 5.79240978e-01
1.39532778e-02 1.45443100e-02 1.13517034e+00 -1.39985156e+00
3.66404384e-01 1.09294784e+00 4.10376132e-01 1.72631480e-02
-7.53310144e-01 -5.48952460e-01 4.92352813e-01 4.89215404e-01
-1.44160330e+00 -5.08357048e-01 1.40573561e+00 -3.05899143e-01
4.04053181e-01 1.15160994e-01 6.93763196e-01 4.19234693e-01
1.65088717e-02 1.25631344e+00 8.19083929e-01 -2.81035304e-01
-5.54002188e-02 3.41725163e-02 1.89833611e-01 6.25679195e-01
4.03451562e-01 1.55768231e-01 -4.65299875e-01 1.16497837e-01
8.48986626e-01 5.59784949e-01 -4.33823586e-01 -5.10609150e-01
-1.21890426e+00 7.45887101e-01 1.11257064e+00 5.93685269e-01
-6.82291627e-01 -1.23335287e-01 2.82997340e-01 -2.67450929e-01
5.16855180e-01 2.54541021e-02 -1.58834949e-01 3.35277528e-01
-7.73186386e-01 1.89226240e-01 1.01164989e-01 9.06504393e-01
1.04914176e+00 1.26623712e-03 -5.29585958e-01 7.93104291e-01
5.16182005e-01 6.41788691e-02 3.78250510e-01 -3.76010031e-01
3.26256514e-01 9.36407804e-01 2.64793485e-01 -1.49018943e+00
-5.78271508e-01 -7.70244181e-01 -7.93256819e-01 7.24551752e-02
6.42537177e-02 -1.13134615e-01 -7.94318378e-01 1.57046473e+00
5.55326998e-01 3.98266673e-01 -6.81302398e-02 1.45126426e+00
1.30975270e+00 5.24703920e-01 2.77483553e-01 -4.59485501e-02
1.56169546e+00 -1.04489088e+00 -6.10276699e-01 -4.23154473e-01
2.48780653e-01 -9.14848983e-01 8.20750058e-01 -2.23281324e-01
-1.22514248e+00 -7.60172546e-01 -1.11035597e+00 -2.84600466e-01
-3.38504344e-01 5.12361526e-01 8.09360147e-01 -6.38237521e-02
-9.53779161e-01 3.00518870e-01 -6.15389466e-01 -1.73904121e-01
6.44655466e-01 2.77154386e-01 -9.69428122e-02 6.38702065e-02
-1.33820248e+00 6.24482453e-01 7.36759007e-01 4.33856010e-01
-6.38282716e-01 -6.12390399e-01 -1.05905318e+00 4.39444631e-01
5.88009536e-01 -4.85793620e-01 1.09351885e+00 -8.34212422e-01
-9.01883483e-01 5.48092782e-01 -2.34939575e-01 -2.08404288e-01
2.89448619e-01 -2.83830643e-01 -3.96638334e-01 3.33670169e-01
2.90725261e-01 8.46976280e-01 6.87008917e-01 -1.33126307e+00
-1.20434213e+00 -3.40974301e-01 8.84142667e-02 4.47980136e-01
-4.90449637e-01 1.68212801e-01 -4.92446393e-01 -7.68045306e-01
3.92470062e-01 -3.36493790e-01 -7.85908550e-02 2.18184054e-01
-4.20088798e-01 -3.30282629e-01 1.22193849e+00 -4.15919781e-01
1.18740594e+00 -2.24799466e+00 -1.12242304e-01 -1.41784474e-01
4.90355879e-01 2.88770765e-01 -3.05962134e-02 6.57327473e-02
-4.98640500e-02 -2.25147262e-01 -1.68163151e-01 -1.59786046e-01
-1.07700713e-01 -3.33481491e-01 -1.44465283e-01 4.03201848e-01
4.68110889e-01 1.11832869e+00 -1.08594537e+00 -7.96410084e-01
4.58997279e-01 5.47317028e-01 -3.51223707e-01 1.12202736e-02
9.61811319e-02 3.00272316e-01 -8.87290835e-01 8.99340749e-01
8.95581067e-01 -2.80788392e-01 -5.29143572e-01 -4.98887807e-01
-4.36666220e-01 4.46218215e-02 -1.40231407e+00 1.40232229e+00
-1.07458062e-01 5.38929164e-01 3.13279212e-01 -1.07832956e+00
1.08855247e+00 1.46095619e-01 4.69154716e-01 -6.63534343e-01
3.82507801e-01 3.68364513e-01 -3.23830433e-02 -4.47234184e-01
7.45794296e-01 -5.36147393e-02 1.38881147e-01 1.31065086e-01
-9.94960889e-02 4.35311347e-01 1.25076398e-01 8.74686688e-02
4.08604920e-01 -3.24975662e-02 4.54316884e-01 -5.22347033e-01
8.55564356e-01 -1.79646447e-01 9.46741283e-01 4.92509574e-01
-5.15545070e-01 7.27012038e-01 9.55947936e-02 -4.46716547e-01
-6.37167752e-01 -7.98624635e-01 -7.59535208e-02 1.13659799e+00
8.80117714e-01 -1.34575263e-01 -5.20232320e-01 -4.68170732e-01
1.12420708e-01 2.56782383e-01 -6.50575399e-01 -3.56579483e-01
-4.60093260e-01 -6.06135011e-01 -6.91098422e-02 8.14454257e-01
9.65674460e-01 -1.29848945e+00 -7.61323690e-01 2.59111822e-01
-1.59044161e-01 -8.49150598e-01 -7.88070619e-01 -7.27729723e-02
-7.26989090e-01 -8.80986810e-01 -1.04055166e+00 -1.22405100e+00
7.33964086e-01 1.14424598e+00 7.91250587e-01 4.47066307e-01
-2.91433722e-01 -3.06100491e-02 -4.68772888e-01 -4.09535527e-01
2.98689723e-01 7.13868067e-02 -1.29237711e-01 2.35767007e-01
3.75350207e-01 -5.62827408e-01 -1.02946544e+00 2.54751235e-01
-9.18552995e-01 4.20955896e-01 8.49099994e-01 7.83983767e-01
6.33352637e-01 1.49626225e-01 7.72231460e-01 -3.67985755e-01
4.80909765e-01 -6.10705137e-01 -5.82034290e-01 1.63019300e-01
-1.66894808e-01 -2.51978874e-01 5.99247694e-01 -3.23905468e-01
-1.29252958e+00 6.86316192e-02 -7.03051537e-02 -2.92579114e-01
-1.63170859e-01 6.61242604e-01 -4.28798318e-01 -1.38799608e-01
1.93591490e-01 4.06771839e-01 -4.06997353e-01 -5.98704815e-01
1.94976300e-01 6.42061412e-01 6.22564793e-01 -4.23533022e-01
6.99048460e-01 5.23382783e-01 -1.97080940e-01 -5.84510267e-01
-1.08488524e+00 -6.73702002e-01 -5.76701522e-01 -1.67861879e-01
6.30738318e-01 -1.04231441e+00 -5.45235872e-01 5.41200042e-01
-9.52924132e-01 1.52161971e-01 -2.14218557e-01 5.19859076e-01
-3.21997285e-01 3.79485220e-01 -4.91018832e-01 -7.24530280e-01
-5.71890950e-01 -1.00481451e+00 1.24852741e+00 1.09966481e+00
3.92594218e-01 -8.06347549e-01 -5.05341291e-01 -2.27598473e-02
5.47256827e-01 3.11652303e-01 4.76516128e-01 -4.95412588e-01
-8.05116415e-01 -1.40591726e-01 -8.76927435e-01 -1.35826483e-01
4.81156796e-01 -1.23113729e-01 -7.43391812e-01 -2.99187630e-01
6.20254800e-02 -5.75361736e-02 1.17061913e+00 6.32992446e-01
1.00494087e+00 -9.26479027e-02 -5.22384346e-01 5.75305998e-01
1.44669509e+00 1.08591333e-01 1.31378531e-01 4.66953516e-01
8.01369429e-01 4.81742978e-01 9.68234599e-01 5.33935368e-01
3.77833664e-01 4.16514248e-01 5.49273849e-01 -5.87913632e-01
-7.71299154e-02 -3.42785537e-01 -4.05375361e-02 5.38107574e-01
-8.97745043e-03 2.87605613e-01 -6.23149216e-01 9.53850091e-01
-2.12097883e+00 -8.43566537e-01 -1.27750700e-02 1.80246365e+00
6.28750741e-01 2.59497255e-01 1.48316652e-01 -2.55026929e-02
1.19788003e+00 3.83012772e-01 -6.80735111e-01 2.69800425e-01
-2.40706518e-01 -4.59521562e-01 2.79991239e-01 1.85057014e-01
-1.41440666e+00 8.68595719e-01 4.77673483e+00 8.82370591e-01
-1.15425587e+00 -4.93531935e-02 5.52032530e-01 1.52926464e-02
-3.08926672e-01 3.36188003e-02 -9.89948630e-01 5.45098424e-01
-4.99291085e-02 -4.00018722e-01 8.22639242e-02 1.06308019e+00
3.52678299e-01 -2.30353355e-01 -6.09451890e-01 9.01842892e-01
-6.97935596e-02 -1.39961696e+00 -5.20397015e-02 -2.88973540e-01
8.11208546e-01 1.54121974e-02 9.12538767e-02 1.58339933e-01
8.95177945e-02 -5.80134630e-01 9.67951655e-01 4.52789426e-01
2.79449970e-01 -9.02954340e-01 8.52526188e-01 4.05205816e-01
-1.92410934e+00 -2.73377925e-01 -6.01043344e-01 -2.15915725e-01
2.24157766e-01 8.23377728e-01 -2.32574061e-01 7.91237593e-01
8.72157693e-01 1.14538479e+00 -5.84183335e-01 1.58457351e+00
-2.24186987e-01 1.27198011e-01 -2.91659623e-01 1.24182375e-02
4.94374245e-01 -5.18470742e-02 7.21169472e-01 1.08785975e+00
3.24554682e-01 2.39327192e-01 3.84835958e-01 1.12873125e+00
-2.90374719e-02 2.53738612e-01 -2.70900667e-01 2.95950413e-01
5.91010392e-01 1.70723343e+00 -9.89367962e-01 -4.22839820e-01
-6.06088996e-01 6.13839805e-01 4.29187506e-01 2.40962297e-01
-7.75222719e-01 -6.43583417e-01 4.62233663e-01 -8.93495604e-02
5.36590338e-01 4.12741080e-02 -2.04907417e-01 -1.15573764e+00
2.23792583e-01 -3.94641101e-01 4.69548911e-01 -8.71194780e-01
-1.12972033e+00 4.21640664e-01 -4.14250093e-03 -1.47688591e+00
2.64091939e-01 -3.77825469e-01 -9.95660782e-01 9.55114841e-01
-2.14768696e+00 -1.32266092e+00 -5.60001433e-01 5.15657485e-01
5.02725780e-01 1.82828560e-01 7.18223453e-02 2.00125977e-01
-6.50841951e-01 3.00096363e-01 3.94883640e-02 2.41900593e-01
4.52301532e-01 -1.04587018e+00 2.42032677e-01 1.07751024e+00
-1.96304321e-01 6.80195510e-01 4.61168289e-01 -6.18301451e-01
-1.00499570e+00 -1.19558346e+00 5.93108237e-01 3.15832883e-01
5.79167426e-01 -1.99865103e-02 -1.14505839e+00 4.84270275e-01
1.59763526e-02 4.68489945e-01 1.51501775e-01 -1.42338485e-01
1.63126111e-01 -1.11232236e-01 -9.42865551e-01 6.78985476e-01
9.86901522e-01 -2.72453785e-01 -7.51428306e-01 -3.35740447e-02
8.52595866e-01 -4.29311037e-01 -6.71271622e-01 7.82583416e-01
3.17827016e-01 -9.72814679e-01 1.00012636e+00 -8.46223310e-02
4.26382720e-01 -7.39073455e-01 5.13372663e-03 -1.09175897e+00
-7.35689759e-01 -3.21145117e-01 -2.87559908e-02 1.35344470e+00
5.34699187e-02 -7.59519815e-01 6.32947743e-01 3.29436332e-01
-4.11464721e-01 -8.79875302e-01 -6.74667478e-01 -4.34166908e-01
-4.27752048e-01 -6.61846204e-03 6.85769975e-01 7.15488493e-01
1.21037304e-01 2.58157790e-01 -2.89303869e-01 2.48134702e-01
5.82477868e-01 8.04153621e-01 3.41718674e-01 -1.49821198e+00
1.45069584e-01 -6.09020710e-01 -3.54761779e-01 -1.23813438e+00
-1.26288757e-01 -7.61613607e-01 2.25147188e-01 -1.68161178e+00
5.87969244e-01 -4.22651738e-01 -8.55132580e-01 5.24179518e-01
-8.84194016e-01 2.42412895e-01 3.23654413e-01 2.09284514e-01
-8.22268128e-01 8.90248179e-01 1.47129726e+00 -6.29486442e-02
-3.69076341e-01 -2.90939927e-01 -1.25862849e+00 9.67245460e-01
9.32140410e-01 -1.91793486e-01 -3.04501802e-01 -3.02629828e-01
-3.13185006e-01 -2.32640252e-01 6.43122792e-01 -1.17471743e+00
4.79658186e-01 -1.40953377e-01 6.21725500e-01 -9.05386984e-01
9.70754400e-02 -7.77567625e-01 -3.55066746e-01 3.76105845e-01
-8.64687040e-02 -2.96415180e-01 4.49714065e-01 5.14359951e-01
-5.08282065e-01 -2.12591723e-01 7.90476322e-01 -1.30822929e-02
-1.30480683e+00 4.57454503e-01 -5.93805723e-02 -6.14404120e-02
1.12317801e+00 -2.81404853e-01 -4.37897265e-01 -1.07080713e-01
-4.95054007e-01 5.79935789e-01 4.53369439e-01 5.95738709e-01
8.47979128e-01 -1.48329914e+00 -6.48207664e-01 3.23697597e-01
2.90800273e-01 1.76467821e-01 5.58972239e-01 1.11496091e+00
-8.27361643e-02 3.86359543e-01 -4.64725047e-01 -6.06846094e-01
-1.25888836e+00 8.17966938e-01 2.88631588e-01 1.54978707e-02
-6.70165598e-01 7.59958446e-01 7.16757953e-01 1.50661021e-01
9.73172486e-02 -4.32554513e-01 -7.05290139e-01 8.70005488e-02
7.18960702e-01 9.26628560e-02 -2.06608534e-01 -8.99908423e-01
-3.98470253e-01 8.29610109e-01 -3.16566288e-01 4.38208938e-01
1.35207379e+00 -3.81549954e-01 2.08643358e-03 1.02907442e-01
1.07463002e+00 -2.31782272e-02 -1.57808399e+00 -6.42162204e-01
-3.23571498e-03 -6.15065098e-01 3.94726962e-01 -2.35049605e-01
-1.45238090e+00 8.80088687e-01 5.26514113e-01 2.30665535e-01
1.42623258e+00 6.84316233e-02 8.98566127e-01 -1.39828131e-01
1.73901469e-01 -9.43426490e-01 -1.20469987e-01 2.06598043e-01
8.28609884e-01 -1.53002143e+00 7.27038011e-02 -7.56272078e-01
-5.28970718e-01 8.37214708e-01 7.27625072e-01 -2.61025548e-01
7.15030730e-01 2.89002284e-02 -1.56678319e-01 -2.71149516e-01
-4.46283162e-01 -6.66880667e-01 5.97740591e-01 3.95124376e-01
2.75577605e-01 -3.54519449e-02 -4.06595498e-01 9.62924302e-01
2.85601646e-01 1.77806020e-01 1.88385665e-01 1.08354485e+00
-1.02062416e+00 -5.13378203e-01 -4.02133018e-01 4.10946071e-01
-4.46257859e-01 -3.02220523e-01 -1.22191273e-01 6.74819767e-01
2.36349657e-01 9.30017471e-01 6.21229857e-02 -3.19269389e-01
2.51479536e-01 -4.51118171e-01 -8.33153054e-02 -4.52891678e-01
-3.00168306e-01 2.87800580e-01 -2.94432700e-01 -3.23135316e-01
-5.15135288e-01 -5.92811704e-01 -1.42488194e+00 5.64168803e-02
-6.08682275e-01 1.85230225e-01 4.51327786e-02 7.59926200e-01
4.93999511e-01 6.50888503e-01 5.88537276e-01 -1.37124193e+00
-1.44512951e-01 -9.07853603e-01 -5.99711537e-01 3.47885132e-01
5.29429078e-01 -1.06297243e+00 -2.57771820e-01 -1.32551670e-01] | [9.725911140441895, -0.4665317237377167] |
012f73a8-4eac-48ed-b1b5-59491924ab5d | investigating-the-successes-and-failures-of | 1905.01758 | null | https://arxiv.org/abs/1905.01758v1 | https://arxiv.org/pdf/1905.01758v1.pdf | Investigating the Successes and Failures of BERT for Passage Re-Ranking | The bidirectional encoder representations from transformers (BERT) model has recently advanced the state-of-the-art in passage re-ranking. In this paper, we analyze the results produced by a fine-tuned BERT model to better understand the reasons behind such substantial improvements. To this aim, we focus on the MS MARCO passage re-ranking dataset and provide potential reasons for the successes and failures of BERT for retrieval. In more detail, we empirically study a set of hypotheses and provide additional analysis to explain the successful performance of BERT. | ['W. Bruce Croft', 'Hamed Zamani', 'Harshith Padigela'] | 2019-05-05 | null | null | null | null | ['passage-re-ranking'] | ['natural-language-processing'] | [-3.30401987e-01 -2.34075755e-01 -1.78745583e-01 -3.45392913e-01
-1.33860862e+00 -6.68025851e-01 8.49491417e-01 3.97765309e-01
-5.79869568e-01 8.34922016e-01 9.59764957e-01 -3.74725968e-01
-5.67478180e-01 -4.77005929e-01 -6.40547872e-01 -1.59430683e-01
-2.77169317e-01 5.51840305e-01 1.92601681e-01 -6.43005073e-01
5.81006765e-01 3.15177351e-01 -1.37725937e+00 8.99292886e-01
5.98446429e-01 3.40205967e-01 8.94102231e-02 9.83798921e-01
2.53352523e-01 1.06705165e+00 -8.75073433e-01 -5.07759929e-01
-7.86767304e-02 -5.60545981e-01 -1.14533699e+00 -6.49884641e-01
3.49520594e-01 -8.42021823e-01 -8.47349048e-01 5.28225899e-01
7.90846944e-01 5.77545702e-01 7.79186964e-01 -6.92467511e-01
-7.99299181e-01 8.88207376e-01 -1.56089514e-02 9.78441298e-01
5.48808336e-01 -5.44646084e-01 1.57553887e+00 -7.58503616e-01
7.62222528e-01 1.00506496e+00 3.94945711e-01 5.86256564e-01
-8.14818203e-01 -3.19725484e-01 5.46640193e-04 7.45936453e-01
-1.43071163e+00 -8.11636150e-01 4.20033127e-01 -1.65482298e-01
1.33884418e+00 4.80191320e-01 7.47039795e-01 8.98204148e-01
7.84060657e-02 1.14295471e+00 8.88346493e-01 -6.26534641e-01
-2.27352530e-01 9.83811393e-02 4.73601252e-01 3.49856526e-01
1.44874796e-01 1.25930056e-01 -7.70202100e-01 -9.75040644e-02
3.88231337e-01 -3.23588669e-01 -4.85572934e-01 -2.98345205e-03
-8.61145139e-01 8.14424276e-01 7.12223589e-01 5.24100482e-01
1.87565684e-02 2.33468145e-01 4.00591463e-01 6.58360898e-01
3.67525786e-01 6.97525144e-01 -4.06948090e-01 -6.08807325e-01
-1.24507189e+00 3.53229493e-01 4.46297050e-01 6.34484053e-01
1.41362399e-01 -3.52591485e-01 -3.98473442e-01 1.04335153e+00
3.77230525e-01 -2.54888356e-01 7.58820474e-01 -5.76444805e-01
5.12340307e-01 1.21490441e-01 2.48746932e-01 -6.83857501e-01
-3.66578907e-01 -7.10193515e-01 -2.04503775e-01 -4.69827235e-01
1.48749575e-01 4.00609225e-01 -6.77952051e-01 1.52597356e+00
-4.79474038e-01 -8.48278701e-02 7.20610619e-02 8.05335522e-01
9.41228747e-01 6.34285092e-01 -1.14730082e-01 8.59855264e-02
7.86528707e-01 -1.15338206e+00 -5.01224458e-01 3.58200558e-02
1.01250494e+00 -7.77797341e-01 1.04458916e+00 2.20355415e-03
-1.09110832e+00 -5.97184420e-01 -1.05323696e+00 -4.16269153e-01
-9.85493436e-02 4.16448474e-01 5.57031572e-01 3.64199162e-01
-1.35615408e+00 7.94976056e-01 -7.64432311e-01 -5.98949730e-01
1.85256656e-02 2.27620006e-01 -1.83105439e-01 -1.32699236e-01
-1.54530644e+00 1.21792328e+00 1.58669114e-01 1.14359759e-01
-9.39240694e-01 -5.08074343e-01 -3.49180311e-01 4.95283633e-01
-3.76886517e-01 -7.31643140e-01 1.39851642e+00 -3.74642402e-01
-1.36475122e+00 7.96353936e-01 -3.06235850e-01 -6.12253606e-01
3.51596206e-01 -5.10134518e-01 -5.12856483e-01 2.37574548e-01
-2.29843892e-02 5.38633585e-01 -1.27245309e-02 -1.02888107e+00
-6.14451230e-01 8.74392875e-03 4.07364309e-01 4.00801778e-01
-3.63150686e-01 2.39667177e-01 -4.53747630e-01 -5.92336655e-01
-2.06930876e-01 -8.70501757e-01 3.13117296e-01 -7.19715953e-01
-4.85832766e-02 -4.44696605e-01 -8.45799819e-02 -7.16608584e-01
1.67216563e+00 -2.34298849e+00 1.35826439e-01 2.02920794e-01
1.64096653e-01 8.84741452e-03 -3.81548852e-01 8.63823295e-01
-5.31855859e-02 4.10680056e-01 4.22965646e-01 -4.36624467e-01
-1.50330365e-01 5.03883176e-02 -3.72351438e-01 3.81398618e-01
-1.51583850e-01 1.02806866e+00 -9.16581750e-01 -3.74067336e-01
-7.11212829e-02 4.76483196e-01 -7.45958269e-01 1.61635563e-01
2.89962888e-01 1.87538266e-01 -3.50522161e-01 4.73632157e-01
-4.60615382e-02 -2.53888011e-01 4.88697290e-02 -2.13798448e-01
-2.24151403e-01 1.18509305e+00 -3.27289969e-01 1.36039412e+00
-2.98186779e-01 1.29712284e+00 -4.58316833e-01 -6.41735971e-01
6.22716308e-01 3.01392943e-01 3.39249432e-01 -9.69890058e-01
1.53322387e-02 4.95620161e-01 1.76794991e-01 -1.60107464e-01
1.02096510e+00 -1.13866769e-01 2.04030499e-01 3.07694465e-01
9.13106054e-02 2.81742036e-01 5.72923005e-01 6.19578958e-01
1.14660978e+00 -7.19502792e-02 2.33692840e-01 -2.02021569e-01
3.98131996e-01 1.13438651e-01 1.82847977e-01 1.15208268e+00
-1.62141398e-01 7.50738144e-01 9.65481997e-02 -2.42840976e-01
-1.01637018e+00 -9.27223980e-01 -1.66311324e-01 9.90088522e-01
7.42185116e-02 -9.14076507e-01 -2.68914700e-01 -5.93037963e-01
-1.50008842e-01 8.78665745e-01 -7.07104862e-01 -4.13945287e-01
-6.52320087e-01 -6.68710709e-01 8.70698154e-01 7.11205423e-01
2.36881956e-01 -8.42460573e-01 -3.44411582e-01 1.79428339e-01
-7.06419528e-01 -8.11857224e-01 -1.22363791e-01 -4.39550504e-02
-1.01205945e+00 -8.46421421e-01 -7.78411925e-01 -6.88513577e-01
2.35536903e-01 5.58189511e-01 1.38906229e+00 5.82585752e-01
1.66797176e-01 5.81955612e-01 -9.77982640e-01 -1.15171924e-01
-2.80567676e-01 5.45673907e-01 2.08663521e-03 -5.88983119e-01
2.36787215e-01 -6.31630123e-02 -5.39611816e-01 2.01944873e-01
-8.79020333e-01 -1.95178151e-01 4.44127530e-01 9.36603904e-01
2.81212687e-01 -1.09943129e-01 7.13442624e-01 -5.44415176e-01
1.28773654e+00 -5.17881989e-01 -1.31812379e-01 4.87969309e-01
-8.30169201e-01 1.96225499e-03 9.51261148e-02 2.49918904e-02
-9.35880542e-01 -8.73611569e-01 -3.95972341e-01 7.25787655e-02
4.54444349e-01 1.08070505e+00 5.62776685e-01 -4.05885018e-02
5.79845726e-01 8.60072300e-02 -4.87329453e-01 -8.10727835e-01
3.15101177e-01 9.33649302e-01 1.07268162e-01 -5.02395093e-01
2.62629837e-01 2.44864628e-01 -3.83564383e-01 -6.08339250e-01
-9.27859545e-01 -6.40144706e-01 -4.38096285e-01 -3.45146507e-01
3.67723733e-01 -1.11183488e+00 -7.86064342e-02 -5.29895239e-02
-9.31308270e-01 -4.20545548e-01 -2.36063406e-01 7.61926889e-01
-5.25887430e-01 2.18587637e-01 -9.47222710e-01 -4.40028280e-01
-2.71237731e-01 -1.01753449e+00 6.70039594e-01 2.43811369e-01
-5.12312770e-01 -8.46166193e-01 5.42865336e-01 4.84924167e-01
5.56759596e-01 -5.62181294e-01 9.79980230e-01 -7.43162811e-01
-4.54805136e-01 -2.39942670e-01 -1.82588249e-01 8.15795437e-02
-2.96951264e-01 9.05043259e-02 -6.07456386e-01 -6.35136306e-01
-4.56805587e-01 -2.57965416e-01 1.17912829e+00 2.39010558e-01
8.40409040e-01 1.53316528e-01 -3.35160911e-01 4.34947759e-01
1.14361787e+00 8.02130029e-02 7.82930195e-01 1.07929718e+00
1.17528245e-01 3.12915921e-01 5.90602517e-01 3.65035564e-01
5.88689148e-01 8.50616038e-01 9.22995526e-03 2.43248001e-01
-3.45323384e-01 -4.81561631e-01 3.56204748e-01 1.20264292e+00
-1.26974970e-01 -6.72180891e-01 -7.36128211e-01 8.91721725e-01
-1.64939141e+00 -9.07584071e-01 -7.48398132e-04 2.12582445e+00
4.95714664e-01 7.10466951e-02 5.20916395e-02 1.47231147e-01
3.31628323e-01 3.16279143e-01 1.70745179e-01 -4.27468956e-01
-2.50366271e-01 1.38377115e-01 3.53679508e-01 6.72638357e-01
-6.34024382e-01 9.10693645e-01 7.78001928e+00 6.17139637e-01
-6.66418910e-01 -7.54020503e-03 3.75656277e-01 -5.61525822e-01
-5.38991451e-01 -2.29712669e-02 -7.14929342e-01 1.51308998e-01
1.17190313e+00 -1.80611566e-01 5.55854201e-01 3.25158149e-01
-2.60233451e-02 1.00224599e-01 -1.26438570e+00 7.93577969e-01
3.30077052e-01 -1.45696187e+00 2.56478578e-01 -1.58893153e-01
6.47112846e-01 4.75132048e-01 -9.57127512e-02 7.16378629e-01
1.45365849e-01 -7.56539643e-01 7.45587707e-01 5.28459132e-01
4.96381700e-01 -4.63250995e-01 8.14454377e-01 3.78093533e-02
-8.16260338e-01 -2.48259202e-01 -3.47609788e-01 -1.42096579e-01
3.40858728e-01 1.64989039e-01 -7.70709336e-01 6.68139398e-01
8.48648906e-01 8.77178073e-01 -8.93215656e-01 1.47643030e+00
-3.82474571e-01 1.00403798e+00 -1.77910119e-01 -1.68837339e-01
2.95136333e-01 2.64958650e-01 6.28709853e-01 1.41715014e+00
3.44110310e-01 7.20897019e-02 -3.64453971e-01 2.82388717e-01
-1.45384505e-01 1.35079697e-01 -6.55166209e-02 -2.88356781e-01
4.19040412e-01 6.33633375e-01 -3.41923386e-01 -4.95239109e-01
-3.90722960e-01 9.69231486e-01 8.25504959e-01 5.38847566e-01
-5.75619280e-01 -2.16146395e-01 3.82170796e-01 8.32594261e-02
1.32956147e-01 -4.02069867e-01 -1.26100834e-02 -1.40530014e+00
-1.59833163e-01 -7.69603193e-01 6.73275113e-01 -9.25354719e-01
-1.08365417e+00 5.57592392e-01 1.89195052e-01 -1.24618900e+00
-1.80996418e-01 -2.58195132e-01 -3.30879003e-01 7.28862166e-01
-1.75685418e+00 -6.82340980e-01 5.41163795e-02 3.20787698e-01
5.79809427e-01 -2.20620841e-01 8.68550181e-01 6.91425443e-01
-3.68018270e-01 7.68741906e-01 7.70078301e-01 1.22387141e-01
9.61283743e-01 -1.01536489e+00 3.79238784e-01 6.47875011e-01
5.71967661e-01 1.08301294e+00 6.10469997e-01 -5.59258282e-01
-9.21779096e-01 -4.89136577e-01 1.41404736e+00 -3.96453023e-01
4.67787653e-01 2.31125697e-01 -6.37104273e-01 8.79802048e-01
1.70589104e-01 -5.01642823e-01 9.09166932e-01 6.32323027e-01
-3.77517700e-01 -9.80934203e-02 -7.62614489e-01 6.68218493e-01
9.14211869e-01 -7.19429672e-01 -9.76966381e-01 1.25464693e-01
2.48445913e-01 -3.62966359e-01 -6.81141734e-01 4.74269718e-01
8.99559557e-01 -9.71899450e-01 1.23076844e+00 -9.35931206e-01
6.22284174e-01 -7.11814612e-02 -3.88062328e-01 -1.53723729e+00
-7.90606141e-01 -2.50610709e-01 -1.75672337e-01 9.76063311e-01
4.57540035e-01 -3.77361059e-01 5.78452766e-01 4.14770544e-01
-1.98548570e-01 -8.21262062e-01 -1.07224262e+00 -6.43190444e-01
3.38642001e-01 -1.74714819e-01 4.64861333e-01 6.86641395e-01
3.53252590e-01 2.91079968e-01 -4.24440861e-01 -2.11087853e-01
1.10533476e-01 4.50796783e-02 4.38234299e-01 -9.14517999e-01
-3.97441059e-01 -5.70861757e-01 -4.62216020e-01 -1.58809221e+00
6.31378125e-03 -9.79103029e-01 -1.56412646e-01 -1.90275586e+00
3.66164774e-01 -4.05095965e-01 -8.67021263e-01 9.85098705e-02
-4.53702658e-01 4.22834635e-01 4.68619734e-01 6.74620211e-01
-9.15660739e-01 5.87112308e-01 7.85443664e-01 -1.06969573e-01
-1.59284785e-01 -8.38259980e-02 -8.97582531e-01 7.43075646e-03
1.26106668e+00 -5.36841810e-01 -3.82524431e-01 -9.84854519e-01
6.49383068e-01 1.00083668e-02 8.87512118e-02 -7.96071470e-01
5.09801395e-02 3.91176313e-01 3.74211580e-01 -7.29352593e-01
3.03704351e-01 -3.18767101e-01 -1.05834238e-01 2.11401954e-01
-9.72847044e-01 4.48491871e-01 1.81374788e-01 4.79250044e-01
-5.00813603e-01 -5.50030291e-01 3.96620572e-01 -8.27550888e-04
-7.06653655e-01 -2.95216262e-01 -6.71143532e-01 -3.78651544e-02
3.61467987e-01 1.59941226e-01 -4.21467811e-01 -8.41246307e-01
-6.57487333e-01 2.19784617e-01 2.77220428e-01 7.78204679e-01
5.81153929e-01 -1.33283401e+00 -9.13187027e-01 -1.46868631e-01
3.78445804e-01 -7.84110546e-01 2.10769966e-01 7.14561939e-01
-5.53498447e-01 8.99799049e-01 -1.29007667e-01 3.23415026e-02
-1.45996010e+00 1.24906018e-01 4.63213921e-01 -6.24916613e-01
-5.68302870e-01 9.48354483e-01 -2.83408582e-01 -6.17384054e-02
-1.53857851e-02 -2.29947586e-02 -6.36821091e-01 4.23118174e-02
5.71165264e-01 8.61425638e-01 4.22046900e-01 -5.54431975e-01
-5.10363579e-01 2.50031769e-01 -5.34072638e-01 -5.77390134e-01
1.29656804e+00 -5.15682280e-01 -1.48652270e-02 4.47372854e-01
1.30746853e+00 1.88126594e-01 -4.84128058e-01 -1.20393656e-01
-6.52732030e-02 -6.66491747e-01 2.34419897e-01 -1.14847040e+00
-5.95719457e-01 8.03194880e-01 4.31834787e-01 1.47619456e-01
7.06089616e-01 5.55503555e-02 7.20196307e-01 4.46940839e-01
4.00267810e-01 -9.07639503e-01 -7.19681829e-02 7.66091764e-01
9.79181468e-01 -8.75131428e-01 1.88971102e-01 7.77624473e-02
-5.17921746e-01 1.04493725e+00 2.73384094e-01 -2.20530421e-01
4.81698900e-01 -4.67238098e-01 8.74871016e-02 -2.80455023e-01
-8.13697755e-01 -1.66189104e-01 5.99752963e-01 1.61087871e-01
1.02225554e+00 8.07911232e-02 -7.73479998e-01 4.31379765e-01
-3.96301687e-01 1.09732799e-01 5.57145774e-01 8.32207620e-01
-3.33321750e-01 -1.31452560e+00 -5.49011007e-02 6.29829764e-01
-7.59884715e-01 -4.88137960e-01 -6.08047366e-01 5.76180279e-01
-5.39411366e-01 1.23016441e+00 -8.03516656e-02 -5.53858817e-01
4.50185597e-01 3.17762569e-02 6.69662058e-01 -4.53743517e-01
-7.55503118e-01 -1.82072762e-02 5.08460701e-01 -3.05350423e-01
-3.00915927e-01 -7.73207068e-01 -9.18854058e-01 -2.23951846e-01
-5.93663216e-01 6.06808364e-01 3.92948955e-01 8.64046872e-01
4.59501565e-01 6.79818153e-01 4.26486701e-01 -4.79810953e-01
-6.37501895e-01 -1.31357121e+00 -3.84349287e-01 4.75262851e-01
3.32449943e-01 -6.34481788e-01 -5.01649618e-01 -2.89517075e-01] | [11.525142669677734, 7.656160354614258] |
dcc80d70-fd4b-4267-86a1-70b87d24cdbc | embarrassingly-simple-unsupervised-aspect | 2004.13580 | null | https://arxiv.org/abs/2004.13580v1 | https://arxiv.org/pdf/2004.13580v1.pdf | Embarrassingly Simple Unsupervised Aspect Extraction | We present a simple but effective method for aspect identification in sentiment analysis. Our unsupervised method only requires word embeddings and a POS tagger, and is therefore straightforward to apply to new domains and languages. We introduce Contrastive Attention (CAt), a novel single-head attention mechanism based on an RBF kernel, which gives a considerable boost in performance and makes the model interpretable. Previous work relied on syntactic features and complex neural models. We show that given the simplicity of current benchmark datasets for aspect extraction, such complex models are not needed. The code to reproduce the experiments reported in this paper is available at https://github.com/clips/cat | ['Stéphan Tulkens', 'Andreas van Cranenburgh'] | 2020-04-28 | embarrassingly-simple-unsupervised-aspect-1 | https://aclanthology.org/2020.acl-main.290 | https://aclanthology.org/2020.acl-main.290.pdf | acl-2020-6 | ['aspect-category-detection'] | ['natural-language-processing'] | [-1.00140549e-01 1.66426510e-01 -2.22434580e-01 -5.77361107e-01
-5.99818230e-01 -7.05662370e-01 8.24849248e-01 3.84174615e-01
-5.78180134e-01 4.23909038e-01 2.56932914e-01 -7.18874216e-01
2.63958901e-01 -6.48085296e-01 -3.17447096e-01 -4.84992355e-01
2.04569608e-01 3.80395323e-01 2.27847621e-02 -3.15978914e-01
2.38481820e-01 3.26279134e-01 -1.29957938e+00 8.35913941e-02
4.88701820e-01 8.67580175e-01 -7.09531903e-02 6.90133750e-01
-3.12021822e-01 6.36858821e-01 -6.65662766e-01 -8.05508554e-01
9.41037461e-02 -2.77752012e-01 -8.87441218e-01 -1.32475913e-01
4.90789190e-02 -1.22822881e-01 -1.18057385e-01 8.15810442e-01
3.72888297e-01 1.41776696e-01 7.26702213e-01 -1.06769967e+00
-1.17810440e+00 6.71660662e-01 -3.58646542e-01 2.69439220e-01
3.21439028e-01 -5.13596013e-02 1.45184398e+00 -1.12654257e+00
4.15411890e-01 8.83356929e-01 7.18665898e-01 4.71181989e-01
-9.09733772e-01 -2.63771117e-01 3.10887605e-01 8.87134345e-04
-1.10431528e+00 -3.97126138e-01 8.65433991e-01 -2.09812850e-01
1.42810810e+00 1.89575747e-01 8.28173816e-01 9.97895956e-01
1.78710088e-01 8.51723671e-01 9.74755049e-01 -6.05503559e-01
1.96919173e-01 3.92781228e-01 4.78434145e-01 6.12995386e-01
5.22633314e-01 -2.10232601e-01 -2.71715045e-01 -2.75023788e-01
5.08820832e-01 8.39792751e-03 -1.08200870e-01 -3.10572207e-01
-8.57669473e-01 1.17854071e+00 3.03390563e-01 3.86717528e-01
-3.55357200e-01 7.66113997e-02 3.15170676e-01 2.15607002e-01
7.45624781e-01 7.09270358e-01 -9.18441951e-01 -4.31117445e-01
-6.15609169e-01 -9.26585402e-03 1.09944892e+00 9.71827984e-01
5.57895899e-01 1.63950369e-01 1.27401441e-01 7.50421345e-01
2.19925463e-01 5.13407290e-01 5.73999405e-01 -7.71586776e-01
7.05255046e-02 6.88890159e-01 6.99169189e-03 -8.09872568e-01
-4.32345957e-01 -3.74928445e-01 -2.65158802e-01 1.14800736e-01
1.11677848e-01 -3.00971478e-01 -9.51844156e-01 1.37220287e+00
1.37137741e-01 -3.55083227e-01 -6.07132390e-02 7.01603115e-01
9.34476256e-01 4.65958774e-01 4.88740318e-02 1.82304859e-01
1.79083478e+00 -1.14819109e+00 -7.96159863e-01 -4.49829847e-01
6.50800109e-01 -8.24792802e-01 1.19472444e+00 1.80331782e-01
-1.09408474e+00 -2.83987131e-02 -1.04449606e+00 -2.84829766e-01
-1.06728947e+00 -2.97280103e-02 1.16798842e+00 8.28001797e-01
-1.12710834e+00 2.92220771e-01 -9.59513366e-01 -5.27757704e-01
5.23638129e-01 4.74136978e-01 -4.21181887e-01 9.64256674e-02
-9.88540292e-01 9.75274205e-01 6.18853308e-02 9.18122679e-02
-1.48712769e-01 -2.40546867e-01 -1.21199405e+00 4.69757952e-02
2.49394730e-01 -5.70795596e-01 1.69859421e+00 -1.33873856e+00
-1.61558449e+00 7.50927329e-01 -4.25592721e-01 -2.80728042e-01
-2.67284185e-01 -4.00530487e-01 -2.41367370e-01 2.05091909e-01
-4.59551327e-02 5.44301033e-01 7.78496623e-01 -8.80346656e-01
-2.06158057e-01 -1.47697851e-01 4.29818153e-01 1.09493054e-01
-5.02495289e-01 4.58051682e-01 -6.12037957e-01 -7.25224018e-01
-4.03850794e-01 -9.88767385e-01 -3.89320344e-01 -3.17016661e-01
-1.94096476e-01 -2.79513776e-01 4.98313278e-01 -6.32024705e-01
1.21034074e+00 -2.10452700e+00 -5.65295666e-02 7.23340176e-03
1.38671264e-01 3.98833483e-01 -2.12045804e-01 4.85573322e-01
-1.01056077e-01 3.88635874e-01 -6.01847887e-01 -5.28086364e-01
1.08081125e-01 1.85515746e-01 -1.77118272e-01 2.22274989e-01
5.58145642e-01 1.22233140e+00 -7.52470434e-01 -1.17898442e-01
4.25427817e-02 7.01431274e-01 -6.04447365e-01 2.47857526e-01
-1.31293997e-01 -1.05266608e-01 -4.48185623e-01 8.17065060e-01
4.83434677e-01 -2.09596932e-01 7.53587484e-02 -8.43708590e-02
-1.06435046e-01 7.95991421e-01 -8.48668158e-01 1.33299756e+00
-7.48418152e-01 8.27030182e-01 -1.38863280e-01 -8.57594550e-01
7.09075809e-01 2.59003520e-01 1.22166835e-01 -5.03188491e-01
4.52514887e-01 2.29368191e-02 -8.52997825e-02 -3.25703472e-01
7.08136559e-01 -1.25063181e-01 -3.56161833e-01 7.68194914e-01
4.22127515e-01 -3.79465073e-01 3.75289738e-01 2.03919441e-01
1.03590858e+00 1.51169047e-01 7.67792284e-01 -4.03828233e-01
2.85629839e-01 -1.47102186e-02 6.19792998e-01 5.93215585e-01
-1.39166772e-01 7.21221864e-01 6.50559604e-01 -6.04116261e-01
-9.97847259e-01 -8.34685683e-01 -1.54920682e-01 1.10831451e+00
-3.04802120e-01 -7.58270800e-01 -6.24336302e-01 -9.35082018e-01
-2.67346263e-01 7.34558761e-01 -7.34697580e-01 8.85303169e-02
-4.60382223e-01 -7.51337826e-01 3.55410963e-01 8.58159363e-01
7.16544464e-02 -1.42293370e+00 -4.25999701e-01 1.82662029e-02
2.05733851e-01 -1.05069721e+00 -3.96117836e-01 5.48593819e-01
-8.31821561e-01 -1.01384032e+00 -4.71149504e-01 -9.60233450e-01
8.88193607e-01 1.80148199e-01 1.32985914e+00 1.01153880e-01
-1.85743943e-01 6.02146566e-01 -6.17966831e-01 -8.24618340e-01
4.36110385e-02 3.46209556e-01 -1.61619693e-01 -3.17369431e-01
1.10974574e+00 -3.36684376e-01 -4.81536418e-01 -1.99898615e-01
-1.02565217e+00 -3.31322670e-01 6.85356736e-01 8.98675621e-01
4.51300681e-01 -2.90802419e-01 2.95241177e-01 -1.29450977e+00
7.79277444e-01 -5.10033965e-01 -6.10761523e-01 -2.57458448e-01
-8.32716584e-01 5.73657155e-02 5.78571737e-01 -2.16699958e-01
-8.53518784e-01 -5.87565824e-04 -4.63268548e-01 8.86276215e-02
-4.50288177e-01 6.83600545e-01 -2.82263439e-02 1.03158861e-01
4.19224590e-01 -4.82053310e-03 -7.14674369e-02 -5.56218028e-01
5.21196365e-01 9.05293226e-01 5.53856231e-02 -1.82990342e-01
6.78141117e-01 4.36601043e-01 -4.37152058e-01 -1.04649913e+00
-8.84850085e-01 -5.60830772e-01 -6.58473253e-01 4.04541045e-01
7.26759970e-01 -9.35883880e-01 -3.93696368e-01 3.79979163e-01
-1.11424136e+00 -2.38266498e-01 -3.58777523e-01 5.26759267e-01
-4.10025239e-01 3.51160109e-01 -8.71263981e-01 -7.24070549e-01
-7.01063812e-01 -9.37371016e-01 9.81185377e-01 2.67632365e-01
-6.04809225e-01 -1.25438392e+00 1.76294565e-01 3.04473639e-01
7.01976836e-01 -2.71749616e-01 7.89550900e-01 -9.37008321e-01
-3.31339031e-01 -3.66412252e-01 -3.31554785e-02 6.66232526e-01
1.65147081e-01 1.35212034e-01 -1.10163891e+00 -1.01756519e-02
5.25147878e-02 -1.26216859e-01 9.08835590e-01 3.18432719e-01
9.08985019e-01 -4.11140859e-01 1.25206085e-02 5.09763718e-01
1.25126672e+00 -2.60722190e-02 5.77367604e-01 6.25608981e-01
7.49301612e-01 5.48382103e-01 4.45924133e-01 2.57245868e-01
7.56427467e-01 4.49595422e-01 4.14925739e-02 -2.05419704e-01
6.70617446e-02 6.67633042e-02 4.82362360e-01 1.26606405e+00
-1.64246112e-01 -2.96767235e-01 -9.34090614e-01 9.56802845e-01
-1.54585302e+00 -6.80953622e-01 2.46862322e-02 1.91559339e+00
7.86945939e-01 2.16722399e-01 6.10598251e-02 -5.55742867e-02
3.23091954e-01 2.79226810e-01 -1.23542443e-01 -1.24848068e+00
-7.87982792e-02 7.74986863e-01 3.60107362e-01 6.91286981e-01
-1.23648977e+00 1.07502294e+00 6.68148851e+00 3.70043546e-01
-9.38882470e-01 1.34211928e-01 3.22618604e-01 -1.14830606e-01
-6.06332600e-01 1.37385219e-01 -6.32166982e-01 1.80943772e-01
1.18561280e+00 -1.70019329e-01 2.29938135e-01 9.85041857e-01
-4.05224450e-02 -6.71206135e-03 -7.98672438e-01 5.77510536e-01
3.60021770e-01 -9.78420258e-01 5.40328473e-02 -1.34204462e-01
4.50162262e-01 3.32213610e-01 1.16144955e-01 4.81129408e-01
3.42971742e-01 -8.89372170e-01 2.86421299e-01 8.38790089e-02
4.04118001e-01 -8.66233289e-01 1.10392344e+00 -2.47882590e-01
-1.03028333e+00 9.31747854e-02 -3.84530306e-01 -3.38694185e-01
1.09354347e-01 6.68166995e-01 -5.49582601e-01 3.69145215e-01
7.02182174e-01 8.72999489e-01 -6.11461043e-01 8.25612068e-01
-7.20182598e-01 8.36796463e-01 -2.68422037e-01 -3.65875095e-01
2.67839342e-01 -9.68581140e-02 5.10098934e-01 1.48736179e+00
1.49977133e-01 -2.22416043e-01 -1.83504537e-01 4.32170421e-01
4.13396694e-02 4.06025797e-01 -9.10875738e-01 -3.79690558e-01
8.51930529e-02 1.51239467e+00 -7.30775714e-01 -3.01803410e-01
-9.95189965e-01 8.79502237e-01 5.48415661e-01 4.58905548e-01
-6.08014166e-01 -8.23206007e-01 8.70295882e-01 -8.09930339e-02
1.00471258e+00 -3.96915764e-01 -4.42407429e-01 -1.54597116e+00
1.82260439e-01 -8.57569575e-01 4.01347309e-01 -5.31208634e-01
-1.29783058e+00 1.09367907e+00 -1.63533643e-01 -1.04475892e+00
-3.81431848e-01 -1.02827847e+00 -6.38037801e-01 6.17073059e-01
-1.77972007e+00 -1.19324386e+00 5.50292321e-02 3.14845622e-01
4.48889464e-01 -1.52598275e-02 1.34884453e+00 1.03248633e-01
-4.56419110e-01 7.55523324e-01 -1.85331672e-01 3.62062573e-01
5.11620104e-01 -1.40002310e+00 7.44794607e-01 7.34660864e-01
3.81877303e-01 1.10690331e+00 6.24593496e-01 -2.70051956e-01
-1.43749797e+00 -8.71534467e-01 1.52157116e+00 -8.79938364e-01
1.04657733e+00 -5.92658937e-01 -7.94998705e-01 1.01093411e+00
8.22477698e-01 -2.38807142e-01 1.23283076e+00 6.11038864e-01
-6.91358209e-01 7.43360296e-02 -8.21009576e-01 6.57488704e-01
5.86490214e-01 -5.78701973e-01 -9.55559731e-01 6.27842546e-02
7.88013875e-01 -1.79148931e-02 -8.33525360e-01 3.44034940e-01
4.69128758e-01 -5.83024204e-01 5.58820367e-01 -6.23455822e-01
3.42511356e-01 -1.77816734e-01 -9.61999223e-02 -1.33423531e+00
-5.07093668e-01 -4.69646424e-01 -3.56002927e-01 1.21580613e+00
9.92766857e-01 -9.33361709e-01 4.56703931e-01 7.32945621e-01
1.07735014e-02 -1.02370453e+00 -6.08112812e-01 -6.46419525e-01
2.61070639e-01 -6.33998513e-01 6.30775392e-01 8.72180521e-01
3.06979775e-01 6.32491171e-01 -1.21578157e-01 -2.18558498e-03
2.73559600e-01 3.73300403e-01 4.48583812e-01 -9.50718164e-01
-5.15847862e-01 -6.08019650e-01 -5.20782948e-01 -8.07995379e-01
2.32339635e-01 -7.66722739e-01 -6.08181283e-02 -1.60376668e+00
2.08815113e-01 -8.88689160e-02 -5.69659472e-01 7.98499584e-01
-3.41355205e-01 5.76632142e-01 5.72959892e-02 -1.23702437e-01
-6.29771829e-01 7.08692610e-01 7.24054456e-01 -2.03340918e-01
-8.33140090e-02 4.72454317e-02 -1.29436934e+00 7.91717410e-01
1.38790143e+00 -6.21258557e-01 -1.73035085e-01 -5.10414720e-01
4.18182343e-01 -8.03984523e-01 -7.48400763e-02 -4.54178721e-01
7.44909346e-02 2.63755143e-01 2.45262772e-01 -9.69977081e-02
4.12944645e-01 -6.69461370e-01 -5.74914932e-01 1.35680974e-01
-6.73215240e-02 3.80805939e-01 5.03520012e-01 2.58894563e-01
-4.45972353e-01 -3.71067017e-01 3.06601346e-01 -1.40483722e-01
-6.26918256e-01 2.48155996e-01 -6.94868147e-01 1.53012807e-03
6.84096754e-01 1.43067852e-01 -2.54818708e-01 -5.01807570e-01
-4.73889619e-01 -8.53972957e-02 5.67267656e-01 6.16980135e-01
5.29849887e-01 -1.22766340e+00 -5.07572174e-01 3.00389469e-01
4.39921111e-01 -2.82628864e-01 -1.96343690e-01 6.78660989e-01
-4.35888559e-01 5.22044301e-01 3.03049367e-02 -8.25138092e-02
-1.07184160e+00 6.95252776e-01 -1.20226331e-02 -1.42947808e-01
-5.49704432e-01 6.99001908e-01 3.26542929e-02 -7.17685282e-01
-1.24319233e-01 -3.33808005e-01 -4.84509170e-01 4.08520065e-02
6.70054197e-01 -2.16663674e-01 1.39170989e-01 -5.44159591e-01
-5.36884785e-01 4.66122389e-01 -3.67888510e-01 -1.32853195e-01
1.52129877e+00 -2.06520967e-02 -1.73720807e-01 6.57466650e-01
1.24728465e+00 2.85080791e-01 -7.64081120e-01 -5.94423153e-02
-3.68836671e-02 -2.28177831e-01 1.60123110e-01 -6.40599549e-01
-9.66184497e-01 7.64745116e-01 -3.44618373e-02 3.50542277e-01
1.16143322e+00 1.34183586e-01 6.73248470e-01 4.71433520e-01
8.32585841e-02 -9.22255456e-01 -3.10978323e-01 8.88420522e-01
6.64781272e-01 -1.25771284e+00 6.33893609e-02 -2.93425620e-01
-8.63016963e-01 1.00017273e+00 4.52588856e-01 -2.57755190e-01
9.78770494e-01 7.02270508e-01 4.96326476e-01 -3.03424299e-01
-8.94370556e-01 -5.45550525e-01 1.54507935e-01 5.83007932e-01
7.93911994e-01 2.27422286e-02 -4.24418300e-01 8.07304800e-01
-3.43136013e-01 -1.83181450e-01 7.22098410e-01 1.27238715e+00
-1.26491129e-01 -1.36080992e+00 1.91952199e-01 5.30851781e-01
-9.07917738e-01 -6.32837474e-01 -4.87894535e-01 7.02662170e-01
-3.54651451e-01 9.75292504e-01 2.20955331e-02 -7.86507800e-02
2.91437328e-01 1.66059107e-01 3.10125172e-01 -8.41678202e-01
-6.18552446e-01 -6.27823547e-02 3.84851128e-01 -5.24735510e-01
-5.33197045e-01 -6.42087638e-01 -1.00754690e+00 -2.28014678e-01
-2.71969259e-01 2.70593315e-01 8.49765718e-01 7.47886956e-01
7.31464207e-01 4.07734901e-01 5.04113555e-01 -6.25511467e-01
-1.19082965e-01 -1.10370684e+00 -4.33276266e-01 2.89736360e-01
2.66701490e-01 -5.07091820e-01 -5.09278953e-01 1.57702807e-02] | [11.405454635620117, 6.753469944000244] |
66e2d1c2-b753-4bc6-94e0-ac7b4bcec950 | road-damage-detection-using-deep-neural | 1801.09454 | null | http://arxiv.org/abs/1801.09454v2 | http://arxiv.org/pdf/1801.09454v2.pdf | Road Damage Detection Using Deep Neural Networks with Images Captured Through a Smartphone | Research on damage detection of road surfaces using image processing
techniques has been actively conducted, achieving considerably high detection
accuracies. Many studies only focus on the detection of the presence or absence
of damage. However, in a real-world scenario, when the road managers from a
governing body need to repair such damage, they need to clearly understand the
type of damage in order to take effective action. In addition, in many of these
previous studies, the researchers acquire their own data using different
methods. Hence, there is no uniform road damage dataset available openly,
leading to the absence of a benchmark for road damage detection. This study
makes three contributions to address these issues. First, to the best of our
knowledge, for the first time, a large-scale road damage dataset is prepared.
