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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
a4520375-3907-4dc9-9d6b-9ca8ea1916db | shapley-head-pruning-identifying-and-removing | 2210.05709 | null | https://arxiv.org/abs/2210.05709v1 | https://arxiv.org/pdf/2210.05709v1.pdf | Shapley Head Pruning: Identifying and Removing Interference in Multilingual Transformers | Multilingual transformer-based models demonstrate remarkable zero and few-shot transfer across languages by learning and reusing language-agnostic features. However, as a fixed-size model acquires more languages, its performance across all languages degrades, a phenomenon termed interference. Often attributed to limited model capacity, interference is commonly addressed by adding additional parameters despite evidence that transformer-based models are overparameterized. In this work, we show that it is possible to reduce interference by instead identifying and pruning language-specific parameters. First, we use Shapley Values, a credit allocation metric from coalitional game theory, to identify attention heads that introduce interference. Then, we show that removing identified attention heads from a fixed model improves performance for a target language on both sentence classification and structural prediction, seeing gains as large as 24.7\%. Finally, we provide insights on language-agnostic and language-specific attention heads using attention visualization. | ['Diyi Yang', 'William Held'] | 2022-10-11 | null | null | null | null | ['sentence-classification'] | ['natural-language-processing'] | [-3.83042581e-02 1.23482376e-01 -1.44835800e-01 5.01072817e-02
-1.12996805e+00 -8.23243260e-01 3.32323730e-01 1.49161428e-01
-7.00031281e-01 8.60191345e-01 4.60304409e-01 -6.80983782e-01
-9.44458395e-02 -4.89630848e-01 -5.64284682e-01 -3.66369605e-01
-1.11586548e-01 5.73476970e-01 -1.31625831e-01 -7.14479089e-01
1.32494554e-01 1.49729356e-01 -1.19144404e+00 1.60908297e-01
1.33162916e+00 2.40900725e-01 4.45181817e-01 6.03026390e-01
-3.51832330e-01 9.62245226e-01 -8.05446386e-01 -6.01489127e-01
2.48502016e-01 -4.95025486e-01 -8.59326065e-01 -2.63079375e-01
3.01790774e-01 1.41466931e-01 -6.68955967e-02 9.43785667e-01
5.11917531e-01 1.83067694e-01 5.23279130e-01 -1.00315678e+00
-9.72022355e-01 1.21104002e+00 -6.25343680e-01 5.87203562e-01
2.15208739e-01 3.87028396e-01 1.48797250e+00 -5.66788375e-01
5.48795164e-01 1.19472921e+00 6.82501614e-01 7.10913718e-01
-1.62068450e+00 -8.26943934e-01 4.46405113e-01 -6.98461011e-02
-1.27412963e+00 -6.02414966e-01 5.96200526e-01 -5.44437468e-01
1.54014850e+00 3.78773659e-01 6.67961717e-01 9.01006997e-01
2.03513831e-01 7.70260930e-01 1.10881937e+00 -6.85157120e-01
-5.68932071e-02 9.93054956e-02 2.65543759e-01 5.78469217e-01
3.37231040e-01 2.03867331e-02 -6.46383345e-01 -1.36786073e-01
2.22693488e-01 -5.11808932e-01 -2.57054925e-01 -3.33352506e-01
-9.15372312e-01 9.44397449e-01 2.61010110e-01 3.70533854e-01
-2.18554050e-01 1.06814057e-01 3.70334864e-01 7.23076701e-01
6.34359717e-01 1.12278843e+00 -8.37723434e-01 -3.11788738e-01
-6.32744253e-01 -7.75263309e-02 7.64436305e-01 8.80062163e-01
6.95901275e-01 2.75124431e-01 -3.72549266e-01 1.04882598e+00
-6.56459853e-02 5.50985217e-01 4.66856092e-01 -8.99591625e-01
5.80097675e-01 6.29737198e-01 -2.15566233e-01 -4.83041614e-01
-5.73533595e-01 -9.37259436e-01 -6.37545586e-01 -6.96970671e-02
5.31142473e-01 -4.10872608e-01 -5.14414966e-01 2.31922579e+00
-2.73978651e-01 -3.78511965e-01 8.20005536e-02 6.02191210e-01
3.90803397e-01 4.13713098e-01 4.18674588e-01 -4.19111133e-01
1.36254084e+00 -9.39525843e-01 -4.44562137e-01 -5.92134893e-01
1.15655470e+00 -6.67124867e-01 1.57765520e+00 3.43996614e-01
-1.27836800e+00 -1.32738605e-01 -7.86210239e-01 3.75555120e-02
-6.22603856e-02 -4.09354180e-01 7.43930221e-01 8.87783289e-01
-1.33417726e+00 3.37294549e-01 -4.16331351e-01 -4.35134470e-01
3.51542532e-01 4.87516433e-01 -2.09062904e-01 1.31672174e-02
-1.28589499e+00 1.03424191e+00 6.86990917e-02 -5.66892505e-01
-6.87300980e-01 -9.84193385e-01 -7.63802052e-01 4.76555526e-01
3.97431642e-01 -7.76696444e-01 1.32326806e+00 -1.21942103e+00
-1.44036829e+00 7.72307694e-01 -6.22247532e-02 -5.52163005e-01
1.51740953e-01 -1.47003168e-02 -2.42498498e-02 -4.73841310e-01
1.17621347e-01 4.68412757e-01 3.56564611e-01 -8.95831585e-01
-5.64619899e-01 -4.37397920e-02 5.64954579e-01 6.09181523e-01
-8.27498794e-01 4.14397210e-01 -2.60823071e-01 -7.08207965e-01
-4.07432079e-01 -8.92574787e-01 -2.58577555e-01 -5.59693575e-01
-3.03458303e-01 -2.03904703e-01 -4.37844358e-02 -8.25801194e-01
1.37324131e+00 -1.94678617e+00 3.50129217e-01 -6.66303858e-02
5.13902307e-01 4.04857025e-02 -6.91018283e-01 3.07704717e-01
-1.98362246e-02 5.78565419e-01 -1.37142286e-01 -3.43402117e-01
4.83951205e-03 -5.51546775e-02 1.93695396e-01 6.89600408e-02
-7.36171007e-03 1.03585899e+00 -8.43222082e-01 -2.54546583e-01
-3.98200266e-02 5.57709597e-02 -1.17864573e+00 -3.23796362e-01
-1.70489982e-01 1.91541106e-01 5.83772548e-02 3.50629926e-01
4.35829490e-01 -3.69575262e-01 6.00269675e-01 1.96338609e-01
4.07624356e-02 5.24284244e-01 -6.47601426e-01 1.52533603e+00
-8.01293373e-01 7.03931093e-01 1.06796324e-01 -9.55627620e-01
5.78087032e-01 9.15916413e-02 1.46004409e-01 -9.20660973e-01
1.02604724e-01 6.55702055e-02 7.46360123e-01 5.17867468e-02
5.96510649e-01 -1.58333883e-01 -4.45689112e-01 8.18277538e-01
1.74635991e-01 9.74752381e-02 2.02132374e-01 4.79583949e-01
1.22404087e+00 -5.36407173e-01 4.89094347e-01 -7.51062453e-01
4.77664709e-01 9.23625194e-03 5.78211784e-01 9.79114532e-01
-2.38739416e-01 7.27362325e-03 6.18407309e-01 -1.61067694e-01
-1.09364021e+00 -9.11296666e-01 2.81788796e-01 1.87388086e+00
-2.51180083e-01 -6.61318421e-01 -7.34667838e-01 -6.56478643e-01
1.49385519e-02 9.69757557e-01 -5.46787381e-01 -4.27021503e-01
-5.35634816e-01 -9.06369627e-01 5.70128262e-01 4.15733486e-01
1.63775235e-01 -9.75184977e-01 -1.64594680e-01 1.24965928e-01
-4.07246321e-01 -6.99440002e-01 -9.20712829e-01 4.50966656e-01
-6.24710679e-01 -8.46604288e-01 -6.86240852e-01 -6.25332177e-01
3.93987447e-01 2.88467199e-01 1.51266515e+00 1.79829136e-01
-5.56106567e-02 5.37030876e-01 -1.54978648e-01 -3.80864054e-01
-6.48855984e-01 6.81687236e-01 1.93989635e-01 -5.72451472e-01
3.17634135e-01 -3.68716061e-01 -2.36256540e-01 -2.99716014e-02
-9.19951573e-02 1.07842922e-01 4.59395915e-01 9.82922792e-01
-8.07601735e-02 -5.66554308e-01 6.79500282e-01 -8.17051768e-01
1.20520389e+00 -3.68933201e-01 -6.37237132e-01 4.13784027e-01
-6.88441098e-01 3.48693699e-01 7.87650764e-01 -4.11971748e-01
-9.20864701e-01 -3.77105147e-01 3.26043189e-01 -7.96843097e-02
3.06776196e-01 6.43357217e-01 -7.89691210e-02 -1.72040641e-01
8.56942713e-01 2.44066164e-01 -1.93687361e-02 -4.04936641e-01
3.20376486e-01 4.60061669e-01 -1.64875060e-01 -7.81385005e-01
5.62562525e-01 -2.16671363e-01 -5.10133326e-01 -7.34208941e-01
-8.78035903e-01 -2.06631556e-01 -2.36540109e-01 -3.13963667e-02
6.69716597e-01 -9.68366444e-01 -8.36118817e-01 2.86307722e-01
-1.04354703e+00 -7.32574224e-01 -4.68415767e-01 2.88536161e-01
-4.26642448e-01 1.26494080e-01 -8.46746027e-01 -7.70086110e-01
-5.30521452e-01 -1.02759242e+00 6.49611950e-01 3.38991396e-02
-2.98126459e-01 -1.16048670e+00 1.27494693e-01 6.28503263e-01
6.27123892e-01 -7.04685509e-01 1.26837194e+00 -9.19857979e-01
-5.14889717e-01 2.87776381e-01 2.54741218e-02 2.59336174e-01
3.69584672e-02 -4.73692060e-01 -7.10759819e-01 -5.99781513e-01
-2.75078744e-01 -3.92619431e-01 7.27957070e-01 4.24581498e-01
8.76133561e-01 -3.74871105e-01 -1.38575330e-01 6.34914696e-01
1.09281301e+00 1.37927800e-01 2.24479392e-01 4.73978400e-01
4.88394380e-01 4.83916253e-01 4.39876504e-02 2.24789768e-01
5.27162969e-01 7.00597942e-01 -1.12589575e-01 4.63088825e-02
-7.28390589e-02 -1.72692925e-01 4.80400234e-01 1.32889366e+00
3.38611640e-02 -4.82140571e-01 -1.18942440e+00 5.48295379e-01
-1.46123993e+00 -8.86912405e-01 3.29498500e-01 2.32539392e+00
1.07960606e+00 3.12005728e-01 1.41800359e-01 -1.30544052e-01
5.57917058e-01 -8.79475400e-02 -5.56021750e-01 -6.53420031e-01
-5.54362297e-01 2.78338939e-01 6.02186382e-01 8.90503109e-01
-6.89964235e-01 1.40261602e+00 7.24135113e+00 9.51860726e-01
-1.09538031e+00 4.09165472e-01 6.73906088e-01 -4.51906800e-01
-6.38755500e-01 -3.05581428e-02 -4.74996090e-01 1.94330662e-01
9.36650872e-01 -9.56818461e-01 8.47971022e-01 4.27693963e-01
1.54189497e-01 1.05461888e-01 -8.48704457e-01 6.24078810e-01
1.68110400e-01 -1.23665202e+00 7.98107088e-02 2.40116492e-01
7.18357384e-01 3.62406582e-01 4.47457097e-02 8.20371509e-01
8.40924203e-01 -1.05825412e+00 6.50530457e-01 2.89112568e-01
7.41617084e-01 -9.10515785e-01 4.59818572e-01 4.69272375e-01
-9.60675657e-01 -2.64275759e-01 -2.82056540e-01 -5.41623473e-01
-7.07954988e-02 7.24024251e-02 -6.85209811e-01 2.20536441e-01
4.66241241e-01 3.41351032e-01 -7.54687071e-01 8.43411863e-01
-9.03744400e-02 1.00009990e+00 -6.18872233e-02 -1.20185956e-01
6.36086985e-02 -2.32581031e-02 5.81967354e-01 1.07737994e+00
1.60551026e-01 1.14113703e-01 1.22746967e-01 6.57141447e-01
-3.53980988e-01 4.98099476e-01 -6.75595880e-01 -5.46666123e-02
4.98711765e-01 1.05344021e+00 -4.54785854e-01 -2.18701914e-01
-5.30379176e-01 8.71811748e-01 8.08833003e-01 4.18137014e-01
-6.86475039e-01 -2.36462146e-01 9.71742928e-01 1.97416451e-02
1.18335173e-01 -1.05748460e-01 -2.91721046e-01 -1.28845656e+00
-3.89632016e-01 -1.11602962e+00 3.76326144e-01 -7.22288907e-01
-1.27848566e+00 3.26536149e-01 -2.85782516e-01 -8.57176006e-01
-1.23640843e-01 -4.01911825e-01 -4.68031049e-01 1.05213082e+00
-1.02773726e+00 -9.22471881e-01 2.00215727e-01 3.43231767e-01
6.65234745e-01 -5.03062308e-01 7.69933999e-01 2.52693623e-01
-5.97514451e-01 1.30320299e+00 2.37202749e-01 2.15250224e-01
5.88415921e-01 -1.23537564e+00 4.95405495e-01 7.92712867e-01
3.66342664e-01 7.66668260e-01 6.60802722e-01 -3.43757868e-01
-1.27984035e+00 -7.25039065e-01 1.13823104e+00 -3.96693617e-01
7.99523592e-01 -6.49627507e-01 -8.44096899e-01 7.24986613e-01
5.06187022e-01 -3.90667409e-01 8.60188127e-01 8.02738905e-01
-5.04234195e-01 -7.33969435e-02 -8.94365013e-01 1.00004756e+00
1.39661002e+00 -6.57462120e-01 -4.43296015e-01 3.19623023e-01
1.07058907e+00 -1.41958743e-02 -6.70267940e-01 -8.18796828e-03
2.16742232e-01 -7.28996933e-01 8.29454124e-01 -9.53294039e-01
1.20838724e-01 2.32768998e-01 -1.60118490e-01 -1.75507128e+00
-8.47394228e-01 -7.12888360e-01 2.47492284e-01 1.03596509e+00
8.44243824e-01 -8.49759758e-01 7.10831642e-01 3.67614895e-01
-1.23426050e-01 -4.35628533e-01 -1.00857985e+00 -1.06815505e+00
8.84299517e-01 -4.62442040e-01 4.24935699e-01 1.27599347e+00
4.37295884e-01 8.80411327e-01 -3.18399847e-01 -2.94366091e-01
5.58131158e-01 -2.80583531e-01 4.87294823e-01 -1.22793233e+00
-3.80859226e-01 -9.32088315e-01 -7.67564401e-02 -8.27502489e-01
4.50109065e-01 -1.37090576e+00 -7.47380853e-02 -1.35712814e+00
5.62747300e-01 -5.61015487e-01 -5.44694185e-01 5.71572065e-01
-2.80983269e-01 8.60621706e-02 7.71748900e-01 1.25425950e-01
-7.96652138e-01 4.50551391e-01 1.15709496e+00 -3.70627314e-01
-1.94136962e-01 -3.39358538e-01 -1.31475127e+00 5.84721625e-01
9.90583539e-01 -4.10204053e-01 -5.24211824e-01 -5.99108696e-01
5.05723000e-01 -2.34053191e-02 -6.49127588e-02 -8.38150501e-01
2.83440888e-01 -2.61088729e-01 -1.05263099e-01 -4.14117910e-02
1.52266994e-01 -3.26497048e-01 -1.64727896e-01 6.83434486e-01
-6.04877710e-01 2.64280647e-01 6.88161075e-01 4.11978453e-01
1.86796069e-01 -1.21120840e-01 5.84912419e-01 -1.81208983e-01
-3.66628289e-01 -1.45157307e-01 -8.54278743e-01 8.55452776e-01
7.70421565e-01 6.34515733e-02 -4.20291543e-01 -5.94939351e-01
-5.58169007e-01 3.32925051e-01 4.26877767e-01 5.68579555e-01
8.84160176e-02 -1.06408513e+00 -9.98128712e-01 8.51037055e-02
2.81878799e-01 -6.79681718e-01 2.96688795e-01 8.70846033e-01
-2.34446958e-01 5.89956701e-01 -1.06334724e-01 -3.99908155e-01
-1.37087345e+00 5.54879487e-01 6.16529346e-01 -6.34765029e-01
-1.66175798e-01 1.06438851e+00 5.54049253e-01 -7.55239904e-01
-5.00212014e-02 -1.37449324e-01 -1.46064550e-01 5.49242459e-02
1.48968995e-01 1.41621545e-01 -1.38160528e-03 -3.47131133e-01
-5.01436114e-01 3.76993537e-01 -2.90775537e-01 -2.79927075e-01
1.02458203e+00 -1.35876372e-01 -5.28194495e-02 3.53045940e-01
7.88273156e-01 3.46573532e-01 -9.15618718e-01 -3.62804174e-01
1.54950067e-01 -6.43118843e-02 -8.48898292e-02 -1.17221868e+00
-1.04973388e+00 7.77745366e-01 2.78576463e-01 1.36371717e-01
1.04318786e+00 3.29230838e-02 4.02147412e-01 5.23625255e-01
5.35961092e-01 -1.00306094e+00 -6.83344081e-02 1.07804096e+00
7.46180177e-01 -1.02955270e+00 -1.97009832e-01 -7.24655166e-02
-5.99791527e-01 4.43077087e-01 9.15411592e-01 1.81708768e-01
5.13941169e-01 5.02017498e-01 -4.48851064e-02 1.66066229e-01
-1.28270352e+00 -2.27169484e-01 1.62136570e-01 6.69876933e-01
8.43867064e-01 2.74045229e-01 -4.22410607e-01 1.00580323e+00
-4.73539323e-01 -2.99455881e-01 6.52763307e-01 5.92965245e-01
-4.91757095e-01 -1.16926682e+00 -1.98225006e-01 7.65504539e-01
-3.86757225e-01 -9.30999100e-01 -2.50302970e-01 8.37223649e-01
-1.59837827e-01 8.33296597e-01 3.23430061e-01 -4.26349789e-01
3.46804678e-01 1.70079798e-01 5.26460767e-01 -7.92983830e-01
-1.14990211e+00 -1.43085271e-01 2.51011938e-01 -1.96347848e-01
1.79961592e-01 -6.83091164e-01 -8.54032636e-01 -6.04650438e-01
-5.16608119e-01 2.76793450e-01 1.18284889e-01 8.42922628e-01
5.46823084e-01 7.09897220e-01 2.72283107e-01 -2.93122143e-01
-7.39444137e-01 -1.06331110e+00 -3.92709196e-01 1.78534120e-01
1.13788851e-01 -4.91298020e-01 -5.55983901e-01 -3.68144840e-01] | [10.696216583251953, 8.460688591003418] |
b479f8b0-7eb7-43c3-ac1b-f0827d779a67 | temporalstereo-efficient-spatial-temporal | 2211.13755 | null | https://arxiv.org/abs/2211.13755v1 | https://arxiv.org/pdf/2211.13755v1.pdf | TemporalStereo: Efficient Spatial-Temporal Stereo Matching Network | We present TemporalStereo, a coarse-to-fine based online stereo matching network which is highly efficient, and able to effectively exploit the past geometry and context information to boost the matching accuracy. Our network leverages sparse cost volume and proves to be effective when a single stereo pair is given, however, its peculiar ability to use spatio-temporal information across frames allows TemporalStereo to alleviate problems such as occlusions and reflective regions while enjoying high efficiency also in the case of stereo sequences. Notably our model trained, once with stereo videos, can run in both single-pair and temporal ways seamlessly. Experiments show that our network relying on camera motion is even robust to dynamic objects when running on videos. We validate TemporalStereo through extensive experiments on synthetic (SceneFlow, TartanAir) and real (KITTI 2012, KITTI 2015) datasets. Detailed results show that our model achieves state-of-the-art performance on any of these datasets. Code is available at \url{https://github.com/youmi-zym/TemporalStereo.git}. | ['Stefano Mattoccia', 'Matteo Poggi', 'Youmin Zhang'] | 2022-11-24 | null | null | null | null | ['stereo-matching-1'] | ['computer-vision'] | [-2.08722159e-01 -2.99072385e-01 -6.50138855e-02 -1.26915202e-01
-6.79466784e-01 -7.61196911e-01 7.59460449e-01 -3.76778305e-01
-3.58861774e-01 4.21553493e-01 2.61691213e-01 -1.28222555e-01
-1.26326993e-01 -8.34649205e-01 -8.87020826e-01 -3.07886899e-01
-4.19129491e-01 3.90295804e-01 5.04471183e-01 -3.39133352e-01
1.48245022e-01 6.05897784e-01 -1.71768606e+00 3.21361721e-01
5.07472157e-01 9.81090188e-01 1.88928336e-01 8.14289391e-01
3.56955618e-01 8.82313848e-01 2.95477957e-01 -4.28791702e-01
9.89207149e-01 5.53657934e-02 -7.71800458e-01 2.98852902e-02
1.24451053e+00 -7.50683308e-01 -1.02215302e+00 6.47662878e-01
5.28389156e-01 2.58247852e-01 9.14955288e-02 -1.14076388e+00
-1.06434762e-01 1.29716083e-01 -4.83387798e-01 3.20859641e-01
6.30742490e-01 6.35860920e-01 1.11400127e+00 -1.24843514e+00
1.11265397e+00 1.03418183e+00 1.06706405e+00 2.06038594e-01
-1.17553711e+00 -7.13713646e-01 -1.02665231e-01 2.68995494e-01
-1.28529239e+00 -7.23099947e-01 5.36558926e-01 -4.56220716e-01
8.41107368e-01 1.56061947e-02 9.07127261e-01 9.49480832e-01
4.95953113e-02 7.64816403e-01 8.79539311e-01 -4.78667617e-02
-1.21802457e-01 -6.11670732e-01 -1.62842333e-01 7.28987694e-01
-1.91676497e-01 6.49330020e-01 -9.74773228e-01 -1.01005338e-01
1.17486084e+00 1.66906074e-01 -3.11629057e-01 -7.60805607e-01
-1.49580240e+00 6.45727694e-01 6.17216945e-01 2.19852686e-01
-3.50363880e-01 3.80272031e-01 3.72396380e-01 4.96607512e-01
4.44731414e-01 2.17691109e-01 -4.05483812e-01 -3.96992952e-01
-1.08948565e+00 4.84055787e-01 6.96413577e-01 1.02021718e+00
1.01558304e+00 -7.90264979e-02 3.65282707e-02 6.86607778e-01
-2.57741064e-01 4.31005657e-01 1.76662859e-02 -1.70575726e+00
5.94538271e-01 1.50375396e-01 7.99393058e-02 -1.11077380e+00
-2.88677335e-01 -2.38586575e-01 -7.62829185e-01 2.50548393e-01
6.48674607e-01 1.67484254e-01 -7.74733722e-01 1.65313017e+00
4.61376876e-01 4.83253866e-01 -4.06093121e-01 1.06061244e+00
8.01316738e-01 4.09585118e-01 -3.33984375e-01 2.16277823e-01
9.15184796e-01 -1.10656559e+00 -1.43594995e-01 -3.37509960e-01
5.05710125e-01 -9.12953854e-01 8.86206031e-01 2.44075388e-01
-1.25436163e+00 -5.01579642e-01 -8.03483009e-01 -3.14565837e-01
-1.67153403e-01 -3.18209112e-01 7.86525965e-01 6.61602318e-02
-1.42559147e+00 1.12335038e+00 -8.11648428e-01 -6.40204668e-01
6.04516029e-01 3.19948137e-01 -6.11017227e-01 -4.76968616e-01
-1.00847030e+00 5.53699672e-01 1.63634792e-01 6.17963560e-02
-9.58576739e-01 -1.13644063e+00 -8.52078378e-01 -1.30508244e-01
5.17583370e-01 -9.38073754e-01 1.28130543e+00 -8.87259722e-01
-1.17772663e+00 1.02729714e+00 -2.25536838e-01 -5.38894594e-01
1.02384150e+00 -3.52877587e-01 7.16697499e-02 4.57945317e-01
2.92007029e-01 1.02935016e+00 5.76771677e-01 -8.51701021e-01
-6.14963591e-01 -2.98797131e-01 4.56123143e-01 1.84457436e-01
-5.15767634e-02 -1.92085028e-01 -6.54301524e-01 -5.29860914e-01
2.32500970e-01 -1.04121172e+00 -1.86968535e-01 6.25691950e-01
-1.65111274e-01 1.44090474e-01 7.81043828e-01 -6.43255353e-01
8.00628662e-01 -2.01155233e+00 -2.36103758e-02 1.26703516e-01
2.12291420e-01 1.66086346e-01 -2.18448400e-01 5.81176877e-01
-1.52710648e-02 -3.06802958e-01 -2.17668951e-01 -4.53472555e-01
-1.15371987e-01 1.03525542e-01 -1.93108141e-01 7.71611512e-01
-2.15819955e-01 9.62966502e-01 -9.32808280e-01 -5.30715704e-01
6.97327971e-01 5.05741000e-01 -7.73313761e-01 2.31563464e-01
4.47530337e-02 5.99599004e-01 -1.12566866e-01 8.70979726e-01
7.39133060e-01 -3.76705796e-01 2.50495542e-02 -3.43246430e-01
-3.87717485e-01 2.62745112e-01 -1.27571964e+00 2.22709322e+00
-4.98943508e-01 9.48081255e-01 1.82603583e-01 -5.62057376e-01
3.04432392e-01 1.45015806e-01 7.57426977e-01 -1.01095855e+00
-7.99893066e-02 4.23781693e-01 -3.63747239e-01 -3.17887425e-01
5.44734299e-01 2.35912353e-01 3.01962465e-01 3.08286607e-01
1.06934883e-01 -3.50648127e-02 4.51748818e-01 4.83493447e-01
1.22841692e+00 4.95068818e-01 1.07748084e-01 -4.45578396e-01
3.67877394e-01 2.29509231e-02 5.52458346e-01 7.95639396e-01
-2.13638216e-01 9.87540543e-01 7.09327534e-02 -8.70742202e-01
-1.44908798e+00 -1.13477218e+00 -1.68085739e-01 9.33572173e-01
4.10550207e-01 -5.75603306e-01 -2.57975727e-01 -2.78491944e-01
1.58814639e-01 2.33137924e-02 -6.23865604e-01 3.50299865e-01
-8.20434332e-01 -1.19101860e-01 2.96467513e-01 5.78655958e-01
7.23788917e-01 -8.48821759e-01 -6.68372691e-01 2.87658781e-01
-3.32233369e-01 -1.40702462e+00 -7.51316726e-01 -3.32664013e-01
-9.59483564e-01 -1.28465080e+00 -7.30749071e-01 -4.15739566e-01
1.94658294e-01 7.21521139e-01 1.45886600e+00 3.63919258e-01
-4.56966370e-01 6.29051149e-01 -2.11627543e-01 2.97612458e-01
3.50161493e-02 8.33673552e-02 -1.13259554e-01 -1.21711679e-01
-3.14488932e-02 -1.19803464e+00 -9.91588712e-01 5.00723362e-01
-7.97759414e-01 3.56896132e-01 2.59893090e-02 7.36099899e-01
3.81825387e-01 -6.48441434e-01 -2.75827289e-01 -5.30420721e-01
-2.05798402e-01 -3.06954235e-01 -8.55039239e-01 -1.30556345e-01
-3.35497171e-01 -2.45856956e-01 5.42264342e-01 -1.18919060e-01
-7.56237507e-01 7.20912963e-02 -1.22370034e-01 -8.44811618e-01
8.86282697e-02 1.18511930e-01 4.04257178e-01 -4.85540181e-01
5.57589948e-01 1.36557534e-01 5.38814999e-02 -5.00447452e-01
2.67045736e-01 3.29228565e-02 7.77918398e-01 -6.03115439e-01
1.03325009e+00 1.16649854e+00 1.43516719e-01 -5.45953512e-01
-8.82050574e-01 -7.39576399e-01 -8.76938581e-01 -4.48977053e-01
6.42571747e-01 -1.20742118e+00 -8.62907767e-01 5.78861654e-01
-1.01714063e+00 -6.69449091e-01 -1.40606388e-01 2.77505338e-01
-8.00777555e-01 4.82332468e-01 -8.84984672e-01 -3.94920677e-01
-3.06995392e-01 -9.31427538e-01 1.15214872e+00 6.49759248e-02
-1.25963483e-02 -1.14450681e+00 1.36567697e-01 3.44350457e-01
4.30956364e-01 4.32771832e-01 5.36100380e-02 1.88560449e-02
-1.27444780e+00 6.71812519e-02 -3.29769492e-01 -8.74677375e-02
-1.15070246e-01 1.24736145e-01 -1.01654840e+00 -5.69001675e-01
-4.25027460e-01 -4.69092727e-01 9.35366690e-01 2.95981765e-01
1.00019193e+00 -1.59771740e-01 -9.98632535e-02 1.20097566e+00
1.72774601e+00 -2.90140569e-01 7.52968848e-01 7.21615672e-01
8.31736863e-01 5.58433473e-01 6.68956220e-01 5.07002831e-01
5.34759164e-01 1.00332725e+00 6.42715633e-01 -1.37974530e-01
-3.45082819e-01 -3.95587832e-01 2.44348049e-01 5.61163187e-01
-3.91541153e-01 6.34120926e-02 -9.19172585e-01 8.01050723e-01
-2.07075572e+00 -1.26151991e+00 -2.40661770e-01 2.17512798e+00
5.20801306e-01 -1.29089773e-01 1.86078936e-01 -1.57753095e-01
5.29416621e-01 5.26774287e-01 -4.16087240e-01 8.31337273e-02
-4.57205474e-01 1.50360644e-01 7.58314192e-01 5.32781184e-01
-1.13322282e+00 9.57816362e-01 5.74079657e+00 6.82739675e-01
-9.93677437e-01 2.30171472e-01 5.02284408e-01 -3.76730174e-01
-1.39012396e-01 2.23851979e-01 -2.80348808e-01 3.70902717e-01
5.40301144e-01 -5.20392880e-02 6.93392038e-01 6.41314745e-01
2.16368929e-01 -2.04747632e-01 -1.21756566e+00 1.18961346e+00
-1.92720205e-01 -1.91566432e+00 -2.06564054e-01 1.74069718e-01
8.68423223e-01 6.52185977e-01 -8.01420398e-03 -4.06275578e-02
2.60317266e-01 -9.05044138e-01 8.78776073e-01 3.38740408e-01
8.20769608e-01 -5.64489961e-01 4.53968346e-01 1.67534143e-01
-1.58491278e+00 1.77000146e-02 -1.86503395e-01 -1.70315117e-01
3.05451155e-01 5.66368341e-01 -3.79937023e-01 7.75806308e-01
1.15752888e+00 1.31528389e+00 -5.97368479e-01 1.12777126e+00
1.97870672e-01 1.38416544e-01 -6.14129066e-01 6.31648839e-01
4.53997403e-01 -3.52123618e-01 6.66574836e-01 1.14294696e+00
3.81604970e-01 1.30315304e-01 2.57653177e-01 5.91325819e-01
-3.05859894e-02 -6.24066740e-02 -9.28110123e-01 4.35266435e-01
4.46547538e-01 1.08932734e+00 -4.81506139e-01 -3.77400517e-01
-4.97769505e-01 1.02742052e+00 4.68261838e-01 2.68651992e-01
-7.42481947e-01 9.47265327e-02 8.75126600e-01 4.15302515e-01
5.52383900e-01 -3.64942372e-01 -1.43121272e-01 -1.48816335e+00
3.21172684e-01 -8.22918296e-01 4.98459399e-01 -9.32264149e-01
-1.13837945e+00 3.73397946e-01 -2.67047919e-02 -1.52162898e+00
-2.10144207e-01 -4.96717393e-01 -5.63406110e-01 5.42899251e-01
-1.56483853e+00 -1.11428630e+00 -6.46111906e-01 1.00368774e+00
6.71867847e-01 7.83598870e-02 2.99683541e-01 6.92661166e-01
-1.85714662e-01 3.60453993e-01 1.38561353e-01 1.97021246e-01
7.28702068e-01 -8.77695739e-01 6.25486135e-01 1.02323425e+00
1.35173082e-01 3.91920358e-01 7.66991198e-01 -3.86347950e-01
-1.56695867e+00 -8.44470859e-01 7.78246701e-01 -4.88904923e-01
7.48399556e-01 -3.41176778e-01 -7.38306344e-01 8.23222160e-01
1.77609816e-01 3.29560131e-01 1.99391525e-02 1.86853856e-02
-6.34971797e-01 -2.66186684e-01 -9.99640882e-01 6.60787344e-01
1.71736693e+00 -7.36651003e-01 -2.34973714e-01 4.84133154e-01
6.27220094e-01 -9.07807171e-01 -9.64085281e-01 4.62612987e-01
8.14909458e-01 -1.89805937e+00 1.22847760e+00 -2.00821310e-01
6.47938490e-01 -2.60494679e-01 -2.58198559e-01 -8.67594600e-01
-2.76854903e-01 -9.75077868e-01 -1.47433624e-01 8.72740030e-01
5.16339652e-02 -7.17002392e-01 8.25218379e-01 4.79589581e-01
-7.19219148e-02 -6.44431233e-01 -1.17534101e+00 -1.07044756e+00
-1.95762739e-01 -6.05317473e-01 6.55650675e-01 9.48371053e-01
-4.02137697e-01 -2.08089843e-01 -6.26820028e-01 -1.49465464e-02
9.71720934e-01 2.84148008e-01 1.12708461e+00 -9.97517765e-01
-3.38464946e-01 -2.84309477e-01 -5.30161977e-01 -1.33379698e+00
-4.83296253e-02 -7.04988003e-01 -2.24678218e-01 -1.09835589e+00
1.88016042e-01 -4.40834969e-01 1.33662313e-01 3.16273838e-01
2.15447813e-01 7.47994244e-01 5.28999150e-01 5.16151965e-01
-5.78766465e-01 4.41021383e-01 1.32313561e+00 1.37739733e-01
5.13508841e-02 -2.20986828e-01 -1.73988547e-02 5.58851600e-01
5.96496940e-01 -2.71151721e-01 -4.44121324e-02 -6.23267114e-01
2.63428479e-01 2.98819423e-01 8.90864849e-01 -1.26932371e+00
4.03830409e-01 -5.25724478e-02 1.45601720e-01 -5.75408161e-01
6.50049448e-01 -7.21547604e-01 5.03703952e-01 3.57180983e-01
8.47368687e-03 4.38877106e-01 1.59662932e-01 4.74411547e-01
-1.97701991e-01 2.26903796e-01 8.30703616e-01 -3.44768763e-01
-9.81188178e-01 9.48675990e-01 1.23119064e-01 3.20242763e-01
6.79180086e-01 -5.59501708e-01 -2.04594851e-01 -6.42785251e-01
-4.78528321e-01 3.91339123e-01 1.05044687e+00 4.27592188e-01
3.68485868e-01 -1.50315320e+00 -6.52840912e-01 1.24477156e-01
2.43217871e-01 6.13350123e-02 5.92330158e-01 1.06145668e+00
-8.74598503e-01 4.18351799e-01 -2.36584574e-01 -1.03711689e+00
-1.22620559e+00 3.29515994e-01 6.21883214e-01 -2.07250610e-01
-1.06582332e+00 6.61723971e-01 3.79700363e-01 -4.39066678e-01
2.15347528e-01 5.18727973e-02 3.64953816e-01 -1.98594555e-01
3.81628811e-01 4.96355593e-01 -2.18663029e-02 -6.77244544e-01
-3.41870606e-01 9.70202863e-01 1.96310133e-01 -1.10947050e-01
1.46530128e+00 -2.70104468e-01 -1.01876780e-01 1.64040834e-01
1.47704732e+00 -1.97384641e-01 -1.77372015e+00 -3.74242723e-01
-2.79263824e-01 -1.00736094e+00 -1.54769033e-01 -3.78657877e-01
-1.29419434e+00 7.01789856e-01 4.15556163e-01 -9.36568305e-02
9.29969192e-01 -6.82277009e-02 1.11133754e+00 2.44562790e-01
8.11614215e-01 -9.70534503e-01 2.30124872e-02 7.58194685e-01
7.33931303e-01 -1.37149751e+00 4.63822819e-02 -4.61453706e-01
-1.96632966e-01 1.31090415e+00 5.93887866e-01 -3.99860948e-01
7.26795256e-01 8.27497616e-02 -4.65488899e-03 -2.91176379e-01
-7.69941807e-01 -2.18008026e-01 1.72062114e-01 3.90395403e-01
1.74450681e-01 -2.20986128e-01 8.55066702e-02 -5.67162097e-01
-2.65998036e-01 9.03295875e-02 4.21496272e-01 1.05700541e+00
-7.48833865e-02 -1.05975878e+00 -2.54402429e-01 2.38665968e-01
-2.95391500e-01 -2.61495322e-01 -5.08688278e-02 1.05367422e+00
-3.76618095e-02 6.27431691e-01 2.04827383e-01 -2.96345860e-01
3.29246193e-01 -3.95401388e-01 5.80258787e-01 -1.84331626e-01
-7.58437574e-01 -3.88845876e-02 2.02604845e-01 -1.44804895e+00
-7.84100652e-01 -8.02567303e-01 -8.12846780e-01 -8.70323896e-01
2.28309914e-01 -3.10455322e-01 2.59931445e-01 7.34295309e-01
5.67327321e-01 -1.00608893e-01 7.30573654e-01 -1.42243087e+00
-1.65915802e-01 -5.67532957e-01 -3.68251383e-01 7.14635491e-01
5.20788312e-01 -6.59290493e-01 -4.70634192e-01 -7.79692903e-02] | [8.578851699829102, -2.064683675765991] |
39727383-fec1-4b1e-88fa-0109b6089e56 | belief-revision-based-caption-re-ranker-with | 2209.08163 | null | https://arxiv.org/abs/2209.08163v1 | https://arxiv.org/pdf/2209.08163v1.pdf | Belief Revision based Caption Re-ranker with Visual Semantic Information | In this work, we focus on improving the captions generated by image-caption generation systems. We propose a novel re-ranking approach that leverages visual-semantic measures to identify the ideal caption that maximally captures the visual information in the image. Our re-ranker utilizes the Belief Revision framework (Blok et al., 2003) to calibrate the original likelihood of the top-n captions by explicitly exploiting the semantic relatedness between the depicted caption and the visual context. Our experiments demonstrate the utility of our approach, where we observe that our re-ranker can enhance the performance of a typical image-captioning system without the necessity of any additional training or fine-tuning. | ['Lluís Padró', 'Pranava Madhyastha', 'Francesc Moreno-Noguer', 'Ahmed Sabir'] | 2022-09-16 | null | https://aclanthology.org/2022.coling-1.487 | https://aclanthology.org/2022.coling-1.487.pdf | coling-2022-10 | ['visual-reasoning', 'visual-reasoning', 'natural-language-visual-grounding'] | ['computer-vision', 'reasoning', 'reasoning'] | [ 4.57901597e-01 3.93335938e-01 -1.21647611e-01 -2.74559140e-01
-1.09268868e+00 -8.34456861e-01 9.55484807e-01 -2.01656595e-02
-1.23172626e-01 5.93258977e-01 4.85110134e-01 -2.87733823e-01
1.71555310e-01 -2.26275608e-01 -1.10082603e+00 -1.71306819e-01
3.91223341e-01 4.28873807e-01 2.67206252e-01 -1.56706646e-01
2.78505683e-01 1.58677205e-01 -1.60294878e+00 6.55967712e-01
7.32595503e-01 7.48605847e-01 5.51123559e-01 8.10705841e-01
1.56434461e-01 5.06488562e-01 -6.04278922e-01 -7.07527220e-01
1.02716357e-01 -4.58476603e-01 -7.04811692e-01 4.04335916e-01
9.98382807e-01 -3.52190644e-01 -4.27528292e-01 9.71185923e-01
1.60230845e-01 8.01434517e-02 7.88560450e-01 -1.25461912e+00
-9.55197215e-01 4.35683399e-01 -5.65427959e-01 1.89177230e-01
4.37099636e-01 1.14141360e-01 1.24212027e+00 -1.07053077e+00
8.19761693e-01 1.32765162e+00 1.69509441e-01 5.20757377e-01
-1.35996497e+00 -3.81755024e-01 3.55008274e-01 3.60296786e-01
-1.17976832e+00 -4.83982563e-01 9.03498530e-01 -5.83290339e-01
3.43852997e-01 1.61716595e-01 4.68052924e-01 1.26461613e+00
-3.61312449e-01 1.00670838e+00 8.82336855e-01 -6.46800756e-01
3.39418985e-02 1.76915407e-01 -1.83422834e-01 5.59470892e-01
3.34350467e-01 -5.92625812e-02 -4.95885640e-01 -2.64153540e-01
7.81660914e-01 -2.87180364e-01 -3.32797945e-01 -8.64198327e-01
-1.16167092e+00 8.29600215e-01 5.58858752e-01 1.84372365e-01
-4.48179126e-01 4.96881366e-01 9.94801149e-02 -3.80645156e-01
2.86017835e-01 7.67700493e-01 -1.12887993e-01 1.28110185e-01
-1.05007255e+00 5.56623414e-02 3.26542735e-01 8.59644473e-01
6.48789823e-01 -2.39106581e-01 -6.26578033e-01 7.84206390e-01
5.28272629e-01 7.34074473e-01 2.47317478e-01 -1.23383319e+00
3.68407518e-01 2.69295841e-01 4.69590306e-01 -1.00839365e+00
2.91258603e-01 -3.51325780e-01 -2.98028409e-01 -2.22332537e-01
-5.08943386e-02 1.17763519e-01 -1.18691850e+00 1.75380659e+00
3.94938439e-02 2.97519088e-01 8.52790400e-02 9.26811218e-01
6.14201784e-01 6.63165629e-01 4.65095580e-01 5.08335344e-02
1.37351072e+00 -1.26882935e+00 -5.25862157e-01 -5.87764621e-01
7.99607858e-02 -8.45011771e-01 1.11362886e+00 -1.05986081e-01
-9.76730645e-01 -5.90195954e-01 -1.06740999e+00 1.01466037e-01
-2.68168328e-03 3.25188011e-01 4.49736059e-01 2.72739261e-01
-1.36826980e+00 2.33790100e-01 -3.28774571e-01 -4.42159235e-01
3.28088701e-01 -4.22531739e-02 -3.23334634e-01 -2.91881740e-01
-1.05001390e+00 9.13004339e-01 4.49354053e-01 -7.62532055e-02
-1.16191602e+00 -4.48939472e-01 -8.48269761e-01 -2.09136121e-02
3.11461091e-01 -1.05777216e+00 1.51239622e+00 -1.36009085e+00
-1.22173858e+00 8.67131233e-01 -4.13500667e-01 -3.34808946e-01
3.72765243e-01 -2.47272521e-01 -8.97120535e-02 5.20775259e-01
2.76226580e-01 1.31826544e+00 9.72589791e-01 -2.08575058e+00
-5.57436824e-01 1.65868074e-01 2.80326426e-01 5.61182201e-01
-2.79579490e-01 -1.75793305e-01 -9.41188157e-01 -5.38769782e-01
-1.37991115e-01 -1.13833940e+00 -2.52597243e-01 -7.75525495e-02
-2.82306761e-01 5.44302166e-03 5.47848523e-01 -5.81398249e-01
9.55998302e-01 -2.13096452e+00 9.79880616e-03 2.12449446e-01
1.22479863e-01 2.91976392e-01 -6.48684502e-01 4.13287789e-01
4.44935486e-02 2.39030704e-01 -2.08691776e-01 -3.11166644e-01
-1.29185200e-01 1.78987026e-01 -4.58621502e-01 9.34179723e-02
2.55057216e-01 1.21260393e+00 -1.28639460e+00 -7.67384350e-01
3.71714473e-01 5.44712245e-01 -3.42761189e-01 3.89129639e-01
-5.86122274e-01 2.03764334e-01 -3.26921523e-01 1.96976811e-01
4.32050973e-01 -7.99226999e-01 2.92291611e-01 -5.79480052e-01
2.63477921e-01 3.30870077e-02 -5.21848798e-01 1.61348283e+00
-5.48226118e-01 8.36930215e-01 -2.53628016e-01 -5.02830446e-01
6.99832499e-01 1.95280999e-01 2.91478306e-01 -5.40523767e-01
-1.71223611e-01 1.53093869e-02 -4.16097939e-01 -4.22328800e-01
6.99686468e-01 -1.26605947e-02 -1.32947171e-03 4.00743306e-01
-8.51780996e-02 -1.02618426e-01 3.05317730e-01 6.27237618e-01
6.46793127e-01 2.54363775e-01 2.29934484e-01 1.73317254e-01
4.16030884e-01 1.58824310e-01 -1.30527496e-01 1.08475113e+00
-1.72130182e-01 8.98985744e-01 2.54062057e-01 -1.45015158e-02
-1.41711938e+00 -1.29885340e+00 2.12633640e-01 1.31040049e+00
3.40153486e-01 -1.97325140e-01 -6.75140679e-01 -7.48933136e-01
-1.50242567e-01 9.53977585e-01 -6.86722994e-01 1.88844334e-02
-3.35617959e-01 -1.73517317e-01 2.66267359e-01 3.95847678e-01
1.74290434e-01 -1.12059641e+00 -3.16351116e-01 -1.76059842e-01
-7.65263617e-01 -1.42058611e+00 -6.58928573e-01 -2.74419159e-01
-7.36204326e-01 -9.08691645e-01 -9.73911524e-01 -9.80460882e-01
1.05350018e+00 6.16438746e-01 1.30003560e+00 7.20344633e-02
1.80224687e-01 7.87746310e-01 -4.24975812e-01 -8.80066678e-02
-5.96453011e-01 -4.50109765e-02 -3.51902515e-01 -5.01558371e-03
-4.49464507e-02 -1.82844818e-01 -7.82484233e-01 2.22171277e-01
-9.52018321e-01 5.94399571e-01 9.16813970e-01 6.32750988e-01
5.89858294e-01 -5.74843585e-01 3.79455417e-01 -6.57388687e-01
5.25232494e-01 -3.70472789e-01 -4.49943334e-01 6.36622012e-01
-4.75231051e-01 3.06857735e-01 2.14964971e-01 -4.76119846e-01
-8.99861872e-01 3.56489122e-01 3.95233840e-01 -6.10041559e-01
7.31941164e-02 3.75605702e-01 -8.35790113e-02 -3.06165852e-02
5.33931136e-01 4.03521955e-01 -2.73776323e-01 -2.45702744e-01
1.08568180e+00 5.93858302e-01 9.19622004e-01 -6.56814575e-01
9.63616073e-01 5.68551600e-01 -1.83508739e-01 -4.24554497e-01
-1.24189246e+00 -5.94433069e-01 -3.68138343e-01 -5.70390165e-01
9.55229759e-01 -1.11842203e+00 -3.95519614e-01 -1.34342015e-01
-1.57541132e+00 2.76251156e-02 -8.36383477e-02 3.06173593e-01
-6.61980271e-01 6.37250781e-01 -2.00787067e-01 -5.88748455e-01
-3.29621881e-01 -1.08356631e+00 1.41181076e+00 1.05133668e-01
-1.06465071e-01 -7.00262487e-01 2.20863625e-01 7.31967926e-01
1.65839672e-01 1.32050991e-01 7.35587776e-01 -4.70276505e-01
-8.63114536e-01 -1.94272511e-02 -7.49743700e-01 2.19302610e-01
-8.29859301e-02 -7.27950186e-02 -1.04003572e+00 -1.82753250e-01
-4.99256343e-01 -3.72166544e-01 8.33763957e-01 3.06887835e-01
9.17301714e-01 -4.30375248e-01 -2.67487735e-01 2.35259533e-01
1.37795126e+00 -1.18977390e-01 5.69977641e-01 4.99535829e-01
8.83594692e-01 6.09740555e-01 5.36733627e-01 1.20582476e-01
5.64128637e-01 7.70749748e-01 4.73723292e-01 -2.25831300e-01
-3.26693416e-01 -7.77668595e-01 3.35050702e-01 3.76351893e-01
2.73455679e-01 -6.09521866e-01 -7.47620165e-01 8.69109869e-01
-2.13002896e+00 -9.48659241e-01 2.37368107e-01 2.01329398e+00
7.97490120e-01 -1.37788117e-01 -1.08242445e-01 -5.37206411e-01
1.23216105e+00 2.19327614e-01 -4.66808677e-01 -1.27675638e-01
-1.02263056e-01 -1.96804434e-01 5.05829215e-01 6.57420874e-01
-1.06498861e+00 1.20831752e+00 7.37949276e+00 6.60997152e-01
-6.85243845e-01 -1.13197245e-01 6.79890275e-01 8.01182613e-02
-6.20925903e-01 1.56319588e-01 -4.64398980e-01 3.02674264e-01
7.85735488e-01 -3.76611352e-01 3.37289542e-01 8.71424377e-01
1.42810509e-01 -2.92561818e-02 -1.25989258e+00 1.13439965e+00
6.80698097e-01 -1.28811252e+00 7.17583179e-01 2.41646431e-02
8.73818994e-01 -1.13631673e-01 3.32885593e-01 -1.65401082e-02
5.40648997e-01 -9.53054190e-01 8.76855195e-01 4.33356404e-01
6.81804657e-01 -3.53756487e-01 5.09113550e-01 -1.98619068e-01
-7.94515848e-01 2.24292696e-01 -1.16784416e-01 2.89093167e-01
5.18963158e-01 4.36305165e-01 -1.31677365e+00 2.38708287e-01
3.11041027e-01 4.60690171e-01 -8.37847054e-01 1.02995634e+00
-5.11385083e-01 3.23772848e-01 -6.94802254e-02 -1.14720454e-02
2.52046317e-01 2.23064259e-01 5.49535096e-01 1.05437028e+00
2.09951401e-01 4.02049609e-02 2.27186993e-01 6.15492404e-01
-2.68670380e-01 1.06894277e-01 -4.10673738e-01 -1.45112038e-01
3.54816973e-01 1.14144027e+00 -5.75312495e-01 -5.50556898e-01
-1.84738129e-01 1.21297407e+00 3.45978051e-01 4.94012713e-01
-8.47088754e-01 1.72804728e-01 3.83714139e-01 2.28103638e-01
6.98601544e-01 -1.04015023e-01 -3.98515686e-02 -1.07774830e+00
4.01408896e-02 -7.92595327e-01 2.81565756e-01 -1.61278260e+00
-1.18447030e+00 7.31355011e-01 3.31699610e-01 -1.06092083e+00
-4.07608986e-01 -5.50404668e-01 -1.79695919e-01 4.76564944e-01
-1.60420728e+00 -1.41127324e+00 -2.99516618e-01 1.91465154e-01
4.20644611e-01 2.48505384e-01 4.26160127e-01 -8.83632004e-02
-1.77318826e-01 4.37961847e-01 7.95501843e-03 -7.02143684e-02
8.68341386e-01 -1.07762086e+00 4.87165242e-01 8.83926749e-01
4.91406113e-01 4.66422230e-01 1.05268264e+00 -7.01714873e-01
-8.24694216e-01 -1.02409136e+00 8.99415433e-01 -6.08444333e-01
7.06128001e-01 -1.77064985e-01 -5.35322428e-01 5.02139926e-01
4.47974831e-01 -2.63660133e-01 6.28763258e-01 1.84638780e-02
-6.78802192e-01 1.15997590e-01 -7.24093378e-01 8.54516745e-01
9.17949915e-01 -7.36447573e-01 -8.05087566e-01 3.06450278e-01
7.46860206e-01 -1.18710317e-01 -3.52432609e-01 3.38570684e-01
4.68882889e-01 -6.67463303e-01 1.23743784e+00 -4.09425229e-01
7.20327079e-01 -5.01918077e-01 -2.12695315e-01 -1.18690705e+00
-5.81369519e-01 -3.21036577e-01 -8.57063569e-03 9.63039815e-01
5.44193923e-01 2.91946065e-02 7.80887902e-01 5.92228174e-01
6.83410540e-02 -3.21642458e-01 -6.21123016e-01 -5.40812612e-01
-4.27134097e-01 -9.36394334e-02 3.93044263e-01 5.35512328e-01
-8.31237212e-02 6.86318815e-01 -6.41164541e-01 2.70226836e-01
6.89620614e-01 2.04379901e-01 7.96634793e-01 -1.01020026e+00
-4.42974299e-01 -2.23077804e-01 -3.28996390e-01 -1.19868171e+00
1.68275222e-01 -7.29168177e-01 5.31117737e-01 -1.89568496e+00
8.55242014e-01 6.79607689e-02 -5.41774035e-01 4.82299596e-01
-5.12163281e-01 5.54330230e-01 6.03639007e-01 4.96053904e-01
-1.35007024e+00 5.64746737e-01 1.39316750e+00 -2.45162070e-01
6.38126433e-02 -5.45583248e-01 -1.07082653e+00 5.66606760e-01
4.92084563e-01 -4.92582172e-01 -6.38551831e-01 -7.17783689e-01
1.98461279e-01 -6.58161789e-02 6.15139782e-01 -7.35702455e-01
3.33515853e-02 -2.45789126e-01 4.44531232e-01 -5.29926956e-01
4.25557613e-01 -6.59245074e-01 8.87761489e-02 1.55416667e-01
-6.76254511e-01 1.80809855e-01 2.24324912e-01 8.85065317e-01
-1.61582455e-01 -4.29555804e-01 6.51799917e-01 -6.14865497e-02
-9.05337334e-01 8.62048492e-02 -2.01603413e-01 6.11822680e-02
1.08071554e+00 -1.40550479e-01 -4.61928636e-01 -8.21480036e-01
-5.59740424e-01 2.23681241e-01 8.16854358e-01 6.51382804e-01
7.13863432e-01 -1.44135344e+00 -7.89047480e-01 -3.79948139e-01
6.11908674e-01 -3.07977319e-01 8.41131508e-02 5.60138077e-02
-3.48263711e-01 4.47784424e-01 -1.32656038e-01 -6.24047995e-01
-1.24987376e+00 6.90529644e-01 7.70031363e-02 -3.32570702e-01
-2.15964362e-01 5.94464064e-01 6.07732773e-01 2.37089455e-01
6.57896101e-02 2.12515444e-01 -2.50142455e-01 -2.15129748e-01
3.71237814e-01 -1.35768726e-01 -3.84057581e-01 -8.85518134e-01
-3.84569287e-01 4.00751740e-01 -3.86835426e-01 -7.84746766e-01
1.06051314e+00 -3.38362902e-01 1.80346191e-01 9.55426842e-02
1.17743731e+00 -1.00543410e-01 -1.36581612e+00 -8.40811431e-02
-6.04979657e-02 -4.70547348e-01 1.21340953e-01 -9.28374290e-01
-6.64277792e-01 4.58930075e-01 4.55999911e-01 -3.02243084e-01
8.90281439e-01 4.33440119e-01 7.25959241e-01 3.68474007e-01
1.33297458e-01 -8.35147560e-01 6.70859694e-01 3.41754913e-01
9.59863663e-01 -1.26761901e+00 8.00497606e-02 -3.63153875e-01
-1.04303610e+00 8.00458133e-01 6.82004809e-01 2.22438900e-03
1.73120797e-01 -4.02689040e-01 2.78101027e-01 -1.04683042e-01
-7.49167442e-01 -4.08999383e-01 7.46874392e-01 7.12332129e-01
2.34440863e-01 -9.02973786e-02 -1.84376538e-01 2.43903503e-01
4.79458794e-02 -1.04298480e-01 5.98281562e-01 6.55140996e-01
-7.13653147e-01 -9.80331957e-01 -3.92186284e-01 9.25318152e-02
-7.74720535e-02 -3.99214625e-01 -6.67866468e-01 2.81038880e-01
-1.56252205e-01 1.07587755e+00 1.07730612e-01 -3.06675434e-01
1.36835814e-01 -3.51446010e-02 5.52765131e-01 -6.71605110e-01
-9.30910408e-02 9.07250345e-02 2.96595171e-02 -4.01314855e-01
-5.81908882e-01 -4.65990394e-01 -7.90015519e-01 2.67835677e-01
-2.29666337e-01 3.36303353e-01 7.65306532e-01 6.85800314e-01
5.47691703e-01 1.76703990e-01 6.26351595e-01 -8.80835235e-01
-2.72071958e-01 -7.14435339e-01 2.11258143e-01 8.29339087e-01
2.75870919e-01 -6.42090380e-01 -4.60575461e-01 4.99038488e-01] | [10.992963790893555, 1.1015948057174683] |
b11ef625-775e-4622-98eb-a5476a6208bf | deep-learning-based-detection-of-the-acute | 2109.12323 | null | https://arxiv.org/abs/2109.12323v1 | https://arxiv.org/pdf/2109.12323v1.pdf | Deep Learning-Based Detection of the Acute Respiratory Distress Syndrome: What Are the Models Learning? | The acute respiratory distress syndrome (ARDS) is a severe form of hypoxemic respiratory failure with in-hospital mortality of 35-46%. High mortality is thought to be related in part to challenges in making a prompt diagnosis, which may in turn delay implementation of evidence-based therapies. A deep neural network (DNN) algorithm utilizing unbiased ventilator waveform data (VWD) may help to improve screening for ARDS. We first show that a convolutional neural network-based ARDS detection model can outperform prior work with random forest models in AUC (0.95+/-0.019 vs. 0.88+/-0.064), accuracy (0.84+/-0.026 vs 0.80+/-0.078), and specificity (0.81+/-0.06 vs 0.71+/-0.089). Frequency ablation studies imply that our model can learn features from low frequency domains typically used for expert feature engineering, and high-frequency information that may be difficult to manually featurize. Further experiments suggest that subtle, high-frequency components of physiologic signals may explain the superior performance of DL models over traditional ML when using physiologic waveform data. Our observations may enable improved interpretability of DL-based physiologic models and may improve the understanding of how high-frequency information in physiologic data impacts the performance our DL model. | ['Jason Adams', 'Chen-Nee Chuah', 'Irene Cortes-Puch', 'Chao Wang', 'Gregory B. Rehm'] | 2021-09-25 | null | null | null | null | ['respiratory-failure'] | ['medical'] | [-7.98811391e-02 -3.83366823e-01 -1.14636414e-01 -2.03928709e-01
-4.96997327e-01 -4.48797584e-01 -2.65829772e-01 3.21032852e-01
-3.33489835e-01 7.68696368e-01 2.90057003e-01 -8.47934008e-01
-3.76175195e-01 -7.00250506e-01 -4.37441885e-01 -4.25597250e-01
-4.79688376e-01 4.12052333e-01 -6.11807033e-02 5.37680015e-02
-5.95434755e-02 8.11986208e-01 -1.00188529e+00 3.93910825e-01
6.44087970e-01 1.05356216e+00 -9.67310295e-02 1.07697272e+00
-2.07610168e-02 7.10450888e-01 -8.13238800e-01 5.09713769e-01
1.94694832e-01 -7.06518829e-01 -2.57752597e-01 -8.11646283e-01
3.55806082e-01 -7.69773364e-01 -4.68848377e-01 2.63211817e-01
1.07595253e+00 -6.49200287e-03 8.77577245e-01 -9.55271125e-01
-3.38722654e-02 2.43695930e-01 -1.86224103e-01 7.54046559e-01
2.17772368e-02 4.91159171e-01 6.15354180e-01 -5.90480983e-01
-6.78414702e-02 7.46802151e-01 1.19790304e+00 7.16519415e-01
-1.33580780e+00 -8.46186638e-01 -4.34762061e-01 9.54765752e-02
-1.29096651e+00 -2.74296522e-01 3.46843004e-01 -6.39298260e-01
1.12221706e+00 1.70543134e-01 8.28390002e-01 9.68203008e-01
6.08412862e-01 -1.10897273e-01 6.46095216e-01 -1.01354279e-01
-8.49982798e-02 -1.01372235e-01 8.41758102e-02 8.19606423e-01
8.44726503e-01 3.66273910e-01 -3.77396733e-01 -4.24196273e-01
1.03668404e+00 2.60450721e-01 -7.75869429e-01 -6.50295839e-02
-9.63703156e-01 8.62795830e-01 3.55881453e-01 8.95856842e-02
-5.08831024e-01 3.55642796e-01 3.49854916e-01 1.99075371e-01
-1.49165750e-01 7.98492014e-01 -7.33564675e-01 -3.46115619e-01
-7.19516039e-01 1.02432422e-01 9.23294365e-01 3.40040445e-01
1.98988527e-01 3.98610055e-01 -2.50884533e-01 7.76661992e-01
3.25885206e-01 6.57608569e-01 5.10435760e-01 -1.18219101e+00
1.42006263e-01 2.06020400e-01 2.83530146e-01 -6.76058352e-01
-1.03952289e+00 -7.23744869e-01 -8.46063912e-01 5.02257235e-02
3.98187310e-01 -5.07983565e-01 -6.83970928e-01 1.60988367e+00
-1.57115936e-01 2.00446412e-01 1.00852966e-01 7.72787690e-01
8.57607484e-01 2.68810689e-01 1.63701400e-01 -4.69439477e-01
1.24006045e+00 -2.79924423e-01 -5.42189181e-01 4.55168895e-02
5.95749140e-01 -4.34053957e-01 1.09514999e+00 2.70497680e-01
-8.98099244e-01 -2.95878321e-01 -1.06859696e+00 3.41527551e-01
2.29515344e-01 -9.58559439e-02 8.92704204e-02 5.67699194e-01
-9.42414165e-01 8.01797271e-01 -1.05102444e+00 -4.29965138e-01
6.44367576e-01 4.95320201e-01 1.63642824e-01 -1.93945188e-02
-1.27900934e+00 1.09289277e+00 -9.80521813e-02 -3.04180741e-01
-6.96901679e-01 -1.41476631e+00 -5.15108645e-01 3.16136420e-01
-3.00713837e-01 -1.25951135e+00 1.31617057e+00 -1.53059781e-01
-9.23866987e-01 2.15356708e-01 -2.60319889e-01 -5.25283098e-01
2.41774261e-01 -4.73226905e-01 -4.33679849e-01 7.23310590e-01
-1.43078700e-01 2.29877710e-01 4.52956766e-01 -8.15375507e-01
-4.89609927e-01 -4.36287411e-02 -3.93404603e-01 1.21123972e-03
-3.11757326e-01 -2.43810743e-01 5.21269977e-01 -4.85136986e-01
1.73835494e-02 -6.31058633e-01 -1.52877301e-01 1.64296478e-01
-1.40088191e-02 9.33483392e-02 7.96625853e-01 -5.79964101e-01
1.39562857e+00 -1.88281846e+00 -6.92863941e-01 5.39212078e-02
8.82994413e-01 3.28410089e-01 3.04421723e-01 4.85017210e-01
-1.45689160e-01 4.04092640e-01 -1.24275880e-02 3.49181235e-01
-4.98086661e-01 7.94167444e-02 -1.23928308e-01 3.79141301e-01
4.35879558e-01 8.78431320e-01 -8.40753615e-01 -1.92343861e-01
5.59810400e-01 6.60686314e-01 -5.32031596e-01 6.07762396e-01
2.31166124e-01 5.50907195e-01 -3.01916152e-01 3.74629736e-01
3.22383642e-01 -5.85325181e-01 5.86768948e-02 -3.95392209e-01
-1.95877209e-01 5.60877323e-01 -5.92695415e-01 1.01120269e+00
-3.14662308e-01 8.75658453e-01 -3.73504430e-01 -8.47571313e-01
6.27328217e-01 6.46624088e-01 1.05491138e+00 -3.30350488e-01
3.43812734e-01 1.97691277e-01 6.56801462e-01 -9.60971415e-01
-3.89057249e-01 -8.57547700e-01 5.69185019e-01 5.87899387e-01
-1.48655996e-01 -1.37970909e-01 -3.88138980e-01 -1.41754925e-01
1.49781919e+00 -4.65933621e-01 3.26723099e-01 -4.83337522e-01
-2.37601884e-02 -1.41869159e-02 5.36990285e-01 9.77922857e-01
-4.56050038e-01 1.00512660e+00 4.46382314e-01 -4.45050597e-01
-7.57678211e-01 -1.15837944e+00 -7.76110590e-01 4.50781763e-01
-3.25647533e-01 -5.72372265e-02 -3.03662807e-01 -3.65918219e-01
5.25682330e-01 8.91660392e-01 -4.07360256e-01 -6.22190893e-01
-5.09907663e-01 -7.48283327e-01 1.16431236e+00 1.04895222e+00
3.61725665e-03 -5.97706974e-01 -1.14686632e+00 6.61526859e-01
-1.02895282e-01 -7.95698702e-01 -2.70037830e-01 5.60009956e-01
-1.00454414e+00 -1.22566903e+00 -6.93501294e-01 -3.37015659e-01
2.57016629e-01 2.55655169e-01 1.08366430e+00 2.28274256e-01
-6.34058058e-01 5.11650145e-01 -8.08890909e-02 -8.87099326e-01
-4.07484621e-01 -7.42621943e-02 6.04519963e-01 -5.46062469e-01
5.61511755e-01 -6.55147910e-01 -1.17664874e+00 1.67373344e-01
-4.89503294e-01 -4.26088810e-01 5.53944647e-01 8.58442903e-01
3.77561510e-01 -3.24997216e-01 1.09707236e+00 -5.50369203e-01
8.48937154e-01 -7.98416793e-01 -2.27083769e-02 -2.26233900e-01
-1.28275871e+00 -1.32676780e-01 8.66830587e-01 -6.69880927e-01
-4.25048798e-01 -4.46911514e-01 -1.67549457e-02 -9.91740167e-01
-1.53511554e-01 3.56094778e-01 6.48193479e-01 1.31186873e-01
1.18208277e+00 -1.17121555e-01 2.79524356e-01 -3.77333283e-01
-3.26293707e-01 5.98249137e-01 4.35075939e-01 -3.78883541e-01
6.29092574e-01 4.78611328e-02 2.82297730e-01 -6.89779639e-01
-5.98119378e-01 -4.77478087e-01 -1.66773438e-01 -3.41756232e-02
9.28246319e-01 -9.69850659e-01 -7.05982566e-01 -4.76021580e-02
-6.56328619e-01 -5.62814355e-01 -4.57003146e-01 1.07906890e+00
-3.97304446e-01 1.09035157e-01 -6.80830419e-01 -6.12233698e-01
-6.47793710e-01 -7.41431713e-01 4.23885524e-01 2.89282739e-01
-9.42465842e-01 -1.07593846e+00 1.87847212e-01 1.53196275e-01
9.82459784e-01 4.70971972e-01 1.25263834e+00 -9.05547023e-01
-1.45490050e-01 -6.47715107e-02 -2.22374022e-01 6.22336745e-01
5.47456741e-01 1.54423058e-01 -9.71628666e-01 -1.88311890e-01
2.32477754e-01 -2.63404876e-01 5.64970732e-01 7.77494907e-01
1.17205989e+00 -2.28797719e-01 -2.82049954e-01 6.92947686e-01
1.31711054e+00 6.29301548e-01 1.88069791e-01 -1.33670673e-01
7.60374308e-01 1.77301213e-01 1.26470953e-01 6.38702154e-01
1.43784180e-01 1.72843095e-02 5.23279458e-02 -1.05227128e-01
-7.88124353e-02 -2.99402513e-02 3.50674987e-02 6.18423581e-01
1.63132042e-01 -3.68680686e-01 -1.37039196e+00 4.54112649e-01
-1.31141484e+00 -7.35416114e-01 -4.65558887e-01 2.18939090e+00
8.63065064e-01 9.04937312e-02 8.50839317e-02 -5.18979542e-02
4.43752289e-01 -2.11500391e-01 -8.60864937e-01 -5.34153521e-01
1.64591283e-01 7.44401097e-01 2.66907007e-01 2.20311388e-01
-6.41843557e-01 8.60578492e-02 6.18796349e+00 -2.13530034e-01
-1.47049916e+00 -1.99305683e-01 4.57670480e-01 -3.30087394e-01
-4.79538133e-03 -5.70340812e-01 -3.25630516e-01 4.66359824e-01
1.43068075e+00 -4.43488419e-01 2.46834442e-01 5.68564415e-01
6.44749522e-01 8.26649815e-02 -1.39970255e+00 9.42453265e-01
-3.60513180e-02 -1.35943174e+00 -2.81775564e-01 -1.29936293e-01
4.00359839e-01 4.02671695e-01 1.10620402e-01 1.61628664e-01
6.04168810e-02 -1.44228697e+00 1.01318829e-01 4.40751404e-01
1.46803594e+00 -4.81133372e-01 9.36666608e-01 3.38956378e-02
-1.03997266e+00 -9.70569700e-02 -2.08152071e-01 -6.41791075e-02
-8.58566985e-02 7.29000688e-01 -1.49581075e+00 6.37651160e-02
8.66595805e-01 3.40071082e-01 1.27726689e-01 1.22884059e+00
7.54735321e-02 1.10648668e+00 -6.10539675e-01 -1.69199072e-02
-2.23672569e-01 4.55994934e-01 5.66661477e-01 1.22418618e+00
4.82609838e-01 5.99872589e-01 -1.63996026e-01 9.07899678e-01
2.08939444e-02 -4.55154665e-02 -7.59042680e-01 -1.92509100e-01
5.47622800e-01 8.29167426e-01 -2.18221858e-01 -2.57047951e-01
-4.63837057e-01 2.23089233e-01 -1.91165149e-01 5.52628160e-01
-6.36379540e-01 -6.69019043e-01 8.30231369e-01 5.22050738e-01
4.85247113e-02 9.59097128e-03 -8.04943442e-01 -9.34466600e-01
-3.53308648e-01 -6.97431862e-01 4.63062853e-01 -8.37854505e-01
-1.22902060e+00 4.05624270e-01 -2.43809357e-01 -1.44131398e+00
-3.87551576e-01 -4.61060643e-01 -7.43805766e-01 1.31734610e+00
-1.33633995e+00 -2.05357783e-02 -6.82637870e-01 1.44602761e-01
4.11198795e-01 4.92971502e-02 1.19886005e+00 1.66911379e-01
-4.74186033e-01 5.16675532e-01 -1.18593842e-01 9.14434567e-02
9.78898883e-01 -1.18827832e+00 -4.08519916e-02 4.41428274e-01
-6.72156990e-01 9.05735672e-01 5.87812543e-01 -6.55073047e-01
-1.17753088e+00 -1.36686587e+00 6.23535275e-01 -8.39212358e-01
4.59907264e-01 4.47057039e-01 -1.29742110e+00 4.08687860e-01
-1.07932374e-01 2.69003734e-02 1.15986276e+00 -2.23698437e-01
-1.24850176e-01 -2.50995725e-01 -1.32040572e+00 3.40269119e-01
8.42367411e-01 -3.54390383e-01 -6.76873624e-01 2.87741888e-03
6.40437484e-01 -3.22593808e-01 -1.30258667e+00 1.02423453e+00
9.18968976e-01 -7.60348141e-01 8.06717157e-01 -8.32135379e-01
3.17053974e-01 -1.41488537e-01 -6.04914781e-03 -1.22426856e+00
-5.15468478e-01 -4.21785831e-01 -3.84604394e-01 5.01562595e-01
3.18109959e-01 -1.01988828e+00 6.48133934e-01 6.78147435e-01
-3.49703521e-01 -1.07606721e+00 -5.94174325e-01 -5.26981950e-01
2.80106932e-01 -4.13448036e-01 3.41210097e-01 1.08410168e+00
-1.98882744e-02 2.44529188e-01 1.47378981e-01 2.73584068e-01
3.21276575e-01 -1.72298953e-01 1.56055450e-01 -1.38319147e+00
-1.73108354e-01 -5.07302582e-01 -5.70535511e-02 -5.57060063e-01
-4.24227417e-01 -7.06450999e-01 -1.41921967e-01 -1.76996744e+00
-1.25896126e-01 -7.49422431e-01 -1.01995909e+00 4.68308300e-01
-3.29648495e-01 -1.26652047e-01 -3.56342137e-01 1.35402784e-01
2.16801703e-01 2.68909335e-01 8.98785591e-01 2.18797699e-01
-4.72103477e-01 1.95309538e-02 -9.77742255e-01 4.39538747e-01
1.41344893e+00 -8.12626243e-01 -6.86397851e-01 -2.08761126e-01
-2.09741592e-01 3.65805298e-01 2.82800555e-01 -1.04041159e+00
-8.55074003e-02 -3.40926111e-01 9.16908026e-01 -2.70639330e-01
1.77311618e-02 -6.46857738e-01 2.01482281e-01 1.01958609e+00
-2.06774414e-01 3.94982278e-01 5.00484109e-01 3.80173057e-01
1.47469223e-01 -1.45714846e-03 9.60014582e-01 -3.07314638e-02
1.77966833e-01 3.38730395e-01 -8.23310494e-01 5.94517231e-01
5.79151750e-01 -1.41518280e-01 -4.60451901e-01 -4.78194445e-01
-5.15379250e-01 6.14404194e-02 2.89880931e-01 1.73439845e-01
6.32518232e-01 -9.72675681e-01 -6.34972394e-01 3.33398402e-01
-4.16867808e-02 5.65136187e-02 2.35392183e-01 1.11809242e+00
-1.06417787e+00 3.90831411e-01 6.10173233e-02 -7.98359275e-01
-8.32681060e-01 1.70533523e-01 8.64279985e-01 2.55659878e-01
-7.64279068e-01 8.58546972e-01 2.62900412e-01 1.80034190e-01
2.98014075e-01 -6.22845232e-01 -7.67157646e-03 -2.95864582e-01
2.81272173e-01 6.70610666e-01 3.49067152e-03 1.59776241e-01
-6.93729341e-01 4.62394983e-01 8.82784203e-02 6.66036606e-02
1.08866024e+00 3.10127050e-01 1.72318816e-01 6.79048657e-01
1.14300978e+00 -9.54259634e-02 -9.62316871e-01 2.35665545e-01
-2.90855020e-01 -1.25247777e-01 -7.52011538e-02 -1.12328982e+00
-7.97143221e-01 1.24608624e+00 1.03437233e+00 1.52632415e-01
1.16633248e+00 -1.53246790e-01 8.98304224e-01 5.27424216e-01
-5.78515530e-02 -7.19405055e-01 1.21411838e-01 5.47842495e-02
6.92208230e-01 -8.94665658e-01 -1.25248358e-01 5.50132953e-02
-4.68401074e-01 1.43926787e+00 6.86340332e-01 9.99266654e-02
8.31022143e-01 3.78340602e-01 4.74073559e-01 -9.41868797e-02
-1.07588136e+00 2.31637120e-01 8.16560239e-02 5.45273066e-01
5.74643195e-01 1.11807793e-01 -2.51805633e-01 7.86022484e-01
-2.18048990e-01 1.40668958e-01 5.28997660e-01 8.78395498e-01
-6.94687784e-01 -7.08492458e-01 -1.54298097e-01 1.36627245e+00
-7.66543627e-01 -2.08856866e-01 1.49547774e-02 6.27462626e-01
2.51584854e-02 1.11535287e+00 2.02280223e-01 -5.38292825e-01
4.86537218e-01 6.93279803e-01 2.38256767e-01 -8.07397187e-01
-5.20278037e-01 1.33976281e-01 1.97732866e-01 -3.80946517e-01
6.09274656e-02 -5.50927877e-01 -1.79016006e+00 -6.08012192e-02
-1.04332104e-01 1.07754037e-01 3.24598610e-01 3.93893391e-01
6.87764168e-01 1.00275719e+00 5.80133498e-01 -1.21691726e-01
-6.52673900e-01 -8.88486505e-01 -4.08624947e-01 2.21547447e-02
9.74076092e-01 -6.75381660e-01 -9.41237032e-01 -2.12369323e-01] | [8.049768447875977, 6.137688636779785] |
5eaf0873-82b4-4d90-ac4c-1559a1736b50 | a-community-based-algorithm-for-large-scale | 1305.0187 | null | http://arxiv.org/abs/1305.0187v1 | http://arxiv.org/pdf/1305.0187v1.pdf | A Community Based Algorithm for Large Scale Web Service Composition | Web service composition is the process of synthesizing a new composite
service using a set of available Web services in order to satisfy a client
request that cannot be treated by any available Web services. The Web services
space is a dynamic environment characterized by a huge number of elements.
Furthermore, many Web services are offering similar functionalities. In this
paper we propose a model for Web service composition designed to address the
scale effect and the redundancy issue. The Web services space is represented by
a two-layered network architecture. A concrete similarity network layer
organizes the Web services operations into communities of functionally similar
operations. An abstract interaction network layer represents the composition
relationships between the sets of communities. Composition synthesis is
performed by a two-phased graph search algorithm. First, the interaction
network is mined in order to discover abstract solutions to the request goal.
Then, the abstract compositions are instantiated with concrete operations
selected from the similarity network. This strategy allows an efficient
exploration of the Web services space. Furthermore, operations grouped in a
community can be easily substituted if necessary during the composition's
synthesis's process. | ['Jean-Francois Santucci', 'Yvan Rivierre', 'Chantal Cherifi'] | 2013-05-01 | null | null | null | null | ['service-composition'] | ['miscellaneous'] | [ 3.10703367e-01 2.57549584e-01 7.73280039e-02 -2.70485610e-01
4.62832861e-02 -7.56210983e-01 6.77447736e-01 -5.93348481e-02
2.78151453e-01 2.92449534e-01 3.96282285e-01 -3.06881875e-01
-4.25050557e-01 -1.12128365e+00 -4.33103135e-03 -8.78875077e-01
-3.60186636e-01 8.33053529e-01 8.77482235e-01 -3.88915241e-01
8.09111595e-02 8.88587832e-01 -2.13847113e+00 5.48594177e-01
8.34011972e-01 8.79397810e-01 3.01842570e-01 8.64742845e-02
-9.97328639e-01 6.08407371e-02 -3.39724779e-01 -2.24943042e-01
4.60346669e-01 -7.76547790e-01 -7.25745082e-01 7.40048110e-01
-5.32697797e-01 1.93100676e-01 2.06982628e-01 1.27374434e+00
3.53852473e-03 1.50309041e-01 1.19504966e-01 -2.04280758e+00
-2.03005206e-02 1.04319501e+00 -2.43905629e-03 -1.72271445e-01
8.13425839e-01 -5.62812872e-02 1.36830151e+00 -7.57875025e-01
9.96266305e-01 1.25120986e+00 -1.24962740e-01 7.06570685e-01
-1.24213088e+00 -4.22913581e-01 6.07594669e-01 2.69884318e-01
-1.36619496e+00 -3.04512262e-01 6.47431135e-01 -2.11463105e-02
9.48634863e-01 8.39024365e-01 1.08602011e+00 4.45001751e-01
-1.46545902e-01 2.13315472e-01 6.53308213e-01 -2.60063887e-01
7.53726602e-01 3.06593012e-02 4.42348905e-02 3.24629992e-01
5.96852422e-01 -4.09652293e-01 -3.69216800e-01 -6.78646982e-01
2.48546898e-01 4.30243075e-01 -3.24971169e-01 -1.01932526e+00
-8.25056076e-01 5.60536981e-01 4.51818526e-01 7.14043617e-01
-5.14204144e-01 -7.49407634e-02 1.90021291e-01 5.22877812e-01
-7.56778270e-02 4.77289706e-01 -5.37747502e-01 2.02084199e-01
-2.84189373e-01 1.20981216e-01 1.57425725e+00 1.06958222e+00
8.59410524e-01 -8.79255459e-02 7.24891782e-01 5.67367673e-01
6.70844078e-01 -2.87588686e-02 2.28863269e-01 -8.84592474e-01
-1.32314384e-01 1.40069747e+00 -2.28538383e-02 -7.23842800e-01
-7.18100145e-02 -1.04262523e-01 -3.03281277e-01 3.42169434e-01
3.70697528e-02 3.88499439e-01 -4.41095173e-01 1.29222906e+00
6.59277141e-01 -9.73338112e-02 1.80567190e-01 7.76279747e-01
5.12694597e-01 4.49795097e-01 -1.14961505e-01 -5.36198676e-01
1.32122242e+00 -9.80689049e-01 -6.73149228e-01 1.77545309e-01
2.48886630e-01 -3.49244744e-01 8.48452508e-01 9.91054550e-02
-1.23470354e+00 3.07008564e-01 -9.92508292e-01 6.61756814e-01
-6.01059794e-01 -1.05180264e+00 7.15748489e-01 4.24873918e-01
-1.36526227e+00 5.17811954e-01 -4.65083718e-01 -7.21652448e-01
-1.32772774e-01 4.03402746e-01 -4.00534421e-01 -2.49070287e-01
-9.52727973e-01 4.62827712e-01 4.35094535e-01 -2.17638969e-01
-9.97446537e-01 5.55664636e-02 -7.15426087e-01 6.72433257e-01
9.96956050e-01 -6.44385636e-01 9.99836445e-01 -1.31033146e+00
-1.22399163e+00 4.04880881e-01 -1.03059672e-01 -9.04035866e-02
1.69237182e-01 1.02097237e+00 -1.01000679e+00 1.45928264e-01
4.83759530e-02 -1.34895205e-01 4.76868033e-01 -1.66424799e+00
-1.38519156e+00 -3.34486216e-01 3.11496973e-01 2.94649124e-01
-2.51715422e-01 4.38068002e-01 -6.29481852e-01 1.18327484e-01
1.04658329e+00 -9.07786191e-01 -5.30229747e-01 -1.44889966e-01
-8.96108337e-03 -2.35310033e-01 5.33000171e-01 -1.79968983e-01
1.33747828e+00 -1.94181979e+00 4.42735910e-01 1.06783700e+00
4.13817883e-01 -4.96332288e-01 -1.46789417e-01 1.16999578e+00
-4.08095084e-02 3.88391376e-01 -2.54906148e-01 3.28111351e-01
2.43939608e-01 5.78283727e-01 3.35373193e-01 4.53642011e-02
-4.52553630e-01 2.70803243e-01 -9.79411364e-01 -2.20646814e-01
-4.67087984e-01 5.50412834e-02 -6.44001305e-01 1.58522263e-01
-6.53583586e-01 2.15622172e-01 -6.23288631e-01 8.94339085e-01
4.88344491e-01 -5.16222477e-01 1.11552382e+00 1.03239782e-01
-3.45271409e-01 6.00691438e-01 -1.56912291e+00 1.38314903e+00
-1.91375807e-01 -2.67164201e-01 5.73663831e-01 -7.69542396e-01
9.50682759e-01 7.14148283e-01 1.02402031e+00 -2.20842764e-01
-9.91927460e-02 7.34539151e-01 5.89846969e-01 -3.35364223e-01
-1.03471249e-01 9.07296464e-02 8.30034092e-02 9.55760479e-01
-2.53613621e-01 1.70305334e-02 6.92352593e-01 1.88114569e-01
1.36303365e+00 -5.97273819e-02 3.17193717e-01 -3.91910791e-01
8.92066300e-01 -8.64462741e-03 7.95867503e-01 3.63031685e-01
1.15303084e-01 -1.28583014e-01 7.18975723e-01 -5.68117380e-01
-9.53418195e-01 -1.24943674e+00 4.55214769e-01 9.80308831e-01
3.23530853e-01 -8.47685575e-01 -4.32339311e-01 -6.23668909e-01
7.05192750e-03 4.51094925e-01 -1.02897160e-01 -1.46754794e-02
-4.33954954e-01 -2.27155179e-01 -2.96089709e-01 -8.83296952e-02
2.18030974e-01 -1.26740336e+00 -4.36046094e-01 3.97425205e-01
-1.54097810e-01 -8.30239534e-01 -3.20459366e-01 -8.37686732e-02
-7.49204755e-01 -1.15778935e+00 3.09883505e-02 -8.72617722e-01
1.11740923e+00 6.28714383e-01 9.51830566e-01 6.84230030e-01
4.03755205e-03 2.66955137e-01 -7.28745162e-01 1.53538510e-01
-6.28499329e-01 -2.28213817e-01 1.24304071e-01 5.05399942e-01
3.01557422e-01 -1.28670716e+00 -5.05858898e-01 6.59844220e-01
-1.22256780e+00 2.60071754e-01 1.68540314e-01 3.08247268e-01
5.08942127e-01 6.15856826e-01 4.16100234e-01 -5.07605016e-01
6.86187983e-01 -7.32036710e-01 -7.56912470e-01 2.59476274e-01
-9.00960684e-01 5.57327047e-02 4.74826694e-01 -3.96615535e-01
-8.75476599e-01 -2.54682843e-02 5.19318730e-02 1.90709099e-01
-2.36631674e-03 8.47920716e-01 -9.91110861e-01 8.84527415e-02
2.14885086e-01 1.52247146e-01 9.87758264e-02 -8.39132071e-01
2.47401133e-01 5.85417628e-01 -1.99373677e-01 -5.05025506e-01
8.95343244e-01 5.74699461e-01 7.07769841e-02 -2.25702569e-01
3.18529665e-01 -4.40108150e-01 -3.78804803e-01 -4.81893331e-01
4.74257320e-01 -1.32127360e-01 -9.02921140e-01 -1.71662927e-01
-7.01992810e-01 1.94265291e-01 -3.41326654e-01 5.52257225e-02
-2.67238259e-01 2.07033888e-01 -3.52015734e-01 -7.96749353e-01
-3.72642800e-02 -1.41453373e+00 2.78766870e-01 1.60625741e-01
-2.44907722e-01 -4.43754315e-01 -2.48830244e-01 2.22994089e-02
4.49944735e-01 1.58333048e-01 1.09670997e+00 -1.42394972e+00
-1.00939059e+00 -2.79125571e-01 4.42607552e-02 -1.93870097e-01
5.35640717e-01 3.97452004e-02 -5.63806407e-02 -3.22083354e-01
-1.84323281e-01 7.85202861e-01 -1.43718153e-01 -3.19982946e-01
7.33263075e-01 -5.54652572e-01 -5.62683463e-01 3.35317403e-01
1.56694007e+00 9.29435968e-01 4.60799187e-01 5.41189730e-01
3.25944006e-01 9.94140923e-01 1.51049256e-01 4.52010423e-01
1.42987758e-01 5.84043503e-01 7.62822688e-01 3.27075690e-01
6.99740574e-02 -1.86087772e-01 1.33517176e-01 9.84368980e-01
-7.35586584e-02 -1.77693367e-01 -1.19237292e+00 3.19963753e-01
-2.14440775e+00 -8.76486838e-01 -1.86450973e-01 2.12514234e+00
5.11324048e-01 1.83819145e-01 1.05842620e-01 4.24278587e-01
7.82815099e-01 -2.63019860e-01 -3.76094460e-01 -5.48912585e-01
1.14683472e-02 -2.30833784e-01 1.21694326e-01 3.96964133e-01
2.83663906e-02 7.86445081e-01 5.48984528e+00 4.48747836e-02
-5.44889927e-01 7.82970116e-02 -2.51745194e-01 1.76855121e-02
-1.10864353e+00 7.41771519e-01 -4.67828691e-01 3.79343659e-01
8.14009130e-01 -7.93136716e-01 1.05457258e+00 8.16236675e-01
2.14918166e-01 2.33757738e-02 -1.00637794e+00 4.50588614e-01
-1.55639440e-01 -1.52108335e+00 2.84711510e-01 3.08980554e-01
6.47170722e-01 -1.03676222e-01 -6.13375008e-01 -3.17458570e-01
6.60375476e-01 -2.90382117e-01 5.81614614e-01 2.71157920e-01
2.41286382e-01 -6.92555070e-01 1.97419047e-01 3.92269284e-01
-1.52434886e+00 -5.91770828e-01 1.52939782e-01 1.42377257e-01
4.61796671e-01 2.48922586e-01 -5.88656604e-01 7.62607396e-01
1.07038116e+00 4.89344485e-02 -8.33691657e-02 1.29540181e+00
-1.76863402e-01 -4.23781760e-02 -1.74053416e-01 -3.17712516e-01
1.30799860e-01 -1.01611710e+00 8.34938347e-01 7.44534850e-01
3.35529149e-01 1.65312052e-01 4.45512503e-01 6.40130699e-01
3.34996395e-02 4.40402806e-01 -5.61997831e-01 -3.21029872e-01
5.92812896e-01 1.17046893e+00 -1.19681859e+00 -5.69100082e-01
-6.49321556e-01 8.88027668e-01 -1.81081861e-01 8.11439693e-01
-3.41011077e-01 -3.41276050e-01 1.25457656e+00 4.65271682e-01
1.52727425e-01 -2.20643744e-01 8.08215365e-02 -1.18118072e+00
-9.45301130e-02 -1.36366069e+00 4.76851851e-01 -7.84596860e-01
-1.02592361e+00 9.32665408e-01 -6.40103500e-03 -1.12609506e+00
-2.35846713e-01 4.64738421e-02 -4.14743572e-01 9.39999461e-01
-1.00193989e+00 -7.46358633e-01 -2.80038953e-01 8.16986978e-01
4.04830933e-01 -3.06805819e-01 1.06418073e+00 1.90182686e-01
-1.33684590e-01 -3.00543487e-01 1.54405693e-03 -4.22594249e-01
-1.00841068e-01 -8.02924752e-01 4.29836869e-01 1.00639486e+00
-8.87382179e-02 6.07226908e-01 8.81485224e-01 -9.61063743e-01
-1.57167649e+00 -3.91641527e-01 1.42599380e+00 3.36761743e-01
1.06146348e+00 -5.86737812e-01 -9.34594154e-01 6.07475638e-01
-1.75344124e-02 -2.50846803e-01 8.46017480e-01 -2.02403575e-01
-5.15151560e-01 -4.36587393e-01 -1.43331182e+00 1.35383725e+00
1.58874869e+00 -2.24788144e-01 -2.47138053e-01 4.49425489e-01
9.94232714e-01 4.60799485e-01 -7.17675447e-01 2.25795448e-01
4.55326021e-01 -7.77261972e-01 6.30899906e-01 -8.52748036e-01
-7.88454041e-02 -7.20701933e-01 -3.69699001e-01 -1.16657054e+00
-4.86290097e-01 -8.48140240e-01 -4.78378609e-02 1.04199946e+00
5.53041518e-01 -1.06507039e+00 7.33265996e-01 8.41804266e-01
-9.51637179e-02 -4.36219066e-01 -8.61286879e-01 -8.36076379e-01
-9.26462054e-01 -1.03175743e-02 1.58926320e+00 7.96639085e-01
5.41027248e-01 2.90353205e-02 3.57480496e-01 1.60345688e-01
7.27860749e-01 5.07984996e-01 4.27644819e-01 -1.74548209e+00
-1.98526323e-01 -8.05434823e-01 -2.67320216e-01 -4.01317239e-01
1.20084016e-02 -1.12611377e+00 -7.09776878e-02 -2.14464951e+00
-1.80469751e-01 -5.20949125e-01 -9.14153382e-02 3.18116248e-01
7.11432874e-01 -4.48594332e-01 2.89391130e-01 5.96175790e-01
-2.95244962e-01 -4.53422703e-02 1.16965282e+00 1.66875497e-01
-3.98550749e-01 3.58435780e-01 -6.92334890e-01 6.24293208e-01
8.72841597e-01 -5.32670736e-01 -6.70856357e-01 1.52718350e-01
5.09873986e-01 3.24877679e-01 -4.79408115e-01 -4.98622537e-01
3.53495419e-01 -8.65271151e-01 -5.16142488e-01 -1.22891806e-01
1.08571663e-01 -1.66333735e+00 1.17556703e+00 7.86379039e-01
-2.62657523e-01 2.78769583e-01 -5.37422299e-01 4.39344168e-01
-4.64919321e-02 -3.40061545e-01 4.11738575e-01 -4.90134120e-01
-7.43595898e-01 4.83641058e-01 -8.46450567e-01 -6.79249823e-01
1.13591373e+00 -4.23894733e-01 2.20176622e-01 -3.79324198e-01
-1.18101549e+00 2.10347533e-01 8.42473090e-01 6.65334344e-01
6.46173596e-01 -1.35517776e+00 -3.28588486e-01 2.62765676e-01
4.00129646e-01 -3.80304486e-01 -4.78571355e-01 4.22984004e-01
-5.86667836e-01 -1.44792691e-01 -4.38902795e-01 -2.16076300e-01
-1.41850019e+00 9.82633531e-01 3.25389147e-01 -6.89427480e-02
-7.72569358e-01 3.77480984e-01 -1.37511596e-01 -2.95219839e-01
3.06157351e-01 1.50103420e-02 -4.14162219e-01 -2.77542379e-02
7.09824502e-01 2.68349916e-01 -8.40618908e-02 -6.39017344e-01
-5.32953739e-01 -5.78134209e-02 5.12752295e-01 -6.34138823e-01
1.33023703e+00 -3.77705961e-01 -1.02419376e+00 1.68423384e-01
8.51271689e-01 -6.25529215e-02 -6.84013426e-01 -3.04682165e-01
6.59841597e-01 -7.07033873e-01 -6.66480482e-01 -6.72437131e-01
-1.21220851e+00 -1.06946938e-01 -4.41366695e-02 1.04792464e+00
1.34620690e+00 2.67565578e-01 3.75241935e-01 1.23601288e-01
7.77327418e-01 -9.62628603e-01 -3.54021005e-02 3.74643117e-01
9.12973702e-01 -2.91361570e-01 -4.41027433e-01 -8.53176534e-01
-4.19464111e-01 1.02956557e+00 3.98651898e-01 9.18311477e-02
9.87043977e-01 4.66320843e-01 -2.34350801e-01 -4.56180185e-01
-1.03752732e+00 -4.95226920e-01 -1.49965987e-01 7.02104032e-01
-1.82822704e-01 4.80551310e-02 -1.03356910e+00 5.62148631e-01
1.87586367e-01 -3.96302253e-01 5.21039546e-01 1.12017322e+00
-8.33490074e-01 -1.67919278e+00 -2.29622632e-01 1.53696030e-01
-2.63648033e-02 -6.81818426e-02 -5.41605532e-01 3.18520904e-01
6.77164644e-02 1.22075188e+00 1.81352451e-01 -3.45055163e-01
5.86211503e-01 3.89524013e-01 -1.43673420e-01 -8.02791715e-01
-7.56061733e-01 3.82788181e-01 3.55781734e-01 -8.43633235e-01
-1.88564390e-01 -7.95439124e-01 -1.76156294e+00 -2.99059242e-01
-2.16146082e-01 8.04149151e-01 1.16025555e+00 6.77188277e-01
3.29850197e-01 6.35219932e-01 8.51512253e-01 -3.52083534e-01
-1.29681602e-01 -1.52746394e-01 -9.02283072e-01 5.63990533e-01
-2.65115499e-01 -3.80698085e-01 -6.17195725e-01 1.81375537e-02] | [8.599237442016602, 6.969568729400635] |
a633de0e-50c8-48dc-a0cc-9ed4b8276501 | fusing-event-based-camera-and-radar-for-slam | 2210.04236 | null | https://arxiv.org/abs/2210.04236v1 | https://arxiv.org/pdf/2210.04236v1.pdf | Fusing Event-based Camera and Radar for SLAM Using Spiking Neural Networks with Continual STDP Learning | This work proposes a first-of-its-kind SLAM architecture fusing an event-based camera and a Frequency Modulated Continuous Wave (FMCW) radar for drone navigation. Each sensor is processed by a bio-inspired Spiking Neural Network (SNN) with continual Spike-Timing-Dependent Plasticity (STDP) learning, as observed in the brain. In contrast to most learning-based SLAM systems%, which a) require the acquisition of a representative dataset of the environment in which navigation must be performed and b) require an off-line training phase, our method does not require any offline training phase, but rather the SNN continuously learns features from the input data on the fly via STDP. At the same time, the SNN outputs are used as feature descriptors for loop closure detection and map correction. We conduct numerous experiments to benchmark our system against state-of-the-art RGB methods and we demonstrate the robustness of our DVS-Radar SLAM approach under strong lighting variations. | ['Georges Gielen', 'Francky Catthoor', 'Hichem Sahli', 'André Bourdoux', 'Ilja Ocket', 'Tim Verbelen', 'Ali Safa'] | 2022-10-09 | null | null | null | null | ['drone-navigation', 'loop-closure-detection'] | ['computer-vision', 'computer-vision'] | [ 6.28558874e-01 -3.74237031e-01 5.26976824e-01 -4.75004017e-01
-2.55695164e-01 -5.62547266e-01 6.22178674e-01 2.69659430e-01
-9.93611336e-01 8.70125055e-01 -5.14371216e-01 1.18073858e-01
-1.45768523e-01 -8.63559663e-01 -1.04982173e+00 -6.33652449e-01
-1.25640899e-01 4.81345356e-01 7.06712961e-01 -4.35954452e-01
5.40236712e-01 5.67147791e-01 -1.72391474e+00 -2.16410473e-01
8.16680968e-01 1.35465252e+00 3.84209156e-01 4.64010954e-01
1.09350741e-01 5.68496287e-01 -4.40299720e-01 3.58128488e-01
4.88440424e-01 -5.41047454e-01 5.72054228e-03 -3.27205032e-01
3.95959496e-01 1.02206826e-01 -6.02079153e-01 1.07304561e+00
4.03072834e-01 2.13086873e-01 3.19382817e-01 -1.10800016e+00
-1.36392182e-02 -9.23573002e-02 -2.17557326e-01 3.13363194e-01
2.73629278e-01 4.53339517e-01 3.16335768e-01 -8.00636590e-01
9.05611336e-01 4.88848120e-01 7.17758238e-01 2.59381860e-01
-1.17148733e+00 -8.29975009e-01 -1.19052358e-01 -1.36416882e-01
-1.38111567e+00 -5.20915747e-01 5.60253203e-01 -2.63125807e-01
1.05544531e+00 -3.17157865e-01 1.32175255e+00 8.32572281e-01
9.03579175e-01 1.30383849e-01 1.15102851e+00 -1.25091402e-02
8.68901074e-01 -4.24155027e-01 -1.37256339e-01 7.19797671e-01
6.38437688e-01 3.69015247e-01 -1.21436381e+00 -3.04194242e-02
7.74345040e-01 1.84010699e-01 -4.67639714e-01 -5.41116536e-01
-1.37775373e+00 5.21245658e-01 7.96298444e-01 -4.95203584e-02
-7.19787717e-01 6.98210776e-01 3.59980054e-02 3.59297663e-01
-2.52507031e-01 6.31727338e-01 -1.28711596e-01 -1.91066757e-01
-1.28924155e+00 2.43022159e-01 6.56531930e-01 5.86249769e-01
1.10242999e+00 3.76729935e-01 4.34557647e-01 7.09947869e-02
2.65134007e-01 1.06772220e+00 5.30672073e-01 -8.12214077e-01
5.08170538e-02 5.69994688e-01 2.15341777e-01 -8.12200427e-01
-5.39527416e-01 -4.68863487e-01 -4.28784341e-01 5.59944272e-01
2.20539689e-01 -1.41646579e-01 -1.03026319e+00 1.55146682e+00
-1.13053575e-01 4.82953757e-01 2.41351902e-01 8.51265311e-01
3.66651893e-01 5.27911484e-01 -4.57742125e-01 -1.13780178e-01
7.94424832e-01 -3.98599982e-01 -5.15433609e-01 -7.95560062e-01
2.12561086e-01 -9.42377970e-02 4.13273811e-01 3.63838375e-01
-5.32846928e-01 -1.63389415e-01 -1.54706836e+00 4.79957879e-01
-4.63858306e-01 -4.74102795e-01 7.72794902e-01 2.07081437e-01
-1.10304236e+00 5.55796564e-01 -1.11106098e+00 -7.07063079e-01
3.01586568e-01 3.91146332e-01 -4.72160608e-01 1.07585855e-01
-9.50181961e-01 1.05306756e+00 1.89763725e-01 2.23556340e-01
-1.11886418e+00 -3.66342187e-01 -8.53649974e-01 -2.87757486e-01
-4.05712314e-02 -6.99660540e-01 9.26198423e-01 -6.55192077e-01
-1.48023820e+00 7.46061623e-01 -2.31256455e-01 -1.08138049e+00
1.36661232e-01 -1.63035274e-01 -9.27871242e-02 3.10733467e-01
-1.41586900e-01 7.25638509e-01 7.49347448e-01 -1.10958123e+00
-6.36554599e-01 -4.47199494e-01 -2.78702676e-01 1.57702237e-01
4.45624501e-01 -4.08409923e-01 5.04567027e-02 -2.84483731e-01
8.11431289e-01 -1.03702402e+00 -2.13393345e-01 2.69693911e-01
4.33118492e-01 5.81057072e-01 5.48357010e-01 -8.63296017e-02
5.44706643e-01 -2.09207106e+00 -4.63769138e-02 2.26635695e-01
-1.37032911e-01 -2.01025046e-02 1.07893669e-04 6.26476645e-01
4.38879520e-01 -6.35464787e-01 -6.16479158e-01 -4.55546826e-02
-3.34477812e-01 2.63492376e-01 -6.54498816e-01 8.18397105e-01
1.61236599e-01 9.70253944e-01 -8.31073642e-01 -1.04220234e-01
1.51192769e-01 3.11256200e-01 -3.34492236e-01 -2.02755202e-02
-2.73214817e-01 8.40131283e-01 -1.61449835e-01 6.56792343e-01
4.57596987e-01 1.84581786e-01 -6.09449446e-02 2.48349309e-01
-5.77135563e-01 3.33241671e-01 -1.06828976e+00 2.37364411e+00
-4.18616325e-01 9.63760078e-01 2.19894703e-02 -8.50784004e-01
1.36786246e+00 -1.67076886e-01 1.40898347e-01 -1.07480180e+00
3.35877478e-01 4.76582140e-01 5.69202863e-02 -2.88308635e-02
3.47900182e-01 -1.62081420e-02 -1.97657019e-01 3.29800308e-01
3.69279891e-01 -4.50258374e-01 -3.59220468e-02 1.18062750e-01
1.48976302e+00 4.05600309e-01 8.28836188e-02 -3.40046406e-01
2.22903602e-02 2.43594110e-01 7.48945415e-01 8.88599277e-01
-2.24093646e-01 4.64959532e-01 -1.77191906e-02 -5.97926438e-01
-4.24784601e-01 -1.20822144e+00 -1.37800083e-01 4.29681867e-01
6.21372581e-01 1.39127821e-02 -4.34486449e-01 1.03286907e-01
9.37485993e-02 7.57962584e-01 -6.08131170e-01 -3.12700003e-01
-3.12244922e-01 -7.39122808e-01 6.43374145e-01 2.34225288e-01
6.95574462e-01 -1.09368777e+00 -1.76435721e+00 5.08110523e-01
3.72382969e-01 -1.01473093e+00 3.02293301e-01 9.74200487e-01
-9.44359660e-01 -9.82188523e-01 -6.40014783e-02 -6.16397202e-01
6.66916251e-01 1.17037818e-01 7.06328332e-01 -2.32455194e-01
-3.28949898e-01 3.44248563e-01 -2.21553728e-01 -6.16723657e-01
3.53893012e-01 -3.36866945e-01 3.67865145e-01 -3.32715288e-02
2.75385380e-01 -1.09687996e+00 -4.89028275e-01 -9.54823196e-02
-7.20466077e-01 4.22731079e-02 6.98491454e-01 6.81761682e-01
8.89423490e-01 -3.13927948e-01 3.74764532e-01 -5.10151565e-01
1.98656723e-01 -1.47723481e-01 -1.17342603e+00 -2.19174966e-04
-4.53230917e-01 2.14951724e-01 3.66023421e-01 -1.30741581e-01
-5.14217019e-01 6.26950443e-01 3.35322887e-01 -8.96191075e-02
-1.22855619e-01 6.58614993e-01 2.34297484e-01 -8.04937243e-01
8.33530903e-01 8.87561262e-01 -4.32274826e-02 2.46698365e-01
-1.15437455e-01 3.11158866e-01 9.21466947e-01 -1.61528304e-01
1.03299057e+00 8.60330224e-01 2.26017997e-01 -7.94356704e-01
-3.68899584e-01 -1.93518981e-01 -5.93746245e-01 -4.56073940e-01
5.61332107e-01 -1.02037144e+00 -7.00041115e-01 8.28412235e-01
-1.16462243e+00 -5.13728559e-01 -2.23399505e-01 7.15100944e-01
-7.59931982e-01 -2.11689442e-01 -1.99516892e-01 -9.79413390e-01
-2.44657815e-01 -7.70282209e-01 6.48479819e-01 5.50454676e-01
1.32918194e-01 -4.26430225e-01 4.02000606e-01 -2.98902661e-01
6.70142591e-01 5.76788068e-01 1.53340742e-01 -3.08805794e-01
-9.34911132e-01 -5.20953894e-01 1.00669101e-01 -3.35395664e-01
-3.11780453e-01 -2.07671970e-01 -9.58334923e-01 -2.39523292e-01
1.80681482e-01 -3.27415407e-01 1.29335618e+00 4.19404119e-01
5.40562749e-01 2.58323252e-01 -4.74702954e-01 1.15211678e+00
1.75044882e+00 2.66233563e-01 6.56433225e-01 4.35803026e-01
-1.45102227e-02 2.49774933e-01 7.06607699e-01 3.32267106e-01
3.68025541e-01 1.48551598e-01 1.00731862e+00 2.97733128e-01
2.12082133e-01 -3.16081852e-01 4.13107574e-01 3.73303711e-01
-1.42185641e-02 2.83358637e-02 -9.28365707e-01 4.69886631e-01
-1.89220512e+00 -7.76985705e-01 2.09673211e-01 2.50637341e+00
4.42168146e-01 4.02047873e-01 -4.89112049e-01 -2.25776508e-02
4.57094431e-01 1.14551552e-01 -8.54403734e-01 -1.09780081e-01
-3.80743444e-01 5.35510898e-01 9.34400618e-01 3.68710637e-01
-8.24508369e-01 1.04524887e+00 5.51080418e+00 -1.48909301e-01
-1.62475717e+00 -2.11311445e-01 -4.10539150e-01 -2.23980173e-01
1.77235284e-04 3.42686266e-01 -6.23432219e-01 3.91577423e-01
1.13091171e+00 -4.26688641e-02 7.26518035e-01 5.00621319e-01
-1.44440174e-01 -9.75247443e-01 -7.80174911e-01 1.22197068e+00
5.21506295e-02 -1.42382252e+00 -3.08226496e-01 -4.92975535e-03
4.19229388e-01 7.02987432e-01 -8.37338194e-02 6.85569970e-03
1.57949015e-01 -8.17930639e-01 8.18960667e-01 8.73346269e-01
5.93551874e-01 -6.39012694e-01 6.42005861e-01 4.43657100e-01
-1.18793166e+00 -1.96754131e-02 -4.68119383e-01 -4.46708500e-01
1.79906368e-01 7.26029336e-01 -5.61597824e-01 3.13420624e-01
7.36480713e-01 6.06323957e-01 -5.59793174e-01 1.41715705e+00
-4.38870549e-01 3.49372119e-01 -6.06609762e-01 -2.46091008e-01
1.86843246e-01 -1.54777944e-01 6.11901045e-01 8.11715424e-01
6.17771327e-01 5.93662523e-02 -7.25753009e-02 8.46296608e-01
9.50310752e-02 -4.68467653e-01 -8.98467720e-01 -3.08436044e-02
7.54428148e-01 1.08977437e+00 -9.14726198e-01 2.67213788e-02
-1.10388151e-03 1.09504092e+00 8.74500424e-02 3.59351248e-01
-8.09733093e-01 -7.28927612e-01 5.25288880e-01 5.38316406e-02
4.36292917e-01 -7.81606197e-01 -1.97202772e-01 -1.06721795e+00
-1.34312168e-01 -1.38862863e-01 -1.36534065e-01 -9.05834973e-01
-6.67300463e-01 7.21635640e-01 -6.71959162e-01 -1.38501477e+00
-1.37206525e-01 -4.61720228e-01 -5.10499537e-01 6.82069063e-01
-1.88715160e+00 -8.81790578e-01 -7.45390356e-01 4.89875615e-01
-6.84264377e-02 -1.04881600e-01 1.07282662e+00 -2.23517776e-01
2.93121785e-02 -1.76279366e-01 3.76921147e-02 7.51403496e-02
9.42501724e-01 -8.63267004e-01 5.58245599e-01 1.02395928e+00
2.82840341e-01 5.29450178e-01 8.28262568e-01 -8.61469209e-01
-1.85287154e+00 -1.04554021e+00 7.23484397e-01 -2.29712456e-01
6.93690002e-01 -6.50196314e-01 -6.54852569e-01 6.51638746e-01
-3.74147147e-02 5.49576700e-01 4.05563861e-01 -3.78330976e-01
-2.34688446e-01 -4.70700890e-01 -1.10896277e+00 8.37230310e-02
1.00464189e+00 -5.34145832e-01 -9.37901258e-01 -7.75998086e-02
5.06666183e-01 -5.82821727e-01 -2.03868330e-01 4.55022097e-01
7.00487733e-01 -1.12476087e+00 3.85502189e-01 -6.39484897e-02
-2.11974993e-01 -9.82837856e-01 -3.05188477e-01 -1.37490678e+00
-8.16551149e-02 -4.81847882e-01 9.60936323e-02 5.25624633e-01
3.26458007e-01 -1.01289284e+00 5.25828362e-01 -3.31392512e-02
-2.55675018e-01 -3.87822270e-01 -1.27657771e+00 -8.73027623e-01
-6.14725113e-01 -4.36517000e-01 1.71560377e-01 4.80142087e-01
-1.66492298e-01 6.60908967e-02 8.52599740e-02 5.68789601e-01
8.43745232e-01 3.43314320e-01 6.07345343e-01 -1.59146357e+00
1.06886856e-01 1.59051389e-01 -8.18435013e-01 -5.29096425e-01
-1.33930007e-02 -8.82986426e-01 7.37587631e-01 -1.52035213e+00
-3.15704703e-01 -6.45289496e-02 -4.35333192e-01 4.23801392e-01
4.53314036e-01 3.69438916e-01 5.48511706e-02 2.33663693e-01
-8.08585286e-01 5.64409137e-01 5.12432039e-01 1.39535964e-01
-1.78958133e-01 -2.80121475e-01 -6.72936141e-02 6.46626890e-01
5.80419421e-01 -8.86029541e-01 -2.20713496e-01 -5.24775803e-01
6.38137102e-01 1.30024090e-01 7.30445325e-01 -1.97259438e+00
9.94692385e-01 -8.13497305e-02 5.44537246e-01 -5.53109527e-01
4.72233593e-01 -7.96254992e-01 1.05572522e-01 1.08865273e+00
2.18988344e-01 3.93312797e-02 3.36106151e-01 8.52064013e-01
-2.89444327e-01 -6.10582940e-02 8.38792741e-01 -2.33428240e-01
-1.09530199e+00 2.02875048e-01 -8.01855266e-01 -7.18779489e-02
8.94074500e-01 -4.88167316e-01 -5.48100650e-01 -2.15299830e-01
-2.52841502e-01 1.83555007e-01 9.47317243e-01 -7.45979995e-02
8.78704429e-01 -7.91278124e-01 -2.96963751e-01 6.81803584e-01
2.94883728e-01 1.09102450e-01 -2.92695444e-02 9.21710730e-01
-7.81135142e-01 4.08744454e-01 -7.84125030e-01 -7.14015007e-01
-3.01953554e-01 1.40246198e-01 4.97727036e-01 2.95105636e-01
-3.79356176e-01 1.04380858e+00 -2.65878052e-01 -3.53935659e-01
7.48198479e-02 -2.85358638e-01 2.32955188e-01 8.70629773e-02
3.59913021e-01 -2.04003900e-01 2.52137482e-01 -2.52730757e-01
-9.13660705e-01 5.13605595e-01 6.40264988e-01 -5.73348165e-01
1.74766469e+00 1.57106426e-02 -1.74159005e-01 8.64324093e-01
3.47677022e-01 -1.95801690e-01 -1.51163590e+00 -1.26106903e-01
1.20853513e-01 -1.70921341e-01 2.50524491e-01 -7.71053195e-01
-7.93335021e-01 8.47904682e-01 7.31667161e-01 -4.97851551e-01
1.12873399e+00 -5.28967738e-01 8.29304099e-01 9.26095843e-01
1.12842393e+00 -1.10575283e+00 -8.06408003e-02 8.54440033e-01
7.37669110e-01 -9.16199446e-01 -7.83362314e-02 3.35038483e-01
-4.99422729e-01 1.11341476e+00 3.32089186e-01 -9.02849019e-01
5.40024340e-01 6.27257168e-01 1.74526535e-02 -1.42142519e-01
-8.77068400e-01 -2.74951220e-01 -2.01146066e-01 6.30107760e-01
-1.46666735e-01 -3.46399933e-01 -8.70029479e-02 4.22682017e-01
-1.61063537e-01 2.19725445e-01 6.35785937e-01 1.34789145e+00
-1.09918237e+00 -5.00375092e-01 -2.61190653e-01 2.91348010e-01
3.25522453e-01 -1.43375814e-01 -3.58164400e-01 6.18085504e-01
1.16051145e-01 4.43082184e-01 2.53370553e-01 -5.50475180e-01
2.30069026e-01 2.36070469e-01 8.15120041e-01 -5.44636309e-01
-5.22423685e-01 -3.67801398e-01 -3.65666598e-01 -8.23285460e-01
-3.52288097e-01 -7.32727170e-01 -1.89093554e+00 1.43453360e-01
-1.57179043e-01 -1.17750429e-01 1.19305313e+00 8.18893433e-01
3.93056184e-01 3.48363936e-01 3.00046742e-01 -9.32926536e-01
-1.37491778e-01 -7.35746384e-01 -8.19204092e-01 -1.15201905e-01
5.44728339e-01 -9.19786930e-01 -4.35209394e-01 -1.61933750e-01] | [7.51577615737915, -1.6917216777801514] |
fd0a5583-8b87-4782-a05a-529d18786629 | comprehension-guided-referring-expressions | 1701.03439 | null | http://arxiv.org/abs/1701.03439v1 | http://arxiv.org/pdf/1701.03439v1.pdf | Comprehension-guided referring expressions | We consider generation and comprehension of natural language referring
expression for objects in an image. Unlike generic "image captioning" which
lacks natural standard evaluation criteria, quality of a referring expression
may be measured by the receiver's ability to correctly infer which object is
being described. Following this intuition, we propose two approaches to utilize
models trained for comprehension task to generate better expressions. First, we
use a comprehension module trained on human-generated expressions, as a
"critic" of referring expression generator. The comprehension module serves as
a differentiable proxy of human evaluation, providing training signal to the
generation module. Second, we use the comprehension module in a
generate-and-rerank pipeline, which chooses from candidate expressions
generated by a model according to their performance on the comprehension task.
We show that both approaches lead to improved referring expression generation
on multiple benchmark datasets. | ['Gregory Shakhnarovich', 'Ruotian Luo'] | 2017-01-12 | comprehension-guided-referring-expressions-1 | http://openaccess.thecvf.com/content_cvpr_2017/html/Luo_Comprehension-Guided_Referring_Expressions_CVPR_2017_paper.html | http://openaccess.thecvf.com/content_cvpr_2017/papers/Luo_Comprehension-Guided_Referring_Expressions_CVPR_2017_paper.pdf | cvpr-2017-7 | ['referring-expression-generation'] | ['computer-vision'] | [ 7.02534616e-01 6.87337697e-01 -6.19966388e-02 -7.44856298e-01
-1.28926075e+00 -5.22654772e-01 8.09336722e-01 8.79983157e-02
-5.20699918e-01 7.12736011e-01 6.06469870e-01 -1.35461360e-01
4.84625220e-01 -8.42257857e-01 -9.28652823e-01 -3.44295949e-01
4.58293617e-01 5.31992555e-01 -2.92581558e-01 -2.37292960e-01
2.77778894e-01 1.88848257e-01 -1.24785185e+00 6.41368151e-01
6.85383022e-01 1.08791137e+00 6.52870089e-02 8.09262216e-01
-8.19563419e-02 1.40217793e+00 -1.04514098e+00 -9.08519924e-01
-7.36492276e-02 -9.41314697e-01 -1.18637633e+00 2.39347026e-01
3.24826211e-01 -3.16077262e-01 1.11634441e-01 1.06452215e+00
3.75742584e-01 -6.43302202e-02 6.99675918e-01 -1.13078141e+00
-9.19779122e-01 8.21345091e-01 -3.64278257e-01 -3.65746245e-02
1.05259562e+00 3.90929073e-01 1.29525077e+00 -7.30917573e-01
9.78825092e-01 1.45962572e+00 3.06273922e-02 9.81988490e-01
-1.33624971e+00 -3.47220898e-01 1.44066021e-01 -1.06447607e-01
-9.34109092e-01 -4.93962616e-01 7.31063664e-01 -4.07354921e-01
5.16282797e-01 2.02619135e-01 3.96904737e-01 1.21888828e+00
-1.81024179e-01 1.24114406e+00 1.21801257e+00 -4.34304178e-01
1.66461632e-01 3.32385182e-01 6.89175949e-02 6.63596451e-01
-3.46579760e-01 -2.47917902e-02 -6.39019132e-01 -1.85654774e-01
5.22831321e-01 -7.06769645e-01 -5.29633880e-01 8.73413831e-02
-1.21512592e+00 9.53790307e-01 7.31299162e-01 1.22210169e-02
-8.78095925e-01 4.15059954e-01 3.12707067e-01 4.13631231e-01
6.32989347e-01 1.03800213e+00 -6.30668774e-02 -2.05465436e-01
-7.59846210e-01 4.54588920e-01 7.46224105e-01 8.80143344e-01
7.01887369e-01 -2.69040018e-01 -1.10346794e+00 7.31618881e-01
2.71390826e-01 4.36402291e-01 3.57565552e-01 -1.06512487e+00
4.27009374e-01 6.51346087e-01 2.43970796e-01 -8.00790608e-01
2.50951052e-02 -4.06513959e-01 -4.30266082e-01 1.19927503e-01
3.77178699e-01 -2.29337186e-01 -6.58011198e-01 2.19293451e+00
8.23302791e-02 -1.99693486e-01 1.75152525e-01 1.19353962e+00
7.86550164e-01 7.37169027e-01 5.54547369e-01 3.27603742e-02
1.61718214e+00 -1.01897001e+00 -4.94167924e-01 -1.57321021e-01
8.18219841e-01 -5.00418842e-01 1.38683355e+00 3.43685895e-01
-1.42636335e+00 -4.24820036e-01 -8.30089390e-01 -2.51149327e-01
-1.68030456e-01 2.90621608e-01 4.81952250e-01 3.48323166e-01
-1.25779879e+00 3.02030027e-01 -9.52542722e-02 -2.07874060e-01
5.89258730e-01 7.35645071e-02 -2.07350388e-01 1.67059854e-01
-1.24138796e+00 1.06895936e+00 -3.56446058e-02 -9.27408338e-02
-1.16197836e+00 -4.84660059e-01 -7.68891633e-01 7.00264052e-02
5.01739085e-02 -1.29520440e+00 1.75438523e+00 -1.76601517e+00
-1.50468600e+00 1.43680358e+00 -4.66088772e-01 -6.55466497e-01
8.34548175e-01 -3.31221730e-01 1.36907950e-01 4.34654713e-01
3.70461375e-01 1.30445135e+00 8.67544949e-01 -1.57497942e+00
-5.03959596e-01 -1.19952284e-01 6.83601737e-01 3.24375033e-01
2.35300258e-01 8.85692239e-02 -3.53132546e-01 -5.42288005e-01
-2.18698338e-01 -6.81918561e-01 -3.29615951e-01 1.75219491e-01
-6.96172893e-01 -4.19344991e-01 4.12090898e-01 -4.30006504e-01
9.83446598e-01 -1.88151312e+00 7.81794339e-02 2.82050878e-01
2.22913995e-01 -3.74535583e-02 -3.94196808e-01 1.47065178e-01
-2.54618406e-01 3.46568584e-01 -6.39784858e-02 -3.26985240e-01
1.22681849e-01 4.71410854e-03 -5.28131664e-01 8.06333497e-02
8.85267377e-01 1.19301426e+00 -1.30613804e+00 -4.10097927e-01
-2.25536361e-01 2.58303165e-01 -4.59135711e-01 7.52506018e-01
-6.53083444e-01 5.38521588e-01 -7.24953115e-01 4.19412792e-01
3.41899574e-01 -5.38153529e-01 -3.05584341e-01 -6.14744760e-02
3.41088802e-01 5.18817127e-01 -4.27782029e-01 1.42790163e+00
-6.72400475e-01 6.14478350e-01 -2.44745627e-01 -6.00913346e-01
9.36722398e-01 3.44591558e-01 3.56504805e-02 -8.45636547e-01
2.50858754e-01 -5.51106147e-02 -3.00910830e-01 -6.15107238e-01
5.36419809e-01 -2.79160976e-01 -3.57885927e-01 7.89076447e-01
-3.02786641e-02 -2.75223851e-01 3.64386588e-01 5.43357968e-01
1.32569933e+00 3.56886029e-01 1.84654325e-01 1.51206320e-02
8.09965432e-01 1.58158168e-01 -4.66064811e-02 9.89260495e-01
1.91500541e-02 6.33861899e-01 1.04374433e+00 -2.67278910e-01
-8.23322415e-01 -1.07096672e+00 3.70461792e-01 1.43296468e+00
-8.08677003e-02 -1.63503498e-01 -1.04319286e+00 -8.76062155e-01
-3.52334201e-01 9.68772411e-01 -7.80270696e-01 -3.18801999e-01
-4.76909965e-01 -3.89654964e-01 7.30930865e-01 4.52805668e-01
4.05100673e-01 -1.53893828e+00 -8.40928614e-01 1.06005475e-01
-4.71522182e-01 -1.03195548e+00 -4.27266538e-01 1.06168604e-02
-4.83030707e-01 -8.71838510e-01 -9.38425183e-01 -5.14494181e-01
1.09529853e+00 -6.81720376e-02 1.77425921e+00 3.29263031e-01
2.34540835e-01 5.16818941e-01 -5.95909834e-01 -4.22409147e-01
-8.70614231e-01 1.20505746e-02 -5.81928551e-01 8.44356492e-02
2.45452657e-01 -6.34959936e-02 -6.64785624e-01 3.26394103e-02
-9.47702706e-01 2.73709714e-01 8.09264779e-01 8.76931548e-01
4.60871637e-01 -9.55477178e-01 6.69766724e-01 -1.08296359e+00
1.28796458e+00 -4.07201618e-01 -5.17742872e-01 2.26052359e-01
-3.68544161e-01 4.25093621e-01 5.61605811e-01 -3.70443881e-01
-1.13577950e+00 2.40542948e-01 -3.63919139e-01 -1.98741153e-01
-1.08178549e-01 3.33593041e-01 -1.72300011e-01 2.26124927e-01
8.37448180e-01 2.14780420e-01 -2.18122497e-01 -5.79634449e-03
8.30039859e-01 4.05655891e-01 7.40574241e-01 -9.59907830e-01
5.84711075e-01 2.17777193e-01 -2.01275274e-01 -2.49262765e-01
-1.25137138e+00 -2.43704841e-01 -2.07087755e-01 -4.09874529e-01
1.14277411e+00 -8.95551741e-01 -8.16637158e-01 4.40802090e-02
-1.75357282e+00 -2.45222643e-01 -4.74406540e-01 -2.47429404e-02
-8.78940523e-01 -8.70609507e-02 -4.06476110e-01 -9.38411415e-01
-5.09671688e-01 -1.24751556e+00 1.61372209e+00 2.03823045e-01
-6.17575645e-01 -7.10672915e-01 -1.18599653e-01 3.79786611e-01
4.55246508e-01 3.74494463e-01 9.31284785e-01 -5.26630521e-01
-7.50667393e-01 -2.60561973e-01 -5.18370032e-01 2.92238683e-01
-3.20056051e-01 -1.94628775e-01 -1.31445169e+00 1.98859811e-01
-1.22239277e-01 -6.94856703e-01 7.80155301e-01 1.30450606e-01
1.31343150e+00 -6.73321307e-01 -8.35804492e-02 3.31703752e-01
1.20852709e+00 1.68389250e-02 8.61797154e-01 1.55463412e-01
3.09613258e-01 9.70053136e-01 6.06504798e-01 3.76318336e-01
5.15458643e-01 5.36109447e-01 3.60013157e-01 -3.53801399e-01
-1.73457246e-02 -5.80275118e-01 5.26134253e-01 1.85188577e-01
1.45127013e-01 -6.17197990e-01 -6.72925830e-01 3.37130129e-01
-1.60760391e+00 -9.80385780e-01 -1.41651958e-01 1.90154743e+00
1.00608277e+00 9.40208137e-02 4.80761426e-03 -3.41241270e-01
4.99738991e-01 1.20606080e-01 -3.55601341e-01 -5.31106651e-01
-1.21411920e-01 2.68239588e-01 1.92882597e-01 4.85974848e-01
-7.77017891e-01 1.14262307e+00 6.30753565e+00 5.87983847e-01
-1.04487503e+00 7.99617842e-02 1.26537538e+00 3.22422594e-01
-6.07941031e-01 1.48113653e-01 -3.34863693e-01 2.14653552e-01
7.58614719e-01 -3.27358633e-01 9.82360840e-02 7.91893661e-01
3.94384742e-01 -2.69054681e-01 -1.49940610e+00 7.95840442e-01
2.21096233e-01 -1.08204436e+00 6.52447879e-01 -3.00127178e-01
5.08079112e-01 -3.47427845e-01 7.71785453e-02 2.84520388e-01
4.33428943e-01 -1.12994635e+00 8.74553680e-01 8.30558658e-01
6.25889719e-01 -4.76163536e-01 7.87911594e-01 2.52027899e-01
-5.84477186e-01 3.56230259e-01 -7.12413490e-02 1.63050741e-02
5.32947481e-01 3.22874725e-01 -1.06073546e+00 1.74076304e-01
4.21837159e-02 1.28947839e-01 -7.99273908e-01 6.40571237e-01
-8.42049420e-01 5.25094569e-01 8.98685530e-02 -3.22006196e-01
4.19757247e-01 -3.00491620e-02 4.41910058e-01 1.33901548e+00
3.46758775e-02 1.61451548e-01 -4.69170418e-03 1.37194288e+00
-5.37222445e-01 1.39267266e-01 -6.99220538e-01 -1.38732120e-01
2.66521066e-01 1.28654218e+00 -2.46524572e-01 -7.21400738e-01
-1.54788196e-01 1.28718972e+00 3.24718744e-01 4.44508612e-01
-7.70663679e-01 -1.11778304e-01 8.00446942e-02 1.91133469e-01
-3.33893776e-01 3.96905899e-01 -1.18477099e-01 -1.01316404e+00
1.38446977e-02 -1.21446145e+00 2.85827219e-01 -1.61188889e+00
-1.24244940e+00 9.97083783e-01 -1.60879269e-01 -1.11005282e+00
-7.02450395e-01 -6.70514941e-01 -6.73999548e-01 1.10058236e+00
-1.53418767e+00 -1.15152383e+00 -5.14556468e-01 3.29892337e-01
5.05788684e-01 9.26214755e-02 7.21482575e-01 -1.27708927e-01
-1.24541789e-01 5.74349165e-01 -9.75459397e-01 2.56227791e-01
5.85167289e-01 -1.41808462e+00 2.65556067e-01 5.77243567e-01
2.34938547e-01 6.12626255e-01 9.65923846e-01 -3.13264281e-01
-1.01249492e+00 -1.04373384e+00 1.06473005e+00 -6.02779925e-01
5.59947252e-01 -1.17320545e-01 -8.07127833e-01 6.07833147e-01
4.92153227e-01 -4.04620349e-01 6.29331470e-01 -1.46040201e-01
-4.40004766e-01 2.03116730e-01 -1.13633299e+00 6.97644174e-01
9.18804348e-01 -6.50968075e-01 -5.71523309e-01 3.63419592e-01
7.07199156e-01 -2.87892073e-01 -5.42616487e-01 9.96121541e-02
3.75166506e-01 -8.58647287e-01 6.25911355e-01 -8.57249320e-01
1.02703345e+00 -1.72768369e-01 -4.83281389e-02 -1.28885722e+00
-5.07886522e-03 -7.32490540e-01 3.90976042e-01 1.21601605e+00
7.68320441e-01 -2.47779295e-01 7.05723107e-01 7.06470966e-01
1.38990775e-01 -6.90697908e-01 -6.39144361e-01 -2.66344517e-01
-1.58819810e-01 -3.78374100e-01 6.07389331e-01 5.95113993e-01
-1.52822409e-03 9.61132109e-01 -2.13444486e-01 -2.40152687e-01
2.00355619e-01 2.73237154e-02 1.08351254e+00 -7.77432263e-01
-3.96533728e-01 -6.42560124e-01 -2.74615079e-01 -1.60438156e+00
5.88482141e-01 -1.01831734e+00 4.16566283e-01 -1.34603631e+00
3.02565366e-01 -4.40240884e-03 7.48060271e-02 3.59316051e-01
-5.35461843e-01 -5.49534149e-02 3.71679574e-01 4.17991988e-02
-8.71041477e-01 5.65734148e-01 1.41972756e+00 -2.28599504e-01
1.82560459e-01 -7.00405762e-02 -1.10282779e+00 7.63492942e-01
5.44496775e-01 -3.75044525e-01 -3.45722616e-01 -7.05965519e-01
5.39233983e-01 4.39957947e-01 6.58158958e-01 -6.67016923e-01
-7.71800503e-02 1.00178123e-01 2.87882507e-01 -3.26296568e-01
2.26074487e-01 -4.43909764e-01 -2.79150814e-01 2.00922474e-01
-1.12153780e+00 5.32363296e-01 -2.67384648e-01 2.23558709e-01
-4.38496023e-01 -5.36838353e-01 6.88618422e-01 -3.40342611e-01
-5.28671026e-01 -2.05452424e-02 -3.34933817e-01 1.22611620e-01
6.50220990e-01 1.90918539e-02 -3.49665195e-01 -1.06038451e+00
-6.74362004e-01 4.65097398e-01 3.94220173e-01 3.82134318e-01
7.96458125e-01 -1.33809304e+00 -1.04675615e+00 -2.44105786e-01
4.54264849e-01 -1.46797225e-01 -3.60507518e-01 5.34418523e-01
-3.97732466e-01 3.82582359e-02 1.26646236e-01 -5.10388494e-01
-9.24013913e-01 4.44680154e-01 4.07009065e-01 -5.24010360e-01
6.17472157e-02 1.07634425e+00 3.92470688e-01 -7.59456381e-02
6.26778081e-02 -2.86405206e-01 -2.48597309e-01 2.81106937e-03
5.61882317e-01 -8.39772001e-02 -3.61618459e-01 -8.60933602e-01
-2.14226469e-02 1.99017122e-01 2.35193148e-02 -5.65614045e-01
8.42954874e-01 -1.02626108e-01 -1.47547856e-01 2.31345445e-01
1.17232096e+00 -8.36591199e-02 -1.03585637e+00 -8.28147605e-02
4.23822194e-01 -3.05954963e-01 -3.23996395e-01 -9.17699873e-01
-7.76882231e-01 9.70691323e-01 3.42174143e-01 1.72945738e-01
1.33631527e+00 3.12490940e-01 4.76909429e-01 4.22108889e-01
3.79920751e-01 -7.93294966e-01 5.87959766e-01 3.70027810e-01
1.30792892e+00 -1.29174328e+00 -3.53325993e-01 -3.53563398e-01
-1.11434340e+00 9.92142677e-01 8.52488399e-01 -1.97255969e-01
-1.20008923e-01 6.58803433e-02 5.28060675e-01 -4.33126479e-01
-1.20255911e+00 -4.64236885e-01 2.46269241e-01 5.15550375e-01
9.27381635e-01 -3.79051156e-02 -3.83363158e-01 4.58853692e-01
-4.66697812e-01 1.74948022e-01 4.99396652e-01 5.41278780e-01
-2.89300114e-01 -1.02491057e+00 -2.31177479e-01 3.58352691e-01
-6.26586556e-01 -2.16290191e-01 -7.97083020e-01 3.65032434e-01
-6.11482486e-02 9.79705572e-01 -8.54454115e-02 -1.24345124e-01
5.43550253e-01 -8.65484122e-03 2.70797044e-01 -6.49120867e-01
-8.30952048e-01 -2.14177758e-01 4.55731392e-01 -7.25169480e-01
-5.67753494e-01 -3.11282277e-01 -1.17717505e+00 2.30171576e-01
-1.53737038e-01 3.76520246e-01 3.83246124e-01 9.33724821e-01
2.98737586e-01 4.79766309e-01 6.77156150e-01 -5.33321977e-01
-7.96121180e-01 -9.43888009e-01 1.83625892e-02 1.10855651e+00
2.10167930e-01 -2.57079154e-01 -1.05539389e-01 4.28064525e-01] | [10.840110778808594, 1.4335837364196777] |
3633e1a3-38a5-42d7-adfa-fbea950f6d46 | codebleu-a-method-for-automatic-evaluation-of | 2009.10297 | null | https://arxiv.org/abs/2009.10297v2 | https://arxiv.org/pdf/2009.10297v2.pdf | CodeBLEU: a Method for Automatic Evaluation of Code Synthesis | Evaluation metrics play a vital role in the growth of an area as it defines the standard of distinguishing between good and bad models. In the area of code synthesis, the commonly used evaluation metric is BLEU or perfect accuracy, but they are not suitable enough to evaluate codes, because BLEU is originally designed to evaluate the natural language, neglecting important syntactic and semantic features of codes, and perfect accuracy is too strict thus it underestimates different outputs with the same semantic logic. To remedy this, we introduce a new automatic evaluation metric, dubbed CodeBLEU. It absorbs the strength of BLEU in the n-gram match and further injects code syntax via abstract syntax trees (AST) and code semantics via data-flow. We conduct experiments by evaluating the correlation coefficient between CodeBLEU and quality scores assigned by the programmers on three code synthesis tasks, i.e., text-to-code, code translation, and code refinement. Experimental results show that our proposed CodeBLEU can achieve a better correlation with programmer assigned scores compared with BLEU and accuracy. | ['Ambrosio Blanco', 'Long Zhou', 'Shuai Ma', 'Shuai Lu', 'Ming Zhou', 'Shuo Ren', 'Shujie Liu', 'Duyu Tang', 'Daya Guo', 'Neel Sundaresan'] | 2020-09-22 | null | null | null | null | ['code-translation'] | ['computer-code'] | [-1.35197386e-01 -1.61487490e-01 -2.61733919e-01 -4.77549374e-01
-7.79483378e-01 -6.91336334e-01 3.95464778e-01 5.03034890e-01
-1.49052635e-01 2.98467219e-01 4.13402230e-01 -6.27295613e-01
1.96244210e-01 -7.94355750e-01 -6.99430466e-01 1.25988185e-01
2.97191203e-01 4.23490480e-02 3.17318350e-01 -2.87432313e-01
7.31324613e-01 -3.11722070e-01 -1.42380416e+00 3.87735754e-01
1.42581952e+00 6.90667450e-01 1.56289279e-01 5.34790456e-01
-7.73470044e-01 1.18260241e+00 -6.24352217e-01 -8.40702772e-01
-5.75456619e-02 -7.62005985e-01 -9.35844362e-01 -5.06588340e-01
5.65522583e-03 9.86381918e-02 1.50219664e-01 1.53467262e+00
1.09792963e-01 -5.86537778e-01 5.59770286e-01 -1.14396906e+00
-7.30576634e-01 1.17355585e+00 -4.38414305e-01 -6.91613648e-03
6.42081022e-01 4.75583263e-02 1.26634836e+00 -8.52122128e-01
4.42258120e-01 1.25088656e+00 6.62902772e-01 5.15242755e-01
-1.14742029e+00 -5.79684138e-01 -3.55468303e-01 -3.04569909e-03
-1.22290635e+00 -2.11880714e-01 4.72132295e-01 -9.55944061e-01
8.53069186e-01 3.76442373e-01 1.98601648e-01 7.00534582e-01
3.75144541e-01 6.80089951e-01 9.63624656e-01 -5.78174293e-01
2.67140210e-01 2.51032948e-01 4.07228261e-01 8.08093965e-01
3.22842747e-01 -6.00330830e-02 -2.24293739e-01 -1.61832154e-01
2.09743917e-01 -1.40893593e-01 -2.23448575e-01 -3.19525123e-01
-1.23836923e+00 8.03551793e-01 3.02898973e-01 4.71544623e-01
2.74114937e-01 2.46091023e-01 6.59463644e-01 5.49328387e-01
-3.59456837e-02 8.00435781e-01 -4.12253767e-01 -7.32559144e-01
-8.76712561e-01 1.58005849e-01 8.78518522e-01 1.06770074e+00
6.25015140e-01 -7.55635649e-02 -3.58417362e-01 7.79544592e-01
4.36634302e-01 4.27986503e-01 7.33584225e-01 -7.91660190e-01
5.29286623e-01 1.20839238e+00 9.49763954e-02 -9.95608985e-01
-1.41869202e-01 -5.16839087e-01 -5.18721700e-01 3.19702297e-01
3.07072043e-01 3.22078168e-02 -5.10716200e-01 1.62433147e+00
-4.27185118e-01 -4.80378985e-01 -7.76910484e-02 6.99143171e-01
7.74553418e-01 4.94856507e-01 -2.07405016e-01 -9.17985365e-02
1.15050864e+00 -1.20765662e+00 -5.77677608e-01 -3.38628411e-01
9.17428732e-01 -1.06833982e+00 1.52151418e+00 2.93695718e-01
-1.08903801e+00 -5.93610466e-01 -1.19029593e+00 3.21712680e-02
-1.98452681e-01 1.33549720e-01 4.23610747e-01 9.28613424e-01
-8.77530277e-01 5.20181000e-01 -5.71248531e-01 -1.22116372e-01
-2.76951101e-02 -6.83626682e-02 -5.04960753e-02 1.20085850e-01
-9.23815131e-01 9.14392710e-01 2.81951517e-01 -5.66321790e-01
-8.48438144e-01 -5.65561235e-01 -6.58261597e-01 3.35707545e-01
1.92183003e-01 -3.14270169e-01 1.48868895e+00 -1.07119727e+00
-1.20568562e+00 7.76544333e-01 -2.10652258e-02 -2.50368655e-01
3.88447165e-01 -8.22208226e-02 -4.03252602e-01 -7.87800252e-01
4.65438515e-02 1.42278835e-01 4.00755107e-01 -1.14301920e+00
-6.28622293e-01 7.05109909e-02 2.98877209e-01 -2.76326537e-01
-3.97846639e-01 2.34645337e-01 -4.24066305e-01 -7.35813856e-01
6.96215825e-03 -7.73480475e-01 6.84264228e-02 -2.44022742e-01
-1.61534622e-02 -2.16550082e-01 2.64972627e-01 -7.70266473e-01
2.12283731e+00 -2.43751025e+00 2.13056371e-01 1.45700008e-01
3.75340432e-01 2.15944856e-01 -1.06945165e-01 3.03196222e-01
-1.54355988e-02 5.08947849e-01 -5.87622344e-01 1.78345546e-01
2.55446583e-01 -7.90523216e-02 -1.24245070e-01 -2.02631094e-02
-4.70622294e-02 9.07271147e-01 -1.22751677e+00 -4.37985092e-01
-1.72045007e-01 -1.69936139e-02 -8.91576350e-01 2.19863400e-01
-3.18154931e-01 -4.02739085e-02 -3.75697523e-01 4.91379857e-01
4.11157668e-01 -3.78622204e-01 1.41309693e-01 1.78564146e-01
-1.98461667e-01 5.55633605e-01 -9.97041643e-01 1.80804551e+00
-8.40399683e-01 6.04710758e-01 -3.63772810e-01 -5.39114058e-01
1.23192394e+00 3.24634314e-02 -1.05735175e-01 -8.82473528e-01
1.10662274e-01 5.73724449e-01 2.36315131e-01 -4.54873443e-01
5.58993220e-01 3.09148937e-01 -5.78699946e-01 4.19071764e-01
-2.07440659e-01 -3.85380566e-01 4.23757911e-01 2.22099900e-01
1.30753684e+00 1.58432260e-01 5.77029407e-01 -5.90775788e-01
6.97104216e-01 -1.00151459e-02 5.06173015e-01 5.34157038e-01
-2.43695769e-02 5.36375165e-01 9.87870276e-01 -2.73499876e-01
-1.14494419e+00 -8.81977737e-01 -3.68182473e-02 9.71306682e-01
1.94243833e-01 -1.02457333e+00 -1.13559508e+00 -9.39017951e-01
-2.34979838e-01 8.33295166e-01 -4.17721510e-01 -3.58848989e-01
-3.95605505e-01 -5.95070124e-01 6.86652720e-01 3.43639225e-01
3.99469018e-01 -7.04312801e-01 -5.97313225e-01 9.02015567e-02
-4.29649919e-01 -5.62079966e-01 -5.91631472e-01 8.58018845e-02
-6.71193838e-01 -1.07282937e+00 -3.47516805e-01 -6.69692576e-01
5.59221506e-01 3.22644264e-02 1.60382509e+00 4.99912053e-01
1.62707463e-01 -3.76518995e-01 -7.37649024e-01 -2.14792922e-01
-1.17201197e+00 2.52307743e-01 -4.62042570e-01 -4.85734880e-01
5.25983393e-01 -3.56205046e-01 -2.42748648e-01 4.25968587e-01
-8.60947549e-01 2.39154980e-01 7.15026855e-01 6.55780673e-01
-2.32671276e-02 -1.26465604e-01 1.19343229e-01 -1.00540709e+00
8.72567475e-01 -3.83354813e-01 -6.39791191e-01 4.92465585e-01
-1.05377913e+00 6.40231371e-01 9.73426402e-01 1.40780389e-01
-7.31783986e-01 -2.96043068e-01 -2.97824085e-01 2.25429893e-01
2.80826271e-01 8.00403893e-01 1.61698218e-02 -1.26960337e-01
1.08309364e+00 1.21330559e-01 -3.45249891e-01 -5.51862001e-01
1.21011145e-01 9.88311470e-01 3.89842629e-01 -6.66354418e-01
7.50656664e-01 -3.15376073e-01 -3.52097958e-01 3.03002633e-03
-5.30977547e-01 -2.85359979e-01 -2.05997169e-01 -3.77484709e-02
6.53075337e-01 -6.68317676e-01 -4.68158305e-01 2.89573014e-01
-1.30836058e+00 7.29144961e-02 -5.57915121e-02 2.98168987e-01
-3.33204418e-01 6.13584638e-01 -4.73132938e-01 -4.93911564e-01
-2.14044929e-01 -1.70297134e+00 9.18765306e-01 3.10078543e-02
-6.43770576e-01 -6.05799496e-01 2.36319184e-01 7.01424479e-02
5.95386088e-01 9.56657529e-02 1.54212523e+00 -2.88925946e-01
-3.74818951e-01 -1.47944182e-01 -4.89625156e-01 4.36663568e-01
-3.64839137e-02 3.14595044e-01 -7.64730513e-01 7.26338848e-03
-1.23247966e-01 -1.56075895e-01 5.73990643e-01 -1.33034393e-01
1.15008879e+00 -3.75146270e-01 -6.86831996e-02 5.27704060e-01
1.74419177e+00 2.75841564e-01 8.05242598e-01 2.97178686e-01
6.26599491e-01 3.73429120e-01 4.14042592e-01 4.80540395e-01
4.42348748e-01 7.93008268e-01 4.69143987e-01 3.45097303e-01
-2.35706046e-01 -4.10022378e-01 6.16851270e-01 1.53877199e+00
1.38992503e-01 -1.55836955e-01 -1.44730270e+00 3.15609902e-01
-1.69630182e+00 -5.07673323e-01 -5.51519275e-01 2.34187388e+00
1.06218731e+00 5.17961442e-01 -6.26149252e-02 4.09909278e-01
4.51531231e-01 -1.97922736e-01 9.99859534e-03 -8.40372622e-01
2.32591838e-01 1.25636846e-01 4.70782310e-01 5.01783669e-01
-5.49722254e-01 6.55700266e-01 6.40680552e+00 9.84104395e-01
-8.93822253e-01 2.48781383e-01 3.22962940e-01 5.78603029e-01
-8.17557871e-01 3.77056509e-01 -4.09557194e-01 8.53522241e-01
1.10449016e+00 -4.63088989e-01 6.16452694e-01 1.12799966e+00
-1.70343995e-01 1.11319534e-01 -1.30212569e+00 9.56199944e-01
-1.00595422e-01 -1.01509428e+00 -3.91989425e-02 -3.94085377e-01
1.01633537e+00 -5.71765900e-02 -2.30172411e-01 5.11338055e-01
5.99601746e-01 -9.45354998e-01 1.15762448e+00 5.66178560e-01
7.23096848e-01 -4.93468821e-01 9.35364902e-01 3.58976513e-01
-1.06315327e+00 -1.23906009e-01 -1.89665616e-01 -2.05746859e-01
-4.49731827e-01 6.57233655e-01 -5.86772859e-01 3.61143827e-01
4.57693309e-01 4.89431024e-01 -1.00177872e+00 1.17636967e+00
-2.21875295e-01 4.80498284e-01 3.02274078e-01 -4.34641987e-01
1.29053742e-01 8.14896673e-02 3.05390328e-01 1.45245695e+00
5.04718363e-01 -4.85203981e-01 -3.59277762e-02 1.12650907e+00
-2.23064676e-01 3.46406788e-01 -3.52552146e-01 -1.37863323e-01
4.78832394e-01 8.82596135e-01 -6.89846456e-01 -2.95352191e-01
-7.48769522e-01 7.44906247e-01 1.03340484e-01 -3.09590623e-02
-1.00587118e+00 -7.25890696e-01 5.59558749e-01 2.03060010e-03
-1.82379052e-01 -1.61886483e-01 -7.76987851e-01 -1.14514816e+00
2.63301641e-01 -1.20488453e+00 5.01577519e-02 -6.90951049e-01
-7.28745520e-01 6.16420388e-01 -2.91166812e-01 -1.39928639e+00
-1.44477248e-01 -4.35679108e-01 -3.49098057e-01 7.74469376e-01
-1.20537937e+00 -4.81517226e-01 -6.22199178e-01 -1.06543414e-01
5.91715276e-01 -1.89774260e-01 8.33936214e-01 5.71927190e-01
-3.13009948e-01 8.67781818e-01 3.00793886e-01 2.53655583e-01
5.77797830e-01 -1.46745956e+00 8.81643772e-01 1.15109003e+00
-5.89529015e-02 8.81045401e-01 9.14673865e-01 -5.00487089e-01
-1.38045335e+00 -8.94269049e-01 1.20318830e+00 -8.32780182e-01
6.45467758e-01 -2.93547302e-01 -7.62024403e-01 1.92210570e-01
-3.89372073e-02 -3.92105937e-01 3.83259892e-01 -8.35744366e-02
-7.33561695e-01 -4.28856798e-02 -7.84448564e-01 4.32857871e-01
8.77793372e-01 -6.47967041e-01 -4.69331026e-01 -3.97471013e-03
8.63684058e-01 -3.20201069e-01 -9.18143570e-01 4.11409020e-01
6.07516706e-01 -1.21034157e+00 4.21907872e-01 -2.55677551e-01
9.91715193e-01 -5.29190302e-01 -3.90050739e-01 -1.25295401e+00
-3.77476811e-01 -4.08318311e-01 2.56588429e-01 1.29021251e+00
7.74809003e-01 -9.79151577e-02 4.20950443e-01 5.18079877e-01
-2.75048196e-01 -4.13124502e-01 -3.38175178e-01 -7.44097114e-01
1.83979601e-01 -4.37022477e-01 1.02749622e+00 1.01443160e+00
4.09832209e-01 3.32131147e-01 -1.69259578e-01 -4.26190972e-01
2.78720051e-01 1.41889870e-01 7.51966834e-01 -1.27328861e+00
-4.84482557e-01 -9.88839746e-01 -4.38309908e-01 -7.42964029e-01
-1.37834214e-02 -1.30101180e+00 1.85149774e-01 -1.37716436e+00
5.40765226e-01 -3.29316735e-01 -1.04200602e-01 2.91492581e-01
-2.66673774e-01 -3.49108018e-02 1.52268618e-01 3.00377399e-01
-6.17644727e-01 1.11365013e-01 9.28539395e-01 -2.88242966e-01
8.39372650e-02 -1.66422427e-01 -8.25356781e-01 7.44681180e-01
5.04431188e-01 -6.80257440e-01 -2.51739293e-01 -7.80025959e-01
1.04350090e+00 1.60637006e-01 -6.85189813e-02 -1.31785822e+00
-3.60506331e-03 -1.70141086e-01 -3.56038421e-01 1.38542503e-01
-7.55265474e-01 -8.80830526e-01 3.05016100e-01 6.87066317e-01
-5.93567550e-01 3.97582024e-01 -1.87343210e-01 1.31062031e-01
-4.75691259e-01 -8.26133192e-01 8.94496500e-01 -3.47828895e-01
-9.56993401e-01 -2.68440187e-01 -1.46300644e-01 3.08522195e-01
7.32933283e-01 -1.29301235e-01 -4.52519804e-01 1.30712138e-02
-1.56768873e-01 -1.27742458e-02 9.37137604e-01 7.14141488e-01
3.47231150e-01 -1.31143248e+00 -7.42375851e-01 2.37582296e-01
7.00455546e-01 -5.67138672e-01 -3.94211620e-01 5.94134867e-01
-8.59798551e-01 4.25009668e-01 -2.27865115e-01 -4.57432210e-01
-1.13908112e+00 4.31743056e-01 2.12900802e-01 -4.28881645e-01
-2.27972753e-02 7.36690760e-01 2.36351237e-01 -5.73713541e-01
2.74628490e-01 -7.78582692e-01 -8.81566480e-02 -3.44564646e-01
5.33120096e-01 4.44841594e-01 4.26822245e-01 -3.29185337e-01
-3.82760018e-01 5.60852349e-01 1.72046855e-01 1.51407942e-01
8.92095625e-01 6.82424521e-03 -5.24312258e-01 5.78010142e-01
1.23650348e+00 1.96969703e-01 -6.28067553e-01 -1.04297057e-01
4.27946717e-01 -6.44299090e-01 -6.63022995e-02 -9.26475942e-01
-7.82468736e-01 9.82649684e-01 6.51963115e-01 3.41057032e-01
8.88621092e-01 -3.44473720e-01 7.36912906e-01 2.94402093e-01
6.96023881e-01 -9.02153671e-01 8.58872607e-02 7.82177627e-01
4.97261047e-01 -1.13898945e+00 -4.57719594e-01 -2.68320918e-01
-4.00738329e-01 1.21107566e+00 6.54487491e-01 1.82315230e-01
3.66367847e-01 5.59163034e-01 -1.03128783e-01 3.60092707e-02
-5.81464410e-01 6.20788820e-02 4.59438533e-01 1.48828685e-01
1.10636675e+00 2.00366572e-01 -7.57798791e-01 6.73183382e-01
-4.11712497e-01 2.35240757e-01 7.08346188e-01 7.52535224e-01
-7.39201605e-01 -1.43119264e+00 -1.51339948e-01 6.25802457e-01
-7.04743683e-01 -2.61942923e-01 -4.99499351e-01 3.19882065e-01
1.52443811e-01 7.94016361e-01 -3.43242854e-01 -8.89160693e-01
3.54656637e-01 7.70416111e-02 3.08775783e-01 -8.18441331e-01
-8.64606857e-01 -5.22634566e-01 -1.18176714e-02 -5.27251661e-01
-1.27360806e-01 -2.20337868e-01 -1.15341926e+00 -6.03250265e-01
-3.72980267e-01 6.09722972e-01 7.10447252e-01 7.93894470e-01
2.15914816e-01 6.92389429e-01 6.29444659e-01 9.79966149e-02
-5.81160963e-01 -8.66615236e-01 -1.45755738e-01 5.56177139e-01
3.04818720e-01 -4.00360107e-01 -2.77868390e-01 2.03628957e-01] | [7.69765567779541, 7.897199630737305] |
f4b00924-d467-4854-ba9b-bb4923159b3b | fisher-efficient-inference-of-intractable | 1805.07454 | null | https://arxiv.org/abs/1805.07454v5 | https://arxiv.org/pdf/1805.07454v5.pdf | Fisher Efficient Inference of Intractable Models | Maximum Likelihood Estimators (MLE) has many good properties. For example, the asymptotic variance of MLE solution attains equality of the asymptotic Cram{\'e}r-Rao lower bound (efficiency bound), which is the minimum possible variance for an unbiased estimator. However, obtaining such MLE solution requires calculating the likelihood function which may not be tractable due to the normalization term of the density model. In this paper, we derive a Discriminative Likelihood Estimator (DLE) from the Kullback-Leibler divergence minimization criterion implemented via density ratio estimation and a Stein operator. We study the problem of model inference using DLE. We prove its consistency and show that the asymptotic variance of its solution can attain the equality of the efficiency bound under mild regularity conditions. We also propose a dual formulation of DLE which can be easily optimized. Numerical studies validate our asymptotic theorems and we give an example where DLE successfully estimates an intractable model constructed using a pre-trained deep neural network. | ['Yu Chen', 'Wittawat Jitkrittum', 'Takafumi Kanamori', 'Song Liu'] | 2018-05-18 | fisher-efficient-inference-of-intractable-1 | http://papers.nips.cc/paper/9083-fisher-efficient-inference-of-intractable-models | http://papers.nips.cc/paper/9083-fisher-efficient-inference-of-intractable-models.pdf | neurips-2019-12 | ['density-ratio-estimation'] | ['methodology'] | [-1.28846571e-01 1.92132011e-01 -1.21204875e-01 -4.20163959e-01
-1.09353018e+00 -2.85373837e-01 1.35462061e-01 -3.38488340e-01
-6.37995243e-01 1.07170343e+00 -3.03481907e-01 -4.51638639e-01
-4.79921490e-01 -6.28364205e-01 -8.60140979e-01 -7.23619998e-01
-1.56608269e-01 3.46608281e-01 -3.90653998e-01 4.49062198e-01
2.23718852e-01 5.99280953e-01 -1.08890021e+00 -7.24206209e-01
9.37990546e-01 1.37102687e+00 2.97106504e-01 6.98878407e-01
2.86843896e-01 5.57775795e-01 -4.07688588e-01 -1.57581404e-01
3.26893151e-01 -5.30236423e-01 -2.49635816e-01 -1.25371128e-01
3.13603431e-01 -7.62581408e-01 -1.95588157e-01 1.54316473e+00
4.24728304e-01 4.97400537e-02 1.29008055e+00 -1.29810369e+00
-6.37265980e-01 2.39568174e-01 -9.25186157e-01 1.42442450e-01
-1.20881749e-02 -6.76145911e-01 8.64522994e-01 -1.11341965e+00
1.89777076e-01 1.09297013e+00 1.11606181e+00 2.69016147e-01
-1.08416438e+00 -6.90589905e-01 -6.09557211e-01 -1.57263324e-01
-2.13588405e+00 -5.68399251e-01 3.10154736e-01 -1.47263646e-01
3.49369973e-01 -6.07587323e-02 1.84571609e-01 6.46600187e-01
3.42633665e-01 8.96301091e-01 8.49362612e-01 -4.95177180e-01
4.11380470e-01 4.01250720e-01 -9.37165972e-03 1.07618773e+00
8.92581344e-01 2.34452961e-03 -1.06955506e-01 -3.29357535e-01
1.32710981e+00 -2.94261098e-01 -4.49942827e-01 -4.02978718e-01
-7.94271410e-01 1.01321375e+00 1.01511210e-01 1.47118583e-01
-5.22093713e-01 4.55168635e-01 -1.95256412e-01 6.45430833e-02
2.52719522e-01 1.40859894e-02 -2.43397444e-01 1.01148255e-01
-1.06022942e+00 1.15123512e-02 1.18745732e+00 1.13826358e+00
5.13318300e-01 1.81270987e-01 1.04706861e-01 7.78571725e-01
8.76843393e-01 1.00365233e+00 6.07127771e-02 -1.21401858e+00
1.95677504e-01 -3.18385690e-01 4.42508608e-01 -8.60670328e-01
-1.68581843e-01 -8.04068089e-01 -1.15263629e+00 -1.43711427e-02
7.23532736e-01 -5.79962969e-01 -3.19417298e-01 1.89914393e+00
1.47441298e-01 2.25550652e-01 4.28266777e-03 8.83327961e-01
1.12991229e-01 7.12601602e-01 -3.21843415e-01 -6.12637699e-01
8.23559582e-01 -3.26548278e-01 -1.00481212e+00 -2.44998708e-01
6.86926246e-01 -4.39965516e-01 6.41656637e-01 3.52734983e-01
-1.06773853e+00 -1.63142029e-02 -1.18341351e+00 1.82248786e-01
3.13594371e-01 4.95561689e-01 3.96731973e-01 8.09567094e-01
-9.21176076e-01 6.90484703e-01 -8.80853891e-01 -1.11086160e-01
2.77460873e-01 2.49175340e-01 -4.13394749e-01 1.08608857e-01
-8.53498161e-01 9.34077322e-01 3.35292220e-01 3.63608629e-01
-8.53904247e-01 -5.40890217e-01 -7.86664248e-01 3.51608813e-01
2.09023908e-01 -6.32294595e-01 1.41254544e+00 -7.46359885e-01
-1.49451029e+00 4.56275821e-01 -3.11889470e-01 -5.16322553e-01
5.29672682e-01 -3.38207245e-01 2.37770621e-02 1.86168060e-01
1.85345903e-01 1.63583457e-01 6.74318910e-01 -9.14830029e-01
-4.41140592e-01 -3.63135904e-01 -3.01112264e-01 -2.18510777e-02
-1.06723063e-01 -2.08762631e-01 -1.82892501e-01 -2.90151685e-01
4.06269461e-01 -6.65470123e-01 -1.37039617e-01 3.63451362e-01
-3.18874687e-01 3.22562568e-02 2.51731306e-01 -7.96784043e-01
1.38097775e+00 -2.35022902e+00 -4.87079740e-01 5.39668798e-01
1.11207560e-01 -8.14239383e-02 1.21192627e-01 2.30503917e-01
6.25125766e-02 -1.49062870e-03 -3.42635423e-01 -2.72025257e-01
2.69143313e-01 -2.34508212e-03 -1.54145345e-01 1.33286071e+00
-1.09808140e-01 5.41993201e-01 -7.07360685e-01 -6.30795002e-01
-3.75262722e-02 4.98613209e-01 -5.33414662e-01 8.37788582e-02
3.44857186e-01 -7.85415620e-02 -5.98437965e-01 3.64104092e-01
9.85322058e-01 -6.69445574e-01 2.13284269e-01 -1.91630632e-01
-2.37139836e-02 -8.37927461e-02 -1.63276005e+00 1.10930109e+00
-5.45346558e-01 8.72292042e-01 4.64156181e-01 -1.20493639e+00
8.62028420e-01 3.05513769e-01 1.19107038e-01 -1.52419001e-01
5.93280852e-01 6.90570116e-01 -1.89526290e-01 -1.12024747e-01
1.85206696e-01 -5.93191743e-01 7.08261207e-02 4.06289816e-01
2.22124774e-02 2.22649619e-01 -1.17091753e-01 2.43923783e-01
7.20506132e-01 -1.43685281e-01 9.21245098e-01 -7.34841287e-01
3.21661711e-01 -7.44864225e-01 4.35831070e-01 1.33939624e+00
-2.62457848e-01 4.85029638e-01 6.02334499e-01 2.12861076e-01
-9.77170408e-01 -1.34098113e+00 -7.88547039e-01 4.03264731e-01
4.45165187e-02 2.26315111e-02 -8.44728410e-01 -2.89011151e-01
-9.90177542e-02 9.12647843e-01 -5.01110971e-01 -4.62736525e-02
-5.62617779e-02 -9.00823355e-01 6.56952679e-01 4.78620410e-01
7.53001273e-01 -1.67118981e-01 -4.83934492e-01 1.30619600e-01
-1.05462097e-01 -9.10318017e-01 -3.72638702e-01 1.88431352e-01
-7.25951850e-01 -8.18050683e-01 -1.03527284e+00 -4.52240139e-01
6.15438521e-01 1.75189376e-01 6.34834230e-01 -2.86467910e-01
1.69883132e-01 4.05718982e-01 1.94057375e-01 -3.38305593e-01
-3.36438537e-01 -2.40871638e-01 4.03826356e-01 -2.81067174e-02
4.42820728e-01 -6.62663102e-01 -4.16971296e-01 2.53843337e-01
-5.71783245e-01 -2.96761811e-01 6.91283584e-01 8.55552971e-01
6.53901994e-01 -3.63826044e-02 9.15219903e-01 -3.80029976e-01
6.50473952e-01 -5.84915638e-01 -1.14424503e+00 3.93471152e-01
-5.77675343e-01 4.01050866e-01 3.73749018e-01 -3.23805511e-01
-1.10125661e+00 -3.47462185e-02 -1.17452005e-02 -4.76761043e-01
1.84803352e-01 6.52124107e-01 -1.23598449e-01 2.49025319e-02
5.59632003e-01 1.53843656e-01 -3.77917755e-03 -6.09231174e-01
2.13273779e-01 1.03939331e+00 6.63591027e-01 -5.10070324e-01
5.17386019e-01 4.17932659e-01 3.89666736e-01 -9.79790747e-01
-1.32845509e+00 -2.89303631e-01 -2.95934707e-01 -1.19864829e-02
5.09000719e-01 -7.95052290e-01 -8.99660945e-01 1.29459307e-01
-1.14420891e+00 -2.70568673e-02 -6.08575009e-02 1.15486670e+00
-8.82189929e-01 5.05796790e-01 -6.11904860e-01 -1.45176971e+00
-1.10271014e-01 -6.85282588e-01 8.26773047e-01 1.52363792e-01
-3.20063606e-02 -1.11045790e+00 5.17460927e-02 -3.86004180e-01
2.62919545e-01 1.46698385e-01 4.41918254e-01 -6.74465716e-01
-3.91057104e-01 -5.37957966e-01 -7.15384722e-01 5.45009434e-01
-8.83263946e-02 -2.24987149e-01 -7.46406555e-01 -2.70931244e-01
5.48181236e-01 -2.56681889e-02 6.84744239e-01 8.87792945e-01
1.07771862e+00 -5.52594841e-01 -3.03494364e-01 8.47500980e-01
1.64618683e+00 1.58276677e-03 5.00603497e-01 5.41910790e-02
2.56655365e-01 3.78774740e-02 4.88829046e-01 8.87442231e-01
2.56746775e-03 1.98131785e-01 3.64710055e-02 4.28378820e-01
3.57156903e-01 -1.36548430e-01 3.28273684e-01 8.88480544e-01
1.95080340e-01 -2.09624082e-01 -6.43612981e-01 3.12343180e-01
-1.82309520e+00 -9.04744208e-01 -2.52438009e-01 2.54117489e+00
8.46013665e-01 -1.87206686e-01 -2.49777719e-01 -1.09241791e-01
7.68065929e-01 -2.88775206e-01 -5.83289385e-01 -2.66393840e-01
7.51100257e-02 1.05806112e-01 9.79224682e-01 6.84898555e-01
-9.92198765e-01 3.40537786e-01 7.41817093e+00 1.26258314e+00
-5.62468052e-01 3.42480928e-01 3.52389723e-01 1.50933787e-01
-2.21516415e-02 -6.82886392e-02 -1.02895296e+00 3.55543435e-01
1.10411251e+00 -5.65258265e-01 3.07451248e-01 1.17828536e+00
1.91817895e-01 -4.43196118e-01 -1.15634012e+00 1.29327345e+00
-2.00425051e-02 -9.19678807e-01 -4.38688934e-01 5.40777683e-01
6.74597144e-01 -6.54964224e-02 4.66664545e-02 1.55130610e-01
1.63301036e-01 -1.00985920e+00 6.33683622e-01 7.64059901e-01
9.33306873e-01 -8.30135942e-01 8.67188096e-01 6.82529986e-01
-9.57917333e-01 1.29431203e-01 -7.74203181e-01 -1.69584639e-02
3.56399983e-01 9.55436349e-01 -8.24984729e-01 2.33911037e-01
1.13967441e-01 3.83195102e-01 2.58441339e-03 1.35836065e+00
-6.70414418e-02 4.84717101e-01 -9.89429772e-01 -2.13685662e-01
2.26563871e-01 -6.94175124e-01 6.85606182e-01 1.34393048e+00
8.77959132e-01 -1.56384949e-02 -3.41291189e-01 1.23550880e+00
-1.48872018e-01 -5.42591102e-02 -5.55602014e-01 -3.63261737e-02
8.27888787e-01 8.43426943e-01 -5.67574024e-01 -4.09813464e-01
-3.62343520e-01 7.34336078e-01 2.21527517e-01 5.33812463e-01
-1.04127741e+00 -6.91885531e-01 4.11825567e-01 -1.36671022e-01
4.89340186e-01 -2.86200136e-01 -2.88113505e-01 -1.17707884e+00
4.33041295e-03 -4.42018867e-01 8.74492154e-02 -6.30748034e-01
-1.28362417e+00 -5.17693944e-02 3.02259147e-01 -1.04545176e+00
-4.74326491e-01 -8.21060240e-01 -2.76991814e-01 9.52487350e-01
-1.03001165e+00 -2.43530542e-01 1.69228017e-01 2.73239911e-01
-1.23323180e-01 -1.14069790e-01 6.12515986e-01 5.42304814e-01
-6.59869790e-01 8.39653075e-01 8.53491843e-01 1.12280868e-01
3.77054602e-01 -1.11691976e+00 -2.80629456e-01 8.77498150e-01
-1.27220169e-01 8.62671971e-01 9.48998570e-01 -5.43147445e-01
-1.32431364e+00 -7.25874960e-01 8.24033678e-01 3.44165452e-02
9.52361763e-01 8.20926130e-02 -7.60961890e-01 1.02353966e+00
-2.07591206e-01 -1.08708657e-01 6.50841236e-01 -5.68815228e-03
-1.38599589e-01 5.58770448e-02 -1.20741713e+00 3.95568699e-01
7.00281978e-01 -4.88977045e-01 -5.48914194e-01 2.68154830e-01
1.22566603e-01 -4.18223143e-02 -7.23549366e-01 5.09486318e-01
9.58975852e-01 -1.10232258e+00 7.14847326e-01 -2.05765650e-01
4.25412357e-02 -1.57380119e-01 -6.81767583e-01 -8.94413292e-01
-4.70337160e-02 -5.04075646e-01 -3.93140823e-01 8.81393909e-01
3.08261454e-01 -6.91492319e-01 4.64700341e-01 5.41446209e-01
3.07186455e-01 -7.44495213e-01 -1.14771974e+00 -1.23485804e+00
-2.14601215e-02 -5.75110078e-01 1.37078032e-01 6.72798336e-01
-1.44751109e-02 1.02193870e-01 -5.88045955e-01 3.24361205e-01
1.20770764e+00 -3.73280525e-01 5.27603924e-01 -1.11680675e+00
-5.89532673e-01 -3.84436071e-01 -1.51113003e-01 -1.87797379e+00
3.30903918e-01 -7.56335735e-01 3.12683225e-01 -1.45057142e+00
4.51306462e-01 -1.76711813e-01 -1.58770040e-01 -2.22666591e-01
1.83035970e-01 -1.06510088e-01 -3.82877477e-02 1.25517128e-02
-5.26649714e-01 7.59339213e-01 9.82193470e-01 3.63750845e-01
7.74120539e-02 6.72595203e-02 -6.10830128e-01 1.27496600e+00
6.53391719e-01 -7.09644616e-01 -2.59587407e-01 -1.48082569e-01
4.60930020e-01 4.29356277e-01 3.83918583e-01 -9.53969061e-01
1.89943537e-01 8.60718712e-02 4.17498678e-01 -5.84620416e-01
2.84513801e-01 -6.09013736e-01 1.94120288e-01 2.62636364e-01
-2.62455165e-01 -4.41655040e-01 -1.87508315e-01 7.25001276e-01
6.63613826e-02 -1.11692834e+00 1.04618776e+00 1.12792991e-01
-1.40893847e-01 1.88401416e-01 -6.34415030e-01 -6.08512312e-02
5.30995429e-01 -3.78818959e-02 6.17709048e-02 -8.67113352e-01
-5.55811286e-01 -1.23944074e-01 7.19912127e-02 -4.72851932e-01
7.32701361e-01 -1.44658875e+00 -6.26835227e-01 1.24824502e-01
-3.84404480e-01 -2.44524270e-01 -7.40648732e-02 1.22672379e+00
-5.91327429e-01 3.83231342e-01 3.86796296e-01 -3.67483705e-01
-6.40628636e-01 2.56671131e-01 6.45479560e-01 -5.89325614e-02
-4.36840653e-01 5.82469642e-01 4.48357165e-01 -1.00661181e-01
2.89191216e-01 -3.85991186e-02 2.65294284e-01 -3.08018208e-01
7.85633266e-01 6.37357295e-01 -4.90564048e-01 -4.63083267e-01
-3.48376393e-01 3.42137903e-01 2.86776304e-01 -6.17508352e-01
9.22191083e-01 -4.67706233e-01 -8.28958452e-02 4.98886943e-01
1.81262970e+00 -2.70519238e-02 -1.15371776e+00 -1.78348124e-01
5.38012981e-02 -5.46302736e-01 3.69625777e-01 -2.06360951e-01
-6.58578813e-01 8.56000662e-01 6.65615857e-01 1.01188749e-01
7.89706588e-01 -6.13620915e-02 6.00141585e-01 9.23057199e-01
3.36506158e-01 -9.01254117e-01 -3.81991655e-01 5.56004584e-01
7.39015400e-01 -1.01638567e+00 1.45787567e-01 -3.46812196e-02
-1.58722103e-01 1.05047107e+00 3.63657363e-02 -5.41014433e-01
1.03207195e+00 2.78527647e-01 -4.14261848e-01 1.74992979e-01
-3.33793610e-01 2.43786573e-02 3.00471753e-01 3.29287648e-01
4.72762972e-01 5.81422336e-02 -6.12961173e-01 7.74262667e-01
-2.40515351e-01 1.60489202e-01 5.36794603e-01 4.03708845e-01
-8.61383379e-01 -3.26057047e-01 -3.32689822e-01 4.92307067e-01
-6.78543389e-01 -1.61536232e-01 2.90673077e-01 9.18128788e-01
-5.79439521e-01 8.48899066e-01 1.96460575e-01 2.72552103e-01
-2.75666744e-01 3.27279456e-02 7.27550864e-01 -1.73526242e-01
6.87310815e-01 4.94245917e-01 -4.05994132e-02 -3.38551402e-01
-3.24130416e-01 -5.79905868e-01 -1.05327547e+00 -5.55092156e-01
-8.32479835e-01 2.60403663e-01 7.19612122e-01 1.20810926e+00
6.68737432e-03 -8.85874256e-02 5.40235758e-01 -5.86605787e-01
-1.24256587e+00 -1.03313720e+00 -1.35723054e+00 -2.53100514e-01
4.71029103e-01 -5.82580626e-01 -8.66824329e-01 -1.53728515e-01] | [7.149463176727295, 4.013396263122559] |
07f73fa8-4519-4d18-a679-faef2638e00e | synthesis-of-a-machine-learning-model-for | null | null | https://www.researchgate.net/publication/347631637_Synthesis_of_a_Machine_Learning_Model_for_Detecting_Computer_Attacks_Based_on_the_CICIDS2017_Dataset | https://ispranproceedings.elpub.ru/jour/article/view/1348/1147 | Synthesis of a Machine Learning Model for Detecting Computer Attacks Based on the CICIDS2017 Dataset | The paper deals with the construction and practical implementation of the model of computer attack detection based on machine learning methods. Among available public datasets one of the most relevant was chosen - CICIDS2017. For this dataset, the procedures of data preprocessing and sampling were developed in detail. In order to reduce computation time, the only class of computer attacks (brute force, XSS, SQL injection) was left in the training set. The procedure of feature space construction is described sequentially, which allowed to significantly reduce its dimensions - from 85 to 10 most important features. The quality assessment of ten most common machine learning models on the obtained pre-processed dataset was made. Among the models (algorithms) that demonstrated the best results (k-nearest neighbors, decision tree, random forest, AdaBoost, logistic regression), taking into account the minimum time of execution, the choice of random forest model was justified. А quasi-optimal selection of hyper parameters was carried out, which made it possible to improve the quality of the model in comparison with the previously published research results. The synthesized model of attack detection was tested on real network traffic. The model has shown its validity only under the condition of training on data collected in a specific network, since important features depend on the physical structure of the network and the settings of the equipment used. The conclusion was made that it is possible to use machine learning methods to detect computer attacks taking into account these limitations. | ['Andrey Matskevich', 'Maxim Goryunov', 'Dmitry Rybolovlev'] | 2020-01-01 | null | null | null | proceedings-of-the-institute-for-system | ['network-intrusion-detection'] | ['miscellaneous'] | [ 1.06145546e-01 -2.13541687e-01 -1.27062351e-01 -1.22388326e-01
-2.62797461e-03 -3.08522493e-01 4.15896267e-01 6.75539553e-01
-8.28243732e-01 6.44210160e-01 -2.34783232e-01 -7.85689175e-01
-6.21599793e-01 -1.13726473e+00 -1.07422955e-01 -5.55197597e-01
-2.93610841e-01 5.88496804e-01 3.63955438e-01 -5.06560542e-02
6.68149650e-01 1.12840307e+00 -1.64155817e+00 2.62838632e-01
4.31068867e-01 8.19974601e-01 -1.49538741e-01 4.80535358e-01
-1.25469148e-01 5.99629343e-01 -8.82804096e-01 -3.05199832e-01
2.97444969e-01 -1.84201494e-01 -7.80617476e-01 -1.30221084e-01
-4.45819825e-01 -2.89836586e-01 1.32363383e-02 7.40944624e-01
4.06020999e-01 -8.36949944e-02 8.22894931e-01 -1.16336942e+00
2.21010700e-01 4.51278895e-01 -1.10255405e-01 3.55901122e-01
5.56530595e-01 -3.12857404e-02 6.03744090e-01 -4.25481498e-01
5.85589468e-01 9.65394557e-01 4.10685539e-01 6.70170337e-02
-1.18112314e+00 -7.28005052e-01 -2.93440312e-01 5.92968762e-01
-1.46263981e+00 1.08206086e-01 6.27224743e-01 -5.55252135e-01
9.22141016e-01 3.46879065e-01 6.52500570e-01 8.80970895e-01
2.25577787e-01 1.89297367e-03 1.41279757e+00 -9.96404171e-01
4.78775859e-01 7.82186568e-01 4.99049634e-01 3.40696752e-01
6.95165813e-01 2.35328674e-01 5.02214096e-02 -5.58610022e-01
2.08911464e-01 -1.05488990e-02 7.32974187e-02 -1.70178011e-01
-4.81335878e-01 9.44581568e-01 -9.47155133e-02 9.49443758e-01
-5.50423801e-01 -6.44732952e-01 6.96219802e-01 3.95738721e-01
7.00860322e-02 3.71263057e-01 -5.61402082e-01 -2.60202140e-01
-8.44542801e-01 -8.32512081e-02 1.01115906e+00 3.17573756e-01
7.08315015e-01 1.60500526e-01 3.82421941e-01 2.56315768e-01
1.79550096e-01 3.42901826e-01 4.19054568e-01 6.00288063e-02
3.87759596e-01 8.08776975e-01 -1.67060778e-01 -1.22101688e+00
-4.59552526e-01 -4.06238675e-01 -5.10814250e-01 4.82443988e-01
6.65848434e-01 -3.15669447e-01 -4.46531951e-01 1.14103651e+00
2.35349044e-01 -5.66462688e-02 1.45384699e-01 3.19390774e-01
3.63430321e-01 5.41527450e-01 3.40658873e-01 -5.35273492e-01
1.18435562e+00 2.13027671e-02 -7.46555090e-01 5.99945903e-01
7.33913958e-01 -8.02137494e-01 5.16981423e-01 8.66529286e-01
-3.05310011e-01 -5.22491038e-01 -1.01822114e+00 1.04853082e+00
-8.42701554e-01 1.44773191e-02 4.71744418e-01 1.38648462e+00
-5.76923966e-01 6.50490522e-01 -3.42830628e-01 -5.58050036e-01
6.35354267e-03 3.97269666e-01 -4.51364279e-01 2.85012186e-01
-1.27442253e+00 1.21033669e+00 6.95465922e-01 3.31497602e-02
-5.49989700e-01 -1.04901172e-01 -1.48480639e-01 2.43270807e-02
3.76942784e-01 -7.32603250e-03 5.57888329e-01 -7.93786108e-01
-1.25669873e+00 7.22585797e-01 1.95356965e-01 -7.17407882e-01
5.25570452e-01 7.33573362e-02 -8.64671350e-01 3.57001036e-01
-3.68150234e-01 -2.59490907e-01 6.60886347e-01 -1.03072238e+00
-4.72361922e-01 -3.73913258e-01 1.28835649e-03 -3.78028989e-01
-5.69495976e-01 4.33664888e-01 2.18119159e-01 -1.97209015e-01
-1.94345191e-01 -6.84949696e-01 -2.32845098e-01 -7.76198208e-01
-2.55641341e-01 -1.59286632e-04 8.75606656e-01 -7.01902211e-01
1.65366614e+00 -1.84226179e+00 -2.18942940e-01 9.99551296e-01
-2.60649323e-01 6.83759809e-01 6.02639794e-01 9.31839943e-01
-4.68385994e-01 1.17354199e-01 4.40858305e-02 4.10946131e-01
-1.72139332e-01 1.00856476e-01 -2.52121717e-01 4.04923260e-01
-1.89347550e-01 -7.75836632e-02 -2.26853624e-01 -7.21875966e-01
6.92001581e-01 3.22476268e-01 -3.29723209e-01 2.41268501e-01
1.33055121e-01 1.06783137e-01 -7.04147577e-01 4.77999568e-01
6.66942358e-01 4.34618771e-01 2.54217148e-01 -2.54398376e-01
-2.19331235e-01 -5.10652326e-02 -1.65295386e+00 5.80137610e-01
-5.36766231e-01 1.97744280e-01 -2.47170419e-01 -1.33792806e+00
1.45358384e+00 7.94859588e-01 7.90601492e-01 -5.36711514e-01
4.18064505e-01 3.80985707e-01 4.62677181e-02 -7.75227726e-01
-1.78675473e-01 2.85442740e-01 1.46661818e-01 5.25498807e-01
-9.30798203e-02 7.15549588e-02 4.06939775e-01 -1.37885809e-01
8.67684424e-01 -2.44503915e-01 5.91995478e-01 -2.81825870e-01
1.02109981e+00 9.48972404e-02 1.66201279e-01 3.59741330e-01
-1.13700725e-01 -2.26567388e-01 8.15202594e-01 -5.54395556e-01
-6.86028838e-01 -6.33372784e-01 -2.51900405e-01 5.65545321e-01
-3.79294991e-01 -1.89093202e-01 -9.46904480e-01 -7.69040942e-01
-3.13397169e-01 7.99057722e-01 -2.69933879e-01 1.95001215e-02
-5.56941390e-01 -9.17514801e-01 5.21980286e-01 -4.12197232e-01
3.10997009e-01 -1.40817845e+00 -8.21005464e-01 2.46761039e-01
2.49345660e-01 -8.90461862e-01 7.94310868e-01 3.72503608e-01
-1.09272480e+00 -1.53734827e+00 1.04288541e-01 -2.50381649e-01
4.98236597e-01 -1.76121265e-01 7.43685305e-01 3.80232096e-01
-6.46951020e-01 3.11202258e-01 -6.59674108e-01 -6.07900381e-01
-7.29998469e-01 1.28722742e-01 -2.91163176e-02 1.70514837e-01
6.79921865e-01 -5.70585907e-01 -3.08388025e-02 1.87743798e-01
-9.24588561e-01 -8.11310053e-01 6.98463619e-01 4.56886560e-01
-1.79338932e-01 6.69999659e-01 6.10280037e-01 -1.13093722e+00
7.32688427e-01 -4.96040970e-01 -7.45056450e-01 6.97296113e-02
-8.25459659e-01 -1.35908067e-01 8.86165857e-01 -2.07823753e-01
-7.53290236e-01 -5.59192970e-02 -3.55516702e-01 1.41949311e-01
-8.14199805e-01 2.15267345e-01 -4.35116440e-01 -1.42573699e-01
8.86410296e-01 1.69254020e-01 1.36662260e-01 -5.76433778e-01
-3.83973807e-01 7.55870104e-01 -2.28924617e-01 -5.30988932e-01
7.80458093e-01 1.21991463e-01 3.36178273e-01 -1.15083992e+00
-8.84898975e-02 -3.48795027e-01 -6.80695057e-01 -4.31987435e-01
6.73057735e-01 -2.72291720e-01 -8.37516129e-01 5.95678866e-01
-9.19813812e-01 2.99255043e-01 1.32950721e-02 8.11797261e-01
-1.78723097e-01 4.29577589e-01 -3.79665822e-01 -1.30530274e+00
-3.16936165e-01 -1.13471580e+00 -1.02020673e-01 -1.18649609e-01
-1.88802928e-01 -9.66144741e-01 -8.95549357e-02 1.00645006e-01
3.00135553e-01 4.08507943e-01 1.09419966e+00 -1.29376900e+00
-1.54797330e-01 -8.05612385e-01 1.19441003e-01 6.15646124e-01
-3.78467632e-03 4.91541684e-01 -8.04117978e-01 -1.13689110e-01
2.59009540e-01 6.06379025e-02 4.06778902e-01 7.61892870e-02
1.01349568e+00 -9.95132476e-02 -2.14568585e-01 5.57142794e-02
1.67286837e+00 7.61315048e-01 6.86506093e-01 6.65408313e-01
-5.26554696e-02 7.72111893e-01 9.34368253e-01 5.85029483e-01
-3.37127090e-01 8.63546669e-01 6.67536259e-01 -1.08092606e-01
4.49044943e-01 9.87340733e-02 2.36316100e-01 2.76263833e-01
-1.83713824e-01 -6.85728788e-02 -9.46227372e-01 2.02082828e-01
-1.23688519e+00 -1.30319440e+00 -4.04183865e-01 2.56682372e+00
1.23282254e-01 8.45747352e-01 6.71274543e-01 1.13945758e+00
8.27000558e-01 -1.10519618e-01 4.00698513e-01 -1.02990389e+00
1.59516096e-01 5.33659816e-01 4.50666875e-01 5.38488328e-01
-1.11994231e+00 3.09028000e-01 5.44627905e+00 8.44513178e-01
-1.21541238e+00 -1.87791258e-01 3.81702036e-01 2.19530791e-01
1.93484858e-01 1.19836897e-01 -9.06254828e-01 7.06445515e-01
1.38665390e+00 1.25798598e-01 1.85821638e-01 8.60751212e-01
4.96222973e-01 -3.38134140e-01 -5.26250184e-01 4.94352043e-01
-1.61846593e-01 -8.44747484e-01 1.96833257e-02 2.66123295e-01
-1.75772160e-02 -4.96520251e-01 -2.59962559e-01 1.32221937e-01
-1.03591405e-01 -7.45250285e-01 3.09689969e-01 4.68744099e-01
2.26803899e-01 -1.12256694e+00 1.07279670e+00 4.44260269e-01
-7.20066667e-01 -5.07914603e-01 -1.08808719e-01 -1.36127010e-01
-2.50971504e-02 7.95270503e-01 -1.00618744e+00 9.02510762e-01
5.01256585e-01 2.71517131e-02 -6.21440709e-01 1.04234767e+00
1.87220275e-02 1.00729728e+00 -5.31327367e-01 -4.63655919e-01
7.76327178e-02 -2.35481054e-01 3.66793573e-01 1.30598629e+00
3.01195949e-01 -2.33515799e-01 -5.86716346e-02 3.00699681e-01
9.34924960e-01 6.72860324e-01 -5.70964217e-01 3.36532444e-01
6.08294547e-01 1.21038926e+00 -7.44182467e-01 -6.72244653e-02
-2.50284880e-01 2.06815124e-01 -1.40039742e-01 -8.41221586e-03
-8.11563551e-01 -6.22712851e-01 1.88549042e-01 6.70648277e-01
1.49495006e-01 3.52585968e-03 -3.68052036e-01 -5.28584182e-01
-1.40236095e-01 -9.25377131e-01 7.14242160e-01 4.08019312e-02
-7.47373462e-01 9.22097802e-01 5.28612316e-01 -1.14817488e+00
-4.92164761e-01 -8.96975398e-01 -6.23926818e-01 8.21756601e-01
-7.91547656e-01 -7.63302207e-01 1.28423393e-01 5.70227027e-01
7.46781453e-02 -6.91162765e-01 1.19371414e+00 6.01282656e-01
-7.41819441e-01 2.52523690e-01 -5.76897003e-02 9.11424607e-02
3.25527221e-01 -8.87843251e-01 -2.79817522e-01 7.57404923e-01
2.50295401e-02 4.28480059e-01 8.27984929e-01 -5.60221612e-01
-6.71514213e-01 -3.27429831e-01 1.00873828e+00 6.12908117e-02
7.04375863e-01 -1.94973171e-01 -7.06772983e-01 1.09205261e-01
-2.04910450e-02 -2.98024088e-01 7.77786136e-01 -1.29796518e-02
-2.09630787e-01 -3.21404070e-01 -1.46656764e+00 2.57513434e-01
1.77415207e-01 -2.48312742e-01 -4.62188929e-01 1.85984626e-01
-9.55283046e-02 3.41317803e-01 -1.01894426e+00 1.99600071e-01
3.38814527e-01 -1.46896887e+00 9.93442833e-01 -5.99903584e-01
-2.19454274e-01 -9.70583260e-02 1.57843426e-01 -7.18759358e-01
2.41784174e-02 -4.91759151e-01 3.58868316e-02 1.29380667e+00
5.02791226e-01 -9.27375317e-01 6.96594894e-01 3.64335775e-01
4.32830572e-01 -7.42934525e-01 -9.79346573e-01 -5.16876996e-01
-3.51132214e-01 -4.21559006e-01 5.05736768e-01 7.72833169e-01
2.35289484e-02 -3.68392877e-02 -1.30481139e-01 5.69414459e-02
5.91088474e-01 -1.19945891e-01 8.52277219e-01 -1.51179469e+00
-4.12130803e-01 -3.83522868e-01 -9.03920889e-01 1.59804001e-01
-1.65752582e-02 -3.14206719e-01 -8.86416078e-01 -9.85286713e-01
-3.83128136e-01 -6.35724306e-01 -3.53146046e-01 2.33805269e-01
2.09099352e-01 2.44618226e-02 1.52426332e-01 -6.17966503e-02
7.92323202e-02 -3.85082811e-01 5.67407727e-01 3.42666090e-01
-1.80932119e-01 7.75693357e-01 -2.41343319e-01 6.01804316e-01
1.09847796e+00 -5.56411326e-01 -3.47407490e-01 3.84839594e-01
-4.64032218e-02 1.55741751e-01 9.22771469e-02 -1.18684745e+00
3.04393806e-02 -1.71902314e-01 3.80471110e-01 -6.63575470e-01
2.64617294e-01 -1.60262227e+00 4.59770352e-01 9.92645025e-01
-7.35570937e-02 1.40706450e-01 6.39159530e-02 2.52279520e-01
-1.58518836e-01 -9.39370811e-01 7.93871462e-01 8.21191818e-02
-5.39698660e-01 -1.73510402e-01 -7.87630200e-01 -3.72801304e-01
1.55903661e+00 -5.73203385e-01 5.57297133e-02 -2.75362521e-01
-9.86772418e-01 -5.95228434e-01 8.89851376e-02 9.18292031e-02
1.82600468e-01 -7.04260230e-01 -4.86809671e-01 3.57926160e-01
-1.60458848e-01 -8.60173643e-01 3.13028276e-01 6.64609075e-01
-8.99469018e-01 6.16953075e-01 -6.11785650e-01 -1.20127924e-01
-1.61126828e+00 9.52918947e-01 1.09068073e-01 -6.23697340e-01
-2.70403445e-01 -5.79908304e-03 -1.00864410e+00 -5.44765890e-02
4.67452258e-01 2.57270951e-02 -8.90891671e-01 2.73028255e-01
3.24239969e-01 8.62894416e-01 4.29475456e-01 -6.66542292e-01
-4.08879668e-01 4.75560784e-01 4.28474583e-02 2.65482105e-02
1.31597197e+00 3.19478750e-01 -3.44013870e-01 1.95962593e-01
1.05449522e+00 3.19462478e-01 -4.74462152e-01 1.58070832e-01
4.68436867e-01 -4.64047462e-01 -2.02631056e-01 -6.85607910e-01
-7.14121938e-01 8.56814384e-01 7.23391414e-01 7.96120584e-01
1.33458447e+00 -8.19772780e-01 6.75633028e-02 3.77294868e-01
6.24006569e-01 -1.17092788e+00 -3.81086737e-01 2.22151399e-01
2.68836409e-01 -8.43441665e-01 3.55805993e-01 -6.26174808e-01
-4.51833844e-01 1.73299873e+00 3.70529383e-01 -2.33656704e-01
1.18555701e+00 3.42757463e-01 -1.71834335e-01 -1.40598953e-01
-5.10761440e-01 -8.42796490e-02 -1.98679775e-01 7.63348818e-01
1.48890868e-01 1.63450360e-01 -1.01009035e+00 4.49240178e-01
-3.36447097e-02 1.55000258e-02 6.47599697e-01 8.98281991e-01
-6.22013152e-01 -1.44685376e+00 -8.21133435e-01 5.21320343e-01
-7.13227451e-01 2.88344443e-01 -3.58854532e-01 1.25564265e+00
3.90045494e-01 1.27629435e+00 -1.74554661e-01 -4.24087405e-01
5.49595177e-01 1.22726679e-01 1.43622965e-01 -2.41135299e-01
-1.02949798e+00 -3.72336030e-01 3.00067514e-01 -3.07548523e-01
-1.44827172e-01 -6.77988291e-01 -8.47589433e-01 -4.56125468e-01
-4.36849684e-01 8.46764505e-01 1.14000428e+00 8.87697220e-01
-1.40941858e-01 3.15844715e-01 9.52092648e-01 -4.78203654e-01
-6.03467166e-01 -9.65727508e-01 -6.04771554e-01 3.85534495e-01
-3.42827767e-01 -7.20443070e-01 -6.72558308e-01 -3.08646679e-01] | [5.266815185546875, 7.188900470733643] |
4b565488-cf68-48a7-baf7-5b55fec50735 | style-transfer-for-co-speech-gesture | 2007.12553 | null | https://arxiv.org/abs/2007.12553v1 | https://arxiv.org/pdf/2007.12553v1.pdf | Style Transfer for Co-Speech Gesture Animation: A Multi-Speaker Conditional-Mixture Approach | How can we teach robots or virtual assistants to gesture naturally? Can we go further and adapt the gesturing style to follow a specific speaker? Gestures that are naturally timed with corresponding speech during human communication are called co-speech gestures. A key challenge, called gesture style transfer, is to learn a model that generates these gestures for a speaking agent 'A' in the gesturing style of a target speaker 'B'. A secondary goal is to simultaneously learn to generate co-speech gestures for multiple speakers while remembering what is unique about each speaker. We call this challenge style preservation. In this paper, we propose a new model, named Mix-StAGE, which trains a single model for multiple speakers while learning unique style embeddings for each speaker's gestures in an end-to-end manner. A novelty of Mix-StAGE is to learn a mixture of generative models which allows for conditioning on the unique gesture style of each speaker. As Mix-StAGE disentangles style and content of gestures, gesturing styles for the same input speech can be altered by simply switching the style embeddings. Mix-StAGE also allows for style preservation when learning simultaneously from multiple speakers. We also introduce a new dataset, Pose-Audio-Transcript-Style (PATS), designed to study gesture generation and style transfer. Our proposed Mix-StAGE model significantly outperforms the previous state-of-the-art approach for gesture generation and provides a path towards performing gesture style transfer across multiple speakers. Link to code, data, and videos: http://chahuja.com/mix-stage | ['Louis-Philippe Morency', 'Chaitanya Ahuja', 'Dong Won Lee', 'Yukiko I. Nakano'] | 2020-07-24 | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/2962_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123630239.pdf | eccv-2020-8 | ['gesture-generation'] | ['robots'] | [ 0.403063 0.09475458 0.1459082 -0.7149108 -0.56050396 -0.8172905
1.0399344 -0.6478888 -0.5096812 0.20246916 0.5276597 -0.04266408
0.33173016 -0.4553102 -0.7777997 -0.7620715 0.04080038 0.81620216
-0.11482365 -0.4375054 -0.12254949 0.62827486 -1.3278494 0.45924202
0.1486134 0.39348692 0.4576045 1.220036 -0.4466239 0.31532606
-0.71913403 -0.08821647 0.22348988 -0.75359696 -0.96080256 0.03748642
0.44960666 -0.63133985 -0.3501452 0.63397473 0.6025311 0.42397252
0.8077421 -1.3968871 -0.6736064 0.9233821 -0.18181686 -0.31569138
0.5583522 0.41851765 0.83912504 -0.63768756 0.7692397 1.7952675
0.32006866 1.5023687 -1.2994049 -0.97303486 0.35010022 -0.1966103
-1.113175 -0.430434 0.82411975 -0.39808783 0.7178246 0.4205307
0.7391771 1.9775207 -0.23521972 1.136935 1.0014935 -0.328834
0.16824211 -0.15254453 -0.08315422 0.45953336 -0.66430914 0.39959496
-0.92686546 0.09403566 1.0223448 0.03115596 -0.183781 -0.27451023
-1.5962921 0.5488342 0.22059603 0.31496912 -0.09373711 0.4062894
0.2405307 0.7864667 -0.0751561 0.38988572 -0.51689714 -0.54078907
-0.56578946 0.4980616 1.067103 1.3805822 0.56227094 0.07238423
-0.55438966 0.65422297 0.55343837 0.7973176 0.55917585 -0.8423403
0.36431944 0.27463835 -0.01396263 -0.22448215 -0.25204262 0.14544746
-0.74531746 0.47887364 0.5757572 -0.38416702 -1.2313393 2.2791257
0.4108556 0.34127313 0.20726827 0.87229407 0.89581645 0.75113726
0.10698485 0.18932486 1.2891821 -0.98508376 -0.5072586 -0.11938276
0.30041605 -0.8695545 1.6677088 0.19476555 -0.75581086 -0.60096025
-0.7037306 -0.11379112 -0.18481192 -0.07449462 0.46317434 0.5513112
-1.0267979 0.50317913 -1.2443451 -0.7500315 0.19212581 0.45441097
-0.37308106 0.3349252 -0.7142868 0.7098056 -0.04408816 -0.26988357
-1.1756089 -0.73586595 -0.8953474 -0.13557068 0.09963101 -0.8430832
1.7751709 -1.0685334 -2.4669316 0.76410896 -0.27135962 -0.04641346
0.7909729 -0.38640133 -0.2683283 -0.12193626 -0.14911747 1.1096562
1.1811562 -1.3408962 -0.57334524 -0.3383632 -0.0183555 0.4051694
-0.17586955 0.14900555 -0.3944038 -0.95077145 -0.15399443 -1.345851
0.11506648 0.22253323 -0.30172127 -0.34315273 1.0387595 -0.2593476
0.6994401 -2.2350452 0.86603385 0.05439521 -0.09126709 0.14508992
-0.680521 0.49112052 0.214385 -0.04179312 -0.32588926 -0.88219625
0.2969951 0.6117086 -0.3562708 0.07336082 0.24101062 0.8637823
-1.0743883 -0.06686579 0.22287206 0.7723843 -0.74484605 0.798281
-0.3430183 1.1984832 -0.31804913 0.31874114 0.28290367 0.18548961
0.17718601 0.11217111 -0.01187341 0.35000935 -1.2472336 2.2155209
-0.7867459 0.46788806 0.3632446 -0.35325783 0.9250633 0.4842434
0.10345861 -0.21664968 0.21422844 0.033183 0.09799279 -0.59491354
0.19723125 -0.48188153 -0.34592336 0.84780204 0.38796458 -0.49921274
-0.10254307 0.04642618 1.0208468 0.50976247 0.04031667 0.07033146
0.16228242 -0.5580452 0.23260649 0.7263613 -0.09618123 0.5821533
0.13380995 -0.3325128 -0.73986626 -1.3967679 0.42467466 1.8711288
-0.06480386 -0.08102772 -0.74283224 -0.8268735 -0.01284811 0.7610105
-0.83797276 0.0538009 -0.9412203 -0.05350506 0.97927487 0.6153629
0.33658153 -1.6362892 -0.6709066 0.15094315 -0.09362618 -0.8927787
-1.1840951 0.19606353 -0.62886274 -0.6782247 -1.0030044 -0.866214
0.4744667 -0.17102925 0.83649015 -0.12652712 0.0616276 0.7009234
-0.5568299 -0.49235144 -0.9728562 0.3545305 0.22219324 0.23005226
0.28725523 -0.86213875 -0.18724519 0.3290457 -1.1522163 0.08678863
0.68068016 0.8811156 0.31371075 -0.87029725 0.16913846 -0.6897708
0.56294173 -0.14550515 -0.12940048 0.1795117 0.03498469 0.5438439
0.41374618 -1.0718048 -1.1611762 0.44684666 -0.33422095 -0.4705859
-0.75969285 -0.10076876 -0.50744736 0.260881 0.54426646 0.25115293
0.335143 -0.66101897 0.81917775 0.63856584 0.7402293 -0.87031615
1.0948904 0.34809378 -0.24902503 -1.1166929 -0.1978846 -0.35770005
-0.90221626 -0.16118434 0.8883195 -0.596104 -0.7799269 0.91307247
-1.2874273 -1.1352565 -0.29567456 0.3256819 -0.7662299 0.05452151
-0.42923322 -0.6131348 -0.18993273 -1.2147653 1.392733 0.28603944
-0.89713115 -0.9514087 0.2338723 -0.12379122 0.4590891 0.16327786
0.526758 -0.61371887 -0.2574564 0.35111123 0.33247107 0.08751323
0.6288203 -0.04961639 -0.98359907 -0.35834843 -0.27373672 -0.2756757
0.6224021 0.00652635 0.62039834 -0.60824144 -0.17645465 0.93641454
0.5949613 0.37443814 0.3069664 0.06938308 1.0120535 0.6091754
0.1943423 0.15676057 0.4202633 0.9448368 0.01953877 0.13641581
-0.38660842 -0.56117606 0.84273064 0.7055465 -0.11145492 -0.10766128
-0.62314063 0.3999653 -1.555187 -1.0091549 0.35667363 1.8411961
1.1401392 -0.25377315 0.5862866 -0.08491443 0.46372405 0.21309724
-0.60287017 -0.59691036 0.18370213 0.5252298 0.02753195 0.891945
-1.1092067 1.2904958 5.485134 0.36601958 -1.5510167 0.05978845
-0.05901691 -0.42883977 -0.40724286 -0.31495824 -0.78158456 0.3040387
0.7006637 0.12572946 0.8578974 0.66688806 0.12610106 0.6078307
-1.7980696 0.92105055 0.15906113 -0.9718603 0.4706315 -0.18044367
0.4674708 -0.01781925 0.225052 0.5433958 0.6880199 -1.0146449
0.98730814 0.3293727 1.0389311 -0.39011422 0.11344945 0.21383083
-1.2484624 0.1801482 0.38551846 -0.07499813 0.30890042 -0.47563124
-1.0966012 0.1418902 0.61382365 0.37850708 -0.20672432 0.41260013
-0.5575453 0.493399 -0.43747225 -0.31118196 0.29140162 0.11275997
0.9256275 1.5343982 0.34897965 0.1751497 0.073379 0.9441268
-0.05713125 -0.24299666 -0.621594 -0.09187423 0.6535914 1.0100352
-0.4143769 -0.3417817 -0.1831223 1.5205946 0.08887701 0.53394455
-0.37319443 -0.23876908 1.1510173 -0.19742838 0.16320935 -0.4487647
-0.2792481 -1.0174967 -0.19152816 -1.1515864 0.0812655 -0.41395277
-1.2491724 0.757483 0.16283002 -1.1749967 -0.6726726 -0.638188
-0.81841546 0.79552484 -0.97918254 -1.7141144 -0.3883686 0.85864526
0.9170965 -0.39939892 1.1610496 -0.07832297 -0.10851292 1.0046905
-0.30903724 0.3346432 0.93146265 -1.4717057 0.749328 0.5399806
0.5344372 0.78706515 0.7980214 -0.38820323 -1.4513104 -1.031826
0.7979997 -0.69314915 0.47538218 -0.73257494 -0.8165104 1.0567858
0.23688293 -0.376839 0.66766226 0.02607916 -0.59334636 0.0694399
-0.9933345 0.9520518 1.4163944 -0.66725135 -0.83812094 -0.16777161
0.8457058 -0.7097915 -0.4260005 0.00768396 0.9915756 -0.5894232
0.82060987 -0.75164276 0.18939446 -0.2925984 -0.22006276 -1.7404531
-0.04491134 -1.1923362 -0.0075 1.3644013 0.36199027 -0.49041313
0.60264564 0.57156837 -0.29095548 -0.12629506 -0.8515083 -0.8544757
0.38877216 -0.38742045 1.0380285 0.93109286 -0.04559533 0.4171316
-0.32362282 0.27091762 0.40780193 -0.00780825 1.3582838 -1.0356529
-0.7514821 -0.7396679 -0.06436614 -1.3867327 0.34179145 -1.0361093
0.32676315 -1.1107972 -0.17737035 -0.37052837 0.11486638 0.6740074
-0.14717728 -0.00976904 0.58667076 0.09310789 -0.11806931 0.77282286
1.5705231 -0.25098163 -0.7178 0.1975954 -0.52809477 0.72274476
0.76406574 -0.43514663 -0.35118574 -0.7197758 -0.36392337 0.07552115
0.32573193 -0.68173856 0.09795405 -0.3419962 0.13154681 -0.20104125
0.44011396 -0.7683648 0.16968778 0.35127312 -0.7258495 0.05321538
0.30522817 0.38181728 0.06858665 0.22231826 0.6723119 -0.02868037
-0.6442776 0.29506794 -0.56399846 -0.11953878 0.752628 0.12318861
0.13268372 -0.54166496 -1.0404476 0.16084214 0.29839563 1.0519155
0.49669838 -1.5527805 -0.85473573 0.5558347 0.22441778 0.20815906
-0.04396865 0.2536394 -0.2615042 -0.10560782 -0.45119682 -0.7382768
-1.5637634 0.07464971 0.18649693 0.08512759 -0.6295817 1.1931528
0.2970507 -1.0033385 0.468836 -0.7990362 0.11741757 -0.09809436
0.72883594 0.20029037 -0.4277616 -0.8423742 -0.34145492 0.57468283
-0.04604155 -0.83688146 1.390104 -0.05201072 0.20327802 0.85549796
1.1715262 -0.05831567 -1.6065056 -0.26004902 -0.2338583 -0.20938626
-0.55094963 -0.92288697 -0.896043 0.984149 0.5996051 -0.3236312
0.78019154 0.19174294 0.742896 0.56293523 0.34761295 -0.7667527
0.5532807 0.9086523 1.42963 -1.062017 -0.8828042 0.11167636
-0.8797792 1.0243541 0.59363717 -0.10539044 0.61397153 0.55994415
0.6785521 0.09340738 -0.46125537 -0.00686077 0.3371684 1.0069286
0.5370511 0.52511275 0.28984743 0.59929746 -0.8148726 -0.1374001
0.18585877 0.8798793 0.0086124 -1.6369038 -0.5041252 -0.05275882
0.02077629 0.06261687 -0.80347556 0.7326609 -0.0389428 0.98678607
0.16576059 -0.6758374 0.42494965 0.5268151 0.81750906 -0.9906492
-0.7567399 0.10242327 -0.18379083 -0.5998101 -0.44801378 -0.97708875
-1.438616 -0.3328363 0.3292084 -0.20183372 0.48874956 1.0599892
0.13816933 0.5750977 0.50643057 -1.5758631 -0.6489879 -1.2836101
-0.3964845 0.69583714 0.71160185 -0.5502902 -0.25917855 0.41296774] | [5.6192216873168945, -0.11828003823757172] |
4aca96d7-517a-45a4-b248-29940a4844c3 | rankings-dependent-preferences-a-real-goods | 2305.03644 | null | https://arxiv.org/abs/2305.03644v1 | https://arxiv.org/pdf/2305.03644v1.pdf | Rankings-Dependent Preferences: A Real Goods Matching Experiment | We investigate whether preferences for objects received via a matching mechanism are influenced by how highly agents rank them in their reported rank order list. We hypothesize that all else equal, agents receive greater utility for the same object when they rank it higher. The addition of rankings-dependent utility implies that it may not be a dominant strategy to submit truthful preferences to a strategyproof mechanism, and that non-strategyproof mechanisms that give more agents objects they report as higher ranked may increase market welfare. We test these hypotheses with a matching experiment in a strategyproof mechanism, the random serial dictatorship, and a non-strategyproof mechanism, the Boston mechanism. A novel feature of our experimental design is that the objects allocated in the matching markets are real goods, which allows us to directly measure rankings-dependence by eliciting values for goods both inside and outside of the mechanism. Our experimental results confirm that the elicited differences in values do decrease for lower-ranked goods. We find no differences between the two mechanisms for the rates of truth-telling and the final welfare. | ['Peter Troyan', 'Andrew Kloosterman'] | 2023-05-05 | null | null | null | null | ['experimental-design'] | ['methodology'] | [-2.87566006e-01 2.51919806e-01 -7.58035839e-01 -4.21194375e-01
-4.94362116e-01 -1.18556702e+00 4.27726239e-01 9.04246699e-03
-9.44784939e-01 1.07145858e+00 5.27133942e-01 -4.12936687e-01
-7.47767150e-01 -8.95933509e-01 -4.10876423e-01 -3.92228723e-01
-9.40343738e-02 6.18554413e-01 8.40175897e-02 -3.20521183e-02
3.08748752e-01 1.50591031e-01 -1.34575045e+00 6.21897817e-01
4.85058844e-01 8.21678579e-01 2.14314297e-01 5.29146552e-01
2.60207862e-01 1.05300987e+00 -4.66119051e-01 -6.54543042e-01
9.87926781e-01 -4.70581561e-01 -8.18120897e-01 1.41396269e-01
-7.32842414e-03 -9.22167003e-01 -2.51398861e-01 1.01038289e+00
1.22217342e-01 2.20214780e-02 7.18506217e-01 -1.40086567e+00
-6.98452532e-01 1.23445868e+00 -4.61914301e-01 6.27273917e-02
4.74703938e-01 5.85831739e-02 1.66438997e+00 -1.61281213e-01
8.03909183e-01 1.22227216e+00 6.25331849e-02 3.97398263e-01
-1.46114111e+00 -4.39799130e-01 1.01491399e-01 -6.13911226e-02
-8.43037665e-01 -4.83353049e-01 6.32884443e-01 -5.01340985e-01
2.59016305e-01 7.83069611e-01 4.35850143e-01 4.55954999e-01
1.22328132e-01 4.15192336e-01 1.82399452e+00 -1.15796283e-01
3.66086721e-01 3.99830014e-01 6.53612316e-02 -6.52316958e-02
1.29926455e+00 4.20542836e-01 -5.36963642e-01 -4.72751528e-01
7.61694610e-01 -3.61194685e-02 -3.80175561e-01 -6.23375535e-01
-1.13913834e+00 7.02810645e-01 3.48521888e-01 1.05575293e-01
-8.35752368e-01 2.62757599e-01 3.74143105e-03 1.03741360e+00
4.64177988e-02 1.10965693e+00 -5.58749497e-01 1.00283086e-01
-4.06796098e-01 8.39469671e-01 8.03964794e-01 7.89394677e-01
7.34148860e-01 -4.85584944e-01 -3.43348980e-01 4.20768023e-01
3.19955021e-01 4.15464997e-01 2.06847891e-01 -1.69143629e+00
6.84023559e-01 5.25858462e-01 1.22129881e+00 -9.45700347e-01
-4.49332952e-01 -3.24791968e-01 5.81962243e-02 4.17378426e-01
8.72201681e-01 -5.26753545e-01 -9.32084396e-02 2.11131740e+00
2.44898908e-02 -9.93429422e-01 5.64474724e-02 1.30216384e+00
2.11946249e-01 6.22597992e-01 -1.97841227e-01 -6.46443129e-01
1.35037792e+00 -4.26843047e-01 -4.64094609e-01 1.06407344e-01
6.13788903e-01 -6.04097724e-01 1.07980299e+00 1.69813797e-01
-1.41909277e+00 1.87535748e-01 -5.71443141e-01 4.02319402e-01
1.33326128e-01 -6.17016017e-01 1.02022874e+00 4.60442275e-01
-7.75696397e-01 7.14712501e-01 -2.51382887e-01 -1.02909163e-01
8.93261060e-02 3.07027310e-01 5.05347103e-02 2.20179066e-01
-1.01788044e+00 9.61595416e-01 -2.13057548e-01 -4.12324011e-01
-7.59407103e-01 -4.93221521e-01 -3.08455229e-01 5.70132852e-01
6.30163610e-01 -7.23156154e-01 1.61579573e+00 -1.30924857e+00
-1.27986968e+00 7.74515450e-01 1.72830686e-01 -8.56390074e-02
9.16452348e-01 2.46115834e-01 1.62113726e-01 -6.87308311e-02
7.58509934e-01 2.68072098e-01 2.39495377e-04 -1.60377824e+00
-9.02126551e-01 -4.53870028e-01 5.76400876e-01 5.86074352e-01
-6.36105165e-02 2.96076626e-01 6.23157382e-01 -2.02003628e-01
2.21223965e-01 -8.00365746e-01 -1.38397992e-01 -4.12327379e-01
-5.02245665e-01 -3.06207716e-01 -3.91019046e-01 1.17583580e-01
7.22384572e-01 -1.77800453e+00 -4.41470146e-01 4.74114865e-01
1.80085465e-01 -8.76414478e-01 -3.04998130e-01 6.63446248e-01
5.73792867e-02 2.76226968e-01 3.51391464e-01 1.02608338e-01
8.43516946e-01 8.26499835e-02 -2.77130276e-01 7.03308880e-01
-5.75758874e-01 8.16966951e-01 -8.10909212e-01 -1.26720279e-01
-6.70905530e-01 -7.03905582e-01 -7.06171870e-01 5.84313385e-02
9.18191597e-02 -2.98198581e-01 -4.50277835e-01 2.98585296e-01
6.64483607e-01 -1.58264905e-01 8.03675950e-01 3.43511134e-01
-5.99738538e-01 1.06439579e+00 -1.32505560e+00 6.07782066e-01
1.82349190e-01 2.78437346e-01 3.24232250e-01 -5.65249026e-01
1.55033663e-01 3.18659186e-01 2.00463265e-01 -8.94036114e-01
3.43408436e-01 4.23682958e-01 5.97949624e-01 -3.14434856e-01
8.01605344e-01 -3.47476035e-01 -2.61018783e-01 1.17090917e+00
-6.44184768e-01 -1.99188422e-02 3.68903458e-01 4.23211336e-01
9.47531641e-01 -3.54111731e-01 4.62791622e-02 -7.68726885e-01
-2.77697593e-01 4.42405134e-01 9.84509408e-01 1.33255970e+00
4.02342640e-02 1.50383592e-01 1.08021164e+00 -1.27267584e-01
-1.07666326e+00 -1.13369131e+00 1.00937989e-02 1.24858212e+00
5.34141541e-01 1.91869125e-01 -3.68022859e-01 -4.44107175e-01
6.83169484e-01 8.52085769e-01 -7.15219975e-01 1.28574744e-01
5.95017821e-02 -4.90546584e-01 -2.62120180e-02 4.92147505e-01
4.44433063e-01 -7.98749745e-01 -9.62299049e-01 -1.81723937e-01
-3.29359204e-01 -3.63234282e-01 -7.82600582e-01 -1.80300012e-01
-8.42852473e-01 -1.16156030e+00 -6.38686776e-01 -4.57361937e-01
8.67729723e-01 7.47350574e-01 8.61739337e-01 2.21006230e-01
6.88036442e-01 4.63630497e-01 -3.15468788e-01 -3.63734037e-01
-2.57266909e-01 -2.50219673e-01 3.35683614e-01 -1.01820074e-01
2.94623494e-01 -1.90625280e-01 -8.83182347e-01 6.21401787e-01
-6.60075009e-01 -1.82977065e-01 4.03441668e-01 3.87226909e-01
-2.14233041e-01 1.03864953e-01 7.53007531e-01 -9.02724802e-01
1.01464522e+00 -6.08292282e-01 -1.26424253e+00 1.57203943e-01
-7.47400582e-01 -1.18839704e-01 4.70118791e-01 -4.34793621e-01
-1.13065064e+00 -4.58766341e-01 7.99992025e-01 6.43798947e-01
1.41499490e-01 6.51352406e-01 -2.14710474e-01 3.01604301e-01
6.26653850e-01 -2.20796034e-01 4.02426980e-02 -5.12267232e-01
7.36655965e-02 6.14499569e-01 1.61324844e-01 -7.39061534e-01
1.02415502e+00 5.82010567e-01 -1.37120737e-02 -1.28671616e-01
-4.03096884e-01 -1.79893821e-01 3.32810223e-01 -1.96233261e-02
4.66296852e-01 -7.76598334e-01 -1.64596772e+00 2.16407329e-01
-1.07181621e+00 -4.03481483e-01 -3.53451163e-01 9.98206913e-01
-7.92919874e-01 1.78280249e-01 -9.38606381e-01 -1.23468387e+00
4.17800874e-01 -9.83228505e-01 5.01699038e-02 1.42753765e-01
-4.45857018e-01 -3.43565106e-01 -1.72449082e-01 4.98775095e-01
5.70894241e-01 -4.28699046e-01 8.38777661e-01 -6.48117125e-01
-1.09332514e+00 -7.87898526e-02 -1.84861407e-01 -4.49592113e-01
1.63185298e-01 -3.55844349e-01 -5.16615629e-01 -3.58796656e-01
1.47843063e-02 -4.15815443e-01 5.82068920e-01 6.59337997e-01
1.80714920e-01 -1.00719810e+00 -2.20142916e-01 -1.62252501e-01
1.24523306e+00 5.46607316e-01 3.89790207e-01 6.12219155e-01
-2.61371404e-01 1.03172922e+00 5.43330848e-01 6.06606662e-01
7.33838499e-01 8.78235102e-01 2.48728663e-01 5.50910756e-02
5.14951348e-01 -4.87749308e-01 4.60385591e-01 6.93171844e-02
-8.59928578e-02 -3.51755768e-01 -1.50461435e-01 7.30387807e-01
-1.87080777e+00 -1.36704624e+00 -7.68058226e-02 2.50997901e+00
1.04047251e+00 2.36366868e-01 7.65252113e-01 1.24386072e-01
6.01520538e-01 -3.20385545e-01 -3.08115989e-01 -3.41505527e-01
-1.69119090e-01 -6.28895700e-01 7.94444680e-01 8.76434267e-01
-2.23682493e-01 4.81148094e-01 7.30570650e+00 -6.35879710e-02
-4.57820684e-01 -5.65112941e-02 7.59751022e-01 -4.93240237e-01
-9.73027766e-01 3.15120190e-01 -3.52381080e-01 4.63940054e-01
4.00799125e-01 -8.27353597e-01 5.28160572e-01 7.64559388e-01
6.31113768e-01 -6.56258464e-01 -1.37355745e+00 2.39337891e-01
-5.08831441e-01 -1.31765699e+00 -1.87142670e-01 4.75160360e-01
9.41362381e-01 -5.25480866e-01 1.13376051e-01 -4.27770913e-02
1.13663268e+00 -6.62863493e-01 1.33601785e+00 1.79987893e-01
3.82422864e-01 -5.86699724e-01 7.88216650e-01 5.06144464e-01
-3.47983211e-01 -3.73017073e-01 -4.92041498e-01 -9.59679246e-01
8.37148055e-02 3.27649176e-01 -6.76355660e-01 1.29453078e-01
4.36364055e-01 -3.09597224e-01 2.28490122e-02 1.20217252e+00
-1.80327535e-01 5.29802263e-01 -8.78490228e-03 -2.48485297e-01
2.17696317e-04 -3.25574845e-01 2.56396830e-01 3.73905122e-01
3.95911150e-02 6.91787660e-01 7.07666948e-02 1.25309157e+00
-4.36096549e-01 -3.27273905e-02 -5.43359816e-01 -8.99619423e-03
7.50945330e-01 8.31744254e-01 -9.62686121e-01 -3.82518291e-01
-3.97309840e-01 4.88730639e-01 -2.59730607e-01 5.31783462e-01
-4.70328748e-01 -1.55118063e-01 8.24958622e-01 3.42230141e-01
-2.09996298e-01 -2.06190702e-02 -7.85021842e-01 -1.20286715e+00
-8.73936862e-02 -5.45421839e-01 5.47648311e-01 -9.24224138e-01
-1.24255967e+00 -2.06302226e-01 3.30790013e-01 -7.05380440e-01
-1.31439716e-01 -2.99920261e-01 -3.30184907e-01 8.09514701e-01
-1.44487786e+00 -1.12790294e-01 2.88562477e-01 1.61033481e-01
2.64182508e-01 2.30873078e-01 2.77051628e-01 -1.12473272e-01
-1.76406428e-02 4.55678344e-01 1.33876950e-01 9.64364633e-02
4.84550506e-01 -1.29385900e+00 -2.34565541e-01 6.40632808e-01
-1.58665001e-01 9.10701096e-01 8.00761700e-01 -6.11102819e-01
-1.30797195e+00 -3.81217301e-01 1.14114344e+00 -4.41901684e-01
4.43682045e-01 -2.33915851e-01 -2.98692703e-01 7.35872746e-01
3.47936213e-01 -9.34917986e-01 6.70177639e-01 3.14269960e-01
-2.38917768e-01 -3.62831712e-01 -1.47061610e+00 9.38488662e-01
1.19289851e+00 -2.55786717e-01 -8.55358839e-01 2.52636999e-01
6.69647098e-01 2.15608016e-01 -4.51770514e-01 -2.50094458e-02
6.79478109e-01 -1.14421046e+00 5.48196614e-01 -7.77895331e-01
6.44663334e-01 -4.32130754e-01 -3.45250309e-01 -1.44388556e+00
-1.01572132e+00 -5.67238748e-01 1.06291735e+00 1.01433098e+00
8.05688858e-01 -1.10026598e+00 9.60487664e-01 1.16498363e+00
2.15761691e-01 2.98156161e-02 -7.32338071e-01 -9.64679897e-01
3.48896712e-01 1.56988606e-01 8.94260049e-01 1.11939514e+00
6.61635876e-01 1.94171458e-01 -1.04145765e-01 6.27619773e-03
7.41214216e-01 7.01988876e-01 6.27445042e-01 -1.28066456e+00
-4.76911813e-01 -6.78003609e-01 1.07516788e-01 -1.07920349e+00
-1.70387447e-01 -7.42275596e-01 2.75619440e-02 -1.66738331e+00
7.11420596e-01 -7.06667304e-01 -1.04950897e-01 3.87056738e-01
2.99859673e-01 7.25799659e-03 5.37085474e-01 4.07187313e-01
-7.40401030e-01 7.60143772e-02 1.12104487e+00 -1.31178172e-02
-3.19669813e-01 3.59588489e-02 -1.54837656e+00 5.03737390e-01
6.26206875e-01 -4.72440273e-01 -4.26259786e-01 -5.40588081e-01
8.25925529e-01 4.26877528e-01 2.51654327e-01 4.55448329e-02
1.45998433e-01 -1.09072888e+00 1.79508686e-01 -2.56198019e-01
-8.17982927e-02 -8.10511053e-01 4.54623669e-01 5.52903056e-01
-8.60875189e-01 3.30722213e-01 -4.27374691e-01 2.73137111e-02
4.75238293e-01 -3.36243808e-01 5.63642085e-01 -4.48201239e-01
8.70786086e-02 -3.70816648e-01 -8.98395360e-01 -5.99565543e-02
8.68683994e-01 -2.98139393e-01 -5.47382116e-01 -1.02173841e+00
-1.45318165e-01 2.58964837e-01 1.10498774e+00 8.98763072e-04
2.51703948e-01 -1.43563783e+00 -8.43544066e-01 -2.82058597e-01
1.19897768e-01 -6.76193595e-01 -1.40197605e-01 6.61582053e-01
-6.94100186e-02 4.64950860e-01 -1.59245342e-01 1.61655441e-01
-1.00436449e+00 5.09147823e-01 4.56339389e-01 1.03274226e-01
1.91344082e-01 5.95836997e-01 6.32194817e-01 -1.31594971e-01
-4.64271381e-03 -3.34141046e-01 1.56569719e-01 9.51994509e-02
5.89172006e-01 6.06618702e-01 -3.71063709e-01 -3.91356319e-01
-3.97489786e-01 3.75030600e-02 -5.55968545e-02 -7.43374228e-01
1.36393774e+00 -3.17061365e-01 -3.12663108e-01 3.51297736e-01
4.32620138e-01 4.82167274e-01 -1.00332153e+00 -2.40953006e-02
1.69670537e-01 -1.20429695e+00 -4.12754834e-01 -1.05115449e+00
-8.75939310e-01 -2.30682343e-01 -3.02773476e-01 9.16445315e-01
6.39405131e-01 1.91751514e-02 -3.74952070e-02 4.06936169e-01
9.84678805e-01 -1.21759236e+00 -2.75983326e-02 -1.43423155e-01
8.26326191e-01 -8.11754227e-01 7.10613951e-02 -3.04076821e-01
-7.70130336e-01 3.33272219e-01 2.55811244e-01 -2.74082050e-02
1.24521516e-01 -1.80239812e-01 1.56121254e-01 -2.03124478e-01
-1.05496538e+00 -3.37779075e-01 -1.75596729e-01 4.19481337e-01
4.06805038e-01 5.59112668e-01 -1.31150854e+00 9.84957993e-01
-6.27498567e-01 -1.65537551e-01 1.42084599e+00 6.69447780e-01
-7.11706340e-01 -1.06964839e+00 -7.57983923e-01 7.85818875e-01
-6.08403504e-01 1.28668323e-01 -8.79171193e-01 6.67194903e-01
-3.43204468e-01 1.48221207e+00 3.15135866e-01 3.24589759e-02
6.27010286e-01 -2.56287009e-01 5.35056651e-01 -6.01256907e-01
-4.88013119e-01 9.16460082e-02 3.92622352e-01 -3.89537603e-01
-2.73785055e-01 -8.46597433e-01 -1.00436318e+00 -8.48707438e-01
-5.58723927e-01 6.92997456e-01 6.48082793e-02 5.61962843e-01
1.95001587e-01 -2.24330708e-01 1.05729187e+00 -1.71433672e-01
-1.21161711e+00 -5.82784295e-01 -1.21481276e+00 9.25783098e-01
2.19523430e-01 -7.10448742e-01 -5.41882813e-01 -5.89531302e-01] | [4.266723155975342, 2.968151569366455] |
4150768d-e67f-4300-a4b9-6a7f86bbd12c | tree-gated-deep-mixture-of-experts-for-pose | 1910.09450 | null | https://arxiv.org/abs/1910.09450v1 | https://arxiv.org/pdf/1910.09450v1.pdf | Tree-gated Deep Mixture-of-Experts For Pose-robust Face Alignment | Face alignment consists of aligning a shape model on a face image. It is an active domain in computer vision as it is a preprocessing for a number of face analysis and synthesis applications. Current state-of-the-art methods already perform well on "easy" datasets, with moderate head pose variations, but may not be robust for "in-the-wild" data with poses up to 90{\deg}. In order to increase robustness to an ensemble of factors of variations (e.g. head pose or occlusions), a given layer (e.g. a regressor or an upstream CNN layer) can be replaced by a Mixture of Experts (MoE) layer that uses an ensemble of experts instead of a single one. The weights of this mixture can be learned as gating functions to jointly learn the experts and the corresponding weights. In this paper, we propose to use tree-structured gates which allows a hierarchical weighting of the experts (Tree-MoE). We investigate the use of Tree-MoE layers in different contexts in the frame of face alignment with cascaded regression, firstly for emphasizing relevant, more specialized feature extractors depending of a high-level semantic information such as head pose (Pose-Tree-MoE), and secondly as an overall more robust regression layer. We perform extensive experiments on several challenging face alignment datasets, demonstrating that our approach outperforms the state-of-the-art methods. | ['Kevin Bailly', 'Estephe Arnaud', 'Arnaud Dapogny'] | 2019-10-21 | null | null | null | null | ['robust-face-alignment'] | ['computer-vision'] | [ 1.14901036e-01 1.46274626e-01 1.15089573e-01 -8.94429326e-01
-5.07771015e-01 -2.08592921e-01 6.11020148e-01 -6.28735125e-02
-4.91783261e-01 4.80632991e-01 1.38873219e-01 9.89146680e-02
4.93725799e-02 -5.24450779e-01 -9.42166805e-01 -8.20012689e-01
1.46359012e-01 7.45478988e-01 2.25617528e-01 -4.14999664e-01
1.60329521e-01 9.13053334e-01 -1.64995551e+00 2.49310374e-01
2.61508465e-01 8.54236841e-01 -3.29528041e-02 3.44462514e-01
5.24538755e-02 1.54024407e-01 -5.18936813e-01 -7.06120491e-01
3.14849287e-01 -2.39741445e-01 -5.18342972e-01 2.58656323e-01
9.60036695e-01 -6.52203485e-02 1.32607684e-01 1.02308238e+00
8.31829965e-01 2.72477809e-02 3.91470909e-01 -1.21335566e+00
-4.26644310e-02 3.79672319e-01 -6.62945032e-01 -4.31797914e-02
3.81865889e-01 1.07348576e-01 7.03314483e-01 -1.09056580e+00
5.57446182e-01 1.41089010e+00 6.82749808e-01 8.71984065e-01
-1.40843928e+00 -6.72551572e-01 3.33772361e-01 9.78219062e-02
-1.32363546e+00 -1.03283644e+00 8.38611960e-01 -4.14201289e-01
8.55231822e-01 1.68198451e-01 4.03012216e-01 1.07552505e+00
1.79167002e-01 3.15111279e-01 9.69603539e-01 -4.31044549e-01
2.13561893e-01 2.98493505e-02 -2.63853610e-01 8.45808029e-01
1.09517701e-01 -1.05061844e-01 -6.95385396e-01 -9.46204662e-02
6.61045194e-01 -2.81386793e-01 -1.69397771e-01 -5.65574050e-01
-7.42917597e-01 7.87527502e-01 5.83171487e-01 3.39262545e-01
-3.27405214e-01 1.70299098e-01 2.55340904e-01 1.86068311e-01
3.51053357e-01 3.21179271e-01 -5.20883203e-01 4.89879668e-01
-1.22073042e+00 4.50022340e-01 7.34777808e-01 7.18375742e-01
8.86391342e-01 1.74962372e-01 -3.16671699e-01 9.88757610e-01
5.70414186e-01 9.44841877e-02 1.90567851e-01 -6.07496679e-01
4.73158211e-01 5.04792213e-01 -3.19885649e-02 -6.92768335e-01
-7.37281919e-01 -5.44658303e-01 -7.14808643e-01 4.94083554e-01
5.56835532e-01 -1.17893964e-01 -1.22512650e+00 2.23665047e+00
5.40167630e-01 2.12506771e-01 -4.35821742e-01 7.92803645e-01
7.47142732e-01 1.40423581e-01 1.54431924e-01 -1.91954419e-01
1.52549791e+00 -8.52598786e-01 -5.65170228e-01 -6.39596343e-01
1.99216262e-01 -1.00614560e+00 6.13852084e-01 2.63918877e-01
-1.27122474e+00 -7.32628167e-01 -8.51791620e-01 1.71459597e-02
-3.71913522e-01 3.18138659e-01 2.55280524e-01 7.10110068e-01
-1.54430461e+00 5.90787292e-01 -6.23416185e-01 -4.94524807e-01
5.98953307e-01 1.01444960e+00 -7.51302063e-01 8.13348070e-02
-8.12514484e-01 1.20023584e+00 -2.77634133e-02 4.70386237e-01
-8.02022219e-01 -4.87146974e-01 -9.89206374e-01 3.15422341e-02
3.04842800e-01 -9.65448380e-01 9.98621702e-01 -1.09114015e+00
-1.82322550e+00 1.11205328e+00 -4.66220319e-01 -2.02875197e-01
5.34753144e-01 -3.23004782e-01 -1.90640509e-01 -1.12563170e-01
-1.81070149e-01 8.88096094e-01 1.40863085e+00 -1.23375928e+00
-2.25225583e-01 -8.41436923e-01 1.15465507e-01 -8.95406958e-03
-1.50713280e-01 5.08399963e-01 -6.87994123e-01 -5.83830953e-01
2.53124963e-02 -1.17386436e+00 -2.53375322e-01 9.20578390e-02
-2.06778586e-01 -1.21514075e-01 6.78366125e-01 -8.44565511e-01
1.11735737e+00 -1.85500383e+00 5.80157220e-01 4.28153425e-01
-5.61095923e-02 3.36617351e-01 -1.94534823e-01 3.06604523e-02
-4.22834754e-01 -3.33196461e-01 -3.20100605e-01 -8.30238283e-01
-1.07940368e-01 1.82293743e-01 1.85481712e-01 6.53746903e-01
5.09770930e-01 7.03520298e-01 -4.99870092e-01 -2.76466668e-01
2.84332484e-01 8.52579832e-01 -8.41301680e-01 2.52224684e-01
-9.70887989e-02 8.26880574e-01 -1.47810116e-01 5.61451972e-01
6.93801284e-01 2.38573983e-01 1.61991403e-01 -5.09612381e-01
-6.03780337e-02 1.30489632e-01 -1.26125360e+00 1.68511260e+00
-6.08776033e-01 3.20042461e-01 3.78801525e-01 -9.94115531e-01
1.06343746e+00 5.04013300e-01 3.07764143e-01 -3.77576888e-01
4.70181704e-01 2.43809447e-01 3.75593334e-01 -2.60559350e-01
7.67606422e-02 -7.95158371e-02 1.78689256e-01 2.84692422e-02
6.11894190e-01 -1.05369855e-02 1.15343302e-01 -4.81560111e-01
7.38884211e-01 2.47243986e-01 2.65927374e-01 -4.92216229e-01
9.93064046e-01 -8.36884558e-01 5.74946463e-01 7.46396035e-02
-8.83166939e-02 8.73478770e-01 4.37040746e-01 -5.08381963e-01
-1.04426956e+00 -5.69910109e-01 -2.12034449e-01 1.32745302e+00
-5.04176080e-01 -3.77193868e-01 -1.26132250e+00 -6.00075424e-01
-1.35669887e-01 3.78836036e-01 -7.27801085e-01 -1.29016221e-01
-1.03502870e+00 -1.04507971e+00 3.44496518e-01 6.87919199e-01
2.34949276e-01 -1.13951612e+00 -4.90683466e-01 3.46106231e-01
2.40304768e-01 -1.26912844e+00 -3.48393142e-01 2.95600444e-01
-7.49078691e-01 -8.61913443e-01 -5.77511013e-01 -5.75842977e-01
1.01862764e+00 -3.40723604e-01 1.19084477e+00 1.83088243e-01
-1.50595099e-01 1.17377840e-01 1.96804777e-02 -4.74740654e-01
-9.02094394e-02 6.51290417e-02 1.52266726e-01 5.32732487e-01
2.35147655e-01 -6.91337883e-01 -5.93861520e-01 3.24155837e-01
-6.58125639e-01 -1.88893050e-01 3.94299865e-01 9.11166072e-01
3.17180485e-01 -3.41351509e-01 3.74880135e-01 -8.18200409e-01
2.80399650e-01 -1.05567276e-01 -7.32660294e-01 1.41275749e-01
-4.85711731e-02 1.88435599e-01 5.36625326e-01 -3.22967231e-01
-8.59421909e-01 5.24890780e-01 -6.27788663e-01 -4.55042809e-01
-2.86731094e-01 2.79599540e-02 -7.31448054e-01 -3.89020890e-01
4.50905263e-01 -2.86026418e-01 -1.43428251e-01 -4.35649753e-01
2.34024778e-01 2.71825284e-01 5.75073183e-01 -6.82358503e-01
8.78489017e-01 4.02262747e-01 3.76086652e-01 -7.47809827e-01
-6.07898116e-01 -1.46355480e-01 -1.04054070e+00 -1.69429615e-01
9.32774782e-01 -7.57241547e-01 -5.59331596e-01 7.11162686e-01
-1.27383399e+00 -2.76420325e-01 9.13186669e-02 2.24572688e-01
-5.38644254e-01 -1.48251057e-01 -3.76627058e-01 -5.72723567e-01
-2.88804263e-01 -1.48329306e+00 1.32949769e+00 5.01438797e-01
-3.15135002e-01 -8.85516524e-01 -2.39040583e-01 3.44167888e-01
4.26540375e-01 2.75901526e-01 7.30524302e-01 -6.62445188e-01
-2.18621045e-01 -2.30971083e-01 1.43934816e-01 3.14883947e-01
-6.87286556e-02 2.06903771e-01 -1.47785389e+00 -5.18694997e-01
-1.62582509e-02 -1.36780620e-01 8.03657889e-01 3.65403533e-01
1.22021616e+00 -2.59179205e-01 -1.64503738e-01 7.26784527e-01
1.21551597e+00 -1.13775790e-01 8.59522700e-01 1.76017925e-01
7.33271956e-01 8.76319230e-01 1.71459153e-01 3.37898880e-01
2.82463759e-01 1.20529485e+00 5.38129508e-01 -1.66110381e-01
-1.94427073e-01 7.60352761e-02 2.95027584e-01 3.78021538e-01
-4.49083418e-01 3.18023115e-02 -7.19301224e-01 3.00084740e-01
-1.65742111e+00 -8.32931995e-01 1.73029214e-01 2.47931838e+00
7.46168256e-01 1.38362184e-01 2.59439796e-01 2.60870129e-01
7.95900226e-01 1.74235299e-01 -3.13042641e-01 -5.53284109e-01
-7.21649593e-03 8.05578768e-01 1.11467220e-01 6.04281127e-01
-1.10661161e+00 1.05863118e+00 5.22334433e+00 5.05032122e-01
-1.41805947e+00 1.34680957e-01 5.56711912e-01 1.35830864e-01
3.90138701e-02 -1.35116251e-02 -1.30467665e+00 4.27983403e-01
6.46320283e-01 6.20525539e-01 4.11079764e-01 6.45337164e-01
