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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
da7283e1-382b-4bf7-bdba-4fe696bb6adf | active-perception-using-light-curtains-for | 2008.02191 | null | https://arxiv.org/abs/2008.02191v1 | https://arxiv.org/pdf/2008.02191v1.pdf | Active Perception using Light Curtains for Autonomous Driving | Most real-world 3D sensors such as LiDARs perform fixed scans of the entire environment, while being decoupled from the recognition system that processes the sensor data. In this work, we propose a method for 3D object recognition using light curtains, a resource-efficient controllable sensor that measures depth at user-specified locations in the environment. Crucially, we propose using prediction uncertainty of a deep learning based 3D point cloud detector to guide active perception. Given a neural network's uncertainty, we derive an optimization objective to place light curtains using the principle of maximizing information gain. Then, we develop a novel and efficient optimization algorithm to maximize this objective by encoding the physical constraints of the device into a constraint graph and optimizing with dynamic programming. We show how a 3D detector can be trained to detect objects in a scene by sequentially placing uncertainty-guided light curtains to successively improve detection accuracy. Code and details can be found on the project webpage: http://siddancha.github.io/projects/active-perception-light-curtains. | ['Srinivasa G. Narasimhan', 'Siddharth Ancha', 'David Held', 'Yaadhav Raaj', 'Peiyun Hu'] | 2020-08-05 | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/4458_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123500732.pdf | eccv-2020-8 | ['3d-object-recognition'] | ['computer-vision'] | [ 2.89959937e-01 3.25506747e-01 2.32028607e-02 -5.20267785e-01
-4.97465372e-01 -5.48252523e-01 1.24013387e-01 -9.52188820e-02
-4.57943976e-01 6.29218966e-02 -4.33324307e-01 -2.75155365e-01
-6.09022677e-02 -6.23234987e-01 -1.05150592e+00 -5.69805503e-01
2.19105244e-01 5.11054218e-01 2.52609372e-01 5.28631330e-01
4.42878962e-01 9.92744923e-01 -1.79568470e+00 -3.02562267e-01
3.61843079e-01 1.33470011e+00 7.88358808e-01 8.16760063e-01
2.13082060e-01 2.26304948e-01 -1.51225701e-01 4.25655484e-01
5.40467858e-01 3.37587476e-01 1.35971904e-01 1.45404235e-01
3.76524091e-01 -5.38878262e-01 -1.14350013e-01 1.11142230e+00
4.71843213e-01 9.37645808e-02 6.02778256e-01 -1.27863801e+00
-3.37749153e-01 -3.82540636e-02 -4.40151542e-01 -7.29730129e-02
3.66826385e-01 2.35850468e-01 7.61616528e-01 -1.08201957e+00
2.57999599e-01 1.13268936e+00 2.51360923e-01 7.06723988e-01
-1.28511584e+00 -4.78257090e-01 3.22707295e-01 1.97766144e-02
-1.34478128e+00 -6.89631045e-01 1.07787120e+00 -5.71844220e-01
1.19316554e+00 2.38462642e-01 6.40939116e-01 8.09283435e-01
1.11597642e-01 9.13603783e-01 7.84173667e-01 -4.14405704e-01
6.63063228e-01 1.98238462e-01 1.72770824e-02 8.85204732e-01
2.82581598e-01 4.24264431e-01 -8.37493777e-01 1.04704477e-01
8.88616264e-01 9.68575403e-02 -7.67622143e-02 -1.07117248e+00
-7.78500378e-01 4.40559804e-01 6.34773433e-01 -1.44328266e-01
-3.16873699e-01 3.86501014e-01 -4.21639234e-01 -6.02470106e-03
4.27248746e-01 4.05224591e-01 -5.58659554e-01 1.24520101e-01
-3.64431918e-01 6.31771758e-02 4.93653834e-01 1.10670376e+00
9.17121649e-01 -2.20753416e-01 3.27714533e-01 4.05238599e-01
9.58775997e-01 1.04078615e+00 -3.15277666e-01 -1.22883332e+00
2.51826048e-01 7.53085315e-01 4.24717456e-01 -7.27892041e-01
-5.23964107e-01 3.56321894e-02 -2.69650578e-01 9.46681142e-01
3.31643736e-03 -8.16236362e-02 -1.04612303e+00 1.33818126e+00
4.99372125e-01 2.54963070e-01 -2.67168880e-01 1.19167209e+00
5.54933667e-01 3.96189660e-01 -5.03953397e-01 -8.24998841e-02
7.93277383e-01 -3.73635620e-01 -1.99557930e-01 -5.80728650e-01
5.77765517e-02 -4.41038311e-01 1.10645235e+00 4.85515416e-01
-1.02650607e+00 -3.67456317e-01 -1.40199387e+00 -1.78002521e-01
-3.58731061e-01 -5.26998416e-02 3.25569391e-01 4.56732899e-01
-1.02680039e+00 1.84279874e-01 -1.32671392e+00 -9.38680023e-02
4.65717435e-01 6.40327096e-01 6.83646426e-02 1.12438314e-01
-2.16230899e-01 9.47290421e-01 1.62417412e-01 1.88026696e-01
-1.16459358e+00 -6.28537536e-01 -8.85066509e-01 -2.16186851e-01
7.41896749e-01 -6.28799200e-01 1.45389223e+00 3.64237167e-02
-1.59967375e+00 1.06368923e+00 -3.61709774e-01 -2.32879564e-01
2.78814316e-01 -6.17216110e-01 2.42449060e-01 -1.18622864e-02
-1.31589428e-01 6.40466988e-01 1.00044847e+00 -1.61274743e+00
-6.70733571e-01 -9.52148318e-01 1.60012901e-01 3.27149421e-01
1.17043935e-01 -4.15610194e-01 -5.35422206e-01 3.76015991e-01
6.46741688e-01 -1.02861655e+00 -5.15958071e-01 5.22828817e-01
-5.99546671e-01 -2.16476470e-01 8.53339434e-01 6.66246340e-02
5.05558670e-01 -2.14848471e+00 1.37400646e-02 3.08844715e-01
3.80534202e-01 -1.32241771e-01 3.03604696e-02 -1.30658135e-01
4.89869803e-01 -2.11305335e-01 -2.80191809e-01 -8.56962740e-01
2.01308236e-01 2.57275760e-01 -4.14002150e-01 5.28730929e-01
1.30242586e-01 7.47986257e-01 -8.91016006e-01 -1.17966935e-01
8.56479526e-01 5.54222107e-01 -4.44046080e-01 3.23197871e-01
-6.65963292e-01 5.02990782e-01 -7.88158655e-01 9.41888392e-01
8.34459007e-01 -7.01082274e-02 -2.23545521e-01 1.36282265e-01
-4.68163759e-01 4.17296469e-01 -1.26163125e+00 2.07490897e+00
-6.56367779e-01 6.87502623e-01 4.70208257e-01 -7.37328053e-01
1.00582337e+00 -1.98026508e-01 4.39587802e-01 -5.42792082e-01
2.61275828e-01 2.01583371e-01 -7.02503622e-01 -4.74338263e-01
3.29649001e-01 2.61080086e-01 -5.05679054e-03 2.73692399e-01
-4.69470888e-01 -7.53559709e-01 -5.23677468e-01 -2.10824639e-01
1.27502120e+00 2.91038781e-01 3.59782055e-02 8.65364075e-02
1.15384787e-01 -2.57275105e-01 3.22412848e-01 9.29582000e-01
-2.36359432e-01 5.20021260e-01 -2.46152550e-01 -3.01133871e-01
-7.91592956e-01 -1.54252529e+00 -2.77833134e-01 6.48272216e-01
4.76457626e-01 7.16177449e-02 -3.87807965e-01 -3.68852317e-01
4.37512606e-01 8.25963080e-01 -2.74110883e-01 4.50176336e-02
-3.27204883e-01 -1.57523453e-01 -2.49166161e-01 5.38779855e-01
2.51935005e-01 -7.42600977e-01 -1.51364481e+00 -5.63791022e-02
3.53525519e-01 -1.11152506e+00 -2.28281654e-02 6.95745170e-01
-7.97802448e-01 -9.61842716e-01 6.38622493e-02 -4.84441400e-01
8.82409453e-01 5.62654793e-01 8.92676532e-01 -1.42436042e-01
-4.84007150e-01 9.29233193e-01 9.48558375e-02 -1.01551986e+00
1.67717680e-01 -2.52322018e-01 4.12754744e-01 -4.44568604e-01
8.14020872e-01 -9.12545264e-01 -6.18424356e-01 2.09933907e-01
-3.14585775e-01 3.63137349e-02 3.85419965e-01 1.10746652e-01
9.84825492e-01 -1.60315409e-01 -3.05234015e-01 -2.91950911e-01
1.43521875e-01 -1.77154364e-03 -1.58111632e+00 -1.42583981e-01
-5.88228524e-01 1.53526098e-01 -8.34073778e-03 -2.76244968e-01
-6.71525002e-01 8.71042311e-01 1.80497006e-01 -8.70285094e-01
-3.73751640e-01 -3.06817610e-02 -4.16805595e-01 -2.45840982e-01
7.50723362e-01 1.70879569e-02 -4.24550176e-01 -5.28149247e-01
3.12703311e-01 8.19642425e-01 3.86224538e-01 -5.31919658e-01
9.83558595e-01 8.47118616e-01 2.99467176e-01 -7.80867338e-01
-9.09407854e-01 -5.51052451e-01 -7.79764831e-01 -5.75810432e-01
6.65120602e-01 -8.36077154e-01 -1.32024646e+00 2.52640545e-01
-1.30219090e+00 -5.38811028e-01 -4.75873411e-01 5.61399579e-01
-8.69707346e-01 -2.48800129e-01 1.14097111e-01 -1.42395043e+00
5.12341000e-02 -1.01770532e+00 1.60037529e+00 3.65310341e-01
1.49931684e-01 -5.64870775e-01 4.39461917e-02 2.08048925e-01
1.50036216e-02 2.48157948e-01 4.74738538e-01 8.29594806e-02
-1.48091400e+00 -1.48442179e-01 -1.47743538e-01 2.15399891e-01
8.43413696e-02 -1.61272958e-01 -1.51895058e+00 -6.07423633e-02
2.53121793e-01 -1.69079959e-01 7.47446120e-01 7.71677732e-01
1.18394184e+00 1.61327988e-01 -6.92547381e-01 6.63308322e-01
1.49718678e+00 4.57934856e-01 3.95303518e-02 1.46625489e-01
5.52893698e-01 3.57366830e-01 5.16488969e-01 5.45618474e-01
2.20365375e-01 5.60020745e-01 1.05868316e+00 3.89445901e-01
2.56374329e-01 -4.72185910e-01 1.85371801e-01 3.48944157e-01
6.29926287e-03 -1.47030145e-01 -9.08247411e-01 2.77144641e-01
-1.87490523e+00 -5.21959960e-01 2.37149000e-01 2.40831041e+00
3.13429177e-01 5.13709009e-01 -2.82096595e-01 7.01391250e-02
6.22392774e-01 2.40185708e-02 -1.28727353e+00 -2.19130918e-01
3.05281132e-01 -1.46955624e-01 8.30756903e-01 1.04764771e+00
-8.89012158e-01 7.13605344e-01 5.78780222e+00 1.17538549e-01
-1.05482745e+00 -6.32179156e-02 8.52370933e-02 -6.98548198e-01
-3.29060972e-01 -1.13176107e-01 -1.18790102e+00 1.47935376e-01
5.41686535e-01 3.91384996e-02 7.88048089e-01 1.03249371e+00
4.90788519e-01 -3.06077451e-01 -1.59879589e+00 1.20852280e+00
-1.18537448e-01 -1.04252338e+00 -5.36085963e-01 2.87130475e-01
2.47918725e-01 4.49679792e-01 3.04760665e-01 -2.21489087e-01
4.53291893e-01 -5.60856879e-01 9.40297961e-01 6.10190570e-01
5.97500205e-01 -3.47839355e-01 -1.10796228e-01 8.20410728e-01
-9.13730979e-01 -1.92294896e-01 -6.31575644e-01 -3.82795215e-01
1.58163205e-01 1.03202832e+00 -1.17968488e+00 -9.33578089e-02
7.98961759e-01 5.09510815e-01 -3.83759588e-02 1.04931653e+00
-6.39186621e-01 1.68168366e-01 -1.03010523e+00 -4.07945454e-01
-1.88783213e-01 -1.74743772e-01 1.05377269e+00 6.73152685e-01
4.74189311e-01 2.11667612e-01 1.77969158e-01 1.26146305e+00
-8.89014080e-03 -6.12887740e-01 -9.81197476e-01 2.87260145e-01
7.51040757e-01 9.23099935e-01 -5.04293323e-01 2.19625458e-01
-3.04453522e-01 6.87961519e-01 3.71168405e-01 2.60316730e-01
-5.21700859e-01 -9.32313949e-02 7.25309312e-01 1.55014232e-01
2.45488659e-01 -7.18849480e-01 -6.33912683e-01 -8.75808477e-01
5.26040435e-01 7.95497298e-02 -2.13288680e-01 -1.29345429e+00
-8.63200486e-01 1.30914822e-01 -4.25917059e-02 -1.16741681e+00
1.40274707e-02 -1.02315366e+00 -1.37409702e-01 6.82715714e-01
-1.59988964e+00 -8.30787420e-01 -3.54420274e-01 5.17169774e-01
4.64466214e-01 1.91234648e-01 6.78465605e-01 -1.87600665e-02
-3.66640627e-01 6.93769604e-02 -1.14248723e-01 -2.48411551e-01
1.71163529e-01 -1.26552045e+00 3.59372497e-01 9.42903697e-01
3.27479810e-01 3.55570465e-01 8.12806845e-01 -6.22350335e-01
-1.78228116e+00 -6.44656539e-01 5.42606235e-01 -8.40688467e-01
5.52115858e-01 -1.01948535e+00 -4.27701354e-01 8.11616778e-01
-2.40558416e-01 -1.23809399e-02 2.80441314e-01 1.61681175e-01
-1.78910479e-01 -2.79753327e-01 -1.28670037e+00 5.41933060e-01
1.54394388e+00 -6.05129421e-01 -5.34601986e-01 4.56582695e-01
9.32883322e-01 -9.27718341e-01 -2.15380579e-01 5.18066287e-01
6.11244917e-01 -8.25685382e-01 1.05339408e+00 -5.00896387e-02
-2.33914442e-02 -4.09105748e-01 -6.20318115e-01 -1.00375772e+00
-1.74568936e-01 -5.90754330e-01 -4.90166783e-01 6.35894775e-01
4.68720168e-01 -6.87364399e-01 1.30314744e+00 9.58686054e-01
-4.62913454e-01 -5.45185387e-01 -1.24019670e+00 -7.48817742e-01
-6.15379453e-01 -1.06262064e+00 4.45006281e-01 1.53833091e-01
-6.75968006e-02 2.65140980e-01 3.72224934e-02 1.03420365e+00
1.03830636e+00 2.03941748e-01 6.18896186e-01 -1.34080100e+00
-1.76470682e-01 -1.88363343e-01 -4.17765647e-01 -1.55881131e+00
3.31331268e-02 -5.70486605e-01 7.14617670e-01 -1.45053923e+00
-1.70985460e-01 -5.87959886e-01 -3.38737458e-01 3.22300464e-01
3.08152914e-01 -9.04480368e-02 2.35321075e-01 1.66226357e-01
-5.64990282e-01 3.30579609e-01 1.06034434e+00 -2.04018757e-01
-4.94571775e-01 4.64548647e-01 -4.07375574e-01 9.14796829e-01
7.36350834e-01 -4.77890402e-01 -5.53220034e-01 -9.87555683e-01
4.38520521e-01 -1.24914579e-01 6.78211629e-01 -1.20308542e+00
5.31215906e-01 -3.30193222e-01 4.20725018e-01 -9.76203978e-01
1.09291923e+00 -1.32205665e+00 -1.48367286e-01 4.28740084e-01
-2.89064914e-01 -3.28350812e-01 3.22536707e-01 5.18099487e-01
4.70792890e-01 -3.21394980e-01 6.91023529e-01 -1.62461534e-01
-6.95248961e-01 4.77181375e-01 -4.49482560e-01 -4.23839748e-01
8.73542845e-01 -3.09742838e-01 -1.49025157e-01 -4.67368253e-02
-7.48522639e-01 3.02766919e-01 7.27592230e-01 4.25794154e-01
1.02899098e+00 -1.00347686e+00 -9.09077600e-02 4.79555309e-01
1.91029698e-01 7.27642119e-01 -1.79818138e-01 2.00752705e-01
-2.10817426e-01 6.24542773e-01 1.93319499e-01 -1.17957830e+00
-1.13005626e+00 4.89581674e-01 5.20314455e-01 3.56285572e-01
-5.01394093e-01 1.15187883e+00 6.17855899e-02 -5.12120664e-01
5.49309492e-01 -8.03822637e-01 4.31108713e-01 -5.09402156e-01
3.82756233e-01 1.57749727e-01 4.21572290e-02 4.18198816e-02
-5.35765529e-01 8.70423019e-01 3.97720903e-01 -3.02476794e-01
1.35911000e+00 -4.01961416e-01 3.06456685e-01 7.87558198e-01
9.25459445e-01 -1.21363357e-01 -1.87727487e+00 -2.44629428e-01
-1.13694996e-01 -7.74397910e-01 5.42995870e-01 -6.37148499e-01
-7.15263724e-01 9.59814370e-01 9.88829017e-01 2.03828961e-01
1.04728425e+00 2.99697876e-01 2.04665229e-01 8.02211523e-01
7.63102472e-01 -1.00640166e+00 7.32588917e-02 3.96791965e-01
8.04539859e-01 -1.38958716e+00 -6.56134933e-02 -4.07240957e-01
3.60607468e-02 6.85895622e-01 6.01898730e-01 -2.04437181e-01
8.28763247e-01 7.39675879e-01 -2.06960961e-02 -3.74307096e-01
-6.21175945e-01 -2.28879258e-01 4.46552858e-02 9.06786621e-01
-3.74912769e-01 2.86996663e-01 7.05609620e-01 1.28404960e-01
-6.94650263e-02 1.29432023e-01 1.45664424e-01 1.07835615e+00
-1.04039073e+00 -6.37135983e-01 -4.00813043e-01 2.38989815e-01
2.88339138e-01 9.20316651e-02 -3.50017726e-01 3.12067389e-01
2.22321153e-01 9.24827278e-01 2.68516630e-01 -4.28109527e-01
6.12153888e-01 -1.18093625e-01 9.20793653e-01 -9.74543452e-01
3.72407079e-01 4.12860652e-03 -3.10839683e-01 -9.48090732e-01
-3.19254905e-01 -9.28198397e-01 -1.30398178e+00 1.70319542e-01
-3.18040341e-01 -3.74039859e-01 1.32786238e+00 8.40999544e-01
4.93146360e-01 1.15702517e-01 7.85124362e-01 -1.27231514e+00
-3.29742551e-01 -4.48009551e-01 -4.88325715e-01 -4.06555653e-01
7.14108944e-01 -8.66092324e-01 -6.45901680e-01 -2.03246757e-01] | [7.685495853424072, -2.716139316558838] |
087e96e1-1c52-4e43-aaa2-21331a2d4822 | prioritycut-occlusion-guided-regularization | 2103.11600 | null | https://arxiv.org/abs/2103.11600v1 | https://arxiv.org/pdf/2103.11600v1.pdf | PriorityCut: Occlusion-guided Regularization for Warp-based Image Animation | Image animation generates a video of a source image following the motion of a driving video. State-of-the-art self-supervised image animation approaches warp the source image according to the motion of the driving video and recover the warping artifacts by inpainting. These approaches mostly use vanilla convolution for inpainting, and vanilla convolution does not distinguish between valid and invalid pixels. As a result, visual artifacts are still noticeable after inpainting. CutMix is a state-of-the-art regularization strategy that cuts and mixes patches of images and is widely studied in different computer vision tasks. Among the remaining computer vision tasks, warp-based image animation is one of the fields that the effects of CutMix have yet to be studied. This paper first presents a preliminary study on the effects of CutMix on warp-based image animation. We observed in our study that CutMix helps improve only pixel values, but disturbs the spatial relationships between pixels. Based on such observation, we propose PriorityCut, a novel augmentation approach that uses the top-k percent occluded pixels of the foreground to regularize warp-based image animation. By leveraging the domain knowledge in warp-based image animation, PriorityCut significantly reduces the warping artifacts in state-of-the-art warp-based image animation models on diverse datasets. | ['Gyeongsu Chae', 'Wai Ting Cheung'] | 2021-03-22 | null | null | null | null | ['image-animation'] | ['computer-vision'] | [ 3.54689181e-01 -9.20135304e-02 -3.00807953e-01 -1.20742777e-02
-4.22980368e-01 -3.67707551e-01 5.99002004e-01 -4.11626101e-01
-3.79329860e-01 6.76468372e-01 5.76355644e-02 5.95086589e-02
2.93726444e-01 -6.27325833e-01 -1.06674707e+00 -9.59672928e-01
2.55853441e-02 1.11415558e-01 4.62503046e-01 -2.79854447e-01
3.02824169e-01 4.18688506e-01 -1.51646042e+00 6.06753469e-01
8.60307276e-01 4.17413980e-01 1.22659594e-01 6.99895799e-01
-3.75757277e-01 8.63124490e-01 -7.70090401e-01 -8.12278092e-02
4.37153608e-01 -1.00962675e+00 -6.42235577e-01 3.87376755e-01
7.79675782e-01 -3.56840283e-01 -7.25627482e-01 1.10615444e+00
1.12143360e-01 2.51306295e-01 4.80308890e-01 -1.43580341e+00
-5.16307294e-01 4.55301225e-01 -1.17644930e+00 3.85672331e-01
8.12038183e-02 4.92253721e-01 5.00619769e-01 -7.54266262e-01
1.02767253e+00 1.28803813e+00 4.02412623e-01 5.96099079e-01
-1.63783085e+00 -9.08089876e-01 -3.32113765e-02 3.04324746e-01
-1.16273439e+00 -2.77922079e-02 1.36802638e+00 -4.17951941e-01
4.65701938e-01 3.93757582e-01 9.41928923e-01 1.26603293e+00
4.75263894e-01 6.84251249e-01 1.25138712e+00 -4.81570482e-01
1.26963496e-01 3.26696336e-02 -4.57679629e-02 7.32182682e-01
-4.67850687e-03 3.63031507e-01 -6.96753085e-01 -3.80147994e-02
9.86133099e-01 -1.57523736e-01 -5.17784774e-01 -9.02017802e-02
-1.21155560e+00 8.21392834e-01 3.12846303e-01 3.03747356e-01
-2.88871229e-01 3.27009857e-01 2.88346559e-01 1.77593797e-01
7.46281624e-01 4.03328121e-01 -1.50196761e-01 6.80502126e-05
-1.70320046e+00 2.86690682e-01 3.35556448e-01 5.38604498e-01
9.48734224e-01 5.11959910e-01 -4.22066189e-02 5.56907594e-01
1.80254467e-02 4.32424992e-01 4.08054978e-01 -1.28903675e+00
3.02335173e-01 2.81153053e-01 -7.53498450e-02 -1.12930620e+00
-6.69108480e-02 6.72682971e-02 -7.49125004e-01 9.24530685e-01
6.00276887e-01 -8.92860517e-02 -1.11076522e+00 1.56018376e+00
2.27061272e-01 7.33439624e-01 -2.24516355e-02 1.09437251e+00
7.38904774e-01 1.16192901e+00 1.42021269e-01 -5.68432868e-01
1.18132091e+00 -1.19130361e+00 -1.00678587e+00 -3.65698099e-01
1.86206385e-01 -1.01184833e+00 1.22344601e+00 5.52087963e-01
-1.16363096e+00 -7.10347414e-01 -1.18027282e+00 -2.41317060e-02
5.03309481e-02 -2.06216380e-01 3.50746423e-01 3.31402808e-01
-6.93695843e-01 9.36317742e-01 -9.40232813e-01 -1.41440988e-01
3.50970179e-01 -4.14931029e-02 -5.63919783e-01 -4.50559966e-02
-1.05108643e+00 8.94057512e-01 3.19741547e-01 -2.92438835e-01
-1.07755625e+00 -1.00070930e+00 -7.78722882e-01 -2.30854407e-01
3.18316102e-01 -2.96256930e-01 6.90495908e-01 -1.65509605e+00
-1.55859935e+00 8.27056706e-01 -8.57253894e-02 -5.18435955e-01
6.92694187e-01 -2.79378891e-01 -3.24085295e-01 2.60050207e-01
-1.20854758e-01 9.89619076e-01 1.57000005e+00 -1.54612625e+00
-3.25185895e-01 1.06117770e-01 -5.07952213e-01 1.51528763e-02
-5.71115799e-02 -1.28084747e-03 -7.10709751e-01 -1.10465050e+00
-2.97952980e-01 -1.09856224e+00 -1.79867879e-01 2.84772515e-01
-2.38346890e-01 2.53830999e-01 1.54496288e+00 -8.16155732e-01
1.10752058e+00 -2.30003667e+00 3.98201197e-01 -5.29187880e-02
1.33384089e-03 5.14947057e-01 -4.94869083e-01 3.10528994e-01
-3.95285547e-01 -8.42049345e-03 -6.73184454e-01 -3.15973103e-01
-7.74347126e-01 6.15104675e-01 -5.99689066e-01 6.49013877e-01
3.63327116e-01 5.82238257e-01 -7.89842427e-01 -8.07508469e-01
5.74092507e-01 8.53470862e-01 -5.13431847e-01 4.12263364e-01
-3.61180276e-01 6.42455697e-01 4.33728397e-02 3.16326261e-01
8.92593920e-01 1.54441252e-01 -9.53984186e-02 -4.36165422e-01
-2.19869450e-01 -4.75260258e-01 -1.02933037e+00 1.93585336e+00
-1.27476767e-01 1.23203313e+00 -4.95099761e-02 -8.48525405e-01
8.23154688e-01 4.42985594e-02 6.65062189e-01 -4.21427608e-01
2.06578165e-01 -1.76403206e-02 5.40942177e-02 -7.68007576e-01
5.76709211e-01 -4.18911539e-02 3.51431072e-01 2.93764740e-01
7.03559369e-02 -4.31009710e-01 3.50825161e-01 2.25309819e-01
8.32455516e-01 5.23483634e-01 5.14124259e-02 -1.84131056e-01
3.52560431e-01 3.44686180e-01 4.49394017e-01 3.18364114e-01
-5.39709702e-02 1.07726693e+00 4.42938507e-01 -5.95064044e-01
-1.25138688e+00 -6.83825970e-01 9.20438692e-02 8.31921935e-01
2.00156316e-01 -2.98168808e-01 -1.16740096e+00 -5.53774834e-01
-1.86687931e-01 7.79443860e-01 -8.75752032e-01 -1.19751833e-01
-1.06863642e+00 -7.37665057e-01 4.86486882e-01 2.77178258e-01
6.74325466e-01 -1.27030098e+00 -5.77974617e-01 9.18665305e-02
-3.01261723e-01 -1.18469822e+00 -7.72658348e-01 -7.95358345e-02
-8.99980187e-01 -1.23127234e+00 -9.66319263e-01 -7.28839397e-01
8.53485763e-01 5.11383533e-01 8.34047377e-01 9.31825042e-02
-3.36240977e-01 -3.02059669e-02 -3.45588624e-01 -2.63593551e-02
-7.96464980e-01 -3.59986991e-01 -2.72905856e-01 2.46428013e-01
-2.13542700e-01 -5.78267217e-01 -5.27746201e-01 2.98195422e-01
-1.23660636e+00 2.60963470e-01 3.96571398e-01 8.35110724e-01
6.18412614e-01 1.82387291e-03 -2.84819633e-01 -1.02111840e+00
3.89177948e-01 -2.36867309e-01 -5.80993891e-01 2.51531918e-02
-3.63644361e-01 3.18157315e-01 6.11373544e-01 -9.89983380e-01
-9.53597426e-01 2.82434553e-01 1.78132102e-01 -1.13744569e+00
3.36106047e-02 9.81380939e-02 1.68173566e-01 -2.68599540e-01
8.24318886e-01 3.22242707e-01 4.07763988e-01 -2.03574389e-01
5.80500901e-01 1.84921354e-01 8.68359506e-01 -2.49780297e-01
9.78361607e-01 7.98426867e-01 9.85301957e-02 -1.12077737e+00
-4.02924240e-01 -7.89336488e-02 -3.21722418e-01 -4.50086087e-01
1.10542536e+00 -7.18672454e-01 -1.94387332e-01 5.43196023e-01
-1.39228070e+00 -6.24981642e-01 -3.55805367e-01 3.46330732e-01
-4.38993007e-01 6.61489010e-01 -5.92913091e-01 -4.54167783e-01
-2.50449687e-01 -1.34557605e+00 6.46578491e-01 2.36507431e-01
-3.86540115e-01 -6.93521202e-01 3.37031662e-01 3.03007483e-01
2.54266351e-01 5.97379029e-01 6.77419722e-01 -2.51885019e-02
-5.29210329e-01 1.74040630e-01 -3.88171822e-02 4.36267108e-01
9.50341970e-02 5.22203147e-01 -6.85266733e-01 -1.74156830e-01
-4.28274460e-02 4.45302986e-02 1.08712459e+00 5.02633572e-01
1.14646804e+00 -2.99978167e-01 -5.32542951e-02 8.85798514e-01
1.37491834e+00 2.35258192e-01 1.07377875e+00 4.24043894e-01
8.38857353e-01 6.77175760e-01 7.70506918e-01 1.66020274e-01
-3.21791053e-01 7.62219727e-01 5.39258182e-01 -4.54542398e-01
-5.65408647e-01 -2.19183832e-01 5.80764890e-01 4.64037836e-01
-3.93148661e-01 -1.02516294e-01 -5.84212482e-01 4.95286971e-01
-1.79868937e+00 -1.07553554e+00 -3.84464979e-01 2.10330892e+00
1.07611227e+00 -4.27857004e-02 1.28406018e-01 1.46570221e-01
7.77559459e-01 6.12972081e-01 -3.75935137e-01 -2.99353004e-01
-2.21321255e-01 3.98790568e-01 7.06747115e-01 7.87464976e-01
-1.00700665e+00 1.14010549e+00 5.62493372e+00 9.02835011e-01
-1.50561273e+00 1.50136590e-01 6.18836641e-01 -1.68881193e-01
-1.20576851e-01 1.75166398e-01 -2.61785924e-01 7.40884423e-01
6.35206044e-01 2.55318940e-01 5.03835976e-01 7.29862392e-01
6.11419857e-01 -3.30365777e-01 -7.91419327e-01 1.00539315e+00
3.79044503e-01 -1.49191380e+00 2.14492887e-01 -2.61801630e-01
7.96737194e-01 -3.23515415e-01 1.00592531e-01 -2.57524222e-01
5.03566451e-02 -8.81719232e-01 7.20717192e-01 2.55721629e-01
8.07924151e-01 -9.01536644e-01 4.85944003e-01 -3.39735076e-02
-8.65697563e-01 2.67940253e-01 -2.43301302e-01 1.96722120e-01
4.03841108e-01 4.30882961e-01 -3.06400627e-01 9.19011831e-02
7.16201186e-01 6.00892425e-01 -3.68817896e-01 6.67269409e-01
-4.99197453e-01 8.36947858e-01 -2.01487280e-02 7.07013309e-01
2.94865459e-01 -4.24257427e-01 7.30195463e-01 1.39285731e+00
2.26472676e-01 2.28387639e-01 2.63354313e-02 8.52977872e-01
1.58922076e-02 -4.88736555e-02 -5.66039801e-01 7.36094415e-02
2.22750276e-01 1.09454584e+00 -8.23212028e-01 -4.64780301e-01
-3.49669725e-01 1.31381476e+00 -1.72546118e-01 5.29482961e-01
-1.14589298e+00 -2.25138634e-01 7.43841231e-01 4.73460406e-01
3.15818906e-01 -3.62075984e-01 -2.17603087e-01 -1.02149951e+00
-3.17991138e-01 -1.00417662e+00 1.63147479e-01 -8.38569582e-01
-7.87830770e-01 6.12316847e-01 2.31436357e-01 -1.41518188e+00
-4.19929959e-02 -2.73922890e-01 -9.03631449e-01 4.97240484e-01
-1.51992297e+00 -1.00109982e+00 -7.13165164e-01 8.15101266e-01
1.04049981e+00 -8.14605057e-02 3.73098522e-01 7.97120780e-02
-5.99313319e-01 2.29033381e-01 -2.80690901e-02 1.93351939e-01
1.05877221e+00 -7.53049910e-01 3.35708946e-01 1.12335455e+00
3.80106926e-01 9.80890021e-02 1.21803367e+00 -7.56086886e-01
-1.24130917e+00 -1.09776485e+00 3.13259810e-01 5.11736833e-02
5.25175512e-01 7.81893060e-02 -1.41090679e+00 4.91187185e-01
9.21795130e-01 5.38845003e-01 2.80904293e-01 -8.21651340e-01
-3.18363219e-01 -3.57359946e-02 -1.11172819e+00 6.09932601e-01
6.70993507e-01 -1.70373052e-01 -4.12144125e-01 1.36755347e-01
5.95053196e-01 -4.51863974e-01 -2.66450316e-01 -8.07293430e-02
2.60065615e-01 -9.67092574e-01 1.03314221e+00 -4.66984838e-01
9.15564477e-01 -6.08147144e-01 2.68813670e-01 -1.45772445e+00
-1.69876441e-01 -1.04896009e+00 -2.32001960e-01 1.12251604e+00
5.53323962e-02 -7.55578205e-02 9.38265920e-01 3.78904194e-01
1.34929523e-01 -2.23693669e-01 -7.51773298e-01 -6.14304006e-01
7.19202310e-02 -2.26875082e-01 -1.00663938e-01 1.15620291e+00
-2.96000779e-01 -6.57659024e-02 -1.03364027e+00 1.78721510e-02
8.07570577e-01 -1.37365565e-01 9.42984819e-01 -7.02379823e-01
-3.42338979e-01 -1.95141733e-01 -3.34654003e-01 -4.81175065e-01
2.90368646e-01 -4.99181926e-01 1.06214851e-01 -1.07460105e+00
7.24348193e-03 1.30387709e-01 1.98833361e-01 5.22023916e-01
-3.08088988e-01 5.60780525e-01 5.56347311e-01 4.70823705e-01
2.11641744e-01 3.64180803e-01 1.49892986e+00 -3.71976912e-01
-4.73941952e-01 -4.72323358e-01 -1.60416916e-01 8.10986996e-01
7.78996885e-01 -7.78429091e-01 -4.35212344e-01 -3.13716561e-01
-3.42862755e-01 7.52145797e-02 3.19320709e-01 -9.20746922e-01
-1.28918691e-02 -6.27582848e-01 3.32193375e-01 -2.43534654e-01
4.20494437e-01 -8.88847709e-01 4.73311871e-01 7.10563302e-01
-2.37881303e-01 1.85781404e-01 4.24521148e-01 4.39787596e-01
-3.54814291e-01 -1.82913408e-01 1.32419825e+00 -2.46924222e-01
-8.62035215e-01 1.06934151e-02 -5.26376903e-01 7.41811320e-02
1.31877077e+00 -1.55338049e-01 -2.79308885e-01 -2.68026650e-01
-3.81255507e-01 -4.46477473e-01 6.01198256e-01 3.19580317e-01
7.88482189e-01 -1.05655050e+00 -8.53385329e-01 2.66733497e-01
-4.45172638e-01 -5.46722487e-02 2.61117995e-01 7.24043787e-01
-1.07758665e+00 -3.12905252e-01 -6.78723514e-01 -5.54007769e-01
-1.73728883e+00 7.88930595e-01 -4.27336246e-02 -1.20542586e-01
-9.66932952e-01 7.06960022e-01 2.68331677e-01 5.18123150e-01
7.25683048e-02 -2.27095768e-01 1.82979908e-02 1.10190198e-01
6.54908001e-01 3.73687923e-01 -3.13534796e-01 -6.79425538e-01
-2.16221631e-01 8.33356500e-01 -2.18074277e-01 -3.13476026e-01
1.24511695e+00 1.64492384e-01 1.91766508e-02 1.68345958e-01
1.21678233e+00 1.95971373e-02 -1.83937776e+00 7.04541132e-02
-4.56069201e-01 -8.22795451e-01 2.56213039e-01 -4.12920028e-01
-1.69107795e+00 8.49703789e-01 5.56077540e-01 -2.43280977e-01
1.29630315e+00 -2.37259075e-01 9.66932833e-01 -2.74692357e-01
-6.06462173e-03 -1.05893219e+00 4.96886760e-01 4.89724018e-02
1.08274913e+00 -1.25393772e+00 3.61655891e-01 -5.82998574e-01
-1.01031494e+00 1.14721227e+00 6.06029391e-01 -5.96455574e-01
2.72141874e-01 6.08362079e-01 2.48174623e-01 1.45937786e-01
-3.93209279e-01 8.92854575e-03 1.32585764e-01 7.40351796e-01
1.21815734e-01 -2.51296222e-01 -5.32511950e-01 -5.00050262e-02
1.01558946e-01 -1.84714366e-02 8.19175839e-01 9.16138053e-01
-2.21259102e-01 -1.08015275e+00 -7.45960295e-01 -1.58799067e-01
-2.69127965e-01 2.26725116e-02 -4.18659300e-01 1.04890370e+00
1.64355412e-01 5.66687167e-01 1.18934862e-01 -4.32250649e-01
1.80893436e-01 -9.76110101e-02 5.26656508e-01 -1.58988371e-01
-7.17534006e-01 2.37092227e-01 -1.89114183e-01 -7.43640125e-01
-6.89639628e-01 -4.84167010e-01 -1.33700836e+00 -4.35768336e-01
-1.50772259e-01 -5.49653508e-02 6.52003884e-01 6.74580395e-01
2.23251924e-01 5.26135564e-01 4.90225077e-01 -1.26611149e+00
-2.54947226e-02 -7.58001208e-01 -4.48551863e-01 7.11791098e-01
4.58691001e-01 -6.07870162e-01 -5.78655899e-01 6.08130515e-01] | [10.910529136657715, -0.8914024233818054] |
10d0726a-5559-48d4-8f88-abf34394a6b6 | chemical-identification-and-indexing-in-1 | null | null | https://doi.org/10.1093/database/baac047 | https://doi.org/10.1093/database/baac047 | Chemical identification and indexing in PubMed full-text articles using deep learning and heuristics | The identification of chemicals in articles has attracted a large interest in the biomedical scientific community, given its importance in drug development research. Most of previous research have focused on PubMed abstracts, and further investigation using full-text documents is required because these contain additional valuable information that must be explored. The manual expert task of indexing Medical Subject Headings (MeSH) terms to these articles later helps researchers find the most relevant publications for their ongoing work. The BioCreative VII NLM-Chem track fostered the development of systems for chemical identification and indexing in PubMed full-text articles. Chemical identification consisted in identifying the chemical mentions and linking these to unique MeSH identifiers. This manuscript describes our participation system and the post-challenge improvements we made. We propose a three-stage pipeline that individually performs chemical mention detection, entity normalization and indexing. Regarding chemical identification, we adopted a deep-learning solution that utilizes the PubMedBERT contextualized embeddings followed by a multilayer perceptron and a conditional random field tagging layer. For the normalization approach, we use a sieve-based dictionary filtering followed by a deep-learning similarity search strategy. Finally, for the indexing we developed rules for identifying the more relevant MeSH codes for each article. During the challenge, our system obtained the best official results in the normalization and indexing tasks despite the lower performance in the chemical mention recognition task. In a post-contest phase we boosted our results by improving our named entity recognition model with additional techniques. The final system achieved 0.8731, 0.8275 and 0.4849 in the chemical identification, normalization and indexing tasks, respectively. The code to reproduce our experiments and run the pipeline is publicly available.
Database URL: https://github.com/bioinformatics-ua/biocreativeVII_track2 | ['Sérgio Matos', 'João R. Almeida', 'João F. Silva', 'Rui Antunes', 'Tiago Almeida'] | 2022-07-01 | null | null | null | database-the-journal-of-biological-databases | ['chemical-indexing'] | ['natural-language-processing'] | [ 2.07434237e-01 1.22262854e-02 -3.80854458e-01 -1.37814701e-01
-9.42538679e-01 -5.92782557e-01 6.13301694e-01 1.17023432e+00
-8.35090637e-01 8.76399159e-01 2.48245835e-01 -4.56822544e-01
-2.00857893e-01 -6.96330726e-01 -8.49651992e-01 -6.72008753e-01
5.99312186e-02 6.56717241e-01 -1.36635289e-01 3.19389522e-01
4.39580053e-01 7.78079867e-01 -1.27905405e+00 3.56094480e-01
8.15795839e-01 8.04857731e-01 3.72609496e-01 4.53271925e-01
-2.55858392e-01 4.80443984e-01 -5.94028354e-01 -3.76184195e-01
-2.34999090e-01 -1.35696024e-01 -8.59537065e-01 -8.26685488e-01
1.72569990e-01 3.72421145e-01 -1.17677905e-01 9.78309214e-01
9.60255384e-01 1.78682909e-03 7.44525969e-01 -8.41125131e-01
-6.03900075e-01 6.18733048e-01 -2.45880544e-01 4.36012745e-02
6.35798037e-01 -1.84905156e-01 9.82361376e-01 -1.11355162e+00
1.03171372e+00 9.34622824e-01 7.12071240e-01 4.78023469e-01
-1.19595301e+00 -8.60386968e-01 -2.81017900e-01 2.70274252e-01
-1.70253110e+00 -5.21928668e-01 1.34519905e-01 -7.32339442e-01
1.37591958e+00 2.39048675e-01 1.83793113e-01 1.16993392e+00
4.76012170e-01 1.57227695e-01 6.78955078e-01 -5.01099944e-01
3.74279708e-01 3.10953319e-01 1.98812485e-01 5.71301281e-01
5.38064241e-01 -1.87204003e-01 -2.95563906e-01 -4.08162713e-01
7.00957105e-02 1.09562121e-01 -1.12411730e-01 1.45364687e-01
-1.19705653e+00 7.34699607e-01 3.35996538e-01 7.37624228e-01
-7.20325410e-01 -2.83578455e-01 4.51406211e-01 -7.77918473e-02
4.73420113e-01 1.12636113e+00 -6.38923049e-01 2.02789962e-01
-1.03101635e+00 1.46594271e-01 1.02775156e+00 7.65957057e-01
4.42686647e-01 -7.85674691e-01 -3.13708097e-01 9.47411001e-01
3.92080039e-01 1.34665981e-01 6.32174671e-01 -5.32893896e-01
4.18491550e-02 8.55981350e-01 -1.30989522e-01 -1.07237530e+00
-8.22448909e-01 -4.36260790e-01 -8.40395689e-01 -2.50623614e-01
1.35548666e-01 -3.25817540e-02 -9.77178335e-01 1.42700708e+00
2.73263603e-01 9.76032689e-02 2.14728191e-01 5.21752238e-01
1.23196566e+00 4.54314679e-01 7.63055086e-01 -5.84164150e-02
1.83761728e+00 -4.82485026e-01 -9.03055370e-01 3.39437455e-01
8.90885353e-01 -1.13266945e+00 3.74888957e-01 1.38685495e-01
-7.43064165e-01 -3.51739645e-01 -1.15468109e+00 -3.01464856e-01
-1.27719724e+00 1.99363008e-01 6.03921473e-01 3.43477160e-01
-8.61185193e-01 8.15239489e-01 -8.00215900e-01 -5.36316276e-01
5.75105608e-01 6.86422765e-01 -6.98013961e-01 -1.58209756e-01
-1.54464328e+00 1.24661112e+00 5.48452497e-01 -1.33055970e-01
-5.00907958e-01 -1.12824452e+00 -8.53066921e-01 -7.40762278e-02
8.89947340e-02 -7.18953371e-01 8.19494724e-01 3.63918543e-02
-9.64990854e-01 1.10635841e+00 -3.82339865e-01 -3.68324637e-01
3.12603116e-02 6.60245791e-02 -6.71717107e-01 1.35693997e-01
5.62897801e-01 8.27949822e-01 -6.41506538e-02 -5.98920584e-01
-7.61059701e-01 -4.10643220e-01 -2.28176966e-01 2.51747705e-02
-4.43548620e-01 2.10758626e-01 -7.83181846e-01 -6.25941813e-01
-2.02220634e-01 -8.62077296e-01 -3.72583181e-01 -3.69893163e-01
-3.98574352e-01 -5.28559983e-01 7.71178585e-03 -7.47063398e-01
1.46151161e+00 -1.94684660e+00 7.85163268e-02 2.16135755e-01
2.58882970e-01 1.25316292e-01 -8.25838279e-03 6.07498527e-01
-5.98189712e-01 5.24564981e-01 -2.17881948e-02 -2.53799796e-01
-4.72865328e-02 -2.58624136e-01 7.01179057e-02 5.55717826e-01
2.87330151e-01 7.96382308e-01 -9.37882781e-01 -5.01092613e-01
2.62725130e-02 7.66802907e-01 -3.94319832e-01 1.51370257e-01
-1.68058470e-01 2.26717070e-01 -3.57994348e-01 8.36831033e-01
5.50208449e-01 -3.54434013e-01 1.12376027e-01 -5.70078731e-01
-4.39200789e-01 4.18101400e-01 -9.83454168e-01 1.81901133e+00
-2.95851767e-01 1.64733097e-01 -1.96152955e-01 -7.68841863e-01
6.70691252e-01 5.60522377e-01 8.97736013e-01 -6.04592860e-01
1.59891456e-01 5.01578450e-01 -9.61439833e-02 -4.96953100e-01
2.77533382e-01 9.15022790e-02 9.58151743e-02 -1.54038267e-02
3.00794750e-01 3.29110175e-01 5.11997104e-01 2.29322225e-01
1.27998209e+00 1.31311221e-02 4.11657393e-01 -3.56000572e-01
8.79700661e-01 2.80752242e-01 3.47075850e-01 5.39043725e-01
1.65425047e-01 4.13195848e-01 3.25657248e-01 -2.53867596e-01
-7.68491268e-01 -3.99079651e-01 -8.34594727e-01 8.61692369e-01
-3.41828287e-01 -6.28710568e-01 -6.48089409e-01 -4.81939256e-01
1.26760006e-01 5.70737422e-01 -7.33845711e-01 -2.19245806e-01
-1.65673599e-01 -1.08116257e+00 7.94216156e-01 1.90445408e-02
-6.63865879e-02 -1.07438684e+00 1.74870051e-03 4.55195338e-01
3.80699933e-02 -9.55456853e-01 -3.23903173e-01 7.09893644e-01
-3.55669171e-01 -1.41716552e+00 -1.05553162e+00 -9.20311928e-01
6.11508787e-01 -4.03163314e-01 6.78242624e-01 -1.44280687e-01
-6.70853615e-01 4.34177779e-02 -1.35658309e-01 -7.47761488e-01
-4.89854336e-01 4.51760203e-01 1.27755389e-01 -3.89204383e-01
7.95764685e-01 -1.95790932e-01 -7.33555853e-01 -3.11311125e-03
-8.84457409e-01 -2.51026481e-01 7.42163062e-01 7.10613549e-01
8.05649340e-01 -2.31245488e-01 5.07645965e-01 -8.30178916e-01
6.39307797e-01 -7.31147170e-01 -6.62068725e-01 1.53710097e-01
-1.00139618e+00 3.19347322e-01 4.11225259e-01 -2.94886082e-01
-5.56824327e-01 3.56967866e-01 -7.46424437e-01 7.56565928e-02
-4.63987052e-01 9.93254542e-01 -3.09303224e-01 -1.07634328e-01
6.42611921e-01 -1.87356800e-01 -2.38272250e-01 -6.88642800e-01
1.44915834e-01 9.72324789e-01 3.05381805e-01 -3.95472825e-01
2.60053843e-01 -7.73862079e-02 1.55065656e-01 -6.42163157e-01
-5.58813214e-01 -8.27415168e-01 -5.22793055e-01 3.83622020e-01
1.21926332e+00 -1.06381905e+00 -1.03373694e+00 1.89676389e-01
-1.41172957e+00 2.76047766e-01 2.79417913e-02 7.39932001e-01
1.02269076e-01 3.01950186e-01 -5.31239569e-01 -2.15216264e-01
-6.28760874e-01 -1.35599613e+00 9.27082121e-01 1.07934028e-01
-6.20322526e-01 -8.97353172e-01 4.99748677e-01 2.30565012e-01
2.08601177e-01 2.42656901e-01 1.17848790e+00 -1.34576988e+00
6.48004711e-02 -4.20588791e-01 -1.52127728e-01 -6.36828244e-02
2.40195662e-01 -9.92680266e-02 -9.95603442e-01 6.22423775e-02
-4.81623560e-01 1.81576416e-01 8.58506203e-01 3.81389886e-01
1.08789861e+00 -2.85245609e-02 -8.15212429e-01 6.61618531e-01
1.25341260e+00 6.89845741e-01 4.48572457e-01 6.64850473e-01
8.05154383e-01 5.16319394e-01 3.19487840e-01 2.55229086e-01
1.78332195e-01 6.65013552e-01 1.38980284e-01 -3.64966154e-01
2.73793843e-02 -1.89191382e-02 -5.44706471e-02 3.28433484e-01
7.22063780e-02 -1.93716958e-01 -1.14763141e+00 5.93553603e-01
-1.45795918e+00 -6.57092631e-01 -2.30085596e-01 2.07126665e+00
1.39933729e+00 -9.70086604e-02 -2.17639282e-01 -1.11911513e-01
6.87373996e-01 -4.35811847e-01 -5.15891910e-01 -4.05103356e-01
-1.00547619e-01 6.77881062e-01 7.86782384e-01 4.54640388e-01
-1.30447042e+00 8.55847299e-01 5.50839043e+00 8.60485315e-01
-1.06911659e+00 2.80672461e-02 6.02791071e-01 2.07063034e-01
1.43605635e-01 -2.80869812e-01 -1.22867072e+00 5.29735565e-01
1.38805377e+00 -2.57913023e-01 -1.37232859e-02 6.73987031e-01
1.67376250e-01 -1.57123618e-02 -1.22775686e+00 7.90944755e-01
-2.64308244e-01 -1.82194698e+00 1.10702969e-01 1.67334422e-01
2.99311578e-01 2.49041796e-01 -1.76569968e-01 1.52834356e-01
5.93264662e-02 -1.10129225e+00 1.79672807e-01 6.69751048e-01
9.87692595e-01 -6.54490650e-01 9.71884191e-01 -1.61428645e-01
-9.03379560e-01 2.40258366e-01 -2.79110640e-01 4.26964998e-01
-3.22160542e-01 9.46357548e-01 -1.10519743e+00 7.78850973e-01
5.50563633e-01 8.92094433e-01 -5.92479885e-01 1.23171341e+00
-1.42614888e-02 2.22907409e-01 -1.42031178e-01 -1.14760548e-01
-6.07609823e-02 1.63993567e-01 2.43811741e-01 1.73063219e+00
2.52499282e-01 4.16076966e-02 4.87911738e-02 8.18655849e-01
-4.33551401e-01 5.86699903e-01 -2.81233102e-01 -3.86557907e-01
5.50449550e-01 1.37930381e+00 -8.61881852e-01 -3.87916923e-01
-2.23568588e-01 7.56291032e-01 -4.69935611e-02 1.68748274e-01
-6.19628549e-01 -8.73316646e-01 7.14340746e-01 5.32651991e-02
1.09227777e-01 2.45521635e-01 -1.23896904e-01 -9.20051038e-01
-3.63652200e-01 -9.11208212e-01 5.88285446e-01 -3.17588627e-01
-1.20381582e+00 6.64552808e-01 -3.66345793e-02 -8.63165617e-01
6.25784919e-02 -7.96804667e-01 2.92095337e-02 1.21660686e+00
-1.30653250e+00 -8.42206895e-01 1.83904290e-01 2.22724497e-01
6.20517135e-02 -2.49947757e-01 1.45827413e+00 9.72492278e-01
-8.20982337e-01 6.96383178e-01 3.76069456e-01 9.99798328e-02
1.25956118e+00 -1.19093406e+00 2.32456133e-01 2.66793787e-01
-1.54074624e-01 1.26043892e+00 6.01551175e-01 -9.19071972e-01
-1.33356512e+00 -1.18407738e+00 1.58022940e+00 -4.65271622e-01
8.78203928e-01 -3.60345721e-01 -9.05845046e-01 1.93383843e-01
2.36039266e-01 -3.41086090e-01 1.25362325e+00 6.37490302e-02
-1.08229503e-01 2.32734412e-01 -1.03771043e+00 4.03029919e-01
6.81821942e-01 -3.91075701e-01 -4.32805270e-01 7.36078322e-01
7.25142360e-01 -3.67445976e-01 -1.52368283e+00 2.05321252e-01
5.52990615e-01 1.81940913e-01 1.07645679e+00 -6.05114102e-01
3.88591617e-01 -4.02667850e-01 1.17475666e-01 -9.34171140e-01
-5.64578891e-01 -2.14326963e-01 1.26394957e-01 1.21766376e+00
8.34674180e-01 -5.37163079e-01 5.15174568e-01 6.18772864e-01
-2.13974595e-01 -6.91418052e-01 -8.35526824e-01 -3.22775900e-01
6.19407818e-02 -1.87279552e-01 4.02510464e-01 1.15997624e+00
2.14989930e-01 4.55904126e-01 1.67543724e-01 1.47682756e-01
3.66822302e-01 -2.71909475e-01 8.46902058e-02 -1.48651934e+00
1.14996806e-01 -5.93547881e-01 -3.93294156e-01 -2.28684574e-01
2.27902651e-01 -1.46665144e+00 -1.51676819e-01 -1.71181476e+00
2.79556185e-01 -1.41984627e-01 -8.28041553e-01 9.13088143e-01
-2.24381208e-01 1.03487581e-01 -2.95955718e-01 2.31018737e-01
-5.74908495e-01 -7.38552958e-02 5.92778683e-01 -4.11988497e-01
-2.35500529e-01 -3.05545896e-01 -1.05363417e+00 3.53805900e-01
6.08125687e-01 -9.91899371e-01 2.63631225e-01 1.68867912e-02
3.57439041e-01 -4.32657391e-01 2.83131506e-02 -9.47881460e-01
5.57977021e-01 1.43910944e-01 7.04136908e-01 -6.19491696e-01
-5.79646528e-02 -8.14054966e-01 5.29948294e-01 7.35277653e-01
-5.26072383e-01 -7.01962486e-02 6.30392730e-01 2.45010629e-01
-1.15550362e-01 -3.66259426e-01 5.58283567e-01 -8.50456506e-02
-4.22905505e-01 2.48981103e-01 -6.59296334e-01 -4.29478049e-01
7.47536421e-01 6.14134520e-02 -1.46383688e-01 3.71073008e-01
-1.01722276e+00 2.42901474e-01 2.66615868e-01 4.20545489e-01
3.45000207e-01 -1.12614036e+00 -6.96427763e-01 -1.04168259e-01
3.58934492e-01 -2.85194069e-01 1.90377086e-02 9.75808680e-01
-4.89555120e-01 8.81345987e-01 -5.03631011e-02 -2.90343195e-01
-1.44461524e+00 7.54620910e-01 2.33213007e-01 -5.14408708e-01
-1.36002168e-01 7.71185458e-01 -7.76192546e-02 -4.05714095e-01
6.27881646e-01 -3.34958792e-01 -8.12846005e-01 5.32747746e-01
6.80549622e-01 3.09815496e-01 7.13679373e-01 -5.47017515e-01
-9.27062631e-01 4.70861048e-01 -4.53737646e-01 2.27155387e-01
1.54284942e+00 4.63519663e-01 -4.13829088e-01 8.62407982e-02
1.66702282e+00 2.62497342e-03 -5.34995757e-02 4.72364277e-02
5.02995133e-01 3.52130115e-01 2.04769611e-01 -1.27851641e+00
-6.31955922e-01 5.19270957e-01 8.51189017e-01 -2.02494740e-01
8.17303836e-01 -6.71986565e-02 3.44770700e-01 5.14448225e-01
-2.00288266e-01 -8.36130440e-01 -7.77017951e-01 6.72778845e-01
7.70045221e-01 -1.14731956e+00 3.99129927e-01 -1.89334601e-01
-1.15592759e-02 1.15664363e+00 4.90028970e-02 4.38084990e-01
7.58920252e-01 3.23151678e-01 -4.07603085e-02 -5.37970960e-01
-5.33512592e-01 -1.88075621e-02 6.80868626e-01 3.32829773e-01
1.06437910e+00 -1.32404834e-01 -8.45004678e-01 9.77367461e-01
2.14426905e-01 1.70000434e-01 2.79731955e-02 8.56928587e-01
-1.34463891e-01 -1.52987576e+00 -3.82159889e-01 4.16323334e-01
-1.14665198e+00 -6.04724765e-01 -6.88328564e-01 4.88461852e-01
3.21274638e-01 9.48311687e-01 -3.19869518e-01 -3.92235428e-01
6.01992548e-01 2.61810988e-01 4.95824628e-02 -8.02910745e-01
-9.88810122e-01 2.81594306e-01 1.10790528e-01 -4.51645494e-01
-4.97188956e-01 -5.67167103e-01 -1.53186476e+00 -9.43003222e-03
-2.92365968e-01 6.68483436e-01 1.06754136e+00 6.66713655e-01
9.12505329e-01 7.96317935e-01 1.61915928e-01 -6.34294689e-01
5.00109456e-02 -1.02997768e+00 -1.44788817e-01 1.96880355e-01
5.04829437e-02 -6.31867588e-01 -2.45093986e-01 2.20136732e-01] | [8.483121871948242, 8.751444816589355] |
92f9f5db-e03b-4c53-bc5c-5488c0dcad6e | infinite-wide-finite-depth-neural-networks | 2112.15577 | null | https://arxiv.org/abs/2112.15577v4 | https://arxiv.org/pdf/2112.15577v4.pdf | How Infinitely Wide Neural Networks Can Benefit from Multi-task Learning -- an Exact Macroscopic Characterization | In practice, multi-task learning (through learning features shared among tasks) is an essential property of deep neural networks (NNs). While infinite-width limits of NNs can provide good intuition for their generalization behavior, the well-known infinite-width limits of NNs in the literature (e.g., neural tangent kernels) assume specific settings in which wide ReLU-NNs behave like shallow Gaussian Processes with a fixed kernel. Consequently, in such settings, these NNs lose their ability to benefit from multi-task learning in the infinite-width limit. In contrast, we prove that optimizing wide ReLU neural networks with at least one hidden layer using L2-regularization on the parameters promotes multi-task learning due to representation-learning - also in the limiting regime where the network width tends to infinity. We present an exact quantitative characterization of this infinite width limit in an appropriate function space that neatly describes multi-task learning. | ['Hanna Wutte', 'Josef Teichmann', 'Jakob Heiss'] | 2021-12-31 | null | null | null | null | ['l2-regularization'] | ['methodology'] | [-1.32727213e-02 3.11332822e-01 -6.65206313e-02 -4.00476724e-01
-6.68296099e-01 -6.08542502e-01 5.23879409e-01 -1.35200039e-01
-6.41709328e-01 5.25782406e-01 -6.76646605e-02 -2.63709813e-01
-6.42863750e-01 -5.94552517e-01 -8.81571114e-01 -1.12307334e+00
-6.61294088e-02 2.85708427e-01 2.14247316e-01 -8.51276219e-02
-1.40270546e-01 4.13812459e-01 -1.14365053e+00 -2.15698462e-02
8.16930771e-01 1.00621212e+00 4.36955467e-02 8.46610785e-01
4.93713357e-02 4.52315956e-01 -2.98282981e-01 -4.31979626e-01
5.14111102e-01 2.31576920e-01 -7.15986431e-01 -2.62572497e-01
5.23491979e-01 1.72254235e-01 -3.85901064e-01 1.24739933e+00
3.19791645e-01 5.14647245e-01 1.27596128e+00 -1.23899889e+00
-1.06972337e+00 5.57207584e-01 -6.34756386e-01 1.97975665e-01
-6.60909295e-01 -3.61850150e-02 1.22744560e+00 -7.41647482e-01
-4.02109139e-03 1.40152168e+00 1.11527240e+00 6.56288803e-01
-1.61019647e+00 -5.49939156e-01 1.87540650e-01 -4.77630109e-01
-1.14109409e+00 -2.05347717e-01 5.58854401e-01 -6.32929683e-01
7.99423933e-01 -1.53941751e-01 1.18287019e-01 1.34026706e+00
5.97459078e-01 7.00920522e-01 9.30675209e-01 -4.20119882e-01
2.23342150e-01 2.36213684e-01 7.25649476e-01 6.43824697e-01
6.13335609e-01 -2.48330072e-01 -3.73294383e-01 -3.71638894e-01
1.03724754e+00 3.40458363e-01 2.88926437e-03 -5.59238791e-01
-9.50064123e-01 1.04889524e+00 2.02264041e-01 2.91339964e-01
-2.90645599e-01 6.23625278e-01 6.10205233e-01 5.59611678e-01
8.74565065e-01 2.81467885e-01 -6.15086019e-01 1.43487602e-01
-5.86038411e-01 1.33934975e-01 6.54413283e-01 9.04461861e-01
8.86224151e-01 1.48531333e-01 -5.49917407e-02 1.08947647e+00
1.11003079e-01 3.78385723e-01 5.56630135e-01 -8.19623947e-01
4.99702990e-01 1.83212519e-01 -6.36599511e-02 -1.63811028e-01
-7.71280110e-01 -6.96957409e-01 -1.24933839e+00 5.11053443e-01
9.83557343e-01 -5.83619773e-01 -4.80062068e-01 2.02794337e+00
-1.56841636e-01 -9.98068750e-02 3.09976757e-01 4.05248344e-01
-2.87239277e-03 3.71838480e-01 3.52751344e-01 -1.35534527e-02
1.11474276e+00 -8.47337723e-01 -3.38763565e-01 -2.63676137e-01
8.14848065e-01 -3.08222413e-01 1.35001111e+00 3.13947409e-01
-1.07510316e+00 -5.82482338e-01 -8.06855023e-01 -4.51417156e-02
-2.95528531e-01 -2.69427985e-01 7.37510860e-01 7.23183453e-01
-1.11208999e+00 9.23554122e-01 -7.29094386e-01 -2.49187902e-01
5.58911920e-01 2.62177289e-01 -3.90361935e-01 8.29616636e-02
-1.12838352e+00 7.59675145e-01 4.31584358e-01 5.05570471e-02
-8.46745193e-01 -1.04796875e+00 -5.98427474e-01 1.75504982e-01
2.77865142e-01 -8.65601540e-01 1.39131427e+00 -9.84021842e-01
-1.17085207e+00 5.40358603e-01 -8.89222249e-02 -5.82096577e-01
6.40051603e-01 -6.86059833e-01 -7.24422485e-02 -1.26317926e-02
-2.57844299e-01 3.49180013e-01 1.23664331e+00 -9.51131225e-01
-3.23330998e-01 -6.33980572e-01 1.39496118e-01 1.37253627e-01
-7.20919728e-01 -1.47959083e-01 3.00963908e-01 -5.85040092e-01
-1.87215120e-01 -9.31372523e-01 -3.95181477e-01 2.17148274e-01
-2.98816025e-01 -5.94264686e-01 8.09750438e-01 1.32366677e-03
8.81058216e-01 -2.29520226e+00 3.91274057e-02 1.09839195e-03
5.23552835e-01 5.18452451e-02 -1.90928131e-01 1.87277377e-01
-7.29825944e-02 3.07045966e-01 3.56942452e-02 -6.10818863e-01
2.55703956e-01 2.13148341e-01 -2.27095202e-01 7.27898657e-01
9.21502188e-02 9.55140352e-01 -6.52807891e-01 -2.44849846e-01
7.66378120e-02 6.79426372e-01 -3.44218850e-01 -1.74770474e-01
-7.95739889e-02 -5.21541871e-02 -6.13798499e-01 -2.76876241e-02
5.40730655e-01 -6.26062989e-01 -2.87514448e-01 -2.84205079e-02
3.38505507e-02 -4.61136252e-01 -1.01643384e+00 1.33566022e+00
-7.91409671e-01 8.48469317e-01 1.20830096e-01 -1.11231196e+00
6.27718329e-01 3.72548014e-01 2.99694657e-01 1.66281946e-02
-2.05443516e-01 2.55770206e-01 3.84124531e-03 -1.24871835e-01
2.83619493e-01 -4.31339741e-01 -7.22340792e-02 4.99752671e-01
3.66808951e-01 3.96097988e-01 -2.75159299e-01 2.69074682e-02
9.46386755e-01 -1.49392471e-01 2.52601027e-01 -7.12208569e-01
2.37468347e-01 -6.47630692e-01 2.39247426e-01 1.17869174e+00
-2.31506124e-01 3.73871177e-01 7.37622380e-01 -4.31779146e-01
-1.40643883e+00 -1.20064139e+00 -4.26467746e-01 1.67147303e+00
-4.00158793e-01 8.82681981e-02 -6.33275509e-01 -5.27880669e-01
2.58158624e-01 6.00673914e-01 -1.09409964e+00 -3.90961945e-01
-4.53319430e-01 -1.17778063e+00 8.64189208e-01 6.79274261e-01
3.91403049e-01 -8.45247090e-01 -5.61915576e-01 1.20008878e-01
5.04822075e-01 -1.03935134e+00 -5.31099081e-01 7.96959400e-01
-1.10726690e+00 -8.15627635e-01 -1.34748638e+00 -5.50777972e-01
2.84372121e-01 2.28003114e-01 7.84493268e-01 -6.88216388e-01
-1.13592364e-01 5.00631154e-01 2.85484582e-01 -4.44489300e-01
-3.31913501e-01 4.45820838e-01 2.07712978e-01 5.81598841e-02
1.41064867e-01 -8.03816080e-01 -4.34362382e-01 2.49229699e-01
-9.77614403e-01 -1.99972644e-01 7.84908414e-01 7.74129331e-01
3.70253921e-01 -5.98994642e-02 9.43159997e-01 -9.62669790e-01
7.74684966e-01 -5.93709230e-01 -5.86450219e-01 6.39015138e-01
-5.16513526e-01 6.78960502e-01 9.19515371e-01 -9.07661676e-01
-1.13774490e+00 -2.54661411e-01 2.27117598e-01 -6.75063014e-01
-1.24252990e-01 1.75916895e-01 1.39960885e-01 -2.43536174e-01
7.94173539e-01 1.43609688e-01 -1.68757707e-01 -5.79685807e-01
4.71984833e-01 3.85378569e-01 3.14552307e-01 -8.97462666e-01
5.78717113e-01 4.74595308e-01 4.99761313e-01 -1.03893745e+00
-1.38874412e+00 -3.50362390e-01 -6.61590278e-01 -3.19569148e-02
9.81864870e-01 -7.96661854e-01 -1.07333708e+00 5.97511351e-01
-1.11584961e+00 -6.20621026e-01 -3.36068332e-01 5.71434975e-01
-8.82411003e-01 1.44604146e-01 -8.25268745e-01 -1.25839710e+00
-3.46166283e-01 -8.93974364e-01 8.14458847e-01 2.71369964e-01
1.55577511e-01 -1.79545903e+00 -7.54541829e-02 -1.79564148e-01
6.61791861e-01 -3.31560075e-02 1.32830179e+00 -1.22722018e+00
9.00669023e-02 1.37125745e-01 -3.62138093e-01 6.84698343e-01
-1.83864295e-01 -2.21833527e-01 -1.34929788e+00 -4.47402634e-02
1.85105190e-01 -3.92224193e-01 1.20993376e+00 8.69273245e-01
1.34358203e+00 -1.08209617e-01 -2.24983573e-01 7.00190306e-01
1.65092766e+00 -3.15871060e-01 1.29659891e-01 1.55948326e-01
7.68775702e-01 4.62992132e-01 -3.28483671e-01 2.58517832e-01
-1.27221867e-01 3.40679646e-01 3.67188215e-01 1.82022646e-01
2.04238072e-01 2.62048859e-02 3.83541346e-01 4.69125569e-01
-3.48019719e-01 1.07202314e-01 -8.27855349e-01 2.01101422e-01
-1.96959412e+00 -8.58487129e-01 -1.18318219e-02 2.36655831e+00
7.66191483e-01 5.18127143e-01 2.08539546e-01 -3.10448140e-01
8.26996326e-01 2.59175330e-01 -8.98996234e-01 -4.11568642e-01
-4.18140531e-01 2.33171545e-02 8.97877514e-01 5.38961530e-01
-1.29088366e+00 7.17578232e-01 6.80333757e+00 1.13423204e+00
-7.82507658e-01 4.99431878e-01 5.51128328e-01 -1.35462031e-01
-1.23316482e-01 -4.33684438e-01 -1.18473756e+00 1.29408047e-01
1.20091605e+00 -3.17880154e-01 1.35797292e-01 1.10126579e+00
-8.72897729e-02 1.42140582e-01 -1.12566578e+00 9.66782272e-01
-3.06187004e-01 -1.30536139e+00 -7.70554394e-02 3.38662833e-01
8.19108486e-01 3.16095293e-01 4.68494147e-01 3.69194984e-01
5.96472204e-01 -9.90025401e-01 6.21008575e-01 6.75872803e-01
7.74640918e-01 -9.49447453e-01 3.37726712e-01 4.87592608e-01
-1.07122719e+00 -4.19342577e-01 -8.13281238e-01 1.82010889e-01
-4.76880707e-02 9.27328050e-01 -4.64970022e-02 2.30294794e-01
2.73019344e-01 5.68956375e-01 -6.58089876e-01 9.70585167e-01
4.45727438e-01 4.72585946e-01 -2.91770905e-01 5.79443872e-02
5.75923979e-01 -3.83737624e-01 4.17424709e-01 1.40374172e+00
2.31411964e-01 -4.56882268e-01 -1.16727553e-01 7.94096589e-01
-1.45019025e-01 -1.77168339e-01 -9.27502334e-01 3.26130502e-02
1.50380746e-01 1.20438445e+00 -5.06312490e-01 -8.01676735e-02
-7.75255978e-01 7.95007050e-01 7.84791470e-01 8.23037863e-01
-5.75514197e-01 -4.72490698e-01 1.03280532e+00 -8.71360600e-02
1.46041662e-01 -4.12215471e-01 -4.64396060e-01 -9.41301167e-01
-6.42327517e-02 -1.16183564e-01 2.70810008e-01 -4.76611853e-01
-1.82591939e+00 5.78770161e-01 8.25764239e-02 -6.68289602e-01
-8.06431249e-02 -1.12248909e+00 -6.06725693e-01 1.04145217e+00
-1.34543383e+00 -1.29193246e+00 3.70817363e-01 7.41519153e-01
4.71812844e-01 -2.43016392e-01 8.11203539e-01 -5.05278334e-02
-5.35177529e-01 7.64780760e-01 7.08773434e-01 7.69959763e-02
6.30654812e-01 -1.52690125e+00 3.14210832e-01 2.04940617e-01
1.54867411e-01 7.98702180e-01 6.74543142e-01 -1.68957561e-01
-1.22236300e+00 -1.18959498e+00 4.37818378e-01 -2.81522900e-01
1.24389875e+00 -5.75334311e-01 -1.20176458e+00 9.85123217e-01
-6.81737363e-02 5.21804690e-01 8.38171840e-01 6.44658089e-01
-7.69803762e-01 3.44147272e-02 -7.66739070e-01 5.31233788e-01
8.84784877e-01 -8.40231538e-01 -4.66056466e-01 4.39282447e-01
8.32160234e-01 2.87536055e-01 -1.03244770e+00 3.99453901e-02
7.75813997e-01 -8.00186694e-01 9.96092319e-01 -1.16834450e+00
3.03778768e-01 4.37892258e-01 -3.33260775e-01 -1.41838062e+00
-2.31605232e-01 -8.87231827e-01 -2.71822482e-01 7.45900095e-01
5.89773417e-01 -9.95750666e-01 4.64737177e-01 6.36933565e-01
-2.85911143e-01 -1.08087456e+00 -1.01765203e+00 -1.22624493e+00
9.61320519e-01 -6.10440195e-01 -1.50931329e-01 4.84477371e-01
-2.71035582e-01 3.36437672e-01 -2.12193757e-01 8.55011865e-02
1.07125318e+00 -3.98126334e-01 1.25735849e-01 -1.80359066e+00
-3.24291706e-01 -8.99632692e-01 -1.52204275e-01 -1.11152339e+00
6.21387422e-01 -8.76398146e-01 -3.64558667e-01 -1.13203037e+00
4.44445461e-01 -6.80019915e-01 -5.70624530e-01 2.15274140e-01
-1.20961651e-01 -5.02979383e-02 1.26206741e-01 3.46655488e-01
-8.12336862e-01 4.74734813e-01 1.21102905e+00 4.00469422e-01
-9.55562219e-02 5.85598171e-01 -7.53117919e-01 1.24480402e+00
8.41006279e-01 -5.20325482e-01 -4.10728693e-01 -3.26928467e-01
5.48941374e-01 -1.60108730e-01 6.93048775e-01 -9.84185994e-01
3.99567991e-01 -1.82876155e-01 4.61972862e-01 -1.65710866e-01
5.10608077e-01 -5.73165953e-01 -4.13250834e-01 8.71653780e-02
-9.00136650e-01 -8.29695687e-02 4.73555475e-02 8.43370736e-01
1.91429496e-01 -6.07476592e-01 1.11926126e+00 -1.48718297e-01
-2.24130884e-01 5.06787837e-01 -4.96586949e-01 5.43052912e-01
8.73472393e-01 -6.10951632e-02 -4.60000813e-01 -3.60915661e-01
-1.04381347e+00 3.89936450e-03 2.43141633e-02 2.14114636e-01
2.96242535e-01 -1.23206127e+00 -5.81784368e-01 -9.50790290e-03
-5.80291077e-02 2.45577823e-02 4.26964909e-01 7.67464757e-01
2.66608119e-01 3.98340136e-01 -1.30297393e-01 -3.88754219e-01
-8.23661387e-01 4.67972636e-01 6.32454693e-01 -3.88094336e-01
-7.03499079e-01 8.92276108e-01 8.94561589e-01 -2.08113685e-01
2.91034967e-01 -2.86619663e-01 8.18227306e-02 1.08427361e-01
5.23115158e-01 5.14578700e-01 -2.02623501e-01 -1.68908834e-01
1.87666729e-01 5.53354383e-01 -2.67228633e-01 -2.44334355e-01
1.26138210e+00 -1.15652755e-01 2.16748893e-01 1.19952083e+00
1.49421656e+00 -3.26947898e-01 -1.76927340e+00 -5.03212631e-01
2.24159315e-01 9.44313928e-02 -5.35258427e-02 -1.78731263e-01
-7.46668875e-01 1.36324036e+00 2.83355504e-01 4.53538418e-01
5.44056535e-01 4.80115205e-01 3.14548254e-01 7.66777277e-01
6.69732839e-02 -1.09125185e+00 3.23176354e-01 9.72965837e-01
7.43694603e-01 -1.10367906e+00 -4.31316584e-01 -1.66692454e-02
-6.46382332e-01 1.50488424e+00 5.19203126e-01 -4.30599660e-01
1.18599451e+00 3.34001601e-01 -3.76969725e-01 7.68260956e-02
-8.94782782e-01 5.91656677e-02 2.92298347e-01 7.85509825e-01
4.25328046e-01 -2.52669957e-02 5.04729271e-01 7.51229346e-01
4.94796373e-02 -3.13418835e-01 4.30209905e-01 3.36246610e-01
-7.76457429e-01 -6.49666369e-01 -5.43690398e-02 6.27655506e-01
-6.52492940e-01 -3.04305017e-01 3.91039729e-01 7.71090388e-01
-2.83425748e-01 5.57632327e-01 1.74413309e-01 1.45682529e-01
3.92494425e-02 5.32447517e-01 6.26915395e-01 -5.62595546e-01
-6.20250046e-01 -1.08774051e-01 -4.12985563e-01 -2.48899415e-01
-1.51806846e-01 -5.53110600e-01 -8.83458853e-01 -2.61171818e-01
-2.64389068e-01 -1.34269610e-01 4.83306646e-01 1.18664658e+00
-2.29009017e-02 4.19915110e-01 3.05698186e-01 -6.25331342e-01
-1.35412908e+00 -1.17212570e+00 -1.08943748e+00 9.56607610e-02
6.39226556e-01 -5.88275850e-01 -7.50719547e-01 -2.18741491e-01] | [7.732889175415039, 3.756253719329834] |
dbef8179-ecaf-44c8-bcc7-4e4eabfe87df | eagermot-3d-multi-object-tracking-via-sensor | 2104.14682 | null | https://arxiv.org/abs/2104.14682v1 | https://arxiv.org/pdf/2104.14682v1.pdf | EagerMOT: 3D Multi-Object Tracking via Sensor Fusion | Multi-object tracking (MOT) enables mobile robots to perform well-informed motion planning and navigation by localizing surrounding objects in 3D space and time. Existing methods rely on depth sensors (e.g., LiDAR) to detect and track targets in 3D space, but only up to a limited sensing range due to the sparsity of the signal. On the other hand, cameras provide a dense and rich visual signal that helps to localize even distant objects, but only in the image domain. In this paper, we propose EagerMOT, a simple tracking formulation that eagerly integrates all available object observations from both sensor modalities to obtain a well-informed interpretation of the scene dynamics. Using images, we can identify distant incoming objects, while depth estimates allow for precise trajectory localization as soon as objects are within the depth-sensing range. With EagerMOT, we achieve state-of-the-art results across several MOT tasks on the KITTI and NuScenes datasets. Our code is available at https://github.com/aleksandrkim61/EagerMOT. | ['Laura Leal-Taixé', 'Aljoša Ošep', 'Aleksandr Kim'] | 2021-04-29 | null | null | null | null | ['3d-multi-object-tracking', 'multi-object-tracking-and-segmentation'] | ['computer-vision', 'computer-vision'] | [-1.13661326e-01 -4.59952474e-01 -2.35784978e-01 -8.17148909e-02
-5.67607999e-01 -9.26873922e-01 5.53623199e-01 4.37092269e-03
-5.35199046e-01 5.27178705e-01 -1.83978707e-01 1.14757665e-01
-1.38260409e-01 -8.03854585e-01 -8.06699753e-01 -7.82386482e-01
1.56352874e-02 6.79167211e-01 7.65251100e-01 -3.86894271e-02
-5.41027933e-02 6.43508017e-01 -1.58572364e+00 -1.00905873e-01
6.09505117e-01 9.97327626e-01 8.49232793e-01 8.45634162e-01
1.37008101e-01 5.06279111e-01 2.75166016e-02 2.26690754e-01
2.31603682e-01 2.09316220e-02 -2.55987763e-01 -2.34140642e-02
6.14807010e-01 -4.56294417e-01 -6.08176112e-01 1.02279580e+00
2.41978779e-01 1.26667395e-01 4.94968742e-01 -1.30758476e+00
-2.20238015e-01 8.69503245e-02 -4.19342041e-01 2.16569275e-01
4.37913746e-01 2.70433724e-01 7.77761459e-01 -1.00034678e+00
7.24600971e-01 1.15191591e+00 6.32681727e-01 5.08495629e-01
-8.43935251e-01 -4.89674568e-01 3.08888316e-01 3.83574992e-01
-1.42340398e+00 -5.15635490e-01 5.03752053e-01 -4.73606050e-01
7.55804062e-01 1.27416566e-01 6.43164873e-01 9.89890754e-01
2.90136755e-01 7.12005019e-01 7.41680384e-01 -2.81017423e-02
1.09364457e-01 -1.93128973e-01 -4.11374820e-03 7.84861982e-01
5.34650743e-01 3.33428472e-01 -7.53207862e-01 2.08671674e-01
7.68108845e-01 4.63349402e-01 -3.14926833e-01 -7.71267772e-01
-1.63232219e+00 5.55409133e-01 9.21194792e-01 1.99084491e-01
-5.11581123e-01 5.26424944e-01 -8.01446363e-02 -1.13007687e-02
1.65126845e-01 -9.64597389e-02 -2.44372636e-01 -3.91779318e-02
-4.43762898e-01 1.42287418e-01 3.25604916e-01 1.28261447e+00
1.04585433e+00 -1.18396573e-01 6.90299794e-02 2.59921640e-01
6.54588699e-01 1.26229405e+00 1.45092696e-01 -1.21565318e+00
5.50415337e-01 4.36107457e-01 6.51264250e-01 -8.33168745e-01
-5.34575582e-01 -2.47588813e-01 -6.21434152e-01 5.14085710e-01
5.56365490e-01 -1.45092383e-01 -8.46035957e-01 1.42854345e+00
6.39162958e-01 3.44343305e-01 1.40588656e-01 1.17448580e+00
7.58721113e-01 5.85416257e-01 -3.43087465e-01 5.93580790e-02
1.21452475e+00 -9.50007796e-01 -5.27778149e-01 -8.78598928e-01
4.99853879e-01 -3.59944314e-01 5.50235331e-01 2.24964291e-01
-6.37744248e-01 -4.54723418e-01 -9.11965072e-01 -2.64434814e-02
-2.17021197e-01 1.58571437e-01 4.26010609e-01 2.57028550e-01
-1.01391685e+00 1.82046577e-01 -1.41370451e+00 -5.91711283e-01
3.81819546e-01 2.68507868e-01 -3.33756715e-01 -5.70893705e-01
-6.21108115e-01 9.63137746e-01 3.47137243e-01 6.00367934e-02
-1.08017528e+00 -3.71767074e-01 -9.46123362e-01 -2.98706651e-01
5.24207532e-01 -7.15959311e-01 1.21312284e+00 -2.91924298e-01
-1.18149066e+00 5.85898995e-01 -5.78946054e-01 -5.56210816e-01
5.60024917e-01 -4.84651148e-01 -5.54405563e-02 2.25859404e-01
2.89455980e-01 7.98009872e-01 3.50875348e-01 -1.31497967e+00
-1.06763768e+00 -6.75362945e-01 -1.71904243e-03 4.61342543e-01
3.35283205e-02 -4.20446843e-01 -6.69523239e-01 -3.91041823e-02
5.01116633e-01 -1.12071168e+00 -3.74050528e-01 5.18683851e-01
-2.26675555e-01 9.07547846e-02 1.00185108e+00 -1.19799204e-01
6.51760042e-01 -1.99732101e+00 2.52081186e-01 -2.10987672e-01
3.55758071e-01 -7.74769261e-02 -2.77392939e-02 3.51222217e-01
7.46595621e-01 -4.23972756e-01 4.80847210e-02 -7.28688478e-01
-1.76323459e-01 4.52215850e-01 -1.10068984e-01 8.44118118e-01
-2.77520627e-01 1.06272304e+00 -1.22638965e+00 -3.37235302e-01
6.70947373e-01 4.63862598e-01 -1.78989530e-01 -1.03909321e-01
-3.11073780e-01 9.74373400e-01 -7.62525976e-01 9.02002752e-01
6.74123049e-01 -4.11143899e-01 -2.21467286e-01 5.83310053e-02
-5.43678820e-01 1.53394297e-01 -1.16804445e+00 1.91002119e+00
-3.16042840e-01 9.03674304e-01 3.11040521e-01 -5.76646090e-01
6.92904115e-01 4.95759286e-02 6.35943472e-01 -6.81586504e-01
9.20621082e-02 4.15030837e-01 -4.37541902e-01 -2.55420655e-01
5.79647720e-01 1.80916443e-01 -1.25929430e-01 -3.94164659e-02
-3.89818102e-01 -3.10420655e-02 6.08702339e-02 2.91224495e-02
1.43277395e+00 8.36860463e-02 3.31930608e-01 1.56155229e-01
2.66771972e-01 5.90360522e-01 6.39223456e-01 9.43339109e-01
-2.36954927e-01 4.48747844e-01 -6.20656788e-01 -2.87518799e-01
-7.50164688e-01 -1.29520023e+00 2.98165400e-02 7.79492736e-01
9.55793440e-01 -3.89021821e-02 6.50931075e-02 -3.56853843e-01
3.56389701e-01 2.46349007e-01 -2.83041060e-01 1.11232229e-01
-6.17501259e-01 -2.40548730e-01 1.13965861e-01 5.02530217e-01
5.90714276e-01 -7.35263824e-01 -1.23796129e+00 2.58857697e-01
-3.22285175e-01 -1.42817676e+00 -2.11101070e-01 2.05784723e-01
-8.55830491e-01 -9.68358278e-01 -5.86606443e-01 -4.06216770e-01
4.53300923e-01 1.11550272e+00 7.87784159e-01 -1.12113848e-01
-5.90923354e-02 6.33621037e-01 -4.30202812e-01 -3.21341217e-01
-3.34907994e-02 -2.52971053e-01 3.04316938e-01 -1.97250262e-01
2.47014642e-01 -4.09252316e-01 -5.87737441e-01 5.36702514e-01
-1.56713590e-01 8.26414376e-02 5.49992919e-01 3.43674719e-01
7.62470424e-01 -2.11797014e-01 4.37589772e-02 -2.39979997e-01
-4.22085881e-01 -6.20872378e-01 -1.05658972e+00 -1.22361466e-01
-2.12908462e-01 -2.46907547e-01 2.00988531e-01 -4.72592443e-01
-7.07948089e-01 4.83978212e-01 2.24478617e-01 -7.32578933e-01
-2.82976687e-01 4.30313170e-01 -1.75074488e-02 -3.15531969e-01
5.75170338e-01 2.24401280e-01 -2.16109276e-01 -4.09423143e-01
3.66017997e-01 4.70379680e-01 7.30730534e-01 -1.62387565e-01
8.76445293e-01 1.20486534e+00 2.71330535e-01 -8.93204808e-01
-8.16856146e-01 -1.03740847e+00 -8.12848032e-01 -5.51029980e-01
8.41289103e-01 -1.39288878e+00 -7.06272662e-01 5.07928729e-01
-1.12689054e+00 -4.90593672e-01 -1.85259581e-02 8.59639645e-01
-5.34004033e-01 1.91914499e-01 -2.38131583e-01 -1.07924557e+00
5.97638115e-02 -7.61314750e-01 1.38168180e+00 2.85167426e-01
6.76781014e-02 -8.69904220e-01 1.24797441e-01 1.43060714e-01
1.29214421e-01 4.28199440e-01 -5.19756228e-02 2.65113004e-02
-1.49515784e+00 -1.92735821e-01 -1.69085547e-01 -5.30753732e-01
1.38156846e-01 -3.87429863e-01 -7.78311312e-01 -4.83019114e-01
-2.03829616e-01 9.84711293e-03 1.14481187e+00 6.73755765e-01
3.27254266e-01 -6.18607923e-02 -1.05957723e+00 6.96992934e-01
1.54734671e+00 2.90291756e-01 1.77949384e-01 5.02898335e-01
8.44962895e-01 2.78319776e-01 1.16819715e+00 4.79265749e-01
7.51391292e-01 1.01712215e+00 9.24124360e-01 2.89832979e-01
-2.34450400e-01 -1.32966802e-01 4.64710563e-01 2.78970122e-01
8.80683511e-02 -7.18282402e-01 -1.07382679e+00 8.56859922e-01
-2.15053868e+00 -8.53648424e-01 -5.28289378e-01 2.13611770e+00
1.75088093e-01 4.74865809e-02 -7.88486674e-02 -3.02498192e-01
6.87153399e-01 1.52775541e-01 -8.59838545e-01 7.39934027e-01
-1.29297063e-01 -5.91727138e-01 1.12233448e+00 7.31453061e-01
-1.07204175e+00 8.80216360e-01 5.02215147e+00 3.36366445e-01
-9.58601058e-01 2.90125877e-01 -3.68596643e-01 -3.81326556e-01
-7.78450370e-02 6.38885573e-02 -1.29484212e+00 3.51976156e-01
6.99671626e-01 -5.82328215e-02 2.50317246e-01 7.29178131e-01
2.58783430e-01 -5.66968024e-01 -1.02996922e+00 1.10039103e+00
-1.34915099e-01 -1.38453662e+00 -5.40279031e-01 3.01873475e-01
6.48354352e-01 9.22132730e-01 -4.07007895e-02 -1.71724945e-01
5.53334475e-01 -5.05779982e-01 1.07154298e+00 5.05268872e-01
5.39177775e-01 -3.15113872e-01 4.48088527e-01 8.83853674e-01
-1.67327178e+00 -3.31918687e-01 -3.85452420e-01 -2.65381187e-01
5.62154889e-01 6.15521908e-01 -8.77322197e-01 6.20809376e-01
9.48244810e-01 1.22443604e+00 -3.25921297e-01 1.46725798e+00
-2.39085078e-01 3.03357691e-02 -7.87466645e-01 -1.27392873e-01
2.69578218e-01 -1.12706959e-01 9.87650812e-01 7.12112248e-01
8.46662581e-01 2.20466882e-01 5.26589870e-01 6.98657095e-01
2.30457351e-01 -6.21617019e-01 -8.63997579e-01 3.10105652e-01
8.43139529e-01 1.04809463e+00 -7.03644872e-01 -1.82629809e-01
-4.60298270e-01 9.22292411e-01 1.78744510e-01 3.11279029e-01
-7.18174636e-01 6.18970916e-02 8.83284509e-01 6.67555556e-02
5.85685372e-01 -9.49773908e-01 -1.49992391e-01 -1.13130260e+00
1.70598999e-01 -6.84980080e-02 1.42294481e-01 -9.76477742e-01
-7.46373713e-01 2.15304092e-01 -1.35711029e-01 -1.67322969e+00
-2.22550899e-01 -5.86197734e-01 -2.46567041e-01 6.62447929e-01
-1.78796327e+00 -1.18165934e+00 -8.57037306e-01 5.32745421e-01
5.08124769e-01 2.40734830e-01 3.11073214e-01 2.43933037e-01
-1.96778685e-01 -2.94066459e-01 4.87020940e-01 -5.84113896e-02
5.41657269e-01 -9.88790631e-01 4.08728868e-01 1.09453762e+00
3.58821809e-01 2.45348006e-01 7.62108147e-01 -9.13871646e-01
-1.77780938e+00 -1.24535751e+00 6.04525864e-01 -8.65087509e-01
4.56285864e-01 -3.57012391e-01 -6.55579150e-01 8.64657521e-01
-3.22117716e-01 3.14140737e-01 -5.45431599e-02 -3.25810820e-01
6.59108162e-02 -7.91424979e-03 -8.95919859e-01 4.23079550e-01
1.39901888e+00 -3.11720431e-01 -2.33576506e-01 2.49788120e-01
6.43360794e-01 -9.05408144e-01 -4.90100473e-01 4.30166394e-01
6.49618208e-01 -8.95403504e-01 1.25962842e+00 1.88501149e-01
-2.33933091e-01 -8.29752326e-01 -4.65243429e-01 -1.06567442e+00
-2.52065867e-01 -2.11823195e-01 -4.36282724e-01 7.95108259e-01
1.29431814e-01 -6.58029020e-01 9.39569831e-01 1.71766728e-01
-3.02267253e-01 -1.22548133e-01 -1.25407088e+00 -1.10723889e+00
-5.49896896e-01 -5.69318295e-01 3.49440992e-01 4.44991559e-01
-3.83800626e-01 1.14045277e-01 -4.46882993e-01 9.37716901e-01
1.07064819e+00 4.31178987e-01 9.19290185e-01 -1.45859075e+00
5.44137731e-02 -1.45201996e-01 -5.39130449e-01 -1.77533984e+00
1.28690274e-02 -6.77689731e-01 5.23331046e-01 -1.94028807e+00
-3.80819105e-02 -6.50230169e-01 3.19618322e-02 3.28352451e-01
1.35435507e-01 3.27490240e-01 2.07894266e-01 4.59582508e-01
-1.07535017e+00 6.08389914e-01 1.12494075e+00 -1.21194951e-01
-1.25690177e-01 1.90894678e-01 -1.77491292e-01 6.58513069e-01
6.18331850e-01 -5.24860501e-01 -1.35745302e-01 -9.38746870e-01
8.02767575e-02 2.51121223e-01 8.60378027e-01 -1.39573622e+00
7.86899030e-01 -3.21932882e-01 4.07132119e-01 -1.26573110e+00
1.02088261e+00 -1.06957626e+00 3.30308229e-01 7.56768525e-01
2.32625127e-01 2.56067067e-02 2.11262986e-01 9.65634346e-01
8.65078047e-02 -6.99467063e-02 6.36569440e-01 -2.13810161e-01
-1.45174396e+00 5.49657047e-01 -3.08845311e-01 -2.77689874e-01
1.15171528e+00 -4.05279666e-01 -6.31313145e-01 -3.89510036e-01
-4.38530982e-01 7.18744576e-01 8.27959776e-01 4.84751374e-01
8.44503522e-01 -1.44697261e+00 -4.47924137e-01 8.11260119e-02
3.29007596e-01 3.67142946e-01 2.87476748e-01 1.04874635e+00
-3.42835605e-01 7.68273234e-01 -1.94024384e-01 -1.25443959e+00
-1.46052516e+00 4.12767977e-01 2.91472226e-01 2.50836968e-01
-8.26991796e-01 8.29337001e-01 2.02587992e-01 -3.70051950e-01
1.10346846e-01 -5.64544201e-01 -7.12526217e-02 -1.63838521e-01
5.61256647e-01 5.11351705e-01 -2.10674435e-01 -9.46651995e-01
-7.29083538e-01 1.13977432e+00 4.13786143e-01 -1.93450183e-01
1.11273086e+00 -8.49724233e-01 2.66285092e-01 6.07834458e-01
7.77926028e-01 -5.79401432e-03 -1.82926297e+00 -4.84536797e-01
-4.25358415e-02 -7.69485295e-01 1.82987809e-01 -4.18400854e-01
-6.21682882e-01 7.79760599e-01 7.36695111e-01 -4.48553897e-02
7.50589192e-01 4.26682711e-01 6.40577257e-01 6.33186877e-01
1.12778473e+00 -4.57930267e-01 9.74451751e-02 8.22463810e-01
5.22054553e-01 -1.48425102e+00 -5.64200915e-02 -4.47626412e-01
-4.31888998e-01 8.51894677e-01 5.95782578e-01 1.88437387e-01
2.38715887e-01 2.35765994e-01 4.22469638e-02 -1.61448568e-01
-6.16984606e-01 -7.34893501e-01 1.09102532e-01 8.80058110e-01
-3.93757194e-01 -1.64966695e-02 5.89160860e-01 -8.12573731e-03
1.80907011e-01 -1.70471430e-01 3.68406087e-01 1.09110463e+00
-1.00424993e+00 -7.44577110e-01 -6.86302006e-01 1.32311463e-01
1.15132004e-01 2.00285375e-01 -9.06780362e-02 7.41872013e-01
2.14458127e-02 1.13367999e+00 1.44832626e-01 -2.83585012e-01
3.29794616e-01 -5.46033442e-01 5.84864497e-01 -6.47326410e-01
2.29798585e-01 7.47888833e-02 -9.95149929e-03 -9.42632198e-01
-5.54495931e-01 -1.04678392e+00 -1.48478520e+00 -1.89591408e-01
-3.60712290e-01 -1.39274359e-01 7.61116743e-01 9.47946787e-01
4.21679854e-01 3.29201549e-01 3.46089125e-01 -1.37524474e+00
-5.39024696e-02 -6.94241285e-01 -4.13398027e-01 -2.13839039e-01
9.86685455e-01 -9.85800266e-01 -1.49070993e-01 -1.54043615e-01] | [6.960919380187988, -2.1672356128692627] |
7ec8e2be-bde9-473d-8258-0e64a3a0d962 | part-aligned-bilinear-representations-for | 1804.07094 | null | http://arxiv.org/abs/1804.07094v1 | http://arxiv.org/pdf/1804.07094v1.pdf | Part-Aligned Bilinear Representations for Person Re-identification | We propose a novel network that learns a part-aligned representation for
person re-identification. It handles the body part misalignment problem, that
is, body parts are misaligned across human detections due to pose/viewpoint
change and unreliable detection. Our model consists of a two-stream network
(one stream for appearance map extraction and the other one for body part map
extraction) and a bilinear-pooling layer that generates and spatially pools a
part-aligned map. Each local feature of the part-aligned map is obtained by a
bilinear mapping of the corresponding local appearance and body part
descriptors. Our new representation leads to a robust image matching
similarity, which is equivalent to an aggregation of the local similarities of
the corresponding body parts combined with the weighted appearance similarity.