This dataset is composed of 9,053 road damage images captured with a smartphone
installed on a car, with 15,435 instances of road surface damage included in
these road images. In order to generate this dataset, we cooperated with 7
municipalities in Japan and acquired road images for more than 40 hours. These
images were captured in a wide variety of weather and illuminance conditions.
In each image, we annotated the bounding box representing the location and type
of damage. Next, we used a state-of-the-art object detection method using
convolutional neural networks to train the damage detection model with our
dataset, and compared the accuracy and runtime speed on both, using a GPU
server and a smartphone. Finally, we demonstrate that the type of damage can be
classified into eight types with high accuracy by applying the proposed object
detection method. The road damage dataset, our experimental results, and the
developed smartphone application used in this study are publicly available
(https://github.com/sekilab/RoadDamageDetector/). | ['Hiroya Maeda', 'Hiroshi Omata', 'Yoshihide Sekimoto', 'Takehiro Kashiyama', 'Toshikazu Seto'] | 2018-01-29 | null | null | null | null | ['road-damage-detection'] | ['computer-vision'] | [ 3.04449886e-01 -3.09321493e-01 2.15723678e-01 1.36785451e-02
-7.87782907e-01 -2.03539699e-01 2.02996776e-01 1.15023933e-01
-3.20076346e-01 5.47248483e-01 1.29827019e-02 -3.99363130e-01
1.36512771e-01 -1.54896843e+00 -7.62829781e-01 -7.52567112e-01
9.01194513e-02 4.36303206e-02 5.96833825e-01 -1.84025213e-01
3.89182001e-01 8.44524741e-01 -2.28500366e+00 -2.50087008e-02
1.23732638e+00 9.98925030e-01 6.02287591e-01 8.18200231e-01
3.60355616e-01 3.88275862e-01 -7.78532922e-01 -1.61581665e-01
1.61170706e-01 2.57496655e-01 -7.34448850e-01 1.00435484e-02
5.61476946e-01 -9.52812195e-01 -5.77104151e-01 8.48997712e-01
7.33784676e-01 2.51710834e-03 7.36373723e-01 -1.02821863e+00
-4.72706914e-01 -1.57779440e-01 -4.72850233e-01 5.32946765e-01
1.63105533e-01 2.90013105e-01 6.15216494e-01 -9.05701816e-01
1.75265536e-01 9.09115672e-01 5.03651619e-01 1.04957402e-01
-7.10155070e-01 -5.43488145e-01 -2.05173403e-01 5.09266913e-01
-1.63187766e+00 -2.43963137e-01 7.91613579e-01 -6.47052646e-01
9.90275800e-01 3.10145587e-01 6.02003336e-01 7.96962261e-01
-4.82814759e-03 5.39455235e-01 9.86732602e-01 -2.31555596e-01
1.08831868e-01 -1.82149544e-01 1.77463487e-01 5.67989469e-01
5.65713167e-01 2.76849955e-01 1.53455362e-01 -8.63302723e-02
5.70913672e-01 1.30576998e-01 -3.50776583e-01 1.73841909e-01
-7.59283900e-01 6.31488919e-01 5.67154408e-01 2.24121854e-01
-4.84015495e-01 -5.15285879e-02 3.33890140e-01 -7.29821920e-02
6.99289441e-01 -3.29267919e-01 -9.36130211e-02 -1.10497966e-01
-3.88277143e-01 -4.14189324e-02 4.78246927e-01 4.05252337e-01
9.58073616e-01 -7.28991553e-02 2.53485590e-01 9.61320102e-01
1.32540911e-01 8.85068417e-01 -7.56500140e-02 -7.53063023e-01
7.36861169e-01 6.04578674e-01 2.97425985e-01 -1.28737962e+00
-5.04125357e-01 1.87207237e-01 -9.19460773e-01 5.55839717e-01
4.90170211e-01 -2.91737556e-01 -8.52401793e-01 1.02930403e+00
4.70603555e-01 3.53849083e-01 4.80132885e-02 8.60490024e-01
8.33947480e-01 6.83845818e-01 -1.31139636e-01 2.37768531e-01
1.37169063e+00 -5.66132545e-01 -4.16443288e-01 -1.24281526e-01
9.04594779e-01 -6.10985696e-01 1.47837234e+00 4.46940839e-01
-5.37829995e-01 -5.04562676e-01 -9.22302246e-01 -1.05522923e-01
-6.69632077e-01 4.24493313e-01 2.58112997e-01 8.13333094e-01
-8.10834348e-01 1.80170402e-01 -6.57887578e-01 -6.90619051e-01
7.46659994e-01 -9.69714448e-02 -2.69717783e-01 -3.58783633e-01
-1.28739059e+00 1.08631539e+00 -4.24141018e-03 3.76029164e-01
-1.11550677e+00 -5.59359848e-01 -7.21589983e-01 -1.57361776e-01
3.46812785e-01 -3.11390698e-01 1.11694515e+00 -4.30800676e-01
-1.07282710e+00 1.00949323e+00 -6.21529147e-02 -2.23083720e-01
3.40589821e-01 -3.43341678e-01 -5.95301211e-01 1.36055350e-01
1.67526379e-01 2.61382759e-01 5.84698200e-01 -1.34201336e+00
-1.05751348e+00 -3.72097969e-01 4.19988602e-01 1.58208743e-01
-2.57871002e-01 1.22212969e-01 -6.55382499e-02 -3.87646765e-01
-4.41025406e-01 -7.33263850e-01 -8.99534393e-03 -6.13621958e-02
-2.81464398e-01 -7.06319362e-02 1.09821296e+00 -9.52239513e-01
1.47707200e+00 -2.12461567e+00 -4.50789690e-01 -1.02610523e-02
-1.48612140e-02 5.99688053e-01 1.53852746e-01 6.07509434e-01
4.92525063e-02 1.87694430e-01 -7.53276408e-01 -1.02550663e-01
-3.46939206e-01 5.03247976e-01 -4.92219687e-01 5.96836686e-01
-7.13354424e-02 4.74375248e-01 -7.60263145e-01 -3.29129070e-01
4.30999190e-01 4.54614311e-01 -4.17081006e-02 2.35345006e-01
4.31320637e-01 2.15369627e-01 -2.48842672e-01 9.56599116e-01
9.45090830e-01 4.95246977e-01 -5.08001506e-01 -7.52253607e-02
-4.49462563e-01 9.42325816e-02 -1.14466572e+00 9.06246006e-01
-6.27923191e-01 6.73355758e-01 1.38113737e-01 -1.05751717e+00
8.70343626e-01 3.21035445e-01 3.41821879e-01 -5.46921492e-01
7.32663348e-02 3.61807138e-01 -3.27710688e-01 -9.84788835e-01
6.21971726e-01 3.15082759e-01 -1.09000742e-01 3.86666656e-01
-7.44698048e-01 -3.21983427e-01 1.38257667e-01 -1.66670382e-01
1.14591241e+00 -3.68188292e-01 -5.62506355e-02 1.20699093e-01
4.61985230e-01 2.42980465e-01 2.49933258e-01 6.88611269e-01
-3.73476326e-01 7.69704401e-01 3.29013348e-01 -6.45308137e-01
-8.41594338e-01 -7.78002620e-01 -3.89295578e-01 7.63565242e-01
2.80967325e-01 2.65905350e-01 -1.14316082e+00 -5.03301680e-01
3.04786526e-02 5.78847885e-01 -5.56954980e-01 -2.42033005e-02
-6.52495980e-01 -1.09790111e+00 9.74773288e-01 5.07682443e-01
8.52946639e-01 -1.21854806e+00 -7.70275176e-01 8.46251845e-02
-4.32985961e-01 -8.32410634e-01 -2.92675197e-01 -4.44885820e-01
-6.06238246e-01 -1.53775048e+00 -5.49994230e-01 -7.35325575e-01
7.39035368e-01 1.01936495e+00 8.07358086e-01 4.58534509e-01
-5.96212566e-01 4.11342204e-01 -2.53478438e-01 -6.24210000e-01
-2.43309453e-01 5.52266017e-02 -1.84164122e-01 -2.60229292e-03
1.47599339e-01 -4.92058396e-01 -7.65942395e-01 7.02555656e-01
-9.58492756e-01 -3.38184774e-01 4.49728489e-01 3.09818923e-01
4.62103605e-01 5.06574869e-01 6.58016026e-01 -2.59613812e-01
5.45222819e-01 -6.27577305e-01 -6.90510571e-01 -1.02331169e-01
-2.83255875e-01 -7.85445869e-01 2.66070515e-01 -1.12079695e-01
-1.26505208e+00 -4.20329981e-02 -5.38894176e-01 1.52010098e-01
-8.91991019e-01 4.73748356e-01 -4.20833439e-01 -1.13660246e-02
8.07328701e-01 -2.87539996e-02 -1.05543517e-01 -5.31269789e-01
2.24521443e-01 1.30196857e+00 6.66600406e-01 -4.14088428e-01
6.74135804e-01 8.38011146e-01 -1.89185292e-01 -1.25383377e+00
-6.49680495e-01 -6.62260294e-01 -7.02092946e-01 -4.70842987e-01
7.80815661e-01 -9.58921015e-01 -4.63536471e-01 1.46864378e+00
-1.15319920e+00 -5.48028946e-01 -2.25044452e-02 2.60697186e-01
-2.92737871e-01 5.37178218e-01 -2.98249215e-01 -9.31236804e-01
-4.37832624e-01 -8.75500619e-01 1.11751580e+00 1.16854601e-01
3.55318904e-01 -6.35579348e-01 6.64476454e-02 6.28359556e-01
3.49032342e-01 3.21549565e-01 5.97019851e-01 1.63242310e-01
-5.16024351e-01 -3.88513625e-01 -5.36386490e-01 4.16693449e-01
5.17770946e-01 2.68212289e-01 -1.05647278e+00 -4.26370166e-02
-1.93389639e-01 1.23669818e-01 9.98256624e-01 4.25217539e-01
1.14708519e+00 -1.45523459e-01 -4.05497819e-01 1.13011397e-01
1.36179173e+00 2.58619636e-01 1.34782541e+00 4.14363742e-01
9.38221931e-01 7.85136759e-01 8.32689464e-01 3.04261655e-01
9.04996216e-01 7.57703483e-01 1.00079250e+00 -2.70453304e-01
-3.28038126e-01 7.63877556e-02 5.55234671e-01 3.25265735e-01
-4.13721710e-01 -4.99155432e-01 -1.25436068e+00 1.09621727e+00
-1.54115868e+00 -1.18676615e+00 -8.77912223e-01 2.36263061e+00
4.24365699e-01 -2.37090830e-02 3.51835251e-01 4.80915606e-01
1.03229535e+00 6.86815381e-02 -3.85873109e-01 -2.64299214e-01
-1.81725889e-01 -2.30511874e-01 7.37643361e-01 4.89235014e-01
-1.35248423e+00 6.15395963e-01 5.84117222e+00 8.13483834e-01
-1.15116620e+00 1.06546417e-01 3.97591859e-01 1.48146391e-01
1.21833138e-01 -2.85769880e-01 -7.82813847e-01 5.40686607e-01
8.18468392e-01 1.86360925e-01 2.75964588e-01 7.67135322e-01
7.23919868e-01 -7.13210523e-01 -1.49115965e-01 6.52373910e-01
-1.30730914e-02 -8.73505533e-01 -3.63188237e-01 2.32406735e-01
6.26103342e-01 3.52284670e-01 -2.69470215e-01 -9.58584249e-02
5.20132594e-02 -7.86746383e-01 5.08565187e-01 6.87088311e-01
7.35402346e-01 -8.34174752e-01 8.63960385e-01 5.01600087e-01
-1.25152040e+00 -3.27667058e-01 -4.33825642e-01 -4.15349126e-01
3.91482025e-01 9.67080057e-01 -6.95807040e-01 4.72461939e-01
1.11221504e+00 6.44050300e-01 -5.24610341e-01 1.24801838e+00
-5.58311403e-01 9.23937619e-01 -4.19310778e-01 3.55731249e-01
-6.17673546e-02 -1.12415902e-01 4.72709715e-01 1.02874744e+00
7.05272794e-01 2.03607738e-01 -3.00307125e-02 3.58330816e-01
1.05854407e-01 -1.00619391e-01 -9.64021027e-01 5.85278690e-01
4.72688675e-01 1.27039731e+00 -4.77336943e-01 -3.26226413e-01
-5.97326100e-01 6.11782193e-01 1.11253649e-01 1.27045780e-01
-9.80989456e-01 -7.13299572e-01 1.04779208e+00 5.37330925e-01
3.45144749e-01 -3.12873483e-01 -5.17026663e-01 -8.35542440e-01
1.65155917e-01 -4.04845864e-01 2.89003223e-01 -9.49827611e-01
-1.19115281e+00 1.82374954e-01 1.39783010e-01 -1.16264474e+00
4.32200938e-01 -4.58091080e-01 -9.32555437e-01 7.81008840e-01
-1.73534262e+00 -1.10029531e+00 -8.93591762e-01 3.85235965e-01
4.60876375e-01 4.67025876e-01 5.70256114e-01 7.98627675e-01
-1.06387091e+00 1.22007579e-01 1.07160628e-01 2.13492543e-01
5.92736363e-01 -9.70852494e-01 6.10861778e-01 9.76913154e-01
-3.44269663e-01 -1.39817342e-01 2.80924678e-01 -8.81199956e-01
-8.20701063e-01 -1.46216559e+00 7.68615067e-01 -5.45040309e-01
5.78316271e-01 -1.18952788e-01 -1.37726200e+00 3.59026521e-01
-1.83971133e-02 1.01800658e-01 9.42420214e-02 -4.35038090e-01
1.36416495e-01 -2.18531087e-01 -1.37159145e+00 3.79421979e-01
1.29588807e+00 -4.24160957e-01 -4.37636554e-01 2.85788566e-01
3.91716033e-01 -4.97928202e-01 -8.25162232e-01 2.88215905e-01
3.03672433e-01 -1.03248727e+00 9.74877357e-01 2.00587079e-01
3.84377420e-01 -6.71868563e-01 -1.81933224e-01 -1.24287009e+00
-3.02142254e-03 -6.44313917e-03 -3.97088192e-02 1.23095119e+00
8.57371837e-02 -1.08785164e+00 3.96608591e-01 3.87485743e-01
-7.65507460e-01 -7.38754332e-01 -1.04639709e+00 -7.46803105e-01
1.10985048e-01 -7.15834379e-01 9.63339686e-01 4.38692123e-01
-5.22470474e-01 -2.80468047e-01 -2.25022927e-01 8.48784924e-01
3.52869004e-01 3.20006832e-02 9.53752875e-01 -1.29990864e+00
7.24057138e-01 -3.33035409e-01 -4.14626122e-01 -8.09008420e-01
-1.30334213e-01 -4.34892893e-01 1.73802197e-01 -1.91433656e+00
-1.57600865e-01 -6.76629841e-01 8.37135762e-02 7.91482627e-01
-1.84308305e-01 3.97205293e-01 -4.16184425e-01 3.87849212e-01
1.85755327e-01 3.84653181e-01 1.06490433e+00 -2.90476352e-01
-1.07435316e-01 8.56133625e-02 -6.17014766e-01 8.62100601e-01
1.19752502e+00 -4.60558027e-01 -1.44156352e-01 -5.44285059e-01
1.02209099e-01 -3.15687567e-01 9.86174464e-01 -1.16935253e+00
1.28766866e-02 -2.45187402e-01 -2.71751225e-01 -9.22044039e-01
8.61004740e-02 -7.64303625e-01 1.90355688e-01 4.45151240e-01
4.85613227e-01 -2.78871894e-01 2.88335711e-01 4.54222888e-01
-1.69528052e-01 -3.24726701e-01 7.63807595e-01 2.41748050e-01
-1.00758445e+00 2.47739092e-01 -7.64056087e-01 -1.53760657e-01
1.24598062e+00 -5.70147872e-01 -7.69089460e-01 -1.63346171e-01
-2.23878130e-01 1.15833580e-02 5.88987648e-01 4.03192937e-01
5.96170247e-01 -1.19162405e+00 -8.42169344e-01 1.74124911e-01
2.96499908e-01 2.74976790e-01 7.39748597e-01 8.97292614e-01
-6.05208039e-01 1.47941276e-01 -8.67573395e-02 -4.56199259e-01
-1.23080647e+00 4.58379775e-01 3.87968689e-01 6.40647933e-02
-6.75363600e-01 1.44694790e-01 1.89395353e-01 -3.87243450e-01
-1.23053014e-01 -5.12859523e-01 -5.97827137e-01 3.19158047e-01
6.55762374e-01 1.02029276e+00 4.58280981e-01 -1.16117084e+00
-2.54321784e-01 8.96301925e-01 4.59020674e-01 3.71638060e-01
1.14848518e+00 -3.76359969e-01 -6.66545331e-02 -2.20234469e-02
9.18759048e-01 -5.10666892e-02 -1.25126088e+00 1.25549003e-01
-3.58228445e-01 -6.28457248e-01 2.42604077e-01 -5.95233202e-01
-1.25620937e+00 8.26786995e-01 9.27948713e-01 4.48950201e-01
1.34710324e+00 -5.03318124e-02 1.19725466e+00 3.60686481e-01
5.35361767e-01 -1.09439969e+00 -1.94502756e-01 5.81051826e-01
1.16894150e+00 -1.36773932e+00 -2.66046047e-01 -6.49338186e-01
-4.66990143e-01 7.86392152e-01 6.64512515e-01 3.91140543e-02
6.02953196e-01 1.59171015e-01 1.59996420e-01 -3.97202998e-01
3.46404016e-02 -6.57524586e-01 -1.08213564e-02 1.04130852e+00
-2.64464378e-01 2.73878783e-01 -2.12713450e-01 2.27779299e-01
1.58884432e-02 -3.93803306e-02 8.47479939e-01 9.67659950e-01
-9.55141068e-01 -8.57478440e-01 -8.73067081e-01 6.20792270e-01
-2.67441422e-01 1.25508621e-01 -1.48241594e-01 7.17619419e-01
4.07167017e-01 1.34090829e+00 3.55154015e-02 -2.75881559e-01
9.66200769e-01 -5.91746747e-01 1.75190177e-02 -5.37508488e-01
-2.61203766e-01 -6.21812940e-01 4.05298650e-01 -3.23540717e-01
-3.36018801e-01 -8.22766662e-01 -1.24212825e+00 -4.27554578e-01
-4.68175799e-01 -3.60873133e-01 6.19042456e-01 7.78968632e-01
5.31438828e-01 1.81577399e-01 8.64751577e-01 -1.09720457e+00
7.12392330e-02 -1.00440371e+00 -6.62563801e-01 1.73061252e-01
2.80184329e-01 -1.03428841e+00 -5.81521928e-01 7.25134313e-02] | [7.447514533996582, 1.1313998699188232] |
bff5c260-4db7-492b-935c-eecd47db72f7 | meta-mask-correction-for-nuclei-segmentation | 2111.12498 | null | https://arxiv.org/abs/2111.12498v1 | https://arxiv.org/pdf/2111.12498v1.pdf | Meta Mask Correction for Nuclei Segmentation in Histopathological Image | Nuclei segmentation is a fundamental task in digital pathology analysis and can be automated by deep learning-based methods. However, the development of such an automated method requires a large amount of data with precisely annotated masks which is hard to obtain. Training with weakly labeled data is a popular solution for reducing the workload of annotation. In this paper, we propose a novel meta-learning-based nuclei segmentation method which follows the label correction paradigm to leverage data with noisy masks. Specifically, we design a fully conventional meta-model that can correct noisy masks using a small amount of clean meta-data. Then the corrected masks can be used to supervise the training of the segmentation model. Meanwhile, a bi-level optimization method is adopted to alternately update the parameters of the main segmentation model and the meta-model in an end-to-end way. Extensive experimental results on two nuclear segmentation datasets show that our method achieves the state-of-the-art result. It even achieves comparable performance with the model training on supervised data in some noisy settings. | ['Chen Li', 'Chunbao Wang', 'Tieliang Gong', 'Zeyu Gao', 'Chang Jia', 'Jiangbo Shi'] | 2021-11-24 | null | null | null | null | ['nuclear-segmentation'] | ['medical'] | [ 5.32675207e-01 3.12495708e-01 -1.13797061e-01 -5.58605790e-01
-1.43066430e+00 -2.40431771e-01 1.73054680e-01 3.42410326e-01
-7.12610424e-01 6.90760016e-01 -9.97323543e-02 -1.51264323e-02
1.88122675e-01 -5.10238349e-01 -5.37717104e-01 -1.15676570e+00
5.73951960e-01 6.21759593e-01 5.30837893e-01 1.44264072e-01
9.69098508e-02 5.64286821e-02 -1.03752720e+00 3.76689881e-01
9.38963056e-01 7.84380794e-01 3.97732973e-01 5.88421702e-01
-2.56852418e-01 6.98098958e-01 -3.31313312e-01 -3.84066761e-01
1.50029689e-01 -4.86240864e-01 -9.48595762e-01 2.92876899e-01
2.76183844e-01 1.10589212e-03 -2.82560717e-02 1.31904292e+00
6.34202480e-01 6.00379705e-02 3.66953284e-01 -6.45019531e-01
5.39223701e-02 6.30482554e-01 -7.70433247e-01 3.47748458e-01
-5.50600767e-01 2.45724052e-01 6.27669096e-01 -6.92490518e-01
5.70064664e-01 8.13385725e-01 7.04949200e-01 7.46830940e-01
-1.29171669e+00 -4.89963979e-01 9.82840955e-02 -7.52945989e-02
-1.39787877e+00 -3.73625845e-01 5.94534218e-01 -2.52350122e-01
3.06699485e-01 2.96062082e-01 5.67000031e-01 5.55479527e-01
4.53263856e-02 7.51344323e-01 1.22404659e+00 -3.62098843e-01
2.55115151e-01 4.70038243e-02 2.34043151e-01 9.17100608e-01
3.50231938e-02 -5.46294510e-01 -1.83148772e-01 -1.23676367e-01
4.00878459e-01 5.77227287e-02 -1.79282248e-01 -2.22731620e-01
-1.29474890e+00 5.13838470e-01 3.72259438e-01 4.53927040e-01
-2.47409686e-01 8.68032873e-02 6.38010681e-01 -1.90690070e-01
8.79658580e-01 1.52531847e-01 -4.48253602e-01 7.81274736e-02
-1.26524997e+00 4.85719405e-02 5.18307030e-01 5.92585742e-01
7.62576103e-01 -4.03748989e-01 -5.44444144e-01 9.09722447e-01
3.28764498e-01 1.95020456e-02 7.06664741e-01 -8.33108664e-01
3.20735961e-01 9.49819207e-01 4.24087374e-03 -6.55154288e-01
-7.20848799e-01 -8.19767952e-01 -9.70203698e-01 -7.52269849e-02
5.44693708e-01 -1.21976122e-01 -1.37244582e+00 1.52803445e+00
9.16883290e-01 6.48548543e-01 -1.70754731e-01 9.03385222e-01
7.32964218e-01 4.64085162e-01 8.73178765e-02 -4.65306342e-01
1.30405009e+00 -1.40537870e+00 -9.74423170e-01 -1.09418593e-01
1.13299823e+00 -6.61143005e-01 8.97086382e-01 3.22608352e-01
-1.02344072e+00 -3.59528929e-01 -8.59887421e-01 -2.82021165e-01
-1.81632191e-01 3.44477475e-01 3.79701763e-01 6.15737796e-01
-7.78072596e-01 5.11705995e-01 -1.28451049e+00 -2.13616192e-01
7.62571573e-01 5.55275977e-01 -2.25058571e-01 4.47329395e-02
-5.63071966e-01 6.30284965e-01 7.85475910e-01 4.80822057e-01
-8.86692882e-01 -7.11888075e-01 -6.89824998e-01 -1.05899394e-01
6.73723638e-01 -5.15842438e-01 1.48939443e+00 -6.94238842e-01
-1.38212037e+00 1.05894089e+00 -2.51415700e-01 -2.57033944e-01
7.35139370e-01 9.69835445e-02 8.06134343e-02 8.95429850e-02
2.40872651e-01 6.54018700e-01 6.11617923e-01 -1.31423199e+00
-9.15138245e-01 -4.45379555e-01 -3.55915397e-01 1.73648596e-01
-2.37994239e-01 -3.15374583e-01 -9.34936225e-01 -5.90173125e-01
4.40599918e-01 -9.84980702e-01 -7.08230019e-01 -1.19961932e-01
-7.37880945e-01 -1.47161365e-01 7.52273858e-01 -7.04137146e-01
1.19150364e+00 -2.16492629e+00 1.38114497e-01 3.89453590e-01
4.53844666e-01 4.17388320e-01 1.96035475e-01 -2.49450609e-01
1.00117907e-01 1.67881697e-01 -7.15204298e-01 -9.45333004e-01
-3.17434609e-01 2.32534438e-01 2.26191550e-01 7.19647586e-01
-5.68484291e-02 7.39875555e-01 -1.03709006e+00 -9.64930236e-01
1.12457260e-01 1.47251561e-01 -4.70821053e-01 2.97041953e-01
-2.29107842e-01 8.18048894e-01 -3.79430532e-01 7.37653434e-01
7.58390605e-01 -4.76599991e-01 3.50434721e-01 -3.88194650e-01
-6.25424599e-03 -2.99979597e-01 -1.07993364e+00 1.91585636e+00
-1.48908496e-01 5.96210919e-02 2.72556454e-01 -1.15099990e+00
5.51598489e-01 1.96745113e-01 5.81141233e-01 -2.90660888e-01
4.67936426e-01 4.78052199e-01 4.42755371e-02 -5.76455951e-01
2.72734761e-01 -2.57628590e-01 7.75148394e-03 3.25449735e-01
1.42102480e-01 -4.81165983e-02 4.00431037e-01 -6.03843527e-03
1.10507858e+00 -5.02228998e-02 1.04159340e-01 -1.73081279e-01
6.80598974e-01 6.76427409e-02 1.01605809e+00 6.51936829e-01
-3.15369755e-01 7.22169280e-01 3.54862392e-01 -3.45129400e-01
-9.35658991e-01 -4.54542071e-01 -5.37859686e-02 8.02429914e-01
4.73837107e-02 -3.46065640e-01 -1.27175987e+00 -1.05740452e+00
-4.38471437e-01 2.48150453e-01 -8.01308095e-01 9.15571582e-03
-6.12398922e-01 -1.36179364e+00 5.35829246e-01 4.33725566e-01
6.07315123e-01 -7.38204241e-01 -3.36404175e-01 2.58271337e-01
-3.14885616e-01 -1.05616903e+00 -5.36315620e-01 3.84832621e-01
-9.99279797e-01 -1.05997765e+00 -7.63506711e-01 -1.01529646e+00
1.16052449e+00 1.97807178e-01 5.40938675e-01 5.44441998e-01
-2.51550943e-01 -2.82750964e-01 -2.08199143e-01 -2.60931969e-01
-4.75037426e-01 4.49871391e-01 -3.28192502e-01 3.08942407e-01
8.95673409e-02 -2.60492355e-01 -6.24939859e-01 1.63587064e-01
-1.23041236e+00 3.34563911e-01 7.23815799e-01 1.08636272e+00
1.26619077e+00 1.16879977e-01 5.11383355e-01 -1.60423851e+00
1.15655415e-01 -4.28220779e-01 -6.22207642e-01 2.89664418e-01
-4.20777529e-01 -2.94089410e-02 5.64976275e-01 -2.67956704e-01
-1.04351079e+00 5.53299665e-01 -2.62617022e-01 -2.64729202e-01
1.17703723e-02 6.07416809e-01 -2.57059723e-01 -1.82078928e-01
2.97444075e-01 8.76197070e-02 -1.35342494e-01 -7.04195559e-01
2.55159616e-01 6.92037821e-01 7.75314867e-01 -1.71371236e-01
6.62391424e-01 8.75795901e-01 2.63300128e-02 -4.12194192e-01
-1.33054054e+00 -7.88799524e-01 -8.10604990e-01 -1.03315659e-01
9.43913221e-01 -6.08068347e-01 -3.65009487e-01 7.12740779e-01
-9.13930535e-01 -5.73475182e-01 -7.94498622e-02 3.02085727e-01
-3.63140732e-01 4.21027273e-01 -7.01503277e-01 -4.75172669e-01
-4.64931607e-01 -1.43775809e+00 1.20387626e+00 4.06073630e-01
1.33135721e-01 -7.45648265e-01 2.92405903e-01 7.85091281e-01
3.64858657e-02 2.58917451e-01 8.80668938e-01 -8.50269377e-01
-5.68530977e-01 -2.31503934e-01 -1.85141385e-01 2.95953691e-01
1.96232975e-01 -2.22000092e-01 -1.02240968e+00 -3.07414830e-01
3.02610081e-02 -3.59983027e-01 1.11491203e+00 3.83282244e-01
1.67928696e+00 -1.26659675e-02 -6.10076129e-01 7.55893528e-01
1.37684178e+00 -7.81157166e-02 3.51791114e-01 2.29123652e-01
8.66280854e-01 3.67132217e-01 7.31563032e-01 1.32525325e-01
4.33525473e-01 5.71137071e-01 2.84035623e-01 -4.49709684e-01
-7.79042989e-02 -7.23499572e-03 -2.73161680e-01 9.81408954e-01
2.14030161e-01 -6.49328902e-02 -1.02211940e+00 6.16020322e-01
-2.25526333e+00 -5.90047300e-01 -2.01581821e-01 1.95367050e+00
1.27306855e+00 2.91420668e-01 -1.79637074e-01 1.99564219e-01
8.26924205e-01 -4.60940860e-02 -6.56279206e-01 1.81121856e-01
2.65182853e-01 1.29508391e-01 5.36305606e-01 3.97392958e-01
-1.34737289e+00 1.08136523e+00 5.74831676e+00 1.15745747e+00
-1.15010810e+00 6.18079364e-01 1.03057218e+00 -8.70885178e-02
4.19488028e-02 -1.00026906e-01 -7.45715678e-01 6.05688334e-01
8.27680826e-01 3.01825792e-01 -6.64873340e-04 6.30761564e-01
4.10019040e-01 -3.70642632e-01 -9.32893097e-01 9.85210896e-01
2.54542120e-02 -1.46791601e+00 -1.92335144e-01 4.72725146e-02
9.70793128e-01 -5.75388260e-02 -2.28508472e-01 3.24507475e-01
4.47037667e-02 -7.46029198e-01 5.71572363e-01 6.06246591e-01
5.80286980e-01 -8.63094509e-01 1.13608372e+00 6.06633186e-01
-9.08267736e-01 6.07513413e-02 -4.15858030e-01 5.00143170e-01
1.19077489e-01 8.52415562e-01 -8.28537703e-01 4.42260265e-01
5.40676415e-01 3.13611358e-01 -7.49946356e-01 1.34379625e+00
-3.99658531e-02 8.40231299e-01 -2.34472826e-01 3.13089758e-01
2.46975347e-01 -7.18720108e-02 2.81247199e-01 1.29946220e+00
-3.60831134e-02 2.17295840e-01 5.52711487e-01 5.55666327e-01
-4.69177455e-01 1.70344591e-01 4.97647002e-03 1.24212883e-01
2.45856717e-01 1.89357471e+00 -1.32724309e+00 -3.81737202e-01
-2.48337805e-01 8.32981825e-01 5.23711085e-01 1.29967406e-02
-8.85674953e-01 -1.60911754e-01 -1.21213116e-01 -2.88750287e-02
4.31417599e-02 8.57653096e-02 -4.92300302e-01 -9.68605697e-01
-8.93759504e-02 -7.20398664e-01 4.85609770e-01 -3.95586729e-01
-1.03974009e+00 4.25898612e-01 -2.80701786e-01 -9.93638456e-01
1.69708356e-01 -2.41061762e-01 -6.39167428e-01 5.54095984e-01
-1.62068176e+00 -1.30832887e+00 -5.02612829e-01 2.29944840e-01
5.38860440e-01 1.55857563e-01 5.80502272e-01 4.79614109e-01
-1.06918156e+00 6.22482240e-01 1.96756274e-01 3.26466054e-01
7.74882555e-01 -1.41135108e+00 -3.32156904e-02 9.48369026e-01
-1.98788166e-01 2.37088814e-01 3.86785895e-01 -7.89805651e-01
-9.69744444e-01 -1.45493042e+00 6.64479375e-01 -9.61821452e-02
3.15930605e-01 -3.07187468e-01 -1.05967176e+00 5.59094787e-01
7.39728808e-02 4.88890111e-01 9.92660105e-01 -1.99962720e-01
4.07086045e-01 -2.83965379e-01 -1.37717783e+00 5.69815636e-01
7.13104129e-01 -1.66318417e-01 -2.37657577e-01 6.22492790e-01
8.42966557e-01 -9.46695685e-01 -7.23605216e-01 4.66770053e-01
1.37291670e-01 -5.88499367e-01 4.59048390e-01 -4.62018937e-01
5.49596958e-02 -6.36123359e-01 2.53658205e-01 -1.23560762e+00
-2.38627478e-01 -6.02012396e-01 -8.37638043e-03 1.34496617e+00
5.61229587e-01 -2.40083590e-01 1.09603834e+00 6.44581378e-01
-5.24535060e-01 -1.09803629e+00 -1.13830125e+00 -3.16585153e-01
-2.06809849e-01 -1.66304022e-01 3.47642809e-01 8.25740576e-01
-1.63163379e-01 5.57188839e-02 -6.94137812e-02 2.31367603e-01
7.72126198e-01 -1.16981670e-01 6.18210196e-01 -1.01399541e+00
-1.03922375e-01 -1.57466486e-01 -2.57375687e-01 -5.81172526e-01
3.39427263e-01 -1.09310639e+00 4.58528280e-01 -1.52301776e+00
5.39502680e-01 -5.28326333e-01 -4.87535864e-01 6.72330439e-01
-5.82284570e-01 5.86742997e-01 -9.08588395e-02 2.34858587e-01
-8.88152659e-01 3.91167641e-01 1.26233375e+00 -2.64776468e-01
-1.50092959e-01 4.36582528e-02 -6.77421510e-01 9.65550363e-01
9.19183254e-01 -8.37792993e-01 -3.15053970e-01 -3.65743577e-01
8.62898584e-03 -2.14924991e-01 2.23964393e-01 -9.64748979e-01
5.58942974e-01 -9.07101408e-02 3.86888057e-01 -7.38639176e-01
1.63517267e-01 -6.59800708e-01 -1.94365367e-01 4.10821825e-01
-4.37066793e-01 -2.74340063e-01 1.94043182e-02 8.15817535e-01
-1.01185001e-01 -4.66856688e-01 1.11178970e+00 -3.39711130e-01
-3.22840959e-01 4.12765712e-01 -1.20814934e-01 8.73702914e-02
1.03598726e+00 -8.59271362e-02 -1.30900323e-01 2.77981013e-01
-9.10277009e-01 5.86117387e-01 3.25018227e-01 -1.54077977e-01
2.91591257e-01 -1.15511417e+00 -6.33929491e-01 -1.49224669e-01
2.06112186e-03 7.08694458e-01 4.37894404e-01 1.22117805e+00
-4.89407569e-01 1.27097592e-01 3.09460044e-01 -8.26864362e-01
-1.26949084e+00 3.24919641e-01 5.37424326e-01 -7.93757558e-01
-5.12214899e-01 9.85802948e-01 -9.92919412e-03 -6.06111526e-01
3.14101040e-01 -5.80018461e-01 -1.90652832e-01 3.80361676e-02
5.93817115e-01 3.43495041e-01 5.11959553e-01 -5.17222941e-01
-3.33851576e-01 2.54662335e-01 -4.83764797e-01 1.02534881e-02
1.38991249e+00 -1.58776343e-01 -3.98170412e-01 3.73077869e-01
1.11351228e+00 -1.21247426e-01 -1.25079191e+00 -3.99435192e-01
1.91697046e-01 -3.57948989e-01 4.61944133e-01 -7.20889449e-01
-1.37155485e+00 6.26295984e-01 7.88339972e-01 -1.89949974e-01
1.13482678e+00 -1.29637495e-01 1.08827686e+00 3.64446521e-01
1.13778815e-01 -1.48414922e+00 9.54767410e-03 1.61695957e-01
2.30515465e-01 -1.30037439e+00 -1.01369649e-01 -5.57122827e-01
-4.44067448e-01 9.17774022e-01 7.52609313e-01 1.71559110e-01
5.93894720e-01 3.40384752e-01 2.56518483e-01 -2.26452455e-01
-6.14826143e-01 -2.21262425e-01 6.65041283e-02 1.87343210e-01
2.68015295e-01 -4.62054722e-02 -4.56132144e-01 8.15205038e-01
2.55146921e-01 1.79799125e-01 4.49563622e-01 1.14534461e+00
-5.54548562e-01 -1.21893299e+00 -3.04540515e-01 6.90930128e-01
-7.33877182e-01 3.69363166e-02 -1.31619066e-01 3.74241650e-01
4.88974214e-01 8.16444397e-01 -2.49862596e-01 -1.61374211e-01
1.29152462e-01 1.57141805e-01 2.63734907e-01 -1.01119757e+00
-7.53678381e-01 4.34392363e-01 -2.75345683e-01 -3.79064798e-01
-6.99367046e-01 -6.07565343e-01 -1.67314625e+00 1.12111226e-01
-5.49559772e-01 1.28759801e-01 3.73592108e-01 1.32180345e+00
1.82572216e-01 6.87084138e-01 5.04332721e-01 -8.00720453e-01
-5.43075919e-01 -1.01271999e+00 -4.93624985e-01 2.55154252e-01
3.40602100e-01 -4.76942003e-01 2.46090088e-02 3.53519827e-01] | [14.780974388122559, -2.612905263900757] |
17b02a34-2da6-417d-85a4-42c8803b1fd4 | auxiliary-learning-as-a-step-towards | 2212.00061 | null | https://arxiv.org/abs/2212.00061v1 | https://arxiv.org/pdf/2212.00061v1.pdf | Auxiliary Learning as a step towards Artificial General Intelligence | Auxiliary Learning is a machine learning approach in which the model acknowledges the existence of objects that do not come under any of its learned categories.The name Auxiliary learning was chosen due to the introduction of an auxiliary class. The paper focuses on increasing the generality of existing narrow purpose neural networks and also highlights the need to handle unknown objects. The Cat & Dog binary classifier is taken as an example throughout the paper. | ['Christeen T. Jose'] | 2022-11-30 | null | null | null | null | ['auxiliary-learning'] | ['methodology'] | [ 2.75831521e-01 5.08128166e-01 -5.34673393e-01 -6.01478100e-01
-2.43222728e-01 -4.05151367e-01 8.38135898e-01 -2.49193814e-02
-4.60890889e-01 1.13376248e+00 -6.75178468e-02 -4.67129648e-01
-2.66369343e-01 -4.95217592e-01 -5.87131262e-01 -9.91681278e-01
-1.06014416e-01 4.73844260e-01 2.07994282e-01 -1.81376174e-01
3.36745530e-01 7.85421073e-01 -1.76632833e+00 4.00182962e-01
2.10476249e-01 1.03814292e+00 2.01839413e-02 4.71258909e-01
-2.27237046e-01 1.04797482e+00 -7.67036974e-01 -4.66931313e-01
3.09915274e-01 -2.49702454e-01 -9.69886661e-01 1.22357579e-03
4.29955930e-01 -1.76487476e-01 -2.60320425e-01 1.01497185e+00
8.16851780e-02 3.97880077e-01 8.53989184e-01 -1.87152994e+00
-7.18662918e-01 6.03526294e-01 5.72421588e-03 6.88749909e-01
-1.76165119e-01 -1.28792256e-01 1.17009914e+00 -9.93872106e-01
3.21286052e-01 1.10694611e+00 9.20550764e-01 5.62268078e-01
-1.07171106e+00 -5.73571026e-01 4.99579072e-01 5.39041460e-01
-1.28886080e+00 -2.95490980e-01 9.87559974e-01 -2.64568508e-01
1.24731421e+00 2.44988024e-01 4.71014380e-01 1.08301020e+00
9.36683416e-02 1.01828408e+00 1.27732277e+00 -6.85959578e-01
2.19001859e-01 4.55674797e-01 7.36427188e-01 4.14346009e-01
3.96007359e-01 4.65016633e-01 -1.28006324e-01 -1.04917116e-01
4.62884843e-01 -1.34044573e-01 5.36882319e-03 -5.46677947e-01
-8.42203021e-01 1.12012184e+00 5.25403678e-01 4.62114394e-01
-2.98772067e-01 -2.41258293e-01 4.16350096e-01 3.83409828e-01