-1.38822123e-02 6.02541491e-02 -1.47849536e+00 8.31408441e-01
3.03736299e-01 -8.33729625e-01 -6.50254861e-02 -6.25440776e-02
4.65806305e-01 -1.20905943e-01 -1.69655588e-02 1.03748031e-01
-9.94945168e-02 -1.25028324e+00 7.77846575e-01 4.72413301e-01
6.22911930e-01 -6.34851873e-01 8.12083125e-01 1.12179674e-01
-1.31122804e+00 -8.56183469e-02 -2.26396561e-01 1.97761089e-01
1.25282422e-01 3.84832203e-01 -5.97432256e-01 4.75479960e-01
6.92222714e-01 1.70146361e-01 -7.54721940e-01 9.02974069e-01
-3.74067217e-01 1.50940418e-01 -5.55296719e-01 4.04816777e-01
1.61095001e-02 7.90433884e-02 3.75214726e-01 1.29416955e+00
3.46783221e-01 -4.95186634e-02 -7.31310109e-03 5.89037299e-01
-1.72828853e-01 1.10989064e-01 -2.79308438e-01 4.71171021e-01
2.62533605e-01 1.61572707e+00 -6.74059808e-01 -6.49407282e-02
-4.65569168e-01 8.71882856e-01 5.24011075e-01 1.95038795e-01
-7.70610571e-01 4.89719138e-02 6.61395431e-01 4.48947579e-01
5.50965726e-01 -1.03887217e-03 -1.47159651e-01 -8.78964961e-01
7.09771216e-02 -9.76446271e-01 2.39195153e-01 -5.81506312e-01
-1.13216364e+00 9.42928970e-01 3.48928943e-02 -9.23831046e-01
-4.21391577e-01 -8.18652868e-01 -8.57348204e-01 1.00372362e+00
-1.49246490e+00 -1.47071648e+00 -3.17255825e-01 8.10512900e-01
4.25879687e-01 -2.81590760e-01 8.82946253e-01 4.18425053e-01
-7.53358006e-01 8.65663528e-01 -7.09963858e-01 1.92221627e-01
8.06288540e-01 -1.06076181e+00 3.90402168e-01 7.76383519e-01
2.06171289e-01 7.87803471e-01 9.34049129e-01 -2.87302196e-01
-1.16890192e+00 -9.74075437e-01 1.12817633e+00 -4.02700126e-01
1.80817515e-01 -6.72553599e-01 -1.01465726e+00 7.34430313e-01
1.62186459e-01 5.14648557e-01 5.31990588e-01 3.64761837e-02
-4.51542735e-01 -4.06491637e-01 -1.28300893e+00 5.73833466e-01
1.02207303e+00 -4.35866475e-01 -4.82888669e-01 -3.57732326e-02
2.71575321e-02 -6.41451716e-01 -6.04195654e-01 7.95436800e-01
6.76362991e-01 -1.09967482e+00 1.03694594e+00 -5.19419611e-01
4.66737822e-02 -5.11751175e-01 -9.94631276e-02 -1.39678395e+00
-3.35172921e-01 -5.71228683e-01 -1.24701776e-01 1.40450680e+00
2.51588315e-01 -4.16208595e-01 6.81945562e-01 4.87642884e-01
-7.74365962e-02 -8.84508610e-01 -1.10037088e+00 -3.21150601e-01
-9.26468596e-02 -1.98443249e-01 7.61373103e-01 6.53436124e-01
-5.07214725e-01 5.80617428e-01 -2.88994610e-01 3.10060740e-01
3.76487702e-01 -3.03595334e-01 9.64562178e-01 -1.36511672e+00
-1.49419576e-01 -6.49883389e-01 -5.47269344e-01 -5.62173188e-01
4.49499786e-01 -6.73036873e-01 3.24485339e-02 -1.15144205e+00
-1.39644563e-01 -2.43670836e-01 -1.42847642e-01 7.14033782e-01
-2.90555358e-01 4.91871983e-01 2.39205867e-01 -3.83851767e-01
-9.18279681e-03 3.29176486e-01 8.52212071e-01 1.15897529e-01
-6.23891652e-02 2.03308180e-01 -4.07056272e-01 1.05813980e+00
6.59892797e-01 -3.37703228e-01 -5.88918179e-02 -4.89054590e-01
9.63173956e-02 -1.71512082e-01 4.12116230e-01 -1.12528181e+00
3.59125048e-01 2.74090350e-01 7.47252762e-01 -1.44249603e-01
5.69078922e-01 -9.35950100e-01 2.10763276e-01 3.32116961e-01
7.14889541e-02 2.14438200e-01 3.89218569e-01 -8.07292759e-02
-1.72369450e-01 -2.73644030e-01 1.10047865e+00 4.49527763e-02
-3.52952808e-01 4.16422784e-01 1.03061967e-01 -3.05188835e-01
7.59383380e-01 -1.97616994e-01 7.04115406e-02 -2.06463456e-01
-7.84178555e-01 -4.51069586e-02 3.47429782e-01 4.80643898e-01
4.11076456e-01 -1.26012099e+00 -9.45506036e-01 6.00248873e-01
-8.18814151e-03 1.05286330e-01 9.97565538e-02 9.25911665e-01
-2.03952894e-01 1.89700559e-01 -5.43795347e-01 -6.55453861e-01
-1.53672862e+00 3.36922795e-01 5.74038982e-01 -3.12439173e-01
-1.09362982e-01 1.15796995e+00 3.00253868e-01 -4.22744751e-01
3.26715797e-01 -4.95820455e-02 -4.87327307e-01 3.48931462e-01
4.45175529e-01 6.13896623e-02 5.56001604e-01 -1.18743348e+00
-6.33203685e-01 1.10768843e+00 2.15819757e-02 -2.28443854e-02
1.42748225e+00 9.61495042e-02 -2.56464660e-01 -6.86416216e-03
1.00947821e+00 1.04601823e-01 -1.22256780e+00 -4.96241637e-02
1.44200623e-01 -3.22721928e-01 -1.74252972e-01 -5.68307102e-01
-1.56606162e+00 1.11350036e+00 6.18939519e-01 -4.96298790e-01
1.26180005e+00 -6.82331473e-02 3.26267123e-01 3.17702219e-02
4.87411499e-01 -9.45517421e-01 4.13435511e-02 3.79115134e-01
1.27648628e+00 -1.06914222e+00 7.74013028e-02 -4.43808258e-01
-3.24560344e-01 1.25501287e+00 7.95675814e-01 -2.39082828e-01
7.05424488e-01 5.20348668e-01 8.20370018e-02 -1.72187835e-01
-5.44903874e-01 -4.11546826e-01 6.50471807e-01 3.68758112e-01
7.15630651e-01 -1.84030056e-01 -2.42643461e-01 5.87828755e-01
-2.66476661e-01 -8.23412836e-02 -1.69925898e-01 6.95082545e-01
-1.46555573e-01 -1.56029129e+00 -5.87597609e-01 1.11076824e-01
-5.95363140e-01 1.04692534e-01 -2.66864359e-01 8.14495206e-01
5.98624647e-01 5.23132563e-01 9.62606817e-02 -3.26006860e-01
5.33098519e-01 4.06902581e-01 9.89187658e-01 -7.11796701e-01
-1.01424587e+00 3.22096884e-01 -1.94845465e-03 -6.50643110e-01
-5.31868935e-01 -7.38143921e-01 -9.53916550e-01 -8.87748301e-02
-3.13358873e-01 -3.41255724e-01 7.38425553e-01 1.07833195e+00
2.34577451e-02 5.50985873e-01 5.23835719e-01 -1.36791420e+00
-2.91010141e-01 -1.07342672e+00 -1.41476199e-01 4.40085620e-01
2.37850457e-01 -9.39444125e-01 -9.55498442e-02 8.61779153e-02] | [13.509075164794922, 0.33108603954315186] |
3468aa77-266f-4b68-865e-dc742af0015c | advances-of-transformer-based-models-for-news | 2007.05044 | null | https://arxiv.org/abs/2007.05044v2 | https://arxiv.org/pdf/2007.05044v2.pdf | Advances of Transformer-Based Models for News Headline Generation | Pretrained language models based on Transformer architecture are the reason for recent breakthroughs in many areas of NLP, including sentiment analysis, question answering, named entity recognition. Headline generation is a special kind of text summarization task. Models need to have strong natural language understanding that goes beyond the meaning of individual words and sentences and an ability to distinguish essential information to succeed in it. In this paper, we fine-tune two pretrained Transformer-based models (mBART and BertSumAbs) for that task and achieve new state-of-the-art results on the RIA and Lenta datasets of Russian news. BertSumAbs increases ROUGE on average by 2.9 and 2.0 points respectively over previous best score achieved by Phrase-Based Attentional Transformer and CopyNet. | ['Ilya Gusev', 'Alexey Bukhtiyarov'] | 2020-07-09 | null | null | null | null | ['headline-generation'] | ['natural-language-processing'] | [ 2.08673805e-01 2.97211915e-01 -1.74006522e-01 -2.51950711e-01
-1.02659428e+00 -6.63956761e-01 7.17148185e-01 4.74175304e-01
-4.72800583e-01 1.03715897e+00 9.78708684e-01 -2.58196503e-01
-1.24290057e-01 -6.58844829e-01 -6.49842680e-01 -3.25822830e-01
1.39660180e-01 7.80603468e-01 2.70985961e-01 -9.60999668e-01
6.18871629e-01 3.26032102e-01 -1.03803730e+00 7.41216958e-01
1.24539649e+00 7.02836692e-01 1.39869660e-01 7.90788233e-01
-7.16686785e-01 1.13496196e+00 -8.58837605e-01 -6.85864031e-01
-3.55546206e-01 -5.87771654e-01 -1.10499144e+00 -3.06043386e-01
5.07782042e-01 8.32428932e-02 -1.46380126e-01 8.28669429e-01
5.68642080e-01 2.77031779e-01 7.47534573e-01 -5.55092692e-01
-9.41765189e-01 1.29988468e+00 -5.39492488e-01 7.67144859e-01
3.88348877e-01 -3.32138926e-01 1.44705200e+00 -7.29158580e-01
6.85672760e-01 1.22965419e+00 4.85376030e-01 6.78280056e-01
-8.81198406e-01 -3.30167830e-01 2.88140297e-01 3.55356127e-01
-6.60709143e-01 -4.07486439e-01 5.87456584e-01 -1.28231630e-01
1.62612009e+00 3.77796590e-01 4.85231906e-01 1.03205073e+00
5.75831950e-01 1.05254757e+00 6.17274940e-01 -3.96101177e-01
-4.08028364e-02 6.56938404e-02 6.47651613e-01 4.08445626e-01
2.44518921e-01 -6.45931244e-01 -7.18444705e-01 5.87555990e-02
2.58468240e-01 -2.79956073e-01 -3.52322370e-01 1.88209936e-01
-1.32284081e+00 9.94466126e-01 3.02724868e-01 6.47071719e-01
-6.76369429e-01 -1.23261195e-02 6.96764827e-01 4.33677137e-01
7.68388867e-01 1.08843148e+00 -8.77337873e-01 -2.53678143e-01
-9.55442965e-01 2.29428589e-01 8.77411664e-01 8.73005807e-01
2.98847884e-01 2.03866675e-01 -7.68668830e-01 9.36831892e-01
-1.88939586e-01 5.09801149e-01 9.73013043e-01 -3.69871378e-01
7.82296836e-01 7.67178416e-01 2.40733940e-03 -7.59312451e-01
-5.51522434e-01 -8.98431540e-01 -8.64626884e-01 -8.50774825e-01
-1.94181432e-03 -2.95413792e-01 -1.04305077e+00 1.54740918e+00
-1.03594184e-01 -8.99783596e-02 4.66677010e-01 4.06790525e-01
1.29603243e+00 1.13953590e+00 -7.36321509e-02 -3.32445055e-01
1.59766781e+00 -1.22760260e+00 -9.25334454e-01 -5.00262260e-01
8.02196920e-01 -6.52112782e-01 1.08606780e+00 3.70847702e-01
-1.08241343e+00 -3.85147542e-01 -1.00625551e+00 -3.97492439e-01
-3.93390685e-01 3.20384383e-01 4.32131678e-01 4.51150417e-01
-1.14331436e+00 6.34991705e-01 -4.13443089e-01 -5.57573795e-01
3.20699930e-01 3.36936653e-01 -4.02995080e-01 1.85830489e-01
-1.48256934e+00 1.13990986e+00 5.77373445e-01 2.13633776e-02
-5.64566433e-01 -8.08500111e-01 -6.94595814e-01 5.00175297e-01
3.15421045e-01 -9.84391391e-01 1.48613405e+00 -7.43903399e-01
-1.49041867e+00 7.99536467e-01 -3.00995082e-01 -1.18592381e+00
1.21039540e-01 -7.01174319e-01 -3.49788517e-01 1.85206905e-01
1.75669119e-01 5.18444419e-01 5.34992933e-01 -7.41982698e-01
-6.24118805e-01 -2.56151944e-01 -4.23930958e-02 3.25420856e-01
-4.26757038e-01 2.63935030e-01 -6.83252439e-02 -6.83184445e-01
-2.01599315e-01 -5.04366398e-01 -8.07852820e-02 -1.16205144e+00
-7.56973505e-01 -6.75355613e-01 5.89964092e-01 -9.37329948e-01
1.28906095e+00 -1.69580007e+00 3.02898914e-01 -5.48069477e-01
1.19790003e-01 4.61981803e-01 -2.57070065e-01 6.84561074e-01
-1.21333838e-01 3.75412703e-01 7.27340393e-03 -2.10075095e-01
-2.59309169e-02 -7.52249733e-02 -9.81155097e-01 -4.52662185e-02
4.90531296e-01 1.13404870e+00 -8.74002337e-01 -2.25265890e-01
-5.74707724e-02 6.98072836e-02 -3.76262993e-01 -6.39542118e-02
-4.96803105e-01 2.22041249e-01 -5.16103327e-01 5.50577231e-02
2.67529696e-01 -1.62670329e-01 -9.60299894e-02 -8.85248706e-02
-1.53337121e-01 1.01581311e+00 -4.06837136e-01 1.57702672e+00
-3.77900213e-01 7.78792202e-01 -4.76294935e-01 -1.08281755e+00
8.71050179e-01 5.02351761e-01 1.65456116e-01 -8.36201608e-01
3.43889743e-01 8.57746303e-02 5.76531738e-02 -2.99880981e-01
1.15999734e+00 -2.79005289e-01 -4.43200827e-01 3.67820472e-01
3.62873644e-01 -1.70801848e-01 6.18239701e-01 6.01261556e-01
1.10588288e+00 -3.02220434e-01 4.32993472e-01 -4.58390772e-01
8.03351402e-01 1.96323052e-01 3.98075283e-01 8.38164926e-01
1.74932539e-01 6.02367997e-01 8.03866506e-01 -3.45729738e-01
-9.41486120e-01 -6.70111775e-01 1.70229480e-01 1.40934455e+00
-2.47698396e-01 -5.19333243e-01 -6.67012632e-01 -7.90694356e-01
-3.72175783e-01 1.38473141e+00 -7.51577199e-01 -3.58335257e-01
-8.59532177e-01 -7.95780659e-01 6.41795397e-01 4.86940742e-01
5.70957184e-01 -1.27998173e+00 -2.55941600e-01 5.07203281e-01
-6.24692738e-01 -1.30929184e+00 -5.06513059e-01 2.57958323e-01
-9.88728583e-01 -6.29648745e-01 -7.97489643e-01 -8.34866822e-01
2.32918069e-01 1.43602893e-01 1.39101624e+00 -4.35846299e-01
2.07664579e-01 2.92090401e-02 -6.42632663e-01 -8.18734884e-01
-4.19365078e-01 7.80366123e-01 -1.85043395e-01 -8.68276954e-02
2.08139986e-01 -3.52660179e-01 -2.11165324e-01 -1.40940845e-01
-7.91332483e-01 1.41038492e-01 7.24814057e-01 9.37172472e-01
3.51369977e-01 -2.31282890e-01 1.07827032e+00 -1.11562204e+00
1.08552670e+00 -2.54335523e-01 -1.62419409e-01 4.82210666e-01
-1.77984655e-01 4.26722020e-01 8.43425453e-01 -2.26223350e-01
-1.33184528e+00 -4.96000618e-01 -3.30760121e-01 1.50004417e-01
2.08019972e-01 8.48225355e-01 5.03807478e-02 5.65243661e-01
7.45333314e-01 5.73312461e-01 -5.33285737e-01 -4.46127653e-01
4.43742245e-01 5.03051937e-01 3.66230398e-01 -1.68502510e-01
3.35743308e-01 2.29163617e-01 -4.48080957e-01 -9.26242292e-01
-1.58696747e+00 -5.69323838e-01 -3.99454713e-01 9.30322707e-02
9.39556003e-01 -8.09646487e-01 -2.59322256e-01 2.55699158e-01
-1.40958452e+00 2.42551081e-02 -5.64405322e-01 2.09712669e-01
-4.04242277e-01 2.41626933e-01 -5.83813787e-01 -4.72969681e-01
-1.10545516e+00 -6.80795610e-01 9.42694783e-01 4.89558578e-01
-1.66421548e-01 -9.86201167e-01 3.61068100e-01 5.36359847e-01
5.78979492e-01 -1.13612071e-01 1.01827705e+00 -1.36152673e+00
-1.42350450e-01 -2.40435928e-01 -3.32719684e-02 4.06511903e-01
-6.58894032e-02 -2.68676937e-01 -8.62275541e-01 -1.88787449e-02
1.48131922e-01 -3.31730485e-01 1.41397691e+00 5.57715058e-01
6.44161463e-01 -4.69520390e-01 -2.91142404e-01 -5.11397049e-02
1.05256534e+00 -4.48756665e-02 7.83098519e-01 2.71023870e-01
6.19698107e-01 5.27801692e-01 3.96327734e-01 2.54400492e-01
4.49183434e-01 4.00199085e-01 9.22246873e-02 1.75576583e-01
-1.90556809e-01 -2.92944908e-01 5.07167339e-01 1.27818716e+00
2.13374496e-01 -7.01099575e-01 -6.85618103e-01 8.38485420e-01
-1.86447275e+00 -1.27083540e+00 -1.97392851e-01 1.86834371e+00
9.50621724e-01 4.10697013e-01 -1.79345638e-01 -2.61301659e-02
7.42867589e-01 3.84182632e-01 -3.88222784e-01 -7.71059513e-01
-4.39756274e-01 1.64976299e-01 3.27194363e-01 1.89939275e-01
-1.04511011e+00 1.41597116e+00 5.82247829e+00 1.08032048e+00
-1.11229503e+00 -5.58745395e-03 5.57380915e-01 -3.55672766e-03
-3.94678950e-01 -1.97122544e-01 -1.05357754e+00 -2.96222977e-02
1.35130847e+00 -5.75339615e-01 -3.81979644e-02 5.93553901e-01
2.14477062e-01 -5.27532250e-02 -8.91245723e-01 6.13392889e-01
4.61153239e-01 -1.56891036e+00 6.51781440e-01 -3.11174244e-01
9.14682448e-01 3.41743618e-01 -1.71765640e-01 7.30370164e-01
3.40803385e-01 -9.64516819e-01 5.13484359e-01 4.16570425e-01
2.77586699e-01 -7.99906671e-01 1.15926373e+00 4.39920664e-01
-8.64296019e-01 1.76372752e-02 -4.81868237e-01 3.97662595e-02
3.78920138e-01 5.50541878e-01 -1.04538345e+00 9.05091465e-01
4.59177762e-01 8.18867028e-01 -5.82244635e-01 1.01380515e+00
-4.39064264e-01 9.68314767e-01 -1.60089284e-01 -4.80286211e-01
5.42947292e-01 1.59143552e-01 8.20921421e-01 1.32683003e+00
1.28368273e-01 1.61134511e-01 -1.37670398e-01 4.29499120e-01
-5.50332963e-01 5.48874319e-01 -3.24427992e-01 -3.51902157e-01
1.80017069e-01 1.26878893e+00 -6.31567419e-01 -6.30889535e-01
-4.64615487e-02 9.73009229e-01 4.21960890e-01 -9.64448694e-03
-6.76316381e-01 -5.80824375e-01 2.05705032e-01 -1.15934663e-01
7.42605507e-01 6.26260191e-02 -2.37460107e-01 -1.42104745e+00
-1.38213381e-01 -8.33263576e-01 4.55045700e-01 -7.41962910e-01
-1.23636961e+00 9.26574230e-01 -1.73789114e-01 -1.00876474e+00
-3.68283778e-01 -4.78131652e-01 -9.06581461e-01 6.50596738e-01
-1.66129899e+00 -1.06195545e+00 1.38258442e-01 3.42827797e-01
1.08653402e+00 -2.98095137e-01 9.20542955e-01 -2.85870377e-02
-5.61148763e-01 5.28467894e-01 3.77336264e-01 2.26197541e-01
6.03438497e-01 -1.39228141e+00 7.06427336e-01 8.08773398e-01
4.41226304e-01 5.14690340e-01 9.52103436e-01 -3.54814857e-01
-1.24629891e+00 -1.02874279e+00 1.48314297e+00 -4.30648029e-01
6.86703444e-01 -2.12797970e-01 -7.56748676e-01 6.81243241e-01
7.96168804e-01 -7.03076363e-01 5.97436607e-01 3.42794955e-01
-3.46829027e-01 -1.37135446e-01 -8.59036624e-01 5.43636322e-01
6.04161918e-01 -2.30210423e-01 -1.33883357e+00 6.03773654e-01
1.05332613e+00 -3.04670691e-01 -5.16859114e-01 2.54860103e-01
1.18551359e-01 -8.83745790e-01 6.04699850e-01 -1.00061774e+00
5.82232416e-01 1.08570047e-02 1.37712806e-01 -1.65938509e+00
-2.43502542e-01 -6.91936493e-01 1.10061117e-01 1.27650082e+00
8.72169256e-01 -6.72425687e-01 5.02446949e-01 -4.27655801e-02
-5.39645970e-01 -5.99104464e-01 -8.32312405e-01 -5.81629992e-01
3.75166655e-01 -2.95420825e-01 4.12925482e-01 7.88431823e-01
3.08441877e-01 1.41957867e+00 -1.98353708e-01 -3.01628917e-01
7.61076286e-02 1.01378337e-01 4.98991996e-01 -1.26255572e+00
1.38786376e-01 -6.26518190e-01 -2.28415623e-01 -1.25544584e+00
3.15972716e-01 -9.26798284e-01 -2.83104405e-02 -2.25226521e+00
2.70769507e-01 3.03657025e-01 -2.51308411e-01 4.86115068e-01
-2.77712226e-01 -1.73942577e-02 1.69318050e-01 -1.66740064e-02
-9.30370808e-01 9.43200946e-01 1.11037838e+00 -3.31524342e-01
-2.12338090e-01 7.64202932e-03 -1.10529470e+00 4.79198039e-01
1.04369271e+00 -4.81475830e-01 -3.32296461e-01 -6.64559782e-01
5.26210785e-01 1.36915118e-01 -1.73341572e-01 -8.06559443e-01
2.69514918e-01 1.85880691e-01 1.32560030e-01 -9.72787261e-01
1.86571032e-01 -1.05360679e-01 -6.04454279e-01 3.98265779e-01
-7.28545785e-01 3.16852987e-01 5.23060560e-01 4.63578939e-01
-4.00426567e-01 -3.66905928e-01 4.06675696e-01 -3.31180066e-01
-5.58415234e-01 -1.25800192e-01 -6.11218214e-01 5.80513358e-01
6.23568654e-01 3.24108377e-02 -6.71775222e-01 -6.05448306e-01
-4.82526660e-01 2.99140066e-01 -2.68056035e-01 7.18158960e-01
6.16185248e-01 -8.44095767e-01 -1.35449755e+00 -2.37737328e-01
4.54071164e-02 -1.96074888e-01 4.93579805e-01 8.09006989e-01
-3.93180430e-01 1.07803190e+00 -3.05529665e-02 -2.21347302e-01
-1.18195689e+00 3.57662320e-01 2.44977579e-01 -1.04695272e+00
-5.46504736e-01 9.35251772e-01 3.09806049e-01 -5.26675820e-01
-1.54443175e-01 -5.16234994e-01 -1.00801551e+00 3.89165491e-01
7.76287854e-01 2.43722662e-01 4.61975574e-01 -5.23746014e-01
-2.89339542e-01 3.04439843e-01 -7.94770002e-01 -7.07700253e-02
1.45050979e+00 1.35253388e-02 -3.77635390e-01 4.29624677e-01
9.55840707e-01 1.35599717e-01 -3.98923784e-01 -1.85372561e-01
2.97204405e-01 3.19023907e-01 1.70032218e-01 -1.11212492e+00
-7.71261513e-01 8.35629225e-01 -1.14820682e-01 5.64143658e-01
9.68760729e-01 1.40612975e-01 9.78419065e-01 8.15384030e-01
1.96153559e-02 -1.09992611e+00 1.25046805e-01 1.17389441e+00
1.28057158e+00 -8.91360998e-01 5.81239024e-03 -6.86428621e-02
-1.01638842e+00 9.98840511e-01 3.91690344e-01 -2.59726197e-01
1.96637645e-01 -1.13931127e-01 -6.77373037e-02 -3.40272337e-01
-1.24500132e+00 -1.67644456e-01 5.61727524e-01 1.32281199e-01
6.25660300e-01 -1.58360705e-01 -6.61700666e-01 7.97269702e-01
-5.25563240e-01 -3.28622192e-01 8.40980768e-01 6.28908753e-01
-8.73500824e-01 -7.90086210e-01 -1.21734127e-01 7.11601973e-01
-8.79438877e-01 -5.67628562e-01 -6.45873427e-01 5.00213981e-01
-5.96408844e-01 1.05570340e+00 -9.28967260e-03 -1.24803565e-01
6.01817191e-01 2.37222821e-01 1.86225146e-01 -8.89237881e-01
-1.07411766e+00 -1.29661039e-02 5.46839118e-01 -3.02306898e-02
-3.07010233e-01 -5.19551694e-01 -1.29046667e+00 -1.01011194e-01
-6.84981942e-01 6.67843044e-01 5.77629983e-01 1.18143332e+00
5.93548298e-01 9.59631860e-01 4.23119932e-01 -5.57900012e-01
-7.38506019e-01 -1.57559466e+00 -2.72005916e-01 -1.40995365e-02
1.88829035e-01 -7.81145766e-02 -2.35699445e-01 -4.63526770e-02] | [12.452510833740234, 9.4715576171875] |
458c7dd8-4e96-416b-8d86-0ab1122590a9 | a-star-test-time-attention-segregation-and | 2306.14544 | null | https://arxiv.org/abs/2306.14544v1 | https://arxiv.org/pdf/2306.14544v1.pdf | A-STAR: Test-time Attention Segregation and Retention for Text-to-image Synthesis | While recent developments in text-to-image generative models have led to a suite of high-performing methods capable of producing creative imagery from free-form text, there are several limitations. By analyzing the cross-attention representations of these models, we notice two key issues. First, for text prompts that contain multiple concepts, there is a significant amount of pixel-space overlap (i.e., same spatial regions) among pairs of different concepts. This eventually leads to the model being unable to distinguish between the two concepts and one of them being ignored in the final generation. Next, while these models attempt to capture all such concepts during the beginning of denoising (e.g., first few steps) as evidenced by cross-attention maps, this knowledge is not retained by the end of denoising (e.g., last few steps). Such loss of knowledge eventually leads to inaccurate generation outputs. To address these issues, our key innovations include two test-time attention-based loss functions that substantially improve the performance of pretrained baseline text-to-image diffusion models. First, our attention segregation loss reduces the cross-attention overlap between attention maps of different concepts in the text prompt, thereby reducing the confusion/conflict among various concepts and the eventual capture of all concepts in the generated output. Next, our attention retention loss explicitly forces text-to-image diffusion models to retain cross-attention information for all concepts across all denoising time steps, thereby leading to reduced information loss and the preservation of all concepts in the generated output. | ['Balaji Vasan Srinivasan', 'Koustava Goswami', 'Apoorv Saxena', 'K J Joseph', 'Srikrishna Karanam', 'Aishwarya Agarwal'] | 2023-06-26 | null | null | null | null | ['image-generation'] | ['computer-vision'] | [ 5.98719239e-01 -9.15857032e-02 4.48905021e-01 8.94466564e-02
-6.56616509e-01 -7.40676641e-01 8.23393285e-01 2.37861261e-01
-1.77436650e-01 5.38177371e-01 3.63634855e-01 -1.01378135e-01
-2.09321544e-01 -8.52176368e-01 -6.85174346e-01 -7.48091936e-01
4.00287867e-01 2.68275529e-01 2.66586512e-01 -2.24945292e-01
2.57438868e-01 2.79061049e-01 -1.52934933e+00 3.61740023e-01
1.10271347e+00 7.22019970e-01 5.77608466e-01 6.23393178e-01
-2.55774498e-01 8.16476345e-01 -8.23028207e-01 -3.05805564e-01
1.18456982e-01 -7.86870837e-01 -5.64664900e-01 3.30939263e-01
7.61532009e-01 -5.35081506e-01 -3.92818183e-01 1.12290967e+00
3.79218668e-01 3.98688078e-01 7.58332491e-01 -8.63570392e-01
-1.06634378e+00 3.79687339e-01 -7.36974239e-01 5.89452200e-02
9.45851654e-02 2.57416666e-01 1.15768492e+00 -1.04883802e+00
7.22737849e-01 1.20125484e+00 4.87592429e-01 2.55315870e-01
-1.49743736e+00 -4.75456178e-01 3.07700425e-01 -1.25492379e-01
-1.27522922e+00 -2.93439120e-01 7.52908945e-01 -5.74059188e-01
8.16371024e-01 2.55067289e-01 8.79578352e-01 1.02857423e+00
3.00264984e-01 6.65551841e-01 9.78404582e-01 -2.28078291e-01
3.69404964e-02 1.28736207e-02 -8.71307775e-02 5.87706566e-01
1.42619491e-01 3.85851134e-03 -7.34844148e-01 1.58728346e-01
9.36046600e-01 1.53498584e-02 -3.62172335e-01 1.27001023e-02
-9.22619402e-01 8.12050879e-01 5.92027605e-01 4.26868081e-01
-6.66492999e-01 1.33144855e-01 -3.85919213e-02 4.46006097e-03
5.30665457e-01 3.86905402e-01 9.41378176e-02 1.60957202e-02
-1.39233732e+00 3.35927486e-01 1.88702941e-01 7.63636351e-01
8.66511166e-01 2.44454905e-01 -3.63812178e-01 8.28317761e-01
1.01328199e-03 3.60143721e-01 3.49144489e-01 -8.08112383e-01
4.56629694e-01 5.95394552e-01 4.54022549e-02 -1.13911355e+00
-4.83009703e-02 -7.15737402e-01 -8.54426742e-01 3.37754756e-01
4.21414733e-01 -2.01477170e-01 -1.13814259e+00 1.86483562e+00
-1.37208387e-01 -6.34838492e-02 -6.64333850e-02 8.22841704e-01
5.13635278e-01 8.13101947e-01 1.77237555e-01 -7.81745538e-02
1.27268612e+00 -8.54257882e-01 -6.63329363e-01 -5.69386363e-01
1.52263492e-01 -9.09017086e-01 1.00073421e+00 1.95122674e-01
-1.38848591e+00 -8.74432564e-01 -9.89253283e-01 -2.13527218e-01
-2.77185202e-01 -2.63969958e-01 4.63131607e-01 2.08042324e-01
-1.08215821e+00 7.06494331e-01 -5.21176338e-01 -3.70738059e-01
5.79973161e-01 9.43483785e-02 -1.59052387e-01 -4.22877640e-01
-1.03095233e+00 6.41666710e-01 2.71102898e-02 9.29253250e-02
-8.42695892e-01 -8.46486628e-01 -5.23646593e-01 3.98814082e-01
2.13970080e-01 -7.79280066e-01 9.59851146e-01 -1.42965901e+00
-7.89645672e-01 4.07924771e-01 -3.27661484e-01 -2.15072110e-01
6.16900802e-01 -3.01710844e-01 -1.82565182e-01 1.85421243e-01
3.21048915e-01 1.15855515e+00 1.15580475e+00 -1.61841130e+00
-7.09126651e-01 -2.54000425e-01 -1.56491119e-02 6.08634353e-01
-3.70494127e-01 -3.22064161e-01 -6.50678158e-01 -1.09388113e+00
2.90465981e-01 -8.34958196e-01 1.69032097e-01 2.46772096e-02
-3.05221200e-01 1.81378171e-01 7.76598275e-01 -7.33414710e-01
1.19043922e+00 -2.17037153e+00 2.98990905e-01 -7.20700901e-03
3.85854483e-01 1.31175350e-02 -4.36497748e-01 4.34883356e-01
-8.50013644e-02 3.29142809e-01 -4.91668910e-01 -3.80377501e-01
-3.62942249e-01 1.69300616e-01 -6.48951292e-01 1.38859022e-02
5.59535682e-01 9.33812082e-01 -8.69357109e-01 -8.72874707e-02
1.41264066e-01 7.28982508e-01 -4.87358242e-01 -9.30575132e-02
-2.65018106e-01 4.42937046e-01 -2.23223627e-01 3.90048176e-01
6.01302803e-01 -1.02882355e-01 -1.64247155e-01 -6.98248446e-02
-1.14724286e-01 1.73625290e-01 -9.29631770e-01 1.60026956e+00
-3.62605542e-01 9.37577665e-01 -1.77609801e-01 -6.17723167e-01
5.54265201e-01 2.27171823e-01 5.14014959e-01 -7.77468503e-01
-1.63243547e-01 1.31314993e-01 1.21961385e-01 -2.12087616e-01
7.78994083e-01 -3.76986206e-01 4.10511255e-01 5.91901600e-01
-1.63238756e-02 -2.91787326e-01 3.02743644e-01 4.77532208e-01
9.75311995e-01 1.39216885e-01 -3.88755471e-01 -8.93687457e-02
6.94046244e-02 1.21551119e-01 3.59205991e-01 1.01787364e+00
8.02426189e-02 1.07487214e+00 3.61662179e-01 2.14069765e-02
-1.05673540e+00 -1.15412617e+00 9.17818919e-02 7.74583876e-01
1.18672527e-01 -3.24678004e-01 -7.60987043e-01 -4.97270107e-01
-1.06549196e-01 9.79524195e-01 -8.37594390e-01 -2.46940762e-01
-3.45630765e-01 -6.89929664e-01 3.64465386e-01 5.91112196e-01
5.00152767e-01 -1.14676678e+00 -5.91767073e-01 4.35432285e-01
-4.29405451e-01 -7.73530304e-01 -5.65845490e-01 2.95871586e-01
-1.10171306e+00 -8.90251517e-01 -1.03652513e+00 -5.84934354e-01
1.05893326e+00 5.08239448e-01 1.10066438e+00 3.08727235e-01
-2.90160596e-01 3.20664793e-01 -1.85585842e-01 -2.23547459e-01
-2.15902001e-01 -2.14672700e-01 -4.96957302e-01 8.45448896e-02
4.62989546e-02 -3.72288436e-01 -5.78432083e-01 5.16745672e-02
-1.08041418e+00 2.86738843e-01 5.96209764e-01 9.88712251e-01
5.68035960e-01 4.19768214e-01 3.12574565e-01 -5.73870599e-01
8.27221870e-01 -4.43845600e-01 -3.03110272e-01 1.54924721e-01
-3.92303705e-01 1.33968234e-01 4.62580621e-01 -6.94373846e-01
-1.15596604e+00 -3.27228874e-01 7.54324719e-02 -3.74942124e-01
9.17114839e-02 6.35142446e-01 7.61990249e-02 2.81495214e-01
6.24264836e-01 4.89913702e-01 -1.15174375e-01 -2.61537462e-01
2.81680286e-01 1.84297577e-01 5.25527775e-01 -3.75111252e-01
7.74632037e-01 4.14411634e-01 -3.15096259e-01 -8.43191147e-01
-8.26215506e-01 -1.97150812e-01 -4.35430467e-01 -2.56299436e-01
9.88964260e-01 -9.02281344e-01 -8.74947831e-02 6.81373656e-01
-1.26597226e+00 -3.90201986e-01 -5.15601635e-01 2.76682645e-01
-2.62957364e-01 1.89789206e-01 -5.62444508e-01 -7.31467903e-01
-1.38277680e-01 -1.32485175e+00 8.80498767e-01 2.56129980e-01
-4.71604586e-01 -1.00171387e+00 -3.23367596e-01 2.91052043e-01
3.64614010e-01 -9.06086937e-02 1.25709939e+00 -4.00881395e-02
-7.40340590e-01 -1.35993332e-01 -3.47178072e-01 3.41028959e-01
2.38056898e-01 8.80328640e-02 -9.07666624e-01 -2.61453748e-01
9.80358645e-02 -1.19201571e-01 1.21088850e+00 5.60048401e-01
6.86841547e-01 -1.39877409e-01 -2.76149243e-01 2.85089970e-01
1.27446532e+00 4.29284781e-01 8.96342993e-01 -1.66627206e-02
6.85923696e-01 7.15345442e-01 2.82295763e-01 -1.81494758e-03
2.02050611e-01 4.59881008e-01 1.28423348e-01 -5.35695791e-01
-5.04374027e-01 -3.93216878e-01 3.38305622e-01 6.20729923e-01
-6.71804324e-02 -6.40915871e-01 -7.03905046e-01 8.42714965e-01
-1.71399319e+00 -1.05419326e+00 -1.87152952e-01 2.36408544e+00
7.30969489e-01 1.62707075e-01 -3.28763396e-01 7.60363461e-03
6.58687770e-01 1.86498970e-01 -6.63698554e-01 -2.29145080e-01
-4.25443888e-01 2.47228086e-01 2.17845246e-01 6.71050429e-01
-7.79226065e-01 1.04201186e+00 6.17282391e+00 8.11695933e-01
-1.06042540e+00 -9.79705015e-04 7.04023063e-01 -2.87737101e-01
-5.91639757e-01 2.53985256e-01 -4.25997347e-01 4.02846217e-01
4.75428432e-01 -7.83937983e-03 5.91150224e-01 2.72624314e-01
1.38812974e-01 -5.76242149e-01 -8.48655462e-01 7.05265582e-01
2.35827044e-01 -1.18040562e+00 3.68113518e-01 6.43122047e-02
1.13148332e+00 -2.51455575e-01 2.22899944e-01 2.39498373e-02
3.70083779e-01 -1.07259476e+00 1.08478522e+00 5.14207363e-01
7.33360350e-01 -7.73562431e-01 4.52919692e-01 1.05951495e-01
-9.89836633e-01 -1.48011371e-01 -3.52424532e-01 -1.90417771e-03
2.01304451e-01 7.53723443e-01 -3.10281515e-01 4.70747322e-01
4.58341628e-01 4.47477669e-01 -4.69096214e-01 7.00007319e-01
-4.19325411e-01 3.34350795e-01 -1.22972220e-01 3.58746111e-01
4.05602694e-01 -2.59501338e-01 6.10890925e-01 8.18881035e-01
6.14345551e-01 1.55786052e-01 -1.27911016e-01 1.37669551e+00
-1.24186635e-01 -2.03713149e-01 -4.79838222e-01 -1.66511148e-01
2.60212213e-01 9.29836333e-01 -8.52389038e-01 -4.75205958e-01
-4.70313191e-01 1.52067649e+00 1.94124684e-01 7.89525449e-01
-7.84928918e-01 -2.92750001e-01 6.34028494e-01 5.39859831e-01
3.68668884e-01 -3.02019238e-01 -6.59462333e-01 -7.92745233e-01
1.45951658e-02 -7.68135786e-01 8.10754597e-02 -1.15009999e+00
-1.17515671e+00 5.36230981e-01 -1.86417833e-01 -7.92843938e-01
8.33562985e-02 -6.52639046e-02 -6.44911587e-01 1.25552058e+00
-1.35152698e+00 -1.19791758e+00 -2.82496542e-01 4.76394862e-01
8.04711819e-01 -4.37494693e-03 5.05751312e-01 2.23987460e-01
-2.31436282e-01 3.32497030e-01 1.16488650e-01 -3.14856470e-02
7.99335718e-01 -1.08439004e+00 4.30080682e-01 1.10668838e+00
3.64439309e-01 8.68524194e-01 4.67083037e-01 -1.11413574e+00
-8.17468703e-01 -9.54362214e-01 9.46594357e-01 -3.48406345e-01
4.53291655e-01 -2.79894173e-01 -1.07860732e+00 4.86397713e-01
2.39866570e-01 -5.31641006e-01 3.45989317e-01 -1.26475349e-01
-3.54500115e-01 1.27416402e-01 -7.32313156e-01 1.00571561e+00
8.35864246e-01 -6.14788830e-01 -4.12782490e-01 6.36746660e-02
3.96553248e-01 -1.19589455e-01 -4.08608854e-01 1.13772646e-01
5.47647536e-01 -9.89996731e-01 8.75253797e-01 -2.16727078e-01
8.85707319e-01 -3.07321161e-01 1.68437541e-01 -1.40505850e+00
-5.33689737e-01 -4.72663045e-01 5.93741760e-02 1.34435034e+00
4.98188585e-01 -4.01854038e-01 3.36255044e-01 6.89426243e-01
-1.24280378e-01 -4.86145943e-01 -6.37029350e-01 -4.70426112e-01
3.61403435e-01 -3.79484385e-01 2.39951596e-01 8.44414115e-01
-4.18580085e-01 4.37651098e-01 -3.95119399e-01 -1.62378117e-01
4.30253655e-01 3.64077240e-02 2.65087396e-01 -1.20645070e+00
-1.66424051e-01 -5.94183326e-01 1.24991015e-01 -1.28329766e+00
-3.39579910e-01 -6.10703409e-01 3.80098343e-01 -1.92121422e+00
3.43957335e-01 -4.89708722e-01 -9.92065519e-02 5.64052403e-01
-5.79366386e-01 4.18949634e-01 4.17159259e-01 4.60318089e-01
2.64454931e-02 5.24210036e-01 1.52483106e+00 -2.74433255e-01
-3.01687956e-01 -4.58871752e-01 -9.50798392e-01 5.02756178e-01
6.57904029e-01 -6.29645586e-01 -6.41990542e-01 -7.77584434e-01
7.62032941e-02 -7.81179965e-02 7.49818921e-01 -1.12774456e+00
1.94254518e-01 -6.47063926e-02 8.07695448e-01 -5.37550688e-01
4.21534538e-01 -6.44198239e-01 2.69101381e-01 3.26057792e-01
-2.49539450e-01 4.95613106e-02 4.40335870e-01 5.57139754e-01
-1.23578742e-01 -3.11893642e-01 8.67615759e-01 -2.60081261e-01
-4.46654588e-01 3.89194041e-02 -6.65936232e-01 1.02878332e-01
7.06486762e-01 -6.30165875e-01 -2.37128720e-01 -5.89404047e-01
-6.78476512e-01 -8.96342471e-02 6.09012902e-01 5.82761288e-01
4.78591770e-01 -1.08408773e+00 -7.30286241e-01 2.85486311e-01
-2.19692200e-01 9.91384462e-02 6.01073980e-01 6.49353802e-01
-2.14893788e-01 2.30068773e-01 -9.76634026e-02 -3.63182455e-01
-1.06726158e+00 3.13670963e-01 2.06015721e-01 -2.92269468e-01
-7.19317257e-01 9.61505532e-01 7.70227075e-01 1.60083354e-01
5.48747629e-02 -1.38006553e-01 8.34761560e-02 2.38147780e-01
3.55811298e-01 5.16288579e-02 -7.29143480e-03 -6.53642118e-01
-1.18936663e-02 6.79791093e-01 -1.92478210e-01 -5.61445713e-01
1.20315182e+00 -1.57507718e-01 4.59521115e-02 2.52625257e-01
7.55992770e-01 8.31400380e-02 -1.72985923e+00 -2.72044707e-02
-3.80854487e-01 -5.01832724e-01 3.79671097e-01 -1.09497523e+00
-1.09560478e+00 1.14221370e+00 4.94138062e-01 -2.93950234e-02
1.26911223e+00 -3.16933215e-01 6.98614597e-01 -1.62059870e-02
-1.92919016e-01 -1.26404345e+00 4.51639891e-01 5.81797361e-01
1.08465207e+00 -9.69350994e-01 6.26176894e-02 -2.66233146e-01
-9.48508859e-01 9.61054623e-01 5.58711588e-01 1.71135768e-01
1.05876453e-01 2.01180458e-01 -1.27638713e-01 -2.89950371e-01
-5.29349327e-01 -2.10075840e-01 3.57551515e-01 6.00374460e-01
4.65650976e-01 -3.38711709e-01 -4.21393067e-02 3.80853325e-01
5.16053922e-02 -2.61812776e-01 3.24737668e-01 7.42261052e-01
-4.52059597e-01 -8.75444591e-01 -4.63905573e-01 4.64547634e-01
-3.16470325e-01 -6.33956313e-01 -4.84512419e-01 6.28523290e-01
2.54364431e-01 1.05212259e+00 4.93253350e-01 -7.33956844e-02
2.09544376e-01 2.81490058e-01 5.07257283e-01 -5.91152668e-01
-6.70283556e-01 5.43368995e-01 -2.85832644e-01 -5.10341600e-02
-8.90262723e-02 -5.44512928e-01 -1.22398007e+00 -2.91641980e-01
-4.30558085e-01 -1.59871221e-01 4.75744516e-01 7.88402379e-01
3.35529953e-01 9.94145632e-01 1.56742364e-01 -7.49309480e-01
-1.17400460e-01 -9.77955103e-01 -5.79561710e-01 6.69816554e-01
2.34055519e-01 -6.60655081e-01 -2.08208859e-01 3.27345729e-01] | [11.35998249053955, -0.1927841305732727] |
125eb2e9-3feb-497d-97c6-72010e327ce3 | hd2reg-hierarchical-descriptors-and-detectors | 2305.03487 | null | https://arxiv.org/abs/2305.03487v1 | https://arxiv.org/pdf/2305.03487v1.pdf | HD2Reg: Hierarchical Descriptors and Detectors for Point Cloud Registration | Feature Descriptors and Detectors are two main components of feature-based point cloud registration. However, little attention has been drawn to the explicit representation of local and global semantics in the learning of descriptors and detectors. In this paper, we present a framework that explicitly extracts dual-level descriptors and detectors and performs coarse-to-fine matching with them. First, to explicitly learn local and global semantics, we propose a hierarchical contrastive learning strategy, training the robust matching ability of high-level descriptors, and refining the local feature space using low-level descriptors. Furthermore, we propose to learn dual-level saliency maps that extract two groups of keypoints in two different senses. To overcome the weak supervision of binary matchability labels, we propose a ranking strategy to label the significance ranking of keypoints, and thus provide more fine-grained supervision signals. Finally, we propose a global-to-local matching scheme to obtain robust and accurate correspondences by leveraging the complementary dual-level features.Quantitative experiments on 3DMatch and KITTI odometry datasets show that our method achieves robust and accurate point cloud registration and outperforms recent keypoint-based methods. | ['Zhiqiang Tian', 'Guofa Wang', 'Shaoyi Du', 'Yiheng Li', 'Canhui Tang'] | 2023-05-05 | null | null | null | null | ['point-cloud-registration'] | ['computer-vision'] | [-1.26193330e-01 -3.08921218e-01 -4.88930792e-01 -3.91321361e-01
-9.84804273e-01 -2.39053577e-01 6.89486861e-01 4.21320617e-01
-2.60466754e-01 1.67717546e-01 1.57275379e-01 3.41729879e-01
-4.59012926e-01 -8.63784909e-01 -5.94411731e-01 -4.15887058e-01
-8.06005485e-03 3.35807443e-01 6.61257446e-01 -2.35534459e-01
5.84303379e-01 9.23056006e-01 -1.94155586e+00 -2.12768987e-02
8.92276883e-01 1.11322570e+00 3.83848011e-01 4.89736274e-02
1.43974843e-02 3.83928686e-01 -2.29773670e-01 1.15025565e-01
3.96461040e-01 -2.69811839e-01 -5.67847311e-01 8.72460306e-02
7.75025845e-01 -1.33540943e-01 -3.47701341e-01 1.27856600e+00
3.55784237e-01 8.88536796e-02 5.42364359e-01 -1.42284572e+00
-6.79369986e-01 -1.17221132e-01 -7.15978563e-01 2.20119745e-01
5.96958578e-01 -1.33016089e-03 1.39309287e+00 -1.25398982e+00
6.06683910e-01 1.16811299e+00 5.30737936e-01 1.27859473e-01
-9.32746708e-01 -8.33395064e-01 2.55108893e-01 2.69008547e-01
-1.75242305e+00 -3.75961512e-01 1.00346303e+00 -5.30668795e-01
7.43784726e-01 2.66018510e-02 6.88896000e-01 4.85764265e-01
4.66363169e-02 7.23579645e-01 1.00795984e+00 -4.22999203e-01
1.73058718e-01 -1.36276990e-01 9.99623090e-02 1.01822114e+00
2.20561296e-01 4.83952016e-01 -7.22425520e-01 -3.20834160e-01
1.04993951e+00 5.35284758e-01 -1.77688804e-02 -9.05184388e-01
-1.39046836e+00 7.39456654e-01 1.00031900e+00 3.81257087e-01
-5.27419508e-01 1.43298492e-01 7.20191328e-03 6.68534487e-02
2.57069677e-01 4.08263087e-01 -1.86474472e-01 3.07261884e-01
-1.02231514e+00 1.24433301e-01 2.53594816e-01 1.31845689e+00
1.44556379e+00 -3.59931737e-01 -3.66863072e-01 6.51446879e-01
5.73523939e-01 6.50611401e-01 4.37837809e-01 -7.83409536e-01
3.92050087e-01 9.94186521e-01 3.22447084e-02 -1.32606900e+00
-3.17782909e-01 -3.93479049e-01 -5.06450295e-01 3.60568404e-01
-8.06250349e-02 5.95389664e-01 -1.08328211e+00 1.38214195e+00
4.21460390e-01 6.12704754e-01 -2.94887990e-01 1.06825113e+00
7.44033813e-01 2.44241998e-01 -9.20941681e-03 2.42540479e-01
1.35813510e+00 -8.72197211e-01 -2.80642271e-01 -2.07309395e-01
4.10591334e-01 -5.58327377e-01 9.01469350e-01 -2.97797769e-01
-9.20804441e-01 -7.30611980e-01 -1.31050336e+00 -1.92962915e-01
-3.44772786e-01 1.40412271e-01 5.39922118e-01 1.20139606e-01
-9.02387261e-01 5.92151046e-01 -8.91785145e-01 -2.63384968e-01
3.76826197e-01 5.02948701e-01 -4.23453838e-01 -9.41801220e-02
-1.10032344e+00 9.97430146e-01 3.51702690e-01 -1.17385909e-01
-6.42203748e-01 -5.67354381e-01 -1.16743934e+00 1.81777999e-02
3.62784900e-02 -6.77976549e-01 7.78571844e-01 -4.17405218e-01
-1.33680630e+00 1.17099392e+00 -3.09941351e-01 -4.01227698e-02
9.37364772e-02 -5.61285429e-02 -2.27139696e-01 3.32882047e-01
6.52825892e-01 8.24270070e-01 8.90380263e-01 -1.31004214e+00
-1.10169840e+00 -5.88769734e-01 7.67743168e-03 3.44328165e-01
-1.65567204e-01 1.77980419e-02 -5.42141080e-01 -4.62346077e-01
8.33794713e-01 -6.81791961e-01 -2.13291943e-01 2.04393223e-01
-1.56159967e-01 -4.47087735e-01 7.00130999e-01 -1.15250491e-01
8.80902231e-01 -2.26112294e+00 5.90763427e-02 5.48068523e-01
4.71302271e-01 -1.47016689e-01 -1.99480727e-01 2.39384845e-01
1.32365897e-01 -3.43465745e-01 1.93307269e-02 -2.57656842e-01
1.39246121e-01 1.82405964e-01 -3.13958168e-01 6.64848387e-01
4.60580438e-01 1.01097095e+00 -1.15535188e+00 -6.60603166e-01
4.43676233e-01 3.53423506e-01 -5.07438779e-01 4.43987101e-02
3.09079707e-01 3.46284330e-01 -9.58978713e-01 1.00561595e+00
6.62474394e-01 -2.49002844e-01 -4.42413658e-01 -4.21611726e-01
-2.51001149e-01 3.02422285e-01 -1.18259668e+00 2.00119448e+00
-4.21769589e-01 3.63819122e-01 -1.25231519e-01 -7.93159544e-01
1.28827488e+00 -1.61066055e-01 5.03535688e-01 -9.08178210e-01
7.08141997e-02 5.26036024e-01 -4.97950047e-01 -2.33554449e-02
6.47341788e-01 -7.19089955e-02 -3.56848001e-01 8.46025869e-02
1.95800245e-01 -2.90849417e-01 -1.87795267e-01 -1.55701619e-02
9.53266144e-01 1.30045399e-01 5.59995711e-01 -2.86837786e-01
7.50292301e-01 5.04509583e-02 7.19314694e-01 7.00616658e-01
-3.43583465e-01 8.31300139e-01 -1.45584762e-01 -3.76530856e-01
-7.79577732e-01 -1.07517970e+00 -2.34718814e-01 1.10513091e+00
9.66638923e-01 -4.98667091e-01 -4.00961637e-01 -6.46415174e-01
3.75270307e-01 1.72075734e-03 -5.45312762e-01 -3.63278091e-01
-5.69264948e-01 -1.57761961e-01 3.25635910e-01 6.45869970e-01
4.60890800e-01 -8.77671957e-01 -6.27685964e-01 5.92712276e-02
1.38209686e-01 -9.84246612e-01 -6.04387760e-01 1.88018575e-01
-9.31533575e-01 -1.02280891e+00 -5.97788930e-01 -1.12187469e+00
8.34868491e-01 7.82532156e-01 8.83869886e-01 3.32920194e-01
-2.23754182e-01 3.74408513e-01 -3.35249543e-01 -1.08546868e-01
1.62494138e-01 1.54952347e-01 1.86903119e-01 -1.13885351e-01
7.02096164e-01 -5.17985404e-01 -6.71067595e-01 6.34926498e-01
-5.59522033e-01 -2.10552409e-01 9.30732608e-01 8.29294801e-01
9.06131089e-01 -4.13526356e-01 2.61856258e-01 -2.46706203e-01
3.98313850e-01 -1.44889265e-01 -8.00071955e-01 1.52632147e-01
-5.08668423e-01 2.84557074e-01 1.32360682e-01 -3.89586762e-02
-5.22863626e-01 4.07628417e-01 -3.93880345e-02 -7.85047412e-01
-2.77729481e-01 2.93391049e-01 -1.72651812e-01 -6.00271702e-01
5.25738299e-01 3.30760598e-01 -2.41502270e-01 -4.50591087e-01
5.12694836e-01 5.54006934e-01 6.64517522e-01 -7.33023942e-01
1.35349667e+00 7.45946646e-01 -5.26197366e-02 -5.31472445e-01
-9.08227086e-01 -1.00073838e+00 -1.11637533e+00 9.39292908e-02
6.81124628e-01 -1.17381012e+00 -5.00053227e-01 3.14071447e-01
-1.03756022e+00 2.43583485e-01 -5.96937895e-01 4.69228357e-01
-7.48123169e-01 3.79987657e-01 -4.31694359e-01 -2.39532039e-01
-1.74881667e-01 -1.24371207e+00 1.69364190e+00 3.17055613e-01
1.01598836e-01 -7.83410132e-01 3.10426056e-01 -8.19010958e-02
1.49652019e-01 1.88507542e-01 3.67983282e-01 -4.98252153e-01
-9.70574260e-01 -3.43941510e-01 -6.11499846e-01 -3.13291699e-02
3.27538610e-01 -2.27087691e-01 -9.32178736e-01 -4.75437045e-01
-1.01361595e-01 -8.90444890e-02 9.20877397e-01 1.50407791e-01
8.01213741e-01 1.76472157e-01 -7.93179214e-01 7.95575559e-01
1.35209203e+00 -1.45023376e-01 3.43987763e-01 6.16971314e-01
8.33899021e-01 4.76805449e-01 1.01750112e+00 4.40201223e-01
6.90832317e-01 8.20267320e-01 3.39555860e-01 -2.80065000e-01
-1.37884051e-01 -6.68239176e-01 1.94630802e-01 8.11538339e-01
-4.12012227e-02 6.24421358e-01 -8.43235016e-01 6.35156453e-01
-1.94581997e+00 -8.09449255e-01 1.08014971e-01 2.21780753e+00
6.26608372e-01 7.55834579e-02 -5.12678698e-02 -5.73001765e-02
9.32812274e-01 3.10087472e-01 -3.70162070e-01 3.50161135e-01
-4.25602905e-02 2.02988401e-01 5.90282679e-01 4.29488420e-01
-1.31990099e+00 1.16016805e+00 6.02664280e+00 7.97222793e-01
-1.10185492e+00 9.23788324e-02 -1.02836363e-01 3.13612908e-01
-3.70832264e-01 3.53037864e-01 -1.04491436e+00 3.17118824e-01
2.08642140e-01 -1.89741954e-01 -7.12018460e-02 9.45503116e-01
-1.53329566e-01 1.66608870e-01 -1.20396316e+00 1.18197536e+00
1.15077034e-01 -1.40205944e+00 6.54606074e-02 1.75853133e-01
7.67129362e-01 2.32012019e-01 -4.98749316e-02 2.17193544e-01
1.63717762e-01 -7.14549899e-01 9.33338225e-01 5.19575834e-01
6.40855968e-01 -6.47219062e-01 5.20714700e-01 2.68717706e-01
-1.76047897e+00 -4.38125059e-03 -7.14465976e-01 -1.13332048e-02
-7.75792673e-02 4.06374156e-01 -4.65236247e-01 7.30318487e-01
7.14002132e-01 1.16822660e+00 -6.16974354e-01 1.38907754e+00
-3.92420173e-01 -1.97577089e-01 -3.92619669e-01 2.29541615e-01
2.72285968e-01 -8.68325531e-02 4.64141130e-01 1.05785191e+00
3.41372788e-01 5.89449033e-02 5.99006832e-01 9.37300563e-01
-1.74064524e-02 3.86297442e-02 -5.54698765e-01 5.20644963e-01
8.70772541e-01 1.26934195e+00 -7.74489701e-01 -2.91525334e-01
-5.23670971e-01 9.67860579e-01 5.37907720e-01 6.38229921e-02
-4.95362967e-01 -6.37008905e-01 8.65387619e-01 1.12829700e-01
3.03200811e-01 -4.25462514e-01 -2.09406465e-01 -1.25221133e+00
8.78079534e-02 -3.61262828e-01 3.06638658e-01 -5.56689680e-01
-1.38168824e+00 3.98335427e-01 -3.28525454e-02 -1.71202981e+00
-7.34388605e-02 -5.09548545e-01 -6.51296377e-01 9.49412942e-01
-2.00711679e+00 -1.42117059e+00 -8.13518882e-01 7.95984685e-01
1.79902464e-01 -5.69032095e-02 5.05113721e-01 3.97400230e-01
-9.01932567e-02 5.61830699e-01 -2.75073089e-02 2.54129797e-01
7.14492381e-01 -1.10798275e+00 4.45083916e-01 6.83148205e-01
3.26734334e-01 7.96174824e-01 3.43955010e-01 -5.32320857e-01
-1.21297681e+00 -1.05959809e+00 7.48466074e-01 -5.63924670e-01
6.38748169e-01 -3.90521079e-01 -1.04461324e+00 5.33760905e-01
-4.35023725e-01 4.89976615e-01 2.61306942e-01 -4.17434960e-04
-4.90995705e-01 -2.68817782e-01 -1.04093337e+00 2.32852608e-01
1.23286438e+00 -9.85919774e-01 -9.50483441e-01 1.91511601e-01
6.82086766e-01 -4.95069444e-01 -8.92327428e-01 5.04402518e-01
4.17715937e-01 -9.03482676e-01 1.25925136e+00 -1.48935378e-01
-9.99192670e-02 -7.02620029e-01 -2.18964607e-01 -1.08212185e+00
-7.18466043e-01 -1.79213613e-01 5.21122664e-02 1.10806143e+00
-1.10785343e-01 -6.55969620e-01 9.51773167e-01 7.83736929e-02
-2.96179175e-01 -5.20876467e-01 -1.09226680e+00 -9.36042547e-01
-9.92424712e-02 -2.13342696e-01 8.89432967e-01 1.07183874e+00
6.37973025e-02 6.19889870e-02 -1.15422629e-01 4.95175749e-01
8.24166834e-01 5.91229677e-01 7.03599334e-01 -1.69629967e+00
2.19873488e-01 -5.74587286e-01 -1.22293830e+00 -1.40220296e+00
3.04848105e-01 -1.12875569e+00 2.52772599e-01 -1.37486088e+00
1.97173268e-01 -7.28918254e-01 -7.48033047e-01 5.67896605e-01
-1.64230794e-01 4.60678697e-01 1.28269702e-01 7.90893316e-01
-8.31415772e-01 8.29588592e-01 1.08563888e+00 -1.87801450e-01
-1.78499997e-01 -1.55333802e-01 -5.26179552e-01 5.82729876e-01
3.68170172e-01 -5.67382693e-01 -9.67156813e-02 -3.18568468e-01
-1.01196192e-01 -2.57253349e-01 7.03105092e-01 -1.16826546e+00
7.12763667e-01 -2.36841902e-01 3.65661293e-01 -7.91942060e-01
2.57788539e-01 -7.49343157e-01 -2.92685211e-01 3.87327075e-01
-1.30969897e-01 -1.21001434e-02 -1.14964597e-01 6.22087240e-01
-6.44549251e-01 3.88702191e-02 8.72275949e-01 -4.69543487e-02
-1.22433341e+00 6.45135343e-01 2.76212305e-01 -1.54081620e-02
1.07596779e+00 -5.94791532e-01 -9.22153071e-02 5.60194217e-02
-4.21397030e-01 4.16154325e-01 9.09384370e-01 7.39073157e-01
9.99946713e-01 -1.72680581e+00 -5.30310035e-01 5.43745458e-01
8.08102727e-01 7.88692534e-02 -7.30804056e-02 7.61158407e-01
-2.99900234e-01 4.78584915e-01 -5.28089762e-01 -1.03177953e+00
-9.49580908e-01 4.96362656e-01 3.42636228e-01 7.63346106e-02
-7.79341817e-01 7.72007585e-01 4.14531082e-01 -5.13851047e-01
5.30486107e-02 -4.21334833e-01 -1.52943939e-01 1.35396197e-01
2.81609654e-01 2.84546942e-01 2.22264633e-01 -1.07078123e+00
-8.17422509e-01 1.33536983e+00 -9.73371416e-02 1.92046165e-01
1.09373021e+00 -2.03003794e-01 -9.64970440e-02 1.39739066e-01
1.34831500e+00 -3.10952663e-02 -1.23134387e+00 -7.11167395e-01
5.66377163e-01 -8.32216620e-01 9.48710591e-02 -1.04305327e-01
-9.01844680e-01 8.39488924e-01 7.51042962e-01 -1.44972101e-01
9.83872533e-01 3.93800139e-01 6.69484675e-01 2.69633830e-01
8.40126038e-01 -9.14373040e-01 1.52476355e-01 5.31119108e-01
7.00420618e-01 -1.45346344e+00 8.50967243e-02 -4.39627558e-01
-2.29423791e-01 8.93824458e-01 6.24388337e-01 -5.54561853e-01
6.17646337e-01 -4.10928838e-02 -2.82187597e-03 -6.49958193e-01
-3.11576307e-01 -6.17228150e-01 7.47298002e-01 6.67134225e-01
1.23549543e-01 -8.76767561e-02 3.80002968e-02 3.06878388e-01
-5.37337214e-02 -1.92871585e-01 -3.83979939e-02 1.09778655e+00
-8.79245520e-01 -1.07548475e+00 -4.17445838e-01 2.22605243e-01
8.06356892e-02 3.32419208e-04 -2.33175099e-01 8.19614828e-01
8.47927630e-02 5.79122782e-01 2.05585912e-01 -5.71971774e-01
6.55290425e-01 -3.91866416e-01 3.29869956e-01 -8.64936829e-01
-2.85114497e-01 -7.86294937e-02 -5.22936940e-01 -8.30488026e-01
-5.42346895e-01 -5.63384235e-01 -1.42246592e+00 1.06685609e-01
-5.67197740e-01 2.06763282e-01 4.68654901e-01 9.83662486e-01
5.96210241e-01 2.33116284e-01 8.33530426e-01 -1.41637325e+00
-4.87222135e-01 -5.08646607e-01 -7.14606702e-01 5.39659441e-01
5.84449291e-01 -1.15364885e+00 -2.88656563e-01 -4.16966408e-01] | [7.822451591491699, -2.181776285171509] |
5e4f1739-2d01-46bf-8d1a-7a7973f82957 | label-name-is-mantra-unifying-point-cloud | 2303.10585 | null | https://arxiv.org/abs/2303.10585v1 | https://arxiv.org/pdf/2303.10585v1.pdf | Label Name is Mantra: Unifying Point Cloud Segmentation across Heterogeneous Datasets | Point cloud segmentation is a fundamental task in 3D vision that serves a wide range of applications. Although great progresses have been made these years, its practical usability is still limited by the availability of training data. Existing approaches cannot make full use of multiple datasets on hand due to the label mismatch among different datasets. In this paper, we propose a principled approach that supports learning from heterogeneous datasets with different label sets. Our idea is to utilize a pre-trained language model to embed discrete labels to a continuous latent space with the help of their label names. This unifies all labels of different datasets, so that joint training is doable. Meanwhile, classifying points in the continuous 3D space by their vocabulary tokens significantly increase the generalization ability of the model in comparison with existing approaches that have fixed decoder architecture. Besides, we also integrate prompt learning in our framework to alleviate data shifts among different data sources. Extensive experiments demonstrate that our model outperforms the state-of-the-art by a large margin. | ['Yingcong Chen', 'Hao Lu', 'Shishi Xiao', 'Hao He', 'Yixun Liang'] | 2023-03-19 | null | null | null | null | ['point-cloud-segmentation'] | ['computer-vision'] | [ 1.18207783e-01 -1.42411605e-01 -5.29340625e-01 -5.02851665e-01
-9.40324366e-01 -6.53694808e-01 5.12202621e-01 -6.61665052e-02
-3.61494511e-01 2.48319685e-01 4.28564698e-02 -2.48999432e-01
2.02495173e-01 -5.46723187e-01 -6.35407865e-01 -6.25484526e-01
5.64138293e-01 5.29615819e-01 4.74328518e-01 3.06148797e-01
1.18595697e-01 3.77343237e-01 -1.39877927e+00 3.05281207e-02
8.73980403e-01 9.90067065e-01 5.71252406e-01 -1.57298282e-01
-6.05293870e-01 4.13915813e-01 -3.62731099e-01 -2.85026670e-01
4.45278436e-01 -8.61950293e-02 -5.11467338e-01 5.13995171e-01
3.99072856e-01 -4.90471452e-01 -1.99081272e-01 1.15708649e+00
3.68883342e-01 -1.21848024e-01 7.18572557e-01 -1.22345495e+00
-6.78876817e-01 2.92999178e-01 -7.75180161e-01 -3.02231938e-01
-4.00568135e-02 -1.31474093e-01 1.01028681e+00 -8.22999716e-01
4.86787111e-01 1.09489310e+00 4.89508539e-01 7.18475282e-01
-1.19273984e+00 -8.84877861e-01 5.71852684e-01 3.50155272e-02
-1.50739634e+00 -1.74939066e-01 1.06041753e+00 -5.35241127e-01
7.29441702e-01 -6.78738952e-02 4.15295303e-01 9.71273303e-01
-3.87125373e-01 9.96356845e-01 9.51195121e-01 -3.81095201e-01
2.12606266e-01 2.41212144e-01 1.15903609e-01 6.75911188e-01
9.87295806e-02 -3.21564823e-01 -4.37402695e-01 -6.75895214e-02
7.94144571e-01 2.78325796e-01 -2.26346463e-01 -8.96361411e-01
-1.22417665e+00 7.12582111e-01 3.62870693e-01 1.28006384e-01
-1.08986571e-01 3.77246514e-02 3.83644134e-01 7.85169825e-02
6.18495286e-01 6.38749972e-02 -4.22703654e-01 2.51816243e-01
-1.00263488e+00 -7.75929242e-02 4.95078146e-01 1.43885636e+00
7.90709019e-01 -3.19244951e-01 8.65917802e-02 8.39070439e-01
7.22640395e-01 5.19729495e-01 5.12942672e-01 -9.62123811e-01
6.17463946e-01 8.16651881e-01 -6.15749881e-02 -7.08861113e-01
-2.85244733e-01 -3.77620578e-01 -7.59306669e-01 1.81773841e-01
3.01131815e-01 3.99645567e-02 -1.04360211e+00 1.75839674e+00
4.39756721e-01 4.67866212e-01 -4.71695811e-02 7.70509124e-01
6.45687759e-01 6.24690115e-01 7.38767255e-03 -3.77243310e-02
1.03831542e+00 -1.05447805e+00 -5.40521502e-01 -4.19815063e-01
5.76499104e-01 -7.07519054e-01 1.01799977e+00 2.24076614e-01
-6.18736446e-01 -7.13394344e-01 -1.04407442e+00 -3.31403852e-01
-1.62409902e-01 1.88607618e-01 5.06688952e-01 5.90204954e-01
-8.15158606e-01 1.08723819e-01 -1.01254940e+00 -2.52018511e-01
6.42860174e-01 4.25950617e-01 -2.82045335e-01 -1.56411856e-01
-7.29926467e-01 5.83530784e-01 3.66947919e-01 -9.59642604e-02
-6.08879328e-01 -3.02370191e-01 -8.05443048e-01 -2.27598757e-01
6.52348220e-01 -6.67956054e-01 1.17387402e+00 -4.68233556e-01
-1.42738926e+00 1.08860159e+00 -3.82337749e-01 -7.64489472e-02
4.96412218e-01 -2.10516185e-01 5.56406286e-03 2.17648200e-03
2.37369940e-01 9.91475344e-01 8.51213336e-01 -1.47119951e+00
-8.17751527e-01 -4.73507106e-01 2.62469828e-01 3.22043061e-01
-4.69083130e-01 -3.06344628e-01 -1.02518952e+00 -4.35477912e-01
7.19520211e-01 -1.23651946e+00 -2.40843117e-01 1.86255991e-01
-4.72317845e-01 -5.04972935e-01 6.66280627e-01 -2.03810439e-01
8.82416546e-01 -2.34079027e+00 2.62994260e-01 6.99519068e-02
2.89273739e-01 3.08385845e-02 -2.13688482e-02 2.44558394e-01
3.16448331e-01 2.57221833e-02 -3.07990134e-01 -6.95205510e-01
2.17239216e-01 3.77452672e-01 -5.00772238e-01 4.73084092e-01
1.09960802e-01 7.51535416e-01 -8.54743063e-01 -6.79653883e-01
3.15511584e-01 3.41301620e-01 -5.04074574e-01 1.14747480e-01
-3.98350775e-01 6.08381450e-01 -7.55304217e-01 5.96295893e-01
7.05832541e-01 -6.25732303e-01 3.37386020e-02 -1.25442594e-01
-5.21973474e-03 3.37565660e-01 -1.17314541e+00 2.36060452e+00
-3.32484484e-01 3.56344908e-01 -2.80113995e-01 -9.87077177e-01
9.97687936e-01 3.65951002e-01 5.56838751e-01 -3.53547156e-01
-5.78944571e-02 4.54578638e-01 -2.62650073e-01 -4.97676164e-01
2.45629191e-01 -2.71315098e-01 -1.60071224e-01 4.38670814e-01
-1.77488737e-02 -2.07336009e-01 -1.40290886e-01 -6.26871362e-02
6.84133708e-01 3.40325922e-01 -9.68526024e-03 2.80493796e-01
4.57907587e-01 -8.46526995e-02 7.04190671e-01 6.13340318e-01
-2.75093019e-01 7.34899223e-01 2.56242633e-01 -7.53634125e-02
-8.65505099e-01 -9.42643166e-01 -3.35266858e-01 7.13178575e-01
5.43408036e-01 -3.55335891e-01 -4.99427050e-01 -9.30076361e-01
8.45694635e-03 5.37047446e-01 -2.37227395e-01 -5.05957846e-03
-3.87784451e-01 -3.67238611e-01 3.57362121e-01 5.01892269e-01
5.45886278e-01 -7.02660084e-01 -4.81867582e-01 1.70491170e-02
-2.23095059e-01 -1.39409387e+00 -3.29777181e-01 2.32530504e-01
-1.05201423e+00 -7.10126519e-01 -7.59362459e-01 -1.01993966e+00
6.62272215e-01 7.98748672e-01 8.59028280e-01 -1.88458711e-01
2.28608146e-01 3.32148403e-01 -3.21660399e-01 -3.13604563e-01
-2.21478567e-01 3.92811626e-01 -1.18323825e-01 -1.90721065e-01
8.48026514e-01 -4.72617269e-01 -4.07400101e-01 3.98089558e-01
-9.04030502e-01 2.62071192e-01 5.22517383e-01 6.37572765e-01
8.63368869e-01 -6.21272298e-03 5.09643912e-01 -1.01579547e+00
1.15517907e-01 -5.01072943e-01 -6.64082646e-01 3.08406860e-01
-5.10187507e-01 7.66992345e-02 2.91133612e-01 -3.10710520e-01
-8.89283955e-01 2.50656992e-01 -1.71338350e-01 -6.99090481e-01
-4.47787642e-01 4.40818489e-01 -5.27787626e-01 1.66611046e-01
2.12586507e-01 8.03894773e-02 -2.59547941e-02 -6.94646239e-01
5.83133101e-01 9.75041151e-01 3.63323539e-01 -4.64750916e-01
7.49956012e-01 5.73734701e-01 -1.79525763e-02 -5.25483787e-01
-1.02431190e+00 -7.36310363e-01 -1.00345004e+00 3.30430865e-02
9.00914550e-01 -1.24687541e+00 -1.83272898e-01 5.41196942e-01
-1.25364292e+00 -8.20346475e-02 -1.38417304e-01 5.65668702e-01
-5.17810166e-01 5.08267939e-01 -4.70961004e-01 -4.96019870e-01
1.08966462e-01 -1.40841258e+00 1.30163348e+00 5.06639481e-02
8.91814381e-02 -9.25239205e-01 -1.91236302e-01 4.68325108e-01
-4.86976020e-02 -3.29391211e-02 9.75106120e-01 -6.61804676e-01
-9.76989269e-01 -2.19556227e-01 -3.73040706e-01 3.56614143e-01
4.94514138e-01 -4.00693238e-01 -1.08433425e+00 -2.28119463e-01
3.71121407e-01 -4.18594748e-01 8.75032961e-01 2.53660232e-01
1.22249925e+00 2.90438920e-01 -6.36352360e-01 6.08735323e-01
1.33075464e+00 5.76563068e-02 1.83470607e-01 3.71697605e-01
1.07253671e+00 7.78769493e-01 5.43584526e-01 1.28129527e-01
6.71572983e-01 7.19444931e-01 5.37330985e-01 -7.17405826e-02
-9.42725912e-02 -3.26423347e-01 1.12997226e-01 1.02698672e+00
3.79127502e-01 -2.52126306e-01 -9.74548638e-01 4.72686470e-01
-1.87783325e+00 -5.68267584e-01 -8.88353959e-03 2.18136740e+00
8.59684348e-01 3.67875189e-01 -1.35997042e-01 -5.94575368e-02
7.49417126e-01 1.78073198e-01 -8.68990660e-01 3.74470174e-01
7.97517821e-02 -3.14790219e-01 5.60506821e-01 2.52681911e-01
-1.20036066e+00 9.55558717e-01 5.51220226e+00 8.30640912e-01
-1.26480019e+00 1.82952389e-01 4.24551904e-01 -5.36006875e-02
-4.01990294e-01 4.27072868e-02 -1.02278018e+00 6.15988433e-01
5.42932332e-01 1.22541934e-01 -2.98549403e-02 7.85893381e-01
-9.72471163e-02 2.52547622e-01 -1.25450861e+00 1.24023056e+00
2.87501842e-01 -1.00137305e+00 -2.42528226e-02 2.05303878e-01
7.37264335e-01 3.98401558e-01 1.73777491e-01 1.18034974e-01
7.01969936e-02 -6.18786931e-01 7.81955242e-01 2.79175252e-01
8.55935454e-01 -4.22407538e-01 5.78771412e-01 7.51482964e-01
-8.46275270e-01 1.26659095e-01 -6.78262413e-01 1.93551883e-01
3.02550107e-01 6.70596421e-01 -7.37976670e-01 5.61034918e-01
4.41280842e-01 9.30536926e-01 -4.34580207e-01 1.01076269e+00
-4.94238377e-01 5.55538118e-01 -5.27312160e-01 3.26787889e-01
2.61666775e-01 -4.00854439e-01 3.14889818e-01 8.06813776e-01
3.92278373e-01 -9.61114690e-02 5.54919183e-01 7.15293646e-01
-1.78962067e-01 7.57311061e-02 -7.08577693e-01 9.71522629e-02
6.84213042e-01 9.23834085e-01 -8.17326903e-01 -2.12021634e-01
-9.95162785e-01 9.94908750e-01 2.52274543e-01 2.58769840e-01
-6.88145578e-01 2.90211998e-02 4.51999009e-01 -9.07448903e-02
3.65036100e-01 -4.97030288e-01 -4.07340288e-01 -1.33287847e+00
1.45376012e-01 -4.80278254e-01 1.04719505e-01 -5.67278028e-01
-1.50225043e+00 4.26798314e-01 5.64373657e-02 -1.69241095e+00
-2.02141494e-01 -5.21606088e-01 6.87468797e-02 8.02402735e-01
-1.85271132e+00 -1.44702697e+00 -2.29363546e-01 4.58492130e-01
7.02673912e-01 -1.67411387e-01 7.66276121e-01 4.49060947e-01
-4.29022700e-01 4.51434433e-01 2.37072438e-01 1.42191082e-01
9.57386613e-01 -1.14198244e+00 4.68197525e-01 7.25268602e-01
5.43884575e-01 6.46369696e-01 2.79273599e-01 -6.27486706e-01
-1.10239446e+00 -1.09105170e+00 8.75309765e-01 -4.46742892e-01
4.88526613e-01 -5.21392345e-01 -1.03969622e+00 7.70429730e-01
-5.36017194e-02 8.08037892e-02 1.04283738e+00 2.00614303e-01
-5.51359713e-01 2.97983196e-02 -7.34707355e-01 5.89802206e-01
1.09251451e+00 -7.91949570e-01 -6.29212081e-01 3.02312046e-01
9.20438647e-01 -6.08228624e-01 -5.27875602e-01 3.83499980e-01
2.99412638e-01 -8.00015092e-01 8.79994929e-01 -2.23728582e-01
3.10028732e-01 -5.92968762e-01 -4.07130867e-01 -9.77998435e-01
-1.97867483e-01 -5.24178632e-02 1.02911200e-02 1.43966484e+00
2.72503793e-01 -5.39835274e-01 9.44612384e-01 7.79126465e-01
-2.11100355e-01 -6.76540434e-01 -9.55718994e-01 -7.04205513e-01
1.17231749e-01 -6.25796258e-01 6.91062450e-01 9.06949937e-01
-2.92315066e-01 6.00961804e-01 -3.17896575e-01 3.68267775e-01
6.18329048e-01 4.81678337e-01 7.91949451e-01 -1.40450418e+00
-1.35485560e-01 -2.55383283e-01 -4.68555778e-01 -1.75275111e+00
3.78711671e-01 -1.20045149e+00 1.71479970e-01 -1.68315017e+00
1.92477822e-01 -1.04012072e+00 -4.60062474e-01 6.67867601e-01
-2.08178699e-01 3.47295642e-01 3.86424989e-01 7.09635913e-01
-7.83321500e-01 6.08148992e-01 1.12317026e+00 -2.02956527e-01
-2.43774638e-01 5.75100929e-02 -6.53241217e-01 8.57159436e-01
7.13467300e-01 -6.07166231e-01 -7.38925457e-01 -1.00348353e+00
2.20309962e-02 -2.44198829e-01 2.21883431e-01 -9.74370837e-01
4.24222022e-01 -4.21958268e-02 2.07283184e-01 -7.90384293e-01
3.51836860e-01 -1.24022007e+00 5.73340524e-03 -8.07374641e-02
-3.70559841e-01 -3.04578722e-01 5.68028167e-02 7.82294393e-01
-2.92627513e-01 -3.77941251e-01 5.20077944e-01 -4.25334014e-02
-8.54888856e-01 4.40211207e-01 1.49770781e-01 -1.11100316e-01
1.00426233e+00 -2.37993479e-01 -1.87811062e-01 -4.15015332e-02
-3.19991380e-01 3.07807535e-01 8.50571156e-01 5.64229608e-01
4.85616714e-01 -1.28090346e+00 -3.59555751e-01 3.60305399e-01
3.40155333e-01 6.49405777e-01 1.05406500e-01 5.38044691e-01
-1.58039987e-01 3.63664657e-01 -8.86596292e-02 -1.08649802e+00
-9.68506455e-01 5.87541580e-01 -5.23983501e-02 1.77394133e-02
-8.41299355e-01 8.37508976e-01 4.12752360e-01 -5.44517338e-01
4.68053490e-01 -2.67726988e-01 -1.47470087e-01 1.41811594e-01
2.02944279e-01 -1.05869681e-01 -4.64742742e-02 -6.56383395e-01
-2.29081139e-01 9.86541927e-01 -2.04968125e-01 -7.72542208e-02
1.24716258e+00 -4.56993580e-01 -6.04938529e-03 8.91283989e-01
1.24954343e+00 1.06408305e-01 -1.58187461e+00 -7.04189718e-01
4.60601971e-02 -4.64683145e-01 4.88317497e-02 -4.47353661e-01
-9.98767972e-01 1.14407063e+00 5.75029910e-01 4.75789011e-02
1.02138114e+00 2.95425296e-01 8.86575818e-01 3.80694509e-01
5.73248386e-01 -8.25066924e-01 -9.69418436e-02 3.74479681e-01
2.62760073e-01 -1.57896745e+00 -5.43896779e-02 -6.72403395e-01
-6.46655858e-01 8.49551380e-01 4.13144559e-01 4.72210795e-02
6.58903062e-01 -5.84573746e-02 4.29542780e-01 -8.67790207e-02
-4.55688328e-01 -3.35053384e-01 2.71434844e-01 6.01529181e-01
2.54978806e-01 -1.66128933e-01 -1.01892360e-01 4.69068974e-01
3.86373363e-02 1.80188883e-02 2.16905311e-01 8.56920600e-01
-4.87878919e-01 -1.58589053e+00 -4.14599441e-02 3.25471789e-01
-3.13134760e-01 1.88832119e-01 -6.64947703e-02 6.02913916e-01
3.96755934e-01 9.13316607e-01 1.00439228e-02 -2.63140529e-01
2.24846929e-01 2.30964676e-01 4.00734246e-01 -9.83072996e-01
4.31404859e-02 4.04895872e-01 -4.86150146e-01 -1.33639887e-01
-8.24498534e-01 -7.88165569e-01 -1.35149789e+00 2.52771884e-01
-4.54723895e-01 -5.74571714e-02 8.17609966e-01 1.05414653e+00
4.11066711e-01 3.37210298e-01 5.75714350e-01 -7.13555574e-01
-6.60900772e-01 -6.50341272e-01 -5.74534774e-01 3.30826938e-01
3.67487609e-01 -8.97340655e-01 -1.95696592e-01 2.74854630e-01] | [8.04272174835205, -3.1417038440704346] |
e5b4682e-e3ee-4ff1-ba50-d25c16292dca | fnr-a-similarity-and-transformer-based-1 | null | null | https://scholar.google.com/citations?view_op=view_citation&hl=en&user=GQTf-pYAAAAJ&citation_for_view=GQTf-pYAAAAJ:u-x6o8ySG0sC | https://link.springer.com/article/10.1007/s13278-023-01065-0 | FNR: a similarity and transformer-based approach to detect multi-modal fake news in social media | Many people today get their news from social media. It is possible to propagate news using textual, visual, or multi-modal information. The popularity of social networks and their wide use by people make them attractive platforms for spreading fake news. Detecting fake news is essential to preventing its spread. Fake news can be a false article or a genuine article with misleading visual information. Adding an actual image to trustworthy unrelated news can also create a fake news story. In this paper, we propose a novel and efficient similarity and transformer-based detection algorithm called Fake News Revealer (FNR), which uses text and images of news to detect fake news. The algorithm uses contrastive loss to consider text and image relations and transformer models to extract contextual and semantic features. According to experiments on two public social media news data sets, the FNR algorithm competes with conventional methods and state-of-the-art fake news detection algorithms by adding a novel mechanism without adding extra parameters or weights. | ['Hamid R. Rabiee', 'Mohammad Amin Fazli', 'Maryam Ramezani', 'Faeze Ghorbanpour'] | 2023-03-28 | null | null | null | social-network-analysis-and-mining-2023-3 | ['fake-news-detection'] | ['natural-language-processing'] | [-1.06197633e-01 1.62956771e-02 -9.26740244e-02 -9.22279730e-02
-4.46439505e-01 -6.22862399e-01 1.10115314e+00 2.06956759e-01
-1.67343393e-01 7.01880932e-01 1.90092996e-01 -7.41851553e-02
5.41527212e-01 -9.45275307e-01 -7.68194258e-01 -2.84547150e-01
3.86261731e-01 3.80686760e-01 8.84401798e-01 -6.64977610e-01
6.48315549e-01 2.54259914e-01 -1.28818893e+00 9.28667784e-01
5.07022858e-01 7.42903829e-01 -1.18220195e-01 2.91074544e-01
-3.88674021e-01 1.05463171e+00 -9.71017897e-01 -1.03810978e+00
2.48887956e-01 -7.89101183e-01 -4.27882880e-01 1.87658042e-01
4.31234479e-01 -4.65471625e-01 -6.27195120e-01 1.37910926e+00
1.70588151e-01 -3.94025266e-01 6.01637363e-01 -1.42799366e+00
-1.07042611e+00 6.49703562e-01 -8.88910353e-01 3.54169369e-01
5.11190355e-01 -2.87369996e-01 4.76767480e-01 -9.24634755e-01
1.12091184e+00 1.33661366e+00 7.90129781e-01 3.26050699e-01
-7.72195160e-01 -5.28188050e-01 -3.53000969e-01 2.77415156e-01
-1.19187427e+00 -3.18818837e-01 1.04231739e+00 -4.72184122e-01
2.41871089e-01 4.18197989e-01 9.81146812e-01 1.29940057e+00
6.72409892e-01 9.47168767e-01 1.49404609e+00 -3.77179861e-01
-1.11335501e-01 7.36680806e-01 -1.68036118e-01 8.58739614e-01
5.70358634e-01 1.90586790e-01 -6.41838253e-01 -5.43923378e-01
4.15593117e-01 1.68367699e-01 -3.00608814e-01 -9.69985276e-02
-1.29366267e+00 1.22568488e+00 5.60261786e-01 4.53141272e-01
-1.45426795e-01 3.92246433e-02 3.33097756e-01 7.86902130e-01
1.19190478e+00 5.30203104e-01 2.64617920e-01 3.27808261e-01
-9.60663378e-01 2.68893540e-01 7.08072245e-01 5.33871830e-01
3.88294786e-01 -1.02930628e-01 -5.35814799e-02 7.26761580e-01
3.11956614e-01 7.99160480e-01 6.51548684e-01 -2.92735815e-01
3.77238959e-01 6.78294241e-01 4.71276402e-01 -2.20467639e+00
-1.70653448e-01 -3.98195207e-01 -5.33923328e-01 1.13618322e-01
3.52948606e-01 3.10911536e-01 -4.80388403e-01 8.57679367e-01
3.65300477e-01 1.86900184e-01 -2.05834806e-01 1.21194983e+00
8.28573167e-01 8.88682723e-01 -5.15791953e-01 -2.69072086e-01
1.54263139e+00 -1.05133617e+00 -1.04841650e+00 -2.36599848e-01
4.97057140e-01 -1.46506143e+00 5.89762211e-01 2.99645692e-01
-7.72197127e-01 1.74488649e-01 -9.82222557e-01 1.01021372e-01
-7.78677881e-01 1.07350394e-01 3.03238750e-01 7.17896044e-01
-5.48796833e-01 3.08338344e-01 -2.60999143e-01 -3.22387546e-01
5.12825191e-01 -2.51639098e-01 -3.94738197e-01 -2.56440878e-01
-1.39316452e+00 1.14575660e+00 1.54288113e-01 -1.55186176e-01
-5.02549887e-01 9.06972364e-02 -4.31027591e-01 -4.61162299e-01
5.37256896e-01 -5.66560388e-01 7.76706219e-01 -1.60456455e+00
-1.25095749e+00 1.11415946e+00 3.04383039e-01 -3.70364606e-01
1.29906368e+00 -7.41557777e-02 -8.73063087e-01 4.74508286e-01
3.23949516e-01 1.51592702e-01 1.66077685e+00 -1.45448852e+00
-3.14237714e-01 -3.17718953e-01 -3.27715456e-01 1.91439204e-02
-1.39769629e-01 3.35653096e-01 -1.89474188e-02 -1.23583937e+00
5.77718437e-01 -8.38116586e-01 3.18398088e-01 2.01682895e-01
-6.90113366e-01 2.61170208e-01 1.37823462e+00 -7.37548411e-01
7.37923384e-01 -1.94345176e+00 -3.46580386e-01 1.65386423e-01
6.60008371e-01 2.33195305e-01 1.14590921e-01 4.06171024e-01
4.23368424e-01 3.98473144e-02 8.52072760e-02 -6.93752319e-02
-3.47313732e-01 -7.17720538e-02 -3.38912934e-01 1.16799319e+00
-2.26793423e-01 9.74272668e-01 -1.19081366e+00 -5.22096872e-01
-3.62332501e-02 4.53523874e-01 -2.27079690e-02 -3.74642581e-01
-9.03321207e-02 1.61055997e-01 -5.62106729e-01 4.96149600e-01
8.16970289e-01 -4.14060265e-01 -2.77329236e-01 -2.03262553e-01
-6.74495250e-02 1.42583355e-01 -6.71616554e-01 6.80830359e-01
2.35192273e-02 1.10584021e+00 -1.24464855e-01 -9.32182670e-01
9.46499050e-01 2.94575989e-01 -6.90613315e-02 -6.39841199e-01
6.31428123e-01 5.50543487e-01 -5.32129169e-01 -5.35057664e-01
8.75430226e-01 -3.20023835e-01 -8.53477269e-02 7.50503600e-01
-3.37743253e-01 -1.52099743e-01 -4.16186810e-01 4.83592659e-01
9.18770492e-01 -2.55380541e-01 3.03524762e-01 2.32104044e-02
1.76093102e-01 3.35940212e-01 2.85847057e-02 9.08981919e-01
-2.18346611e-01 4.04134512e-01 3.64185095e-01 -7.25541413e-01
-1.28128994e+00 -6.48781657e-01 1.57416835e-01 6.95493996e-01
7.78409958e-01 3.40017080e-02 -5.09047270e-01 -9.36769485e-01
9.41990763e-02 4.81114745e-01 -5.95570207e-01 -2.02475160e-01
-1.83303908e-01 -8.46440554e-01 5.60413897e-01 -3.87988269e-01
8.23798954e-01 -7.89745748e-01 -1.27582148e-01 2.35714793e-01
-6.31621420e-01 -1.12849700e+00 -4.53615278e-01 -8.16416085e-01
-4.97454911e-01 -1.02705336e+00 -9.34729576e-01 -5.77508450e-01
8.45989108e-01 1.05310941e+00 7.93606222e-01 4.15777653e-01
-2.70909876e-01 5.37048429e-02 -8.12915862e-01 -2.20600441e-01
-1.08450460e+00 -4.29790378e-01 4.73985542e-03 5.02315462e-01
1.51897579e-01 9.70137268e-02 -3.60016793e-01 7.24814951e-01
-1.05491149e+00 2.17443138e-01 5.20206213e-01 9.22908068e-01
-3.65148745e-02 1.58618003e-01 5.34009814e-01 -1.11513436e+00
8.54578853e-01 -7.63289750e-01 -6.13884151e-01 2.07507581e-01
-5.82211733e-01 -3.62662196e-01 4.78981584e-01 -6.52540982e-01
-8.47272992e-01 -2.86363214e-01 5.04247606e-01 -1.88337684e-01
2.22566098e-01 3.71556431e-01 5.73364675e-01 -4.72250223e-01
9.27322328e-01 4.08706456e-01 1.87324911e-01 -1.76388621e-01
3.40583026e-01 8.58648837e-01 1.72202826e-01 2.74614066e-01
1.11630976e+00 1.15931237e+00 -2.57913172e-01 -8.91484797e-01
-1.07131732e+00 -5.60977459e-01 6.98861405e-02 -5.56821644e-01
4.67198282e-01 -6.04704678e-01 -3.70269835e-01 8.33585620e-01
-1.58395994e+00 6.20048463e-01 3.51507783e-01 4.73260671e-01
-5.76972887e-02 6.60159051e-01 -8.20929408e-01 -7.66681194e-01
-1.05019286e-02 -7.80085206e-01 7.00848758e-01 -4.72429007e-01
1.55722409e-01 -7.66387820e-01 -1.58083007e-01 8.83206308e-01
6.91976845e-01 3.55728745e-01 7.15067610e-02 -8.83662522e-01
-6.45254135e-01 -7.75705099e-01 -6.40348434e-01 2.05763131e-01
3.74109694e-03 -1.27406225e-01 -6.01256549e-01 -1.03831984e-01
2.29824319e-01 -9.26970169e-02 1.01587212e+00 1.11891981e-02
1.88338831e-01 -1.05125463e+00 -5.45866787e-01 -9.63607132e-02
1.31768084e+00 -2.05137298e-01 7.25971580e-01 5.74673355e-01
7.16594696e-01 6.32711709e-01 5.77295542e-01 3.98107022e-01
1.06143966e-01 8.41987848e-01 4.14348096e-01 3.56164500e-02
-4.75599132e-02 -2.18518496e-01 4.89039868e-01 7.61563361e-01
2.56875344e-02 -6.58771515e-01 -6.42069459e-01 3.88954610e-01
-1.83566749e+00 -1.57938421e+00 -6.53321028e-01 1.92192209e+00
4.02899146e-01 2.87156165e-01 7.37586170e-02 1.21717989e-01
1.36182427e+00 2.90906727e-01 5.99844614e-03 -1.74886376e-01
-6.01079881e-01 -8.45806837e-01 8.69389176e-01 4.77515429e-01
-1.11913049e+00 1.08416045e+00 5.70457792e+00 1.14789498e+00
-1.22916186e+00 6.70243919e-01 6.83695436e-01 1.91935882e-01
-4.44516897e-01 -2.33636990e-01 -3.29780012e-01 7.06740916e-01
2.35083461e-01 3.86930034e-02 3.93977910e-01 6.91688776e-01
3.61201108e-01 -2.99557298e-01 -1.48191288e-01 9.37337160e-01
6.85929596e-01 -1.83806729e+00 2.45244622e-01 -1.81928962e-01
9.28841770e-01 2.99466569e-02 7.38081941e-03 -4.16602373e-01
1.98162813e-03 -5.89786947e-01 1.14249265e+00 4.27495360e-01
3.15572411e-01 -6.11518025e-01 1.19172287e+00 2.73866236e-01
-4.57038313e-01 2.57946134e-01 -2.85403281e-01 -1.33148640e-01
3.69097680e-01 8.82761300e-01 -6.80657744e-01 2.12289944e-01
4.92074281e-01 7.15115666e-01 -6.18365169e-01 8.77563357e-01
-2.60220826e-01 4.10925001e-01 -4.08119619e-01 -5.49600959e-01
3.36327493e-01 -9.17631686e-02 1.22513771e+00 9.85000670e-01
3.55089784e-01 -7.54752308e-02 -1.78253502e-01 7.84326255e-01
-1.49091110e-01 2.50697821e-01 -9.04896975e-01 -2.95093209e-01
2.60937393e-01 9.33363318e-01 -1.34010589e+00 -5.75660527e-01
-4.11564529e-01 1.55268645e+00 -1.70501098e-01 -4.05386686e-02
-9.55915928e-01 -2.58651614e-01 -2.38919809e-01 7.17198193e-01
5.82102351e-02 2.88658321e-01 6.64149970e-02 -1.54081976e+00
8.41692090e-02 -7.34790802e-01 -1.06132738e-01 -9.59712029e-01
-1.65815043e+00 8.43040049e-01 -2.68389195e-01 -1.63639283e+00
3.81967008e-01 -2.08514810e-01 -2.34868869e-01 6.01869784e-02
-1.31036246e+00 -1.30594933e+00 -2.56357342e-01 5.60858250e-01
1.92868173e-01 -5.08573651e-01 2.17806578e-01 2.16779977e-01
4.55908291e-02 3.05334538e-01 3.89312625e-01 1.36555120e-01
7.98536777e-01 -6.31524205e-01 2.78910667e-01 7.20455408e-01
1.93990678e-01 8.47593099e-02 1.29877448e+00 -9.30954635e-01
-9.24463630e-01 -8.26989532e-01 1.09823048e+00 -4.44537282e-01
9.79708314e-01 -1.90596417e-01 -6.64088726e-01 3.93842489e-01
1.08492970e-01 1.81814954e-01 3.06384057e-01 -6.53837383e-01
-7.84344852e-01 2.96049207e-01 -1.63627493e+00 5.61366737e-01
6.79389000e-01 -3.62382293e-01 -7.20321655e-01 8.05827379e-01
5.97722113e-01 -2.21221700e-01 -1.07121050e-01 -2.59332955e-01
6.46936655e-01 -1.34725249e+00 8.70266497e-01 -2.53128648e-01
7.75923073e-01 -3.50765914e-01 2.69607216e-01 -1.40389645e+00
-1.23982944e-01 -7.04896748e-01 2.37161294e-01 7.14045763e-01
4.00473177e-01 -1.00518835e+00 5.57376385e-01 -2.77455002e-01
1.49727270e-01 -1.22139126e-01 -9.21401262e-01 -8.80400658e-01
-6.22928739e-01 -5.40322140e-02 2.30340317e-01 1.70626223e+00
1.77766755e-01 2.83804178e-01 -1.06833100e+00 7.52044097e-02
7.97805071e-01 6.28741831e-02 6.26278698e-01 -1.15837371e+00
-1.94165975e-01 -3.76904696e-01 -8.69466364e-01 -7.33385921e-01
-2.93600291e-01 -4.38691974e-01 -1.91577479e-01 -1.27367640e+00
1.67459771e-01 -1.92759633e-01 2.78066754e-01 1.73302352e-01
2.56458700e-01 1.01558554e+00 6.88710809e-02 7.52024591e-01
-6.35248899e-01 3.52145523e-01 1.60277486e+00 -3.51464868e-01
1.21955335e-01 4.88665141e-02 -2.94611216e-01 9.19653535e-01
7.95605838e-01 -1.10514629e+00 7.34853223e-02 -8.17011297e-02
9.03954685e-01 1.19065553e-01 7.75098085e-01 -5.58097184e-01
1.91440910e-01 -2.06528492e-02 8.58348161e-02 -4.51792419e-01
3.69631618e-01 -7.35932589e-01 9.94138047e-02 7.56675005e-01
-9.00204927e-02 5.42723760e-02 -2.56181598e-01 1.09771466e+00
-3.85408580e-01 -1.36822686e-01 9.32568491e-01 -4.45329458e-01
-2.73729444e-01 -2.56876290e-01 -8.56934786e-01 -1.53441191e-01
1.15730774e+00 -3.30374569e-01 -1.03387189e+00 -9.27073777e-01
-5.69454193e-01 -5.14361501e-01 6.10224783e-01 5.62948883e-01
9.66605961e-01 -1.33421159e+00 -1.06081390e+00 -1.32866055e-01
4.00433421e-01 -8.62141728e-01 1.51919229e-02 9.66746092e-01
-1.07376444e+00 1.97768658e-02 -1.55958146e-01 -1.19570218e-01
-1.40098131e+00 6.70174420e-01 7.31135458e-02 1.02914616e-01
-6.58392489e-01 8.08720887e-01 -3.19229126e-01 -9.95787084e-02
-4.80217904e-01 2.24454552e-01 -5.96296974e-02 1.73394531e-01
7.41458476e-01 5.25852561e-01 -6.88768029e-02 -1.29187381e+00
-2.55060941e-01 1.59468204e-01 -1.83798075e-01 -1.59189567e-01
1.13841021e+00 -3.37735444e-01 -5.47725379e-01 3.28751266e-01
1.16071081e+00 4.62171167e-01 -3.98368269e-01 -2.59947270e-01
-1.78731933e-01 -9.96047676e-01 3.83237481e-01 -7.96162367e-01
-1.15242410e+00 1.73430845e-01 2.13770196e-01 9.19354856e-01
4.11439806e-01 3.48142564e-01 1.04279697e+00 1.12584308e-01
4.92703915e-01 -1.07156694e+00 4.21703964e-01 8.62808228e-02
1.06023431e+00 -1.61323440e+00 3.95274431e-01 -7.76359439e-01
-6.63759530e-01 9.90392745e-01 -2.49966178e-02 -4.29488063e-01
6.97704792e-01 -1.31128654e-01 2.57864386e-01 -6.02095306e-01
-2.50890195e-01 3.73139679e-01 4.12677556e-01 2.13291913e-01
-8.43133032e-02 1.16700277e-01 -8.55329871e-01 2.21690498e-02
-2.00729351e-02 -1.59680769e-01 1.02699232e+00 8.53870451e-01
-8.16669524e-01 -6.51282191e-01 -9.48926091e-01 5.46831846e-01
-7.76239634e-01 -4.74411957e-02 -7.62867928e-01 6.46163583e-01
1.15856394e-01 1.04849863e+00 -1.80874005e-01 -4.19116944e-01
-3.18770140e-01 -4.61680949e-01 2.90314049e-01 -2.14486495e-01
-5.17725289e-01 1.71302296e-02 2.32616514e-01 -3.92589688e-01
-7.35427201e-01 -3.77198011e-01 -6.62385225e-01 -7.08081365e-01
-1.03257716e+00 8.15475062e-02 9.65264738e-01 1.09331298e+00
2.16742322e-01 5.56456298e-03 6.87099278e-01 -6.12674534e-01
-1.18186757e-01 -7.73469925e-01 -5.28014779e-01 6.41483128e-01
3.39787453e-01 -8.16201925e-01 -9.07455087e-01 4.45645675e-02] | [8.137957572937012, 10.25755500793457] |
58ef7180-30df-436b-9694-86253fe5e39a | sign-language-production-with-avatar-layering | null | null | https://aclanthology.org/2022.lrec-1.163 | https://aclanthology.org/2022.lrec-1.163.pdf | Sign Language Production With Avatar Layering: A Critical Use Case over Rare Words | Sign language production (SLP) is the process of generating sign language videos from spoken language expressions. Since sign languages are highly under-resourced, existing vision-based SLP approaches suffer from out-of-vocabulary (OOV) and test-time generalization problems and thus generate low-quality translations. To address these problems, we introduce an avatar-based SLP system composed of a sign language translation (SLT) model and an avatar animation generation module. Our Transformer-based SLT model utilizes two additional strategies to resolve these problems: named entity transformation to reduce OOV tokens and context vector generation using a pretrained language model (e.g., BERT) to reliably train the decoder. Our system is validated on a new Korean-Korean Sign Language (KSL) dataset of weather forecasts and emergency announcements. Our SLT model achieves an 8.77 higher BLEU-4 score and a 4.57 higher ROUGE-L score over those of our baseline model. In a user evaluation, 93.48% of named entities were successfully identified by participants, demonstrating marked improvement on OOV issues. | ['Jong Park', 'Du Hui Lee', 'Sukmin Cho', 'Eui Jun Hwang', 'Jung-Ho Kim'] | null | null | null | null | lrec-2022-6 | ['sign-language-translation', 'sign-language-production'] | ['computer-vision', 'natural-language-processing'] | [ 8.04691166e-02 1.65651768e-01 -1.37675256e-01 -3.34151089e-01
-1.35858893e+00 -7.21019626e-01 6.72399580e-01 -8.44669640e-01
-4.86573726e-01 6.33064747e-01 7.63579369e-01 -1.85889274e-01
8.57710660e-01 -2.91803479e-01 -6.87693536e-01 -2.85087347e-01
3.40022802e-01 5.55072725e-01 2.85057932e-01 -3.13006818e-01
1.04072124e-01 1.12378351e-01 -1.34887028e+00 4.93075550e-01
1.16244113e+00 4.98066723e-01 2.22324394e-03 9.19386387e-01
-3.40012401e-01 9.66557503e-01 -8.11570168e-01 -4.68889713e-01
1.78477392e-01 -8.09007466e-01 -7.89068520e-01 -1.41946003e-01
7.05314815e-01 -6.83196783e-01 -2.19994232e-01 8.40280950e-01
8.38225007e-01 -7.67379403e-02 6.14966810e-01 -1.61253834e+00
-8.90884042e-01 5.06437778e-01 -2.50651538e-01 -4.66166288e-01
5.96231282e-01 5.60961723e-01 1.14996552e+00 -1.26003575e+00
1.34054005e+00 1.18966782e+00 4.76288915e-01 1.13181925e+00
-7.84308970e-01 -8.59979570e-01 1.00564025e-01 7.84974620e-02
-1.33957982e+00 -4.43345368e-01 3.14933151e-01 -3.92961115e-01
1.10067499e+00 8.37526545e-02 7.93082356e-01 1.38216329e+00
-1.64287865e-01 1.27238107e+00 1.13289475e+00 -4.89741713e-01
5.85767590e-02 2.74245963e-02 -1.43745095e-01 8.90940309e-01
-5.82419746e-02 4.55868691e-02 -9.68973577e-01 6.95826486e-02
8.49680603e-01 -7.94030428e-01 -3.99348706e-01 -2.84809291e-01
-1.51756644e+00 7.07012117e-01 2.45288182e-02 -2.97560915e-02
-3.24160695e-01 2.24968061e-01 4.31643337e-01 2.60891378e-01
-5.06518409e-02 3.90311360e-01 -2.87123650e-01 -6.29239440e-01
-8.05176556e-01 3.98277491e-01 1.00658607e+00 1.35228801e+00
-1.04344718e-01 3.21476102e-01 -5.02269804e-01 7.37204731e-01
6.09781742e-01 1.09880483e+00 6.06830537e-01 -8.05618227e-01
6.25943482e-01 3.00764859e-01 2.17949912e-01 -4.55934793e-01
-9.84838232e-02 2.31772333e-01 -2.53902406e-01 3.35875481e-01
5.44927657e-01 -2.97519922e-01 -1.50966501e+00 1.75360060e+00
1.63682923e-01 1.56905159e-01 3.24276596e-01 1.17356491e+00
1.25225472e+00 7.85166562e-01 4.38158184e-01 1.70484886e-01
1.30117607e+00 -1.23672462e+00 -6.01069868e-01 -1.74789071e-01
6.55939043e-01 -8.15892816e-01 1.39778292e+00 1.61681563e-01
-9.50655222e-01 -2.33300388e-01 -6.63935781e-01 -8.97208452e-02
-1.58980101e-01 4.73153293e-01 2.09672943e-01 4.08433259e-01
-1.19607830e+00 -1.86448574e-01 -7.82378376e-01 -6.23330235e-01
4.88679856e-02 -2.15527304e-02 -4.16478902e-01 3.52512300e-02
-1.09128356e+00 1.04325438e+00 5.37849255e-02 8.12433474e-03
-8.18121076e-01 -5.51484108e-01 -9.58554327e-01 -3.93715382e-01
6.74494058e-02 -5.98134160e-01 1.75847352e+00 -9.65209842e-01
-1.87630963e+00 8.94580364e-01 -3.52414608e-01 -2.29515210e-01
8.42308581e-01 -2.90319771e-01 -5.37075341e-01 2.05152154e-01
1.19085982e-01 9.79908109e-01 7.39653766e-01 -1.06932938e+00
-7.73969829e-01 2.68689930e-01 -2.63445437e-01 5.32320797e-01
2.54046440e-01 3.93079996e-01 -6.98080659e-01 -7.88670301e-01
-8.54157060e-02 -1.08005965e+00 1.51813403e-02 2.20587566e-01
-1.58260614e-01 -2.47092664e-01 8.00852597e-01 -1.23145401e+00
1.07879591e+00 -1.82958519e+00 -1.57821365e-02 4.52866405e-02
8.32711607e-02 6.14669263e-01 -7.32038796e-01 4.37121719e-01
3.47125769e-01 -1.44608114e-02 -1.19391106e-01 -1.24488115e-01
8.03834498e-02 2.32006013e-01 -4.30975378e-01 -7.94521943e-02
3.67768556e-01 1.23341691e+00 -1.14147019e+00 -5.95564842e-01
1.23972803e-01 6.88831449e-01 -5.27298450e-01 1.77261814e-01
-4.04351622e-01 6.09175265e-01 -2.39581913e-01 9.82710838e-01
2.28275195e-01 -8.12176839e-02 1.71925351e-01 -3.63310426e-02
-2.00501189e-01 3.28326672e-01 -9.20858979e-01 1.55942166e+00
-5.21307588e-01 1.07665861e+00 -2.75833488e-01 -2.10037291e-01
9.32490587e-01 5.35679102e-01 2.21481159e-01 -7.46302307e-01
1.73478816e-02 4.93850231e-01 -3.03904295e-01 -8.21390510e-01
5.75775564e-01 1.71354916e-02 -2.57032633e-01 3.83735031e-01
1.99892178e-01 -1.88988060e-01 2.44603634e-01 2.71273762e-01
1.01503599e+00 4.99437720e-01 1.86626837e-01 3.21812391e-01
3.35220933e-01 3.27828676e-01 4.68145758e-01 6.49013460e-01
-5.30615509e-01 7.54199445e-01 2.41740495e-01 -9.38472599e-02
-1.06565535e+00 -1.10742295e+00 4.39210206e-01 9.02042806e-01
-1.75360478e-02 -5.23449302e-01 -7.35006988e-01 -9.00661826e-01
-2.14145362e-01 1.02201653e+00 -4.07963470e-02 8.94360691e-02
-9.02931869e-01 -1.06852017e-01 1.08443046e+00 7.74095297e-01
5.02458751e-01 -1.45383286e+00 -6.87701046e-01 1.69247717e-01
-6.99073732e-01 -1.47381961e+00 -8.90554726e-01 -6.75171971e-01
-3.64793926e-01 -1.01223814e+00 -1.17128468e+00 -1.25831139e+00
7.83757865e-01 -1.84141695e-02 9.47129846e-01 -3.69714946e-01
-6.24121130e-02 5.05805194e-01 -7.39480615e-01 -5.75328290e-01
-9.01680350e-01 -3.40719193e-01 -1.83901023e-02 -2.25004449e-01
6.73382282e-01 1.03866644e-01 -2.56789535e-01 3.11753660e-01
-5.39057076e-01 5.64706981e-01 7.09071577e-01 1.17682898e+00
4.56441849e-01 -1.03802037e+00 3.08586180e-01 -7.22704902e-02
6.71142101e-01 2.73089737e-01 -7.54655004e-01 5.65404475e-01
-5.38074732e-01 1.78806067e-01 2.76728034e-01 -7.96553850e-01
-1.12577617e+00 2.29008257e-01 -1.82711869e-01 -1.47841960e-01
-1.72201032e-03 3.51015359e-01 -1.10551417e-01 -1.32407889e-01
3.93894523e-01 5.84267855e-01 1.06101945e-01 -1.91676810e-01
6.35488868e-01 1.05685794e+00 6.67613626e-01 -1.68774918e-01
9.15548086e-01 5.27534224e-02 -5.15029669e-01 -1.02925038e+00
-3.11574608e-01 -2.54999548e-01 -4.45784062e-01 -5.88654578e-01
9.47495162e-01 -1.17358637e+00 -5.49000919e-01 8.23646545e-01
-1.35769737e+00 -6.51070952e-01 -3.01213354e-01 7.03196585e-01
-5.63161314e-01 4.13263708e-01 -4.97248381e-01 -6.96798921e-01
-5.72501123e-01 -1.15556812e+00 1.27910721e+00 3.60775560e-01
-5.59529305e-01 -3.17373991e-01 1.96836725e-01 5.39148331e-01
3.70812595e-01 2.26021886e-01 3.92254233e-01 -4.35146302e-01
-6.83646917e-01 -2.08669424e-01 -3.47735167e-01 4.01651502e-01
-1.51401535e-01 1.06859356e-02 -5.91284037e-01 -1.55917510e-01
-8.39035988e-01 -4.48998362e-01 5.88016748e-01 1.48979008e-01
1.66031331e-01 -3.71154904e-01 -9.99533460e-02 4.78592843e-01
9.61444378e-01 2.57607639e-01 4.75489020e-01 1.42777950e-01
7.78865159e-01 3.26707602e-01 7.44859099e-01 3.63198787e-01
7.76047945e-01 8.94967675e-01 -1.99661434e-01 -7.09312689e-03
-7.00050354e-01 -9.85779643e-01 1.00356352e+00 1.07214010e+00
-2.43051574e-01 -3.08144003e-01 -1.06687260e+00 8.08887362e-01
-1.78456020e+00 -9.23612654e-01 -2.64268845e-01 1.93574691e+00
9.68753576e-01 -3.20280433e-01 1.98931292e-01 -3.92648131e-01
4.41149414e-01 9.87428501e-02 -3.89495373e-01 -2.90771902e-01
-2.64906913e-01 7.86964223e-02 5.16985714e-01 6.15845501e-01
-9.56467450e-01 1.61939752e+00 6.13708925e+00 3.20736825e-01
-1.35440218e+00 7.82645196e-02 -1.81840599e-01 -6.94188476e-02
-2.05456942e-01 -5.59944250e-02 -9.32582200e-01 3.65871519e-01
7.79131472e-01 -2.35822767e-01 1.99983120e-01 8.68001401e-01
5.22397578e-01 1.97988093e-01 -7.40513027e-01 1.18402934e+00
4.92282063e-01 -1.18670022e+00 4.00860727e-01 -1.06794238e-01
6.54454112e-01 3.72791648e-01 -2.75881231e-01 5.40998638e-01
5.70834935e-01 -7.82899857e-01 1.05087698e+00 5.00091374e-01
1.42280006e+00 -2.61241734e-01 5.88259339e-01 7.55147338e-02
-1.25899422e+00 2.42550209e-01 2.02702358e-01 3.21541876e-01
9.40239549e-01 -2.42126599e-01 -1.17794240e+00 3.80590372e-02
4.73830909e-01 3.22710246e-01 -2.50460893e-01 1.08819318e+00
-8.49437237e-01 9.18936551e-01 -3.76059949e-01 -5.12550414e-01
2.62009919e-01 2.42924094e-02 8.58439147e-01 1.40025413e+00
4.21563864e-01 8.68616998e-02 1.16988666e-01 6.42824411e-01
-5.61835282e-02 3.13106507e-01 -5.54846287e-01 -3.90713930e-01
3.90189439e-01 8.34791958e-01 -3.78240556e-01 -5.92846692e-01
-5.52719235e-01 1.44919026e+00 -1.84406549e-01 5.65454483e-01
-1.09830344e+00 -5.26781619e-01 8.63651037e-01 -8.57881606e-02
3.55258435e-01 -4.42134351e-01 4.13127318e-02 -1.56055856e+00
2.22082287e-01 -1.22677565e+00 7.66409561e-02 -1.12204754e+00
-9.78349447e-01 6.79093421e-01 -9.85857919e-02 -1.67589355e+00
-7.50749946e-01 -6.06457412e-01 -2.99119741e-01 7.82152951e-01
-1.47958601e+00 -1.70226467e+00 -5.25730431e-01 3.51021141e-01
8.73194814e-01 -2.77162969e-01 9.07186091e-01 1.85642242e-01
-2.74252743e-02 9.06863272e-01 -2.75431216e-01 7.02528298e-01
9.24998522e-01 -9.59361315e-01 8.61101747e-01 1.18216920e+00
3.50190103e-01 2.54585654e-01 5.46643674e-01 -9.02931929e-01
-1.19966745e+00 -1.11048877e+00 1.69650531e+00 -6.34364426e-01
6.53418839e-01 -2.05003425e-01 -4.08074498e-01 6.22556329e-01
8.13246369e-02 -2.68382192e-01 4.88023520e-01 -6.30277276e-01
-6.35010540e-01 4.38261479e-01 -9.19558704e-01 1.06851268e+00
1.33448792e+00 -6.20012522e-01 -7.62764931e-01 1.97171733e-01
6.12250686e-01 -8.29535007e-01 -4.73689795e-01 2.42928907e-01
1.12510026e+00 -3.72887522e-01 5.58160305e-01 -7.38003433e-01
4.22211856e-01 -5.92985749e-01 -1.29073523e-02 -1.22810078e+00
1.23390123e-01 -8.29218507e-01 -1.44972235e-01 1.09700954e+00
6.85389400e-01 -3.50097656e-01 6.17983103e-01 8.52039397e-01
-5.98989874e-02 -2.29189128e-01 -8.48644495e-01 -9.59434330e-01
-2.66782552e-01 -4.27486837e-01 3.66152734e-01 7.95647025e-01
2.21523836e-01 4.85242039e-01 -5.40624857e-01 3.94488387e-02
4.49222594e-01 -8.03682655e-02 1.11924148e+00 -7.56563425e-01
-1.05755746e-01 -4.30867046e-01 -3.83801043e-01 -1.39773190e+00
1.11805297e-01 -9.01474535e-01 5.20754516e-01 -1.83864820e+00
-4.44962718e-02 8.58082995e-02 2.86086500e-01 8.37671459e-01
2.73708496e-02 2.51951963e-01 6.53519392e-01 2.76497662e-01
-4.31499749e-01 4.76818502e-01 1.36214137e+00 -1.45677567e-01
-5.05070746e-01 -2.36571655e-01 -3.13635051e-01 6.54700398e-01
6.35336995e-01 -1.84899136e-01 -7.67795071e-02 -7.02372551e-01
-2.27050483e-01 4.56578173e-02 4.55240309e-01 -7.11410582e-01
2.09380522e-01 -2.23143175e-01 -1.27119049e-01 -5.88886797e-01
2.01075241e-01 -3.08454752e-01 -2.02476308e-01 4.64741826e-01
-3.89345378e-01 4.82111201e-02 5.17149717e-02 2.36719221e-01
-3.40534300e-01 1.30503371e-01 6.43216491e-01 3.47442320e-03
-1.31577945e+00 -7.49778152e-02 -7.11171567e-01 2.28128895e-01
8.46561193e-01 -2.66934156e-01 -3.56956154e-01 -8.66202950e-01
-4.27314758e-01 3.34987521e-01 3.21899295e-01 8.95929694e-01
9.56496000e-01 -1.47283614e+00 -1.01059282e+00 2.53903866e-01
5.59018552e-01 -2.05523863e-01 -1.04863577e-01 7.63328373e-01
-8.86209369e-01 3.65031093e-01 -2.16784000e-01 -3.27881396e-01
-1.74048436e+00 -3.18530977e-01 2.34807327e-01 -1.25308663e-01
-9.57331240e-01 1.07871079e+00 -2.57404834e-01 -5.92397451e-01
3.10459465e-01 -5.33039033e-01 -7.47141754e-03 -1.96210682e-01
6.58685684e-01 1.06212303e-01 -4.24932867e-01 -1.11817920e+00
-4.27007407e-01 5.86800992e-01 1.08572513e-01 -8.61888647e-01
9.39855695e-01 1.49717346e-01 4.17170703e-01 8.87760520e-02
8.72827530e-01 1.44263923e-01 -1.07167923e+00 -1.47375375e-01
5.05117178e-02 -1.93389401e-01 -3.04928184e-01 -1.23835146e+00
-6.69319868e-01 5.46781480e-01 4.86743569e-01 -7.76264369e-01
8.48155856e-01 3.38317305e-02 1.27158415e+00 4.58169878e-01
4.84674722e-01 -1.38454223e+00 -7.91617855e-02 9.09177184e-01
1.17395413e+00 -1.29319406e+00 -6.12306476e-01 -1.59280971e-01
-1.36913741e+00 9.87194419e-01 6.75341427e-01 1.34185642e-01
2.15321556e-01 3.41720492e-01 8.58033419e-01 8.58421028e-02
-5.24390280e-01 -3.41875106e-01 4.61137980e-01 8.65588903e-01
4.07448679e-01 2.03412563e-01 -7.03436494e-01 4.91722792e-01
-3.55531067e-01 4.11084384e-01 6.04739845e-01 1.03201842e+00
-2.12312892e-01 -1.02036154e+00 -1.76998809e-01 5.26906401e-02
1.42775014e-01 -2.68025935e-01 -7.19221711e-01 9.13148046e-01
-2.60277003e-01 8.24255586e-01 -1.25128910e-01 -4.83353198e-01
6.26412451e-01 4.77062136e-01 4.08487737e-01 -4.06407267e-01
-3.87404740e-01 4.35373466e-03 4.96742636e-01 -7.98919201e-01
-5.16026974e-01 -7.72780955e-01 -1.52002180e+00 1.24047667e-01
-2.22073898e-01 -2.02617377e-01 6.87379897e-01 8.78627241e-01
4.08771902e-01 1.07831120e-01 1.14474893e-01 -4.86647964e-01
-6.94257677e-01 -9.68548834e-01 -1.59065187e-01 5.90980113e-01
8.40635151e-02 -2.81643003e-01 -3.41914520e-02 4.61708575e-01] | [9.19657039642334, -6.520587921142578] |
07046d1c-7b01-4bf2-ac63-53ff93c2a543 | solving-learn-to-race-autonomous-racing | 2207.01275 | null | https://arxiv.org/abs/2207.01275v2 | https://arxiv.org/pdf/2207.01275v2.pdf | Solving Learn-to-Race Autonomous Racing Challenge by Planning in Latent Space | Learn-to-Race Autonomous Racing Virtual Challenge hosted on www<dot>aicrowd<dot>com platform consisted of two tracks: Single and Multi Camera. Our UniTeam team was among the final winners in the Single Camera track. The agent is required to pass the previously unknown F1-style track in the minimum time with the least amount of off-road driving violations. In our approach, we used the U-Net architecture for road segmentation, variational autocoder for encoding a road binary mask, and a nearest-neighbor search strategy that selects the best action for a given state. Our agent achieved an average speed of 105 km/h on stage 1 (known track) and 73 km/h on stage 2 (unknown track) without any off-road driving violations. Here we present our solution and results. | ['Andrew Melnik', 'Helge Ritter', 'Rahul Kala', 'Fabian Heinrich', 'Shivansh Beohar'] | 2022-07-04 | null | null | null | null | ['road-segementation'] | ['computer-vision'] | [-2.15801701e-01 1.49648115e-01 -3.49892318e-01 -3.32951695e-01
-7.80848444e-01 -7.99341738e-01 3.53769362e-01 -3.17196906e-01
-7.61280179e-01 8.99170637e-01 -3.96531671e-01 -5.44856548e-01
-3.92406255e-01 -7.21295297e-01 -9.72669661e-01 -4.40247327e-01
2.42967941e-02 6.16189718e-01 5.31226516e-01 -3.11373025e-01
2.20825106e-01 4.73079652e-01 -1.58111453e+00 -1.92433760e-01
9.64717150e-01 7.30795503e-01 1.88027337e-01 1.26823282e+00
3.99365067e-01 9.42940891e-01 -9.66925249e-02 -3.72849822e-01
6.25539124e-01 -2.55160064e-01 -7.18929112e-01 -1.44583821e-01
6.81189537e-01 -2.71229774e-01 -9.16883171e-01 9.75744784e-01
6.90971315e-02 7.45077610e-01 4.86450911e-01 -1.65607166e+00
8.15532878e-02 6.20208502e-01 -5.07248282e-01 5.69230497e-01
-2.09732100e-01 8.12128961e-01 6.97502255e-01 -3.40366572e-01
6.67180896e-01 9.55702066e-01 4.83612895e-01 5.21855652e-01
-8.88358116e-01 -6.90699995e-01 3.13171417e-01 4.30314183e-01
-1.46330702e+00 -4.85677451e-01 2.48413458e-01 -6.33393824e-01
9.28286254e-01 -1.18679896e-01 8.01128387e-01 9.26903784e-01
5.70401669e-01 7.67548859e-01 1.00185680e+00 4.95217919e-01
4.51111197e-01 -1.65987343e-01 9.43814069e-02 9.82057810e-01
1.32986307e-01 6.15322590e-01 -3.47267807e-01 5.30702829e-01
6.72108173e-01 -4.59957361e-01 2.12619632e-01 -3.79316807e-01
-1.15912759e+00 7.01946676e-01 4.54603761e-01 -3.03261697e-01
-3.25853825e-01 7.56380498e-01 2.40408763e-01 1.04809098e-01
-6.14612550e-02 4.13970232e-01 -4.17763412e-01 -7.41668463e-01
-9.72921789e-01 5.46509564e-01 3.75141889e-01 9.97131705e-01
8.72956634e-01 3.37605476e-01 -3.47626992e-02 2.09754631e-01
2.41664983e-02 6.50748730e-01 -2.53753066e-01 -1.89592326e+00
7.52203405e-01 1.00627966e-01 3.68503153e-01 -5.64293683e-01
-6.49235606e-01 -3.79080534e-01 -4.76228774e-01 6.87543929e-01
6.59460604e-01 -9.34305012e-01 -1.05802500e+00 1.58489048e+00
2.84780532e-01 5.95295310e-01 2.14723289e-01 1.17964792e+00
5.60357906e-02 6.76490605e-01 -1.03202976e-01 1.82415962e-01
9.70340490e-01 -1.46239686e+00 -5.62559724e-01 -6.43555462e-01
6.60717607e-01 -4.32777971e-01 6.56821549e-01 3.35834503e-01
-1.13509703e+00 -7.05347002e-01 -1.14018667e+00 1.87232658e-01
-3.07984024e-01 2.45528460e-01 5.18160522e-01 6.05259657e-01
-1.03884101e+00 6.05546355e-01 -1.03986633e+00 -1.77751914e-01
5.09673417e-01 3.44449639e-01 -2.77236640e-01 -1.48213848e-01
-9.73641276e-01 1.22895765e+00 3.37697238e-01 3.36869270e-01
-1.66623688e+00 -7.02443838e-01 -9.46577668e-01 -3.24436873e-01
8.01438153e-01 -5.89117825e-01 1.26855838e+00 -7.77956784e-01
-1.83745229e+00 5.13910949e-01 1.34353012e-01 -7.86054909e-01
9.66962516e-01 -3.22996080e-01 -5.29525936e-01 -1.56645626e-02
2.03873351e-01 9.55011308e-01 4.91186500e-01 -1.06993949e+00
-1.27836275e+00 -1.61966771e-01 2.08079740e-01 6.63624108e-01
7.48095453e-01 -4.05019164e-01 -5.08433104e-01 9.35965776e-02
-3.19192857e-01 -1.37495840e+00 -4.78601515e-01 -3.04019213e-01
-5.00368357e-01 1.07395642e-01 8.48546326e-01 -4.08597261e-01
8.05591404e-01 -1.97546959e+00 3.39794308e-01 1.03778191e-01
7.57454783e-02 -1.69139262e-02 -1.79590359e-01 6.54359162e-02
4.78611052e-01 -1.35735095e-01 -4.92841214e-01 -2.21034810e-01
-8.27739132e-04 2.76882142e-01 -1.93070933e-01 7.50514150e-01
-9.75633338e-02 7.20016241e-01 -1.03403187e+00 -3.31098735e-01
4.54043061e-01 2.20772862e-01 -5.12534380e-01 3.12671959e-02
-1.07727952e-01 4.55133587e-01 -2.63174355e-01 1.83103397e-01
6.91267490e-01 5.18387794e-01 -2.15950459e-01 2.45207980e-01
-7.99857140e-01 -1.38947144e-01 -1.51557505e+00 1.75958896e+00
-2.79928029e-01 1.08529854e+00 4.24289554e-01 -3.71072024e-01
7.18520522e-01 -2.50789404e-01 4.74104315e-01 -5.45707226e-01
2.62138993e-01 4.66705002e-02 1.59900069e-01 -5.35480320e-01
1.01349378e+00 3.24008346e-01 -1.59633443e-01 7.39762411e-02
-3.54131460e-01 -3.50202858e-01 4.20564801e-01 1.99118629e-01
1.05914569e+00 5.67673504e-01 -4.83362108e-01 -3.75571847e-01
1.68814778e-01 7.53396392e-01 6.44705474e-01 7.93292642e-01
-8.21553528e-01 4.18745607e-01 5.67133427e-01 -4.27532345e-01
-1.15295088e+00 -1.17531514e+00 2.13559404e-01 7.26486206e-01
8.03469360e-01 7.17511177e-02 -9.33748126e-01 -5.94561040e-01
-1.88133240e-01 1.00826693e+00 -5.99896014e-01 -1.22288890e-01
-6.51185215e-01 -3.00393939e-01 7.84838736e-01 4.06512201e-01
8.74431789e-01 -7.55173862e-01 -9.90144968e-01 1.15019523e-01
-4.17970605e-02 -1.30142903e+00 -7.51968384e-01 2.15985730e-01
-5.03249764e-01 -1.30038702e+00 -4.40123886e-01 -4.98666495e-01
4.92568791e-01 1.30631402e-01 5.03728747e-01 -3.22316647e-01
-6.27275854e-02 2.54399300e-01 5.19814529e-02 -1.97499339e-02
-1.51423186e-01 4.37339425e-01 2.35532030e-01 9.56474319e-02
-4.17873031e-03 1.95086747e-01 -6.75414443e-01 5.94105124e-01
-1.85021669e-01 1.91763222e-01 3.15974832e-01 2.62933463e-01
6.53748512e-01 3.23933512e-01 6.23440444e-02 -2.65819848e-01
2.48361126e-01 -5.85758567e-01 -1.17632222e+00 -1.83840439e-01
-4.20538396e-01 -1.54740810e-01 5.22105873e-01 -2.33089864e-01
-9.70620394e-01 5.31076610e-01 7.43856728e-02 -6.13873065e-01
-3.26041698e-01 -9.51967314e-02 1.08567648e-01 8.19313303e-02
4.81805980e-01 -4.81586009e-02 -7.91821331e-02 9.58326906e-02
5.64138412e-01 3.73811007e-01 9.83146608e-01 -3.29197496e-01
8.58345687e-01 6.48064792e-01 1.43434063e-01 -7.36249089e-01
-7.11332440e-01 -2.56502867e-01 -8.65162969e-01 -8.44824076e-01
1.54191387e+00 -1.04739344e+00 -1.15595388e+00 6.94958091e-01
-6.56453431e-01 -1.22102833e+00 -3.48981500e-01 8.13815296e-01
-8.35819960e-01 -1.26626104e-01 -4.32165146e-01 -8.98941576e-01
3.09833109e-01 -1.49412310e+00 5.84581792e-01 7.68398881e-01
1.44465536e-01 -6.93514407e-01 2.79187173e-01 6.13674939e-01
3.49972188e-01 9.79138985e-02 2.65217632e-01 -6.82212114e-02
-9.55101430e-01 -4.48148213e-02 -1.85460269e-01 8.33617225e-02
-4.34320539e-01 3.96910816e-01 -7.46934414e-01 -3.12740415e-01
-8.04464877e-01 -1.18142851e-01 9.10948813e-01 8.46994162e-01
8.98652732e-01 2.74046391e-01 -3.81355941e-01 6.91248238e-01
1.41008866e+00 4.65105116e-01 6.45067513e-01 5.08309543e-01
8.62974823e-01 4.16916430e-01 8.00543964e-01 9.03151482e-02
8.94215882e-01 7.38278687e-01 7.81172454e-01 1.08499996e-01
1.93879548e-02 -4.01458740e-01 7.81460941e-01 1.85996860e-01
-1.21084806e-02 -3.91625017e-01 -9.68494534e-01 7.63934374e-01
-2.05465817e+00 -1.08372879e+00 -5.76693058e-01 2.34018636e+00
1.36120487e-02 5.76786399e-01 3.48040521e-01 -4.58516628e-01
8.34614873e-01 5.12930714e-02 -9.24243391e-01 -6.49111032e-01
-8.36033598e-02 -3.90944362e-01 1.40355396e+00 1.18562400e+00
-1.17092347e+00 1.42206311e+00 6.45769739e+00 6.62592947e-01
-7.43086457e-01 -1.57872692e-01 5.97776175e-01 -4.13740367e-01
4.62045297e-02 3.19094688e-01 -1.04423010e+00 3.73066366e-01
1.39084172e+00 1.00871705e-01 8.56908977e-01 8.20624888e-01
5.49514174e-01 -5.88517725e-01 -6.63543284e-01 6.97408020e-01
-2.68719167e-01 -1.50320065e+00 -6.62288666e-01 1.44919857e-01
9.57841456e-01 6.14083111e-01 5.57601601e-02 4.75427538e-01
1.04749393e+00 -1.02570593e+00 1.04565656e+00 5.99696338e-01
7.33233452e-01 -1.21366513e+00 5.40966034e-01 2.91452557e-01
-1.41925967e+00 -2.18050420e-01 2.24794704e-03 -9.29825753e-02
5.52037716e-01 -1.46800548e-01 -3.35869849e-01 3.96793336e-01
6.82131052e-01 6.31742120e-01 -1.28706023e-01 1.26069665e+00
-2.21713424e-01 2.86315709e-01 -4.85325962e-01 1.48329064e-01
9.36191320e-01 -6.63711786e-01 8.88869643e-01 8.61222804e-01
9.20284316e-02 2.09650807e-02 4.92336690e-01 6.23014748e-01
2.16490090e-01 -5.31520307e-01 -6.43573403e-01 4.14069921e-01
5.39356947e-01 1.21245074e+00 -9.13161457e-01 -3.59793246e-01
5.04343286e-02 1.00916314e+00 1.76140293e-01 3.41014832e-01
-1.31689835e+00 -5.95952511e-01 1.15961564e+00 6.22373112e-02
3.31035316e-01 -5.79944432e-01 -2.82470495e-01 -7.51468480e-01
-3.47651660e-01 -3.19294631e-01 2.24870920e-01 -9.03878033e-01
-5.54962695e-01 4.22485888e-01 8.60044807e-02 -9.82155085e-01
7.45939314e-02 -2.89154440e-01 -6.06582224e-01 5.40929914e-01
-1.60112166e+00 -7.67237782e-01 -2.96279699e-01 5.29072583e-01
7.91196704e-01 -2.23606005e-01 6.03850745e-02 4.08246309e-01
-1.18551242e+00 2.65505701e-01 1.68210357e-01 2.73247492e-02
2.93354481e-01 -1.30327702e+00 5.37098229e-01 1.07235467e+00
-3.71611297e-01 -1.93011209e-01 7.93762743e-01 -8.31924260e-01
-1.34119177e+00 -1.27076304e+00 3.40167582e-01 -5.50159931e-01
7.53107965e-01 1.84595957e-02 -4.11375701e-01 1.04048038e+00
2.91429281e-01 -2.58210659e-01 2.01244708e-02 -4.13415700e-01
2.03359663e-01 -1.08559906e-01 -1.09572816e+00 8.96555960e-01
9.67433274e-01 -2.49096721e-01 -1.32490173e-01 3.38029824e-02
5.54684937e-01 -8.94193947e-01 -4.89253521e-01 -1.55884251e-01
3.54921430e-01 -6.68328643e-01 4.96493906e-01 -4.22918141e-01
3.07119310e-01 -7.59592712e-01 -1.10873967e-01 -1.51116729e+00
-1.70372382e-01 -9.99646485e-01 2.71635681e-01 5.87405324e-01
8.68891120e-01 -3.61048102e-01 1.00941992e+00 6.56781912e-01
-6.78678393e-01 -4.77944434e-01 -1.26278055e+00 -8.33552778e-01
2.64627099e-01 -7.40009129e-01 3.22943270e-01 4.41472232e-01
-5.67129195e-01 1.52169615e-01 -5.69662809e-01 5.92403352e-01
1.01411247e+00 -3.31309855e-01 1.02690756e+00 -8.44715476e-01
7.40421042e-02 -6.40905857e-01 -2.22196743e-01 -1.21843016e+00
2.39029378e-01 -7.28750288e-01 3.29591125e-01 -1.66150796e+00
-1.72501862e-01 -4.35126632e-01 1.20734900e-01 2.90999264e-01
2.49378398e-01 -1.89460516e-02 3.67697746e-01 -1.46304891e-01
-6.75478041e-01 4.11123455e-01 1.29858220e+00 -2.12761015e-01
-4.13713098e-01 1.62721038e-01 -3.03388923e-01 5.52261949e-01
8.88508677e-01 -5.56828976e-01 -4.42780823e-01 -5.75010538e-01
2.12952688e-01 3.05577338e-01 2.37758726e-01 -1.39379871e+00
5.29826403e-01 -5.88082910e-01 9.86216357e-04 -8.32702816e-01
6.88806713e-01 -8.60837340e-01 2.10214496e-01 6.80498183e-01
-3.00334185e-01 4.36443806e-01 4.81584430e-01 5.98468602e-01
1.94719329e-01 1.27920201e-02 1.09328341e+00 1.25865713e-01
-1.18023598e+00 4.52970535e-01 -1.07235980e+00 2.06842616e-01
1.66353965e+00 -4.78867292e-01 -5.13319671e-01 -2.73115247e-01
-6.74507916e-01 1.14224732e+00 4.95125115e-01 7.55855083e-01
3.72137189e-01 -9.21804309e-01 -7.53146529e-01 -4.22163904e-02
-2.48763368e-01 1.24146812e-01 7.63738632e-01 8.04713547e-01
-8.91815245e-01 1.99164256e-01 -5.79074681e-01 -4.92915541e-01
-8.18377912e-01 2.30860084e-01 1.02202404e+00 -1.69827528e-02
-7.24640846e-01 7.41773486e-01 -3.88805628e-01 -5.57160676e-01
-2.65325326e-02 -6.10728562e-02 -9.90184397e-02 -8.98792669e-02
3.26015621e-01 1.08581030e+00 -3.20206881e-01 -9.35508490e-01
-3.84495258e-01 8.44619691e-01 1.68937549e-01 -6.07646644e-01
8.82837772e-01 -4.24582273e-01 6.77637219e-01 5.33199668e-01
7.47725844e-01 -3.67632538e-01 -2.27584195e+00 5.09299278e-01
-2.30915293e-01 -2.93238789e-01 6.80302918e-01 -8.70212197e-01
-1.41443884e+00 7.01531947e-01 8.33427012e-01 -1.53917506e-01
3.12991261e-01 -3.77038836e-01 8.85041058e-01 2.77469337e-01
2.34637752e-01 -1.66771340e+00 -3.91194165e-01 7.57002711e-01
2.03515708e-01 -1.28253794e+00 -1.67942733e-01 8.05736557e-02
-1.28885460e+00 8.30461085e-01 1.14164758e+00 -3.10690433e-01
4.17841554e-01 2.92775720e-01 2.41977796e-01 -3.90733331e-01
-8.27306449e-01 -4.46830183e-01 -1.03988923e-01 6.10165417e-01
-4.45044905e-01 3.72224420e-01 3.81771743e-01 -1.69473961e-01
-3.43926907e-01 -1.00277394e-01 1.02486992e+00 6.88403487e-01
-6.05795562e-01 -3.48062366e-01 -1.00302286e-01 2.54303366e-01
-1.06820324e-02 4.47791129e-01 -4.50550281e-02 9.42715764e-01
1.64669454e-01 1.04404414e+00 6.44207418e-01 -5.72292507e-01
6.88957632e-01 -2.63727695e-01 2.03252420e-01 -5.09373061e-02
-6.96199656e-01 -3.71852875e-01 3.08329433e-01 -9.23955023e-01
5.12060383e-03 -7.95331657e-01 -1.57297218e+00 -8.79640460e-01
-1.77706014e-02 1.75216664e-02 9.07308578e-01 9.50737834e-01
3.53060126e-01 6.06010139e-01 7.06216633e-01 -1.00283492e+00
7.28961453e-02 -3.65268260e-01 -4.94211078e-01 -3.14078450e-01
5.43098688e-01 -6.40000463e-01 -3.11318487e-01 -2.73863286e-01] | [5.191470623016357, 1.1756867170333862] |
9a772ada-6575-4e7c-a1f5-ea096f126fee | bros-a-layout-aware-pre-trained-language | 2108.04539 | null | https://arxiv.org/abs/2108.04539v5 | https://arxiv.org/pdf/2108.04539v5.pdf | BROS: A Pre-trained Language Model Focusing on Text and Layout for Better Key Information Extraction from Documents | Key information extraction (KIE) from document images requires understanding the contextual and spatial semantics of texts in two-dimensional (2D) space. Many recent studies try to solve the task by developing pre-trained language models focusing on combining visual features from document images with texts and their layout. On the other hand, this paper tackles the problem by going back to the basic: effective combination of text and layout. Specifically, we propose a pre-trained language model, named BROS (BERT Relying On Spatiality), that encodes relative positions of texts in 2D space and learns from unlabeled documents with area-masking strategy. With this optimized training scheme for understanding texts in 2D space, BROS shows comparable or better performance compared to previous methods on four KIE benchmarks (FUNSD, SROIE*, CORD, and SciTSR) without relying on visual features. This paper also reveals two real-world challenges in KIE tasks-(1) minimizing the error from incorrect text ordering and (2) efficient learning from fewer downstream examples-and demonstrates the superiority of BROS over previous methods. Code is available at https://github.com/clovaai/bros. | ['Sungrae Park', 'Daehyun Nam', 'Wonseok Hwang', 'Mingi Ji', 'Donghyun Kim', 'Teakgyu Hong'] | 2021-08-10 | null | null | null | null | ['key-information-extraction'] | ['natural-language-processing'] | [ 3.17808688e-01 -1.71032324e-01 5.93094379e-02 -3.19709152e-01
-7.94539332e-01 -7.17075109e-01 8.97503138e-01 2.48845890e-01
-3.69243473e-01 2.34274343e-01 3.72627616e-01 -5.53595424e-01
-1.19925193e-01 -4.51764375e-01 -6.72636092e-01 -5.72655976e-01
1.67961746e-01 5.01190960e-01 2.71908075e-01 -1.00901239e-01
8.30383837e-01 4.63960886e-01 -1.37826574e+00 6.86924577e-01
5.10653675e-01 7.58666873e-01 7.09438264e-01 1.00590813e+00
-7.28894532e-01 7.52572536e-01 -6.08518124e-01 -1.35923535e-01
1.93701684e-01 -3.46010506e-01 -8.43374491e-01 1.14166029e-01
7.02914655e-01 -2.14416593e-01 -4.11486655e-01 8.56966734e-01
4.64859277e-01 -6.51010871e-02 8.96987855e-01 -9.76391077e-01
-1.11920798e+00 4.76629734e-01 -8.88355136e-01 2.83141345e-01
4.43163127e-01 -1.53241724e-01 1.16304111e+00 -1.25853157e+00
7.85985231e-01 1.22958386e+00 5.70534766e-01 3.28665763e-01
-1.07866514e+00 -2.86157489e-01 4.26012307e-01 4.02563155e-01
-1.47224259e+00 -2.08495110e-01 7.62794137e-01 -6.05917454e-01
1.17764378e+00 4.37671155e-01 3.47533733e-01 1.14582157e+00
9.36302394e-02 1.28225851e+00 1.14183795e+00 -9.59333479e-01
-8.32644105e-02 2.86709338e-01 2.71426409e-01 7.30571330e-01
9.74171162e-02 -1.71864823e-01 -8.60910654e-01 3.50433469e-01
6.62500679e-01 8.12303089e-03 -3.19061726e-01 -6.02767944e-01
-1.28447747e+00 4.39518422e-01 2.44378537e-01 6.54734671e-01
2.08983839e-01 -1.49285048e-01 3.67712021e-01 1.97641224e-01
4.80645895e-01 5.64274609e-01 -4.10656780e-01 6.48386776e-03
-1.13753438e+00 -1.06228944e-02 5.11834443e-01 1.26563919e+00
5.47813773e-01 -3.60436320e-01 -2.20851421e-01 5.73738933e-01
2.72167057e-01 6.44113004e-01 4.59728152e-01 -2.51333445e-01
1.05748355e+00 7.32782066e-01 -1.16651908e-01 -1.12783027e+00
-5.84456861e-01 -1.22377716e-01 -7.17956603e-01 2.75238380e-02
5.48295259e-01 3.10502946e-01 -1.12773752e+00 1.12997246e+00
1.32940244e-03 -1.18638026e-02 -9.88238528e-02 8.64124298e-01
1.00407314e+00 9.01684642e-01 -2.79180199e-01 1.08998396e-01
1.30300796e+00 -1.13011074e+00 -7.90276289e-01 -4.56969708e-01
8.17139924e-01 -9.14791524e-01 1.51967025e+00 3.96773785e-01
-9.34910297e-01 -4.89718467e-01 -1.06888473e+00 -7.15046704e-01
-9.82331336e-01 7.03028083e-01 2.92759269e-01 5.01202166e-01
-1.18159711e+00 3.01250726e-01 -5.01206696e-01 -6.29785717e-01
3.49177480e-01 2.53791481e-01 -3.71473551e-01 1.90148745e-02
-8.80289733e-01 7.59272754e-01 2.64347136e-01 1.86841547e-01
-4.79638219e-01 -6.57868981e-01 -8.86771381e-01 -5.82331344e-02
5.77781022e-01 -3.65879774e-01 1.15620339e+00 -6.59899056e-01
-1.37631691e+00 1.10826993e+00 -3.89964789e-01 -5.92735037e-02
5.23802042e-01 -5.14234960e-01 -4.23680767e-02 3.19773674e-01
1.39615223e-01 6.86411560e-01 9.44035828e-01 -1.44735110e+00
-6.54526830e-01 -6.56675518e-01 -8.95218998e-02 4.88458157e-01
-5.01206994e-01 -1.64647579e-01 -9.64007735e-01 -8.38846326e-01
1.33816853e-01 -5.83567321e-01 2.16744751e-01 1.62421644e-01
-6.93106651e-01 -1.94680542e-01 1.14655745e+00 -9.53657568e-01
1.44101810e+00 -2.07282543e+00 3.31472397e-01 9.98184904e-02
2.43341401e-01 2.98222840e-01 -1.65975884e-01 6.27651930e-01
9.99271683e-03 1.93445891e-01 -1.19388118e-01 -5.13664067e-01
1.74556347e-03 -8.54699910e-02 -3.75980943e-01 3.40068102e-01
1.20804541e-01 1.10846901e+00 -8.34738433e-01 -7.11730719e-01
5.84330261e-01 2.76635885e-01 -1.24920040e-01 1.08922925e-02
-3.33291888e-01 3.28614801e-01 -5.34372211e-01 5.27455151e-01
7.56635070e-01 -5.26227355e-01 4.01708335e-01 -2.97975361e-01
-2.28931606e-01 2.81055480e-01 -1.29885185e+00 1.90644968e+00
-4.71268445e-01 1.15015829e+00 -1.27961352e-01 -1.18917537e+00
8.24515402e-01 1.06907427e-01 1.54013708e-01 -1.08015501e+00
6.00379985e-03 -2.92950496e-02 -6.22086763e-01 -7.21556783e-01
6.19697690e-01 3.89193952e-01 -3.46464477e-02 5.69597423e-01
1.06291790e-02 -2.06996754e-01 2.79345214e-01 4.75516796e-01
8.12767744e-01 3.88901591e-01 3.05055529e-01 -2.00229898e-01
6.02218747e-01 1.12274580e-01 -2.62193382e-01 1.04457092e+00
6.74708039e-02 7.19060302e-01 5.54008067e-01 -4.81421411e-01
-1.18579304e+00 -9.23192918e-01 -3.90320458e-02 1.16590941e+00
3.58181238e-01 -6.80883646e-01 -6.04438365e-01 -9.54991102e-01
-3.61601524e-02 7.33690023e-01 -8.58162820e-01 4.36702669e-01
-7.26304412e-01 -3.68402541e-01 4.21624213e-01 7.79333651e-01
3.74669194e-01 -6.76607728e-01 -4.97425765e-01 -4.97505903e-01
-1.83118537e-01 -1.09710145e+00 -7.49625266e-01 3.51501346e-01
-6.68671906e-01 -1.03334105e+00 -8.29948843e-01 -1.11047840e+00
9.86538708e-01 6.88639820e-01 1.00004685e+00 1.41959623e-01
-5.56107938e-01 6.15043819e-01 -4.08856630e-01 -3.52256924e-01
8.64934847e-02 2.10103765e-01 -4.48163062e-01 -2.87637472e-01
5.47369421e-01 -8.59540179e-02 -4.30691898e-01 2.84515917e-01
-9.17413116e-01 6.30086243e-01 7.97210515e-01 7.65911460e-01
4.80371326e-01 -8.14691558e-02 7.66581669e-03 -1.02421808e+00
5.81414700e-01 -1.58370599e-01 -5.30981481e-01 7.20321596e-01
-5.24354398e-01 3.35962415e-01 7.60494411e-01 -2.31821418e-01
-1.12060022e+00 5.62838465e-02 1.42479211e-01 -2.11220950e-01
-3.88590723e-01 2.34843880e-01 -1.64913341e-01 2.64984697e-01
6.35852575e-01 4.57527667e-01 -3.03514391e-01 -6.92577302e-01
5.05319536e-01 1.04195094e+00 3.00603867e-01 -5.01245320e-01
8.67642760e-01 6.82493865e-01 -4.72896248e-02 -1.12551260e+00
-1.01211500e+00 -8.26638758e-01 -1.22871196e+00 -4.11041230e-02
7.62908280e-01 -6.64342940e-01 -4.15057391e-01 3.97166073e-01
-1.16676641e+00 -4.94332761e-01 -2.08043270e-02 2.14699447e-01
-4.54274744e-01 6.87229872e-01 -4.70845014e-01 -7.01672316e-01
-2.22526237e-01 -9.15040612e-01 1.55765879e+00 -1.71595637e-03
-1.39890164e-02 -1.21995604e+00 -1.41708210e-01 3.42518032e-01
6.24769926e-02 -6.51900396e-02 9.62113142e-01 -4.90440160e-01
-6.63996458e-01 3.92532907e-02 -5.96047580e-01 7.74623547e-03
6.97312951e-02 -8.60626027e-02 -9.23859894e-01 -2.07968667e-01
-1.95478022e-01 -1.16664805e-01 1.00497544e+00 4.14907068e-01
1.21055269e+00 -1.33436039e-01 -5.99604845e-01 5.76768994e-01
1.41523802e+00 1.60255939e-01 3.72534722e-01 5.76816201e-01
8.24902236e-01 7.46935666e-01 6.49206579e-01 2.35143453e-01
3.77132475e-01 7.62038767e-01 2.03094319e-01 -1.63989663e-01
-3.17050189e-01 -4.33716774e-01 1.83673948e-01 6.69929385e-01
2.05858842e-01 -5.35764635e-01 -1.20121789e+00 5.75859010e-01
-2.04882216e+00 -8.07903171e-01 -2.95927465e-01 1.80299771e+00
6.37081206e-01 1.31307393e-01 -2.07009524e-01 9.73530859e-02
7.14976013e-01 2.50321001e-01 -4.37928408e-01 -2.13033542e-01
-3.78698081e-01 7.43228272e-02 4.67431247e-01 5.28239131e-01
-1.26221669e+00 1.13107574e+00 5.83069658e+00 9.44502831e-01
-1.09168315e+00 -1.71997949e-01 5.61921537e-01 -8.76949430e-02
-2.78777450e-01 -1.07779853e-01 -1.15596914e+00 2.14086145e-01
5.39186180e-01 1.01621434e-01 2.86779523e-01 5.44091344e-01
1.61599338e-01 -2.34318793e-01 -1.11777186e+00 1.01407468e+00
6.26016021e-01 -1.60966146e+00 2.44320869e-01 -2.04510480e-01
6.10209107e-01 -3.26841325e-01 1.65388465e-01 -5.86436503e-02
3.95616591e-02 -1.01692855e+00 9.55469429e-01 4.85240608e-01
1.01629937e+00 -4.02114391e-01 5.49921632e-01 4.47084308e-01
-1.28099084e+00 1.95225030e-02 -2.53388733e-01 1.36110753e-01
-4.02317122e-02 3.16183329e-01 -1.03635025e+00 5.96459925e-01
8.43023002e-01 9.54732835e-01 -1.08928347e+00 6.06835008e-01
-4.30977970e-01 2.73775429e-01 -1.87722638e-01 -3.89328390e-01
4.79179949e-01 -5.82825318e-02 3.58884513e-01 1.67069829e+00
4.09016103e-01 -9.34839472e-02 -1.13157719e-01 6.47241175e-01
6.68295026e-02 3.02124888e-01 -7.87124991e-01 -2.05475867e-01
2.08922118e-01 1.18282187e+00 -9.63875711e-01 -1.97973371e-01
-6.49759173e-01 1.14334011e+00 3.35984290e-01 3.78201813e-01
-5.16881466e-01 -7.25074291e-01 1.24575518e-01 2.89404333e-01
5.95162749e-01 -4.44070816e-01 -5.03329754e-01 -1.05745697e+00
1.42523184e-01 -6.34708166e-01 4.19515818e-01 -1.10589731e+00
-9.86972868e-01 4.70520824e-01 -3.51910144e-02 -1.07835329e+00
1.65935494e-02 -1.21337640e+00 -2.61958748e-01 7.78245211e-01
-1.59737790e+00 -1.37539566e+00 -3.07004333e-01 6.36032045e-01
9.53695178e-01 -9.75306928e-02 5.93542099e-01 4.60584909e-02
-4.56559509e-01 5.01504302e-01 4.84500796e-01 3.10578972e-01
8.71311605e-01 -1.63572550e+00 6.85622215e-01 6.81566179e-01
6.43581212e-01 5.59147835e-01 5.20169318e-01 -6.20862067e-01
-1.50852370e+00 -7.96260953e-01 1.09442866e+00 -8.64356399e-01
7.02750385e-01 -8.71806979e-01 -7.74139464e-01 6.99273586e-01
4.74312752e-01 -3.63782853e-01 4.36141580e-01 -5.96293956e-02
-4.41730350e-01 2.17849791e-01 -6.89859986e-01 8.52159619e-01
1.13778853e+00 -6.89310193e-01 -6.22785628e-01 5.14432430e-01
4.75813627e-01 -7.12889552e-01 -2.71341294e-01 -6.57424629e-02
3.95347863e-01 -1.00227427e+00 1.06677902e+00 -5.39663911e-01
4.99815553e-01 -3.88463199e-01 -4.09762468e-03 -1.05069184e+00
-1.29366100e-01 -4.72115964e-01 -7.74777979e-02 9.28388238e-01
4.55758810e-01 -4.60242294e-02 8.05538476e-01 -7.56473690e-02
2.32467558e-02 -7.10524678e-01 -7.00273156e-01 -5.93787134e-01
4.88539413e-02 -5.44824302e-01 1.89418331e-01 8.32405508e-01
4.62585427e-02 6.20021939e-01 -4.40557897e-01 2.27467045e-01
3.97388130e-01 1.66483000e-01 7.89307117e-01 -9.34321523e-01
-5.28229820e-03 -5.25584102e-01 -1.68756694e-01 -1.72344160e+00
5.35024591e-02 -9.31198657e-01 5.33390744e-03 -1.86023176e+00
2.68272191e-01 -1.17227688e-01 1.82930335e-01 3.96135092e-01
-1.15166090e-01 3.14696617e-02 2.75227815e-01 3.11671019e-01
-8.86197567e-01 1.66835666e-01 1.39471257e+00 -2.71003783e-01
-9.97965708e-02 -1.80919692e-01 -6.52858853e-01 6.47763610e-01
7.91427970e-01 -1.10357367e-01 -4.31902468e-01 -8.54570746e-01
4.14584756e-01 -1.78990230e-01 4.02305365e-01 -6.09007180e-01
6.01546109e-01 -1.73173547e-01 5.98930597e-01 -1.25494719e+00
2.07321420e-01 -7.90586293e-01 -5.07989466e-01 5.80054224e-02
-5.18560708e-01 2.22706303e-01 3.68774176e-01 6.56323612e-01
-1.24873020e-01 -5.29254675e-01 3.84262651e-01 -2.46190980e-01
-1.01832688e+00 -2.20961779e-01 -3.26546520e-01 7.43430033e-02
9.42806661e-01 -4.79819417e-01 -7.04752564e-01 -2.80298918e-01
-6.88418031e-01 2.57863641e-01 3.48715007e-01 6.02186680e-01
8.08419287e-01 -1.04138064e+00 -3.56900871e-01 3.47904682e-01
2.13054448e-01 4.24829647e-02 2.94233084e-01 7.42939234e-01
-7.57063329e-01 1.11017966e+00 1.12336710e-01 -8.00187349e-01
-1.42479563e+00 6.65297925e-01 1.47377610e-01 -5.30304074e-01
-8.43882501e-01 9.31361735e-01 4.26979810e-01 -4.53075200e-01
5.43568552e-01 -4.13939863e-01 -2.18739614e-01 1.67595670e-01
5.83372295e-01 1.66870072e-01 1.65244386e-01 -4.96437222e-01
-4.03288841e-01 9.41470325e-01 -3.58141333e-01 -1.91946819e-01
1.39157093e+00 -3.95411879e-01 1.69602990e-01 5.09637654e-01
1.16020775e+00 1.25287667e-01 -1.24995995e+00 -4.14299220e-01
5.00554562e-01 -6.00256383e-01 1.52518868e-01 -8.37964773e-01
-7.47471392e-01 1.36168039e+00 5.60543835e-01 2.77984619e-01
1.02960157e+00 1.53717220e-01 4.69338626e-01 6.01526260e-01
-9.64141041e-02 -1.30017972e+00 6.22511804e-01 5.79623997e-01
9.59200501e-01 -1.28481662e+00 8.08258448e-03 -2.74886847e-01
-8.71703863e-01 1.51827919e+00 5.05215287e-01 3.07233721e-01
6.89090312e-01 4.62321222e-01 1.60960108e-01 -3.79613310e-01
-5.72514892e-01 -2.92073369e-01 6.03712916e-01 6.23409569e-01
5.47969222e-01 -3.57224077e-01 9.75041986e-02 2.62487888e-01
7.64515577e-03 -2.57758081e-01 2.61325002e-01 1.16954434e+00
-4.31367904e-01 -1.06364465e+00 -5.11062384e-01 4.11534369e-01
-1.46362647e-01 -3.44839692e-01 -7.09005713e-01 9.94199276e-01
-3.09331920e-02 6.00789249e-01 2.09491670e-01 -9.99114141e-02
4.16014940e-01 1.56266570e-01 6.01709545e-01 -7.64204860e-01
-2.98567474e-01 2.27633044e-01 -2.00630873e-01 -2.82075733e-01
-3.52441370e-01 -6.99215353e-01 -1.08727729e+00 -1.47989154e-01
-3.79347682e-01 -6.91641122e-03 7.95803607e-01 8.58514845e-01
3.81067544e-01 4.50444043e-01 3.16060007e-01 -7.72725761e-01
-3.01899016e-01 -7.86377192e-01 -5.37258267e-01 3.67827624e-01
4.20796901e-01 -5.04746139e-01 -3.97021443e-01 2.82890350e-01] | [11.586295127868652, 2.2901511192321777] |
0b25c745-f320-4425-8e67-b531bd0ac8d9 | multi-view-reconstruction-of-bullet-time | 2304.00330 | null | https://arxiv.org/abs/2304.00330v1 | https://arxiv.org/pdf/2304.00330v1.pdf | Multi-view reconstruction of bullet time effect based on improved NSFF model | Bullet time is a type of visual effect commonly used in film, television and games that makes time seem to slow down or stop while still preserving dynamic details in the scene. It usually requires multiple sets of cameras to move slowly with the subject and is synthesized using post-production techniques, which is costly and one-time. The dynamic scene perspective reconstruction technology based on neural rendering field can be used to solve this requirement, but most of the current methods are poor in reconstruction accuracy due to the blurred input image and overfitting of dynamic and static regions. Based on the NSFF algorithm, this paper reconstructed the common time special effects scenes in movies and television from a new perspective. To improve the accuracy of the reconstructed images, fuzzy kernel was added to the network for reconstruction and analysis of the fuzzy process, and the clear perspective after analysis was input into the NSFF to improve the accuracy. By using the optical flow prediction information to suppress the dynamic network timely, the network is forced to improve the reconstruction effect of dynamic and static networks independently, and the ability to understand and reconstruct dynamic and static scenes is improved. To solve the overfitting problem of dynamic and static scenes, a new dynamic and static cross entropy loss is designed. Experimental results show that compared with original NSFF and other new perspective reconstruction algorithms of dynamic scenes, the improved NSFF-RFCT improves the reconstruction accuracy and enhances the understanding ability of dynamic and static scenes. | ['Wentao Zeng', 'Yangtian Yan', 'Yan Gao', 'Linquan Yu'] | 2023-04-01 | null | null | null | null | ['neural-rendering'] | ['computer-vision'] | [ 1.02939025e-01 -5.05576372e-01 9.70061794e-02 -1.79966763e-01
6.00526392e-01 -2.99822122e-01 3.24060284e-02 -8.76509249e-01
-2.82650739e-01 4.89845455e-01 1.97661206e-01 4.33080122e-02
-2.50740409e-01 -8.31381202e-01 -4.58297879e-01 -7.41450727e-01
8.82557705e-02 -3.25875670e-01 6.36994064e-01 -2.67084032e-01
3.01584184e-01 3.93179178e-01 -1.36620069e+00 4.79855627e-01
1.02384746e+00 8.98926020e-01 7.02616394e-01 6.31680131e-01
-2.52185673e-01 1.34883320e+00 -3.55963320e-01 -2.50771910e-01
5.56323409e-01 -6.43818200e-01 -3.76987517e-01 -7.16247931e-02
1.85436323e-01 -8.58113587e-01 -1.00281584e+00 1.43365574e+00
4.77802724e-01 4.67918783e-01 1.84606627e-01 -9.25517976e-01
-8.75379026e-01 2.78730392e-01 -6.52617753e-01 6.10220790e-01
4.41657692e-01 2.95011312e-01 1.08286180e-01 -5.97503543e-01
6.88122869e-01 1.47505772e+00 7.80952990e-01 6.30136430e-01
-8.52104604e-01 -1.01646936e+00 3.14164072e-01 5.48780143e-01
-1.07254887e+00 -3.68858099e-01 9.93216395e-01 -2.42273048e-01
5.45884907e-01 3.51583451e-01 1.09189832e+00 7.48753488e-01
4.43560928e-01 6.15502715e-01 1.25715125e+00 -7.17041716e-02
-4.16590154e-01 1.09682791e-01 -9.46477614e-03 8.66164744e-01
-8.12911466e-02 6.00457728e-01 -8.73785391e-02 5.02497554e-01
1.39515924e+00 2.66969711e-01 -9.42344546e-01 1.43159866e-01
-1.09321630e+00 3.40957612e-01 4.96673256e-01 5.33490539e-01
3.71282198e-03 -1.55812234e-01 4.75955218e-01 4.13727194e-01
4.57286775e-01 3.00177068e-01 -2.20707893e-01 -2.34332681e-01
-1.00365508e+00 -1.21126384e-01 5.47810972e-01 5.98886251e-01
4.93503690e-01 4.62032318e-01 -1.14057429e-01 6.78844094e-01
6.49681166e-02 5.71217775e-01 5.93185306e-01 -1.33473897e+00
3.64282817e-01 4.91374433e-01 -1.77621115e-02 -1.72605693e+00
-3.71535540e-01 -4.26052034e-01 -1.18187940e+00 5.02990246e-01
3.73049945e-01 1.76965863e-01 -8.21093738e-01 1.59172750e+00
2.60979384e-01 5.53970098e-01 -1.66905284e-01 1.35201395e+00
8.86480808e-01 1.25993025e+00 -4.08010185e-01 -8.08012426e-01
7.60261655e-01 -1.09548509e+00 -1.48204637e+00 3.42578650e-01
-4.96325791e-02 -9.29469287e-01 1.07101691e+00 7.03026235e-01
-1.27570581e+00 -1.14383900e+00 -1.10023868e+00 -1.03561692e-01
-8.29389021e-02 -2.18369544e-01 4.98547286e-01 3.87923509e-01
-9.24036980e-01 8.89097810e-01 -5.55792868e-01 1.41195104e-01
3.09614778e-01 1.88660353e-01 -1.76948711e-01 -3.41559798e-01
-1.40176880e+00 1.07373881e+00 5.20430267e-01 4.36766446e-01
-6.17465854e-01 -9.33264315e-01 -4.21919882e-01 2.48524453e-02
3.87629926e-01 -6.59481049e-01 7.19974875e-01 -1.53691959e+00
-1.79574740e+00 8.75380784e-02 2.40502264e-02 -3.31008844e-02
8.71451557e-01 -8.55433866e-02 -8.82855594e-01 5.34004748e-01
-1.52764589e-01 2.25329757e-01 9.40943182e-01 -1.22964406e+00
-5.91486573e-01 3.44417840e-02 1.18178122e-01 3.95743340e-01
2.55917497e-02 1.20138209e-02 -4.79190618e-01 -6.26220167e-01
2.34746099e-01 -5.16641200e-01 5.40099144e-02 4.31942016e-01
9.15159807e-02 4.26451892e-01 1.50341225e+00 -1.26313293e+00
1.41838241e+00 -2.22205997e+00 2.10492492e-01 5.39238825e-02
3.53038549e-01 4.99711096e-01 3.17051373e-02 -1.98779389e-01
-3.12714994e-01 -1.19702384e-01 -1.23830579e-01 3.58436406e-01
-7.15964258e-01 2.70432383e-01 -1.49962366e-01 5.17314136e-01
-6.01863503e-01 6.49833024e-01 -9.46699321e-01 -6.12024307e-01
7.50316441e-01 5.34842610e-01 -4.85086381e-01 2.23414078e-01
1.07113712e-01 7.53236532e-01 -2.73507684e-01 2.54638314e-01
1.18125188e+00 4.92165312e-02 -3.43402445e-01 -6.03497148e-01
-2.73393542e-01 -3.76415133e-01 -1.42712426e+00 1.58521295e+00
-4.79244381e-01 6.47814572e-01 2.54085541e-01 -7.41366804e-01
6.79349005e-01 3.77895445e-01 6.90654397e-01 -9.61572289e-01
2.60859579e-01 -5.52984774e-02 9.71210375e-02 -1.33251190e+00
2.91432440e-01 -3.01664263e-01 7.13950992e-01 1.32760435e-01
-2.33527929e-01 -8.27655643e-02 -5.84839731e-02 1.39113918e-01
5.55724621e-01 4.45348978e-01 -6.48234561e-02 -5.05334176e-02
8.30986500e-01 -3.14457893e-01 9.75320399e-01 3.17891836e-01
-2.01978967e-01 6.81379735e-01 -1.98572725e-02 -7.54558086e-01
-1.05549419e+00 -9.61316884e-01 3.76685821e-02 3.80844116e-01
8.83810461e-01 4.49175946e-02 -6.01157129e-01 -6.08968675e-01
-3.84854615e-01 5.32271326e-01 -3.47293884e-01 -5.49147010e-01
-7.22613633e-01 -4.20822144e-01 2.13120893e-01 1.72986612e-01
1.38219178e+00 -1.00968421e+00 -1.94096237e-01 3.27890515e-01
-6.12180889e-01 -1.11033571e+00 -7.62626708e-01 -6.52680159e-01
-1.07108009e+00 -1.26844728e+00 -7.25071311e-01 -6.68776751e-01
6.29085064e-01 6.84231937e-01 5.51003277e-01 1.88427940e-01
-3.18422616e-02 6.97609922e-03 -1.34364411e-01 9.76385176e-02
-4.38144952e-01 -7.93716431e-01 7.16525316e-02 2.13882491e-01
-2.36306772e-01 -7.69700408e-01 -7.00455308e-01 4.31274205e-01
-1.03583264e+00 5.18125951e-01 1.71149060e-01 8.10164630e-01
1.85533240e-01 8.04566443e-01 2.81248894e-02 -6.43280268e-01
3.49822670e-01 -4.91986237e-02 -6.39795601e-01 1.92496568e-01
-7.22903490e-01 -3.33574057e-01 1.15221190e+00 -8.51717293e-01
-1.67731416e+00 -4.25378770e-01 9.10638943e-02 -9.96383190e-01
2.85636008e-01 9.61054042e-02 -2.69326121e-01 -4.79918569e-01
3.05414349e-01 5.13812363e-01 2.58819222e-01 -2.53210753e-01
7.87445754e-02 3.61974567e-01 6.19650483e-01 7.00898701e-03
7.36326993e-01 6.17585957e-01 -1.29002407e-02 -5.01770675e-01
-7.00550556e-01 -1.76623091e-01 -3.41431141e-01 -1.03151298e+00
1.10411096e+00 -7.68389285e-01 -8.40467155e-01 7.84180939e-01
-1.30862021e+00 -8.87263566e-03 -2.81363577e-01 8.95782471e-01
-3.91461164e-01 7.23992765e-01 -9.55331385e-01 -5.98946989e-01
-1.21798076e-01 -1.15055978e+00 1.79601789e-01 7.70143747e-01
4.58612055e-01 -1.05075419e+00 -2.97131956e-01 3.89773488e-01
5.75408220e-01 1.42012194e-01 6.14892781e-01 4.27763402e-01
-8.76099706e-01 1.37040615e-01 -4.30374801e-01 8.47732246e-01
1.86233893e-01 2.64810830e-01 -8.24573040e-01 -4.25390713e-02
8.30801606e-01 4.46776927e-01 7.65514553e-01 7.86533177e-01
1.49480975e+00 -5.12342513e-01 -7.70890340e-02 1.20516360e+00
1.63286245e+00 9.60639536e-01 1.14693224e+00 3.13901335e-01
8.10488343e-01 6.27833903e-01 5.61108530e-01 -2.60035060e-02
1.18763089e-01 3.45705599e-01 2.87777781e-01 -2.77408093e-01
-4.13807303e-01 -2.17310980e-01 4.11643744e-01 1.26686692e+00
-6.78629816e-01 -1.41582295e-01 -1.62548661e-01 9.79541689e-02
-1.79447699e+00 -1.74520099e+00 -3.18718731e-01 1.88370800e+00
6.80831909e-01 1.48501053e-01 -4.57117587e-01 1.99422643e-01
1.06184685e+00 3.04470420e-01 -4.90433484e-01 -3.89526397e-01
-3.41625482e-01 -3.31815898e-01 3.41746897e-01 6.35256827e-01
-7.60327816e-01 6.10671282e-01 6.24466562e+00 1.13533962e+00
-1.42541826e+00 2.62297153e-01 6.07779384e-01 -1.33931980e-01
-2.43599489e-01 6.71789944e-02 -2.02852741e-01 9.22978103e-01
4.37143207e-01 -2.45715715e-02 1.02146327e+00 4.52105850e-01
7.47676849e-01 -2.90625185e-01 -5.46333849e-01 1.32227194e+00
1.18199870e-01 -1.24167204e+00 5.45572191e-02 -3.01584572e-01
9.00843740e-01 -4.73687828e-01 7.59621561e-02 5.30277751e-02
-1.91571131e-01 -8.90662670e-01 5.01280129e-01 1.23430967e+00
9.27174747e-01 -6.41109586e-01 6.92714691e-01 4.48955894e-01
-1.07251799e+00 -2.72413075e-01 -4.36686844e-01 -3.33354533e-01
3.53658557e-01 7.69941151e-01 1.06525555e-01 7.37663269e-01
8.54812741e-01 1.15881753e+00 -1.59361407e-01 1.03693759e+00
5.16985320e-02 2.77550668e-01 -1.35122851e-01 3.33459258e-01
5.33095188e-02 -8.33809435e-01 8.60031903e-01 9.15658891e-01
2.54079103e-01 4.70830977e-01 6.97761402e-02 8.90524626e-01
3.44658077e-01 -1.55473545e-01 -6.18862689e-01 3.97466838e-01
-2.46822983e-02 1.12960207e+00 -5.31533778e-01 -5.60115397e-01
-4.36906248e-01 1.02285540e+00 -3.09247673e-01 5.40857911e-01
-1.25090623e+00 -4.50508982e-01 3.75387482e-02 3.24520677e-01
-1.07843980e-01 -1.63983285e-01 -2.59587675e-01 -1.45677221e+00
-6.26825169e-02 -6.32527888e-01 1.35574207e-01 -1.47449458e+00
-1.06335604e+00 4.34617966e-01 -5.34558371e-02 -1.64208198e+00
4.10439044e-01 -2.75715411e-01 -6.34248137e-01 8.50844800e-01
-1.53329563e+00 -6.43535733e-01 -5.53551853e-01 9.44763064e-01
6.85339093e-01 -1.48752481e-01 1.35765716e-01 7.25969672e-01
-4.87533629e-01 1.73839658e-01 2.18489751e-01 -1.79431722e-01
6.94006503e-01 -7.20269620e-01 -5.48633993e-01 1.16880214e+00
-4.60500896e-01 3.75168324e-01 6.08283043e-01 -7.94926941e-01
-1.06096411e+00 -9.10105646e-01 3.10099006e-01 7.24466220e-02
1.84083328e-01 9.57500935e-02 -1.07047880e+00 2.86036938e-01
1.48464322e-01 8.02126676e-02 1.72324181e-02 -6.46994293e-01
1.02709517e-01 -4.73093271e-01 -1.39695036e+00 6.37595534e-01
1.18495572e+00 -4.17506665e-01 -5.33169925e-01 1.23302579e-01
9.63061392e-01 -5.15416861e-01 -7.37838268e-01 4.75416988e-01
6.43492103e-01 -1.37013268e+00 1.03724635e+00 -2.68051066e-02
8.05658996e-01 -6.70558929e-01 1.76108479e-01 -1.08654153e+00
-7.28956878e-01 -6.18520081e-01 -2.00035483e-01 1.06814921e+00
-1.94340989e-01 -7.58320212e-01 4.00531977e-01 3.56107116e-01
-2.72043526e-01 -5.12841225e-01 -8.02294254e-01 -6.46824658e-01
-5.10646999e-01 -2.28065759e-01 4.14862245e-01 1.15705466e+00
-4.18223619e-01 6.52687028e-02 -6.71506464e-01 1.21418767e-01
4.88008708e-01 3.61211076e-02 3.72978598e-01 -8.63364100e-01
-2.43368864e-01 -3.72427762e-01 -2.98695594e-01 -1.20088816e+00
-2.02764884e-01 -6.99172616e-01 -2.84788430e-01 -1.50259900e+00
2.13742077e-01 -2.86150932e-01 -2.06310645e-01 -2.65690982e-01
-2.28414342e-01 9.45478678e-02 3.24515909e-01 4.67729360e-01
-2.04204679e-01 6.55452371e-01 2.49476242e+00 -1.08853161e-01
-2.01532245e-01 -4.98086736e-02 -1.94771290e-01 8.55043232e-01
3.88198078e-01 -1.48337498e-01 -7.08117127e-01 -5.92740595e-01
7.82170147e-03 5.56374490e-01 3.35619807e-01 -1.17202997e+00
3.97094935e-01 -3.06495309e-01 8.49736273e-01 -5.20352483e-01
3.94423425e-01 -1.29160607e+00 4.62730497e-01 6.97555959e-01
-1.73625350e-02 3.57970074e-02 -3.50480638e-02 6.22185171e-01
-4.03516322e-01 -8.67745429e-02 1.19669402e+00 -5.12445629e-01
-8.53928089e-01 4.54180896e-01 -1.36937335e-01 -1.59344763e-01
1.12261426e+00 -8.63961339e-01 -3.68043244e-01 -6.40439987e-01
-6.30276263e-01 3.15523520e-02 3.51474822e-01 2.77067035e-01
9.21455622e-01 -1.28876853e+00 -2.70947397e-01 4.51394737e-01
-8.11043084e-01 6.58804029e-02 9.92096186e-01 9.88429606e-01
-1.16528130e+00 -3.97776365e-02 -5.32332420e-01 -4.50812429e-01
-1.09669828e+00 9.36129034e-01 7.21338093e-01 -1.44751385e-01
-9.57004368e-01 6.21090233e-01 5.14952242e-01 -1.18034735e-01
-3.77785303e-02 -2.96122998e-01 -4.99386311e-01 -2.50810504e-01
6.68947279e-01 7.39230990e-01 -5.35812438e-01 -5.18294573e-01
1.62716359e-01 9.96241987e-01 6.33246899e-02 -5.62098157e-03
1.18157816e+00 -5.95470369e-01 -3.59481901e-01 3.69022697e-01
1.15551913e+00 9.12392214e-02 -1.49736357e+00 4.33195755e-02
-9.53046322e-01 -1.00008810e+00 2.52046525e-01 -9.12019610e-01
-1.63422465e+00 7.64541149e-01 8.11726868e-01 8.84829387e-02
1.67715013e+00 -8.22131634e-01 1.16693950e+00 -2.35466957e-01
2.44833857e-01 -1.21726835e+00 1.51720926e-01 4.12048221e-01
7.91845083e-01 -8.76695275e-01 1.54432535e-01 -6.26494110e-01
-5.96026421e-01 1.48567677e+00 9.28858876e-01 -3.69870633e-01
6.21670604e-01 3.85181963e-01 -1.19364589e-01 1.22072279e-01
-3.95884454e-01 5.04276693e-01 1.66853890e-01 5.40600300e-01
-1.31605938e-01 -4.01107103e-01 -3.63530666e-01 3.85680288e-01
-9.54402797e-03 2.96938866e-01 8.35214734e-01 5.33275545e-01
-3.33826602e-01 -4.59090978e-01 -5.00454843e-01 3.89768749e-01
-5.05531251e-01 3.94231081e-02 4.11925554e-01 7.31168211e-01
6.39564395e-01 8.65171611e-01 1.69924996e-03 -6.71620965e-01
3.34449917e-01 -4.80620742e-01 6.36668980e-01 8.50191712e-02
-4.86215770e-01 1.12884521e-01 -2.65397012e-01 -8.51838171e-01
-7.05035925e-01 -3.19063097e-01 -1.27714550e+00 -6.58330917e-01
-5.87399781e-01 -5.29517196e-02 4.32160169e-01 7.39876688e-01
-5.58674783e-02 9.39422309e-01 8.68510842e-01 -5.79303265e-01
-4.34708176e-03 -7.11725116e-01 -6.87820256e-01 4.91774023e-01
3.89293313e-01 -6.69298470e-01 -7.04188287e-01 1.41881302e-01] | [11.00754165649414, -1.8016690015792847] |
bf078e02-fddf-468b-baa8-1306bbbea9fd | abstraction-and-analogy-making-in-artificial | 2102.10717 | null | https://arxiv.org/abs/2102.10717v2 | https://arxiv.org/pdf/2102.10717v2.pdf | Abstraction and Analogy-Making in Artificial Intelligence | Conceptual abstraction and analogy-making are key abilities underlying humans' abilities to learn, reason, and robustly adapt their knowledge to new domains. Despite of a long history of research on constructing AI systems with these abilities, no current AI system is anywhere close to a capability of forming humanlike abstractions or analogies. This paper reviews the advantages and limitations of several approaches toward this goal, including symbolic methods, deep learning, and probabilistic program induction. The paper concludes with several proposals for designing challenge tasks and evaluation measures in order to make quantifiable and generalizable progress in this area. | ['Melanie Mitchell'] | 2021-02-22 | null | null | null | null | ['program-induction'] | ['computer-code'] | [-5.48034087e-02 1.39857173e-01 -1.68717191e-01 -6.24070287e-01
-3.96327168e-01 -5.49638033e-01 8.83085608e-01 5.03742993e-01
-1.71896845e-01 8.13864410e-01 1.62415430e-01 -5.72406411e-01
-2.49126852e-01 -9.00819361e-01 -5.18882155e-01 -6.43809140e-02
-1.34022519e-01 9.01854694e-01 1.90657645e-01 -4.06220496e-01
6.21058166e-01 8.04025233e-01 -1.53367627e+00 3.08900952e-01
1.11335528e+00 5.88990331e-01 -8.64843875e-02 6.60065770e-01
-3.95265490e-01 1.22329271e+00 -6.08911872e-01 -5.12352109e-01
-8.53189826e-02 -4.05766010e-01 -1.20325923e+00 -4.82988626e-01
3.71877015e-01 -3.72678459e-01 -3.12777728e-01 8.26395988e-01
1.10854432e-01 4.30941254e-01 7.57386565e-01 -1.47644377e+00
-1.28319442e+00 8.50287020e-01 5.11567444e-02 2.36516833e-01
9.56157148e-01 2.89694488e-01 9.29735482e-01 -5.44489324e-01
3.70390087e-01 1.47912240e+00 7.47798979e-01 6.20750725e-01
-1.29127979e+00 -7.77752876e-01 8.05194527e-02 2.59392351e-01
-1.21954012e+00 -2.49045640e-01 6.35159671e-01 -6.64962888e-01
1.18195260e+00 2.99792383e-02 8.75956893e-01 1.06263125e+00
1.60512641e-01 7.59079456e-01 1.29815197e+00 -7.23503888e-01
3.64222765e-01 3.19262505e-01 3.35530132e-01 7.33208835e-01
3.23337644e-01 4.07656521e-01 -5.21344125e-01 -4.24923748e-01
9.96253908e-01 -1.64818823e-01 2.41491839e-01 -4.56418723e-01
-1.10163319e+00 7.64314592e-01 6.03688538e-01 6.68368101e-01
-3.35804790e-01 4.26079273e-01 3.95612806e-01 2.73906350e-01
-1.74304083e-01 1.16832900e+00 -2.35335305e-01 -2.48394430e-01
-7.55216837e-01 8.34583580e-01 1.02037418e+00 1.18358696e+00
4.84352350e-01 3.86077970e-01 9.87443700e-02 3.98527384e-01
3.32687832e-02 2.02788472e-01 4.96190548e-01 -1.37791479e+00
5.60905822e-02 6.53043151e-01 1.33814707e-01 -8.46197546e-01
-4.30096745e-01 -1.16014741e-01 -1.99635446e-01 4.49915648e-01
2.99917549e-01 -6.54857233e-02 -6.76768601e-01 1.71197188e+00
-1.16238974e-01 -2.08995551e-01 1.71810657e-01 3.65853935e-01
6.50414228e-01 6.64223313e-01 6.46886766e-01 4.26477581e-01
1.23169506e+00 -6.94703221e-01 -3.01399350e-01 -4.21047926e-01
5.38618088e-01 -3.74024481e-01 1.28109848e+00 3.92495304e-01
-1.33907688e+00 -6.54077411e-01 -1.11424291e+00 -3.78039569e-01
-5.37605762e-01 -5.89446545e-01 1.41131568e+00 8.09046268e-01
-8.55416834e-01 6.45249546e-01 -8.09059858e-01 -4.12194908e-01
4.69698459e-01 2.14439571e-01 -1.49897844e-01 -7.34587088e-02
-1.26201987e+00 1.49881315e+00 8.59055758e-01 -4.28512514e-01
-6.69826448e-01 -6.44343317e-01 -8.26388776e-01 1.29171222e-01
6.80505186e-02 -9.58669007e-01 1.57077670e+00 -9.62901831e-01
-1.37325966e+00 6.71670973e-01 -2.22773939e-01 -6.26517057e-01
8.21284577e-02 -2.51457661e-01 -3.15457821e-01 -1.54024005e-01
1.48739270e-03 7.34278619e-01 9.81367230e-02 -1.26046860e+00
-6.77517354e-01 -2.94196814e-01 3.81374657e-01 2.46780127e-01
-1.16421625e-01 1.97259083e-01 3.75543833e-01 -6.38300061e-01
-4.26089652e-02 -9.28909838e-01 -2.07422793e-01 -2.22707051e-03
6.70728013e-02 -6.37260258e-01 1.48367733e-01 -4.85265166e-01
8.71554077e-01 -1.87808800e+00 2.15808585e-01 1.81846678e-01
2.46438056e-01 1.57762349e-01 -1.42490994e-02 5.99905014e-01
5.12985811e-02 2.41389841e-01 -2.04173684e-01 1.76547974e-01
4.23474371e-01 3.12723011e-01 -6.10797524e-01 -1.29642799e-01
2.24686936e-01 1.13560522e+00 -1.30831730e+00 -4.81529862e-01
2.73313552e-01 1.59443319e-01 -6.57238424e-01 3.30720931e-01
-3.93992543e-01 3.38711709e-01 -5.31040847e-01 6.78203225e-01
1.45629361e-01 4.08193171e-02 2.88126469e-01 3.75532687e-01
-9.81978700e-02 6.52444839e-01 -9.28113520e-01 1.56898201e+00
-7.34025538e-01 7.76036084e-01 -5.27281284e-01 -1.04886687e+00
1.05002987e+00 2.91701615e-01 -6.67572245e-02 -6.18077219e-01
1.09921545e-01 2.78162301e-01 2.05274925e-01 -4.06649262e-01
5.30333579e-01 -4.23654616e-01 -2.61544466e-01 5.14081836e-01
1.14725031e-01 -9.10044789e-01 5.20562381e-02 5.85018136e-02
8.52373838e-01 1.93549603e-01 6.57136917e-01 -3.67969483e-01
4.05350059e-01 2.44948879e-01 1.95085511e-01 9.98576343e-01
-1.00807466e-01 -4.04068232e-02 3.13580602e-01 -9.29966211e-01
-1.18857777e+00 -1.37295926e+00 -4.72824797e-02 1.46237123e+00
-1.59555331e-01 -1.43484622e-01 -6.07430458e-01 -1.50726199e-01
1.69126466e-01 1.66337276e+00 -5.08350194e-01 -2.44545117e-01
-4.97237921e-01 -7.47784823e-02 8.71903241e-01 1.10782385e+00
4.62368071e-01 -1.45943773e+00 -7.92348504e-01 1.97483554e-01
-1.33385081e-02 -7.93506503e-01 3.25981408e-01 5.65690249e-02
-1.00381899e+00 -7.23994911e-01 -2.51551092e-01 -9.87005591e-01
3.32814455e-01 -1.50511451e-02 1.36380577e+00 1.16930790e-01
-1.70089170e-01 4.59892750e-01 -1.46618158e-01 -6.79535806e-01
-6.57630622e-01 -2.58493349e-02 1.69522032e-01 -1.08777809e+00
8.35696220e-01 -8.06549489e-01 -4.19793837e-02 -3.85623574e-02
-7.66790032e-01 -1.77173793e-01 6.19165301e-01 8.33792150e-01
3.59332040e-02 1.20083340e-01 6.13820016e-01 -6.69630468e-01
1.11917114e+00 -5.80960274e-01 -4.96748239e-01 4.25081998e-01
-7.20690727e-01 2.49136165e-01 5.31412005e-01 -4.26166534e-01
-1.19259334e+00 -8.48827511e-02 1.28973676e-02 1.22975811e-01
-4.58574593e-01 5.29431164e-01 -8.05264898e-03 -2.08344653e-01
1.12327683e+00 1.82862699e-01 -2.59728670e-01 -1.99493513e-01
7.89711475e-01 5.70945680e-01 7.18503237e-01 -1.52724540e+00
7.17646301e-01 6.92510754e-02 -1.73933759e-01 -7.44157135e-01
-6.03925943e-01 1.59557298e-01 -7.45487154e-01 2.88722843e-01
7.23461390e-01 -6.82290733e-01 -6.76671743e-01 1.76424429e-01
-1.20993412e+00 -4.23642576e-01 -4.95197058e-01 3.73110831e-01
-9.41708446e-01 2.74082452e-01 -5.77427030e-01 -7.28026569e-01
-1.08285092e-01 -9.28924322e-01 3.56221944e-01 3.67126107e-01
-1.10721159e+00 -1.08302855e+00 2.52580345e-01 2.50618368e-01
6.23792887e-01 3.03218722e-01 1.37095082e+00 -8.82499635e-01
-5.23283362e-01 -3.26272815e-01 -1.55486196e-01 3.10237259e-01
-2.33292282e-01 3.94713357e-02 -8.01354825e-01 1.78225815e-01
8.17414653e-03 -7.82485843e-01 2.97502100e-01 1.11699313e-01
1.19641161e+00 -3.22567523e-01 -3.64239424e-01 3.53632361e-01
9.50850368e-01 6.44289970e-01 6.42333686e-01 4.29536968e-01
3.69076401e-01 5.94084918e-01 2.30082661e-01 2.23772451e-01
8.35852742e-01 4.27803636e-01 -2.11698905e-01 3.89903992e-01
-1.02551682e-02 -5.17172217e-01 5.20735830e-02 3.58832061e-01
-3.97045135e-01 1.63496256e-01 -1.45221698e+00 6.68397844e-01
-1.64498091e+00 -1.18793011e+00 3.47527295e-01 1.82973564e+00
9.99686420e-01 3.71565044e-01 -4.38164324e-02 -4.22779880e-02
4.02829230e-01 -2.06572518e-01 -5.69202602e-01 -1.02840602e+00
6.05081916e-02 5.82574546e-01 -3.17821167e-02 3.59170884e-01
-6.78813457e-01 1.28907359e+00 8.38532448e+00 2.65277654e-01
-6.87242389e-01 -2.99317658e-01 2.97961831e-01 3.91360343e-01
-5.23940146e-01 2.70507336e-01 -3.86002690e-01 -4.27388325e-02
1.04236972e+00 -5.54772675e-01 8.16277027e-01 1.11639035e+00
-5.94973862e-01 5.40421382e-02 -1.60977578e+00 5.59069514e-01
-4.05407809e-02 -1.40583944e+00 3.55954170e-01 -5.06851315e-01
7.19938219e-01 -1.33093610e-01 1.24970160e-01 6.49348259e-01
1.04079926e+00 -1.35691607e+00 9.05118644e-01 6.35934472e-01
4.60085034e-01 -7.14226723e-01 4.50618058e-01 3.69313687e-01
-8.51296842e-01 -3.74255717e-01 -3.72941405e-01 -8.25308263e-01
-2.23609060e-01 5.35288751e-02 -7.67729282e-01 8.02259296e-02
6.52270913e-01 3.18668962e-01 -4.77181166e-01 8.27654183e-01
-7.49115527e-01 4.23156738e-01 -5.90095557e-02 -4.44689870e-01
9.91684049e-02 1.94273964e-01 3.41008902e-01 1.27814460e+00
1.82337284e-01 5.93212366e-01 2.32687965e-01 1.29868603e+00
2.16471449e-01 -2.81097800e-01 -8.54640484e-01 -4.71928269e-01
1.13390303e+00 5.74005902e-01 -4.08837140e-01 -5.86553872e-01
-4.49607790e-01 5.15513480e-01 4.51191783e-01 3.29004794e-01
-7.29609191e-01 -5.96917629e-01 6.20992959e-01 -9.27665383e-02
-2.06415370e-01 -6.64780974e-01 -7.90775418e-01 -8.51963401e-01
-2.26969466e-01 -1.20128644e+00 2.93835163e-01 -9.65705037e-01
-1.38889968e+00 5.15647769e-01 4.29702550e-01 -4.31604356e-01
-6.36177778e-01 -6.66702569e-01 -8.59323084e-01 1.00927711e+00
-8.73364151e-01 -1.27095068e+00 -2.14421675e-01 4.18725371e-01
3.20311576e-01 -4.62921411e-01 1.38431919e+00 -4.00782287e-01
6.60392195e-02 5.68899870e-01 -1.30544856e-01 -3.60063417e-03
3.01669061e-01 -1.28740931e+00 9.50027883e-01 4.40560967e-01
-6.91131502e-02 1.12396526e+00 8.49589050e-01 -4.45641160e-01
-1.32986510e+00 -3.78466427e-01 7.67457128e-01 -9.19785500e-01
9.87115324e-01 -3.24901015e-01 -1.14173484e+00 1.11916828e+00
1.09433264e-01 -4.95429993e-01 7.53358126e-01 6.06891096e-01
-8.34191501e-01 1.09886087e-01 -1.19400883e+00 7.68448770e-01
9.57626104e-01 -8.81287038e-01 -1.54682684e+00 1.59276888e-01
8.62891972e-01 -2.90040821e-01 -1.07787704e+00 2.14848056e-01
7.74327934e-01 -8.78389537e-01 1.12183869e+00 -9.62329447e-01
5.80805600e-01 -4.61744256e-02 -1.93915337e-01 -1.18799043e+00
-7.64286697e-01 -6.75337553e-01 -1.59077317e-01 1.01208937e+00
2.44995251e-01 -7.61214972e-01 6.56199634e-01 1.19840360e+00
-2.43585959e-01 -4.82857138e-01 -6.58919752e-01 -8.48174155e-01
8.66022289e-01 -3.88122022e-01 1.15890872e+00 1.08390021e+00
5.90168536e-01 2.24996939e-01 3.19434464e-01 3.23893651e-02
5.41487217e-01 2.61755735e-01 8.60246658e-01 -1.43500459e+00
-2.02469751e-01 -9.26944792e-01 -5.50290644e-01 -5.54848135e-01
5.16646564e-01 -9.63713050e-01 -4.22996320e-02 -1.63777018e+00
1.65155366e-01 -5.85635960e-01 -2.38936171e-01 5.30968666e-01
-1.94813814e-02 -4.40587431e-01 1.90329999e-01 1.53402403e-01
-3.34810644e-01 2.96747744e-01 8.80759716e-01 -7.83189386e-02
-2.04971090e-01 -1.88916862e-01 -1.20769882e+00 1.06776977e+00
1.01836264e+00 -2.56443113e-01 -4.54099923e-01 -5.80818355e-01
2.17991769e-01 -8.40946734e-02 5.41482806e-01 -1.43520808e+00
4.44140702e-01 -6.42908752e-01 6.61136329e-01 -1.72415778e-01
1.87805280e-01 -5.43449521e-01 5.80119751e-02 5.40098071e-01
-5.62758088e-01 4.22044933e-01 6.27549052e-01 1.62858799e-01
-8.73623192e-02 -3.44947696e-01 6.87558830e-01 -3.30639184e-01
-1.16100395e+00 -2.48635530e-01 -5.97042620e-01 1.80793211e-01
9.42739189e-01 -2.46909469e-01 -2.93062717e-01 -2.97484547e-01
-6.71284616e-01 3.66784900e-01 6.17301226e-01 5.83397627e-01
6.04629099e-01 -1.19328332e+00 -6.55634046e-01 -3.73121607e-03
2.10824892e-01 -1.90431416e-01 -6.19416945e-02 -5.72681502e-02
-6.72851801e-01 6.04611874e-01 -8.20286691e-01 -9.41486359e-02
-6.82262123e-01 9.42430258e-01 2.97590017e-01 -5.26465550e-02
-4.13695902e-01 9.00585055e-01 1.23877311e-02 -6.10296547e-01
3.08100522e-01 -4.02102500e-01 -1.16562434e-01 -6.70498073e-01
6.09233260e-01 5.01235664e-01 -4.30960029e-01 -2.01545984e-01
-1.83829471e-01 2.57701486e-01 -1.27332598e-01 -2.74916947e-01
1.07202196e+00 2.74532318e-01 -3.58468533e-01 6.19446933e-01
5.96713066e-01 -3.50717068e-01 -8.86235058e-01 -1.68871075e-01
3.41088772e-01 -2.19178811e-01 -2.10279092e-01 -1.00926745e+00
-1.61113605e-01 1.06445670e+00 2.02999502e-01 1.63106993e-01
7.79002905e-01 1.14530273e-01 6.81530476e-01 9.77812886e-01
6.25586689e-01 -9.25012171e-01 1.57306029e-03 7.46850431e-01
9.45477068e-01 -9.09143686e-01 2.99385190e-01 1.01536512e-01
-6.62359476e-01 1.14399529e+00 8.37569714e-01 -3.91958147e-01
4.92028654e-01 1.52124032e-01 -4.12771434e-01 -2.11633965e-01
-6.03263617e-01 1.08902566e-01 2.39412576e-01 1.03157032e+00
6.90532207e-01 3.55841994e-01 1.17935957e-02 5.18221974e-01
-7.74483979e-01 2.69057304e-01 2.87397474e-01 1.09236610e+00
-7.74160981e-01 -1.00021350e+00 -3.44860286e-01 2.05196798e-01
6.39159046e-03 -1.07088961e-01 -6.22262180e-01 1.01065719e+00
1.16854265e-01 6.59975111e-01 8.43395367e-02 -3.34875375e-01
4.25045222e-01 4.23276663e-01 7.87272930e-01 -7.37609267e-01
-4.12455022e-01 -1.07442427e+00 -2.20728870e-02 -4.06718522e-01
-2.18511187e-02 -6.65390015e-01 -1.52539527e+00 -7.29945004e-01
1.59761742e-01 1.62034810e-01 4.86934543e-01 8.47275853e-01
5.95611557e-02 3.23942274e-01 6.92408308e-02 -8.12637866e-01
-7.46407747e-01 -7.34167695e-01 -2.50109494e-01 2.69599855e-01
-7.12408572e-02 -6.74717247e-01 -6.42140508e-02 -6.63775057e-02] | [9.214394569396973, 7.011345386505127] |
75ad603c-0a7e-4527-b27e-baa732a623db | semi-supervised-disentangled-framework-for | 2012.11805 | null | https://arxiv.org/abs/2012.11805v1 | https://arxiv.org/pdf/2012.11805v1.pdf | Semi-Supervised Disentangled Framework for Transferable Named Entity Recognition | Named entity recognition (NER) for identifying proper nouns in unstructured text is one of the most important and fundamental tasks in natural language processing. However, despite the widespread use of NER models, they still require a large-scale labeled data set, which incurs a heavy burden due to manual annotation. Domain adaptation is one of the most promising solutions to this problem, where rich labeled data from the relevant source domain are utilized to strengthen the generalizability of a model based on the target domain. However, the mainstream cross-domain NER models are still affected by the following two challenges (1) Extracting domain-invariant information such as syntactic information for cross-domain transfer. (2) Integrating domain-specific information such as semantic information into the model to improve the performance of NER. In this study, we present a semi-supervised framework for transferable NER, which disentangles the domain-invariant latent variables and domain-specific latent variables. In the proposed framework, the domain-specific information is integrated with the domain-specific latent variables by using a domain predictor. The domain-specific and domain-invariant latent variables are disentangled using three mutual information regularization terms, i.e., maximizing the mutual information between the domain-specific latent variables and the original embedding, maximizing the mutual information between the domain-invariant latent variables and the original embedding, and minimizing the mutual information between the domain-specific and domain-invariant latent variables. Extensive experiments demonstrated that our model can obtain state-of-the-art performance with cross-domain and cross-lingual NER benchmark data sets. | ['Boyan Xu', 'Wen Wen', 'Ruichu Cai', 'Zijian Li', 'Di Lv', 'Zhifeng Hao'] | 2020-12-22 | null | null | null | null | ['cross-lingual-ner'] | ['natural-language-processing'] | [-8.66469890e-02 -1.51186273e-01 -3.43744576e-01 -5.71383297e-01
-7.98897624e-01 -8.25340867e-01 3.71256769e-01 2.54271515e-02
-7.90817857e-01 7.69708097e-01 4.37584490e-01 -2.05899036e-04
-1.28376573e-01 -7.42295504e-01 -3.05394232e-01 -8.20062816e-01
3.87373418e-01 5.45078278e-01 -1.02537973e-02 -2.73894608e-01
-2.93109477e-01 3.07797700e-01 -9.13014710e-01 -9.73144397e-02
1.32431865e+00 5.74255884e-01 2.45981306e-01 -1.51389584e-01
-5.49352288e-01 6.88360706e-02 -3.95697892e-01 -3.90688360e-01
8.10427964e-02 -4.93507057e-01 -7.03181148e-01 -2.13911206e-01
-1.18684754e-01 1.43698364e-01 -2.82561898e-01 1.32890141e+00
4.68257517e-01 2.27440640e-01 7.17647731e-01 -1.05279851e+00
-8.83248448e-01 5.68347573e-01 -3.28426689e-01 -3.92154381e-02
-2.96771936e-02 -1.48742989e-01 1.15550041e+00 -7.69154370e-01
5.75135648e-01 1.27181911e+00 3.32688630e-01 7.03348458e-01
-1.44105721e+00 -1.04517055e+00 1.61446378e-01 2.44695134e-02
-1.31865454e+00 -3.19067270e-01 9.94407654e-01 -4.79250818e-01
6.97730541e-01 -1.12939432e-01 1.11979842e-01 1.04750061e+00
-1.01133324e-01 5.36763549e-01 1.13249254e+00 -3.45828772e-01
1.38002232e-01 4.00040925e-01 3.36825490e-01 3.62749994e-01
1.63330704e-01 1.38968378e-01 -2.78431207e-01 -1.43089697e-01
6.65862203e-01 -2.12659943e-03 -1.62451059e-01 -3.60028923e-01
-1.14513814e+00 9.28549767e-01 3.53490353e-01 7.11733460e-01
-2.84894079e-01 -6.94059134e-01 4.36876833e-01 1.58220440e-01
6.66908085e-01 3.98207039e-01 -1.01358724e+00 7.02472404e-02
-6.45479202e-01 -1.72993392e-01 7.78112948e-01 8.71834934e-01
1.04907346e+00 4.13668938e-02 9.69444052e-04 1.14617038e+00
5.00275254e-01 5.93384385e-01 6.44879699e-01 -2.64444411e-01
8.67884517e-01 8.82514358e-01 1.70998573e-02 -8.48541439e-01
-2.77313888e-01 -2.99689680e-01 -9.21961665e-01 -3.01823467e-01
2.67369181e-01 -5.23986816e-01 -7.67077565e-01 2.25632834e+00
3.62602383e-01 1.54511649e-02 4.54971880e-01 8.29058647e-01
9.34193790e-01 7.56414771e-01 6.70560241e-01 -2.61088938e-01
1.47783196e+00 -8.57835650e-01 -7.75512695e-01 -5.68354905e-01
6.35212779e-01 -6.26671076e-01 1.00783038e+00 -4.12602186e-01
-4.74590838e-01 -4.11786169e-01 -8.46113563e-01 -1.70902759e-01
-5.22436261e-01 2.72374690e-01 3.32293749e-01 3.51313293e-01
-3.53310764e-01 1.85689762e-01 -7.21392989e-01 -3.62645149e-01
4.43669874e-03 4.31859195e-01 -9.03286159e-01 -2.35097975e-01
-1.71663165e+00 9.55049396e-01 8.33088040e-01 -1.19395532e-01
-3.42324644e-01 -5.97179830e-01 -1.01216328e+00 2.55966812e-01
2.09073693e-01 -4.30675089e-01 6.72932446e-01 -7.09967971e-01
-1.36862099e+00 8.48724067e-01 -2.93803066e-01 1.32008702e-01
5.99111244e-02 -2.08888054e-01 -7.09801316e-01 -3.64402294e-01
4.59363848e-01 3.12887609e-01 3.48037332e-01 -1.04762912e+00
-4.70219314e-01 -5.86623371e-01 -2.34469086e-01 4.49182123e-01
-7.94549346e-01 7.11462349e-02 -3.83214831e-01 -5.89566588e-01
1.95578769e-01 -8.08663368e-01 -1.02615513e-01 -4.91106331e-01
-3.09968650e-01 -4.77569491e-01 7.30890810e-01 -8.93996239e-01
1.17999399e+00 -2.46087980e+00 4.42057312e-01 -1.40033081e-01
2.54309317e-03 4.10486698e-01 -3.63772064e-01 2.78705001e-01
-1.40230983e-01 2.30160862e-01 -3.78300160e-01 -1.28953934e-01
1.24540545e-01 3.36093694e-01 -1.18710518e-01 2.83393055e-01
3.48697096e-01 5.99292636e-01 -1.06977308e+00 -5.38337350e-01
2.77749598e-01 6.46060526e-01 -1.79441288e-01 4.07723606e-01
2.91718692e-02 7.46573269e-01 -6.89429283e-01 1.68237090e-01
7.28695273e-01 7.65165463e-02 2.40028337e-01 -4.63967234e-01
-1.01550438e-01 5.32100320e-01 -1.11358333e+00 1.64735484e+00
-5.55695832e-01 2.07561418e-01 -2.89135464e-02 -1.00835156e+00
1.21919501e+00 5.90624988e-01 4.27849472e-01 -5.40098190e-01
7.64090717e-02 3.73399675e-01 -1.50044799e-01 -3.85343194e-01
1.67687628e-02 -8.11326027e-01 -5.01754105e-01 3.18677008e-01
4.40806210e-01 2.93940842e-01 6.09488599e-02 -1.65785462e-01
8.17690492e-01 7.86324367e-02 5.99671245e-01 -1.60653189e-01
6.03929162e-01 2.32252176e-03 1.23839736e+00 -9.76428241e-02
-1.72530130e-01 1.88476831e-01 3.50806296e-01 -1.48993522e-01
-7.78993011e-01 -1.15497017e+00 -2.71483362e-01 1.11134470e+00
2.79116571e-01 -1.98461950e-01 -5.76976955e-01 -1.13597524e+00
-2.02614501e-01 9.37372804e-01 -3.71196568e-01 -2.58552730e-01
-4.67396438e-01 -7.69026279e-01 7.81355798e-01 4.61609781e-01
7.01758862e-01 -1.01133728e+00 3.11030626e-01 2.62550503e-01
-5.11412501e-01 -1.19642138e+00 -6.26242280e-01 3.68112445e-01
-9.25752819e-01 -8.43793035e-01 -7.07027256e-01 -1.05701756e+00
7.09412932e-01 4.49447706e-02 8.58267009e-01 -4.64426756e-01
3.64015430e-01 -1.14851594e-01 -2.46588707e-01 -2.45969072e-02
-3.26841444e-01 2.75158256e-01 2.45928571e-01 -5.20143025e-02
9.46197152e-01 -4.45657104e-01 -6.05026968e-02 5.56974649e-01
-9.77912128e-01 -1.92984939e-01 7.20659852e-01 1.02679896e+00
5.85795879e-01 2.22018957e-01 6.04282498e-01 -9.19178128e-01
5.96303761e-01 -6.67421281e-01 -4.08469290e-01 4.03611094e-01
-4.86406863e-01 4.41051096e-01 8.76919150e-01 -5.93596458e-01
-1.46685791e+00 1.13892458e-01 -1.67507622e-02 -1.10952362e-01
-3.93777221e-01 6.83587730e-01 -1.00508857e+00 4.31762546e-01
5.22109330e-01 2.40722984e-01 -5.20520985e-01 -7.52822816e-01
4.48315382e-01 7.90963829e-01 3.74074876e-01 -6.66767001e-01
1.02562940e+00 4.96443920e-02 -2.70327270e-01 -6.87947035e-01
-1.08519232e+00 -8.11143100e-01 -1.09211063e+00 4.89364922e-01
1.15706134e+00 -1.07257354e+00 2.99166087e-02 3.17751229e-01
-1.37707603e+00 2.63233244e-01 -1.81221753e-01 8.70544791e-01
7.30230054e-03 3.79287183e-01 -3.21460098e-01 -3.51511300e-01
-2.76289999e-01 -9.99820411e-01 8.53832722e-01 5.66137969e-01
-2.15495586e-01 -1.52433717e+00 4.52784777e-01 3.20926964e-01
1.14817679e-01 -5.16702905e-02 1.31815088e+00 -1.29125869e+00
-7.55050704e-02 -2.84167528e-01 -4.04419839e-01 6.89118922e-01
5.59879124e-01 -5.21493316e-01 -7.44578719e-01 -2.10970312e-01
5.96534312e-02 -7.89935738e-02 4.31953520e-01 -4.61954344e-03
3.43658149e-01 -3.28055948e-01 -3.62509787e-01 6.61220431e-01
1.34996223e+00 -1.81430969e-02 2.95308411e-01 3.11691284e-01
9.82389808e-01 6.33040428e-01 6.75341725e-01 1.07370364e-02
5.13329148e-01 5.42879403e-01 -1.34741873e-01 -1.45909414e-01
1.17576122e-01 -5.13373375e-01 4.61071402e-01 1.35622263e+00
1.37218997e-01 -1.29287392e-01 -9.97846782e-01 6.84842944e-01
-1.61797452e+00 -7.16807961e-01 1.11811951e-01 2.11700678e+00
1.10100150e+00 -2.06374362e-01 -3.31946999e-01 -3.28998804e-01
1.10963452e+00 1.64399788e-01 -7.54431665e-01 -8.76855552e-02
-1.71424270e-01 1.96536109e-01 2.72623688e-01 2.69453496e-01
-1.24848843e+00 1.23698783e+00 4.49789286e+00 8.57182622e-01
-1.01666677e+00 1.85094878e-01 -6.68637156e-02 4.23937321e-01
-4.21494812e-01 2.68090665e-01 -9.96484578e-01 5.16320705e-01
8.12144279e-01 -5.57584465e-01 3.02009881e-01 1.00300205e+00
7.06652999e-02 5.09572148e-01 -1.10499561e+00 7.75591254e-01
-6.16443269e-02 -6.74441457e-01 -3.52493078e-02 1.88293487e-01
7.57575274e-01 1.13702320e-01 -2.36994714e-01 7.18223751e-01
4.38743174e-01 -7.33911216e-01 1.06798366e-01 1.29897848e-01
9.48030293e-01 -7.10818291e-01 1.00676763e+00 5.55309951e-01
-1.29108047e+00 2.61151642e-01 -5.37546158e-01 3.02649915e-01
1.58829689e-01 6.38843536e-01 -6.03868365e-01 6.28240228e-01
4.82833087e-01 8.23079288e-01 -1.92265674e-01 6.62606120e-01
-7.22031176e-01 6.08005226e-01 -2.90243715e-01 2.04992443e-01
8.64248946e-02 -5.74466109e-01 6.05584979e-01 1.24617636e+00
1.66806847e-01 1.89440131e-01 1.23905703e-01 8.37083638e-01
-3.16192776e-01 4.48908746e-01 -5.53333819e-01 -3.35956335e-01
6.64359570e-01 1.25061548e+00 -3.23073566e-01 -1.76005960e-02
-5.17471671e-01 1.03595102e+00 5.77220559e-01 4.29069340e-01
-6.53482795e-01 -6.29591763e-01 8.06366324e-01 -4.61487472e-01
9.15859789e-02 -2.80560821e-01 -3.37875038e-01 -1.57826006e+00
4.14807089e-02 -5.90934694e-01 5.14734149e-01 -2.38806054e-01
-1.89159358e+00 6.52658284e-01 -1.72041580e-01 -1.24766624e+00
-1.89987123e-02 -4.43978131e-01 -5.15872359e-01 1.34749639e+00
-1.62183726e+00 -1.17456985e+00 -5.13098426e-02 6.38044775e-01
3.15884769e-01 -3.47070575e-01 1.26630402e+00 5.00525177e-01
-6.71998620e-01 6.70835793e-01 5.41837275e-01 7.51172900e-01
1.12242770e+00 -1.06092095e+00 1.78152233e-01 8.13810825e-01
8.29013065e-02 1.02969325e+00 3.86132091e-01 -7.29922891e-01
-9.21243489e-01 -1.15955007e+00 1.40574133e+00 -3.33507568e-01
6.69083953e-01 -3.21417034e-01 -1.26153350e+00 7.41550267e-01
-2.65068114e-01 2.39297692e-02 1.14142001e+00 4.99927998e-01
-5.91430366e-01 -9.98554006e-02 -1.10422421e+00 5.07017612e-01
6.54160619e-01 -8.71102989e-01 -1.05537498e+00 7.85448104e-02
9.39273059e-01 -4.88265269e-02 -1.06593239e+00 3.67236614e-01
3.01022500e-01 -4.12136734e-01 7.94616699e-01 -9.28623199e-01
1.88341275e-01 -3.50952685e-01 -2.15783194e-01 -1.61108088e+00
-4.53987032e-01 -7.23692924e-02 3.61383140e-01 1.83888698e+00
5.38497329e-01 -6.96408629e-01 4.76729214e-01 9.18282390e-01
7.16788918e-02 -1.38731062e-01 -1.11909842e+00 -9.13193405e-01
3.81738335e-01 -5.67785613e-02 5.50090790e-01 1.40835595e+00
8.12884718e-02 1.05839539e+00 -2.53165901e-01 4.48211282e-01
6.05972171e-01 6.81422045e-03 4.83338833e-01 -1.48366189e+00
-1.08210966e-02 -8.47117156e-02 -3.09473485e-01 -9.49466705e-01
6.53565943e-01 -1.12842429e+00 1.97407991e-01 -1.55465817e+00
3.60313028e-01 -6.33873105e-01 -5.86579800e-01 7.42645025e-01
-4.92653847e-01 -4.10685956e-01 1.84131861e-01 3.59989166e-01
-3.79369169e-01 9.27573085e-01 1.01122940e+00 -7.28758723e-02
-3.94860834e-01 -1.72722623e-01 -7.50800788e-01 8.41078758e-01
7.33297050e-01 -1.05107951e+00 -2.78384984e-01 -6.36917591e-01
-1.32592589e-01 9.41895694e-02 3.14725051e-03 -6.54001355e-01
2.02453762e-01 -4.73925501e-01 1.50692344e-01 -1.19091310e-01
1.56975433e-01 -1.05457044e+00 -1.03581414e-01 1.06573738e-01
-3.66869539e-01 -2.69133955e-01 1.37247264e-01 6.91003203e-01
-5.53939998e-01 -2.72766829e-01 8.20798695e-01 7.34886378e-02
-7.29854643e-01 4.01738226e-01 8.98641124e-02 5.47349632e-01
8.30307961e-01 2.06302553e-01 -3.25369507e-01 7.80299306e-02
-7.07057476e-01 1.78983659e-01 3.39635670e-01 6.74483597e-01
1.86266273e-01 -1.50850832e+00 -8.52025330e-01 2.81152636e-01
2.47342661e-01 2.50659697e-02 3.70706588e-01 4.12690938e-01
1.82347059e-01 4.11889374e-01 -2.19287246e-01 -3.93999755e-01
-1.14545560e+00 3.47389460e-01 1.21032096e-01 -7.20624387e-01
-1.22130677e-01 5.85829735e-01 6.16980672e-01 -1.10746169e+00
-1.41057581e-01 2.04270929e-01 -3.36218685e-01 1.27156049e-01
1.65611103e-01 2.62521684e-01 -1.94499373e-01 -1.01897836e+00
-6.16622269e-01 6.77413404e-01 -1.28318146e-01 -9.54815149e-02
1.47535002e+00 -2.59642243e-01 -2.20308930e-01 4.22562301e-01
1.41986072e+00 1.78640887e-01 -9.49926853e-01 -7.41371751e-01
2.33425379e-01 -1.19444430e-01 8.06808919e-02 -8.45395267e-01
-8.26492906e-01 1.11229002e+00 2.90614545e-01 -2.91738570e-01
1.04581356e+00 7.71725699e-02 9.35930431e-01 2.88471818e-01
2.46681735e-01 -1.04891980e+00 -3.05910915e-01 9.25768495e-01
3.96488428e-01 -1.27537096e+00 -1.88641980e-01 -3.78598213e-01
-7.99653649e-01 9.28068042e-01 7.31348991e-01 5.93608730e-02
6.77975297e-01 -1.28039479e-01 2.65919387e-01 1.07284682e-02
-2.81338692e-01 -2.14059025e-01 6.21630669e-01 6.19546473e-01
6.59796178e-01 2.75714964e-01 -4.07738835e-01 1.33071160e+00
2.70176912e-03 -2.61955798e-01 -1.82403654e-01 4.71427053e-01
-1.40615463e-01 -1.54883122e+00 -1.74039930e-01 -7.56499637e-03
-3.24192584e-01 -2.87794977e-01 -3.37490052e-01 7.55893528e-01
2.64924020e-01 8.80021393e-01 -3.98651928e-01 -3.12022507e-01
5.12032628e-01 3.77714038e-01 3.19397455e-04 -9.43494618e-01
-4.10007626e-01 -7.47274533e-02 -4.69667986e-02 -6.00937381e-02
-3.60222965e-01 -5.00967145e-01 -1.38928497e+00 8.14803690e-02
-5.55723786e-01 4.77336109e-01 7.63960898e-01 1.25451601e+00
3.71269107e-01 2.30134040e-01 6.47693992e-01 -2.76971161e-01
-7.25990772e-01 -9.20496285e-01 -6.12013400e-01 7.69109905e-01
-1.16412784e-03 -5.96702456e-01 -5.33066630e-01 9.94791612e-02] | [9.77961540222168, 9.561347007751465] |
a10057b1-a89b-45ce-bea3-3c064219249a | robust-category-level-6d-pose-estimation-with | 2209.05624 | null | https://arxiv.org/abs/2209.05624v1 | https://arxiv.org/pdf/2209.05624v1.pdf | Robust Category-Level 6D Pose Estimation with Coarse-to-Fine Rendering of Neural Features | We consider the problem of category-level 6D pose estimation from a single RGB image. Our approach represents an object category as a cuboid mesh and learns a generative model of the neural feature activations at each mesh vertex to perform pose estimation through differentiable rendering. A common problem of rendering-based approaches is that they rely on bounding box proposals, which do not convey information about the 3D rotation of the object and are not reliable when objects are partially occluded. Instead, we introduce a coarse-to-fine optimization strategy that utilizes the rendering process to estimate a sparse set of 6D object proposals, which are subsequently refined with gradient-based optimization. The key to enabling the convergence of our approach is a neural feature representation that is trained to be scale- and rotation-invariant using contrastive learning. Our experiments demonstrate an enhanced category-level 6D pose estimation performance compared to prior work, particularly under strong partial occlusion. | ['Adam Kortylewski', 'Alan Yuille', 'Angtian Wang', 'Wufei Ma'] | 2022-09-12 | null | null | null | null | ['6d-pose-estimation-1'] | ['computer-vision'] | [ 2.00258374e-01 2.27845415e-01 -9.48558673e-02 -4.94380146e-01
-9.25577462e-01 -5.31779528e-01 6.62585437e-01 -6.42687827e-02
-1.69851363e-01 3.44608665e-01 1.41556010e-01 2.41864592e-01
1.99355692e-01 -7.52453268e-01 -1.06568038e+00 -3.51545006e-01
4.76234630e-02 8.63527894e-01 1.64868653e-01 -4.41336669e-02
3.35776985e-01 1.04620266e+00 -1.84857535e+00 2.79915214e-01
5.59997678e-01 1.35480976e+00 -1.08836301e-01 5.84119558e-01
-2.84686368e-02 3.27720761e-01 -5.97606421e-01 -8.44638944e-02
4.46746081e-01 2.60931421e-02 -6.59483731e-01 4.94535834e-01
8.75580013e-01 -6.50959671e-01 -1.55585602e-01 7.83130705e-01
2.56763607e-01 1.92364320e-01 9.51786757e-01 -9.74840879e-01
-2.05462441e-01 -1.04640566e-01 -6.40092015e-01 -2.47885272e-01
5.53266406e-01 -3.52484770e-02 9.56997573e-01 -1.33007646e+00
9.74970400e-01 1.61862481e+00 5.78515112e-01 4.93229389e-01
-1.52872217e+00 -4.36965108e-01 4.67761546e-01 -3.72231632e-01
-1.46064687e+00 -3.02460998e-01 9.95120645e-01 -4.73337859e-01
8.69772077e-01 3.24091375e-01 1.00250340e+00 6.78324938e-01
2.34277278e-01 6.97970152e-01 1.18161654e+00 -3.58453840e-02
3.86576474e-01 -1.63775712e-01 -2.69496679e-01 8.75402331e-01
2.05846667e-01 8.69392604e-03 -7.36334205e-01 -3.71757060e-01
1.50546050e+00 4.01087515e-02 -5.78055941e-02 -1.08688748e+00
-1.06431258e+00 7.65242100e-01 8.75075877e-01 -3.64705682e-01
-5.38089514e-01 5.18221438e-01 7.02307969e-02 -1.18078016e-01
7.55961239e-01 4.94208992e-01 -4.42076653e-01 4.88726003e-03
-8.44698548e-01 6.71979070e-01 7.62797892e-01 8.73927653e-01
9.48292077e-01 8.27806443e-02 -1.85171776e-02 4.30603981e-01
6.60312057e-01 3.84542942e-01 -1.44488215e-01 -1.29795480e+00
1.13666803e-01 7.16627777e-01 2.06536263e-01 -8.96921933e-01
-3.62581402e-01 -5.69312513e-01 -3.20559561e-01 7.19923854e-01
3.22631866e-01 3.36101145e-01 -1.19556248e+00 1.54174078e+00
7.96529591e-01 1.85401589e-01 -4.60073590e-01 1.03687179e+00
7.52675891e-01 2.89912671e-01 1.64782815e-02 1.05110541e-01
1.04895461e+00 -5.31104445e-01 -1.91861749e-01 -2.46648267e-01
-9.66551714e-03 -6.10828400e-01 9.57578182e-01 2.79841810e-01
-1.47990942e+00 -5.53535342e-01 -9.98647332e-01 -3.52784574e-01
-1.39329240e-01 -8.03985372e-02 8.05172145e-01 3.62693369e-01
-1.01843774e+00 6.31734967e-01 -1.13228452e+00 1.71076860e-02
6.69956028e-01 6.55765414e-01 -2.84118921e-01 -1.40450429e-02
-3.98904830e-01 9.46499705e-01 1.46181583e-01 9.74454731e-02
-1.09338629e+00 -8.49232614e-01 -9.91052747e-01 -1.32771030e-01
7.94807151e-02 -1.02378535e+00 1.14452767e+00 -7.30727136e-01
-1.46669102e+00 1.16815579e+00 -2.35716701e-01 1.91800967e-02
6.25071228e-01 -3.76627982e-01 4.09691304e-01 2.67802745e-01
-5.49484529e-02 8.79993677e-01 1.27646756e+00 -1.76507926e+00
-3.65330130e-01 -7.34015048e-01 2.15289101e-01 6.73272789e-01
2.38219306e-01 -4.12931055e-01 -7.52757311e-01 -5.72651982e-01
8.85628700e-01 -9.23228800e-01 -3.94306660e-01 6.23880267e-01
-2.16701388e-01 -6.54435158e-02 9.16004419e-01 -4.96046394e-01
4.11109954e-01 -1.95408082e+00 4.36075211e-01 3.97952676e-01
4.56602186e-01 -2.85977989e-01 -2.13599466e-02 -1.27359740e-02
1.45439595e-01 1.20441976e-03 -1.01767100e-01 -6.33849025e-01
1.38723012e-03 8.72701481e-02 -3.38387966e-01 7.81261623e-01
6.10560000e-01 9.44338262e-01 -8.43648970e-01 -1.97058395e-01
4.99068648e-01 1.10188198e+00 -8.72850716e-01 3.94826114e-01
-5.54112256e-01 7.13932812e-01 -7.60094583e-01 8.18352282e-01
7.36413419e-01 -3.07477295e-01 -1.52253062e-01 -5.48411965e-01
6.53823838e-02 4.28111404e-01 -1.25997400e+00 2.17330909e+00
-4.32574987e-01 2.70100147e-01 2.80730486e-01 -5.47022045e-01
8.86813581e-01 -2.17816178e-02 6.20085537e-01 -1.71186626e-01
2.44461283e-01 1.73542932e-01 -5.59115827e-01 1.43046528e-01
4.11720514e-01 1.77711379e-02 4.29793671e-02 3.87853593e-01
3.16709541e-02 -8.40409756e-01 -4.26891774e-01 1.21074766e-01
7.51313388e-01 8.76591206e-01 9.39477012e-02 -1.74351290e-01
6.14983998e-02 -6.64402097e-02 3.05934757e-01 5.01269341e-01
2.66125560e-01 9.83606100e-01 2.50258029e-01 -5.08679748e-01
-1.16155648e+00 -1.35726464e+00 -4.25108582e-01 7.22511768e-01
2.55689114e-01 -1.91002920e-01 -6.00637436e-01 -4.46677655e-01
3.53347242e-01 3.44125867e-01 -7.16766536e-01 -7.09876567e-02
-8.22875798e-01 -4.39842641e-01 -1.81072861e-01 5.70757210e-01
1.52030870e-01 -6.65159583e-01 -9.68280613e-01 3.54479730e-01
2.57289797e-01 -1.02075243e+00 -1.58893868e-01 2.45213732e-01
-1.27976930e+00 -9.47107613e-01 -6.69018984e-01 -5.64824224e-01
1.02690268e+00 -1.27658481e-02 1.29496908e+00 9.69629660e-02
-4.21362132e-01 6.47146165e-01 1.53354168e-01 -3.27790678e-01
-1.06769346e-01 -1.17061965e-01 3.64730619e-02 -7.48829767e-02
-1.11300446e-01 -7.02999473e-01 -8.16575527e-01 2.62903094e-01
-4.04044360e-01 3.53747793e-02 4.59239483e-01 5.46827137e-01
1.11390507e+00 -4.67541844e-01 -2.67301966e-02 -6.29625142e-01
2.63498038e-01 -4.19324599e-02 -6.25626385e-01 -1.47722512e-01
-1.55057341e-01 9.81841385e-02 3.55563802e-03 -5.35793722e-01
-9.63090062e-01 5.12385845e-01 -8.23115259e-02 -7.84973741e-01
-2.07369104e-01 8.50640163e-02 -7.45615885e-02 -4.26085442e-01
5.63860834e-01 -1.02307744e-01 -1.27960116e-01 -5.85720837e-01
5.48607171e-01 2.85945833e-02 5.91913164e-01 -8.67350161e-01
9.48127687e-01 7.66476274e-01 2.28833839e-01 -6.31652176e-01
-7.67055213e-01 -3.29786837e-01 -9.10107255e-01 -3.71472567e-01
8.88102293e-01 -1.09844148e+00 -1.01788664e+00 2.18899682e-01
-1.18202639e+00 -3.28340054e-01 -4.71918702e-01 4.54458088e-01
-8.45577180e-01 -9.68407094e-02 -5.87055326e-01 -8.55957687e-01
-1.04741678e-01 -1.29095209e+00 1.89546525e+00 -1.79782044e-02
-3.15542877e-01 -6.10430241e-01 -1.87035397e-01 8.03412497e-02
2.37305328e-01 8.02983642e-01 7.91760981e-01 -4.81134467e-02
-1.07776546e+00 -3.08837980e-01 -1.78031102e-01 -2.08455846e-01
-1.64955094e-01 -4.89675440e-03 -1.19460213e+00 -4.07256901e-01
4.26551327e-02 -6.06162846e-01 6.26965523e-01 5.72351336e-01
1.40303135e+00 -1.06445663e-01 -3.50863487e-01 1.01353383e+00
1.37110651e+00 -3.04697275e-01 2.52530932e-01 6.93693059e-04
9.20268953e-01 4.60120499e-01 6.24590337e-01 3.84792089e-01
3.75575393e-01 7.02166557e-01 9.18356061e-01 -1.99908957e-01
-2.76639670e-01 -4.93580997e-01 -1.00574762e-01 4.49952483e-01
-4.10184622e-01 2.98360705e-01 -7.43759215e-01 2.10092086e-02
-1.50319648e+00 -3.76289874e-01 1.66578412e-01 2.36731791e+00
6.12124622e-01 3.21092188e-01 9.58634168e-02 -1.57595631e-02
3.21628928e-01 1.19931616e-01 -9.26667750e-01 -1.68999031e-01
2.89000392e-01 3.12194586e-01 2.14368165e-01 5.80897868e-01
-9.13126290e-01 9.87772703e-01 7.00983095e+00 3.37262899e-01
-1.13280034e+00 -3.68989781e-02 5.39148867e-01 -1.30630672e-01
-5.92765212e-01 -8.78518671e-02 -8.09363842e-01 -1.85326919e-01
3.71970266e-01 3.41499984e-01 4.54803765e-01 9.19535041e-01
-1.29285425e-01 -4.28743325e-02 -1.40101612e+00 1.02356505e+00
1.14422940e-01 -1.31687129e+00 1.15673326e-01 2.91183680e-01
8.08363974e-01 6.05981685e-02 2.95337498e-01 7.56440833e-02
3.46080065e-01 -1.06249928e+00 1.03519607e+00 5.19804657e-01
8.83434176e-01 -7.72266924e-01 -5.12790354e-03 2.68602937e-01
-1.27919769e+00 2.52840608e-01 -4.66559172e-01 -2.12913528e-02
1.73548147e-01 3.64225119e-01 -8.46528530e-01 9.84175950e-02
7.74661481e-01 5.41707814e-01 -3.87784272e-01 9.15388763e-01
-4.02191997e-01 8.97870064e-02 -5.80835521e-01 1.63615957e-01
5.42854108e-02 -7.22469613e-02 7.07113743e-01 7.66978502e-01
9.24226940e-02 2.39984334e-01 3.79872352e-01 1.24186027e+00
-1.98980093e-01 -4.84177656e-02 -4.66615051e-01 3.36281031e-01
4.31650698e-01 1.20116413e+00 -9.36825454e-01 -1.83917731e-01
-7.18864575e-02 1.01952064e+00 5.85626662e-01 4.95283544e-01
-5.23011804e-01 1.84832722e-01 7.64442742e-01 3.00307333e-01
4.85293418e-01 -6.33424222e-01 -5.11958122e-01 -1.10804975e+00
1.98330414e-02 -6.62357867e-01 -2.75113191e-02 -9.14888144e-01
-9.27028358e-01 5.24917483e-01 9.66570377e-02 -1.01398575e+00
-3.46348703e-01 -7.46545076e-01 -2.82661378e-01 9.02585387e-01
-1.29526663e+00 -1.52382433e+00 -4.43440616e-01 4.88186389e-01
5.96100748e-01 3.71551961e-01 8.17955792e-01 -2.02193350e-01
9.86497328e-02 2.65150666e-01 -4.17887658e-01 -1.48320034e-01
2.42509067e-01 -1.30265462e+00 4.85480219e-01 3.83784443e-01
2.51259476e-01 7.01348603e-01 6.20351613e-01 -7.11637914e-01
-1.77086294e+00 -9.17746246e-01 3.83409709e-01 -9.20069158e-01
1.40791059e-01 -7.89355159e-01 -8.46765578e-01 7.11281776e-01
-4.27298456e-01 5.28665960e-01 1.90346718e-01 1.99481249e-01
-5.48588276e-01 1.41505867e-01 -1.23161495e+00 5.03912926e-01
1.20550501e+00 -7.75213778e-01 -5.30301452e-01 1.39379323e-01
5.52542925e-01 -1.17859697e+00 -1.08441722e+00 4.13160503e-01
8.95397961e-01 -6.86436534e-01 1.35878587e+00 -5.78851700e-01
2.01350614e-01 -3.52951884e-01 -3.35008085e-01 -1.06587970e+00
-1.45096749e-01 -2.91366905e-01 -4.97049451e-01 5.90308309e-01
3.76289487e-02 -2.60645300e-01 1.06732059e+00 8.16384792e-01
-1.56512663e-01 -9.42944884e-01 -8.18981588e-01 -4.06146407e-01
2.14328598e-02 -4.30776447e-01 6.22351646e-01 6.17723346e-01
-5.84365487e-01 4.35677804e-02 -6.97089508e-02 3.57627839e-01
9.76978838e-01 3.35283309e-01 9.49529588e-01 -1.47773004e+00
-2.14858681e-01 -3.27327371e-01 -6.07798040e-01 -1.35134840e+00
1.63361177e-01 -8.08563709e-01 2.10136101e-01 -1.46635640e+00
-1.56551190e-02 -4.93252099e-01 6.56200247e-03 3.46906394e-01
3.52508724e-02 6.54648662e-01 1.22902300e-02 2.46410951e-01
-3.66896689e-01 5.68240464e-01 1.49498856e+00 5.73009327e-02
-2.98497796e-01 -4.07161340e-02 -2.69093990e-01 9.43419576e-01
2.59768158e-01 -4.39109027e-01 -2.50805944e-01 -5.60401678e-01
1.88453585e-01 -9.05665234e-02 6.53036892e-01 -6.78243876e-01
-2.75959838e-02 -7.74010345e-02 9.90666926e-01 -9.76881027e-01
1.00229168e+00 -7.27891266e-01 1.41771287e-01 4.70319301e-01
-3.98942530e-01 -3.77815776e-03 4.53772880e-02 6.16192639e-01
1.55487150e-01 2.99508810e-01 7.56014049e-01 -4.04735178e-01
-5.57700396e-01 7.67094612e-01 8.34738165e-02 -6.81536198e-02
6.52001619e-01 -3.54396731e-01 3.34567726e-01 -3.05912435e-01
-8.78239810e-01 -2.20161080e-01 9.38523769e-01 3.16307604e-01
8.37472558e-01 -1.41946101e+00 -7.11211026e-01 4.03866291e-01
2.34587165e-03 6.49546623e-01 -1.88490167e-01 3.81775469e-01
-4.48502153e-01 3.34616266e-02 -1.83786139e-01 -1.10939527e+00
-9.39607024e-01 2.89175391e-01 4.20935869e-01 1.16651751e-01
-7.69848466e-01 1.12060630e+00 4.17840838e-01 -5.37873149e-01
3.90001744e-01 -4.88823295e-01 8.80918056e-02 -1.99349195e-01
2.69611329e-01 1.42003998e-01 1.53240815e-01 -8.77574503e-01
-4.23873574e-01 9.42186177e-01 1.75565198e-01 -2.95759827e-01
1.49192393e+00 -5.63595735e-04 4.89134565e-02 4.98128682e-01
1.30904973e+00 -1.56990796e-01 -1.76477015e+00 -1.79182231e-01
-4.24343050e-01 -8.46501350e-01 2.15431690e-01 -5.32498002e-01
-9.64377642e-01 8.83295000e-01 5.75931311e-01 -3.03949147e-01
7.42557287e-01 2.67116755e-01 3.11594129e-01 2.80574679e-01
5.58572590e-01 -8.34428132e-01 2.94332951e-01 2.94752121e-01
1.31459248e+00 -1.10480022e+00 3.42605293e-01 -6.74328089e-01
-9.02203619e-02 1.06813717e+00 6.64221942e-01 -5.70172548e-01
5.92934191e-01 3.44164759e-01 -9.19573605e-02 -4.52385992e-01
-4.02193546e-01 -6.83176070e-02 8.46085846e-01 6.42931283e-01
3.63075137e-01 -9.49874669e-02 1.09399684e-01 -2.74988124e-03
-1.88970789e-01 -4.17198956e-01 -9.43713188e-02 1.06069911e+00
-3.89749110e-01 -8.40124488e-01 -4.34692383e-01 4.61222380e-01
-3.18134934e-01 3.30217123e-01 -3.45486730e-01 7.62622714e-01
-9.00082290e-03 2.29520887e-01 3.78690481e-01 -7.98510537e-02
3.83646786e-01 -1.48902357e-01 1.03553736e+00 -8.25648606e-01
-3.77010375e-01 5.14634311e-01 -1.23482928e-01 -1.17797887e+00
-4.82066482e-01 -7.12058246e-01 -1.27856588e+00 2.12156847e-01
-2.26558790e-01 -2.85286576e-01 1.04259038e+00 7.31150031e-01
1.49001047e-01 3.93949479e-01 5.25803804e-01 -1.74918211e+00
-4.69977975e-01 -3.64472717e-01 -3.96179438e-01 5.11788011e-01
4.49892581e-01 -1.09223092e+00 -2.02834100e-01 -1.36033803e-01] | [7.641812324523926, -2.546248197555542] |
4cf0c87d-77cb-491f-b8dc-11e71ef06064 | event-based-vision-meets-deep-learning-on | 1804.01310 | null | http://arxiv.org/abs/1804.01310v1 | http://arxiv.org/pdf/1804.01310v1.pdf | Event-based Vision meets Deep Learning on Steering Prediction for Self-driving Cars | Event cameras are bio-inspired vision sensors that naturally capture the
dynamics of a scene, filtering out redundant information. This paper presents a
deep neural network approach that unlocks the potential of event cameras on a
challenging motion-estimation task: prediction of a vehicle's steering angle.