This part-aligned representation reduces the part misalignment problem
significantly. Our approach is also advantageous over other pose-guided
representations (e.g., extracting representations over the bounding box of each
body part) by learning part descriptors optimal for person re-identification.
For training the network, our approach does not require any part annotation on
the person re-identification dataset. Instead, we simply initialize the part
sub-stream using a pre-trained sub-network of an existing pose estimation
network, and train the whole network to minimize the re-identification loss. We
validate the effectiveness of our approach by demonstrating its superiority
over the state-of-the-art methods on the standard benchmark datasets, including
Market-1501, CUHK03, CUHK01 and DukeMTMC, and standard video dataset MARS. | ['Kyoung Mu Lee', 'Yumin Suh', 'Jingdong Wang', 'Tao Mei', 'Siyu Tang'] | 2018-04-19 | part-aligned-bilinear-representations-for-1 | http://openaccess.thecvf.com/content_ECCV_2018/html/Yumin_Suh_Part-Aligned_Bilinear_Representations_ECCV_2018_paper.html | http://openaccess.thecvf.com/content_ECCV_2018/papers/Yumin_Suh_Part-Aligned_Bilinear_Representations_ECCV_2018_paper.pdf | eccv-2018-9 | ['2d-human-pose-estimation'] | ['computer-vision'] | [ 1.19239658e-01 1.36387218e-02 -1.58909755e-03 -3.93804312e-01
-5.86154282e-01 -4.64603901e-01 4.15776163e-01 9.46652442e-02
-6.96599185e-01 4.19979215e-01 1.28911570e-01 5.71437061e-01
1.78830549e-01 -5.74709296e-01 -8.34885776e-01 -5.59590578e-01
-2.21166629e-02 6.45526230e-01 5.99857494e-02 -2.38230646e-01
-1.59612939e-01 4.25801337e-01 -1.68048346e+00 -1.20435596e-01
5.24849772e-01 1.21505630e+00 -4.20955569e-01 4.24005926e-01
4.56395447e-01 -7.53289973e-03 -6.79901719e-01 -8.23827982e-01
7.20647573e-01 -2.51308322e-01 -7.90686607e-01 3.22205335e-01
1.15898168e+00 -4.04368877e-01 -4.29278493e-01 1.02040088e+00
8.30488205e-01 2.66340405e-01 4.52706814e-01 -1.28174615e+00
-2.48801067e-01 6.32498637e-02 -8.49309266e-01 -2.71324553e-02
5.76865017e-01 3.01531013e-02 6.08644247e-01 -8.93025279e-01
6.03492200e-01 1.40038550e+00 1.11949730e+00 7.34879971e-01
-1.26825404e+00 -8.62872124e-01 3.02646965e-01 4.38902639e-02
-1.69663572e+00 -4.54399168e-01 5.57084322e-01 -3.80058557e-01
6.85191929e-01 2.04587996e-01 9.27384377e-01 1.02159083e+00
-7.79731199e-02 7.57641137e-01 6.46177769e-01 -2.72903800e-01
-2.19800293e-01 -2.27547344e-02 2.00366572e-01 8.13263714e-01
5.77445149e-01 9.35229510e-02 -5.00125170e-01 -2.04975069e-01
8.40315819e-01 2.82651454e-01 -2.69720823e-01 -8.57603133e-01
-1.16240680e+00 3.48046422e-01 5.48184037e-01 -1.71400487e-01
-2.95710534e-01 1.99762821e-01 4.73135561e-01 1.14506416e-01
2.67141610e-01 1.27482042e-01 -2.65352428e-01 1.14646658e-01
-1.02547526e+00 6.81107759e-01 6.56209707e-01 1.07406700e+00
5.85131347e-01 -4.01319683e-01 -5.16445875e-01 9.35311794e-01
4.07188684e-01 6.83757186e-01 5.37397265e-01 -4.60579097e-01
6.63520277e-01 7.71987677e-01 2.73609430e-01 -9.68768358e-01
-5.53385794e-01 -4.60107297e-01 -1.02726102e+00 -7.08719417e-02
7.69795001e-01 -9.51028764e-02 -1.03378952e+00 1.97555840e+00
5.69332480e-01 -4.89150360e-02 -2.77312756e-01 1.15626991e+00
8.96085918e-01 -7.71320518e-03 1.93161845e-01 2.82160610e-01
1.73606288e+00 -1.06212759e+00 -2.53036439e-01 -3.91077131e-01
3.69546771e-01 -5.30702293e-01 3.53077710e-01 -4.07826491e-02
-1.04051864e+00 -9.63443458e-01 -1.17896092e+00 9.00491886e-03
-4.05287623e-01 5.91511130e-01 3.38259816e-01 7.07511544e-01
-9.24907744e-01 7.50773191e-01 -5.89708924e-01 -5.09224713e-01
2.39673361e-01 7.32406676e-01 -8.45395982e-01 8.21992978e-02
-1.00178635e+00 6.83364928e-01 3.95640016e-01 3.62642556e-01
-4.91205484e-01 -4.93787587e-01 -1.31555212e+00 -6.82472959e-02
1.90676078e-01 -1.04505074e+00 1.04883206e+00 -1.07754934e+00
-1.32865417e+00 1.22238171e+00 -9.69430879e-02 -3.67643803e-01
7.98117399e-01 -3.73997092e-01 -4.35807288e-01 -2.19259784e-02
2.40574628e-01 8.18914950e-01 1.15306365e+00 -9.88301158e-01
-5.75555980e-01 -7.73378074e-01 -2.28704602e-01 4.07262087e-01
-2.38026530e-01 5.00768535e-02 -9.81775045e-01 -7.81330705e-01
2.57285029e-01 -1.19577110e+00 -8.52808878e-02 2.67781645e-01
-5.43401718e-01 -1.57559618e-01 2.38041937e-01 -1.04276133e+00
1.02274406e+00 -2.08809376e+00 2.34659731e-01 3.96286398e-01
3.39981943e-01 2.64190614e-01 -1.59374535e-01 1.67147160e-01
-3.47694904e-01 -2.24195242e-01 -6.68093264e-02 -9.30551529e-01
9.65567422e-04 -2.79419929e-01 2.51721025e-01 9.21682358e-01
5.26575791e-03 9.93711650e-01 -6.88412309e-01 -6.39057517e-01
1.83554083e-01 4.35566366e-01 -2.92893052e-01 2.34084621e-01
4.90570754e-01 2.61898994e-01 -8.59546140e-02 7.62803912e-01
9.85778213e-01 -6.84009939e-02 1.07670650e-01 -5.06621659e-01
1.01145998e-01 -9.83517468e-02 -1.48313570e+00 1.76525748e+00
-4.96558696e-02 9.43956077e-02 -8.49722326e-02 -7.47460127e-01
9.21126962e-01 2.08154649e-01 5.92111528e-01 -5.12336195e-01
1.81424960e-01 4.10258211e-02 -1.98895544e-01 -8.17478150e-02
4.50093389e-01 2.62729675e-01 -2.31795385e-01 3.46902490e-01
2.25462332e-01 5.22727787e-01 9.12282541e-02 -1.05113849e-01
6.01736903e-01 3.17585617e-01 3.94504547e-01 -6.53154254e-02
7.36034870e-01 -4.05409813e-01 5.99003196e-01 6.41112864e-01
-5.28710485e-01 8.74195457e-01 -1.51445255e-01 -8.36118281e-01
-1.00583017e+00 -1.07318485e+00 -3.83048877e-02 1.01949108e+00
5.88886857e-01 -5.59684515e-01 -8.57620895e-01 -8.59422982e-01
3.75691712e-01 -2.45495394e-01 -9.07653034e-01 -3.36757004e-01
-8.59592736e-01 -7.61455715e-01 6.44010901e-01 8.88489366e-01
8.99863064e-01 -8.41800094e-01 -4.41217691e-01 -1.71167590e-02
-3.33362758e-01 -9.52777684e-01 -1.04741073e+00 -3.00305873e-01
-6.20210469e-01 -1.25241995e+00 -1.26566052e+00 -8.03212821e-01
1.02042925e+00 2.99017310e-01 9.47568536e-01 1.52901888e-01
-5.24124563e-01 4.64762092e-01 -9.20770224e-03 -2.53589839e-01
8.26656222e-02 6.92991763e-02 4.92349833e-01 4.16972458e-01
4.26071852e-01 -1.98030874e-01 -9.64623570e-01 6.24680281e-01
-2.56062180e-01 -1.11642256e-01 5.57093620e-01 8.18291306e-01
7.53765583e-01 -4.42956656e-01 1.69893861e-01 -2.34673128e-01
3.50473017e-01 5.55064231e-02 -4.63441610e-01 4.90220875e-01
-3.79076302e-01 -1.17956862e-01 2.80499846e-01 -5.77813864e-01
-7.00031221e-01 5.85889578e-01 1.27210943e-02 -3.35770637e-01
-9.08063948e-02 -1.59448743e-01 -3.10961396e-01 -3.95966649e-01
5.04628658e-01 3.16752046e-01 2.02720672e-01 -6.45045757e-01
3.29693675e-01 6.35365784e-01 1.02931094e+00 -4.60596651e-01
1.16378224e+00 4.67816621e-01 -2.44935006e-02 -5.00274897e-01
-6.25643015e-01 -9.43290353e-01 -1.16173244e+00 -2.77682990e-01
8.62912118e-01 -1.18912780e+00 -9.09757495e-01 7.96836555e-01
-1.12056792e+00 1.78518772e-01 -4.00895745e-01 4.00968224e-01
-3.53998929e-01 7.02370942e-01 -5.06038666e-01 -6.47340655e-01
-8.12489808e-01 -7.78071463e-01 1.50737107e+00 4.82540965e-01
-3.00465554e-01 -4.92318451e-01 1.65108368e-01 2.70111322e-01
7.60605484e-02 3.72752964e-01 1.64989635e-01 -5.97941220e-01
-1.21901408e-01 -8.07451546e-01 -3.68780017e-01 9.20083523e-02
5.50253764e-02 -4.94268745e-01 -9.83963370e-01 -8.67749691e-01
-5.37548423e-01 -3.69611025e-01 8.82363796e-01 2.28259072e-01
1.04111886e+00 -2.48668835e-01 -6.39048815e-01 8.68109643e-01
1.08157265e+00 -3.12927872e-01 4.17539597e-01 4.76324201e-01
1.01060748e+00 5.67890048e-01 4.20963705e-01 5.07992625e-01
6.14868820e-01 1.20329976e+00 2.50095934e-01 -3.24656934e-01
-2.82496840e-01 -4.39742416e-01 2.52893895e-01 1.74321294e-01
-5.70796669e-01 2.44016021e-01 -5.24253130e-01 3.61827731e-01
-1.95436716e+00 -9.71883059e-01 2.99876094e-01 2.76880312e+00
5.96398473e-01 -2.59805709e-01 7.84478486e-01 -4.76203002e-02
1.01314926e+00 -9.13619772e-02 -5.34264028e-01 1.51386380e-01
5.61633445e-02 -1.33604137e-02 6.75994217e-01 8.31429809e-02
-1.65215969e+00 7.34683275e-01 5.69551706e+00 4.80010241e-01
-7.65916407e-01 -6.45433515e-02 3.24061334e-01 -1.26183748e-01
4.39088970e-01 -4.37687159e-01 -1.21267009e+00 3.79927695e-01
3.49410594e-01 9.72540826e-02 2.34774977e-01 9.01796997e-01
-3.04921895e-01 1.64283410e-01 -1.31155849e+00 1.52575874e+00
6.02862477e-01 -8.06864262e-01 -1.23215199e-01 1.61321372e-01
4.97740626e-01 -3.89853418e-01 -7.89103135e-02 2.94824451e-01
-1.73599944e-01 -1.01575375e+00 9.38173950e-01 5.51212907e-01
9.39413726e-01 -6.91089392e-01 1.01606905e+00 1.76465716e-02
-1.78158700e+00 -1.00815490e-01 -4.52499002e-01 1.65199354e-01
2.29871683e-02 1.31698279e-02 -5.79387903e-01 7.56709337e-01
9.89526987e-01 6.22443795e-01 -9.23983514e-01 1.45527172e+00
-5.03476011e-03 -5.34099862e-02 -4.03956175e-01 4.19316202e-01
-4.32149172e-01 7.76489526e-02 5.10691702e-01 1.25941515e+00
2.93704495e-03 -2.80345768e-01 5.65772831e-01 7.29807794e-01
-1.92753613e-01 1.29052356e-01 -2.01725453e-01 5.34398794e-01
3.41237724e-01 1.36230743e+00 -4.33189571e-01 -4.88141835e-01
-3.11130404e-01 1.35901320e+00 4.22957450e-01 1.51678264e-01
-7.80408561e-01 -3.29672754e-01 8.51145864e-01 1.80418193e-01
3.13736171e-01 1.07604444e-01 9.60889757e-02 -1.22212899e+00
3.27572823e-01 -8.89996588e-01 7.02088594e-01 -2.00606659e-01
-1.47699583e+00 6.36737764e-01 1.65633783e-01 -1.38562226e+00
-4.40753043e-01 -4.39485013e-01 -4.46182728e-01 1.04640448e+00
-1.04653084e+00 -1.64646173e+00 -7.99160898e-01 7.06078112e-01
1.82634085e-01 -7.65576437e-02 8.31039846e-01 3.66424233e-01
-9.32627141e-01 1.34237695e+00 -3.94140899e-01 6.69721365e-01
1.01903033e+00 -1.28047276e+00 8.75320196e-01 9.37897980e-01
-9.70500484e-02 8.12999666e-01 3.40806067e-01 -8.09181273e-01
-1.17718291e+00 -1.21823454e+00 8.70883763e-01 -5.94645798e-01
-8.99748784e-03 -6.34899795e-01 -6.08653963e-01 6.74003601e-01
-3.94130081e-01 3.90445024e-01 5.66176713e-01 1.15913853e-01
-5.96381903e-01 -3.07654202e-01 -1.14908612e+00 3.93273085e-01
1.41025877e+00 -3.33608329e-01 -6.04384184e-01 1.47954509e-01
2.75057971e-01 -7.02276051e-01 -9.19485569e-01 4.48969245e-01
1.09707725e+00 -6.38032854e-01 1.49148977e+00 -6.52635694e-01
-2.19737723e-01 -4.48398679e-01 1.93066463e-01 -1.05509353e+00
-6.55653238e-01 -4.66073483e-01 -1.34473205e-01 1.16923821e+00
4.33301600e-03 -4.74989384e-01 9.34052289e-01 7.89252579e-01
1.91377610e-01 -5.56568086e-01 -9.37306762e-01 -9.44163740e-01
-3.09024930e-01 2.23339453e-01 7.80231714e-01 5.31434774e-01
-6.18446879e-02 4.53348570e-02 -6.24704719e-01 1.87513039e-01
9.58479524e-01 2.63185531e-01 1.17733228e+00 -1.41539109e+00
-3.36438656e-01 -3.17828447e-01 -8.53505373e-01 -1.08603203e+00
1.93154234e-02 -7.91689336e-01 1.39461219e-01 -1.22339046e+00
5.70279121e-01 -2.70952761e-01 -2.46374503e-01 8.04981232e-01
-3.59568685e-01 6.12174213e-01 4.61229324e-01 4.11506921e-01
-6.82921708e-01 4.57288265e-01 1.06488490e+00 -3.89037758e-01
-3.58816326e-01 3.02289248e-01 -5.41741729e-01 6.62281275e-01
5.31510234e-01 -1.58563688e-01 3.38405580e-03 -1.37930680e-02
-1.94832131e-01 -4.09385264e-01 9.88961101e-01 -1.39121306e+00
3.06061357e-01 4.13072586e-01 1.21019995e+00 -5.78622818e-01
5.36723971e-01 -5.73132873e-01 1.06836215e-01 5.54843783e-01
-3.72922719e-02 1.62088498e-01 2.15187296e-01 5.47478914e-01
-1.94097739e-02 6.56841248e-02 7.84190655e-01 -1.89042315e-01
-6.26729667e-01 7.42859483e-01 3.12100500e-01 -5.28921597e-02
8.38446200e-01 -5.57798266e-01 -1.36037022e-01 -1.96576223e-01
-7.30882287e-01 2.36121565e-01 5.77732921e-01 6.69362128e-01
5.31052232e-01 -1.56690764e+00 -6.85832620e-01 4.00080860e-01
4.26956713e-01 -5.49171381e-02 3.73318821e-01 8.50153029e-01
-1.38917297e-01 2.30727240e-01 -4.71991599e-01 -5.84831059e-01
-1.63339996e+00 4.61870909e-01 7.67388165e-01 -3.05408210e-01
-8.15109134e-01 9.11123276e-01 3.57399523e-01 -5.08076608e-01
5.00959277e-01 -1.19309664e-01 -2.88373590e-01 -5.50141484e-02
7.62314141e-01 3.90933871e-01 -6.81106420e-03 -1.22072029e+00
-7.60138571e-01 1.03310680e+00 -1.33162484e-01 1.01571344e-01
8.78612995e-01 -8.80182460e-02 7.98282698e-02 -1.54805988e-01
1.14033544e+00 -1.96146742e-01 -1.23531413e+00 -3.53051692e-01
-3.85345072e-01 -4.02946472e-01 -5.05163431e-01 -7.37440884e-01
-9.60723996e-01 5.30902803e-01 1.08922720e+00 -3.59453648e-01
9.11938250e-01 6.63563842e-03 8.87623429e-01 2.76122451e-01
4.55916196e-01 -1.24436796e+00 -1.69498392e-03 2.78063387e-01
9.69430327e-01 -1.31139946e+00 2.44841382e-01 -4.15432125e-01
-3.71470839e-01 1.00896657e+00 8.64610493e-01 -1.78902596e-02
4.57066655e-01 -2.61475295e-01 -4.75653224e-02 -1.88134775e-01
1.70051053e-01 -4.41877574e-01 8.53189409e-01 7.86277771e-01
2.63659626e-01 2.19772100e-01 -7.17451200e-02 8.60396206e-01
-4.13523227e-01 -1.64845809e-01 -3.30485761e-01 6.25515699e-01
-1.96427062e-01 -1.15440249e+00 -7.34863698e-01 2.88264990e-01
-2.41447762e-01 2.06491396e-01 -6.44691646e-01 9.07487154e-01
5.19200504e-01 5.24504542e-01 1.27629846e-01 -5.63590169e-01
5.68069577e-01 1.03057377e-01 7.21458733e-01 -4.29177672e-01
-7.01035500e-01 -6.92893341e-02 9.04105455e-02 -8.12327623e-01
-2.83244491e-01 -8.07149649e-01 -8.60688925e-01 -1.87966302e-01
-2.44129479e-01 -1.90851271e-01 5.05096912e-01 9.03972387e-01
3.43027532e-01 2.20696449e-01 3.15079212e-01 -1.32030272e+00
-8.04377794e-01 -1.03898311e+00 -4.69491005e-01 9.47850347e-01
1.73947588e-01 -9.41467285e-01 4.06576954e-02 -1.99531429e-02] | [14.666483879089355, 0.9584407806396484] |
622c1d76-5743-496c-a70b-655e15fc32ce | exploiting-deep-learning-for-persian | 1808.05077 | null | http://arxiv.org/abs/1808.05077v1 | http://arxiv.org/pdf/1808.05077v1.pdf | Exploiting Deep Learning for Persian Sentiment Analysis | The rise of social media is enabling people to freely express their opinions
about products and services. The aim of sentiment analysis is to automatically
determine subject's sentiment (e.g., positive, negative, or neutral) towards a
particular aspect such as topic, product, movie, news etc. Deep learning has
recently emerged as a powerful machine learning technique to tackle a growing
demand of accurate sentiment analysis. However, limited work has been conducted
to apply deep learning algorithms to languages other than English, such as
Persian. In this work, two deep learning models (deep autoencoders and deep
convolutional neural networks (CNNs)) are developed and applied to a novel
Persian movie reviews dataset. The proposed deep learning models are analyzed
and compared with the state-of-the-art shallow multilayer perceptron (MLP)
based machine learning model. Simulation results demonstrate the enhanced
performance of deep learning over state-of-the-art MLP. | ['Kia Dashtipour', 'Amir Hussain', 'Mandar Gogate', 'Hadi Larijani', 'Cosimo Ieracitano', 'Ahsan Adeel'] | 2018-08-15 | null | null | null | null | ['persian-sentiment-anlysis'] | ['natural-language-processing'] | [-3.76503706e-01 -1.03931658e-01 -3.13755065e-01 -8.75999212e-01
-5.04806265e-02 -2.27173612e-01 4.83085394e-01 4.41868663e-01
-6.04537308e-01 7.21267700e-01 2.28677884e-01 -4.20492381e-01
4.32555109e-01 -7.99230456e-01 -3.76294345e-01 -5.37648678e-01
2.79326081e-01 2.49759346e-01 -3.47082108e-01 -7.54270375e-01
4.03969854e-01 2.79496938e-01 -1.25242424e+00 4.91072953e-01
5.35054982e-01 1.15096545e+00 -1.98985413e-01 4.63581294e-01
-4.56501216e-01 1.33436882e+00 -7.50344276e-01 -7.37874568e-01
-2.23374054e-01 8.88439566e-02 -9.46045160e-01 -1.14945523e-01
-2.61992700e-02 -9.62275490e-02 9.79784131e-02 1.29446936e+00
4.66335267e-01 1.22031868e-01 6.62780702e-01 -8.94171476e-01
-1.20287621e+00 6.27499580e-01 -5.95026433e-01 3.62818182e-01
-1.17411027e-02 -5.23030162e-01 1.03202021e+00 -7.85719872e-01
1.83320016e-01 1.10540116e+00 6.18753612e-01 3.71452659e-01
-4.74530101e-01 -6.06396377e-01 3.64921182e-01 1.99496359e-01
-7.52795756e-01 -2.36515850e-02 1.06694543e+00 -5.00710189e-01
1.16847658e+00 -2.16646329e-01 6.24892175e-01 9.60236907e-01
9.13981736e-01 9.69348311e-01 1.25740290e+00 -3.54748219e-01
4.51594293e-01 8.83110940e-01 3.63623917e-01 2.90325522e-01
-5.24934866e-02 -4.68970805e-01 -4.72069144e-01 7.22953081e-02
6.33242354e-02 8.65495875e-02 3.06536257e-01 2.22798392e-01
-4.71851230e-01 1.23381603e+00 6.35708511e-01 4.66669738e-01
-7.85554528e-01 -2.60001868e-01 7.29974866e-01 7.20019400e-01
1.03934097e+00 4.07291144e-01 -1.07743096e+00 -1.12335272e-01
-4.85259861e-01 1.17001727e-01 9.22416866e-01 3.62851590e-01
6.55190825e-01 4.66501981e-01 5.48568308e-01 1.12275147e+00
4.80264723e-01 3.78033698e-01 1.01444876e+00 -1.79454461e-01
1.20919622e-01 8.61457288e-01 -5.36224470e-02 -1.51843202e+00
-5.76503575e-01 -5.54492772e-01 -1.17614019e+00 2.21454218e-01
-3.06645125e-01 -7.21287608e-01 -5.88302195e-01 1.11946154e+00
1.32476479e-01 -4.64455217e-01 6.67828858e-01 9.30086076e-01
1.22832036e+00 1.21350384e+00 8.79739225e-02 1.54338732e-01
1.36939573e+00 -1.00102448e+00 -6.88843548e-01 -6.52171493e-01
5.36348701e-01 -9.55677688e-01 6.49760842e-01 6.87498868e-01
-7.71075785e-01 -7.86792159e-01 -1.04682767e+00 -6.01850525e-02
-9.12860036e-01 3.72037143e-01 8.99384618e-01 6.62862301e-01
-8.31077039e-01 5.07503808e-01 -6.25337958e-01 -4.23129320e-01
3.82637829e-01 7.44304597e-01 -2.83462018e-01 3.75352085e-01
-1.45520067e+00 1.02725863e+00 1.15157276e-01 2.26346105e-01
-3.84192973e-01 -8.63082260e-02 -8.29479992e-01 -5.71037307e-02
-3.29143077e-01 -1.19106032e-01 1.53148127e+00 -1.94697845e+00
-1.86020052e+00 1.02046585e+00 1.38864443e-02 -5.12958705e-01
-2.42378712e-01 -4.22259599e-01 -9.12198901e-01 -2.42133617e-01
-1.50239795e-01 5.80589712e-01 6.98128462e-01 -8.71849775e-01
-7.10912466e-01 -5.65255582e-01 3.97270530e-01 1.74150974e-01
-9.27476585e-01 4.05833125e-01 1.84187293e-01 -3.90145034e-01
-2.68012255e-01 -8.65336776e-01 -3.42925012e-01 -5.19172430e-01
-2.57521510e-01 -5.25698185e-01 8.60946178e-01 -7.04552114e-01
7.45702326e-01 -2.07400608e+00 -2.84365028e-01 3.31897587e-02
-5.39119355e-02 5.19289374e-01 1.35527715e-01 3.87980878e-01
-1.46697819e-01 3.70248742e-02 1.21357866e-01 -2.44064525e-01
-8.41183811e-02 2.86628813e-01 -1.03681602e-01 4.46265221e-01
3.85357141e-01 5.07650137e-01 -6.09869242e-01 -2.12475628e-01
2.00213343e-01 6.85042679e-01 -4.57836509e-01 1.70499295e-01
1.11185797e-01 1.23820961e-01 -6.66430533e-01 6.45114422e-01
6.10331833e-01 -2.39915684e-01 1.98144183e-01 -1.56377852e-01
-3.33833784e-01 3.80848140e-01 -7.72160113e-01 1.04836762e+00
-8.25987339e-01 9.66790318e-01 -5.66172451e-02 -1.27916682e+00
1.51423883e+00 4.15573329e-01 1.09804414e-01 -7.71609485e-01
7.35465586e-01 7.73747042e-02 2.23689303e-01 -5.85036695e-01
8.26680243e-01 -4.52297330e-01 -1.90573394e-01 2.34990492e-01
2.38312423e-01 2.66096860e-01 -5.45808785e-02 -2.74271011e-01
3.70538920e-01 -4.09251094e-01 3.66715878e-01 -5.53195894e-01
8.94250274e-01 -5.32777980e-02 5.06695569e-01 2.42815927e-01
-2.46738762e-01 1.44806027e-01 6.34933889e-01 -1.05296195e+00
-9.16178226e-01 -2.32682362e-01 -4.90774661e-02 1.33346760e+00
-1.85798407e-01 8.20795596e-02 -5.75437307e-01 -4.87332553e-01
-1.16715148e-01 5.26070416e-01 -7.52126455e-01 -9.93138831e-03
-1.65176570e-01 -8.94595921e-01 3.43924761e-02 6.36829793e-01
7.55778909e-01 -1.39310122e+00 -6.37396455e-01 1.58592641e-01
2.74940401e-01 -1.00710011e+00 3.01229388e-01 4.09844756e-01
-7.32443333e-01 -5.21824419e-01 -7.02897251e-01 -1.36145473e+00
4.70678717e-01 -2.61677623e-01 1.17185867e+00 -3.31734389e-01
3.33216310e-01 -1.20533220e-01 -5.67827702e-01 -8.67990911e-01
-4.37284380e-01 4.82538730e-01 3.34495813e-01 2.88019449e-01
9.84470725e-01 -3.88200700e-01 -5.62620223e-01 -5.17645478e-02
-7.20287621e-01 -1.17665142e-01 6.96799338e-01 6.65794671e-01
6.07836768e-02 4.85274941e-01 1.07768190e+00 -1.29425740e+00
1.10787201e+00 -6.93293273e-01 -3.11238110e-01 -2.11099327e-01
-5.95552921e-01 -1.25433832e-01 1.02897000e+00 -2.07102329e-01
-1.25615335e+00 -2.75628179e-01 -6.73372984e-01 2.49059141e-01
-2.61880189e-01 1.15860832e+00 1.64561704e-01 7.51957521e-02
4.89693373e-01 -9.56707373e-02 -4.38774601e-02 -3.87256533e-01
5.33275194e-02 1.43055737e+00 2.60721743e-01 -3.73016372e-02
-1.34000378e-02 3.85203391e-01 -5.38124859e-01 -8.94039929e-01
-1.05300641e+00 -5.67439854e-01 -5.53111434e-01 -2.08345249e-01
9.58899140e-01 -1.02437353e+00 -8.14961314e-01 7.54325628e-01
-1.08467090e+00 9.60327536e-02 1.23034850e-01 5.86842835e-01
-2.22558454e-02 -9.51409563e-02 -9.46852863e-01 -8.18530142e-01
-8.19999933e-01 -1.02667391e+00 6.99740291e-01 5.97497880e-01
-4.65449363e-01 -1.44108987e+00 3.32037032e-01 2.87332594e-01
5.81828296e-01 1.58299729e-02 8.69619250e-01 -1.14420402e+00
6.42015576e-01 -7.81219900e-01 -5.28758131e-02 1.06319427e+00
7.86645114e-02 6.11636005e-02 -1.14836025e+00 -3.96239981e-02
3.44551533e-01 -6.10806704e-01 5.76772869e-01 4.61875200e-01
8.12562644e-01 -5.48059344e-01 2.00725228e-01 1.23778135e-01
1.46480012e+00 5.60690820e-01 6.41595960e-01 6.92477405e-01
4.45549965e-01 6.85024381e-01 3.40119362e-01 5.69844365e-01
4.75866377e-01 -3.96883860e-03 4.27259058e-01 -3.25554639e-01
8.24034989e-01 2.45026916e-01 4.64433223e-01 1.26003301e+00
8.92613381e-02 -3.79967779e-01 -7.05328822e-01 6.13321602e-01
-1.60502672e+00 -5.77730536e-01 -7.54542574e-02 1.40978861e+00
5.87166131e-01 4.11009252e-01 -1.54521301e-01 3.22319269e-01
5.51118612e-01 4.19749618e-01 -4.30673689e-01 -1.56248426e+00
-1.80804163e-01 3.76654536e-01 3.20110649e-01 2.24493816e-01
-1.51274884e+00 1.01454103e+00 5.57061958e+00 4.39791858e-01
-1.77139699e+00 3.18198875e-02 9.63006854e-01 2.86463618e-01
-4.67928592e-03 -6.04869306e-01 -6.82249606e-01 1.42411366e-01
1.26290524e+00 1.59239203e-01 -1.40921855e-02 1.48175085e+00
1.02981731e-01 1.29038952e-02 -5.36109447e-01 8.30333650e-01
3.31824183e-01 -1.21210957e+00 -7.55002946e-02 -2.55855680e-01
1.03891790e+00 4.23026025e-01 4.50533748e-01 5.53740144e-01
3.34066212e-01 -9.62778330e-01 4.09922630e-01 1.51052490e-01
6.02203384e-02 -1.18905234e+00 1.61861598e+00 2.19823718e-01
-5.40198863e-01 -1.61981046e-01 -6.33205593e-01 -6.68932199e-01
-6.25419244e-03 6.94743752e-01 -6.11774385e-01 1.56156480e-01
1.07799971e+00 1.22674668e+00 -3.63143682e-01 2.90217787e-01
-2.04164773e-01 9.32941616e-01 1.63145110e-01 -8.36964011e-01
7.97841132e-01 -3.93599309e-02 -4.06225724e-03 1.36686063e+00
-7.85730779e-02 1.01413824e-01 -8.36149082e-02 3.98231149e-01
-1.55928805e-01 5.34423769e-01 -5.61901987e-01 -4.63733822e-01
-4.05734330e-01 1.64849627e+00 -6.23378158e-01 -3.06556582e-01
-7.88087368e-01 8.56501043e-01 6.67896643e-02 2.37475082e-01
-2.69960940e-01 -4.86340642e-01 7.04527497e-01 -1.61877468e-01
2.83766627e-01 1.57937154e-01 -4.78170782e-01 -1.01118541e+00
-3.41381788e-01 -1.01199281e+00 1.66572601e-01 -8.94333780e-01
-1.50545394e+00 1.08876050e+00 -6.90441489e-01 -9.65754390e-01
-2.73377001e-01 -1.16471899e+00 -7.46811450e-01 9.12735939e-01
-1.73985255e+00 -9.89799917e-01 1.75772890e-01 5.70041597e-01
6.40711486e-01 -8.74583125e-01 9.42832232e-01 2.90470153e-01
-4.41467375e-01 2.25147814e-01 3.55862945e-01 4.66473669e-01
4.76510227e-01 -1.10777354e+00 3.36991638e-01 3.52524191e-01
-3.30935478e-01 5.58051407e-01 7.20942140e-01 -3.25776279e-01
-1.18842709e+00 -8.89334738e-01 1.33859980e+00 1.70660801e-02
7.21990824e-01 -1.20435752e-01 -5.68417311e-01 5.58872759e-01
8.40791881e-01 -3.12125534e-01 1.16073966e+00 5.95511198e-01
-1.03537682e-02 -4.10432845e-01 -1.08973849e+00 5.18128216e-01
-3.41707289e-01 -4.04134423e-01 -4.75827426e-01 8.81414711e-02
3.82752150e-01 -1.59051105e-01 -8.32778156e-01 3.35724682e-01
6.50521219e-01 -1.05033183e+00 5.05911171e-01 -9.10130620e-01
1.15907717e+00 5.79894930e-02 -3.34078908e-01 -1.52733433e+00
-2.77776986e-01 -9.77879465e-02 1.89513251e-01 1.00654185e+00
6.34591043e-01 -6.68095410e-01 9.46746171e-01 4.03684527e-01
-1.58579037e-01 -1.12517154e+00 -4.87882912e-01 8.98140371e-02
2.45049641e-01 -4.05195832e-01 3.81162226e-01 1.13962209e+00
1.91521898e-01 7.80983686e-01 -3.98774594e-01 1.74775459e-02
-1.76234454e-01 4.17734161e-02 5.07423818e-01 -1.34383059e+00
9.67464224e-02 -3.53299916e-01 -7.95721591e-01 -8.12380075e-01
3.69747549e-01 -5.31388104e-01 -2.37374231e-01 -1.71594822e+00
-1.38908714e-01 -3.84965800e-02 -7.06775367e-01 3.33153963e-01
3.69603664e-01 2.05478549e-01 -6.19343817e-02 -1.98025554e-01
-4.82569575e-01 5.39665878e-01 9.80480790e-01 -4.83292520e-01
-3.00079919e-02 2.53626168e-01 -1.24428928e+00 1.14313710e+00
1.21890414e+00 -4.20069754e-01 -1.74345061e-01 -3.74733508e-01
9.83478248e-01 -2.53526002e-01 -1.42010614e-01 -8.15030038e-01
1.35637730e-01 6.35828152e-02 5.72645187e-01 -7.16439009e-01
1.83790222e-01 -7.24311411e-01 -5.34616828e-01 4.02140647e-01
-6.38312459e-01 3.78981084e-01 3.72069418e-01 7.53916502e-02
-8.35531533e-01 -2.73942769e-01 6.03305817e-01 -2.37039894e-01
-9.68777120e-01 4.06970084e-02 -8.87390792e-01 -3.61784369e-01
5.75304031e-01 2.01515794e-01 -1.73016578e-01 -5.18796802e-01
-6.18603766e-01 1.22302279e-01 -9.18786228e-02 8.16342473e-01
7.14643359e-01 -1.07304978e+00 -7.08113670e-01 1.64053023e-01
1.99093651e-02 -2.60977894e-01 3.78191739e-01 4.50484425e-01
-8.46310914e-01 5.48468888e-01 -4.35025334e-01 -3.01795751e-01
-1.10037482e+00 3.99867445e-01 2.98430830e-01 -2.62888044e-01
-1.36842623e-01 9.79307055e-01 -6.62883697e-03 -9.08862412e-01
6.18632771e-02 -3.63004357e-01 -1.13564432e+00 1.87671423e-01
4.90038365e-01 5.12250289e-02 3.78139585e-01 -9.05729532e-01
-4.14152056e-01 4.93694186e-01 -4.56090063e-01 1.25485152e-01
1.50367594e+00 1.21334977e-01 -2.96720237e-01 7.74654329e-01
1.54217851e+00 -3.46258551e-01 -5.72734296e-01 -5.87497950e-02
-2.72754699e-01 7.48398304e-02 3.88230324e-01 -9.53783274e-01
-1.40507650e+00 1.18728673e+00 7.37637997e-01 4.20686096e-01
1.07441401e+00 -3.02479953e-01 9.39394176e-01 7.28317559e-01
-4.25009876e-02 -1.65452492e+00 -3.45273055e-02 7.63021231e-01
6.19723201e-01 -1.71652031e+00 -1.46578163e-01 3.54147255e-01
-1.13910544e+00 1.29668784e+00 6.91463530e-01 -5.17621994e-01
1.22226226e+00 1.45993248e-01 6.28225684e-01 -2.29424492e-01
-7.79235601e-01 1.07931517e-01 1.06548935e-01 3.90301675e-01
1.02528560e+00 1.68745860e-01 -2.02604696e-01 9.12875056e-01
-5.30163467e-01 2.01443523e-01 5.92049062e-01 8.99969816e-01
-4.29189086e-01 -6.32092178e-01 -2.01402307e-02 4.34217423e-01
-1.20263863e+00 -1.50380805e-01 -5.03915250e-01 4.15619612e-01
1.47140697e-01 1.31594646e+00 3.00343215e-01 -5.28118253e-01
1.39309928e-01 -2.54547924e-01 -2.39762157e-01 -5.28466344e-01
-1.03705513e+00 -1.12986602e-01 2.25510240e-01 1.12873070e-01
-8.37682545e-01 -3.10048997e-01 -1.10568082e+00 -4.26913619e-01
-2.29607075e-01 3.46853644e-01 1.09968162e+00 1.06309557e+00
1.20466255e-01 5.83281338e-01 8.51318657e-01 -5.29090941e-01
-2.61269629e-01 -1.04874122e+00 -7.57723510e-01 1.84706002e-01
3.66998255e-01 -1.69057831e-01 -2.01491266e-01 4.97058369e-02] | [11.152005195617676, 7.017779350280762] |
d1dcb3df-51d6-468a-a4ec-617de1621955 | an-effective-motion-centric-paradigm-for-3d | 2303.12535 | null | https://arxiv.org/abs/2303.12535v1 | https://arxiv.org/pdf/2303.12535v1.pdf | An Effective Motion-Centric Paradigm for 3D Single Object Tracking in Point Clouds | 3D single object tracking in LiDAR point clouds (LiDAR SOT) plays a crucial role in autonomous driving. Current approaches all follow the Siamese paradigm based on appearance matching. However, LiDAR point clouds are usually textureless and incomplete, which hinders effective appearance matching. Besides, previous methods greatly overlook the critical motion clues among targets. In this work, beyond 3D Siamese tracking, we introduce a motion-centric paradigm to handle LiDAR SOT from a new perspective. Following this paradigm, we propose a matching-free two-stage tracker M^2-Track. At the 1st-stage, M^2-Track localizes the target within successive frames via motion transformation. Then it refines the target box through motion-assisted shape completion at the 2nd-stage. Due to the motion-centric nature, our method shows its impressive generalizability with limited training labels and provides good differentiability for end-to-end cycle training. This inspires us to explore semi-supervised LiDAR SOT by incorporating a pseudo-label-based motion augmentation and a self-supervised loss term. Under the fully-supervised setting, extensive experiments confirm that M^2-Track significantly outperforms previous state-of-the-arts on three large-scale datasets while running at 57FPS (~8%, ~17% and ~22% precision gains on KITTI, NuScenes, and Waymo Open Dataset respectively). While under the semi-supervised setting, our method performs on par with or even surpasses its fully-supervised counterpart using fewer than half labels from KITTI. Further analysis verifies each component's effectiveness and shows the motion-centric paradigm's promising potential for auto-labeling and unsupervised domain adaptation. | ['Zhen Li', 'Shuguang Cui', 'Shenghui Cheng', 'Baoyuan Wang', 'Haiming Zhang', 'Xu Yan', 'Chaoda Zheng'] | 2023-03-21 | null | null | null | null | ['3d-single-object-tracking', 'pseudo-label'] | ['computer-vision', 'miscellaneous'] | [ 5.69390915e-02 -3.25885624e-01 -4.45355147e-01 -3.13531309e-01
-7.68444896e-01 -6.03461087e-01 6.14862978e-01 -1.13143995e-01
-5.47572792e-01 4.39992249e-01 -4.37809885e-01 -3.30866903e-01
1.92998443e-02 -3.81847382e-01 -8.85453284e-01 -5.65364957e-01
1.27866879e-01 6.51701212e-01 7.52040207e-01 -1.86568826e-01
3.20441537e-02 5.77079892e-01 -1.66641319e+00 -4.17523205e-01
8.80317867e-01 9.40245032e-01 2.69104034e-01 3.78730685e-01
-1.99013054e-01 3.53171438e-01 -1.27027512e-01 -3.74361098e-01
6.48827136e-01 1.86605737e-01 -1.57687157e-01 1.78715393e-01
1.21726274e+00 -1.44030347e-01 -4.78190392e-01 9.30942655e-01
3.56361836e-01 1.29309878e-01 4.35776919e-01 -1.60221326e+00
-2.33283252e-01 -1.30863816e-01 -9.06481624e-01 1.23904087e-01
7.94214159e-02 4.47654366e-01 7.92682290e-01 -1.14511418e+00
5.80875456e-01 1.32275712e+00 9.94863033e-01 6.55422390e-01
-1.26387787e+00 -1.06065679e+00 3.53749692e-01 -1.33490292e-02
-1.48072219e+00 -3.96456599e-01 7.41074324e-01 -5.58967531e-01
6.01816177e-01 -5.62671795e-02 5.92966914e-01 8.78709078e-01
9.29865688e-02 8.08918297e-01 1.06531537e+00 -4.26642597e-02
9.93951485e-02 -1.69957783e-02 -1.45473452e-02 8.51209342e-01
5.32182693e-01 5.27277052e-01 -6.39665067e-01 8.65444094e-02
5.81047833e-01 3.05361524e-02 7.01686442e-02 -1.00634408e+00
-1.50371325e+00 5.82191527e-01 5.68369210e-01 -1.61405742e-01
1.06448047e-01 3.10141742e-01 3.65565658e-01 1.28128588e-01
3.96055162e-01 -1.18259646e-01 -2.68212378e-01 -8.19006264e-02
-1.08435297e+00 2.71688640e-01 2.88536370e-01 1.48897183e+00
9.03310895e-01 3.11279655e-01 -5.99852987e-02 4.08775628e-01
4.51319844e-01 1.06736183e+00 1.32503375e-01 -1.01762736e+00
5.94051600e-01 5.36285222e-01 3.03999573e-01 -6.83667302e-01
-2.69043833e-01 -5.25217533e-01 -5.29802918e-01 6.73680961e-01
4.86305594e-01 7.07959682e-02 -1.09213674e+00 1.80233967e+00
8.60430598e-01 5.49893498e-01 -4.83506434e-02 9.39817250e-01
5.72681487e-01 1.92209914e-01 1.68012232e-01 -9.57357511e-02
1.25446630e+00 -1.12425280e+00 -5.01167178e-01 -4.50401217e-01
5.60535729e-01 -7.10471749e-01 9.65452909e-01 1.42874837e-01
-7.18827784e-01 -9.41447914e-01 -1.08658624e+00 -2.80604381e-02
-1.87438279e-01 2.42551625e-01 5.71291745e-01 7.06453860e-01
-8.72618496e-01 3.51682097e-01 -1.08073270e+00 -4.89143461e-01
6.15435421e-01 4.63657796e-01 -5.17314315e-01 -2.44966313e-01
-5.80708265e-01 8.66923451e-01 2.02276185e-01 -1.04092687e-01
-9.43682909e-01 -8.79523993e-01 -9.01706636e-01 -5.58896780e-01
4.59016651e-01 -8.35250080e-01 1.25243723e+00 -3.89420658e-01
-1.34302104e+00 8.51222098e-01 -5.73222876e-01 -6.17689490e-01
7.62120426e-01 -3.78310204e-01 -3.37008089e-01 6.75558001e-02
3.98319811e-01 1.16955018e+00 1.01641595e+00 -1.40834701e+00
-1.14703619e+00 -4.10202980e-01 -1.56401068e-01 1.87566996e-01
-9.13243890e-02 -4.27149773e-01 -7.33910024e-01 -4.72946197e-01
3.41567636e-01 -1.35453987e+00 -1.46904513e-01 5.49763262e-01
-1.18211746e-01 -2.69135833e-01 1.32118917e+00 4.60093981e-03
7.30005682e-01 -2.26994967e+00 -1.75327763e-01 -2.77749598e-01
4.03700054e-01 3.29689264e-01 -8.22880343e-02 4.16981168e-02
1.88198492e-01 -3.47681046e-01 -2.94499099e-01 -8.68644774e-01
1.11759938e-01 3.33294868e-01 -2.59002566e-01 6.60450101e-01
3.88350666e-01 1.06141031e+00 -1.10287428e+00 -6.96856618e-01
5.10352075e-01 4.77278799e-01 -2.86862701e-01 -1.12562627e-01
-2.41020501e-01 6.80045843e-01 -4.94686097e-01 8.81037354e-01
9.43886757e-01 -7.03645572e-02 -3.28911185e-01 -1.73404261e-01
-3.75683218e-01 -3.52571197e-02 -1.07623386e+00 1.93786180e+00
-2.31950402e-01 7.15626240e-01 2.97111683e-02 -6.25686705e-01
9.62333441e-01 -6.51816949e-02 6.69830441e-01 -6.50009036e-01
-4.97019328e-02 4.32388723e-01 -1.05267689e-01 -2.07732901e-01
6.52303398e-01 -1.76729307e-01 -3.83818112e-02 5.68017960e-02
-1.85984239e-01 -1.81342825e-01 7.20316991e-02 2.47304127e-01
1.00735188e+00 7.38913059e-01 -1.20851792e-01 -6.85489625e-02
4.64824200e-01 4.35128897e-01 7.45828509e-01 5.40541589e-01
-6.21185303e-01 6.11718893e-01 -3.39696258e-01 -2.70045072e-01
-9.50311124e-01 -1.12391114e+00 -3.36939186e-01 7.08442330e-01
6.89849496e-01 -1.38880670e-01 -3.84911537e-01 -8.29365075e-01
5.40934801e-01 3.75807852e-01 -4.87274736e-01 -8.75386298e-02
-6.42428577e-01 -2.31940731e-01 5.71833849e-01 6.71364307e-01
5.91490746e-01 -5.44819355e-01 -7.77212441e-01 1.39499575e-01
-8.21182281e-02 -1.62340927e+00 -6.25462115e-01 1.87036529e-01
-1.11885989e+00 -8.50333571e-01 -6.05639160e-01 -5.77197015e-01
5.90031207e-01 1.00505936e+00 7.61235356e-01 -2.17141122e-01
-1.06187627e-01 4.61017907e-01 -1.44826353e-01 -4.44162935e-01
-1.00164048e-01 9.05200315e-04 4.40669507e-01 3.12633291e-02
5.02550304e-01 -6.00576222e-01 -6.72006249e-01 5.65813363e-01
-4.54060942e-01 -1.55472279e-01 7.53061950e-01 4.90070730e-01
8.43871713e-01 -2.45996580e-01 2.78295249e-01 -4.79857832e-01
-2.66512483e-01 -2.31972724e-01 -8.95208836e-01 -1.53846279e-01
-7.60627031e-01 -6.71712682e-02 2.59862959e-01 -7.21629441e-01
-8.70776892e-01 6.45456910e-01 1.90703273e-01 -1.21383774e+00
-1.73624516e-01 -1.97814301e-01 -1.77624837e-01 -5.27084947e-01
5.24284422e-01 2.29299441e-01 9.87700149e-02 -3.52920562e-01
4.52890813e-01 3.44098181e-01 9.26433623e-01 -5.19742846e-01
1.71533549e+00 1.01332128e+00 1.86473891e-01 -5.94197810e-01
-8.00550461e-01 -7.84037828e-01 -8.36416066e-01 -3.16678435e-01
8.56370807e-01 -1.36868179e+00 -8.56830359e-01 3.09163094e-01
-9.68618214e-01 -2.47292593e-01 -4.95763808e-01 5.91460705e-01
-4.79683608e-01 5.06608427e-01 -1.00299343e-01 -1.00487041e+00
-1.31198555e-01 -1.05412877e+00 1.52258694e+00 1.08907729e-01
1.29372418e-01 -6.91266537e-01 5.28815202e-02 5.58449030e-01
1.26901150e-01 3.52489352e-01 2.57024795e-01 -3.08950990e-01
-9.08000290e-01 -2.64119685e-01 -3.67534518e-01 9.40401107e-02
8.50324985e-03 -1.19753249e-01 -1.06397402e+00 -5.27175546e-01
-3.01677227e-01 -1.80810243e-01 8.66262734e-01 1.23398393e-01
5.38703620e-01 4.84650224e-01 -6.59905314e-01 7.69910991e-01
1.28702974e+00 3.28668915e-02 3.09587806e-01 4.83645290e-01
9.97548997e-01 5.10100007e-01 1.28870785e+00 8.85870308e-02
7.66747236e-01 9.44435060e-01 6.79692745e-01 -1.20654099e-01
-4.94088858e-01 -4.96286929e-01 6.73037052e-01 6.87141657e-01
1.47499576e-01 9.85288173e-02 -7.70866632e-01 5.93957782e-01
-1.89923131e+00 -7.68215418e-01 -5.53803682e-01 2.43894935e+00
5.25909364e-01 4.83211607e-01 3.11983973e-01 4.70328983e-03
6.39830768e-01 7.82067999e-02 -8.54231417e-01 3.17012072e-01
-8.83179978e-02 -7.10101873e-02 8.04550946e-01 4.10026252e-01
-1.16821551e+00 1.27814102e+00 5.09393501e+00 9.98706281e-01
-1.09824038e+00 3.20460230e-01 1.55584970e-02 8.38642940e-03
2.04439182e-02 2.31453195e-01 -1.41674304e+00 4.08306092e-01
6.80788159e-01 -3.68722379e-02 -6.83113933e-02 9.39204693e-01
2.79990375e-01 -1.29526570e-01 -1.05053174e+00 8.92503202e-01
-1.42859221e-01 -1.06311512e+00 -2.38888681e-01 3.34899038e-01
6.45240545e-01 4.54956621e-01 7.60406554e-02 4.72784907e-01
2.21386999e-01 -5.98242700e-01 1.19189799e+00 1.81470975e-01
1.04978883e+00 -5.08252859e-01 3.14880610e-01 4.99958843e-01
-1.87620044e+00 7.64140412e-02 -4.19392735e-01 1.10379443e-01
3.03268731e-01 3.70367497e-01 -6.99937284e-01 8.17102671e-01
7.36879647e-01 9.43635643e-01 -6.34776473e-01 1.11503768e+00
-6.65265396e-02 3.87227893e-01 -5.92117488e-01 2.92217463e-01
3.49554926e-01 -1.37083203e-01 8.33077013e-01 1.11764848e+00
3.77898574e-01 -3.49090137e-02 4.96715844e-01 6.18695557e-01
1.08104683e-01 -2.63848245e-01 -8.20733130e-01 2.44063169e-01
7.17305124e-01 1.37517810e+00 -7.19263256e-01 -3.13685268e-01
-4.00654793e-01 7.67436326e-01 1.60594538e-01 1.80058464e-01
-1.16314840e+00 -9.76811126e-02 7.91471183e-01 3.22219253e-01
6.14134192e-01 -5.92709780e-01 -2.06945568e-01 -1.00805271e+00
2.53741205e-01 -5.46907425e-01 5.51623479e-02 -6.13114595e-01
-9.83157992e-01 5.24905860e-01 4.19379473e-02 -1.99592304e+00
9.42878723e-02 -4.83892113e-01 -4.03191477e-01 5.12241662e-01
-1.98950851e+00 -1.52328730e+00 -6.25049770e-01 7.42068172e-01
7.20062256e-01 2.05793511e-02 2.86902219e-01 5.32624304e-01
-4.50091600e-01 6.28025949e-01 -1.14631921e-01 -1.46276906e-01
8.69195819e-01 -1.13229060e+00 6.17604792e-01 9.51187372e-01
2.95554221e-01 4.58087951e-01 6.30833328e-01 -8.31572950e-01
-1.65546155e+00 -1.44130969e+00 7.22191095e-01 -8.93588185e-01
6.83031857e-01 -3.97040069e-01 -8.06770921e-01 5.68148136e-01
-1.77712873e-01 4.05038893e-01 1.14466421e-01 -3.64057869e-01
-4.60344195e-01 -3.58843893e-01 -1.01849985e+00 5.33334434e-01
1.54061615e+00 -2.80656397e-01 -4.04070407e-01 3.09707284e-01
9.71667528e-01 -6.85572565e-01 -6.77804291e-01 7.65573025e-01
4.88789380e-01 -6.69742227e-01 1.14780879e+00 -1.13934778e-01
-1.49629802e-01 -9.59795177e-01 -2.49443531e-01 -6.22599304e-01
-1.28477111e-01 -8.45242977e-01 -2.66999751e-01 1.16048777e+00
1.08501822e-01 -5.70813298e-01 1.09887254e+00 8.24884772e-02
-4.21722144e-01 -4.97299880e-01 -1.20621812e+00 -1.30499494e+00
-3.71156372e-02 -7.16942847e-01 2.87866950e-01 8.13118577e-01
-5.61823368e-01 3.25857729e-01 -3.06418419e-01 4.83561903e-01
1.16156018e+00 2.04919919e-01 1.42010236e+00 -1.45892441e+00
-2.90730037e-04 -2.70307958e-01 -4.88321573e-01 -1.57743645e+00
1.86055541e-01 -7.69647121e-01 3.09582293e-01 -1.09987164e+00
-7.61233643e-02 -8.59213769e-01 -4.66213264e-02 4.71759409e-01
-1.27830163e-01 5.74003696e-01 2.65290856e-01 4.71631497e-01
-7.28258431e-01 7.72548735e-01 1.21281970e+00 -2.12433487e-01
-1.11027621e-01 1.19995750e-01 -3.15705657e-01 6.27894521e-01
4.92582142e-01 -7.12624133e-01 -3.94184530e-01 -4.49959308e-01
-2.76321530e-01 -1.66482151e-01 5.75358391e-01 -1.40186238e+00
4.76708323e-01 -1.95625708e-01 7.86487460e-02 -1.03768086e+00
6.48846507e-01 -1.03792083e+00 1.34982243e-01 5.09099483e-01
2.49126181e-01 1.65765584e-01 5.12302339e-01 1.00903118e+00
-7.32790977e-02 1.64469168e-01 8.02380741e-01 3.39829266e-01
-1.01409781e+00 7.19327450e-01 -3.50769274e-02 7.33396038e-02
1.20345747e+00 -8.17762136e-01 -1.57703832e-01 -5.65327257e-02
-3.66374761e-01 5.36839545e-01 7.19252348e-01 6.03174090e-01
6.46255076e-01 -1.47218275e+00 -5.58099568e-01 1.47073552e-01
5.06967187e-01 3.99720967e-01 8.90163053e-03 1.28717518e+00
-2.02368125e-01 4.05417621e-01 7.25706592e-02 -1.34022987e+00
-1.26811779e+00 4.66005653e-01 1.17971055e-01 -2.35239845e-02
-8.93165827e-01 6.83376789e-01 3.39104056e-01 -3.94366831e-01
2.68392950e-01 -3.67444545e-01 1.84395328e-01 -2.44573310e-01
1.16993338e-01 2.16359690e-01 6.43846160e-03 -1.02764654e+00
-4.96332496e-01 1.13448632e+00 -4.67206584e-03 -1.63403541e-01
9.10552979e-01 -2.45789930e-01 5.74541926e-01 4.49575573e-01
9.46325839e-01 1.79351673e-01 -1.91154635e+00 -4.24877137e-01
1.55166104e-01 -6.05648875e-01 -3.16149890e-02 -3.31928521e-01
-1.09668005e+00 9.99351680e-01 7.87006140e-01 -3.57931167e-01
7.55318820e-01 -7.72328004e-02 9.42847073e-01 3.57547879e-01
6.57013118e-01 -6.80774331e-01 1.98585272e-01 4.76197869e-01
3.70700538e-01 -1.56849754e+00 3.35202143e-02 -5.18627107e-01
-5.90828359e-01 6.89623654e-01 8.31132889e-01 -5.39368726e-02
2.28993639e-01 3.23717147e-01 2.49364942e-01 -1.47604987e-01
-6.16275430e-01 -5.93675911e-01 3.82855415e-01 7.71039128e-01
-1.19981810e-01 -2.44906008e-01 1.24611676e-01 -1.19938217e-01
-9.80649590e-02 -1.17542617e-01 2.53022499e-02 1.07516837e+00
-6.16101921e-01 -1.13426709e+00 -4.21306998e-01 -1.61843970e-02
1.36998609e-01 2.05207869e-01 -1.26803711e-01 1.24705589e+00
3.28219026e-01 8.74423623e-01 2.70688673e-03 -5.56420207e-01
5.05105793e-01 -2.99804583e-02 4.45787162e-01 -4.75442141e-01
-2.99324512e-01 2.72808492e-01 -2.23081738e-01 -7.55017757e-01
-6.39430761e-01 -1.00593364e+00 -1.44951129e+00 -2.83279955e-01
-5.30180454e-01 -9.70643312e-02 7.68363953e-01 8.19567561e-01
4.97943461e-01 3.39279741e-01 5.68936765e-01 -1.18438733e+00
-5.36654174e-01 -6.69093728e-01 -1.27225399e-01 2.94915885e-01
5.43582380e-01 -1.16936052e+00 -2.93327034e-01 6.54813573e-02] | [6.603890419006348, -2.2706778049468994] |
75b3c15b-121d-4a3f-b5a2-f166e82c2ead | an-analysis-of-gpt-3-s-performance-in | 2303.14342 | null | https://arxiv.org/abs/2303.14342v2 | https://arxiv.org/pdf/2303.14342v2.pdf | Analyzing the Performance of GPT-3.5 and GPT-4 in Grammatical Error Correction | GPT-3 and GPT-4 models are powerful, achieving high performance on a variety of Natural Language Processing tasks. However, there is a relative lack of detailed published analysis of their performance on the task of grammatical error correction (GEC). To address this, we perform experiments testing the capabilities of a GPT-3.5 model (text-davinci-003) and a GPT-4 model (gpt-4-0314) on major GEC benchmarks. We compare the performance of different prompts in both zero-shot and few-shot settings, analyzing intriguing or problematic outputs encountered with different prompt formats. We report the performance of our best prompt on the BEA-2019 and JFLEG datasets, finding that the GPT models can perform well in a sentence-level revision setting, with GPT-4 achieving a new high score on the JFLEG benchmark. Through human evaluation experiments, we compare the GPT models' corrections to source, human reference, and baseline GEC system sentences and observe differences in editing strategies and how they are scored by human raters. | ['Kentaro Inui', 'Michael Zock', 'Diana Galvan-Sosa', 'Keisuke Sakaguchi', 'Steven Coyne'] | 2023-03-25 | null | null | null | null | ['grammatical-error-correction'] | ['natural-language-processing'] | [ 1.18694186e-01 2.52307862e-01 4.43863004e-01 -5.09520590e-01
-1.21623182e+00 -3.75911832e-01 5.10342777e-01 9.43384886e-01
-7.43732274e-01 5.27917266e-01 4.05246556e-01 -5.05522728e-01
-1.67327300e-02 -2.42804438e-01 -7.20770836e-01 2.41265401e-01
8.24205652e-02 7.23321497e-01 3.36886853e-01 -8.05095077e-01
7.23483682e-01 -2.69744545e-01 -1.08363998e+00 5.53126633e-01
1.33846426e+00 3.45310688e-01 2.50225544e-01 1.04311073e+00
1.43474087e-01 6.21496916e-01 -9.25985515e-01 -1.05635166e+00
-8.49299505e-02 -2.84941643e-01 -1.16485202e+00 -6.65251911e-01
7.73875475e-01 -6.98881503e-03 2.52726227e-01 1.06646466e+00
8.14546406e-01 8.05314630e-02 6.05596185e-01 -7.71672547e-01
-1.04566622e+00 9.93865728e-01 -2.64643669e-01 6.13461792e-01
8.03674936e-01 4.42991436e-01 1.07161522e+00 -1.01071882e+00
9.54654336e-01 1.40454245e+00 1.05253530e+00 7.47824311e-01
-1.21580470e+00 -1.34516910e-01 1.09625675e-01 2.32573926e-01
-9.78947818e-01 -6.79496109e-01 -6.22002967e-02 -2.66776562e-01
2.13760567e+00 2.19735727e-01 1.48849443e-01 1.25614893e+00
4.42686439e-01 5.32692373e-01 7.83883095e-01 -8.95989120e-01
-3.22917290e-02 9.06414911e-03 6.53137088e-01 7.46419728e-01
2.82102406e-01 -6.65285513e-02 -9.17306185e-01 -1.44009367e-01
-3.41640562e-01 -7.52873600e-01 -6.04885221e-01 7.23914921e-01
-1.05888343e+00 5.70165396e-01 -5.79100884e-02 2.50843525e-01
-2.56725699e-01 1.43317178e-01 7.57230937e-01 7.69794226e-01
8.24022889e-01 1.03632891e+00 -7.38790929e-01 -7.20142543e-01
-8.19028378e-01 6.37810946e-01 9.76182520e-01 1.36314690e+00
9.35029313e-02 -1.17131017e-01 -8.68112385e-01 1.08833015e+00
2.13311091e-01 5.10278761e-01 5.11654258e-01 -8.07693064e-01
1.18267083e+00 2.72254109e-01 1.45428970e-01 -1.02516425e+00
-5.07264435e-01 -2.84293383e-01 -3.23300272e-01 -8.33372846e-02
5.43668509e-01 -1.83015376e-01 -7.25955069e-01 1.73118353e+00
-3.86192381e-01 -6.07929051e-01 4.96272780e-02 3.58188152e-01
8.38231206e-01 6.26593113e-01 3.80833447e-01 -4.68943901e-02
1.16413462e+00 -9.15793955e-01 -7.02758014e-01 -5.64521074e-01
1.23673797e+00 -1.10754693e+00 1.63421738e+00 3.52780044e-01
-1.32407665e+00 -4.80908871e-01 -9.23407853e-01 -5.39086938e-01
-2.20223516e-01 3.14522237e-01 7.93423876e-02 4.45347637e-01
-1.29230726e+00 1.01513803e+00 -7.58665025e-01 -9.43264544e-01
-1.68121934e-01 -3.34194928e-01 -3.43073994e-01 -5.65816239e-02
-1.15805399e+00 1.50336599e+00 3.65858018e-01 -1.61342308e-01
-4.22580808e-01 -9.79014754e-01 -7.06499219e-01 1.54322132e-01
1.12126902e-01 -6.01099968e-01 1.93901110e+00 -1.18816063e-01
-1.11409998e+00 9.75018561e-01 -2.71632910e-01 -6.99483633e-01
5.68045795e-01 -5.45537889e-01 -6.99208736e-01 -2.13003129e-01
4.22060639e-01 5.32841980e-01 3.67125809e-01 -6.77844942e-01
-7.60296941e-01 -1.43470719e-01 -2.62776852e-01 2.16670603e-01
2.67626438e-02 6.51454210e-01 -1.82445943e-01 -7.75143385e-01
-3.22175473e-01 -6.89342499e-01 3.45155895e-02 -3.16009581e-01
-6.42998159e-01 -4.87509608e-01 2.92778134e-01 -1.25583220e+00
1.69931650e+00 -2.09658670e+00 2.80851256e-02 -1.57178238e-01
-7.35942572e-02 5.35426199e-01 -4.65152502e-01 6.85960054e-01
9.89370793e-02 5.81787288e-01 -4.97535199e-01 -8.26010168e-01
3.64203081e-02 -8.38665385e-03 -1.98710397e-01 -1.52907833e-01
3.83815765e-01 1.03705728e+00 -1.04145372e+00 -3.78294706e-01
-3.75027567e-01 -1.48713678e-01 -8.22028339e-01 -7.07156062e-02
-3.32447797e-01 4.90545630e-02 2.08419085e-01 5.56439102e-01
8.42772350e-02 -1.20481670e-01 -6.36356845e-02 1.96295694e-01
-7.21234083e-02 8.06898355e-01 -5.75570703e-01 1.67920005e+00
-1.81915089e-01 5.99835992e-01 -2.85149395e-01 -1.95828483e-01
7.14525163e-01 3.65369201e-01 -5.17161071e-01 -1.01292825e+00
-9.27144513e-02 4.99947160e-01 2.20694229e-01 -5.58504164e-01
1.25157750e+00 1.43808173e-03 -3.62586886e-01 6.72688186e-01
5.22870839e-01 -1.94711924e-01 8.05443466e-01 7.09976077e-01
1.41920245e+00 -4.07852232e-01 2.62230307e-01 -4.57904726e-01
4.83803749e-02 2.62237221e-01 5.04432499e-01 1.42715263e+00
-3.06505799e-01 8.21591973e-01 5.34011602e-01 8.04906487e-02
-1.01306140e+00 -7.17052817e-01 1.05837531e-01 1.23427224e+00
-3.22843462e-01 -1.10136449e+00 -9.03597295e-01 -7.53109038e-01
1.12289049e-01 1.56456041e+00 -5.44021249e-01 -5.56502521e-01
-6.83784246e-01 -6.53849125e-01 9.69965696e-01 3.83801788e-01
1.62169144e-01 -1.22047925e+00 -5.04012823e-01 5.56915700e-01
-5.33320129e-01 -9.24163342e-01 -7.46926546e-01 1.11613005e-01
-6.81334496e-01 -1.04684389e+00 -3.38257879e-01 -5.88740110e-01
3.19554120e-01 -1.07665747e-01 1.68184733e+00 3.88977796e-01
-6.24558143e-02 5.79436600e-01 -8.56265306e-01 -8.19210589e-01
-1.03445899e+00 3.57116580e-01 -3.57514769e-02 -7.87072420e-01
4.33270842e-01 6.43667416e-04 6.02995120e-02 -1.09089531e-01
-3.44638109e-01 8.24520364e-02 2.02359051e-01 8.56555939e-01
8.76236185e-02 -5.80627978e-01 5.92679322e-01 -1.35969651e+00
1.39637172e+00 -3.32265824e-01 -1.68608099e-01 6.50971532e-01
-1.05487478e+00 1.03729919e-01 4.82294589e-01 1.68685451e-01
-1.18025863e+00 -7.81190932e-01 -4.75425243e-01 2.85325408e-01
2.32510522e-01 7.65179336e-01 2.27307275e-01 2.26750001e-01
1.05936742e+00 -4.55581248e-02 -2.77424067e-01 -8.83554816e-01
2.58678049e-01 6.19738936e-01 8.95943105e-01 -8.75916481e-01
4.09899861e-01 -5.67696154e-01 -7.36773431e-01 -3.91092181e-01
-1.08593190e+00 -1.26606613e-01 -3.08779836e-01 2.46346835e-02
6.04803562e-01 -6.85337424e-01 -2.10815296e-02 6.53758645e-01
-1.82438362e+00 -3.01116884e-01 -2.48517439e-01 1.05305739e-01
-3.61732483e-01 4.65059519e-01 -9.59214270e-01 -4.45015937e-01
-9.15046096e-01 -9.49608743e-01 1.01635778e+00 4.90206555e-02
-7.70280421e-01 -8.81518841e-01 2.85387546e-01 1.23006098e-01
5.96782207e-01 -2.48864070e-01 1.43642509e+00 -8.81554425e-01
2.09795800e-03 -7.31258616e-02 4.02339501e-03 4.31470841e-01
-3.94919425e-01 1.80367962e-01 -5.34433961e-01 -3.80886614e-01
-3.16549331e-01 -3.47215444e-01 8.81896496e-01 -3.50646414e-02
8.14834356e-01 -3.40909928e-01 -4.29174080e-02 2.99672633e-01
1.11122870e+00 5.84298186e-02 5.75128257e-01 3.57805818e-01
3.91735196e-01 3.76998693e-01 6.88742161e-01 2.16641426e-01
4.62787271e-01 6.07253075e-01 5.31508140e-02 5.30353904e-01
-3.98351759e-01 -5.11879146e-01 3.55009913e-01 1.21137905e+00
3.17146704e-02 -7.66016901e-01 -1.16119254e+00 6.66068614e-01
-1.84270179e+00 -8.06612670e-01 -5.63669264e-01 2.05640054e+00
1.08776963e+00 4.10358518e-01 -3.60224843e-01 -1.17662782e-02
6.00333929e-01 -1.34477109e-01 -2.00578049e-01 -1.08663106e+00
-3.87195915e-01 3.10315102e-01 1.66027442e-01 4.50771391e-01
-7.04191566e-01 1.03297603e+00 7.26226997e+00 5.37998378e-01
-6.42973542e-01 3.73102248e-01 3.89435291e-01 -1.81743428e-01
-1.50197655e-01 -4.62421440e-02 -1.18361008e+00 8.67643237e-01
1.41883743e+00 -6.19017839e-01 2.58693248e-01 5.07491827e-01
2.75236756e-01 -1.94026321e-01 -1.38183999e+00 7.18296230e-01
2.58577704e-01 -1.29101479e+00 1.25744969e-01 -5.13565302e-01
6.37277126e-01 2.79225856e-01 -2.88762331e-01 9.46009934e-01
4.68349963e-01 -7.82430291e-01 1.15959775e+00 7.06658185e-01
7.15429962e-01 -2.74848044e-01 6.45746112e-01 4.17869776e-01
-3.38989198e-01 4.91570085e-02 -5.31958759e-01 -1.41463995e-01
4.83434439e-01 6.40961885e-01 -8.47178102e-01 5.80095112e-01
9.18764889e-01 8.35857272e-01 -1.25433505e+00 1.13973498e+00
-5.55934727e-01 9.35238600e-01 -1.25100717e-01 -8.55498537e-02
-1.53002828e-01 4.15997624e-01 9.83422995e-01 1.66510916e+00
4.59661990e-01 -5.54007925e-02 -2.62216598e-01 7.54586816e-01
-4.19960350e-01 2.02535197e-01 -2.71315455e-01 -7.00921565e-02
8.23262990e-01 9.27394688e-01 -4.95500676e-02 -5.15952408e-01
-2.52220184e-01 1.02497196e+00 8.38836670e-01 8.03252012e-02
-5.15167475e-01 -7.46213257e-01 3.77043217e-01 -2.28164531e-02
-9.77723226e-02 -4.65052016e-02 -3.12156081e-01 -1.20370770e+00
3.60394865e-01 -1.28306985e+00 4.07447010e-01 -1.08139265e+00
-1.52765417e+00 6.37723029e-01 -2.58257929e-02 -8.71614516e-01
-2.82687575e-01 -4.78585124e-01 -9.32616472e-01 1.12124896e+00
-1.30985785e+00 -5.71685314e-01 -2.09950581e-02 1.33256480e-01
6.94392383e-01 -1.44996181e-01 7.63984561e-01 4.55687851e-01
-6.16467237e-01 1.04804897e+00 1.21222809e-01 -8.44882894e-03
1.20923591e+00 -1.54162860e+00 1.18744862e+00 1.33074975e+00
-3.92048806e-02 9.20346022e-01 9.13782179e-01 -9.24147904e-01
-7.78117776e-01 -1.26942575e+00 1.75144553e+00 -8.09435546e-01
4.89692777e-01 -2.08573252e-01 -1.24915433e+00 8.35383773e-01
3.65478814e-01 -4.26321656e-01 2.67300934e-01 5.35610437e-01
-4.12161112e-01 3.45283389e-01 -1.13417256e+00 4.02413398e-01
1.21784627e+00 -5.94379485e-01 -1.08343554e+00 4.95074511e-01
1.05301380e+00 -7.65855074e-01 -9.14914310e-01 2.44520321e-01
1.03372566e-01 -7.43954301e-01 3.87364566e-01 -9.44763541e-01
8.47263157e-01 -1.72076114e-02 3.56902443e-02 -2.00203586e+00
-5.01019299e-01 -5.25785327e-01 1.84306040e-01 1.56458163e+00
9.79491174e-01 -5.94914913e-01 -1.77493498e-01 7.20070064e-01
-8.50963414e-01 -4.87915099e-01 -8.43511283e-01 -9.15744424e-01
2.48366684e-01 -4.17868972e-01 3.81869137e-01 8.02790999e-01
2.62919188e-01 4.90529239e-01 -1.69707254e-01 -2.29483217e-01
1.78210348e-01 -5.58301032e-01 5.13907492e-01 -1.24134195e+00
-5.61874509e-01 -4.28035945e-01 2.35706642e-02 -5.75570285e-01
-1.61714807e-01 -1.20551324e+00 3.66171449e-01 -1.52332342e+00
2.39632323e-01 -1.36555210e-01 8.25475231e-02 7.79932678e-01
-6.83785260e-01 -2.33001873e-01 5.23349345e-01 2.63746768e-01
-7.47138441e-01 5.38729072e-01 7.11847961e-01 -1.16966978e-01
-5.07770851e-02 -4.51757193e-01 -1.01873267e+00 3.47141773e-01
6.33128762e-01 -6.61270499e-01 8.53980035e-02 -1.29257202e+00
6.12792730e-01 -1.43620208e-01 1.64190665e-01 -1.11921382e+00
3.49473596e-01 3.96894932e-01 -8.63577202e-02 -3.98454726e-01
-3.44654471e-01 1.51927143e-01 -1.73589975e-01 2.42823869e-01
-8.10290098e-01 7.01176405e-01 3.91370803e-01 4.40270871e-01
-4.72179465e-02 -6.81757033e-01 7.15945542e-01 -2.13404343e-01
-6.66984320e-01 -2.27593511e-01 -5.21354675e-01 7.15743363e-01
2.99767137e-01 1.05247512e-01 -1.05992520e+00 -1.03930227e-01
-4.36584681e-01 5.12826622e-01 4.24745470e-01 7.52641618e-01
3.96203727e-01 -7.73715436e-01 -1.08050144e+00 5.59640341e-02
6.76856995e-01 -4.04747099e-01 -1.19671104e-02 8.86218250e-01
-4.38174665e-01 5.13469577e-01 -9.15065110e-02 -4.33796972e-01
-1.29031801e+00 2.94607561e-02 4.01766270e-01 -5.17492712e-01
-6.60341620e-01 1.05753613e+00 -5.55453897e-01 -9.17041540e-01
1.15875944e-01 -3.99680495e-01 3.38842464e-03 -3.90839279e-01
5.79808056e-01 4.74547952e-01 7.92303920e-01 -2.48280615e-01
-1.82331711e-01 1.09750196e-01 -6.35599494e-01 -6.01616688e-02
1.13275707e+00 -2.46161982e-01 -1.55237034e-01 3.68777633e-01
8.42561305e-01 -1.09802864e-01 -3.74798149e-01 -3.32019061e-01
6.21565282e-01 -2.46413544e-01 -2.17595562e-01 -1.62177503e+00
-3.13650161e-01 7.75332332e-01 2.93531179e-01 -6.15577288e-02
7.72171557e-01 -1.27636358e-01 7.62018681e-01 6.41757488e-01
5.10000706e-01 -1.37305367e+00 -1.46785513e-01 1.32782030e+00
1.20515633e+00 -1.22186577e+00 -3.82444143e-01 -1.87879100e-01
-6.77936733e-01 8.71169031e-01 9.48099494e-01 2.66046047e-01
1.52220398e-01 -1.60746962e-01 -6.72146976e-02 -4.05162424e-01
-1.17043293e+00 2.46536061e-01 9.55266133e-02 3.76308054e-01
8.51859391e-01 -1.83913317e-02 -8.64033878e-01 8.51250291e-01
-5.61814904e-01 -2.15867937e-01 9.09226060e-01 1.02135730e+00
-4.23864603e-01 -1.10343206e+00 1.28727868e-01 7.78228998e-01
-4.72215086e-01 -5.60834706e-01 -4.76866871e-01 6.94314182e-01
-2.44468033e-01 1.26777995e+00 -1.02685474e-01 -1.95310920e-01
8.76262605e-01 3.57515574e-01 5.19831955e-01 -1.19313467e+00
-1.20775306e+00 -5.39911389e-01 7.07633555e-01 -5.50964057e-01
2.32465371e-01 -9.09137011e-01 -9.70052660e-01 -3.93879712e-01
-2.59579360e-01 3.43331814e-01 4.68844235e-01 8.94922733e-01
7.29418099e-01 4.35013413e-01 -1.69406787e-01 -2.52092987e-01
-9.85665143e-01 -1.55809307e+00 -3.78299594e-01 6.45740151e-01
-9.19852406e-02 -2.17893705e-01 -2.70468593e-01 -3.45870070e-02] | [11.065794944763184, 10.727232933044434] |