3.73396516e-01 4.34192419e-01 -8.85580182e-01 2.73110047e-02
-6.39840603e-01 1.81952193e-02 9.00775790e-01 1.02068996e+00
5.97727120e-01 5.34353197e-01 1.43216908e-01 7.90794253e-01
-2.37466637e-02 1.27602130e-01 5.36445796e-01 -7.18025625e-01
9.59477276e-02 6.42076612e-01 -3.55165452e-02 -5.50528884e-01
-3.41725647e-01 -6.59439325e-01 -7.26113856e-01 5.23491800e-01
2.54548371e-01 1.41435474e-01 -1.18886209e+00 1.54613996e+00
7.83760392e-04 8.40025321e-02 3.40506613e-01 3.24138105e-01
1.32087708e+00 3.06116402e-01 3.75003159e-01 -2.82499194e-01
6.85617983e-01 -1.08099103e+00 -6.74256742e-01 -3.64528000e-01
3.29823971e-01 -4.94852394e-01 6.22820139e-01 5.18762887e-01
-7.45544791e-01 -8.70237231e-01 -9.74776387e-01 8.84933695e-02
-1.10950804e+00 8.06094706e-02 1.06839013e+00 5.82419217e-01
-9.05315280e-01 6.80858612e-01 -6.29898071e-01 -1.47673577e-01
4.85020787e-01 7.14773893e-01 -6.56331062e-01 2.33787298e-01
-9.95005190e-01 1.41646469e+00 9.84979689e-01 5.36080673e-02
-6.35815203e-01 -1.99084952e-01 -8.77570391e-01 -1.29211351e-01
1.98180437e-01 -1.49848118e-01 1.24279749e+00 -1.36221373e+00
-1.14024889e+00 1.04749596e+00 2.95583270e-02 -3.75171840e-01
4.01427865e-01 -1.55990094e-01 -7.04713047e-01 -1.31422421e-03
1.72672570e-02 8.16703439e-01 1.08143950e+00 -1.24643564e+00
-1.07612646e+00 -1.98386028e-01 2.56704599e-01 2.90308986e-02
-1.15909621e-01 -1.78827852e-01 6.44255280e-02 -9.97451603e-01
2.37448394e-01 -8.51288378e-01 -1.29861802e-01 -3.16035211e-01
-2.95436203e-01 -7.62887001e-01 1.20019674e+00 -2.56168306e-01
9.60421503e-01 -2.16239309e+00 -1.22122787e-01 1.12874217e-01
2.17562601e-01 3.45624208e-01 -1.75550014e-01 3.27166729e-02
-9.53900218e-01 -2.06935138e-01 -9.59673524e-02 3.71739864e-02
-1.27293512e-01 6.75459802e-01 -3.96111637e-01 5.16657710e-01
3.17036062e-01 1.07037485e+00 -7.63721049e-01 -3.73256326e-01
3.83814394e-01 1.95282847e-01 -1.38833821e-01 1.58532754e-01
1.16500504e-01 9.95700732e-02 -1.76524773e-01 9.57879901e-01
5.37952900e-01 -1.40855327e-01 -2.84436882e-01 -2.36112416e-01
2.56243479e-02 2.96973735e-01 -1.25861776e+00 8.65010619e-01
1.98234141e-01 7.45554924e-01 -2.21018661e-02 -1.72249281e+00
8.72941017e-01 3.37851256e-01 4.29173738e-01 -2.95245290e-01
4.11811471e-01 3.14320832e-01 6.06945395e-01 -4.35769945e-01
3.22950661e-01 -3.51324558e-01 4.95382734e-02 3.23933989e-01
3.60639930e-01 -2.21072361e-01 1.36531174e-01 -2.46732578e-01
6.74557149e-01 1.30969822e-01 9.36124504e-01 -3.97034794e-01
6.69736862e-01 2.24818483e-01 3.84444118e-01 1.03109813e+00
-4.76832330e-01 1.93330809e-01 1.27676800e-01 -9.21595693e-01
-9.51110125e-01 -1.02625573e+00 -5.61662138e-01 1.52753389e+00
-9.45185423e-02 1.49246633e-01 -2.92473704e-01 -9.78977859e-01
1.11709155e-01 8.53709340e-01 -1.11619830e+00 -2.10999608e-01
-4.69771445e-01 -9.13796186e-01 3.47605765e-01 9.81736541e-01
3.23647529e-01 -1.43139088e+00 -5.53670466e-01 1.01078846e-01
-2.91103553e-02 -6.99762881e-01 3.06899607e-01 1.17798591e+00
-1.17774463e+00 -1.47308826e+00 -5.39341629e-01 -1.02726603e+00
8.52016807e-01 1.30054504e-01 1.16675949e+00 1.11677341e-01
-1.09067693e-01 5.90070844e-01 -3.69744182e-01 -9.03076708e-01
-4.90399718e-01 1.44084170e-01 2.90823877e-01 -4.36629236e-01
1.28485680e+00 -5.40460408e-01 1.72543973e-01 1.47642940e-02
-6.72249436e-01 -3.86037380e-01 5.45626283e-01 1.26575685e+00
4.28914428e-01 2.18799710e-01 7.35982776e-01 -9.00677383e-01
4.22872543e-01 -6.00095630e-01 -4.43306804e-01 1.76562056e-01
-6.35132492e-01 -9.98420045e-02 3.92265648e-01 -8.25520217e-01
-7.34330833e-01 1.87219456e-01 -3.42689157e-01 -3.48203570e-01
-5.50679386e-01 3.68242532e-01 -1.76095426e-01 -3.66138726e-01
7.50646293e-01 3.37233156e-01 7.51223508e-03 -4.77809757e-01
2.20880508e-01 6.49365127e-01 4.29897785e-01 -2.77397573e-01
8.06450844e-01 2.46658042e-01 1.37414455e-01 -8.52982044e-01
-1.06097102e+00 -4.98705506e-01 -1.29659843e+00 3.88705097e-02
2.31277213e-01 -6.47169054e-01 -4.09698397e-01 3.81465226e-01
-9.51766133e-01 -1.98355794e-01 -8.45754325e-01 5.98821282e-01
-7.82008410e-01 9.86828357e-02 -5.30836999e-01 -8.14324915e-01
-1.95757970e-02 -9.61503863e-01 2.69871235e-01 -1.54704507e-02
-2.24612907e-01 -1.18073905e+00 -3.01227301e-01 -2.29132563e-01
3.29300195e-01 5.05437851e-02 9.59622741e-01 -1.43265474e+00
-2.46263683e-01 -4.85900968e-01 1.67210996e-02 7.42551506e-01
5.15880644e-01 -2.56577641e-01 -1.27990663e+00 -2.65681386e-01
2.10718274e-01 -4.18127477e-01 9.49221909e-01 4.11107212e-01
1.08332360e+00 -2.93007433e-01 -4.31788504e-01 5.09742916e-01
1.36044908e+00 5.87505996e-01 3.43836725e-01 6.70079410e-01
4.58472043e-01 7.57278144e-01 5.69640100e-01 -1.77839622e-01
-2.08645329e-01 2.34049022e-01 5.03065646e-01 -1.26640663e-01
6.41860738e-02 1.54688284e-01 -7.40342438e-02 4.14043427e-01
-2.70776361e-01 1.80558022e-03 -8.05436194e-01 6.66091979e-01
-1.53989267e+00 -1.15751445e+00 -2.63516694e-01 1.76118910e+00
5.00137448e-01 5.68749130e-01 3.63065787e-02 7.13700831e-01
6.69124246e-01 -1.09917268e-01 -5.92041850e-01 -5.57234883e-01
-3.72177482e-01 4.90753651e-01 3.12055111e-01 4.49635029e-01
-1.66108513e+00 7.05559909e-01 8.44339466e+00 7.08927155e-01
-1.10132587e+00 1.34567350e-01 5.85984588e-01 4.03915226e-01
3.20692688e-01 -3.77260774e-01 -1.02541637e+00 1.69261351e-01
7.69186318e-01 1.06344570e-03 -6.36246055e-02 1.49141443e+00
-4.22210842e-01 -7.83673897e-02 -1.51717734e+00 8.08232009e-01
2.81311363e-01 -1.08726931e+00 1.69772923e-01 -6.77449033e-02
5.01368940e-01 1.83724195e-01 5.21019660e-02 7.19694018e-01
4.22844619e-01 -1.11403239e+00 6.22166216e-01 2.90453821e-01
5.91862738e-01 -7.54758120e-01 9.96425927e-01 4.32221204e-01
-8.03632438e-01 -3.66940558e-01 -3.57907653e-01 -4.88206059e-01
-2.29123250e-01 -3.12239490e-02 -7.01968610e-01 -5.10359742e-02
6.68559670e-01 6.14584684e-01 -8.06760013e-01 1.14112759e+00
-4.66888070e-01 6.73995018e-01 -4.62704182e-01 -1.41776219e-01
5.80912054e-01 -3.05884182e-02 6.28845572e-01 1.26298678e+00
-4.56053466e-02 3.50826472e-01 3.27540219e-01 4.17449772e-01
1.55614689e-01 6.96538156e-03 -9.86813724e-01 2.17598826e-01
2.08160073e-01 9.06119287e-01 -9.51771975e-01 -6.41282916e-01
-6.90969646e-01 7.13279843e-01 3.25518161e-01 2.99862564e-01
-4.54043686e-01 -5.75967491e-01 5.25970459e-01 -1.35753870e-01
5.36623657e-01 4.06735957e-01 -3.78637999e-01 -8.68586302e-01
-1.15108952e-01 -6.68594003e-01 6.39757574e-01 -5.34829855e-01
-1.53360057e+00 7.56196916e-01 2.47564450e-01 -1.17771995e+00
-3.91981125e-01 -1.17346323e+00 -4.63600993e-01 8.94346833e-01
-1.52212203e+00 -1.21787834e+00 -2.14912593e-01 7.05374479e-01
6.89995766e-01 -8.42271745e-01 1.15183115e+00 -4.85107452e-02
-2.63274282e-01 6.07116163e-01 3.12313437e-01 2.47630984e-01
3.36795747e-01 -1.34928107e+00 1.45740271e-01 6.37405813e-01
2.60993931e-02 6.35208309e-01 9.54983592e-01 -3.99923086e-01
-5.34037352e-01 -1.12364697e+00 1.00596750e+00 -8.47591281e-01
6.02930009e-01 -1.31328367e-02 -9.99572217e-01 1.12716520e+00
2.52070367e-01 2.20176518e-01 7.82542169e-01 2.16987655e-01
-4.12717491e-01 5.10718524e-02 -1.21829152e+00 2.42517129e-01
6.79259121e-01 -4.54252750e-01 -1.49582613e+00 2.44919538e-01
3.13372433e-01 -2.91397035e-01 -4.60589856e-01 7.02382922e-01
4.40969974e-01 -7.85841882e-01 1.15884137e+00 -1.28700328e+00
2.32901603e-01 -3.17192376e-02 -3.88455689e-01 -1.20703566e+00
-7.48708010e-01 -4.75441664e-02 -5.45824289e-01 9.25063550e-01
2.03193620e-01 -6.16575658e-01 1.09556782e+00 5.14208674e-01
-8.00795332e-02 -5.65814018e-01 -1.09885097e+00 -1.08388674e+00
2.23241597e-01 -5.28122783e-01 2.86951005e-01 1.24600518e+00
-7.41256168e-03 6.44778252e-01 -1.87711552e-01 -2.94922799e-01
6.78545713e-01 5.64262345e-02 3.78730208e-01 -1.69220531e+00
-1.80686393e-03 -6.52431369e-01 -7.06753075e-01 -8.63035262e-01
4.09468681e-01 -8.32710981e-01 7.76769593e-02 -1.07748759e+00
1.05554812e-01 -5.85043490e-01 -7.80440748e-01 7.97597289e-01
-2.41617858e-01 7.58480430e-01 6.71415180e-02 1.46573320e-01
-3.98348927e-01 3.06495488e-01 8.19291651e-01 -4.36711848e-01
-7.47541562e-02 7.45491207e-01 -9.96598601e-01 1.16925192e+00
1.01627183e+00 -5.53933322e-01 -4.42386627e-01 6.84693828e-02
-2.65359730e-01 -5.74834049e-01 4.24366593e-01 -9.14165020e-01
-3.26133892e-02 -1.64333597e-01 8.05474043e-01 -8.53189886e-01
7.04465210e-01 -1.41052210e+00 -1.16569348e-01 7.40716279e-01
-4.77724165e-01 9.18553993e-02 3.72869641e-01 4.38176781e-01
-4.04199451e-01 -7.79878199e-01 1.05057585e+00 -3.64086747e-01
-1.17335391e+00 1.48052722e-01 -3.44957650e-01 -3.15228254e-02
1.26501441e+00 -7.07612574e-01 -4.09193225e-02 -7.24946558e-02
-1.21635997e+00 -2.32711077e-01 -6.64738491e-02 5.99343836e-01
5.48015058e-01 -1.43628240e+00 -4.90539908e-01 2.10019976e-01
3.89190346e-01 -4.60556686e-01 -3.48288238e-01 5.25930822e-01
-2.16342378e-02 9.16679561e-01 -6.07130587e-01 -4.70070779e-01
-1.41440296e+00 1.06947398e+00 3.31549704e-01 2.52055302e-02
-7.34512746e-01 1.01490843e+00 1.22191034e-01 -4.95033741e-01
7.76328206e-01 -1.09531380e-01 -7.59761810e-01 1.85080469e-01
6.05748832e-01 2.97795624e-01 1.47510171e-01 -9.43411469e-01
-3.24629664e-01 -5.75798079e-02 -2.26552889e-01 4.25739288e-01
1.75075030e+00 1.30402654e-01 -6.88174963e-02 8.96470070e-01
1.13518989e+00 -4.86070961e-01 -9.65362549e-01 -4.78819638e-01
2.58552760e-01 -3.16519797e-01 -1.13698497e-01 -7.87385881e-01
-7.17471957e-01 8.42999697e-01 8.11671913e-01 3.28745037e-01
1.14690733e+00 -5.14642373e-02 -5.36887534e-02 9.15686846e-01
5.19370377e-01 -1.32514679e+00 -1.88064829e-01 9.44404602e-01
8.42128754e-01 -1.49957299e+00 1.91538796e-01 -2.85898209e-01
-2.48500481e-01 1.35311747e+00 9.06531274e-01 -2.82897145e-01
9.53711331e-01 2.93915451e-01 5.04963398e-02 -2.83171143e-02
-6.76125705e-01 -1.90372333e-01 3.08655828e-01 1.16676509e+00
3.82475048e-01 -1.53155208e-01 -2.77914315e-01 5.77902555e-01
-2.75683194e-01 -5.80574349e-02 3.78726900e-01 1.10173571e+00
-6.26609147e-01 -9.05206382e-01 -4.48502660e-01 8.64111006e-01
-5.92521012e-01 -2.75656670e-01 -3.69932264e-01 1.28929639e+00
6.62254810e-01 5.93809247e-01 7.01460466e-02 7.33964965e-02
2.87796825e-01 3.05798173e-01 4.03877407e-01 -7.12068856e-01
-6.27715409e-01 -2.60527223e-01 -9.50232223e-02 -2.14356869e-01
-8.79465044e-01 -4.53018039e-01 -5.69749713e-01 1.03641720e-02
-4.69941199e-01 4.39306498e-01 4.30674821e-01 1.09425735e+00
-4.14182723e-01 5.22166908e-01 3.35601062e-01 -8.64140749e-01
-5.13803542e-01 -1.31882799e+00 -6.53498888e-01 3.72935116e-01
7.73334622e-01 -1.17750728e+00 -5.25443673e-01 2.65302718e-01] | [9.793664932250977, 2.912135601043701] |
f8500adf-4138-4273-8a0f-40d7a6c531ca | integrating-user-history-into-heterogeneous | null | null | https://aclanthology.org/2020.coling-main.372 | https://aclanthology.org/2020.coling-main.372.pdf | Integrating User History into Heterogeneous Graph for Dialogue Act Recognition | Dialogue Act Recognition (DAR) is a challenging problem in Natural Language Understanding, which aims to attach Dialogue Act (DA) labels to each utterance in a conversation. However, previous studies cannot fully recognize the specific expressions given by users due to the informality and diversity of natural language expressions. To solve this problem, we propose a Heterogeneous User History (HUH) graph convolution network, which utilizes the user{'}s historical answers grouped by DA labels as additional clues to recognize the DA label of utterances. To handle the noise caused by introducing the user{'}s historical answers, we design sets of denoising mechanisms, including a History Selection process, a Similarity Re-weighting process, and an Edge Re-weighting process. We evaluate the proposed method on two benchmark datasets MSDialog and MRDA. The experimental results verify the effectiveness of integrating user{'}s historical answers, and show that our proposed model outperforms the state-of-the-art methods. | ['Ying Shen', 'Haitao Zheng', 'Ziran Li', 'Dong Wang'] | 2020-12-01 | null | null | null | coling-2020-8 | ['dialogue-act-classification'] | ['natural-language-processing'] | [ 8.52960870e-02 6.49529099e-02 1.62615523e-01 -8.49247694e-01
-1.75066128e-01 -4.11346287e-01 9.06857789e-01 1.91725671e-01
-6.16262078e-01 6.08799458e-01 8.66418958e-01 -1.42242059e-01
9.03624445e-02 -6.55924976e-01 8.92640725e-02 -5.31591475e-01
9.66506749e-02 4.35170412e-01 2.32955292e-01 -5.03254175e-01
2.64652789e-01 2.20538154e-01 -1.11470938e+00 4.07732159e-01
9.12924647e-01 9.65396047e-01 1.23086132e-01 6.23526335e-01
-6.36571705e-01 1.31159651e+00 -7.47039557e-01 -4.68845129e-01
-2.17315212e-01 -9.23276186e-01 -1.09352875e+00 4.40273672e-01
-9.43165049e-02 -6.30039990e-01 -6.77385926e-01 1.14806104e+00
5.76189876e-01 7.86500990e-01 5.33958256e-01 -1.03642702e+00
-4.98839200e-01 7.19377339e-01 -3.86544764e-01 4.39594358e-01
6.77889287e-01 1.60433084e-01 1.11991072e+00 -6.89608037e-01
6.75841272e-01 1.51227713e+00 4.77966577e-01 8.75035048e-01
-6.79046869e-01 -3.54692161e-01 4.69428390e-01 3.86858284e-01
-1.10134864e+00 -5.26890457e-01 1.14169359e+00 -2.27839231e-01
7.54079521e-01 2.20916703e-01 3.08446199e-01 1.17613125e+00
-2.70848930e-01 1.14860594e+00 9.78380322e-01 -2.32204437e-01
1.63184389e-01 2.01004837e-02 6.77325010e-01 9.73823011e-01
-7.06558228e-01 -5.35393596e-01 -3.64453375e-01 -2.71122634e-01
5.47488272e-01 -9.21245813e-02 -5.03610790e-01 3.13694596e-01
-8.11278999e-01 9.74177480e-01 2.03061938e-01 4.38309580e-01
-3.39337021e-01 -2.37200156e-01 7.33721435e-01 4.09938574e-01
4.93506372e-01 1.19427525e-01 -2.21949130e-01 -2.79253125e-01
-3.87790143e-01 7.98386931e-02 1.00411880e+00 9.04390275e-01
7.24814415e-01 -4.99963164e-02 -6.68568730e-01 1.12760735e+00
2.91362554e-01 -2.75781572e-01 5.28641224e-01 -8.69907141e-01
4.19194430e-01 7.97378123e-01 3.09627075e-02 -1.08747423e+00
-7.17137098e-01 -2.76310205e-01 -1.03977907e+00 -3.58616292e-01
3.83616894e-01 -5.68045199e-01 -6.97102070e-01 1.88936162e+00
4.93250370e-01 3.16909432e-01 8.39231536e-02 9.80029762e-01
1.46070409e+00 8.53093028e-01 1.25017405e-01 -6.29012465e-01
1.45814073e+00 -1.04940903e+00 -1.27766538e+00 -1.54277042e-01
6.73888862e-01 -6.26872599e-01 9.83729720e-01 1.79704979e-01
-8.35882306e-01 -4.95930314e-01 -8.37928772e-01 -1.39760509e-01
-5.45669608e-02 7.66209587e-02 6.67976558e-01 4.35438931e-01
-7.61468530e-01 3.56317312e-01 -4.80852425e-01 -3.91697019e-01
1.36765346e-01 1.98870644e-01 -1.38610587e-01 1.88053064e-02
-1.51795113e+00 9.01936173e-01 2.49456480e-01 3.70186090e-01
-7.19014525e-01 -1.30627334e-01 -1.09477139e+00 1.49713874e-01
6.38864577e-01 -4.31103110e-01 1.64765525e+00 -8.42031837e-01
-1.76304340e+00 7.27282166e-01 -2.52897918e-01 -2.81932533e-01
3.30449790e-01 1.36565849e-01 -5.73006511e-01 1.01526231e-01
-2.67295539e-01 9.03496072e-02 4.42321420e-01 -1.12084639e+00
-7.38847375e-01 -5.17474949e-01 4.58853930e-01 6.78151488e-01
-2.93088615e-01 1.90889567e-01 -7.30964899e-01 -4.79399711e-01
8.90130624e-02 -6.12495482e-01 -2.51856714e-01 -4.90744531e-01
-3.49245042e-01 -7.84591436e-01 7.97464967e-01 -8.84937823e-01
1.65560663e+00 -2.13161969e+00 3.96971941e-01 1.01012014e-01
5.50803125e-01 2.76592851e-01 4.26087864e-02 5.60437202e-01
3.61242175e-01 -2.07636029e-01 -2.48032019e-01 -6.63077295e-01
1.15834430e-01 4.42560554e-01 3.25581804e-02 3.37332189e-01
-1.22923627e-01 5.24246633e-01 -9.85200524e-01 -4.72012490e-01
1.76895216e-01 2.02772498e-01 -1.62580743e-01 7.22139299e-01
-3.94630194e-01 5.72517574e-01 -8.28416646e-01 2.75525242e-01
4.77537930e-01 -2.39742205e-01 3.81786078e-01 -2.48053312e-01
1.50756419e-01 4.73819584e-01 -1.15022790e+00 1.70511675e+00
-3.62044156e-01 3.80874097e-01 3.41367960e-01 -8.70398402e-01
1.04540443e+00 5.02962947e-01 3.13326538e-01 -5.29044807e-01
5.10545313e-01 -9.53091606e-02 1.44932091e-01 -8.11909676e-01
4.17361557e-01 2.98444815e-02 -1.70012489e-01 6.79863930e-01
2.86604285e-01 6.05811924e-02 1.70947000e-01 4.23743784e-01
1.31756473e+00 -2.88189948e-01 3.34190279e-01 7.91755840e-02
1.05389142e+00 -4.11039054e-01 8.09376597e-01 5.05664051e-01
-5.40665746e-01 3.62364292e-01 6.86682403e-01 -5.54969311e-01
-4.90112096e-01 -5.31428695e-01 2.72865474e-01 1.35697258e+00
1.17436819e-01 -3.68233979e-01 -9.07452583e-01 -1.03012133e+00
-4.16114151e-01 8.09558153e-01 -6.67766571e-01 -9.86751243e-02
-5.85316896e-01 -6.96144044e-01 6.11746311e-01 3.61036569e-01
1.11713243e+00 -1.35287488e+00 -7.20686615e-02 4.72710013e-01
-4.56395596e-01 -1.12499738e+00 -8.28731835e-01 -1.79418713e-01
-2.86771029e-01 -8.86391640e-01 -4.99418557e-01 -8.38044763e-01
4.81796563e-01 1.04461767e-01 9.77565706e-01 3.86543274e-01
1.58151656e-01 4.08194393e-01 -5.99499762e-01 3.53374816e-02
-4.95613366e-01 3.83270904e-02 -8.86200666e-02 5.67742467e-01
5.03028393e-01 -5.42327940e-01 -5.79021454e-01 3.03671777e-01
-8.01055431e-01 -8.44733603e-03 1.36884093e-01 8.72551203e-01
-3.25904489e-02 6.19352348e-02 7.72286057e-01 -1.24078822e+00
1.35832894e+00 -5.46856642e-01 -2.48792857e-01 3.73234808e-01
-3.81065398e-01 -1.15202218e-02 6.05866373e-01 -3.12920988e-01
-1.61844635e+00 1.21704070e-02 -5.05621552e-01 -1.99095383e-02
-3.17443639e-01 7.58140147e-01 -2.91168571e-01 2.41662666e-01
4.42727536e-01 2.44749874e-01 -6.85959607e-02 -4.85594064e-01
4.65457439e-01 9.45600152e-01 6.69820368e-01 -4.90599334e-01
8.49183947e-02 1.04347065e-01 -5.26514113e-01 -9.01856303e-01
-8.86303782e-01 -7.28902996e-01 -3.28098834e-01 -5.36499798e-01
1.14894772e+00 -6.18283093e-01 -1.07011938e+00 7.29420543e-01
-1.51003623e+00 4.59129475e-02 2.71309823e-01 3.98021787e-01
-3.45468879e-01 7.39919484e-01 -1.15088785e+00 -1.10901225e+00
-4.71168518e-01 -1.11289489e+00 5.64525485e-01 5.62783360e-01
-4.21860605e-01 -1.05836105e+00 -6.00209273e-02 5.42362452e-01
1.49069384e-01 8.97874609e-02 9.22791839e-01 -1.00593948e+00
-9.48623568e-02 -1.24557167e-01 -2.29066223e-01 3.47503632e-01
2.56799817e-01 -2.54042149e-01 -9.85988200e-01 -3.66521031e-02
4.50814217e-01 -3.43525797e-01 7.93658495e-01 7.03570247e-02
8.31804872e-01 -5.84100127e-01 -8.07865858e-02 1.54418498e-01
6.63658857e-01 6.41716778e-01 6.52529776e-01 -2.04456270e-01
6.32152617e-01 8.65932226e-01 3.79553825e-01 7.70759106e-01
7.34110713e-01 5.25401413e-01 2.42833734e-01 3.51222381e-02
1.63856804e-01 -1.09188646e-01 2.23954022e-01 1.16763520e+00
-7.09282160e-02 -5.87677360e-01 -7.05347896e-01 3.78484309e-01
-2.05588460e+00 -9.44386721e-01 -2.80364007e-01 1.70989871e+00
9.93942559e-01 2.61880398e-01 1.25776082e-01 -7.58725628e-02
8.58132720e-01 5.74146867e-01 -4.39798236e-01 -3.92743170e-01
-4.46730293e-02 2.65418389e-03 -8.80062506e-02 6.83004558e-01
-1.00849402e+00 8.39197457e-01 5.04578781e+00 8.61205280e-01
-7.84010410e-01 2.17440546e-01 6.76983714e-01 5.08395910e-01
-2.87469506e-01 -2.57336318e-01 -6.66296303e-01 4.14943606e-01
5.92811406e-01 -2.26749286e-01 4.61021721e-01 5.62839687e-01
2.58994401e-01 8.83599669e-02 -9.91375208e-01 8.23858380e-01
2.79701471e-01 -1.29270256e+00 -2.17131317e-01 -4.95967120e-01
3.97666126e-01 -2.78785825e-01 -4.99221504e-01 5.11220634e-01
5.98762155e-01 -8.42266858e-01 4.08851914e-02 6.38254642e-01
2.76028067e-01 -7.38338232e-01 9.16544795e-01 5.22843897e-01
-1.20856166e+00 1.13466308e-02 -3.55712734e-02 -2.29164243e-01
1.70642599e-01 2.10336253e-01 -1.02922404e+00 5.04374683e-01
3.75538617e-01 5.47689557e-01 -1.69996575e-01 5.95880628e-01
-4.01999325e-01 7.82773376e-01 -1.06527604e-01 -5.06058037e-01
5.55360138e-01 -3.96526426e-01 3.64727437e-01 1.32129383e+00
-7.21531808e-02 7.77621388e-01 3.94312263e-01 6.08765602e-01
-4.12896991e-01 3.20287228e-01 -4.61054355e-01 -2.18157873e-01
5.34755290e-01 1.29304147e+00 -5.25334239e-01 -3.49858373e-01
-6.12358928e-01 1.33068120e+00 4.44178134e-01 3.55537117e-01
-5.36498487e-01 -4.94486272e-01 4.30102944e-01 -3.51233423e-01
-2.16384619e-01 -1.62841141e-01 -3.68898734e-02 -1.02317965e+00
-6.60986453e-02 -8.65503728e-01 7.91014671e-01 -6.46141350e-01
-1.45355487e+00 8.10242951e-01 -3.59076917e-01 -9.01115358e-01
-2.96195805e-01 -2.17525885e-01 -9.91866052e-01 8.28941941e-01
-9.84758973e-01 -9.18780446e-01 -4.87557232e-01 6.53503954e-01
8.88360679e-01 -2.20495805e-01 9.09861445e-01 4.47596401e-01
-6.31111622e-01 5.31770408e-01 -2.42661104e-01 4.86521184e-01
5.02476573e-01 -1.18150735e+00 2.74316937e-01 5.48534870e-01
-1.94885567e-01 5.39549291e-01 8.03969860e-01 -5.80351174e-01
-1.08256388e+00 -8.27280760e-01 1.06795967e+00 -1.63538922e-02
4.75469232e-01 -2.65902072e-01 -1.20318878e+00 7.31709719e-01
6.61429763e-01 -3.83404702e-01 7.30512202e-01 1.60120010e-01
-2.55312002e-03 2.79811770e-02 -1.18902242e+00 7.26185262e-01
1.03605199e+00 -4.64108139e-01 -8.98782074e-01 3.44285697e-01
8.34645808e-01 -4.40104425e-01 -6.69851363e-01 2.40982324e-01
1.92822531e-01 -8.07051361e-01 5.79240859e-01 -7.19442070e-01
2.77196407e-01 -1.68921009e-01 1.00909390e-01 -1.40401351e+00
-2.35139444e-01 -8.67243230e-01 1.32098585e-01 1.55727839e+00
1.97126374e-01 -3.31114322e-01 5.80152929e-01 8.77212703e-01
-4.48088109e-01 -5.08183718e-01 -9.61396754e-01 -1.14758722e-01
-5.20979106e-01 -2.00322211e-01 6.29122198e-01 1.36523342e+00
3.92691404e-01 1.03281748e+00 -7.56628871e-01 1.52951432e-02
9.81072858e-02 -1.22686058e-01 6.41508162e-01 -1.10792542e+00
-3.39031518e-01 -5.32264292e-01 -1.64572015e-01 -1.52811325e+00
3.58116180e-01 -6.13950491e-01 1.82801932e-01 -1.67682278e+00
-1.37854874e-01 -1.93215236e-01 -1.41260698e-01 2.56642044e-01
-4.21142668e-01 -3.60312641e-01 -1.07427083e-01 -1.11325033e-01
-7.37224102e-01 9.86407399e-01 1.36023927e+00 -2.94632256e-01
-4.40684378e-01 1.44068852e-01 -3.25790495e-01 9.07810628e-01
6.15703404e-01 -6.11106381e-02 -6.57047033e-01 -3.59910846e-01
-1.02221996e-01 7.13661790e-01 -2.11968407e-01 -6.12612486e-01
5.04599750e-01 -2.47314975e-01 -7.16220364e-02 -5.37743211e-01
5.35501301e-01 -5.42974055e-01 -1.98946744e-01 1.56234980e-01
-5.47935307e-01 4.60721813e-02 -1.24467000e-01 5.66746056e-01
-3.37636232e-01 -4.69798446e-01 6.14018738e-01 -3.88400167e-01
-9.23104048e-01 2.52117068e-01 -7.24591792e-01 1.28217593e-01
6.93302095e-01 2.27946937e-01 -4.13782775e-01 -9.18466508e-01
-8.30938518e-01 7.05201268e-01 -3.03569943e-01 3.94289136e-01
8.28770399e-01 -1.25314450e+00 -7.32562542e-01 1.18144425e-02
1.77705690e-01 -2.73676738e-02 5.76500118e-01 6.09321594e-01
-3.89631242e-01 -2.07997069e-01 -1.04653519e-02 -7.20610246e-02
-1.40830636e+00 2.81318188e-01 3.67466688e-01 -3.80111963e-01
-4.31350082e-01 1.10521281e+00 2.57110279e-02 -6.56312883e-01
6.16014183e-01 -3.71228494e-02 -8.08958888e-01 3.24248463e-01
4.70560580e-01 3.45712125e-01 -1.58418357e-01 -7.07785130e-01
-7.83934519e-02 -5.98820001e-02 -4.08434153e-01 -2.10611969e-01
1.09456420e+00 -4.19186175e-01 -2.32824832e-01 5.87317348e-01
9.66305196e-01 -2.46156648e-01 -8.35294783e-01 -7.59301662e-01
8.71088281e-02 -2.59551167e-01 -4.94496562e-02 -6.83641732e-01
-1.01790583e+00 8.42988014e-01 3.01657200e-01 3.79043788e-01
1.05803657e+00 -1.34175628e-01 1.10298955e+00 5.19258440e-01
1.68722928e-01 -1.42376113e+00 1.78104341e-01 9.02289271e-01
9.59869385e-01 -1.15991426e+00 -3.12591583e-01 -4.31331217e-01
-1.05753279e+00 1.03977895e+00 9.05359089e-01 2.25785881e-01
5.38201630e-01 -2.81357348e-01 3.26340199e-01 -3.96026105e-01
-6.80048883e-01 -3.74993920e-01 9.48037207e-02 1.91911310e-01
5.07962584e-01 -6.14073798e-02 -8.49834919e-01 8.68646801e-01
9.72271562e-02 -1.95850134e-01 6.19362056e-01 8.64206731e-01
-4.27075684e-01 -1.05077910e+00 1.16473801e-01 4.84776646e-01
-2.70598561e-01 -6.71640784e-02 -6.94439530e-01 3.74336153e-01
-8.35806578e-02 1.42510831e+00 -6.57619759e-02 -7.73747444e-01
3.57704788e-01 4.04424250e-01 1.01327047e-01 -7.06461549e-01
-9.16468084e-01 1.22041948e-01 6.59404039e-01 -1.77118376e-01
-4.15467560e-01 -4.80091602e-01 -1.71841145e+00 -4.12600547e-01
-2.28813365e-01 3.00564200e-01 2.09356442e-01 1.28227937e+00
5.23774736e-02 6.65557683e-01 9.51233149e-01 -1.81666270e-01
-6.95049047e-01 -1.41791010e+00 -6.25375330e-01 9.34474885e-01
2.48285845e-01 -5.89275599e-01 -3.94424587e-01 -1.14261508e-01] | [12.6980619430542, 7.727793216705322] |
56af1c04-9543-4706-8aee-f4b3f1c01bb4 | understanding-diffusion-models-a-unified | 2208.11970 | null | https://arxiv.org/abs/2208.11970v1 | https://arxiv.org/pdf/2208.11970v1.pdf | Understanding Diffusion Models: A Unified Perspective | Diffusion models have shown incredible capabilities as generative models; indeed, they power the current state-of-the-art models on text-conditioned image generation such as Imagen and DALL-E 2. In this work we review, demystify, and unify the understanding of diffusion models across both variational and score-based perspectives. We first derive Variational Diffusion Models (VDM) as a special case of a Markovian Hierarchical Variational Autoencoder, where three key assumptions enable tractable computation and scalable optimization of the ELBO. We then prove that optimizing a VDM boils down to learning a neural network to predict one of three potential objectives: the original source input from any arbitrary noisification of it, the original source noise from any arbitrarily noisified input, or the score function of a noisified input at any arbitrary noise level. We then dive deeper into what it means to learn the score function, and connect the variational perspective of a diffusion model explicitly with the Score-based Generative Modeling perspective through Tweedie's Formula. Lastly, we cover how to learn a conditional distribution using diffusion models via guidance. | ['Calvin Luo'] | 2022-08-25 | null | null | null | null | ['3d-absolute-human-pose-estimation'] | ['computer-vision'] | [ 3.76439877e-02 3.32491934e-01 6.79526180e-02 -1.47864595e-01
-9.14553702e-01 -6.05127394e-01 1.00234354e+00 -4.96887475e-01
-1.69166043e-01 6.05015218e-01 5.22273898e-01 -2.87801236e-01
-3.73365462e-01 -9.43288803e-01 -7.35839844e-01 -1.24693024e+00
1.19959213e-01 6.82409108e-01 -1.65712595e-01 -7.64819235e-02
-4.59604460e-04 2.92821944e-01 -1.32027602e+00 1.36161357e-01
7.15819538e-01 6.85326099e-01 2.70614803e-01 1.11502433e+00
-4.01914269e-02 1.01147270e+00 -7.55791068e-01 -3.92834276e-01
7.61338621e-02 -8.34908485e-01 -6.74745917e-01 2.60441393e-01
3.69164854e-01 -4.80722100e-01 -7.53400266e-01 1.18873656e+00
4.09520596e-01 3.77366960e-01 1.26232827e+00 -1.27362192e+00
-1.46284223e+00 6.46657705e-01 -3.57829630e-01 3.22157621e-01
-8.12619105e-02 2.72741139e-01 1.24663854e+00 -7.28927612e-01
1.01176560e+00 1.12791860e+00 2.36658663e-01 8.89949441e-01
-1.45539379e+00 -5.62249720e-02 3.12913209e-01 -8.41353089e-03
-1.12538421e+00 -2.27533415e-01 6.46564841e-01 -7.60852396e-01
7.90902197e-01 1.43327892e-01 8.63037705e-01 1.54813302e+00
3.51602346e-01 1.24749005e+00 9.36164200e-01 -3.36175114e-01
4.46186870e-01 -2.96813343e-02 -2.45280713e-01 6.92679524e-01
-2.17566386e-01 2.69052774e-01 -6.19292140e-01 6.58325776e-02
1.04092562e+00 -1.80664584e-02 -2.11178496e-01 -3.72682929e-01
-9.76565540e-01 1.23083544e+00 5.12534618e-01 2.57867664e-01
-4.43360329e-01 6.45986438e-01 -1.95228517e-01 1.82511136e-01
5.36918879e-01 2.26185173e-01 -1.32697940e-01 7.66200125e-02
-1.31051099e+00 4.18780297e-01 6.76126480e-01 8.14184368e-01
7.18656182e-01 4.64456141e-01 -5.19160271e-01 4.97004420e-01
6.84639156e-01 8.20864022e-01 1.83725789e-01 -1.21925676e+00
1.36408746e-01 -9.10151079e-02 3.69977131e-02 -5.37930012e-01
3.49713974e-02 -4.41311091e-01 -8.71501327e-01 3.81769896e-01
2.64738739e-01 -5.93729973e-01 -1.28462803e+00 1.93693066e+00
8.03206488e-02 7.24719837e-02 2.45030802e-02 9.02151644e-01
7.08174646e-01 1.00798428e+00 -1.71796810e-02 -1.81063339e-01
7.01845527e-01 -9.25851405e-01 -7.42312193e-01 -5.23347631e-02
1.90920442e-01 -6.03355646e-01 7.12621272e-01 4.09760207e-01
-1.31526041e+00 -2.36347139e-01 -9.27569032e-01 -1.19861610e-01
-1.82020739e-01 -6.47088587e-02 7.12202609e-01 4.84869987e-01
-1.40680337e+00 7.28321910e-01 -1.08417630e+00 -1.68894947e-01
3.65177900e-01 6.29569450e-03 1.54859737e-01 9.73838381e-03
-1.06269085e+00 8.26353014e-01 -4.68002409e-02 6.05830643e-03
-1.71499169e+00 -6.83611572e-01 -6.47312105e-01 5.63360937e-02
2.33347818e-01 -1.15926182e+00 1.38689995e+00 -8.81553411e-01
-1.66127753e+00 5.98876774e-01 -1.75263941e-01 -2.86442876e-01
7.41290748e-01 2.95393374e-02 -2.25779220e-01 8.07780474e-02
7.85748959e-02 8.76668811e-01 1.04562604e+00 -1.39884448e+00
-3.99262935e-01 -1.27279311e-01 1.11582667e-01 2.66827136e-01
-7.94098601e-02 -4.69982535e-01 -3.17678183e-01 -6.67675972e-01
-3.63492370e-02 -7.35851824e-01 -3.95704597e-01 -1.71920750e-03
-6.36563241e-01 -2.31677204e-01 6.31232202e-01 -3.77780497e-01
1.02272952e+00 -1.97819376e+00 7.49718547e-01 2.21146077e-01
6.76414430e-01 -2.17656255e-01 -2.93860823e-01 5.96442521e-01
1.66074380e-01 2.54765272e-01 -2.59961784e-01 -7.00488746e-01
3.77231836e-01 4.05469298e-01 -4.90203917e-01 3.08897793e-01
1.88015789e-01 1.27163041e+00 -9.69801724e-01 -3.27836245e-01
1.89158499e-01 8.49009216e-01 -8.69426370e-01 9.34695527e-02
-4.21712041e-01 4.21960264e-01 -3.76225650e-01 2.86411315e-01
4.12175089e-01 -4.35202926e-01 -4.42556068e-02 1.68504089e-01