To make the best out of this sensor-algorithm combination, we adapt
state-of-the-art convolutional architectures to the output of event sensors and
extensively evaluate the performance of our approach on a publicly available
large scale event-camera dataset (~1000 km). We present qualitative and
quantitative explanations of why event cameras allow robust steering prediction
even in cases where traditional cameras fail, e.g. challenging illumination
conditions and fast motion. Finally, we demonstrate the advantages of
leveraging transfer learning from traditional to event-based vision, and show
that our approach outperforms state-of-the-art algorithms based on standard
cameras. | ['Davide Scaramuzza', 'Guillermo Gallego', 'Antonio Loquercio', 'Ana I. Maqueda', 'Narciso Garcia'] | 2018-04-04 | event-based-vision-meets-deep-learning-on-1 | http://openaccess.thecvf.com/content_cvpr_2018/html/Maqueda_Event-Based_Vision_Meets_CVPR_2018_paper.html | http://openaccess.thecvf.com/content_cvpr_2018/papers/Maqueda_Event-Based_Vision_Meets_CVPR_2018_paper.pdf | cvpr-2018-6 | ['event-based-vision'] | ['computer-vision'] | [ 1.96334794e-01 -3.27713698e-01 9.15399939e-02 -4.17327434e-01
-5.25715649e-01 -5.44670045e-01 8.40244055e-01 -2.66961098e-01
-6.39181972e-01 2.86823183e-01 4.83552784e-01 -9.84798595e-02
3.86489183e-02 -5.56369245e-01 -1.15123570e+00 -6.01584256e-01
-1.06238022e-01 -6.28326610e-02 6.31774783e-01 -1.41306579e-01
3.23989630e-01 7.99955666e-01 -1.80431771e+00 4.53646094e-01
-2.16627773e-02 1.07495272e+00 1.41943499e-01 9.97844636e-01
3.78741920e-01 9.57109749e-01 -1.56946644e-01 -2.71308929e-01
3.44004154e-01 -1.26301631e-01 -2.40190491e-01 1.37717173e-01
9.67258155e-01 -7.58053064e-01 -9.71439421e-01 6.86274290e-01
2.30284303e-01 8.73562917e-02 4.50008571e-01 -1.35383141e+00
-3.48334044e-01 -3.37120444e-01 -2.04686970e-01 6.83514416e-01
2.67877609e-01 8.64220738e-01 7.50495970e-01 -8.55318487e-01
9.23512876e-01 1.08157718e+00 9.59640861e-01 4.21107113e-01
-1.02241552e+00 -4.46021616e-01 1.13449633e-01 7.51433551e-01
-9.74163353e-01 -9.01095867e-01 6.83520794e-01 -6.61500931e-01
1.52142751e+00 -2.45253429e-01 7.93225646e-01 1.60044110e+00
4.51343298e-01 8.95146608e-01 5.38520575e-01 1.49422795e-01
2.24161312e-01 -5.15208840e-01 8.75896960e-02 7.85156071e-01
3.79389673e-01 6.81212604e-01 -1.08291543e+00 2.00122625e-01
5.46172202e-01 2.72805065e-01 -2.33653679e-01 -5.36537766e-01
-1.57695425e+00 6.48317456e-01 4.03738767e-01 -3.81674796e-01
-5.02576590e-01 8.32418919e-01 3.66754711e-01 7.69633502e-02
-1.30119259e-02 9.55952480e-02 -4.57545817e-01 -1.79853246e-01
-8.89920413e-01 3.22952449e-01 6.06031299e-01 9.13952231e-01
8.57208312e-01 2.44922012e-01 5.24515621e-02 1.74527168e-02
4.79321331e-01 6.88412309e-01 4.32322100e-02 -1.48914731e+00
2.06763133e-01 3.14666778e-01 4.29171056e-01 -8.46589029e-01
-5.83784759e-01 1.20231338e-01 -4.73499388e-01 5.30149996e-01
5.88508606e-01 -3.29093397e-01 -7.34745383e-01 1.35374784e+00
9.72013399e-02 6.95263863e-01 1.12761721e-01 1.07197201e+00
7.86008954e-01 6.86012328e-01 -1.13813847e-01 -1.03253029e-01
1.05787218e+00 -9.16677594e-01 -3.57708931e-01 -5.87748945e-01
6.59775063e-02 -3.63413870e-01 5.60774982e-01 5.17235219e-01
-7.10956812e-01 -6.69119000e-01 -1.12881362e+00 7.43361786e-02
-4.73335952e-01 1.17405914e-01 6.15844071e-01 3.11911315e-01
-1.05923307e+00 6.70861959e-01 -1.32847583e+00 -8.61050010e-01
4.82835621e-01 2.90799499e-01 -5.42613745e-01 -1.52592391e-01
-7.37351120e-01 9.87424016e-01 -1.56789906e-02 6.89521655e-02
-1.55546498e+00 -8.01183343e-01 -8.04517627e-01 -4.56621572e-02
2.46122986e-01 -5.99514127e-01 1.51882482e+00 -1.03687012e+00
-1.30100834e+00 7.60347962e-01 -3.52752179e-01 -7.50916719e-01
4.32053864e-01 -4.83890504e-01 -4.35685039e-01 5.38654447e-01
-8.15890208e-02 9.11136866e-01 9.11071301e-01 -1.00282657e+00
-7.87973404e-01 -7.23407939e-02 -9.31378528e-02 -2.16265589e-01
-1.78620473e-01 8.03488120e-02 -3.58801365e-01 -1.35644600e-01
-5.09715378e-01 -1.00303864e+00 -3.26502800e-01 4.27797347e-01
2.39837945e-01 4.05500606e-02 1.03789735e+00 -3.05874676e-01
5.07645726e-01 -2.15606999e+00 -2.21695527e-01 -4.36152577e-01
1.22927129e-01 2.62870550e-01 -2.95974851e-01 6.54845536e-01
-1.81100387e-02 -4.09062058e-01 1.38932735e-01 -3.21419716e-01
-1.37894094e-01 2.73084044e-01 -4.11985964e-01 8.48187447e-01
5.97654819e-01 9.93173540e-01 -8.98450553e-01 -3.08697373e-01
6.73569620e-01 5.13269722e-01 -3.64656925e-01 2.03977481e-01
-2.50083625e-01 5.24816394e-01 -1.82496011e-01 5.95997512e-01
2.37460628e-01 -2.50675946e-01 -1.83241546e-01 -4.98692453e-01
-4.25748229e-01 1.28452718e-01 -1.10564125e+00 1.75529408e+00
-5.27599826e-02 1.48803139e+00 -4.13204357e-02 -9.00925517e-01
4.94839758e-01 2.71230429e-01 7.03265667e-01 -6.61255717e-01
8.87704194e-02 -1.86530977e-01 -2.38908291e-01 -8.42123687e-01
5.15575111e-01 1.86108425e-01 6.25227988e-02 8.60080868e-02
2.64136940e-01 9.98007730e-02 1.15492076e-01 8.94685909e-02
1.60331607e+00 4.26090330e-01 2.27355614e-01 8.58619884e-02
8.05721730e-02 3.80977392e-01 5.41142404e-01 9.59238231e-01
-6.25435770e-01 6.91366673e-01 1.58887610e-01 -9.33188617e-01
-1.17420828e+00 -1.10457885e+00 2.10643247e-01 1.02479959e+00
2.71999836e-01 -1.87724665e-01 -3.11746627e-01 -3.89526457e-01
1.82147190e-01 6.50286257e-01 -6.42956853e-01 -2.20587149e-01
-7.09657371e-01 -8.12747836e-01 6.10033274e-01 9.29818988e-01
3.92341614e-01 -9.04500842e-01 -1.34744811e+00 3.07946026e-01
5.94079569e-02 -1.51043463e+00 -6.59144968e-02 2.15223163e-01
-5.75927079e-01 -1.28617489e+00 -7.37094432e-02 -2.98254967e-01
1.06812924e-01 6.91547751e-01 1.24433148e+00 -3.59953612e-01
-4.26456243e-01 9.48321640e-01 -1.53473467e-01 -6.54549062e-01
-1.45322710e-01 -4.14637893e-01 1.08987816e-01 1.89964339e-01
5.28099895e-01 -5.25224984e-01 -1.03485191e+00 1.63428113e-01
-6.15992010e-01 -2.71540005e-02 5.16075671e-01 4.77325022e-01
2.26485789e-01 -4.11684066e-01 4.37635750e-01 -2.21061587e-01
-2.76290290e-02 -6.33338869e-01 -1.07737148e+00 -1.26739696e-01
-3.42884809e-01 -8.87743980e-02 5.40088356e-01 -3.74046355e-01
-1.08570588e+00 8.01164567e-01 1.83075890e-01 -8.49649966e-01
-5.87749422e-01 1.16818421e-01 2.26670995e-01 -2.05994695e-01
7.20156014e-01 1.83033705e-01 -1.32024422e-01 4.95841093e-02
4.33453053e-01 2.72378743e-01 1.10586846e+00 -1.04614748e-02
5.72416961e-01 1.25805342e+00 3.11566204e-01 -8.53107989e-01
-6.39382482e-01 -8.56998146e-01 -6.38598859e-01 -6.63041055e-01
1.15565538e+00 -1.34683299e+00 -9.68569577e-01 5.67276478e-01
-1.39746094e+00 -3.90645862e-01 -1.94149241e-01 6.17460072e-01
-7.69816816e-01 1.84135273e-01 -4.67035174e-01 -8.13159883e-01
1.34836927e-01 -9.36152697e-01 1.46840847e+00 3.37940007e-01
-8.19565877e-02 -7.72493243e-01 3.27383995e-01 2.21943021e-01
4.02566314e-01 3.52715254e-01 -1.92371588e-02 -4.98159856e-01
-1.06306100e+00 -3.91349465e-01 -2.65045911e-01 5.26086567e-03
-3.59416783e-01 3.25754374e-01 -1.31159735e+00 -9.84401107e-02
-1.86091706e-01 -3.88395190e-01 1.38327801e+00 5.71050942e-01
8.70396852e-01 3.37749086e-02 -5.14418185e-01 8.65659118e-01
1.55871046e+00 1.30875744e-02 6.64474130e-01 4.37682927e-01
6.50461674e-01 3.45714808e-01 4.20168489e-01 5.85021317e-01
5.15663564e-01 3.29014719e-01 1.01188409e+00 5.26340194e-02
-1.87879473e-01 5.34078255e-02 7.98312604e-01 8.82697478e-02
3.08497064e-02 -5.58578312e-01 -1.00299883e+00 9.36695814e-01
-2.09161782e+00 -1.36734688e+00 -3.66424739e-01 1.93582988e+00
2.21113980e-01 2.76911259e-01 2.05225106e-02 -2.04131469e-01
3.87212068e-01 4.42721069e-01 -7.08648443e-01 -1.47961572e-01
-2.35050559e-01 -2.39200845e-01 8.77825499e-01 1.63808778e-01
-1.40285325e+00 7.54636168e-01 7.26448870e+00 4.94838553e-03
-1.15848339e+00 -9.53210071e-02 7.03227371e-02 -4.57831532e-01
2.03845352e-01 -7.62757426e-03 -9.68678236e-01 3.16928297e-01
1.62789643e+00 1.68481946e-01 4.15564984e-01 7.82042086e-01
5.75374544e-01 -2.78209299e-01 -1.49803233e+00 9.99898195e-01
3.77828836e-01 -1.76952136e+00 -2.34605268e-01 -6.54516593e-02
7.36087501e-01 9.60060894e-01 -3.46646100e-01 7.62146898e-03
5.47103882e-01 -9.78617072e-01 7.99880147e-01 9.24276352e-01
3.25485229e-01 -3.98866206e-01 4.86272573e-01 2.63217747e-01
-1.10986078e+00 -2.44318485e-01 -2.03084737e-01 -2.86239624e-01
2.66412258e-01 3.18812221e-01 -8.02833557e-01 1.77802175e-01
1.00112104e+00 1.30491602e+00 -7.27337182e-01 1.18689203e+00
1.05128832e-01 7.26354659e-01 -3.92204374e-01 1.16638370e-01
3.08217824e-01 2.19936162e-01 7.66915143e-01 1.57740426e+00
2.28220105e-01 -9.15646832e-03 1.92369938e-01 6.78274453e-01
6.04014471e-03 -8.92497480e-01 -1.18835735e+00 1.00370638e-01
2.81173617e-01 1.30031633e+00 -4.28926706e-01 -3.32548022e-01
-1.02887404e+00 1.08172739e+00 1.39523730e-01 5.10809660e-01
-8.65911782e-01 -5.30806519e-02 9.15710211e-01 -9.55119282e-02
9.32379305e-01 -5.82344949e-01 -2.37757042e-02 -1.20782971e+00
-1.88278645e-01 -4.13027138e-01 2.33947873e-01 -1.24747634e+00
-1.28085506e+00 -3.08657680e-02 -7.04969913e-02 -1.31344879e+00
-4.05462116e-01 -9.65454578e-01 -8.38487208e-01 2.22570106e-01
-1.69236267e+00 -1.31012034e+00 -7.46453226e-01 6.69516444e-01
6.94610596e-01 -2.40298972e-01 3.91551882e-01 1.50432169e-01
-4.26784724e-01 -7.25250840e-02 1.70210600e-01 1.63528249e-01
8.28064978e-01 -1.02482367e+00 6.03598177e-01 1.22157276e+00
1.43939823e-01 9.18586627e-02 8.18461299e-01 -3.70300204e-01
-2.28142667e+00 -1.38477421e+00 4.94226813e-01 -9.24711764e-01
9.32604849e-01 -2.21056283e-01 -5.28017402e-01 1.08316207e+00
4.31177169e-01 5.21134794e-01 3.88924539e-01 -3.39015514e-01
-4.19020087e-01 -4.90450442e-01 -6.39527738e-01 6.00940645e-01
1.01453984e+00 -6.54577971e-01 -7.14097679e-01 2.05715448e-01
3.71939600e-01 -1.71000391e-01 -5.00407517e-01 3.31765175e-01
7.46681035e-01 -1.11412454e+00 1.16780198e+00 -7.35141277e-01
5.54556549e-01 -5.72317600e-01 -3.53430748e-01 -1.02724147e+00
-4.47873473e-01 -7.47046351e-01 -3.21155548e-01 8.15399349e-01
-6.57827407e-02 -1.47089347e-01 7.09275126e-01 2.49151796e-01
-4.77631539e-01 -1.41302869e-01 -1.16878295e+00 -7.03603387e-01
-2.63972700e-01 -8.19577038e-01 2.44986285e-02 6.07542813e-01
-3.60239387e-01 3.48015696e-01 -4.61027622e-01 5.02063870e-01
7.96919107e-01 3.18792723e-02 9.50955987e-01 -1.19927335e+00
-6.21524751e-02 -2.64936835e-01 -7.72330642e-01 -8.74232292e-01
2.63376683e-01 -3.51139963e-01 3.91154826e-01 -1.32046294e+00
1.85232311e-01 7.34645367e-01 -2.68437505e-01 3.63811523e-01
7.59027302e-02 3.80220950e-01 7.84745067e-02 9.59627107e-02
-1.16436517e+00 3.34122866e-01 6.14502132e-01 -1.23331130e-01
1.78797737e-01 -5.54084361e-01 -2.98899800e-01 1.01130772e+00
5.58978498e-01 -2.68635392e-01 -5.63788451e-02 -8.36498737e-01
3.36650461e-01 2.09104642e-01 1.21172345e+00 -1.30696654e+00
7.24588394e-01 -3.76174420e-01 7.18233705e-01 -7.80048609e-01
4.00513709e-01 -1.03214991e+00 1.16163276e-01 3.99037927e-01
-1.09737568e-01 2.56091475e-01 4.22302753e-01 1.17835462e+00
1.52624268e-02 3.74330133e-01 7.24218845e-01 -2.58489788e-01
-1.42533827e+00 1.71585038e-01 -9.51361358e-01 -8.86200517e-02
9.84793544e-01 -3.30353796e-01 -7.18954742e-01 -3.95353794e-01
-2.97244579e-01 2.53465205e-01 4.89063799e-01 4.93170559e-01
6.14159584e-01 -1.18675601e+00 -6.99433148e-01 6.48588464e-02
4.35006946e-01 -3.76874566e-01 -5.92941768e-04 8.16037178e-01
-5.75589597e-01 5.39432049e-01 -4.26380545e-01 -9.59605813e-01
-9.60809410e-01 6.48679554e-01 4.11884218e-01 2.52458632e-01
-6.99551105e-01 5.04709661e-01 1.38038307e-01 1.29969776e-01
2.24217638e-01 -4.73680615e-01 7.77246431e-02 1.06707038e-02
4.16880518e-01 5.27684748e-01 1.29901618e-01 -5.17024696e-01
-8.05555046e-01 5.07287323e-01 3.75018537e-01 -1.84924811e-01
1.44847894e+00 -2.87873030e-01 4.49601650e-01 5.87245345e-01
1.02786350e+00 -2.32190862e-01 -2.15861797e+00 1.95868731e-01
-1.33691102e-01 -1.74512982e-01 2.64088511e-01 -6.17238581e-01
-9.46926773e-01 1.08187354e+00 8.12164903e-01 -1.07037060e-01
1.06160903e+00 -1.79726064e-01 8.31609011e-01 9.29760933e-01
1.14629366e-01 -9.48795676e-01 2.37508252e-01 6.29942119e-01
6.20447755e-01 -1.74894667e+00 9.45336670e-02 5.36180213e-02
-4.76768672e-01 1.29590547e+00 5.51488757e-01 -5.07890821e-01
5.10259509e-01 3.03102225e-01 -9.68837216e-02 -2.86120445e-01
-1.38562191e+00 -4.42254573e-01 2.03504786e-01 7.20405996e-01
2.60168817e-02 -3.37113231e-01 3.94449830e-01 2.94249266e-01
2.43349269e-01 4.15392190e-01 7.04814434e-01 9.61928666e-01
-5.22332668e-01 -2.77004808e-01 -3.87929857e-01 2.16009155e-01
-2.42883995e-01 -2.05169376e-02 -5.19104838e-01 8.16263318e-01
2.64159620e-01 1.13873947e+00 4.14643973e-01 -4.56469446e-01
4.17517781e-01 1.43063888e-01 3.69413525e-01 -1.18489958e-01
-5.74647486e-01 -1.19099773e-01 2.81835377e-01 -1.12126005e+00
-7.98103571e-01 -1.00923765e+00 -1.02685952e+00 -2.41731420e-01
1.04946196e-01 -7.25560784e-01 5.76229751e-01 9.92601514e-01
6.28259003e-01 3.98338079e-01 4.14073497e-01 -1.29281914e+00
-2.65316695e-01 -4.75858033e-01 -4.92725335e-02 4.74975675e-01
9.93753731e-01 -5.78764558e-01 -4.03722167e-01 7.96337008e-01] | [8.491018295288086, -1.0363715887069702] |
d1a0ce02-2cf4-4e6e-a086-083318c9a84d | temporal-information-extraction-from-korean | null | null | https://aclanthology.org/K15-1028 | https://aclanthology.org/K15-1028.pdf | Temporal Information Extraction from Korean Texts | null | ['Ho-Jin Choi', 'Chae-Gyun Lim', 'Hyun-Woo Do', 'Young-Seob Jeong', 'Zae Myung Kim'] | 2015-07-01 | null | null | null | conll-2015-7 | ['temporal-information-extraction'] | ['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] | [-7.157752990722656, 3.9809765815734863] |
77944cc8-e793-41cb-9823-d66dc1707df2 | graph-convolutional-neural-networks-with | 2212.02055 | null | https://arxiv.org/abs/2212.02055v2 | https://arxiv.org/pdf/2212.02055v2.pdf | Graph Convolutional Neural Networks with Diverse Negative Samples via Decomposed Determinant Point Processes | Graph convolutional networks (GCNs) have achieved great success in graph representation learning by extracting high-level features from nodes and their topology. Since GCNs generally follow a message-passing mechanism, each node aggregates information from its first-order neighbour to update its representation. As a result, the representations of nodes with edges between them should be positively correlated and thus can be considered positive samples. However, there are more non-neighbour nodes in the whole graph, which provide diverse and useful information for the representation update. Two non-adjacent nodes usually have different representations, which can be seen as negative samples. Besides the node representations, the structural information of the graph is also crucial for learning. In this paper, we used quality-diversity decomposition in determinant point processes (DPP) to obtain diverse negative samples. When defining a distribution on diverse subsets of all non-neighbouring nodes, we incorporate both graph structure information and node representations. Since the DPP sampling process requires matrix eigenvalue decomposition, we propose a new shortest-path-base method to improve computational efficiency. Finally, we incorporate the obtained negative samples into the graph convolution operation. The ideas are evaluated empirically in experiments on node classification tasks. These experiments show that the newly proposed methods not only improve the overall performance of standard representation learning but also significantly alleviate over-smoothing problems. | ['Jie Lu', 'Maoying Qiao', 'Junyu Xuan', 'Wei Duan'] | 2022-12-05 | null | null | null | null | ['point-processes'] | ['methodology'] | [ 5.65641858e-02 1.08829886e-01 -1.16229601e-01 -3.63902539e-01
-8.45571980e-02 -3.34450454e-01 5.21252811e-01 5.25186002e-01
-3.20890546e-01 6.24167860e-01 2.09034141e-02 -2.42078722e-01
-1.07406318e-01 -1.45681763e+00 -5.37711203e-01 -1.11916924e+00
-5.53411901e-01 2.93896198e-01 1.96597993e-01 -2.56605595e-01
-7.50235170e-02 6.74563706e-01 -1.16471338e+00 9.07114893e-02
8.44958127e-01 8.97581041e-01 9.52340066e-02 3.90695870e-01
-4.52056527e-01 6.01236880e-01 -6.11382604e-01 -4.21299964e-01
1.18880458e-01 -2.76647270e-01 -4.67977583e-01 3.65970358e-02
4.76264544e-02 9.23891366e-02 -6.33019745e-01 1.23590422e+00
3.83207500e-01 4.61802542e-01 8.65298629e-01 -1.30055237e+00
-7.29781866e-01 7.93025076e-01 -9.72213984e-01 1.66142046e-01
-1.70105379e-02 -9.41794086e-03 1.21900296e+00 -6.28761113e-01
4.32803869e-01 1.35126269e+00 6.70495212e-01 2.32508332e-01
-1.20205116e+00 -7.39761949e-01 5.07004440e-01 2.18599901e-01
-1.25940120e+00 -2.71340758e-02 1.22573054e+00 -1.32719755e-01
5.58715403e-01 4.02529389e-01 9.04947996e-01 9.29289639e-01
9.28066000e-02 8.42261553e-01 3.90888333e-01 -1.46570325e-01
9.70522463e-02 -2.36329958e-02 2.60308176e-01 7.54508615e-01
5.67470074e-01 -3.41506809e-01 -1.43535361e-01 -2.15019464e-01
9.25678611e-01 2.52906919e-01 -5.43873072e-01 -6.25911295e-01
-9.44939375e-01 9.96432662e-01 1.18734264e+00 3.90702724e-01
-6.89160347e-01 4.04613703e-01 4.19643730e-01 2.82067776e-01
5.58065414e-01 -2.03224309e-02 -2.40660131e-01 3.13097358e-01
-3.03458720e-01 -2.66302913e-01 7.93555796e-01 6.90542102e-01
1.16635299e+00 1.06671080e-02 -3.07625920e-01 9.47031200e-01
5.37162304e-01 2.06746846e-01 3.11805248e-01 -2.12510109e-01
6.19639039e-01 1.00342309e+00 -2.98698813e-01 -1.84842360e+00
-5.53040028e-01 -1.02246749e+00 -1.73259592e+00 -1.44158274e-01
2.07760826e-01 -7.56340921e-02 -9.86159801e-01 1.69248104e+00
3.13789487e-01 3.58274400e-01 -1.05487667e-02 7.20404446e-01
1.03002405e+00 9.86799896e-01 2.13409081e-01 -2.06353128e-01
1.07906890e+00 -6.35200620e-01 -4.46034372e-01 -5.07177860e-02
6.40047967e-01 -4.21777606e-01 7.68252075e-01 1.48761034e-01
-7.44116127e-01 -4.31547672e-01 -8.83818865e-01 2.01050341e-01
-2.20755637e-01 2.71312356e-01 9.15572643e-01 4.53323781e-01
-1.01867247e+00 7.90601850e-01 -7.38152802e-01 -8.20183009e-02
6.16926968e-01 3.85956496e-01 -1.94649339e-01 -1.58257574e-01
-1.26843643e+00 1.76669002e-01 2.84636110e-01 4.05967116e-01
-4.79662299e-01 -5.35899520e-01 -9.74111974e-01 4.48380560e-01
1.93256840e-01 -4.39111710e-01 6.55232310e-01 -8.90623987e-01
-9.42865729e-01 2.88292110e-01 -1.49222583e-01 -1.65805966e-01
2.77296364e-01 4.94059175e-01 -5.27791262e-01 1.95044845e-01
-1.10164009e-01 3.74535769e-01 9.35869157e-01 -1.11208248e+00
-3.62529308e-01 -2.78467238e-01 3.36308144e-02 2.01706573e-01
-6.48749352e-01 -5.55266678e-01 -7.26125658e-01 -7.45845914e-01
5.77527046e-01 -6.69119775e-01 -5.31320155e-01 1.52656855e-02
-5.09609461e-01 -4.36271727e-01 6.30888402e-01 -3.60459566e-01
1.32847595e+00 -2.26352811e+00 7.40429536e-02 8.19859207e-01
8.47026944e-01 2.95140117e-01 -3.82780403e-01 4.92902458e-01
-1.40253395e-01 2.38313586e-01 -1.79655939e-01 -1.01594038e-01
-2.65777916e-01 2.47977749e-01 -4.18689251e-02 4.39005882e-01
4.85349298e-01 8.35945666e-01 -1.18601871e+00 -2.80932099e-01
2.55491018e-01 8.10744762e-01 -3.68421048e-01 -1.85939282e-01
-5.30243479e-02 1.05622314e-01 -7.15926170e-01 2.77936012e-01
9.80089366e-01 -6.44004941e-01 2.38055527e-01 -3.73854488e-01
4.78123963e-01 1.74018800e-01 -1.40923357e+00 1.09801447e+00
-4.14318830e-01 5.39700985e-01 1.85469374e-01 -1.37177444e+00
1.22674477e+00 -3.01990882e-02 3.48293304e-01 -4.71100330e-01
1.87127071e-03 -1.07889771e-01 2.04096481e-01 -2.52440907e-02
4.27828789e-01 1.41964793e-01 1.99173152e-01 2.68744081e-01
-1.00868180e-01 3.51379752e-01 2.58090258e-01 5.80432534e-01
1.17644036e+00 -5.26104510e-01 2.03427002e-01 -1.92403093e-01
7.87332416e-01 -5.17868042e-01 7.42726386e-01 5.59720218e-01
6.29560575e-02 3.82445782e-01 1.04603791e+00 -4.11936283e-01
-6.00356579e-01 -1.01349497e+00 1.76738590e-01 9.07355368e-01
1.79793224e-01 -5.94278038e-01 -3.25825095e-01 -7.33504653e-01
2.40196865e-02 3.51516157e-01 -7.36542881e-01 -4.62838382e-01
-4.49959725e-01 -9.78493690e-01 2.44159207e-01 4.90298748e-01
4.07355160e-01 -1.02167785e+00 1.85240567e-01 3.88933986e-01
1.49873808e-01 -5.48574209e-01 -3.42660934e-01 1.97059199e-01
-8.77531707e-01 -1.16141474e+00 -8.95197570e-01 -9.60211873e-01
1.15116680e+00 5.86944520e-01 9.15106297e-01 5.21994829e-01
-4.62320186e-02 9.68575925e-02 -3.87517422e-01 6.18221238e-02
-3.25108081e-01 2.94654816e-01 -2.41091907e-01 4.12300527e-01
2.07029134e-01 -8.36799979e-01 -5.69090545e-01 1.57306790e-01
-7.40835130e-01 -3.07289213e-02 8.00327420e-01 9.29410279e-01
5.65523624e-01 3.81661147e-01 6.37718141e-01 -9.46005404e-01
9.13255036e-01 -5.35764337e-01 -4.42396164e-01 2.72921652e-01
-2.68338978e-01 2.19321430e-01 1.06989706e+00 -6.21684670e-01
-7.40202725e-01 -5.34005873e-02 1.26607269e-02 -3.76106501e-01
2.49587893e-01 8.29174757e-01 -2.67466068e-01 -8.22979584e-02
5.14477372e-01 2.63940811e-01 8.85293186e-02 -3.43784094e-01
2.66107738e-01 2.38092646e-01 5.07846437e-02 -4.10558403e-01
9.45431948e-01 2.94801027e-01 3.11444998e-01 -1.10527360e+00
-3.71290743e-01 -4.04293656e-01 -2.89658159e-01 -1.95362031e-01
2.69761115e-01 -7.50537276e-01 -7.92099476e-01 5.45517087e-01
-1.01771224e+00 -3.98291610e-02 -1.82936862e-01 4.45551842e-01
2.28723004e-01 6.42379522e-01 -6.93749785e-01 -8.81546021e-01
-3.42144579e-01 -9.48660612e-01 6.87188268e-01 4.03374016e-01
1.57988429e-01 -1.27289927e+00 -2.85171747e-01 -5.14647365e-01
4.17624384e-01 1.20229922e-01 1.08522177e+00 -6.92295849e-01
-6.25869870e-01 -3.64967555e-01 -5.08829594e-01 3.41992080e-01
1.39149472e-01 7.29689449e-02 -5.44581831e-01 -4.78001624e-01
-4.09992099e-01 1.06451713e-01 1.26147318e+00 4.46469575e-01
1.20872533e+00 -1.71395540e-01 -5.48878193e-01 6.36240900e-01
1.16424286e+00 -1.65460557e-01 6.23525321e-01 -1.32581651e-01
1.02669394e+00 4.73728955e-01 1.90115646e-01 5.38633525e-01
1.59221828e-01 2.05721885e-01 4.58648860e-01 -2.39278540e-01
1.13550117e-02 -2.01534241e-01 1.89126909e-01 1.07552218e+00
-8.05995893e-03 -4.58536148e-01 -7.21403360e-01 3.92374605e-01
-1.75531423e+00 -8.51376832e-01 -6.24853730e-01 2.11790133e+00
6.34877563e-01 1.87599510e-01 -1.39133930e-01 2.73692608e-01
1.03315222e+00 5.90631008e-01 -5.69180250e-01 -3.99697907e-02
-4.09067981e-02 2.40408257e-01 4.29727435e-01 2.82370418e-01
-8.97501528e-01 6.32996738e-01 4.76951075e+00 1.01129365e+00
-1.01823223e+00 -3.65100384e-01 7.75106311e-01 2.39089012e-01
-7.74433732e-01 -1.76329121e-01 -5.91930091e-01 2.74420798e-01
4.28414673e-01 -2.28943184e-01 4.56598699e-01 8.25626373e-01
2.05895994e-02 2.26894543e-01 -7.10282743e-01 9.34512019e-01
-2.91278034e-01 -1.27178240e+00 2.93708324e-01 1.18280418e-01
6.38034165e-01 6.42933846e-02 -1.83415767e-02 4.07042414e-01
4.27531451e-01 -9.33217347e-01 1.79992557e-01 4.18578535e-01
5.90447605e-01 -1.04349399e+00 7.79462636e-01 2.94342548e-01
-1.69945276e+00 2.95090433e-02 -9.08855021e-01 5.48798777e-03
-1.57151505e-01 1.26919997e+00 -7.23316431e-01 7.93166995e-01
4.65865552e-01 1.14404809e+00 -5.56366920e-01 1.10307872e+00
-3.57182771e-01 6.64903343e-01 -4.70838368e-01 -4.78422403e-01
3.70852560e-01 -6.20258570e-01 5.27309060e-01 9.08093870e-01
3.14371616e-01 6.65241554e-02 1.51165396e-01 9.37519491e-01
-4.11024719e-01 3.52095127e-01 -5.12816966e-01 -2.21898630e-01
4.74008799e-01 1.51167250e+00 -1.02447069e+00 -2.81893849e-01
-4.17056918e-01 6.83536887e-01 6.31133914e-01 6.52232468e-01
-5.05066633e-01 -6.00727320e-01 7.55425632e-01 -1.17617227e-01
4.18578863e-01 -1.47848502e-01 -4.43037562e-02 -1.04191446e+00
5.68859950e-02 -4.44658041e-01 3.80961031e-01 -3.41428488e-01
-1.59908175e+00 5.14394104e-01 -3.87574911e-01 -1.23684859e+00
5.35427928e-02 -4.57160383e-01 -8.17500353e-01 1.03254151e+00
-1.38628364e+00 -8.22148383e-01 -4.98758614e-01 5.20505428e-01
3.11173890e-02 -6.19297624e-02 5.81701636e-01 2.98674166e-01
-6.85174465e-01 6.09270632e-01 2.09375858e-01 4.39620197e-01
1.77238852e-01 -1.01704657e+00 5.36599159e-01 5.87515414e-01
2.45377079e-01 7.69767523e-01 1.99200541e-01 -7.91556001e-01
-1.33558345e+00 -1.28518474e+00 5.45367181e-01 4.47344691e-01
6.11108243e-01 -3.74823451e-01 -1.30009186e+00 4.40514624e-01
-3.42317253e-01 2.53465593e-01 5.03375411e-01 3.52193266e-01
-2.16514185e-01 -2.31389791e-01 -8.91140938e-01 6.71126604e-01
1.04351723e+00 -2.55761236e-01 -1.64315611e-01 2.00629026e-01
5.55630147e-01 -1.51182469e-02 -6.49664998e-01 4.41768318e-01
2.26127252e-01 -7.86720753e-01 1.02885735e+00 -4.77022648e-01
1.99527487e-01 -2.62640238e-01 3.15989494e-01 -1.74730515e+00
-6.05722368e-01 -4.45048779e-01 -1.65703714e-01 1.20068419e+00
4.47482377e-01 -9.17937160e-01 1.01136494e+00 2.16881663e-01
9.54786781e-03 -9.00758803e-01 -7.11022437e-01 -5.11923075e-01
-2.20426470e-01 -4.21669662e-01 6.73187733e-01 9.36044753e-01
-2.23576888e-01 6.32833600e-01 -1.94145679e-01 1.40435159e-01
5.80882311e-01 -5.05479090e-02 6.08620465e-01 -1.77471364e+00
-1.90718278e-01 -5.93663633e-01 -7.73386955e-01 -1.26169550e+00
1.14238024e-01 -1.26418364e+00 -2.38509759e-01 -1.83297110e+00
1.17428444e-01 -7.76809931e-01 -4.30582106e-01 3.57675254e-01
-3.63930434e-01 -2.70840619e-03 -7.79160485e-02 7.80291259e-02
-3.16686809e-01 7.49673545e-01 1.48908770e+00 -2.33401760e-01
-4.00428087e-01 2.70544261e-01 -6.58696473e-01 5.44821620e-01
8.97116899e-01 -3.36346090e-01 -6.69837713e-01 -1.29958406e-01
4.51646954e-01 -1.77691445e-01 2.48359501e-01 -9.68769670e-01
4.12940327e-03 1.64738953e-01 7.00647533e-01 -7.00220883e-01
2.62773544e-01 -8.37341905e-01 9.52666849e-02 5.65896034e-01
-3.67650926e-01 -2.59727240e-01 2.60893386e-02 9.92245972e-01
-2.35459581e-02 -2.18471184e-01 6.01144135e-01 3.96414287e-02
-4.94429439e-01 6.68039262e-01 -2.71390438e-01 -1.73331141e-01
8.22738051e-01 -1.30415738e-01 -2.83351898e-01 -6.54394329e-01
-7.35535502e-01 3.79583538e-01 3.12704951e-01 1.40359849e-01
7.33741879e-01 -1.46043599e+00 -6.86183631e-01 5.31099737e-01
-1.00212060e-02 2.61209697e-01 2.08128095e-01 7.83504128e-01
-3.26524347e-01 -1.17626086e-01 6.36257827e-02 -5.60852528e-01
-1.10344684e+00 3.91701251e-01 1.16529919e-01 -3.74730319e-01
-7.03072488e-01 9.87045765e-01 3.02516133e-01 -4.32172149e-01
3.51639479e-01 -4.16837394e-01 -5.35549641e-01 3.16585124e-01
4.41262424e-01 2.22110003e-01 -1.20471381e-02 -4.98000741e-01
-1.85854003e-01 2.64685243e-01 -2.79198110e-01 3.43534559e-01
1.43138933e+00 7.21797124e-02 -4.10458922e-01 4.22493517e-01
1.48866475e+00 -5.69485612e-02 -8.62576365e-01 -4.54587668e-01
-1.40832558e-01 -4.38482046e-01 1.94563180e-01 -1.44369960e-01
-1.73604488e+00 1.02505744e+00 1.59249857e-01 6.49892211e-01
1.02042639e+00 7.72091374e-02 6.22028708e-01 4.97098625e-01
2.17460319e-01 -6.79685235e-01 6.16967380e-02 4.90347415e-01
7.40535140e-01 -8.22626770e-01 3.77978645e-02 -8.45752299e-01
-3.67181629e-01 1.30253184e+00 5.77960134e-01 -2.59505630e-01
7.54042208e-01 -1.41170815e-01 -3.70257348e-01 -2.72630215e-01
-5.51611900e-01 -1.87534004e-01 3.62479419e-01 7.14974582e-01
3.93402159e-01 1.63478389e-01 -2.91823000e-01 5.33137321e-01
5.65549731e-02 -5.21079957e-01 3.99569720e-01 4.78111625e-01
-4.50453579e-01 -1.06941450e+00 -1.11472331e-01 1.02297819e+00
8.11458305e-02 -1.18881874e-01 -3.65811795e-01 7.53487408e-01
-4.72425163e-01 6.96591198e-01 2.51557738e-01 -5.79249382e-01
1.08976670e-01 -2.15092316e-01 2.43548244e-01 -6.38193667e-01
-4.42402333e-01 -5.54501414e-02 9.14094076e-02 -2.62965113e-01
-1.51840746e-01 -4.70057517e-01 -1.56902432e+00 -5.81222951e-01
-4.00125384e-01 3.40248466e-01 4.33211118e-01 4.36258584e-01
3.65380317e-01 8.61618042e-01 8.34240437e-01 -7.76735365e-01
-4.92374301e-01 -1.02706289e+00 -8.16440821e-01 3.10811758e-01
1.64721891e-01 -5.55347860e-01 -5.87549627e-01 -6.74763680e-01] | [7.194202423095703, 6.194403648376465] |
744e19b1-f7e6-4d81-aa11-758622d9b028 | encoding-sentiment-information-into-word | null | null | https://aclanthology.org/C18-1085 | https://aclanthology.org/C18-1085.pdf | Encoding Sentiment Information into Word Vectors for Sentiment Analysis | General-purpose pre-trained word embeddings have become a mainstay of natural language processing, and more recently, methods have been proposed to encode external knowledge into word embeddings to benefit specific downstream tasks. The goal of this paper is to encode sentiment knowledge into pre-trained word vectors to improve the performance of sentiment analysis. Our proposed method is based on a convolutional neural network (CNN) and an external sentiment lexicon. Experiments on four popular sentiment analysis datasets show that this method improves the accuracy of sentiment analysis compared to a number of benchmark methods. | ['Timothy Baldwin', 'Fang Li', 'Zhe Ye'] | 2018-08-01 | encoding-sentiment-information-into-word-1 | https://aclanthology.org/C18-1085 | https://aclanthology.org/C18-1085.pdf | coling-2018-8 | ['learning-word-embeddings'] | ['methodology'] | [ 6.88268915e-02 3.04387906e-03 -3.56485963e-01 -7.24157453e-01
-2.83719033e-01 -4.35881764e-01 7.00916886e-01 4.45703804e-01
-9.16492641e-01 3.70312154e-01 6.72927022e-01 -2.20401451e-01
4.57549602e-01 -9.93733466e-01 -1.16615601e-01 -4.22591686e-01
3.29994410e-01 -1.17385224e-01 6.88191205e-02 -6.53546810e-01
3.56578052e-01 8.90768468e-02 -1.21618736e+00 2.61538535e-01
4.59085763e-01 1.12703729e+00 -1.37000129e-01 4.40270513e-01
-5.65697014e-01 5.40989816e-01 -5.83284259e-01 -7.95096099e-01
1.43293887e-01 -1.31013349e-01 -7.37800002e-01 -2.79103935e-01
-3.29373538e-01 1.28622204e-01 -9.78474170e-02 1.04795337e+00
4.50610340e-01 3.68120372e-01 6.04952097e-01 -8.08665931e-01
-1.17934096e+00 6.17170036e-01 -4.17215735e-01 3.25146973e-01
3.95111479e-02 -1.06424592e-01 1.51951683e+00 -1.05918229e+00
2.58206874e-01 1.08387530e+00 7.89128125e-01 3.30907196e-01
-8.05640638e-01 -7.09073901e-01 3.20782751e-01 1.12565286e-01
-1.06055117e+00 -1.47685483e-01 9.98514891e-01 -2.93058366e-01
1.33569336e+00 -2.92980820e-01 6.54263377e-01 9.08252239e-01
4.63950962e-01 6.99137032e-01 7.17509747e-01 -5.49823582e-01
1.96706310e-01 5.04777133e-01 5.19074738e-01 5.84789693e-01
5.09146035e-01 -3.81352305e-01 -6.00011349e-01 3.78998369e-02
2.75065601e-01 1.84649318e-01 8.70449767e-02 -1.76691785e-01
-7.99760461e-01 1.32926750e+00 6.59128964e-01 4.85783875e-01
-2.65980661e-01 2.31070563e-01 8.04872274e-01 9.95808542e-02
1.09517133e+00 8.96394789e-01 -9.24123466e-01 -2.71034032e-01
-4.94744658e-01 -6.89705536e-02 6.52799964e-01 3.60953063e-01
7.49142468e-01 1.72124729e-02 -2.37011425e-02 9.77821231e-01
3.50867748e-01 4.33153749e-01 1.21003330e+00 8.11135545e-02
3.85157913e-01 1.09873259e+00 -2.09416196e-01 -1.39376223e+00
-6.06618285e-01 -3.17082226e-01 -5.38475096e-01 -1.93409383e-01
-1.40506014e-01 -3.64175081e-01 -8.99704278e-01 1.44006407e+00
1.39080003e-01 -9.44229364e-02 3.90404433e-01 6.78776145e-01
8.25236678e-01 7.54579663e-01 3.26679409e-01 2.72816926e-01
1.44063461e+00 -1.22261047e+00 -9.31696236e-01 -5.60304046e-01
1.02154517e+00 -8.45922589e-01 1.11608708e+00 1.80661604e-01
-5.62105775e-01 -6.41061008e-01 -9.60202992e-01 -2.49978796e-01
-1.18423986e+00 1.09978346e-02 1.04707479e+00 9.02073622e-01
-7.11658418e-01 1.05512850e-01 -7.37477779e-01 -3.56586665e-01
7.53326833e-01 2.78989285e-01 -5.58201253e-01 1.58976745e-02
-1.55573165e+00 1.04650056e+00 5.13391554e-01 6.82482868e-02
-1.98998004e-01 -6.13463104e-01 -1.23622668e+00 2.89055109e-01
-1.17652774e-01 -4.16651636e-01 1.12028491e+00 -1.15429509e+00
-1.58594966e+00 7.42309511e-01 -2.27726325e-01 -2.35160396e-01
-3.77221435e-01 -5.97804010e-01 -7.03075945e-01 -1.71240211e-01
6.52476102e-02 4.66625065e-01 5.63610673e-01 -8.51162791e-01
-5.75621009e-01 -3.67958486e-01 1.33524612e-01 1.71984285e-01
-1.41836882e+00 6.82592243e-02 -6.00587189e-01 -7.09924877e-01
-3.36000144e-01 -6.26602054e-01 -5.82384348e-01 -3.52912843e-01
-1.59463391e-01 -5.95360756e-01 7.26287544e-01 -2.55376160e-01
1.40856838e+00 -2.06325865e+00 2.86174640e-02 1.06447428e-01
1.49081042e-02 6.73128605e-01 -3.66286516e-01 5.12758553e-01
-1.18472576e-01 2.32205108e-01 -1.61024451e-01 -5.92763364e-01
1.12394266e-01 2.51371622e-01 -5.15857100e-01 2.71625042e-01
6.29453599e-01 1.26838291e+00 -1.03405070e+00 -1.00122273e-01
2.05359712e-01 6.53597057e-01 -7.04840362e-01 2.03185171e-01
-8.71277377e-02 -1.15381710e-01 -6.55414581e-01 4.00712043e-01
3.68685782e-01 -2.04177842e-01 1.24223113e-01 -2.49886394e-01
-4.69889631e-03 4.92641658e-01 -6.15574896e-01 1.58770752e+00
-7.29992032e-01 6.02048814e-01 -4.85893250e-01 -1.17165685e+00
1.07472908e+00 1.67462885e-01 9.20656025e-02 -6.70378506e-01
8.30970645e-01 -1.02889128e-01 -8.64001364e-02 -4.24189121e-01
8.73484135e-01 -5.33316553e-01 -5.74042618e-01 5.28822184e-01
3.61506581e-01 -3.51428568e-01 2.71725714e-01 2.14383215e-01
9.20922041e-01 -2.25754380e-01 5.27430654e-01 -1.60578087e-01
7.18733907e-01 4.09956574e-02 4.58615422e-01 1.20969899e-01
-1.55545145e-01 4.40182358e-01 5.32798648e-01 -6.46830678e-01
-7.00680792e-01 -4.63555634e-01 -1.43622324e-01 1.31471419e+00
3.67876105e-02 -8.32347691e-01 -4.48013932e-01 -9.17051017e-01
1.26561239e-01 4.10141826e-01 -9.74416554e-01 -5.69739580e-01
-3.37971956e-01 -9.21468437e-01 4.81165290e-01 9.67882276e-01
1.13347434e-01 -1.21127677e+00 -2.46338561e-01 3.34197015e-01
2.52566367e-01 -1.34851944e+00 -1.56029135e-01 5.92598438e-01
-7.82287300e-01 -8.86149585e-01 -3.74900669e-01 -1.01108944e+00
8.07887018e-01 2.64607757e-01 9.17358279e-01 3.71013843e-02
-8.02547298e-03 9.72079560e-02 -9.06218886e-01 -7.68971264e-01
1.72697946e-01 5.54498494e-01 2.03481227e-01 1.17980860e-01
1.16180050e+00 -1.01699077e-01 -4.65459913e-01 -1.10352352e-01
-1.30204260e+00 -2.05628321e-01 4.22747523e-01 9.55335915e-01
4.33905423e-01 -5.90959191e-02 9.23811316e-01 -1.29028010e+00
1.18473077e+00 -5.95739841e-01 -4.12125498e-01 -2.56447941e-01
-7.31403708e-01 1.96550146e-01 1.05304527e+00 -2.21674934e-01
-9.83265281e-01 -1.23999231e-01 -5.88757694e-01 4.67473194e-02
8.87583047e-02 1.12070334e+00 1.13198254e-02 1.02652781e-01
3.91100585e-01 -1.02713834e-02 -1.35959387e-01 -4.12182212e-01
7.34958708e-01 8.90936971e-01 1.48560166e-01 -2.53788471e-01
6.40509963e-01 4.38308269e-01 -3.75641286e-01 -6.20414615e-01
-1.50487757e+00 -7.77569830e-01 -6.02743924e-01 2.52698392e-01
9.26720560e-01 -9.87883270e-01 -1.58467069e-01 3.23536307e-01
-1.10953128e+00 -1.01330668e-01 -2.93073684e-01 4.11412597e-01
-3.55741084e-02 1.40658885e-01 -4.58205909e-01 -3.77646357e-01
-5.88053048e-01 -1.14354682e+00 9.11449552e-01 4.24787074e-01
-5.02497435e-01 -1.61276507e+00 4.94974852e-01 2.85409331e-01
7.11313009e-01 -1.96235269e-01 7.16194272e-01 -1.01646614e+00
3.24896932e-01 -6.79187179e-01 -3.04734141e-01 7.95747697e-01
4.41147625e-01 -2.94391781e-01 -1.18361843e+00 -1.20831229e-01
-2.55385011e-01 -4.25952017e-01 1.17886508e+00 1.18389010e-01
1.10410762e+00 1.07028648e-01 -3.07120264e-01 6.25567436e-01
1.42759097e+00 -6.42546341e-02 3.62537950e-01 5.86748004e-01
6.19990885e-01 5.28897762e-01 4.22846019e-01 4.87006426e-01
4.64216918e-01 1.74412027e-01 2.79889494e-01 -2.19780043e-01
1.32515445e-01 -1.58276096e-01 3.51011366e-01 1.36849952e+00
3.28952312e-01 -2.75018126e-01 -9.41717684e-01 9.91044521e-01
-1.66474676e+00 -3.80533725e-01 2.67566293e-02 1.35858655e+00
8.20899844e-01 3.51786315e-01 -4.26462740e-01 1.49119660e-01
3.02508742e-01 4.98303980e-01 -4.68028724e-01 -1.05917811e+00
-6.21417090e-02 8.19206357e-01 3.76935929e-01 3.16362441e-01
-1.22017956e+00 1.46710491e+00 6.36493492e+00 3.73827368e-01
-1.31025255e+00 2.48155177e-01 3.78894091e-01 6.33681267e-02
-3.88670832e-01 -3.64985406e-01 -9.01184618e-01 1.91446096e-01
1.08152187e+00 -2.78267473e-01 -4.51865226e-01 9.94173944e-01
-7.65080601e-02 3.02551538e-01 -7.40500689e-01 5.79610348e-01
4.51912135e-01 -1.22274697e+00 2.54282176e-01 -3.81123841e-01
9.97950494e-01 7.56444931e-02 2.15848938e-01 8.32578778e-01
4.50583518e-01 -1.09037101e+00 9.21116769e-02 7.52360672e-02
7.13355601e-01 -1.08353758e+00 1.41438329e+00 -2.04243645e-01
-1.23486650e+00 2.33384315e-02 -5.71044087e-01 -4.98538494e-01
3.87017548e-01 7.96693444e-01 -5.94319701e-01 3.34293455e-01
7.78408825e-01 1.14482486e+00 -6.07671857e-01 3.99480015e-01
-7.42000043e-01 7.33265817e-01 3.71159129e-02 -3.81234497e-01
6.58763826e-01 1.10465012e-01 -6.24150895e-02 1.51333892e+00
-1.45188451e-01 5.81654906e-02 -6.52119219e-02 3.88550907e-01
-5.22196352e-01 4.52179939e-01 -5.25649905e-01 -7.64645875e-01
-2.27830157e-01 1.64099312e+00 -6.43719494e-01 -3.06019485e-01
-8.88923347e-01 8.26977789e-01 5.86110592e-01 2.04621986e-01
-4.36590105e-01 -9.61061776e-01 1.34475398e+00 -4.94249582e-01
6.77671552e-01 -1.88821137e-01 -4.49908823e-01 -1.41151702e+00
-2.51659006e-01 -5.57371080e-01 2.88196325e-01 -5.99118590e-01
-1.56779623e+00 8.32043648e-01 -6.04043365e-01 -9.14804697e-01
5.91418520e-02 -1.19735110e+00 -7.11111307e-01 9.73589957e-01
-2.16040421e+00 -1.08418941e+00 -6.80027381e-02 4.35877889e-01
6.40558302e-01 -4.16871846e-01 1.23321307e+00 2.08263651e-01
-5.79344809e-01 6.87590420e-01 1.49280578e-01 4.80681807e-01
8.46673012e-01 -1.10404623e+00 6.26685381e-01 7.90688515e-01
1.86764687e-01 9.01205719e-01 4.47523683e-01 -3.92120779e-01
-1.38080502e+00 -1.27414298e+00 1.15894330e+00 -5.49124122e-01
1.11012757e+00 -5.02468348e-01 -7.16293037e-01 7.06043303e-01
4.72963244e-01 2.45624870e-01 1.38452673e+00 5.00489593e-01
-5.76231658e-01 -1.43101171e-01 -7.59523034e-01 4.64860976e-01
4.71083283e-01 -5.76382041e-01 -9.52864707e-01 -1.58186276e-02
9.62872565e-01 5.09790145e-02 -6.15479767e-01 2.79202402e-01
6.09411120e-01 -2.83586949e-01 7.68851101e-01 -1.06973624e+00
9.40183759e-01 -1.26259834e-01 -1.23001307e-01 -1.87869203e+00
-2.62224823e-01 -5.42996302e-02 -1.26048967e-01 1.14267945e+00
5.78470945e-01 -6.36989892e-01 7.47396827e-01 4.40420955e-01
-1.19816139e-01 -1.03182924e+00 -3.61790597e-01 -3.78438860e-01
3.25815290e-01 -6.82990491e-01 5.26603758e-01 1.02803326e+00
5.12223125e-01 7.91555166e-01 -8.42319280e-02 -4.24843729e-02
-1.13459066e-01 1.94604248e-01 6.98907912e-01 -1.26329291e+00
1.87162295e-01 -5.54341972e-01 -7.31744647e-01 -1.10685003e+00
6.03515327e-01 -1.18021655e+00 -1.30265551e-02 -1.87904525e+00
9.31584686e-02 -1.65197119e-01 -1.07667816e+00 6.26712441e-01
-4.94297713e-01 6.48871958e-01 -8.93372819e-02 -5.49448967e-01
-6.59589231e-01 8.47447693e-01 1.13661075e+00 -2.48436466e-01
-3.53236571e-02 -4.77447420e-01 -1.43558347e+00 8.36392581e-01
9.33053076e-01 -6.71010256e-01 -2.74026155e-01 -6.49399936e-01
1.00652349e+00 -8.20412397e-01 -4.87509608e-01 -5.09697855e-01
2.55637556e-01 -1.05618313e-02 3.45808297e-01 -3.61622244e-01
2.44769067e-01 -7.87362397e-01 -9.06175435e-01 2.67328858e-01
-4.32188421e-01 1.67768344e-01 5.14882386e-01 5.42762220e-01
-7.16915905e-01 -2.70642847e-01 5.05245030e-01 1.46746755e-01
-1.05075014e+00 1.66024938e-01 -3.89541179e-01 3.50881256e-02
9.11202312e-01 1.74745023e-01 -6.42946959e-02 -1.81193009e-01
-1.36060253e-01 2.07477078e-01 7.71887004e-02 9.16850984e-01
7.04464316e-01 -1.29924786e+00 -3.22848648e-01 4.53799427e-01
6.33664191e-01 -2.39639729e-01 -3.08151603e-01 4.82149720e-01
-2.52345175e-01 7.53224850e-01 -7.80596659e-02 -7.28951171e-02
-1.08932555e+00 2.97623366e-01 -1.25266671e-01 -5.05042911e-01
-3.56963128e-01 1.14402092e+00 2.64431149e-01 -7.05838978e-01
-1.42801538e-01 -5.17965794e-01 -8.82835746e-01 3.55816543e-01
8.09911191e-01 -3.56975973e-01 2.28398755e-01 -7.70397544e-01
-4.16933775e-01 6.74294531e-01 -4.01903927e-01 2.14090347e-01
1.80621672e+00 -3.45375836e-02 -1.17664233e-01 4.79157627e-01
1.64494157e+00 4.82816510e-02 -4.80610102e-01 -3.77216190e-01
1.51786417e-01 -4.15295064e-01 5.13281167e-01 -5.12366354e-01
-1.27653062e+00 1.08309162e+00 1.74774379e-01 3.31200689e-01
1.00442743e+00 -1.28444016e-01 1.15572071e+00 5.21300197e-01
1.05943941e-02 -1.08820701e+00 2.51552671e-01 1.10271764e+00
3.17751229e-01 -1.46676135e+00 -1.38779003e-02 5.47198728e-02
-8.04335177e-01 1.36990285e+00 6.21766090e-01 -5.06967485e-01
9.69993591e-01 2.32978597e-01 4.51718807e-01 -2.83348113e-01
-6.46661520e-01 -4.38742548e-01 4.30003911e-01 2.81319439e-01
8.22219968e-01 -1.12982333e-01 -6.43384218e-01 1.31496537e+00
-2.78447688e-01 -2.30979137e-02 4.40151215e-01 1.01661325e+00
-3.96303266e-01 -1.42014468e+00 2.24345595e-01 5.81546426e-01
-9.31044519e-01 -3.62438560e-01 -4.06716287e-01 4.86651272e-01
-4.03784700e-02 1.02221572e+00 1.46857470e-01 -4.34407741e-01
4.22098458e-01 2.22045675e-01 -1.01621419e-01 -1.20656681e+00
-8.96858752e-01 -4.43511188e-01 4.31640036e-02 -4.87158954e-01
-5.49782932e-01 -1.35201603e-01 -1.25452006e+00 -7.87872064e-04
-5.27984142e-01 2.92938739e-01 9.81264293e-01 1.03090000e+00
4.29123968e-01 8.64867866e-01 5.99438310e-01 -5.36292434e-01
-9.98814330e-02 -1.07755899e+00 -4.85293657e-01 6.91319108e-01
2.53738850e-01 -5.38786650e-01 -4.91929293e-01 6.23813719e-02] | [10.539731979370117, 8.398943901062012] |
0aeaad2d-0fc6-4af4-a225-efa15168a353 | meta-gmvae-mixture-of-gaussian-vae-for | null | null | https://openreview.net/forum?id=wS0UFjsNYjn | https://openreview.net/pdf?id=wS0UFjsNYjn | Meta-GMVAE: Mixture of Gaussian VAE for Unsupervised Meta-Learning | Unsupervised learning aims to learn meaningful representations from unlabeled data which can captures its intrinsic structure, that can be transferred to downstream tasks. Meta-learning, whose objective is to learn to generalize across tasks such that the learned model can rapidly adapt to a novel task, shares the spirit of unsupervised learning in that the both seek to learn more effective and efficient learning procedure than learning from scratch. The fundamental difference of the two is that the most meta-learning approaches are supervised, assuming full access to the labels. However, acquiring labeled dataset for meta-training not only is costly as it requires human efforts in labeling but also limits its applications to pre-defined task distributions. In this paper, we propose a principled unsupervised meta-learning model, namely Meta-GMVAE, based on Variational Autoencoder (VAE) and set-level variational inference. Moreover, we introduce a mixture of Gaussian (GMM) prior, assuming that each modality represents each class-concept in a randomly sampled episode, which we optimize with Expectation-Maximization (EM). Then, the learned model can be used for downstream few-shot classification tasks, where we obtain task-specific parameters by performing semi-supervised EM on the latent representations of the support and query set, and predict labels of the query set by computing aggregated posteriors. We validate our model on Omniglot and Mini-ImageNet datasets by evaluating its performance on downstream few-shot classification tasks. The results show that our model obtain impressive performance gains over existing unsupervised meta-learning baselines, even outperforming supervised MAML on a certain setting. | ['Sung Ju Hwang', 'Seanie Lee', 'Dongchan Min', 'Dong Bok Lee'] | 2021-01-01 | null | null | null | iclr-2021-1 | ['unsupervised-few-shot-learning', 'unsupervised-few-shot-image-classification'] | ['computer-vision', 'computer-vision'] | [ 2.64994860e-01 2.61283338e-01 -6.08136773e-01 -5.37038982e-01
-1.00028551e+00 -2.68115729e-01 8.54096889e-01 -1.26658678e-02
-5.07796943e-01 5.28854787e-01 1.53503448e-01 1.97700679e-01
-2.37083621e-02 -6.98610246e-01 -7.49847174e-01 -7.75351763e-01
3.90471548e-01 6.16903961e-01 -6.98797256e-02 1.36809677e-01
-9.91127044e-02 -4.43613939e-02 -1.73218572e+00 3.91251773e-01
6.69768691e-01 8.86245966e-01 5.22940516e-01 3.41847062e-01
-1.04760863e-01 8.96556199e-01 -2.88372070e-01 -3.77391517e-01
-7.67821074e-02 -5.63464701e-01 -8.17631245e-01 4.79900956e-01
2.65332889e-02 -1.91389993e-01 -2.67115563e-01 9.99810398e-01
5.57439923e-02 7.20683634e-01 1.43459976e+00 -1.15957832e+00
-6.98533654e-01 5.31254351e-01 -2.59373426e-01 5.88722294e-03
-3.66328895e-01 2.30298936e-01 1.12237811e+00 -1.09701300e+00
6.86944842e-01 1.05869186e+00 3.94698709e-01 8.43645513e-01
-1.53602302e+00 -2.48980597e-01 1.68180197e-01 2.19981298e-01
-1.28636801e+00 -6.15597844e-01 6.51165724e-01 -5.42126656e-01
9.42173481e-01 -2.91035116e-01 3.07723194e-01 1.64016843e+00
-9.76070762e-02 1.01323652e+00 8.15075636e-01 -4.76253539e-01
7.29241729e-01 2.59188652e-01 8.79574269e-02 5.59542120e-01
-3.68468225e-01 8.58179554e-02 -4.71980125e-01 2.23225206e-02
3.32326055e-01 4.28322136e-01 3.70126963e-02 -5.44726610e-01
-1.04575610e+00 1.13567650e+00 1.99601948e-01 2.90556729e-01
-4.81056720e-01 2.54860759e-01 3.71475637e-01 2.20062539e-01
8.05329800e-01 1.47544980e-01 -5.04377902e-01 1.43512264e-02
-1.07625067e+00 -1.06782511e-01 5.38359344e-01 1.04199374e+00
1.11871898e+00 1.96867689e-01 -4.30922896e-01 1.10927629e+00
5.89731634e-01 3.03493708e-01 9.62323248e-01 -1.19875884e+00
2.19532952e-01 2.71217018e-01 -1.64950550e-01 -2.00975850e-01
-1.58115849e-01 -3.17871898e-01 -6.53225422e-01 -4.33484130e-02
-6.28408277e-03 -2.67082036e-01 -1.24376178e+00 2.03435254e+00
2.22176552e-01 5.84244192e-01 2.74377435e-01 5.28074145e-01
7.08510518e-01 8.64130974e-01 4.62483287e-01 -5.96163452e-01
1.00940812e+00 -1.18609107e+00 -6.34879827e-01 -3.23990911e-01
8.07838857e-01 -1.02884941e-01 1.10315859e+00 1.75742269e-01
-9.09140825e-01 -8.12168896e-01 -9.08009410e-01 -8.92973542e-02
-4.40614760e-01 -1.19213559e-01 4.08600658e-01 4.74931538e-01
-8.93513620e-01 6.37854338e-01 -1.03993666e+00 -5.05411923e-01
5.43205917e-01 1.60569139e-02 -4.56449855e-03 -6.90486059e-02
-1.02253807e+00 7.18487680e-01 7.89993227e-01 -3.99093270e-01
-1.61723959e+00 -8.49614918e-01 -1.13146687e+00 2.40299612e-01
5.81768453e-01 -8.41313362e-01 1.57860231e+00 -9.44850802e-01
-1.64174163e+00 8.88197184e-01 -3.33571732e-01 -5.30359447e-01
1.28203586e-01 6.28100932e-02 -2.80426502e-01 7.62213096e-02
9.07860249e-02 9.11290467e-01 1.28988039e+00 -1.21883106e+00
-5.88545263e-01 -2.91544765e-01 -1.60172939e-01 1.28654212e-01
-5.26674807e-01 -4.93489891e-01 -4.36210126e-01 -5.30372024e-01
-1.18660137e-01 -7.28689075e-01 -2.89765805e-01 -4.31784898e-01
-1.27941519e-01 -5.74313760e-01 8.10523927e-01 -2.29410350e-01
9.67031658e-01 -2.34039330e+00 4.43226308e-01 -1.95422217e-01
6.71492219e-02 1.52161136e-01 -2.26216301e-01 4.66548055e-01
2.22291902e-01 -1.98059723e-01 -4.77312893e-01 -9.04769838e-01
1.92751780e-01 5.25235474e-01 -6.24054790e-01 3.11732143e-01
2.10953191e-01 1.25336254e+00 -1.06195164e+00 -4.67521489e-01
3.63009036e-01 2.98416317e-01 -5.48582673e-01 3.80257964e-01
-7.45152414e-01 4.32169080e-01 -3.25204253e-01 4.68290716e-01
1.08686492e-01 -5.85467994e-01 2.96813637e-01 -5.22198826e-02
2.38210365e-01 -9.32559073e-02 -7.85752654e-01 2.22351909e+00
-6.44502223e-01 4.29079622e-01 -3.57209086e-01 -1.62634468e+00
6.99697375e-01 4.62389708e-01 5.39865792e-01 -4.45850492e-01
2.64458120e-01 -1.29805878e-01 -4.90647525e-01 -6.46009803e-01
1.46069378e-01 -4.09989595e-01 -9.14674774e-02 6.81677341e-01
9.25223887e-01 -4.38741110e-02 1.26446411e-01 2.72015333e-01
7.44327962e-01 3.76527786e-01 2.47784749e-01 -6.88993081e-04
1.33144200e-01 3.68270464e-03 5.38614571e-01 9.14579093e-01
-1.10450253e-01 4.50521767e-01 4.21970226e-02 -2.49481499e-02
-8.60832751e-01 -1.35260570e+00 -1.05745681e-01 1.52338719e+00
-1.17293596e-01 -3.10545653e-01 -6.70635760e-01 -7.85492718e-01
-1.58734962e-01 1.19491971e+00 -7.08043277e-01 -5.69646835e-01
1.60093512e-02 -6.69603288e-01 1.02265358e-01 6.39678001e-01
2.51785010e-01 -1.25148761e+00 -6.50354445e-01 2.31038973e-01
-1.06182136e-01 -1.10822070e+00 -1.84288099e-01 3.54716271e-01
-1.22271943e+00 -8.11892629e-01 -7.15815246e-01 -9.29282248e-01
5.37588954e-01 2.08947688e-01 1.01946211e+00 -4.31640714e-01
-1.31339327e-01 9.22454238e-01 -3.44423681e-01 -2.48664424e-01
-3.70549589e-01 8.42214599e-02 2.47445390e-01 2.94893801e-01
6.54208839e-01 -7.18764663e-01 -2.49812081e-01 -3.42535861e-02
-1.02763283e+00 -1.54848192e-02 4.88911599e-01 1.05006170e+00
8.80314469e-01 -1.88503005e-02 8.94081414e-01 -1.21789503e+00
3.53282809e-01 -9.71963227e-01 -2.85544068e-01 3.71867597e-01
-7.89302826e-01 1.52093872e-01 5.46787262e-01 -5.94577909e-01
-1.49238682e+00 -2.85246447e-02 1.45259470e-01 -9.93026197e-01
-3.97083521e-01 7.46985555e-01 -1.37472957e-01 7.12988853e-01
6.27772152e-01 4.54540849e-01 -9.08382386e-02 -5.76736152e-01
9.36645150e-01 7.31978297e-01 5.37631810e-01 -6.29772007e-01
6.02848470e-01 6.99496269e-01 -3.26005429e-01 -1.07004523e+00
-1.43566811e+00 -4.61969316e-01 -8.64469409e-01 -1.43321589e-01
1.05804241e+00 -1.07188272e+00 -3.67996842e-01 1.52781636e-01
-7.86852300e-01 -7.36057103e-01 -7.02147961e-01 5.33470392e-01
-1.01131356e+00 1.30817398e-01 -5.30536592e-01 -9.34100211e-01
4.78728004e-02 -9.31253970e-01 1.00427008e+00 1.73095167e-01
-1.65562317e-01 -1.50280726e+00 3.14838260e-01 3.28278214e-01
1.44898564e-01 -6.15042227e-04 1.10685205e+00 -7.91858912e-01
-2.72103637e-01 1.21526811e-02 1.21160001e-01 5.36446452e-01
1.24403425e-01 -3.32148343e-01 -1.41912937e+00 -4.03259903e-01
2.91695535e-01 -1.01988518e+00 1.26213801e+00 5.74239373e-01
1.37317145e+00 -1.04871072e-01 -4.83847976e-01 6.93001688e-01
1.27279282e+00 -8.16233978e-02 2.30983242e-01 -1.07122604e-02
4.77014959e-01 7.12815762e-01 6.84622288e-01 4.64372396e-01
3.66793066e-01 3.78313750e-01 2.90619045e-01 3.90993029e-01
7.87702128e-02 -4.84099090e-01 5.04219830e-01 8.55462074e-01
8.30819383e-02 -4.94121388e-03 -5.86238444e-01 6.11450016e-01
-2.18901682e+00 -9.94965196e-01 5.09334981e-01 2.09005737e+00
7.59556770e-01 -1.17146038e-01 6.85184896e-02 -3.88032198e-01
5.20664275e-01 2.99461335e-01 -9.49171305e-01 5.29883755e-03
1.72531292e-01 2.71831274e-01 -8.51945207e-02 2.53872395e-01
-1.12040663e+00 1.12766337e+00 6.05126715e+00 8.67071390e-01
-8.30849946e-01 7.20606565e-01 5.76306164e-01 -3.51509988e-01
-1.77898392e-01 1.34104937e-02 -6.61344945e-01 3.99406433e-01
1.32744801e+00 -2.01046690e-01 4.41341251e-01 9.19559479e-01
2.77553909e-02 1.95027396e-01 -1.45724773e+00 1.02067256e+00
1.81093872e-01 -1.38503182e+00 7.68241882e-02 8.66774693e-02
1.06231058e+00 3.24398845e-01 4.53779221e-01 1.11181664e+00
4.44395214e-01 -8.16043079e-01 5.47338665e-01 6.21445775e-01
7.10371375e-01 -6.26543880e-01 3.40695262e-01 9.20717716e-01
-8.45478475e-01 -2.12287396e-01 -7.25563169e-01 6.00200929e-02
1.42804191e-01 2.05703825e-01 -5.65949619e-01 2.69804418e-01
4.46493417e-01 1.05638802e+00 -2.07625628e-01 5.90355754e-01
-3.72294724e-01 7.65972257e-01 1.38055876e-01 1.75330535e-01
3.30991060e-01 -2.67242491e-01 3.46750855e-01 1.16641212e+00
2.72947431e-01 1.49859041e-02 4.13984329e-01 1.24646306e+00
-3.43978226e-01 5.20828553e-02 -6.75288498e-01 -3.71263713e-01
4.94197786e-01 1.16730583e+00 -4.19576317e-01 -6.63109779e-01
-4.68179882e-01 1.08458519e+00 6.78622723e-01 7.92184472e-01
-7.24698722e-01 2.89094411e-02 4.13214803e-01 -3.01702291e-01
5.48513174e-01 -3.72199751e-02 1.34572953e-01 -1.45322299e+00
-4.57085937e-01 -5.15285850e-01 7.48256981e-01 -6.30165040e-01
-1.51220453e+00 1.90634862e-01 2.45896623e-01 -1.17228973e+00
-7.90855587e-01 -5.18650353e-01 -7.30670869e-01 5.60899138e-01
-1.52313828e+00 -1.24489582e+00 -1.71385650e-02 7.22497225e-01
1.10305119e+00 -4.22028154e-01 9.35946167e-01 -3.50251645e-02
-6.25244975e-01 4.85936850e-01 4.02288795e-01 4.65633683e-02
5.11819363e-01 -1.08380711e+00 -7.32924929e-03 5.93923748e-01
7.37190068e-01 5.23067534e-01 4.60618645e-01 -4.47211862e-01
-1.17990160e+00 -1.45973134e+00 6.10832214e-01 -6.38053000e-01
6.11075461e-01 -3.80005121e-01 -1.05309463e+00 1.15373147e+00
1.27185494e-01 1.04042098e-01 1.02134776e+00 2.57452101e-01
-3.57873946e-01 2.66867936e-01 -9.28967297e-01 4.63056087e-01
1.01960778e+00 -7.56630599e-01 -9.50157404e-01 5.35970151e-01
9.31297362e-01 7.16606155e-02 -8.58925343e-01 2.15932623e-01
1.48884147e-01 -7.92746603e-01 8.60003352e-01 -1.20591223e+00
5.90681672e-01 8.13268200e-02 -4.51207548e-01 -1.60008335e+00
-3.87598932e-01 -3.10781062e-01 -7.82690108e-01 1.35832202e+00
4.03897792e-01 -2.93113142e-01 6.78425670e-01 4.28599477e-01
-2.48770401e-01 -6.79095447e-01 -6.86981916e-01 -9.49306965e-01
1.64441198e-01 -8.11762154e-01 2.32321039e-01 1.09013116e+00
1.23395063e-01 6.62774980e-01 -3.96398783e-01 1.38554201e-01
9.62624669e-01 1.54486895e-01 6.23816073e-01 -1.23320293e+00
-6.64097905e-01 -2.26172015e-01 -4.27646451e-02 -1.00594628e+00
8.24080586e-01 -1.27937782e+00 2.49363467e-01 -1.39217114e+00
4.31937963e-01 -4.07113060e-02 -7.05885053e-01 4.31820273e-01
-1.10010341e-01 3.00537031e-02 9.32350755e-02 4.97819006e-01
-1.02511966e+00 1.03461611e+00 7.82300234e-01 -4.10650373e-01
-3.83465469e-01 6.50405511e-02 -4.42276150e-01 8.23018730e-01
4.95531559e-01 -6.68750763e-01 -8.12231660e-01 -3.99085104e-01
-1.05701014e-01 -5.78748584e-02 2.66559362e-01 -8.68006051e-01
1.90446153e-01 -2.89615631e-01 3.05388123e-01 -3.05417567e-01
6.50584579e-01 -5.90094090e-01 -2.65069425e-01 2.83622563e-01
-5.58262169e-01 -7.01089799e-01 -1.66546986e-01 1.05789840e+00
-1.63404867e-01 -6.73434794e-01 7.69768476e-01 -2.66198128e-01
-1.15661979e+00 6.83707476e-01 -4.16534722e-01 3.00057471e-01
1.19547594e+00 -3.63541092e-03 -1.94060821e-02 -4.03109074e-01
-1.18707490e+00 2.82003373e-01 4.24385250e-01 4.83881265e-01
6.51164293e-01 -1.36953115e+00 -4.96657759e-01 7.97528997e-02
6.02044582e-01 -4.15886082e-02 4.53972310e-01 7.55805969e-01
4.28611457e-01 2.12524176e-01 1.13332368e-01 -8.79782200e-01
-5.36008120e-01 9.71550524e-01 1.72506750e-01 -9.12984535e-02
-6.92879975e-01 7.62065172e-01 4.21635300e-01 -4.97131765e-01
3.15349847e-01 -3.83136608e-02 -2.98440188e-01 3.07996303e-01
5.51401436e-01 2.74590343e-01 -3.26103866e-01 -5.16997457e-01
1.07889563e-01 2.34868765e-01 -1.89131439e-01 -3.53214622e-01
1.47400260e+00 -3.10281694e-01 2.59683341e-01 1.18690836e+00
1.39787018e+00 -7.62786865e-01 -1.68117762e+00 -7.86320508e-01
2.20921606e-01 -1.30494922e-01 2.13241816e-01 -3.64009649e-01
-8.13303888e-01 1.17010880e+00 4.68961149e-01 -1.55088827e-01
6.42207444e-01 4.10911351e-01 6.61344588e-01 6.96050584e-01
3.96549672e-01 -1.36506891e+00 4.99219209e-01 4.91974115e-01
2.50217021e-01 -1.56521988e+00 -3.69864345e-01 2.61598825e-01
-9.10455704e-01 7.56081820e-01 5.65240860e-01 -7.27442056e-02
8.67768168e-01 -2.69280016e-01 -6.78976299e-03 -3.19528133e-01
-1.03680992e+00 -3.88434589e-01 4.24959868e-01 7.15321720e-01
1.24755502e-01 8.78377724e-03 3.49848598e-01 8.12491596e-01
2.41664588e-01 7.81300589e-02 1.69447795e-01 9.78087246e-01
-7.52440333e-01 -8.62460077e-01 1.82338342e-01 6.11337483e-01
-1.00084417e-01 3.78769487e-02 1.86007079e-02 5.02664566e-01
1.07345544e-01 9.81331110e-01 2.60524988e-01 -2.06184983e-01
3.98785202e-03 7.76534081e-01 4.59309399e-01 -1.38064337e+00
7.35233799e-02 1.49884224e-01 -4.27893281e-01 -4.24837977e-01
-5.91648161e-01 -6.26038730e-01 -1.18137205e+00 2.59897679e-01
-1.26105264e-01 2.35194594e-01 4.29603547e-01 1.48713148e+00
2.79600501e-01 5.86612642e-01 5.65385401e-01 -8.92700851e-01
-9.64931071e-01 -1.20448124e+00 -8.80464733e-01 6.71833456e-01
1.82758138e-01 -8.88512492e-01 -3.21528792e-01 4.64529485e-01] | [9.771493911743164, 2.9120562076568604] |
13c0b6bc-48c4-4db0-8502-324d7a45cee8 | di-nids-domain-invariant-network-intrusion | 2210.08252 | null | https://arxiv.org/abs/2210.08252v1 | https://arxiv.org/pdf/2210.08252v1.pdf | DI-NIDS: Domain Invariant Network Intrusion Detection System | The performance of machine learning based network intrusion detection systems (NIDSs) severely degrades when deployed on a network with significantly different feature distributions from the ones of the training dataset. In various applications, such as computer vision, domain adaptation techniques have been successful in mitigating the gap between the distributions of the training and test data. In the case of network intrusion detection however, the state-of-the-art domain adaptation approaches have had limited success. According to recent studies, as well as our own results, the performance of an NIDS considerably deteriorates when the `unseen' test dataset does not follow the training dataset distribution. In some cases, swapping the train and test datasets makes this even more severe. In order to enhance the generalisibility of machine learning based network intrusion detection systems, we propose to extract domain invariant features using adversarial domain adaptation from multiple network domains, and then apply an unsupervised technique for recognising abnormalities, i.e., intrusions. More specifically, we train a domain adversarial neural network on labelled source domains, extract the domain invariant features, and train a One-Class SVM (OSVM) model to detect anomalies. At test time, we feedforward the unlabeled test data to the feature extractor network to project it into a domain invariant space, and then apply OSVM on the extracted features to achieve our final goal of detecting intrusions. Our extensive experiments on the NIDS benchmark datasets of NFv2-CIC-2018 and NFv2-UNSW-NB15 show that our proposed setup demonstrates superior cross-domain performance in comparison to the previous approaches. | ['Marius Portmann', 'Mahsa Baktashmotlagh', 'Siamak Layeghy'] | 2022-10-15 | null | null | null | null | ['network-intrusion-detection'] | ['miscellaneous'] | [ 6.21502399e-01 -3.65319341e-01 -2.01852858e-01 -3.77007872e-01
-5.03691435e-02 -9.43216085e-01 6.25948131e-01 5.53666055e-02
-3.92576307e-01 8.06344092e-01 -5.68297327e-01 -5.78362584e-01
-2.62216717e-01 -9.81163025e-01 -4.18378800e-01 -8.08450103e-01
8.11800547e-03 7.31871068e-01 5.98310947e-01 -2.82182246e-01
2.59947836e-01 9.31819141e-01 -1.21885836e+00 2.00002044e-01
6.40741229e-01 1.18417072e+00 -6.34703100e-01 5.47992229e-01
-2.85635591e-01 4.14513499e-01 -1.14949417e+00 -3.98679450e-02
4.87105995e-01 -4.11779314e-01 -7.38719642e-01 -1.45220831e-01
1.68986320e-02 -7.36271515e-02 -5.21572292e-01 1.09823871e+00
2.21370548e-01 1.21405460e-01 9.20422733e-01 -1.68829453e+00
-4.07410324e-01 -6.15451150e-02 -3.96272868e-01 5.78279912e-01
1.52752668e-01 2.48486981e-01 3.77821684e-01 -3.78941834e-01
6.03701353e-01 1.00651550e+00 4.67207700e-01 7.72478104e-01
-1.44309139e+00 -1.05061698e+00 -7.62522072e-02 2.48684600e-01
-1.11428404e+00 -2.77309954e-01 9.46869254e-01 -3.03979903e-01
6.12091839e-01 2.00290814e-01 1.26781315e-01 1.50196624e+00
9.36621577e-02 4.13501203e-01 1.09322309e+00 -2.90673792e-01
5.40518224e-01 4.46096480e-01 1.40473887e-01 2.15335220e-01
2.56479353e-01 3.27044785e-01 7.54890312e-03 -5.32393157e-01
4.32671607e-01 2.67831951e-01 -8.47266763e-02 -4.48605776e-01
-6.26736224e-01 9.66376305e-01 3.58285129e-01 5.19647360e-01
-4.70884442e-01 -6.97462916e-01 7.30433047e-01 7.99839795e-01
3.79170686e-01 3.94739658e-01 -7.37809479e-01 1.78544968e-01
-6.35280788e-01 3.88947539e-02 9.52724814e-01 4.48269874e-01
5.60455859e-01 1.81375995e-01 2.14305460e-01 9.67113495e-01
1.81958422e-01 4.06353295e-01 7.14412153e-01 -1.12220660e-01
2.44904265e-01 9.05435681e-01 -2.99943954e-01 -9.63450015e-01
-3.02733094e-01 -2.59733319e-01 -9.17961717e-01 5.25159419e-01
6.39095485e-01 -2.87498981e-01 -1.17763853e+00 1.56867528e+00
4.66244936e-01 3.79729569e-01 4.46569443e-01 5.03501832e-01
4.39979762e-01 4.08809066e-01 2.41630912e-01 4.72273827e-02
8.19251359e-01 -2.47159407e-01 -2.10213065e-01 -3.02992433e-01
4.92306948e-01 -5.90515494e-01 6.20043635e-01 3.30148637e-01
-4.17661935e-01 -5.18566608e-01 -1.33265102e+00 8.91366422e-01
-9.67523277e-01 -5.78349054e-01 2.95966923e-01 9.03651416e-01
-4.84349459e-01 6.23595715e-01 -6.55677557e-01 -6.09206915e-01
7.28239596e-01 6.53942049e-01 -6.29291058e-01 -2.47386545e-01
-1.39671195e+00 6.55149698e-01 8.51229131e-01 -4.99090165e-01
-8.65765095e-01 -7.23923981e-01 -5.69899797e-01 -1.81764811e-01
2.62125909e-01 -1.82734981e-01 7.30303109e-01 -1.27096021e+00
-1.34946227e+00 8.26467633e-01 3.41528058e-01 -7.10449457e-01
2.09003910e-01 2.71648884e-01 -1.19122338e+00 1.27448425e-01
-2.86913291e-02 4.08253586e-03 1.16811323e+00 -1.05402172e+00
-7.95855343e-01 -5.03937125e-01 -8.44893530e-02 -3.66096884e-01
-6.32173002e-01 1.34520128e-01 8.75110850e-02 -5.78855336e-01
-5.98709099e-02 -8.95562828e-01 1.22167906e-02 -2.16579020e-01
-4.62417662e-01 -8.70950520e-02 1.83145738e+00 -2.81880200e-01
8.53976130e-01 -2.53113842e+00 -9.32307169e-02 7.93596745e-01
2.63468530e-02 1.05717397e+00 -1.92166075e-01 2.48576850e-01
-6.92912579e-01 2.23480221e-02 -4.92880762e-01 3.06337893e-01
-3.12382311e-01 5.47259867e-01 -4.85039681e-01 5.16812027e-01
5.91700554e-01 3.83799255e-01 -6.59145176e-01 -4.05369438e-02
2.71602482e-01 1.96378976e-01 -4.12554920e-01 4.96075451e-01
-1.39036611e-01 7.07332790e-01 -5.75981319e-01 6.09253883e-01
7.74329841e-01 4.48632985e-02 6.17837347e-02 2.48242617e-01
4.71144408e-01 2.11363868e-03 -1.17217076e+00 9.49012756e-01
-1.37537971e-01 4.65841204e-01 -1.09299377e-01 -1.62712026e+00
1.23242164e+00 3.45020980e-01 5.41953623e-01 -7.54353404e-01
5.10744303e-02 2.62443423e-01 5.33865690e-01 -2.24387273e-01
-3.12657177e-01 -3.33331168e-01 -6.97245523e-02 3.21272433e-01
3.33788842e-01 2.37782270e-01 -6.72092079e-04 2.10602079e-02
1.65652490e+00 -4.68205571e-01 5.02015650e-01 1.37353301e-01
1.04837525e+00 2.48635579e-02 5.47199130e-01 6.58985913e-01
-4.75234896e-01 2.54601061e-01 7.34604657e-01 -4.62151080e-01
-9.13389921e-01 -1.37698758e+00 -3.81361842e-01 9.54255342e-01
-1.44618899e-01 1.25627175e-01 -6.30732536e-01 -1.43078816e+00
3.11866790e-01 6.36122108e-01 -5.51806152e-01 -7.42748201e-01
-4.13097382e-01 -8.10974896e-01 1.05509782e+00 4.13384438e-01
7.08411813e-01 -1.28909993e+00 -1.90643430e-01 2.78146207e-01
4.73248750e-01 -1.10384023e+00 1.59344792e-01 4.96449262e-01
-6.63559794e-01 -1.45478022e+00 -3.07263792e-01 -7.51348436e-01
6.82162344e-01 -2.82883286e-01 7.14731872e-01 2.73521654e-02
-3.75462919e-01 1.96423933e-01 -4.27416205e-01 -6.01218045e-01
-7.47718394e-01 1.49441227e-01 4.42414284e-01 3.46097022e-01
7.74444044e-01 -7.63997018e-01 -2.82579362e-01 5.26017904e-01
-1.31965387e+00 -1.01678169e+00 5.77082217e-01 9.82306838e-01
2.64554799e-01 6.73067927e-01 1.04644632e+00 -1.33592737e+00
6.40066743e-01 -9.41048324e-01 -4.24685061e-01 -5.96216880e-02
-5.08487344e-01 -1.84836134e-01 1.28359389e+00 -8.18722308e-01
-8.91587198e-01 -2.57521987e-01 -1.77465111e-01 -5.06767690e-01
-8.78229976e-01 1.86142519e-01 -4.18406188e-01 -3.18550885e-01
1.07144964e+00 3.33792120e-01 1.78060606e-01 -2.37378135e-01
-2.78004259e-01 8.81782889e-01 5.51862657e-01 -3.70806485e-01
1.41313028e+00 2.96034783e-01 8.17523673e-02 -9.44974840e-01
-4.66912597e-01 -5.14422774e-01 -7.01811314e-01 2.33533561e-01
4.90608007e-01 -2.67539710e-01 -3.09067726e-01 6.64929092e-01
-7.96794415e-01 -5.20524494e-02 -3.10671598e-01 2.07874209e-01
-2.76848465e-01 1.37192190e-01 -1.32873490e-01 -5.11932790e-01
-2.00019583e-01 -8.95601690e-01 2.49914169e-01 3.97740632e-01
-2.81663150e-01 -1.25896597e+00 2.86381990e-01 -1.59844324e-01
7.08384216e-01 6.03354573e-01 1.27342725e+00 -1.85506439e+00
2.20243528e-01 -7.56062031e-01 -2.14437589e-01 8.39533687e-01
4.40568566e-01 -1.71862185e-01 -9.34633434e-01 -3.62427264e-01
1.85202107e-01 -1.48072809e-01 4.37376976e-01 -8.89247656e-02
1.06431019e+00 -5.90965524e-02 -4.46438283e-01 5.99269390e-01
1.31160796e+00 8.03330898e-01 5.87202132e-01 7.33285248e-01
2.88257629e-01 3.48029554e-01 5.79032362e-01 3.31883937e-01
-5.41716754e-01 4.10034090e-01 5.37440836e-01 -1.61235571e-01
2.34857500e-01 7.87180103e-03 2.43363842e-01 -1.15366884e-01
3.30190510e-01 -1.26787186e-01 -9.96966660e-01 3.49156886e-01
-1.46176469e+00 -8.51038277e-01 3.37213010e-01 2.15713286e+00
6.36693954e-01 5.96298575e-01 3.38668823e-01 5.34030557e-01
7.23378599e-01 5.35049774e-02 -9.57415879e-01 -7.06679285e-01
-2.13522166e-02 8.14584494e-01 4.80377018e-01 -1.12075597e-01
-1.42122674e+00 8.46550286e-01 5.12935781e+00 6.84995770e-01
-1.38934219e+00 -5.66947125e-02 3.71362150e-01 2.07058579e-01
4.02355969e-01 -2.02783719e-01 -3.84727359e-01 6.39256954e-01
1.30265129e+00 -2.02272803e-01 3.86556685e-01 9.49422240e-01
-2.30693072e-01 3.40894222e-01 -1.08754981e+00 6.91348553e-01
-1.38428817e-02 -9.01790500e-01 -8.39063227e-02 1.86213344e-01
4.33166116e-01 1.78602301e-02 2.31931627e-01 7.48498917e-01
1.54376119e-01 -1.03942168e+00 -1.44514367e-01 1.32583261e-01
7.00220466e-01 -1.10162616e+00 1.04589832e+00 3.73684853e-01
-7.79114485e-01 -2.58554012e-01 -3.08160424e-01 1.94063798e-01
-4.56030965e-01 3.27568918e-01 -1.14018023e+00 4.84404385e-01
5.99877059e-01 4.62232947e-01 -5.69746256e-01 8.97797763e-01
2.63989642e-02 7.44311750e-01 -3.46056253e-01 2.86923170e-01
1.84233308e-01 3.33707370e-02 7.05012977e-01 9.40897107e-01
-1.09775282e-01 2.02426538e-02 2.69654691e-01 5.10392427e-01
-1.22097611e-01 4.67212545e-03 -1.13017750e+00 -3.02235126e-01
4.15753365e-01 1.06744182e+00 -5.28489828e-01 -2.81413496e-01
-3.92672628e-01 9.61866617e-01 1.17421158e-01 4.08557892e-01
-7.84248590e-01 -7.88237870e-01 1.09976673e+00 1.91896502e-02
2.90380597e-01 1.18484482e-01 -1.68528691e-01 -9.31185842e-01
-2.03145370e-01 -1.19087887e+00 6.92805946e-01 5.07033579e-02
-1.59817398e+00 7.95993626e-01 -1.04553610e-01 -1.37166321e+00
-4.94088888e-01 -8.93232822e-01 -8.61350179e-01 9.37262535e-01
-1.23692238e+00 -8.32780004e-01 8.90454575e-02 1.14853549e+00
2.62002409e-01 -8.53574216e-01 1.19424093e+00 3.34352314e-01
-6.42689407e-01 8.44809949e-01 2.99028754e-01 6.46899104e-01
8.76105011e-01 -1.01061940e+00 3.80939782e-01 8.21466923e-01
-9.01125446e-02 4.22554612e-01 5.92126846e-01 -5.46168327e-01
-1.02876985e+00 -1.40931618e+00 4.89876308e-02 -2.60873973e-01
8.76216590e-01 -1.70945406e-01 -1.31993020e+00 6.78716183e-01
-1.41047180e-01 5.00801682e-01 9.23011780e-01 -1.37541175e-01
-6.77024841e-01 -1.15158200e-01 -1.97699594e+00 4.01458263e-01
5.79216421e-01 -4.95309114e-01 -7.38305271e-01 1.95817620e-01
2.85376430e-01 -8.27132091e-02 -8.81313086e-01 6.47206962e-01
2.77690589e-01 -8.11000586e-01 1.08723497e+00 -1.22481656e+00
2.88022850e-02 -2.18277022e-01 -2.40185872e-01 -1.45575011e+00
-1.17410749e-01 -1.93103999e-01 -1.91636458e-01 1.37195933e+00
2.07596153e-01 -1.27053463e+00 9.82383847e-01 2.11906850e-01
3.50425661e-01 -4.24195379e-01 -9.03362930e-01 -8.98684084e-01
1.00415885e-01 -3.04849684e-01 6.56741858e-01 1.08571231e+00
-3.58621299e-01 1.64430588e-01 1.08355008e-01 5.54303646e-01
6.10006571e-01 -2.72146523e-01 8.45070243e-01 -1.62918437e+00
-1.57893419e-01 -2.90920824e-01 -1.00684941e+00 -1.80876553e-01
4.78779405e-01 -8.30379903e-01 -3.58024180e-01 -5.36359787e-01
-3.25589806e-01 -5.74717283e-01 -7.29326427e-01 6.50191069e-01
1.89101353e-01 4.68124539e-01 -1.64206582e-03 3.66542973e-02
-2.73325443e-01 1.66200668e-01 6.73509896e-01 -2.63469070e-01
-3.64324242e-01 3.76393974e-01 -5.33627391e-01 8.43492508e-01
1.05031884e+00 -7.04918802e-01 -3.79810989e-01 2.94311345e-01
-7.48619914e-01 -3.40002775e-01 2.83711225e-01 -1.38490295e+00
1.06525220e-01 -2.36931711e-01 8.35971594e-01 -1.69342831e-01
8.79345834e-02 -1.14769292e+00 -1.84113577e-01 5.30339658e-01
-2.01374292e-02 -1.76895067e-01 4.93644595e-01 6.31097078e-01
-3.14054310e-01 -2.36318052e-01 1.26221216e+00 1.96860045e-01
-7.32403994e-01 4.45873260e-01 -3.23452055e-01 1.85369715e-01
1.38641059e+00 -2.36231357e-01 -3.02104175e-01 1.19178489e-01
-7.00447381e-01 -4.55108099e-02 4.49481964e-01 4.55847204e-01
6.78395152e-01 -1.15801179e+00 -5.35291016e-01 9.78459001e-01
2.47613996e-01 -2.20141232e-01 3.97367179e-02 3.89908463e-01
-2.68315613e-01 3.47230494e-01 -7.27117538e-01 -6.11268103e-01
-1.22326863e+00 6.77472293e-01 2.73432910e-01 -3.95423204e-01
-3.92621279e-01 4.29109812e-01 3.44729945e-02 -7.05869794e-01
5.51175028e-02 4.23462480e-01 -3.19199145e-01 -1.94458425e-01
6.04485035e-01 1.29852578e-01 3.47360671e-01 -4.84441102e-01
-6.58427775e-01 1.44452706e-01 -5.74429274e-01 2.02493683e-01
1.21455061e+00 5.49613953e-01 9.95732471e-02 -2.35320092e-03
1.42314112e+00 -2.76941746e-01 -7.18476117e-01 -4.26938891e-01
3.68851870e-01 -4.47117209e-01 -4.33960229e-01 -8.72099400e-01
-1.03088689e+00 5.96385479e-01 9.46578741e-01 6.79437757e-01
1.52795768e+00 -2.05050096e-01 6.38813078e-01 3.30688268e-01
1.30264238e-01 -7.58410752e-01 5.76064140e-02 7.26678193e-01
3.40192080e-01 -1.27203476e+00 -4.68792170e-01 -4.70381677e-02
-4.38965797e-01 1.27127409e+00 7.39184439e-01 -5.75489104e-01
7.90725172e-01 2.34398127e-01 4.62005958e-02 1.85173601e-02
-4.79546845e-01 7.27213696e-02 1.61667824e-01 1.19451809e+00
-1.18952401e-01 -7.40681142e-02 1.54671595e-01 3.98302704e-01
-1.37812523e-02 -2.66725242e-01 2.48190150e-01 9.44141924e-01
-3.25818986e-01 -1.46572053e+00 -6.16155803e-01 6.87232137e-01
-6.18281126e-01 2.76756376e-01 -6.81586564e-01 9.88957167e-01
8.57497528e-02 8.91970992e-01 1.37450531e-01 -6.15193248e-01
6.72509789e-01 5.27253866e-01 -7.86492154e-02 -7.44061112e-01
-5.23899019e-01 -5.69875002e-01 -1.81308106e-01 -4.37439799e-01
-6.53185248e-02 -4.60171640e-01 -1.27972829e+00 -3.85616213e-01
7.91928619e-02 1.39064923e-01 5.72085261e-01 1.21245670e+00
2.32230321e-01 7.04123199e-01 1.02807283e+00 -4.22351271e-01
-7.90097475e-01 -8.58341813e-01 -5.94732463e-01 8.49394739e-01
3.93859625e-01 -7.45055020e-01 -5.56179404e-01 -3.21108341e-01] | [5.37531852722168, 7.337593078613281] |
67b56eb9-28c3-4d68-82b5-2bc118b68785 | r2former-unified-retrieval-and-reranking | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Zhu_R2Former_Unified_Retrieval_and_Reranking_Transformer_for_Place_Recognition_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Zhu_R2Former_Unified_Retrieval_and_Reranking_Transformer_for_Place_Recognition_CVPR_2023_paper.pdf | R2Former: Unified Retrieval and Reranking Transformer for Place Recognition | Visual Place Recognition (VPR) estimates the location of query images by matching them with images in a reference database. Conventional methods generally adopt aggregated CNN features for global retrieval and RANSAC-based geometric verification for reranking. However, RANSAC only employs geometric information but ignores other possible information that could be useful for reranking, e.g. local feature correlations, and attention values. In this paper, we propose a unified place recognition framework that handles both retrieval and reranking with a novel transformer model, named R2Former. The proposed reranking module takes feature correlation, attention value, and xy coordinates into account, and learns to determine whether the image pair is from the same location. The whole pipeline is end-to-end trainable and the reranking module alone can also be adopted on other CNN or transformer backbones as a generic component. Remarkably, R2Former significantly outperforms state-of-the-art methods on major VPR datasets with much less inference time and memory consumption. It also achieves the state-of-the-art on the hold-out MSLS challenge set and could serve as a simple yet strong solution for real-world large-scale applications. Experiments also show vision transformer tokens are comparable and sometimes better than CNN local features on local matching. The code is released at https://github.com/Jeff-Zilence/R2Former. | ['Heng Wang', 'Xiaohui Shen', 'Mubarak Shah', 'Chen Chen', 'Linjie Yang', 'Sijie Zhu'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['visual-place-recognition'] | ['computer-vision'] | [-4.47105557e-01 -4.54463661e-01 -3.50259513e-01 -3.87491226e-01
-1.19397211e+00 -6.09190106e-01 5.76826513e-01 1.84620366e-01
-5.61133325e-01 2.86605954e-01 1.29098371e-01 -6.43794090e-02
-1.35769919e-01 -7.75650144e-01 -9.39403892e-01 -5.37512004e-01
3.75910550e-02 5.51722050e-01 3.72641265e-01 -1.47096798e-01
5.09943426e-01 9.17791724e-01 -1.64075446e+00 1.55815780e-01
4.64031994e-01 1.34582496e+00 4.16056097e-01 3.12223434e-01
3.03695619e-01 7.34544277e-01 -2.69237101e-01 -2.18351081e-01
4.76791918e-01 3.55124891e-01 -5.94813287e-01 -4.10207778e-01
1.10146928e+00 -7.19753623e-01 -8.17788601e-01 9.24878657e-01
6.74338281e-01 3.29257369e-01 2.50082254e-01 -1.24885249e+00
-1.01933169e+00 2.44446516e-01 -6.40008450e-01 2.05420226e-01
3.96860212e-01 1.77447200e-01 1.37263119e+00 -1.30841208e+00
6.65802717e-01 1.03173447e+00 5.87123156e-01 1.19398311e-01
-8.57395172e-01 -8.25867832e-01 2.04850212e-01 6.66737199e-01
-1.95272171e+00 -3.34399313e-01 7.42673099e-01 -1.75767064e-01
1.02692664e+00 2.95414567e-01 5.08870840e-01 6.99361563e-01
-2.04127923e-01 1.12705231e+00 6.50240660e-01 1.53439522e-01
-2.55072191e-02 -4.07190621e-01 -2.69854050e-02 8.05362582e-01
-5.35526201e-02 6.26792088e-02 -7.97263861e-01 -2.00026676e-01
8.96178663e-01 4.89267915e-01 -3.74469429e-01 -5.55426002e-01
-1.70950091e+00 6.16080582e-01 1.34842551e+00 3.14238608e-01
-3.46716106e-01 6.06789410e-01 1.17981985e-01 7.66648427e-02
2.20926404e-01 4.07393187e-01 -3.66711557e-01 1.40798926e-01
-1.07110929e+00 4.30954546e-01 2.57685065e-01 1.09822857e+00
1.02048206e+00 -3.85198146e-01 -2.79592156e-01 8.74340534e-01
5.42328537e-01 6.17300212e-01 5.07131457e-01 -8.33352447e-01
5.32739162e-01 5.75665891e-01 1.81125030e-01 -1.35992169e+00
-2.02203304e-01 -4.32944715e-01 -5.80225706e-01 -8.97883847e-02
1.75118551e-01 5.84331214e-01 -1.07677817e+00 1.38733006e+00
2.90706545e-01 5.69444120e-01 -3.80236298e-01 1.41139627e+00
1.20438612e+00 5.47576487e-01 -3.07510853e-01 6.80444837e-01
1.37299645e+00 -1.25919127e+00 -1.08691245e-01 -2.33382419e-01
4.29896027e-01 -9.16181922e-01 7.71189094e-01 8.31486210e-02
-8.79979849e-01 -4.55969751e-01 -1.08148026e+00 -6.98433995e-01
-6.46690845e-01 5.56401432e-01 5.11898041e-01 -2.82139126e-02
-1.47509658e+00 7.23721325e-01 -7.41997778e-01 -4.20672566e-01
3.31184447e-01 3.87121201e-01 -7.07467437e-01 -2.86202669e-01
-9.97755885e-01 8.50670874e-01 1.92764811e-02 5.01579881e-01
-8.25288951e-01 -7.75232136e-01 -1.01620317e+00 7.18858615e-02
1.56599134e-02 -4.90612119e-01 1.18430865e+00 -3.03589374e-01
-1.12554169e+00 1.05726492e+00 -3.19452703e-01 -3.39196593e-01
6.42961502e-01 -2.96688646e-01 -3.96649614e-02 1.26165226e-01
3.54073793e-01 8.13941061e-01 6.68003261e-01 -1.03787899e+00
-6.64605498e-01 -5.13131201e-01 5.29524907e-02 2.20753580e-01
2.82657072e-02 -1.02218732e-01 -9.22415793e-01 -5.40117800e-01
6.43778622e-01 -9.38101709e-01 -1.69068456e-01 5.25911212e-01
-3.56854826e-01 -3.21255326e-01 6.60421133e-01 -5.08166730e-01
7.30813265e-01 -2.19751477e+00 -1.67155430e-01 4.63939279e-01
4.21105683e-01 3.87449190e-02 -3.93751383e-01 2.96454221e-01
-9.93479937e-02 -1.14069074e-01 2.12284401e-01 -4.13489938e-01
1.90371916e-01 -1.39856458e-01 -5.88492692e-01 9.87756252e-01
7.35919923e-02 1.23554730e+00 -1.10603201e+00 -3.16104770e-01
6.23527348e-01 7.24361420e-01 -3.88973445e-01 4.88934033e-02
2.71048039e-01 2.04662442e-01 -4.48650241e-01 9.52818751e-01
7.78980732e-01 -3.89659643e-01 -2.96479255e-01 -4.86273199e-01
-2.99029857e-01 4.37008202e-01 -1.07729137e+00 1.82022452e+00
-3.31415772e-01 8.97521734e-01 -2.98802406e-01 -6.50177240e-01
1.05144191e+00 -1.74604997e-01 4.78248715e-01 -9.82471108e-01
1.90034974e-02 4.30985719e-01 -4.93405610e-01 -7.33204409e-02
9.04899657e-01 6.59878671e-01 6.65258709e-03 -1.55879930e-02
1.66249946e-01 2.35967249e-01 -2.05545038e-01 -3.19293537e-03
1.00453722e+00 1.18426591e-01 2.73869693e-01 -8.83681104e-02
4.24812526e-01 -5.55910245e-02 4.82823282e-01 7.63118207e-01
-2.77044266e-01 1.16406345e+00 9.41551290e-03 -7.13392138e-01
-8.05164337e-01 -1.13732576e+00 -1.45814806e-01 1.09013140e+00
7.73046672e-01 -5.46263695e-01 -8.90053883e-02 -4.12646651e-01
3.83716911e-01 2.01812387e-01 -6.78863525e-01 3.52557823e-02
-5.68367183e-01 -2.23685041e-01 6.25619292e-01 7.17522860e-01
6.47861898e-01 -8.13181937e-01 -4.71198022e-01 -7.54924193e-02
-4.14249718e-01 -1.01211059e+00 -7.17410147e-01 -1.21014401e-01
-5.42926550e-01 -1.08701909e+00 -1.00985134e+00 -8.17931890e-01
7.25043774e-01 8.44409943e-01 8.34742725e-01 4.19789612e-01
-1.78870216e-01 3.49718004e-01 -3.68416935e-01 2.04106465e-01
5.68250358e-01 2.84381092e-01 -1.00625254e-01 7.28312433e-02
2.78024346e-01 -3.72603834e-01 -1.08925176e+00 5.99915326e-01
-4.37685251e-01 -3.31944674e-01 6.33009374e-01 8.25518429e-01
9.35324848e-01 -5.50525248e-01 -1.15439162e-01 -2.99370378e-01
1.24895558e-01 -3.62058192e-01 -7.39906490e-01 4.09816414e-01
-3.33012313e-01 9.58301499e-02 3.45443845e-01 -3.62298101e-01
-3.80715072e-01 1.62620917e-01 -1.77625164e-01 -1.07562768e+00
-3.75998169e-02 3.58096421e-01 -1.48498407e-02 -5.69599211e-01
2.74411500e-01 3.51074547e-01 -4.17755306e-01 -5.28844476e-01
3.93108249e-01 4.56981063e-01 6.95032716e-01 -4.46931154e-01
1.08768082e+00 6.41933680e-01 -1.67289183e-01 -5.59121311e-01
-5.30389607e-01 -1.00939691e+00 -5.67808747e-01 -1.03440516e-01
6.67182267e-01 -1.29089224e+00 -9.77711201e-01 5.18841147e-01
-1.21533310e+00 -2.44560450e-01 1.62758559e-01 4.13408637e-01
-3.92978549e-01 8.80901739e-02 -5.34539878e-01 -3.19079340e-01
-3.81187856e-01 -1.27695954e+00 1.66667795e+00 2.03997284e-01
1.50526300e-01 -5.93020201e-01 6.07630648e-02 3.71366739e-01
4.88156021e-01 -7.14088976e-03 1.69831142e-01 -7.16136217e-01
-1.27331150e+00 -4.91960526e-01 -5.36165118e-01 -3.32590103e-01
-2.22220272e-01 -2.15915944e-02 -1.11844945e+00 -4.35352415e-01
-5.64920366e-01 -1.84713930e-01 1.19411945e+00 2.03465652e-02
1.33647466e+00 -2.42766097e-01 -3.56204748e-01 8.92299891e-01
1.45508194e+00 -1.84751287e-01 9.06793654e-01 5.70440233e-01
1.07090223e+00 1.98761210e-01 8.30087543e-01 3.45422983e-01
8.49471927e-01 9.33150172e-01 6.33112609e-01 -2.28178456e-01
9.63387191e-02 -5.42162120e-01 2.36992717e-01 5.69445252e-01
2.23153420e-02 3.77823748e-02 -1.08287001e+00 7.61337757e-01
-2.12151170e+00 -1.05827582e+00 8.57858434e-02 2.25755501e+00
4.04227853e-01 -3.18511248e-01 -3.44221503e-01 -2.63052553e-01
7.91585863e-01 5.66609263e-01 -5.12011826e-01 3.35491687e-01
-2.57713590e-02 -6.30952138e-03 1.00036049e+00 5.52679956e-01
-1.32406783e+00 1.07761204e+00 4.97272539e+00 8.56793702e-01
-1.48180032e+00 1.01816073e-01 5.78957796e-01 -4.57140096e-02
-7.45320469e-02 -2.07002088e-02 -7.76479781e-01 1.68782040e-01
2.11205751e-01 1.20938949e-01 6.40767992e-01 1.13283312e+00
-2.33597070e-01 4.88159098e-02 -1.16258442e+00 1.59577823e+00
3.61501306e-01 -1.50991321e+00 9.28744897e-02 2.88659125e-04
5.04988313e-01 8.20839286e-01 2.97344059e-01 2.48175681e-01
2.21651290e-02 -8.82003307e-01 1.09515190e+00 4.50951159e-01
7.83415675e-01 -5.57122588e-01 8.15503895e-01 -4.37168516e-02
-1.65897572e+00 -4.02281582e-02 -5.98306477e-01 2.59081870e-01
-1.72335058e-01 2.38195434e-01 -7.27582932e-01 5.25231719e-01
1.05946994e+00 1.01957417e+00 -1.06780601e+00 1.52417624e+00
-1.53375939e-01 -1.08146943e-01 -6.02158368e-01 1.13845877e-01
4.05912310e-01 1.80557668e-01 3.99960756e-01 8.94090533e-01
5.46353042e-01 -5.38429879e-02 8.09807330e-02 7.64232337e-01
-2.09898517e-01 2.61200741e-02 -6.58602476e-01 3.30700994e-01
7.84219682e-01 1.50100577e+00 -7.31541514e-01 -2.03972965e-01
-1.52937636e-01 9.93523121e-01 6.06042087e-01 2.62342393e-01
-8.73744488e-01 -4.50993448e-01 9.22224998e-01 1.23134062e-01
5.68625450e-01 -2.50746518e-01 8.54354128e-02 -1.52342105e+00
1.38991594e-01 -5.91288149e-01 2.30197400e-01 -1.10717416e+00
-1.26043570e+00 6.24246716e-01 -2.85607219e-01 -1.61633968e+00
-1.51364555e-04 -7.59802997e-01 -6.03667557e-01 6.58356249e-01
-1.81745315e+00 -1.29761350e+00 -7.63401926e-01 7.56669879e-01
2.36857861e-01 5.28203929e-03 5.10249078e-01 6.06925011e-01
-3.38007987e-01 9.33043361e-01 1.53194770e-01 6.63784683e-01
9.66810346e-01 -1.10819626e+00 5.39855540e-01 7.75428176e-01
4.14293170e-01 9.97718573e-01 2.15837762e-01 -3.64462912e-01
-1.65789688e+00 -1.27411807e+00 1.02825761e+00 -5.08943021e-01
7.72058845e-01 -3.87795269e-01 -6.14774585e-01 6.33458912e-01
-1.16535358e-01 5.17325580e-01 2.84077734e-01 3.09013855e-02
-8.91808450e-01 -4.36335415e-01 -9.46726918e-01 7.12466240e-01
1.01698399e+00 -1.09028327e+00 -3.07292074e-01 2.55161852e-01
6.24794900e-01 -9.35580015e-01 -8.05306613e-01 3.81969780e-01
6.19596303e-01 -7.03752339e-01 1.32104599e+00 -4.44541611e-02
2.44222805e-01 -8.12822759e-01 -4.89694625e-01 -8.29349875e-01
-4.51740772e-01 -3.15814912e-01 1.47082120e-01 1.05849040e+00
2.41800904e-01 -7.58182943e-01 5.67650676e-01 5.80163479e-01
-3.07070892e-02 -8.94878209e-01 -1.16727531e+00 -5.82277715e-01
-3.63469690e-01 -4.12064612e-01 9.07447219e-01 1.08052921e+00
-2.08573341e-01 -5.81234694e-02 -2.66074806e-01 6.11054301e-01
6.46447837e-01 5.01936138e-01 7.69345522e-01 -1.00522065e+00
1.28461933e-02 -5.81666470e-01 -9.50436115e-01 -1.36123097e+00
1.25118747e-01 -1.14656508e+00 2.36902595e-01 -1.62238002e+00
-2.65712608e-02 -5.09861410e-01 -6.32416487e-01 8.17930877e-01
7.24214613e-02 6.40223563e-01 5.54527104e-01 5.69908798e-01
-1.04770589e+00 7.15179682e-01 9.13994730e-01 -3.98248345e-01
-1.07166814e-02 -2.98337549e-01 -3.93922210e-01 4.98048097e-01
6.80420995e-01 -3.00671518e-01 -1.26311304e-02 -7.40337312e-01
2.10228518e-01 -2.20746532e-01 9.63672101e-01 -1.12267208e+00
7.36489892e-01 1.11836798e-01 5.99579811e-01 -1.00683677e+00
3.42544585e-01 -9.57038641e-01 1.15369417e-01 1.08543649e-01
-3.94300699e-01 5.42511225e-01 -7.87180662e-02 4.44327056e-01
-3.22909027e-01 1.80158705e-01 4.36556160e-01 5.05994409e-02
-1.09202039e+00 7.75505722e-01 2.21754745e-01 -2.48304382e-01
8.10460925e-01 -1.33017123e-01 -6.66323960e-01 -2.61631131e-01
-2.75597632e-01 4.69206393e-01 7.08412528e-01 7.71635950e-01
9.98902500e-01 -1.63528132e+00 -4.89653319e-01 1.40360743e-01
5.17972946e-01 1.12403788e-01 1.29069895e-01 8.62444222e-01
-8.63254189e-01 5.08811176e-01 -6.84839040e-02 -7.69102573e-01
-1.06083083e+00 4.57886279e-01 4.86823916e-01 -3.91776711e-02
-5.95339179e-01 7.38484979e-01 1.55454114e-01 -6.41696095e-01
4.18803364e-01 -3.95325899e-01 -4.94908020e-02 1.83023214e-01
6.76825345e-01 1.97610080e-01 2.14189485e-01 -1.02208400e+00
-8.50991488e-01 9.13210750e-01 -2.04862788e-01 1.01220079e-01
1.31514239e+00 -5.44262677e-02 -2.66889006e-01 2.38458246e-01
1.58124471e+00 -1.78836897e-01 -1.19462812e+00 -4.36696976e-01
-6.57009184e-02 -7.78255939e-01 1.08673178e-01 -3.77117455e-01
-1.50578439e+00 8.17021966e-01 6.10969901e-01 -4.68326181e-01
8.06970477e-01 1.98816866e-01 7.86573529e-01 8.11104059e-01
8.36031020e-01 -7.84759045e-01 -6.74991263e-03 6.65177763e-01
1.15625310e+00 -1.28652787e+00 -1.13270981e-02 -1.16484359e-01
-4.76705551e-01 9.92214859e-01 7.02892661e-01 -5.66909015e-01
7.34471023e-01 -1.35931820e-01 9.27858055e-02 -1.29723430e-01
-4.74824250e-01 -3.68134767e-01 5.81382394e-01 3.98980826e-01
1.59232482e-01 -3.87396067e-02 1.80797651e-01 2.09294572e-01
-3.09539527e-01 -1.78669557e-01 -6.71403408e-02 6.56919599e-01
-2.25260064e-01 -5.98689795e-01 -3.55389863e-01 2.83110946e-01
-1.07333452e-01 -3.05578500e-01 -2.76331037e-01 6.06332421e-01
8.89626667e-02 5.01334786e-01 2.35073909e-01 -7.10505724e-01
2.92986006e-01 -4.36759800e-01 2.87047714e-01 -1.59396127e-01
-7.16476738e-01 -1.34744793e-01 -3.48007947e-01 -1.24239731e+00
-2.22819671e-01 -5.12044013e-01 -1.15768218e+00 -5.79675496e-01
-3.36627632e-01 -1.90813452e-01 4.68381882e-01 7.05065727e-01
6.58313751e-01 1.01923861e-01 5.17937183e-01 -1.41817176e+00
-3.51955086e-01 -5.58665156e-01 -2.02820331e-01 3.04333299e-01
6.81618989e-01 -7.47923791e-01 -1.27537400e-01 -3.86148900e-01] | [7.805339336395264, -1.8683099746704102] |
215c4e5f-4e8b-419b-8050-980b8e8e7f16 | scalable-end-to-end-ml-platforms-from-automl | 2302.14139 | null | https://arxiv.org/abs/2302.14139v3 | https://arxiv.org/pdf/2302.14139v3.pdf | Scalable End-to-End ML Platforms: from AutoML to Self-serve | ML platforms help enable intelligent data-driven applications and maintain them with limited engineering effort. Upon sufficiently broad adoption, such platforms reach economies of scale that bring greater component reuse while improving efficiency of system development and maintenance. For an end-to-end ML platform with broad adoption, scaling relies on pervasive ML automation and system integration to reach the quality we term self-serve that we define with ten requirements and six optional capabilities. With this in mind, we identify long-term goals for platform development, discuss related tradeoffs and future work. Our reasoning is illustrated on two commercially-deployed end-to-end ML platforms that host hundreds of real-time use cases -- one general-purpose and one specialized. | ['Mia R. Garrard', 'Norm Zhou', 'Ryan Maghsoudian', 'Tanya Qie', 'George Han', 'Cesar Cardoso', 'Anika Li', 'Tanvi Gupta', 'Yin Huang', 'Pavlos A. Apostolopoulos', 'Igor L. Markov'] | 2023-02-27 | null | null | null | null | ['automl'] | ['methodology'] | [-5.33071101e-01 2.67928779e-01 -6.93905056e-01 -3.92880291e-01
-8.78049016e-01 -8.51673961e-01 3.20961416e-01 -1.34306848e-01
2.56031573e-01 2.54570514e-01 1.02413744e-01 -7.07970858e-01
-1.19483672e-01 -3.29081029e-01 -2.72542596e-01 2.56158233e-01
-2.19139025e-01 1.58680350e-01 1.52251422e-01 -9.54547152e-02
-3.16709206e-02 5.13470531e-01 -1.74502432e+00 3.87699753e-01
6.85376167e-01 1.28030109e+00 5.22762313e-02 8.54827464e-01
-1.49174869e-01 1.20717072e+00 -5.13229609e-01 -1.56018808e-01
3.90275478e-01 3.34776312e-01 -3.82970899e-01 2.58200020e-01
-2.47193389e-02 -3.26959521e-01 8.32920615e-03 1.71183556e-01
6.35568738e-01 -5.02587736e-01 7.55738243e-02 -1.94406021e+00
-4.01924938e-01 5.57463229e-01 -5.24479866e-01 -3.78330797e-01
5.57687342e-01 7.88255632e-01 1.07439840e+00 -7.84754813e-01
4.61787343e-01 6.70812547e-01 1.12035036e+00 4.11764793e-02
-1.34721875e+00 -3.40502828e-01 2.17508137e-01 -5.62377691e-01
-1.61864042e+00 -1.42632902e+00 5.07858813e-01 -6.80564642e-01
1.51050556e+00 6.01880729e-01 5.47365963e-01 4.41406459e-01
3.79854828e-01 7.77129531e-01 3.65553170e-01 -2.09865078e-01
5.31260073e-01 2.82895356e-01 3.08564454e-01 3.15136522e-01
7.02674031e-01 -7.31003225e-01 -4.30313796e-01 -5.76249421e-01
2.99683899e-01 -3.13956678e-01 9.59451422e-02 -4.74789739e-01
-1.08786058e+00 1.88366145e-01 -3.31473082e-01 5.12580991e-01
-6.92008257e-01 5.27168334e-01 5.36225557e-01 3.06721330e-01
2.42685616e-01 4.35003400e-01 -1.10265779e+00 -6.68657005e-01
-1.16135800e+00 1.69863746e-01 1.31374633e+00 1.76704478e+00
8.13193977e-01 1.46915868e-01 2.37323552e-01 5.13511598e-01
7.71489441e-01 3.81208479e-01 2.44467422e-01 -1.16077065e+00
3.19650650e-01 9.56725419e-01 2.61728376e-01 -4.12699819e-01
-5.89689255e-01 -6.02914393e-01 -1.78630754e-01 3.36179584e-01
-8.39056149e-02 -3.57247174e-01 -2.68731695e-02 1.47238493e+00
1.35649502e-01 -3.60380858e-01 2.28618085e-01 1.59075096e-01
4.44735587e-01 2.61843741e-01 1.52929246e-01 -4.99042511e-01
1.32928038e+00 -5.69033742e-01 -8.21743309e-01 -7.23746657e-01
7.76635051e-01 -9.74056661e-01 1.50652277e+00 5.83368361e-01
-1.35805643e+00 -4.51753408e-01 -1.49457598e+00 6.45392537e-02
-4.62032408e-02 2.21730098e-01 7.76910484e-01 8.41055036e-01
-1.24747872e+00 8.91263634e-02 -1.04863834e+00 -5.95597506e-01
1.55890375e-01 3.52689475e-01 3.90205905e-02 4.33677614e-01
-2.77230889e-01 6.47056460e-01 -8.91715810e-02 -4.67628062e-01
-3.92855167e-01 -1.17200959e+00 -3.29342484e-01 6.31265938e-02
2.83298761e-01 -1.06366217e+00 1.91304374e+00 -6.62613571e-01
-1.63450062e+00 6.40623629e-01 2.88456529e-01 -1.38126045e-01
2.46324748e-01 -3.15825552e-01 -9.84863937e-01 -2.57382661e-01
2.93337226e-01 -8.48838910e-02 5.10895848e-01 -1.13712311e+00
-9.53158081e-01 -1.81596190e-01 -5.21733314e-02 7.73115903e-02
-5.04088819e-01 1.61831349e-01 -5.16771615e-01 -8.55214819e-02
-1.13122292e-01 -5.86163461e-01 -5.42382970e-02 2.40305066e-02
-2.10348144e-03 1.66205596e-02 1.15265870e+00 -4.80297208e-01
1.60097504e+00 -2.32056665e+00 -4.86984521e-01 1.79628865e-03
7.25528061e-01 -2.03308895e-01 2.14455575e-02 8.88496816e-01
2.05446690e-01 2.89450526e-01 3.71127963e-01 -1.36577666e-01
5.70642114e-01 -2.77658105e-01 -5.89017719e-02 4.90948617e-01
4.14330140e-02 9.91428912e-01 -4.53839093e-01 -3.58876109e-01
4.40412253e-01 1.89416990e-01 -5.64217925e-01 3.59008759e-01
-6.67340100e-01 -3.43192935e-01 -5.33863366e-01 9.38769460e-01
4.85063672e-01 -6.03303730e-01 5.23480237e-01 -4.75445122e-01
-4.66970026e-01 2.70173699e-01 -1.35334170e+00 1.86902678e+00
-8.89114082e-01 6.03134692e-01 7.86411762e-01 9.87974405e-02
6.33651257e-01 2.72999883e-01 9.18643296e-01 -4.55413699e-01
4.55301926e-02 2.56202638e-01 6.14946410e-02 -7.46342540e-01
5.03525615e-01 3.24677825e-01 -4.57455099e-01 6.93801165e-01
-4.57068086e-01 -3.85062635e-01 -3.89516242e-02 2.86464810e-01
1.49876153e+00 2.24870101e-01 7.97240674e-01 -5.03971636e-01
4.30101864e-02 -4.18164097e-02 3.90591979e-01 3.62063944e-01
-2.11140826e-01 1.05749637e-01 4.16392148e-01 6.47449493e-02
-1.09806991e+00 -1.00341284e+00 -1.45401508e-01 1.62329042e+00
-3.10557246e-01 -9.98767018e-01 -6.50513768e-01 -3.42163414e-01
1.54966667e-01 1.22124076e+00 4.02911335e-01 3.90539974e-01
-1.48090586e-01 -3.48860592e-01 8.19368124e-01 3.98905993e-01
3.70416433e-01 -5.65096319e-01 -1.02430856e+00 5.73520303e-01
1.73033640e-01 -1.16074479e+00 -3.44283253e-01 -4.74869274e-02
-4.95565414e-01 -8.87054086e-01 1.72469199e-01 -3.36060464e-01
-1.06180161e-01 2.75459558e-01 1.52001452e+00 5.02208173e-02
-3.00512195e-01 1.04903364e+00 -2.60472465e-02 -3.24453294e-01
-5.12258947e-01 5.63043617e-02 2.44988650e-01 -4.83406484e-01
2.62707323e-01 -8.86104822e-01 -5.47798991e-01 2.10941494e-01
-6.06243193e-01 -2.34612525e-02 8.90306354e-01 -1.50037944e-01
1.56765878e-01 -7.77923167e-02 1.00528407e+00 -4.64795113e-01
1.04814088e+00 -7.15042830e-01 -6.30892456e-01 3.23612422e-01
-1.10528898e+00 -2.20952854e-01 5.53002238e-01 -3.37860644e-01
-9.19903219e-01 1.08170591e-01 -3.24270308e-01 3.25297415e-02
-3.25125419e-02 5.17724574e-01 -7.75495946e-01 1.08632013e-01
8.34647655e-01 -1.45317167e-01 2.92760283e-01 -2.88600504e-01
8.73087645e-01 1.27068663e+00 4.10083711e-01 -7.68656969e-01
3.41328859e-01 -9.45100375e-03 -6.77880049e-01 -8.29877138e-01
-5.89443445e-02 -2.20609218e-01 -2.33831406e-01 -1.65313274e-01
2.43858501e-01 -1.21373415e+00 -1.11089599e+00 9.50141177e-02
-8.21415782e-01 -6.09957039e-01 -3.72335821e-01 -1.44788295e-01
-9.05258179e-01 1.64188705e-02 -3.92823040e-01 -9.42454755e-01
-5.79588175e-01 -1.16530430e+00 1.26790607e+00 -1.00053273e-01
-7.22201228e-01 -6.47121310e-01 -9.97889340e-02 -4.70930599e-02
9.13581550e-01 2.16787815e-01 6.19767964e-01 -4.58665907e-01
-7.42151916e-01 -4.00532901e-01 -1.30487204e-01 1.82800025e-01
2.76865423e-01 5.63585818e-01 -9.54478502e-01 -4.97842580e-01
1.66705072e-01 1.73641786e-01 -6.20141387e-01 1.29484385e-01
7.48289406e-01 -4.29799974e-01 -6.19998574e-01 3.61370444e-01
1.53711128e+00 2.04328634e-02 3.76000762e-01 6.27063513e-02
4.52502578e-01 2.28722781e-01 7.79215693e-01 8.07810724e-01
5.71203053e-01 7.52046347e-01 9.79154781e-02 -2.53153164e-02
-2.47307673e-01 5.63042425e-02 3.96058142e-01 9.70903695e-01
5.08582890e-01 -2.07291543e-01 -1.17999554e+00 4.05085146e-01
-2.17855477e+00 -6.31129324e-01 -1.89307004e-01 2.18238735e+00
6.17147982e-01 4.23809111e-01 4.41599756e-01 -4.90956008e-02
3.77210416e-02 -9.87277403e-02 -8.98060501e-01 -8.77553448e-02
3.81161749e-01 -2.83465296e-01 5.28953731e-01 3.57432902e-01
-4.99872327e-01 5.62083900e-01 7.62468290e+00 4.53300208e-01
-1.06849205e+00 3.46578956e-01 3.92637312e-01 -2.69753337e-01
-3.22447777e-01 1.28916085e-01 -5.22138178e-01 2.64077395e-01
1.69515061e+00 -7.91654229e-01 3.22900265e-01 1.59920943e+00
5.12651324e-01 4.17914502e-02 -1.62749839e+00 1.04623878e+00
-3.84264857e-01 -1.63986015e+00 -7.95722723e-01 2.87441880e-01
2.17869967e-01 2.81295180e-01 -6.28425106e-02 4.23443913e-01
3.55849862e-01 -5.75643063e-01 1.18771279e+00 4.83400673e-01
1.29888833e+00 -6.53764307e-01 2.98657298e-01 4.74105418e-01
-1.59730291e+00 7.03219557e-03 4.46123004e-01 -5.13937712e-01
1.65864334e-01 6.71969652e-01 -9.68630791e-01 1.73150271e-01
4.12023604e-01 4.02543187e-01 -4.20350194e-01 6.32223487e-01
2.49415472e-01 4.10139918e-01 -3.71660680e-01 -4.76484448e-02
-3.13079715e-01 3.34198594e-01 2.36986428e-01 1.23653316e+00
2.59207219e-01 -7.26425946e-02 2.73461819e-01 7.97303975e-01
2.23445997e-01 7.72913620e-02 -6.68553233e-01 -1.10274307e-01
1.09206367e+00 1.60098457e+00 -3.62606198e-01 -2.84148961e-01
-6.83913112e-01 5.67513883e-01 -7.93635845e-03 2.68319488e-01
-8.12837303e-01 -3.39690745e-01 1.37962735e+00 6.37345493e-01
-1.57407895e-01 -5.51250637e-01 -8.79991412e-01 -9.88628805e-01
1.85253248e-01 -1.29879344e+00 5.10417372e-02 -5.24052382e-01
-1.35855782e+00 4.38218087e-01 -8.90282691e-02 -1.30592835e+00
-6.75650895e-01 -1.49094194e-01 -1.77737281e-01 6.33339643e-01
-7.63554811e-01 -1.50188494e+00 -2.83610195e-01 7.14307604e-03
4.53310966e-01 -2.83957183e-01 8.47408831e-01 4.94561136e-01
-5.43215513e-01 6.09884143e-01 3.27333319e-03 -6.14460289e-01
5.56548655e-01 -9.41491008e-01 7.83867300e-01 8.21814954e-01
-5.53061366e-01 1.12865841e+00 7.89818645e-01 -6.70008361e-01
-2.54515982e+00 -1.17449856e+00 4.35120821e-01 -7.78993726e-01
9.54275370e-01 -7.52003789e-01 -3.29146981e-01 1.21514571e+00
4.10929620e-01 -1.50716364e-01 8.64553332e-01 4.23857272e-01
-1.22163825e-01 -4.22635436e-01 -1.43145227e+00 8.01759481e-01
1.03434122e+00 -5.96780360e-01 -1.76085085e-01 2.94928938e-01
1.05091417e+00 -1.18218340e-01 -1.47955465e+00 3.53961378e-01
7.09759295e-01 -9.85086024e-01 8.13624620e-01 -8.52552280e-02
-2.14390114e-01 -6.52429640e-01 -6.09224677e-01 -6.78817689e-01
-4.63397026e-01 -1.45585537e+00 -4.47524339e-01 1.43527269e+00
4.36721951e-01 -8.75025749e-01 5.73830187e-01 1.27691281e+00
-5.27556360e-01 -6.15335464e-01 -3.34702581e-01 -9.02205229e-01
-3.99892658e-01 -1.14204681e+00 1.04296088e+00 6.99833274e-01
6.51463628e-01 7.94901431e-01 9.77311507e-02 3.11658531e-01
5.66778183e-01 -5.60435504e-02 1.29070675e+00 -1.01791143e+00
-6.60061538e-01 -6.05686545e-01 -2.82742471e-01 -1.23189557e+00
-3.69624496e-01 -4.97809559e-01 -3.60800438e-02 -1.63670099e+00
-5.10363206e-02 -7.44268894e-01 2.23883256e-01 6.93567038e-01
4.10359979e-01 -3.08997463e-02 2.28465289e-01 4.08854008e-01
-1.01475286e+00 2.00154111e-02 2.15167686e-01 1.92789137e-01
-5.58646858e-01 -1.14091001e-02 -1.44566870e+00 7.86128461e-01
5.93347847e-01 1.86612085e-01 -7.14933157e-01 -3.41361761e-01
3.96373630e-01 2.49702558e-01 -3.27608019e-01 -1.33292592e+00
3.04491639e-01 -3.28666091e-01 -3.97663191e-02 -6.72850847e-01
2.66937256e-01 -1.18987119e+00 8.98193359e-01 2.12164357e-01
1.55420512e-01 7.83147812e-02 4.64739174e-01 1.73788462e-02
3.60427499e-01 6.55065104e-02 5.22484064e-01 3.08994234e-01
-8.24992299e-01 1.61296606e-01 -7.30368853e-01 -2.50725240e-01
1.66022074e+00 -7.55412355e-02 -5.82172275e-01 -3.96374166e-01
-5.04448652e-01 2.13833645e-01 1.02966869e+00 3.21800470e-01
2.41049081e-01 -1.16256237e+00 -5.47311246e-01 2.06286073e-01
5.44368148e-01 -4.10337657e-01 -3.02386224e-01 6.70562506e-01
-6.13779187e-01 3.03564221e-01 2.81821992e-02 -6.87110245e-01
-1.02345419e+00 7.86109686e-01 3.30216169e-01 1.46997511e-01
-6.99294925e-01 1.86774358e-01 -1.21629305e-01 -3.80303800e-01
2.03816950e-01 -6.04276955e-01 7.95526683e-01 -3.06790501e-01
7.17980266e-01 4.56849843e-01 3.33437771e-01 3.76421325e-02
-7.51208782e-01 1.59292724e-02 2.80608982e-01 -1.62638605e-01
1.17547464e+00 -5.17096937e-01 -1.91309452e-01 8.86438668e-01
7.47805893e-01 5.61792493e-01 -1.17612684e+00 3.08595926e-01
2.66713768e-01 -1.63760602e-01 8.88804197e-02 -9.23347235e-01
-5.01159370e-01 5.91840073e-02 4.92976308e-01 7.83928812e-01
1.25348997e+00 2.05790624e-01 5.48876345e-01 2.38964140e-01
7.75637209e-01 -9.16529775e-01 -2.37910748e-01 -1.58722356e-01
8.89015496e-01 -7.94851303e-01 3.03683072e-01 -5.67088485e-01
-3.79180223e-01 9.57435310e-01 5.01586676e-01 2.10598841e-01
6.69884443e-01 1.33168709e+00 -1.53105035e-01 -1.82177827e-01
-1.33303535e+00 -5.02148084e-02 -2.83644676e-01 7.76303649e-01
9.49288130e-01 4.15580899e-01 -4.29334044e-01 9.76040721e-01
-1.12099372e-01 3.96934509e-01 4.89329875e-01 1.31925201e+00
-5.51014185e-01 -1.26966035e+00 -2.47935712e-01 7.02082932e-01
-1.63813323e-01 -7.73965418e-02 -2.10791945e-01 1.14384782e+00
-5.57440594e-02 1.49788356e+00 -6.06704019e-02 -4.58170056e-01
5.15134513e-01 6.06460050e-02 1.15896225e-01 -5.46420455e-01
-7.82806933e-01 3.02157611e-01 8.58805656e-01 -8.05295289e-01
3.27892691e-01 -7.51853883e-01 -1.39770365e+00 -7.44060397e-01
-3.85811143e-02 -3.04290891e-01 1.02153254e+00 5.01398027e-01
1.24833393e+00 9.12689865e-01 4.53487515e-01 -3.74779642e-01
-7.06946254e-01 -8.85460198e-01 -4.55880791e-01 -7.42858127e-02
5.32731228e-02 -3.46623287e-02 8.70991871e-02 2.43171960e-01] | [8.527649879455566, 7.216139793395996] |
943170c7-13ca-4160-b7ef-ceac4417e817 | plan-eliminate-and-track-language-models-are | 2305.02412 | null | https://arxiv.org/abs/2305.02412v2 | https://arxiv.org/pdf/2305.02412v2.pdf | Plan, Eliminate, and Track -- Language Models are Good Teachers for Embodied Agents | Pre-trained large language models (LLMs) capture procedural knowledge about the world. Recent work has leveraged LLM's ability to generate abstract plans to simplify challenging control tasks, either by action scoring, or action modeling (fine-tuning). However, the transformer architecture inherits several constraints that make it difficult for the LLM to directly serve as the agent: e.g. limited input lengths, fine-tuning inefficiency, bias from pre-training, and incompatibility with non-text environments. To maintain compatibility with a low-level trainable actor, we propose to instead use the knowledge in LLMs to simplify the control problem, rather than solving it. We propose the Plan, Eliminate, and Track (PET) framework. The Plan module translates a task description into a list of high-level sub-tasks. The Eliminate module masks out irrelevant objects and receptacles from the observation for the current sub-task. Finally, the Track module determines whether the agent has accomplished each sub-task. On the AlfWorld instruction following benchmark, the PET framework leads to a significant 15% improvement over SOTA for generalization to human goal specifications. | ['Shrimai Prabhumoye', 'Tom Mitchell', 'Yuanzhi Li', 'Amos Azaria', 'Ruslan Salakhutdinov', 'Yonatan Bisk', 'So Yeon Min', 'Yue Wu'] | 2023-05-03 | null | null | null | null | ['instruction-following'] | ['natural-language-processing'] | [ 2.90192038e-01 5.41825950e-01 -8.88575912e-02 -3.09100866e-01
-7.77358174e-01 -6.70076072e-01 8.59862268e-01 1.43220618e-01
-4.35209364e-01 5.87228835e-01 6.14445925e-01 -6.16370797e-01
1.22610256e-01 -7.45154440e-01 -8.22694540e-01 -3.65349263e-01
1.69814810e-01 7.50420392e-01 3.87861311e-01 -1.40822113e-01
2.17736617e-01 3.59125614e-01 -1.28739285e+00 6.47858799e-01
8.88848186e-01 5.16995907e-01 5.34168601e-01 6.24935091e-01
-8.43193009e-02 1.63726699e+00 -7.40157485e-01 -1.01446435e-01
2.08388150e-01 -4.47325617e-01 -1.06817091e+00 7.11119100e-02
3.72275740e-01 -6.54784262e-01 -2.58159190e-01 6.58826590e-01
1.76852867e-01 5.13212264e-01 6.30920649e-01 -1.25292468e+00
-4.62697327e-01 9.48184133e-01 -1.56051055e-01 -1.54495791e-01
4.23040062e-01 6.69432938e-01 8.58099818e-01 -4.55224097e-01
6.03926241e-01 1.57847142e+00 2.57556409e-01 7.02037752e-01
-1.31204915e+00 -3.99502695e-01 8.04391265e-01 1.03286991e-03
-1.04702461e+00 -6.40036225e-01 2.85107315e-01 -5.70922613e-01
1.49613822e+00 2.80416571e-03 5.16091704e-01 1.13546658e+00
5.32172024e-01 1.10764992e+00 1.03439295e+00 -3.94954592e-01
5.22768497e-01 -9.73939225e-02 8.16806927e-02 8.91597867e-01
7.08868206e-02 1.77174330e-01 -5.10990024e-01 2.80549172e-02
9.26099300e-01 -1.31028309e-01 -6.83956817e-02 -6.53623760e-01
-1.26739037e+00 6.65523589e-01 2.40565553e-01 1.00049123e-01
-4.93751168e-01 4.15242165e-01 3.96133155e-01 3.71709317e-01
-1.24802671e-01 1.12862408e+00 -5.92510104e-01 -3.58424932e-01
-5.72947800e-01 4.67782289e-01 8.58951807e-01 1.24364805e+00
7.10492551e-01 2.78481543e-01 -6.51296616e-01 3.73731226e-01
2.02818006e-01 1.30411655e-01 5.33064961e-01 -1.43853676e+00
8.96321535e-01 6.68356061e-01 2.93338865e-01 -3.53150249e-01
-6.01962268e-01 -2.49261349e-01 -2.54952103e-01 4.72910792e-01
3.80450696e-01 -1.57447830e-01 -1.08841825e+00 2.05358624e+00
2.53587961e-01 6.17096461e-02 1.58590242e-01 7.20392227e-01
5.15122056e-01 7.20421612e-01 6.30002677e-01 -5.91661334e-02
1.25847554e+00 -1.53327632e+00 -4.78678256e-01 -9.34257150e-01
1.01522517e+00 -3.81579965e-01 1.33818829e+00 5.20436287e-01
-1.38472247e+00 -5.37575781e-01 -9.15446997e-01 -2.60748059e-01
-1.60383448e-01 -3.35630588e-02 8.87363017e-01 9.80805606e-02
-1.08498240e+00 4.90750968e-01 -1.19802380e+00 -9.26751494e-02
3.49549711e-01 3.82083714e-01 -3.72550100e-01 -1.59656480e-01
-8.31451297e-01 1.25429964e+00 5.74695110e-01 -3.85110289e-01
-1.72930336e+00 -6.76120341e-01 -1.32546008e+00 3.72934312e-01
7.34075665e-01 -8.51783276e-01 1.82248259e+00 -6.45390928e-01
-1.88785565e+00 6.26614928e-01 -7.43325651e-02 -3.65384012e-01
2.61790633e-01 -3.50336701e-01 1.43956318e-01 -1.86099246e-01
1.50984094e-01 7.27170646e-01 7.68689811e-01 -1.13284147e+00
-7.18069017e-01 -1.95197746e-01 7.55432010e-01 5.10939181e-01
1.54406071e-01 -9.35856402e-02 -5.29103696e-01 -5.67810953e-01
-1.80397800e-03 -1.22183287e+00 -3.91009122e-01 -6.22841060e-01
-3.48603100e-01 -3.67102087e-01 2.07509398e-01 -3.98307592e-01
1.19009864e+00 -2.03982639e+00 5.16469777e-01 -2.46363059e-01
2.87005007e-01 1.21305779e-01 -7.57994711e-01 3.48819554e-01
-1.84583917e-01 1.23108074e-01 1.39660582e-01 -5.67947447e-01
2.04534769e-01 2.56918699e-01 -3.33926976e-01 1.06218025e-01
1.86526120e-01 1.00132585e+00 -1.03687668e+00 -4.62067306e-01
3.24082792e-01 -4.97616529e-02 -9.74407673e-01 5.14043987e-01
-8.30074608e-01 4.59251344e-01 -7.82870233e-01 3.20787936e-01
1.25085115e-01 -6.12777434e-02 2.91434169e-01 1.59476578e-01
-1.23715721e-01 1.03404474e+00 -1.08181262e+00 2.01334405e+00
-8.75124395e-01 2.34365448e-01 2.32663453e-01 -5.36717772e-01
2.48269588e-01 3.09894711e-01 1.44015282e-01 -8.89617503e-01
-4.79759611e-02 -1.63603902e-01 2.02316657e-01 -5.64918935e-01
3.07210326e-01 -4.64797616e-02 -3.17061335e-01 7.64708936e-01
1.05980322e-01 -5.07925689e-01 4.57051843e-01 2.58209795e-01
1.39022517e+00 7.33826697e-01 4.64071840e-01 -1.98110774e-01
3.81059825e-01 2.59908825e-01 6.12862825e-01 1.03978539e+00
-1.98318496e-01 2.53495753e-01 5.52821279e-01 -5.48125565e-01
-7.59935975e-01 -8.70231092e-01 6.21268809e-01 1.75928581e+00
-1.32242978e-01 -6.82251513e-01 -8.32099795e-01 -8.33095968e-01
-1.27595693e-01 1.30848372e+00 -4.67971027e-01 -3.89713258e-01
-1.12057126e+00 -2.60265768e-01 3.42650324e-01 6.67894781e-01
2.06374958e-01 -1.51098347e+00 -1.04800189e+00 4.10205722e-01
-1.46818966e-01 -1.02507663e+00 -6.69452190e-01 5.76498985e-01
-7.81633079e-01 -8.55138779e-01 4.67377529e-02 -7.79902995e-01
8.04446578e-01 9.21577588e-02 1.55103266e+00 6.21496327e-02
1.24803506e-01 2.49146372e-01 -1.35095477e-01 -4.74395424e-01
-6.21315539e-01 1.79902360e-01 5.75343035e-02 -4.76343930e-01
2.10972071e-01 -4.23391372e-01 -2.92004019e-01 1.36216372e-01
-5.69357991e-01 5.65995514e-01 8.36203158e-01 6.56489134e-01
5.14209986e-01 -1.42728072e-02 2.75789052e-01 -8.99999857e-01
7.10669339e-01 -1.50689363e-01 -7.19449639e-01 2.64037162e-01
-4.39208210e-01 6.33562446e-01 7.02973068e-01 -5.02705812e-01
-1.23111522e+00 2.93033689e-01 2.03942031e-01 -1.99120030e-01
-3.64853770e-01 4.25538957e-01 -4.56090778e-01 3.13277096e-01
6.53942049e-01 1.84837535e-01 -2.53922075e-01 -2.19547510e-01
4.32261795e-01 6.59810901e-02 3.63606513e-01 -1.16140759e+00
7.17342317e-01 -7.98601061e-02 -1.97033167e-01 -1.14781305e-01
-1.20458925e+00 -1.73262388e-01 -3.81705195e-01 1.84999958e-01
1.02196980e+00 -1.10889840e+00 -9.92058575e-01 2.50695109e-01
-1.18513620e+00 -1.35390806e+00 -4.16828901e-01 3.41567934e-01
-9.92268145e-01 -2.37893716e-01 -8.33516359e-01 -5.97857535e-01
-1.67958960e-02 -1.75728142e+00 1.07714891e+00 1.25450969e-01
-6.37044668e-01 -7.62148857e-01 -6.29726648e-02 4.98362899e-01
4.64048952e-01 -6.81845173e-02 1.30202138e+00 -6.25470757e-01
-7.55235255e-01 7.00893328e-02 2.59779423e-01 -4.40518511e-03
5.16306050e-02 -5.17320991e-01 -8.07159960e-01 -3.00361603e-01
2.20079020e-01 -7.63420880e-01 5.28795481e-01 2.06032485e-01
1.25855553e+00 -5.16923428e-01 -3.44819546e-01 5.61470330e-01
9.96515810e-01 4.87910300e-01 6.06355906e-01 5.95574439e-01
7.41320491e-01 5.23379147e-01 6.56392455e-01 1.56765059e-01
6.35930300e-01 7.36343443e-01 4.67410982e-01 2.58392572e-01
-3.07974100e-01 -5.15976071e-01 7.88326323e-01 4.11621928e-01
9.35200527e-02 -2.30211869e-01 -1.01302278e+00 4.05037642e-01
-1.95110381e+00 -9.28392112e-01 3.92007768e-01 2.06027317e+00
1.18596840e+00 4.77100134e-01 -2.33793765e-01 -3.76832396e-01
-7.73123279e-02 2.86977947e-01 -6.85828805e-01 -5.66434860e-01
4.98761714e-01 1.86467499e-01 2.30046406e-01 9.96677399e-01
-9.85404432e-01 1.45498884e+00 6.18670940e+00 6.37679219e-01
-7.85553157e-01 7.66481608e-02 3.87969375e-01 -2.05914065e-01
-1.91195488e-01 7.86244497e-02 -1.03127456e+00 -3.40049453e-02
9.90583599e-01 -4.98527512e-02 7.50517607e-01 8.28459382e-01
2.88387001e-01 -1.54722139e-01 -1.79241943e+00 5.69785714e-01
2.84522530e-02 -1.03044903e+00 3.32378179e-01 -1.59019694e-01
7.51133323e-01 3.56288031e-02 1.71895735e-02 1.13903022e+00
1.04189205e+00 -1.21315110e+00 1.12617552e+00 1.76367044e-01
4.60740238e-01 -4.24217135e-01 2.30084196e-01 9.32390571e-01
-1.03257287e+00 -3.06884736e-01 -6.06552772e-02 -4.87370282e-01
8.34744722e-02 -1.16485938e-01 -1.03669393e+00 4.84848432e-02
1.76830336e-01 2.72844166e-01 -4.35408056e-01 5.44367969e-01
-8.15300107e-01 4.15602386e-01 -9.38715413e-02 1.60792872e-01
5.32506227e-01 -9.42875296e-02 3.89123797e-01 9.52348351e-01
-1.49163082e-01 4.18396771e-01 6.67392492e-01 9.40796554e-01
2.53821649e-02 -2.43352726e-01 -4.30893064e-01 -2.23444611e-01
3.97985637e-01 1.03839481e+00 -3.21953058e-01 -5.23306072e-01
-4.12112623e-01 7.43557215e-01 7.45088041e-01 6.15033746e-01
-7.48044550e-01 3.95340547e-02 7.89187074e-01 1.11204490e-01
6.00706227e-02 -4.26993251e-01 -4.32049572e-01 -1.21259809e+00
-2.41931245e-01 -1.53021061e+00 3.04802746e-01 -9.53569770e-01
-4.86674517e-01 5.34138501e-01 1.84619799e-01 -7.70893097e-01
-5.72184563e-01 -5.67824721e-01 -6.28417253e-01 8.31693351e-01
-1.29585803e+00 -1.17857039e+00 -1.94608390e-01 5.46281934e-01
1.05418217e+00 -1.42422155e-01 1.08864951e+00 -7.10340440e-02
-6.74231410e-01 3.02073181e-01 -6.25476122e-01 -5.48827238e-02
5.18702209e-01 -1.47040832e+00 5.43690979e-01 7.86153197e-01
-1.67779088e-01 9.21429336e-01 7.76323736e-01 -7.34549344e-01
-1.48254514e+00 -1.08287871e+00 8.88451755e-01 -1.06098926e+00
5.06605387e-01 -5.89963138e-01 -6.01901412e-01 1.31820059e+00
9.96382833e-02 -3.80367547e-01 2.54114717e-01 2.40013957e-01
-4.13828045e-01 7.82370716e-02 -7.90737331e-01 1.09269583e+00
1.08467782e+00 -4.73790586e-01 -1.12658918e+00 4.45130497e-01
1.00273585e+00 -7.86439896e-01 -5.88867247e-01 5.78916669e-02
2.65211403e-01 -6.56213522e-01 1.01749408e+00 -1.07672942e+00
5.97172737e-01 -3.82825792e-01 8.23675618e-02 -1.54459310e+00
-6.99547768e-01 -8.15486372e-01 -2.32207179e-01 7.91658580e-01
5.54500461e-01 -3.49366218e-01 5.90352654e-01 1.08951223e+00
-6.51741564e-01 -7.11159587e-01 -4.36339706e-01 -6.66760802e-01
1.58352852e-01 -4.86113191e-01 7.13493049e-01 5.74976504e-01
2.20810100e-01 7.11459756e-01 -1.86888739e-01 1.97090566e-01
3.39474112e-01 2.07117707e-01 9.72953737e-01 -9.47629392e-01
-6.27776563e-01 -5.56287766e-01 4.62239146e-01 -1.32970464e+00
5.53210080e-01 -7.98910737e-01 4.35308933e-01 -1.79149365e+00
1.68445170e-01 -5.67552269e-01 -2.36565217e-01 1.14187717e+00
-1.39068827e-01 -4.72800255e-01 4.74251449e-01 1.61565557e-01
-9.48235035e-01 3.65157366e-01 1.59527278e+00 -2.43941233e-01
-5.93883395e-01 -1.21148430e-01 -8.21734905e-01 1.00379086e+00
7.10625112e-01 -5.05813181e-01 -7.21867740e-01 -1.03219259e+00
2.32428893e-01 3.07408214e-01 -6.50630444e-02 -1.13733375e+00
3.63147229e-01 -7.86250353e-01 1.82627097e-01 -2.83093303e-02
3.70602876e-01 -7.73758948e-01 -1.54970903e-02 5.92437327e-01
-8.08401048e-01 2.76248157e-01 4.57544327e-01 9.43994597e-02
-1.24980994e-01 -2.93594480e-01 6.55523121e-01 -4.83500838e-01
-8.05899203e-01 1.83348551e-01 -6.04534209e-01 5.47212586e-02
1.07687640e+00 5.61000407e-02 -5.05264044e-01 -3.13551098e-01
-7.86362469e-01 5.59394479e-01 4.71341521e-01 4.60429192e-01
2.65535146e-01 -1.01320982e+00 -5.60047030e-01 1.84622034e-01
1.54710755e-01 4.27243292e-01 -2.13922448e-02 6.87583625e-01
-3.73849839e-01 6.48328424e-01 -2.03964561e-01 -8.33663866e-02
-8.76762450e-01 8.41402292e-01 3.67838919e-01 -7.97853410e-01
-6.41676188e-01 8.53439867e-01 8.74960065e-01 -5.03330171e-01
4.76550609e-01 -8.24204803e-01 -2.68057346e-01 -2.16150776e-01
6.50717795e-01 4.78868112e-02 -1.18280619e-01 -2.17967838e-01
-3.62595916e-01 2.26217031e-01 -3.31569135e-01 -1.81264371e-01
1.33220756e+00 1.32396847e-01 -2.91151982e-02 3.65467399e-01
6.02369189e-01 -2.41037048e-02 -1.55548692e+00 -1.11483529e-01
6.28477484e-02 8.75664577e-02 1.98650363e-04 -1.20592844e+00
-6.17961586e-01 7.28626251e-01 -1.64397433e-01 -1.02529898e-01
8.68906856e-01 -1.00899555e-01 4.43102807e-01 8.00684094e-01
6.33351922e-01 -1.12438011e+00 4.05053318e-01 1.13026905e+00
1.17363930e+00 -1.04569125e+00 -4.48952988e-03 -1.57921717e-01
-8.58353615e-01 7.94643462e-01 1.23772967e+00 1.35477722e-01
-5.28085530e-02 4.64625806e-01 -8.16111490e-02 -1.84473738e-01
-1.33675468e+00 -3.45997908e-03 -4.29177145e-03 4.57437813e-01
4.45381314e-01 1.74163133e-01 1.13164440e-01 6.00520611e-01
-2.90684342e-01 4.57227714e-02 5.59389353e-01 1.18006694e+00
-4.76548940e-01 -9.24016118e-01 -2.10167348e-01 4.34672803e-01
-2.41426498e-01 -1.78737596e-01 -2.03546897e-01 7.20255613e-01
8.36813599e-02 8.50406587e-01 -1.45826526e-02 -1.51285827e-01
6.65781915e-01 3.31018567e-01 6.00453973e-01 -1.32031202e+00
-9.40708637e-01 -1.58197850e-01 3.05604875e-01 -1.21552014e+00
-5.32736853e-02 -5.37276626e-01 -1.57907605e+00 2.55799461e-02
1.23483717e-01 1.05242088e-01 2.54602373e-01 1.06546903e+00
3.18471551e-01 8.75194907e-01 7.72170648e-02 -8.88892889e-01
-1.01161432e+00 -1.03703105e+00 1.20715601e-02 4.77224022e-01
3.74885827e-01 -6.40994310e-01 -8.12273398e-02 1.53533876e-01] | [4.323593616485596, 1.002880334854126] |
804e2d29-2f23-46aa-89b2-f876ae1d546f | fully-convolutional-line-parsing | 2104.11207 | null | https://arxiv.org/abs/2104.11207v3 | https://arxiv.org/pdf/2104.11207v3.pdf | Fully Convolutional Line Parsing | We present a one-stage Fully Convolutional Line Parsing network (F-Clip) that detects line segments from images. The proposed network is very simple and flexible with variations that gracefully trade off between speed and accuracy for different applications. F-Clip detects line segments in an end-to-end fashion by predicting each line's center position, length, and angle. We further customize the design of convolution kernels of our fully convolutional network to effectively exploit the statistical priors of the distribution of line angles in real image datasets. We conduct extensive experiments and show that our method achieves a significantly better trade-off between efficiency and accuracy, resulting in a real-time line detector at up to 73 FPS on a single GPU. Such inference speed makes our method readily applicable to real-time tasks without compromising any accuracy of previous methods. Moreover, when equipped with a performance-improving backbone network, F-Clip is able to significantly outperform all state-of-the-art line detectors on accuracy at a similar or even higher frame rate. In other word, under same inference speed, F-Clip always achieving best accuracy compare with other methods. Source code https://github.com/Delay-Xili/F-Clip. | ['Yi Ma', 'Shuai Wu', 'Haigang Gong', 'Xiaojun Yuan', 'Xili Dai'] | 2021-04-22 | null | null | null | null | ['line-segment-detection'] | ['computer-vision'] | [-1.66433468e-01 -2.12148041e-01 -9.08673629e-02 -3.67160797e-01
-6.59227908e-01 -7.41624951e-01 1.27679020e-01 3.02956492e-01
-4.97115612e-01 1.58131585e-01 -4.97620046e-01 -5.23603261e-01
3.78583014e-01 -9.63850796e-01 -1.07548416e+00 -2.22798824e-01
-7.85792544e-02 2.17566252e-01 7.37737298e-01 -1.27667069e-01
3.04546922e-01 8.16521704e-01 -1.05863619e+00 1.47384778e-01
6.24024630e-01 1.21347928e+00 -2.64528599e-02 1.04607987e+00
-4.69155274e-02 5.36280751e-01 -2.69112110e-01 -5.32802403e-01
3.64299387e-01 3.44599672e-02 -4.63761896e-01 -3.26216333e-02
7.39203751e-01 -6.85650766e-01 -5.92982471e-01 6.65482104e-01
5.13381124e-01 -3.34431469e-01 2.24855497e-01 -9.91471171e-01
-7.83050358e-02 5.36925554e-01 -8.44959855e-01 2.13195264e-01
2.64060467e-01 1.25153258e-01 6.43107891e-01 -7.66762674e-01
3.46945375e-01 8.22970748e-01 1.15991712e+00 -4.61096503e-02
-9.82649088e-01 -4.68671083e-01 -5.00335395e-02 6.73247874e-02
-1.35046732e+00 -2.61113226e-01 6.96208179e-01 -2.80366391e-01
8.62273932e-01 7.87661299e-02 5.89893281e-01 7.16892481e-01
2.87723362e-01 9.15993333e-01 6.80205226e-01 -4.64638501e-01
1.02220647e-01 -2.61527449e-01 1.57607689e-01 1.02894604e+00
1.72335416e-01 -2.70489454e-01 -4.74803954e-01 2.06359625e-01
1.26272237e+00 -1.36987269e-01 -3.17506999e-01 -5.52157760e-01
-1.32753623e+00 6.04933321e-01 5.53196132e-01 6.32751063e-02
-3.96982133e-02 4.04328138e-01 5.60788155e-01 -5.81825618e-03
2.41191179e-01 4.60084565e-02 -5.03221929e-01 -2.57651627e-01
-1.00788558e+00 2.16400027e-01 7.56178081e-01 1.16643667e+00
5.74908972e-01 -1.63873807e-01 -1.33523092e-01 5.42226076e-01
4.39066812e-03 5.42351902e-01 1.02822371e-01 -8.71459603e-01
6.04170263e-01 3.04210365e-01 1.00418225e-01 -1.07470727e+00
-9.66014266e-01 -8.63470495e-01 -5.97849846e-01 2.68387586e-01
6.97329938e-01 -4.13046092e-01 -5.91166258e-01 1.33587694e+00
3.08575183e-01 1.35844916e-01 -3.89392018e-01 9.35114801e-01
4.74696547e-01 6.09651625e-01 -2.92589664e-01 1.59456640e-01
1.58503640e+00 -1.27939689e+00 -1.70341447e-01 -3.73204589e-01
7.39372671e-01 -1.04748929e+00 9.65832233e-01 5.54230869e-01
-1.10208249e+00 -6.43865526e-01 -1.34734595e+00 -3.84934276e-01
-3.03798210e-04 6.60706162e-01 9.60780740e-01 6.16995096e-01
-9.14382219e-01 6.41983032e-01 -1.11074865e+00 -2.62169331e-01
5.16188860e-01 4.52330798e-01 -1.82577878e-01 1.76053256e-01
-4.89789754e-01 1.75540447e-01 3.24973077e-01 2.19623819e-01
-5.18729948e-02 -1.02843106e+00 -7.94366062e-01 2.12506548e-01
5.02345324e-01 -7.93503046e-01 1.60760629e+00 -8.73819232e-01
-1.73133230e+00 6.12715483e-01 -1.26263753e-01 -6.33077443e-01
8.95382464e-01 -7.57930279e-01 -1.26880363e-01 3.93971980e-01
3.10051274e-02 7.18745291e-01 5.92093885e-01 -8.69329393e-01
-8.20561409e-01 -1.37487680e-01 7.78483823e-02 -8.10328200e-02
-1.41038105e-01 -1.39310002e-01 -1.09979117e+00 -5.07232428e-01
-7.22482055e-02 -1.02508938e+00 -3.01180370e-02 5.90307474e-01
-5.17277122e-01 -2.29147524e-02 8.48916769e-01 -2.49629810e-01
1.08274806e+00 -2.07353187e+00 -6.63307786e-01 4.81644832e-02
1.55263290e-01 5.14627695e-01 1.88942015e-01 5.12316704e-01
5.67794293e-02 -3.11960340e-01 -6.02704845e-02 -3.23505968e-01
-1.92912251e-01 -1.98839694e-01 -2.63084322e-01 6.94045007e-01
2.05008209e-01 7.70083964e-01 -6.88679814e-01 -3.74990046e-01
6.39393091e-01 6.90071881e-01 -6.44029915e-01 2.76409723e-02
-3.83341089e-02 2.07313076e-01 -4.20344800e-01 5.35020411e-01
8.33552480e-01 -3.74025434e-01 1.02640741e-01 -4.40412790e-01
-4.99722302e-01 -6.79803938e-02 -1.03474975e+00 1.94227338e+00
-5.63437402e-01 1.07874560e+00 -2.33971313e-01 -6.75011039e-01
1.09357953e+00 -1.47752285e-01 4.22521323e-01 -8.07752550e-01
3.64415050e-01 2.22266078e-01 -1.28334686e-01 -2.26724491e-01
5.42940080e-01 7.67591178e-01 -4.05954719e-02 4.49763164e-02
-1.13230154e-01 4.61491868e-02 3.45975041e-01 1.25156045e-01
9.80463505e-01 4.10392672e-01 1.89560369e-01 -2.86775023e-01
3.62147719e-01 -8.54782239e-02 4.54534560e-01 8.49212646e-01
-1.79623291e-01 8.03609848e-01 6.68506145e-01 -6.95042968e-01
-1.39172578e+00 -9.82907236e-01 -2.31546119e-01 8.38448346e-01
2.74649590e-01 -4.40187633e-01 -9.47896540e-01 -3.77621591e-01
-1.13190778e-01 3.56178105e-01 -3.81546438e-01 3.25253487e-01
-8.71258736e-01 -3.59153509e-01 6.65951431e-01 9.50680673e-01
8.75546157e-01 -5.59928000e-01 -9.84192193e-01 2.76606590e-01
3.38117301e-01 -1.60961759e+00 -6.38913989e-01 1.11522399e-01
-6.78064823e-01 -1.09596896e+00 -6.58202708e-01 -8.54120791e-01
5.84823489e-01 3.78792107e-01 1.08965039e+00 -2.82842182e-02
-4.28771555e-01 8.72234348e-03 -3.05040896e-01 -2.39422515e-01
1.28711626e-01 2.70550042e-01 -4.72818315e-01 -1.68555483e-01
1.65882841e-01 -3.68080825e-01 -1.06772840e+00 2.80762881e-01
-6.70168936e-01 4.33800876e-01 3.85370046e-01 4.50655550e-01
5.29297829e-01 -2.09432736e-01 -5.66170812e-02 -8.92242908e-01
2.08819583e-01 1.65905766e-02 -1.08133721e+00 -2.06310581e-02
-3.74732614e-01 -4.77130525e-03 8.94111812e-01 -8.87486264e-02
-7.69768894e-01 4.20974880e-01 -3.83774906e-01 -2.59657353e-01
-2.40882814e-01 2.29392946e-02 3.20536435e-01 -2.34638810e-01
4.63428974e-01 -2.18377076e-02 -2.13971794e-01 -1.89093709e-01
4.31893736e-01 3.82526249e-01 1.04691207e+00 -3.93663317e-01
5.18919170e-01 7.57528901e-01 1.58127427e-01 -9.27785397e-01
-8.85510445e-01 -5.05360067e-01 -9.35525060e-01 -1.93997756e-01
5.58764935e-01 -8.94775033e-01 -1.00863051e+00 9.33467627e-01
-1.21602929e+00 -5.43226480e-01 1.83270574e-01 3.52435350e-01
-6.34328902e-01 5.47963023e-01 -8.87446642e-01 -3.47599268e-01
-6.14699006e-01 -1.13285303e+00 1.22415245e+00 5.98225057e-01
1.30738437e-01 -8.91515970e-01 -2.00985566e-01 1.31571591e-01
3.39816660e-01 4.51054901e-01 4.36496019e-01 -2.36002058e-01
-5.92063248e-01 -3.10873866e-01 -6.45083368e-01 -5.13524516e-03
-2.94630378e-01 5.29365361e-01 -8.05529535e-01 -2.80091912e-01
-6.64674461e-01 -1.37407526e-01 7.86919653e-01 4.79299963e-01
1.51126218e+00 9.00019482e-02 -3.03000242e-01 1.08348286e+00
1.77037156e+00 3.72145697e-02 5.95460773e-01 6.00150168e-01
7.55796134e-01 6.34010062e-02 5.09801149e-01 6.10616028e-01
5.52608013e-01 9.20113742e-01 2.50600547e-01 -5.21882176e-01
-1.71606168e-01 -2.25349024e-01 6.92464262e-02 4.91451919e-01
1.76758036e-01 -5.08638799e-01 -9.60606098e-01 3.05223435e-01
-1.95205879e+00 -6.98386967e-01 -5.53640723e-01 2.19534254e+00
5.60805678e-01 6.93066061e-01 3.49877775e-01 7.99736679e-02
6.05156600e-01 2.32766904e-02 -5.09332836e-01 -6.50852859e-01
5.80040365e-02 2.89139688e-01 1.09528482e+00 3.12699109e-01
-1.41981244e+00 9.80230868e-01 5.99056244e+00 7.87857473e-01
-1.49516356e+00 -4.10940677e-01 7.71032453e-01 1.52445734e-01
1.42096281e-01 -5.14957830e-02 -8.98424089e-01 3.38390142e-01
7.26458609e-01 5.73331751e-02 4.35522124e-02 9.24845695e-01
3.99098337e-01 -2.18704775e-01 -9.25335169e-01 1.08709586e+00
-1.72183923e-02 -1.51278710e+00 -4.03485298e-01 -2.19994605e-01
4.21673387e-01 2.78920949e-01 -1.34864956e-01 1.31687475e-02
-4.85639349e-02 -6.47505879e-01 1.02617776e+00 3.03784668e-01
9.20706809e-01 -1.02695274e+00 7.47336864e-01 1.80363640e-01
-1.28409708e+00 2.08827242e-01 -2.96910822e-01 -1.05884135e-01
3.16388965e-01 7.94820189e-01 -8.66310000e-01 4.41854298e-01
5.97715795e-01 4.28628981e-01 -6.74338877e-01 1.38069820e+00
-2.15079039e-01 6.77688599e-01 -6.58859968e-01 6.60296604e-02
5.02504408e-01 -1.58956181e-02 1.99521147e-02 1.75492072e+00
2.76775330e-01 -1.85406744e-01 1.94477543e-01 5.03477335e-01
1.26490658e-02 5.02525680e-02 -2.67629176e-01 3.40890735e-01
6.06092036e-01 1.38418460e+00 -1.18401146e+00 -1.87986746e-01
-6.77188218e-01 1.14986455e+00 6.14305496e-01 1.89855155e-02
-1.26111674e+00 -7.86905706e-01 4.60242718e-01 1.66944340e-01
7.40606666e-01 -7.58780241e-01 -4.22197521e-01 -9.61920857e-01
1.48665681e-01 -2.78783828e-01 2.06184089e-02 -6.71157718e-01
-7.73972273e-01 7.10788965e-01 -4.00771290e-01 -1.11918044e+00
-1.05961137e-01 -9.43611026e-01 -6.13643110e-01 2.93755502e-01
-1.72586417e+00 -1.17286885e+00 -6.67278230e-01 5.14119089e-01
7.16265619e-01 2.67323554e-01 5.43837786e-01 2.82137692e-01
-6.58273757e-01 9.95019674e-01 1.84187844e-01 6.19360983e-01
6.93927109e-01 -1.16926575e+00 9.22535181e-01 9.37523782e-01
9.34788063e-02 3.01177502e-01 6.48775518e-01 -2.11570770e-01
-1.52383804e+00 -1.00826788e+00 3.34656626e-01 7.47230500e-02
6.14521086e-01 -4.76833284e-01 -5.96026838e-01 5.58517814e-01
1.65566549e-01 3.20821434e-01 4.51840967e-01 -1.09500684e-01
-5.34005046e-01 -2.34083593e-01 -7.40505874e-01 6.40812814e-01
9.20411587e-01 -1.44762337e-01 8.89502838e-02 3.44683737e-01
6.87653780e-01 -9.31751132e-01 -7.55469441e-01 3.38685900e-01
6.92096829e-01 -1.63303423e+00 9.77829039e-01 2.02467591e-01
4.82005358e-01 -3.04496467e-01 3.28877717e-01 -8.01216781e-01
-3.87058437e-01 -6.99995577e-01 2.23079994e-02 9.51137960e-01
3.13570857e-01 -5.87077737e-01 8.31428349e-01 1.39400661e-01
-3.01616907e-01 -1.06327665e+00 -6.00766242e-01 -6.58446550e-01
-9.90723297e-02 -7.35475302e-01 3.94593388e-01 4.37514305e-01
-2.25130677e-01 5.62301427e-02 -1.92218781e-01 4.81436521e-01
7.34700799e-01 3.44755322e-01 9.98031557e-01 -6.71776116e-01
-3.20767581e-01 -5.48278570e-01 -5.45986712e-01 -1.63212252e+00
-1.73536837e-01 -3.60679179e-01 -9.45988894e-02 -1.20577407e+00
-1.83994904e-01 -5.37570536e-01 1.81394652e-01 4.89426970e-01
1.13958667e-03 5.99957228e-01 3.27575892e-01 5.30554056e-02
-6.08911574e-01 -5.93698118e-03 9.84657347e-01 2.72393793e-01
-1.74872220e-01 -5.09545468e-02 -3.94127637e-01 1.00863981e+00
9.82347965e-01 -7.21840039e-02 -1.16500273e-01 -8.40689003e-01
1.01846077e-01 1.84119686e-01 2.90684700e-01 -1.47061718e+00
5.41455448e-01 2.41162896e-01 6.67793036e-01 -7.62430131e-01
9.85365584e-02 -5.41321635e-01 -2.16742098e-01 4.40356493e-01
-6.30803928e-02 2.02509090e-01 4.33460891e-01 2.46274710e-01
5.00762127e-02 -9.76547375e-02 8.01241934e-01 2.76166946e-01
-9.50077891e-01 4.68651384e-01 -8.43039528e-02 -1.96889520e-01
1.19259179e+00 -4.09268200e-01 -6.21472776e-01 -2.41500005e-01
-1.38294041e-01 2.27134034e-01 6.51525736e-01 3.71970922e-01
4.24052924e-01 -9.48901236e-01 -5.97841680e-01 3.27346623e-01
6.70959204e-02 2.22747743e-01 2.62758642e-01 7.03419030e-01
-1.61181700e+00 7.40737677e-01 -1.28150418e-01 -8.89283955e-01
-9.37477887e-01 3.58305842e-01 2.91848123e-01 -1.22466445e-01
-1.07042325e+00 7.46535301e-01 1.53120607e-02 2.56961081e-02
7.82899112e-02 -5.18806517e-01 2.84390390e-01 -4.29310560e-01
4.95251209e-01 2.70218432e-01 2.33644232e-01 -1.99043036e-01
-3.56044322e-01 9.20385122e-01 -2.11218715e-01 2.43107826e-01
1.23598230e+00 -9.07843583e-04 3.80291373e-01 1.29866853e-01
1.16937089e+00 3.81428719e-01 -1.77915013e+00 3.29380035e-02
-3.71831298e-01 -4.43532348e-01 2.20863685e-01 -5.84638238e-01
-1.23011470e+00 6.44995272e-01 4.68961120e-01 -5.13978582e-03
1.01720321e+00 -3.65142763e-01 1.17797804e+00 1.62034944e-01
4.96790648e-01 -7.53378451e-01 -8.30584988e-02 3.79820228e-01
3.60529363e-01 -1.14061797e+00 3.32520530e-02 -9.22990978e-01
-3.74329358e-01 1.82076716e+00 5.61170638e-01 -5.51032662e-01
4.13646609e-01 8.73822749e-01 1.11896314e-01 3.75953428e-02
-4.05147344e-01 3.03856321e-02 6.59988075e-02 3.29548091e-01
5.69775045e-01 9.18267369e-02 -2.51317501e-01 7.25639537e-02
-3.89192522e-01 1.50905222e-01 6.01561427e-01 8.88600588e-01
-4.23387080e-01 -8.84687245e-01 -2.69314408e-01 6.40984923e-02
-3.53488833e-01 5.29652685e-02 1.29296511e-01 9.48255122e-01
-1.49142951e-01 7.43950427e-01 4.99740064e-01 -1.38149172e-01
3.53517145e-01 -5.12066662e-01 4.59345281e-01 -2.78132055e-02
-3.98257434e-01 9.94640961e-02 3.22530977e-02 -1.05192137e+00
-8.84485915e-02 -4.26316023e-01 -1.49415183e+00 -6.49862587e-01
-2.85526246e-01 -3.58125806e-01 7.71135330e-01 6.75184190e-01
5.70227683e-01 4.60592240e-01 6.06733680e-01 -8.43475938e-01
-2.10578561e-01 -4.24832970e-01 -3.34171861e-01 8.53068084e-02
2.50288814e-01 -3.34954709e-01 1.33819327e-01 -7.47442991e-02] | [8.33554744720459, -1.6019787788391113] |
5ceb95a7-105c-46d1-972c-8700996cc1b8 | understanding-tables-with-intermediate-pre | 2010.00571 | null | https://arxiv.org/abs/2010.00571v2 | https://arxiv.org/pdf/2010.00571v2.pdf | Understanding tables with intermediate pre-training | Table entailment, the binary classification task of finding if a sentence is supported or refuted by the content of a table, requires parsing language and table structure as well as numerical and discrete reasoning. While there is extensive work on textual entailment, table entailment is less well studied. We adapt TAPAS (Herzig et al., 2020), a table-based BERT model, to recognize entailment. Motivated by the benefits of data augmentation, we create a balanced dataset of millions of automatically created training examples which are learned in an intermediate step prior to fine-tuning. This new data is not only useful for table entailment, but also for SQA (Iyyer et al., 2017), a sequential table QA task. To be able to use long examples as input of BERT models, we evaluate table pruning techniques as a pre-processing step to drastically improve the training and prediction efficiency at a moderate drop in accuracy. The different methods set the new state-of-the-art on the TabFact (Chen et al., 2020) and SQA datasets. | ['Syrine Krichene', 'Thomas Müller', 'Julian Martin Eisenschlos'] | 2020-10-01 | null | https://aclanthology.org/2020.findings-emnlp.27 | https://aclanthology.org/2020.findings-emnlp.27.pdf | findings-of-the-association-for-computational | ['table-based-fact-verification'] | ['natural-language-processing'] | [ 0.40808877 0.39329994 -0.32722855 -0.63234746 -1.1184582 -0.90446556
0.59523755 0.856365 -0.1364945 0.93683654 0.26654014 -1.0410935
-0.00970453 -1.2024466 -1.0534503 0.21309094 0.03395933 0.8245433
0.15095785 -0.45883343 0.37050965 0.14983389 -1.5497577 1.261355
0.9707079 1.4014025 -0.44277453 0.6728484 -0.23389077 1.3172756
-0.38140535 -1.2485255 0.2744694 -0.46820313 -1.208505 -0.25560194
0.60495013 -0.39153084 0.06484563 0.6423872 0.1066292 -0.18825164
0.68021387 -1.1195954 -0.30128932 1.4299295 -0.04023279 0.10631977
0.68906224 0.19066173 1.6456103 -0.72976285 0.62542737 1.1456453
0.85273796 0.26203707 -1.2376986 -0.22570759 -0.0162735 0.51763743
-0.88764673 -0.63947564 0.35311309 -0.08438394 1.6071299 0.6826292
0.7061695 0.7424948 0.1383329 0.96994543 0.9245021 -0.6632731
0.33395043 0.08614861 0.14080912 0.8421177 0.5508618 -0.5457631
-0.66324884 -0.15630746 0.10390909 -0.3194268 0.27026868 -0.22363244
-1.1880314 0.9106327 0.29704964 -0.07712536 -0.19933206 -0.03119739
0.73004 0.7763429 0.19779563 0.79387796 -0.82750803 -0.5101039
-1.0703855 0.82208675 1.3194602 0.9031836 0.6354384 -0.48990902
-0.42222294 0.4175031 -0.03898063 0.15045173 0.33767435 -0.90381724
1.196089 1.0912468 0.06666908 -0.62726957 -0.22437252 -0.25412577
-0.64306164 -0.10803588 0.893233 0.1138406 -0.4384855 1.3774494
0.33435005 -0.35942703 0.29684606 0.3601168 0.5513882 0.49731615
-0.35836607 -0.2285551 1.4617964 -0.8420005 -0.5786665 -0.39889473
1.2730246 -0.5913706 1.4358108 0.7383974 -1.5326656 -0.16262347
-1.229147 -0.51507336 -0.6403032 -0.01694841 1.0233355 0.708637
-0.68708676 0.7589866 -0.5742503 0.04008223 0.62647074 0.15838626
-0.21747266 -0.44572353 -1.4865894 1.1835352 0.30238795 -0.07589632
-0.28889868 -0.84940714 -1.1169897 0.22496669 0.9358633 -0.7412492
1.5090928 -0.31689864 -1.0879123 0.98294383 -0.25281236 -1.2160921
0.8514796 -0.20118459 -0.03799095 -0.28320163 0.23234636 0.2860624
0.25703198 -0.55666435 -0.5542291 -0.48289573 0.54550844 0.1918599
0.08267903 -0.01482029 -0.12714382 -0.32899415 0.14346667 -0.4329122
-0.02566637 -0.39687574 -0.74812245 -0.4201656 -0.01904982 -0.7079303
1.6673502 -1.5020925 -0.1875399 0.3188926 0.01370826 -0.14261352
0.12328725 0.609295 0.01508624 0.41148382 -0.71118504 -0.3607237
0.44826266 0.22140858 -0.34169054 0.09642436 0.69568235 1.0870472
-0.51384556 -0.6220932 0.03475419 -0.39871815 -1.1412373 0.09780397
-0.85768276 -0.23528276 0.03734564 0.91048974 0.46430033 -0.23178448
0.3531352 -0.18677565 0.10946111 1.2729352 -1.4938056 1.5145842
-0.5294613 0.3582505 -0.4644046 -0.9611598 0.5805352 0.01547392
0.22100112 -0.87034464 0.153595 0.45964545 0.13212486 -0.5496533
0.61613655 -0.16511874 -0.58488864 0.61978644 -0.20955108 -0.31682184
1.0467935 0.42170894 1.2794689 0.08538172 0.7684642 -0.15596421
0.7865363 0.5122333 0.3815485 0.7601506 0.23452617 0.4060416
1.0944527 -0.7220474 -1.0911953 -0.6933272 -0.30345017 1.0848821
-0.45554525 -0.9437748 -0.84515643 -0.9234127 0.37232882 1.017434
-0.74745953 -0.15919389 -0.845865 -0.54071814 0.69243765 0.49234545
0.25662178 -1.1932304 -0.7886197 0.2703315 -0.5509796 -1.1222868
-0.11802964 0.60349196 -0.8633207 -1.2675152 0.14320293 -0.3455272
0.33573598 -0.61778367 1.8037289 0.05705637 -0.22183916 -0.255246
-0.26857376 -0.5191528 -0.81128585 0.29539147 -0.15217009 -0.27347136
0.37661242 -0.30521092 -0.10203672 -0.03033482 -0.7232281 0.16405971
0.54955614 0.77627337 0.67580205 -0.13750921 0.43449733 -1.425412
0.6527975 -0.37293237 -0.6242858 0.45695806 -0.9848612 0.43457186
0.96426284 0.02004798 -0.67951155 -0.10786407 -0.3634757 0.32334667
0.0205864 0.9850612 -0.46780017 0.3961315 0.82557803 0.06653648
-0.01061574 -0.35273626 0.6049168 0.48088473 0.48529273 -0.7609687
0.55141675 0.03874971 0.36043376 0.08609286 -1.2439055 -0.24599178
-0.9179336 0.25779575 0.31277937 -0.7441223 -0.8099394 -0.14293194
-0.886952 -0.5285158 -0.46397758 0.00987171 -0.77253014 0.27342242
-0.66215587 -0.76985925 -0.46532932 -0.92267233 0.7878198 -0.4851434
-0.70543134 -0.6720074 -0.08807078 0.5358241 0.13852738 0.2621199
1.4517909 -0.9906792 -0.707359 -0.3568223 -0.0445734 0.2369634
-0.08938336 -0.1524475 -0.6959752 0.23859309 0.04208613 -1.0285295
0.8248011 0.01679017 1.1647575 -0.9234171 0.2394098 0.33863026
1.1543381 -0.2474422 0.7682722 0.76307505 0.32062814 0.72012013
0.98166764 0.4915588 0.6357208 0.34428167 0.3871034 0.44159594
0.13738123 -0.39802644 0.21086189 0.46005198 0.25985563 -0.07750044
-1.1056362 0.39637592 -1.6956223 -1.0285913 -0.33485696 2.1061149
1.455616 0.79277587 0.23126875 0.91768587 -0.05087617 -0.07823617
-0.48861778 -0.8596453 -0.13856155 0.32777295 0.36154863 0.6134069
-1.0862908 0.8474667 5.6379766 0.74974316 -0.5665685 -0.29909724
0.99683815 -0.23322695 -0.51725936 0.03255248 -0.7184882 0.29248488
1.2999903 -0.1459526 0.6204654 0.9398706 -0.33855006 -0.17589614
-1.722835 0.790073 0.03402781 -1.7081825 0.19374174 -0.34075186
0.3552057 -0.37651235 -0.06348857 0.80161494 0.03889734 -1.1976478
1.1045406 0.1494876 0.79048246 -0.89984775 1.0373485 0.47287422
-0.94575804 -0.15731859 -0.319397 -0.56010956 -0.06450119 0.6839091
-0.96996903 0.58673465 0.45145568 0.69651943 -1.0596642 0.4892335
-0.29458377 0.7498778 -0.45776328 -0.23121652 0.24816078 -0.00612267
-0.02437213 1.2330258 -0.04831066 -0.09477615 -0.20054981 0.8832255
-0.467083 0.11971474 -0.57656693 0.04192242 0.36398375 1.0299219
-0.52515495 -0.71857375 -0.2971636 0.6197639 0.58004856 -0.22642225
-0.73576164 -0.22592181 0.23927991 0.35115007 0.39022976 0.19298238
-1.1115882 -1.0853562 0.58703226 -1.3481063 0.77530086 -0.48085442
-1.0958474 0.33627284 0.03584311 -1.0115973 -0.72446525 -0.73189884
-0.5135574 0.7949842 -1.3057506 -0.9328664 0.00797805 0.3243694
0.41639763 0.10872315 0.8648492 0.12194355 -0.4085085 0.91465706
-0.23047617 0.13815413 0.6454216 -1.608282 0.9179978 0.80606467
0.24279474 0.7921237 0.65156776 -0.58553714 -1.783089 -0.99817556
1.5491587 -1.0993139 0.7308055 -0.6821073 -0.79848045 0.91117257
0.01870078 -0.20980667 0.6122397 0.4986985 -0.5875503 -0.30968434
-1.284738 0.5741995 0.9867163 -0.6102669 -0.87117046 0.22092772
0.58427775 -0.63043666 -0.9846184 0.53205746 0.65937454 -1.1698861
0.8326156 -0.9230451 1.1327558 0.09931245 -0.3990376 -1.0136114
0.11479422 -0.48536023 -0.622742 1.1915834 1.0695701 -0.43199658
0.90287924 0.7897218 -0.08543156 -1.3485795 -0.9092469 -0.5675009
0.22761375 -0.7307569 0.9790286 0.71953 0.6095574 0.66366136
-0.03830108 -0.39240938 0.33769742 0.5350915 0.88719314 -0.98906434
-0.41480413 -0.476805 0.05589115 -0.7517639 -0.22976616 -1.3041561
-0.04738864 -1.8159815 0.14262342 -0.34653547 -0.06619205 0.5111501
-0.2753372 0.04513353 0.09890893 -0.02395149 -0.7948311 0.0871084
0.9983422 -0.16744862 0.18243353 0.10761857 -0.9243953 0.53280056
0.74384606 -0.41991624 -0.39090303 -0.22502518 0.97115153 0.26763135
-0.02841163 -0.8875871 0.10536484 -0.01421012 0.29285952 -0.94299555
-0.00811758 -0.51375526 -0.35616887 0.6025255 -0.7810864 0.4798692
0.25980106 0.17623231 -0.28743747 -0.33906427 0.39764223 -0.36772203
-0.14902598 0.02649378 -0.29914358 0.47566774 0.54449016 -0.08204651
-0.27996182 -0.16501449 -0.3187179 0.2513372 0.24379191 -0.09637079
0.495764 -0.92269164 -0.7619952 0.26527265 0.2712232 0.5045334
-0.20272067 0.9060706 -0.70875585 0.6593515 -0.19350037 -0.21991524
-1.0960075 0.5869128 0.08025459 -0.9560589 -0.31730613 0.8023449
-0.60947496 -0.6820593 0.27526706 -1.1114197 -0.13663687 0.11540947
0.5257362 0.19630654 0.73711586 0.16966997 -0.5000035 -0.08896345
-0.13218226 -0.00969917 1.2245903 0.18892325 -0.35386166 0.616151
0.94230795 0.21554239 -0.59832036 -0.12295776 0.49042287 -0.32823196
-0.5143223 -1.3590071 -0.25147355 0.9236633 -0.2526196 0.46757898
0.9564488 -0.0508153 0.7923861 0.9661754 0.3104998 -0.813724
-0.05293908 0.7486671 0.97967654 -1.3572253 0.1055341 -0.44967657
-0.7712203 1.0573483 0.5103513 0.083413 0.10131015 0.6116791
-0.342232 -0.0315236 -1.2088811 -0.06543445 0.25851902 0.13241151
0.60108936 0.03548448 -0.22713289 1.1062655 -0.891829 -0.02460376
0.44309568 1.1379621 -0.22748913 -1.1765128 -0.29023066 1.0708386
-0.591545 -0.6471831 -0.6250888 0.81227547 0.04507696 0.75794125
-0.06754683 -0.2946567 0.5377213 0.42742762 0.8339699 -0.66191226
-0.9077977 -0.5625321 0.856622 -0.623674 0.20981292 -0.8513662
-1.2266711 -0.65126634 0.09988604 0.19229995 0.29762703 1.1241453
0.0295136 0.56372815 0.2804972 -0.05399362 -0.97847164 -1.192013
-0.17398728 0.6075717 0.19280098 -0.20078696 -0.22167298 0.1198437 ] | [9.721380233764648, 7.8270158767700195] |
660bf105-76a7-4ea0-8659-8d0ac16fc6f6 | part-aware-context-network-for-human-parsing | null | null | http://openaccess.thecvf.com/content_CVPR_2020/html/Zhang_Part-Aware_Context_Network_for_Human_Parsing_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Zhang_Part-Aware_Context_Network_for_Human_Parsing_CVPR_2020_paper.pdf | Part-Aware Context Network for Human Parsing | Recent works have made significant progress in human parsing by exploiting rich contexts. However, human parsing still faces a challenge of how to generate adaptive contextual features for the various sizes and shapes of human parts. In this work, we propose a Part-aware Context Network (PCNet), a novel and effective algorithm to deal with the challenge. PCNet mainly consists of three modules, including a part class module, a relational aggregation module, and a relational dispersion module. The part class module extracts the high-level representations of every human part from a categorical perspective. We design a relational aggregation module to capture the representative global context by mining associated semantics of human parts, which adaptively augments the context for human parts. We propose a relational dispersion module to generate the discriminative and effective local context and neglect disturbing one by making the affinity of human parts dispersed. The relational dispersion module ensures that features in the same class will be close to each other and away from those of different classes. By fusing the outputs of the relational aggregation module, the relational dispersion module and the backbone network, our PCNet generates adaptive contextual features for various sizes of human parts, improving the parsing accuracy. We achieve a new state-of-the-art segmentation performance on three challenging human parsing datasets, i.e., PASCAL-Person-Part, LIP, and CIHP.
| [' Ming Tang', ' Jinqiao Wang', ' Bingke Zhu', ' Yingying Chen', 'Xiaomei Zhang'] | 2020-06-01 | null | null | null | cvpr-2020-6 | ['human-parsing'] | ['computer-vision'] | [ 2.42506370e-01 2.22639427e-01 -3.56025510e-02 -6.83477640e-01
-5.40077746e-01 -4.14260179e-01 1.86940461e-01 1.75226793e-01
-4.04653490e-01 4.53202367e-01 3.50080699e-01 2.01631799e-01
1.29406974e-01 -1.00733149e+00 -5.44128835e-01 -6.58878982e-01
2.36636266e-01 5.69210649e-01 8.19646239e-01 -2.13295043e-01
-3.46040428e-02 2.74080127e-01 -1.59158695e+00 5.65011382e-01
9.63056505e-01 9.98204827e-01 3.52554560e-01 5.31483829e-01
-5.70412934e-01 4.42311317e-01 -6.81340337e-01 -4.74343449e-01
2.38502026e-01 -4.72403437e-01 -7.78950453e-01 -4.17553671e-02
3.69376481e-01 -4.19912674e-02 6.13028482e-02 1.03107560e+00
6.94256783e-01 1.46635160e-01 3.31548184e-01 -1.05186534e+00
-3.81379008e-01 9.17212307e-01 -5.64309061e-01 1.69470757e-01
2.94814587e-01 6.60638511e-03 1.22228396e+00 -5.42389333e-01
7.56364465e-01 1.49910867e+00 7.17756569e-01 8.90131235e-01
-9.20434296e-01 -6.46276116e-01 7.14227021e-01 6.58344775e-02
-9.26384687e-01 5.20450398e-02 9.10747290e-01 -1.54266432e-01
6.13937795e-01 2.11963922e-01 8.34471643e-01 1.16617155e+00
-2.79523768e-02 1.18568766e+00 7.33403981e-01 -2.05210730e-01
2.76246518e-01 -2.71804035e-01 5.88524759e-01 5.40287018e-01
2.29602486e-01 -2.03655675e-01 -3.82383883e-01 -1.82995275e-01
5.58558822e-01 4.14565243e-02 6.44751862e-02 -3.96642238e-01
-8.80871475e-01 8.54552090e-01 5.83427250e-01 2.42721558e-01
-3.91668409e-01 -5.54608786e-03 4.91887182e-01 -2.85960764e-01
1.58675373e-01 1.45307079e-01 -8.39602351e-01 1.36823207e-01
-6.35803640e-01 4.64934975e-01 8.58568668e-01 1.34545743e+00
8.24727416e-01 -7.31759906e-01 -5.10642290e-01 1.10687721e+00
2.93980241e-01 2.11366609e-01 4.75204736e-01 -1.02329147e+00
6.82335258e-01 1.11304629e+00 -2.81374991e-01 -1.00145662e+00
-6.84887826e-01 -3.96331966e-01 -7.38342583e-01 -3.97691160e-01
3.14666122e-01 -3.17168027e-01 -1.01480162e+00 2.03785896e+00
1.01680052e+00 1.19191892e-01 -1.74800605e-01 8.82975459e-01
1.01506674e+00 5.25539994e-01 6.19015396e-01 1.67737752e-01
1.79778779e+00 -1.20178425e+00 -4.88338143e-01 -5.56182981e-01
3.97479206e-01 -6.56279862e-01 1.02131402e+00 4.86000329e-02
-8.08767319e-01 -7.30962396e-01 -9.04445291e-01 -4.46490824e-01
-4.36634451e-01 2.45493963e-01 7.09255219e-01 2.94276804e-01
-5.77858031e-01 5.92018366e-01 -7.46458173e-01 -2.71780163e-01
6.09690428e-01 1.57094508e-01 -3.91294569e-01 -2.04297289e-01
-1.12252593e+00 2.94879228e-01 5.57727575e-01 1.16964310e-01
-2.68394232e-01 -5.21589577e-01 -1.14749622e+00 8.48905891e-02
6.52846515e-01 -9.52351987e-01 1.04632032e+00 -1.11692488e+00
-1.15355599e+00 7.97712862e-01 -1.56073183e-01 -2.26871789e-01
5.02891779e-01 -4.76976216e-01 -1.63008079e-01 2.42285654e-01
5.08521855e-01 9.46654379e-01 6.54804468e-01 -1.18254030e+00
-1.09397030e+00 -6.17375076e-01 -1.43296495e-01 1.87676311e-01
1.17898598e-01 -3.49227972e-02 -9.61772084e-01 -7.98172474e-01
3.61770242e-01 -7.12057710e-01 -3.88129592e-01 -1.80786535e-01
-6.30288780e-01 -4.68909413e-01 7.89155006e-01 -5.92670381e-01
1.24979722e+00 -2.25615001e+00 1.11322217e-01 1.30332187e-01
1.78861201e-01 4.01079386e-01 -1.49677396e-01 7.32484460e-02
1.38524547e-01 6.05068915e-02 -6.09061003e-01 -4.68252301e-01
-9.75599419e-03 5.76112211e-01 -2.05174573e-02 -7.15597570e-02
4.30743814e-01 8.50394726e-01 -1.08257067e+00 -9.03107822e-01
4.99860831e-02 3.57720673e-01 -7.59348571e-01 2.95186043e-01
-4.11335379e-01 2.73899734e-01 -8.12043965e-01 7.88416564e-01
8.35670352e-01 -1.69158857e-02 3.41968358e-01 -8.05299506e-02
1.53563693e-01 1.67299956e-01 -1.29596663e+00 1.68174934e+00
-2.82404525e-03 -1.02887601e-01 3.43259752e-01 -7.34651804e-01
8.84377301e-01 -1.06652461e-01 3.27834845e-01 -5.42317510e-01
1.98530123e-01 -1.52863916e-02 1.72655750e-02 -4.63707000e-01
4.99117017e-01 1.06758088e-01 -4.05624151e-01 1.84876546e-01
1.09655991e-01 2.57707506e-01 2.64624923e-01 2.39525422e-01
1.11726105e+00 3.39903533e-01 2.63729274e-01 -3.24379325e-01
5.91894805e-01 -1.82011053e-02 1.32033253e+00 6.10777020e-01
-6.66007578e-01 6.95711970e-01 7.57291019e-01 -5.88630676e-01
-7.26771891e-01 -1.03514862e+00 1.17103480e-01 1.51137912e+00
5.46872735e-01 -5.36166072e-01 -1.22265828e+00 -1.24661028e+00
3.62670384e-02 3.04957241e-01 -8.40038240e-01 -4.66634519e-02
-1.09541857e+00 -5.94908357e-01 4.01399553e-01 1.01964343e+00
9.07039881e-01 -1.49473011e+00 -6.85944974e-01 3.35088164e-01
-3.79373223e-01 -1.25467110e+00 -7.60031104e-01 2.82551259e-01
-6.67328358e-01 -1.28911591e+00 -3.27301532e-01 -1.04377246e+00
6.49032235e-01 -5.41131198e-02 1.29057717e+00 3.85792941e-01
-2.88653314e-01 -7.74101168e-02 -5.78100741e-01 -3.17476869e-01
-1.25127435e-01 2.82794833e-01 -4.83200133e-01 -7.35732391e-02
4.98805344e-01 -4.37303066e-01 -8.35361242e-01 6.13825202e-01
-7.15063691e-01 1.01857699e-01 6.43204331e-01 7.33595788e-01
1.15361476e+00 4.70269285e-02 4.54699397e-01 -1.40757179e+00
2.61045873e-01 -5.46588957e-01 -2.33866900e-01 2.74855286e-01
-5.91235310e-02 -4.60212678e-02 6.50486887e-01 -4.58952248e-01
-1.11545348e+00 3.85562152e-01 -3.94142568e-01 -5.08922823e-02
-4.86978710e-01 -3.14861573e-02 -9.93754268e-01 6.00170672e-01
3.97454590e-01 -8.60502943e-02 -4.42172498e-01 -5.47912717e-01
5.76578200e-01 4.90376621e-01 8.38211477e-01 -1.04536426e+00
5.03700435e-01 3.00501049e-01 -1.15451679e-01 -5.06794214e-01
-9.85501587e-01 -6.50012791e-01 -8.50983083e-01 6.29025996e-02
1.32880366e+00 -7.81114817e-01 -1.67317837e-01 4.86392528e-01
-9.87910926e-01 -3.43092918e-01 -5.09614170e-01 -9.36709419e-02
-2.99380630e-01 4.08492476e-01 -6.94046915e-01 -3.85517836e-01
-5.32260895e-01 -1.12484443e+00 1.40353322e+00 7.37922370e-01
-1.07494786e-01 -5.45450866e-01 -8.92433077e-02 4.90139127e-01
-1.04263835e-01 4.63337272e-01 9.96616244e-01 -9.02070105e-01
-4.50415581e-01 -4.10593897e-02 -2.77849138e-01 3.48143727e-01
7.97335729e-02 3.64473276e-02 -8.93709421e-01 7.64347538e-02
-2.75673062e-01 5.73884062e-02 9.92794037e-01 3.46536726e-01
1.28538501e+00 -1.84975281e-01 -6.26379669e-01 6.36346400e-01
9.66640830e-01 7.31956288e-02 4.92795289e-01 9.57572460e-02
9.15986598e-01 8.20901155e-01 8.62650335e-01 4.60594036e-02
6.32138729e-01 4.34721529e-01 2.89188653e-01 -1.05191432e-02
-2.40236998e-01 -4.96582747e-01 2.58381907e-02 4.42128807e-01
1.85132936e-01 -6.47903830e-02 -6.70041978e-01 5.96589208e-01
-2.09430242e+00 -8.55643809e-01 3.10948752e-02 1.72966814e+00
8.23341250e-01 3.46485794e-01 3.48585814e-01 -1.74865469e-01
1.17485321e+00 4.13843878e-02 -5.94678164e-01 -3.56672317e-01
1.22193089e-02 2.08242893e-01 2.72501744e-02 1.50358379e-01
-1.46777809e+00 1.36421347e+00 5.19667864e+00 7.90148854e-01
-5.63000262e-01 3.95721979e-02 7.26957858e-01 1.84216410e-01
-1.24577813e-01 -1.53966825e-02 -1.11432922e+00 6.44846022e-01
2.96754003e-01 5.93004882e-01 -6.21226542e-02 1.27713478e+00
-2.87788033e-01 -2.43555643e-02 -9.76162076e-01 6.26759768e-01
-1.51098937e-01 -8.74068379e-01 1.30191356e-01 -1.03375420e-01
4.92710143e-01 -1.26755267e-01 -4.31998581e-01 5.09117305e-01
3.26643795e-01 -7.05964804e-01 7.79230118e-01 3.31827343e-01
2.55601853e-01 -9.54988599e-01 7.82572448e-01 4.49774683e-01
-1.66219592e+00 -2.41591349e-01 -3.86680961e-01 3.00009131e-01
5.32461852e-02 7.46746719e-01 -3.17290574e-01 6.66327000e-01
9.35949981e-01 4.34698522e-01 -6.19952261e-01 8.04762304e-01
-5.54584861e-01 4.34806019e-01 -4.06338781e-01 8.45071152e-02
2.54944414e-01 -9.39526856e-02 4.61916268e-01 1.51726627e+00
-1.58708870e-01 3.01173449e-01 6.34811699e-01 8.10816169e-01
-1.66515619e-01 1.82130501e-01 -1.99314788e-01 3.37062061e-01
7.67522037e-01 1.63207471e+00 -1.04883242e+00 -4.44044530e-01
-2.73316830e-01 9.59825873e-01 5.96881211e-01 -4.78987843e-02
-7.49895930e-01 -5.14645040e-01 7.70817757e-01 -3.69589068e-02
5.48575819e-01 3.43220532e-02 -4.26457107e-01 -9.39109147e-01
2.85849839e-01 -8.42620730e-01 1.01194513e+00 -3.81444842e-01
-1.48488927e+00 6.90715313e-01 4.26937714e-02 -8.11324477e-01
7.53186718e-02 -6.36304498e-01 -9.01112735e-01 7.59548783e-01
-1.18117869e+00 -1.43440664e+00 -4.72525805e-01 6.24605417e-01
6.66489959e-01 9.39844474e-02 4.54022020e-01 1.38888195e-01
-8.35555553e-01 6.55167341e-01 -7.78473675e-01 5.53819776e-01
5.28191090e-01 -1.59140313e+00 5.67486286e-01 9.13711607e-01
-1.06525965e-01 8.27292442e-01 2.80289829e-01 -1.01035988e+00
-8.13727975e-01 -1.34078765e+00 8.49659204e-01 -3.31317246e-01
1.23487704e-01 -5.89612782e-01 -9.08084810e-01 6.58416629e-01
-2.06013158e-01 3.15336019e-01 5.65859973e-01 2.32370064e-01
-4.56039280e-01 -5.37025668e-02 -1.39935124e+00 6.13587916e-01
1.51687598e+00 -2.19493672e-01 -9.58328187e-01 -1.24205882e-02
1.06104469e+00 -4.47072148e-01 -6.50503039e-01 5.25282323e-01
4.53346997e-01 -1.04348123e+00 1.05454648e+00 -4.30406451e-01
4.71141368e-01 -3.78528535e-01 -2.66115993e-01 -1.01011086e+00
-4.37214255e-01 -4.17502254e-01 -1.43863440e-01 1.66616094e+00
2.78779477e-01 -4.78982836e-01 9.12859797e-01 5.75960159e-01
-3.68708462e-01 -1.06900525e+00 -9.34640884e-01 -4.45202768e-01
1.80714563e-01 -3.91874433e-01 9.56785560e-01 6.67959273e-01
-3.85838002e-01 4.41478282e-01 1.68669611e-01 2.29407787e-01
4.65119839e-01 3.23353916e-01 7.19650447e-01 -1.21837795e+00
-3.86640221e-01 -4.78281200e-01 -3.18851352e-01 -1.09198999e+00
1.33269876e-01 -7.50136375e-01 3.29892725e-01 -1.46087921e+00
3.98360312e-01 -6.94178343e-01 -2.26328328e-01 6.64307892e-01
-8.41681302e-01 -2.03194663e-01 2.31705114e-01 4.99607548e-02
-6.76715851e-01 2.47760117e-01 1.42163897e+00 1.93966981e-02
-3.86283278e-01 1.14416838e-01 -1.00411677e+00 1.11734688e+00
7.98268259e-01 -4.59436953e-01 -3.09643209e-01 -4.25549418e-01
-2.33155340e-02 -3.81887972e-01 3.19605768e-01 -1.09118557e+00
1.26806155e-01 -3.57962251e-02 7.47402191e-01 -8.62341106e-01
6.63399100e-02 -7.28435814e-01 -2.06965268e-01 2.27629229e-01
-1.72289312e-01 1.03244118e-01 -5.01634320e-03 6.74909532e-01
-5.03905490e-02 -1.78857654e-01 8.88538718e-01 -4.66459721e-01
-8.93415511e-01 4.77242559e-01 2.70499527e-01 4.12704915e-01
9.14762080e-01 -2.33826209e-02 -3.56400341e-01 3.50489676e-01
-8.31118166e-01 6.97624803e-01 2.78969944e-01 4.78663296e-01
4.57737297e-01 -1.15138531e+00 -4.59116429e-01 3.15244526e-01
6.01027347e-02 6.92313492e-01 4.27832246e-01 3.60140175e-01
-3.92591208e-01 -1.96936026e-01 -4.88700457e-02 -6.88811243e-01
-1.28469670e+00 5.51063538e-01 3.27984214e-01 -5.03070474e-01
-9.67767179e-01 1.07141578e+00 4.62496132e-01 -8.51618946e-01
1.58534303e-01 -5.78201413e-01 -4.44288313e-01 1.21293500e-01
5.72885573e-01 3.01126599e-01 -2.60712624e-01 -8.04400742e-01
-5.27479768e-01 7.89184034e-01 -1.58099324e-01 2.35861033e-01
1.08119440e+00 3.74450833e-02 -1.72866136e-01 -1.64850731e-03
9.22482371e-01 7.64845535e-02 -1.41137981e+00 -1.73523128e-01
1.31954014e-01 -1.76150709e-01 -4.53946501e-01 -1.05252516e+00
-1.23561454e+00 7.52694011e-01 3.90294939e-01 9.22622532e-02
1.25191271e+00 3.63368094e-01 1.26279712e+00 8.49109143e-02
4.46323991e-01 -1.46705484e+00 -1.45583376e-01 6.21821344e-01
5.51219940e-01 -8.99552703e-01 -1.27749428e-01 -8.69290888e-01
-8.31695795e-01 8.75775337e-01 1.01474833e+00 -1.87825710e-01
5.70660532e-01 2.15771481e-01 1.17728725e-01 -1.80245042e-01
-4.28580105e-01 -3.68146569e-01 3.27788830e-01 6.56129837e-01
2.10276961e-01 3.29360247e-01 -4.89763379e-01 1.37305295e+00
-2.96775520e-01 -4.68156189e-01 -1.54720798e-01 1.12650466e+00
-6.11020684e-01 -1.36180103e+00 -2.45684698e-01 2.29954347e-01
-5.50466716e-01 8.05598646e-02 -6.48379087e-01 8.70083153e-01
8.79427969e-01 9.11297381e-01 1.19248286e-01 -4.27001715e-01
5.91751754e-01 2.90647626e-01 2.26091787e-01 -7.93735743e-01
-9.18826699e-01 7.90444016e-02 1.58228710e-01 -8.83691370e-01
-2.18786493e-01 -7.52720118e-01 -1.79637885e+00 9.92641151e-02
-2.46043608e-01 2.09411867e-02 3.27827483e-01 1.02276266e+00
4.83271956e-01 6.60307884e-01 3.53037417e-01 -6.55415058e-01
-2.37735227e-01 -8.79891992e-01 -4.44253296e-01 7.24112391e-01
-2.02516583e-03 -6.30449235e-01 1.57065205e-02 -2.54867375e-01] | [8.74142074584961, 0.06326030939817429] |
986bc358-c826-4c6c-a48e-ba6613d0097d | synwmd-syntax-aware-word-mover-s-distance-for | 2206.10029 | null | https://arxiv.org/abs/2206.10029v1 | https://arxiv.org/pdf/2206.10029v1.pdf | SynWMD: Syntax-aware Word Mover's Distance for Sentence Similarity Evaluation | Word Mover's Distance (WMD) computes the distance between words and models text similarity with the moving cost between words in two text sequences. Yet, it does not offer good performance in sentence similarity evaluation since it does not incorporate word importance and fails to take inherent contextual and structural information in a sentence into account. An improved WMD method using the syntactic parse tree, called Syntax-aware Word Mover's Distance (SynWMD), is proposed to address these two shortcomings in this work. First, a weighted graph is built upon the word co-occurrence statistics extracted from the syntactic parse trees of sentences. The importance of each word is inferred from graph connectivities. Second, the local syntactic parsing structure of words is considered in computing the distance between words. To demonstrate the effectiveness of the proposed SynWMD, we conduct experiments on 6 textual semantic similarity (STS) datasets and 4 sentence classification datasets. Experimental results show that SynWMD achieves state-of-the-art performance on STS tasks. It also outperforms other WMD-based methods on sentence classification tasks. | ['C. -C. Jay Kuo', 'Bin Wang', 'Chengwei Wei'] | 2022-06-20 | null | null | null | null | ['sentence-classification'] | ['natural-language-processing'] | [ 3.44912350e-01 -1.42567858e-01 -1.01489238e-01 -5.76287270e-01
-4.84250993e-01 -2.85057902e-01 4.56643850e-01 9.58019555e-01
-7.44541645e-01 2.39023492e-01 6.91858232e-01 -4.93353575e-01
-3.43005210e-01 -8.30512822e-01 1.27555002e-02 -4.46248740e-01
-2.91745905e-02 1.31403685e-01 5.71526110e-01 -5.22582829e-01
9.38721895e-01 9.82725341e-03 -1.09184730e+00 3.40138614e-01
1.15031922e+00 7.08730817e-01 7.39584804e-01 4.16258007e-01
-8.71017635e-01 6.17341220e-01 -5.75791478e-01 -4.53993917e-01
-6.89322874e-02 -7.23393798e-01 -1.17391622e+00 -3.20201397e-01
2.54879653e-01 2.21368372e-01 -2.76496798e-01 1.24455905e+00
6.11595094e-01 5.59223235e-01 6.65420473e-01 -8.94470572e-01
-8.47264886e-01 8.64165962e-01 -6.84334934e-01 9.46027815e-01
7.60358036e-01 -3.88817549e-01 1.55690551e+00 -1.04895997e+00
4.78906214e-01 1.55619514e+00 8.45076621e-01 2.18709365e-01
-7.29437649e-01 -9.93641615e-02 2.55001456e-01 5.16026378e-01
-1.16478372e+00 1.53846100e-01 8.32264066e-01 -1.92207858e-01
1.63953626e+00 2.80359864e-01 3.01150024e-01 5.30289829e-01
4.08866316e-01 7.44549215e-01 6.59875929e-01 -8.01950037e-01
2.39425868e-01 -4.42649096e-01 8.64612639e-01 6.85182393e-01
2.12370366e-01 -6.38467252e-01 -6.36238933e-01 -3.31509441e-01
-1.14345491e-01 -4.52916473e-02 -1.64577663e-01 -8.67739916e-02
-1.08202362e+00 8.99000645e-01 3.46856028e-01 8.61075997e-01
-5.79065159e-02 -1.63662493e-01 7.82637119e-01 4.65154797e-01
7.59080708e-01 3.33207667e-01 -1.77639708e-01 -2.13176548e-01
-6.41124487e-01 5.02942055e-02 3.87820005e-01 8.24453771e-01
5.16334474e-01 -2.92713702e-01 -3.23414028e-01 9.63529348e-01
3.51800084e-01 1.99379414e-01 1.01432908e+00 -4.32376146e-01
9.98618424e-01 9.90596652e-01 -5.59807777e-01 -1.60712111e+00
-4.23226684e-01 -5.25133461e-02 -6.35800242e-01 -4.97998923e-01
2.19160831e-03 2.62480438e-01 -5.58411479e-01 1.60801482e+00
3.84292245e-01 2.42639810e-01 1.89313039e-01 7.74518073e-01
1.33781540e+00 5.59724808e-01 1.58082470e-01 -2.70075768e-01
1.39913869e+00 -1.02161288e+00 -8.01456451e-01 -3.45377803e-01
1.23368120e+00 -8.60038400e-01 1.19344664e+00 -4.34902042e-01
-1.06290722e+00 -6.36959434e-01 -1.10188663e+00 -1.27010912e-01
-5.34484923e-01 -6.05277658e-01 2.51810610e-01 4.49550718e-01
-9.96870518e-01 7.81032205e-01 -5.26638269e-01 -7.19340026e-01
-5.12189567e-02 -9.25795734e-03 -1.55592099e-01 -3.76867652e-02
-1.70765102e+00 1.21385241e+00 7.13083506e-01 -2.53125370e-01
2.35917673e-01 -5.89919448e-01 -1.21274507e+00 1.22977160e-01
1.49644405e-01 -6.69045031e-01 9.92906451e-01 -6.49344444e-01
-1.00714481e+00 9.25910473e-01 -6.27219796e-01 -3.23252946e-01
-8.13242327e-03 1.35363042e-01 -5.09633541e-01 3.20010573e-01
2.07428381e-01 1.87658906e-01 4.44863290e-01 -5.34625351e-01
-4.57759112e-01 -3.95161688e-01 -1.46751814e-02 5.41452348e-01
-7.15119362e-01 2.77471483e-01 -1.91069141e-01 -9.84869778e-01
2.95622110e-01 -6.52923942e-01 -1.01869814e-01 -2.43306190e-01
-1.50861368e-01 -7.42786050e-01 7.69339740e-01 -6.66505992e-01
1.84466803e+00 -2.10739684e+00 2.24943444e-01 2.35980049e-01
2.91129977e-01 3.93459082e-01 -5.03272355e-01 8.68059278e-01
-4.73390296e-02 1.60940364e-01 -5.99702597e-01 -2.65378714e-01
6.37900010e-02 1.31415710e-01 -4.67507727e-02 1.77635893e-01
-4.32109199e-02 9.63596940e-01 -1.28760719e+00 -8.54644239e-01
1.02831595e-01 1.85088485e-01 -1.74594879e-01 9.98122152e-03
3.03097695e-01 -2.95416921e-01 -6.41724586e-01 8.63084495e-02
6.58465505e-01 -1.40923485e-01 3.98683578e-01 -1.51377067e-01
2.86753327e-01 7.53162622e-01 -9.37403619e-01 2.02310562e+00
-3.37782443e-01 4.44299459e-01 -6.86758459e-01 -1.22222626e+00
1.07614374e+00 6.10835925e-02 3.11790973e-01 -9.42685902e-01
6.83407485e-02 1.39111444e-01 1.24919996e-01 -7.31144369e-01
7.98708856e-01 -4.82735932e-02 -2.19665617e-01 5.84971130e-01
-6.55112341e-02 -1.37342632e-01 4.05563831e-01 6.67914331e-01
1.37574482e+00 -4.38560545e-01 7.86715329e-01 -7.38749802e-01
9.53171313e-01 -3.17204565e-01 4.49495882e-01 4.58989233e-01
-3.03575486e-01 3.75091493e-01 3.18910152e-01 -3.09933007e-01
-8.03768098e-01 -1.16725206e+00 8.33582431e-02 1.08506286e+00
5.85220456e-01 -9.36914802e-01 -8.95117223e-01 -8.43952000e-01
-2.87476107e-02 8.74488533e-01 -5.43539464e-01 -5.36254346e-01
-6.12139821e-01 -7.17671931e-01 4.73812461e-01 6.50043070e-01
5.13312995e-01 -8.38387549e-01 -3.55406433e-01 1.72275767e-01
-6.06721759e-01 -1.07930267e+00 -1.09351337e+00 -3.23544681e-01
-1.02443397e+00 -1.01562011e+00 -3.74964595e-01 -1.13052654e+00
6.52283251e-01 8.74099553e-01 1.16180360e+00 3.16615343e-01
-3.30734015e-01 2.84870923e-01 -8.35883498e-01 -8.06881264e-02
-3.66551578e-01 1.57436788e-01 6.71340078e-02 -3.35708678e-01
7.96781540e-01 -3.38944316e-01 -4.28019136e-01 2.52187699e-01
-8.65890563e-01 -1.78309485e-01 8.15331787e-02 7.67778218e-01
2.48414189e-01 7.85486549e-02 6.33836031e-01 -5.29933453e-01
1.29325986e+00 -4.89239693e-01 1.35078728e-01 5.68686664e-01
-7.30981886e-01 3.72270405e-01 4.58142102e-01 -1.98524028e-01
-9.69883442e-01 -5.97914875e-01 -1.11106217e-01 3.44394982e-01
2.70024836e-01 8.46654952e-01 1.32169038e-01 1.72461078e-01
3.77170473e-01 3.14466923e-01 -1.55249268e-01 -3.92493367e-01
2.26498157e-01 8.93575788e-01 2.93392360e-01 -4.50929970e-01
5.03892839e-01 1.17478326e-01 5.02332039e-02 -7.68046796e-01
-9.08823431e-01 -7.61451125e-01 -7.66363144e-01 -8.95456597e-02
9.88223195e-01 -4.56392735e-01 -4.85467166e-01 1.58240810e-01
-1.30501997e+00 3.61033678e-01 3.30437138e-03 6.29490376e-01
-8.49257782e-02 1.20262289e+00 -5.80367446e-01 -5.98709881e-01
-8.32574308e-01 -7.17468977e-01 8.93464625e-01 2.81098392e-02
-6.93071306e-01 -1.55073524e+00 1.32691681e-01 4.27087903e-01
3.28329712e-01 1.22126192e-02 1.35673118e+00 -9.26788747e-01
3.23848873e-01 -2.00968817e-01 -1.21905945e-01 1.86453402e-01
4.40688342e-01 -1.43693998e-01 -3.78469467e-01 -2.64503747e-01
1.64234698e-01 7.38338605e-02 8.48933756e-01 1.71771646e-01
9.58595395e-01 -1.54708490e-01 -2.73724049e-01 -1.00880980e-01
1.21074772e+00 9.86250564e-02 3.88136417e-01 4.16225046e-01
6.16866469e-01 7.58804381e-01 8.37851942e-01 3.24788839e-01
5.48664093e-01 7.16813982e-01 2.35455394e-01 2.99305648e-01
-1.08883463e-01 -2.29663864e-01 3.87166917e-01 1.76867044e+00
1.28327116e-01 -4.85381603e-01 -1.01941502e+00 3.95836800e-01
-2.07451940e+00 -1.01320755e+00 -7.40089715e-01 1.96799350e+00
6.68538809e-01 4.47169036e-01 -1.06165901e-01 3.60900044e-01
1.13800049e+00 5.48269689e-01 -2.31056392e-01 -8.13116133e-01
-1.65735096e-01 2.32480958e-01 9.26308632e-02 7.56571233e-01
-6.65116489e-01 9.22233403e-01 5.62197971e+00 1.13112521e+00
-4.88329500e-01 1.64600834e-01 7.65115470e-02 2.22414941e-01
-5.19699991e-01 -6.91730380e-02 -4.36580986e-01 6.48231864e-01
7.38825619e-01 -5.91693819e-01 -8.45531449e-02 5.77811003e-01
1.93818927e-01 -1.51850432e-01 -8.08860600e-01 8.49236965e-01
4.41769868e-01 -1.09374797e+00 1.57115132e-01 -4.82167751e-01
4.91529435e-01 -1.79594114e-01 -2.00130060e-01 -3.55234183e-03
7.12287426e-02 -6.54086709e-01 4.17752922e-01 3.66476566e-01
9.08893719e-02 -8.61606121e-01 1.06142902e+00 3.26153725e-01
-1.76336443e+00 2.44821459e-01 -5.91750264e-01 -2.96656489e-01
3.18793803e-01 8.38092089e-01 -6.71491683e-01 7.78599858e-01
5.00400245e-01 8.87063026e-01 -6.98327541e-01 4.99062866e-01
-3.14921260e-01 4.58221406e-01 1.26025245e-01 -6.18175268e-01
5.19811511e-01 -1.57462075e-01 5.87980390e-01 1.52734053e+00
3.49232048e-01 7.86473826e-02 7.41607696e-02 4.21427041e-01
5.95633462e-02 5.44330060e-01 -5.80386341e-01 2.19910309e-01
7.61550128e-01 1.12096655e+00 -8.32676828e-01 -3.69796783e-01
-4.43135947e-01 1.24628663e+00 4.55679476e-01 -1.07505672e-01
-6.90902293e-01 -9.52574313e-01 7.75828362e-01 -1.69948652e-01
1.58322364e-01 -4.88290161e-01 -2.31852546e-01 -8.93107653e-01
2.43827716e-01 -3.31132859e-01 7.48951197e-01 -7.57179499e-01
-1.63562453e+00 6.66702569e-01 8.61713942e-03 -1.12731004e+00
-1.01231202e-01 -4.36287791e-01 -9.25697207e-01 8.93720090e-01
-1.44655263e+00 -5.65002322e-01 -3.40421379e-01 3.70416999e-01
9.43764091e-01 -1.88349485e-01 6.85069799e-01 1.06338136e-01
-4.61520851e-01 7.61431813e-01 2.42714118e-02 9.55182537e-02
5.43664932e-01 -1.18421519e+00 1.11015773e+00 8.20701480e-01
8.59960690e-02 6.39323652e-01 6.47351682e-01 -7.65349805e-01
-1.19356704e+00 -9.52026367e-01 1.76608074e+00 -3.10916811e-01
8.88003588e-01 -1.09696157e-01 -1.22080696e+00 1.13223800e-02
3.82335603e-01 -2.37841442e-01 9.91151035e-01 -2.67107964e-01
-4.54346776e-01 1.84939772e-01 -9.65808988e-01 6.46750808e-01
1.60369182e+00 -6.46904171e-01 -1.48005366e+00 3.60330075e-01
1.15969431e+00 -3.88965942e-02 -8.43904793e-01 3.45134556e-01
3.94787267e-02 -8.00256312e-01 9.74980533e-01 -9.05228198e-01
5.13993740e-01 -1.65666491e-01 -3.65392417e-01 -1.42235482e+00
-4.95361179e-01 -2.93313980e-01 -6.08852543e-02 1.42744243e+00
4.59511101e-01 -7.01758742e-01 3.36762935e-01 4.38408494e-01
-3.47231001e-01 -8.35329652e-01 -9.61812496e-01 -1.20944333e+00
9.99434218e-02 -3.95827413e-01 6.95748210e-01 1.34825993e+00
7.84562647e-01 6.19053602e-01 7.58379772e-02 -2.12651893e-01
3.82537365e-01 -3.80102620e-02 2.94524655e-02 -1.03760862e+00
-6.52769431e-02 -6.09890223e-01 -7.26085901e-01 -8.69316936e-01
6.05910003e-01 -1.41327870e+00 -1.00844532e-01 -1.87596190e+00
4.02886093e-01 1.73406675e-01 -4.50886667e-01 1.70507237e-01
-9.00000274e-01 -2.58691221e-01 1.72040790e-01 1.55553445e-01
-7.66834438e-01 7.46746540e-01 1.04507351e+00 -2.60335028e-01
6.36437982e-02 -4.43184733e-01 -5.12421191e-01 7.52042890e-01
1.01411247e+00 -5.71512163e-01 -6.93799198e-01 -6.37303054e-01
2.09101021e-01 -2.37989709e-01 -1.20461270e-01 -6.08126819e-01
4.39890206e-01 -2.24573016e-01 -2.92387187e-01 -4.98626053e-01
-4.73463312e-02 -5.71056843e-01 -4.17076558e-01 8.11837077e-01
-7.55537093e-01 8.09165895e-01 -2.98081517e-01 6.79957211e-01
-3.01881641e-01 -7.22531974e-01 5.17836452e-01 -8.63051489e-02
-9.20498908e-01 -4.16152999e-02 -3.86075884e-01 5.28951526e-01
9.71940041e-01 -4.94766235e-01 -3.87124449e-01 -3.39814126e-02
-3.87438536e-02 2.77910978e-01 1.59136817e-01 8.15381289e-01
9.34126854e-01 -1.67770422e+00 -6.75990701e-01 -3.27398807e-01
4.30177361e-01 -4.30699348e-01 3.23034376e-01 6.86233997e-01
-3.91523361e-01 3.56305838e-01 1.71833649e-01 -3.02447706e-01
-1.85971630e+00 7.30924785e-01 -3.64493988e-02 -3.71458441e-01
-6.61983371e-01 7.58479238e-01 -1.19882114e-01 -4.79807079e-01
2.95310970e-02 -4.27201837e-01 -3.71389180e-01 1.26484439e-01
6.16668165e-01 5.86831093e-01 2.03985140e-01 -7.21790314e-01
-6.88419819e-01 8.70210052e-01 -1.56921312e-01 7.61787370e-02
1.06791484e+00 -3.68571430e-01 -4.25394326e-01 4.47936445e-01
1.61534953e+00 -2.50881821e-01 -2.69283921e-01 -6.84575021e-01
6.71823621e-01 -6.99823141e-01 -8.40214361e-03 -1.96103036e-01
-7.28257418e-01 9.64958370e-01 1.78519592e-01 3.55839938e-01
9.33458269e-01 -5.76198567e-03 1.30787718e+00 6.09888554e-01
4.70668897e-02 -1.48354828e+00 5.57809711e-01 8.07162821e-01
5.72161317e-01 -1.00834107e+00 -1.54774198e-02 -6.25129342e-01
-7.30863214e-01 1.28596461e+00 5.94506443e-01 2.00323626e-01
5.29134452e-01 1.14964577e-03 -2.42568851e-01 -1.84854910e-01
-4.94063765e-01 -3.08566481e-01 6.57265902e-01 3.51385951e-01
6.75022483e-01 -2.01013044e-01 -1.11868024e+00 3.59419048e-01
-1.23461761e-01 -6.51769876e-01 1.62422314e-01 1.24558342e+00
-9.77056980e-01 -1.11706364e+00 7.32209487e-03 4.29152697e-01
-2.22534284e-01 -5.24196267e-01 -6.80801213e-01 2.62534499e-01
-2.12945744e-01 1.15824747e+00 1.25960723e-01 -6.70932055e-01
3.33033562e-01 6.78985864e-02 3.65215480e-01 -6.67903185e-01
-6.06859744e-01 -6.81161523e-01 1.62868589e-01 -3.29558015e-01
-4.85351413e-01 -5.18164277e-01 -1.74531603e+00 -6.06770992e-01
-3.46053720e-01 4.16808128e-01 5.56537092e-01 1.22738051e+00
4.45517272e-01 5.78359425e-01 7.66897082e-01 -2.40157709e-01
-6.51357949e-01 -9.19099569e-01 -5.67930341e-01 7.52320826e-01
-2.47460112e-01 -4.34467852e-01 -3.94740433e-01 -2.84669548e-01] | [11.009225845336914, 8.788773536682129] |
8dd28013-6b10-4909-9da9-6a1373e70933 | star-a-structure-aware-lightweight | null | null | http://openaccess.thecvf.com//content/ICCV2021/html/Zhang_STAR_A_Structure-Aware_Lightweight_Transformer_for_Real-Time_Image_Enhancement_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Zhang_STAR_A_Structure-Aware_Lightweight_Transformer_for_Real-Time_Image_Enhancement_ICCV_2021_paper.pdf | STAR: A Structure-Aware Lightweight Transformer for Real-Time Image Enhancement | Image and video enhancement such as color constancy, low light enhancement, and tone mapping on smartphones is challenging because high-quality images should be achieved efficiently with a limited resource budget. Unlike prior works that either used very deep CNNs or large Transformer models, we propose a \underline s eman\underline t ic-\underline a wa\underline r e lightweight Transformer, termed STAR, for real-time image enhancement. STAR is formulated to capture long-range dependencies between image patches, which naturally and implicitly captures the semantic relationships of different regions in an image. STAR is a general architecture that can be easily adapted to different image enhancement tasks. Extensive experiments show that STAR can effectively boost the quality and efficiency of many tasks such as illumination enhancement, auto white balance, and photo retouching, which are indispensable components for image processing on smartphones. For example, STAR reduces model complexity and improves image quality compared to the recent state-of-the-art [??] on the MIT-Adobe FiveK dataset [??] (i.e., 1.8dB PSNR improvements with 25% parameters and 13% float operations.) | ['Jinwei Gu', 'Ping Luo', 'Xiaogang Wang', 'Jun Jiang', 'Yitong Jiang', 'Zhaoyang Zhang'] | 2021-01-01 | null | null | null | iccv-2021-1 | ['photo-retouching', 'color-constancy', 'video-enhancement', 'tone-mapping'] | ['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision'] | [ 4.37818915e-01 -3.83948714e-01 1.55517295e-01 -2.84703881e-01
-4.11450177e-01 -1.23868302e-01 7.16764182e-02 -2.89553136e-01
-3.96709710e-01 5.11543751e-01 -9.60999951e-02 -2.80901104e-01
-7.74239898e-02 -6.68041646e-01 -9.61518764e-01 -7.03701794e-01
1.44108951e-01 -6.98971808e-01 2.81176955e-01 -6.10033751e-01
8.51453394e-02 5.36653511e-02 -1.70650196e+00 2.68691450e-01
1.12193418e+00 1.54956949e+00 5.07850885e-01 7.76792765e-01
1.86082497e-01 8.52997184e-01 -3.65448594e-01 -7.69370317e-01
2.72693515e-01 -2.43265495e-01 -3.07638735e-01 4.23319310e-01
8.05805683e-01 -5.93615115e-01 -5.45716286e-01 1.30430543e+00
7.17916310e-01 6.81870580e-02 9.59427953e-02 -1.20444679e+00
-1.04549026e+00 2.55529583e-01 -8.69306386e-01 3.59503061e-01
-2.18676835e-01 2.52348036e-01 6.73847973e-01 -8.98074508e-01
2.34880492e-01 9.43413138e-01 8.18627656e-01 3.79232943e-01
-1.00513041e+00 -8.12656403e-01 3.28904651e-02 5.23048997e-01
-1.26896536e+00 -6.54795945e-01 6.04849696e-01 1.62428841e-01
8.31282556e-01 3.91111344e-01 6.21665895e-01 7.27479815e-01
3.19545329e-01 7.97413230e-01 1.35465050e+00 -3.54723632e-01
-2.16712087e-01 5.57318665e-02 -2.14620009e-01 7.10973084e-01
3.43256779e-02 1.98309973e-01 -7.67845929e-01 4.52715218e-01
1.08022344e+00 -9.64201638e-04 -5.07058740e-01 8.48710760e-02
-1.18933785e+00 2.52565384e-01 6.25529528e-01 2.72983350e-02
-2.38545850e-01 3.80340755e-01 2.45170876e-01 2.97653407e-01
5.11366785e-01 1.96718991e-01 -4.51764226e-01 -1.81860000e-01
-7.45407403e-01 -1.01433448e-01 2.04694018e-01 1.06788778e+00
7.55018830e-01 3.27598095e-01 -1.26265332e-01 1.14220226e+00
-4.34097312e-02 8.00200105e-01 2.15036780e-01 -1.11200476e+00
5.52955210e-01 4.05776799e-01 2.29948580e-01 -1.05037165e+00
-1.48589492e-01 -6.22220576e-01 -1.22522664e+00 3.06151778e-01
7.15906471e-02 -2.06721008e-01 -7.23283708e-01 1.66551781e+00
2.07553148e-01 4.67778444e-01 -2.89553404e-02 9.01229560e-01
7.92532682e-01 5.69846451e-01 1.03297848e-02 -1.57623455e-01
1.50862181e+00 -1.31071055e+00 -1.08657706e+00 -3.15015912e-01
3.49608138e-02 -1.02250648e+00 1.33010685e+00 5.46516180e-01
-1.54564583e+00 -1.08107090e+00 -1.23958850e+00 -4.92801487e-01
-2.68688381e-01 3.16718906e-01 7.72973061e-01 9.34740663e-01
-1.62106812e+00 5.87462842e-01 -5.82932055e-01 5.30709624e-02
5.02618551e-01 4.31621522e-01 -1.88884482e-01 -3.40860695e-01
-1.24752128e+00 6.06664658e-01 6.56033456e-02 4.63159114e-01
-6.23476148e-01 -8.26340377e-01 -9.44962859e-01 1.41608715e-01
5.55798471e-01 -6.86163247e-01 1.13310647e+00 -1.23701060e+00
-1.69824731e+00 6.10791326e-01 -1.84096381e-01 -3.79171222e-01
1.80345669e-01 -4.06671166e-01 -7.69485295e-01 1.28810495e-01
-3.39368135e-01 7.64816105e-01 1.18810010e+00 -1.06575465e+00
-8.22868049e-01 -2.73957312e-01 2.23018214e-01 4.64115977e-01
-8.01050901e-01 2.77983904e-01 -1.00763893e+00 -8.13442826e-01
-1.84660554e-01 -7.30102837e-01 -2.05854788e-01 4.66613412e-01
-6.71093166e-02 4.52134341e-01 1.03724098e+00 -9.77430046e-01
1.20060670e+00 -2.31902289e+00 -1.89048782e-01 -1.86909214e-01
4.23523903e-01 5.06893218e-01 -3.85327637e-01 -3.52975875e-01
-9.42124724e-02 -1.03839546e-01 -1.78455204e-01 -2.95252025e-01
4.63667661e-02 3.34894583e-02 -2.10016862e-01 3.58898789e-01
1.24300897e-01 1.17376912e+00 -6.28823340e-01 -4.21196014e-01
5.54208934e-01 1.02374065e+00 -3.57844263e-01 1.44531712e-01
8.02762583e-02 4.48429883e-02 1.91625401e-01 6.26739085e-01
1.03217864e+00 -4.47226673e-01 -1.31758023e-02 -5.79438448e-01
-1.10474281e-01 -2.35393152e-01 -1.11108553e+00 1.49591541e+00
-8.36731613e-01 1.04790866e+00 3.47894967e-01 -5.82459271e-01
8.30807447e-01 2.18732730e-01 3.18212330e-01 -1.41002703e+00
2.02130944e-01 1.47519067e-01 -2.69595772e-01 -4.06280011e-01
9.96239960e-01 4.41975862e-01 3.70779872e-01 2.37848476e-01
-2.68799454e-01 -9.18804377e-04 1.25674397e-01 5.54736853e-02
5.88853657e-01 9.84835476e-02 9.91495792e-03 -1.00585602e-01
4.31131899e-01 -5.80631554e-01 3.81697267e-01 6.81400836e-01
-2.16252983e-01 5.77088594e-01 8.96055773e-02 -2.85554528e-01
-1.18928540e+00 -1.03614664e+00 -6.63125739e-02 1.25936651e+00
7.14964569e-01 -3.26910377e-01 -8.19305956e-01 -1.42155513e-01
-6.56324446e-01 2.30575204e-01 -5.34793973e-01 -5.47710955e-02
-5.19285560e-01 -9.49483812e-01 2.87179232e-01 5.54648757e-01
1.36579776e+00 -9.27372098e-01 -5.77271461e-01 8.64501595e-02
-6.27692044e-01 -1.59947908e+00 -8.33259463e-01 4.02586311e-02
-7.79592156e-01 -9.48511600e-01 -8.55435431e-01 -8.19666326e-01
6.51398480e-01 8.18353236e-01 1.21796548e+00 2.06114233e-01
-2.48328805e-01 1.68476105e-01 -3.40254873e-01 -3.63636047e-01
1.27421424e-01 -3.24585825e-01 -1.83889300e-01 2.19766214e-01
2.80478969e-02 -5.00490308e-01 -1.27127695e+00 6.51113987e-01
-1.21712065e+00 2.55403250e-01 7.32731342e-01 1.02762949e+00
7.83555925e-01 5.42700708e-01 1.43814340e-01 -5.64451218e-01
4.31201845e-01 1.46226242e-01 -6.11666739e-01 3.82170141e-01
-6.48982227e-01 -2.20244765e-01 5.67913294e-01 -6.13516390e-01
-1.54080391e+00 -1.93483621e-01 -1.43096566e-01 -3.38420928e-01
1.31154805e-01 6.46120980e-02 -3.86533737e-01 -5.54644108e-01
5.38664460e-01 5.12587845e-01 -1.04957275e-01 -2.56586492e-01
3.15404356e-01 6.65744662e-01 9.23944473e-01 -1.45763487e-01
6.43942356e-01 7.33337045e-01 -6.31163493e-02 -7.73472130e-01
-7.19312370e-01 -2.15672046e-01 -2.41566852e-01 -4.34805721e-01
7.75354147e-01 -1.26214957e+00 -8.57547402e-01 9.18753624e-01
-8.59231532e-01 -7.61798561e-01 -1.43935129e-01 1.92132190e-01
-3.62749934e-01 6.05121970e-01 -6.89641476e-01 -4.36231762e-01
-5.78902125e-01 -1.28032768e+00 1.07100570e+00 6.14639878e-01
3.84640157e-01 -9.38703239e-01 -6.27269149e-01 3.72655243e-01
8.93155634e-01 -8.81188065e-02 5.99497378e-01 6.74301565e-01
-6.20336413e-01 2.50258446e-01 -7.65642464e-01 6.20372593e-01
1.17958903e-01 -1.64921910e-01 -1.12727618e+00 -3.03529561e-01
-1.98764205e-02 -3.16156521e-02 8.31195354e-01 5.97373247e-01
1.67064822e+00 -2.18155935e-01 -3.01495679e-02 1.07516003e+00
1.52570581e+00 1.89964578e-01 1.28724384e+00 3.22699010e-01
5.97422123e-01 2.31944412e-01 7.75485516e-01 3.93155247e-01
4.08620983e-01 9.98471081e-01 5.91231585e-01 -9.15011346e-01
-6.49540961e-01 2.22736243e-02 4.97530133e-01 6.32030904e-01
-3.50473002e-02 -2.40833297e-01 -3.57751846e-01 5.12275636e-01
-1.61120415e+00 -6.68786705e-01 -1.35397941e-01 1.96289647e+00
9.70135868e-01 -5.32087423e-02 -2.57589102e-01 1.29048616e-01
7.72334754e-01 2.51158595e-01 -7.40416765e-01 -2.22353309e-01
-5.62786520e-01 4.53050971e-01 7.83318937e-01 3.39313805e-01
-1.07224274e+00 8.62700641e-01 6.21480274e+00 1.20039773e+00
-1.27517366e+00 3.14384192e-01 1.08092642e+00 -1.04046181e-01
-3.57842296e-02 -3.22897017e-01 -5.70187032e-01 4.97620881e-01
8.54644656e-01 2.09768146e-01 6.70816600e-01 5.64190865e-01
3.26062769e-01 -1.70445532e-01 -5.71843147e-01 1.44000804e+00
2.88760841e-01 -1.25950611e+00 -3.01664799e-01 -1.50673576e-02
9.61977243e-01 -1.84665546e-01 7.42597878e-01 5.91056310e-02
2.04157487e-01 -1.01256382e+00 6.56306386e-01 4.86890882e-01
1.32033539e+00 -7.15576410e-01 6.02259099e-01 -1.78603649e-01
-1.45071805e+00 -1.83749184e-01 -4.04453784e-01 1.58735454e-01
1.77806363e-01 6.90611601e-01 -9.19259191e-02 5.53262711e-01
1.34783232e+00 8.82093489e-01 -8.13335896e-01 1.05920267e+00
-2.52551287e-01 2.98361242e-01 -1.06295720e-01 2.12955266e-01
-1.58940896e-03 -2.09884390e-01 1.59458593e-01 1.10287416e+00
4.42227244e-01 2.45801106e-01 -3.21717352e-01 5.25299430e-01
-1.89350843e-01 -1.12459935e-01 -1.53654426e-01 3.73480499e-01
1.64230093e-01 1.36773372e+00 -5.39607465e-01 -3.51333350e-01
-4.67681736e-01 1.41545248e+00 -1.35658264e-01 6.30509913e-01
-1.20809591e+00 -6.99602425e-01 8.08764160e-01 1.08759142e-01
6.39205992e-01 -1.58514753e-01 -3.24006587e-01 -1.03237808e+00
1.55093372e-01 -1.09963572e+00 6.71678483e-02 -1.49171913e+00
-8.51418674e-01 8.19952428e-01 -5.65070271e-01 -1.37145984e+00
4.43323880e-01 -7.83236623e-01 -2.84074247e-01 6.28246546e-01
-2.07369828e+00 -1.19599855e+00 -8.13861072e-01 1.01173592e+00
5.18965840e-01 1.00624338e-01 3.62162948e-01 6.47972822e-01
-5.91235816e-01 7.76256859e-01 2.86341012e-01 -7.62000158e-02
9.51986372e-01 -1.10017407e+00 4.11203742e-01 1.18630147e+00
-2.79694912e-03 3.75994503e-01 5.01413286e-01 -2.81289816e-01
-1.36093128e+00 -1.27823091e+00 5.75537682e-01 4.50391956e-02
4.72180545e-01 -2.66297787e-01 -7.24394739e-01 2.84972727e-01
5.92312634e-01 2.60487705e-01 3.98278594e-01 -1.43386900e-01
-4.47134137e-01 -6.15891993e-01 -9.57442045e-01 9.93425071e-01
1.11439061e+00 -4.74418283e-01 2.40567759e-01 1.79576144e-01
7.74083853e-01 -7.25146830e-01 -9.53169167e-01 3.46005559e-01
5.47793388e-01 -1.23722470e+00 1.29893076e+00 -1.96562894e-02
5.61273515e-01 -4.54031557e-01 -1.39670774e-01 -1.14234960e+00
-1.94021851e-01 -9.36859787e-01 -1.83041602e-01 9.87646341e-01
1.67884663e-01 -4.17531103e-01 3.20797473e-01 5.48547387e-01
-1.21194191e-01 -6.07329965e-01 -5.96630573e-01 -5.26363313e-01
-4.57934260e-01 -6.44244850e-01 8.50688756e-01 6.58391297e-01
-4.31348085e-01 -4.15748917e-02 -8.75336349e-01 2.98217833e-01
6.81840837e-01 -1.04666008e-02 6.97961450e-01 -6.53104484e-01
-2.24158958e-01 -3.46220851e-01 -1.82526663e-01 -1.42515469e+00
-1.67370349e-01 -2.10834533e-01 -6.19848706e-02 -1.26053870e+00
2.19046444e-01 -2.77199805e-01 -4.26514745e-01 4.75094587e-01
-4.66640651e-01 9.89118755e-01 1.62998155e-01 -2.05407113e-01
-1.02821398e+00 7.03661144e-01 1.55394673e+00 -2.18479410e-01
-5.41850887e-02 -2.39514694e-01 -1.07599366e+00 5.45953214e-01
5.93989015e-01 8.88492242e-02 -4.46002990e-01 -8.36745620e-01
4.29853737e-01 -3.08287553e-02 6.85362697e-01 -1.01218629e+00
3.13404351e-01 -3.92632298e-02 5.95152497e-01 -4.08143371e-01
6.05185151e-01 -9.42717254e-01 8.68382975e-02 2.25412726e-01
-2.23804116e-01 2.46131301e-01 4.05422628e-01 4.96505290e-01
-4.99184698e-01 3.10159087e-01 1.05566955e+00 8.15602541e-02
-1.24607766e+00 5.13909578e-01 1.93843767e-02 -3.28190565e-01
8.48459303e-01 -4.72500414e-01 -5.77368557e-01 -6.23534799e-01
-3.56016010e-01 -9.71323904e-03 3.04708987e-01 4.82050806e-01
9.77025449e-01 -1.30052209e+00 -5.34075439e-01 2.23644629e-01
-1.00552291e-01 -2.66091347e-01 9.41722691e-01 9.61047053e-01
-4.38362926e-01 1.10176846e-01 -5.23606420e-01 -4.88588691e-01
-1.53379071e+00 4.18030113e-01 5.32378733e-01 -2.05393344e-01
-5.56013703e-01 8.65158498e-01 3.36469024e-01 5.14073148e-02
1.35423556e-01 -2.27872208e-01 -9.56854075e-02 -3.94624978e-01
8.42220664e-01 3.92962664e-01 1.05201393e-01 -5.83531737e-01
5.43539878e-03 7.55174458e-01 -1.15179814e-01 1.24003783e-01
1.34085655e+00 -7.85188437e-01 -8.79763067e-02 -2.05973476e-01
9.02563095e-01 -7.41548613e-02 -1.69575810e+00 -5.16756415e-01
-7.02581167e-01 -8.24567735e-01 4.83142287e-01 -9.70052660e-01
-1.67067528e+00 9.07073200e-01 1.21479750e+00 -7.80433416e-02
1.95814621e+00 -4.78718281e-01 1.11609817e+00 2.35257939e-01
5.14013506e-02 -1.27466142e+00 4.80054647e-01 2.68672705e-01
7.81431854e-01 -1.42371285e+00 -4.29303236e-02 -5.49926877e-01
-7.13709712e-01 9.34560716e-01 7.45062053e-01 3.23003620e-01
4.77752507e-01 5.29284477e-01 1.32450923e-01 8.40688124e-03
-5.21033108e-01 -3.27843964e-01 2.55942732e-01 7.71252215e-01
2.13773787e-01 -2.56544113e-01 1.31558254e-01 3.00845742e-01
-6.69050515e-02 1.86748356e-01 4.39303100e-01 3.20602536e-01
-1.99271485e-01 -8.39916289e-01 -4.60026473e-01 2.30445161e-01
-5.26898444e-01 -5.64254165e-01 1.74940035e-01 6.62231982e-01
2.96747386e-01 1.38748205e+00 1.93334762e-02 -5.14838159e-01
3.15925419e-01 -7.33767986e-01 4.36562866e-01 9.70513076e-02
-6.08225882e-01 7.04147369e-02 -1.81872487e-01 -8.15041125e-01
-6.67555928e-01 -2.15732977e-01 -6.42049849e-01 -5.59056759e-01
-2.71752149e-01 -3.37362975e-01 7.33877659e-01 6.08486056e-01
4.88496602e-01 7.93511748e-01 5.85950911e-01 -6.17527068e-01
3.51395272e-03 -7.25874543e-01 -7.16659606e-01 4.54483658e-01
3.47627580e-01 -4.00276601e-01 -9.93772876e-03 3.64309013e-01] | [10.92416763305664, -2.1779685020446777] |
f82e6df6-6795-4573-88e8-386a56e6b24f | espt-a-self-supervised-episodic-spatial | 2304.13287 | null | https://arxiv.org/abs/2304.13287v1 | https://arxiv.org/pdf/2304.13287v1.pdf | ESPT: A Self-Supervised Episodic Spatial Pretext Task for Improving Few-Shot Learning | Self-supervised learning (SSL) techniques have recently been integrated into the few-shot learning (FSL) framework and have shown promising results in improving the few-shot image classification performance. However, existing SSL approaches used in FSL typically seek the supervision signals from the global embedding of every single image. Therefore, during the episodic training of FSL, these methods cannot capture and fully utilize the local visual information in image samples and the data structure information of the whole episode, which are beneficial to FSL. To this end, we propose to augment the few-shot learning objective with a novel self-supervised Episodic Spatial Pretext Task (ESPT). Specifically, for each few-shot episode, we generate its corresponding transformed episode by applying a random geometric transformation to all the images in it. Based on these, our ESPT objective is defined as maximizing the local spatial relationship consistency between the original episode and the transformed one. With this definition, the ESPT-augmented FSL objective promotes learning more transferable feature representations that capture the local spatial features of different images and their inter-relational structural information in each input episode, thus enabling the model to generalize better to new categories with only a few samples. Extensive experiments indicate that our ESPT method achieves new state-of-the-art performance for few-shot image classification on three mainstay benchmark datasets. The source code will be available at: https://github.com/Whut-YiRong/ESPT. | ['Shengwu Xiong', 'Yaxiong Chen', 'Zhaoyang Sun', 'Xiongbo Lu', 'Yi Rong'] | 2023-04-26 | null | null | null | null | ['few-shot-image-classification'] | ['computer-vision'] | [ 1.44028276e-01 -1.83504313e-01 -5.67265809e-01 -4.63111281e-01
-6.00640833e-01 -8.94922391e-02 7.03246057e-01 1.21226266e-01
-1.23790279e-01 4.98545885e-01 1.20314904e-01 3.72124553e-01
-3.36987495e-01 -8.79665792e-01 -8.16498160e-01 -8.14439476e-01
6.70852140e-02 1.51470631e-01 2.83201993e-01 -1.09321922e-01
1.76260114e-01 2.84583569e-01 -1.88704789e+00 3.72961849e-01
6.98298216e-01 1.24250507e+00 5.06984353e-01 2.52101719e-01
-2.26436466e-01 9.52402771e-01 -2.19261721e-01 2.26917148e-01
7.59718493e-02 -7.41150320e-01 -6.60084844e-01 3.34600747e-01
3.86202991e-01 -1.19967215e-01 -6.88749492e-01 9.13406849e-01
3.94943774e-01 6.60759509e-01 6.40714943e-01 -1.22480237e+00
-7.85989821e-01 3.37283164e-01 -5.11207998e-01 5.22766292e-01
2.70710677e-01 2.41535887e-01 9.32196200e-01 -1.21776831e+00
8.61160159e-01 9.40902412e-01 4.05202270e-01 4.02391374e-01
-1.20227742e+00 -3.85119736e-01 1.05670951e-01 7.07023442e-01
-1.42990804e+00 -4.32185441e-01 1.09993315e+00 -2.56043106e-01
9.10071850e-01 3.88538316e-02 8.73357713e-01 1.05099940e+00
1.02376461e-01 1.05823934e+00 1.20442545e+00 -4.30093616e-01
4.59118307e-01 2.83124983e-01 2.05431029e-01 7.79474556e-01
-2.53512204e-01 1.30255803e-01 -1.01891077e+00 3.56344670e-01
5.59920192e-01 4.86036599e-01 -2.42038772e-01 -7.22487926e-01
-8.38111222e-01 8.85160744e-01 8.45826209e-01 5.18107355e-01
-3.22063029e-01 2.76497975e-02 3.09336215e-01 3.63383979e-01
6.31992996e-01 2.03381762e-01 1.66761801e-02 -7.58561939e-02
-8.91265094e-01 -5.89079894e-02 3.37398112e-01 9.16558146e-01
1.23676252e+00 -5.04631586e-02 -3.18012387e-01 9.31303740e-01
-2.16657165e-02 2.90137917e-01 8.21293294e-01 -6.79191768e-01
2.34468743e-01 6.88040614e-01 -2.68647373e-01 -1.09014308e+00
-1.21178694e-01 -4.57307309e-01 -8.09660852e-01 -1.25528043e-02
-1.45275161e-01 3.98115158e-01 -9.77436960e-01 1.58748078e+00
3.40831757e-01 6.53636158e-01 -1.15094306e-02 7.48108923e-01
8.88447106e-01 9.51819003e-01 -3.08721457e-02 -3.71232957e-01
9.94807422e-01 -1.17078507e+00 -5.26737690e-01 -4.87057269e-01
6.75755322e-01 -3.67367357e-01 1.30632722e+00 7.15878904e-02
-8.73382986e-01 -8.70376110e-01 -1.22931385e+00 9.68890712e-02
-6.50915563e-01 -7.76715651e-02 4.67897594e-01 1.07491538e-01
-6.38358176e-01 7.44818985e-01 -7.49717236e-01 -5.33730268e-01
8.29090178e-01 -1.43438682e-01 -2.53774643e-01 -5.99636555e-01
-1.20689142e+00 6.76418245e-01 4.83534127e-01 -2.35079840e-01
-8.84882629e-01 -8.42338800e-01 -1.11318266e+00 8.58398899e-02
6.05563581e-01 -2.92486906e-01 9.04204905e-01 -7.92958856e-01
-1.23886812e+00 8.20364118e-01 -2.35088274e-01 -4.30177271e-01
2.80177146e-01 1.62697844e-02 -2.89031386e-01 4.01958019e-01
3.96831751e-01 6.73499107e-01 1.07015085e+00 -1.18819416e+00
-4.23592865e-01 -3.55619550e-01 -1.81572691e-01 3.06010306e-01
-6.38641357e-01 -4.12668884e-01 -4.23029393e-01 -7.31187999e-01
3.44010815e-02 -7.84925699e-01 7.77294263e-02 2.40982711e-01
-3.09830271e-02 -2.83377767e-01 9.69697237e-01 -1.68342710e-01
1.08688891e+00 -2.57273841e+00 3.48519832e-01 -9.09605473e-02
-7.72863626e-02 3.52101058e-01 -3.95100117e-01 4.33843285e-01
-9.99548882e-02 -3.25708389e-01 -2.39806324e-01 -4.64212149e-01
-1.46728933e-01 4.33724403e-01 -5.34560859e-01 4.52409178e-01
5.32539077e-02 1.24432099e+00 -1.07434177e+00 -5.63246846e-01
7.44844913e-01 3.10017437e-01 -1.74317062e-01 2.81002581e-01
-1.81800872e-01 3.53143722e-01 -2.60956794e-01 4.89699423e-01
4.02523547e-01 -5.01617849e-01 -2.47565567e-01 -1.72057778e-01
1.65691655e-02 -1.30906418e-01 -9.98517692e-01 2.12142134e+00
-5.63342452e-01 4.91195291e-01 -6.81979001e-01 -1.22681785e+00
1.01461887e+00 5.18302768e-02 6.97136581e-01 -1.08340073e+00
4.15291227e-02 -5.23391478e-02 -4.15350914e-01 -4.90160465e-01
7.27435946e-03 -2.77560264e-01 -2.77725011e-02 5.54657102e-01
5.83806932e-01 -6.10275939e-02 2.58908272e-01 2.67146975e-01
7.98905194e-01 3.64731881e-03 6.30964339e-01 2.23922823e-02
2.89519727e-01 -1.60472453e-01 4.96976912e-01 8.46309960e-01
-3.75562519e-01 6.10000372e-01 7.47070275e-03 -5.02731323e-01
-9.33518648e-01 -1.25655150e+00 -1.32536620e-01 1.09060180e+00
4.89492983e-01 -4.18778002e-01 -3.54963452e-01 -6.79920673e-01
-1.27788022e-01 7.03728378e-01 -9.40918088e-01 -7.16251910e-01
-3.02836955e-01 -5.17238259e-01 -2.52162516e-02 5.49968660e-01
8.25356066e-01 -1.27715015e+00 -8.01107109e-01 2.34749645e-01
-2.80799866e-02 -1.00101185e+00 -4.75705773e-01 2.21779913e-01
-7.68170714e-01 -9.07966256e-01 -7.24083722e-01 -9.16262686e-01
5.53508222e-01 7.21557677e-01 6.40505314e-01 -5.35725690e-02
-6.55834854e-01 3.31590861e-01 -6.97733164e-01 -9.14678257e-03
-8.00354779e-03 -8.94112699e-03 -2.28013936e-02 3.53790551e-01
3.02893400e-01 -7.28349626e-01 -5.57600021e-01 3.30846041e-01
-9.15942013e-01 1.28598511e-01 5.62830508e-01 1.06909442e+00
8.23139489e-01 2.45992869e-01 6.37720764e-01 -7.57710457e-01
2.35695928e-01 -6.61721110e-01 -5.71527109e-02 4.20770675e-01
-5.89624584e-01 2.37941518e-02 7.67636776e-01 -6.04379416e-01
-1.14957726e+00 -8.86294320e-02 2.52423435e-01 -9.97374296e-01
-1.99586153e-01 5.36860585e-01 -7.95048401e-02 -8.56931731e-02
7.07908750e-01 8.79295170e-01 6.41447455e-02 -3.02821428e-01
5.21854877e-01 4.64852035e-01 4.53368038e-01 -2.73375124e-01
7.84791410e-01 6.23771727e-01 -1.26690656e-01 -9.61693048e-01
-1.33499479e+00 -6.87511206e-01 -8.80502820e-01 -4.31260645e-01
6.35148287e-01 -7.90136576e-01 3.69328968e-02 5.71379721e-01
-6.34140909e-01 -4.11623031e-01 -7.89674819e-01 4.00709748e-01
-8.04218888e-01 2.50330120e-01 -3.52626890e-01 -6.26267076e-01
-1.18721455e-01 -8.84505510e-01 9.56140339e-01 4.36971188e-01
-1.37135282e-01 -1.07830977e+00 1.42192885e-01 1.52416930e-01
3.04477721e-01 1.24963127e-01 9.07896578e-01 -3.88177514e-01
-5.31054854e-01 -2.38666505e-01 -1.50147080e-01 2.83865899e-01
3.71980995e-01 -4.82810527e-01 -9.58199501e-01 -5.41277945e-01
2.58777440e-01 -7.32450068e-01 1.16962063e+00 2.98853904e-01
1.25184274e+00 -3.40168774e-01 -2.97134548e-01 7.76324928e-01
1.65435731e+00 1.18307203e-01 4.90223497e-01 2.41687283e-01
6.21036410e-01 4.50038880e-01 9.25588489e-01 5.77404261e-01
4.41964529e-02 6.60818040e-01 1.13460816e-01 1.85935482e-01
-4.49496448e-01 -4.62036490e-01 3.21739346e-01 7.83926666e-01
7.74447620e-02 6.21785372e-02 -7.12731719e-01 6.10407472e-01
-2.09633970e+00 -1.21184683e+00 4.95940328e-01 1.98325324e+00
6.63350701e-01 1.76874667e-01 -1.33914560e-01 4.65809330e-02
6.24213874e-01 7.77227342e-01 -9.64100838e-01 2.05542669e-02
-1.40391722e-01 1.80195555e-01 4.37869206e-02 1.91438287e-01
-1.03368819e+00 1.00636435e+00 4.97223425e+00 1.16825736e+00
-1.17433167e+00 2.32323840e-01 5.41030288e-01 -3.41823429e-01
-8.03267509e-02 -6.60571828e-02 -6.77042067e-01 5.58913469e-01
5.38713336e-01 -5.09128928e-01 5.92051387e-01 9.58640516e-01
1.74967926e-02 -1.52396977e-01 -1.00475645e+00 1.16711748e+00
5.03209591e-01 -1.58030760e+00 2.00687781e-01 -1.56224474e-01
9.95520771e-01 -1.17810005e-02 3.75402510e-01 5.29807210e-01
-2.42452741e-01 -9.43496823e-01 4.76082683e-01 6.65349722e-01
8.11790884e-01 -6.17808163e-01 5.20290017e-01 4.67941374e-01
-1.40567470e+00 -3.64987105e-01 -4.51335341e-01 2.01192982e-02
1.32851824e-01 4.34265733e-01 -4.65801954e-01 5.64224005e-01
7.71394789e-01 1.47579479e+00 -8.50954413e-01 9.59049404e-01
-3.11036259e-01 2.70998955e-01 2.63990797e-02 2.31074356e-02
4.27774578e-01 -1.16172992e-01 6.12986028e-01 7.39486098e-01
2.75959820e-01 1.19794324e-01 1.96336478e-01 8.61178935e-01
1.68617386e-02 8.99318084e-02 -9.06380057e-01 -1.54358987e-02
5.34939885e-01 1.04138553e+00 -7.53047943e-01 -4.42585826e-01
-5.03092825e-01 1.06291544e+00 5.38269699e-01 2.71046370e-01
-6.58551812e-01 -3.34512472e-01 3.47893357e-01 1.69334132e-02
5.91516554e-01 1.21938631e-01 -9.29262713e-02 -1.17070115e+00
1.19254135e-01 -5.96959472e-01 5.26471317e-01 -8.73557866e-01
-1.43555629e+00 4.81555015e-01 1.28234386e-01 -1.51634443e+00
-2.13002428e-01 -2.16085792e-01 -9.04725015e-01 4.00495678e-01
-1.50064504e+00 -1.25368249e+00 -5.91189444e-01 7.69773722e-01
1.03292537e+00 -3.99199396e-01 7.21318781e-01 4.24276777e-02
-5.53320467e-01 5.00373185e-01 2.00579867e-01 -8.79831687e-02
5.83131850e-01 -1.01196408e+00 6.94638938e-02 7.50465155e-01
4.39318299e-01 3.97492051e-01 5.31917095e-01 -4.92077351e-01
-1.22287261e+00 -1.40853441e+00 5.01691580e-01 -5.79416044e-02
6.58225715e-01 -2.36293748e-01 -1.27129531e+00 5.75329423e-01
2.12754738e-02 4.92262602e-01 5.37912369e-01 -1.54413894e-01
-4.31705385e-01 -3.41001064e-01 -8.32169116e-01 4.78207856e-01
1.20756984e+00 -8.52788210e-01 -7.17368543e-01 3.14772874e-01
5.69199264e-01 3.37868407e-02 -7.32254148e-01 2.72786498e-01
2.70034850e-01 -8.96244109e-01 1.04916990e+00 -4.09218341e-01
5.82805037e-01 -9.74993259e-02 -2.80802667e-01 -1.52865756e+00
-3.72984022e-01 -1.47274062e-01 -4.48881745e-01 1.05589068e+00
-1.30612344e-01 -4.56662297e-01 6.39341593e-01 1.03651762e-01
-1.51384562e-01 -1.05903852e+00 -9.44316208e-01 -1.06972635e+00
-1.34034529e-01 -2.41726339e-01 3.32089782e-01 9.06685114e-01
1.99600607e-01 3.60232234e-01 -5.75524628e-01 -1.43735126e-01
9.11761224e-01 5.58308005e-01 5.75216591e-01 -9.64692175e-01
-3.27686816e-01 -2.80264497e-01 -6.76787019e-01 -8.07148755e-01
3.53956372e-01 -1.17443001e+00 1.21930741e-01 -1.45239127e+00
5.21924436e-01 -2.87291229e-01 -5.31198859e-01 5.28490782e-01
-1.65571094e-01 3.23413163e-01 3.58288229e-01 4.01842594e-01
-9.58364666e-01 1.05406821e+00 1.25640440e+00 -4.21173692e-01
-1.60179824e-01 -1.21863469e-01 -3.54701757e-01 5.26764035e-01
8.32187057e-01 -5.18895090e-01 -7.78340161e-01 -5.26679866e-02
-2.94296503e-01 -1.50406927e-01 4.75502163e-01 -1.29856622e+00
4.55473542e-01 -3.21040988e-01 4.90709692e-01 -4.80014473e-01
5.18554389e-01 -5.01836717e-01 -9.77313071e-02 5.04002810e-01
-3.18536609e-01 -5.78777790e-01 -2.35627014e-02 8.52477431e-01
-4.68790293e-01 -3.01186234e-01 1.06074250e+00 -1.80281654e-01
-1.34035873e+00 6.21038854e-01 -1.33710466e-02 1.03734747e-01
1.44939148e+00 -2.43467942e-01 -3.52871776e-01 -2.81507045e-01
-7.94215858e-01 1.51213869e-01 5.74790835e-01 6.18383944e-01
1.00327241e+00 -1.54949987e+00 -2.56897748e-01 4.63424891e-01
6.50271416e-01 -4.83856462e-02 7.63610423e-01 7.95459688e-01
7.53610488e-03 3.70170891e-01 -5.61009288e-01 -6.86019719e-01
-1.13217938e+00 8.75317156e-01 1.80620998e-01 1.44686056e-02
-1.01966357e+00 7.01887488e-01 2.99476802e-01 -1.41612232e-01
1.89819515e-01 7.56612644e-02 -1.85256928e-01 2.55401254e-01
7.41864800e-01 3.27854484e-01 -1.11049503e-01 -6.16948843e-01
-1.74886242e-01 7.57922709e-01 -1.77440047e-01 1.17395734e-02
1.60287106e+00 -2.02649161e-01 2.37508044e-01 1.04928744e+00
1.70672667e+00 -6.07277811e-01 -1.46740818e+00 -7.42077470e-01
-1.50864020e-01 -7.10558951e-01 1.58158377e-01 -3.20783287e-01
-1.05722106e+00 9.17823136e-01 6.58368289e-01 -8.74312297e-02
1.02285314e+00 3.89160395e-01 7.64331698e-01 4.11883473e-01
4.53894526e-01 -1.11084962e+00 7.66836107e-01 3.55410218e-01
8.58037412e-01 -1.35713124e+00 2.04989649e-02 -1.81982175e-01
-7.65315711e-01 8.98694932e-01 6.64904952e-01 -3.09090048e-01
8.23362708e-01 -2.59362191e-01 -2.31453463e-01 -3.25766832e-01
-8.06596160e-01 -2.89576203e-01 3.69578600e-01 4.93045688e-01
-1.44848287e-01 -1.56182602e-01 1.36043087e-01 3.36341172e-01
9.85994413e-02 2.05598846e-01 2.46370673e-01 1.12687612e+00
-5.41071653e-01 -7.98216701e-01 1.44147322e-01 5.40266275e-01
3.82415712e-01 7.73704350e-02 -1.29960313e-01 5.50834835e-01
2.68564820e-02 6.09971702e-01 1.56660140e-01 -3.75009328e-01
2.33261004e-01 1.77296534e-01 5.74381590e-01 -9.86043274e-01
1.30115703e-01 -1.32981345e-01 -4.67267305e-01 -7.04502702e-01
-4.81493622e-01 -6.42112255e-01 -1.20004070e+00 -8.26743096e-02
-4.43253852e-02 3.81375365e-02 2.16735095e-01 1.11268950e+00
3.18602383e-01 5.48351228e-01 1.01523149e+00 -9.36912179e-01
-4.65833575e-01 -9.03978050e-01 -8.34561229e-01 6.48405552e-01
2.20000073e-01 -1.03899789e+00 -4.73955542e-01 -2.99187135e-02] | [9.967394828796387, 2.6007578372955322] |
2dd267a4-5f42-4e51-98c3-9e42e562dae6 | guided-perturbations-self-corrective-behavior | null | null | http://openaccess.thecvf.com/content_iccv_2017/html/Sankaranarayanan_Guided_Perturbations_Self-Corrective_ICCV_2017_paper.html | http://openaccess.thecvf.com/content_ICCV_2017/papers/Sankaranarayanan_Guided_Perturbations_Self-Corrective_ICCV_2017_paper.pdf | Guided Perturbations: Self-Corrective Behavior in Convolutional Neural Networks | Convolutional Neural Networks have been a subject of great importance over the past decade and great strides have been made in their utility for producing state of the art performance in many computer vision problems. However, the behavior of deep networks is yet to be fully understood and is still an active area of research. In this work, we present an intriguing behavior: pre-trained CNNs can be made to improve their predictions by structurally perturbing the input. We observe that these perturbations - referred as Guided Perturbations - enable a trained network to improve its prediction performance without any learning or change in network weights. We perform various ablative experiments to understand how these perturbations affect the local context and feature representations. Furthermore, we demonstrate that this idea can improve performance of several existing approaches on semantic segmentation and scene labeling tasks on the PASCAL VOC dataset and supervised classification tasks on MNIST and CIFAR10 datasets.