3dcf48ec-77f0-4468-a880-1a256f6e454a | lqr-trees-with-sampling-based-exploration-of | 2303.00553 | null | https://arxiv.org/abs/2303.00553v1 | https://arxiv.org/pdf/2303.00553v1.pdf | LQR-trees with Sampling Based Exploration of the State Space | This paper introduces an extension of the LQR-tree algorithm, which is a feedback-motion-planning algorithm for stabilizing a system of ordinary differential equations from a bounded set of initial conditions to a goal. The constructed policies are represented by a tree of exemplary system trajectories, so called demonstrations, and linear-quadratic regulator (LQR) feedback controllers. Consequently, the crucial component of any LQR-tree algorithm is a demonstrator that provides suitable demonstrations. In previous work, such a demonstrator was given by a local trajectory optimizer. However, these require appropriate initial guesses of solutions to provide valid results, which was pointed out, but largely unresolved in previous implementations. In this paper, we augment the LQR-tree algorithm with a randomized motion-planning procedure to discover new valid demonstration candidates to initialize the demonstrator in parts of state space not yet covered by the LQR-tree. In comparison to the previous versions of the LQR-tree algorithm, the resulting exploring LQR-tree algorithm reliably synthesizes feedback control laws for a far more general set of problems. | ['Stefan Ratschan', 'Jiří Fejlek'] | 2023-03-01 | null | null | null | null | ['motion-planning'] | ['robots'] | [-4.46567163e-02 5.28633833e-01 -4.31663036e-01 1.54396236e-01
-7.13917375e-01 -8.30577731e-01 7.24711537e-01 -1.15891330e-01
-2.10915625e-01 1.17658007e+00 -8.66957828e-02 -6.29517138e-01
-4.13120866e-01 -3.31847548e-01 -7.94743717e-01 -7.72349715e-01
-3.06457728e-01 3.48886669e-01 4.39919412e-01 -5.53352773e-01
3.26653689e-01 7.45157599e-01 -1.29262280e+00 -3.38260740e-01
8.94637942e-01 5.03630459e-01 3.30289543e-01 9.78282273e-01
3.97688508e-01 5.17812073e-01 -3.40483785e-01 4.81942356e-01
5.30669749e-01 -6.49685979e-01 -7.30205894e-01 2.69276261e-01
8.08905885e-02 -1.23160154e-01 -3.94776255e-01 1.08984208e+00
3.24757576e-01 5.30085564e-01 6.25861526e-01 -1.59200788e+00
-1.14577189e-01 2.83382595e-01 8.12459812e-02 -1.03092588e-01
4.77263719e-01 6.72684431e-01 6.73101723e-01 -7.12489367e-01
1.01416051e+00 1.41485679e+00 5.52807212e-01 8.08461666e-01
-1.48558605e+00 -2.75719941e-01 2.04105720e-01 -1.25757739e-01
-1.35094690e+00 -4.24773276e-01 4.73263055e-01 -6.52848244e-01
8.38464141e-01 4.78579879e-01 8.68807018e-01 7.54193008e-01
4.93676066e-01 6.13053322e-01 9.50160384e-01 -3.23932558e-01
5.93248725e-01 1.41748309e-01 -9.78703573e-02 8.62669110e-01
1.28893957e-01 6.00816190e-01 -7.19748512e-02 -2.95846611e-01
1.10459208e+00 -4.37278688e-01 -5.42311013e-01 -1.08035529e+00
-1.34960115e+00 7.17646897e-01 4.68208820e-01 1.30513147e-01
-5.20359814e-01 3.51985008e-01 1.54959425e-01 7.22296298e-01
-7.27899596e-02 9.96273398e-01 -3.98773491e-01 2.53265370e-02
-6.18012667e-01 7.68385470e-01 8.42753649e-01 1.09229362e+00
6.31180882e-01 4.48884189e-01 -2.53727913e-01 -3.89661714e-02
2.93922633e-01 3.38352233e-01 1.67470977e-01 -1.33878219e+00
2.75130808e-01 5.91269910e-01 8.91235054e-01 -6.59919381e-01
-3.52383018e-01 -1.70932695e-01 -3.58638167e-01 8.10333729e-01
4.56884503e-01 -3.15672070e-01 -6.64285660e-01 1.61160576e+00
7.05879092e-01 -4.39308397e-02 3.07514429e-01 1.17111766e+00
-5.48964785e-03 9.15485978e-01 -5.22039235e-01 -6.51926279e-01
5.87463260e-01 -8.96033227e-01 -6.59453750e-01 1.23403236e-01
7.22205341e-01 -3.11313242e-01 1.09953451e+00 4.19227540e-01
-9.87136483e-01 -5.85650086e-01 -1.00955415e+00 4.40655351e-01
-3.01765986e-02 4.64654639e-02 7.03354627e-02 2.06494197e-01
-1.36582375e+00 1.14517987e+00 -8.08859944e-01 -4.65236425e-01
-4.86390889e-01 5.02647102e-01 -2.19310492e-01 3.85404915e-01
-1.04378712e+00 1.30381131e+00 3.98066640e-01 2.35153228e-01
-1.39377010e+00 -6.01460814e-01 -8.52932274e-01 -3.68004322e-01
7.21800089e-01 -5.59228182e-01 1.62830520e+00 -6.11793041e-01
-1.93349612e+00 2.53846407e-01 1.28935929e-02 -4.40076888e-01
6.17053211e-01 -1.61449909e-01 4.18394879e-02 -1.00322263e-02
-8.14734399e-03 6.31118059e-01 1.05390286e+00 -1.30475092e+00
-3.69268596e-01 6.02121130e-02 1.80302456e-01 2.57794261e-01
3.71228993e-01 -1.43455761e-02 -9.53355059e-02 -2.85330981e-01
-1.45062521e-01 -1.22162676e+00 -9.25054193e-01 7.57023841e-02
-3.59337568e-01 -2.53094316e-01 9.18492615e-01 -1.27981067e-01
1.25464010e+00 -1.80495250e+00 8.81826878e-01 1.53060317e-01
-1.23496056e-01 4.34265882e-01 -9.81945321e-02 9.21952546e-01
-1.12218931e-01 -2.84281485e-02 -1.18074812e-01 8.83083194e-02
5.03082201e-02 3.00920397e-01 -6.82311714e-01 8.84030879e-01
1.80966735e-01 7.85494685e-01 -1.18649042e+00 -3.40675324e-01
3.97239685e-01 -5.32323606e-02 -4.48796511e-01 3.67857933e-01
-7.07069814e-01 9.63751495e-01 -8.84205282e-01 4.10683721e-01
-9.77173913e-03 2.48593703e-01 -1.23695061e-01 2.44104728e-01
-8.80935729e-01 1.12118058e-01 -1.41389275e+00 1.42518723e+00
-2.23885924e-01 4.73181725e-01 6.13891184e-01 -9.73731220e-01
1.21188712e+00 4.80181247e-01 6.03560150e-01 4.38977731e-03
2.77292460e-01 4.14213002e-01 4.43308242e-02 -5.48967421e-01
4.02232468e-01 -1.46729261e-01 -1.41706154e-01 1.09624416e-01
-2.20435098e-01 -7.99734354e-01 2.35481501e-01 1.89042129e-02
1.17604566e+00 6.75806284e-01 5.53066373e-01 -7.12014973e-01
9.41303551e-01 6.10226274e-01 6.81603789e-01 5.13805985e-01
-4.58262473e-01 3.21703285e-01 4.43603545e-01 -2.96964228e-01
-1.18520916e+00 -6.42621815e-01 1.29808336e-01 5.02664864e-01
2.40355328e-01 -4.97327775e-01 -7.36371398e-01 -4.18731242e-01
1.17880747e-01 8.45430195e-01 -4.05251503e-01 -1.60170019e-01
-8.17393601e-01 1.81715548e-01 1.79252625e-01 1.55223206e-01
-3.60563248e-02 -1.00681925e+00 -1.22254395e+00 5.46759129e-01
2.90062994e-01 -6.12503171e-01 -5.88186026e-01 3.04801971e-01
-9.04737711e-01 -1.24230301e+00 -7.22770751e-01 -6.17417455e-01
7.87224710e-01 -3.97711173e-02 5.95472038e-01 -5.36390804e-02
-1.97909892e-01 4.94709820e-01 -2.09091201e-01 -1.85623437e-01
-8.82548034e-01 -3.76910686e-01 5.38190663e-01 -2.10017368e-01
-6.45047724e-01 -3.08381557e-01 -2.78294772e-01 6.04990363e-01
-4.79047835e-01 -3.20447646e-02 3.29627573e-01 8.40214491e-01
7.49948561e-01 2.11123517e-03 5.81985891e-01 -1.47361130e-01
8.74042809e-01 -9.49641839e-02 -1.28793633e+00 1.08279869e-01
-4.12180662e-01 4.41565722e-01 1.12440133e+00 -7.16458678e-01
-7.12172925e-01 6.15419209e-01 1.63890466e-01 -8.48037601e-01
-4.21528928e-02 1.69412404e-01 -6.38501272e-02 -3.03835630e-01
8.62668037e-01 1.46546900e-01 3.76105219e-01 -2.22988054e-01
8.01465929e-01 3.67367774e-01 4.93441433e-01 -6.17024481e-01
1.03326380e+00 -1.85484178e-02 3.69721562e-01 -7.35753357e-01
-3.56523603e-01 -5.79428196e-01 -7.88286924e-01 -4.91109371e-01
6.92475438e-01 -4.18062925e-01 -8.37065279e-01 2.57081930e-02
-1.06232643e+00 -8.34504664e-01 -7.74088144e-01 4.40577328e-01
-1.24208748e+00 1.52057605e-02 -9.45363790e-02 -1.30010140e+00
4.55931686e-02 -1.51679599e+00 1.04201984e+00 1.05511650e-01
-4.40117925e-01 -7.74216413e-01 5.00266254e-01 -4.88512456e-01
2.15386093e-01 6.00861371e-01 6.17465377e-01 -1.44734055e-01
-6.80684984e-01 -1.87787443e-01 5.29909730e-01 1.49507634e-02
-8.89313817e-02 1.64806828e-01 -4.31458861e-01 -7.85534620e-01
1.44039974e-01 -4.02102619e-01 3.94799799e-01 5.12771308e-01
3.83826017e-01 -5.01606047e-01 -7.17469990e-01 3.06931168e-01
1.33994544e+00 7.60892212e-01 2.11252764e-01 4.41530854e-01
2.97060311e-01 6.87379241e-01 1.11742067e+00 8.93384516e-02
1.65309049e-02 8.29691648e-01 4.80698436e-01 9.64438841e-02
1.54351175e-01 -4.14398104e-01 7.80968785e-01 3.52147937e-01
1.49075747e-01 2.12729037e-01 -7.07732737e-01 6.52196646e-01
-2.31504250e+00 -7.33893454e-01 -3.70326400e-01 2.25393891e+00
6.32600367e-01 -1.40419647e-01 4.99449193e-01 4.18435074e-02
6.46091998e-01 -6.91044563e-03 -9.83218133e-01 -6.03005886e-01
3.68879795e-01 -3.51823151e-01 5.38024962e-01 8.51614594e-01
-9.74065185e-01 8.68741572e-01 6.88201427e+00 3.99914086e-01
-8.87669563e-01 -3.39171052e-01 -7.25528821e-02 1.79250076e-01
-1.37513191e-01 3.84860903e-01 -8.78162026e-01 1.47574961e-01
9.49461997e-01 -6.91551983e-01 6.50635779e-01 1.05076694e+00
8.69777620e-01 -1.25523105e-01 -1.29651296e+00 5.38055480e-01
-4.09607202e-01 -1.12930942e+00 -5.66424489e-01 -1.29457656e-02
8.72392654e-01 -5.06359279e-01 -1.09360792e-01 5.26636720e-01
6.36236906e-01 -7.37589121e-01 9.23681974e-01 3.86950344e-01
4.93604302e-01 -5.81979454e-01 1.22683205e-01 9.05628920e-01
-1.20282543e+00 -4.26280886e-01 -3.66840899e-01 -8.96011218e-02
4.32050765e-01 4.58793268e-02 -8.29816639e-01 6.36956573e-01
9.59902778e-02 4.38020200e-01 -1.13542713e-01 1.18099320e+00
-4.72288251e-01 2.99689829e-01 -3.08429331e-01 -5.32790065e-01
5.69528997e-01 -4.57487494e-01 1.16798389e+00 8.31635416e-01
2.43352741e-01 2.06750020e-01 6.19135618e-01 1.09000647e+00
7.32553899e-01 -6.84160814e-02 -1.13436019e+00 1.88296139e-01
5.03242798e-02 1.11294734e+00 -4.96657789e-01 -2.42359012e-01
2.92991370e-01 5.34860313e-01 1.75363630e-01 3.89183491e-01
-8.03131998e-01 -3.77608955e-01 7.27039516e-01 5.59060052e-02
2.92303026e-01 -6.81885362e-01 2.57413685e-01 -1.03067553e+00
-3.08045596e-01 -9.25388217e-01 -5.94302155e-02 -7.55534768e-01
-4.88090008e-01 4.41941828e-01 4.42704886e-01 -1.56649613e+00
-8.83348346e-01 -3.49649161e-01 -5.83775163e-01 8.54553759e-01
-1.10335696e+00 -5.44665277e-01 1.88618556e-01 4.67252344e-01
8.18630457e-01 -4.00008372e-04 6.87614262e-01 -3.97224665e-01
-4.59253192e-01 -2.43915796e-01 2.51922488e-01 -5.91505706e-01
2.99698293e-01 -1.58825576e+00 4.21744317e-01 1.00710142e+00
-2.38769084e-01 6.57543123e-01 1.37013638e+00 -7.92863190e-01
-1.92701793e+00 -1.19579995e+00 3.92668396e-01 -3.28700811e-01
1.00034857e+00 -2.12376773e-01 -8.00105035e-01 6.71182990e-01
8.44534189e-02 -1.52937084e-01 -5.58275282e-01 -5.91879070e-01
4.38845545e-01 7.65434951e-02 -9.54835236e-01 9.76529241e-01
9.20391023e-01 -1.17684575e-02 -6.70856774e-01 4.04980510e-01
8.75292957e-01 -6.99919224e-01 -7.17714071e-01 2.65510470e-01
1.56718418e-01 -4.43587661e-01 7.60199904e-01 -8.21794808e-01
-1.35970548e-01 -8.75637710e-01 2.60006785e-01 -1.81647742e+00
-1.99371666e-01 -1.58709574e+00 -3.40462983e-01 6.84524536e-01
3.26147139e-01 -3.86649847e-01 6.79746032e-01 4.07515734e-01
-4.65372860e-01 -9.18700874e-01 -9.85271573e-01 -1.37096179e+00
2.15484202e-01 -5.81660569e-02 1.36996076e-01 4.31235731e-01
3.13576430e-01 2.98400253e-01 -4.29168612e-01 2.30204284e-01
7.26291955e-01 3.37747484e-01 1.08646142e+00 -7.91279137e-01
-3.29563558e-01 -4.33620363e-01 -1.08912595e-01 -1.24485600e+00
2.83891380e-01 -6.87969029e-01 6.38651788e-01 -1.73564231e+00
-5.64971149e-01 -3.66038144e-01 3.18111509e-01 3.57321769e-01
6.27167523e-02 -6.77376211e-01 3.23175341e-01 2.55670905e-01
-4.17135805e-01 7.65746593e-01 1.56549489e+00 9.67912599e-02
-8.44155669e-01 4.57663178e-01 -7.82063007e-02 6.30767286e-01
7.45746374e-01 -2.85501629e-01 -7.57280111e-01 8.46988782e-02
-1.98173285e-01 8.55101466e-01 3.76954049e-01 -9.40540016e-01
3.54286581e-01 -6.99220359e-01 -3.46171141e-01 -6.41639233e-01
3.46068591e-01 -7.69451678e-01 4.15605962e-01 1.01642954e+00
-4.98460174e-01 2.74629831e-01 1.83839537e-02 6.87960744e-01
-6.16966672e-02 -2.93836653e-01 7.75792003e-01 -9.96462703e-02
-9.22162235e-01 1.18945464e-02 -7.53075123e-01 -2.78669894e-01
1.51297247e+00 -2.90619254e-01 -7.62328207e-02 -4.32924569e-01
-8.69383931e-01 6.86329663e-01 6.05201721e-01 2.35355496e-01
6.47496939e-01 -1.17754233e+00 -2.41688088e-01 2.29442772e-02
-2.80750871e-01 6.91510588e-02 -3.88690144e-01 9.12796974e-01
-2.94680208e-01 7.33683109e-01 -2.78546900e-01 -6.09236717e-01
-1.11226225e+00 1.05243468e+00 5.63675344e-01 -7.53342509e-02
-8.96356404e-01 3.60374659e-01 -7.51609802e-02 -5.58843732e-01
1.98798880e-01 -7.95932531e-01 -1.18719310e-01 -4.51961488e-01
1.87719613e-01 3.86885315e-01 -2.71419138e-01 -4.62975979e-01
-2.57383972e-01 6.64273083e-01 6.90686524e-01 -5.20117819e-01
1.11301446e+00 -5.02536781e-02 1.71389997e-01 5.62565446e-01
8.17933917e-01 -3.81014735e-01 -1.60293341e+00 2.79247314e-01
1.33883014e-01 -3.01102638e-01 -3.02790642e-01 -3.10438484e-01
-3.64354104e-01 4.42564398e-01 1.18096136e-01 3.72871041e-01
7.44874477e-01 -2.66242027e-01 2.87178010e-01 6.05180621e-01
9.50572670e-01 -8.21914136e-01 8.15199502e-03 6.00201309e-01
1.37422121e+00 -7.28937328e-01 -1.17117083e-02 -2.95972168e-01
-6.84412241e-01 1.26786983e+00 6.77118123e-01 -5.11060238e-01
4.01214093e-01 3.12562972e-01 -1.73662081e-01 5.13455132e-03
-1.04083395e+00 -3.06079566e-01 2.05636799e-01 6.46215200e-01
-1.34785280e-01 -1.91648707e-01 -5.53068995e-01 -1.11141510e-03
-8.68496969e-02 8.78483281e-02 6.26023531e-01 1.13770151e+00
-9.21413660e-01 -1.09109282e+00 -6.28299952e-01 -3.37023109e-01
3.70392323e-01 6.20674133e-01 -3.95535856e-01 1.22000444e+00
-2.60887891e-01 9.60861146e-01 -2.16787323e-01 -2.63369352e-01
8.40070665e-01 -4.47485819e-02 4.65828001e-01 -6.84713304e-01
-6.25890493e-01 1.81795612e-01 2.87125438e-01 -9.46187198e-01
-1.90038994e-01 -7.02397406e-01 -1.36760211e+00 5.29440530e-02
-4.46173042e-01 4.05229062e-01 4.60823745e-01 6.63294435e-01
1.50643289e-01 4.10109371e-01 7.55216718e-01 -1.25174797e+00
-1.31917143e+00 -6.74288034e-01 -3.30900311e-01 -1.44062907e-01
6.59701943e-01 -8.76105011e-01 -2.28873342e-01 6.46760687e-02] | [4.909358024597168, 2.029085397720337] |
6cf79960-756d-4082-a3b6-6ce9c3873003 | diffconv-analyzing-irregular-point-clouds | 2111.14658 | null | https://arxiv.org/abs/2111.14658v3 | https://arxiv.org/pdf/2111.14658v3.pdf | diffConv: Analyzing Irregular Point Clouds with an Irregular View | Standard spatial convolutions assume input data with a regular neighborhood structure. Existing methods typically generalize convolution to the irregular point cloud domain by fixing a regular "view" through e.g. a fixed neighborhood size, where the convolution kernel size remains the same for each point. However, since point clouds are not as structured as images, the fixed neighbor number gives an unfortunate inductive bias. We present a novel graph convolution named Difference Graph Convolution (diffConv), which does not rely on a regular view. diffConv operates on spatially-varying and density-dilated neighborhoods, which are further adapted by a learned masked attention mechanism. Experiments show that our model is very robust to the noise, obtaining state-of-the-art performance in 3D shape classification and scene understanding tasks, along with a faster inference speed. | ['Aasa Feragen', 'Manxi Lin'] | 2021-11-29 | null | null | null | null | ['3d-shape-retrieval', '3d-object-classification'] | ['computer-vision', 'computer-vision'] | [-9.39245597e-02 6.53883144e-02 1.63940579e-01 -4.71721917e-01
-3.72202247e-02 -6.38620973e-01 7.26011634e-01 5.99764176e-02
-2.55369157e-01 3.29862982e-01 -3.25631686e-02 -5.64904392e-01
-1.60579812e-02 -1.38249612e+00 -1.18463421e+00 -6.50854707e-01
1.41197052e-02 3.73062372e-01 3.39905947e-01 6.68095201e-02
2.51656413e-01 9.94284570e-01 -1.35365760e+00 3.85643452e-01
7.09835947e-01 9.61421490e-01 3.97017241e-01 6.76934600e-01
-5.02101243e-01 2.90197402e-01 -3.21509123e-01 -2.99785614e-01
3.87879431e-01 5.69091178e-02 -9.07364368e-01 1.46297336e-01
8.22114706e-01 -1.65972427e-01 -6.19130850e-01 1.14567530e+00
1.66076154e-01 3.89820188e-01 4.96035278e-01 -1.01283157e+00
-1.39098454e+00 3.63867760e-01 -5.00956237e-01 2.35580817e-01
4.18346096e-03 1.68706983e-01 8.13700199e-01 -1.06541944e+00
4.82728511e-01 1.38841856e+00 8.35514843e-01 5.40378451e-01
-1.53102720e+00 -4.39622819e-01 7.27850378e-01 -2.05487281e-01
-1.50914717e+00 6.54302016e-02 7.30674088e-01 -3.97715926e-01
1.20612419e+00 1.01376601e-01 6.34761751e-01 8.11044455e-01
7.69860521e-02 2.99768358e-01 9.08124268e-01 -1.67526677e-01
3.40133756e-01 -1.12019852e-01 -3.61748487e-02 6.84227407e-01
2.71921128e-01 1.00575387e-01 -1.06324986e-01 -5.98111264e-02
1.38411295e+00 3.28035742e-01 -3.98353308e-01 -6.65693820e-01
-1.33087027e+00 8.38762701e-01 1.14220333e+00 3.14588845e-01
-2.83100814e-01 3.31845522e-01 1.27524659e-01 4.22073185e-01
7.87626684e-01 1.89993784e-01 -3.98280323e-01 3.81935328e-01
-5.89944243e-01 1.89279780e-01 4.89500284e-01 1.21718132e+00
1.02583063e+00 6.23216331e-02 -7.00415894e-02 6.02548540e-01
4.60881479e-02 5.00574112e-01 1.85893551e-02 -8.25531542e-01
3.55734080e-01 6.64611280e-01 -2.05819190e-01 -1.01587832e+00
-3.49852979e-01 -3.52175057e-01 -9.93223011e-01 7.54012346e-01
3.71639669e-01 1.00646123e-01 -1.40414965e+00 1.72440457e+00
2.10162401e-01 7.19817579e-01 -1.39140010e-01 8.72962594e-01
1.06447244e+00 4.05721724e-01 -5.60100041e-02 4.72570479e-01
1.03580189e+00 -6.94339454e-01 -1.45594329e-01 -6.98807314e-02
4.46314901e-01 -4.05113667e-01 1.21008563e+00 -1.19183816e-01
-9.62086141e-01 -6.90119565e-01 -8.92625570e-01 -2.46277496e-01
-6.98139608e-01 -3.33206117e-01 7.52642453e-01 4.26006883e-01
-1.42203665e+00 8.67181599e-01 -8.95514309e-01 -5.37778854e-01
6.50745511e-01 3.45259041e-01 -5.26643574e-01 -1.93493173e-01
-5.93622148e-01 6.37981296e-01 5.86051419e-02 -1.58813685e-01
-4.58968461e-01 -1.08355415e+00 -1.07869077e+00 5.23554608e-02
4.24442589e-02 -1.10719156e+00 9.32978272e-01 -6.44767582e-01
-1.25918603e+00 1.03004336e+00 -3.68069142e-01 -4.54953581e-01
3.23199987e-01 -5.30087389e-02 -1.34555295e-01 -6.60408363e-02
9.44169089e-02 6.88199222e-01 9.06793714e-01 -1.43528640e+00
-2.08429411e-01 -5.67547083e-01 3.62945795e-01 3.56788576e-01
1.27699047e-01 -3.66124332e-01 -4.76333052e-01 -6.87291384e-01
4.49119925e-01 -7.79428303e-01 -4.05548781e-01 3.92655790e-01
-2.33976483e-01 -2.50415385e-01 1.14892888e+00 -2.34361421e-02
7.16702580e-01 -2.26283288e+00 1.50473351e-02 3.85522842e-01
4.92564201e-01 2.70960294e-02 -2.23731995e-01 7.34018758e-02
-3.56770724e-01 2.57414073e-01 -4.25349325e-01 -3.83329749e-01
-2.00574368e-01 5.18106222e-01 -4.40670699e-01 4.62381899e-01
3.80898952e-01 1.08654881e+00 -1.04258323e+00 1.47725083e-03
5.43114603e-01 7.00410783e-01 -6.46179795e-01 7.85656869e-02
-3.07517856e-01 4.90335763e-01 -4.57410425e-01 4.39681441e-01
1.21763670e+00 -7.16836333e-01 -3.04122299e-01 -2.71228820e-01
-1.78625956e-01 2.87362307e-01 -1.20772171e+00 1.85440576e+00
-4.37001556e-01 4.02341276e-01 1.40212208e-01 -9.54602897e-01
9.23864543e-01 -1.00297481e-01 3.33229721e-01 -3.63037765e-01
-3.75283472e-02 -1.44714694e-02 -1.63070202e-01 -9.96873900e-03
2.91459471e-01 -8.80570561e-02 3.80141050e-01 3.65239739e-01
-4.47262377e-02 -3.78095418e-01 -4.58114892e-01 9.70958248e-02
1.14033413e+00 5.34308776e-02 -4.59393673e-02 -4.04368669e-01
2.61393547e-01 -1.67750418e-01 2.04149216e-01 8.91150057e-01
7.45299608e-02 1.03715253e+00 2.16052130e-01 -7.21990407e-01
-1.02242947e+00 -1.41909385e+00 -3.76094580e-01 6.80167377e-01
3.48335356e-01 -1.15918159e-01 -6.03367686e-01 -8.89484167e-01
5.74426591e-01 5.97970963e-01 -8.49429846e-01 2.13516857e-02
-6.20564640e-01 -3.68480593e-01 3.09719324e-01 9.04810369e-01
6.76164865e-01 -1.08310103e+00 -2.20283642e-01 1.04884719e-02
4.72951323e-01 -1.16670334e+00 -6.78908408e-01 1.80420473e-01
-1.17032969e+00 -1.00243926e+00 -6.07375383e-01 -8.83408308e-01
9.95980859e-01 7.43464112e-01 1.45772159e+00 2.58989185e-01
8.75490680e-02 4.96799827e-01 -1.19425096e-01 -3.28499794e-01
-1.34603214e-02 -3.39090601e-02 -8.57735574e-02 3.54494452e-02
5.74929416e-01 -9.82769132e-01 -5.76889515e-01 1.37819927e-02
-7.95917928e-01 -1.08525932e-01 3.45041037e-01 8.18167806e-01
9.43219185e-01 -2.99603671e-01 1.78356931e-01 -1.00853753e+00
4.71734852e-01 -4.39445913e-01 -7.48162210e-01 -3.42603512e-02
-3.90516430e-01 1.49694622e-01 6.95012867e-01 -4.11111355e-01
-7.98954070e-01 1.62728980e-01 -1.23862006e-01 -1.08958399e+00
-5.11790752e-01 1.66354448e-01 -8.19792449e-02 -5.18188536e-01
5.94475329e-01 8.04521590e-02 1.43896431e-01 -4.72614110e-01
6.77545369e-01 1.40043288e-01 5.44622183e-01 -4.96421963e-01
9.71460462e-01 9.04012084e-01 1.48114279e-01 -8.69505525e-01
-6.35474622e-01 -2.57944643e-01 -9.31659460e-01 1.25077680e-01
1.07819116e+00 -7.46193051e-01 -7.58766651e-01 3.91303569e-01
-1.30624986e+00 -6.23756766e-01 -6.61580503e-01 4.60030466e-01
-5.35133064e-01 2.09152699e-01 -4.83761102e-01 -4.41146672e-01
-4.85163182e-02 -9.55199182e-01 1.23430467e+00 4.36505452e-02
5.72660528e-02 -1.25215590e+00 -1.26350746e-01 -3.70462656e-01
4.62933093e-01 2.53074825e-01 9.91099536e-01 -3.39103252e-01
-1.05346477e+00 3.17114532e-01 -6.69385970e-01 3.19680065e-01
3.16662431e-01 -2.33635291e-01 -1.17909241e+00 -2.86037356e-01
-6.04293607e-02 1.10264413e-01 1.08351135e+00 6.35781407e-01
1.82249951e+00 -2.66635958e-02 -4.95351493e-01 1.18120837e+00
1.49639177e+00 -6.02155589e-02 7.07122445e-01 -9.89790112e-02
1.07276106e+00 1.35242641e-01 -1.22926630e-01 6.21001348e-02
3.14152807e-01 2.57921666e-01 7.98985958e-01 -4.72454578e-01
-3.58710676e-01 -3.17087978e-01 -2.92141020e-01 4.50841159e-01
-2.65708357e-01 -1.51989341e-01 -7.66160071e-01 6.55498207e-01
-1.73158538e+00 -9.74351823e-01 -5.89928068e-02 2.31190872e+00
2.88421422e-01 6.92635626e-02 -2.00502694e-01 -1.96251601e-01
7.55056441e-01 4.55578774e-01 -7.42464483e-01 -3.27988416e-01
-2.34028921e-01 6.13309920e-01 8.19668591e-01 7.18696237e-01
-1.20186877e+00 8.79038990e-01 6.99355507e+00 4.15279746e-01
-1.25790834e+00 -9.19431075e-02 4.88210976e-01 8.04198012e-02
-6.52285457e-01 -1.20989293e-01 -4.46457177e-01 3.02504152e-01
3.60407978e-01 1.26259267e-01 5.50533235e-01 8.51202428e-01
-1.52043968e-01 2.60961652e-01 -1.08287406e+00 1.13133073e+00
-9.22992527e-02 -1.77692497e+00 2.83028662e-01 2.48269916e-01
7.02754617e-01 5.63499629e-01 1.75839052e-01 -3.22465859e-02
5.97575009e-01 -1.30564415e+00 4.74630922e-01 4.68714565e-01
9.44920897e-01 -6.85375929e-01 4.23028320e-01 1.62976593e-01
-1.38698959e+00 2.90175796e-01 -6.98668897e-01 -1.58231363e-01
-3.02042644e-02 6.29574299e-01 -4.91351336e-01 4.64549571e-01
9.65759277e-01 9.78433073e-01 -3.72307628e-01 8.78039598e-01
1.52178437e-01 2.41642758e-01 -4.68746841e-01 1.21482782e-01
3.75708848e-01 -2.52278119e-01 6.66665614e-01 1.15439510e+00
3.61251563e-01 2.44942859e-01 1.66026086e-01 1.36847651e+00
-2.40630269e-01 -3.42817336e-01 -1.22722232e+00 4.41183239e-01
4.67169881e-01 9.78417575e-01 -8.03948760e-01 -1.96675628e-01
-8.91628742e-01 1.12995410e+00 6.01310432e-01 6.33828580e-01
-5.38565040e-01 -3.47101659e-01 1.26582420e+00 3.29766691e-01
8.07040215e-01 -4.45966959e-01 -7.86189079e-01 -1.01320994e+00
1.85563024e-02 -1.74662918e-01 2.88790852e-01 -7.95430541e-01
-1.79336953e+00 6.78433359e-01 1.09118894e-02 -1.17259765e+00
3.34246755e-01 -9.29691732e-01 -8.68690014e-01 1.23673344e+00
-1.63891494e+00 -1.20993698e+00 -5.02527893e-01 8.11350167e-01
2.18815446e-01 7.43190423e-02 8.38111997e-01 1.40171081e-01
8.82112682e-02 5.10181844e-01 -1.68173522e-01 2.71915168e-01
2.42293850e-01 -1.56826210e+00 1.00058091e+00 6.06775939e-01
2.40399808e-01 7.96691716e-01 2.09159389e-01 -5.92309296e-01
-1.23596025e+00 -1.48581922e+00 6.76215410e-01 -6.63616538e-01
4.18807626e-01 -4.82196033e-01 -1.41365874e+00 9.42561507e-01
1.79499775e-01 7.40873039e-01 1.77353889e-01 9.81491506e-02
-6.87570989e-01 1.65237486e-01 -1.38273084e+00 5.53338051e-01
1.67033803e+00 -8.39432299e-01 -4.80692804e-01 2.00334504e-01
1.07840562e+00 -8.12722087e-01 -7.41651237e-01 5.70978880e-01
5.89098632e-02 -9.25051510e-01 1.19683814e+00 -7.94140279e-01
5.06335758e-02 -5.29203773e-01 -2.25680575e-01 -1.45351934e+00
-7.76143432e-01 -1.54888228e-01 -1.39465958e-01 6.72345102e-01
3.25526953e-01 -8.46056819e-01 9.65872526e-01 4.19878572e-01
-4.29764807e-01 -7.93610811e-01 -9.23506856e-01 -9.31331336e-01
3.44018906e-01 -5.43138683e-01 9.81826067e-01 1.19548857e+00
-4.63003367e-01 1.86167046e-01 3.10176790e-01 6.46764278e-01
5.54056168e-01 3.65937203e-01 6.69634283e-01 -1.39450145e+00
-1.02575026e-01 -5.55222750e-01 -6.69533610e-01 -1.45457327e+00
1.72925383e-01 -1.20287228e+00 -2.77677327e-01 -1.55919158e+00
-9.58868489e-02 -6.00563943e-01 -2.77960867e-01 3.42678249e-01
-1.04355730e-01 2.79257387e-01 -4.46520858e-02 -8.77081677e-02
-2.19225913e-01 5.12219310e-01 1.61790395e+00 -1.79035351e-01
-2.52382189e-01 7.02977106e-02 -6.49959445e-01 7.52615213e-01
6.69022262e-01 -1.06137514e-01 -5.94353259e-01 -8.80753279e-01
-9.76851359e-02 -5.44833720e-01 7.47777104e-01 -8.32170248e-01
1.68468505e-01 -2.07933202e-01 5.32844782e-01 -7.53249228e-01
3.14815462e-01 -8.30662012e-01 9.28631648e-02 4.89021428e-02
-2.99522337e-02 4.04969543e-01 4.31318164e-01 7.48656332e-01
1.54615687e-02 1.23401381e-01 8.22434008e-01 -4.37106222e-01
-6.63922071e-01 7.86392033e-01 1.32764995e-01 -1.17848121e-01
9.21234965e-01 -5.40983617e-01 -2.79750854e-01 -3.18198204e-01
-8.90117884e-01 -7.06800744e-02 7.66111910e-01 4.36762482e-01
9.21901584e-01 -1.59651756e+00 -5.70476949e-01 6.10600293e-01
1.22280285e-01 7.01353848e-01 1.16556756e-01 3.65436941e-01
-4.88572955e-01 1.12152226e-01 6.51470348e-02 -9.80006397e-01
-8.81473124e-01 7.71106958e-01 5.16039729e-01 1.91134304e-01
-1.28147542e+00 9.85335290e-01 6.44009113e-01 -7.46255696e-01
1.99773610e-01 -8.50498259e-01 1.99186608e-01 -5.34887016e-01
4.46740717e-01 -1.73288945e-03 2.58829355e-01 -4.86838967e-01
-2.84164757e-01 8.56806278e-01 2.21898958e-01 2.11379677e-01
1.26809752e+00 1.33658303e-02 -7.30874017e-02 4.67407137e-01
1.07638955e+00 4.28548902e-02 -1.45080066e+00 -4.59095865e-01
-4.16304350e-01 -7.86007166e-01 9.14758965e-02 -3.34258765e-01
-1.25982368e+00 9.97887969e-01 4.54322368e-01 4.19462740e-01
7.83189237e-01 3.28491211e-01 5.63756645e-01 3.19176584e-01
4.64691728e-01 -4.47901905e-01 -6.84669092e-02 9.59219873e-01
1.03369451e+00 -1.15795898e+00 -2.32787102e-01 -5.66268563e-01
-2.90047526e-01 9.45059061e-01 8.12588453e-01 -5.15013516e-01
1.23901653e+00 4.92091984e-01 -5.92106208e-03 -4.55286980e-01
-4.67025310e-01 -4.37941670e-01 3.32565993e-01 1.02536404e+00
3.11768174e-01 1.24094881e-01 4.51318204e-01 7.17073828e-02
-9.10703465e-02 -2.48924628e-01 7.20842332e-02 6.76312387e-01
-2.50366300e-01 -7.58278847e-01 -3.05816382e-01 5.48460603e-01
-2.14274190e-02 -2.09393322e-01 -2.90582001e-01 8.54359806e-01
1.80755079e-01 4.22919869e-01 8.12210083e-01 -4.61176813e-01
6.10878289e-01 -2.34947167e-02 6.47062182e-01 -8.77440870e-01
-3.93706858e-01 -3.40292424e-01 -4.65713769e-01 -7.27192760e-01
-4.12914127e-01 -4.21595007e-01 -1.52972353e+00 -6.28336608e-01
-1.03722833e-01 -2.61911124e-01 3.33424896e-01 6.12411022e-01
7.18662322e-01 6.19459391e-01 4.79004651e-01 -1.04293799e+00
-2.20662951e-01 -7.12878406e-01 -4.48097795e-01 5.94364762e-01
6.06345773e-01 -7.55510628e-01 -2.91251630e-01 -1.20922804e-01] | [7.937928676605225, -3.6849215030670166] |
445483b4-b09e-4e1b-8bad-3153af325c0f | sequential-ensemble-learning-for-outlier | 1609.05528 | null | http://arxiv.org/abs/1609.05528v1 | http://arxiv.org/pdf/1609.05528v1.pdf | Sequential Ensemble Learning for Outlier Detection: A Bias-Variance Perspective | Ensemble methods for classification and clustering have been effectively used
for decades, while ensemble learning for outlier detection has only been
studied recently. In this work, we design a new ensemble approach for outlier
detection in multi-dimensional point data, which provides improved accuracy by
reducing error through both bias and variance. Although classification and
outlier detection appear as different problems, their theoretical underpinnings
are quite similar in terms of the bias-variance trade-off [1], where outlier
detection is considered as a binary classification task with unobserved labels
but a similar bias-variance decomposition of error.
In this paper, we propose a sequential ensemble approach called CARE that
employs a two-phase aggregation of the intermediate results in each iteration
to reach the final outcome. Unlike existing outlier ensembles which solely
incorporate a parallel framework by aggregating the outcomes of independent
base detectors to reduce variance, our ensemble incorporates both the parallel
and sequential building blocks to reduce bias as well as variance by ($i$)
successively eliminating outliers from the original dataset to build a better
data model on which outlierness is estimated (sequentially), and ($ii$)
combining the results from individual base detectors and across iterations
(parallelly). Through extensive experiments on sixteen real-world datasets
mainly from the UCI machine learning repository [2], we show that CARE performs
significantly better than or at least similar to the individual baselines. We
also compare CARE with the state-of-the-art outlier ensembles where it also
provides significant improvement when it is the winner and remains close
otherwise. | ['Wen Zhong', 'Shebuti Rayana', 'Leman Akoglu'] | 2016-09-18 | null | null | null | null | ['outlier-ensembles'] | ['methodology'] | [-1.66190177e-01 -3.64255279e-01 2.96843439e-01 -3.21350694e-01
-8.68099630e-01 -3.23088169e-01 4.34809327e-01 5.47717392e-01
-3.63195926e-01 4.56633329e-01 1.27185494e-01 -1.96504354e-01
-2.68081754e-01 -4.99085724e-01 -5.14275074e-01 -7.99613655e-01
-2.67281324e-01 4.62005228e-01 7.51333609e-02 1.02859735e-01
4.70594764e-01 1.79181844e-01 -1.66331005e+00 1.78751186e-01
1.43562126e+00 9.32625294e-01 -6.74793124e-01 3.99527073e-01
3.77881788e-02 5.29304326e-01 -7.16630995e-01 -1.58605292e-01
4.07717079e-01 -4.23695177e-01 -3.54008555e-01 -4.42447327e-02
3.57920855e-01 1.64252147e-02 -2.48009190e-02 8.08100164e-01
5.96210539e-01 2.90297359e-01 8.76960099e-01 -1.50497699e+00
-3.33647370e-01 3.97839010e-01 -8.24413836e-01 2.35416934e-01
1.96641505e-01 6.35073259e-02 8.30784440e-01 -9.78141427e-01
1.97256859e-02 9.11010265e-01 1.09540498e+00 1.27472237e-01
-1.43081248e+00 -8.00077081e-01 3.61637145e-01 -1.86787229e-02
-1.45837045e+00 -2.50045568e-01 5.51131785e-01 -4.85042900e-01
1.01639247e+00 2.96492189e-01 2.93748796e-01 8.72639954e-01
3.62954944e-01 6.49312615e-01 9.17216599e-01 -2.59412438e-01
5.87007403e-01 -2.21775383e-01 4.71314609e-01 3.79326969e-01
6.95582330e-01 8.33658725e-02 -3.18757236e-01 -7.67013073e-01
2.52010971e-02 5.83216727e-01 -2.12840363e-01 -2.57701486e-01
-9.34752643e-01 6.75190866e-01 3.72231424e-01 1.95272908e-01
-5.00231802e-01 7.16480240e-03 5.08819282e-01 4.45855170e-01
6.66097343e-01 4.94350255e-01 -5.14564335e-01 -5.11207134e-02
-1.16439438e+00 3.71692836e-01 7.99662173e-01 6.80420816e-01
5.98180532e-01 -7.84069002e-02 -3.12294394e-01 8.30584764e-01
3.67407918e-01 3.52294922e-01 4.98259902e-01 -6.11954212e-01
5.32597125e-01 1.00195301e+00 2.63294309e-01 -9.60734725e-01
-5.61987281e-01 -4.82590765e-01 -1.19349825e+00 3.66079718e-01
3.80592346e-01 -2.42588237e-01 -1.01503229e+00 1.62995017e+00
1.93581417e-01 8.10070455e-01 -4.08600643e-02 4.77434993e-01
4.11829948e-01 3.01565319e-01 1.07908338e-01 -2.76902199e-01
9.67149496e-01 -6.99517488e-01 -4.11457300e-01 2.75982507e-02
8.52103233e-01 -4.79801744e-01 6.08798146e-01 7.18553007e-01
-6.39806509e-01 -4.69084054e-01 -1.12318146e+00 3.49883944e-01
-3.70212823e-01 -1.15996018e-01 2.56474406e-01 7.70268559e-01
-8.59926581e-01 7.64459133e-01 -1.07151401e+00 -2.32400760e-01
5.27659178e-01 4.31968391e-01 -3.03848833e-01 -1.64719336e-02
-7.62831271e-01 7.04649031e-01 3.20649475e-01 1.38718724e-01
-4.37061727e-01 -6.66476548e-01 -6.92847788e-01 -1.55616209e-01
3.73005182e-01 -9.05836105e-01 8.60215902e-01 -6.49630189e-01
-1.05622101e+00 3.88643712e-01 -4.86273587e-01 -5.68900287e-01
7.16112077e-01 -6.05881989e-01 -5.22121370e-01 -5.53497791e-01
1.00439355e-01 -2.03522258e-02 7.34643161e-01 -1.36283743e+00
-9.95458484e-01 -7.21092999e-01 -5.47797501e-01 9.72078443e-02
-3.39706503e-02 -3.38458680e-02 -3.15936625e-01 -7.65717864e-01
5.80288470e-01 -9.94649827e-01 -5.66717684e-01 -5.99043190e-01
-5.06538868e-01 -3.72070611e-01 9.77903068e-01 -5.14451623e-01
1.80937624e+00 -2.22392273e+00 2.53720805e-02 6.37708008e-01
4.54179347e-01 8.08051527e-02 1.00899071e-01 4.35001761e-01
-3.09860945e-01 2.07060248e-01 -6.09319389e-01 -7.04396963e-01
-1.01237543e-01 1.20266482e-01 -4.15705502e-01 7.18077898e-01
6.07130937e-02 5.21505713e-01 -8.22959065e-01 -5.77175617e-02
2.07402229e-01 5.86676635e-02 -6.60035133e-01 6.28630221e-02
1.38634682e-01 5.33533216e-01 -1.12350315e-01 7.92406917e-01
6.88252687e-01 7.74829686e-02 -1.80980474e-01 1.18487291e-01
-3.95004675e-02 -1.65686235e-01 -1.72016537e+00 1.26132405e+00
-1.72961459e-01 2.59013236e-01 -3.70686322e-01 -1.04398870e+00
1.05924356e+00 3.10823888e-01 6.83009624e-01 -2.51518101e-01
-5.24362177e-02 5.83422422e-01 8.14240649e-02 -2.27296874e-01
3.82353842e-01 5.88817336e-02 -4.80097950e-01 5.75236499e-01
-3.09377879e-01 1.33174703e-01 2.37194389e-01 6.75484389e-02
1.48917091e+00 -2.46029459e-02 3.99426579e-01 -1.19467467e-01
5.69861174e-01 -1.69066072e-01 8.65555704e-01 1.19372249e+00
-3.18233401e-01 8.67967069e-01 3.84347856e-01 -5.19601643e-01
-8.20761621e-01 -1.18134654e+00 -2.01929152e-01 7.04863369e-01
6.97087646e-02 -5.08373857e-01 -5.25126338e-01 -9.45086122e-01
4.95608151e-01 8.71762872e-01 -5.96903682e-01 -2.15052754e-01
-4.78718579e-01 -1.24439335e+00 5.61970770e-01 5.45190573e-01
4.82096702e-01 -9.12843108e-01 -3.72149527e-01 2.62977123e-01
-8.45721513e-02 -6.23078465e-01 -2.60491222e-01 4.09917057e-01
-1.14367342e+00 -1.21156955e+00 -1.68565810e-01 -1.97161466e-01
6.70191944e-01 3.43446657e-02 1.18629110e+00 3.57607901e-01
-1.30223066e-01 9.13652033e-02 -4.01648790e-01 -7.88691819e-01
-7.38083292e-03 -1.01860754e-01 5.38383842e-01 -4.75597754e-03
6.94361687e-01 -6.63707078e-01 -5.33866227e-01 3.39987904e-01
-8.57989728e-01 -7.57156789e-01 4.43664759e-01 9.35630322e-01
4.59157497e-01 3.03620607e-01 7.02616632e-01 -8.80772412e-01
6.88181996e-01 -8.68445814e-01 -4.06135172e-01 -1.22377641e-01
-8.12671423e-01 5.17854989e-02 7.20971227e-01 2.06781253e-02
-6.94172084e-01 -1.34584770e-01 1.19822770e-02 -3.84619534e-01
-3.48646224e-01 3.20823342e-01 5.96011169e-02 2.33985126e-01
5.83018064e-01 -4.75992728e-03 -1.33669674e-01 -5.22133946e-01
5.26436456e-02 9.55135286e-01 3.65859747e-01 -6.24657929e-01
7.50695527e-01 4.93737578e-01 -4.61612083e-02 -4.89861339e-01
-7.84263194e-01 -8.45669389e-01 -7.94493318e-01 9.02287811e-02
4.72746491e-01 -9.43000257e-01 -6.47086322e-01 7.47137904e-01
-8.38191688e-01 2.12014943e-01 -2.34971836e-01 4.09200311e-01
-3.54213715e-01 4.86310363e-01 -2.54000187e-01 -1.02412760e+00
-4.34878081e-01 -1.11421633e+00 1.05723512e+00 6.23976402e-02
-4.48479950e-01 -7.86530375e-01 4.94027138e-01 -3.51297557e-02
1.89205363e-01 7.06176817e-01 6.58361614e-01 -1.27751589e+00
-2.45657131e-01 -5.12193084e-01 -9.76428017e-03 3.59538764e-01
2.01966703e-01 6.60973117e-02 -8.24746251e-01 -5.41289508e-01
-5.74350171e-02 2.86518484e-01 1.16264153e+00 3.01904470e-01
1.24561000e+00 -4.69491854e-02 -5.28029203e-01 5.88061750e-01
1.57475817e+00 1.74163446e-01 6.04370952e-01 5.39424896e-01
6.19124174e-01 2.92571694e-01 4.57700104e-01 5.84228218e-01
3.42759192e-01 5.43902755e-01 2.71204948e-01 7.06705302e-02
4.29536611e-01 1.09604679e-01 4.55522925e-01 7.10220814e-01
-3.03181469e-01 -1.96862951e-01 -1.14616907e+00 5.44328868e-01
-2.27162337e+00 -1.25869477e+00 -5.39970040e-01 2.73714232e+00
3.50272834e-01 8.29281062e-02 2.65927523e-01 3.45377654e-01
6.12659037e-01 -5.70663624e-02 -5.88930368e-01 -4.53601301e-01
2.00481080e-02 1.37876347e-01 3.44363302e-01 1.92147806e-01
-1.33865690e+00 4.03916091e-01 6.76511574e+00 5.15452325e-01
-5.91300309e-01 -4.83727232e-02 6.49756908e-01 -2.94159174e-01
1.64973527e-01 1.10941783e-01 -6.36485159e-01 8.94867361e-01
9.74628747e-01 3.22292037e-02 7.46888444e-02 7.29737759e-01
2.07739562e-01 -3.80725086e-01 -1.31312537e+00 9.80190873e-01
3.36749181e-02 -8.07026207e-01 -2.00120941e-01 2.18006164e-01
1.06516922e+00 3.05152565e-01 -1.58203855e-01 5.74896336e-01
5.47643185e-01 -9.07777488e-01 4.68713194e-01 7.31776237e-01
1.01191640e-01 -1.03442657e+00 1.29114342e+00 5.05345225e-01
-1.10047340e+00 -5.69801569e-01 -1.02975100e-01 -1.61364362e-01
-4.23258319e-02 1.17150915e+00 -2.78554678e-01 9.87935901e-01
1.16906381e+00 6.54349327e-01 -6.49077594e-01 1.29097140e+00
8.13931823e-02 5.75009525e-01 -6.54467106e-01 3.35411131e-01
6.36121854e-02 -5.17754912e-01 7.29962587e-01 1.06703675e+00
5.87741017e-01 8.33858550e-02 3.65639627e-01 4.90413100e-01
2.66041048e-02 6.93456456e-02 -6.50623679e-01 5.70686519e-01
6.10508263e-01 9.41909671e-01 -5.27164221e-01 -4.27613109e-01
-5.37343025e-01 9.33507860e-01 4.09752309e-01 3.66837919e-01
-6.98495924e-01 -5.64993143e-01 8.66470814e-01 -1.41898513e-01
4.09310788e-01 -4.99912947e-02 -8.15488338e-01 -1.21358311e+00
1.42688558e-01 -9.22049344e-01 7.28231788e-01 -1.37443915e-01
-1.71124351e+00 4.33858842e-01 -1.57456264e-01 -1.57710588e+00
-9.21792760e-02 -4.84167457e-01 -1.02143991e+00 8.43595624e-01
-1.02036643e+00 -5.85937321e-01 -4.39691275e-01 3.81440341e-01
2.76696593e-01 -7.60817230e-02 6.67659104e-01 1.18291095e-01
-1.08026457e+00 6.36512399e-01 6.57019794e-01 4.53678407e-02
1.07747126e+00 -1.49376798e+00 3.33233446e-01 1.09822118e+00
7.51232877e-02 7.69286335e-01 7.45678544e-01 -9.03782248e-01
-8.25520694e-01 -1.26883554e+00 6.97275460e-01 -1.02995086e+00
3.25685382e-01 -6.54552206e-02 -1.12174642e+00 6.67582154e-01
-6.70353472e-02 2.43250038e-02 9.57269788e-01 5.52206218e-01
-2.71204591e-01 -1.88309029e-01 -1.33760262e+00 5.42624176e-01
9.79595304e-01 1.25581687e-02 -7.56856859e-01 -1.07510984e-01
1.54497638e-01 -3.41339767e-01 -8.63408625e-01 7.03296781e-01
4.40093100e-01 -1.44465399e+00 6.80717409e-01 -5.00693858e-01
1.06885970e-01 -7.08619475e-01 -2.48218507e-01 -1.54120409e+00
-3.24270546e-01 -2.90963322e-01 -3.32851827e-01 1.16689110e+00
3.60647053e-01 -1.06789601e+00 4.62175250e-01 5.32266378e-01
-2.89827943e-01 -8.48241150e-01 -9.07428205e-01 -8.94921482e-01
-9.95960087e-04 -7.27202475e-01 7.52411664e-01 8.38132143e-01
4.73027525e-04 -7.69688189e-02 -2.40283996e-01 4.88302439e-01
8.85528862e-01 3.24533954e-02 1.08538973e+00 -1.62411094e+00
-2.47948281e-02 -4.79437947e-01 -5.31078756e-01 -5.67328930e-01
-6.47436157e-02 -8.05355906e-01 1.59064740e-01 -1.29180574e+00
2.70899445e-01 -4.70585138e-01 -7.88753331e-01 4.43219543e-01
-8.41890395e-01 2.10124820e-01 -4.11195867e-02 5.89314044e-01
-7.98383236e-01 4.49975371e-01 2.00327575e-01 2.24466980e-01
-5.68804860e-01 2.07735628e-01 -7.98935115e-01 9.76088881e-01
7.02133834e-01 -7.37449288e-01 8.15261155e-03 -1.63822293e-01
-9.15052078e-04 -4.19792026e-01 2.26273432e-01 -1.37203586e+00
2.72174448e-01 2.08354130e-01 6.87188745e-01 -8.53612185e-01
-1.77694753e-01 -9.06645536e-01 1.79939151e-01 5.20121336e-01
1.07304744e-01 4.14200544e-01 1.20204404e-01 1.07104647e+00
-3.92033100e-01 7.67641282e-03 6.14692450e-01 7.13839084e-02
-5.24118662e-01 3.80576879e-01 -1.38922468e-01 -4.09845524e-02
1.34942365e+00 -3.32783461e-01 -2.04448223e-01 -1.48451015e-01
-7.47735202e-01 6.10182106e-01 7.11763084e-01 4.06109631e-01
3.85334104e-01 -1.25667524e+00 -8.94850791e-01 2.99814075e-01
3.62632602e-01 1.57393992e-01 1.76534697e-01 1.13395333e+00
-2.09351316e-01 1.30560324e-01 2.54361987e-01 -9.05028701e-01
-1.05210888e+00 3.82884145e-01 2.75154173e-01 -5.02081275e-01
-4.39903617e-01 6.66009963e-01 -2.53062278e-01 -7.11178780e-01
3.04055452e-01 -2.58916616e-01 -1.21417865e-02 1.23975389e-01
4.75559622e-01 8.32953751e-01 3.30668688e-01 -4.64343935e-01
-5.78151345e-01 6.12886608e-01 -1.46632552e-01 2.30328724e-01
1.28090847e+00 -1.13647372e-01 -3.30523461e-01 6.90151334e-01
8.27128708e-01 1.13818914e-01 -9.24782336e-01 -2.41386011e-01
4.99388158e-01 -4.77428526e-01 -1.70403704e-01 -5.62294960e-01
-7.23195374e-01 4.08902496e-01 6.28452837e-01 2.71444201e-01
1.31400478e+00 -2.54332989e-01 6.43981814e-01 2.60490298e-01
3.43472332e-01 -1.18304706e+00 -8.00393429e-03 6.42622471e-01
5.64275503e-01 -1.47931111e+00 1.90885544e-01 -6.66002259e-02
-5.68790495e-01 9.30929959e-01 6.85661793e-01 -4.84814554e-01
6.83024287e-01 1.62510917e-01 -3.06056924e-02 -7.91532099e-02
-7.72326112e-01 -9.14277807e-02 3.53709459e-01 3.80256891e-01
3.61004084e-01 -6.88873604e-02 -3.78971756e-01 7.92233825e-01
6.77960441e-02 -1.47119612e-01 2.16504484e-01 9.30893958e-01
-5.13343632e-01 -1.10851526e+00 -7.03206122e-01 9.97388482e-01
-4.86362338e-01 -4.82413284e-02 -1.78120598e-01 5.70878267e-01
4.83607113e-01 1.22392297e+00 3.25328141e-01 -4.46394175e-01
7.13134766e-01 4.63330775e-01 -8.94078389e-02 -6.25858963e-01
-7.97327399e-01 -2.03443140e-01 -1.69591993e-01 -7.35245824e-01
-1.36522874e-01 -1.04794610e+00 -1.14147842e+00 -2.43607119e-01
-4.48725075e-01 3.21948081e-01 4.73514885e-01 1.08299041e+00
6.46264791e-01 5.17813742e-01 9.35449839e-01 -7.68214881e-01
-7.56268084e-01 -1.03536296e+00 -5.80209196e-01 5.47601223e-01
4.01968271e-01 -6.73833609e-01 -9.01363373e-01 -3.06104243e-01] | [7.547979831695557, 2.7002744674682617] |
dfaaaafb-62dd-44fc-921b-c3fe1147fabd | md-gcn-a-multi-scale-temporal-dual-graph | null | null | https://kns.cnki.net/kcms2/article/abstract?v=LeQIq0pPraN7z56UFBXYmp5cqSpFXzXCFpgvv08RLM-paCwYX2_gXb2meUAoCMzJBhBlFELjQyQx1hYWJOzD5oGwuoPshYZpIjSZ02gYn8LByhWo_x9WmW2F6JdeiXmK&uniplatform=NZKPT | https://www.mdpi.com/1424-8220/23/2/841 | MD-GCN: A Multi-Scale Temporal Dual Graph Convolution Network for Traffic Flow Prediction | The spatial–temporal prediction of traffic flow is very important for traffic management and planning. The most difficult challenges of traffic flow prediction are the temporal feature extraction and the spatial correlation extraction of nodes. Due to the complex spatial correlation between different roads and the dynamic trend of time patterns, traditional forecasting methods still have limitations in obtaining spatial–temporal correlation, which makes it difficult to extract more valid information. In order to improve the accuracy of the forecasting, this paper proposes a multi-scale temporal dual graph convolution network for traffic flow prediction (MD-GCN). Firstly, we propose a gated temporal convolution based on a channel attention and inception structure to extract multi-scale temporal dependence. Then, aiming at the complexity of the traffic spatial structure, we develop a dual graph convolution module including the graph sampling and aggregation submodule (GraphSAGE) and the mix-hop propagation graph convolution submodule (MGCN) to extract the local correlation and global correlation between neighbor nodes. Finally, extensive experiments are carried out on several public traffic datasets, and the experimental results show that our proposed algorithm outperforms the existing methods. | ['Huang Xiaohui;Wang Junyang;Lan Yuanchun;Jiang Chaojie;Yuan Xinhua'] | 2023-01-11 | null | null | null | iaengqi-kan-2023-1 | ['graph-sampling'] | ['graphs'] | [-1.59607098e-01 -5.99303186e-01 -1.44998759e-01 -2.40920514e-01
1.79493561e-01 -4.71597426e-02 4.35259759e-01 -2.14175954e-01
-1.13861084e-01 6.20030761e-01 5.05383722e-02 -8.11820686e-01
-5.03177822e-01 -1.29845095e+00 -2.60710746e-01 -6.97138786e-01
-5.71686924e-01 6.21818379e-02 7.18424737e-01 -3.05095345e-01
1.84146598e-01 6.16302967e-01 -1.43332374e+00 6.24062866e-02
1.00879037e+00 1.17115867e+00 2.00135291e-01 5.74788094e-01
-5.68684220e-01 9.19135213e-01 -2.08375454e-01 -3.31170052e-01
9.04032663e-02 -3.00190032e-01 -6.01243734e-01 4.69274409e-02
1.66436672e-01 -2.19816282e-01 -1.02898896e+00 8.93410861e-01
3.36848199e-01 4.82690573e-01 2.82876521e-01 -1.57004905e+00
-4.09421533e-01 4.95345533e-01 -6.11564517e-01 9.24703360e-01
-3.17105532e-01 4.02755052e-01 8.35960686e-01 -3.82283807e-01
1.08216047e-01 1.34381485e+00 6.87033415e-01 -3.42711061e-02
-7.76180506e-01 -8.69693577e-01 6.25321925e-01 7.99722433e-01
-1.51738203e+00 -1.38474837e-01 1.01939511e+00 -4.01923180e-01
8.52838457e-01 2.24693194e-01 7.85786450e-01 5.43710589e-01
2.22731784e-01 6.02984011e-01 8.04938018e-01 2.39647359e-01
-3.51959974e-01 -2.67455041e-01 1.39624238e-01 6.88364267e-01
-1.08620062e-01 2.52995163e-01 -1.33420721e-01 3.07755113e-01
7.61748493e-01 2.93569177e-01 -1.17902994e-01 3.67698848e-01
-9.57617939e-01 5.48316419e-01 1.05172217e+00 5.99749863e-01
-4.68166977e-01 4.81032312e-01 5.04028797e-01 1.26535028e-01
8.12642097e-01 -4.70553398e-01 -1.25010252e-01 -2.23375142e-01
-8.75895202e-01 -7.28225335e-02 3.49304348e-01 9.83550489e-01
9.51710165e-01 4.70197648e-02 -3.81368101e-01 6.52125895e-01
1.77493647e-01 5.55651248e-01 9.51843411e-02 -4.10096049e-01
9.46748316e-01 8.34493160e-01 -4.18765754e-01 -1.91377580e+00
-7.31541395e-01 -6.04098320e-01 -1.30448484e+00 -2.76261687e-01
2.86538213e-01 -3.21281731e-01 -6.74532652e-01 1.31649148e+00
3.16478282e-01 9.85664606e-01 -3.47392470e-01 9.81341064e-01
9.24348414e-01 1.04147100e+00 3.78093898e-01 -2.17126220e-01
1.21703935e+00 -1.04319346e+00 -8.50754142e-01 -1.78155661e-01
8.32523048e-01 -5.82367063e-01 7.13993073e-01 -4.45125759e-01
-7.24377394e-01 -7.71548510e-01 -5.82333028e-01 5.10863848e-02
-7.07202673e-01 3.55125442e-02 9.40439045e-01 3.05917144e-01
-8.14009309e-01 4.50232923e-01 -5.40083706e-01 -2.57624745e-01
5.87584555e-01 2.95262516e-01 -1.92693323e-02 -1.62554815e-01
-1.58182263e+00 5.15371859e-01 5.03901780e-01 8.33497047e-01
-1.58106893e-01 -7.39997566e-01 -6.22972429e-01 3.05402726e-01
3.23747158e-01 -3.47920656e-01 6.15107417e-01 -6.14641070e-01
-9.16926742e-01 3.92252766e-02 -3.49697948e-01 -3.92921209e-01
4.19030130e-01 4.64873344e-01 -1.22652721e+00 7.66316196e-03
1.50392935e-01 1.18549399e-01 4.12778735e-01 -6.39683962e-01
-1.05681396e+00 -2.32512891e-01 -4.03044969e-02 3.32162939e-02
-4.19292271e-01 -5.45893088e-02 -7.95204818e-01 -5.72179914e-01
-1.24761358e-01 -5.50344408e-01 -4.92807031e-01 -4.84847009e-01
-3.66035938e-01 -3.56624812e-01 1.13250065e+00 -8.31104875e-01
1.99314594e+00 -2.08922601e+00 -3.73159021e-01 6.29126310e-01
6.07356906e-01 3.73111010e-01 -1.57499894e-01 5.23081064e-01
-7.13615939e-02 -5.37601933e-02 -5.84150665e-02 1.72623731e-02
-2.07561195e-01 2.19173476e-01 -1.80863887e-01 2.63950408e-01
1.60427049e-01 1.21236014e+00 -1.04385984e+00 -7.77132213e-01
5.67309260e-01 5.02194166e-01 -2.67894119e-01 -9.50483531e-02
-4.38561216e-02 5.99259973e-01 -8.23382258e-01 2.55080402e-01
1.08824039e+00 -4.30110514e-01 -7.97901601e-02 -3.26266289e-01
-4.36464429e-01 3.13313395e-01 -1.11704397e+00 9.90683854e-01
-6.39039278e-01 7.47289658e-01 -2.95748830e-01 -1.13032854e+00
9.64926779e-01 1.56368420e-01 6.53083920e-01 -1.16470838e+00
1.59260511e-01 1.30601957e-01 1.65909678e-01 -7.47954249e-01
4.51560915e-01 1.40346721e-01 9.38554704e-02 2.14951009e-01
-3.18849772e-01 6.85040593e-01 4.86371756e-01 2.63090909e-01
1.12082124e+00 -4.94516879e-01 -4.41548645e-01 -8.37105289e-02
9.95715201e-01 -1.72163397e-02 6.25419259e-01 1.66716203e-01
-2.70865589e-01 8.27883407e-02 6.51440024e-01 -8.73226345e-01
-6.02507055e-01 -6.66541934e-01 2.01708168e-01 7.03534067e-01
2.52229065e-01 -3.52104396e-01 -3.43654275e-01 -7.19178557e-01
1.03101127e-01 3.04471612e-01 -7.02266455e-01 -1.94431782e-01
-1.03666174e+00 -8.85992706e-01 3.13532859e-01 6.08061671e-01
1.15156794e+00 -9.83097851e-01 1.82777256e-01 4.54344481e-01
-3.23592842e-01 -1.47585535e+00 -8.08289528e-01 -5.72163641e-01
-6.80336773e-01 -1.01696706e+00 -3.16992581e-01 -7.06043839e-01
5.30509591e-01 8.13710511e-01 8.45746636e-01 5.78525126e-01
-1.15512736e-01 -2.34962329e-01 -1.99803472e-01 5.84819280e-02
1.43017232e-01 4.71027225e-01 -4.28725868e-01 5.28252244e-01
6.02066934e-01 -1.05385137e+00 -7.17657745e-01 4.85028207e-01
-5.73694408e-01 1.76839635e-01 6.72231317e-01 4.22136784e-01
2.76198030e-01 6.92197800e-01 5.65627515e-01 -8.27687502e-01
7.23252356e-01 -8.47456753e-01 -7.48914301e-01 1.33871004e-01
-5.65173924e-01 -3.32985848e-01 8.42157781e-01 -2.02867091e-01
-9.75343406e-01 -3.44540536e-01 -1.07946575e-01 -4.46643203e-01
-3.38943563e-02 7.62448549e-01 -1.14787392e-01 7.36714457e-04
1.60988629e-01 4.82338756e-01 -6.19720854e-02 -2.55373269e-01
2.97525018e-01 5.52573860e-01 2.65660971e-01 -1.03288591e-01
9.46032465e-01 4.66553479e-01 5.01244187e-01 -8.76987517e-01
-5.70406795e-01 -5.25568783e-01 -7.44872212e-01 -7.10905373e-01
9.41831052e-01 -6.54576540e-01 -1.16964900e+00 4.02808547e-01
-1.12172520e+00 -2.10646078e-01 1.65058166e-01 5.81384301e-01
-4.77599688e-02 4.30681795e-01 -4.69167173e-01 -7.25487053e-01
-6.31385669e-02 -9.94701743e-01 7.33192384e-01 2.52677500e-01
4.30919379e-01 -1.46418178e+00 -2.64245450e-01 1.73062280e-01
7.47437179e-01 1.59518301e-01 8.68524015e-01 9.48283914e-03
-1.02895689e+00 -1.98918775e-01 -9.11940336e-01 5.26635759e-02
2.04258829e-01 1.35013446e-01 -5.30227602e-01 1.00506283e-01
-5.93004882e-01 5.71892917e-01 1.04885530e+00 5.61954200e-01
1.44224286e+00 -3.82047325e-01 -6.36340559e-01 7.41486251e-01
1.13350379e+00 2.02393025e-01 7.50581264e-01 -3.46023366e-02
1.15116727e+00 8.02698970e-01 4.65189934e-01 1.74151912e-01
7.62193739e-01 4.75945234e-01 3.11344922e-01 -2.52681911e-01
-3.75406086e-01 -3.73276293e-01 -8.25802423e-03 1.18608236e+00
-3.44056457e-01 -3.74640137e-01 -9.50052142e-01 6.86309040e-01
-2.16225863e+00 -1.43520105e+00 -8.64191651e-01 1.70206690e+00
-1.91618055e-02 2.03087389e-01 3.93062264e-01 1.13567315e-01
1.05823636e+00 5.10951936e-01 -2.24349737e-01 -1.99541762e-01
-6.54617697e-02 -1.63119853e-01 8.58668804e-01 6.57184660e-01
-9.51969802e-01 1.08416712e+00 5.57036734e+00 1.26930356e+00
-1.22872210e+00 4.05409709e-02 6.05808437e-01 2.08899468e-01
-3.15625995e-01 -4.78554107e-02 -7.98120499e-01 1.00136805e+00
9.87059116e-01 -6.67090639e-02 6.26554012e-01 3.18158507e-01
5.88611603e-01 2.25522324e-01 -3.21855694e-01 9.16815102e-01
-4.88396257e-01 -1.48703969e+00 6.35815738e-03 2.42905408e-01
4.72316295e-01 2.04428345e-01 -5.35826664e-03 3.40349466e-01
2.62320995e-01 -9.90547419e-01 9.73657072e-02 7.66048908e-01
5.05585670e-01 -9.72366810e-01 7.15077102e-01 2.95092255e-01
-2.13413119e+00 -1.96128368e-01 -2.16898277e-01 -2.67390758e-01
4.64908510e-01 9.09084022e-01 -5.34854054e-01 9.17577147e-01
5.83408177e-01 1.25571752e+00 -5.42987347e-01 1.02045369e+00
5.15088066e-02 7.78329253e-01 -2.79912055e-01 -2.71433383e-01
5.08095980e-01 -5.33898950e-01 3.83865654e-01 1.33597255e+00
1.88374907e-01 1.87711567e-01 2.19674975e-01 8.00680995e-01
1.54564716e-02 7.73063004e-02 -4.80222493e-01 -6.00915812e-02
5.01393974e-01 1.40231740e+00 -7.81061471e-01 -2.61929989e-01
-7.26458073e-01 5.75290620e-01 3.12105685e-01 6.94265366e-01
-1.06601703e+00 -7.40029991e-01 6.93471730e-01 3.03784311e-01
3.81160945e-01 -5.63145757e-01 -1.90707177e-01 -8.91247809e-01
1.81423664e-01 -1.07444681e-01 6.06301188e-01 -4.50010091e-01
-1.42817163e+00 6.63039029e-01 -9.31445733e-02 -1.40089297e+00
2.19299018e-01 -3.55577976e-01 -1.03174448e+00 1.16427219e+00
-1.82927811e+00 -1.30252600e+00 -6.44695282e-01 8.86761665e-01
1.66414559e-01 -1.07432440e-01 5.99689521e-02 9.52998638e-01
-8.51399660e-01 3.67391676e-01 -2.99179137e-01 4.12961572e-01
-7.44367717e-03 -6.97763801e-01 5.84269226e-01 9.37319875e-01
-3.56584638e-01 2.07633957e-01 1.84260011e-01 -7.48877108e-01
-1.18397248e+00 -1.59460044e+00 1.19300854e+00 -6.65650470e-03
9.50418174e-01 -2.41052777e-01 -8.54006469e-01 6.26939952e-01
-4.42262366e-02 5.19755661e-01 3.67206931e-01 1.78547636e-01
-6.40986348e-03 -6.75013304e-01 -7.39557683e-01 6.79852843e-01
1.31827724e+00 -4.35643435e-01 1.70280322e-01 4.13174123e-01
8.44670832e-01 -7.36181345e-03 -7.15841770e-01 3.45735520e-01
3.37052763e-01 -6.62776947e-01 8.20916533e-01 -5.05908132e-01
2.44209260e-01 -6.27172589e-01 1.41039506e-01 -1.13760662e+00
-7.59333909e-01 -4.04039919e-01 -1.20517999e-01 1.32678795e+00
2.41130129e-01 -9.91241336e-01 6.90356255e-01 1.75736234e-01
-2.18717709e-01 -6.75565541e-01 -8.80572081e-01 -7.42002904e-01
-2.44552538e-01 -8.22513759e-01 1.16496956e+00 1.01074409e+00
-2.66401887e-01 6.41790390e-01 -3.24988067e-01 3.62876415e-01
3.47746789e-01 3.84285361e-01 8.84562850e-01 -1.23263121e+00
7.80618116e-02 -8.17593157e-01 -7.04139113e-01 -1.33369970e+00
1.25862390e-01 -9.71368730e-01 -4.45028961e-01 -1.57361317e+00
-2.94916511e-01 -6.01547420e-01 -3.51182580e-01 6.97693974e-02
-2.83838749e-01 -4.74869125e-02 -1.04932211e-01 1.15980156e-01
-5.93437016e-01 7.29706407e-01 1.73886466e+00 -1.41267940e-01
-8.33614096e-02 1.30305171e-01 -3.33728969e-01 2.49416441e-01
7.88657606e-01 -1.13500617e-01 -6.24777436e-01 -5.05836189e-01
7.38444701e-02 1.29642650e-01 5.02128720e-01 -1.04941523e+00
5.20075560e-01 -3.72459173e-01 1.82772368e-01 -1.04936373e+00
2.16637086e-02 -1.04811633e+00 7.48667419e-02 3.66480440e-01
6.69759139e-02 2.01108664e-01 2.53862262e-01 6.44980788e-01
-3.81271034e-01 5.66630900e-01 2.86148965e-01 1.47144511e-01
-9.84255195e-01 1.05491269e+00 -4.56792086e-01 -1.28357515e-01
9.98744965e-01 -1.66295096e-01 -4.40536737e-01 -2.70834506e-01
-5.54756045e-01 8.92840862e-01 -2.37722591e-01 5.82857072e-01
4.83977318e-01 -1.79505646e+00 -6.30089164e-01 2.71059215e-01
6.15042970e-02 -1.52274117e-01 8.59369516e-01 1.24494278e+00
-5.09545207e-01 5.80773354e-01 -1.00709543e-01 -4.40514863e-01
-9.45681751e-01 6.78961456e-01 4.19660360e-01 -4.03645247e-01
-6.04605556e-01 5.54086566e-01 3.65311764e-02 -3.12753886e-01
-2.31891662e-01 -3.55014175e-01 -5.09917617e-01 1.40861189e-02
4.93187994e-01 7.07484901e-01 2.49370784e-02 -9.89132226e-01
-3.07892889e-01 6.25083387e-01 1.92346156e-01 3.28588247e-01
1.17079449e+00 -4.87838537e-01 -4.04667854e-01 2.16295347e-01
1.32237709e+00 -2.04455182e-01 -9.10016656e-01 -4.22101438e-01
-3.01126808e-01 -7.22505689e-01 2.50339568e-01 -1.86106741e-01
-1.77295649e+00 1.29283750e+00 3.13608944e-01 7.48444736e-01
1.31343913e+00 -3.97349477e-01 1.38998222e+00 9.28762252e-04
7.36552402e-02 -9.41688597e-01 -3.79619271e-01 6.75215900e-01
4.35935527e-01 -9.99471784e-01 -2.78658539e-01 -9.38512623e-01
-2.38029629e-01 1.18623888e+00 8.14956605e-01 -8.48486349e-02
1.35951650e+00 -3.96897197e-02 -3.02826196e-01 -5.67392826e-01
-7.09976017e-01 -7.03605294e-01 4.25138205e-01 3.75908852e-01
2.58704960e-01 2.33455002e-01 -2.71941841e-01 3.00843239e-01
-7.69415647e-02 6.60385787e-02 -9.04924273e-02 4.43493307e-01
-3.01162511e-01 -8.50203156e-01 1.23841219e-01 6.47217095e-01
-5.03112003e-02 -9.40354988e-02 1.00273177e-01 6.32164776e-01
3.04016829e-01 1.11528432e+00 5.17621279e-01 -1.06331730e+00
3.34824055e-01 -3.77446920e-01 -5.35540618e-02 -4.50908579e-02
-3.69085401e-01 -2.30290771e-01 1.94815904e-01 -5.90634465e-01
-3.75093192e-01 -4.29268718e-01 -1.29956448e+00 -9.87917662e-01
-2.62381136e-01 2.60765165e-01 3.50657463e-01 1.01938105e+00
5.08844614e-01 9.88717735e-01 1.00983822e+00 -4.89810050e-01
4.62911040e-01 -9.17362034e-01 -5.47734141e-01 3.50185007e-01
3.92031997e-01 -7.17324376e-01 -1.53186843e-01 -3.15878034e-01] | [6.448951244354248, 2.0573604106903076] |