1.22323502e-02 -1.62621066e-01 -1.15387774e+00 1.67006803e+00
-3.71175259e-01 9.36193407e-01 2.45569542e-01 -5.90601146e-01
5.19913554e-01 2.64558941e-01 3.89822155e-01 -3.61522049e-01
1.40298441e-01 -5.97427376e-02 -1.75340474e-01 -3.40270668e-01
6.22685909e-01 -5.16300797e-01 7.68334195e-02 7.06360996e-01
5.75811744e-01 -3.42409372e-01 2.10475862e-01 7.65847147e-01
9.54607606e-01 7.09093362e-02 -2.09763139e-01 -2.47937500e-01
-1.02614880e-01 -2.27003217e-01 7.58612081e-02 1.25221109e+00
-1.18884958e-01 7.96559274e-01 5.74316025e-01 -1.71087515e-02
-1.27759326e+00 -1.76786613e+00 5.96966296e-02 9.50953245e-01
-9.99541506e-02 -3.37263584e-01 -9.59183395e-01 -4.14609045e-01
-1.93129212e-01 9.36865747e-01 -1.06454980e+00 -1.36627093e-01
-6.53634891e-02 -1.03301990e+00 3.37028384e-01 4.55741197e-01
1.61421686e-01 -9.61223841e-01 -2.15304658e-01 1.20616015e-02
-1.44275382e-01 -4.90819693e-01 -5.84950387e-01 1.56828284e-01
-7.65480220e-01 -5.87498546e-01 -1.04437172e+00 -4.43044811e-01
6.73046172e-01 -1.41762570e-01 1.19167709e+00 -1.09826684e-01
-1.89312100e-01 6.62617445e-01 4.71162051e-03 -2.97750562e-01
-6.15908861e-01 -2.34444663e-02 -9.90006924e-02 1.26776755e-01
7.28109777e-02 -6.03357375e-01 -7.66774118e-01 -1.30553335e-01
-1.13888741e+00 -2.77710636e-03 5.23169339e-01 7.43752360e-01
7.39739239e-01 1.42295942e-01 1.64091736e-01 -7.05499649e-01
9.32976723e-01 -6.49025679e-01 -4.72235322e-01 1.33029535e-01
-5.98694444e-01 3.43739092e-01 2.78516352e-01 -4.58296090e-01
-1.31413174e+00 -3.11049491e-01 -3.62643301e-01 -6.82677984e-01
4.51112390e-02 4.91941094e-01 8.26525763e-02 5.27002573e-01
7.50221431e-01 4.07379091e-01 -1.82409421e-01 -2.33943000e-01
1.04099202e+00 2.06650376e-01 5.43922842e-01 -7.07827091e-01
8.43325913e-01 8.46378207e-01 -1.28124550e-01 -7.43433714e-01
-8.39493632e-01 -1.34556321e-02 -6.45542979e-01 -3.92182857e-01
1.16975856e+00 -7.09903777e-01 -6.54825687e-01 5.48869252e-01
-1.09839153e+00 -7.04589069e-01 -7.51961827e-01 2.03470275e-01
-7.60612667e-01 6.36666175e-03 -7.13930488e-01 -1.00454175e+00
-1.30824875e-02 -1.19131804e+00 1.00310254e+00 3.56809616e-01
-1.28384486e-01 -1.59534895e+00 2.39510104e-01 1.54772341e-01
5.90991914e-01 1.10015400e-01 8.61542106e-01 -1.66787729e-01
-9.98097420e-01 -5.99345900e-02 7.36030787e-02 4.11300391e-01
-4.06262577e-01 4.43251401e-01 -8.68576229e-01 -2.05759034e-01
2.54057255e-02 -1.47369727e-01 1.16382289e+00 9.54040945e-01
6.31088972e-01 -3.55823994e-01 -2.24213228e-01 7.83425689e-01
1.47593534e+00 -1.08527228e-01 5.57938874e-01 -8.55170712e-02
6.09753251e-01 2.94470340e-01 -2.97619492e-01 4.48844492e-01
2.37589195e-01 1.66074052e-01 4.42999691e-01 -6.77639432e-03
-4.14947897e-01 -7.19906628e-01 3.42254639e-01 9.36496377e-01
-1.29618391e-01 -5.98831952e-01 -5.50800443e-01 6.63092434e-01
-1.79596817e+00 -1.13075340e+00 -1.18533984e-01 1.83639717e+00
7.11598456e-01 7.86817924e-04 -1.41412944e-01 -1.89716518e-01
4.62879866e-01 4.48528022e-01 -6.05365992e-01 -1.74051210e-01
-2.74986148e-01 1.63754568e-01 2.35655874e-01 1.04055703e+00
-8.94568503e-01 9.63335454e-01 7.65199184e+00 9.76021767e-01
-9.12112176e-01 3.49503547e-01 5.59244394e-01 -4.56515640e-01
-9.99947548e-01 -3.52995060e-02 -7.63075411e-01 2.51978248e-01
8.60144496e-01 -1.72933862e-01 7.60518968e-01 8.29471469e-01
1.23635039e-01 -1.60965025e-01 -1.11730766e+00 6.99273288e-01
2.19553173e-01 -1.57879794e+00 2.73237348e-01 3.91973704e-01
1.25789213e+00 1.79996371e-01 6.74424231e-01 1.93156332e-01
1.03419530e+00 -1.19092560e+00 1.02479756e+00 9.13416624e-01
5.40390790e-01 -4.15221602e-01 2.84422904e-01 4.95562673e-01
-5.57675600e-01 5.03470637e-02 -3.36657852e-01 4.39824611e-02
5.61003268e-01 7.58407593e-01 -3.53877842e-01 1.30487338e-01
4.74550307e-01 6.29622579e-01 -8.21533874e-02 7.01607287e-01
-6.93557024e-01 7.11209118e-01 -2.39747211e-01 -3.24140564e-02
4.88953769e-01 -3.27765077e-01 8.34322870e-01 1.00383759e+00
3.97303522e-01 1.82427734e-01 -1.75874650e-01 1.61751616e+00
-1.71279848e-01 -2.66356081e-01 -4.79253918e-01 -1.52891189e-01
9.79759246e-02 1.07687485e+00 -6.14580333e-01 -4.05233443e-01
-9.77162868e-02 1.16182709e+00 2.44947195e-01 9.61177170e-01
-8.58478963e-01 -2.86109280e-02 7.64910221e-01 -1.58271924e-01
4.35158402e-01 -2.20480770e-01 -2.99721718e-01 -1.40093768e+00
-4.50468391e-01 -4.17984247e-01 1.60573229e-01 -1.09427226e+00
-1.39037025e+00 4.35604662e-01 8.11150856e-03 -6.69458508e-01
-4.12648618e-01 -4.90108877e-01 -7.88146138e-01 1.06511009e+00
-1.25978529e+00 -1.05490959e+00 -2.13255454e-02 5.88797688e-01
5.03741562e-01 3.41449268e-02 6.88357055e-01 1.45739183e-01
-5.31007230e-01 2.71057039e-01 5.40434659e-01 1.10658638e-01
2.06784755e-01 -1.50686002e+00 3.86445552e-01 1.04376316e+00
6.40955031e-01 6.17677212e-01 8.98426294e-01 -5.97390592e-01
-1.12154543e+00 -9.37330723e-01 7.47636497e-01 -8.99050891e-01
7.54256189e-01 -3.81217062e-01 -5.68566799e-01 9.55215931e-01
4.30979818e-01 -2.26058438e-01 4.16106433e-01 -2.80017480e-02
-1.72723889e-01 3.74266446e-01 -7.58393526e-01 8.60169232e-01
9.73604023e-01 -6.36440396e-01 -3.63124043e-01 3.35428029e-01
5.12349367e-01 -5.18261671e-01 -6.90318763e-01 -2.39062086e-01
4.05231535e-01 -1.10114586e+00 9.86488998e-01 -7.99810290e-01
8.85515094e-01 -4.08292077e-02 -3.27828616e-01 -1.55836463e+00
-5.50318182e-01 -7.97583818e-01 -3.15669596e-01 1.01042354e+00
6.16867900e-01 -2.49780864e-01 8.14538121e-01 5.91239512e-01
-6.82004318e-02 -7.74818361e-01 -8.43668938e-01 -5.45580089e-01
6.28168881e-01 -5.97536147e-01 2.23991960e-01 5.44054627e-01
-3.16046238e-01 3.25291187e-01 -4.21890378e-01 6.53908849e-02
8.74979854e-01 -1.67131007e-01 5.10158539e-01 -9.57252085e-01
-6.30813539e-01 -8.26271474e-01 -5.15646255e-03 -1.55075550e+00
-5.72067089e-02 -1.11411643e+00 9.22033489e-02 -1.90255034e+00
3.72501850e-01 -1.39143094e-01 -1.79253355e-01 9.68773961e-02
-1.33255482e-01 2.02178165e-01 4.72504914e-01 3.52455527e-01
-5.40891886e-01 7.52457738e-01 1.53299463e+00 -3.57539386e-01
-1.10166252e-01 -4.78428975e-02 -8.28760445e-01 5.85349917e-01
3.77399951e-01 -6.07218206e-01 -7.76793659e-01 -7.83671856e-01
6.22685969e-01 1.45920575e-01 5.19598186e-01 -5.13557255e-01
3.61745387e-01 -3.26748759e-01 4.80377287e-01 -3.98571283e-01
3.32644850e-01 -2.68621892e-01 1.73404545e-01 1.32461518e-01
-5.91216505e-01 -2.77009219e-01 -1.72435641e-01 8.41908455e-01
-1.30986571e-02 -3.36493224e-01 7.91049600e-01 -2.28043273e-01
-5.40800035e-01 4.46767569e-01 -7.84236610e-01 3.54954511e-01
9.05419409e-01 4.82405759e-02 -2.87800610e-01 -8.84696543e-01
-1.22437763e+00 2.69031376e-02 2.78139174e-01 2.41969794e-01
5.48894703e-01 -1.31184828e+00 -7.88344324e-01 2.26039425e-01
-3.49427283e-01 -7.14105442e-02 6.02309823e-01 7.10466921e-01
-1.94062993e-01 2.86458224e-01 1.58979237e-01 -5.73183060e-01
-6.40237153e-01 4.37957555e-01 5.70113003e-01 -3.33045602e-01
-4.64636803e-01 1.38106334e+00 5.18283844e-01 -1.55085474e-01
1.04884572e-01 -1.99729726e-01 2.38129154e-01 -1.95901264e-02
3.17428976e-01 3.39085519e-01 -2.89305091e-01 -5.40178001e-01
9.16252658e-02 3.27207357e-01 7.71082193e-02 -8.41890097e-01
1.13673222e+00 -2.95127630e-01 1.48112878e-01 4.96284962e-01
1.09598494e+00 -7.82751292e-02 -1.78096914e+00 -1.84633523e-01
-6.69895172e-01 -8.99389163e-02 3.73805642e-01 -9.06150699e-01
-1.25643587e+00 1.28614640e+00 2.78016835e-01 3.28980386e-01
7.66697228e-01 4.12380338e-01 6.31755352e-01 8.33389014e-02
-6.08359650e-02 -1.16169810e+00 3.29006851e-01 4.61687624e-01
9.00093317e-01 -9.78475988e-01 -2.47091725e-01 2.42390558e-01
-8.40912104e-01 8.02492499e-01 1.87616527e-01 -1.38658926e-01
1.04892123e+00 2.86607981e-01 1.01414360e-01 -4.08268660e-01
-9.91102755e-01 -1.91777617e-01 3.85228574e-01 8.56559217e-01
2.57885844e-01 1.88064605e-01 3.01643074e-01 3.81545007e-01
-3.18494916e-01 -1.67256504e-01 5.46518207e-01 5.94135582e-01
-3.81919533e-01 -8.07073891e-01 -1.12045489e-01 4.66205090e-01
-3.75309616e-01 -3.91527772e-01 -1.66231394e-01 5.17496347e-01
5.05301356e-02 9.00095880e-01 2.10229382e-01 -1.56486377e-01
-1.81860924e-02 2.00766727e-01 6.92798555e-01 -5.23662627e-01
-9.17657465e-02 2.30312854e-01 -5.98097742e-01 -3.10091496e-01
-3.71493042e-01 -8.63930583e-01 -1.02838612e+00 -5.71528316e-01
-2.10533425e-01 4.32984233e-02 7.12999880e-01 1.05253923e+00
1.21343024e-01 6.21187150e-01 1.72319099e-01 -9.34173107e-01
-6.60579920e-01 -9.07700717e-01 -8.41795921e-01 2.39135936e-01
5.82153320e-01 -4.27551597e-01 -5.43256700e-01 3.07348281e-01] | [11.346065521240234, -0.11371827870607376] |
7e8e517b-004c-4174-a2fa-004282eb1072 | neural-network-models-for-paraphrase | 1806.04330 | null | http://arxiv.org/abs/1806.04330v2 | http://arxiv.org/pdf/1806.04330v2.pdf | Neural Network Models for Paraphrase Identification, Semantic Textual Similarity, Natural Language Inference, and Question Answering | In this paper, we analyze several neural network designs (and their
variations) for sentence pair modeling and compare their performance
extensively across eight datasets, including paraphrase identification,
semantic textual similarity, natural language inference, and question answering
tasks. Although most of these models have claimed state-of-the-art performance,
the original papers often reported on only one or two selected datasets. We
provide a systematic study and show that (i) encoding contextual information by
LSTM and inter-sentence interactions are critical, (ii) Tree-LSTM does not help
as much as previously claimed but surprisingly improves performance on Twitter
datasets, (iii) the Enhanced Sequential Inference Model is the best so far for
larger datasets, while the Pairwise Word Interaction Model achieves the best
performance when less data is available. We release our implementations as an
open-source toolkit. | ['Wuwei Lan', 'Wei Xu'] | 2018-06-12 | neural-network-models-for-paraphrase-1 | https://aclanthology.org/C18-1328 | https://aclanthology.org/C18-1328.pdf | coling-2018-8 | ['sentence-pair-modeling'] | ['natural-language-processing'] | [ 2.60293454e-01 -2.50324588e-02 -3.86610061e-01 -5.34086585e-01
-9.30138290e-01 -4.80255216e-01 7.16291368e-01 5.72168231e-01
-5.35536468e-01 7.83012033e-01 6.44537151e-01 -6.34830892e-01
-3.05168808e-01 -5.05652905e-01 -7.37191260e-01 -9.85392630e-02
1.02318965e-01 5.86921573e-01 8.31118226e-02 -4.10774410e-01
5.69474697e-01 -1.23315275e-01 -1.38518214e+00 6.49997592e-01
8.87709141e-01 8.36555958e-01 6.17922954e-02 7.01578856e-01
-6.35998547e-01 1.32954943e+00 -4.35956270e-01 -7.56686032e-01
-1.44315362e-01 -4.88783717e-01 -1.51813567e+00 -8.50898266e-01
8.77526045e-01 -1.82123184e-01 -3.98878753e-01 8.32506001e-01
6.35382235e-01 1.80519685e-01 5.72635770e-01 -9.59183991e-01
-1.07326925e+00 1.09412122e+00 -1.82712451e-01 3.54761243e-01
7.91057289e-01 -1.84552372e-01 1.31891572e+00 -8.31350148e-01
4.99835074e-01 1.09191501e+00 1.10815108e+00 3.79628271e-01
-1.21221185e+00 -4.72686678e-01 -1.10257208e-01 6.19378328e-01
-9.73515034e-01 -4.40759003e-01 6.67444408e-01 -1.09449834e-01
1.60531437e+00 4.13619101e-01 2.55972952e-01 1.67473161e+00
3.01649392e-01 8.49339008e-01 9.14591074e-01 -4.88740265e-01
-1.49153411e-01 1.35214776e-01 7.11860657e-01 4.76207048e-01
-1.17197149e-01 -3.59227628e-01 -6.23201907e-01 -4.51054335e-01
1.32124022e-01 -1.00154765e-02 -3.36141721e-03 1.31174654e-01
-1.22288024e+00 8.65657210e-01 4.48671997e-01 8.36924314e-01
-1.49114177e-01 8.94029513e-02 8.84384871e-01 8.49905133e-01
7.61226594e-01 6.71338499e-01 -3.87077004e-01 -3.49329501e-01
-1.08495021e+00 4.37233508e-01 1.14524806e+00 8.17142427e-01
4.30450112e-01 -3.06761950e-01 -4.30100381e-01 1.28745651e+00
-1.59711987e-01 9.55691636e-02 7.66648054e-01 -8.97174001e-01
6.90454960e-01 4.56991583e-01 -1.25769883e-01 -1.27560914e+00
-4.45885241e-01 -3.22806865e-01 -9.83728588e-01 -8.69638443e-01
3.94574195e-01 -2.23516431e-02 -2.10168406e-01 1.71315062e+00
-2.89870888e-01 1.44976109e-01 -3.25969327e-03 3.98311675e-01
1.50959110e+00 4.35339540e-01 -3.78406234e-02 1.66924186e-02
1.20743442e+00 -1.15980709e+00 -7.57060766e-01 -5.71669042e-01
9.07519698e-01 -7.57330656e-01 9.32916462e-01 -4.07171547e-02
-1.40804994e+00 -6.18612111e-01 -8.60955477e-01 -4.71342653e-01
-6.08505547e-01 -2.18670338e-01 4.98759538e-01 1.65526420e-01
-1.32484293e+00 9.63635325e-01 -3.24083656e-01 -8.30724299e-01
7.78570166e-03 1.78642496e-01 -2.47717261e-01 4.22715805e-02
-1.84044707e+00 1.40381384e+00 1.35400206e-01 -6.94765896e-02
-8.62640068e-02 -7.31497467e-01 -8.87891233e-01 4.79365289e-02
9.61333439e-02 -1.18174720e+00 1.41292834e+00 -7.93487191e-01
-1.24549663e+00 1.04188633e+00 -5.00983953e-01 -9.57115293e-01
3.80852550e-01 -1.28610536e-01 -2.39642158e-01 -3.07693053e-03
6.57003075e-02 5.68826973e-01 3.30425292e-01 -8.42933655e-01
-1.04676709e-01 -2.28835493e-01 1.23702034e-01 2.40399659e-01
-4.40228194e-01 3.71805251e-01 -3.13137509e-02 -6.49119914e-01
-2.76722133e-01 -5.09809613e-01 -1.00702085e-01 -3.42528015e-01
-3.95981550e-01 -6.86542392e-01 5.41706324e-01 -1.02204514e+00
1.54587007e+00 -1.59647214e+00 9.21121836e-02 -4.39158410e-01
1.02764063e-01 3.09715420e-01 -3.37113380e-01 1.09177446e+00
-2.78415084e-01 2.44533494e-01 -3.61758322e-01 -8.22975993e-01
1.28931254e-01 1.52709961e-01 -4.36028928e-01 2.81967279e-02
-6.27695173e-02 1.43996918e+00 -9.33807194e-01 -6.57323062e-01
1.59936920e-01 1.39373437e-01 -2.25277841e-01 8.11140761e-02
-1.21783853e-01 2.68613081e-02 -1.62099317e-01 2.75084406e-01
4.29362267e-01 -5.03470838e-01 2.62273759e-01 -1.82817027e-01
2.09624082e-01 9.07749116e-01 -4.33390796e-01 1.84534442e+00
-6.39626205e-01 1.03877890e+00 -1.26928926e-01 -1.18386781e+00
8.23338032e-01 3.31582844e-01 2.58812547e-01 -9.93633568e-01
-9.11093652e-02 1.10766448e-01 1.85661595e-02 -1.00606191e+00
7.09648907e-01 -6.28425404e-02 1.09031622e-04 7.21000969e-01
-1.04853958e-01 3.07056010e-02 7.87425935e-02 3.98843288e-01
1.27869785e+00 -1.85222015e-01 3.52480114e-01 -2.15213060e-01
5.14349699e-01 -2.98112899e-01 -5.30452328e-03 1.10883093e+00
-1.79778576e-01 4.77683961e-01 4.25295889e-01 -2.63734311e-01
-1.07755315e+00 -8.14454675e-01 -8.38577524e-02 1.22830331e+00
-1.07091390e-01 -4.42130059e-01 -8.44628632e-01 -5.58065593e-01
4.52084169e-02 1.11543369e+00 -7.02374339e-01 -1.26411840e-01
-6.23180866e-01 -3.29741627e-01 8.29918921e-01 7.58174598e-01
6.41976595e-01 -1.10635090e+00 -1.16976470e-01 4.09884751e-02
-6.96035564e-01 -1.16346073e+00 -5.04732549e-01 7.81885013e-02
-7.89865494e-01 -7.94882536e-01 -3.62459749e-01 -9.76569653e-01
4.85589355e-03 2.88910091e-01 1.58661854e+00 1.27524897e-01
8.68524015e-02 2.56193280e-01 -4.27967161e-01 -3.59058119e-02
-4.48086381e-01 2.78153360e-01 -1.37546510e-02 -4.05232072e-01
6.70466185e-01 -6.68256462e-01 -3.43082964e-01 5.87932207e-03
-6.60013974e-01 -1.69890210e-01 3.42078626e-01 1.03870642e+00
-1.78775281e-01 -6.37213647e-01 9.43762839e-01 -8.05186033e-01
1.28298557e+00 -6.93225741e-01 3.20413619e-01 4.43887502e-01
-4.84579653e-01 7.00866953e-02 7.43008971e-01 -1.51777714e-01
-8.68791461e-01 -6.83547974e-01 -4.03119624e-01 -1.85909614e-01
-2.76305407e-01 9.05711710e-01 5.61602354e-01 7.82949701e-02
6.59741044e-01 3.91732603e-01 1.12332940e-01 -5.89054644e-01
2.56433457e-01 9.18761671e-01 3.78222615e-01 -4.21830922e-01
2.94947773e-01 8.12485814e-02 -3.73882264e-01 -7.98119128e-01
-1.09407043e+00 -3.81244600e-01 -6.28325760e-01 8.28995109e-02
8.20198953e-01 -5.07682681e-01 -8.60172212e-01 3.07120681e-01
-1.44792891e+00 -2.04718292e-01 1.04292855e-01 1.36705965e-01
-2.61810035e-01 9.32537138e-01 -8.61278653e-01 -7.40312576e-01
-7.50718713e-01 -7.96937108e-01 8.63798976e-01 -5.54531105e-02
-8.18303704e-01 -1.57372868e+00 1.87161192e-01 7.09910631e-01
7.95922399e-01 -1.23928271e-01 1.15322781e+00 -1.15362489e+00
4.79646921e-02 -2.00755358e-01 -3.88993561e-01 2.24752590e-01
9.75917131e-02 -7.32310489e-02 -7.70321310e-01 -4.79106680e-02
1.37549371e-01 -6.32123411e-01 1.01826131e+00 3.75582665e-01
1.42349839e+00 -3.63545865e-01 -3.05521071e-01 1.11167692e-01
9.36732590e-01 -1.49692059e-01 3.69384617e-01 2.25403637e-01
4.54969257e-01 8.53251100e-01 1.06415950e-01 8.70619193e-02
9.40248907e-01 7.31223762e-01 1.98924303e-01 1.00049525e-01
1.49866641e-02 -4.04756397e-01 4.07434970e-01 1.19752109e+00
4.45081592e-01 -2.70685554e-01 -9.39906418e-01 5.50215662e-01
-2.26667285e+00 -1.34183955e+00 -4.48569834e-01 2.01539278e+00
9.39719439e-01 1.35272205e-01 1.74384654e-01 -1.72800943e-01
6.21928513e-01 4.69506532e-01 -4.41005290e-01 -8.87549341e-01
-2.81446666e-01 2.90457875e-01 1.77792490e-01 6.81961536e-01
-8.45081568e-01 8.08510303e-01 7.52567482e+00 8.84810567e-01
-6.58250451e-01 1.73067346e-01 5.48694491e-01 -9.86392424e-02
-4.57849026e-01 -1.10292278e-01 -4.43076313e-01 5.95082223e-01
1.29246652e+00 -1.55873805e-01 2.58572578e-01 5.82589388e-01
1.45066291e-01 -1.63423896e-01 -1.32495427e+00 8.96384716e-01
3.54890138e-01 -1.45393789e+00 7.06257820e-02 -4.19424355e-01
3.34467709e-01 2.66969562e-01 -2.20100850e-01 4.52643543e-01
2.61009932e-01 -1.20823002e+00 2.40706429e-01 7.39747465e-01
2.70986944e-01 -6.26646757e-01 8.58607411e-01 5.56533575e-01
-8.54759216e-01 -6.59701228e-02 -2.12452590e-01 -4.78792161e-01
1.48239091e-01 4.86811459e-01 -4.82238501e-01 7.67507732e-01
7.23395407e-01 1.12786937e+00 -1.01788688e+00 7.59228170e-01
-9.81460884e-02 7.85099685e-01 -1.31672755e-01 -6.03484035e-01
4.48027760e-01 -1.96574688e-01 4.95311797e-01 1.51644623e+00
-2.32945070e-01 -3.04395705e-01 -7.79931024e-02 8.69658530e-01
-1.37934044e-01 1.07335776e-01 -7.93034196e-01 -4.00843844e-02
6.72264576e-01 1.01832604e+00 -1.18448697e-01 -4.86533165e-01
-4.70136315e-01 9.45064604e-01 7.44254112e-01 3.03221136e-01
-7.96254694e-01 -5.23636878e-01 5.67285180e-01 -2.45667174e-01
-8.51513818e-03 -1.51034132e-01 -5.83579063e-01 -1.13931429e+00
9.53136981e-02 -8.33681822e-01 5.57756782e-01 -8.65096390e-01
-2.00646734e+00 5.72019339e-01 8.60128254e-02 -6.85639799e-01
-7.10332692e-01 -4.87283319e-01 -9.61988807e-01 7.36779392e-01
-1.15495479e+00 -1.07973742e+00 -4.89620417e-02 2.08762854e-01
6.57411754e-01 -1.28653273e-01 8.93932939e-01 3.72387588e-01
-5.90154290e-01 8.30193579e-01 4.11141783e-01 1.96412206e-01
9.01664913e-01 -1.14949310e+00 8.68351460e-01 2.63089806e-01
1.77530810e-01 9.50387597e-01 7.92704344e-01 -4.05507892e-01
-1.16914499e+00 -5.75026095e-01 1.67435932e+00 -7.10619986e-01
1.06539130e+00 -4.60504591e-01 -1.09988415e+00 6.74199998e-01
1.07259083e+00 -7.27441669e-01 7.89982140e-01 4.32654560e-01
-4.38852638e-01 1.31138116e-01 -9.05660629e-01 7.03678012e-01
1.08213997e+00 -1.10259581e+00 -1.01748765e+00 6.24816239e-01
1.01931357e+00 -1.58097103e-01 -9.50772285e-01 3.89823735e-01
4.72118437e-01 -1.19984889e+00 9.43320155e-01 -1.09335923e+00
1.04498017e+00 3.55273217e-01 -1.45787984e-01 -1.16812313e+00
-2.24354595e-01 -3.99446189e-01 -1.82933524e-01 1.30704594e+00
5.72363615e-01 -8.54420662e-01 5.88834524e-01 6.74662232e-01
-7.97600448e-02 -1.19200099e+00 -8.53799820e-01 -6.77035332e-01
4.37256098e-01 -4.29164380e-01 2.97130466e-01 1.36521947e+00
4.67261642e-01 8.42793524e-01 -3.46827328e-01 -6.93878174e-01
3.36441278e-01 2.83613980e-01 4.55898523e-01 -7.62895107e-01
-3.45959485e-01 -8.33225846e-01 -8.43705088e-02 -1.19954872e+00
7.43992686e-01 -1.19228947e+00 -4.02773112e-01 -1.92486882e+00
6.31422222e-01 2.09207833e-01 4.23065983e-02 4.47070360e-01
-4.00029093e-01 1.06091495e-03 -1.88924581e-01 -1.16290055e-01
-6.09506607e-01 5.35447955e-01 6.87699437e-01 -2.81750798e-01
1.88805059e-01 1.93574615e-02 -7.36225009e-01 4.99302238e-01
7.21383393e-01 -1.73336774e-01 -2.15932891e-01 -8.18210363e-01
4.01878774e-01 1.26546443e-01 6.10601366e-01 -5.46863735e-01
5.59051573e-01 3.07090700e-01 1.17065027e-01 -7.29348421e-01
3.78876001e-01 -4.34905648e-01 -1.31665871e-01 4.25167352e-01
-1.06498861e+00 3.98417830e-01 1.96583241e-01 4.79816288e-01
-3.52140367e-01 -6.46041334e-01 5.13832808e-01 -2.61006027e-01
-1.98254064e-01 -5.61797097e-02 -4.21116024e-01 3.90320867e-01
3.44470888e-01 -1.65611967e-01 -4.58480656e-01 -9.91480589e-01
-4.22299743e-01 4.62706447e-01 1.11833215e-01 8.06074619e-01
4.41324472e-01 -1.21486568e+00 -8.75246644e-01 -3.26403499e-01
8.55028778e-02 -5.31869352e-01 2.68655390e-01 1.01199389e+00
-6.07754886e-02 7.01660991e-01 3.69519413e-01 -5.31061590e-01
-1.35746336e+00 4.46709812e-01 4.51724678e-01 -5.47122717e-01
-2.34015763e-01 9.54906583e-01 -3.15230221e-01 -8.59821737e-01
2.67552018e-01 -2.31345102e-01 -2.44982198e-01 2.99773842e-01
3.12850416e-01 3.78548622e-01 2.38405153e-01 -3.79806697e-01
-3.87128085e-01 4.49887365e-01 -3.64041805e-01 2.31993556e-01
1.22426796e+00 -1.67256057e-01 -5.72492838e-01 7.26960123e-01
1.67403173e+00 -4.89587128e-01 -1.75271124e-01 -6.30929291e-01
1.82319790e-01 -1.14228286e-01 -1.30279765e-01 -6.50642216e-01
-3.65082920e-01 9.91443396e-01 -3.04363500e-02 6.11694634e-01
6.72834337e-01 -1.06381282e-01 1.13883412e+00 7.13871002e-01
-2.77225487e-02 -9.97348905e-01 1.61838308e-01 1.13056874e+00
8.93723547e-01 -1.22356820e+00 -8.63763168e-02 1.44886643e-01
-6.75522208e-01 1.02405608e+00 5.40373027e-01 -1.34547120e-02
5.53312480e-01 7.86640272e-02 -3.70705605e-01 -1.99326575e-01
-1.10151398e+00 3.58193405e-02 3.73924047e-01 1.02238759e-01
8.71098757e-01 -3.66952121e-01 -4.46099401e-01 5.02738774e-01
-3.84278774e-01 -1.98909685e-01 1.11342452e-01 5.31722367e-01
-1.40112817e-01 -1.06709552e+00 1.68934047e-01 8.34545910e-01
-3.91074598e-01 -6.37064755e-01 -1.12624228e+00 5.08360386e-01
-4.96876657e-01 1.20376682e+00 1.81481659e-01 -5.72540820e-01
2.01781958e-01 4.06022280e-01 5.67680240e-01 -3.35693419e-01
-1.31270695e+00 -8.43331993e-01 7.06316173e-01 -5.23490071e-01
-4.21926171e-01 -6.01901054e-01 -7.90245771e-01 -6.01850271e-01
-2.10964784e-01 1.46952763e-01 5.30553758e-01 1.36708426e+00
7.10608184e-01 2.48101443e-01 5.21622062e-01 -6.05871320e-01
-8.02831590e-01 -1.46076298e+00 -3.44290547e-02 3.50779533e-01
2.37444893e-01 -1.35091931e-01 -3.75425011e-01 -3.93777102e-01] | [11.191493034362793, 8.615317344665527] |
263e9f3c-db83-499d-b13d-afe052a9405d | in-distribution-interpretability-for | 2007.00758 | null | https://arxiv.org/abs/2007.00758v2 | https://arxiv.org/pdf/2007.00758v2.pdf | In-Distribution Interpretability for Challenging Modalities | It is widely recognized that the predictions of deep neural networks are difficult to parse relative to simpler approaches. However, the development of methods to investigate the mode of operation of such models has advanced rapidly in the past few years. Recent work introduced an intuitive framework which utilizes generative models to improve on the meaningfulness of such explanations. In this work, we display the flexibility of this method to interpret diverse and challenging modalities: music and physical simulations of urban environments. | ['Cosmas Heiß', 'Ron Levie', 'Joan Bruna', 'Gitta Kutyniok', 'Cinjon Resnick'] | 2020-07-01 | null | null | null | null | ['physical-simulations'] | ['miscellaneous'] | [ 1.68857083e-01 4.63568151e-01 1.74907491e-01 -5.21386325e-01
1.54507622e-01 -4.20186341e-01 1.15634894e+00 -3.22105885e-01
1.52105480e-01 6.12335145e-01 3.59337896e-01 -6.71640575e-01
-4.97262955e-01 -7.65919864e-01 -4.56489325e-01 -4.60728943e-01
-1.33764222e-01 1.57554194e-01 -9.45866033e-02 -3.31073761e-01
2.62201339e-01 6.74096584e-01 -1.77765822e+00 3.44761759e-01
6.90145612e-01 6.65722966e-01 1.49188712e-02 6.07897460e-01
-1.17968298e-01 6.95854366e-01 -3.69958252e-01 -4.07043964e-01
1.07486993e-02 -5.02504170e-01 -5.58325708e-01 -2.51504183e-01
1.32277101e-01 -2.09521011e-01 -4.68837947e-01 8.52646708e-01
3.36162329e-01 1.91812843e-01 6.29228354e-01 -1.37244296e+00
-6.51055813e-01 6.24151170e-01 1.30003408e-01 2.13726133e-01
1.20061085e-01 6.34333715e-02 1.12903142e+00 -6.35959864e-01
4.49558675e-01 1.33186495e+00 7.76271999e-01 5.85876942e-01
-1.22508991e+00 -6.77141368e-01 1.30650163e-01 2.09045723e-01
-1.12692034e+00 -5.47813594e-01 8.70813608e-01 -3.50476831e-01
1.06458855e+00 2.65305430e-01 9.16556954e-01 1.27342689e+00
2.82217473e-01 6.39107108e-01 1.03395665e+00 -4.12557840e-01
2.26626113e-01 -6.29629269e-02 2.05979813e-02 5.98311067e-01
2.86622316e-01 5.32328188e-01 -7.35428333e-01 -9.74638388e-02
1.10772741e+00 -1.52614877e-01 -7.21846987e-03 -2.89188027e-01
-8.99276376e-01 9.50258672e-01 5.06892264e-01 3.72683972e-01
-1.48396507e-01 4.81412232e-01 1.00775592e-01 -1.14940673e-01
4.38311458e-01 5.81100702e-01 -3.07650745e-01 -3.46944422e-01
-9.17065918e-01 5.31509519e-01 7.39174068e-01 4.66588736e-01
3.24363858e-01 3.99882525e-01 4.74992186e-01 5.96740901e-01
7.45116770e-01 1.75200418e-01 2.42030412e-01 -1.17614353e+00
7.82007724e-02 2.47658163e-01 1.40198907e-02 -1.20817530e+00
-4.05277580e-01 -5.41248679e-01 -9.26704586e-01 3.91309112e-01
3.44834417e-01 -7.01870173e-02 -8.11895192e-01 1.83360088e+00
-9.12022311e-03 2.60310799e-01 1.14698991e-01 8.56621385e-01
8.67246807e-01 5.13343453e-01 1.78286761e-01 2.18838379e-01
4.98376608e-01 -7.36755431e-01 -6.38489544e-01 -2.52093822e-01
3.03088516e-01 -6.07188582e-01 8.15751970e-01 4.13309425e-01
-1.09091914e+00 -6.76549315e-01 -1.01157868e+00 1.29691988e-01
-2.92742848e-01 -2.14726195e-01 1.28654635e+00 7.63376832e-01
-1.02958727e+00 1.05631459e+00 -1.13105083e+00 -5.80899537e-01
2.83838362e-01 4.58714455e-01 -3.46574746e-02 3.59743863e-01
-1.14939320e+00 9.44225132e-01 3.89957339e-01 2.81263351e-01
-4.27210569e-01 -4.93501037e-01 -7.13499010e-01 1.57342419e-01
-1.76929116e-01 -9.49874938e-01 1.28277493e+00 -1.24674904e+00
-1.61341262e+00 5.89353979e-01 -1.81798980e-01 -3.46991777e-01
2.75001913e-01 -3.01039249e-01 -4.53704953e-01 -7.43033439e-02
-3.72320056e-01 7.67778397e-01 3.43973875e-01 -1.28820026e+00
-2.12529555e-01 -2.22491741e-01 2.89767027e-01 -2.32765786e-02
1.41583607e-01 -1.29795447e-01 -5.29941693e-02 -6.67136014e-01
3.33477914e-01 -1.05507398e+00 -3.50550622e-01 9.36812907e-02
-4.98950571e-01 1.26752049e-01 4.37994838e-01 -2.32356593e-01
9.90603805e-01 -2.19502878e+00 5.57327643e-02 2.35726789e-01
2.03573227e-01 7.48656020e-02 4.16994728e-02 5.40888608e-01
-3.55151623e-01 3.85459512e-01 -1.23945981e-01 -4.25495654e-01
2.33004734e-01 7.57780671e-01 -8.42251420e-01 9.29666832e-02
2.13425368e-01 9.31639969e-01 -5.59819400e-01 -1.47755012e-01
3.06738198e-01 7.63642490e-01 -6.57248199e-01 7.79633829e-03
-1.74770042e-01 4.40297574e-01 -4.11675334e-01 4.62191701e-01
2.53129303e-01 -2.96906769e-01 1.56634286e-01 2.53563542e-02
-3.24251175e-01 6.94678485e-01 -1.12903309e+00 1.27749372e+00
-1.57242082e-02 1.15035069e+00 -1.98959231e-01 -1.06266427e+00
7.50232041e-01 3.71123970e-01 2.87139326e-01 -5.69474101e-01
1.83996841e-01 9.78353769e-02 3.50174785e-01 -4.22879636e-01
5.18998504e-01 -5.73017299e-01 1.16111338e-01 5.62106490e-01
-9.58031267e-02 -1.28928766e-01 -2.04099074e-01 -6.83545461e-03
7.15944648e-01 5.93603194e-01 3.98453325e-01 -3.57612632e-02
1.82268284e-02 7.84140229e-02 2.84880102e-01 8.75148118e-01
-3.00724264e-02 5.94416916e-01 2.78524637e-01 -7.71039903e-01
-1.06613994e+00 -1.37163365e+00 -1.94892526e-01 9.46266234e-01
1.57615677e-01 -5.10162711e-01 -5.52356362e-01 -9.62873548e-02
-6.01009279e-02 1.08134341e+00 -7.49919057e-01 -1.55456275e-01
-3.65213901e-01 -8.16606522e-01 9.16265488e-01 7.85631180e-01
2.82454997e-01 -1.16071498e+00 -8.21073472e-01 1.28211528e-01
-4.56612781e-02 -9.59915817e-01 6.67721450e-01 4.45987880e-01
-1.11407399e+00 -7.78510213e-01 -8.81583914e-02 -1.94796190e-01
3.01318139e-01 1.77617610e-01 1.24449372e+00 2.70340413e-01
-9.00307149e-02 3.21821719e-01 -1.85738087e-01 -5.64369559e-01
-7.04186976e-01 -1.25340104e-01 8.22162330e-02 -5.26513159e-01
1.57844186e-01 -1.02792847e+00 -4.18592542e-01 9.79496017e-02
-1.12704015e+00 5.77232182e-01 6.10907257e-01 6.17201626e-01
2.74951637e-01 -1.96163967e-01 5.76026201e-01 -5.75625718e-01
7.17278957e-01 -4.96577829e-01 -1.57916397e-01 -1.10411063e-01
-6.71489775e-01 3.62166613e-01 5.71572483e-01 -3.79501849e-01
-1.02285314e+00 -1.73791498e-01 -4.50615674e-01 -1.80562720e-01
-6.43617272e-01 7.70025730e-01 6.09574430e-02 -6.91257641e-02
4.65987235e-01 6.35828003e-02 -1.68702722e-01 -3.60979825e-01
4.06724304e-01 2.66610682e-01 6.81492746e-01 -6.23845577e-01
8.52187753e-01 4.06054795e-01 1.88654840e-01 -8.48120391e-01
-7.55917847e-01 2.27167010e-01 -5.23605764e-01 -4.84407723e-01
5.17128110e-01 -4.05169606e-01 -4.55413163e-01 1.22079536e-01
-1.13501358e+00 -4.35119539e-01 -2.02409074e-01 5.65982759e-01
-7.02923059e-01 1.63196653e-01 -4.13300633e-01 -8.72162282e-01
4.32156548e-02 -8.95666540e-01 8.26546371e-01 1.83383614e-01
-7.77582109e-01 -1.33278644e+00 8.55462998e-02 1.75409153e-01
4.67767298e-01 3.37219119e-01 1.10450661e+00 -5.41188419e-01
-6.42027140e-01 1.91862941e-01 -1.26970023e-01 -3.98327447e-02
-2.11549610e-01 5.36575496e-01 -1.43340302e+00 4.40632939e-01
-1.27242610e-01 -1.28213033e-01 8.31189871e-01 5.51151633e-01
1.16081536e+00 5.99636734e-02 -3.37712795e-01 6.23721063e-01
1.22311580e+00 5.20254001e-02 8.12970638e-01 4.02248472e-01
2.46457845e-01 5.50979793e-01 2.48185486e-01 2.96129465e-01