| ['Ser Nam Lim', 'Swami Sankaranarayanan', 'Arpit Jain'] | 2017-10-01 | null | null | null | iccv-2017-10 | ['scene-labeling'] | ['computer-vision'] | [ 7.15428054e-01 3.21056157e-01 -9.59093962e-03 -7.84448326e-01
-1.93520039e-01 -7.25768209e-01 5.48743069e-01 -6.29458949e-02
-6.34077430e-01 6.47401273e-01 -7.02614263e-02 -3.13393444e-01
2.41495803e-01 -6.60766602e-01 -1.11581433e+00 -7.36348093e-01
1.77966210e-03 2.48361886e-01 6.37092054e-01 -2.99877673e-01
3.09534103e-01 7.60260224e-01 -1.41229391e+00 5.25293350e-01
5.17405033e-01 7.95391440e-01 -4.80352677e-02 6.40505433e-01
1.37706861e-01 8.40310931e-01 -6.28372967e-01 -2.46878684e-01
3.03907901e-01 -3.67706805e-01 -1.24922502e+00 -2.49621514e-02
7.68646657e-01 7.80469775e-02 -1.78744540e-01 1.11372733e+00
4.08673316e-01 1.98903352e-01 5.60656369e-01 -8.03822875e-01
-5.34530342e-01 6.22990012e-01 -4.38843727e-01 4.76103067e-01
-1.10982828e-01 2.35989824e-01 8.73456717e-01 -3.99319649e-01
6.22313201e-01 1.17819071e+00 6.33185089e-01 6.78455651e-01
-1.44167817e+00 -5.13707221e-01 3.72500211e-01 3.76382083e-01
-1.02864182e+00 -4.08980727e-01 8.84183466e-01 -3.48617196e-01
1.03672755e+00 4.69189025e-02 5.61214924e-01 1.01524007e+00
9.84898284e-02 8.17001164e-01 9.74296510e-01 -5.04196882e-01
2.36782283e-01 7.93523639e-02 1.30495697e-01 7.60920465e-01
2.24197596e-01 2.14655846e-01 -2.37606302e-01 3.75959337e-01
5.30373454e-01 -1.85179770e-01 -3.38238090e-01 -4.47262883e-01
-9.81602252e-01 8.76438022e-01 8.98621976e-01 4.27923650e-01
-8.03308263e-02 4.60350454e-01 5.23588777e-01 1.29825011e-01
4.62691873e-01 1.13902545e+00 -7.02252209e-01 1.20340455e-02
-6.70033634e-01 3.21310014e-01 4.62261409e-01 4.03111279e-01
8.23982239e-01 2.10809618e-01 -1.58251852e-01 7.47550726e-01
-1.04198821e-01 3.01364297e-03 4.92612660e-01 -1.00828135e+00
2.25908607e-01 7.05156624e-01 -7.93879852e-02 -9.43412185e-01
-6.49897993e-01 -5.91371477e-01 -6.56101465e-01 4.56579387e-01
6.59540176e-01 -2.75359690e-01 -1.10043347e+00 1.75915575e+00
9.62262452e-02 2.85801828e-01 1.25676975e-01 8.69643152e-01
7.82747686e-01 4.61322308e-01 1.44094899e-01 4.15553480e-01
9.20801222e-01 -1.19243622e+00 -3.46825063e-01 -5.20379126e-01
7.82570839e-01 -6.35263860e-01 1.02476943e+00 2.56016195e-01
-9.64345157e-01 -6.93534136e-01 -1.15333927e+00 3.40594389e-02
-5.96371770e-01 -5.22509515e-02 5.90620220e-01 5.65256298e-01
-9.92128193e-01 9.83457386e-01 -1.07132518e+00 -6.71224415e-01
8.85307074e-01 5.44610441e-01 -1.91295028e-01 -1.11959182e-01
-7.98278749e-01 8.96357954e-01 6.38959229e-01 1.70665696e-01
-1.06414008e+00 -7.77748346e-01 -6.49990916e-01 2.17196718e-01
3.50009531e-01 -3.67578089e-01 1.39348590e+00 -1.19462299e+00
-1.41002560e+00 1.03201175e+00 1.11204516e-02 -8.11640024e-01
3.35206300e-01 -5.47937155e-02 -1.86651722e-01 1.15299830e-02
-2.61735052e-01 1.16871047e+00 7.48202205e-01 -1.26567745e+00
-5.76190293e-01 -2.99008310e-01 1.77811369e-01 4.65216488e-02
-1.89781874e-01 2.66220653e-03 -4.39832285e-02 -6.86758399e-01
1.09777801e-01 -1.26850176e+00 -5.21314204e-01 -1.82147890e-01
-7.13284314e-01 -4.45794165e-02 9.44322884e-01 -2.05803275e-01
4.46598947e-01 -2.08207178e+00 2.34964024e-02 1.63766239e-02
-1.19002566e-01 6.89717054e-01 -4.12345797e-01 1.69002384e-01
-2.56766528e-01 5.01282454e-01 -4.32038993e-01 -2.30259538e-01
-3.34793150e-01 3.85718524e-01 -4.46702093e-01 5.08288443e-01
4.85156804e-01 1.07196200e+00 -6.41103268e-01 2.92424746e-02
3.47132474e-01 4.85183209e-01 -6.81198895e-01 1.34217918e-01
-5.56404948e-01 7.69347429e-01 -2.42731109e-01 2.34529689e-01
4.95462239e-01 -1.44853562e-01 7.26196766e-02 -2.42450282e-01
-2.57131495e-02 2.86987662e-01 -5.91062665e-01 1.42117870e+00
-1.01998419e-01 1.05906999e+00 -2.22383514e-01 -1.63131666e+00
7.63659716e-01 3.07686012e-02 1.97251543e-01 -8.09128523e-01
3.10153335e-01 -1.97363928e-01 4.11044270e-01 -4.02364969e-01
3.19338202e-01 -1.14369392e-01 2.61461198e-01 2.13790938e-01
-5.05217686e-02 -9.61690322e-02 1.83127135e-01 -2.74457276e-01
1.10627544e+00 2.73536053e-02 6.33766968e-03 -3.25305820e-01
4.76420790e-01 4.51760024e-01 4.17535245e-01 7.39047170e-01
-3.46012115e-01 7.07309663e-01 5.22017598e-01 -7.14012206e-01
-1.07010877e+00 -6.98010981e-01 -9.45159942e-02 1.25968897e+00
4.18252312e-02 7.88736567e-02 -9.97906566e-01 -8.05494249e-01
-1.16356947e-01 6.76751971e-01 -9.46350098e-01 -4.32874799e-01
-7.09513605e-01 -8.09177935e-01 8.40429068e-01 8.14340830e-01
7.87693501e-01 -1.54496717e+00 -6.43349171e-01 1.36767820e-01
2.54444420e-01 -1.42258620e+00 9.56356823e-02 5.42916536e-01
-1.01191711e+00 -1.15334702e+00 -4.55382794e-01 -1.07066596e+00
8.55530143e-01 6.34488091e-02 1.11221480e+00 1.99868292e-01
-4.53630209e-01 1.43913090e-01 -4.31410193e-01 -5.34796655e-01
-4.30782527e-01 4.35586303e-01 -2.89555371e-01 -1.24706373e-01
3.18801165e-01 -6.44268334e-01 -5.51674843e-01 3.89901757e-01
-1.02751720e+00 3.75222452e-02 4.69177753e-01 7.91887581e-01
5.82179785e-01 6.55871183e-02 4.98533189e-01 -1.28029144e+00
4.13158298e-01 -8.48781615e-02 -6.17474616e-01 -1.26149252e-01
-4.19771165e-01 2.49517545e-01 8.77108097e-01 -2.99036890e-01
-1.04190624e+00 3.44135642e-01 -4.62674856e-01 -8.75289068e-02
-8.37568998e-01 3.50941211e-01 4.54913042e-02 -3.98301452e-01
9.20948863e-01 9.06596798e-03 -3.26920360e-01 -5.20376921e-01
4.58556086e-01 2.23517746e-01 6.88349545e-01 -6.55402124e-01
7.57120550e-01 7.15838134e-01 -3.46700065e-02 -7.77833819e-01
-1.23449624e+00 -3.36848110e-01 -8.25044453e-01 1.50958285e-01
1.23631954e+00 -5.11278391e-01 -4.72105294e-01 5.40382564e-01
-1.10105217e+00 -8.21300805e-01 -3.53909403e-01 6.80388957e-02
-5.14255941e-01 -8.71709362e-02 -5.61135292e-01 -2.00254783e-01
1.04698241e-02 -1.33463275e+00 6.51737511e-01 5.02446532e-01
-1.67741030e-01 -1.07970655e+00 3.21581177e-02 4.32874739e-01
3.87307197e-01 3.58831286e-01 1.11680019e+00 -9.12297606e-01
-5.31628788e-01 -2.87163481e-02 -3.84407043e-01 5.61248422e-01
1.57215983e-01 2.90381964e-02 -1.26271224e+00 -3.20866197e-01
-3.15024018e-01 -5.20865560e-01 1.18469298e+00 4.13094461e-01
1.63544011e+00 -1.49542163e-03 -3.84192735e-01 8.51113498e-01
1.32707644e+00 3.42188835e-01 7.00301170e-01 6.07110858e-01
7.43368447e-01 5.91320753e-01 2.40242541e-01 -5.25886677e-02
-2.34816089e-01 5.23541212e-01 8.56235743e-01 -3.82248610e-01
-3.07681888e-01 -5.75341517e-04 -1.15586422e-01 6.19411692e-02
1.68613717e-02 -1.98472843e-01 -1.08307421e+00 6.77416623e-01
-1.76474142e+00 -8.23411226e-01 -1.51961308e-03 1.77023852e+00
6.60026014e-01 2.52469063e-01 -2.27899611e-01 1.05851859e-01
6.46699131e-01 3.76799852e-01 -8.29747498e-01 -5.70745468e-01
-1.32809460e-01 3.59264970e-01 5.25169313e-01 2.85686016e-01
-1.23797810e+00 1.58062363e+00 6.79233074e+00 3.81418198e-01
-1.51369381e+00 -2.51868397e-01 9.46250677e-01 1.07863039e-01
6.09208755e-02 -1.03089035e-01 -6.86498463e-01 6.54938295e-02
6.68487608e-01 3.99686575e-01 4.00040358e-01 9.81885076e-01
3.85167189e-02 -1.07615076e-01 -1.26896298e+00 6.12569273e-01
-6.60569593e-02 -1.61104536e+00 -1.31120766e-02 -8.32451507e-02
1.05323875e+00 4.02672619e-01 2.66117513e-01 2.56115705e-01
6.77602291e-01 -1.36001587e+00 3.74586552e-01 2.15090290e-01
3.92417729e-01 -7.27829576e-01 7.94831693e-01 2.86934763e-01
-7.05288470e-01 8.04407895e-02 -5.83095253e-01 -1.92842230e-01
-1.54707596e-01 4.62379277e-01 -1.06610155e+00 -5.95127866e-02
9.45791245e-01 6.05587304e-01 -8.79575074e-01 1.15855765e+00
-4.63923424e-01 1.15540886e+00 -1.24873213e-01 -1.61118880e-02
4.75794405e-01 -2.31492110e-02 3.44979227e-01 1.12784815e+00
-2.26299971e-01 -2.26263747e-01 1.25627089e-02 9.46130395e-01
-4.56655860e-01 -1.80945233e-01 -5.25923967e-01 -1.32786810e-01
-1.62059776e-02 1.12306154e+00 -9.75364745e-01 -2.05002859e-01
-2.62727767e-01 9.38918352e-01 6.73954308e-01 6.46045625e-01
-8.03623855e-01 -1.90294906e-01 9.53622878e-01 -8.46204609e-02
5.87523460e-01 -1.42884031e-01 -4.96901691e-01 -8.32248747e-01
-1.93631694e-01 -7.81969726e-01 1.12632461e-01 -5.96420646e-01
-1.03692365e+00 5.73251784e-01 -2.97760665e-01 -5.88332295e-01
-1.68730244e-02 -8.09830904e-01 -8.14692318e-01 5.11392891e-01
-1.66533279e+00 -6.95032895e-01 -3.41256142e-01 4.33727473e-01
4.80785459e-01 -1.50449604e-01 7.21309423e-01 -4.67800945e-02
-7.94177532e-01 5.02544165e-01 1.46539897e-01 4.03589994e-01
6.10041618e-01 -1.17908275e+00 6.53166831e-01 9.07228768e-01
3.86944532e-01 3.01260918e-01 9.08640504e-01 -2.81983674e-01
-7.91430593e-01 -1.53063893e+00 3.52846265e-01 -3.27749968e-01
5.00893176e-01 -3.97242337e-01 -1.02457106e+00 9.58025038e-01
3.09836656e-01 4.08854097e-01 5.68078220e-01 1.23621695e-01
-4.13411856e-01 -1.38249695e-01 -1.09704030e+00 8.12973082e-01
1.06783402e+00 -3.96135002e-01 -5.28271735e-01 3.73652697e-01
8.54663789e-01 -3.94970924e-01 -4.42697376e-01 6.72014475e-01
2.14594290e-01 -1.18895519e+00 9.45761919e-01 -1.05340219e+00
4.28322107e-01 -2.28075698e-01 -1.57940120e-01 -1.58925843e+00
-3.05912822e-01 -2.02614009e-01 3.08571994e-01 1.16666937e+00
5.07219315e-01 -6.80620670e-01 1.23417354e+00 4.73855555e-01
-4.02500451e-01 -8.85332108e-01 -5.68087935e-01 -7.29806304e-01
3.90776038e-01 -4.72707093e-01 3.94179940e-01 8.95066500e-01
-5.33156991e-01 2.47730404e-01 -8.84437561e-02 2.06989765e-01
2.43694633e-01 5.37536964e-02 8.35970223e-01 -1.01512969e+00
-1.68730423e-01 -5.71088374e-01 -4.87695754e-01 -9.73075628e-01
5.11851609e-01 -8.48205507e-01 2.03124404e-01 -1.46236980e+00
-9.33598913e-03 -3.74742955e-01 -2.98322856e-01 7.50062644e-01
-3.51257622e-02 6.12775862e-01 2.40202546e-01 4.94376197e-02
-5.62008202e-01 4.04733390e-01 1.23257101e+00 -2.53188878e-01
-7.27525428e-02 -3.09406803e-03 -7.17162788e-01 9.88154233e-01
1.16455710e+00 -7.00025082e-01 -3.83041739e-01 -5.13178170e-01
1.40744269e-01 -5.49423099e-01 4.32411969e-01 -1.25845468e+00
-7.33894855e-02 -2.48655006e-02 5.41314006e-01 -2.82901019e-01
2.46113762e-01 -8.08622718e-01 -4.42172021e-01 4.88861471e-01
-6.94956362e-01 -1.23517625e-01 5.69005549e-01 3.94276917e-01
-3.73386949e-01 -3.61804545e-01 1.25780571e+00 -1.31769568e-01
-1.00798309e+00 2.09107757e-01 -4.85230774e-01 2.02787206e-01
1.07943404e+00 -5.88387847e-02 -5.40016353e-01 -7.08470270e-02
-7.89778531e-01 -3.75581048e-02 5.55732012e-01 4.04665232e-01
2.85878032e-01 -9.05543804e-01 -4.06015128e-01 3.38825211e-02
-4.21146303e-02 3.35772961e-01 4.12912518e-02 4.95305568e-01
-7.39131689e-01 5.02928495e-01 -4.05049086e-01 -8.89275730e-01
-1.26281881e+00 5.50922811e-01 7.47260630e-01 -2.28791520e-01
-4.37565416e-01 1.15382469e+00 2.40041584e-01 -7.17656076e-01
3.75233769e-01 -5.17835975e-01 -3.03501129e-01 -2.83545345e-01
3.84123653e-01 -3.60325873e-02 2.38312602e-01 -4.44372296e-01
-2.53776312e-01 5.84501445e-01 -2.84204662e-01 3.49246949e-01
1.41379774e+00 8.27141553e-02 2.35380512e-02 1.28150940e-01
1.30094647e+00 -4.04300183e-01 -1.47874033e+00 -8.71031508e-02
5.69992401e-02 -1.20948493e-01 3.88310105e-02 -1.04918075e+00
-1.35278392e+00 1.18686306e+00 6.90054834e-01 3.30956131e-01
1.03881085e+00 1.10883363e-01 7.21777260e-01 7.49019325e-01
2.27403035e-03 -8.03934157e-01 2.43141770e-01 7.30926156e-01
6.52823806e-01 -1.40629387e+00 -3.45385253e-01 -4.42850351e-01
-4.46884513e-01 9.81188476e-01 9.01519716e-01 -4.95113552e-01
3.97821456e-01 1.44324735e-01 3.01180601e-01 -1.77537739e-01
-4.92940277e-01 -3.69765937e-01 3.19314778e-01 7.68978775e-01
5.20899236e-01 -1.76370859e-01 2.16300666e-01 1.80214822e-01
-3.69610727e-01 -3.29732060e-01 5.20297766e-01 8.16580594e-01
-5.81314802e-01 -1.02729559e+00 -2.36944869e-01 3.54291409e-01
-6.94151759e-01 -5.78140728e-02 -6.38088286e-01 7.56278813e-01
2.17785776e-01 7.09898174e-01 8.07836279e-02 -3.45457107e-01
3.10730875e-01 1.85086042e-01 4.55684155e-01 -8.83697689e-01
-7.20251083e-01 -4.23787922e-01 7.78539404e-02 -6.58471465e-01
-4.37060744e-01 -5.94541728e-01 -1.51399183e+00 -4.54266295e-02
-8.69191810e-02 -1.08640222e-02 7.56653547e-01 1.31279337e+00
2.41297945e-01 8.98104966e-01 2.11628303e-01 -9.70132649e-01
-2.73853481e-01 -8.02382648e-01 -2.10568070e-01 6.41944468e-01
2.25976408e-01 -4.93650198e-01 -2.64676094e-01 2.56817281e-01] | [9.41407585144043, 1.947060465812683] |
385069ba-f554-4498-97c5-acea0fe22b44 | deep-transfer-tensor-factorization-for-multi | 2302.06133 | null | https://arxiv.org/abs/2302.06133v1 | https://arxiv.org/pdf/2302.06133v1.pdf | Deep Transfer Tensor Factorization for Multi-View Learning | This paper studies the data sparsity problem in multi-view learning. To solve data sparsity problem in multiview ratings, we propose a generic architecture of deep transfer tensor factorization (DTTF) by integrating deep learning and cross-domain tensor factorization, where the side information is embedded to provide effective compensation for the tensor sparsity. Then we exhibit instantiation of our architecture by combining stacked denoising autoencoder (SDAE) and CANDECOMP/ PARAFAC (CP) tensor factorization in both source and target domains, where the side information of both users and items is tightly coupled with the sparse multi-view ratings and the latent factors are learned based on the joint optimization. We tightly couple the multi-view ratings and the side information to improve cross-domain tensor factorization based recommendations. Experimental results on real-world datasets demonstrate that our DTTF schemes outperform state-of-the-art methods on multi-view rating predictions. | ['Chunxi Li', 'Ke Xin', 'Penghao Jiang'] | 2023-02-13 | null | null | null | null | ['multi-view-learning'] | ['computer-vision'] | [-0.73879576 -0.51522434 -0.4562071 -0.6672969 -0.6671911 -0.5450509
0.3658083 -0.6275893 0.3891027 0.16270673 1.1238569 -0.01506319
-0.30008823 -0.5652731 -0.73219836 -0.5203514 -0.01042493 0.42104042
-0.55737287 -0.6208074 -0.10498082 -0.1752168 -1.2617844 1.2978095
0.65796614 1.2090728 -0.19427681 0.3959513 0.15344995 0.9419461
-0.11729421 -0.7921122 0.54117006 0.24281609 -0.4819513 0.43911105
0.6297031 -0.88265485 -0.4091626 0.7441273 0.35715476 0.224878
0.5408188 -1.2099563 -1.5994322 0.73028415 -0.7425433 -0.07991974
0.16731417 -0.54613817 1.5809847 -1.5123909 0.18522659 1.402799
0.49093017 0.12222257 -0.998643 -0.4251285 0.6893968 0.2264729
-0.6850687 -0.15329444 0.73506165 -0.7039455 0.76441365 0.0479871
0.62061006 1.2996473 -0.01947262 1.0080835 0.8072489 0.1369407
-0.04347623 0.06628508 0.0539786 0.41818452 -0.01232951 0.02595365
-0.7320478 -0.48831025 0.94672143 0.9309166 0.05281926 -0.563372
-1.5245204 1.2487919 0.48588538 0.12084866 -0.68148464 -0.2050685
0.74133486 0.73779744 0.82915896 -0.03933208 -0.90147996 0.32641742
-0.59007835 0.29157555 0.56156546 0.8895295 0.49784097 0.48215327
0.08793139 1.318173 0.57292455 0.46871883 0.6214025 -0.96079445
0.8326912 0.580599 0.15653692 -1.112626 0.08000903 -0.7552833
-1.208703 -0.37772286 -0.11003207 -0.14887011 -0.6329945 1.408924
0.33556914 0.15722685 0.10271642 1.4354563 0.9230319 0.72640693
-0.3723774 -0.16754161 1.4777277 -1.377743 -0.7053428 0.06350586
0.5628872 -0.7546915 0.778472 0.9843362 -0.89636445 -0.8437976
-1.0522419 -0.27884063 -0.04442391 0.572469 1.0075358 0.21515895
-0.63613695 0.38615716 -0.608397 0.17840508 0.09910544 0.301183
-0.62897086 -0.6263192 -1.1507324 0.20625944 -0.32891062 0.18489906
-1.1126044 -0.9136529 -0.59589463 0.07116143 0.02310516 -1.1446745
1.0371791 -0.92612326 -1.2162962 0.23629136 -0.06936382 0.03263902
-0.13530388 -0.7062946 -0.82264674 -0.0796699 0.22741827 -0.11910121
1.1735245 -1.3961535 -0.43372974 -0.8362173 0.35766128 0.3596543
-0.78033113 0.00908248 -0.2000188 -1.0193639 0.23745802 -0.7654495
-0.2761592 -0.50960374 -0.05604048 -0.09175288 0.90196544 -1.0125161
1.0976659 -2.0018091 0.7643015 0.03843196 0.46999678 -0.06791848
-0.53908014 0.74998045 -0.44681168 -0.26740444 0.70306796 -0.61328626
-0.03275593 0.52245134 -0.86393815 0.29261065 -0.2950068 0.78185046
-0.75566214 0.15810083 -0.0530737 0.6617348 -1.0190575 0.6150527
0.07032824 0.41978097 -0.6331534 0.68588686 0.8894493 -0.57741994
0.45288035 -1.0034541 0.1569129 0.37966642 -1.1886961 2.082723
-0.6234863 -0.444878 0.2915858 -1.1560115 0.78996694 0.6588488
0.84119284 -0.45699617 -0.02240906 -0.09802527 -0.28653222 -0.42651993
0.6960072 -0.3309204 0.03479991 0.82716817 0.67532057 0.46048462
-0.10294064 0.5406799 0.43220744 0.12355572 -0.13900761 0.09248945
0.41824043 -0.4283075 0.68214136 0.06590237 0.42886758 0.5619048
0.2541304 -0.969614 -1.1815059 -0.97788817 0.27442876 1.806564
-0.32410455 -0.88902944 -0.15657282 -0.8447855 0.20482387 -0.01203006
-0.8370882 0.11005711 -0.12156354 -0.77620035 -0.28422695 0.9092157
-0.09562872 -0.05590851 0.81918776 0.11648104 -0.41587955 -1.1565717
-0.71922934 -0.08018079 -1.3184328 -0.7836178 -0.68548435 -0.44852737
0.5563982 1.126066 1.2420684 -0.13716035 0.45411777 0.5090997
-0.7793952 0.2508462 0.15521897 -0.33461437 0.6207411 0.6065583
0.19227806 -0.99990577 -0.79656154 0.72489315 -0.9064168 -0.1825124
0.46328533 1.2117207 0.6154915 -0.38091826 0.56298333 -0.8716745
0.79034555 -0.90957093 -0.15196964 0.25077394 -0.5975379 -0.08431239
0.899526 -0.5131563 -0.90552944 -0.42624226 0.1357193 -1.1359739
0.27684212 0.7977278 -0.21192923 0.26100692 0.45345208 -0.0301994
-0.1305258 -1.059404 0.8580579 0.61179394 -0.05090323 -0.7596096
0.67003226 0.6590697 -0.44114226 -0.08293342 -1.539161 -0.7222708
-0.7079987 -0.11444028 0.6883397 -1.9089339 -0.5613341 0.09947772
-0.9079585 0.46932727 -0.09670148 0.84009725 -0.46239883 0.46224466
-0.93591326 -0.28690118 -0.48764858 -1.1349012 1.2158865 -0.35171178
0.4389798 -1.1248032 0.27091765 1.2691572 0.1534056 -0.36017576
0.8415446 -0.5344557 -0.32429028 -0.21212773 -0.25594848 1.0450016
-0.01857289 -0.3253921 -0.79441935 -0.56043017 0.15912047 -0.64425766
0.80054754 0.30675292 1.0289003 -0.45055747 0.14204584 0.7878912
1.3337866 -0.4004272 0.2958999 -0.08291306 1.3672588 0.38437825
0.682094 0.86376023 1.1096562 0.5171806 0.67143506 0.05111097
0.3375638 -0.5050448 0.55107504 1.8495177 -0.6375105 0.13324006
-0.17477417 0.5439294 -2.272265 -1.0320249 -0.24191426 1.9258863
0.0691724 -0.5554061 0.40900412 -0.08900581 0.35801348 0.44814914
-0.41241637 -0.53055763 -0.10988034 -0.24388403 0.16437495 0.2689366
-1.0571283 0.46975917 6.1426797 0.36840695 -0.8036399 0.64046115
0.40379944 -0.36037794 -0.7642202 -0.13794479 -0.45277107 -0.01790682
0.75624377 0.3143082 0.8748864 0.9411059 0.18803878 0.7413356
-0.9614603 1.0245166 0.34166574 -1.2046632 0.41340435 0.37741745
1.1785057 0.28625757 0.49020207 0.55424756 0.79799867 -0.62402445
0.42334124 0.5312366 0.42971462 -0.72409993 0.52840185 0.29245067
-1.3179567 -0.6600117 -0.8894395 -0.1666541 0.15994693 0.9503757
-0.19770283 1.2025677 0.95718485 1.7478694 -0.04810711 0.2715338
-0.01361417 0.5211205 0.18887182 0.8508892 0.21887656 -0.8609613
0.16225173 0.5034314 0.39922705 0.06390219 0.3913566 0.42520648
-0.13761154 0.1973481 -0.27298427 -0.12294938 -0.07092073 1.5773481
0.42410615 -0.36834493 -1.0351232 0.8973203 0.6052771 0.7276815
-0.5993839 0.6737222 1.330817 -0.15357663 0.8764774 -0.31316885
-0.19006385 -2.2744381 0.12189214 -1.2526451 0.60827404 -0.8279126
-2.062551 0.42118818 -0.38636863 -1.8209704 -0.06284349 -0.6817301
-0.31897673 0.8716397 -1.4904649 -1.9165789 0.08576763 1.1595021
0.27255687 -0.59013003 0.95870095 0.8060422 -0.3562852 0.4937489
0.5619288 0.17040853 0.94346505 -1.1351225 0.17657061 0.47653177
0.23181678 0.958508 0.08983821 -0.44185656 -1.9744203 -1.1580433
0.3971119 -0.46503523 1.0227797 -0.39465198 -0.59060633 1.0624698
0.05047026 0.38749894 1.4437646 0.95574534 -1.4391001 -0.38708696
-0.6185168 0.0447872 0.80400425 -0.84877545 -0.52523774 0.53959656
0.89906114 0.01296543 -1.8225898 0.3022744 0.89577615 -1.208778
1.3699467 -1.4678526 0.98955965 -0.3425279 -0.9007177 -1.7089027
-1.0082359 -0.03131005 -0.53224474 0.9452546 0.23310342 -0.19081478
0.5777019 0.25530085 -0.37003508 -0.91888875 -0.5782202 -0.268624
0.07616827 -0.32851583 0.80947876 1.3340715 0.08093826 0.694077
-1.4239502 0.35722765 0.6343373 0.6189304 0.7882379 -1.179032
-0.91019803 0.37341052 0.01880991 -1.3714468 -0.12564017 -0.8450468
-0.8405772 -1.5622672 0.37955418 -0.02709151 -0.55578697 0.2506571
-0.18360557 0.10072254 0.34102374 0.42282042 -0.72877145 0.8098361
1.7569996 -0.10220309 0.2569473 -0.08877992 -1.2237244 0.4090685
0.13755932 -0.2037213 -0.69866693 -1.0775353 0.77613497 0.50650656
-0.07069746 -0.29698393 -0.05038543 -0.2311658 0.5859577 -0.6676819
0.6596585 -0.99992025 0.08087035 -0.23279037 -0.25043342 0.33604214
-0.5192051 0.9514715 -0.5590809 0.6000754 0.25111008 -0.27554134
-0.2944508 0.8479299 0.18135516 -0.34021404 0.36101332 0.26449648
-0.47856194 -0.5580331 -1.2419266 0.3266995 0.05791627 0.832894
0.82579553 -1.9809418 -0.7809314 0.34570202 0.3797105 -0.40088758
0.93789464 0.89880294 0.40719077 0.25155595 -0.345876 -0.33506867
-0.89675283 1.0595533 0.15143469 -0.66885144 -0.16449606 0.96446395
0.52090895 -0.9814409 -0.08319215 -0.22797695 -0.50264126 0.44464436
0.74387044 0.2512788 0.20711713 -0.754226 0.08868442 0.586623
-0.5141698 0.22622865 1.9441582 -0.37529266 -0.41812226 0.5549303
1.3928401 -0.08846126 -0.9587658 -0.6125978 -0.77294856 -0.71278083
0.41544107 -0.78563523 -1.6520349 0.9599971 0.17927733 0.16096345
1.0721673 -0.22079395 1.0769881 0.2963764 0.4100065 -0.9360741
0.46441442 0.5769767 1.1265272 -1.325959 0.33265886 -0.2110617
-1.2238584 0.96149176 0.6016817 -0.6093409 1.2551028 -0.18456347
0.15395582 -0.26076812 -1.3913947 0.28323013 0.73525995 0.49596626
0.4886674 0.42807105 0.12803495 1.3003635 0.07819644 0.06599685
0.21634011 0.39131057 -0.03042634 -1.3771622 -0.5454044 0.6720337
-0.47396758 -0.28979927 -0.12302314 -0.05100625 0.19581394 1.0280327
-0.25023806 -1.1257696 0.37602213 -0.15893145 0.19950368 -0.5903242
-0.86088216 0.5471203 0.09316357 -0.8181917 -0.6937064 -0.53396785
-0.4213052 -0.3770907 -0.17794587 0.3793924 0.70978105 0.8950733
0.73962045 0.30564073 1.1980574 -0.92548305 -0.900859 -1.0384854
-1.1480755 0.6470618 0.41791293 -0.78357786 -0.28111222 0.17032 ] | [8.672783851623535, 4.633760452270508] |
5a4eb508-e1c2-471b-9d2e-bba0be38fc0c | table-based-fact-verification-with-salience | 2109.04053 | null | https://arxiv.org/abs/2109.04053v1 | https://arxiv.org/pdf/2109.04053v1.pdf | Table-based Fact Verification with Salience-aware Learning | Tables provide valuable knowledge that can be used to verify textual statements. While a number of works have considered table-based fact verification, direct alignments of tabular data with tokens in textual statements are rarely available. Moreover, training a generalized fact verification model requires abundant labeled training data. In this paper, we propose a novel system to address these problems. Inspired by counterfactual causality, our system identifies token-level salience in the statement with probing-based salience estimation. Salience estimation allows enhanced learning of fact verification from two perspectives. From one perspective, our system conducts masked salient token prediction to enhance the model for alignment and reasoning between the table and the statement. From the other perspective, our system applies salience-aware data augmentation to generate a more diverse set of training instances by replacing non-salient terms. Experimental results on TabFact show the effective improvement by the proposed salience-aware learning techniques, leading to the new SOTA performance on the benchmark. Our code is publicly available at https://github.com/luka-group/Salience-aware-Learning . | ['Muhao Chen', 'Pedro Szekely', 'Jay Pujara', 'Kexuan Sun', 'Fei Wang'] | 2021-09-09 | null | https://aclanthology.org/2021.findings-emnlp.338 | https://aclanthology.org/2021.findings-emnlp.338.pdf | findings-emnlp-2021-11 | ['table-based-fact-verification'] | ['natural-language-processing'] | [ 2.36765280e-01 3.52691919e-01 -9.11111712e-01 -3.22101414e-01
-1.15503788e+00 -4.39463109e-01 7.23122537e-01 7.81500459e-01
-1.88001655e-02 9.36261952e-01 8.45091581e-01 -3.92598927e-01
8.57390165e-02 -8.77556205e-01 -8.52473140e-01 -4.40149069e-01
3.50642018e-02 2.07066476e-01 1.73081666e-01 -2.15400010e-01
6.43434227e-01 -1.59692481e-01 -1.52326107e+00 7.73793340e-01
9.28728938e-01 9.61460769e-01 -2.36989949e-02 4.48465385e-02
-4.12987620e-01 1.42964733e+00 -3.31938863e-01 -6.96884394e-01
2.81359226e-01 -2.52410918e-01 -7.90338933e-01 -4.80540209e-02
5.64009845e-01 -1.15824707e-01 -3.78811546e-02 1.18142760e+00
3.11171085e-01 6.15022704e-03 3.68023723e-01 -1.35190117e+00
-6.24404550e-01 1.58330774e+00 -8.70801926e-01 3.15619260e-01
5.74251831e-01 -6.11456819e-02 1.61369205e+00 -1.06385183e+00
6.07927024e-01 1.18752468e+00 3.81673843e-01 1.91215068e-01
-8.59019935e-01 -8.90172839e-01 5.50639391e-01 7.51188397e-01
-1.16737545e+00 -4.77558434e-01 1.38012195e+00 -3.16641808e-01
5.18363655e-01 4.79985982e-01 5.37790537e-01 1.05319595e+00
-5.97711243e-02 1.11204910e+00 1.30698013e+00 -4.86341357e-01
2.38315314e-01 4.17911202e-01 7.44398385e-02 5.66231370e-01
5.35957217e-01 -1.94215477e-01 -8.98937464e-01 -1.95106328e-01
2.24588513e-01 -1.29902601e-01 -2.94000089e-01 -4.24372703e-01
-1.54087353e+00 7.29416251e-01 4.07181412e-01 2.34510705e-01
-4.96365488e-01 -6.81850761e-02 7.24509954e-01 4.99407612e-02
4.93116379e-01 5.53012669e-01 -4.86686766e-01 -9.78119224e-02
-1.02322614e+00 4.87404823e-01 6.51298046e-01 8.82354677e-01
4.62418705e-01 -9.35697742e-03 -5.54791868e-01 5.14834404e-01
1.02131620e-01 3.40962380e-01 4.26483929e-01 -4.47999716e-01
1.07433951e+00 8.52078676e-01 2.59856850e-01 -1.40962434e+00
-2.72716641e-01 -6.21917844e-01 -7.14126527e-01 -2.13680908e-01
3.68775845e-01 1.00969948e-01 -4.00859028e-01 1.59416091e+00
6.54427171e-01 1.52104050e-01 1.33181423e-01 9.91926908e-01
7.75421858e-01 4.95778263e-01 3.89857851e-02 -3.20972174e-01
1.63463783e+00 -8.28910232e-01 -1.02067828e+00 -9.90747586e-02
7.95774996e-01 -7.72778928e-01 1.27676070e+00 3.83134425e-01
-8.67688000e-01 -2.86230773e-01 -1.15101027e+00 -4.21659686e-02
-4.04783487e-01 -7.23757921e-03 7.94367552e-01 5.26623547e-01
-3.69814813e-01 1.99798658e-01 -3.31126243e-01 1.87842250e-01
5.64223468e-01 -7.74565190e-02 -7.81549960e-02 6.61642849e-02
-1.63906205e+00 7.41034389e-01 8.09251785e-01 -6.09921627e-02
-6.19779706e-01 -9.41738963e-01 -9.98932242e-01 3.37611258e-01
1.05206680e+00 -4.02446300e-01 1.16164052e+00 -5.75640023e-01
-8.91739666e-01 6.30476534e-01 -2.90075421e-01 -5.46870708e-01
6.24703228e-01 -2.22338736e-01 -4.10997152e-01 -9.61261019e-02
2.75490135e-01 2.77282238e-01 5.87391734e-01 -1.50993168e+00
-6.69946611e-01 -1.77847520e-01 3.02438021e-01 2.19050184e-01
-4.78778988e-01 -1.68847680e-01 -6.47985488e-02 -8.40055346e-01
1.20486133e-01 -4.54612553e-01 -8.30158219e-02 -4.06949848e-01
-8.89052868e-01 -1.92300931e-01 6.07763648e-01 -6.35163188e-01
1.55725157e+00 -1.76197684e+00 -3.61590028e-01 2.08042562e-01
2.63188332e-01 1.65013960e-04 9.82605368e-02 5.42263508e-01
-4.26810235e-01 3.12610358e-01 -1.74040347e-01 -6.58232998e-03
1.63617104e-01 -4.24351186e-01 -7.38283455e-01 2.01739207e-01
2.07479522e-01 8.21406960e-01 -1.08402860e+00 -9.29945290e-01
8.92687589e-02 -9.72665250e-02 -6.40606999e-01 -2.21607070e-02
-5.38254440e-01 1.89828742e-02 -5.23812354e-01 8.48358572e-01
7.53306150e-01 -3.26035440e-01 4.32085156e-01 -3.56985569e-01
-3.29390019e-02 8.56090546e-01 -1.30742967e+00 1.34787583e+00
-4.55000937e-01 3.30458701e-01 -5.12843668e-01 -9.40305531e-01
9.69424307e-01 2.96586186e-01 3.92532825e-01 -6.56319976e-01
1.39621809e-01 9.97961909e-02 -7.51647279e-02 -5.63694417e-01
8.43407333e-01 -2.72193521e-01 -2.78262228e-01 4.44966227e-01
-3.55061859e-01 6.55011162e-02 2.46577173e-01 4.60750014e-01
7.89985955e-01 -2.28763577e-02 8.43297720e-01 -2.36800984e-01
8.49180162e-01 3.46947372e-01 7.75275230e-01 4.67198700e-01
-1.35803163e-01 1.18002154e-01 8.18367660e-01 -4.26052272e-01
-8.68624866e-01 -7.73202896e-01 -8.67644325e-02 8.12762260e-01
2.14592367e-01 -7.09075391e-01 -3.11487764e-01 -9.12582755e-01
2.27703407e-01 1.16106689e+00 -8.57338130e-01 6.83522448e-02
-3.80077869e-01 -5.69675922e-01 1.46824673e-01 4.94646579e-01
6.04861200e-01 -1.03226042e+00 -3.33007395e-01 6.55351067e-03
-6.61881208e-01 -1.08341622e+00 -3.09800237e-01 -4.56921495e-02
-6.87257588e-01 -1.18868589e+00 -5.71981631e-02 -5.95599174e-01
6.80077672e-01 3.08447093e-01 1.08640492e+00 4.48727496e-02
1.37855574e-01 -2.48456076e-01 -4.34064507e-01 -5.80154419e-01
-2.57626176e-01 1.78368017e-03 -4.02417816e-02 -6.00740165e-02
4.44629908e-01 -5.42152345e-01 -3.78742456e-01 -1.48186833e-02
-7.76716888e-01 3.17075759e-01 5.83774149e-01 9.38305557e-01
5.20220816e-01 1.59257606e-01 7.47525156e-01 -1.33423185e+00
6.05325699e-01 -6.76487327e-01 -6.70630038e-01 2.97846943e-01
-6.46989703e-01 2.01323643e-01 7.66161501e-01 -2.82677084e-01
-1.30821168e+00 -1.81574017e-01 1.41353279e-01 -2.29578376e-01
2.07565904e-01 1.00546086e+00 -3.77530813e-01 5.48333347e-01
6.50133848e-01 2.53953755e-01 -5.52982748e-01 -1.07937559e-01
4.65953022e-01 5.08904099e-01 4.69516367e-01 -7.95002162e-01
1.11144972e+00 4.65645075e-01 -1.99540611e-02 -1.38716757e-01
-1.29332745e+00 -4.34476674e-01 -5.46513021e-01 -2.68801123e-01
3.44682604e-01 -9.93050277e-01 -9.44013476e-01 -1.38343647e-01
-8.95164847e-01 -1.46358656e-02 -3.06093931e-01 2.99845427e-01
-3.54996949e-01 3.39841038e-01 -7.82456845e-02 -1.07007527e+00
-3.10305387e-01 -7.08725035e-01 8.00898433e-01 6.38646409e-02
-2.87734836e-01 -9.53704894e-01 6.18743487e-02 6.70791388e-01
2.16449657e-03 5.58957219e-01 9.43546236e-01 -9.36208785e-01
-8.74822140e-01 -3.05953175e-01 -1.80867031e-01 -1.94772869e-01
4.22159761e-01 -1.55895323e-01 -9.00915861e-01 7.33181685e-02
1.36424331e-02 -5.29278398e-01 8.63729775e-01 4.66485396e-02
1.41185308e+00 -9.20499742e-01 -2.51631021e-01 1.11361973e-01
1.46880746e+00 -7.86630809e-02 3.66746247e-01 6.27399862e-01
7.11713731e-01 6.48321152e-01 1.33443952e+00 8.20637941e-01
7.72138059e-01 5.93547463e-01 4.47452486e-01 5.61574735e-02
1.53355241e-01 -6.63287580e-01 2.43828520e-01 7.21432626e-01
2.63150603e-01 -1.76814031e-02 -1.09986246e+00 9.17578697e-01
-1.97930467e+00 -1.34647465e+00 -1.17648765e-01 2.03455305e+00
1.18032765e+00 6.04610443e-01 9.36691090e-02 5.03304958e-01
7.04826176e-01 5.72711647e-01 -3.10859442e-01 -7.15036616e-02
-6.66986555e-02 -3.45698625e-01 2.25208208e-01 4.96760607e-01
-1.12642717e+00 8.32232356e-01 4.69811249e+00 9.68584895e-01
-9.71787274e-01 8.61499384e-02 6.69605672e-01 -2.14337602e-01
-9.08627570e-01 1.31775945e-01 -7.97105908e-01 3.79827201e-01
5.66436052e-01 -7.94534624e-01 -1.30998105e-01 1.06490600e+00
3.30729455e-01 -1.66020766e-01 -1.09282184e+00 8.04021955e-01
3.36128920e-02 -1.68146229e+00 4.16991711e-01 -2.03216314e-01
9.04490590e-01 -5.54146588e-01 1.71112061e-01 3.32468182e-01
2.66174167e-01 -6.39593661e-01 1.02297485e+00 2.74154484e-01
5.23986101e-01 -8.63446593e-01 9.95586812e-01 3.40127259e-01
-1.17148221e+00 4.72474843e-03 -1.99786380e-01 -1.69774055e-01
-1.43334121e-01 9.87516701e-01 -1.20855272e+00 8.19377303e-01
2.47102097e-01 6.93106890e-01 -5.95872641e-01 7.70631492e-01
-4.56979275e-01 7.06972837e-01 1.86430737e-01 -2.79291719e-01
-2.71602664e-02 3.23915064e-01 5.52202106e-01 1.19890475e+00
-1.06575519e-01 1.86834596e-02 3.02786559e-01 9.58949804e-01
-4.32854682e-01 3.93530428e-01 -6.04268193e-01 7.54317418e-02
7.75624156e-01 1.32057977e+00 -6.09495580e-01 -5.85042179e-01
-3.68670046e-01 2.12620258e-01 4.11475986e-01 -5.37579022e-02
-1.02074564e+00 -4.87833172e-02 4.14906055e-01 1.86350226e-01
2.53092706e-01 1.92834660e-01 -8.34793925e-01 -1.45196927e+00
4.61869121e-01 -1.09441149e+00 5.98896503e-01 -5.51773727e-01
-1.11890435e+00 2.38058209e-01 1.80439517e-01 -1.43379569e+00
-1.97849065e-01 -2.11553931e-01 -5.99172890e-01 6.72910631e-01
-1.79596436e+00 -1.09764397e+00 -2.25752875e-01 3.61427039e-01
4.53413397e-01 -2.51428166e-04 5.03124833e-01 6.91321865e-02
-5.08117139e-01 8.09084415e-01 -4.70163763e-01 3.17962974e-01
6.95577681e-01 -1.37142479e+00 4.50437725e-01 1.10285103e+00
1.78848132e-01 8.66126001e-01 9.06885505e-01 -8.98312628e-01
-1.24913907e+00 -1.06418288e+00 1.21904922e+00 -4.73259002e-01
9.34274197e-01 -4.35973495e-01 -1.04605401e+00 6.72341883e-01
2.30332360e-01 -2.00258166e-01 8.24792445e-01 5.60725689e-01
-8.21061909e-01 -2.57282317e-01 -1.11254299e+00 8.89973819e-01
7.40922928e-01 -5.56017935e-01 -8.81086767e-01 4.28668380e-01
6.93203747e-01 -5.83908200e-01 -7.44967401e-01 3.02075773e-01
3.80573630e-01 -7.53894567e-01 7.43124902e-01 -6.21101260e-01
1.05893099e+00 -5.34439147e-01 -1.68980047e-01 -1.04513121e+00
-2.68001825e-01 -4.25614178e-01 -3.22028875e-01 1.58203983e+00
6.92770422e-01 -4.51034755e-01 8.44036520e-01 4.27236944e-01
3.49876322e-02 -8.28398943e-01 -8.56823385e-01 -6.91646338e-01
-1.34064898e-01 -4.57535654e-01 8.77018332e-01 1.48507345e+00
8.50402772e-01 1.95626378e-01 -5.17286479e-01 2.43503377e-01
6.13898039e-01 6.32900715e-01 7.30752170e-01 -8.38539302e-01
-1.50282308e-01 -3.98923129e-01 -1.29861668e-01 -6.03181660e-01
2.24008292e-01 -1.08250690e+00 -2.27443804e-03 -1.46155596e+00
5.95836043e-01 -3.42540503e-01 -4.69968975e-01 6.84056938e-01
-7.14195371e-01 -1.91929787e-01 2.91521639e-01 1.96131412e-02
-7.04955518e-01 6.14030838e-01 1.17462504e+00 -2.87770838e-01
3.70021835e-02 -2.35698014e-01 -1.14916360e+00 6.06723309e-01
1.18679404e+00 -6.19352639e-01 -3.96477133e-01 1.23626240e-01
5.85121214e-01 1.67040601e-01 3.55091035e-01 -8.98976207e-01
2.51496166e-01 -4.94664103e-01 2.08755776e-01 -9.34155166e-01
3.56731899e-02 -7.90737510e-01 -2.40331993e-01 4.27190095e-01
-7.90722311e-01 1.68244299e-02 3.59780222e-01 4.71564263e-01
-3.73402417e-01 -5.84357269e-02 4.49122399e-01 -1.22928865e-01
-6.62481606e-01 -3.75688635e-02 6.37319237e-02 3.17718238e-01
8.16651702e-01 1.55295745e-01 -6.60338283e-01 -3.69493276e-01
-1.69677004e-01 3.71195555e-01 2.95144439e-01 5.06721020e-01
6.87204778e-01 -1.52845097e+00 -8.78934383e-01 -2.47682273e-01
5.38921356e-01 -1.43995762e-01 2.40152016e-01 8.27463627e-01
4.45217751e-02 5.98782241e-01 -1.28926694e-01 -1.76662654e-01
-1.23078609e+00 9.23928618e-01 -1.67668208e-01 -6.94802642e-01
-3.38566363e-01 6.40648961e-01 3.50201614e-02 -3.00618440e-01
6.65569603e-02 -5.48748255e-01 -3.04368585e-01 2.63644040e-01
6.15183532e-01 8.48327726e-02 6.12122566e-02 -2.24626169e-01
-4.33736801e-01 -5.97145967e-02 -3.67834002e-01 -8.38658065e-02
1.19241881e+00 8.00191388e-02 -9.74271446e-02 6.44873381e-01
6.78439558e-01 5.73542774e-01 -7.99650729e-01 -5.64887226e-01
4.62769628e-01 -8.92543316e-01 -2.61500536e-04 -9.18112814e-01
-9.10081565e-01 5.05732656e-01 -2.24121705e-01 2.00101942e-01
1.05356073e+00 -1.54243276e-01 5.68090022e-01 3.91667634e-01
2.79130012e-01 -8.49051535e-01 1.30386069e-01 2.94156492e-01
1.12415779e+00 -1.49159133e+00 2.64436930e-01 -7.63820827e-01
-8.87002349e-01 8.23630154e-01 7.26244032e-01 1.86168969e-01
2.99606681e-01 3.34600896e-01 9.35800001e-02 -2.76911497e-01
-1.13989663e+00 -3.48083004e-02 2.76551187e-01 3.17745477e-01
7.45519221e-01 7.21532777e-02 -4.61018354e-01 7.60348260e-01
-5.85716367e-01 -1.39793590e-01 7.69598544e-01 7.62790322e-01
-3.17835212e-01 -9.09184158e-01 -5.97741127e-01 5.37709773e-01
-3.59796345e-01 -5.68558097e-01 -3.51954341e-01 8.46742272e-01
1.08908694e-02 8.62210095e-01 -3.32537800e-01 -5.11911929e-01
3.38508874e-01 1.83891118e-01 1.27333730e-01 -5.77003598e-01
-6.42482519e-01 -2.39478022e-01 3.94957304e-01 -3.77313375e-01
-4.53393817e-01 -1.12649393e+00 -1.25820208e+00 -4.62464005e-01
-2.45661423e-01 4.15159196e-01 3.37659925e-01 8.54728937e-01
1.82584103e-03 6.55514657e-01 7.22314656e-01 -7.14966357e-02
-5.57353795e-01 -9.04232442e-01 -4.03206825e-01 4.83347982e-01
4.74589139e-01 -7.80684233e-01 -2.32148737e-01 2.70845536e-02] | [9.389863014221191, 7.867068290710449] |
c85f7929-7795-4961-8c4a-161e74d73e30 | cityscapes-panoptic-parts-and-pascal-panoptic | 2004.07944 | null | https://arxiv.org/abs/2004.07944v1 | https://arxiv.org/pdf/2004.07944v1.pdf | Cityscapes-Panoptic-Parts and PASCAL-Panoptic-Parts datasets for Scene Understanding | In this technical report, we present two novel datasets for image scene understanding. Both datasets have annotations compatible with panoptic segmentation and additionally they have part-level labels for selected semantic classes. This report describes the format of the two datasets, the annotation protocols, the merging strategies, and presents the datasets statistics. The datasets labels together with code for processing and visualization will be published at https://github.com/tue-mps/panoptic_parts. | ['Gijs Dubbelman', 'Panagiotis Meletis', 'Xiaoxiao Wen', 'Chenyang Lu', 'Daan de Geus'] | 2020-04-16 | null | null | null | null | ['human-part-segmentation', 'part-level-panoptic-segmentation'] | ['computer-vision', 'computer-vision'] | [ 9.89474878e-02 -1.64768815e-01 -2.57525206e-01 -5.73046684e-01
-2.25546539e-01 -1.07233691e+00 5.14060438e-01 2.28921950e-01
1.53925940e-01 4.32988435e-01 -6.68782517e-02 -3.55536550e-01
-3.40284675e-01 -7.82556593e-01 -1.37202427e-01 -6.29244566e-01
-3.73113781e-01 3.94545525e-01 5.94016254e-01 -2.60131452e-02
3.27692896e-01 6.06109381e-01 -1.76631403e+00 4.85191494e-01
5.22114635e-01 1.25473559e+00 4.93231624e-01 9.54390049e-01
-2.57506341e-01 3.59375954e-01 -5.07777035e-01 6.44683614e-02
6.01338506e-01 -3.43940586e-01 -1.02711129e+00 5.71632564e-01
8.19444656e-01 -9.01757255e-02 -3.12301695e-01 1.00191402e+00
1.48620933e-01 5.55764511e-02 3.94488782e-01 -1.52104032e+00
9.32245888e-03 3.69610548e-01 -4.55894947e-01 3.95671636e-01
1.34659067e-01 9.27392915e-02 7.71241784e-01 -4.59678173e-01
1.02277446e+00 8.60759616e-01 7.91317523e-01 -9.55752954e-02
-8.68145525e-01 -6.16412282e-01 -1.42636254e-01 1.23702101e-02
-1.41280532e+00 -2.90629864e-01 3.77298743e-01 -7.33474672e-01
7.96580076e-01 7.97253013e-01 8.67182732e-01 2.75966197e-01
-1.05926573e-01 5.44701278e-01 1.49777532e+00 -2.67146200e-01
-8.49228259e-03 2.10217699e-01 4.97191608e-01 6.57624066e-01
3.80342782e-01 -2.50946939e-01 -1.62873611e-01 -8.71785544e-03
7.16988504e-01 -3.71090710e-01 -1.07849896e-01 -3.13152075e-01
-1.02229607e+00 5.75671911e-01 4.63954091e-01 3.83877665e-01
-1.37897477e-01 -1.55855995e-02 4.59376037e-01 -2.49516964e-02
5.82694232e-01 3.62292260e-01 -7.60956109e-01 1.90083101e-01
-1.05216610e+00 1.43970266e-01 4.99522567e-01 1.31757379e+00
9.14910674e-01 -9.26984623e-02 3.07730228e-01 8.85519028e-01
2.53881723e-01 5.02103448e-01 -1.15445912e-01 -1.54392874e+00
1.98978841e-01 5.52400351e-01 2.55623311e-02 -1.03724968e+00
-8.17321360e-01 -8.49680379e-02 -3.65440100e-01 1.21686757e-01
2.77661353e-01 -1.39096500e-02 -1.34757602e+00 8.68898690e-01
4.65901375e-01 -4.31618579e-02 -1.15037821e-01 6.23861492e-01
1.50677836e+00 7.64786720e-01 2.39878073e-01 3.18902209e-02
1.83599854e+00 -9.09105420e-01 -9.06083405e-01 -2.25404520e-02
3.77194911e-01 -1.15208411e+00 8.34679365e-01 1.54364198e-01
-7.84906507e-01 -5.55072069e-01 -7.11050749e-01 -9.03974399e-02
-1.12311924e+00 2.64286995e-01 6.19420946e-01 5.86508512e-01
-1.12461317e+00 3.65067244e-01 -7.50739098e-01 -1.11008334e+00
5.70707262e-01 -7.03008398e-02 -2.01454639e-01 3.12340736e-01
-8.14102650e-01 5.60959458e-01 8.29818726e-01 -1.06762640e-01
-6.43430829e-01 -4.64584798e-01 -6.92987561e-01 -3.78152430e-01
2.53748268e-01 -3.61794502e-01 1.39467859e+00 -2.94552594e-01
-1.17408347e+00 1.32848418e+00 8.34095031e-02 -2.43388712e-01
2.18230277e-01 -8.64784978e-03 -5.72569311e-01 7.14045107e-01
2.81495482e-01 1.17184091e+00 3.63417953e-01 -1.60977185e+00
-1.16498232e+00 -1.70471251e-01 2.31760278e-01 1.79783374e-01
2.72186399e-01 2.61536539e-01 -8.18333566e-01 -5.90608418e-01
3.14861894e-01 -7.72953153e-01 -2.43751228e-01 1.21372260e-01
-4.73492205e-01 1.73424333e-01 1.12985468e+00 -7.93947995e-01
1.08208060e+00 -2.24609542e+00 -6.69008076e-01 7.62154907e-02
-2.45355852e-02 2.36430928e-01 -7.66109824e-02 5.56248486e-01
-9.91929770e-02 5.13778985e-01 -7.57376969e-01 -2.85929531e-01
-3.08974504e-01 4.20393556e-01 -4.65027809e-01 5.60130119e-01
-3.41261327e-01 4.76258755e-01 -6.33097470e-01 -6.08981252e-01
8.59890103e-01 1.20086588e-01 8.96438509e-02 1.06689118e-01
-3.48218948e-01 7.89289653e-01 -2.33712927e-01 8.72950435e-01
1.34313035e+00 1.23372534e-02 1.92162231e-01 -3.18230003e-01
-7.93755531e-01 2.00541422e-01 -1.27262080e+00 1.60788846e+00
-1.01937056e-01 1.00262117e+00 4.13098842e-01 -6.71046436e-01
7.72003174e-01 2.77918160e-01 6.38998210e-01 -5.36835253e-01
8.96146148e-02 2.74901092e-01 -5.37661552e-01 -6.16154253e-01
8.17708135e-01 4.32035625e-01 4.86504845e-02 1.12962447e-01
1.46495089e-01 -8.53088915e-01 9.32145417e-01 1.49400920e-01
2.89980829e-01 3.75753015e-01 4.34700787e-01 -6.75711572e-01
3.07721406e-01 8.56805086e-01 5.05683720e-01 6.45857275e-01
-1.80416703e-01 8.36307883e-01 4.55277741e-01 -4.97978866e-01
-1.00873435e+00 -9.74576354e-01 -8.17696929e-01 7.63725817e-01
4.75526661e-01 -6.64088845e-01 -6.85875952e-01 -2.24386975e-01
-1.79322705e-01 6.63723350e-01 -6.15925431e-01 8.12932432e-01
-3.62119973e-01 -9.10521686e-01 6.05403364e-01 3.15538079e-01
7.77591527e-01 -9.99552548e-01 -9.91376102e-01 -4.03952688e-01
-3.06578398e-01 -1.28991568e+00 2.11530745e-01 1.91895530e-01
-9.52187836e-01 -1.49057019e+00 -2.30899423e-01 -6.85520589e-01
4.62229162e-01 4.04746920e-01 1.13903654e+00 6.25128448e-02
-6.62273765e-01 3.26797515e-01 -5.73493481e-01 -4.22618747e-01
6.13847040e-02 1.20994814e-01 -4.97861028e-01 -4.93102908e-01
7.87516162e-02 -2.77947336e-01 -6.69627547e-01 4.70895737e-01
-1.22866416e+00 3.74209166e-01 -8.32945406e-02 -1.83189645e-01
1.02367330e+00 1.26844883e-01 -3.22481215e-01 -1.03568220e+00
-8.55481997e-03 -5.26921093e-01 -1.25648320e+00 2.70387679e-01
-1.94061637e-01 -6.09813511e-01 -5.63335381e-02 5.54668725e-01
-1.09479058e+00 2.85741657e-01 -1.09342568e-01 7.30137378e-02
-9.99440610e-01 1.71726763e-01 -2.62674540e-02 1.86099291e-01
4.16926026e-01 -1.83166698e-01 -3.98888886e-01 -9.99703467e-01
6.95440054e-01 8.96613181e-01 6.13717675e-01 -2.68146545e-01
4.67611343e-01 1.07254684e+00 -1.14131691e-02 -1.26066709e+00
-9.52480257e-01 -1.13980341e+00 -1.10698855e+00 -5.68522990e-01
1.11394083e+00 -7.00113177e-01 5.57265542e-02 7.87235916e-01
-9.61555541e-01 -5.21382511e-01 -5.83789289e-01 4.12117720e-01
-4.70985949e-01 1.57114416e-01 -4.79883492e-01 -5.09526432e-01
-1.45417452e-01 -1.13299823e+00 1.06303489e+00 6.72022164e-01
-1.14483587e-01 -1.14016116e+00 2.30918258e-01 4.21589911e-01
3.48434716e-01 5.07777512e-01 3.33529204e-01 -3.86627436e-01
-3.98383081e-01 8.17245767e-02 -6.12308860e-01 7.84603804e-02
2.15263590e-01 5.69998860e-01 -1.14714921e+00 1.32217661e-01
-1.75208822e-01 -1.77838752e-04 9.62981701e-01 8.34406376e-01
1.30418897e+00 -1.34351507e-01 -6.09892190e-01 9.03798997e-01
1.57267725e+00 3.36079419e-01 5.50379097e-01 4.45474923e-01
4.53977168e-01 1.14247859e+00 8.09057355e-01 5.40710151e-01
2.20617071e-01 5.06883204e-01 5.55608928e-01 -8.34377408e-02
-5.07762134e-01 2.33315185e-01 -3.04448545e-01 7.41611063e-01
-4.98361029e-02 -5.17379105e-01 -1.37557626e+00 7.21099079e-01
-1.67016017e+00 -6.04354322e-01 -1.14998198e+00 1.83114135e+00
3.84410650e-01 -4.78173524e-01 1.31478965e-01 -1.88497767e-01
7.45908916e-01 5.46982348e-01 -2.74558067e-02 -3.49440336e-01
-4.94998097e-01 1.22070335e-01 9.08339977e-01 6.71565235e-01
-1.72431731e+00 1.09687030e+00 7.31112337e+00 6.74333930e-01
-9.22230840e-01 3.51949781e-01 4.16300833e-01 2.60347903e-01
1.65415481e-01 4.20471340e-01 -9.68368232e-01 3.06194335e-01
1.03926492e+00 2.06756204e-01 -1.96557189e-03 4.75715101e-01
3.31104875e-01 -7.59986162e-01 -7.94659555e-02 5.43952048e-01
-1.84155807e-01 -1.46098149e+00 7.71371126e-02 -1.21472152e-02
8.08826149e-01 6.16635919e-01 -2.44972467e-01 -5.56596398e-01
2.61990130e-01 -6.23270988e-01 7.48890877e-01 3.91253978e-01
9.19557333e-01 -2.93990135e-01 7.56299913e-01 -3.85584891e-01
-1.53001487e+00 9.46248472e-02 -3.97564441e-01 -1.32758785e-02
4.12024468e-01 5.76711595e-01 -3.13077867e-01 1.14500868e+00
1.23359668e+00 9.26566660e-01 -1.11874926e+00 1.51303232e+00
-3.88756990e-01 6.46064699e-01 -7.64408231e-01 7.72132695e-01
2.57644445e-01 -5.12762606e-01 4.29971278e-01 1.62713325e+00
4.26236331e-01 4.10578437e-02 3.71465266e-01 3.73607099e-01
2.33677030e-01 2.25003108e-01 -6.14569247e-01 -1.07377551e-01
5.29307365e-01 1.57350028e+00 -1.60621035e+00 -6.60594046e-01
-2.35709667e-01 5.87479353e-01 -5.07358730e-01 2.59119630e-01
-6.36937439e-01 -6.65423512e-01 7.47634113e-01 2.06329823e-01
1.13600329e-01 -3.49048138e-01 -7.26490200e-01 -7.90743351e-01
-3.13150376e-01 -2.52901226e-01 8.75235319e-01 -1.27375138e+00
-8.48697722e-01 6.53657079e-01 7.24174976e-01 -1.40198791e+00
3.70776623e-01 -7.20598519e-01 -2.87997365e-01 3.24074239e-01
-1.11370409e+00 -1.37439048e+00 -7.70256937e-01 1.78646043e-01
5.33068836e-01 2.03776523e-01 7.39734471e-01 3.83752495e-01
-5.56112528e-01 -4.51722533e-01 3.23964983e-01 8.67510661e-02
4.11236942e-01 -1.15681589e+00 3.34280461e-01 8.13795924e-01
2.28771642e-01 5.96743412e-02 8.48061144e-01 -6.23349190e-01
-4.49417114e-01 -1.19913471e+00 6.04294360e-01 -3.03361952e-01
8.16329420e-01 -1.72432140e-01 -7.41129160e-01 9.02475178e-01
9.07264650e-01 -1.12742878e-01 6.96505547e-01 -4.29938763e-01
-1.42733390e-02 -4.46834862e-02 -1.07000613e+00 1.92246526e-01
9.64451134e-01 -2.90566325e-01 -3.02897930e-01 9.56148684e-01
4.68931258e-01 -5.97385049e-01 -8.24626148e-01 4.99499321e-01
3.85712296e-01 -9.08122897e-01 6.98890686e-01 2.89179608e-02
1.72340900e-01 -9.19847608e-01 -2.64324963e-01 -8.95565212e-01
-3.58185824e-03 -2.03284502e-01 3.99820507e-01 1.29434180e+00
2.91496813e-01 -5.26561081e-01 3.13349664e-01 -1.53224513e-01
-4.83275354e-01 -2.18866527e-01 -7.05265105e-01 -7.96700001e-01
-7.51573890e-02 -7.75497139e-01 5.14566004e-01 1.00313640e+00
-3.94535422e-01 -1.11461066e-01 -5.59142493e-02 4.74510401e-01
5.87566972e-01 5.12848854e-01 6.59686565e-01 -1.22928309e+00
3.37571681e-01 -6.90699160e-01 -2.91413188e-01 -6.17316306e-01
-4.74397928e-01 -8.16850483e-01 -1.52526051e-01 -2.13000464e+00
-9.04497951e-02 -5.68612874e-01 1.47199944e-01 7.42619336e-01
4.71846282e-01 9.68127847e-01 1.34786040e-01 5.66605806e-01
-5.54591358e-01 -1.29119545e-01 7.96061814e-01 2.23453809e-02
-1.47164509e-01 -1.92285791e-01 -1.41062275e-01 9.12859321e-01
1.95181179e+00 -5.18996239e-01 -2.21905470e-01 -4.62012470e-01
-1.10895839e-02 -7.49343574e-01 2.85487801e-01 -1.23870790e+00
-5.33789210e-02 -2.66366273e-01 -9.77102965e-02 -1.11408424e+00
2.60392904e-01 -8.21418047e-01 6.75349414e-01 4.78080630e-01
5.45015819e-02 1.71611030e-02 7.03503013e-01 7.19412267e-02
-4.75178152e-01 -4.79751825e-01 1.03804898e+00 -3.20575744e-01
-1.20549393e+00 1.00152098e-01 -7.00794578e-01 -1.67837769e-01
1.26629448e+00 -2.57178605e-01 -6.96516871e-01 4.22273427e-02
-1.13163853e+00 5.85159481e-01 7.88100302e-01 3.26043278e-01
2.81289369e-02 -9.12286758e-01 -5.65745115e-01 2.82255784e-02
3.33503842e-01 -1.27061298e-02 5.06666780e-01 1.01908636e+00
-1.63741851e+00 6.45314336e-01 -5.85769892e-01 -6.86091721e-01
-1.58561254e+00 2.17549160e-01 4.99353021e-01 9.45985317e-02
-6.36821091e-01 5.37959337e-01 2.56825238e-01 -6.21459007e-01
-2.18743891e-01 -3.58370394e-01 -4.80634958e-01 5.01405597e-01
3.18907231e-01 5.66998720e-01 3.21516097e-02 -8.34587812e-01
-5.79278827e-01 8.70452404e-01 6.08146012e-01 -6.44826591e-02
1.25443339e+00 -6.27518773e-01 -5.91934860e-01 6.86050594e-01
8.65636289e-01 -8.19236785e-02 -1.02177191e+00 -2.46904846e-02
2.22734377e-01 -5.93680918e-01 -1.26106083e-01 -9.52879786e-01
-1.37977815e+00 9.39970255e-01 9.59478021e-01 4.94223297e-01
1.20617008e+00 4.83253807e-01 4.73737687e-01 -5.98986074e-02
3.18038426e-02 -1.16716552e+00 -7.12913811e-01 6.47120833e-01
9.16735947e-01 -9.38259006e-01 4.33900297e-01 -1.11899972e+00
-5.01378179e-01 1.04896772e+00 3.14797878e-01 1.88777059e-01
1.09946287e+00 3.83581638e-01 5.90138018e-01 -6.94228172e-01
-2.45152026e-01 -6.80844009e-01 2.37225488e-01 8.08762848e-01
4.89587933e-01 3.17047894e-01 -3.81186962e-01 -2.36211166e-01
-3.86333495e-01 -2.63552725e-01 6.16744816e-01 9.78608608e-01
-6.41855717e-01 -9.49499488e-01 -4.41652507e-01 4.28540260e-01
-3.01789522e-01 2.78741643e-02 -7.06660926e-01 1.01251364e+00
6.48796678e-01 1.03461301e+00 7.15725124e-01 -1.29588217e-01
2.43466407e-01 -2.51604803e-02 1.46628916e-01 -6.20131135e-01
-3.85967225e-01 2.21715629e-01 5.04548967e-01 -5.96546590e-01
-1.07835686e+00 -7.16223955e-01 -1.30803800e+00 -3.41007203e-01
-2.48664860e-02 3.58557880e-01 7.46790290e-01 2.21513301e-01
4.39806819e-01 3.40158612e-01 2.13651001e-01 -8.66883993e-01
7.51169920e-01 -1.14794147e+00 -8.45164955e-01 1.14399940e-01
-3.65148559e-02 -5.15472829e-01 -4.21956956e-01 5.58141708e-01] | [8.974746704101562, -1.4717501401901245] |
4d72e50a-cc25-4dfa-9980-2b061d89613b | hierarchical-graph-generation-with-k-2-trees | 2305.19125 | null | https://arxiv.org/abs/2305.19125v2 | https://arxiv.org/pdf/2305.19125v2.pdf | Hierarchical Graph Generation with $K^2$-trees | Generating graphs from a target distribution is a significant challenge across many domains, including drug discovery and social network analysis. In this work, we introduce a novel graph generation method leveraging $K^2$-tree representation which was originally designed for lossless graph compression. Our motivation stems from the ability of the $K^2$-trees to enable compact generation while concurrently capturing the inherent hierarchical structure of a graph. In addition, we make further contributions by (1) presenting a sequential $K^2$-tree representation that incorporates pruning, flattening, and tokenization processes and (2) introducing a Transformer-based architecture designed to generate the sequence by incorporating a specialized tree positional encoding scheme. Finally, we extensively evaluate our algorithm on four general and two molecular graph datasets to confirm its superiority for graph generation. | ['Sungsoo Ahn', 'Dongwoo Kim', 'Yunhui Jang'] | 2023-05-30 | null | null | null | null | ['drug-discovery'] | ['medical'] | [ 8.41672003e-01 5.04721999e-01 -1.72979563e-01 -4.09645736e-02
-5.72878301e-01 -4.59477007e-01 2.09394261e-01 7.05194056e-01
-3.49683538e-02 9.73809183e-01 -8.88415799e-03 -7.13722944e-01
-1.76577494e-01 -1.35132539e+00 -5.47773063e-01 -4.01890010e-01
-6.73748016e-01 2.62799710e-01 2.72316992e-01 -2.29422256e-01
3.51596415e-01 5.06009042e-01 -1.13496661e+00 1.54516771e-01
7.74231315e-01 7.08006620e-01 7.01152533e-02 7.62746215e-01
-1.84013546e-01 7.51874387e-01 -5.13434768e-01 -8.99813175e-01
2.96396047e-01 -7.14485288e-01 -1.06015420e+00 -2.55453680e-02
3.12606573e-01 -1.00388259e-01 -5.68039119e-01 7.69303083e-01
5.88519633e-01 -3.76955383e-02 4.92501915e-01 -9.23946679e-01
-1.47791848e-01 8.33111942e-01 -4.71218914e-01 1.70999169e-01
5.61871350e-01 7.79444724e-02 1.21964777e+00 -1.98027179e-01
9.33716178e-01 1.01464820e+00 7.28357196e-01 3.77608776e-01
-1.23540294e+00 -5.59695780e-01 -1.15578182e-01 2.02030689e-01
-1.50232446e+00 -2.90248901e-01 8.82931232e-01 -1.02375247e-01
1.28268886e+00 3.60156953e-01 8.41374278e-01 5.14162838e-01
3.92817378e-01 3.55530113e-01 9.15813863e-01 -4.69592273e-01
3.00323129e-01 -3.90476674e-01 -3.12009025e-02 1.10670280e+00
4.88591880e-01 -1.21210314e-01 -4.30370718e-01 -5.84904611e-01
5.28591931e-01 -3.65244478e-01 -1.79296255e-01 -3.33759606e-01
-6.87065244e-01 9.06535268e-01 4.12856311e-01 -5.01910411e-03
-3.17674875e-01 6.27309978e-01 7.13418305e-01 2.44101226e-01
3.40322793e-01 2.95378864e-01 -3.52807529e-02 -3.60593013e-02
-1.03179979e+00 2.14777440e-01 7.99856842e-01 1.36220038e+00
7.67277598e-01 1.46208346e-01 -2.84314722e-01 4.13772672e-01
8.38899240e-02 -1.83670036e-02 2.67125398e-01 -5.97844541e-01
5.14998496e-01 7.83128142e-01 -5.22487402e-01 -1.20881712e+00
-1.74202353e-01 -4.27053183e-01 -1.14577472e+00 -3.32030475e-01
-3.48281674e-02 1.75837949e-01 -9.62477148e-01 1.70561409e+00
4.48772371e-01 5.08865230e-02 -7.78389126e-02 1.09323584e-01
8.19719076e-01 2.83821523e-01 5.65484874e-02 -5.19341946e-01
1.46377754e+00 -6.55363023e-01 -5.28103828e-01 1.97644785e-01
7.84602046e-01 -6.11279368e-01 6.61704957e-01 1.49817541e-01
-1.38838649e+00 -1.98442027e-01 -8.99384201e-01 -1.76187456e-01
-3.26655865e-01 -4.43055034e-01 9.92482305e-01 9.04906273e-01
-1.26022863e+00 8.11857581e-01 -5.14743984e-01 -2.88241982e-01
5.38325071e-01 4.58615065e-01 -4.11312073e-01 -2.30122238e-01
-9.55471694e-01 4.17308748e-01 5.58111012e-01 -3.02076340e-01
-7.33694255e-01 -4.49770808e-01 -1.08156645e+00 1.19734652e-01
3.12577397e-01 -9.88980293e-01 7.77430415e-01 -1.41607717e-01
-1.29241586e+00 6.98868215e-01 -3.87017310e-01 -9.11890090e-01
1.75084174e-01 5.19025743e-01 -4.09528501e-02 3.50696921e-01
-1.76117316e-01 4.52101171e-01 5.53800344e-01 -8.29104543e-01
-1.82913944e-01 -3.77308488e-01 -6.81345239e-02 1.78117469e-01
-1.97632805e-01 -2.16612756e-01 -3.78520340e-01 -1.05291176e+00
7.12459683e-02 -7.72955537e-01 -6.91429853e-01 -3.48324150e-01
-7.71763146e-01 -2.52408977e-03 6.48853719e-01 -6.00928843e-01
1.85980666e+00 -1.82540011e+00 1.32561937e-01 6.85492039e-01
7.51195610e-01 3.06853503e-01 -1.98974147e-01 1.11538804e+00
-1.21364847e-01 3.05534631e-01 -6.09995604e-01 -1.64736509e-02
-2.33825639e-01 1.48405135e-01 -2.48664021e-01 2.14319959e-01
6.71645403e-02 1.02906740e+00 -8.12852323e-01 -6.64363861e-01
-1.49318308e-01 2.50715494e-01 -6.81292951e-01 -7.28669465e-02
-2.49555409e-01 -1.55633450e-01 -3.21716070e-01 6.07439339e-01
7.20553398e-01 -3.90320987e-01 8.43375862e-01 -2.22181708e-01
3.03954422e-01 3.95344913e-01 -9.07519996e-01 1.49185109e+00
5.51882833e-02 7.26162046e-02 -3.46339904e-02 -1.03126180e+00
1.06925905e+00 1.44688874e-01 6.10527098e-01 -5.56491017e-01
9.05368775e-02 1.25220209e-01 4.27952036e-02 9.84971747e-02
7.55469322e-01 -2.42879972e-01 -2.05485955e-01 6.07520759e-01
-1.30023837e-01 -4.98327464e-01 4.81392413e-01 7.11005092e-01
1.79725456e+00 -2.89850980e-01 4.97308820e-01 -4.20110598e-02
4.99137521e-01 1.20010898e-02 3.25956643e-01 6.53580427e-01
-8.61687437e-02 3.12458545e-01 6.65062904e-01 -4.73616362e-01
-1.11195946e+00 -8.21090341e-01 1.44537359e-01 5.31776726e-01
-2.23696083e-01 -1.17202616e+00 -6.61366343e-01 -8.23026180e-01
-8.45557377e-02 5.03279090e-01 -3.79256010e-01 -2.94289619e-01
-6.63044333e-01 -8.87472928e-01 8.80628824e-01 1.94706917e-01
3.61832023e-01 -7.64724433e-01 -4.14523512e-01 5.86024642e-01
-1.65129408e-01 -1.00378680e+00 -6.72728837e-01 2.08482355e-01
-1.08409333e+00 -1.29031265e+00 -3.11706841e-01 -8.73990357e-01
7.84567535e-01 3.18106890e-01 1.17411065e+00 2.71293163e-01
-6.32493436e-01 2.02975273e-01 -4.17825609e-01 -1.53217301e-01
-5.88833630e-01 3.83573025e-01 -3.67839247e-01 -5.63282251e-01
6.29440397e-02 -1.00999475e+00 -5.28478384e-01 -1.88524351e-01
-1.21902537e+00 1.43255785e-01 3.32851350e-01 7.37857819e-01
8.65305007e-01 5.84125936e-01 4.77619797e-01 -1.30908930e+00
9.91849244e-01 -4.09189910e-01 -5.60297191e-01 4.51781973e-02
-8.81571710e-01 2.44672745e-01 8.84441376e-01 9.91994664e-02
-4.47550446e-01 2.43109703e-01 -3.87785822e-01 1.27567984e-02
4.43165272e-01 5.89950860e-01 1.24554284e-01 -2.96927124e-01
4.80165243e-01 5.59269309e-01 3.02543521e-01 -1.98621005e-01
5.23379624e-01 2.81287402e-01 3.22719485e-01 -4.33398545e-01
7.36091733e-01 4.44932699e-01 6.01172268e-01 -6.97152257e-01
-5.20449653e-02 -2.27828220e-01 -3.82952780e-01 1.18894830e-01
2.41568178e-01 -6.55274987e-01 -8.16519916e-01 3.48942250e-01
-9.85979319e-01 -3.03322971e-01 -3.85914117e-01 3.38192955e-02
-5.83399594e-01 1.02258897e+00 -8.34406376e-01 -5.40684760e-01
-9.53254640e-01 -9.79524910e-01 8.71080995e-01 -3.55620772e-01
-2.53548861e-01 -9.04133558e-01 2.35935956e-01 1.22138470e-01
4.13742572e-01 6.69544101e-01 1.42240131e+00 -3.64620149e-01
-7.85762966e-01 -9.46955904e-02 -3.09474736e-01 -6.24855608e-02
2.32388392e-01 -1.09031029e-01 -2.99076140e-01 -5.71675062e-01
-5.70417821e-01 -2.43701026e-01 7.78108180e-01 1.42768070e-01
1.24668801e+00 -6.44452035e-01 -5.77959657e-01 6.68559432e-01
1.60259271e+00 1.16012260e-01 8.69906723e-01 -2.85070129e-02
6.06906295e-01 3.24573606e-01 1.49884960e-02 5.87886631e-01
4.99622703e-01 7.07187712e-01 3.60109329e-01 -6.91109523e-02
-4.43881541e-01 -5.64167142e-01 1.25334933e-01 1.14998269e+00
1.87416092e-01 -3.43615711e-01 -6.12773001e-01 6.10300064e-01
-1.44333959e+00 -9.41499829e-01 -2.62638390e-01 2.18433738e+00
1.16230595e+00 3.80406231e-02 2.98245400e-01 3.75314921e-01
6.45968676e-01 1.32475346e-01 -4.46156532e-01 -7.81093001e-01
-7.25337490e-02 1.03154016e+00 7.60106325e-01 5.24138093e-01
-5.48362613e-01 1.01447296e+00 7.10718966e+00 1.19425511e+00
-8.30807388e-01 -3.70119452e-01 5.10842085e-01 2.09178045e-01
-6.89989507e-01 1.76159188e-01 -7.10032880e-01 3.73734504e-01
1.01786888e+00 -7.60375917e-01 4.25935268e-01 5.50696254e-01
-2.84851920e-02 2.38869503e-01 -6.48069322e-01 7.99076498e-01
-1.86692148e-01 -1.73694897e+00 5.27719319e-01 3.65078181e-01
4.11822230e-01 -2.71745294e-01 -2.13738665e-01 -9.13971290e-02
6.66367769e-01 -1.19009423e+00 1.37685835e-01 -9.50001553e-02
1.22569561e+00 -8.55418921e-01 3.23673964e-01 1.00445837e-01
-1.63248980e+00 2.42296964e-01 -3.08164805e-01 3.66691612e-02
3.00687611e-01 8.76125276e-01 -1.40128577e+00 1.07069743e+00
3.21582049e-01 4.96145904e-01 -5.11542976e-01 9.51422930e-01
-1.55501096e-02 5.95756948e-01 -3.93359423e-01 2.91414782e-02
9.86229032e-02 -1.18576072e-01 5.34020066e-01 1.38226223e+00
3.88713479e-01 2.51306146e-01 -4.33944725e-02 4.81618881e-01
-3.88082802e-01 2.13423550e-01 -1.10260916e+00 -2.73518711e-01
7.33379602e-01 9.26788151e-01 -7.59125292e-01 -2.39164546e-01
-6.51750788e-02 9.21405017e-01 5.30406237e-01 -4.25492190e-02
-6.31030858e-01 -9.65006709e-01 3.71347308e-01 2.32792288e-01
5.53909481e-01 -5.25667906e-01 -6.87591061e-02 -8.04174602e-01
-1.15192391e-01 -1.02625155e+00 6.75060689e-01 -2.49833673e-01
-9.63923573e-01 8.59849393e-01 1.04466900e-02 -7.44419992e-01
-2.46569738e-01 -2.05705121e-01 -3.77617240e-01 7.89447308e-01
-1.49477112e+00 -8.75757873e-01 -1.37690902e-01 6.07578695e-01
-1.98541563e-02 3.54792289e-02 9.04924214e-01 3.40799987e-01
-4.70984310e-01 9.27275419e-01 -2.10703351e-02 -2.04759449e-01
9.29107219e-02 -1.04836798e+00 9.27655697e-01 7.88325369e-01
-9.35511515e-02 7.55485356e-01 5.01748681e-01 -8.53941202e-01
-1.69471192e+00 -1.37048197e+00 9.62213337e-01 1.75682560e-01
3.39767307e-01 -3.45453471e-01 -7.31924772e-01 5.65136313e-01
2.90832203e-02 -2.72739679e-01 9.16024148e-01 -3.67490798e-01
-3.42553258e-01 -8.12197030e-02 -1.59197187e+00 7.17323959e-01
1.49636233e+00 -3.03083420e-01 1.04531929e-01 3.84616137e-01
8.11303318e-01 -4.41147953e-01 -1.21602929e+00 4.40151870e-01
4.29762304e-01 -7.80612648e-01 9.30753946e-01 -4.72941071e-01
3.95241201e-01 -2.35115588e-01 -7.55976811e-02 -9.64065075e-01
-5.77382386e-01 -1.22482157e+00 -2.80420572e-01 9.36057329e-01
2.74956971e-01 -6.53637350e-01 1.30911839e+00 5.99323437e-02
1.75579470e-02 -9.61597025e-01 -1.07282865e+00 -6.13376498e-01
-2.48682499e-01 -2.64559001e-01 8.18858027e-01 6.73529446e-01
2.55343348e-01 5.62727273e-01 -3.95240426e-01 -2.79056907e-01
7.81748056e-01 9.17016342e-02 9.39432502e-01 -7.37277567e-01
-4.88935620e-01 -4.04969513e-01 -6.15705431e-01 -9.93981302e-01
-1.95170090e-01 -1.31507611e+00 -3.55862856e-01 -1.56489110e+00
2.93179274e-01 -5.51123798e-01 7.02920929e-02 4.56773013e-01
-1.01489097e-01 3.46131772e-01 1.03919171e-01 1.78977698e-02
-3.79510313e-01 6.80189908e-01 1.33635068e+00 -9.72667430e-03
-4.72012535e-02 -1.42034397e-01 -9.63548720e-01 1.12548977e-01
8.94000590e-01 -4.22946781e-01 -8.25661838e-01 -1.50480762e-01
2.60555238e-01 2.93775558e-01 3.02105234e-03 -6.55811369e-01
3.67464982e-02 -7.02452939e-03 -1.95234686e-01 -5.60231924e-01
1.88516319e-01 -5.38507998e-01 5.65617681e-01 8.36472690e-01
-1.66634455e-01 4.34398949e-01 2.86512762e-01 8.07132423e-01
-1.09938845e-01 -7.26239160e-02 7.56469011e-01 -3.85853499e-01
-2.69713491e-01 6.73877418e-01 -1.47376269e-01 6.59737512e-02
1.13461435e+00 -5.64913571e-01 -3.64426494e-01 -3.84010881e-01
-4.47764784e-01 7.87017345e-02 6.00240588e-01 -2.35259056e-01
8.69145811e-01 -1.16820610e+00 -6.85722589e-01 2.06137806e-01
1.01999957e-02 6.16659932e-02 5.79830213e-03 3.78571898e-01
-8.82599354e-01 4.02834445e-01 1.43130794e-02 -9.66967270e-02
-1.62823415e+00 6.17565751e-01 -4.98813298e-03 -8.38203549e-01
-6.29279554e-01 7.99018979e-01 -1.04544461e-01 -2.43190333e-01
4.42659259e-02 -1.04064025e-01 2.43492588e-01 -4.21874672e-01
3.20492923e-01 4.56262589e-01 2.01191455e-01 -3.07073563e-01
-3.63404185e-01 4.13546622e-01 -2.72718668e-01 1.19148947e-01
1.22656131e+00 4.46723290e-02 -5.48066914e-01 -4.40774500e-01
1.20501411e+00 1.17168002e-01 -4.58024740e-01 -1.21049970e-01
-1.35612309e-01 -2.92233586e-01 -2.44883224e-01 -3.65141720e-01
-1.00430405e+00 4.66155529e-01 -5.74618652e-02 4.61313248e-01
1.26093698e+00 -2.41427705e-01 1.30372131e+00 4.69835624e-02
6.34290040e-01 -7.22657263e-01 -2.52726357e-02 3.90970141e-01
4.80553925e-01 -2.96190023e-01 4.24407303e-01 -9.18507516e-01
-2.36082688e-01 1.00061584e+00 1.34473011e-01 1.20688215e-01
4.74572212e-01 2.43569300e-01 -6.94668829e-01 -5.66994429e-01
-8.61494601e-01 -1.44651487e-01 -1.35134384e-01 8.81530762e-01
5.06580710e-01 9.55463722e-02 -8.40618312e-01 1.05800927e-01
-4.37638909e-01 2.13323489e-01 5.89851260e-01 1.19802129e+00
-3.87022406e-01 -1.86389267e+00 1.46790147e-01 7.20995426e-01
-5.13914466e-01 -5.62452078e-01 -4.93930161e-01 4.72374439e-01
-1.23555109e-01 7.84790576e-01 -2.77739435e-01 -5.08854091e-01
1.16575971e-01 -1.32776320e-01 6.76692963e-01 -7.53190935e-01
-6.70687079e-01 -2.47994810e-01 3.56253237e-01 -4.40198958e-01
7.00131350e-04 -2.34741673e-01 -1.15605223e+00 -8.83182406e-01
-2.29001105e-01 4.49914843e-01 2.78893083e-01 2.20099851e-01
9.60799456e-01 5.38271189e-01 6.12559140e-01 -2.25070521e-01
-2.91103840e-01 -6.40577734e-01 -6.69578910e-01 2.17499495e-01
-6.76175803e-02 -2.22372249e-01 3.60904336e-02 1.86180770e-02] | [7.014904975891113, 6.035726070404053] |
54a24166-cace-4423-b922-b56e851d011d | promoting-the-knowledge-of-source-syntax-in | 1910.11218 | null | https://arxiv.org/abs/1910.11218v1 | https://arxiv.org/pdf/1910.11218v1.pdf | Promoting the Knowledge of Source Syntax in Transformer NMT Is Not Needed | The utility of linguistic annotation in neural machine translation seemed to had been established in past papers. The experiments were however limited to recurrent sequence-to-sequence architectures and relatively small data settings. We focus on the state-of-the-art Transformer model and use comparably larger corpora. Specifically, we try to promote the knowledge of source-side syntax using multi-task learning either through simple data manipulation techniques or through a dedicated model component. In particular, we train one of Transformer attention heads to produce source-side dependency tree. Overall, our results cast some doubt on the utility of multi-task setups with linguistic information. The data manipulation techniques, recommended in previous works, prove ineffective in large data settings. The treatment of self-attention as dependencies seems much more promising: it helps in translation and reveals that Transformer model can very easily grasp the syntactic structure. An important but curious result is, however, that identical gains are obtained by using trivial "linear trees" instead of true dependencies. The reason for the gain thus may not be coming from the added linguistic knowledge but from some simpler regularizing effect we induced on self-attention matrices. | ['Ondřej Bojar', 'Dominik Macháček', 'Thuong-Hai Pham'] | 2019-10-24 | null | null | null | null | ['small-data'] | ['computer-vision'] | [ 1.78305358e-01 5.06117880e-01 -8.16935077e-02 -3.22857559e-01
-9.55825031e-01 -6.35601282e-01 8.52995336e-01 -3.53683122e-02
-4.07174796e-01 8.71325791e-01 5.93602121e-01 -7.69441783e-01
9.22201276e-02 -4.15184289e-01 -8.85626614e-01 -8.30667019e-01
1.05378740e-01 7.46871293e-01 6.77868798e-02 -6.72453523e-01
1.58719152e-01 1.71725437e-01 -1.07863545e+00 4.61755484e-01
6.00030065e-01 3.51074666e-01 4.01141584e-01 5.01143277e-01
-3.73227447e-01 9.66866195e-01 -5.58809280e-01 -7.60360718e-01
1.86972439e-01 -6.15889728e-01 -1.04404950e+00 -1.60555899e-01
2.62215406e-01 7.95096010e-02 1.66974500e-01 8.65698755e-01
3.95385295e-01 -2.88232803e-01 4.09522891e-01 -6.92056715e-01
-8.86904418e-01 1.13438559e+00 -4.58447754e-01 4.44538176e-01
1.89112037e-01 2.48900637e-01 1.36142826e+00 -8.18561554e-01
6.65064454e-01 1.42789125e+00 6.56253397e-01 4.94023860e-01
-1.54266536e+00 -2.73996413e-01 2.21836656e-01 8.61210003e-02
-9.36557412e-01 -6.52236521e-01 6.55574381e-01 -4.58515227e-01
1.40634692e+00 2.96628892e-01 1.72837868e-01 1.29816818e+00
3.39739501e-01 6.14869058e-01 1.42114520e+00 -8.24400306e-01
-2.92054087e-01 4.06420141e-01 1.82829037e-01 6.93716884e-01
1.03091419e-01 4.33810577e-02 -3.71837944e-01 3.71423066e-02
5.79857767e-01 -4.95101720e-01 -3.25398743e-01 -1.55313268e-01
-1.33541393e+00 9.01220679e-01 3.33841532e-01 9.02082324e-01
-3.04685205e-01 1.52307879e-02 6.77631140e-01 8.20652068e-01
7.88560271e-01 5.48049510e-01 -9.31775570e-01 -1.76406905e-01
-6.47024095e-01 -1.54793620e-01 7.50945270e-01 6.43851638e-01
6.57824278e-01 1.20504005e-02 -9.44856107e-02 7.35880375e-01
9.03456137e-02 2.07049340e-01 7.08346784e-01 -6.29086196e-01
7.32299089e-01 4.36610401e-01 -1.52029365e-01 -5.68414509e-01
-3.94293576e-01 -4.49620545e-01 -5.43023944e-01 3.37331705e-02
8.64117086e-01 -2.79187024e-01 -7.39182174e-01 2.07398462e+00
-5.71539067e-02 -4.35750902e-01 1.05384536e-01 7.83575475e-01
2.24673405e-01 5.14502823e-01 8.56922846e-03 -1.95928842e-01
1.30706275e+00 -9.73068655e-01 -6.83904707e-01 -4.47595209e-01
1.17583704e+00 -8.78921986e-01 1.43385911e+00 1.80305094e-01
-1.20386839e+00 -5.15839815e-01 -6.17506087e-01 -2.49051690e-01
-4.14204508e-01 1.75949156e-01 7.95452237e-01 6.20517850e-01
-1.13434124e+00 7.12329090e-01 -8.60128284e-01 -6.49652064e-01
5.59742004e-02 4.74556357e-01 -3.39768767e-01 5.62722534e-02
-1.13379896e+00 1.49611187e+00 2.25121662e-01 1.65304780e-01
-4.93288368e-01 -4.45258409e-01 -7.88521588e-01 1.04885630e-01
3.52317482e-01 -8.28100324e-01 1.24200153e+00 -1.61767530e+00
-1.55686605e+00 1.00073993e+00 -2.59588540e-01 -4.65430200e-01
3.94993216e-01 -2.82383978e-01 1.27225518e-01 -2.87432164e-01
-3.28239389e-02 3.13443363e-01 5.33853352e-01 -1.16502142e+00
-1.61413208e-01 -4.37612593e-01 8.53758976e-02 5.29059060e-02
-1.82596728e-01 4.70258296e-01 3.24375406e-02 -6.57864213e-01
-2.99561530e-01 -1.00431550e+00 -1.65620670e-01 -6.40105546e-01
-2.95935929e-01 -3.90000135e-01 3.30266863e-01 -7.88031340e-01
1.16214752e+00 -1.92712712e+00 6.50900304e-01 -1.48402154e-01
-1.60261050e-01 3.22097093e-01 -3.19655240e-01 7.08880186e-01
-4.92616296e-01 3.59277666e-01 -9.67617556e-02 -3.92111242e-01
-1.61236733e-01 3.75819981e-01 -3.11555207e-01 2.90361494e-01
4.91831601e-01 1.18457615e+00 -6.45086646e-01 -2.79167563e-01
4.30590846e-02 4.05574888e-01 -4.13606197e-01 -6.06953241e-02
-2.61641473e-01 4.42841619e-01 -2.81125754e-01 3.08439791e-01
1.78914592e-01 -1.78528592e-01 4.76090938e-01 -1.17252044e-01
-2.52607316e-01 9.96622562e-01 -5.05603552e-01 1.88133609e+00
-5.90566456e-01 6.31106317e-01 7.77369514e-02 -1.37729311e+00
6.62314057e-01 4.38118309e-01 1.00058325e-01 -8.98564279e-01
2.81846106e-01 4.18099761e-01 6.51613176e-01 -5.76977849e-01
2.11540684e-01 -5.19817829e-01 5.05669527e-02 6.42014086e-01
2.19766110e-01 2.14820206e-01 1.98235348e-01 -3.99697982e-02
9.39329982e-01 5.16126573e-01 1.14412390e-01 -5.83075821e-01
3.76257986e-01 1.32736713e-01 4.43423331e-01 5.80984652e-01
3.16650987e-01 4.13482994e-01 7.30287135e-01 -4.78117764e-01
-1.24385881e+00 -4.97054756e-01 4.85599972e-02 1.46150827e+00
-5.78027248e-01 -4.91936713e-01 -6.84188306e-01 -7.11541533e-01
-3.59135747e-01 7.89763570e-01 -7.52259731e-01 -3.93076800e-02
-1.18210053e+00 -1.02003956e+00 6.15161359e-01 5.24659574e-01
-7.00782835e-02 -1.14334500e+00 -5.11554658e-01 4.64119881e-01
-2.70347685e-01 -9.45983112e-01 -2.10813493e-01 7.17659652e-01
-1.08343089e+00 -6.46523952e-01 -7.90704846e-01 -8.15795958e-01
3.66730273e-01 5.44851124e-02 1.33425927e+00 1.16241284e-01
2.47424453e-01 -8.39154869e-02 -4.42948163e-01 -3.55544329e-01
-7.18414187e-01 5.09392500e-01 -9.96907651e-02 -2.06130162e-01
4.18403387e-01 -7.94224322e-01 1.03495717e-02 9.76345986e-02
-5.86941004e-01 8.95739123e-02 1.00122476e+00 9.94425952e-01
-3.28553282e-02 -5.60083807e-01 5.23620605e-01 -1.14390731e+00
5.61259151e-01 -2.53016621e-01 -3.26506376e-01 2.86585808e-01
-4.79725271e-01 6.16557181e-01 8.47726762e-01 -3.91758293e-01
-1.05371368e+00 -1.35564551e-01 -2.84347743e-01 -1.56716079e-01
-1.05394281e-01 5.53993225e-01 -1.22496836e-01 2.23686308e-01
6.83743954e-01 1.24034338e-01 1.45862997e-01 -7.47695565e-01
4.42127109e-01 4.65949148e-01 9.86792818e-02 -8.82248461e-01
6.54527605e-01 1.27347201e-01 -1.68073490e-01 -6.64831877e-01
-8.18542361e-01 -1.41926721e-01 -7.80676901e-01 4.65106040e-01
6.65416062e-01 -7.95899093e-01 -4.58199620e-01 1.06174067e-01
-1.44678044e+00 -8.11560392e-01 -2.04174951e-01 1.94232643e-01
-5.56182086e-01 3.52167189e-01 -1.02937925e+00 -6.49183095e-01
-2.12215945e-01 -1.09444773e+00 9.99930561e-01 -4.63539720e-01
-3.52467239e-01 -1.16282547e+00 1.12822220e-01 3.57562155e-01
6.88233912e-01 -2.20960259e-01 1.37729776e+00 -9.49612439e-01
-6.10834658e-01 1.99927106e-01 -1.79209545e-01 4.01460469e-01
2.17916787e-01 -2.00260758e-01 -1.07516468e+00 -1.91543847e-01
1.40897259e-01 -5.00865102e-01 8.15680683e-01 1.03846759e-01
6.02600515e-01 -5.36184430e-01 -1.86668083e-01 2.04161838e-01
1.42142832e+00 -1.16879016e-01 5.99811971e-01 3.75348419e-01
7.53279448e-01 1.00953829e+00 1.94285005e-01 -1.76082000e-01
3.17194581e-01 9.62685823e-01 6.50223717e-02 -2.81670272e-01
-2.76273936e-01 -1.21076420e-01 6.38083577e-01 1.00846851e+00
-2.41361514e-01 -1.38446152e-01 -9.95057285e-01 5.97235084e-01
-1.88690710e+00 -8.93775344e-01 -2.75827765e-01 2.04129887e+00
1.00366783e+00 4.76609766e-01 2.60275006e-01 1.85623206e-02
3.52260560e-01 -5.33794872e-02 7.07293376e-02 -7.95888007e-01
-2.24698275e-01 2.91528106e-01 5.25991499e-01 7.12336183e-01
-5.73696196e-01 1.08322442e+00 6.00659370e+00 6.43215299e-01
-1.40378881e+00 3.85291129e-01 5.57058930e-01 -8.67598504e-02
-4.52799350e-01 1.96847394e-01 -7.34327674e-01 3.43284488e-01
1.22781622e+00 5.21844253e-02 3.33415866e-01 4.62719232e-01
2.67645478e-01 -6.17082678e-02 -1.43740296e+00 4.32265162e-01
8.67644176e-02 -1.12394404e+00 1.06729247e-01 1.79068387e-01
2.79028445e-01 3.51795197e-01 -1.46518439e-01 4.16123956e-01
3.85961264e-01 -1.01930487e+00 7.34813690e-01 1.08114749e-01
5.07925868e-01 -4.70618546e-01 8.21369946e-01 6.73270166e-01
-6.83861554e-01 -9.96842086e-02 -4.10275042e-01 -4.71666664e-01
2.51709104e-01 3.55063260e-01 -7.31661737e-01 6.84944570e-01
3.09081256e-01 3.96478117e-01 -7.20396161e-01 4.94221985e-01
-2.31556594e-01 8.46416473e-01 -3.17292511e-01 -3.62480357e-02
5.37881255e-01 -3.22792828e-01 4.52436477e-01 1.60717559e+00
3.50339264e-02 -2.28864729e-01 -8.72009322e-02 6.25666022e-01
1.89515457e-01 4.24727201e-01 -9.08610523e-01 -4.47979830e-02
-1.11215577e-01 1.14356411e+00 -5.37971735e-01 -2.45556980e-01
-8.29835415e-01 9.76830006e-01 7.68426657e-01 1.84471726e-01
-6.13609374e-01 -2.27406640e-02 4.54711169e-01 2.06676826e-01
4.91959333e-01 -2.14542747e-01 -4.45947617e-01 -1.34861553e+00
1.99645102e-01 -1.17051697e+00 7.75515288e-02 -6.22146189e-01
-1.08390248e+00 8.02417636e-01 -1.05183631e-01 -5.51975071e-01
-3.82052839e-01 -7.43254244e-01 -5.77679694e-01 1.13437045e+00
-1.47157395e+00 -1.34136486e+00 5.16008556e-01 4.11317170e-01
8.17354679e-01 8.83535221e-02 1.05766129e+00 4.09897298e-01
-5.70122838e-01 6.41864657e-01 -6.99819624e-03 1.76311642e-01
7.65297472e-01 -1.30463588e+00 5.43675303e-01 8.73976588e-01
4.36321378e-01 9.16929603e-01 8.53132129e-01 -3.52733493e-01
-1.36592710e+00 -5.26093423e-01 1.59735394e+00 -9.55262065e-01
1.02148104e+00 -6.64614081e-01 -1.15101671e+00 1.03271782e+00
8.04703116e-01 -3.68238896e-01 4.16853219e-01 5.92436254e-01
-3.37538391e-01 1.11082144e-01 -5.13652384e-01 4.67656881e-01
9.99667704e-01 -5.03556967e-01 -8.75966966e-01 2.67188013e-01
7.64794290e-01 -1.62316337e-01 -5.57486057e-01 2.52844483e-01
3.51354927e-01 -8.49108100e-01 6.56570256e-01 -9.59424496e-01
5.95525742e-01 7.33435079e-02 -1.02613151e-01 -1.45073473e+00
-4.74642932e-01 -5.99535108e-01 4.71261650e-01 1.48213327e+00
9.75424647e-01 -6.34707749e-01 4.25037205e-01 4.16561216e-01
-3.40140045e-01 -8.04302454e-01 -8.37931454e-01 -7.89378226e-01
5.25024712e-01 -2.72879243e-01 1.54458702e-01 1.14050639e+00
2.36468121e-01 1.06404734e+00 -3.90658170e-01 -2.04306543e-01
1.84297994e-01 5.44295162e-02 7.27838874e-01 -1.12940955e+00
-6.81334794e-01 -5.06709576e-01 1.39203072e-01 -8.68957281e-01
3.38331491e-01 -9.90949988e-01 -5.51139154e-02 -1.23689020e+00
6.88425750e-02 -4.87493515e-01 -2.36511379e-01 7.12953746e-01
-2.27789059e-01 5.64526953e-02 3.89937401e-01 2.77014971e-01
-2.00194314e-01 2.68888503e-01 1.03922093e+00 1.23488724e-01
-3.04061789e-02 -1.61538333e-01 -9.34851468e-01 6.06626034e-01
8.85468423e-01 -5.57594538e-01 -2.25119755e-01 -9.65267837e-01
5.95267713e-01 5.99830225e-02 2.05065146e-01 -3.28018844e-01
-6.65763170e-02 1.16548322e-01 -6.07629679e-03 -5.05736135e-02
1.67425781e-01 -7.42216468e-01 -5.76829500e-02 4.45117503e-01
-3.85389060e-01 3.97158325e-01 3.40826690e-01 8.10257569e-02
-1.53107913e-02 -2.75756121e-01 5.30988514e-01 -5.78571737e-01
-3.64537001e-01 -4.00227517e-01 -4.67373222e-01 -6.68716878e-02
4.91703719e-01 -1.39549104e-02 -1.32130578e-01 -8.01879615e-02
-8.02429259e-01 -9.27814171e-02 4.66744602e-01 5.58675170e-01
-1.53963313e-01 -9.71818745e-01 -8.49355280e-01 9.95412394e-02
5.83911203e-02 -3.40449035e-01 -3.92928511e-01 1.12638879e+00
-8.33170663e-04 6.96344554e-01 -1.87632456e-01 -4.22795117e-01
-1.21973550e+00 7.69307613e-01 3.06346625e-01 -5.13339162e-01
-6.35437131e-01 7.36947000e-01 2.35585585e-01 -3.84455442e-01
1.53183237e-01 -5.26482165e-01 7.60460198e-02 2.27795109e-01
2.31027022e-01 -7.36999661e-02 3.42232734e-01 -5.52169442e-01
-3.89142126e-01 5.49224377e-01 -3.59749347e-01 -2.00772151e-01
1.49594331e+00 -1.91656798e-01 -1.14720896e-01 6.56760991e-01
9.38144684e-01 1.83500648e-01 -8.18747938e-01 -3.13300163e-01
3.91019136e-01 -8.57373234e-04 -2.86258578e-01 -1.02404201e+00
-7.51049459e-01 1.22811997e+00 1.24312893e-01 2.05534890e-01
8.81749094e-01 1.54203430e-01 5.59952855e-01 5.28490126e-01
3.50116789e-01 -7.52937853e-01 -2.50861824e-01 6.27831161e-01
8.75479758e-01 -1.24919522e+00 -2.92652905e-01 -1.35489821e-01
-6.84380114e-01 1.04368806e+00 4.00542945e-01 -8.73388126e-02
5.63039705e-02 3.80717844e-01 3.11706454e-01 -8.56540799e-02
-1.03713059e+00 -4.33728129e-01 -3.44093144e-02 3.65927458e-01
1.21215677e+00 -1.62122160e-01 -5.62580168e-01 2.54757702e-01
-3.11077148e-01 -6.25847280e-02 4.26426023e-01 5.86231232e-01
-2.58357733e-01 -1.72074330e+00 -2.00316817e-01 1.22266963e-01
-9.27655458e-01 -5.24629354e-01 -6.40685141e-01 1.02661920e+00
-6.04982786e-02 8.01994383e-01 -1.37041986e-01 -4.89888936e-02
2.14244395e-01 6.03695691e-01 7.08049119e-01 -8.73030722e-01
-1.02021146e+00 2.04589590e-01 4.30109143e-01 -4.85385180e-01
-4.09389824e-01 -6.55024290e-01 -7.14569986e-01 -1.56340808e-01
-4.27995861e-01 2.34522328e-01 6.43398702e-01 1.18792641e+00
2.99690157e-01 4.39640969e-01 3.17139924e-01 -8.24442625e-01
-8.04276943e-01 -1.35919178e+00 9.88465324e-02 4.02230561e-01
3.77696335e-01 -3.07537794e-01 -2.24262357e-01 -1.89429391e-02] | [11.118412971496582, 9.713634490966797] |
5c789a6a-c635-4e97-ae9b-d94708c2e9ad | graph-representation-learning-for-interactive | 2304.02656 | null | https://arxiv.org/abs/2304.02656v1 | https://arxiv.org/pdf/2304.02656v1.pdf | Graph Representation Learning for Interactive Biomolecule Systems | Advances in deep learning models have revolutionized the study of biomolecule systems and their mechanisms. Graph representation learning, in particular, is important for accurately capturing the geometric information of biomolecules at different levels. This paper presents a comprehensive review of the methodologies used to represent biological molecules and systems as computer-recognizable objects, such as sequences, graphs, and surfaces. Moreover, it examines how geometric deep learning models, with an emphasis on graph-based techniques, can analyze biomolecule data to enable drug discovery, protein characterization, and biological system analysis. The study concludes with an overview of the current state of the field, highlighting the challenges that exist and the potential future research directions. | ['Yu Guang Wang', 'Bingxin Zhou', 'Xinye Xiong'] | 2023-04-05 | null | null | null | null | ['drug-discovery'] | ['medical'] | [ 2.86310345e-01 6.90542394e-03 1.16120186e-02 -3.71792875e-02
-8.12277421e-02 -4.92473066e-01 4.13285017e-01 8.96169662e-01
2.16172218e-01 7.74872482e-01 -1.81765631e-01 -6.07330620e-01
5.04886881e-02 -1.07362604e+00 -8.79305601e-01 -9.22714889e-01
-5.32608628e-01 5.02327263e-01 -1.38747454e-01 -2.20160514e-01
3.30119967e-01 1.31696451e+00 -9.27198768e-01 3.59452307e-01
6.69689357e-01 8.41832578e-01 -5.11475168e-02 9.78754520e-01
-4.47733998e-01 7.67092049e-01 -5.91284633e-01 -3.48764360e-01
-1.81251168e-01 -3.76786947e-01 -4.59437728e-01 -7.40913525e-02
7.91473806e-01 2.89111316e-01 -4.77896452e-01 6.76527321e-01
2.78328091e-01 1.43032476e-01 1.17239118e+00 -4.81326938e-01
-1.15664709e+00 1.12919221e-02 -1.77235872e-01 6.67956891e-03
3.91952813e-01 2.51782447e-01 7.40422130e-01 -8.52433324e-01
7.74947762e-01 1.54349470e+00 6.70764506e-01 5.91270149e-01
-1.24416220e+00 -2.72210956e-01 2.08517060e-01 1.77002832e-01
-1.06283367e+00 -1.16956986e-01 8.37478936e-01 -6.75623715e-01
1.42503631e+00 1.01692080e-01 7.51052976e-01 8.12048554e-01
1.31486523e+00 3.22127551e-01 7.20660865e-01 -2.88460553e-01
1.17523424e-01 -8.19267392e-01 5.16225517e-01 1.14159977e+00
7.99803734e-01 -8.52455646e-02 -1.84404597e-01 -2.80911565e-01
8.51874232e-01 3.64003897e-01 -7.14714453e-02 -7.50805557e-01
-7.77373254e-01 7.38687038e-01 7.38863409e-01 3.44229132e-01
-4.96299863e-01 4.98058528e-01 2.29355752e-01 -2.79876161e-02
2.09806457e-01 7.91359484e-01 -3.82966511e-02 5.81002891e-01
-3.98706704e-01 4.12101418e-01 8.17703366e-01 6.86191797e-01
7.50159383e-01 3.61439943e-01 1.54879838e-01 3.81039470e-01
4.22908992e-01 5.82814991e-01 -3.70124787e-01 -2.49004528e-01
8.68819281e-02 9.28397715e-01 -5.24463058e-02 -1.40828466e+00
-9.43855286e-01 -5.08835204e-02 -1.07356000e+00 8.11741799e-02
4.32705462e-01 2.07722150e-02 -1.13235164e+00 1.08264554e+00
1.21994130e-01 -7.82568306e-02 -2.64634937e-02 7.41659462e-01
1.39662039e+00 6.96285129e-01 5.26248634e-01 5.85177764e-02
1.29103911e+00 -4.55041438e-01 -6.66914940e-01 2.28665620e-01
7.21324205e-01 -5.18270791e-01 4.49742109e-01 1.24525987e-01
-9.16112959e-01 -5.17111897e-01 -1.12019956e+00 -3.40465873e-01
-1.06279218e+00 -2.20559701e-01 1.15221012e+00 6.01771653e-01
-7.71376669e-01 7.32364655e-01 -1.10548639e+00 -4.47256535e-01
6.96500003e-01 6.86853945e-01 -3.47173184e-01 -3.95114310e-02
-1.07237971e+00 8.62522662e-01 1.86288401e-01 1.76458478e-01
-1.09008181e+00 -6.92027271e-01 -1.07262313e+00 8.19741786e-02
-2.27720767e-01 -6.97029889e-01 5.31178176e-01 -4.48101521e-01
-1.27022076e+00 8.94164503e-01 -2.66434819e-01 -6.74379468e-01
4.17030707e-04 3.17896366e-01 -3.55599970e-01 1.07841052e-01
-8.20315361e-01 3.35681528e-01 2.04421476e-01 -9.85202670e-01
1.98820576e-01 -7.44557381e-01 -2.56992169e-02 -1.11666784e-01
2.44621441e-01 -3.12043607e-01 3.01962912e-01 -2.67781556e-01
-9.26155597e-02 -6.71254396e-01 -5.70430696e-01 4.83910218e-02
-4.89897430e-01 -2.42505863e-01 7.02535808e-01 -1.82708576e-01
7.31520712e-01 -1.66628861e+00 3.55085164e-01 3.54083687e-01
9.33048546e-01 5.32268286e-01 -3.57173264e-01 1.04504681e+00
-1.95323482e-01 1.44196078e-01 -7.56882802e-02 2.62007922e-01
-4.23649520e-01 -1.61286190e-01 -2.07570016e-01 8.01933169e-01
3.40645045e-01 1.56046343e+00 -9.60626900e-01 1.20375648e-01
6.03545308e-01 8.28468561e-01 4.22373414e-02 1.37913138e-01
-6.72865212e-01 4.47462201e-01 -6.84726000e-01 8.84457529e-01
7.80537665e-01 -7.61550486e-01 7.00888276e-01 -2.75856316e-01
-4.21703905e-02 1.64002478e-01 -3.35663438e-01 9.99096811e-01
3.63906510e-02 6.74339592e-01 -1.01748452e-01 -1.23093736e+00
1.30333710e+00 5.15222689e-03 5.31517565e-01 -6.99624836e-01
-3.72370221e-02 1.89263895e-01 3.04698557e-01 -3.13725531e-01
2.80771349e-02 -5.61481016e-03 4.13104564e-01 1.93673998e-01
-3.58411580e-01 8.54041353e-02 1.63311183e-01 1.20194502e-01
8.96143138e-01 -5.56404628e-02 7.02466309e-01 -4.39973831e-01
8.56239438e-01 -2.98283789e-02 -1.58711478e-01 5.83430290e-01
-8.85994956e-02 8.87370631e-02 5.64876139e-01 -1.24891341e+00
-1.05537581e+00 -9.66083109e-01 -1.34581998e-01 9.04149413e-01
1.97362006e-01 -2.74824828e-01 -7.78491676e-01 -2.98886150e-01
5.95767617e-01 -2.76433885e-01 -9.59539592e-01 -1.61671504e-01
-8.25949013e-01 -9.93736267e-01 3.31265539e-01 4.89448667e-01
-2.03051016e-01 -1.20952415e+00 -1.82563677e-01 4.49356377e-01
6.13919556e-01 -7.24683166e-01 -6.09491486e-03 1.78277180e-01
-9.67203438e-01 -1.68007922e+00 -6.16945922e-01 -8.15633655e-01
8.65960062e-01 4.41355854e-01 1.13521624e+00 2.81845689e-01
-6.90248787e-01 1.95095301e-01 -6.72777370e-03 -5.34004509e-01
-4.45486933e-01 -4.09746580e-02 -1.24377847e-01 -2.61669368e-01
6.46418214e-01 -2.18005076e-01 -8.20657492e-01 -1.65579975e-01
-7.97667682e-01 -2.73018718e-01 3.43188763e-01 6.56216681e-01
8.52609932e-01 -6.11896276e-01 6.39238358e-01 -1.19262278e+00
7.64354527e-01 -2.63793468e-01 -6.54849112e-01 4.97653693e-01
-3.85888517e-01 1.63893729e-01 1.02021968e+00 -2.72172153e-01
-3.81785840e-01 -1.19342655e-01 8.01413581e-02 1.08099811e-01
-1.21816769e-01 5.51275194e-01 -2.48398751e-01 -8.71146917e-01
7.96990216e-01 2.28779525e-01 2.38397524e-01 -2.46202826e-01
3.53550166e-01 1.24857247e-01 2.62038726e-02 -5.49570918e-01
1.48275448e-02 5.78393877e-01 8.79172742e-01 -1.37229979e+00
-2.79599547e-01 -2.51064926e-01 -6.62116051e-01 -2.17199013e-01
8.32375526e-01 -2.47701377e-01 -1.23775983e+00 6.25654578e-01
-1.39429915e+00 -4.46842164e-01 2.56727546e-01 2.40337513e-02
-4.69822735e-01 6.86240017e-01 -8.05383801e-01 -4.56211478e-01
-5.51965892e-01 -1.24659276e+00 1.13917041e+00 8.87004137e-02
-2.57890195e-01 -1.73479724e+00 2.89790690e-01 4.37160917e-02
3.03779453e-01 1.05441797e+00 1.55721068e+00 -6.89556479e-01
-9.09396887e-01 -2.59983987e-01 -2.81741947e-01 -3.37386489e-01
4.15412307e-01 1.65581435e-01 -6.30025268e-01 -5.09305000e-01
-5.55900872e-01 -2.15551049e-01 9.56934333e-01 7.97951162e-01
1.23691380e+00 -2.18665376e-01 -8.29031110e-01 7.54549503e-01
1.28272331e+00 5.80766320e-01 6.88933969e-01 -7.04773217e-02
1.16212535e+00 4.61392313e-01 -8.17150101e-02 -4.78662550e-02
-8.03661160e-03 3.01903397e-01 6.38654411e-01 -4.09622550e-01
-4.33683962e-01 -7.37006292e-02 -1.27891541e-01 6.11856043e-01
-2.21304625e-01 -8.77659142e-01 -1.07879221e+00 -1.21919893e-01
-1.62507129e+00 -8.21367383e-01 -4.81168598e-01 1.92171323e+00
2.37035319e-01 -2.58635879e-01 8.59350488e-02 -3.38707477e-01
7.32178271e-01 3.60077679e-01 -1.03682268e+00 -7.26958930e-01
-4.39210445e-01 4.41251278e-01 5.25044382e-01 7.65477180e-01
-9.94068801e-01 1.18905902e+00 8.09261513e+00 1.52644917e-01
-1.54442072e+00 -7.90822029e-01 7.27993309e-01 5.06751895e-01
-3.23734611e-01 -3.56980413e-01 -6.06617868e-01 2.24038109e-01
1.00762284e+00 -1.57182619e-01 3.37359935e-01 3.16275030e-01
2.96440929e-01 3.99325490e-01 -1.23175859e+00 7.59896994e-01
8.00762512e-03 -2.19670534e+00 8.97394955e-01 3.64576072e-01
6.12472415e-01 -2.75415815e-02 2.19324782e-01 -3.41993779e-01
-8.99662450e-02 -1.49560952e+00 -2.45156903e-02 8.97826195e-01
7.78654456e-01 -6.94591224e-01 4.66193616e-01 -2.59141505e-01
-1.34047413e+00 4.66203958e-01 -6.51722729e-01 -2.85584956e-01
-1.02320738e-01 5.09326935e-01 -9.22418237e-01 4.64580208e-01
-1.30318701e-01 1.10660243e+00 -3.81007999e-01 7.72760034e-01
2.26201013e-01 2.39620179e-01 2.32926667e-01 -6.64258897e-01
3.39331925e-01 -6.16540670e-01 3.73996198e-01 1.56917679e+00
-1.78829327e-01 2.57517904e-01 2.82815307e-01 9.06138420e-01
-3.24889600e-01 3.25527728e-01 -8.34828079e-01 -6.30545497e-01
2.42927626e-01 8.89003992e-01 -8.25586677e-01 -2.27627859e-01
-3.54977578e-01 2.06613541e-01 5.58583021e-01 6.81476176e-01
-5.07861555e-01 -6.80455208e-01 9.63758171e-01 4.18748319e-01
1.11491047e-01 -7.68926740e-01 -1.26137927e-01 -7.95620382e-01
-4.60542291e-01 -9.71714795e-01 1.29359141e-01 -4.45554763e-01
-1.20716155e+00 2.02191919e-01 -5.68242550e-01 -4.41834331e-01
2.37223238e-01 -1.65609145e+00 -6.07279778e-01 7.44722068e-01
-1.29005885e+00 -1.10631824e+00 -2.05200300e-01 9.29310396e-02
1.19388223e-01 -9.67438295e-02 9.94384527e-01 -1.21377811e-01
-5.06054163e-01 2.51374602e-01 6.17966592e-01 4.19694595e-02
3.36105496e-01 -1.18532860e+00 1.12365401e+00 1.54413551e-01
-1.08786717e-01 1.07486534e+00 5.34179032e-01 -9.64826405e-01
-2.14851999e+00 -1.15906298e+00 5.20140827e-01 -4.31573451e-01
6.06967628e-01 -6.63633406e-01 -1.35187221e+00 4.40467685e-01
-1.81253031e-01 -2.83157881e-02 9.65710878e-01 -8.23510252e-03
-3.20464939e-01 1.49035946e-01 -9.73767638e-01 6.07080936e-01
9.49501574e-01 -4.67482656e-01 -5.51670603e-02 6.78030789e-01
3.69318128e-01 -5.21361828e-01 -1.05834246e+00 3.88482958e-01
7.74073422e-01 -6.86445534e-01 1.19283211e+00 -1.47805595e+00
1.86501682e-01 -1.16665021e-01 2.48382509e-01 -9.17107165e-01
-7.89798081e-01 -6.27670527e-01 -6.53941631e-01 5.52099980e-02
3.07618231e-01 -7.25097299e-01 1.11320782e+00 3.33918244e-01
-1.07341789e-01 -1.08144164e+00 -3.82968009e-01 -6.82527304e-01
5.14684677e-01 1.61634058e-01 7.33019829e-01 6.63257837e-01
1.01505190e-01 4.27957684e-01 5.79653271e-02 -1.84475541e-01
5.65037727e-01 4.61345404e-01 7.55872667e-01 -1.47061634e+00
4.39673930e-01 -7.17698157e-01 -1.04709375e+00 -9.99617994e-01
3.61173242e-01 -1.08261991e+00 -4.87296134e-01 -1.82661426e+00
1.66113675e-01 1.22565813e-01 -4.05776531e-01 1.28065214e-01
-5.51243536e-02 -1.00960471e-01 -1.69825390e-01 8.70819576e-03
-6.58735871e-01 3.10207635e-01 1.74447381e+00 -5.70358157e-01
-8.57452955e-03 -3.10329378e-01 -7.01218963e-01 3.91783774e-01
8.58049214e-01 1.07135803e-01 -6.69416338e-02 -3.47716689e-01
3.10163409e-01 -2.64855474e-01 2.11986557e-01 -7.58964956e-01
1.40396833e-01 -5.19751668e-01 7.58910179e-01 -5.00859559e-01
2.41617963e-01 -5.34215689e-01 5.32841608e-02 8.21013808e-01
-4.60262626e-01 1.18455261e-01 4.42052156e-01 9.63063478e-01
1.46035761e-01 3.75706643e-01 9.83318627e-01 -3.20265621e-01
-3.74859899e-01 8.56373906e-01 -7.63145447e-01 -2.42364988e-01
1.05096376e+00 -4.52502400e-01 -6.97368503e-01 -6.11733310e-02
-7.22240865e-01 1.12530574e-01 7.84426212e-01 7.63541907e-02
8.51976573e-01 -1.03565693e+00 -4.58745360e-01 2.33709067e-01
2.54867435e-01 -3.49934101e-01 2.39488438e-01 2.43130714e-01
-1.15125799e+00 1.17536664e+00 -3.23804379e-01 -4.86037672e-01
-1.24121010e+00 8.15144420e-01 8.00269008e-01 -1.46506459e-01
-3.32379818e-01 4.53952640e-01 7.45064974e-01 -1.97730027e-02
1.16675049e-01 -5.68237007e-01 -3.37580323e-01 -5.71767092e-01
4.60651875e-01 4.15284365e-01 1.62423000e-01 -5.36824226e-01
-6.66438043e-01 9.02511299e-01 -3.20461810e-01 9.65778887e-01
1.19871044e+00 3.48308951e-01 -4.84264702e-01 1.23133987e-01
1.23098469e+00 -2.72119284e-01 -9.97398138e-01 1.81803897e-01
8.75855889e-03 1.15431108e-01 -3.84866387e-01 -4.72021103e-01
-6.95804417e-01 1.29979646e+00 3.80741239e-01 3.27555954e-01
3.49266410e-01 -1.62194259e-02 7.65131772e-01 8.63792717e-01
2.02348709e-01 -3.57028633e-01 1.91682979e-01 8.03024828e-01
9.16287005e-01 -9.45644259e-01 4.32071477e-01 -5.45543313e-01
1.09802723e-01 1.54775810e+00 4.98757064e-01 -4.87058580e-01
6.25683129e-01 4.52995300e-02 -7.22289924e-03 -8.67280006e-01
-6.84301198e-01 1.29431129e-01 6.34677112e-01 1.04833770e+00
1.28244460e+00 3.32153141e-01 -1.32327959e-01 -7.68986717e-02
2.03690827e-01 -3.86927336e-01 3.36633593e-01 8.36396396e-01
-7.97103584e-01 -1.23942864e+00 -1.69754982e-01 5.16164482e-01
-3.49035859e-01 -1.14828244e-01 -1.36229098e+00 5.18804371e-01
-3.02201241e-01 5.69332540e-01 1.21290781e-01 -1.59187421e-01
3.26253831e-01 -1.40410975e-01 8.96167099e-01 -6.56333625e-01
-3.08106065e-01 -2.99916118e-01 -3.50254148e-01 -4.52103943e-01
-2.46793091e-01 -1.16818756e-01 -1.54678810e+00 -4.83451992e-01
-1.25528932e-01 2.44692042e-01 5.03183246e-01 7.35113442e-01
9.93433297e-01 6.09683037e-01 1.18106760e-01 -8.71794999e-01
6.49749339e-02 -5.15914619e-01 -4.55932528e-01 2.21091792e-01
6.59164310e-01 -6.31974876e-01 8.44098330e-02 2.77392529e-02] | [5.115475654602051, 5.79793643951416] |
f28be5cd-8459-4a8f-ac48-b4af04199de2 | achieving-reliable-human-assessment-of-open | null | null | https://openreview.net/forum?id=7SQh5IvfeLq | https://openreview.net/pdf?id=7SQh5IvfeLq | Achieving Reliable Human Assessment of Open-Domain Dialogue Systems | Evaluation of open-domain dialogue systems is highly challenging and development of better techniques is highlighted time and again as desperately needed. Despite substantial efforts to carry out reliable live evaluation of systems in recent competitions, annotations have been abandoned and reported as too unreliable to yield sensible results. This is a serious problem since automatic metrics are not known to provide a good indication of what may or may not be a high-quality conversation. Answering the distress call of competitions that have emphasized the urgent need for better evaluation techniques in dialogue, we present the successful development of human evaluation that is highly reliable while still remaining feasible and low cost. Self-replication experiments reveal almost perfectly repeatable results with a correlation of $r=0.969$. Furthermore, due to the lack of appropriate methods of statistical significance testing, the likelihood of potential improvements to systems occurring due to chance is rarely taken into account in dialogue evaluation, and the evaluation we propose facilitates application of standard tests. Since we have developed a highly reliable evaluation method, new insights into system performance can be revealed. We therefore include a comparison of state-of-the-art models (i) with and without personas, to measure the contribution of personas to conversation quality, as well as (ii) prescribed versus freely chosen topics. Interestingly with respect to personas, results indicate that personas do not positively contribute to conversation quality as expected. | ['Anonymous'] | 2021-09-17 | null | null | null | acl-arr-september-2021-9 | ['dialogue-evaluation'] | ['natural-language-processing'] | [-7.55537674e-02 2.87946135e-01 3.26565117e-01 -5.61136901e-01
-9.47480142e-01 -6.95885420e-01 1.07429683e+00 3.15702021e-01
-8.14694047e-01 1.12872362e+00 6.61730826e-01 -2.70932585e-01
-1.15990080e-01 -3.45661640e-01 -1.68673739e-01 -3.74213427e-01
1.86774004e-02 6.65667653e-01 1.86704755e-01 -5.95227420e-01
4.84585762e-01 5.39259724e-02 -1.38666594e+00 2.41967171e-01
7.51243293e-01 4.06502545e-01 -7.73676112e-02 9.50976193e-01
-8.55947211e-02 8.32762063e-01 -1.28436387e+00 -7.43973434e-01
-2.06894144e-01 -7.77725339e-01 -1.26385021e+00 2.46273771e-01
2.34042019e-01 -4.28060383e-01 1.37761563e-01 5.85904717e-01
8.03676546e-01 2.58690864e-01 4.84275699e-01 -1.01543558e+00
-1.01311162e-01 4.06015873e-01 -1.79310888e-02 1.41622931e-01
8.41213822e-01 3.07335228e-01 1.15094638e+00 -4.83624935e-01
6.53418660e-01 1.24394202e+00 5.55133224e-01 5.47366083e-01
-1.34780002e+00 -1.86052188e-01 -8.55139717e-02 -1.63904890e-01
-9.93259490e-01 -8.76816988e-01 2.19279706e-01 -3.29672903e-01
1.10826147e+00 5.27293742e-01 5.10452688e-01 1.03409493e+00
-1.02414332e-01 4.36851233e-01 1.22200298e+00 -5.08651435e-01
3.81492786e-02 8.18873107e-01 -3.88699397e-03 4.01932478e-01
8.64914581e-02 -2.45269343e-01 -6.93683863e-01 -4.32928860e-01
3.77450913e-01 -7.83571184e-01 -4.03010547e-01 -5.23950644e-02
-1.29023170e+00 8.13857317e-01 1.27932278e-03 6.52966142e-01
-2.42261142e-01 -4.36450481e-01 7.88802147e-01 5.55043876e-01
5.12948394e-01 8.21880162e-01 -3.85701507e-01 -1.03987491e+00
-7.27721393e-01 7.01006770e-01 1.45917463e+00 4.23651338e-01
1.90996021e-01 -2.19231606e-01 -1.00233032e-04 1.23981905e+00
9.53104272e-02 1.59889236e-02 5.29814005e-01 -1.12959361e+00
4.00396943e-01 6.65220320e-01 5.42346478e-01 -1.10739005e+00
-4.40261543e-01 -1.46695137e-01 -3.14512402e-01 2.31119365e-01
9.41233099e-01 -2.81884611e-01 1.33250624e-01 1.71008587e+00
-1.30294869e-02 -6.86496913e-01 1.83992803e-01 8.23805869e-01
6.83817208e-01 4.16517973e-01 3.60843837e-02 -4.16616350e-01
1.34424281e+00 -6.35560691e-01 -7.81019270e-01 -6.21740194e-03
8.07045519e-01 -1.06864250e+00 1.27627409e+00 5.54748774e-01
-1.27341223e+00 -4.10524428e-01 -8.96745622e-01 1.99373066e-01
-2.08572775e-01 -2.99187064e-01 6.17344081e-01 1.02400947e+00
-1.04200685e+00 7.01735556e-01 -4.40236509e-01 -6.05498672e-01
-2.37120718e-01 2.31493369e-01 -5.62181115e-01 1.96203604e-01
-1.19144869e+00 1.32028651e+00 -7.79339857e-03 1.17269352e-01
-2.98172772e-01 -9.11870003e-02 -5.49321413e-01 -2.94357300e-01
3.54000449e-01 -3.15132946e-01 1.59360039e+00 -9.49688733e-01
-1.54217541e+00 8.96001160e-01 -1.30906850e-01 -2.64732242e-01
6.81570053e-01 -1.72206447e-01 -3.04827958e-01 -2.88804509e-02
1.42894417e-01 3.36029917e-01 1.39763147e-01 -1.00434935e+00
-7.23674357e-01 -1.53161019e-01 3.74445260e-01 5.95121264e-01
-2.28372052e-01 5.18331170e-01 -1.27646655e-01 -1.15267925e-01
-2.47088879e-01 -8.07745993e-01 -9.36735272e-02 -4.05296236e-01
-1.09943338e-01 -5.67394853e-01 2.99520284e-01 -8.03324282e-01
1.36307621e+00 -1.76809371e+00 -1.17034517e-01 -3.44227515e-02
2.01561496e-01 2.77687132e-01 2.16020837e-01 8.10514390e-01
2.34987542e-01 2.90685296e-01 -9.55600590e-02 -2.99197823e-01
2.53077894e-01 5.19597903e-02 1.82668775e-01 3.62769812e-01
1.89978480e-01 5.21905124e-01 -9.41321731e-01 -4.64388698e-01
2.25200176e-01 2.72235900e-01 -3.14080983e-01 4.54097182e-01
-6.50459304e-02 3.18387985e-01 -2.23772898e-01 2.43267521e-01
-5.53458855e-02 -1.00068592e-01 1.85434774e-01 2.05951467e-01
-4.10463333e-01 8.73015821e-01 -1.10451770e+00 1.44321227e+00
-3.96850467e-01 7.34945953e-01 1.55196816e-01 -6.95374608e-01
9.35404480e-01 7.26509213e-01 1.68606609e-01 -6.42157435e-01
1.48499027e-01 2.62488127e-01 5.10298371e-01 -5.78270257e-01
8.67718101e-01 -3.86192411e-01 -9.86237898e-02 7.05599129e-01
-2.36598868e-02 -3.29359949e-01 3.37925643e-01 3.38340402e-01
1.14239657e+00 -2.04555001e-02 2.24266052e-01 -2.53691107e-01
5.61559021e-01 1.43703759e-01 2.85899848e-01 6.88649178e-01
-4.65825617e-01 5.97110093e-01 7.73231685e-01 -3.43262888e-02
-1.08274996e+00 -5.06337821e-01 -1.29704192e-01 1.02511692e+00
-3.27114969e-01 -5.47072053e-01 -8.82752776e-01 -4.83337939e-01
-4.06467617e-01 7.25790560e-01 -3.33109200e-01 1.93755463e-01
-2.56772786e-01 -7.90681303e-01 6.56370878e-01 1.07037768e-01
3.45478803e-01 -1.07320237e+00 -6.84433281e-01 4.14780974e-01
-5.51604033e-01 -1.04396224e+00 -1.31187797e-01 3.86023484e-02
-6.21580601e-01 -1.00874662e+00 -9.31035876e-01 -4.21948284e-01
2.32461721e-01 2.03619853e-01 1.30742705e+00 3.01371545e-01
8.20451155e-02 4.53017116e-01 -5.37723064e-01 -2.85608977e-01
-8.29137146e-01 1.35512454e-02 -1.10991903e-01 -3.42491239e-01
4.53364670e-01 -1.68220624e-01 -5.62537789e-01 6.25949502e-01
-5.01232386e-01 -2.55945653e-01 3.07019979e-01 9.52395618e-01
-4.39364940e-01 -3.50856893e-02 8.63716900e-01 -1.06532621e+00
1.40674496e+00 -2.45974764e-01 -5.64530417e-02 1.74438041e-02
-8.51182103e-01 -1.62965998e-01 3.08066130e-01 -3.34065943e-03
-1.18520164e+00 -6.13139927e-01 -3.36890161e-01 5.54322779e-01
-4.51226979e-01 5.44857562e-01 8.04281682e-02 2.93102339e-02
9.87407923e-01 -3.07989806e-01 3.47166270e-01 -2.63390899e-01
7.44859651e-02 1.00515807e+00 1.94357723e-01 -6.08639181e-01
3.35936934e-01 3.49417143e-02 -6.58843994e-01 -1.26451373e+00
-3.90511274e-01 -7.15839803e-01 -3.12402159e-01 -3.99504751e-01
6.89287901e-01 -8.15676212e-01 -7.68600166e-01 2.19641760e-01
-1.09132421e+00 -4.55503434e-01 -5.40694632e-02 3.92708778e-01
-4.92785096e-01 6.15017533e-01 -5.65890670e-01 -1.17488599e+00
-5.88414073e-02 -1.11107731e+00 5.88355124e-01 2.49488484e-02
-1.18754816e+00 -1.12817323e+00 1.71623975e-01 8.79038572e-01
4.92611945e-01 1.30427286e-01 5.70068121e-01 -9.91046906e-01
-3.56392972e-02 -6.62367940e-01 6.10763393e-03 5.13247609e-01
1.96273237e-01 7.69395679e-02 -1.02113843e+00 -1.56537667e-01
5.17771095e-02 -7.27496028e-01 1.39784247e-01 -1.34458229e-01
2.77735531e-01 -2.54459739e-01 1.35069579e-01 -5.41707575e-01
7.79881477e-01 1.41377985e-01 6.62386775e-01 3.94657582e-01
4.90312539e-02 1.22988653e+00 6.93560123e-01 4.64787692e-01
5.55432379e-01 9.91287351e-01 -1.46417826e-01 5.87525405e-02
1.80977598e-01 -2.52019819e-02 3.65921915e-01 7.91389108e-01
-2.21726522e-01 -3.69374633e-01 -7.90511131e-01 6.16319478e-01
-1.65789616e+00 -1.05755436e+00 -4.26064074e-01 2.48665595e+00
8.79036248e-01 5.69595695e-01 6.60864115e-01 3.50186825e-01
5.08580208e-01 2.04616770e-01 9.25882161e-02 -8.63021433e-01
-1.07437201e-01 -7.54254013e-02 -4.17620763e-02 7.20526755e-01
-5.44964790e-01 5.50778210e-01 6.64685822e+00 3.92219275e-01
-6.59794629e-01 -1.16660528e-01 7.34730124e-01 7.95020685e-02
-2.50077665e-01 1.47024304e-01 -4.53494340e-01 3.26321125e-01
1.28117657e+00 -2.56403714e-01 1.84238896e-01 5.15962660e-01
5.75834572e-01 -5.64135253e-01 -9.98745322e-01 5.75601876e-01
5.02566434e-02 -7.38590956e-01 -5.10726631e-01 1.80464596e-01
2.88461298e-01 -2.80654758e-01 -5.01969934e-01 4.58252162e-01
3.03144723e-01 -8.53463054e-01 4.65892226e-01 2.18758836e-01
3.45581263e-01 -6.27762914e-01 1.02726912e+00 5.26499093e-01
-4.18663502e-01 3.07476103e-01 -3.01261425e-01 -4.22271580e-01
2.69895971e-01 2.68245220e-01 -1.16483557e+00 2.50370473e-01
4.00060445e-01 1.15089551e-01 -3.18666250e-01 8.10234845e-01
-1.38501421e-01 7.07159400e-01 -1.96048811e-01 -5.07451653e-01
2.96029598e-01 -1.58904158e-02 5.00096202e-01 1.34375429e+00
-1.20333001e-01 2.42178306e-01 -1.81895532e-02 4.03226435e-01
2.29254246e-01 5.17063975e-01 -6.91596627e-01 -2.89899141e-01
2.87312508e-01 1.23745871e+00 -7.18462348e-01 -1.32090002e-01
-6.48406565e-01 9.48207796e-01 2.29795679e-01 -5.02953008e-02
-3.21999788e-01 -2.32489049e-01 4.10978705e-01 1.46917164e-01
-3.11482906e-01 -1.94740176e-01 -1.88682094e-01 -8.35149050e-01
3.19590382e-02 -1.29325116e+00 1.63382262e-01 -4.82351750e-01
-1.11293161e+00 7.66554296e-01 -8.18144083e-02 -8.24877620e-01
-7.19774485e-01 -3.29411000e-01 -5.83510220e-01 1.16207778e+00
-9.87009168e-01 -4.04345632e-01 -1.84066340e-01 2.74965495e-01
6.08955264e-01 -1.31407812e-01 1.21129274e+00 1.23269685e-01
-3.29288453e-01 6.54073715e-01 -1.15818277e-01 -1.32737592e-01
1.14056587e+00 -1.39144373e+00 2.87712991e-01 3.24812382e-01
3.91279459e-02 8.40958059e-01 1.21978116e+00 -4.71620589e-01
-8.33450019e-01 -2.24557132e-01 1.39772260e+00 -7.16763437e-01
7.41719484e-01 -2.37295538e-01 -1.01411021e+00 7.36915041e-03
5.12358189e-01 -8.51855874e-01 9.10117924e-01 6.87588871e-01
-2.07202118e-02 2.45270908e-01 -9.87472653e-01 5.29313624e-01
6.66097581e-01 -5.25842667e-01 -7.50210285e-01 3.65193367e-01
3.31097245e-01 -2.43170440e-01 -9.40009594e-01 1.26468062e-01
6.76061749e-01 -1.40523076e+00 5.49160302e-01 -3.44051272e-01
3.81552339e-01 2.06317574e-01 -2.82307137e-02 -1.22966719e+00
6.18647188e-02 -7.13026226e-01 6.80517375e-01 1.53011692e+00
6.90463245e-01 -4.97911066e-01 8.03727806e-01 1.25043225e+00
-1.39786318e-01 -5.98027885e-01 -7.94448197e-01 -4.66257095e-01
2.02756599e-02 -3.53242129e-01 1.78005457e-01 1.00278080e+00
6.07142270e-01 7.01086819e-01 -4.94190395e-01 -4.22998905e-01
1.63645029e-01 -3.60252321e-01 1.23750508e+00 -1.21244025e+00
-3.87177169e-01 -6.85774446e-01 -4.40448850e-01 -7.95927167e-01
-1.14178032e-01 -2.22386479e-01 2.60328144e-01 -1.53030956e+00
1.37051985e-01 -3.16047072e-01 5.11777438e-02 3.76042053e-02
-3.58926982e-01 1.81527853e-01 9.71730947e-02 1.42126933e-01
-6.18759930e-01 3.70619655e-01 9.95003104e-01 4.20508564e-01
-2.79189199e-01 3.52281809e-01 -1.00810027e+00 5.65296352e-01
7.75457919e-01 -6.06198534e-02 -4.48809087e-01 -1.14013371e-03
1.63192421e-01 3.26932043e-01 1.59617469e-01 -8.72503102e-01
5.64728118e-02 6.16971962e-02 -2.55251545e-02 -1.60881177e-01
4.80117470e-01 -4.43957657e-01 -4.37870733e-02 7.93303549e-02
-6.15141034e-01 1.65300459e-01 7.89872184e-02 2.71295160e-01
-3.53458256e-01 -5.24365127e-01 3.52521360e-01 -4.72608089e-01
-4.27692980e-01 -4.27837461e-01 -6.17649317e-01 1.67258054e-01
5.78862667e-01 -3.99434775e-01 -1.85929343e-01 -1.00635993e+00
-5.39013803e-01 2.54456013e-01 5.35698235e-01 4.51505393e-01
1.18685648e-01 -7.35266268e-01 -8.79614234e-01 -2.96859652e-01
2.41813466e-01 -3.60307693e-01 1.35111153e-01 7.20840573e-01
-3.79008770e-01 6.11534178e-01 -3.04342099e-02 -3.97527337e-01
-1.53444588e+00 -1.01063140e-01 1.63164258e-01 -2.53642976e-01
-4.74182010e-01 7.30648994e-01 -2.30937004e-01 -4.54274118e-01
4.60531145e-01 2.57374853e-01 -2.69119710e-01 3.07634503e-01
6.03022575e-01 4.20457006e-01 2.10256666e-01 -7.32167959e-01
-5.47084436e-02 -2.07287073e-01 -2.03803867e-01 -6.12696111e-01
1.22648311e+00 -3.55330229e-01 1.19750053e-02 9.05071616e-01
8.61952484e-01 8.89241695e-02 -8.79640400e-01 -6.13668673e-02
2.44475991e-01 -4.97423530e-01 -2.83199817e-01 -1.13442290e+00
-1.28030568e-01 8.22356522e-01 2.17390835e-01 7.74673760e-01
5.70997596e-01 -2.66460776e-01 3.92702788e-01 4.36854303e-01
4.65666264e-01 -1.29268253e+00 3.32336240e-02 4.40808535e-01
1.00813532e+00 -1.45342672e+00 1.41848370e-01 -1.75522283e-01
-9.50033188e-01 8.95775020e-01 5.32725096e-01 2.90733635e-01
1.47569507e-01 -1.40368804e-01 3.48893940e-01 -1.84694499e-01
-1.00349951e+00 -9.67210382e-02 -7.52653033e-02 3.71949822e-01
1.17377126e+00 7.82980490e-03 -9.36311662e-01 2.84669966e-01
-4.35691833e-01 -1.85629040e-01 6.54923022e-01 8.37366164e-01
-4.76578355e-01 -1.32172275e+00 -3.51756692e-01 4.16911751e-01
-6.23489499e-01 2.71322906e-01 -8.12255740e-01 1.11527956e+00
-4.47037548e-01 1.44398594e+00 -2.67174512e-01 -2.84064859e-01
6.77716732e-01 3.53395849e-01 3.12736094e-01 -7.32261837e-01
-1.03023195e+00 1.63886651e-01 9.93605137e-01 -2.70172387e-01
-5.20394683e-01 -8.52535069e-01 -7.16191292e-01 -5.60259879e-01
-6.97606266e-01 7.87418962e-01 7.77308583e-01 1.00052595e+00
-1.13553867e-01 2.64969915e-01 6.02867305e-01 -6.17272973e-01
-8.30763638e-01 -1.29226375e+00 -5.18582761e-01 5.96123040e-01
5.91284782e-03 -4.40089524e-01 -4.25242364e-01 -2.18544468e-01] | [12.867280006408691, 8.050719261169434] |
dee688f4-5867-4f80-9c10-b46891999e25 | seizure-prediction-using-bidirectional-lstm | 1912.06385 | null | https://arxiv.org/abs/1912.06385v1 | https://arxiv.org/pdf/1912.06385v1.pdf | Seizure Prediction Using Bidirectional LSTM | Approximately, 50 million people in the world are affected by epilepsy. For patients, the anti-epileptic drugs are not always useful and these drugs may have undesired side effects on a patient's health. If the seizure is predicted the patients will have enough time to take preventive measures. The purpose of this work is to investigate the application of bidirectional LSTM for seizure prediction. In this paper, we trained EEG data from canines on a double Bidirectional LSTM layer followed by a fully connected layer. The data was provided in the form of a Kaggle competition by American Epilepsy Society. The main task was to classify the interictal and preictal EEG clips. Using this model, we obtained an AUC of 0.84 on the test dataset. Which shows that our classifier's performance is above chance level on unseen data. The comparison with the previous work shows that the use of bidirectional LSTM networks can achieve significantly better results than SVM and GRU networks. | ['Junaid Javed Qureshi', 'Mohammad Farhad Bulbul', 'Hazrat Ali', 'Feroz Karim', 'Adnan Omer Abuassba'] | 2019-12-13 | null | null | null | null | ['seizure-prediction'] | ['medical'] | [-1.99919999e-01 9.32854190e-02 5.70248030e-02 -4.03713256e-01
-4.49013263e-01 -1.53885201e-01 3.34967643e-01 7.59614706e-02
-5.17343938e-01 1.20117879e+00 9.08523425e-02 -3.61926854e-01
1.26339495e-01 -5.13981462e-01 -6.66568816e-01 -6.86930776e-01
-5.77294409e-01 3.05571735e-01 1.84807926e-01 -1.12123333e-01
2.15503365e-01 5.17482162e-01 -1.14677489e+00 7.77195394e-01
5.46143711e-01 1.16233516e+00 2.60596275e-01 3.91119212e-01
2.40683302e-01 1.01943004e+00 -9.40491498e-01 1.48749530e-01
9.10642669e-02 -3.23137194e-01 -9.65167880e-01 -4.66700882e-01
-5.85548989e-02 -3.66708964e-01 -2.82794327e-01 8.32169652e-01
7.16235518e-01 -2.94335723e-01 3.74393344e-01 -9.96184051e-01
-2.38694072e-01 6.28520012e-01 -3.02143902e-01 6.04623079e-01
3.41646999e-01 -8.90601799e-02 5.06360173e-01 -4.80174273e-01
2.82516509e-01 5.85881531e-01 6.16330624e-01 6.10591173e-01
-1.18585777e+00 -9.92929459e-01 -2.09974423e-01 6.24169648e-01
-1.28828371e+00 -3.09372187e-01 5.24106622e-01 -5.84987164e-01
1.35720074e+00 1.35139152e-01 1.02235663e+00 1.39750886e+00
8.39479685e-01 5.44419050e-01 1.41143370e+00 -2.65079468e-01
1.31796658e-01 1.79203093e-01 2.14822680e-01 9.26022455e-02
-4.68557924e-02 4.97560322e-01 -6.23160779e-01 -2.00010478e-01
4.86127675e-01 -1.32317141e-01 -6.65351510e-01 -8.38581398e-02
-1.17766225e+00 7.09743202e-01 7.57786393e-01 1.17378306e+00
-5.63847899e-01 -8.11774358e-02 5.55881500e-01 5.96326590e-01
7.12308705e-01 5.92535436e-01 -5.94238043e-01 -1.57317996e-01
-1.24722588e+00 -1.90603122e-01 6.42485082e-01 9.68291759e-02
5.72429076e-02 1.64966851e-01 1.47881389e-01 8.05261314e-01
-1.11479275e-02 1.77540407e-01 1.18557656e+00 -6.12650020e-03
1.89449787e-01 5.39555609e-01 -9.91421714e-02 -8.07658970e-01
-6.50788844e-01 -6.20326936e-01 -8.05521786e-01 1.53912231e-01
5.41252308e-02 -3.25194746e-01 -1.24331796e+00 1.56429231e+00
-4.91664529e-01 1.75610691e-01 2.24270910e-01 7.73559391e-01
8.62902939e-01 5.94877183e-01 2.86086082e-01 -1.29693344e-01
1.01785231e+00 -5.37292421e-01 -7.72194266e-01 -3.33384454e-01
1.00977516e+00 -5.22140145e-01 4.91016358e-01 5.91830075e-01
-8.11225355e-01 -2.28776008e-01 -1.12531078e+00 2.92729855e-01
-4.44639564e-01 2.13991910e-01 4.89273220e-01 3.52750152e-01
-1.23900318e+00 6.93627357e-01 -1.06956053e+00 -4.99908179e-01
2.98279762e-01 8.51953804e-01 -6.32772326e-01 3.98706287e-01
-1.40593123e+00 1.45464814e+00 5.71570098e-01 2.42889956e-01
-8.45969021e-01 -3.51758361e-01 -4.58619922e-01 -5.01120687e-02
-4.15813476e-01 -1.96480498e-01 9.68219578e-01 -8.30531597e-01
-1.07946563e+00 7.99009979e-01 -1.04832627e-01 -1.16152990e+00
5.58234632e-01 -3.42397764e-02 -4.42109913e-01 -2.05775753e-01
-2.41061244e-02 8.86113346e-01 3.42656374e-01 -6.87720239e-01
-7.54694998e-01 -7.02855110e-01 -3.29397500e-01 -9.78021994e-02
-2.09502637e-01 4.42439951e-02 3.40229690e-01 -6.33782864e-01
-9.41900089e-02 -9.19019938e-01 6.39155135e-02 -8.86880100e-01
-3.73756230e-01 -4.16249156e-01 8.99617195e-01 -9.74527240e-01
1.04304826e+00 -2.01887131e+00 -8.98789912e-02 4.36341390e-03
1.35363758e-01 2.99088210e-01 1.12682581e-01 2.11187452e-01
-6.54637873e-01 7.18643703e-03 -2.18854159e-01 2.37621099e-01
-4.54681963e-01 -1.77458212e-01 -4.75442976e-01 4.79399860e-01
-2.66510367e-01 7.94429958e-01 -5.41521370e-01 1.28806844e-01
9.41243991e-02 6.38148069e-01 -4.55055609e-02 3.61227483e-01
1.45452231e-01 5.02116442e-01 -2.10547343e-01 3.80815566e-01
1.61309645e-01 2.45580152e-02 -7.16703385e-02 -6.02264889e-02
3.36150452e-02 4.17914808e-01 -1.21454120e-01 1.41683424e+00
-3.18150252e-01 1.19775820e+00 -6.38763428e-01 -1.26763797e+00
8.47126842e-01 8.24659884e-01 3.58956993e-01 -9.14399505e-01
2.61584133e-01 5.37856817e-01 7.17808902e-01 -4.51867551e-01
-2.57024854e-01 -1.78067029e-01 3.02335262e-01 2.25418627e-01
2.73294635e-02 2.57721066e-01 -4.67021875e-02 -1.58346981e-01
1.20210147e+00 2.73947529e-02 2.14340493e-01 -5.47731519e-01
3.06825042e-01 -4.86011803e-02 4.55778420e-01 2.83886731e-01
3.30137759e-02 1.09800674e-01 4.63911951e-01 -8.60933721e-01
-6.41785383e-01 -7.07213521e-01 -4.83235896e-01 4.03681129e-01
-4.60949600e-01 1.08685153e-05 -9.78623033e-01 -5.46022117e-01
-3.57829869e-01 9.14327979e-01 -8.49148214e-01 -4.08507764e-01
-3.39746505e-01 -9.71164405e-01 4.15813267e-01 5.98178208e-01
6.96891963e-01 -1.56851602e+00 -1.00620294e+00 4.05130029e-01
-2.29030788e-01 -9.45602059e-01 7.28990734e-02 6.25855327e-01
-9.17307377e-01 -1.01052582e+00 -9.04908836e-01 -1.07230127e+00
4.52295542e-01 -4.27133888e-01 6.61318719e-01 -9.12791342e-02
-2.30145112e-01 -2.61433601e-01 -2.87146986e-01 -4.54436898e-01
-4.48524654e-02 1.92608848e-01 2.32409462e-01 -2.92582154e-01
9.86664653e-01 -7.99450397e-01 -5.99639118e-01 8.32533389e-02
-4.79969203e-01 1.93683431e-02 5.92371881e-01 8.71869147e-01
2.97150344e-01 -1.64774224e-01 1.02055550e+00 -6.27682090e-01
8.03476512e-01 -3.77837211e-01 -4.50173229e-01 1.67214900e-01
-5.11848450e-01 -2.90165044e-04 4.55804139e-01 -4.39894676e-01
-4.60057914e-01 1.73794270e-01 -1.27082974e-01 -2.30340585e-01
-3.04882407e-01 4.77848530e-01 2.61010081e-01 -2.08117932e-01
5.22297084e-01 2.49144152e-01 -5.16628802e-01 -3.89968455e-01
-6.12736166e-01 9.73718584e-01 2.48310894e-01 2.22133964e-01
4.58013592e-03 -4.10560742e-02 -3.56144458e-01 -1.02265871e+00
-4.89838719e-01 -9.74600241e-02 -3.07234913e-01 -1.01568609e-01
9.27078485e-01 -7.92601287e-01 -5.85241437e-01 6.56843245e-01
-1.33220696e+00 -5.45173943e-01 3.28098714e-01 8.96489441e-01
-4.26134616e-01 -1.61649302e-01 -7.39196837e-01 -5.51889241e-01
-6.16577983e-01 -1.44915783e+00 3.68957639e-01 -1.25788435e-01
-5.09619355e-01 -9.64327753e-01 1.52762160e-01 1.79538071e-01
3.54756325e-01 3.80251497e-01 8.71410310e-01 -1.10731757e+00
-7.41714984e-02 -3.53457361e-01 -2.61767711e-02 4.44378406e-01
2.68610954e-01 -5.06708920e-01 -1.21614599e+00 -4.13738370e-01
4.69310939e-01 -4.19751167e-01 8.64829242e-01 7.06994593e-01
1.13530302e+00 -1.67829514e-01 -4.67005342e-01 4.43133801e-01
1.12384284e+00 9.83196974e-01 8.80746007e-01 6.03673041e-01
2.51094908e-01 4.96886373e-01 1.41328340e-02 -5.06090336e-02
1.20182440e-01 5.93350172e-01 1.78082243e-01 -7.05472603e-02
1.50794894e-01 -2.24360786e-02 3.95066470e-01 6.51634574e-01
-2.85568461e-02 -6.99430853e-02 -1.18119848e+00 8.39101791e-01
-1.43962932e+00 -7.72261322e-01 -9.60321352e-02 2.39203811e+00
6.72399521e-01 2.11019650e-01 -9.99383256e-03 2.27712885e-01
5.64142764e-01 -3.61860096e-01 -3.24718654e-01 -6.15313232e-01
3.18026170e-02 4.12596703e-01 4.68531668e-01 2.96439826e-01
-9.54846323e-01 8.09213638e-01 6.49849272e+00 5.05966425e-01
-1.77714467e+00 9.30495411e-02 8.43722701e-01 -2.52624393e-01
2.30975524e-01 -3.50950062e-01 -4.57875997e-01 8.38812649e-01
1.47480309e+00 -2.10930705e-01 3.62739027e-01 4.50906545e-01
3.01821977e-01 -3.59225661e-01 -1.23692679e+00 9.36048090e-01
3.32706720e-02 -8.72435451e-01 -2.61343360e-01 2.75957018e-01
4.50518608e-01 6.71845436e-01 -1.36954740e-01 2.85328954e-01
-4.66891788e-02 -1.35149372e+00 2.73813695e-01 4.37391967e-01
5.61308205e-01 -7.42030144e-01 1.05905294e+00 5.10330915e-01
-6.31113052e-01 -1.24304935e-01 -2.01728970e-01 -8.02846551e-02
-4.02401835e-02 3.54535490e-01 -1.13151515e+00 -4.99131950e-03
1.09165382e+00 6.54570401e-01 -5.91530919e-01 1.33246255e+00
-4.03751194e-01 6.95672870e-01 -2.09408000e-01 -1.71883032e-01
3.69722009e-01 2.37608757e-02 3.14669997e-01 9.81540263e-01
5.41369081e-01 1.56974941e-01 -2.91539788e-01 4.94282961e-01
-3.48034911e-02 1.86484963e-01 -8.76591861e-01 -1.90690354e-01
-1.28074333e-01 8.51720929e-01 -6.59839332e-01 -2.33342275e-01
-1.67067319e-01 8.63980293e-01 2.47454599e-01 2.91096032e-01
-6.02491677e-01 -4.81251568e-01 -8.41934606e-02 5.62788248e-02
-1.62151486e-01 3.61494213e-01 -2.92842895e-01 -1.07030082e+00
7.82596916e-02 -8.09494197e-01 1.61115199e-01 -8.88763607e-01
-8.49508882e-01 1.36799324e+00 -2.26049051e-01 -1.14541566e+00
-5.37312925e-01 -5.42253792e-01 -6.66076124e-01 9.43651140e-01
-1.24620736e+00 -9.28673923e-01 9.81125310e-02 5.36662400e-01
4.37813729e-01 -2.84299672e-01 1.12442005e+00 3.22808892e-01
-2.77794361e-01 2.15811685e-01 -1.78849727e-01 2.12432280e-01
6.62307262e-01 -9.71819818e-01 1.16558120e-01 5.76723516e-01
1.60394669e-01 4.59287554e-01 7.61904180e-01 -6.35984659e-01
-3.96441221e-01 -7.62286484e-01 1.28440893e+00 3.34790871e-02
8.21045518e-01 -2.20065832e-01 -1.04191077e+00 1.00029552e+00
6.51735842e-01 -1.92304790e-01 7.38579512e-01 -1.73919760e-02
2.63585359e-01 -3.79136279e-02 -1.20947289e+00 1.65876120e-01
4.17197794e-01 -3.86882424e-01 -1.03975451e+00 4.19491619e-01
1.63493842e-01 -1.25532195e-01 -8.48583519e-01 7.07604289e-01
5.37395120e-01 -1.00966942e+00 3.87980372e-01 -6.52045190e-01
3.73560190e-01 1.75256819e-01 6.20927773e-02 -1.87542689e+00
7.44145289e-02 -4.41463180e-02 1.91279128e-01 5.00077367e-01
7.46155500e-01 -1.02100074e+00 6.15995705e-01 6.13595009e-01
-7.27089271e-02 -1.10101640e+00 -9.83376682e-01 -6.90990925e-01
1.44306019e-01 -1.73664629e-01 2.14060202e-01 8.58662784e-01
5.59159458e-01 4.66495693e-01 -3.39005202e-01 -1.50414810e-01
4.25268859e-01 -4.86403033e-02 -1.45640643e-03 -1.20999742e+00
1.27002463e-01 -2.31594995e-01 -9.36973214e-01 -3.35444212e-01
3.00173163e-01 -9.42343831e-01 -1.95004210e-01 -1.72716200e+00
1.63236216e-01 -1.57877848e-01 -7.10957468e-01 9.35719371e-01
4.93300349e-01 3.33139360e-01 -1.78057626e-01 2.72389799e-02
1.94584265e-01 2.27386013e-01 7.42489517e-01 -3.33653808e-01
-4.67664480e-01 2.17801139e-01 -2.67899305e-01 8.43646049e-01
1.40347564e+00 -4.76122230e-01 -3.14942986e-01 -4.59442377e-01
-9.51130241e-02 3.08953732e-01 2.40160361e-01 -1.33007479e+00
3.45648676e-01 4.54625607e-01 8.10000479e-01 -5.12969017e-01
4.70364273e-01 -7.56693661e-01 2.21097231e-01 9.20340955e-01
-3.32521349e-01 4.06172201e-02 4.36442524e-01 6.11295402e-02
-5.06435692e-01 8.85410011e-02 7.48374879e-01 -8.95235240e-02
-3.63934070e-01 6.53415024e-02 -6.71529472e-01 -3.96988153e-01
1.20569229e+00 -1.91510901e-01 -1.56565070e-01 -3.42304796e-01
-1.18379176e+00 1.58865765e-01 8.78230408e-02 5.69486737e-01
6.66513145e-01 -1.14865124e+00 -6.12922490e-01 2.51024693e-01
-7.69793317e-02 -6.60302520e-01 9.95918363e-02 1.14819002e+00
-4.64681447e-01 9.82115448e-01 -5.18010437e-01 -4.89748031e-01
-1.37469006e+00 2.33113989e-01 6.82621658e-01 -8.58402848e-02
-6.91641152e-01 8.54604185e-01 1.63520530e-01 2.01763399e-03
2.80820072e-01 -4.47156698e-01 -6.45756245e-01 1.40335664e-01
5.50184608e-01 3.21009718e-02 6.27570510e-01 -6.29465342e-01
-4.02387828e-01 4.90341000e-02 -3.16537499e-01 -5.43781333e-02
1.78839409e+00 6.13765657e-01 -3.90460789e-01 5.23758233e-01
1.24758768e+00 -5.23492277e-01 -4.22935575e-01 4.73276615e-01
1.41069427e-01 -1.00793861e-01 5.26621878e-01 -1.30116796e+00
-1.32113445e+00 1.22624505e+00 1.21235931e+00 4.34127718e-01
1.04019809e+00 -1.77829593e-01 8.53571057e-01 4.39182699e-01
8.00910592e-01 -8.49116623e-01 -5.89224696e-01 2.02507794e-01
9.99934494e-01 -9.00134087e-01 -4.65288639e-01 3.05667907e-01
-5.21726906e-01 1.14031410e+00 5.17486870e-01 -3.34262997e-01
6.07361317e-01 2.09659562e-01 9.60328057e-02 -2.43074939e-01
-8.40911269e-01 3.49222749e-01 2.12593690e-01 3.41717333e-01
7.67738163e-01 3.27753693e-01 -6.51828468e-01 4.80049223e-01
-4.90524203e-01 3.77642781e-01 5.86601615e-01 6.36573851e-01
-4.57790375e-01 -1.10824394e+00 -1.19492806e-01 9.67479646e-01
-8.65627885e-01 -2.16504008e-01 -5.65256536e-01 6.00578010e-01
1.98501851e-02 1.06859267e+00 -5.90312071e-02 -5.11075437e-01
1.31415293e-01 3.54153961e-01 5.31110525e-01 -4.22159433e-01
-6.27568781e-01 3.25285792e-02 5.04198559e-02 -4.22349989e-01
-3.71231556e-01 -4.32079226e-01 -1.17294121e+00 3.19637438e-05
-3.94465119e-01 5.24790585e-01 7.68320382e-01 1.04106617e+00
2.41002172e-01 5.99613667e-01 4.59556073e-01 -6.07883692e-01
-6.91796914e-02 -1.42679560e+00 -7.18199134e-01 -3.35629508e-02
4.75132108e-01 -6.57948971e-01 -5.88408947e-01 -4.68061976e-02] | [13.225833892822266, 3.535714864730835] |
844cca18-ebb8-4f84-9fc2-2789fbe8e2e1 | ablation-study-of-how-run-time-assurance | 2207.04117 | null | https://arxiv.org/abs/2207.04117v1 | https://arxiv.org/pdf/2207.04117v1.pdf | Ablation Study of How Run Time Assurance Impacts the Training and Performance of Reinforcement Learning Agents | Reinforcement Learning (RL) has become an increasingly important research area as the success of machine learning algorithms and methods grows. To combat the safety concerns surrounding the freedom given to RL agents while training, there has been an increase in work concerning Safe Reinforcement Learning (SRL). However, these new and safe methods have been held to less scrutiny than their unsafe counterparts. For instance, comparisons among safe methods often lack fair evaluation across similar initial condition bounds and hyperparameter settings, use poor evaluation metrics, and cherry-pick the best training runs rather than averaging over multiple random seeds. In this work, we conduct an ablation study using evaluation best practices to investigate the impact of run time assurance (RTA), which monitors the system state and intervenes to assure safety, on effective learning. By studying multiple RTA approaches in both on-policy and off-policy RL algorithms, we seek to understand which RTA methods are most effective, whether the agents become dependent on the RTA, and the importance of reward shaping versus safe exploration in RL agent training. Our conclusions shed light on the most promising directions of SRL, and our evaluation methodology lays the groundwork for creating better comparisons in future SRL work. | ['Kerianne L Hobbs', 'Taylor T Johnson', 'Kyle Dunlap', 'Nathaniel Hamilton'] | 2022-07-08 | null | null | null | null | ['safe-exploration'] | ['robots'] | [ 1.35553390e-01 1.32472053e-01 -6.09805346e-01 -1.13952033e-01
-7.68238902e-01 -8.77928317e-01 8.84890199e-01 2.23563790e-01
-8.43849242e-01 1.09039581e+00 1.93670485e-02 -7.77473807e-01
-3.43926549e-01 -5.39623260e-01 -6.89938188e-01 -9.05120969e-01
-5.82153678e-01 1.04016431e-01 5.31255342e-02 -1.79950848e-01
4.21194315e-01 4.98346090e-01 -1.54520595e+00 -4.79011089e-01
7.87703335e-01 5.74924886e-01 -3.82093370e-01 5.67291856e-01
4.85970259e-01 1.12379897e+00 -8.82528841e-01 1.97334252e-02
3.98313791e-01 -5.96394956e-01 -6.45261109e-01 -3.18269610e-01
1.32842794e-01 -7.08302379e-01 -1.08899243e-01 9.44139004e-01
5.78771353e-01 3.78563613e-01 1.89774543e-01 -1.64700878e+00
-1.52840212e-01 8.85454834e-01 -5.92164695e-01 4.10730720e-01
1.79272652e-01 8.12010884e-01 7.94071198e-01 1.42493397e-01
3.58322442e-01 1.39217174e+00 2.72837698e-01 7.89613307e-01
-1.20866001e+00 -8.80061865e-01 4.32963103e-01 6.69731647e-02
-9.23578739e-01 -5.55455983e-01 2.92380661e-01 -1.50516093e-01
8.75696898e-01 8.39001015e-02 6.47317111e-01 1.25591767e+00
4.87191647e-01 7.89553702e-01 1.51426530e+00 -4.54737991e-01
7.91857839e-01 1.02603473e-01 -6.75064027e-02 4.90281165e-01
5.50461948e-01 1.07720828e+00 -2.69078642e-01 -4.48825270e-01
8.64522874e-01 -3.53248507e-01 -9.18273777e-02 -4.59251255e-01
-8.15505564e-01 1.06937921e+00 8.68784562e-02 3.99963707e-02
-3.68493855e-01 4.90525573e-01 5.03030837e-01 4.82145190e-01
-9.64476392e-02 1.12798703e+00 -3.33646804e-01 -7.03231633e-01
-4.53884751e-01 5.87031186e-01 7.59830236e-01 4.14685965e-01
4.18972015e-01 4.54766363e-01 -3.24322253e-01 3.57566684e-01
1.55072883e-01 3.10441703e-01 2.71381885e-01 -1.29518163e+00
2.47747093e-01 2.24444285e-01 5.18088698e-01 -4.64351147e-01
-3.26275021e-01 -4.29313153e-01 -2.12442558e-02 9.36137736e-01
4.41592753e-01 -6.76282525e-01 -5.21834612e-01 1.96389866e+00
2.43095860e-01 2.01116838e-02 3.16350609e-01 7.57579505e-01
-1.87649474e-01 4.60821062e-01 2.94677705e-01 -5.54274440e-01
7.98327506e-01 -7.47113407e-01 -5.53860724e-01 -3.33972901e-01
7.43614614e-01 -4.03073400e-01 1.22925985e+00 6.22677207e-01
-1.04738450e+00 -1.21468589e-01 -1.20817149e+00 7.27399409e-01
-1.08960420e-01 -4.34220135e-01 7.71853030e-01 9.42540467e-01
-8.64976943e-01 9.09978747e-01 -1.14148414e+00 -2.21571013e-01
3.84984106e-01 4.67032939e-01 2.10595965e-01 1.59179047e-01
-1.09073818e+00 1.23093235e+00 2.38080651e-01 -3.33364420e-02
-1.50910079e+00 -4.98600543e-01 -7.53706217e-01 -2.14162797e-01
9.78948891e-01 -1.27407223e-01 1.67676473e+00 -8.98090422e-01
-1.80749047e+00 6.59273341e-02 4.33890343e-01 -8.41302693e-01
5.04385769e-01 -2.71134317e-01 -3.47505063e-01 -1.84195206e-01
-9.13733989e-02 5.69529772e-01 8.74045193e-01 -1.36010325e+00
-6.89917386e-01 -1.09579995e-01 4.53708708e-01 2.98928291e-01
-1.91459828e-03 -7.09213838e-02 4.16022897e-01 -3.93918693e-01
-7.46520400e-01 -1.19077718e+00 -4.04737473e-01 -5.30659974e-01
-6.05215132e-02 -2.01471984e-01 7.08346188e-01 -3.70616056e-02
1.35968256e+00 -2.11278391e+00 -1.38823792e-01 2.63954610e-01
-1.10660404e-01 4.24054414e-01 -3.17127287e-01 5.23687184e-01
1.46621019e-01 2.30696678e-01 3.49186398e-02 2.59635806e-01
2.05876872e-01 4.38973039e-01 -6.19017422e-01 5.52966535e-01
4.36492339e-02 5.58651328e-01 -1.17931259e+00 -3.55139077e-01
2.87263185e-01 2.15723991e-01 -4.60906565e-01 3.57396454e-01
-3.84014398e-01 5.44256866e-01 -7.08839417e-01 5.28541744e-01
4.72231135e-02 8.83247256e-02 1.67906418e-01 3.95741463e-01
-4.38142151e-01 4.39413309e-01 -1.01189113e+00 9.65609789e-01
-4.44539338e-01 3.31840098e-01 -1.51663711e-02 -6.61870658e-01
6.58035636e-01 3.05879176e-01 4.34818625e-01 -1.14616001e+00
2.57778764e-01 -1.35932583e-02 4.09742594e-01 -3.65762979e-01
3.92456859e-01 -6.90580457e-02 -1.63909703e-01 7.90023088e-01
-4.55402672e-01 -1.78805098e-01 3.25983405e-01 -6.87615126e-02
1.23742294e+00 3.76976848e-01 4.27713037e-01 -2.72026807e-01
8.50815848e-02 1.25513405e-01 5.69934249e-01 1.04290104e+00
-6.21819258e-01 -3.14506769e-01 7.83385873e-01 -3.73909384e-01
-7.59834290e-01 -7.82876372e-01 -2.49850061e-02 1.14290905e+00
2.20121294e-01 -1.92975506e-01 -7.08940923e-01 -9.97360051e-01
1.91388726e-01 1.31271231e+00 -7.24841714e-01 -6.67071283e-01
-6.03977561e-01 -7.50135005e-01 7.62067020e-01 4.97361153e-01
3.56583804e-01 -1.33455944e+00 -1.49782765e+00 3.31596732e-01
3.33291352e-01 -5.98640978e-01 -4.04352367e-01 4.48170125e-01
-7.94602275e-01 -1.09326553e+00 -2.00165436e-01 -8.66238624e-02
4.68504012e-01 1.09242775e-01 9.23722446e-01 5.08200191e-02
-1.44183546e-01 7.21564412e-01 -5.20541310e-01 -4.95942831e-01
-6.89356565e-01 -1.59355015e-01 3.60540032e-01 -5.60236096e-01
1.59256846e-01 -3.15881640e-01 -5.34977019e-01 4.14515942e-01
-7.79586911e-01 -5.35257578e-01 5.90831995e-01 7.28191972e-01
3.26167941e-01 2.38161743e-01 7.87664950e-01 -6.91961825e-01
1.10323822e+00 -3.70155394e-01 -1.11136508e+00 3.16292673e-01
-1.16107845e+00 6.26290977e-01 6.42539501e-01 -6.84711456e-01
-8.36777329e-01 -4.05698836e-01 1.69470757e-01 -3.06662291e-01
-7.63538405e-02 7.06731752e-02 2.60629028e-01 -2.34179676e-01
8.35925341e-01 -4.26920429e-02 3.19865763e-01 5.66253439e-02
1.83830813e-01 2.79149175e-01 -6.12230673e-02 -9.93559539e-01
6.57763004e-01 1.19674671e-02 -1.35342687e-01 -5.18847167e-01
-6.32274806e-01 2.14166120e-01 2.27229983e-01 -3.15020770e-01
4.93666977e-01 -5.29021621e-01 -1.16038811e+00 1.73980314e-02
-3.58735770e-01 -8.73996437e-01 -5.91774881e-01 4.66303438e-01
-7.09524095e-01 9.80818942e-02 -2.99463809e-01 -1.25053477e+00
-1.06661260e-01 -1.49907482e+00 5.30177593e-01 5.85294962e-01
-4.96011794e-01 -7.06522882e-01 4.18833911e-01 1.05307736e-01
5.18024802e-01 2.85439879e-01 9.51469064e-01 -5.93477726e-01
-5.52709639e-01 1.83193132e-01 2.59251595e-01 2.10663050e-01
1.13882385e-01 -3.49376281e-03 -7.53137290e-01 -8.50102842e-01
-9.89184231e-02 -8.58847439e-01 3.07939887e-01 3.66988301e-01
8.86911333e-01 -6.89181030e-01 -8.09190515e-03 2.04094350e-01
1.28090191e+00 7.34509051e-01 4.16300207e-01 9.06641424e-01
9.37724859e-02 5.03495276e-01 1.12296522e+00 7.22257793e-01
-1.21249643e-03 4.56930161e-01 6.36358798e-01 2.05467746e-01
3.51007968e-01 -2.98001498e-01 9.67372417e-01 -2.39094850e-02
-4.25307378e-02 -1.00728974e-01 -5.96653581e-01 1.43201321e-01
-1.78676009e+00 -9.70794439e-01 6.04433358e-01 2.75860190e+00
9.74086642e-01 4.74058360e-01 5.76022089e-01 2.38208920e-02
4.02838677e-01 1.35198072e-01 -1.01427960e+00 -8.19323659e-01
2.85388619e-01 1.58869475e-01 7.66397655e-01 6.47258162e-01
-8.47442389e-01 8.24623883e-01 6.81586075e+00 6.27842963e-01
-1.17723382e+00 -3.25118810e-01 5.88841140e-01 -2.57741123e-01
-1.21610768e-01 2.32279435e-01 -7.13250339e-01 2.50401825e-01
1.13868546e+00 -3.05077583e-01 9.34825301e-01 9.70427632e-01
5.27193129e-01 -5.40664554e-01 -1.23809278e+00 2.81493515e-01
-3.15745175e-01 -7.68592715e-01 -3.31709117e-01 1.97238728e-01
5.12089908e-01 -6.93092868e-02 1.74457163e-01 7.91306376e-01
9.27945554e-01 -1.14983344e+00 7.56410420e-01 7.10965991e-02
3.90481651e-01 -1.11607897e+00 5.20055354e-01 2.65082389e-01
-7.55486786e-01 -3.94484192e-01 2.26080008e-02 -2.58073300e-01
-2.62166232e-01 -1.77247480e-01 -9.80432153e-01 3.52143161e-02
4.22482699e-01 1.66091904e-01 -5.30719757e-01 8.96315694e-01
-3.88887227e-01 9.21685457e-01 -2.00578496e-01 -4.79771465e-01
6.78193867e-01 -5.96795268e-02 4.51021820e-01 6.20443761e-01
-1.64806083e-01 -1.52174577e-01 3.38644147e-01 7.07014501e-01
5.46464741e-01 -2.34211132e-01 -6.36485398e-01 -4.74347562e-01
6.45576179e-01 9.90972996e-01 -7.76043594e-01 7.01915249e-02
4.38961089e-02 1.94288567e-01 3.22852701e-01 4.21654642e-01
-9.15159523e-01 -1.10956989e-01 7.48796582e-01 -8.56504291e-02
-1.33701274e-02 -3.19901377e-01 -1.34273693e-01 -4.79328483e-01
-2.73996323e-01 -1.40255487e+00 4.89571840e-01 -2.90872723e-01
-9.86529410e-01 4.32577640e-01 3.24885488e-01 -1.12309945e+00
-6.02827251e-01 -1.36484846e-01 -4.86663878e-01 3.76911789e-01
-1.52592134e+00 -2.04139307e-01 2.66832054e-01 3.79402846e-01
4.30082709e-01 -2.30383333e-02 7.21370220e-01 -2.01933756e-01
-8.72915089e-01 7.58748829e-01 1.62708730e-01 -3.57757449e-01
6.98200643e-01 -1.11035919e+00 3.52255516e-02 7.23850608e-01
-1.82543531e-01 6.06887698e-01 1.02138948e+00 -6.59931183e-01
-1.67980397e+00 -8.18676710e-01 -1.24219447e-01 -4.28415000e-01
6.59287810e-01 6.66163256e-03 -8.08135509e-01 6.22516751e-01
2.62456000e-01 -4.25119787e-01 3.41289163e-01 2.17811480e-01
-1.90407231e-01 -1.54148042e-01 -1.16485822e+00 9.84287739e-01
5.11879146e-01 -6.46087155e-02 -3.70642930e-01 -9.13478807e-02
7.05173135e-01 -1.77852452e-01 -8.35015178e-01 3.79842699e-01
3.70211393e-01 -9.66003418e-01 6.99793637e-01 -5.41958332e-01
1.29352491e-02 -2.74045080e-01 1.35741264e-01 -1.47298801e+00
-8.77185613e-02 -1.04232144e+00 -4.46986221e-02 8.50397885e-01
3.68646592e-01 -7.14590132e-01 7.46437669e-01 7.60115802e-01
1.04960024e-01 -9.15142238e-01 -8.09851110e-01 -1.18457079e+00
1.81482852e-01 -1.55141950e-01 5.26360452e-01 8.25284183e-01
1.88975722e-01 2.01955169e-01 -4.93356615e-01 6.98314160e-02
6.64797306e-01 9.94916540e-03 6.84458911e-01 -5.90205908e-01
-4.96223450e-01 -7.17603624e-01 2.02066243e-01 -4.40725565e-01
1.38945833e-01 -2.70173311e-01 3.27013999e-01 -1.00186515e+00
-9.32171047e-02 -6.96000040e-01 -3.75272781e-01 8.18587780e-01
-2.14581370e-01 -4.94102240e-01 2.98042595e-01 -2.45143846e-02
-8.50648820e-01 7.63609827e-01 1.28454340e+00 1.59786806e-01
-6.73318088e-01 1.25818327e-01 -7.76614010e-01 4.90580469e-01
1.11750329e+00 -4.72677857e-01 -8.60510767e-01 -5.57010099e-02
2.43821308e-01 3.00691783e-01 3.14135283e-01 -9.06940579e-01
-1.21157682e-02 -7.99155176e-01 -1.14728421e-01 1.54823139e-02
-4.59309742e-02 -7.76721597e-01 -1.25580177e-01 7.54104614e-01
-7.72380471e-01 2.75572360e-01 4.96728480e-01 5.00212073e-01
2.93111682e-01 -2.40622461e-01 8.89552593e-01 -1.52188279e-02
-5.60833216e-01 1.39486656e-01 -6.27135754e-01 4.53489184e-01
1.32453227e+00 -2.13890538e-01 -3.14668655e-01 -3.93302798e-01
-1.84921637e-01 7.02377498e-01 5.08659065e-01 5.35139978e-01
3.39239299e-01 -8.04959476e-01 -3.40106547e-01 5.67974038e-02
1.58359250e-03 -2.88195938e-01 -9.00224373e-02 5.84652424e-01
-1.51469996e-02 2.37457335e-01 -3.13937426e-01 -1.60744682e-01
-9.78176594e-01 6.78002238e-01 4.82032269e-01 -3.47628683e-01
-6.39353693e-01 4.91522193e-01 -1.56458020e-01 -1.14727437e-01
8.05604756e-01 -7.99537823e-02 -7.25872144e-02 -2.86841184e-01
4.87948745e-01 6.39575362e-01 -1.64436191e-01 6.66224882e-02
-2.55137801e-01 7.04244450e-02 -4.28117841e-01 -4.18699205e-01
1.07517397e+00 2.11751938e-01 3.46037477e-01 4.70851839e-01
5.17204225e-01 -1.78368911e-01 -1.88343251e+00 2.28892252e-01
1.04258798e-01 -4.45053995e-01 1.90214962e-01 -1.08524656e+00
-7.96602428e-01 4.87594426e-01 7.07080185e-01 4.65623111e-01
9.89818037e-01 -3.98186356e-01 3.08049709e-01 5.08812010e-01
7.48562694e-01 -1.32795048e+00 1.15124144e-01 3.86645943e-01
7.03752458e-01 -1.25538301e+00 1.88019067e-01 3.42510343e-01
-9.74020660e-01 8.71954143e-01 1.00705349e+00 -1.54923260e-01
1.22901827e-01 4.79017437e-01 2.93865930e-02 -3.94183062e-02
-1.00838268e+00 -3.99351157e-02 -5.67232788e-01 5.17264903e-01
8.58318582e-02 7.10859522e-02 -2.28681177e-01 1.99364386e-02
-1.27216518e-01 4.53659445e-02 7.39649713e-01 1.45683241e+00
-5.72298944e-01 -1.35747755e+00 -5.01949668e-01 3.94693255e-01
-3.21562231e-01 2.38676384e-01 -3.20722610e-01 1.20316446e+00
-2.01927736e-01 1.09238756e+00 -1.66409343e-01 -2.77013421e-01
3.24955225e-01 -1.20187081e-01 6.11882150e-01 -3.34058911e-01
-7.08061099e-01 -9.37768742e-02 2.62477666e-01 -8.17003071e-01
-2.73434609e-01 -8.02014112e-01 -1.43358779e+00 -3.40914339e-01
-2.31672421e-01 4.07806545e-01 4.53089178e-01 1.08150721e+00
2.43294761e-01 5.98563731e-01 8.56958568e-01 -5.38034439e-01
-1.29477310e+00 -4.76100415e-01 -4.39627796e-01 -4.77445610e-02
5.75690866e-01 -9.72081602e-01 -5.09999812e-01 -6.17329419e-01] | [4.504345417022705, 2.1707725524902344] |
646bd43e-5322-40ea-b255-932afe0ad638 | learning-to-play-the-chess-variant-crazyhouse | 1908.06660 | null | https://arxiv.org/abs/1908.06660v2 | https://arxiv.org/pdf/1908.06660v2.pdf | Learning to play the Chess Variant Crazyhouse above World Champion Level with Deep Neural Networks and Human Data | Deep neural networks have been successfully applied in learning the board games Go, chess and shogi without prior knowledge by making use of reinforcement learning. Although starting from zero knowledge has been shown to yield impressive results, it is associated with high computationally costs especially for complex games. With this paper, we present CrazyAra which is a neural network based engine solely trained in supervised manner for the chess variant crazyhouse. Crazyhouse is a game with a higher branching factor than chess and there is only limited data of lower quality available compared to AlphaGo. Therefore, we focus on improving efficiency in multiple aspects while relying on low computational resources. These improvements include modifications in the neural network design and training configuration, the introduction of a data normalization step and a more sample efficient Monte-Carlo tree search which has a lower chance to blunder. After training on 569,537 human games for 1.5 days we achieve a move prediction accuracy of 60.4%. During development, versions of CrazyAra played professional human players. Most notably, CrazyAra achieved a four to one win over 2017 crazyhouse world champion Justin Tan (aka LM Jann Lee) who is more than 400 Elo higher rated compared to the average player in our training set. Furthermore, we test the playing strength of CrazyAra on CPU against all participants of the second Crazyhouse Computer Championships 2017, winning against twelve of the thirteen participants. Finally, for CrazyAraFish we continue training our model on generated engine games. In ten long-time control matches playing Stockfish 10, CrazyAraFish wins three games and draws one out of ten matches. | ['Johannes Fürnkranz', 'Johannes Czech', 'Moritz Willig', 'Kristian Kersting', 'Alena Beyer'] | 2019-08-19 | null | null | null | null | ['board-games'] | ['playing-games'] | [-3.56451511e-01 -2.34829523e-02 1.37620121e-01 1.27553836e-01
-7.85446584e-01 -5.01371443e-01 1.65280268e-01 -2.10697725e-01
-1.03092396e+00 8.37281346e-01 -3.45475584e-01 -6.42486393e-01
-3.93316805e-01 -1.07823622e+00 -8.73713911e-01 -3.88982892e-01
-1.37039572e-01 7.64487088e-01 3.88679415e-01 -8.49261880e-01
2.20705673e-01 2.45777547e-01 -1.54318500e+00 3.53850543e-01
5.81670582e-01 8.27662051e-01 1.61842003e-01 1.22787786e+00
5.69211006e-01 1.23030615e+00 -8.31254661e-01 -7.93090999e-01
7.38735616e-01 -2.00597078e-01 -7.62862742e-01 -7.67811835e-01
3.26871544e-01 -5.27679384e-01 -3.78922313e-01 6.26411438e-01
6.85320735e-01 2.56731957e-01 1.94866553e-01 -1.16478157e+00
2.20471233e-01 1.14527023e+00 -4.52386051e-01 3.73284280e-01
-4.97011319e-02 5.35035014e-01 1.16491413e+00 -3.83653551e-01
4.57125038e-01 5.26699603e-01 1.11757982e+00 4.58853304e-01
-8.00491095e-01 -1.04635251e+00 -6.26794457e-01 2.95972854e-01
-1.23967659e+00 -2.44374767e-01 1.71389282e-01 -3.27059418e-01
1.38079464e+00 4.05195132e-02 1.05501485e+00 9.52069402e-01
2.60473639e-01 6.64517403e-01 9.37807143e-01 -4.17191714e-01
3.59895498e-01 -2.65744120e-01 -1.68241113e-01 7.40234137e-01
2.01790631e-01 6.54066980e-01 -4.13846731e-01 9.97615531e-02
7.81971514e-01 -3.84936094e-01 2.91119635e-01 -2.28172794e-01
-7.84546673e-01 1.04053223e+00 5.21062493e-01 2.84065187e-01
-3.51434201e-01 6.96173608e-01 5.27496755e-01 3.52328598e-01
-1.96284622e-01 1.03455555e+00 -5.32418728e-01 -1.00749373e+00
-1.13170540e+00 7.81415999e-01 9.94183540e-01 5.31326473e-01
4.41884041e-01 3.94933105e-01 1.68506384e-01 4.06950831e-01
-3.20197493e-01 1.40278056e-01 6.17247283e-01 -1.03505456e+00
4.36094761e-01 4.09348518e-01 -2.93194316e-02 -8.33714545e-01
-7.27523446e-01 -7.56321490e-01 -7.02402234e-01 9.73999441e-01
8.29340518e-01 -5.49368143e-01 -6.54232621e-01 1.46217835e+00
-1.41011178e-01 9.86052603e-02 5.01591749e-02 9.01545227e-01
6.59339309e-01 4.76352811e-01 -8.99047628e-02 5.38795948e-01
1.22057891e+00 -9.97707069e-01 2.72952151e-02 -2.47788012e-01
8.25619876e-01 -5.54861844e-01 9.71686304e-01 1.03984308e+00
-1.49612856e+00 -6.61828637e-01 -1.47202301e+00 2.04122782e-01
-2.23553032e-01 1.96038902e-01 1.16347575e+00 1.05941057e+00
-9.14758503e-01 1.09038007e+00 -1.01788366e+00 1.69837490e-01
3.59723181e-01 7.91944206e-01 -2.88076282e-01 2.22124562e-01
-1.35903466e+00 1.08110642e+00 8.47624302e-01 -2.96063125e-01
-1.12017286e+00 -7.81461775e-01 -6.10442162e-01 2.66054124e-01
8.44814718e-01 -4.13222432e-01 1.61601353e+00 -7.37345755e-01
-1.83478570e+00 8.17059994e-01 8.20891798e-01 -1.11296296e+00
8.21713448e-01 -1.53639868e-01 -2.91059792e-01 -2.49763161e-01
7.20817875e-03 4.69188541e-01 1.78479552e-01 -4.36447769e-01
-1.04741597e+00 1.27153366e-03 5.56820452e-01 2.38095060e-01
1.21144176e-01 -4.75193672e-02 -4.35567677e-01 -3.94162238e-01
-4.44899768e-01 -9.81603265e-01 -2.61225283e-01 -8.21376145e-01
-1.02095395e-01 -1.97318405e-01 -5.37314452e-02 -6.31924987e-01
1.18276012e+00 -1.81783473e+00 -1.26841947e-01 2.96499819e-01
3.97739530e-01 4.65372056e-01 4.55182716e-02 3.39739740e-01
-2.71461815e-01 -6.52969778e-02 3.67888808e-01 1.16228312e-01
1.61475897e-01 4.32171002e-02 1.40297115e-01 1.66095674e-01
-2.02117190e-01 9.98518467e-01 -7.76242793e-01 4.50043706e-03
-1.79252438e-02 -1.50633290e-01 -1.03862333e+00 -4.27811593e-02
3.59380106e-03 -9.73102376e-02 -2.56794388e-03 4.82316077e-01
3.36838514e-01 -2.33738068e-02 -1.57408544e-03 3.33719701e-01
-1.94135070e-01 4.08138573e-01 -1.23152530e+00 1.77560902e+00
-6.03819013e-01 5.85000098e-01 -1.97650373e-01 -8.48111808e-01
6.67136669e-01 -1.56563222e-02 2.29590878e-01 -7.82038450e-01
5.93116701e-01 2.58957475e-01 7.86669433e-01 -5.25507890e-02
1.07065797e+00 -2.27801010e-01 -3.39828610e-01 4.49098140e-01
1.94446251e-01 -2.81225592e-01 7.02023745e-01 2.46211722e-01
1.52857995e+00 2.37271667e-01 2.48739943e-01 -8.29716213e-03
-1.74615402e-02 3.78097326e-01 4.88397241e-01 1.30254292e+00
-1.31037742e-01 3.98121625e-01 5.96715569e-01 -5.75548112e-01
-1.36029673e+00 -8.58349025e-01 3.84292960e-01 1.48235059e+00
-1.57663062e-01 -8.78816009e-01 -9.48851287e-01 -1.83767006e-01
-2.81946868e-01 8.16221178e-01 -6.13031030e-01 -4.64945376e-01
-6.50030017e-01 -6.08670056e-01 1.29439485e+00 6.37014866e-01
8.71278822e-01 -1.16662371e+00 -1.16762865e+00 3.75856638e-01
1.54045463e-01 -6.58798635e-01 -1.13033660e-01 4.96950656e-01
-4.14605618e-01 -1.24228299e+00 -4.64517921e-01 -4.38366324e-01
-2.75698125e-01 -2.33783305e-01 1.49721718e+00 2.55639315e-01
-5.06023586e-01 -1.12413637e-01 -3.68540406e-01 -5.41638792e-01
-4.13535506e-01 7.21149385e-01 1.19321436e-01 -9.94456172e-01
3.79356384e-01 -6.06691778e-01 -3.98591012e-01 2.82239377e-01
-5.08560240e-01 3.72980207e-01 8.64810705e-01 1.16249597e+00
-5.21234088e-02 5.12458503e-01 -2.04776265e-02 -7.68907845e-01
6.93254411e-01 -2.01346949e-01 -8.70976865e-01 -1.58419102e-01
-5.58889449e-01 -4.63920794e-02 7.13743567e-01 -2.73648858e-01
-7.18623996e-01 -7.49678090e-02 -3.59235406e-01 -2.57910162e-01
1.42825618e-01 4.62409794e-01 2.30558336e-01 -9.38764513e-02
1.32146263e+00 -7.17602968e-02 -8.64817202e-02 -1.24853082e-01
1.52488157e-01 2.06012905e-01 7.51591206e-01 -6.96232021e-01
7.79810429e-01 -1.37232170e-01 -1.25675350e-01 -2.81854779e-01
-3.88178051e-01 -9.44030061e-02 -3.16360444e-01 -2.74490923e-01
7.21273482e-01 -8.77698064e-01 -1.72156155e+00 8.46573055e-01
-6.69348896e-01 -8.42829049e-01 -3.86627316e-01 4.49870676e-01
-5.94586313e-01 -1.84963360e-01 -7.25428581e-01 -5.90389609e-01
-2.29752928e-01 -1.13613856e+00 2.20581189e-01 3.74148279e-01
-3.30589116e-01 -5.52882791e-01 2.40246147e-01 6.08534813e-01
3.48800272e-01 1.81500614e-01 5.36660790e-01 -7.33660221e-01
-2.72252053e-01 -4.04342443e-01 1.10839084e-01 3.41484964e-01
-5.12284219e-01 -1.31651565e-01 -5.96935272e-01 -3.34349781e-01
-4.36693162e-01 -6.92746341e-01 6.56712055e-01 2.82743573e-01
8.43251824e-01 1.72414288e-01 8.43886733e-02 6.52046323e-01
1.34329045e+00 4.46354151e-01 7.19591558e-01 1.06085646e+00
3.87067199e-01 1.26473203e-01 5.02788246e-01 5.49449027e-01
1.31988809e-01 7.43474185e-01 5.71388900e-01 1.19032919e-01
1.34363040e-01 -3.99630904e-01 4.47894633e-01 4.14168209e-01
-8.20651591e-01 1.49336252e-02 -1.10421968e+00 4.10802037e-01
-1.56335521e+00 -1.16127062e+00 -1.27590239e-01 2.05985880e+00
8.71469676e-01 8.64022970e-01 6.03957534e-01 4.31177855e-01
3.40427786e-01 -3.28873575e-01 -4.83260989e-01 -6.80058241e-01
5.08280396e-02 1.07537925e+00 1.07984149e+00 3.28700960e-01
-8.65367770e-01 1.26416266e+00 5.97962475e+00 1.30307376e+00
-8.15242469e-01 8.98895785e-03 4.36273217e-01 -7.03926563e-01
4.22021568e-01 1.19514622e-01 -7.80252278e-01 3.49990547e-01
1.10725892e+00 -1.43896610e-01 8.37500334e-01 1.25691593e+00
-1.29730463e-01 -5.48053145e-01 -7.91923285e-01 1.01382494e+00
-2.59474218e-01 -1.69355464e+00 -5.44237196e-01 2.84371436e-01
6.62374854e-01 1.41447052e-01 1.91375073e-02 1.09901083e+00
1.17283261e+00 -1.52456748e+00 9.57584381e-01 2.09564075e-01
6.31028533e-01 -1.34784293e+00 1.01703537e+00 5.91986239e-01
-7.27642119e-01 -1.75915092e-01 -3.79845947e-01 -8.47440302e-01
-2.28808045e-01 8.32424685e-02 -9.97422576e-01 5.32462716e-01
9.17984426e-01 4.98283356e-02 -4.78087217e-01 1.09630024e+00
-3.21862638e-01 8.50547552e-01 -4.51704681e-01 -2.51347512e-01
6.77387774e-01 9.97988135e-02 2.16801226e-01 7.71174133e-01
2.18957528e-01 2.79800951e-01 -1.45374015e-01 6.17762685e-01
4.83704843e-02 -1.68327406e-01 -1.28434762e-01 4.77928817e-02
2.81966060e-01 1.23898613e+00 -6.58188283e-01 -1.96000770e-01
1.27879456e-01 8.04291368e-01 4.18436408e-01 -3.61228049e-01
-1.03783286e+00 -6.55108273e-01 4.60046679e-01 2.92910337e-01
4.18580621e-01 -1.15351640e-01 -4.13698375e-01 -7.41463184e-01
-3.78526747e-01 -1.33805394e+00 3.69270593e-01 -7.82374680e-01
-7.96132743e-01 6.69910967e-01 -1.56009585e-01 -9.42801654e-01
-6.73084736e-01 -9.14026797e-01 -7.74803817e-01 8.71641219e-01
-6.60596371e-01 -7.48692214e-01 -2.79924423e-01 4.72448617e-01
2.45197847e-01 -7.99578488e-01 6.02079153e-01 1.94827318e-01
-4.81947243e-01 1.00265872e+00 1.80284619e-01 3.11709821e-01
2.33480170e-01 -1.38668978e+00 6.11482561e-01 5.26489556e-01
1.08137019e-01 5.35106838e-01 7.87260711e-01 -4.79311854e-01
-1.06618845e+00 -4.23423886e-01 1.51062444e-01 -3.05219471e-01
7.37048149e-01 -2.19817132e-01 -4.47549105e-01 5.65554321e-01
1.42075568e-01 -6.18875980e-01 5.81379950e-01 4.48388070e-01
-5.99652231e-02 9.88297015e-02 -9.13384974e-01 7.53220856e-01
8.48698854e-01 -8.52200612e-02 -5.29878914e-01 -7.45643601e-02
1.16993092e-01 -1.09706771e+00 -6.48156047e-01 3.55005004e-02
7.00377762e-01 -1.49077785e+00 7.98709989e-01 -7.21867085e-01
7.77848065e-01 -6.64907172e-02 1.63649395e-01 -1.43602514e+00
-3.66906524e-01 -6.41595304e-01 4.08480495e-01 7.39696383e-01
3.37230712e-01 -1.56831965e-01 1.52738941e+00 5.08655488e-01
-1.26948670e-01 -7.25082695e-01 -9.16269839e-01 -9.09372449e-01
5.24771690e-01 -9.12523746e-01 4.18793082e-01 6.69352114e-01
1.59162417e-01 3.11094433e-01 -7.95305192e-01 -3.60141546e-01
2.52592236e-01 -4.35195744e-01 1.28085732e+00 -1.00642776e+00
-1.18925416e+00 -7.71689892e-01 -6.24415815e-01 -7.41490841e-01
-2.18079209e-01 -9.54125047e-01 -4.29076701e-02 -1.00713027e+00
1.22755952e-01 -4.93036538e-01 -1.27272978e-02 6.73069656e-01
1.67695016e-01 4.71576095e-01 3.49553734e-01 -9.12298262e-03
-4.39454585e-01 4.38080840e-02 9.21935141e-01 4.32119593e-02
-1.48473233e-01 2.48238161e-01 -7.66826332e-01 8.12796772e-01
9.58592355e-01 -4.58156288e-01 -1.60048097e-01 -1.77147701e-01
8.74522984e-01 3.15734074e-02 3.00385714e-01 -1.65120029e+00
3.85831207e-01 3.46321575e-02 3.94605756e-01 -4.07534808e-01
2.68874228e-01 -4.47896451e-01 3.42585146e-01 8.74442756e-01
-9.02453884e-02 1.60258934e-01 5.82146466e-01 4.43220176e-02
4.43158634e-02 -6.20342374e-01 7.08594263e-01 -3.72635961e-01
-8.52028847e-01 -4.53310460e-02 -7.39426970e-01 1.46586746e-01
1.06217754e+00 -4.99556035e-01 3.35100815e-02 -5.52130044e-01
-6.56938672e-01 1.63300157e-01 3.18697065e-01 1.90171644e-01
4.14083488e-02 -1.00873840e+00 -7.11785495e-01 -1.84986182e-02
-2.61617213e-01 -1.26760334e-01 4.71511573e-01 5.59108198e-01
-1.30512500e+00 2.31884599e-01 -7.08732486e-01 -2.23352984e-02
-1.20503390e+00 2.09225908e-01 5.78021228e-01 -9.55707312e-01
-5.32118618e-01 1.01073420e+00 -3.52039546e-01 -5.83846509e-01
3.17095593e-03 -1.82310045e-01 1.11207953e-02 -2.80621916e-01
4.63453859e-01 6.84010506e-01 3.77071440e-01 -1.90645725e-01
-2.11669117e-01 1.61521398e-02 -2.61020005e-01 -3.16882253e-01
1.39772904e+00 8.64133954e-01 3.13602030e-01 9.29305106e-02
4.58669066e-01 3.57430093e-02 -1.24867165e+00 2.21749738e-01
-1.18065640e-01 -2.13237122e-01 1.49231985e-01 -1.19955194e+00
-1.17706561e+00 8.30610156e-01 5.44582188e-01 -2.03505307e-02
8.79024923e-01 -5.24389863e-01 7.51613498e-01 6.82120681e-01
7.41441011e-01 -1.27496600e+00 9.82811674e-02 9.64829445e-01
4.35004443e-01 -9.68114555e-01 4.65417430e-02 3.03454101e-01
-6.93123639e-01 1.15253663e+00 9.49443817e-01 -5.79249203e-01
1.08455911e-01 5.56424975e-01 -1.95881665e-01 -1.78312883e-01
-8.46072018e-01 -1.50447339e-01 -1.70324132e-01 4.22647923e-01
1.13646403e-01 1.36089563e-01 -3.44924659e-01 1.26193082e+00
-1.19484067e+00 3.31053376e-01 7.71529377e-01 8.58939171e-01
-3.78719836e-01 -9.92613614e-01 -4.05973375e-01 4.66725528e-01
-6.01738393e-01 -4.21523482e-01 -3.07455733e-02 1.30513990e+00
4.34038520e-01 5.79420269e-01 8.58186632e-02 -7.75005221e-01
4.43066835e-01 -2.72780269e-01 6.71180964e-01 -3.23007524e-01
-1.32109118e+00 -4.08118665e-01 2.27306172e-01 -4.80453104e-01
2.07912207e-01 -4.54916805e-01 -1.18490136e+00 -1.18921101e+00
-3.01115662e-01 3.09229165e-01 5.30013919e-01 8.61163914e-01
2.61887163e-02 7.61313915e-01 -6.75805509e-02 -9.03655469e-01
-5.65441728e-01 -9.33906794e-01 -9.44612741e-01 7.72789717e-02
-4.76115078e-01 -7.91471958e-01 -6.44277632e-02 -4.46425557e-01] | [3.4454593658447266, 1.4243119955062866] |
67e4f936-4fe8-4044-94a7-ee6910107bf9 | context-aware-self-supervised-learning-of | 2306.04763 | null | https://arxiv.org/abs/2306.04763v1 | https://arxiv.org/pdf/2306.04763v1.pdf | Context-Aware Self-Supervised Learning of Whole Slide Images | Presenting whole slide images (WSIs) as graph will enable a more efficient and accurate learning framework for cancer diagnosis. Due to the fact that a single WSI consists of billions of pixels and there is a lack of vast annotated datasets required for computational pathology, the problem of learning from WSIs using typical deep learning approaches such as convolutional neural network (CNN) is challenging. Additionally, WSIs down-sampling may lead to the loss of data that is essential for cancer detection. A novel two-stage learning technique is presented in this work. Since context, such as topological features in the tumor surroundings, may hold important information for cancer grading and diagnosis, a graph representation capturing all dependencies among regions in the WSI is very intuitive. Graph convolutional network (GCN) is deployed to include context from the tumor and adjacent tissues, and self-supervised learning is used to enhance training through unlabeled data. More specifically, the entire slide is presented as a graph, where the nodes correspond to the patches from the WSI. The proposed framework is then tested using WSIs from prostate and kidney cancers. To assess the performance improvement through self-supervised mechanism, the proposed context-aware model is tested with and without use of pre-trained self-supervised layer. The overall model is also compared with multi-instance learning (MIL) based and other existing approaches. | ['Nasim Yahyasoltani', 'Milan Aryal'] | 2023-06-07 | null | null | null | null | ['whole-slide-images'] | ['computer-vision'] | [ 5.59804142e-01 5.49320042e-01 -2.48598814e-01 -2.44923711e-01
-5.68784595e-01 -2.48235032e-01 3.27123761e-01 7.89153397e-01
-2.34500736e-01 7.38415897e-01 -5.17299213e-02 -3.46782804e-01
-1.93044350e-01 -1.08639300e+00 -6.92132056e-01 -9.48351741e-01
-2.21497536e-01 2.36746311e-01 4.32742000e-01 -2.17171639e-01
2.08371222e-01 7.35429466e-01 -1.09211123e+00 3.46869588e-01
9.51318085e-01 1.01727140e+00 4.52952296e-01 6.59998715e-01
-5.96922636e-01 7.36575186e-01 -5.47524452e-01 -1.21303402e-01
-5.54423183e-02 -2.35747561e-01 -7.64739692e-01 6.51383325e-02
5.98835349e-01 9.23280939e-02 -2.28166655e-01 9.87742901e-01
3.23714077e-01 -7.32647106e-02 5.32079101e-01 -9.44852412e-01
-1.24351040e-01 3.89490545e-01 -6.46649420e-01 3.25666428e-01
-1.28241688e-01 1.89289749e-01 9.47700500e-01 -4.36508119e-01
9.36350882e-01 6.85929418e-01 6.23795807e-01 3.32338840e-01
-7.86486030e-01 -4.69316423e-01 1.09237634e-01 8.00343901e-02
-1.24796903e+00 2.13237032e-02 1.07089460e+00 -1.58594236e-01
5.83095312e-01 3.21629465e-01 8.20058286e-01 6.08485043e-01
5.32083333e-01 8.57434154e-01 1.12157381e+00 -4.66750413e-01
1.29174709e-01 1.68380871e-01 1.51308998e-01 9.69862700e-01
3.82971138e-01 -2.77167737e-01 -2.08141327e-01 2.04629585e-01
7.06489801e-01 5.32325745e-01 -2.08909392e-01 -2.49533176e-01
-9.79424894e-01 5.02382696e-01 1.19661522e+00 5.06790340e-01
-5.57733066e-02 2.35465970e-02 5.42376280e-01 -2.62340717e-02
4.59408879e-01 1.70350522e-01 -2.07484484e-01 3.35599363e-01
-1.00492752e+00 -1.65666491e-01 5.34758210e-01 5.28425932e-01
8.97044480e-01 -1.60622030e-01 -1.41526967e-01 5.14396369e-01
2.67295361e-01 9.49924290e-02 6.22645199e-01 -1.60378385e-02
3.68740380e-01 1.49055040e+00 -4.21988040e-01 -1.11835170e+00
-7.32648551e-01 -8.50102425e-01 -1.26504910e+00 1.47980303e-01
4.60033119e-01 1.16355024e-01 -1.19724035e+00 1.19286478e+00
5.41793525e-01 5.16136825e-01 5.94895594e-02 6.81340396e-01
1.17774403e+00 3.69691521e-01 1.63400441e-01 -8.34019259e-02
1.18314612e+00 -8.55850279e-01 -4.65609342e-01 -2.34196723e-01
8.90024960e-01 -4.72334534e-01 8.58614385e-01 -1.46921858e-01
-7.21904397e-01 -4.52466905e-01 -1.05196059e+00 2.39094123e-02
-6.58621311e-01 2.63593465e-01 8.01053643e-01 2.15876788e-01
-1.01865637e+00 4.49836284e-01 -8.50350380e-01 -6.20540142e-01
8.84937108e-01 5.10445058e-01 -5.77314675e-01 -2.06387252e-01
-7.95129299e-01 5.94496071e-01 4.51437384e-01 2.98303604e-01
-9.11706924e-01 -7.37377226e-01 -8.63060832e-01 9.57546830e-02
2.33616695e-01 -2.27174386e-01 5.10418057e-01 -1.07607055e+00
-1.04797125e+00 8.87966812e-01 3.76577750e-02 -3.24844182e-01
5.06116033e-01 4.48543340e-01 -1.86434492e-01 4.22069103e-01
-1.66515559e-01 4.67093319e-01 5.67638218e-01 -1.19091892e+00
-5.13843417e-01 -6.82327151e-01 1.25383034e-01 3.60414833e-01
-6.05926991e-01 -6.08949363e-01 -5.72357953e-01 -4.26280051e-01
2.04015464e-01 -9.30964887e-01 -3.61035347e-01 2.08347425e-01
-7.69699693e-01 1.40598547e-02 1.06361902e+00 -7.24175096e-01
1.17330980e+00 -2.13043499e+00 -1.81307554e-01 4.94784176e-01
3.05065453e-01 4.29272324e-01 -1.22086532e-01 2.14433402e-01
-5.42486571e-02 -3.25926056e-04 -2.99713492e-01 -1.61172487e-02
-5.23522556e-01 3.84372413e-01 5.44792652e-01 4.63036746e-01
4.95094985e-01 1.00544894e+00 -8.47302198e-01 -9.96189892e-01
3.01268816e-01 3.73636812e-01 -2.37725958e-01 1.72025785e-01
-1.62074551e-01 5.79013467e-01 -4.82560217e-01 9.56645250e-01
7.34439254e-01 -5.18521488e-01 3.56345922e-01 -4.45477664e-01
2.42676616e-01 -1.21105291e-01 -9.64501023e-01 1.58525324e+00
-3.62062514e-01 4.78494495e-01 5.57500981e-02 -1.40820467e+00
8.69436085e-01 1.27330959e-01 5.46219110e-01 -5.63027978e-01
1.75958183e-02 -1.17452722e-02 2.05833629e-01 -5.80653369e-01
2.61092894e-02 8.00327733e-02 3.85590762e-01 9.91674140e-02
-1.91571303e-02 1.29916847e-01 2.71539599e-01 3.35292399e-01
1.39696097e+00 -1.42278150e-01 4.54194337e-01 -2.38171145e-01
7.15393126e-01 2.67517716e-01 6.01779819e-01 3.30507308e-01
-1.55136004e-01 2.92976081e-01 6.10460997e-01 -5.17496705e-01
-7.46742308e-01 -8.84525418e-01 -1.53391436e-01 6.37476444e-01
2.63406307e-01 -1.26853390e-02 -4.10193920e-01 -9.58575010e-01
4.05795462e-02 -5.33744842e-02 -9.10501003e-01 -5.09191118e-02
-5.94226837e-01 -8.71731937e-01 2.68695027e-01 6.20450377e-01
6.65974379e-01 -9.57080901e-01 -4.61611688e-01 2.28059664e-01
1.74689546e-01 -8.21728170e-01 -2.28462704e-02 3.49617064e-01
-1.21921468e+00 -1.41176307e+00 -6.45030499e-01 -1.08971786e+00
1.41159093e+00 2.50153720e-01 9.25697386e-01 6.62661850e-01
-8.44259143e-01 -1.91303510e-02 -2.56290913e-01 -4.10952359e-01
-3.89040947e-01 9.30151492e-02 -6.97703481e-01 9.42246690e-02
1.87463567e-01 -3.59177440e-01 -9.66330051e-01 6.11528615e-03
-1.01876283e+00 1.50496036e-01 1.15021598e+00 1.19991314e+00
9.58756089e-01 2.82438070e-01 3.98482174e-01 -1.36622393e+00
2.29280174e-01 -5.09501994e-01 -3.74839365e-01 5.10136187e-01
-3.33439112e-01 -9.47800651e-02 7.05689907e-01 -2.62254328e-01
-9.85342503e-01 1.98036641e-01 -8.00989289e-03 2.25522323e-03
-1.50649175e-01 8.73894215e-01 -9.40889046e-02 -5.20106852e-01
5.64601421e-01 9.79495943e-02 2.76793927e-01 6.21249229e-02
-9.41573605e-02 4.73574340e-01 1.68890804e-01 -4.66588065e-02
6.01579964e-01 5.28070271e-01 5.07711887e-01 -7.80567110e-01
-7.82311857e-01 -6.13168716e-01 -6.13542020e-01 -2.80961096e-01
6.52078807e-01 -5.65190494e-01 -6.02992117e-01 3.79257083e-01
-5.56054354e-01 -4.73367214e-01 -1.03634156e-01 3.44360173e-01
-2.76776627e-02 1.55728459e-01 -7.60090470e-01 -5.41036010e-01
-4.75119621e-01 -1.06122017e+00 1.09095001e+00 5.77043414e-01
2.30710208e-01 -1.39279926e+00 -8.49216282e-02 3.43068957e-01
3.52311760e-01 7.61483014e-01 1.06240571e+00 -8.75238538e-01
-7.18125224e-01 -6.95478618e-01 -3.99843365e-01 1.94538161e-01
4.60957885e-01 2.01951161e-01 -8.37735832e-01 -3.59936923e-01
-4.77646440e-01 -3.38079393e-01 8.48507345e-01 4.73890096e-01
1.24019980e+00 -3.58239353e-01 -9.28877831e-01 6.10807538e-01
2.00627875e+00 1.54207284e-02 5.72635889e-01 2.24037334e-01
8.27485442e-01 5.07694006e-01 4.53568280e-01 6.38163388e-02
2.64783710e-01 4.19142321e-02 7.21996605e-01 -7.03376174e-01
-1.89823747e-01 -5.55275474e-03 -3.00645232e-01 5.92669904e-01
-1.39302490e-02 -2.16043413e-01 -1.06102216e+00 5.20400167e-01
-1.51900518e+00 -7.09119320e-01 -1.27773676e-02 1.98841894e+00
6.70876145e-01 2.05004096e-01 -3.55893523e-01 3.67512226e-01
6.38165832e-01 3.25492807e-02 -7.30006337e-01 -5.89205548e-02
1.21344678e-01 4.47523892e-01 4.49708849e-01 1.59232616e-01
-1.09240329e+00 6.16227090e-01 5.31024647e+00 7.31434464e-01
-1.56883943e+00 -1.80892989e-01 1.19101608e+00 2.23989576e-01
-4.82549705e-02 -2.76548326e-01 -5.86862862e-01 2.42756322e-01
4.83795285e-01 1.21043220e-01 -1.34460837e-01 7.43910134e-01
6.62400797e-02 -2.92983413e-01 -8.99276674e-01 6.49712980e-01
4.42821942e-02 -1.62415910e+00 7.87147284e-02 2.02255160e-01
7.92055011e-01 -4.87767048e-02 -1.73973337e-01 -3.83180310e-03
7.95504376e-02 -9.75545049e-01 -2.23065063e-01 5.99916458e-01
9.43909585e-01 -6.98953032e-01 1.23579586e+00 3.15124273e-01
-1.40667951e+00 -1.67685032e-01 -3.81224006e-01 3.83967280e-01
-5.41343987e-01 6.76667750e-01 -1.39560640e+00 8.18510056e-01
3.76058489e-01 8.04705501e-01 -9.46170449e-01 1.11057472e+00
1.19935170e-01 7.05686271e-01 -2.76088327e-01 -2.45169982e-01
3.18250984e-01 9.45229456e-02 7.06998631e-02 1.32537436e+00
3.05825114e-01 -1.29687404e-02 3.08319658e-01 4.61100429e-01
-9.39383209e-02 3.08074832e-01 -5.82432687e-01 -1.13846049e-01
2.01305583e-01 1.82931483e+00 -1.23922431e+00 -2.74066716e-01
-4.74421084e-01 7.16217577e-01 5.75667262e-01 2.79141784e-01
-4.11761582e-01 -3.91391784e-01 4.35831323e-02 4.45243865e-01
4.58611362e-02 2.08350465e-01 -1.23094507e-01 -7.38988519e-01
-1.08553372e-01 -4.17349517e-01 5.65932810e-01 -3.36272240e-01
-1.19042361e+00 4.98528063e-01 -3.81941080e-01 -1.32853341e+00
2.82194078e-01 -6.32380545e-01 -1.24468815e+00 6.11977160e-01
-1.98009408e+00 -1.52437878e+00 -9.36245620e-01 5.84207475e-01
2.28541687e-01 -5.50806634e-02 8.02328825e-01 7.24604204e-02
-6.59360945e-01 7.60330141e-01 -8.16011429e-02 4.68782246e-01
4.77371335e-01 -1.46017337e+00 -2.46212080e-01 5.17978549e-01
-2.24091396e-01 5.68200231e-01 2.43034616e-01 -7.82087803e-01
-1.47209263e+00 -1.48847580e+00 3.80747974e-01 1.23243213e-01
5.70297480e-01 -1.29204437e-01 -9.83712792e-01 4.19208318e-01
-5.11566773e-02 7.38569438e-01 9.19570327e-01 -1.30976170e-01
2.08001196e-01 -4.51624066e-01 -1.29282093e+00 4.27496046e-01
6.84498549e-01 -2.04681307e-01 1.95576120e-02 4.98165876e-01
3.72370064e-01 -5.93190134e-01 -9.44713891e-01 5.15284538e-01
2.29734689e-01 -6.96394503e-01 7.21347451e-01 -3.77943784e-01
4.54947501e-01 -3.07472080e-01 2.82707959e-01 -1.38806427e+00
-1.28635898e-01 3.16215679e-02 1.35920435e-01 1.04126382e+00
4.67399716e-01 -4.89609390e-01 1.34924281e+00 1.56859294e-01
-4.59270924e-01 -1.13389719e+00 -8.30711365e-01 -3.50870669e-01
-9.68704000e-02 2.23101303e-01 4.10354406e-01 9.18223202e-01
-4.50604334e-02 1.38978362e-01 2.93178052e-01 2.57323146e-01
5.93147337e-01 1.17154978e-02 5.70329130e-01 -1.29905343e+00
-1.12043723e-01 -3.64818633e-01 -1.05349457e+00 -3.39380354e-01
-2.16809705e-01 -1.24684680e+00 -1.23714939e-01 -1.93039918e+00
2.63278306e-01 -7.48382568e-01 -6.55441046e-01 5.51069677e-01
-4.39212561e-01 2.79164255e-01 -2.18199179e-01 4.96676750e-02
-6.03254735e-01 2.25123554e-01 1.76540768e+00 -6.03390992e-01
-1.75345875e-02 -8.01459923e-02 -4.78093117e-01 5.68032861e-01
7.72618890e-01 -3.28299701e-02 -5.19745946e-01 -4.82323319e-02
2.48382408e-02 1.55567512e-01 3.54382664e-01 -1.20594454e+00
5.06843090e-01 -9.81683731e-02 8.08044910e-01 -7.06566513e-01
7.63181448e-02 -9.13423181e-01 -4.32953052e-02 6.02122307e-01
-2.45847985e-01 -3.86006773e-01 2.02089503e-01 6.84493005e-01
-4.55793113e-01 -1.51898414e-01 6.41808450e-01 -4.92119431e-01
-5.30329704e-01 6.05152965e-01 1.52231097e-01 -3.07074100e-01
1.28640079e+00 -5.57892978e-01 -3.88889492e-01 1.95063025e-01
-8.26000631e-01 2.68708348e-01 2.47121871e-01 -2.24690497e-01
7.07237184e-01 -1.04140377e+00 -4.96038854e-01 8.41824487e-02
5.27055025e-01 6.08946860e-01 6.12794161e-01 8.27772975e-01
-9.10495102e-01 1.59373298e-01 -2.29128301e-01 -8.31594527e-01
-1.27936769e+00 4.39210117e-01 4.01256353e-01 -8.35757852e-01
-6.46141231e-01 9.85582709e-01 2.52768815e-01 -3.88342649e-01
3.46572548e-01 -4.72240180e-01 -4.49556589e-01 -2.47746870e-01
4.03831601e-01 2.94688437e-02 1.92759499e-01 -3.16260815e-01
-2.35379905e-01 2.80612499e-01 -3.36326212e-01 6.54580474e-01
1.32632196e+00 1.24929644e-01 -2.52781630e-01 2.14157894e-01
1.16050506e+00 -1.46897405e-01 -1.26240063e+00 -3.09360385e-01
7.37108812e-02 -9.82206017e-02 1.23074293e-01 -8.00730944e-01
-1.36909175e+00 9.14068520e-01 8.20967317e-01 -4.01803898e-03
1.17925406e+00 -1.08399257e-01 6.99855924e-01 3.02021205e-01
3.82807739e-02 -1.01010764e+00 3.29573661e-01 -6.44443110e-02
4.91296560e-01 -1.53286242e+00 9.55569670e-02 -5.00110209e-01
-3.53312939e-01 1.42126501e+00 8.82819891e-01 -2.89412051e-01
7.55539417e-01 3.45104575e-01 1.11932218e-01 -5.87097764e-01
-5.63969672e-01 -3.29404920e-02 2.04450756e-01 5.32796383e-01
4.54086065e-01 1.75773010e-01 -4.13791537e-02 2.97996610e-01
1.19217001e-01 1.00102119e-01 2.83522278e-01 1.09884071e+00
-4.69130307e-01 -7.63303757e-01 -1.01146832e-01 1.00243998e+00
-4.40159321e-01 5.36348671e-02 -3.09455574e-01 1.01485610e+00
2.17633605e-01 4.70117211e-01 1.23379327e-01 -9.95549187e-02
-2.63055470e-02 -2.48372898e-01 2.95787871e-01 -7.63639808e-01
-6.48710608e-01 -1.11379484e-02 -2.82008529e-01 -3.11764896e-01
-4.81058538e-01 -2.18318790e-01 -1.52740145e+00 9.27362293e-02
-4.75672007e-01 -2.50987653e-02 6.93503261e-01 9.47452068e-01
-2.06911340e-02 9.87058282e-01 7.28547037e-01 -4.34010625e-01
5.44876046e-02 -1.00313795e+00 -5.09836078e-01 4.63734031e-01
3.26926380e-01 -4.84883726e-01 -1.97147280e-01 -1.23583097e-02] | [15.05008316040039, -2.912081480026245] |
3425879e-668e-4661-a3b6-ca05ea559c17 | a-lightweight-machine-learning-pipeline-for | 2208.03130 | null | https://arxiv.org/abs/2208.03130v1 | https://arxiv.org/pdf/2208.03130v1.pdf | A Lightweight Machine Learning Pipeline for LiDAR-simulation | Virtual testing is a crucial task to ensure safety in autonomous driving, and sensor simulation is an important task in this domain. Most current LiDAR simulations are very simplistic and are mainly used to perform initial tests, while the majority of insights are gathered on the road. In this paper, we propose a lightweight approach for more realistic LiDAR simulation that learns a real sensor's behavior from test drive data and transforms this to the virtual domain. The central idea is to cast the simulation into an image-to-image translation problem. We train our pix2pix based architecture on two real world data sets, namely the popular KITTI data set and the Audi Autonomous Driving Dataset which provide both, RGB and LiDAR images. We apply this network on synthetic renderings and show that it generalizes sufficiently from real images to simulated images. This strategy enables to skip the sensor-specific, expensive and complex LiDAR physics simulation in our synthetic world and avoids oversimplification and a large domain-gap through the clean synthetic environment. | ['Marc Stamminger', 'Bernhard Egger', 'Niklas Knoop', 'Richard Marcus'] | 2022-08-05 | null | null | null | null | ['machine-learning', 'machine-learning'] | ['methodology', 'miscellaneous'] | [ 2.01514602e-01 7.21283183e-02 3.03717196e-01 -7.84444749e-01
-6.24043822e-01 -6.37222528e-01 6.51570559e-01 -3.24456424e-01
-5.64082265e-01 8.80358636e-01 -6.61347806e-01 -7.55681217e-01
1.45434558e-01 -1.17405355e+00 -1.13712347e+00 -4.10820633e-01
2.14619748e-02 8.29116642e-01 6.47533476e-01 -5.10368943e-01
-6.24081828e-02 7.74273813e-01 -2.23732948e+00 -2.48291388e-01
9.28118110e-01 7.94994235e-01 3.93038154e-01 7.79565275e-01
-5.19431308e-02 4.93728191e-01 -4.63847578e-01 -2.36858502e-01
5.46761811e-01 -1.10769384e-01 -2.05737561e-01 1.04381174e-01
4.17085350e-01 -2.98776984e-01 -2.32071415e-01 9.99520361e-01
5.40695012e-01 -1.02432832e-01 3.99975032e-01 -1.93268871e+00
2.89435744e-01 2.39603311e-01 -2.97625810e-01 -3.36660296e-01
-4.30915020e-02 7.06417620e-01 2.61991918e-01 -5.29846966e-01
5.92449784e-01 1.15679479e+00 6.03489876e-01 2.88554162e-01
-1.14277840e+00 -9.49838877e-01 -1.97033167e-01 2.26626843e-01
-1.49212718e+00 -2.29576990e-01 8.10303211e-01 -4.86267775e-01
4.67799366e-01 1.42953888e-01 7.75642872e-01 1.28819191e+00
1.57823116e-01 3.59446704e-01 1.37912798e+00 9.17466953e-02
4.90043044e-01 5.14370084e-01 -1.26135677e-01 5.46992660e-01
2.85574317e-01 6.71715260e-01 -2.38411739e-01 3.11068982e-01
5.28539956e-01 -3.34711373e-01 -1.63517952e-01 -7.50549793e-01
-1.01839936e+00 6.71309173e-01 3.51257414e-01 -5.41856766e-01
-1.47603258e-01 4.46780115e-01 3.99513334e-01 3.40021938e-01
-7.11277574e-02 4.94515710e-03 -3.05899620e-01 -3.18135560e-01
-7.81187594e-01 6.08740449e-01 8.20178390e-01 8.81620824e-01
1.15820658e+00 3.07625055e-01 4.60922658e-01 3.24838638e-01
1.14094004e-01 9.05601561e-01 1.49850205e-01 -1.29016507e+00
3.67815971e-01 3.79683971e-01 3.14750701e-01 -5.50983012e-01
-3.28416973e-01 -2.99777597e-01 -5.84752798e-01 1.02236605e+00
2.78738827e-01 -1.44579306e-01 -1.07679033e+00 1.57657385e+00
3.50168765e-01 5.36445737e-01 1.01806343e-01 8.89428496e-01
5.04034877e-01 5.17663717e-01 -1.15747407e-01 2.84472734e-01
1.09289563e+00 -4.80676383e-01 -3.13205540e-01 -5.43452799e-01
4.25297916e-01 -3.58446687e-01 1.42920101e+00 5.24512053e-01
-8.62814903e-01 -8.62248838e-01 -1.55874217e+00 1.27260372e-01
-7.85648286e-01 -1.28583893e-01 5.71398377e-01 5.87112129e-01
-1.07102573e+00 6.32024050e-01 -7.37403631e-01 -2.87791699e-01
3.07532877e-01 2.90443927e-01 -3.21122706e-01 -1.57925352e-01
-1.30518687e+00 1.06894267e+00 3.93988401e-01 2.26031616e-02
-1.27175307e+00 -7.52336144e-01 -9.34363782e-01 -2.99483776e-01
4.81144130e-01 -5.55440903e-01 1.23601067e+00 -3.02150935e-01
-1.44718349e+00 7.81463921e-01 4.73202765e-02 -7.16188014e-01
1.04990935e+00 7.26516396e-02 -3.54731888e-01 -2.89460778e-01
1.28690988e-01 9.84210789e-01 7.74318457e-01 -1.81367826e+00
-6.16440296e-01 -1.94540128e-01 1.17359906e-01 -1.14988998e-01
4.05362576e-01 -6.15398467e-01 -4.56766635e-01 1.97538231e-02
-8.76382142e-02 -1.07189143e+00 -3.80113989e-01 2.93586016e-01
-3.29805344e-01 3.03142905e-01 1.00645006e+00 -3.34888995e-02
3.94308090e-01 -2.06764269e+00 -3.23779017e-01 2.66966224e-01
-8.87240544e-02 1.53434888e-01 -4.49343957e-03 2.76465684e-01
-1.95906460e-01 -8.82392302e-02 -5.44759810e-01 -3.76417190e-01
1.60086572e-01 7.55020976e-01 -6.35632217e-01 3.06456953e-01
2.68805355e-01 9.96331155e-01 -8.63288105e-01 -6.44896865e-01
6.65381610e-01 6.12219810e-01 -2.35038012e-01 2.22639188e-01
-4.33194607e-01 7.39518642e-01 -2.49648303e-01 3.73380989e-01
1.15042055e+00 3.24907392e-01 -1.76594138e-01 -1.48017928e-01
-4.67418760e-01 1.12368390e-01 -1.24957490e+00 1.77424204e+00
-6.79547071e-01 6.79046392e-01 2.27843314e-01 -7.60510087e-01
1.03693902e+00 -2.43795633e-01 3.12954158e-01 -1.02756035e+00
3.39979649e-01 1.46111101e-01 -1.13985389e-01 -5.77371180e-01
6.11009121e-01 -2.97123700e-01 -2.58318931e-01 3.67324054e-01
-4.08258259e-01 -1.18701792e+00 1.15658782e-01 3.55653092e-02
9.01208758e-01 6.74685001e-01 -2.52095431e-01 -4.50521819e-02
3.58224660e-01 5.35142064e-01 4.90060300e-01 5.47615707e-01
-9.41842198e-02 6.80695474e-01 4.62435931e-01 -3.64350498e-01
-1.19248831e+00 -1.38464415e+00 -2.37714529e-01 2.85644531e-01
4.90917653e-01 -9.23661795e-03 -6.96223021e-01 -5.54741263e-01
3.69042635e-01 1.13895237e+00 -5.56418002e-01 -3.67610008e-01
-4.93831098e-01 -4.49484676e-01 6.75081730e-01 4.32194829e-01
7.91472316e-01 -8.25183451e-01 -1.16331458e+00 8.55382159e-02
9.61900130e-02 -1.38731754e+00 3.66432369e-01 3.43422413e-01
-5.48198581e-01 -9.42146242e-01 3.10451947e-02 -1.77323833e-01
2.45934248e-01 2.70274162e-01 1.25634885e+00 -1.06389374e-01
-3.31205249e-01 9.90655273e-02 -1.28752450e-02 -8.69678915e-01
-6.30278111e-01 -2.87488520e-01 9.34858471e-02 -2.42946073e-01
1.29230440e-01 -8.83140802e-01 -3.63298774e-01 5.70557415e-01
-9.13873672e-01 2.11968064e-01 5.28967202e-01 5.01220405e-01
6.54075742e-01 3.45163643e-01 1.04576133e-01 -7.56601870e-01
2.02410698e-01 -2.77926743e-01 -1.15165949e+00 -2.70224631e-01
-4.94058907e-01 7.00256377e-02 7.84303129e-01 -2.43345395e-01
-8.43476832e-01 3.68132263e-01 -2.09184036e-01 -4.80727106e-01
-4.55484986e-01 1.90418407e-01 -4.48377699e-01 -2.56054759e-01
6.81779206e-01 1.09327160e-01 1.05223484e-01 -2.13064104e-01
2.57501841e-01 4.70018506e-01 1.03646374e+00 -6.22858346e-01
1.39723134e+00 8.65507603e-01 3.75222951e-01 -8.33895922e-01
-3.11503828e-01 6.48367330e-02 -6.02059066e-01 -4.06781733e-01
5.97752035e-01 -9.37072754e-01 -9.76405978e-01 3.59469324e-01
-9.85200167e-01 -8.07322264e-01 -7.02856183e-01 3.94473284e-01
-8.18090856e-01 1.78574547e-01 1.62399083e-01 -7.47357070e-01
4.04649824e-01 -1.53882766e+00 1.34304404e+00 6.12935834e-02
2.04066873e-01 -5.86784959e-01 1.49147123e-01 -4.36180979e-02
3.33072543e-01 6.78947985e-01 6.54231906e-01 1.23379737e-01
-8.94619823e-01 -2.23186225e-01 -4.29883748e-01 4.33576852e-01
-3.43819380e-01 2.63745606e-01 -1.36055207e+00 2.79683415e-02
2.24300809e-02 -4.62192416e-01 7.10389376e-01 3.97208519e-03
1.33607757e+00 5.36179483e-01 -3.11590582e-01 7.76556432e-01
1.55266953e+00 6.51199445e-02 8.69514048e-01 2.25711539e-01
6.21465981e-01 7.34413803e-01 1.03475010e+00 1.91009879e-01
7.45462179e-01 6.71161115e-01 1.07597649e+00 -4.55110550e-01
-8.79872590e-02 -3.90949488e-01 3.89457971e-01 3.43609065e-01
1.92607939e-01 -2.45361760e-01 -1.15604031e+00 4.50102150e-01
-1.77363253e+00 -5.84176898e-01 -3.99727583e-01 2.40603948e+00
3.88035476e-01 5.91153741e-01 -1.08816609e-01 3.02865982e-01
3.59770685e-01 -7.66429827e-02 -7.46834278e-01 -4.35914665e-01
-1.05586894e-01 4.15833384e-01 8.51280630e-01 5.99058568e-01
-7.94800103e-01 1.05625367e+00 5.87067366e+00 5.73006094e-01
-1.34023392e+00 1.03117414e-01 1.81750163e-01 1.29798815e-01
-2.41745666e-01 1.97295263e-01 -5.43294311e-01 4.11796659e-01
1.38596225e+00 -8.31190199e-02 2.56897300e-01 9.52575862e-01
4.86952305e-01 -5.39386928e-01 -1.19776213e+00 9.69981849e-01
-3.32772613e-01 -1.15655243e+00 -1.33514792e-01 2.16785043e-01
2.64253676e-01 5.55635095e-01 9.57249105e-02 4.56171185e-01
7.97159731e-01 -1.22149825e+00 8.86295557e-01 4.23839957e-01
1.11935914e+00 -7.95910180e-01 5.85730255e-01 6.11070991e-01
-1.03005743e+00 1.92947581e-01 -3.97646368e-01 -2.16903444e-02
4.10674572e-01 4.80045944e-01 -1.04775417e+00 5.60831070e-01
7.90587008e-01 4.84766543e-01 -6.44300997e-01 8.65501225e-01
-3.08850467e-01 4.23786759e-01 -5.02306998e-01 2.42704913e-01
2.26251692e-01 -4.61457193e-01 3.35169911e-01 8.86054873e-01
3.54574353e-01 -3.82203639e-01 1.33130476e-01 1.04503942e+00
1.73934817e-01 -5.34685314e-01 -1.19101882e+00 4.00643796e-01
3.21209908e-01 1.25643396e+00 -5.41063130e-01 -3.08000714e-01
-1.31438076e-01 7.43948936e-01 -6.50903359e-02 3.53365511e-01
-1.37931967e+00 -3.15221637e-01 1.03871560e+00 3.38497132e-01
1.85537666e-01 -4.30997938e-01 -3.97100300e-01 -8.81858289e-01
2.60642618e-02 -6.46025240e-01 -3.91621858e-01 -1.17948306e+00
-6.21786833e-01 6.21584058e-01 2.63681650e-01 -1.61749053e+00
-4.58896428e-01 -7.31649756e-01 -5.00425160e-01 9.53568637e-01
-1.98622346e+00 -1.12335157e+00 -1.00119781e+00 5.43999672e-01
3.09731871e-01 2.68011093e-01 5.48170209e-01 4.13434625e-01
-2.90812552e-01 2.00426713e-01 -2.00942725e-01 -1.99131817e-01
6.49220884e-01 -1.19362867e+00 8.35736632e-01 7.83416688e-01
-2.49420986e-01 1.10163532e-01 1.08188891e+00 -5.12278199e-01
-1.57453477e+00 -1.36192858e+00 2.99565673e-01 -6.30930781e-01
7.30761170e-01 -6.41165972e-01 -8.48449349e-01 4.94461179e-01
4.54702415e-03 1.24123544e-01 -9.85069051e-02 -7.17062354e-01
-1.05530068e-01 -4.71920758e-01 -1.23690736e+00 7.22719848e-01
1.21865964e+00 -4.50055122e-01 -9.80085582e-02 2.87201375e-01
6.98031068e-01 -5.18797994e-01 -5.56914210e-01 6.23490691e-01
5.36486030e-01 -1.29420602e+00 1.04239249e+00 -1.91076502e-01
2.15525880e-01 -8.28660190e-01 -2.29481444e-01 -1.37341583e+00
3.93397361e-01 -5.05755067e-01 4.00602937e-01 9.78842795e-01
2.98862934e-01 -8.49393189e-01 9.30915713e-01 3.18903059e-01
-3.48238111e-01 -3.14715803e-01 -1.02195704e+00 -8.94992471e-01
2.07463857e-02 -1.27333975e+00 9.35740650e-01 5.81898570e-01
-8.00826132e-01 1.18690923e-01 1.03400846e-03 3.27511102e-01
8.63427520e-01 2.64017913e-03 1.45850027e+00 -1.41126692e+00
-1.41126901e-01 -9.84825939e-02 -8.04094195e-01 -8.20754468e-01
2.96863973e-01 -6.82678282e-01 3.78017694e-01 -1.17512167e+00
-3.80637020e-01 -7.51785338e-01 1.89096048e-01 9.68830511e-02
3.62051576e-01 4.14442539e-01 -1.96885735e-01 -2.64321536e-01
-5.10260537e-02 5.08439422e-01 1.18790257e+00 -1.48173660e-01
5.60844131e-02 9.33823436e-02 -3.23626786e-01 7.00133801e-01
7.74980128e-01 -3.65919054e-01 -8.74084651e-01 -5.06931901e-01
3.23165923e-01 -4.80439328e-02 9.35521305e-01 -1.36038971e+00
1.75102297e-02 -2.92739898e-01 3.28407586e-02 -9.53062594e-01
7.38990009e-01 -1.19658029e+00 4.07143593e-01 4.09415632e-01
3.01807791e-01 1.51796743e-01 4.53022927e-01 3.12775403e-01
-1.73830301e-01 1.12699546e-01 8.19134295e-01 -1.19052194e-01
-1.03410029e+00 3.71102929e-01 2.28624279e-03 -6.96850270e-02
1.25376999e+00 -5.26240170e-01 -3.32079440e-01 -5.11565268e-01
-4.33854640e-01 5.03533304e-01 9.86617267e-01 3.55051726e-01
6.47309721e-01 -1.13444340e+00 -5.76518774e-01 5.85793257e-01
3.46947581e-01 5.48088849e-01 4.52141315e-02 5.30899942e-01
-8.52526009e-01 1.08126007e-01 -3.24136257e-01 -1.07717133e+00
-9.15180326e-01 5.41148365e-01 3.99984062e-01 1.76561326e-01
-7.12998033e-01 5.84806740e-01 1.86521932e-01 -8.49425495e-01
1.46562504e-02 -6.48489833e-01 1.70653746e-01 -2.54410416e-01
1.81284443e-01 1.79198399e-01 3.47082287e-01 -7.55166769e-01
-2.19153076e-01 7.80461073e-01 5.24292588e-01 -4.81784344e-01
1.20257699e+00 -3.60401832e-02 2.22914413e-01 5.98857701e-01
1.05288243e+00 -2.39972174e-01 -1.65272450e+00 2.34008357e-01
-2.40602195e-01 -4.98719215e-01 1.12489998e-01 -6.32762432e-01
-9.36798871e-01 1.33509231e+00 8.16957653e-01 -6.85553951e-03
8.87626410e-01 -1.73351571e-01 6.72794759e-01 4.87865120e-01
7.97911823e-01 -1.05855012e+00 -4.08466011e-01 4.20787871e-01
7.75407314e-01 -1.30820870e+00 -1.06529213e-01 -4.72665608e-01
-6.37073755e-01 7.35825717e-01 7.01267362e-01 -1.57029688e-01
6.24546885e-01 8.26673925e-01 3.56635034e-01 -2.59265065e-01
-5.86562932e-01 -4.47176099e-01 -4.61945027e-01 8.23597729e-01
-3.10837567e-01 9.83752981e-02 4.09440070e-01 1.07085891e-02
-7.33159721e-01 2.40534246e-01 7.76075065e-01 1.02914727e+00
-3.40987623e-01 -1.17859268e+00 -3.84502888e-01 8.80845562e-02
2.70896077e-01 3.07067394e-01 -2.58591026e-01 1.32930923e+00
4.68144149e-01 6.88948512e-01 2.59138972e-01 -7.05462277e-01
6.83898866e-01 -2.97296923e-02 4.47208613e-01 -4.39948797e-01
-1.92837834e-01 -4.18053001e-01 1.37092054e-01 -9.14515316e-01
-7.55326748e-02 -5.62878728e-01 -1.41290414e+00 -4.76423591e-01
1.08442351e-01 6.04871707e-03 1.33722520e+00 7.28981674e-01
2.98635900e-01 8.34589183e-01 6.95187449e-01 -1.15162444e+00
-4.07291740e-01 -6.00456417e-01 -4.98057276e-01 2.95880675e-01
4.38702017e-01 -9.77367401e-01 -2.38477960e-01 -1.36323884e-01] | [8.078468322753906, -2.474738359451294] |
7caa0702-0bd0-45e8-a179-473a176f9894 | orthonormal-product-quantization-network-for | 2107.00327 | null | https://arxiv.org/abs/2107.00327v4 | https://arxiv.org/pdf/2107.00327v4.pdf | Orthonormal Product Quantization Network for Scalable Face Image Retrieval | Existing deep quantization methods provided an efficient solution for large-scale image retrieval. However, the significant intra-class variations like pose, illumination, and expressions in face images, still pose a challenge for face image retrieval. In light of this, face image retrieval requires sufficiently powerful learning metrics, which are absent in current deep quantization works. Moreover, to tackle the growing unseen identities in the query stage, face image retrieval drives more demands regarding model generalization and system scalability than general image retrieval tasks. This paper integrates product quantization with orthonormal constraints into an end-to-end deep learning framework to effectively retrieve face images. Specifically, a novel scheme that uses predefined orthonormal vectors as codewords is proposed to enhance the quantization informativeness and reduce codewords' redundancy. A tailored loss function maximizes discriminability among identities in each quantization subspace for both the quantized and original features. An entropy-based regularization term is imposed to reduce the quantization error. Experiments are conducted on four commonly-used face datasets under both seen and unseen identities retrieval settings. Our method outperforms all the compared deep hashing/quantization state-of-the-arts under both settings. Results validate the effectiveness of the proposed orthonormal codewords in improving models' standard retrieval performance and generalization ability. Combing with further experiments on two general image datasets, it demonstrates the broad superiority of our method for scalable image retrieval. | ['Hong Yan', 'Xuefei Zhe', 'Ming Zhang'] | 2021-07-01 | null | null | null | null | ['face-image-retrieval'] | ['computer-vision'] | [-5.51557578e-02 -5.71334779e-01 -2.58389920e-01 -6.49174750e-01
-1.05025160e+00 -4.12076741e-01 4.95125353e-01 -8.32192153e-02
-3.02827448e-01 4.39401925e-01 3.66406664e-02 3.30503434e-01
-3.91115725e-01 -6.03435457e-01 -4.32589740e-01 -9.34102058e-01
2.06121653e-02 2.85230875e-01 -4.70285952e-01 -8.30376148e-02
2.78786480e-01 6.71739042e-01 -1.93228137e+00 -3.20239626e-02
3.67850780e-01 1.52846920e+00 -3.01381201e-02 2.05724999e-01
1.26076505e-01 1.45590454e-01 -4.98868495e-01 -6.43888116e-01
5.14402449e-01 -8.64492580e-02 -3.82684290e-01 -6.71082968e-03
9.56050217e-01 -8.42354000e-01 -8.15444887e-01 1.02412653e+00
1.05432642e+00 -3.40161123e-03 7.92428970e-01 -1.50403833e+00
-1.29429638e+00 -7.25529045e-02 -4.94841069e-01 2.26013303e-01
2.82811940e-01 -3.61774303e-02 1.09977841e+00 -1.23426485e+00
4.63967949e-01 1.55552077e+00 4.25929695e-01 6.63832664e-01
-9.00098324e-01 -1.03637421e+00 -2.72469372e-01 5.02785861e-01
-2.12739944e+00 -8.21380734e-01 6.80869520e-01 -6.18665293e-02