fc4f8011-ea8f-48bf-a75b-ea75bb097cd5 | doubly-robust-nearest-neighbors-in-factor | 2211.14297 | null | https://arxiv.org/abs/2211.14297v2 | https://arxiv.org/pdf/2211.14297v2.pdf | Doubly robust nearest neighbors in factor models | In this technical note, we introduce an improved variant of nearest neighbors for counterfactual inference in panel data settings where multiple units are assigned multiple treatments over multiple time points, each sampled with constant probabilities. We call this estimator a doubly robust nearest neighbor estimator and provide a high probability non-asymptotic error bound for the mean parameter corresponding to each unit at each time. Our guarantee shows that the doubly robust estimator provides a (near-)quadratic improvement in the error compared to nearest neighbor estimators analyzed in prior work for these settings. | ['Devavrat Shah', 'Susan Murphy', 'Predrag Klasnja', 'Sabina Tomkins', 'Katherine Tian', 'Raaz Dwivedi'] | 2022-11-25 | null | null | null | null | ['counterfactual-inference'] | ['miscellaneous'] | [-2.31353827e-02 8.15466493e-02 -8.23730588e-01 -5.64985871e-01
-1.07525373e+00 -8.84590566e-01 5.95725954e-01 2.33823314e-01
-3.96689564e-01 1.36317301e+00 6.38167500e-01 -5.38338304e-01
-5.08072317e-01 -9.73773599e-01 -1.04920363e+00 -5.95122457e-01
-3.80328476e-01 4.36240137e-01 -4.10589159e-01 3.75564426e-01
4.34457392e-01 2.95175582e-01 -1.19350123e+00 -4.52574082e-02
9.59403276e-01 7.12682426e-01 -7.46231377e-01 3.98223579e-01
4.86723989e-01 2.12260216e-01 -9.54338312e-02 -5.23645759e-01
6.07062161e-01 -4.02502567e-01 -3.11491191e-01 -4.15663391e-01
8.65330756e-01 -5.87212324e-01 -4.23248142e-01 1.02798486e+00
7.56764114e-01 4.35349256e-01 1.31133425e+00 -1.37893879e+00
-6.79195762e-01 1.16119921e+00 -1.02215886e+00 3.36427361e-01
5.36544025e-01 -2.38179952e-01 8.26434135e-01 -5.77149749e-01
3.73645067e-01 1.38814533e+00 9.76388454e-01 1.17900558e-01
-1.86885905e+00 -9.05833542e-01 -8.67111832e-02 -1.65415734e-01
-1.37946761e+00 -6.01430297e-01 3.14513505e-01 -4.96656299e-01
2.87346125e-01 1.57033056e-01 3.72382045e-01 1.38854957e+00
4.14378285e-01 4.47779238e-01 1.35296082e+00 -4.83097643e-01
7.12952435e-01 -1.00040264e-01 3.24532121e-01 6.03040829e-02
8.69018674e-01 7.97435999e-01 -3.38103443e-01 -9.46370661e-01
8.08919370e-01 3.82014304e-01 -3.42175245e-01 -4.86059546e-01
-1.02189445e+00 1.20466900e+00 1.59296796e-01 3.14454213e-02
-6.92753315e-01 2.99641728e-01 4.30211395e-01 8.30821246e-02
6.41264200e-01 -1.73833519e-01 -5.83896339e-01 1.37326956e-01
-8.07253003e-01 7.54604697e-01 1.03046203e+00 7.10321486e-01
3.35609585e-01 -1.62625402e-01 -8.06613982e-01 3.83686364e-01
-1.98504925e-02 1.00535297e+00 7.47805685e-02 -1.47223914e+00
7.52281487e-01 -2.14183405e-01 9.42096651e-01 -6.49213135e-01
-2.55795509e-01 -2.68680602e-01 -8.55830491e-01 1.73452199e-01
8.58573258e-01 -7.08398581e-01 -1.63866848e-01 2.12997198e+00
5.14112294e-01 4.34703469e-01 -1.70510918e-01 4.42053258e-01
-3.92361253e-01 5.40350974e-01 2.49298543e-01 -8.50415587e-01
1.10803413e+00 -1.71807483e-01 -7.85081804e-01 2.59128153e-01
3.57246161e-01 -4.76800591e-01 6.83696389e-01 2.11829767e-01
-9.63863313e-01 1.27147168e-01 -6.40502930e-01 3.23056936e-01
-1.06656507e-01 -5.41762948e-01 6.31047368e-01 9.52379882e-01
-5.69868505e-01 1.05366361e+00 -1.26196727e-01 -1.42615527e-01
2.86957115e-01 -1.48978978e-02 -1.86547592e-01 -1.19580559e-01
-1.08529091e+00 4.35659796e-01 8.00057426e-02 -4.44274485e-01
-6.11708581e-01 -1.37336731e+00 -7.77637422e-01 2.27549583e-01
3.21239591e-01 -9.34805095e-01 1.40641165e+00 -3.86829764e-01
-1.50067925e+00 3.11574310e-01 -2.70738870e-01 -8.34742427e-01
7.80260026e-01 2.01614220e-02 -3.06962281e-01 -2.98113793e-01
6.17335439e-01 -9.19911936e-02 5.39121389e-01 -9.38217998e-01
-6.82859659e-01 -7.49393523e-01 -7.80658796e-02 1.73070788e-01
1.98127553e-01 -1.38353243e-01 7.21879661e-01 -8.66667867e-01
-2.15824127e-01 -7.33318269e-01 -2.93931335e-01 -1.17392763e-01
-1.71505228e-01 -4.22301859e-01 2.18028188e-01 -2.63321459e-01
1.05640137e+00 -1.81306183e+00 -4.84338433e-01 1.29894063e-01
-1.64922446e-01 -4.25245613e-01 1.67052120e-01 4.47083026e-01
-3.05877864e-01 2.65347272e-01 -3.04525346e-01 -4.33475822e-02
5.25007129e-01 -3.33496630e-01 -4.78125513e-01 1.09335613e+00
-8.25275064e-01 7.26915121e-01 -9.63361979e-01 -3.26675087e-01
4.24766958e-01 -3.35825980e-02 -3.67274672e-01 -1.54504508e-01
4.19301927e-01 -3.30075324e-01 -2.72314042e-01 1.14988945e-01
1.19235933e+00 8.95682201e-02 2.97027797e-01 1.35830358e-01
-1.99391738e-01 -9.83659998e-02 -1.36395276e+00 1.11196363e+00
-4.93967116e-01 1.50045022e-01 -4.98322733e-02 -1.19661355e+00
1.89125344e-01 7.55136907e-01 3.27710718e-01 -3.86446327e-01
2.01373845e-01 1.62734553e-01 -4.87966418e-01 -1.52613655e-01
-5.93172573e-02 -9.72277343e-01 -5.00350595e-01 6.15642607e-01
-2.69693345e-01 2.91395396e-01 -1.41851962e-01 1.60886031e-02
6.97168052e-01 -3.88684303e-01 9.71153140e-01 -8.26497078e-01
-6.23106770e-02 -4.67957258e-01 8.47574174e-01 1.46234334e+00
-4.14068341e-01 3.40405524e-01 4.55014199e-01 -1.90051854e-01
-1.09219515e+00 -1.54776645e+00 -6.12868786e-01 7.77662456e-01
-1.97562099e-01 3.40295047e-01 -5.83515882e-01 -6.72489583e-01
9.17672455e-01 1.24871111e+00 -1.11181986e+00 5.20962812e-02
-1.07846269e-02 -6.62766814e-01 2.81718701e-01 6.79325163e-01
2.21521780e-01 -3.38534325e-01 -7.11748451e-02 2.06697747e-01
-2.29071286e-02 -4.40464228e-01 -1.11873114e+00 -8.00122768e-02
-1.03869784e+00 -1.02899253e+00 -1.00370193e+00 -1.38215482e-01
1.55744806e-01 4.54780012e-01 8.25692952e-01 -9.98213589e-01
3.53481084e-01 2.41934106e-01 1.39562532e-01 -4.14072007e-01
7.57694542e-02 -4.64120626e-01 4.89387870e-01 2.09606692e-01
5.66502690e-01 -6.84757590e-01 -9.92914021e-01 1.39306784e-02
-4.88672465e-01 -8.57844949e-01 1.94640830e-01 8.70004177e-01
4.05875266e-01 -1.74795270e-01 9.41766143e-01 -1.05474877e+00
7.20593154e-01 -1.09435999e+00 -1.15002728e+00 2.44208872e-01
-6.93374574e-01 1.18741848e-01 7.81332314e-01 -4.60106254e-01
-1.28706026e+00 -2.75320619e-01 7.44507968e-01 -1.00440584e-01
-2.68512726e-01 4.31928605e-01 -1.00040525e-01 2.53580749e-01
8.50204170e-01 -4.84804213e-01 -1.50846377e-01 -4.16272610e-01
5.47745109e-01 6.40506268e-01 5.77797294e-01 -7.64126658e-01
5.04226744e-01 7.24314570e-01 2.24873468e-01 -1.75846554e-02
-1.03265083e+00 -3.34684134e-01 -9.17858034e-02 3.22502673e-01
7.00259209e-01 -8.81338298e-01 -1.48282123e+00 4.09079120e-02
-6.76826060e-01 -4.68525738e-01 -5.93734622e-01 1.05335987e+00
-8.63229871e-01 2.42347315e-01 -2.27892026e-01 -1.37148929e+00
-6.11371286e-02 -3.52593690e-01 7.45406389e-01 3.01386416e-01
-1.23042770e-01 -1.40518820e+00 4.31829870e-01 3.83020602e-02
3.55799720e-02 5.60882330e-01 4.86723483e-01 -5.37607551e-01
7.54088089e-02 -5.50987780e-01 -2.98100740e-01 -1.96178421e-01
1.44684047e-01 -3.19619387e-01 -7.76595354e-01 -3.31239283e-01
3.11996173e-02 7.52473325e-02 6.85502231e-01 1.41902244e+00
1.23121047e+00 -8.59125197e-01 -6.51366115e-01 2.16332540e-01
1.69797122e+00 1.07401796e-01 3.62799317e-01 -1.08615339e-01
-3.58354338e-02 3.82141769e-01 5.73347032e-01 9.58427727e-01
3.68254811e-01 5.36791980e-01 -1.85398862e-01 2.56873876e-01
6.59016609e-01 -5.11297286e-01 2.45831832e-01 -4.53345597e-01
2.81437308e-01 -3.13325584e-01 -3.96279246e-01 1.00910461e+00
-2.01411462e+00 -1.80100334e+00 -1.97233886e-01 3.05411983e+00
7.06355095e-01 -3.50505680e-01 6.23975337e-01 -1.45380855e-01
1.25032032e+00 -5.80975004e-02 -6.88883960e-01 -4.66591299e-01
-4.49054055e-02 2.09090143e-01 1.36777508e+00 7.87357032e-01
-1.01071596e+00 1.25926644e-01 8.44892216e+00 9.26629722e-01
-8.27324167e-02 1.31312862e-01 7.44549453e-01 -1.03831582e-01
-5.35574675e-01 4.00811344e-01 -6.45010054e-01 8.31308663e-01
1.41930795e+00 -1.13419878e+00 4.35742080e-01 7.17321992e-01
8.27040315e-01 -4.21902210e-01 -1.33986104e+00 6.13902092e-01
-3.57964724e-01 -1.32645094e+00 -1.56032890e-01 5.79102337e-01
1.33141875e+00 -3.56834412e-01 1.75981611e-01 2.12526083e-01
1.08316612e+00 -7.83179402e-01 4.42867160e-01 5.99584818e-01
7.26514876e-01 -1.17049563e+00 7.93745875e-01 3.13884914e-01
-7.92433083e-01 -4.75968987e-01 -5.86880147e-01 -5.35197139e-01
3.26738179e-01 1.24524128e+00 -4.93681014e-01 4.92316097e-01
3.83370697e-01 2.79478222e-01 4.54883873e-01 1.25942135e+00
1.52410343e-01 7.87003994e-01 -6.27388120e-01 1.90235928e-01
-6.60442561e-02 -3.26513171e-01 5.17619491e-01 8.53582859e-01
6.32895589e-01 5.22717535e-01 -2.11030114e-02 9.33567524e-01
-2.13643193e-01 -1.14793129e-01 -1.18102050e+00 5.99152267e-01
1.20879519e+00 6.02107048e-01 -1.70304567e-01 -4.88496959e-01
-3.23201895e-01 5.76868594e-01 7.07309395e-02 5.09342432e-01
-8.42464209e-01 -5.38055182e-01 7.77753770e-01 -6.42208084e-02
3.11804354e-01 4.04111505e-01 -4.68507439e-01 -1.03979588e+00
-7.71125108e-02 -4.35643613e-01 1.03321719e+00 -3.56878757e-01
-2.00028419e+00 -8.56006563e-01 6.23497725e-01 -1.04034948e+00
-5.13991475e-01 -4.00755763e-01 -6.89929187e-01 1.02365303e+00
-8.72799933e-01 -3.94744933e-01 5.08635819e-01 3.92762065e-01
-5.79935079e-03 3.26418102e-01 5.65472901e-01 -1.20579593e-01
-4.23806608e-01 8.35016787e-01 1.00091290e+00 -1.92618489e-01
8.04332376e-01 -1.73655903e+00 2.00396597e-01 6.14023149e-01
-4.09954220e-01 6.38126016e-01 9.90709484e-01 -6.18851364e-01
-9.51698184e-01 -8.45671952e-01 9.07325447e-01 -3.36560160e-01
7.63645887e-01 -1.64057970e-01 -4.12698925e-01 1.15252781e+00
1.06419027e-01 1.32310718e-01 9.32790637e-01 6.17372036e-01
-3.46229047e-01 -2.65065670e-01 -1.82981610e+00 6.96362257e-01
1.02371287e+00 -4.02156264e-01 -6.81964397e-01 4.28970635e-01
4.34559077e-01 -4.48843054e-02 -1.18353617e+00 2.35034332e-01
9.20533836e-01 -9.78503883e-01 1.03364265e+00 -6.67240500e-01
-3.62483002e-02 -1.33782312e-01 -4.44160521e-01 -1.39408672e+00
-4.95136112e-01 -8.83674920e-01 1.24173425e-01 1.27114153e+00
3.03759366e-01 -1.22604740e+00 4.69403744e-01 8.92268062e-01
5.84992766e-01 -7.68275335e-02 -1.38465178e+00 -1.16196764e+00
8.49966705e-01 -2.29514882e-01 7.72549689e-01 1.01729608e+00
4.49321538e-01 9.59084108e-02 -4.67713058e-01 2.71998167e-01
1.47171593e+00 3.10456127e-01 4.84076083e-01 -1.12447166e+00
-2.33419329e-01 -3.72463942e-01 -1.55720443e-01 -7.38443911e-01
4.59956735e-01 -5.22607267e-01 -2.10286856e-01 -1.05957878e+00
9.66310799e-01 -8.34553167e-02 -2.24314511e-01 -1.71956997e-02
-4.54261392e-01 -3.11277453e-02 -2.45265722e-01 -4.09269094e-01
-1.85593292e-01 5.77790082e-01 6.88584685e-01 1.62497871e-02
1.97694153e-01 3.71010572e-01 -9.93052185e-01 4.97002870e-01
6.32584751e-01 -7.04791546e-01 -2.58909285e-01 1.10176861e-01
-1.68119714e-01 8.25297952e-01 6.04098678e-01 -5.02069592e-01
2.43219882e-02 -7.73034871e-01 4.85232800e-01 -6.23783171e-01
-2.96704113e-01 -5.39127231e-01 2.99199134e-01 3.88838708e-01
-5.62231839e-01 -2.37842500e-01 -6.16149120e-02 1.06573749e+00
3.16964477e-01 -2.95294344e-01 8.67803037e-01 4.82224897e-02
2.77614594e-01 1.74174130e-01 -3.91447812e-01 2.94290543e-01
1.06790328e+00 -5.50728245e-03 -3.57300878e-01 -7.89931953e-01
-5.73363960e-01 3.12893480e-01 5.23368180e-01 -3.16246599e-01
-7.13789091e-02 -1.80433452e+00 -1.02294183e+00 -5.72710931e-01
-1.89066395e-01 -7.42132068e-01 6.55080855e-01 7.23394513e-01
2.74303377e-01 4.76934582e-01 1.34800285e-01 -1.48374885e-01
-5.34361124e-01 1.00411284e+00 4.31176305e-01 -1.63997397e-01
-3.41323793e-01 3.25037241e-01 4.00401205e-01 -4.51120228e-01
-2.15386584e-01 5.40018547e-03 3.63171637e-01 5.53837270e-02
8.86501193e-01 1.28815234e+00 -4.50576216e-01 -1.70516714e-01
-2.67853022e-01 3.49069208e-01 2.19564468e-01 -6.70751035e-01
1.09681988e+00 -4.48445439e-01 1.30946651e-01 9.99105990e-01
1.35506022e+00 1.41646057e-01 -1.38184059e+00 -1.05773741e-02
-4.85896803e-02 -8.31796825e-01 -2.39930600e-02 -7.01969504e-01
-5.13935566e-01 3.15441608e-01 7.54773557e-01 4.60167795e-01
6.13090217e-01 -2.35594258e-01 3.17798823e-01 3.19313221e-02
5.52937090e-01 -1.00754499e+00 -8.13668430e-01 -1.38529718e-01
7.46033728e-01 -1.17907584e+00 1.95233107e-01 -6.34444728e-02
-1.29297659e-01 5.73884845e-01 -1.46315932e-01 -5.95129788e-01
9.02219117e-01 2.82215774e-01 -3.00274551e-01 3.20187122e-01
-5.74016869e-01 1.93432301e-01 1.11334659e-01 8.87037218e-01
6.01980150e-01 7.49909937e-01 -8.78746450e-01 4.66785192e-01
-1.75822273e-01 1.82897821e-01 7.05124915e-01 4.80111331e-01
-1.56662196e-01 -7.74513245e-01 -7.37508833e-01 7.41414368e-01
-7.69273460e-01 1.60776097e-02 7.96819627e-02 8.55180562e-01
-3.77913415e-01 1.19518161e+00 4.54956353e-01 4.38614875e-01
4.22857255e-01 3.64146642e-02 6.27223849e-01 -1.65977657e-01
-5.04455790e-02 -1.47145420e-01 -1.40078753e-01 -5.65820873e-01
-3.70788008e-01 -1.41523242e+00 -6.48866534e-01 -1.28589880e+00
-4.65880632e-01 3.76941025e-01 6.58459589e-02 8.12389016e-01
2.73998290e-01 -1.64608404e-01 1.19642138e+00 -9.35431361e-01
-1.41175747e+00 -9.16717947e-01 -1.14136875e+00 3.76678884e-01
2.60487854e-01 -7.43801177e-01 -9.12486374e-01 -4.31382447e-01] | [7.996118068695068, 5.233266353607178] |
016f39d3-9df2-425f-bcf5-1d444fe108ec | advancing-learned-video-compression-with-in | 2211.07004 | null | https://arxiv.org/abs/2211.07004v3 | https://arxiv.org/pdf/2211.07004v3.pdf | Advancing Learned Video Compression with In-loop Frame Prediction | Recent years have witnessed an increasing interest in end-to-end learned video compression. Most previous works explore temporal redundancy by detecting and compressing a motion map to warp the reference frame towards the target frame. Yet, it failed to adequately take advantage of the historical priors in the sequential reference frames. In this paper, we propose an Advanced Learned Video Compression (ALVC) approach with the in-loop frame prediction module, which is able to effectively predict the target frame from the previously compressed frames, without consuming any bit-rate. The predicted frame can serve as a better reference than the previously compressed frame, and therefore it benefits the compression performance. The proposed in-loop prediction module is a part of the end-to-end video compression and is jointly optimized in the whole framework. We propose the recurrent and the bi-directional in-loop prediction modules for compressing P-frames and B-frames, respectively. The experiments show the state-of-the-art performance of our ALVC approach in learned video compression. We also outperform the default hierarchical B mode of x265 in terms of PSNR and beat the slowest mode of the SSIM-tuned x265 on MS-SSIM. The project page: https://github.com/RenYang-home/ALVC. | ['Luc van Gool', 'Radu Timofte', 'Ren Yang'] | 2022-11-13 | null | null | null | null | ['ms-ssim'] | ['computer-vision'] | [ 2.72214741e-01 -1.90050989e-01 -2.86505908e-01 -2.14318812e-01
-8.22974384e-01 1.52313545e-01 4.48505133e-01 -2.18422011e-01
-3.85099769e-01 5.54913044e-01 6.31900191e-01 -1.11144498e-01
-8.15839544e-02 -5.47277927e-01 -7.48998523e-01 -8.36405039e-01
-3.97323400e-01 -1.64978489e-01 5.97503901e-01 -2.70206910e-02
4.02652174e-01 1.55622393e-01 -1.51128805e+00 5.91601968e-01
4.94548142e-01 1.09325814e+00 7.67014682e-01 9.85008776e-01
3.36331218e-01 1.18801558e+00 -2.14833334e-01 -3.54604214e-01
2.96296597e-01 -7.15677083e-01 -4.99306828e-01 1.17697939e-01
2.93865472e-01 -7.45923340e-01 -8.12568486e-01 9.12002325e-01
6.50643706e-01 5.49126901e-02 1.90899074e-02 -9.61394310e-01
-1.39392585e-01 6.30334854e-01 -5.83943546e-01 4.04024005e-01
4.08820331e-01 2.80491978e-01 7.35989034e-01 -8.03728580e-01
7.21920907e-01 1.17872727e+00 3.81212085e-01 5.28432608e-01
-7.91751027e-01 -6.68757021e-01 -5.37137054e-02 7.78051853e-01
-1.37690902e+00 -7.44077325e-01 6.96999490e-01 -1.28576472e-01
8.48305345e-01 2.37320781e-01 7.13162839e-01 8.85666490e-01
4.55290258e-01 7.48955548e-01 5.47024906e-01 -2.81335592e-01
2.41553888e-01 -4.99690115e-01 -3.39216650e-01 5.13761938e-01
-2.72201091e-01 3.89957130e-01 -9.42952573e-01 1.20247506e-01
7.57121027e-01 2.44209066e-01 -7.60316551e-01 5.63158169e-02
-1.42789865e+00 4.73944277e-01 2.63929456e-01 2.72314489e-01
-6.49496436e-01 5.25251806e-01 5.14185369e-01 3.08187783e-01
4.42656696e-01 -2.96920240e-01 -2.53384858e-01 -5.62227726e-01
-1.62111354e+00 1.95437536e-01 5.83443880e-01 1.08656371e+00
5.90941668e-01 1.01524077e-01 -3.33031625e-01 7.01597273e-01
4.65909511e-01 3.22396904e-01 6.49572492e-01 -1.45442963e+00
6.10324144e-01 -6.05127849e-02 -1.46428764e-01 -1.00853479e+00
1.83218852e-01 -4.05145526e-01 -1.01429963e+00 9.56273600e-02
-1.35445401e-01 7.41277710e-02 -5.61113954e-01 1.59344327e+00
1.41357556e-01 8.62968504e-01 6.37888238e-02 9.78613377e-01
4.69306558e-01 1.09297860e+00 -1.01068556e-01 -6.86262488e-01
9.74181533e-01 -1.17676556e+00 -7.39579022e-01 2.10180432e-02
3.34891498e-01 -9.47835684e-01 5.31962931e-01 4.13103640e-01
-1.58210230e+00 -9.65315282e-01 -1.14938474e+00 -1.63445860e-01
4.72361922e-01 8.66006687e-03 -2.30402481e-02 1.10753439e-02
-1.18828154e+00 9.66336906e-01 -1.03389657e+00 6.71006320e-03
3.99081051e-01 1.75572991e-01 -1.61415964e-01 -3.92625451e-01
-1.06484318e+00 4.31523979e-01 6.71562731e-01 -3.23293984e-01
-1.11903644e+00 -7.16849923e-01 -7.02322364e-01 3.32408458e-01
4.08788115e-01 -6.64050341e-01 1.12516630e+00 -9.92503941e-01
-1.40157282e+00 2.79618293e-01 -4.34514672e-01 -9.07817185e-01
4.90090966e-01 -4.26330715e-01 -3.78569186e-01 5.72197258e-01
-1.18476205e-01 9.64209735e-01 1.27235198e+00 -9.20238733e-01
-1.01599920e+00 2.00184565e-02 -1.74659729e-01 2.08930716e-01
-9.76625010e-02 -2.33363330e-01 -9.74298716e-01 -1.01383686e+00
-3.22547071e-02 -7.77529955e-01 -1.39916386e-03 1.99760720e-01
6.31655827e-02 1.16898187e-01 1.23896432e+00 -1.01615250e+00
1.80420530e+00 -2.42375803e+00 2.56796181e-01 -2.11907893e-01
1.04482353e-01 4.30928499e-01 -1.47353530e-01 3.89823228e-01
-2.67648380e-02 -1.40530974e-01 -5.21412268e-02 -5.96924067e-01
-2.60134310e-01 2.15633586e-01 -5.11779308e-01 3.22682232e-01
-2.54584223e-01 5.47350764e-01 -8.78473163e-01 -6.90188348e-01
3.22798222e-01 8.59589815e-01 -8.28880966e-01 4.90296900e-01
-9.80716199e-02 4.32406366e-01 6.08102279e-03 3.37549210e-01
7.14321077e-01 -1.19048171e-01 1.19842313e-01 -4.45959479e-01
-1.85060337e-01 2.41451859e-01 -9.17856216e-01 2.02004075e+00
-1.87075436e-01 6.57116890e-01 -5.32786548e-02 -7.57418096e-01
6.71163142e-01 5.15311778e-01 6.07062638e-01 -7.41176307e-01
-1.39363810e-01 3.36693913e-01 -8.77835751e-02 -4.54294235e-01
5.17533600e-01 2.16240898e-01 5.57108879e-01 5.64342737e-02
5.37094176e-02 8.19669440e-02 4.37195122e-01 3.36872995e-01
9.90450740e-01 4.56909239e-01 4.07510251e-01 1.86769627e-02
8.23737860e-01 -5.64048231e-01 6.36712193e-01 3.64239603e-01
-1.35767683e-01 8.36714685e-01 2.75933653e-01 -4.02701110e-01
-1.50032401e+00 -8.25060487e-01 2.12295949e-01 8.83637965e-01
1.64674863e-01 -9.48850334e-01 -7.94337153e-01 -1.88837200e-01
-4.78653699e-01 8.19301844e-01 -3.95353399e-02 -8.31573084e-02
-8.46004426e-01 -8.59810933e-02 2.87302524e-01 1.71428680e-01
1.00870275e+00 -8.01160395e-01 -9.62870777e-01 4.45763916e-01
-5.71496248e-01 -1.13452113e+00 -8.52091312e-01 -3.19117665e-01
-1.11268151e+00 -8.38803232e-01 -1.00700760e+00 -6.91793144e-01
1.12807542e-01 4.36198831e-01 1.04398417e+00 3.17204475e-01
2.40697667e-01 3.18696916e-01 -5.73278725e-01 8.83613899e-02
-5.84891319e-01 -1.91347405e-01 -3.20166260e-01 1.18869938e-01
3.87239680e-02 -8.51493657e-01 -1.10171068e+00 4.68873382e-01
-1.12310815e+00 6.13684177e-01 4.26552087e-01 7.87798405e-01
9.70081091e-01 1.24478273e-01 2.32074097e-01 -2.74018258e-01
8.32958519e-02 -5.38092256e-01 -4.43511158e-01 2.52486970e-02
-5.57036579e-01 -3.23187970e-02 5.53073943e-01 -2.59264231e-01
-8.35525870e-01 -3.08915153e-02 -3.72284830e-01 -8.41142476e-01
1.92765206e-01 4.19616133e-01 1.45365493e-02 1.53287157e-01
1.54608816e-01 6.66755855e-01 3.16132009e-02 -4.47538853e-01
-5.45880906e-02 5.31237602e-01 7.72505224e-01 -3.24367106e-01
4.98251319e-01 3.59380156e-01 1.01169765e-01 -7.39287853e-01
-4.38272864e-01 -3.55763733e-01 -3.74288589e-01 -3.70455384e-01
6.88648582e-01 -1.30721653e+00 -4.08842802e-01 4.32948321e-01
-1.25330794e+00 -3.33582312e-01 -2.43564695e-01 6.68298960e-01
-9.03037190e-01 8.74184310e-01 -7.11094201e-01 -4.25449103e-01
-4.95121092e-01 -1.47130680e+00 1.03728461e+00 1.03588244e-02
8.54617134e-02 -7.63743222e-01 8.59141815e-03 1.62899449e-01
6.08771443e-01 6.72487244e-02 5.64612389e-01 -1.52623251e-01
-9.61848557e-01 1.36829033e-01 5.52805364e-02 5.55822253e-01
-1.10300459e-01 -8.27035531e-02 -6.90088630e-01 -6.54222071e-01
3.65829080e-01 7.32356012e-02 1.08747315e+00 4.75957096e-01
1.23478627e+00 -5.46820164e-01 -5.38885556e-02 9.69207525e-01
1.59101510e+00 3.43292236e-01 1.29863906e+00 2.83205181e-01
4.28461254e-01 1.01411171e-01 7.33816266e-01 8.04726183e-01
2.20963031e-01 8.35536122e-01 5.26196480e-01 3.91359597e-01
-4.87045348e-01 -3.76028955e-01 7.00540066e-01 1.12210119e+00
-4.42043185e-01 -4.72789168e-01 -6.19977534e-01 3.45127255e-01
-1.94869196e+00 -1.34452868e+00 2.70927787e-01 2.28534532e+00
8.87493014e-01 1.67528838e-01 -1.87307060e-01 4.06215191e-01
7.99333215e-01 5.52951992e-01 -4.48291868e-01 -5.00169024e-02
-8.58247206e-02 4.03208993e-02 4.52974319e-01 6.06327236e-01
-8.68683100e-01 6.14358902e-01 5.30629826e+00 1.30553102e+00
-1.13865495e+00 2.09020257e-01 8.01098645e-01 -2.72759438e-01
-5.40482588e-02 2.08997726e-01 -6.69634461e-01 9.18944955e-01
1.30688369e+00 -2.48852670e-01 6.84976935e-01 7.80182481e-01
5.78687727e-01 -3.29268873e-01 -1.15228987e+00 1.39858031e+00
1.04295097e-01 -1.58512044e+00 2.47596920e-01 1.10486686e-01
5.63957036e-01 -4.40262891e-02 1.61686018e-02 1.85591862e-01
-5.14830530e-01 -6.61140859e-01 1.00844622e+00 7.38823175e-01
1.00367999e+00 -7.56367624e-01 7.00828373e-01 4.08119380e-01
-1.44809544e+00 -1.02069467e-01 -5.65322876e-01 1.29915118e-01
5.63307822e-01 6.25342965e-01 -4.28189099e-01 5.03591478e-01
7.36162663e-01 1.09762025e+00 -3.81556213e-01 1.02635801e+00
-6.16060905e-02 8.90959978e-01 -8.10208023e-02 6.69366479e-01
2.43374333e-01 -6.28411248e-02 7.50982702e-01 1.25596225e+00
8.36548507e-01 2.75339127e-01 9.83333960e-03 2.85637319e-01
-3.22628506e-02 -1.25919372e-01 -2.06730887e-01 2.55247682e-01
3.67409319e-01 5.98529994e-01 -2.60621250e-01 -6.92354321e-01
-5.70005715e-01 1.16489255e+00 -1.07900620e-01 2.79548824e-01
-1.10218549e+00 -9.77321416e-02 5.06556809e-01 2.84004152e-01
7.85059154e-01 -2.63765007e-01 2.76892751e-01 -1.17288828e+00
2.05991678e-02 -1.10356009e+00 3.18864644e-01 -9.56042409e-01
-6.30260527e-01 5.00155568e-01 5.92317134e-02 -1.62002218e+00
-5.70422530e-01 -7.61162341e-02 -4.06166434e-01 6.66816294e-01
-1.63372505e+00 -6.89131021e-01 -4.57570910e-01 9.27862167e-01
1.01025307e+00 -3.24678689e-01 4.54576701e-01 5.19873083e-01
-1.15297765e-01 4.62400347e-01 3.01014364e-01 -1.68781593e-01
7.88905203e-01 -5.28328300e-01 1.13305829e-01 1.21558142e+00
-1.49556518e-01 3.02617162e-01 7.33086586e-01 -7.02800333e-01
-1.47086525e+00 -1.18077683e+00 8.11724246e-01 5.57873785e-01
3.07928205e-01 2.55061328e-01 -8.74250412e-01 4.80015814e-01
4.56717312e-01 1.09689429e-01 3.89994651e-01 -9.51485515e-01
-1.61727130e-01 -4.41687495e-01 -1.02576458e+00 4.93014395e-01
9.57326233e-01 -4.32981879e-01 -2.50689179e-01 3.32408585e-02
1.01001215e+00 -4.75932717e-01 -8.72775078e-01 3.20545226e-01
5.07099748e-01 -1.53159571e+00 1.02489030e+00 7.38117993e-02
1.02029550e+00 -4.04674798e-01 -6.36545837e-01 -8.06135476e-01
-4.30411607e-01 -9.29105401e-01 -7.62124658e-01 1.08977425e+00
-1.17276028e-01 -1.13007881e-01 5.80785871e-01 2.95663089e-01
-2.65394449e-01 -9.09582376e-01 -1.16250062e+00 -6.08636737e-01
-4.80215102e-01 -6.19337440e-01 4.52788949e-01 3.47227812e-01
-1.54873699e-01 2.51616966e-02 -9.60355282e-01 3.73205654e-02
6.72532737e-01 -1.21432327e-01 6.67739987e-01 -4.67436641e-01
-7.56150186e-01 -1.19296759e-01 -6.84753656e-01 -1.71953857e+00
-3.23410302e-01 -6.71322227e-01 -7.16784894e-02 -1.19695675e+00
1.93982914e-01 6.69455826e-02 -3.23554486e-01 1.49407193e-01
-4.71306145e-02 1.63368046e-01 7.57268786e-01 5.82273722e-01
-7.63577580e-01 7.11141050e-01 1.23767745e+00 -1.55342713e-01
-5.97244687e-02 -1.63848594e-01 -2.72245646e-01 5.43464482e-01
7.68986821e-01 -3.80087942e-01 -6.00538313e-01 -6.18553221e-01
-1.38957754e-01 6.83806956e-01 3.70052725e-01 -1.46182323e+00
4.39134449e-01 6.40011430e-02 3.11166048e-01 -8.53457570e-01
5.01922846e-01 -9.77204740e-01 5.90806246e-01 8.07212770e-01
-4.28831071e-01 3.88345987e-01 -6.55250847e-02 6.43034995e-01
-5.20938039e-01 -2.73732573e-01 1.05178607e+00 -9.27418694e-02
-8.90941799e-01 5.31209111e-01 -1.44930929e-01 -1.83069736e-01
9.30619657e-01 -2.95215219e-01 -9.55252200e-02 -6.58673108e-01
-5.54169834e-01 -4.19929251e-02 5.73052168e-01 1.22091509e-01
1.12688541e+00 -1.32614434e+00 -9.74525750e-01 1.67965829e-01
-4.07934904e-01 -1.15612902e-01 4.88936812e-01 8.93456340e-01
-8.55259776e-01 2.86205828e-01 -1.74900621e-01 -7.22360253e-01
-1.47932076e+00 7.88244009e-01 7.08279386e-02 -4.18525636e-01
-7.89938748e-01 4.21235263e-01 -3.25834304e-02 7.87966132e-01
4.27521557e-01 -1.28971010e-01 -2.70635020e-02 -3.80849540e-01
1.02054477e+00 6.33084655e-01 -3.25373471e-01 -7.19323218e-01
1.75187923e-02 4.87267047e-01 5.31173162e-02 -2.02670813e-01
1.54191756e+00 -5.39646029e-01 -1.13313556e-01 1.41934961e-01
1.46861649e+00 -9.78112742e-02 -1.59523320e+00 -2.84317076e-01
-2.78239995e-01 -8.95161211e-01 1.57010198e-01 -4.21994537e-01
-1.39856195e+00 7.15062201e-01 8.54761600e-01 -2.93093354e-01
1.69919014e+00 -3.10462654e-01 1.21680379e+00 6.66683689e-02
5.28785408e-01 -8.19302440e-01 1.04385689e-01 4.77077961e-01
1.02758110e+00 -8.49981785e-01 3.96522731e-01 -2.32892349e-01
-5.72624981e-01 1.44815266e+00 7.60666728e-02 -1.39993280e-01
7.61650026e-01 2.40018055e-01 -3.94578397e-01 2.58551151e-01
-1.14228332e+00 1.78536519e-01 5.19122593e-02 4.63281780e-01
4.56490099e-01 -3.27102125e-01 -5.62336862e-01 1.11819357e-01
-5.47865629e-02 4.27603722e-01 5.11624336e-01 8.29348743e-01
-4.79249299e-01 -9.88973081e-01 -3.21692973e-01 1.28495321e-01
-6.20392203e-01 -3.62239063e-01 5.64123750e-01 4.44451630e-01
2.21017778e-01 8.81972849e-01 -1.61755527e-03 -6.51928663e-01
3.23768072e-02 -1.21637434e-01 2.55190581e-01 5.78659736e-02
-2.25723431e-01 5.10793328e-01 -2.15013608e-01 -1.00248694e+00
-7.85480797e-01 -5.31607807e-01 -1.01253402e+00 -6.94941700e-01
1.98922575e-01 2.05301419e-02 4.17816490e-01 4.71710503e-01
5.56150138e-01 4.16116029e-01 7.85564899e-01 -1.30312765e+00
-3.09200644e-01 -7.92975247e-01 -2.94749945e-01 5.13541400e-01
4.87955868e-01 -1.18452244e-01 -4.02162790e-01 6.31881893e-01] | [11.303062438964844, -1.6635611057281494] |
689a8448-12ab-48cd-9b98-c797af06533e | exploiting-asymmetry-for-synthetic-training | 2303.04132 | null | https://arxiv.org/abs/2303.04132v1 | https://arxiv.org/pdf/2303.04132v1.pdf | Exploiting Asymmetry for Synthetic Training Data Generation: SynthIE and the Case of Information Extraction | Large language models (LLMs) show great potential for synthetic data generation. This work shows that useful data can be synthetically generated even for tasks that cannot be solved directly by the LLM: we show that, for problems with structured outputs, it is possible to prompt an LLM to perform the task in the opposite direction, to generate plausible text for the target structure. Leveraging the asymmetry in task difficulty makes it possible to produce large-scale, high-quality data for complex tasks. We demonstrate the effectiveness of this approach on closed information extraction, where collecting ground-truth data is challenging, and no satisfactory dataset exists to date. We synthetically generate a dataset of 1.8M data points, demonstrate its superior quality compared to existing datasets in a human evaluation and use it to finetune small models (220M and 770M parameters). The models we introduce, SynthIE, outperform existing baselines of comparable size with a substantial gap of 57 and 79 absolute points in micro and macro F1, respectively. Code, data, and models are available at https://github.com/epfl-dlab/SynthIE. | ['Robert West', 'Maxime Peyrard', 'Marija Sakota', 'Martin Josifoski'] | 2023-03-07 | null | null | null | null | ['synthetic-data-generation', 'synthetic-data-generation'] | ['medical', 'miscellaneous'] | [ 3.55686605e-01 7.61759341e-01 7.99780246e-03 -3.77665013e-01
-1.48458612e+00 -9.09883142e-01 9.89117444e-01 -1.29413143e-01
-3.86032939e-01 1.10012031e+00 5.13315797e-01 -4.23503131e-01
2.90425241e-01 -4.95580167e-01 -1.09212983e+00 -1.09646738e-01
2.47479826e-01 7.35175490e-01 1.58241577e-02 -3.19708645e-01
2.40758389e-01 3.64354849e-02 -1.42397904e+00 8.41118038e-01
8.21579278e-01 4.70952570e-01 2.86103189e-01 7.54641593e-01
-7.99602270e-02 7.12117136e-01 -8.84304523e-01 -6.15639269e-01
3.87560666e-01 -3.62888068e-01 -1.03399730e+00 -4.27447781e-02
6.69621050e-01 -2.99953043e-01 -2.28046272e-02 5.63188374e-01
5.84490120e-01 -1.34145752e-01 8.03585708e-01 -1.37358785e+00
-6.45697117e-01 1.15393043e+00 -2.56472647e-01 -8.88449177e-02
5.59862196e-01 4.33089375e-01 1.08541965e+00 -1.23526180e+00
9.12750959e-01 1.50160289e+00 6.38200104e-01 9.43972945e-01
-1.50845766e+00 -7.74458349e-01 -1.56435713e-01 -4.53579724e-01
-1.12713420e+00 -8.28852415e-01 2.14607239e-01 -2.74511337e-01
1.10530984e+00 1.98060617e-01 2.69247830e-01 1.81363523e+00
-1.07441232e-01 9.51725066e-01 1.12456012e+00 -5.45733452e-01
-6.50635213e-02 2.66796619e-01 -1.56900883e-01 5.51561952e-01
4.95508134e-01 1.77017629e-01 -6.48270309e-01 -1.79134645e-02
3.77169669e-01 -6.62700295e-01 -1.77930132e-01 2.67165694e-02
-1.55961168e+00 7.00512588e-01 2.26666719e-01 3.30526941e-02
-1.31580085e-01 2.30619118e-01 2.53981352e-01 4.03613865e-01
5.10411501e-01 1.25124991e+00 -7.11372495e-01 -3.71246278e-01
-1.02474463e+00 9.80466902e-01 1.03926575e+00 1.47094464e+00
3.12022328e-01 8.04941654e-02 -5.69319427e-01 7.29354799e-01
1.00521058e-01 6.57627642e-01 5.99019885e-01 -1.20829105e+00
1.06014717e+00 4.07539368e-01 4.72906411e-01 -5.59801519e-01
-3.64880085e-01 -2.75195032e-01 -6.12061977e-01 6.43231496e-02
7.17364550e-01 -4.91785377e-01 -8.52331102e-01 1.78035355e+00
2.94929259e-02 -2.11282521e-01 3.18545073e-01 4.72415060e-01
8.62489402e-01 7.74537504e-01 -9.22287405e-02 1.31415084e-01
1.18572426e+00 -1.04654896e+00 -4.36712980e-01 -4.85381782e-01
8.92960727e-01 -9.70925689e-01 1.59384167e+00 5.10461569e-01
-1.35182440e+00 -4.46456343e-01 -9.12420869e-01 -2.47971550e-01
-2.78851926e-01 1.16023839e-01 5.55811048e-01 4.35892671e-01
-1.13889873e+00 4.90721315e-01 -5.88638306e-01 -2.13483825e-01
4.21785712e-01 4.40841243e-02 -3.25469792e-01 -5.03158793e-02
-1.15434921e+00 8.08620155e-01 6.41120493e-01 -1.39384404e-01
-1.13904738e+00 -8.39163244e-01 -8.57523799e-01 -2.49682620e-01
4.45901901e-01 -7.80331254e-01 1.82220614e+00 -6.81337118e-01
-1.14447832e+00 7.22977638e-01 -3.62596959e-02 -6.61449313e-01
8.34770441e-01 -3.26238364e-01 -1.18161663e-01 -2.46105686e-01
2.51159132e-01 1.37904751e+00 5.31243801e-01 -1.26857102e+00
-4.75644588e-01 1.42718017e-01 6.28087670e-02 1.70302480e-01
-1.27055541e-01 -7.82940350e-03 -3.24917763e-01 -7.80903876e-01
-3.90395463e-01 -9.26276445e-01 -4.29450959e-01 -4.70239699e-01
-8.78325701e-01 -1.92721337e-01 2.73620546e-01 -6.45873964e-01
1.22826171e+00 -1.70709801e+00 -1.19929224e-01 -1.67614609e-01
7.88631365e-02 2.90814608e-01 -5.65997660e-01 5.80546200e-01
1.38083547e-01 6.67955220e-01 -4.38557684e-01 -6.39478028e-01
1.17746718e-01 3.92080769e-02 -4.36101913e-01 -2.96697885e-01
4.97820258e-01 1.32962847e+00 -9.29099977e-01 -5.31078815e-01
-2.49775782e-01 1.11312337e-01 -7.38574564e-01 2.71175802e-01
-7.39264011e-01 2.43957967e-01 -2.25035563e-01 2.85249501e-01
2.79961556e-01 -3.92847180e-01 -3.72062158e-03 1.23314179e-01
4.55477908e-02 7.23144531e-01 -1.12715816e+00 1.73119891e+00
-6.39245391e-01 7.86487699e-01 -1.61968440e-01 -4.87116635e-01
9.13824558e-01 2.85410047e-01 -1.30347908e-01 -5.85899055e-01
-1.43765673e-01 3.71786833e-01 -4.84974422e-02 -3.38337302e-01
8.98191690e-01 1.11809492e-01 -3.21239561e-01 7.26754665e-01
6.91221729e-02 -6.75467551e-01 7.07382798e-01 4.19218630e-01
1.24452078e+00 2.40196809e-01 -4.73843608e-03 -1.67399600e-01
9.34363436e-03 2.70486236e-01 1.70944467e-01 1.01521206e+00
3.63921553e-01 8.37163150e-01 5.18291712e-01 -2.76946396e-01
-1.50106776e+00 -9.52971101e-01 -2.04353221e-02 8.13288152e-01
-3.41240317e-01 -7.76568413e-01 -1.04967928e+00 -7.62009382e-01
5.26055985e-04 1.10582960e+00 -5.03607869e-01 6.43548146e-02
-6.29104316e-01 -6.92371786e-01 1.03022420e+00 4.30337846e-01
2.70685136e-01 -1.45753109e+00 -3.91273588e-01 2.66399711e-01
-5.87075531e-01 -1.27062154e+00 -4.58707929e-01 -1.36830583e-01
-8.73932838e-01 -9.03325200e-01 -5.27796984e-01 -6.15893126e-01
6.12401962e-01 6.58073872e-02 1.77715302e+00 -8.00518505e-03
-1.53328419e-01 -8.29457045e-02 -2.65324563e-01 -6.83819771e-01
-1.12802517e+00 5.53076744e-01 -1.74121946e-01 -6.27385676e-01
1.49467543e-01 -3.49944115e-01 -3.16879898e-01 2.90099859e-01
-1.01872909e+00 7.40968227e-01 7.67472148e-01 9.21236038e-01
4.40339357e-01 -4.25216496e-01 8.33980262e-01 -1.41279101e+00
1.01302075e+00 -5.45890391e-01 -4.32687879e-01 1.59865618e-01
-8.51652145e-01 3.91091913e-01 8.00307930e-01 -3.06636304e-01
-1.02968180e+00 -9.92997885e-02 -1.26816571e-01 6.89162016e-02
-2.45826140e-01 4.13081646e-01 -6.99379891e-02 6.25018418e-01
1.28890514e+00 8.45362023e-02 -5.27825132e-02 -5.98726571e-01
8.36874008e-01 7.79144406e-01 4.47944283e-01 -8.51543427e-01
7.84462452e-01 7.53584951e-02 -3.91658962e-01 -4.75531012e-01
-1.02223337e+00 1.47672206e-01 -3.86530548e-01 2.55407244e-01
4.15679097e-01 -1.08538604e+00 -2.36335307e-01 2.58548826e-01
-1.30984819e+00 -8.56439233e-01 -5.09258986e-01 1.46186188e-01
-7.24199355e-01 -5.16721718e-02 -5.88925481e-01 -4.81366128e-01
-5.46689689e-01 -9.94839311e-01 1.29276943e+00 -1.26922235e-01
-6.52490556e-01 -7.57598102e-01 -7.42286295e-02 4.88283902e-01
3.80589873e-01 2.90669620e-01 7.14137971e-01 -7.55872011e-01
-6.45355880e-01 -5.26681505e-02 -1.62933290e-01 3.41531157e-01
-3.97400968e-02 2.30774134e-01 -8.44599128e-01 -1.60131782e-01
-4.79808599e-01 -8.51793826e-01 7.38456011e-01 5.55404238e-02
1.10459852e+00 -7.32573688e-01 -2.15488091e-01 4.34914142e-01
9.46867764e-01 -3.19806308e-01 5.53391635e-01 2.41514057e-01
6.13426268e-01 7.28717268e-01 7.01425433e-01 4.00451928e-01
6.16305172e-01 6.70507431e-01 6.54261187e-02 -4.84622791e-02
-4.15053934e-01 -8.06587338e-01 5.37831068e-01 6.15750909e-01
2.73221701e-01 -5.40009916e-01 -1.10047340e+00 7.03119338e-01
-1.61510479e+00 -8.53720725e-01 -1.38076201e-01 2.07505727e+00
1.46345007e+00 5.35465479e-01 6.21643849e-02 -6.34422377e-02
3.58653665e-01 -4.78057424e-03 -3.55275065e-01 -4.22230601e-01
-2.58149117e-01 2.24273220e-01 4.22882706e-01 6.82917535e-01
-8.12519073e-01 1.12274563e+00 6.86849594e+00 9.69269335e-01
-6.67397380e-01 -3.33132558e-02 8.36544633e-01 -4.12205040e-01
-8.68043721e-01 3.30523103e-02 -1.14303732e+00 4.98823464e-01
1.37369514e+00 -2.88998008e-01 4.69630718e-01 7.25803018e-01
3.85015160e-01 1.61290035e-01 -1.33518136e+00 7.24269509e-01
-7.48156905e-02 -1.53102684e+00 3.69608194e-01 8.10013153e-03
8.80343139e-01 1.19149752e-01 -2.51829084e-02 5.48859775e-01
8.05606902e-01 -1.42303240e+00 9.94906485e-01 3.81752878e-01
9.64909673e-01 -5.54884255e-01 3.87824059e-01 7.03844309e-01
-6.99951947e-01 1.04203336e-01 -2.78682977e-01 -1.56476542e-01
1.61847010e-01 5.16203582e-01 -1.49419522e+00 2.96872973e-01
3.40412259e-01 4.99290109e-01 -1.05799413e+00 5.17033756e-01
-4.88242179e-01 7.94294894e-01 -3.82549107e-01 -1.86857373e-01
1.44075483e-01 2.06772536e-01 3.88335466e-01 1.42772150e+00
3.03468585e-01 -3.02282631e-01 1.48526691e-02 1.25854099e+00
-4.56792027e-01 8.13688859e-02 -9.81573045e-01 -2.85877317e-01
6.69179797e-01 1.21891356e+00 -2.95867264e-01 -4.28595692e-01
-1.68421820e-01 7.76403010e-01 5.13833761e-01 3.00992936e-01
-6.75163150e-01 -4.10523444e-01 3.05586487e-01 1.55844286e-01
-8.74456838e-02 -1.34879634e-01 -5.82460999e-01 -1.10105979e+00
2.29301721e-01 -1.53203654e+00 2.25000247e-01 -8.96572053e-01
-1.14130127e+00 8.31289113e-01 1.53672040e-01 -1.03649795e+00
-9.33654428e-01 -4.97107595e-01 -2.14829206e-01 1.00556397e+00
-1.09113061e+00 -1.28708136e+00 -1.85531139e-01 1.74226224e-01
9.01315033e-01 -1.53723925e-01 8.47667456e-01 -4.86532301e-02
-2.92969823e-01 9.08943415e-01 -8.63523185e-02 9.30452272e-02
7.55331159e-01 -1.51127172e+00 1.36194491e+00 8.26504707e-01
4.71181095e-01 5.59408307e-01 8.02698791e-01 -6.93057537e-01
-1.21444356e+00 -1.28744793e+00 1.20224965e+00 -1.20665109e+00
4.48114514e-01 -8.42421830e-01 -6.26506627e-01 7.64723361e-01
1.84030980e-01 -3.03516597e-01 3.96611810e-01 4.78588268e-02
-5.10050774e-01 2.21139550e-01 -1.06737411e+00 8.35677087e-01
1.31491804e+00 -2.22992435e-01 -5.38186312e-01 5.81216156e-01
1.22622478e+00 -6.09085262e-01 -6.97747290e-01 3.45449269e-01
5.47969759e-01 -8.16106021e-01 8.00731957e-01 -8.50657523e-01
9.90292966e-01 -2.85495147e-02 -1.96613908e-01 -1.52372015e+00
1.54898435e-01 -8.61302376e-01 -1.40541568e-01 1.30579710e+00
1.30889428e+00 -4.00196820e-01 7.13464499e-01 6.67380214e-01
-1.84025675e-01 -8.68338227e-01 -4.69722390e-01 -9.45925176e-01
3.72984618e-01 -6.63807273e-01 7.93076515e-01 6.58023000e-01
-1.60118118e-01 6.99832976e-01 -2.48014629e-01 -2.57636607e-01
4.62618142e-01 2.96496786e-02 1.25053656e+00 -8.69426548e-01
-5.19870341e-01 -3.71692985e-01 3.52375150e-01 -1.13577926e+00
2.07765192e-01 -1.14440918e+00 1.35438532e-01 -1.63601327e+00
9.14362669e-02 -7.02919126e-01 2.56369621e-01 6.30656302e-01
-2.06851304e-01 2.66505420e-01 4.08323020e-01 1.27715409e-01
-3.49466085e-01 3.85394961e-01 1.33096862e+00 2.01596413e-02
-2.15975538e-01 -1.19705603e-01 -1.19161594e+00 5.86780131e-01
9.22457278e-01 -4.67980415e-01 -5.79097986e-01 -6.69182897e-01
3.73683006e-01 3.74831236e-03 1.63476869e-01 -9.71629560e-01
-2.23069727e-01 -1.40087619e-01 3.37359130e-01 -3.93664896e-01
3.06291312e-01 -2.10910305e-01 1.28685057e-01 2.38810673e-01
-9.07761097e-01 2.47085288e-01 2.55018413e-01 1.94706008e-01
-8.19625780e-02 -3.03979874e-01 4.44140285e-01 -2.52124518e-01
-2.83095211e-01 1.27853051e-01 -1.57649502e-01 7.55981565e-01
6.44545615e-01 1.18256941e-01 -5.99029422e-01 -5.57353616e-01
-3.33350986e-01 3.15312296e-01 5.38880050e-01 7.33236074e-01
5.22814989e-01 -1.32016265e+00 -1.25611579e+00 8.11621696e-02
3.21466893e-01 3.97583455e-01 -2.95625567e-01 3.92484874e-01
-4.62029785e-01 4.68143433e-01 1.69997245e-01 -3.86941105e-01
-9.17436600e-01 4.27536011e-01 1.52350038e-01 -5.09473383e-01
-4.69728947e-01 8.25623274e-01 4.50677350e-02 -6.86086535e-01
5.49701713e-02 -5.39539039e-01 1.04223371e-01 9.91284009e-03
6.22144997e-01 1.53676808e-01 7.15285391e-02 -2.24014282e-01
3.99392247e-02 -6.06136210e-02 -1.18081085e-01 -6.50197983e-01
1.22110939e+00 1.24313854e-01 2.71336675e-01 3.81155461e-01
9.11266088e-01 1.81041360e-01 -1.42585492e+00 -2.10812956e-01
2.96795309e-01 -3.77407610e-01 -5.54268301e-01 -1.15870965e+00
-5.04510641e-01 7.86087334e-01 -9.42218825e-02 1.58639744e-01
8.27941179e-01 1.03733175e-01 7.69458771e-01 6.67648613e-01
4.40079570e-01 -8.47764909e-01 2.31116787e-01 6.85434997e-01
1.37283659e+00 -1.23941362e+00 -3.22599530e-01 -3.28106523e-01
-8.73805523e-01 8.53934288e-01 7.67548621e-01 1.06337056e-01
8.99867341e-02 5.82700431e-01 8.24984387e-02 5.65713644e-02
-1.40448999e+00 1.29573390e-01 1.78611115e-01 5.96221149e-01
7.07620084e-01 2.35909924e-01 -1.11679323e-01 6.88132584e-01
-9.55876946e-01 -7.29938969e-02 9.18349922e-01 6.89856648e-01
-3.50795597e-01 -1.32186830e+00 -2.16748700e-01 6.69315279e-01
-4.57355469e-01 -4.80712503e-01 -5.28370023e-01 8.96193385e-01
-1.96803227e-01 1.06816673e+00 -1.65992543e-01 -3.14909846e-01
4.91921365e-01 2.02247456e-01 4.43292141e-01 -9.18028951e-01
-4.74836528e-01 -3.31640720e-01 7.23224819e-01 -5.52749813e-01
4.24821936e-02 -5.72216034e-01 -1.14050245e+00 -2.73349047e-01
-3.34303098e-04 1.19640030e-01 6.55485511e-01 6.37535512e-01
5.90852618e-01 2.49761105e-01 5.79483628e-01 -7.33492136e-01
-8.40367317e-01 -1.37407959e+00 -2.96431612e-02 6.56687975e-01
1.89252794e-01 -2.60972351e-01 -2.56098211e-01 2.61724204e-01] | [11.419493675231934, 8.81213665008545] |
043d33b8-164e-4d53-a8d8-bee147e8f35d | few-shot-class-incremental-audio-1 | 2305.19539 | null | https://arxiv.org/abs/2305.19539v1 | https://arxiv.org/pdf/2305.19539v1.pdf | Few-shot Class-incremental Audio Classification Using Dynamically Expanded Classifier with Self-attention Modified Prototypes | Most existing methods for audio classification assume that the vocabulary of audio classes to be classified is fixed. When novel (unseen) audio classes appear, audio classification systems need to be retrained with abundant labeled samples of all audio classes for recognizing base (initial) and novel audio classes. If novel audio classes continue to appear, the existing methods for audio classification will be inefficient and even infeasible. In this work, we propose a method for few-shot class-incremental audio classification, which can continually recognize novel audio classes without forgetting old ones. The framework of our method mainly consists of two parts: an embedding extractor and a classifier, and their constructions are decoupled. The embedding extractor is the backbone of a ResNet based network, which is frozen after construction by a training strategy using only samples of base audio classes. However, the classifier consisting of prototypes is expanded by a prototype adaptation network with few samples of novel audio classes in incremental sessions. Labeled support samples and unlabeled query samples are used to train the prototype adaptation network and update the classifier, since they are informative for audio classification. Three audio datasets, named NSynth-100, FSC-89 and LS-100 are built by choosing samples from audio corpora of NSynth, FSD-MIX-CLIP and LibriSpeech, respectively. Results show that our method exceeds baseline methods in average accuracy and performance dropping rate. In addition, it is competitive compared to baseline methods in computational complexity and memory requirement. The code for our method is given at https://github.com/vinceasvp/FCAC. | ['Emmanouil Benetos', 'Jialong Li', 'Wei Xie', 'Wenchang Cao', 'Yanxiong Li'] | 2023-05-31 | null | null | null | null | ['audio-classification'] | ['audio'] | [ 2.50898242e-01 -1.11566477e-01 -7.65256360e-02 -4.27151442e-01
-8.58477116e-01 -6.42792106e-01 8.09823647e-02 2.26914391e-01
-4.31246519e-01 6.27563238e-01 6.09828681e-02 2.58508831e-01
1.96827412e-01 -6.68046474e-01 -4.32885826e-01 -6.11669838e-01
-3.36672127e-01 4.59403813e-01 5.69340527e-01 -2.97372080e-02
-5.65760694e-02 3.22852522e-01 -2.09240460e+00 5.96541822e-01
3.80879849e-01 1.20397556e+00 8.79049748e-02 8.53854239e-01
-1.11666374e-01 5.88638127e-01 -6.91623569e-01 1.44534290e-01
1.58099115e-01 -4.30657089e-01 -1.06641543e+00 6.45279288e-02
2.78793514e-01 -2.15837315e-01 -3.94618303e-01 7.09665775e-01
8.75858843e-01 5.83430827e-01 2.08297044e-01 -1.46771300e+00
-1.90215558e-01 1.09604776e+00 -1.28047377e-01 2.93552667e-01
4.36170936e-01 -7.18098134e-02 1.01887250e+00 -1.30866647e+00
5.60544908e-01 1.01118469e+00 7.15331554e-01 7.96717703e-01
-9.53942120e-01 -1.11043799e+00 4.94744420e-01 6.96409047e-01
-1.67932737e+00 -8.75017405e-01 9.87983584e-01 -3.46564442e-01
6.59136474e-01 4.72179592e-01 9.84208822e-01 9.51710939e-01
-4.25597757e-01 9.36288893e-01 4.30017978e-01 -5.90927541e-01
6.08858943e-01 3.62824619e-01 4.72573459e-01 1.58403531e-01
-4.17458773e-01 -6.50724918e-02 -7.96392441e-01 -1.84833929e-01
1.47716433e-01 5.54820523e-02 -4.86361951e-01 -6.10961430e-02
-7.76201963e-01 4.94240940e-01 2.48883098e-01 4.07628328e-01
-2.83762425e-01 -9.59029645e-02 7.47731924e-01 7.52514899e-01
5.95263660e-01 1.51022941e-01 -5.56489706e-01 -3.79231721e-01
-9.60075080e-01 1.31622002e-01 6.66799784e-01 1.15018129e+00
7.34774172e-01 3.57964933e-01 1.12431258e-01 1.28554893e+00
-8.75576437e-02 1.32864654e-01 8.83252203e-01 -7.38521457e-01
2.59800136e-01 5.11643469e-01 -3.58131975e-01 -6.32856846e-01
-3.07619810e-01 -5.53011477e-01 -6.99661613e-01 -3.19695085e-01
-2.92522669e-01 -9.29873809e-02 -8.94609988e-01 1.67634475e+00
6.47272289e-01 6.73237264e-01 3.18654478e-02 5.78405797e-01
1.01914918e+00 9.89139140e-01 -4.09832411e-02 -4.04423088e-01
1.01498020e+00 -9.98949945e-01 -4.78583068e-01 -3.35288532e-02
3.03897083e-01 -7.06461489e-01 1.12576973e+00 5.14829397e-01
-9.12748337e-01 -1.03053248e+00 -1.04084015e+00 3.39768082e-01
-4.19892192e-01 2.81585078e-03 1.70513287e-01 3.59441966e-01
-9.12221432e-01 5.12521505e-01 -6.29496157e-01 -3.00420463e-01
3.78026247e-01 2.97808886e-01 -1.39120355e-01 8.25138912e-02
-1.21946967e+00 -3.79491784e-02 8.69542420e-01 1.13345467e-01
-1.35972714e+00 -7.36357868e-01 -6.91083252e-01 1.68741599e-01
4.25995141e-01 -1.10014290e-01 1.60933900e+00 -1.15810406e+00
-1.55448020e+00 3.77471119e-01 -1.55626945e-02 -3.81560415e-01
1.52186096e-01 -7.84756467e-02 -8.43705177e-01 3.31008524e-01
1.86116751e-02 8.33675742e-01 1.06769001e+00 -1.03218329e+00
-1.19516754e+00 -1.72244236e-02 2.01678146e-02 2.39721015e-01
-7.00276434e-01 -6.86990470e-02 -2.45333925e-01 -6.70767605e-01
2.29344219e-01 -8.74990106e-01 1.97038651e-01 -1.95737451e-01
-3.20305705e-01 -3.60577852e-01 9.72782254e-01 -3.35426003e-01
1.59463239e+00 -2.57757449e+00 6.34531258e-03 1.01547666e-01
8.18982646e-02 2.92257577e-01 -4.63419348e-01 4.31630909e-01
-3.83525908e-01 -1.15821518e-01 -5.31823896e-02 -2.52855659e-01
-1.29913926e-01 6.70194775e-02 -7.76053071e-01 -4.53359932e-02
-3.14607173e-02 4.34102654e-01 -9.56031263e-01 -3.61402541e-01
-6.81855083e-02 2.71378636e-01 -6.61184430e-01 2.56425798e-01
-5.46114519e-02 2.38163486e-01 -9.72966999e-02 8.01081181e-01
4.44985002e-01 1.09587938e-01 9.70721990e-02 -7.26161897e-02
-3.48755717e-02 4.18322325e-01 -1.52195001e+00 1.58421218e+00
-3.53832901e-01 3.58140737e-01 -1.46281093e-01 -1.15684307e+00
8.50923121e-01 6.90739572e-01 4.02439356e-01 -1.50834218e-01
-1.02061518e-01 1.98487759e-01 -6.47548288e-02 -4.00894970e-01
4.27110374e-01 -1.60167634e-01 -1.84047535e-01 4.08305705e-01
5.53101540e-01 -7.28024021e-02 3.74864817e-01 1.40405491e-01
1.10852540e+00 -2.25394607e-01 5.53385243e-02 2.41372287e-01
6.96418285e-01 -2.86085278e-01 9.64669704e-01 6.31880224e-01
-1.71260342e-01 4.47728038e-01 -7.84082860e-02 -6.87695205e-01
-7.94504166e-01 -1.12640917e+00 -1.70092896e-01 1.55935514e+00
-2.22075522e-01 -8.19742441e-01 -5.43870032e-01 -6.24938130e-01
-1.81277215e-01 4.51449603e-01 -2.84408331e-01 -4.35785830e-01
-5.81647336e-01 -3.11964959e-01 6.71360791e-01 4.94879276e-01
3.45605165e-01 -1.47871685e+00 -3.85254055e-01 7.03239799e-01
-2.75095701e-01 -8.69172573e-01 -4.99694884e-01 3.25806379e-01
-8.96700621e-01 -1.04667175e+00 -4.35321242e-01 -1.01257253e+00
4.44545090e-01 1.23587325e-01 7.80927896e-01 6.98909210e-03
-2.64586389e-01 4.86565769e-01 -7.92711437e-01 -5.60684264e-01
-4.17569190e-01 4.87143487e-01 4.51894879e-01 3.28392476e-01
2.54952371e-01 -9.76823568e-01 -3.19914252e-01 2.12506145e-01
-9.25271571e-01 -1.47253856e-01 2.35864997e-01 7.98797965e-01
8.00270736e-01 2.36800179e-01 1.02044249e+00 -7.38584936e-01
3.94271910e-01 -4.61215347e-01 -2.48901621e-01 5.87939695e-02
-3.68743628e-01 -5.66743553e-01 9.38572466e-01 -1.01222038e+00
-6.85975552e-01 2.39901111e-01 -2.66843945e-01 -6.05206728e-01
-3.05209845e-01 4.92307931e-01 -4.00669798e-02 2.88558334e-01
5.91949880e-01 4.68524873e-01 -4.36504781e-01 -8.11779618e-01
1.20582722e-01 1.10933661e+00 4.47326571e-01 -4.54137534e-01
8.42010021e-01 1.37458324e-01 -7.61783063e-01 -1.12865829e+00
-8.02491903e-01 -5.86654484e-01 -6.59870744e-01 -5.02507448e-01
1.96086407e-01 -1.04294276e+00 -2.51472652e-01 5.73835611e-01
-7.93104112e-01 -2.47172773e-01 -1.03395534e+00 5.57342887e-01
-2.83275574e-01 3.64335217e-02 -6.76752269e-01 -8.30388010e-01
-3.61212343e-01 -8.80310237e-01 9.16855395e-01 2.36006185e-01
-2.72625715e-01 -4.89839613e-01 2.53129542e-01 -8.15756917e-02
3.20817322e-01 -2.24707514e-01 8.09527099e-01 -9.47143018e-01
-4.10435617e-01 -4.02244657e-01 4.98123348e-01 5.70307910e-01
2.45036229e-01 -1.20275132e-01 -1.45042133e+00 -6.41965151e-01
-1.49809808e-01 -5.78128815e-01 7.96521604e-01 -7.32752122e-03
1.52305067e+00 -4.86863881e-01 -2.51659960e-01 4.86948133e-01
9.05005634e-01 7.17217088e-01 3.18310976e-01 1.74824838e-02
2.83122748e-01 2.13921815e-01 4.77543622e-01 6.19906366e-01
2.45882526e-01 6.69227064e-01 -6.20067157e-02 2.51663864e-01
-2.13188991e-01 -3.22557509e-01 6.23304188e-01 1.82919335e+00
9.51981097e-02 -1.32399546e-02 -7.37828016e-01 6.47617459e-01
-1.55335414e+00 -1.08986688e+00 5.54027021e-01 2.21035433e+00
1.21725130e+00 1.61217153e-01 2.36643001e-01 8.87066126e-01
8.55230153e-01 -9.62964147e-02 -6.60507202e-01 -1.08311594e-01
1.96883053e-01 5.91921806e-01 -5.18685997e-01 3.12435418e-01
-1.16796446e+00 8.79996777e-01 6.02115822e+00 8.57311010e-01
-1.11754954e+00 2.56910503e-01 3.47394586e-01 -4.90320742e-01
1.35573685e-01 1.05431639e-01 -9.87529993e-01 4.45864022e-01
1.22410691e+00 -2.59083867e-01 4.13437068e-01 9.37508285e-01
-1.72101185e-01 2.36876786e-01 -1.25340986e+00 1.09555638e+00
1.36604771e-01 -1.18933523e+00 1.83310539e-01 -5.32898962e-01
3.42905849e-01 -5.42955622e-02 -1.28955338e-02 8.67248356e-01
-8.38029161e-02 -2.55157799e-01 8.02826464e-01 4.63883579e-01
9.60465968e-01 -8.77754509e-01 6.68902278e-01 2.01241955e-01
-1.74319792e+00 -4.02391225e-01 -4.55796123e-01 -3.41348015e-02
-1.45455054e-03 5.05344272e-01 -9.42624390e-01 4.98726636e-01
1.11816740e+00 8.56892884e-01 -6.15473926e-01 1.22394323e+00
-1.09992914e-01 1.07149255e+00 -4.46113497e-01 1.77445352e-01
-4.14320603e-02 2.80591697e-01 6.08815014e-01 1.09353888e+00
3.71033788e-01 5.48168533e-02 4.95274961e-01 2.88181663e-01
-2.78489292e-01 2.86922395e-01 -2.51538396e-01 9.81576070e-02
1.06785691e+00 1.21215260e+00 -6.63233697e-01 -6.69791818e-01
-1.16207756e-01 7.68961847e-01 1.41700819e-01 3.42917800e-01
-7.08493173e-01 -8.81479144e-01 3.78250837e-01 4.12091166e-02
3.81356239e-01 3.44902307e-01 5.16494989e-01 -1.25803125e+00
1.89535230e-01 -9.86889780e-01 6.79004669e-01 -6.78601682e-01
-1.21801794e+00 9.85037148e-01 -7.22206943e-03 -1.73241735e+00
-4.34881836e-01 4.54410119e-03 -5.50282419e-01 2.58367419e-01
-1.23523867e+00 -8.43801975e-01 -4.10925597e-01 6.61238194e-01
9.77210343e-01 -5.72276592e-01 1.06846917e+00 7.15449870e-01
-4.92136836e-01 7.89459348e-01 -2.55324636e-02 2.14336962e-01
6.62059009e-01 -8.79399598e-01 1.44119605e-01 5.40623128e-01
4.91357535e-01 1.71439931e-01 3.23467255e-01 -3.74636799e-01
-9.48333621e-01 -1.34427261e+00 8.33805978e-01 2.81912535e-02
6.31065249e-01 -6.78503633e-01 -1.10558748e+00 7.22513258e-01
-1.48844019e-01 5.33761792e-02 1.05019033e+00 1.55702755e-01
-2.81914949e-01 -7.02741921e-01 -7.56769538e-01 3.07649314e-01
1.06225157e+00 -6.54616416e-01 -7.69462049e-01 3.20620537e-01
1.06050336e+00 -2.59210736e-01 -7.81660497e-01 3.93197179e-01
6.10112011e-01 -4.16533023e-01 7.82925725e-01 -5.68717182e-01
-1.47743270e-01 -4.24159527e-01 -2.84674644e-01 -1.30287719e+00
-2.01075852e-01 -5.92432559e-01 -2.50652671e-01 1.44697833e+00
4.12384748e-01 -5.48188686e-01 5.90470791e-01 -1.29937842e-01
-3.97442281e-01 -6.24395311e-01 -1.32228744e+00 -9.90718603e-01
-2.38621041e-01 -7.56887317e-01 8.71345043e-01 1.09817350e+00
1.33972540e-01 5.20752072e-01 -2.34918147e-01 2.72835866e-02
3.43295723e-01 1.99485466e-01 7.41914272e-01 -1.51133513e+00
-3.15314412e-01 -6.84749559e-02 -6.54394329e-01 -8.18573773e-01
6.46939874e-02 -1.13033986e+00 3.93251032e-02 -9.15945351e-01
-1.90975256e-02 -6.88728929e-01 -6.80188596e-01 9.28136647e-01
2.37231553e-01 2.17017069e-01 2.03630865e-01 3.44488174e-01
-7.14643478e-01 7.59837806e-01 6.78931177e-01 -4.11637247e-01
-7.85210967e-01 3.86925429e-01 -3.31396729e-01 7.17112482e-01
1.03262389e+00 -7.18346775e-01 -6.62397385e-01 -7.34721720e-02
-7.29180872e-02 -1.67703837e-01 7.53781125e-02 -1.64522040e+00
3.69810641e-01 1.78589523e-01 4.31256801e-01 -8.06228817e-01
5.02705157e-01 -8.08844984e-01 3.49604100e-01 2.93909431e-01
-6.56254411e-01 -1.06152529e-02 3.60919982e-01 4.02913600e-01
-4.64744747e-01 -5.25137603e-01 6.75224304e-01 9.48607847e-02
-7.13150918e-01 5.74288130e-01 -4.90584940e-01 1.02281138e-01
9.03926313e-01 -2.45604709e-01 7.03895465e-02 -3.59597594e-01
-1.28640580e+00 9.98409390e-02 -7.30047226e-02 7.29039490e-01
9.74328041e-01 -1.73272634e+00 -6.20572388e-01 4.65318143e-01
2.54544586e-01 1.34956956e-01 4.43878651e-01 4.46465999e-01
-1.06151380e-01 2.23230906e-02 2.36516465e-02 -6.19540572e-01
-1.37862325e+00 6.26099586e-01 9.66834202e-02 1.22761101e-01
-6.51663482e-01 9.64219749e-01 -3.91344316e-02 -6.09020889e-01
6.80549979e-01 -4.05287564e-01 -2.43229106e-01 4.54036534e-01
8.51223409e-01 2.69462973e-01 1.00037396e-01 -4.38665211e-01
-2.69582897e-01 2.98933864e-01 -3.73748541e-01 -7.01695085e-02
1.69347620e+00 -5.34858480e-02 1.17485644e-02 1.19523799e+00
1.13223827e+00 -1.56654969e-01 -7.80000448e-01 -5.12913167e-01
-1.20983779e-01 -1.66520983e-01 -2.56920308e-01 -6.26234829e-01
-1.12286246e+00 9.49252009e-01 1.01552236e+00 2.81393945e-01
1.51686120e+00 2.99880393e-02 7.36416280e-01 5.56552529e-01
3.74268860e-01 -1.35540283e+00 1.73037827e-01 5.96164584e-01
9.01739538e-01 -5.88764310e-01 -2.04965904e-01 -4.49564271e-02
-3.41091096e-01 1.17246532e+00 6.39052570e-01 6.75780997e-02
9.83797133e-01 1.78050131e-01 3.82241793e-02 1.66131034e-01
-1.08054399e+00 9.00838990e-03 2.52021164e-01 5.28237820e-01
1.77422166e-01 -1.05167255e-02 2.95781553e-01 8.52021039e-01
-5.36206365e-01 -2.37454344e-02 4.48641509e-01 1.12696290e+00
-7.04060018e-01 -1.18756163e+00 -1.36511043e-01 4.18408155e-01
-9.53683853e-02 -1.18661560e-01 -2.38820210e-01 6.38028204e-01
3.89663547e-01 8.44097853e-01 3.06175292e-01 -8.52507770e-01
5.07606983e-01 6.52353287e-01 7.69921243e-02 -9.29042399e-01
-5.17125607e-01 2.20159218e-01 -1.49771735e-01 -2.62653828e-01
-3.83363783e-01 -5.34696996e-01 -1.10702515e+00 1.17906556e-01
-6.04589343e-01 6.42091274e-01 1.83312491e-01 5.55293441e-01
4.35254127e-01 7.22612619e-01 1.04940057e+00 -7.98676789e-01
-4.30935949e-01 -1.16129422e+00 -7.22006917e-01 8.43225196e-02
3.71906519e-01 -5.64314008e-01 -5.06560445e-01 4.53752249e-01] | [15.096755027770996, 5.18921422958374] |