2.54893541e-01 5.64744174e-01 7.24140525e-01 -1.51849702e-01
8.70184973e-02 -3.30900311e-01 2.40103349e-01 7.06832051e-01
-5.42731941e-01 -2.30797082e-01 -1.05962408e+00 2.31290802e-01
-1.90255404e+00 -1.29696858e+00 -1.27592549e-01 1.67110980e+00
3.35428715e-01 4.18375432e-01 -2.15679452e-01 3.82727593e-01
3.16884488e-01 2.02397138e-01 -4.71240819e-01 -6.45100892e-01
-1.19317405e-01 2.12500691e-01 -8.58760849e-02 4.57302839e-01
-7.79150248e-01 8.15071106e-01 8.32401085e+00 4.23181027e-01
-1.19968593e+00 -2.29002118e-01 5.12593806e-01 1.41828716e-01
-6.25394881e-01 5.86451665e-02 -4.37815249e-01 5.54599985e-02
1.11655569e+00 1.80430740e-01 4.63384211e-01 7.49771595e-01
4.87805247e-01 -9.27100107e-02 -1.20717204e+00 6.39590442e-01
-5.20262085e-02 -1.51558471e+00 8.82690921e-02 2.25456461e-01
4.88937706e-01 1.47252709e-01 3.79170805e-01 2.50660509e-01
4.04253662e-01 -1.37140858e+00 9.96582925e-01 6.48030758e-01
2.51681089e-01 -5.63085496e-01 4.65293646e-01 4.49912190e-01
-8.79774451e-01 -8.47803131e-02 -1.85157031e-01 -9.15365994e-01
2.24055380e-01 3.79740000e-01 -7.68216014e-01 3.53309005e-01
6.73605442e-01 5.39959550e-01 -5.38171828e-01 1.06820905e+00
-3.14300507e-01 8.65371525e-01 -4.48121250e-01 6.82864264e-02
3.78459185e-01 -1.29290387e-01 5.48494816e-01 1.05178189e+00
5.28926671e-01 8.13188180e-02 -2.23532781e-01 1.30657971e+00
4.86161768e-01 -2.91961432e-01 -9.87993300e-01 -2.73529172e-01
1.87901661e-01 1.00959897e+00 -9.18973327e-01 -3.96602191e-02
-2.43928045e-01 5.65005898e-01 1.75239086e-01 4.16397095e-01
-1.10173190e+00 2.74931669e-01 8.30762506e-01 5.15561774e-02
1.60755068e-01 -6.87456250e-01 -7.36023784e-01 -9.53854620e-01
-3.55955094e-01 -5.89900136e-01 -1.17411092e-01 -1.33847225e+00
-9.97317791e-01 3.39562744e-01 2.06516355e-01 -9.46840703e-01
-5.33160031e-01 -7.54229367e-01 -8.39141369e-01 7.07974494e-01
-1.16741145e+00 -1.10245514e+00 -2.58653402e-01 1.21885084e-01
2.79090226e-01 -6.74863765e-03 1.26235151e+00 1.53483823e-02
-4.26600993e-01 1.37326032e-01 1.43659860e-01 -1.65520951e-01
3.62529308e-01 -1.05076146e+00 5.94793677e-01 6.14679992e-01
4.60886776e-01 8.84908736e-01 1.22783971e+00 -4.15804416e-01
-1.12431705e+00 -7.05565572e-01 7.93919265e-01 -2.76767761e-01
6.30988061e-01 -1.94371641e-01 -7.93521523e-01 7.22060084e-01
3.40726465e-01 -3.14729482e-01 1.05922711e+00 3.65815461e-01
-3.17420125e-01 3.01991284e-01 -6.64551258e-01 8.96653354e-01
8.72533381e-01 -4.24953014e-01 -8.35017145e-01 -2.07751811e-01
4.02988940e-01 -2.06019565e-01 -5.17708421e-01 4.37649697e-01
8.97370160e-01 -1.40564442e+00 9.48711753e-01 -7.21941352e-01
6.28972769e-01 -1.21546902e-01 -3.54000092e-01 -1.27872384e+00
-2.85393596e-01 -4.74190086e-01 -2.44797856e-01 8.21815908e-01
3.28164220e-01 -4.65755701e-01 7.03532815e-01 7.10920095e-01
-3.48707110e-01 -8.09495032e-01 -8.11221898e-01 -3.11785966e-01
1.69668630e-01 -9.38195288e-01 4.63731110e-01 7.67208457e-01
3.06872148e-02 1.99435934e-01 -8.86263847e-02 2.46014223e-01
3.35644007e-01 2.64091253e-01 7.21290767e-01 -1.77442002e+00
-4.53198314e-01 -6.92372918e-01 -5.34711421e-01 -1.07519543e+00
2.98093259e-01 -6.96880937e-01 -7.46118650e-02 -1.58638203e+00
1.47376955e-01 -2.72866011e-01 -1.52625874e-01 3.12803030e-01
1.08231038e-01 4.55126375e-01 2.50510037e-01 1.06704280e-01
-3.33719492e-01 4.50617671e-01 9.41547573e-01 1.64475307e-01
2.81131063e-02 -7.32840896e-02 -8.11099648e-01 1.30162215e+00
1.14742100e+00 -2.51289636e-01 -5.18348694e-01 -4.23384130e-01
6.44580841e-01 -1.05235912e-01 9.37279105e-01 -1.08158314e+00
-1.27226725e-01 -1.96750075e-01 7.75385976e-01 -5.09783447e-01
5.18376887e-01 -6.65907264e-01 4.74017859e-01 3.67810905e-01
-5.47497869e-01 6.77340850e-02 4.24686521e-01 5.89441121e-01
-1.53200701e-01 -1.07446201e-01 5.72227895e-01 -8.03552754e-03
-7.51855195e-01 -3.14703226e-01 -5.74390948e-01 -3.26994717e-01
5.56250811e-01 -4.12186176e-01 -8.47065672e-02 -7.51428366e-01
-9.34979737e-01 -5.00984788e-01 4.44888651e-01 4.95757490e-01
6.30464554e-01 -1.18704426e+00 -1.08830951e-01 4.54715602e-02
-3.31302136e-01 -3.27805668e-01 -6.48100153e-02 6.96664453e-01
-5.05640507e-01 6.11777306e-01 -4.63552147e-01 -4.19935942e-01
-1.12240124e+00 1.53792486e-01 5.86273432e-01 -4.06921469e-02
-7.10913062e-01 4.24458295e-01 2.67567545e-01 -3.71990651e-01
7.25708576e-03 -5.41745663e-01 -4.34140742e-01 -3.95115107e-01
2.71110713e-01 3.21817845e-01 -2.72604734e-01 -7.71085978e-01
-5.17411120e-02 3.21635634e-01 4.98395741e-01 -3.54419261e-01
1.49901450e+00 -2.23216727e-01 1.78854033e-01 8.84813070e-01
5.16328812e-01 -7.05172196e-02 -1.23151481e+00 2.55644470e-01
-1.42786220e-01 -1.30215973e-01 3.42127196e-02 -8.46866131e-01
-7.82391250e-01 1.21261954e+00 4.75643873e-01 5.15124381e-01
9.05059338e-01 1.52266840e-03 4.25471216e-01 4.23254251e-01
4.59317118e-02 -7.32145727e-01 -1.64615214e-01 7.36447036e-01
7.41891444e-01 -9.07962739e-01 -1.97210774e-01 -3.82015437e-01
-2.30128780e-01 1.50981140e+00 4.12362307e-01 -1.77101605e-02
6.83830023e-01 3.52481812e-01 -1.47350822e-02 -2.87410021e-01
-7.82746494e-01 -8.25962052e-02 3.59434009e-01 5.30914485e-01
7.30413616e-01 1.10756882e-01 1.53599739e-01 8.42116296e-01
-6.60363138e-01 4.26424630e-02 4.59519506e-01 6.35506451e-01
-5.44355750e-01 -9.84800398e-01 -3.90895307e-01 9.18805301e-02
-4.34127480e-01 -1.28173366e-01 -6.27621830e-01 1.02851760e+00
2.86754489e-01 9.55558419e-01 -1.70734860e-02 -3.48374695e-01
-6.86997920e-02 3.07412028e-01 5.04463851e-01 -4.06295657e-01
-4.16783720e-01 -2.19318410e-03 1.77587554e-01 -5.04772365e-01
-8.04314971e-01 -7.01831222e-01 -1.41688573e+00 -3.42552364e-01
-1.55418396e-01 -1.43505245e-01 6.40380383e-01 1.36574328e+00
1.91351831e-01 7.00174332e-01 -8.23517367e-02 -1.25950956e+00
-1.34228349e-01 -8.16060126e-01 -3.81498158e-01 2.88362265e-01
9.68624949e-02 -9.85126674e-01 -3.40928495e-01 2.20643148e-01] | [8.9204683303833, 5.612634181976318] |
376b75e5-0184-4299-a4de-747e8505d3d9 | extractive-adversarial-networks-high-recall | 1809.01499 | null | http://arxiv.org/abs/1809.01499v2 | http://arxiv.org/pdf/1809.01499v2.pdf | Extractive Adversarial Networks: High-Recall Explanations for Identifying Personal Attacks in Social Media Posts | We introduce an adversarial method for producing high-recall explanations of
neural text classifier decisions. Building on an existing architecture for
extractive explanations via hard attention, we add an adversarial layer which
scans the residual of the attention for remaining predictive signal. Motivated
by the important domain of detecting personal attacks in social media comments,
we additionally demonstrate the importance of manually setting a semantically
appropriate `default' behavior for the model by explicitly manipulating its
bias term. We develop a validation set of human-annotated personal attacks to
evaluate the impact of these changes. | ['Samuel Carton', 'Paul Resnick', 'Qiaozhu Mei'] | 2018-09-01 | extractive-adversarial-networks-high-recall-1 | https://aclanthology.org/D18-1386 | https://aclanthology.org/D18-1386.pdf | emnlp-2018-10 | ['hard-attention'] | ['methodology'] | [ 5.41728914e-01 1.04296362e+00 -2.30043709e-01 -7.67230511e-01
-8.69107008e-01 -7.94879377e-01 9.80508626e-01 6.25593215e-02
-2.63625294e-01 5.35592079e-01 5.45756638e-01 -7.26258099e-01
3.41496795e-01 -5.52923441e-01 -1.07659185e+00 -9.22605991e-02
-6.94167465e-02 3.40909302e-01 1.09937809e-01 -2.48385698e-01
2.28333712e-01 2.58930266e-01 -9.24499273e-01 9.17300999e-01
5.62355757e-01 6.54635131e-01 -5.94870985e-01 8.02048802e-01
2.15217024e-01 9.06786442e-01 -1.03487837e+00 -1.22835958e+00
2.02158034e-01 -4.62129474e-01 -9.45021749e-01 -1.33197471e-01
5.88635623e-01 -5.74484229e-01 -4.78225648e-01 1.05236208e+00
1.18204325e-01 9.40430462e-02 7.60475218e-01 -1.27965331e+00
-1.02352309e+00 1.24468267e+00 -1.45942196e-01 5.50119877e-01
3.30496818e-01 4.77812558e-01 1.37298727e+00 -5.80636978e-01
6.01697266e-01 1.35318696e+00 6.27416551e-01 1.02980042e+00
-1.41532433e+00 -8.69053125e-01 5.93495369e-01 -5.36171421e-02
-7.49959469e-01 -5.58124006e-01 7.46653140e-01 -2.72927105e-01
1.07740784e+00 5.04792333e-01 1.41636431e-01 1.97533274e+00
1.38267830e-01 6.30637407e-01 7.31897771e-01 -3.58363658e-01
1.83411345e-01 3.95278245e-01 4.61945832e-01 7.56880462e-01
2.13223174e-01 3.12788725e-01 -3.89976293e-01 -6.07340395e-01
8.48622471e-02 -1.26444055e-02 -3.41995806e-01 1.24497682e-01
-7.23205328e-01 1.07949173e+00 6.65928960e-01 -7.25151971e-02
-2.04921722e-01 4.48055297e-01 4.08189327e-01 3.57445538e-01
6.15457118e-01 1.09764493e+00 -8.20513666e-01 2.34045669e-01
-5.94499528e-01 2.00445086e-01 9.79240298e-01 7.24047899e-01
3.04236323e-01 2.11979315e-01 -4.20202076e-01 3.51514488e-01
2.13186696e-01 2.75301337e-01 3.89499545e-01 -7.59905815e-01
5.33097088e-01 4.50566620e-01 7.23428801e-02 -7.70509064e-01
-1.63225666e-01 -4.31927323e-01 -3.43071997e-01 -2.41282489e-02
2.31235340e-01 -4.98876035e-01 -9.73256648e-01 1.71457303e+00
-6.69569746e-02 2.05105066e-01 1.31256014e-01 6.01805508e-01
6.04807138e-01 3.50503981e-01 5.52203238e-01 1.45556957e-01
1.07570136e+00 -8.05796087e-01 -4.43986356e-01 -5.84156275e-01
3.71072948e-01 -2.32401222e-01 1.16861629e+00 3.21056396e-01
-8.88747633e-01 -1.86927870e-01 -1.14408600e+00 -6.09635040e-02
-3.27323794e-01 -2.32793197e-01 6.70763493e-01 5.72600842e-01
-5.98434985e-01 8.03573549e-01 -6.95229053e-01 8.03196877e-02
6.95816457e-01 4.05993193e-01 -1.63281128e-01 2.12216467e-01
-1.53101420e+00 9.26827133e-01 1.49943620e-01 -1.93714201e-01
-1.13084781e+00 -7.17709720e-01 -7.17119396e-01 3.33180070e-01
3.61943275e-01 -6.04243159e-01 1.56579220e+00 -1.57321537e+00
-1.25048971e+00 8.31034899e-01 7.78890923e-02 -8.27987254e-01
5.25408804e-01 -3.89648199e-01 -4.90316123e-01 1.62881628e-01
-2.34428316e-01 6.48301303e-01 1.27439952e+00 -1.13225162e+00
-3.99185233e-02 -1.72397092e-01 5.95650136e-01 -3.04526240e-01
-5.24807036e-01 4.82826233e-02 6.55252859e-02 -1.02260303e+00
-4.93885279e-01 -1.03017211e+00 -4.77567375e-01 -2.32546136e-01
-9.19950843e-01 -2.29608510e-02 8.03545535e-01 -7.86030710e-01
1.08082187e+00 -2.04510951e+00 -1.05609605e-02 3.44038099e-01
4.21426833e-01 2.62510538e-01 -1.90778151e-01 1.22381300e-01
-3.44166249e-01 8.55923593e-01 -2.52445191e-01 -4.70102996e-01
1.71402499e-01 1.08885773e-01 -1.07405031e+00 5.62262833e-02
6.09448731e-01 1.03707218e+00 -7.24713504e-01 -6.55603632e-02
-2.23109886e-01 4.55135792e-01 -9.43026781e-01 2.96862066e-01
-7.67669559e-01 7.21613392e-02 -6.61432683e-01 2.43220478e-01
9.04266089e-02 -2.88461149e-01 1.14362173e-01 5.52006476e-02
6.07438028e-01 6.59299493e-01 -3.98208350e-01 1.16630197e+00
-3.44803661e-01 7.19754159e-01 3.91721390e-02 -5.32800198e-01
6.02315784e-01 3.73968363e-01 -2.44049340e-01 -1.44919734e-02
4.67758626e-01 -1.50008932e-01 3.49257141e-02 -4.45185661e-01
2.44399503e-01 -2.18953818e-01 -3.39258939e-01 6.74739599e-01
2.12009922e-02 -7.84874037e-02 -5.09168625e-01 5.79605997e-01
1.49602520e+00 -1.20770410e-01 1.60373613e-01 -5.31329885e-02
3.33733618e-01 8.04933533e-03 1.16529092e-01 8.67621601e-01
-2.03752741e-01 7.17006981e-01 8.20632041e-01 -6.39512181e-01
-8.88268292e-01 -6.82567477e-01 1.23848103e-01 1.29662335e+00
-4.35390323e-01 -4.85594511e-01 -9.81607854e-01 -1.61393189e+00
1.89241081e-01 1.36453497e+00 -1.09123552e+00 -4.85345900e-01
-5.56393683e-01 -5.28843343e-01 8.57792258e-01 6.33687854e-01
1.49406537e-01 -1.12546146e+00 -3.82491380e-01 -8.26976150e-02
2.82913223e-02 -1.08746934e+00 -5.37047207e-01 4.76350278e-01
-4.39309388e-01 -1.24930549e+00 2.20807627e-01 -2.13019058e-01
7.19223142e-01 -2.26932317e-01 1.26393318e+00 6.42987430e-01
2.92228144e-02 2.91323990e-01 -2.76192576e-01 -7.28915870e-01
-8.93930852e-01 4.39149618e-01 -2.82079995e-01 -2.24007145e-01
5.15661120e-01 -3.98996919e-01 -2.67728806e-01 -2.44233832e-02
-1.04168320e+00 -2.42291689e-01 2.68616498e-01 8.19371760e-01
-2.85637882e-02 -3.02955806e-01 6.36505485e-01 -1.69560397e+00
1.00439560e+00 -6.92104220e-01 -3.28128785e-01 -4.02783565e-02
-6.77441955e-01 4.47022945e-01 9.78064835e-01 -6.80233181e-01
-9.15445030e-01 -6.27454147e-02 -3.71836603e-01 -5.40394604e-01
-2.16529176e-01 1.45788193e-01 -2.10605562e-01 2.92371899e-01
1.14349055e+00 -4.83584106e-01 -2.51757145e-01 -9.69383493e-02
5.52166641e-01 5.20789742e-01 5.05584657e-01 -4.13438290e-01
1.02404535e+00 2.33297244e-01 -3.24339271e-01 -2.48196468e-01
-1.33547318e+00 2.52227634e-01 -4.10220146e-01 1.88505068e-01
8.50227296e-01 -4.40431178e-01 -7.55470276e-01 -1.67733505e-01
-1.31201839e+00 -3.68950874e-01 -4.31723356e-01 -3.56564373e-02
-2.49583483e-01 -4.38401774e-02 -7.54517674e-01 -5.03966331e-01
-5.06667256e-01 -9.94806468e-01 9.28760648e-01 -2.59635478e-01
-8.10788631e-01 -1.09599066e+00 -4.28708382e-02 2.99426585e-01
4.23518211e-01 2.93392062e-01 1.19078016e+00 -1.57168269e+00
-2.53068089e-01 -4.78951842e-01 -1.20579330e-02 3.73384446e-01
-3.14752072e-01 -6.23254068e-02 -1.54965019e+00 -2.04915870e-02
-5.67538366e-02 -4.94741112e-01 1.21696043e+00 -7.85025135e-02
1.39025211e+00 -9.88284111e-01 -5.14161825e-01 4.30942059e-01
8.91437888e-01 -2.48788536e-01 4.84013468e-01 2.38752738e-01
7.28302300e-01 6.19635642e-01 1.45411845e-02 8.86517614e-02
1.56836938e-02 3.36798668e-01 7.13311195e-01 -8.84459317e-02
1.33213207e-01 -5.68316638e-01 5.21246493e-01 -2.17734903e-01
2.96646118e-01 -6.52339816e-01 -6.47315621e-01 2.26051673e-01
-1.46484780e+00 -1.15243495e+00 1.97728470e-01 1.78370571e+00
8.88536453e-01 7.82796323e-01 -1.38012305e-01 5.73828891e-02
5.79965234e-01 2.78468609e-01 -7.12683737e-01 -9.83138621e-01
2.55051881e-01 4.29290831e-01 5.49038589e-01 9.03331399e-01
-1.22625506e+00 9.76184249e-01 7.32928371e+00 2.37743169e-01
-9.93995905e-01 1.17913693e-01 8.26468110e-01 -2.90609986e-01
-9.01464880e-01 7.77064217e-03 -7.75877655e-01 2.98057884e-01
1.39526343e+00 1.74009934e-01 3.87215853e-01 8.98735404e-01
-1.71639755e-01 7.09773898e-01 -1.41428030e+00 2.27254797e-02
1.96595639e-01 -1.17778444e+00 3.59388381e-01 3.74295725e-03
4.91466731e-01 -3.04319453e-03 2.23052204e-01 5.42171180e-01
9.56282318e-01 -1.21232307e+00 7.06111252e-01 3.26831162e-01
7.60915160e-01 -7.18427896e-01 5.74596703e-01 2.28227824e-01
-1.45497382e-01 -3.39455485e-01 -1.37607798e-01 -1.66629061e-01
-1.25220507e-01 3.44699264e-01 -1.28828418e+00 -1.82309389e-01
3.42133820e-01 4.62413996e-01 -8.42887640e-01 1.28307277e-02
-9.62789834e-01 1.10162437e+00 -8.66277143e-02 -1.34260720e-03
1.77058548e-01 6.66458189e-01 6.09631658e-01 1.35779929e+00
-3.16408247e-01 2.18501166e-01 -3.81843969e-02 1.07818437e+00
-5.79039454e-01 -3.78976524e-01 -8.32165062e-01 -1.57466710e-01
2.98247248e-01 1.05442703e+00 -2.33230010e-01 -3.58320296e-01
-2.25758910e-01 1.05433345e+00 5.01185536e-01 4.49786782e-01
-1.00048923e+00 -1.90279722e-01 6.35072768e-01 2.47880742e-01
3.23607981e-01 3.15206259e-01 -3.66312921e-01 -1.11533773e+00
-1.88452169e-01 -1.13312638e+00 5.43950617e-01 -8.08944166e-01
-1.39893651e+00 7.69903898e-01 -1.47123441e-01 -5.21020532e-01
-5.87450624e-01 -5.53603292e-01 -1.00438476e+00 7.94300258e-01
-1.19552302e+00 -1.13744736e+00 1.24732712e-02 3.55542392e-01
6.86056554e-01 -1.57865465e-01 1.16987479e+00 -2.64331251e-01
-7.15836287e-01 1.02866900e+00 -6.63968861e-01 2.10203186e-01
6.22758806e-01 -1.40526199e+00 1.14210021e+00 9.15965915e-01
1.04872115e-01 9.43529189e-01 1.19513869e+00 -6.76850736e-01
-9.92167592e-01 -1.23603916e+00 7.28225708e-01 -1.25333142e+00
8.34391475e-01 -6.18540168e-01 -1.09271538e+00 1.44214904e+00
3.82152110e-01 -1.84904963e-01 7.14647770e-01 1.64716303e-01
-8.78703117e-01 4.09599721e-01 -1.26031578e+00 7.21309721e-01
9.43318069e-01 -6.43392563e-01 -9.25468922e-01 4.26125616e-01
1.30783904e+00 -2.09842592e-01 -3.79402310e-01 2.43675977e-01
4.60129917e-01 -6.46958649e-01 9.18754458e-01 -1.30770898e+00
7.83146262e-01 2.99202025e-01 -1.60172638e-02 -1.41061342e+00
-2.29437783e-01 -7.58972287e-01 -2.89810389e-01 1.06830645e+00
9.04337764e-01 -6.41470194e-01 9.08122480e-01 1.18990552e+00
-2.22531796e-01 -5.56072950e-01 -5.47926486e-01 3.67936748e-03
2.35918164e-01 -4.88478422e-01 6.12551808e-01 1.02768517e+00
2.35824153e-01 6.79102600e-01 -2.78690279e-01 3.05529267e-01
2.20353678e-01 -4.42853659e-01 6.35023296e-01 -1.08675015e+00
-7.69102156e-01 -3.91380042e-01 -6.06025718e-02 -6.70546353e-01
8.47926617e-01 -8.99382174e-01 8.76596421e-02 -9.11492884e-01
1.09590091e-01 -5.75487204e-02 -2.26972878e-01 8.38629842e-01
-4.05199528e-01 4.12510306e-01 4.06011045e-02 1.41473442e-01
-3.97057772e-01 2.00082883e-01 7.81615674e-01 -3.19788516e-01
1.68006584e-01 1.86706945e-01 -1.23807967e+00 9.24815774e-01
8.27867150e-01 -8.13487113e-01 -4.69518334e-01 -4.01315778e-01
4.60169345e-01 -2.71988839e-01 4.52878833e-01 -3.56883466e-01
-1.63308665e-01 -4.67594750e-02 4.35132325e-01 5.89093715e-02
2.86731601e-01 -7.49641836e-01 -4.57215905e-01 5.07583916e-01
-1.27401733e+00 1.56354114e-01 3.71948898e-01 7.30982900e-01
8.15576315e-02 -3.22290540e-01 7.34196305e-01 -1.78615168e-01
-1.68215096e-01 -7.31932558e-03 -3.97446722e-01 1.00561179e-01
8.71569335e-01 2.90469110e-01 -7.64675856e-01 -6.05818987e-01
-8.10465217e-01 7.09599629e-02 4.66879368e-01 5.31634569e-01
6.09549403e-01 -7.04300225e-01 -5.11877000e-01 3.15951437e-01
-7.94985611e-03 -5.34500599e-01 -3.48227084e-01 8.59884545e-02
-2.31926199e-02 2.14793324e-01 1.65936053e-01 1.34823427e-01
-1.08665073e+00 9.17786479e-01 5.40459454e-01 -1.63361102e-01
-5.26421845e-01 1.01663709e+00 2.04835266e-01 -3.02637935e-01
3.04142237e-01 -4.28877145e-01 -1.71986759e-01 -4.11370516e-01
5.43945193e-01 3.25263143e-02 1.21658623e-01 -1.92263842e-01
-2.72694290e-01 -4.06252444e-01 -3.13415885e-01 -2.80014157e-01
1.24044204e+00 3.31128776e-01 1.60502091e-01 1.09447964e-01
1.05628026e+00 4.33538347e-01 -1.23364425e+00 1.43947795e-01
-5.68976961e-02 -1.51202351e-01 -6.17219508e-02 -1.16722322e+00
-8.16234887e-01 8.75383139e-01 1.53916270e-01 4.55832958e-01
6.74722910e-01 1.60834212e-02 6.12662911e-01 5.96594870e-01
-2.69325227e-01 -3.98066670e-01 1.91267699e-01 4.73192275e-01
1.07486105e+00 -1.15277338e+00 1.24532714e-01 -3.19918722e-01
-7.36221850e-01 9.26193297e-01 7.68504560e-01 -2.92400539e-01
5.46520114e-01 4.07596201e-01 9.44330469e-02 -4.34086859e-01
-1.20197916e+00 4.32534039e-01 4.30903107e-01 4.75367755e-01
5.32347918e-01 -7.25084469e-02 7.22819567e-02 9.09370899e-01
-5.47186553e-01 -5.49470663e-01 6.94692194e-01 5.62739789e-01
-2.63929844e-01 -8.95278513e-01 -1.12586066e-01 6.08797491e-01
-1.00292420e+00 -3.51495922e-01 -1.12289500e+00 5.91998339e-01
-2.35822350e-01 9.21796143e-01 -2.26542100e-01 -6.00803196e-01
3.65595907e-01 4.21389252e-01 2.76937950e-02 -9.90691423e-01
-1.28596735e+00 -5.97634673e-01 5.05857646e-01 -4.47933167e-01
1.55084386e-01 -5.59549630e-01 -1.19998670e+00 -1.71724901e-01
-3.98157597e-01 1.48742348e-01 6.01351261e-01 1.08608341e+00
3.70808333e-01 5.76796591e-01 4.62296456e-01 -4.98576999e-01
-9.90563929e-01 -1.11323464e+00 1.54528975e-01 8.58594418e-01
5.59149206e-01 -2.78742909e-01 -7.46392131e-01 5.56228459e-02] | [6.003941059112549, 8.079684257507324] |
0d138be7-7f88-4619-9684-52dc75fc39cf | detecting-beats-in-the-photoplethysmogram | null | null | https://doi.org/10.1088/1361-6579/ac826d | https://iopscience.iop.org/article/10.1088/1361-6579/ac826d/pdf | Detecting beats in the photoplethysmogram: benchmarking open-source algorithms | The photoplethysmogram (PPG) signal is widely used in pulse oximeters and smartwatches. A fundamental step in analysing the PPG is the detection of heartbeats. Several PPG beat detection algorithms have been proposed, although it is not clear which performs best. Objective: This study aimed to: (i) develop a framework with which to design and test PPG beat detectors; (ii) assess the performance of PPG beat detectors in different use cases; and (iii) investigate how their performance is affected by patient demographics and physiology. Approach: Fifteen beat detectors were assessed against electrocardiogram-derived heartbeats using data from eight datasets. Performance was assessed using the F1 score, which combines sensitivity and positive predictive value. Main results: Eight beat detectors performed well in the absence of movement, with F1 scores of ≥90% on hospital data and wearable data collected at rest. Their performance was poorer during exercise, with F1 scores of 55-91%; poorer in neonates than adults with F1 scores of 84-96% in neonates compared to 98-99% in adults; and poorer in atrial fibrillation (AF), with F1 scores of 92-97% in AF, compared to 99-100% in normal sinus rhythm. Significance: Two PPG beat detectors denoted 'MSPTD' and 'qppg' performed best, with complementary performance characteristics. This evidence can be used to inform the choice of PPG beat detector algorithm. The algorithms, datasets, and assessment framework are freely available. | ['Joachim A Behar and Panayiotis A Kyriacou', 'Callum Pettit', 'Jonathan Mant', 'Karthik Budidha', 'Philip J Aston', 'Elisa Mejía-Mejía', 'Kevin Kotzen', 'Peter H Charlton'] | 2022-07-19 | null | null | null | physiological-measurement-2022-7 | ['photoplethysmography-ppg-heart-rate', 'photoplethysmography-ppg-beat-detection'] | ['medical', 'medical'] | [ 1.92454934e-01 8.72481018e-02 -6.22911192e-02 1.92010682e-02
-2.11818859e-01 -6.05826497e-01 -1.27360329e-01 5.24433255e-01
-3.92014116e-01 5.81682742e-01 2.66353339e-01 -5.53528488e-01
-2.40340263e-01 -3.89612645e-01 -7.64724314e-02 -5.84983766e-01
-3.38293105e-01 1.47615716e-01 9.22071040e-02 3.97991568e-01
1.29969448e-01 3.72775733e-01 -1.14824355e+00 2.24401087e-01
6.85599387e-01 9.87584412e-01 -3.39634061e-01 1.18692064e+00
7.03247488e-02 4.92489904e-01 -7.60046303e-01 -7.38765462e-04
3.29992473e-01 -9.67523754e-01 -1.65730074e-01 -6.42523348e-01
2.58507669e-01 -2.96846181e-01 1.60976723e-01 3.30341041e-01
1.32746911e+00 -2.99356401e-01 4.07585859e-01 -8.90012920e-01
2.41056398e-01 3.83866549e-01 -4.50883526e-03 5.93682110e-01
5.46404958e-01 6.03043497e-01 2.87673324e-01 -4.00965095e-01
7.14804679e-02 5.28971732e-01 1.07908833e+00 4.92118180e-01
-1.31240678e+00 -4.94670480e-01 -7.51864314e-01 -2.01410070e-01
-1.29345143e+00 -4.75683451e-01 5.36608994e-01 -5.21842718e-01
1.04798460e+00 4.74555910e-01 1.22766387e+00 4.93028551e-01
4.27746713e-01 -1.02887429e-01 1.42173016e+00 -4.22886610e-01
2.20449865e-02 1.98766425e-01 1.42755061e-01 3.67107153e-01
7.03895032e-01 1.49584085e-01 -6.51261866e-01 -3.78063351e-01
7.09408998e-01 -1.07932605e-01 -7.70560682e-01 1.11298814e-01
-1.01324379e+00 5.24706066e-01 -4.08660144e-01 4.24688667e-01
-6.60077393e-01 -1.32109642e-01 3.53720367e-01 2.83182502e-01
-3.99671644e-02 5.77491641e-01 -3.97165328e-01 -1.06416321e+00
-8.81581247e-01 -1.08485520e-01 1.26524246e+00 4.60306078e-01
3.27642202e-01 -2.17418388e-01 -4.85500544e-01 9.18481410e-01
2.04435185e-01 6.39872372e-01 4.20273960e-01 -1.02319741e+00
3.09695691e-01 5.04061222e-01 3.51739913e-01 -7.35764563e-01
-9.56623614e-01 -9.93793085e-02 -6.06219769e-01 7.88540095e-02
7.37092316e-01 -4.39466178e-01 -5.13988912e-01 1.32392776e+00
-3.92764844e-02 7.31619000e-02 -3.22810784e-02 9.10938382e-01
1.07511628e+00 1.80764031e-02 2.64701396e-01 -6.48057938e-01
1.55245006e+00 1.12516275e-02 -4.67749298e-01 2.76523102e-02
6.03723288e-01 -6.47669077e-01 8.22862148e-01 5.08934200e-01
-1.22579551e+00 -5.58503985e-01 -8.98019791e-01 4.67352718e-01
5.32453135e-02 -1.87501758e-01 5.03216609e-02 1.57399893e+00
-1.17429495e+00 8.78235638e-01 -8.43958259e-01 -4.78915364e-01
4.53659803e-01 2.94127196e-01 6.91457912e-02 1.85751826e-01
-1.15550458e+00 1.32622945e+00 1.69443756e-01 -6.65878206e-02
-9.89079997e-02 -9.95021999e-01 -5.54196000e-01 -6.59919009e-02
-2.46001318e-01 -7.42277682e-01 6.66814268e-01 -1.58106416e-01
-1.45277143e+00 7.74427831e-01 -4.01191898e-02 -2.84489363e-01
7.07901955e-01 -1.52917907e-01 -6.79321706e-01 6.15603507e-01
-1.52653009e-01 -5.69565743e-02 3.73431355e-01 -5.21596551e-01
-3.28540891e-01 -3.40493977e-01 -3.61577302e-01 3.15588176e-01
1.42523006e-01 8.44831169e-02 1.27521113e-01 -4.44961712e-02
1.29098520e-01 -6.68532372e-01 8.86431411e-02 -1.50748014e-01
1.19549386e-01 1.09385639e-01 4.22299355e-01 -7.64274240e-01
1.44394052e+00 -1.85313714e+00 -8.03278148e-01 4.25406307e-01
4.81098831e-01 6.93409264e-01 3.65827471e-01 4.31210756e-01
-4.13818192e-03 2.17178464e-01 -1.62789524e-01 3.32731545e-01
2.52421480e-02 8.54120925e-02 1.01531349e-01 7.17051208e-01
-4.37433049e-02 8.63789439e-01 -9.90515590e-01 -3.88032913e-01
8.69580865e-01 6.46977067e-01 -2.22434148e-01 2.60960370e-01
7.27140486e-01 6.18733585e-01 8.85760188e-02 5.92481732e-01
3.89354974e-01 1.35518447e-01 1.67921945e-01 -3.60338241e-01
-5.44373214e-01 4.68256801e-01 -1.07487786e+00 1.11291099e+00
-5.22675924e-02 6.96077108e-01 -2.24324867e-01 -7.11780548e-01
1.54133308e+00 8.73320878e-01 7.69768357e-01 -8.23052943e-01
2.08923191e-01 3.32151949e-01 6.03868544e-01 -1.06385839e+00
-4.62428778e-01 -8.00197780e-01 5.23583949e-01 5.85822284e-01
-3.39626133e-01 -2.96314359e-01 7.46695697e-02 -2.63730586e-01
1.23509645e+00 -9.23251510e-02 5.26471794e-01 -6.53504908e-01
2.70100057e-01 -5.08576632e-01 3.85466069e-01 1.08739495e+00
-6.54934824e-01 9.42571938e-01 7.60454297e-01 -6.84672177e-01
-4.14017469e-01 -8.77878010e-01 -7.00624704e-01 2.97079712e-01
-6.63027167e-02 -1.49363801e-01 -4.89403665e-01 -9.23257768e-02
1.85465455e-01 5.18613279e-01 -4.17167693e-01 -2.52249092e-01
-4.54706460e-01 -9.50252175e-01 1.03794241e+00 6.01498127e-01
3.06450486e-01 -1.19845772e+00 -1.69502139e+00 5.99900186e-01
-1.29185349e-01 -7.37410247e-01 -1.44217955e-03 5.73447831e-02
-9.29755569e-01 -1.28550828e+00 -8.72428358e-01 -2.34849676e-02
1.17888920e-01 -2.53548533e-01 1.23893201e+00 1.75468594e-01
-7.23030329e-01 6.93277359e-01 -3.43239695e-01 -8.55279386e-01
-4.49880183e-01 -4.27981406e-01 2.43367150e-01 -2.39100710e-01
6.07156992e-01 -7.89049923e-01 -1.29098189e+00 3.12150657e-01
-3.75327408e-01 -3.12216550e-01 4.42965418e-01 2.78207004e-01
2.71835804e-01 -9.05666947e-01 7.50992477e-01 -5.83636820e-01
7.11197853e-01 -3.49740982e-01 -1.91012233e-01 7.81984534e-03
-1.11527812e+00 -6.06141090e-01 4.56908166e-01 -2.79741853e-01
-4.46273163e-02 -4.91460890e-01 -4.29548249e-02 -2.13276923e-01
-1.55012861e-01 1.68212116e-01 2.76273221e-01 4.44997959e-02
1.06484818e+00 1.56397913e-02 4.86709982e-01 -2.24130452e-01
-4.39211190e-01 5.15696108e-01 6.48207009e-01 -5.70837438e-01
1.93884954e-01 1.97419658e-01 3.01246136e-01 -1.08418465e+00
-1.55244932e-01 -7.25076914e-01 -6.83482587e-01 -5.34334719e-01
8.09045196e-01 -4.61495489e-01 -9.59692121e-01 2.19335422e-01
-5.65225720e-01 -4.21171427e-01 -4.33946222e-01 1.06584215e+00
-2.72993773e-01 5.50700784e-01 -1.71791703e-01 -1.18252075e+00
-9.29514349e-01 -7.00977385e-01 3.34328502e-01 4.03704017e-01
-8.57079327e-01 -1.14878082e+00 4.52416211e-01 5.36599159e-02
7.82687545e-01 1.07425451e+00 5.15535176e-01 -5.87705851e-01
4.01909083e-01 -4.55715299e-01 1.95599403e-02 4.17584151e-01
6.67355299e-01 2.03393489e-01 -9.93476868e-01 -3.06915343e-01
7.81323016e-02 2.80782193e-01 3.11002702e-01 6.91865206e-01
4.40970778e-01 -1.32857963e-01 -1.17758319e-01 6.50917351e-01
1.32219088e+00 7.64306664e-01 9.74723279e-01 1.21570282e-01
3.25917870e-01 5.79268098e-01 1.54027075e-01 5.64160347e-01
3.91090997e-02 1.79598555e-01 5.52076586e-02 -2.72693425e-01
-8.88710916e-02 4.94487844e-02 1.77044645e-01 3.28590930e-01
-6.10701323e-01 1.93609834e-01 -1.15750909e+00 4.54592913e-01
-1.18914127e+00 -8.83011281e-01 -6.41660750e-01 2.51437688e+00
3.56643528e-01 1.66018933e-01 7.51564205e-01 3.31696302e-01
7.62039065e-01 -6.84752688e-02 -3.90212744e-01 -9.03155148e-01
-5.11061847e-02 8.00225496e-01 4.85046864e-01 1.51027471e-01
-6.21141791e-01 -4.38200831e-02 6.79357004e+00 -6.58604026e-01
-1.19808221e+00 -1.15697809e-01 4.16931540e-01 -1.81157872e-01
3.07830960e-01 1.40856683e-01 -3.26606691e-01 1.05510426e+00
1.30332077e+00 -2.56973028e-01 3.55561152e-02 3.09506863e-01
6.36671543e-01 -3.61740053e-01 -1.07275474e+00 1.26458001e+00
-9.74609982e-03 -1.09652603e+00 -7.58714676e-01 -2.34884247e-01
6.46486878e-02 2.60219034e-02 -4.93893951e-01 -1.33565605e-01
-5.80278754e-01 -1.03981900e+00 2.44251087e-01 7.97418237e-01
1.38764381e+00 -1.47382393e-01 7.86919355e-01 -4.67164256e-02
-9.52017188e-01 1.94306061e-01 -1.89392939e-01 -3.59738827e-01
1.67051896e-01 6.28307939e-01 -1.12421167e+00 2.05447450e-01
7.31086791e-01 2.84472227e-01 -2.03315109e-01 1.45759439e+00
-2.39302024e-01 1.04697716e+00 -8.20739865e-01 -1.30697772e-01
-3.64052922e-01 -7.03063160e-02 6.32563651e-01 1.37462604e+00
3.57217968e-01 6.22501552e-01 -4.29025143e-01 8.66323054e-01
5.74716270e-01 1.62234724e-01 -5.53212881e-01 1.47503868e-01
6.65428698e-01 1.00760460e+00 -7.32254088e-01 -3.46802294e-01
-6.79175377e-01 4.04681861e-02 -6.12695932e-01 2.65853405e-01
-5.38591027e-01 -7.42000878e-01 3.63148302e-01 7.01345265e-01
-3.67992461e-01 3.22137624e-01 -7.19312310e-01 -6.52845263e-01
5.63754253e-02 -6.64528728e-01 6.24207556e-01 -5.33981800e-01