7.49061108e-01 1.45157591e-01 2.47923657e-01 8.57099116e-01
9.63156149e-02 6.77680850e-01 5.77681661e-01 -1.60000786e-01
1.50499195e-01 1.47231966e-01 -1.57488987e-01 6.55117214e-01
1.51898727e-01 -1.78838875e-02 -7.00063586e-01 -3.29574168e-01
5.15738189e-01 2.54429758e-01 -2.39784434e-01 -2.07206741e-01
-7.89383471e-01 1.00528300e+00 5.13528824e-01 1.31069139e-01
-3.58319968e-01 3.84439170e-01 4.59810913e-01 2.79634684e-01
2.58062959e-01 -9.85870734e-02 -1.00336142e-01 1.54397488e-01
-1.09467828e+00 3.66331667e-01 4.54360455e-01 9.52825963e-01
8.96352589e-01 -1.70152694e-01 -4.76669133e-01 1.07656050e+00
5.56807816e-01 8.94401491e-01 5.38855731e-01 -9.94734228e-01
2.06152663e-01 3.95074636e-01 -1.21000618e-01 -1.59810197e+00
1.01799950e-01 -3.92358750e-01 -8.25371802e-01 -5.26091635e-01
-1.30261153e-01 5.51796079e-01 -7.87066400e-01 1.83588874e+00
4.60663795e-01 7.60782510e-02 -9.80167314e-02 1.08824503e+00
8.13654006e-01 5.45845211e-01 1.82452984e-02 -2.64222711e-01
1.35802472e+00 -6.02747560e-01 -1.03839743e+00 2.44940683e-01
4.10200745e-01 -9.19385433e-01 9.37080026e-01 1.44670263e-01
-1.03700018e+00 -6.03391647e-01 -1.04316163e+00 -4.67243880e-01
-3.74521822e-01 2.35958397e-01 6.01722062e-01 7.71864772e-01
-1.40264440e+00 2.84692496e-01 -4.57530200e-01 -1.59640104e-01
6.91441238e-01 7.47780263e-01 -4.66313839e-01 -4.83582884e-01
-1.40805495e+00 4.72569555e-01 1.63873613e-01 3.96275073e-01
-9.46648359e-01 -5.44538021e-01 -8.62027586e-01 3.64153266e-01
-8.98087993e-02 -3.65069270e-01 9.52955961e-01 -6.19306564e-01
-1.17900133e+00 9.39782321e-01 -3.23179990e-01 -3.85535806e-02
1.89591467e-01 -2.99281068e-02 -3.33894581e-01 5.54645002e-01
1.18824549e-01 1.07735503e+00 1.18956172e+00 -1.00416803e+00
-1.78796843e-01 -7.72112966e-01 -9.52421501e-02 2.29047418e-01
-1.21741962e+00 6.68284670e-03 -8.74503195e-01 -5.21331728e-01
3.53322774e-02 -8.10841322e-01 3.39183062e-01 5.39300263e-01
-6.90662628e-03 -6.31637752e-01 1.10638177e+00 -4.29482311e-01
1.37636578e+00 -2.28303456e+00 4.61518839e-02 1.59653276e-01
7.18097091e-02 2.49402955e-01 -4.16631639e-01 5.08921325e-01
1.80509150e-01 -1.49133690e-02 4.54213992e-02 -6.62316084e-01
4.06644404e-01 2.44474858e-01 -2.52314359e-01 7.40156710e-01
2.97325253e-01 9.37350154e-01 -6.06624424e-01 -7.98403025e-01
1.04019672e-01 1.09144318e+00 -6.71872973e-01 1.70705333e-01
2.82311231e-01 -2.49730982e-02 -4.74630266e-01 1.05946684e+00
9.34371829e-01 -1.85361624e-01 -1.43250182e-01 -2.66712010e-01
4.85598564e-01 -1.33237153e-01 -9.55285847e-01 1.79417360e+00
-4.27790731e-01 5.68967819e-01 1.53857172e-01 -9.09839571e-01
8.30023885e-01 3.09190452e-01 3.84941369e-01 -9.77793813e-01
1.10805213e-01 3.16596061e-01 -3.58823121e-01 -4.25793052e-01
5.38067162e-01 5.11993803e-02 9.62055400e-02 3.21326017e-01
1.78423330e-01 8.40805098e-02 9.52717885e-02 2.38782495e-01
6.29656017e-01 -3.21040511e-01 -7.36541748e-02 -2.78570324e-01
7.65834212e-01 -7.12517083e-01 4.02532279e-01 4.79750574e-01
-5.33452630e-01 6.15673602e-01 5.02435640e-02 -4.24734980e-01
-9.29042399e-01 -8.21211159e-01 -5.80984056e-01 1.23721874e+00
3.47570062e-01 -5.39426446e-01 -6.79948032e-01 -3.50503206e-01
3.10842454e-01 -6.58988878e-02 -6.24580383e-01 -3.54582131e-01
-3.39335978e-01 -6.64422691e-01 7.70677090e-01 1.85590923e-01
5.53779364e-01 -8.81512344e-01 -2.27902412e-01 -1.27442211e-01
-1.43981338e-01 -1.16757691e+00 -8.62776875e-01 -4.22016025e-01
-4.28253740e-01 -8.45615327e-01 -7.78436899e-01 -1.03720737e+00
6.37763262e-01 4.85656470e-01 7.85025775e-01 3.71677279e-01
-8.19027603e-01 6.96800768e-01 -2.49822542e-01 -2.67837364e-02
7.66019374e-02 -6.43608645e-02 3.28692228e-01 1.83367252e-01
6.97300732e-01 -3.24777335e-01 -1.15570164e+00 2.90551960e-01
-1.28575087e+00 -8.31625581e-01 4.94307339e-01 1.13663280e+00
4.93576437e-01 7.02701434e-02 6.41104460e-01 9.54646841e-02
7.23879814e-01 -4.61226463e-01 -4.62682545e-01 4.59623337e-01
-6.53824151e-01 7.48154745e-02 3.53916556e-01 -3.73506218e-01
-4.96814817e-01 -2.17336804e-01 9.63357389e-02 -8.17876637e-01
3.37551236e-01 1.81452468e-01 -2.50868231e-01 -4.69801575e-01
2.61710703e-01 5.19445717e-01 9.27428007e-02 -1.76006526e-01
3.33324730e-01 1.06865287e+00 2.56907284e-01 -5.36341190e-01
7.60710597e-01 4.83550698e-01 5.18716276e-02 -8.73437881e-01
-3.95873755e-01 -5.12498915e-01 -4.39939529e-01 -5.51899523e-03
5.73959649e-01 -1.11093616e+00 -1.08358419e+00 3.59342515e-01
-1.07532024e+00 4.68244582e-01 2.58618146e-01 1.72260344e-01
-2.10016832e-01 6.57073855e-01 -7.42760122e-01 -8.80917847e-01
-8.13764393e-01 -1.51524162e+00 1.80865479e+00 2.07884684e-01
2.15991735e-01 -7.56027937e-01 -2.55683035e-01 5.14144719e-01
6.07343853e-01 -2.37422898e-01 7.92752445e-01 -4.23396230e-01
-6.52106941e-01 -4.33755636e-01 -5.50877988e-01 4.98070121e-01
2.08236828e-01 -1.06548481e-01 -1.08565450e+00 -8.49590302e-01
-1.31055027e-01 -9.08243120e-01 8.06140482e-01 -9.30766836e-02
1.55203474e+00 -3.91821384e-01 -1.26572743e-01 6.57531440e-01
1.34320271e+00 2.43262965e-02 5.37561536e-01 1.99427456e-02
4.89943296e-01 4.50573474e-01 5.09503961e-01 7.65315711e-01
4.26401168e-01 8.57724845e-01 4.22860563e-01 2.64181763e-01
-8.47226828e-02 -1.57822832e-01 1.63097247e-01 7.69675612e-01
4.79540914e-01 -3.70094091e-01 -5.87357640e-01 5.39117217e-01
-1.42051721e+00 -8.30022931e-01 6.54542208e-01 2.19537115e+00
9.21432912e-01 -5.25760472e-01 -3.67320180e-01 1.74926877e-01
6.31719112e-01 2.02396616e-01 -6.51256323e-01 -2.16043308e-01
-7.28470385e-02 3.46283019e-01 1.83416203e-01 2.77522743e-01
-1.12899435e+00 8.87593627e-01 5.76335526e+00 1.19758677e+00
-1.32003725e+00 6.40277639e-02 6.99737966e-01 -3.37165177e-01
-2.72851706e-01 -4.61339086e-01 -1.12357116e+00 4.82773900e-01
8.37995470e-01 -9.57421511e-02 5.71168900e-01 7.93488503e-01
-2.99565103e-02 3.76207471e-01 -1.15180409e+00 1.65521872e+00
4.81475711e-01 -9.88967597e-01 5.42035460e-01 2.89527118e-01
5.93859971e-01 -2.47868583e-01 7.36894011e-01 3.97448570e-01
-4.49540615e-01 -1.21190846e+00 4.91150171e-01 1.26936018e-01
1.07016647e+00 -7.28362203e-01 6.23396933e-01 -1.73416644e-01
-1.12003589e+00 -2.44966581e-01 -7.43569195e-01 3.10639769e-01
-2.32878730e-01 2.15040490e-01 -6.38370275e-01 3.17163467e-01
6.74111366e-01 6.03843212e-01 -6.20719612e-01 7.22941220e-01
1.80934563e-01 6.66858852e-02 -5.04956007e-01 1.15584932e-01
3.48216623e-01 -8.00656974e-02 1.13176018e-01 1.10581386e+00
3.55981290e-01 1.46041453e-01 6.98960125e-02 5.25035918e-01
-5.46297729e-01 4.04665142e-01 -6.09559476e-01 -2.15017855e-01
9.18221056e-01 1.30080342e+00 -2.37473592e-01 -2.26100817e-01
-2.60264874e-01 1.14068127e+00 2.87960231e-01 5.00467658e-01
-7.20709741e-01 -2.97212690e-01 8.99806559e-01 -6.76033869e-02
4.44032013e-01 -6.98673278e-02 3.39084685e-01 -1.08458996e+00
2.75338560e-01 -8.82658541e-01 2.94285625e-01 -3.19280237e-01
-1.35849011e+00 4.79296446e-01 -5.63667044e-02 -1.03590286e+00
-2.17643782e-01 -3.97241265e-01 2.22316403e-02 7.36736953e-01
-1.92429221e+00 -1.34854507e+00 -3.81727070e-01 8.91876459e-01
3.39517832e-01 -2.91769683e-01 9.03825283e-01 9.59881365e-01
-6.32380426e-01 1.36103523e+00 4.53570575e-01 1.08909354e-01
9.41283047e-01 -6.58797920e-01 -1.30252376e-01 3.01279604e-01
8.67875367e-02 1.00236785e+00 2.26960033e-01 -2.24228427e-01
-2.03432274e+00 -1.06564236e+00 7.68669188e-01 -7.55019411e-02
4.54392374e-01 -5.48183560e-01 -8.41034710e-01 1.66395813e-01
6.74163252e-02 4.52021956e-01 9.83681262e-01 -3.69015366e-01
-7.83882380e-01 -5.05948663e-01 -1.32439113e+00 3.72219086e-01
8.57167244e-01 -1.03227806e+00 -1.06952108e-01 5.74594498e-01
5.41825414e-01 -7.95871168e-02 -1.07796967e+00 4.70901310e-01
9.83671069e-01 -6.60095811e-01 1.18732345e+00 -4.32800263e-01
2.35318393e-01 -5.08545749e-02 -7.20099926e-01 -6.10901833e-01
-1.66010454e-01 -5.34601569e-01 -9.14704055e-02 1.19382679e+00
-1.71319649e-01 -4.86052632e-01 6.68786585e-01 6.94689453e-01
3.40627253e-01 -9.13685322e-01 -1.31731892e+00 -6.81638420e-01
1.21990301e-01 6.26594052e-02 6.63326144e-01 7.79312611e-01
-3.21476012e-01 2.33677626e-02 -3.87126654e-01 1.45560592e-01
9.74400759e-01 5.76175340e-02 5.55687547e-01 -9.85600412e-01
1.21026061e-01 -3.37547898e-01 -8.98302019e-01 -1.00343680e+00
5.98521054e-01 -9.34795558e-01 -1.43161580e-01 -1.03847945e+00
4.20435637e-01 -4.88598496e-01 -5.46320200e-01 5.26865959e-01
-1.86764866e-01 8.63788784e-01 2.14514226e-01 5.13093531e-01
-8.64289463e-01 1.13443589e+00 1.15540826e+00 -4.80874926e-01
2.31838956e-01 -5.23330867e-01 -5.44627905e-01 9.44752842e-02
4.92895216e-01 -2.86049694e-01 -6.28120363e-01 -7.11070836e-01
-2.70440914e-02 -1.02511823e-01 2.82877594e-01 -8.18630993e-01
3.47223341e-01 2.69002408e-01 5.21830559e-01 -4.36178923e-01
8.26296091e-01 -8.32817078e-01 -1.82608441e-01 3.24552327e-01
-4.25349534e-01 7.26697743e-02 1.59894645e-01 6.49861753e-01
-5.09386420e-01 1.75273810e-02 7.75977850e-01 2.53687590e-01
-4.96786833e-01 8.65270376e-01 2.33595580e-01 -9.20633152e-02
9.28334534e-01 -2.20231056e-01 -9.77852941e-02 -5.35338461e-01
-3.38468134e-01 3.85736734e-01 3.39706093e-01 6.65682495e-01
9.95177329e-01 -1.70209193e+00 -7.15389907e-01 4.83824730e-01
3.33772659e-01 -3.71573478e-01 3.19632322e-01 3.65650684e-01
-4.41108108e-01 7.48997569e-01 -7.12151602e-02 -6.56461060e-01
-1.39232254e+00 5.72662354e-01 2.08480254e-01 1.59964219e-01
1.96863841e-02 1.02950561e+00 2.95454115e-01 -1.12498596e-01
5.70173562e-01 2.01178372e-01 3.63175720e-02 2.22938478e-01
8.54667246e-01 1.15989335e-01 1.96876273e-01 -9.95609581e-01
-4.18572903e-01 7.61681318e-01 -4.91533428e-01 1.30252942e-01
1.05650759e+00 -3.83057624e-01 -3.13083082e-01 -5.85085824e-02
1.99974489e+00 -5.68257093e-01 -9.14146245e-01 -3.14989716e-01
-2.78801501e-01 -8.39915752e-01 2.95761466e-01 -3.30153346e-01
-1.27519524e+00 1.23649549e+00 1.10752690e+00 -2.12816685e-01
1.25463080e+00 -1.64137021e-01 1.13863683e+00 6.38506889e-01
5.09175122e-01 -9.57354784e-01 2.59983331e-01 1.72750011e-01
9.57168400e-01 -1.50114894e+00 -1.80779491e-02 -1.23071484e-01
-1.29066467e-01 1.02871156e+00 4.50759649e-01 1.65306732e-01
7.72776186e-01 -2.41538808e-01 -1.09830545e-02 -2.21323237e-01
-7.32817829e-01 1.41285196e-01 3.14204246e-01 2.90829003e-01
2.85448253e-01 -2.05174297e-01 -3.88391972e-01 9.21579227e-02
4.28076647e-03 -6.41772524e-03 -1.59282893e-01 8.59932482e-01
-2.89101511e-01 -1.06418478e+00 -3.50459665e-01 2.64807910e-01
-5.57947159e-01 -2.39431083e-01 -1.97954670e-01 5.30608237e-01
3.06393132e-02 8.92904341e-01 -2.80734468e-02 -2.47329131e-01
-6.22816682e-02 -2.51824297e-02 6.04403198e-01 -3.21410060e-01
-3.13647717e-01 -3.48690674e-02 -6.79928303e-01 -6.14311635e-01
-2.70259470e-01 -4.04944599e-01 -9.80883360e-01 -3.75562847e-01
-4.95790184e-01 1.75911531e-01 8.56343448e-01 6.41523778e-01
6.72841132e-01 -1.34870440e-01 9.36555386e-01 -8.76579881e-01
-1.14014125e+00 -8.13841939e-01 -6.21615708e-01 6.76045537e-01
4.47777361e-01 -8.74898970e-01 -5.06456971e-01 -9.09540057e-02] | [11.46330451965332, 0.9127699732780457] |
9582b444-fcb6-4046-bcb1-04c3b2db13fc | context-or-no-context-a-preliminary | null | null | https://aclanthology.org/2021.newsum-1.11 | https://aclanthology.org/2021.newsum-1.11.pdf | Context or No Context? A preliminary exploration of human-in-the-loop approach for Incremental Temporal Summarization in meetings | Incremental meeting temporal summarization, summarizing relevant information of partial multi-party meeting dialogue, is emerging as the next challenge in summarization research. Here we examine the extent to which human abstractive summaries of the preceding increments (context) can be combined with extractive meeting dialogue to generate abstractive summaries. We find that previous context improves ROUGE scores. Our findings further suggest that contexts begin to outweigh the dialogue. Using keyphrase extraction and semantic role labeling (SRL), we find that SRL captures relevant information without overwhelming the the model architecture. By compressing the previous contexts by ~70%, we achieve better ROUGE scores over our baseline models. Collectively, these results suggest that context matters, as does the way in which context is presented to the model. | ['Ramesh Manuvinakurike', 'Saurav Sahay', 'Shachi H Kumar', 'Nicole Beckage'] | null | null | null | null | emnlp-newsum-2021-11 | ['semantic-role-labeling', 'keyphrase-extraction'] | ['natural-language-processing', 'natural-language-processing'] | [ 4.87415135e-01 6.84600949e-01 -5.03170490e-01 -1.73138782e-01
-1.31241071e+00 -8.99077773e-01 1.01904905e+00 8.55338693e-01
-5.07330477e-01 1.14289236e+00 1.75053561e+00 -2.08773632e-02
3.23748678e-01 -1.68685809e-01 -2.89604723e-01 -1.87841896e-02
3.31616938e-01 3.78226697e-01 6.10159412e-02 -5.86203277e-01
6.39408410e-01 -1.54803935e-02 -1.11688876e+00 9.30535018e-01
9.15364444e-01 7.94842094e-02 2.33924687e-02 1.10797215e+00
-3.50006342e-01 1.09532881e+00 -1.38620496e+00 -4.31783870e-02
-2.21906513e-01 -7.53241181e-01 -1.23668766e+00 4.26595099e-02
7.55830050e-01 -5.03702998e-01 -5.25225401e-01 4.32472527e-01
5.47375798e-01 3.48732412e-01 7.44453609e-01 -7.79636264e-01
-2.14715570e-01 1.21297085e+00 -3.63821089e-01 5.40324509e-01
9.60553527e-01 -5.53624518e-02 1.44565511e+00 -6.82016611e-01
8.59642684e-01 1.42328382e+00 4.18706119e-01 7.01307416e-01
-1.08940852e+00 -2.93773472e-01 4.49937642e-01 -2.27599218e-01
-5.98316669e-01 -8.89369011e-01 6.78064167e-01 -1.88607842e-01
1.38015497e+00 8.07280362e-01 6.26254916e-01 1.04355311e+00
2.49187469e-01 1.13046026e+00 5.22385836e-01 -5.09701192e-01
5.98929115e-02 -1.92809999e-01 6.70214713e-01 2.32036635e-01
5.76006293e-01 -9.40070510e-01 -9.46177304e-01 -6.58679485e-01
9.33334157e-02 -4.03971583e-01 -3.29923630e-01 5.01088500e-01
-1.10014057e+00 6.49981856e-01 -2.40742058e-01 4.09754455e-01
-5.02060771e-01 2.49751806e-01 7.22117245e-01 1.34046435e-01
5.51714957e-01 1.22263169e+00 -4.49849278e-01 -3.83038878e-01
-1.20609319e+00 6.78212941e-01 1.22653842e+00 9.85120952e-01
3.67168128e-01 1.39586730e-02 -7.89162755e-01 7.83641815e-01
-1.02052309e-01 2.60953575e-01 3.82612795e-01 -1.27166510e+00
8.04572940e-01 6.19187295e-01 4.60491508e-01 -7.84942925e-01
-3.48654479e-01 -9.94208306e-02 -3.37218523e-01 -1.05124879e+00
1.96704909e-01 -5.58216035e-01 -5.30017734e-01 1.70052397e+00
-2.45747082e-02 -2.53754914e-01 4.21954483e-01 2.49691680e-01
1.27792764e+00 1.04134786e+00 3.11709523e-01 -9.03400779e-01
1.48363781e+00 -1.04957938e+00 -1.16291809e+00 -6.35168910e-01
7.19648242e-01 -9.93422985e-01 1.02581286e+00 6.77919835e-02
-1.53991401e+00 -1.87689528e-01 -1.15929508e+00 -3.92265022e-01
2.16082618e-01 1.02325842e-01 3.13859642e-01 1.55086011e-01
-1.28153694e+00 4.52523142e-01 -5.94713330e-01 -5.96133113e-01
-1.00998200e-01 7.33155757e-03 -9.88245457e-02 9.37063172e-02
-1.21307063e+00 8.14764202e-01 3.48810047e-01 -3.52854729e-01
-4.60410774e-01 -6.94374084e-01 -9.83426452e-01 7.34318346e-02
5.29767632e-01 -9.85400558e-01 1.89927828e+00 -2.46076494e-01
-1.17801487e+00 4.02267456e-01 -8.45970511e-01 -5.35486579e-01
1.13115691e-01 -5.97025335e-01 1.56545714e-02 5.62413096e-01
2.85431772e-01 4.99063015e-01 3.89157057e-01 -1.34302676e+00
-9.09657121e-01 -1.14878222e-01 1.69870064e-01 9.68672395e-01
-3.76698792e-01 3.22138757e-01 -3.54979992e-01 -6.67593181e-01
4.35282812e-02 -7.82401204e-01 -1.58254713e-01 -1.10978293e+00
-6.52614713e-01 -6.04606986e-01 6.04549468e-01 -1.15534139e+00
2.00371599e+00 -1.68254542e+00 2.43567184e-01 -2.58683771e-01
4.97884929e-01 -2.38944143e-02 3.49671696e-03 1.06784797e+00
4.28734690e-01 4.83230680e-01 -2.28144124e-01 -4.69233930e-01
-4.96701822e-02 -3.68602201e-02 -5.40175557e-01 -7.22554252e-02
-8.05876181e-02 1.09785306e+00 -9.80993748e-01 -7.29622483e-01
-2.69355267e-01 -1.12644218e-01 -4.20751810e-01 1.54869080e-01
-5.36179662e-01 1.92439228e-01 -5.66739321e-01 1.78957909e-01
8.53440538e-02 -9.89630148e-02 3.83688748e-01 6.90632313e-02
-1.46418184e-01 1.04537010e+00 -6.36725664e-01 1.70239282e+00
-2.07039565e-01 8.89720142e-01 2.06975788e-01 -3.87206644e-01
5.60407758e-01 4.96365309e-01 2.91562080e-01 -2.67029494e-01
-9.11999792e-02 -8.20449144e-02 -4.66493480e-02 -4.35677618e-01
1.51696551e+00 -2.78081391e-02 -6.98627353e-01 8.71903121e-01
-2.50682473e-01 -4.99329418e-01 5.68263829e-01 1.06252468e+00
1.10470366e+00 -4.07073319e-01 5.95665276e-01 -2.39234731e-01
3.50926012e-01 5.20160496e-01 4.85904604e-01 9.77338016e-01
-1.66617647e-01 4.73393351e-01 7.61579037e-01 9.80044827e-02
-1.01687908e+00 -5.09851456e-01 4.78750408e-01 1.26413178e+00
1.84277184e-02 -1.03533399e+00 -8.99132371e-01 -8.27065170e-01
-1.48370668e-01 1.26025426e+00 -2.97384083e-01 -4.48576398e-02
-1.22070026e+00 -5.12889504e-01 7.87124515e-01 5.20032525e-01
2.66078055e-01 -9.26592469e-01 -4.07254875e-01 3.54569316e-01
-9.34006989e-01 -1.14330411e+00 -1.00284100e+00 -2.18724206e-01
-1.07999992e+00 -8.38019550e-01 -4.67906058e-01 -6.57917857e-01
3.62490356e-01 5.32022715e-01 1.22392917e+00 3.73712853e-02
3.31432283e-01 8.08966637e-01 -2.98348457e-01 -5.35902321e-01
-8.31021130e-01 4.64503109e-01 -9.96879786e-02 -7.38048375e-01
2.73454577e-01 -3.32823753e-01 -5.51320076e-01 -3.17921937e-01
-8.77810955e-01 2.26631448e-01 4.30635542e-01 5.10231614e-01
8.63756910e-02 -3.04437935e-01 9.77626026e-01 -1.10573292e+00
1.51917911e+00 -3.44107181e-01 3.79889816e-01 3.43156785e-01
-3.60231936e-01 9.59776714e-02 4.42905039e-01 -7.26988316e-02
-1.37036407e+00 -6.03815079e-01 1.19976602e-01 5.04014790e-01
2.89521553e-02 6.47901356e-01 -3.10604181e-02 8.33454490e-01
7.40692437e-01 1.09970011e-01 -1.96483344e-01 -2.95739889e-01
4.09237951e-01 6.73620701e-01 4.64024931e-01 -5.95916927e-01
3.72426569e-01 2.70400256e-01 -5.53371906e-01 -1.32373738e+00
-9.69315708e-01 -7.82306373e-01 -3.12241137e-01 -1.35922790e-01
8.87112617e-01 -1.07010365e+00 -2.39576027e-01 -8.83702412e-02
-1.48054910e+00 -2.43886799e-01 -4.23873752e-01 2.10431606e-01
-2.94708371e-01 7.43101597e-01 -9.93096709e-01 -8.74962628e-01
-7.99750924e-01 -6.15032971e-01 9.97504711e-01 4.46674258e-01
-1.24687588e+00 -8.55135143e-01 5.20292632e-02 7.39605010e-01
2.55433973e-02 1.26108885e-01 9.82717693e-01 -1.26912928e+00
-9.05310884e-02 -1.29226238e-01 1.51051730e-01 -3.60024199e-02
3.11044008e-01 -7.47476965e-02 -7.18269944e-01 -1.70658052e-01
1.80042356e-01 -2.27850825e-01 1.02131116e+00 4.90798563e-01
4.53941524e-01 -1.02964509e+00 -3.55120987e-01 -4.10274118e-01
6.34072304e-01 6.47688657e-02 2.79756635e-01 1.00860238e-01
6.61124408e-01 9.44153428e-01 7.57725358e-01 5.16480088e-01
8.26333225e-01 2.81456530e-01 -4.87644762e-01 3.19331825e-01
-3.09274226e-01 -6.94331467e-01 5.83477318e-01 1.33673644e+00
2.18901396e-01 -4.81686503e-01 -7.98263252e-01 6.63041770e-01
-1.97150326e+00 -1.09344018e+00 -3.90061662e-02 1.85232425e+00
1.21073270e+00 2.51023352e-01 1.04812540e-01 -1.81248218e-01
6.76017463e-01 7.73139477e-01 -2.48037964e-01 -6.96369290e-01
-1.41127288e-01 -2.07800910e-01 1.57475337e-01 9.59466577e-01
-7.45634019e-01 1.05528677e+00 7.13084698e+00 5.72392642e-01
-5.23383021e-01 -8.32444131e-02 5.47157049e-01 -3.61529201e-01
-8.41848850e-01 2.07749039e-01 -1.02596605e+00 2.02451080e-01
1.12288141e+00 -7.51319051e-01 -9.09140110e-02 4.75005478e-01
6.38612866e-01 -4.43032682e-01 -1.29675102e+00 3.91818851e-01
4.13690656e-01 -1.43559837e+00 4.08375800e-01 -1.05793718e-02
9.54546154e-01 -3.72880399e-01 -3.26980442e-01 4.88083541e-01
4.91473854e-01 -7.79028773e-01 4.60145533e-01 2.61338651e-01
4.46747154e-01 -6.46861970e-01 5.57658494e-01 6.47471130e-01
-9.82035279e-01 3.10382098e-02 -1.28158331e-01 -2.14198753e-01
4.06141192e-01 1.04554102e-01 -1.52577066e+00 5.35376251e-01
-2.52753831e-02 5.06179452e-01 -5.74503422e-01 4.11100626e-01
-3.75654846e-01 8.38286281e-01 -1.50899172e-01 -2.29336977e-01
2.86733896e-01 7.96425641e-02 1.14525580e+00 1.68879414e+00
-7.91068524e-02 5.75168908e-01 4.05412763e-01 2.27887899e-01
-3.87476742e-01 1.42993048e-01 -5.66372633e-01 -2.89205015e-01
9.24792290e-01 8.71095657e-01 -4.82433617e-01 -9.31889057e-01
-1.43660069e-01 9.19483542e-01 -1.23418495e-01 4.92949575e-01
-3.37416083e-01 -3.41430038e-01 3.26113522e-01 1.71605647e-02
-1.79152355e-01 -3.65768552e-01 -4.81460512e-01 -1.22892404e+00
9.77454633e-02 -1.14438879e+00 6.73089683e-01 -6.02492690e-01
-6.81341290e-01 2.72057831e-01 3.85673702e-01 -6.69820964e-01
-5.90991020e-01 4.11451817e-01 -8.21397424e-01 6.45613253e-01
-1.22685647e+00 -8.49882007e-01 -4.51693200e-02 -5.59718423e-02
1.44057226e+00 1.91579938e-01 6.78940415e-01 -3.20293188e-01
-5.81360459e-01 4.92154211e-01 -2.12523654e-01 -1.24016413e-02
9.47465122e-01 -1.35747778e+00 6.52551174e-01 1.01246858e+00
5.17593026e-02 1.02205849e+00 1.25751412e+00 -1.13859141e+00
-1.30906463e+00 -6.45973742e-01 1.59162450e+00 -7.80835032e-01
4.74276781e-01 -6.56526908e-02 -9.29742992e-01 7.56689727e-01
9.47200775e-01 -1.13305783e+00 9.90593135e-01 4.38889973e-02
-3.25356238e-02 2.70649910e-01 -8.52049708e-01 1.09267211e+00
9.17511761e-01 -5.01090884e-01 -1.49361587e+00 1.66042894e-01
1.35756648e+00 -3.61208886e-01 -5.66465974e-01 2.46955127e-01
3.08831960e-01 -3.49507391e-01 5.62695146e-01 -7.39750087e-01
6.49166107e-01 1.42343584e-02 -1.02814414e-01 -1.42982483e+00
-2.45285826e-03 -1.12428129e+00 -5.23134708e-01 1.32591009e+00
6.95263267e-01 -3.12787145e-01 7.24429727e-01 1.22346759e+00
-5.20496070e-01 -3.58605742e-01 -5.55518806e-01 -1.53933287e-01
1.08354114e-01 2.24526022e-02 3.48675877e-01 8.05055022e-01
8.02597880e-01 1.15969181e+00 -1.05397061e-01 -2.29214102e-01
3.18825841e-01 -7.92003050e-02 8.68835628e-01 -1.17614913e+00
-6.12820983e-02 -5.73919296e-01 5.73182464e-01 -1.43979776e+00
2.82319725e-01 -6.41152322e-01 1.42309591e-01 -2.39742970e+00
4.42578226e-01 1.68058366e-01 1.67306229e-01 2.45551378e-01
-6.68539882e-01 -2.60478318e-01 3.63494456e-01 4.46007431e-01
-8.54797840e-01 5.08075655e-01 1.17246163e+00 -2.35124916e-01
-7.91055620e-01 -1.04744628e-01 -1.45349789e+00 7.79809892e-01
7.57990003e-01 -2.46348768e-01 -5.34758210e-01 -3.31020713e-01
4.93715778e-02 5.09348929e-01 -4.44295973e-01 -4.72180754e-01
5.18845737e-01 -3.65152746e-01 1.80643991e-01 -7.72157490e-01
4.10068363e-01 -2.64953785e-02 -4.66724783e-01 1.95733473e-01
-1.11614680e+00 2.32401207e-01 5.01119554e-01 5.09731650e-01
-2.39626363e-01 -3.68921161e-01 1.81606188e-01 -3.50235164e-01
-3.43730360e-01 -3.33469033e-01 -1.00434744e+00 6.60485685e-01
4.34281170e-01 -2.96117127e-01 -5.54572284e-01 -8.73877227e-01
-5.87106705e-01 5.29889405e-01 3.66262585e-01 2.34998301e-01
4.68250364e-01 -8.39826107e-01 -1.12907350e+00 -6.25221014e-01
9.18011665e-02 2.07529664e-01 2.25698665e-01 5.33829510e-01
-1.78508192e-01 7.64791548e-01 3.92533541e-01 -1.53942347e-01
-1.60571027e+00 -1.19808570e-01 -2.77182758e-01 -6.50142968e-01
-7.10243046e-01 6.36847019e-01 1.59389168e-01 -1.44754155e-02
3.58062685e-01 -3.84247988e-01 -7.00375080e-01 6.32623494e-01
7.82154262e-01 5.88268161e-01 -6.87697306e-02 -4.39397395e-01
-6.34740442e-02 1.19715405e-03 -5.59007764e-01 -7.13894784e-01
1.11278534e+00 -5.16799331e-01 -2.27166742e-01 6.21353805e-01
8.26179802e-01 6.97998524e-01 -9.14567769e-01 -2.57501990e-01
4.69796032e-01 5.85464425e-02 -2.03579620e-01 -7.22939610e-01
6.31245747e-02 2.65996248e-01 -7.83227146e-01 4.23538923e-01
7.26559639e-01 1.75999627e-01 9.70228851e-01 7.71427870e-01
-1.04342960e-01 -1.39495730e+00 3.63308668e-01 9.03256595e-01
1.22278142e+00 -8.99008155e-01 5.60567796e-01 -2.74287164e-01
-1.20395303e+00 7.98810720e-01 7.60266006e-01 1.82050616e-01
-2.57904474e-02 4.11447473e-02 -1.40691206e-01 -2.12812036e-01
-1.23064208e+00 1.94382578e-01 2.77725071e-01 7.53149241e-02
7.88662136e-01 -2.43909117e-02 -8.64417672e-01 8.23067009e-01
-6.18404150e-01 -7.14825928e-01 1.21352637e+00 1.08496070e+00
-9.02583480e-01 -8.00148010e-01 -3.73978943e-01 6.66552663e-01
-6.75294876e-01 -3.10322702e-01 -1.21294510e+00 6.74847722e-01
-7.94916570e-01 1.47471070e+00 2.40905341e-02 -8.33235681e-02
5.28162837e-01 4.62799400e-01 1.69996113e-01 -1.26744580e+00
-1.08277869e+00 4.49584067e-01 9.75330174e-01 -2.14019924e-01
-1.09312214e-01 -7.96434700e-01 -1.48326218e+00 -3.58157545e-01
-1.89440891e-01 7.67518759e-01 2.94010341e-01 8.45221341e-01
5.63405037e-01 6.01943612e-01 4.27825123e-01 -6.02208197e-01
-5.96800327e-01 -1.48817217e+00 2.42038313e-02 8.60441178e-02
5.70945442e-01 1.44862458e-01 -4.54500616e-01 4.38896306e-02] | [12.570655822753906, 9.362797737121582] |
64a72cc5-1a7c-41a0-b5fb-1212e07d170f | srcn3d-sparse-r-cnn-3d-surround-view-camera | 2206.14451 | null | https://arxiv.org/abs/2206.14451v3 | https://arxiv.org/pdf/2206.14451v3.pdf | SRCN3D: Sparse R-CNN 3D for Compact Convolutional Multi-View 3D Object Detection and Tracking | Detection and tracking of moving objects is an essential component in environmental perception for autonomous driving. In the flourishing field of multi-view 3D camera-based detectors, different transformer-based pipelines are designed to learn queries in 3D space from 2D feature maps of perspective views, but the dominant dense BEV query mechanism is computationally inefficient. This paper proposes Sparse R-CNN 3D (SRCN3D), a novel two-stage fully-sparse detector that incorporates sparse queries, sparse attention with box-wise sampling, and sparse prediction. SRCN3D adopts a cascade structure with the twin-track update of both a fixed number of query boxes and latent query features. Our novel sparse feature sampling module only utilizes local 2D region of interest (RoI) features calculated by the projection of 3D query boxes for further box refinement, leading to a fully-convolutional and deployment-friendly pipeline. For multi-object tracking, motion features, query features and RoI features are comprehensively utilized in multi-hypotheses data association. Extensive experiments on nuScenes dataset demonstrate that SRCN3D achieves competitive performance in both 3D object detection and multi-object tracking tasks, while also exhibiting superior efficiency compared to transformer-based methods. Code and models are available at https://github.com/synsin0/SRCN3D. | ['Diange Yang', 'Kun Jiang', 'Shiqi Sun', 'Jiaxin Li', 'Yunlong Wang', 'Yifan Sun', 'Jingyan Shen', 'Yining Shi'] | 2022-06-29 | null | null | null | null | ['3d-multi-object-tracking'] | ['computer-vision'] | [-3.78985316e-01 -5.43010354e-01 -2.53277481e-01 -4.13128316e-01
-7.02536464e-01 -4.09858972e-01 4.76666152e-01 -2.76250064e-01
-4.72394466e-01 -9.04127955e-02 7.05163330e-02 4.16624807e-02
1.87646061e-01 -6.66926265e-01 -8.21863294e-01 -3.54917884e-01
3.44339490e-01 4.01904106e-01 1.21560121e+00 -1.68490723e-01
7.85078555e-02 9.15296316e-01 -1.97609866e+00 2.10157022e-01
1.52122691e-01 1.11363947e+00 4.70431805e-01 7.20476151e-01
5.43491468e-02 4.42772001e-01 -1.80288807e-01 -1.22405365e-01
5.07786393e-01 2.10156381e-01 7.68553913e-02 6.76486548e-03
9.57597673e-01 -6.05208516e-01 -7.01017499e-01 8.32903147e-01
6.93449259e-01 1.98175058e-01 2.51734614e-01 -1.31793439e+00
-3.32092732e-01 -1.97389662e-01 -8.54180336e-01 6.27853155e-01
1.92769572e-01 4.48394328e-01 9.27950203e-01 -1.59100318e+00
5.05455017e-01 1.36399317e+00 7.13579416e-01 5.04302740e-01
-9.50104952e-01 -1.00493336e+00 3.19676369e-01 3.37393790e-01
-1.62247431e+00 -5.50165951e-01 7.46787012e-01 -5.61361015e-01
1.21216381e+00 1.87083393e-01 8.71425927e-01 9.57965970e-01
2.40097433e-01 8.26525271e-01 6.27211988e-01 2.68078297e-01
-4.92160730e-02 1.17037728e-01 2.40156859e-01 9.20390487e-01
4.67680722e-01 4.34315771e-01 -9.87272084e-01 -1.19956918e-01
9.47823465e-01 5.09721577e-01 1.99926436e-01 -8.47013533e-01
-1.23200178e+00 8.37347329e-01 6.16990089e-01 -2.03438446e-01
-1.49905056e-01 2.10256949e-01 2.25769922e-01 -1.41132623e-01
5.61460376e-01 -7.50138462e-02 -3.65077019e-01 2.36561283e-01
-7.74069667e-01 5.63715696e-01 1.70599788e-01 1.41267574e+00
8.59984338e-01 9.49192271e-02 -2.88611054e-01 5.09424210e-01
5.66535771e-01 1.04452264e+00 1.63684100e-01 -9.21995044e-01
3.30727309e-01 9.17044342e-01 2.87937284e-01 -8.57192278e-01
-4.88934666e-01 -6.44790411e-01 -6.45706117e-01 3.88626695e-01
-4.14982177e-02 2.53713608e-01 -8.34007621e-01 1.14665306e+00
9.40209270e-01 1.71456248e-01 -2.30204448e-01 1.25686359e+00
1.37860358e+00 5.06592870e-01 -3.35008591e-01 4.02899623e-01
1.47744513e+00 -1.04915416e+00 -2.63120234e-01 -4.33275908e-01
3.63786459e-01 -6.49497747e-01 6.53105378e-01 -8.47418755e-02
-9.70309377e-01 -1.07039022e+00 -8.03547442e-01 -5.58965206e-01
-3.17973584e-01 4.80789393e-01 3.76306862e-01 3.06114703e-01
-8.72429490e-01 -2.43084490e-01 -9.73626494e-01 -2.33270958e-01
7.59392023e-01 3.66652787e-01 -3.86209786e-01 -2.79456884e-01
-6.31302178e-01 9.27867711e-01 8.93242285e-02 1.91656262e-01
-1.27185631e+00 -8.14936519e-01 -1.10750818e+00 -1.03748530e-01
5.33442020e-01 -9.78965759e-01 1.09605539e+00 1.34505779e-01
-1.04581213e+00 9.31794524e-01 -7.27847695e-01 -5.18620133e-01
3.51210982e-01 -5.91246426e-01 -1.48691401e-01 2.55682785e-02
4.95950311e-01 9.83844340e-01 1.16239417e+00 -9.58662987e-01
-1.02310812e+00 -6.30902708e-01 -1.43701389e-01 3.64052504e-01
1.55803785e-01 9.48250219e-02 -7.72871315e-01 -1.75543070e-01
4.26752537e-01 -8.50683272e-01 -4.84853059e-01 4.89620686e-01
-1.65060103e-01 -4.75849599e-01 1.29994047e+00 -1.46171540e-01
7.75581360e-01 -2.36834121e+00 -1.04467561e-02 -2.06401125e-01
6.37976348e-01 1.09611444e-01 8.81993994e-02 5.83706163e-02
3.56902152e-01 -4.86943126e-01 1.15145497e-01 -6.06173694e-01
-1.08610615e-01 -9.07451436e-02 -3.59490126e-01 7.68851578e-01
6.44279361e-01 1.06031561e+00 -8.46074045e-01 -4.09371138e-01
6.78073943e-01 5.07953644e-01 -6.72795832e-01 1.32539764e-01
-8.97537544e-02 3.52146953e-01 -7.71877050e-01 1.02907169e+00
7.31830299e-01 -3.41063648e-01 -6.87957764e-01 -3.97782564e-01
-6.63494706e-01 8.01732764e-02 -1.29721165e+00 1.75003707e+00
-1.28055781e-01 6.38458371e-01 1.88307554e-01 -5.38662314e-01
1.12815964e+00 -1.00387737e-01 4.43173826e-01 -6.34475589e-01
5.01975082e-02 2.34723046e-01 -3.01757604e-01 -2.68837273e-01
8.58166277e-01 4.03744102e-01 -1.19582020e-01 -7.03767464e-02
1.98817365e-02 -3.77091408e-01 -7.22857416e-02 2.81781554e-01
1.12251270e+00 3.27876091e-01 3.47136438e-01 3.65447104e-02
5.43596625e-01 3.37846309e-01 6.85556889e-01 8.21880221e-01
-4.22327340e-01 5.58164477e-01 -8.06895718e-02 -5.02282262e-01
-1.13948631e+00 -9.13344264e-01 -2.51910925e-01 9.45484757e-01
6.36573493e-01 -4.27063197e-01 2.25889891e-01 -5.35328209e-01
5.44742405e-01 2.74016321e-01 -4.60349768e-01 -1.21985218e-02
-7.33455241e-01 -1.42936558e-01 1.35310248e-01 6.32648706e-01
4.35819715e-01 -5.84952056e-01 -1.27632678e+00 -3.03724315e-03
2.52873987e-01 -1.31343842e+00 -7.79704809e-01 3.33710283e-01
-7.57290959e-01 -1.02573419e+00 -5.56699097e-01 -5.04959702e-01
5.70180893e-01 1.32421982e+00 7.95551300e-01 -2.47507572e-01
-4.33629423e-01 6.19361281e-01 -2.72337526e-01 -6.21786416e-01
2.30483904e-01 -1.45580783e-01 1.42426893e-01 -2.66323425e-03
6.85455382e-01 -2.53694564e-01 -7.52054155e-01 6.17574215e-01
-3.38504195e-01 9.38151702e-02 7.81927347e-01 6.80993974e-01
9.50885773e-01 -4.18206543e-01 1.64358933e-02 -4.17152375e-01
-2.76915103e-01 -5.26617885e-01 -1.18429887e+00 -4.31191534e-01
-1.37987539e-01 -3.08516622e-01 1.14343710e-01 -3.01697493e-01
-7.63699710e-01 6.31593347e-01 -1.85494691e-01 -1.25628281e+00
-2.12059706e-01 -1.79091901e-01 -1.13499857e-01 -2.36575767e-01
5.73557436e-01 4.46133435e-01 -6.37271777e-02 -4.92456377e-01
3.47794026e-01 3.20346951e-01 3.75114888e-01 7.20912442e-02
1.17419815e+00 9.49726701e-01 8.86302218e-02 -9.08246934e-01
-1.02288568e+00 -1.32000017e+00 -7.19275892e-01 -4.52077359e-01
1.07558465e+00 -1.70946527e+00 -8.06404352e-01 4.41996455e-01
-1.20770586e+00 -2.25285180e-02 -4.07648534e-01 6.66112840e-01
-3.10958028e-01 -8.11284631e-02 -4.64957394e-03 -8.80791485e-01
-2.00386688e-01 -1.18773508e+00 1.86247957e+00 3.98250312e-01
2.15808839e-01 -2.86022604e-01 1.04143319e-03 5.23100793e-01
3.96265864e-01 -8.56381804e-02 6.34270981e-02 -4.73006248e-01
-1.26074529e+00 -3.54705364e-01 -4.45947498e-01 -7.92918610e-04
-2.11121693e-01 -2.78703213e-01 -1.08726442e+00 -3.38879466e-01
-1.25718072e-01 -3.19158316e-01 1.15745234e+00 5.57161093e-01
8.46132815e-01 3.24733853e-01 -6.97980046e-01 8.66003990e-01
1.29262006e+00 -8.72995332e-03 -5.99342547e-02 1.94291957e-02
1.03952587e+00 3.10347080e-01 8.82919550e-01 6.35810316e-01
6.68101132e-01 8.43115807e-01 6.75198317e-01 -3.34474631e-02
-2.46535704e-01 -3.57135534e-01 5.76087952e-01 5.11080503e-01
2.42836639e-01 1.81111306e-01 -7.68198967e-01 5.85174441e-01
-1.73037851e+00 -1.15510786e+00 -3.61001074e-01 1.92678535e+00
2.00843707e-01 4.32268918e-01 3.53849828e-01 -3.45006257e-01
6.50903046e-01 4.68614936e-01 -1.16562235e+00 4.22329277e-01
-1.20964020e-01 -4.78169993e-02 8.57577205e-01 3.12391371e-01
-1.29695129e+00 1.00255322e+00 5.03852701e+00 6.18204355e-01
-9.05472457e-01 3.64903688e-01 4.66887429e-02 -5.36121905e-01
-1.38693437e-01 -6.36468679e-02 -2.03282619e+00 1.29386991e-01
5.21153986e-01 4.47622128e-03 -2.36708790e-01 1.22083771e+00
3.57202977e-01 -1.59423530e-01 -8.54078352e-01 1.20741117e+00
1.07134372e-01 -1.58718216e+00 -1.91890538e-01 2.02514231e-02
5.16624987e-01 8.36215913e-01 3.59730721e-02 3.12784463e-01
3.35805833e-01 -4.94329512e-01 1.13569093e+00 4.12229568e-01
6.10343099e-01 -4.50069934e-01 3.60453546e-01 5.64887226e-01
-1.79836953e+00 -2.94077665e-01 -6.61843359e-01 1.45839199e-01
1.81939125e-01 5.60470939e-01 -7.14183271e-01 4.05871272e-01
1.28721988e+00 1.28357220e+00 -7.30849028e-01 1.26217592e+00
1.59169585e-01 2.37126768e-01 -5.45570850e-01 -7.06981122e-02
1.75063893e-01 7.06215426e-02 9.34665442e-01 1.00073016e+00
3.47317547e-01 1.48946270e-01 4.09213692e-01 9.99151051e-01
1.62521660e-01 -2.83061922e-01 -8.56980979e-01 5.49379289e-01
6.87839270e-01 1.48949993e+00 -5.75171769e-01 -2.86005914e-01
-8.43275249e-01 4.92798060e-01 2.38478199e-01 1.21651068e-02
-9.25337672e-01 -8.51397142e-02 9.95673299e-01 3.69329959e-01
1.06020856e+00 -5.98641455e-01 4.47812229e-02 -1.12048423e+00
1.20857157e-01 -4.01225150e-01 1.05633825e-01 -8.56980026e-01
-1.18762434e+00 3.26889455e-01 1.34995759e-01 -1.64802277e+00
2.69568145e-01 -4.76524323e-01 -4.23798651e-01 8.73094022e-01
-1.79026234e+00 -1.40743971e+00 -6.11025751e-01 7.45302737e-01
9.23513353e-01 -1.84618801e-01 2.88480937e-01 3.53341728e-01
-6.11802220e-01 1.03037998e-01 -1.93113655e-01 -4.62285206e-02
5.29459655e-01 -8.88585627e-01 8.22421014e-01 9.11760092e-01
8.89527351e-02 3.09916794e-01 2.40435243e-01 -8.13032329e-01
-1.82007146e+00 -1.61693466e+00 5.59689701e-01 -9.03989017e-01
4.37286526e-01 -6.96797013e-01 -7.34714568e-01 5.98232210e-01
-2.18378708e-01 6.79390669e-01 3.50399017e-01 -3.06523770e-01
-2.71263480e-01 -2.63772309e-01 -5.80123186e-01 2.17901230e-01
1.42447460e+00 -4.93507981e-01 -4.23619032e-01 3.73565286e-01
9.72619951e-01 -9.77476239e-01 -6.29692078e-01 3.94000441e-01
5.59639335e-01 -1.21508873e+00 1.28459275e+00 -1.94972500e-01
-1.39749616e-01 -9.05102193e-01 -2.56297231e-01 -5.30091941e-01
-5.80192447e-01 -3.72831255e-01 -4.50160950e-01 6.49819672e-01
1.09950542e-01 -5.64835310e-01 1.07995296e+00 1.42078266e-01
-6.62042260e-01 -7.01286316e-01 -1.15417278e+00 -5.72343111e-01
-7.08747625e-01 -6.13141894e-01 3.97198439e-01 2.77324349e-01
-8.93042207e-01 5.50362229e-01 -3.41929257e-01 6.80535972e-01
8.34286332e-01 4.08685118e-01 1.38645017e+00 -1.35649824e+00
-6.34913668e-02 -3.06020051e-01 -6.63876414e-01 -1.78519106e+00
-1.17998771e-01 -8.52541327e-01 8.10330212e-02 -1.31094980e+00
8.02624896e-02 -3.29459250e-01 2.95234602e-02 1.35658264e-01
-1.28345266e-01 2.48008817e-01 2.14799210e-01 4.27785248e-01
-9.36166644e-01 8.78523052e-01 1.28530359e+00 -1.61488339e-01
-2.67406493e-01 2.83663452e-01 -2.98536003e-01 6.21287286e-01
3.23965281e-01 -5.97460985e-01 -2.72016048e-01 -5.32744527e-01
8.75989422e-02 -4.96645160e-02 9.30037796e-01 -1.19252837e+00
6.57827497e-01 2.46333070e-02 7.82623947e-01 -1.77020037e+00
8.60494375e-01 -9.00297999e-01 5.09139383e-04 4.70115423e-01
1.22877536e-02 3.26709986e-01 3.09732348e-01 9.07689869e-01
-1.47375762e-01 1.41238526e-01 8.20264697e-01 -3.20210189e-01
-1.19301558e+00 7.40686476e-01 -1.13226831e-01 -8.07404965e-02
1.15975928e+00 -5.22357941e-01 -1.17948443e-01 2.13246152e-01
-4.91077542e-01 5.80273449e-01 3.17839473e-01 6.11674130e-01
1.15048647e+00 -1.37082851e+00 -6.93940043e-01 6.60532475e-01
3.74853253e-01 7.78703153e-01 4.74647552e-01 1.00894570e+00
-2.40395844e-01 8.17489445e-01 -2.31360104e-02 -1.27987003e+00
-1.29924059e+00 3.81318539e-01 3.73292297e-01 1.02543086e-01
-9.21518922e-01 1.29282439e+00 4.18228030e-01 -3.38783741e-01
2.87122309e-01 -6.04853272e-01 -1.83948934e-01 -1.03241630e-01
5.29348671e-01 3.47462296e-01 -5.41518256e-03 -8.59646261e-01
-5.85668862e-01 8.42624843e-01 -2.04591349e-01 1.86580241e-01
1.30740273e+00 -2.15958223e-01 3.65491331e-01 4.58894700e-01
8.47076952e-01 9.92973521e-03 -1.63014543e+00 -5.60368598e-01
-3.18643630e-01 -7.23858476e-01 3.09462100e-01 1.10038809e-01
-9.00482178e-01 7.91117728e-01 6.49863422e-01 -2.95971096e-01
6.89797163e-01 3.99211645e-01 7.15367794e-01 4.62727487e-01
3.88688356e-01 -5.24906456e-01 2.18812153e-01 6.84674382e-01
8.23932767e-01 -1.59016573e+00 1.92416355e-01 -3.41018230e-01
-4.08049732e-01 8.07410717e-01 1.17300534e+00 -2.95174330e-01
8.56268883e-01 1.99097484e-01 -2.30727121e-01 -5.64448893e-01
-9.29426968e-01 -5.93292475e-01 4.38976437e-01 5.43840349e-01
-1.75653353e-01 -3.47784579e-01 5.04283845e-01 4.60207760e-01
1.07106708e-01 -2.79410243e-01 8.92846882e-02 9.49786961e-01
-6.70806408e-01 -4.99526978e-01 -4.98057187e-01 5.74767172e-01
1.68543786e-01 8.09202567e-02 -2.61318684e-02 8.06003511e-01
4.12067443e-01 5.53119957e-01 2.41315246e-01 -4.25622642e-01
6.95937097e-01 -2.79247463e-01 2.12837473e-01 -9.77315426e-01
-4.42881584e-01 2.24819779e-01 -2.59918422e-01 -9.76291597e-01
-5.48049688e-01 -1.03309095e+00 -1.00013423e+00 1.48037657e-01
-5.75921059e-01 -3.16575676e-01 5.70542276e-01 6.03526175e-01
7.83073425e-01 4.37054396e-01 5.38255215e-01 -1.24719965e+00
-4.63433623e-01 -7.14317322e-01 -2.85060078e-01 -2.25440919e-01
6.64221287e-01 -1.04048789e+00 -1.66378841e-01 -1.45841062e-01] | [7.70259428024292, -2.3428804874420166] |
297d6be7-2105-4030-aca4-2c0dc82022d9 | phmospell-phonological-and-morphological | null | null | https://aclanthology.org/2021.acl-long.464 | https://aclanthology.org/2021.acl-long.464.pdf | PHMOSpell: Phonological and Morphological Knowledge Guided Chinese Spelling Check | Chinese Spelling Check (CSC) is a challenging task due to the complex characteristics of Chinese characters. Statistics reveal that most Chinese spelling errors belong to phonological or visual errors. However, previous methods rarely utilize phonological and morphological knowledge of Chinese characters or heavily rely on external resources to model their similarities. To address the above issues, we propose a novel end-to-end trainable model called PHMOSpell, which promotes the performance of CSC with multi-modal information. Specifically, we derive pinyin and glyph representations for Chinese characters from audio and visual modalities respectively, which are integrated into a pre-trained language model by a well-designed adaptive gating mechanism. To verify its effectiveness, we conduct comprehensive experiments and ablation tests. Experimental results on three shared benchmarks demonstrate that our model consistently outperforms previous state-of-the-art models. | ['Jing Xiao', 'Shaojun Wang', 'Minchuan Chen', 'ZhiYu Zhang', 'Weiwei Jiang', 'Junjie Li', 'Li Huang'] | 2021-08-01 | null | null | null | acl-2021-5 | ['chinese-spell-checking'] | ['natural-language-processing'] | [ 1.77131057e-01 -7.36960292e-01 -9.77868773e-03 -1.33729890e-01
-1.08500338e+00 -6.37379467e-01 5.35481095e-01 1.18755065e-01
-6.37610435e-01 3.80645186e-01 5.35600483e-01 -4.15870041e-01
7.35171497e-01 -4.14356321e-01 -5.54698050e-01 -4.25177574e-01
3.71733785e-01 1.50984265e-02 3.52528840e-01 -1.47801623e-01
5.34987450e-01 -1.83631510e-01 -1.22259474e+00 8.24442446e-01
1.29309416e+00 7.22929537e-01 6.64192200e-01 6.53796136e-01
-2.97735333e-01 7.72549510e-01 -5.60133994e-01 -5.45630991e-01
-2.25249320e-01 -5.16065598e-01 -5.10013998e-01 -1.41692340e-01
2.73442596e-01 -6.70220181e-02 -1.29597515e-01 1.29917693e+00
7.62541592e-01 1.51594039e-02 3.78005028e-01 -6.82511449e-01
-1.40415442e+00 7.45291531e-01 -4.11836386e-01 1.51383653e-01
5.90351164e-01 1.11163810e-01 1.18147290e+00 -1.30024171e+00
4.61143404e-01 9.72420871e-01 6.67200744e-01 7.21911490e-01
-7.34656870e-01 -7.18999386e-01 5.69502473e-01 4.02241945e-01
-1.49423230e+00 -3.55756342e-01 5.31783640e-01 9.12718382e-03
1.00228274e+00 2.13437796e-01 5.26335180e-01 1.28333378e+00
-1.02062270e-01 1.44383454e+00 1.04734957e+00 -7.14847624e-01
-4.57239710e-03 -1.50990173e-01 -8.95522237e-02 6.73600137e-01
-2.69457176e-02 -2.08603919e-01 -7.75419116e-01 1.54202044e-01
4.07521218e-01 3.20824496e-02 -4.28799778e-01 1.63559303e-01
-1.25619018e+00 4.70734179e-01 6.67055845e-02 3.59122872e-01
3.20340209e-02 2.63506353e-01 4.31318820e-01 1.49629861e-01
1.84382871e-01 3.27469826e-01 -4.91929770e-01 -6.38904452e-01
-9.15771663e-01 -3.99155021e-02 4.01147634e-01 1.37242079e+00
4.29661542e-01 1.64877791e-02 -2.82299072e-01 9.02734339e-01
3.92549723e-01 6.08914673e-01 8.15945685e-01 -2.46863350e-01
8.22266698e-01 4.55752850e-01 -2.54845083e-01 -6.50638700e-01
-1.43514305e-01 -4.26676512e-01 -5.57608664e-01 -3.98631305e-01
1.12332806e-01 -1.24156646e-01 -9.36023951e-01 1.47850156e+00
-2.82763302e-01 5.07174432e-01 -8.48325193e-02 8.48141730e-01
9.42084670e-01 7.02412546e-01 1.91543579e-01 1.70349687e-01
1.25904238e+00 -1.18826890e+00 -7.79772818e-01 -5.34899592e-01
6.94297910e-01 -1.08092558e+00 1.68654644e+00 4.20745671e-01
-1.18994606e+00 -5.52960336e-01 -9.82051611e-01 -2.53204048e-01
-3.14742506e-01 6.18784726e-01 3.28743100e-01 8.22579741e-01
-7.66268492e-01 3.70080411e-01 -9.69477177e-01 -6.67271465e-02
2.36713544e-01 -1.93798006e-01 -9.54150036e-02 -1.17897660e-01
-1.15946913e+00 5.30858099e-01 3.94037694e-01 1.76160946e-01
-7.39600003e-01 -5.00257015e-01 -9.01792526e-01 2.29229718e-01
2.90947080e-01 -2.11985096e-01 1.22422040e+00 -9.25079703e-01
-1.63407862e+00 8.44634950e-01 -4.62708920e-01 -1.20618328e-01
2.77612835e-01 -6.03180289e-01 -8.64994287e-01 2.07798272e-01
-4.29436378e-02 5.10257185e-01 7.06837952e-01 -9.72476900e-01
-6.98468983e-01 1.22436807e-02 -2.90676445e-01 2.80562818e-01
-6.01495743e-01 5.42105138e-01 -1.31833315e+00 -1.20379043e+00
-1.66223533e-02 -7.45446384e-01 -9.63352248e-02 -2.17377126e-01
-6.74156129e-01 6.13472983e-03 7.05668032e-01 -7.34240413e-01
1.79328430e+00 -2.48118091e+00 -1.55366749e-01 2.79840846e-02
-1.59231097e-01 5.65928996e-01 -4.94592696e-01 4.78934795e-01
2.81802028e-01 3.54924351e-01 -1.98836997e-01 -8.13235044e-01
1.73027106e-02 -1.71353146e-01 -4.79906797e-01 2.67982930e-01
4.05212522e-01 9.66898322e-01 -9.07541096e-01 -6.37480617e-01
4.49673878e-03 2.67789632e-01 -8.06613982e-01 1.18711524e-01
-2.41047829e-01 2.61806518e-01 -3.81886989e-01 1.04100418e+00
7.85651386e-01 -5.22364676e-01 1.85641691e-01 2.60994524e-01
-2.18417689e-01 7.55754232e-01 -9.91251647e-01 2.05546570e+00
-4.13205504e-01 4.64636534e-01 -4.31503743e-01 -5.59144974e-01
8.03350925e-01 3.14465344e-01 -2.98583686e-01 -1.06134045e+00
-2.17286777e-02 4.06733811e-01 -6.53090924e-02 -3.64387304e-01
7.03853846e-01 2.63513058e-01 -2.48725981e-01 3.22543740e-01
-1.51960673e-02 -1.38748484e-02 2.45933861e-01 3.73411149e-01
9.65677559e-01 3.40040743e-01 1.75309837e-01 -9.34780613e-02
7.51382470e-01 -2.87227333e-01 8.28273177e-01 8.25936496e-01
-4.04848725e-01 1.18582451e+00 3.66513073e-01 -9.64283347e-02
-7.02945113e-01 -1.16306090e+00 1.35553688e-01 1.30254149e+00
1.65242419e-01 -9.27771986e-01 -7.24903524e-01 -6.18882000e-01
-4.11289692e-01 4.75027502e-01 -4.67395753e-01 -1.30650967e-01
-6.36935711e-01 -5.34103811e-01 8.69873464e-01 9.19491231e-01
5.98510027e-01 -1.28664911e+00 -3.21250528e-01 2.16854453e-01
-2.99043864e-01 -1.40711415e+00 -8.27453017e-01 2.85847355e-02
-6.19478285e-01 -8.26919913e-01 -8.37665081e-01 -1.11604631e+00
4.87762719e-01 3.55108202e-01 1.10867822e+00 5.07969677e-01
-4.47728932e-02 -8.28049779e-02 -9.36057866e-01 -5.19651473e-01
-1.74787581e-01 2.88524657e-01 -3.21019769e-01 -9.09507349e-02
6.51295602e-01 -2.12746844e-01 -4.62603480e-01 1.25722751e-01
-6.88323557e-01 1.98239118e-01 6.66899204e-01 7.43683994e-01
7.13158965e-01 -4.71046060e-01 3.70808929e-01 -9.87633348e-01
5.34553766e-01 -1.67858392e-01 -5.53266585e-01 4.51018810e-01
-2.37047002e-01 -3.47343162e-02 8.58573675e-01 -5.15876412e-01
-1.14018071e+00 6.56967089e-02 -2.36180276e-01 -2.11075902e-01
-1.05239287e-01 9.24852490e-01 -3.97787511e-01 1.20323069e-01
3.20119411e-01 7.87867248e-01 -5.88381290e-01 -6.53743804e-01
2.64481753e-01 8.00739050e-01 8.16760540e-01 -6.59049869e-01
6.43537164e-01 3.47972959e-01 -6.99593842e-01 -7.87647128e-01
-8.53338420e-01 -4.59124327e-01 -3.55178446e-01 5.52047975e-02
8.69899333e-01 -1.30896330e+00 -4.80690539e-01 7.61621594e-01
-1.05513120e+00 -2.89008647e-01 4.05965865e-01 4.75292176e-01
-3.63384873e-01 7.38384008e-01 -9.30137277e-01 -6.39616847e-01
-3.39830279e-01 -1.06898212e+00 1.19071352e+00 2.62219429e-01
-8.02881122e-02 -8.55949461e-01 -8.20226874e-03 6.78613111e-02
3.41269404e-01 -3.52067828e-01 9.06132340e-01 -5.42936921e-01
-6.60459042e-01 -1.74943924e-01 -3.97521526e-01 2.53940940e-01
-9.00998414e-02 1.77343592e-01 -9.72327411e-01 -1.92242652e-01
-5.83499670e-01 -4.69139546e-01 1.18720293e+00 8.93550441e-02
1.61520278e+00 1.37259647e-01 -1.95125222e-01 9.16809976e-01
1.37934959e+00 8.68256614e-02 7.34352767e-01 3.68038148e-01
7.78709173e-01 -2.73549482e-02 4.97763753e-01 7.92800725e-01
4.74580079e-01 3.88076991e-01 3.92213017e-01 1.77470431e-01
-1.72390670e-01 -6.87187076e-01 5.06085098e-01 1.32986724e+00
1.95424538e-02 -4.83551055e-01 -9.71696019e-01 6.15963995e-01
-1.63799381e+00 -8.37074995e-01 -9.42522660e-02 1.88709462e+00
1.04810131e+00 3.97315443e-01 -3.35690945e-01 5.85814677e-02
7.21788943e-01 2.95668989e-01 -1.14340201e-01 -3.31908733e-01
-6.11047268e-01 2.96022892e-01 2.52694637e-01 8.22949708e-02
-1.33964980e+00 1.47853863e+00 6.43766260e+00 1.08449805e+00
-1.15745354e+00 4.73167142e-03 3.95873576e-01 8.77336040e-02
-5.71463704e-01 1.31453767e-01 -1.06427336e+00 8.05779099e-01
6.77815378e-01 1.28178880e-01 3.24826151e-01 5.70783436e-01
1.11197844e-01 9.36486349e-02 -8.59072089e-01 8.28626335e-01
2.65991330e-01 -1.46676743e+00 1.13041878e-01 -4.33941364e-01
8.93504441e-01 2.45875999e-01 3.03454816e-01 4.82618541e-01
3.09745759e-01 -9.12790954e-01 1.16177380e+00 4.94716913e-01
8.93226206e-01 -6.99367285e-01 4.21006739e-01 2.76962847e-01
-1.50565910e+00 1.19859762e-01 -3.39485466e-01 -7.06770197e-02
2.78206527e-01 1.59618512e-01 -7.87771568e-02 3.31712097e-01
6.21522546e-01 1.09522605e+00 -9.59764123e-01 1.47170568e+00
-7.79953957e-01 1.00717449e+00 5.26262522e-02 -4.35840130e-01
4.18137908e-01 1.38744503e-01 1.20491989e-01 1.64267886e+00
5.50554752e-01 -8.56317356e-02 1.34706423e-01 8.25882733e-01
-1.55698910e-01 1.72635257e-01 -1.04959458e-01 -4.56524730e-01
8.27486873e-01 9.89654541e-01 -4.68156576e-01 -3.01622540e-01
-1.04174876e+00 1.21283746e+00 5.19780934e-01 4.04669493e-01
-8.94053578e-01 -6.94144785e-01 5.21366596e-01 -5.30462086e-01
6.84618592e-01 -2.47246757e-01 -4.06307667e-01 -1.64200950e+00
1.53146490e-01 -1.18953013e+00 3.59343618e-01 -8.72160792e-01
-1.13510716e+00 5.76361358e-01 -7.37156153e-01 -1.67755508e+00
5.03563061e-02 -6.31630659e-01 -9.50056434e-01 1.04844654e+00
-1.84345460e+00 -1.24535382e+00 -1.10587753e-01 6.41572356e-01
8.13684046e-01 -3.00556451e-01 8.75772476e-01 4.12859499e-01
-7.36768007e-01 1.04727161e+00 1.43848196e-01 4.96704936e-01
9.27622736e-01 -1.19820476e+00 8.86013150e-01 1.47340357e+00
3.33565384e-01 8.47158909e-01 3.09762657e-01 -6.85370147e-01
-1.21355772e+00 -1.09992635e+00 1.37187231e+00 -3.59506398e-01
6.61150157e-01 -5.68517327e-01 -1.05063176e+00 5.32798886e-01
3.43565792e-01 2.57677197e-01 8.91701519e-01 1.44663930e-01
-5.66475332e-01 4.15635675e-01 -2.73397118e-01 8.32494438e-01
1.02836668e+00 -9.50995803e-01 -6.14369452e-01 -3.61281559e-02
7.90263355e-01 -5.25809646e-01 -1.25738978e-01 8.25489089e-02
3.92685115e-01 -8.59655440e-01 5.87582350e-01 -7.29682744e-01
7.47253954e-01 -3.13189983e-01 -2.04326242e-01 -1.36982548e+00
-3.39472234e-01 -7.68620968e-01 3.07688303e-02 1.21466589e+00
7.52991796e-01 2.15908308e-02 6.97856128e-01 7.79576004e-02
-7.14387298e-01 -4.70255107e-01 -6.74373507e-01 -7.69847453e-01
1.29697219e-01 -7.14111149e-01 6.12193823e-01 9.93737757e-01
3.57063174e-01 1.14313895e-02 -4.99114513e-01 2.94327706e-01
9.13515165e-02 2.19349176e-01 4.19200599e-01 -8.65780890e-01
-4.43197876e-01 -6.69516444e-01 3.67480963e-02 -1.37249935e+00
1.80017978e-01 -8.66790056e-01 2.50172794e-01 -1.25548935e+00
3.24622244e-01 -4.46671158e-01 -3.81131798e-01 2.58413255e-01
-8.13123941e-01 2.44060516e-01 5.12494683e-01 9.69743505e-02
-1.15395248e+00 6.75894260e-01 1.21041250e+00 3.43707800e-02
-6.51924114e-04 4.60055145e-03 -7.24924922e-01 8.40660810e-01
5.94451308e-01 -2.16913000e-01 4.74869534e-02 -1.17398286e+00
6.99463308e-01 -2.79630899e-01 2.59959400e-02 -1.09141624e+00
4.27417368e-01 -4.47925553e-02 3.37306589e-01 -7.97317147e-01
2.09712595e-01 -4.00018096e-01 -7.48398006e-01 3.56130660e-01
-5.26629150e-01 5.43298066e-01 2.25250468e-01 5.87991536e-01
-5.71152449e-01 -3.08316767e-01 5.35314441e-01 -7.98608661e-02
-1.05270815e+00 2.09768862e-01 -7.06451178e-01 4.10188615e-01
4.90050554e-01 1.77819319e-02 -4.13450122e-01 -3.31992835e-01
-3.41604620e-01 3.11337948e-01 3.36732090e-01 6.07142985e-01
8.36339116e-01 -1.43660796e+00 -6.55772805e-01 4.45691437e-01
5.57597041e-01 -3.45191419e-01 1.64961398e-01 7.17289627e-01
-8.29425156e-01 4.81095761e-01 -6.18634261e-02 -2.48628378e-01
-1.20850253e+00 5.04281759e-01 9.16137695e-02 -1.50872722e-01
-4.45530951e-01 1.17045033e+00 -3.84144820e-02 -4.31646496e-01
4.55199420e-01 -5.19870460e-01 -1.73897147e-01 -1.61334112e-01
9.02502120e-01 1.84995681e-02 1.60341755e-01 -4.54354048e-01
-5.39474964e-01 5.06102979e-01 -2.79815882e-01 -2.24007353e-01
9.61668253e-01 -1.79574221e-01 2.41866916e-01 2.78093517e-01
9.32569861e-01 4.63742763e-01 -1.22160161e+00 -6.34946465e-01
7.80828819e-02 -6.62469506e-01 -1.01965711e-01 -8.10562074e-01
-9.42485750e-01 1.27606142e+00 1.49985418e-01 -3.66201460e-01
1.19051135e+00 -2.36695692e-01 9.60918367e-01 5.55390656e-01
1.81998238e-01 -1.51712036e+00 2.31333509e-01 1.06010449e+00
4.52495009e-01 -1.28486323e+00 -4.29200053e-01 -4.97387588e-01
-9.51862216e-01 1.09159410e+00 8.14015508e-01 -1.61809102e-01
5.28380454e-01 3.65802705e-01 3.20909530e-01 3.99128973e-01
-9.17469263e-01 -3.99786144e-01 3.48682612e-01 5.28106987e-01
8.41680348e-01 -6.31306767e-02 -3.90869707e-01 1.04077923e+00
-6.08490370e-02 -4.08142917e-02 5.35412610e-01 1.12920964e+00
-4.40576136e-01 -1.23828804e+00 -5.25704697e-02 1.34051172e-02
-7.64571071e-01 -9.35308695e-01 -3.41072053e-01 5.34355938e-01
6.70117908e-04 8.63531530e-01 1.72408551e-01 -3.30527216e-01
1.00921743e-01 -1.71178460e-01 2.15866104e-01 -7.86708236e-01
-7.83929646e-01 3.88424188e-01 -8.59822035e-02 -5.61840832e-01
-1.99251115e-01 -8.28014076e-01 -1.31765532e+00 7.99815636e-03
-2.69544691e-01 -1.45096213e-01 3.57884258e-01 9.89084899e-01
4.49105889e-01 4.54292476e-01 3.61430645e-01 -3.78049225e-01
-4.80686456e-01 -8.84287357e-01 -3.53517681e-01 4.84565377e-01
1.57495141e-01 -7.07354322e-02 -1.13251116e-02 6.25497028e-02] | [10.9162015914917, 10.810404777526855] |
0b4255b3-40d5-440a-97de-3d63093edd97 | cloud-removal-in-satellite-images-using | 1912.06838 | null | https://arxiv.org/abs/1912.06838v1 | https://arxiv.org/pdf/1912.06838v1.pdf | Cloud Removal in Satellite Images Using Spatiotemporal Generative Networks | Satellite images hold great promise for continuous environmental monitoring and earth observation. Occlusions cast by clouds, however, can severely limit coverage, making ground information extraction more difficult. Existing pipelines typically perform cloud removal with simple temporal composites and hand-crafted filters. In contrast, we cast the problem of cloud removal as a conditional image synthesis challenge, and we propose a trainable spatiotemporal generator network (STGAN) to remove clouds. We train our model on a new large-scale spatiotemporal dataset that we construct, containing 97640 image pairs covering all continents. We demonstrate experimentally that the proposed STGAN model outperforms standard models and can generate realistic cloud-free images with high PSNR and SSIM values across a variety of atmospheric conditions, leading to improved performance in downstream tasks such as land cover classification. | ['Vishnu Sarukkai', 'Burak Uzkent', 'Anirudh Jain', 'Stefano Ermon'] | 2019-12-14 | null | null | null | null | ['cloud-removal'] | ['computer-vision'] | [ 2.89698362e-01 -5.48607230e-01 1.06351666e-01 -2.60980099e-01
-8.79068613e-01 -9.33237135e-01 7.62341440e-01 -1.99343145e-01
-3.01528186e-01 8.11895847e-01 1.61589421e-02 -4.99135703e-01
3.20861578e-01 -1.10167670e+00 -8.38611186e-01 -8.60719264e-01
-3.57073635e-01 4.02703732e-02 1.27477109e-01 -2.03055978e-01
-1.87939554e-01 5.42217791e-01 -1.58304393e+00 9.66404453e-02
1.36304712e+00 6.82721019e-01 5.20733714e-01 7.95805216e-01
1.68084174e-01 4.90373313e-01 -5.42121053e-01 -5.33878766e-02
7.13123739e-01 -3.80836904e-01 -3.89967829e-01 2.57593274e-01
6.79645300e-01 -4.21123177e-01 -4.48138207e-01 1.19115520e+00
4.35971946e-01 -9.64494422e-04 3.18648785e-01 -1.06106448e+00
-5.41452706e-01 6.21312149e-02 -6.27719939e-01 1.86564833e-01
-3.11986148e-01 5.49277604e-01 7.61723757e-01 -8.18934500e-01
5.79061747e-01 9.18180287e-01 7.11687267e-01 4.85378429e-02
-1.29984653e+00 -7.97234416e-01 2.92899042e-01 -2.35091597e-01
-1.60083652e+00 -3.30222994e-01 9.82605442e-02 -4.84922200e-01
8.59875143e-01 6.33662462e-01 8.81425381e-01 5.83525181e-01
2.49047447e-02 5.63879788e-01 1.29503655e+00 -7.27053508e-02
8.21131840e-02 -2.04268321e-01 -4.06233519e-01 1.93594009e-01
5.43987989e-01 2.90956378e-01 -1.64313018e-01 5.99176846e-02
8.12162817e-01 2.52486914e-01 -4.91228491e-01 2.15741903e-01
-1.19818592e+00 6.32820785e-01 7.97684848e-01 -1.98341623e-01
-5.04445732e-01 3.99747670e-01 -1.36817887e-01 3.75604600e-01
9.85987186e-01 3.32371950e-01 -2.64377326e-01 4.96331722e-01
-1.36804175e+00 7.60943830e-01 3.18715453e-01 1.27519023e+00
7.50708997e-01 5.27791858e-01 -2.44584277e-01 4.68095154e-01
1.12553924e-01 1.38785577e+00 -2.11788207e-01 -9.78070974e-01
5.02720237e-01 1.28081992e-01 5.87354720e-01 -9.19918299e-01
-2.04709042e-02 -8.16948891e-01 -9.76291895e-01 3.15549791e-01
-4.14889157e-02 -3.29595596e-01 -1.39065886e+00 1.41371584e+00
3.46385211e-01 6.34623408e-01 9.94947702e-02 1.08697093e+00
6.20806754e-01 9.47689414e-01 2.92432066e-02 1.95582863e-02
1.26611233e+00 -8.51715207e-01 -6.33243680e-01 -6.64092481e-01
3.15629661e-01 -5.36039650e-01 7.70969748e-01 1.00181416e-01
-7.01784611e-01 -2.58163512e-01 -8.50595057e-01 2.62582958e-01
-4.26556140e-01 4.20908928e-02 7.80698419e-01 2.62066066e-01
-1.21320879e+00 4.91110623e-01 -9.46264029e-01 -2.20754132e-01
4.50486541e-01 1.11329872e-02 -9.90016609e-02 -2.62080520e-01
-1.13044512e+00 7.05543280e-01 1.04521811e-01 6.23009741e-01
-1.15007126e+00 -8.29842925e-01 -8.83960783e-01 1.48091853e-01
2.71190424e-02 -7.25307107e-01 8.22986543e-01 -9.39910054e-01
-9.12887752e-01 9.19176638e-01 -2.51127124e-01 -7.68927872e-01
5.63756764e-01 -3.41669083e-01 -6.13793671e-01 1.30614311e-01
3.22765440e-01 7.97250152e-01 9.44564998e-01 -1.39891768e+00
-8.03273261e-01 1.01190601e-02 -1.27749026e-01 2.74405897e-01
-1.08452410e-01 1.25376239e-01 -6.93506837e-01 -9.91620958e-01
1.19460270e-01 -1.29286361e+00 -6.07903242e-01 1.14573322e-01
-3.16379189e-01 6.89240873e-01 8.21951032e-01 -8.68561804e-01
8.98929477e-01 -2.05551744e+00 -2.05254942e-01 5.44292890e-02
1.55657947e-01 3.83677483e-01 -5.23580194e-01 2.79721588e-01
-3.99309248e-02 4.63375717e-01 -7.13423729e-01 -4.13979411e-01
-3.31502080e-01 3.37210149e-01 -7.50404179e-01 5.58045506e-01
6.01889670e-01 9.48452413e-01 -9.72115338e-01 -2.14336202e-01
1.96004897e-01 4.68452543e-01 -4.60243076e-01 3.62854242e-01
-5.89710414e-01 7.99643040e-01 -2.82709748e-01 6.69515371e-01
1.37200725e+00 -2.36505151e-01 1.25378177e-01 4.16049570e-01
-3.27643156e-01 2.11300880e-01 -8.50589871e-01 1.33839524e+00
-3.79858166e-01 8.85670066e-01 3.66886407e-01 -2.88720131e-01
6.91967547e-01 8.69164616e-02 2.75655299e-01 -8.37715626e-01
-3.40019315e-01 2.24585593e-01 -1.48589849e-01 -5.08420110e-01
6.16745949e-01 -7.09152073e-02 1.15957946e-01 1.21105194e-01
-4.94482666e-01 -7.13109493e-01 -1.73946336e-01 1.32395238e-01
8.00609052e-01 2.65106201e-01 -3.59908670e-01 -3.36023659e-01
-7.79777467e-02 4.70016658e-01 5.69709599e-01 7.75287032e-01
-1.73985269e-02 1.09696794e+00 1.08541891e-01 -6.43928409e-01
-1.20457304e+00 -1.00534844e+00 -1.03182584e-01 5.55680573e-01
6.82254806e-02 -1.43823862e-01 -3.59532923e-01 -2.17782974e-01
2.02458516e-01 4.16665524e-01 -2.63588846e-01 1.79059356e-01
-5.53336918e-01 -1.33328724e+00 5.59101880e-01 2.34130651e-01
7.86836684e-01 -8.21728528e-01 -3.37894350e-01 1.18805379e-01
-5.27781725e-01 -1.50306630e+00 -1.75703198e-01 -8.48063529e-02
-1.00121868e+00 -6.49806857e-01 -6.90833211e-01 -3.92292261e-01
5.81978440e-01 8.72052431e-01 1.47730148e+00 2.88536936e-01
-3.04765105e-01 -2.93401450e-01 -2.06107885e-01 -5.74580789e-01
-1.32143246e-02 1.10284232e-01 -2.27836654e-01 -1.13716982e-02
-3.46739143e-02 -7.07667053e-01 -9.77338791e-01 4.91477661e-02
-1.09612632e+00 2.18902349e-01 4.47235107e-01 6.52877033e-01
7.02077806e-01 1.66474536e-01 1.22543061e-02 -7.36817360e-01
1.26270158e-02 -6.13759100e-01 -1.27416539e+00 -1.25020787e-01
-3.84475201e-01 -1.94558024e-01 2.18652517e-01 1.35599300e-02
-1.13174725e+00 3.33898723e-01 -4.12246846e-02 -6.19011104e-01
-2.16627978e-02 6.40492916e-01 -4.44237739e-02 -2.46490672e-01
5.62773168e-01 6.71596646e-01 -3.46631825e-01 -3.25153619e-01
3.01892489e-01 4.80521500e-01 8.36238742e-01 -2.88654208e-01
1.46701503e+00 9.36106920e-01 -7.65316710e-02 -9.99713540e-01
-7.65337527e-01 -3.91688347e-01 -4.15539205e-01 -6.01117536e-02
7.84452796e-01 -1.82667387e+00 -1.66021571e-01 6.65065229e-01
-9.79175806e-01 -7.02688158e-01 6.54446706e-02 2.75938064e-01
8.05908218e-02 1.31268367e-01 -2.21456662e-01 -9.97839451e-01
-5.21255612e-01 -7.67317951e-01 1.36061680e+00 1.04611525e-02
3.64939570e-01 -7.29640365e-01 -7.91937634e-02 3.08360178e-02
7.80410469e-01 7.63342798e-01 3.05587471e-01 3.85607958e-01
-1.41503429e+00 -1.95092529e-01 -5.22278845e-01 3.13408941e-01
1.74974903e-01 1.88899502e-01 -1.26024199e+00 -5.96396267e-01
-2.34944329e-01 2.57579517e-02 1.47806931e+00 4.74504411e-01
1.17912352e+00 -4.35686380e-01 -3.30000937e-01 1.30555260e+00
1.73664165e+00 -2.17271313e-01 1.01233733e+00 4.73726660e-01
7.57514596e-01 3.53240728e-01 7.00347126e-01 3.48926455e-01
3.65595549e-01 4.05992866e-01 1.01505578e+00 -6.95527375e-01
-2.12371022e-01 2.64771134e-02 5.95749617e-02 3.70566368e-01
-3.27018738e-01 -5.09318948e-01 -1.08114421e+00 1.03892183e+00
-1.74332762e+00 -9.74007010e-01 -5.10053396e-01 2.22112393e+00
6.51794076e-01 -7.35326484e-02 -3.28830093e-01 -5.19606113e-01
4.94329095e-01 6.69495702e-01 -4.52051342e-01 4.58244175e-01
-6.23062909e-01 3.68146032e-01 1.26582599e+00 6.10427499e-01
-1.45247817e+00 1.34949803e+00 6.91113949e+00 2.89111048e-01
-1.54222906e+00 1.74808100e-01 4.91613179e-01 -2.63005018e-01
-5.51335037e-01 2.07677349e-01 -6.11802936e-01 5.67668676e-01
9.77386951e-01 2.07060918e-01 3.07982564e-01 5.69293082e-01
7.31612682e-01 -1.17835306e-01 -1.81008428e-01 7.39981651e-01
-3.98093700e-01 -1.47037220e+00 -1.42750302e-02 1.96149051e-01
1.15179861e+00 9.04006422e-01 1.05520085e-01 -1.88667029e-02
7.79858887e-01 -1.14502370e+00 7.72142470e-01 3.72340500e-01
1.05685890e+00 -3.70129019e-01 4.09043014e-01 1.72463223e-01
-1.31612051e+00 2.63507634e-01 -3.95175785e-01 -1.85338795e-01
-4.91395853e-02 8.62268686e-01 -5.95142245e-01 6.97478592e-01
8.30695331e-01 7.57362425e-01 -4.56580639e-01 1.40293646e+00
-4.93819535e-01 9.08593178e-01 -6.28373444e-01 6.47416890e-01
3.76095504e-01 -3.72365952e-01 5.86795390e-01 1.31647599e+00
6.12621546e-01 3.45278770e-01 1.11128017e-01 8.23343337e-01
-1.58559918e-01 -3.52416903e-01 -8.19443703e-01 8.56620520e-02
6.19458497e-01 1.24207389e+00 -5.28053939e-01 -4.37627524e-01
-3.58741611e-01 1.14055872e+00 -1.19341157e-01 6.32265627e-01
-9.46469247e-01 -2.47806221e-01 1.40143347e+00 1.42241418e-01
5.27956069e-01 -5.10715902e-01 -2.48295322e-01 -1.65801167e+00
9.33168232e-02 -8.81424963e-01 -4.86678593e-02 -9.90940809e-01
-7.94249058e-01 5.61321497e-01 -1.43841386e-01 -1.70212579e+00
7.00117797e-02 -3.03734541e-01 -8.17389190e-01 1.28777564e+00
-2.30436826e+00 -1.48673093e+00 -9.36805367e-01 2.69192964e-01
2.75442451e-01 3.59158278e-01 6.61076963e-01 3.77457291e-01
-3.87677103e-01 -9.50438753e-02 3.84805143e-01 5.97629622e-02
6.18126929e-01 -1.14631820e+00 1.29505825e+00 1.52952278e+00
-4.63501457e-03 3.85065079e-01 8.53366613e-01 -7.36955047e-01
-1.43904018e+00 -2.12789917e+00 7.98881888e-01 -2.61990100e-01
3.87990683e-01 -5.48669338e-01 -9.99998868e-01 8.49439561e-01
2.58771837e-01 6.34733021e-01 6.81945682e-02 -3.39372247e-01
-3.94817084e-01 -3.44759315e-01 -8.65148664e-01 4.74136919e-01
1.04825842e+00 -4.94549364e-01 2.20713019e-01 8.05046082e-01
8.49786997e-01 -6.12929046e-01 -6.46589160e-01 4.57435280e-01
4.53298539e-01 -7.75702596e-01 1.01564384e+00 -5.30523837e-01
6.01111948e-01 -7.40629017e-01 -1.81822091e-01 -1.53140295e+00
-5.50322354e-01 -4.81580466e-01 3.55443686e-01 9.74126339e-01
4.36962694e-01 -6.24794006e-01 6.71672702e-01 4.23691064e-01
-2.50223041e-01 8.65206271e-02 -5.58285356e-01 -1.04019415e+00
1.53902382e-01 -3.43597919e-01 1.00482380e+00 1.19348598e+00
-9.32960987e-01 -2.02048719e-01 -5.77810526e-01 1.06017685e+00
7.29036391e-01 6.53112769e-01 9.16182995e-01 -1.04941928e+00
-1.06931262e-01 -1.28144369e-01 -7.57314041e-02 -8.06821465e-01
-1.24557532e-01 -6.03850603e-01 2.74974972e-01 -1.45093250e+00
5.18898033e-02 -7.54553497e-01 -6.91513938e-04 6.63097143e-01
-5.09412587e-01 6.36129677e-01 1.33670270e-01 4.49456394e-01
-7.89232627e-02 5.58415115e-01 1.29876852e+00 -3.44787151e-01
1.99509994e-03 -1.58456147e-01 -3.80475283e-01 3.75065118e-01
7.34276474e-01 -6.68297172e-01 -1.36823148e-01 -1.27286208e+00
3.26377451e-01 -2.96429321e-02 6.99615717e-01 -1.14642203e+00
-1.23934053e-01 -7.02144563e-01 4.88826036e-01 -7.49695480e-01
2.29092777e-01 -6.59473121e-01 6.40359282e-01 4.66677576e-01
1.28627867e-01 -1.80632249e-02 4.65352148e-01 4.34038222e-01
-3.96752626e-01 4.14690316e-01 7.68570423e-01 -3.38956714e-01
-7.71051526e-01 7.51957715e-01 -2.76579440e-01 -2.54362971e-01
7.37102628e-01 2.49879524e-01 -3.70351106e-01 -5.20672262e-01
-4.96315837e-01 5.48574090e-01 8.28940809e-01 3.29002351e-01
3.91094536e-01 -1.05036175e+00 -1.21814966e+00 3.10055912e-01
3.35018426e-01 4.20574754e-01 1.84009016e-01 5.43475091e-01
-8.69352996e-01 2.60367334e-01 1.75973549e-01 -7.50102758e-01
-1.03682959e+00 2.44878456e-01 5.29231906e-01 -1.33353084e-01
-8.65133703e-01 8.30195367e-01 2.74912715e-01 -5.51241159e-01
-3.81238550e-01 -5.41568995e-01 2.95306981e-01 -1.76853433e-01
5.20342350e-01 -3.08430552e-01 3.72365683e-01 -5.20600200e-01
-1.74498901e-01 3.10997605e-01 4.98518288e-01 -1.02671847e-01
1.61206806e+00 -1.43389657e-01 -1.69281900e-01 -1.19200602e-01
7.95801997e-01 6.11586357e-03 -1.45394850e+00 -3.82489234e-01
-2.89071977e-01 -9.60164487e-01 3.06846678e-01 -6.45717084e-01
-1.52531230e+00 8.43860567e-01 6.38997614e-01 1.59855038e-02
1.26957345e+00 -5.20890951e-01 7.82675207e-01 3.08624506e-01
2.44467676e-01 -5.59311450e-01 -5.14573574e-01 5.55492282e-01
8.67283642e-01 -1.55667007e+00 6.83652163e-02 -6.11244380e-01
-5.12510896e-01 5.92706919e-01 3.69359344e-01 -2.08176270e-01
5.26928067e-01 4.95393634e-01 2.29080036e-01 -1.47106349e-01
-8.26019466e-01 -6.85887754e-01 -4.87007871e-02 6.34489119e-01
1.42614082e-01 3.94691139e-01 2.72527963e-01 -3.94092612e-02
-9.55283195e-02 1.53153077e-01 5.64315438e-01 1.03738368e+00
-4.29923385e-01 -7.33059525e-01 -7.09149301e-01 4.80419785e-01
-2.80694246e-01 -6.07281268e-01 7.49117509e-02 5.49775839e-01
1.63017046e-02 7.69515693e-01 3.32122743e-01 -1.08095475e-01
-2.98613328e-02 -3.86722952e-01 7.89598972e-02 -6.93384409e-01
-3.65325660e-01 2.59477317e-01 2.40855485e-01 -5.06362915e-01
-4.91736621e-01 -8.05305302e-01 -7.71561921e-01 -4.35704917e-01
-3.49780321e-01 -4.40227687e-02 7.27834940e-01 6.94454014e-01
7.31239140e-01 4.17629421e-01 6.93252027e-01 -1.13156521e+00
-2.52296209e-01 -1.02068567e+00 -6.15583658e-01 2.99940884e-01
9.26096439e-01 -3.68939131e-01 -2.35549837e-01 3.13629091e-01] | [9.725188255310059, -1.673419713973999] |
994a88c9-0c21-4afd-8312-8b8a5f37577d | tido-source-free-task-incremental-learning-in | 2301.12055 | null | https://arxiv.org/abs/2301.12055v1 | https://arxiv.org/pdf/2301.12055v1.pdf | TIDo: Source-free Task Incremental Learning in Non-stationary Environments | This work presents an incremental learning approach for autonomous agents to learn new tasks in a non-stationary environment. Updating a DNN model-based agent to learn new target tasks requires us to store past training data and needs a large labeled target task dataset. Few-shot task incremental learning methods overcome the limitation of labeled target datasets by adapting trained models to learn private target classes using a few labeled representatives and a large unlabeled target dataset. However, the methods assume that the source and target tasks are stationary. We propose a one-shot task incremental learning approach that can adapt to non-stationary source and target tasks. Our approach minimizes adversarial discrepancy between the model's feature space and incoming incremental data to learn an updated hypothesis. We also use distillation loss to reduce catastrophic forgetting of previously learned tasks. Finally, we use Gaussian prototypes to generate exemplar instances eliminating the need to store past training data. Unlike current work in task incremental learning, our model can learn both source and target task updates incrementally. We evaluate our method on various problem settings for incremental object detection and disease prediction model update. We evaluate our approach by measuring the performance of shared class and target private class prediction. Our results show that our approach achieved improved performance compared to existing state-of-the-art task incremental learning methods. | ['Leong Tze Yun', 'Abhinit Kumar Ambastha'] | 2023-01-28 | null | null | null | null | ['disease-prediction'] | ['medical'] | [ 4.89294142e-01 1.12918161e-01 -3.81635725e-01 -5.69063485e-01
-7.28586972e-01 -3.46867770e-01 8.41817319e-01 -9.15337205e-02
-9.27843153e-01 1.40447831e+00 -1.30635545e-01 2.33505294e-01
7.74458647e-02 -6.05332077e-01 -1.00863147e+00 -6.10853970e-01
-2.10900202e-01 1.04375875e+00 7.61809409e-01 1.09055683e-01
3.95687344e-03 2.74809122e-01 -1.65676677e+00 4.44771260e-01
8.99339437e-01 8.04316521e-01 6.33077502e-01 9.07102525e-01
-1.82962716e-01 9.79298413e-01 -9.85711277e-01 -2.64852494e-01
4.34097111e-01 -1.92377552e-01 -6.61216259e-01 -7.64074549e-02
4.52109635e-01 -5.30608654e-01 -3.37988436e-01 8.17297280e-01
7.79769897e-01 4.48727667e-01 8.34599435e-01 -1.55696738e+00
-8.09275448e-01 4.00613427e-01 -7.17592239e-02 5.44117868e-01
-1.26464829e-01 2.28674591e-01 2.03907311e-01 -1.09397256e+00
8.08379889e-01 1.14882362e+00 9.60018814e-01 1.14393055e+00
-1.20212591e+00 -7.81555057e-01 5.78008294e-01 6.76591516e-01
-8.77695739e-01 -6.33858919e-01 6.37547255e-01 -3.10058117e-01
1.17314208e+00 -3.61466080e-01 6.48979247e-01 1.74728417e+00
4.48759407e-01 8.85255992e-01 9.77959692e-01 -3.26107979e-01
8.09436917e-01 2.70056218e-01 2.06347317e-01 4.54866171e-01
2.50072241e-01 3.39226604e-01 -5.14075816e-01 -4.15708661e-01
3.62608194e-01 6.22206569e-01 2.04846442e-01 -5.43947399e-01
-1.22486508e+00 7.06887603e-01 1.61834896e-01 1.19241051e-01
-5.83312273e-01 3.16509277e-01 7.94028640e-01 7.17273116e-01
9.16036844e-01 1.71099588e-01 -9.45681572e-01 -1.10232309e-01
-4.49003577e-01 8.37275758e-02 7.50628293e-01 1.05519426e+00
7.09490895e-01 4.99410182e-01 -5.08852363e-01 7.84197807e-01
-3.32552433e-01 6.62919223e-01 1.16182971e+00 -9.63975072e-01
2.60713696e-01 2.84353554e-01 2.50252545e-01 1.20816380e-01
-3.80727023e-01 -5.47500730e-01 -5.04402995e-01 2.69389957e-01
9.11194682e-02 -5.01818120e-01 -1.31312442e+00 1.77064443e+00
5.18753707e-01 7.95200884e-01 5.69368839e-01 1.40483314e-02
3.46799523e-01 6.21409595e-01 2.78134376e-01 -6.07136369e-01
7.70286739e-01 -1.25089693e+00 -6.24052703e-01 -4.01029646e-01
6.40296400e-01 -1.97569653e-01 9.32029128e-01 1.30343944e-01
-7.64478326e-01 -8.73469949e-01 -7.54465818e-01 3.63253236e-01
-4.14937884e-01 -2.51372129e-01 5.23840129e-01 2.87404329e-01
-9.60857928e-01 4.40996528e-01 -1.16088355e+00 -3.59361470e-01
7.20126748e-01 3.12888891e-01 -8.76018927e-02 -2.37500399e-01
-1.13014472e+00 1.15748155e+00 7.55046427e-01 -6.19135976e-01
-1.69751465e+00 -1.15429676e+00 -7.78825939e-01 -2.53452599e-01
3.62642139e-01 -9.22194421e-01 1.85173714e+00 -1.24872196e+00
-1.59640503e+00 2.83591509e-01 -4.03395951e-01 -1.22779751e+00
4.43476915e-01 -1.46756440e-01 -4.10985798e-01 -1.25926182e-01
9.83774066e-02 8.45292091e-01 1.23635209e+00 -1.02804363e+00
-9.70780671e-01 -3.33995104e-01 -1.93803996e-01 6.42566741e-01
-4.93338108e-01 -5.72051644e-01 2.08905473e-01 -6.61145627e-01
-3.46703053e-01 -9.20735598e-01 -2.37558633e-01 3.77490260e-02
3.40184957e-01 -5.19744575e-01 1.31376493e+00 -3.18822682e-01
5.96564114e-01 -2.03043246e+00 -1.14042118e-01 -5.46510100e-01
-3.01484317e-02 3.96909267e-01 -2.84679353e-01 7.99025148e-02
7.89073929e-02 -5.00757515e-01 -1.61239415e-01 -6.67871416e-01
-3.92723352e-01 6.11301064e-01 -4.85434830e-01 5.31619564e-02
-2.11521924e-01 1.14500403e+00 -1.21241498e+00 -3.39693129e-01
-4.58671972e-02 4.26366955e-01 -1.89305082e-01 1.31086364e-01
-6.73884749e-01 3.81848216e-01 -2.08750293e-01 5.46225429e-01
4.11830723e-01 -5.56884892e-02 -4.41158675e-02 2.56104261e-01
4.43770617e-01 2.59266142e-02 -7.22447634e-01 2.00458717e+00
-5.81049979e-01 5.36507249e-01 -4.67179865e-01 -1.28351402e+00
8.99915338e-01 5.23398936e-01 4.80277151e-01 -6.58533573e-01
-3.68196666e-01 5.16857021e-02 2.70099211e-02 -3.65733504e-01
2.58469462e-01 -3.76136512e-01 -8.03837776e-02 5.40697753e-01
4.60079402e-01 7.63968006e-02 -2.99678054e-02 1.25257894e-01
1.31958425e+00 2.30618849e-01 4.19791281e-01 1.83717653e-01
-3.48032005e-02 2.86553591e-01 7.92455256e-01 1.13782203e+00
-5.77643394e-01 6.44372031e-02 -1.22569531e-01 -1.06233549e+00
-1.12966442e+00 -1.42081439e+00 -4.23863754e-02 1.65789855e+00
-3.30052257e-01 3.75338882e-01 -3.04183394e-01 -1.27807844e+00
2.63437778e-01 1.13263524e+00 -1.06611764e+00 -6.74815536e-01
-6.30199611e-01 -7.09743559e-01 2.13610783e-01 6.64805830e-01
4.05429870e-01 -1.41927719e+00 -1.01540983e+00 4.65088010e-01
5.89438304e-02 -6.97679162e-01 -5.16061902e-01 5.68146586e-01
-1.28765845e+00 -1.02891517e+00 -1.00778055e+00 -1.16741157e+00
8.18103194e-01 1.80768833e-01 9.66419399e-01 -7.14040518e-01
-3.53072435e-01 9.53949511e-01 -2.27440894e-01 -1.07250857e+00
-5.78438699e-01 2.46898815e-01 7.69855440e-01 -1.74526379e-01
4.45609361e-01 -4.05508995e-01 -3.58579308e-01 6.60169348e-02
-7.55392134e-01 -2.06534758e-01 6.05066597e-01 1.19313240e+00
7.09125102e-01 -4.66269515e-02 1.34243083e+00 -1.20180357e+00
5.84507346e-01 -7.32046604e-01 -3.14669520e-01 6.08689427e-01
-7.83091962e-01 2.10511565e-01 8.13999891e-01 -1.46023333e+00
-1.44997513e+00 3.06651324e-01 3.88563603e-01 -6.93351746e-01
-8.52006376e-02 9.06282291e-02 3.70883673e-01 1.51345655e-01
9.94379222e-01 6.48151338e-01 2.50369281e-01 -3.35261136e-01
4.27699506e-01 5.24933457e-01 4.58561540e-01 -4.52934474e-01
5.62174082e-01 5.58580101e-01 -1.68843970e-01 -3.94132197e-01
-9.73970711e-01 -2.28808954e-01 -8.06137204e-01 -4.33675088e-02
4.70754713e-01 -1.04732966e+00 -2.97295004e-01 7.76140213e-01
-1.11262465e+00 -1.04160500e+00 -1.30178761e+00 5.34249783e-01
-8.20573807e-01 -4.13347818e-02 -3.91351432e-01 -7.38835812e-01
-5.42264283e-01 -6.85580790e-01 7.92375028e-01 -1.75102688e-02
1.16621993e-01 -1.23534369e+00 5.27804792e-01 -3.20805132e-01
8.65370989e-01 -1.80777892e-01 7.13331103e-01 -1.24920952e+00
-1.59233540e-01 -1.58082291e-01 3.21438074e-01 4.38362628e-01
4.55377549e-01 -9.11927462e-01 -8.80934000e-01 -8.07802856e-01
2.32000127e-01 -7.06932127e-01 1.10362530e+00 4.23827469e-01
8.22498024e-01 -3.88956666e-01 -5.97169101e-01 3.36156964e-01
9.94763017e-01 6.85186327e-01 3.17775369e-01 2.68858016e-01
2.54978091e-01 2.22924367e-01 8.07727933e-01 5.43237984e-01
2.51699954e-01 2.76222765e-01 2.24122807e-01 4.79824334e-01
-4.61333126e-01 -2.50524431e-01 7.13985801e-01 3.25148493e-01
3.59509230e-01 7.01653659e-02 -8.77859831e-01 7.62936413e-01
-1.95153439e+00 -1.20051336e+00 8.28248918e-01 2.22363067e+00
1.08320761e+00 2.84472674e-01 -5.16170524e-02 -4.28005874e-01
6.67527437e-01 -2.55624056e-01 -1.57826710e+00 2.69476715e-02
1.92474604e-01 2.44914785e-01 3.01387042e-01 4.77454424e-01
-1.16826284e+00 1.05846298e+00 6.51701164e+00 6.47495449e-01
-1.06452978e+00 7.91553855e-01 6.34370208e-01 -6.10166728e-01
2.37666190e-01 -4.38339978e-01 -1.13115335e+00 5.36325991e-01
1.23649406e+00 -5.18814147e-01 1.10100187e-01 1.37275326e+00
-4.52766091e-01 5.13697788e-02 -1.25397992e+00 8.49447548e-01
4.72035468e-01 -1.15584588e+00 7.65405297e-02 -3.84142041e-01
1.19320786e+00 4.45634872e-01 3.70545447e-01 1.11315930e+00
8.40470254e-01 -3.50118071e-01 3.84164959e-01 9.48254585e-01
8.32165003e-01 -4.42412168e-01 6.74002409e-01 5.78294039e-01
-9.70682442e-01 -5.60862780e-01 -6.50162458e-01 4.51782495e-02
8.56547505e-02 2.74950773e-01 -1.53205419e+00 -1.74035788e-01
6.83913112e-01 7.60591984e-01 -6.22262061e-01 1.18925118e+00
1.29034087e-01 5.08769333e-01 -2.77870417e-01 7.38901794e-02
-9.39628854e-02 5.63319206e-01 5.75757682e-01 6.49214745e-01
4.72060680e-01 -1.21673398e-01 5.68376482e-01 2.11933047e-01
-1.28610283e-01 -2.64135718e-01 -1.07918715e+00 4.94528413e-01
7.78245747e-01 6.73211336e-01 -2.80907631e-01 -9.20951486e-01
-2.25997195e-01 1.48767745e+00 6.13146067e-01 6.07440591e-01
-7.04974711e-01 -1.13180839e-01 3.95967215e-01 -6.52032122e-02
4.51657534e-01 -1.27270728e-01 1.37583524e-01 -1.03128743e+00
-1.00470878e-01 -4.50489968e-01 6.67607486e-01 -6.20770216e-01
-1.56836116e+00 6.49862647e-01 2.02739120e-01 -1.26934040e+00
-8.49064469e-01 -2.83524960e-01 -5.81632197e-01 4.38416988e-01
-1.48551106e+00 -1.26985931e+00 -1.94785580e-01 8.35368216e-01
1.38566446e+00 -8.44926000e-01 9.75338221e-01 -4.79835458e-02
-1.88406706e-01 6.67748570e-01 5.38738847e-01 -3.33172262e-01
1.16883254e+00 -1.08062994e+00 6.57496929e-01 4.55269575e-01
-2.69600842e-02 2.46865392e-01 3.33134174e-01 -1.15006757e+00
-9.40806091e-01 -1.59890831e+00 5.13409853e-01 -7.78888166e-01
4.43852752e-01 -5.15974522e-01 -9.52909529e-01 1.25657392e+00
5.64716831e-02 4.41141099e-01 5.52762806e-01 -2.14757100e-01
-3.31294715e-01 -2.90607244e-01 -1.37443030e+00 3.26788038e-01
1.17984176e+00 -2.63914853e-01 -8.94405246e-01 6.39073253e-01
1.19360268e+00 -2.22327158e-01 -4.31320697e-01 4.23885494e-01
3.65656734e-01 -2.27893725e-01 9.52345610e-01 -1.07962692e+00
-4.08504725e-01 1.62245676e-01 6.49513379e-02 -1.77713418e+00
-4.19070810e-01 -2.86160409e-01 -7.35368729e-01 5.89474261e-01
5.94132066e-01 -9.96812940e-01 9.21966553e-01 5.36044538e-01
-4.24256086e-01 -5.82338154e-01 -1.24198949e+00 -1.27188158e+00
2.06644181e-02 -2.63151407e-01 2.92542785e-01 8.50137711e-01
-3.13820153e-01 4.92716730e-01 -4.67464387e-01 -2.42531642e-01
8.14326942e-01 -1.76843062e-01 6.58930540e-01 -1.56336117e+00
-3.03293794e-01 1.65552467e-01 -1.55370757e-01 -7.48962164e-01
5.45749187e-01 -9.64053810e-01 3.43537237e-03 -1.48202157e+00
3.00457507e-01 -6.10029042e-01 -6.15864336e-01 8.09099555e-01
-2.29650721e-01 -3.17361914e-02 1.83494482e-02 5.32234550e-01
-1.22558832e+00 7.48314381e-01 9.59998012e-01 -3.85208726e-01
-6.28458619e-01 4.51771677e-01 -3.19246113e-01 5.50463259e-01
9.36231971e-01 -8.41998875e-01 -9.13162529e-01 -4.86351252e-01
-1.50561020e-01 -3.92580062e-01 2.26985157e-01 -1.23796356e+00
6.31790161e-01 -1.15247004e-01 8.26432765e-01 -5.28229415e-01
6.41377091e-01 -9.71861541e-01 -1.05305187e-01 1.08028579e+00
-4.41163719e-01 -9.79797244e-02 2.89452553e-01 1.09705484e+00
3.75209659e-01 -4.93622303e-01 7.54652917e-01 -5.55726707e-01
-1.07746911e+00 5.54323137e-01 -3.13521415e-01 1.24200337e-01
1.65997016e+00 -8.39107931e-02 -5.08644342e-01 -2.51077384e-01
-1.22259474e+00 2.44449973e-01 2.09375918e-01 5.49699903e-01
8.99084389e-01 -1.18894327e+00 -7.84821749e-01 3.31742555e-01
3.62027325e-02 2.47942880e-02 3.88430566e-01 2.85612226e-01
1.30209178e-01 1.86179399e-01 -3.55451465e-01 -5.60357034e-01
-1.02395701e+00 1.14139962e+00 2.98697889e-01 -3.70959610e-01
-5.55733860e-01 9.17114377e-01 6.76685497e-02 -7.32121646e-01
6.06488466e-01 -6.48251846e-02 -1.06415637e-01 2.33417198e-01
6.81623816e-01 5.74492693e-01 -1.70494065e-01 -7.21027283e-03
-2.90291272e-02 -1.12203590e-04 -8.60757113e-01 -2.62696207e-01
1.27226496e+00 -7.24562928e-02 3.65731508e-01 1.17011094e+00
9.66503859e-01 -8.29721510e-01 -1.75412393e+00 -8.85207117e-01
-8.74520689e-02 -5.51538728e-02 -3.69566888e-01 -1.00609422e+00
-4.72331047e-01 6.83457196e-01 1.29585528e+00 -3.23126942e-01
7.16734231e-01 1.62831292e-01 7.69603968e-01 1.00144649e+00
7.08298028e-01 -1.34036756e+00 7.87989020e-01 7.96487570e-01
7.15108931e-01 -1.50737715e+00 1.23362476e-02 3.45500380e-01
-8.21539998e-01 7.09056318e-01 1.02636433e+00 2.51491833e-02
8.31983030e-01 1.34652704e-01 -9.19609517e-02 2.11577132e-01
-1.28331053e+00 -4.36721742e-02 1.11412197e-01 1.13465011e+00
-2.30689600e-01 -3.24093625e-02 3.11770290e-01 4.15231198e-01
9.63739157e-02 3.53802174e-01 2.21002802e-01 1.29772341e+00
-7.40990222e-01 -1.01597512e+00 -1.44833967e-01 8.52048516e-01
-1.81029558e-01 -2.09717199e-01 9.32519790e-03 5.79156458e-01
1.82761922e-01 3.64004344e-01 4.54928488e-01 -6.47828951e-02
2.45811731e-01 9.29367840e-01 6.71661139e-01 -1.04774630e+00
-2.96122193e-01 -5.40782213e-01 -3.07150245e-01 -1.21282212e-01
-1.80582866e-01 -6.41339779e-01 -1.12203574e+00 3.75830144e-01
-9.59210694e-02 6.16406575e-02 6.76693439e-01 6.87532961e-01
6.96736753e-01 6.67152405e-01 6.05505168e-01 -8.56646955e-01
-9.82042551e-01 -1.37281203e+00 -2.45251670e-01 2.83355772e-01
4.44121778e-01 -8.66731107e-01 -2.33941317e-01 3.82203192e-01] | [9.852714538574219, 3.3735783100128174] |
3b538bf3-7abe-47f3-aed1-07fe22124c85 | basen-time-domain-brain-assisted-speech | 2305.09994 | null | https://arxiv.org/abs/2305.09994v1 | https://arxiv.org/pdf/2305.09994v1.pdf | BASEN: Time-Domain Brain-Assisted Speech Enhancement Network with Convolutional Cross Attention in Multi-talker Conditions | Time-domain single-channel speech enhancement (SE) still remains challenging to extract the target speaker without any prior information on multi-talker conditions. It has been shown via auditory attention decoding that the brain activity of the listener contains the auditory information of the attended speaker. In this paper, we thus propose a novel time-domain brain-assisted SE network (BASEN) incorporating electroencephalography (EEG) signals recorded from the listener for extracting the target speaker from monaural speech mixtures. The proposed BASEN is based on the fully-convolutional time-domain audio separation network. In order to fully leverage the complementary information contained in the EEG signals, we further propose a convolutional multi-layer cross attention module to fuse the dual-branch features. Experimental results on a public dataset show that the proposed model outperforms the state-of-the-art method in several evaluation metrics. The reproducible code is available at https://github.com/jzhangU/Basen.git. | ['Zhen-Hua Ling', 'Qiu-Shi Zhu', 'Qing-Tian Xu', 'Jie Zhang'] | 2023-05-17 | null | null | null | null | ['speech-enhancement'] | ['speech'] | [ 1.25541061e-01 -2.11783841e-01 5.60560584e-01 -3.37550253e-01
-1.23058105e+00 -2.00553373e-01 2.90173113e-01 -2.09141284e-01
-3.82257074e-01 5.52931786e-01 4.05040920e-01 7.92124942e-02
-1.13061070e-01 -1.32113338e-01 -6.52617991e-01 -8.16421807e-01
1.48933064e-02 -2.50236392e-01 -1.46062493e-01 -5.16651431e-03
9.55478176e-02 1.48191199e-01 -1.62552011e+00 5.12962461e-01
9.75904167e-01 1.44183266e+00 4.97229904e-01 7.64130652e-01
3.57072324e-01 6.04287863e-01 -6.35313094e-01 -4.51784544e-02
-3.12358625e-02 -7.08948195e-01 -3.41149688e-01 -1.92312703e-01
1.18956916e-01 -1.37178510e-01 -6.16901159e-01 1.02119493e+00
1.00766098e+00 1.06709950e-01 4.09147769e-01 -1.27009416e+00
-2.18657330e-01 4.96056318e-01 -2.95316398e-01 6.50881827e-01
1.48061767e-01 -4.09388766e-02 7.80847073e-01 -1.13802361e+00
-1.22059949e-01 7.32007146e-01 5.28728127e-01 3.54384750e-01
-8.62849534e-01 -1.15952897e+00 2.53669880e-02 6.76981091e-01
-1.47813928e+00 -9.35404718e-01 1.16952133e+00 -2.26940170e-01
9.24716234e-01 1.61888525e-01 5.76029539e-01 1.41554546e+00
2.98462272e-01 8.95958722e-01 1.35876846e+00 -2.42953449e-01
1.96744785e-01 7.83663839e-02 4.05313410e-02 2.33706072e-01
-3.46317053e-01 7.19030127e-02 -1.02460778e+00 -2.21606549e-02
3.27594966e-01 -1.25491828e-01 -6.71030402e-01 3.14708799e-01
-1.00290120e+00 4.12544519e-01 3.69787931e-01 5.84770143e-01
-8.46258402e-01 1.61939263e-02 3.08034033e-01 9.61950421e-02
7.11993277e-01 2.23956928e-01 -3.79530221e-01 -3.67095739e-01
-1.24553871e+00 -1.13209181e-01 6.25237286e-01 4.69972193e-01
3.87200505e-01 4.03076977e-01 -2.32920229e-01 9.10222411e-01
1.53612897e-01 4.21834737e-01 6.72214270e-01 -7.20469713e-01
1.89392865e-01 -4.52696458e-02 -1.30561218e-01 -5.86141944e-01
-2.11198822e-01 -9.22999084e-01 -8.07426333e-01 -8.46062601e-02
2.18833350e-02 -4.58352834e-01 -6.06772125e-01 1.86920249e+00
1.98372930e-01 8.10953736e-01 1.29135042e-01 1.11438775e+00
9.29832518e-01 6.52286112e-01 -8.29925910e-02 -1.48556054e-01
1.43000960e+00 -1.00959373e+00 -8.99222910e-01 -4.49925542e-01
-2.90096104e-01 -6.35658801e-01 6.15474939e-01 6.61932051e-01
-1.13033080e+00 -6.15930855e-01 -1.14188600e+00 1.54638916e-01
-3.81468609e-02 2.57046551e-01 2.56274670e-01 5.08722425e-01
-9.64530408e-01 3.50248188e-01 -1.07866597e+00 5.46648987e-02
4.45278108e-01 3.32837880e-01 -3.48972887e-01 1.53173208e-01
-1.28176177e+00 6.23458266e-01 5.07356972e-02 3.29414994e-01
-1.38872492e+00 -7.86596239e-01 -6.92496538e-01 3.28517258e-01
2.19982505e-01 -4.68616694e-01 1.39506543e+00 -1.06728971e+00
-1.63341546e+00 2.53684253e-01 -5.23953795e-01 -5.27332306e-01
3.59801799e-02 -3.91055435e-01 -6.06097877e-01 4.76685524e-01
-1.54118538e-01 3.81996959e-01 1.01205683e+00 -9.36296105e-01
-5.05109906e-01 -6.00485027e-01 -4.89921987e-01 2.54237413e-01
-4.75259840e-01 2.67416835e-01 -1.78477928e-01 -8.12910974e-01
-3.25941704e-02 -5.43949246e-01 3.62820625e-01 -4.25592482e-01
-3.45837563e-01 -1.13181919e-01 5.08170962e-01 -1.19447553e+00
1.04618573e+00 -2.46765995e+00 1.74650133e-01 -1.12583898e-01
3.06845307e-01 1.60489634e-01 -2.28230700e-01 2.96084136e-01
-3.62601370e-01 -5.13104558e-01 -2.33999342e-01 -5.72790325e-01
7.27898404e-02 -5.00048757e-01 -3.34094763e-01 5.30937135e-01
2.61206180e-01 6.62384629e-01 -5.69603086e-01 -4.67705876e-02
-3.06558255e-02 1.18823636e+00 -3.36934000e-01 5.02637506e-01
4.17378455e-01 8.03358138e-01 -5.66284508e-02 3.95899802e-01
7.49471426e-01 1.06025212e-01 -2.59729534e-01 -3.18311960e-01
-1.70450255e-01 8.12080026e-01 -9.24312830e-01 1.93169260e+00
-5.74305117e-01 8.32412422e-01 5.42761981e-01 -9.01907742e-01
8.69062006e-01 8.53875875e-01 4.14097786e-01 -7.65295565e-01
5.23248851e-01 2.94708014e-01 3.95949870e-01 -3.30266923e-01
-1.22772455e-01 -2.44622350e-01 1.65919110e-01 4.18271929e-01
6.07522786e-01 1.36005193e-01 -3.16875845e-01 -3.15596983e-02
1.18466210e+00 -2.68990517e-01 2.18174934e-01 -4.02798206e-02
5.80131471e-01 -7.85702169e-01 4.92719918e-01 5.16665399e-01
-3.23682457e-01 5.57304323e-01 -2.16870587e-02 2.92858452e-01
-5.65675259e-01 -1.15698171e+00 -8.23445767e-02 1.20058525e+00
-2.49538213e-01 -2.87694246e-01 -1.11594725e+00 -1.38602659e-01
-3.96735340e-01 8.50843608e-01 -4.78065372e-01 -3.85058641e-01
-1.23622246e-01 -5.43045282e-01 6.80415630e-01 6.43624306e-01
5.41139185e-01 -1.03498769e+00 -5.91272354e-01 4.24149036e-01
-7.14631915e-01 -1.27347457e+00 -5.92923999e-01 4.39635038e-01
-4.08813685e-01 -5.31244934e-01 -8.73347282e-01 -5.68745375e-01
9.98792872e-02 1.63771436e-01 5.58262527e-01 -5.48241377e-01
-2.22221851e-01 6.57355487e-01 -2.58004785e-01 -9.23993528e-01
-2.97147054e-02 -1.58584967e-01 2.60392189e-01 5.86793542e-01
6.29559755e-01 -1.04170477e+00 -7.37461865e-01 1.82852209e-01
-4.64765787e-01 1.07245333e-03 6.44984543e-01 6.31513953e-01
4.04186785e-01 2.55733252e-01 1.09904063e+00 9.25106481e-02
7.70834446e-01 -5.59287369e-01 -2.98259407e-01 -5.51242568e-02
-6.74654767e-02 -2.13278830e-01 3.56856376e-01 -7.02779531e-01
-1.37642729e+00 -7.73780271e-02 -3.39860857e-01 -5.20144224e-01
-5.03850877e-01 4.08686340e-01 -4.02570784e-01 1.19647928e-01
2.91471809e-01 6.50583029e-01 -2.22105935e-01 -7.18862593e-01
-1.64595634e-01 1.22370422e+00 9.02070582e-01 -3.20009530e-01
2.21142098e-01 3.46711725e-01 -5.20407617e-01 -8.88467491e-01
-7.74174094e-01 -5.97079396e-01 -3.18342119e-01 -3.29272062e-01
8.81058156e-01 -1.32177877e+00 -5.91171920e-01 8.40387940e-01
-1.29940927e+00 -2.51300097e-01 8.22723955e-02 1.04717040e+00
-5.88454962e-01 8.95842612e-02 -5.08952796e-01 -1.19355249e+00
-7.85238147e-01 -1.07051468e+00 8.70090008e-01 1.51196301e-01
-1.04947403e-01 -4.25695628e-01 1.71430573e-01 5.63487411e-01
5.69497824e-01 -2.98613101e-01 2.89195746e-01 -1.03041875e+00
-1.45501047e-01 -1.64134011e-01 1.44928262e-01 5.55596769e-01
1.17391840e-01 -6.77036643e-01 -1.81923366e+00 -2.18267187e-01
7.88731992e-01 -6.61115572e-02 8.36852312e-01 7.14708745e-01
1.15744662e+00 -1.30750895e-01 7.62696285e-03 6.65566862e-01
1.00534761e+00 3.49417001e-01 6.26547337e-01 -6.03255033e-02
3.55182111e-01 5.05662739e-01 1.60256118e-01 5.34483612e-01
4.01640892e-01 4.32993859e-01 3.07398766e-01 -1.23068869e-01
-3.00816506e-01 -2.26193257e-02 6.08996749e-01 1.20898414e+00
9.56353173e-03 -1.90882578e-01 -8.42247069e-01 8.06259513e-01
-1.33675563e+00 -7.76341140e-01 1.72896892e-01 2.10526204e+00
8.45479965e-01 -1.14080958e-01 5.86285181e-02 4.54196900e-01
7.01877654e-01 -2.07391486e-01 -6.00947559e-01 -1.03975497e-01
-5.49233630e-02 6.60789669e-01 -8.19226727e-02 2.68651783e-01
-1.06828868e+00 2.99619645e-01 4.57221508e+00 8.54658484e-01
-1.40020144e+00 6.83188677e-01 2.92641848e-01 -3.99341494e-01
2.67807424e-01 -6.09620392e-01 -4.69032735e-01 5.00460684e-01
1.63754404e+00 -4.18587804e-01 7.25623369e-01 2.88765997e-01
4.04137105e-01 3.73282842e-02 -1.10716033e+00 1.19900703e+00
3.31541270e-01 -5.88840485e-01 -6.78087473e-01 -2.25879718e-02
1.91896796e-01 4.28394079e-01 2.73985505e-01 2.77180821e-01
-3.21664810e-01 -8.68401289e-01 9.21344519e-01 6.14888847e-01
7.23103166e-01 -8.08887243e-01 6.87856317e-01 4.24955487e-01
-1.33200896e+00 -3.45770031e-01 1.05241098e-01 -2.47875880e-02
-8.31753686e-02 5.03438711e-01 -7.03596890e-01 5.77863276e-01
1.01571643e+00 6.54680729e-01 -3.67347360e-01 1.36864519e+00
-3.40729803e-01 1.08108175e+00 -1.88399673e-01 3.17513168e-01
-6.56191185e-02 2.65616626e-01 8.49271178e-01 1.27110088e+00
6.18350863e-01 3.39516908e-01 -4.48193550e-01 1.01528978e+00
-9.37682167e-02 1.12773152e-02 -3.55865628e-01 -2.42818594e-02
3.39235187e-01 1.36131918e+00 -2.92838693e-01 -6.00517355e-03
-3.79368246e-01 1.02216589e+00 -5.44428872e-03 5.59421599e-01
-9.73415315e-01 -7.51350164e-01 6.01230621e-01 -2.06600547e-01
5.48050165e-01 -3.96133922e-02 -8.82441085e-03 -1.10629487e+00
2.66858101e-01 -9.46289599e-01 1.31182656e-01 -1.12282860e+00
-1.25434601e+00 1.13179815e+00 -3.38203609e-01 -1.19786334e+00
2.40475535e-02 -4.25444216e-01 -7.00326324e-01 1.30165350e+00
-1.64354193e+00 -9.58041728e-01 -3.45267028e-01 8.81201982e-01
7.15401351e-01 -9.18180123e-02 7.99787998e-01 6.39476061e-01
-6.12524986e-01 6.42207146e-01 -3.23283561e-02 2.17725575e-01
8.78343523e-01 -8.79393578e-01 -5.65075055e-02 9.96300519e-01
2.63214228e-03 4.24172610e-01 5.51944613e-01 -2.95636415e-01
-1.24471366e+00 -1.09201527e+00 7.52466261e-01 -4.48190980e-02
6.32968307e-01 -7.57535458e-01 -1.13717484e+00 4.28974688e-01
7.82396853e-01 -2.20264897e-01 9.04773593e-01 -2.30632648e-01
-3.33490968e-01 -4.59447503e-01 -8.23372006e-01 7.69286379e-02
4.66987491e-01 -9.49623644e-01 -8.36795568e-01 -6.03436902e-02
6.50088370e-01 -3.80043909e-02 -7.46285796e-01 2.73117870e-01
4.91123259e-01 -7.72268832e-01 8.86995316e-01 -2.22560346e-01
2.60667771e-01 -1.79687575e-01 -3.47096145e-01 -1.62063646e+00
-1.63855895e-01 -6.79099262e-01 -2.47208223e-01 1.28378952e+00
1.86233371e-01 -8.00763905e-01 8.67017433e-02 2.66718447e-01
-4.23725486e-01 -5.16529679e-01 -1.16417873e+00 -7.86097229e-01
-1.20849304e-01 -8.23726594e-01 5.21511078e-01 4.88527954e-01
3.16137463e-01 5.22402108e-01 -3.65856498e-01 5.62549293e-01
6.56145930e-01 -1.91956773e-01 1.30330309e-01 -1.25824332e+00
-4.05026913e-01 -3.47956508e-01 -3.03582132e-01 -8.52539718e-01
4.06480670e-01 -9.80125666e-01 4.43954110e-01 -1.32160723e+00
2.86272943e-01 3.26189250e-01 -1.00815642e+00 4.27056432e-01
-1.79520428e-01 1.83663487e-01 -1.37333758e-04 -2.67290145e-01
-4.69964594e-01 1.05691850e+00 8.04093897e-01 -2.44663924e-01
-1.76777944e-01 1.53788477e-02 -9.09730196e-01 7.40472496e-01
8.33284080e-01 -6.39367461e-01 -3.77936512e-01 -3.50971103e-01
-3.65261704e-01 3.82818222e-01 5.83870053e-01 -1.48948097e+00
6.22159541e-01 4.57617819e-01 3.24682891e-01 -5.19114435e-01
8.54473710e-01 -8.16513062e-01 2.41509508e-02 1.74176738e-01
-3.97855282e-01 -3.66053283e-01 5.48950255e-01 5.70216060e-01
-5.16051829e-01 1.22555524e-01 7.76383698e-01 2.40302876e-01
-1.38413012e-01 1.52257875e-01 -4.98466641e-01 -6.12338632e-02
6.26204908e-01 1.69382870e-01 -2.52627939e-01 -6.35664761e-01
-7.58782804e-01 -8.34688172e-02 -4.19039309e-01 3.63460183e-01
7.98116565e-01 -1.27245975e+00 -1.01622617e+00 4.01400745e-01
3.81341553e-03 -5.14861822e-01 6.87123954e-01 1.42860758e+00
5.27179062e-01 5.94875693e-01 -3.55988950e-01 -4.40997064e-01
-1.18380499e+00 3.68066490e-01 6.93153024e-01 3.76514047e-01
-4.89746094e-01 1.09167898e+00 5.41430295e-01 -5.37061412e-03
4.27856654e-01 -2.78938591e-01 -3.03949088e-01 1.96114276e-02
9.39437509e-01 4.26259935e-01 4.89924312e-01 -7.52262592e-01
-5.45586288e-01 1.30750343e-01 1.60751641e-01 -5.31894565e-01
1.73168290e+00 -2.19099447e-01 -4.44313735e-02 6.82629704e-01
1.17275143e+00 -7.40931779e-02 -1.24094748e+00 -3.92697811e-01
-3.93704921e-01 -2.68270969e-01 6.18084371e-01 -1.05811524e+00
-1.19950080e+00 1.45291841e+00 8.96655142e-01 -1.80410624e-01
1.83645916e+00 -8.72690678e-02 8.53665411e-01 1.86519787e-01
2.31928974e-01 -7.23788083e-01 1.19717054e-01 2.34766260e-01
1.15029740e+00 -9.34539080e-01 -5.38008511e-01 -1.88011918e-02
-6.66495264e-01 8.49655390e-01 4.80899096e-01 1.30410260e-02
9.35168684e-01 3.85512948e-01 5.25678992e-02 -1.05233446e-01
-9.33033586e-01 -1.70933127e-01 5.19501030e-01 4.95874763e-01
4.02366787e-01 2.28136666e-02 1.66651830e-01 1.60512328e+00
-3.22902411e-01 6.75191656e-02 2.96501219e-01 5.61098278e-01
-3.40137988e-01 -5.56937695e-01 -6.02137744e-01 3.09611440e-01
-7.64450729e-01 -4.92309421e-01 -3.16235572e-01 1.07591882e-01
-1.03992000e-01 1.40032840e+00 -1.38559481e-02 -5.77873945e-01
3.40373486e-01 6.14050627e-01 3.85027677e-01 -2.96946585e-01
-7.13145792e-01 7.20922709e-01 -9.12032723e-02 -4.64829743e-01
-4.28606719e-01 -7.49247909e-01 -1.13761222e+00 2.43329287e-01
-2.86195457e-01 2.34856606e-01 8.48296583e-01 8.25081050e-01
7.02117324e-01 1.01858139e+00 6.50939941e-01 -1.00723612e+00
-2.43410319e-01 -1.35692751e+00 -7.28127480e-01 -3.06390021e-02
7.92940021e-01 -6.41007960e-01 -6.46712005e-01 9.66186002e-02] | [15.193021774291992, 5.533141613006592] |
5b28fe7a-cfa5-4586-85fe-940bf29acc04 | unsupervised-summarization-re-ranking | 2212.09593 | null | https://arxiv.org/abs/2212.09593v3 | https://arxiv.org/pdf/2212.09593v3.pdf | Unsupervised Summarization Re-ranking | With the rise of task-specific pre-training objectives, abstractive summarization models like PEGASUS offer appealing zero-shot performance on downstream summarization tasks. However, the performance of such unsupervised models still lags significantly behind their supervised counterparts. Similarly to the supervised setup, we notice a very high variance in quality among summary candidates from these models while only one candidate is kept as the summary output. In this paper, we propose to re-rank summary candidates in an unsupervised manner, aiming to close the performance gap between unsupervised and supervised models. Our approach improves the unsupervised PEGASUS by up to 7.27% and ChatGPT by up to 6.86% relative mean ROUGE across four widely-adopted summarization benchmarks ; and achieves relative gains of 7.51% (up to 23.73% from XSum to WikiHow) averaged over 30 zero-shot transfer setups (finetuning on a dataset, evaluating on another). | ['Nancy Chen', 'Shafiq Joty', 'Mathieu Ravaut'] | 2022-12-19 | null | null | null | null | ['abstractive-text-summarization'] | ['natural-language-processing'] | [ 3.75332624e-01 5.63590467e-01 -3.77746880e-01 -2.04832152e-01
-1.62752855e+00 -5.75849593e-01 1.01566482e+00 7.37620711e-01
-5.47021985e-01 1.00847936e+00 1.00356841e+00 3.66795212e-02
-9.93950516e-02 -4.27810282e-01 -6.41101360e-01 -4.11214560e-01
1.15188986e-01 6.27213597e-01 2.41850272e-01 -3.97742748e-01
6.19587541e-01 -1.58937857e-01 -1.35150766e+00 3.46494794e-01
1.27409434e+00 3.92043978e-01 1.23514026e-01 1.03990114e+00
-1.37895122e-01 6.81503296e-01 -9.19540167e-01 -5.83293021e-01
-1.55307710e-01 -5.07179856e-01 -1.13482583e+00 -1.42452670e-02
8.58365893e-01 -3.08260769e-01 -3.42318863e-01 7.87915766e-01
7.57643223e-01 3.62538010e-01 7.90811062e-01 -7.75278270e-01
-6.28236711e-01 1.41035748e+00 -4.77105170e-01 3.13645989e-01
3.06496054e-01 -1.19057782e-02 1.55396938e+00 -6.21259987e-01
8.61343563e-01 9.08052742e-01 5.15523493e-01 5.04532456e-01
-1.26546347e+00 -2.14521989e-01 -1.19160973e-02 -1.64981224e-02
-8.70123744e-01 -8.53183806e-01 1.89253435e-01 -1.69548661e-01
1.38617814e+00 5.06380498e-01 3.31098169e-01 9.69320118e-01
4.36852016e-02 1.06700397e+00 5.94777882e-01 -2.84453422e-01
1.27798378e-01 -1.10064447e-01 7.54132032e-01 3.26481968e-01
5.39189875e-01 -6.86662495e-01 -8.21772575e-01 -7.34241530e-02
-1.15216993e-01 -4.09337103e-01 -2.30536565e-01 1.40573353e-01
-1.35554469e+00 5.98652780e-01 1.93878934e-01 2.13741153e-01
-2.74350405e-01 1.98621005e-01 9.23077106e-01 3.86477441e-01
7.69455254e-01 9.74792719e-01 -2.91919440e-01 -4.98188734e-01
-1.42869163e+00 4.07187670e-01 9.31711614e-01 1.24796593e+00
6.37132466e-01 3.79036590e-02 -1.04533947e+00 8.78313482e-01
-4.66944158e-01 2.74432987e-01 6.77160859e-01 -1.06194293e+00
8.96681249e-01 2.64793217e-01 1.85736150e-01 -5.05406439e-01
-2.48035237e-01 -7.14180946e-01 -7.69217372e-01 -5.99929035e-01
1.69977620e-01 -1.92100003e-01 -7.88068473e-01 1.57722688e+00
-2.80861497e-01 9.19800904e-03 2.25104541e-01 4.06077564e-01
1.18585706e+00 9.95549023e-01 6.33626580e-02 -5.34119189e-01
1.02302861e+00 -1.48241162e+00 -6.78746104e-01 -2.88586080e-01
9.54081774e-01 -7.29120195e-01 1.28591585e+00 2.89486647e-01
-1.44797528e+00 -3.22504908e-01 -1.16634405e+00 -2.77608782e-01
-5.01128063e-02 1.89565867e-01 1.91798538e-01 2.97733933e-01
-1.35202491e+00 1.15813029e+00 -6.93545580e-01 -7.78723717e-01
2.58318543e-01 5.75561635e-02 -1.56907141e-01 1.18647598e-01
-8.61063123e-01 7.52454817e-01 6.17587566e-01 -5.47851980e-01
-7.34966338e-01 -1.04958129e+00 -6.05155468e-01 4.52930540e-01
4.18170929e-01 -1.01238382e+00 1.63985646e+00 -4.46624666e-01
-1.58161271e+00 7.64553308e-01 -1.38632998e-01 -8.44290733e-01
7.03771114e-01 -7.19494581e-01 -5.08138500e-02 2.30348036e-01
4.15581673e-01 5.60225964e-01 4.13724512e-01 -9.02553976e-01
-7.77767122e-01 9.61496755e-02 -9.83286183e-03 5.86449742e-01
-6.15199625e-01 1.39384076e-01 -5.34089625e-01 -5.94820917e-01
-4.85377729e-01 -5.72284281e-01 -1.44574568e-01 -8.54059637e-01
-8.52019191e-01 -3.80554289e-01 5.19061029e-01 -8.28487873e-01
1.58939362e+00 -1.71399581e+00 4.16957378e-01 -5.37924230e-01
3.23410660e-01 5.57010233e-01 -2.85051882e-01 9.39397931e-01
2.27742061e-01 3.03274393e-01 -4.69758391e-01 -7.80151665e-01
6.29694387e-02 -3.56392339e-02 -4.12347198e-01 9.56884995e-02
1.76009446e-01 1.11167943e+00 -1.26603186e+00 -4.89133596e-01
-1.73251092e-01 -2.76601553e-01 -6.09048665e-01 2.25416571e-01
-2.16955841e-01 2.02411920e-01 -3.10025185e-01 2.46827930e-01
1.87795714e-01 -3.52436453e-01 -6.08704537e-02 1.58373922e-01
-1.66702658e-01 8.20440650e-01 -4.86620218e-01 2.03500032e+00
-4.07209903e-01 8.21962178e-01 -4.03326839e-01 -9.28024590e-01
7.02714443e-01 3.47158074e-01 2.65260041e-01 -4.83755231e-01
-1.27169248e-02 1.24436088e-01 -8.05758089e-02 -2.41981655e-01
1.31306267e+00 1.27427250e-01 -4.87000138e-01 7.07136095e-01
5.80899477e-01 -1.32879063e-01 8.15917373e-01 9.99691486e-01
1.38629150e+00 -1.19917400e-01 4.64913338e-01 -4.71680790e-01
2.33048946e-01 1.16498224e-01 3.22744161e-01 1.17853284e+00
1.17127001e-01 9.30941999e-01 8.23448002e-01 2.09385231e-01
-1.15321398e+00 -9.64194477e-01 1.78197980e-01 1.31699586e+00
1.43307038e-02 -1.08631170e+00 -9.05013442e-01 -7.20714867e-01
-1.00312926e-01 1.29181588e+00 -4.39195782e-01 -4.24214572e-01
-5.57359397e-01 -5.54306567e-01 8.93209279e-01 4.03597265e-01
4.86698121e-01 -9.56215143e-01 -4.75639343e-01 2.66704172e-01
-3.83690745e-01 -9.92973506e-01 -6.30632579e-01 9.67546552e-02
-1.02492285e+00 -4.42173302e-01 -1.00990033e+00 -4.35480624e-01
1.95444167e-01 4.18582380e-01 1.41916811e+00 -1.70132235e-01
9.56137702e-02 2.65484273e-01 -6.24922097e-01 -4.18653131e-01
-5.79367518e-01 9.03680742e-01 -3.56271900e-02 -3.37139994e-01
-1.74903244e-01 -5.57840288e-01 -5.28816283e-01 -2.33065203e-01
-7.65350759e-01 1.60627976e-01 5.51253855e-01 7.51599967e-01
2.26084113e-01 -5.11298299e-01 1.10161638e+00 -1.21844494e+00
1.00325418e+00 -4.86251235e-01 8.23057070e-02 3.90542090e-01
-8.10110211e-01 1.66368112e-02 8.01001787e-01 -7.85901025e-03
-1.26829588e+00 -6.59340084e-01 1.58771306e-01 -2.72530448e-02
5.92870899e-02 6.92123652e-01 1.31670401e-01 7.36952782e-01
8.20530713e-01 3.44111204e-01 -1.72615454e-01 -5.96715629e-01
6.63249254e-01 7.66566932e-01 7.10591137e-01 -3.31047207e-01
5.26556909e-01 6.48440570e-02 -6.26902819e-01 -9.33393121e-01
-1.26567447e+00 -7.29694426e-01 -5.40328979e-01 2.88605466e-02
6.07963145e-01 -9.78880763e-01 1.98571868e-02 2.46282503e-01
-1.04431570e+00 -4.33614224e-01 -5.58906615e-01 8.81248564e-02
-6.23259664e-01 7.47381926e-01 -7.83173382e-01 -4.36844736e-01
-1.20457196e+00 -7.08902001e-01 1.07363939e+00 2.66269594e-01
-6.69844329e-01 -7.99702883e-01 1.98875889e-01 3.95070136e-01
3.56426656e-01 -6.80291802e-02 7.63235509e-01 -1.09530807e+00
-1.58530280e-01 -2.59416729e-01 -2.27183044e-01 3.18201482e-01
1.36996344e-01 2.78775424e-01 -7.62973189e-01 -4.07790661e-01
-4.99545187e-01 -3.50859672e-01 1.45343268e+00 3.82840991e-01
8.15609992e-01 -5.21921575e-01 -1.74707472e-01 1.65382028e-01
1.03741360e+00 -2.56571800e-01 5.53290069e-01 2.46613681e-01
5.49588680e-01 5.16482115e-01 7.79600859e-01 4.83230501e-01
3.03621680e-01 4.75318700e-01 -2.34284885e-02 4.14102435e-01
-4.33915854e-01 -3.81191641e-01 5.81341982e-01 1.41801798e+00
-3.81670780e-02 -6.07846439e-01 -7.15602160e-01 8.47977281e-01
-2.14873838e+00 -1.04761147e+00 -2.64685094e-01 2.14171338e+00
9.61183429e-01 3.43520701e-01 3.13629776e-01 -6.10230975e-02
6.69360697e-01 5.14465451e-01 -4.32592899e-01 -6.53725922e-01
-1.09375373e-01 1.26354665e-01 4.68712121e-01 3.24422270e-01
-9.28621590e-01 1.17940724e+00 6.39167738e+00 1.02394927e+00
-8.40566039e-01 -1.36577552e-02 5.04656672e-01 -4.28513497e-01
-4.07134593e-01 9.78754014e-02 -7.71965086e-01 5.21530092e-01
1.28230166e+00 -1.03189051e+00 -5.00379875e-02 6.76493049e-01
2.70177871e-01 -1.27738982e-01 -1.14430797e+00 6.09847188e-01
2.44210303e-01 -1.59812784e+00 3.59754056e-01 -2.15544179e-01
1.05023229e+00 2.89708465e-01 -1.49344772e-01 6.60126448e-01
4.78750765e-01 -7.88295269e-01 7.89131761e-01 2.77486295e-01
7.69907296e-01 -8.39752674e-01 6.37347579e-01 5.99874020e-01
-6.69044018e-01 1.55652881e-01 -4.85150874e-01 -1.32759467e-01
4.14426029e-01 6.11159205e-01 -9.63528693e-01 1.04341364e+00
4.16881084e-01 1.01873064e+00 -7.78313160e-01 1.11698794e+00
-2.72116750e-01 9.46589589e-01 3.63798179e-02 -2.08543152e-01
5.16165495e-01 -9.19421092e-02 9.00137067e-01 1.68972623e+00
2.68021613e-01 -1.88352130e-02 -7.79592618e-02 4.17078048e-01
-5.92745245e-01 3.19019079e-01 -3.71884882e-01 -2.48144582e-01
4.23230380e-01 1.34187627e+00 -5.23086250e-01 -1.01131320e+00
2.13169269e-02 1.04585361e+00 5.13447404e-01 2.05025464e-01
-9.06445861e-01 -7.99066365e-01 2.66966403e-01 -8.91391095e-03
2.92799324e-01 1.51869375e-02 -3.87503415e-01 -1.32630920e+00
-4.88666855e-02 -6.00110710e-01 4.66137022e-01 -5.43676019e-01
-1.10390818e+00 5.53688526e-01 2.03557193e-01 -1.13488233e+00
-3.39340866e-01 1.74147308e-01 -1.23253107e+00 5.45312345e-01
-1.47745121e+00 -8.44909489e-01 -2.46070862e-01 -1.53316155e-01
1.08797824e+00 -1.69482857e-01 6.72956526e-01 1.45183146e-01
-7.47011065e-01 7.14046419e-01 5.30780673e-01 -2.35873625e-01
1.24803162e+00 -1.58778095e+00 8.80887389e-01 9.85845864e-01
1.52932122e-01 5.38469911e-01 1.06327200e+00 -6.14757776e-01
-1.04769695e+00 -1.32212079e+00 1.27418888e+00 -4.52309519e-01
7.39316821e-01 -5.73587306e-02 -9.91089284e-01 7.01442659e-01
8.74989748e-01 -6.64574921e-01 6.46419227e-01 2.80136645e-01
-3.44641238e-01 5.43683693e-02 -6.91252589e-01 8.42896461e-01
1.20032406e+00 -2.30081603e-01 -9.81907308e-01 3.78080010e-01
1.10076642e+00 -2.17373356e-01 -8.43038917e-01 2.90715963e-01
3.04805338e-01 -9.42048371e-01 6.32938743e-01 -8.94313872e-01
1.09991932e+00 1.79502219e-01 1.55454904e-01 -1.72017908e+00
-3.33532989e-01 -1.04215598e+00 -2.52184570e-01 1.71054482e+00
6.08961701e-01 -4.17413116e-01 7.63585627e-01 2.74530381e-01
-6.54936552e-01 -5.09989381e-01 -5.40619612e-01 -7.69969523e-01
2.97460914e-01 1.01680383e-01 1.56535998e-01 7.81704783e-01
3.91721606e-01 1.01900113e+00 -2.93241799e-01 -4.81195718e-01
5.55677235e-01 2.18751013e-01 1.06481218e+00 -1.00382745e+00
-2.92732894e-01 -7.63744891e-01 9.53083634e-02 -1.20896149e+00
1.12686723e-01 -1.28051567e+00 5.64430021e-02 -2.01891184e+00
6.23165309e-01 6.71370551e-02 -2.38936692e-01 3.02785963e-01
-6.28599226e-01 1.30882844e-01 3.43039334e-01 3.01069170e-01
-1.19850612e+00 7.80604780e-01 9.62772787e-01 -2.42009729e-01
-2.60757238e-01 -1.01529555e-02 -1.17524099e+00 4.28684771e-01
9.55504298e-01 -3.42628479e-01 -3.67663860e-01 -7.61697650e-01
5.05514592e-02 1.71450868e-01 -1.52179971e-01 -9.60311472e-01
2.64959514e-01 2.08040744e-01 -3.10384035e-01 -7.05756545e-01
3.84639464e-02 2.61791319e-01 -2.47106910e-01 2.24000692e-01
-9.03142989e-01 -2.03920484e-01 1.02688439e-01 7.22450554e-01
-3.15213740e-01 -4.19515312e-01 5.16511619e-01 -5.61143178e-03
-5.17833412e-01 4.71271537e-02 -4.80938375e-01 4.71524060e-01
6.30747318e-01 -8.17441642e-02 -7.24383831e-01 -6.19973719e-01
-4.95611548e-01 4.51560944e-01 3.35519820e-01 3.59132916e-01
1.93004027e-01 -9.00391459e-01 -1.23061681e+00 -6.59660041e-01
3.88779610e-01 3.16392481e-02 3.53739381e-01 8.97684693e-01
-4.17275131e-01 6.32763624e-01 -9.87419039e-02 -4.50734288e-01
-1.28685236e+00 1.39224246e-01 -3.65348905e-01 -7.67588496e-01
-8.38744104e-01 7.75274932e-01 3.26671172e-04 -3.29886615e-01
1.22967422e-01 -2.45805383e-01 -2.25231677e-01 4.75632459e-01
5.98215401e-01 9.47648287e-01 4.52363878e-01 -3.19009751e-01
3.51534151e-02 7.90248439e-02 -5.57625592e-01 -7.34719485e-02
1.33781898e+00 -2.11243451e-01 -2.37514470e-02 5.86718798e-01
1.28156161e+00 9.27484185e-02 -9.63106453e-01 -4.57157761e-01
4.28915739e-01 -1.00343831e-01 -1.75539508e-01 -7.32911348e-01
-4.81591970e-01 7.13470757e-01 -4.25890833e-01 3.33205462e-01
8.07049990e-01 1.50784135e-01 9.20632660e-01 6.07375920e-01
7.28416964e-02 -1.14836979e+00 1.64422885e-01 8.29025865e-01
9.01050389e-01 -1.02530706e+00 3.62725019e-01 -3.14206988e-01
-9.24655676e-01 7.99504817e-01 2.44691968e-01 -3.03718269e-01
-1.78848386e-01 -9.92256254e-02 -4.09730375e-01 -4.39322256e-02
-1.20624232e+00 -9.88367796e-02 2.41769403e-01 2.95044929e-01
5.25529385e-01 1.64558172e-01 -6.26314580e-01 6.62018299e-01
-5.21701396e-01 -3.51217628e-01 9.12598133e-01 7.42063880e-01
-7.33379126e-01 -8.14479887e-01 2.35915780e-01 9.50295746e-01
-6.10389650e-01 -2.54350424e-01 -4.83780921e-01 6.57495916e-01
-9.11566973e-01 1.00201571e+00 1.64060414e-01 -1.76529318e-01
4.36833084e-01 1.49713561e-01 2.01477692e-01 -1.14478242e+00
-9.01770294e-01 1.11341223e-01 6.99387789e-01 -2.87567347e-01
-1.69577330e-01 -7.18771160e-01 -1.11028194e+00 -4.02668029e-01
-2.73954660e-01 4.53319579e-01 1.56729132e-01 7.22247541e-01
5.28030813e-01 8.17404270e-01 4.61347312e-01 -9.28448915e-01
-9.15922523e-01 -1.32421684e+00 -3.95101696e-01 3.01340729e-01
8.57587159e-02 -8.54744539e-02 -3.67530823e-01 1.38150603e-01] | [12.371068954467773, 9.394088745117188] |
8c73b27c-7f35-4131-a8dd-f9250c7f7470 | scene-flow-estimation-a-survey | 1612.02590 | null | http://arxiv.org/abs/1612.02590v3 | http://arxiv.org/pdf/1612.02590v3.pdf | Scene Flow Estimation: A Survey | This paper is the first to review the scene flow estimation field, which
analyzes and compares methods, technical challenges, evaluation methodologies
and performance of scene flow estimation. Existing algorithms are categorized
in terms of scene representation, data source, and calculation scheme, and the
pros and cons in each category are compared briefly. The datasets and
evaluation protocols are enumerated, and the performance of the most
representative methods is presented. A future vision is illustrated with few
questions arisen for discussion. This survey presents a general introduction
and analysis of scene flow estimation. | ['Xuezhi Xiang', 'Zike Yan'] | 2016-12-08 | null | null | null | null | ['scene-flow-estimation'] | ['computer-vision'] | [ 1.83203951e-01 -6.17966831e-01 -3.04790109e-01 -2.80705184e-01
2.43916765e-01 -6.50418222e-01 5.11522412e-01 1.08448096e-01
-2.99596727e-01 7.36201525e-01 4.15125668e-01 -2.28860602e-01
-2.89442122e-01 -8.10593724e-01 2.20264301e-01 -6.36859477e-01
-3.68134886e-01 -1.96234345e-01 5.45234382e-01 -3.91260535e-02
6.25276923e-01 8.99077535e-01 -1.79994059e+00 1.22711919e-02
8.18381548e-01 9.31548655e-01 1.80179060e-01 1.33487177e+00
-2.70077497e-01 1.49458170e+00 -1.06299949e+00 -2.95843363e-01
1.62528411e-01 -6.34316504e-01 -1.16997957e+00 2.70192176e-01
7.96986639e-01 -6.42433882e-01 -9.81151819e-01 5.10551691e-01
6.43783450e-01 5.34809351e-01 6.85874522e-01 -1.62930608e+00
2.03596056e-01 -1.52608026e-02 -1.38453975e-01 1.03815484e+00
6.35464132e-01 8.11770260e-02 2.88325250e-01 -8.31051707e-01
7.59109139e-01 8.96630764e-01 5.96029937e-01 4.39128190e-01
-6.68032706e-01 -1.59633607e-01 1.48262441e-01 4.17609006e-01
-1.18217695e+00 -4.63863641e-01 8.20832312e-01 -6.08405530e-01
1.06640422e+00 6.00944519e-01 1.09038079e+00 2.98660398e-01
1.53977975e-01 1.02082884e+00 5.75613916e-01 -3.75124812e-01
2.19814423e-02 1.48921892e-01 3.52213740e-01 8.15740824e-01
3.32522631e-01 -1.06024435e-02 -5.84160030e-01 1.22900568e-01
8.11642706e-01 -4.91585612e-01 -2.45353758e-01 -8.44370544e-01
-1.19697809e+00 6.22819126e-01 4.42739129e-01 4.76865411e-01
1.87207153e-03 7.46395662e-02 9.19769287e-01 -2.04413831e-01
6.44983649e-02 2.05349326e-01 6.46503940e-02 -4.16684687e-01
-9.61239278e-01 3.60976249e-01 6.57089353e-01 1.26292586e+00
5.11522532e-01 5.91774821e-01 -3.09903264e-01 7.22372174e-01
1.29272208e-01 5.03505290e-01 1.88502058e-01 -1.30873168e+00
4.47415590e-01 6.42003790e-02 -1.59513205e-01 -1.46703029e+00
-4.24225360e-01 1.48271903e-01 -5.93576372e-01 2.62707919e-01
1.80228487e-01 -2.67833173e-01 -4.47027713e-01 8.15345526e-01
2.20845014e-01 2.71951735e-01 7.42726400e-02 9.37048733e-01
1.64791107e+00 7.74482727e-01 2.55619913e-01 -5.02608061e-01
1.16954374e+00 -1.23903930e+00 -1.10027134e+00 -4.24055979e-02
4.32659984e-01 -1.06191945e+00 5.34195483e-01 -5.87063096e-02
-1.29619575e+00 -8.45159173e-01 -7.94275522e-01 -2.80015707e-01
-4.74096268e-01 -1.03282653e-01 1.20625710e+00 1.09534597e+00
-8.59451413e-01 2.45535851e-01 -7.78214633e-01 -9.66440141e-01
5.72093904e-01 -4.06126082e-02 -5.56477457e-02 6.06812015e-02
-8.62950742e-01 9.39761341e-01 3.49926889e-01 3.35009813e-01
-6.94693685e-01 -8.86807919e-01 -1.20953429e+00 -4.13608044e-01
-9.03072879e-02 -8.46591413e-01 1.31073654e+00 1.19203039e-01
-1.65583718e+00 7.35366702e-01 -4.97135729e-01 -2.41623864e-01
6.42263114e-01 4.86375168e-02 -6.14826202e-01 5.95720530e-01
1.06947616e-01 5.97775340e-01 2.42361769e-01 -1.06605256e+00
-1.28476238e+00 3.23002845e-01 3.27512920e-01 5.44394732e-01
2.23280221e-01 4.14443702e-01 -5.40306449e-01 -4.37635779e-01
-5.18990792e-02 -2.23011091e-01 -5.13714612e-01 2.39859551e-01
-2.72381425e-01 1.62005559e-01 1.28506243e+00 -9.99572966e-03
1.78706133e+00 -1.88366294e+00 -5.16332388e-01 9.97538716e-02
3.25499654e-01 3.11556935e-01 3.73998582e-01 6.89503253e-01
1.58804134e-01 -2.42992178e-01 -1.32997543e-01 -2.99707949e-01
-6.01478875e-01 8.26128498e-02 -5.66647947e-01 9.11027014e-01
-4.05070424e-01 7.32762992e-01 -1.20368648e+00 -1.03661776e+00
1.63365042e+00 4.44092423e-01 -2.68891364e-01 1.43241152e-01
7.48795211e-01 5.54286599e-01 -4.12113905e-01 7.78234005e-01
9.70841706e-01 4.27337810e-02 -1.26977235e-01 -5.46486020e-01
-8.28092337e-01 4.10083503e-01 -1.54044509e+00 1.53784120e+00
-4.96414244e-01 1.22513473e+00 -2.38319993e-01 -8.04944634e-01
8.08507979e-01 3.25934023e-01 1.13534796e+00 -4.65759903e-01
9.68630239e-02 1.90111130e-01 -2.37519681e-01 -7.70323873e-01
8.23701918e-01 3.46439004e-01 4.54778150e-02 1.50205612e-01
1.69841215e-01 -5.77234805e-01 1.08167851e+00 3.66543621e-01
8.74828875e-01 3.70315164e-02 1.00504005e+00 -3.91315430e-01
1.18451405e+00 5.36363304e-01 4.20986563e-01 9.42069888e-01
-1.16172838e+00 5.42460918e-01 2.62482733e-01 -9.96946216e-01
-5.38984716e-01 -1.28164721e+00 -5.04938602e-01 5.98058939e-01
1.12512958e+00 -7.12651014e-01 -4.48369205e-01 -6.71456218e-01
-1.76051632e-01 2.17041761e-01 -5.10334373e-01 4.08673674e-01
-8.14599633e-01 -7.41230667e-01 4.61140513e-01 6.95050895e-01
9.00699615e-01 -1.10679507e+00 -1.48126674e+00 1.04045883e-01
-4.59665537e-01 -1.48859251e+00 -7.48780072e-02 -4.85641062e-01
-1.15351915e+00 -1.38422227e+00 -5.79480886e-01 -7.92764544e-01
4.45112497e-01 8.89752865e-01 1.28213274e+00 2.16151178e-01
-7.54442155e-01 8.52542460e-01 -2.50243306e-01 -3.02442402e-01
3.47503386e-02 -1.11972034e-01 -3.68474796e-02 -3.72327089e-01
3.54502052e-01 -2.64909714e-01 -8.68290961e-01 2.93158770e-01
-4.58784521e-01 -1.14472717e-01 -8.90639275e-02 2.03936547e-01
2.19899148e-01 1.46158919e-01 -1.43942796e-02 -6.59524143e-01
3.58718663e-01 -1.48110643e-01 -7.67031193e-01 8.58085975e-03
-3.08046222e-01 -6.16103709e-01 4.28061455e-01 1.91444322e-01
-1.46455622e+00 -4.15997393e-03 3.67854163e-02 -1.10188097e-01
-5.54836154e-01 -3.75213057e-01 -3.44023877e-03 -4.91963446e-01
8.54455590e-01 1.07412010e-01 -2.06286922e-01 -1.00102894e-01
4.47372794e-01 1.76164687e-01 7.36852169e-01 -2.26176128e-01
6.07861578e-01 1.13088882e+00 1.96966946e-01 -1.46046650e+00
-6.48396552e-01 -1.22191322e+00 -1.08670175e+00 -1.14390087e+00
8.09659600e-01 -3.68653715e-01 -7.47437418e-01 6.91344500e-01
-1.27164459e+00 -1.10787675e-01 -6.30944312e-01 1.01208603e+00
-1.05183864e+00 4.35103416e-01 -7.50860214e-01 -7.53880382e-01
-1.13166444e-01 -1.36495006e+00 7.89717019e-01 4.58630234e-01
-1.44648537e-01 -1.54446650e+00 3.76811147e-01 2.01111406e-01
5.70033073e-01 5.04778445e-01 4.64760810e-02 3.91745299e-01
-5.89865923e-01 -5.41578680e-02 -3.23668718e-01 -1.73471272e-01
2.74584442e-01 5.16814590e-01 -1.21950781e+00 -1.31534031e-02
-2.65531957e-01 1.40600815e-01 4.11047965e-01 1.10117447e+00
1.23462200e+00 1.16946742e-01 -7.82360315e-01 9.30802524e-01
1.58934081e+00 4.27890420e-01 8.12051296e-01 2.81886131e-01
5.81490815e-01 7.75687635e-01 8.85392904e-01 3.13679457e-01
1.30567864e-01 4.54238713e-01 2.75938749e-01 -6.19199909e-02
-7.64581978e-01 -3.36628705e-02 -1.35480255e-01 8.78785849e-01
-5.00767231e-01 -5.04158080e-01 -6.80471063e-01 4.46015120e-01
-1.53863049e+00 -1.40882850e+00 -7.90779114e-01 1.78973329e+00
-1.19829990e-01 -1.83583260e-01 4.36135322e-01 4.73487884e-01
8.45166743e-01 7.18923509e-01 -5.02593741e-02 -4.20132816e-01
-1.87690556e-01 -2.01078340e-01 5.74724674e-01 8.97197008e-01
-1.62922728e+00 8.26535523e-01 8.72656345e+00 4.21521991e-01
-9.62301612e-01 -1.65143684e-01 5.66966012e-02 2.25665659e-01
5.16695976e-01 -3.37627158e-02 -8.33862245e-01 3.13379556e-01
5.42951345e-01 -4.34278429e-01 -8.43823031e-02 7.55641818e-01
2.81129688e-01 -6.02770507e-01 -5.85217237e-01 1.32073224e+00
3.04048330e-01 -1.68006301e+00 8.18238929e-02 -2.69512147e-01
4.38922375e-01 7.26510659e-02 -3.39160919e-01 -3.52549106e-02
-1.25052467e-01 -4.43612069e-01 5.85796833e-01 2.26114243e-01
8.13923120e-01 -4.44818735e-01 6.43742621e-01 -2.54643649e-01
-1.97241759e+00 -3.09368223e-03 -4.59135771e-01 -3.41056854e-01
9.86808479e-01 3.57025683e-01 -4.53494281e-01 8.71002495e-01
9.54568207e-01 1.22556984e+00 -5.80646813e-01 1.87562764e+00
2.41264910e-03 3.65397871e-01 -2.65123546e-01 -3.63886133e-02
9.53889042e-02 -2.91344225e-01 7.65596688e-01 1.82707691e+00
1.28223091e-01 1.16136856e-01 1.70964524e-01 4.14391965e-01
6.24426663e-01 2.22986221e-01 -9.62728024e-01 5.93964040e-01
4.93922055e-01 1.40666556e+00 -1.10873568e+00 -6.03439450e-01
-6.88889802e-01 5.57379305e-01 -1.99708909e-01 3.92888129e-01
-9.09272671e-01 -9.79566097e-01 8.27971756e-01 8.63466561e-02
-3.92769367e-01 -2.78411269e-01 -4.98447627e-01 -9.99581397e-01
-3.65136951e-01 1.00596465e-01 8.86373639e-01 -6.50122344e-01
-8.19779992e-01 5.19516110e-01 7.64912546e-01 -1.90411747e+00
-8.02307650e-02 -7.41707206e-01 -6.08407080e-01 5.23468852e-01
-1.57550824e+00 -5.69962442e-01 -1.13836586e+00 5.66532850e-01
8.32833409e-01 -9.52090174e-02 4.48257715e-01 5.54568350e-01
-6.11509085e-01 1.36355117e-01 -2.40555733e-01 3.66164774e-01
4.43531394e-01 -9.57119882e-01 1.89073533e-01 1.09142935e+00
-2.53811419e-01 2.16278657e-01 7.83140957e-01 -3.44072849e-01
-9.97627497e-01 -1.03367460e+00 9.34503973e-01 -5.96399367e-01
4.14017230e-01 -2.01854438e-01 -4.21514869e-01 5.32842398e-01
3.42876405e-01 4.57681239e-01 5.22011936e-01 -4.34883744e-01
2.21093595e-01 -9.19914320e-02 -1.33547807e+00 6.25590205e-01
1.17671406e+00 -6.88799918e-02 -2.47524679e-01 2.10932851e-01
1.36603415e-01 -8.36628258e-01 -7.26201296e-01 4.36647922e-01
6.00117147e-01 -1.40731561e+00 1.23036122e+00 -2.48756945e-01
-7.06740539e-04 -5.83040535e-01 5.30117527e-02 -6.91484809e-01
-3.68423045e-01 -5.29302061e-01 -4.74981517e-01 7.57248044e-01
-3.14395547e-01 -4.45146888e-01 9.20877516e-01 1.87811837e-01
-3.26951623e-01 -1.00297607e-01 -6.40326023e-01 -6.97279334e-01
-3.82863104e-01 -7.75237560e-01 3.10331851e-01 7.92236328e-01
3.27238798e-01 3.53885233e-01 -9.51141790e-02 -6.62511215e-02
9.06812251e-01 2.42795110e-01 9.99927938e-01 -8.77627671e-01
6.20535374e-01 -7.63318956e-01 -1.05656755e+00 -1.38860524e+00
-8.04851130e-02 -4.98242915e-01 -1.33524045e-01 -2.05114388e+00
-1.27647281e-01 -3.55413228e-01 2.76542991e-01 -2.50035644e-01
-3.72147770e-03 3.20260316e-01 4.52947557e-01 1.57572970e-01
-6.50302887e-01 9.18697640e-02 1.48975384e+00 6.46672770e-02
-2.76455671e-01 2.67006606e-01 -1.37142792e-01 7.19013929e-01
9.72810984e-01 -1.74576789e-02 -8.61586094e-01 -4.12138611e-01
-4.87011790e-01 -1.02922529e-01 1.87905431e-01 -1.50061798e+00
4.88502085e-01 -3.77207786e-01 2.35805094e-01 -1.19863784e+00
3.96023422e-01 -9.44311500e-01 -2.03418940e-01 7.72436619e-01
-8.78264979e-02 2.02337518e-01 2.50667363e-01 2.37379387e-01
-5.08248210e-01 -3.86499256e-01 1.13121891e+00 -9.50253531e-02
-1.52557230e+00 4.22291368e-01 -8.85079563e-01 9.89389345e-02
1.51982141e+00 -8.74399245e-01 -3.45451206e-01 -3.41808677e-01
-3.79103512e-01 9.57412869e-02 -7.85588697e-02 4.59275395e-01
6.44762158e-01 -1.46364450e+00 -4.75272685e-01 2.91492820e-01
2.05642298e-01 -4.41238396e-02 7.06413686e-01 8.19959998e-01
-1.39270616e+00 8.79422367e-01 -3.40762138e-01 -9.59615707e-01
-1.20011342e+00 5.59431374e-01 7.87967384e-01 9.01397094e-02
-6.75345123e-01 5.68714201e-01 3.92784804e-01 -3.31932008e-01
7.84113333e-02 -3.01218450e-01 -7.23864317e-01 -2.89571911e-01
7.12987840e-01 1.30758679e+00 9.82699096e-02 -9.43244815e-01
-6.53722644e-01 8.94370377e-01 5.94509542e-01 2.59378582e-01
3.98284763e-01 -7.29672253e-01 -2.01659240e-02 4.24126595e-01
1.10837221e+00 -1.58244908e-01 -9.77314293e-01 1.69648230e-01
-5.23133993e-01 -1.08796477e+00 -1.31829873e-01 -2.55473763e-01
-1.21438801e+00 1.07211995e+00 4.58603680e-01 3.21488768e-01
1.32529426e+00 5.06307296e-02 5.64130723e-01 -8.49856809e-02
4.31503743e-01 -9.21911716e-01 1.83585789e-02 7.88123965e-01
4.92196977e-01 -9.23194170e-01 2.06050977e-01 -1.15048289e+00
-5.08321643e-01 1.18670595e+00 8.84498179e-01 -2.75102764e-01
1.14545369e+00 6.92348063e-01 2.38603100e-01 -3.20182651e-01
-2.53726363e-01 -2.26838455e-01 1.52903348e-01 1.29743481e+00
6.25194252e-01 -3.55152637e-01 -2.20404699e-01 -6.94451809e-01
-4.42263991e-01 -3.22176516e-02 6.04336679e-01 1.32033479e+00
-3.87378514e-01 -6.03395820e-01 -5.01273572e-01 3.16463053e-01
-2.42137313e-01 4.07626837e-01 1.81616917e-01 1.05026865e+00
-5.84489889e-02 1.16529596e+00 5.14334142e-01 -4.76011075e-02
8.16846907e-01 -7.03909695e-01 5.55265963e-01 -2.35676229e-01
-3.41175616e-01 -3.68911952e-01 5.25036827e-02 -9.61004496e-01
-1.12297368e+00 -6.12145483e-01 -1.04660118e+00 -8.03882062e-01
5.90000600e-02 2.68715888e-01 3.94020319e-01 3.34871471e-01
8.26238319e-02 7.01441348e-01 5.72888136e-01 -9.30140316e-01
6.58795953e-01 -3.91698092e-01 -5.48506677e-01 1.98307127e-01
6.29529595e-01 -8.58205020e-01 -5.27533770e-01 5.24128735e-01] | [8.707001686096191, -1.815508246421814] |
eca67481-8e79-499c-8311-ab231291db1e | learning-environment-models-with-continuous | 2306.17204 | null | https://arxiv.org/abs/2306.17204v1 | https://arxiv.org/pdf/2306.17204v1.pdf | Learning Environment Models with Continuous Stochastic Dynamics | Solving control tasks in complex environments automatically through learning offers great potential. While contemporary techniques from deep reinforcement learning (DRL) provide effective solutions, their decision-making is not transparent. We aim to provide insights into the decisions faced by the agent by learning an automaton model of environmental behavior under the control of an agent. However, for most control problems, automata learning is not scalable enough to learn a useful model. In this work, we raise the capabilities of automata learning such that it is possible to learn models for environments that have complex and continuous dynamics. The core of the scalability of our method lies in the computation of an abstract state-space representation, by applying dimensionality reduction and clustering on the observed environmental state space. The stochastic transitions are learned via passive automata learning from observed interactions of the agent and the environment. In an iterative model-based RL process, we sample additional trajectories to learn an accurate environment model in the form of a discrete-state Markov decision process (MDP). We apply our automata learning framework on popular RL benchmarking environments in the OpenAI Gym, including LunarLander, CartPole, Mountain Car, and Acrobot. Our results show that the learned models are so precise that they enable the computation of policies solving the respective control tasks. Yet the models are more concise and more general than neural-network-based policies and by using MDPs we benefit from a wealth of tools available for analyzing them. When solving the task of LunarLander, the learned model even achieved similar or higher rewards than deep RL policies learned with stable-baselines3. | ['Bettina Könighofer', 'Bernhard K. Aichernig', 'Edi Muškardin', 'Martin Tappler'] | 2023-06-29 | null | null | null | null | ['dimensionality-reduction', 'benchmarking', 'openai-gym', 'acrobot', 'decision-making', 'benchmarking'] | ['methodology', 'miscellaneous', 'playing-games', 'playing-games', 'reasoning', 'robots'] | [-5.23914993e-02 1.24089740e-01 -1.83850378e-01 1.05487704e-01
-5.54844618e-01 -7.52500951e-01 9.07535553e-01 2.17278242e-01
-5.25258780e-01 9.32600021e-01 1.78983301e-01 -2.71452010e-01
-2.82203138e-01 -9.09943759e-01 -8.88875663e-01 -9.46294010e-01
-4.26634699e-01 9.13054645e-01 3.72437760e-02 -5.34940422e-01
-1.30470544e-01 5.83198130e-01 -1.68532753e+00 -1.77519202e-01
6.44779503e-01 6.37557745e-01 3.04452956e-01 9.31164742e-01
1.11124873e-01 1.04754019e+00 -2.50182569e-01 3.62216830e-01
3.69211674e-01 -5.81449032e-01 -7.00474083e-01 -2.61077695e-02
-6.94788992e-02 -4.27597761e-02 -3.80629718e-01 7.24807918e-01
3.08650315e-01 6.06573462e-01 6.21884704e-01 -1.01139796e+00
9.76656303e-02 6.67926610e-01 1.73906580e-01 -3.62300612e-02
1.60879806e-01 7.59534538e-01 8.98332238e-01 1.42823784e-02
5.86660504e-01 1.40334964e+00 4.20918643e-01 7.34634876e-01
-1.54318023e+00 -1.97620422e-01 4.49466705e-01 1.06741272e-01
-8.58343720e-01 -3.32824409e-01 3.45645994e-01 -3.12323719e-01
1.04289794e+00 1.80374354e-01 1.19682980e+00 1.44449937e+00
3.73856664e-01 9.21621144e-01 1.36288381e+00 -3.48257571e-01
8.56398404e-01 -1.21309087e-01 -3.93352136e-02 7.34262288e-01
-1.25426520e-02 7.51765072e-01 -2.36781925e-01 -3.14254850e-01
7.33575165e-01 6.30727559e-02 4.15523397e-03 -7.85002589e-01
-1.20144594e+00 8.19005787e-01 2.96989530e-01 1.03713371e-01
-4.38920796e-01 6.42202318e-01 3.73253167e-01 4.92866933e-01
-6.83542714e-02 7.54456758e-01 -5.27374387e-01 -5.24208963e-01
-4.19790357e-01 8.23496521e-01 1.22918117e+00 5.82524896e-01
8.17613423e-01 2.40031242e-01 -7.71118999e-02 1.87079102e-01
1.16024256e-01 7.33564496e-01 3.92075241e-01 -1.40106249e+00
1.75455697e-02 3.36200505e-01 6.74779594e-01 -6.32623851e-01
-5.79878926e-01 -4.24140066e-01 -5.72878838e-01 6.63393974e-01
6.41399980e-01 -3.59704584e-01 -7.08405137e-01 1.83523536e+00
3.33232343e-01 2.18753010e-01 4.07473892e-01 7.49777973e-01
-3.26738507e-01 7.34759152e-01 5.87388203e-02 -2.51650661e-01
8.28159690e-01 -9.47753549e-01 -6.16773129e-01 -3.34276170e-01
7.72149265e-01 2.40108371e-02 1.28109205e+00 5.76958120e-01
-9.49933648e-01 -4.58899021e-01 -9.28318501e-01 3.74538481e-01
-2.39540622e-01 -2.68794954e-01 7.62645245e-01 2.95531601e-01
-9.28685784e-01 9.06887650e-01 -1.51571417e+00 -4.38939184e-01
3.24842930e-02 4.31218207e-01 6.83203060e-03 2.05258325e-01
-1.08392942e+00 1.13208795e+00 4.11874682e-01 -1.07086442e-01
-1.77039492e+00 -1.83290452e-01 -8.80428493e-01 1.22100376e-01
8.45614791e-01 -5.99087119e-01 1.69436300e+00 -1.03675342e+00
-2.03106117e+00 1.28526688e-01 6.96182996e-02 -9.83779311e-01
5.83144188e-01 -1.41914815e-01 6.17309585e-02 -1.28275380e-01
-1.48897082e-01 4.27003473e-01 7.77119458e-01 -1.21329641e+00
-6.29996419e-01 -1.99261457e-01 2.32680231e-01 3.71882290e-01
1.17238306e-01 -5.37387431e-01 2.31647380e-02 -8.82453322e-02
-4.87526804e-01 -1.35141468e+00 -8.25842679e-01 -4.40364182e-01
-2.82360613e-02 -2.22331241e-01 6.13369882e-01 -2.15766951e-01
9.80534911e-01 -1.99974823e+00 5.25007844e-01 1.18742302e-01
-5.01652472e-02 1.22392975e-01 -1.17354393e-01 7.27175355e-01
3.55252087e-01 -2.00077981e-01 -3.21960390e-01 -2.73797125e-01
4.21219736e-01 8.55994582e-01 -4.81116682e-01 3.58652323e-01
1.16202785e-02 9.69544053e-01 -1.19020009e+00 -1.35841697e-01
2.97186255e-01 1.32075578e-01 -5.78984737e-01 1.90987825e-01
-1.02181327e+00 7.67585397e-01 -7.21703887e-01 1.53231576e-01
-1.52603760e-01 3.21317725e-02 4.92956698e-01 6.35195017e-01
-1.98587939e-01 4.36339289e-01 -1.33211493e+00 1.78663123e+00
-7.41286874e-01 2.79388458e-01 4.62864228e-02 -1.00889742e+00
7.87769198e-01 6.10498413e-02 5.65996170e-01 -7.29694486e-01
2.21692701e-03 2.65045427e-02 2.38755107e-01 -4.51716185e-01
2.67800391e-01 -1.13286875e-01 -4.04518545e-01 7.32467532e-01
1.09003246e-01 -5.40021658e-01 2.74353415e-01 7.18353665e-04
1.20156169e+00 7.11552203e-01 2.86845475e-01 -3.97182018e-01
2.23250672e-01 3.20669502e-01 5.02295494e-01 1.15037107e+00
-1.50611863e-01 -1.98420897e-01 6.46274924e-01 -8.51149380e-01
-9.91363525e-01 -1.02080226e+00 2.76664108e-01 9.31890309e-01
-9.11224484e-02 -5.82944095e-01 -8.35056186e-01 -5.45998693e-01
-4.29773889e-02 8.47284436e-01 -7.16709912e-01 -2.84785539e-01
-7.87601650e-01 -5.42227387e-01 2.14051396e-01 2.52588481e-01
2.65056998e-01 -1.34172416e+00 -1.16664219e+00 5.34464240e-01
7.95991719e-02 -8.64643216e-01 -1.95343122e-01 5.38841128e-01
-8.44184101e-01 -1.05176508e+00 -7.38473088e-02 -2.98037916e-01
1.81899875e-01 -3.25809151e-01 1.01184034e+00 -1.90968990e-01
-2.46270031e-01 8.53728890e-01 2.99887657e-02 -4.90294367e-01
-7.09147155e-01 2.45674372e-01 4.78853047e-01 -1.31615356e-01
4.96049691e-03 -6.31543815e-01 -2.36887068e-01 1.44325253e-02
-6.89788640e-01 -6.27572984e-02 4.07771111e-01 8.71523499e-01
7.22463131e-01 1.88572243e-01 3.52906793e-01 -5.96189678e-01
6.77891552e-01 -3.87974024e-01 -1.03882861e+00 9.21845809e-02
-6.07981801e-01 7.42355645e-01 8.68823171e-01 -5.73685110e-01
-9.77026343e-01 4.39373314e-01 8.77903998e-02 -1.40558153e-01
-4.47882354e-01 2.06184626e-01 -1.06452685e-02 2.91334629e-01
7.66220450e-01 4.51275259e-01 2.75992930e-01 -3.21435452e-01
5.65442264e-01 1.99222371e-01 2.94465065e-01 -1.13063693e+00
6.14937127e-01 3.80542576e-01 2.58755028e-01 -6.23614609e-01
-7.20057130e-01 2.80757248e-02 -5.05251348e-01 -1.54127985e-01
8.17649484e-01 -6.64312005e-01 -1.25408041e+00 2.54863083e-01
-5.56836247e-01 -1.22393143e+00 -9.09586310e-01 4.07997608e-01
-1.34255409e+00 -3.65351066e-02 -5.05590975e-01 -1.25485444e+00
1.38051659e-01 -1.24882770e+00 9.00560558e-01 1.85085207e-01
-8.66023600e-02 -1.09569168e+00 6.21351600e-01 -1.84480309e-01
3.69995624e-01 3.66000026e-01 9.05537426e-01 -5.69706917e-01
-7.17653751e-01 1.19880721e-01 7.25931048e-01 7.13483840e-02
1.65429339e-02 -1.91234291e-01 -7.05162525e-01 -3.62121493e-01
4.78556287e-03 -4.67450291e-01 6.69395983e-01 2.66193360e-01
1.06370223e+00 -5.40771127e-01 -3.27780873e-01 3.09045374e-01
1.27536321e+00 2.22851798e-01 3.50608110e-01 4.91768032e-01
3.12190533e-01 5.98170638e-01 5.39420307e-01 3.71281087e-01
5.12947202e-01 7.00954437e-01 6.30563140e-01 2.91712373e-01
2.88385779e-01 -5.70644617e-01 9.08841908e-01 4.10291851e-01
-1.45598426e-01 1.14083432e-01 -1.04377675e+00 3.71075153e-01
-2.25986314e+00 -1.09654403e+00 1.61220595e-01 2.31474710e+00
7.74039626e-01 1.53597996e-01 4.98583972e-01 -1.38390854e-01
-3.38142812e-02 1.10210992e-01 -9.11304474e-01 -4.82609093e-01
7.04923719e-02 1.86401397e-01 4.13641036e-01 8.06099713e-01
-9.12801147e-01 1.17610812e+00 6.53499603e+00 4.84374911e-01
-9.35133576e-01 -5.32774441e-02 4.87037003e-01 -3.75071734e-01
-8.16054456e-03 1.53680757e-01 -8.64823937e-01 3.14192444e-01
1.52420068e+00 7.30649829e-02 1.13986921e+00 8.79759967e-01
5.84313989e-01 -3.43237877e-01 -1.26633704e+00 6.41376019e-01
-4.52871889e-01 -1.34667802e+00 -3.28327090e-01 3.51408005e-01
6.99949861e-01 2.82629162e-01 -6.23353198e-02 8.41957688e-01
1.22695410e+00 -1.04006803e+00 7.02517867e-01 7.93775916e-01
9.78964567e-02 -9.03312504e-01 3.39348853e-01 9.32525754e-01
-7.64857173e-01 -6.05447888e-01 -1.87339664e-01 -5.57970166e-01
-1.05895840e-01 8.94176192e-04 -7.20321655e-01 2.61892378e-01
5.26241958e-01 7.42819071e-01 -2.74157643e-01 5.45027733e-01
-2.10450932e-01 7.63908803e-01 -5.51339447e-01 -4.17811930e-01
6.35646224e-01 -6.18976593e-01 5.26092410e-01 8.04137409e-01
3.26038226e-02 -3.95519063e-02 5.89317799e-01 9.20825839e-01
3.76174510e-01 -2.06258371e-01 -9.68671203e-01 -6.30317181e-02
3.35289841e-03 1.13071167e+00 -5.78139901e-01 -2.55985737e-01
-1.60425119e-02 6.20004058e-01 6.29705608e-01 4.20780659e-01
-5.31555355e-01 2.06484109e-01 9.74457741e-01 -3.15146446e-02
1.87982798e-01 -6.96867228e-01 1.03832945e-01 -1.14148760e+00
-3.53971153e-01 -1.26822317e+00 1.17269829e-01 -3.73730272e-01
-7.88073897e-01 2.59306490e-01 1.08874694e-01 -8.30771148e-01
-9.13257122e-01 -5.11254907e-01 -4.62343454e-01 4.83445466e-01
-1.18421555e+00 -7.31830001e-01 3.60319875e-02 4.94303703e-01
5.93592942e-01 -1.43240780e-01 1.10505509e+00 -4.66553360e-01
-5.72357535e-01 -1.45339131e-01 4.14329827e-01 -1.73929140e-01
1.83970422e-01 -1.72137618e+00 3.47640067e-01 6.02664769e-01
3.68171304e-01 2.93248743e-01 8.76586318e-01 -4.55719560e-01
-1.89715540e+00 -8.80190372e-01 1.44924760e-01 -7.57285714e-01
8.59490395e-01 -6.64437413e-01 -6.98020935e-01 8.45152199e-01
2.09687948e-01 1.37582077e-02 1.67348191e-01 -8.11706632e-02
1.56297535e-01 -1.07947774e-02 -7.68139184e-01 8.11057448e-01
9.84201252e-01 -3.29478621e-01 -4.40314710e-01 3.32438648e-01
6.00930154e-01 -3.83737087e-01 -6.64921403e-01 -7.44264945e-02
3.42981875e-01 -7.22007275e-01 7.54911005e-01 -1.01929247e+00
7.73295313e-02 -3.40419590e-01 -1.00900903e-01 -1.67786944e+00
-7.79081136e-02 -1.01670611e+00 -5.38605630e-01 5.93310833e-01
2.96754450e-01 -5.93956470e-01 5.50395846e-01 4.40569490e-01
2.12763362e-02 -7.25008547e-01 -7.98354983e-01 -7.46171236e-01
3.28101307e-01 -4.96540129e-01 5.99160612e-01 5.56097686e-01
-1.96038075e-02 2.80527651e-01 -3.30128998e-01 1.14661798e-01
7.37104356e-01 2.29863524e-01 9.98968005e-01 -1.10273182e+00
-8.37789237e-01 -3.05996120e-01 6.28137449e-03 -1.00462174e+00
6.54918551e-01 -7.11670756e-01 2.40946442e-01 -1.46493661e+00
-1.70500711e-01 -6.34003282e-01 -2.20676512e-01 4.41688925e-01
1.55644745e-01 -6.14314675e-01 3.20871413e-01 1.34241194e-01
-7.88154960e-01 7.55656600e-01 1.16207767e+00 -6.91317543e-02
-5.00405252e-01 2.96343416e-01 -1.58553839e-01 7.52084255e-01
9.88053799e-01 -3.89987588e-01 -5.04239738e-01 -2.38359600e-01
4.29987282e-01 2.79506505e-01 3.56197834e-01 -1.02670002e+00
1.21957228e-01 -5.56289077e-01 6.06526211e-02 -1.03305772e-01
4.10329998e-01 -7.25276828e-01 6.93193451e-02 1.02894306e+00
-5.83010316e-01 9.55648813e-03 8.91582891e-02 9.45350528e-01
1.29345089e-01 -2.91154590e-02 5.43115735e-01 -4.99095529e-01
-7.11618483e-01 2.82305837e-01 -9.30614114e-01 4.60088141e-02
1.04108667e+00 3.78941506e-01 1.20183721e-01 -6.15512967e-01
-1.21416962e+00 4.90118206e-01 5.17193675e-01 3.05646747e-01
2.24569246e-01 -1.05456758e+00 -3.94898653e-01 1.33150071e-01
-1.00596927e-01 1.35295749e-01 -1.54560998e-01 6.32361293e-01
-2.95071900e-01 2.60485083e-01 -3.09547007e-01 -5.33341765e-01
-6.34111047e-01 6.07730389e-01 9.64219272e-01 -6.87975645e-01
-7.16232777e-01 -3.78695354e-02 -7.24078938e-02 -7.53139377e-01
2.83262432e-01 -6.88878655e-01 -4.03530449e-02 -2.16193646e-01
4.52411741e-01 2.00354367e-01 -2.40735412e-01 -1.39633149e-01
1.42814115e-01 1.66291028e-01 3.39600414e-01 -4.55224752e-01
1.49222255e+00 1.90880373e-02 1.73536733e-01 8.54287922e-01
3.85004133e-01 -3.30479056e-01 -1.91516244e+00 -1.49244115e-01
2.63460457e-01 9.14315209e-02 -5.15687000e-03 -7.25657821e-01
-4.20940161e-01 9.30786490e-01 6.27095222e-01 2.97385633e-01
7.12197661e-01 -9.51373354e-02 4.85172749e-01 1.03018892e+00
7.87896454e-01 -1.28810477e+00 1.45750642e-01 8.96646202e-01
6.45282745e-01 -1.09372139e+00 -4.30260986e-01 4.95383650e-01
-6.97178543e-01 1.06080794e+00 5.40489316e-01 -4.14294153e-01
3.42986017e-01 5.11704803e-01 -7.02910796e-02 4.13074866e-02
-1.22171175e+00 -4.97789562e-01 -2.28971809e-01 7.94504166e-01
-2.66307890e-01 3.41055632e-01 2.04160005e-01 3.41889769e-01
-2.83219725e-01 -2.02397734e-01 7.21181989e-01 1.00013709e+00
-6.75922692e-01 -1.21570110e+00 -2.76091635e-01 1.61242560e-01
-1.04859449e-01 3.26938480e-01 -2.76227355e-01 9.32679176e-01
-1.56621382e-01 6.79013133e-01 2.36844514e-02 1.49390716e-02
1.71727866e-01 2.21667811e-01 6.05495334e-01 -6.46498382e-01
-5.64004123e-01 -4.45658714e-02 3.33210826e-02 -9.24809098e-01
-2.37024456e-01 -9.97874618e-01 -1.45498359e+00 -1.10625409e-01
2.16949135e-01 3.55621994e-01 7.51972795e-01 1.09062672e+00
3.10468674e-01 5.10169208e-01 5.63906550e-01 -1.06002712e+00
-1.09615505e+00 -8.07558656e-01 -4.59595203e-01 1.55731857e-01
6.08068168e-01 -6.16146207e-01 -1.91113368e-01 -1.02709725e-01] | [4.149211406707764, 1.7296420335769653] |
7d7b1962-7358-47dd-b0f0-48b3ea5b515a | leveraging-commonsense-knowledge-on | 2108.03731 | null | https://arxiv.org/abs/2108.03731v1 | https://arxiv.org/pdf/2108.03731v1.pdf | Leveraging Commonsense Knowledge on Classifying False News and Determining Checkworthiness of Claims | Widespread and rapid dissemination of false news has made fact-checking an indispensable requirement. Given its time-consuming and labor-intensive nature, the task calls for an automated support to meet the demand. In this paper, we propose to leverage commonsense knowledge for the tasks of false news classification and check-worthy claim detection. Arguing that commonsense knowledge is a factor in human believability, we fine-tune the BERT language model with a commonsense question answering task and the aforementioned tasks in a multi-task learning environment. For predicting fine-grained false news types, we compare the proposed fine-tuned model's performance with the false news classification models on a public dataset as well as a newly collected dataset. We compare the model's performance with the single-task BERT model and a state-of-the-art check-worthy claim detection tool to evaluate the check-worthy claim detection. Our experimental analysis demonstrates that commonsense knowledge can improve performance in both tasks. | ['Zeyd Boukhers', 'Oul Han', 'Selma Tekir', 'Erhan Sezerer', 'Ipek Baris Schlicht'] | 2021-08-08 | null | null | null | null | ['news-classification'] | ['natural-language-processing'] | [ 1.43818960e-01 3.88113782e-02 -3.57332766e-01 -1.32220894e-01
-1.13891375e+00 -3.39023411e-01 8.02679300e-01 5.15940011e-01
-3.97544295e-01 8.25854897e-01 3.37872386e-01 -5.29857457e-01
-5.67921028e-02 -7.94501662e-01 -6.32077992e-01 4.43501547e-02
4.27333504e-01 3.60466003e-01 5.59165239e-01 -6.38897538e-01
9.04623508e-01 -8.81623328e-02 -1.38417399e+00 8.88235748e-01
1.26634717e+00 1.19215536e+00 -2.01323643e-01 4.91728157e-01
-1.40528247e-01 1.85502315e+00 -9.05666471e-01 -1.18018997e+00
-1.38303265e-01 -4.59545970e-01 -1.10657907e+00 -2.27022961e-01
5.15931308e-01 -2.50382423e-02 1.87539123e-02 1.31485581e+00
2.83242762e-01 -1.46639079e-01 6.12694979e-01 -1.35140228e+00
-1.39836574e+00 9.09506381e-01 -4.89010900e-01 6.83161616e-01
3.90112877e-01 -4.90057655e-02 1.10323822e+00 -6.36335135e-01
4.54105407e-01 1.01889467e+00 8.57396126e-01 3.99604976e-01
-7.73338258e-01 -8.29877317e-01 -2.55022347e-01 6.69513464e-01
-1.03971517e+00 -5.34947276e-01 1.26173651e+00 -6.58816397e-01
6.90520644e-01 4.71585363e-01 4.41831797e-01 1.35832620e+00
2.47993514e-01 9.49574828e-01 1.77311718e+00 -5.64413786e-01
2.84779727e-01 3.93177062e-01 5.23787916e-01 8.89432132e-01
4.91029590e-01 -2.86328405e-01 -8.60672176e-01 -5.49071670e-01
1.22000374e-01 -9.81214419e-02 -1.89973176e-01 5.38061619e-01
-1.11966729e+00 1.18808663e+00 3.37156743e-01 6.27228796e-01
-5.21296501e-01 1.92208678e-01 7.17343211e-01 4.27901566e-01
1.08686340e+00 6.97921157e-01 -3.84421319e-01 -3.75609189e-01
-1.29392326e+00 3.90824407e-01 9.03248250e-01 4.11368370e-01
1.34581346e-02 -2.42340505e-01 -5.86270213e-01 8.33120227e-01
1.07062692e-02 3.14869672e-01 6.59437776e-01 -7.08154142e-01
3.77808452e-01 8.82724762e-01 2.03540057e-01 -1.64916563e+00
-2.21277326e-01 -9.93870854e-01 -6.66802466e-01 -1.91322014e-01
3.82599771e-01 3.20946783e-01 -2.67563254e-01 1.55528998e+00
3.13631594e-01 5.40544428e-02 -2.24003851e-01 8.83690298e-01
7.39205539e-01 2.24494174e-01 1.91912934e-01 -2.92566150e-01
1.74152648e+00 -8.37786019e-01 -8.38397026e-01 -1.84727877e-01
6.83592498e-01 -7.79737532e-01 1.62765944e+00 5.08126438e-01
-7.36800373e-01 -2.31089756e-01 -1.05372548e+00 -3.92970294e-01
-3.09586287e-01 2.23770291e-01 6.38049662e-01 6.88331306e-01
-2.79711396e-01 3.25502753e-01 -2.08592206e-01 1.33715812e-02
8.15406442e-01 -6.81997299e-01 1.06635176e-01 1.31061405e-01
-1.55891645e+00 1.23212457e+00 3.66993040e-01 -3.24272275e-01
-7.06214547e-01 -6.33256972e-01 -2.76538759e-01 -4.54400927e-02
8.68994236e-01 -5.15840232e-01 1.07519627e+00 -8.33791316e-01
-1.08129382e+00 1.41573501e+00 1.40129983e-01 -6.80559337e-01
8.08596492e-01 -3.94267708e-01 -7.28935659e-01 4.00917195e-02
6.13714993e-01 -3.24334651e-01 9.09044325e-01 -1.09875071e+00
-5.08513510e-01 -1.87538102e-01 2.32339561e-01 -5.01814365e-01
-3.74406904e-01 3.35335463e-01 7.13388562e-01 -9.10637200e-01
-1.22357756e-01 -4.82533723e-01 5.13861775e-01 -1.27823159e-01
-7.55002141e-01 -5.68172216e-01 7.01186776e-01 -7.86780596e-01
1.43907177e+00 -1.90034819e+00 -6.60928071e-01 -2.10626081e-01
6.33803189e-01 1.22018315e-01 2.34230235e-01 2.05706224e-01
2.43624374e-01 3.27554494e-01 -6.81790039e-02 -7.64962956e-02
8.02204311e-02 -1.83339864e-01 -6.34880126e-01 4.10719603e-01
1.76614732e-01 1.04739928e+00 -1.14406681e+00 -8.17053378e-01
-4.56443220e-01 3.24204788e-02 -4.54435080e-01 -1.69720098e-01
-4.95986640e-01 -6.34219572e-02 -6.54635012e-01 7.66296685e-01
4.02850747e-01 -7.94662833e-01 -2.26678610e-01 -2.05742493e-01
6.08154722e-02 8.05931389e-01 -4.56246912e-01 1.15740073e+00
-2.18034491e-01 5.68927646e-01 -3.40578765e-01 -9.68687296e-01
9.37931359e-01 4.71935719e-02 -1.34037077e-01 -9.00921106e-01
2.79065698e-01 3.00080121e-01 -1.59875620e-02 -8.07896614e-01
6.35246217e-01 -8.69007528e-01 -2.91907221e-01 7.11975694e-01
-3.44852537e-01 2.12986954e-02 3.38501632e-02 6.36304796e-01
1.24690759e+00 -2.22707361e-01 5.04489005e-01 -4.86128241e-01
6.00346863e-01 5.45647144e-01 3.40561867e-01 8.70074689e-01
-5.42045355e-01 -2.54193783e-01 6.03804171e-01 -5.38354099e-01
-6.98806584e-01 -6.69943213e-01 -6.03329483e-03 1.43689275e+00
1.30745098e-01 -2.82413810e-01 -4.16787148e-01 -9.57314372e-01
1.52808845e-01 1.44504046e+00 -9.40052688e-01 -2.15929329e-01
-8.20407569e-02 -5.62295973e-01 1.01283479e+00 -3.49187013e-03
8.19040060e-01 -8.71542454e-01 -9.48233128e-01 1.42606661e-01
-8.54644477e-01 -1.10234272e+00 -9.02986974e-02 -1.20893098e-01
-2.73254842e-01 -1.43168819e+00 -1.01461172e-01 -4.24918592e-01
-9.37136821e-03 2.32566074e-01 1.39795351e+00 4.79573458e-01
-1.36111125e-01 -1.15638427e-01 -7.05143631e-01 -6.34331942e-01
-8.87741923e-01 -1.62549585e-01 -1.66309640e-01 4.29826006e-02
5.33248305e-01 -2.88610399e-01 -2.53445029e-01 7.02678561e-02
-9.36347663e-01 1.46568090e-01 3.43566984e-01 8.53952229e-01
1.25215694e-01 1.18818410e-01 1.21266055e+00 -9.89718497e-01
1.27179551e+00 -7.46345997e-01 -1.22466214e-01 4.77078617e-01
-7.18793571e-01 -4.96442104e-03 4.47070450e-01 -5.51789343e-01
-9.75845933e-01 -9.32266593e-01 1.26323029e-01 1.99750792e-02
3.66987288e-01 7.41210639e-01 4.00203466e-01 2.14477420e-01
1.28437579e+00 2.59902418e-01 -2.12717608e-01 -2.71790802e-01
4.26403672e-01 9.23150778e-01 5.14747560e-01 -6.68491066e-01
7.20087349e-01 4.85178888e-01 -3.44072461e-01 -2.84595728e-01
-2.02011251e+00 -4.06485587e-01 -1.50768295e-01 -3.30213428e-01
6.73729658e-01 -7.10599959e-01 -9.26821768e-01 9.92751941e-02
-1.53786147e+00 5.63873835e-02 -1.50936142e-01 1.70833200e-01
-4.73045915e-01 4.31072265e-01 -8.27869713e-01 -1.08520639e+00
-6.22202039e-01 -4.92488384e-01 7.74821997e-01 -3.53242725e-01
-4.25219715e-01 -8.44531178e-01 1.34956211e-01 1.15950799e+00
6.09028280e-01 4.48299766e-01 1.10246181e+00 -1.13528168e+00
-1.58591732e-01 -3.25357765e-01 -5.24875045e-01 3.47341657e-01
-1.02072306e-01 -4.20340925e-01 -1.03710389e+00 4.02483523e-01
5.87423027e-01 -1.09685767e+00 1.07649374e+00 -1.37892678e-01
9.43232417e-01 -8.89930785e-01 5.24741272e-03 -2.38027081e-01
1.25784731e+00 -4.57481772e-01 4.59740460e-01 6.49524927e-01
2.40671709e-01 3.66549850e-01 6.72410131e-01 7.39755273e-01
6.46400034e-01 2.77398497e-01 2.90512770e-01 4.10395861e-01
-9.40428525e-02 -4.85437900e-01 5.35857230e-02 6.72271729e-01
-2.29852960e-01 -9.68059376e-02 -8.95849943e-01 4.71902013e-01
-1.82850158e+00 -1.75280237e+00 -5.33996582e-01 1.48780167e+00
1.33360624e+00 4.13577735e-01 6.44894540e-02 5.29816449e-01
6.43059731e-01 2.75723606e-01 -1.27828076e-01 -3.50788146e-01
-2.26128370e-01 -1.78444460e-01 -1.55393118e-02 3.51377010e-01
-9.72310722e-01 9.81857657e-01 5.46003342e+00 1.33236468e+00
-9.49024618e-01 9.04836357e-01 6.56191409e-01 3.89531329e-02
-4.45462853e-01 -1.72894612e-01 -2.07748681e-01 6.18633568e-01
6.75305307e-01 -2.58362353e-01 2.15534493e-01 9.87845182e-01
8.13358650e-02 -1.94237515e-01 -8.39891315e-01 9.29031014e-01
6.15162373e-01 -1.68279815e+00 4.52414453e-02 -1.99220270e-01
6.23773634e-01 -9.70151871e-02 -1.82568684e-01 7.58877873e-01
2.58819848e-01 -7.08734691e-01 1.28829670e+00 5.31711578e-01
4.20284092e-01 -3.57503384e-01 8.23052645e-01 9.69869018e-01
-2.28493124e-01 -1.04905941e-01 -2.38139972e-01 -5.11289954e-01
2.11236000e-01 1.32432485e+00 -7.09771872e-01 8.65122154e-02
4.23269540e-01 3.05824161e-01 -7.84669638e-01 4.82342243e-01
-3.70090961e-01 1.00219071e+00 7.44581297e-02 -4.30635065e-01
-1.04969554e-02 4.92903352e-01 5.24641156e-01 1.24337113e+00
2.46050078e-02 1.75664723e-01 2.49857783e-01 1.41249275e+00
-5.80484807e-01 1.86998606e-01 -2.52870381e-01 -3.62133503e-01
4.39918488e-01 1.06909490e+00 -6.17006481e-01 -7.90133834e-01
-1.53447781e-02 1.02101636e+00 6.70874715e-01 -2.32427374e-01
-1.21988595e+00 -1.14231110e-01 -1.45347230e-02 3.17857385e-01
-9.28224325e-02 1.86642140e-01 -7.81107306e-01 -1.24148512e+00
1.39361918e-01 -9.37850237e-01 4.41034973e-01 -8.52710843e-01
-1.90204608e+00 5.79802871e-01 -2.11677089e-01 -8.47788274e-01
1.35025680e-01 -4.15361375e-01 -4.98732060e-01 4.75793719e-01
-1.82703769e+00 -1.41010988e+00 -2.30769753e-01 5.23554742e-01
4.08730596e-01 -1.22942999e-01 5.98231018e-01 1.50255486e-01
-5.58191426e-02 4.86787409e-01 -5.00198662e-01 2.26337627e-01
7.37548649e-01 -7.72750914e-01 2.08729357e-02 7.18967617e-01
1.18439846e-01 6.24435723e-01 7.93485403e-01 -1.08164215e+00
-8.94787073e-01 -7.94026077e-01 1.39790273e+00 -7.95659423e-01
1.34805834e+00 -1.23167545e-01 -1.00202477e+00 3.20636898e-01
-1.20703354e-01 -5.40784374e-03 1.03529155e+00 3.69033694e-01
-1.18530309e+00 4.00421560e-01 -1.41493714e+00 1.60160094e-01
1.04733145e+00 -8.97786498e-01 -1.23565769e+00 7.42562711e-01
7.06999183e-01 5.73331825e-02 -4.51603711e-01 4.48206328e-02
4.56061095e-01 -1.00762618e+00 5.61080933e-01 -9.22362804e-01
1.06727624e+00 -9.14161503e-02 -3.37686539e-01 -1.07098699e+00
-5.95501482e-01 -3.89177382e-01 -2.02094913e-01 1.14129424e+00
4.79742706e-01 -5.02500832e-01 3.09856445e-01 4.97713417e-01
-8.04053433e-03 -6.06890082e-01 -1.14987493e+00 -7.58772969e-01
-1.38040990e-01 -7.42382884e-01 2.15507284e-01 1.65807414e+00
2.95201004e-01 8.34320664e-01 -4.07153726e-01 -1.17350474e-01
6.06959581e-01 3.49223733e-01 4.48430270e-01 -1.46890271e+00
-4.09136772e-01 -4.45605725e-01 -3.94649297e-01 -5.05741358e-01
1.90914288e-01 -1.02344942e+00 -1.29162326e-01 -1.27912307e+00
8.26648951e-01 -4.48755205e-01 -2.61627048e-01 5.31501114e-01
-5.19997895e-01 3.41442257e-01 1.41849905e-01 4.35654938e-01
-8.81443262e-01 4.72092777e-01 1.18786800e+00 -4.00468677e-01
1.81076467e-01 -2.47940376e-01 -1.18126142e+00 9.06073630e-01
8.15715849e-01 -7.65486360e-01 -2.45989770e-01 -3.06969136e-01
1.04274702e+00 -2.87394762e-01 9.60069895e-01 -7.65938818e-01
1.38733193e-01 -4.28853631e-01 -2.75402348e-02 -3.73774052e-01
1.46912068e-01 -3.72153282e-01 -4.46022153e-01 4.90548223e-01
-7.57676184e-01 -4.20699567e-02 -1.99427336e-01 8.00192058e-01
8.79424624e-03 -1.38347805e-01 7.49983907e-01 -1.45294473e-01
-3.37112397e-01 -4.71179217e-01 -1.92629620e-02 7.91692615e-01
8.16873133e-01 -8.67369398e-02 -1.16265273e+00 -3.14025134e-01
-2.55914122e-01 -4.40378249e-01 1.63165241e-01 3.90523404e-01
7.08940983e-01 -1.08167171e+00 -1.23575294e+00 -3.73218477e-01
4.07900184e-01 -8.09416473e-01 1.35796130e-01 1.19036579e+00
-2.40480408e-01 8.74787942e-02 2.98801064e-02 -9.36837792e-02
-1.02903795e+00 9.01649177e-01 -6.91723600e-02 -4.38645780e-01
-4.12585616e-01 9.12342012e-01 -4.04976696e-01 3.74622159e-02
-1.87279209e-01 -4.73519146e-01 -1.06679358e-01 -1.02449946e-01
7.29167640e-01 6.28881097e-01 -1.86708108e-01 -4.76277798e-01
-4.06411558e-01 -1.51249185e-01 3.06441132e-02 1.32500619e-01
1.19783080e+00 1.12532742e-01 -5.36005616e-01 7.07218826e-01
8.60965192e-01 4.87264961e-01 -2.41252720e-01 -3.31484169e-01
3.83591622e-01 -6.18969560e-01 3.77552480e-01 -1.52023458e+00
-3.92954379e-01 5.04168570e-01 -3.18177640e-02 7.13998914e-01
5.40265203e-01 4.75491554e-01 9.55065370e-01 3.16286504e-01
5.40669858e-01 -1.19721651e+00 4.04898137e-01 5.57744265e-01
1.24803901e+00 -1.55835247e+00 8.29637200e-02 -5.70676863e-01
-8.30973566e-01 6.57892287e-01 2.76703686e-01 -1.51765764e-01
5.72154224e-01 -1.24681275e-02 4.36673202e-02 -8.27453136e-01
-8.67695868e-01 -5.38987751e-06 5.77256680e-01 1.03413790e-01
6.82016969e-01 2.98585773e-01 -7.01285720e-01 1.07121742e+00
-4.73676682e-01 4.26237643e-01 4.59972292e-01 7.57271945e-01
-7.89220989e-01 -2.56944925e-01 -4.36066300e-01 6.75155878e-01
-8.04627359e-01 -3.40672106e-01 -8.07307780e-01 5.00286460e-01
3.89667541e-01 1.30679917e+00 -3.99505347e-01 -5.54097414e-01
-1.16966199e-02 2.70445049e-01 2.82553971e-01 -4.05827582e-01
-7.35554278e-01 -6.04086637e-01 6.47390842e-01 -3.80969137e-01
-7.29713738e-01 -1.05286345e-01 -1.04775882e+00 -6.59667253e-01
-6.40026212e-01 2.34609962e-01 5.95101833e-01 1.35351288e+00
3.91010851e-01 3.08523774e-01 2.13163093e-01 2.20727429e-01
-1.12139225e+00 -1.24346268e+00 -3.91404510e-01 8.12683046e-01
3.18846941e-01 -8.78845215e-01 -5.05527377e-01 4.78494167e-02] | [8.704639434814453, 9.75560188293457] |
1e434176-1fe7-44e3-a27f-df4e371b27fa | what-transfers-in-morphological-inflection | null | null | https://aclanthology.org/2021.sigmorphon-1.18 | https://aclanthology.org/2021.sigmorphon-1.18.pdf | What transfers in morphological inflection? Experiments with analogical models | We investigate how abstract processes like suffixation can be learned from morphological inflection task data using an analogical memory-based framework. In this framework, the inflection target form is specified by providing an example inflection of another word in the language. We show that this model is capable of near-baseline performance on the SigMorphon 2020 inflection challenge. Such a model can make predictions for unseen languages, allowing us to perform one-shot inflection on natural languages and investigate morphological transfer with synthetic probes. Accuracy for one-shot transfer can be unexpectedly high for some target languages (88% in Shona) and language families (53% across Romance). Probe experiments show that the model learns partially generalizable representations of prefixation, suffixation and reduplication, aiding its ability to transfer. We argue that the degree of generality of these process representations also helps to explain transfer results from previous research. | ['Micha Elsner'] | null | null | null | null | acl-sigmorphon-2021-8 | ['morphological-inflection'] | ['natural-language-processing'] | [ 4.17807013e-01 1.03967801e-01 -2.33119667e-01 -5.29835343e-01
-6.48840606e-01 -1.22168565e+00 9.22996461e-01 2.35152841e-01
-7.70873547e-01 6.62437320e-01 5.57786226e-01 -6.23166859e-01
1.70740962e-01 -8.67871344e-01 -1.15091789e+00 -3.18198800e-01
-1.69198625e-02 8.93976748e-01 5.98433390e-02 -5.33408642e-01
3.63270581e-01 4.76657361e-01 -1.14773607e+00 6.53395891e-01
8.13411355e-01 1.58654526e-01 2.16002494e-01 1.83648080e-01
-3.25437814e-01 1.16213366e-01 -5.50599277e-01 -6.71140075e-01
2.50771523e-01 -1.74987301e-01 -1.09821200e+00 -3.47229809e-01
8.71465147e-01 -1.11922473e-01 -2.20458940e-01 8.75605643e-01
3.11021358e-01 1.25122711e-01 1.01755595e+00 -5.86754501e-01
-1.27311587e+00 1.30441892e+00 -4.63971227e-01 5.36903381e-01
4.58392441e-01 4.35645074e-01 1.14925134e+00 -1.10867155e+00
7.76632965e-01 1.65069580e+00 7.85859048e-01 6.30790412e-01
-1.68997109e+00 -7.36826360e-01 7.35220760e-02 -3.78738903e-02
-1.08576214e+00 -3.03011954e-01 3.25623333e-01 -3.12936604e-01
1.15128171e+00 5.43940924e-02 5.04701376e-01 1.28837883e+00
1.27687812e-01 8.09196115e-01 1.31602740e+00 -4.81373698e-01
-3.30377966e-02 -2.20419168e-01 5.50389051e-01 4.88043636e-01
4.64465678e-01 5.55285633e-01 -5.94471216e-01 -2.87047792e-02
7.21135378e-01 -4.62250441e-01 -2.18064055e-01 -2.84602121e-02
-1.42887914e+00 6.43477738e-01 8.05688918e-01 5.05480170e-01
-1.29362643e-01 1.42895550e-01 5.67196071e-01 7.97458410e-01
1.26609311e-01 9.64754283e-01 -6.37419939e-01 8.39165971e-02
-6.31318331e-01 3.57066572e-01 7.07860529e-01 1.02011955e+00
8.31922829e-01 1.56993419e-01 -1.47292286e-01 8.38420272e-01
-4.57472168e-02 5.39600432e-01 7.31524944e-01 -6.75182581e-01
5.36658943e-01 1.57752857e-01 -4.22412567e-02 -3.52719337e-01
-2.53025413e-01 -3.52097571e-01 -1.79825336e-01 -6.62081093e-02
7.33784616e-01 4.89002950e-02 -1.06933367e+00 2.42480636e+00
-1.61219016e-01 -1.00983486e-01 6.86391890e-02 4.75888014e-01
3.44179779e-01 5.96901000e-01 7.08914638e-01 -1.12827875e-01
1.37566459e+00 -5.46500742e-01 -2.06361800e-01 -6.57788694e-01
8.17619503e-01 -6.46335840e-01 1.84779251e+00 2.43747041e-01
-1.22590280e+00 -6.44751310e-01 -1.13221741e+00 -2.25530013e-01
-5.40558338e-01 -6.41871274e-01 9.00559545e-01 6.78669274e-01
-1.04249358e+00 8.66445661e-01 -7.53001034e-01 -8.32677603e-01
2.02963248e-01 3.14297885e-01 -2.84611136e-01 -2.71665990e-01
-1.26040864e+00 1.14018643e+00 7.90201426e-01 -3.94850224e-01
-8.78626823e-01 -1.16407120e+00 -7.09111869e-01 9.27377045e-02
-2.62600303e-01 -7.96383500e-01 1.31147981e+00 -1.30395317e+00
-1.01175916e+00 1.37052703e+00 -7.13064298e-02 -4.96206164e-01
5.99443838e-02 -2.14729667e-01 -5.22434413e-01 -6.29882589e-02
-2.75030844e-02 1.03970695e+00 5.22434056e-01 -1.19928312e+00
-2.55121917e-01 -5.12183070e-01 -1.31894201e-01 1.76781267e-01
-3.30176026e-01 3.56212318e-01 3.83596718e-01 -9.93892610e-01
1.01623878e-01 -8.70529890e-01 1.20305300e-01 -2.18693644e-01
3.11209038e-02 -3.03676516e-01 1.55269831e-01 -7.08764434e-01
8.59900594e-01 -1.97606683e+00 3.79471667e-02 -3.31600346e-02
-3.16881180e-01 1.50942713e-01 -7.98559129e-01 6.01434708e-01
-3.78366768e-01 3.65319014e-01 -6.11316264e-01 -6.66686594e-02
1.86064824e-01 3.04388136e-01 -8.24932754e-01 2.34294310e-01
3.69348317e-01 1.43108296e+00 -1.03298140e+00 -1.15909293e-01
-2.79358834e-01 1.32713065e-01 -7.80271649e-01 -1.67549383e-02
-3.01084787e-01 2.26544231e-01 3.35627615e-01 5.20818233e-01
7.21242249e-01 3.79656136e-01 3.59304041e-01 6.43709823e-02
5.02467044e-02 8.80635083e-01 -6.71523511e-01 1.92080820e+00
-5.33537984e-01 3.22291583e-01 -2.38890618e-01 -6.07506633e-01
8.67478490e-01 1.04379028e-01 -3.95540416e-01 -6.59872055e-01
-8.06079730e-02 5.29484689e-01 6.92361832e-01 -3.39661203e-02
5.61772883e-01 -8.29409063e-01 -3.15035939e-01 8.09608161e-01
3.93709868e-01 -3.95238936e-01 6.86691403e-02 9.74439830e-02
9.59439099e-01 1.26970768e-01 3.21081966e-01 -8.81357253e-01
9.73044112e-02 -3.53957415e-02 3.46718073e-01 8.02244306e-01
-3.27737443e-02 3.74528825e-01 2.45006885e-02 -4.16543424e-01
-1.12241423e+00 -2.00304747e+00 -2.92580366e-01 1.66657674e+00
-2.71523744e-01 -7.77980089e-02 -7.16147602e-01 -5.85868537e-01
2.04505786e-01 1.47180176e+00 -7.84943998e-01 -6.49978995e-01
-1.13163137e+00 -6.65845275e-01 9.38382566e-01 8.57260644e-01
4.37313179e-03 -1.76690865e+00 -8.18130672e-02 2.14051783e-01
-6.37849867e-02 -4.66129273e-01 -6.58223927e-01 3.97719413e-01
-1.21684778e+00 -5.00600636e-01 -4.67357248e-01 -1.29371083e+00
4.48247761e-01 4.40965779e-02 1.66423237e+00 8.70581716e-02
-4.17641625e-02 2.08154812e-01 -6.59008250e-02 -3.74414206e-01
-9.09331143e-01 4.82501239e-01 3.46410185e-01 -4.82875437e-01
7.82971323e-01 -9.65974987e-01 -3.27391744e-01 7.86351785e-02
-1.21849597e+00 -4.45604563e-01 5.39415896e-01 6.61810756e-01
2.46051729e-01 -9.74860787e-01 8.70599627e-01 -1.12431490e+00
6.63660586e-01 -7.01674461e-01 -3.11523408e-01 2.26980582e-01
-4.74365264e-01 2.57036179e-01 6.00152791e-01 -8.27896059e-01
-1.14656579e+00 -2.12767139e-01 -1.34508073e-01 1.00468993e-01
-2.98521698e-01 3.87499690e-01 -2.35581622e-01 3.38039994e-01
1.14049459e+00 3.58217120e-01 -1.75158754e-01 -5.96271157e-01
8.91293943e-01 3.88210684e-01 9.25022840e-01 -1.21001112e+00
8.60993564e-01 1.47766784e-01 -3.87243688e-01 -6.16722286e-01
-8.37396026e-01 1.21967467e-02 -8.84545445e-01 5.68426788e-01
5.99729896e-01 -7.74737895e-01 -4.53940004e-01 2.00354204e-01
-1.20043576e+00 -5.32403708e-01 -5.74568391e-01 2.91633189e-01
-8.45984578e-01 2.04477012e-01 -1.05975318e+00 -1.20098561e-01
-3.66143733e-01 -6.07038736e-01 7.97871530e-01 -4.95305136e-02
-8.06308985e-01 -1.13804102e+00 2.87505478e-01 -6.19319119e-02
4.57166284e-01 -2.18656331e-01 1.67276275e+00 -1.01039517e+00
-4.89769399e-01 2.24839345e-01 4.37411517e-02 1.67077065e-01
1.68421939e-01 -3.33346784e-01 -9.36373770e-01 -6.32215679e-01
-5.73103428e-02 -7.44467735e-01 1.19237030e+00 -6.93756491e-02
7.59707093e-01 -2.19133899e-01 -2.41159856e-01 6.97820902e-01
1.23029506e+00 -9.76518318e-02 7.02058494e-01 2.57080793e-01
2.98614770e-01 6.64029658e-01 3.75244796e-01 -3.15960228e-01
2.06441522e-01 3.26754123e-01 -9.76767391e-02 2.97942966e-01
-5.63357294e-01 -8.47466528e-01 6.84745669e-01 9.46196079e-01
3.73987198e-01 -1.98777318e-01 -1.10210395e+00 1.01629877e+00
-1.16066253e+00 -9.50383127e-01 1.05100907e-01 2.41758442e+00
1.38022935e+00 1.84047461e-01 -1.66029371e-02 -1.86185971e-01
6.85788453e-01 6.61842674e-02 -5.26949763e-01 -8.94351602e-01
-4.23050821e-01 7.96643138e-01 2.39450157e-01 7.95264184e-01
-6.63539290e-01 1.57172239e+00 7.73211813e+00 6.93523586e-01
-9.56463516e-01 7.49101713e-02 2.91012555e-01 1.86877459e-01
-8.06809843e-01 2.45593637e-02 -7.71787465e-01 1.80024266e-01
1.30904269e+00 -3.66948277e-01 7.69699931e-01 2.06257537e-01
-4.07182574e-01 3.27948809e-01 -1.73764765e+00 4.11990613e-01
1.71114624e-01 -1.05388522e+00 6.19630218e-01 -1.89166129e-01
6.50637031e-01 3.76781821e-01 2.84655422e-01 6.80271506e-01
6.28890097e-01 -1.23329318e+00 7.75481641e-01 2.61069715e-01
1.00717247e+00 -7.53973901e-01 -1.83152892e-02 3.57005715e-01
-6.87215507e-01 -3.23079266e-02 -6.25183523e-01 -1.73511401e-01
1.92849904e-01 -1.52655378e-01 -1.07986736e+00 2.98064351e-02
1.02815770e-01 2.60854155e-01 -8.62934709e-01 9.37997341e-01
-6.29704773e-01 1.01534975e+00 -3.45197409e-01 8.19748789e-02
6.47639632e-02 -1.24429323e-01 5.69778025e-01 1.57253838e+00
2.81013578e-01 5.14424331e-02 7.16051161e-02 1.08856571e+00
-3.97263914e-01 3.17706704e-01 -7.86889911e-01 -1.02301724e-01
7.67837763e-01 8.63296926e-01 -5.67237794e-01 -3.04304689e-01
-2.24605575e-01 1.20745707e+00 8.27487350e-01 3.40880007e-01
-5.10342777e-01 -3.23209941e-01 7.64710307e-01 2.91269660e-01
2.64602393e-01 -4.41575885e-01 -5.78018725e-01 -1.07983375e+00
-3.60718042e-01 -1.02325559e+00 4.73334163e-01 -7.08840489e-01
-1.61126029e+00 4.62271810e-01 -5.35942242e-02 -7.04553425e-01
-3.57145011e-01 -8.59559417e-01 -6.46864772e-01 9.63211954e-01
-1.08643901e+00 -1.38334680e+00 3.59443545e-01 3.31959367e-01
5.99356651e-01 -5.27136363e-02 1.04873705e+00 1.43270949e-02
-7.73288161e-02 1.05099118e+00 -1.17064886e-01 2.43435249e-01
1.12340474e+00 -1.42760324e+00 1.24650240e+00 8.09672594e-01
7.78475583e-01 1.24255502e+00 7.11661458e-01 -7.04080462e-01
-1.03442097e+00 -1.09344018e+00 1.10100460e+00 -1.01758337e+00
7.93961644e-01 -5.47448218e-01 -1.38862813e+00 1.34507632e+00
5.51069856e-01 -6.76954463e-02 8.12512457e-01 5.41177332e-01
-1.02661419e+00 2.92628825e-01 -1.03813887e+00 9.71852779e-01
1.63615525e+00 -7.75813282e-01 -1.59023547e+00 8.11591372e-02
1.05613458e+00 1.25565082e-01 -6.37124896e-01 2.22345740e-01
3.91617805e-01 -4.37991828e-01 9.76838589e-01 -1.37239599e+00
4.18422639e-01 -9.06558633e-02 -3.80094826e-01 -1.75709176e+00
-6.49141490e-01 -4.69896019e-01 9.51334611e-02 1.38577986e+00
6.41057193e-01 -5.89881718e-01 4.02336568e-01 2.16011256e-01
-2.75123119e-01 -7.91487843e-02 -7.33390868e-01 -1.23443997e+00
1.23829544e+00 -3.29812914e-01 7.25890875e-01 1.13004673e+00
7.52740279e-02 8.45087469e-01 3.20195645e-01 -1.21035613e-01
4.82247412e-01 1.90161556e-01 3.29801440e-01 -1.22673726e+00
-5.86036384e-01 -4.05015528e-01 1.30576007e-02 -1.03735554e+00
6.85792327e-01 -1.88566899e+00 -5.29533485e-03 -9.17926133e-01
2.88863450e-01 -3.61431986e-01 -4.44899827e-01 4.17673826e-01
-3.09150904e-01 3.39359432e-01 4.01070476e-01 7.56295919e-02
-8.19178522e-02 4.62122351e-01 8.43572199e-01 -4.49139290e-02
-3.86689268e-02 -4.06093776e-01 -9.85532820e-01 7.76367664e-01
8.24624777e-01 -6.30009234e-01 -2.38386124e-01 -1.00251496e+00
2.12692097e-01 -3.17859679e-01 7.70959482e-02 -6.59617722e-01
-1.30233422e-01 -2.12981343e-01 3.96981448e-01 -2.05461696e-01
1.21727355e-01 -3.11507672e-01 -2.70158082e-01 7.20855892e-01
-6.13850355e-01 3.78783911e-01 5.86312652e-01 3.46079826e-01
9.29525271e-02 -3.36905152e-01 9.39488888e-01 -2.66652316e-01
-6.36104345e-01 8.60587806e-02 -5.16416907e-01 6.23431087e-01
4.34787482e-01 -1.61474407e-01 -4.95227188e-01 3.59799154e-02
-7.49585986e-01 -1.36027411e-01 8.12271833e-01 6.92388773e-01
2.41142824e-01 -1.42726433e+00 -1.01924706e+00 3.24696988e-01
3.92588943e-01 -7.46065795e-01 -4.17380840e-01 1.62857026e-01
-2.91652381e-01 2.27871284e-01 -5.70932388e-01 -2.38396183e-01
-7.31049061e-01 1.10069513e+00 2.54060179e-01 2.74874605e-02
-1.99278921e-01 9.41755593e-01 5.66247225e-01 -9.95964408e-01
-2.52116352e-01 -4.33022946e-01 2.40489662e-01 2.89188568e-02
5.50592840e-01 3.81207131e-02 2.47350726e-02 -4.44065511e-01
-2.47765079e-01 4.78025556e-01 -4.72726852e-01 -4.95394051e-01
1.18763602e+00 3.39101516e-02 -2.01139301e-01 7.02401459e-01
1.04726291e+00 3.01668972e-01 -8.01105797e-01 -3.80986363e-01
5.88555276e-01 -2.06940994e-01 -5.45482576e-01 -1.27394986e+00
-4.00510669e-01 1.09841061e+00 4.17472392e-01 -4.10286069e-01
7.76955009e-01 1.73901692e-01 8.50193143e-01 7.12303519e-01
6.15527093e-01 -7.31963933e-01 -6.93471283e-02 9.05772388e-01
1.20243359e+00 -7.30452836e-01 -5.42649508e-01 -3.40779662e-01
-2.71066606e-01 8.18644226e-01 6.61053240e-01 -4.05632377e-01
2.87579179e-01 3.33911955e-01 5.22919595e-02 1.36450112e-01
-7.79965699e-01 -1.50766661e-02 7.02205971e-02 7.84309924e-01
8.48077297e-01 2.38231108e-01 -6.20226324e-01 7.69145012e-01
-6.83075130e-01 -4.21955645e-01 4.46394086e-01 6.11693978e-01
-6.77715778e-01 -1.35846221e+00 -3.02942127e-01 2.65352607e-01
-3.06003451e-01 -7.46521473e-01 -6.46703243e-01 8.77763212e-01
-7.89634436e-02 4.31563795e-01 4.52828795e-01 3.24828885e-02
3.79162103e-01 7.44383872e-01 9.92803514e-01 -1.02319300e+00
-9.94362831e-01 -4.08021063e-01 1.40962023e-02 -2.83692598e-01
4.08783555e-01 -8.77296865e-01 -1.40294921e+00 -2.94596106e-01
1.53122917e-01 -2.82293912e-02 1.45044848e-01 6.14800572e-01
1.98628142e-01 -1.18331099e-02 2.04085052e-01 -5.76387584e-01
-1.03312898e+00 -1.31466818e+00 -6.22056007e-01 8.96283031e-01
-5.30350432e-02 -2.97167629e-01 -2.64171392e-01 1.03707209e-01] | [10.757308006286621, 9.468131065368652] |
dd9d4393-4097-4795-b0c8-26c4686a4c0a | selformer-molecular-representation-learning | 2304.04662 | null | https://arxiv.org/abs/2304.04662v2 | https://arxiv.org/pdf/2304.04662v2.pdf | SELFormer: Molecular Representation Learning via SELFIES Language Models | Automated computational analysis of the vast chemical space is critical for numerous fields of research such as drug discovery and material science. Representation learning techniques have recently been employed with the primary objective of generating compact and informative numerical expressions of complex data. One approach to efficiently learn molecular representations is processing string-based notations of chemicals via natural language processing (NLP) algorithms. Majority of the methods proposed so far utilize SMILES notations for this purpose; however, SMILES is associated with numerous problems related to validity and robustness, which may prevent the model from effectively uncovering the knowledge hidden in the data. In this study, we propose SELFormer, a transformer architecture-based chemical language model that utilizes a 100% valid, compact and expressive notation, SELFIES, as input, in order to learn flexible and high-quality molecular representations. SELFormer is pre-trained on two million drug-like compounds and fine-tuned for diverse molecular property prediction tasks. Our performance evaluation has revealed that, SELFormer outperforms all competing methods, including graph learning-based approaches and SMILES-based chemical language models, on predicting aqueous solubility of molecules and adverse drug reactions. We also visualized molecular representations learned by SELFormer via dimensionality reduction, which indicated that even the pre-trained model can discriminate molecules with differing structural properties. We shared SELFormer as a programmatic tool, together with its datasets and pre-trained models. Overall, our research demonstrates the benefit of using the SELFIES notations in the context of chemical language modeling and opens up new possibilities for the design and discovery of novel drug candidates with desired features. | ['Tunca Doğan', 'Atabey Ünlü', 'Erva Ulusoy', 'Atakan Yüksel'] | 2023-04-10 | null | null | null | null | ['drug-discovery', 'molecular-property-prediction'] | ['medical', 'miscellaneous'] | [ 4.94296163e-01 -3.57339345e-02 -3.06518734e-01 -1.52147561e-01
-5.18326998e-01 -8.27306688e-01 4.35666561e-01 8.30127001e-01
6.21116944e-02 1.10713208e+00 1.07322305e-01 -8.74306440e-01
-3.80902588e-01 -1.06680727e+00 -7.79477715e-01 -7.40879238e-01
-3.45408201e-01 3.66572112e-01 -3.53440881e-01 -2.49822795e-01
5.04339278e-01 9.17245984e-01 -1.35700715e+00 4.86307412e-01
1.09267139e+00 8.38544071e-01 1.68452024e-01 3.43819946e-01
-4.98317361e-01 7.45708942e-01 -4.27014023e-01 -3.65126163e-01
-3.26202735e-02 -2.89635599e-01 -7.19538033e-01 -5.95367610e-01
1.47867829e-01 4.32677567e-01 -1.76999465e-01 7.79004872e-01
4.91286010e-01 2.26274520e-01 1.07826686e+00 -7.24634290e-01
-9.52806950e-01 5.59790075e-01 -2.79605091e-01 -8.40651616e-02
6.98687375e-01 3.75175536e-01 1.06843579e+00 -9.09797788e-01
7.19594836e-01 1.20424056e+00 4.99107420e-01 4.96802956e-01
-1.44337559e+00 -8.59883964e-01 -1.18784167e-01 -1.56004131e-02
-1.34165549e+00 -1.09615825e-01 6.96234167e-01 -7.61967838e-01
1.36736035e+00 2.94509411e-01 4.29978162e-01 1.04799521e+00
5.78531265e-01 2.72074431e-01 9.05399382e-01 -2.27412894e-01
5.09262323e-01 9.26700234e-02 1.78952098e-01 8.39551091e-01
4.56436634e-01 2.26595357e-01 -4.36564386e-01 -5.64507127e-01
4.54985440e-01 2.81958103e-01 -2.05137119e-01 -3.31724435e-01
-1.03139043e+00 1.17353201e+00 5.13013721e-01 4.66296047e-01
-4.37913984e-01 4.07428481e-03 4.49026912e-01 1.57271817e-01
2.91349530e-01 1.17013800e+00 -5.97607195e-01 6.92978352e-02
-6.18121624e-01 1.21771768e-01 6.59659386e-01 7.29416728e-01
7.39641726e-01 3.14956576e-01 -2.03539193e-01 4.65604901e-01
2.99275696e-01 7.59388059e-02 6.38973117e-01 -5.23992665e-02
3.34308267e-01 1.05294645e+00 -3.86507124e-01 -1.10442948e+00
-4.88982737e-01 -4.74364579e-01 -9.52820003e-01 1.18685447e-01
-4.47509997e-02 7.09070265e-02 -1.04295337e+00 1.60291803e+00
1.05244689e-01 -1.91451252e-01 4.76961464e-01 4.88734573e-01
1.06452215e+00 1.03318906e+00 8.13587368e-01 -2.48164013e-01
1.09279168e+00 -3.07200223e-01 -3.98952365e-01 3.81097823e-01
1.12462699e+00 -5.69845617e-01 9.85225141e-01 6.58143878e-01
-7.73149312e-01 -3.32385510e-01 -1.27911794e+00 6.65740520e-02
-1.11754620e+00 6.78139329e-02 1.15291560e+00 7.24038064e-01
-5.71955502e-01 1.26654577e+00 -4.12845075e-01 -3.07567548e-02
7.07397819e-01 7.41872489e-01 -7.08754838e-01 -1.12143330e-01
-1.25621223e+00 8.44579458e-01 7.35597491e-01 -9.28330794e-02
-6.09411478e-01 -9.95370030e-01 -1.07900834e+00 1.60442829e-01
6.41051903e-02 -8.17017376e-01 4.39270407e-01 -6.10932231e-01
-1.56858754e+00 3.79076481e-01 9.36827362e-02 -5.52036464e-01
-3.86470221e-02 1.36835039e-01 -5.49030840e-01 -6.56817704e-02
-2.40471080e-01 3.42766374e-01 3.80478024e-01 -7.54756927e-01
2.63576478e-01 -3.75290722e-01 -1.17785960e-01 -2.36551136e-01
-2.17868730e-01 -1.19541451e-01 2.95342118e-01 -6.81431651e-01
-2.78175116e-01 -5.81071138e-01 -5.48198462e-01 -1.27357006e-01
-5.59293389e-01 -3.14244747e-01 5.01602709e-01 -3.14446241e-01
1.29771638e+00 -1.76267505e+00 2.29525268e-01 7.77563751e-01
3.70706499e-01 5.52537322e-01 -3.53376746e-01 8.36140871e-01
-7.98455656e-01 6.69565856e-01 -3.29495817e-01 3.15936685e-01
-1.21557057e-01 -8.13737512e-02 -4.74465400e-01 3.28270584e-01
6.13224387e-01 9.59282160e-01 -9.63088393e-01 -2.01373577e-01
2.87135899e-01 7.55250633e-01 -7.16774702e-01 2.71296263e-01
-7.57373095e-01 4.02998507e-01 -8.04478407e-01 7.69107878e-01
5.80202043e-01 -3.13162088e-01 3.70815814e-01 -3.36241752e-01
-1.31904870e-01 4.06766474e-01 -8.30562413e-01 1.59560895e+00
-3.36493641e-01 1.22364327e-01 -8.10472548e-01 -9.32947218e-01
1.22387862e+00 2.54200965e-01 5.04089236e-01 -8.57005596e-01
-8.01922381e-02 1.88176438e-01 -1.01551423e-02 -3.88275772e-01
2.47733831e-01 -2.45420069e-01 1.68358296e-01 1.89043716e-01
2.67401338e-02 9.62662511e-03 4.53783780e-01 2.20413685e-01
1.02414536e+00 3.35456789e-01 6.11681283e-01 -8.89400020e-02
6.24497294e-01 1.60818007e-02 1.51026696e-01 2.01455608e-01
6.67417049e-01 1.30880147e-01 5.20180702e-01 -6.49246991e-01
-8.00489843e-01 -8.42084527e-01 -1.69455409e-01 9.15196776e-01
-4.70868886e-01 -9.30790901e-01 -3.90748560e-01 -4.70057905e-01
1.21004924e-01 6.58721328e-01 -6.84732735e-01 -3.35756123e-01
-2.20767811e-01 -7.51882970e-01 5.83237588e-01 2.46467024e-01
-1.70232534e-01 -9.98281360e-01 -4.17010710e-02 4.42215055e-01
6.26749456e-01 -4.61220205e-01 -1.28870726e-01 5.20845771e-01
-8.95311415e-01 -1.35028744e+00 -5.09003818e-01 -5.28924406e-01
7.38942146e-01 -1.36763975e-01 9.13555622e-01 -7.85820857e-02
-7.00248301e-01 -2.19529733e-01 -1.72859594e-01 -4.38768983e-01
-6.39292538e-01 5.69649637e-02 -7.59169608e-02 -2.40610152e-01
2.09845021e-01 -7.05740273e-01 -5.46656847e-01 -9.95434225e-02
-1.28463995e+00 -2.82452218e-02 9.10028994e-01 7.74399519e-01
7.52219081e-01 -2.38044903e-01 7.47473896e-01 -1.18171513e+00
9.43963647e-01 -7.20650673e-01 -5.72813630e-01 3.65218610e-01
-6.71530604e-01 8.27195942e-01 1.06601346e+00 -5.19487739e-01
-4.55049038e-01 3.65043551e-01 -2.80443192e-01 -1.85846299e-01
-1.02262688e-03 1.01028037e+00 -4.17147011e-01 -2.86930382e-01
9.05966520e-01 3.13188642e-01 -7.21578719e-03 -3.72157544e-01
4.60341364e-01 3.54226708e-01 -3.97102386e-02 -8.66146803e-01
5.75387418e-01 -2.55247176e-01 5.46772480e-01 -9.15918469e-01
-2.31264099e-01 -2.57844239e-01 -3.38086903e-01 4.10010576e-01
5.83021760e-01 -7.12686718e-01 -1.10794830e+00 -2.82499462e-01
-1.17994773e+00 1.15000069e-01 -8.09360147e-02 3.55637640e-01
-3.80068541e-01 5.63429356e-01 -2.57717907e-01 -6.65120542e-01
-7.83199430e-01 -1.12123954e+00 7.52986610e-01 4.56123892e-03
-5.11647761e-01 -9.56228435e-01 2.56574780e-01 -8.21531862e-02
3.45726371e-01 8.30338001e-01 1.73416269e+00 -9.73831475e-01
-6.08152509e-01 -1.51845142e-01 -1.21063069e-01 4.32525575e-02
3.45988750e-01 1.67871580e-01 -7.54577160e-01 -8.32461864e-02
-6.97180450e-01 -4.21891868e-01 7.63756990e-01 9.33585316e-02
1.53820455e+00 -6.22474372e-01 -3.82396758e-01 6.44549489e-01
1.44178593e+00 5.01358926e-01 8.64999235e-01 -3.38364728e-02
6.12347066e-01 3.93829107e-01 3.10679168e-01 4.45036739e-01
-2.01269418e-01 4.01987195e-01 4.62836802e-01 -2.16919616e-01
3.53399754e-01 -5.82722485e-01 3.11220586e-01 3.01546216e-01
-1.86436012e-01 -3.32219720e-01 -7.05704749e-01 -3.28082591e-02
-1.37401068e+00 -1.07779372e+00 1.27870888e-01 2.16133404e+00
1.18379927e+00 -6.77307844e-02 2.63506733e-02 1.40539795e-01
2.31598899e-01 -1.24091223e-01 -7.36913919e-01 -8.21927726e-01
-2.10657895e-01 8.97404730e-01 4.01994973e-01 1.71334505e-01
-7.47719824e-01 9.57117140e-01 5.85174131e+00 9.15558219e-01
-1.49977803e+00 -6.06896818e-01 7.02289402e-01 3.35189879e-01
-6.22878850e-01 -1.13697313e-01 -6.06191516e-01 3.45529199e-01
1.29532242e+00 -4.40119863e-01 2.97914445e-01 8.94452393e-01
3.40663075e-01 4.70547885e-01 -1.40344143e+00 9.91466939e-01
-2.21280500e-01 -2.10413408e+00 9.79820669e-01 1.60948515e-01
4.35417324e-01 -3.78476977e-01 1.50750756e-01 2.48099312e-01
2.29517892e-01 -1.82294703e+00 8.03280920e-02 5.73891938e-01
1.01608145e+00 -9.14819658e-01 4.04598355e-01 -2.90179010e-02
-1.06560993e+00 4.60398644e-02 -4.99933094e-01 2.40293033e-02
-4.74453598e-01 3.98759454e-01 -1.23815060e+00 8.42149258e-01
-8.27821270e-02 9.65763569e-01 -6.06146216e-01 9.17730093e-01
1.68212190e-01 4.21038985e-01 -9.19054523e-02 -4.10245389e-01
2.71868557e-01 -4.45458680e-01 2.82379538e-02 1.35799587e+00
3.89057130e-01 1.22583024e-01 1.39938325e-01 1.20415545e+00
-2.39037752e-01 5.79586625e-01 -9.80216920e-01 -8.91319335e-01
3.12899739e-01 9.51059401e-01 -2.97479391e-01 -1.98890164e-01
-1.38897106e-01 4.88423496e-01 1.55013010e-01 3.73414785e-01
-6.59266770e-01 -5.72952747e-01 6.24657929e-01 2.95998424e-01
3.03894132e-02 -2.62205809e-01 -1.80723503e-01 -7.51760304e-01
-4.24501121e-01 -1.05490601e+00 5.13794005e-01 -6.65507317e-01
-1.12210023e+00 7.75056660e-01 -1.71610773e-01 -1.23836374e+00
-2.74780929e-01 -1.20581698e+00 -5.53889573e-01 1.07596040e+00
-1.47204530e+00 -1.13584673e+00 2.00405091e-01 5.04527569e-01
2.76406318e-01 -4.04506624e-01 1.40317392e+00 2.28060678e-01
-6.09039962e-01 4.83674198e-01 2.77568191e-01 -2.16821000e-01
5.16611278e-01 -9.90479648e-01 1.25927135e-01 1.23451218e-01
1.94604605e-01 1.23275197e+00 6.23236775e-01 -6.57135844e-01
-1.79468381e+00 -1.32028782e+00 6.21691287e-01 -2.01964721e-01
6.36689842e-01 -2.38352045e-01 -9.29778755e-01 1.69958279e-01
-1.93168640e-01 -1.18114345e-01 1.41496480e+00 -1.70534879e-01
-6.17168844e-01 1.37820899e-01 -1.04479468e+00 4.56008345e-01
7.77202964e-01 -4.44193453e-01 -3.37545067e-01 6.39263570e-01
8.44878376e-01 -1.26031607e-01 -1.17613530e+00 2.81191528e-01
4.31340992e-01 -4.15505916e-01 1.18989742e+00 -1.26862931e+00
5.42605639e-01 -2.97087193e-01 -4.37918343e-02 -1.27978647e+00
-2.29816958e-01 -6.91993833e-01 -1.65369421e-01 8.74491513e-01
9.03419375e-01 -6.96940064e-01 7.34124720e-01 6.17229879e-01
-1.21921860e-01 -1.05952609e+00 -6.43030226e-01 -5.63884735e-01
2.64548600e-01 -1.52705133e-01 8.11116576e-01 8.50314021e-01
2.82322228e-01 5.80978513e-01 -1.69111758e-01 -1.49878666e-01
1.09166682e-01 1.74440011e-01 5.77987790e-01 -1.48504746e+00
-2.18671009e-01 -4.87866312e-01 -6.75402820e-01 -5.96686363e-01
1.90681770e-01 -1.39390647e+00 -7.18770206e-01 -1.36336279e+00
9.45220813e-02 -2.50434011e-01 -4.76008683e-01 6.34184480e-01
3.62839371e-01 -2.77110070e-01 -1.49541602e-01 -6.11130968e-02
-1.43858790e-01 5.83950758e-01 1.17569852e+00 -6.38200104e-01
-3.87772411e-01 -1.26679629e-01 -1.03935075e+00 2.60313421e-01
7.95214832e-01 -4.52051163e-01 -5.29101074e-01 2.47035936e-01
5.53193331e-01 -1.47012621e-01 4.20309901e-02 -7.08310723e-01
2.41277535e-02 -5.08820713e-01 6.44977093e-01 -4.42855537e-01
2.53148466e-01 -7.19751000e-01 4.40315306e-01 5.54608941e-01
-5.40022552e-01 6.73949420e-02 4.75571185e-01 5.82073510e-01
-6.71024173e-02 -2.99970433e-02 6.25219345e-01 -1.89586267e-01
-4.96275485e-01 6.46201789e-01 -2.70771950e-01 -2.79790789e-01
9.57971156e-01 -3.97424877e-01 -2.76119530e-01 1.39281362e-01
-5.39279938e-01 -1.94577307e-01 3.04506063e-01 2.18813851e-01
1.02015615e+00 -9.93861675e-01 -4.07214522e-01 4.02669966e-01
4.57005709e-01 -2.38703072e-01 1.27287470e-02 1.97250083e-01
-6.91881955e-01 7.22176611e-01 -2.64429450e-01 -2.00090304e-01
-1.22835958e+00 8.80825818e-01 2.28429615e-01 -4.27875698e-01
-2.48857409e-01 3.51856679e-01 1.92654684e-01 -2.52095371e-01
1.12446994e-01 -6.88182294e-01 -5.00061989e-01 2.00694101e-03
5.08393228e-01 -3.89414616e-02 3.10140491e-01 -3.26391757e-01
-4.04304445e-01 6.39571965e-01 -2.58611649e-01 6.17964685e-01
1.68432367e+00 7.89500713e-01 -9.48230773e-02 -1.03650652e-02
1.38936496e+00 -6.76466897e-02 -5.96323311e-01 5.84178753e-02
2.21337736e-01 -1.01485148e-01 -1.13629214e-01 -9.67419147e-01
-6.02103770e-01 1.00615299e+00 5.18057585e-01 -3.61095697e-01
7.94645965e-01 -2.38338515e-01 4.49776739e-01 9.34435070e-01
1.80452719e-01 -7.67195880e-01 1.57538399e-01 3.97793025e-01
1.04606366e+00 -9.61438656e-01 2.63669610e-01 -3.98522854e-01
-3.83429587e-01 1.51316142e+00 1.79288909e-01 2.30862960e-01
2.93579191e-01 -6.74695075e-02 -5.25434196e-01 -5.21232963e-01
-6.82336092e-01 3.93148623e-02 5.21327853e-01 7.20993161e-01
1.01136696e+00 5.89042492e-02 -2.68147945e-01 7.18971670e-01
-8.40931684e-02 9.19878557e-02 1.24141879e-01 9.19762671e-01
-2.95262009e-01 -1.51775312e+00 -2.21702725e-01 4.11314487e-01
-3.34909797e-01 -5.25915027e-01 -8.14138353e-01 6.26917958e-01
4.25829925e-02 6.53315365e-01 -5.10644078e-01 -4.02316540e-01
3.08574915e-01 1.48367912e-01 3.57481807e-01 -8.42823982e-01
-6.29624426e-01 -4.32224065e-01 1.30889803e-01 -3.69749784e-01
-1.88133836e-01 7.00657442e-02 -1.44629157e+00 -2.24426627e-01
-2.33885154e-01 5.89197755e-01 8.01160336e-01 6.10592186e-01
8.07479382e-01 6.28634334e-01 5.37377477e-01 -6.43812478e-01
-4.66068804e-01 -5.77996969e-01 -3.66358608e-01 3.09344798e-01
1.46611691e-01 -6.20345712e-01 3.64185013e-02 -8.71417820e-02] | [5.1146674156188965, 5.85244607925415] |
643cb166-a91a-4991-8dbe-e597d0463847 | tensor-representations-via-kernel | 1604.00239 | null | http://arxiv.org/abs/1604.00239v2 | http://arxiv.org/pdf/1604.00239v2.pdf | Tensor Representations via Kernel Linearization for Action Recognition from 3D Skeletons (Extended Version) | In this paper, we explore tensor representations that can compactly capture
higher-order relationships between skeleton joints for 3D action recognition.