4a490d47-3085-4a6f-b6e7-1b42895eb024 | semi-supervised-medical-image-classification | 2005.07377 | null | https://arxiv.org/abs/2005.07377v1 | https://arxiv.org/pdf/2005.07377v1.pdf | Semi-supervised Medical Image Classification with Relation-driven Self-ensembling Model | Training deep neural networks usually requires a large amount of labeled data to obtain good performance. However, in medical image analysis, obtaining high-quality labels for the data is laborious and expensive, as accurately annotating medical images demands expertise knowledge of the clinicians. In this paper, we present a novel relation-driven semi-supervised framework for medical image classification. It is a consistency-based method which exploits the unlabeled data by encouraging the prediction consistency of given input under perturbations, and leverages a self-ensembling model to produce high-quality consistency targets for the unlabeled data. Considering that human diagnosis often refers to previous analogous cases to make reliable decisions, we introduce a novel sample relation consistency (SRC) paradigm to effectively exploit unlabeled data by modeling the relationship information among different samples. Superior to existing consistency-based methods which simply enforce consistency of individual predictions, our framework explicitly enforces the consistency of semantic relation among different samples under perturbations, encouraging the model to explore extra semantic information from unlabeled data. We have conducted extensive experiments to evaluate our method on two public benchmark medical image classification datasets, i.e.,skin lesion diagnosis with ISIC 2018 challenge and thorax disease classification with ChestX-ray14. Our method outperforms many state-of-the-art semi-supervised learning methods on both single-label and multi-label image classification scenarios. | ['Lequan Yu', 'Quande Liu', 'Qi Dou', 'Luyang Luo', 'Pheng Ann Heng'] | 2020-05-15 | null | null | null | null | ['multi-label-image-classification', 'semi-supervised-medical-image-classification'] | ['computer-vision', 'medical'] | [ 5.39099455e-01 4.11192566e-01 -6.22328699e-01 -9.00885940e-01
-1.00824428e+00 -3.70422244e-01 1.50114670e-01 2.86754698e-01
-1.90377414e-01 7.85083711e-01 -1.59716681e-01 -2.98337042e-01
-1.01739056e-01 -2.98344791e-01 -7.34062850e-01 -7.78477311e-01
3.12833160e-01 7.99315512e-01 -6.44936711e-02 2.62825906e-01
-4.10470098e-01 1.07318885e-03 -1.23308969e+00 5.62904596e-01
9.29239750e-01 1.41876125e+00 -1.79073289e-01 2.68748313e-01
5.62799396e-03 1.36262679e+00 -3.48020047e-01 -4.84473407e-01
1.93878710e-01 -5.78578472e-01 -1.19697618e+00 5.95939279e-01
4.29491132e-01 -2.38969073e-01 2.10419372e-02 1.36332381e+00
4.48735982e-01 -3.22585613e-01 8.43463540e-01 -1.33121073e+00
-9.10758793e-01 5.84329009e-01 -5.52757084e-01 -4.28423844e-02
-1.70649022e-01 1.20357960e-01 9.44999695e-01 -6.35820270e-01
8.52531433e-01 7.51382649e-01 8.92691553e-01 8.66057694e-01
-1.09938991e+00 -7.37788200e-01 1.34560108e-01 3.22884828e-01
-1.39552307e+00 -2.15638325e-01 7.57651925e-01 -4.58022892e-01
4.21077818e-01 2.75903344e-01 3.37990254e-01 1.29066694e+00
-8.74351989e-03 8.95149589e-01 1.22701895e+00 -3.50794375e-01
4.18853372e-01 2.28584915e-01 4.33836490e-01 9.34160292e-01
2.17364892e-01 2.09354833e-02 -3.06269467e-01 -2.97430307e-01
3.06775540e-01 2.34472886e-01 -3.59294534e-01 -4.14195150e-01
-1.15873849e+00 4.80162233e-01 6.26962841e-01 9.83530357e-02
-3.25622678e-01 -1.19736284e-01 6.91345572e-01 1.75420955e-01
7.30033875e-01 3.97302210e-01 -7.22516835e-01 3.31355035e-01
-9.77099657e-01 -2.44773582e-01 5.48564732e-01 1.07015574e+00
5.04398048e-01 -3.32454354e-01 -3.57879072e-01 1.03220332e+00
3.83726031e-01 2.91731000e-01 7.41076171e-01 -8.87279212e-01
2.41032451e-01 8.42179656e-01 -2.53464073e-01 -6.18716657e-01
-4.99355972e-01 -6.72561705e-01 -1.43482733e+00 -7.27207735e-02
2.44445801e-01 1.14446739e-02 -1.33084095e+00 1.78285670e+00
6.45875692e-01 3.37452739e-01 2.17981473e-01 8.83371115e-01
1.19892132e+00 1.50209069e-02 2.07998976e-01 -3.28270435e-01
1.36595440e+00 -1.19778001e+00 -9.38950956e-01 7.38694519e-02
9.91966426e-01 -3.34664375e-01 8.19760621e-01 3.81644487e-01
-7.94474006e-01 -4.63547289e-01 -9.72798586e-01 1.04540654e-01
-6.90802559e-02 3.40180337e-01 3.96996081e-01 3.17064315e-01
-8.36420536e-01 5.93446791e-01 -9.37107980e-01 -1.44532010e-01
8.83741200e-01 2.06657872e-01 -5.73509872e-01 -2.89704502e-01
-1.18814731e+00 8.36564183e-01 5.29744148e-01 5.00223227e-02
-8.87766182e-01 -8.57029438e-01 -8.67842317e-01 -2.11920261e-01
5.67366719e-01 -6.61818266e-01 1.43055928e+00 -1.29669750e+00
-1.03434050e+00 1.40247929e+00 -8.51603225e-03 -4.38223273e-01
6.95976257e-01 1.39197230e-01 -4.33271796e-01 3.55465770e-01
2.24185735e-01 9.25140142e-01 7.56309032e-01 -1.44431221e+00
-5.27057469e-01 -2.81105876e-01 -3.16115081e-01 6.37793764e-02
-3.57939661e-01 -2.95112818e-01 -3.99694949e-01 -6.85788870e-01
3.49247962e-01 -1.14633834e+00 -2.56983221e-01 4.20140028e-01
-8.97006273e-01 -3.63731086e-01 7.36943126e-01 -4.30410862e-01
9.11358476e-01 -2.14958024e+00 -1.21899314e-01 2.14761421e-01
5.55678785e-01 2.45208457e-01 -5.31058647e-02 -2.84827411e-01
-3.75173002e-01 1.54315874e-01 -5.64102113e-01 -5.53651094e-01
-1.91726491e-01 5.23591042e-01 -1.76989138e-01 5.33200920e-01
4.46373254e-01 1.05391324e+00 -9.07841623e-01 -1.02572274e+00
-3.16234827e-02 2.85362959e-01 -2.67839372e-01 2.72222281e-01
-3.10295582e-01 7.69705772e-01 -3.76359880e-01 1.12907219e+00
4.74604815e-01 -1.24471569e+00 4.17382896e-01 -4.76521015e-01
9.14988935e-01 4.67978306e-02 -8.07596326e-01 1.64290822e+00
-2.60800868e-01 1.13794960e-01 -1.70494959e-01 -1.26891875e+00
6.53676391e-01 6.62099123e-01 8.61314893e-01 -4.13016438e-01
1.94029748e-01 2.99290121e-01 -9.68372151e-02 -6.90680742e-01
-1.42331064e-01 -3.01978588e-01 2.08434165e-01 5.73278666e-01
1.36625886e-01 3.18513157e-05 -7.09939376e-02 3.09394628e-01
1.10270190e+00 4.35732603e-02 4.51294959e-01 -3.10118616e-01
3.59324783e-01 1.58469379e-01 7.85211086e-01 5.71558475e-01
-6.08842492e-01 7.71336854e-01 2.57441401e-01 -6.71062529e-01
-9.26625133e-01 -8.06391060e-01 -5.34005582e-01 6.53605998e-01
2.91973114e-01 -2.07283989e-01 -6.50251985e-01 -1.51718080e+00
-7.43964538e-02 2.82944232e-01 -1.08536255e+00 -2.73960859e-01
-1.45383000e-01 -8.47032368e-01 5.45868039e-01 7.34487474e-01
5.22797644e-01 -9.08885121e-01 -1.02730751e-01 -3.06267012e-02
-3.92546415e-01 -1.24244881e+00 -4.60686237e-01 5.51469743e-01
-7.37994909e-01 -1.46092272e+00 -4.73046273e-01 -9.05561149e-01
1.25183499e+00 -8.40752870e-02 1.27275217e+00 4.25907463e-01
-4.66949582e-01 2.28409842e-01 -4.31836784e-01 -2.81738430e-01
-7.44216144e-01 -8.80263671e-02 4.65251096e-02 2.50578653e-02
3.89851630e-01 -1.49438366e-01 -4.80409920e-01 4.55967635e-01
-1.10094273e+00 4.14101452e-01 5.37205517e-01 1.36780024e+00
1.17489541e+00 1.32795036e-01 6.28413558e-01 -1.46034968e+00
1.14959531e-01 -6.23464644e-01 -4.63782176e-02 7.40217090e-01
-9.74678278e-01 8.59923661e-02 7.03527331e-01 -6.14652991e-01
-9.08925891e-01 3.54876757e-01 -7.71929175e-02 -8.05494189e-01
-3.06830525e-01 4.87020671e-01 9.27530453e-02 1.94939032e-01
9.23330069e-01 5.36929630e-02 1.11599140e-01 -3.33060354e-01
2.11739108e-01 8.77559185e-01 6.49584770e-01 -5.70658863e-01
4.34653014e-01 7.08846629e-01 6.95844442e-02 3.99166420e-02
-1.68488479e+00 -6.14592254e-01 -6.51206791e-01 -1.30142510e-01
8.04927289e-01 -9.35225904e-01 -4.10457432e-01 4.63866860e-01
-7.84988761e-01 -3.86594534e-01 -4.36101615e-01 3.66711974e-01
-3.79734993e-01 3.08045715e-01 -8.08789074e-01 -2.81837314e-01
-6.06539071e-01 -1.34566903e+00 1.34975290e+00 -7.62153342e-02
-3.58506769e-01 -1.14944661e+00 -7.37537518e-02 6.10414803e-01
2.05548164e-02 3.45586270e-01 8.28184664e-01 -1.10257149e+00
-1.47042587e-01 -2.76052773e-01 -3.29284161e-01 5.20570874e-01
5.66588104e-01 -2.03252807e-01 -1.01121151e+00 -2.96802431e-01
8.23282599e-02 -1.16336668e+00 8.51529479e-01 1.78138644e-01
1.70031691e+00 -3.57461423e-01 -4.69872802e-01 6.24120355e-01
1.17076361e+00 -2.25068599e-01 5.69603071e-02 1.80781931e-02
9.84883130e-01 7.11349070e-01 7.44084835e-01 3.01551402e-01
4.83727604e-01 3.91193658e-01 4.19315934e-01 -3.68764997e-01
-2.19058678e-01 8.84527154e-03 -3.49472582e-01 6.83942199e-01
3.81971925e-01 -8.37101936e-02 -1.00879610e+00 3.78331661e-01
-2.02879596e+00 -4.57823038e-01 -1.78675354e-02 1.72866607e+00
1.65472412e+00 -2.86529623e-02 -2.14964375e-01 2.13933185e-01
8.77154469e-01 -3.05066526e-01 -1.04825044e+00 3.96847785e-01
3.22471075e-02 2.52066366e-02 3.10178578e-01 1.07075423e-01
-1.38316011e+00 5.38186550e-01 5.99222517e+00 9.53584015e-01
-1.23922133e+00 3.86580586e-01 1.32732677e+00 -7.97478035e-02
-1.28157333e-01 -4.74297673e-01 -5.68481565e-01 4.84312445e-01
6.22898519e-01 3.61551315e-01 -7.71128982e-02 1.12608707e+00
-2.66996473e-01 7.64031634e-02 -1.48918569e+00 9.52698648e-01
1.55518472e-01 -1.37656736e+00 1.67797096e-02 -2.06137970e-01
1.06643593e+00 5.77272661e-03 1.76954255e-01 7.19990879e-02
6.10817790e-01 -1.08066571e+00 5.82101941e-01 3.85308683e-01
1.15325761e+00 -2.98470855e-01 9.28071558e-01 4.33223665e-01
-7.91425288e-01 1.40407801e-01 -1.13710634e-01 5.37729383e-01
-1.44976765e-01 9.36668396e-01 -9.73719835e-01 6.44933522e-01
8.00933182e-01 1.12412095e+00 -7.03350782e-01 5.38878024e-01
-2.95207679e-01 8.05741906e-01 -7.81967640e-02 3.46839041e-01
1.26215309e-01 3.07469785e-01 -1.70866027e-01 1.00688207e+00
-1.56458333e-01 1.46926761e-01 4.39564556e-01 8.10966551e-01
-3.69237840e-01 3.80987953e-03 -1.67571321e-01 2.08419144e-01
4.53385442e-01 1.42590642e+00 -7.90017426e-01 -7.74199426e-01
-1.95343599e-01 8.81778479e-01 5.18705249e-01 1.04837865e-01
-9.80189800e-01 3.11029226e-01 5.83757199e-02 -2.96003401e-01
-8.37690458e-02 5.35910308e-01 -6.29981935e-01 -1.19287074e+00
1.05289795e-01 -1.09531868e+00 7.86931455e-01 -6.81666315e-01
-1.86532903e+00 7.59544551e-01 -2.15888694e-01 -1.55784309e+00
-2.06823170e-01 -5.09189487e-01 -1.37796998e-01 5.71405530e-01
-1.69314504e+00 -1.40384161e+00 -4.37924951e-01 6.28303349e-01
4.73618388e-01 -1.27309546e-01 9.63379443e-01 2.44702548e-01
-7.35105932e-01 8.72764885e-01 -1.05682174e-02 2.85971045e-01
1.06307328e+00 -1.35428166e+00 -3.03197633e-02 5.39439023e-01
1.95121452e-01 4.87353742e-01 2.59357125e-01 -7.26330638e-01
-7.74809659e-01 -1.60348880e+00 5.69369853e-01 -5.09824157e-01
4.28614199e-01 5.87376468e-02 -1.23583591e+00 6.74620092e-01
-3.37138996e-02 9.13098037e-01 1.29432261e+00 -5.06807864e-02
-6.98944271e-01 1.60771539e-03 -1.29579604e+00 2.24163368e-01
1.11449838e+00 -5.74589789e-01 -3.03873837e-01 8.92924309e-01
7.34269619e-01 -5.85391462e-01 -1.05243278e+00 9.93643582e-01
2.10236892e-01 -5.69142163e-01 7.44619548e-01 -9.11824763e-01
6.93448722e-01 -3.97654474e-01 -1.11772910e-01 -1.21284163e+00
-2.37947017e-01 -6.29475489e-02 -2.47632861e-01 9.82890785e-01
6.24955058e-01 -4.45334405e-01 9.16973829e-01 1.00290811e+00
-2.12631822e-01 -1.20596755e+00 -7.53545880e-01 -7.60366380e-01
-7.49325156e-02 -3.56997848e-01 4.17951912e-01 1.60705423e+00
5.71474582e-02 1.04042999e-01 -4.07599747e-01 2.49063015e-01
6.99302197e-01 1.16661176e-01 2.38853395e-01 -1.06631386e+00
-5.87255836e-01 -1.71712086e-01 -2.01631859e-01 -4.81555104e-01
4.94848609e-01 -1.21792912e+00 2.45624244e-01 -1.37823915e+00
6.10365689e-01 -9.01342988e-01 -5.69480360e-01 1.05197310e+00
-6.51154816e-01 7.20026135e-01 -3.44896257e-01 7.45900154e-01
-1.13287830e+00 2.97297329e-01 1.29246271e+00 -4.49307889e-01
2.93562561e-01 -1.00601308e-01 -7.30492473e-01 7.86523581e-01
6.81332707e-01 -6.67398691e-01 -6.32012129e-01 -3.49890679e-01
1.39068946e-01 -1.02000751e-01 3.36558789e-01 -7.30080962e-01
2.35713348e-01 -1.53737426e-01 4.88958985e-01 -3.66977483e-01
-3.01904995e-02 -1.02727246e+00 2.01219782e-01 5.44545352e-01
-9.55230951e-01 -4.55372691e-01 -1.32374868e-01 7.37743437e-01
-3.35508615e-01 -1.85648054e-01 1.09496784e+00 -1.63750038e-01
-4.51971680e-01 5.86574972e-01 2.01444894e-01 2.78367788e-01
1.16432953e+00 1.47536814e-01 -4.41485584e-01 -8.78807083e-02
-9.84420061e-01 3.82314205e-01 3.72012049e-01 2.03954011e-01
4.90527213e-01 -1.46760106e+00 -6.93142176e-01 1.29481871e-03
5.58271885e-01 5.23318231e-01 4.54369009e-01 9.43854034e-01
-2.63670027e-01 8.18025768e-02 1.26364678e-01 -1.08623707e+00
-1.35001886e+00 8.76470149e-01 5.45231342e-01 -5.49827576e-01
-4.96680677e-01 1.06041765e+00 2.57694989e-01 -6.37390912e-01
4.29353267e-01 -3.17214549e-01 4.18833010e-02 -3.11400920e-01
5.09492397e-01 -1.52383298e-01 2.63707519e-01 -4.23886448e-01
-4.07998562e-01 2.71060437e-01 -4.09567356e-01 3.31345797e-01
9.91142035e-01 2.05894057e-02 -1.88874185e-01 3.69478643e-01
1.34280849e+00 -6.43645883e-01 -1.30220127e+00 -8.39165270e-01
-1.83832310e-02 -5.50769493e-02 -3.73383537e-02 -1.26190412e+00
-1.31405425e+00 6.86864674e-01 5.63595176e-01 -3.08135450e-02
1.25729811e+00 2.11443126e-01 6.16796494e-01 5.81958950e-01
1.70573786e-01 -1.11857402e+00 3.18481207e-01 -4.50989753e-02
5.77039063e-01 -1.93590796e+00 6.19374029e-02 -7.23950028e-01
-1.04723918e+00 7.73412347e-01 8.70814264e-01 3.02946806e-01
8.48903537e-01 3.16375405e-01 4.16356385e-01 -3.12966704e-01
-9.26994383e-01 8.89488086e-02 5.46131551e-01 4.21197295e-01
3.96928668e-01 1.73245400e-01 6.21291213e-02 7.34783173e-01
2.26403594e-01 7.33033344e-02 2.58510541e-02 9.12856460e-01
-5.78079000e-02 -1.11905229e+00 -1.88477099e-01 7.92850554e-01
-6.42281651e-01 -1.96618453e-01 -4.18392092e-01 4.20322090e-01
3.62220734e-01 8.33270967e-01 -1.64876670e-01 -3.95364910e-01
-1.05971001e-01 1.64199054e-01 2.75270522e-01 -8.79903316e-01
-4.27283615e-01 -1.48668721e-01 3.06633487e-02 -3.82186651e-01
-8.15758049e-01 -4.44976360e-01 -1.46822894e+00 2.87880212e-01
-4.88867402e-01 -2.36743595e-02 4.14252728e-01 1.26241291e+00
4.49175745e-01 6.31421983e-01 7.13660896e-01 -9.50113833e-02
-8.20031524e-01 -1.07409263e+00 -5.00925183e-01 1.03365970e+00
3.22045475e-01 -6.38253868e-01 -1.16230063e-01 5.35688579e-01] | [14.870950698852539, -2.185311794281006] |
c5c55286-3277-4b50-a259-ea887653cdfa | interpreting-gnn-based-ids-detections-using | 2306.00934 | null | https://arxiv.org/abs/2306.00934v2 | https://arxiv.org/pdf/2306.00934v2.pdf | Interpreting GNN-based IDS Detections Using Provenance Graph Structural Features | The black-box nature of complex Neural Network (NN)-based models has hindered their widespread adoption in security domains due to the lack of logical explanations and actionable follow-ups for their predictions. To enhance the transparency and accountability of Graph Neural Network (GNN) security models used in system provenance analysis, we propose PROVEXPLAINER, a framework for projecting abstract GNN decision boundaries onto interpretable feature spaces. We first replicate the decision-making process of GNNbased security models using simpler and explainable models such as Decision Trees (DTs). To maximize the accuracy and fidelity of the surrogate models, we propose novel graph structural features founded on classical graph theory and enhanced by extensive data study with security domain knowledge. Our graph structural features are closely tied to problem-space actions in the system provenance domain, which allows the detection results to be explained in descriptive, human language. PROVEXPLAINER allowed simple DT models to achieve 95% fidelity to the GNN on program classification tasks with general graph structural features, and 99% fidelity on malware detection tasks with a task-specific feature package tailored for direct interpretation. The explanations for malware classification are demonstrated with case studies of five real-world malware samples across three malware families. | ['Kangkook Jee', 'Murat Kantarcioglu', 'Feng Chen', 'Muhyun Kim', 'Tianhao Wang', 'Joshua Wiedemeier', 'Kunal Mukherjee'] | 2023-06-01 | null | null | null | null | ['explainable-models', 'malware-classification'] | ['computer-vision', 'miscellaneous'] | [ 3.99697930e-01 5.93399286e-01 -4.15228605e-01 -4.91414547e-01
2.69054055e-01 -7.00313032e-01 7.10235000e-01 2.24130258e-01
4.04263914e-01 1.70042634e-01 -2.43015066e-02 -1.56287324e+00
-2.48524591e-01 -6.99324191e-01 -7.63291895e-01 7.14575499e-02
-6.59083247e-01 -1.93836316e-01 1.17233712e-02 -1.55881509e-01
3.17905694e-01 6.72374487e-01 -1.23035455e+00 4.06310678e-01
6.60037458e-01 7.88763940e-01 -6.34818792e-01 8.88212502e-01
6.64269626e-02 1.46037149e+00 -7.63436496e-01 -5.03190279e-01
2.67903686e-01 3.08654215e-02 -7.30505109e-01 -4.16457146e-01
4.10590708e-01 -6.85103416e-01 -4.19024289e-01 1.18131006e+00
-4.20030206e-01 -4.18919712e-01 4.77305532e-01 -2.27523541e+00
-1.10255873e+00 1.16486239e+00 -3.48408818e-01 -1.31270722e-01
-4.91333157e-02 5.38843513e-01 1.13689232e+00 -1.57314450e-01
4.13064271e-01 1.40469015e+00 1.06708157e+00 9.43095028e-01
-1.50121605e+00 -7.75774837e-01 3.29881400e-01 1.38945004e-03
-9.16096509e-01 -1.52162448e-01 6.25408351e-01 -7.22739995e-01
1.64807618e+00 4.57551807e-01 5.22573471e-01 1.60643780e+00
8.25425565e-01 6.39456064e-02 7.97450900e-01 -1.69812873e-01
2.97481835e-01 -8.12381785e-03 9.71159160e-01 1.00955033e+00
1.12578714e+00 5.74576318e-01 -3.12934428e-01 -6.64687395e-01
5.70216358e-01 3.84087622e-01 -2.19380990e-01 -4.22676742e-01
-6.15065813e-01 9.93934512e-01 5.62696040e-01 4.94138226e-02
6.72936998e-03 8.98100376e-01 7.02497005e-01 4.32177752e-01
3.89582276e-01 6.10619128e-01 -7.22959518e-01 4.80799451e-02
-6.29827201e-01 -1.17004208e-01 9.64149714e-01 1.04685402e+00
6.73453927e-01 6.91453755e-01 1.94240920e-02 -3.75052869e-01
7.89414287e-01 2.93898404e-01 -1.25365436e-01 -6.73991561e-01
2.32322931e-01 1.06394851e+00 -6.78119063e-01 -1.22685790e+00
-2.41114497e-01 -5.47129273e-01 -6.69634700e-01 7.43992925e-01
1.25584543e-01 6.09861836e-02 -9.45244968e-01 1.73807275e+00
-1.63028389e-01 -2.61279251e-02 5.27253039e-02 3.83584410e-01
4.67473447e-01 3.95782381e-01 2.11709291e-01 2.00636625e-01
1.18879402e+00 -5.05230904e-01 -2.20133573e-01 -2.76364714e-01
1.15134311e+00 2.30253726e-01 1.12148368e+00 4.07801658e-01
-1.91431522e-01 -2.09924594e-01 -1.29130292e+00 2.65132755e-01
-3.13106507e-01 -3.15796465e-01 9.02562320e-01 1.41180098e+00
-1.35076535e+00 8.46767187e-01 -1.03172481e+00 -2.13030785e-01
5.94543576e-01 6.30164623e-01 -3.33569795e-01 2.37660483e-01
-9.03109670e-01 6.67150140e-01 4.99866694e-01 1.04299523e-01
-1.45409167e+00 -7.47761130e-01 -9.00154650e-01 1.37044802e-01
3.95736873e-01 -7.17553914e-01 1.02203822e+00 -6.37919068e-01
-1.00693393e+00 2.39871740e-01 4.01856273e-01 -7.01652646e-01
5.36603928e-02 2.58682728e-01 -5.31233251e-01 -1.79080486e-01
-2.30838269e-01 1.79918095e-01 1.17347634e+00 -1.27008533e+00
-4.94832247e-02 -3.00785631e-01 6.13177001e-01 -7.17438877e-01
-6.32937193e-01 8.69036019e-02 3.89894575e-01 -2.71021634e-01
-1.97158590e-01 -8.01699758e-01 -1.84194982e-01 5.53001687e-02
-9.21093702e-01 1.08155444e-01 1.18574715e+00 -1.03322172e+00
1.33690286e+00 -2.06278610e+00 -3.26090127e-01 5.99894583e-01
1.31306076e+00 2.04545483e-01 -8.63622501e-02 3.18378270e-01
-4.27292198e-01 1.09755981e+00 -8.18316936e-02 -1.59468919e-01
3.94151300e-01 2.76274174e-01 -6.97102070e-01 3.83925885e-01
5.26597440e-01 1.03777051e+00 -7.92781532e-01 -1.48950219e-01
-1.03365080e-02 2.70869553e-01 -6.89493299e-01 -4.44938280e-02
-5.17934203e-01 1.46283641e-01 -4.55257624e-01 7.00626135e-01
5.11959553e-01 -3.97172838e-01 3.09268951e-01 -4.81268764e-02
3.60308200e-01 1.90202564e-01 -4.99708861e-01 8.03797841e-01
-3.45845103e-01 9.48673725e-01 1.73681796e-01 -4.85641539e-01
8.76466334e-01 1.08309850e-01 -2.85887271e-01 -1.76951081e-01
1.22527570e-01 -1.66229084e-01 3.57724369e-01 -3.00991416e-01
4.28593576e-01 1.40645117e-01 -1.52344629e-01 9.45796013e-01
-2.23042876e-01 -1.16333976e-01 -3.79157841e-01 4.67532158e-01
1.77779222e+00 -1.63057849e-01 4.61184353e-01 -3.35642010e-01
2.10922644e-01 2.20925093e-01 4.39235806e-01 1.00611401e+00
-2.52235681e-01 -1.06674932e-01 1.15682840e+00 -6.00124002e-01
-1.08828998e+00 -8.01430702e-01 2.71144480e-01 8.37736666e-01
-3.59835804e-01 -1.00922728e+00 -1.09202611e+00 -1.28771043e+00
1.08657954e-02 1.33468843e+00 -9.55682635e-01 -6.66523933e-01
-4.69503164e-01 -2.40768373e-01 1.22920847e+00 6.57763898e-01
8.95538256e-02 -8.42620254e-01 -4.66231018e-01 -1.96747005e-01
5.17072797e-01 -8.26260328e-01 -1.77469358e-01 3.88807923e-01
-1.06963217e+00 -1.46597958e+00 6.89548671e-01 -1.35679454e-01
9.02280271e-01 6.42069243e-03 1.08620334e+00 8.60720277e-01
-2.02177390e-01 5.95465302e-01 -2.81188369e-01 -4.60145772e-01
-1.39149702e+00 -1.18105814e-01 2.48534620e-01 -6.47803664e-01
2.07441986e-01 -6.41638935e-01 4.36935946e-02 1.74088538e-01
-9.37822163e-01 4.34910841e-02 2.27363765e-01 6.71218932e-01
-8.80275890e-02 2.74190873e-01 -2.77238861e-02 -1.07167876e+00
1.01808906e+00 -4.86776531e-01 -8.80831420e-01 4.46997792e-01
-1.43923521e+00 3.95340323e-01 1.16493142e+00 -5.09277701e-01
-3.68118167e-01 -4.95110720e-01 4.95865643e-01 -7.16592193e-01
8.50419104e-02 5.99594057e-01 -2.04603434e-01 -4.71636057e-01
1.38737929e+00 -1.06548250e-01 3.02852511e-01 8.44229683e-02
3.63374352e-01 3.24013948e-01 2.46567592e-01 -7.65168428e-01
1.23056424e+00 -1.07480958e-01 2.48500913e-01 -4.10808653e-01
1.12037528e-02 6.04664683e-01 -2.91750908e-01 -3.31755280e-02
4.90318656e-01 -3.40464234e-01 -1.08873379e+00 1.99922532e-01
-1.35761130e+00 -3.42449576e-01 -2.62101948e-01 -3.52086425e-02
-1.61556244e-01 4.48739856e-01 -6.89514160e-01 -1.04473913e+00
-6.11562729e-01 -1.34738350e+00 7.10691929e-01 -1.56394497e-01
-5.53787470e-01 -1.28850901e+00 -2.45213121e-01 -5.08403629e-02
5.87098300e-01 4.53010499e-01 1.84011424e+00 -1.24314320e+00
-6.63715005e-01 -3.36391538e-01 -4.57711846e-01 5.88919222e-01
1.35759100e-01 5.91049731e-01 -1.05796015e+00 -3.93030584e-01
2.01800719e-01 9.32835639e-02 1.68598384e-01 4.08838876e-02
1.15808618e+00 -1.00090778e+00 -4.20378506e-01 5.56820869e-01
1.14901340e+00 4.18355577e-02 1.67560846e-01 3.94506007e-02
1.14809477e+00 5.47358394e-01 -7.05305934e-02 1.09632872e-01
-2.25042198e-02 2.78734505e-01 1.13441670e+00 1.95434108e-01
8.58931895e-03 -6.48602009e-01 6.42180920e-01 4.03616160e-01
7.14073628e-02 -3.00153494e-01 -1.57074678e+00 -1.64611191e-02
-1.72329342e+00 -7.56365478e-01 -5.27113020e-01 1.80691302e+00
1.67779118e-01 4.56878692e-01 -1.19010024e-01 1.23851955e-01
8.09897721e-01 4.35026549e-02 -7.89417624e-01 -7.87278891e-01
2.78415829e-01 2.77690534e-02 6.45902336e-01 4.88822907e-01
-5.00940084e-01 5.73312640e-01 6.84457493e+00 6.79037690e-01
-9.15516615e-01 1.01697877e-01 4.46897984e-01 3.58580738e-01
-7.64894128e-01 6.78070903e-01 -5.06986678e-01 -4.85335663e-02
1.67841697e+00 -4.92182493e-01 8.95027041e-01 1.18626559e+00
2.45220646e-01 6.36905491e-01 -1.62346244e+00 5.85705698e-01
-3.42893809e-01 -1.52076316e+00 2.51317859e-01 4.30914134e-01
1.13486327e-01 -1.38530731e-01 2.23840058e-01 4.83998060e-01
8.64568949e-01 -1.46038651e+00 1.06911957e+00 1.84461817e-01
9.00900185e-01 -4.92413908e-01 5.16883612e-01 4.23943013e-01
-1.10314000e+00 -5.28950572e-01 -1.04609743e-01 -3.98634285e-01
-5.67098498e-01 2.60576129e-01 -1.66667819e+00 5.28069556e-01
3.61641318e-01 6.16297662e-01 -9.44214463e-01 1.86837554e-01
-6.09291077e-01 1.02638185e+00 2.81847529e-02 -2.73126632e-01
7.47031868e-02 3.01158130e-01 5.82589388e-01 1.06166422e+00
8.98844525e-02 -3.39636594e-01 -2.44764522e-01 1.62344503e+00
1.20877750e-01 -6.41263366e-01 -9.77025867e-01 -6.65455461e-01
4.69944388e-01 1.21079266e+00 -6.62582278e-01 -2.75347456e-02
-3.46044153e-02 2.92648733e-01 2.71468252e-01 4.35841382e-01
-1.00608146e+00 -2.15771422e-01 7.84848332e-01 3.22816074e-01
-2.38645703e-01 -3.55522096e-01 -6.31149530e-01 -8.56789172e-01
-3.66597064e-02 -1.26321268e+00 9.52705666e-02 -6.36004865e-01
-1.35673165e+00 9.24297690e-01 4.17365432e-01 -1.03673339e+00
-6.50904894e-01 -1.08084166e+00 -9.18463886e-01 9.91467059e-01
-5.69821060e-01 -1.40983987e+00 -1.64979056e-01 3.55391800e-01
-1.61596984e-01 -3.85700136e-01 9.51702178e-01 -4.33891445e-01
-7.94874907e-01 7.44032443e-01 -3.19108725e-01 1.95245132e-01
-1.32986128e-01 -1.12919021e+00 1.36601186e+00 1.28247607e+00
-4.98030968e-02 1.38188601e+00 9.51698601e-01 -1.15166390e+00
-1.68661821e+00 -1.26076102e+00 3.01995575e-01 -8.66095066e-01
1.24014306e+00 -7.42611825e-01 -9.62364078e-01 9.87958252e-01
-1.86855048e-01 3.33394073e-02 4.76670057e-01 1.95288777e-01
-1.14331138e+00 2.07710758e-01 -1.32671607e+00 6.26764059e-01
1.37361312e+00 -1.12472618e+00 -1.72074571e-01 2.76113212e-01
1.34564233e+00 -2.11242046e-02 -7.23923802e-01 2.95146912e-01
2.35814393e-01 -8.79561484e-01 5.82732320e-01 -1.19282079e+00
3.22199792e-01 -4.61909354e-01 -3.52912813e-01 -8.76353025e-01
-3.22798640e-01 -8.07357192e-01 -6.60104394e-01 9.44482684e-01
6.42145634e-01 -9.48348463e-01 9.02149975e-01 1.20826709e+00
-2.16411635e-01 -6.48473978e-01 -6.56458437e-01 -9.56108510e-01
-7.46670291e-02 -1.02290702e+00 1.02098107e+00 8.94905269e-01
1.97136924e-01 3.87051292e-02 -8.33786558e-03 6.12243593e-01
9.67126727e-01 -2.49646053e-01 6.62327826e-01 -1.46656847e+00
-6.64180160e-01 -4.84343052e-01 -8.59983087e-01 -2.13474363e-01
5.10290623e-01 -1.16471720e+00 -4.50294733e-01 -1.22781026e+00
2.12778866e-01 -2.09630266e-01 6.86345324e-02 1.12160075e+00
2.94960976e-01 -4.44157451e-01 2.75855839e-01 4.09861594e-01
-9.40428600e-02 2.51326323e-01 4.84493166e-01 -4.74499106e-01
4.81293304e-03 -1.41147316e-01 -1.03089225e+00 6.47906125e-01
6.21227562e-01 -8.77018750e-01 -9.43154395e-01 -3.60597461e-01
1.74698025e-01 1.12441063e-01 9.30866957e-01 -8.77064705e-01
1.36089325e-01 -2.92923391e-01 4.12762240e-02 -9.59424227e-02
-9.14295688e-02 -8.95305216e-01 7.81063676e-01 9.38116670e-01
-2.96974480e-01 3.76075506e-01 4.91625667e-01 7.48901606e-01
2.76134074e-01 -2.82940477e-01 7.89320618e-02 3.42118233e-01
-6.86098397e-01 4.53988492e-01 -4.31556165e-01 -3.51718247e-01
9.68208134e-01 -6.19383156e-01 -1.16851687e+00 -5.10364592e-01
-5.20455420e-01 -9.97172818e-02 9.66321230e-01 5.20514846e-01
7.93709815e-01 -8.63816440e-01 -2.79327691e-01 4.06426281e-01
2.52813548e-01 -4.09910172e-01 -8.55134204e-02 5.14621973e-01
-5.82720160e-01 4.48730111e-01 -4.13352668e-01 -5.76364756e-01
-1.31216013e+00 8.69555414e-01 4.62048590e-01 -4.07008797e-01
-4.55245882e-01 4.54521924e-01 3.39384019e-01 -6.82664454e-01
1.06246062e-01 -8.14114273e-01 1.70471102e-01 -8.66474092e-01
4.16352451e-01 2.82961786e-01 5.42528220e-02 -6.84420764e-02
-6.51099861e-01 -2.23305181e-01 -2.55511850e-01 1.54615939e-01
1.30798912e+00 5.03466129e-01 -4.43357766e-01 9.57758799e-02
9.97486830e-01 -2.66587406e-01 -1.14065313e+00 2.53461152e-01
1.00523867e-01 -1.21843390e-01 7.67909214e-02 -8.25326324e-01
-7.23543942e-01 9.75400805e-01 2.82504380e-01 7.60256112e-01
9.31106687e-01 -3.52805734e-01 6.72281254e-04 6.15688860e-01
7.72807121e-01 -1.59030855e-01 -2.05203876e-01 4.84950811e-01
6.57892346e-01 -5.92070043e-01 -7.19755562e-03 -4.53005373e-01
-1.22317262e-01 1.49597597e+00 8.59677732e-01 1.95685297e-01
5.51903427e-01 4.57001328e-01 -4.38260913e-01 -5.51197767e-01
-8.54946911e-01 8.27198505e-01 2.63750404e-02 9.93545830e-01
9.24365968e-02 2.25119695e-01 5.64005136e-01 1.00862801e+00
-1.61213905e-01 -1.69648588e-01 8.76105130e-01 9.77208436e-01
-9.30832624e-02 -9.40826654e-01 -3.94418776e-01 6.74119890e-01
8.68116226e-03 -4.40549225e-01 -8.15042496e-01 1.08195281e+00
-2.62168527e-01 1.15807903e+00 -4.19538468e-01 -1.39426613e+00
9.17616412e-02 -1.23164423e-01 1.25669762e-01 -7.48808920e-01
-9.64966416e-01 -9.87809241e-01 4.94392365e-01 -8.29761028e-01
3.98970276e-01 -2.52562344e-01 -1.36009502e+00 -1.14929521e+00
-3.76262456e-01 -2.40115169e-02 7.97193944e-01 7.29038715e-01
6.09897733e-01 7.00349510e-01 3.74991357e-01 -5.55438042e-01
-1.07569957e+00 -5.10571182e-01 -5.05705178e-01 1.12089574e-01
3.80278111e-01 -2.92079061e-01 -6.50134861e-01 -1.62057653e-01] | [6.984769821166992, 7.524519920349121] |
c7771cea-6355-445e-8ad9-414590483ff9 | a-novel-adaptive-learning-rate-scheduler-for | 1902.07399 | null | https://arxiv.org/abs/1902.07399v4 | https://arxiv.org/pdf/1902.07399v4.pdf | LipschitzLR: Using theoretically computed adaptive learning rates for fast convergence | Optimizing deep neural networks is largely thought to be an empirical process, requiring manual tuning of several hyper-parameters, such as learning rate, weight decay, and dropout rate. Arguably, the learning rate is the most important of these to tune, and this has gained more attention in recent works. In this paper, we propose a novel method to compute the learning rate for training deep neural networks with stochastic gradient descent. We first derive a theoretical framework to compute learning rates dynamically based on the Lipschitz constant of the loss function. We then extend this framework to other commonly used optimization algorithms, such as gradient descent with momentum and Adam. We run an extensive set of experiments that demonstrate the efficacy of our approach on popular architectures and datasets, and show that commonly used learning rates are an order of magnitude smaller than the ideal value. | ['Snehanshu Saha', 'Tejas Prashanth', 'Rahul Yedida'] | 2019-02-20 | null | null | null | null | ['handwritten-digit-recognition'] | ['computer-vision'] | [-3.17053080e-01 -1.37742996e-01 -2.43468627e-01 -5.38206220e-01
-3.50191474e-01 -5.32821774e-01 3.36236149e-01 9.80611742e-02
-1.06463766e+00 7.94125080e-01 -1.77948445e-01 -3.65042776e-01
-1.49876894e-02 -5.72872698e-01 -8.04813385e-01 -7.06504464e-01
-9.12775472e-02 1.00074142e-01 5.01110852e-01 -1.89029723e-01
2.59031236e-01 4.82993186e-01 -1.20824611e+00 -3.05454165e-01
6.17636919e-01 1.08133376e+00 -1.61403447e-01 5.50361753e-01
5.72967306e-02 6.47333264e-01 -5.86311579e-01 -5.77542305e-01
3.13251436e-01 -4.20992464e-01 -7.31568575e-01 -1.95977986e-01
1.76206484e-01 -3.77974868e-01 -1.85091227e-01 1.04431450e+00
4.97612566e-01 3.12850863e-01 4.36547399e-01 -9.13841784e-01
-3.31247091e-01 7.30585635e-01 -2.26919264e-01 4.30095196e-01
-3.75608891e-01 1.08767964e-01 1.06696033e+00 -6.25863612e-01
3.27137858e-01 1.05918157e+00 8.49159777e-01 7.25113511e-01
-1.10782886e+00 -6.35507941e-01 4.55743104e-01 6.31451979e-02
-1.25890517e+00 -3.37492377e-01 5.68606675e-01 -3.69516671e-01
7.99504161e-01 -1.77183881e-01 5.04242599e-01 7.87212670e-01
-4.54394743e-02 6.03449404e-01 5.65140069e-01 -6.18754268e-01
5.92033505e-01 4.11391556e-01 1.10103860e-01 8.00761878e-01
3.21771264e-01 -1.42576844e-01 -3.31586033e-01 -5.90216890e-02
1.07285726e+00 -2.14506984e-01 -2.58836240e-01 -2.72296250e-01
-7.49371529e-01 1.09938502e+00 4.15947288e-01 4.42018062e-02
-2.81625807e-01 6.30334079e-01 5.38595915e-01 5.08942604e-01
5.56160510e-01 5.09183466e-01 -5.80417752e-01 -3.49661559e-01
-7.00944602e-01 1.84689298e-01 9.25065279e-01 4.37591225e-01
6.54180944e-01 3.42225954e-02 5.12321927e-02 1.08493388e+00
3.21489483e-01 -1.84187684e-02 6.48539126e-01 -9.49059248e-01
4.94265586e-01 3.68867010e-01 3.17324579e-01 -7.25510359e-01
-3.65662098e-01 -4.83838946e-01 -6.76008105e-01 4.08230007e-01
7.67403603e-01 -6.95086718e-01 -5.84361911e-01 1.88715982e+00
3.97396088e-01 5.52765504e-02 -1.77054852e-01 1.02981138e+00
4.02563334e-01 3.00427705e-01 1.59851983e-02 6.13920565e-04
7.21983075e-01 -1.08707821e+00 -3.69025528e-01 -1.01929560e-01
5.05626976e-01 -4.82491344e-01 1.22274768e+00 4.05052841e-01
-1.24706244e+00 -1.87676981e-01 -1.16180217e+00 7.00843558e-02
-3.23131606e-02 2.84179688e-01 6.60426378e-01 8.00636947e-01
-9.84663248e-01 1.08562493e+00 -1.08061028e+00 -7.13058189e-02
3.78280103e-01 4.68324423e-01 5.03230169e-02 4.07462060e-01
-1.01122856e+00 8.37901592e-01 6.09946191e-01 1.88454673e-01
-8.02454591e-01 -6.57218814e-01 -3.38693798e-01 1.75158158e-01
1.08038761e-01 -5.83246887e-01 1.40197611e+00 -8.14502597e-01
-2.09792972e+00 6.90607727e-01 1.44830748e-01 -7.79447734e-01
8.98530900e-01 -5.21096885e-01 1.51200458e-01 -8.94329026e-02
-6.95813179e-01 3.57576728e-01 9.98540163e-01 -7.92304516e-01
-4.62570459e-01 -1.63119808e-01 3.68577331e-01 5.33062890e-02
-9.76639152e-01 1.64864659e-01 -4.17932928e-01 -3.54572773e-01
-8.87094140e-02 -8.97476852e-01 -3.15298647e-01 3.36074233e-01
-1.85217783e-01 -2.80947834e-01 2.30393544e-01 -2.70570904e-01
1.20184422e+00 -2.15600181e+00 9.06789154e-02 1.85891226e-01
1.71503901e-01 2.91889668e-01 1.00309335e-01 -1.43111087e-02
1.53803125e-01 1.11899585e-01 -1.81082726e-01 -4.21364635e-01
6.66905865e-02 4.99203391e-02 -2.75870025e-01 6.04841471e-01
2.10538819e-01 6.79029167e-01 -7.62155592e-01 -7.98667669e-02
-1.53210804e-01 7.55634010e-01 -7.26840317e-01 1.36524260e-01
-1.12113059e-01 1.30055636e-01 -3.79751235e-01 1.37960881e-01
3.10565472e-01 -3.62242103e-01 5.16344383e-02 -4.80604246e-02
-1.45422161e-01 4.68600124e-01 -1.26624894e+00 1.23579633e+00
-5.17950237e-01 7.25229383e-01 1.11679301e-01 -1.00950539e+00
1.07419550e+00 1.64093524e-01 3.18276107e-01 -2.46990949e-01
4.17179108e-01 2.16982067e-01 -1.39164343e-01 -2.44935811e-01
3.20002794e-01 -1.04912482e-01 4.83767807e-01 6.25317991e-01
-2.53310371e-02 2.80957986e-02 3.34984869e-01 -2.97477037e-01
9.96374011e-01 -1.39225513e-01 -5.22240065e-02 -3.02861512e-01
5.17685354e-01 -2.74739206e-01 3.55364680e-01 6.09665632e-01
-1.83573857e-01 5.43864906e-01 7.31798232e-01 -5.81160367e-01
-1.20477366e+00 -7.93254673e-01 -2.50857472e-01 1.38912368e+00
-1.43561289e-01 -2.22351313e-01 -9.79172051e-01 -6.62676752e-01
3.50623601e-03 2.77733654e-01 -6.52409136e-01 -1.77092433e-01
-7.02260435e-01 -1.21064019e+00 6.26004755e-01 5.56771457e-01
4.80037421e-01 -9.19982076e-01 -7.92859733e-01 2.14951098e-01
4.09768373e-01 -1.02784288e+00 -2.91961432e-01 4.27138716e-01
-1.26096761e+00 -8.88538659e-01 -9.15262341e-01 -5.94364703e-01
7.65904665e-01 -2.57250637e-01 1.05470502e+00 3.51935238e-01
-8.00517350e-02 9.96166542e-02 -1.16164535e-01 -4.63929534e-01
-3.48918200e-01 5.83943486e-01 8.79946575e-02 -1.71931051e-02
-6.71144575e-02 -7.45386422e-01 -7.81402230e-01 2.45504707e-01
-9.32280779e-01 -2.45052800e-01 4.78493661e-01 7.63642371e-01
5.09486198e-01 3.49520892e-02 4.94975537e-01 -1.00640392e+00
1.11087692e+00 -4.24870938e-01 -1.08047462e+00 8.83571431e-02
-8.30742240e-01 4.15495813e-01 8.55905473e-01 -8.66981506e-01
-6.98430479e-01 4.28555869e-02 -3.20754111e-01 -4.02601838e-01
1.94471180e-01 3.88806790e-01 3.45459312e-01 -5.40615380e-01
8.02566230e-01 -7.55443946e-02 -1.17909618e-01 -5.23067176e-01
3.52098733e-01 9.87518728e-02 2.95310050e-01 -6.62424922e-01
6.52301669e-01 4.75660980e-01 -2.58655399e-01 -4.52337682e-01
-9.73836303e-01 -5.15756942e-02 -4.44079250e-01 -6.07548580e-02
1.85626939e-01 -5.98102570e-01 -8.52172971e-01 3.80924493e-01
-9.72687066e-01 -8.36252928e-01 -3.38964522e-01 7.21463859e-01
-4.05116558e-01 -8.40602741e-02 -8.14677477e-01 -6.40786052e-01
-6.15210652e-01 -1.02756524e+00 4.94014084e-01 3.15604985e-01
9.54628661e-02 -1.37099099e+00 1.75435245e-01 -2.17565998e-01
7.56757736e-01 9.49668512e-02 7.98269153e-01 -5.64147413e-01
-1.52775139e-01 -1.50983766e-01 -2.27679074e-01 6.43148065e-01
-4.83999774e-02 1.68090954e-01 -8.29737246e-01 -4.32927936e-01
1.93214700e-01 -4.03663486e-01 1.00762796e+00 4.44232762e-01
1.29358542e+00 -2.60721594e-01 4.44372483e-02 1.05012834e+00
1.58771837e+00 -2.43288904e-01 5.21483839e-01 6.74160659e-01
4.96830046e-01 2.28262484e-01 6.28766865e-02 7.54951775e-01
1.34409532e-01 4.64736283e-01 5.07397890e-01 1.20610677e-01
1.49558157e-01 2.14776080e-02 3.01333785e-01 8.23604226e-01
-3.04554254e-01 2.14795455e-01 -7.80873537e-01 3.30488771e-01
-1.77575243e+00 -6.34387255e-01 3.68832231e-01 2.33729601e+00
1.25626278e+00 5.27807951e-01 3.53163362e-01 2.42868721e-01
5.77304780e-01 -5.59725650e-02 -8.57001066e-01 -4.07121509e-01
1.07616320e-01 2.32773781e-01 7.07855463e-01 5.28567731e-01
-9.57309186e-01 9.60448682e-01 7.35653019e+00 4.71971095e-01
-1.54968297e+00 3.65055464e-02 8.20648372e-01 -3.97489309e-01
-1.37191325e-01 -1.83043718e-01 -1.14969468e+00 4.72658962e-01
9.65454102e-01 -9.50678587e-02 6.58368587e-01 1.16719484e+00
9.58726108e-02 1.70638964e-01 -1.06441438e+00 9.76813436e-01
-4.22163248e-01 -1.22311676e+00 -3.22491795e-01 -1.22035362e-01
8.85922432e-01 3.42147976e-01 1.84254870e-01 2.17679650e-01
4.43356633e-01 -9.94226396e-01 7.37940550e-01 1.44729748e-01
4.23551053e-01 -8.69217038e-01 5.94784796e-01 2.42077053e-01
-8.03776264e-01 -1.48659796e-01 -7.07753181e-01 -1.81716401e-02
-2.48915274e-02 9.86500442e-01 -5.84731042e-01 -3.37941766e-01
6.92551613e-01 3.58741730e-01 -3.89940739e-01 1.47857988e+00
-4.48281735e-01 9.04661417e-01 -6.59037888e-01 -4.61572140e-01
3.27993125e-01 -2.22035438e-01 1.36179581e-01 1.32512081e+00
1.09669432e-01 -1.71568900e-01 -2.07414925e-01 7.67802000e-01
-5.76582551e-01 2.45037422e-01 -4.44780141e-02 1.40665144e-01
4.44579005e-01 1.28126180e+00 -6.75380409e-01 2.91328114e-02
-2.07415283e-01 6.58740282e-01 7.75974393e-01 5.17811537e-01
-1.04949677e+00 -4.40331399e-01 7.21793115e-01 1.30968532e-02
5.73452950e-01 -4.85647142e-01 -2.47413561e-01 -9.57472146e-01
4.17402953e-01 -5.12351334e-01 1.04210816e-01 -7.20216660e-03
-1.03746915e+00 6.91484690e-01 -1.33923247e-01 -7.86297321e-01
-3.14817250e-01 -7.22293317e-01 -6.83438182e-01 6.71486795e-01
-1.73055267e+00 -2.01970816e-01 -2.28449985e-01 4.46609974e-01
3.38373482e-01 -6.70051947e-02 5.32697141e-01 3.92030209e-01
-8.66348743e-01 9.46258426e-01 3.68541181e-01 1.67070791e-01
3.91291142e-01 -1.05581367e+00 4.30825770e-01 5.33433676e-01
4.50337352e-03 6.77863061e-01 9.19888198e-01 3.00727356e-02
-1.19568419e+00 -9.12745237e-01 3.95748049e-01 9.67452396e-03
8.84387791e-01 -2.71538854e-01 -1.00429130e+00 5.71533084e-01
-1.85672879e-01 2.27495566e-01 4.94758517e-01 1.33029476e-01
-3.50704074e-01 -2.97468454e-01 -9.11370993e-01 5.32508135e-01
8.13264251e-01 -2.95078218e-01 -8.43860954e-02 3.88818860e-01
5.21971881e-01 -5.69493949e-01 -8.83321464e-01 1.90632477e-01
7.33586371e-01 -9.72047508e-01 8.40739191e-01 -6.24451399e-01
2.36808360e-01 1.68717206e-01 1.42546207e-01 -1.44817543e+00
1.84682105e-02 -7.80988097e-01 -3.57486725e-01 1.05913663e+00
6.17712438e-01 -7.96272099e-01 1.07038713e+00 8.92642021e-01
2.70682335e-01 -1.24988532e+00 -8.40900540e-01 -8.86412859e-01
3.25998902e-01 -3.45456511e-01 3.96160036e-01 6.89655304e-01
-3.41108143e-01 7.54020512e-02 -1.34987220e-01 -4.88065965e-02
4.90988374e-01 -2.66250610e-01 5.49798012e-01 -1.24649990e+00
-4.27904814e-01 -7.93266833e-01 -1.90643370e-01 -1.17857361e+00
2.01104984e-01 -4.87110317e-01 4.46195155e-02 -1.05670571e+00
-9.56204832e-02 -6.83172524e-01 -5.47359765e-01 4.19134289e-01
-1.55992508e-01 4.25569620e-03 -3.43385339e-02 2.68194854e-01
-5.83397865e-01 6.71380877e-01 9.43981647e-01 1.67635560e-01
-5.35400450e-01 1.98123187e-01 -5.18854082e-01 8.90273392e-01
1.11771929e+00 -6.41723216e-01 -4.25671548e-01 -8.26924086e-01
4.60299522e-01 -5.89546561e-01 1.16888352e-01 -1.01375556e+00
3.01463753e-01 -2.35417664e-01 2.00192153e-01 1.70315400e-01
2.38624111e-01 -4.79739308e-01 -3.59174877e-01 3.98846835e-01
-7.05306053e-01 1.21437348e-01 2.34370649e-01 2.39671409e-01
-2.08876990e-02 -6.51263237e-01 9.66287851e-01 3.46892402e-02
-3.86621743e-01 2.52377659e-01 9.44665149e-02 2.53162235e-01
6.02023542e-01 1.09247327e-01 -1.66493515e-03 -3.11769992e-01
-4.82650280e-01 1.33684412e-01 1.81962624e-01 1.68443173e-01
5.81429064e-01 -1.05732358e+00 -6.50694430e-01 5.30467229e-03
-3.45045924e-01 -7.23923519e-02 -5.07498980e-01 7.32616246e-01
-7.78068006e-01 1.68725193e-01 -5.31595387e-03 -3.76354873e-01
-9.73189652e-01 1.06246829e-01 7.20546484e-01 -2.86906004e-01
-6.60539746e-01 1.08875000e+00 -2.46856228e-01 -2.11759105e-01
7.48403013e-01 -4.49157059e-01 -7.96421394e-02 -2.67693877e-01
7.86747634e-01 3.66755784e-01 1.56736240e-01 -7.24272355e-02
-2.28928819e-01 5.05293131e-01 -9.32513922e-02 -9.82058421e-02
1.52151883e+00 -1.96942650e-02 3.72860618e-02 4.25557733e-01
1.33153522e+00 -4.68074620e-01 -1.64429283e+00 -2.44926199e-01
8.07112008e-02 -1.67826965e-01 3.00078839e-01 -3.05030465e-01
-1.34194529e+00 7.97294438e-01 7.36992300e-01 2.80491561e-01
9.50740159e-01 -1.86581239e-01 8.13107669e-01 7.41574049e-01
2.26831302e-01 -1.34169316e+00 9.47766453e-02 7.05812871e-01
5.17073214e-01 -1.27163196e+00 1.88652705e-02 -7.66346510e-03
-2.03414097e-01 1.29857111e+00 5.34355640e-01 -5.28908730e-01
7.32428491e-01 3.13746899e-01 7.39552453e-02 1.31337553e-01
-9.36782479e-01 4.30835746e-02 1.09744407e-01 5.59165888e-02
6.40934944e-01 -2.59572059e-01 -4.59741414e-01 2.97267437e-01
-2.87678242e-01 2.06922203e-01 1.86139882e-01 7.07945108e-01
-6.19752347e-01 -9.88757074e-01 -1.06480084e-01 2.34559774e-01
-8.30413997e-01 -7.42832422e-02 -2.10000332e-02 4.80784714e-01
-3.71898681e-01 5.79289973e-01 -2.99406052e-03 -9.99066755e-02
3.32769752e-01 -7.80271441e-02 6.28144205e-01 -3.53143781e-01
-8.04945707e-01 -5.00851750e-01 -1.75713971e-01 -3.53576899e-01
-1.57781839e-01 -2.07137734e-01 -1.32866395e+00 -4.21796471e-01
-5.23142576e-01 4.23570275e-02 1.19967294e+00 1.02681434e+00
-4.61352197e-03 3.17418098e-01 6.65487230e-01 -7.16064274e-01
-1.10993946e+00 -6.69771373e-01 -4.10358280e-01 4.48473305e-01
3.25269639e-01 -6.64000273e-01 -6.37888074e-01 -1.14523262e-01] | [7.832232475280762, 3.7470712661743164] |
9e3a31a6-1b1c-49da-a3e6-8b0d6e2d6939 | automatic-grammatical-error-detection-for | null | null | https://aclanthology.org/W16-4908 | https://aclanthology.org/W16-4908.pdf | Automatic Grammatical Error Detection for Chinese based on Conditional Random Field | In the process of learning and using Chinese, foreigners may have grammatical errors due to negative migration of their native languages. Currently, the computer-oriented automatic detection method of grammatical errors is not mature enough. Based on the evaluating task {---} CGED2016, we select and analyze the classification model and design feature extraction method to obtain grammatical errors including Mission(M), Disorder(W), Selection (S) and Redundant (R) automatically. The experiment results based on the dynamic corpus of HSK show that the Chinese grammatical error automatic detection method, which uses CRF as classification model and n-gram as feature extraction method. It is simple and efficient which play a positive effect on the research of Chinese grammatical error automatic detection and also a supporting and guiding role in the teaching of Chinese as a foreign language. | ['Hong-ying Zan', 'Liyan Zhuo', 'Yingjie Han', 'Yajun Liu'] | 2016-12-01 | null | null | null | ws-2016-12 | ['grammatical-error-detection'] | ['natural-language-processing'] | [-5.84438801e-01 -3.12904716e-01 5.30211687e-01 -6.48056149e-01
-3.36417645e-01 -5.02731279e-02 -7.95130059e-02 6.59272015e-01
-7.82248974e-01 7.96710432e-01 2.49837697e-01 -5.72332501e-01
1.60940439e-01 -8.09164345e-01 -3.58406007e-01 -2.48609319e-01
1.00038514e-01 1.91019356e-01 1.06315292e-01 -4.72495228e-01
9.35153246e-01 9.34154391e-02 -1.38527083e+00 1.46128424e-02
1.87419057e+00 3.79797846e-01 8.02962363e-01 3.68287325e-01
-4.55812901e-01 7.12522805e-01 -1.04917204e+00 -4.18666303e-01
-4.05468404e-01 -4.90387768e-01 -1.15339983e+00 -2.86590755e-01
-3.03280115e-01 -2.44055390e-01 1.24228276e-01 1.41149676e+00
7.16592848e-01 2.19220847e-01 4.24908221e-01 -5.93811691e-01
-9.71514583e-01 8.22867811e-01 9.96970683e-02 2.82227516e-01
4.37015086e-01 -3.24519008e-01 3.60831589e-01 -9.44632292e-01
5.79890788e-01 1.14677787e+00 6.74924314e-01 7.93721676e-01
-4.79931980e-01 -9.55602944e-01 8.03944189e-03 2.61076689e-01
-1.36191916e+00 -2.17757672e-01 3.12560767e-01 -4.64804590e-01
1.29421794e+00 3.09970099e-02 8.13828528e-01 3.99977267e-01
5.68162382e-01 7.15297341e-01 1.14467633e+00 -1.08009017e+00
-1.01674303e-01 3.31024706e-01 5.36381483e-01 7.50974119e-01
2.48743266e-01 2.71748211e-02 -2.68173009e-01 3.68570566e-01
2.26829141e-01 -1.71227589e-01 -2.47868657e-01 1.21083724e+00
-7.22247899e-01 7.15526879e-01 -6.61878437e-02 8.68615925e-01
6.40128478e-02 -1.46154359e-01 1.48063153e-01 5.70703506e-01
3.94476116e-01 2.85526186e-01 -8.70321393e-01 -6.67134941e-01
-7.07058430e-01 1.63484097e-01 4.63810861e-01 1.47491634e+00
6.56280398e-01 1.75302818e-01 1.21253189e-02 8.24836373e-01
5.21393239e-01 5.80189109e-01 1.16857457e+00 -1.51024193e-01
3.89343709e-01 1.04591715e+00 -1.82736188e-01 -9.15680110e-01
-4.05980438e-01 -2.31961772e-01 -4.90680516e-01 -1.59637377e-01
-6.03085421e-02 -3.47329497e-01 -9.17865574e-01 1.51062119e+00
1.65930912e-01 -4.63170737e-01 3.89575064e-02 3.58211696e-01
1.17647576e+00 7.99713671e-01 4.82914984e-01 -3.89182180e-01
1.15389848e+00 -6.04770958e-01 -9.89611208e-01 2.25575402e-01
1.51938546e+00 -1.34296966e+00 1.13616228e+00 5.03832817e-01
-8.79739463e-01 -6.69313908e-01 -8.26398194e-01 -1.14296347e-01
-6.51697993e-01 5.98459065e-01 4.03138399e-01 8.54036212e-01
-8.05291831e-01 5.79219997e-01 -6.03599012e-01 -4.90084887e-01
-4.98343892e-02 2.94478923e-01 -4.63355601e-01 -8.02075565e-02
-1.29957318e+00 1.09008431e+00 8.13735306e-01 1.15796566e-01
-3.24440897e-01 -4.22998160e-01 -8.35983753e-01 -2.90584952e-01
-3.15259337e-01 5.74792102e-02 1.07038558e+00 -8.90108585e-01
-1.11555421e+00 8.05482030e-01 -3.22652012e-01 5.03911190e-02
7.53304809e-02 -3.30458015e-01 -1.01250517e+00 -3.26483190e-01
3.28505337e-01 4.49516997e-02 3.72120664e-02 -5.76691568e-01
-1.19097424e+00 -4.83923048e-01 -6.56403959e-01 1.32062390e-01
-4.53946978e-01 8.62258196e-01 -3.81076112e-02 -8.04201305e-01
3.13463926e-01 -5.80223441e-01 -4.17427011e-02 -9.36587572e-01
-1.88778654e-01 -9.31385696e-01 4.82687175e-01 -1.57395828e+00
2.16770005e+00 -2.08026767e+00 -1.70601025e-01 3.40196282e-01
-3.36356550e-01 4.30854261e-01 3.56691301e-01 4.62812155e-01
7.06998340e-04 7.25896537e-01 -8.24728683e-02 2.02583671e-02
-2.94139594e-01 1.02928117e-01 2.48337090e-01 1.80673137e-01
6.05679937e-02 3.68905365e-01 -9.05824840e-01 -7.84532249e-01
1.49062276e-02 -1.46982953e-01 -4.59648579e-01 5.56391299e-01
2.01431528e-01 4.34691608e-01 -6.70915723e-01 8.65139842e-01
7.40180671e-01 4.18607205e-01 -1.02136090e-01 4.51139092e-01
-6.92291439e-01 7.25030422e-01 -1.09032500e+00 1.48679650e+00
-5.33844292e-01 3.89020473e-01 -3.16698343e-01 -7.07518280e-01
1.38112366e+00 3.42970580e-01 -2.49107048e-01 -8.43408227e-01
2.68649071e-01 7.62037754e-01 1.42379940e-01 -1.15458965e+00
7.36102223e-01 4.06815782e-02 -3.46267037e-02 1.27471760e-01
2.12526411e-01 1.82108894e-01 3.46891314e-01 2.54893839e-01
7.93247938e-01 1.60052136e-01 2.80450761e-01 -6.93936825e-01
6.36838615e-01 1.07510455e-01 8.72367024e-01 2.83007562e-01
-7.53741618e-03 1.94468230e-01 8.57474431e-02 -9.27350894e-02
-6.82929933e-01 -4.04885620e-01 -5.21970928e-01 7.09960461e-01
-8.11361447e-02 -5.43831944e-01 -9.19095695e-01 -8.23728263e-01
-2.37082571e-01 9.05286610e-01 -1.43144578e-01 -1.26976863e-01
-8.36801171e-01 -7.13910520e-01 5.25161207e-01 2.13368803e-01
6.09924197e-01 -1.55137348e+00 -9.83170345e-02 5.77107370e-01
-2.60166436e-01 -6.25451922e-01 -3.93956035e-01 1.39214560e-01
-8.43590260e-01 -1.06074488e+00 -1.81710854e-01 -1.54757285e+00
1.11038625e+00 -2.99078345e-01 1.01433766e+00 1.08729458e+00
-1.53642625e-01 -3.74824479e-02 -9.45938766e-01 -7.29645312e-01
-7.59166360e-01 -9.59255546e-02 -8.43771361e-03 -8.62748086e-01
7.87625551e-01 -1.63355783e-01 6.92186737e-03 -7.04199448e-02
-5.30490875e-01 -1.13616720e-01 2.92366982e-01 7.91892290e-01
2.47837752e-02 4.39072847e-01 7.14080632e-01 -1.04165745e+00
7.37091660e-01 -3.48568916e-01 -5.75944006e-01 5.06384909e-01
-1.18576002e+00 -6.97566867e-02 5.15067875e-01 1.32150486e-01
-1.36943388e+00 -3.82821262e-01 -8.01478326e-01 5.68491280e-01
-4.60691214e-01 7.37601161e-01 -3.18599194e-01 -5.97468801e-02
2.65214115e-01 4.41107422e-01 -3.39805871e-01 -8.42462182e-01
-4.97941375e-01 1.21916807e+00 1.13063969e-01 -8.22130322e-01
1.78610250e-01 -6.85643375e-01 -4.17460084e-01 -7.86355138e-01
-1.38637289e-01 -3.41528714e-01 -6.82557166e-01 -3.42002869e-01
8.63572061e-01 -1.12420893e+00 -5.71133137e-01 1.00668693e+00
-1.43185866e+00 -3.90548445e-03 4.01393086e-01 9.48819041e-01
2.53079116e-01 5.59391558e-01 -9.79881763e-01 -8.39337111e-01
-4.87187207e-01 -1.09863663e+00 4.68751580e-01 5.56570113e-01
5.11518791e-02 -7.57847726e-01 -8.72964635e-02 5.53610846e-02
1.83656335e-01 -2.89130747e-01 1.43713319e+00 -6.19882703e-01
-2.20185757e-01 2.61912793e-02 1.66834831e-01 6.89036071e-01
-2.22844526e-01 3.56793255e-01 -3.47874016e-01 3.26337703e-02
-1.00913092e-01 -1.55867130e-01 4.38402236e-01 1.11330852e-01
1.08337498e+00 -3.69726121e-01 2.13411655e-02 3.78268749e-01
1.69984293e+00 7.96376824e-01 8.88515949e-01 4.60511357e-01
4.67185438e-01 6.13027811e-01 1.03081965e+00 3.23111683e-01
4.93018657e-01 2.33139336e-01 -2.92771030e-02 4.75540519e-01
-2.57296786e-02 -5.57193637e-01 6.39340520e-01 1.85546017e+00
-1.40344262e-01 -1.72708884e-01 -1.23181701e+00 8.10883641e-01
-1.44825518e+00 -6.45168304e-01 -8.36094797e-01 2.12465739e+00
1.22015727e+00 -6.01567924e-02 -4.18764204e-01 1.70939147e-01
9.78545427e-01 -7.69992828e-01 4.89725649e-01 -6.38915658e-01
-8.95559862e-02 6.42627239e-01 1.24183953e-01 5.05670667e-01
-6.13268852e-01 1.44342351e+00 5.57943916e+00 9.73933876e-01
-9.91477787e-01 1.76643372e-01 4.13044214e-01 6.71789646e-01
-5.03176510e-01 2.64740884e-01 -1.51706004e+00 1.02727377e+00
9.35806274e-01 -2.83369590e-02 1.53185531e-01 6.40160739e-01
3.68244499e-01 -3.16811413e-01 -3.88266742e-01 8.24956238e-01
-1.28310159e-01 -9.28228915e-01 -4.45527658e-02 -1.62437066e-01
6.24544680e-01 -1.86337605e-01 -5.55983365e-01 6.28543556e-01
3.71576130e-01 -9.76574063e-01 8.68630350e-01 4.69679415e-01
5.00074863e-01 -8.94449711e-01 1.08860540e+00 5.07229269e-01
-8.44260156e-01 8.19275454e-02 -8.44147801e-01 -4.74512815e-01
-7.21503422e-02 8.22145343e-01 -3.83670479e-01 6.61730647e-01
1.02544022e+00 8.07649493e-01 -8.45931828e-01 9.03836668e-01
-6.70269489e-01 1.09073150e+00 -1.98438093e-01 -5.46427846e-01
-9.35534835e-02 -4.86606359e-01 2.63737202e-01 1.40028059e+00
9.23559725e-01 4.11759138e-01 6.42905012e-03 4.44551051e-01
1.04082905e-01 9.35380399e-01 -2.66374439e-01 -1.81805007e-02
7.97731519e-01 8.12552512e-01 -4.70114559e-01 -2.22781003e-01
-3.31642091e-01 6.68420672e-01 4.99050915e-01 -1.60100486e-03
-5.10905385e-01 -9.94086325e-01 3.16526890e-02 -5.40821664e-02
-1.10582806e-01 -2.45899752e-01 -5.49617112e-01 -1.08558333e+00
1.61167324e-01 -9.76245880e-01 1.68334797e-01 -5.56354642e-01
-8.26825559e-01 5.52099347e-01 -4.43891019e-01 -1.00434208e+00
3.04141104e-01 -6.24703765e-01 -8.88669670e-01 1.26358795e+00
-1.19481611e+00 -9.63126063e-01 -4.80161682e-02 5.14172375e-01
4.49544191e-01 -5.44688344e-01 9.25650120e-01 8.69211495e-01
-8.92604172e-01 8.11737537e-01 1.70228884e-01 4.04078245e-01
7.58525252e-01 -1.16227663e+00 -2.31819466e-01 1.17160285e+00
-3.15814883e-01 6.92854643e-01 4.91748959e-01 -1.13509393e+00
-7.64078140e-01 -8.60861182e-01 2.12063956e+00 -7.89579526e-02
3.72527130e-02 -3.23311761e-02 -7.48691261e-01 5.45627594e-01
1.42482415e-01 -5.59228897e-01 5.52032173e-01 3.31929058e-01
6.30670488e-01 3.99716310e-02 -1.17651951e+00 2.13028774e-01
8.26016188e-01 -5.23236208e-02 -6.89531446e-01 2.80716091e-01
8.19705784e-01 -3.53195518e-01 -1.01375413e+00 3.06547970e-01
1.28065675e-01 -7.10875630e-01 1.17565379e-01 -8.54692101e-01
7.12629020e-01 -1.98797226e-01 1.22680888e-01 -1.31556952e+00
-2.87779510e-01 -2.44265780e-01 6.83138192e-01 1.91873205e+00
5.36644220e-01 -4.64750916e-01 3.09706926e-01 6.62294209e-01
-8.80708039e-01 -6.57302201e-01 -8.05523217e-01 -5.78194499e-01
3.84235650e-01 -3.99707735e-01 8.62831235e-01 1.13358974e+00
2.06313014e-01 -9.15817320e-02 -2.21975937e-01 -1.32685052e-02
-3.57810780e-02 -3.36917639e-01 3.77592146e-01 -1.24318039e+00
2.27131248e-01 -1.87567964e-01 -4.47877795e-01 -8.44409645e-01
1.95567042e-01 -7.91132808e-01 1.49064749e-01 -1.22656155e+00
-1.60598546e-01 -9.24926579e-01 -9.15628299e-02 4.58094597e-01
-5.36398590e-01 -4.81527925e-01 -1.20571025e-01 -2.71550417e-02
-2.14845568e-01 5.69891632e-01 1.06912816e+00 2.68681079e-01
-1.68885112e-01 -2.97235698e-02 -4.79723781e-01 7.37098813e-01
7.86142468e-01 -8.10203075e-01 1.93702310e-01 -7.18143761e-01
5.91559350e-01 -1.41599718e-02 -4.10657972e-01 -7.31977522e-01
1.31228194e-01 -2.56637752e-01 3.16039413e-01 -4.01528984e-01
-8.98792267e-01 -7.11230218e-01 -2.89337784e-01 7.85530090e-01
-5.42927012e-02 7.19265819e-01 9.86980274e-02 -5.09236865e-02
-4.70366329e-01 -1.10612857e+00 6.36731029e-01 -3.51848185e-01
-9.96476233e-01 4.12145592e-02 -6.54036760e-01 3.41799587e-01
7.04180300e-01 6.63603982e-03 -1.96134776e-01 -5.08664437e-02
-5.85651755e-01 4.46653366e-01 1.79731756e-01 3.84332806e-01
6.52109981e-01 -1.11333573e+00 -8.31309021e-01 3.85592878e-01
1.18636429e-01 -1.22823147e-02 2.79981285e-01 5.20056665e-01
-1.17958689e+00 1.18375503e-01 -1.91473231e-01 -2.59625688e-02
-1.37007463e+00 -1.18245576e-02 1.45454049e-01 -3.60914767e-01
-2.90294021e-01 9.97070849e-01 -5.63179851e-01 -7.94559121e-01
3.16029340e-01 -8.67660418e-02 -7.72200525e-01 -4.42274988e-01
3.68376404e-01 5.58246613e-01 3.81438583e-01 -7.78232038e-01
-2.50818342e-01 3.62906277e-01 -3.10085211e-02 2.03018486e-01
1.12003016e+00 -2.71656811e-01 -7.85317004e-01 2.54150510e-01
8.38211417e-01 4.69080210e-01 2.84022521e-02 1.92257062e-01
3.34800243e-01 -5.43656051e-01 -2.26628438e-01 -9.82479990e-01
-6.59850895e-01 9.14733052e-01 7.14720428e-01 -5.83427511e-02
1.02663314e+00 -4.44192261e-01 8.93690586e-01 2.45110333e-01
5.12470245e-01 -1.84086478e+00 -5.59436858e-01 9.90023375e-01
5.56369305e-01 -1.32113338e+00 -2.66025394e-01 -4.31733847e-01
-3.01611960e-01 1.40377104e+00 1.15486705e+00 -2.72226389e-02
1.22179902e+00 8.52668956e-02 8.34123045e-02 -5.25701046e-02
-3.59240621e-01 -5.22826053e-02 -3.16746086e-02 3.66915405e-01
1.22407913e+00 2.68405080e-01 -1.59072399e+00 1.00902307e+00
-5.81053495e-01 -2.22137533e-02 5.99914551e-01 1.08764744e+00
-8.75761688e-01 -1.67020440e+00 -2.38364771e-01 5.64970076e-01
-7.76927590e-01 -4.58760023e-01 9.49264392e-02 7.99791634e-01
8.28608572e-01 1.21165907e+00 -5.96471503e-02 -5.26114702e-01
2.74012625e-01 3.60744864e-01 4.41057384e-01 -8.54688168e-01
-1.19143057e+00 -2.61856496e-01 7.68342614e-02 -1.08574197e-01
-3.08310747e-01 -5.77881336e-01 -1.69806230e+00 -5.00566781e-01
-8.65313947e-01 7.50396729e-01 8.24885607e-01 1.33373129e+00
-1.42217672e-03 2.80414075e-01 4.81573552e-01 3.66541326e-01
-3.84315491e-01 -1.43388677e+00 -8.59567642e-01 4.81658757e-01
-3.00377429e-01 -2.33347163e-01 -2.66022682e-01 -6.45461306e-02] | [11.031722068786621, 10.782458305358887] |