-5.80430090e-01 2.58431226e-01 -2.01139748e-02 -1.33603859e+00
-2.84786224e-01 -4.21778291e-01 -1.06477988e+00 1.42837548e+00
-1.20369494e+00 -1.61792055e-01 -7.16340840e-01 2.12813139e-01
-3.66169393e-01 2.34769017e-01 1.04593241e+00 1.01382703e-01
-2.53050208e-01 6.59580827e-01 -3.61070395e-01 3.63533720e-02
5.93308449e-01 -1.27101839e+00 -7.61796758e-02 6.93228483e-01
-6.75304294e-01 7.47929990e-01 5.53686202e-01 -4.74558443e-01
-1.10307574e+00 -5.35896301e-01 7.45781183e-01 -7.10520148e-01
8.92310366e-02 4.41601396e-01 -9.09844220e-01 -1.30587772e-01
-3.25235933e-01 1.84095651e-01 1.10473347e+00 -2.23388478e-01
9.21582207e-02 -2.75650412e-01 -1.61541450e+00 -1.09955696e-02
4.18110013e-01 -3.07916939e-01 -6.73474967e-01 -1.82803050e-01
-2.94187874e-01 -5.62711656e-01 -1.65346670e+00 7.58522213e-01
1.29641533e+00 -1.26939249e+00 7.38120794e-01 -1.98137522e-01
-9.89744440e-02 -2.19940990e-01 4.40554738e-01 -8.45785618e-01
-3.69551927e-01 -1.21691656e+00 -8.82544816e-02 9.68414426e-01
1.91427693e-01 -1.52452910e+00 5.20724535e-01 1.15562010e+00
-2.61885226e-01 -7.21195281e-01 -7.66407847e-01 -5.41579425e-01
-1.87545568e-01 -1.92721680e-01 2.90538520e-01 8.84833097e-01
5.63612998e-01 1.32313687e-02 -1.23000413e-01 -1.20936558e-01
3.74108851e-01 -1.91552676e-02 5.31506300e-01 -1.25524247e+00
-2.62126207e-01 -4.75949764e-01 -4.52524781e-01 -7.70988837e-02
-1.03964984e+00 -5.94025254e-01 -2.36080915e-01 -1.85512519e+00
-1.34982586e-01 -4.64004368e-01 -7.02869058e-01 6.57872677e-01
-3.64016533e-01 3.76121938e-01 -2.23807041e-02 9.09672976e-02
2.52025276e-01 -4.94227976e-01 8.30151081e-01 5.28867126e-01
-9.53378677e-01 6.54893816e-02 -6.39738917e-01 2.09629923e-01
8.75542283e-01 -5.07716060e-01 -3.43750298e-01 3.49149138e-01
5.37793711e-02 4.01010573e-01 2.64935464e-01 -1.18861675e+00
3.70764360e-03 -7.82720447e-02 6.23080909e-01 -3.47755522e-01
-3.13936435e-02 -4.43398327e-01 6.14207327e-01 1.10443914e+00
2.75886208e-01 2.10784763e-01 3.93396318e-01 -2.23372191e-01
1.43615931e-01 -1.97937772e-01 9.92641449e-01 -1.73457488e-01
5.71289174e-02 1.27035901e-01 -4.11833555e-01 4.66298312e-01
7.00209737e-01 -1.00998211e+00 -6.35256231e-01 -3.59599054e-01
-6.90140903e-01 -3.63602042e-02 4.46013957e-01 2.50673622e-01
5.77105939e-01 -8.46077740e-01 -8.85840774e-01 3.42954308e-01
2.31743768e-01 -2.64461875e-01 2.22173139e-01 1.81047213e+00
-1.14820683e+00 2.43681133e-01 -2.04584986e-01 -7.24138677e-01
-1.11059415e+00 4.98035401e-02 6.97116256e-01 3.93758476e-01
-9.58210528e-01 5.87905765e-01 -1.66424736e-01 5.88396847e-01
7.57878497e-02 -3.14189225e-01 -2.29139596e-01 1.47331208e-01
7.09151208e-01 8.31018627e-01 2.09280729e-01 -2.05908924e-01
-6.98888898e-01 7.28449881e-01 4.87006426e-01 9.55474749e-02
8.90627325e-01 6.48364276e-02 -2.65423860e-02 4.50082421e-01
8.71207833e-01 -6.63610846e-02 -8.24048519e-01 4.86162722e-01
-2.36473545e-01 -4.67478782e-01 -3.59033793e-01 -1.01112151e+00
-6.37723207e-01 8.24173927e-01 1.15402424e+00 5.81530511e-01
1.16900241e+00 -2.60230809e-01 3.29862893e-01 -2.52505124e-01
1.43336385e-01 -1.06930518e+00 -2.47866988e-01 -2.77385265e-01
4.57748353e-01 -5.04058838e-01 8.37933943e-02 6.56042248e-02
-5.47151685e-01 1.26286399e+00 1.08761236e-01 1.73894823e-01
6.14242673e-01 7.63446018e-02 5.49626946e-01 -2.87240893e-01
-3.75827581e-01 6.06922759e-03 6.46064356e-02 8.01028430e-01
7.87555277e-01 2.52569437e-01 -1.07656085e+00 4.88745600e-01
-2.21110508e-01 2.92907655e-01 5.96552908e-01 8.98465574e-01
-5.20744026e-01 -8.52142155e-01 -2.76126325e-01 1.04861772e+00
-7.85116553e-01 7.59405047e-02 -2.26851687e-01 7.20722556e-01
2.19472229e-01 1.34797096e+00 -2.28976190e-01 -1.28928646e-01
6.63875639e-01 6.48491442e-01 4.34682816e-01 -4.32874054e-01
-7.90616214e-01 3.43391486e-02 3.63643765e-01 -3.42289656e-01
-5.86735189e-01 -7.35878348e-01 -1.08482194e+00 9.38549936e-02
-2.00631693e-01 4.29802150e-01 5.90625286e-01 5.85724354e-01
3.48890513e-01 3.50418687e-01 4.92446899e-01 -2.59786814e-01
-1.71309650e-01 -1.17797422e+00 -6.57182038e-01 2.48371989e-01
4.41729993e-01 -4.42402363e-01 -7.40478396e-01 1.36779873e-02] | [14.040369033813477, 3.033963441848755] |
0e07620f-e935-4ef6-af92-413b2a82208a | tweester-at-semeval-2016-task-4-sentiment | null | null | https://aclanthology.org/S16-1023 | https://aclanthology.org/S16-1023.pdf | Tweester at SemEval-2016 Task 4: Sentiment Analysis in Twitter Using Semantic-Affective Model Adaptation | null | ['ros', 'Haris Papageorgiou', 'Mal', 'Fenia Christopoulou', 'Elisavet Palogiannidi', 'Filippos Kokkinos', 'Alex Potamianos', 'Shrikanth Narayanan', 'Nikolaos rakis', 'Elias Iosif', 'Athanasia Kolovou'] | 2016-06-01 | null | null | null | semeval-2016-6 | ['twitter-sentiment-analysis'] | ['natural-language-processing'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.291708946228027, 3.7703092098236084] |
63ba5046-337d-41ba-9e61-16f666aa5c77 | recent-applications-of-machine-learning | 2306.04566 | null | https://arxiv.org/abs/2306.04566v1 | https://arxiv.org/pdf/2306.04566v1.pdf | Recent applications of machine learning, remote sensing, and iot approaches in yield prediction: a critical review | The integration of remote sensing and machine learning in agriculture is transforming the industry by providing insights and predictions through data analysis. This combination leads to improved yield prediction and water management, resulting in increased efficiency, better yields, and more sustainable agricultural practices. Achieving the United Nations' Sustainable Development Goals, especially "zero hunger," requires the investigation of crop yield and precipitation gaps, which can be accomplished through, the usage of artificial intelligence (AI), machine learning (ML), remote sensing (RS), and the internet of things (IoT). By integrating these technologies, a robust agricultural mobile or web application can be developed, providing farmers and decision-makers with valuable information and tools for improving crop management and increasing efficiency. Several studies have investigated these new technologies and their potential for diverse tasks such as crop monitoring, yield prediction, irrigation management, etc. Through a critical review, this paper reviews relevant articles that have used RS, ML, cloud computing, and IoT in crop yield prediction. It reviews the current state-of-the-art in this field by critically evaluating different machine-learning approaches proposed in the literature for crop yield prediction and water management. It provides insights into how these methods can improve decision-making in agricultural production systems. This work will serve as a compendium for those interested in yield prediction in terms of primary literature but, most importantly, what approaches can be used for real-time and robust prediction. | ['Abdelghani Chehbouni', 'Ayoub Kechchour', 'Terence Epule Epule', 'Fatima Zahra Bassine'] | 2023-06-07 | null | null | null | null | ['crop-yield-prediction', 'crop-yield-prediction'] | ['computer-vision', 'miscellaneous'] | [ 1.21782668e-01 -3.43270212e-01 -7.68414259e-01 -7.58410990e-03
1.31540254e-01 -6.82853162e-01 1.99992750e-02 7.86024153e-01
1.99047819e-01 7.01866627e-01 -2.35490471e-01 -9.09105957e-01
-2.68230289e-01 -1.69352686e+00 -2.48612776e-01 -9.13995564e-01
-1.43325448e-01 -1.97339416e-01 -2.35854119e-01 -5.45648277e-01
3.22076716e-02 7.99436331e-01 -1.98798239e+00 5.16602173e-02
1.13886464e+00 1.07999551e+00 8.82703185e-01 5.83972514e-01
-2.43663147e-01 7.71058574e-02 -3.01047526e-02 9.89825651e-02
2.07584631e-02 -8.08923542e-02 -3.67257714e-01 -3.55494082e-01
-6.93081737e-01 -3.41219515e-01 5.82224369e-01 1.00426781e+00
4.55849290e-01 -3.15122306e-01 4.30783123e-01 -8.62070501e-01
-6.74693644e-01 6.22098625e-01 -7.32649863e-01 -2.44385511e-01
1.21200085e-01 -1.49242012e-02 3.92216831e-01 -5.41938543e-01
3.14961039e-02 1.05169201e+00 7.42888868e-01 -1.07458606e-01
-9.28401947e-01 -7.22933292e-01 -2.61699259e-01 1.73163399e-01
-9.85311210e-01 -2.22722828e-01 2.82600343e-01 -3.03610176e-01
1.01669323e+00 4.24122661e-01 1.18549240e+00 9.15826187e-02
7.10679710e-01 5.14806926e-01 9.92821693e-01 -8.29416573e-01
3.99635553e-01 -6.36241063e-02 -1.16789028e-01 2.10526168e-01
8.10651600e-01 3.47150117e-01 2.45229434e-03 6.80788308e-02
5.33668280e-01 4.54430610e-01 8.20681825e-02 -1.35757476e-01
-1.04388511e+00 9.30382729e-01 7.42289007e-01 5.47971845e-01
-1.15199494e+00 -1.66752726e-01 1.75853103e-01 2.61714637e-01
7.51958489e-01 3.28887135e-01 -1.14720213e+00 3.45013514e-02
-7.81408310e-01 6.16230890e-02 8.75110269e-01 7.47024059e-01
8.59284699e-01 5.23660719e-01 2.90061653e-01 5.83033562e-01
5.54942071e-01 1.60442054e+00 2.36518651e-01 -9.24173653e-01
1.29729047e-01 6.52445018e-01 3.70972544e-01 -1.28040314e+00
-5.99700212e-01 7.48061482e-03 -1.03631639e+00 1.60217464e-01
-9.80363786e-02 -6.61815703e-01 -6.21214271e-01 1.04643846e+00
2.17278779e-01 -1.82188883e-01 5.78510046e-01 4.64066565e-01
8.19754004e-01 1.03874254e+00 6.67574406e-01 -2.90057987e-01
1.33444726e+00 -3.43229622e-01 -9.04163718e-01 -6.89080060e-02
9.45010185e-01 -8.76629412e-01 5.03192842e-01 4.92380187e-02
-6.13839507e-01 -4.40447539e-01 -8.77834320e-01 6.06883645e-01
-1.25309825e+00 3.54314774e-01 1.42349994e+00 8.58616471e-01
-8.07416260e-01 8.24281812e-01 -1.06613684e+00 -9.79394615e-01
2.62637258e-01 2.14826167e-01 -1.15791306e-01 -4.83428739e-04
-1.18270421e+00 1.33907461e+00 5.61551750e-01 4.06321377e-01
-2.22531065e-01 -7.33403742e-01 -8.39176297e-01 1.46657646e-01
-1.78799137e-01 -3.91742647e-01 8.01742494e-01 -5.42182446e-01
-1.49426782e+00 7.64100611e-01 -1.72658548e-01 -4.84571844e-01
-2.30340555e-01 -4.85961795e-01 -5.87854922e-01 -3.16979527e-01
4.29443978e-02 4.15311366e-01 1.21536754e-01 -1.07153571e+00
-1.09208381e+00 -8.96431267e-01 -4.02283132e-01 -2.59929448e-02
-3.40378970e-01 1.19230434e-01 7.21501291e-01 -3.05200219e-01
4.65777427e-01 -1.02465189e+00 -4.38968301e-01 -3.10271800e-01
1.60015151e-01 1.47368401e-01 1.09192133e+00 -8.78515720e-01
1.09731054e+00 -1.87736416e+00 -4.09423739e-01 1.77956298e-01
-3.51192951e-01 8.47334802e-01 -7.86736682e-02 7.37583220e-01
1.87056467e-01 2.20275775e-01 -2.82576345e-02 6.80057883e-01
-4.31157082e-01 5.11148274e-01 -2.56679028e-01 2.03802004e-01
3.36519778e-01 9.96826470e-01 -1.08570552e+00 3.64333130e-02
9.39213812e-01 4.30988520e-01 2.18355417e-01 1.01545677e-01
-1.51856571e-01 5.47932744e-01 -8.34756911e-01 1.20467937e+00
1.06042922e+00 1.00175075e-01 3.49096090e-01 -2.71393135e-02
-7.43791342e-01 -4.26670969e-01 -8.85561109e-01 1.08035123e+00
-8.64310622e-01 3.42704386e-01 -1.06005803e-01 -1.30424583e+00
1.48909581e+00 5.04974008e-01 5.79755902e-01 -4.77789760e-01
-1.11712232e-01 7.10820496e-01 -1.94991782e-01 -8.13508868e-01
2.39420041e-01 4.76016760e-01 3.32442135e-01 8.09745193e-02
-2.54151046e-01 -2.78091758e-01 -1.60616990e-02 -7.35462010e-01
3.83491963e-01 5.79817533e-01 5.98555207e-01 -6.24136508e-01
3.22330713e-01 2.29337692e-01 3.31765950e-01 2.63574511e-01
-2.10486010e-01 -2.51775801e-01 -8.80103856e-02 -8.57783318e-01
-9.07384515e-01 -5.95992029e-01 -2.26713508e-01 1.44227314e+00
-2.55913213e-02 4.23783749e-01 -4.36014056e-01 1.39164433e-01
4.56099421e-01 8.20846081e-01 -2.77257890e-01 1.78929362e-02
-1.68964207e-01 -1.53010213e+00 5.09765372e-02 5.79015255e-01
8.49505663e-01 -1.48633778e+00 -1.09573412e+00 4.75839049e-01
-3.05148602e-01 -8.60886335e-01 9.18545842e-01 5.23839355e-01
-1.46440995e+00 -5.97353637e-01 -6.12758279e-01 -4.94291633e-01
1.99068457e-01 7.82301366e-01 1.23936844e+00 4.89268862e-02
-1.33685425e-01 -1.39487579e-01 -7.64586031e-01 -1.35390723e+00
-3.50463867e-01 3.96242827e-01 -1.09920360e-01 -6.03910625e-01
1.00438607e+00 -6.99145854e-01 -7.06889212e-01 -2.74183378e-02
-7.18013585e-01 -4.09659483e-02 8.14746916e-01 4.19728190e-01
4.58965838e-01 7.22763911e-02 1.15753365e+00 -6.14944816e-01
1.12528250e-01 -1.01514745e+00 -9.26125288e-01 5.86903930e-01
-1.02725875e+00 -2.02047840e-01 1.93947837e-01 4.53267619e-02
-9.51638937e-01 3.96825641e-01 7.29292631e-02 4.79706496e-01
-5.53987324e-01 1.03997421e+00 -2.84315407e-01 -9.86934975e-02
4.93036300e-01 5.51735125e-02 9.22668427e-02 -4.24829543e-01
4.17604536e-01 9.98531342e-01 1.63978100e-01 -5.40297404e-02
4.29084718e-01 1.00238226e-01 2.88367450e-01 -1.14229155e+00
-4.58577991e-01 -5.02407312e-01 -9.53643918e-01 -2.40139708e-01
6.93029165e-01 -9.07076418e-01 -5.92195392e-01 6.57265425e-01
-9.36030507e-01 -2.09002748e-01 -8.12817886e-02 8.19959462e-01
-3.72529328e-01 -3.36760789e-01 -1.30888119e-01 -1.14120340e+00
-9.18842435e-01 -7.67430723e-01 7.37646639e-01 7.73766518e-01
-1.56147536e-02 -1.01916409e+00 -1.60137296e-01 -6.27237782e-02
1.13232911e+00 6.36698186e-01 8.98181200e-01 -1.05180174e-01
-1.11594677e-01 -4.83285040e-01 -4.34279561e-01 1.90587074e-01
7.43767738e-01 4.43844289e-01 -1.09054518e+00 6.18300922e-02
-4.59786385e-01 4.03939607e-03 5.49196362e-01 1.03751624e+00
8.30668747e-01 -1.38167888e-01 -6.27896965e-01 5.57978749e-01
1.73434496e+00 4.94418979e-01 6.17830813e-01 3.64314795e-01
1.82125345e-01 9.15174425e-01 1.18440068e+00 7.05433965e-01
3.87473017e-01 2.96668801e-02 8.47085953e-01 -5.19655347e-01
3.70102972e-01 -3.04133864e-03 5.88800684e-02 8.09570968e-01
-8.34552228e-01 -5.09544872e-02 -1.08392775e+00 6.45483315e-01
-2.09594631e+00 -1.08298910e+00 -2.55580902e-01 2.20635915e+00
5.31339526e-01 -5.62504113e-01 -1.74057722e-01 4.48705614e-01
5.37776053e-01 -3.00324485e-02 -4.66845930e-01 -7.60066092e-01
-3.07650506e-01 5.19417405e-01 1.34376085e+00 -1.09065503e-01
-1.28010726e+00 1.39430225e+00 6.26393270e+00 4.14142817e-01
-1.51780915e+00 -2.16645837e-01 5.76497734e-01 7.97908247e-01
-1.57617539e-01 2.89322417e-02 -7.35065579e-01 1.19979434e-01
1.08953798e+00 -1.00232989e-01 3.99418175e-01 9.45994973e-01
9.30320024e-01 -5.58855534e-01 -2.31712475e-01 3.07404429e-01
-8.29133689e-01 -1.34837687e+00 -1.85265958e-01 -9.39049497e-02
8.91279697e-01 2.84037381e-01 -1.81065515e-01 -1.67245939e-02
5.85672021e-01 -7.62399137e-01 -7.95838609e-03 7.45310009e-01
6.11009240e-01 -7.60226071e-01 1.29590797e+00 2.40588605e-01
-1.60248840e+00 -3.50584030e-01 -7.92445779e-01 -5.45877099e-01
-3.56126666e-01 1.00075996e+00 -2.29639396e-01 1.00823855e+00
1.06829345e+00 9.30445790e-01 -1.24176316e-01 8.43160152e-01
-8.15427676e-03 7.59104788e-01 -3.62953365e-01 -3.52352321e-01
1.37759447e-01 -5.44546187e-01 -7.18471855e-02 1.01704955e+00
6.84680820e-01 4.34846342e-01 1.34158686e-01 4.35617775e-01
6.81989551e-01 4.67360914e-01 -8.98110807e-01 -1.12565644e-01
7.87414134e-01 1.26486897e+00 -7.82201529e-01 3.22584212e-02
-4.36100423e-01 3.05288702e-01 -5.78866005e-01 2.02684313e-01
-2.49369323e-01 -6.52683318e-01 6.65376365e-01 -9.32507068e-02
6.53226748e-02 -3.68309826e-01 -7.33347714e-01 -7.30260730e-01
-3.79343599e-01 -4.38342124e-01 -4.71459031e-02 -7.20501065e-01
-8.58341753e-01 -1.41156008e-02 -6.72836825e-02 -1.08760703e+00
-3.13814789e-01 -6.18471563e-01 -5.05247176e-01 1.20355260e+00
-2.14927053e+00 -1.52449977e+00 -5.64591765e-01 3.11912224e-02
1.60769850e-01 -2.43622810e-01 1.73252034e+00 -1.49636371e-02
-3.02956760e-01 -1.63578361e-01 8.14879715e-01 -3.23119789e-01
2.47756749e-01 -8.97095978e-01 1.45886779e-01 4.67719436e-01
-4.52007413e-01 -2.24630758e-01 4.48806077e-01 -7.10044146e-01
-1.62123191e+00 -1.36209142e+00 1.07597768e+00 2.46672899e-01
5.79464436e-01 5.01294732e-01 -6.86091244e-01 4.27033365e-01
-1.54894397e-01 3.96595821e-02 7.44112730e-01 5.00359870e-02
2.79273421e-01 -7.18418479e-01 -1.32085001e+00 2.17423618e-01
3.39248031e-01 1.77342579e-01 2.14580640e-01 4.01048452e-01
5.87955654e-01 8.93691257e-02 -1.31797624e+00 7.99683690e-01
1.19968200e+00 -5.87854922e-01 8.59238088e-01 -6.13065541e-01
5.66903651e-01 1.35712355e-01 -3.64846975e-01 -1.52812958e+00
-4.56459552e-01 -2.84351222e-02 -5.00894189e-02 1.10763109e+00
4.69635546e-01 -7.41840482e-01 5.23113430e-01 5.37178516e-01
2.39090547e-01 -6.33778334e-01 -2.90471703e-01 -5.20359457e-01
3.47987562e-01 -2.73520172e-01 1.21084452e+00 8.80315244e-01
8.67450908e-02 -3.11507553e-01 -9.01151896e-02 4.97127652e-01
2.89340496e-01 3.81529719e-01 5.95675766e-01 -1.65701830e+00
7.41605997e-01 -2.34332234e-01 -4.98696566e-01 -1.43567696e-01
-2.95768559e-01 -3.53534460e-01 -2.25419909e-01 -1.65652883e+00
-1.26628101e-01 -7.62660444e-01 -1.93745941e-01 8.69109273e-01
-2.66716003e-01 -2.53582466e-02 1.36041299e-01 1.89877912e-01
4.26752776e-01 2.16267601e-01 1.20027518e+00 7.28122815e-02
-7.61465728e-01 7.03281581e-01 -5.77530146e-01 7.62374520e-01
1.50842476e+00 -2.28830650e-01 -1.57163441e-02 -5.23146152e-01
4.85263497e-01 3.47414196e-01 2.60814782e-02 -9.58734751e-01
-3.62588018e-01 -8.78980815e-01 6.46086991e-01 -7.54128218e-01
-4.19287592e-01 -1.10854816e+00 2.40583375e-01 9.85425472e-01
2.04780236e-01 1.48028940e-01 5.70722938e-01 4.40062732e-02
4.49692421e-02 -1.03452779e-01 5.54531932e-01 -6.77793995e-02
-9.28757906e-01 1.11063391e-01 -6.21908128e-01 -1.00993943e+00
1.28937304e+00 -7.47805685e-02 -2.18374729e-01 -1.44247189e-01
-8.02138090e-01 2.67598659e-01 6.58579497e-03 5.39220870e-01
2.93852001e-01 -1.07788968e+00 -1.10520005e+00 3.92942607e-01
1.25937432e-01 -2.91189283e-01 -2.27945708e-02 5.06883800e-01
-1.07720840e+00 6.68679476e-01 -6.61385298e-01 -5.55857420e-01
-9.83497620e-01 5.17589629e-01 3.79910171e-02 -2.15023637e-01
-3.89580168e-02 1.18880734e-01 -6.14460588e-01 -6.58056676e-01
-1.75077260e-01 -5.66086173e-01 -8.74998569e-01 1.12464257e-01
4.14243758e-01 6.37642741e-01 2.28718504e-01 -4.31414813e-01
-3.08262259e-01 6.90267920e-01 8.06005120e-01 3.47777605e-01
1.62958467e+00 -5.11379361e-01 -3.64960939e-01 6.18961334e-01
4.32028383e-01 -4.90075439e-01 -6.63878322e-01 5.32140955e-02
3.51194084e-01 -2.92774409e-01 3.74495357e-01 -1.04576933e+00
-1.29686868e+00 9.33178663e-01 1.36760437e+00 5.87660849e-01
1.45390952e+00 -6.00355446e-01 6.47171974e-01 6.15729690e-01
5.51540554e-01 -8.54859114e-01 -8.40614736e-01 3.78278911e-01
6.55443251e-01 -1.79046309e+00 1.09809980e-01 -5.16410649e-01
-1.66552290e-01 1.37731850e+00 8.50388408e-02 1.95329159e-01
1.40082455e+00 4.88865942e-01 3.95079941e-01 3.12716037e-01
-2.62110054e-01 -5.75862408e-01 -2.71504283e-01 1.12084544e+00
8.44715774e-01 1.09481204e+00 -4.51247156e-01 3.15420568e-01
-9.59059298e-02 7.38140821e-01 -4.10870602e-03 1.19227493e+00
-9.55010951e-01 -1.16275728e+00 -6.47295356e-01 8.27544391e-01
-4.75236505e-01 -2.05329493e-01 1.31270707e-01 3.11291307e-01
1.83822691e-01 1.39847660e+00 -6.12084158e-02 -2.58267730e-01
1.75957978e-01 -1.89307377e-01 7.48661757e-02 -3.60536933e-01
-3.73917371e-01 -1.31071076e-01 -2.49273762e-01 -2.88195819e-01
-9.03409660e-01 -6.68765903e-01 -8.51321161e-01 -9.29157913e-01
-8.10042202e-01 -6.98631257e-02 1.36912501e+00 9.85337019e-01
6.17988110e-01 2.63239026e-01 1.00318599e+00 -1.00538862e+00
-2.67880619e-01 -1.15054274e+00 -9.33997571e-01 -4.76077497e-01
-1.45271361e-01 -5.09290040e-01 2.51207769e-01 1.64037153e-01] | [9.341012001037598, -1.5977263450622559] |
41e9b8aa-9bc4-4e83-8566-393551e89c30 | fine-grained-visual-recognition-with-batch | 1910.12423 | null | https://arxiv.org/abs/1910.12423v3 | https://arxiv.org/pdf/1910.12423v3.pdf | ACE: Adaptive Confusion Energy for Natural World Data Distribution | With the development of deep learning, standard classification problems have achieved good results. However, conventional classification problems are often too idealistic. Most data in the natural world usually have imbalanced distribution and fine-grained characteristics. Recently, many state-of-the-art approaches tend to focus on one or another separately, but rarely on both. In this paper, we introduce a novel and adaptive batch-wise regularization based on the proposed Adaptive Confusion Energy (ACE) to flexibly address the nature world distribution, which usually involves fine-grained and long-tailed properties at the same time. ACE increases the difficulty of the training process and further alleviates the overfitting problem. Through the datasets with the technical issue in fine-grained (CUB, CAR, AIR) and long-tailed (ImageNet-LT), or comprehensive issues (CUB-LT, iNaturalist), the result shows that the ACE is not only competitive to some state-of-the-art on performance but also demonstrates the effectiveness of training. | ['Wan-Cyuan Fan', 'Tyng-Luh Liu', 'Ming-Sui Lee', 'Cheng-Yao Hong', 'Yen-Chi Hsu', 'Davi Geiger'] | 2019-10-28 | null | null | null | null | ['fine-grained-visual-recognition'] | ['computer-vision'] | [-5.42097151e-01 -7.82785177e-01 -8.53121206e-02 -4.74353820e-01
-5.93527079e-01 -1.82979062e-01 4.54261988e-01 2.92588174e-02
-3.52968574e-01 9.24862385e-01 1.14521710e-02 -1.39815032e-01
-5.17984331e-01 -7.44165957e-01 -5.28413892e-01 -9.24162805e-01
1.21235510e-03 6.59685433e-01 1.68528154e-01 -1.92754865e-01
3.54339540e-01 3.92683834e-01 -1.82991278e+00 2.13204652e-01
1.25005555e+00 1.40202916e+00 -2.38714516e-02 2.68950224e-01
-4.98008072e-01 8.87299597e-01 -6.46969080e-01 -4.81226653e-01
8.51641372e-02 -1.76525801e-01 -5.78887999e-01 -5.90037694e-03
3.58189911e-01 -9.82018933e-02 -2.03405395e-01 1.16804624e+00
6.52077973e-01 -3.41892801e-02 9.29538369e-01 -1.55774009e+00
-7.82572448e-01 1.70640349e-01 -1.12505686e+00 3.17595005e-01
-3.41804028e-01 1.53221954e-02 7.23981857e-01 -6.59918189e-01
1.10549711e-01 1.32408464e+00 8.09514344e-01 2.48101637e-01
-8.01752925e-01 -9.08990681e-01 4.43934917e-01 5.06199658e-01
-1.42051852e+00 9.78259221e-02 6.15962863e-01 -5.89410007e-01
5.99785984e-01 2.09245175e-01 4.07465518e-01 9.00874913e-01
3.42448771e-01 6.26064122e-01 1.42686784e+00 -8.28951970e-02
7.93784037e-02 1.67878166e-01 2.87871480e-01 2.65587896e-01
4.53185081e-01 2.57086270e-02 -7.50944242e-02 -1.25093572e-02
3.70104790e-01 1.33445680e-01 7.07586333e-02 -2.00434119e-01
-9.45336223e-01 8.58010113e-01 4.53562051e-01 3.06971103e-01
-2.60044873e-01 -1.56011716e-01 6.37607276e-01 6.25402853e-02
7.10076272e-01 1.65834993e-01 -6.13979995e-01 -3.05705547e-01
-7.91107297e-01 3.91621500e-01 6.97717786e-01 7.17852652e-01
7.18656659e-01 4.36510779e-02 -2.74379104e-01 1.08691907e+00
1.58106372e-01 3.82194102e-01 7.63857424e-01 -4.44594413e-01
4.66853678e-01 7.14614213e-01 8.55044127e-02 -1.09596908e+00
-5.63669622e-01 -1.01249385e+00 -1.43918812e+00 7.27220178e-02
3.31733346e-01 -6.43332377e-02 -1.16326737e+00 1.74171436e+00
4.61986989e-01 -2.63997205e-02 -1.30870894e-01 8.53721797e-01
8.69863570e-01 7.18955576e-01 2.12841690e-01 7.47087877e-03
1.20788658e+00 -1.10970283e+00 -7.67920077e-01 -3.69869769e-02
3.49145800e-01 -6.70724630e-01 1.40823090e+00 5.70212305e-01
-6.94303155e-01 -6.62689149e-01 -9.26425278e-01 1.54804885e-01
-6.39861763e-01 7.53564090e-02 6.60362184e-01 5.85221708e-01
-6.10430837e-01 4.08613473e-01 -3.34344059e-01 -5.81690110e-02
6.22139156e-01 1.63406089e-01 -2.80500412e-01 -2.24178731e-01
-1.28305280e+00 8.79152954e-01 4.29845482e-01 5.98119386e-02
-7.21018195e-01 -7.20780969e-01 -4.28815514e-01 1.88349470e-01
2.46205196e-01 -4.40030724e-01 8.26279879e-01 -1.02073324e+00
-1.15262496e+00 6.33762836e-01 3.31305355e-01 -6.12006821e-02
7.00091124e-01 -3.26919019e-01 -6.66917622e-01 -3.41670364e-01
2.30490074e-01 2.78297514e-01 6.57801151e-01 -1.18764949e+00
-6.53287590e-01 -4.90531087e-01 -1.37442857e-01 1.40993625e-01
-5.47321498e-01 -7.82548264e-02 -2.87922233e-01 -6.82583869e-01
-1.94555402e-01 -6.07328832e-01 -1.76566672e-02 -1.51824772e-01
-3.75271767e-01 -5.10346055e-01 8.94283891e-01 -3.78846467e-01
1.48340487e+00 -2.32720947e+00 -1.89817503e-01 -1.01736009e-01
6.51441291e-02 3.84489000e-01 -5.24449274e-02 2.84012944e-01
-2.97013342e-01 2.53133029e-01 -1.90912083e-01 -1.24471858e-01
1.22964822e-01 3.02493095e-01 2.93768402e-02 6.88581645e-01
7.59326965e-02 6.10182762e-01 -7.24769294e-01 -4.93379861e-01
6.55881166e-02 2.70390391e-01 -4.71072853e-01 3.95648032e-01
-2.11711824e-01 3.94260973e-01 -4.81536865e-01 6.48387790e-01
1.26159406e+00 -4.28501815e-01 -2.66793936e-01 -4.27610070e-01
-1.11722194e-01 -1.84007764e-01 -1.28962529e+00 1.31918490e+00
-2.63002038e-01 1.91219509e-01 4.61524986e-02 -1.22334480e+00
9.16451514e-01 -1.16041802e-01 3.93646181e-01 -9.20887351e-01
7.34678358e-02 4.79841828e-01 3.27892244e-01 -7.13884890e-01
3.35309654e-01 -3.70219558e-01 -1.13505945e-01 1.59061164e-01
-6.38184920e-02 7.78800771e-02 1.97583139e-01 -1.69700369e-01
6.57152355e-01 9.62677002e-02 5.03781401e-02 -5.41837037e-01
6.02381229e-01 -1.57824486e-01 6.83281183e-01 4.97370213e-01
-3.63117456e-01 5.52869081e-01 4.84900951e-01 -5.59929371e-01
-1.10137534e+00 -8.62245083e-01 -4.11941946e-01 1.10371494e+00
4.30703849e-01 -2.27202326e-02 -7.22139597e-01 -6.83685422e-01
3.09517235e-01 4.52459335e-01 -6.39553368e-01 -2.61780351e-01
-2.72007585e-01 -1.45890868e+00 6.08224213e-01 5.46600461e-01
1.06515849e+00 -9.04995084e-01 -1.98080674e-01 5.59366941e-02
-1.75772220e-01 -9.55856442e-01 -2.93756902e-01 2.14981541e-01
-6.10478699e-01 -7.74933159e-01 -7.15846598e-01 -5.47624111e-01
4.22953874e-01 -3.28315701e-03 1.30819917e+00 7.42585957e-02
-3.08493465e-01 -3.31603855e-01 -3.26694041e-01 -5.22501111e-01
6.76601753e-02 3.41578424e-01 8.90052989e-02 2.45622486e-01
4.27962035e-01 -4.61914808e-01 -6.40665948e-01 4.88434732e-01
-1.04595423e+00 -2.72703499e-01 6.67737961e-01 9.56910372e-01
5.21983624e-01 6.99960828e-01 1.07812452e+00 -8.40045989e-01
6.73457921e-01 -7.91724563e-01 -3.48016411e-01 1.50309995e-01
-8.76514971e-01 -2.11074680e-01 8.72898161e-01 -4.14842606e-01
-1.25795376e+00 -5.77722907e-01 -3.70536238e-01 -4.66253459e-01
-4.32655036e-01 6.20129347e-01 -3.69699568e-01 1.81246802e-01
5.15680373e-01 2.38665134e-01 -1.96067706e-01 -7.26038516e-01
-2.57939398e-02 1.02684724e+00 3.09897959e-01 -7.83238649e-01
5.70861876e-01 2.68462896e-01 -1.51097775e-01 -4.55650508e-01
-1.18395352e+00 -3.62993866e-01 -4.02774155e-01 1.58817563e-02
8.18432689e-01 -1.11748397e+00 -6.14888012e-01 1.22504437e+00
-7.85090744e-01 -2.39544079e-01 -2.37698555e-01 3.88590187e-01
-2.74209857e-01 1.10483646e-01 -5.82165718e-01 -7.70511448e-01
-2.84880280e-01 -1.05754530e+00 9.57477629e-01 4.64553803e-01
3.17347795e-01 -8.85474026e-01 -6.40163124e-02 3.82981569e-01
7.65478849e-01 2.43954971e-01 1.21229780e+00 -7.02844203e-01
-3.47051769e-01 -2.83222079e-01 -7.60018647e-01 5.79110205e-01
1.56420931e-01 1.22970760e-01 -9.74340260e-01 -2.74002343e-01
-2.79308483e-03 -5.48735082e-01 7.91835904e-01 2.46414214e-01
1.84933460e+00 -1.03300773e-01 -8.23347420e-02 7.12128103e-01
1.59133577e+00 1.65664047e-01 8.67837667e-01 6.17833614e-01
7.73151457e-01 6.06427252e-01 7.54089057e-01 4.36024219e-01
5.89976490e-01 3.92135561e-01 5.82771003e-01 -2.92332560e-01
-2.88383868e-02 -3.12243644e-02 -5.77357896e-02 9.57585871e-01
3.76633257e-02 -3.92266423e-01 -8.14566612e-01 6.18640304e-01
-1.81514490e+00 -7.54549146e-01 -2.58664459e-01 1.88057077e+00
8.40348363e-01 1.64644346e-01 8.76381770e-02 1.78996012e-01
8.10557067e-01 1.31780759e-01 -7.95357764e-01 -3.69392931e-01
-2.44965658e-01 -1.93428904e-01 3.98035556e-01 5.06413840e-02
-1.28097129e+00 5.30380607e-01 6.15280914e+00 1.66770136e+00
-1.21978843e+00 1.18409507e-01 1.23715615e+00 1.52151868e-01
-2.45130941e-01 -6.14839494e-01 -8.53981674e-01 9.51412857e-01
6.88660383e-01 7.98072293e-02 2.77728111e-01 8.64048183e-01
-1.01622306e-01 -1.91794738e-01 -7.72223234e-01 1.18313730e+00
6.21781759e-02 -9.45146680e-01 -8.36737901e-02 1.68627929e-02
7.90007830e-01 8.83179083e-02 1.87023684e-01 6.31691873e-01
2.58155968e-02 -1.29352868e+00 6.87407315e-01 4.61994290e-01
8.80962074e-01 -8.71442080e-01 1.15509236e+00 7.12909400e-01
-9.88420486e-01 -3.02428871e-01 -5.25421023e-01 -1.86336845e-01
-2.15533778e-01 1.07540631e+00 1.32123381e-01 7.17290044e-01
1.15351772e+00 7.14503944e-01 -5.77687144e-01 1.24436760e+00
1.23341501e-01 5.75504720e-01 -3.48221540e-01 -5.24739064e-02
4.14326847e-01 -2.76463419e-01 2.15991318e-01 1.19144642e+00
3.01397473e-01 -1.67144880e-01 3.79850805e-01 6.28281057e-01
-1.76060140e-01 1.27756312e-01 -1.55410737e-01 6.06148951e-02
2.75415957e-01 1.39680004e+00 -7.53987968e-01 -3.46216977e-01
-3.64970028e-01 4.57104713e-01 4.17477518e-01 1.56080395e-01
-1.01781082e+00 -3.34330112e-01 5.93216658e-01 1.24152064e-01
4.22631860e-01 7.21911266e-02 -4.32377070e-01 -1.27399504e+00
8.84286314e-02 -1.08153832e+00 4.93325561e-01 -4.17659193e-01
-2.14041734e+00 7.76112616e-01 2.62781233e-02 -1.25096190e+00
3.10791433e-01 -7.26481974e-01 -5.51205397e-01 8.83675575e-01
-1.75487924e+00 -9.73878980e-01 -7.02485323e-01 7.40361392e-01
4.24757451e-01 -1.81697577e-01 6.76310599e-01 1.01860440e+00
-8.31007421e-01 6.41552448e-01 4.82613504e-01 -8.65392387e-02
8.37804019e-01 -1.19320309e+00 -1.57188680e-02 4.46606189e-01
-5.57570219e-01 2.35790774e-01 6.04685485e-01 -4.53237295e-01
-8.97573411e-01 -1.12057328e+00 4.44058418e-01 -1.94531813e-01
5.86703718e-01 -3.39124113e-01 -9.76286709e-01 1.01685472e-01
1.48891881e-01 3.92621666e-01 5.50235689e-01 1.61819831e-01
-4.18823093e-01 -5.54476202e-01 -1.39210820e+00 1.02327883e-01