We first define RBF kernels on 3D joint sequences, which are then linearized to
form kernel descriptors. The higher-order outer-products of these kernel
descriptors form our tensor representations. We present two different kernels
for action recognition, namely (i) a sequence compatibility kernel that
captures the spatio-temporal compatibility of joints in one sequence against
those in the other, and (ii) a dynamics compatibility kernel that explicitly
models the action dynamics of a sequence. Tensors formed from these kernels are
then used to train an SVM. We present experiments on several benchmark datasets
and demonstrate state of the art results, substantiating the effectiveness of
our representations. | ['Piotr Koniusz', 'Fatih Porikli', 'Anoop Cherian'] | 2016-04-01 | null | null | null | null | ['3d-human-action-recognition'] | ['computer-vision'] | [ 1.30097230e-03 -6.06941640e-01 -4.03627008e-01 -1.21637933e-01
-1.47409484e-01 -5.16391814e-01 6.76799476e-01 -1.86638579e-01
-2.83489645e-01 3.44745815e-02 5.86693645e-01 -1.10752605e-01
-2.63582468e-01 -2.47255713e-01 -4.40095246e-01 -7.53740847e-01
-6.09063268e-01 7.42495954e-02 5.09798825e-01 -1.23340033e-01
3.24051350e-01 8.93037856e-01 -1.13507485e+00 4.38915372e-01
4.71808136e-01 1.08332014e+00 -3.42062145e-01 8.43931794e-01
1.91490203e-01 1.30183363e+00 -3.54593724e-01 -1.92019999e-01
4.64300439e-02 -1.53867602e-01 -9.86872554e-01 4.36659873e-01
4.48418319e-01 -5.55454433e-01 -8.41064870e-01 7.09109426e-01
-1.74774937e-02 2.95617491e-01 8.99944246e-01 -1.11395800e+00
-8.29388440e-01 1.79758705e-02 -4.17112321e-01 4.75496083e-01
6.77985966e-01 1.98258281e-01 1.09526396e+00 -6.93186343e-01
5.02216101e-01 1.20620990e+00 6.09714091e-01 2.75689572e-01
-1.23515069e+00 -4.62383628e-02 1.58333093e-01 5.36367595e-01
-9.33041453e-01 -1.77780297e-02 8.75126660e-01 -8.44649076e-01
1.17855561e+00 2.53032953e-01 9.61079121e-01 1.26652479e+00
7.12873101e-01 1.10777867e+00 8.99602354e-01 -2.00796947e-01
4.91713174e-02 -6.98588133e-01 4.96838033e-01 9.30876911e-01
-1.43431023e-01 2.59076878e-02 -7.03816950e-01 -4.36070651e-01
1.24802554e+00 1.78935826e-01 -9.63217765e-02 -8.73741388e-01
-1.51559389e+00 7.77470767e-01 2.40929842e-01 2.73955137e-01
-5.51883757e-01 4.56818581e-01 6.28708303e-01 1.54446602e-01
5.29248714e-01 5.17712953e-03 -1.45437926e-01 -6.33586287e-01
-3.92766058e-01 2.55911052e-01 7.14862823e-01 5.41917562e-01
4.90365714e-01 -1.98997627e-03 -2.50993133e-01 8.94848883e-01
2.08123296e-01 2.32632354e-01 5.50245583e-01 -1.00552762e+00
3.48516285e-01 8.28891695e-01 -2.97158062e-02 -9.77588356e-01
-2.24027053e-01 1.46609485e-01 -6.07862473e-01 1.85325295e-01
5.10775208e-01 4.00533646e-01 -7.26885438e-01 1.40351677e+00
3.52635264e-01 5.28148353e-01 -2.57098734e-01 1.00900114e+00
1.90863669e-01 3.39417756e-01 -1.21414088e-01 3.37952971e-01
1.24603653e+00 -1.02204287e+00 -5.58520854e-01 3.17997187e-01
9.25025702e-01 -5.97709298e-01 7.23097444e-01 3.21332440e-02
-9.49599922e-01 -4.53977346e-01 -1.00610292e+00 -3.72376218e-02
-1.74535871e-01 4.04355019e-01 8.94277513e-01 2.70608336e-01
-8.87675941e-01 1.01811552e+00 -1.34865761e+00 -1.62316322e-01
1.05345905e-01 1.02612749e-01 -6.23217463e-01 1.31768867e-01
-1.13626420e+00 9.24395800e-01 -2.06380747e-02 7.74417669e-02
-8.92776310e-01 -5.86692333e-01 -9.56747711e-01 -3.58685642e-01
8.30787942e-02 -3.94148082e-01 1.16159582e+00 -3.64815891e-01
-1.57137442e+00 8.43145847e-01 2.00091705e-01 -2.36345381e-01
4.75344241e-01 -4.55581367e-01 -3.65126014e-01 4.83682126e-01
6.70373533e-03 1.84155017e-01 1.01013839e+00 -5.06153166e-01
-3.65851313e-01 -4.94405568e-01 2.83675998e-01 1.37805268e-01
-2.36997873e-01 8.73651877e-02 -4.83267248e-01 -9.82032001e-01
2.63818711e-01 -1.33882415e+00 -1.85617685e-01 3.39527547e-01
-3.31596136e-01 -5.22792637e-01 8.43049109e-01 -7.08475530e-01
1.04595876e+00 -2.20766830e+00 6.50619209e-01 1.64020568e-01
7.82383904e-02 2.96174228e-01 -2.94095844e-01 4.72076625e-01
-3.81420523e-01 -2.87919402e-01 9.74688977e-02 3.95457260e-03
2.66859550e-02 4.22115356e-01 -2.97358871e-01 7.01389015e-01
6.35285437e-01 9.04746473e-01 -8.99311543e-01 -1.93983644e-01
3.00688356e-01 6.59729660e-01 -3.90547395e-01 2.44231761e-01
9.93670523e-02 3.09980363e-01 -8.89825165e-01 6.27070725e-01
1.17543973e-01 -1.71040982e-01 2.62920465e-02 -6.85861707e-01
1.56120598e-01 4.85113919e-01 -1.03041351e+00 1.64329779e+00
4.78559621e-02 3.44618052e-01 -4.48831409e-01 -1.09580719e+00
8.76508534e-01 1.86986759e-01 9.90541339e-01 -2.37978145e-01
2.07194872e-03 -9.60713327e-02 -2.05228344e-01 -6.09287918e-01
3.56869280e-01 3.09033185e-01 2.70791333e-02 6.28547311e-01
1.95662558e-01 -1.33269848e-02 3.83529007e-01 9.34302360e-02
1.45033050e+00 5.37777364e-01 -2.28530094e-02 -1.15409702e-01
6.21680200e-01 -3.06559503e-01 2.30122179e-01 2.77681828e-01
-5.26681006e-01 2.38764986e-01 9.16305721e-01 -7.82818556e-01
-9.86096263e-01 -1.35836756e+00 1.44798696e-01 8.30454350e-01
-1.48888811e-01 -4.97508138e-01 -5.34507215e-01 -9.97209609e-01
4.83521849e-01 1.41563535e-01 -1.01304996e+00 -4.81116503e-01
-5.40472209e-01 -1.67157859e-01 7.71289527e-01 1.01935077e+00
2.68065006e-01 -5.23048043e-01 -7.19639897e-01 -6.15840629e-02
-6.58321530e-02 -1.23341274e+00 -9.30743635e-01 1.11462459e-01
-1.20725191e+00 -1.29560220e+00 -7.68980503e-01 -4.30658787e-01
4.25206751e-01 3.05449128e-01 5.60687482e-01 -2.37325832e-01
-4.31439310e-01 9.74078000e-01 -4.85168070e-01 3.51766437e-01
-2.49896973e-01 -5.31971037e-01 3.92453998e-01 3.46900970e-01
1.69617847e-01 -6.26353145e-01 -3.51104051e-01 5.33610761e-01
-1.09593594e+00 -2.58934349e-01 3.78509134e-01 8.26219022e-01
4.48501498e-01 -3.09024334e-01 -2.35443532e-01 -7.13157356e-02
6.62682414e-01 -4.05591391e-02 -3.07986677e-01 3.93448442e-01
7.54310712e-02 4.34055448e-01 1.09726228e-01 -1.05694997e+00
-6.61863565e-01 2.62846738e-01 2.97764391e-01 -1.07940853e+00
1.20465038e-02 5.73964715e-01 2.00350329e-01 -2.76456654e-01
4.67700928e-01 3.41103435e-01 2.52814263e-01 -7.35684216e-01
5.90540946e-01 2.42538109e-01 4.43275899e-01 -9.06977057e-01
6.53055847e-01 6.28551304e-01 1.79221615e-01 -1.05462110e+00
-6.37456238e-01 -4.86093760e-01 -1.12770891e+00 -3.90477270e-01
1.10480654e+00 -4.92796868e-01 -7.41421938e-01 8.11836600e-01
-9.94537532e-01 -2.80832261e-01 -3.11361253e-01 9.23736274e-01
-1.07577991e+00 8.36619854e-01 -1.19646478e+00 -6.68156862e-01
1.32033497e-01 -1.04691982e+00 1.12131166e+00 -1.59618929e-01
-2.51435071e-01 -1.01499748e+00 5.95139205e-01 4.59864885e-01
-1.17004558e-01 2.87792295e-01 1.01753509e+00 -4.13583696e-01
-3.76715869e-01 -5.02482355e-01 -1.16710387e-01 7.13107228e-01
3.04059863e-01 3.95010203e-01 -2.96133518e-01 -6.47857264e-02
-1.50377765e-01 -4.14315671e-01 8.56745005e-01 1.38778687e-01
7.99980879e-01 -1.81818493e-02 -2.19416276e-01 4.40109044e-01
7.02507973e-01 1.02872945e-01 7.07310021e-01 1.10009737e-01
9.21944737e-01 5.05848110e-01 5.94052672e-01 5.56929052e-01
1.49356753e-01 9.48704958e-01 7.80004263e-02 1.27119198e-01
4.23623696e-02 -2.73096532e-01 6.60847902e-01 9.42380607e-01
-6.68699205e-01 5.96418619e-01 -8.27474833e-01 3.84986192e-01
-2.11317587e+00 -8.62481594e-01 -1.23218760e-01 1.80784917e+00
7.62184620e-01 1.48506602e-02 4.24703270e-01 3.00641432e-02
3.80058646e-01 4.85068530e-01 -5.48917472e-01 -4.22451288e-01
1.97332159e-01 3.48152995e-01 2.54248112e-01 8.92007574e-02
-1.46709824e+00 8.93432200e-01 7.05039978e+00 5.31689882e-01
-9.71437931e-01 -2.11827889e-01 -2.35669431e-03 2.08939075e-01
4.97241877e-02 5.17052114e-02 -4.20078367e-01 1.55211434e-01
6.01089358e-01 1.55755430e-01 1.63087055e-01 9.19025302e-01
-5.33538908e-02 -4.06654365e-02 -1.20727074e+00 7.42125988e-01
6.25689924e-02 -1.30674279e+00 2.24993482e-01 2.17416331e-01
4.86968607e-01 -2.15599954e-01 1.77203473e-02 -2.34321784e-02
4.54189122e-01 -6.80339158e-01 6.25117362e-01 8.71443629e-01
3.11301172e-01 -6.07944548e-01 2.24174723e-01 1.21974833e-02
-1.33787644e+00 1.05035834e-01 -3.63127857e-01 -1.44641250e-01
1.56650260e-01 4.87382174e-01 -5.41624606e-01 6.22106612e-01
6.38977230e-01 1.23088896e+00 -5.40847957e-01 8.95728886e-01
-3.58116567e-01 3.39951217e-01 -6.88676611e-02 -1.25450017e-02
4.19206202e-01 -4.57363367e-01 6.04999006e-01 1.04427862e+00
-6.73364401e-02 8.40047076e-02 1.99709818e-01 7.54405737e-01
5.13908863e-01 -2.41629362e-01 -6.01117492e-01 -6.27885580e-01
-3.28746170e-01 9.86374080e-01 -3.50576699e-01 -1.99968666e-01
-5.48654616e-01 1.47249591e+00 6.38591766e-01 5.23788333e-01
-8.99235547e-01 -1.68238342e-01 1.54210174e+00 -3.50171417e-01
5.51681697e-01 -1.03668010e+00 2.94502139e-01 -1.48931503e+00
2.13868633e-01 -7.38382697e-01 4.81245399e-01 -7.61030734e-01
-1.49834049e+00 3.87367159e-02 1.69479802e-01 -1.32272494e+00
-2.36724511e-01 -1.05660474e+00 -3.84899408e-01 6.06801450e-01
-8.92579198e-01 -1.19223738e+00 1.21821657e-01 8.95862162e-01
1.16891652e-01 5.33969961e-02 9.09823358e-01 4.13644314e-02
-5.87168872e-01 3.06169808e-01 4.77073565e-02 5.49654067e-01
3.07198375e-01 -1.04099989e+00 5.67042053e-01 4.74647880e-01
4.81644571e-01 7.00178564e-01 8.79900753e-02 -8.56441438e-01
-1.83789861e+00 -6.83431685e-01 6.29798710e-01 -5.78431427e-01
1.29707468e+00 -1.02113448e-01 -8.86817575e-01 8.77014577e-01
-5.32521188e-01 2.99448133e-01 7.26165831e-01 3.86430509e-02
-1.01211905e+00 6.89199567e-02 -5.39915681e-01 6.57728434e-01
1.09942532e+00 -9.51628149e-01 -7.53372669e-01 3.16601485e-01
3.42843592e-01 -3.55060846e-01 -1.24065030e+00 4.44354385e-01
9.24035192e-01 -9.56947565e-01 1.30138862e+00 -1.43648767e+00
3.80646139e-01 -2.93857396e-01 -1.69012502e-01 -1.10724211e+00
-6.13799393e-01 -3.04246008e-01 -4.78186309e-01 4.65807408e-01
-1.19932897e-01 -3.70180517e-01 7.30171144e-01 4.56802368e-01
-5.87106459e-02 -1.01995087e+00 -1.04540682e+00 -1.28743136e+00
-4.81983796e-02 -3.85131150e-01 1.86579451e-01 9.98457074e-01
4.19601887e-01 1.19696908e-01 -4.06816334e-01 -6.00902140e-02
5.02984226e-01 -2.62510069e-02 6.35458171e-01 -8.04641902e-01
-3.85057688e-01 -5.16373754e-01 -1.32873988e+00 -1.41820252e+00
2.93593675e-01 -7.07978249e-01 -3.11448932e-01 -1.11389601e+00
1.93601862e-01 4.35950942e-02 -3.42735678e-01 6.67047858e-01
-1.36520723e-02 6.93530366e-02 3.17444489e-03 3.29442203e-01
-4.08091336e-01 8.94043565e-01 1.37599027e+00 -2.35542327e-01
-1.15200784e-02 -1.29968524e-01 1.99149743e-01 7.93894351e-01
3.80491793e-01 8.15777034e-02 -3.58115703e-01 -2.03476787e-01
-3.81311506e-01 3.85558940e-02 6.12797320e-01 -9.66307223e-01
-1.14651412e-01 -4.20279920e-01 5.23146570e-01 -5.66854656e-01
6.45209491e-01 -5.72838604e-01 -2.38391459e-02 5.69458485e-01
-4.04883355e-01 9.16446969e-02 -5.87187819e-02 7.86338031e-01
-1.31053403e-01 7.76863843e-02 5.43850541e-01 2.11441740e-01
-7.06242442e-01 3.54085743e-01 -5.55576384e-01 -1.40127257e-01
1.06491148e+00 -3.39845091e-01 1.48259522e-02 -2.79593587e-01
-9.54013288e-01 -1.90080509e-01 4.51555401e-01 7.32693315e-01
8.27325463e-01 -1.87847340e+00 -3.96844506e-01 1.78861856e-01
2.29093209e-01 -9.16515648e-01 1.47506654e-01 1.20204687e+00
-5.09265363e-01 4.57692057e-01 -5.62592328e-01 -7.50386238e-01
-1.39518178e+00 5.17203808e-01 4.27919179e-01 -4.43260640e-01
-7.13167846e-01 6.93478942e-01 1.08265720e-01 -3.23806852e-01
3.56785446e-01 -7.97187805e-01 -2.39607673e-02 -1.24485217e-01
5.55864394e-01 5.89737594e-01 -1.42875537e-01 -8.63692522e-01
-7.65792668e-01 7.06812024e-01 3.97276832e-03 -1.62865490e-01
1.14755452e+00 4.68164146e-01 -1.55737579e-01 6.30016685e-01
1.46322155e+00 -6.05227828e-01 -1.55317628e+00 -3.12769234e-01
2.98314661e-01 -7.39677668e-01 -2.50581294e-01 -3.40886921e-01
-9.00406003e-01 7.15467811e-01 3.77685070e-01 1.82970881e-01
8.28401923e-01 1.77801773e-01 7.03483999e-01 3.35892469e-01
2.72978425e-01 -1.02422643e+00 7.88122714e-01 9.49600697e-01
1.02593029e+00 -4.82439935e-01 4.54030521e-02 -4.16410506e-01
-7.84798086e-01 1.54904199e+00 2.16282889e-01 -5.48319042e-01
8.26183438e-01 -8.40416178e-02 -9.59552974e-02 -3.74505639e-01
-6.51908934e-01 -3.47641200e-01 8.18662763e-01 7.61709213e-01
4.54059780e-01 9.76335704e-02 -2.07116142e-01 1.24764271e-01
2.83348769e-01 2.05711741e-02 8.26128200e-02 1.15021539e+00
-2.33314469e-01 -1.24868715e+00 -2.42153272e-01 2.27257818e-01
-6.73257932e-02 4.68695939e-01 -7.62362242e-01 5.17337918e-01
-2.75223613e-01 6.53452277e-01 -2.32204776e-02 -8.49903464e-01
7.37488806e-01 2.85765350e-01 1.03355157e+00 -3.40473413e-01
-1.44759357e-01 3.23872827e-02 2.47729361e-01 -1.23861766e+00
-3.95351350e-01 -9.02471125e-01 -1.07398176e+00 -5.45917340e-02
4.40989286e-02 -1.92305177e-01 5.31465828e-01 9.80674267e-01
3.49686176e-01 1.82966411e-01 5.73308349e-01 -9.15024877e-01
-1.21395481e+00 -8.20996761e-01 -9.37414527e-01 7.88766623e-01
1.96751058e-01 -1.15878892e+00 -2.68092424e-01 2.13434651e-01] | [7.894011497497559, 0.3995693027973175] |
0b86d38c-0093-4c8d-bc45-eace2df55c7c | semantic-image-translation-for-repairing-the | 2303.17418 | null | https://arxiv.org/abs/2303.17418v2 | https://arxiv.org/pdf/2303.17418v2.pdf | Semantic Image Translation for Repairing the Texture Defects of Building Models | The accurate representation of 3D building models in urban environments is significantly hindered by challenges such as texture occlusion, blurring, and missing details, which are difficult to mitigate through standard photogrammetric texture mapping pipelines. Current image completion methods often struggle to produce structured results and effectively handle the intricate nature of highly-structured fa\c{c}ade textures with diverse architectural styles. Furthermore, existing image synthesis methods encounter difficulties in preserving high-frequency details and artificial regular structures, which are essential for achieving realistic fa\c{c}ade texture synthesis. To address these challenges, we introduce a novel approach for synthesizing fa\c{c}ade texture images that authentically reflect the architectural style from a structured label map, guided by a ground-truth fa\c{c}ade image. In order to preserve fine details and regular structures, we propose a regularity-aware multi-domain method that capitalizes on frequency information and corner maps. We also incorporate SEAN blocks into our generator to enable versatile style transfer. To generate plausible structured images without undesirable regions, we employ image completion techniques to remove occlusions according to semantics prior to image inference. Our proposed method is also capable of synthesizing texture images with specific styles for fa\c{c}ades that lack pre-existing textures, using manually annotated labels. Experimental results on publicly available fa\c{c}ade image and 3D model datasets demonstrate that our method yields superior results and effectively addresses issues associated with flawed textures. The code and datasets will be made publicly available for further research and development. | ['Libin Wang', 'Qing Zhu', 'Bo Xu', 'Haojia Yu', 'Han Hu', 'Qisen Shang'] | 2023-03-30 | null | null | null | null | ['texture-synthesis'] | ['computer-vision'] | [ 6.58980727e-01 8.09307694e-02 3.61794740e-01 -2.47236893e-01
-6.22683942e-01 -6.26916111e-01 6.82918489e-01 -2.44483218e-01
5.55424809e-01 4.79802728e-01 1.43765047e-01 -1.48299724e-01
-2.83784747e-01 -1.17601299e+00 -7.27909148e-01 -5.39786458e-01
2.33442500e-01 5.73805690e-01 2.81075776e-01 -5.96600771e-01
1.79977328e-01 7.38947690e-01 -1.92847753e+00 4.80468601e-01
7.97882199e-01 9.91216362e-01 3.43429625e-01 5.49421966e-01
-2.81500489e-01 3.91001254e-01 -3.81997108e-01 -3.04626942e-01
4.93566811e-01 -2.21684664e-01 -5.92616260e-01 6.49158895e-01
8.61666083e-01 -3.77415150e-01 -2.08488435e-01 8.15357924e-01
1.46670461e-01 -2.12519914e-01 8.09620261e-01 -1.00935638e+00
-5.22798419e-01 -2.96977341e-01 -5.81418216e-01 -5.97653985e-01
3.99631917e-01 2.65040427e-01 7.73676455e-01 -9.95539367e-01
6.75605834e-01 1.31082511e+00 8.68257105e-01 2.64673382e-01
-1.62966013e+00 -5.52615285e-01 1.08308919e-01 -2.83659846e-01
-1.64292932e+00 -4.06470895e-01 1.19760239e+00 -4.18381542e-01
5.08026242e-01 6.56948149e-01 8.19719970e-01 9.50470626e-01
1.81277156e-01 4.91338581e-01 1.39388204e+00 -5.56351840e-01
2.37161726e-01 -2.33780757e-01 -4.94753599e-01 8.73209774e-01
2.75356006e-02 1.59272090e-01 -3.84657115e-01 -1.23129718e-01
1.28636920e+00 -1.89538643e-01 -3.18513095e-01 -4.87003088e-01
-1.24349332e+00 4.54197526e-01 2.46756315e-01 -1.38726190e-01
-2.93645293e-01 1.75870225e-01 -1.26131818e-01 2.02757984e-01
6.24961019e-01 9.87476632e-02 -1.97492063e-01 2.09114805e-01
-8.25643241e-01 4.55154926e-01 4.70281333e-01 1.45460737e+00
1.22587967e+00 2.25507110e-01 1.01506233e-01 1.16268682e+00
3.30317914e-01 8.96294534e-01 -3.84704828e-01 -1.36720812e+00
3.35847765e-01 6.74212813e-01 2.17683673e-01 -1.24339712e+00
-8.76935944e-02 -1.80635676e-01 -1.02138638e+00 5.49939275e-01
2.13769302e-01 5.40290594e-01 -1.06846750e+00 1.21064866e+00
3.71406466e-01 -2.02489197e-01 -2.26461798e-01 6.63683236e-01
7.20315695e-01 3.90674382e-01 -1.15550093e-01 3.41966927e-01
1.34646809e+00 -4.94902760e-01 -4.93262649e-01 -3.99813503e-01
1.96163937e-01 -1.25270975e+00 1.51333272e+00 2.44990140e-01
-8.38202178e-01 -3.21141839e-01 -1.06768036e+00 -1.66836008e-01
1.37832105e-01 1.92363769e-01 5.76522231e-01 8.44949007e-01
-9.88272786e-01 1.42996311e-01 -6.05236530e-01 -2.48148844e-01
4.78502303e-01 1.42184854e-01 -3.25979203e-01 -2.91647404e-01
-6.39000237e-01 5.57871580e-01 1.57906547e-01 1.77981645e-01
-7.26997793e-01 -7.72096813e-01 -1.01837707e+00 -2.81793058e-01
3.41845363e-01 -8.40650380e-01 8.52516353e-01 -9.24594164e-01
-1.60184002e+00 9.31108952e-01 -5.70224114e-02 2.70942450e-01
5.49202979e-01 7.21243769e-02 -3.76083612e-01 -5.10394061e-03
2.57769436e-01 7.07034051e-01 8.45005333e-01 -1.82683933e+00
-4.31971908e-01 -1.40935138e-01 -4.64961343e-02 1.08212434e-01
1.77026750e-03 -4.82863992e-01 -4.25112009e-01 -1.14083862e+00
5.52603781e-01 -9.66486335e-01 -1.75274253e-01 4.68546271e-01
-4.03203040e-01 4.83591586e-01 1.10580468e+00 -6.63109303e-01
1.02506661e+00 -2.13537240e+00 -1.88199282e-01 6.47440672e-01
8.94966349e-02 -6.06581718e-02 -8.66627023e-02 3.53132337e-01
2.40575120e-01 7.70299807e-02 -5.82214236e-01 -3.38744700e-01
2.66364347e-02 5.96954703e-01 -3.44177723e-01 1.29342914e-01
2.89599597e-01 5.28217494e-01 -6.20386243e-01 -3.58006954e-01
6.13455772e-01 6.34081960e-01 -8.05216908e-01 -9.50645953e-02
-5.98414242e-01 7.65303850e-01 -6.68744326e-01 9.68018711e-01
8.67715418e-01 -5.41484728e-02 2.04564497e-01 -3.89977545e-01
-1.84544623e-01 -1.43457912e-02 -1.46034384e+00 1.66058111e+00
-8.09511900e-01 3.48648578e-01 3.04455131e-01 -6.72695041e-01
1.39994109e+00 1.41847983e-01 3.03979129e-01 -8.62114191e-01
-1.59437969e-01 5.52536547e-01 -6.29515290e-01 -6.69934228e-02
6.42983794e-01 -2.37676337e-01 -1.66596979e-01 3.26522321e-01
-5.01579225e-01 -1.04273474e+00 -1.91164508e-01 -1.08339809e-01
8.81173849e-01 5.42669654e-01 -2.22483292e-01 -5.15704334e-01
5.69769442e-01 2.02253506e-01 5.36375344e-01 5.71718335e-01
4.34228718e-01 1.31102800e+00 1.41684741e-01 -9.05175388e-01
-1.68562615e+00 -1.14975178e+00 -2.72505313e-01 3.54523361e-01
2.51401335e-01 -4.65239078e-01 -6.81775987e-01 -1.77844048e-01
-1.46228492e-01 5.27436674e-01 -5.51417708e-01 1.85524955e-01
-7.76404202e-01 -5.11643708e-01 4.88254964e-01 1.66764691e-01
8.69801879e-01 -6.77586973e-01 -5.37214637e-01 3.84903550e-01
-4.60497469e-01 -1.08788979e+00 -3.96037161e-01 -1.85827643e-01
-7.03988373e-01 -1.00238109e+00 -6.75989687e-01 -7.93362558e-01
9.37654912e-01 2.66677111e-01 1.46300220e+00 5.12242913e-01
-4.27765071e-01 4.19607311e-01 -2.74634540e-01 -2.12589115e-01
-5.31692624e-01 -2.49194667e-01 -4.54296887e-01 2.56348968e-01
-1.50711998e-01 -7.43746161e-01 -6.85589790e-01 8.09983075e-01
-1.23724699e+00 7.85437882e-01 3.10599566e-01 7.76335895e-01
1.02753448e+00 2.45401487e-01 -7.18595088e-02 -7.58584619e-01
1.63356140e-01 5.98950908e-02 -6.22404814e-01 2.27675259e-01
-3.06022376e-01 1.57461807e-01 5.70162475e-01 -1.19605385e-01
-1.44479525e+00 2.20354095e-01 -1.70051813e-01 -2.80265599e-01
-4.21982855e-01 1.38087362e-01 -3.24878424e-01 -1.78048804e-01
6.13343477e-01 2.20439389e-01 -1.18010022e-01 -6.05878413e-01
2.75980145e-01 3.49882901e-01 6.57341242e-01 -1.16368258e+00
1.05074894e+00 7.96519876e-01 9.26831588e-02 -1.07878006e+00
-6.23770773e-01 1.96973696e-01 -5.11126161e-01 -4.39477175e-01
5.95098853e-01 -8.65497887e-01 -4.19669837e-01 5.84644854e-01
-9.46050286e-01 -6.13844156e-01 -2.07403377e-01 1.39267862e-01
-8.01711917e-01 3.54918957e-01 -3.17879200e-01 -4.86269563e-01
-4.16787826e-02 -1.26334643e+00 1.73381007e+00 -1.00929968e-01
-3.12678486e-01 -8.09862554e-01 -1.23174742e-01 5.58791339e-01
6.20079935e-01 5.51287174e-01 1.28026175e+00 9.15947855e-01
-1.11417067e+00 2.54397154e-01 -1.62226543e-01 1.29081443e-01
3.09153080e-01 2.27645397e-01 -8.44716787e-01 1.87200420e-02
-3.05532038e-01 -7.47296363e-02 3.24148118e-01 1.70776874e-01
1.09720206e+00 -3.69975686e-01 4.12092358e-02 7.28403389e-01
1.47320068e+00 -5.77035770e-02 9.86370564e-01 6.70166135e-01
8.29434931e-01 7.26845860e-01 4.20707345e-01 4.89000112e-01
6.12090945e-01 9.63317752e-01 2.85479039e-01 -2.50538468e-01
-6.72622323e-01 -4.18025196e-01 -3.07689514e-02 5.54242492e-01
-3.41310173e-01 -7.29924142e-02 -1.00100422e+00 4.49000806e-01
-1.46020448e+00 -7.76832163e-01 -5.13685405e-01 2.09759164e+00
8.17126453e-01 3.66874002e-02 -4.07886118e-01 3.92123133e-01
5.55998921e-01 1.68929473e-01 -4.59857881e-02 -3.36836994e-01
-4.84649748e-01 6.33576870e-01 5.34043074e-01 6.71471834e-01
-8.29978228e-01 1.06077397e+00 5.61287642e+00 9.99669492e-01
-1.05669105e+00 -2.08453089e-01 8.02465379e-01 3.52301300e-01
-1.05092478e+00 2.74847120e-01 -4.84976888e-01 2.34816045e-01
2.48226792e-01 3.22692752e-01 5.09691119e-01 7.15975463e-01
2.79307038e-01 -1.94719329e-01 -6.86320722e-01 1.03195262e+00
-1.84912950e-01 -1.60024786e+00 1.28309995e-01 1.40166253e-01
1.10591042e+00 -4.56668526e-01 6.19403012e-02 -2.02736810e-01
5.15499473e-01 -9.74269092e-01 1.20626974e+00 4.81977075e-01
1.32581389e+00 -6.34341955e-01 1.77721441e-01 1.14593156e-01
-1.61578953e+00 4.80379462e-01 -2.28473157e-01 6.53274264e-03
2.61281133e-01 8.63428891e-01 -4.51907665e-01 7.69592166e-01
8.42983603e-01 5.05116224e-01 -5.39125681e-01 6.81076586e-01
-2.38867134e-01 2.85032809e-01 -6.38244748e-01 5.35484850e-01
1.25474229e-01 -4.85127926e-01 3.33389133e-01 8.59761000e-01
6.42309189e-01 3.72854620e-02 2.70484567e-01 1.16778576e+00
2.35312641e-01 -6.82423040e-02 -7.59155452e-01 4.48019266e-01
5.04224360e-01 9.81550455e-01 -1.09192669e+00 -1.47778973e-01
-3.17373931e-01 9.67567980e-01 -1.57557894e-02 4.63221908e-01
-6.38899565e-01 -1.02747194e-02 5.79549491e-01 4.66633260e-01
3.29156905e-01 -5.61348855e-01 -8.32048357e-01 -1.03409553e+00
2.49370307e-01 -1.02842307e+00 -1.46111503e-01 -8.92418146e-01
-1.22299266e+00 6.40178859e-01 4.40907069e-02 -1.41232681e+00
1.51463047e-01 -4.20990586e-01 -2.89951742e-01 7.98611760e-01
-1.32490551e+00 -1.72482574e+00 -6.52377307e-01 7.30699539e-01
4.62939829e-01 9.79307070e-02 1.02296889e+00 5.33351600e-01
-5.10438494e-02 1.53157935e-01 -1.07247561e-01 -2.42269009e-01
3.30763161e-01 -9.22225356e-01 5.37955046e-01 7.33987451e-01
-1.22896150e-01 2.33676329e-01 6.66372359e-01 -7.79431045e-01
-1.44692874e+00 -1.32891953e+00 6.95622504e-01 -3.05669785e-01
1.05259471e-01 -5.11811078e-01 -6.87874019e-01 6.80059731e-01
-2.55542696e-01 -8.23298190e-03 3.21275711e-01 -2.99894363e-01
-2.91195810e-01 -1.12983681e-01 -1.19683695e+00 8.90877962e-01
1.33848679e+00 -5.97926497e-01 -1.92869440e-01 1.33244216e-01
2.89139827e-03 -5.55022895e-01 -9.55661595e-01 7.40696251e-01
6.07557654e-01 -1.20872724e+00 1.26289773e+00 4.44011033e-01
4.02160794e-01 -7.84775853e-01 -3.89205784e-01 -1.04620814e+00
-2.61802286e-01 -6.04800522e-01 4.74008828e-01 1.13444650e+00
2.58996468e-02 -4.08035040e-01 8.90851855e-01 7.13680208e-01
-3.89476597e-01 -1.83232650e-01 -8.42663109e-01 -5.89916825e-01
-1.71505377e-01 -5.52046657e-01 9.17620659e-01 9.29330289e-01
-9.46650088e-01 -2.32690707e-01 -4.66364413e-01 2.21619263e-01
8.56692672e-01 3.98741394e-01 1.16890228e+00 -1.12490559e+00
-9.80020389e-02 -4.76442546e-01 -2.95150816e-01 -9.76739764e-01
-7.34296814e-02 -5.01119316e-01 7.19746128e-02 -1.42051482e+00
-2.25838780e-01 -1.36314702e+00 6.43203676e-01 3.79652292e-01
2.16740578e-01 9.07275558e-01 -2.15001076e-01 2.20698267e-01
9.77052143e-04 8.19714665e-01 1.47360826e+00 -2.28912055e-01
-2.45741326e-02 -3.13088000e-01 -4.48318154e-01 8.38002920e-01
6.52029574e-01 -2.41508245e-01 -5.51150799e-01 -8.21632147e-01
1.84400424e-01 -1.00588448e-01 6.47861540e-01 -1.01743627e+00
-2.58894384e-01 -2.87530959e-01 4.37739223e-01 -5.66177607e-01
4.56981659e-01 -1.02429378e+00 1.14293456e+00 4.03012969e-02
1.67645946e-01 8.28943327e-02 1.94715574e-01 3.51881325e-01
-2.41014034e-01 2.21742079e-01 8.27392220e-01 -2.73161769e-01
-7.19724894e-01 3.09953928e-01 -3.22902679e-01 -1.84579179e-01
7.62522638e-01 -5.39965928e-01 -2.01619923e-01 -4.01014954e-01
-5.45581281e-01 -1.10989593e-01 1.23326635e+00 2.95719087e-01
6.78378105e-01 -1.72900987e+00 -6.39576018e-01 8.05061400e-01
3.20015669e-01 4.94424671e-01 4.13625687e-01 3.32731038e-01
-1.34667540e+00 -3.72866131e-02 -3.90529633e-01 -8.81778121e-01
-9.36812758e-01 -5.36763705e-02 3.10568482e-01 -2.45483462e-02
-1.09958506e+00 4.09926206e-01 5.72241902e-01 -7.24206924e-01
-1.41005009e-01 -4.42883819e-01 3.92932236e-01 -5.10382771e-01
1.34665251e-01 1.49015978e-01 4.51981664e-01 -8.66344810e-01
-1.06537215e-01 1.07100749e+00 3.10287535e-01 -1.10608429e-01
1.38385546e+00 -3.10270727e-01 -4.65889797e-02 -1.04414113e-01
7.11381018e-01 2.26265252e-01 -1.51957726e+00 -2.08521724e-01
-1.12005919e-01 -9.54267383e-01 9.44587514e-02 -6.20497763e-01
-1.15874314e+00 6.03978455e-01 4.86438453e-01 -1.47822604e-01
1.21656549e+00 -2.11981386e-01 7.29434431e-01 8.26675724e-03
8.95910740e-01 -1.13705242e+00 8.20271149e-02 4.46777672e-01
1.21330905e+00 -8.72104883e-01 5.00640040e-03 -9.14090395e-01
-3.78414214e-01 1.27593660e+00 3.50563973e-01 -1.14333376e-01
6.04518831e-01 3.35074008e-01 3.05823356e-01 -4.43610847e-01
-3.26727867e-01 -5.67623302e-02 3.50756168e-01 6.23624444e-01
2.18399182e-01 1.04639284e-01 3.39711197e-02 -4.58901003e-02
-4.92135018e-01 -1.26971647e-01 4.24358845e-01 1.24572122e+00
-5.75748608e-02 -1.57615066e+00 -8.65776658e-01 1.75154731e-01
6.20160392e-03 -4.48819995e-02 -1.00416251e-01 7.44979441e-01
1.59076095e-01 8.24289382e-01 8.06386396e-02 -2.73850203e-01
4.95878398e-01 -1.56055063e-01 6.49866939e-01 -4.10841912e-01
-2.41705462e-01 5.26355207e-01 3.15555990e-01 -5.26201129e-01
-4.60236818e-01 -4.15006578e-01 -8.31359386e-01 -6.05766177e-01
4.10520248e-02 -4.76552308e-01 4.77057636e-01 6.73574865e-01
3.69631052e-01 3.71120244e-01 5.51964581e-01 -1.11692202e+00
2.77040362e-01 -4.99898672e-01 -6.69893682e-01 8.23434591e-01
8.30662027e-02 -1.04878497e+00 -8.36875811e-02 3.35590929e-01] | [9.270792961120605, -3.1941323280334473] |
abc5994d-9c3c-48f1-8319-9777d7c460a2 | a-safe-genetic-algorithm-approach-for-energy | 2306.14237 | null | https://arxiv.org/abs/2306.14237v2 | https://arxiv.org/pdf/2306.14237v2.pdf | A Safe Genetic Algorithm Approach for Energy Efficient Federated Learning in Wireless Communication Networks | Federated Learning (FL) has emerged as a decentralized technique, where contrary to traditional centralized approaches, devices perform a model training in a collaborative manner, while preserving data privacy. Despite the existing efforts made in FL, its environmental impact is still under investigation, since several critical challenges regarding its applicability to wireless networks have been identified. Towards mitigating the carbon footprint of FL, the current work proposes a Genetic Algorithm (GA) approach, targeting the minimization of both the overall energy consumption of an FL process and any unnecessary resource utilization, by orchestrating the computational and communication resources of the involved devices, while guaranteeing a certain FL model performance target. A penalty function is introduced in the offline phase of the GA that penalizes the strategies that violate the constraints of the environment, ensuring a safe GA process. Evaluation results show the effectiveness of the proposed scheme compared to two state-of-the-art baseline solutions, achieving a decrease of up to 83% in the total energy consumption. | ['Ramin Khalili', 'M. A. Gutierrez-Estevez', 'Nikolaos Petropouleas', 'Theodora Panagea', 'Alexandros-Ioannis Thanopoulos', 'Nikolaos Koursioumpas', 'Lina Magoula'] | 2023-06-25 | null | null | null | null | ['total-energy'] | ['miscellaneous'] | [ 4.43572134e-01 1.91942632e-01 -1.69699460e-01 -1.26296088e-01
-2.94892609e-01 -6.81614161e-01 4.98795480e-01 3.47890258e-01
-3.34386528e-01 6.58947885e-01 -3.58284026e-01 -2.51716524e-01
-5.83133757e-01 -1.02400792e+00 -3.54992360e-01 -1.17392921e+00
1.53407156e-02 6.67565241e-02 -3.04987848e-01 5.09772837e-01
1.07547976e-01 6.14976645e-01 -1.39192939e+00 -2.44133070e-01
8.77169430e-01 1.61653006e+00 -1.83066815e-01 1.83236420e-01
4.09594029e-02 5.96575499e-01 -6.30171895e-01 -5.96538723e-01
5.17939329e-01 -4.08353597e-01 -4.64757502e-01 -4.30150591e-02
-1.29678816e-01 -2.48733219e-02 6.88571334e-02 1.36074376e+00
5.33507586e-01 4.71493676e-02 1.55891880e-01 -1.54559731e+00
-1.40272021e-01 5.34964502e-01 -2.99028665e-01 -4.77582425e-01
-1.63953211e-02 5.27005792e-02 6.83299780e-01 -1.50843337e-01
4.93685693e-01 6.04734480e-01 4.24783856e-01 6.33916557e-01
-1.23816252e+00 -8.39487612e-01 4.68535461e-02 4.48335819e-02
-1.38085103e+00 -4.63319063e-01 1.03373384e+00 -2.26511285e-02
6.58731520e-01 6.46309614e-01 6.45061016e-01 6.75277472e-01
2.37880394e-01 2.63723820e-01 1.04029596e+00 -6.30810440e-01
9.93337095e-01 1.75902739e-01 -4.40897882e-01 4.27327275e-01
7.83699095e-01 -1.11485131e-01 -5.44580638e-01 -5.28075993e-01
-4.46693413e-02 9.19899046e-02 -7.21177831e-02 -9.75250065e-01
-6.59367204e-01 3.68867606e-01 1.43110558e-01 4.44106817e-01
-6.12618029e-01 8.73174295e-02 4.97068077e-01 -8.76805782e-02
4.16965783e-01 9.03885663e-02 -4.10563976e-01 1.07775636e-01
-8.67529988e-01 -5.62841557e-02 8.93999100e-01 7.10459888e-01
6.34923697e-01 -5.54842018e-02 -9.42917243e-02 -1.20065331e-01
3.68021548e-01 3.38628352e-01 1.33598909e-01 -9.50289369e-01
2.51954138e-01 1.21163082e+00 2.94692248e-01 -1.09320307e+00
-2.16351897e-01 -5.46408415e-01 -9.68519330e-01 1.03612296e-01
8.19127411e-02 -5.31705976e-01 -8.06628391e-02 1.73298657e+00
6.41687453e-01 2.17577934e-01 1.91158354e-01 4.38556701e-01
-6.34677708e-02 3.85478675e-01 5.47048330e-01 -7.81094015e-01
8.55876148e-01 -6.47233069e-01 -9.66457307e-01 4.78725731e-02
1.94852948e-01 -3.93035918e-01 3.72245967e-01 4.92860407e-01
-1.05314755e+00 6.26141429e-02 -1.11890960e+00 5.19386709e-01
-3.94510478e-01 2.44009033e-01 7.65043616e-01 1.34884906e+00
-7.89903283e-01 3.95630270e-01 -9.43114281e-01 -4.60651845e-01
6.46119297e-01 7.69146621e-01 -2.16896012e-02 7.36983567e-02
-5.47338009e-01 4.97349471e-01 3.06220412e-01 3.01032782e-01
-8.44745219e-01 -9.50597465e-01 -4.73951399e-01 3.70174587e-01
3.70547622e-01 -7.79920995e-01 8.67296219e-01 -7.40497112e-01
-1.45087016e+00 4.25714165e-01 9.66021717e-02 -8.58249724e-01
9.25439954e-01 4.04802710e-02 -5.84183574e-01 2.14422736e-02
-4.28603590e-01 -2.53577046e-02 6.84099972e-01 -1.35934198e+00
-8.43532205e-01 -5.76392829e-01 8.82474706e-02 1.33957103e-01
-1.13176346e+00 -1.47561684e-01 -2.16831133e-01 -3.23537081e-01
-3.01250786e-01 -9.63639796e-01 -5.11845052e-01 1.38550669e-01
-3.70672286e-01 -7.81688914e-02 1.26436841e+00 -4.05922681e-01
1.48158789e+00 -2.11031723e+00 -1.02235213e-01 5.64229369e-01
-2.78427545e-02 3.18411767e-01 3.46987665e-01 4.96770561e-01
5.87878823e-01 2.43258148e-01 -5.00212252e-01 -5.25542676e-01
1.63182214e-01 2.73101106e-02 2.93413460e-01 7.89905667e-01
-3.00336927e-01 4.10996020e-01 -7.35157907e-01 -2.11854041e-01
4.36980933e-01 5.15037119e-01 -1.75266698e-01 2.22820491e-01
-4.14514452e-01 3.62909913e-01 -7.22407699e-01 7.03307867e-01
8.31220806e-01 2.89773755e-02 1.08574605e+00 -2.71083146e-01
-2.01171726e-01 -2.99064666e-01 -1.27414775e+00 1.59628439e+00
-5.53769827e-01 -8.82625878e-02 7.35539854e-01 -9.32065070e-01
9.01444316e-01 4.56374258e-01 1.20309019e+00 -7.24592566e-01
4.72366571e-01 2.79377788e-01 -3.43787849e-01 -2.58909971e-01
-8.41972325e-03 2.87437975e-01 -2.49203295e-01 5.15927136e-01
-2.91636407e-01 4.55492169e-01 -4.16741930e-02 -4.81653921e-02
1.23431504e+00 2.41804108e-01 3.89199436e-01 -4.31360304e-01
9.42965746e-01 -1.52597025e-01 7.52333879e-01 3.91950876e-01
-4.04374421e-01 -4.27832216e-01 -4.20604972e-03 -4.11017060e-01
-5.59597015e-01 -5.35474658e-01 3.80947173e-01 5.70838392e-01
5.04562378e-01 -1.47723779e-01 -1.08395004e+00 -8.11439097e-01
1.01178758e-01 7.57829309e-01 -3.99193317e-01 -3.55085880e-01
-3.58264089e-01 -7.16796815e-01 5.16346335e-01 -1.29769534e-01
9.20904160e-01 -7.85090387e-01 -1.52506936e+00 2.42742121e-01
1.61659285e-01 -9.60892200e-01 -2.60993093e-01 3.43654931e-01
-5.78591645e-01 -1.23091054e+00 1.19673885e-01 -3.41338366e-01
8.11361372e-01 8.73003751e-02 6.19219303e-01 8.71063098e-02
-3.99607688e-01 2.80781031e-01 -8.18660632e-02 -5.19205153e-01
-3.52345824e-01 9.08392444e-02 9.38546844e-03 5.05662143e-01
2.52206743e-01 -5.15795708e-01 -8.47769618e-01 1.35865942e-01
-8.81458342e-01 -4.60290819e-01 4.75721955e-01 3.20235670e-01
6.05119288e-01 8.05007398e-01 6.97560787e-01 -8.97917628e-01
5.03644228e-01 -5.19604623e-01 -9.20425475e-01 6.61671042e-01
-1.23035479e+00 -1.72202170e-01 1.00927353e+00 -1.30805179e-01
-1.37792659e+00 5.23719549e-01 4.43586856e-01 -1.90597847e-01
4.07586396e-02 -6.74310327e-02 -9.00745511e-01 -5.99115431e-01
8.25292245e-02 1.13069400e-01 7.71264657e-02 -3.35139275e-01
1.83507100e-01 5.14667571e-01 3.35591495e-01 -4.08865333e-01
9.32538033e-01 7.57674456e-01 5.61425388e-01 -2.47622013e-01
-4.68107820e-01 -3.84833485e-01 -1.62008986e-01 -4.81441677e-01
6.27291679e-01 -5.79835773e-01 -1.28042853e+00 3.04888308e-01
-6.94762707e-01 2.76223302e-01 -2.74681062e-01 1.46039933e-01
-3.02137882e-01 2.14036271e-01 2.71953136e-01 -1.27402329e+00
-1.13725674e+00 -7.82935739e-01 3.83050442e-01 4.42587763e-01
-9.00370777e-02 -8.84260476e-01 -2.10569084e-01 4.40214664e-01
7.86845624e-01 8.56817245e-01 1.04404891e+00 -5.27798235e-01
-6.08457386e-01 -1.88336402e-01 3.40728283e-01 2.79881328e-01
3.82126510e-01 -1.46817371e-01 -8.39402020e-01 -7.43429124e-01
2.62391925e-01 6.63077906e-02 1.72843989e-02 -3.56344171e-02
1.38462961e+00 -7.73812473e-01 -3.62141758e-01 4.17162448e-01
2.08948302e+00 6.66040719e-01 2.71739930e-01 4.15323019e-01
2.46782079e-01 5.09882808e-01 6.48425519e-01 9.24692869e-01
9.67432559e-02 3.72542381e-01 1.09137857e+00 -1.93479343e-03
1.35580987e-01 -2.36166805e-01 1.23751633e-01 6.49377480e-02
3.66122961e-01 -5.36914647e-01 -5.28325558e-01 5.15379727e-01
-1.89855742e+00 -7.66830504e-01 1.76306948e-01 2.52109599e+00
2.10100636e-01 1.09396055e-02 -4.18112688e-02 5.15687466e-01
6.59569979e-01 1.96269318e-01 -1.03010499e+00 -5.55976927e-01
4.80874740e-02 1.27878979e-01 9.51464474e-01 -1.45543367e-01
-9.57767010e-01 3.60529721e-01 5.27816296e+00 5.91209054e-01
-1.22966266e+00 1.20938085e-01 6.07873857e-01 -2.84058869e-01
-3.41854334e-01 3.92961837e-02 -1.49901688e-01 4.74350631e-01
1.18034756e+00 -5.30715227e-01 6.84746742e-01 9.72992301e-01
5.86378157e-01 -2.69719213e-02 -8.06098759e-01 7.16936111e-01
-1.39268175e-01 -1.26220405e+00 -2.33624265e-01 2.22834781e-01
7.83788979e-01 -3.69480103e-01 -5.50446883e-02 -2.69957721e-01
3.39440018e-01 -5.56728065e-01 9.01093304e-01 4.13413078e-01
3.97433221e-01 -1.42763650e+00 5.11748433e-01 4.71866399e-01
-1.24532998e+00 -5.92326283e-01 8.45043585e-02 2.14458868e-01
-7.43943155e-02 8.07385743e-01 -5.73891401e-01 1.11392784e+00
8.59377623e-01 1.59334261e-02 -4.02942568e-01 9.84412849e-01
9.40391943e-02 5.19756734e-01 -3.45928907e-01 -1.01646557e-01
-5.02743423e-02 -3.36253285e-01 3.75054955e-01 7.04090893e-01
3.03197861e-01 -3.22090447e-01 1.04329437e-01 6.70910895e-01
-3.78392339e-01 2.74556577e-01 -3.64069790e-01 -9.18287933e-02
1.02125084e+00 1.52428889e+00 -5.14945507e-01 2.90513486e-01
-8.78242925e-02 7.47328639e-01 -1.01730190e-01 5.00571392e-02
-6.46988392e-01 -1.47583053e-01 5.89842796e-01 -1.02890640e-01
2.00858817e-01 2.32628025e-02 -5.58717728e-01 -3.92304868e-01
3.15803468e-01 -5.69908023e-01 4.93965715e-01 4.40620370e-02
-9.67097580e-01 2.66957045e-01 -4.11375999e-01 -1.11508560e+00
-2.31099799e-02 7.54603744e-02 -4.85377789e-01 4.50926036e-01
-1.49332547e+00 -1.25004363e+00 -2.96003342e-01 6.37342989e-01
5.42558283e-02 -1.21030584e-01 1.07257140e+00 4.65223432e-01
-8.09706509e-01 5.85527241e-01 4.34231281e-01 -6.54519320e-01
1.45670623e-01 -8.26787174e-01 -2.43603796e-01 1.20749366e+00
-3.76133382e-01 5.87206423e-01 6.25389993e-01 -5.60838521e-01
-2.10738945e+00 -1.34791493e+00 7.11320579e-01 3.53764176e-01
3.85288268e-01 -4.48160827e-01 -2.03364082e-02 2.11110696e-01
2.90617645e-01 2.06191167e-01 8.65460575e-01 -5.25547624e-01
7.28252158e-02 -6.62583947e-01 -2.26049781e+00 3.95083755e-01
9.25799549e-01 -1.97776064e-01 5.18584430e-01 7.08547160e-02
4.60020810e-01 1.50131732e-01 -9.92596924e-01 4.56111580e-01
5.04305601e-01 -7.19174445e-01 5.70686638e-01 -1.15557268e-01
-4.81298953e-01 -5.15947580e-01 -1.60949931e-01 -9.12949622e-01
9.93245374e-03 -1.11268675e+00 -7.23151863e-01 1.80681527e+00
6.32685097e-03 -3.94155443e-01 1.20951593e+00 9.46736157e-01
3.32698345e-01 -6.21576548e-01 -1.22439301e+00 -7.56431401e-01
-4.45769727e-01 -8.59962329e-02 9.17912066e-01 7.47573614e-01
-2.72186518e-01 -3.77937675e-01 -3.69249135e-01 4.36661422e-01
1.26930952e+00 -2.75238641e-02 4.40634936e-01 -1.27298152e+00
-1.77771912e-03 -6.44631758e-02 -2.31603086e-01 1.42398894e-01
3.03077459e-01 -6.42755270e-01 -2.81725705e-01 -1.42566299e+00
-1.00641347e-01 -5.67437112e-01 -6.42760515e-01 5.82033575e-01
4.00738388e-01 -3.14666033e-02 2.72849947e-01 6.12366088e-02
-7.13224471e-01 3.07009846e-01 5.95407724e-01 -3.03144336e-01
-6.27615228e-02 7.90351704e-02 -8.71073127e-01 4.07824516e-01
1.06388605e+00 -5.66625535e-01 -7.34728098e-01 -2.40989700e-01
-1.97092518e-01 -8.01285878e-02 1.23922840e-01 -1.29883432e+00
6.63231611e-01 -4.54350650e-01 -3.08633931e-02 -4.47616816e-01
9.35623143e-03 -1.99474573e+00 1.04313195e+00 9.73342597e-01
-1.26371309e-01 -1.77266344e-01 -1.47034302e-01 7.62991011e-01
1.64676756e-01 -9.94945541e-02 7.35400200e-01 5.20528555e-02
-3.19819123e-01 2.46606737e-01 -1.72883391e-01 -7.46907830e-01
1.90138173e+00 -1.05046518e-01 -3.28697324e-01 2.15507835e-01
-1.53261423e-01 4.38302428e-01 5.78661084e-01 2.55122334e-01
-5.27157001e-02 -8.06758821e-01 -1.49805704e-02 -7.70933554e-02
-8.08814764e-02 -1.13238573e-01 3.93141836e-01 4.24221367e-01
-2.89045602e-01 3.77923578e-01 -1.57736361e-01 1.92902312e-02
-1.27685916e+00 7.14526653e-01 3.85298818e-01 -5.01634419e-01
-3.41122866e-01 2.63451636e-01 -7.18782425e-01 -2.65682757e-01
6.75008953e-01 2.75210440e-01 7.86147639e-02 -1.28094122e-01
2.17837915e-01 1.05895281e+00 6.05379760e-01 -2.36406818e-01
-6.69671297e-01 1.68032899e-01 3.17186952e-01 2.99307734e-01
1.43457866e+00 -4.41014320e-01 -3.37871611e-01 -3.49685490e-01
9.80129600e-01 2.66485929e-01 -1.22640610e+00 2.40322202e-02
3.27410579e-01 -2.39259794e-01 3.43031377e-01 -1.19360375e+00
-1.41481137e+00 1.97219793e-02 1.11549938e+00 2.51975924e-01
1.65518117e+00 -6.94301367e-01 6.55299842e-01 7.92721212e-02
1.00994813e+00 -1.00436902e+00 -5.33074498e-01 -4.22226638e-01
6.90643787e-02 -5.54783762e-01 1.09678045e-01 -5.46345055e-01
-1.11998275e-01 8.22063327e-01 4.44474041e-01 2.19724938e-01
5.45549870e-01 4.91092592e-01 -3.48138809e-01 3.43150226e-03
-5.33779860e-01 4.38413680e-01 -4.06361848e-01 5.03331363e-01
-2.43313417e-01 2.17351645e-01 -7.76857793e-01 6.80629075e-01
3.19093645e-01 2.37357244e-01 1.81476519e-01 1.37530661e+00
-9.39092860e-02 -1.38780129e+00 -2.61821419e-01 3.38593990e-01
-6.93454981e-01 5.60292006e-01 -4.81776744e-01 3.82671773e-01
7.62727857e-01 1.38318372e+00 -3.82990479e-01 -1.52482435e-01
4.00360018e-01 -1.65151786e-02 -2.55692620e-02 -8.93830135e-02
-1.07088757e+00 -1.51760608e-01 1.14365980e-01 -6.55399859e-01
-7.76590824e-01 -7.37111986e-01 -1.23105383e+00 -2.09274486e-01
-1.17024414e-01 4.18779939e-01 1.23020911e+00 7.07451522e-01
5.94477713e-01 5.48150063e-01 1.16171885e+00 -3.54574293e-01
-6.87300801e-01 -2.66611993e-01 -4.88064647e-01 9.17324498e-02
4.79830196e-03 -2.93387026e-01 -4.35806513e-01 -1.10996865e-01] | [5.865357398986816, 6.057925224304199] |
45a8bb08-bb09-442b-9a48-5c60f1a6344c | proteinnet-a-standardized-data-set-for | 1902.00249 | null | http://arxiv.org/abs/1902.00249v1 | http://arxiv.org/pdf/1902.00249v1.pdf | ProteinNet: a standardized data set for machine learning of protein structure | Rapid progress in deep learning has spurred its application to bioinformatics
problems including protein structure prediction and design. In classic machine
learning problems like computer vision, progress has been driven by
standardized data sets that facilitate fair assessment of new methods and lower
the barrier to entry for non-domain experts. While data sets of protein
sequence and structure exist, they lack certain components critical for machine
learning, including high-quality multiple sequence alignments and insulated
training / validation splits that account for deep but only weakly detectable
homology across protein space. We have created the ProteinNet series of data
sets to provide a standardized mechanism for training and assessing data-driven
models of protein sequence-structure relationships. ProteinNet integrates
sequence, structure, and evolutionary information in programmatically
accessible file formats tailored for machine learning frameworks. Multiple
sequence alignments of all structurally characterized proteins were created
using substantial high-performance computing resources. Standardized data
splits were also generated to emulate the difficulty of past CASP (Critical
Assessment of protein Structure Prediction) experiments by resetting protein
sequence and structure space to the historical states that preceded six prior
CASPs. Utilizing sensitive evolution-based distance metrics to segregate
distantly related proteins, we have additionally created validation sets
distinct from the official CASP sets that faithfully mimic their difficulty.
ProteinNet thus represents a comprehensive and accessible resource for training
and assessing machine-learned models of protein structure. | ['Mohammed AlQuraishi'] | 2019-02-01 | null | null | null | null | ['protein-secondary-structure-prediction'] | ['medical'] | [ 5.31000316e-01 -1.67887494e-01 -1.19669646e-01 -7.47607589e-01
-9.08419609e-01 -7.00135767e-01 2.58311719e-01 8.31178427e-01
-7.26115763e-01 1.17683113e+00 -1.06110632e-01 -6.90876901e-01
-6.00243807e-02 -1.98828638e-01 -7.44759023e-01 -7.88331509e-01
-2.56367356e-01 9.11022902e-01 1.45104125e-01 -2.99780011e-01
2.67480105e-01 6.42629325e-01 -1.26990402e+00 4.82153118e-01
9.49287891e-01 4.64961827e-01 5.47427356e-01 5.65031826e-01
9.51756447e-05 2.54585743e-01 -4.41349596e-01 -1.19631626e-01
9.89239141e-02 -6.00533247e-01 -1.01377273e+00 -4.20735389e-01
4.23169762e-01 1.34773161e-02 1.40266582e-01 5.83285213e-01
6.71111524e-01 -3.24823521e-02 5.91562152e-01 -7.31538117e-01
-6.57032669e-01 -1.19049057e-01 -3.69260013e-01 2.76043028e-01
3.22846085e-01 7.90522397e-01 1.19028342e+00 -7.25425720e-01
1.06325006e+00 8.67764175e-01 9.89997864e-01 5.21912158e-01
-1.77911973e+00 -3.36075008e-01 -3.74865502e-01 2.98658073e-01
-9.34957147e-01 -1.74667269e-01 9.97914970e-02 -6.53953552e-01
1.86690962e+00 5.40698431e-02 7.13688374e-01 9.73213434e-01
4.40948695e-01 7.51634464e-02 6.24194860e-01 -9.36067253e-02
3.22888315e-01 -4.62528914e-01 1.84929177e-01 5.87646246e-01
1.37841955e-01 -3.30656283e-02 -3.43305945e-01 -7.54074335e-01
3.75404388e-01 -9.41364020e-02 -2.44861051e-01 -7.37296045e-01
-1.24605608e+00 8.35928261e-01 2.49790832e-01 2.39695031e-02
-5.23286283e-01 -3.69862109e-01 5.33923805e-01 2.91390121e-01
-6.51386008e-02 8.51080060e-01 -1.11435831e+00 -1.55581847e-01
-8.38001132e-01 5.62512159e-01 7.92248368e-01 4.32946056e-01
7.60597587e-01 -2.14436516e-01 4.20598179e-01 7.81488538e-01
6.26044050e-02 3.73363271e-02 6.58922017e-01 -8.86043787e-01
3.03002559e-02 5.71766317e-01 3.59856695e-01 -6.99177742e-01
-6.30302846e-01 -7.56782964e-02 -4.05909479e-01 5.02978981e-01
6.62372351e-01 1.76152900e-01 -9.05560613e-01 1.77400482e+00
4.91651684e-01 -4.02632833e-01 1.98339373e-01 7.39105463e-01
5.53700566e-01 4.42646682e-01 3.08518469e-01 -2.04058617e-01
1.30278027e+00 -4.12131757e-01 -4.17754613e-02 8.22306275e-02
9.33948457e-01 -7.56719351e-01 1.05933356e+00 4.19546425e-01
-9.35510397e-01 -4.89421219e-01 -1.36131847e+00 -3.25156420e-01
-3.99184406e-01 -3.05613548e-01 5.64363003e-01 2.29451448e-01
-9.61983979e-01 1.06608903e+00 -9.04463172e-01 -4.57625449e-01
5.09520292e-01 5.29562891e-01 -8.68023634e-01 2.29241401e-01
-1.15169287e+00 1.43716550e+00 7.94768989e-01 -3.14496040e-01
-9.53796387e-01 -8.39632392e-01 -6.29116297e-01 -3.72711904e-02
-1.86284840e-01 -5.49907744e-01 1.28090763e+00 -9.75343704e-01
-8.50335777e-01 1.18836772e+00 -2.02629685e-01 -8.01243722e-01
5.31721115e-02 8.98877233e-02 -1.71588838e-01 2.37169445e-01
-9.52932052e-03 8.38636577e-01 1.93353102e-01 -6.13186300e-01
-4.10578370e-01 -4.65480179e-01 -3.03468376e-01 2.38115266e-01
2.73688972e-01 2.57105619e-01 2.97111779e-01 -2.26745024e-01
-1.24031994e-02 -6.65828943e-01 -5.17654598e-01 2.09218234e-01
2.06401885e-01 1.56346168e-02 4.86722380e-01 -7.79132366e-01
5.50047815e-01 -1.58781779e+00 1.61672324e-01 1.02076866e-02
5.37900031e-01 5.81508040e-01 -3.34692627e-01 7.75148034e-01
-6.44742310e-01 8.67722407e-02 -3.84106696e-01 3.84398729e-01
-4.49602574e-01 3.30018133e-01 1.29864350e-01 6.74010396e-01
5.13676882e-01 9.09326553e-01 -8.17553520e-01 -1.47909358e-01
5.78566678e-02 5.87684751e-01 -6.15348697e-01 1.77969888e-01
-5.20740449e-01 4.37897861e-01 -1.79741383e-01 4.72210854e-01
5.15428424e-01 -6.00340724e-01 6.76402390e-01 -1.10503137e-01
-1.19806096e-01 6.68733954e-01 -5.03786445e-01 1.64135516e+00
1.95254937e-01 2.29704320e-01 1.33364707e-01 -1.09365535e+00
1.02517056e+00 2.01180995e-01 8.24692488e-01 -3.11931640e-01
-9.50997099e-02 1.14783369e-01 7.04461575e-01 -2.34097302e-01
5.41836396e-02 -3.08831066e-01 4.57936853e-01 3.49790186e-01
3.22413027e-01 -7.61842355e-03 1.61807165e-01 1.94332786e-02
1.34738576e+00 5.22498429e-01 4.51766551e-01 -3.45008135e-01
3.82152081e-01 7.19740689e-01 9.31451440e-01 2.62630671e-01
-4.26119089e-01 5.21668613e-01 3.94342244e-01 -8.70792270e-01
-1.80627131e+00 -1.05462646e+00 -4.48579669e-01 1.14501095e+00
-2.44936839e-01 -7.03162253e-01 -7.60883152e-01 -4.15730327e-01
4.29471061e-02 2.25712016e-01 -2.92773098e-01 -3.11230868e-01
-5.33903480e-01 -1.20222068e+00 5.98902404e-01 9.28292051e-02
-1.28117770e-01 -1.24955404e+00 -7.50302255e-01 6.04436576e-01
7.28353932e-02 -5.02462268e-01 -1.29884154e-01 8.66492331e-01
-7.84575701e-01 -1.70888102e+00 -5.47647059e-01 -1.02087843e+00
2.47186020e-01 -5.76938614e-02 1.27041256e+00 -3.65342349e-02
-8.88094544e-01 -2.49544397e-01 -8.33190046e-03 -4.27835286e-01
-8.18937182e-01 -5.69786271e-03 3.29186320e-01 -7.01829612e-01
8.69429410e-01 -6.76901042e-01 -6.35081828e-01 5.06408572e-01
-8.10634315e-01 1.02548420e-01 4.42792386e-01 1.24096477e+00
9.89225268e-01 -4.23712790e-01 6.81140721e-01 -5.97755730e-01
5.91435730e-01 -3.86338919e-01 -6.10824287e-01 3.14548343e-01
-7.40210474e-01 1.90662533e-01 7.08786726e-01 -1.98700115e-01
-7.94089139e-01 2.78032631e-01 -4.36102152e-01 1.19450673e-01
-4.76723433e-01 5.58590710e-01 -1.27606601e-01 -1.19210429e-01
9.45520282e-01 3.08598697e-01 5.19094229e-01 -7.80104876e-01
6.25939146e-02 4.18699205e-01 5.61239123e-01 -7.37571955e-01
3.62046987e-01 2.31848750e-03 -1.38626322e-01 -7.35996485e-01
-3.19324523e-01 -4.60262418e-01 -8.56686652e-01 5.47651350e-01
9.07930851e-01 -7.07990468e-01 -1.06957090e+00 3.56953830e-01
-9.69994128e-01 -2.34733149e-01 -8.02552029e-02 5.22469819e-01
-8.02381575e-01 6.88526332e-01 -7.08866239e-01 -1.35967940e-01
-5.70446372e-01 -1.37418020e+00 7.74170518e-01 -1.35854408e-01
-7.55711079e-01 -7.71113336e-01 5.27238250e-01 4.58809167e-01
1.62913471e-01 5.27515948e-01 1.43914104e+00 -1.18177271e+00
-2.39346236e-01 1.87869638e-01 6.28131256e-02 4.00148958e-01
2.43776023e-01 2.09498480e-01 -5.70235789e-01 -3.56736243e-01
-1.92217335e-01 -8.53885412e-01 7.61586905e-01 1.56899363e-01
7.83234179e-01 -1.34730950e-01 -3.89937788e-01 6.73256040e-01
1.25303328e+00 5.57355344e-01 3.69350076e-01 5.95635235e-01
4.24510777e-01 6.78175449e-01 4.26464409e-01 7.47423619e-02
-2.30443496e-02 6.10808790e-01 2.81846166e-01 -3.81418139e-01
4.61299896e-01 -2.00046852e-01 2.11002901e-02 1.98753357e-01
-6.08120821e-02 -2.11242009e-02 -1.27865720e+00 2.32565627e-01
-1.73857367e+00 -1.05722976e+00 -4.84687909e-02 2.04777598e+00
1.39229572e+00 1.53085038e-01 1.66306406e-01 -3.48715663e-01
5.97212791e-01 -2.59312004e-01 -1.21926773e+00 -4.38979983e-01
-4.48260605e-01 5.68037033e-01 3.23396146e-01 4.55114603e-01
-8.77241492e-01 7.66386211e-01 7.73466253e+00 5.83188832e-01
-9.95513797e-01 -3.72490197e-01 6.58571005e-01 -2.96481967e-01
-1.34106591e-01 5.35507612e-02 -7.27234900e-01 4.06595379e-01
1.34301901e+00 -3.13787580e-01 2.91802257e-01 8.68730187e-01
4.34219629e-01 1.57798931e-01 -1.31543291e+00 6.82931304e-01
-6.54017687e-01 -1.99250102e+00 -5.83050363e-02 2.04414502e-01
4.05195057e-01 8.60289752e-01 -1.04686134e-01 -1.56873897e-01
7.74210215e-01 -1.27398920e+00 1.04219884e-01 1.76190004e-01
8.34932804e-01 -6.62958980e-01 5.78000069e-01 1.49052754e-01
-6.41874373e-01 3.94622117e-01 -5.54207325e-01 -2.12780219e-02
-4.46724407e-02 3.46785426e-01 -1.19999313e+00 1.86699390e-01
5.99547625e-01 6.29859328e-01 -3.15959454e-01 6.02038085e-01
3.00897568e-01 4.25767273e-01 -2.70531863e-01 1.68030873e-01
2.65291899e-01 -4.07183766e-01 2.59230316e-01 9.15826321e-01
-3.01798016e-01 -6.01546243e-02 2.35087574e-01 9.44742680e-01
6.35772943e-02 1.92902848e-01 -2.01142550e-01 -3.28702658e-01
4.43179876e-01 1.04837143e+00 -4.75953847e-01 -1.91379339e-01
-4.65102196e-01 7.19342887e-01 6.02306545e-01 2.37017363e-01
-7.61875391e-01 -3.37463319e-01 1.56848621e+00 1.88196823e-01
2.64689118e-01 -3.14571530e-01 2.10406393e-01 -9.75535929e-01
-1.46698982e-01 -1.46463287e+00 4.04614449e-01 -7.29551315e-01
-1.37109351e+00 3.95954192e-01 -3.49098980e-01 -4.21066940e-01
-2.34521881e-01 -9.34394956e-01 -3.42652828e-01 1.35352397e+00
-1.00702357e+00 -7.55627990e-01 3.11078578e-01 1.40312806e-01
3.03701669e-01 -2.79361695e-01 1.17614543e+00 -5.41698299e-02
-4.99510825e-01 4.50196385e-01 8.25935066e-01 -2.95535654e-01
9.43141162e-01 -1.13406539e+00 9.49969888e-01 2.19457537e-01
-2.39873126e-01 1.18409014e+00 1.02082419e+00 -8.37850869e-01
-1.07974434e+00 -8.61587822e-01 7.09574103e-01 -5.46967089e-01
6.57414258e-01 -3.76745313e-01 -1.45118678e+00 5.97355366e-01
-5.83066978e-02 -4.09458041e-01 1.19729972e+00 1.25264242e-01
-4.75277722e-01 2.87374735e-01 -1.45934367e+00 3.03354949e-01
8.92060339e-01 -5.85394442e-01 -7.66882598e-01 4.37098444e-01
5.13305128e-01 -3.54274437e-02 -1.11963975e+00 4.89800870e-01
6.45545781e-01 -1.07699847e+00 1.03976834e+00 -1.44797397e+00
3.59917372e-01 -3.50081384e-01 -1.47087738e-01 -9.98033822e-01
-6.07237816e-01 -5.48563957e-01 1.72230378e-01 6.35172307e-01
6.71998858e-01 -6.96855724e-01 1.04839492e+00 6.54317260e-01
-3.08955103e-01 -1.01624668e+00 -7.61430740e-01 -6.00387752e-01
4.32325572e-01 2.27189258e-01 4.80801731e-01 9.77796018e-01
1.82831168e-01 4.62292224e-01 1.83942288e-01 -3.11886787e-01
3.33954990e-01 1.07220188e-01 6.80857837e-01 -1.27500021e+00
-6.67185783e-01 -4.66014147e-01 -5.99724054e-01 -7.07833707e-01
-6.92301244e-02 -9.73201275e-01 -2.24918857e-01 -1.31450617e+00
5.04507244e-01 -1.04098301e-02 -4.25782442e-01 7.15809047e-01
-2.27779195e-01 -5.78039885e-02 -5.82331836e-01 3.23055416e-01
-2.28398845e-01 3.53237033e-01 7.44035304e-01 9.08814296e-02
1.16533684e-02 -5.17510295e-01 -7.22969294e-01 6.73914850e-01
9.47298825e-01 -3.22038561e-01 -3.13598275e-01 7.91990682e-02
1.90037899e-02 -4.15643662e-01 1.89503103e-01 -8.38252187e-01
-3.81425738e-01 -5.25400400e-01 6.50457740e-01 -5.88944554e-01
2.85632253e-01 -4.71570075e-01 5.15023291e-01 8.52343142e-01
-5.36721587e-01 2.59875000e-01 2.91631311e-01 3.63753945e-01
6.57114014e-02 7.53546208e-02 1.22830534e+00 -3.84665698e-01
-6.03116632e-01 3.47127259e-01 -5.72231829e-01 1.44424494e-02
1.01403356e+00 -4.65591609e-01 -3.81460011e-01 9.32918489e-02
-9.27366734e-01 1.29843935e-01 1.11844063e+00 4.44302559e-02
3.74787390e-01 -6.34761333e-01 -6.29815340e-01 3.55941772e-01
2.99806356e-01 -3.21055144e-01 -3.97057198e-02 4.48506653e-01
-1.08324754e+00 6.18727028e-01 -7.47839689e-01 -6.03058875e-01
-1.68402505e+00 6.99992836e-01 6.34007096e-01 -1.24540053e-01
-4.41896707e-01 6.96930885e-01 2.73311853e-01 -6.30132437e-01
-1.17386458e-02 -6.61480427e-03 2.45539024e-01 -3.66684198e-01
5.72991967e-01 -6.53477460e-02 2.20508903e-01 -5.73983073e-01
-4.70729321e-01 -7.41706640e-02 -5.27914345e-01 4.52585995e-01
1.75528514e+00 2.50907600e-01 -1.01707056e-01 1.33560091e-01
1.23055029e+00 -6.25196934e-01 -1.36662161e+00 -1.15861660e-02
2.60568112e-01 6.95414841e-02 -5.17452776e-01 -1.11012769e+00
-9.60097909e-02 7.88151383e-01 7.14797258e-01 -3.60387415e-01
6.94484115e-01 -1.15390554e-01 7.83069730e-01 8.40567827e-01
3.26825052e-01 -8.74687314e-01 -3.56923640e-01 6.75839007e-01
5.18716156e-01 -1.37561774e+00 2.68300641e-02 1.55241311e-01
-2.06511945e-01 1.01846147e+00 6.46928608e-01 3.20297509e-01
1.04641512e-01 2.80942798e-01 1.46519005e-01 -4.10181940e-01
-9.97641504e-01 1.71810046e-01 -1.83707499e-03 9.57255781e-01
9.58015382e-01 -3.31086069e-01 -5.12512982e-01 4.11728173e-01
-1.42061636e-01 -7.95647502e-02 1.70976780e-02 1.01165271e+00
-8.40109646e-01 -1.61116695e+00 -1.39155105e-01 3.84251952e-01
-6.63314223e-01 -3.52091372e-01 -9.66071188e-01 6.24992967e-01
-3.09644919e-02 5.13376057e-01 -7.82220513e-02 -1.17037483e-01
-2.55524442e-02 4.04473066e-01 6.39993906e-01 -6.86323881e-01
-5.87230265e-01 -1.22081153e-01 2.03728199e-01 -4.00812477e-01
-1.22815855e-01 -5.00483871e-01 -1.57619321e+00 -4.00271654e-01
-2.26172522e-01 5.61924756e-01 6.32905722e-01 6.57507241e-01
8.93695593e-01 1.60535663e-01 3.59260738e-02 -7.99891651e-01
-7.44343936e-01 -8.40832531e-01 -3.25618744e-01 5.75772345e-01
2.86376387e-01 -3.63338739e-01 5.80275059e-02 3.66910100e-01] | [4.736576557159424, 5.5799384117126465] |
e2113d45-0287-4c56-98b7-49db41e40272 | imperceptible-transfer-attack-and-defense-on | 2111.10990 | null | https://arxiv.org/abs/2111.10990v2 | https://arxiv.org/pdf/2111.10990v2.pdf | Imperceptible Transfer Attack and Defense on 3D Point Cloud Classification | Although many efforts have been made into attack and defense on the 2D image domain in recent years, few methods explore the vulnerability of 3D models. Existing 3D attackers generally perform point-wise perturbation over point clouds, resulting in deformed structures or outliers, which is easily perceivable by humans. Moreover, their adversarial examples are generated under the white-box setting, which frequently suffers from low success rates when transferred to attack remote black-box models. In this paper, we study 3D point cloud attacks from two new and challenging perspectives by proposing a novel Imperceptible Transfer Attack (ITA): 1) Imperceptibility: we constrain the perturbation direction of each point along its normal vector of the neighborhood surface, leading to generated examples with similar geometric properties and thus enhancing the imperceptibility. 2) Transferability: we develop an adversarial transformation model to generate the most harmful distortions and enforce the adversarial examples to resist it, improving their transferability to unknown black-box models. Further, we propose to train more robust black-box 3D models to defend against such ITA attacks by learning more discriminative point cloud representations. Extensive evaluations demonstrate that our ITA attack is more imperceptible and transferable than state-of-the-arts and validate the superiority of our defense strategy. | ['Wei Hu', 'Daizong Liu'] | 2021-11-22 | null | null | null | null | ['point-cloud-classification'] | ['computer-vision'] | [ 2.66571462e-01 3.00403386e-02 1.83496207e-01 1.91475004e-01
-5.44820964e-01 -1.08590627e+00 7.30889678e-01 -5.68240732e-02
4.64944504e-02 9.43178963e-03 -2.32182741e-01 -4.26241159e-01
9.86585468e-02 -9.24180329e-01 -9.65385079e-01 -7.35936403e-01
-2.19308063e-01 8.60861391e-02 2.82818437e-01 -3.65055710e-01
3.04387748e-01 1.23847485e+00 -9.67005908e-01 -2.89375901e-01
9.17113185e-01 7.40478039e-01 -4.41327661e-01 4.92810369e-01
3.26267868e-01 1.52588889e-01 -1.08241916e+00 -4.75875735e-01
8.78340423e-01 1.31720202e-02 -2.38553613e-01 -1.13604888e-01
6.05247974e-01 -3.49050313e-01 -6.36672795e-01 1.46753466e+00
4.51404572e-01 -6.98346198e-02 6.99837506e-01 -1.50109553e+00
-9.68425512e-01 -8.08564797e-02 -6.36048973e-01 -5.94196171e-02
4.18201327e-01 4.14356053e-01 2.99287587e-01 -5.28221726e-01
4.07915264e-01 1.60524023e+00 4.79084939e-01 8.73054564e-01
-1.23601687e+00 -1.12176454e+00 1.86318234e-01 -1.36821583e-01
-1.52084756e+00 8.46061707e-02 1.28428853e+00 -2.45929554e-01
5.11699021e-01 7.45381534e-01 3.31908882e-01 1.48258555e+00
2.94478118e-01 3.99910420e-01 1.08757961e+00 -9.09961313e-02
2.43685573e-01 1.78308189e-02 -5.04760325e-01 2.77328908e-01
5.39322913e-01 6.81537867e-01 -3.55861709e-02 -6.63714409e-01
1.00856245e+00 2.00379550e-01 -4.67250079e-01 -6.30022824e-01
-1.08468401e+00 7.16394186e-01 9.02238071e-01 -1.78523008e-02
-1.04671039e-01 5.94924986e-02 1.74689144e-01 3.63504171e-01
3.23800862e-01 6.31690264e-01 -1.01071492e-01 3.20420623e-01
-2.81900615e-01 3.34960312e-01 5.88002145e-01 9.59098876e-01
5.42414308e-01 2.84977883e-01 2.34104559e-01 2.36829847e-01
3.41142833e-01 1.09282374e+00 -9.08857435e-02 -5.65322816e-01
6.76406801e-01 5.72968066e-01 -2.22526826e-02 -1.69262123e+00
-7.85462260e-02 -2.34114632e-01 -1.06657004e+00 8.78705382e-01
1.14373319e-01 -7.48942271e-02 -8.65629554e-01 1.66602612e+00
6.74911916e-01 5.47854722e-01 9.72827002e-02 9.26858187e-01
3.01005751e-01 6.93775177e-01 -9.86015350e-02 2.12832108e-01
8.58852327e-01 -3.09498727e-01 -2.45019510e-01 -2.48203557e-02
3.99919927e-01 -7.72318661e-01 9.16098416e-01 2.76609004e-01
-7.44867802e-01 -4.78452086e-01 -1.27054822e+00 4.75207269e-01
-3.50161254e-01 -6.32470608e-01 9.68077034e-02 1.04108298e+00
-5.54513752e-01 6.01325214e-01 -9.77198005e-01 -6.62972927e-02
5.48769414e-01 4.07671869e-01 -5.03829539e-01 1.80328891e-01
-1.16838443e+00 9.83692884e-01 -1.18193477e-02 -1.58231389e-02
-1.23704350e+00 -8.96269321e-01 -7.06080198e-01 -2.47781798e-01
2.98156619e-01 -6.22726142e-01 5.80518842e-01 -5.69532275e-01
-1.52631152e+00 7.93790042e-01 3.29752654e-01 -4.31168377e-01
8.66703629e-01 -3.03696990e-01 -6.72782838e-01 3.14997911e-01
-3.07110786e-01 4.20961618e-01 1.51200974e+00 -1.83192718e+00
-4.20805477e-02 -6.10213459e-01 1.70231014e-01 2.10153069e-02
-4.35397387e-01 3.80422967e-03 -6.93236738e-02 -1.10427773e+00
7.09547028e-02 -1.18075323e+00 -3.55041474e-01 4.21341985e-01
-8.20925057e-01 2.67338365e-01 1.42263925e+00 -2.14545637e-01
7.83081472e-01 -2.53308606e+00 1.53718323e-01 5.33234835e-01
3.33952874e-01 6.00793481e-01 -2.26997927e-01 5.44220686e-01
-2.74361163e-01 5.57286143e-01 -2.23850265e-01 -1.39008500e-02
7.20045120e-02 1.69796973e-01 -1.13625348e+00 8.37725163e-01
4.33962286e-01 8.24451983e-01 -7.94017911e-01 3.11414171e-02
4.23099369e-01 7.07898736e-01 -6.81925476e-01 1.85330257e-01
5.51530607e-02 8.00105333e-01 -9.15676057e-01 6.74950719e-01
1.28183043e+00 2.90871084e-01 -4.58258301e-01 -5.53226061e-02
2.64394879e-01 -1.80197790e-01 -1.01509225e+00 1.16030848e+00
-2.50470400e-01 1.91982970e-01 -2.31620204e-02 -5.67766428e-01
1.31385517e+00 5.56628853e-02 2.01049402e-01 -1.59633800e-01
2.34173179e-01 1.94947809e-01 -1.66682616e-01 -1.07255809e-01
1.61949083e-01 -6.52535111e-02 -2.87247956e-01 1.38035774e-01
-6.07545078e-01 -7.06614852e-01 -9.28951144e-01 7.09110796e-02
1.11248565e+00 -7.77623057e-02 -1.29526258e-01 -4.43295389e-02
7.03122914e-01 -2.36063644e-01 5.02738774e-01 7.09079981e-01
-2.20684350e-01 6.96642399e-01 3.51852894e-01 -3.95753682e-01
-1.04338217e+00 -1.52802849e+00 -6.70596883e-02 -5.58489785e-02
8.43119621e-01 -1.79397479e-01 -7.68253326e-01 -1.16674173e+00
3.99909139e-01 7.07231402e-01 -6.57762170e-01 -7.31607318e-01
-8.52519989e-01 -1.06928140e-01 9.93458152e-01 3.43430370e-01
5.60018480e-01 -4.54160213e-01 -4.03279692e-01 -2.25245714e-01
3.70564818e-01 -1.08613920e+00 -4.57840294e-01 -3.70959997e-01
-9.54028726e-01 -1.05627978e+00 -5.60924411e-01 -4.53211725e-01
1.07774603e+00 4.58112866e-01 5.73474765e-01 2.67073065e-01
-1.03707857e-01 3.60002607e-01 -3.71885568e-01 -6.97333038e-01
-7.26244807e-01 -3.29624385e-01 4.39724475e-01 6.39801621e-02
1.95758507e-01 -6.66314125e-01 -5.89713275e-01 7.44405687e-01
-1.18502557e+00 -4.58567709e-01 3.72397244e-01 4.72569495e-01
5.48162639e-01 2.51951963e-01 8.67295861e-02 -6.16752863e-01
4.10532266e-01 -2.40997970e-01 -6.80254221e-01 -1.11574106e-01
-1.72826007e-01 -2.34363019e-01 8.98560166e-01 -8.96918535e-01
-4.70665872e-01 -1.15789995e-01 -1.78516954e-01 -1.22739661e+00
-3.62295091e-01 -2.01423332e-01 -5.91370821e-01 -9.48593318e-01
8.81616652e-01 1.36635080e-01 4.68493514e-02 -3.89921397e-01
3.75702143e-01 3.07810009e-01 6.65531576e-01 -7.63530493e-01
2.29649210e+00 8.88178945e-01 3.54556084e-01 -7.97858655e-01
-3.37389320e-01 5.64541519e-02 -4.63017017e-01 -9.25178528e-02
7.10056663e-01 -6.62047803e-01 -7.26226807e-01 6.22379839e-01
-1.21697199e+00 2.72813868e-02 -1.70424536e-01 1.21848427e-01
-5.07924139e-01 8.00704062e-01 -2.61634737e-01 -5.22138357e-01
-2.05035537e-01 -1.23322964e+00 1.21735859e+00 -9.56808180e-02
-1.11749440e-01 -7.67949939e-01 9.35064107e-02 -1.05945989e-02
3.71719897e-01 1.03270662e+00 1.01738000e+00 -7.08369195e-01
-7.16278195e-01 -6.91088915e-01 4.17488515e-02 4.97317731e-01
2.16429740e-01 1.82142064e-01 -7.74620175e-01 -6.90010786e-01
4.70481128e-01 7.61989430e-02 2.52366126e-01 -2.39889219e-01
1.32469881e+00 -5.70518136e-01 -4.29100364e-01 9.37627077e-01
1.16698289e+00 1.63493931e-01 7.97584593e-01 4.81709093e-01
8.52830648e-01 2.71197766e-01 5.03832638e-01 1.49878234e-01
-2.31085822e-01 8.26948166e-01 1.04282355e+00 -2.11775631e-01
2.23989964e-01 -5.86500645e-01 3.53336096e-01 3.99062067e-01
4.48823161e-03 -1.90083876e-01 -8.89858365e-01 1.36933222e-01
-1.34080064e+00 -7.69566774e-01 1.59348115e-01 2.29118490e+00
4.42773193e-01 4.95919526e-01 -8.55692998e-02 1.85831621e-01
8.81781280e-01 4.47390944e-01 -6.28058434e-01 -2.42763355e-01
-5.08554168e-02 1.38743445e-01 7.18806386e-01 3.12780023e-01
-1.26049066e+00 9.09311235e-01 5.58424330e+00 8.40932190e-01
-1.35882330e+00 -9.98076126e-02 2.64247924e-01 8.11222009e-03
-4.49921370e-01 1.33895233e-01 -5.83553970e-01 5.03237069e-01
4.59847987e-01 -3.10574532e-01 1.82755306e-01 1.00839281e+00
7.48273954e-02 7.76995897e-01 -8.97178173e-01 7.62460053e-01
1.28331050e-01 -1.17887104e+00 6.07124209e-01 4.44725603e-01
6.70816600e-01 -3.09765220e-01 5.84387779e-01 1.75876338e-02
3.62240881e-01 -9.90683138e-01 7.01764524e-01 1.75429016e-01
6.72529459e-01 -1.01159549e+00 3.07134658e-01 4.05592233e-01
-9.54209864e-01 7.87578747e-02 -5.64373493e-01 3.19297224e-01
3.19339000e-02 1.26877740e-01 -5.49018741e-01 8.31129611e-01
6.10333920e-01 4.85654503e-01 -5.01447678e-01 9.26962376e-01
-5.91018975e-01 5.01567721e-01 -4.97286052e-01 1.76319838e-01
4.00182426e-01 -1.29298747e-01 1.44120824e+00 6.62221968e-01
4.01152760e-01 1.30607158e-01 1.85970932e-01 1.03754830e+00
-4.69643734e-02 -1.87063962e-01 -1.09254849e+00 3.29385221e-01
7.26709962e-01 8.45853984e-01 -3.61975938e-01 2.01447174e-01
-6.55332059e-02 9.12852585e-01 -8.88888761e-02 4.64746922e-01
-9.36136782e-01 -5.61866343e-01 1.22528553e+00 1.14452787e-01
2.52802253e-01 -5.50743341e-01 -3.33297551e-01 -1.18672991e+00
8.94194171e-02 -1.09549677e+00 5.87586947e-02 -4.87270832e-01
-1.61412680e+00 7.13280082e-01 -1.16534680e-02 -1.92538178e+00
1.96794882e-01 -7.19825685e-01 -9.81734335e-01 7.56095827e-01
-1.28673530e+00 -1.27262402e+00 -2.59064376e-01 8.93884540e-01
-7.67171830e-02 -1.52880013e-01 7.45191395e-01 2.86152074e-03
-3.32829684e-01 8.16921175e-01 -1.61741793e-01 3.05071473e-01
5.84950805e-01 -8.54300559e-01 1.04316819e+00 1.16837335e+00
1.00601427e-01 8.38585556e-01 6.81364655e-01 -9.01638329e-01
-1.62685370e+00 -1.29758096e+00 -3.83808604e-03 -8.71189296e-01
8.18803549e-01 -6.05727494e-01 -1.13636863e+00 4.99592572e-01
-2.02040285e-01 6.57775253e-02 3.82490307e-01 -5.05033374e-01
-8.81236017e-01 4.06082049e-02 -1.49443221e+00 1.02987897e+00
9.28570569e-01 -6.25283659e-01 -7.44768262e-01 2.58868456e-01
1.10617125e+00 -5.75953424e-01 -8.01967919e-01 7.84001350e-01
-3.71814054e-03 -6.61702871e-01 1.50591671e+00 -5.98773122e-01
1.46967217e-01 -6.92396045e-01 -1.38033152e-01 -1.20925677e+00
-1.24062903e-01 -1.11772823e+00 -2.80599207e-01 9.94595826e-01
-3.12620923e-02 -9.33951914e-01 8.72041762e-01 3.09706509e-01
-1.03289023e-01 -5.31453907e-01 -1.32417881e+00 -1.30015504e+00
5.91176629e-01 -4.12305802e-01 9.92885411e-01 1.13104308e+00
-4.70681608e-01 -4.54473138e-01 -3.47784758e-01 1.16248822e+00
8.63259137e-01 -1.55715033e-01 1.35855436e+00 -1.05527020e+00
1.56877637e-02 -4.36559379e-01 -1.06640768e+00 -1.01940751e+00
1.44595966e-01 -6.97615087e-01 -2.91975170e-01 -6.05793357e-01
-4.36215162e-01 -5.79124749e-01 -6.73212633e-02 3.61753553e-01
-2.31504038e-01 3.89193773e-01 5.15593410e-01 4.86447096e-01
7.79363140e-02 7.33881176e-01 1.25552285e+00 -3.62881005e-01
-4.74351048e-02 2.44962931e-01 -5.30958056e-01 8.70931208e-01
5.21059811e-01 -6.51089013e-01 -3.03541929e-01 -4.70679253e-01
-3.54704913e-03 -2.69897401e-01 8.35069001e-01 -1.15569186e+00
-8.72547701e-02 -2.14628741e-01 2.22158745e-01 -5.87033451e-01
3.93085063e-01 -1.41238844e+00 2.13059023e-01 6.10136986e-01
1.01663433e-01 4.55035195e-02 4.29220974e-01 9.16718602e-01
-1.32503122e-01 6.58144280e-02 1.02794218e+00 2.37856016e-01
-1.08634673e-01 6.89440548e-01 2.65803393e-02 -6.71240389e-02
1.54868603e+00 -3.87598276e-01 -5.22850037e-01 -3.16283941e-01
-3.78608823e-01 1.33680310e-02 1.09487069e+00 6.83278263e-01
9.49116766e-01 -1.47206509e+00 -6.87570870e-01 5.79259932e-01
1.10887520e-01 9.30847675e-02 2.08042488e-01 3.31922770e-01
-7.72560120e-01 -1.01812370e-01 -1.20350234e-01 -8.51273954e-01
-1.18816459e+00 1.13698065e+00 3.78825575e-01 1.35715589e-01
-8.93443465e-01 6.08553290e-01 5.55589020e-01 -6.58113122e-01
1.33544102e-01 1.60742714e-03 1.92888230e-01 -6.75271451e-01
3.63294512e-01 1.79475293e-01 -1.17802218e-01 -7.44129241e-01
-5.03472924e-01 1.05677128e+00 -1.08508468e-01 3.24063957e-01
1.08144104e+00 2.13832572e-01 8.29925612e-02 -2.52399653e-01
1.14851987e+00 4.18703228e-01 -1.38168526e+00 -1.84296090e-02
-2.61095494e-01 -1.15147901e+00 -2.63067305e-01 -2.98906356e-01
-1.01026714e+00 8.44304085e-01 5.48350155e-01 3.92454594e-01
9.44966078e-01 -1.70206010e-01 8.78612041e-01 2.58734524e-01
5.34849823e-01 -2.37422600e-01 2.07867473e-01 2.56353021e-01
1.04315197e+00 -8.47274959e-01 -4.08393033e-02 -8.31513107e-01
-4.76388216e-01 9.29698765e-01 5.72646797e-01 -6.62027121e-01
7.14220166e-01 2.40158230e-01 3.14893067e-01 -1.60615012e-01
-2.62597799e-01 5.93395531e-01 3.97431463e-01 9.60721374e-01
-6.76139951e-01 -1.44094780e-01 4.27473873e-01 1.30922124e-01
-3.57666433e-01 -7.46009707e-01 4.11225766e-01 9.10842597e-01
-1.07897758e-01 -1.12591147e+00 -9.87908244e-01 -2.79040277e-01
-2.20292047e-01 2.39886671e-01 -6.97681963e-01 9.79384303e-01
-1.33417204e-01 8.55524778e-01 -2.07522571e-01 -7.20232427e-01
6.66874468e-01 -4.35755879e-01 2.26776034e-01 -4.17547703e-01
-5.48605144e-01 -1.36063844e-01 -5.31089842e-01 -6.00727141e-01
2.67635826e-02 -4.65287030e-01 -8.85562956e-01 -5.73030353e-01
-3.34684283e-01 -5.93050849e-03 5.24794161e-01 4.99019474e-01
6.29985273e-01 1.84107751e-01 1.39168453e+00 -1.18045986e+00
-1.11855483e+00 -3.76009732e-01 -3.09169024e-01 7.44109631e-01
4.59647626e-01 -7.93764889e-01 -8.79323065e-01 -5.03228009e-01] | [7.702663898468018, -4.469666481018066] |
2827bb4d-bc21-4cfd-95bd-c2d41e77e9fd | d-terminer-online-demo-for-monolingual-and | null | null | https://aclanthology.org/2022.term-1.7 | https://aclanthology.org/2022.term-1.7.pdf | D-Terminer: Online Demo for Monolingual and Bilingual Automatic Term Extraction | This contribution presents D-Terminer: an open access, online demo for monolingual and multilingual automatic term extraction from parallel corpora. The monolingual term extraction is based on a recurrent neural network, with a supervised methodology that relies on pretrained embeddings. Candidate terms can be tagged in their original context and there is no need for a large corpus, as the methodology will work even for single sentences. With the bilingual term extraction from parallel corpora, potentially equivalent candidate term pairs are extracted from translation memories and manual annotation of the results shows that good equivalents are found for most candidate terms. Accompanying the release of the demo is an updated version of the ACTER Annotated Corpora for Term Extraction Research (version 1.5). | ['Els Lefever', 'Veronique Hoste', 'Ayla Rigouts Terryn'] | null | null | null | null | term-lrec-2022-6 | ['term-extraction'] | ['natural-language-processing'] | [ 1.06705293e-01 2.49217704e-01 -5.78563273e-01 -1.09906875e-01
-1.01654971e+00 -8.33730757e-01 7.75794327e-01 6.19103968e-01
-8.65235269e-01 8.64784479e-01 6.27372086e-01 -5.93780220e-01
-1.00224763e-01 -4.04010415e-01 -3.77442032e-01 -2.01090753e-01
-1.12265505e-01 7.15283453e-01 -2.92577833e-01 -7.74861753e-01
8.58917609e-02 2.05926418e-01 -1.10107505e+00 2.82969207e-01
7.90289938e-01 4.18077320e-01 4.38948244e-01 2.12919176e-01
-4.19850111e-01 4.48441327e-01 -4.73519087e-01 -7.10236311e-01
3.26250136e-01 -2.54721910e-01 -1.24653614e+00 -4.99273181e-01
1.72369227e-01 -2.48360112e-02 -2.19733521e-01 9.85644877e-01
4.79114622e-01 -2.05270890e-02 4.62519020e-01 -5.00979602e-01
-8.36853862e-01 1.32805574e+00 1.05603576e-01 2.32658848e-01
6.08447790e-01 -4.13448513e-01 1.35033071e+00 -1.20466542e+00
1.23936403e+00 9.81284618e-01 6.89858973e-01 2.84361035e-01
-1.19898701e+00 -4.26713407e-01 -2.40892187e-01 1.25032961e-01
-1.61477196e+00 -6.27261817e-01 5.11621237e-01 -3.52259040e-01
1.89469612e+00 1.26602158e-01 9.01030779e-01 1.24149740e+00
5.65125763e-01 3.67684036e-01 1.10246968e+00 -1.17826366e+00
-1.26259610e-01 5.08263111e-01 4.49509501e-01 2.76891947e-01
3.13319564e-01 1.50333971e-01 -4.97578889e-01 -8.87529254e-02
3.50904524e-01 -4.19823080e-01 -1.30833074e-01 1.17864013e-01
-1.37367249e+00 7.44612098e-01 -3.50295305e-02 1.16510046e+00
-4.90854055e-01 -3.25437427e-01 9.20122623e-01 7.39911318e-01
6.29896820e-01 7.16702163e-01 -5.28697610e-01 -6.55661672e-02
-1.10773468e+00 8.48039240e-02 7.57015407e-01 1.17716157e+00
6.66404068e-01 -2.63539225e-01 -9.64639112e-02 9.89761651e-01
2.98245251e-01 5.40472806e-01 1.14186454e+00 -5.78768730e-01
4.01016802e-01 6.20642483e-01 -1.38337269e-01 -7.19003260e-01
-3.61367613e-01 -4.88294423e-01 -3.49584609e-01 -4.34326917e-01
-1.57026932e-01 -2.36767098e-01 -8.52929831e-01 1.46759558e+00
4.26807962e-02 -8.78470480e-01 3.62834096e-01 3.52754027e-01
8.44303131e-01 6.31362736e-01 -3.98555584e-02 -5.55646718e-01
1.62601829e+00 -9.30944145e-01 -1.25675559e+00 -4.50090170e-01
8.54453444e-01 -1.15858924e+00 5.71969330e-01 9.26982786e-04
-1.16813302e+00 -1.99114546e-01 -1.22466612e+00 -4.50958192e-01
-1.10558224e+00 1.06562413e-01 6.19576693e-01 5.86285233e-01
-1.19982266e+00 3.41120362e-01 -6.35064960e-01 -6.23848617e-01
-3.45387578e-01 4.07562971e-01 -5.37653923e-01 6.21250309e-02
-1.80855680e+00 1.70234442e+00 9.93530691e-01 1.30692899e-01
-5.64733088e-01 -2.54659057e-01 -1.04567611e+00 -2.35217243e-01
8.62079933e-02 -3.67590249e-01 1.15459037e+00 -9.75875676e-01
-1.17730772e+00 1.23085368e+00 -2.57479578e-01 -6.00036800e-01
-6.29848242e-02 -2.54686236e-01 -7.80951262e-01 1.55223355e-01
6.08151495e-01 5.65461993e-01 3.19732100e-01 -6.77384555e-01
-5.63498378e-01 -1.93530932e-01 2.86369361e-02 3.72997403e-01
-5.46779394e-01 6.61698818e-01 -1.92216292e-01 -9.23317969e-01
-9.85142216e-02 -8.75092506e-01 -3.96640971e-02 -7.78388202e-01
6.72035739e-02 -3.46678704e-01 1.16068810e-01 -1.14047039e+00
1.49590921e+00 -1.60980248e+00 1.04562342e-01 -6.35764748e-02
-1.80673450e-01 1.84592351e-01 -1.12782747e-01 1.12800419e+00
-2.48320699e-01 2.80323029e-01 -1.16189532e-01 -6.30690157e-02
1.14583448e-01 8.33892077e-02 -3.46789122e-01 2.36620322e-01
-5.86644262e-02 9.59906876e-01 -9.51153576e-01 -6.65998578e-01
-8.49398784e-03 2.33833238e-01 1.93624616e-01 -1.46840319e-01
8.75875801e-02 -4.91961837e-02 -4.33638245e-02 5.86702287e-01
1.20832428e-01 2.76978880e-01 4.27151918e-01 4.61286493e-02
-4.98317689e-01 1.04185200e+00 -6.21766210e-01 2.05026340e+00
-9.16148484e-01 8.51164997e-01 -7.79319927e-02 -5.87598622e-01
7.72047281e-01 1.05735910e+00 2.43672431e-01 -7.79114366e-01
2.16001913e-01 7.78058052e-01 -1.25400290e-01 -4.00189549e-01
1.00807714e+00 -3.15722018e-01 -1.20730184e-01 5.33243477e-01
6.36344314e-01 -1.13168284e-02 8.10129941e-01 1.09212175e-01
8.28367352e-01 2.34924167e-01 7.28645563e-01 -5.28136313e-01
4.38592136e-01 4.95739698e-01 3.99111986e-01 3.03544700e-01
1.60083890e-01 1.64432943e-01 -1.42623201e-01 -4.88026321e-01
-1.13143480e+00 -6.36584818e-01 -7.19178498e-01 1.10969615e+00
-3.30736965e-01 -7.78440833e-01 -6.54533267e-01 -4.75079656e-01
-5.79173565e-01 9.11770165e-01 -3.39928359e-01 1.33359089e-01
-8.38672638e-01 -5.70152104e-01 8.01086783e-01 5.00792190e-02
-2.76530813e-02 -1.18643391e+00 -2.77585238e-01 5.13982296e-01
-5.87004542e-01 -1.03687453e+00 -5.11344314e-01 5.91418445e-01
-7.26995349e-01 -5.35234392e-01 -7.48308837e-01 -1.39208925e+00
3.42364609e-01 -6.94378316e-02 1.14367151e+00 -1.82175592e-01
1.78423822e-01 -9.33487937e-02 -4.96675491e-01 -2.51080662e-01
-6.07330739e-01 4.64060783e-01 3.55342925e-01 -6.59596026e-01
9.04573202e-01 -7.38672972e-01 -1.41731918e-01 -3.25071901e-01
-8.45596313e-01 -1.44873917e-01 6.87716126e-01 9.77588475e-01
4.41942394e-01 -4.72194582e-01 6.54856980e-01 -4.63325799e-01
1.02278185e+00 -4.75930870e-01 -4.83370006e-01 4.66431975e-01
-1.11498749e+00 3.25219572e-01 3.82008493e-01 -3.68368357e-01
-8.69990766e-01 -2.73393542e-01 -2.52077937e-01 5.27875125e-02
-1.32947728e-01 9.92658556e-01 2.29861721e-01 1.53460726e-01
8.89800847e-01 1.89338341e-01 -5.81077218e-01 -6.05888188e-01
6.75299168e-01 1.04876137e+00 3.54568511e-01 -3.18417698e-01
4.77706850e-01 -3.28061357e-02 -4.95168149e-01 -6.51336610e-01
-4.70866710e-01 -5.06042778e-01 -8.31431985e-01 -1.40421569e-01
8.22771788e-01 -9.77814972e-01 1.05777197e-01 -1.74633980e-01
-1.45522475e+00 2.27167711e-01 -2.13348642e-01 9.53816652e-01
-1.47506252e-01 8.34101886e-02 -8.75735343e-01 -5.85838974e-01
-7.95493186e-01 -9.37267900e-01 8.92271698e-01 -8.96947458e-02
-7.56971717e-01 -1.24752033e+00 8.66267323e-01 1.96686506e-01
3.87064606e-01 -6.51875958e-02 1.05217421e+00 -1.21433866e+00
-3.68229784e-02 -2.63700545e-01 2.60130495e-01 1.28635600e-01
2.68085986e-01 -3.16494137e-01 -7.36198425e-01 -2.68386930e-01
1.49218455e-01 -3.77109498e-01 6.06508434e-01 -1.04729377e-01
-3.86872202e-01 -5.22756159e-01 -3.15124363e-01 4.60667051e-02
1.35536742e+00 3.92516971e-01 1.87737271e-01 7.30944812e-01
5.96806407e-01 8.31125200e-01 2.10511506e-01 -2.64107227e-01
3.28116804e-01 5.32489002e-01 -2.88497597e-01 3.77766609e-01
-2.62518395e-02 -1.44569069e-01 8.71605277e-01 1.71687412e+00
2.18604207e-01 1.19298182e-01 -1.22896969e+00 1.19317937e+00
-1.54761493e+00 -8.25395763e-01 -1.07980900e-01 2.11385441e+00
1.38371634e+00 4.49629873e-03 -2.32147634e-01 -1.75507888e-01
7.52585769e-01 1.87787443e-01 1.77120239e-01 -8.72963607e-01
-1.70313194e-01 7.44967282e-01 5.42541623e-01 7.90692389e-01
-6.55057251e-01 1.12551737e+00 7.21789980e+00 9.87434328e-01
-9.91075993e-01 5.01210213e-01 -4.27892134e-02 -1.94901556e-01
-5.90529978e-01 5.03308952e-01 -8.15267742e-01 1.27409637e-01
1.61041522e+00 -6.12046063e-01 5.75873077e-01 6.44388556e-01
4.36201431e-02 5.71889542e-02 -1.04809201e+00 7.84788191e-01
2.64192253e-01 -1.27424240e+00 1.82630420e-02 3.24930921e-02
6.31663084e-01 4.24872547e-01 -3.24873835e-01 5.65378487e-01
8.84933621e-02 -8.87849689e-01 7.52904356e-01 1.89951062e-01
8.00710559e-01 -9.37818050e-01 9.97692704e-01 1.50368124e-01
-1.14063108e+00 2.65240580e-01 -3.06388140e-01 9.83835086e-02
3.91516656e-01 5.72767079e-01 -7.09540665e-01 9.63449299e-01
2.96086580e-01 5.31477332e-01 -7.03934312e-01 6.30625069e-01
-3.75739366e-01 3.99017215e-01 -2.98656493e-01 -1.20673977e-01
3.22493017e-01 -3.45761031e-01 8.59873891e-01 1.54464614e+00
2.61819750e-01 -5.43035984e-01 -5.06371930e-02 5.98260760e-01
-1.79654986e-01 8.17146122e-01 -8.47696006e-01 -5.15590668e-01
3.98415804e-01 1.23749185e+00 -6.91723049e-01 -3.85275126e-01
-4.46280122e-01 1.04919267e+00 4.70601201e-01 1.68308377e-01
-1.17472023e-01 -7.78999090e-01 -1.02462899e-02 -3.45919967e-01
2.00321674e-01 -2.21182719e-01 -1.43878683e-01 -1.26135755e+00
4.21718806e-01 -1.15899861e+00 2.67888814e-01 -5.69203436e-01
-1.23290861e+00 1.16398430e+00 2.68437654e-01 -1.07760680e+00
-8.93703878e-01 -4.57392931e-01 -2.05238611e-01 1.05728269e+00
-1.00415802e+00 -1.32029164e+00 8.43692899e-01 2.58467913e-01
5.59520066e-01 -1.96550593e-01 1.39729786e+00 5.92043757e-01
-2.41518706e-01 4.21401292e-01 2.80529320e-01 2.13621616e-01
8.51350605e-01 -1.22513163e+00 4.13095325e-01 7.44525850e-01
5.12484431e-01 1.34771609e+00 6.26354635e-01 -9.72804904e-01
-1.01023698e+00 -7.49019682e-01 2.12242603e+00 -3.69463980e-01
1.23674452e+00 -4.63417619e-01 -7.10701227e-01 8.89088750e-01
1.28444481e+00 -7.79454350e-01 8.48825872e-01 3.80175114e-01
-4.99988236e-02 1.53030530e-01 -7.56134331e-01 8.32904935e-01
9.10817266e-01 -1.16384673e+00 -1.48046112e+00 7.93290734e-01
9.61317837e-01 -1.58446819e-01 -1.08636141e+00 1.10142134e-01
6.58939362e-01 -2.62191057e-01 5.45537233e-01 -1.94159061e-01
1.17767259e-01 -9.81240347e-02 2.18730792e-02 -1.10125446e+00
-4.22849208e-02 -9.07605112e-01 5.85431159e-02 1.43748403e+00
1.05549586e+00 -4.83512163e-01 -1.28200993e-01 2.28768378e-01
-5.25818504e-02 -5.27913392e-01 -1.24501216e+00 -9.71393585e-01
3.58318031e-01 -3.89988184e-01 3.83127123e-01 1.26080108e+00
6.72448218e-01 1.06955147e+00 -1.96474940e-01 -3.21939468e-01
2.08869755e-01 -7.59524852e-02 -1.53795779e-01 -1.04832494e+00
-1.05742067e-01 -3.68284851e-01 -1.61548123e-01 -6.78645492e-01
5.10354877e-01 -1.49362206e+00 -5.75823933e-02 -1.56876266e+00
-2.65861899e-02 1.31214797e-01 -1.97125137e-01 4.43191946e-01
7.00285286e-02 1.07002586e-01 -3.02363813e-01 3.57072413e-01
-2.32413337e-01 6.30221605e-01 4.12266046e-01 -2.34584317e-01
-3.20788920e-01 -5.24445832e-01 -4.47759241e-01 4.90790009e-01
7.14774728e-01 -1.08906710e+00 -1.22501686e-01 -5.97662270e-01
6.87822640e-01 -5.46812862e-02 -2.87643254e-01 -7.20405757e-01
5.26191629e-02 1.72810063e-01 -1.21274672e-01 -6.27339602e-01
-3.94436270e-02 -8.54761660e-01 2.34425843e-01 1.97464123e-01
-5.18786311e-01 8.53154600e-01 4.02117580e-01 -2.97512170e-02
-3.92120987e-01 -8.89327288e-01 3.47089887e-01 -3.97814333e-01
-3.14979017e-01 -1.59698814e-01 -9.46150064e-01 3.41482125e-02
5.80401361e-01 2.65940223e-02 6.87184781e-02 -1.48427352e-01
-8.76365483e-01 1.10475525e-01 4.18172121e-01 5.90801239e-01
1.49646908e-01 -1.76247978e+00 -7.49280035e-01 -2.28831694e-01
2.35796526e-01 -6.06788814e-01 -4.15904820e-01 8.13161790e-01
-4.69823360e-01 9.14838076e-01 -2.73478359e-01 8.25755522e-02
-1.02958477e+00 7.79522300e-01 3.22567552e-01 -7.25105047e-01
-5.04311085e-01 3.03696513e-01 -5.31543851e-01 -5.33424556e-01
-6.94541782e-02 -2.22970411e-01 -5.69144428e-01 3.93040895e-01
5.44440627e-01 1.77141234e-01 4.61568832e-01 -1.10645545e+00
-4.86496061e-01 2.58373410e-01 -4.20618057e-01 -9.87308443e-01
1.36183226e+00 -4.66002792e-01 -6.71165407e-01 9.19091284e-01
1.41468155e+00 3.51897001e-01 3.84054869e-01 -1.91860586e-01
5.36278725e-01 2.69527465e-01 2.49320775e-01 -7.28384435e-01
-3.74994814e-01 3.55188131e-01 6.88252211e-01 1.55113876e-01
7.97904909e-01 -9.71373618e-02 9.25655246e-01 7.43911922e-01
3.42128009e-01 -1.50312006e+00 -8.38273406e-01 1.10756993e+00
9.06351507e-01 -9.42018867e-01 -2.09351815e-02 2.62752801e-01
-2.14080930e-01 1.20588422e+00 -7.11699054e-02 6.86564436e-03
5.55594385e-01 6.09128922e-02 4.34308380e-01 -4.69795704e-01
-6.29242361e-01 -1.71313927e-01 4.11492705e-01 3.95940721e-01
9.62274075e-01 -1.64475664e-01 -1.34394753e+00 4.93632764e-01
-4.85311508e-01 -3.53651404e-01 1.70315593e-01 1.17602932e+00
-4.88347337e-02 -1.81758833e+00 -3.74936581e-01 1.41485453e-01
-8.84705842e-01 -9.48334455e-01 -7.91792035e-01 6.90742373e-01
1.60477102e-01 1.07551646e+00 -2.30969101e-01 -1.12888463e-01
5.08899875e-02 8.04563344e-01 3.28986853e-01 -6.92665458e-01
-1.24265146e+00 2.30459094e-01 6.04705751e-01 -2.79927790e-01
-7.47193396e-01 -4.78039324e-01 -1.04235744e+00 7.55297393e-03
-6.07256711e-01 6.19416714e-01 7.79173672e-01 1.21622550e+00
2.63307810e-01 2.94473380e-01 3.45926255e-01 -7.39743590e-01
4.73643318e-02 -1.42190790e+00 -1.95729315e-01 -9.35808271e-02
1.58571497e-01 -2.47457385e-01 -3.35002154e-01 1.08266838e-01] | [10.53520679473877, 9.793434143066406] |
85899996-3de8-48ff-99ab-69be70dac3ec | learning-semantic-person-image-generation-by | 2104.06650 | null | https://arxiv.org/abs/2104.06650v1 | https://arxiv.org/pdf/2104.06650v1.pdf | Learning Semantic Person Image Generation by Region-Adaptive Normalization | Human pose transfer has received great attention due to its wide applications, yet is still a challenging task that is not well solved. Recent works have achieved great success to transfer the person image from the source to the target pose. However, most of them cannot well capture the semantic appearance, resulting in inconsistent and less realistic textures on the reconstructed results. To address this issue, we propose a new two-stage framework to handle the pose and appearance translation. In the first stage, we predict the target semantic parsing maps to eliminate the difficulties of pose transfer and further benefit the latter translation of per-region appearance style. In the second one, with the predicted target semantic maps, we suggest a new person image generation method by incorporating the region-adaptive normalization, in which it takes the per-region styles to guide the target appearance generation. Extensive experiments show that our proposed SPGNet can generate more semantic, consistent, and photo-realistic results and perform favorably against the state of the art methods in terms of quantitative and qualitative evaluation. The source code and model are available at https://github.com/cszy98/SPGNet.git. | ['WangMeng Zuo', 'Dongliang He', 'Tianwei Lin', 'Fu Li', 'Xin Li', 'Xiaoming Li', 'Zhengyao Lv'] | 2021-04-14 | null | http://openaccess.thecvf.com//content/CVPR2021/html/Lv_Learning_Semantic_Person_Image_Generation_by_Region-Adaptive_Normalization_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Lv_Learning_Semantic_Person_Image_Generation_by_Region-Adaptive_Normalization_CVPR_2021_paper.pdf | cvpr-2021-1 | ['pose-transfer'] | ['computer-vision'] | [ 2.27186948e-01 -1.76376089e-01 2.47164339e-01 -4.88989294e-01
-4.54308212e-01 -2.46468857e-01 4.34515625e-01 -3.31030428e-01
-2.69853830e-01 6.37765169e-01 2.48338819e-01 4.04614478e-01
2.06038743e-01 -8.30832541e-01 -6.80132866e-01 -6.03523195e-01
7.39361525e-01 5.44490099e-01 5.39220095e-01 -3.30290765e-01
5.78655936e-02 1.86248779e-01 -1.39933264e+00 8.50626752e-02
1.00470030e+00 8.92906189e-01 2.77525008e-01 1.60695180e-01
-9.58981812e-02 2.59216577e-01 -5.13517320e-01 -7.86957800e-01
3.27580184e-01 -6.60045445e-01 -5.46424866e-01 2.04151511e-01
4.50760722e-01 -2.32156530e-01 -2.17589736e-01 1.23363435e+00
7.83222914e-01 1.18207991e-01 4.39015120e-01 -1.33258140e+00
-5.21619558e-01 1.11867681e-01 -9.37605023e-01 -3.07349950e-01
5.40381968e-01 1.45975500e-01 4.66229647e-01 -7.01189399e-01
6.02884829e-01 1.58265066e+00 5.16362607e-01 6.80073500e-01
-9.79124963e-01 -9.49125111e-01 3.20182800e-01 2.67280072e-01
-1.47393334e+00 -2.55964249e-01 9.83256996e-01 -3.10959488e-01
4.12191590e-03 2.37097591e-01 8.18944097e-01 1.24956369e+00
6.13782601e-03 7.61253357e-01 1.34513438e+00 -4.44495559e-01
1.52350673e-02 2.55163968e-01 -2.40926728e-01 6.79321826e-01
1.74449086e-01 1.84965541e-03 -3.18180531e-01 2.17988387e-01
1.06157422e+00 3.02815139e-02 -4.02893901e-01 -4.82815623e-01
-1.17939651e+00 5.28837085e-01 6.70210779e-01 1.37934506e-01
-2.53957748e-01 -3.01542543e-02 3.36475283e-01 -2.45662704e-01
5.38384438e-01 1.01492174e-01 -2.21511126e-01 2.91890353e-02
-5.81519067e-01 5.12221873e-01 3.01657140e-01 9.32641268e-01
5.98203838e-01 -2.29425043e-01 -3.15784156e-01 1.10632193e+00
3.72778326e-01 7.65978754e-01 2.96401650e-01 -7.02057838e-01
6.15280569e-01 6.06475592e-01 2.63058841e-01 -1.29686546e+00
-2.18315348e-01 -5.67558587e-01 -8.08232069e-01 1.09524019e-01
5.39661765e-01 -1.08228572e-01 -9.39931035e-01 1.74947083e+00
6.04787946e-01 -2.67437380e-02 -2.03977272e-01 1.35361576e+00
8.03294837e-01 8.10857475e-01 2.79133111e-01 3.87243591e-02
1.59537244e+00 -1.17842472e+00 -7.59332716e-01 -5.89253426e-01
2.66627371e-02 -1.07185507e+00 1.14232850e+00 2.39260674e-01
-1.00885439e+00 -7.21947312e-01 -8.20799112e-01 1.27966832e-02
-4.30992134e-02 4.75701541e-01 4.91255999e-01 5.28145134e-01
-7.30156004e-01 2.49974817e-01 -6.21857941e-01 -6.23156607e-01
2.92997062e-01 3.32681499e-02 -4.27609921e-01 -2.59871066e-01
-1.33988106e+00 8.43229711e-01 4.79404956e-01 3.78276139e-01
-3.82548779e-01 -2.56468415e-01 -8.17680657e-01 -2.00277135e-01
5.06587803e-01 -8.14348400e-01 1.13150322e+00 -1.25190222e+00
-1.59661448e+00 7.40877211e-01 -7.77012110e-02 2.45516509e-01
8.47443819e-01 -2.48089164e-01 -3.11267018e-01 3.09294555e-02
2.59824157e-01 9.49311495e-01 6.82258606e-01 -1.57365441e+00
-5.78047276e-01 -5.57475150e-01 -9.63288993e-02 6.91340983e-01
-3.69902968e-01 2.28136382e-03 -1.12386882e+00 -8.72718275e-01
1.18108533e-01 -1.05089867e+00 -1.56122848e-01 2.77975380e-01
-4.03945595e-01 1.69704501e-02 5.42573690e-01 -9.53446448e-01
8.95159960e-01 -1.97855210e+00 3.36151034e-01 7.12218881e-02
-1.45117924e-01 2.34404132e-01 -1.34255206e-02 2.02321902e-01
1.12543702e-01 -2.40383878e-01 -2.22081676e-01 -5.63462198e-01
-1.07976623e-01 -9.49614346e-02 5.24134226e-02 2.65384436e-01
-1.26424404e-02 7.33583212e-01 -7.65335917e-01 -7.22439885e-01
4.43008035e-01 6.78418934e-01 -3.54376256e-01 3.96604449e-01
-1.12862810e-01 8.53695750e-01 -7.00896323e-01 4.48448122e-01
9.88038480e-01 -5.90362847e-02 -4.90287021e-02 -4.99597639e-01
6.90693930e-02 -2.73079813e-01 -1.35293031e+00 1.77606750e+00
-2.50427872e-01 1.16804570e-01 8.56791530e-03 -6.14753664e-01
9.48365033e-01 1.21831700e-01 3.51490170e-01 -8.16954911e-01
3.44787478e-01 1.87379122e-01 -1.70440674e-01 -4.21412438e-01
3.03920209e-01 -3.28779191e-01 -4.60756347e-02 6.37371838e-02
-2.88299680e-01 -5.43204248e-02 8.96821767e-02 -9.54626724e-02
3.52291256e-01 6.00521684e-01 -9.16504767e-03 -2.02822965e-02
7.97789931e-01 -4.28048000e-02 7.87793696e-01 8.13204348e-02
-1.74313113e-01 1.05633116e+00 2.06210405e-01 -3.48814905e-01
-1.00103605e+00 -1.10478318e+00 9.93701667e-02 8.05843413e-01
7.97414601e-01 -2.01551437e-01 -1.14849722e+00 -5.83437026e-01
-2.38884628e-01 4.31464702e-01 -4.66230541e-01 -1.48117587e-01
-5.93560457e-01 -6.90640867e-01 2.85367161e-01 5.66052854e-01
1.01015294e+00 -1.25243187e+00 -2.21517220e-01 5.91154695e-02
-6.65654063e-01 -1.17845440e+00 -7.88883746e-01 -7.06812978e-01
-6.36415660e-01 -8.95550609e-01 -1.35287809e+00 -9.69045997e-01
1.08690965e+00 1.76343888e-01 6.78905487e-01 -6.59165680e-02
-2.78028816e-01 4.87983739e-03 -5.49275756e-01 -3.20431918e-01
-1.57249570e-01 -3.58999078e-03 -6.66957423e-02 2.92580903e-01
8.36443305e-02 -3.08388293e-01 -9.83139932e-01 6.81348145e-01
-8.05419862e-01 6.80536330e-01 7.35400200e-01 5.35680234e-01
6.18777633e-01 7.63418078e-02 3.04568738e-01 -7.20655143e-01
4.15022731e-01 -3.50956842e-02 -3.89559627e-01 2.96264261e-01
-4.09740180e-01 -1.70309246e-01 6.70529306e-01 -3.82089704e-01
-1.55818844e+00 1.65616199e-01 -3.52662742e-01 -2.60690212e-01
-2.83683002e-01 -2.57103331e-03 -6.46901488e-01 -7.17970077e-03
4.06289160e-01 2.89819032e-01 1.51178107e-01 -5.90158343e-01
1.79044500e-01 5.86712837e-01 5.04439890e-01 -7.79073060e-01
9.91519511e-01 4.89266455e-01 -2.85465628e-01 -3.74357522e-01
-8.16600025e-01 -2.33021811e-01 -5.61478376e-01 -4.54482138e-01
1.10171032e+00 -9.71087337e-01 -4.00730342e-01 8.82533073e-01
-1.30246174e+00 -1.37386829e-01 2.98707448e-02 4.58116978e-01
-4.71341819e-01 4.71912652e-01 -4.86350566e-01 -6.15207255e-01
-4.22161579e-01 -1.26084852e+00 1.27214181e+00 5.72027326e-01
2.62467330e-03 -6.94445848e-01 -4.03900892e-02 7.36102581e-01
3.23329329e-01 3.68453592e-01 6.18989468e-01 3.60291498e-03
-4.14458692e-01 -1.36027649e-01 -4.30102289e-01 4.04052019e-01
2.14225024e-01 -3.35218966e-01 -7.25928664e-01 -3.32255304e-01
-1.97304651e-01 -1.15072958e-01 5.13114870e-01 2.40536541e-01
1.00381768e+00 -2.80774124e-02 -3.74743640e-01 5.84766686e-01
1.31720591e+00 1.59590006e-01 7.94925213e-01 3.32427204e-01
9.03393269e-01 8.47283065e-01 1.02059376e+00 2.27242172e-01
4.66954350e-01 1.08458543e+00 1.23156674e-01 -4.44959700e-01
-3.83676648e-01 -6.15503848e-01 2.92516083e-01 6.41131520e-01
-3.82659435e-01 -2.88677216e-01 -6.54718161e-01 2.05676064e-01
-1.99130929e+00 -8.42923760e-01 -9.18050855e-02 2.22392392e+00
7.62222528e-01 1.02811262e-01 2.18301252e-01 -1.57393828e-01
1.13799500e+00 2.64572762e-02 -3.32664460e-01 9.69988480e-02
3.69132347e-02 -9.53223184e-02 3.16525728e-01 3.55678767e-01
-1.01875353e+00 1.08909702e+00 4.74418211e+00 9.84665811e-01
-1.00919247e+00 2.03163370e-01 7.49630809e-01 1.89467639e-01
-1.52785748e-01 -1.21386498e-01 -7.73936689e-01 7.30389416e-01
1.04068272e-01 2.14522891e-02 1.96358100e-01 7.53500700e-01
4.22351956e-01 -1.95799489e-02 -6.87389135e-01 1.22696972e+00
2.93618798e-01 -7.56371975e-01 2.02349022e-01 -7.03424811e-02
6.78976655e-01 -7.88056076e-01 -1.35260737e-02 1.81070328e-01
-1.82036608e-01 -8.38968933e-01 1.06869960e+00 6.88596904e-01
8.83068681e-01 -8.35297823e-01 9.25093174e-01 3.04967105e-01
-1.40734160e+00 1.84278280e-01 -3.48835617e-01 7.99274370e-02
4.44705814e-01 3.68584007e-01 -3.21876287e-01 7.61587381e-01
7.88443446e-01 6.05139256e-01 -7.26024151e-01 1.14344323e+00
-4.14842933e-01 8.58682692e-02 -5.50183393e-02 7.98406303e-02
-1.99414134e-01 -4.29158360e-01 3.87730747e-01 9.12211657e-01
4.28141505e-01 5.32276817e-02 4.21135008e-01 8.68116140e-01
1.85770586e-01 4.65066552e-01 -2.24761441e-01 3.52354556e-01
3.55589658e-01 1.24926889e+00 -7.85463452e-01 -2.73089081e-01
-2.38438681e-01 1.40716195e+00 1.12528302e-01 2.90132612e-01
-1.17139256e+00 -2.64852971e-01 3.60843718e-01 5.50920248e-01
-7.60667548e-02 9.24830064e-02 -2.24427015e-01 -1.30030155e+00
3.97967547e-01 -8.78472149e-01 2.90761411e-01 -9.24778163e-01
-1.30494142e+00 6.52603269e-01 1.88248321e-01 -1.45527184e+00
2.06725840e-02 -4.18382376e-01 -5.41309774e-01 9.53962684e-01
-1.14548337e+00 -1.50022268e+00 -7.72731185e-01 5.48447967e-01
6.29059076e-01 9.56588835e-02 5.20095468e-01 5.88855147e-01
-7.34643340e-01 7.47260153e-01 -2.63268858e-01 1.37067527e-01
1.09597337e+00 -9.18087184e-01 3.24536860e-01 8.95955324e-01
-3.63060534e-01 4.67130035e-01 8.37960541e-01 -7.80153990e-01
-9.56429839e-01 -1.14357507e+00 4.83372092e-01 -2.61450350e-01
7.94821009e-02 -3.13591123e-01 -6.07568383e-01 4.70912129e-01
-5.40448027e-03 -1.93351388e-01 1.04985654e-01 -1.95369303e-01
-9.84464511e-02 -3.85177284e-01 -1.17972767e+00 8.09198260e-01
1.15231645e+00 6.34618625e-02 -4.13964748e-01 1.96879566e-01
5.15581429e-01 -6.38943017e-01 -5.99204719e-01 4.95882571e-01
5.84006548e-01 -9.45728600e-01 1.04843318e+00 1.66481987e-01
5.09269476e-01 -6.55685961e-01 1.50583655e-01 -1.27227390e+00
-4.22180265e-01 -1.86153874e-01 3.72486174e-01 1.45711470e+00
1.46244481e-01 -6.27447724e-01 8.29285860e-01 5.09855926e-01
2.61348803e-02 -6.61052883e-01 -5.93028545e-01 -6.96546257e-01
-9.75788236e-02 4.50939015e-02 5.39488435e-01 5.89354992e-01
-3.87703747e-01 3.84127736e-01 -6.96079195e-01 8.34112689e-02
7.97877610e-01 1.91990748e-01 9.96244729e-01 -1.08058429e+00
-3.36913556e-01 -3.03297013e-01 -5.42868018e-01 -1.07544601e+00
-1.42902076e-01 -5.94134331e-01 2.69306481e-01 -1.80787528e+00
5.00266612e-01 -4.62876529e-01 -7.25268340e-03 4.70965952e-01
-5.02416730e-01 5.33000112e-01 2.85454631e-01 2.78020710e-01
-4.86337364e-01 8.19971919e-01 1.73294997e+00 2.64496338e-02
-2.16498207e-02 1.46810785e-02 -7.17656851e-01 7.26720929e-01
7.91789174e-01 -1.72226369e-01 -5.83358705e-01 -4.61558402e-01
-1.62164703e-01 -7.39151016e-02 5.47342598e-01 -1.26520836e+00
-1.02606289e-01 -2.18597829e-01 6.99640572e-01 -5.04271030e-01
6.07532442e-01 -7.66258657e-01 4.07846153e-01 5.94935775e-01
-8.24064538e-02 1.28278844e-02 4.09730002e-02 5.55529356e-01
-2.03544810e-01 1.45796344e-01 1.05752885e+00 -2.78617918e-01
-7.80902684e-01 5.52612185e-01 3.56509954e-01 -3.19704600e-02
1.19294560e+00 -2.97438532e-01 -2.06609234e-01 -4.52627033e-01
-5.55210054e-01 2.92355508e-01 7.80173600e-01 7.73363233e-01
6.99368477e-01 -1.54908776e+00 -8.38980198e-01 1.15020759e-01
2.26307556e-01 1.38528481e-01 6.88829720e-01 7.61016607e-01
-7.05234051e-01 6.57150894e-02 -5.01158416e-01 -5.15126586e-01
-1.29709411e+00 4.51008171e-01 3.72759610e-01 -1.20250441e-01
-7.47392595e-01 6.89118445e-01 7.10129261e-01 -4.37955111e-01
1.53274640e-01 2.58502096e-01 -1.65606260e-01 -2.81683475e-01
4.34931785e-01 2.15211511e-01 -2.27729529e-01 -9.66483533e-01
-3.41218352e-01 1.04413044e+00 -5.64371273e-02 -2.08533049e-01
9.87887084e-01 -2.69225806e-01 8.72677565e-03 3.88735197e-02
9.20708418e-01 8.48327647e-04 -1.33750570e+00 -3.52012292e-02
-5.94280839e-01 -7.96445429e-01 -4.23561692e-01 -9.71595466e-01
-1.24784493e+00 7.77155757e-01 8.22791159e-01 -3.55558753e-01
1.17943144e+00 -1.32130742e-01 1.09397185e+00 -4.00956661e-01
6.88350022e-01 -1.05595243e+00 1.82010263e-01 2.03959122e-01
1.00958741e+00 -1.17669165e+00 -5.66917434e-02 -1.03550553e+00
-7.74273217e-01 8.14821422e-01 1.18168962e+00 -8.42825323e-02
1.75699204e-01 -1.99777573e-01 2.60844529e-01 -2.59617157e-02
-4.84065227e-02 -7.49000832e-02 4.15620565e-01 5.81940353e-01
3.76541078e-01 1.31554425e-01 -5.53381562e-01 6.07865393e-01
-1.80793986e-01 -1.07353143e-01 4.67554443e-02 6.15899324e-01
-3.18043858e-01 -1.35235310e+00 -8.00144792e-01 2.30543073e-02
-1.18314520e-01 1.90088809e-01 -2.77448028e-01 7.01518536e-01
4.07329440e-01 8.17611933e-01 -2.45578781e-01 -4.39456731e-01
6.62625611e-01 -1.36836618e-01 5.65612674e-01 -6.14619911e-01
-2.37923384e-01 3.10017347e-01 -2.91674528e-02 -5.24859190e-01
-2.51632601e-01 -4.83620346e-01 -1.15589404e+00 -2.88145930e-01
-1.38931334e-01 1.12864584e-01 5.30457258e-01 7.54104018e-01
2.25423425e-01 5.53894818e-01 5.21157980e-01 -1.02619803e+00
-3.03532243e-01 -9.69153106e-01 -4.02119726e-01 9.48639452e-01
-3.34970802e-01 -8.94661725e-01 -1.26409397e-01 4.43902127e-02] | [11.9827241897583, -0.860615611076355] |
2b81653a-9d88-427e-84f0-92bb73d2e330 | emph-mensa-mix-up-ensemble-average-for | 2304.01554 | null | https://arxiv.org/abs/2304.01554v2 | https://arxiv.org/pdf/2304.01554v2.pdf | MEnsA: Mix-up Ensemble Average for Unsupervised Multi Target Domain Adaptation on 3D Point Clouds | Unsupervised domain adaptation (UDA) addresses the problem of distribution shift between the unlabelled target domain and labelled source domain. While the single target domain adaptation (STDA) is well studied in the literature for both 2D and 3D vision tasks, multi-target domain adaptation (MTDA) is barely explored for 3D data despite its wide real-world applications such as autonomous driving systems for various geographical and climatic conditions. We establish an MTDA baseline for 3D point cloud data by proposing to mix the feature representations from all domains together to achieve better domain adaptation performance by an ensemble average, which we call Mixup Ensemble Average or MEnsA. With the mixed representation, we use a domain classifier to improve at distinguishing the feature representations of source domain from those of target domains in a shared latent space. In empirical validations on the challenging PointDA-10 dataset, we showcase a clear benefit of our simple method over previous unsupervised STDA and MTDA methods by large margins (up to 17.10% and 4.76% on averaged over all domain shifts). | ['Jonghyun Choi', 'Ashish Sinha'] | 2023-04-04 | null | null | null | null | ['multi-target-domain-adaptation'] | ['computer-vision'] | [ 2.88796335e-01 -3.44618894e-02 -1.55459523e-01 -3.96715581e-01
-8.81809056e-01 -8.69410992e-01 1.14909732e+00 6.49295300e-02
-2.52863348e-01 7.42223084e-01 -2.38170219e-03 -1.47925511e-01
-2.88573533e-01 -5.39848447e-01 -6.47498190e-01 -9.69518840e-01
6.99282512e-02 8.64546359e-01 4.72654372e-01 -1.79065108e-01
1.98615007e-02 6.60905063e-01 -1.54948294e+00 -1.03275150e-01
1.03316748e+00 7.94412851e-01 1.40630648e-01 2.07296342e-01
-1.96027756e-01 1.43637091e-01 -5.49698412e-01 4.46363240e-02
6.61127746e-01 -2.85603762e-01 -5.83652377e-01 3.02385598e-01
4.64051902e-01 -4.83118445e-02 1.42155811e-02 8.64746213e-01
6.55833185e-01 1.61379755e-01 1.32109785e+00 -1.41680825e+00
-3.43802571e-01 -2.41283476e-02 -7.60491729e-01 1.45397097e-01
-1.44762406e-03 3.68604697e-02 5.96655250e-01 -9.16669309e-01
8.25958967e-01 1.14038193e+00 6.90215528e-01 4.41000134e-01
-1.46295989e+00 -8.89297366e-01 2.24842459e-01 6.30290806e-02
-1.28866446e+00 -4.05920655e-01 8.65395904e-01 -9.18033540e-01
9.33302879e-01 -2.64596581e-01 2.02836856e-01 1.31387830e+00
1.43843532e-01 6.97074950e-01 1.36309719e+00 -3.10028493e-01
4.97197121e-01 3.55227083e-01 8.46055709e-03 -7.90595040e-02
1.57295927e-01 2.70088166e-01 -4.67527121e-01 -1.47773758e-01
5.21078706e-01 -1.82495743e-01 9.43404883e-02 -1.13665271e+00
-1.31128907e+00 8.88258278e-01 3.59621525e-01 -6.14076247e-03
-4.53645438e-01 -5.40534854e-01 3.26349229e-01 4.91306067e-01
9.11369860e-01 5.20639002e-01 -6.31189525e-01 -5.61369881e-02
-7.96082020e-01 6.18446708e-01 5.30033886e-01 1.10778403e+00
9.29893970e-01 7.36896973e-03 9.52305570e-02 1.29572809e+00
1.52687982e-01 9.32290316e-01 5.03387928e-01 -7.93612778e-01
3.08104515e-01 7.56327331e-01 2.50122964e-01 -4.85064924e-01
-4.23755825e-01 -4.80653346e-01 -6.63043618e-01 6.00695133e-01
4.61284041e-01 -2.26559982e-01 -1.22174966e+00 1.72052586e+00
6.70938730e-01 1.82376474e-01 3.81408423e-01 8.21400821e-01
6.62996769e-01 5.62112689e-01 1.04866363e-01 1.98288709e-02
1.02670920e+00 -6.49979889e-01 -2.01868728e-01 -5.75542986e-01
6.73253953e-01 -5.86906493e-01 6.68618083e-01 8.26617852e-02
-4.76885736e-01 -6.86073363e-01 -1.07766259e+00 2.72618532e-01
-5.87534785e-01 -7.69825876e-02 1.23470321e-01 4.73311126e-01
-7.68185496e-01 2.20714629e-01 -5.14789104e-01 -9.64423060e-01
5.22969604e-01 3.02937567e-01 -6.48404419e-01 -9.31762233e-02
-1.07198381e+00 1.13497424e+00 4.92699355e-01 -6.80311084e-01
-9.58330333e-01 -1.06000209e+00 -7.42736757e-01 -5.37689090e-01
2.06499726e-01 -6.27646327e-01 1.19842434e+00 -7.53701866e-01
-1.60965300e+00 1.18928826e+00 -1.47381559e-01 -6.94370627e-01
4.36786830e-01 -2.81959921e-01 -5.42920589e-01 -1.86391503e-01
3.19428891e-01 8.31607103e-01 1.09401810e+00 -1.37351978e+00
-8.62060964e-01 -5.32418251e-01 -2.27388710e-01 3.59210581e-01
-2.34297905e-02 -3.70666772e-01 1.78398825e-02 -6.41328990e-01
1.90876976e-01 -1.33168435e+00 -6.20379634e-02 -1.18451677e-01
5.33183552e-02 -3.68552893e-01 1.07363451e+00 -2.32210428e-01
6.28634393e-01 -2.37263250e+00 2.82741666e-01 1.08697988e-01
1.59239009e-01 2.81238109e-01 -1.28383145e-01 3.86247784e-01
-3.35003912e-01 -3.85812670e-01 -7.54540384e-01 -2.45899335e-01
7.54331425e-02 4.64785159e-01 -4.77055848e-01 5.69018304e-01
4.88559276e-01 4.92112458e-01 -8.55454743e-01 -2.67696619e-01
3.91816676e-01 1.64860412e-01 -3.98427874e-01 1.23766296e-01
-1.16892099e-01 7.72226930e-01 -3.74673486e-01 5.17211795e-01
1.00744796e+00 1.98614839e-02 -6.00787960e-02 2.91606784e-01
1.93384066e-02 1.89376831e-01 -1.09356856e+00 1.76215577e+00
-2.39070475e-01 6.77004457e-01 -1.57241285e-01 -1.01538265e+00
1.48297703e+00 1.03091426e-01 6.79193079e-01 -7.80995667e-01
-1.43980712e-01 4.51027125e-01 5.97869139e-03 -7.12156445e-02
4.77083832e-01 -5.07904708e-01 -4.73508865e-01 1.13428622e-01
3.45357358e-01 -2.95649916e-01 -3.32130700e-01 3.92360687e-02
1.00162637e+00 3.55461985e-01 4.90581363e-01 -5.07676423e-01
4.46553946e-01 4.41977382e-01 4.68904614e-01 7.52399802e-01
-5.72340548e-01 8.09815943e-01 2.81227082e-01 -2.11464629e-01
-1.25487053e+00 -1.19580829e+00 -3.74689341e-01 1.10354888e+00
3.23370636e-01 1.87158018e-01 -4.51495051e-01 -9.79075253e-01
5.04866779e-01 8.83074284e-01 -5.81493437e-01 -4.18456107e-01
-2.39158198e-01 -6.85446799e-01 4.16788161e-01 4.33376908e-01
5.88678837e-01 -7.00048625e-01 -3.87163222e-01 1.28228724e-01
1.52080327e-01 -1.24353850e+00 4.51222658e-02 5.26715219e-01
-9.18115973e-01 -7.35063910e-01 -1.04905999e+00 -6.11300290e-01
3.10763359e-01 2.92086959e-01 1.18688035e+00 -1.04747307e+00
2.98856974e-01 5.12521088e-01 -4.37765986e-01 -7.53540754e-01
-6.23739541e-01 2.99534440e-01 4.67370301e-01 4.90440652e-02
7.30338514e-01 -6.85933232e-01 -2.17962638e-01 5.15566051e-01
-5.74801266e-01 -2.41231129e-01 7.30096400e-01 6.91217840e-01
5.45509040e-01 -1.04977459e-01 7.48426676e-01 -9.74304318e-01
4.07742202e-01 -9.05694544e-01 -4.74773288e-01 -1.82463974e-01
-6.98687196e-01 1.18897237e-01 2.34885678e-01 -5.00362337e-01
-1.05403841e+00 2.32195124e-01 1.11884728e-01 -7.30373979e-01
-8.06598485e-01 1.37283534e-01 -2.68678367e-01 2.40735747e-02
1.00056279e+00 2.95792043e-01 1.82586253e-01 -5.96841216e-01
3.41359586e-01 7.55326867e-01 5.95779955e-01 -5.98441422e-01
1.09855604e+00 4.85528708e-01 8.47659782e-02 -9.46498752e-01
-5.44195354e-01 -9.18287516e-01 -1.00708842e+00 -7.79275596e-02
8.33269715e-01 -1.21886504e+00 2.45582119e-01 6.05125487e-01
-9.29285467e-01 -3.06469291e-01 -4.98067945e-01 4.88503486e-01
-6.79061472e-01 2.99761683e-01 4.42609936e-01 -6.01744533e-01
2.13926300e-01 -1.03439832e+00 1.30764234e+00 1.03489999e-02
-3.40123415e-01 -1.04054737e+00 4.77162778e-01 -6.44231215e-02
2.08474085e-01 5.51561952e-01 8.63759220e-01 -1.18039870e+00
-1.44875403e-02 -7.99084604e-02 -1.61189616e-01 4.78917927e-01
2.61194587e-01 -4.44411278e-01 -1.12227690e+00 -3.53100002e-01
-2.36878879e-02 -7.06481561e-02 9.01473701e-01 4.09215093e-01
5.54740548e-01 4.21269208e-01 -4.91310120e-01 3.93703341e-01
1.24950778e+00 3.14375997e-01 2.93179482e-01 6.10210121e-01
4.96474862e-01 7.09540129e-01 8.96741390e-01 4.49264854e-01
3.49666715e-01 8.36702943e-01 5.01352906e-01 -4.51405458e-02
-3.00823957e-01 -1.28818840e-01 4.18709785e-01 3.82887185e-01
1.15704335e-01 1.38109876e-02 -1.39759958e+00 1.03102732e+00
-1.78281367e+00 -5.44432044e-01 -8.67533777e-03 2.34425449e+00
4.13032591e-01 2.65064418e-01 5.33434153e-01 -9.21843275e-02
7.33058274e-01 3.63313407e-02 -9.91773784e-01 -2.85850763e-01
-2.49926597e-01 1.25163823e-01 8.58705699e-01 1.65618926e-01
-1.47721601e+00 7.78184950e-01 6.17078876e+00 9.08394754e-01
-9.34713364e-01 9.86186266e-02 6.53966144e-02 -2.05289796e-02
-3.08216596e-03 -3.87981124e-02 -8.25009406e-01 3.59259814e-01
9.93465602e-01 -2.50629783e-01 -5.45672337e-05 1.09396827e+00
-4.25097868e-02 4.69503328e-02 -1.06171024e+00 1.01981366e+00
-1.79820105e-01 -8.57320547e-01 4.02660295e-02 4.63849545e-01
1.08476818e+00 4.88866061e-01 1.23236790e-01 6.64586902e-01
5.32128513e-01 -4.89793897e-01 6.39026344e-01 2.39672393e-01
7.50940084e-01 -7.67556489e-01 6.97893620e-01 6.14256263e-01
-9.08419013e-01 -2.79080980e-02 -4.94474918e-01 1.01706106e-02
-1.75638899e-01 6.49365187e-01 -1.17501271e+00 8.60194504e-01
7.69580245e-01 1.17470813e+00 -6.05652511e-01 9.17651117e-01
1.80511177e-01 6.49196088e-01 -4.53186095e-01 4.35538888e-01
3.74905288e-01 -1.73410892e-01 1.16342723e+00 1.02365124e+00
4.24784869e-01 -2.41978601e-01 8.08757618e-02 4.77867186e-01
1.79943621e-01 -1.43134490e-01 -1.20707798e+00 1.60669670e-01
5.24903655e-01 8.24109852e-01 -3.97633612e-01 -3.24409902e-01
-1.98343873e-01 9.02535141e-01 3.60298692e-03 4.11361128e-01
-7.62563050e-01 -2.81817466e-01 1.32218695e+00 1.46350771e-01
5.58552086e-01 -3.93752784e-01 -4.08510119e-01 -1.00125873e+00
-1.80783853e-01 -6.49165213e-01 3.97660911e-01 -6.65519238e-01
-1.69917083e+00 4.89760309e-01 5.04768550e-01 -1.97121859e+00
-4.08196956e-01 -6.88552916e-01 -2.81867445e-01 1.03811455e+00
-1.71459711e+00 -1.12527454e+00 -3.02276045e-01 6.23530030e-01
6.82231247e-01 -6.08291328e-01 8.50464761e-01 1.27627552e-01
-1.40657470e-01 4.06480074e-01 7.80563474e-01 -3.12116921e-01
1.24841559e+00 -1.45466638e+00 5.89016199e-01 6.58497870e-01
-1.08040243e-01 2.00899854e-01 8.01909029e-01 -4.68771696e-01
-9.84187543e-01 -1.40785456e+00 6.83324158e-01 -8.50222409e-01
6.00466788e-01 -4.11054701e-01 -1.12615156e+00 4.30695146e-01
1.81823969e-01 1.40456983e-03 5.39563954e-01 1.16176896e-01
-5.24407625e-01 -2.35873371e-01 -1.24160552e+00 1.90426961e-01
9.90368247e-01 -4.10540104e-01 -8.79825890e-01 1.48723572e-01
5.29030442e-01 -2.76486367e-01 -9.84211087e-01 6.65257871e-01
3.43240619e-01 -9.98061240e-01 1.04529202e+00 -4.35319662e-01
3.50969344e-01 -5.96514344e-01 -4.21843737e-01 -1.73256135e+00
-4.78326112e-01 -1.29628882e-01 9.20591280e-02 1.31148911e+00
2.37315953e-01 -8.85484874e-01 6.74091458e-01 3.03987153e-02
-2.51387626e-01 -1.65510550e-01 -1.22899461e+00 -1.26193213e+00
6.72923803e-01 -4.27938521e-01 6.84303641e-01 1.10859275e+00
-4.70524579e-01 4.63553399e-01 -1.93683967e-01 4.37767863e-01
7.07842708e-01 1.63615480e-01 1.15755320e+00 -1.67069316e+00
1.05085745e-01 -4.04507965e-01 -6.21196985e-01 -1.09741843e+00
1.98451504e-01 -1.03586709e+00 7.23369196e-02 -1.36313117e+00
-1.57098457e-01 -5.04885733e-01 -4.45933700e-01 2.86461800e-01
1.08014509e-01 4.77504246e-02 1.31086484e-01 5.64120591e-01
-4.67660725e-01 7.19000280e-01 9.27902520e-01 -2.79798836e-01
-2.52003312e-01 -9.78318881e-03 -6.23033702e-01 5.59475362e-01
6.69838727e-01 -6.29753470e-01 -5.29463351e-01 -3.41993660e-01
-4.47117865e-01 -4.52873558e-01 3.73647064e-01 -1.33992553e+00
-6.40600771e-02 -1.99497789e-01 2.83715636e-01 -7.45479345e-01
2.73349732e-01 -1.07955945e+00 1.87273115e-01 8.16137046e-02
-2.50332728e-02 -4.45034981e-01 5.49548090e-01 8.76336813e-01
-3.29559058e-01 1.31810829e-01 1.15293109e+00 1.91759348e-01
-1.39307189e+00 1.19358122e-01 -3.22135776e-01 4.93171252e-03
1.35149503e+00 -5.75620413e-01 -1.74620494e-01 -1.93285868e-01
-7.67652869e-01 2.50921369e-01 7.05299437e-01 7.24806786e-01
2.94222146e-01 -1.51294112e+00 -9.40008759e-01 4.76439029e-01
8.35200310e-01 2.85084099e-01 2.33864143e-01 6.10922277e-01
1.42071601e-02 3.98883045e-01 -3.78315926e-01 -1.27453613e+00
-9.76484895e-01 3.16954941e-01 4.05359030e-01 -9.00754854e-02
-5.86253345e-01 6.37133718e-01 5.69096446e-01 -9.64525998e-01
-1.64678097e-01 -2.09662452e-01 -2.06959397e-01 2.80597568e-01
6.38420088e-03 3.28787625e-01 2.71332234e-01 -8.55133951e-01
-6.69793367e-01 7.58803844e-01 -2.37978483e-03 1.18186384e-01
1.31472027e+00 -2.82903135e-01 3.91701728e-01 6.71018064e-01
1.11945283e+00 -2.99722582e-01 -1.64953554e+00 -5.69067895e-01
1.92270249e-01 -3.30735624e-01 -1.63494885e-01 -8.86502326e-01
-4.60590482e-01 8.84509087e-01 9.53760803e-01 2.10567996e-01
1.19129229e+00 2.52878487e-01 3.16249311e-01 1.97276831e-01
3.46487194e-01 -9.53481853e-01 -2.15851113e-01 7.92636395e-01
8.66046011e-01 -1.49519300e+00 -6.09631352e-02 -7.99086597e-03
-1.14873850e+00 7.64799118e-01 4.62164640e-01 -3.48615557e-01
6.78406060e-01 -1.15625881e-01 2.54211605e-01 -4.82133999e-02
-6.16410077e-01 -4.78649318e-01 2.09128529e-01 1.20537043e+00
-1.66655377e-01 1.45335868e-01 2.46548533e-01 3.93715918e-01
-9.60013866e-02 -2.15797946e-01 1.75186440e-01 8.18287909e-01
-4.64247465e-01 -1.21708024e+00 -4.98425126e-01 3.36121768e-01
7.95233920e-02 4.35984403e-01 -5.81419647e-01 1.06789565e+00
2.40232438e-01 6.76474988e-01 2.58281767e-01 -5.09767115e-01
6.80912614e-01 4.92234081e-01 1.44907877e-01 -6.44171178e-01
-1.58342168e-01 5.52204102e-02 -3.01695410e-02 -1.67912349e-01
-8.07266951e-01 -1.07327282e+00 -9.23079312e-01 -8.18886012e-02
-1.03644319e-02 -4.09116037e-02 7.65625536e-01 9.63160813e-01
6.71767414e-01 3.96127969e-01 6.77574098e-01 -1.02012110e+00
-4.80548888e-01 -1.10181177e+00 -7.95855939e-01 5.74713111e-01
5.46573400e-01 -1.37648094e+00 -4.11196083e-01 1.11876339e-01] | [8.141557693481445, -2.509727716445923] |
089c9683-7d1b-4e31-892d-0fb2f649bda0 | graph-ladling-shockingly-simple-parallel-gnn | 2306.10466 | null | https://arxiv.org/abs/2306.10466v1 | https://arxiv.org/pdf/2306.10466v1.pdf | Graph Ladling: Shockingly Simple Parallel GNN Training without Intermediate Communication | Graphs are omnipresent and GNNs are a powerful family of neural networks for learning over graphs. Despite their popularity, scaling GNNs either by deepening or widening suffers from prevalent issues of unhealthy gradients, over-smoothening, information squashing, which often lead to sub-standard performance. In this work, we are interested in exploring a principled way to scale GNNs capacity without deepening or widening, which can improve its performance across multiple small and large graphs. Motivated by the recent intriguing phenomenon of model soups, which suggest that fine-tuned weights of multiple large-language pre-trained models can be merged to a better minima, we argue to exploit the fundamentals of model soups to mitigate the aforementioned issues of memory bottleneck and trainability during GNNs scaling. More specifically, we propose not to deepen or widen current GNNs, but instead present a data-centric perspective of model soups tailored for GNNs, i.e., to build powerful GNNs by dividing giant graph data to build independently and parallelly trained multiple comparatively weaker GNNs without any intermediate communication, and combining their strength using a greedy interpolation soup procedure to achieve state-of-the-art performance. Moreover, we provide a wide variety of model soup preparation techniques by leveraging state-of-the-art graph sampling and graph partitioning approaches that can handle large graph data structures. Our extensive experiments across many real-world small and large graphs, illustrate the effectiveness of our approach and point towards a promising orthogonal direction for GNN scaling. Codes are available at: \url{https://github.com/VITA-Group/graph_ladling}. | ['Zhangyang Wang', 'Ying Ding', 'Tianlong Chen', 'Shiwei Liu', 'Ajay Jaiswal'] | 2023-06-18 | null | null | null | null | ['graph-sampling', 'graph-partitioning'] | ['graphs', 'graphs'] | [ 6.59772754e-02 3.64183456e-01 -3.89618784e-01 1.11656368e-01
-4.05220121e-01 -4.60728884e-01 4.62135017e-01 2.85629392e-01
-1.59919426e-01 6.10523343e-01 5.67638725e-02 -4.93126124e-01
-1.49909616e-01 -1.22559226e+00 -7.43862391e-01 -5.30349493e-01
-3.10937256e-01 4.84659225e-01 2.60735333e-01 -4.69963789e-01
5.54031543e-02 5.47288179e-01 -1.00883865e+00 -2.82360256e-01
1.09575546e+00 4.21990752e-01 1.23686932e-01 8.52158129e-01
-9.50310603e-02 7.16931641e-01 -3.85441929e-01 -7.66321123e-01
4.13231611e-01 -3.19067299e-01 -8.26886594e-01 1.91260546e-01
5.29971898e-01 -8.87752548e-02 -7.05767512e-01 1.24165499e+00
5.15948057e-01 2.39980686e-02 1.80599168e-01 -1.17939878e+00
-8.37251842e-01 1.14782858e+00 -8.81481469e-01 1.52807832e-01
-1.27161846e-01 2.71810412e-01 1.05387998e+00 -3.88896734e-01
5.69873154e-01 1.17432225e+00 8.87779653e-01 6.80854142e-01
-1.44295371e+00 -7.50482321e-01 2.32736871e-01 -2.42570370e-01
-1.44950867e+00 -3.52191567e-01 8.28429639e-01 -1.49567828e-01
8.86208534e-01 2.25711733e-01 9.74543452e-01 1.01163602e+00
7.40958378e-04 5.98494351e-01 7.09074676e-01 -4.68354642e-01
8.64909515e-02 -1.60160288e-01 1.78112969e-01 1.11834896e+00
7.37695634e-01 -2.02760085e-01 -3.19282621e-01 -3.34577411e-01
1.12876356e+00 -4.35981266e-02 -2.44089186e-01 -5.51070929e-01
-1.04134429e+00 9.08178389e-01 8.78422976e-01 3.37843567e-01
-2.64684141e-01 4.82183874e-01 5.50842941e-01 3.58761102e-01
5.20284295e-01 6.30960345e-01 -4.44762319e-01 1.85133457e-01
-9.13381994e-01 4.38270010e-02 9.16729450e-01 1.06172025e+00
9.57901001e-01 1.57452777e-01 2.17915609e-01 8.05757165e-01
1.37616679e-01 1.20912068e-01 2.56011635e-01 -5.34892201e-01
6.90337777e-01 8.68362248e-01 -6.98184192e-01 -9.91638780e-01
-5.43302178e-01 -8.30480099e-01 -1.43806851e+00 -3.44350189e-01
3.93760115e-01 -1.61425874e-01 -1.15046322e+00 1.91678309e+00
3.52458328e-01 1.60742506e-01 -2.10793138e-01 4.05260801e-01
7.26472199e-01 4.13970590e-01 1.08843923e-01 1.02994263e-01
9.91826117e-01 -1.29796100e+00 -1.09017976e-01 -4.59457308e-01
9.44929600e-01 -2.78941631e-01 1.32511890e+00 4.52479869e-01
-1.24860132e+00 -5.00283599e-01 -1.05615830e+00 -7.98313543e-02
-4.25447702e-01 -3.56963038e-01 1.04685962e+00 7.72335708e-01
-1.63052404e+00 1.04714346e+00 -8.31948459e-01 -6.45700216e-01
6.19498610e-01 4.11528409e-01 -3.59610468e-01 -2.35424563e-01
-1.11377978e+00 6.14769161e-01 7.65991807e-01 -1.08588025e-01
-8.90032887e-01 -6.18024528e-01 -8.13951671e-01 1.69401854e-01
5.41417241e-01 -9.73892748e-01 6.35559678e-01 -6.85917854e-01
-1.21231306e+00 6.92106783e-01 3.02981764e-01 -5.89235485e-01
2.47483641e-01 1.22791991e-01 -3.66254598e-01 -3.36225778e-02
-2.11611480e-01 5.02027035e-01 7.06248760e-01 -1.08747566e+00
-3.81079651e-02 -4.32260901e-01 1.97940081e-01 1.56641677e-01
-9.31599736e-01 -2.50748962e-01 -7.53618419e-01 -5.99822938e-01
1.74266025e-01 -9.27917659e-01 -6.47269607e-01 -4.07920480e-01
-7.93736696e-01 -7.89775550e-02 2.88923115e-01 -5.32080472e-01
1.73876429e+00 -1.82421350e+00 2.51677006e-01 4.98262346e-01
1.07570302e+00 4.03992444e-01 -5.11345148e-01 7.67517030e-01
-3.89161780e-02 4.91266638e-01 -1.26148969e-01 -1.89876080e-01
-3.48272473e-02 1.61707848e-01 8.07441920e-02 3.88170451e-01
-4.24328912e-03 1.19392383e+00 -9.75008547e-01 -3.79862815e-01
1.90529022e-02 3.76128703e-01 -7.13445246e-01 -1.69238344e-01
-2.42710754e-01 2.32681543e-01 -3.87259901e-01 6.15016222e-01
8.42253923e-01 -8.13840747e-01 4.88537103e-01 3.52848531e-03
3.67243052e-01 2.89265513e-01 -1.09275699e+00 1.72797608e+00
-2.67046034e-01 1.24010593e-01 1.21193096e-01 -1.11337948e+00
9.12731230e-01 -2.67645031e-01 1.74283192e-01 -5.42978585e-01
-9.87696927e-04 2.18638629e-01 1.49721384e-01 3.33048329e-02
5.00771523e-01 4.66853799e-03 -3.69201079e-02 3.28556538e-01
3.44994307e-01 -2.32405171e-01 6.81951940e-01 6.69741333e-01
1.45497048e+00 -2.62988389e-01 3.15754354e-01 -5.12384415e-01
3.62401664e-01 -1.80584341e-01 1.90648615e-01 8.28307569e-01
-1.13713503e-01 3.57468188e-01 8.06570828e-01 -4.91960764e-01
-1.29026449e+00 -9.34738338e-01 2.93452561e-01 1.39114130e+00
-8.96811187e-02 -7.76810765e-01 -8.48877013e-01 -7.99523890e-01
-1.35501698e-01 3.35080296e-01 -6.04177773e-01 -2.80771643e-01
-6.35925531e-01 -1.08838844e+00 7.74329960e-01 3.90258759e-01
4.56567615e-01 -8.47541034e-01 2.63606936e-01 2.50544935e-01
2.78550386e-01 -8.43987823e-01 -4.47023094e-01 2.68356323e-01
-1.10226119e+00 -8.39771688e-01 -8.09245169e-01 -8.43240321e-01
8.68551493e-01 3.58981967e-01 1.56137192e+00 7.22395241e-01
1.76397618e-02 1.17794042e-02 -2.17085913e-01 -7.20490888e-02
-7.19959855e-01 8.30398381e-01 1.48946308e-02 -4.21836108e-01
3.27734351e-02 -1.08725655e+00 -5.58722794e-01 4.24000435e-02
-9.69554305e-01 3.12346041e-01 6.87353730e-01 7.65782177e-01
2.81343132e-01 1.64216906e-01 5.70335925e-01 -1.16195822e+00
9.76728439e-01 -4.70113605e-01 -5.71943879e-01 2.59629577e-01
-8.98149490e-01 1.93420425e-01 9.10680950e-01 -4.89455312e-01
-3.99956524e-01 -4.57072288e-01 -2.01787487e-01 -1.95340350e-01
1.85464308e-01 5.81650555e-01 -1.70314871e-02 -4.90450233e-01
9.56444919e-01 2.12887421e-01 -2.87606269e-02 -2.42746115e-01
7.70198762e-01 9.55845043e-02 1.43763483e-01 -5.42446494e-01
1.10386443e+00 3.29590499e-01 5.65025061e-02 -6.44648850e-01
-5.85465491e-01 -3.34378511e-01 -4.43464607e-01 -3.21949199e-02
4.77891594e-01 -8.53032410e-01 -4.84957159e-01 4.93450254e-01
-8.45718205e-01 -8.10636878e-01 -2.24342570e-01 9.69238281e-02
-1.76783442e-01 6.19354844e-01 -1.09355748e+00 -2.89026290e-01
-7.00843453e-01 -8.70289624e-01 6.81991160e-01 3.43071163e-01
7.71386735e-03 -1.30554760e+00 1.15833655e-01 2.18345493e-01
7.00689316e-01 1.86079174e-01 1.04328430e+00 -8.29632461e-01
-5.71434498e-01 -7.54143596e-02 -6.10968053e-01 3.86852473e-01
-1.29576519e-01 -1.49026483e-01 -5.71930349e-01 -8.04452837e-01
-4.46072549e-01 -5.25640309e-01 1.01870918e+00 2.87530392e-01
1.13328505e+00 -3.94690335e-01 -4.52075869e-01 8.71200740e-01
1.70972836e+00 -3.78104120e-01 6.18469834e-01 1.45566791e-01
1.27296996e+00 1.80032447e-01 -1.83513582e-01 3.10511440e-01
5.33568382e-01 8.91492516e-03 5.67356229e-01 -4.57350940e-01
-3.10906708e-01 -5.24292886e-01 1.83369610e-02 1.37824237e+00
-1.99817479e-01 -4.52591896e-01 -1.04268169e+00 5.32906950e-01
-1.60530770e+00 -5.00806808e-01 -1.95294961e-01 2.11007047e+00
7.63795078e-01 3.94489735e-01 2.49890298e-01 -2.35442683e-01
8.35472465e-01 5.05527914e-01 -6.44946814e-01 -2.10331976e-01
8.93454999e-03 2.82663137e-01 9.19473946e-01 3.59910905e-01
-7.44938970e-01 1.28214145e+00 5.93739748e+00 1.21708822e+00
-9.77206767e-01 6.94653913e-02 8.06422591e-01 7.49631226e-02
-4.82852548e-01 1.83907941e-01 -6.88434124e-01 1.70836627e-01
1.03790581e+00 -3.05864036e-01 1.00075686e+00 8.43459368e-01
-2.51469851e-01 2.72318393e-01 -7.85178900e-01 7.25355744e-01
-1.61842123e-01 -1.60933745e+00 2.57878393e-01 3.49598467e-01
9.32271063e-01 5.02682388e-01 -2.58013904e-01 6.38788700e-01
8.55176032e-01 -9.81605947e-01 2.70453125e-01 1.03724055e-01
6.43509269e-01 -7.13598311e-01 3.09202701e-01 4.09982473e-01
-1.48920226e+00 4.05844264e-02 -6.05297148e-01 -4.79078628e-02
-7.04198629e-02 7.17029154e-01 -8.11254203e-01 7.30319500e-01
4.80717719e-01 5.56345880e-01 -9.55995381e-01 8.88545215e-01
-5.69675826e-02 6.51583374e-01 -3.79675776e-01 -1.77249104e-01
2.84689248e-01 -3.94105852e-01 3.23908180e-01 1.10384691e+00
2.23037258e-01 -2.86342114e-01 1.97901383e-01 9.46845233e-01
-5.49975097e-01 2.10540041e-01 -7.64855146e-01 -4.42775071e-01
4.95940626e-01 1.44883585e+00 -1.08493936e+00 -5.53908169e-01
-3.89715612e-01 7.38643706e-01 9.09221768e-01 4.09505397e-01
-9.53071356e-01 -4.06114697e-01 2.52898335e-01 3.30202520e-01
2.33909175e-01 -3.26714426e-01 -2.12523445e-01 -1.28061819e+00
-2.03387111e-01 -9.77195680e-01 5.07820189e-01 -3.49370331e-01
-1.37689292e+00 7.65120387e-01 -2.10372165e-01 -7.42974102e-01
2.81194896e-01 -3.44448715e-01 -7.38723397e-01 6.62673414e-01
-1.32214046e+00 -1.39353216e+00 -3.82549554e-01 6.42654002e-01
2.66456813e-01 8.37562904e-02 5.40517986e-01 3.92962754e-01
-5.42855442e-01 7.26096511e-01 7.17810914e-03 2.19543830e-01
4.31713760e-01 -1.21824694e+00 1.15089941e+00 1.07886696e+00
1.18857943e-01 7.94546902e-01 5.47602057e-01 -8.05016696e-01
-1.55411959e+00 -1.13464344e+00 5.32079399e-01 -6.07667193e-02
1.13326573e+00 -7.29106069e-01 -1.03035307e+00 8.08372676e-01
1.52013218e-03 1.38230100e-01 2.92046368e-01 4.50338572e-01
-3.89565736e-01 -9.77148190e-02 -9.62697148e-01 9.39144075e-01
1.59136331e+00 -3.29448730e-01 7.80302212e-02 5.56066930e-01
1.04155600e+00 -3.38074684e-01 -9.20779884e-01 3.50638956e-01
1.60234600e-01 -9.74806964e-01 9.86636341e-01 -6.03325307e-01
2.57318914e-01 4.08486575e-02 1.21166423e-01 -1.47524416e+00
-6.68182969e-01 -1.03370380e+00 -2.25976050e-01 1.23067677e+00
6.02395177e-01 -9.94275510e-01 1.21788347e+00 3.46163958e-01
-7.73540884e-02 -8.52975428e-01 -5.45258701e-01 -7.90495932e-01
3.03809613e-01 -1.79713205e-01 8.14151287e-01 1.10329127e+00
2.00445252e-03 5.86513102e-01 -3.75856519e-01 5.62091023e-02
8.00059080e-01 -2.39104122e-01 1.16690743e+00 -1.24483967e+00
-4.46763963e-01 -7.41179645e-01 -3.13556492e-01 -1.07254839e+00
-1.04366496e-01 -1.24043882e+00 -4.29516196e-01 -1.73273194e+00
4.70992178e-01 -5.43652534e-01 -2.33662039e-01 4.37307745e-01
-2.26720482e-01 3.03344995e-01 2.44542569e-01 -1.55101372e-02
-7.71578193e-01 3.46429616e-01 1.57035744e+00 -5.02992608e-02
-2.71351814e-01 -1.88269556e-01 -1.19083786e+00 5.56023061e-01
9.99694109e-01 -3.68485123e-01 -6.04099154e-01 -3.79291058e-01
6.19255781e-01 -5.23834005e-02 2.11216196e-01 -1.06890988e+00
2.56583512e-01 -1.98873729e-02 2.85585858e-02 -2.34884769e-01
-9.18312073e-02 -2.65565515e-01 2.09116220e-01 6.04272485e-01
-1.53348356e-01 1.17264740e-01 1.88880727e-01 5.93775272e-01
1.51908681e-01 -5.69848903e-02 8.12688887e-01 -4.31878358e-01
-4.93169934e-01 7.44295657e-01 1.56211466e-01 3.13805491e-01
7.19479620e-01 -5.98345883e-02 -6.18986130e-01 -3.90872747e-01
-5.75361729e-01 2.79579282e-01 6.64780736e-01 1.29170433e-01
3.62046808e-01 -1.11985230e+00 -7.56628513e-01 2.45605782e-01
2.11297516e-02 2.01075286e-01 3.84054512e-01 8.97213280e-01
-6.97326779e-01 2.08621789e-02 -5.87034561e-02 -3.14839989e-01
-9.93380129e-01 8.59155178e-01 2.71109521e-01 -1.03414929e+00
-6.13914073e-01 1.11450684e+00 3.06556731e-01 -8.03406298e-01
8.14974904e-02 -3.52784336e-01 2.23243803e-01 -2.82059282e-01
1.78496838e-01 4.19516295e-01 6.73027784e-02 -1.04553089e-01
-1.32516354e-01 2.60527343e-01 -3.14925969e-01 4.68655288e-01
1.41576707e+00 -1.93465918e-01 -3.99520606e-01 -1.49936020e-01
1.08799410e+00 1.04938664e-01 -1.04638946e+00 -3.29518616e-01
-7.14415088e-02 -5.89026101e-02 -1.24432959e-01 -4.99904752e-01
-1.21493196e+00 6.08871222e-01 8.23729187e-02 8.28436852e-01
1.15356100e+00 2.70194322e-01 9.34473157e-01 2.36069188e-01
3.34720969e-01 -8.71976495e-01 1.54030457e-01 5.67250907e-01
6.51116610e-01 -1.11409211e+00 2.17441246e-01 -3.58579963e-01
-3.51605743e-01 9.71522510e-01 6.69005513e-01 -4.30895567e-01
5.34774601e-01 2.28701025e-01 -4.42043096e-01 -4.98453945e-01
-7.00437725e-01 -2.20695123e-01 7.64626116e-02 6.08329713e-01
2.52975345e-01 2.40063071e-01 -1.78657860e-01 2.98421532e-01
-3.00630778e-01 -1.60011306e-01 5.13104975e-01 5.48664153e-01
-3.61364663e-01 -1.09108937e+00 -4.67281649e-03 6.83732152e-01
-2.79361606e-01 -6.23641908e-01 -3.51864815e-01 1.03716373e+00
-2.03782573e-01 5.27053297e-01 -3.96941036e-01 -5.46640515e-01
1.25052094e-01 -2.59439975e-01 5.24406672e-01 -4.87850487e-01
-5.83543420e-01 -3.90061103e-02 2.59136677e-01 -5.12652040e-01
7.97208995e-02 -3.15880775e-02 -1.02831912e+00 -9.37694669e-01
-5.67663372e-01 -8.54121372e-02 3.83464158e-01 5.51098764e-01
4.86915290e-01 6.30895257e-01 3.42400044e-01 -1.04198802e+00
-5.70504308e-01 -1.09423316e+00 -7.36611962e-01 2.76815951e-01
8.36701766e-02 -1.87461436e-01 -4.79963899e-01 -4.46141392e-01] | [6.9795122146606445, 6.137726306915283] |
08ae7e79-8626-4df3-bc1c-94c19048d5f9 | a-survey-on-efficient-processing-of | 2204.07922 | null | https://arxiv.org/abs/2204.07922v1 | https://arxiv.org/pdf/2204.07922v1.pdf | A Survey on Efficient Processing of Similarity Queries over Neural Embeddings | Similarity query is the family of queries based on some similarity metrics. Unlike the traditional database queries which are mostly based on value equality, similarity queries aim to find targets "similar enough to" the given data objects, depending on some similarity metric, e.g., Euclidean distance, cosine similarity and so on. To measure the similarity between data objects, traditional methods normally work on low level or syntax features(e.g., basic visual features on images or bag-of-word features of text), which makes them weak to compute the semantic similarities between objects. So for measuring data similarities semantically, neural embedding is applied. Embedding techniques work by representing the raw data objects as vectors (so called "embeddings" or "neural embeddings" since they are mostly generated by neural network models) that expose the hidden semantics of the raw data, based on which embeddings do show outstanding effectiveness on capturing data similarities, making it one of the most widely used and studied techniques in the state-of-the-art similarity query processing research. But there are still many open challenges on the efficiency of embedding based similarity query processing, which are not so well-studied as the effectiveness. In this survey, we first provide an overview of the "similarity query" and "similarity query processing" problems. Then we talk about recent approaches on designing the indexes and operators for highly efficient similarity query processing on top of embeddings (or more generally, high dimensional data). Finally, we investigate the specific solutions with and without using embeddings in selected application domains of similarity queries, including entity resolution and information retrieval. By comparing the solutions, we show how neural embeddings benefit those applications. | ['Yifan Wang'] | 2022-04-17 | null | null | null | null | ['entity-resolution'] | ['natural-language-processing'] | [-9.36017632e-02 -4.17496681e-01 -2.66258001e-01 -4.76235896e-01
-3.92999589e-01 -4.61149096e-01 5.43364942e-01 8.82012129e-01
-6.82146788e-01 2.97263619e-02 3.20592642e-01 8.58077332e-02
-5.26463449e-01 -1.31313181e+00 -3.73039305e-01 -4.66332734e-01
1.23487751e-03 4.62513298e-01 4.18624729e-01 -6.23977482e-01
5.45773089e-01 6.29345179e-01 -2.09091783e+00 1.37677267e-01
5.01585662e-01 1.18451464e+00 2.44648486e-01 3.98070276e-01
-8.31488192e-01 2.59545833e-01 -5.76442480e-01 -4.47629541e-01
8.96450132e-02 -1.47142177e-02 -9.05239999e-01 -6.62707448e-01
2.32401609e-01 -3.02984547e-02 -7.04863250e-01 1.32223094e+00
6.74482882e-01 4.08991247e-01 7.01655984e-01 -1.45303476e+00
-1.32522357e+00 2.04074532e-01 1.00787036e-01 3.63508254e-01
8.23424160e-01 -4.82025892e-01 1.35263932e+00 -1.18935537e+00
5.57820678e-01 1.26316166e+00 4.86451387e-01 2.66515166e-01
-9.57677066e-01 -1.65547416e-01 -2.50037581e-01 8.30736816e-01
-1.71305990e+00 -5.21843471e-02 7.08425939e-01 -2.39726186e-01
1.00819802e+00 6.05791092e-01 4.58233476e-01 5.89747250e-01
1.13297865e-01 6.85223579e-01 4.29028243e-01 -4.24238741e-01
2.32308269e-01 4.28512484e-01 4.44361538e-01 4.35900599e-01
2.54530996e-01 -3.37939262e-02 -2.57426262e-01 -4.41197723e-01
3.42563629e-01 5.75764656e-01 -1.69028953e-01 -5.06299675e-01
-1.27161503e+00 1.13032508e+00 7.30867565e-01 8.59533489e-01
-2.88201302e-01 -7.25844596e-03 7.84555018e-01 6.15351498e-01
1.83539480e-01 6.32554948e-01 -1.75097927e-01 1.26810655e-01
-5.39707601e-01 4.44529057e-01 6.92808747e-01 1.06274879e+00
1.09923780e+00 -3.57995331e-01 -3.45830619e-01 9.67693031e-01
2.95890361e-01 4.33695912e-01 8.37361157e-01 -6.02426171e-01
1.62659630e-01 9.63686109e-01 -7.08112568e-02 -1.52647221e+00
-2.26701736e-01 7.46349171e-02 -6.95775509e-01 -1.11142576e-01
-1.34067193e-01 4.41396743e-01 -4.36491132e-01 1.59377134e+00
3.16622943e-01 -9.26136822e-02 2.73904353e-01 7.71827042e-01
1.15585625e+00 7.94353783e-01 -1.40639618e-01 4.15200591e-02
1.60664344e+00 -7.26563036e-01 -7.67467082e-01 2.89944887e-01
9.32710767e-01 -9.78561759e-01 1.22868073e+00 -1.86701283e-01
-1.03826964e+00 -6.68267190e-01 -9.23382223e-01 -3.84108245e-01
-1.43567121e+00 -4.32357103e-01 5.52030265e-01 4.53463078e-01
-1.14442885e+00 6.82898819e-01 -2.76299477e-01 -7.04417765e-01
2.59811943e-03 4.45481181e-01 -3.94069821e-01 -2.03918502e-01
-1.59166968e+00 9.78820622e-01 6.42097294e-01 -3.88007939e-01
-2.35974431e-01 -7.08268225e-01 -9.96541321e-01 3.13479125e-01
3.09639245e-01 -6.92525566e-01 8.96229208e-01 -4.39808160e-01
-8.30901980e-01 9.00464535e-01 -2.33111948e-01 -2.34072402e-01
-7.57165924e-02 -1.54604122e-01 -6.90497756e-01 1.54194623e-01
1.55111387e-01 6.35459185e-01 5.22360265e-01 -8.77513289e-01
-6.66950703e-01 -4.48039412e-01 2.00530231e-01 2.67425567e-01
-9.65956926e-01 4.83432531e-01 -4.83249068e-01 -7.24897385e-01
1.41598716e-01 -5.73938370e-01 -1.41369298e-01 3.50623786e-01
-2.28608120e-02 -8.15524995e-01 1.02226698e+00 -2.97163904e-01
1.60144114e+00 -2.40167761e+00 -5.11519425e-02 2.86036432e-01
3.07728231e-01 3.43029201e-01 -3.58516186e-01 9.75678623e-01
-1.91521272e-01 1.54934898e-01 -1.54343411e-01 7.59679750e-02
3.64835978e-01 3.89454722e-01 -3.81917387e-01 2.09146932e-01
6.79497197e-02 1.04711545e+00 -9.90119815e-01 -7.40453064e-01
1.79938033e-01 4.49342519e-01 -3.57799977e-01 5.79288065e-01
-1.59015413e-02 -3.52677435e-01 -5.97983420e-01 5.53599954e-01
4.52272475e-01 -2.66037341e-02 -3.14324707e-01 -5.54214895e-01
-3.84688750e-02 6.81197420e-02 -1.26203179e+00 1.73966241e+00
-4.71425891e-01 5.09834945e-01 -3.84399325e-01 -1.38186240e+00
1.26597393e+00 3.24731082e-01 7.76563823e-01 -9.11097705e-01
6.00791313e-02 2.27745980e-01 -2.35205039e-01 -6.62286043e-01
9.40596521e-01 1.23515718e-01 -3.03703040e-01 5.39494157e-01
-1.37058184e-01 -1.43417880e-01 2.46370837e-01 1.56324863e-01
1.01300454e+00 -5.57906806e-01 3.70820045e-01 -1.03252023e-01
8.05344999e-01 6.67960718e-02 8.37484151e-02 5.49916625e-01
-9.69777405e-02 5.56303442e-01 2.06247419e-01 -5.19963443e-01
-1.11487424e+00 -1.18471348e+00 -1.53102830e-01 1.36010242e+00
6.27012610e-01 -6.92367435e-01 -4.99673128e-01 -3.74778092e-01
2.47922003e-01 3.05054426e-01 -5.93848705e-01 -6.21357560e-01
-3.41079354e-01 -4.26746517e-01 4.68080997e-01 8.09877694e-01
2.36538947e-01 -1.18323600e+00 -3.90786052e-01 2.24461481e-01
1.36780635e-01 -7.01864600e-01 -5.32479763e-01 -7.92250410e-02
-7.69035101e-01 -9.12051201e-01 -8.33091438e-01 -1.06410217e+00
4.30123568e-01 6.60823762e-01 1.08232260e+00 2.10085854e-01
-5.26495576e-01 6.21817887e-01 -6.53875291e-01 -3.73251885e-01
5.94579615e-02 -1.00482479e-01 3.91135156e-01 -8.77115037e-03
9.21499610e-01 -4.18895483e-01 -4.69712138e-01 3.75491679e-01
-1.45700300e+00 -7.68673062e-01 5.79789758e-01 8.71908605e-01
7.83711433e-01 5.02025001e-02 3.62686783e-01 -4.76997554e-01
1.23834324e+00 -6.05605006e-01 -3.14341575e-01 7.62594640e-01
-7.68833876e-01 3.45761538e-01 6.02337122e-01 -6.02256298e-01
-4.12299186e-01 -5.49035490e-01 -5.53243123e-02 -6.85602188e-01
-3.93156707e-02 5.05028486e-01 -1.03099994e-01 -1.70363501e-01
5.74782610e-01 5.08601248e-01 -5.67727163e-02 -5.56492031e-01
6.54151142e-01 7.53876269e-01 4.33600366e-01 -5.48709452e-01
7.12454140e-01 4.44189280e-01 -1.01416074e-01 -7.53213406e-01
-3.68266791e-01 -9.78713453e-01 -4.75225896e-01 4.08956766e-01
1.07520366e+00 -2.77245164e-01 -7.42235184e-01 -1.27952471e-01
-1.31607318e+00 6.25199735e-01 -6.75805688e-01 7.23830640e-01
-5.88287711e-01 5.13401449e-01 -4.67130005e-01 -4.83228713e-01
-3.91490996e-01 -9.30151165e-01 9.84584689e-01 3.18112165e-01
-2.14855701e-01 -1.20084095e+00 4.18584079e-01 -1.56556278e-01
7.86733449e-01 -2.44891241e-01 1.42574072e+00 -1.17356467e+00
-3.01202536e-01 -4.23632979e-01 -5.67137837e-01 3.23293895e-01
2.22681984e-01 -2.06875816e-01 -7.57764161e-01 -1.46431699e-01
-4.49854434e-02 -6.45260811e-02 5.79634309e-01 3.42577510e-02
1.27763224e+00 -2.35766903e-01 -4.22512650e-01 5.62921643e-01
1.60401440e+00 2.84867287e-01 6.55184448e-01 2.33923033e-01
4.51399475e-01 8.82485628e-01 7.89650440e-01 5.76379538e-01
2.74583608e-01 7.51268268e-01 4.27884072e-01 -7.07772970e-02
2.82864124e-01 -1.88339129e-01 1.03454784e-01 1.01685941e+00
2.55651236e-01 -8.67696404e-02 -6.63058579e-01 6.77433491e-01
-1.80986059e+00 -9.78931665e-01 1.05316184e-01 2.32054329e+00
4.53551084e-01 -3.00257385e-01 6.65373122e-03 1.51717365e-01
8.94596338e-01 2.74256170e-01 -3.01627189e-01 -7.90504277e-01
-1.55332342e-01 4.26627755e-01 2.01577246e-01 1.78153560e-01
-9.14698780e-01 8.53064358e-01 6.01500463e+00 9.50028181e-01
-9.80797648e-01 6.43829107e-02 -2.03229830e-01 2.00316414e-01
-6.22692227e-01 1.63848437e-02 -6.43133879e-01 3.62081110e-01
5.98760784e-01 -6.74165487e-01 1.59900993e-01 9.83793259e-01
-2.13634491e-01 2.51013756e-01 -1.33876586e+00 1.45379055e+00
2.46010303e-01 -1.28805351e+00 3.83074850e-01 -1.17448390e-01
2.84918070e-01 -2.40673646e-01 1.81692123e-01 4.04432774e-01
-1.07806772e-01 -1.02392697e+00 1.91817433e-01 7.54268527e-01
7.02106059e-01 -6.35752797e-01 9.79028881e-01 -3.54674868e-02
-1.47455478e+00 -2.52927095e-03 -1.04419172e+00 2.62763888e-01
9.79735106e-02 3.93732190e-01 -4.19902205e-01 7.14999437e-01
8.34583342e-01 7.06207991e-01 -5.78472733e-01 1.11036098e+00
2.79106200e-01 -2.86576122e-01 -3.17635238e-01 -3.70705247e-01
2.30148822e-01 -2.34858543e-01 2.97773927e-01 1.21011055e+00
5.10222197e-01 1.06377855e-01 9.25036073e-02 9.15593982e-01
-3.52038182e-02 5.03098190e-01 -1.07467353e+00 -2.51628637e-01
6.74669862e-01 1.05897975e+00 -2.22369730e-01 -5.90837061e-01
-6.67498648e-01 8.82456660e-01 -2.27707531e-02 1.87862173e-01
-5.66469073e-01 -1.05792511e+00 1.15319002e+00 -1.16404407e-01
1.29011869e-01 -2.09698603e-01 9.94841475e-03 -9.52501416e-01
3.54674041e-01 -4.82412547e-01 7.26763725e-01 -6.46725297e-01
-1.64947212e+00 5.74972868e-01 2.26989120e-01 -1.47547102e+00
-1.73386171e-01 -6.51411831e-01 -7.50060558e-01 8.86579752e-01
-1.46265829e+00 -6.53388083e-01 -3.95067602e-01 9.40075040e-01
3.36120218e-01 -2.82084972e-01 1.07172811e+00 4.86762136e-01
-1.66344762e-01 6.54119074e-01 3.47132355e-01 1.04771145e-01
7.49147832e-01 -1.02005720e+00 3.06361496e-01 1.12552814e-01
4.98704940e-01 8.14875960e-01 3.55129004e-01 -2.97483295e-01
-1.53901970e+00 -6.86586022e-01 1.35644233e+00 -2.78706312e-01
7.18745410e-01 -2.79795378e-01 -1.34938788e+00 2.03390241e-01
3.75126079e-02 2.79363245e-01 9.47841048e-01 -1.55298755e-01
-6.46970570e-01 -2.99988031e-01 -1.16539598e+00 5.98988891e-01
1.10266805e+00 -9.29560721e-01 -1.09865904e+00 4.16538298e-01
1.18474424e+00 2.51593053e-01 -1.25521505e+00 3.81163210e-01
5.42479813e-01 -1.01162660e+00 1.35774434e+00 -1.10780513e+00
-1.01347484e-01 -3.77945542e-01 -8.24317753e-01 -9.83280599e-01
-4.51340914e-01 -9.19071212e-02 3.27416137e-03 1.01391017e+00
8.68196860e-02 -7.24487185e-01 5.88935316e-01 4.15352792e-01
1.11947902e-01 -9.21992660e-01 -6.54052556e-01 -1.00148797e+00
-1.01633877e-01 -4.21638548e-01 1.04676867e+00 1.08115852e+00
-8.23078379e-02 4.90834452e-02 2.43505314e-01 5.68533689e-02
2.20871747e-01 2.78569430e-01 5.37740707e-01 -1.33895159e+00
2.72127204e-02 -7.21204817e-01 -1.12014008e+00 -1.01612473e+00
9.44198444e-02 -1.00904834e+00 -4.17720139e-01 -1.51703429e+00
-2.90277433e-02 -5.33311427e-01 -5.25203228e-01 1.30276710e-01
1.84780024e-02 1.16374418e-02 1.29494578e-01 3.90298218e-01
-5.84761798e-01 7.45734274e-01 8.07079256e-01 -3.98195624e-01
-9.39646140e-02 -1.71190262e-01 -5.40445626e-01 3.07985425e-01
4.73847806e-01 -5.16240239e-01 -4.74694669e-01 -6.69550359e-01
3.46511751e-01 -1.79266781e-01 2.82853246e-01 -8.93474340e-01
5.77799976e-01 -1.73297331e-01 -4.11562212e-02 -5.40040255e-01
4.81457531e-01 -1.02997208e+00 -1.81171879e-01 3.98187101e-01
-4.49458212e-01 7.41349578e-01 -3.30498785e-01 4.99659151e-01
-8.29786658e-01 -5.20362020e-01 4.41832304e-01 -2.91838441e-02
-1.03120232e+00 5.38894176e-01 5.32147177e-02 1.48405164e-01
1.06067038e+00 -3.88380945e-01 -4.30868901e-02 -3.79325598e-01
-5.90807080e-01 1.44224867e-01 4.30611223e-01 6.90183878e-01
1.04458654e+00 -1.68161559e+00 -5.67575037e-01 2.21663326e-01
7.16800988e-01 -1.93512052e-01 -7.37503767e-02 5.56584477e-01
-4.41588998e-01 6.78175569e-01 -2.48805478e-01 -4.53282237e-01
-1.24645424e+00 1.31787992e+00 -2.53312336e-03 -7.71024972e-02
-3.35117906e-01 6.67420924e-01 2.90952891e-01 -4.38478857e-01
3.39046299e-01 -2.97694594e-01 -4.98088330e-01 2.42501497e-01
6.38573587e-01 4.87893701e-01 8.54405463e-02 -5.82266867e-01
-4.95455831e-01 1.05222905e+00 -1.06592290e-01 4.69049253e-03
1.03867054e+00 7.78801963e-02 -3.51607144e-01 3.71102750e-01
1.90144575e+00 -3.41096699e-01 3.09624206e-02 -7.41248786e-01
3.79956782e-01 -6.25461936e-01 -4.03795183e-01 6.23457916e-02
-7.70904958e-01 1.22845316e+00 8.46226156e-01 5.53590000e-01
1.12024450e+00 3.08440596e-01 1.04871547e+00 9.04052198e-01
3.64365906e-01 -1.17372799e+00 2.32067391e-01 5.24580777e-01
8.79444718e-01 -1.09845889e+00 -2.05365449e-01 -2.05562726e-01
-3.85683924e-01 1.23677087e+00 3.52920771e-01 -2.81275988e-01
8.97470415e-01 -1.01187997e-01 -1.23816408e-01 -4.04069871e-01
-4.79840696e-01 -4.63226646e-01 4.84103203e-01 7.10931063e-01
2.75369346e-01 -5.83378337e-02 -6.50339544e-01 3.41513574e-01
-2.07343459e-01 -4.44682211e-01 -1.36844918e-01 1.03236699e+00
-6.30832791e-01 -1.37350190e+00 -3.73972297e-01 5.14537394e-01
-7.04081059e-02 -1.09479465e-01 -4.35505688e-01 6.17628872e-01
1.95395872e-01 7.18013048e-01 3.92624646e-01 -6.67571068e-01
6.34110212e-01 3.40111628e-02 9.82347429e-02 -6.85597360e-01
-5.90446293e-01 -7.34262764e-01 -3.87401521e-01 -6.60735488e-01
-3.00784975e-01 -1.86809793e-01 -1.18759966e+00 -3.23066622e-01
-3.37657690e-01 3.93418908e-01 8.15496624e-01 7.19563007e-01
4.87250090e-01 -8.28811675e-02 8.46401334e-01 -3.87994975e-01
-7.82924950e-01 -6.84105754e-01 -6.29210114e-01 8.26644719e-01
-1.70627609e-02 -6.22491598e-01 -2.02504307e-01 -3.20705593e-01] | [10.464266777038574, 8.558061599731445] |
01060343-0266-44f9-996c-d9ff492421e2 | learning-data-teaching-strategies-via | 2111.07083 | null | https://arxiv.org/abs/2111.07083v1 | https://arxiv.org/pdf/2111.07083v1.pdf | Learning Data Teaching Strategies Via Knowledge Tracing | Teaching plays a fundamental role in human learning. Typically, a human teaching strategy would involve assessing a student's knowledge progress for tailoring the teaching materials in a way that enhances the learning progress. A human teacher would achieve this by tracing a student's knowledge over important learning concepts in a task. Albeit, such teaching strategy is not well exploited yet in machine learning as current machine teaching methods tend to directly assess the progress on individual training samples without paying attention to the underlying learning concepts in a learning task. In this paper, we propose a novel method, called Knowledge Augmented Data Teaching (KADT), which can optimize a data teaching strategy for a student model by tracing its knowledge progress over multiple learning concepts in a learning task. Specifically, the KADT method incorporates a knowledge tracing model to dynamically capture the knowledge progress of a student model in terms of latent learning concepts. Then we develop an attention pooling mechanism to distill knowledge representations of a student model with respect to class labels, which enables to develop a data teaching strategy on critical training samples. We have evaluated the performance of the KADT method on four different machine learning tasks including knowledge tracing, sentiment analysis, movie recommendation, and image classification. The results comparing to the state-of-the-art methods empirically validate that KADT consistently outperforms others on all tasks. | ['Qing Wang', 'Ghodai Abdelrahman'] | 2021-11-13 | null | null | null | null | ['movie-recommendation'] | ['miscellaneous'] | [ 3.77382427e-01 1.03013083e-01 -2.76819319e-01 -6.18542194e-01
-2.37408772e-01 -4.70864207e-01 6.24156177e-01 6.01749361e-01
-5.64245582e-01 2.96274960e-01 1.19632157e-02 -2.46054262e-01
-3.61213356e-01 -1.06835866e+00 -7.07860410e-01 -6.75871670e-01
3.90720755e-01 4.46814448e-01 5.98170340e-01 -2.12984905e-01
3.61730248e-01 3.82017106e-01 -1.56097472e+00 5.46655357e-01
1.02118039e+00 9.84912157e-01 5.02639234e-01 5.11045098e-01
-5.22536337e-01 1.25126338e+00 -5.97143352e-01 -5.68022013e-01
-2.80934781e-01 -4.45603788e-01 -1.11770248e+00 2.51864130e-03
4.91241902e-01 -2.01090649e-01 1.99207608e-02 1.11542690e+00
-3.98518378e-03 4.94374782e-01 8.41547489e-01 -9.68387246e-01
-1.04335713e+00 8.00767303e-01 -2.99303830e-01 3.72307569e-01
-1.22480296e-01 -2.10564226e-01 9.96761918e-01 -8.95765960e-01
1.41592443e-01 1.01722038e+00 3.76460969e-01 6.95896566e-01
-7.22649038e-01 -5.40007651e-01 5.99229872e-01 4.50810462e-01
-9.62794721e-01 1.39426634e-01 9.37842250e-01 -6.98949397e-01
4.08819050e-01 -1.32434428e-01 8.04713190e-01 5.65881491e-01
5.02550006e-02 1.23140776e+00 1.03869081e+00 -4.28780854e-01
1.18345059e-01 5.90418875e-01 6.73178077e-01 8.69930267e-01
-2.87491947e-01 -3.42879891e-02 -7.64677465e-01 3.69080603e-01
3.72085512e-01 3.59539777e-01 -2.08830491e-01 -2.17517003e-01
-8.79856348e-01 8.18023503e-01 5.54289818e-01 3.11184406e-01
-3.78236353e-01 -1.77532434e-02 2.73808181e-01 4.45909917e-01
5.51691294e-01 4.09845561e-01 -7.83070028e-01 -1.67897809e-02
-6.81645095e-01 3.72593403e-02 2.29341730e-01 7.41167784e-01
7.67038882e-01 -6.26444593e-02 -5.56137741e-01 6.59869313e-01
4.63196486e-01 1.43185124e-01 9.18147624e-01 -6.79079890e-01
1.60447985e-01 1.20515013e+00 -3.74344200e-01 -6.71373188e-01
-4.40892018e-02 -6.17342591e-01 -3.96424204e-01 1.75836578e-01
2.59311289e-01 9.77619588e-02 -8.35741818e-01 1.75029194e+00
4.35040534e-01 3.37497801e-01 3.89668643e-01 5.23504436e-01
9.76372719e-01 8.22234333e-01 1.96450949e-01 -2.70557880e-01
1.29307640e+00 -1.33250153e+00 -6.03973866e-01 -2.38233078e-02
6.79868519e-01 -5.32319129e-01 1.47395194e+00 8.53327215e-01
-1.02081406e+00 -1.08898568e+00 -6.86962247e-01 -4.63624485e-02
-5.02030194e-01 1.08618572e-01 3.66691947e-01 4.59924370e-01
-8.40355337e-01 6.86433375e-01 -6.41528249e-01 -2.30961680e-01
7.06637561e-01 2.81962961e-01 9.92910191e-02 -3.51309367e-02
-1.11171639e+00 8.01481366e-01 4.58338648e-01 -1.50833189e-01
-1.40042341e+00 -9.59251940e-01 -5.34762025e-01 6.43986642e-01
5.21823227e-01 -2.74975032e-01 1.54315627e+00 -1.17853904e+00
-1.92098105e+00 7.50270009e-01 1.94593504e-01 -3.70947689e-01
1.71767056e-01 -4.34887409e-01 -1.00741968e-01 6.69288039e-02
-3.79751086e-01 6.80653572e-01 9.68922913e-01 -1.00955570e+00
-9.20805573e-01 -3.03756326e-01 2.99532026e-01 3.24710310e-01
-1.21652472e+00 -1.48783430e-01 -3.51242661e-01 -5.87552905e-01
-1.46182373e-01 -6.07447922e-01 -8.45836103e-02 -2.59807706e-01
1.25000238e-01 -8.49652290e-01 9.94660556e-01 -3.07675093e-01
1.32526672e+00 -2.00308037e+00 1.95645720e-01 8.51014704e-02
1.94556817e-01 6.32095754e-01 -8.19577202e-02 8.27839598e-02
6.30126819e-02 -1.63752869e-01 -6.20896108e-02 -5.28329052e-02
-6.28445670e-02 2.94037282e-01 -4.54119354e-01 -1.41384125e-01
-1.24672791e-02 8.62042427e-01 -1.18802321e+00 -2.80273139e-01
5.38673773e-02 3.17071140e-01 -9.45377111e-01 5.90885580e-01
-5.20297647e-01 4.76338774e-01 -6.07989490e-01 1.91648960e-01
4.12352383e-02 -4.39550936e-01 -5.02371490e-02 6.74577579e-02
4.69845720e-02 3.34226817e-01 -8.12823832e-01 1.57326925e+00
-5.10555267e-01 4.39134508e-01 -3.83268952e-01 -1.26594675e+00
1.08003807e+00 3.50108743e-01 1.85594797e-01 -6.16814196e-01
1.40481278e-01 -1.53961033e-01 2.09411427e-01 -7.11125791e-01
4.99618769e-01 -1.43300250e-01 2.73148537e-01 6.40179694e-01
3.41186881e-01 -1.12475734e-02 -1.34117916e-01 3.28734905e-01
7.78987885e-01 1.87813178e-01 -7.44384006e-02 -4.39663619e-01
6.67526484e-01 -8.35230052e-02 3.55626553e-01 5.61961889e-01
-2.22042426e-01 3.81103484e-03 2.33682901e-01 -4.53965366e-01
-3.89099270e-01 -7.99203634e-01 1.32519826e-01 1.85232294e+00
-2.31250212e-01 -3.74827206e-01 -9.71494734e-01 -1.19669497e+00
-1.54947817e-01 7.71840632e-01 -7.93535471e-01 -7.14947283e-01
-3.94389123e-01 -4.52191591e-01 -3.31521556e-02 7.86497653e-01
5.87284625e-01 -1.29750276e+00 -7.37750888e-01 1.00274391e-01
2.58427292e-01 -8.19829047e-01 -6.02744102e-01 2.60186553e-01
-1.12106061e+00 -1.01653659e+00 -5.19206882e-01 -1.07094371e+00
1.06664884e+00 2.72572339e-01 1.06503344e+00 3.22660297e-01
4.60227653e-02 7.12318063e-01 -5.67032933e-01 -5.71227670e-01
-3.92545938e-01 4.80029956e-02 2.54710048e-01 2.19840318e-01
6.34405315e-01 -3.55900615e-01 -6.25586689e-01 2.33476743e-01
-9.62547481e-01 2.37113744e-01 5.67216992e-01 7.34811783e-01
6.26325786e-01 6.21839166e-01 7.98862338e-01 -1.23358917e+00
6.54648840e-01 -4.03723538e-01 -5.98198891e-01 4.97858882e-01
-9.27358747e-01 2.23250970e-01 7.02836990e-01 -6.43150866e-01
-1.38792896e+00 -1.53901070e-01 -1.24374591e-01 -6.64240301e-01
-1.09977268e-01 9.38328326e-01 -5.12560159e-02 1.70578927e-01
5.60088933e-01 3.85005176e-01 -3.56196314e-01 -6.59914315e-01
2.88594425e-01 4.39208359e-01 4.72281277e-01 -1.03309667e+00
4.61433738e-01 8.81681070e-02 -3.41314405e-01 -4.67018515e-01
-1.72473478e+00 -5.23926020e-01 -7.92627096e-01 -2.90823758e-01
7.51816928e-01 -5.70809484e-01 -9.10121381e-01 6.22051716e-01
-7.97498107e-01 -8.15410316e-01 -6.44122124e-01 5.18469751e-01
-2.56811678e-01 -1.67617947e-01 -5.08671463e-01 -5.31348825e-01
-1.67686120e-01 -1.24732888e+00 3.47893357e-01 6.04035854e-01
1.08429097e-01 -1.44637740e+00 2.07626000e-01 5.84446013e-01
3.90360892e-01 -4.84119952e-01 1.21790946e+00 -1.07901013e+00
-3.56212169e-01 5.05888797e-02 4.61445227e-02 6.50774360e-01
2.35789254e-01 -3.29850577e-02 -1.21519220e+00 -2.98385292e-01
8.62729996e-02 -6.24500334e-01 9.35791731e-01 1.58827990e-01
1.81513166e+00 -5.10090768e-01 -5.07050380e-02 2.99927503e-01
9.65650499e-01 3.14359576e-01 2.19658911e-01 1.51929423e-01
8.07107329e-01 1.01402366e+00 5.97058654e-01 5.28494455e-02
4.57347900e-01 5.46980262e-01 2.56646454e-01 2.42740437e-01
-1.08851351e-01 -3.29443753e-01 5.16072690e-01 1.27540624e+00
-2.26836968e-02 -6.44926354e-02 -9.21599209e-01 6.41788125e-01
-1.64102626e+00 -7.68626869e-01 2.30899099e-02 2.02237344e+00
1.29614925e+00 2.97048450e-01 -1.12661272e-01 1.63291097e-01
3.08116436e-01 -1.86658755e-01 -5.76171398e-01 -4.74001527e-01
4.76584166e-01 3.04792076e-01 -1.04190782e-01 4.79807734e-01
-7.98657238e-01 1.15402961e+00 5.22477245e+00 8.03336203e-01
-1.16537607e+00 1.88859552e-01 4.89682525e-01 1.46751180e-01
-1.89003274e-01 -3.57213646e-01 -1.12668097e+00 2.54171371e-01
1.12867582e+00 -2.48952538e-01 1.17336541e-01 9.47331488e-01
-5.15865088e-02 1.04501881e-02 -1.26806223e+00 5.58580637e-01
1.13283202e-01 -1.20050001e+00 4.85670626e-01 -1.55313894e-01
1.05203640e+00 -4.17308360e-01 5.45138180e-01 8.27043891e-01
5.04553556e-01 -8.07462156e-01 4.90183383e-01 6.36618257e-01
2.40164891e-01 -8.18082452e-01 5.70628703e-01 5.46073496e-01
-7.67586529e-01 -2.40352839e-01 -2.47452930e-01 -1.14334188e-01
-4.43701267e-01 2.91500807e-01 -9.80942369e-01 4.84262258e-02
8.71906638e-01 7.74868071e-01 -7.68628776e-01 7.87093759e-01
-7.44607031e-01 1.25017297e+00 2.62983173e-01 -1.31231055e-01
4.50404435e-01 -3.81907597e-02 -4.90720607e-02 9.14614022e-01
3.18469882e-01 6.08477056e-01 2.99965829e-01 7.06124127e-01
-4.21593904e-01 1.76762298e-01 -5.61496913e-02 -2.75285244e-01
2.40976259e-01 1.19695532e+00 -9.03920770e-01 -4.79188323e-01
-5.40232480e-01 7.62205184e-01 5.07463932e-01 3.09073359e-01
-3.47996891e-01 -1.76744446e-01 4.26665068e-01 1.29671484e-01
4.24881250e-01 1.27729073e-01 -3.73154432e-02 -8.48438263e-01
-2.38394544e-01 -5.97546995e-01 6.12220764e-01 -8.85411263e-01
-1.10007405e+00 5.54189026e-01 -1.16217628e-01 -9.56289947e-01
5.41834012e-02 -8.96102309e-01 -8.00108790e-01 6.38106048e-01
-1.83678484e+00 -8.18552017e-01 -4.53793943e-01 9.18204486e-01
7.45303154e-01 -5.39319158e-01 6.55419827e-01 1.34516716e-01
-6.28490388e-01 6.05622470e-01 -1.51168928e-02 2.09700316e-01
6.57048821e-01 -1.48144829e+00 -8.61026347e-02 6.74054086e-01
3.89856488e-01 6.58075213e-01 4.82982427e-01 -4.58835661e-01
-1.09290135e+00 -1.13996243e+00 8.25025201e-01 -6.11365020e-01
6.31133914e-01 4.83442964e-05 -1.42913198e+00 8.40039134e-01
3.02284539e-01 -1.53653175e-02 8.73113811e-01 3.45275939e-01
-2.62391984e-01 -3.13801527e-01 -8.15866709e-01 2.10824147e-01
4.40105885e-01 -4.98685896e-01 -1.03975105e+00 2.68653065e-01
9.00313556e-01 -2.68649668e-01 -8.55481148e-01 3.76138359e-01
2.91134298e-01 -6.65991485e-01 6.91438973e-01 -1.05558527e+00
6.24458671e-01 -2.46483847e-01 2.06559002e-01 -1.62353575e+00
-2.79939532e-01 -1.32522583e-01 -4.38747793e-01 1.28127122e+00
1.53341591e-01 -2.08114982e-01 1.05932295e+00 3.49741846e-01
-6.23036861e-01 -1.02331090e+00 -1.70642972e-01 -4.51133192e-01
3.98228854e-01 -2.69424736e-01 5.57863653e-01 1.35703158e+00
1.64807215e-02 5.00987947e-01 5.61133549e-02 1.88696384e-01
4.20562238e-01 2.57207334e-01 4.24986035e-01 -1.64826453e+00
-2.06785068e-01 -5.97753048e-01 -1.76141068e-01 -1.16363943e+00
2.39504263e-01 -1.12871301e+00 -7.10103288e-02 -1.38819551e+00
4.19059515e-01 -5.58780432e-01 -8.68814051e-01 6.88863516e-01
-5.50447762e-01 -4.71975505e-02 -9.72785950e-02 4.34134528e-03
-8.67103398e-01 6.21124148e-01 1.59952724e+00 -2.01448619e-01
-2.82081723e-01 3.18613678e-01 -7.97740757e-01 8.90113533e-01
6.76340461e-01 -5.33081472e-01 -9.31330204e-01 -6.32441998e-01
4.88788813e-01 -2.01607347e-01 3.43946427e-01 -9.44023788e-01
6.14907086e-01 -3.06772321e-01 3.43754411e-01 -5.42921543e-01
4.99375015e-02 -9.54333782e-01 -6.27443671e-01 4.88523096e-01
-8.17828655e-01 -1.63516402e-01 2.72876531e-01 4.11258727e-01
-2.93870926e-01 -5.88367999e-01 8.99972320e-01 -9.40543860e-02
-1.03884768e+00 5.68103194e-01 -1.86964080e-01 8.14882293e-02
9.39743459e-01 1.47581175e-01 -3.27557832e-01 2.87178401e-02
-1.01109552e+00 5.15772879e-01 -8.58865902e-02 5.74032962e-01
9.52015758e-01 -1.30665386e+00 -6.67110920e-01 2.37313211e-01
2.78172076e-01 1.64370432e-01 3.03296775e-01 6.99208856e-01
-1.59241334e-01 4.33313310e-01 -2.00962484e-01 -6.44933164e-01
-1.29423904e+00 9.05427039e-01 3.88614714e-01 -3.60779256e-01
-4.60865408e-01 1.29806924e+00 6.34736657e-01 -6.99641764e-01
5.41715741e-01 -5.28718412e-01 -7.89240122e-01 2.99275845e-01
8.92468452e-01 1.93887070e-01 2.06092685e-01 -2.51971811e-01
1.98975980e-01 5.60894370e-01 -5.49917161e-01 3.12826872e-01
1.29996622e+00 1.53160524e-02 2.03043550e-01 6.40864432e-01
1.04933321e+00 -2.09380910e-01 -1.42398024e+00 -7.65806973e-01
1.47712678e-01 -5.90793788e-02 3.76395941e-01 -1.12844408e+00
-1.40907598e+00 1.42852759e+00 5.58641315e-01 2.32031364e-02
1.22851562e+00 9.17638168e-02 5.35258710e-01 6.92563176e-01
9.94977877e-02 -1.08540642e+00 7.93822527e-01 6.98142231e-01
4.84578133e-01 -1.35756528e+00 -2.16600627e-01 -2.57459260e-03
-6.42162144e-01 1.16678894e+00 1.06755781e+00 7.34719709e-02
8.37148070e-01 -1.66364461e-01 1.13917470e-01 -4.85778123e-01
-8.98230314e-01 -1.32213701e-02 6.62740409e-01 9.52759571e-03
6.19672477e-01 -1.78772405e-01 2.67641097e-01 9.52234745e-01
-1.26558170e-01 7.79752210e-02 3.39969367e-01 9.42430854e-01
-9.47916269e-01 -9.41017926e-01 -1.30325288e-01 4.37049896e-01
-3.57220471e-01 -2.53321707e-01 -2.69167721e-01 3.05668116e-01
2.35595852e-01 6.96681499e-01 3.36660817e-02 -3.56963485e-01
5.21746278e-01 3.89821470e-01 5.76230049e-01 -1.18173707e+00
-9.75318253e-01 -2.71134555e-01 -6.85049176e-01 -2.67517388e-01
-6.47603631e-01 -5.24488389e-01 -1.35616422e+00 7.17804674e-03
-2.64120817e-01 6.66762054e-01 2.55677015e-01 1.25914514e+00
-1.22402579e-01 1.22394216e+00 4.87417459e-01 -1.89334646e-01
-4.92424399e-01 -8.15761566e-01 -3.82532984e-01 2.73405880e-01
2.40482509e-01 -6.56893075e-01 -1.18378706e-01 4.42050517e-01] | [10.131913185119629, 7.023647308349609] |
739f551d-765a-46f6-af28-6c593efe76fe | adaptive-re-calibration-of-channel-wise | 2210.11722 | null | https://arxiv.org/abs/2210.11722v1 | https://arxiv.org/pdf/2210.11722v1.pdf | Adaptive re-calibration of channel-wise features for Adversarial Audio Classification | DeepFake Audio, unlike DeepFake images and videos, has been relatively less explored from detection perspective, and the solutions which exist for the synthetic speech classification either use complex networks or dont generalize to different varieties of synthetic speech obtained using different generative and optimization-based methods. Through this work, we propose a channel-wise recalibration of features using attention feature fusion for synthetic speech detection and compare its performance against different detection methods including End2End models and Resnet-based models on synthetic speech generated using Text to Speech and Vocoder systems like WaveNet, WaveRNN, Tactotron, and WaveGlow. We also experiment with Squeeze Excitation (SE) blocks in our Resnet models and found that the combination was able to get better performance. In addition to the analysis, we also demonstrate that the combination of Linear frequency cepstral coefficients (LFCC) and Mel Frequency cepstral coefficients (MFCC) using the attentional feature fusion technique creates better input features representations which can help even simpler models generalize well on synthetic speech classification tasks. Our models (Resnet based using feature fusion) trained on Fake or Real (FoR) dataset and were able to achieve 95% test accuracy with the FoR data, and an average of 90% accuracy with samples we generated using different generative models after adapting this framework. | ['Nikhitha Reddeddy', 'Abhinav Thimma Reddy', 'Vardhan Dongre'] | 2022-10-21 | null | null | null | null | ['synthetic-speech-detection'] | ['audio'] | [ 1.91346914e-01 2.01396957e-01 4.22845662e-01 -1.02061376e-01
-8.08409631e-01 -4.24012870e-01 1.01966012e+00 -5.36401153e-01
-2.15429828e-01 7.20625758e-01 5.41323721e-01 -4.69062738e-02
-9.07931700e-02 -6.67123854e-01 -5.87947786e-01 -6.48196220e-01
-2.16760281e-02 4.66069654e-02 2.46600807e-01 -4.94259298e-01
1.21293306e-01 3.73676866e-01 -1.95057881e+00 7.14724064e-01
5.68542361e-01 9.55103517e-01 2.40270227e-01 1.28816855e+00
2.17151001e-01 6.31389737e-01 -1.35978186e+00 -2.77545393e-01
3.77422571e-01 -6.22510254e-01 -3.88943195e-01 -1.07976057e-01
5.88886917e-01 -8.42474326e-02 -5.19148946e-01 1.05562484e+00
1.00386906e+00 8.80493373e-02 7.16517806e-01 -1.27832460e+00
-8.59815180e-01 8.83325815e-01 1.30256712e-01 4.51958805e-01
5.50955534e-01 4.03648347e-01 4.80657429e-01 -1.15857935e+00
4.58655119e-01 1.61045182e+00 8.85523438e-01 5.95997989e-01
-1.09716225e+00 -8.25856507e-01 -6.05289757e-01 3.61614168e-01
-1.37649548e+00 -8.79032612e-01 6.14516377e-01 -2.59109974e-01
1.39066374e+00 5.16821742e-01 8.34329367e-01 1.69726670e+00
2.03167796e-01 5.80598831e-01 1.14749479e+00 -5.19370735e-01
-1.44750765e-02 5.15103698e-01 -2.67131120e-01 3.87393773e-01
-1.77113816e-01 6.59391046e-01 -6.71470463e-01 1.37288481e-01
5.81887424e-01 -4.62099105e-01 -6.81468546e-01 3.99996787e-01
-1.42533815e+00 8.37694526e-01 3.12463462e-01 6.79959238e-01
-3.58262211e-01 3.28522742e-01 3.60879570e-01 7.64189065e-01
3.52264762e-01 5.69218516e-01 -1.62918299e-01 -1.60474733e-01
-1.34100723e+00 1.11304440e-01 7.86963165e-01 7.19347239e-01
4.41204101e-01 1.06124568e+00 -4.14931178e-01 8.90165925e-01
6.21606335e-02 7.26612389e-01 1.25991225e+00 -5.85389674e-01
2.09819153e-01 -2.13614702e-01 -2.27841690e-01 -1.04860485e+00
-2.96705693e-01 -7.93821514e-01 -6.93565726e-01 3.61830294e-01
5.78299016e-02 -4.16831106e-01 -1.14642775e+00 1.43536067e+00
-2.01480284e-01 4.85140562e-01 5.66169322e-01 8.77776980e-01
1.17619240e+00 8.92127633e-01 -4.93547112e-01 -3.88333648e-02
1.10613489e+00 -1.00120962e+00 -9.17387366e-01 2.10628957e-01
2.26563975e-01 -1.15567744e+00 8.49582374e-01 6.89539790e-01
-7.77218521e-01 -1.03464282e+00 -1.29174781e+00 3.55928838e-01
-7.05888867e-01 2.89402038e-01 1.66666582e-01 1.30910897e+00
-1.45652437e+00 5.35655975e-01 -3.36349219e-01 -4.10294503e-01
1.12175234e-01 2.70085037e-01 -3.81945759e-01 4.14266229e-01
-1.35593891e+00 9.23876762e-01 6.15690947e-01 -1.15023024e-01
-1.28658795e+00 -3.43354285e-01 -7.18076110e-01 1.62561715e-01
-4.71279547e-02 -4.47112083e-01 1.07579744e+00 -1.19568443e+00
-1.86101317e+00 1.76643148e-01 2.79302716e-01 -9.99972343e-01
5.00422657e-01 -4.97753322e-02 -1.00760102e+00 1.53909281e-01
-2.50679165e-01 9.33431625e-01 1.46663392e+00 -1.06602824e+00
-3.64302129e-01 3.25011700e-01 -2.85154849e-01 -3.58192623e-02
-4.19471025e-01 -1.81469485e-01 6.12203777e-01 -1.10714614e+00
-4.70650569e-02 -6.39091551e-01 3.67441535e-01 -6.33677959e-01
-5.16969919e-01 7.30054453e-02 1.44454062e+00 -7.71709085e-01
1.06186867e+00 -2.29372668e+00 -1.99261680e-01 -1.65085811e-02
1.08514063e-01 7.80866742e-01 -2.81128794e-01 5.51413953e-01
-4.77849573e-01 2.76577681e-01 1.73195645e-01 -1.89566478e-01
-8.24891552e-02 -3.72283459e-02 -4.50200349e-01 2.11282223e-01
2.77397454e-01 7.82424092e-01 -7.58470476e-01 6.36011139e-02
4.63006288e-01 8.18679094e-01 -6.29300594e-01 -1.71705231e-01
5.97107112e-02 1.08113952e-01 3.98972966e-02 4.24476177e-01
5.18262982e-01 3.81980509e-01 -5.15309155e-01 -2.72532403e-01
-3.88158485e-03 2.84621000e-01 -1.11133838e+00 1.25072896e+00
-6.26358747e-01 1.30969596e+00 -3.07108797e-02 -9.99648452e-01
1.15435874e+00 8.08860242e-01 -1.54448748e-01 -4.42159474e-01
2.44524822e-01 3.69080484e-01 3.76557738e-01 -4.66537446e-01
5.80247521e-01 -1.23698466e-01 1.90320477e-01 9.36037824e-02
6.79744244e-01 -6.68363094e-01 -2.18702063e-01 6.14658147e-02
9.14585948e-01 -2.99515992e-01 7.28930086e-02 -1.46682158e-01
5.57124257e-01 -2.42708743e-01 -9.67198014e-02 9.72166181e-01
7.60436431e-03 8.84165585e-01 8.67716745e-02 -1.59958676e-01
-1.06925857e+00 -9.72072244e-01 -1.91641852e-01 7.37580657e-01
-3.78257036e-01 -3.20505708e-01 -8.80415499e-01 -4.17411953e-01
-2.15743959e-01 1.03029120e+00 -3.80933225e-01 -3.94664735e-01
-2.71942317e-01 -7.32308567e-01 1.16065133e+00 6.53197169e-02
7.86176264e-01 -1.18549609e+00 -3.69056165e-01 4.51642066e-01
-2.76218657e-03 -1.00533795e+00 -2.04753667e-01 2.95809388e-01
-2.54947364e-01 -5.78086674e-01 -7.99994767e-01 -6.73129499e-01
-5.25380187e-02 -6.45075738e-02 5.96390009e-01 -3.64262849e-01
-2.74661630e-01 4.01717991e-01 -7.71554768e-01 -6.02798760e-01
-1.11666131e+00 -1.43309832e-01 3.16299379e-01 1.93316385e-01
1.62109844e-02 -6.33311987e-01 -2.66811818e-01 1.50770679e-01
-9.05331552e-01 -1.44270435e-01 4.59119737e-01 1.12001288e+00
-1.40321210e-01 9.49563980e-02 1.07561183e+00 -3.64034534e-01
1.13767505e+00 -3.70208174e-01 -1.88301370e-01 -2.78549083e-02
-3.76155257e-01 -1.02180958e-01 7.00018108e-01 -8.10190558e-01
-8.02067399e-01 -2.73892611e-01 -4.73486483e-01 -8.39175522e-01
-2.14106560e-01 8.89784917e-02 1.93702608e-01 -2.35122368e-01
1.30796373e+00 6.51755333e-01 -5.50465174e-02 -2.76853561e-01
3.23618352e-01 1.20700896e+00 5.01788199e-01 1.19594350e-01
7.01952994e-01 1.53247878e-01 -3.79708230e-01 -1.23535109e+00
-4.48640548e-02 5.68876937e-02 -5.07687815e-02 -3.65893513e-01
8.51264715e-01 -8.91341627e-01 -3.00445139e-01 7.17908740e-01
-1.24015725e+00 1.10796928e-01 -5.50492883e-01 8.51542532e-01
-5.25114179e-01 2.60770589e-01 -4.84966874e-01 -1.01340318e+00
-2.06388652e-01 -1.23578429e+00 9.97235477e-01 1.09591737e-01
-1.10148616e-01 -7.51100898e-01 -4.76805530e-02 7.13104159e-02
1.11636674e+00 1.56995729e-01 5.10380208e-01 -1.04783154e+00
-3.91138166e-01 -2.28842154e-01 1.17102914e-01 9.09070432e-01
2.28379548e-01 1.35664389e-01 -1.53549588e+00 -1.80019766e-01
1.24097630e-01 -2.99703538e-01 1.19604135e+00 2.69020438e-01
7.42829919e-01 -7.33367145e-01 4.22802158e-02 4.98342395e-01
9.64625955e-01 3.77723426e-01 8.36926341e-01 -1.64594457e-01
3.30383897e-01 1.96578667e-01 2.01179206e-01 3.46065372e-01
-3.24686766e-01 7.43570447e-01 4.17084336e-01 1.83279254e-02
-9.66932893e-01 -1.92299768e-01 7.69657135e-01 9.08295274e-01
6.09934628e-02 -7.70407200e-01 -5.44892490e-01 4.64259028e-01
-1.28045475e+00 -1.38882697e+00 3.32988836e-02 2.06383681e+00
4.99083459e-01 1.74789727e-01 2.11566295e-02 6.29767537e-01
9.02900577e-01 1.78293422e-01 -1.84443474e-01 -7.04501569e-01
-5.46104074e-01 6.87009931e-01 3.53144646e-01 4.87942785e-01
-1.02642727e+00 9.70311403e-01 6.75384092e+00 1.41521156e+00
-1.69815409e+00 3.75299454e-01 4.86537337e-01 2.93530617e-02
-1.85123906e-01 -4.95146453e-01 -4.53429669e-01 4.55795646e-01
1.34955740e+00 -1.25032902e-01 6.97360396e-01 7.40746379e-01
1.35554895e-01 2.12892428e-01 -8.09247792e-01 1.20865941e+00
3.91302973e-01 -1.37198019e+00 2.10179165e-01 -1.83351085e-01
7.24344075e-01 3.45074654e-01 3.86129498e-01 5.35987258e-01
1.88690409e-01 -1.29962826e+00 1.04590249e+00 5.39535224e-01
9.50986862e-01 -5.09113193e-01 7.73325503e-01 2.93173760e-01
-9.77915347e-01 -9.17499661e-02 -2.73500174e-01 3.43692116e-02
-1.44550160e-01 3.52083951e-01 -1.78542066e+00 4.64482397e-01
5.13399065e-01 3.52242619e-01 -4.05133158e-01 1.08407533e+00
1.06564194e-01 6.28438830e-01 -3.87966365e-01 -3.36236000e-01
2.36474156e-01 5.19429743e-01 1.04552042e+00 1.49420738e+00
8.42804432e-01 -5.82940400e-01 -2.84557074e-01 7.58670866e-01
1.81902915e-01 -3.91184352e-03 -7.96259522e-01 -2.82140642e-01
4.08604413e-01 1.02212334e+00 -5.24517953e-01 -4.55653757e-01
-9.69508961e-02 8.25281084e-01 -4.43591923e-01 5.48230827e-01
-9.50783432e-01 -6.61130428e-01 4.77952003e-01 1.96961299e-01
3.74836236e-01 4.94560637e-02 2.74947643e-01 -1.12417960e+00
-3.61267835e-01 -9.58010972e-01 -2.10897297e-01 -1.13192928e+00
-1.07270181e+00 1.26848841e+00 -2.59866305e-02 -1.49498308e+00
-7.98823774e-01 -5.96442461e-01 -5.84033906e-01 9.36467588e-01
-1.00112200e+00 -7.80568838e-01 -2.80759126e-01 6.59868777e-01
8.87608528e-01 -8.53856623e-01 7.93956757e-01 3.16650450e-01
-3.40954363e-02 7.37355649e-01 1.32814825e-01 -3.42533849e-02
5.14679074e-01 -9.76938009e-01 4.26687539e-01 7.84167409e-01
5.27625561e-01 2.25826159e-01 8.44248176e-01 -2.94750452e-01
-8.69176447e-01 -1.12066996e+00 5.14624953e-01 -1.15740724e-01
3.24656278e-01 -4.13412333e-01 -5.41784406e-01 3.86860728e-01
5.94636202e-01 -4.02187891e-02 2.95466274e-01 -6.51754737e-01
-2.39543870e-01 -2.04354115e-02 -1.55014789e+00 3.34218144e-01
8.45756531e-01 -6.56132519e-01 -6.92655981e-01 3.52708727e-01
8.68640840e-01 -3.05652738e-01 -4.23677117e-01 3.92155111e-01
4.24717754e-01 -1.39429641e+00 9.49692369e-01 -3.80434453e-01
1.02702856e-01 -4.09755260e-01 -4.87639010e-01 -1.80971551e+00
-8.90259352e-03 -7.92521000e-01 1.71462536e-01 1.21363175e+00
6.83056891e-01 -1.04179788e+00 3.53419334e-01 -5.05411267e-01
-5.20713508e-01 -3.75629961e-01 -1.24338460e+00 -9.85582113e-01
-3.63068670e-01 -6.66473985e-01 5.85081756e-01 9.07092929e-01
-1.60898313e-01 3.56876761e-01 -6.07102871e-01 7.74677694e-02
-6.43770117e-03 -4.53403473e-01 8.34885359e-01 -9.25244510e-01
-5.27349234e-01 -5.56881487e-01 -1.01018298e+00 -7.20505357e-01
-7.57028833e-02 -9.55789685e-01 -2.08970159e-01 -1.21301270e+00
-4.64557856e-01 -8.13491046e-02 2.59595253e-02 3.46395046e-01
4.23025936e-01 4.47841525e-01 2.87420839e-01 -1.28476173e-01
1.06666476e-01 5.94824553e-01 1.09895325e+00 -1.97383434e-01
-6.00793678e-03 5.16171381e-02 -2.36790791e-01 5.18122911e-01
6.32418633e-01 -4.86205995e-01 -4.67016637e-01 9.35848430e-02
-1.93905741e-01 3.09690744e-01 6.93753064e-01 -1.71863174e+00
6.14861287e-02 5.76545954e-01 5.66741467e-01 -4.18091208e-01
8.49929810e-01 -5.98795712e-01 4.64226067e-01 6.18400693e-01
-1.72101885e-01 -1.86463948e-02 1.97176710e-01 4.11472142e-01
-6.32278800e-01 -1.99339017e-01 7.81018674e-01 -2.12696746e-01
-5.90755701e-01 -3.63559246e-01 -6.70339465e-01 -3.05919647e-01
8.18373382e-01 -4.93611515e-01 -5.40517569e-01 -1.06503868e+00
-1.12777698e+00 -4.97970283e-01 -6.47311881e-02 7.07086444e-01
8.76681328e-01 -1.19850755e+00 -8.52739036e-01 5.78814328e-01
-3.21011871e-01 -6.41436875e-01 1.52180612e-01 6.36499703e-01
-5.82436919e-01 5.31922519e-01 -3.45367134e-01 -6.66527510e-01
-1.21143365e+00 4.53503370e-01 7.82667816e-01 3.50044936e-01
-2.28406325e-01 1.01383281e+00 9.48535204e-02 -3.86489809e-01
1.65455759e-01 -5.53869009e-01 -2.34090284e-01 1.45693704e-01
3.21939349e-01 4.96852666e-01 4.22995538e-01 -7.63197422e-01
-2.39271536e-01 2.39075497e-01 3.21022093e-01 -4.75199312e-01
1.06973553e+00 2.19949350e-01 4.74931806e-01 1.87062845e-01
1.16479135e+00 3.53881001e-01 -8.14237535e-01 1.39671579e-01
-5.01001418e-01 -3.55792493e-01 1.36930406e-01 -9.74735081e-01
-1.06526566e+00 8.71312499e-01 1.08989346e+00 8.14266384e-01
1.12493944e+00 -7.65226260e-02 5.76494098e-01 2.64864415e-01
2.99514621e-01 -6.62640572e-01 2.81799614e-01 4.51426923e-01
1.17383516e+00 -7.89095938e-01 -5.77244163e-01 -3.00351471e-01
-5.37687659e-01 1.35624754e+00 3.26412559e-01 -3.26355249e-01
5.65495670e-01 3.67010117e-01 7.47275129e-02 1.17197499e-01
-8.50887537e-01 -2.31432319e-01 4.46048565e-02 8.13162386e-01
1.76720038e-01 5.42256720e-02 -9.01253670e-02 3.21909964e-01
-9.87932980e-01 -3.45943034e-01 8.97429764e-01 3.23507130e-01
-5.90647638e-01 -8.22283924e-01 -7.06992686e-01 5.22634506e-01
-4.44007337e-01 -4.60833013e-01 -3.41153681e-01 8.65714729e-01
3.98731053e-01 1.22531974e+00 2.76450962e-02 -9.29524302e-01
2.37969607e-01 2.75061071e-01 3.19694161e-01 -5.82379520e-01
-9.15271580e-01 1.74993262e-01 1.80090919e-01 -7.83743858e-02
-2.83671647e-01 -4.88466054e-01 -8.22143435e-01 -6.14594109e-02
-7.77381241e-01 2.24972889e-01 9.54828084e-01 6.19174898e-01
3.37353021e-01 8.42130840e-01 6.78728640e-01 -1.06507564e+00
-4.73843485e-01 -1.58841515e+00 -4.71168995e-01 5.57860471e-02
7.92980611e-01 -5.38451314e-01 -7.91870654e-01 2.78276540e-02] | [15.160063743591309, 6.243263244628906] |
c2a61d49-bb6d-4c7a-b33f-51ef0ee7e25f | medical-image-denoising-using-convolutional | 1608.04667 | null | http://arxiv.org/abs/1608.04667v2 | http://arxiv.org/pdf/1608.04667v2.pdf | Medical image denoising using convolutional denoising autoencoders | Image denoising is an important pre-processing step in medical image
analysis. Different algorithms have been proposed in past three decades with
varying denoising performances. More recently, having outperformed all
conventional methods, deep learning based models have shown a great promise.