b484951d-1883-43a7-95f4-6096cd7a3f1f | a-crf-based-framework-for-tracklet | 2011.14594 | null | https://arxiv.org/abs/2011.14594v2 | https://arxiv.org/pdf/2011.14594v2.pdf | A CRF-based Framework for Tracklet Inactivation in Online Multi-Object Tracking | Online multi-object tracking (MOT) is an active research topic in the domain of computer vision. Although many previously proposed algorithms have exhibited decent results, the issue of tracklet inactivation has not been sufficiently studied. Simple strategies such as using a fixed threshold on classification scores are adopted, yielding undesirable tracking mistakes and limiting the overall performance. In this paper, a conditional random field (CRF) based framework is put forward to tackle the tracklet inactivation issue in online MOT problems. A discrete CRF which exploits the intra-frame relationship between tracking hypotheses is developed to improve the robustness of tracklet inactivation. Separate sets of feature functions are designed for the unary and binary terms in the CRF, which take into account various tracking challenges in practical scenarios. To handle the problem of varying CRF nodes in the MOT context, two strategies named as hypothesis filtering and dummy nodes are employed. In the proposed framework, the inference stage is conducted by using the loopy belief propagation algorithm, and the CRF parameters are determined by utilizing the maximum likelihood estimation method followed by slight manual adjustment. Experimental results show that the tracker combined with the CRF-based framework outperforms the baseline on the MOT16 and MOT17 benchmarks. The extensibility of the proposed framework is further validated by an extensive experiment. | ['Huijun Gao', 'Zidong Wang', 'Huihui Pan', 'Tianze Gao'] | 2020-11-30 | null | null | null | null | ['online-multi-object-tracking'] | ['computer-vision'] | [ 2.05454856e-01 -3.43204230e-01 -1.12482816e-01 -2.98824608e-01
-4.14222449e-01 -3.57120395e-01 5.53576589e-01 4.56649698e-02
-6.49308085e-01 7.50541568e-01 -3.00764292e-01 -7.11094737e-02
1.03513792e-01 -5.08890569e-01 -6.01554453e-01 -1.13030386e+00
2.01566860e-01 1.98907450e-01 8.51823270e-01 2.26318449e-01
2.65425891e-01 3.54101002e-01 -1.63325572e+00 -3.22525471e-01
9.04459536e-01 9.02806222e-01 4.99052554e-01 5.93898535e-01
-9.98752117e-02 7.44953275e-01 -7.35146999e-01 -3.21430594e-01
2.80973345e-01 -2.38379285e-01 -3.30983438e-02 1.82882681e-01
5.10942221e-01 -2.23190576e-01 -1.28368437e-01 1.25948524e+00
5.86225927e-01 2.17740119e-01 3.29027265e-01 -1.29919422e+00
-1.18065432e-01 4.21194613e-01 -8.64016533e-01 4.82800931e-01
2.13870816e-02 2.30749771e-01 5.87361157e-01 -6.18035376e-01
2.55655020e-01 1.22724557e+00 9.13440466e-01 3.23450834e-01
-9.60057318e-01 -9.46995795e-01 2.94445366e-01 2.63175160e-01
-1.68413508e+00 -3.83483261e-01 5.60347497e-01 -4.96143997e-01
3.67558777e-01 1.93532139e-01 5.39065182e-01 6.57266557e-01
7.89589167e-01 6.06692791e-01 1.23132300e+00 -4.60646182e-01
2.11130992e-01 3.31944525e-01 3.46005976e-01 6.85383260e-01
5.68557799e-01 4.09415990e-01 -4.49681491e-01 -3.11699241e-01
7.09160805e-01 -4.77170013e-02 -2.40490362e-02 -3.28819841e-01
-1.05237925e+00 7.44253099e-01 3.56930137e-01 2.00864688e-01
-3.30398381e-01 1.41871691e-01 2.99585938e-01 -1.07224986e-01
3.44942719e-01 -3.37865293e-01 -1.74662888e-01 1.11747622e-01
-1.12332189e+00 2.33780444e-01 4.17895496e-01 1.07609689e+00
5.79404652e-01 -2.53293719e-02 -6.08569801e-01 3.75033796e-01
8.02926660e-01 7.67089605e-01 2.22037286e-01 -5.71488440e-01
2.47259185e-01 3.76766741e-01 2.49370784e-01 -1.11013544e+00
-3.37090999e-01 -6.31287634e-01 -5.41917741e-01 2.32369348e-01
4.48098570e-01 -2.34552681e-01 -9.96444881e-01 1.73068762e+00
9.67713833e-01 6.08333647e-01 -1.70634300e-01 8.02918911e-01
7.04045534e-01 3.68288189e-01 4.16819245e-01 -4.46076512e-01
1.44815087e+00 -7.80040979e-01 -1.17110097e+00 1.68939605e-02
3.29535395e-01 -9.75809872e-01 1.40562996e-01 1.64009705e-01
-7.27048159e-01 -8.82819474e-01 -1.06275189e+00 3.58312190e-01
-2.71997690e-01 2.20966920e-01 4.61162835e-01 9.43953395e-01
-8.73415351e-01 1.97154894e-01 -1.11526239e+00 -4.39941883e-01
3.64829391e-01 6.38657570e-01 1.60123289e-01 -3.61744352e-02
-9.52751040e-01 9.23796833e-01 6.04433060e-01 4.60544050e-01
-6.40142739e-01 -3.24075222e-01 -6.15091443e-01 -2.35899910e-01
4.66264367e-01 -6.69183433e-01 1.07998776e+00 -5.44785202e-01
-1.42079103e+00 2.53992558e-01 -2.13123068e-01 -5.21021605e-01
5.81266403e-01 -1.52022779e-01 -4.61532056e-01 -5.22018336e-02
9.04871523e-02 6.01137161e-01 1.03004718e+00 -1.10829055e+00
-1.01024508e+00 -2.20661059e-01 2.71702954e-03 2.46556550e-01
-7.86023214e-02 2.34098658e-01 -6.38044178e-01 -6.65299475e-01
-2.42338032e-02 -1.17321181e+00 -2.53834069e-01 1.32867560e-01
-2.30239972e-01 -2.54654557e-01 9.83436525e-01 -4.94635761e-01
1.30926466e+00 -2.08411074e+00 -3.77037197e-01 7.61708170e-02
7.21082166e-02 3.51338863e-01 3.83020014e-01 1.87159367e-02
5.31001031e-01 -5.37925482e-01 -1.53479110e-02 -3.15171003e-01
-1.06859878e-01 1.55386180e-01 -1.13435546e-02 8.16718161e-01
-5.52448444e-02 5.11120141e-01 -8.31241608e-01 -1.17810369e+00
5.83650768e-01 5.98424077e-01 -2.90268213e-01 2.26004332e-01
-2.37050608e-01 7.66186535e-01 -7.42504835e-01 6.76746130e-01
1.15874255e+00 -2.66425997e-01 6.61812499e-02 -2.91437894e-01
-4.91464138e-01 -2.60111928e-01 -1.47621095e+00 1.30721676e+00
2.93124355e-02 2.98413903e-01 2.79162843e-02 -4.82130885e-01
8.59068394e-01 2.78861523e-01 6.07187331e-01 -3.94460768e-01
3.72993678e-01 -1.75611451e-01 1.58040836e-01 -2.91398227e-01
7.92686880e-01 -6.76624626e-02 8.41071480e-04 5.69197237e-02
-1.03038520e-01 4.40517306e-01 1.92496106e-01 -1.16921766e-02
1.02982235e+00 4.81678873e-01 3.67806703e-01 -1.39449939e-01
6.77347183e-01 2.41954610e-01 9.41112876e-01 1.05698407e+00
-6.02726221e-01 1.97153568e-01 -4.58779246e-01 -7.63735399e-02
-4.67152178e-01 -8.57729137e-01 -3.75418454e-01 9.29196298e-01
7.27615893e-01 -2.24461094e-01 -5.05791187e-01 -5.97133338e-01
3.64087597e-02 3.93095404e-01 -4.08998460e-01 -8.35275836e-03
-6.43200040e-01 -9.95517790e-01 5.56019187e-01 4.24415559e-01
7.00645208e-01 -1.04192388e+00 -1.03113544e+00 4.67396885e-01
-1.18035249e-01 -1.29280531e+00 -3.61511827e-01 1.39291823e-01
-8.22658718e-01 -1.02326334e+00 -5.91837764e-01 -5.07381737e-01
6.10634446e-01 5.25717139e-01 5.33215761e-01 2.66136080e-01
-2.07366899e-01 2.43553370e-01 -4.62148845e-01 -4.09417808e-01
-3.29629719e-01 -2.05281854e-01 6.72622165e-03 1.52496517e-01
4.03252691e-01 -4.55386527e-02 -5.50909936e-01 6.39782488e-01
-7.50211716e-01 -1.14361882e-01 8.08567584e-01 7.02204466e-01
6.14474952e-01 2.76590168e-01 3.93100053e-01 -8.01731884e-01
2.38341466e-02 -4.14048553e-01 -9.63227868e-01 4.10934061e-01
-5.11450708e-01 -7.96914399e-02 4.18657005e-01 -6.28188491e-01
-1.30743945e+00 3.81088078e-01 -1.44459251e-02 -5.23843169e-01
-1.25692174e-01 9.10587460e-02 -2.07616597e-01 -5.61004639e-01
1.67786211e-01 2.34148383e-01 -1.34963393e-01 -3.58582109e-01
2.02055693e-01 5.35947263e-01 5.44855773e-01 -3.45139802e-01
1.07695210e+00 6.14540875e-01 3.10968086e-02 -7.17544258e-01
-8.03161323e-01 -7.18706310e-01 -5.07753372e-01 -6.79706812e-01
8.91851366e-01 -1.06975055e+00 -7.95576751e-01 4.81339186e-01
-1.05197215e+00 2.01310962e-01 1.69768810e-01 7.76044309e-01
-3.57186615e-01 6.70500696e-01 -3.64465117e-01 -1.21326315e+00
-2.68360645e-01 -1.13648295e+00 9.67922270e-01 8.14619303e-01
2.01607540e-01 -8.83888066e-01 8.65524188e-02 2.55273700e-01
3.65654618e-01 2.39733189e-01 2.41598099e-01 -5.74023187e-01
-9.69170690e-01 -1.15844131e-01 -1.29464492e-01 -1.42550319e-01
7.08683282e-02 -1.15650855e-01 -8.84990573e-01 -5.28496206e-01
4.42226306e-02 5.72623014e-02 7.62494981e-01 6.10929847e-01
7.05326200e-01 1.46895453e-01 -8.22584212e-01 3.63640279e-01
1.47938466e+00 3.35798353e-01 5.09492457e-01 3.69244099e-01
6.35672092e-01 9.64196846e-02 1.34715986e+00 6.24511123e-01
4.52466816e-01 8.66203666e-01 4.19743299e-01 1.19539246e-01
-1.52160734e-01 -1.86168887e-02 4.31156605e-01 6.48766637e-01
1.76356956e-01 -4.51619565e-01 -3.70824754e-01 3.17115098e-01
-2.01638222e+00 -9.53852475e-01 -3.44303250e-01 2.36075807e+00
5.49235642e-01 4.20525521e-01 3.24755870e-02 -8.50819722e-02
1.26742065e+00 8.74481201e-02 -5.32875240e-01 1.94982260e-01
1.41967922e-01 -3.86433750e-02 8.03424478e-01 3.73056680e-01
-1.27766001e+00 8.92888725e-01 5.65851927e+00 8.67534935e-01
-1.01211154e+00 2.09694698e-01 -1.60942703e-01 3.56212705e-01
4.02826548e-01 2.52979606e-01 -1.48879600e+00 8.04292977e-01
6.98366523e-01 2.80097753e-01 -1.30349537e-02 5.75028360e-01
3.70511055e-01 -4.85777289e-01 -5.54615736e-01 7.63875663e-01
-9.99944210e-02 -7.51772642e-01 -4.57205683e-01 6.86889067e-02
3.30169916e-01 -2.51114070e-02 -9.76747796e-02 3.67547929e-01
3.74071926e-01 -2.94263333e-01 8.35045159e-01 7.32561648e-01
3.73105526e-01 -5.02723336e-01 8.27508271e-01 4.62132871e-01
-1.69532919e+00 2.45557129e-02 -4.58825827e-01 2.36627325e-01
2.95952916e-01 5.79960167e-01 -8.87513936e-01 9.57454324e-01
7.05858827e-01 5.18453181e-01 -7.70943761e-01 1.69226360e+00
2.21967101e-01 6.69581950e-01 -6.42381489e-01 -8.13266113e-02
-9.34832096e-02 4.09475006e-02 8.09314907e-01 1.20735955e+00
1.48233891e-01 -7.77041167e-02 5.06845832e-01 6.22738421e-01
4.58982378e-01 -6.03337884e-02 -2.30101347e-01 3.02798420e-01
6.33284271e-01 1.49310708e+00 -1.12689555e+00 -2.75127679e-01
-4.10910070e-01 4.56099600e-01 3.80162224e-02 3.65092121e-02
-1.32961547e+00 5.09411506e-02 5.64120747e-02 -6.58634603e-02
9.66297090e-01 -1.72913581e-01 -4.32862379e-02 -8.85319948e-01
-7.23136514e-02 -4.92917627e-01 5.57729244e-01 -3.67137700e-01
-1.09582901e+00 4.84492123e-01 2.54407912e-01 -1.47803557e+00
-6.35977983e-02 -1.72141403e-01 -6.43628478e-01 5.21155059e-01
-1.71008611e+00 -1.21514213e+00 -4.36321288e-01 5.72488010e-01
5.20543039e-01 1.67407930e-01 3.14317495e-01 6.71430767e-01
-8.25514257e-01 7.24849164e-01 7.74800181e-02 -1.97673082e-01
7.53543079e-01 -8.09080184e-01 -1.06508709e-01 1.06532228e+00
-1.65869355e-01 6.78601682e-01 1.07610786e+00 -1.10878849e+00
-1.31525791e+00 -1.30022669e+00 6.88858271e-01 -2.23870978e-01
4.06886518e-01 -2.31156811e-01 -7.87289321e-01 6.41894817e-01
1.43464074e-01 3.33109945e-01 5.46965599e-01 -2.85849929e-01
3.66891325e-01 7.07852542e-02 -1.22027874e+00 2.17635050e-01
7.66670942e-01 2.21540153e-01 -4.75870460e-01 1.70239016e-01
4.00976539e-01 -5.56831300e-01 -8.11628699e-01 5.32032132e-01
6.05747879e-01 -7.22946525e-01 8.29344571e-01 1.57277748e-01
-6.07087255e-01 -1.07668340e+00 -3.36833745e-02 -6.93912923e-01
-3.16003799e-01 -4.76653934e-01 -4.10423189e-01 1.58000433e+00
-1.09667495e-01 -4.75530744e-01 7.86551893e-01 2.76228040e-01
-5.90535216e-02 -2.82437325e-01 -1.07549655e+00 -8.77355278e-01
-6.31599367e-01 9.55567323e-03 2.12207451e-01 5.28870821e-01
-6.44600630e-01 1.24657825e-01 -5.36451578e-01 5.74114263e-01
1.07446873e+00 6.77656159e-02 8.86170745e-01 -1.39222825e+00
-2.60211259e-01 1.40730634e-01 -4.52064902e-01 -1.14552152e+00
-1.04425661e-01 -5.53250313e-01 5.56827724e-01 -1.11802602e+00
2.91765898e-01 -6.81836486e-01 -5.70598125e-01 1.83704808e-01
-4.89774317e-01 1.89241827e-01 3.08189541e-01 3.33536357e-01
-1.15592599e+00 6.21286869e-01 9.57492530e-01 2.40598619e-02
-6.08623326e-02 3.32434237e-01 -2.58633435e-01 5.88767409e-01
5.76388538e-01 -1.00843704e+00 -2.24111192e-02 -1.49957761e-01
-3.87707919e-01 8.76486078e-02 4.91059929e-01 -1.31413805e+00
8.17723572e-01 -2.23456603e-02 4.85334694e-01 -1.30600476e+00
4.31278586e-01 -1.13788068e+00 4.76712525e-01 6.40388310e-01
1.24881856e-01 1.33702010e-01 3.02417576e-01 1.02080524e+00
6.89549297e-02 -1.98819548e-01 9.00769651e-01 8.78167450e-02
-7.39487708e-01 2.27076575e-01 -2.88956344e-01 -3.69977027e-01
1.23303294e+00 -4.56867397e-01 -1.42931357e-01 9.71707106e-02
-4.66871083e-01 5.31373262e-01 3.54052335e-01 3.28953028e-01
3.84918571e-01 -1.23918760e+00 -4.84441340e-01 -6.93677459e-04
1.07400537e-01 -9.03761014e-02 2.69032538e-01 1.26054192e+00
-1.10769928e-01 2.85222590e-01 -1.33005470e-01 -1.00138724e+00
-1.47011685e+00 5.16106129e-01 9.94830802e-02 -5.46536803e-01
-7.64006257e-01 4.92572576e-01 3.50298956e-02 -7.08753467e-02
4.54695255e-01 -1.53589398e-01 -4.10396159e-01 -2.12424397e-01
4.28819686e-01 3.07620972e-01 -1.56703979e-01 -8.45310569e-01
-5.82548499e-01 6.46442294e-01 -3.36733788e-01 1.71172246e-01
8.26254070e-01 -5.26726305e-01 3.07785004e-01 2.17201322e-01
3.96418989e-01 2.11387519e-02 -1.52177966e+00 -3.22431237e-01
9.39197168e-02 -5.63464522e-01 1.58322215e-01 -6.61175966e-01
-8.28102589e-01 3.30378950e-01 1.08962190e+00 3.84804383e-02
9.23058808e-01 -2.85096705e-01 7.67940760e-01 1.22918878e-02
5.52518666e-01 -9.51316297e-01 -4.41489995e-01 3.46777946e-01
8.92409682e-02 -1.29461789e+00 1.85078129e-01 -4.26338375e-01
-3.67937356e-01 8.90597343e-01 7.36681044e-01 -1.27299326e-02
5.15357316e-01 5.00720203e-01 4.05220948e-02 -9.05516818e-02
-5.01708508e-01 -4.07571822e-01 8.69007334e-02 4.35432613e-01
1.40998825e-01 -1.80045336e-01 -5.53795755e-01 2.68971890e-01
2.14444101e-01 2.98445791e-01 2.49674112e-01 1.34583259e+00
-7.36237943e-01 -1.05906487e+00 -9.64657664e-01 3.12635362e-01
-6.05197966e-01 2.46973455e-01 2.60297898e-02 8.03286076e-01
4.40838546e-01 1.23791969e+00 -1.09789588e-01 -3.78344238e-01
1.19290603e-02 -2.59506524e-01 5.60495853e-01 -3.26669276e-01
-8.45591426e-01 3.42932731e-01 -1.78355739e-01 -1.48646235e-01
-9.98873532e-01 -8.74938905e-01 -1.26182246e+00 -1.29154399e-01
-1.22219729e+00 1.37108207e-01 6.35087669e-01 1.14240539e+00
9.75891873e-02 5.55157185e-01 5.45620382e-01 -7.83590674e-01
-5.72617769e-01 -9.13540184e-01 -3.53353858e-01 -2.74970513e-02
3.35560858e-01 -1.33290792e+00 -2.54488200e-01 3.76292728e-02] | [6.4819655418396, -2.029845714569092] |
3151c0f0-b28e-438e-948e-9464287ec62d | simultaneous-segmentation-and-recognition | 1909.08606 | null | https://arxiv.org/abs/1909.08606v1 | https://arxiv.org/pdf/1909.08606v1.pdf | Simultaneous Segmentation and Recognition: Towards more accurate Ego Gesture Recognition | Ego hand gestures can be used as an interface in AR and VR environments. While the context of an image is important for tasks like scene understanding, object recognition, image caption generation and activity recognition, it plays a minimal role in ego hand gesture recognition. An ego hand gesture used for AR and VR environments conveys the same information regardless of the background. With this idea in mind, we present our work on ego hand gesture recognition that produces embeddings from RBG images with ego hands, which are simultaneously used for ego hand segmentation and ego gesture recognition. To this extent, we achieved better recognition accuracy (96.9%) compared to the state of the art (92.2%) on the biggest ego hand gesture dataset available publicly. We present a gesture recognition deep neural network which recognises ego hand gestures from videos (videos containing a single gesture) by generating and recognising embeddings of ego hands from image sequences of varying length. We introduce the concept of simultaneous segmentation and recognition applied to ego hand gestures, present the network architecture, the training procedure and the results compared to the state of the art on the EgoGesture dataset | ['Tejo Chalasani', 'Aljosa Smolic'] | 2019-09-18 | null | null | null | null | ['hand-segmentation'] | ['computer-vision'] | [ 4.51229215e-01 1.23784147e-01 -3.23720276e-02 -4.43255663e-01
-1.87104523e-01 -7.04164982e-01 9.13310409e-01 -8.14216852e-01
-6.06541574e-01 8.32101610e-03 4.12873387e-01 4.46018055e-02
1.47546172e-01 -5.21331251e-01 -5.76548934e-01 -9.28307772e-01
2.73411512e-01 6.68343902e-01 5.01838047e-03 1.27855778e-01
2.05146775e-01 9.91293550e-01 -1.68333769e+00 3.07432890e-01
-2.32431084e-01 7.11582839e-01 1.88721478e-01 1.20800388e+00
-2.19625890e-01 7.99667537e-01 -5.43280959e-01 7.79015720e-02
1.88337371e-01 -3.87718290e-01 -9.73701656e-01 2.57905453e-01
6.00507498e-01 -1.11803508e+00 -1.00266683e+00 6.45351052e-01
8.89702499e-01 5.59019864e-01 9.20549214e-01 -1.01072609e+00
-4.59397495e-01 6.65118396e-01 -2.60760218e-01 1.92317665e-01
5.91281354e-01 1.40757143e-01 6.50701702e-01 -8.37283194e-01
1.23370922e+00 1.26376498e+00 -1.80507019e-01 1.18506777e+00
-8.25730979e-01 -6.04950488e-01 2.69302249e-01 3.40688735e-01
-1.59596062e+00 -6.25244141e-01 5.48426330e-01 -4.72158462e-01
1.43024349e+00 4.21599537e-01 7.10749447e-01 1.66657567e+00
-2.23886594e-01 1.37728322e+00 6.33281231e-01 -4.36531931e-01
2.12124139e-01 -2.85304695e-01 2.28063643e-01 5.00214040e-01
-5.87467194e-01 1.73631776e-02 -4.52418000e-01 4.14988726e-01
1.44918299e+00 3.25718105e-01 -2.22082525e-01 -3.41984034e-01
-1.49916744e+00 6.12601340e-01 3.32303166e-01 5.16620576e-01
-7.23917663e-01 6.27690792e-01 2.46194109e-01 2.71975905e-01
7.00660497e-02 -3.37549180e-01 -2.30781049e-01 -5.47900796e-01
-7.98192859e-01 2.00373128e-01 8.34639966e-01 1.10930133e+00
1.33483917e-01 2.65615433e-01 -2.83883482e-01 4.22178209e-01
5.31664789e-01 4.43059355e-01 7.00626969e-01 -8.94166529e-01
1.98554680e-01 5.34257233e-01 -5.28444536e-02 -3.28408748e-01
-3.51125449e-01 2.03933448e-01 -2.83899695e-01 2.37034887e-01
5.80029726e-01 6.53243735e-02 -1.54156518e+00 1.55127025e+00
3.10019791e-01 3.39878112e-01 1.24205783e-01 1.36174238e+00
1.63659239e+00 2.30448961e-01 1.49374530e-01 1.25669092e-01
1.54462838e+00 -7.32301354e-01 -8.77315700e-01 4.34374139e-02
4.95090574e-01 -5.43981731e-01 8.69582415e-01 3.16366225e-01
-9.33973610e-01 -4.23653722e-01 -6.79144204e-01 -4.65921313e-02
-4.03689057e-01 1.38828665e-01 6.60677373e-01 1.01492155e+00
-1.07625151e+00 2.81977355e-01 -1.02869189e+00 -7.26100385e-01
4.23653126e-01 6.28639221e-01 -7.01729178e-01 9.52079799e-03
-5.78792036e-01 4.54492003e-01 3.31063509e-01 1.99032232e-01
-1.20343673e+00 -2.82658301e-02 -9.14774835e-01 -1.89854562e-01
2.04186082e-01 -4.85012352e-01 1.23598039e+00 -8.09299827e-01
-1.76630080e+00 1.28520894e+00 -3.98711681e-01 -2.23725170e-01
9.52531219e-01 -4.83163327e-01 -1.01685993e-01 3.75424325e-01
-3.62225235e-01 9.34137464e-01 1.07938945e+00 -8.75686109e-01
-3.01712394e-01 -8.95981610e-01 1.01097874e-01 5.92396021e-01
2.15748504e-01 2.47051507e-01 -7.82569706e-01 -4.00895447e-01
4.23370421e-01 -1.00352168e+00 1.20191589e-01 -5.81678271e-01
-2.41128743e-01 -6.53799891e-01 1.11493325e+00 -7.31361747e-01
5.82180083e-01 -2.10141587e+00 4.67477888e-01 4.38569002e-02
2.76625037e-01 9.59631354e-02 -3.42534393e-01 3.03494632e-01
-1.59844011e-01 -1.78342089e-01 2.64324009e-01 -2.42209449e-01
2.63752431e-01 4.51107413e-01 -4.15092766e-01 6.71642900e-01
-2.07724214e-01 1.33205080e+00 -8.39334071e-01 -3.30322862e-01
8.50332439e-01 1.02521932e+00 -3.12825769e-01 5.93221247e-01
-5.94914593e-02 8.03555667e-01 -5.99589229e-01 8.10988069e-01
9.26720053e-02 1.38386846e-01 1.16969928e-01 -4.94478643e-02
1.05049565e-01 1.18241213e-01 -1.38909280e+00 1.97794569e+00
-3.36391121e-01 1.08781457e+00 1.27564013e-01 -6.48183644e-01
7.13056922e-01 7.35854149e-01 4.75890040e-01 -5.63869536e-01
6.14575326e-01 -2.43273824e-02 -1.22425079e-01 -8.88505101e-01
4.24157262e-01 2.14088976e-01 3.40635091e-01 9.82232988e-01
1.77726507e-01 2.20769849e-02 7.92145450e-03 6.62039518e-02
1.07641649e+00 5.60594261e-01 -1.06386049e-02 4.40916926e-01
3.47631663e-01 -1.49984628e-01 -1.72467217e-01 7.78296232e-01
-3.50683481e-01 8.68475318e-01 4.87301424e-02 -5.40903568e-01
-6.79814816e-01 -1.14850593e+00 6.11534528e-02 1.45253706e+00
2.20894869e-02 7.63295665e-02 -1.01301217e+00 -6.08725965e-01
-4.85092729e-01 7.98661232e-01 -7.05968082e-01 3.85786295e-01
-1.02744794e+00 -2.42039993e-01 6.51756048e-01 1.03048420e+00
4.10134792e-01 -1.81143701e+00 -1.20348537e+00 -4.90126833e-02
-4.91504371e-02 -1.35694551e+00 -3.04216295e-01 -8.28358233e-02
-8.72779071e-01 -1.23074663e+00 -1.20103359e+00 -6.39946401e-01
4.78483498e-01 1.69801310e-01 8.97309542e-01 -2.44983256e-01
-7.20625758e-01 1.34212744e+00 -6.98735416e-01 -2.12558776e-01
-2.76032239e-01 2.80398633e-02 1.62770838e-01 -1.59308180e-01
9.76641357e-01 -5.23643076e-01 -7.33562648e-01 2.85762191e-01
-9.58614945e-01 -2.28369579e-01 4.68978077e-01 5.24609864e-01
3.32680225e-01 -9.02843654e-01 -1.43169388e-01 -4.97779042e-01
4.45988774e-01 8.19167793e-02 -1.75787270e-01 3.93035114e-02
5.80240078e-02 -2.68777400e-01 -2.77450323e-01 -5.69751322e-01
-8.64004850e-01 5.13580143e-01 -2.64030099e-01 -9.20527279e-01
-8.81514192e-01 -3.79974365e-01 -1.50627404e-01 7.85956159e-02
5.05655944e-01 5.25405169e-01 1.83443446e-02 -5.24177670e-01
9.19537604e-01 1.11810780e+00 4.50999498e-01 -1.80095002e-01
1.70318171e-01 7.02585876e-01 -3.37672830e-01 -1.31008756e+00
5.71416244e-02 -8.27191114e-01 -1.27891588e+00 -3.43184382e-01
1.34140098e+00 -7.68365443e-01 -1.10192156e+00 4.43828106e-01
-1.36794949e+00 -6.51886404e-01 -4.38516647e-01 5.37606895e-01
-1.02998114e+00 2.69612223e-01 -5.07545650e-01 -1.24104238e+00
-4.39546049e-01 -1.37574446e+00 1.55480099e+00 1.35748222e-01
-4.71757650e-01 -8.19996774e-01 -1.39707506e-01 2.38304004e-01
-4.45675179e-02 3.80758226e-01 2.50261366e-01 -9.01704371e-01
-8.80905569e-01 -2.58571059e-01 -3.51872563e-01 -1.14928544e-01
2.26852193e-01 -3.17698956e-01 -1.44620585e+00 -2.45509334e-02
-1.34822085e-01 -1.47155091e-01 8.54438365e-01 4.56198752e-01
8.15030158e-01 -2.99133938e-02 -3.41161907e-01 6.10733151e-01
9.52399969e-01 5.12332797e-01 7.21346617e-01 1.09144762e-01
1.03246009e+00 8.59611630e-01 4.24304307e-01 2.93841720e-01
2.52735838e-02 8.08040023e-01 5.86862922e-01 1.60009891e-01
-2.97519416e-01 9.84828770e-02 3.46390635e-01 2.64841795e-01
-6.65238917e-01 -2.22847313e-01 -8.07419002e-01 6.18851304e-01
-1.79248416e+00 -1.07048392e+00 -3.40488181e-02 1.69903445e+00
1.17057636e-01 -7.20856667e-01 2.83150285e-01 1.91035435e-01
4.80034441e-01 2.30195031e-01 -7.00689912e-01 -5.02657712e-01
1.27285749e-01 3.63434136e-01 4.15104032e-01 2.94021547e-01
-1.36156619e+00 1.37436903e+00 6.48591137e+00 2.13450447e-01
-1.22664177e+00 2.20234185e-01 -2.40681879e-03 -3.83412242e-01
3.41786981e-01 -6.25312030e-01 -6.72548890e-01 -2.71537036e-01
6.73736334e-01 4.76864606e-01 7.94020653e-01 1.04429090e+00
1.85020447e-01 4.23105121e-01 -1.58791113e+00 1.56134582e+00
4.14665818e-01 -9.77325439e-01 1.38178110e-01 2.05615088e-01
2.54634410e-01 5.33631563e-01 -5.64100854e-02 -1.36190569e-02
1.38875544e-01 -1.42504370e+00 7.75825441e-01 3.88848364e-01
1.05517650e+00 -4.52662677e-01 6.72023058e-01 3.50647509e-01
-1.10523462e+00 9.42891091e-02 -4.11285199e-02 -4.76981141e-02
2.85132825e-01 -5.18255055e-01 -1.24708760e+00 6.55431971e-02
6.82697058e-01 5.39620936e-01 -4.88818809e-02 1.93349272e-01
-4.25578982e-01 2.33144134e-01 -3.85151803e-01 -1.12583816e-01
3.74146909e-01 -2.33370066e-01 7.93993950e-01 1.49355066e+00
1.68913119e-02 5.66152930e-01 -1.47538468e-01 9.35608149e-01
-4.90764752e-02 1.26799941e-01 -8.08009744e-01 -4.91722584e-01
8.12869966e-02 1.01360834e+00 -1.07716632e+00 -4.50706512e-01
-6.68053478e-02 1.68017530e+00 -2.65999615e-01 6.58820391e-01
-3.63040149e-01 -3.93294811e-01 8.81552219e-01 -7.31628686e-02
4.24753755e-01 -4.61785376e-01 3.89521495e-02 -1.02117026e+00
-1.80183381e-01 -6.23986423e-01 2.99635440e-01 -8.42808425e-01
-6.71145141e-01 6.68895006e-01 5.28381132e-02 -9.75781798e-01
-1.14959240e+00 -1.14235270e+00 -4.21870112e-01 8.29380214e-01
-9.80910599e-01 -1.28965092e+00 -8.91848505e-01 7.54996061e-01
1.07222867e+00 -2.10454032e-01 1.03172410e+00 7.49406144e-02
-2.49056160e-01 3.76018792e-01 -2.09436998e-01 7.46276081e-01
4.28507894e-01 -1.22167170e+00 7.46289015e-01 4.69082415e-01
7.20336735e-01 9.59248066e-01 5.02163351e-01 -5.45218289e-01
-1.92179310e+00 -6.46777391e-01 4.31365967e-01 -1.05482686e+00
2.28583783e-01 -2.77451158e-01 -3.28001469e-01 1.20948684e+00
1.19702019e-01 -1.18985780e-01 6.35080695e-01 -1.85121953e-01
-1.86616600e-01 6.70568764e-01 -1.15233493e+00 8.08949053e-01
1.68296301e+00 -7.10758865e-01 -7.94072866e-01 1.91998079e-01
4.00875360e-01 -8.80300164e-01 -6.78815365e-01 5.85922450e-02
1.30677187e+00 -7.98463643e-01 1.19950438e+00 -8.54383528e-01
1.56280950e-01 -7.02338144e-02 -5.01914084e-01 -7.35683799e-01
2.33690903e-01 -4.90527093e-01 -3.67309511e-01 9.01737392e-01
-3.60770375e-01 -4.85571891e-01 1.32026482e+00 7.31478751e-01
2.39755765e-01 -2.05496773e-02 -1.09267008e+00 -5.68338633e-01
-3.38645011e-01 -7.68197954e-01 5.49987912e-01 7.63344228e-01
-1.16247848e-01 8.72874558e-02 9.01192650e-02 -8.70499387e-02
6.16792381e-01 5.46696559e-02 1.19992840e+00 -1.07305682e+00
6.69077188e-02 -5.32322407e-01 -7.38756776e-01 -1.21450365e+00
2.29812428e-01 -9.00745630e-01 3.62423770e-02 -1.75424755e+00
1.52492253e-02 9.37327519e-02 1.28708053e-02 2.80470550e-01
4.44451958e-01 7.94790745e-01 4.66101617e-01 3.44513208e-01
-3.32942367e-01 1.65833056e-01 1.19230866e+00 -2.56218821e-01
-6.18775606e-01 -2.17261136e-01 -6.25698417e-02 8.17994654e-01
4.67168242e-01 -1.46361291e-01 -2.68961728e-01 -3.29111665e-01
-4.05963391e-01 -7.51895383e-02 5.13832271e-01 -4.47953671e-01
8.61442313e-02 4.68613841e-02 5.43997109e-01 -8.84025931e-01
7.04655051e-01 -7.43265271e-01 1.24270424e-01 4.36990768e-01
-1.67672738e-01 -2.59745955e-01 -8.61743987e-02 4.61637527e-01
4.63638790e-02 -2.01534152e-01 1.22988030e-01 -4.79822993e-01
-1.18442869e+00 1.27239898e-01 -4.43388343e-01 -5.31501353e-01
8.21194947e-01 -8.10506940e-01 8.44035372e-02 -4.16320890e-01
-1.37382090e+00 -1.45763904e-01 4.31218565e-01 8.76437247e-01
1.04428959e+00 -1.00666177e+00 -6.70551062e-01 4.85745758e-01
1.41105279e-01 -8.12195800e-03 2.61628866e-01 7.63408661e-01
-6.68560982e-01 8.38922679e-01 -2.94492424e-01 -1.04324973e+00
-1.81497085e+00 1.23050600e-01 1.67794466e-01 2.88379401e-01
-1.10999155e+00 9.02910471e-01 4.81209457e-01 -4.66371477e-01
5.54129899e-01 -3.30538303e-01 -6.96521342e-01 1.10703535e-01
9.44185138e-01 5.56920111e-01 -1.43962562e-01 -1.19091773e+00
-4.65442300e-01 8.27087283e-01 8.87676030e-02 -5.84991455e-01
1.33379567e+00 -2.11437442e-03 2.27570012e-01 4.51101810e-01
9.89101887e-01 -4.25304472e-01 -1.16968524e+00 5.13187200e-02
-4.35357690e-01 -5.62411666e-01 8.51864815e-02 -5.93686879e-01
-9.77621317e-01 1.32210290e+00 1.21393061e+00 -1.27603412e-01
7.65706837e-01 3.75703245e-01 4.88138169e-01 7.05807805e-01
3.82567078e-01 -9.60833251e-01 1.95376858e-01 5.65931261e-01
9.94082272e-01 -9.90562201e-01 -3.61001790e-01 -3.01138043e-01
-6.50941372e-01 1.23239648e+00 1.91248775e-01 -2.74728119e-01
3.48302543e-01 9.84538496e-02 3.85687649e-01 -4.70285356e-01
-3.69128883e-02 -5.42724490e-01 3.04841101e-01 9.82119262e-01
6.11642003e-01 3.04488391e-01 -1.47955030e-01 3.16309482e-01
-2.63153851e-01 1.82380781e-01 4.35901582e-01 8.76166344e-01
-1.26268551e-01 -7.99915910e-01 -4.64861423e-01 1.97664231e-01
-5.43833077e-01 1.98849499e-01 -8.44741583e-01 9.50454712e-01
-3.30081463e-01 7.11206675e-01 4.03354853e-01 -3.53456587e-01
3.90362859e-01 5.70792139e-01 9.67213809e-01 -6.36042535e-01
-5.74639320e-01 2.90380299e-01 -1.44347455e-02 -9.82425034e-01
-6.30560040e-01 -8.61364424e-01 -1.49635780e+00 -9.12409052e-02
-2.44554933e-02 -5.22607923e-01 1.24625373e+00 1.12605298e+00
-9.68981907e-02 7.06198990e-01 -1.17946195e-03 -1.55051851e+00
-1.84847891e-01 -1.19496036e+00 -6.92519069e-01 4.07695979e-01
1.56993538e-01 -6.09127223e-01 -9.47771221e-02 2.82283604e-01] | [6.6451568603515625, -0.24887032806873322] |
3bdcec06-6277-4a3b-b832-500139d424c6 | distributed-sketching-for-randomized | 2203.09755 | null | https://arxiv.org/abs/2203.09755v1 | https://arxiv.org/pdf/2203.09755v1.pdf | Distributed Sketching for Randomized Optimization: Exact Characterization, Concentration and Lower Bounds | We consider distributed optimization methods for problems where forming the Hessian is computationally challenging and communication is a significant bottleneck. We leverage randomized sketches for reducing the problem dimensions as well as preserving privacy and improving straggler resilience in asynchronous distributed systems. We derive novel approximation guarantees for classical sketching methods and establish tight concentration results that serve as both upper and lower bounds on the error. We then extend our analysis to the accuracy of parameter averaging for distributed sketches. Furthermore, we develop unbiased parameter averaging methods for randomized second order optimization for regularized problems that employ sketching of the Hessian. Existing works do not take the bias of the estimators into consideration, which limits their application to massively parallel computation. We provide closed-form formulas for regularization parameters and step sizes that provably minimize the bias for sketched Newton directions. Additionally, we demonstrate the implications of our theoretical findings via large scale experiments on a serverless cloud computing platform. | ['Mert Pilanci', 'Burak Bartan'] | 2022-03-18 | null | null | null | null | ['distributed-optimization'] | ['methodology'] | [-2.91701943e-01 -3.61206144e-01 -1.47166038e-02 -9.84852612e-02
-1.10155165e+00 -1.09415221e+00 9.82016698e-02 1.46960586e-01
-4.69388932e-01 7.95034766e-01 2.25079164e-01 -4.32112336e-01
-4.00720179e-01 -5.23710430e-01 -7.68421650e-01 -8.95917356e-01
-4.90148693e-01 1.18305482e-01 -2.75060833e-01 -1.23408481e-01
6.08583212e-01 6.88938260e-01 -6.53537929e-01 -1.65489420e-01
7.07003117e-01 1.01057470e+00 -3.50148410e-01 1.16760159e+00
2.32758939e-01 5.56179702e-01 -5.80214798e-01 -3.92028153e-01
9.16973472e-01 -3.33573252e-01 -5.01472473e-01 1.92737691e-02
7.19127178e-01 -1.04583085e+00 -5.48178554e-01 1.19695771e+00
7.30524540e-01 2.76782691e-01 2.70336568e-01 -1.56816316e+00
-3.83629531e-01 6.59348905e-01 -6.16616964e-01 -1.40231326e-01
-1.76361110e-02 2.96444576e-02 1.10437310e+00 -8.25464308e-01
6.26557350e-01 8.57992053e-01 1.04377615e+00 3.48550767e-01
-1.38986194e+00 -6.78555906e-01 1.34105414e-01 -2.02638358e-01
-1.54059756e+00 -7.07218349e-01 6.10136688e-01 -2.26416960e-01
6.20073557e-01 6.77911341e-01 3.07738006e-01 2.93774098e-01
1.44414186e-01 6.51539326e-01 7.54969180e-01 -1.84373468e-01
5.63250244e-01 1.45791590e-01 1.37754723e-01 8.64712775e-01
8.32561612e-01 -1.97737381e-01 -7.06010699e-01 -9.92936313e-01
9.48689640e-01 2.40886778e-01 -2.70961761e-01 -5.51405191e-01
-1.01066923e+00 9.70830917e-01 8.78012851e-02 -2.50483066e-01
-4.21697527e-01 6.85037613e-01 4.96376127e-01 5.44075191e-01
5.82215667e-01 2.68557638e-01 -4.66863483e-01 -1.49115086e-01
-1.01264560e+00 5.72596133e-01 1.45540583e+00 1.09677470e+00
7.00262070e-01 -6.16160780e-02 -2.53607512e-01 4.45557594e-01
-1.09791629e-01 9.35538590e-01 -3.83189142e-01 -1.44247293e+00
7.52713084e-01 -7.03580305e-02 6.33854747e-01 -1.39784968e+00
-6.89407587e-02 -5.84760681e-02 -8.97841752e-01 1.19072922e-01
5.86345196e-01 -6.79660082e-01 1.69865638e-01 1.65203822e+00
5.40463507e-01 -2.30824277e-01 -1.74380720e-01 1.11679113e+00
-1.98619261e-01 6.49028480e-01 -6.25124276e-01 -3.07873428e-01
7.67227650e-01 -1.08435202e+00 -6.77187741e-01 3.18175852e-01
8.01243961e-01 -5.62378764e-01 7.61413932e-01 3.95855188e-01
-1.19690645e+00 4.20003653e-01 -8.98759007e-01 -2.50142157e-01
1.04948826e-01 9.16584581e-02 8.41970980e-01 7.44811535e-01
-1.13894963e+00 9.56897795e-01 -9.87227917e-01 1.93798414e-03
3.41059148e-01 3.46873552e-01 -1.66375175e-01 6.73797280e-02
-4.43005025e-01 1.77067950e-01 -5.31295598e-01 3.02644551e-01
-6.20568335e-01 -1.09086382e+00 -2.49234393e-01 2.75966197e-01
4.43386942e-01 -7.86611080e-01 1.13069463e+00 -5.33139169e-01
-1.55313313e+00 2.55281270e-01 -1.64864361e-01 -5.85479975e-01
8.44608009e-01 -3.07614326e-01 5.97045422e-01 3.13989967e-01
-9.81017500e-02 -3.04048836e-01 7.36084342e-01 -9.73571539e-01
-3.37280363e-01 -4.17050928e-01 8.45295265e-02 2.30580941e-01
-7.83509612e-01 -2.50849724e-01 -5.80215491e-02 -4.80034620e-01
-1.15480714e-01 -9.76345658e-01 -6.70856595e-01 8.19367766e-01
-3.92785609e-01 2.72662491e-01 4.36995178e-01 -7.21373498e-01
1.19250131e+00 -2.09352827e+00 9.80582610e-02 5.95676541e-01
7.58544683e-01 -1.43092126e-01 1.66754462e-02 1.03852963e+00
7.58071721e-01 2.67766714e-01 -1.01485610e-01 -5.48316121e-01
4.98402506e-01 5.51286004e-02 -6.30065322e-01 1.05795538e+00
-6.26182914e-01 5.64837217e-01 -7.39180863e-01 -1.71837524e-01
-2.36990318e-01 9.53087658e-02 -1.01424861e+00 3.12410325e-01
2.13188618e-01 -4.77524325e-02 -6.58316553e-01 1.55865476e-01
1.02315664e+00 -4.81400430e-01 4.81850743e-01 2.89949365e-02
-1.30791560e-01 1.00842744e-01 -1.54479492e+00 1.53830683e+00
-6.47907197e-01 5.70452094e-01 1.26483643e+00 -7.12067127e-01
4.37367886e-01 2.01102510e-01 5.56045055e-01 4.62591760e-02
-1.11975946e-01 4.31729108e-01 -6.25571489e-01 -2.42563918e-01
6.90439880e-01 -1.15859494e-01 -3.04382369e-02 9.90928411e-01
-5.87114811e-01 -3.07777077e-01 -1.31259859e-01 7.02516854e-01
1.30427170e+00 -6.34075761e-01 6.29314631e-02 -5.81977844e-01
1.11486681e-01 -2.92420536e-01 3.68938565e-01 1.13004601e+00
-1.75127134e-01 3.56509477e-01 8.87359142e-01 -4.17046577e-01
-1.40392935e+00 -8.26256633e-01 1.66057140e-01 1.24524188e+00
1.87469035e-01 -7.23219335e-01 -5.89979470e-01 -3.49854290e-01
7.92484820e-01 2.30545640e-01 -5.10536671e-01 2.82554448e-01
-2.20365345e-01 -3.99931967e-01 4.75389093e-01 3.29252750e-01
1.93325698e-01 -1.01094861e-02 -3.06374431e-01 3.19509879e-02
1.82653651e-01 -8.70908737e-01 -1.23804045e+00 -2.07396522e-01
-1.03433025e+00 -9.59810019e-01 -7.72617161e-01 -1.59547657e-01
8.84101808e-01 6.57194555e-01 5.12127697e-01 4.16213423e-01
-2.43183792e-01 8.77796769e-01 3.82623821e-01 -1.32197618e-01
1.04479678e-02 1.28458127e-01 1.05524682e-01 7.83247277e-02
-4.71027464e-01 -6.93494081e-01 -8.71876359e-01 2.08555773e-01
-9.02166069e-01 -7.64947459e-02 1.32788315e-01 9.33688760e-01
3.66427928e-01 -4.79345351e-01 2.49398813e-01 -9.47270691e-01
1.05645800e+00 -5.01970649e-01 -1.29186189e+00 2.20797390e-01
-8.46485734e-01 1.28816709e-01 8.95487487e-01 -4.13619339e-01
-7.09047735e-01 -1.91802531e-02 4.29346561e-01 -4.84409779e-01
7.91975439e-01 2.27931723e-01 4.52109784e-01 -7.40437388e-01
4.78035629e-01 2.62505591e-01 4.22834575e-01 -3.27505618e-01
5.65406084e-01 5.17087281e-01 1.68299302e-01 -1.18360817e+00
7.88035333e-01 9.00850832e-01 3.45838904e-01 -9.39944208e-01
-4.38484192e-01 -3.25669408e-01 2.26106700e-02 2.90140897e-01
-1.11359656e-01 -6.74817681e-01 -1.41757798e+00 7.62523189e-02
-1.24234450e+00 -4.62263256e-01 -3.15965891e-01 2.11706996e-01
-6.58871472e-01 6.72394156e-01 -1.00807786e+00 -1.30684614e+00
-7.34979451e-01 -7.80313611e-01 8.57728302e-01 -8.20000470e-02
1.13635004e-01 -1.05543947e+00 4.05729711e-01 -8.17983076e-02
1.03849280e+00 5.88969551e-02 3.65664601e-01 -5.20542204e-01
-9.41177547e-01 -3.17440331e-01 -3.90564203e-01 1.57324657e-01
-4.58233654e-01 -3.35258767e-02 -2.80733645e-01 -6.70254052e-01
1.62799191e-02 -2.37608269e-01 3.97196412e-01 2.79774934e-01
1.26151204e+00 -1.23097229e+00 -1.83916569e-01 1.05638742e+00
1.51519763e+00 -8.08156192e-01 2.40668327e-01 -1.51491344e-01
6.24384344e-01 2.26374969e-01 2.19219387e-01 1.40387297e+00
2.00115845e-01 4.77508217e-01 -8.12937319e-02 2.41770566e-01
4.19197947e-01 -2.00969741e-01 3.02567512e-01 1.15797305e+00
7.76318014e-02 4.22141179e-02 -5.31462967e-01 5.26914477e-01
-2.04586339e+00 -6.61422968e-01 -2.41679139e-03 2.66686058e+00
1.04624045e+00 -5.86442173e-01 3.27165633e-01 -3.03969771e-01
5.97232342e-01 2.52857320e-02 -6.76680803e-01 -4.51551855e-01
3.88854407e-02 1.28067732e-01 1.29746103e+00 8.59473765e-01
-6.25335753e-01 5.34994483e-01 6.90261412e+00 7.41315424e-01
-8.78661633e-01 2.99785703e-01 4.31191444e-01 -7.35643148e-01
-5.97366452e-01 2.55575955e-01 -4.43842560e-01 2.50640452e-01
8.28300893e-01 -7.65633285e-01 1.27439439e+00 9.85791624e-01
5.84151447e-02 7.54753221e-03 -1.01923919e+00 1.15820825e+00
-3.54271114e-01 -1.56312871e+00 -4.59031552e-01 4.58964348e-01
1.18199742e+00 -3.30765992e-02 7.13740960e-02 -4.06822115e-01
3.92503560e-01 -7.33053446e-01 3.70342612e-01 4.29208249e-01
7.54417777e-01 -7.68107116e-01 4.40606713e-01 1.59321934e-01
-1.00656426e+00 -1.01015441e-01 -7.00282276e-01 -3.98585498e-01
1.17589995e-01 7.32130826e-01 -4.48334694e-01 1.53960630e-01
2.83988446e-01 2.32416913e-01 -5.17091453e-02 1.10518360e+00
1.57733738e-01 6.53188348e-01 -1.06698406e+00 -4.33748543e-01
1.00842074e-01 -6.70397699e-01 7.05972314e-01 1.03091633e+00
3.20152253e-01 2.30700582e-01 8.67675245e-02 7.78290391e-01
-3.58807296e-01 4.24183249e-01 -6.82896495e-01 -3.66863072e-01
9.25658584e-01 1.30721426e+00 -4.42198813e-01 -3.21894526e-01
-2.33941227e-01 1.08236444e+00 4.94964808e-01 6.70581639e-01
-5.39441526e-01 -8.63505781e-01 1.13569343e+00 -1.08575493e-01
5.35662591e-01 -8.08214664e-01 -6.57327652e-01 -1.36929572e+00
4.65568304e-01 -7.67228007e-01 1.99787050e-01 1.43590868e-02
-1.53659225e+00 -1.19458027e-01 -3.37678730e-01 -5.44086337e-01
-1.29403159e-01 -2.42777944e-01 -5.80096841e-01 7.75076330e-01
-9.96715367e-01 -7.21596360e-01 8.06659758e-02 5.39631367e-01
-2.14117393e-01 1.66081339e-01 6.71861827e-01 3.33284706e-01
-4.84151691e-01 1.01539731e+00 9.83766317e-01 -1.16837285e-01
6.97195888e-01 -1.07986796e+00 1.76592812e-01 8.77699256e-01
-6.18935823e-02 9.99841273e-01 8.11095715e-01 -5.84033191e-01
-2.53464699e+00 -5.30921757e-01 5.91847599e-01 -1.24292560e-01
1.07334363e+00 -6.60690248e-01 -6.65524662e-01 5.66534460e-01
-9.53698382e-02 4.42365795e-01 5.53405762e-01 3.54447186e-01
-4.42217022e-01 -4.51667786e-01 -1.28380203e+00 5.44041693e-01
9.34563100e-01 -7.67142534e-01 5.34509003e-01 8.34339499e-01
5.14255226e-01 -5.36353707e-01 -9.11546350e-01 -6.16016448e-01
7.95270443e-01 -6.40785933e-01 9.18256760e-01 -7.20935464e-01
1.75049111e-01 1.59801412e-02 -3.82688254e-01 -8.83118033e-01
-3.34921442e-02 -1.60517895e+00 -3.74463648e-01 8.69295835e-01
1.22378372e-01 -1.00345325e+00 9.49429035e-01 1.26440513e+00
4.94222254e-01 -6.40176833e-01 -9.17632580e-01 -9.18156922e-01
2.46334404e-01 -1.35601565e-01 5.14867783e-01 8.68785739e-01
1.93672940e-01 -2.20709071e-01 -7.88268805e-01 3.62850308e-01
9.93640959e-01 2.14008331e-01 1.25484371e+00 -4.64051515e-01
-5.22258937e-01 -4.02147889e-01 -2.83111576e-02 -1.30354273e+00
3.81078534e-02 -7.80780852e-01 -1.11393280e-01 -9.66189325e-01
2.48594269e-01 -6.77741349e-01 1.69291645e-01 2.03081846e-01
-6.62830546e-02 -1.90704335e-02 4.74144965e-01 2.89840072e-01
-8.88323605e-01 7.25966156e-01 8.52150202e-01 1.86702609e-01
-9.12730843e-02 4.65507284e-02 -5.87451756e-01 2.41580963e-01
4.71085221e-01 -6.23162985e-01 -3.10870647e-01 -6.12617433e-01
5.88272333e-01 4.09649760e-01 3.63155872e-01 -4.73296374e-01
7.94672430e-01 -4.69566524e-01 -2.28219599e-01 -1.47882879e-01
1.37888923e-01 -7.89482355e-01 3.64089161e-02 5.54340661e-01
-6.84614837e-01 1.68575823e-01 -3.90011892e-02 8.33329082e-01
5.21042228e-01 -1.29423589e-01 5.10839224e-01 3.51766765e-01
2.80895680e-01 5.70012271e-01 -3.11526746e-01 3.05090338e-01
8.26671481e-01 4.46758926e-01 -2.61968315e-01 -9.66888011e-01
-2.00960308e-01 4.33332711e-01 6.61713660e-01 -4.72731650e-01
5.57129204e-01 -1.27484012e+00 -6.91778541e-01 -3.83806266e-02
-4.71333742e-01 -4.56658155e-01 2.82441050e-01 1.02318978e+00
-8.55982125e-01 1.68999031e-01 2.26036459e-01 -1.56480074e-01
-1.19720387e+00 5.73541760e-01 1.24399126e-01 -5.30372038e-02
-5.83008289e-01 8.28883350e-01 -1.62108541e-01 -4.13246989e-01
4.16815221e-01 -3.52122486e-02 1.27427340e+00 -5.12942672e-01
7.46822715e-01 8.95915210e-01 -1.65662333e-01 3.58868152e-01
-3.51339161e-01 1.39281243e-01 -3.68253551e-02 -5.39467156e-01
1.44189918e+00 -5.71335673e-01 -5.85856795e-01 1.81624442e-01
1.61145401e+00 5.86663783e-01 -1.42321682e+00 -3.51610273e-01
-4.26981866e-01 -8.64720523e-01 -2.54125185e-02 -2.07865104e-01
-1.15649855e+00 7.52926707e-01 1.62899226e-01 2.00221092e-01
7.48288572e-01 -4.21887994e-01 1.00691557e+00 9.56410050e-01
6.20741189e-01 -1.09169471e+00 -3.24274421e-01 3.40748012e-01
8.44767272e-01 -8.55301559e-01 5.48904598e-01 -3.03707302e-01
-2.67366886e-01 1.31649637e+00 -5.90981916e-02 -5.68900347e-01
6.13242209e-01 5.62830806e-01 -3.35143030e-01 3.96902002e-02
-9.57424104e-01 5.74526072e-01 -8.28318149e-02 1.37565657e-01
2.13140994e-01 1.46912292e-01 -7.22199321e-01 6.45486891e-01
-2.13661306e-02 -1.38632581e-02 6.74683928e-01 1.01931143e+00
-2.12367103e-01 -8.81329834e-01 -4.63529736e-01 4.72367674e-01
-5.15849233e-01 -2.11616069e-01 -2.62162089e-01 3.53693068e-01
-9.54428256e-01 6.93418920e-01 -1.17600597e-01 -5.95838390e-02
-1.58088803e-01 -2.66437531e-01 5.78897357e-01 -6.19262159e-02
-7.13334143e-01 -1.73806667e-01 -5.78155229e-03 -7.92005241e-01
4.58459020e-01 -4.37141746e-01 -9.46840703e-01 -1.41629207e+00
-3.30601960e-01 4.29897428e-01 9.68940914e-01 6.66952014e-01
9.34391558e-01 -2.88963437e-01 1.06402564e+00 -7.90187359e-01
-1.70042169e+00 -3.03737640e-01 -8.22835386e-01 1.63156331e-01
4.49035972e-01 4.93639931e-02 -6.91963911e-01 -4.19818103e-01] | [6.23539400100708, 5.023632526397705] |
e7810e46-df6f-4fad-bc17-ede3d918783a | bayesian-nvh-metamodels-to-assess-interior | 2207.02120 | null | https://arxiv.org/abs/2207.02120v1 | https://arxiv.org/pdf/2207.02120v1.pdf | Bayesian NVH metamodels to assess interior cabin noise using measurement databases | In recent years, a great emphasis has been put on engineering the acoustic signature of vehicles that represents the overall comfort level for passengers. Due to highly uncertain behavior of production cars, probabilistic metamodels or surrogates can be useful to estimate the NVH dispersion and assess different NVH risks. These metamodels follow physical behaviors and shall aid as a design space exploration tool during the early stage design process to support the NVH optimization. The measurement databases constitute different noise paths such as aerodynamic noise (wind-tunnel test), tire-pavement interaction noise (rolling noise), and noise due to electric motors (whining noise). This research work proposes a global NVH metamodeling technique for broadband noises such as aerodynamic and rolling noises exploiting the Bayesian framework that takes into account the prior (domain-expert) knowledge about complex physical mechanisms. Generalized additive models (GAMs) with polynomials and Gaussian basis functions are used to model the dependency of sound pressure level (SPL) on predictor variables. Moreover, parametric bootstrap algorithm based on data-generating mechanism using the point estimates is used to estimate the dispersion in unknown parameters. Probabilistic modelling is carried out using an open-source library PyMC3 that utilizes No-U-Turn sampler (NUTS) and the developed models are validated using Cross-Validation technique. | ['L. Gagliardini', 'J. Antoni', 'O. Sauvage', 'V. Prakash'] | 2022-06-12 | null | null | null | null | ['additive-models'] | ['methodology'] | [-4.02038634e-01 -3.34112972e-01 3.85637790e-01 -2.14568362e-01
-6.16765618e-01 -3.35233271e-01 4.99866605e-01 9.18529406e-02
-1.00142397e-01 9.61311340e-01 1.17329337e-01 -6.46981657e-01
-7.22153008e-01 -1.00119400e+00 -4.22668964e-01 -7.44663239e-01
2.27411062e-01 4.25468117e-01 4.51629519e-01 -2.19111472e-01
2.03307241e-01 6.05878294e-01 -2.06527710e+00 -6.33230209e-02
8.18434954e-01 7.62902677e-01 3.10089618e-01 7.24916220e-01
-1.06416449e-01 1.74268112e-01 -7.11865127e-01 -1.46129690e-02
-1.14993833e-01 1.49072558e-02 -2.03247741e-01 -5.01955092e-01
-4.64706153e-01 -7.92021453e-02 3.23901445e-01 9.08551991e-01
4.72428411e-01 6.13261640e-01 1.18754852e+00 -1.23185670e+00
-9.49799735e-03 2.42518276e-01 -6.77078366e-02 2.76335352e-03
-5.86497150e-02 3.23944628e-01 2.60708392e-01 -6.79191649e-01
3.93474996e-02 1.21271431e+00 5.39131343e-01 8.37970227e-02
-1.15137148e+00 -8.71129513e-01 -4.80146766e-01 3.45052361e-01
-1.43036604e+00 4.60614488e-02 8.91932428e-01 -8.08155954e-01
6.47399485e-01 4.88776147e-01 6.00779533e-01 9.66145813e-01
5.89952946e-01 8.11885223e-02 1.34802699e+00 -2.52588630e-01
6.40965164e-01 5.76175928e-01 4.90351975e-01 -3.66045758e-02
3.90757591e-01 6.68403864e-01 -7.05749989e-02 -3.77052963e-01
2.61344053e-02 -5.76954901e-01 2.18339726e-01 1.99231789e-01
-4.23641354e-02 8.37230027e-01 -3.18261206e-01 1.81156248e-01
-5.94551802e-01 1.86103344e-01 4.52726334e-01 -1.43223107e-01
2.43587822e-01 -1.56897828e-02 -3.92828047e-01 -4.19499189e-01
-7.66511858e-01 5.67723989e-01 8.69911671e-01 6.68551087e-01
5.72135448e-01 4.82115567e-01 -6.69019818e-02 7.26251960e-01
9.18125451e-01 9.28559124e-01 -8.50752816e-02 -7.43733823e-01
1.15834340e-01 -1.23469949e-01 3.48798662e-01 -9.73080516e-01
-3.66170198e-01 -4.86736119e-01 -4.46519762e-01 6.91859245e-01
2.04420481e-02 -3.78984898e-01 -7.90848911e-01 1.33366954e+00
5.75075209e-01 1.18400916e-01 -3.05923820e-02 3.17925543e-01
6.36978686e-01 8.95177722e-01 4.15746361e-01 -1.67736232e-01
1.15897322e+00 -7.26138204e-02 -9.19438064e-01 4.48519021e-01
3.16136658e-01 -9.92462039e-01 7.55895972e-01 6.99273288e-01
-6.71153784e-01 -7.98568130e-01 -1.02874863e+00 8.38290095e-01
-5.98859668e-01 -6.35917904e-03 -5.27950525e-02 1.48950005e+00
-4.58075315e-01 4.08393532e-01 -7.45928049e-01 9.93850082e-02
-6.66501820e-02 5.34097478e-02 2.67909896e-02 2.59291559e-01
-1.38057756e+00 1.04878008e+00 2.57429369e-02 5.97550690e-01
-1.09125173e+00 -9.45589900e-01 -6.10798597e-01 -2.42586404e-01
4.86818962e-02 -4.85022455e-01 1.00880015e+00 1.15626074e-01
-1.92767191e+00 -2.13097900e-01 1.11293942e-01 -1.51409119e-01
4.56016272e-01 -2.27181181e-01 -7.96826422e-01 -2.04504669e-01
-1.90821305e-01 -4.62333858e-01 6.24450266e-01 -1.44401562e+00
-2.95262367e-01 -3.57126333e-02 -4.93734807e-01 -2.73334742e-01
3.77400041e-01 2.53194183e-01 4.02366698e-01 -3.18775594e-01
-1.65180996e-01 -8.54572594e-01 -2.09373251e-01 -7.45254636e-01
-6.50612652e-01 -2.74289995e-01 9.42611635e-01 -8.84637773e-01
1.23557901e+00 -1.84958899e+00 -4.64188874e-01 9.01577234e-01
-5.23030937e-01 3.24494690e-01 4.74219501e-01 8.07860613e-01
3.59038673e-02 -3.88203491e-03 -9.49268341e-02 -5.46484180e-02
2.53292799e-01 3.77476215e-02 -1.18678156e-03 4.22578514e-01
-1.51262268e-01 8.72958004e-02 -5.18619299e-01 -2.76003510e-01
7.73089826e-01 5.34014881e-01 -3.61597925e-01 3.00593019e-01
4.12177332e-02 4.36180115e-01 -5.62760472e-01 2.67757744e-01
1.14739132e+00 8.09617996e-01 -3.31722975e-01 -5.64305663e-01
-5.60677350e-01 -1.78131253e-01 -1.54754317e+00 6.34029210e-01
-9.46414769e-01 3.06293994e-01 3.40027481e-01 -8.25430334e-01
1.18671274e+00 3.56676012e-01 3.11770216e-02 -1.51219204e-01
3.56162548e-01 3.23371232e-01 3.93975433e-03 -7.93484807e-01
4.00501072e-01 -2.48246729e-01 2.07600251e-01 -1.25798717e-01
-1.64417222e-01 -6.45250976e-01 -1.75875932e-01 -3.68034661e-01
6.06366694e-01 1.74151868e-01 -1.23236865e-01 -5.17554641e-01
7.44045079e-01 -2.42202878e-01 2.70060807e-01 3.79269838e-01
1.91340089e-01 8.15567598e-02 3.94524336e-01 2.10038409e-01
-8.98660958e-01 -1.20167887e+00 -5.46207905e-01 3.94152343e-01
-1.34217560e-01 1.83454081e-01 -5.81221879e-01 1.93328649e-01
1.24206029e-01 1.55841386e+00 -3.81351084e-01 -2.88891524e-01
-2.96049923e-01 -7.78038681e-01 3.84804755e-01 4.29038942e-01
7.79005587e-02 -6.72031045e-01 -4.80215877e-01 4.37394887e-01
4.88860518e-01 -8.41389775e-01 4.43212837e-01 2.23000407e-01
-5.21990895e-01 -8.73569906e-01 -1.49159983e-01 2.64008716e-02
2.04848424e-01 -3.39173794e-01 7.00353622e-01 -3.99545282e-01
-4.56231743e-01 6.08594239e-01 -2.40322500e-01 -8.45487416e-01
-9.78766620e-01 -5.60362220e-01 -2.10770406e-02 1.99305620e-02
7.28318393e-02 -6.37879670e-01 -4.47851717e-01 8.00409794e-01
-7.04516530e-01 -4.80228037e-01 2.17914253e-01 4.56780642e-01
4.82737392e-01 7.79667914e-01 7.18289733e-01 -6.50717616e-01
8.21153343e-01 -8.87124419e-01 -1.01413071e+00 -5.37990108e-02
-4.84128118e-01 -2.50653863e-01 5.51346004e-01 -1.80509925e-01
-1.59417427e+00 -4.71728951e-01 -4.88205463e-01 -2.76619315e-01
-6.82180166e-01 6.12190962e-01 -6.03575468e-01 1.69819623e-01
2.97236711e-01 -1.53792307e-01 5.92508204e-02 -6.43972754e-01
1.02642246e-01 8.40739965e-01 3.37986290e-01 -9.87521648e-01
9.17342842e-01 6.26792014e-02 5.63333035e-01 -1.33742464e+00
-1.35410381e-02 -2.90359139e-01 -6.97434172e-02 -7.70855010e-01
9.14799511e-01 -4.96754736e-01 -9.62229073e-01 6.02441549e-01
-9.82810616e-01 -5.77105060e-02 6.38887435e-02 9.51624691e-01
-4.69645053e-01 1.61657572e-01 -1.44808218e-01 -1.87883055e+00
-6.18762970e-02 -1.39862692e+00 4.74728465e-01 1.55926317e-01
-2.29569063e-01 -8.01878333e-01 1.04445696e-01 4.32476848e-01
5.34730196e-01 5.02520978e-01 1.19092238e+00 -5.72402716e-01
-3.54726613e-01 -4.06860292e-01 2.69382358e-01 1.02104616e+00
-1.81172997e-01 5.90828776e-01 -9.15298402e-01 3.40228498e-01
1.87340647e-01 1.81419715e-01 2.61571050e-01 8.39717448e-01
9.84351218e-01 1.28376111e-01 -1.30445302e-01 9.81087983e-02
1.57324815e+00 6.68962777e-01 7.03240395e-01 8.04734677e-02
3.53725761e-01 1.01699555e+00 7.76312947e-01 6.39541686e-01
-3.36864917e-03 5.78240693e-01 6.50867105e-01 3.84745270e-01
1.46101534e-01 -1.33560151e-01 1.18410841e-01 6.04710698e-01
-5.74995540e-02 -3.43957603e-01 -6.96096122e-01 3.87681901e-01
-1.18061566e+00 -1.01691663e+00 -8.35914135e-01 2.45083261e+00
4.20308560e-01 3.29414219e-01 4.94399704e-02 4.48877394e-01
8.13673735e-01 -2.35433877e-01 8.89915302e-02 -9.41947579e-01
2.42111400e-01 3.43988568e-01 7.20891535e-01 7.23984301e-01
-6.59754395e-01 1.26246244e-01 5.86563730e+00 1.27599883e+00
-5.58943212e-01 6.12789690e-02 4.59714204e-01 3.21810663e-01
-6.53015077e-01 9.54204425e-02 -9.79802191e-01 7.82516837e-01
1.38782024e+00 3.56309377e-02 8.96385387e-02 7.72659123e-01
9.88173366e-01 -6.54484093e-01 -3.54596168e-01 4.25516546e-01
-6.16833627e-01 -9.39127505e-01 -3.97611082e-01 1.82096109e-01
6.56751990e-01 -3.67858052e-01 1.95457533e-01 1.41658321e-01
4.06165928e-01 -9.22939062e-01 7.09846079e-01 1.14671230e+00
3.14394325e-01 -1.19978154e+00 1.14859974e+00 3.88891101e-01
-9.62958038e-01 -4.93372791e-03 -2.08764583e-01 2.38781482e-01
6.64120555e-01 1.02141440e+00 -9.10742462e-01 7.78200507e-01
5.75013936e-01 -4.75455523e-01 -7.75504392e-03 1.17640102e+00
8.89886990e-02 1.28715074e+00 -7.86919415e-01 -3.67452621e-01
-2.84406487e-02 -5.24534583e-01 8.23441386e-01 1.01932335e+00
7.51142204e-01 -2.19129533e-01 -4.27877784e-01 1.19916236e+00
9.58715439e-01 2.94609278e-01 -3.78061980e-01 1.16565615e-01
8.19288135e-01 1.07618022e+00 -3.58197719e-01 2.30010644e-01
-1.89907566e-01 -2.40541145e-01 -7.82748163e-01 6.70952141e-01
-1.06264412e+00 -5.28946579e-01 5.77152908e-01 4.87029970e-01
2.81577408e-01 -1.82399571e-01 -2.15568379e-01 2.82980293e-01
-3.18254888e-01 -2.85723954e-01 -1.70623034e-01 -7.79515386e-01
-1.25804961e+00 3.01863581e-01 7.89425135e-01 -1.06790769e+00
-3.09386402e-01 -8.53603661e-01 -1.05468035e+00 1.32634807e+00
-9.30765927e-01 -9.30962265e-01 -9.13995877e-02 9.60372463e-02
1.22419551e-01 -2.54505754e-01 5.77972889e-01 2.40724489e-01
-4.68317449e-01 2.53830701e-01 6.23866737e-01 -4.03729379e-01
9.26075503e-02 -7.35419095e-01 -2.02919364e-01 4.37558144e-01
-7.94290006e-01 5.67332506e-01 1.53240669e+00 -9.43815649e-01
-1.06484628e+00 -9.04317439e-01 2.94418067e-01 -2.47421041e-01
8.42104435e-01 -3.71453166e-02 -7.38671243e-01 -1.39280334e-01
6.91702887e-02 -1.46010548e-01 8.46102595e-01 -5.70217259e-02
2.77911097e-01 -2.80435294e-01 -1.33576369e+00 2.03772858e-01
5.36825173e-02 -3.22693229e-01 -3.99249852e-01 -1.50284037e-01
5.81414402e-01 1.51014403e-02 -9.63746727e-01 4.87719804e-01
5.77291787e-01 -1.19053149e+00 7.83462465e-01 -3.86620104e-01
-1.16149984e-01 -5.71006656e-01 -2.96484262e-01 -1.15416062e+00
1.63130626e-01 -5.41335225e-01 1.57818288e-01 1.58898640e+00
4.27255690e-01 -6.41354561e-01 4.34307367e-01 7.71464825e-01
-4.09560978e-01 -8.22376847e-01 -1.26847923e+00 -9.21248317e-01
-1.03891954e-01 -1.19000697e+00 7.80194044e-01 3.05554606e-02
-6.67331994e-01 -2.11228147e-01 -2.48451829e-01 4.63137895e-01
8.76062632e-01 -6.22656763e-01 7.52638102e-01 -1.40101969e+00
-3.15131813e-01 -1.60192221e-01 -3.96252722e-01 -1.21204004e-01
8.83757100e-02 -2.31553748e-01 2.32580721e-01 -1.27788734e+00
-4.15592819e-01 -5.00816345e-01 -2.50788867e-01 -4.97628987e-01
1.48453683e-01 -3.85471255e-01 -2.57849872e-01 -5.96754730e-01
3.91766846e-01 8.80836844e-01 7.27396309e-01 2.47494981e-01
-1.46847829e-01 8.63924503e-01 -7.09438324e-02 7.63220847e-01
6.31298065e-01 -6.01325274e-01 -7.48621881e-01 4.48901057e-01
3.57468516e-01 2.59531230e-01 7.40158916e-01 -1.19994414e+00
-3.56967151e-02 -2.31013089e-01 -1.89495027e-01 -1.29400873e+00
2.80041248e-01 -1.08200192e+00 9.90667820e-01 2.40044236e-01
5.02874404e-02 -1.47947567e-02 4.31469113e-01 8.12358499e-01
-1.40038446e-01 -6.82697117e-01 6.29261255e-01 2.21595168e-01
-1.61928028e-01 -1.15309015e-01 -1.05134618e+00 -3.18390697e-01
1.05244339e+00 -5.77732146e-01 -1.45900827e-02 -3.63201648e-01
-7.03083515e-01 -1.10639364e-01 -8.59763697e-02 -2.22010911e-02
3.96307260e-01 -1.11202240e+00 -5.53191364e-01 -3.99460958e-04
-2.46391788e-01 -3.11440229e-01 1.01422143e+00 6.88081920e-01
-6.09622538e-01 2.47215018e-01 4.83065099e-02 -3.47121894e-01
-9.79895532e-01 3.10995162e-01 4.61657643e-01 4.74571921e-02
1.18123032e-01 7.98512042e-01 -9.88116637e-02 -4.30870861e-01
-1.32925391e-01 -2.82318473e-01 -4.19503480e-01 2.79592704e-02
1.45374015e-01 1.23062527e+00 1.23792671e-01 -6.97304308e-01
-3.70353013e-01 5.29147446e-01 6.80114448e-01 -4.39697266e-01
1.00445378e+00 -2.18479395e-01 -6.27523065e-02 6.45822704e-01
1.16883230e+00 4.11167502e-01 -9.42343354e-01 3.61507177e-01
-1.18416764e-01 -1.87253490e-01 3.49153966e-01 -8.75163436e-01
-5.22324741e-01 7.50715256e-01 7.50217974e-01 1.42847613e-01
8.91659498e-01 -3.35196346e-01 2.94455022e-01 -7.76575506e-02
2.22329229e-01 -1.59659076e+00 -5.42129815e-01 2.74239063e-01
7.80550361e-01 -6.78858995e-01 -3.19768414e-02 -4.09702599e-01
-4.61108655e-01 9.70353723e-01 3.79634440e-01 3.58556286e-02
1.55753374e+00 6.53438389e-01 -1.06316328e-01 4.49807011e-02
-5.69311976e-01 2.12724116e-02 4.43397880e-01 7.71982670e-01
1.68255985e-01 3.47093612e-01 -4.06522989e-01 1.21759975e+00
-2.78452098e-01 -1.60156637e-01 3.22810024e-01 5.56031108e-01
-6.16185546e-01 -9.99723196e-01 -8.64264369e-01 3.91807824e-01
-2.87001878e-01 1.45706221e-01 4.77117538e-01 7.77619421e-01
4.32105035e-01 1.47001505e+00 -1.53538734e-01 -5.74743390e-01
6.44301653e-01 2.11463273e-02 -5.10908552e-02 -3.08439970e-01
-3.32760185e-01 7.36015514e-02 5.92023671e-01 -1.17336504e-01
-8.05541649e-02 -7.30703175e-01 -1.06217766e+00 -3.49090666e-01
-5.92498481e-01 6.73406124e-01 1.29679406e+00 1.02963340e+00
-1.50922099e-02 7.59828389e-01 7.61756420e-01 -5.22237837e-01
-7.05448151e-01 -1.07028532e+00 -1.03840208e+00 -2.68245220e-01
-1.19439915e-01 -1.17987835e+00 -7.69236267e-01 -4.64096129e-01] | [6.260730266571045, 3.2482244968414307] |