7.62007475e-01 -1.70976922e-01 -2.98350722e-01 3.44814897e-01
4.86131072e-01 -4.12179053e-01 -8.42942953e-01 6.77195966e-01
6.22348070e-01 -1.28617775e+00 8.03065360e-01 -7.20571339e-01
5.08186996e-01 -3.85493368e-01 -3.97642672e-01 -1.49966216e+00
-5.00046551e-01 -7.86715245e-04 -4.71365117e-02 1.41293776e+00
7.92243853e-02 -7.72138119e-01 7.86422491e-01 2.42409363e-01
-7.37546682e-02 -1.11517262e+00 -8.39845598e-01 -1.02176273e+00
5.23660481e-01 -8.19536671e-02 9.82284546e-01 8.71567428e-01
-6.63301051e-01 3.64649028e-01 -5.60488343e-01 -2.20112354e-01
5.65636694e-01 3.58890593e-01 6.68258548e-01 -1.46859384e+00
-1.45829543e-01 -5.19353449e-01 -3.41802329e-01 -7.53823519e-01
-1.76987108e-02 -4.62071329e-01 1.43953547e-01 -1.42646587e+00
4.75834817e-01 -9.47342396e-01 -5.17431319e-01 2.67682701e-01
-3.74374449e-01 3.23202729e-01 1.45406844e-02 2.62391806e-01
-7.09450722e-01 7.83917129e-01 1.44756639e+00 -1.59904405e-01
2.24493429e-01 -1.63175371e-02 -9.65244055e-01 8.87656331e-01
8.29276919e-01 -4.49157298e-01 -4.66867596e-01 -4.80676889e-01
1.87398061e-01 -4.43024457e-01 9.68259200e-02 -1.04709840e+00
1.36457505e-02 -2.64315069e-01 5.66278815e-01 -7.05396771e-01
4.90003489e-02 -9.10616279e-01 7.85166249e-02 2.53056705e-01
-5.98760433e-02 1.78342074e-01 2.77693868e-01 4.99476463e-01
-5.60156465e-01 -4.04068977e-02 1.09112525e+00 -1.89939380e-01
-5.66925466e-01 7.72561967e-01 -6.75496981e-02 4.84197199e-01
1.05355334e+00 -1.39844175e-02 -6.59252763e-01 -2.89421946e-01
-3.90521705e-01 2.77051389e-01 2.80862302e-01 3.36345792e-01
1.84819043e-01 -1.41557980e+00 -8.55491221e-01 3.87660354e-01
1.47925407e-01 2.90467381e-01 6.32043540e-01 8.96444976e-01
-4.74018782e-01 1.03844404e-01 -4.65143472e-01 -6.67434931e-01
-7.17418134e-01 6.92104936e-01 5.50743878e-01 -6.77096963e-01
-3.14210117e-01 9.91914153e-01 5.91259122e-01 -7.28264809e-01
2.73924202e-01 -9.22213867e-02 -1.98877260e-01 2.59904951e-01
3.35793912e-01 4.96099919e-01 3.41987461e-01 -6.21696949e-01
-4.85608637e-01 7.32328057e-01 -2.02453405e-01 5.64446151e-01
1.39056098e+00 -2.22254276e-01 -2.90888429e-01 4.56535637e-01
1.33937907e+00 -1.13280930e-01 -1.22508681e+00 8.18316862e-02
-2.28494421e-01 -6.67594075e-01 -1.48807094e-02 -1.17911685e+00
-1.31288803e+00 1.06290817e+00 7.86849618e-01 6.03726506e-01
1.23314059e+00 -2.45902017e-01 8.85389745e-01 -1.58776138e-02
3.11555266e-01 -1.13394356e+00 7.37822875e-02 7.11640835e-01
7.40328133e-01 -1.34625685e+00 3.76583822e-02 -3.40150893e-01
-6.05906665e-01 8.85060191e-01 9.77847397e-01 -3.24626237e-01
9.98173773e-01 4.40705419e-01 9.58421305e-02 -1.51918873e-01
-6.47840321e-01 -4.32071872e-02 2.93015510e-01 5.39569974e-01
4.52830076e-01 3.10583096e-02 -2.51406163e-01 6.86226428e-01
-9.52483341e-02 -2.62642726e-02 1.79322913e-01 5.89986861e-01
-3.16363692e-01 -9.96250212e-01 -4.10566270e-01 8.98217797e-01
-8.18023205e-01 -8.12873151e-03 6.37505725e-02 7.88063109e-01
7.13283479e-01 7.98181653e-01 2.09682629e-01 -2.53520608e-01
3.73344600e-01 -3.14961821e-01 2.78590232e-01 -2.76030719e-01
-4.63929236e-01 4.16339226e-02 -3.09416298e-02 -4.24141109e-01
-4.89593208e-01 -3.40457112e-01 -8.36119771e-01 -5.55010021e-01
-5.01994610e-01 1.84092164e-01 5.19005179e-01 9.69655812e-01
3.85802716e-01 7.60972142e-01 5.93030989e-01 -6.65557742e-01
-7.97259092e-01 -1.14226282e+00 -1.02751291e+00 6.42196596e-01
3.72671425e-01 -1.11303878e+00 -5.86052537e-01 -3.62425029e-01] | [9.124969482421875, 3.8758511543273926] |
d02d9690-f228-461d-a11f-32ddead5e2e9 | improved-deep-spectral-convolution-network | 1808.01104 | null | https://arxiv.org/abs/1808.01104v4 | https://arxiv.org/pdf/1808.01104v4.pdf | Improved Deep Spectral Convolution Network For Hyperspectral Unmixing With Multinomial Mixture Kernel and Endmember Uncertainty | In this study, we propose a novel framework for hyperspectral unmixing by using an improved deep spectral convolution network (DSCN++) combined with endmember uncertainty. DSCN++ is used to compute high-level representations which are further modeled with Multinomial Mixture Model to estimate abundance maps. In the reconstruction step, a new trainable uncertainty term based on a nonlinear neural network model is introduced to provide robustness to endmember uncertainty. For the optimization of the coefficients of the multinomial model and the uncertainty term, Wasserstein Generative Adversarial Network (WGAN) is exploited to improve stability and to capture uncertainty. Experiments are performed on both real and synthetic datasets. The results validate that the proposed method obtains state-of-the-art hyperspectral unmixing performance particularly on the real datasets compared to the baseline techniques. | ['Savas Ozkan', 'Gozde Bozdagi Akar'] | 2018-08-03 | null | null | null | null | ['hyperspectral-unmixing'] | ['computer-vision'] | [ 4.49879616e-01 -4.11346555e-01 2.63407350e-01 -1.31730869e-01
-6.61087394e-01 -4.71752584e-01 6.55969024e-01 -2.23735943e-01
-1.94961995e-01 9.45388854e-01 9.89459381e-02 -1.38475567e-01
-1.20866254e-01 -9.20175493e-01 -8.90167058e-01 -1.02614808e+00
8.01539496e-02 1.96219534e-01 -5.05602002e-01 1.21888414e-01
-2.51593977e-01 4.47047621e-01 -1.33834386e+00 -6.76727220e-02
1.47407401e+00 1.09382784e+00 2.48335060e-02 4.22024995e-01
2.50927415e-02 7.31936872e-01 -3.52884084e-01 -3.92581010e-03
7.79633820e-01 -2.70031571e-01 -2.66936868e-01 2.19649687e-01
4.30276155e-01 -5.35334170e-01 -4.26845580e-01 1.67765796e+00
4.83517468e-01 5.42737365e-01 1.03480518e+00 -9.57445741e-01
-5.79625070e-01 8.26309144e-01 -6.83874369e-01 -2.52466947e-01
-4.29391950e-01 2.16950849e-01 6.21981084e-01 -7.51890242e-01
6.26506582e-02 1.24845338e+00 8.08704913e-01 -4.13047895e-02
-1.40505576e+00 -9.52803969e-01 -9.38461274e-02 -1.36972442e-01
-1.62750232e+00 -1.73337877e-01 6.94585621e-01 -5.20437062e-01
5.19270957e-01 1.81429029e-01 4.02883440e-01 7.06054449e-01
2.66679302e-02 5.07483661e-01 1.38690209e+00 -3.37772369e-01
2.40983844e-01 -1.19308658e-01 1.37537986e-01 5.17348468e-01
3.08207989e-01 5.42683303e-01 -8.68244395e-02 -2.97665238e-01
8.47748041e-01 2.31371924e-01 -4.92847919e-01 -1.72507390e-01
-9.10121024e-01 8.71834934e-01 8.76907706e-01 -6.86070696e-02
-5.34604371e-01 2.26254955e-01 -7.24847838e-02 1.02090321e-01
8.31762314e-01 3.50937277e-01 -1.85347632e-01 5.95803678e-01
-1.12388372e+00 1.47680119e-01 5.18368244e-01 6.77375913e-01
1.15285349e+00 6.03965700e-01 -2.58216769e-01 9.66384470e-01
9.20908988e-01 1.05015683e+00 4.68121231e-01 -8.28586042e-01
2.25772843e-01 4.87165838e-01 3.18643808e-01 -7.61678934e-01
-2.01149881e-01 -7.86708713e-01 -1.28722501e+00 5.62486112e-01
7.38709643e-02 -3.77049685e-01 -1.34749138e+00 1.53237844e+00
1.87989101e-01 9.03142273e-01 3.71229619e-01 9.04891193e-01
6.06055021e-01 1.08707881e+00 2.54873365e-01 -1.90330505e-01
8.98266196e-01 -9.58306313e-01 -7.22282350e-01 -1.18196554e-01
-2.78882347e-02 -5.72654784e-01 5.76050043e-01 2.39117399e-01
-6.72893882e-01 -5.41733861e-01 -1.38705325e+00 2.44989946e-01
-4.55408961e-01 2.42979974e-01 5.41269839e-01 6.59920871e-01
-5.50309896e-01 8.87509465e-01 -9.60352957e-01 2.44644746e-01
5.02689481e-01 2.28806138e-01 -1.09991729e-01 1.90449152e-02
-1.35863495e+00 6.97200775e-01 7.54157186e-01 5.51856637e-01
-1.35643816e+00 -7.36928821e-01 -1.00771940e+00 4.64601777e-02
-2.72140559e-02 -7.16916025e-01 7.89883971e-01 -1.21914697e+00
-1.75455654e+00 2.78342426e-01 3.92952859e-01 -5.45966804e-01
4.50032234e-01 -3.14422965e-01 -5.42372525e-01 1.10035144e-01
-2.12388664e-01 5.17523706e-01 1.38208795e+00 -1.49836576e+00
-1.47558495e-01 -1.73493147e-01 -3.56035054e-01 3.22299808e-01
-2.69827813e-01 -3.39765280e-01 2.69802064e-01 -9.92433608e-01
3.28480512e-01 -7.50730813e-01 -2.52259165e-01 -8.83162618e-02
-3.71456146e-01 4.00692850e-01 6.73480630e-01 -1.10800564e+00
1.00383258e+00 -2.16231942e+00 3.40190321e-01 5.05679369e-01
-4.56090048e-02 4.38038379e-01 -2.79708564e-01 6.94677904e-02
-3.01029980e-01 9.51878726e-02 -1.12997806e+00 -4.44262445e-01
1.43451482e-01 1.60643235e-01 -3.41855168e-01 6.50370657e-01
3.33281279e-01 7.77621806e-01 -9.03972268e-01 9.13030580e-02
5.76734602e-01 7.50326216e-01 6.72714459e-03 4.22933608e-01
-6.58918917e-01 4.89199400e-01 7.23695476e-03 6.62477195e-01
1.30117536e+00 -7.14930594e-02 -1.89197864e-02 -3.11078936e-01
-9.56683129e-04 -4.27810371e-01 -1.48674965e+00 1.67034161e+00
-2.80881733e-01 1.12328008e-02 3.37907672e-01 -6.94523513e-01
7.56580114e-01 1.58299282e-01 6.09385297e-02 2.03612432e-01
2.71481067e-01 2.69828916e-01 -1.18893728e-01 -1.82628185e-01
3.65846097e-01 -5.38537383e-01 5.04688263e-01 2.58987397e-01
1.80940479e-01 -5.47290087e-01 -2.16823205e-01 -1.86263457e-01
2.86701381e-01 5.20143628e-01 3.87726612e-02 -5.66174090e-01
6.89044178e-01 -1.82605848e-01 2.85643756e-01 3.54571253e-01
-2.59712432e-03 5.95358968e-01 -6.58485368e-02 -9.44200009e-02
-9.71465707e-01 -1.03609431e+00 -4.32370275e-01 6.60472155e-01
8.52757022e-02 4.49430406e-01 -6.45574152e-01 -3.83933008e-01
1.61412239e-01 8.28272760e-01 -5.36601841e-01 -4.07853164e-03
2.82251149e-01 -1.49351668e+00 5.86366892e-01 3.95650864e-01
7.75674939e-01 -7.37796426e-01 1.52030215e-01 1.13936499e-01
-2.13752706e-02 -6.79459155e-01 -2.99201429e-01 2.88650423e-01
-8.93790543e-01 -8.89304817e-01 -7.22919762e-01 -2.02430278e-01
5.62500536e-01 -2.54502930e-02 5.16248941e-01 -6.59546971e-01
-7.66416267e-02 -5.38562648e-02 -2.60394305e-01 -6.29628301e-01
-6.03116632e-01 -2.28485048e-01 4.82141748e-02 4.99487221e-01
1.87150702e-01 -9.54628229e-01 -5.06536841e-01 -1.40422761e-01
-1.66062701e+00 -1.38547286e-01 6.51717246e-01 9.52195525e-01
6.81362152e-01 3.98668170e-01 5.35029650e-01 -6.90606058e-01
3.48822504e-01 -7.48296618e-01 -9.61265326e-01 1.79746449e-01
-7.38618791e-01 2.69984156e-01 4.15255308e-01 -5.56447089e-01
-1.32783365e+00 6.02451526e-02 9.63786170e-02 -7.41450369e-01
-4.65847217e-02 8.26890111e-01 -2.42282048e-01 -1.82058647e-01
7.66393840e-01 2.64161319e-01 1.91541106e-01 -4.45265383e-01
4.30201739e-01 8.14818740e-01 6.83953881e-01 -3.55224907e-01
1.12583971e+00 5.44639349e-01 1.16080277e-01 -8.66646111e-01
-7.88548887e-01 -3.48275393e-01 -5.39500654e-01 2.35145018e-02
6.52650535e-01 -1.43847048e+00 -1.09187188e-02 1.09011233e+00
-8.27596009e-01 -4.65381831e-01 -3.02967101e-01 6.51480079e-01
-2.68213004e-01 6.55494392e-01 -5.98053873e-01 -1.08122063e+00
-7.20558643e-01 -1.04110706e+00 8.77631664e-01 4.23219740e-01
4.98082668e-01 -1.02157116e+00 8.76264423e-02 2.24482223e-01
5.31359911e-01 6.66736603e-01 7.47949898e-01 -2.94161886e-01
-5.28062403e-01 1.90549984e-03 -4.73095536e-01 1.19532537e+00
1.49281353e-01 7.00536445e-02 -1.23292267e+00 -2.81065106e-01
6.39104545e-02 -3.37817013e-01 1.30334127e+00 6.12281561e-01
1.04326510e+00 -2.02218622e-01 8.38613603e-03 1.03569436e+00
1.78152275e+00 -1.35238186e-01 9.12643194e-01 6.87405542e-02
7.55676687e-01 1.53089508e-01 1.87638044e-01 5.10286033e-01
-8.32086429e-02 -2.05943838e-01 9.54091728e-01 -1.24316690e-02
1.20848864e-01 -1.05612837e-01 3.40885788e-01 6.01016045e-01
-5.83402961e-02 -2.25411400e-01 -5.97986639e-01 2.67216951e-01
-1.84857035e+00 -8.41311991e-01 -2.11745590e-01 2.31813693e+00
8.56772006e-01 -3.46535653e-01 -4.67275113e-01 -1.26082990e-02
8.37461591e-01 3.77866864e-01 -7.38782823e-01 3.71434361e-01
-5.00578165e-01 4.19763923e-01 7.15574980e-01 8.13878655e-01
-1.50325525e+00 8.16254437e-01 5.94061708e+00 7.86085367e-01
-1.20881021e+00 1.78882912e-01 5.88805974e-01 2.61182994e-01
-3.94311905e-01 -2.60875911e-01 -2.50191301e-01 4.62676108e-01
7.41281986e-01 2.82899290e-01 8.52425039e-01 5.06182373e-01
1.47010043e-01 3.41759026e-02 -3.74404907e-01 8.81421387e-01
9.65807587e-02 -7.85955310e-01 1.83241233e-01 -1.90665618e-01
1.29612851e+00 4.32161868e-01 1.46194711e-01 1.74219713e-01
4.79725897e-01 -1.26008165e+00 4.91471797e-01 1.07440412e+00
9.69640434e-01 -1.03954613e+00 9.04408753e-01 2.91829169e-01
-9.28985238e-01 -4.34907898e-02 -7.13987827e-01 -3.24131548e-02
-7.97817856e-02 8.60172391e-01 -3.45148355e-01 9.60770190e-01
2.09120840e-01 7.18818903e-01 -2.78139919e-01 8.28704655e-01
-3.67382884e-01 6.28984392e-01 -5.33900321e-01 4.08116043e-01
2.73482263e-01 -8.24788511e-01 6.52361929e-01 9.97732222e-01
5.98291636e-01 1.33475572e-01 8.94541964e-02 1.24536681e+00
-3.48092318e-01 5.45515027e-03 -3.15417916e-01 -4.08497870e-01
3.46420139e-01 1.50444949e+00 -1.95736170e-01 -3.51144820e-01
-1.83757097e-02 1.02984989e+00 1.65891889e-02 6.49695575e-01
-8.13171029e-01 -2.74913579e-01 6.60484552e-01 -2.60795444e-01
1.29821673e-01 -7.65011013e-02 7.57544208e-03 -1.40803611e+00
-4.18601513e-01 -9.10588682e-01 2.24596485e-01 -1.02655840e+00
-1.64489317e+00 5.55068791e-01 -1.35963202e-01 -1.12433732e+00
3.96536514e-02 -9.31098998e-01 -5.91352224e-01 1.40544379e+00
-1.99871874e+00 -1.49288177e+00 -6.21894062e-01 3.97788614e-01
4.28053215e-02 -1.58111438e-01 9.49883461e-01 -5.17322170e-03
-7.18950570e-01 -2.27691252e-02 8.11644256e-01 -1.18142202e-01
6.93964899e-01 -1.34746563e+00 -2.07493559e-01 1.23365271e+00
-2.46033609e-01 4.53599632e-01 5.86467922e-01 -7.89818943e-01
-1.06290650e+00 -1.79836702e+00 3.43647189e-02 -6.27373531e-02
7.98928797e-01 1.93370178e-01 -8.76921296e-01 6.63713813e-01
2.94787407e-01 2.52780259e-01 8.93112957e-01 -5.25424004e-01
-4.74735379e-01 -1.74750790e-01 -1.42868423e+00 1.09855406e-01
3.95612478e-01 -5.35529494e-01 -4.04988438e-01 2.53486603e-01
8.39242399e-01 -2.47716919e-01 -1.07292378e+00 8.35743606e-01
5.85969031e-01 -6.33290052e-01 8.79652500e-01 -4.21370208e-01
6.01614594e-01 -7.10210860e-01 -4.59656924e-01 -1.70615101e+00
-3.81971121e-01 -5.82187831e-01 -5.53341568e-01 1.01880610e+00
3.73103023e-01 -7.00379014e-01 3.12891632e-01 2.65275508e-01
-2.51662374e-01 -1.59504071e-01 -5.31715691e-01 -5.94845355e-01
2.33199582e-01 -3.06680501e-01 9.05163884e-01 1.09206569e+00
-4.72913235e-01 6.05688803e-02 -5.30069351e-01 9.39498127e-01
9.51581359e-01 -7.20631927e-02 3.26885343e-01 -1.35199344e+00
-4.29252326e-01 -3.43795002e-01 1.23174181e-02 -6.06123984e-01
4.77503896e-01 -9.31546867e-01 1.87850058e-01 -1.24087441e+00
8.94980505e-02 -1.84057921e-01 -7.68337667e-01 2.12083116e-01
-5.46324074e-01 4.35445875e-01 -1.11661345e-01 2.18538359e-01
1.08861491e-01 8.68141592e-01 9.74855840e-01 -6.57515049e-01
-2.72924721e-01 -3.62956747e-02 -5.12658417e-01 6.72306001e-01
8.89730513e-01 -2.41645619e-01 -2.83235162e-01 -4.11038816e-01
1.80260818e-02 -1.12138957e-01 4.43267465e-01 -1.23066044e+00
-1.86945930e-01 -1.93535432e-01 5.62254548e-01 -5.24672687e-01
3.29035610e-01 -8.89167845e-01 4.99167562e-01 3.57485741e-01
-8.50753337e-02 -7.05959737e-01 4.78244990e-01 8.29632401e-01
-2.05488265e-01 -4.22103167e-01 1.11619556e+00 -4.34000343e-02
-3.65433782e-01 6.57320619e-01 -5.34185953e-02 -5.84620357e-01
7.22697198e-01 3.77421230e-01 -2.53982067e-01 -1.95156246e-01
-7.77968407e-01 2.44064242e-01 3.23915631e-01 -1.68883771e-01
4.24635291e-01 -1.39306879e+00 -1.08311903e+00 3.32072616e-01
1.62749901e-01 1.89089239e-01 4.18167919e-01 5.78109980e-01
-7.11620867e-01 -3.20436954e-02 -1.22190587e-01 -3.46208751e-01
-6.20261788e-01 4.65462565e-01 8.05675685e-01 -3.81030619e-01
1.35090694e-01 9.14161563e-01 -9.64215174e-02 -9.59216177e-01
3.86044309e-02 -4.98306721e-01 -1.18728280e-01 1.45875756e-02
6.71281934e-01 6.46952152e-01 -2.09140331e-01 -6.07424021e-01
8.53067636e-02 -5.83198480e-03 6.14859402e-01 -1.52479365e-01
1.41240954e+00 -1.23202763e-01 -7.41307378e-01 4.08103228e-01
9.45546806e-01 -1.48552552e-01 -1.65959859e+00 -3.14970195e-01
-5.38219631e-01 -2.29379311e-01 5.79729915e-01 -1.09971154e+00
-8.83929133e-01 9.71472323e-01 1.00188625e+00 -8.90033022e-02
1.13741136e+00 -7.68734574e-01 3.23991865e-01 3.15186560e-01
-1.17092133e-01 -7.99423754e-01 -2.47272491e-01 5.35110295e-01
9.51525033e-01 -1.44171679e+00 -2.56050751e-03 -1.58653975e-01
-2.12477341e-01 1.18364751e+00 3.97058398e-01 -2.59850293e-01
8.70070100e-01 1.36594921e-01 1.01137020e-01 2.06055418e-01
2.73088545e-01 -1.47780061e-01 4.43110585e-01 5.14288902e-01
1.59377635e-01 4.31731969e-01 1.36196434e-01 5.91452599e-01
3.50467712e-01 -1.99117251e-02 2.68120944e-01 3.65571976e-01
-6.63770795e-01 -8.92270029e-01 -7.55205452e-01 5.53485334e-01
-2.97924221e-01 -5.78169882e-01 3.05182133e-02 1.94434181e-01
3.34950805e-01 8.70787501e-01 -1.19145922e-01 -2.55512923e-01
-2.11125642e-01 5.09310246e-01 2.52695411e-01 -5.90357721e-01
-2.66600490e-01 2.74942040e-01 -3.11964035e-01 -1.21017233e-01
-6.64890766e-01 -4.45607990e-01 -9.59774435e-01 -9.89187807e-02
-6.97821856e-01 1.36523709e-01 7.03716338e-01 7.74799466e-01
1.85628325e-01 6.06248796e-01 5.75365663e-01 -1.11096859e+00
-1.19883645e+00 -1.71291959e+00 -1.21978247e+00 1.92963108e-01
4.13415343e-01 -4.04359907e-01 -5.49556017e-01 4.10686582e-02] | [10.013647079467773, -1.9585403203964233] |
73aafe54-c115-4b12-b2d0-8c0cb45ef7f5 | ap-10k-a-benchmark-for-animal-pose-estimation | 2108.12617 | null | https://arxiv.org/abs/2108.12617v2 | https://arxiv.org/pdf/2108.12617v2.pdf | AP-10K: A Benchmark for Animal Pose Estimation in the Wild | Accurate animal pose estimation is an essential step towards understanding animal behavior, and can potentially benefit many downstream applications, such as wildlife conservation. Previous works only focus on specific animals while ignoring the diversity of animal species, limiting the generalization ability. In this paper, we propose AP-10K, the first large-scale benchmark for mammal animal pose estimation, to facilitate the research in animal pose estimation. AP-10K consists of 10,015 images collected and filtered from 23 animal families and 54 species following the taxonomic rank and high-quality keypoint annotations labeled and checked manually. Based on AP-10K, we benchmark representative pose estimation models on the following three tracks: (1) supervised learning for animal pose estimation, (2) cross-domain transfer learning from human pose estimation to animal pose estimation, and (3) intra- and inter-family domain generalization for unseen animals. The experimental results provide sound empirical evidence on the superiority of learning from diverse animals species in terms of both accuracy and generalization ability. It opens new directions for facilitating future research in animal pose estimation. AP-10k is publicly available at https://github.com/AlexTheBad/AP10K. | ['DaCheng Tao', 'Ziyu Guan', 'Wei Zhao', 'Jing Zhang', 'Yufei Xu', 'Hang Yu'] | 2021-08-28 | null | null | null | null | ['animal-pose-estimation'] | ['computer-vision'] | [-2.15392470e-01 -3.17031056e-01 -4.51459199e-01 -5.48346162e-01
-5.49265087e-01 -6.90529346e-01 1.61127582e-01 1.69877350e-01
-7.49457359e-01 7.87558794e-01 8.33631083e-02 2.92180687e-01
-4.59422916e-02 -5.64141452e-01 -1.07872558e+00 -4.17488307e-01
-7.12384939e-01 5.14105856e-01 3.26486528e-01 -5.39084002e-02
1.91070866e-02 6.43647373e-01 -1.37718487e+00 1.10604256e-01
3.97119075e-01 1.01330805e+00 1.48060232e-01 5.09472668e-01
5.80375850e-01 2.80438483e-01 -3.29610288e-01 -6.11275613e-01
2.71856785e-01 -1.96311489e-01 -6.51872754e-01 -3.33156407e-01
6.46271110e-01 -6.53382540e-01 -2.83276260e-01 8.30130577e-01
2.85101205e-01 2.11317483e-02 7.07746863e-01 -1.41246569e+00
-5.03786385e-01 5.63964307e-01 -8.59872639e-01 4.02035952e-01
1.38038263e-01 3.75111610e-01 9.94753122e-01 -5.41239440e-01
5.84001005e-01 1.41488159e+00 1.20295870e+00 6.79950058e-01
-1.44356394e+00 -1.27533853e+00 6.87608793e-02 4.10413355e-01
-1.39434087e+00 -2.73655713e-01 3.95202368e-01 -7.33097017e-01
5.36034107e-01 2.08884198e-02 8.34464908e-01 1.25780606e+00
3.32153797e-01 8.38289976e-01 1.00810277e+00 3.55201513e-01
4.24942411e-02 -1.74094558e-01 1.69080496e-01 5.05861580e-01
5.27037084e-01 4.17739928e-01 -7.00003922e-01 -3.32923293e-01
8.46098900e-01 -9.52265039e-02 -1.63206533e-01 -7.57716894e-01
-1.42799437e+00 9.40483272e-01 1.13809741e+00 -4.21944827e-01
-2.48727411e-01 3.91597867e-01 5.52114904e-01 1.83821350e-01
4.14612055e-01 7.87161291e-01 -8.09899271e-01 1.65605605e-01
-6.00957274e-01 6.37013137e-01 7.38606930e-01 1.09543204e+00
5.30193090e-01 -2.52222210e-01 -2.04673540e-02 9.96384799e-01
5.32481015e-01 1.06720686e+00 2.39537567e-01 -7.06739545e-01
3.98020595e-01 2.41151556e-01 1.76195309e-01 -1.11416221e+00
-8.65601957e-01 -4.69458431e-01 -5.31035125e-01 -1.28601387e-01
4.13577735e-01 2.01088879e-02 -7.38729179e-01 1.95279801e+00
5.51930308e-01 8.55287686e-02 -1.81559965e-01 9.68442202e-01
1.35636210e+00 7.21241057e-01 4.03269678e-01 2.19147220e-01
1.62430835e+00 -9.61297214e-01 -8.67218152e-02 -5.82760751e-01
6.06447697e-01 -5.48352659e-01 6.56512558e-01 1.25919238e-01
-4.84132230e-01 -3.18111360e-01 -9.82559144e-01 1.26737386e-01
-3.64196956e-01 4.49621946e-01 7.12048352e-01 2.47410029e-01
-6.17012739e-01 4.32940543e-01 -1.06197643e+00 -5.41727722e-01
6.95577204e-01 3.97312224e-01 -7.12707460e-01 1.51964143e-01
-1.03350782e+00 9.33198631e-01 4.42934364e-01 2.54015207e-01
-1.13676572e+00 -8.85232568e-01 -9.34704959e-01 -4.38317865e-01
2.61950463e-01 -3.12857687e-01 1.51447403e+00 -5.52153289e-01
-9.69944417e-01 1.23867142e+00 4.12404388e-01 -7.92703211e-01
4.83361751e-01 -6.76872790e-01 -1.52097210e-01 7.65640140e-02
4.46933866e-01 1.33993244e+00 7.20371187e-01 -1.12248158e+00
-8.29068244e-01 -6.45152748e-01 -5.43870591e-02 1.12081870e-01
-4.61759754e-02 -1.06369674e-01 -3.31941128e-01 -1.03007257e+00
6.66956827e-02 -1.39100337e+00 -1.71268970e-01 6.08537316e-01
1.58369616e-02 -4.95877154e-02 5.30714691e-01 -8.36062610e-01
9.56552565e-01 -2.02081537e+00 3.26777250e-01 -1.74032688e-01
5.45011722e-02 1.16870597e-01 -3.20583791e-01 4.07106519e-01
-3.19821499e-02 -3.18696886e-01 -2.08759069e-01 1.00523047e-01
-2.26008862e-01 9.78932455e-02 -1.53113261e-01 9.52810585e-01
1.68581739e-01 9.66262937e-01 -9.02350724e-01 -5.63320577e-01
6.35972619e-02 7.73304626e-02 -9.61224377e-01 1.10206328e-01
-5.80883957e-02 5.61707497e-01 -5.64113081e-01 8.14244151e-01
7.80226111e-01 1.90956180e-03 -2.69873883e-03 -4.64232892e-01
2.24591315e-01 -1.36472717e-01 -7.06020057e-01 1.41246557e+00
-2.89384604e-01 3.72194439e-01 1.41657278e-01 -7.99958527e-01
5.94975054e-01 -3.33353400e-01 5.26456714e-01 -4.62049007e-01
1.93982780e-01 2.33996615e-01 1.54314414e-01 -2.36132666e-01
4.55141157e-01 -4.90604341e-02 -5.88850141e-01 -1.70910563e-02
3.52828711e-01 -2.27533579e-01 2.70053566e-01 -4.61795591e-02
6.26342297e-01 2.88625687e-01 7.22772837e-01 -5.79282224e-01
1.77297011e-01 1.72456011e-01 4.82008815e-01 5.32656491e-01
-5.72600245e-01 1.89543650e-01 2.45217130e-01 -5.02881348e-01
-7.25603044e-01 -1.21416342e+00 -6.00994945e-01 1.56947315e+00
6.20275676e-01 -3.42709661e-01 -7.42908359e-01 -7.00795710e-01
7.21365929e-01 3.36397320e-01 -1.05561900e+00 -4.86163467e-01
-7.28208303e-01 -7.50882983e-01 6.54705107e-01 9.77976382e-01
4.40319657e-01 -1.40210211e+00 -6.89123333e-01 -6.93428367e-02
-3.64929080e-01 -1.09400213e+00 -4.69126612e-01 2.46414870e-01
-7.77871847e-01 -1.04271281e+00 -6.69040978e-01 -8.46009314e-01
4.24768746e-01 2.91578531e-01 9.43661988e-01 -1.35273010e-01
-4.28724200e-01 2.07922429e-01 -4.74621952e-01 -4.02607352e-01
-1.78512841e-01 1.98343948e-01 5.01612961e-01 -4.91744936e-01
3.32467437e-01 -3.14137518e-01 -7.56745696e-01 9.11849201e-01
-4.65079546e-01 -2.08632424e-01 7.94705153e-01 7.22102821e-01
6.79234207e-01 -5.06943583e-01 5.22814214e-01 -5.78952789e-01
-5.62680066e-02 -7.55412579e-01 -7.18001068e-01 -3.74832153e-02
2.74863243e-01 -1.33898377e-01 4.36387241e-01 -5.91591775e-01
-6.31681681e-01 2.96765208e-01 -2.31788889e-01 6.08226769e-02
-1.70698389e-01 2.60529011e-01 -3.16097923e-02 -2.88047314e-01
7.37240672e-01 -4.16049324e-02 -8.55178535e-02 -7.08408475e-01
1.52882978e-01 4.01667058e-01 7.47312427e-01 -5.87292910e-01
9.05747652e-01 4.35004056e-01 -2.00018287e-02 -1.02520442e+00
-8.42334986e-01 -5.80972850e-01 -7.08156228e-01 -3.33734185e-01
1.12391901e+00 -1.19324660e+00 -6.72869265e-01 7.47669101e-01
-6.62298679e-01 -4.89597201e-01 1.02502123e-01 8.23027551e-01
-6.75537407e-01 2.19331354e-01 -6.00960493e-01 -2.15557396e-01
-4.07849491e-01 -1.08919799e+00 1.44505668e+00 1.10845633e-01
-3.40026766e-01 -3.76222104e-01 -2.25884262e-02 5.31182110e-01
2.20171064e-02 2.06952497e-01 4.32793081e-01 -6.18000805e-01
-4.15138274e-01 -4.37332034e-01 -2.68467069e-01 2.43102938e-01
-1.56820893e-01 -1.54863924e-01 -4.93514150e-01 -7.01319933e-01
-5.15276313e-01 -7.72641540e-01 8.11389923e-01 5.65833628e-01
1.23192620e+00 -4.14836034e-02 -6.57986522e-01 8.13988149e-01
1.02964056e+00 -9.98096243e-02 1.29437625e-01 3.75202298e-01
6.40534461e-01 5.87852895e-01 1.10602665e+00 3.05118978e-01
3.70427996e-01 9.72078681e-01 6.42810166e-01 1.61581203e-01
-2.62030419e-02 -6.72494292e-01 2.11414337e-01 3.82469267e-01
-7.40150586e-02 -8.98356065e-02 -1.04659295e+00 5.41768789e-01
-1.65117240e+00 -8.38140249e-01 1.28167987e-01 2.15213346e+00
5.72748423e-01 -6.12767227e-02 5.58035254e-01 -3.74230146e-01
6.56697512e-01 -1.32090542e-02 -7.04406500e-01 1.34295508e-01
2.66626209e-01 -8.91556740e-02 1.13888288e+00 -3.69729176e-02
-1.76089370e+00 9.84465539e-01 6.34246445e+00 1.01688504e+00
-1.04103780e+00 -1.16540842e-01 2.68699437e-01 -2.84381844e-02
5.03735840e-01 -5.22933006e-01 -1.19056964e+00 5.19283414e-01
4.05726343e-01 5.10795228e-02 1.09172113e-01 1.44264841e+00
-2.14088455e-01 -1.44843251e-01 -1.34337258e+00 9.26039755e-01
1.17987931e-01 -6.07517540e-01 -1.06513351e-01 5.44953458e-02
5.68484485e-01 3.12837809e-01 4.00325842e-02 4.90642399e-01
1.99585646e-01 -1.07390738e+00 9.94795382e-01 1.25792116e-01
5.40702879e-01 -7.27070451e-01 8.75457168e-01 4.85162407e-01
-1.56332231e+00 -4.92269248e-02 -4.03348833e-01 1.75734758e-01
-3.56824137e-02 -3.49549145e-01 -5.14603019e-01 2.35034525e-01
1.29509723e+00 9.40471888e-01 -9.17730689e-01 1.36261988e+00
-4.07717191e-02 5.94463050e-01 -6.93198323e-01 -1.50379077e-01
1.98905826e-01 6.94204718e-02 6.09149933e-01 1.21591246e+00
2.50360016e-02 9.21991765e-02 3.88797820e-01 2.42639840e-01
-1.30321309e-01 2.14918792e-01 -2.17333972e-01 1.29865902e-02
4.95058358e-01 1.19082236e+00 -1.09875870e+00 -1.82417023e-03
-2.40976453e-01 6.75579727e-01 3.05110127e-01 -2.07349345e-01
-1.38356364e+00 7.56749064e-02 8.30941558e-01 2.83135235e-01
5.59762657e-01 -9.02667940e-02 1.16184875e-01 -9.56242025e-01
-2.43021660e-02 -9.24707830e-01 6.12647474e-01 -4.37925160e-01
-1.54591763e+00 3.10469836e-01 6.98451817e-01 -1.55996168e+00
1.76792592e-01 -8.84829879e-01 2.10573703e-01 1.88144132e-01
-1.12103236e+00 -1.57220757e+00 -3.86957586e-01 9.56727192e-02
5.20706773e-01 4.75180037e-02 5.44380069e-01 4.34889942e-01
-1.39862776e-01 8.15039933e-01 2.12853357e-01 2.04685479e-01
8.25276017e-01 -7.54585207e-01 5.07075131e-01 3.90237302e-01
-1.73603967e-01 4.02367830e-01 7.11024582e-01 -7.77922094e-01
-1.30686259e+00 -1.39211380e+00 2.17725605e-01 -7.09727466e-01
7.74013400e-01 -4.78039652e-01 -6.76513195e-01 8.52592230e-01
-6.24667704e-01 1.79126903e-01 6.05973363e-01 -6.68672845e-02
-2.86812127e-01 -2.03328028e-01 -1.15274858e+00 4.32125032e-01
1.32196867e+00 1.71028487e-02 -6.48977757e-01 3.72828990e-01
5.58042586e-01 -6.62317038e-01 -9.18277264e-01 8.94555569e-01
1.10106075e+00 -4.33661163e-01 1.27288270e+00 -7.22714067e-01
8.18056881e-01 -3.72731537e-01 -3.44138205e-01 -1.17540479e+00
-5.40843606e-01 4.81249124e-01 1.99821308e-01 8.33041489e-01
2.57757634e-01 -4.29415286e-01 6.78220451e-01 -1.38123617e-01
1.02710791e-01 -6.71661377e-01 -8.41514707e-01 -1.09819293e+00
3.25183898e-01 -1.77268207e-01 4.14736509e-01 5.95822334e-01
-6.32184923e-01 3.01797420e-01 -8.39174390e-01 3.49340886e-01
8.12398970e-01 4.04287010e-01 1.15895069e+00 -1.32259905e+00
-3.36685359e-01 -3.41334432e-01 -9.45382118e-01 -1.47709835e+00
1.79900780e-01 -7.95487761e-01 4.95366186e-01 -1.14225173e+00
4.32390511e-01 -1.70586884e-01 9.26150680e-02 6.37466252e-01
-1.02096327e-01 7.96301365e-01 4.02130187e-03 2.60458410e-01
-7.59104729e-01 4.82951313e-01 1.25616086e+00 5.77908643e-02
1.29511714e-01 2.67260492e-01 -4.39841658e-01 1.09270990e+00
5.72627544e-01 -5.56182623e-01 -1.65175170e-01 -2.80590087e-01
-1.16081953e-01 -1.12202317e-02 8.10636163e-01 -1.16377795e+00
-2.37697959e-01 -3.04799020e-01 4.89459157e-01 -8.89177322e-01
4.55395818e-01 -9.30658340e-01 2.13104323e-01 9.38863575e-01
-2.73040593e-01 -2.71886081e-01 4.50550377e-01 9.20265734e-01
7.95956179e-02 6.96061254e-02 1.23846233e+00 -3.15503031e-02
-1.15133560e+00 5.06799877e-01 -1.91717699e-01 3.36216569e-01
1.33419418e+00 -8.17250907e-02 -2.35648796e-01 -4.97188196e-02
-5.74471474e-01 5.71802318e-01 3.87774825e-01 8.02257359e-01
2.08582535e-01 -1.25246763e+00 -8.95825088e-01 9.34063345e-02
7.07012832e-01 -2.77169734e-01 3.83090496e-01 7.08924472e-01
-1.03007495e+00 3.55090380e-01 -6.46970153e-01 -9.53243732e-01
-1.70403779e+00 3.83397460e-01 2.79146194e-01 5.12033105e-02
-5.90299487e-01 1.33281946e+00 8.05556893e-01 -7.81792939e-01
2.45561898e-01 -4.40920860e-01 -1.57279134e-01 7.90666044e-02
3.51128012e-01 3.17252398e-01 -1.30778983e-01 -1.23821270e+00