These methods are however limited for requirement of large training sample size
and high computational costs. In this paper we show that using small sample
size, denoising autoencoders constructed using convolutional layers can be used
for efficient denoising of medical images. Heterogeneous images can be combined
to boost sample size for increased denoising performance. Simplest of networks
can reconstruct images with corruption levels so high that noise and signal are
not differentiable to human eye. | ['Lovedeep Gondara'] | 2016-08-16 | null | null | null | null | ['medical-image-denoising'] | ['computer-vision'] | [ 1.60046950e-01 -5.49996532e-02 4.52950001e-01 -3.27715963e-01
-5.99262714e-01 2.33009402e-02 2.32392982e-01 7.45330751e-02
-6.75953865e-01 7.24642158e-01 1.56084567e-01 1.16176486e-01
-2.66683679e-02 -8.43781888e-01 -2.98551053e-01 -1.08759522e+00
-2.07787454e-02 -1.32744551e-01 2.01283336e-01 -3.11145395e-01
2.34733522e-03 4.32169288e-01 -1.33992290e+00 3.99833679e-01
7.63856471e-01 9.31783617e-01 2.96414703e-01 7.40970969e-01
3.57980700e-03 6.65910065e-01 -7.56455600e-01 -1.68052822e-01
3.63168508e-01 -5.07597089e-01 -3.30235332e-01 1.50912404e-01
5.64834923e-02 -5.51444530e-01 -5.44651926e-01 1.18585622e+00
1.06616747e+00 1.20321391e-02 3.28988284e-01 -5.63865662e-01
-6.71302676e-01 3.20072085e-01 -5.11731029e-01 4.60409224e-01
-6.42149523e-02 6.81561753e-02 1.83659196e-01 -7.56683767e-01
4.77348149e-01 1.11345625e+00 9.85707164e-01 6.95769727e-01
-1.22881591e+00 -5.29159665e-01 -5.08876145e-01 3.83411884e-01
-9.60167170e-01 -3.87977391e-01 9.93002832e-01 -2.70469803e-02
7.95919120e-01 2.77207404e-01 5.42948782e-01 1.04225135e+00
5.93348801e-01 7.03357816e-01 1.23604620e+00 -2.44834647e-01
5.65332286e-02 -6.12692200e-02 -4.51117717e-02 4.79676038e-01
2.01090157e-01 1.06893837e-01 -2.80380875e-01 1.60803393e-01
7.71105349e-01 3.24131280e-01 -3.95089149e-01 2.66851872e-01
-1.08623779e+00 9.11214769e-01 6.03866279e-01 7.26276994e-01
-7.64492989e-01 1.99355036e-01 7.56314754e-01 5.44795215e-01
5.41687191e-01 2.78424412e-01 -2.27106795e-01 2.19882190e-01
-1.08641517e+00 -8.30542520e-02 4.58045959e-01 3.99313837e-01
3.08123618e-01 3.96376073e-01 -5.90407327e-02 9.27910089e-01
-1.11863367e-01 2.93981522e-01 8.61104786e-01 -1.11053944e+00
1.77511703e-02 2.60145545e-01 -2.25461110e-01 -1.16183543e+00
-5.17875612e-01 -6.76736414e-01 -1.78341031e+00 6.32331908e-01
3.23865205e-01 -1.15304716e-01 -1.17437303e+00 1.13241971e+00
3.07597756e-01 2.38506705e-01 1.83935046e-01 7.43158400e-01
1.05499291e+00 7.21856833e-01 -8.75722393e-02 -3.02511811e-01
1.34668851e+00 -7.09525824e-01 -1.23267043e+00 -1.41427983e-02
1.66136742e-01 -1.07019532e+00 5.90851247e-01 1.03701186e+00
-1.37038958e+00 -7.52216280e-01 -1.10592020e+00 -1.43182084e-01
-1.63769111e-01 -1.39863491e-01 6.38839662e-01 8.62188220e-01
-1.18293989e+00 1.01553297e+00 -1.05542529e+00 -1.71580866e-01
8.58578086e-01 5.33202529e-01 -5.97034335e-01 -3.34582925e-01
-9.73159194e-01 9.11524117e-01 1.94131389e-01 5.92711389e-01
-1.04022527e+00 -5.78984320e-01 -6.52628899e-01 5.30617572e-02
-1.81387402e-02 -7.91409552e-01 9.02920187e-01 -1.07486475e+00
-1.39913428e+00 6.54478371e-01 6.10033087e-02 -6.88242674e-01
4.69329685e-01 -2.22503349e-01 -5.39243758e-01 3.64119381e-01
-3.28222603e-01 2.96602041e-01 1.20545101e+00 -1.20783257e+00
-4.07539666e-01 -3.34839314e-01 -4.09236789e-01 -9.79927108e-02
-3.59991550e-01 -7.58771459e-03 -6.30983710e-02 -9.45483148e-01
4.33534831e-01 -4.46162850e-01 -6.04390442e-01 5.22213429e-02
-1.40038505e-01 1.96212679e-01 8.37114513e-01 -9.42716539e-01
9.70947266e-01 -2.02654243e+00 -1.98086575e-02 -3.16248164e-02
3.40353131e-01 6.23887837e-01 -1.97572187e-01 4.55377221e-01
-1.62824050e-01 1.90838687e-02 -2.52956450e-01 -3.48176152e-01
-5.85844100e-01 3.55983227e-01 1.79511175e-01 7.75478125e-01
1.00392215e-01 6.92276120e-01 -6.63101673e-01 -4.21991825e-01
5.12279212e-01 1.02178395e+00 -4.40306336e-01 2.00433329e-01
4.29766208e-01 6.73468173e-01 -2.02205300e-01 4.60431665e-01
9.75558996e-01 1.22595010e-02 -1.67688519e-01 -6.15750372e-01
2.53320873e-01 -3.45309466e-01 -1.15212834e+00 1.37323225e+00
-4.75895792e-01 7.81639218e-01 5.97467422e-01 -1.56396782e+00
7.69856572e-01 6.65992081e-01 6.48472071e-01 -8.34273398e-01
3.87906283e-01 2.34422952e-01 2.22713098e-01 -8.38927031e-01
9.10284594e-02 -7.64628530e-01 4.98749763e-01 -6.08626567e-02
1.34332404e-01 1.42908869e-02 1.15784705e-01 -1.57987103e-01
1.20752025e+00 -3.66866440e-01 3.46918285e-01 -1.83438882e-01
6.58072472e-01 -1.89190611e-01 5.89149773e-01 6.93340063e-01
-2.80510277e-01 1.00682282e+00 6.71033859e-02 -6.93150520e-01
-1.24338305e+00 -6.75516188e-01 -2.73933232e-01 4.94332105e-01
-1.13163213e-03 5.87291978e-02 -8.29088151e-01 -1.31643072e-01
-3.24909210e-01 7.67405629e-02 -6.79983854e-01 -1.91343158e-01
-8.94058764e-01 -1.00688148e+00 4.42334503e-01 4.70097244e-01
7.57676005e-01 -1.22372413e+00 -6.24033749e-01 5.62362432e-01
-1.36834517e-01 -9.13855195e-01 -4.48514894e-02 3.86676639e-01
-1.39640236e+00 -8.90830278e-01 -1.21946597e+00 -1.04889560e+00
8.63828659e-01 3.64828527e-01 1.06809330e+00 5.64698219e-01
-6.55744851e-01 2.09700212e-01 -4.39463943e-01 -5.37552238e-01
-5.93589306e-01 -3.82328540e-01 -1.79682642e-01 -7.66661838e-02
1.46207660e-02 -7.72309542e-01 -1.21776462e+00 -6.31545484e-02
-1.40784061e+00 -1.86338544e-01 8.06682825e-01 1.27932954e+00
6.55462265e-01 6.75482333e-01 6.43097699e-01 -9.61849213e-01
7.93302476e-01 -2.79450148e-01 -1.83256298e-01 -2.13886604e-01
-5.10469258e-01 -4.26714942e-02 9.20729995e-01 -2.30252013e-01
-1.16252279e+00 -1.57443479e-01 -7.41108119e-01 -1.82948500e-01
-3.58479083e-01 7.00888097e-01 2.11074993e-01 -1.11661926e-01
7.12003291e-01 3.43385518e-01 5.26736319e-01 -6.66308343e-01
-1.10054553e-01 5.35717368e-01 5.83468974e-01 1.55918866e-01
5.82847655e-01 8.99059772e-01 2.42828026e-01 -1.03497827e+00
-5.42308569e-01 -3.77330661e-01 -4.71666247e-01 -1.11900315e-01
9.75572050e-01 -9.89554226e-01 -5.15322924e-01 8.03404152e-01
-9.53400850e-01 -7.06542581e-02 -1.85021609e-01 4.39669818e-01
-1.06900722e-01 6.52388394e-01 -9.84365761e-01 -5.62699199e-01
-6.61738038e-01 -1.12456322e+00 6.73003197e-01 3.64166737e-01
1.97827592e-01 -1.10723543e+00 -2.15284064e-01 2.61354387e-01
8.76045346e-01 5.50795138e-01 6.42999411e-01 -3.22646141e-01
-1.75918490e-01 -6.80494189e-01 8.07741005e-03 9.61136639e-01
3.69357765e-01 -2.89411038e-01 -9.36870694e-01 -5.22624969e-01
6.29012704e-01 -3.96277905e-02 1.10856259e+00 1.03615725e+00
1.39940727e+00 -1.15128249e-01 -1.69086069e-01 7.54771829e-01
1.66137147e+00 2.26233244e-01 1.13285291e+00 4.37578559e-01
4.31580722e-01 3.73486489e-01 1.71505645e-01 1.96347311e-01
-2.98845172e-01 2.45087631e-02 4.86385226e-01 -5.65257430e-01
-5.78755736e-01 4.23170269e-01 7.46195242e-02 1.01218891e+00
-3.86540145e-01 -1.02784127e-01 -5.68266809e-01 6.08994484e-01
-1.46442187e+00 -1.08988631e+00 -4.67682689e-01 2.00232434e+00
6.86041057e-01 6.67361617e-02 -2.10929543e-01 5.38003862e-01
4.87559915e-01 -8.21543932e-02 -3.98341030e-01 -1.83876291e-01
-2.89013594e-01 5.76671362e-01 4.38131869e-01 3.93006474e-01
-1.05033457e+00 2.79228300e-01 7.05039454e+00 8.25532258e-01
-1.07635331e+00 2.39742890e-01 9.06482637e-01 8.14635381e-02
-1.87069010e-02 -4.57168430e-01 -1.70376688e-01 5.02037346e-01
7.67243147e-01 2.38719463e-01 2.02675268e-01 5.13924658e-01
4.46330011e-01 -3.16436648e-01 -6.78732574e-01 1.27493489e+00
-2.73548495e-02 -1.31705773e+00 -1.07266352e-01 -4.93013933e-02
6.71510994e-01 -9.35699716e-02 9.87108871e-02 3.35091073e-03
-7.59698004e-02 -1.28020728e+00 6.55404478e-02 5.97546577e-01
3.09409052e-01 -9.07001615e-01 1.37651753e+00 3.32593799e-01
-7.41887629e-01 -2.06043094e-01 -6.01513267e-01 -4.86448407e-02
3.20358008e-01 1.05977035e+00 -2.00004101e-01 4.84609544e-01
1.08043921e+00 6.02019906e-01 -3.08248252e-01 1.19810343e+00
-8.25585425e-02 8.34809840e-01 -2.32973546e-01 3.28693479e-01
1.75066248e-01 -3.45399708e-01 4.61102962e-01 1.14622343e+00
4.52470213e-01 2.24609002e-01 -2.40509450e-01 3.51085484e-01
3.72753292e-02 -9.06376690e-02 -5.11260688e-01 2.47410983e-01
-1.20979421e-01 1.31851780e+00 -9.15127635e-01 -4.78143007e-01
-4.00954217e-01 1.18049002e+00 -2.58102208e-01 3.94736767e-01
-5.07691145e-01 -4.17719722e-01 4.53966498e-01 7.77636617e-02
4.13644344e-01 -1.53049290e-01 -4.40131605e-01 -8.73517454e-01
-2.67078519e-01 -1.06248295e+00 3.98859560e-01 -5.76535881e-01
-1.33458912e+00 7.52823591e-01 -4.22692895e-01 -1.26296437e+00
2.00738817e-01 -7.02721357e-01 -6.23757601e-01 8.18402350e-01
-1.52010369e+00 -9.12149251e-01 -4.65595990e-01 6.15182281e-01
7.47702122e-01 -2.22994059e-01 8.03108513e-01 5.63176572e-01
-5.36958635e-01 1.72945678e-01 4.85417217e-01 5.70539422e-02
5.57453215e-01 -1.18774211e+00 -6.58624843e-02 9.62370396e-01
-1.68844461e-01 5.32876134e-01 1.07559276e+00 -4.43188339e-01
-1.32835531e+00 -7.03023493e-01 4.58118051e-01 8.83589312e-02
1.23419270e-01 -4.72414307e-03 -1.14295030e+00 1.77130222e-01
7.29671359e-01 2.58772254e-01 6.29078686e-01 -2.94398040e-01
2.16067359e-01 -4.86398011e-01 -1.41930819e+00 4.17341202e-01
3.98043603e-01 -8.34950805e-02 -3.99244517e-01 1.54291138e-01
1.58764750e-01 -3.66914928e-01 -1.07384467e+00 3.24491680e-01
3.41123700e-01 -1.24388194e+00 1.21005499e+00 -2.96915233e-01
4.84866679e-01 -2.88720727e-01 2.13372946e-01 -1.30816317e+00
-5.39303482e-01 -5.48047364e-01 5.54189086e-02 8.92169714e-01
-1.08509390e-02 -4.77041632e-01 7.24160731e-01 2.76950270e-01
-1.79301664e-01 -6.06442630e-01 -9.36593711e-01 -6.73092008e-01
1.97682813e-01 -1.33467793e-01 1.78765625e-01 5.79895675e-01
-3.96188319e-01 1.41993180e-01 -6.55809820e-01 3.74879055e-02
8.35576236e-01 -2.60443717e-01 3.47236514e-01 -1.12433791e+00
-4.16072421e-02 -4.13121939e-01 -4.56097692e-01 -5.84710658e-01
-3.22355896e-01 -6.59976363e-01 7.00057149e-02 -1.92274165e+00
1.84190169e-01 2.15275101e-02 -5.00250041e-01 3.05628389e-01
-2.77335554e-01 7.73954332e-01 -1.32679015e-01 1.05040200e-01
-3.78708579e-02 3.20990711e-01 1.51846087e+00 -2.71007836e-01
-1.91129539e-02 -1.02369905e-01 -5.97458601e-01 6.86551452e-01
1.03058398e+00 -5.71699321e-01 -2.90249556e-01 -5.56936562e-01
9.68691260e-02 -5.65412939e-02 3.44095707e-01 -1.30478203e+00
1.92627966e-01 3.85651857e-01 8.27299654e-01 -4.22676980e-01
1.72676608e-01 -1.12756860e+00 5.20371079e-01 9.16374445e-01
-1.21633545e-01 -1.63473651e-01 1.65791199e-01 6.26266956e-01
-7.30905533e-01 -4.57337856e-01 1.05872881e+00 -5.60654163e-01
-5.56977272e-01 1.26938835e-01 -6.47280991e-01 -6.32581949e-01
8.67933452e-01 -5.61627388e-01 2.15194702e-01 -4.70769763e-01
-8.93247604e-01 -1.89709112e-01 -1.17116445e-03 -8.45982358e-02
8.75589311e-01 -1.11224902e+00 -8.62691402e-01 1.85356345e-02
-4.95731115e-01 2.71378569e-02 6.94807708e-01 1.14035702e+00
-1.01643908e+00 -5.63054718e-02 -4.23533052e-01 -5.44375062e-01
-1.48191071e+00 7.11268723e-01 3.90075773e-01 -2.76329339e-01
-1.03896308e+00 7.96882212e-01 -4.66412604e-02 5.09795696e-02
1.04651250e-01 -2.93572634e-01 -3.02474380e-01 -5.01941741e-02
7.95778036e-01 4.66403455e-01 2.91390687e-01 -4.00328189e-01
-9.97200087e-02 4.28890854e-01 -8.83360729e-02 3.11798811e-01
1.98465478e+00 -2.20983729e-01 -2.65499800e-01 -4.73603513e-03
1.22393572e+00 -3.11110675e-01 -1.18495941e+00 7.46705979e-02
-2.31426522e-01 -3.93418789e-01 5.79535782e-01 -6.78332925e-01
-1.68741453e+00 1.09116423e+00 1.14820564e+00 4.54143524e-01
1.75719965e+00 -4.20698345e-01 1.06740749e+00 1.60256132e-01
1.20214470e-01 -1.23341048e+00 1.64579898e-01 -2.12028399e-02
9.17673767e-01 -1.58179319e+00 2.42837444e-01 -2.89057404e-01
-3.95680845e-01 1.41477156e+00 1.45913616e-01 -5.36297202e-01
8.41604471e-01 5.13106346e-01 4.54717785e-01 -2.73856759e-01
-2.60501266e-01 -9.26926956e-02 1.72863618e-01 7.94826746e-01
6.91394389e-01 -2.86764830e-01 -5.74564695e-01 5.03534138e-01
2.65731394e-01 1.37221932e-01 5.95202088e-01 8.59620392e-01
-5.25933504e-01 -1.16669929e+00 -6.23668194e-01 6.60264194e-01
-1.15132403e+00 -9.78603289e-02 3.40860873e-01 6.07689857e-01
2.77082205e-01 1.00021017e+00 -1.85438678e-01 -5.36136553e-02
3.30058426e-01 -3.63536596e-01 5.34756839e-01 -1.84851855e-01
-9.28268909e-01 4.33705032e-01 -1.21006534e-01 -4.99295264e-01
-6.65569305e-01 -3.17889512e-01 -9.18310821e-01 -2.88573354e-01
-2.88658291e-01 -1.44707397e-01 6.32643521e-01 7.63272345e-01
1.93641692e-01 7.94739485e-01 4.31218356e-01 -1.01309347e+00
-4.77823108e-01 -1.10464394e+00 -8.84275913e-01 7.10074604e-01
6.42327070e-01 -1.35219455e-01 -2.93109775e-01 5.27420878e-01] | [13.145330429077148, -2.524794578552246] |
b35276b4-a834-437d-96bd-5ca5736c43d7 | a-robust-learning-approach-to-domain-adaptive | 1904.02361 | null | https://arxiv.org/abs/1904.02361v3 | https://arxiv.org/pdf/1904.02361v3.pdf | A Robust Learning Approach to Domain Adaptive Object Detection | Domain shift is unavoidable in real-world applications of object detection. For example, in self-driving cars, the target domain consists of unconstrained road environments which cannot all possibly be observed in training data. Similarly, in surveillance applications sufficiently representative training data may be lacking due to privacy regulations. In this paper, we address the domain adaptation problem from the perspective of robust learning and show that the problem may be formulated as training with noisy labels. We propose a robust object detection framework that is resilient to noise in bounding box class labels, locations and size annotations. To adapt to the domain shift, the model is trained on the target domain using a set of noisy object bounding boxes that are obtained by a detection model trained only in the source domain. We evaluate the accuracy of our approach in various source/target domain pairs and demonstrate that the model significantly improves the state-of-the-art on multiple domain adaptation scenarios on the SIM10K, Cityscapes and KITTI datasets. | ['Mani Ranjbar', 'Mehran Khodabandeh', 'Arash Vahdat', 'William G. Macready'] | 2019-04-04 | a-robust-learning-approach-to-domain-adaptive-1 | http://openaccess.thecvf.com/content_ICCV_2019/html/Khodabandeh_A_Robust_Learning_Approach_to_Domain_Adaptive_Object_Detection_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/Khodabandeh_A_Robust_Learning_Approach_to_Domain_Adaptive_Object_Detection_ICCV_2019_paper.pdf | iccv-2019-10 | ['robust-object-detection'] | ['computer-vision'] | [ 4.02220070e-01 1.35985196e-01 -7.38685951e-02 -5.55665910e-01
-8.66347373e-01 -7.39580214e-01 5.63101530e-01 5.19641265e-02
-5.40124238e-01 7.62664616e-01 -4.53263015e-01 -9.44911465e-02
1.12335540e-01 -5.78355789e-01 -8.74278188e-01 -6.64652288e-01
1.07350163e-01 8.25642586e-01 8.57992649e-01 -6.62156716e-02
-1.30677551e-01 5.31853616e-01 -1.46156561e+00 2.21600056e-01
5.61373174e-01 1.01661777e+00 1.92621619e-01 7.17951000e-01
1.37220755e-01 3.67957026e-01 -7.73461044e-01 -2.15163454e-01
8.01080704e-01 1.49613753e-01 -4.83202636e-01 5.50857425e-01
7.29888320e-01 -4.77951854e-01 -2.78501451e-01 1.16821253e+00
2.76958019e-01 1.17800355e-01 8.11677158e-01 -1.35734618e+00
-2.89844334e-01 -2.12425925e-03 -6.03004813e-01 3.10668409e-01
-1.03534028e-01 -2.91454848e-02 3.54281008e-01 -6.56136811e-01
7.06509054e-01 1.00784123e+00 5.42633295e-01 6.16501093e-01
-1.32119751e+00 -8.89511645e-01 4.28536624e-01 1.90547630e-01
-1.59538066e+00 -4.56449956e-01 6.56927466e-01 -5.33259213e-01
4.41817462e-01 2.90808431e-03 -2.82943528e-02 1.21940970e+00
-2.68814176e-01 7.08710492e-01 9.02125537e-01 -3.90226930e-01
4.62789536e-01 7.98961937e-01 7.29173645e-02 2.27786079e-01
6.03230715e-01 3.27522665e-01 -2.24479228e-01 -2.65946507e-01
4.20055598e-01 -7.02845231e-02 1.18336163e-01 -9.66233730e-01
-9.73504126e-01 7.83430576e-01 2.29355618e-01 -7.60018602e-02
-4.54099327e-02 -3.09825748e-01 5.35581708e-01 2.10975736e-01
6.17552042e-01 1.64998963e-01 -7.12546945e-01 2.89378881e-01
-7.28118718e-01 3.63683730e-01 7.41981030e-01 1.58346081e+00
6.37629867e-01 -1.77873537e-01 6.23624884e-02 7.98269451e-01
3.28562818e-02 8.46701920e-01 1.71599492e-01 -8.21676910e-01
5.53265810e-01 5.11979580e-01 5.38969338e-01 -6.24126196e-01
-4.53064412e-01 -3.25435311e-01 -5.70257723e-01 2.65633672e-01
8.98824275e-01 -2.57599562e-01 -1.06621611e+00 1.57048416e+00
7.98799813e-01 3.45468730e-01 2.10995942e-01 7.01958716e-01
6.39540851e-01 3.36278111e-01 1.68818831e-01 -1.62184089e-02
1.21810877e+00 -6.65959358e-01 -4.50796008e-01 -7.20036745e-01
7.88970947e-01 -5.61878920e-01 5.94737887e-01 3.24809670e-01
-3.56006593e-01 -6.55518651e-01 -9.96363044e-01 3.03463131e-01
-7.74686575e-01 3.08776200e-01 -3.29561606e-02 8.41058016e-01
-4.93565381e-01 -3.19030648e-03 -5.94396114e-01 -9.03943241e-01
5.05355537e-01 2.72446603e-01 -4.98063654e-01 -3.42259437e-01
-1.06601238e+00 9.79948819e-01 6.98552787e-01 -2.79629529e-01
-1.13473666e+00 -5.22509456e-01 -6.99332237e-01 -2.65963525e-01
8.60734046e-01 -2.17548713e-01 1.45776367e+00 -9.13353860e-01
-1.14363730e+00 1.02287149e+00 1.64861400e-02 -6.25656664e-01
8.70867074e-01 -6.06795475e-02 -6.95645988e-01 5.37891465e-04
3.01943630e-01 5.96407890e-01 1.07296598e+00 -1.41337013e+00
-1.34826779e+00 -4.14704025e-01 -7.89402872e-02 -5.26921265e-02
-8.18189681e-02 -6.88195452e-02 -2.70935208e-01 -4.23705101e-01
-3.02817225e-01 -1.16505730e+00 -3.56084406e-01 1.40000969e-01
-1.32176414e-01 3.96112613e-02 1.44976318e+00 -3.98839176e-01
6.96487188e-01 -2.35822797e+00 -4.90205199e-01 3.15098047e-01
-5.65879568e-02 4.88669246e-01 -9.31911767e-02 -1.12595499e-01
5.15243486e-02 -1.97152525e-01 -2.34214664e-01 -2.10344553e-01
-1.55604154e-01 6.11583054e-01 -3.46315056e-01 7.55966961e-01
3.12014669e-01 4.10130173e-01 -9.95885134e-01 -5.81476629e-01
2.88385063e-01 1.51879907e-01 -3.00701886e-01 2.72248149e-01
-1.45265400e-01 5.22196651e-01 -5.63900232e-01 6.49547040e-01
9.97949719e-01 -9.67866108e-02 1.90596998e-01 2.59972304e-01
7.30536357e-02 -9.27820802e-02 -1.35639381e+00 1.23587692e+00
-2.44935811e-01 6.61004841e-01 2.43728578e-01 -1.14493942e+00
1.03302979e+00 2.33945534e-01 2.18128383e-01 -5.30422509e-01
-1.67868659e-02 1.63083553e-01 -5.27487248e-02 -4.80960160e-01
5.04041910e-01 1.19402900e-01 -2.77531117e-01 2.52373833e-02
-1.02615155e-01 3.01210606e-03 1.97095752e-01 -2.22882647e-02
1.06072021e+00 -2.32969508e-01 5.90423226e-01 -1.81867436e-01
5.69677711e-01 4.30796832e-01 7.22036302e-01 9.97588217e-01
-6.80238485e-01 6.00057423e-01 3.62437636e-01 -5.12998521e-01
-1.28930843e+00 -1.11140847e+00 -3.83825243e-01 1.36943507e+00
1.39145121e-01 3.00803483e-01 -8.60656381e-01 -1.19674635e+00
1.23597436e-01 7.22642541e-01 -7.18826234e-01 -1.99789286e-01
-5.54849803e-01 -6.68662965e-01 5.19319177e-01 5.83017349e-01
5.15104592e-01 -6.05394661e-01 -7.49488890e-01 2.38422662e-01
-5.66384681e-02 -1.64289975e+00 -3.73499006e-01 2.39843458e-01
-6.12314761e-01 -1.27358723e+00 -6.45728469e-01 -7.56533623e-01
9.01873887e-01 3.78083736e-01 1.00841701e+00 -2.80320913e-01
-2.69558847e-01 4.11662161e-01 -6.07293308e-01 -8.85542393e-01
-7.34858394e-01 2.26679772e-01 1.39964446e-01 1.39853761e-01
7.84381449e-01 -6.36376441e-02 -2.53708392e-01 7.47351706e-01
-8.79056513e-01 -4.64247704e-01 4.86705333e-01 7.48619199e-01
5.88834167e-01 2.04292342e-01 6.95444703e-01 -1.14826226e+00
4.21461016e-02 -6.16338909e-01 -1.10050941e+00 2.97945589e-01
-3.31413954e-01 -3.35133851e-01 4.89075065e-01 -8.19005013e-01
-1.22261035e+00 6.37538791e-01 3.39648426e-01 -3.22130293e-01
-9.02106345e-01 -2.72908241e-01 -3.89686435e-01 -2.19885424e-01
1.04318404e+00 1.59370914e-01 -1.53992608e-01 -4.85965371e-01
2.11050659e-01 9.89625573e-01 5.65274656e-01 -5.23875237e-01
9.88530755e-01 8.54667187e-01 1.90010127e-02 -9.69054341e-01
-7.42677987e-01 -9.33174551e-01 -9.21384215e-01 -8.18635821e-02
5.04939556e-01 -1.13819957e+00 -2.68065613e-02 2.96977907e-01
-1.14356136e+00 -2.64737964e-01 -3.57945114e-01 2.48021379e-01
-5.28487384e-01 2.80121118e-01 1.32895574e-01 -8.13328505e-01
1.99265644e-01 -1.01739097e+00 1.13787961e+00 1.31620660e-01
-4.00991626e-02 -8.23546112e-01 -1.10088907e-01 1.68290749e-01
1.80085987e-01 3.33572477e-01 4.43952173e-01 -1.30308402e+00
-5.20918548e-01 -7.16771543e-01 -3.51015031e-01 4.94348764e-01
1.91840395e-01 -2.76676029e-01 -1.15978110e+00 -3.99548799e-01
-3.11649621e-01 -7.50283375e-02 6.18947029e-01 2.10379437e-01
9.97565925e-01 -1.08191796e-01 -7.65595078e-01 3.13502640e-01
1.14583957e+00 2.79822528e-01 1.66859344e-01 4.25868511e-01
3.18216324e-01 8.37800324e-01 1.03872097e+00 5.98404884e-01
3.34415674e-01 7.62825966e-01 4.43674892e-01 -1.66826457e-01
5.76699451e-02 -6.11713640e-02 1.21401280e-01 -4.04796362e-01
3.93211544e-01 -2.02761844e-01 -1.05883288e+00 1.02354431e+00
-1.97649467e+00 -7.90987968e-01 2.52587115e-03 2.28833365e+00
3.50862056e-01 3.61132443e-01 4.76184368e-01 -1.12372123e-01
9.33515847e-01 -2.67663002e-01 -7.46394575e-01 -1.37149528e-01
4.94176475e-03 -3.65278423e-01 1.16720998e+00 1.70199320e-01
-1.56675017e+00 8.33881795e-01 6.48533058e+00 7.27038562e-01
-8.79835606e-01 4.30951476e-01 4.94207710e-01 -7.92596117e-02
6.14234567e-01 -3.95712465e-01 -1.18158102e+00 5.01214743e-01
9.00675237e-01 -6.96809441e-02 3.57583544e-04 1.42988336e+00
3.80134918e-02 -1.53559685e-01 -1.25315011e+00 6.71337664e-01
-8.99415165e-02 -8.62607121e-01 -4.53717798e-01 4.96606305e-02
8.53330374e-01 2.96875238e-01 1.57467037e-01 4.84330028e-01
5.46815932e-01 -5.62123179e-01 5.80020547e-01 -2.04932056e-02
8.73754442e-01 -6.60262227e-01 8.18197250e-01 6.92209661e-01
-1.09980953e+00 -3.21274221e-01 -5.89723051e-01 2.52727777e-01
-2.02382430e-02 3.78369659e-01 -1.46557558e+00 2.77485847e-01
9.55241323e-01 3.85132551e-01 -5.72294831e-01 1.20551574e+00
-5.93834259e-02 5.38900971e-01 -6.21189058e-01 2.56953299e-01
9.25050825e-02 2.62062371e-01 7.16763318e-01 1.44828057e+00
2.15282798e-01 8.06990042e-02 4.35680032e-01 5.18472314e-01
1.44716203e-02 -1.66933641e-01 -9.77774382e-01 5.30269980e-01
6.43819749e-01 1.09641039e+00 -5.89063048e-01 -4.78662163e-01
-6.62031233e-01 7.03365684e-01 1.69241056e-01 5.69223285e-01
-8.89067709e-01 -1.37820974e-01 9.62933183e-01 4.55718696e-01
6.24423265e-01 2.88796742e-02 -2.78262794e-01 -8.49926591e-01
-4.26432118e-02 -6.08571172e-01 5.77874243e-01 -2.93302208e-01
-1.35614860e+00 4.46265221e-01 4.31826562e-01 -1.46077836e+00
-3.44723046e-01 -9.02070761e-01 -2.98087656e-01 5.65092564e-01
-1.74476480e+00 -1.22938848e+00 -2.60602474e-01 6.35785758e-01
5.44146955e-01 -3.14123273e-01 5.49275935e-01 4.14095134e-01
-5.19515753e-01 7.53084123e-01 5.10526240e-01 3.65286052e-01
1.02217877e+00 -1.06717849e+00 4.68738973e-01 1.05313873e+00
-1.98518336e-01 1.53283328e-01 9.27802444e-01 -5.51246464e-01
-6.46030366e-01 -1.83962297e+00 6.21065199e-01 -7.90925443e-01
5.30962527e-01 -6.71591222e-01 -9.44553018e-01 8.53693664e-01
-3.62026572e-01 6.95233703e-01 4.29604232e-01 -2.06765369e-01
-4.82528597e-01 -2.94625014e-01 -1.65482163e+00 9.55768749e-02
9.02551770e-01 -2.13692024e-01 -5.35945535e-01 6.26958072e-01
5.05543590e-01 -7.00053453e-01 -4.76367354e-01 3.34872514e-01
2.79189974e-01 -5.40664852e-01 1.06911302e+00 -7.03664482e-01
-3.61474872e-01 -5.54524779e-01 -3.59558225e-01 -1.18543553e+00
-2.50294119e-01 -1.62650451e-01 -9.50302742e-03 1.25280643e+00
4.24408108e-01 -7.02006459e-01 1.06142795e+00 7.67594755e-01
1.90009952e-01 1.15091637e-01 -1.37628233e+00 -1.23137796e+00
9.53017399e-02 -4.22530204e-01 6.60755575e-01 6.89567983e-01
-4.48305368e-01 -5.01340553e-02 -3.77967745e-01 7.66437650e-01
7.70325303e-01 -1.58106834e-01 1.13101244e+00 -1.42011786e+00
-4.02724259e-02 1.04366280e-01 -4.45260704e-01 -1.02445030e+00
2.46171266e-01 -3.70765507e-01 4.15003181e-01 -8.73456359e-01
3.49939689e-02 -5.86155415e-01 -2.22035065e-01 3.44242662e-01
4.31799702e-02 1.67287126e-01 1.03063963e-01 5.15495948e-02
-8.01485837e-01 2.38913624e-03 7.25007653e-01 -1.97268993e-01
-1.52332887e-01 3.40941608e-01 -4.59424883e-01 8.36404383e-01
7.84518600e-01 -8.07084441e-01 -2.52065867e-01 -2.82868475e-01
-2.54221469e-01 -3.39259863e-01 5.43057680e-01 -1.08792317e+00
2.21939668e-01 -4.02871668e-01 3.85030478e-01 -5.61412692e-01
2.79730976e-01 -1.54535782e+00 -1.13190606e-01 2.34499723e-01
-1.20702006e-01 -4.53243852e-01 3.96289080e-01 9.11511123e-01
2.12487560e-02 -1.62075162e-01 1.16639841e+00 -7.73551613e-02
-1.12559068e+00 2.62042910e-01 -3.16088915e-01 8.00079033e-02
1.52395248e+00 -4.62686896e-01 -3.53449553e-01 -1.24672875e-01
-5.59533358e-01 4.39898849e-01 6.02618933e-01 5.25761724e-01
3.34750533e-01 -1.11086690e+00 -7.51264095e-01 3.42124045e-01
7.91113317e-01 3.92485738e-01 6.50902689e-02 3.77789974e-01
-8.65584314e-02 4.62544471e-01 1.14522241e-02 -8.33035588e-01
-1.43998063e+00 9.90687847e-01 3.65632832e-01 -1.68078169e-01
-2.79543042e-01 6.78294599e-01 6.23451889e-01 -6.06478512e-01
3.32671851e-01 -3.14433426e-01 -9.94943380e-02 -1.13173693e-01
6.51234984e-01 4.14277285e-01 1.33891940e-01 -9.35611367e-01
-5.02283454e-01 4.00870174e-01 -3.46030116e-01 1.99632436e-01
9.89747465e-01 -4.06842381e-01 6.02277279e-01 5.36642596e-03
9.16244864e-01 -4.30371553e-01 -1.65501809e+00 -6.23285711e-01
2.63584882e-01 -5.61748207e-01 -3.20887744e-01 -7.88250804e-01
-7.41420448e-01 6.29306793e-01 1.02344108e+00 8.15009028e-02
1.01512527e+00 2.24131241e-01 4.98213619e-01 5.42306006e-01
4.00470108e-01 -1.46798670e+00 -1.67262852e-01 4.42377359e-01
5.48038363e-01 -1.78146636e+00 -2.26698294e-01 -4.96099323e-01
-6.67556405e-01 8.53706539e-01 7.73594379e-01 -2.38486864e-02
7.17697978e-01 3.14562857e-01 2.54683733e-01 1.87275589e-01
-5.47488093e-01 -3.25085849e-01 2.21372634e-01 1.37572432e+00
-3.24922651e-01 2.53813535e-01 3.59839678e-01 4.52806801e-01
2.34545112e-01 -6.58587962e-02 5.02124965e-01 9.61455703e-01
-6.65568113e-01 -1.02063727e+00 -9.10488427e-01 3.12670767e-01
-3.27338457e-01 3.56604308e-01 -3.97265881e-01 1.10016370e+00
5.84606230e-01 1.17425144e+00 1.04027219e-01 -3.97182368e-02
7.00025797e-01 -1.38375303e-02 5.41612431e-02 -7.65637040e-01
-4.31000032e-02 -1.05997980e-01 1.13006771e-01 -3.24730545e-01
-3.50465596e-01 -9.69567597e-01 -9.40251946e-01 -4.48333621e-02
-3.59109312e-01 -1.10177524e-01 5.55105031e-01 9.32125092e-01
2.57275939e-01 2.85523713e-01 4.15252328e-01 -6.27183855e-01
-7.40274727e-01 -8.73113573e-01 -7.02709258e-01 5.49404919e-01
7.75731444e-01 -8.60533774e-01 -2.47618780e-01 3.46007735e-01] | [9.542756080627441, 1.7562556266784668] |
3a98fe54-8e06-494f-abfd-60dfd16f138e | the-p-destre-a-fully-annotated-dataset-for | 2004.02782 | null | https://arxiv.org/abs/2004.02782v1 | https://arxiv.org/pdf/2004.02782v1.pdf | The P-DESTRE: A Fully Annotated Dataset for Pedestrian Detection, Tracking, Re-Identification and Search from Aerial Devices | Over the last decades, the world has been witnessing growing threats to the security in urban spaces, which has augmented the relevance given to visual surveillance solutions able to detect, track and identify persons of interest in crowds. In particular, unmanned aerial vehicles (UAVs) are a potential tool for this kind of analysis, as they provide a cheap way for data collection, cover large and difficult-to-reach areas, while reducing human staff demands. In this context, all the available datasets are exclusively suitable for the pedestrian re-identification problem, in which the multi-camera views per ID are taken on a single day, and allows the use of clothing appearance features for identification purposes. Accordingly, the main contributions of this paper are two-fold: 1) we announce the UAV-based P-DESTRE dataset, which is the first of its kind to provide consistent ID annotations across multiple days, making it suitable for the extremely challenging problem of person search, i.e., where no clothing information can be reliably used. Apart this feature, the P-DESTRE annotations enable the research on UAV-based pedestrian detection, tracking, re-identification and soft biometric solutions; and 2) we compare the results attained by state-of-the-art pedestrian detection, tracking, reidentification and search techniques in well-known surveillance datasets, to the effectiveness obtained by the same techniques in the P-DESTRE data. Such comparison enables to identify the most problematic data degradation factors of UAV-based data for each task, and can be used as baselines for subsequent advances in this kind of technology. The dataset and the full details of the empirical evaluation carried out are freely available at http://p-destre.di.ubi.pt/. | ['Hugo Proença', 'S. V. Aruna Kumar', 'Abhijit Das', 'Ehsan Yaghoubi', 'B. S. Harish'] | 2020-04-06 | null | null | null | null | ['person-search'] | ['computer-vision'] | [-2.40951404e-01 -5.42965770e-01 7.09191263e-02 3.24360691e-02
-1.88318133e-01 -7.95242965e-01 9.07277882e-01 4.33250040e-01
-8.28401387e-01 8.05535316e-01 -3.71983111e-01 -5.02896011e-02
-1.03802122e-01 -8.57431591e-01 -3.83719206e-01 -8.27689409e-01
-1.51146770e-01 4.30898488e-01 5.22451639e-01 -3.60658109e-01
-1.58386618e-01 9.38209891e-01 -2.04388213e+00 -1.76637664e-01
5.49468040e-01 1.05192661e+00 2.58874506e-01 7.10111856e-01
5.15218735e-01 2.37416103e-01 -5.27569234e-01 -7.89085507e-01
3.47211689e-01 2.25378916e-01 -5.24109483e-01 1.37207463e-01
6.36332512e-01 -4.40911919e-01 -5.01198709e-01 9.47012722e-01
6.39506817e-01 1.88418508e-01 4.52703029e-01 -1.38869762e+00
-3.62557173e-01 -1.95294574e-01 -3.19067210e-01 4.21855569e-01
6.04360759e-01 2.45695651e-01 4.91276115e-01 -7.37000883e-01
5.34442127e-01 9.01993930e-01 8.23754072e-01 4.80423987e-01
-9.18534935e-01 -3.36434335e-01 1.16934348e-02 4.45754021e-01
-1.71297753e+00 -5.89199543e-01 4.69526589e-01 -7.06035316e-01
4.96161014e-01 6.50848389e-01 8.80540490e-01 1.24418283e+00
-2.96548516e-01 6.12398088e-01 1.03753650e+00 -4.41576302e-01
3.70962881e-02 5.60087144e-01 1.91700727e-01 6.43024504e-01
6.65481925e-01 2.32896775e-01 -2.37883195e-01 -9.04120728e-02
3.77444565e-01 7.06732571e-02 -2.71439254e-01 -4.00954694e-01
-1.08321977e+00 4.90314662e-01 3.09311539e-01 5.39368629e-01
-4.50644344e-01 -2.74698585e-01 4.49678123e-01 2.14490503e-01
4.62421268e-01 1.69939399e-01 -1.86531365e-01 -1.14359841e-01
-8.65213037e-01 4.49752688e-01 5.32231450e-01 9.33071911e-01
5.67131996e-01 -2.27434978e-01 -2.86340117e-01 5.14289677e-01
1.86718434e-01 8.99827003e-01 1.33854911e-01 -4.49575245e-01
4.03088242e-01 5.99010170e-01 6.96966469e-01 -1.44158208e+00
-3.84019911e-01 -2.31718004e-01 -8.21579814e-01 8.91550779e-02
7.18552411e-01 -1.03940353e-01 -4.62317795e-01 1.39936531e+00
5.96687615e-01 -9.54567790e-02 -5.55690266e-02 9.67759550e-01
8.19060266e-01 1.97264433e-01 7.93050751e-02 -7.39272460e-02
1.75215256e+00 -5.57255268e-01 -4.23022091e-01 -8.18294138e-02
2.28273481e-01 -8.39671910e-01 5.34214497e-01 3.15373391e-01
-5.97162783e-01 -8.02194238e-01 -8.21268082e-01 4.21961874e-01
-9.48238254e-01 6.10997021e-01 2.34331161e-01 1.08909571e+00
-1.20538390e+00 2.62342453e-01 -5.14786303e-01 -9.25728917e-01
2.97149777e-01 3.34650248e-01 -6.28628254e-01 -1.02134481e-01
-1.13834643e+00 1.00432134e+00 3.18651408e-01 3.09567064e-01
-7.43261516e-01 -2.02823877e-01 -5.71076810e-01 -2.84386158e-01
4.49169636e-01 -4.14763033e-01 7.14883149e-01 -7.04030752e-01
-8.27882588e-01 1.21741033e+00 -6.88265115e-02 -6.59109712e-01
9.20446992e-01 -7.12731779e-02 -6.53465033e-01 9.26420763e-02
2.26307750e-01 4.34895337e-01 9.62896228e-01 -1.26862204e+00
-7.52449751e-01 -4.91105586e-01 1.43584743e-01 -2.27062091e-01
-5.72588682e-01 3.72736752e-01 -3.66097122e-01 -5.88301957e-01
-5.48583031e-01 -1.11055696e+00 9.10302345e-03 -6.36305660e-02
-2.48334825e-01 -1.59888312e-01 8.85835350e-01 -8.56508493e-01
1.13091385e+00 -1.98064113e+00 1.24557219e-01 1.41605377e-01
2.42333755e-01 9.01913881e-01 1.64398208e-01 5.24346948e-01
2.75685579e-01 -7.32011124e-02 -3.99424471e-02 -4.95056838e-01
-7.26888627e-02 3.79375406e-02 1.32251373e-02 7.08319962e-01
-1.36685938e-01 7.07097650e-01 -8.92587602e-01 -4.73822057e-01
6.44123495e-01 4.58568633e-01 3.10751170e-01 1.15924709e-01
2.76262283e-01 5.23544908e-01 -2.81913877e-01 8.23793530e-01
7.45675266e-01 2.60504097e-01 -1.43763125e-01 -2.48817354e-01
-6.56398952e-01 -6.80568159e-01 -1.28249753e+00 1.05581415e+00
-1.08419642e-01 6.54173672e-01 2.41672054e-01 -8.10241461e-01
9.38006580e-01 3.95699352e-01 5.46336651e-01 -5.05551279e-01
3.49797457e-01 1.99844316e-01 -4.93484765e-01 -5.36558807e-01
8.49676192e-01 4.78757083e-01 -5.65559752e-02 -7.56053925e-02
-1.04553543e-01 3.26263100e-01 6.53307259e-01 -1.05229273e-01
7.45221019e-01 -2.87987441e-01 4.45988834e-01 -2.19728202e-01
1.02065551e+00 8.70403126e-02 1.97758421e-01 6.97005928e-01
-5.21480620e-01 2.51633704e-01 -8.28415528e-02 -8.25596452e-01
-1.03426218e+00 -9.22113001e-01 -2.82721579e-01 6.85551345e-01
2.69846350e-01 -7.53352344e-02 -9.30059254e-01 -5.96960008e-01
1.04332052e-01 2.04945356e-01 -5.69259405e-01 2.74395674e-01
-3.05122942e-01 -7.79006004e-01 8.26375961e-01 2.22606674e-01
7.86637783e-01 -6.33869708e-01 -8.71076703e-01 1.31358311e-01
-2.74162114e-01 -1.34898329e+00 -1.41484410e-01 -4.89908665e-01
-4.03843462e-01 -1.51811635e+00 -1.21099174e+00 -3.90922368e-01
5.81680059e-01 7.82470047e-01 8.81236076e-01 3.75133038e-01
-4.81034815e-01 1.01760316e+00 -5.93349099e-01 -4.66176718e-01
-1.98005423e-01 7.31276814e-03 5.80879569e-01 5.39538503e-01
4.88339126e-01 -1.63784638e-01 -4.52796042e-01 7.99348176e-01
-5.54612935e-01 -3.79054248e-01 2.35444963e-01 5.34704864e-01
3.47896695e-01 8.12188089e-02 2.92731822e-01 -2.37363160e-01
5.02748847e-01 -4.25631613e-01 -1.01818776e+00 4.22033221e-01
-3.87787342e-01 -6.38128877e-01 5.78204215e-01 -2.99585938e-01
-7.23552704e-01 1.54797569e-01 -1.35095557e-02 -2.46003598e-01
-6.54430807e-01 -1.46507323e-01 -1.00281633e-01 -6.06755435e-01
5.93648493e-01 2.32759044e-01 -1.55501157e-01 -4.69621450e-01
6.78141490e-02 9.45455551e-01 6.03854239e-01 -3.15825254e-01
9.62555945e-01 5.98499894e-01 1.10049404e-01 -1.39280653e+00
-3.26700807e-01 -8.91401172e-01 -1.08623993e+00 -6.95200920e-01
1.06943631e+00 -9.18416977e-01 -7.95330524e-01 9.44095790e-01
-1.24062943e+00 1.17282718e-01 -1.43286899e-01 2.58692920e-01
-1.24402970e-01 7.94956505e-01 -1.09237291e-01 -1.35424256e+00
-2.60034263e-01 -1.07211339e+00 1.01527059e+00 4.02832210e-01
4.54381183e-02 -8.03450942e-01 -1.05974421e-01 4.50677782e-01
3.46490353e-01 5.69583237e-01 2.62803268e-02 -4.11782324e-01
-4.95893300e-01 -6.99389160e-01 -2.91781217e-01 3.51464331e-01
5.60666397e-02 1.25858476e-02 -1.15022540e+00 -6.49941444e-01
-4.01081324e-01 1.80123314e-01 6.51950538e-01 1.68617457e-01
6.26775384e-01 -1.31890729e-01 -4.76603001e-01 1.42616376e-01
1.23217046e+00 -1.20390290e-02 3.85788560e-01 5.27536571e-01
4.91313308e-01 7.10677981e-01 8.90851140e-01 6.05857909e-01
6.16878390e-01 1.30928278e+00 5.76802671e-01 -9.43860337e-02
-1.62159309e-01 2.47091949e-02 1.91638961e-01 3.05259287e-01
-7.64420748e-01 -3.32070112e-01 -9.31262493e-01 7.10729063e-01
-2.02953124e+00 -1.06887448e+00 -4.43902522e-01 2.61405182e+00
-3.93884815e-02 -3.02852005e-01 7.27132678e-01 3.71018171e-01
9.98252630e-01 7.65659064e-02 7.23699033e-02 1.50209352e-01
-1.93605304e-01 -2.10259184e-01 8.53398860e-01 3.03779572e-01
-1.64138365e+00 5.88504910e-01 5.07261705e+00 6.74676120e-01
-7.21341431e-01 2.60576546e-01 1.89856589e-01 3.08358312e-01
5.69368362e-01 -4.51109141e-01 -9.73336995e-01 8.11650813e-01
7.05594182e-01 4.17498015e-02 5.02245963e-01 8.09595048e-01
3.42557609e-01 -3.61611724e-01 -7.06694126e-01 1.26510155e+00
1.32510439e-01 -9.71959949e-01 -3.20899338e-01 2.14115933e-01
3.56662512e-01 -2.10360140e-01 6.06931224e-02 -1.56869426e-01
-1.04365014e-01 -5.68243802e-01 8.76644015e-01 6.87304378e-01
7.62943029e-01 -7.19036937e-01 1.10166323e+00 4.20819372e-01
-1.67592549e+00 -2.66298860e-01 -5.62485397e-01 -8.78490042e-03
1.78047881e-01 4.23076391e-01 -6.88186705e-01 9.77892995e-01
9.41720724e-01 4.21300799e-01 -1.03418434e+00 1.40607905e+00
-8.16276670e-02 9.47445482e-02 -2.74258733e-01 -2.60148555e-01
7.46501014e-02 -1.90452468e-02 8.52944493e-01 1.31826246e+00
4.58533347e-01 -1.59333214e-01 3.50312382e-01 3.70236993e-01
1.90650806e-01 6.09979965e-02 -8.83800507e-01 3.42673004e-01
4.88036245e-01 1.33880818e+00 -7.48175681e-01 -1.64222330e-01
-4.31343466e-01 1.01299238e+00 5.82331456e-02 9.56893712e-02
-8.29905391e-01 -7.43909329e-02 6.98794842e-01 3.25378984e-01
2.49673009e-01 -4.84303147e-01 2.54755288e-01 -1.11029971e+00
1.53319791e-01 -6.86095834e-01 4.09332752e-01 -4.23746467e-01
-1.20262158e+00 7.02244878e-01 3.05707037e-01 -1.29224432e+00
-9.38142538e-02 -7.23588884e-01 -2.45558292e-01 7.73375332e-01
-1.46178973e+00 -1.51974535e+00 -8.15192878e-01 7.05347955e-01
2.60377765e-01 -4.00941730e-01 8.96654665e-01 6.39808774e-01
-5.78622818e-01 5.80933750e-01 1.76203903e-02 2.61844486e-01
5.31740069e-01 -8.98154259e-01 4.45422888e-01 1.23860860e+00
1.25978902e-01 2.31130168e-01 7.25944161e-01 -6.40537143e-01
-1.45782435e+00 -1.07240570e+00 8.07988167e-01 -8.23164463e-01
4.68896776e-01 -3.75034750e-01 -4.26991642e-01 3.20318282e-01
-1.90858230e-01 1.08022958e-01 4.08770591e-01 -2.22058490e-01
2.36524437e-02 -2.85917729e-01 -1.43216920e+00 2.70485848e-01
9.78607237e-01 -3.36690903e-01 -3.02332312e-01 4.28280354e-01
4.20767739e-02 -2.85953939e-01 -8.66873145e-01 1.34299453e-02
5.97441196e-01 -1.13767743e+00 1.27508593e+00 -2.10475668e-01
-4.08063740e-01 -5.54618716e-01 -2.42153063e-01 -9.50108290e-01
-2.73637414e-01 -4.07705992e-01 -5.62508684e-03 1.54589498e+00
-1.40022114e-02 -7.51911342e-01 6.19075060e-01 6.31647229e-01
2.20647126e-01 -2.34524384e-01 -1.38636875e+00 -1.05726647e+00
-6.10808909e-01 -2.47340128e-01 6.67207837e-01 6.64442360e-01
-6.29789114e-01 -3.06993365e-01 -8.04378569e-01 5.17323017e-01
9.55863059e-01 -2.64987350e-01 1.15720582e+00 -1.58540130e+00
9.58102476e-03 -1.46016210e-01 -9.33510959e-01 -6.33462667e-01
-1.41956016e-01 -4.15205091e-01 -2.98682600e-01 -1.20343971e+00
-2.42371317e-02 -5.85275471e-01 1.21197149e-01 2.77967066e-01
-5.98002411e-03 5.70911467e-01 6.21978223e-01 3.60151291e-01
-6.21572554e-01 1.64831400e-01 7.03447461e-01 -2.53273219e-01
7.82934129e-02 3.68368089e-01 -2.68383205e-01 6.28744304e-01
6.37963831e-01 -2.63117224e-01 1.97267905e-01 -2.71062016e-01
-1.36709571e-01 -3.23465824e-01 9.96458352e-01 -1.37083447e+00
4.15660620e-01 1.09371580e-01 4.79947299e-01 -4.67997164e-01
5.61416626e-01 -1.21338344e+00 2.67513692e-01 5.09716332e-01
5.52485645e-01 3.01751077e-01 2.76838601e-01 5.65059662e-01
-1.17543899e-01 -3.88524264e-01 6.96701288e-01 -2.57615536e-01
-1.11044085e+00 3.85343015e-01 -5.56386054e-01 -2.41923839e-01
1.48053408e+00 -4.49453682e-01 -2.81580418e-01 -1.29253626e-01
-5.22521675e-01 7.88687095e-02 7.49748290e-01 4.20019150e-01
3.90990019e-01 -1.19704092e+00 -7.84331262e-01 2.02914014e-01
2.35166267e-01 -4.03098494e-01 4.78907704e-01 8.98371100e-01
-3.67678761e-01 5.30142903e-01 -4.08369690e-01 -6.30130529e-01
-1.86609685e+00 8.26221168e-01 2.13533580e-01 -1.22232810e-01
-2.22257435e-01 5.44766366e-01 -4.32468653e-01 -1.58230558e-01
1.72703102e-01 2.05608457e-01 -4.81657654e-01 4.82104242e-01
7.96619356e-01 9.31606174e-01 2.10297540e-01 -1.25276613e+00
-6.64752662e-01 6.60506964e-01 2.94881284e-01 2.31025144e-01
1.08897531e+00 -3.72833848e-01 -9.64898989e-02 3.21355127e-02
6.89262390e-01 9.64803547e-02 -9.65657651e-01 4.90471274e-02
-5.24330735e-02 -7.99534440e-01 -1.12210512e-01 -5.12334883e-01
-7.56207347e-01 5.70410788e-01 1.06545210e+00 7.71208525e-01
1.04375911e+00 -1.37704119e-01 5.12781978e-01 3.00511479e-01
8.99619520e-01 -9.51248229e-01 -4.17509586e-01 2.25312769e-01
7.17670143e-01 -1.36665964e+00 1.17699407e-01 -4.46432710e-01
-3.56604934e-01 1.11807656e+00 2.26697564e-01 1.84727252e-01
2.22521454e-01 -5.56310592e-03 -1.55023530e-01 1.36267031e-02
1.15025908e-01 -6.64849281e-01 2.38520443e-01 1.09676206e+00
-8.79877061e-02 3.34903508e-01 -2.34440252e-01 3.97114426e-01
3.45677632e-04 1.10076983e-02 4.08329546e-01 7.47832060e-01
-3.09344441e-01 -1.24013376e+00 -9.85996068e-01 2.60112554e-01
-1.83308348e-01 2.25676239e-01 -5.98575115e-01 9.65149105e-01
4.02116776e-01 1.36274552e+00 -3.50038171e-01 -5.78450918e-01
7.12926447e-01 -2.62550741e-01 5.28277159e-01 -7.66235664e-02
-7.07164764e-01 -6.29450381e-01 3.35518062e-01 -2.76913404e-01
-7.17900038e-01 -9.67722178e-01 -2.35427856e-01 -7.17634499e-01
-3.42276096e-01 2.63442732e-02 6.71139359e-01 7.19977438e-01
1.98723182e-01 1.50899678e-01 5.68706214e-01 -1.27338243e+00
-3.40006411e-01 -6.58611476e-01 -5.21525979e-01 3.55577439e-01
2.90103048e-01 -1.08568609e+00 -1.05994090e-01 -1.70978773e-02] | [14.393697738647461, 1.0420860052108765] |
ca755666-a2eb-44ca-b144-fd378887f1ae | astrometric-calibration-and-source | 2211.09939 | null | https://arxiv.org/abs/2211.09939v1 | https://arxiv.org/pdf/2211.09939v1.pdf | Astrometric Calibration and Source Characterisation of the Latest Generation Neuromorphic Event-based Cameras for Space Imaging | As an emerging approach to space situational awareness and space imaging, the practical use of an event-based camera in space imaging for precise source analysis is still in its infancy. The nature of event-based space imaging and data collection needs to be further explored to develop more effective event-based space image systems and advance the capabilities of event-based tracking systems with improved target measurement models. Moreover, for event measurements to be meaningful, a framework must be investigated for event-based camera calibration to project events from pixel array coordinates in the image plane to coordinates in a target resident space object's reference frame. In this paper, the traditional techniques of conventional astronomy are reconsidered to properly utilise the event-based camera for space imaging and space situational awareness. This paper presents the techniques and systems used for calibrating an event-based camera for reliable and accurate measurement acquisition. These techniques are vital in building event-based space imaging systems capable of real-world space situational awareness tasks. By calibrating sources detected using the event-based camera, the spatio-temporal characteristics of detected sources or `event sources' can be related to the photometric characteristics of the underlying astrophysical objects. Finally, these characteristics are analysed to establish a foundation for principled processing and observing techniques which appropriately exploit the capabilities of the event-based camera. | ['Gregory Cohen', 'André van Schaik', 'Nicholas Tothill', 'Saeed Afshar', 'Alexandre Marcireau', 'Nicholas Owen Ralph'] | 2022-11-17 | null | null | null | null | ['camera-calibration', 'astronomy'] | ['computer-vision', 'miscellaneous'] | [ 5.75353026e-01 -7.62354672e-01 2.44890720e-01 -4.59014952e-01
-5.79129159e-01 -7.24227071e-01 1.07430053e+00 2.26017218e-02
-4.48781937e-01 5.92552423e-01 -2.41864711e-01 -4.28194135e-01
-6.01626992e-01 -7.00168073e-01 -3.04237753e-01 -8.54074478e-01
1.27964526e-01 2.99366504e-01 8.62203240e-01 -5.93898930e-02
5.47228098e-01 1.23380256e+00 -1.61702347e+00 -8.88865218e-02
3.59996587e-01 1.05044079e+00 4.53827530e-01 1.08339155e+00
-1.00037709e-01 9.21005309e-01 -7.65568197e-01 3.50228995e-01
3.43403876e-01 -8.57273757e-01 -1.28898891e-02 1.82745516e-01
3.11210245e-01 -1.93850935e-01 -3.86955738e-01 9.86440420e-01
2.16856539e-01 4.50571179e-01 4.33278263e-01 -9.81110215e-01
1.36954740e-01 -5.10927916e-01 -3.09048546e-03 1.06582904e+00
3.43173802e-01 7.62505829e-02 1.99625358e-01 -6.39225364e-01
5.27626336e-01 6.91171229e-01 2.10651442e-01 -3.27588767e-01
-8.86852801e-01 -3.30724746e-01 -5.50155878e-01 7.59871185e-01
-1.13981998e+00 -5.39575398e-01 8.24438035e-01 -5.82230449e-01
1.01208401e+00 6.67099833e-01 7.56606221e-01 7.83670485e-01
3.24397236e-01 -5.29659986e-02 1.26196408e+00 -9.08765376e-01
5.88473022e-01 3.09622467e-01 -1.45798288e-02 1.15974307e-01
2.74629414e-01 7.83508241e-01 -9.00913060e-01 -1.02305971e-02
1.25056767e+00 6.34558499e-02 -3.71401697e-01 -5.02876282e-01
-1.56450355e+00 8.63357842e-01 4.06219065e-01 7.24919081e-01
-5.92173100e-01 -8.89457762e-02 2.83894707e-02 7.22113475e-02
2.55753100e-01 7.14903951e-01 7.91053772e-02 -1.53078794e-01
-1.08638990e+00 1.81208864e-01 3.66954774e-01 6.32662237e-01
3.62548858e-01 3.48648071e-01 -4.61247601e-02 1.97237059e-01
1.89090058e-01 7.34665573e-01 3.04637134e-01 -8.15986335e-01
-1.23746417e-01 6.44417107e-01 2.16256291e-01 -8.46621215e-01
-2.80924261e-01 -5.68626583e-01 2.19986752e-01 6.73223197e-01
3.36747855e-01 3.14818412e-01 -6.00588262e-01 8.52906406e-01
6.42440379e-01 5.63051760e-01 9.65801701e-02 9.23807025e-01
3.02884668e-01 8.32079053e-01 -1.74001902e-01 -3.55733722e-01
1.32512963e+00 -2.89586097e-01 -5.51049471e-01 -4.24112797e-01
2.09065825e-01 -1.05723369e+00 4.63275045e-01 3.45056087e-01
-6.73837662e-01 -6.06166601e-01 -1.19462359e+00 4.54775691e-01
-3.35877895e-01 8.52175988e-03 6.05878174e-01 5.82943201e-01
-6.19740486e-01 5.66745028e-02 -9.80458379e-01 -7.57834077e-01
-3.17699790e-01 -2.78032590e-02 -1.02873430e-01 4.52645302e-01
-7.67351449e-01 1.14317286e+00 7.51391292e-01 -2.22023904e-01
-8.71572793e-01 -8.21410894e-01 -5.94006836e-01 -1.95418656e-01
1.89021155e-01 -3.94925445e-01 1.43656254e+00 -7.94929147e-01
-1.11629367e+00 8.09000134e-01 7.42085371e-03 -8.31929386e-01
1.70072570e-01 2.02179074e-01 -1.16306663e+00 7.52364278e-01
3.10986400e-01 -3.25105518e-01 7.74624884e-01 -8.52295399e-01
-1.05213022e+00 -4.01166618e-01 -2.54948825e-01 2.22916961e-01
1.62281543e-01 6.18950963e-01 -4.26286310e-02 -2.95822859e-01
5.02826273e-01 -6.94878340e-01 5.95884062e-02 1.76165909e-01
2.49484688e-01 3.75632167e-01 1.14295685e+00 -4.47714508e-01
7.28225112e-01 -2.36861062e+00 -2.54491389e-01 9.07556096e-05
-1.33775741e-01 3.19282174e-01 4.94291365e-01 7.07640886e-01
-3.63967210e-01 -1.07286406e+00 -1.58416033e-01 4.73685354e-01
-3.26177269e-01 4.74252366e-02 -7.07737803e-01 6.92416191e-01
-1.87098935e-01 4.22146171e-01 -7.33302891e-01 -2.02482969e-01
1.25029373e+00 3.11598748e-01 4.83616292e-02 3.71512115e-01
2.03544851e-02 9.58168805e-01 -4.34004694e-01 5.74957192e-01
2.66179740e-01 9.01584402e-02 -2.58259296e-01 -2.12881476e-01
-1.03647768e+00 1.91687822e-01 -1.17305887e+00 1.34560430e+00
-2.90578812e-01 9.31924284e-01 1.84687465e-01 -9.81755435e-01
1.14850163e+00 3.25131685e-01 5.22542536e-01 -1.16940248e+00
-9.12775751e-03 3.54314029e-01 -2.14726105e-01 -5.75286627e-01
8.04309547e-01 -6.30691171e-01 2.34039024e-01 2.64893860e-01
5.53028919e-02 -3.27384263e-01 8.33749026e-02 2.45539829e-01
1.03398263e+00 -1.07574143e-01 4.07063782e-01 -2.06166476e-01
6.89816535e-01 5.36318541e-01 1.37344420e-01 5.53654909e-01
-2.09265038e-01 3.90859127e-01 -4.52320904e-01 -4.73680019e-01
-1.17706501e+00 -1.36580300e+00 -7.21867442e-01 6.72587216e-01
2.88126141e-01 -8.93355384e-02 -2.15012088e-01 -1.43433362e-01
-2.68586487e-01 1.31932998e+00 -7.62833059e-02 -2.11918145e-01
-2.80562669e-01 -9.66182470e-01 1.67340226e-02 2.63745248e-01
3.93247902e-01 -7.75874972e-01 -1.38365543e+00 5.65712988e-01
3.19774784e-02 -8.38786244e-01 4.52614218e-01 2.68838465e-01
-7.54132271e-01 -1.28199506e+00 -2.47734010e-01 2.56707259e-02
3.16795319e-01 8.12831998e-01 6.95327044e-01 -4.42928165e-01
-5.65533698e-01 1.05956697e+00 -5.17481863e-01 -9.14501846e-01
-2.95178145e-01 -8.25031579e-01 5.31781837e-02 3.05426866e-01
4.05694246e-01 -4.79058564e-01 -5.84957242e-01 4.76673454e-01
-1.20260465e+00 -3.16687435e-01 3.13598752e-01 2.84418136e-01
4.36037689e-01 2.89729804e-01 2.02083915e-01 -1.06119767e-01
-1.59573197e-01 -3.86553019e-01 -1.29226148e+00 3.07423413e-01
-2.17362523e-01 -3.75381857e-01 1.26712397e-01 -1.48417294e-01
-1.59920013e+00 -4.60671820e-03 3.02183151e-01 -2.23061830e-01
-3.68074685e-01 1.22075938e-01 1.65668756e-01 -6.84342086e-01
1.29510665e+00 4.83197480e-01 -1.93611100e-01 -1.06634505e-01
-9.93757546e-02 5.88766038e-01 1.03649378e+00 -2.47226089e-01
8.59814942e-01 1.08152056e+00 2.26875708e-01 -1.33345854e+00
-8.46930623e-01 -1.22634375e+00 -5.80964029e-01 -6.78417683e-01
1.00184011e+00 -8.86845350e-01 -2.36193184e-03 1.28237098e-01
-1.00230885e+00 2.22869560e-01 -6.12201154e-01 1.18518925e+00
-5.88570237e-01 2.34990671e-01 2.79338568e-01 -1.30758214e+00
2.50730664e-01 -7.55929768e-01 1.18540514e+00 5.36610007e-01
-6.30240887e-02 -1.20478296e+00 5.27834415e-01 3.15260708e-01
5.93574464e-01 4.23798472e-01 1.62200034e-01 -5.12733877e-01
-1.07864022e+00 -5.86367309e-01 -2.09875315e-01 -1.21882604e-02
1.07745588e-01 -1.35955140e-01 -9.74087358e-01 -1.43733501e-01
1.19436300e+00 3.27882558e-01 3.39812875e-01 5.99455833e-01
5.65377653e-01 4.75909799e-01 -3.43584865e-01 9.43951905e-01
1.81063759e+00 9.80964661e-01 6.71135247e-01 8.65469992e-01
2.10738853e-01 4.69996631e-01 9.23984528e-01 5.13755918e-01
-2.57405877e-01 7.54937172e-01 3.58174920e-01 1.61798596e-01
-6.55356376e-03 2.69917697e-01 1.76713273e-01 2.50112683e-01
-2.99569108e-02 1.54961407e-01 -6.97546184e-01 5.30048847e-01
-1.52036321e+00 -1.34147859e+00 -7.38079965e-01 2.55401969e+00
8.07845518e-02 6.00289442e-02 -2.01645434e-01 1.95017904e-01
8.11536908e-01 2.53195554e-01 -3.21162581e-01 6.71540350e-02
-2.08959609e-01 2.56933242e-01 7.68240988e-01 2.55017757e-01
-8.83541942e-01 2.86331415e-01 6.33042955e+00 3.12348872e-01
-1.32385349e+00 7.77505711e-02 -2.63691783e-01 -2.78737068e-01
-1.49275571e-01 4.70614582e-01 -9.61480916e-01 4.85249072e-01
1.38028467e+00 -5.08720160e-01 3.17489207e-01 6.29313171e-01
3.99328619e-01 -7.85696328e-01 -7.65787601e-01 1.28257346e+00
2.79173940e-01 -1.41367412e+00 -4.88701701e-01 -5.88105097e-02
4.06914473e-01 6.18780740e-02 -4.43583935e-01 -1.99318588e-01
-2.55935103e-01 -2.13443890e-01 1.03700459e+00 8.52731943e-01
4.42916006e-01 -4.90916193e-01 1.21425010e-01 2.80586034e-01
-1.16822910e+00 -5.33133186e-03 -3.70820791e-01 -2.05638811e-01
9.46567059e-01 8.09970140e-01 -1.34784126e+00 7.73509145e-01
6.77667379e-01 3.22201431e-01 -7.92806506e-01 1.49912417e+00
6.28735945e-02 6.53553843e-01 -5.85871756e-01 1.98938385e-01
1.79856285e-01 -4.57459837e-01 1.20072341e+00 7.68470168e-01
8.70766163e-01 3.00718606e-01 -2.29967102e-01 8.75614285e-01
9.28418875e-01 -5.23391485e-01 -8.74993920e-01 -1.69356629e-01
4.29904670e-01 1.30137622e+00 -8.37290645e-01 -4.48793530e-01
-5.19826651e-01 7.59996235e-01 -5.66803694e-01 1.93601266e-01
-8.28644633e-01 -2.92022020e-01 1.45298317e-01 3.92589539e-01
2.96961397e-01 -7.83827305e-01 -1.66986465e-01 -8.75782549e-01
2.49118079e-03 -3.16949725e-01 5.74093997e-01 -1.40930271e+00
-9.44879532e-01 4.17907774e-01 4.64872181e-01 -1.43618631e+00
-3.53197396e-01 -7.19763100e-01 -8.40389073e-01 9.29032743e-01
-1.26446068e+00 -1.05596399e+00 -5.31862080e-01 6.79414093e-01
2.90979236e-01 -3.89912844e-01 5.98258913e-01 4.62967716e-02
4.34992462e-03 -6.28710389e-01 3.11847538e-01 -5.23097098e-01
6.14620507e-01 -1.17660403e+00 9.15747881e-02 1.42224240e+00
4.54797357e-01 1.47874177e-01 1.01543295e+00 -1.00171518e+00
-1.46152925e+00 -8.93509805e-01 6.17995560e-01 -7.81915605e-01
1.06537473e+00 -4.51852828e-02 -8.77998412e-01 7.09592760e-01
6.65237894e-03 -3.49345617e-02 6.44807398e-01 -3.32633972e-01
2.28942335e-01 -2.85838246e-01 -9.15346086e-01 2.59418011e-01
7.04575002e-01 -1.01977789e+00 -1.05609596e+00 6.98741198e-01
2.80544072e-01 -3.76304239e-01 -5.09310424e-01 2.55394399e-01
3.13672349e-02 -1.27310610e+00 1.34660959e+00 -1.11475244e-01
-3.55049640e-01 -8.03566635e-01 -1.57520652e-01 -7.80981600e-01
-3.49596828e-01 -3.72173339e-01 2.73925781e-01 9.48166132e-01
-2.86376327e-01 -7.62769341e-01 5.03642559e-01 2.46879295e-01
-4.01380390e-01 5.55551529e-01 -1.17538452e+00 -1.13465202e+00
-8.67415309e-01 -6.54571533e-01 1.96300611e-01 7.89557755e-01
-5.50409079e-01 3.04716080e-02 1.49921924e-01 8.84464800e-01
9.94748771e-01 1.78172469e-01 4.94672269e-01 -1.15230370e+00
-4.34986621e-01 6.77947775e-02 -6.56869888e-01 -6.32965611e-03
-5.23092031e-01 -5.99035680e-01 -6.70383126e-02 -1.26778996e+00
-3.37407500e-01 -2.07888097e-01 -2.54550576e-03 -5.77538908e-01
1.47631109e-01 2.15519555e-02 -7.17408732e-02 5.14017582e-01
-3.16426873e-01 2.60065049e-01 6.51530802e-01 4.64858055e-01
-6.79655606e-03 -1.83877386e-02 6.64270446e-02 6.62925363e-01
5.50605774e-01 -2.97949016e-01 -4.79947835e-01 -2.22465217e-01
2.66881406e-01 4.52991128e-01 1.23591614e+00 -1.67991281e+00
9.25633609e-01 -2.76854724e-01 5.43808043e-01 -9.87358332e-01
5.65683365e-01 -9.63602543e-01 8.91109049e-01 3.09025198e-01
1.12635933e-01 -1.60606518e-01 1.50906816e-01 6.10669494e-01
-4.05796945e-01 -5.84271789e-01 1.01992738e+00 -3.30772191e-01
-1.29856849e+00 -9.35281888e-02 -5.04727125e-01 -5.18166602e-01
1.67131078e+00 -5.20852089e-01 -2.39781871e-01 3.46864574e-03
-4.84635144e-01 -5.38927555e-01 6.13511622e-01 -4.04761471e-02
5.08805692e-01 -1.31270289e+00 -3.21477443e-01 4.26622808e-01
4.40746099e-01 -1.84188813e-01 5.03015339e-01 8.02135289e-01
-9.30168748e-01 8.45342517e-01 -3.76475096e-01 -9.67941344e-01
-1.08755493e+00 6.69818819e-01 5.43074608e-01 4.04545903e-01
-6.31153524e-01 7.75416195e-01 2.83696562e-01 -7.06295716e-03
-3.12929332e-01 1.43796466e-02 -2.99892034e-02 -2.13962525e-01
1.07326496e+00 4.46806133e-01 3.34654361e-01 -6.84733987e-01
-3.05878639e-01 2.73227185e-01 4.02697682e-01 -5.73843539e-01
1.24963236e+00 -2.47005299e-01 -7.01256767e-02 8.11028063e-01
5.89986503e-01 -8.42276067e-02 -1.32969570e+00 -2.98907071e-01
1.30873993e-01 -9.37368989e-01 5.49735725e-01 -8.29865515e-01
-4.14511174e-01 6.36403441e-01 8.38926256e-01 5.66549420e-01
1.31248176e+00 1.50134832e-01 1.89414755e-01 4.51582000e-02
5.58216929e-01 -1.02480721e+00 -2.30547830e-01 3.18365656e-02
1.11112475e+00 -9.24195588e-01 1.96699187e-01 -4.07644242e-01
-2.71805316e-01 1.26975632e+00 2.88082045e-02 -3.53795402e-02
3.47270131e-01 1.66298494e-01 -1.40190125e-01 -7.21475363e-01
-2.74474323e-01 -2.31267080e-01 4.06152010e-01 6.33360505e-01
-5.39124757e-02 6.17580786e-02 -1.98182896e-01 -1.44797847e-01
-1.29427180e-01 -1.30636886e-01 4.48334187e-01 1.27423561e+00
-8.77156019e-01 -9.44406033e-01 -1.45823991e+00 2.17809647e-01
4.34138440e-02 3.24986726e-01 -3.16542894e-01 6.83072805e-01
6.46362975e-02 7.74889767e-01 4.54638600e-01 3.14191818e-01
4.80909497e-01 6.82200342e-02 6.58192515e-01 -5.70012212e-01
-3.08688253e-01 8.27701390e-02 2.64384419e-01 -4.58509266e-01
-7.70347178e-01 -1.07186961e+00 -1.03144419e+00 7.67760351e-02
-3.10869575e-01 7.29012340e-02 1.42622972e+00 9.94823515e-01
-8.44775495e-05 5.99848628e-01 5.60255408e-01 -8.18997681e-01
-2.15064853e-01 -5.47998071e-01 -7.18647182e-01 -5.49548045e-02
1.07214600e-01 -7.78024673e-01 -6.62487328e-01 2.84471571e-01] | [10.925222396850586, -2.5027825832366943] |
b05d9680-27d5-4d2e-a9a8-71af3026aa43 | federated-variational-inference-methods-for | 2302.03314 | null | https://arxiv.org/abs/2302.03314v2 | https://arxiv.org/pdf/2302.03314v2.pdf | Federated Variational Inference Methods for Structured Latent Variable Models | Federated learning methods enable model training across distributed data sources without data leaving their original locations and have gained increasing interest in various fields. However, existing approaches are limited, excluding many structured probabilistic models. We present a general and elegant solution based on structured variational inference, widely used in Bayesian machine learning, adapted for the federated setting. Additionally, we provide a communication-efficient variant analogous to the canonical FedAvg algorithm. The proposed algorithms' effectiveness is demonstrated, and their performance is compared with hierarchical Bayesian neural networks and topic models. | ['Kerrie Mengersen', 'Robert Salomone', 'Conor Hassan'] | 2023-02-07 | null | null | null | null | ['topic-models'] | ['natural-language-processing'] | [-2.69485235e-01 2.16799095e-01 -4.23624545e-01 -5.22369623e-01
-1.12812567e+00 -2.85837352e-01 7.20639110e-01 4.02250923e-02
-1.80161372e-01 9.93203044e-01 1.03492305e-01 -1.94810748e-01
-7.20123827e-01 -8.70591760e-01 -7.35943556e-01 -9.73126292e-01
-2.42076844e-01 9.06325281e-01 2.13321492e-01 5.56394577e-01
-1.09678313e-01 2.30522022e-01 -1.31363750e+00 -2.02410892e-01
6.94765091e-01 9.31999266e-01 3.40449922e-02 4.27205265e-01
-3.95830750e-01 7.86234677e-01 -3.93717825e-01 -7.75176585e-01
-7.18103051e-02 6.83642328e-02 -7.06772685e-01 -5.02295911e-01
3.04178834e-01 -5.74311376e-01 -5.84338307e-01 1.09363508e+00
2.24146277e-01 3.83302182e-01 7.37048388e-01 -1.29655325e+00
-3.95513535e-01 1.01056135e+00 -3.56932670e-01 -9.69744548e-02
-7.33253881e-02 -6.35735929e-01 1.00275338e+00 -8.56227338e-01
4.71014291e-01 1.44085467e+00 6.99418187e-01 4.25741613e-01
-1.19323027e+00 -6.33731008e-01 1.55037761e-01 5.07094085e-01
-1.45974672e+00 -5.12254000e-01 6.50661051e-01 -1.82065427e-01
6.13391161e-01 9.28776115e-02 1.77905589e-01 1.20501173e+00
-9.44257602e-02 9.87098813e-01 1.00119948e+00 -4.38878387e-01
8.89570534e-01 1.57240301e-01 3.08373481e-01 6.36843145e-01
4.92217302e-01 1.19319789e-01 -1.10558450e+00 -9.92632151e-01
3.13271195e-01 4.73948926e-01 -1.45920575e-01 -6.87798679e-01
-8.27890098e-01 1.14620054e+00 2.72263497e-01 9.46184769e-02
-4.59566057e-01 4.56661493e-01 1.85601905e-01 -1.53065324e-01
7.63976336e-01 -5.90670705e-01 -6.66263640e-01 -1.54615015e-01
-1.30144370e+00 3.09864759e-01 1.26095378e+00 1.07077861e+00
8.77523184e-01 -5.87946400e-02 -1.14636369e-01 7.89521337e-01
8.98511052e-01 7.63549805e-01 2.87309557e-01 -1.18001211e+00
1.21402830e-01 1.09173149e-01 3.27228397e-01 -7.83194244e-01
-6.70201257e-02 -1.12995066e-01 -9.69167352e-01 -3.18092734e-01
3.21795225e-01 -2.06182227e-01 -3.83578062e-01 1.73314559e+00
8.10443461e-01 5.86615801e-01 7.82003254e-02 5.73782861e-01
4.85569626e-01 7.84197152e-01 -5.70960063e-03 -2.25199133e-01
1.12013149e+00 -7.47792423e-01 -9.17323768e-01 2.91590363e-01
2.28256404e-01 -3.46689850e-01 3.27059627e-01 8.35544884e-01
-9.27050471e-01 7.14248717e-02 -6.76970065e-01 9.02830437e-02
-5.45605421e-01 -3.47930968e-01 1.02594841e+00 1.15096188e+00
-1.12192237e+00 6.26192331e-01 -1.21883416e+00 -2.71806717e-01
6.96839809e-01 2.46197909e-01 -4.37534899e-02 -4.29028541e-01
-1.12176156e+00 4.94979978e-01 2.90197700e-01 3.85738388e-02
-1.35900545e+00 -6.95624530e-01 -6.09430015e-01 2.33413339e-01
3.36614013e-01 -8.98885787e-01 1.63524151e+00 1.44432351e-01
-1.70725143e+00 9.18120667e-02 -5.02667129e-01 -8.27997863e-01
4.79172766e-01 -2.93517023e-01 -2.60101676e-01 9.32846442e-02
-2.31891096e-01 1.13764420e-01 8.91153216e-01 -1.08995652e+00
-7.71697879e-01 -6.96825087e-01 -3.41715425e-01 -3.89111638e-01
-7.94578254e-01 8.18576142e-02 -5.39138258e-01 -4.38458204e-01
2.72248127e-02 -4.87821728e-01 -4.01764899e-01 5.87675087e-02
-4.65654016e-01 -6.05659783e-01 1.08867288e+00 -6.38757765e-01
1.13702667e+00 -1.82153916e+00 3.75789553e-02 4.32242453e-01
2.94158518e-01 -1.85481355e-01 3.66637170e-01 5.32757044e-01
8.06631386e-01 -5.28904982e-02 -3.53934288e-01 -8.23547781e-01
6.08789563e-01 3.95985276e-01 -5.34618139e-01 6.34406328e-01
-6.20432019e-01 5.18483639e-01 -7.99948394e-01 -7.74256229e-01
1.14240117e-01 7.82840192e-01 -5.44060647e-01 2.59782284e-01
-5.15552163e-01 1.58835247e-01 -6.68618679e-01 6.39259279e-01
7.65172839e-01 -3.97217184e-01 6.05653048e-01 1.17086433e-02
1.94647580e-01 3.36529426e-02 -1.39072728e+00 1.89529335e+00
-6.02398336e-01 2.32627317e-01 6.68294370e-01 -1.05663824e+00
8.52568865e-01 6.38272882e-01 5.34481287e-01 6.41423389e-02
-1.06408089e-01 9.69386697e-02 -7.73125350e-01 -1.45191565e-01
3.85861218e-01 -3.31111364e-02 -7.50142932e-02 8.26929510e-01
3.89889717e-01 1.67238653e-01 -2.85362098e-02 5.96991956e-01
9.79824185e-01 1.84648097e-01 1.28290847e-01 -1.97854877e-01
5.60496561e-02 -1.37404397e-01 6.00104153e-01 1.31029737e+00
-8.60839561e-02 1.48901671e-01 9.81977284e-02 -3.66952926e-01
-5.68613052e-01 -1.35120726e+00 -4.88860160e-01 1.36151230e+00
1.63069125e-02 -7.25248992e-01 -7.91891038e-01 -8.00197899e-01
1.12324364e-01 7.58290410e-01 -4.41593051e-01 2.18790606e-01
-3.65826078e-02 -8.29137504e-01 5.37338912e-01 2.88494110e-01
4.88546908e-01 -4.33094949e-01 -5.57464540e-01 3.12125742e-01
-2.64046937e-01 -7.77677894e-01 3.36748883e-02 1.11910962e-01
-1.22078419e+00 -9.58311319e-01 -6.55269206e-01 -7.57108815e-03
2.30033264e-01 2.33617082e-01 9.82353091e-01 -4.52876538e-01
-2.17832491e-01 7.58510709e-01 -7.56649002e-02 -6.16738498e-01
-1.68144360e-01 7.91573152e-02 3.21324885e-01 2.05066323e-01
4.99799788e-01 -6.65664256e-01 -4.12416965e-01 1.93281874e-01
-8.08879733e-01 -4.21486616e-01 3.13922465e-01 9.22538757e-01
4.68341827e-01 4.82890531e-02 5.83136261e-01 -1.13425851e+00
2.74449587e-01 -9.30713594e-01 -9.57832694e-01 4.32909548e-01
-8.07347357e-01 1.44453809e-01 3.72077376e-01 2.12004054e-02
-1.66279900e+00 -2.26156622e-01 5.24365902e-02 -5.29508829e-01
-4.82327878e-01 5.73844492e-01 -2.55615890e-01 1.01339184e-01
4.00878012e-01 2.98328754e-02 -1.72991768e-01 -1.15073204e+00
5.66112697e-01 8.64684224e-01 3.77298623e-01 -1.01552534e+00
5.73712289e-01 9.39664483e-01 -1.20787799e-01 -7.30663300e-01
-8.17220330e-01 -3.71206999e-01 -4.05074716e-01 9.38006565e-02
3.92087698e-01 -8.43485534e-01 -6.88690722e-01 3.43014121e-01
-1.30462039e+00 -1.25971332e-01 -2.48705328e-01 7.41836011e-01
-5.64735293e-01 2.91300327e-01 -5.72299123e-01 -1.28725767e+00
-4.71464157e-01 -6.54519975e-01 8.60413849e-01 1.42581195e-01
2.25686148e-01 -1.29823053e+00 5.69421232e-01 3.23069751e-01
6.63139462e-01 -1.40667498e-01 8.94601464e-01 -1.03059816e+00
-7.75540054e-01 -2.92995095e-01 -1.76477924e-01 -6.77746758e-02
-5.12039103e-02 8.40623048e-04 -1.25248516e+00 -2.38061234e-01
-1.83136798e-02 -2.61500150e-01 8.15728903e-01 5.99897623e-01
1.12766838e+00 -5.03429353e-01 -9.00071025e-01 5.85640192e-01
1.48920870e+00 -3.85953724e-01 -5.73760122e-02 -1.75606355e-01
3.97528470e-01 4.91939157e-01 3.27156708e-02 8.10250878e-01
4.59626943e-01 5.27617216e-01 4.23570991e-01 2.93715805e-01
3.08255613e-01 -3.47985029e-01 9.60693657e-02 7.42086351e-01
1.62841722e-01 -9.83137935e-02 -7.90919781e-01 7.00277686e-01
-2.19324946e+00 -8.90724599e-01 -1.04998834e-01 2.31055021e+00
7.22267866e-01 -4.04167533e-01 6.61043078e-02 -2.17717364e-01
8.02916050e-01 1.03557266e-01 -6.02270484e-01 -1.35089666e-01
-1.12888642e-01 2.41786376e-01 3.67616236e-01 2.39320204e-01
-9.40544963e-01 6.59182787e-01 7.52854013e+00 1.03598535e+00
-3.07472914e-01 1.04093134e+00 3.84946465e-01 -2.94425458e-01
-3.11965227e-01 1.83878437e-01 -9.32374179e-01 3.87280017e-01
1.48584378e+00 -2.31428802e-01 3.73691559e-01 1.22362542e+00
-4.85876463e-02 -7.22299004e-03 -1.02230322e+00 9.53585029e-01
-1.28383994e-01 -1.46227443e+00 -1.09874934e-01 1.97099090e-01
8.35593045e-01 5.37555635e-01 -1.30562663e-01 2.82578051e-01
1.12838614e+00 -7.40915239e-01 4.99812365e-01 7.82111883e-01
2.33015195e-01 -7.76066661e-01 4.86664414e-01 5.28688371e-01
-8.40411782e-01 -2.51280636e-01 -6.99355185e-01 1.50823206e-01
3.71108204e-01 9.54074740e-01 -4.20309752e-01 8.21647227e-01
1.21267855e+00 4.82510239e-01 -1.23727284e-01 1.26398778e+00
-1.20083064e-01 1.06896532e+00 -7.67350256e-01 -2.24603236e-01
1.29939942e-03 -2.04663336e-01 5.31787038e-01 9.92640018e-01
4.62693363e-01 -3.20403695e-01 4.12411615e-03 9.67900395e-01
-2.10364476e-01 8.87622163e-02 -6.79008424e-01 3.93028110e-01
8.73187304e-01 1.13900578e+00 -2.76444942e-01 -5.26177824e-01
-5.42356193e-01 4.60672200e-01 4.05371547e-01 4.66303945e-01
-7.19584882e-01 -1.00257143e-01 5.66100240e-01 -4.59473401e-01
5.57351768e-01 -3.58598642e-02 4.95413691e-02 -1.28568947e+00
-3.21867347e-01 -3.94920200e-01 9.59727466e-01 -4.58515942e-01
-1.73968709e+00 3.43225688e-01 6.09533668e-01 -5.69019437e-01
-5.32063425e-01 -4.02582139e-01 -4.04869229e-01 6.00130022e-01
-1.33849049e+00 -1.23906279e+00 -4.67132777e-02 1.08473945e+00
1.08282916e-01 -2.69404620e-01 1.20399916e+00 2.48676851e-01
-6.89004958e-01 5.48468888e-01 1.09222698e+00 -9.49868634e-02
6.14763916e-01 -1.12514281e+00 -1.24006569e-01 8.30032229e-01
5.83538234e-01 8.96392703e-01 6.35555744e-01 -3.63833427e-01
-1.71834362e+00 -1.20090306e+00 7.71728396e-01 -2.98702568e-01
6.99039042e-01 -4.18567449e-01 -7.25754797e-01 8.04464877e-01
4.19727534e-01 5.71361072e-02 8.94028187e-01 6.42902911e-01
-6.92635596e-01 -4.27739322e-01 -1.51237857e+00 1.76366210e-01
6.42450273e-01 -3.07393402e-01 -4.76958513e-01 5.78588784e-01
6.04285896e-01 7.59245381e-02 -8.88016164e-01 -3.06695681e-02
4.90212470e-01 -8.48615408e-01 7.76164711e-01 -7.57083237e-01
-2.70897120e-01 -3.55446786e-02 -5.28215706e-01 -1.09047258e+00
-1.13816753e-01 -1.03217900e+00 -1.20231616e+00 1.31588662e+00
1.06116474e-01 -8.70909631e-01 1.05591524e+00 9.80856061e-01
3.01648766e-01 -2.31059641e-01 -1.61982727e+00 -6.34623706e-01
1.90385178e-01 -7.89764524e-01 8.46294940e-01 7.02846229e-01
1.08040139e-01 -1.62617147e-01 -6.10103190e-01 3.18906069e-01
1.52967513e+00 2.78103322e-01 7.45748878e-01 -1.62027657e+00
-4.95541185e-01 -1.59110680e-01 -3.73600163e-02 -1.11301374e+00
4.03055698e-01 -6.76382482e-01 -1.36602588e-03 -1.42340553e+00
4.23886836e-01 -5.98959029e-01 -4.93262112e-01 5.41488767e-01
1.72290206e-01 1.24542592e-02 -7.87148848e-02 3.32318902e-01
-9.46151733e-01 7.68980086e-01 1.44604534e-01 -3.42854530e-01
2.05239192e-01 2.31398374e-01 -4.54225719e-01 6.53419137e-01
5.94123423e-01 -7.90710568e-01 -3.11284751e-01 -5.09285212e-01
5.53176552e-02 8.39354470e-02 6.26929820e-01 -7.88722873e-01
7.38241851e-01 -2.71356087e-02 1.46974474e-01 -8.57825518e-01
4.38275367e-01 -1.03720224e+00 3.83904248e-01 1.68763101e-01
-1.10474065e-01 -4.55298662e-01 -1.70485243e-01 1.28013718e+00
-1.24160022e-01 -4.21642423e-01 4.96566266e-01 -1.00183278e-01
-3.56110901e-01 5.74132562e-01 -4.22208637e-01 -2.49256164e-01
7.96881497e-01 2.82287598e-01 -2.40481272e-01 -4.82561648e-01
-8.80822241e-01 7.01818541e-02 9.47318748e-02 -5.49367666e-02
6.22641385e-01 -1.03057694e+00 -6.19940162e-01 -1.23140626e-01
-1.45780995e-01 6.99410737e-02 5.63957214e-01 8.59543741e-01
1.26389280e-01 7.31594503e-01 6.20047748e-01 -7.50817120e-01
-5.99751890e-01 6.39841080e-01 1.30491510e-01 -1.71755582e-01
-6.59929633e-01 7.37269819e-01 -1.80053249e-01 -7.40292728e-01
6.74808383e-01 6.14444576e-02 2.91252673e-01 -2.21273810e-01
7.45785296e-01 8.11616004e-01 1.82314858e-01 -6.65680394e-02
-3.25593174e-01 -1.25140592e-01 3.04388069e-02 -3.09146821e-01
1.46709335e+00 -2.72969693e-01 -4.14151430e-01 7.00076044e-01
9.91280258e-01 -2.26482749e-01 -1.24407935e+00 -8.74536753e-01
1.71989545e-01 -4.49817240e-01 4.57513034e-01 -6.03214681e-01
-1.01180112e+00 8.76480818e-01 6.24045312e-01 3.12984139e-01
7.34983206e-01 2.56507963e-01 6.62152469e-01 7.45700955e-01
1.03489316e+00 -8.11214745e-01 -5.63347101e-01 2.66869009e-01
2.00529560e-01 -1.05233526e+00 -4.49586809e-02 1.34467529e-02
8.57960209e-02 9.83277261e-01 1.02370538e-01 2.22808480e-01
1.28011751e+00 3.85095239e-01 -2.69513369e-01 -1.84589297e-01
-1.07625425e+00 2.96706587e-01 -2.35681161e-01 8.95647764e-01
4.38798964e-02 -7.00111464e-02 1.84609517e-01 9.77097452e-01
2.22553387e-01 2.44851857e-01 1.61731720e-01 8.95500958e-01
-2.64185160e-01 -1.07026851e+00 -6.96853340e-01 5.12210071e-01
-6.59433603e-01 -1.64443836e-01 1.00709572e-01 4.31355625e-01
-3.84926945e-01 1.18816769e+00 3.25860381e-02 1.63990676e-01
-3.78861755e-01 3.73364508e-01 2.76405126e-01 -3.14361423e-01
-1.51384175e-01 1.66785061e-01 -2.05702916e-01 -5.08223414e-01
-6.37988031e-01 -8.53716373e-01 -7.19144881e-01 -4.57964897e-01
-5.15069067e-01 8.37162554e-01 1.05605078e+00 8.89780939e-01
5.80104589e-01 -3.84880081e-02 6.49685204e-01 -8.60765517e-01
-1.11464417e+00 -1.12876093e+00 -8.77177179e-01 -2.03619659e-01
2.01166824e-01 -7.98321903e-01 -4.08837199e-01 1.56731740e-01] | [5.850080490112305, 6.3189592361450195] |
3443002b-d960-49cb-8a44-33783f921f5c | long-range-language-modeling-with-self | 2306.13421 | null | https://arxiv.org/abs/2306.13421v1 | https://arxiv.org/pdf/2306.13421v1.pdf | Long-range Language Modeling with Self-retrieval | Retrieval-augmented language models (LMs) have received much attention recently. However, typically the retriever is not trained jointly as a native component of the LM, but added to an already-pretrained LM, which limits the ability of the LM and the retriever to adapt to one another. In this work, we propose the Retrieval-Pretrained Transformer (RPT), an architecture and training procedure for jointly training a retrieval-augmented LM from scratch for the task of modeling long texts. Given a recently generated text chunk in a long document, the LM computes query representations, which are then used to retrieve earlier chunks in the document, located potentially tens of thousands of tokens before. Information from retrieved chunks is fused into the LM representations to predict the next target chunk. We train the retriever component with a semantic objective, where the goal is to retrieve chunks that increase the probability of the next chunk, according to a reference LM. We evaluate RPT on four long-range language modeling tasks, spanning books, code, and mathematical writing, and demonstrate that RPT improves retrieval quality and subsequently perplexity across the board compared to strong baselines. | ['Jonathan Berant', 'Ohad Rubin'] | 2023-06-23 | null | null | null | null | ['retrieval'] | ['methodology'] | [ 1.82174519e-01 1.42561167e-01 -2.63877064e-01 -1.39604911e-01
-1.37597179e+00 -3.86004984e-01 7.61654496e-01 8.98692980e-02
-5.45655370e-01 4.82876629e-01 5.57027280e-01 -2.73033194e-02
8.61361921e-02 -5.48980594e-01 -1.10743260e+00 -4.51079279e-01
2.66915798e-01 1.02297533e+00 2.46417597e-01 -2.86127895e-01
2.37865657e-01 1.58703223e-01 -1.20585978e+00 7.54433811e-01
8.36254478e-01 5.56618333e-01 9.53312278e-01 6.27195716e-01
-4.56853718e-01 1.10992777e+00 -6.29945457e-01 -3.09232384e-01
-3.24418753e-01 -3.77378732e-01 -8.03314447e-01 -3.05968523e-01
2.30856895e-01 -3.68115425e-01 -6.01544499e-01 4.94800389e-01
4.30209696e-01 4.16491061e-01 9.52049613e-01 -4.48678941e-01
-8.98017585e-01 1.44296014e+00 -3.17633897e-01 1.45838931e-01
2.50340372e-01 -3.69400382e-01 1.20407665e+00 -1.29984879e+00
6.85189426e-01 1.41174817e+00 -1.39113413e-02 6.51174366e-01
-1.01334453e+00 -5.21471798e-01 3.77920181e-01 5.36702946e-02
-1.53748631e+00 -7.39739537e-01 4.52872813e-01 -1.52033791e-01
1.56900823e+00 -6.07217960e-02 1.71323955e-01 1.00261343e+00
6.00003079e-02 1.37458527e+00 1.45743653e-01 -8.04856598e-01
-1.95139289e-01 3.53791147e-01 1.66221425e-01 2.43520543e-01
-1.89124629e-01 -2.69427568e-01 -9.10924435e-01 9.50194746e-02
2.99492449e-01 -3.08069289e-02 -1.76765248e-01 -6.16550306e-03
-1.11841214e+00 6.04414701e-01 3.11298281e-01 3.59515995e-01
-4.28784609e-01 3.12787533e-01 1.87287137e-01 2.14865193e-01
7.14761496e-01 5.66585958e-01 -4.18606251e-01 -8.24789256e-02
-1.37266052e+00 1.91309601e-01 7.88362443e-01 1.23861730e+00
6.56931221e-01 -2.90424496e-01 -5.74268520e-01 1.21256506e+00
6.27932847e-01 7.84026027e-01 7.90517151e-01 -4.51900631e-01
8.17418218e-01 3.79844576e-01 -3.14700641e-02 -5.06874442e-01
-5.88074476e-02 -6.20085239e-01 -4.35124665e-01 -7.28240311e-01
-2.16879845e-01 2.43453756e-01 -8.03151011e-01 1.63183486e+00
-1.90964699e-01 -1.71389114e-02 2.97641397e-01 5.53160965e-01
7.85452843e-01 1.37280989e+00 2.35137299e-01 -6.74844757e-02
9.34677899e-01 -1.36272776e+00 -6.05976820e-01 -6.10880554e-01
9.85330820e-01 -9.88383353e-01 1.02329540e+00 2.28920937e-01
-1.44861257e+00 -5.25089502e-01 -8.60264540e-01 -4.86806363e-01
-3.63406003e-01 5.43534696e-01 -1.22406974e-01 -1.33253038e-01
-9.84665871e-01 4.47499424e-01 -5.97480357e-01 -2.20384538e-01
1.75126255e-01 5.93500957e-02 -3.19592208e-02 -3.44393134e-01
-1.39467239e+00 1.06999171e+00 4.46130246e-01 -9.62130260e-03
-9.37022865e-01 -8.21414292e-01 -6.74413025e-01 5.91221273e-01
1.46007612e-01 -6.74176991e-01 1.42621410e+00 -8.23186278e-01
-1.55947971e+00 6.61148667e-01 -4.39538658e-01 -5.79124868e-01
4.16464247e-02 -6.61650956e-01 -2.53991127e-01 5.43355010e-02
1.21509969e-01 9.29222822e-01 1.07733870e+00 -9.70647991e-01
-5.74108899e-01 -2.09571734e-01 -1.44004166e-01 5.59333682e-01
-3.16486180e-01 1.33760154e-01 -1.08650136e+00 -9.11854982e-01
-2.11886004e-01 -7.87372649e-01 2.82619983e-01 -5.73544621e-01
-3.54669929e-01 -6.29588366e-01 3.84656131e-01 -9.44399834e-01
1.53163171e+00 -2.37724924e+00 6.65302277e-01 2.64033936e-02
-2.20396951e-01 1.01838388e-01 -7.03673124e-01 7.02048004e-01
2.10927531e-01 4.84839268e-02 1.31988198e-01 -8.95126820e-01
1.01481445e-01 2.82510556e-02 -1.01261723e+00 -2.01875605e-02
2.45371461e-01 1.24198580e+00 -8.23615968e-01 -4.83348399e-01
-1.64584517e-01 4.81592357e-01 -3.68383348e-01 5.10389924e-01
-7.03542709e-01 -5.04867658e-02 -2.59520382e-01 3.18670720e-01
1.39581382e-01 -3.74347001e-01 -4.98283596e-04 2.20812768e-01
2.45354995e-01 8.71166587e-01 -4.40353096e-01 2.05342937e+00
-8.70506048e-01 6.00368738e-01 -5.42744279e-01 -7.19264686e-01
9.43610728e-01 3.10286164e-01 1.30145073e-01 -1.05459452e+00
-1.70205295e-01 2.65269458e-01 -3.76289129e-01 -2.80497700e-01
1.11584079e+00 1.48956418e-01 -2.03725040e-01 9.02574897e-01
1.84052438e-01 -8.24849382e-02 1.89400405e-01 7.84567535e-01
1.01429057e+00 1.84093237e-01 -1.48115784e-01 1.39501780e-01
4.01550084e-01 -1.03822060e-01 -2.08354548e-01 1.00611961e+00
7.64639378e-01 5.62241673e-01 1.48420542e-01 1.79733217e-01
-1.08294833e+00 -1.07766044e+00 2.40958154e-01 1.86002219e+00
4.35180515e-02 -5.45149684e-01 -3.51753533e-01 -5.53155184e-01
1.00291841e-01 1.13521183e+00 -3.53024781e-01 -7.21962512e-01
-8.30291331e-01 -1.71476468e-01 4.39116836e-01 5.69277823e-01
3.00282277e-02 -1.33195591e+00 -1.30606130e-01 4.53562737e-01
-3.07264268e-01 -9.62051868e-01 -6.88158512e-01 7.71522447e-02
-9.14185643e-01 -5.19515932e-01 -1.02120090e+00 -8.90403032e-01
7.24480033e-01 3.39187145e-01 1.36641693e+00 8.33436549e-02
1.34459108e-01 5.15763164e-01 -6.48978353e-01 -1.47816807e-01
-5.83414495e-01 7.50137866e-01 -2.86264062e-01 -3.05963129e-01
2.61898845e-01 -1.26855806e-01 -2.30370507e-01 -1.03692442e-01
-8.88549328e-01 2.08520800e-01 1.17499018e+00 9.14822698e-01
6.11231565e-01 -3.28572422e-01 8.63165259e-01 -8.46530676e-01
6.79490924e-01 -7.11211622e-01 -5.21667182e-01 7.42355227e-01
-4.92190868e-01 4.07828987e-01 4.13810939e-01 -7.79805481e-01
-1.20785749e+00 -2.97545135e-01 1.44470081e-01 -5.97395480e-01
6.10411346e-01 8.59898269e-01 6.38636723e-02 6.21528208e-01
4.28706348e-01 7.10056901e-01 -3.29682499e-01 -7.69086659e-01
7.49147952e-01 8.22497427e-01 4.16210383e-01 -6.69216573e-01
5.38238585e-01 -1.46812201e-01 -6.32605672e-01 -5.44932067e-01
-1.04484868e+00 -6.58726096e-01 -6.69681132e-01 -3.50819118e-02
3.53307694e-01 -1.28232658e+00 -4.02518697e-02 1.73493385e-01
-1.48578370e+00 -6.05213225e-01 -5.21543205e-01 3.59819710e-01
-3.34798187e-01 -1.03317089e-01 -6.62764549e-01 -5.67195952e-01
-6.71999753e-01 -9.18180823e-01 1.50444806e+00 5.88658378e-02
-2.20928833e-01 -9.43537891e-01 1.62404135e-01 1.98903486e-01
5.46500444e-01 -9.28306997e-01 1.28503203e+00 -8.65144670e-01
-7.67905056e-01 -4.22391057e-01 -2.59498835e-01 4.37688649e-01
-2.53509253e-01 -1.50420576e-01 -9.15780008e-01 -4.29090142e-01
-4.37005073e-01 -6.40424073e-01 1.26464999e+00 2.01319814e-01
1.07300806e+00 -4.38750625e-01 -6.56399071e-01 3.36096168e-01
9.84573543e-01 -2.28342842e-02 6.08992398e-01 2.01390252e-01
5.24440229e-01 5.02765715e-01 6.40045345e-01 2.70714581e-01
4.23555166e-01 7.22701550e-01 1.98166054e-02 4.29234654e-01
-3.92101467e-01 -6.88145578e-01 8.27969909e-01 1.44073856e+00
4.82100099e-01 -5.94718456e-01 -7.45348752e-01 5.59483945e-01
-1.68840206e+00 -7.18973994e-01 5.15210390e-01 2.17522192e+00
1.17947066e+00 5.23627689e-03 -4.32381481e-01 -5.56342065e-01
3.42394650e-01 1.96160048e-01 -7.08695948e-01 -1.02570176e-01
-1.26107141e-01 3.64941508e-01 1.90193608e-01 5.48564494e-01
-6.27182603e-01 1.47838783e+00 6.08779716e+00 1.09691226e+00
-1.09922683e+00 1.06499732e-01 3.20927233e-01 -5.61440527e-01
-4.82122064e-01 1.52352065e-01 -1.22109878e+00 3.83061171e-01
1.26950812e+00 -5.53767920e-01 5.68140864e-01 7.74435043e-01
-7.29711652e-02 -8.40694085e-02 -1.40038347e+00 7.58326948e-01
4.29832101e-01 -1.16963434e+00 7.71742404e-01 -7.16241077e-02
6.76966131e-01 3.54792118e-01 4.02671844e-01 9.85968113e-01
2.01568112e-01 -1.14078820e+00 9.41790402e-01 9.48058844e-01
6.28915906e-01 -5.56749523e-01 4.77720320e-01 6.78376019e-01
-8.31150293e-01 7.70145580e-02 -8.34472656e-01 4.19742882e-01
9.04189646e-02 4.76945698e-01 -1.00276494e+00 2.74392575e-01
3.34220678e-01 1.03247762e+00 -5.81218541e-01 8.13695431e-01
-3.87065440e-01 5.57242870e-01 -1.67268679e-01 -1.36759281e-01
2.03438565e-01 1.93040237e-01 4.83316064e-01 1.31832933e+00
4.20088351e-01 -7.57913515e-02 -1.14824176e-01 1.08668280e+00
-6.83257520e-01 3.66325021e-01 -3.43392998e-01 -4.42179829e-01
8.63286316e-01 9.98802543e-01 -3.93321272e-03 -5.72418749e-01
-3.43565524e-01 1.10394812e+00 7.22186029e-01 6.02963686e-01
-5.74435771e-01 -3.45472187e-01 1.35117680e-01 -2.79037617e-02
4.39934433e-01 -8.68225843e-02 2.77578920e-01 -1.18185151e+00
1.57652825e-01 -7.48505235e-01 3.66182387e-01 -1.07272613e+00
-1.23714685e+00 5.68692327e-01 -1.15032874e-01 -1.04483771e+00
-5.99616826e-01 -1.73697829e-01 -2.79592782e-01 1.21720576e+00
-1.82582927e+00 -1.16787076e+00 2.05037266e-01 3.29028815e-01
1.07057834e+00 -3.95490289e-01 8.07923913e-01 2.56646067e-01
-4.36531812e-01 6.04526103e-01 5.85747004e-01 5.81964850e-02
9.31601346e-01 -9.58809614e-01 1.70962989e-01 6.03284001e-01
4.54192758e-01 8.66306305e-01 2.56420046e-01 -7.01836467e-01
-1.45118654e+00 -1.26872432e+00 1.33477235e+00 -5.07469416e-01
6.21136487e-01 -3.82861346e-01 -1.19591916e+00 9.28972363e-01
2.62877166e-01 -4.10791159e-01 5.29086411e-01 1.52880669e-01
-4.95259434e-01 -8.00954103e-02 -2.34151065e-01 4.33089823e-01
4.91928965e-01 -9.57471251e-01 -9.40526843e-01 4.87979352e-01
1.07692325e+00 -4.10657436e-01 -6.44711971e-01 9.99044999e-02
6.16167009e-01 -2.02071279e-01 1.09068215e+00 -6.63845420e-01
5.27661145e-01 1.94321305e-01 -1.78419515e-01 -1.33626604e+00
-1.01240076e-01 -3.22731823e-01 -6.04073644e-01 1.24557173e+00
8.16952229e-01 -1.91689417e-01 3.77476692e-01 4.33028132e-01
-3.29890311e-01 -6.62797868e-01 -8.29834819e-01 -8.40358496e-01
4.21328664e-01 -3.44273448e-01 5.52456498e-01 4.78677124e-01
2.35942885e-01 8.44748378e-01 -1.77323118e-01 4.95027527e-02
1.44321635e-01 8.72938633e-02 7.39323199e-01 -1.01230955e+00
-4.53316361e-01 -5.42902648e-01 4.24265742e-01 -1.81786978e+00
5.87626696e-01 -1.42491293e+00 4.33234215e-01 -1.62672913e+00
5.66077948e-01 -6.79244459e-01 -2.90453434e-01 4.60119039e-01
-2.64414996e-01 -2.98851371e-01 3.26261848e-01 7.71262169e-01
-9.63535905e-01 9.31892693e-01 9.90277588e-01 -5.49781561e-01
-4.38086182e-01 -8.91717896e-02 -5.73130608e-01 1.48944139e-01
4.32883173e-01 -7.95366824e-01 -5.36044419e-01 -1.04803252e+00
5.86274445e-01 2.04223767e-01 5.14824800e-02 -7.65045583e-01
6.51872873e-01 4.21767265e-01 3.35091919e-01 -8.42677176e-01
5.05770862e-01 -5.60792923e-01 -2.29822442e-01 -5.28544635e-02
-9.80111957e-01 -2.79066842e-02 2.59777278e-01 4.94718760e-01
-3.31328094e-01 -6.02849603e-01 3.99980158e-01 -3.76923289e-03
-4.18849409e-01 2.18664601e-01 -3.10025394e-01 7.94409215e-02
5.29860079e-01 3.95758837e-01 -4.52051014e-01 -5.07583916e-01
-7.58063078e-01 3.73650163e-01 2.09423184e-01 7.45439112e-01
7.49110341e-01 -1.09919572e+00 -7.97162354e-01 5.19332327e-02
3.10253412e-01 1.98369071e-01 1.49675146e-01 6.59309924e-01
-5.95733412e-02 9.23305929e-01 4.11316246e-01 -4.70354199e-01
-9.59152937e-01 5.73759675e-01 2.56308198e-01 -7.42230594e-01
-5.78234673e-01 9.82590437e-01 1.57262549e-01 -3.55559707e-01
5.91036558e-01 -3.32472980e-01 -3.10575515e-01 7.72398412e-02
6.66215479e-01 3.82612199e-02 8.28470513e-02 -4.91818517e-01
3.61978933e-02 3.71880233e-01 -7.17672586e-01 -4.32394624e-01
1.34274983e+00 -3.35888773e-01 -3.76103997e-01 6.52066708e-01
1.26870596e+00 -1.00438178e-01 -7.84700334e-01 -9.28179979e-01
3.14578563e-01 -6.26667589e-02 1.67350143e-01 -9.16802645e-01
-7.91182220e-01 1.02109087e+00 -1.45466132e-02 -3.32159519e-01
8.14720988e-01 4.99070942e-01 8.81449282e-01 7.90614843e-01
3.20922911e-01 -1.04374969e+00 4.52361405e-01 9.52433765e-01
1.04642034e+00 -8.41154456e-01 -1.09224997e-01 1.72002822e-01
-6.69320703e-01 1.08186924e+00 4.81790036e-01 1.49002910e-01
4.94876504e-01 -1.15126900e-01 -2.38040060e-01 -6.75881654e-02
-1.22802448e+00 -5.63942827e-02 6.98328912e-01 1.54347435e-01
5.72664857e-01 -2.27350637e-01 2.28036001e-01 7.33933449e-01
-9.06209573e-02 -6.71555102e-02 6.81382194e-02 7.89848626e-01
-6.89225376e-01 -1.13504016e+00 -1.27191897e-02 6.44016266e-01
-1.47613555e-01 -7.17651427e-01 -2.95439601e-01 4.19439673e-01
-4.47962701e-01 6.51283085e-01 2.62787998e-01 -1.03554666e-01
1.47173956e-01 5.31815410e-01 3.31748843e-01 -1.11612940e+00
-3.28580558e-01 2.01134771e-01 -1.53174520e-01 -3.92568111e-01
-1.43664882e-01 -5.38349867e-01 -1.26163411e+00 9.25953910e-02
-5.47219932e-01 3.74983698e-01 6.50236607e-01 9.51233804e-01
5.82476914e-01 5.86483121e-01 4.06497538e-01 -5.80614388e-01
-6.08422279e-01 -1.45011151e+00 -5.06303608e-01 4.78170812e-02
1.76631033e-01 -2.44376466e-01 -1.72514498e-01 7.88475722e-02] | [11.462100982666016, 7.797205924987793] |
fec4c822-69e3-4dfb-8b3e-aa6b195305f2 | when-the-majority-is-wrong-leveraging | 2305.06626 | null | https://arxiv.org/abs/2305.06626v3 | https://arxiv.org/pdf/2305.06626v3.pdf | When the Majority is Wrong: Modeling Annotator Disagreement for Subjective Tasks | Though majority vote among annotators is typically used for ground truth labels in natural language processing, annotator disagreement in tasks such as hate speech detection may reflect differences in opinion across groups, not noise. Thus, a crucial problem in hate speech detection is determining whether a statement is offensive to the demographic group that it targets, when that group may constitute a small fraction of the annotator pool. We construct a model that predicts individual annotator ratings on potentially offensive text and combines this information with the predicted target group of the text to model the opinions of target group members. We show gains across a range of metrics, including raising performance over the baseline by 22% at predicting individual annotators' ratings and by 33% at predicting variance among annotators, which provides a metric for model uncertainty downstream. We find that annotator ratings can be predicted using their demographic information and opinions on online content, without the need to track identifying annotator IDs that link each annotator to their ratings. We also find that use of non-invasive survey questions on annotators' online experiences helps to maximize privacy and minimize unnecessary collection of demographic information when predicting annotators' opinions. | ['Dan Klein', 'Rediet Abebe', 'Eve Fleisig'] | 2023-05-11 | null | null | null | null | ['hate-speech-detection'] | ['natural-language-processing'] | [-1.71016902e-01 5.59117079e-01 -3.92040163e-01 -6.32993698e-01
-8.23140919e-01 -1.24087870e+00 3.13265413e-01 6.13070726e-01
-4.93642598e-01 4.46682811e-01 7.78504074e-01 -1.13291092e-01
5.43840766e-01 -2.15996146e-01 -2.13661883e-02 -3.31332147e-01
4.02853131e-01 2.29911506e-01 -2.20827311e-01 -9.94832441e-03
2.33328730e-01 6.04435951e-02 -7.69090414e-01 5.63146710e-01
9.16230440e-01 9.94115174e-01 -6.85579836e-01 5.76033056e-01
1.71199813e-01 1.33415711e+00 -1.06190968e+00 -1.22373664e+00
1.78509384e-01 -3.01229805e-01 -7.90010571e-01 -3.88697572e-02
9.19925034e-01 -4.74658132e-01 -2.14074790e-01 1.18917704e+00
6.18534923e-01 -6.26364499e-02 7.44585991e-01 -1.09290111e+00
-1.14079392e+00 1.05779648e+00 -6.28861487e-01 1.67691514e-01
4.02101070e-01 1.01408921e-01 1.24744403e+00 -7.07476556e-01
4.66667414e-01 1.03081381e+00 9.86454666e-01 7.13798106e-01
-1.30366158e+00 -8.40629280e-01 -4.20327485e-02 -1.16864063e-01
-1.26109409e+00 -6.06302977e-01 5.59595942e-01 -1.11512303e+00
2.21777216e-01 3.85071516e-01 8.42669606e-02 1.36624014e+00
-2.89321214e-01 5.21643996e-01 1.29366159e+00 1.79641191e-02
3.79516512e-01 9.48352218e-01 5.80974281e-01 5.17611265e-01
2.61838704e-01 -5.02411008e-01 -8.41118634e-01 -9.72203851e-01
-2.54965037e-01 -2.79129624e-01 -2.61510819e-01 2.45473787e-01
-6.66879714e-01 1.06761515e+00 3.10971230e-01 2.15694413e-01
-1.95241138e-01 -2.96263158e-01 6.27232790e-01 2.56268829e-01
1.01898015e+00 9.47459340e-01 -2.94993967e-01 -3.35780494e-02
-9.23438787e-01 1.87685847e-01 1.28668702e+00 4.95449126e-01
6.06858432e-01 -2.07463533e-01 -3.90387446e-01 1.11279559e+00
6.78482279e-02 7.59706438e-01 4.48469847e-01 -1.03742230e+00
6.04010284e-01 4.88331854e-01 6.59337878e-01 -1.47215199e+00
-2.21290782e-01 -2.72037208e-01 -2.32737660e-01 -1.09433614e-01
7.73102462e-01 -9.54601943e-01 -4.17964429e-01 1.68115342e+00
2.25585416e-01 -5.77508569e-01 -4.55750734e-01 9.69131529e-01
4.52234775e-01 4.48146880e-01 4.98517156e-01 -4.67723817e-01
1.50296009e+00 -5.08866012e-01 -8.03730786e-01 -4.55766141e-01
1.10579872e+00 -7.65322208e-01 7.34066904e-01 4.81170535e-01
-7.00736761e-01 -1.81236193e-01 -6.05933785e-01 -5.35240769e-02
-3.17146957e-01 -1.01146810e-01 4.91005689e-01 1.23204052e+00
-7.95914650e-01 4.63659734e-01 -1.48537651e-01 -3.66566777e-01
4.53500211e-01 2.33250558e-01 -4.82807070e-01 2.36489430e-01
-1.36272919e+00 1.29196632e+00 -2.44747400e-01 6.37624264e-02
-4.71570462e-01 -6.79002702e-01 -6.60619199e-01 6.71350397e-03
2.25095853e-01 3.12129268e-03 1.39172065e+00 -1.18226063e+00
-7.26837814e-01 1.06141508e+00 -2.32125431e-01 -3.50980461e-01
5.32111764e-01 -4.11621094e-01 -5.37083507e-01 1.92179251e-02
3.30940038e-01 3.84038657e-01 7.32385516e-01 -1.02139866e+00
-3.94037187e-01 -6.47488713e-01 -2.30395317e-01 1.64761499e-01
-9.77594376e-01 5.35146594e-01 5.94678760e-01 -3.28570127e-01
-2.97511637e-01 -1.08070409e+00 5.66156954e-02 2.06963811e-02
-5.72008550e-01 -3.83546293e-01 9.50390279e-01 -1.12856996e+00
1.44854915e+00 -1.95184326e+00 -3.29143494e-01 2.81979501e-01
6.35727763e-01 4.72309202e-01 2.27436602e-01 3.75208318e-01
6.46195561e-02 7.30513871e-01 2.19682008e-01 -5.15713453e-01
3.03192157e-02 -3.26378614e-01 -5.49436569e-01 7.59441853e-01
-7.20232278e-02 6.44674182e-01 -9.69478905e-01 -4.39311773e-01
-1.51060134e-01 1.04946837e-01 -3.50326598e-01 3.04253101e-01
2.14663953e-01 1.96557030e-01 -3.50164115e-01 6.43369913e-01
4.34126973e-01 2.56935284e-02 1.86315775e-01 -6.52120188e-02
2.41607497e-03 4.62727576e-01 -4.43150818e-01 9.22888100e-01
-2.69633055e-01 8.58743608e-01 5.71575642e-01 -4.19183344e-01
1.28273332e+00 5.42590022e-01 5.77212237e-02 -4.35875282e-02
4.43552017e-01 2.43385017e-01 1.25599056e-01 -6.48385108e-01
5.50463200e-01 -2.65657902e-01 -4.64538574e-01 8.11515629e-01
-2.62095600e-01 8.05887878e-02 -9.13089961e-02 3.88608664e-01
1.13961077e+00 -7.48417437e-01 2.96846896e-01 -2.96564966e-01
2.03793615e-01 1.28946051e-01 4.56130147e-01 9.47323442e-01
-8.32165182e-01 3.60111982e-01 8.29345584e-01 -6.38957202e-01
-1.21483350e+00 -6.43576086e-01 6.46623224e-03 1.54061484e+00
-2.78868079e-01 -2.74218500e-01 -9.11524534e-01 -1.20048678e+00
1.34304799e-02 9.71552670e-01 -9.32129204e-01 -3.20033431e-01
-1.39664903e-01 -6.00953162e-01 7.41739213e-01 3.23285997e-01
5.92572168e-02 -6.71743274e-01 5.09540085e-04 -9.91586223e-02
-4.20913428e-01 -1.00783849e+00 -8.55518579e-01 1.49725884e-01
-2.70086557e-01 -9.39851820e-01 -4.53041106e-01 -4.30252641e-01
7.53282011e-01 1.75345987e-01 7.55007565e-01 6.24773353e-02
1.29201502e-01 1.24098636e-01 -4.56437856e-01 -5.37492990e-01
-6.27910912e-01 2.97588021e-01 3.25431079e-01 3.75777990e-01
9.78936017e-01 -3.61867487e-01 -4.58771795e-01 4.67172682e-01
-4.88439918e-01 -4.26762015e-01 -2.64633261e-02 5.35710990e-01
-4.92473483e-01 -5.59679627e-01 4.74690855e-01 -1.37021589e+00
9.87819612e-01 -8.48322988e-01 -4.61152196e-02 4.73521128e-02
-3.13599676e-01 -2.22307086e-01 7.28003979e-01 -8.42242002e-01
-8.98523450e-01 -1.97685882e-02 3.05489540e-01 -2.60032147e-01
-1.77613378e-01 1.46055654e-01 3.22771102e-01 1.67696327e-01
1.20464218e+00 -4.32886064e-01 1.29482284e-01 -3.74798715e-01
8.58464688e-02 1.25146222e+00 3.58421564e-01 -3.70765388e-01
9.64180291e-01 2.81353027e-01 -6.80830061e-01 -6.14964902e-01
-1.74927175e+00 -6.47782624e-01 -5.28266609e-01 -3.55413169e-01
9.82451737e-01 -9.89607513e-01 -9.82979238e-01 1.67993814e-01
-1.34724617e+00 -8.63873065e-02 1.56936035e-01 3.65183502e-01
2.45128945e-01 3.24549705e-01 -8.68875444e-01 -1.43775296e+00
-5.66696644e-01 -7.81294167e-01 7.85262823e-01 2.61997022e-02
-1.34919095e+00 -1.15339899e+00 1.61880881e-01 9.92718577e-01
4.17810112e-01 2.00426355e-01 5.57968974e-01 -1.64011443e+00
3.90550196e-01 -6.91308439e-01 -1.63841605e-01 7.00477839e-01
-3.84467724e-03 -2.51300652e-02 -1.22732842e+00 -1.82803512e-01
1.70356631e-01 -9.18811440e-01 3.48575205e-01 -4.00432274e-02
8.07418168e-01 -8.16352367e-01 -2.55415648e-01 -1.91386878e-01
8.92259181e-01 -1.83242992e-01 4.57747392e-02 -1.33836553e-01
8.05990398e-01 1.02384448e+00 2.39891022e-01 6.14411294e-01
2.26697028e-01 5.76703668e-01 1.83809772e-01 2.49702066e-01
5.12359083e-01 -5.32994509e-01 6.93657398e-01 6.70683920e-01
2.14723825e-01 -2.04654977e-01 -1.02131283e+00 4.93058950e-01
-1.73048210e+00 -1.15307724e+00 -2.53519595e-01 2.11288571e+00
8.23163033e-01 -4.92870770e-02 4.47653711e-01 -2.11005390e-01
1.32069457e+00 2.93054849e-01 -4.56768215e-01 -8.62151623e-01
1.19700447e-01 -3.21635276e-01 8.07247758e-01 7.49409437e-01
-1.06433284e+00 7.46381521e-01 6.79677439e+00 3.64842147e-01
-9.06798124e-01 3.54964495e-01 1.15103841e+00 -3.92196536e-01
-1.98796049e-01 -2.01312453e-01 -9.15594220e-01 8.00517797e-01
1.17822671e+00 -1.66621134e-01 3.78435880e-01 1.12089932e+00
4.37000096e-01 9.60152596e-02 -1.07391429e+00 6.55876935e-01
5.86159587e-01 -7.57830203e-01 -5.15610158e-01 3.66095543e-01
9.68823254e-01 -1.42758399e-01 1.07038304e-01 1.30980730e-01
8.64455998e-01 -9.43776369e-01 8.58503759e-01 5.41050881e-02
4.16179508e-01 -6.62014484e-01 9.29180264e-01 4.48289275e-01
-3.24669242e-01 -4.41987634e-01 -4.44319814e-01 -2.75964677e-01
2.08539531e-01 8.09808731e-01 -1.24939489e+00 -5.73341072e-01
3.70675206e-01 3.76082867e-01 -5.89855790e-01 5.30967474e-01
-3.72369379e-01 1.02446389e+00 -2.02821836e-01 -2.96483874e-01
1.76530644e-01 2.95597285e-01 3.83474171e-01 1.26853681e+00
1.50977805e-01 6.89053386e-02 2.11438596e-01 7.98451483e-01
-4.96702135e-01 2.85463274e-01 -8.00071239e-01 -3.31646293e-01
9.89134073e-01 1.65876424e+00 -1.18156709e-01 -2.82089382e-01
-4.08077031e-01 8.40207756e-01 7.63811469e-01 2.12979451e-01
-7.32816756e-01 -2.92259634e-01 6.75021708e-01 2.59441614e-01
-1.88940391e-01 1.19004548e-01 -5.91430008e-01 -1.11764991e+00
-5.84189482e-02 -1.07734191e+00 4.60534930e-01 -9.02311802e-01
-1.68561888e+00 4.69690949e-01 -5.41885912e-01 -7.33741999e-01
-2.17627868e-01 -7.10994184e-01 -5.17331719e-01 1.03024590e+00
-5.00092506e-01 -8.35164964e-01 -1.89765677e-01 1.87160701e-01
2.21712172e-01 3.62966508e-02 8.01603913e-01 -1.04757562e-01
-6.14702702e-01 7.76048481e-01 -3.45794410e-01 8.08649063e-01
1.13444746e+00 -1.28732181e+00 1.11841582e-01 5.43728888e-01
-4.22687270e-02 8.24968696e-01 9.96218920e-01 -7.41186261e-01
-7.20016658e-01 -8.31409693e-01 1.38213444e+00 -1.26443243e+00
1.25324106e+00 -6.82480991e-01 -9.84870791e-01 7.32136250e-01
1.39348894e-01 -1.46120444e-01 1.38426232e+00 6.36647582e-01
-1.11917543e+00 1.89251557e-01 -1.23046410e+00 5.34207821e-01
8.58981490e-01 -1.08767807e+00 -6.18616641e-01 6.02494121e-01
5.08040726e-01 -8.43154937e-02 -1.08843803e+00 -2.35736638e-01
6.83085620e-01 -7.72044063e-01 2.09232017e-01 -8.09065819e-01
5.73088050e-01 -1.61637776e-02 1.06685199e-01 -1.54754663e+00
-4.71854925e-01 -7.93213427e-01 3.45890857e-02 1.50655651e+00
7.42192805e-01 -2.51758963e-01 9.27472711e-01 1.89927435e+00
2.60541975e-01 -2.83155978e-01 -4.49840218e-01 -2.79478043e-01
1.54990464e-01 -1.43937603e-01 7.31932521e-02 1.41192424e+00
7.95712650e-01 7.31755793e-01 -8.60532284e-01 2.57443279e-01
4.55185145e-01 -6.23339415e-01 5.68172157e-01 -1.37431145e+00
-1.53008774e-01 -1.10283889e-01 -2.08615676e-01 -5.31007171e-01
4.62497801e-01 -6.44536853e-01 -3.37799005e-02 -9.95698035e-01
4.82306480e-01 -2.98580468e-01 1.39617354e-01 6.50548041e-01
-4.49418217e-01 5.30660927e-01 2.07453519e-01 4.53088582e-01
-8.05693746e-01 -1.25999063e-01 8.23094547e-01 -1.15321748e-01
-5.71666062e-02 -1.74943969e-01 -1.42743254e+00 7.85139620e-01
6.82608902e-01 -8.32175255e-01 -1.13403387e-01 -3.50236833e-01
4.26073819e-01 -2.37590387e-01 2.85630003e-02 -7.31208444e-01
2.26502597e-01 -1.16088510e-01 4.45439458e-01 -2.21681111e-02
2.13905826e-01 -5.95614970e-01 -2.65763015e-01 2.85181314e-01
-1.11425638e+00 1.04124010e-01 -3.26820135e-01 3.91757399e-01
8.32342654e-02 -6.95158660e-01 7.43797779e-01 -1.49628401e-01
-1.25234323e-02 1.42884359e-01 -7.63875961e-01 3.30405980e-01
8.17334950e-01 9.78148580e-02 -9.45385098e-01 -9.84382391e-01
-8.63177776e-01 4.86448072e-02 6.15166068e-01 2.57072181e-01
-1.87723741e-01 -1.03138816e+00 -8.74706149e-01 -4.71639514e-01
1.75464407e-01 -7.26328671e-01 1.58752166e-02 7.21967220e-01
-3.62218730e-02 9.25544351e-02 2.16809601e-01 -7.17259385e-03
-1.47157502e+00 5.51379800e-01 2.54941463e-01 -1.79573502e-02
4.26625997e-01 9.77408409e-01 1.56533167e-01 -3.70457023e-01
9.72221941e-02 4.58600849e-01 -7.22620189e-02 4.24049854e-01
9.85657394e-01 3.87679279e-01 -2.01383024e-01 -1.03023708e+00
-2.00674579e-01 -2.16397122e-01 -5.44191837e-01 -9.30989385e-02
9.17625785e-01 -1.28168717e-01 -3.16687137e-01 5.99027872e-01
1.40029776e+00 6.27104521e-01 -8.38556945e-01 -1.74457312e-01
5.89583367e-02 -5.95637679e-01 -1.49936289e-01 -1.03194845e+00
-6.13512754e-01 7.53513038e-01 6.74256086e-02 8.31808448e-01
2.65411824e-01 1.19343646e-01 5.26367724e-01 2.98429012e-01
2.61665583e-01 -1.20279920e+00 1.54314697e-01 3.51336092e-01
5.49834609e-01 -1.46172535e+00 -1.53787779e-02 -4.45041597e-01
-1.17117500e+00 7.73977637e-01 8.99145007e-01 1.59825444e-01
4.41528052e-01 -7.43718147e-02 6.18285596e-01 -1.70627967e-01
-7.84512341e-01 3.29816252e-01 3.16909462e-01 7.60083139e-01
8.80919755e-01 2.20727786e-01 -3.48640084e-01 5.54142833e-01
-5.10425270e-01 -6.42482460e-01 6.44932926e-01 2.96980888e-01
-7.85056472e-01 -8.49474967e-01 -4.92720962e-01 7.64676750e-01
-1.00108421e+00 -2.61233330e-01 -1.24742115e+00 1.92438781e-01
9.24475640e-02 1.34682906e+00 1.49141680e-02 -6.83394253e-01
5.08867241e-02 4.35622811e-01 -2.75790513e-01 -9.13933635e-01
-1.15250695e+00 -5.50412536e-01 8.37180853e-01 -1.42811522e-01
-1.43834725e-01 -8.33513737e-01 -2.93809831e-01 -7.08383620e-01
-5.72231770e-01 6.98251128e-01 4.93481547e-01 9.86227095e-01
3.28909338e-01 -6.50801361e-01 8.14701796e-01 -2.77494192e-01
-8.34798038e-01 -1.20051360e+00 -5.15556633e-01 1.03442764e+00
2.07258672e-01 -2.04981595e-01 -8.37144673e-01 9.37808305e-03] | [8.76060962677002, 10.468401908874512] |
c5f87b86-a795-4ad0-9fd8-f3ce9205485b | robust-video-stabilization-by-optimization-in | null | null | http://openaccess.thecvf.com/content_CVPR_2019/html/Yu_Robust_Video_Stabilization_by_Optimization_in_CNN_Weight_Space_CVPR_2019_paper.html | http://openaccess.thecvf.com/content_CVPR_2019/papers/Yu_Robust_Video_Stabilization_by_Optimization_in_CNN_Weight_Space_CVPR_2019_paper.pdf | Robust Video Stabilization by Optimization in CNN Weight Space | We propose a novel robust video stabilization method. Unlike traditional video stabilization techniques that involve complex motion models, we directly model the appearance change of the frames as the dense optical flow field of consecutive frames. We introduce a new formulation of the video stabilization task based on first principles, which leads to a large scale non-convex problem. This problem is hard to solve, so previous optical flow based approaches have resorted to heuristics. In this paper, we propose a novel optimization routine that transfers this problem into the convolutional neural network parameter domain. While we exploit the general benefits of CNNs, including standard gradient-based optimization techniques, our method is a new approach to using CNNs purely as an optimizer rather than learning from data.Our method trains the CNN from scratch on each specific input example, and intentionally overfits the CNN parameters to produce the best result on the input example. By solving the problem in the CNN weight space rather than directly for image pixels, we make it a viable formulation for video stabilization. Our method produces both visually and quantitatively better results than previous work, and is robust in situations acknowledged as limitations in current state-of-the-art methods.