cc9fe56a-4e3c-4b26-b8f9-ec43a2bf1dcd | why-you-should-try-the-real-data-for-the | 2107.13938 | null | https://arxiv.org/abs/2107.13938v1 | https://arxiv.org/pdf/2107.13938v1.pdf | Why You Should Try the Real Data for the Scene Text Recognition | Recent works in the text recognition area have pushed forward the recognition results to the new horizons. But for a long time a lack of large human-labeled natural text recognition datasets has been forcing researchers to use synthetic data for training text recognition models. Even though synthetic datasets are very large (MJSynth and SynthTest, two most famous synthetic datasets, have several million images each), their diversity could be insufficient, compared to natural datasets like ICDAR and others. Fortunately, the recently released text-recognition annotation for OpenImages V5 dataset has comparable with synthetic dataset number of instances and more diverse examples. We have used this annotation with a Text Recognition head architecture from the Yet Another Mask Text Spotter and got comparable to the SOTA results. On some datasets we have even outperformed previous SOTA models. In this paper we also introduce a text recognition model. The model's code is available. | ['Vladimir Loginov'] | 2021-07-29 | why-you-should-try-the-real-data-for-the-1 | https://arxiv.org/abs/2107.13938 | https://arxiv.org/pdf/2107.13938 | null | ['scene-text-recognition'] | ['computer-vision'] | [ 5.81018329e-01 -6.71834573e-02 6.57804608e-02 -6.39055252e-01
-5.83506525e-01 -3.58383715e-01 1.15757227e+00 -2.69568473e-01
-4.15053993e-01 7.03858852e-01 1.49674699e-01 -8.09931755e-02
2.67176211e-01 -5.63742697e-01 -6.12425089e-01 -7.11843610e-01
6.45473123e-01 1.17627084e+00 2.74300575e-01 2.25245580e-02
4.83879536e-01 9.06854272e-02 -1.82476425e+00 8.60647023e-01
7.47500718e-01 8.45269501e-01 1.57509401e-01 7.23232925e-01
-5.15677392e-01 9.15704429e-01 -6.84819698e-01 -4.81963098e-01
2.95273662e-01 -4.69280422e-01 -7.17244267e-01 7.15486109e-01
9.29280519e-01 -6.75235614e-02 -3.63709003e-01 7.02726901e-01
5.91736555e-01 -1.99304029e-01 7.85172522e-01 -1.11995399e+00
-6.80810213e-01 8.46481323e-01 -5.76829791e-01 -1.72003686e-01
3.63099515e-01 -3.77702415e-02 7.37835586e-01 -1.15410912e+00
9.86096084e-01 1.00659621e+00 7.10217595e-01 5.10354996e-01
-1.21090364e+00 -4.70369518e-01 -3.66714686e-01 -4.50829007e-02
-1.44591141e+00 -5.30992270e-01 3.09823543e-01 -3.62587065e-01
1.21108222e+00 4.50914055e-01 4.19226587e-01 1.68110943e+00
-6.75853118e-02 1.20857143e+00 1.27308834e+00 -8.86938512e-01
7.84685463e-03 3.86465043e-01 2.44032204e-01 5.38338482e-01
3.81215185e-01 -2.53460765e-01 -4.57915515e-01 1.85175717e-01
3.57283324e-01 1.30316662e-02 -7.36044273e-02 -1.92927554e-01
-1.71024370e+00 6.17129743e-01 -2.87022799e-01 6.40006423e-01
3.00803874e-02 -8.04755613e-02 7.61839032e-01 5.75110078e-01
4.31624919e-01 3.12314689e-01 -3.99974704e-01 -1.72721535e-01
-1.39495552e+00 2.27820128e-01 1.04854393e+00 1.03646040e+00
4.22210604e-01 3.34794611e-01 -1.10156655e-01 1.21817505e+00
-5.68436794e-02 8.19639683e-01 1.24291837e+00 -2.11144581e-01
7.92738497e-01 8.07056129e-01 -1.43477812e-01 -8.73505831e-01
-3.24090004e-01 -1.12671547e-01 -1.05308807e+00 7.67314658e-02
7.87385821e-01 1.32586911e-01 -1.39957118e+00 6.62801325e-01
-3.51968199e-01 -3.08019757e-01 1.12067692e-01 5.78215599e-01
7.75883019e-01 8.44908237e-01 -4.63519931e-01 1.38653308e-01
1.07038820e+00 -1.15907609e+00 -7.19529569e-01 -2.48854250e-01
9.19528008e-01 -1.10355592e+00 1.36695981e+00 8.71121228e-01
-4.18501139e-01 -3.42895180e-01 -1.05487502e+00 7.33293295e-02
-8.70343447e-01 4.91393030e-01 4.21794057e-01 1.20183718e+00
-8.20508540e-01 2.64205694e-01 -6.23856843e-01 -1.13445354e+00
3.92132223e-01 1.22169323e-01 -4.57675099e-01 -1.31191030e-01
-8.13743532e-01 7.75391400e-01 7.43023038e-01 -2.04387501e-01
-6.93252146e-01 1.53130531e-01 -4.75572318e-01 -2.69506007e-01
4.76994574e-01 -2.12918967e-01 1.04787838e+00 -1.27555561e+00
-1.20212793e+00 1.27799928e+00 1.35524794e-01 -7.83687711e-01
9.83207941e-01 8.15367419e-03 -8.25001299e-01 -1.55147865e-01
3.05055529e-02 8.96603763e-01 1.22374833e+00 -1.03500009e+00
-3.94451171e-01 -4.54432905e-01 -7.14736462e-01 1.03643453e-02
-4.06036705e-01 1.05544448e-01 -3.57275158e-01 -1.13466990e+00
1.02447875e-01 -8.91494274e-01 8.15414265e-02 -1.05957888e-01
-7.43526697e-01 4.81299721e-02 1.11934710e+00 -3.78482789e-01
9.90197659e-01 -1.97104156e+00 -1.81309417e-01 -1.14612289e-01
-8.46499056e-02 1.58814400e-01 -1.36180460e-01 5.22617817e-01
-3.23808551e-01 2.45819554e-01 -2.26021364e-01 -4.52663422e-01
7.90306255e-02 3.82442176e-01 -7.38593638e-01 3.15336347e-01
2.38900911e-02 7.98028827e-01 -2.34201774e-01 -5.34809768e-01
3.37129414e-01 1.70113131e-01 -1.71826974e-01 -2.60833621e-01
-5.09739578e-01 -4.82714251e-02 -3.88701051e-01 7.31405318e-01
5.79050660e-01 -2.40918964e-01 -2.95697242e-01 1.44360393e-01
4.96132951e-03 -1.36835113e-01 -1.29670417e+00 1.49776363e+00
-1.07701763e-03 1.33039558e+00 -5.50851524e-01 -1.03426719e+00
1.09292614e+00 3.33882689e-01 4.72539753e-01 -7.79082239e-01
1.81982771e-01 2.80627817e-01 -2.67589808e-01 -6.13210976e-01
9.75715160e-01 -4.53745527e-03 -2.05496494e-02 6.71523511e-01
4.16738205e-02 -4.34928656e-01 4.28011060e-01 1.50123522e-01
8.49243164e-01 3.11875135e-01 -8.86401832e-02 -1.11439340e-01
4.84077543e-01 7.05654860e-01 -1.14262573e-01 9.88195062e-01
1.05947398e-01 1.17267346e+00 3.80317509e-01 -8.52839291e-01
-1.54806852e+00 -5.53369820e-01 -4.61079061e-01 9.74728167e-01
-4.27260101e-01 -4.59523410e-01 -9.70948100e-01 -6.45804882e-01
-6.48715049e-02 7.48405695e-01 -8.72264981e-01 4.04484749e-01
-3.08812857e-01 -1.02399850e+00 1.29774141e+00 2.30958417e-01
7.64208794e-01 -1.18675244e+00 -3.07782739e-01 -2.22787485e-01
3.66027802e-02 -1.35660744e+00 -2.60547638e-01 3.51570576e-01
-8.88613880e-01 -8.65599930e-01 -9.61650789e-01 -7.61767864e-01
5.39662302e-01 9.25742164e-02 9.91870224e-01 -1.27966791e-01
-6.58197999e-01 1.88469693e-01 -8.02023411e-01 -6.79595292e-01
-7.44744420e-01 2.13631436e-01 3.17032002e-02 1.18540972e-01
6.29300356e-01 3.55651453e-02 1.38598189e-01 5.92146277e-01
-1.34552467e+00 4.57280248e-01 5.31592786e-01 1.09399009e+00
3.51138115e-01 1.15939014e-01 4.29987311e-01 -1.13205492e+00
4.18259352e-01 3.32978629e-02 -4.53444004e-01 2.56411493e-01
-7.79452920e-01 -1.08380895e-02 6.93438351e-01 -5.04740357e-01
-1.01541924e+00 3.22981209e-01 9.85442996e-02 -2.08871156e-01
-6.46564543e-01 3.19382161e-01 6.44593090e-02 3.02092254e-01
1.14512682e+00 7.29443908e-01 -9.92429852e-02 -3.61626834e-01
1.23726688e-01 1.10791290e+00 2.58089602e-01 -3.13943386e-01
5.06212950e-01 4.32941407e-01 -4.17489856e-01 -1.30167830e+00
-7.24228382e-01 -2.33926162e-01 -7.54088104e-01 6.64850026e-02
6.99747682e-01 -8.03941071e-01 -1.42449036e-01 1.02769685e+00
-7.92871594e-01 -4.89808470e-01 -4.04565066e-01 3.94773543e-01
-6.34546459e-01 6.00795329e-01 -3.01182121e-01 -6.95718408e-01
-2.26403251e-01 -1.05500615e+00 1.25551450e+00 -1.27022475e-01
1.51816541e-02 -9.44121659e-01 -9.67064034e-03 5.95296443e-01
3.83028209e-01 1.06396519e-01 5.10932267e-01 -1.04229462e+00
-3.59433860e-01 -5.83322108e-01 -2.84550607e-01 3.66607994e-01
1.24104545e-01 2.43865296e-01 -1.24352074e+00 -5.52982930e-03
-1.03321493e-01 -6.27338052e-01 1.11631954e+00 7.53670260e-02
1.07382488e+00 -1.16118863e-01 -3.88605803e-01 3.02008122e-01
1.06205845e+00 -1.11010019e-02 1.00205982e+00 5.19828439e-01
6.93089247e-01 4.99477565e-01 5.32540560e-01 3.73588234e-01
1.64268930e-02 6.46945596e-01 -4.46669050e-02 -1.82920769e-02
-1.27970472e-01 -1.54617429e-01 4.25789624e-01 8.31562161e-01
2.99092114e-01 -7.61682808e-01 -1.36126781e+00 2.60541171e-01
-1.64392638e+00 -1.06027520e+00 -4.75789368e-01 2.00869155e+00
6.90488040e-01 4.06475276e-01 1.39250597e-02 6.06941581e-01
7.60913610e-01 7.68426359e-02 -3.87650222e-01 -2.82640427e-01
-7.80441642e-01 -8.69046301e-02 5.87528765e-01 1.00528924e-02
-1.21099472e+00 1.10365021e+00 6.85394764e+00 1.14771032e+00
-1.26612020e+00 -2.32528239e-01 6.67585611e-01 7.86146242e-03
2.35787362e-01 -2.22004548e-01 -8.22561383e-01 3.68851304e-01
1.10642910e+00 -8.11606571e-02 2.28962108e-01 8.26685667e-01
-1.07905157e-01 -3.01444679e-01 -1.19291997e+00 1.36391401e+00
6.74633741e-01 -1.32326996e+00 5.79477012e-01 -1.40936794e-02
7.60844290e-01 5.19531012e-01 -9.59657133e-02 2.95891136e-01
3.49559598e-02 -1.35112691e+00 7.65713513e-01 4.41931844e-01
1.02895927e+00 -3.39539170e-01 5.63392043e-01 5.83439350e-01
-7.07676589e-01 1.86156079e-01 -6.05554283e-01 2.24958718e-01
-4.32060003e-01 4.51075137e-01 -1.20988309e+00 4.41919714e-01
5.44017255e-01 1.02642643e+00 -1.21129966e+00 7.95651793e-01
4.05057102e-01 7.98238158e-01 -4.23802376e-01 -4.16606188e-01
2.36528337e-01 -1.79051906e-01 3.44042927e-01 1.51864159e+00
2.06645295e-01 -5.14285386e-01 -1.42524922e-02 5.61902225e-01
-5.70199564e-02 4.75351006e-01 -1.00966847e+00 -5.75450242e-01
-1.21576719e-01 1.09545112e+00 -1.14029443e+00 -6.57536328e-01
-4.74233180e-01 1.25251412e+00 -3.41811121e-01 1.10479273e-01
-6.26900733e-01 -5.85510731e-01 -1.95327774e-01 -1.57502815e-02
1.24938890e-01 -3.84967737e-02 -4.81327951e-01 -1.63683176e+00
5.54485023e-02 -1.25914812e+00 3.76426518e-01 -1.25639296e+00
-1.12150431e+00 8.05828154e-01 -3.70889008e-01 -1.38668418e+00
-2.13071838e-01 -1.03175795e+00 -1.83926746e-01 5.08345068e-01
-7.18262196e-01 -1.31639421e+00 -3.85712922e-01 5.43235004e-01
1.23030865e+00 -7.66482174e-01 1.05696130e+00 3.91725153e-01
-6.81399643e-01 6.07607067e-01 6.13873839e-01 5.67641139e-01
1.05809450e+00 -1.07720315e+00 7.86532402e-01 4.81868625e-01
7.36006081e-01 3.32412422e-02 7.28708982e-01 -7.08171308e-01
-1.57165253e+00 -8.27654839e-01 6.61492825e-01 -9.64477479e-01
7.86008835e-01 -6.80735111e-01 -1.09202957e+00 8.17415655e-01
2.38272026e-01 -1.88201025e-01 3.83683085e-01 -3.75737906e-01
-4.71773207e-01 1.37124732e-01 -1.08480000e+00 5.52453578e-01
6.74543500e-01 -2.99598664e-01 -6.95617855e-01 5.87740481e-01
2.38428131e-01 -2.78874964e-01 -4.80565488e-01 6.22295961e-02
6.71858191e-01 -1.06628704e+00 4.92018431e-01 -5.31253159e-01
4.45626676e-01 -1.74748719e-01 -3.76387209e-01 -8.84683609e-01
4.15380925e-01 -2.90927976e-01 4.33089525e-01 1.18470109e+00
6.19780183e-01 -7.56505132e-01 1.10926545e+00 2.57688433e-01
5.40224835e-02 -1.70517385e-01 -7.74337590e-01 -9.16355371e-01
1.13219164e-01 -6.68369234e-01 2.59990573e-01 1.45377624e+00
-8.01233500e-02 5.15884876e-01 -5.87137997e-01 -4.59414482e-01
3.86949718e-01 1.97817050e-02 1.06279123e+00 -1.40927732e+00
-1.30665287e-01 -5.12460291e-01 -6.84279203e-01 -7.71825194e-01
2.89304387e-02 -1.12707102e+00 -1.03102885e-01 -1.32691777e+00
2.94789046e-01 -3.43515545e-01 3.80681455e-01 6.59372091e-01
3.51893425e-01 5.18504560e-01 2.25496925e-02 3.18448424e-01
-4.15924102e-01 3.37168813e-01 1.12448502e+00 -4.78433132e-01
2.00900242e-01 -2.71327525e-01 -1.32149845e-01 6.73820078e-01
7.69342184e-01 -5.57893038e-01 -1.41425669e-01 -4.46627021e-01
3.18666637e-01 -1.87211171e-01 8.76856744e-02 -1.22764504e+00
2.94344336e-01 8.81052315e-02 7.14718342e-01 -9.03073668e-01
2.13737741e-01 -9.60171878e-01 2.25300550e-01 3.18075031e-01
-4.75750268e-01 -1.72502115e-01 1.83979228e-01 3.29030484e-01
-2.89496422e-01 -3.31204921e-01 6.80035710e-01 -9.83150955e-03
-5.14333189e-01 -3.45535502e-02 -7.52745330e-01 -1.39823770e-02
7.96031237e-01 -7.84841537e-01 -4.98302072e-01 -3.30620438e-01
-6.13102138e-01 -9.08688903e-02 8.82096171e-01 8.10076118e-01
5.44760764e-01 -8.76622319e-01 -8.69342029e-01 4.30297673e-01
4.97670680e-01 -1.93727493e-01 -3.01692337e-01 7.51274168e-01
-8.19078982e-01 6.37937367e-01 -4.13683891e-01 -8.87514055e-01
-1.17801487e+00 5.11520445e-01 4.78121966e-01 -1.53368711e-01
-8.40669215e-01 1.35439098e-01 -1.97249576e-01 -6.51753783e-01
3.59636366e-01 -3.16921145e-01 5.20564709e-03 1.44535229e-01
6.20342612e-01 3.80569130e-01 2.67244488e-01 -6.88778877e-01
-3.74123380e-02 6.43099308e-01 -2.85027146e-01 -1.47231042e-01
1.25460279e+00 6.44013658e-02 5.16555347e-02 8.40965509e-01
7.35462129e-01 1.15327556e-02 -6.21072054e-01 -2.11161271e-01
4.50687796e-01 -5.19094586e-01 -2.40565702e-01 -9.77925420e-01
-8.34242523e-01 9.07197058e-01 7.59386599e-01 5.36615551e-01
9.70973730e-01 -2.28542879e-01 2.72691607e-01 1.19706154e+00
3.97338033e-01 -1.27169168e+00 1.01787463e-01 8.64374697e-01
8.86432409e-01 -1.48886061e+00 1.45206496e-01 -2.28697553e-01
-9.20876801e-01 1.36658835e+00 4.53927308e-01 1.02318250e-01
2.71778017e-01 6.01916134e-01 3.05910289e-01 -8.08173493e-02
-8.94045889e-01 -1.37181118e-01 -3.69009115e-02 4.46015537e-01
7.33450949e-01 -1.31570727e-01 5.40965796e-02 5.89169078e-02
-1.31563127e-01 1.46067321e-01 9.38636124e-01 8.20801139e-01
-5.36858737e-01 -1.39559972e+00 -6.23895764e-01 8.52946043e-01
-5.90695381e-01 -8.67409036e-02 -9.31717932e-01 1.02092803e+00
-2.45542750e-01 6.13185942e-01 8.07714686e-02 -3.83339763e-01
3.49620879e-01 5.93975127e-01 4.64666337e-01 -6.39492095e-01
-5.98414660e-01 5.97960241e-02 2.12424681e-01 -3.10114585e-02
-3.12168032e-01 -7.19809413e-01 -1.18493354e+00 -2.67275125e-01
-5.36275268e-01 -2.36152075e-02 9.75419283e-01 6.58203602e-01
8.49287137e-02 7.35354722e-02 4.47661012e-01 -7.89136469e-01
-2.57804632e-01 -1.48203146e+00 -7.99840569e-01 4.16173339e-01
-8.32888111e-02 -2.95698375e-01 -2.56898314e-01 7.72113562e-01] | [11.84131908416748, 2.3011057376861572] |
5e3ab08b-2439-4ca2-bd64-2882da24cda1 | diffusion-models-beat-gans-on-image-synthesis | 2105.05233 | null | https://arxiv.org/abs/2105.05233v4 | https://arxiv.org/pdf/2105.05233v4.pdf | Diffusion Models Beat GANs on Image Synthesis | We show that diffusion models can achieve image sample quality superior to the current state-of-the-art generative models. We achieve this on unconditional image synthesis by finding a better architecture through a series of ablations. For conditional image synthesis, we further improve sample quality with classifier guidance: a simple, compute-efficient method for trading off diversity for fidelity using gradients from a classifier. We achieve an FID of 2.97 on ImageNet 128$\times$128, 4.59 on ImageNet 256$\times$256, and 7.72 on ImageNet 512$\times$512, and we match BigGAN-deep even with as few as 25 forward passes per sample, all while maintaining better coverage of the distribution. Finally, we find that classifier guidance combines well with upsampling diffusion models, further improving FID to 3.94 on ImageNet 256$\times$256 and 3.85 on ImageNet 512$\times$512. We release our code at https://github.com/openai/guided-diffusion | ['Alex Nichol', 'Prafulla Dhariwal'] | 2021-05-11 | null | http://proceedings.neurips.cc/paper/2021/hash/49ad23d1ec9fa4bd8d77d02681df5cfa-Abstract.html | http://proceedings.neurips.cc/paper/2021/file/49ad23d1ec9fa4bd8d77d02681df5cfa-Paper.pdf | neurips-2021-12 | ['conditional-image-generation'] | ['computer-vision'] | [ 1.00938253e-01 2.59187669e-01 -2.60372400e-01 -4.35471386e-02
-1.03242517e+00 -5.62320292e-01 8.07077885e-01 -4.14744049e-01
-4.83313560e-01 6.17531836e-01 1.19433895e-01 -3.75596821e-01
1.82807580e-01 -8.90920401e-01 -7.98432648e-01 -6.63896263e-01
-3.05453151e-01 4.23265733e-02 -1.36152506e-01 -8.86683166e-02
4.18702364e-02 1.02654830e-01 -1.21229804e+00 2.79469222e-01
8.18624973e-01 9.84736681e-01 9.50199440e-02 1.02506089e+00
3.64495665e-01 7.85540402e-01 -4.88929749e-01 -5.32597065e-01
7.16644585e-01 -8.14649522e-01 -5.03732562e-01 1.05886541e-01
6.27980113e-01 -7.03778327e-01 -8.82489085e-01 1.20815504e+00
4.28313017e-01 -5.14106229e-02 6.54939890e-01 -1.11021006e+00
-9.53568518e-01 9.77137089e-01 -7.99470127e-01 1.55792341e-01
-8.14273879e-02 9.16973829e-01 1.04868305e+00 -7.38864779e-01
6.83493018e-01 1.20159781e+00 5.29349566e-01 7.32769370e-01
-1.33836889e+00 -1.16975319e+00 1.21086948e-02 -1.47614777e-01
-1.35039389e+00 -7.39928961e-01 3.68477494e-01 -4.30508018e-01
8.05354536e-01 3.65272490e-03 8.79930615e-01 1.06450057e+00
1.23021618e-01 6.52329624e-01 1.24420607e+00 -2.11317703e-01
1.46262348e-01 -1.79936588e-01 -2.38083586e-01 1.04739535e+00
1.60573110e-01 4.93016809e-01 -4.66764390e-01 -4.73318761e-03
1.08345008e+00 -1.90230712e-01 -2.48060644e-01 1.91813439e-01
-1.11878860e+00 1.04361975e+00 8.10718536e-01 1.12492420e-01
-3.16263050e-01 9.07495558e-01 -2.58659869e-02 4.28017676e-01
3.82602513e-01 4.56507742e-01 -1.65551379e-01 -1.37060955e-01
-1.34111702e+00 3.59724194e-01 4.96241719e-01 1.19751537e+00
1.16523004e+00 5.77770650e-01 -2.06050396e-01 7.37790167e-01
2.18223110e-01 9.70202982e-01 1.80696309e-01 -1.49282098e+00
1.42336354e-01 4.14647721e-02 -2.00325236e-01 -6.16495252e-01
2.48257760e-02 -7.00256884e-01 -1.15136230e+00 3.55878413e-01
4.64409709e-01 -4.39861566e-01 -1.38892674e+00 2.06319976e+00
-2.24887013e-01 1.87501177e-01 -3.67812961e-02 5.22431910e-01
4.92516667e-01 6.57390118e-01 9.59349126e-02 1.71241194e-01
1.43325269e+00 -1.08637345e+00 -1.46732464e-01 -6.17288351e-01
4.73891079e-01 -7.93508291e-01 1.14058590e+00 3.17786902e-01
-1.30037773e+00 -4.88972068e-01 -1.19342172e+00 1.46807596e-01
1.37308374e-01 2.57584542e-01 6.84231699e-01 9.29037988e-01
-1.42619169e+00 6.50720239e-01 -9.34493184e-01 9.53499824e-02
8.81975830e-01 3.36072564e-01 -1.22040570e-01 -4.77412969e-01
-9.61082041e-01 5.58374047e-01 -3.55597660e-02 -4.00327533e-01
-1.52004695e+00 -9.83576596e-01 -8.62362742e-01 6.24186918e-03
-4.28578518e-02 -9.00963366e-01 1.16063178e+00 -1.04713857e+00
-1.40996659e+00 8.08780551e-01 -1.63750857e-01 -7.94777155e-01
5.21998942e-01 3.98143791e-02 -2.56501555e-01 1.40177980e-01
2.14064285e-01 1.39466381e+00 1.10052621e+00 -1.15484297e+00
-5.31423986e-01 -1.98624521e-01 1.02815785e-01 -9.05677974e-02
-2.91355342e-01 -4.59261358e-01 -5.09433150e-01 -7.92850435e-01
-6.15886832e-03 -1.28793812e+00 -4.13257420e-01 -7.22474535e-04
-4.80192423e-01 2.92334169e-01 5.19597769e-01 -7.47072458e-01
9.65996146e-01 -2.11601496e+00 -3.06878462e-02 2.24766284e-01
6.84938133e-01 1.64315104e-01 -4.74377751e-01 3.11516207e-02
1.93311483e-01 4.04994905e-01 -4.97739106e-01 -3.37394655e-01
-1.19710237e-01 -6.09009080e-02 -1.87598616e-01 3.57486010e-01
2.86452800e-01 1.08784640e+00 -6.20106280e-01 -1.15250304e-01
1.34547548e-02 5.47431350e-01 -9.91011143e-01 -2.14902878e-01
-2.23003119e-01 3.57406884e-01 -2.20423236e-01 5.02476931e-01
8.48080873e-01 -5.11161864e-01 4.68708947e-02 1.46053387e-02
2.88146645e-01 5.25087416e-02 -7.76042819e-01 1.92710090e+00
-5.45054376e-01 9.81525838e-01 2.20669076e-01 -5.46132684e-01
8.48847866e-01 1.94368083e-02 3.24149162e-01 -7.06225276e-01
6.07174411e-02 3.56678993e-01 4.30416614e-01 1.03410996e-01
4.07264769e-01 -1.67983413e-01 4.77614366e-02 8.21704566e-01
3.97792399e-01 -5.37678719e-01 2.04605699e-01 4.63833690e-01
1.35533762e+00 -1.56320661e-01 -1.95334703e-01 -4.89516646e-01
-3.28098945e-02 -2.06940651e-01 2.89534062e-01 9.84572291e-01
-9.61316079e-02 6.19341612e-01 4.55750465e-01 -9.29415822e-02
-1.52211225e+00 -1.03777170e+00 8.80961046e-02 7.73670733e-01
-1.39506131e-01 -6.19761050e-01 -1.08723354e+00 -6.46775126e-01
-9.98695716e-02 6.82773590e-01 -6.31581485e-01 -1.64927006e-01
-4.39368576e-01 -7.57506073e-01 9.07787025e-01 2.38465384e-01
1.10819864e+00 -7.88439810e-01 -3.14699322e-01 -1.05509862e-01
5.36472276e-02 -9.37887967e-01 -6.68538392e-01 -2.39419304e-02
-8.48279893e-01 -6.97048664e-01 -1.12863362e+00 -5.44570625e-01
6.28192365e-01 1.72807515e-01 1.30659878e+00 2.78280795e-01
-2.33643189e-01 1.44826353e-01 -1.36678413e-01 -9.49122831e-02
-6.23899400e-01 3.90852958e-01 -1.66423246e-01 -3.95373017e-01
-1.00617163e-01 -7.91512251e-01 -1.05884230e+00 2.23563567e-01
-8.96967411e-01 1.00694075e-01 8.35855961e-01 8.84619832e-01
4.25854921e-01 -4.58550379e-02 3.88122499e-01 -9.01485562e-01
5.06930470e-01 -4.12414491e-01 -5.14409244e-01 -2.00567365e-01
-8.76028895e-01 2.71723628e-01 4.56449926e-01 -4.48072582e-01
-7.77088106e-01 -1.81644976e-01 -3.45112771e-01 -6.08012676e-01
1.80498123e-01 1.36289716e-01 1.90793768e-01 -1.77423373e-01
1.16516089e+00 4.17162895e-01 2.42579862e-01 2.74238829e-02
7.20805407e-01 3.62744212e-01 3.54545385e-01 -4.74966407e-01
8.79520297e-01 5.02468765e-01 -2.86252886e-01 -7.56145537e-01
-4.03982580e-01 2.93085814e-01 3.53006534e-02 3.39279398e-02
5.70478320e-01 -1.35340130e+00 -4.28573757e-01 5.88357031e-01
-5.17926455e-01 -1.04304063e+00 -4.53708053e-01 3.80245268e-01
-6.04982436e-01 -2.41082050e-02 -8.13697815e-01 -3.90174359e-01
-4.87790257e-01 -1.43274546e+00 7.55451083e-01 5.36043718e-02
-1.50825322e-01 -8.28358412e-01 -2.38391533e-01 1.39413819e-01
8.20103407e-01 -4.98152487e-02 7.19285488e-01 -2.37830766e-02
-9.55488443e-01 1.92582849e-02 -4.54573244e-01 6.04986906e-01
8.77668802e-03 -2.12755445e-02 -8.96128595e-01 -6.24919593e-01
-1.89443946e-01 -4.61595595e-01 1.40524185e+00 5.62026680e-01
9.90476131e-01 -3.52456987e-01 -2.48263463e-01 1.02143681e+00
1.35837853e+00 1.02044865e-01 1.02386165e+00 -2.32761726e-02
5.21000624e-01 4.14203443e-02 7.93885812e-02 4.98227954e-01
2.30443984e-01 1.23408258e-01 3.00051808e-01 -1.13261156e-01
-8.36416960e-01 -4.39025342e-01 3.43501449e-01 6.37944877e-01
6.61761090e-02 -2.32821077e-01 -8.54972899e-01 5.28483868e-01
-1.17382979e+00 -8.88728678e-01 3.16743791e-01 1.90840006e+00
9.84798729e-01 1.99010029e-01 4.59093712e-02 -1.82321414e-01
5.67897856e-01 3.89768451e-01 -7.77344942e-01 8.35548863e-02
-1.46070525e-01 4.99004751e-01 8.45022678e-01 7.08090544e-01
-8.28901827e-01 1.19369984e+00 6.38255930e+00 1.14915788e+00
-1.21473181e+00 1.16171740e-01 1.33814216e+00 -2.76056886e-01
-7.15865672e-01 1.72454357e-01 -7.59116411e-01 6.30415857e-01
8.83782685e-01 -4.89875376e-02 7.57865369e-01 8.24789524e-01
-2.19467267e-01 -2.13864014e-01 -9.28220391e-01 1.12079597e+00
-3.28733362e-02 -1.71419883e+00 6.26356676e-02 4.85455334e-01
1.28663266e+00 2.79832721e-01 7.01995432e-01 2.24665284e-01
8.82608056e-01 -1.29086363e+00 7.94467866e-01 8.41924623e-02
1.38816094e+00 -7.09429979e-01 1.28772244e-01 1.78228199e-01
-9.72062230e-01 -2.36170124e-02 -4.06500041e-01 1.90260828e-01
-1.67692214e-01 7.96945214e-01 -5.40416300e-01 -7.04648793e-02
6.22828007e-01 7.32399225e-01 -5.46692133e-01 5.88512897e-01
-4.86733615e-01 6.50966644e-01 -4.94339108e-01 1.17259748e-01
3.57675403e-01 1.34846658e-01 3.94442588e-01 1.21139634e+00
7.73467898e-01 9.68064740e-02 -2.13689744e-01 1.17136681e+00
-5.71130693e-01 -3.95116538e-01 -7.01874912e-01 -1.79563779e-02
6.16364956e-01 1.01021123e+00 -6.61959171e-01 -5.21213591e-01
1.57283414e-02 1.00858665e+00 1.19487077e-01 4.39124703e-01
-9.19532120e-01 -3.52117211e-01 8.10468554e-01 1.15998097e-01
4.44737762e-01 -4.06511366e-01 -3.10178697e-01 -1.30099714e+00
-3.62426698e-01 -1.24114990e+00 1.26168758e-01 -6.37333333e-01
-1.11926866e+00 7.76600301e-01 -1.72724217e-01 -9.97439802e-01
-3.23573917e-01 -3.54815692e-01 -3.76321584e-01 8.50106537e-01
-1.21667719e+00 -1.13739824e+00 -2.95742631e-01 5.83077133e-01
5.01028061e-01 -4.00609970e-01 6.81331694e-01 2.90910691e-01
-3.04537743e-01 9.84915435e-01 -3.42963003e-02 1.20789178e-01
4.92699027e-01 -9.17743385e-01 8.79885197e-01 9.73256767e-01
3.26312304e-01 5.69985628e-01 5.65293193e-01 -5.06391823e-01
-1.34792578e+00 -1.02400815e+00 2.86700040e-01 -7.22735301e-02
3.74636918e-01 -3.31545115e-01 -4.02248621e-01 8.69187534e-01
6.70674086e-01 1.08453445e-02 3.95618230e-01 -1.20388269e-01
-7.19839096e-01 -7.21991733e-02 -1.30535293e+00 8.38241458e-01
1.26207447e+00 -5.11567235e-01 4.06153530e-01 1.29826054e-01
5.91251910e-01 -4.33353990e-01 -8.81985545e-01 9.28179249e-02
5.08900106e-01 -1.06902421e+00 9.19654846e-01 1.08992182e-01
8.03491533e-01 6.26126081e-02 -3.11009645e-01 -1.47066438e+00
-4.28241402e-01 -1.08157766e+00 -1.04251336e-02 8.49398136e-01
7.61313379e-01 -8.30363572e-01 1.16261613e+00 3.74097466e-01
3.98451239e-02 -5.98076940e-01 -6.98518276e-01 -6.52936280e-01
5.90688407e-01 -5.19300938e-01 6.52291894e-01 6.83232486e-01
-4.01365817e-01 1.43406183e-01 -4.85887885e-01 -2.04535425e-01
8.98265004e-01 -2.11788923e-01 8.14015269e-01 -4.12025422e-01
-6.90332711e-01 -8.39550674e-01 -1.44799277e-01 -1.57358599e+00
-1.06543088e-02 -8.86083364e-01 -1.24131382e-01 -1.16037822e+00
3.00053596e-01 -8.99896085e-01 1.81086250e-02 4.74742472e-01
3.18088848e-03 8.30031216e-01 4.56065685e-01 4.19882536e-01
-1.05762571e-01 4.92661655e-01 1.61758125e+00 -3.94088119e-01
5.33221290e-03 -3.58277649e-01 -1.11779356e+00 5.08763731e-01
1.16345227e+00 -3.99418682e-01 -6.16571367e-01 -7.72120416e-01
1.51427805e-01 2.22137589e-02 3.32562953e-01 -1.31335580e+00
3.77504863e-02 1.39222875e-01 5.68035603e-01 -5.98906726e-02
5.45291126e-01 -2.32751504e-01 2.86875248e-01 8.49933326e-01
-5.53636491e-01 -4.85801622e-02 2.09259376e-01 3.54464024e-01
-4.80163060e-02 -6.83489740e-02 1.14556682e+00 -3.68171573e-01
-5.06086349e-01 6.52349710e-01 -3.03543717e-01 2.54816502e-01
8.91608059e-01 9.56696495e-02 -5.38260341e-01 -8.70812476e-01
-6.84180081e-01 -3.01577356e-02 6.00161850e-01 1.56155899e-01
7.93831289e-01 -1.37549257e+00 -9.68678951e-01 5.91187179e-01
-2.55724788e-01 -2.03240812e-01 2.79957622e-01 5.40483773e-01
-7.59589911e-01 1.39519572e-01 -2.62456536e-01 -6.12611055e-01
-8.82836103e-01 7.57338852e-02 4.50273722e-01 -1.15090519e-01
-4.29459691e-01 1.34000993e+00 2.99255580e-01 -1.46955863e-01
-8.33564475e-02 -1.28811821e-01 5.93159854e-01 -3.23871434e-01
4.68298942e-01 3.18304688e-01 -3.00163001e-01 -2.70207465e-01
-6.30646870e-02 5.90749979e-01 -1.25402793e-01 -7.48583376e-01
1.07825565e+00 2.70510651e-02 5.84480502e-02 -2.79104650e-01
1.30730951e+00 7.00690597e-02 -1.80991578e+00 -2.01898292e-01
-7.67235875e-01 -5.51721811e-01 2.29958355e-01 -7.22753525e-01
-1.60548079e+00 9.49659288e-01 5.58247626e-01 8.06560963e-02
1.13334608e+00 -2.25904249e-02 8.75423372e-01 1.06102854e-01
4.72330093e-01 -5.86630166e-01 3.82113129e-01 5.02498686e-01
7.29409099e-01 -1.11493301e+00 1.49700329e-01 -8.52091610e-02
-5.77447236e-01 6.48691416e-01 3.33341062e-01 -6.59235239e-01
8.98836136e-01 5.58575869e-01 -9.06042680e-02 -1.12011388e-01
-8.55013788e-01 -6.79715723e-02 -4.70364951e-02 6.02528334e-01
3.82026464e-01 1.41045079e-01 1.11419238e-01 1.05331987e-01
-6.25693858e-01 -1.20009810e-01 4.66521084e-01 5.75718999e-01
-3.69339526e-01 -1.12312794e+00 -2.87812389e-02 5.97818136e-01
-5.43617547e-01 -6.18551731e-01 1.61764417e-02 7.91042030e-01
4.09103222e-02 9.34838057e-01 1.54664338e-01 -6.29303396e-01
-3.10106039e-01 -1.60245195e-01 7.99526215e-01 -3.27917844e-01
-4.66499835e-01 2.65990883e-01 -4.38900897e-03 -5.35547256e-01
-1.43796399e-01 -4.94450688e-01 -1.00179529e+00 -9.55696404e-01
-1.80502430e-01 -1.27459392e-01 6.93945408e-01 3.52055311e-01
6.31524384e-01 4.22207654e-01 6.74824774e-01 -7.32764125e-01
-4.86359358e-01 -9.01331604e-01 -5.28885186e-01 1.08371116e-01
2.98211932e-01 -1.11263752e-01 -4.14642930e-01 2.10639462e-01] | [11.464679718017578, -0.3924142122268677] |
327fe971-b9e6-4c9f-9ba6-ee59fb97d0e3 | influencer-backdoor-attack-on-semantic | 2303.12054 | null | https://arxiv.org/abs/2303.12054v2 | https://arxiv.org/pdf/2303.12054v2.pdf | Influencer Backdoor Attack on Semantic Segmentation | When a small number of poisoned samples are injected into the training dataset of a deep neural network, the network can be induced to exhibit malicious behavior during inferences, which poses potential threats to real-world applications. While they have been intensively studied in classification, backdoor attacks on semantic segmentation have been largely overlooked. Unlike classification, semantic segmentation aims to classify every pixel within a given image. In this work, we explore backdoor attacks on segmentation models to misclassify all pixels of a victim class by injecting a specific trigger on non-victim pixels during inferences, which is dubbed Influencer Backdoor Attack (IBA). IBA is expected to maintain the classification accuracy of non-victim pixels and misleads classifications of all victim pixels in every single inference. Specifically, we consider two types of IBA scenarios, i.e., 1) Free-position IBA: the trigger can be positioned freely except for pixels of the victim class, and 2) Long-distance IBA: the trigger can only be positioned somewhere far from victim pixels, given the possible practical constraint. Based on the context aggregation ability of segmentation models, we propose techniques to improve IBA for the scenarios. Concretely, for free-position IBA, we propose a simple, yet effective Nearest Neighbor trigger injection strategy for poisoned sample creation. For long-distance IBA, we propose a novel Pixel Random Labeling strategy. Our extensive experiments reveal that current segmentation models do suffer from backdoor attacks, and verify that our proposed techniques can further increase attack performance. | ['Hengshuang Zhao', 'Philip Torr', 'Jindong Gu', 'Haoheng Lan'] | 2023-03-21 | null | null | null | null | ['backdoor-attack'] | ['adversarial'] | [ 6.30631626e-01 2.20388189e-01 -2.45762855e-01 -1.44503891e-01
-5.98617494e-01 -1.20211363e+00 5.42566955e-01 -1.39482850e-02
-3.80531073e-01 4.85468686e-01 -5.77469528e-01 -6.89119279e-01
1.15634635e-01 -1.27139044e+00 -1.08242226e+00 -1.00084436e+00
2.15807542e-01 3.23337652e-02 6.84963763e-01 2.43024379e-01
2.39094198e-01 7.38557458e-01 -1.12424493e+00 2.11750329e-01
7.62347341e-01 8.68402719e-01 -1.67359293e-01 2.50371695e-01
-7.30011165e-02 5.89171231e-01 -1.23282564e+00 -5.91827750e-01
4.93063569e-01 -3.49166304e-01 -6.87022984e-01 -7.02148303e-02
2.99963623e-01 -5.01374364e-01 -4.03247297e-01 1.69094610e+00
1.00698046e-01 -2.80294120e-01 4.00454730e-01 -1.59031308e+00
-3.96896929e-01 1.02457976e+00 -7.64900804e-01 1.22421823e-01
-1.45785287e-02 3.57888490e-01 6.04977429e-01 -3.24341208e-01
3.68857175e-01 1.09205651e+00 4.98677433e-01 7.65206337e-01
-8.79007220e-01 -1.23200691e+00 3.70387495e-01 3.08464952e-02
-1.44514644e+00 1.73878972e-04 9.52755749e-01 -6.42820001e-02
1.94899395e-01 6.09917819e-01 5.07393003e-01 1.41814375e+00
1.83981478e-01 9.39471900e-01 1.22142065e+00 -1.60225481e-02
5.86401105e-01 1.30570114e-01 3.59215736e-01 4.56431866e-01
6.59217656e-01 1.17004111e-01 -2.38408744e-01 -5.23950934e-01
5.09834826e-01 4.43848521e-02 -3.17242861e-01 -3.06854546e-02
-8.28036010e-01 1.04372430e+00 6.52038455e-01 6.07393356e-03
-6.28921613e-02 2.64858633e-01 1.70484856e-01 -1.64417192e-01
9.63812917e-02 3.35763574e-01 -2.89849162e-01 1.85417727e-01
-6.83410764e-01 4.00591373e-01 5.74983001e-01 7.60014176e-01
7.58763969e-01 -4.27339338e-02 -1.94445804e-01 4.72168863e-01
1.51370689e-01 6.77972496e-01 1.18780464e-01 -5.13640404e-01
4.55317765e-01 4.82486516e-01 -6.32072240e-02 -1.21349585e+00
-8.22717845e-02 -5.14164329e-01 -6.37690544e-01 2.50347465e-01
6.58398151e-01 -2.74418384e-01 -1.11888039e+00 1.86699927e+00
6.61998093e-01 5.88451922e-01 7.22183213e-02 8.73606563e-01
3.75348151e-01 6.44704163e-01 -2.06057765e-02 -9.63767096e-02
1.36166620e+00 -6.03492975e-01 -4.29450214e-01 -2.15746611e-01
5.77764928e-01 -3.36647481e-01 9.99820530e-01 4.49600667e-01
-6.51726961e-01 -8.97117630e-02 -1.33237088e+00 4.47127610e-01
-6.75492346e-01 -4.17635202e-01 4.94006813e-01 1.28941000e+00
-4.74496782e-01 2.44258076e-01 -7.32424915e-01 2.55434722e-01
9.86149728e-01 2.56962836e-01 1.29379645e-01 -1.25562906e-01
-1.34218180e+00 1.89301118e-01 3.50772083e-01 1.81715652e-01
-1.25091863e+00 -6.66274250e-01 -5.15986621e-01 -1.88222259e-01
7.02712476e-01 -1.58272296e-01 9.86482382e-01 -7.00149894e-01
-1.04247248e+00 6.73867345e-01 -5.27363410e-03 -7.55435288e-01
8.53078008e-01 -1.25986472e-01 -2.91560888e-01 1.48831487e-01
2.97802806e-01 6.73386514e-01 1.08601880e+00 -1.64973330e+00
-4.91980374e-01 -6.52230263e-01 3.55014265e-01 -1.37688532e-01
-5.31976461e-01 -2.22259790e-01 -3.39388341e-01 -7.50807166e-01
2.28398353e-01 -9.26794767e-01 -4.06828016e-01 -4.11875509e-02
-1.24298298e+00 -2.21074093e-03 1.40668857e+00 1.21855801e-02
1.07238257e+00 -2.28405714e+00 -6.55471146e-01 4.70997453e-01
2.43648499e-01 7.03089118e-01 7.14726225e-02 -1.57270238e-01
3.62014510e-02 5.70443392e-01 -5.95609725e-01 4.80266884e-02
-1.79918066e-01 3.97865087e-01 -1.01529789e+00 5.40011108e-01
-1.07522160e-01 7.36139536e-01 -9.36519623e-01 -1.23986073e-01
1.09316103e-01 3.43987674e-01 -4.98646557e-01 -5.14803231e-02
-3.68883610e-01 1.20421387e-01 -7.37767458e-01 8.15511584e-01
1.12042475e+00 -4.70546074e-02 -2.75403142e-01 -1.40705451e-01
3.05266947e-01 9.23286006e-02 -8.88307452e-01 8.35095644e-01
-3.09557058e-02 2.83341885e-01 -8.97247717e-02 -6.39366806e-01
8.17294955e-01 -4.66316007e-03 5.81577383e-02 -3.71844798e-01
2.69883901e-01 2.29171947e-01 4.93438616e-02 -7.77356979e-03
3.22149068e-01 9.01798680e-02 -2.91823417e-01 5.52697122e-01
-6.72710836e-01 1.78742483e-01 -3.23194116e-01 3.40753645e-01
1.31811190e+00 -3.62146229e-01 -2.65972614e-02 -1.63897097e-01
3.73247504e-01 -1.00054787e-02 8.10749650e-01 1.34618461e+00
-4.89399254e-01 5.02830267e-01 7.69818127e-01 -2.59232372e-01
-5.15646100e-01 -1.29809511e+00 -2.77303606e-01 5.51227570e-01
7.29621887e-01 -4.53781188e-02 -1.34194148e+00 -1.33411586e+00
1.07068710e-01 8.19778860e-01 -5.57994068e-01 -4.35202271e-01
-3.60196441e-01 -8.69959712e-01 1.28099513e+00 3.65487665e-01
1.08269119e+00 -1.04524672e+00 -6.92292809e-01 -7.09207654e-02
8.71336609e-02 -1.22066438e+00 -4.03268665e-01 1.23854451e-01
-6.28043056e-01 -1.33985603e+00 -4.11553115e-01 -3.15234125e-01
8.38568747e-01 4.75499779e-01 4.12683636e-01 7.61309043e-02
-2.07497433e-01 -1.39150359e-02 -2.13645816e-01 -4.86470789e-01
-3.66897106e-01 -3.46662402e-02 -6.62428066e-02 2.98549116e-01
4.80791867e-01 -3.63326281e-01 -5.60123920e-01 6.15668833e-01
-1.17681456e+00 -4.62415695e-01 4.02600378e-01 4.32221830e-01
5.50891340e-01 4.66338336e-01 6.86533988e-01 -1.43314946e+00
4.29570556e-01 -6.23118401e-01 -8.96006882e-01 7.82646686e-02
-2.76341766e-01 -3.63810390e-01 8.58278930e-01 -8.01780403e-01
-5.66991687e-01 1.17539845e-01 -2.28489369e-01 -8.86543453e-01
-3.68763894e-01 1.76172331e-02 -8.87347877e-01 -2.45410472e-01
6.23195231e-01 2.33243600e-01 -3.23811710e-01 -1.57970190e-01
3.62693310e-01 6.52755916e-01 5.63142836e-01 -5.18910527e-01
1.07950449e+00 9.40753460e-01 -5.78774065e-02 -8.13391387e-01
-7.56447732e-01 -2.98740063e-02 3.04846582e-03 -2.25309372e-01
8.75759780e-01 -6.39002383e-01 -7.01938212e-01 1.03037047e+00
-1.14562261e+00 -2.65511274e-01 -1.53960278e-02 -6.85701892e-02
3.78913209e-02 3.30953658e-01 -5.30168295e-01 -8.44213367e-01
-1.49542883e-01 -1.59237945e+00 8.04656565e-01 3.70761812e-01
-2.19167862e-02 -8.86729777e-01 -7.08769262e-01 5.09867549e-01
-4.18846197e-02 4.15457606e-01 9.08135414e-01 -1.13781774e+00
-9.93912935e-01 -3.08402717e-01 -1.12982310e-01 2.88627446e-01
1.59922764e-01 -2.53311004e-02 -1.20745671e+00 -1.52930617e-01
3.46360922e-01 -9.39117819e-02 8.18118572e-01 1.70346931e-01
1.73770750e+00 -7.08768785e-01 -5.92233539e-01 6.19854391e-01
1.07764864e+00 5.97020626e-01 6.79041386e-01 3.88230950e-01
9.28228498e-01 2.56054401e-01 6.01413310e-01 2.68759876e-01
-1.44887179e-01 3.58977526e-01 9.03103352e-01 5.97113967e-02
2.63046682e-01 -4.55087602e-01 2.44397953e-01 -8.65185633e-02
7.37154245e-01 -5.59655786e-01 -8.01213562e-01 4.92363721e-01
-1.44847786e+00 -6.60707593e-01 -2.64840454e-01 2.29096222e+00
7.94870019e-01 6.44970596e-01 -9.42120478e-02 3.98103029e-01
1.02258027e+00 5.13362288e-01 -9.25484776e-01 -3.18727821e-01
-1.02268562e-01 -5.64029813e-02 1.01365972e+00 5.37229002e-01
-1.21840966e+00 1.04853535e+00 5.54310036e+00 1.38155150e+00
-1.07820189e+00 1.81403890e-01 9.97847736e-01 -3.93567188e-03
-6.57113433e-01 1.37150452e-01 -1.09225130e+00 8.15870285e-01
2.85604835e-01 1.70281291e-01 2.14004755e-01 9.50719893e-01
-8.02475214e-02 -8.51417631e-02 -8.30457032e-01 5.96423984e-01
1.03420638e-01 -1.31021750e+00 4.29099649e-01 3.32640260e-01
6.93785965e-01 -3.29613626e-01 3.89844984e-01 7.70856142e-02
3.47709566e-01 -8.08876514e-01 7.36786962e-01 -2.37104312e-01
4.23825085e-01 -1.20293832e+00 4.30153102e-01 5.88937938e-01
-7.04660296e-01 -2.18933597e-01 -2.51849502e-01 2.36276582e-01
-1.62839279e-01 9.44677472e-01 -5.38929999e-01 2.22840846e-01
5.46180725e-01 2.03820199e-01 -4.66740936e-01 6.86136186e-01
-6.33081853e-01 9.44932818e-01 -4.30461645e-01 2.81079244e-02
6.27810776e-01 -1.26694530e-01 8.76968563e-01 7.03062296e-01
-4.67895456e-02 9.74793434e-02 1.57546014e-01 1.21536565e+00
-3.86334509e-01 -4.44818258e-01 -9.35924351e-01 2.39888921e-01
9.77854431e-01 1.10931385e+00 -1.14353216e+00 -2.70136714e-01
3.31489258e-02 9.20922875e-01 -9.37518850e-02 4.41805333e-01
-1.33348286e+00 -4.80817258e-01 8.34937394e-01 5.24818152e-02
2.82238066e-01 2.67563194e-01 -7.07899630e-01 -7.39449024e-01
7.51534998e-02 -9.65431392e-01 3.31119835e-01 -5.02969503e-01
-1.20449460e+00 3.10037017e-01 -5.95323853e-02 -9.32742178e-01
4.14001554e-01 -4.48116004e-01 -8.88364673e-01 4.89134461e-01
-1.10525584e+00 -9.40620840e-01 6.97645918e-03 7.55834699e-01
2.90406972e-01 7.50503391e-02 3.97456110e-01 1.36203051e-01
-8.06324065e-01 9.93659556e-01 -2.50491470e-01 5.09303272e-01
1.42728433e-01 -8.34492028e-01 4.15533394e-01 1.29064536e+00
3.23441297e-01 8.28857839e-01 4.72960055e-01 -9.70182180e-01
-1.16140056e+00 -1.54080009e+00 3.34475666e-01 -2.61957765e-01
4.68916148e-01 -6.29720628e-01 -1.06490457e+00 7.74851501e-01
-3.31709921e-01 8.65697190e-02 5.66592932e-01 -2.67405182e-01
-6.49950266e-01 -1.15740970e-01 -1.55467629e+00 9.75983083e-01
8.19279134e-01 -6.32931769e-01 -2.94412345e-01 4.23154682e-01
7.80353427e-01 -3.91853571e-01 -1.92648172e-01 5.38670778e-01
1.98637620e-01 -1.02521229e+00 9.79368746e-01 -1.71188533e-01
2.46217489e-01 -5.31614542e-01 -9.05457661e-02 -8.03620636e-01
3.14832926e-01 -7.61065900e-01 -1.07803054e-01 1.25538552e+00
3.65557492e-01 -1.04887569e+00 1.19017565e+00 4.60498512e-01
9.63160545e-02 -7.85560489e-01 -1.23079491e+00 -1.05618262e+00
2.88164198e-01 -7.17089117e-01 8.69736135e-01 1.00805914e+00
-3.14377576e-01 -1.38531104e-01 -2.16224328e-01 9.59568977e-01
9.96264458e-01 3.61667350e-02 6.30171061e-01 -7.41684079e-01
-2.19195440e-01 -4.07017082e-01 -3.72120440e-01 -1.18779278e+00
1.78083390e-01 -6.69025302e-01 1.21408716e-01 -7.67802536e-01
1.71695396e-01 -7.74255693e-01 -2.60821700e-01 6.42470360e-01
-3.36693555e-01 6.31285548e-01 1.83984235e-01 1.89117879e-01
-3.04306656e-01 1.47419617e-01 1.10028684e+00 -2.90191978e-01
1.17971199e-02 3.89601678e-01 -6.30604088e-01 9.68371451e-01
8.68291974e-01 -8.71321261e-01 -7.55105734e-01 -1.70789883e-01
2.24437434e-02 -1.63489550e-01 6.60095513e-01 -1.01998758e+00
1.51173934e-01 -5.73846996e-02 1.61664918e-01 -8.51128578e-01
2.06760302e-01 -8.49171579e-01 -4.46521454e-02 6.91665947e-01
-2.23231927e-01 -4.76477414e-01 1.40762165e-01 7.80499160e-01
1.16635270e-01 -4.71168190e-01 8.69324207e-01 -2.60224715e-02
-4.50270027e-01 3.00870210e-01 -5.69906771e-01 4.53219041e-02
1.52358115e+00 -4.97774571e-01 -6.56422317e-01 -7.36699924e-02
-4.34401929e-01 1.29847795e-01 5.28399527e-01 2.91390806e-01
6.25888407e-01 -1.03114378e+00 -1.38157547e-01 5.01410186e-01
-4.04605083e-02 3.21178108e-01 1.82028666e-01 2.91497797e-01
-2.95028359e-01 8.59639794e-02 2.98294961e-01 -7.01061070e-01
-1.27723682e+00 7.93911099e-01 3.81307691e-01 4.55062389e-02
-5.52331090e-01 1.02464545e+00 6.09660923e-01 -3.57877702e-01
3.38337183e-01 -2.31888205e-01 1.61437422e-01 -1.94530278e-01
4.84551162e-01 1.84238404e-01 -2.28079334e-01 -4.56551760e-01
-3.92092735e-01 2.70938933e-01 -4.58648384e-01 1.70643274e-02
5.61389506e-01 2.58387089e-01 -4.35906425e-02 1.80126160e-01
1.04379582e+00 2.87003368e-01 -1.33528388e+00 6.62058666e-02
-2.31749028e-01 -6.99630141e-01 -3.75742614e-02 -8.32988262e-01
-1.37886012e+00 9.20818985e-01 2.80409008e-01 4.09619808e-01
1.04796267e+00 -1.12482414e-01 1.13805246e+00 3.22691321e-01
5.72311699e-01 -6.88742697e-01 -2.30322983e-02 2.90379882e-01
2.18595207e-01 -7.21101642e-01 -2.18179658e-01 -7.62944400e-01
-4.49815422e-01 5.43839276e-01 7.99986660e-01 -9.56686735e-02
6.14326596e-01 4.20871645e-01 2.66993612e-01 -5.17886467e-02
-2.25115582e-01 3.42499256e-01 -3.81479174e-01 5.24925947e-01
-5.83027661e-01 7.12095723e-02 -1.62824355e-02 6.42408133e-01
-1.40383229e-01 -4.15871441e-01 7.12925732e-01 9.88158405e-01
-4.26260442e-01 -9.30334747e-01 -6.63704336e-01 3.37911934e-01
-6.86769724e-01 -1.45394104e-02 -6.13881409e-01 7.29392290e-01
3.51802140e-01 1.01857078e+00 -1.96362962e-03 -5.04480481e-01
7.92471692e-02 -1.11364901e-01 7.92815117e-04 -4.17344123e-01
-5.91925681e-01 -5.21839634e-02 -3.18187654e-01 -6.18321657e-01
1.43705636e-01 -5.05935729e-01 -1.46079314e+00 -2.49768287e-01
-4.80217069e-01 9.89093259e-02 4.70425099e-01 1.04862607e+00
1.71414778e-01 3.58060807e-01 8.98593783e-01 -1.96353242e-01
-7.68278658e-01 -3.97883892e-01 -7.10216820e-01 2.22620651e-01
2.35101551e-01 -5.59801459e-01 -7.15427399e-01 -3.68599623e-01] | [5.667520523071289, 7.7415852546691895] |
7add29be-3332-4d88-ab00-0d9426ff85cd | bayesian-methods-for-semi-supervised-text | 2010.14872 | null | https://arxiv.org/abs/2010.14872v1 | https://arxiv.org/pdf/2010.14872v1.pdf | Bayesian Methods for Semi-supervised Text Annotation | Human annotations are an important source of information in the development of natural language understanding approaches. As under the pressure of productivity annotators can assign different labels to a given text, the quality of produced annotations frequently varies. This is especially the case if decisions are difficult, with high cognitive load, requires awareness of broader context, or careful consideration of background knowledge. To alleviate the problem, we propose two semi-supervised methods to guide the annotation process: a Bayesian deep learning model and a Bayesian ensemble method. Using a Bayesian deep learning method, we can discover annotations that cannot be trusted and might require reannotation. A recently proposed Bayesian ensemble method helps us to combine the annotators' labels with predictions of trained models. According to the results obtained from three hate speech detection experiments, the proposed Bayesian methods can improve the annotations and prediction performance of BERT models. | ['Marko Robnik-Sikonja', 'Gregor Pirs', 'Kristian Miok'] | 2020-10-28 | null | https://aclanthology.org/2020.law-1.1 | https://aclanthology.org/2020.law-1.1.pdf | coling-law-2020-12 | ['text-annotation'] | ['natural-language-processing'] | [ 1.45734698e-02 2.97867268e-01 -3.44922170e-02 -4.34894115e-01
-4.56732273e-01 -5.49634695e-01 4.43035662e-01 3.46905500e-01
-4.27881122e-01 7.28772700e-01 3.70653331e-01 -1.67267993e-01
-1.18255056e-02 -3.65329832e-01 -2.01925471e-01 -5.82717240e-01
6.66333079e-01 5.51611722e-01 3.48424643e-01 -1.26462772e-01
3.97354245e-01 4.75012250e-02 -1.46004570e+00 3.49757999e-01
1.05239153e+00 8.91282022e-01 3.15616995e-01 4.03885871e-01
-3.74985546e-01 1.15460479e+00 -7.58444488e-01 -7.74643302e-01
-2.04405516e-01 -1.25268728e-01 -9.99495029e-01 2.37851385e-02
-5.12909517e-02 -2.90895104e-01 3.25787440e-02 1.26716685e+00
3.29772085e-01 1.77167132e-01 7.57603407e-01 -9.94385660e-01
-5.81196845e-01 1.09923744e+00 -3.36593658e-01 1.94746569e-01
2.73879200e-01 -1.61217600e-01 9.63309467e-01 -7.86127150e-01
3.25518638e-01 1.17082131e+00 8.82871568e-01 5.56669831e-01
-1.01638508e+00 -5.62466264e-01 2.86140531e-01 4.13025081e-01
-1.28463411e+00 -3.72110337e-01 7.21502125e-01 -1.02570927e+00
3.34860414e-01 -1.49128854e-01 1.85277924e-01 1.56728625e+00
-2.72907708e-02 6.46668613e-01 1.13269913e+00 -6.00850582e-01
2.48751193e-01 6.45431101e-01 4.98924792e-01 5.99521101e-01
1.39611006e-01 -2.82443047e-01 -8.46065164e-01 -3.34179431e-01
-5.69317229e-02 -9.48265120e-02 -3.60356838e-01 1.25513628e-01
-9.43048894e-01 8.27591479e-01 1.77357066e-02 4.59077716e-01
-2.60875195e-01 -2.10215554e-01 5.10343969e-01 -3.03576857e-01
7.14069903e-01 4.99084473e-01 -4.51970309e-01 -1.65835708e-01
-9.35350835e-01 5.24266139e-02 1.00715029e+00 6.46947443e-01
5.23851693e-01 -3.17973375e-01 -3.41602802e-01 8.67948771e-01
8.28039289e-01 3.30646902e-01 5.41784108e-01 -7.83911765e-01
1.34789363e-01 5.17112672e-01 5.91030419e-01 -1.27137947e+00
-3.42029393e-01 -3.08346778e-01 -6.34010077e-01 2.63109505e-02
5.52396655e-01 -2.56674826e-01 -5.48652112e-01 1.31975555e+00
2.93460339e-01 -1.13855571e-01 -1.47796035e-01 7.40838170e-01
6.61843836e-01 4.42260563e-01 4.47259784e-01 -2.20777243e-01
1.53678465e+00 -8.14134836e-01 -1.36202681e+00 -1.00657992e-01
7.45969832e-01 -9.22007442e-01 7.74401128e-01 6.02698743e-01
-3.68099153e-01 -5.01059532e-01 -1.00061440e+00 4.96664690e-03
-3.93360227e-01 4.85436231e-01 1.77264392e-01 6.94451272e-01
-5.47684491e-01 6.36581421e-01 -7.53460288e-01 -2.43239984e-01
3.88797492e-01 9.76851559e-04 -1.17798150e-01 9.34338868e-02
-1.36965787e+00 1.18434227e+00 6.58535540e-01 4.67239112e-01
-7.95531631e-01 -1.45380169e-01 -4.65942204e-01 2.70722628e-01
5.07942796e-01 -2.05268502e-01 1.56617308e+00 -1.04874682e+00
-1.39968443e+00 6.22509956e-01 -1.63207516e-01 -2.59087354e-01
5.82050323e-01 -6.54374421e-01 -2.46031865e-01 -2.18477651e-01
3.98733057e-02 2.42835850e-01 8.63790572e-01 -1.30042696e+00
-6.27612352e-01 -4.04369891e-01 -6.52410612e-02 -1.44687966e-01
-4.11061943e-01 3.05736452e-01 1.33755207e-01 -4.88427192e-01
-3.98027152e-02 -1.02302504e+00 -6.40664389e-03 -2.44709551e-01
-6.29487157e-01 -6.33357048e-01 1.00366318e+00 -9.80893493e-01
1.39701116e+00 -2.24162745e+00 2.38141287e-02 -5.60503500e-03
4.57727313e-01 4.37389553e-01 4.62124765e-01 1.58424899e-01
2.83291608e-01 4.69096869e-01 -4.18721922e-02 -4.61707503e-01
2.37001464e-01 1.58343256e-01 -3.86274785e-01 2.12546773e-02
-1.23642281e-01 1.59690812e-01 -1.06138289e+00 -6.30921245e-01
-9.89981219e-02 4.67596173e-01 -3.08990143e-02 5.68759322e-01
-3.71965170e-01 5.72369874e-01 -4.32738066e-01 3.55157733e-01
3.06527585e-01 -4.58381385e-01 3.06228757e-01 -2.72938162e-01
-1.59678403e-02 4.21128660e-01 -9.48951185e-01 1.49784350e+00
-1.99717239e-01 8.12453628e-01 -2.68299639e-01 -8.58204305e-01
1.10735095e+00 6.62863374e-01 -8.46260414e-02 1.10351875e-01
3.93281013e-01 2.29194239e-01 6.30982965e-02 -8.13243270e-01
6.12961054e-01 4.39519882e-02 1.21336751e-01 6.52988493e-01
1.64131626e-01 1.15717664e-01 1.08012252e-01 2.13881120e-01
8.90913308e-01 2.82136708e-01 3.16298276e-01 -2.16771498e-01
3.40188891e-01 -6.55794367e-02 9.20253336e-01 7.49598503e-01
-4.60919440e-01 1.67351663e-01 7.03829229e-01 -6.16985500e-01
-1.04815722e+00 -4.12796468e-01 -1.33034557e-01 1.39801383e+00
-1.29838407e-01 -4.62085366e-01 -8.40210557e-01 -1.17061603e+00
-3.18372369e-01 8.67445290e-01 -5.08048594e-01 -1.85402334e-01
8.49764571e-02 -7.54228294e-01 5.03726900e-01 4.16551024e-01
4.08207208e-01 -8.96708667e-01 -5.89389026e-01 2.01958597e-01
-4.74962890e-01 -1.19554687e+00 -1.41080737e-01 3.89336348e-01
-3.12995464e-01 -9.58916306e-01 -4.61247921e-01 -3.02923232e-01
3.84151489e-01 -8.42958614e-02 8.57757747e-01 3.08660954e-01
3.14321697e-01 8.19718912e-02 -8.55414093e-01 -7.68457472e-01
-7.55040467e-01 2.37842157e-01 2.23530307e-01 1.69761881e-01
8.17011654e-01 -2.58092612e-01 -2.06737429e-01 5.42284131e-01
-7.08764553e-01 1.50584146e-01 1.58353910e-01 8.42986703e-01
6.34884909e-02 2.36149460e-01 5.81611812e-01 -1.10952008e+00
7.18401790e-01 -6.01208687e-01 -6.23604059e-01 3.77869993e-01
-6.68903530e-01 1.66452214e-01 2.16982201e-01 -5.05331755e-01
-1.42509127e+00 1.00129075e-01 2.89554652e-02 -2.61155069e-01
-3.67697507e-01 7.09987342e-01 -6.81780726e-02 3.19017828e-01
9.17441607e-01 -4.48986053e-01 -3.33178550e-01 -6.98714912e-01
7.23982602e-02 1.06614459e+00 1.58364847e-01 -5.37716269e-01
3.25397313e-01 3.11057782e-03 -5.19497335e-01 -4.59466010e-01
-1.73088360e+00 -3.34022045e-01 -1.08623624e+00 -4.69309419e-01
1.08854198e+00 -7.21716344e-01 -4.25843447e-01 5.40217340e-01
-1.71662903e+00 -3.02414030e-01 3.29509646e-01 5.02633989e-01
-2.37803627e-02 4.34245080e-01 -3.76054794e-01 -1.20081973e+00
-3.31378914e-02 -1.30477786e+00 7.49716818e-01 3.70826423e-01
-7.34181225e-01 -1.04151654e+00 4.24393900e-02 7.83839047e-01
3.19751620e-01 -5.91345392e-02 8.88146400e-01 -1.38605762e+00
-2.49734789e-01 -2.02374518e-01 -1.63019270e-01 4.57481116e-01
-1.07380837e-01 2.09469378e-01 -1.45213473e+00 3.00629675e-01
-7.35951513e-02 -4.72053140e-01 6.51021898e-01 1.15853911e-02
1.25828362e+00 -3.55926692e-01 -4.42836493e-01 -2.32660428e-01
7.78926075e-01 1.75295919e-01 2.56952792e-01 1.38574407e-01
7.52672791e-01 8.62539589e-01 4.62220222e-01 7.13516831e-01
4.47117478e-01 7.42205739e-01 3.07061791e-01 4.65673327e-01
2.47285530e-01 -2.26556584e-01 1.85314283e-01 9.18536901e-01
-2.27895543e-01 -5.10109603e-01 -1.27509236e+00 3.27463418e-01
-2.27323174e+00 -9.23319638e-01 -2.07937568e-01 1.94819283e+00
1.00833881e+00 7.59419054e-02 -1.37747630e-01 2.43006855e-01
1.02478600e+00 -1.04954146e-01 -3.33151340e-01 -1.84402928e-01
1.44697130e-01 -3.86176318e-01 2.34797925e-01 1.98189408e-01
-1.16009963e+00 8.89917612e-01 6.31346512e+00 7.87875533e-01
-6.68817103e-01 5.35436094e-01 6.76245034e-01 3.86443377e-01
-8.17881525e-02 -7.76167065e-02 -1.02587855e+00 7.10722923e-01
1.09152377e+00 1.15330189e-01 1.22933008e-01 1.02629530e+00
2.81418830e-01 -5.11639178e-01 -1.05322635e+00 7.70419002e-01
2.23118767e-01 -9.96767819e-01 -3.88527632e-01 4.54924256e-02
6.51755571e-01 -1.57331467e-01 -3.32456499e-01 2.80876815e-01
7.21821606e-01 -9.14314806e-01 7.70866930e-01 8.48864257e-01
5.17306924e-02 -5.43645024e-01 1.24611616e+00 6.74263418e-01
-4.86077517e-01 -2.12894589e-01 -2.81094164e-01 -1.01441436e-01
1.82773888e-01 1.04050589e+00 -9.38181162e-01 9.51727256e-02
7.85002410e-01 4.91045475e-01 -4.76677924e-01 9.39601362e-01
-8.66349041e-01 8.57061148e-01 1.17282588e-02 -2.42065564e-01
-5.79843558e-02 2.18805552e-01 4.01431948e-01 1.06007099e+00
2.30278611e-01 1.13032728e-01 4.26290810e-01 1.00156832e+00
-9.87140760e-02 1.44731894e-01 -5.21132708e-01 -2.16274798e-01
6.86875105e-01 1.38824356e+00 -8.13059092e-01 -3.68723571e-01
-2.67836094e-01 7.67966509e-01 4.68509853e-01 2.32917115e-01
-6.37386382e-01 -1.38339221e-01 1.61676034e-01 -9.43968445e-02
5.72636090e-02 -1.73903212e-01 -3.47656101e-01 -1.02748322e+00
-3.77506435e-01 -9.11469400e-01 2.30936497e-01 -1.17266989e+00
-1.39520502e+00 7.32843101e-01 4.24296502e-03 -7.67853856e-01
-9.89491642e-02 -7.04216361e-01 -3.13016087e-01 7.67984450e-01
-1.11652589e+00 -8.50864351e-01 -3.42076391e-01 6.41478226e-02
5.94714761e-01 -1.72803193e-01 9.06015158e-01 1.25691488e-01
-7.85834014e-01 2.13322446e-01 1.37464236e-02 4.10668880e-01
1.02127945e+00 -1.26630068e+00 -2.77965933e-01 7.81294167e-01
2.05498233e-01 3.94463480e-01 8.58916163e-01 -6.85457468e-01
-5.22620022e-01 -7.59867132e-01 1.12807477e+00 -7.78073668e-01
7.70383239e-01 -7.82661587e-02 -1.29339182e+00 5.17667592e-01
3.52310121e-01 -3.01958919e-01 1.21408033e+00 5.95111728e-01
-5.24693608e-01 2.94254392e-01 -8.82578552e-01 1.06821701e-01
5.17503679e-01 -7.25467682e-01 -9.33414817e-01 4.57201272e-01
5.14933765e-01 -2.94526666e-01 -7.09378242e-01 2.08628371e-01
6.23489380e-01 -8.13010633e-01 1.54850423e-01 -4.33476746e-01
3.85872722e-01 -4.42471653e-01 -7.15481266e-02 -1.49876368e+00
-2.04909787e-01 -3.34132284e-01 -7.93582201e-02 1.47519457e+00
6.09395862e-01 -7.12667406e-02 2.63612300e-01 1.21423030e+00
3.57773900e-02 -2.08899409e-01 -4.85470206e-01 -3.97811770e-01
-4.33358133e-01 -4.68764156e-01 5.11304855e-01 1.26354325e+00
3.76849741e-01 5.70344985e-01 -5.91133535e-01 3.22842896e-01
2.90713102e-01 -4.61705714e-01 6.91090941e-01 -1.93263388e+00
-2.56929006e-02 -3.23952019e-01 -1.57800153e-01 -6.21393442e-01
5.16892254e-01 -5.49950182e-01 5.47888398e-01 -1.26461709e+00
4.85868722e-01 -3.95765841e-01 -2.66701609e-01 8.10750902e-01
-3.90133172e-01 2.08998285e-03 8.79650097e-03 3.70887876e-01
-7.41025686e-01 3.73668373e-01 7.28787363e-01 -1.07249938e-01
4.79871891e-02 1.45607620e-01 -6.32431567e-01 1.25026619e+00
6.91208601e-01 -8.29458356e-01 -2.63120621e-01 -3.96023154e-01
6.85163796e-01 -2.82994151e-01 1.83097377e-01 -7.53948987e-01
2.25132748e-01 -5.96902408e-02 1.79335654e-01 -3.38080883e-01
5.89226335e-02 -8.48362923e-01 7.70093352e-02 -3.07336599e-02
-6.81244254e-01 -2.39473611e-01 -1.66130304e-01 7.09247172e-01
-1.17899783e-01 -9.40551043e-01 5.76187491e-01 -1.76661208e-01
-5.50058007e-01 -1.41209945e-01 -7.61008501e-01 -2.43797660e-01
6.77858233e-01 1.05624944e-01 -4.56231505e-01 -3.50170523e-01
-1.09975970e+00 8.38769078e-02 2.03389362e-01 4.22339320e-01
1.03425294e-01 -1.11298060e+00 -5.58375239e-01 -3.54620785e-01
2.06534490e-01 6.84815124e-02 6.08740970e-02 8.21684003e-01
-1.14861898e-01 1.21474996e-01 7.48832151e-02 -5.77232897e-01
-1.13100243e+00 3.48698646e-01 2.27994069e-01 -3.50105733e-01
-8.68668035e-02 7.56787598e-01 2.45790143e-04 -2.05664232e-01
4.75113004e-01 -3.92503627e-02 -7.10096002e-01 3.95673007e-01
8.25691998e-01 4.57221538e-01 3.60175259e-02 -7.59069502e-01
1.49498107e-02 -1.76360384e-02 -2.35342175e-01 -1.03353687e-01
1.03700340e+00 -3.41609478e-01 -1.67934641e-01 9.04785872e-01
3.58741283e-01 -5.00107482e-02 -1.04467523e+00 -4.56219435e-01
4.50077683e-01 -4.65390086e-01 4.00385708e-01 -9.56018925e-01
-6.67750537e-01 1.07818758e+00 4.88576144e-01 5.86500466e-01
5.77633321e-01 -9.81184691e-02 3.23718458e-01 5.45607686e-01
2.54885286e-01 -1.39050066e+00 1.71032891e-01 7.40247369e-01
6.18866146e-01 -1.63662720e+00 -1.08067669e-01 -2.88958281e-01
-8.21075976e-01 1.24347317e+00 8.41757774e-01 5.39123535e-01
7.56297708e-01 3.64413708e-02 2.32968435e-01 -1.27068326e-01
-7.78199494e-01 -7.97825605e-02 4.79295164e-01 5.10449767e-01
7.31025159e-01 8.72851536e-02 -2.88210183e-01 9.42982137e-01
1.38502583e-01 -1.67218789e-01 6.39710069e-01 5.46854615e-01
-7.09433436e-01 -9.83354092e-01 -5.50452948e-01 1.68159738e-01
-6.43426836e-01 -1.01028644e-01 -5.20950258e-01 1.65240005e-01
4.10960883e-01 1.18848062e+00 -2.33952120e-01 -3.68550509e-01
-1.57607391e-01 7.40040362e-01 -7.43062869e-02 -7.89994180e-01
-3.50235313e-01 1.69375129e-02 3.14630240e-01 -4.04470712e-02
-8.15851986e-01 -5.56229770e-01 -8.32094848e-01 -9.42813531e-02
-9.74961221e-01 3.94797266e-01 8.08813214e-01 1.57209969e+00
2.53837764e-01 4.16892648e-01 3.24811280e-01 -6.90299392e-01
-4.83837843e-01 -1.49455225e+00 -4.72326070e-01 3.35936606e-01
-8.27859342e-02 -9.30223167e-01 -3.09686482e-01 6.15816355e-01] | [9.667580604553223, 4.656586647033691] |
5c7872fb-5a18-4370-98cd-45cfcbff1672 | unsupervised-learning-based-long-term | 1902.09596 | null | http://arxiv.org/abs/1902.09596v1 | http://arxiv.org/pdf/1902.09596v1.pdf | Unsupervised learning-based long-term superpixel tracking | Finding correspondences between structural entities decomposing images is of
high interest for computer vision applications. In particular, we analyze how
to accurately track superpixels - visual primitives generated by aggregating
adjacent pixels sharing similar characteristics - over extended time periods
relying on unsupervised learning and temporal integration. A two-step video
processing pipeline dedicated to long-term superpixel tracking is proposed.
First, unsupervised learning-based superpixel matching provides correspondences
between consecutive and distant frames using new context-rich features extended
from greyscale to multi-channel and forward-backward consistency contraints.