-8.39751840e-01 8.43104243e-01 -4.24414314e-02 4.49770749e-01
1.54035068e+00 9.34176445e-02 1.40812680e-01 9.19837356e-02
1.18238330e+00 -3.10458362e-01 -1.16786933e+00 -2.70596892e-01
-5.45160435e-02 -5.30750632e-01 -3.36400032e-01 -6.49159312e-01
-9.27927673e-01 5.85017860e-01 6.47405088e-01 -3.60197544e-01
7.98337221e-01 4.34370309e-01 7.79136717e-01 5.04468322e-01
7.43291378e-01 -9.30321515e-01 1.11171573e-01 2.81068027e-01
1.30890703e+00 -1.48723876e+00 3.84631455e-01 -4.24390137e-01
-7.76457489e-01 5.75463414e-01 1.06977093e+00 -3.64939928e-01
7.00251341e-01 9.88900289e-02 -1.25987366e-01 -2.26478577e-01
-3.88043106e-01 -1.20859794e-01 5.08197486e-01 6.47421479e-01
3.97965461e-01 4.94682461e-01 -3.50088865e-01 7.65367448e-01
-7.13282645e-01 -2.25369796e-01 -1.52455389e-01 9.04498398e-01
-4.36223447e-01 -5.07418573e-01 -4.90163445e-01 5.97032189e-01
-2.92131007e-01 1.31241649e-01 -4.90208268e-01 9.91483033e-01
3.73189598e-01 3.43965292e-01 -2.25613669e-01 -1.98132023e-01
4.84986722e-01 -4.67314869e-01 6.82690561e-01 -6.63482487e-01
-3.18365633e-01 -3.01250994e-01 6.47566468e-02 -4.90420938e-01
-3.44395489e-01 -6.21122301e-01 -8.25930715e-01 -3.79636467e-01
-4.05591041e-01 6.37536496e-02 4.36421096e-01 4.71928388e-01
-1.19189650e-01 1.01588167e-01 2.15716377e-01 -1.26872087e+00
-7.06860125e-01 -9.66108024e-01 -4.29306209e-01 3.95634711e-01
2.75919586e-01 -1.18923867e+00 -3.18937242e-01 3.45709808e-02] | [7.620604991912842, -0.9199326038360596] |
17f33e21-6dc0-47a0-93c5-35d798e1d5e5 | continuous-conditional-video-synthesis-by | 2210.05810 | null | https://arxiv.org/abs/2210.05810v2 | https://arxiv.org/pdf/2210.05810v2.pdf | A unified model for continuous conditional video prediction | Different conditional video prediction tasks, like video future frame prediction and video frame interpolation, are normally solved by task-related models even though they share many common underlying characteristics. Furthermore, almost all conditional video prediction models can only achieve discrete prediction. In this paper, we propose a unified model that addresses these two issues at the same time. We show that conditional video prediction can be formulated as a neural process, which maps input spatio-temporal coordinates to target pixel values given context spatio-temporal coordinates and context pixel values. Specifically, we feed the implicit neural representation of coordinates and context pixel features into a Transformer-based non-autoregressive conditional video prediction model. Our task-specific models outperform previous work for video future frame prediction and video interpolation on multiple datasets. Importantly, the model is able to interpolate or predict with an arbitrary high frame rate, i.e., continuous prediction. Our source code is available at \url{https://npvp.github.io}. | ['Guillaume-Alexandre Bilodeau', 'Xi Ye'] | 2022-10-11 | null | null | null | null | ['video-prediction', 'video-frame-interpolation'] | ['computer-vision', 'computer-vision'] | [ 2.77694881e-01 -9.50298011e-02 -4.28644061e-01 -5.15520215e-01
-8.07003796e-01 -1.70461014e-01 6.47749364e-01 -3.84859622e-01
-5.95905930e-02 7.84242690e-01 3.61306578e-01 -3.12079281e-01
3.78116846e-01 -7.29153335e-01 -1.17637062e+00 -5.63832402e-01
2.73838416e-02 -3.37208770e-02 4.81152087e-01 3.55852872e-01
2.05404758e-01 1.12711169e-01 -1.55937409e+00 8.29130232e-01
6.63694441e-01 1.12810040e+00 4.97147262e-01 1.16719759e+00
1.02639765e-01 1.53608358e+00 2.69689765e-02 -2.96514958e-01
8.67412835e-02 -3.61679077e-01 -7.31462777e-01 -1.19847603e-01
5.25665998e-01 -6.52181923e-01 -6.07471228e-01 7.07246721e-01
6.15082681e-03 1.67035267e-01 5.56775630e-01 -1.38041925e+00
-9.22925174e-01 1.41679287e-01 -3.27504426e-01 2.06639007e-01
5.66972435e-01 2.03803942e-01 9.80904818e-01 -1.03481710e+00
5.50715148e-01 8.44288170e-01 9.77752030e-01 6.56290829e-01
-1.47002196e+00 -6.20557845e-01 3.66741121e-01 5.49525261e-01
-1.39717698e+00 -5.90680540e-01 6.36041641e-01 -7.85414219e-01
1.06444979e+00 2.70365655e-01 7.71165967e-01 1.13418818e+00
3.07188004e-01 9.03065443e-01 4.70694244e-01 -1.71971187e-01
2.01483052e-02 -2.99659252e-01 -3.67995769e-01 4.59303379e-01
-6.07199609e-01 1.81319430e-01 -7.02835441e-01 1.02123156e-01
1.04994214e+00 2.86211073e-01 -4.70179290e-01 -3.20073932e-01
-1.51981914e+00 5.32785714e-01 9.90179479e-02 1.49852127e-01
-6.40512645e-01 6.66448653e-01 1.55263707e-01 1.02876343e-01
7.38118768e-01 -2.95837164e-01 -7.40945995e-01 -3.72724473e-01
-1.18472278e+00 3.34060997e-01 6.39987767e-01 1.14692068e+00
9.67145920e-01 1.64610907e-01 -3.76430392e-01 4.67458874e-01
3.16738993e-01 3.44298720e-01 2.82115430e-01 -1.45449173e+00
4.79059786e-01 -1.67370006e-01 3.70035499e-01 -9.68697190e-01
9.00596566e-03 1.03557482e-01 -6.91899896e-01 2.56999535e-03
5.27056813e-01 -2.50630766e-01 -6.31990552e-01 1.82348359e+00
5.89420386e-02 1.28354597e+00 -1.29113466e-01 8.23247492e-01
6.93015933e-01 1.18033218e+00 6.50834501e-01 -3.90097499e-01
9.06826973e-01 -9.43470657e-01 -6.36435688e-01 1.13941930e-01
6.05524898e-01 -7.23290086e-01 8.30180168e-01 1.12978846e-01
-1.33676887e+00 -9.19744790e-01 -6.15486741e-01 -4.29028273e-01
-2.35253901e-04 2.08630502e-01 3.65226299e-01 1.51687369e-01
-1.54639447e+00 5.91753721e-01 -1.02450120e+00 -1.29949421e-01
2.72226244e-01 3.04899007e-01 -2.97430336e-01 3.83400195e-03
-1.13354588e+00 4.64773148e-01 3.51682514e-01 -2.31896508e-02
-7.62528777e-01 -1.13665473e+00 -9.02948797e-01 1.57238290e-01
1.37751192e-01 -9.33782458e-01 1.49249148e+00 -1.48361194e+00
-1.41285217e+00 6.05158150e-01 -9.06965971e-01 -7.21558690e-01
5.88221550e-01 -3.41039628e-01 -2.77959943e-01 8.63069594e-02
1.11605935e-01 1.09461772e+00 1.06223464e+00 -1.05921972e+00
-9.76980031e-01 3.78752351e-02 7.17157423e-02 2.86765277e-01
-1.60119534e-01 2.05926560e-02 -7.60053933e-01 -8.16935301e-01
-6.08311519e-02 -7.27321804e-01 -2.11544320e-01 2.94656724e-01
-2.61721071e-02 -1.83377743e-01 1.05158842e+00 -9.68360543e-01
1.21405375e+00 -2.14626336e+00 2.48247877e-01 -3.13563794e-01
1.90449432e-01 -6.01242520e-02 5.72114810e-02 -6.53926060e-02
-3.54007781e-01 1.81478828e-01 -6.34492636e-02 -5.75032353e-01
-1.48742333e-01 1.46376193e-01 -5.11414766e-01 3.29713494e-01
3.20675910e-01 1.13372648e+00 -9.34565425e-01 -5.90218842e-01
5.71675897e-01 1.00493360e+00 -7.47148514e-01 2.72912890e-01
-6.39510751e-01 7.54588842e-01 -2.87680209e-01 3.61835688e-01
5.02412915e-01 -3.62515777e-01 2.36613601e-01 -3.27273399e-01
-2.84191430e-01 6.38712868e-02 -7.35231161e-01 1.76777112e+00
-3.72622162e-01 1.15836406e+00 -3.34661454e-01 -1.11621487e+00
6.84169650e-01 7.85426974e-01 9.09055769e-01 -5.74835777e-01
-3.80717784e-01 -1.75001815e-01 -6.26950026e-01 -5.49087286e-01
8.57512534e-01 -1.21986575e-01 3.13636899e-01 9.28300098e-02
-6.91423146e-03 3.19785565e-01 -1.15153410e-01 3.32060754e-02
7.53172755e-01 9.34112251e-01 1.81629688e-01 4.27872948e-02
5.47713280e-01 -7.26369694e-02 8.92034113e-01 5.12072563e-01
-2.58506387e-01 9.16081131e-01 4.84014034e-01 -7.47391939e-01
-1.28573537e+00 -1.22374678e+00 -2.53498666e-02 1.22999537e+00
8.86220336e-02 -6.14693284e-01 -4.83260185e-01 -3.29939961e-01
-3.07105839e-01 6.88861310e-01 -6.71636999e-01 2.25125074e-01
-9.18788910e-01 -1.56791687e-01 1.18269183e-01 6.74599051e-01
1.59999758e-01 -1.05578208e+00 -4.70251888e-01 3.23510855e-01
-5.81631541e-01 -1.36175656e+00 -4.84436035e-01 -2.40232483e-01
-9.39257205e-01 -8.15040350e-01 -9.87367213e-01 -9.51114118e-01
3.63444537e-01 3.09221953e-01 1.37996995e+00 2.01275721e-01
2.74876386e-01 6.28167272e-01 -1.94066107e-01 8.86637196e-02
-2.43319079e-01 -1.05570190e-01 -1.08837210e-01 1.75426081e-01
5.23258507e-01 -7.53552794e-01 -7.89146066e-01 1.99862584e-01
-6.27823591e-01 8.29628050e-01 -1.38503755e-03 6.45078242e-01
1.06493819e+00 -3.90729010e-01 3.11936557e-01 -6.83560371e-01
1.00014262e-01 -7.14351594e-01 -4.68667984e-01 2.09267348e-01
2.18001055e-03 -3.35218370e-01 7.57203221e-01 -3.80843252e-01
-1.25679946e+00 3.89312178e-01 -1.91780284e-01 -8.64130259e-01
-4.25478756e-01 3.93565387e-01 6.06017709e-02 2.39199206e-01
3.23564738e-01 6.08590662e-01 -2.74365932e-01 -3.49294782e-01
3.01055789e-01 9.58846360e-02 6.23756170e-01 -4.60811943e-01
4.99057859e-01 4.88028467e-01 -1.07496634e-01 -6.92289591e-01
-6.04841292e-01 -8.89592320e-02 -9.15723503e-01 -5.44652283e-01
1.25447345e+00 -1.28907585e+00 -6.35866582e-01 2.70591825e-01
-1.41287911e+00 -9.12333369e-01 -1.64015695e-01 5.45672059e-01
-1.09862387e+00 5.23034155e-01 -7.78584719e-01 -7.87266850e-01
1.98958665e-02 -1.15196371e+00 1.09252667e+00 3.29398997e-02
-1.36129975e-01 -1.19579923e+00 -2.25828826e-01 9.24624726e-02
2.70613074e-01 3.09953928e-01 5.36170542e-01 6.40546009e-02
-1.05746663e+00 3.19103077e-02 -3.45410764e-01 1.60821021e-01
-4.30753862e-04 3.94281596e-01 -7.49345064e-01 -3.51939835e-02
-1.10692382e-01 1.51100447e-02 7.81519651e-01 9.46975768e-01
1.66125095e+00 -3.47488582e-01 -3.27404946e-01 9.19814289e-01
1.35270953e+00 1.54341623e-01 8.28039467e-01 9.11303163e-02
8.71166885e-01 4.03310627e-01 5.94617009e-01 6.10502839e-01
7.36664355e-01 8.79146755e-01 3.49322319e-01 2.78697968e-01
-1.96845710e-01 -5.07655144e-01 3.38210732e-01 6.30788982e-01
-4.93498355e-01 -1.16619430e-01 -9.58177269e-01 5.18587530e-01
-2.15239263e+00 -1.38768303e+00 -4.44565982e-01 2.01721621e+00
6.34661674e-01 -1.03341609e-01 -2.82982830e-02 -1.75387010e-01
8.46476018e-01 2.44382083e-01 -4.77492154e-01 7.02252332e-03
5.36596291e-02 -2.39814624e-01 2.66052365e-01 6.37006104e-01
-1.34000254e+00 1.04797733e+00 6.27682400e+00 5.96880078e-01
-1.12587667e+00 2.47198001e-01 1.11179829e+00 -2.20661521e-01
-3.78975034e-01 5.29171377e-02 -5.96672833e-01 7.23656178e-01
1.14437127e+00 -1.23760983e-01 3.95411372e-01 7.61571050e-01
4.51475233e-01 1.34414449e-01 -1.25812340e+00 1.14753640e+00
-6.91464245e-02 -1.83416128e+00 1.33977890e-01 -1.75282061e-01
7.83255875e-01 5.96827874e-03 2.56510764e-01 2.76644886e-01
8.85574967e-02 -9.65876460e-01 9.85453069e-01 1.04393601e+00
9.95735407e-01 -4.08544898e-01 2.06904814e-01 2.60108680e-01
-1.49035704e+00 2.29823627e-02 -3.00506502e-01 -3.02450925e-01
6.74723089e-01 1.68955103e-01 -2.54972249e-01 2.87350357e-01
8.57743919e-01 1.39799142e+00 -2.75978833e-01 9.09085095e-01
-8.17164555e-02 7.02673614e-01 -6.29020780e-02 4.44216192e-01
9.39730778e-02 -2.17655659e-01 2.04424977e-01 1.23329091e+00
6.43339217e-01 1.75503701e-01 1.35972396e-01 7.20300615e-01
1.05963089e-01 -1.06352819e-02 -5.73608637e-01 5.09936571e-01
3.08937520e-01 6.43911958e-01 -3.21080536e-01 -5.39650917e-01
-8.94091487e-01 1.10950661e+00 4.37478274e-01 8.43531609e-01
-1.40334535e+00 3.35721076e-01 9.38440800e-01 1.65735766e-01
5.45309424e-01 -4.18185681e-01 -4.70357299e-01 -1.40450072e+00
1.80850208e-01 -2.50196069e-01 1.60345346e-01 -1.15746081e+00
-8.62256706e-01 4.90598619e-01 -1.43252477e-01 -1.58223546e+00
-6.29767954e-01 -4.90100771e-01 -5.89517474e-01 8.40836048e-01
-1.57895529e+00 -1.29539669e+00 -2.52226681e-01 9.61025894e-01
7.95939744e-01 1.08179478e-02 7.78424680e-01 4.24493730e-01
-3.34523737e-01 3.76339972e-01 9.62333679e-02 1.74341500e-01
5.02260923e-01 -9.85503376e-01 4.87472653e-01 9.89411831e-01
1.71052814e-01 1.97383329e-01 7.15210617e-01 -5.92758417e-01
-1.26473057e+00 -1.40228069e+00 1.16346693e+00 -6.23680532e-01
6.61465824e-01 -2.35291094e-01 -9.87895906e-01 1.22250021e+00
3.22285779e-02 3.70345682e-01 6.86163425e-01 -2.22336277e-01
-1.57493800e-01 1.49868438e-02 -6.78539455e-01 7.92389810e-01
1.02807486e+00 -6.78119421e-01 -1.02983296e-01 4.21006471e-01
1.04216242e+00 -5.05101383e-01 -1.01935530e+00 3.20305854e-01
7.36453176e-01 -1.22143042e+00 1.36943126e+00 -6.35338187e-01
8.66223693e-01 -2.86713243e-01 -4.42551404e-01 -6.24734700e-01
-5.26404679e-01 -4.54052866e-01 -6.70500219e-01 9.41883385e-01
4.87086117e-01 -2.46139154e-01 1.12555826e+00 1.08066463e+00
-7.17981756e-02 -7.63251841e-01 -1.13533604e+00 -3.98003578e-01
1.47112548e-01 -1.04757953e+00 4.44118977e-01 7.99742281e-01
-2.07312360e-01 -1.27382755e-01 -9.71379519e-01 1.58045188e-01
4.24816430e-01 1.54034376e-01 6.34788573e-01 -8.63687694e-01
-3.71120334e-01 -2.85205185e-01 -4.13259715e-01 -1.64355230e+00
4.71658140e-01 -6.07057810e-01 1.10519350e-01 -1.51250589e+00
4.88249511e-02 -3.63698661e-01 -1.92579567e-01 2.87827849e-01
-2.24778667e-01 5.25627077e-01 4.15670812e-01 4.47372526e-01
-7.61066496e-01 4.39100057e-01 1.10959184e+00 1.40789047e-01
-1.59399122e-01 3.82519588e-02 -1.42724201e-01 6.88259900e-01
1.09115863e+00 -1.96285412e-01 -3.92764330e-01 -8.38134170e-01
1.36121452e-01 6.51801109e-01 7.47748554e-01 -1.10446048e+00
3.99164975e-01 -4.83530611e-01 6.34603620e-01 -5.80451667e-01
7.55212903e-01 -8.41535151e-01 5.33615589e-01 2.07544968e-01
-4.52157229e-01 1.31629601e-01 -5.95281087e-02 8.00559342e-01
-2.41586179e-01 1.83037758e-01 5.31695604e-01 -1.04197629e-01
-1.20990694e+00 7.85397947e-01 -7.23202586e-01 -2.66406775e-01
1.20398080e+00 -5.20188868e-01 8.44830349e-02 -7.07894266e-01
-1.03144467e+00 1.47638517e-03 4.15947050e-01 4.37241405e-01
8.67132246e-01 -1.50522113e+00 -7.66423941e-01 9.41329151e-02
-1.59704030e-01 -8.76406580e-02 6.07577503e-01 9.33655262e-01
-5.05930424e-01 4.12120521e-01 -1.05690613e-01 -9.47177708e-01
-1.22753251e+00 7.75974631e-01 2.81671673e-01 1.06274173e-01
-7.88808584e-01 7.50089586e-01 4.67379361e-01 -3.30938399e-02
1.88282371e-01 -4.00139540e-01 -1.30782917e-01 -3.40038210e-01
7.24509299e-01 -9.84091312e-03 -6.55831873e-01 -1.11125088e+00
1.30393893e-01 5.53595364e-01 2.45035082e-01 1.07642896e-01
1.23986971e+00 -3.66675347e-01 -1.50430971e-03 5.08294880e-01
1.28223753e+00 -5.59903800e-01 -2.08463526e+00 -2.51507342e-01
-1.26576781e-01 -6.72005951e-01 -1.08676970e-01 -2.28510588e-01
-1.23134804e+00 8.76715958e-01 3.33741665e-01 1.76603287e-01
1.21744943e+00 -8.16939026e-02 6.92242980e-01 2.99040508e-03
2.60403216e-01 -7.37075508e-01 -1.73335671e-01 6.13655567e-01
6.92795277e-01 -1.22698939e+00 -2.35130802e-01 -5.35417616e-01
-6.77674711e-01 1.22384131e+00 7.51538396e-01 -1.97507828e-01
8.85469973e-01 1.49104401e-01 -2.72719204e-01 4.16697711e-01
-1.11964643e+00 5.53947948e-02 2.12324947e-01 8.23195636e-01
9.00136948e-01 1.11386843e-01 2.59257853e-02 4.03308392e-01
6.16255496e-03 5.73469818e-01 2.18189016e-01 5.54802001e-01
-8.19894001e-02 -9.33745205e-01 -1.65859133e-01 4.78751957e-01
-5.26429296e-01 -3.21883202e-01 4.09566849e-01 3.05866897e-01
1.05594918e-01 6.75620556e-01 4.24392581e-01 -5.41988075e-01
-1.60927191e-01 1.17016450e-01 2.46531054e-01 -1.77979469e-01
-1.44627243e-01 -1.13262534e-01 -1.36788845e-01 -8.82895529e-01
-7.79093623e-01 -8.43663216e-01 -1.03497362e+00 -5.25661111e-01
2.96546966e-01 -2.16507792e-01 2.18057126e-01 7.27708519e-01
4.70268369e-01 3.75269353e-01 2.33022466e-01 -1.30872130e+00
2.50725597e-01 -6.27192080e-01 -1.70026124e-01 5.09126782e-01
4.20925021e-01 -2.93552250e-01 9.49479863e-02 9.35191154e-01] | [10.486515045166016, -0.7490968704223633] |
8a4a76bd-7556-474b-a0f0-e879fbdad470 | multi-person-articulated-tracking-with | 1903.09214 | null | http://arxiv.org/abs/1903.09214v1 | http://arxiv.org/pdf/1903.09214v1.pdf | Multi-person Articulated Tracking with Spatial and Temporal Embeddings | We propose a unified framework for multi-person pose estimation and tracking.
Our framework consists of two main components,~\ie~SpatialNet and TemporalNet.
The SpatialNet accomplishes body part detection and part-level data association
in a single frame, while the TemporalNet groups human instances in consecutive
frames into trajectories. Specifically, besides body part detection heatmaps,
SpatialNet also predicts the Keypoint Embedding (KE) and Spatial Instance
Embedding (SIE) for body part association. We model the grouping procedure into
a differentiable Pose-Guided Grouping (PGG) module to make the whole part
detection and grouping pipeline fully end-to-end trainable. TemporalNet extends
spatial grouping of keypoints to temporal grouping of human instances. Given
human proposals from two consecutive frames, TemporalNet exploits both
appearance features encoded in Human Embedding (HE) and temporally consistent
geometric features embodied in Temporal Instance Embedding (TIE) for robust
tracking. Extensive experiments demonstrate the effectiveness of our proposed
model. Remarkably, we demonstrate substantial improvements over the
state-of-the-art pose tracking method from 65.4\% to 71.8\% Multi-Object
Tracking Accuracy (MOTA) on the ICCV'17 PoseTrack Dataset. | ['Sheng Jin', 'Wentao Liu', 'Chen Qian', 'Wanli Ouyang'] | 2019-03-21 | multi-person-articulated-tracking-with-1 | http://openaccess.thecvf.com/content_CVPR_2019/html/Jin_Multi-Person_Articulated_Tracking_With_Spatial_and_Temporal_Embeddings_CVPR_2019_paper.html | http://openaccess.thecvf.com/content_CVPR_2019/papers/Jin_Multi-Person_Articulated_Tracking_With_Spatial_and_Temporal_Embeddings_CVPR_2019_paper.pdf | cvpr-2019-6 | ['multi-person-pose-estimation-and-tracking'] | ['computer-vision'] | [-3.40216100e-01 -5.21193147e-02 -3.64986718e-01 -2.31780231e-01
-6.32770956e-01 -3.75910759e-01 3.85191441e-01 -1.80961639e-01
-5.06179929e-01 5.84240735e-01 2.95453131e-01 4.36739206e-01
-4.08123024e-02 -4.82521147e-01 -7.41806507e-01 -4.03581083e-01
-5.84664702e-01 5.93430281e-01 5.18066227e-01 -2.11733533e-03
-3.78249615e-01 5.37342131e-01 -1.36868906e+00 -1.78838685e-01
4.76162076e-01 9.22513306e-01 -6.78496808e-02 6.77740455e-01
3.64770532e-01 2.07001597e-01 -3.44862878e-01 -3.25857848e-01
4.89244848e-01 -1.43824115e-01 -4.45623130e-01 1.02819785e-01
1.20426619e+00 -3.39718193e-01 -7.86318541e-01 9.33218837e-01
6.55164421e-01 3.24758083e-01 3.39582115e-01 -1.50856280e+00
-2.81847030e-01 2.28144005e-01 -8.09616446e-01 2.93458581e-01
5.19836485e-01 4.14612442e-01 7.58732796e-01 -8.83606851e-01
8.67855191e-01 1.55493116e+00 1.13200545e+00 7.53193557e-01
-1.00334775e+00 -8.67468596e-01 4.83234823e-01 -3.40124965e-02
-1.46626472e+00 -2.60328799e-01 6.78991854e-01 -5.25940776e-01
6.73381269e-01 1.12645730e-01 1.26578021e+00 6.58373177e-01
4.61094618e-01 1.15392959e+00 4.87167656e-01 -4.26507406e-02
-3.96398038e-01 -4.20258433e-01 2.76078910e-01 1.15239894e+00
3.63486946e-01 4.13581610e-01 -6.39805496e-01 -1.21052116e-01
9.68816102e-01 3.42917383e-01 -1.57160178e-01 -7.18387783e-01
-1.51895058e+00 4.81532365e-01 9.33727384e-01 -3.46427056e-04
-4.67538953e-01 8.59229505e-01 3.87679130e-01 -1.48499936e-01
4.05499518e-01 -1.04675323e-01 -3.97287697e-01 1.28789797e-01
-1.06606495e+00 7.02512860e-01 2.86059499e-01 1.35715199e+00
3.96684349e-01 -2.92950645e-02 -6.93595052e-01 4.92884338e-01
7.50867784e-01 7.68327832e-01 1.28305167e-01 -7.81267464e-01
5.03258049e-01 6.67795062e-01 2.71191239e-01 -1.05564082e+00
-5.94232678e-01 -5.46883702e-01 -5.84574759e-01 -8.79571959e-03
5.17850816e-01 -1.45992383e-01 -1.11059821e+00 1.95550847e+00
8.45091820e-01 3.98248196e-01 -6.02221310e-01 1.15367043e+00
8.53614390e-01 2.98633635e-01 6.60711884e-01 -2.21208110e-03
1.73919594e+00 -1.06720746e+00 -7.39270329e-01 -1.90803468e-01
4.65031445e-01 -4.14924562e-01 3.36126626e-01 -1.93792075e-01
-1.28006923e+00 -1.06534159e+00 -7.82337606e-01 -3.70200560e-03
-1.16137713e-01 4.89034534e-01 6.28109574e-01 5.07525980e-01
-1.08032000e+00 5.11801422e-01 -1.13135910e+00 -6.05551839e-01
4.58171815e-01 7.80581355e-01 -3.64212960e-01 2.13656742e-02
-9.72908676e-01 5.24317205e-01 4.31600630e-01 4.45337325e-01
-7.82858014e-01 -8.19007933e-01 -1.21687949e+00 -3.44077289e-01
2.98256546e-01 -1.27446103e+00 1.04222035e+00 -2.02173680e-01
-9.74083662e-01 8.33889544e-01 -3.79171282e-01 -5.81825733e-01
7.65541077e-01 -7.17143774e-01 -5.50522923e-01 1.31826311e-01
4.09353673e-01 1.15862787e+00 7.68271983e-01 -1.03938890e+00
-9.69605148e-01 -5.87469101e-01 -4.47958112e-01 2.27343544e-01
6.65230677e-02 2.89092422e-01 -1.16600466e+00 -8.91376674e-01
3.07663143e-01 -1.25029027e+00 -2.38856837e-01 7.20781147e-01
-5.45463741e-01 -5.53413093e-01 1.01458573e+00 -8.10540438e-01
1.43251109e+00 -1.95913148e+00 3.33439142e-01 1.14805467e-01
5.22332072e-01 8.27989951e-02 1.14584677e-01 -4.76614684e-02
6.27372041e-02 -3.31818461e-01 2.70455271e-01 -6.78633273e-01
1.54861271e-01 -6.13646097e-02 5.90129308e-02 9.59238291e-01
-1.33903086e-01 1.41378772e+00 -1.00879288e+00 -1.01595545e+00
5.11809886e-01 5.39209902e-01 -5.40760517e-01 -1.34743014e-02
-1.43012002e-01 4.52907026e-01 -6.02246940e-01 9.50028539e-01
4.16723490e-01 -2.83545285e-01 1.60588194e-02 -5.50041854e-01
-1.95545554e-01 -2.17244700e-01 -1.28434980e+00 2.10105729e+00
3.19067359e-01 3.14734429e-01 -8.21390375e-02 -1.03946641e-01
5.64555705e-01 4.12714690e-01 9.59572732e-01 -3.46567541e-01
1.32234752e-01 -1.58035517e-01 -1.06155522e-01 -5.66477738e-02
6.76231325e-01 8.30476359e-02 -2.70651370e-01 -7.26415915e-03
2.36389935e-01 6.72280014e-01 8.94260481e-02 9.05365571e-02
8.15898538e-01 6.78366661e-01 1.79295003e-01 -2.75532991e-01
3.61356676e-01 9.00449604e-02 8.92214656e-01 6.01274908e-01
-8.90630662e-01 4.03359503e-01 -3.00011605e-01 -7.98700511e-01
-8.39347482e-01 -1.32825208e+00 1.19757086e-01 1.32971466e+00
5.64779580e-01 -6.19710803e-01 -3.51343900e-01 -7.48897731e-01
4.26250875e-01 -1.07738897e-01 -8.86499226e-01 1.14221565e-01
-1.20311081e+00 -4.04174358e-01 5.27422011e-01 1.01880610e+00
5.69591880e-01 -1.10631740e+00 -6.97676718e-01 4.28403169e-01
-8.48508552e-02 -9.81065571e-01 -1.18909967e+00 -3.27643692e-01
-9.09052670e-01 -1.09775841e+00 -9.55557227e-01 -9.02045548e-01
5.33037782e-01 1.84919253e-01 1.03209352e+00 9.93348211e-02
-5.92316508e-01 6.77859902e-01 3.31500135e-02 -1.09278664e-01
3.43821585e-01 -1.21284232e-01 4.88634318e-01 -2.59528309e-01
1.30407602e-01 -1.09858453e-01 -1.18919647e+00 5.70095837e-01
-5.95930144e-02 -7.54108280e-02 5.42431951e-01 4.91362542e-01
8.38554263e-01 -2.79813051e-01 -1.45548107e-02 -1.57285392e-01
-3.75708081e-02 -1.37457609e-01 -4.59343165e-01 3.73844713e-01
-1.55798957e-01 -3.48832220e-01 6.71556294e-02 -4.94050443e-01
-6.73700154e-01 4.60996926e-01 7.08792433e-02 -8.62639010e-01
-1.43603876e-01 -1.35137409e-01 -1.14348605e-01 -2.49916032e-01
3.05302322e-01 3.10528815e-01 1.52292838e-02 -3.96644711e-01
5.03213763e-01 -2.03048274e-01 1.08123779e+00 -6.88772917e-01
1.12618840e+00 6.05477750e-01 -3.39606591e-02 -5.22801220e-01
-7.27252364e-01 -1.01303935e+00 -1.04367244e+00 -7.07200348e-01
1.36004925e+00 -1.31708133e+00 -1.04982805e+00 3.12593162e-01
-1.02893484e+00 -9.70183760e-02 -2.61124015e-01 5.75802088e-01
-4.77900803e-01 3.38456303e-01 -6.79061770e-01 -7.59936988e-01
-6.22242928e-01 -8.08890700e-01 1.75740385e+00 2.79787898e-01
-2.94600010e-01 -1.04865921e+00 2.00021848e-01 2.04387624e-02
-1.93721414e-01 6.69824481e-01 3.89168672e-02 -1.26831725e-01
-8.42847168e-01 -4.33466583e-01 -1.66159078e-01 -5.21641672e-01
-1.74462780e-01 -2.51216650e-01 -5.12246490e-01 -8.87006044e-01
-6.65226042e-01 -3.66441682e-02 7.08041430e-01 9.17216957e-01
8.90873253e-01 -9.76217613e-02 -1.07069921e+00 8.69956315e-01
1.22020984e+00 -1.22721922e-02 3.94555449e-01 2.61018813e-01
1.23097587e+00 2.17880026e-01 1.11547601e+00 2.90298581e-01
5.24540961e-01 1.27735543e+00 2.52365053e-01 -1.39416113e-01
-5.23732662e-01 -4.74347323e-01 3.76248121e-01 4.48892266e-01
-2.76446998e-01 8.80802274e-02 -7.77003646e-01 6.67231321e-01
-2.09186888e+00 -1.17321181e+00 -2.04586267e-01 1.94212627e+00
4.90066171e-01 1.10309385e-01 8.51664782e-01 -3.65224034e-01
9.95157957e-01 1.94921434e-01 -5.13316810e-01 6.63712800e-01
3.12575817e-01 -1.31270155e-01 7.02916443e-01 2.62038440e-01
-1.72391891e+00 1.09235168e+00 5.92438889e+00 5.50989151e-01
-5.58829963e-01 1.35929167e-01 5.36055900e-02 -1.83457240e-01
3.03039044e-01 -3.63699943e-01 -1.50766921e+00 4.28968906e-01
4.12888110e-01 5.91561310e-02 -2.62274891e-01 8.79745603e-01
2.24089343e-02 2.69497871e-01 -1.09876645e+00 1.18404579e+00
-3.62993032e-02 -1.12973046e+00 -1.10693969e-01 1.94530204e-01
4.35671985e-01 -1.36420414e-01 8.46714005e-02 4.42494154e-01
2.73480684e-01 -7.48190880e-01 1.06950259e+00 6.49430931e-01
8.43851626e-01 -6.26359463e-01 3.82358193e-01 1.16420023e-01
-2.32222009e+00 -6.37798533e-02 -1.89412385e-01 2.32319728e-01
5.00389457e-01 -1.36084706e-01 -5.79014659e-01 6.61082685e-01
9.55478430e-01 1.03369260e+00 -7.01481402e-01 1.50232959e+00
1.00090303e-01 2.25021049e-01 -4.29429650e-01 2.93346137e-01
1.83434740e-01 6.06218353e-02 7.45643675e-01 1.20573390e+00
1.01995952e-01 2.26107433e-01 8.83571446e-01 6.41185701e-01
1.01253301e-01 -2.48339295e-01 -2.63982803e-01 3.93500984e-01
5.31844914e-01 1.22438109e+00 -7.10763752e-01 -3.84836078e-01
-2.35977262e-01 1.04486299e+00 7.49209598e-02 1.37958780e-01
-1.24676836e+00 -1.60791613e-02 1.02402520e+00 2.64337629e-01
4.42072123e-01 -2.94410914e-01 2.56415512e-02 -9.30803299e-01
8.94327685e-02 -3.92590225e-01 8.62516940e-01 -4.69905823e-01
-1.26502657e+00 2.87469596e-01 1.59936905e-01 -1.63698351e+00
-1.80289254e-01 -4.44425464e-01 -4.65604216e-01 7.75204897e-01
-1.01818883e+00 -1.79945982e+00 -4.56789643e-01 8.34801257e-01
6.26299858e-01 1.28875852e-01 4.05892253e-01 4.24967200e-01
-6.48516417e-01 8.58652115e-01 -5.45999408e-01 7.16561913e-01
7.34040916e-01 -1.39636576e+00 7.35604882e-01 9.07883406e-01
2.88408808e-02 9.23347175e-01 6.33298576e-01 -1.29071045e+00
-1.48363590e+00 -1.64429271e+00 7.52776384e-01 -8.86306584e-01
4.70381886e-01 -3.28826457e-01 -5.00341833e-01 9.87531602e-01
-2.29278013e-01 4.57582921e-01 4.34597671e-01 1.15251154e-01
-1.56224012e-01 8.21239501e-02 -9.92768228e-01 5.82853913e-01
1.56303144e+00 -2.30632529e-01 -6.97412968e-01 3.05827945e-01
8.11233640e-01 -9.03964281e-01 -1.16899276e+00 4.96366560e-01
9.78978693e-01 -4.24416989e-01 1.54118299e+00 -5.97503483e-01
-2.65819818e-01 -8.39653313e-01 1.52879208e-01 -6.87697470e-01
-8.29213679e-01 -6.11582339e-01 -6.87816441e-01 7.24083483e-01
-1.35139123e-01 -1.14647672e-01 1.06626475e+00 5.44875145e-01
-3.22488993e-01 -6.62780881e-01 -9.70460474e-01 -9.52732384e-01
-2.77423292e-01 -3.35506648e-01 3.63156706e-01 6.27643824e-01
-3.47679317e-01 -2.01570511e-01 -7.03605354e-01 4.91978228e-01
1.32916391e+00 2.34836057e-01 1.00928795e+00 -1.30430174e+00
-1.85222194e-01 -2.55647480e-01 -8.00716460e-01 -1.36791873e+00
-8.37542117e-02 -7.18081892e-01 1.83481455e-01 -1.46241105e+00
3.85709524e-01 -3.02709937e-01 -3.33626479e-01 6.03107870e-01
-4.93871540e-01 5.47020972e-01 6.04356825e-01 4.10633802e-01
-1.25630474e+00 5.69233954e-01 1.33924365e+00 -7.01698363e-02
-3.10038954e-01 1.90554447e-02 -1.52537599e-01 6.31353438e-01
1.99687645e-01 -4.14687753e-01 -6.50563277e-03 -1.59626052e-01
-3.82869691e-01 1.70725331e-01 1.01252925e+00 -1.50984550e+00
5.77797890e-01 2.63269365e-01 1.01881981e+00 -1.27001393e+00
5.76736450e-01 -7.83843935e-01 4.16588277e-01 8.56931627e-01
-3.81996483e-02 1.52021959e-01 4.06473339e-01 9.45447803e-01
1.28650814e-01 7.51687646e-01 7.16443837e-01 2.58048810e-02
-1.12567282e+00 1.08616114e+00 4.15394366e-01 -6.74043521e-02
1.10868227e+00 -6.08005404e-01 -4.76015136e-02 1.14357024e-01
-1.24643111e+00 6.82610273e-01 4.24045891e-01 6.92731380e-01
6.66819751e-01 -1.74713111e+00 -4.52457219e-01 -2.80647241e-02
1.21453419e-01 -1.11632116e-01 4.24161851e-01 1.12203360e+00
-1.81543246e-01 5.65265954e-01 -2.12164134e-01 -1.05530059e+00
-1.64349771e+00 4.65464383e-01 4.38465923e-01 -2.31332198e-01
-1.16299045e+00 1.01809561e+00 3.92942339e-01 -2.70954967e-01
4.45457876e-01 -3.64415109e-01 -7.83163588e-03 -1.69992000e-01
4.44750458e-01 4.80610102e-01 -6.04490101e-01 -1.07278502e+00
-6.71766877e-01 1.05485916e+00 -9.66671631e-02 2.93725040e-02
9.95418906e-01 -9.10400599e-02 2.45596275e-01 9.40643549e-02
9.83312547e-01 -2.45825514e-01 -1.71894741e+00 -2.55452275e-01
-1.45842150e-01 -4.99565482e-01 -3.17756891e-01 -6.17111087e-01
-1.02385354e+00 3.81679058e-01 9.89270449e-01 -3.81080091e-01
7.17572272e-01 2.37370834e-01 1.07570827e+00 1.12955673e-02
7.30234623e-01 -9.09337997e-01 3.59862633e-02 2.71004766e-01
6.75538301e-01 -1.07120574e+00 2.08174050e-01 -5.36748052e-01
-4.58679765e-01 7.17689216e-01 8.22465956e-01 -2.43532509e-01
6.81724727e-01 -1.46236002e-01 -2.10702643e-01 -7.12902188e-01
-3.40121806e-01 -4.88733828e-01 1.02793479e+00 5.23058951e-01
4.10607904e-01 1.04854323e-01 -1.33090436e-01 4.47556198e-01
-1.34263009e-01 5.58810483e-04 -5.38515806e-01 9.80999231e-01
-5.05806804e-01 -6.87360704e-01 -6.83394432e-01 2.98331320e-01
-2.02029765e-01 3.61309171e-01 -5.79301380e-02 1.33893490e+00
5.09601772e-01 4.36674654e-01 1.00901097e-01 -4.29047823e-01
5.41195929e-01 5.74341379e-02 6.73661470e-01 -6.33218765e-01
-5.93382239e-01 4.31965351e-01 -7.75449052e-02 -1.10368633e+00
-5.22625208e-01 -1.05573952e+00 -1.33713162e+00 -2.21823722e-01
-1.95756003e-01 -1.09939374e-01 1.49854496e-01 7.84608603e-01
2.37469986e-01 7.41376221e-01 -5.55864312e-02 -1.34812117e+00
-1.82150006e-01 -8.15590203e-01 -5.56614697e-01 5.13099015e-01
2.98682302e-01 -1.10843813e+00 3.49159747e-01 2.21918404e-01] | [6.9854044914245605, -0.9724116325378418] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.