| [' Ravi Ramamoorthi', 'Jiyang Yu'] | 2019-06-01 | null | null | null | cvpr-2019-6 | ['video-stabilization'] | ['computer-vision'] | [-1.35913137e-02 -1.39107570e-01 -2.29941040e-01 -2.54229996e-02
-4.13701385e-01 -4.84916091e-01 3.11044991e-01 -2.98640490e-01
-5.81942439e-01 7.04631507e-01 2.27975801e-01 -2.37003326e-01
4.28021193e-01 -2.60655731e-01 -9.69064593e-01 -7.46222019e-01
1.46924600e-01 -1.69834778e-01 2.76566356e-01 -3.72330874e-01
3.00025433e-01 4.78009313e-01 -1.28730953e+00 -1.08836800e-01
6.46124780e-01 8.60706508e-01 -1.07816584e-01 7.64205754e-01
2.73145616e-01 1.07639730e+00 -1.30640537e-01 -7.51830265e-02
5.16565442e-01 -5.72133064e-01 -9.39667583e-01 4.07187432e-01
1.13992918e+00 -4.32662576e-01 -5.78180194e-01 9.35169458e-01
3.80408853e-01 6.28939629e-01 2.17844650e-01 -1.07777619e+00
-6.63420439e-01 -5.96250184e-02 -6.31903410e-01 3.84391278e-01
2.50028431e-01 3.67744118e-01 8.37699234e-01 -7.35037982e-01
8.04964542e-01 1.20251906e+00 7.77185917e-01 8.17490160e-01
-1.45050240e+00 -1.21180408e-01 4.39275384e-01 1.13587573e-01
-9.63526070e-01 -8.19661438e-01 7.73808241e-01 -5.56747496e-01
9.73423064e-01 -5.65412920e-04 8.29562187e-01 7.01207042e-01
2.27867737e-01 7.59111702e-01 6.49769902e-01 -4.56786841e-01
7.16032311e-02 -3.65042031e-01 -9.08600613e-02 1.05424798e+00
6.19364753e-02 5.00308387e-02 -5.45157373e-01 2.57337522e-02
1.16743934e+00 -2.42088154e-01 -6.81712627e-01 -7.68224061e-01
-1.39136291e+00 7.87354708e-01 5.76336980e-01 3.52579243e-02
-2.40524098e-01 7.61138916e-01 2.36093909e-01 1.73227414e-01
5.60236573e-01 5.32324433e-01 -4.58241731e-01 -3.48834023e-02
-1.27620208e+00 4.70510930e-01 6.96943462e-01 6.54589355e-01
9.81019914e-01 3.99812460e-01 -1.43698528e-01 5.66850543e-01
4.42879230e-01 1.25255659e-01 3.71297151e-01 -1.67429960e+00
2.54134864e-01 1.78237766e-01 3.29741985e-01 -1.19003928e+00
-3.03777814e-01 -1.25266775e-01 -5.80828547e-01 6.35053039e-01
8.11058223e-01 -3.90249014e-01 -8.34132552e-01 1.99049222e+00
4.78826374e-01 4.99040246e-01 -2.60125399e-01 1.02656138e+00
3.85204315e-01 6.96028292e-01 -2.29533762e-01 -3.34692150e-01
7.30949402e-01 -1.61551416e+00 -7.56333768e-01 -3.12723488e-01
5.71255684e-01 -8.12319040e-01 8.57048571e-01 2.38189518e-01
-1.50009596e+00 -4.31179017e-01 -1.06775105e+00 -3.99392724e-01
2.86616366e-02 -4.62707505e-02 5.08687854e-01 3.32161456e-01
-1.74549079e+00 1.05094409e+00 -1.13017011e+00 -5.17663419e-01
3.48394543e-01 6.30563676e-01 -5.50820172e-01 2.11275622e-01
-6.86729968e-01 1.02837229e+00 9.77094918e-02 3.31813544e-01
-7.09614396e-01 -7.13707209e-01 -1.08425188e+00 2.03984454e-02
3.73408914e-01 -8.97348881e-01 1.33551288e+00 -1.49820280e+00
-1.97873569e+00 7.43162751e-01 -5.69989741e-01 -3.40761602e-01
6.11052275e-01 -2.59205014e-01 1.94100589e-01 1.49853557e-01
-1.72162652e-01 9.22671020e-01 1.15347433e+00 -1.18812466e+00
-4.69293505e-01 2.44121596e-01 2.56721467e-01 1.13038614e-01
-2.85137773e-01 1.16523728e-01 -7.46407747e-01 -6.54708207e-01
1.84583862e-03 -1.10816479e+00 -4.73559439e-01 6.11262798e-01
-1.00233652e-01 2.48974174e-01 8.17176402e-01 -5.62306702e-01
1.14586139e+00 -1.92903197e+00 5.42314470e-01 -2.25996654e-02
4.22700256e-01 4.79973584e-01 -1.85848013e-01 -7.22107664e-02
-2.51361817e-01 -6.17429279e-02 -1.57589287e-01 -6.18358016e-01
-4.08753276e-01 2.17197970e-01 -1.21074550e-01 9.19968843e-01
3.46393049e-01 9.21642900e-01 -1.11645949e+00 -5.10963440e-01
3.77303630e-01 6.32997394e-01 -9.67295647e-01 1.94553033e-01
-1.25213936e-01 5.36741197e-01 -4.06747423e-02 4.63604838e-01
6.02359176e-01 -3.11672211e-01 4.03292030e-02 -4.06089276e-01
-3.14431280e-01 -7.84633607e-02 -1.21208966e+00 1.83662128e+00
-2.13604882e-01 1.02088082e+00 4.36439663e-01 -1.15879512e+00
3.84785801e-01 2.67505765e-01 8.49108994e-01 -2.14305907e-01
2.51140445e-01 7.28367493e-02 -1.48826227e-01 -5.33219218e-01
5.54050267e-01 -5.85105531e-02 6.99214101e-01 2.13467196e-01
3.86685133e-01 -1.96452409e-01 4.49285656e-01 6.35092631e-02
9.27370250e-01 6.97599709e-01 9.87838060e-02 -6.18014574e-01
6.65558696e-01 -7.05762254e-03 6.81573331e-01 5.60439646e-01
-6.11696720e-01 1.04896879e+00 5.80258608e-01 -8.89915228e-01
-1.21321726e+00 -5.84494829e-01 1.31178871e-01 8.89167845e-01
1.75836667e-01 -3.42156112e-01 -9.04238582e-01 -4.65288013e-01
-9.23084319e-02 -6.22870624e-02 -8.15436542e-01 3.12280152e-02
-9.49895024e-01 -6.09904766e-01 2.13803858e-01 3.41203541e-01
3.64871711e-01 -9.18746829e-01 -5.93607664e-01 3.94831777e-01
-2.64321536e-01 -1.17102087e+00 -1.03896570e+00 7.99155235e-02
-1.12600863e+00 -1.12364745e+00 -8.06432903e-01 -9.03827846e-01
8.13752770e-01 4.92969006e-01 1.13773310e+00 6.33983910e-01
-1.11931428e-01 3.97591472e-01 5.70548102e-02 9.25446302e-02
-1.32209539e-01 3.94896343e-02 1.80312470e-01 2.82986134e-01
-1.77498057e-01 -3.42901766e-01 -7.71271348e-01 1.48293763e-01
-8.83358657e-01 3.24789621e-02 -1.08412713e-01 9.30833340e-01
4.81272429e-01 -5.51125586e-01 -4.28364724e-02 -5.47330379e-01
4.16504145e-01 -1.38786688e-01 -8.09616327e-01 6.72725588e-02
-4.69435781e-01 2.59582460e-01 5.32283783e-01 -5.01530945e-01
-7.33708680e-01 5.63919067e-01 6.18650839e-02 -9.24826801e-01
2.32099697e-01 2.97700256e-01 3.18386286e-01 -7.26835549e-01
7.66071916e-01 -1.35444194e-01 4.79829818e-01 7.56707927e-03
2.86830395e-01 -1.56757802e-01 6.31306648e-01 -5.81584692e-01
9.01050270e-01 7.82564878e-01 1.94074243e-01 -8.67442608e-01
-8.42957437e-01 -5.48885345e-01 -7.74510026e-01 -5.60418487e-01
1.02488947e+00 -9.20727313e-01 -7.21253395e-01 6.11750960e-01
-1.37111831e+00 -7.04344392e-01 -1.79206207e-01 4.68874454e-01
-7.04760432e-01 5.94429076e-01 -7.12750077e-01 -3.81820083e-01
-4.01933715e-02 -1.47334659e+00 9.08883333e-01 2.82807171e-01
-1.94228396e-01 -1.35610962e+00 3.38115573e-01 8.27978365e-03
5.54259837e-01 4.64645505e-01 1.65479109e-01 1.65895328e-01
-7.65391529e-01 -5.40007986e-02 -1.10304162e-01 4.30815667e-01
1.16266720e-01 6.47120833e-01 -8.63427520e-01 -5.28047740e-01
5.24658114e-02 -3.50185603e-01 9.37119544e-01 8.86553943e-01
1.07569039e+00 -1.27851442e-01 -5.27120605e-02 1.14619768e+00
1.61058712e+00 -1.95575818e-01 8.09511065e-01 6.16575181e-01
1.03130639e+00 3.84595454e-01 1.31986618e-01 7.90793821e-02
2.70234674e-01 7.16681242e-01 5.47579646e-01 -4.75461930e-01
-1.39427960e-01 1.64388239e-01 6.14093602e-01 5.61362982e-01
-4.69631642e-01 -1.94365382e-01 -6.27841115e-01 6.20567203e-01
-2.30870652e+00 -1.08635688e+00 -1.84339494e-01 2.03356147e+00
8.16237390e-01 -8.26683417e-02 3.26295048e-02 -2.55553514e-01
7.40097463e-01 4.30849373e-01 -4.05880511e-01 -4.52828437e-01
-1.29417822e-01 8.26383978e-02 7.04380333e-01 9.36800420e-01
-1.31310236e+00 1.11843443e+00 7.39791298e+00 2.66677886e-01
-1.42187917e+00 -9.44168717e-02 5.95782042e-01 -4.23982322e-01
3.23974676e-02 1.49213344e-01 -6.09937489e-01 2.55973518e-01
5.80582678e-01 4.57924008e-02 6.73042417e-01 5.63594401e-01
6.71297252e-01 -8.87504667e-02 -1.23962069e+00 1.03464401e+00
2.13929489e-01 -1.73907888e+00 -1.43043205e-01 3.06939278e-02
1.08367705e+00 9.85886529e-02 5.03051169e-02 -2.75334090e-01
3.60160321e-01 -9.32741821e-01 9.56551850e-01 4.78875101e-01
4.00427669e-01 -3.25634003e-01 3.88106227e-01 -1.66760594e-01
-9.74878669e-01 9.66740623e-02 -3.50977570e-01 -2.25470006e-01
3.90663892e-01 4.17399585e-01 -2.24860907e-02 1.29935265e-01
7.19688416e-01 1.14855623e+00 -4.48294371e-01 1.28574049e+00
-1.02704801e-01 4.32765573e-01 -2.71815211e-01 2.79204279e-01
4.82990056e-01 -3.44154090e-01 4.86442834e-01 1.04493821e+00
2.10551634e-01 -7.26990625e-02 2.07197681e-01 6.32589817e-01
-2.28543460e-01 5.03789932e-02 -5.30634105e-01 2.52103090e-01
-3.01431865e-01 1.35752201e+00 -7.02476740e-01 -3.63792658e-01
-6.51056349e-01 9.06632006e-01 6.43599093e-01 6.90988839e-01
-8.60253334e-01 -1.61235586e-01 8.83461177e-01 1.11571908e-01
4.41061169e-01 -5.04524767e-01 -1.55003175e-01 -1.81141567e+00
5.35533056e-02 -8.65936697e-01 2.74181888e-02 -9.08413410e-01
-8.28161538e-01 5.38604259e-01 -2.13439748e-01 -1.30246186e+00
-2.11037025e-01 -7.63441265e-01 -8.61621618e-01 7.53647387e-01
-1.76705778e+00 -8.93234909e-01 -2.82399982e-01 8.35318089e-01
5.13244331e-01 1.82867855e-01 3.20003986e-01 3.28138649e-01
-7.59618938e-01 3.37912112e-01 2.51823992e-01 6.59803003e-02
9.94992912e-01 -1.18247247e+00 2.93189913e-01 1.39306593e+00
-1.72133133e-01 6.77841842e-01 9.00884271e-01 -2.71862000e-01
-1.27395678e+00 -8.26638877e-01 8.68480504e-01 -4.75130677e-01
8.84796381e-01 -1.44166797e-01 -9.34227884e-01 8.67170393e-01
6.58430398e-01 5.26628077e-01 1.80434600e-01 -2.32803226e-01
-8.92016813e-02 -6.15909621e-02 -7.92229474e-01 8.34514678e-01
8.74168158e-01 -3.63097042e-01 -2.49337122e-01 3.79014581e-01
5.46691060e-01 -8.06831300e-01 -4.90523577e-01 -1.17837312e-02
4.82807994e-01 -8.03725898e-01 9.91405964e-01 -8.05692613e-01
7.73830712e-01 -5.88844359e-01 1.45210400e-01 -1.38846016e+00
-5.64128399e-01 -1.13835454e+00 -3.08518589e-01 6.81215405e-01
4.52782571e-01 -3.17048013e-01 8.70411575e-01 9.59873438e-01
-2.63469011e-01 -5.56007266e-01 -7.68883049e-01 -4.22133595e-01
1.44126400e-01 -1.51291385e-01 4.71539348e-02 1.01494253e+00
-7.23764747e-02 2.26902440e-02 -6.44117057e-01 6.69879466e-02
6.80022418e-01 -2.02036619e-01 8.05315077e-01 -8.18296909e-01
-1.99356198e-01 -5.39971828e-01 -3.69337499e-01 -1.23371625e+00
3.75993639e-01 -4.91963595e-01 3.98006320e-01 -1.25885010e+00
-1.49573624e-01 -4.60031331e-02 -1.58483088e-01 5.53088367e-01
-3.33523393e-01 4.62937564e-01 3.51024300e-01 1.90198898e-01
-5.59488833e-01 3.92179966e-01 1.49606788e+00 -2.70741791e-01
-2.97512680e-01 -2.30402768e-01 -4.54363525e-01 7.03990340e-01
5.96093178e-01 -2.42261901e-01 -2.97513157e-01 -8.75786364e-01
9.34958830e-02 -1.50320306e-01 4.47885245e-01 -9.53461409e-01
4.31637138e-01 -3.77983689e-01 3.91597420e-01 -7.20662589e-04
2.07158014e-01 -6.78528666e-01 -8.81507173e-02 4.72335875e-01
-2.48504996e-01 3.49175125e-01 2.91321456e-01 3.60369593e-01
-2.74883926e-01 -2.18362704e-01 1.19994795e+00 -3.01279992e-01
-6.61227345e-01 5.98547578e-01 -5.37887514e-01 1.06856786e-01
7.37317860e-01 -2.99953341e-01 -1.23168088e-01 -4.91274774e-01
-9.14454222e-01 2.21145406e-01 8.93305898e-01 2.24182904e-01
3.55843425e-01 -1.30513477e+00 -4.48733211e-01 1.11097373e-01
-4.44535553e-01 1.02924660e-01 -6.09344468e-02 9.86065626e-01
-1.17869055e+00 1.66203529e-01 -4.05662417e-01 -7.29635358e-01
-1.17412138e+00 7.06159830e-01 7.52776086e-01 -3.58021050e-03
-4.54330891e-01 8.32040846e-01 1.46981567e-01 -1.04127154e-01
2.01759502e-01 -4.66370016e-01 -1.92868412e-01 -1.21914402e-01
4.26795632e-01 2.75087059e-01 8.52431953e-02 -8.85417759e-01
-1.91245645e-01 8.36868048e-01 3.66497934e-02 -1.51379913e-01
1.40989125e+00 -2.77784199e-01 -3.90189528e-01 -6.70946017e-02
1.45692694e+00 -1.01462580e-01 -1.95692003e+00 -5.99438176e-02
-3.56385767e-01 -7.13662803e-01 3.56298089e-01 -1.97754055e-02
-1.63772130e+00 7.02349722e-01 3.41062099e-01 4.46120836e-02
9.69286621e-01 -4.47280824e-01 6.31601512e-01 4.05420393e-01
-2.34252632e-01 -1.20954561e+00 1.56257495e-01 7.94282317e-01
6.79004788e-01 -1.44171035e+00 2.17263535e-01 -2.17833415e-01
-3.03267956e-01 1.38594258e+00 8.01357090e-01 -5.45925677e-01
7.33433843e-01 7.86221623e-02 3.81223708e-01 -2.65931115e-02
-6.77798927e-01 -5.67649975e-02 2.55637228e-01 4.70456600e-01
5.94246030e-01 -7.34992743e-01 -1.05886780e-01 -5.21111369e-01
2.76674926e-01 3.51562798e-01 8.32825184e-01 1.17117977e+00
-3.17650974e-01 -1.16335928e+00 -3.71153623e-01 9.60229337e-02
-5.80599666e-01 -5.48858307e-02 -6.53448030e-02 6.59446597e-01
-6.44652918e-02 7.30650067e-01 5.81539760e-04 -1.18549816e-01
2.98738837e-01 -1.41499370e-01 5.67093015e-01 -3.20676148e-01
-5.68802893e-01 1.54313311e-01 -1.93439737e-01 -9.71089005e-01
-1.07264841e+00 -7.82040954e-01 -9.45733190e-01 -6.31928146e-01
-2.55954623e-01 -1.61744550e-01 3.42742264e-01 9.62856591e-01
1.90965459e-01 3.30094546e-01 5.98595977e-01 -1.62826431e+00
-1.97770327e-01 -3.85726482e-01 -1.56388268e-01 4.57771093e-01
9.68723714e-01 -6.69670343e-01 -4.25148875e-01 6.27263844e-01] | [10.568838119506836, -1.3896687030792236] |
e08b7abf-2feb-4b1c-a506-a9c9b0de6b90 | a-simple-and-effective-approach-to-the-story | 1803.05547 | null | http://arxiv.org/abs/1803.05547v1 | http://arxiv.org/pdf/1803.05547v1.pdf | A Simple and Effective Approach to the Story Cloze Test | In the Story Cloze Test, a system is presented with a 4-sentence prompt to a
story, and must determine which one of two potential endings is the 'right'
ending to the story. Previous work has shown that ignoring the training set and
training a model on the validation set can achieve high accuracy on this task
due to stylistic differences between the story endings in the training set and
validation and test sets. Following this approach, we present a simpler
fully-neural approach to the Story Cloze Test using skip-thought embeddings of
the stories in a feed-forward network that achieves close to state-of-the-art
performance on this task without any feature engineering. We also find that
considering just the last sentence of the prompt instead of the whole prompt
yields higher accuracy with our approach. | ['Siddarth Srinivasan', 'Mark Riedl', 'Richa Arora'] | 2018-03-15 | a-simple-and-effective-approach-to-the-story-1 | https://aclanthology.org/N18-2015 | https://aclanthology.org/N18-2015.pdf | naacl-2018-6 | ['cloze-test'] | ['natural-language-processing'] | [-4.13474962e-02 1.53761178e-01 7.88966194e-02 -5.44854581e-01
-7.96246588e-01 -6.49336994e-01 7.32570052e-01 2.64140695e-01
-4.95938629e-01 3.52287114e-01 6.69976652e-01 -3.71726900e-01
1.54175863e-01 -6.39074326e-01 -5.53713381e-01 -1.94984362e-01
2.03938186e-01 4.54994678e-01 1.62675753e-01 -4.06409711e-01
2.29046226e-01 -1.15036573e-02 -1.23346961e+00 6.96108401e-01
2.24795192e-01 7.77067721e-01 -4.41696756e-02 1.05236709e+00
1.21036507e-01 1.19514596e+00 -8.61614883e-01 -5.41097581e-01
5.10122143e-02 -8.39464366e-01 -9.62789714e-01 -1.32194176e-01
5.31233370e-01 -6.65401816e-01 -6.61555231e-01 5.57987452e-01
3.89895707e-01 1.47493929e-01 7.94632971e-01 -9.39282775e-01
-9.45964813e-01 7.62578368e-01 -3.35120797e-01 3.21133852e-01
7.17402279e-01 1.08371221e-01 1.25990057e+00 -1.02846539e+00
6.45179570e-01 8.82668734e-01 7.07155824e-01 6.17600620e-01
-1.17862999e+00 -4.82076108e-01 1.23947691e-02 4.06569541e-01
-1.20465875e+00 -5.94282448e-01 7.52668142e-01 -6.93107307e-01
1.30444074e+00 1.49598375e-01 5.12256622e-01 1.21247828e+00
-2.36271992e-02 6.30210996e-01 9.51789796e-01 -5.57920694e-01
2.75916517e-01 2.77316183e-01 5.08204103e-01 2.32672408e-01
-1.49252176e-01 7.34147727e-02 -5.54822505e-01 -1.68672293e-01
3.52959126e-01 -5.84084451e-01 -2.35355780e-01 7.93553069e-02
-8.46318245e-01 1.05954993e+00 3.70137841e-01 5.12236774e-01
-1.36137813e-01 8.63107219e-02 5.79085350e-01 3.54209214e-01
4.74973351e-01 6.55798912e-01 -1.57186523e-01 -4.84385639e-01
-1.14494205e+00 6.15429282e-01 8.04545760e-01 5.90659440e-01
6.70406641e-03 2.03355271e-02 -2.56238312e-01 7.29956508e-01
8.33973289e-03 -7.35949501e-02 4.59022343e-01 -5.27627409e-01
5.71253717e-01 2.92490304e-01 2.20842376e-01 -8.73698235e-01
-4.47982609e-01 -4.27691877e-01 -1.36982620e-01 3.28171581e-01
7.65309155e-01 -1.73403800e-01 -6.16711617e-01 1.74345350e+00
-7.74899591e-03 -2.29906868e-02 1.23483948e-01 9.05137122e-01
1.02292299e+00 5.67015231e-01 -4.26003784e-02 2.08280664e-02
1.31869769e+00 -7.52905965e-01 -5.23285329e-01 -4.60836440e-01
8.61552894e-01 -6.05871201e-01 1.59838939e+00 4.32636946e-01
-9.95159864e-01 -4.35043722e-01 -1.64677608e+00 -3.18902642e-01
-3.10170323e-01 9.00131464e-03 3.07556391e-01 4.57788259e-01
-8.97083223e-01 6.95402086e-01 -3.45391124e-01 -3.28661114e-01
3.29020321e-01 -1.44349635e-02 -3.52739871e-01 -3.08856606e-01
-1.16504180e+00 1.38068771e+00 3.70483637e-01 -3.35113764e-01
-7.01696217e-01 -5.26624918e-01 -9.19792414e-01 1.17560439e-01
2.02916086e-01 -2.02808172e-01 1.68499422e+00 -8.75827909e-01
-1.28569412e+00 9.95197296e-01 -1.92441136e-01 -1.98866352e-01
6.06656909e-01 -4.63831335e-01 -4.21852618e-01 -3.20493914e-02
4.27859202e-02 3.91642720e-01 5.79536796e-01 -9.79816258e-01
-2.38228738e-01 5.37243225e-02 3.71520251e-01 1.29471287e-01
-2.22697586e-01 5.14736354e-01 5.30894510e-02 -5.59960127e-01
2.38062385e-02 -5.79352796e-01 3.55141282e-01 -3.18789601e-01
-3.87992710e-01 -2.97126830e-01 7.93046653e-01 -8.31401825e-01
1.38365746e+00 -2.48307800e+00 -2.55503684e-01 -8.16381574e-02
2.02097401e-01 7.36810789e-02 -1.99578598e-01 7.24392653e-01
-5.00333369e-01 2.44432926e-01 1.29693300e-01 -5.17249525e-01
1.05564915e-01 -8.64175037e-02 -3.71166974e-01 6.29266679e-01
4.02763218e-01 7.24160850e-01 -9.50171828e-01 -1.98663354e-01
-2.18179449e-02 2.68793881e-01 -4.53806907e-01 2.96554893e-01
-2.02653900e-01 2.04271391e-01 1.91169024e-01 4.35153618e-02
3.22555572e-01 -5.87460846e-02 -7.92882964e-02 3.93773377e-01
1.65703386e-01 8.06999803e-01 -7.89313376e-01 1.36861706e+00
-3.85553628e-01 1.28136528e+00 -4.64380264e-01 -3.24105144e-01
7.29882538e-01 4.98960942e-01 -2.76202857e-01 -7.74402022e-01
4.67211664e-01 3.58425491e-02 1.71862930e-01 -7.20403850e-01
6.38328254e-01 -5.22015452e-01 -2.54420251e-01 8.08289230e-01
2.37201527e-02 -1.75961122e-01 3.69009644e-01 3.05655718e-01
1.27255273e+00 7.32618198e-02 1.95796773e-01 2.89799105e-02
-7.84682333e-02 6.78428486e-02 1.00413665e-01 7.85902739e-01
3.52408327e-02 1.17787039e+00 8.82271767e-01 -2.50959337e-01
-1.33162236e+00 -8.94682169e-01 1.45675801e-02 1.19888330e+00
-3.82647663e-01 -5.93955636e-01 -1.01200104e+00 -5.39019167e-01
-2.33862355e-01 1.66910148e+00 -9.18393195e-01 -1.60565153e-01
-5.45949459e-01 -2.42968991e-01 6.80349827e-01 6.68724775e-01
1.27807278e-02 -8.96294713e-01 -8.13559234e-01 1.51992232e-01
-1.18217632e-01 -1.04935598e+00 -3.59579861e-01 4.09046769e-01
-4.00223047e-01 -1.01254344e+00 -2.54808009e-01 -8.25732470e-01
3.92851084e-01 -3.90130356e-02 1.05531955e+00 7.18770996e-02
2.71692783e-01 -2.77063072e-01 -7.09757447e-01 -2.13150546e-01
-6.70447469e-01 -1.96623251e-01 -1.00520350e-01 -2.02021167e-01
6.40564561e-01 -4.93643731e-01 -3.29136588e-02 -5.86813949e-02
-9.18157876e-01 3.33433539e-01 7.82426074e-02 7.58764744e-01
4.98034467e-04 3.38212661e-02 3.89780968e-01 -8.67923677e-01
1.03060246e+00 -6.28965139e-01 -7.02778623e-02 -1.24882542e-01
-1.47323862e-01 -2.67275479e-02 9.15718198e-01 -6.61365271e-01
-4.65982318e-01 -2.58994609e-01 -2.27674678e-01 -1.98943809e-01
-2.57723838e-01 7.27524400e-01 -1.33702323e-01 5.52659392e-01
1.12106287e+00 -1.13015942e-01 -1.76519006e-01 -4.84028727e-01
3.58325094e-01 6.76577806e-01 7.34105289e-01 -1.71827674e-01
7.76560843e-01 4.97976094e-02 -4.23446745e-01 -5.20648599e-01
-8.73057187e-01 -2.97830969e-01 -6.91024363e-01 -3.46751809e-01
6.49522245e-01 -7.83394516e-01 -4.18152034e-01 2.62979299e-01
-1.53305769e+00 -4.34962928e-01 -3.52165461e-01 5.11245430e-01
-6.88357651e-01 -1.51678160e-01 -6.64271414e-01 -6.59973562e-01
1.18717723e-01 -7.58275509e-01 6.00533485e-01 2.66889599e-03
-1.17207301e+00 -7.58604825e-01 1.96016073e-01 -2.68367957e-02
2.18635350e-01 4.83059078e-01 1.30854046e+00 -1.28640532e+00
1.29127055e-01 -7.45291889e-01 -1.63215995e-01 3.59564543e-01
-1.42584115e-01 1.48348778e-01 -1.33589053e+00 -1.26843676e-01
3.33144575e-01 -5.98480701e-01 6.62213922e-01 1.18439846e-01
3.58836621e-01 -3.10950756e-01 7.62869120e-02 3.09550256e-01
1.28992057e+00 9.07112062e-02 7.06833780e-01 4.61908549e-01
4.96907473e-01 6.05460167e-01 4.02944922e-01 3.71721983e-01
3.43505025e-01 5.81526577e-01 1.89517289e-01 2.18157336e-01
-2.70955563e-01 -5.94830811e-01 6.64527833e-01 5.08809865e-01
5.64500630e-01 -5.44316888e-01 -9.29661214e-01 6.25846088e-01
-1.56895506e+00 -1.18892586e+00 -3.48932534e-01 2.00824976e+00
9.23441887e-01 5.56078672e-01 2.08905339e-01 7.18688428e-01
5.50497413e-01 3.75299484e-01 -3.11842173e-01 -8.74977887e-01
-1.18309326e-01 2.91793048e-02 1.12617560e-01 6.94774866e-01
-1.04617083e+00 7.68177807e-01 7.50595140e+00 7.36024916e-01
-1.22640812e+00 2.81335413e-01 5.43604136e-01 -7.55049586e-01
-2.50400901e-01 -3.67737561e-02 -5.56414664e-01 3.69757921e-01
1.20878494e+00 -7.80072659e-02 5.20994604e-01 7.82101154e-01
5.09568691e-01 -5.84291995e-01 -1.84435225e+00 7.65485108e-01
5.27257264e-01 -1.17562151e+00 -3.41638029e-01 -3.57774138e-01
5.54733932e-01 -1.50665194e-01 1.21854722e-01 2.40953684e-01
1.90568089e-01 -1.39847898e+00 1.36654854e+00 1.40425652e-01
1.00763619e+00 -6.69815421e-01 6.65974915e-01 5.94831526e-01
-5.20460546e-01 -3.78993787e-02 -5.07469289e-02 -8.51349056e-01
8.29437897e-02 2.30614483e-01 -1.26700616e+00 -9.90094841e-02
3.72677505e-01 3.43655735e-01 -8.27862322e-01 9.89042044e-01
-5.95240295e-01 8.32895875e-01 -2.15025634e-01 -3.99731964e-01
2.93893963e-01 5.20633459e-01 5.48638284e-01 1.32143617e+00
-6.00977354e-02 6.25395775e-02 -1.77151725e-01 8.36704373e-01
5.89398704e-02 9.34538841e-02 -7.02799320e-01 -2.19573721e-01
2.13927194e-01 6.54910088e-01 -6.45461261e-01 -2.91440845e-01
-4.14383173e-01 7.58636475e-01 5.32847345e-01 2.90868968e-01
-7.40520835e-01 -5.13571799e-01 2.90710300e-01 4.44315672e-01
4.22237426e-01 -2.36893520e-01 -7.74536610e-01 -7.41979003e-01
2.56633043e-01 -7.50643551e-01 6.06666915e-02 -1.21663380e+00
-1.22781777e+00 7.93324649e-01 1.81259230e-01 -1.13644266e+00
-6.33273780e-01 -4.92204249e-01 -1.05012786e+00 1.17448878e+00
-1.02944922e+00 -9.49273407e-01 2.33012233e-02 1.70765534e-01
5.23964465e-01 1.03644818e-01 9.08223808e-01 1.91725083e-02
-1.69893548e-01 6.94512308e-01 1.48092777e-01 5.09626448e-01
4.88191187e-01 -1.13862944e+00 6.98202014e-01 8.95422637e-01
2.60591418e-01 4.35590774e-01 1.25623846e+00 -5.68060756e-01
-6.22357070e-01 -6.51070178e-01 1.46333098e+00 -7.87060142e-01
8.37800503e-01 -5.84530413e-01 -9.67698455e-01 6.73289001e-01
4.76128042e-01 -5.10041595e-01 8.65487397e-01 4.49858367e-01
-6.99485660e-01 5.33213139e-01 -9.94305015e-01 6.70617700e-01
6.10013068e-01 -9.45024669e-01 -1.17330480e+00 1.93989202e-01
7.55584717e-01 -5.79219341e-01 -2.07178131e-01 -2.42559895e-01
6.65408075e-01 -1.00825644e+00 1.62876472e-01 -9.16097999e-01
1.11043894e+00 -1.03115968e-01 -2.10771382e-01 -1.45707321e+00
-4.14956927e-01 -3.03413600e-01 4.14291732e-02 1.08375204e+00
4.56595570e-01 -2.90483907e-02 8.22300434e-01 6.67077184e-01
-1.35061607e-01 -8.57704580e-01 -9.78333056e-01 -6.43072844e-01
4.33492541e-01 -7.66494632e-01 3.99728745e-01 7.00491250e-01
5.06471038e-01 7.40812898e-01 -2.61115104e-01 -1.84725329e-01
-1.81126207e-01 -3.46640140e-01 5.66666186e-01 -1.03330123e+00
-4.96601045e-01 -5.74316502e-01 -3.19225013e-01 -7.70431161e-01
1.15747139e-01 -8.61400008e-01 2.42057279e-01 -1.56581306e+00
3.05511892e-01 9.27693397e-02 -7.67055303e-02 3.38517725e-01
-9.43062454e-02 2.65771031e-01 3.29679608e-01 -7.39040002e-02
-3.01682711e-01 8.32563788e-02 8.84454072e-01 -1.97191015e-02
-8.94012675e-02 -2.46894106e-01 -8.00296366e-01 7.25963831e-01
6.88024163e-01 -8.27322364e-01 -4.36384976e-01 -5.90251327e-01
3.98144692e-01 1.40358120e-01 4.75716352e-01 -9.98809278e-01
1.94022223e-01 5.72566390e-02 3.98949265e-01 -3.81651968e-01
4.51650470e-01 -4.40346152e-01 1.04882881e-01 1.92882735e-02
-7.80164361e-01 3.12355280e-01 3.26382309e-01 1.57870844e-01
-1.27939358e-01 -8.38865221e-01 7.12923646e-01 7.09082410e-02
-3.24840516e-01 -5.05263746e-01 -5.46972513e-01 2.76358694e-01
7.77640939e-01 -5.21870315e-01 -3.46243262e-01 -7.80459225e-01
-7.25622475e-01 -2.00640902e-01 7.06938505e-01 4.66929913e-01
6.27024531e-01 -1.30739677e+00 -1.05383503e+00 1.15317665e-01
1.35491401e-01 -2.63997912e-01 3.41336951e-02 5.49686968e-01
-5.75268686e-01 2.65725911e-01 -2.07928643e-02 -1.04599059e-01
-1.32975650e+00 5.49621403e-01 2.06599563e-01 -2.03182578e-01
-4.65321183e-01 1.07719481e+00 -8.62152949e-02 1.84631720e-01
2.20958501e-01 -2.86423653e-01 -3.72740775e-01 2.15169981e-01
8.63814831e-01 3.82271931e-02 1.54064432e-01 -9.47527945e-01
-2.25494951e-01 1.35779262e-01 -2.74953365e-01 -5.74633181e-01
1.18195558e+00 2.84415126e-01 2.69857615e-01 1.09449005e+00
1.45050085e+00 1.65183038e-01 -1.21419024e+00 -1.31152943e-01
1.88690387e-02 -5.48430324e-01 9.97735932e-02 -1.03444636e+00
-3.39814395e-01 9.49483573e-01 1.85223639e-01 4.59478557e-01
5.63540578e-01 1.11282773e-01 6.40717208e-01 2.21887946e-01
3.65536064e-02 -1.12233615e+00 1.22382671e-01 7.90244460e-01
1.17930365e+00 -9.30212677e-01 6.60846848e-03 -3.38271745e-02
-9.43168223e-01 1.17339778e+00 4.74607885e-01 -4.43562806e-01
3.63190293e-01 2.77177602e-01 1.86786190e-01 -1.19925523e-02
-8.85528386e-01 9.04972702e-02 7.02513456e-02 5.90744793e-01
6.09920084e-01 4.22368050e-02 -1.48336902e-01 9.54108894e-01
-7.75339484e-01 -1.98921531e-01 8.91990244e-01 8.89324546e-01
-4.15623158e-01 -5.97046018e-01 -2.73265839e-01 5.51083267e-01
-4.44959313e-01 -3.73319447e-01 -7.00313866e-01 8.52039337e-01
5.04995361e-02 1.19328058e+00 2.18347043e-01 -8.18950891e-01
4.91787314e-01 3.86457354e-01 4.52202916e-01 -9.48313713e-01
-9.17845368e-01 -2.16930881e-01 4.66469526e-01 -1.43508583e-01
2.22396716e-01 -8.96180153e-01 -1.29592073e+00 -6.71908736e-01
-3.29523683e-01 1.93072245e-01 5.17246723e-01 1.22718990e+00
-9.39999372e-02 4.42351818e-01 5.13587892e-01 -7.38258660e-01
-6.13405049e-01 -1.37365353e+00 -5.14765143e-01 4.97663617e-01
4.77668583e-01 -5.88806927e-01 -3.88261408e-01 1.78150665e-02] | [11.307680130004883, 8.829998016357422] |
e314dc09-b8a0-4776-80da-f296eea9d975 | hierarchical-region-learning-for-nested-named | null | null | https://aclanthology.org/2020.findings-emnlp.430 | https://aclanthology.org/2020.findings-emnlp.430.pdf | Hierarchical Region Learning for Nested Named Entity Recognition | Named Entity Recognition (NER) is deeply explored and widely used in various tasks. Usually, some entity mentions are nested in other entities, which leads to the nested NER problem. Leading region based models face both the efficiency and effectiveness challenge due to the high subsequence enumeration complexity. To tackle these challenges, we propose a hierarchical region learning framework to automatically generate a tree hierarchy of candidate regions with nearly linear complexity and incorporate structure information into the region representation for better classification. Experiments on benchmark datasets ACE-2005, GENIA and JNLPBA demonstrate competitive or better results than state-of-the-art baselines. | ['Yucheng Li', 'Shuzi Niu', 'Xinwei Long'] | 2020-11-01 | null | null | null | findings-of-the-association-for-computational | ['nested-named-entity-recognition'] | ['natural-language-processing'] | [-3.08092743e-01 -8.42160285e-02 -5.75577021e-01 -4.95930314e-01
-1.16229796e+00 -9.29262161e-01 2.82805294e-01 4.80067223e-01
-7.51584113e-01 9.60498095e-01 6.24536693e-01 -3.01171213e-01
2.34991893e-01 -8.05026948e-01 -7.03722060e-01 -2.20971227e-01
-1.22491486e-01 2.00297445e-01 4.81155634e-01 -9.79469940e-02
2.56985247e-01 5.72946906e-01 -8.38944852e-01 4.47766930e-01
1.21238434e+00 6.55605376e-01 -7.41835535e-02 2.60465205e-01
-6.96145594e-01 8.40651214e-01 -6.95942283e-01 -6.54527009e-01
9.49771106e-02 -9.79774371e-02 -1.09098482e+00 -3.77323657e-01
1.69485852e-01 -4.72224690e-02 -3.48873228e-01 8.58551264e-01
3.82790118e-01 5.58630645e-01 6.59142852e-01 -8.05538177e-01
-6.21133983e-01 1.06492782e+00 -6.68249369e-01 4.36445415e-01
4.39456999e-01 -5.33239543e-01 1.36543715e+00 -9.95761931e-01
1.02433693e+00 9.56021726e-01 7.25868464e-01 3.70630890e-01
-9.17524934e-01 -6.89923406e-01 5.28605282e-01 2.38618687e-01
-1.70731771e+00 -9.43932831e-02 3.29219639e-01 -1.43892141e-02
1.24177468e+00 1.18233502e-01 1.44348010e-01 9.94307458e-01
1.56349950e-02 9.87602830e-01 7.66963482e-01 -1.87468618e-01
2.68342167e-01 1.49861258e-02 4.07295108e-01 4.81688231e-01
2.63332218e-01 -1.16982706e-01 -7.74170533e-02 -3.36819619e-01
6.37772858e-01 -1.40636086e-01 -1.24761336e-01 -1.53577432e-01
-1.02281570e+00 8.19148660e-01 7.34478772e-01 4.55069453e-01
-4.56745088e-01 -2.66517639e-01 6.36608899e-01 4.05690596e-02
4.60227072e-01 4.95913357e-01 -8.55871916e-01 -9.08160768e-03
-1.01647282e+00 2.02239156e-01 8.48254800e-01 1.21725249e+00
5.37990391e-01 -3.53085786e-01 -3.46197307e-01 9.26680386e-01
1.44776151e-01 -9.38641652e-02 1.23632751e-01 -1.92633182e-01
1.11154783e+00 1.00006497e+00 2.35250413e-01 -6.97072923e-01
-6.24878943e-01 -6.40686750e-01 -9.35307264e-01 -5.27879894e-01
3.21101129e-01 -2.12495431e-01 -9.64369714e-01 1.49876750e+00
7.41694152e-01 3.65736425e-01 1.88770309e-01 6.92865431e-01
9.78037775e-01 9.90680218e-01 4.65585113e-01 -5.23376390e-02
1.61285865e+00 -1.10044968e+00 -6.27959728e-01 -2.58743972e-01
9.76756930e-01 -3.56076598e-01 5.40231407e-01 -6.97432905e-02
-6.89093888e-01 -3.52245867e-01 -8.46056759e-01 -2.94304579e-01
-6.57395482e-01 4.23444241e-01 5.78792393e-01 4.79351789e-01
-4.19845879e-01 3.58281195e-01 -7.77514040e-01 -2.07089514e-01
4.35697526e-01 3.12634170e-01 -6.63387001e-01 -1.74487293e-01
-1.48532569e+00 9.30199981e-01 8.91802728e-01 3.09146613e-01
-3.62015158e-01 -1.07858932e+00 -1.22014177e+00 1.25302359e-01
5.27341902e-01 -3.01004082e-01 1.06883919e+00 -1.52759954e-01
-9.22865570e-01 6.28495514e-01 -3.45757812e-01 -6.20753825e-01
3.76743704e-01 -4.65318263e-01 -7.68797457e-01 -9.26601961e-02
3.69992048e-01 4.96312499e-01 -4.56800535e-02 -8.09448838e-01
-7.69701481e-01 -3.19745004e-01 6.13179319e-02 2.03479305e-01
-6.62197545e-02 5.08453488e-01 -5.53093970e-01 -7.38660395e-01
7.59546161e-02 -6.96582317e-01 -7.42493987e-01 -6.76228404e-01
-6.34407878e-01 -7.11395979e-01 5.71671963e-01 -8.53879392e-01
1.68286729e+00 -1.98346806e+00 -1.46759614e-01 1.40264541e-01
2.39985227e-03 2.51608253e-01 -1.00237995e-01 4.15145934e-01
-2.11046010e-01 4.67478156e-01 -8.77619311e-02 1.35843143e-01
-1.18610509e-01 -1.14786997e-01 -4.45807546e-01 4.73115236e-01
3.64276737e-01 9.60365534e-01 -9.09761131e-01 -7.93382466e-01
-1.67922154e-01 2.81813234e-01 -3.31159949e-01 1.32623464e-01
-2.40065213e-02 2.55935252e-01 -6.38983846e-01 7.01845944e-01
7.42354989e-01 -1.58762902e-01 1.77246168e-01 -2.19288886e-01
-4.08326715e-01 7.76745081e-01 -1.30544603e+00 1.52058244e+00
-3.37956965e-01 1.13221936e-01 -1.18722178e-01 -8.11068475e-01
9.92559135e-01 4.98047054e-01 1.19352989e-01 -4.53522295e-01
-9.65701044e-02 2.44424388e-01 -3.06184113e-01 -1.05131470e-01
8.42736661e-01 -1.46044157e-02 -6.84423864e-01 1.09552786e-01
4.82279137e-02 6.45387828e-01 2.87698537e-01 1.15660667e-01
1.24610043e+00 2.10965127e-01 8.85891140e-01 -1.73551023e-01
5.28364360e-01 2.61448234e-01 1.21996379e+00 6.60306394e-01
-3.16072106e-01 4.11668867e-01 4.39208895e-01 -6.08275473e-01
-9.33074415e-01 -1.03080070e+00 -6.50345832e-02 1.29653573e+00
1.76013455e-01 -4.46189165e-01 -5.65156639e-01 -1.30040503e+00
-3.01547289e-01 7.85503864e-01 -6.68863356e-01 2.35234737e-01
-1.07226074e+00 -4.79511678e-01 9.68694150e-01 9.87907827e-01
6.43175960e-01 -1.18895233e+00 -1.70551419e-01 4.61246997e-01
-3.92516583e-01 -1.37757206e+00 -8.62377405e-01 2.43094921e-01
-9.41586137e-01 -7.98749626e-01 -9.29835021e-01 -1.13588202e+00
5.52208662e-01 -5.29242083e-02 1.22908938e+00 -3.32350194e-01
-2.83365041e-01 -4.97051269e-01 -4.89051610e-01 -2.61893451e-01
-7.68583342e-02 8.09884369e-01 -3.50260466e-01 -5.19187391e-01
4.07039016e-01 -8.81696194e-02 -4.33172047e-01 4.26613539e-01
-6.71857536e-01 -3.49335998e-01 8.88903558e-01 7.28695035e-01
7.13378668e-01 9.28211808e-02 7.66020000e-01 -1.22908294e+00
2.19558075e-01 -7.54409373e-01 -7.21949995e-01 7.11297214e-01
-2.43517593e-01 3.64947110e-01 7.14283407e-01 -2.14050546e-01
-1.44942939e+00 2.89241105e-01 -1.72438502e-01 -7.89128989e-02
-3.26406896e-01 7.44638979e-01 -4.72521752e-01 4.79583383e-01
6.00588322e-01 2.19696596e-01 -1.09713256e+00 -5.82777560e-01
5.74923694e-01 5.76107860e-01 6.32494152e-01 -6.70124114e-01
6.86347961e-01 1.74140364e-01 -3.74484122e-01 -5.63717008e-01
-1.19983912e+00 -1.18768954e+00 -8.18219900e-01 2.15438992e-01
8.31090689e-01 -1.24023807e+00 -1.44550130e-01 -6.71175197e-02
-1.44742978e+00 1.80729121e-01 5.44611402e-02 4.54887271e-01
1.01029001e-01 4.06182438e-01 -9.08407450e-01 -8.01989615e-01
-6.03251040e-01 -6.97509944e-01 1.02685428e+00 5.92192173e-01
-1.57401040e-01 -8.23488116e-01 2.39816979e-01 2.83832848e-01
-7.74174780e-02 1.41467080e-01 8.33386302e-01 -1.09754419e+00
-5.38915813e-01 -1.78525314e-01 -4.47847664e-01 -3.03407937e-01
-1.18973441e-01 -3.11530650e-01 -5.00630975e-01 -1.06165912e-02
-5.82053602e-01 -2.80609578e-02 8.86848271e-01 -2.96705775e-03
8.40587854e-01 -4.17908013e-01 -6.98431194e-01 4.43974137e-01
1.52946723e+00 2.23782405e-01 6.54054463e-01 4.48357880e-01
6.23935223e-01 6.58623278e-01 1.04989672e+00 3.85151476e-01
4.82180595e-01 6.08952165e-01 -1.34583220e-01 -5.03403209e-02
3.07921946e-01 -5.22675931e-01 2.43286133e-01 6.44873023e-01
1.88875020e-01 -2.76888520e-01 -1.03971457e+00 1.03712046e+00
-1.76255107e+00 -1.13064170e+00 -1.28280029e-01 1.87849748e+00
1.03210545e+00 -7.36719891e-02 1.78972065e-01 -3.84041071e-01
1.12842607e+00 2.09121644e-01 -4.82086271e-01 -5.18096626e-01
-1.29507899e-01 1.27599269e-01 7.34306812e-01 -6.91767856e-02
-1.59611189e+00 1.33971632e+00 6.32658911e+00 8.95196617e-01
-7.09946275e-01 -2.93594282e-02 8.05143237e-01 3.87056589e-01
-8.79406780e-02 -1.69292316e-02 -1.42173398e+00 6.63963929e-02
1.10420763e+00 -1.16285212e-01 -4.72175218e-02 1.06728303e+00
-1.65981445e-02 2.36892581e-01 -8.55227232e-01 7.46312439e-01
-2.53560662e-01 -1.31286347e+00 -1.03978306e-01 9.18162689e-02
1.04070389e+00 2.37201706e-01 -4.56559986e-01 7.85462856e-01
4.82841194e-01 -9.60070848e-01 5.24920583e-01 5.29491492e-02
5.13714492e-01 -1.06855452e+00 1.00270200e+00 3.98547143e-01
-2.00382733e+00 2.63117477e-02 -6.30567133e-01 4.43242639e-01
3.96522492e-01 5.96703351e-01 -9.81832683e-01 9.56917405e-01
6.79546177e-01 3.70981216e-01 -2.96071410e-01 1.42696810e+00
-4.12869126e-01 7.69582212e-01 -4.36449617e-01 -1.46551400e-01
4.44853663e-01 -1.53711036e-01 4.25937891e-01 1.67027748e+00
1.31763503e-01 2.61946589e-01 2.32679889e-01 8.81192505e-01
-6.76758647e-01 6.62717998e-01 -5.64332426e-01 -3.97650413e-02
8.73575568e-01 1.29133070e+00 -9.76279497e-01 -2.31981933e-01
-5.64792633e-01 1.12266028e+00 8.54723573e-01 1.81005090e-01
-1.14646065e+00 -8.33936453e-01 4.54139560e-01 -3.99345100e-01
8.41923535e-01 -1.42791271e-01 -3.27921420e-01 -1.25644720e+00
1.16257086e-01 -6.08119845e-01 8.90871584e-01 -1.45461395e-01
-1.36271799e+00 6.63981557e-01 -6.51617646e-02 -1.14101827e+00
2.87712272e-02 -3.42796683e-01 -6.39537215e-01 6.35864615e-01
-1.43141997e+00 -1.05844712e+00 1.79148868e-01 2.12995529e-01
4.97366130e-01 1.88425645e-01 6.43071830e-01 4.53584701e-01
-7.97482550e-01 9.41695929e-01 8.79815221e-02 1.02716231e+00
7.20291495e-01 -1.35885155e+00 8.46070528e-01 1.01521599e+00
4.37950194e-01 9.37814832e-01 3.22953582e-01 -7.72952557e-01
-1.00276828e+00 -1.54715037e+00 1.42454672e+00 -4.35446292e-01
4.59337741e-01 -5.06697536e-01 -9.36739743e-01 6.83920085e-01
6.33424297e-02 2.60163635e-01 8.07762086e-01 5.64869940e-01
-8.31968486e-01 1.64082527e-01 -1.06054270e+00 7.21662104e-01
1.02482021e+00 -5.48390031e-01 -1.08869827e+00 -7.67470300e-02
8.91480625e-01 -4.06765491e-01 -1.20115566e+00 4.39301819e-01
3.79445165e-01 -4.49282765e-01 8.77046883e-01 -8.13551366e-01
2.99009662e-02 -4.30049330e-01 7.57148191e-02 -1.07467711e+00
-2.12411001e-01 -4.54233974e-01 -1.85817540e-01 1.83595145e+00
1.04261351e+00 -3.42702687e-01 8.51506114e-01 6.20654404e-01
-6.95023220e-03 -6.62964761e-01 -1.04528880e+00 -8.42147470e-01
1.43220454e-01 -1.60856143e-01 6.99936032e-01 9.93275285e-01
1.45744801e-01 7.27187455e-01 -1.09601907e-01 4.94678617e-01
3.69706690e-01 2.35730842e-01 2.80882210e-01 -1.04342818e+00
2.20992967e-01 -2.78861642e-01 -3.12045634e-01 -1.24858510e+00
3.68536919e-01 -7.78009057e-01 3.78268510e-01 -1.73294282e+00
3.08391660e-01 -5.22251129e-01 -4.97564256e-01 6.57662511e-01
-5.34181297e-01 -1.88858002e-01 -1.72882844e-02 6.05916530e-02
-1.19723868e+00 5.30636549e-01 6.56324327e-01 -1.98050844e-03
-5.02321601e-01 -1.25767499e-01 -6.68033719e-01 4.47880417e-01
6.79884672e-01 -7.08210468e-01 -1.31742340e-02 -1.68474168e-01
3.97543699e-01 2.05139533e-01 -3.17871094e-01 -8.36022139e-01
3.87023211e-01 -7.89193660e-02 3.38378221e-01 -1.23759604e+00
-2.10152045e-01 -5.87310553e-01 -1.37001067e-01 8.35845247e-02
-5.22108257e-01 2.94164598e-01 2.37841025e-01 7.89930880e-01
-4.58157688e-01 -2.35170439e-01 6.44745409e-01 -1.28351739e-02
-1.01004767e+00 3.64208072e-01 -2.33294219e-01 4.96307611e-01
1.10648298e+00 1.71341877e-02 -2.45959193e-01 4.59831767e-02
-4.16122824e-01 3.86503935e-01 9.14954543e-02 5.31552851e-01
4.62458879e-01 -1.19712842e+00 -8.92805874e-01 -3.08532596e-01
3.71965796e-01 2.84012586e-01 3.43935698e-01 6.19767606e-01
-5.67673802e-01 6.41013205e-01 3.29456627e-01 -1.83698744e-01
-1.20838535e+00 6.28222585e-01 5.72155714e-02 -1.03669965e+00
-6.62602544e-01 8.32809567e-01 2.83155262e-01 -9.45846021e-01
4.01618779e-02 -4.33237851e-01 -5.44376493e-01 2.12049093e-02
6.44185841e-01 4.28233862e-01 7.91218728e-02 -6.11041605e-01
-5.93153536e-01 1.89958677e-01 -4.86795783e-01 1.52246267e-01
1.14608264e+00 -2.53537390e-03 3.74435112e-02 1.97723970e-01
1.17754865e+00 2.03149572e-01 -7.10650921e-01 -2.86541820e-01
9.49692965e-01 1.43198907e-01 -1.37244508e-01 -7.94833541e-01
-7.51726568e-01 4.21888292e-01 7.48367757e-02 -2.34039009e-01
8.22524667e-01 2.13644400e-01 1.03264475e+00 6.36411071e-01
5.58001041e-01 -1.10909164e+00 -6.26795352e-01 8.80178034e-01
3.99594903e-01 -1.04133070e+00 1.44249603e-01 -6.98452771e-01
-7.97779679e-01 8.45854223e-01 7.82604694e-01 -1.52020186e-01
6.29867435e-01 3.24387550e-01 -3.77868898e-02 9.01389793e-02
-8.01623523e-01 -1.92605093e-01 4.65650916e-01 2.34492525e-01
8.09254766e-01 -8.33161548e-02 -6.11223936e-01 1.03314722e+00
1.55407280e-01 -3.03075284e-01 -1.21609094e-02 9.27894354e-01
-3.16335618e-01 -1.19319820e+00 -3.11583847e-01 3.22871119e-01
-1.06290555e+00 -3.25070560e-01 -3.64322960e-01 8.82588804e-01
-8.47068280e-02 9.07265067e-01 -1.30689144e-01 -9.13435072e-02
4.33302194e-01 1.15466490e-01 3.14877294e-02 -6.97563112e-01
-1.10335374e+00 -1.70773845e-02 4.28119302e-01 -4.52152312e-01
-2.42618740e-01 -7.11586833e-01 -1.77954471e+00 2.31947705e-01
-7.11725771e-01 7.93385029e-01 4.27964061e-01 1.00527108e+00
6.49012685e-01 3.07144403e-01 5.77468097e-01 -6.46620840e-02
-6.08239710e-01 -1.01143432e+00 -4.68070954e-01 2.46056035e-01
-1.42780691e-01 -4.30640310e-01 2.52113882e-02 -3.78828764e-01] | [9.54858684539795, 9.505696296691895] |
573df1db-0386-43b5-953b-1c183ad79442 | a-closer-look-at-temporal-ordering-in-the | 2209.15501 | null | https://arxiv.org/abs/2209.15501v2 | https://arxiv.org/pdf/2209.15501v2.pdf | A Closer Look at Temporal Ordering in the Segmentation of Instructional Videos | Understanding the steps required to perform a task is an important skill for AI systems. Learning these steps from instructional videos involves two subproblems: (i) identifying the temporal boundary of sequentially occurring segments and (ii) summarizing these steps in natural language. We refer to this task as Procedure Segmentation and Summarization (PSS). In this paper, we take a closer look at PSS and propose three fundamental improvements over current methods. The segmentation task is critical, as generating a correct summary requires each step of the procedure to be correctly identified. However, current segmentation metrics often overestimate the segmentation quality because they do not consider the temporal order of segments. In our first contribution, we propose a new segmentation metric that takes into account the order of segments, giving a more reliable measure of the accuracy of a given predicted segmentation. Current PSS methods are typically trained by proposing segments, matching them with the ground truth and computing a loss. However, much like segmentation metrics, existing matching algorithms do not consider the temporal order of the mapping between candidate segments and the ground truth. In our second contribution, we propose a matching algorithm that constrains the temporal order of segment mapping, and is also differentiable. Lastly, we introduce multi-modal feature training for PSS, which further improves segmentation. We evaluate our approach on two instructional video datasets (YouCook2 and Tasty) and observe an improvement over the state-of-the-art of $\sim7\%$ and $\sim2.5\%$ for procedure segmentation and summarization, respectively. | ['Shreyank N Gowda', 'Frank Keller', 'Laura Sevilla-Lara', 'Anil Batra'] | 2022-09-30 | null | null | null | null | ['dense-video-captioning'] | ['computer-vision'] | [ 5.78052342e-01 2.03063134e-02 -4.20724809e-01 -4.29913849e-01
-9.48396623e-01 -8.33767056e-01 5.01614571e-01 5.04620850e-01
-4.23042774e-01 4.37581688e-01 1.83028176e-01 -1.77626297e-01
-1.72176078e-01 -6.47162974e-01 -1.05435920e+00 -3.99216592e-01
1.58207253e-01 4.19475943e-01 5.76089025e-01 5.67871854e-02
5.25011003e-01 3.47315669e-01 -1.62778986e+00 3.50963026e-01
1.22477078e+00 9.16321099e-01 1.66283727e-01 7.62698472e-01
-1.94965944e-01 6.37558997e-01 -6.15185380e-01 -3.10929894e-01
1.57749489e-01 -8.14942658e-01 -1.30256474e+00 3.18961263e-01
7.89706349e-01 -1.35589570e-01 -6.56901598e-02 9.99794185e-01
3.62898484e-02 5.01440108e-01 7.80406356e-01 -1.10672665e+00
-4.23082747e-02 7.81481028e-01 -4.42179114e-01 2.09483951e-01
7.30187654e-01 7.62554780e-02 1.16493559e+00 -3.97053093e-01
7.70249486e-01 7.86173165e-01 5.67496836e-01 5.88651359e-01
-1.13479865e+00 -3.42715830e-01 5.26658654e-01 2.47828707e-01
-9.79380310e-01 -2.42112622e-01 5.62664509e-01 -7.43553758e-01
6.07877314e-01 3.60870719e-01 8.44537199e-01 6.74650550e-01
-9.15309042e-02 1.36427534e+00 9.41412508e-01 -4.10729289e-01
1.99997947e-01 -1.15746446e-01 4.70551521e-01 8.06129694e-01
-1.51807010e-01 -1.06811546e-01 -5.77641308e-01 4.93629247e-01
5.24875641e-01 -2.40807161e-01 -3.76550317e-01 -3.67749155e-01
-1.25484836e+00 6.28792763e-01 1.61670223e-01 3.63269597e-01
-2.27030024e-01 1.81780562e-01 4.08857763e-01 1.40715733e-01
2.15137869e-01 6.91538990e-01 -1.86261997e-01 -6.17347538e-01
-1.54166842e+00 3.72628540e-01 7.65848219e-01 1.12895441e+00
7.10675359e-01 -2.93875813e-01 -3.87539893e-01 5.07245600e-01
-2.20821146e-02 -2.83416547e-02 3.66032660e-01 -1.38906384e+00
4.43606198e-01 6.55865490e-01 4.09497768e-02 -7.83639908e-01
-2.40737811e-01 -2.38273814e-01 -2.35367507e-01 -6.31065816e-02
8.49414706e-01 5.44167049e-02 -8.72164428e-01 1.60139394e+00
2.63669640e-01 3.98024887e-01 -9.30725113e-02 7.82618821e-01
8.46953750e-01 7.95273244e-01 1.31032225e-02 -3.25495690e-01
1.13892329e+00 -1.31351340e+00 -6.95357740e-01 -1.63862169e-01
6.81349456e-01 -7.62048304e-01 1.08815074e+00 5.10789633e-01
-1.56206179e+00 -6.47403240e-01 -8.35579991e-01 2.27794889e-03
-1.80068359e-01 5.45394979e-02 4.70679015e-01 2.20538899e-01
-7.77631342e-01 1.13901901e+00 -9.82024729e-01 -2.43511349e-01
3.06268990e-01 3.24336141e-01 -2.08051726e-01 1.53029375e-02
-8.36563885e-01 6.73448145e-01 3.09902400e-01 -1.66054294e-01
-7.95648456e-01 -9.82294321e-01 -1.00959015e+00 1.48750722e-01
6.84785247e-01 -4.05354232e-01 1.54948187e+00 -1.11064124e+00
-1.52288461e+00 9.09946442e-01 -3.51922929e-01 -5.32007635e-01
7.14982271e-01 -3.80805641e-01 6.03239834e-02 4.60226804e-01
1.64508045e-01 9.16859150e-01 5.81914842e-01 -1.17970574e+00
-1.02323937e+00 -1.15962010e-02 2.73345143e-01 3.14843655e-01
1.99816693e-02 -2.13134617e-01 -7.23111570e-01 -4.68996793e-01
1.17103405e-01 -9.58308101e-01 -2.20822498e-01 -1.23438276e-01
-4.42150325e-01 -5.14560521e-01 4.11698014e-01 -7.04687953e-01
1.67094147e+00 -2.07823563e+00 5.00476539e-01 5.78483101e-03
9.11670998e-02 3.81295919e-01 -5.03874733e-04 2.92911589e-01
9.69915017e-02 1.49647027e-01 -4.81491387e-01 -4.38756019e-01
1.22315526e-01 3.08203016e-04 -1.06204741e-01 2.68417954e-01
-2.58678291e-02 7.97074974e-01 -1.03573418e+00 -6.32696807e-01
1.65443927e-01 7.13607445e-02 -6.18576229e-01 2.02873707e-01
-3.20240617e-01 3.78252417e-01 -2.38481224e-01 3.29724640e-01
2.31172115e-01 -3.65808420e-02 -1.47741288e-01 -2.23220587e-01
-3.84727240e-01 4.09006447e-01 -1.24663675e+00 1.96244693e+00
-2.83658087e-01 7.88268387e-01 -2.49596700e-01 -1.23753762e+00
6.08641207e-01 2.66259402e-01 7.30297863e-01 -4.97188896e-01
3.75495143e-02 1.04495414e-01 -1.99984819e-01 -6.19498134e-01
6.65008962e-01 1.37372419e-01 -2.09466562e-01 3.61625016e-01
1.21891245e-01 -5.02531707e-01 8.32588613e-01 3.06670159e-01
8.49748194e-01 6.28724217e-01 2.70475835e-01 -2.39458501e-01
4.98782486e-01 2.57306576e-01 6.07520998e-01 7.14798272e-01
-3.71420383e-01 7.66346633e-01 7.70498574e-01 -2.56595224e-01
-7.36909211e-01 -9.25256193e-01 1.25340417e-01 9.21275496e-01
3.70474875e-01 -4.97111648e-01 -1.25134516e+00 -1.05705559e+00
-2.13011816e-01 8.38841736e-01 -5.68116486e-01 -1.10651530e-01
-8.18222165e-01 -2.35322136e-02 3.08740556e-01 6.40616536e-01
3.77591610e-01 -1.01922619e+00 -8.87215078e-01 1.34075284e-01
-4.58842635e-01 -1.14451563e+00 -1.02035105e+00 -4.73506674e-02
-9.17915940e-01 -1.27955461e+00 -5.51172376e-01 -9.61246371e-01
7.25209415e-01 1.57158822e-01 1.07275546e+00 5.41605093e-02
-2.53417026e-02 5.41627884e-01 -3.09181899e-01 -1.86118737e-01
-3.88895273e-01 1.24897532e-01 -1.75306037e-01 -1.19151458e-01
1.90041661e-01 -2.57879645e-01 -5.97803831e-01 2.53260434e-01
-9.58835721e-01 3.89908761e-01 4.77582723e-01 5.08480251e-01
8.61249864e-01 -8.15220997e-02 3.07834297e-01 -1.11115444e+00
4.42175448e-01 -1.15054816e-01 -6.03633106e-01 4.11038816e-01
-4.84256834e-01 1.20040089e-01 7.85712421e-01 -3.95269632e-01
-8.35010529e-01 2.94148862e-01 -1.02408446e-01 -4.34307486e-01
-4.43687439e-01 6.72440052e-01 1.15269693e-02 1.72660321e-01
3.41717690e-01 2.90727288e-01 -1.80137545e-01 -3.21669847e-01
2.78270215e-01 1.86190590e-01 6.16379440e-01 -6.89826548e-01
4.62601602e-01 1.06822200e-01 8.23756959e-03 -7.26747751e-01
-1.04546988e+00 -8.31123531e-01 -9.06768620e-01 -5.14976382e-01
9.23195481e-01 -4.89706039e-01 -8.09031665e-01 2.53932327e-01
-9.29446101e-01 -6.00228846e-01 -6.08931839e-01 4.87615794e-01
-9.26541269e-01 6.04399681e-01 -5.18603861e-01 -3.76549661e-01
-2.19457652e-02 -1.45054436e+00 9.52169299e-01 4.61198598e-01
-4.44532543e-01 -9.57883000e-01 -4.57617007e-02 6.64971888e-01
-1.09206242e-02 3.81023049e-01 7.79575109e-01 -6.33334100e-01
-6.18707597e-01 -1.79597184e-01 5.22713065e-02 3.16207916e-01
-2.47327369e-02 3.98252279e-01 -5.97346365e-01 -5.43896982e-04
-3.33248734e-01 -7.81917274e-02 9.07700539e-01 6.51525557e-01
1.14584982e+00 -3.28689873e-01 -2.59138763e-01 4.91430968e-01
1.24552345e+00 3.65021020e-01 5.06225288e-01 2.47317433e-01
6.35568678e-01 9.47459579e-01 9.72481191e-01 1.22496285e-01
4.05675560e-01 7.62105703e-01 1.64190441e-01 2.00764149e-01
-3.08211952e-01 -4.20808464e-01 4.74051833e-01 9.91757035e-01
-3.73169445e-02 -2.64053732e-01 -8.95907879e-01 8.88335586e-01
-1.96474588e+00 -1.02473605e+00 -2.34945342e-01 2.24236012e+00
1.03434181e+00 3.85473192e-01 4.28717405e-01 3.16460341e-01
5.29412031e-01 9.27712098e-02 -3.42674077e-01 -6.11443818e-01
3.67672473e-01 3.38017166e-01 1.60568774e-01 6.91175044e-01
-1.35999966e+00 1.04161894e+00 5.69763756e+00 9.16333437e-01
-1.01610589e+00 -3.41303885e-01 4.49369669e-01 -2.58196685e-02
-2.30686873e-01 1.80178843e-02 -8.18623662e-01 5.52955508e-01
8.64047348e-01 -2.42105097e-01 3.51145178e-01 5.66106141e-01
2.80817777e-01 -3.90113205e-01 -1.57544243e+00 6.06851041e-01
1.76849082e-01 -1.39145875e+00 -6.26972318e-02 -3.82670581e-01
9.68185842e-01 -5.41159928e-01 3.07436120e-02 2.60460079e-01
1.07498854e-01 -9.57321942e-01 9.49811995e-01 5.95862448e-01
4.69084114e-01 -8.31628561e-01 4.83727425e-01 3.86508197e-01
-1.27901828e+00 2.53088236e-01 2.33705282e-01 9.63885561e-02
1.77087247e-01 3.81760299e-01 -7.15411961e-01 5.89134812e-01
5.01066685e-01 9.72447097e-01 -5.26563525e-01 1.41021800e+00
-5.39324701e-01 7.73310065e-01 -9.55848694e-02 5.27086668e-02
4.69667971e-01 -3.06052059e-01 6.22173786e-01 1.37709892e+00
8.86116102e-02 1.65938139e-01 4.69028562e-01 8.33499014e-01
3.94738046e-03 -9.71601252e-03 -3.67415220e-01 -1.06805265e-01
3.37009728e-01 1.02594078e+00 -9.92389858e-01 -4.19290692e-01
-3.20593953e-01 9.11368549e-01 6.89573884e-02 2.43977353e-01
-8.93039525e-01 -4.19178218e-01 5.16962469e-01 5.38724624e-02
3.54702115e-01 -1.68439493e-01 -3.64752740e-01 -1.00810599e+00
1.75912213e-02 -8.01908553e-01 3.03753644e-01 -3.79208267e-01
-6.87854111e-01 2.09983036e-01 1.46495536e-01 -1.35270953e+00
-2.41256669e-01 -1.83926642e-01 -6.64513052e-01 3.94060552e-01
-1.49000800e+00 -7.87435770e-01 -2.38019139e-01 1.49346039e-01
1.11495829e+00 4.66023415e-01 4.65828627e-01 2.07353279e-01
-6.97139680e-01 5.84126294e-01 -1.73511535e-01 7.14573860e-02
5.47652125e-01 -1.55027771e+00 3.21691453e-01 1.17066658e+00
1.78130627e-01 4.63754028e-01 8.52626324e-01 -4.78358150e-01
-1.04595935e+00 -1.02406394e+00 1.04401934e+00 -2.64375865e-01
3.87356013e-01 -2.59007085e-02 -9.49359298e-01 7.70775676e-01
9.74999443e-02 -3.29469293e-01 6.77481055e-01 -2.02859879e-01
6.17557801e-02 4.57181484e-02 -9.06674325e-01 6.62017703e-01
1.04749167e+00 -1.93268806e-01 -7.23289073e-01 2.12870866e-01
7.23273516e-01 -6.09941721e-01 -9.52110469e-01 2.83977419e-01
5.43548167e-01 -1.01112974e+00 7.54771113e-01 -6.11091495e-01
8.39068890e-01 -2.68676430e-01 2.04958797e-01 -1.17168832e+00
-4.94315363e-02 -6.91285908e-01 -9.37745944e-02 1.19638979e+00
4.04568851e-01 1.15401119e-01 1.01976967e+00 6.26661539e-01
-6.28265083e-01 -9.97699261e-01 -6.14896357e-01 -8.27883601e-01
-4.33941744e-02 -3.53685528e-01 2.69469559e-01 9.40247536e-01
2.17534482e-01 2.96919215e-02 3.13566960e-02 -6.58669695e-02
5.27677417e-01 3.98761690e-01 6.04727983e-01 -1.10439384e+00
-7.73587972e-02 -7.96389759e-01 -2.40584537e-01 -1.41844881e+00
3.74869883e-01 -7.87025571e-01 4.21522886e-01 -1.85641873e+00
8.11983272e-02 -2.28801355e-01 -2.28645094e-02 3.43509883e-01
-3.02382648e-01 -2.46943738e-02 3.72326553e-01 3.16517912e-02
-1.06500447e+00 2.03948393e-01 1.44239080e+00 -1.00635394e-01
-3.60576510e-01 3.17084253e-01 -3.39383632e-01 8.65470886e-01
5.32625854e-01 -3.11250716e-01 -3.05171758e-01 -4.39624697e-01
1.31880408e-02 3.43924433e-01 8.84955972e-02 -1.19307005e+00
5.13912737e-01 -4.28719103e-01 -8.91420841e-02 -6.40299559e-01
1.68361455e-01 -7.65384614e-01 -1.16053261e-01 4.93343413e-01
-7.40409374e-01 -1.14724226e-02 2.77743816e-01 3.66852552e-01
-4.22489375e-01 -6.27586424e-01 7.85516322e-01 -2.06185505e-01
-9.68904912e-01 9.66994911e-02 -4.53208327e-01 2.39864454e-01
1.30414104e+00 -5.64252198e-01 -5.56917787e-02 -3.04190367e-01
-8.50427210e-01 4.99979645e-01 5.13748229e-01 2.33774215e-01
6.32120669e-01 -1.06072831e+00 -4.18046206e-01 -3.22976671e-02
-1.33371696e-01 3.16195995e-01 1.62641689e-01 1.04784942e+00
-7.68128514e-01 4.16729748e-01 -6.24642745e-02 -6.25745654e-01
-1.41321814e+00 3.94815266e-01 1.48013189e-01 -4.05201316e-01
-3.16288322e-01 1.02471936e+00 6.57692924e-02 -2.20917583e-01
5.73056042e-01 -7.68991053e-01 -4.94198054e-01 3.83361250e-01
2.60223836e-01 6.19127333e-01 -5.80852963e-02 -5.30018151e-01
-1.06138043e-01 8.11599433e-01 -9.79402363e-02 -6.21571541e-02
1.20613992e+00 -1.38382884e-02 4.04465050e-02 6.11092985e-01
1.05883563e+00 -5.14218397e-02 -1.47370243e+00 -1.52592704e-01
2.42618322e-01 -3.39251697e-01 -1.89762592e-01 -7.41806149e-01
-9.15015519e-01 1.00461709e+00 7.84625188e-02 3.08084935e-01
1.12694454e+00 -1.85821280e-01 1.00698888e+00 9.12273489e-03
1.22872524e-01 -1.22665370e+00 2.67880112e-01 5.26516438e-01
4.36641067e-01 -1.17012632e+00 1.96325742e-02 -6.21641994e-01
-8.19222271e-01 1.18303621e+00 7.02800333e-01 -1.68897212e-02
1.74731836e-01 -1.39549822e-01 -1.96445838e-01 -1.14507355e-01
-5.33799708e-01 -2.58946210e-01 7.33853340e-01 2.81294465e-01
4.92286772e-01 7.07131624e-02 -5.01536906e-01 5.30381620e-01
-2.74021119e-01 1.04560122e-01 6.51717484e-01 1.04105437e+00
-5.81017017e-01 -9.91010249e-01 -4.04686630e-02 4.27907735e-01
-5.06009102e-01 3.67815346e-02 -4.41506892e-01 6.83025062e-01
6.27717823e-02 8.78488779e-01 1.16707921e-01 -2.21561432e-01
6.04839206e-01 7.09339902e-02 6.12865865e-01 -8.37045550e-01
-8.10132027e-01 -2.59173196e-02 -5.49780205e-02 -7.43543863e-01
-5.31191051e-01 -1.00264585e+00 -1.49408042e+00 -2.31990427e-01
-2.79516131e-01 4.54285264e-01 5.01173258e-01 1.04104149e+00
1.02352373e-01 7.74431825e-01 3.90323132e-01 -7.36205101e-01
-3.47397059e-01 -5.11624932e-01 -1.85642362e-01 6.72402203e-01
2.76478022e-01 -4.87448633e-01 -2.06203565e-01 4.65762645e-01] | [8.785028457641602, 0.5623630285263062] |
ec80d79b-eaa5-4f28-892a-c066b683382a | nisq-ready-community-detection-based-on | 2212.14717 | null | https://arxiv.org/abs/2212.14717v2 | https://arxiv.org/pdf/2212.14717v2.pdf | NISQ-ready community detection based on separation-node identification | The analysis of network structure is essential to many scientific areas, ranging from biology to sociology. As the computational task of clustering these networks into partitions, i.e., solving the community detection problem, is generally NP-hard, heuristic solutions are indispensable. The exploration of expedient heuristics has led to the development of particularly promising approaches in the emerging technology of quantum computing. Motivated by the substantial hardware demands for all established quantum community detection approaches, we introduce a novel QUBO based approach that only needs number-of-nodes many qubits and is represented by a QUBO-matrix as sparse as the input graph's adjacency matrix. The substantial improvement on the sparsity of the QUBO-matrix, which is typically very dense in related work, is achieved through the novel concept of separation-nodes. Instead of assigning every node to a community directly, this approach relies on the identification of a separation-node set, which -- upon its removal from the graph -- yields a set of connected components, representing the core components of the communities. Employing a greedy heuristic to assign the nodes from the separation-node sets to the identified community cores, subsequent experimental results yield a proof of concept. This work hence displays a promising approach to NISQ ready quantum community detection, catalyzing the application of quantum computers for the network structure analysis of large scale, real world problem instances. | ['Mirco Schoenfeld', 'David Bucher', 'Jonas Nüßlein', 'Sebastian Feld', 'Dominik Ott', 'Jonas Stein'] | 2022-12-30 | null | null | null | null | ['community-detection'] | ['graphs'] | [ 2.85905659e-01 2.93750405e-01 1.51031390e-01 2.42216632e-01
-3.59938622e-01 -7.14655161e-01 2.95896202e-01 5.26334405e-01
-2.34034166e-01 6.43187344e-01 -3.24843913e-01 -5.10998666e-01
-5.00631034e-01 -1.21525037e+00 -2.70762175e-01 -1.06862152e+00
-4.70448762e-01 8.97888839e-01 2.64862090e-01 -3.97545844e-01
5.66420376e-01 3.69089514e-01 -1.06596398e+00 -1.95746139e-01
6.65841699e-01 4.93094563e-01 -1.10772595e-01 6.58080816e-01
-1.61047578e-02 4.52175409e-01 -1.19871721e-01 -2.64064610e-01
3.11432838e-01 -7.59993374e-01 -9.79413450e-01 1.91443816e-01
-2.05668494e-01 4.33576792e-01 -2.25581959e-01 1.40554667e+00
2.09697947e-01 -6.73275366e-02 4.46916997e-01 -1.19676781e+00
-2.43101403e-01 7.57735610e-01 -6.30646884e-01 1.96610525e-01
4.98863429e-01 -1.65843830e-01 1.45018029e+00 -2.37311378e-01
9.44825232e-01 8.22039425e-01 4.86590385e-01 2.13918701e-01
-1.62467074e+00 -4.78711128e-01 -5.69689155e-01 3.37492287e-01
-1.79202318e+00 -2.41182104e-01 7.64452338e-01 -4.85007286e-01
7.40205228e-01 2.85379708e-01 8.82306278e-01 3.48511785e-01
4.10036743e-02 -1.37609810e-01 1.44173813e+00 -7.18693137e-01
5.06699920e-01 6.63450882e-02 3.85613710e-01 8.41099381e-01
7.48266399e-01 -3.27963904e-02 -5.00725567e-01 -3.17688495e-01
3.47299159e-01 -1.39516354e-01 -1.21094055e-01 -8.24673414e-01
-1.01400840e+00 9.48318481e-01 5.93749166e-01 6.97370410e-01
-2.97829270e-01 3.39863032e-01 3.85952115e-01 2.83613861e-01
5.33478037e-02 4.20181483e-01 2.89400071e-01 1.71046719e-01
-8.69339883e-01 -9.63889062e-03 1.10739458e+00 7.12065399e-01
1.16992569e+00 -6.18079066e-01 4.45386052e-01 -9.33413580e-02
2.01363489e-01 3.32162470e-01 -3.16827446e-01 -7.58207023e-01
8.32955912e-02 8.11611950e-01 -8.80458951e-02 -1.26359713e+00
-4.32825148e-01 -5.45567989e-01 -1.19094777e+00 -1.75640136e-01
3.64293188e-01 2.58004554e-02 -4.99117196e-01 1.56129885e+00
6.89335346e-01 -7.86514729e-02 -4.69467882e-03 7.98463702e-01
1.58544794e-01 3.92590940e-01 -3.96351397e-01 -3.72610182e-01
1.52822840e+00 -2.22112969e-01 -3.40850115e-01 3.42322797e-01
6.56081378e-01 -4.20931488e-01 1.74203232e-01 4.78326708e-01
-6.77215040e-01 7.06469491e-02 -9.90369618e-01 3.66229266e-01
-2.62676328e-01 -4.12251621e-01 9.27557707e-01 9.50952709e-01
-1.35119450e+00 5.73026538e-01 -6.07659698e-01 -7.17404842e-01
1.90370172e-01 6.53742075e-01 -5.47891915e-01 -2.42253810e-01
-8.96296561e-01 5.93158901e-01 6.28242016e-01 2.73637354e-01
-6.50627673e-01 1.54530719e-01 -3.61624271e-01 4.23666984e-01
5.67453563e-01 -7.52631783e-01 4.07891482e-01 -7.53508091e-01
-1.14327037e+00 9.29149806e-01 2.00397596e-02 -4.79048789e-01
-1.80861093e-02 8.71305048e-01 9.83050466e-03 6.80166245e-01
1.96653172e-01 1.93912983e-01 7.65812218e-01 -1.00602341e+00
-3.71861875e-01 -4.05883014e-01 1.62400872e-01 -1.43337205e-01
-3.88228178e-01 9.46855769e-02 -4.36649472e-02 2.43400469e-01
7.53269672e-01 -1.25397682e+00 -6.51662827e-01 -5.53345859e-01
-6.44112051e-01 -3.89481813e-01 2.93651521e-01 9.75881070e-02
1.27499926e+00 -2.03933525e+00 6.18246853e-01 1.03104496e+00
9.50373530e-01 -1.29351705e-01 1.09657170e-02 9.50125873e-01
-2.87413567e-01 2.29874253e-01 -2.37214565e-01 1.17307112e-01
1.67271979e-02 1.04492372e-02 5.73905893e-02 9.00581300e-01
5.36715277e-02 4.69619095e-01 -1.07453561e+00 -4.26351786e-01
1.05843864e-01 9.26346183e-02 -7.25853980e-01 -4.25344408e-01
4.09912132e-02 2.73425370e-01 -5.73345184e-01 3.51942003e-01
6.72708273e-01 -6.31050229e-01 8.44444633e-01 -1.50198452e-02
-3.24932635e-01 2.63855457e-01 -1.43868017e+00 1.23884702e+00
2.93226659e-01 5.30479193e-01 3.46819967e-01 -1.46659100e+00
7.42535889e-01 3.89583975e-01 7.30077922e-01 -3.94798934e-01
3.29134881e-01 3.82166326e-01 4.74941701e-01 -1.05101198e-01
4.03797448e-01 -4.21163738e-01 -2.74948686e-01 5.82061410e-01
9.24615189e-02 -2.08374555e-03 7.35300064e-01 7.89838612e-01
1.64961517e+00 -5.28731167e-01 2.64346242e-01 -6.53005660e-01
6.39743388e-01 2.85192877e-01 3.53375614e-01 7.35706806e-01
-4.08228576e-01 1.68837324e-01 9.79122341e-01 -8.26706365e-02
-1.09259379e+00 -8.59637439e-01 8.80325492e-03 3.93411726e-01
2.30545044e-01 -8.23106766e-01 -8.40185821e-01 -1.59180716e-01
-1.52996615e-01 -1.34528205e-01 -5.08450806e-01 -1.54097930e-01
-3.33835781e-01 -9.11625028e-01 2.72279173e-01 -3.38887215e-01
2.16275472e-02 -7.04247952e-01 -4.75445956e-01 4.09001768e-01
-1.94216132e-01 -1.06667244e+00 1.45054162e-01 5.30302763e-01
-9.02480960e-01 -1.28167856e+00 -1.79262102e-01 -5.99221408e-01
7.77872145e-01 4.65721130e-01 7.80619264e-01 4.88634735e-01
-4.40739959e-01 2.69710332e-01 -5.57620883e-01 2.16850057e-01
-4.42394525e-01 1.37642220e-01 2.65837908e-01 1.84773847e-01
3.20453465e-01 -8.80116582e-01 -4.35767084e-01 -1.87031016e-01
-6.87625647e-01 -2.03203365e-01 4.74928588e-01 6.28502667e-01
2.37927750e-01 4.79541957e-01 2.68091381e-01 -7.75575221e-01
4.48828936e-01 -6.29375041e-01 -8.69875729e-01 8.68092105e-02
-4.02198523e-01 1.38930336e-01 6.91774905e-01 7.91278780e-02
-1.53440952e-01 2.64904350e-01 4.30488706e-01 9.28753838e-02
2.79262930e-01 8.80258203e-01 1.49839774e-01 -4.85553145e-01
5.39958358e-01 1.95680842e-01 -1.32483125e-01 -4.82070744e-02
4.49675351e-01 6.74742758e-01 1.18312381e-01 -4.17768121e-01
1.06801784e+00 7.63890564e-01 7.10445404e-01 -1.18427932e+00
-3.74128729e-01 -1.04309821e+00 -5.91375828e-01 -1.19335726e-01
8.00476968e-01 -6.95832193e-01 -1.15513802e+00 3.47757004e-02
-1.11427999e+00 2.97488898e-01 -2.40378585e-02 4.10795659e-01
-2.35120028e-01 7.23224461e-01 -6.81433201e-01 -1.13293481e+00
-8.82473737e-02 -9.46689785e-01 5.55736840e-01 1.22824430e-01
-5.14723808e-02 -7.04763889e-01 4.37711477e-01 4.51422960e-01
1.09417006e-01 1.61748216e-01 9.75450397e-01 -4.69813347e-01
-1.06390178e+00 -3.01334590e-01 -3.54817659e-01 -8.21271315e-02
-2.27525294e-01 1.27159834e-01 -5.01017749e-01 -5.30584276e-01
-2.32881352e-01 -1.31427959e-01 5.78702867e-01 6.40984997e-02
3.10072362e-01 2.32163027e-01 -4.35165793e-01 2.10267767e-01
1.70006573e+00 -1.59015983e-01 5.26436448e-01 1.14210844e-01
5.73384404e-01 4.95717973e-01 9.16790888e-02 4.70132917e-01
2.25714669e-01 4.77156639e-01 4.58076954e-01 1.90743834e-01
2.20327273e-01 1.95550397e-01 9.85003933e-02 1.39120579e+00
-3.05548787e-01 -9.16201398e-02 -1.01299167e+00 5.65479338e-01
-1.76209760e+00 -1.15013659e+00 -7.73391187e-01 2.15049434e+00
6.00548148e-01 2.16951042e-01 1.83448255e-01 3.20347399e-01
9.39479351e-01 -2.18260691e-01 4.83530834e-02 -3.57415676e-01
3.69156897e-02 5.34196436e-01 5.12073994e-01 4.82342720e-01
-8.35132241e-01 7.40905106e-01 5.60379696e+00 6.43578768e-01
-7.04570293e-01 2.24020198e-01 5.13065793e-02 3.14173043e-01
-1.97678208e-01 8.08451891e-01 -5.93382180e-01 2.67824084e-01
1.19599366e+00 -3.15456122e-01 7.05730319e-01 3.79157841e-01
2.11663619e-01 -6.49721563e-01 -9.85622704e-01 9.25597906e-01
-2.16788635e-01 -1.36090922e+00 -2.51093239e-01 5.71056187e-01
8.11870575e-01 9.66871753e-02 -3.92387837e-01 -9.39315259e-02
3.64782035e-01 -7.17720151e-01 5.68432987e-01 2.17528090e-01
6.81997597e-01 -6.72063828e-01 5.36571860e-01 3.68033081e-01
-1.28631353e+00 -1.75200373e-01 -4.66816336e-01 -4.15216297e-01
-1.02992237e-01 9.17360485e-01 -7.12693036e-01 7.65000045e-01
5.03967702e-01 4.26805198e-01 -2.72908866e-01 1.13269126e+00
8.12342167e-02 6.28062725e-01 -6.71459794e-01 -3.50882381e-01
3.79930258e-01 -8.93199563e-01 6.41766131e-01 8.31738234e-01
1.80392385e-01 2.73251921e-01 1.14969812e-01 9.49401498e-01
-1.69324264e-01 1.23940498e-01 -7.10891008e-01 -4.43395078e-01
5.29422820e-01 1.45814800e+00 -1.42413759e+00 -2.94153899e-01
-3.61681208e-02 7.68136978e-01 4.76530999e-01 9.20885056e-02
-3.96285504e-01 -6.07692540e-01 1.60431206e-01 1.99611276e-01
5.18091142e-01 -4.64783370e-01 -1.00649640e-01 -1.03159475e+00
-1.32174402e-01 -8.56802881e-01 1.22935832e-01 -2.99877554e-01
-9.21637416e-01 3.05165380e-01 -2.30446890e-01 -9.04892504e-01
1.49333835e-01 -3.06719363e-01 -5.11081517e-01 8.64775419e-01
-1.14663494e+00 -7.18199313e-01 7.68701434e-02 4.97821301e-01
-5.79341054e-01 3.21809053e-01 8.88485670e-01 3.12402964e-01
-6.73384488e-01 -9.65373293e-02 6.52646244e-01 8.92957151e-02
1.06291756e-01 -1.14442706e+00 -4.80601601e-02 1.02766991e+00
1.41984150e-01 9.59698856e-01 6.98305964e-01 -5.83001018e-01
-1.73223484e+00 -3.81990314e-01 1.23490465e+00 -2.56776780e-01
1.22779882e+00 -7.86375999e-01 -3.21159482e-01 2.06900105e-01
1.24606937e-01 -1.41963148e-02 6.63261533e-01 2.20125899e-01
-1.81264997e-01 -1.60553604e-01 -8.97060752e-01 2.09146887e-01
8.69936764e-01 -8.52142274e-01 -2.27802753e-01 6.40176415e-01
1.79675385e-01 2.07864940e-01 -7.22340822e-01 -1.64305061e-01
2.12790772e-01 -1.08622324e+00 7.12340295e-01 -4.48772430e-01
2.23635793e-01 -6.25775456e-01 7.71002173e-02 -8.48142862e-01
-5.92216372e-01 -9.64979947e-01 1.53470278e-01 8.38456094e-01
4.82897460e-02 -5.33797979e-01 1.06765425e+00 -3.56960818e-02
3.68570924e-01 -2.29605302e-01 -1.50273752e+00 -5.17689884e-01
-2.26562113e-01 -5.68765495e-03 1.40662789e-01 1.16041827e+00
4.82230693e-01 8.62513363e-01 1.00597953e-02 1.71544969e-01
1.02322054e+00 2.64154583e-01 5.62938631e-01 -1.59618437e+00
-4.72015589e-01 -3.91544312e-01 -8.10653150e-01 -4.88339812e-01
-2.17224658e-01 -1.11050403e+00 4.36961129e-02 -1.27964032e+00
5.62012374e-01 -6.50452793e-01 -2.32469663e-01 1.96345493e-01
2.36887425e-01 4.28351730e-01 2.34294981e-01 2.76741982e-01
-1.04267502e+00 8.89128819e-02 9.54076111e-01 -1.21426932e-03
1.59000624e-02 -1.67327106e-01 -4.53050017e-01 2.96303600e-01
4.47505563e-01 -9.27509427e-01 -1.53570011e-01 2.97502995e-01
8.80919993e-01 3.56210351e-01 3.95781636e-01 -1.15945172e+00
4.33188856e-01 7.91142359e-02 -4.84217644e-01 -4.47358012e-01
9.65256244e-02 -8.02126288e-01 3.90574723e-01 9.70202804e-01
2.33486667e-01 -7.79957324e-02 -4.10124123e-01 8.12352479e-01
-3.43882181e-02 -5.23542404e-01 6.27067268e-01 -2.83817828e-01
-3.43016595e-01 4.44350317e-02 -5.41416109e-01 -1.33632556e-01
1.11862874e+00 -3.46342295e-01 -2.39521906e-01 -1.28793985e-01
-7.94546902e-01 1.02913454e-01 6.07091248e-01 -3.95375252e-01
1.87629372e-01 -8.73530984e-01 -6.87724352e-01 -1.59709349e-01
1.02227889e-02 -2.29620859e-01 3.95278707e-02 1.39658916e+00
-7.54397988e-01 6.24535799e-01 -1.68620765e-01 -6.88324332e-01
-1.27442300e+00 5.75443268e-01 2.02471331e-01 -4.89133358e-01
-3.81741405e-01 7.15098500e-01 -1.65549174e-01 -7.02646151e-02
-3.29817921e-01 -4.05009240e-02 1.17788367e-01 1.67271316e-01
8.15581717e-03 2.91911632e-01 6.97657987e-02 -7.53209651e-01
-5.03348172e-01 4.04920876e-01 1.56357452e-01 -6.20076843e-02
1.36160040e+00 -1.73878849e-01 -1.05452812e+00 1.72670946e-01
1.04164577e+00 9.86949131e-02 -3.26574445e-01 -3.44336778e-02
3.64137620e-01 -1.12386353e-01 -1.83812231e-02 -2.09700055e-02
-7.03654766e-01 8.38402390e-01 2.48008057e-01 9.55658376e-01
7.88108468e-01 6.05963208e-02 3.54636401e-01 8.08954060e-01
1.01369989e+00 -9.16188359e-01 -1.27260879e-01 5.57431340e-01
-6.75418153e-02 -8.69536102e-01 1.91100329e-01 -6.86675549e-01
3.13576758e-01 1.10527062e+00 -1.67399034e-01 -3.61697346e-01
4.76677060e-01 -1.08374514e-01 -6.38195992e-01 -6.65522039e-01
-4.16284919e-01 -5.67365766e-01 -2.02921391e-01 3.48533630e-01
1.13387041e-01 3.98432404e-01 -6.23605847e-01 1.26998350e-01
-6.71529071e-03 -8.80282745e-02 1.05187249e+00 8.03208053e-01
-7.18669176e-01 -1.51693857e+00 -5.86630106e-01 4.13352758e-01
-4.18637022e-02 -2.79710054e-01 -6.66684031e-01 7.16766715e-01
2.38161176e-01 9.80243981e-01 -1.38173983e-01 -3.96920413e-01
-1.25322178e-01 -1.53768227e-01 6.47033989e-01 -9.83062327e-01
-6.25761509e-01 -1.69169307e-02 2.74024699e-02 -2.90007055e-01
-6.53455973e-01 -7.43535757e-01 -1.32688439e+00 -7.39885747e-01
-8.84684026e-01 7.33523548e-01 5.78217387e-01 9.81113136e-01
1.34577051e-01 2.23613933e-01 7.54078746e-01 -6.78048134e-01
-6.24249339e-01 -5.77642381e-01 -9.78837192e-01 1.48842543e-01
1.46988496e-01 -5.83630443e-01 -5.15669644e-01 -3.85055214e-01] | [6.933660984039307, 5.1836256980896] |
89a86eca-ebcc-405e-80db-dcdfed5122a4 | an-efficient-convex-hull-based-vehicle-pose | 2302.01034 | null | https://arxiv.org/abs/2302.01034v2 | https://arxiv.org/pdf/2302.01034v2.pdf | An Efficient Convex Hull-based Vehicle Pose Estimation Method for 3D LiDAR | Vehicle pose estimation with LiDAR is essential in the perception technology of autonomous driving. However, due to incomplete observation measurements and sparsity of the LiDAR point cloud, it is challenging to achieve satisfactory pose extraction based on 3D LiDAR by using the existing pose estimation methods. In addition, the requirement for real-time performance further increases the difficulty of the pose estimation task. In this paper, we proposed a novel convex hull-based vehicle pose estimation method. The extracted 3D cluster is reduced to the convex hull, reducing the computation burden and retaining contour information. Then a novel criterion based on the minimum occlusion area is developed for the search-based algorithm, which can achieve accurate pose estimation. This criterion also makes the proposed algorithm especially suitable for obstacle avoidance. The proposed algorithm is validated on the KITTI dataset and a manually labeled dataset acquired at an industrial park. The results show that our proposed method can achieve better accuracy than the state-of-the-art pose estimation method while maintaining real-time speed. | ['Ningning Ding'] | 2023-02-02 | null | null | null | null | ['vehicle-pose-estimation'] | ['computer-vision'] | [ 1.49228182e-02 -2.15399146e-01 2.21743062e-02 -3.76812994e-01
-5.17254114e-01 -2.03000978e-01 2.08541781e-01 1.53291270e-01
-5.78985572e-01 5.41288495e-01 -6.15573168e-01 -3.50469977e-01
-3.53557378e-01 -8.19021225e-01 -4.94364411e-01 -5.38282275e-01
2.45941281e-01 6.37642860e-01 4.31391597e-01 -1.45095244e-01
4.84955341e-01 9.27485704e-01 -1.96790373e+00 -6.59573376e-01
1.20021403e+00 9.98393357e-01 7.67642796e-01 -4.93763760e-03
-1.41165048e-01 -2.58291811e-01 -2.14711607e-01 4.20387238e-02
3.30718249e-01 9.52174366e-02 1.32300720e-01 3.30056101e-01
3.10168922e-01 -4.12036657e-01 1.09326974e-01 1.08002770e+00
3.20536524e-01 3.37293111e-02 5.66359282e-01 -1.38321698e+00
4.27853256e-01 -2.63614893e-01 -8.65747392e-01 -2.32163191e-01
-6.59557208e-02 -3.40129048e-01 5.55320442e-01 -1.14797640e+00
5.77071309e-01 9.87689734e-01 5.39950848e-01 -3.22914533e-02
-8.08690846e-01 -9.70022202e-01 -9.91712734e-02 6.11695945e-01
-1.92745161e+00 -2.59487331e-01 9.60840166e-01 -3.85265470e-01
4.79178131e-01 1.79357119e-02 6.80379868e-01 1.81253590e-02
4.18931872e-01 2.95578778e-01 1.12350345e+00 -2.10112303e-01
4.46555950e-02 2.71712869e-01 -2.32119914e-02 7.15599298e-01
8.30881119e-01 1.64591551e-01 -2.18639508e-01 1.65166736e-01
4.90000933e-01 1.59333572e-01 3.25017087e-02 -9.36671138e-01
-8.29189599e-01 8.35467756e-01 4.46281999e-01 -1.20247267e-01
-3.26952130e-01 -2.27261066e-01 2.13998139e-01 -2.76125699e-01
3.02948803e-01 -3.83389071e-02 -2.44711667e-01 -8.40774551e-02
-9.21032488e-01 3.26190293e-01 2.92928278e-01 1.12412429e+00
1.03895152e+00 8.07136893e-02 6.77565396e-01 4.60775018e-01
6.61851883e-01 8.81355226e-01 -1.31721660e-01 -6.63856030e-01
5.89677811e-01 8.88349712e-01 1.28631428e-01 -1.20668590e+00
-6.31880641e-01 -4.83576834e-01 -4.32793230e-01 5.61179698e-01
7.39914626e-02 5.84751256e-02 -8.37972641e-01 1.14895034e+00
6.31504714e-01 -7.21805617e-02 7.13738874e-02 8.52575600e-01
6.59685552e-01 5.54678142e-01 -2.19599247e-01 -4.94985133e-01
1.19822991e+00 -4.62960809e-01 -8.95752251e-01 -3.83722335e-01
5.26622176e-01 -9.02803421e-01 5.18699825e-01 4.95089352e-01
-5.38076162e-01 -6.92818165e-01 -1.48234475e+00 1.88262194e-01
-2.22822413e-01 6.63772583e-01 3.44603360e-01 5.60150743e-01
-4.02077377e-01 1.33364950e-03 -8.26480150e-01 -5.10623276e-01
1.74492598e-01 3.93426329e-01 -4.06581819e-01 -2.23501295e-01
-7.62589633e-01 1.41181469e+00 5.47726929e-01 3.90010238e-01
-4.33197975e-01 -3.23265076e-01 -9.21575308e-01 -1.90350428e-01
5.46634734e-01 -1.78999543e-01 9.22165096e-01 1.09355070e-01
-1.17701328e+00 5.59785306e-01 -4.28102911e-01 -3.87423098e-01
5.90378642e-01 -4.10002381e-01 -3.02292645e-01 1.35296911e-01
1.97946519e-01 5.38783789e-01 6.58752441e-01 -1.46586096e+00
-7.77143776e-01 -7.84604967e-01 -4.38412189e-01 4.55758065e-01
-1.47091150e-01 -4.22080249e-01 -4.16712791e-01 1.89624771e-01
6.93821847e-01 -1.08967245e+00 -3.31819594e-01 1.49116218e-01
-9.90132689e-02 -2.42508724e-01 1.60110009e+00 -5.06533921e-01
1.03624892e+00 -2.24020147e+00 -3.03747684e-01 3.80349308e-01
8.74293894e-02 3.13006461e-01 3.02656978e-01 3.96750540e-01
4.51944679e-01 -3.56109530e-01 -2.88060158e-01 -2.08910048e-01
-3.18377495e-01 2.95199066e-01 7.17033297e-02 7.69714355e-01
2.16654673e-01 2.96152741e-01 -5.05732000e-01 -9.68891501e-01
7.49664068e-01 5.15211403e-01 -2.65335768e-01 3.74216586e-02
9.03709009e-02 2.28492439e-01 -6.77061081e-01 6.82292879e-01
1.33730471e+00 5.37434578e-01 2.94370372e-02 -3.56659859e-01
-7.02976882e-01 -7.33991386e-03 -1.52011979e+00 1.43174982e+00
-5.30815780e-01 5.13350844e-01 4.33602124e-01 -8.45745087e-01
1.62564743e+00 -2.46483553e-02 5.59851944e-01 -4.50579673e-01
2.46982768e-01 5.88705778e-01 3.74978147e-02 -4.23640996e-01
7.89448440e-01 -9.10033807e-02 -1.44626215e-01 -1.57674402e-01
-5.69246173e-01 -6.49725914e-01 1.80726886e-01 -1.68670997e-01
4.06292558e-01 3.48786116e-01 5.20679593e-01 -2.91116029e-01
7.42680013e-01 3.78982246e-01 7.96094656e-01 -7.76605215e-04
-2.39947498e-01 1.59188330e-01 -3.07672709e-01 -2.00678736e-01
-9.95491743e-01 -1.06309700e+00 -5.42054236e-01 -7.53117725e-02
8.41592610e-01 -2.41053551e-01 -4.27060246e-01 -2.80822366e-01
3.53338718e-01 5.31512678e-01 -1.32882297e-02 -1.80203523e-02
-6.48395896e-01 -3.54869545e-01 -4.13813479e-02 4.75742370e-01
7.31145144e-01 -5.42419076e-01 -1.03092325e+00 2.84900159e-01
-5.46497889e-02 -1.21837139e+00 1.69376228e-02 -2.90154368e-02
-1.04861140e+00 -1.00385690e+00 -1.28035499e-02 -7.57765412e-01
7.07485855e-01 7.53750682e-01 4.21711773e-01 2.19245747e-01
-2.52343297e-01 -2.58085787e-01 -1.79204702e-01 -8.73483181e-01
-3.78406905e-02 3.56109813e-02 2.77579367e-01 -1.97485954e-01
5.69895148e-01 -2.51521170e-01 -4.60000902e-01 6.10348463e-01
-3.64924461e-01 -5.74091710e-02 9.21923578e-01 4.44596469e-01
7.00868428e-01 5.65859139e-01 5.70377052e-01 -3.49335968e-01
2.12998062e-01 -7.19836950e-02 -1.15331078e+00 -3.84111553e-01
-8.69385660e-01 -2.55220354e-01 2.98796266e-01 -2.11273916e-02
-1.09126484e+00 7.14200556e-01 -1.00628905e-01 -3.27230275e-01
-9.53495950e-02 5.37204266e-01 -3.84401530e-01 -2.47139469e-01
2.15236992e-01 1.72246918e-01 5.02741098e-01 -4.34361637e-01
1.19181827e-01 9.09667134e-01 4.93673950e-01 -4.02600504e-02
1.23400700e+00 5.79352856e-01 5.89493036e-01 -1.19267106e+00
-2.78953105e-01 -7.41581857e-01 -9.17086601e-01 -4.57836032e-01
8.30929458e-01 -1.00284684e+00 -9.00682211e-01 1.20517530e-01
-1.18707955e+00 6.51884556e-01 2.61383384e-01 9.73685145e-01
-4.80193704e-01 4.89995927e-01 3.95455919e-02 -1.23140883e+00
-2.13880584e-01 -1.38085032e+00 9.83473420e-01 2.13858739e-01
-2.79625300e-02 -4.39186603e-01 -3.24705482e-01 5.24652719e-01
1.84859902e-01 3.29292864e-01 5.07858813e-01 4.33093309e-03
-9.92874920e-01 -6.22480094e-01 -2.72294104e-01 -5.07933423e-02
1.35274313e-03 1.52952239e-01 -6.10096514e-01 -2.80451328e-01
1.61115840e-01 -3.39815952e-02 5.94738007e-01 2.34938145e-01
7.04985082e-01 3.97159606e-01 -7.52286911e-01 2.91022450e-01
1.74457955e+00 3.75708789e-01 5.49068391e-01 6.29617095e-01
5.18235743e-01 6.41081572e-01 1.64048779e+00 3.09096217e-01
4.85344648e-01 8.68081152e-01 8.66584301e-01 -5.72803691e-02
2.47497931e-01 -3.51159215e-01 7.88128525e-02 9.15546656e-01
1.00077495e-01 2.34262779e-01 -8.18023086e-01 5.80976486e-01
-1.76113737e+00 -6.96096182e-01 -5.85470378e-01 2.37148619e+00
3.27214062e-01 3.99325937e-01 -1.67242229e-01 6.20945275e-01
7.16599405e-01 -1.19581610e-01 -3.21184874e-01 -4.77070391e-01
4.68378782e-01 -8.40252712e-02 8.31057847e-01 6.48226738e-01
-1.00174177e+00 8.13626349e-01 5.26838875e+00 8.23095143e-01
-1.12802303e+00 -1.14320442e-01 -3.20568413e-01 2.47922719e-01
3.30498554e-02 4.93726209e-02 -1.37885427e+00 1.67069152e-01
6.18997455e-01 -3.00369561e-01 -2.59025127e-01 1.15852392e+00
5.27915478e-01 -5.75102568e-01 -5.84581196e-01 1.01465666e+00
1.79492608e-01 -9.09730852e-01 -2.24081129e-01 3.76240432e-01
3.41884792e-01 -2.66539693e-01 -1.66004762e-01 4.19476256e-02
-4.12025422e-01 -7.13819981e-01 6.66315436e-01 1.98990121e-01
7.47968256e-01 -1.15975285e+00 1.01836085e+00 8.46892178e-01
-1.67959857e+00 -5.34501150e-02 -5.20311594e-01 -2.86051482e-01
5.66945314e-01 7.38308251e-01 -1.40501595e+00 7.58321106e-01
3.66461903e-01 5.71104944e-01 -4.37830001e-01 1.28695130e+00
-1.37600452e-01 1.48723900e-01 -5.48681557e-01 -2.80539453e-01
2.17394102e-02 -5.99684358e-01 7.20365763e-01 7.43503988e-01
6.90499067e-01 -1.75213784e-01 5.77995121e-01 6.51133835e-01
3.76184046e-01 2.99289882e-01 -9.84369695e-01 3.00267428e-01
8.16423833e-01 1.51357830e+00 -7.85215020e-01 -7.87607394e-03
-3.30912411e-01 1.96002498e-01 -9.82318893e-02 -1.53932095e-01
-8.50354075e-01 -5.19294262e-01 4.46992636e-01 5.38042724e-01
3.06731462e-01 -8.79890084e-01 -5.44168591e-01 -4.92132336e-01
1.71237156e-01 -2.30875969e-01 -1.89172164e-01 -6.35927439e-01
-5.20953298e-01 4.87857431e-01 2.98846513e-01 -1.82177639e+00
-1.55136853e-01 -6.50211394e-01 -3.52242470e-01 7.49404073e-01
-1.67008591e+00 -1.14816570e+00 -6.06248319e-01 1.20839059e-01
6.16722286e-01 9.31967199e-02 5.11809230e-01 4.03894573e-01
-4.35498774e-01 1.86100438e-01 -1.38611749e-01 -4.32217658e-01
4.78595436e-01 -5.51591814e-01 -1.36325672e-01 9.21225190e-01
-3.05253595e-01 3.75432700e-01 8.69175255e-01 -9.59043801e-01
-1.53067005e+00 -1.03047466e+00 8.47467542e-01 -3.02798245e-02
2.26050079e-01 -2.75011450e-01 -6.82763875e-01 3.27744901e-01
-2.37172633e-01 -2.20388949e-01 1.70992225e-01 -3.35733056e-01
3.34274203e-01 -4.96263951e-01 -1.36167812e+00 2.98580080e-01
8.39831591e-01 -7.97674134e-02 -4.86951083e-01 -1.35036662e-01
4.28472191e-01 -5.93204498e-01 -6.51950538e-01 9.40064669e-01
7.91494012e-01 -6.38770640e-01 7.86879659e-01 2.64789492e-01
-4.26522344e-02 -9.55879450e-01 -5.76058999e-02 -9.53459203e-01
-3.38060260e-02 -9.67880264e-02 2.50755548e-01 9.98106718e-01
3.37460846e-01 -7.54295111e-01 1.02947450e+00 2.11883470e-01
-3.19799840e-01 -6.84048116e-01 -1.18594027e+00 -9.07637894e-01
-4.99420017e-01 -4.58176106e-01 3.83339733e-01 3.88040125e-01
-2.73128539e-01 4.77951646e-01 -1.51686952e-01 5.26125133e-01
8.77587438e-01 3.19912672e-01 1.13891041e+00 -1.79780281e+00
5.01802385e-01 1.46152362e-01 -8.96568894e-01 -1.00655639e+00
9.73219126e-02 -5.19802928e-01 3.91551435e-01 -1.65071952e+00
-3.06891967e-02 -6.09144211e-01 3.63819718e-01 -5.56003526e-02
5.49491346e-02 3.57032388e-01 2.39060476e-01 1.38316944e-01
-5.64730577e-02 7.23876417e-01 1.31221831e+00 1.88358784e-01
-1.99850276e-01 2.10044652e-01 -2.06924662e-01 8.74143064e-01
9.13571656e-01 -3.75414491e-01 -6.41658425e-01 -2.13870376e-01
-3.06868628e-02 6.64087310e-02 5.21820039e-02 -1.18082702e+00
2.98553199e-01 -2.01544464e-01 2.55955070e-01 -1.72001112e+00
8.67570996e-01 -1.38523686e+00 1.38585150e-01 7.82234311e-01
4.98956740e-01 2.58594900e-01 3.09549421e-01 5.70143878e-01
-2.46984527e-01 -3.75764221e-01 7.81286716e-01 1.44842386e-01
-8.95538926e-01 2.07134247e-01 -2.70392716e-01 -5.47106683e-01
1.56486523e+00 -7.57619977e-01 -4.01330478e-02 -2.52178192e-01
-2.68084705e-01 3.71544331e-01 6.55352056e-01 3.26624781e-01
1.03094661e+00 -1.45355070e+00 -6.42499924e-01 4.24406528e-01
3.79519641e-01 3.08977067e-01 -3.41181308e-02 9.45718408e-01
-7.98281848e-01 5.66817343e-01 -2.65481740e-01 -1.04777539e+00
-1.64289033e+00 4.05191869e-01 -2.17926487e-01 1.80602446e-01
-4.83894229e-01 1.48586780e-01 -1.83851764e-01 -3.56413096e-01
-1.87712446e-01 -2.94868380e-01 -3.93981963e-01 -3.70048583e-02
7.83884674e-02 5.74418485e-01 3.18193316e-01 -1.00827122e+00
-6.42428100e-01 1.26587498e+00 1.74535796e-01 -1.05033934e-01
1.02340233e+00 -3.88799936e-01 9.96613950e-02 2.38913432e-01
1.09249091e+00 3.27650756e-01 -1.03041923e+00 1.27421170e-01
-1.01645395e-01 -7.32611656e-01 2.21977144e-01 -3.15951966e-02
-7.50543833e-01 9.69467998e-01 7.47603714e-01 -1.89567223e-01
8.13761175e-01 -4.01850373e-01 8.92568886e-01 3.70232046e-01
7.90122867e-01 -1.17100239e+00 -3.19246143e-01 3.29776824e-01
7.50305951e-01 -1.29081702e+00 5.48357010e-01 -1.08257318e+00
-3.31257313e-01 1.09038079e+00 8.08121860e-01 -9.72543955e-02
6.42496586e-01 3.37406099e-01 1.29868865e-01 -4.34055999e-02
-1.86325639e-01 -4.31993216e-01 1.89767823e-01 6.04176104e-01
7.04196841e-02 1.26805753e-01 -9.32746828e-01 8.10106918e-02
-4.01953101e-01 -2.87846234e-02 5.55256188e-01 1.16279459e+00
-1.11209726e+00 -9.84287381e-01 -7.04174459e-01 3.83704722e-01
3.24534737e-02 4.13732588e-01 2.42367722e-02 1.15899622e+00
3.25710624e-01 1.12750316e+00 1.48218825e-01 -4.89630014e-01
4.84627366e-01 -1.13216855e-01 4.15410966e-01 -5.66424668e-01
3.39656621e-01 1.93389133e-01 1.61248684e-01 -3.20727825e-01
-2.86444753e-01 -6.79242194e-01 -1.63698077e+00 -1.01072066e-01
-8.67423356e-01 2.85776138e-01 1.29333735e+00 8.08611691e-01
2.22297296e-01 5.81432804e-02 8.08467865e-01 -9.06290233e-01
-6.06764615e-01 -9.09696639e-01 -5.58749318e-01 8.05332065e-02
2.53947023e-02 -1.32395959e+00 -2.87661642e-01 -2.14067027e-01] | [7.7153754234313965, -2.2947449684143066] |
6e59c014-b2fc-4eb5-b49e-1a9766bd82ef | meta-learned-models-of-cognition | 2304.06729 | null | https://arxiv.org/abs/2304.06729v1 | https://arxiv.org/pdf/2304.06729v1.pdf | Meta-Learned Models of Cognition | Meta-learning is a framework for learning learning algorithms through repeated interactions with an environment as opposed to designing them by hand. In recent years, this framework has established itself as a promising tool for building models of human cognition. Yet, a coherent research program around meta-learned models of cognition is still missing. The purpose of this article is to synthesize previous work in this field and establish such a research program. We rely on three key pillars to accomplish this goal. We first point out that meta-learning can be used to construct Bayes-optimal learning algorithms. This result not only implies that any behavioral phenomenon that can be explained by a Bayesian model can also be explained by a meta-learned model but also allows us to draw strong connections to the rational analysis of cognition. We then discuss several advantages of the meta-learning framework over traditional Bayesian methods. In particular, we argue that meta-learning can be applied to situations where Bayesian inference is impossible and that it enables us to make rational models of cognition more realistic, either by incorporating limited computational resources or neuroscientific knowledge. Finally, we reexamine prior studies from psychology and neuroscience that have applied meta-learning and put them into the context of these new insights. In summary, our work highlights that meta-learning considerably extends the scope of rational analysis and thereby of cognitive theories more generally. | ['Eric Schulz', 'Jane X. Wang', 'Matthew Botvinick', 'Akshay Jagadish', 'Ishita Dasgupta', 'Marcel Binz'] | 2023-04-12 | null | null | null | null | ['bayesian-inference'] | ['methodology'] | [ 2.02640463e-02 3.45283121e-01 -2.92912513e-01 -4.57924277e-01
-4.06540245e-01 -3.91459852e-01 7.04887211e-01 3.87901425e-01
-6.42117023e-01 5.10832369e-01 1.10957950e-01 -5.85117579e-01
-8.05497885e-01 -7.81230032e-01 -8.13445508e-01 -3.83726150e-01
1.52134567e-01 3.51396620e-01 -3.27524878e-02 5.15584871e-02
7.55550742e-01 3.03502440e-01 -1.92297006e+00 1.80814192e-01
1.04791999e+00 1.65480614e-01 5.16925752e-01 4.10497248e-01
1.08999744e-01 8.76925349e-01 -3.83771837e-01 -6.95655882e-01
-2.43312836e-01 -3.83687347e-01 -7.98383415e-01 -9.68971252e-02
3.88025224e-01 -4.20165211e-01 -1.78561494e-01 9.14115369e-01
3.20242435e-01 2.87497193e-01 8.21299434e-01 -1.04614413e+00
-7.82704771e-01 8.35912406e-01 -1.12378351e-01 3.62893999e-01
3.11364174e-01 -3.37217911e-03 8.55457485e-01 -5.86643934e-01
2.63800591e-01 1.71709621e+00 4.95344758e-01 5.93213558e-01
-1.33118451e+00 -6.39416635e-01 5.66371799e-01 6.74802780e-01
-8.97076666e-01 -3.52667361e-01 6.08253360e-01 -6.28617942e-01
6.64834023e-01 -8.07997510e-02 1.18990326e+00 1.15033901e+00
5.75873256e-01 8.43735278e-01 1.91575336e+00 -8.86636436e-01
5.21478653e-01 1.96410194e-01 7.45811701e-01 4.92436379e-01
5.67436695e-01 7.39676356e-01 -8.01790595e-01 7.88031518e-02
6.73572063e-01 -8.05839002e-02 2.41346285e-01 -1.79575294e-01
-8.67165923e-01 9.06132102e-01 1.17371902e-01 4.00685549e-01
-4.23996389e-01 3.21490973e-01 8.63701478e-03 3.76034155e-02
3.45825613e-01 6.13185883e-01 -4.37928736e-01 -7.83451349e-02
-9.23292518e-01 4.99265105e-01 6.12502933e-01 3.33133727e-01
5.21428287e-01 1.40516460e-01 5.21936640e-02 7.76027441e-01
9.52595890e-01 5.60889363e-01 5.18300474e-01 -1.29494965e+00
-2.54564881e-01 2.03187242e-01 4.96816076e-02 -8.90580058e-01
-5.78238130e-01 -5.48902214e-01 1.13791320e-02 3.55413765e-01
4.07711029e-01 -6.79227933e-02 -4.90132809e-01 1.99557161e+00
4.90994640e-02 7.49848410e-02 -1.44638911e-01 4.71560001e-01
6.27995908e-01 3.70523483e-01 3.72496545e-01 -5.42462409e-01
1.38687038e+00 -6.72042072e-01 -7.95941651e-01 -5.47418058e-01
4.00142223e-01 -5.40905476e-01 9.93121028e-01 7.13031292e-01
-1.42160165e+00 -6.84888721e-01 -1.11216271e+00 9.43697914e-02
-3.31022203e-01 -3.59445363e-01 1.34783137e+00 1.33022392e+00
-1.12799168e+00 5.14089882e-01 -1.08279431e+00 -3.99816036e-01
4.61382747e-01 1.86450351e-02 3.17373157e-01 -1.00714818e-01
-1.02452731e+00 1.43634164e+00 6.64810061e-01 -1.44451112e-02
-1.01075613e+00 -7.29716480e-01 -6.76014960e-01 -9.07322615e-02
4.55306172e-01 -1.11634707e+00 1.59832919e+00 -1.08869040e+00
-1.57312238e+00 7.47823775e-01 -5.09761691e-01 -2.44917214e-01
6.96395189e-02 -2.68359542e-01 -3.14173490e-01 1.17816284e-01
-9.94160399e-02 3.71780396e-01 4.97448951e-01 -1.38513482e+00
-6.33660734e-01 -6.86361790e-01 2.45233148e-01 2.44646087e-01
-9.69782993e-02 2.28006653e-02 8.57032686e-02 -5.99371791e-01
2.74498880e-01 -8.67147624e-01 -3.47329564e-02 -5.07752597e-01
2.17131838e-01 -5.35159886e-01 1.08380932e-02 -1.99735075e-01
1.15136516e+00 -1.68230391e+00 -6.15513846e-02 1.04296684e-01
4.90335882e-01 2.67203152e-02 5.41363172e-02 3.86619568e-01
-2.05761224e-01 2.66557157e-01 1.55955032e-01 -1.09734580e-01
3.26236874e-01 1.25886187e-01 -2.92048484e-01 2.64457375e-01
-3.16673934e-01 1.12907994e+00 -1.05150509e+00 -2.74047166e-01
3.74593049e-01 3.97946090e-01 -7.66592979e-01 -5.38850315e-02
-1.52433842e-01 6.12748191e-02 -1.77034512e-01 3.25399548e-01
6.04977727e-01 -1.37074813e-01 5.91465235e-01 1.25710756e-01
-3.60911161e-01 5.67236006e-01 -1.04303467e+00 1.35808432e+00
-3.64771694e-01 8.43179166e-01 -5.05945861e-01 -1.12036026e+00
5.35388887e-01 1.34657204e-01 1.02577163e-02 -6.55273259e-01
1.75768137e-01 7.61185363e-02 5.42371750e-01 -5.72046995e-01
1.26541540e-01 -6.15704179e-01 4.80630368e-01 9.57825720e-01
2.42959827e-01 -2.97092557e-01 1.23887859e-01 1.15254931e-01
4.67577964e-01 2.33589306e-01 4.95765418e-01 -4.72167790e-01
1.29035696e-01 -2.69039869e-01 4.32317495e-01 1.25612569e+00
-8.24801400e-02 -3.51285785e-01 1.61024496e-01 -2.83784330e-01
-6.32015765e-01 -1.28094864e+00 -3.79166871e-01 1.39243102e+00
-2.64348745e-01 -4.36521828e-01 -8.48329127e-01 -2.38912970e-01
-3.16525102e-01 1.46382713e+00 -7.61973619e-01 -3.20724964e-01
-4.99679297e-02 -1.16977274e+00 2.25751057e-01 5.19901276e-01
4.18324053e-01 -8.87251019e-01 -1.01967037e+00 1.63385779e-01
-9.56540108e-02 -4.21882808e-01 5.21516144e-01 1.50114879e-01
-1.41012788e+00 -1.12228429e+00 -5.91662005e-02 -5.21012485e-01
3.06240052e-01 6.53714538e-01 1.04056442e+00 3.34860682e-01
7.50135779e-02 1.12826645e+00 -2.61842936e-01 -1.13094187e+00
-3.47001970e-01 -3.03989798e-01 1.73929080e-01 -7.19829977e-01
8.62429798e-01 -7.04760075e-01 -3.83616924e-01 -2.41504572e-02
-9.43012416e-01 -1.01356609e-02 6.25219524e-01 7.21386135e-01
1.45910278e-01 4.48745728e-01 9.32793200e-01 -9.50793684e-01
8.45674992e-01 -5.43662548e-01 -5.21986663e-01 3.86321604e-01
-1.09714198e+00 -1.10608317e-01 -2.36952603e-01 -3.45360070e-01
-1.64892733e+00 -6.02685988e-01 1.57370567e-01 4.06071305e-01
-3.34838539e-01 1.00823367e+00 6.02678908e-03 1.13694228e-01
7.87858665e-01 2.06355810e-01 5.12353424e-03 -1.69736356e-01
4.34505612e-01 3.70823711e-01 -1.93226598e-02 -1.06884992e+00
4.95788902e-01 3.10355306e-01 -1.20124474e-01 -9.10757244e-01
-1.20095038e+00 1.44905433e-01 -6.08621299e-01 -6.25038326e-01
7.24984825e-01 -8.03955555e-01 -1.03401339e+00 3.41724366e-01
-8.51642609e-01 -5.32852709e-01 9.31614786e-02 9.69896555e-01
-1.05526292e+00 1.17630653e-01 -3.79111707e-01 -1.22756362e+00
3.58641207e-01 -1.07574022e+00 2.77623981e-01 3.97790581e-01
-4.84496623e-01 -1.55226314e+00 2.49232221e-02 6.92356586e-01
1.90959319e-01 -2.35366568e-01 1.27138543e+00 -6.52809381e-01
-6.21174037e-01 2.05404714e-01 3.21871489e-01 -3.61931808e-02
-1.07440099e-01 7.55555704e-02 -1.32370400e+00 7.58248046e-02
7.12782145e-01 -6.31113052e-01 9.67707455e-01 8.36540937e-01
1.03742218e+00 3.00749242e-02 -2.61695981e-01 1.94875985e-01
1.23736608e+00 4.41825330e-01 4.95240390e-01 6.63477719e-01
5.84921464e-02 1.05026901e+00 5.48761308e-01 1.48673669e-01
8.95419896e-01 3.81308496e-01 3.32846120e-02 7.68253803e-01
8.84470940e-02 -3.85269672e-01 2.90308923e-01 7.69197345e-01
-2.62993187e-01 1.82617158e-01 -1.00865149e+00 2.14616612e-01
-1.98080969e+00 -1.44100833e+00 -8.61294754e-03 2.36189628e+00
8.41454327e-01 2.68359423e-01 1.46242350e-01 1.51884764e-01
4.82785493e-01 -2.12963074e-01 -5.55393577e-01 -2.96991020e-01
-3.27854455e-02 1.72596142e-01 -1.54311806e-01 7.44754791e-01
-5.56322992e-01 9.49242592e-01 8.10721207e+00 4.99849319e-01
-4.92023945e-01 1.93977743e-01 4.44442004e-01 -4.11328524e-02
-6.20747745e-01 1.88032061e-01 -6.90753520e-01 4.92184386e-02
1.39487219e+00 -4.30255741e-01 5.87597132e-01 6.43483281e-01
4.68582660e-01 -5.79064071e-01 -1.34181094e+00 7.14818418e-01
1.75097451e-01 -1.08041871e+00 7.91062564e-02 1.70433179e-01
5.74239194e-01 -4.82787907e-01 3.56333166e-01 5.45546114e-01
5.47521174e-01 -1.09864271e+00 1.00361967e+00 5.58761001e-01
-1.19155824e-01 -7.67719805e-01 4.02122229e-01 6.69800639e-01
-3.87868017e-01 -2.13278532e-01 -6.78900063e-01 -8.01061571e-01
-1.88100219e-01 6.06229126e-01 -6.87914789e-01 1.08697474e-01
5.71039855e-01 5.60507715e-01 -6.97607577e-01 1.16433942e+00
-5.67354083e-01 8.19987655e-01 1.69249520e-01 -1.31741807e-01
-1.61225304e-01 5.49484678e-02 3.14557552e-01 1.00198710e+00
5.03046364e-02 4.09643829e-01 -1.77706853e-01 1.26035535e+00
5.47363520e-01 -2.23740805e-02 -4.98796701e-01 -1.24765269e-01
8.50936115e-01 9.25153136e-01 -9.81386542e-01 -3.40168804e-01
-4.87468809e-01 2.34152079e-01 4.95107919e-01 3.47869813e-01
-5.03261685e-01 1.44671291e-01 3.36964369e-01 -2.59559870e-01
-9.46379229e-02 -3.05289954e-01 -6.63741529e-01 -1.17064500e+00
-6.98072731e-01 -1.01258457e+00 2.49856874e-01 -1.02541173e+00
-1.21297038e+00 -1.89908475e-01 7.02317178e-01 -3.39457840e-01
-4.27240729e-01 -7.88274348e-01 -2.49476627e-01 6.98919237e-01
-1.21228004e+00 -9.91584063e-01 -6.77245557e-02 3.00956488e-01
6.51311934e-01 4.28333282e-02 8.86599839e-01 -3.87994558e-01
-5.28195083e-01 2.73681670e-01 9.79116634e-02 -4.57949191e-01
6.88772798e-01 -1.28271294e+00 -9.77009237e-02 5.14577568e-01
2.04132080e-01 1.56590009e+00 8.60264957e-01 -7.91671216e-01
-1.23420525e+00 -2.58638352e-01 6.36795521e-01 -8.70859683e-01
5.32349646e-01 -1.01874374e-01 -8.07741642e-01 9.96788085e-01
3.20992738e-01 -8.46590877e-01 1.34613681e+00 7.48896062e-01
-4.98200268e-01 1.74483970e-01 -9.09859776e-01 9.96041059e-01
9.72733319e-01 -3.82341385e-01 -1.42110145e+00 2.99609005e-01
3.35759014e-01 -1.00562781e-01 -7.46450305e-01 -1.98884122e-02
9.91027832e-01 -1.37524748e+00 1.09850299e+00 -6.77427053e-01
4.64289963e-01 3.52313146e-02 -1.62542239e-01 -1.57529294e+00
-7.67637372e-01 -1.90889508e-01 -1.34377733e-01 1.01584005e+00
2.44161278e-01 -7.07421482e-01 4.95507598e-01 9.18200552e-01
-1.76108077e-01 -3.26018929e-01 -4.81569707e-01 -7.33949780e-01
7.19864428e-01 -1.13938320e+00 3.54916543e-01 8.55795503e-01
3.60767275e-01 2.12937146e-01 1.57045782e-01 3.49084064e-02
1.14309132e+00 -9.77197140e-02 4.23025161e-01 -1.83774126e+00
-3.25914949e-01 -7.31190324e-01 6.12834189e-03 -8.24321032e-01
6.27417803e-01 -8.27916801e-01 -1.38670653e-01 -1.56709683e+00
7.24498391e-01 -8.46010000e-02 -2.63608396e-01 1.65651903e-01
-3.31575602e-01 -2.37105250e-01 3.95814538e-01 1.23519488e-01
-3.42703462e-01 2.74115950e-01 1.00621355e+00 3.09062690e-01
-1.40823498e-01 1.87174492e-02 -1.42567563e+00 1.32411957e+00
1.00793278e+00 -4.73676383e-01 -8.38857889e-01 -3.56541902e-01
7.90334761e-01 -1.44045800e-01 6.33179247e-01 -7.46271670e-01
3.20178568e-01 -5.62465131e-01 7.18897879e-01 -4.08752948e-01
2.19872341e-01 -6.66526258e-01 -1.62400454e-01 5.04925668e-01
-4.32518512e-01 -1.72010232e-02 5.68881869e-01 6.80252850e-01
3.83699715e-01 -7.99778640e-01 7.08034337e-01 -5.38555503e-01
-7.38114774e-01 -5.58602929e-01 -1.04357302e+00 -1.40335476e-02
8.45734119e-01 -3.47373575e-01 -4.79772598e-01 -5.04430175e-01
-8.61852050e-01 -2.53309142e-02 2.90061682e-01 3.29457372e-01
6.37344480e-01 -1.09322083e+00 -3.31567556e-01 -2.31410161e-01
-9.86670032e-02 -6.51746273e-01 4.46475565e-01 7.13772476e-01
5.69466501e-03 9.00770307e-01 -2.75526553e-01 -3.19067389e-01
-9.36998785e-01 6.37070537e-01 3.44594747e-01 2.97522038e-01
-1.05806798e-01 5.34635186e-01 1.73572078e-01 -2.92765915e-01
1.96764767e-01 -1.75402090e-01 -3.58059198e-01 3.73854399e-01
9.21897888e-01 8.03053141e-01 -2.97100097e-01 -1.20486818e-01
-2.47318581e-01 3.33600760e-01 -1.55390874e-01 -6.95228875e-01
1.46494591e+00 -4.84497517e-01 -3.33778918e-01 1.17514288e+00
2.42204875e-01 -5.42235747e-02 -1.06081402e+00 -9.48861837e-02
8.96102786e-02 -4.18717891e-01 4.96918052e-01 -1.14358914e+00
-4.73766863e-01 9.20594275e-01 5.43673277e-01 -3.15897055e-02
8.35095644e-01 6.96681291e-02 -3.52319330e-01 6.68169916e-01
4.08233374e-01 -1.23016858e+00 4.91427392e-01 5.05020559e-01
6.29689336e-01 -1.19083416e+00 2.63584614e-01 -7.82964081e-02
-3.72514397e-01 1.27452898e+00 6.33330762e-01 6.54553846e-02
8.84007394e-01 -1.77961156e-01 -2.14617580e-01 -3.74291211e-01
-1.08642709e+00 -2.20927998e-01 3.06681991e-01 9.75326896e-01
1.00089121e+00 1.41586646e-01 -6.43930316e-01 8.90983105e-01
-5.26217878e-01 3.38414222e-01 7.93758452e-01 9.57269073e-01
-1.00030291e+00 -1.22054768e+00 -8.28199327e-01 3.61949950e-01
-3.09209526e-01 -2.21293107e-01 -3.24541181e-01 9.18504953e-01
9.24521238e-02 1.48218882e+00 -2.57979706e-02 -1.80977911e-01
-4.07500751e-02 4.49801952e-01 1.10830057e+00 -8.92060399e-01
-3.36423248e-01 2.59088874e-01 -1.68247387e-01 -3.27323407e-01
-1.02820504e+00 -1.11812997e+00 -7.56956220e-01 -3.84199411e-01
-2.77053028e-01 6.17000088e-02 5.34777164e-01 1.29161406e+00
-1.14988916e-01 6.95489943e-01 -1.65411457e-01 -6.39192939e-01
-5.30228198e-01 -9.03784275e-01 -6.20034039e-01 -3.90949845e-01
-3.57533921e-03 -1.09616184e+00 -4.11124617e-01 5.58491386e-02] | [9.462590217590332, 7.113786220550537] |
71d4a7f4-0916-4254-bc44-c2f48b42e484 | location-aware-web-service-qos-prediction-via | null | null | https://scholar.lanfanshu.cn/scholar?q=Location-Aware+Web+Service+QoS+Prediction+via+Deep+Collaborative+Filtering | https://scholar.lanfanshu.cn/scholar?q=Location-Aware+Web+Service+QoS+Prediction+via+Deep+Collaborative+Filtering | Location-Aware Web Service QoS Prediction via Deep Collaborative Filtering | Abstract—Nowadays, there is a large number of web ser-vices with similar functions, from which users choose the best
according to the quality of service (QoS). Hence, QoS prediction is a primary challenge in service recommendation. Most existing approaches model the user-service interaction relationship.However, the low-dimensional linear and the high-dimensional nonlinear relationships between users and services are seldom considered simultaneously. In addition, although location information, including the local location information of users and services, is incorporated to overcome data sparsity in most approaches. The global location information is seldom considered.Aiming at the above shortcomings, we propose a new QoS prediction model that fuses local and global location information of users and services in the interaction layer of the model. The proposed model uses a multilayer perceptron (MLP) to acquire high-dimensional nonlinear relationships of users and services,where the dot product is employed in complementing the learning of low-dimensional linear relationships. Experimental results on the real world dataset WS-Dream validate the prediction performance of the proposed model. | ['Zhaohong Jia，Li Jin，Yiwen Zhang，Chuang Liu，Kai Li，Yun Yang'] | 2022-11-04 | null | null | null | journal-2022-11 | ['collaborative-filtering'] | ['miscellaneous'] | [-4.25731272e-01 -7.70942748e-01 -6.04055703e-01 -7.15828180e-01
-1.61686108e-01 -5.03074601e-02 2.54268587e-01 1.33727426e-02
-3.10964942e-01 3.76909763e-01 3.37800056e-01 -8.24400038e-02
-6.78987920e-01 -8.84294331e-01 -1.07324757e-01 -8.14347982e-01
-2.25414902e-01 5.32961965e-01 5.26605360e-02 -3.36513996e-01
1.90004468e-01 4.65735197e-01 -1.61576426e+00 5.24226665e-01
1.06710672e+00 1.50567245e+00 4.54019189e-01 1.68829456e-01
-4.79496062e-01 1.82533115e-01 -1.94322705e-01 -1.76947102e-01
2.36993551e-01 -5.65397516e-02 -1.96558550e-01 1.10655963e-01
-2.80677557e-01 -2.89795280e-01 -3.51982504e-01 9.25404787e-01
4.79763418e-01 4.37840611e-01 5.11478245e-01 -1.33143222e+00
-6.77170813e-01 5.06817400e-01 -4.17906135e-01 1.20989449e-01
3.62465799e-01 -4.72343624e-01 1.29922426e+00 -1.03626120e+00
1.79636534e-02 1.02422845e+00 4.78782475e-01 1.75277978e-01
-9.07036543e-01 -3.49565059e-01 6.50531948e-01 6.80608213e-01
-1.51392496e+00 -1.79059669e-01 9.57017720e-01 -2.59522557e-01
7.46962249e-01 4.90274221e-01 6.76636755e-01 7.40048945e-01
-1.94410533e-02 7.71530509e-01 5.77177465e-01 -5.95413223e-02
8.56715217e-02 6.51983202e-01 6.93467781e-02 3.69326919e-01
-1.01855382e-01 -1.81043848e-01 -1.71717465e-01 -3.66450727e-01
7.03198791e-01 6.91814303e-01 -3.20244372e-01 -4.26606089e-01
-8.39224160e-01 6.15673602e-01 4.22358066e-01 5.37515640e-01
-5.77326119e-01 -7.33397543e-01 2.19674692e-01 2.66449094e-01
3.72771382e-01 -5.42433448e-02 -8.27382445e-01 -2.62250394e-01
-6.50515795e-01 -7.98939764e-02 1.09700978e+00 1.06351566e+00
4.92898941e-01 3.33554931e-02 3.53302866e-01 1.09081638e+00
5.54383934e-01 2.02408835e-01 8.32516193e-01 -4.03124034e-01
5.35180867e-01 8.40311944e-01 3.23545277e-01 -1.62907660e+00
-5.03164053e-01 -8.34889710e-01 -1.03521919e+00 -4.97382969e-01
2.20841140e-01 2.47293245e-02 -1.92914948e-01 1.23887658e+00
1.42562464e-01 4.56015825e-01 1.13922328e-01 1.27291656e+00
8.46825361e-01 9.61911142e-01 -8.80045965e-02 -5.97853720e-01
1.04116333e+00 -1.06935441e+00 -7.09175944e-01 1.78067893e-01
4.26324636e-01 -5.73804617e-01 1.07699049e+00 4.17016923e-01
-9.48935688e-01 -5.55997431e-01 -6.92936897e-01 2.98881054e-01
-6.07500315e-01 2.81545132e-01 7.77314246e-01 4.53438818e-01
-5.90508461e-01 5.45498610e-01 -3.96036237e-01 -3.22172254e-01
-6.60520643e-02 5.97163081e-01 -5.70364185e-02 -1.49478540e-01
-1.45461738e+00 5.40861011e-01 3.65484655e-01 2.95286238e-01
-3.35725956e-02 -3.77427787e-01 -4.15357679e-01 4.83329445e-01
3.43596160e-01 -2.52669752e-01 7.66930819e-01 -1.00632262e+00
-1.32881486e+00 -4.97151054e-02 -2.39170954e-01 6.38884157e-02
3.87102753e-01 3.39842364e-02 -1.29493821e+00 -3.43222439e-01
-3.11085373e-01 -2.24461555e-01 4.84546244e-01 -1.26938021e+00
-1.31308985e+00 -4.22150880e-01 -3.52057330e-02 5.33104181e-01
-9.51756775e-01 -1.68980166e-01 -7.92013526e-01 -3.69613260e-01
6.38232768e-01 -4.82399851e-01 -1.98332503e-01 -3.72409433e-01
1.04784146e-01 -2.54986852e-01 8.25213671e-01 -7.90373504e-01
1.48458958e+00 -2.07008123e+00 2.08646595e-01 5.81981301e-01
-9.28572714e-02 3.63386944e-02 -1.50886059e-01 4.16960835e-01
-4.69585843e-02 -2.53161162e-01 1.44984394e-01 -1.89089656e-01
1.49403781e-01 6.03832185e-01 -4.82389070e-02 3.15317303e-01
-3.60270590e-01 4.38016921e-01 -7.80134141e-01 -3.41597557e-01
7.05658123e-02 6.80229306e-01 -6.00861371e-01 2.85987884e-01
-1.87378954e-02 6.49235606e-01 -7.51172841e-01 9.91903245e-01
7.81683624e-01 -2.78781354e-01 2.19374374e-01 -4.35874581e-01
-5.36392666e-02 1.03650674e-01 -1.48741329e+00 1.54127824e+00
-6.22571766e-01 -1.61287084e-01 2.29931489e-01 -1.40425873e+00
1.03387523e+00 6.41930580e-01 1.14941788e+00 -1.00407887e+00
1.28988355e-01 5.61488092e-01 -5.49374484e-02 -9.18702781e-01
2.16755584e-01 -3.20293345e-02 7.33612329e-02 4.96386066e-02
-1.46409243e-01 6.69290125e-01 1.35543793e-01 -2.55047202e-01
6.04054511e-01 8.09839144e-02 6.64187893e-02 9.14012734e-03
1.04255557e+00 -4.80518758e-01 9.52980459e-01 2.63033211e-01
-8.08265209e-02 2.79932112e-01 3.28081548e-01 -6.15026891e-01
-7.79552817e-01 -7.75489271e-01 -2.46528476e-01 1.24183917e+00
5.18577039e-01 -2.97148943e-01 -1.65874258e-01 -7.50390947e-01
1.01214997e-01 4.72743899e-01 -1.48367167e-01 -1.76140398e-01
-3.04307193e-01 -8.65810156e-01 -1.39170825e-01 2.98503935e-01
2.41576463e-01 -8.86405170e-01 2.59841472e-01 4.31183010e-01
-2.27194712e-01 -1.09062588e+00 -4.67238218e-01 -2.46542260e-01
-1.02063644e+00 -5.85950673e-01 -6.65668786e-01 -9.76849318e-01
6.87072992e-01 3.35496873e-01 9.38138485e-01 9.96977314e-02
2.44362488e-01 7.79355466e-02 -4.23652649e-01 1.89431176e-01
2.45040968e-01 1.49894759e-01 6.03381574e-01 6.74558461e-01
6.04288161e-01 -9.90611076e-01 -6.54934585e-01 7.64419079e-01
-6.97943747e-01 -1.78820372e-01 8.03676069e-01 8.02515686e-01
4.37552929e-01 7.80224979e-01 8.24242055e-01 -7.68854320e-01
6.83538616e-01 -1.07266021e+00 -4.52825308e-01 1.22557946e-01
-9.94081378e-01 -4.32509214e-01 1.03931391e+00 -6.40489936e-01
-1.03236210e+00 -9.30365771e-02 -3.28838319e-01 -1.61063135e-01
-2.74731189e-01 9.91136611e-01 -7.76284993e-01 -4.90281694e-02
1.52741047e-02 4.35670644e-01 -2.36312211e-01 -1.13568115e+00
1.89182132e-01 1.03841984e+00 1.72182322e-01 -4.69037443e-01
4.17169005e-01 2.39976496e-01 -1.56868860e-01 -6.77456975e-01
-6.01882994e-01 -8.51756036e-01 -6.23313487e-01 -6.79052547e-02
3.45304400e-01 -5.51351249e-01 -1.06779444e+00 1.85938757e-02
-8.32934678e-01 6.73341155e-01 5.03519140e-02 8.80608559e-01
-3.68367642e-01 2.69226134e-01 -4.82391328e-01 -9.90364313e-01
5.17206490e-02 -1.20512402e+00 4.89062726e-01 4.08764631e-01
3.95809621e-01 -1.22280085e+00 -3.92496198e-01 3.67927551e-01
5.68633139e-01 -2.48489127e-01 9.84508693e-01 -1.02155888e+00
-3.22209954e-01 -7.60473847e-01 -4.18958187e-01 3.29840273e-01
4.44516867e-01 -4.78165001e-01 -3.06751907e-01 -2.57606894e-01
1.71513334e-01 4.94867474e-01 2.00422600e-01 1.76534787e-01
1.23093915e+00 -6.96418226e-01 -3.12384814e-01 5.74949384e-01
1.49000990e+00 7.12765396e-01 2.31710732e-01 1.98833287e-01
9.13959503e-01 8.21094990e-01 7.60602593e-01 7.88136840e-01
6.88877225e-01 7.19315231e-01 5.10787904e-01 -1.13991506e-01
6.92243338e-01 1.03366338e-01 9.00320187e-02 1.41594112e+00
-3.44089448e-01 -2.04501733e-01 -6.06199026e-01 2.71451622e-01
-2.35117698e+00 -7.54228473e-01 -1.94322020e-01 2.20992303e+00
4.13664222e-01 1.43345565e-01 2.27597564e-01 2.16915235e-01
4.84798372e-01 -1.41870797e-01 -5.91064870e-01 -2.33307570e-01
-5.19035384e-02 -6.96548700e-01 3.94389957e-01 3.37652117e-01
-9.22499120e-01 4.71386164e-01 5.14090729e+00 7.64997721e-01
-1.18232977e+00 5.57123385e-02 3.76633763e-01 -1.99055135e-01
-1.70880556e-01 -2.65923679e-01 -7.70354629e-01 7.68232524e-01
8.97196829e-01 -3.71576883e-02 7.51894474e-01 9.41690445e-01
6.59285486e-01 4.01114374e-01 -8.97900343e-01 1.12300539e+00
4.18308824e-02 -8.21584880e-01 -3.95980589e-02 9.97800082e-02
5.65798402e-01 -2.93679446e-01 1.59338385e-01 3.18359673e-01
-1.69597566e-01 -6.44332230e-01 2.62589276e-01 1.16597211e+00
2.03226477e-01 -1.04346788e+00 1.05972135e+00 7.99359918e-01
-1.29691958e+00 -6.06585145e-01 -4.73301023e-01 3.60812386e-03
3.84697855e-01 6.03685021e-01 -4.50141281e-01 9.32180226e-01
8.31083775e-01 1.05960226e+00 -2.60942280e-02 1.14014757e+00
4.88700867e-01 2.76410788e-01 -2.31082723e-01 -2.77133137e-02
2.24733680e-01 -8.77420604e-01 3.67933214e-01 1.00120294e+00
6.56631231e-01 3.10542762e-01 6.27133906e-01 2.91498929e-01
3.23599935e-01 7.70986617e-01 -1.33599922e-01 -1.52016029e-01
1.29428074e-01 1.26212227e+00 -1.75114632e-01 -2.48952895e-01
-1.00621891e+00 8.33624721e-01 -5.35195172e-02 6.32821441e-01
-6.53009236e-01 -3.01183224e-01 8.51417780e-01 3.24691385e-01
1.33536175e-01 -3.74750137e-01 -3.24742913e-01 -1.33104026e+00
1.19656637e-01 -7.55428851e-01 5.40334404e-01 -3.55094045e-01
-1.58475256e+00 5.07254839e-01 -4.16469187e-01 -1.77986455e+00
1.20519809e-02 -4.72670645e-01 -2.86160439e-01 1.18479764e+00
-1.53555000e+00 -1.00518382e+00 -1.83938369e-01 6.62032843e-01
4.22730297e-01 -5.38647771e-01 9.54626858e-01 1.13261878e+00
-6.44132912e-01 4.19002473e-01 5.95573485e-01 -2.67666221e-01
4.55210745e-01 -9.32190001e-01 -3.08200419e-01 2.57253468e-01
-1.53886437e-01 7.88647473e-01 5.39486051e-01 -4.90746439e-01
-1.43832898e+00 -7.01157272e-01 1.10299814e+00 1.60631478e-01
7.06568718e-01 -1.63979501e-01 -1.03920865e+00 3.92352253e-01
-3.81105900e-01 1.37664557e-01 1.08912480e+00 2.86452442e-01
5.09480312e-02 -6.66516364e-01 -1.19838941e+00 6.44373953e-01
8.01683307e-01 -2.12006345e-01 -1.24671429e-01 4.16179627e-01
5.56982398e-01 -3.61490063e-02 -1.35103750e+00 3.79892945e-01
8.83524656e-01 -8.07582736e-01 1.15852749e+00 -8.73135924e-01
6.67906692e-03 -3.04477751e-01 -6.61820590e-01 -1.20911562e+00
-5.98131001e-01 4.99455407e-02 -5.18317282e-01 1.23801076e+00
4.77954417e-01 -7.64870644e-01 9.75147963e-01 8.95532250e-01
6.57990947e-02 -1.04267728e+00 -8.26573610e-01 -5.73294461e-01
-6.46901667e-01 -3.00620884e-01 9.86976683e-01 1.07970905e+00
2.26935133e-01 3.50628942e-01 -7.55465627e-01 3.11917931e-01
3.47378880e-01 2.59518743e-01 2.89640814e-01 -1.75823629e+00
-5.24611354e-01 -4.84743744e-01 -3.41309547e-01 -1.37876129e+00
-1.15450509e-01 -6.89164281e-01 -3.19612086e-01 -1.70011246e+00
-1.75357699e-01 -8.34478855e-01 -9.85816896e-01 1.00107759e-01
3.00234348e-01 -1.48791999e-01 -2.29244933e-01 5.53520024e-01
-6.75872505e-01 6.47155702e-01 1.32949126e+00 4.67470959e-02
-5.96952140e-01 8.39624345e-01 -7.12364256e-01 7.94785321e-01
5.58212638e-01 -2.03126967e-01 -5.23104608e-01 -3.56857151e-01
1.57249108e-01 4.65187550e-01 -3.79908711e-01 -7.44829357e-01
4.63166416e-01 -5.48626125e-01 4.81853694e-01 -5.72014451e-01
7.60102272e-01 -1.62287664e+00 6.25302270e-02 9.60181504e-02
-1.46197945e-01 3.40292491e-02 -3.52753550e-01 6.74275577e-01
-2.75522590e-01 -5.17048053e-02 3.18210244e-01 -7.17274845e-02
-7.41754174e-01 9.09846008e-01 -1.38111368e-01 -6.90964997e-01
7.37068951e-01 -1.26639321e-01 2.93949991e-01 -5.68824410e-01
-1.11484265e+00 2.90152401e-01 -3.47988121e-02 7.86709189e-01
6.98826790e-01 -1.60677505e+00 -3.70079935e-01 3.45918208e-01
1.19988047e-01 -3.08729112e-01 3.17397058e-01 8.42637777e-01
-1.49238020e-01 3.73306632e-01 -1.82131439e-01 -2.99294323e-01
-1.01916182e+00 8.27993095e-01 2.39542708e-01 3.57118025e-02
-3.57473224e-01 4.71094489e-01 -2.35072494e-01 -5.80990493e-01
6.06202543e-01 6.85478076e-02 -1.03380930e+00 2.28670686e-01
5.06993234e-01 4.63457525e-01 8.96654278e-02 -1.10961890e+00
-2.33393520e-01 4.55126643e-01 -3.52501795e-02 1.25703197e-02
1.56218517e+00 -5.72358489e-01 -3.28310490e-01 6.59433663e-01
1.41142714e+00 -1.19604282e-01 -9.32877600e-01 -7.42448866e-01
2.20051721e-01 -9.20798957e-01 1.46950036e-01 -7.10101366e-01
-1.32743669e+00 7.47034132e-01 6.85075104e-01 6.54809475e-01
1.41573071e+00 -4.39568073e-01 1.03255570e+00 2.73916125e-01
4.33679938e-01 -1.16575134e+00 -2.49308765e-01 3.72484833e-01
6.57916486e-01 -1.20196629e+00 -3.14155906e-01 -3.98779780e-01
-6.20291412e-01 9.66550112e-01 6.01400614e-01 6.65432513e-02
1.23959887e+00 -1.09810002e-01 -1.75315852e-03 3.23212475e-01
-5.15301704e-01 -5.31918481e-02 5.42795539e-01 3.70454460e-01
4.61569846e-01 1.80854395e-01 -6.85369074e-01 1.38321340e+00
6.69795722e-02 -1.94795221e-01 -9.08748955e-02 8.25153887e-01
-3.34131420e-01 -1.36051631e+00 -2.39495769e-01 6.29717946e-01
-4.19335663e-01 -6.95109144e-02 2.91539431e-01 2.56490801e-02
3.17665577e-01 1.17828417e+00 8.62042233e-02 -7.17178404e-01
6.56543076e-01 -1.39246598e-01 -1.75871372e-01 -2.06681639e-01
-2.42893487e-01 3.56156647e-01 -7.73361176e-02 -5.29971540e-01
4.39927578e-02 -5.95535159e-01 -1.35174477e+00 -3.19721133e-01
-2.95088172e-01 5.08714497e-01 1.13189316e+00 9.89787757e-01
4.28865403e-01 6.08022988e-01 1.08822131e+00 -8.52247119e-01
-5.10148525e-01 -7.57596195e-01 -1.16893470e+00 4.29142237e-01
1.95152536e-01 -7.45275974e-01 -3.01263779e-01 -2.23149419e-01] | [10.049534797668457, 5.67059850692749] |
6a2010e4-0556-4d21-ac86-870ce710c9c2 | cross-directional-feature-fusion-network-for | 2010.14014 | null | https://arxiv.org/abs/2010.14014v2 | https://arxiv.org/pdf/2010.14014v2.pdf | Cross-directional Feature Fusion Network for Building Damage Assessment from Satellite Imagery | Fast and effective responses are required when a natural disaster (e.g., earthquake, hurricane, etc.) strikes. Building damage assessment from satellite imagery is critical before an effective response is conducted. High-resolution satellite images provide rich information with pre- and post-disaster scenes for analysis. However, most existing works simply use pre- and post-disaster images as input without considering their correlations. In this paper, we propose a novel cross-directional fusion strategy to better explore the correlations between pre- and post-disaster images. Moreover, the data augmentation method CutMix is exploited to tackle the challenge of hard classes. The proposed method achieves state-of-the-art performance on a large-scale building damage assessment dataset -- xBD. | ['Chen Chen', 'Taojiannan Yang', 'Sijie Zhu', 'Yu Shen'] | 2020-10-27 | null | null | null | null | ['2d-semantic-segmentation'] | ['computer-vision'] | [ 4.23424006e-01 -6.55269325e-01 1.40732929e-01 -2.71643579e-01
-1.25714958e+00 -3.59736755e-02 5.10982394e-01 5.70667088e-01
-6.26840472e-01 5.84552109e-01 5.54910898e-01 3.30762081e-02
-3.24684978e-01 -1.34634483e+00 -2.09994048e-01 -1.10889828e+00
-8.08091536e-02 2.95559615e-01 -1.51999444e-02 -6.13221765e-01
1.39404520e-01 5.79472661e-01 -1.55629730e+00 2.86624700e-01
8.46140623e-01 7.24276721e-01 6.11564815e-01 4.62807775e-01
4.27138984e-01 4.61409301e-01 -4.03628170e-01 2.71196216e-01
3.26335251e-01 -5.68904616e-02 -8.73985171e-01 3.85952383e-01
1.08722523e-02 -6.89000785e-01 -4.04591352e-01 9.46915329e-01
8.32366526e-01 1.18981309e-01 5.76638758e-01 -9.13120747e-01
-2.46757045e-01 3.98291707e-01 -9.58907783e-01 5.82042396e-01
3.53825629e-01 3.27061862e-01 7.69256711e-01 -8.82617474e-01
2.93607384e-01 1.22350347e+00 6.45840347e-01 -2.26361006e-01
-1.23110771e+00 -5.08265316e-01 -4.43157963e-02 3.70534420e-01
-1.60078943e+00 -1.23167224e-01 8.49962592e-01 -5.74096322e-01
7.34175563e-01 2.92859793e-01 4.40326780e-01 6.53506398e-01
-1.56332478e-01 6.45594597e-01 1.36642587e+00 -2.64521867e-01
7.62782469e-02 -6.26436293e-01 -3.26506910e-03 9.80630293e-02
5.38938120e-02 1.41850919e-01 -2.83849418e-01 -1.47564784e-01
3.15528870e-01 3.80936891e-01 -4.93205339e-01 4.65158910e-01
-1.17433834e+00 7.75194705e-01 7.50595748e-01 3.46992016e-01
-1.00131011e+00 -3.32816929e-01 3.49308342e-01 -3.53073329e-02
5.65965593e-01 1.80031415e-02 -9.57397670e-02 2.44621232e-01
-1.16922677e+00 2.65258551e-01 -1.41522944e-01 1.22186780e-01
9.37505066e-01 -2.38167103e-02 1.74586624e-02 8.49857450e-01
1.56791613e-01 6.34139895e-01 6.59033731e-02 -5.20485997e-01
7.97730088e-01 6.11130714e-01 -9.67004988e-03 -1.55973673e+00
-7.91440666e-01 -4.05642539e-01 -1.62382293e+00 7.11407661e-02
8.54248106e-02 7.24330693e-02 -8.79997671e-01 1.42669010e+00
6.14708602e-01 1.11501448e-01 -9.90090966e-02 1.23864532e+00
5.28270781e-01 8.40895355e-01 2.95008868e-01 -4.37281787e-01
1.28215396e+00 -3.60268354e-01 -6.07134581e-01 -4.47531670e-01
1.81578889e-01 -6.44936383e-01 1.13217342e+00 3.79496634e-01
-5.84501028e-01 -4.87294137e-01 -7.77102649e-01 3.27773035e-01
-1.34524986e-01 1.18011154e-01 2.33632401e-01 3.42579372e-02
-6.51765883e-01 2.79967040e-01 -7.48174369e-01 -4.16036546e-01
4.02045548e-01 4.63104621e-02 -5.95134854e-01 -6.51597142e-01
-1.13723874e+00 7.40995824e-01 5.01738012e-01 3.75948370e-01
-1.12309444e+00 -6.30556226e-01 -5.10799587e-01 -2.04065084e-01
3.57230246e-01 -3.76211017e-01 6.81467950e-01 -1.03563257e-01
-3.72975320e-01 9.67190444e-01 4.33439642e-01 -2.38531902e-01
2.68478453e-01 -3.63391191e-01 -2.77379513e-01 4.00283277e-01
3.30850840e-01 4.16784823e-01 5.45711517e-01 -1.32762277e+00
-7.41208136e-01 -8.95207524e-01 -2.22913157e-02 3.63043368e-01
-4.83764172e-01 3.33533764e-01 3.13833728e-02 -1.02197945e+00
4.28491652e-01 -7.94956625e-01 -4.56366509e-01 -4.28410649e-01
-5.58027983e-01 1.20789602e-01 1.01177704e+00 -1.06318629e+00
1.51412117e+00 -2.09927130e+00 8.18436742e-02 -4.77276817e-02
-2.69547030e-02 2.73088396e-01 -1.94675907e-01 8.37450266e-01
-2.42819175e-01 1.80768803e-01 -8.18652630e-01 -7.49889121e-04
-3.32430363e-01 4.67657983e-01 -4.18657929e-01 5.49165666e-01
1.58382446e-01 1.90395251e-01 -7.15130925e-01 -5.10123551e-01
2.32189715e-01 4.39718872e-01 -1.36437654e-01 3.43982518e-01
2.94069141e-01 6.88470900e-01 -4.64920372e-01 7.75647044e-01
9.46780860e-01 2.55657256e-01 -2.17401892e-01 -3.37972790e-01
-3.55419278e-01 -4.57156561e-02 -1.18566883e+00 1.33988380e+00
-2.19948456e-01 7.97254518e-02 2.48570368e-01 -1.29926431e+00
9.12842274e-01 3.04107755e-01 7.35953212e-01 -6.89562500e-01
-4.13122587e-02 1.10266760e-01 -4.40270603e-01 -6.54313505e-01
4.80617464e-01 -5.01011014e-01 -2.57724136e-01 6.36887550e-01
-5.92095196e-01 -1.91994995e-01 2.19346344e-01 7.70947710e-02
1.16820073e+00 -3.15526068e-01 2.91394264e-01 -1.42767698e-01
4.04990703e-01 1.95626616e-01 7.48923838e-01 3.58015954e-01
-1.51915401e-01 9.82249737e-01 -5.66687947e-03 -5.40485978e-01
-8.98062289e-01 -8.48496139e-01 -1.44124731e-01 7.31461525e-01
-1.69452891e-01 -1.71062108e-02 -4.94156092e-01 -4.27339405e-01
-3.94336432e-02 5.35802484e-01 -4.15155500e-01 -7.68143460e-02
-5.07805526e-01 -1.54536009e+00 6.95188463e-01 4.92148578e-01
1.01096642e+00 -8.78264010e-01 -7.01137662e-01 4.31500167e-01
-8.99696767e-01 -9.05845642e-01 -7.53866211e-02 -8.40582401e-02
-9.24691319e-01 -1.23509538e+00 -5.90273917e-01 -4.75676715e-01
7.25951672e-01 7.78131247e-01 8.49772990e-01 3.35592657e-01
-5.28712749e-01 3.28138143e-01 -5.24102271e-01 1.38642952e-01
5.66051938e-02 -1.98678404e-01 7.09272027e-02 3.34431008e-02
7.37241730e-02 -9.40639019e-01 -7.69096255e-01 4.91161436e-01
-1.28157926e+00 -6.11320660e-02 6.23997629e-01 8.58281434e-01
8.03011358e-01 7.40094960e-01 5.42747855e-01 -1.21130943e-01
5.00591457e-01 -6.33268774e-01 -6.99772984e-02 1.64520100e-01
-3.83225888e-01 -3.12344134e-01 2.20268637e-01 -1.18164942e-01
-1.23713648e+00 2.77612835e-01 -1.74846157e-01 1.56783730e-01
-6.62025332e-01 1.24165583e+00 -2.64540344e-01 4.04798537e-01
6.69035017e-01 1.77101478e-01 -7.09918499e-01 -8.35003376e-01
-3.36835273e-02 9.23885822e-01 9.16240215e-01 -5.27754366e-01
9.94991720e-01 7.10005343e-01 -9.72573739e-03 -1.03388131e+00
-9.58739460e-01 -9.13849056e-01 -8.61066580e-01 -3.11369270e-01
8.15465569e-01 -1.23459840e+00 -1.67254210e-02 1.03242099e+00
-9.37306046e-01 -9.12853479e-02 -9.45957452e-02 5.60254097e-01
-9.39636081e-02 4.94030982e-01 -3.91479999e-01 -8.96208346e-01
-4.13275510e-01 -7.65518188e-01 1.03671730e+00 9.51424837e-02
2.65747756e-01 -4.24454719e-01 3.67087424e-01 7.53088355e-01
3.18138450e-01 7.07438350e-01 5.83678186e-01 -1.33885950e-01
-3.77869666e-01 -2.52028078e-01 -4.17692035e-01 3.21796030e-01
4.91317451e-01 -7.72170648e-02 -6.48394763e-01 -3.66817355e-01
6.68143705e-02 -4.01453078e-01 1.11135614e+00 3.33991408e-01
7.99008489e-01 -4.49635714e-01 -1.55926989e-02 2.09160194e-01
1.57500875e+00 -1.30570769e-01 9.34986591e-01 5.53127170e-01
5.67494929e-01 8.32133830e-01 1.20625210e+00 1.04835188e+00
7.49400139e-01 6.79146767e-01 6.17129266e-01 -3.06688249e-01
3.49373892e-02 1.53841957e-01 1.41841173e-01 5.21659255e-01
-2.71237403e-01 -1.77460775e-01 -1.64445770e+00 1.04538810e+00
-1.70688415e+00 -1.14563227e+00 -6.18638456e-01 2.17933536e+00
7.49803066e-01 -1.84320718e-01 1.25462431e-02 7.04696476e-01
7.54203737e-01 3.75263512e-01 -1.93198368e-01 5.20882130e-01
-5.68474829e-01 1.44602507e-01 2.70119876e-01 2.32829392e-01
-1.50884557e+00 5.59009075e-01 5.27039051e+00 6.76782846e-01
-1.02688658e+00 2.29298204e-01 6.68336451e-01 1.35129794e-01
-8.84550139e-02 -4.98566218e-02 -4.32305753e-01 2.20585823e-01
4.82986838e-01 1.27724409e-01 1.34328499e-01 4.64151978e-01
6.28671527e-01 -6.79165483e-01 -2.02736527e-01 8.48141015e-01
-3.91249843e-02 -7.30131269e-01 -3.87899637e-01 7.68562853e-02
7.69522548e-01 1.20358445e-01 -1.63809843e-02 -9.95929316e-02
2.97278911e-01 -5.80450177e-01 2.79777974e-01 5.13110638e-01
5.07381916e-01 -9.14229691e-01 7.67155766e-01 6.42874181e-01
-1.40380001e+00 -2.18575940e-01 -5.55538945e-02 -3.50800633e-01
4.64510858e-01 1.06415689e+00 -6.66340828e-01 8.33400249e-01
9.56318378e-01 5.58119357e-01 -6.63461387e-01 1.09497392e+00
-6.61095440e-01 8.14065874e-01 -6.23872876e-01 9.92838323e-01
2.26052135e-01 -1.21108569e-01 5.15525937e-01 1.02674925e+00
6.06567800e-01 8.11285973e-01 7.21340954e-01 1.73111722e-01
3.95477861e-01 -3.70983407e-02 -5.61447322e-01 2.41301119e-01
1.76988319e-01 1.16542459e+00 -5.90139508e-01 -2.23128885e-01
-1.95041336e-02 7.21843600e-01 -5.23663126e-02 1.22056387e-01
-5.79830170e-01 8.04618746e-02 5.99501133e-01 3.29875052e-01
7.76435137e-02 -5.25838137e-01 -3.34065765e-01 -9.67634082e-01
-1.46128237e-01 -9.76077735e-01 7.65151858e-01 -8.33518386e-01
-1.20657086e+00 4.68383044e-01 3.31148356e-01 -1.32456911e+00
1.07737653e-01 1.69502363e-01 -8.19113731e-01 6.82224154e-01
-1.66262138e+00 -1.47355497e+00 -7.52463996e-01 6.86520100e-01
3.94899696e-01 2.89413869e-01 6.16709590e-01 6.73464000e-01
-8.93240631e-01 -1.39637500e-01 -2.03165978e-01 9.73400399e-02
5.83616853e-01 -8.69629323e-01 4.86176498e-02 1.40809417e+00
-3.26846182e-01 -2.67239008e-02 7.69675493e-01 -9.88862872e-01
-8.91591012e-01 -1.24092007e+00 8.07823360e-01 3.48940492e-01
3.84478629e-01 3.49011123e-01 -1.09306276e+00 1.85587302e-01
6.53823987e-02 -2.90605724e-02 6.34730399e-01 -9.70720723e-02
-1.11842170e-01 -3.06758672e-01 -1.27797544e+00 1.97416618e-01
7.30663359e-01 -6.00918114e-01 -5.13399839e-01 4.76651400e-01
2.84120947e-01 1.85337327e-02 -8.45544457e-01 6.81089103e-01
8.48967209e-02 -8.61790121e-01 1.29112244e+00 -2.73915470e-01
5.86213470e-01 -5.66818357e-01 -6.44604802e-01 -1.36699927e+00
-5.38012981e-01 -3.01115997e-02 4.67013478e-01 1.39251828e+00
-2.05569908e-01 -2.43026286e-01 2.63782978e-01 4.62051332e-01
-1.89238325e-01 -2.86808223e-01 -9.28485096e-01 -5.87299287e-01
-3.63940150e-01 -5.59260428e-01 5.42959571e-01 1.08902872e+00
-4.19756263e-01 6.78873211e-02 -6.27660811e-01 7.97210515e-01
8.88500214e-01 1.18093513e-01 7.01843023e-01 -1.24819434e+00
1.73416451e-01 -1.55041248e-01 -3.53772759e-01 -3.57988894e-01
-3.67329121e-01 -3.83176744e-01 1.99145347e-01 -1.69611847e+00
4.30472016e-01 -5.34761786e-01 -4.35011387e-01 9.83246148e-01
-5.52862346e-01 5.46941400e-01 1.62083596e-01 3.53224397e-01
-1.15925081e-01 7.82505393e-01 7.50224113e-01 -4.35298443e-01
-9.58058685e-02 -5.30957393e-02 -3.88683945e-01 6.66901469e-01
1.22642636e+00 -9.23352957e-01 -5.56390844e-02 -5.05241513e-01
1.99665263e-01 2.28892595e-01 7.36468077e-01 -1.25245297e+00
1.12687744e-01 -6.01533651e-01 -1.46339133e-01 -1.21041536e+00
1.29137173e-01 -6.50343060e-01 4.89485472e-01 5.45213819e-01
6.99403137e-02 1.05647676e-01 1.26651838e-01 5.24594247e-01
-6.13852501e-01 -7.54974186e-02 8.40532541e-01 -1.26218135e-02
-6.05726182e-01 3.57577950e-01 -4.23202366e-01 -7.27108419e-02
8.73666525e-01 1.30606562e-01 -4.34807509e-01 -2.37548143e-01
-6.11272216e-01 4.10514355e-01 3.68417352e-01 2.21657097e-01
8.63568604e-01 -1.41918492e+00 -1.38934624e+00 -5.88550940e-02
1.58938825e-01 2.91865736e-01 9.84414697e-01 1.10857511e+00
-4.27026957e-01 -1.56063184e-01 -4.07657355e-01 -4.84183252e-01
-1.51762795e+00 4.17759150e-01 -1.66678891e-01 -7.40421474e-01
-4.79443967e-01 5.38060308e-01 7.16262832e-02 -4.17030573e-01
-2.82493323e-01 -6.43019080e-02 -4.54447806e-01 6.63612843e-01
8.73707294e-01 5.62967658e-01 1.35423020e-01 -1.06069720e+00
-4.74556059e-01 5.97499788e-01 3.13464165e-01 -1.04558602e-01
1.90915227e+00 -3.21187377e-01 -3.20263296e-01 1.27563044e-01
8.05633962e-01 -3.84735733e-01 -1.09243417e+00 -4.25827116e-01
-2.48202700e-02 -8.32338214e-01 7.00210154e-01 -7.01777339e-01
-1.34656894e+00 1.09642386e+00 8.17255199e-01 7.03785662e-03
1.87725592e+00 -1.42020777e-01 1.03011072e+00 4.00771767e-01
3.77938390e-01 -1.27363503e+00 2.37616166e-01 1.52627140e-01
1.59342253e+00 -1.36075330e+00 4.85419631e-01 -1.90421790e-01
-6.26666367e-01 8.07237744e-01 3.06767672e-01 2.10232437e-01
5.03278792e-01 1.77217290e-01 -6.28098324e-02 -2.35609472e-01
-3.38438094e-01 -6.41953588e-01 5.12016565e-02 6.74397886e-01
-2.20968828e-01 2.75064409e-01 -1.52123809e-01 3.52768332e-01
3.30497980e-01 -1.93491012e-01 3.75297487e-01 1.17204821e+00
-7.74091780e-01 -9.09648299e-01 -1.13759696e+00 2.28035286e-01
-2.45152816e-01 -1.16036586e-01 -1.49731949e-01 5.57410777e-01
2.38389999e-01 1.21768332e+00 -2.54035324e-01 -6.01491213e-01
4.90167767e-01 -4.14307952e-01 2.16318835e-02 -4.72202450e-01
-6.16109014e-01 2.52545267e-01 -1.14595048e-01 -2.78836876e-01
-8.35038722e-01 -9.82687831e-01 -1.19080329e+00 -3.43854904e-01
-2.29716420e-01 -2.84187883e-01 5.06474018e-01 8.29015851e-01
6.83692619e-02 2.96808779e-01 1.12687457e+00 -1.07917237e+00
-4.15154576e-01 -1.11157620e+00 -6.35606587e-01 5.48386931e-01
2.40793616e-01 -6.06925666e-01 -2.42273524e-01 1.13966845e-01] | [9.555876731872559, -1.3296061754226685] |
20e957e2-1b88-4d4b-af52-526b13f131da | rhythmnet-end-to-end-heart-rate-estimation | 1910.11515 | null | https://arxiv.org/abs/1910.11515v2 | https://arxiv.org/pdf/1910.11515v2.pdf | RhythmNet: End-to-end Heart Rate Estimation from Face via Spatial-temporal Representation | Heart rate (HR) is an important physiological signal that reflects the physical and emotional status of a person. Traditional HR measurements usually rely on contact monitors, which may cause inconvenience and discomfort. Recently, some methods have been proposed for remote HR estimation from face videos; however, most of them focus on well-controlled scenarios, their generalization ability into less-constrained scenarios (e.g., with head movement, and bad illumination) are not known. At the same time, lacking large-scale HR databases has limited the use of deep models for remote HR estimation. In this paper, we propose an end-to-end RhythmNet for remote HR estimation from the face. In RyhthmNet, we use a spatial-temporal representation encoding the HR signals from multiple ROI volumes as its input. Then the spatial-temporal representations are fed into a convolutional network for HR estimation. We also take into account the relationship of adjacent HR measurements from a video sequence via Gated Recurrent Unit (GRU) and achieves efficient HR measurement. In addition, we build a large-scale multi-modal HR database (named as VIPL-HR, available at 'http://vipl.ict.ac.cn/view_database.php?id=15'), which contains 2,378 visible light videos (VIS) and 752 near-infrared (NIR) videos of 107 subjects. Our VIPL-HR database contains various variations such as head movements, illumination variations, and acquisition device changes, replicating a less-constrained scenario for HR estimation. The proposed approach outperforms the state-of-the-art methods on both the public-domain and our VIPL-HR databases. | ['Shiguang Shan', 'Xilin Chen', 'Xuesong Niu', 'Hu Han'] | 2019-10-25 | null | null | null | null | ['heart-rate-estimation'] | ['medical'] | [-6.27975762e-02 -3.95773619e-01 -6.13371693e-02 -4.82877612e-01
-6.54699862e-01 1.55640754e-03 -1.48556352e-01 -5.10483205e-01
-8.59516561e-02 7.47714281e-01 3.63011271e-01 2.21378312e-01
2.20183551e-01 -4.08900440e-01 -2.24125251e-01 -8.60667706e-01
-2.67912876e-02 -5.68622112e-01 -5.40258586e-01 -1.68015659e-01
-4.65618968e-02 5.42726517e-01 -1.53557956e+00 -1.75770074e-02
3.71016532e-01 1.29088271e+00 -1.51660219e-01 5.11906505e-01
6.04072392e-01 6.15145385e-01 -7.37685919e-01 1.21000580e-01
1.22441076e-01 -1.00805104e+00 -2.97912896e-01 -2.02594131e-01
3.70889217e-01 -4.79336321e-01 -8.38070810e-01 5.16083717e-01
1.41910934e+00 3.88631701e-01 -1.25824939e-02 -1.03314757e+00
-7.44413078e-01 3.66155729e-02 -7.56285667e-01 6.39616132e-01
6.87099576e-01 5.07787824e-01 4.75254834e-01 -8.04382682e-01
3.65520090e-01 9.82825220e-01 6.40902042e-01 7.69217491e-01
-8.27346563e-01 -1.00578237e+00 -7.09088221e-02 5.27456105e-01
-1.63343048e+00 -8.44466865e-01 1.06145990e+00 -1.60620227e-01
7.39733458e-01 4.50400889e-01 8.11645925e-01 1.42810071e+00
1.35652930e-01 3.13863486e-01 1.26910007e+00 7.93023556e-02
7.33262720e-03 -8.94407108e-02 -1.74251676e-01 3.04530799e-01
-2.32353017e-01 1.90838978e-01 -7.20770180e-01 1.35691747e-01
9.31249976e-01 2.47373581e-01 -1.04591072e+00 4.08939868e-01
-1.17146957e+00 4.13563788e-01 5.52403033e-01 4.14404750e-01
-3.20820183e-01 -5.50277866e-02 5.84529936e-01 3.96748960e-01
4.66373473e-01 5.49288988e-02 -4.51586396e-01 -9.22305882e-02
-8.12049687e-01 -2.45652676e-01 5.56404829e-01 6.81606829e-01
4.08001840e-01 2.11229011e-01 -5.35317242e-01 9.57931399e-01
1.46867365e-01 5.57145476e-01 6.50616169e-01 -7.07909882e-01
2.74774730e-01 -1.78670809e-02 -8.35711136e-02 -1.09723091e+00
-7.82764196e-01 -4.89866674e-01 -1.22211611e+00 -2.67537892e-01
-5.41605204e-02 -2.98511088e-01 -4.85888541e-01 1.97866917e+00
4.09745246e-01 6.57764316e-01 -1.02082320e-01 1.35086262e+00
1.45157349e+00 6.25866950e-01 -7.44676590e-03 -1.16405237e+00
1.29161036e+00 -7.19060600e-01 -1.01598227e+00 3.65139134e-02
4.63111475e-02 -5.45566022e-01 9.46462452e-01 4.27208930e-01
-8.99410188e-01 -9.80977535e-01 -8.33637357e-01 -1.20240122e-01
2.26744190e-02 2.31376648e-01 3.53593946e-01 3.64477664e-01
-1.01383865e+00 5.60016453e-01 -3.22462887e-01 -3.48948210e-01
3.06380510e-01 9.00725722e-02 -3.48714232e-01 -1.06191576e-01
-1.56861198e+00 6.78048909e-01 -1.08310968e-01 1.05998302e+00
-9.25869882e-01 -5.06117761e-01 -8.54800284e-01 -2.19114885e-01
3.82669896e-01 -5.21076858e-01 9.18599367e-01 -7.43318677e-01
-1.99305868e+00 7.36037195e-01 -2.77483106e-01 1.96890563e-01
5.15154064e-01 -2.83642620e-01 -9.67032790e-01 2.75490969e-01
-5.07636309e-01 2.12568358e-01 9.80471551e-01 -7.85036981e-01
2.04107627e-01 -5.39933980e-01 -2.80120939e-01 5.22797883e-01
-1.13690935e-01 2.32725263e-01 -4.22867388e-01 -4.19954509e-01
-1.14926822e-01 -8.08794022e-01 1.50738463e-01 3.37021872e-02
-3.40529084e-01 -9.40251872e-02 6.08886659e-01 -1.13742876e+00
1.26439345e+00 -2.22754335e+00 1.03153862e-01 -6.00467026e-02
3.51792783e-01 1.09427877e-01 -1.07604042e-01 6.57414719e-02
-3.37615281e-01 1.44282440e-02 1.05343275e-01 -2.36310944e-01
-2.70759910e-01 -2.43954822e-01 -6.47015646e-02 1.13322759e+00
-3.22485656e-01 7.79031515e-01 -4.93258387e-01 -5.37237704e-01
4.62005138e-01 1.02241194e+00 1.46450130e-02 6.99701369e-01
4.65506643e-01 1.07186055e+00 -1.86331436e-01 6.42183363e-01
6.98322296e-01 7.88462907e-02 -4.29935940e-02 -5.56749046e-01
-4.74280491e-02 -1.54483512e-01 -9.63059783e-01 1.77233887e+00
-6.70559466e-01 6.69536650e-01 -2.18230709e-01 -7.51578152e-01
1.20958745e+00 7.96526551e-01 7.60572731e-01 -9.61415231e-01
2.83043802e-01 -1.30697742e-01 -1.23089865e-01 -7.94527471e-01
-1.37865543e-01 -1.88278303e-01 7.58232027e-02 1.17641263e-01
-7.81856775e-02 2.93312728e-01 -3.45928907e-01 -4.94831026e-01
8.15309048e-01 1.46745518e-01 4.67916220e-01 2.74067611e-01
6.40683889e-01 -1.00792348e+00 9.64979291e-01 3.53920966e-01
-7.01079726e-01 9.11655247e-01 2.59363145e-01 -4.50755239e-01
-4.61965412e-01 -7.17811584e-01 -3.26629162e-01 8.67236078e-01
1.30933940e-01 -1.04220361e-01 -4.84004110e-01 -2.02378884e-01
-2.72097588e-01 2.65767992e-01 -7.66040206e-01 -3.91674548e-01
-5.37823737e-01 -8.72126043e-01 5.63939750e-01 3.03124040e-01
8.71311903e-01 -1.61740780e+00 -7.44485438e-01 3.98195945e-02
-6.05937064e-01 -1.14169288e+00 -5.93555331e-01 -3.48264396e-01
-6.53953910e-01 -9.79275584e-01 -8.61453831e-01 -2.43041813e-01
1.74799100e-01 2.97338039e-01 1.09403980e+00 4.50476333e-02
-8.83628964e-01 3.69638860e-01 -2.12924466e-01 -1.31791845e-01
3.97546291e-01 -3.12887818e-01 8.12790170e-02 2.81995058e-01
2.95156628e-01 -6.24110222e-01 -1.23659849e+00 4.37067658e-01
-3.39059711e-01 -1.70910344e-01 3.49017560e-01 5.67861557e-01
6.45573974e-01 -1.24560826e-01 8.29924405e-01 -3.39405000e-01
4.31462467e-01 -4.78894740e-01 -2.93243587e-01 1.19901992e-01
-3.76307249e-01 -6.62367642e-01 5.89477718e-01 -5.15296578e-01
-1.19012296e+00 -1.63745180e-01 -1.79024637e-01 -8.71365905e-01
-3.44467193e-01 2.08256513e-01 -3.71926785e-01 4.63219471e-02
6.36316121e-01 1.54899359e-01 -9.94323939e-02 -3.11833203e-01
7.49399737e-02 7.30468690e-01 6.04857981e-01 -1.12285495e-01
6.00966811e-01 2.28139952e-01 -5.54280542e-03 -1.02695227e+00
-8.59295666e-01 -4.70228195e-01 -4.35873240e-01 -6.61105990e-01
9.82882917e-01 -1.37056947e+00 -1.04867637e+00 6.42043471e-01
-8.34189832e-01 -4.23659354e-01 2.56274324e-02 7.42906928e-01
-6.46908820e-01 2.04750165e-01 -8.29795957e-01 -9.02438223e-01
-8.34691346e-01 -6.86154068e-01 8.73574972e-01 7.30743885e-01
1.36023372e-01 -5.44014692e-01 1.63231477e-01 5.94171286e-01
6.61043227e-01 7.01233149e-01 8.14594179e-02 7.34481812e-02
-3.46788429e-02 -6.64668977e-02 -6.52700104e-03 3.13584775e-01
2.50628054e-01 -2.03768507e-01 -1.61352825e+00 -4.23059702e-01
3.24421495e-01 -5.10200024e-01 5.91523588e-01 6.11765087e-01
1.65589690e+00 -1.89261004e-01 1.22245811e-01 9.45944250e-01
1.41290748e+00 1.81027815e-01 1.11261892e+00 -1.68449134e-01
8.96305978e-01 5.61739802e-01 6.40896142e-01 7.20107496e-01
1.37642667e-01 7.99558759e-01 2.36593157e-01 -5.37151039e-01
-2.69307066e-02 1.21757962e-01 5.41150451e-01 7.58822143e-01
-5.86538851e-01 -8.65324736e-02 -3.68656576e-01 -5.47344498e-02
-1.49170387e+00 -1.31170332e+00 1.06150925e-01 2.42334795e+00
8.92640948e-01 -4.84302521e-01 2.27506816e-01 -5.52090351e-04
8.99202824e-01 4.57130730e-01 -1.03256166e+00 -2.72499681e-01
-9.19065326e-02 1.95874423e-01 3.67670469e-02 1.34206906e-01
-9.61578548e-01 3.95594060e-01 4.90332031e+00 3.36524606e-01
-1.44459784e+00 2.60144889e-01 9.24358249e-01 -6.02167726e-01
1.28313571e-01 -4.79479223e-01 -4.93989974e-01 5.05200565e-01
1.18823433e+00 -4.61089090e-02 7.82416940e-01 5.41999459e-01
8.04441035e-01 1.50132785e-02 -9.67984319e-01 1.82493544e+00
5.50808966e-01 -5.25444746e-01 -8.03202331e-01 -4.45682928e-02
1.70773670e-01 -5.77432383e-03 1.93044625e-03 5.10530353e-01
-7.27492332e-01 -1.30728626e+00 1.02093525e-01 9.34488237e-01
1.47355235e+00 -8.18501174e-01 8.27408969e-01 -7.91506320e-02
-1.34542501e+00 -6.15393110e-02 -4.20158058e-01 1.83843106e-01
-5.43307215e-02 6.67649329e-01 -1.89121254e-02 5.44617593e-01
9.63979185e-01 9.67025459e-01 -4.43947703e-01 7.95425713e-01
-2.10724756e-01 4.46078092e-01 -6.09105304e-02 4.90235180e-01
-6.16281331e-01 -1.79704323e-01 2.26308942e-01 1.04976892e+00
3.69260222e-01 6.80240989e-01 8.71560574e-02 7.43948936e-01
-1.85936242e-01 1.41366020e-01 -7.52161503e-01 2.72846699e-01
2.51684189e-01 1.65382981e+00 -1.46277070e-01 1.02042174e-02
-5.03655970e-01 1.06175590e+00 7.15925917e-03 5.92234135e-01
-1.38151097e+00 -5.09793222e-01 7.24885464e-01 -2.93901443e-01
-1.68221787e-01 6.48781508e-02 2.41989538e-01 -1.41004157e+00
1.44678205e-01 -9.39133644e-01 5.27216256e-01 -1.07264245e+00
-1.15778029e+00 8.41463745e-01 -2.32152879e-01 -1.26986110e+00
3.61101050e-03 2.90823616e-02 -6.51958227e-01 1.17398119e+00
-1.63378477e+00 -6.07788265e-01 -1.18486845e+00 1.08991623e+00
5.04757941e-01 1.95311308e-01 7.10380137e-01 6.49915993e-01
-1.28779387e+00 8.30832601e-01 -4.74257678e-01 8.37721899e-02
1.14319265e+00 -6.88463628e-01 -2.00314060e-01 5.87903917e-01
-3.66346717e-01 4.79457587e-01 5.78897238e-01 -2.01858386e-01
-1.52449191e+00 -1.07293379e+00 5.09971380e-01 -1.06506258e-01
6.58059418e-02 -3.81286889e-01 -9.37536240e-01 5.57774007e-01
9.88831595e-02 7.68070996e-01 7.52907753e-01 2.19475534e-02
-2.38582000e-01 -6.11189485e-01 -1.25214279e+00 3.39899838e-01
1.19920945e+00 -7.48743117e-01 -1.40642032e-01 3.20836872e-01
5.57956696e-01 -5.67771077e-01 -1.28066814e+00 3.54548812e-01
6.75314546e-01 -1.09328520e+00 9.18826163e-01 -1.79433435e-01
2.41897389e-01 -2.65197963e-01 -1.57692358e-02 -1.17250407e+00
-2.79401243e-01 -8.04684162e-01 -2.15558261e-01 1.16500020e+00
-1.99891642e-01 -7.62650669e-01 2.55926698e-01 5.94437599e-01
-1.35848835e-01 -7.44612217e-01 -9.26045716e-01 -4.83586401e-01
-6.28606975e-01 -2.04833150e-01 4.17549103e-01 9.95014727e-01
-2.51597855e-02 3.83551985e-01 -9.26166892e-01 8.23080912e-02
5.81710935e-01 3.02553624e-01 5.73821425e-01 -1.03362465e+00
-2.33055130e-01 -5.78170456e-02 -3.94716203e-01 -4.70774055e-01
1.64942220e-01 -4.10507202e-01 1.88994825e-01 -1.26161087e+00
2.62431592e-01 -1.48805855e-02 -7.53405333e-01 6.66879058e-01
-1.61109835e-01 4.60474968e-01 6.15630448e-02 2.13519648e-01
-4.30614889e-01 8.84186208e-01 1.26977444e+00 2.63876677e-01
-5.47908247e-01 -1.93137348e-01 -5.92842042e-01 2.96228081e-01
1.00720501e+00 -6.13616705e-02 -4.95334923e-01 6.24351948e-02
-9.20445994e-02 7.07471013e-01 4.80164915e-01 -1.20856416e+00
-1.67160913e-01 -2.62082845e-01 9.03295636e-01 -2.39388525e-01
5.64059734e-01 -6.46543741e-01 4.66812789e-01 3.18000287e-01
-4.16219652e-01 9.07781050e-02 7.18975514e-02 5.18556297e-01
-2.23932877e-01 6.04842126e-01 1.02298093e+00 -2.67102450e-01
-3.97285879e-01 8.83820713e-01 1.66651025e-01 3.98040898e-02
9.15323198e-01 -9.34625119e-02 -6.31640732e-01 -4.75912601e-01
-7.35252678e-01 6.07614145e-02 -2.45575551e-02 5.23896158e-01
8.74189198e-01 -1.41630328e+00 -7.60839403e-01 2.28232235e-01
1.94611698e-01 -1.95064023e-01 1.09234059e+00 1.31301284e+00
-1.31010279e-01 2.39152730e-01 -4.81247872e-01 -4.29806292e-01
-1.32105780e+00 6.23877466e-01 8.22720587e-01 3.04953188e-01
-1.00715530e+00 5.22498250e-01 1.84472218e-01 -1.64908860e-02
2.87818730e-01 6.48346469e-02 -6.96481764e-01 1.95861310e-01
8.36004198e-01 4.85233486e-01 2.22482961e-02 -7.06387937e-01
-5.17725587e-01 8.05107236e-01 4.46463048e-01 1.89469054e-01
1.24351609e+00 -5.70207179e-01 -6.04858659e-02 8.13888788e-01
1.51427865e+00 -2.31430143e-01 -1.14642620e+00 -1.66406721e-01
-7.58631408e-01 -4.64679450e-01 1.89833805e-01 -6.12615883e-01
-1.74996722e+00 9.02571321e-01 1.22398198e+00 -4.63572443e-01
1.70371783e+00 -2.83168107e-01 8.31270456e-01 1.12098165e-01
3.13180774e-01 -1.05829298e+00 1.21702440e-01 -4.81961071e-02
1.20820117e+00 -1.27543867e+00 1.25649467e-01 -1.93523377e-01
-8.23601842e-01 1.22803020e+00 8.79548430e-01 2.92885989e-01
5.54916680e-01 -3.92991394e-01 2.79439747e-01 -1.61229327e-01
-7.19300807e-01 -5.38757890e-02 1.08273208e-01 4.45880830e-01
6.60642385e-01 -2.61201449e-02 -1.30720362e-01 3.73917907e-01
-6.69299439e-02 3.70201617e-01 4.45932359e-01 4.85174239e-01
4.23441418e-02 -1.31996542e-01 -2.59076148e-01 3.01288366e-01
-4.79028046e-01 4.09680605e-02 7.47731850e-02 6.18697882e-01
1.15087003e-01 1.28199768e+00 -1.51587933e-01 -4.90538746e-01
3.37598443e-01 -6.92874240e-03 4.09992695e-01 -2.84110874e-01
-4.93514359e-01 1.54090002e-01 -7.69210830e-02 -1.03990006e+00
-5.40550411e-01 -5.36299288e-01 -9.97196615e-01 -4.14680004e-01
3.01199732e-04 -1.62615940e-01 4.67903465e-01 6.16217017e-01
4.00108635e-01 6.91365182e-01 1.19096863e+00 -9.03846443e-01
-2.44574636e-01 -1.08682954e+00 -9.16245639e-01 4.01424915e-01
4.76918370e-01 -5.10429382e-01 -6.66330874e-01 -1.49649262e-01] | [13.887093544006348, 2.7075819969177246] |
b140d949-e637-4937-b4a6-18f8537c57bd | space-time-correspondence-as-a-contrastive | 2006.14613 | null | https://arxiv.org/abs/2006.14613v2 | https://arxiv.org/pdf/2006.14613v2.pdf | Space-Time Correspondence as a Contrastive Random Walk | This paper proposes a simple self-supervised approach for learning a representation for visual correspondence from raw video. We cast correspondence as prediction of links in a space-time graph constructed from video. In this graph, the nodes are patches sampled from each frame, and nodes adjacent in time can share a directed edge. We learn a representation in which pairwise similarity defines transition probability of a random walk, so that long-range correspondence is computed as a walk along the graph. We optimize the representation to place high probability along paths of similarity. Targets for learning are formed without supervision, by cycle-consistency: the objective is to maximize the likelihood of returning to the initial node when walking along a graph constructed from a palindrome of frames. Thus, a single path-level constraint implicitly supervises chains of intermediate comparisons. When used as a similarity metric without adaptation, the learned representation outperforms the self-supervised state-of-the-art on label propagation tasks involving objects, semantic parts, and pose. Moreover, we demonstrate that a technique we call edge dropout, as well as self-supervised adaptation at test-time, further improve transfer for object-centric correspondence. | ['Alexei A. Efros', 'Allan Jabri', 'Andrew Owens'] | 2020-06-25 | null | http://proceedings.neurips.cc/paper/2020/hash/e2ef524fbf3d9fe611d5a8e90fefdc9c-Abstract.html | http://proceedings.neurips.cc/paper/2020/file/e2ef524fbf3d9fe611d5a8e90fefdc9c-Paper.pdf | neurips-2020-12 | ['dense-pixel-correspondence-estimation'] | ['computer-vision'] | [ 3.61796081e-01 3.88125181e-01 -5.31480193e-01 -5.73894799e-01
-6.73443913e-01 -5.09056628e-01 6.40435219e-01 2.70454496e-01
-3.64475280e-01 3.76492113e-01 3.46557647e-01 1.94814369e-01
3.80313359e-02 -8.73808920e-01 -1.40996242e+00 -4.62299138e-01
-5.03143847e-01 7.57765591e-01 5.38425803e-01 1.18928313e-01
4.13250923e-02 4.36974704e-01 -1.28411007e+00 3.66223782e-01
4.58014518e-01 7.69507766e-01 2.60552764e-01 5.86368203e-01
1.10568821e-01 8.07872713e-01 -1.22542679e-01 -4.47133392e-01
2.69217998e-01 -8.18946719e-01 -1.18569767e+00 4.78398681e-01
9.71193910e-01 -2.05760285e-01 -7.50983536e-01 1.02650571e+00
3.34598422e-02 3.24169189e-01 7.56785989e-01 -1.41476405e+00
-6.47478402e-01 4.74530280e-01 -5.21267712e-01 1.00957081e-01
4.54991341e-01 1.47828981e-01 1.53860915e+00 -7.55543768e-01
1.15596688e+00 1.27201831e+00 6.17121696e-01 5.26891768e-01
-1.52048981e+00 -1.50273964e-01 4.14283454e-01 4.25653964e-01
-1.24324489e+00 -2.12963507e-01 9.91320074e-01 -5.14821708e-01
8.92255664e-01 -3.96686606e-02 1.01303613e+00 1.00579131e+00
2.46635765e-01 8.89133275e-01 5.92456937e-01 -2.34348297e-01
3.20268154e-01 -3.07683527e-01 -1.42647130e-02 1.30923247e+00
-1.24826595e-01 1.60493866e-01 -8.97349477e-01 -4.89598922e-02
8.88859391e-01 4.31489535e-02 -1.79166242e-01 -1.00042951e+00
-1.29547775e+00 8.17743242e-01 1.10053587e+00 -5.70298694e-02
-5.03439568e-02 5.75499833e-01 1.83909297e-01 4.05474037e-01
5.50852478e-01 1.95536852e-01 -2.03136548e-01 2.30785698e-01
-9.25914943e-01 7.13906437e-02 7.34253407e-01 1.23084331e+00
1.15865135e+00 -3.77731323e-01 -2.23009333e-01 3.78997803e-01
3.09381932e-01 2.24616617e-01 7.40689710e-02 -1.25591242e+00
3.16261858e-01 5.15571713e-01 -2.39277110e-01 -9.63901818e-01
-2.35990927e-01 -5.14616549e-01 -5.87752879e-01 2.11804643e-01
4.49862748e-01 3.67525399e-01 -1.06997454e+00 1.97060621e+00
2.00723186e-01 5.68748593e-01 -3.14583063e-01 9.70495582e-01
6.40151322e-01 6.92215204e-01 -3.65668721e-02 -7.15259984e-02
9.84295964e-01 -1.32235050e+00 -2.13753715e-01 -3.19010615e-01
5.57169557e-01 -5.26430190e-01 1.00116754e+00 -7.99210146e-02
-1.30370128e+00 -6.27803385e-01 -8.52535903e-01 -1.35911614e-01
2.28108577e-02 -3.39980096e-01 3.66553247e-01 -2.01359808e-01
-1.30183220e+00 1.07841301e+00 -1.05806327e+00 -5.46824217e-01
5.38627446e-01 8.55065212e-02 -5.15548646e-01 -2.42819101e-01
-8.88115764e-01 6.43409550e-01 1.76633559e-02 -1.06046580e-01
-1.18878281e+00 -6.38948917e-01 -9.53531504e-01 -1.01607248e-01
2.27213740e-01 -9.45411980e-01 8.47505331e-01 -1.17650020e+00
-1.12909734e+00 1.48084545e+00 -2.62810707e-01 -5.04562974e-01
5.96252680e-01 -1.62881166e-01 -2.51200050e-02 5.08038223e-01
4.44415092e-01 1.33978593e+00 1.03300500e+00 -1.36706102e+00
-5.71364641e-01 -2.85406053e-01 1.62366852e-01 4.06429321e-01
-1.87591687e-02 -1.76937282e-01 -8.92977715e-01 -5.64129651e-01
3.11392248e-01 -1.12745762e+00 -1.88270688e-01 7.40793169e-01
-4.74258810e-01 -3.35028768e-01 6.33136570e-01 -6.27296448e-01
7.85625100e-01 -2.15599704e+00 5.81937551e-01 2.40597069e-01
3.48134547e-01 -4.72922951e-01 -4.36188698e-01 3.06908906e-01
1.56475846e-02 1.93843478e-03 -5.67858458e-01 -4.78319228e-01
-2.29797989e-01 3.29855114e-01 2.43804324e-02 7.49048531e-01
4.86813188e-01 1.05198514e+00 -1.33536196e+00 -5.48314095e-01
4.78196256e-02 3.13170731e-01 -8.36281240e-01 5.29095471e-01
-5.75179935e-01 5.76510429e-01 -1.90399051e-01 3.15503478e-01
2.96224475e-01 -7.10902154e-01 -1.15278140e-01 -4.67247546e-01
1.23285383e-01 3.27822119e-01 -8.51608515e-01 2.29212618e+00
-2.06812382e-01 6.10710859e-01 -3.19922239e-01 -1.14258206e+00
8.21845353e-01 -6.72328621e-02 4.94645149e-01 -5.07977247e-01
-1.95729882e-01 -1.45508468e-01 -1.85712963e-01 -4.00069922e-01
1.36258408e-01 3.87081236e-01 1.41446590e-01 4.98360723e-01
4.04312193e-01 -5.85175268e-02 3.02259922e-01 5.73988318e-01
1.22747254e+00 6.23204112e-01 -1.48912504e-01 -1.38374969e-01
1.63211301e-01 9.85761806e-02 5.06281257e-01 7.03752875e-01
-1.36954933e-01 1.01909566e+00 6.14693284e-01 -3.95245880e-01
-1.15598083e+00 -1.47790396e+00 4.83139493e-02 1.17385650e+00
4.74720567e-01 -4.05421853e-01 -6.56312346e-01 -9.53509748e-01
2.93067587e-03 2.84536630e-01 -7.38533139e-01 -4.05407965e-01
-7.91820288e-01 -4.76561487e-02 -3.19821723e-02 6.26572549e-01
4.38560516e-01 -1.01074648e+00 -3.58626992e-01 1.88559443e-01
-3.18070024e-01 -1.07478833e+00 -1.00002205e+00 2.86777437e-01
-1.06133592e+00 -1.28260481e+00 -7.60048568e-01 -1.32380009e+00
1.14916945e+00 2.47558311e-01 1.50888479e+00 1.92466542e-01
-3.18604559e-01 6.04356647e-01 -2.32926324e-01 6.31670237e-01
-3.26573908e-01 -6.06451519e-02 -3.55397463e-01 1.82546198e-01
-2.85206325e-02 -5.89737475e-01 -8.01057935e-01 3.14562470e-01
-4.76825565e-01 1.76609755e-01 3.46426994e-01 9.97313917e-01
9.76187170e-01 -5.06837428e-01 4.17802893e-02 -8.54496479e-01
-4.01282124e-02 -4.50322807e-01 -3.91479492e-01 4.43160415e-01
-3.83655131e-01 2.38253102e-01 4.09308463e-01 -3.32459092e-01
-5.78546643e-01 3.28650445e-01 1.46439075e-01 -7.61211038e-01
-6.37118518e-02 3.53829265e-01 9.86972079e-02 -6.87047690e-02
6.64083302e-01 1.63766935e-01 1.71464086e-01 -3.06054920e-01
7.48608828e-01 -8.88978168e-02 6.70083225e-01 -6.17676437e-01
7.88953006e-01 5.90532064e-01 2.45215610e-01 -4.60552573e-01
-9.40750360e-01 -5.30299187e-01 -9.83528614e-01 -4.10454243e-01
1.03960717e+00 -1.06649029e+00 -4.54302758e-01 2.23523021e-01
-1.23385441e+00 -6.15322948e-01 -6.04926527e-01 5.09498775e-01
-8.42986405e-01 2.87831098e-01 -7.31156528e-01 -2.21025124e-01
1.06532559e-01 -7.98755705e-01 1.05868864e+00 4.03486714e-02
-4.41891849e-01 -1.27357590e+00 1.50437295e-01 1.23604521e-01
-4.92766984e-02 2.04716071e-01 9.02755439e-01 -3.71782482e-01
-1.05953598e+00 9.60616744e-04 -1.70735553e-01 2.27979958e-01
8.54965076e-02 3.93690355e-02 -6.47048235e-01 -5.70316613e-01
-5.22416353e-01 -5.74019313e-01 1.06101668e+00 4.36176866e-01
1.09242153e+00 -3.56516182e-01 -4.87423986e-01 8.14075053e-01
1.23386931e+00 -2.95417666e-01 3.63585979e-01 1.56188563e-01
1.06742465e+00 7.53118932e-01 4.27144438e-01 1.47669762e-01
4.31639999e-01 6.82438374e-01 6.61722839e-01 -2.62084872e-01
-5.00152290e-01 -6.52606547e-01 3.29631209e-01 6.10190392e-01
-2.68558171e-02 -2.11936772e-01 -6.15194738e-01 5.61365426e-01
-1.95232844e+00 -9.52192783e-01 9.57253948e-03 2.43791461e+00
7.80927062e-01 3.54110420e-01 2.90895641e-01 -2.57151186e-01
9.28476214e-01 4.35776681e-01 -6.38144016e-01 1.97845593e-01
-1.00296833e-01 7.42100598e-03 4.08378065e-01 8.80518079e-01
-1.14706981e+00 1.06138456e+00 5.98551750e+00 3.21922451e-01
-8.62381220e-01 6.22149669e-02 8.14982057e-01 1.46190152e-01
-4.62732375e-01 1.90556943e-01 -5.18992662e-01 4.10836846e-01
3.64114851e-01 1.33990139e-01 4.38047677e-01 6.23878300e-01
-3.30911539e-02 8.32784176e-02 -1.82705808e+00 7.46070623e-01
3.62306267e-01 -1.47997367e+00 1.17738023e-01 -2.45806396e-01
8.97764444e-01 1.76947922e-01 1.04256719e-02 -1.52682573e-01
3.81204188e-01 -6.48846984e-01 9.31463242e-01 5.15873611e-01
7.54880130e-01 -4.89036560e-01 2.09899887e-01 -3.63107175e-02
-1.41939199e+00 3.64405751e-01 -2.83622086e-01 1.13096058e-01
4.14932042e-01 3.93185794e-01 -5.13866365e-01 2.72963345e-01
6.35985136e-01 1.31674767e+00 -5.59432983e-01 1.18669796e+00
-5.00202596e-01 5.71824253e-01 -1.97399139e-01 2.44328260e-01
2.93795913e-01 -4.45806265e-01 5.19073904e-01 1.17362392e+00
4.35408875e-02 -1.48863643e-01 6.61835909e-01 9.38665509e-01
-3.56813014e-01 -2.42495611e-01 -6.89185798e-01 2.65747815e-01
5.10332525e-01 1.01551974e+00 -9.56981301e-01 -3.32574248e-01
-5.34150660e-01 1.54081702e+00 8.00335884e-01 5.25166154e-01
-7.65156686e-01 1.12971425e-01 5.78083336e-01 3.66081685e-01
2.53881186e-01 -2.14358136e-01 -1.01548024e-02 -8.81695747e-01
1.20605320e-01 -3.02349329e-01 5.46593010e-01 -8.61270726e-01
-1.49966145e+00 5.56705296e-01 -2.66627446e-02 -1.48868263e+00
-1.68029100e-01 -2.03388959e-01 -7.75984943e-01 4.68394548e-01
-1.42434955e+00 -1.19502354e+00 -4.01948661e-01 7.31413245e-01
5.76577604e-01 1.91090610e-02 5.66820323e-01 6.69781342e-02
-2.83196807e-01 5.76008976e-01 -1.75519809e-01 3.54827464e-01
7.66135871e-01 -1.22975898e+00 6.90409601e-01 8.38356853e-01
7.10329115e-01 2.86497265e-01 3.86488140e-01 -7.78063118e-01
-1.08045912e+00 -1.50051785e+00 1.02148843e+00 -2.72252470e-01
7.12028027e-01 -4.60158139e-01 -9.93351579e-01 9.48358834e-01
4.78164144e-02 7.33967602e-01 2.39328712e-01 4.08482030e-02
-6.38877869e-01 -6.92179287e-03 -8.34450841e-01 5.40824473e-01
1.76406980e+00 -8.92079532e-01 -4.73250091e-01 7.20047235e-01
8.09456706e-01 -3.67504030e-01 -6.58810258e-01 1.32680997e-01
2.54927784e-01 -9.01434362e-01 1.02902699e+00 -8.81858647e-01
7.17558265e-01 -2.18998417e-01 3.31141725e-02 -1.39104009e+00
-6.66720271e-01 -8.73567045e-01 -2.03341633e-01 9.79457736e-01
4.20259058e-01 -1.62637696e-01 1.18276095e+00 1.71660289e-01
-3.97304118e-01 -8.53449523e-01 -8.63374472e-01 -9.35810506e-01
-1.26838014e-01 1.00842558e-01 -3.92233171e-02 8.79108191e-01
-1.51445761e-01 5.63677788e-01 -3.26038599e-01 2.25620627e-01
1.05431235e+00 6.71889037e-02 5.00419617e-01 -1.11447668e+00
-4.30543512e-01 -3.21584672e-01 -7.97714829e-01 -1.57189810e+00
4.06593055e-01 -1.23706830e+00 2.54633725e-01 -1.60850644e+00
4.11897004e-01 -5.14616549e-01 -1.26496941e-01 3.37314904e-01
-1.76369622e-01 3.74306828e-01 2.12081015e-01 4.63503659e-01
-9.35122371e-01 5.79117775e-01 1.43335927e+00 -3.95820349e-01
6.22833706e-03 -1.48151562e-01 -2.89842784e-02 6.48009956e-01
4.20203447e-01 -7.52867401e-01 -4.96772528e-01 -6.53451800e-01
6.57722056e-02 2.39763960e-01 5.76820612e-01 -9.27336872e-01
4.67809379e-01 4.93254215e-02 2.33369321e-01 -4.29193228e-01
3.42036247e-01 -7.14325488e-01 1.37857199e-01 5.45832932e-01
-7.97765017e-01 9.52947140e-02 -5.70947886e-01 9.42017019e-01
-2.34569371e-01 -6.36276901e-02 7.79611707e-01 -1.11021534e-01
-8.30925286e-01 8.25319409e-01 4.60112318e-02 5.60343564e-01
1.05363226e+00 -4.06795740e-01 -1.47916391e-01 -4.33630735e-01
-1.08772039e+00 3.19735527e-01 7.15823710e-01 3.46101135e-01
7.53738642e-01 -1.57059646e+00 -8.09179544e-01 1.86460689e-01
3.50535184e-01 1.91539958e-01 -7.19243139e-02 5.86192727e-01
-5.12157857e-01 -1.76657617e-01 -3.08062434e-01 -1.10002768e+00
-1.11740232e+00 4.52451915e-01 3.44690830e-01 -5.34319459e-03
-1.00344861e+00 1.37575197e+00 3.37939471e-01 -2.04967827e-01
6.75568938e-01 -1.75544590e-01 4.91827428e-02 -5.81375174e-02
1.29211441e-01 2.13118762e-01 -1.00686081e-01 -6.63803518e-01
-3.31504166e-01 8.99920583e-01 -7.25374445e-02 -1.88600674e-01
1.08682239e+00 -1.32163778e-01 -1.44493327e-01 6.91973448e-01
1.70875907e+00 -4.13988233e-01 -1.98398137e+00 -5.68531156e-01
-9.79053453e-02 -6.07448697e-01 -2.91465431e-01 -4.01696235e-01
-1.27798593e+00 6.94698632e-01 4.87255335e-01 -1.51090935e-01
5.52152872e-01 6.35270774e-01 5.75590372e-01 4.14360762e-01
3.72590035e-01 -8.06288600e-01 8.15496206e-01 5.07911742e-01
8.97257507e-01 -1.32175922e+00 -1.49863169e-01 -6.89886451e-01
-5.62391102e-01 9.87754166e-01 7.14441478e-01 -6.14590168e-01
7.49403238e-01 6.08212352e-02 -2.18850195e-01 -3.02925557e-01
-7.21801639e-01 -2.81806111e-01 4.13511813e-01 7.93478787e-01
5.05917192e-01 -1.52789712e-01 1.51983380e-01 -3.25570792e-01
4.69536632e-02 -3.43483567e-01 2.69498587e-01 8.64324987e-01
-4.90856260e-01 -8.44545603e-01 8.94382372e-02 4.65455592e-01
9.37001333e-02 -1.21173531e-01 -3.67490441e-01 4.80357230e-01
-1.09921984e-01 5.43964565e-01 7.89393723e-01 -2.91838914e-01
2.24769413e-01 -1.17156111e-01 8.44096243e-01 -8.53676558e-01
-2.38589555e-01 -1.96890961e-02 1.44220859e-01 -9.36195374e-01
-5.94620407e-01 -7.74487376e-01 -1.40095949e+00 -1.75240189e-01
-1.30343631e-01 1.12269804e-01 8.38078111e-02 8.04941177e-01
2.35713825e-01 4.64761406e-01 1.00708020e+00 -9.71574664e-01
-2.90448010e-01 -5.32243013e-01 -3.83085281e-01 9.45584238e-01
3.61660451e-01 -5.81572056e-01 -4.02433485e-01 5.53519726e-01] | [8.88036823272705, -0.3429740071296692] |
63c11fb3-4f65-414f-9626-accb2d1469c5 | data-interpretation-over-plots | 1909.00997 | null | https://arxiv.org/abs/1909.00997v3 | https://arxiv.org/pdf/1909.00997v3.pdf | PlotQA: Reasoning over Scientific Plots | Existing synthetic datasets (FigureQA, DVQA) for reasoning over plots do not contain variability in data labels, real-valued data, or complex reasoning questions. Consequently, proposed models for these datasets do not fully address the challenge of reasoning over plots. In particular, they assume that the answer comes either from a small fixed size vocabulary or from a bounding box within the image. However, in practice, this is an unrealistic assumption because many questions require reasoning and thus have real-valued answers which appear neither in a small fixed size vocabulary nor in the image. In this work, we aim to bridge this gap between existing datasets and real-world plots. Specifically, we propose PlotQA with 28.9 million question-answer pairs over 224,377 plots on data from real-world sources and questions based on crowd-sourced question templates. Further, 80.76% of the out-of-vocabulary (OOV) questions in PlotQA have answers that are not in a fixed vocabulary. Analysis of existing models on PlotQA reveals that they cannot deal with OOV questions: their overall accuracy on our dataset is in single digits. This is not surprising given that these models were not designed for such questions. As a step towards a more holistic model which can address fixed vocabulary as well as OOV questions, we propose a hybrid approach: Specific questions are answered by choosing the answer from a fixed vocabulary or by extracting it from a predicted bounding box in the plot, while other questions are answered with a table question-answering engine which is fed with a structured table generated by detecting visual elements from the image. On the existing DVQA dataset, our model has an accuracy of 58%, significantly improving on the highest reported accuracy of 46%. On PlotQA, our model has an accuracy of 22.52%, which is significantly better than state of the art models. | ['Mitesh M. Khapra', 'Nitesh Methani', 'Pratyush Kumar', 'Pritha Ganguly'] | 2019-09-03 | null | null | null | null | ['chart-question-answering', 'chart-question-answering'] | ['computer-code', 'computer-vision'] | [ 6.16002269e-02 4.59841222e-01 1.50197059e-01 -4.02731717e-01
-1.07004952e+00 -1.23845792e+00 4.58644450e-01 1.45241827e-01
1.38641953e-01 5.11230767e-01 1.34891585e-01 -6.08097970e-01
-1.06874302e-01 -1.25661552e+00 -8.89998555e-01 -1.89693049e-02
6.01064861e-01 6.01408184e-01 5.63042521e-01 -4.10137594e-01
-1.71751510e-02 2.83744216e-01 -1.67608249e+00 7.92350590e-01
9.56301510e-01 1.12944090e+00 -1.30896747e-01 8.95170093e-01
-7.45885968e-01 1.62234008e+00 -1.08610547e+00 -1.10958827e+00
3.29962820e-01 -4.87282395e-01 -1.02485538e+00 2.98278868e-01
1.13066566e+00 -4.10993457e-01 -1.44760847e-01 8.99202228e-01
1.99518383e-01 -3.09719503e-01 7.55496085e-01 -1.61094630e+00
-9.68196452e-01 4.94956881e-01 -2.66856104e-01 -1.09637909e-01
6.44426942e-01 3.72834742e-01 1.51461017e+00 -9.16647673e-01
9.74512398e-01 1.21946168e+00 5.71513295e-01 2.97924042e-01
-1.14277232e+00 -4.64714617e-01 4.04128581e-02 1.30869925e-01
-1.33260071e+00 -5.89351840e-02 6.60456419e-01 -6.08212948e-01
6.73255324e-01 7.85823226e-01 6.74390733e-01 7.65211880e-01
-1.11026563e-01 7.57958591e-01 1.08384490e+00 -2.50330031e-01
2.92266101e-01 3.77585590e-01 -1.04141206e-01 6.46139443e-01
1.33237634e-02 -5.97659826e-01 -3.86430293e-01 -1.00525267e-01
6.64560437e-01 -3.48351181e-01 -2.19002351e-01 -6.62812769e-01
-1.23355854e+00 9.82747018e-01 6.60293162e-01 1.17740005e-01
-2.12296724e-01 1.05005972e-01 1.94500804e-01 3.99938613e-01
9.06725451e-02 5.87218583e-01 -4.33558762e-01 9.88550186e-02
-8.54263484e-01 7.04578340e-01 1.02093136e+00 1.12439048e+00
8.94824505e-01 -2.91395277e-01 -5.61184168e-01 4.66345876e-01
2.56185830e-01 1.04776418e+00 -3.01336974e-01 -1.25191689e+00
8.24591100e-01 1.35074615e+00 3.49121571e-01 -1.42670882e+00
-8.67465287e-02 -1.94026798e-01 -4.41898704e-01 3.31393450e-01
1.06648517e+00 1.63599804e-01 -6.79998994e-01 1.44298410e+00
5.48370361e-01 -5.94350934e-01 5.63673861e-02 9.52129304e-01
1.41035235e+00 6.26770675e-01 -9.11274776e-02 4.55192685e-01
1.85570705e+00 -1.03043067e+00 -8.28357875e-01 -2.98897356e-01
5.13921797e-01 -7.50868797e-01 1.88262165e+00 4.12419021e-01
-8.69078457e-01 -4.50395942e-01 -9.05437708e-01 -4.20437604e-01
-8.30127537e-01 3.87097299e-01 2.65860021e-01 8.82302821e-01
-1.05238044e+00 -3.03943485e-01 1.47789538e-01 -4.72174853e-01
4.22488034e-01 -9.13753882e-02 -1.20206334e-01 -2.56097585e-01
-1.15959632e+00 6.90482199e-01 -2.32167289e-01 -1.73273250e-01
-7.91546822e-01 -8.38463426e-01 -8.18439186e-01 1.38226733e-01
9.40108418e-01 -4.90826666e-01 1.29673004e+00 -8.50321352e-01
-8.21324468e-01 1.14139295e+00 -1.16220228e-01 -4.75737095e-01
7.44030476e-01 -4.97147664e-02 -3.73903692e-01 4.80911285e-01
4.14909840e-01 7.67181695e-01 7.86632478e-01 -1.31598043e+00
-3.28176290e-01 -1.74641266e-01 1.00538778e+00 -2.15022609e-01
-3.81817408e-02 -1.63674042e-01 -3.06728810e-01 -4.55938548e-01
-2.08884820e-01 -7.33095229e-01 1.13690339e-01 7.66156435e-01
-3.20780903e-01 -1.57739043e-01 8.52151573e-01 -7.38033056e-01
1.29420745e+00 -1.92943001e+00 -4.03186768e-01 4.25957628e-02
5.23524940e-01 1.48183435e-01 -1.72118455e-01 6.01946473e-01
1.57373101e-01 3.68708581e-01 -3.00423533e-01 3.18365067e-01
3.37592661e-01 2.26154119e-01 -8.42383683e-01 -2.16392502e-02
5.74026406e-01 1.21874869e+00 -8.81629288e-01 -8.06317449e-01
2.50165403e-01 4.72686738e-02 -4.08715039e-01 2.34373823e-01
-9.11510527e-01 2.75207665e-02 -3.17075253e-01 8.66698265e-01
9.22596157e-01 -5.58149338e-01 8.81469175e-02 -1.67544633e-01
1.21760309e-01 -5.92413079e-03 -1.30763769e+00 1.32684004e+00
-3.36923957e-01 7.95615017e-01 -2.14055955e-01 -5.31325281e-01
1.32728922e+00 -6.28275424e-03 1.86979502e-01 -8.59469771e-01
-1.81573927e-02 7.41265342e-02 -3.03392947e-01 -7.98413217e-01
4.81489867e-01 2.02542514e-01 -2.70296693e-01 3.89660001e-01
-1.86300233e-01 -7.60024071e-01 4.83641773e-01 4.48786795e-01
1.45405483e+00 7.05704615e-02 2.67121077e-01 6.23184955e-03
8.53841484e-01 7.43465245e-01 8.98969918e-02 1.00885201e+00
-3.27253670e-01 9.32139516e-01 1.08312929e+00 -5.61766088e-01
-1.15810013e+00 -1.30156362e+00 1.74044341e-01 7.85771251e-01
8.01575854e-02 -7.31880665e-01 -8.72855425e-01 -9.63628888e-01
-2.83045061e-02 7.20851839e-01 -8.72298479e-01 3.47111404e-01
-2.85098016e-01 1.77723318e-01 6.58241451e-01 3.96431059e-01
6.93466246e-01 -1.15311778e+00 -8.92905116e-01 -8.54523107e-02
-5.26403546e-01 -1.41017735e+00 -1.01772673e-01 -2.08654583e-01
-2.73604482e-01 -1.54069650e+00 -5.67935050e-01 -4.34773296e-01
4.61051404e-01 3.54442686e-01 1.76182675e+00 1.09534122e-01
-7.26298392e-02 7.00824499e-01 -4.47965831e-01 -5.55243075e-01
-4.03787971e-01 9.85286459e-02 -9.92695153e-01 -8.69131461e-02
4.80542272e-01 6.18142262e-02 -5.05607545e-01 4.59341407e-01
-1.27437055e+00 -5.27152345e-02 3.72954279e-01 4.38138604e-01
3.79434437e-01 -2.02368274e-01 3.55403215e-01 -9.03076470e-01
4.45720196e-01 -4.29341108e-01 -6.44097269e-01 7.52215743e-01
-1.19273990e-01 8.87176320e-02 7.13648856e-01 -4.30112541e-01
-9.47299957e-01 4.81772050e-03 1.02248214e-01 -3.01941037e-01
-7.63365999e-02 2.71942884e-01 -3.38992864e-01 3.10981899e-01
9.60321486e-01 -2.86272671e-02 -1.74244329e-01 7.11547583e-02
6.60365820e-01 4.78376865e-01 3.07975560e-01 -4.91727293e-01
1.10698569e+00 6.68759465e-01 -1.62394419e-01 -5.51180184e-01
-1.09337449e+00 -3.09299588e-01 -4.46844310e-01 -6.27863944e-01
1.04044259e+00 -8.19950759e-01 -8.72151375e-01 1.45182610e-01
-1.24374342e+00 -4.01357770e-01 -6.17259562e-01 -1.60237893e-01
-5.40980399e-01 2.52243709e-02 -8.62729773e-02 -9.47810411e-01
-1.11471996e-01 -1.04032981e+00 1.40446687e+00 9.73463356e-02
-3.86605293e-01 -6.96152925e-01 -1.72870591e-01 8.05417657e-01
4.34166729e-01 6.13253415e-01 1.10157132e+00 -3.65463406e-01
-9.02900636e-01 -1.08382963e-01 -6.74856007e-01 1.62429623e-02
2.83469027e-03 2.10328728e-01 -9.52072978e-01 4.57214683e-01
-1.31290495e-01 -7.84004867e-01 2.65653372e-01 -1.33176833e-01
8.93032372e-01 -1.31087422e-01 2.24691346e-01 -3.58098149e-02
1.52301121e+00 -9.67076719e-02 7.35985935e-01 4.51428771e-01
5.36203325e-01 9.69343007e-01 9.79442179e-01 4.89677012e-01
8.40713441e-01 3.70103747e-01 8.54185283e-01 -7.59421289e-02
-3.83092105e-01 -4.86438274e-01 1.21983126e-01 2.30565056e-01
5.43667138e-01 -5.17069995e-01 -1.30083358e+00 8.58606398e-01
-1.74229074e+00 -8.58385682e-01 -9.58041668e-01 1.66379368e+00
6.52091682e-01 4.05231975e-02 9.00779888e-02 2.30839014e-01
3.95968169e-01 4.74825829e-01 -6.42863333e-01 -4.64509279e-01
-4.93430793e-01 2.67670542e-01 2.91425556e-01 2.23169699e-01
-9.48727787e-01 7.63642967e-01 6.01652384e+00 7.49306381e-01
-7.92174637e-01 -8.33075866e-02 5.00579000e-01 1.70178622e-01
-7.88287520e-01 2.46832073e-01 -4.30120438e-01 -8.52408409e-02
7.30706632e-01 2.67632492e-02 2.40856215e-01 8.61366987e-01
-1.78705119e-02 -3.06462228e-01 -8.84533823e-01 9.82077062e-01
2.96658605e-01 -1.33739805e+00 2.56072998e-01 -3.94446589e-02
5.60343742e-01 -7.16697395e-01 4.49549183e-02 5.14899790e-01
4.79992896e-01 -1.23736429e+00 1.22699726e+00 6.79178953e-01
7.97023356e-01 -4.34994876e-01 6.38662696e-01 4.34143722e-01
-1.37296534e+00 5.46374209e-02 -3.57648402e-01 -1.52077124e-01
-1.71015188e-01 4.19213325e-01 -9.22477543e-01 4.25996423e-01
9.14801180e-01 2.94091493e-01 -1.24997616e+00 5.82870066e-01
-4.70786721e-01 6.00028932e-01 -2.17876717e-01 -3.55036318e-01
3.74444276e-01 4.51788045e-02 8.20549652e-02 7.56352365e-01
2.14965552e-01 -5.36631532e-02 -1.59410968e-01 1.12980413e+00
-1.88344181e-01 3.33230734e-01 -8.30273569e-01 2.16564909e-02
2.55214602e-01 1.24593484e+00 -7.46486902e-01 -3.94580603e-01
-6.62884653e-01 6.40964508e-01 4.06088442e-01 2.76210248e-01
-1.34675300e+00 -4.59879279e-01 3.28967452e-01 4.59775925e-01
5.75105071e-01 6.08752221e-02 -4.64014709e-01 -1.06707954e+00
5.35942495e-01 -1.30862415e+00 3.99015754e-01 -1.47301137e+00
-1.16043425e+00 6.57416463e-01 -1.28812060e-01 -1.36869061e+00
-1.68453097e-01 -6.85842931e-01 -1.01387553e-01 5.84221601e-01
-1.32105851e+00 -1.48614001e+00 -9.48942363e-01 5.75435400e-01
5.05769432e-01 2.79055864e-01 6.83739841e-01 1.00940622e-01
1.48739293e-01 2.91905910e-01 -3.80454093e-01 4.18512940e-01
8.45206559e-01 -1.48451912e+00 5.01463890e-01 6.15123332e-01
4.19368654e-01 1.99125618e-01 6.61038995e-01 -4.54295993e-01
-1.36524832e+00 -1.14443886e+00 9.98620212e-01 -1.18174922e+00
9.32259917e-01 -8.32027733e-01 -1.04559207e+00 4.55995232e-01
3.45301986e-01 1.71620950e-01 5.39669156e-01 -4.74358648e-01
-7.41621435e-01 -1.40217826e-01 -1.33784556e+00 7.06543803e-01
1.06824636e+00 -7.27227330e-01 -5.95708251e-01 3.94708335e-01
9.08068120e-01 -4.40939307e-01 -8.29140186e-01 1.63197860e-01
5.57746053e-01 -1.19097602e+00 8.59306395e-01 -5.37155211e-01
8.87131214e-01 -6.76282048e-01 -5.48422575e-01 -6.51362717e-01
1.26611456e-01 -1.21736161e-01 -8.13604072e-02 1.52096319e+00
5.19989192e-01 -2.25008622e-01 9.58264291e-01 8.08131456e-01
5.32929838e-01 -7.39541590e-01 -5.98919988e-01 -5.05363226e-01
2.77362280e-02 -6.87187672e-01 1.00878716e+00 8.15049350e-01
-6.17391944e-01 3.06698591e-01 -1.97507203e-01 9.09785628e-02
2.30131671e-01 3.47426891e-01 1.43212748e+00 -1.02432644e+00
3.22474651e-02 9.29153990e-04 -3.22099358e-01 -1.00262761e+00
1.72799197e-03 -4.90651757e-01 -2.06806794e-01 -2.10620952e+00
-3.42587419e-02 -2.51168340e-01 5.58461547e-01 3.64736766e-01
-4.70904075e-02 3.13624978e-01 4.18995231e-01 2.62299702e-02
-6.38642490e-01 3.65108341e-01 1.44552159e+00 -5.34073472e-01
2.73541361e-01 -3.14512104e-01 -6.81832671e-01 6.60970151e-01
6.06115937e-01 -4.01534557e-01 -5.91594875e-01 -5.32393754e-01
8.51438463e-01 1.96566284e-01 7.43047416e-01 -9.73707974e-01
9.35197175e-02 -2.01468736e-01 2.89557278e-01 -8.89328241e-01
2.90493101e-01 -1.13771701e+00 2.07872335e-02 2.58500695e-01
-2.97085285e-01 1.80721968e-01 -3.84777971e-02 4.41890240e-01
-2.90287793e-01 -3.16725701e-01 3.35173309e-01 -2.78247833e-01
-5.52873254e-01 -2.04580024e-01 -3.56982589e-01 5.10701239e-01
1.24445581e+00 -3.30251634e-01 -8.35947752e-01 -7.99637496e-01
-5.00085533e-01 3.29455078e-01 5.86150289e-01 5.11860847e-01
5.76450169e-01 -1.33581865e+00 -7.64801204e-01 -2.36505345e-01
6.29004240e-01 1.16691612e-01 2.69886434e-01 3.83060366e-01
-9.11498308e-01 3.38686466e-01 2.26386022e-02 -7.09412217e-01
-1.04091585e+00 7.47156441e-01 2.23336726e-01 -5.00945330e-01
-1.70412511e-01 4.07928675e-01 1.25872672e-01 -9.01384056e-01
-2.65897419e-02 -8.05377960e-01 -1.83871374e-01 2.79917300e-01
4.15697873e-01 1.09429687e-01 9.01418796e-04 -6.39602423e-01
-4.24735874e-01 7.21481621e-01 5.19772291e-01 -1.32576361e-01
9.19366717e-01 8.33982676e-02 2.54702512e-02 6.00750506e-01
8.38926256e-01 4.73487973e-01 -9.66291010e-01 -2.54533142e-01
-2.37866025e-02 -5.81011891e-01 -6.21983111e-01 -9.35112178e-01
-8.56023908e-01 9.40369308e-01 3.53658557e-01 6.95159435e-01
1.06986618e+00 3.61458480e-01 6.14422858e-01 4.74988580e-01
2.91034371e-01 -7.26086497e-01 4.68874007e-01 3.95543605e-01
1.31639981e+00 -1.40953362e+00 -1.38691291e-01 -7.00696766e-01
-9.24970746e-01 1.10488796e+00 8.84131253e-01 -1.36220127e-01
3.42300713e-01 2.30205253e-01 4.60508943e-01 -6.45474076e-01
-8.74207020e-01 -4.64798123e-01 3.02765191e-01 7.07650304e-01
2.50162333e-01 -1.01299137e-01 1.29071206e-01 4.60069060e-01
-5.47188103e-01 -1.17901474e-01 7.11218178e-01 8.40950429e-01
-3.82262737e-01 -7.45673478e-01 -7.90017545e-01 4.23385710e-01
-1.28681004e-01 7.67740011e-02 -1.06572962e+00 1.06742692e+00
2.25821152e-01 1.30366695e+00 4.78971563e-02 -1.12718530e-01
8.06300223e-01 -8.28083828e-02 2.30286449e-01 -4.44535345e-01
-9.68662858e-01 -5.96499562e-01 1.83015376e-01 -4.68159378e-01
-4.05089766e-01 -4.14238393e-01 -1.00784600e+00 -2.84431994e-01
-1.60573691e-01 -9.32300016e-02 3.53130668e-01 6.72788978e-01
2.17356309e-01 5.46785295e-01 2.19165474e-01 2.42661089e-01
-4.96421158e-01 -7.62461424e-01 -2.69155979e-01 5.54873228e-01
2.09834099e-01 -5.31679809e-01 -2.92317092e-01 2.51995713e-01] | [11.10448169708252, 1.9622236490249634] |
ae76cd7a-9cab-484a-88cf-76e627102018 | neural-architecture-transfer-2-a-paradigm-for | 2307.00960 | null | https://arxiv.org/abs/2307.00960v1 | https://arxiv.org/pdf/2307.00960v1.pdf | Neural Architecture Transfer 2: A Paradigm for Improving Efficiency in Multi-Objective Neural Architecture Search | Deep learning is increasingly impacting various aspects of contemporary society. Artificial neural networks have emerged as the dominant models for solving an expanding range of tasks. The introduction of Neural Architecture Search (NAS) techniques, which enable the automatic design of task-optimal networks, has led to remarkable advances. However, the NAS process is typically associated with long execution times and significant computational resource requirements. Once-For-All (OFA) and its successor, Once-For-All-2 (OFAv2), have been developed to mitigate these challenges. While maintaining exceptional performance and eliminating the need for retraining, they aim to build a single super-network model capable of directly extracting sub-networks satisfying different constraints. Neural Architecture Transfer (NAT) was developed to maximise the effectiveness of extracting sub-networks from a super-network. In this paper, we present NATv2, an extension of NAT that improves multi-objective search algorithms applied to dynamic super-network architectures. NATv2 achieves qualitative improvements in the extractable sub-networks by exploiting the improved super-networks generated by OFAv2 and incorporating new policies for initialisation, pre-processing and updating its networks archive. In addition, a post-processing pipeline based on fine-tuning is introduced. Experimental results show that NATv2 successfully improves NAT and is highly recommended for investigating high-performance architectures with a minimal number of parameters. | ['Matteo Matteucci', 'Eugenio Lomurno', 'Simone Sarti'] | 2023-07-03 | null | null | null | null | ['architecture-search'] | ['methodology'] | [ 3.12684417e-01 4.36722562e-02 -1.27173528e-01 -3.99961144e-01
-4.19375271e-01 -5.80168605e-01 5.11222184e-01 -1.23037338e-01
-8.43927622e-01 7.46826887e-01 -8.48094150e-02 -3.51745963e-01
-7.71410227e-01 -5.71944892e-01 -6.24799252e-01 -5.03730714e-01
-9.14444923e-02 7.71891952e-01 2.46754259e-01 -1.97864518e-01
2.79276699e-01 8.07456553e-01 -1.90069175e+00 3.28082144e-01
7.99395502e-01 1.23085427e+00 6.20197475e-01 4.98550743e-01
-2.73354828e-01 3.14837635e-01 -6.25553429e-01 -2.73497552e-01
5.00732124e-01 -2.25829184e-01 -9.91616964e-01 -5.00752866e-01
3.41430813e-01 7.62204006e-02 -1.12106070e-01 8.24538112e-01
6.94551587e-01 4.41559702e-01 3.52716774e-01 -1.20854545e+00
-2.61810660e-01 7.48662531e-01 5.14642186e-02 3.72442961e-01
-3.33512336e-01 1.98129743e-01 1.05870783e+00 -6.11669302e-01
5.67763090e-01 1.17520440e+00 6.06954753e-01 7.76381373e-01
-1.22597170e+00 -6.88986778e-01 1.10679209e-01 5.00297546e-01
-1.37510276e+00 -8.07281375e-01 7.56326139e-01 -2.06846111e-02
1.69400358e+00 4.62681234e-01 7.24988818e-01 1.04543960e+00
9.64907929e-03 7.94675767e-01 7.35962987e-01 -5.54471910e-01
5.22767723e-01 5.27599491e-02 -4.61252518e-02 6.55260205e-01
2.02528566e-01 1.62356600e-01 -6.45448267e-01 3.41370702e-02
5.92168808e-01 -4.36382443e-01 -1.65452380e-02 -2.71491945e-01
-8.74097943e-01 6.53028488e-01 5.86367905e-01 6.12556636e-01
-7.74029255e-01 1.81919605e-01 4.94409978e-01 4.77031291e-01
1.34551093e-01 1.17641461e+00 -7.30915427e-01 -2.18768105e-01
-1.15430295e+00 3.21422219e-01 7.13678479e-01 7.61689901e-01
6.99608982e-01 4.32709068e-01 -1.29763126e-01 1.10714626e+00
5.42798126e-03 -5.80954850e-02 9.60283101e-01 -1.19788611e+00
2.54553616e-01 1.11176705e+00 -3.30339551e-01 -8.66992593e-01
-6.25122488e-01 -9.78327870e-01 -9.37658846e-01 2.13721201e-01
5.62138148e-02 3.93837802e-02 -1.11422575e+00 1.77323747e+00
1.24353506e-01 -2.48372369e-02 7.97491968e-02 7.21943974e-01
5.16903043e-01 4.97015983e-01 -1.42327890e-01 1.10129267e-01
1.09829259e+00 -1.15488350e+00 -3.12760592e-01 -6.80405021e-01
5.48317671e-01 -2.57014811e-01 8.52271259e-01 5.03813326e-01
-1.27337539e+00 -5.74794412e-01 -1.08640754e+00 1.62044793e-01
-6.54671252e-01 -2.50915401e-02 4.81116652e-01 6.33908927e-01
-1.51994359e+00 8.66488874e-01 -7.68087745e-01 -1.80995256e-01
6.07222617e-01 1.01271653e+00 -2.40821838e-01 1.42191947e-01
-1.11257267e+00 1.01089621e+00 1.04031086e+00 3.82342190e-01
-6.78925455e-01 -6.37818992e-01 -5.78962147e-01 2.89881229e-01
6.13374829e-01 -7.97914445e-01 1.06045008e+00 -1.32876611e+00
-1.57559502e+00 4.71190661e-01 -1.03632197e-01 -8.82000864e-01
2.25736037e-01 -8.53248909e-02 -4.20361042e-01 -1.39524257e-02
-3.20156097e-01 1.05906069e+00 8.62498760e-01 -9.31627333e-01
-6.72845304e-01 -2.64875114e-01 3.91002893e-02 2.77850509e-01
-9.32494223e-01 2.53763460e-02 -8.23136389e-01 -5.45880318e-01
-1.83961689e-01 -1.05684745e+00 -3.57630014e-01 -2.66950667e-01
-2.76184469e-01 -4.30072933e-01 6.86960161e-01 -5.56070328e-01
1.47488475e+00 -1.79457474e+00 4.87526655e-01 4.49403584e-01
1.91496775e-01 8.52891743e-01 -6.63326263e-01 -3.46948169e-02
-3.48725952e-02 2.74982512e-01 -2.09232584e-01 -4.46396530e-01
2.59894710e-02 5.20889282e-01 1.88230127e-01 -1.14023022e-01
3.50635827e-01 9.80009496e-01 -5.44603944e-01 -3.80376875e-01
2.29784450e-03 4.68122125e-01 -6.35207832e-01 -9.18698013e-02
-4.15913492e-01 -1.24357343e-01 -2.91870922e-01 5.65202117e-01
8.69912356e-02 -3.23460072e-01 1.82338253e-01 -2.40487605e-01
-9.31313727e-03 2.09132686e-01 -1.12057650e+00 1.68389595e+00
-5.80600381e-01 6.85971618e-01 2.41994545e-01 -1.16774905e+00
1.07052898e+00 9.29140821e-02 3.35430920e-01 -1.00280607e+00
4.19586331e-01 4.30562288e-01 4.28021371e-01 -2.72415221e-01
5.84945798e-01 4.42567438e-01 1.82427704e-01 2.86552578e-01
2.49018669e-01 2.86534369e-01 5.84626973e-01 -1.70124009e-01
1.17344356e+00 1.19875073e-01 1.15449257e-01 -4.42266166e-01
7.81224370e-01 1.04731254e-01 6.35895073e-01 7.60377049e-01
-1.45023495e-01 3.19125175e-01 1.25549406e-01 -7.72629738e-01
-1.16105270e+00 -5.05357146e-01 1.70644537e-01 1.27643049e+00
-2.60004938e-01 -2.56569624e-01 -7.94435859e-01 -5.15940249e-01
-1.20267920e-01 8.38517725e-01 -5.46187282e-01 -2.99683213e-01
-9.33477283e-01 -6.58095658e-01 7.48984218e-01 4.98598635e-01
6.73911631e-01 -1.42847288e+00 -9.41403568e-01 4.23579544e-01
4.72175442e-02 -9.39362109e-01 -1.96911320e-01 5.10651052e-01
-1.04037917e+00 -8.50450456e-01 -5.60088694e-01 -7.32599437e-01
5.96157610e-01 -1.19076043e-01 1.21252596e+00 3.65367085e-01
-4.13837940e-01 -1.11983292e-01 -1.69509891e-02 -3.43101382e-01
-5.25160134e-01 9.39163327e-01 2.11017013e-01 -1.65458843e-01
2.65597731e-01 -7.26207793e-01 -3.62296194e-01 2.88926184e-01
-1.00534821e+00 1.34904847e-01 1.09983647e+00 7.98822224e-01
5.42063594e-01 2.07809865e-01 7.65112460e-01 -5.27959824e-01
9.50390339e-01 -2.24847808e-01 -7.41534054e-01 3.30494195e-01
-1.24973750e+00 6.32046521e-01 8.23796511e-01 -5.03785789e-01
-8.54187787e-01 5.11181988e-02 -7.18974397e-02 -8.31483781e-01
-2.66498178e-01 6.47258997e-01 7.61953071e-02 -1.23428211e-01
7.81724095e-01 1.65526554e-01 3.09759974e-01 -6.55989945e-01
2.55306046e-02 5.33048868e-01 6.02931321e-01 -4.63905722e-01
6.47502065e-01 -5.35368547e-02 1.26016840e-01 -8.00498664e-01
-6.64415777e-01 -3.36334944e-01 -5.49294233e-01 -1.70880616e-01
5.40301561e-01 -4.92095411e-01 -5.21176159e-01 3.63800406e-01
-1.07258594e+00 -4.92611706e-01 -1.69750378e-01 1.69942752e-01
-1.35743707e-01 -2.08498582e-01 -8.82937908e-02 -5.60430348e-01
-8.36465418e-01 -1.34657598e+00 4.53282446e-01 4.32908028e-01
-4.21481729e-01 -1.05481672e+00 -1.70309320e-02 3.64323497e-01
9.20448661e-01 -2.41804738e-02 9.82601941e-01 -1.06063902e+00
-3.81087005e-01 1.71429589e-02 -9.37771127e-02 5.38346231e-01
-1.35264754e-01 -2.49331102e-01 -9.52717364e-01 -3.41206223e-01
-3.52149606e-01 -2.64376044e-01 7.36112475e-01 3.14057916e-01
1.31895959e+00 -5.65384209e-01 -2.65435010e-01 7.53815532e-01
1.40479040e+00 4.03538764e-01 4.28014219e-01 1.03431165e+00
4.99102175e-01 6.05122030e-01 1.25475023e-02 -1.73132792e-02
4.80161011e-02 7.59976625e-01 7.32504427e-01 1.87001035e-01
-1.77615330e-01 3.45953137e-01 2.58881003e-01 7.71672308e-01
-2.21991822e-01 -2.31117621e-01 -1.10057187e+00 8.29823375e-01
-1.75317168e+00 -7.13222444e-01 3.59690070e-01 1.81002033e+00
8.51330459e-01 3.72278899e-01 6.36492148e-02 1.41063109e-01
4.41219985e-01 5.91134019e-02 -1.06056285e+00 -6.97991312e-01
-1.10566758e-01 5.15207410e-01 5.06265283e-01 2.95156300e-01
-8.20599258e-01 8.93385768e-01 5.78887224e+00 1.02570546e+00
-1.11876547e+00 -1.89391360e-01 5.03097236e-01 -4.60390985e-01
-1.10803485e-01 -2.87766486e-01 -1.00245047e+00 2.87270904e-01
1.24051011e+00 -1.08202808e-01 8.73676181e-01 1.04446518e+00
3.93180288e-02 2.66863406e-01 -9.58030522e-01 8.98512542e-01
-2.59710033e-03 -1.74811530e+00 2.49036804e-01 -1.95374563e-02
6.54096544e-01 4.48582709e-01 -5.59655689e-02 4.66371357e-01
2.10877329e-01 -1.04681277e+00 7.60063648e-01 3.74751896e-01
5.08476615e-01 -1.08446646e+00 8.44645262e-01 3.79869431e-01
-1.09780610e+00 -5.44824183e-01 -1.87362835e-01 1.79605410e-01
-2.12640435e-01 2.62315035e-01 -9.57803845e-01 6.50972724e-01
9.35378909e-01 3.78336668e-01 -9.33318138e-01 9.88702655e-01
1.42181262e-01 3.49333137e-01 -5.36485910e-01 -3.50650460e-01
5.99991143e-01 -2.61020325e-02 7.06629872e-01 1.06704867e+00
2.57241964e-01 -4.15159971e-01 -1.45589486e-01 9.27235246e-01
-1.83666334e-01 -1.75614953e-01 -2.56898135e-01 -2.14864045e-01
7.93592572e-01 1.33470941e+00 -6.77454889e-01 -1.37451738e-01
1.66279271e-01 6.99171245e-01 5.33965170e-01 1.31135777e-01
-4.58281994e-01 -5.64211905e-01 6.81750119e-01 -1.14596181e-01
3.42876643e-01 -4.10893336e-02 -3.35732609e-01 -5.16916990e-01
-3.68785821e-02 -1.16894126e+00 3.09716731e-01 -4.89436686e-01
-7.80546844e-01 1.11205268e+00 8.98865089e-02 -6.31198943e-01
-6.76164746e-01 -6.32340133e-01 -3.60194981e-01 8.22744012e-01
-1.45137477e+00 -9.23331380e-01 -2.39455879e-01 4.57544744e-01
8.67342472e-01 -6.50410235e-01 8.16533148e-01 2.85246283e-01
-8.75340104e-01 7.95874655e-01 2.47072661e-04 -2.59628743e-01
2.18160421e-01 -1.03749192e+00 4.39655513e-01 7.36537576e-01
8.03758726e-02 5.74213862e-01 5.82752526e-01 -2.40232542e-01
-1.46886444e+00 -1.06677270e+00 7.59747148e-01 -1.47799868e-02
3.88279498e-01 -1.67827457e-01 -9.56898332e-01 3.32819045e-01
1.91209018e-01 -5.14272153e-01 2.63511091e-01 1.55344263e-01
-5.49380993e-03 -3.67728531e-01 -8.96690309e-01 5.96623659e-01
1.14374411e+00 -2.66427994e-01 -5.02304375e-01 -5.84976301e-02
6.65815115e-01 -2.24480391e-01 -7.63330638e-01 4.45262760e-01
4.88603473e-01 -9.40645158e-01 1.12882173e+00 -5.14382124e-01
1.86629266e-01 -2.89188307e-02 1.14740901e-01 -1.39786744e+00
-4.95674521e-01 -7.73833692e-01 -3.88495058e-01 9.74491417e-01
6.07379615e-01 -7.95259178e-01 1.17821443e+00 6.37793064e-01
-3.85795474e-01 -1.10548306e+00 -1.07610428e+00 -1.04489458e+00
-2.60062426e-01 -2.39829227e-01 8.27231765e-01 7.89704263e-01
-6.78647876e-01 3.67454529e-01 -1.24239482e-01 -5.34225628e-02
4.46259707e-01 -2.44310901e-01 4.45377022e-01 -1.48976552e+00
-3.93073320e-01 -1.09565270e+00 -1.59247369e-01 -6.33032024e-01
1.59202009e-01 -9.42064345e-01 -1.19539294e-02 -1.42440772e+00
-2.68560022e-01 -5.18226564e-01 -6.32583976e-01 9.35791671e-01
1.50140777e-01 1.42367348e-01 1.62753299e-01 2.88642973e-01
-3.89978766e-01 4.79399234e-01 8.86168063e-01 -1.08999185e-01
-5.77032864e-01 -9.05956104e-02 -6.18460834e-01 6.73807740e-01
1.24454236e+00 -4.98159736e-01 -6.24822795e-01 -6.39359772e-01
5.51026285e-01 -6.08622015e-01 1.63011760e-01 -1.23110032e+00
5.51319599e-01 1.10169739e-01 2.47260153e-01 -5.17311513e-01
3.98818225e-01 -8.66011798e-01 3.63887221e-01 6.15395665e-01
-4.72563297e-01 4.88407284e-01 6.58234179e-01 3.09487343e-01
-1.44201428e-01 -6.92698240e-01 6.54536605e-01 -2.11244956e-01
-1.03316724e+00 1.39216501e-02 -3.14070135e-01 -1.08228356e-01
8.50488424e-01 -4.77277339e-01 -2.12939650e-01 1.15308963e-01
-5.15306890e-01 4.64076102e-01 1.31622583e-01 5.96561611e-01
5.95517874e-01 -1.07587528e+00 -4.52686667e-01 1.35204151e-01
-2.63080537e-01 3.53444844e-01 8.16682726e-02 5.17271757e-01
-5.08128405e-01 6.83933794e-01 -4.66131210e-01 -3.74407798e-01
-1.46668017e+00 3.14738482e-01 5.09937584e-01 -7.43657947e-01
-3.65871012e-01 9.18120444e-01 -4.04361665e-01 -7.01777458e-01
4.42917585e-01 -1.57512836e-02 -3.17235470e-01 3.07885744e-02
3.89701754e-01 5.56977272e-01 3.90238672e-01 -2.81364590e-01
-3.96391213e-01 1.33873314e-01 -4.26272571e-01 -9.08694714e-02
1.84214449e+00 2.00254157e-01 -1.15445696e-01 -7.43821561e-02
1.06297398e+00 -7.33014703e-01 -1.04851818e+00 -2.69413412e-01
3.64890605e-01 2.04608351e-01 5.72283864e-01 -9.86489475e-01
-1.38116872e+00 4.96120751e-01 5.81433535e-01 8.07528943e-02
1.46106696e+00 -3.45332712e-01 7.09832728e-01 9.69386339e-01
2.16638073e-01 -1.29982328e+00 4.69632540e-03 6.39398694e-01
9.74683642e-01 -7.41049469e-01 3.35615575e-02 3.37044269e-01
-2.30486259e-01 1.21130693e+00 8.93358350e-01 8.99294317e-02
3.09396833e-01 2.33960226e-01 -2.35852286e-01 -2.40439206e-01
-1.05613005e+00 -6.73415884e-03 5.84814847e-01 4.71023768e-01
-2.37512797e-01 -4.04803693e-01 7.03577250e-02 3.49901497e-01
-2.92853028e-01 -2.44677048e-02 -1.17913848e-02 1.05324113e+00
-3.40281188e-01 -1.27256644e+00 -1.89487591e-01 6.88194454e-01
-2.90509760e-01 -1.43757343e-01 -5.28041959e-01 8.22580397e-01
9.16707590e-02 5.70886374e-01 8.53473842e-02 -6.31684244e-01
3.64926547e-01 3.19154352e-01 2.75851101e-01 -1.48955151e-01
-1.07737350e+00 -1.98516726e-01 4.06458318e-01 -7.25866020e-01
-3.15488398e-01 -3.67763311e-01 -9.69319642e-01 -8.97997543e-02
-3.19502592e-01 4.08912487e-02 1.09545720e+00 9.83355999e-01
9.38493431e-01 9.52207327e-01 2.07591087e-01 -1.07312393e+00
-6.64838135e-01 -9.44446087e-01 2.77155619e-02 -1.17672332e-01
1.17889985e-01 -5.80291867e-01 -1.86996385e-01 -1.19544953e-01] | [8.447500228881836, 3.2792813777923584] |
d9c000ca-6f14-4c6d-b3c5-613750c834e1 | visual-aware-text-to-speech | 2306.12020 | null | https://arxiv.org/abs/2306.12020v1 | https://arxiv.org/pdf/2306.12020v1.pdf | Visual-Aware Text-to-Speech | Dynamically synthesizing talking speech that actively responds to a listening head is critical during the face-to-face interaction. For example, the speaker could take advantage of the listener's facial expression to adjust the tones, stressed syllables, or pauses. In this work, we present a new visual-aware text-to-speech (VA-TTS) task to synthesize speech conditioned on both textual inputs and sequential visual feedback (e.g., nod, smile) of the listener in face-to-face communication. Different from traditional text-to-speech, VA-TTS highlights the impact of visual modality. On this newly-minted task, we devise a baseline model to fuse phoneme linguistic information and listener visual signals for speech synthesis. Extensive experiments on multimodal conversation dataset ViCo-X verify our proposal for generating more natural audio with scenario-appropriate rhythm and prosody. | ['Tao Mei', 'Tiejun Zhao', 'Ting Yao', 'Wei zhang', 'Yalong Bai', 'Mohan Zhou'] | 2023-06-21 | null | null | null | null | ['speech-synthesis'] | ['speech'] | [ 2.92803973e-01 3.02093595e-01 7.52013028e-02 -6.34805441e-01
-8.53766799e-01 -6.05819881e-01 7.15419412e-01 -4.51422542e-01
2.06756309e-01 5.05580664e-01 7.39799976e-01 -1.18396088e-01
6.97379529e-01 -2.89631635e-01 -3.44481349e-01 -7.33753741e-01
5.18858433e-01 9.24623162e-02 -1.54576108e-01 -3.56657147e-01
1.27894953e-01 6.12132132e-01 -1.91415012e+00 1.00569248e+00
4.97335047e-02 1.02286303e+00 3.04794550e-01 1.22990370e+00
-3.05389553e-01 7.08898306e-01 -9.21727479e-01 -8.61646514e-03
-9.00569856e-02 -8.39886129e-01 -4.52231616e-01 4.51464981e-01
4.33887094e-01 -7.39449337e-02 -2.22721249e-02 6.96986556e-01
9.22001243e-01 1.31415695e-01 5.92298687e-01 -1.38925290e+00
-2.39824876e-01 6.33623183e-01 -4.67708498e-01 -1.22981913e-01
9.70640600e-01 4.47328240e-01 7.27550745e-01 -1.27197433e+00
6.71562910e-01 1.77188194e+00 -1.91148911e-02 1.05820251e+00
-1.22406054e+00 -6.60105348e-01 1.78753793e-01 -7.25782365e-02
-1.28790152e+00 -1.30922318e+00 1.05967414e+00 -2.93193668e-01
7.51484752e-01 7.12733865e-01 4.68384534e-01 1.56433403e+00
-4.90195416e-02 6.77029312e-01 1.03470647e+00 -7.20790446e-01
-1.35692833e-02 3.98076743e-01 -5.66187501e-01 2.21528798e-01
-9.90428925e-01 2.21440136e-01 -1.01313722e+00 -2.52298471e-02
3.49600434e-01 -6.49065793e-01 -5.76472640e-01 2.04044133e-01
-1.10084498e+00 4.97856796e-01 1.27095610e-01 3.80427450e-01
-2.72062629e-01 5.34454128e-03 4.61077601e-01 3.33438337e-01
2.51207024e-01 6.00110441e-02 1.09048918e-01 -1.83621064e-01
-9.46131229e-01 -7.34777078e-02 7.08299756e-01 7.64487028e-01
2.74362355e-01 6.33334756e-01 -6.79970741e-01 1.03581357e+00
4.00865197e-01 8.06809366e-01 3.77984613e-01 -8.72552574e-01
3.28195155e-01 -3.05427518e-02 6.85758330e-03 -6.29057527e-01
-4.25270289e-01 7.36587271e-02 -4.94860321e-01 5.65116704e-01
4.84450944e-02 -4.56711918e-01 -7.38515377e-01 1.82518351e+00
4.90894943e-01 -1.88662276e-01 1.25987053e-01 1.05296373e+00
1.24962044e+00 9.45543528e-01 -6.39986992e-02 -6.66417003e-01
1.45599401e+00 -7.31404841e-01 -1.31163466e+00 -1.45648956e-01
1.81552302e-03 -1.01203489e+00 1.28215647e+00 3.49714279e-01
-1.32184112e+00 -8.35288942e-01 -7.18339860e-01 -5.97463809e-02
1.41443297e-01 3.22854578e-01 -3.88904475e-02 6.22183681e-01
-1.25969386e+00 -1.02028109e-01 -2.09818229e-01 -3.21930617e-01
-1.33110210e-01 3.48871350e-02 -4.79456782e-01 5.94401717e-01
-9.33653831e-01 5.17426372e-01 3.99637073e-02 2.58996844e-01
-8.51205111e-01 -1.83565289e-01 -9.12624419e-01 2.56914049e-02
2.94641286e-01 -5.30802488e-01 1.64412296e+00 -1.50500453e+00
-2.31923771e+00 9.10537481e-01 -7.38671064e-01 -4.28620493e-04
4.38321680e-01 2.09147081e-01 -6.83583081e-01 4.77862835e-01
-2.74720788e-01 1.15208387e+00 1.32620561e+00 -1.66156983e+00
-2.41123378e-01 -1.97811797e-01 -4.30723250e-01 4.15467709e-01
3.82385738e-02 4.52359200e-01 -3.72673512e-01 -8.49648714e-01
-2.00663075e-01 -7.76702940e-01 4.12578344e-01 1.08903505e-01
-7.90110886e-01 -9.98950005e-02 1.07275271e+00 -5.49322009e-01
1.01529980e+00 -2.44065380e+00 3.43814373e-01 1.59358323e-01
-7.79946297e-02 1.95228428e-01 -3.30059648e-01 5.48483908e-01
-4.28858668e-01 -3.03149313e-01 3.29835825e-02 -7.86779463e-01
-1.43129885e-01 -3.91136073e-02 -4.74645436e-01 1.76760867e-01
1.81832284e-01 8.29932213e-01 -4.55875576e-01 -7.26708531e-01
2.45287850e-01 6.67830110e-01 -3.83658350e-01 7.07364440e-01
-2.10619628e-01 8.73440802e-01 3.76184881e-02 7.26309896e-01
3.76243562e-01 3.51429671e-01 2.05675095e-01 -2.28476003e-01
-9.89945754e-02 2.97125995e-01 -8.79109085e-01 1.31545889e+00
-6.24664903e-01 9.31001723e-01 6.90506876e-01 -3.42642397e-01
1.14104092e+00 8.62957656e-01 1.17736394e-02 -6.81187868e-01
3.90874147e-01 -1.78564474e-01 1.80926267e-02 -6.89402699e-01
1.45607561e-01 -4.15460527e-01 -8.24385062e-02 2.71069378e-01
-2.59080413e-03 -5.79712272e-01 -2.17779815e-01 -1.57093685e-02
4.52535748e-01 -1.04927175e-01 4.30994630e-01 2.27385774e-01
7.31857002e-01 -6.46161437e-01 -7.36582279e-02 2.76744097e-01
-1.25167280e-01 8.32857370e-01 5.19209087e-01 2.53533542e-01
-4.85475212e-01 -1.20858574e+00 3.03374171e-01 1.51149774e+00
-3.70083570e-01 -2.53943264e-01 -1.03452528e+00 -3.04038435e-01
-5.42172790e-01 9.84725356e-01 -4.78256971e-01 2.46885885e-03
-4.34381276e-01 2.06461608e-01 5.94364762e-01 2.17583880e-01
-4.08048257e-02 -1.78491020e+00 -4.37750995e-01 2.44930580e-01
-5.26993930e-01 -1.17042565e+00 -1.09354544e+00 -1.33501455e-01
-2.24451885e-01 -4.28654641e-01 -8.75791907e-01 -7.67439902e-01
4.59192008e-01 2.06109971e-01 7.56821394e-01 -4.00513321e-01
-2.65650690e-01 5.73944688e-01 -3.42263401e-01 -4.41013604e-01
-9.07994151e-01 -5.29709578e-01 8.37157294e-02 5.68669915e-01
-2.08685622e-01 -5.94571114e-01 -3.43364865e-01 4.41633016e-01
-6.46268368e-01 4.09155786e-01 2.35869214e-01 4.43735182e-01
2.89596230e-01 -6.07295573e-01 7.90567994e-01 -3.20324242e-01
1.02408123e+00 -8.02253038e-02 -1.95925266e-01 1.12324886e-01
2.83274084e-01 -2.68698126e-01 5.32556236e-01 -8.88387263e-01
-1.44859910e+00 2.73158729e-01 -3.10369372e-01 -6.84593678e-01
-4.20741946e-01 3.11501082e-02 -5.78646898e-01 1.46079645e-01
6.38556480e-01 1.27292320e-01 1.06186189e-01 -1.25915587e-01
5.68387687e-01 1.26953781e+00 1.06718481e+00 -3.21673989e-01
5.11993349e-01 4.14531201e-01 -1.59142077e-01 -1.28558910e+00
-2.83064663e-01 -2.11507931e-01 -3.14852685e-01 -7.50203431e-01
9.33195293e-01 -8.63447070e-01 -1.25481260e+00 4.32413280e-01
-1.48421788e+00 -4.13323551e-01 -1.10275857e-01 2.95629948e-01
-7.01387942e-01 1.48449063e-01 -4.21758443e-01 -1.18001842e+00
-4.16132033e-01 -1.26371109e+00 1.29907155e+00 2.17602611e-01
-5.53699613e-01 -4.32686538e-01 -9.61059798e-03 5.38982391e-01
3.78304273e-01 5.88755570e-02 5.55874705e-01 -2.63665110e-01
-2.63552442e-02 1.15725756e-01 6.49064705e-02 2.23461226e-01
4.54763144e-01 3.99310589e-01 -1.59652531e+00 7.60553079e-03
1.82996783e-02 -3.39482456e-01 4.57459956e-01 5.08765578e-01
1.06132102e+00 -5.85032582e-01 5.76819517e-02 3.97422343e-01
5.02230406e-01 4.93407071e-01 5.58448076e-01 -7.26406634e-01
5.40125370e-01 1.14762509e+00 6.24183595e-01 5.66605687e-01
1.70829490e-01 1.04210556e+00 3.85183811e-01 -2.56002963e-01
-6.21999383e-01 -3.03585082e-01 8.73077393e-01 6.15399063e-01
2.81231225e-01 -6.35307789e-01 -4.07223225e-01 2.27599978e-01
-1.25785375e+00 -1.06677377e+00 9.09193326e-03 1.98447597e+00
9.95754063e-01 -1.32418334e-01 2.92741597e-01 2.01193824e-01
9.28696275e-01 3.64429206e-01 -3.65194649e-01 -1.08637595e+00
-2.20373943e-01 1.32386953e-01 -3.13265145e-01 1.03077710e+00
-6.18463695e-01 9.90115643e-01 5.89005327e+00 9.13304448e-01
-1.90581524e+00 -4.23708037e-02 5.18877149e-01 -5.26339173e-01
-4.92755413e-01 -5.02091825e-01 -6.25600696e-01 2.85216302e-01
9.69187021e-01 1.40289411e-01 7.17909515e-01 5.17360330e-01
8.82752299e-01 -9.47924703e-02 -1.20647824e+00 1.24035215e+00
4.70528126e-01 -1.09959686e+00 8.66700336e-02 -2.40407139e-01
1.27698094e-01 -4.78797555e-01 2.59941041e-01 3.63616762e-03
-2.60262340e-01 -1.23926187e+00 1.07387459e+00 3.53775173e-01
1.32946861e+00 -5.43561399e-01 -4.97406647e-02 8.06932226e-02
-1.35810256e+00 1.41629830e-01 5.11563480e-01 2.67475963e-01
5.39746225e-01 -1.11541659e-01 -1.37130749e+00 4.27622208e-03
3.63171786e-01 2.39451788e-02 -2.02379987e-01 3.67674768e-01
-4.26188260e-01 6.73658788e-01 -5.18454388e-02 -6.01453893e-02
-1.92247659e-01 3.65639657e-01 9.59846675e-01 1.44186974e+00
2.97293633e-01 1.75967664e-01 -7.83271194e-02 7.46395171e-01
2.68296730e-02 4.21084881e-01 -7.65734732e-01 -1.94745809e-01
5.08341432e-01 1.28079093e+00 -3.50102335e-01 -1.70261756e-01
-1.74437329e-01 1.13557065e+00 -2.59595543e-01 5.98795354e-01
-7.19267905e-01 -3.16277683e-01 5.54141045e-01 2.25810170e-01
1.76717639e-01 6.33016648e-03 2.38279458e-02 -6.40376449e-01
6.18849974e-03 -1.23668087e+00 2.19058804e-02 -1.45263791e+00
-7.91529715e-01 9.11949933e-01 -6.05259500e-02 -1.22149813e+00
-7.28406429e-01 -3.73766989e-01 -9.59131062e-01 9.92439628e-01
-1.07324517e+00 -1.22350895e+00 -2.22613126e-01 7.73581445e-01
9.48583841e-01 -7.24790320e-02 6.68737173e-01 -3.61595824e-02
-2.24484369e-01 7.46293843e-01 -7.40828216e-01 -5.82501814e-02
1.06908965e+00 -8.27749550e-01 2.50232846e-01 3.93362284e-01
1.20184362e-01 2.02892289e-01 1.12182891e+00 -3.32394451e-01
-1.22106457e+00 -6.47262812e-01 1.00371695e+00 9.87328030e-03
2.42526829e-01 -7.29867935e-01 -6.98329329e-01 4.29190934e-01
5.89412332e-01 -2.53188729e-01 7.67366290e-01 -4.63487595e-01
-2.88192213e-01 -2.17439294e-01 -1.20665133e+00 9.08688128e-01
5.67881763e-01 -8.03970397e-01 -4.52371866e-01 -1.29608419e-02
8.16747904e-01 -3.90259445e-01 -2.39379674e-01 1.16446428e-01
6.23214543e-01 -1.08804023e+00 9.36905622e-01 -3.44167769e-01
7.57740960e-02 -3.77200335e-01 -2.51184464e-01 -1.43333554e+00
3.30777735e-01 -1.34982729e+00 4.07293707e-01 1.51503003e+00
2.86776483e-01 -3.93998086e-01 2.19576746e-01 1.41587391e-01
-1.36709064e-01 -5.18156663e-02 -1.06285942e+00 -1.24450818e-01
-4.24492240e-01 -5.04442990e-01 4.68191475e-01 6.73374116e-01
3.87815922e-01 6.41353190e-01 -8.05677950e-01 1.45569026e-01
9.69770998e-02 -6.75434992e-02 1.17837322e+00 -7.96915770e-01
-4.31548625e-01 -3.23714823e-01 1.11574806e-01 -9.44544315e-01
2.50464588e-01 -6.35186553e-01 4.75886792e-01 -1.11061347e+00
-2.53458142e-01 4.73495573e-01 2.67937422e-01 4.76219952e-01
4.15243804e-02 -3.68903503e-02 4.57699865e-01 -3.14416826e-01
-6.52261823e-02 7.34806299e-01 1.64865124e+00 -1.81399081e-02
-5.70128739e-01 7.44863674e-02 -4.33300078e-01 5.78299105e-01
5.37496448e-01 -2.01784432e-01 -4.12506312e-01 1.48209155e-01
-1.87186494e-01 1.19316649e+00 3.09322715e-01 -5.54362178e-01
1.91523015e-01 -3.26526016e-01 1.49611667e-01 -5.92514336e-01
1.11821926e+00 -6.33553743e-01 5.94336055e-02 1.34667277e-01
-6.13473058e-01 -2.55957637e-02 3.72407794e-01 2.52378076e-01
-4.03558254e-01 2.51200885e-01 9.40690160e-01 3.35881971e-02
-1.88824579e-01 -8.56002569e-02 -9.07527685e-01 -9.66080725e-02
7.47799814e-01 -2.90391326e-01 -2.58712918e-01 -1.28601348e+00
-1.05056775e+00 -2.22088885e-03 5.97832315e-02 6.39263213e-01
8.50029051e-01 -1.37478650e+00 -8.29917610e-01 4.06602830e-01
1.92416489e-01 -5.12810230e-01 5.00533879e-01 6.68042243e-01
-3.41672637e-02 3.84547114e-01 -2.92766839e-01 -7.45929420e-01
-1.93875790e+00 3.66173238e-01 4.59313124e-01 5.75981081e-01
-2.34705180e-01 9.67975318e-01 6.32914960e-01 -3.59281041e-02
5.17421365e-01 -2.28882551e-01 -3.46098065e-01 3.63190830e-01
5.96722305e-01 1.66676551e-01 -2.03735888e-01 -1.10804391e+00
-3.10919642e-01 4.27708209e-01 4.37202662e-01 -9.11117494e-01
7.05389500e-01 -4.55843508e-01 1.56922683e-01 8.15225840e-01
1.12065160e+00 6.22639894e-01 -1.04760730e+00 -2.31669750e-02
-6.32388413e-01 -3.96500856e-01 -1.06415242e-01 -8.35146666e-01
-1.00075424e+00 1.34237921e+00 4.63528603e-01 5.70381507e-02
1.23442602e+00 1.36293843e-01 5.14531195e-01 1.44079477e-01
1.02993306e-02 -9.49291706e-01 6.25023186e-01 1.01466447e-01
1.65845954e+00 -1.03487730e+00 -6.52043462e-01 -3.38365018e-01
-1.25859630e+00 1.25157332e+00 5.93657792e-01 6.57266200e-01
4.57077920e-01 4.51808244e-01 8.15757930e-01 3.57106663e-02
-1.12193215e+00 -1.82312280e-01 6.13820314e-01 7.19939709e-01
5.49861133e-01 1.93623841e-01 4.13913101e-01 2.82456160e-01
-5.83621204e-01 -3.27797920e-01 3.33105356e-01 3.20532233e-01
-6.02241576e-01 -7.85138607e-01 -8.35565090e-01 -3.81067365e-01
-2.80229270e-01 -2.99193896e-02 -9.54425991e-01 3.11096311e-01
-2.13380456e-01 1.67823911e+00 9.11541060e-02 -2.73513943e-01
3.05398166e-01 4.65047985e-01 5.21467149e-01 -5.86326778e-01
-7.99734533e-01 8.27613115e-01 2.80176371e-01 -5.46523452e-01
-2.83028275e-01 -5.83949685e-01 -1.37286103e+00 6.47937879e-02
1.46160852e-02 -2.01767251e-01 7.56309390e-01 6.77188754e-01
2.34273404e-01 6.40526593e-01 8.72754753e-01 -1.31487560e+00
-9.85004827e-02 -1.17516243e+00 -4.00524855e-01 2.68677890e-01
7.35484421e-01 -3.33081484e-01 -6.50110006e-01 8.98084417e-02] | [13.276756286621094, -0.4130769371986389] |
7e51a2b7-05d0-4db9-8f5a-9401c1661c7c | characterizing-the-efficiency-of-graph-neural | 2211.03021 | null | https://arxiv.org/abs/2211.03021v1 | https://arxiv.org/pdf/2211.03021v1.pdf | Characterizing the Efficiency of Graph Neural Network Frameworks with a Magnifying Glass | Graph neural networks (GNNs) have received great attention due to their success in various graph-related learning tasks. Several GNN frameworks have then been developed for fast and easy implementation of GNN models. Despite their popularity, they are not well documented, and their implementations and system performance have not been well understood. In particular, unlike the traditional GNNs that are trained based on the entire graph in a full-batch manner, recent GNNs have been developed with different graph sampling techniques for mini-batch training of GNNs on large graphs. While they improve the scalability, their training times still depend on the implementations in the frameworks as sampling and its associated operations can introduce non-negligible overhead and computational cost. In addition, it is unknown how much the frameworks are 'eco-friendly' from a green computing perspective. In this paper, we provide an in-depth study of two mainstream GNN frameworks along with three state-of-the-art GNNs to analyze their performance in terms of runtime and power/energy consumption. We conduct extensive benchmark experiments at several different levels and present detailed analysis results and observations, which could be helpful for further improvement and optimization. | ['Chul-Ho Lee', 'Bradley Rees', 'Jongryool Kim', 'Xin Huang'] | 2022-11-06 | null | null | null | null | ['graph-sampling'] | ['graphs'] | [-4.49667126e-02 5.72880879e-02 -3.71070445e-01 -3.06565195e-01
1.63344759e-02 -3.07522416e-01 3.66844177e-01 1.81633487e-01
-2.46481746e-01 5.21379769e-01 -4.23311681e-01 -5.62800646e-01
-1.92416117e-01 -1.21615911e+00 -6.73464358e-01 -6.90018535e-01
-3.99572879e-01 4.14046794e-01 2.67804384e-01 -1.82397515e-01
-9.70124975e-02 4.75419641e-01 -1.30837083e+00 -3.73044938e-01
8.54339302e-01 8.40501189e-01 1.13608293e-01 6.44952297e-01
-3.94485332e-02 1.01616633e+00 -5.60075402e-01 -6.09355450e-01
3.22699666e-01 -4.49835718e-01 -5.01904368e-01 -1.64474785e-01
3.50519240e-01 -1.46069542e-01 -8.11137438e-01 1.27686036e+00
6.64434373e-01 2.41813496e-01 1.03911720e-01 -1.56825137e+00
-7.56442726e-01 8.96433532e-01 -4.53686148e-01 -9.21430532e-03
-1.24651931e-01 1.06876791e-01 9.90443647e-01 -2.66203851e-01
4.16659027e-01 1.06797779e+00 8.34097803e-01 5.19793153e-01
-9.77984965e-01 -8.77833247e-01 2.08212316e-01 4.52139765e-01
-1.28266346e+00 -3.35011572e-01 9.27492738e-01 3.00619841e-01
1.29565144e+00 1.31266490e-01 1.08845913e+00 1.06162071e+00
2.24542618e-01 7.10391939e-01 6.21149361e-01 -5.56934416e-01
3.98622662e-01 -1.15623601e-01 2.39794329e-02 8.26114714e-01
7.96994090e-01 -1.33326873e-01 -2.90479809e-01 3.69203165e-02
7.62321532e-01 2.25522161e-01 -1.34973317e-01 -5.22513568e-01
-6.15056276e-01 9.48911786e-01 1.12559175e+00 2.83672214e-01
-3.02959770e-01 8.69128168e-01 7.25503981e-01 2.42585689e-01
5.13728142e-01 1.50012821e-01 -2.25000933e-01 -8.44958872e-02
-9.47355986e-01 -1.11527503e-01 1.19030845e+00 1.23242080e+00
8.14885855e-01 4.78498995e-01 2.66340107e-01 7.28738666e-01
3.22256863e-01 1.92906111e-01 3.47488403e-01 -4.23140764e-01
3.48165989e-01 9.89874601e-01 -8.17738175e-01 -1.28083432e+00
-5.41441202e-01 -7.12727904e-01 -1.42000759e+00 -1.29814684e-01
9.24135000e-02 -1.69013098e-01 -1.01981592e+00 1.46780813e+00
2.36143902e-01 3.75204057e-01 -2.13424325e-01 5.15088916e-01
1.00842869e+00 6.25605941e-01 1.41636774e-01 1.92587733e-01
9.88485277e-01 -1.60172844e+00 -4.70037162e-01 -5.58710158e-01
7.73287237e-01 -2.60033369e-01 8.87026668e-01 3.78257722e-01
-8.10490310e-01 -3.43997955e-01 -1.17827237e+00 8.35091993e-03
-6.90977275e-01 5.54846600e-03 1.18890440e+00 1.10422778e+00
-1.46299183e+00 9.86685753e-01 -1.06760085e+00 -8.22545469e-01
6.27235949e-01 4.84954059e-01 -2.27896869e-02 -3.34728181e-01
-9.42249537e-01 7.21101940e-01 5.67816913e-01 2.53721565e-01
-1.04354453e+00 -3.37198555e-01 -9.02026057e-01 2.27804869e-01
5.91390848e-01 -6.44517362e-01 1.05125165e+00 -7.83478737e-01
-1.51697063e+00 4.11663085e-01 4.92683798e-01 -6.69333637e-01
2.15410233e-01 9.94465053e-02 -5.70293486e-01 -5.79482056e-02
-3.83440912e-01 4.80730146e-01 3.46871048e-01 -7.40444779e-01
-2.79286236e-01 -1.55008689e-01 5.29799879e-01 2.40728214e-01
-7.41025686e-01 -1.05943851e-01 -7.01121509e-01 -4.82152820e-01
-1.84224203e-01 -1.01221800e+00 -5.84536254e-01 -4.35916260e-02
-4.98485595e-01 -2.21363276e-01 7.78102040e-01 -1.62628964e-01
1.45530117e+00 -1.85122979e+00 -3.77524704e-01 2.20641896e-01
6.50233626e-01 5.18452227e-01 -1.12087667e-01 6.16611004e-01
1.30935133e-01 2.61056244e-01 8.04398730e-02 -9.23524424e-02
1.00945480e-01 3.40706259e-01 2.70755529e-01 5.83805680e-01
-1.00543447e-01 1.14803040e+00 -1.06123090e+00 -3.22863698e-01
4.24949050e-01 6.39986992e-01 -4.87559378e-01 1.11689240e-01
-2.35331580e-01 -1.18736893e-01 -3.33748966e-01 6.93945408e-01
4.64276284e-01 -6.69263542e-01 6.63939118e-01 -1.20268360e-01
4.53225970e-01 2.81004280e-01 -1.00819135e+00 1.46253300e+00
-3.62398028e-01 7.78972745e-01 -4.20750901e-02 -1.41846371e+00
8.66383076e-01 -6.32349327e-02 3.13904464e-01 -8.30196440e-01
2.39158317e-01 1.24209009e-01 2.24886686e-01 -4.78967279e-02
4.89126056e-01 2.97724158e-01 1.44381583e-01 5.07830203e-01
2.79083133e-01 2.24036798e-01 5.94833851e-01 2.49025762e-01
1.50266409e+00 1.04703270e-01 4.16076779e-01 -2.27370411e-01
1.34597138e-01 -6.85984567e-02 1.73185095e-01 6.99365795e-01
-2.28432134e-01 1.09308809e-01 4.89244878e-01 -6.79455101e-01
-9.29103673e-01 -6.25256956e-01 3.70924771e-01 1.26768696e+00
1.74328923e-01 -8.63975525e-01 -8.60111296e-01 -7.49203563e-01
-4.22953904e-01 5.83023846e-01 -3.57894361e-01 -3.04159403e-01
-5.68580389e-01 -1.12869656e+00 6.68954611e-01 6.09388947e-01
6.90640390e-01 -1.17494643e+00 -7.01768637e-01 4.47022587e-01
3.98170024e-01 -1.00957465e+00 -2.13938788e-01 5.46840489e-01
-1.17858398e+00 -1.22111702e+00 -4.56179470e-01 -8.52813184e-01
9.54192460e-01 5.16421020e-01 1.53102553e+00 5.33813477e-01
-2.21248999e-01 1.97441146e-01 -4.55401540e-01 -5.70249677e-01
-2.32223496e-01 5.22017479e-01 -1.42625034e-01 -4.75722373e-01
3.33679557e-01 -8.09129477e-01 -7.01305985e-01 9.57536846e-02
-8.63911331e-01 9.74656269e-02 6.62303507e-01 5.78144372e-01
4.09425765e-01 4.30638343e-01 3.13165367e-01 -1.20617497e+00
7.59957433e-01 -5.34477949e-01 -6.79177225e-01 3.48865390e-01
-1.08823180e+00 5.95584735e-02 9.98871565e-01 -3.51931900e-01
-3.98032427e-01 -3.25878590e-01 -1.41305670e-01 -4.16958332e-01
1.61018804e-01 7.11106420e-01 -1.87432449e-02 -3.73311639e-01
6.81687772e-01 7.45171383e-02 -2.41216585e-01 -2.97838867e-01
3.47088933e-01 1.99371964e-01 9.09145027e-02 -4.44803089e-01
7.98550367e-01 7.87416697e-02 1.63825020e-01 -7.23111749e-01
-4.92176443e-01 -2.61538625e-01 -1.97281376e-01 -3.41522217e-01
4.97575283e-01 -7.96050847e-01 -6.66076541e-01 5.77923775e-01
-6.77738130e-01 -8.41321588e-01 -7.17736259e-02 5.85875846e-02
-8.32276121e-02 3.08449388e-01 -7.45893598e-01 -7.12715685e-01
-9.01266575e-01 -1.06002259e+00 5.52111328e-01 5.14977157e-01
1.94819108e-01 -1.47261488e+00 -1.53055429e-01 -2.71630436e-01
9.71037090e-01 2.40085989e-01 8.34683180e-01 -7.84550011e-01
-7.11532116e-01 -3.50815475e-01 -4.59574759e-01 3.40250254e-01
1.23319505e-02 1.55451447e-01 -7.43853688e-01 -8.10705543e-01
-3.32712263e-01 -2.90941209e-01 7.12560594e-01 5.16424477e-01
1.40164793e+00 -5.04181981e-01 -6.16243660e-01 9.15799677e-01
2.01913667e+00 6.83953092e-02 6.37337506e-01 3.06947231e-01
1.04816103e+00 4.17743139e-02 1.02858610e-01 5.41343614e-02
2.99735695e-01 1.96543291e-01 1.06680059e+00 -2.78602213e-01
-1.34229362e-01 -2.55722612e-01 4.27733839e-01 1.23439264e+00
-2.88245738e-01 -8.42960835e-01 -1.00376296e+00 4.04021621e-01
-1.77521849e+00 -6.26409471e-01 -1.31623402e-01 1.98500967e+00
2.71334589e-01 2.94523984e-01 1.44729763e-01 1.54896304e-01
8.00256670e-01 4.80316162e-01 -8.06783855e-01 -3.06877285e-01
2.42768049e-01 4.81915623e-01 7.29176641e-01 -9.65065733e-02
-8.30324471e-01 8.72988045e-01 6.49506760e+00 9.52816188e-01
-1.15609705e+00 2.98694540e-02 7.37776756e-01 -2.76921373e-02
-9.21363533e-02 1.47143885e-01 -6.61551535e-01 3.01887989e-01
1.60103202e+00 -4.33838874e-01 8.32858741e-01 1.49928570e+00
-3.23096246e-01 2.78385337e-02 -1.07187247e+00 1.04737008e+00
-7.21951248e-03 -1.36147273e+00 -5.47980443e-02 1.40810221e-01
6.02219522e-01 5.99298775e-01 -4.18383271e-01 7.39114165e-01
7.67988503e-01 -1.13986230e+00 4.31011587e-01 -2.06615925e-02
8.33550930e-01 -1.01493859e+00 7.79918194e-01 2.91894615e-01
-1.62161362e+00 -3.19141056e-03 -8.81075621e-01 -3.19767267e-01
-7.24450126e-02 8.47684205e-01 -7.83940971e-01 7.56614804e-01
8.48592877e-01 7.41710663e-01 -8.96407425e-01 1.08293056e+00
-3.13299239e-01 8.36751103e-01 -4.51038808e-01 -6.56517923e-01
3.40297818e-01 -3.75846982e-01 3.66989076e-02 1.13174474e+00
3.70137662e-01 -2.66255707e-01 2.14084387e-01 8.44844341e-01
-5.96430779e-01 1.43441305e-01 -5.74722171e-01 -3.80168468e-01
5.76969862e-01 1.60779858e+00 -1.13604224e+00 -2.85760194e-01
-5.10405600e-01 6.43500030e-01 6.21839046e-01 2.52634346e-01
-1.23378634e+00 -6.48241460e-01 3.88834924e-01 1.10176548e-01
1.03945725e-01 -2.06952989e-01 1.53505430e-01 -9.72868443e-01
-1.92544311e-02 -6.77116632e-01 4.29047287e-01 -5.45349777e-01
-1.32432413e+00 8.00046325e-01 -1.83064342e-01 -8.24529588e-01
-3.07710562e-02 -7.59380221e-01 -7.79616773e-01 4.45488244e-01
-1.46414626e+00 -1.00422955e+00 -6.82898760e-01 4.56286192e-01
1.42644659e-01 2.14384515e-02 7.84844100e-01 5.51011920e-01
-9.45356846e-01 7.17318416e-01 9.83315632e-02 3.64632040e-01
3.43835801e-01 -1.22805607e+00 9.74929631e-01 1.02839553e+00
1.90552205e-01 6.19463146e-01 4.88801330e-01 -5.25330961e-01
-1.83916903e+00 -1.41146195e+00 3.75415534e-01 8.23262110e-02
6.72437191e-01 -5.88328898e-01 -6.90305412e-01 8.82229686e-01
3.14207584e-01 4.38804179e-01 3.89187932e-01 2.41336882e-01
-1.21972198e-02 -3.80774170e-01 -9.48708832e-01 8.31132293e-01
1.42668056e+00 -3.03397238e-01 3.30965102e-01 5.58663905e-01
6.17843449e-01 -4.60840255e-01 -7.61271000e-01 2.99340844e-01
2.64630973e-01 -1.13818932e+00 8.82252574e-01 -3.34348351e-01
1.85190305e-01 -8.77226666e-02 1.23896629e-01 -1.30106115e+00
-3.89588237e-01 -6.08384490e-01 -4.63011414e-01 1.07893193e+00
1.71373263e-01 -8.35792661e-01 1.15899074e+00 3.52284759e-01
-1.45925105e-01 -1.09498429e+00 -5.91408789e-01 -9.44482267e-01
-3.37427497e-01 -5.36082327e-01 7.11110234e-01 8.66110861e-01
-1.93759426e-01 5.72684705e-01 -4.33105528e-01 -7.18419999e-02
5.70289612e-01 -8.23775306e-02 9.47771370e-01 -1.19211102e+00
-2.17209443e-01 -5.81867516e-01 -6.26679301e-01 -9.05267119e-01
-7.49701634e-02 -1.21184790e+00 -1.58559829e-01 -1.93322480e+00
3.19336057e-01 -4.16701764e-01 -4.33240920e-01 7.20816016e-01
-1.05489455e-01 1.97244033e-01 1.80581808e-02 3.22167240e-02
-9.71796453e-01 4.31672841e-01 8.47372651e-01 -1.57984942e-01
-1.43445119e-01 -1.77630827e-01 -6.84162736e-01 6.18800342e-01
1.01869166e+00 -5.81471443e-01 -8.21386278e-01 -5.22661746e-01
4.82136548e-01 -4.12101775e-01 2.39160433e-01 -1.49118686e+00
4.78593826e-01 -2.50940658e-02 1.96354702e-01 -4.21110004e-01
-1.37911364e-01 -8.29070628e-01 5.10008156e-01 6.18952990e-01
1.90722123e-01 4.13031012e-01 1.46014616e-01 7.06887662e-01
5.54715171e-02 -1.59857407e-01 7.58460462e-01 -2.71154165e-01
-7.76807606e-01 7.00321198e-01 -7.91949183e-02 -4.46266234e-02
1.06174016e+00 -4.04602796e-01 -4.76727784e-01 -4.52305257e-01
-1.24641292e-01 1.52625814e-01 5.98863900e-01 1.08413339e-01
3.99059057e-01 -1.23676455e+00 -3.35038006e-01 8.20691660e-02
-1.33133046e-02 1.95766389e-01 1.33163780e-01 6.97001398e-01
-9.84840512e-01 3.37238818e-01 -2.31424272e-01 -4.60353613e-01
-8.54256213e-01 7.77188897e-01 4.71704900e-01 -8.27392280e-01
-7.09097803e-01 8.86799157e-01 -7.38764331e-02 -5.96243322e-01
4.37408239e-01 -2.56580710e-01 1.67246699e-01 -3.98887843e-01
1.22809038e-01 6.16622150e-01 3.05587560e-01 -1.09695764e-02
-3.58884841e-01 1.29546091e-01 4.00100015e-02 7.26630270e-01
1.66068947e+00 9.56054479e-02 -3.02596837e-01 1.36378974e-01
9.38752234e-01 -3.56975079e-01 -8.74279976e-01 -1.04994550e-01
-8.44227597e-02 3.74275781e-02 1.23612747e-01 -5.26588023e-01
-1.80624306e+00 8.67393732e-01 4.14027512e-01 5.90844274e-01
1.41264093e+00 -2.28848338e-01 9.35948849e-01 4.41141754e-01
5.38594007e-01 -1.15499306e+00 2.65397560e-02 5.08679926e-01
3.95690531e-01 -9.18656111e-01 3.82721275e-01 -4.73026603e-01
3.36610107e-03 1.25562048e+00 7.64870524e-01 -2.99476296e-01
7.66130626e-01 1.81817561e-01 -3.17614466e-01 -4.74609226e-01
-7.17242241e-01 -3.58144119e-02 3.24375629e-02 6.21646225e-01
4.84900445e-01 1.42708898e-01 -9.63186845e-02 3.27676535e-01
-2.64287770e-01 -6.79653138e-02 4.35210377e-01 8.76434207e-01
-1.79184645e-01 -1.05964744e+00 2.30202094e-01 8.56827259e-01
-3.00062418e-01 -2.61564314e-01 -2.35070363e-01 1.01439929e+00
-2.10386291e-01 8.04493725e-01 -2.37768307e-01 -6.73368275e-01
1.96128428e-01 -2.23892465e-01 4.75309044e-01 -6.66484356e-01
-6.05678618e-01 -2.70935506e-01 2.77897358e-01 -6.85828567e-01
-2.34659269e-01 2.12525055e-02 -1.15657198e+00 -9.76779222e-01
-6.37288749e-01 3.61722745e-02 8.00156951e-01 5.64528942e-01
3.88300180e-01 9.36617911e-01 1.79083064e-01 -1.11215496e+00
-4.75054801e-01 -9.87911224e-01 -6.81143820e-01 2.04613637e-02
-2.48711288e-01 -2.74605513e-01 -2.11517140e-01 -5.18214703e-01] | [7.010208606719971, 5.908237934112549] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.