Resulting elementary matches are then combined along multi-step paths running
through the whole sequence with various inter-frame distances. This produces a
large set of candidate long-term superpixel pairings upon which majority voting
is performed. Video object tracking experiments demonstrate the accuracy of our
elementary estimator against state-of-the-art methods and proves the ability of
multi-step integration to provide accurate long-term superpixel matches
compared to usual direct and sequential integration. | ['Gwenolé Quellec', 'Pierre-Henri Conze', 'Florian Tilquin', 'Fabrice Heitz', 'Mathieu Lamard'] | 2019-02-25 | null | null | null | null | ['video-object-tracking'] | ['computer-vision'] | [ 4.25505757e-01 -3.81656766e-01 -1.61785200e-01 -1.47282928e-01
-1.04531276e+00 -6.38405800e-01 5.81621706e-01 1.91376105e-01
-6.53048992e-01 8.14226270e-01 -3.53388309e-01 2.41246879e-01
-7.24462867e-02 -5.48400819e-01 -8.02784681e-01 -6.59293354e-01
-2.86946446e-01 3.37565660e-01 1.06377614e+00 1.80562899e-01
2.43123993e-01 7.73818135e-01 -1.70839119e+00 3.89298648e-01
7.14050114e-01 1.04645383e+00 3.64120722e-01 8.55705202e-01
7.52782747e-02 5.70133984e-01 -1.53106675e-01 -2.83517271e-01
6.40016437e-01 -2.94705600e-01 -6.32454515e-01 4.78030771e-01
1.19217110e+00 -4.17606950e-01 1.03187658e-01 1.26354861e+00
4.51398492e-02 2.37842634e-01 3.82397264e-01 -1.18686867e+00
-1.47472441e-01 2.24558040e-01 -4.89805758e-01 4.27424520e-01
4.59227204e-01 1.72448874e-01 1.04147875e+00 -1.02170527e+00
1.13751602e+00 9.13163900e-01 9.25266683e-01 1.20581560e-01
-1.44706595e+00 -4.40430284e-01 9.40063000e-02 2.10347354e-01
-1.34064662e+00 -2.98683852e-01 6.33923650e-01 -6.77404642e-01
7.79532790e-01 2.57752270e-01 9.54686403e-01 3.99140000e-01
7.97084421e-02 6.96234047e-01 1.03048706e+00 -5.02929568e-01
2.16493770e-01 -3.93127687e-02 4.64633666e-02 9.13724005e-01
2.86624402e-01 2.52647609e-01 -5.69933593e-01 -3.45662124e-02
1.05189919e+00 7.03276768e-02 -1.37143016e-01 -6.24906540e-01
-1.70770168e+00 2.67592847e-01 4.54887390e-01 5.68927348e-01
-6.97866142e-01 3.04028690e-01 8.79881084e-02 2.22164229e-01
5.00691772e-01 2.22892985e-01 -4.07825798e-01 1.63968027e-01
-1.64803231e+00 1.68462813e-01 3.73810828e-01 9.50778186e-01
1.14724410e+00 -1.45676389e-01 -2.57985204e-01 3.35931420e-01
8.92715231e-02 3.91673386e-01 1.02837123e-02 -1.35246003e+00
9.78018567e-02 4.65785325e-01 3.20878297e-01 -9.02197421e-01
-2.54988611e-01 -5.30343771e-01 -8.07948351e-01 6.61552191e-01
7.28712797e-01 1.18505172e-01 -6.51426136e-01 1.31026673e+00
6.78199708e-01 7.11231828e-01 -2.09082171e-01 7.89414227e-01
3.02262574e-01 5.32245815e-01 1.78542450e-01 -6.55327082e-01
1.23550367e+00 -9.09964621e-01 -3.12773913e-01 1.95818529e-01
2.24622563e-01 -9.01697516e-01 1.87232614e-01 2.28463516e-01
-1.45603883e+00 -1.07549584e+00 -7.90178597e-01 1.05856530e-01
-3.20932478e-01 3.42531234e-01 4.34933811e-01 4.91626173e-01
-1.24682140e+00 8.08618546e-01 -8.30466747e-01 -2.91776419e-01
4.69545037e-01 4.86090571e-01 -5.65946162e-01 1.65492669e-01
-7.35739052e-01 6.36774480e-01 7.06913888e-01 8.80053788e-02
-7.22734332e-01 -8.90143037e-01 -7.14134276e-01 -1.72452003e-01
4.14150208e-01 -9.10115480e-01 8.62085998e-01 -1.37173450e+00
-1.05289364e+00 1.05038500e+00 -5.45027435e-01 -6.62489533e-01
9.15248930e-01 -1.42682806e-01 -1.62878320e-01 4.02106017e-01
3.15566599e-01 9.19179499e-01 1.14241922e+00 -1.14451718e+00
-1.43025768e+00 -1.99479192e-01 -2.01903265e-02 2.06493303e-01
-2.99848337e-02 1.90016776e-01 -8.69569600e-01 -7.78509319e-01
4.08448219e-01 -9.12698269e-01 -3.62911403e-01 5.20448148e-01
-2.57340878e-01 -1.39277056e-01 7.76991010e-01 -6.22036219e-01
1.07676303e+00 -1.77227390e+00 3.12700957e-01 2.47114584e-01
7.22892582e-02 2.75654882e-01 5.74914552e-02 -1.08934656e-01
-7.65937939e-02 -4.60473984e-01 -2.30794683e-01 -4.40964431e-01
-3.56306762e-01 -2.00544551e-01 2.89109107e-02 5.81088603e-01
1.42482013e-01 9.84984875e-01 -1.16554797e+00 -9.25793469e-01
7.29346573e-01 1.94839999e-01 -1.31433889e-01 1.34921577e-02
-3.27072561e-01 4.83705878e-01 -3.38470161e-01 8.91843498e-01
7.09794879e-01 -2.32484609e-01 -1.03911616e-01 -7.11826265e-01
-4.78272706e-01 -4.88927722e-01 -1.40870678e+00 1.78930402e+00
-5.62922796e-03 8.07276070e-01 -5.20470999e-02 -8.14927638e-01
5.62062860e-01 5.53537980e-02 9.54883814e-01 -3.65706801e-01
-1.72218964e-01 3.70848358e-01 -5.71324944e-01 -3.48330468e-01
7.37740993e-01 1.57840520e-01 2.46632963e-01 2.44760305e-01
1.23711433e-02 1.18317008e-01 6.39134645e-01 5.18931299e-02
7.22270846e-01 5.15794992e-01 5.14577150e-01 -2.33565867e-01
1.08652520e+00 2.49143586e-01 5.37928581e-01 6.57837987e-01
-4.23369616e-01 6.95128441e-01 -4.04188894e-02 -5.51547289e-01
-1.22757232e+00 -1.05635738e+00 3.41554582e-02 9.29108858e-01
4.10906911e-01 -8.60328302e-02 -6.98808849e-01 -5.65481305e-01
1.62590891e-01 1.73437849e-01 -6.34274423e-01 4.68401045e-01
-7.46606648e-01 -4.54424500e-01 1.77063629e-01 6.20773673e-01
5.80888748e-01 -7.99826264e-01 -1.00209832e+00 4.20716077e-01
-1.27254978e-01 -1.19422674e+00 -6.50700212e-01 -7.27270991e-02
-1.00919437e+00 -1.14365578e+00 -1.07635796e+00 -9.43130195e-01
8.01183939e-01 4.75012302e-01 1.06371677e+00 -1.37339178e-02
-5.69392502e-01 5.47686040e-01 -3.30632478e-02 3.58721793e-01
-3.82318616e-01 -3.93273145e-01 3.91297378e-02 4.14085299e-01
1.12302393e-01 -5.08812666e-01 -7.23410726e-01 3.30333740e-01
-6.57509983e-01 1.05964229e-01 6.82969034e-01 6.75540268e-01
1.12226653e+00 -4.03086990e-02 -2.85554565e-02 -4.38125759e-01
-5.35772666e-02 8.55455175e-02 -1.33953238e+00 8.45904648e-01
-3.09468776e-01 -5.88421002e-02 2.04413891e-01 -4.62270796e-01
-1.07932007e+00 5.92019200e-01 3.45612794e-01 -6.89132869e-01
-1.11562274e-01 -1.00370347e-01 8.10020193e-02 -4.54773605e-01
3.35586697e-01 4.20929909e-01 -2.47133106e-01 -5.48311286e-02
6.50570512e-01 2.05589741e-01 1.01426852e+00 -3.34930718e-01
9.05265450e-01 8.63119721e-01 1.40105858e-01 -8.14984679e-01
-3.99411678e-01 -1.01647615e+00 -1.37295032e+00 -7.40429699e-01
1.29508412e+00 -9.74517763e-01 -4.50007439e-01 5.86299479e-01
-1.34240568e+00 -2.15847239e-01 -4.68595386e-01 6.45015776e-01
-6.72491908e-01 6.06869757e-01 -5.17027438e-01 -7.73274064e-01
-2.92610079e-01 -9.78730917e-01 1.32745719e+00 3.08902025e-01
-1.00024819e-01 -1.08019900e+00 1.04919814e-01 3.24195266e-01
6.93718623e-03 3.53718072e-01 1.02617308e-01 -7.56577924e-02
-1.29418790e+00 -7.17775673e-02 -3.56511325e-01 3.05419475e-01
4.06454615e-02 3.97169441e-01 -7.10993767e-01 -2.43574485e-01
-5.00376999e-01 1.46418855e-01 9.61775362e-01 8.22141111e-01
5.31057894e-01 2.42192328e-01 -5.88085532e-01 6.69363558e-01
1.71138966e+00 7.28166625e-02 2.94722378e-01 5.14599979e-01
6.77002847e-01 4.50578451e-01 1.10345423e+00 2.46564314e-01
-2.81088892e-02 8.87139797e-01 1.77352667e-01 -1.49429202e-01
-3.36209893e-01 1.23909257e-01 2.82356709e-01 2.42612824e-01
-3.31579715e-01 2.69640774e-01 -6.22471035e-01 7.77796507e-01
-1.98058951e+00 -1.41112339e+00 -4.81494486e-01 2.45549369e+00
4.84518677e-01 1.70683131e-01 1.90518633e-01 -1.56881243e-01
9.81480718e-01 9.77276266e-02 -4.67596084e-01 4.74024564e-01
-3.47992897e-01 -1.56985559e-02 8.79105151e-01 5.68956673e-01
-1.65146768e+00 9.91070449e-01 5.86681986e+00 9.46749687e-01
-8.98029387e-01 2.52028674e-01 6.17205977e-01 2.30674461e-01
5.38404286e-02 1.68136820e-01 -8.89840603e-01 3.48503411e-01
2.10785598e-01 -1.55833542e-01 -1.19702891e-02 7.42126107e-01
1.55732587e-01 -5.77587903e-01 -1.15372169e+00 1.04503500e+00
7.63207227e-02 -1.77472556e+00 -2.23543823e-01 -1.80020064e-01
1.40614676e+00 2.08369568e-02 -7.20089674e-02 -4.89110649e-01
1.47375137e-01 -3.67078274e-01 1.03916800e+00 7.22883224e-01
6.86932564e-01 -4.06170279e-01 4.08153564e-01 1.05632998e-01
-1.83963823e+00 -1.53418854e-02 -2.43959293e-01 3.37839216e-01
5.93433499e-01 6.87512815e-01 -5.10612071e-01 8.01257968e-01
6.00113273e-01 9.80051935e-01 -6.12048924e-01 1.32637429e+00
8.94167461e-03 -3.66788618e-02 -5.72048783e-01 3.21359247e-01
2.12775856e-01 -5.49185276e-01 5.88956952e-01 1.42371273e+00
4.14675206e-01 7.05282763e-02 3.35079491e-01 8.42643499e-01
2.07253262e-01 -2.44697377e-01 -2.99891025e-01 4.17231649e-01
4.40702617e-01 1.49540031e+00 -1.26602936e+00 -8.54462802e-01
-4.15535361e-01 1.41347873e+00 1.01740018e-01 3.21480840e-01
-7.88320839e-01 2.01325566e-01 5.20926952e-01 2.42946502e-02
6.91781998e-01 -3.92703891e-01 -8.08717608e-02 -1.16289485e+00
1.67208508e-01 -3.91126961e-01 5.54460287e-01 -7.93017149e-01
-1.03703880e+00 4.82750118e-01 -1.64582487e-02 -1.77072644e+00
-3.34537476e-01 -2.75914699e-01 -5.67306876e-01 6.37645841e-01
-1.59611404e+00 -1.56471753e+00 -4.39433813e-01 8.16145062e-01
7.49648273e-01 3.26645002e-02 3.37989777e-01 2.11975232e-01
-1.62551954e-01 2.20080256e-01 3.05056870e-01 1.25155002e-01
5.05721867e-01 -1.11581421e+00 3.23950499e-01 1.42505550e+00
5.15707433e-01 3.26126397e-01 5.78756273e-01 -8.04281235e-01
-9.60626006e-01 -1.40825450e+00 8.74939144e-01 -4.59103316e-01
6.36835098e-01 6.69413134e-02 -7.85396099e-01 4.62611973e-01
2.66207326e-02 4.25273359e-01 1.56616360e-01 -6.34014249e-01
-1.17007531e-01 -3.56920570e-01 -8.73638630e-01 3.64705116e-01
1.01920331e+00 -5.61315238e-01 -4.61248487e-01 2.45710045e-01
2.18405575e-01 -2.80068785e-01 -9.09658432e-01 2.65884161e-01
7.30500638e-01 -1.18362033e+00 1.38048816e+00 -2.98377901e-01
1.32196233e-01 -8.50325942e-01 -1.23225013e-02 -6.24988377e-01
-3.73120219e-01 -7.88252711e-01 9.96656194e-02 9.85217810e-01
9.90949422e-02 -1.88811079e-01 9.10104454e-01 4.99029249e-01
1.27282575e-01 -3.05955082e-01 -9.93681788e-01 -9.19107616e-01
-4.98027235e-01 -2.86746651e-01 -9.73884314e-02 7.50434518e-01
-3.29742581e-01 -2.81630248e-01 -2.57939756e-01 3.76039416e-01
1.38459039e+00 5.75911582e-01 8.35590184e-01 -1.15510464e+00
-3.34155321e-01 -5.54936826e-01 -8.13015163e-01 -1.14657235e+00
9.97083485e-02 -5.25216460e-01 6.62193894e-02 -1.34260118e+00
2.67029285e-01 -4.76227790e-01 -3.43744099e-01 1.69092029e-01
-4.06211942e-01 7.67322958e-01 2.46494740e-01 4.08270538e-01
-1.05025327e+00 -1.65302795e-03 9.75852907e-01 -1.25819728e-01
-2.04093665e-01 4.53792475e-02 3.29175144e-01 7.62198091e-01
3.15051645e-01 -2.95313090e-01 -1.50275007e-01 -7.81861469e-02
-2.45184019e-01 7.93343335e-02 6.90381348e-01 -1.38011289e+00
8.01191151e-01 -1.42180964e-01 4.77555722e-01 -8.80210817e-01
3.43688279e-01 -9.50812995e-01 6.12951398e-01 5.52632749e-01
-2.68768966e-01 -2.88385283e-02 -6.04953654e-02 8.03253949e-01
-4.12227720e-01 -1.02014445e-01 8.49492371e-01 -2.29466051e-01
-1.40340102e+00 4.57235932e-01 -9.23836678e-02 -3.20314378e-01
1.71383262e+00 -7.36370742e-01 2.85941780e-01 3.86326388e-02
-9.97723997e-01 4.95485775e-02 7.10114717e-01 2.20975265e-01
6.19907439e-01 -1.19126332e+00 -8.04597616e-01 1.68520629e-01
-1.56758819e-02 -1.97465420e-01 4.01206523e-01 1.21573639e+00
-7.60922551e-01 3.57921124e-01 -4.02656168e-01 -1.03904271e+00
-2.05275726e+00 5.76453805e-01 4.35081512e-01 -2.17720792e-01
-7.31163263e-01 1.06080055e+00 1.12830512e-01 3.79538745e-01
1.46474823e-01 -5.55500627e-01 7.60120600e-02 2.86199182e-01
4.17972952e-01 5.39007664e-01 -9.16602090e-02 -9.44714963e-01
-3.84049445e-01 1.18669343e+00 1.15219139e-01 -1.57664463e-01
1.13829386e+00 -3.94491732e-01 -1.42911062e-01 3.03824753e-01
1.24760163e+00 -1.19741924e-01 -1.78963900e+00 -4.49856043e-01
1.24098085e-01 -7.49468207e-01 -7.91470110e-02 -1.28381148e-01
-1.17452264e+00 4.99086976e-01 8.19189370e-01 -5.56309000e-02
1.16617966e+00 8.35273974e-03 5.76225877e-01 1.36342585e-01
6.02921307e-01 -1.14829826e+00 1.22560216e-02 3.68920714e-03
3.10482979e-01 -1.32788241e+00 3.50536078e-01 -6.44193947e-01
-3.32130432e-01 1.44500661e+00 2.65894651e-01 -2.72143275e-01
4.90922153e-01 1.86695218e-01 -1.26724839e-01 7.21970350e-02
-4.74638075e-01 -5.42547405e-01 5.66046894e-01 6.22784972e-01
9.89695415e-02 -1.76384941e-01 -1.77416056e-01 -4.27619100e-01
4.65537280e-01 2.33981326e-01 1.69421583e-01 8.03639233e-01
-4.50620979e-01 -1.05650151e+00 -5.63068628e-01 8.16430300e-02
-7.98527300e-02 1.00909978e-01 6.59733266e-02 7.05091894e-01
5.31177878e-01 6.01641238e-01 2.68209577e-01 4.93080840e-02
2.72953063e-02 -1.04624987e-01 7.01255441e-01 -1.26297966e-01
-5.70121706e-01 3.54875863e-01 -1.33034125e-01 -9.77212071e-01
-1.24328506e+00 -1.18653965e+00 -1.04266882e+00 2.07986057e-01
-4.58912045e-01 -1.62346572e-01 3.66756916e-01 8.43213081e-01
1.37815386e-01 4.13876057e-01 4.05632198e-01 -1.29719055e+00
-1.78694010e-01 -3.69410366e-01 -3.72727484e-01 6.39147222e-01
4.75968838e-01 -4.58024532e-01 -1.26547009e-01 8.52292478e-01] | [9.081554412841797, -0.3596699833869934] |
1af0594d-062c-4058-94e8-b2915f8a6cb1 | finding-the-needle-in-a-haystack-unsupervised | 2303.07991 | null | https://arxiv.org/abs/2303.07991v1 | https://arxiv.org/pdf/2303.07991v1.pdf | Finding the Needle in a Haystack: Unsupervised Rationale Extraction from Long Text Classifiers | Long-sequence transformers are designed to improve the representation of longer texts by language models and their performance on downstream document-level tasks. However, not much is understood about the quality of token-level predictions in long-form models. We investigate the performance of such architectures in the context of document classification with unsupervised rationale extraction. We find standard soft attention methods to perform significantly worse when combined with the Longformer language model. We propose a compositional soft attention architecture that applies RoBERTa sentence-wise to extract plausible rationales at the token-level. We find this method to significantly outperform Longformer-driven baselines on sentiment classification datasets, while also exhibiting significantly lower runtimes. | ['Marek Rei', 'Helen Yannakoudakis', 'Andrew Caines', 'Kamil Bujel'] | 2023-03-14 | null | null | null | null | ['document-classification'] | ['natural-language-processing'] | [ 4.08862233e-01 5.81414938e-01 -2.64769107e-01 -4.57916617e-01
-1.46063566e+00 -5.81785977e-01 7.05715299e-01 3.09223384e-01
-3.39548051e-01 5.68762422e-01 1.18044555e+00 -7.12082863e-01
2.40100071e-01 -3.65205556e-01 -7.81998813e-01 -2.72670031e-01
3.07332933e-01 3.07512164e-01 -1.94486499e-01 -2.75116891e-01
5.07896960e-01 -6.10891245e-02 -9.14155424e-01 1.15343857e+00
5.35787582e-01 5.37836730e-01 -2.14379039e-02 8.19809496e-01
-2.83481836e-01 1.08734524e+00 -6.15061283e-01 -8.21029663e-01
-4.82137278e-02 -6.42404035e-02 -9.49710369e-01 -2.26430297e-01
7.17969954e-01 -3.35205227e-01 -5.68022244e-02 5.16238809e-01
2.14173153e-01 -1.88438460e-01 7.61245251e-01 -6.01733506e-01
-9.37899411e-01 1.73525596e+00 -3.81608069e-01 4.34794843e-01
1.60039291e-01 2.95484483e-01 2.08522582e+00 -1.14402258e+00
5.05231678e-01 1.40249789e+00 6.93747461e-01 4.68335927e-01
-1.10233891e+00 -3.70194733e-01 7.60761619e-01 2.67923832e-01
-8.48425329e-01 -6.76566899e-01 6.77853584e-01 -3.03490669e-01
1.78143501e+00 1.20671920e-01 1.89963564e-01 1.44063318e+00
4.71716434e-01 1.11981082e+00 4.66421366e-01 -4.60794836e-01
-1.33819401e-01 -2.48798206e-01 5.68931103e-01 6.72346354e-01
4.34537768e-01 -4.13838118e-01 -9.13526058e-01 -1.52096391e-01
-7.11360648e-02 -2.83395708e-01 -5.97485155e-02 3.11881691e-01
-1.07601893e+00 9.77725446e-01 4.06901985e-01 4.33203638e-01
-4.87961441e-01 7.18988597e-01 6.68259025e-01 3.44717759e-03
6.00431085e-01 9.20381367e-01 -9.31917071e-01 -1.29507333e-01
-1.11637962e+00 3.25407654e-01 5.42258680e-01 8.12393963e-01
1.03278577e-01 3.90960604e-01 -6.09532118e-01 4.68969613e-01
4.39134449e-01 9.66299698e-02 6.21437490e-01 -5.26797414e-01
7.96081364e-01 3.42749894e-01 7.37614334e-02 -5.80196857e-01
-3.84585261e-01 -8.70673358e-01 -2.18737513e-01 -4.24094200e-01
2.52478451e-01 -2.18373612e-01 -7.66565382e-01 1.63999963e+00
-3.38463843e-01 -3.55272144e-01 1.61785990e-01 4.35419530e-01
7.56257415e-01 7.65185654e-01 2.63268650e-01 -2.46178024e-02
1.25170088e+00 -1.22620678e+00 -6.06646836e-01 -6.70771956e-01
1.05452180e+00 -6.97801471e-01 1.54316080e+00 5.17984271e-01
-1.33006287e+00 -4.65351343e-01 -1.25280881e+00 -7.06248462e-01
-5.79628050e-02 2.38784343e-01 5.68236947e-01 6.63772583e-01
-1.00200951e+00 6.77987635e-01 -6.98017716e-01 -5.57596609e-02
4.74573910e-01 -3.92845348e-02 1.67453513e-01 4.24021967e-02
-1.13783658e+00 1.10667348e+00 2.46722475e-01 1.15494914e-01
-6.91873074e-01 -1.08901036e+00 -8.48556697e-01 5.07785678e-01
1.32338807e-01 -7.38091946e-01 1.72277617e+00 -6.27301216e-01
-1.30474532e+00 5.47888875e-01 -4.86316293e-01 -6.13026321e-01
3.13107669e-01 -6.53158545e-01 7.80553147e-02 -3.44141304e-01
1.59359515e-01 5.19290686e-01 6.21085465e-01 -7.93672442e-01
-3.41538966e-01 -1.54292703e-01 -1.46900080e-02 1.65762767e-01
-4.65413243e-01 2.65130460e-01 4.06803824e-02 -8.27649415e-01
-4.10835207e-01 -6.05530202e-01 -3.56006891e-01 -4.74277735e-01
-6.87673271e-01 -6.36022508e-01 3.95015150e-01 -7.51872003e-01
1.26101410e+00 -1.93769431e+00 -1.28505304e-01 -7.36307055e-02
-8.04313794e-02 8.36887881e-02 -4.48761791e-01 6.24155819e-01
-2.71350682e-01 6.90471113e-01 -9.92894098e-02 -6.86066568e-01
2.16849789e-01 3.85995544e-02 -1.07235992e+00 1.12626180e-02
8.03362370e-01 1.36168218e+00 -7.46675253e-01 -2.10927695e-01
-2.04669163e-01 2.62968957e-01 -9.60476577e-01 -1.47990547e-02
-6.15722716e-01 -6.88850060e-02 -3.93280596e-01 2.53465652e-01
3.29902954e-03 -4.51850533e-01 3.15423280e-01 -1.83771372e-01
-7.70944133e-02 1.43037701e+00 -3.87811273e-01 1.66632795e+00
-7.19094753e-01 7.41305470e-01 -2.73551255e-01 -6.97778046e-01
3.41449022e-01 4.55004245e-01 -7.46855810e-02 -4.68360215e-01
7.92149603e-02 1.44216776e-01 3.08368057e-01 -2.93259740e-01
9.11268711e-01 -3.54578972e-01 -2.06134215e-01 6.20095193e-01
-4.64850888e-02 -2.08209246e-01 2.26414099e-01 5.41186571e-01
1.28412271e+00 2.21576571e-01 2.31537059e-01 -2.84877717e-01
3.06829900e-01 6.82603493e-02 1.22083314e-01 1.01827216e+00
3.87136787e-01 5.36583185e-01 7.74994433e-01 -2.47413129e-01
-1.07400131e+00 -6.78233504e-01 9.46574882e-02 1.43397093e+00
-6.81834698e-01 -1.01735580e+00 -4.63661522e-01 -8.77233982e-01
6.27425909e-02 1.44459057e+00 -6.48145795e-01 -4.08540070e-02
-7.21157789e-01 -7.05754817e-01 9.60529685e-01 9.32024717e-01
-3.43905210e-01 -8.42462540e-01 -3.30552340e-01 4.20457363e-01
-2.21354872e-01 -1.20087636e+00 -5.02823651e-01 5.67952335e-01
-8.70497286e-01 -8.25073481e-01 -2.96593070e-01 -2.92141169e-01
3.07238609e-01 -2.06541106e-01 1.32464933e+00 3.32876742e-01
1.25459120e-01 -1.01013422e-01 -3.26858282e-01 -5.34444332e-01
-5.20030856e-01 4.12539363e-01 -2.80302376e-01 -3.33916336e-01
2.25418150e-01 -2.58391529e-01 -1.75900936e-01 -4.14988905e-01
-7.29724228e-01 3.86205353e-02 7.97173262e-01 1.01558840e+00
1.94205076e-01 -4.36749458e-01 5.77451229e-01 -1.09634352e+00
1.00794435e+00 -3.19544911e-01 -1.71101727e-02 1.32440150e-01
-8.08012247e-01 7.41766751e-01 9.83399093e-01 5.60350297e-03
-1.14460945e+00 -3.77309561e-01 -4.85856384e-01 1.36726081e-01
1.61583647e-01 9.23543453e-01 -1.28140956e-01 8.19339037e-01
6.74006939e-01 -4.50555868e-02 -6.01110995e-01 -6.60554111e-01
7.84379900e-01 4.54447120e-01 3.95581365e-01 -8.24153244e-01
6.68191493e-01 2.24511087e-01 -3.48317385e-01 -5.69226205e-01
-1.52838206e+00 -2.81961352e-01 -3.04556578e-01 3.55161279e-01
7.29503572e-01 -8.75591934e-01 -2.38028586e-01 1.25791151e-02
-1.54627132e+00 -5.30668795e-01 -2.56674975e-01 1.28610581e-01
-2.31481686e-01 1.62974209e-01 -8.74849617e-01 -6.63907528e-01
-6.43262684e-01 -1.01122582e+00 1.18575442e+00 -3.88859361e-01
-7.83625901e-01 -1.00879574e+00 -1.60926759e-01 5.03344357e-01
3.40530634e-01 -2.99654394e-01 1.46989095e+00 -9.03210700e-01
-4.36992884e-01 5.01561584e-03 -1.75086856e-01 1.64903253e-01
-1.96614236e-01 1.77158937e-01 -1.25478256e+00 2.15106502e-01
-2.33985469e-01 -4.09171879e-01 1.36549926e+00 3.56084555e-01
1.21851122e+00 -9.08156872e-01 -1.45150557e-01 1.42842948e-01
1.06504846e+00 -3.05559754e-01 3.52938652e-01 1.78892747e-01
7.03615308e-01 5.76171994e-01 2.75941461e-01 2.95193344e-01
4.12835151e-01 4.32619780e-01 3.10505241e-01 1.34539008e-01
-4.04692829e-01 -6.37964070e-01 8.20497990e-01 8.97848189e-01
4.00619060e-01 -5.47492325e-01 -9.99398530e-01 6.18253708e-01
-1.75505710e+00 -1.11145175e+00 -4.88478422e-01 1.30293477e+00
1.07256699e+00 7.10290015e-01 -2.37498119e-01 1.10098645e-01
1.28603876e-01 3.94399881e-01 -2.90288091e-01 -9.07262444e-01
-1.81303948e-01 4.89366531e-01 4.22296017e-01 7.82870293e-01
-7.40364015e-01 1.12790167e+00 6.76824236e+00 5.89702010e-01
-8.95832837e-01 2.42811769e-01 6.62261784e-01 -4.40280467e-01
-1.04885375e+00 2.12785065e-01 -1.05281353e+00 1.05478756e-01
1.17213750e+00 -1.10820271e-01 5.64835109e-02 7.53203988e-01
4.16961044e-01 2.30525747e-01 -1.55220413e+00 4.14462417e-01
1.99482128e-01 -1.72734249e+00 4.81765181e-01 -1.16824158e-01
8.01694036e-01 2.71471262e-01 1.15787573e-01 4.35919881e-01
5.52190125e-01 -1.13928199e+00 1.30134666e+00 1.08771287e-01
3.09551090e-01 -5.43725133e-01 6.68424129e-01 3.25838506e-01
-8.28541815e-01 -2.77308822e-01 -1.74622253e-01 -4.85943973e-01
2.14756563e-01 6.65186703e-01 -1.18266952e+00 2.88860649e-01
2.53107905e-01 9.48572934e-01 -7.76405275e-01 4.80375797e-01
-7.76534796e-01 1.13854778e+00 -9.07293633e-02 -4.68227386e-01
6.48549259e-01 5.22239804e-01 4.34385061e-01 1.66626084e+00
6.26123101e-02 -2.45864615e-01 -1.91481352e-01 1.01948059e+00
-4.32705700e-01 1.32002711e-01 -3.67319882e-01 -4.10937309e-01
8.52783024e-02 1.04388380e+00 -3.72235447e-01 -5.00830352e-01
-3.40862930e-01 6.09380603e-01 4.20262069e-01 1.87723637e-01
-6.74565017e-01 -1.24736898e-01 4.87760425e-01 -6.41138256e-02
5.02213180e-01 -3.31135243e-01 -8.37495327e-01 -1.29160464e+00
9.21700001e-02 -8.47111046e-01 5.04422903e-01 -1.07680666e+00
-1.34827554e+00 4.08286780e-01 -3.46291035e-01 -4.39540058e-01
-4.31069046e-01 -8.55878413e-01 -8.85375857e-01 8.42202604e-01
-1.64715624e+00 -1.22059500e+00 4.13867056e-01 8.56718719e-02
1.09034657e+00 1.87452793e-01 7.80887604e-01 -2.57141352e-01
-5.60434580e-01 6.36785388e-01 3.82468775e-02 1.75054744e-01
6.52493060e-01 -1.26850772e+00 8.16643000e-01 1.24451554e+00
4.76841152e-01 1.10486650e+00 7.83943713e-01 -6.91587389e-01
-1.28996122e+00 -1.06750274e+00 1.38870442e+00 -1.03159535e+00
1.01865280e+00 -4.98110890e-01 -7.21234262e-01 9.37116146e-01
6.23763442e-01 -4.81079817e-01 7.69485712e-01 7.51076698e-01
-7.77639866e-01 1.82405129e-01 -3.72992933e-01 7.99305141e-01
8.59410703e-01 -7.34066308e-01 -1.23905110e+00 3.67390871e-01
9.10776615e-01 -3.54013182e-02 -3.90270114e-01 9.80444327e-02
5.55138290e-01 -4.97058302e-01 5.88702142e-01 -1.10752285e+00
1.20717335e+00 1.15996569e-01 -5.52084148e-02 -1.31610620e+00
-3.93939286e-01 -6.51182711e-01 -1.06972396e-01 1.16156256e+00
1.15374374e+00 -2.64317244e-01 5.38223743e-01 6.07223749e-01
-5.65031111e-01 -8.12424958e-01 -6.81389809e-01 -5.59732795e-01
5.91194868e-01 -9.14474726e-01 4.03300464e-01 6.09370053e-01
4.44595993e-01 9.70764160e-01 -2.00774968e-02 -1.25501037e-01
4.29099590e-01 1.00386359e-01 3.11259329e-01 -8.54086936e-01
-4.87233669e-01 -8.97936046e-01 3.35992754e-01 -1.07861733e+00
5.69470465e-01 -1.33583319e+00 3.05437356e-01 -1.82868850e+00
2.36847565e-01 -1.05767827e-02 -3.28042686e-01 9.08518255e-01
-4.68307704e-01 3.60256098e-02 1.12312764e-01 -4.40592431e-02
-4.46693808e-01 5.36083341e-01 7.11371720e-01 -2.47107640e-01
8.16474929e-02 -4.05689299e-01 -1.29703784e+00 8.50986123e-01
7.41119504e-01 -6.65981412e-01 -1.96948379e-01 -9.09388721e-01
8.58546197e-01 -3.23463768e-01 5.07791266e-02 -4.58007574e-01
-1.51207233e-02 -1.34577483e-01 4.02780801e-01 -6.84406519e-01
-2.94839349e-02 -1.25111058e-01 -6.39514029e-01 4.51558799e-01
-1.26527333e+00 2.24316597e-01 1.96265116e-01 4.40136909e-01
1.25502676e-01 -3.11612457e-01 4.63457793e-01 -2.66927958e-01
-2.37183928e-01 -2.01806456e-01 -7.70707965e-01 3.45115781e-01
8.20939988e-02 2.34915372e-02 -5.26041925e-01 -1.97301477e-01
-3.27596545e-01 2.91996589e-03 2.35173419e-01 6.82201862e-01
4.50667679e-01 -8.77223969e-01 -9.11324561e-01 5.19122789e-03
1.13073692e-01 -1.70522869e-01 -3.10867667e-01 7.37778842e-01
2.39046980e-02 1.08202243e+00 3.87239456e-01 -5.55041134e-02
-9.08835888e-01 2.88383991e-01 1.15733922e-01 -7.41145074e-01
-5.08853734e-01 1.22215867e+00 2.11979493e-01 -1.14030555e-01
1.00574538e-01 -9.87905264e-01 -9.93496180e-02 1.04512006e-01
4.62536722e-01 -1.50887936e-03 5.10921955e-01 -1.08338311e-01
-5.55734634e-01 1.71064541e-01 -4.09401625e-01 -3.11502904e-01
1.53331733e+00 1.37235582e-01 2.39538401e-02 4.01562601e-01
9.63848472e-01 6.67571306e-01 -1.08978760e+00 -7.79260173e-02
5.34980237e-01 1.38960704e-01 1.43070951e-01 -1.03878880e+00
-6.06873512e-01 1.23628461e+00 -4.95506287e-01 -1.32884178e-03
4.31098640e-01 7.79676735e-02 7.80915558e-01 6.31697655e-01
-3.14596504e-01 -9.30558562e-01 3.15786690e-01 1.03987491e+00
1.18753397e+00 -8.75294626e-01 -1.61078777e-02 3.19466516e-02
-6.82731152e-01 1.25391006e+00 5.06587207e-01 -1.23839155e-01
2.54779875e-01 5.84306300e-01 -1.32085547e-01 -1.18880466e-01
-1.49420524e+00 -6.57809451e-02 3.83916914e-01 -1.74459573e-02
1.03001857e+00 -1.46257535e-01 -3.06010813e-01 1.21010411e+00
-5.05407870e-01 -3.52293640e-01 7.65645027e-01 5.54630399e-01
-4.02929872e-01 -1.07028913e+00 -1.10041730e-01 5.38449645e-01
-9.19718802e-01 -1.01512218e+00 -7.13722110e-01 1.37058645e-01
-2.14462906e-01 1.17896700e+00 -1.27495855e-01 -1.15557261e-01
8.65623131e-02 6.63299799e-01 3.58077675e-01 -1.15221167e+00
-1.09578252e+00 1.00228963e-02 7.75017023e-01 -5.23756683e-01
-1.13921776e-01 -7.21425533e-01 -1.40465152e+00 -2.32008975e-02
-2.27804691e-01 2.77628422e-01 5.53378582e-01 1.17798328e+00
3.88194025e-01 7.42877603e-01 1.59661725e-01 -5.60254335e-01
-8.92646194e-01 -1.06589031e+00 1.55076191e-01 1.19670674e-01
4.14168596e-01 3.42631452e-02 -3.29000652e-01 7.46638626e-02] | [11.65208911895752, 8.93753719329834] |
01e1e148-c830-4339-adbf-35d6a6dd22a5 | few-shot-one-class-classification-via-meta-1 | 2007.04146 | null | https://arxiv.org/abs/2007.04146v2 | https://arxiv.org/pdf/2007.04146v2.pdf | Few-Shot One-Class Classification via Meta-Learning | Although few-shot learning and one-class classification (OCC), i.e., learning a binary classifier with data from only one class, have been separately well studied, their intersection remains rather unexplored. Our work addresses the few-shot OCC problem and presents a method to modify the episodic data sampling strategy of the model-agnostic meta-learning (MAML) algorithm to learn a model initialization particularly suited for learning few-shot OCC tasks. This is done by explicitly optimizing for an initialization which only requires few gradient steps with one-class minibatches to yield a performance increase on class-balanced test data. We provide a theoretical analysis that explains why our approach works in the few-shot OCC scenario, while other meta-learning algorithms fail, including the unmodified MAML. Our experiments on eight datasets from the image and time-series domains show that our method leads to better results than classical OCC and few-shot classification approaches, and demonstrate the ability to learn unseen tasks from only few normal class samples. Moreover, we successfully train anomaly detectors for a real-world application on sensor readings recorded during industrial manufacturing of workpieces with a CNC milling machine, by using few normal examples. Finally, we empirically demonstrate that the proposed data sampling technique increases the performance of more recent meta-learning algorithms in few-shot OCC and yields state-of-the-art results in this problem setting. | ['Hans-Georg Köpken', 'Denis Krompaß', 'Ahmed Frikha', 'Volker Tresp'] | 2020-07-08 | null | https://openreview.net/forum?id=B1ltfgSYwS | https://openreview.net/pdf?id=B1ltfgSYwS | null | ['one-class-classifier'] | ['methodology'] | [ 4.92499858e-01 -7.65248537e-02 -2.73883402e-01 -2.07452163e-01
-7.61397064e-01 6.55076131e-02 8.08270633e-01 4.08022165e-01
-4.06949371e-01 4.15714025e-01 -6.35286629e-01 -4.06181104e-02
-4.57296431e-01 -6.88563347e-01 -6.47250235e-01 -1.03624725e+00
-1.76963463e-01 7.85152078e-01 4.64093089e-01 -3.60164851e-01
2.91679472e-01 3.61343384e-01 -2.23954320e+00 2.93598920e-01
6.71961129e-01 1.22316420e+00 -6.96491748e-02 7.83623934e-01
2.32073516e-02 9.11055386e-01 -8.00524771e-01 -2.86557358e-02
2.42685482e-01 -5.19944251e-01 -6.05616271e-01 4.14664626e-01
3.41847837e-01 1.02785088e-01 1.78877056e-01 8.86025488e-01
3.40524226e-01 6.12848699e-01 8.50140214e-01 -1.46859920e+00
-2.43430823e-01 3.50662917e-01 -3.93845469e-01 4.20825750e-01
-8.92172903e-02 3.24570507e-01 7.21223354e-01 -9.56045747e-01
3.35123122e-01 8.59742343e-01 6.03629172e-01 5.09015679e-01
-1.20285571e+00 -2.39928558e-01 3.10137141e-02 6.05545819e-01
-9.87674534e-01 -1.88913718e-01 8.20067167e-01 -4.35128748e-01
1.10484016e+00 1.29672945e-01 7.56406784e-01 1.43130124e+00
4.78301138e-01 9.29074705e-01 1.02880061e+00 -7.35054493e-01
8.69998574e-01 4.96876128e-02 6.21281624e-01 4.71842378e-01
2.52217114e-01 2.82880098e-01 -3.37756902e-01 -7.69058913e-02
2.78660148e-01 4.29741949e-01 2.07937256e-01 -6.74350321e-01
-8.33452582e-01 9.22306299e-01 -1.75905988e-01 6.03903949e-01
-4.88369226e-01 1.23546012e-02 7.34110951e-01 8.38661849e-01
7.14704931e-01 6.36669159e-01 -4.33005899e-01 -3.39689493e-01
-8.42572510e-01 2.34150186e-01 1.09852290e+00 8.01985085e-01
7.98191845e-01 5.00811458e-01 -1.04047939e-01 9.62350368e-01
-3.82933766e-01 2.88139731e-01 9.70820546e-01 -6.43438578e-01
-4.70076501e-02 4.72408295e-01 4.00022380e-02 -3.86396945e-01
-3.82446349e-01 -5.10231495e-01 -6.63715243e-01 4.54399765e-01
4.42640960e-01 -1.90574545e-02 -1.23030615e+00 1.22663105e+00
1.93450108e-01 4.30907667e-01 1.47002861e-01 6.00919843e-01
1.26527861e-01 5.96152186e-01 -1.35917962e-01 -5.85653961e-01
1.08411396e+00 -1.01733994e+00 -7.42799044e-01 -3.04221332e-01
8.08857262e-01 -3.77853364e-01 1.33338821e+00 9.33314502e-01
-7.88706362e-01 -7.17952549e-01 -1.40748024e+00 5.14596283e-01
-7.60809064e-01 -2.75515765e-01 5.77653706e-01 7.71183074e-01
-2.24303514e-01 1.02382886e+00 -1.21934569e+00 -4.26700205e-01
3.37828726e-01 2.13215277e-01 2.32713427e-02 -7.25938454e-02
-9.36280549e-01 8.61478508e-01 7.17331946e-01 -2.43405893e-01
-1.11869812e+00 -7.06893027e-01 -9.37259674e-01 -6.68686209e-03
8.93622398e-01 -1.98075861e-01 1.43595028e+00 -9.19820130e-01
-1.62096906e+00 6.36663020e-01 1.47707775e-01 -9.27931070e-01
3.36110651e-01 -3.70714456e-01 -6.43727183e-01 1.13221943e-01
-2.52749920e-01 3.42822522e-02 1.20625460e+00 -1.04406786e+00
-7.97189713e-01 -2.97407150e-01 -2.23922521e-01 -2.55729049e-01
-4.51749861e-01 -3.51625621e-01 1.26416966e-01 -5.92178285e-01
-3.06090117e-02 -6.33738101e-01 -2.31821895e-01 -3.59384716e-01
-1.64097428e-01 -3.15323979e-01 1.11419046e+00 1.63051352e-01
9.58852112e-01 -2.24588752e+00 6.17639758e-02 1.72680378e-01
-2.74694979e-01 3.68008822e-01 -3.28758992e-02 4.35939789e-01
-2.99719423e-01 -4.62588370e-01 -4.71641093e-01 -3.65138501e-01
-4.46651392e-02 6.27522886e-01 -3.84894013e-01 5.71933031e-01
3.41239482e-01 7.19287634e-01 -1.07652736e+00 -2.41621464e-01
6.18583560e-01 -1.78659204e-02 -2.80143261e-01 2.30205357e-01
-4.04131144e-01 1.29829019e-01 4.05168161e-02 9.34604347e-01
1.90860137e-01 -2.55466193e-01 -1.28589675e-01 1.67916030e-01
1.02047883e-02 -3.63074899e-01 -1.22444773e+00 1.53636527e+00
-6.00399196e-01 3.62078041e-01 -3.80829930e-01 -1.61793482e+00
1.00750303e+00 3.06332260e-01 5.80533862e-01 -6.86827004e-01
3.21262330e-01 4.50798482e-01 6.81299418e-02 -6.28586292e-01
2.33941823e-01 -4.28550363e-01 -1.06669106e-01 3.85067433e-01
5.94703376e-01 -2.22752079e-01 4.70872045e-01 -3.00876468e-01
1.18416631e+00 -1.45762905e-01 5.97640932e-01 -1.84713483e-01
4.74483490e-01 1.31938949e-01 4.06613201e-01 1.17775667e+00
-2.42930979e-01 6.21116400e-01 3.69215429e-01 -6.38355732e-01
-1.10359514e+00 -1.04572105e+00 -3.18095177e-01 1.17790079e+00
1.57235280e-01 -2.85952181e-01 -5.78563809e-01 -7.05095589e-01
-3.12877037e-02 1.12846386e+00 -7.53358781e-01 -6.35322928e-01
-5.87904572e-01 -1.13091481e+00 8.18516836e-02 4.49381560e-01
4.86985333e-02 -1.16702223e+00 -1.01473331e+00 4.27005529e-01
4.24114734e-01 -8.76005292e-01 3.15227658e-01 8.82915318e-01
-1.06762528e+00 -1.34803331e+00 -6.61244869e-01 -4.87495512e-01
3.63301963e-01 -9.22152326e-02 9.21916485e-01 -1.45562649e-01
-7.33210444e-01 5.84744334e-01 -5.37071466e-01 -7.77656555e-01
-5.18746793e-01 1.78239033e-01 2.30679557e-01 3.41812342e-01
6.68704510e-01 -7.24025607e-01 -1.72064170e-01 3.15291137e-01
-9.03037488e-01 -5.92139423e-01 5.44996440e-01 1.34890926e+00
7.19949007e-01 2.82756984e-01 8.00888181e-01 -1.11244678e+00
3.53400558e-01 -6.74797118e-01 -5.36026597e-01 8.11168775e-02
-1.03973830e+00 -3.13087106e-02 9.36243355e-01 -8.55058134e-01
-8.26324761e-01 7.32013164e-03 -4.43350077e-02 -8.76074553e-01
-4.85402495e-01 2.45545387e-01 1.88233599e-01 7.92797878e-02
9.31500494e-01 3.38113964e-01 1.22198522e-01 -6.10684216e-01
1.01967968e-01 5.32506764e-01 7.80156970e-01 -6.38156354e-01
6.71147287e-01 4.81643468e-01 1.01937294e-01 -1.16479874e+00
-1.10332906e+00 -7.20345378e-01 -7.80114830e-01 -2.12239996e-01
5.66283464e-01 -4.87322599e-01 -3.70702803e-01 6.92776680e-01
-5.28842866e-01 -4.10541296e-01 -1.12801337e+00 5.62541604e-01
-1.06456995e+00 1.50166959e-01 -4.83488023e-01 -1.17513347e+00
-2.11537063e-01 -8.30639958e-01 1.01189458e+00 2.92294770e-02
-1.53195322e-01 -9.21872258e-01 2.97006637e-01 -2.54652560e-01
2.96326965e-01 3.88737261e-01 1.03142047e+00 -1.38826537e+00
-2.16631100e-01 -4.68012154e-01 5.59759378e-01 4.75278854e-01
1.14755079e-01 -1.30242616e-01 -1.24264407e+00 -5.35417914e-01
4.43631947e-01 -5.51828027e-01 1.06781363e+00 2.22201228e-01
1.27882564e+00 -9.18163627e-04 -1.65920928e-01 4.41360950e-01
1.40404570e+00 4.47416931e-01 5.31691313e-01 4.33164656e-01
4.03100342e-01 2.78549403e-01 1.14033544e+00 5.96745849e-01
-4.82857317e-01 5.57894170e-01 6.40177548e-01 1.78494036e-01
1.89154968e-01 2.25697160e-01 2.14456826e-01 6.01315498e-01
-5.89692257e-02 -1.29066736e-01 -8.07648242e-01 7.93473840e-01
-1.98432994e+00 -1.12351990e+00 1.42902836e-01 2.31349444e+00
5.16816914e-01 5.53446949e-01 2.94115752e-01 8.39909434e-01
5.56956589e-01 2.00728863e-01 -6.59747958e-01 -4.67310518e-01
-2.05035647e-03 6.99435592e-01 1.40851229e-01 1.25208110e-01
-1.48115206e+00 5.93061626e-01 6.09041834e+00 9.32063639e-01
-1.15333331e+00 3.88212144e-01 4.33416545e-01 -4.13288653e-01
4.69099402e-01 -2.72067964e-01 -7.98757613e-01 4.65561181e-01
1.50942254e+00 -1.45613968e-01 2.09801123e-01 1.37980199e+00
-3.17173123e-01 -1.76924303e-01 -1.44007957e+00 1.01958036e+00
2.74358064e-01 -1.19725025e+00 -2.20713943e-01 -8.02959129e-02
8.32087100e-01 1.30400043e-02 1.50908791e-02 8.28595817e-01
-1.31685838e-01 -6.19146168e-01 5.84719539e-01 4.32591766e-01
3.12574893e-01 -9.92056727e-01 9.51868117e-01 5.55655837e-01
-8.53669465e-01 -7.12982416e-01 -4.92086381e-01 -1.11830086e-01
7.23645836e-02 7.01005697e-01 -5.41261494e-01 6.38771415e-01
5.15784085e-01 7.50404954e-01 -5.63242614e-01 1.17716193e+00
2.51249641e-01 6.33750975e-01 -2.39214197e-01 -2.44081784e-02
3.77010077e-01 1.96521599e-02 7.26526797e-01 9.54709172e-01
3.17799658e-01 -1.87322125e-01 2.64433533e-01 5.67493618e-01
4.34498042e-01 -6.14498183e-03 -7.96884179e-01 -1.33217545e-02
2.07050011e-01 9.98155773e-01 -8.04109812e-01 -4.53346878e-01
-3.69150579e-01 8.87400746e-01 1.24643065e-01 9.89825949e-02
-6.26371205e-01 -7.15716958e-01 3.14353228e-01 -2.51944456e-02
5.43476045e-01 1.37014994e-02 -1.14089422e-01 -9.65298295e-01
-2.90769376e-02 -9.43996727e-01 6.87293887e-01 -3.16430986e-01
-1.55716240e+00 3.74775052e-01 3.48665178e-01 -1.57054794e+00
-8.35872769e-01 -9.01188254e-01 -8.47616792e-01 2.66192883e-01
-1.28291118e+00 -7.68891096e-01 -2.09518492e-01 4.87784088e-01
1.04409635e+00 -4.64228868e-01 9.31600749e-01 1.56212568e-01
-6.20504975e-01 4.80378687e-01 2.56905943e-01 -2.70561337e-01
5.02332509e-01 -1.45652974e+00 2.46780932e-01 8.21159124e-01
2.57971227e-01 3.44911009e-01 8.55077744e-01 -4.41134572e-01
-1.43605602e+00 -1.08823192e+00 2.68469900e-01 -3.16016793e-01
7.93200314e-01 -2.98556596e-01 -1.26212645e+00 5.04725039e-01
-1.36910498e-01 3.97997469e-01 4.99394327e-01 9.05651003e-02
-5.52068055e-02 -2.19237328e-01 -1.13387275e+00 3.42649043e-01
7.35648036e-01 -3.27577382e-01 -9.80525613e-01 5.18537700e-01
4.03945237e-01 -3.00005555e-01 -8.99053514e-01 5.27554154e-01
3.12101662e-01 -9.96284306e-01 9.21851814e-01 -7.71143913e-01
2.71120280e-01 4.31947149e-02 -2.07623333e-01 -1.38352668e+00
1.39200717e-01 -4.86784816e-01 -9.62770700e-01 8.18000913e-01
1.00674324e-01 -7.20944881e-01 6.85716391e-01 7.73121342e-02
-4.68849212e-01 -9.89090979e-01 -1.05276883e+00 -1.50208211e+00
-7.95306489e-02 -5.26955783e-01 3.18454981e-01 8.59130859e-01
-4.88383211e-02 1.33425027e-01 -4.77723449e-01 -7.37901181e-02
8.21277738e-01 2.50834823e-01 7.08583832e-01 -1.68576419e+00
-5.36370993e-01 -2.52808660e-01 -6.39382660e-01 -4.11757320e-01
1.53243795e-01 -6.56814396e-01 4.78686616e-02 -7.82436490e-01
-3.11989542e-02 -1.34082422e-01 -6.16394758e-01 5.37490666e-01
1.46359969e-02 2.29870304e-01 -8.09891224e-02 6.95483088e-02
-5.69332361e-01 6.16385937e-01 6.88796520e-01 -2.83035576e-01
-2.79869199e-01 3.32976669e-01 -2.61016935e-02 6.87665045e-01
7.28826880e-01 -4.87077832e-01 -5.84492445e-01 3.98826480e-01
-2.51556635e-01 -2.56288737e-01 4.86503482e-01 -1.47997832e+00
2.25079656e-01 -1.36635840e-01 3.72518659e-01 -8.08737516e-01
5.60642838e-01 -9.37388539e-01 -1.62041456e-01 7.87010193e-01
-1.91929996e-01 -1.72220051e-01 1.42448127e-01 8.50326359e-01
-1.77237853e-01 -7.52894521e-01 1.12707818e+00 -2.20542237e-01
-1.04989100e+00 1.51667088e-01 -3.82878333e-01 1.16738617e-01
1.23194969e+00 -4.31166410e-01 -2.50044733e-01 -6.77806092e-03
-1.03144991e+00 4.33878638e-02 2.45926708e-01 5.10521889e-01
4.56731975e-01 -1.22707486e+00 -3.56496155e-01 6.34580851e-01
4.94087517e-01 -1.91929802e-01 1.74884245e-01 1.01022208e+00
-7.79142231e-02 2.79067457e-01 -2.82902271e-01 -1.10927200e+00
-9.82154369e-01 8.77892554e-01 4.47205007e-01 -2.68693507e-01
-1.03398252e+00 3.29652190e-01 -3.70669752e-01 -2.92166471e-01
4.33283955e-01 -3.19287956e-01 4.89317987e-04 3.17052543e-01
7.18554497e-01 6.65781915e-01 5.44911444e-01 2.53243987e-02
-5.17570674e-02 3.24998230e-01 -3.33848804e-01 7.53942281e-02
1.44684148e+00 3.09926659e-01 4.38854098e-01 1.36553979e+00
1.11201096e+00 -7.01990664e-01 -1.18226230e+00 -2.03463748e-01
3.67407918e-01 -4.21098560e-01 -1.03041217e-01 -5.50060570e-01
-7.42315829e-01 9.63113308e-01 8.75417411e-01 5.30199409e-01
1.06292927e+00 -1.16824701e-01 5.62933326e-01 9.63523090e-01
5.27492702e-01 -1.41599953e+00 5.05832553e-01 6.68972611e-01
4.53899324e-01 -1.41058576e+00 -1.00642681e-01 2.77328230e-02
-4.15024161e-01 1.27218521e+00 7.26753592e-01 -2.62117982e-01
7.05639601e-01 1.73241034e-01 -1.82672694e-01 -3.71509641e-01
-1.01829481e+00 -2.94225514e-01 8.65311325e-02 5.56421816e-01
-9.06135887e-02 -2.81540453e-01 -1.42335109e-02 5.03951848e-01
1.27706230e-01 2.66552866e-02 4.89745557e-01 1.43304539e+00
-6.92531228e-01 -9.64894235e-01 -1.63673371e-01 6.03986144e-01
-4.52982843e-01 2.99952805e-01 3.33756804e-02 1.18967116e+00
-1.06729448e-01 7.41994083e-01 4.49375659e-01 -2.24383682e-01
4.87533361e-01 7.29914784e-01 5.13342798e-01 -8.66463721e-01
-5.85923851e-01 1.31289372e-02 -1.00920781e-01 -6.63078606e-01
-2.72386342e-01 -5.93886554e-01 -9.93384361e-01 1.42953381e-01
-3.31134230e-01 1.99092478e-01 4.94040251e-01 1.24516940e+00
-4.24028337e-02 7.66524971e-01 8.38969588e-01 -1.11099780e+00
-1.05392802e+00 -9.47189271e-01 -9.37866867e-01 4.66691494e-01
4.70462382e-01 -9.97358918e-01 -7.88973153e-01 -1.87732115e-01] | [9.911348342895508, 3.124608278274536] |
06b277cb-b732-4de2-91b8-046d93527071 | video-inpainting-by-jointly-learning-temporal | 1806.08482 | null | http://arxiv.org/abs/1806.08482v2 | http://arxiv.org/pdf/1806.08482v2.pdf | Video Inpainting by Jointly Learning Temporal Structure and Spatial Details | We present a new data-driven video inpainting method for recovering missing
regions of video frames. A novel deep learning architecture is proposed which
contains two sub-networks: a temporal structure inference network and a spatial
detail recovering network. The temporal structure inference network is built
upon a 3D fully convolutional architecture: it only learns to complete a
low-resolution video volume given the expensive computational cost of 3D
convolution. The low resolution result provides temporal guidance to the
spatial detail recovering network, which performs image-based inpainting with a
2D fully convolutional network to produce recovered video frames in their
original resolution. Such two-step network design ensures both the spatial
quality of each frame and the temporal coherence across frames. Our method
jointly trains both sub-networks in an end-to-end manner. We provide
qualitative and quantitative evaluation on three datasets, demonstrating that
our method outperforms previous learning-based video inpainting methods. | ['Xiaoguang Han', 'Haibin Huang', 'Chuan Wang', 'Jue Wang'] | 2018-06-22 | null | null | null | null | ['video-inpainting'] | ['computer-vision'] | [ 1.19740620e-01 1.82737067e-01 -1.62160039e-01 -2.93678045e-01
-9.08341765e-01 -1.90743238e-01 3.82142544e-01 -6.33175015e-01
-2.27220029e-01 6.19198561e-01 5.69704175e-01 -2.47417223e-02
1.72913015e-01 -6.76211774e-01 -1.29434586e+00 -3.54699463e-01
-2.16427997e-01 6.49229065e-02 2.21471578e-01 3.53754684e-02
2.35942416e-02 6.33092701e-01 -1.16868556e+00 6.62947059e-01
3.70743781e-01 1.11734676e+00 2.06988096e-01 9.13336396e-01
1.64802596e-01 1.49937320e+00 -1.70907795e-01 1.31601930e-01
5.19375861e-01 -5.08060932e-01 -8.45699489e-01 4.21980858e-01
6.66290581e-01 -1.33729696e+00 -1.07242656e+00 5.71486294e-01
1.76810578e-01 1.96897000e-01 3.12284499e-01 -6.69815481e-01
-7.17425704e-01 3.15823257e-01 -7.21655905e-01 4.03647155e-01
5.74094057e-01 3.23196828e-01 7.31319785e-01 -1.17741537e+00
8.70736659e-01 1.41897106e+00 8.61082077e-01 6.11538768e-01
-1.34544444e+00 -4.05138910e-01 1.37680396e-01 -9.03704464e-02
-1.18382227e+00 -5.40335000e-01 9.96457815e-01 -4.72556829e-01
1.15806270e+00 -2.74386436e-01 7.09839225e-01 1.01511312e+00
2.25953713e-01 7.08147943e-01 7.29020000e-01 2.00891234e-02
8.83914903e-02 -7.62249589e-01 -6.91377282e-01 9.03622389e-01
-2.11834177e-01 6.10755205e-01 -8.64262879e-01 2.44574308e-01
1.76505733e+00 4.09008026e-01 -3.23351622e-01 -2.48031124e-01
-1.13348305e+00 4.71314967e-01 6.08899772e-01 7.31721967e-02
-6.06237769e-01 6.32174253e-01 2.06914261e-01 3.88111740e-01
8.59395981e-01 1.13285147e-01 -3.48982424e-01 5.96672371e-02
-1.45044959e+00 3.62025440e-01 2.19620407e-01 1.01942515e+00
7.63123393e-01 2.96905488e-01 -1.79359347e-01 4.88649070e-01
2.15166956e-01 1.00047894e-01 4.46144752e-02 -1.65079355e+00
3.70325655e-01 2.80237764e-01 4.09096181e-01 -8.41824114e-01
-1.11017518e-01 -2.31604185e-02 -9.84391391e-01 5.53424656e-01
3.61584574e-02 -6.94518210e-03 -9.12867546e-01 1.58676279e+00
2.08327368e-01 7.09972024e-01 -1.50154650e-01 1.16346252e+00
9.18621361e-01 7.84636796e-01 -1.32500827e-01 -2.40922838e-01
9.97217834e-01 -1.29928887e+00 -7.48580098e-01 -4.32797670e-02
8.44379440e-02 -5.92430949e-01 6.57234251e-01 8.60561579e-02
-1.88837552e+00 -7.24162936e-01 -1.02043712e+00 -8.39463413e-01
3.27701688e-01 -2.37388283e-01 3.75832170e-01 -3.12140346e-01
-1.43522727e+00 9.04595435e-01 -9.73955691e-01 1.72658369e-01
8.82850170e-01 -1.33879893e-02 -7.38744617e-01 -3.52104723e-01
-9.11677539e-01 5.96582294e-01 1.30591929e-01 1.01039939e-01
-1.63288367e+00 -1.07992589e+00 -1.13368368e+00 9.96249840e-02
7.43241906e-02 -1.09767628e+00 1.33483422e+00 -9.32633758e-01
-1.51605225e+00 9.21963871e-01 -2.17194051e-01 -5.70983231e-01
7.18917191e-01 -3.86137098e-01 7.52905086e-02 7.08817363e-01
3.81601155e-01 9.24873352e-01 1.17382276e+00 -1.25670981e+00
-7.04316914e-01 4.39504907e-02 4.87918034e-02 1.59003600e-01
3.99632275e-01 -8.82308930e-02 -7.70469725e-01 -1.08159912e+00
-5.87118454e-02 -2.90186673e-01 -1.50941476e-01 7.35342443e-01
-3.01091187e-02 2.36061424e-01 9.85431612e-01 -1.13452244e+00
9.39696789e-01 -2.15392995e+00 5.12759984e-01 -4.44161862e-01
6.02608979e-01 9.30058863e-03 -3.46903473e-01 2.39927337e-01
-2.24780828e-01 -8.46570209e-02 -5.43161571e-01 -7.85033166e-01
-4.83623058e-01 1.54355660e-01 -5.74262500e-01 3.72344673e-01
6.80021286e-01 1.18303657e+00 -1.04197359e+00 -2.38852680e-01
5.08207917e-01 1.08821034e+00 -7.73044288e-01 8.45996320e-01
-3.61036479e-01 6.82286978e-01 -1.76342770e-01 6.94110274e-01
7.46457160e-01 -3.38756531e-01 -4.90446612e-02 -4.53418285e-01
-2.85642266e-01 2.76573867e-01 -6.35190904e-01 2.42158580e+00
-4.60689902e-01 8.31508577e-01 4.11567122e-01 -7.50462413e-01
4.88773972e-01 4.29854125e-01 7.21630931e-01 -8.75987887e-01
9.45989322e-03 1.51848681e-02 -7.29635119e-01 -6.49878740e-01
4.69979167e-01 -2.71606535e-01 2.32608214e-01 3.75203937e-01
2.28423253e-01 -1.15883574e-01 -6.16892837e-02 2.48229161e-01
1.27103913e+00 7.93240786e-01 -1.98586419e-01 1.44983912e-02
7.43658543e-02 -2.81671733e-01 5.67841351e-01 3.09624791e-01
-7.86827281e-02 1.26885891e+00 4.98516113e-01 -1.08101356e+00
-1.59213746e+00 -1.14412284e+00 2.34818369e-01 6.91572845e-01
1.42161652e-01 -2.74046361e-01 -6.27835989e-01 -4.67974603e-01
-1.71764076e-01 1.74465299e-01 -8.82898450e-01 -1.30981073e-01
-8.85794997e-01 1.86307266e-01 3.06091309e-01 6.45165443e-01
7.37266302e-01 -1.21380126e+00 -7.22408473e-01 4.08618838e-01
-4.63917553e-01 -1.41864550e+00 -8.35811973e-01 2.82238945e-02
-1.17189455e+00 -1.25357890e+00 -8.38140070e-01 -9.64206636e-01
5.36468267e-01 5.57037711e-01 1.42284250e+00 3.16640228e-01
-1.62728786e-01 2.01598182e-01 -1.98719800e-01 3.41697991e-01
-3.99302989e-01 -3.37005496e-01 -2.46888310e-01 -1.36756718e-01
-2.06348464e-01 -1.10508168e+00 -1.00964212e+00 6.44714311e-02
-1.27307057e+00 5.27765155e-01 6.01640165e-01 8.88094366e-01
7.77350128e-01 -1.79679319e-02 8.09476152e-02 -4.07060564e-01
1.37834907e-01 -4.38275158e-01 -5.13675869e-01 -2.03743801e-01
-1.58242844e-02 1.00168355e-01 5.03820896e-01 -2.07392246e-01
-9.69422400e-01 1.26826912e-01 -2.58431166e-01 -1.17595708e+00
3.61052752e-02 1.54162869e-01 6.83188587e-02 -2.02336628e-02
2.94883281e-01 3.99420857e-01 2.64324039e-01 -5.71502090e-01
2.51036048e-01 -5.53559698e-02 8.78874421e-01 -4.02261436e-01
8.00614476e-01 7.49785244e-01 -5.65261468e-02 -4.89496380e-01
-1.09286106e+00 3.45918871e-02 -8.04942310e-01 -2.97329277e-01
1.06329441e+00 -1.55712676e+00 -5.10561228e-01 3.92907470e-01
-1.39006209e+00 -7.88433135e-01 -5.29835463e-01 2.99760878e-01
-9.56126809e-01 3.36928964e-01 -1.13602722e+00 -2.17450827e-01
-4.21551764e-01 -1.02519441e+00 1.60244548e+00 -5.39267026e-02
7.66813308e-02 -8.59766722e-01 1.05341941e-01 2.75894046e-01
3.11273843e-01 6.16116583e-01 6.27488434e-01 5.25300503e-01
-1.15687788e+00 3.52400333e-01 -5.54935575e-01 3.57715249e-01
1.32478431e-01 -6.16692975e-02 -8.27545643e-01 -3.45065325e-01
5.78005686e-02 -3.64506334e-01 1.15879416e+00 7.97392309e-01
1.16122603e+00 -3.78862619e-01 8.93011615e-02 1.14101517e+00
1.47516751e+00 -7.84725249e-02 8.92750263e-01 1.15828030e-01
9.92912352e-01 3.18488181e-01 4.05363172e-01 3.46525043e-01
3.74499202e-01 4.06974792e-01 7.17449903e-01 -4.43907470e-01
-6.43386304e-01 -5.34955382e-01 3.68552417e-01 3.85741204e-01
-2.52646804e-01 4.65683788e-02 -3.85643840e-01 5.98148704e-01
-1.99312842e+00 -1.30471051e+00 4.01348144e-01 1.93097389e+00
1.00243044e+00 -5.07629551e-02 1.22188292e-01 -1.35504633e-01
5.15372634e-01 5.02498448e-01 -6.66018486e-01 -2.41066571e-02
-3.69558297e-02 1.56849712e-01 -8.19204943e-05 9.45562899e-01
-9.85346913e-01 9.39123511e-01 6.75930166e+00 4.09165770e-01
-8.73570502e-01 6.10349774e-02 1.01468694e+00 -5.28376937e-01
-3.64421159e-01 -1.49655208e-01 -2.19882458e-01 4.15909231e-01
6.53087080e-01 2.87040949e-01 6.07357144e-01 5.00278294e-01
6.47687614e-01 1.13729775e-01 -1.33340788e+00 1.22396886e+00
-1.34571552e-01 -2.08799887e+00 3.31580222e-01 -1.21383322e-02
8.96049857e-01 5.14084697e-02 -1.37725502e-01 1.06560573e-01
1.23631060e-01 -1.29135060e+00 1.02036166e+00 7.30780065e-01
1.20087004e+00 -1.00129032e+00 2.71620750e-01 1.10890232e-01
-1.32653713e+00 -2.78257802e-02 -2.33727634e-01 -2.31259003e-01
5.39412200e-01 6.58221185e-01 -5.09497942e-04 3.49812567e-01
9.90823984e-01 1.20914519e+00 -3.31999511e-02 5.81819475e-01
-1.54146627e-01 1.38157979e-01 6.19914047e-02 1.02207208e+00
2.71323085e-01 -1.53962806e-01 5.56813121e-01 1.08745825e+00
3.25618088e-01 3.48418534e-01 1.22888703e-02 1.13373423e+00
-4.92231518e-01 -7.23796964e-01 -7.86924541e-01 1.21592186e-01
3.27906221e-01 1.07214344e+00 -3.31195772e-01 -4.16049600e-01
-5.39438665e-01 1.33121085e+00 6.14902973e-01 4.82377738e-01
-8.61310065e-01 -3.17791216e-02 8.93282592e-01 4.46415305e-01
6.44013762e-01 -3.49194109e-01 -5.29756248e-02 -1.28788531e+00
1.21850498e-01 -7.23891735e-01 6.14394918e-02 -1.14807999e+00
-1.13426161e+00 6.84124827e-01 -2.52002507e-01 -1.37310624e+00
-3.43249053e-01 -1.98166817e-01 -4.95940417e-01 8.54505122e-01
-1.81705534e+00 -1.19242930e+00 -4.78549182e-01 7.77792513e-01
8.97765458e-01 -1.01447748e-02 4.06432003e-01 4.61694300e-01
-3.58098924e-01 2.07126617e-01 -3.90798718e-01 3.41808945e-01
5.89326322e-01 -8.86933267e-01 5.62961280e-01 1.06516862e+00
-3.00451934e-01 3.40985894e-01 4.57322985e-01 -6.33196890e-01
-1.56349719e+00 -1.54844522e+00 7.82626987e-01 -2.81419933e-01
1.56859249e-01 -6.54316694e-02 -9.36380684e-01 9.05039489e-01
3.87992978e-01 6.11942291e-01 6.93339556e-02 -4.89091486e-01
-6.24027431e-01 -1.81104895e-03 -1.24734354e+00 5.33208668e-01
1.32691789e+00 -8.03940177e-01 -4.49065924e-01 8.25476199e-02
1.15669107e+00 -7.21963286e-01 -1.00083089e+00 3.31927925e-01
4.78423297e-01 -1.34790957e+00 1.21136546e+00 -3.52793306e-01
1.32488108e+00 -5.44810891e-01 -6.20046109e-02 -9.20283973e-01
-6.05944514e-01 -9.01529312e-01 -7.87650228e-01 7.11510122e-01
-1.09580278e-01 1.63476571e-01 7.45164454e-01 5.25793076e-01
-3.01758975e-01 -9.08845961e-01 -7.96598673e-01 -2.95229524e-01
-9.09373239e-02 -3.30526173e-01 4.30224031e-01 8.55280876e-01
-5.50371408e-01 3.51944745e-01 -7.31161296e-01 1.33957162e-01
8.29371691e-01 1.92776263e-01 4.97026265e-01 -6.95454478e-01
-2.92251378e-01 -3.13754588e-01 -1.48412332e-01 -1.52785695e+00
7.57179707e-02 -4.47158039e-01 9.82772335e-02 -1.52294958e+00
3.60895842e-01 1.93509713e-01 -7.13470131e-02 3.89311254e-01
-1.42050497e-02 5.93964338e-01 -5.87241948e-02 1.99521422e-01
-6.45524979e-01 7.53805995e-01 1.58613753e+00 -6.29449263e-03
-2.00584501e-01 -4.40572917e-01 -5.53593159e-01 5.69558382e-01
3.84183586e-01 -3.10397863e-01 -5.28501749e-01 -9.29638743e-01
7.73510486e-02 7.31175601e-01 8.19821775e-01 -8.79158258e-01
-3.25924158e-02 -4.87137474e-02 9.29530501e-01 -7.64884710e-01
3.80596131e-01 -8.88782203e-01 1.69909760e-01 3.14799249e-01
-3.94074053e-01 1.01237223e-01 2.02443928e-01 6.35461271e-01
-3.74109179e-01 5.40531099e-01 1.11227190e+00 -3.35458398e-01
-6.42020345e-01 1.00319433e+00 -1.45437360e-01 1.03636302e-01
6.96589172e-01 -3.02618206e-01 1.55877054e-01 -5.14912844e-01
-6.74090922e-01 2.58888509e-02 6.75914466e-01 4.20971930e-01
1.12286031e+00 -1.57743454e+00 -8.00857544e-01 3.44912320e-01
-3.32876623e-01 7.63142228e-01 5.82680345e-01 6.12909973e-01
-9.52700317e-01 7.41960332e-02 -4.36367810e-01 -6.88597143e-01
-8.11493933e-01 6.23536706e-01 5.41553795e-01 -2.58092582e-01
-1.20953596e+00 7.95802891e-01 4.50072348e-01 -1.29967285e-02
3.36300522e-01 -4.12804812e-01 1.90225869e-01 -4.44815487e-01
9.10217464e-01 1.05646439e-01 -2.79003501e-01 -4.57728982e-01
-4.93936017e-02 5.82097113e-01 -1.23588555e-01 -1.48250848e-01
1.77508616e+00 -3.25043678e-01 -1.59324169e-01 1.07630394e-01
1.43997717e+00 -4.32086378e-01 -2.47021818e+00 -3.24654847e-01
-5.37073255e-01 -6.51848853e-01 2.45782375e-01 -3.49990189e-01
-1.42021179e+00 7.11839437e-01 2.09806785e-01 -2.10180461e-01
1.33794045e+00 -1.84251994e-01 1.12491858e+00 -5.37469983e-02
2.21458152e-01 -8.21256638e-01 4.54903960e-01 5.91720760e-01
1.19361567e+00 -1.24142206e+00 1.23322681e-01 -1.69350669e-01
-2.61697471e-01 1.11025167e+00 5.94040871e-01 -6.89019501e-01
6.10475540e-01 5.00803769e-01 -2.03902259e-01 -2.14703470e-01
-1.13418543e+00 1.56414255e-01 1.80432305e-01 6.69771910e-01
5.32317936e-01 -4.83681619e-01 2.32887253e-01 2.71927714e-01
2.83322692e-01 3.45291436e-01 3.38245690e-01 9.09668982e-01
-2.40010664e-01 -6.45579100e-01 -6.11051638e-03 -5.11662662e-02
-3.35768133e-01 -1.35282129e-01 -8.47797990e-02 6.88963115e-01
1.64987352e-02 6.77119493e-01 3.14279795e-01 -2.81239569e-01
2.08496585e-01 -3.65806878e-01 7.41648018e-01 -3.43915999e-01
-3.14129204e-01 3.50808412e-01 -1.01719230e-01 -1.26745677e+00
-5.85625768e-01 -1.89429656e-01 -1.27775848e+00 -6.45847082e-01
5.21474779e-01 -2.71551549e-01 1.61689594e-01 7.58261263e-01
7.29187071e-01 8.69141281e-01 8.24638367e-01 -1.70183015e+00
1.31868184e-01 -7.50105083e-01 -4.27765101e-01 4.15473759e-01
1.03943384e+00 -3.78932744e-01 -1.30080432e-01 5.27964592e-01] | [10.797579765319824, -1.3117905855178833] |
6f5279e2-1d4e-4ddf-9863-1d6360f6f811 | acoustic-echo-cancellation-with-cross-domain | null | null | https://www.isca-speech.org/archive/pdfs/interspeech_2021/pfeifenberger21_interspeech.pdf | https://www.isca-speech.org/archive/pdfs/interspeech_2021/pfeifenberger21_interspeech.pdf | Acoustic Echo Cancellation with Cross-Domain Learning | This paper proposes the Cross-Domain Echo-Controller(CDEC), submitted to the Interspeech 2021 AEC-Challenge.The algorithm consists of three building blocks: (i) a Time-Delay Compensation (TDC) module, (ii) a frequency-domainblock-based Acoustic Echo Canceler (AEC), and (iii) a Time-Domain Neural-Network (TD-NN) used as a post-processor.Our system achieves an overall MOS score of 3.80, while onlyusing 2.1 million parameters at a system latency of 32ms. | ['Franz Pernkopf', 'Matthias Zoehrer', 'Lukas Pfeifenberger'] | 2021-08-30 | null | null | null | interspeech-2021-8 | ['acoustic-echo-cancellation', 'acoustic-echo-cancellation'] | ['medical', 'speech'] | [ 2.54815698e-01 -2.44428858e-01 3.18702430e-01 -2.01392770e-01
-1.45342255e+00 -4.67807889e-01 3.76635045e-01 -6.86101764e-02
-6.62143648e-01 2.87812591e-01 3.73404205e-01 -5.71152747e-01
1.89546585e-01 7.87017420e-02 -6.09064341e-01 -4.25476521e-01
-4.86002564e-01 2.01243777e-02 4.45392460e-01 -1.03727810e-01
-8.31721574e-02 4.68967438e-01 -1.39025009e+00 1.57109886e-01
7.49092877e-01 1.25641239e+00 -9.60954577e-02 1.39529240e+00
5.12136340e-01 3.16271842e-01 -8.28248441e-01 -2.07273856e-01
4.51204032e-02 -3.49359542e-01 -7.09097832e-02 -7.97285259e-01
4.89011228e-01 -4.68850940e-01 -4.04232204e-01 8.47239852e-01
9.62826610e-01 3.57169271e-01 4.53937531e-01 -1.07213426e+00
3.80410165e-01 6.07273817e-01 -2.65406221e-01 5.74774034e-02
2.88533807e-01 8.35789442e-02 7.00717390e-01 -1.42064083e+00
1.35253593e-01 1.20787072e+00 9.83168781e-01 5.37213445e-01
-1.01699615e+00 -1.12725079e+00 -4.23888862e-01 2.85461873e-01
-1.40851605e+00 -1.28329015e+00 9.15976048e-01 -2.78831669e-03
1.43109584e+00 2.81117886e-01 5.44995785e-01 1.14929521e+00
3.99905533e-01 5.69354773e-01 8.99137855e-01 -7.16665983e-01
4.70033944e-01 -3.74360442e-01 1.03022300e-01 1.32777289e-01
-3.63098741e-01 9.18002129e-01 -9.89640653e-01 -4.65527713e-01
3.57876509e-01 -8.17634583e-01 -2.48567149e-01 3.28980684e-01
-8.01527679e-01 3.24910223e-01 -1.82800680e-01 -1.67113915e-02
-5.44737518e-01 7.79807687e-01 7.66309261e-01 5.97790778e-01
2.72747338e-01 1.31113961e-01 -4.91298229e-01 -6.02945149e-01
-1.31221557e+00 3.54315072e-01 1.25520802e+00 7.46374667e-01
2.33565748e-01 9.68086600e-01 2.81035900e-01 1.09167564e+00
4.96623427e-01 1.23805559e+00 5.61173379e-01 -1.23439789e+00
4.22022581e-01 -8.42984438e-01 1.47281110e-01 -1.10530186e+00
-4.87909347e-01 -3.65238070e-01 -8.66938412e-01 1.32429563e-02
-8.59975219e-02 -7.15933740e-01 -7.74032712e-01 1.80279171e+00
9.17115137e-02 5.58288455e-01 -8.89317840e-02 1.03543448e+00
3.14240932e-01 9.25682962e-01 -3.40027474e-02 -2.46185333e-01
1.00249398e+00 -9.56990957e-01 -1.28169405e+00 -3.29181761e-01
3.75479400e-01 -1.11749172e+00 4.03181046e-01 8.87544334e-01
-1.56628859e+00 -7.93536425e-01 -1.55561638e+00 2.67088830e-01
-1.06701232e-01 -1.08591437e-01 -9.64889769e-03 7.78467655e-01
-1.40538228e+00 4.85362291e-01 -7.82254815e-01 2.36618951e-01
-7.50733316e-01 4.98038620e-01 1.07838817e-01 5.67695439e-01
-1.56638980e+00 4.69216406e-01 -5.76636903e-02 1.50719345e-01
-8.06967318e-01 -8.22162330e-01 -7.84652352e-01 9.66118351e-02
2.57948965e-01 -5.32400645e-02 1.75146818e+00 -8.64351511e-01
-2.07997251e+00 4.69182879e-01 -3.04572374e-01 -7.63280571e-01
2.31029838e-01 -8.73212934e-01 -1.42630017e+00 2.31137320e-01
-2.97807068e-01 5.95324099e-01 1.36752391e+00 -1.04125881e+00
-9.36222136e-01 1.10680126e-01 -8.25121760e-01 -4.21547368e-02
5.05764857e-02 7.59232566e-02 -8.22609961e-01 -1.18704939e+00
1.50713667e-01 -1.19291365e+00 -1.10435672e-01 -3.01137328e-01
-5.24801433e-01 1.47070184e-01 6.73303545e-01 -1.10628557e+00
1.96443212e+00 -2.40873003e+00 -1.73924968e-01 5.29316068e-01
1.77618340e-02 8.11554432e-01 -4.25327331e-01 5.70623040e-01
-3.80616486e-01 -3.13265890e-01 -5.48664182e-02 -5.70727170e-01
1.68609545e-01 -4.48108435e-01 -2.87168294e-01 3.91885489e-01
3.61713879e-02 5.56225359e-01 -5.35126925e-01 -1.54813737e-01
2.70726174e-01 4.32631642e-01 -5.60236990e-01 1.72521889e-01
1.73551962e-02 -7.92500079e-02 1.69087410e-01 5.65499365e-01
1.02589762e+00 4.28253055e-01 3.29491913e-01 -4.00887877e-01
-4.69415158e-01 5.24545968e-01 -1.71483827e+00 1.48042095e+00
-4.94164854e-01 1.03840280e+00 9.37007487e-01 -2.56036848e-01
7.72634923e-01 7.70862043e-01 4.05351430e-01 -1.02942574e+00
1.37618169e-01 8.70856881e-01 9.22293141e-02 -1.99975684e-01
7.77022123e-01 8.97260234e-02 -5.78016937e-02 5.84661305e-01
1.10019259e-01 -2.54029371e-02 -8.28746185e-02 7.14559406e-02
1.22408557e+00 -3.31905633e-01 -7.94302225e-02 -3.64676505e-01
4.93153691e-01 -5.24301469e-01 5.38510799e-01 6.60173297e-01
-5.05670905e-01 4.35681522e-01 -1.99418008e-01 9.86674130e-02
-1.00977015e+00 -1.10122561e+00 2.47628435e-01 1.06989264e+00
-2.60440499e-01 -5.26720166e-01 -8.43700051e-01 1.94114000e-01
4.42296229e-02 5.73480189e-01 -8.13264623e-02 -1.60202056e-01
-1.00619292e+00 3.78041953e-01 1.16863012e+00 4.90789324e-01
3.73483479e-01 -8.19499075e-01 -5.59044003e-01 6.37916982e-01
-3.91704559e-01 -1.11617839e+00 -1.42792225e+00 6.93506896e-01
-7.01976240e-01 -3.61870438e-01 -7.30847418e-01 -7.72467792e-01
-2.91112751e-01 1.73382834e-01 1.06323040e+00 -3.27616900e-01
2.82121927e-01 3.48714054e-01 -1.01073161e-01 -6.49977803e-01
-4.22112644e-01 -1.56832144e-01 4.85038131e-01 -5.12030832e-02
3.71149838e-01 -7.76364863e-01 -7.56000459e-01 2.05095336e-01
-4.51915562e-01 -3.10935229e-01 4.64856744e-01 6.81756437e-01
3.39454144e-01 -3.85786593e-01 6.46259725e-01 -4.49090183e-01
8.72067988e-01 -5.44289835e-02 -8.90103877e-01 -6.28343821e-02
-7.86127746e-01 -3.39284182e-01 6.00728154e-01 -6.89570069e-01
-9.59076703e-01 1.36444211e-01 -6.27953470e-01 -5.28607488e-01
2.39201784e-01 4.96380895e-01 -3.77566516e-02 8.16086009e-02
6.08489931e-01 3.25724870e-01 -1.78689271e-01 -3.73304546e-01
3.05709541e-01 1.02114522e+00 9.72248912e-01 -3.18661928e-01
7.78523445e-01 -9.77445170e-02 -1.43410653e-01 -1.08297014e+00
-8.46431628e-02 -6.97016418e-01 -2.53278911e-01 -3.27739090e-01
6.20242894e-01 -1.30654204e+00 -6.77162766e-01 8.17461610e-01
-1.42005920e+00 -3.82993966e-01 2.35149473e-01 9.94605422e-01
-2.73812711e-01 1.22949369e-01 -8.96101177e-01 -1.01529408e+00
-7.43129134e-01 -5.51846921e-01 8.05778980e-01 2.76548173e-02
-4.27985430e-01 -6.15057886e-01 5.21766305e-01 -1.48001656e-01
7.47914195e-01 -1.69703722e-01 3.29349697e-01 -4.91509110e-01
-3.74873400e-01 -2.54343480e-01 2.25744210e-02 6.66953146e-01
-4.84824449e-01 8.07082094e-03 -1.35431707e+00 -2.88886458e-01
-2.06633173e-02 -2.19653547e-01 5.80053508e-01 7.24135637e-01
7.41983593e-01 -7.21564591e-02 -2.66184509e-01 7.42579639e-01
1.14574480e+00 8.81186664e-01 5.15569150e-01 -4.39181477e-01
1.84655935e-01 2.30931565e-01 4.58435565e-01 5.79002261e-01
1.58648923e-01 7.79136479e-01 -2.43340880e-01 -1.85995504e-01
-2.69343883e-01 -2.41313428e-01 5.81703484e-01 1.85710990e+00
2.75352210e-01 -6.40183330e-01 -7.58685470e-01 3.74232799e-01
-1.41366577e+00 -6.19800448e-01 -2.81863093e-01 2.18858218e+00
9.81474340e-01 1.81848615e-01 8.89701843e-02 2.96811074e-01
4.08145815e-01 3.01344395e-01 -6.01775050e-01 -9.88220096e-01
1.76821396e-01 5.86658180e-01 5.92731535e-01 8.38191152e-01
-1.02289188e+00 5.85699916e-01 7.13687754e+00 1.02450967e+00
-1.30860960e+00 -2.33043404e-03 1.73606560e-01 5.70195727e-02
2.05252450e-02 -4.41550076e-01 -5.92337787e-01 3.80064785e-01
1.98710835e+00 3.25486399e-02 6.89501584e-01 4.53241259e-01
3.01681817e-01 -3.16562620e-03 -8.80662382e-01 1.19378757e+00
9.65410322e-02 -7.56052613e-01 -6.82951927e-01 -4.29839827e-02
2.78715968e-01 2.77989149e-01 8.27247277e-02 5.77758491e-01
-5.38679175e-02 -6.78364992e-01 1.01489663e+00 5.49028039e-01
1.33485460e+00 -8.26663911e-01 2.77343214e-01 -1.93484500e-02
-1.61285424e+00 1.91135183e-01 1.38574317e-01 4.27861810e-01
3.83863598e-01 7.18876183e-01 -3.93661201e-01 1.09407879e-01
7.30207860e-01 1.54114395e-01 3.58325183e-01 9.81287956e-01
6.38921112e-02 1.19505274e+00 -5.72523952e-01 -7.03568682e-02
1.56312305e-02 2.19139814e-01 8.72796595e-01 1.77731586e+00
3.71410429e-01 4.22579318e-01 -3.50801051e-01 1.31500751e-01
-1.14308931e-01 -3.86403233e-01 1.14081586e-02 -9.63337868e-02
1.07196164e+00 9.30346072e-01 -6.44767582e-02 -2.69987941e-01
-2.20778465e-01 1.09572816e+00 -3.42792779e-01 6.27743602e-01
-1.03042758e+00 -1.22372031e+00 6.09745204e-01 -3.74857605e-01
4.85958219e-01 -5.11334062e-01 1.38597311e-02 -5.19519031e-01
-2.05935061e-01 -1.30336177e+00 -1.97135583e-01 -7.39319563e-01
-9.21399832e-01 5.68471730e-01 -4.41353619e-01 -1.31145537e+00
-5.50325334e-01 -5.04478991e-01 -4.43167597e-01 1.20509386e+00
-1.44789875e+00 -2.74933547e-01 1.08662263e-01 5.12281895e-01
4.49329317e-01 -1.66941971e-01 1.05698597e+00 1.04485023e+00
-1.80849299e-01 1.00294292e+00 4.88777876e-01 -5.61668016e-02
1.07984972e+00 -1.27120638e+00 7.51964331e-01 7.41841674e-01
-2.59184301e-01 6.75653875e-01 7.71073997e-01 -7.20822453e-01
-1.70505238e+00 -8.83910894e-01 1.44378865e+00 1.11131467e-01
6.03562474e-01 -6.64944530e-01 -7.96464145e-01 1.67256922e-01
6.10007823e-01 -3.17797214e-01 6.77322984e-01 1.72520816e-01
-3.39121521e-01 -4.01594937e-01 -5.64823866e-01 5.81126571e-01
5.90660155e-01 -8.73914897e-01 -2.03308091e-01 -2.35805780e-01
9.57473457e-01 -6.89014912e-01 -7.14770436e-01 1.47931457e-01
1.16696239e+00 -8.51072967e-01 1.05586982e+00 -6.98093325e-02
1.77855521e-01 -3.59888673e-02 -4.73274857e-01 -1.35361338e+00
-1.34752706e-01 -1.32807422e+00 -6.39565766e-01 1.02029645e+00
4.81097877e-01 -4.69092548e-01 5.64073026e-01 3.81091267e-01
-5.78054190e-01 -3.21753323e-01 -1.05028248e+00 -1.01144826e+00
-1.61000222e-01 -1.09036875e+00 3.69351804e-01 4.22776669e-01
-1.73301041e-01 3.71787250e-01 -6.54188633e-01 2.27523729e-01
5.00795484e-01 -6.82501078e-01 4.71729100e-01 -6.74310625e-01
-5.65838277e-01 -4.61694568e-01 3.48270595e-01 -1.60710084e+00
-3.98308039e-01 -1.10567749e-01 7.58981586e-01 -5.90976894e-01
-7.00639665e-01 -2.30206534e-01 -5.76713681e-01 -9.50653851e-02
1.45509213e-01 -1.57297328e-01 3.39191437e-01 -2.38656670e-01
-5.94355047e-01 5.29048741e-01 6.02683187e-01 5.01516424e-02
-2.56617814e-01 1.35973856e-01 -4.44922447e-02 5.89181960e-01
5.50863206e-01 -7.39500582e-01 -3.61452013e-01 -3.54693443e-01
-1.12795226e-01 9.81333971e-01 1.22513980e-01 -1.33804607e+00
8.16180885e-01 2.57983059e-01 1.29291937e-01 -1.13587236e+00
8.57078791e-01 -8.36487949e-01 1.80082768e-01 8.31101060e-01
-3.87675494e-01 3.57508838e-01 6.65743291e-01 5.90292692e-01
-4.69255656e-01 1.13633707e-01 6.44822061e-01 5.86723089e-01
-4.45758939e-01 -2.04963952e-01 -9.71465945e-01 2.13414505e-01
1.74545377e-01 -2.29834225e-02 2.16894466e-02 -8.03450942e-01
-4.86607343e-01 1.31047964e-01 -4.44077402e-01 2.35301912e-01
7.64054537e-01 -1.49415910e+00 -8.04603219e-01 4.27676290e-01
-2.18868941e-01 -9.00665164e-01 4.98187512e-01 6.77946389e-01
-4.47593629e-01 8.81431460e-01 4.20863599e-01 -6.03100061e-01
-1.27985084e+00 -1.85088038e-01 5.71143985e-01 -2.55567789e-01
-2.58560628e-01 1.07206726e+00 -2.91800380e-01 -3.48633558e-01
9.05013025e-01 -1.95267498e-01 2.86482930e-01 -1.66317195e-01
7.60811925e-01 6.83074951e-01 3.70417625e-01 -4.10556912e-01
-5.99660993e-01 4.80992049e-01 2.67985493e-01 -1.02669346e+00
8.02098095e-01 -2.48943225e-01 3.16173583e-01 4.34837520e-01
1.46155620e+00 2.22159520e-01 -9.50029969e-01 -4.30072725e-01
1.25788808e-01 -3.33815329e-02 3.22445691e-01 -1.24365163e+00
-7.65659750e-01 8.67415905e-01 1.14281750e+00 -1.58011064e-01
1.60697401e+00 -8.92116070e-01 1.48226821e+00 3.82950068e-01
1.74481466e-01 -1.70285916e+00 1.81825265e-01 1.02134299e+00
9.38459694e-01 -8.03015232e-01 -4.24123555e-01 1.21047504e-01
-4.57683325e-01 1.23085856e+00 2.08966181e-01 -1.11013256e-01
1.40431619e+00 8.81523728e-01 4.82643247e-01 2.32713476e-01
-1.18773544e+00 3.29345196e-01 5.98159969e-01 6.59616649e-01
5.43704629e-01 1.67119756e-01 -3.37918460e-01 8.51895332e-01
-1.99795306e-01 -1.05923764e-01 2.21294444e-02 7.36947596e-01
-4.25267458e-01 -8.78457069e-01 -4.05584991e-01 1.13491103e-01
-4.66316909e-01 -3.38135213e-01 -9.32275057e-02 5.19808829e-01
-1.05295032e-01 1.38467753e+00 4.58785566e-03 -9.53417063e-01
6.28408849e-01 -1.25293553e-01 -7.12157264e-02 5.89242466e-02
-1.04156911e+00 9.59609926e-01 3.21432799e-01 -9.49913025e-01
-1.87984146e-02 -5.91022074e-01 -1.21963799e+00 -4.91695791e-01
-3.24067712e-01 1.88562900e-01 1.05205107e+00 5.08073390e-01
4.72340316e-01 8.32850695e-01 8.11698318e-01 -7.97464192e-01
-5.89256346e-01 -9.52184975e-01 -7.13254690e-01 -3.20155233e-01
8.28240812e-01 1.05532944e-01 -4.32065427e-01 3.84642854e-02] | [15.064191818237305, 6.004395008087158] |
c86482a6-3310-42ad-96fe-8632c14c50b9 | arbitrary-video-style-transfer-via-multi | 2009.08003 | null | https://arxiv.org/abs/2009.08003v2 | https://arxiv.org/pdf/2009.08003v2.pdf | Arbitrary Video Style Transfer via Multi-Channel Correlation | Video style transfer is getting more attention in AI community for its numerous applications such as augmented reality and animation productions. Compared with traditional image style transfer, performing this task on video presents new challenges: how to effectively generate satisfactory stylized results for any specified style, and maintain temporal coherence across frames at the same time. Towards this end, we propose Multi-Channel Correction network (MCCNet), which can be trained to fuse the exemplar style features and input content features for efficient style transfer while naturally maintaining the coherence of input videos. Specifically, MCCNet works directly on the feature space of style and content domain where it learns to rearrange and fuse style features based on their similarity with content features. The outputs generated by MCC are features containing the desired style patterns which can further be decoded into images with vivid style textures. Moreover, MCCNet is also designed to explicitly align the features to input which ensures the output maintains the content structures as well as the temporal continuity. To further improve the performance of MCCNet under complex light conditions, we also introduce the illumination loss during training. Qualitative and quantitative evaluations demonstrate that MCCNet performs well in both arbitrary video and image style transfer tasks. | ['Wei-Ming Dong', 'Changsheng Xu', 'Chongyang Ma', 'Haibin Huang', 'Fan Tang', 'Yingying Deng'] | 2020-09-17 | null | null | null | null | ['video-style-transfer'] | ['computer-vision'] | [ 3.83650154e-01 -3.51754248e-01 8.00408572e-02 -4.27994877e-01
-1.81368336e-01 -5.99233925e-01 6.30210161e-01 -4.57197487e-01
-1.14606053e-01 6.48445845e-01 2.65514255e-01 1.91981494e-01
2.53681898e-01 -6.85096562e-01 -8.46172154e-01 -6.45821989e-01
3.02241832e-01 -2.19557047e-01 1.42584950e-01 -2.66977131e-01
1.88688800e-01 3.25851023e-01 -1.35970712e+00 4.09494221e-01
9.44071293e-01 1.06688738e+00 4.39063102e-01 6.71404362e-01
-1.74880758e-01 9.53169644e-01 -4.56140459e-01 -3.56245607e-01
1.10443257e-01 -7.86545038e-01 -7.28486955e-01 3.26746315e-01
6.89026237e-01 -4.93044496e-01 -6.78338945e-01 1.08568418e+00
3.08554381e-01 2.73434490e-01 7.08189607e-01 -1.35847199e+00
-1.07982993e+00 1.30241990e-01 -5.48737586e-01 -8.07980914e-03
6.13182783e-01 4.14893419e-01 8.11025739e-01 -8.57519090e-01
7.72390962e-01 1.45036829e+00 2.17267230e-01 7.17790008e-01
-1.22319984e+00 -7.45505154e-01 4.32233006e-01 2.45890155e-01
-1.06153619e+00 -3.08835119e-01 1.00251532e+00 -2.39443481e-01
3.97629619e-01 2.37572506e-01 9.70813394e-01 1.31435609e+00
3.31964403e-01 9.92701828e-01 9.43589389e-01 -2.29635060e-01
1.83273088e-02 6.76662028e-02 -4.87940550e-01 6.83259487e-01
-2.72190392e-01 4.04041354e-03 -8.08211505e-01 3.94120783e-01
1.31239080e+00 3.08836978e-02 -5.66013873e-01 -3.24793965e-01
-1.42450285e+00 6.60410285e-01 4.83593643e-01 1.94943920e-01
-1.94259346e-01 2.78130054e-01 3.68993372e-01 5.61930418e-01
3.76207173e-01 4.73831117e-01 -2.65520781e-01 -1.30256191e-01
-6.78270400e-01 1.94555700e-01 4.03512776e-01 1.39686823e+00
8.18778217e-01 3.98150533e-01 -5.09905159e-01 8.18679512e-01
1.26793668e-01 5.73592305e-01 3.83722812e-01 -1.11938834e+00
4.01124716e-01 5.14028728e-01 2.82526053e-02 -1.25015211e+00
2.45775748e-02 -2.73731261e-01 -9.71961737e-01 1.83455318e-01
1.67971537e-01 -1.26188129e-01 -6.52468622e-01 2.05060601e+00
1.80137172e-01 5.54365635e-01 -4.44159210e-02 9.65875208e-01
8.30762446e-01 9.37864304e-01 1.03697486e-01 -1.10380031e-01
1.16155195e+00 -9.39288676e-01 -8.71758938e-01 -1.13988221e-01
3.10948789e-01 -9.12432492e-01 1.51512718e+00 1.38819143e-01
-1.24015617e+00 -8.55285823e-01 -9.51046348e-01 -2.91676909e-01
1.62379831e-01 1.51418243e-03 5.00239074e-01 6.75141215e-02
-9.88397300e-01 6.73349857e-01 -5.92340648e-01 -2.81786531e-01
4.34302956e-01 2.45715231e-01 -3.85970175e-01 -8.62249732e-02
-1.11394060e+00 5.83862066e-01 1.81775257e-01 5.58301955e-02
-6.87094688e-01 -8.90940666e-01 -1.04515207e+00 -3.68468091e-02
1.20736063e-01 -9.22856271e-01 1.02233493e+00 -1.65127516e+00
-1.96574664e+00 7.12631762e-01 -1.08114600e-01 2.03625530e-01
4.51883525e-01 -3.98015708e-01 -4.76427108e-01 2.65242308e-01
6.70668259e-02 1.07006681e+00 1.04862964e+00 -1.31725633e+00
-8.39078844e-01 8.89612734e-02 1.10097706e-01 5.39393663e-01
-6.83966756e-01 -1.55761212e-01 -7.07341671e-01 -1.12155378e+00
-1.65005535e-01 -8.60606730e-01 6.86003417e-02 4.83637124e-01
-1.74842745e-01 3.93174365e-02 1.13484430e+00 -4.85908180e-01
9.54730392e-01 -2.30913830e+00 7.83393681e-01 -4.10588682e-02
1.72025055e-01 1.66040778e-01 -4.78713900e-01 1.84763178e-01
6.55231625e-02 -3.11561614e-01 -7.13081285e-02 -1.35769233e-01
-2.39667848e-01 2.94449300e-01 -3.76564741e-01 2.76747435e-01
6.34361088e-01 1.15312874e+00 -1.10181475e+00 -4.04773861e-01
3.46756786e-01 4.39242661e-01 -9.20298219e-01 6.56059265e-01
-2.39603132e-01 9.89409804e-01 -5.09223461e-01 3.24234605e-01
4.70637530e-01 -2.32007533e-01 5.48455911e-03 -5.07279992e-01
1.64902240e-01 -2.40597799e-01 -9.83557284e-01 1.94241917e+00
-6.13703549e-01 6.99812710e-01 -5.32476814e-04 -8.65437508e-01
9.07736897e-01 2.16038972e-01 4.56411332e-01 -8.65749836e-01
2.03984454e-01 3.39153484e-02 -1.97788447e-01 -4.56529349e-01
5.41167617e-01 -1.97121307e-01 -9.40246284e-02 3.89767617e-01
2.86924511e-01 -4.58486170e-01 -2.24879488e-01 2.14469209e-01
5.45580566e-01 4.88464117e-01 -3.23377289e-02 -3.15554827e-01
6.48111284e-01 -4.61345255e-01 5.83055615e-01 2.85492390e-01
1.23387948e-02 7.50753045e-01 2.89571434e-01 -4.22047466e-01
-1.18823075e+00 -1.05905378e+00 2.50391603e-01 1.12450039e+00
5.61220765e-01 -1.60576284e-01 -6.68807626e-01 -6.25021338e-01
-1.95317507e-01 4.55532849e-01 -7.07889259e-01 -5.20502150e-01
-6.58127367e-01 6.51378408e-02 2.19111294e-01 4.97387022e-01
7.79031396e-01 -1.34219718e+00 -4.00634885e-01 2.68337131e-01
-2.59449303e-01 -1.11697674e+00 -1.24117720e+00 -2.65551418e-01
-5.67945182e-01 -8.53230476e-01 -9.54420328e-01 -1.12291324e+00
6.96290970e-01 3.87189656e-01 9.49861407e-01 2.55842894e-01
-1.08637340e-01 4.42772031e-01 -4.37708139e-01 -3.26873176e-02
-3.88255030e-01 -1.02157384e-01 2.69891992e-02 4.74729270e-01
-8.50737914e-02 -9.01389241e-01 -6.90793097e-01 3.13871235e-01
-1.28043056e+00 5.58360815e-01 3.66730243e-01 1.14068806e+00
3.56389731e-01 -3.30717325e-01 6.19685531e-01 -7.04550087e-01
5.13038754e-01 -2.38258392e-01 -3.05409014e-01 2.86991656e-01
-7.17288703e-02 2.23066629e-04 8.10307801e-01 -7.78879166e-01
-1.17038834e+00 2.65394747e-02 -6.44870773e-02 -8.19971263e-01
7.32091516e-02 1.47538438e-01 -3.96493167e-01 -1.74344897e-01
4.50882822e-01 4.89420801e-01 1.92143887e-01 -1.62960202e-01
5.69389522e-01 3.97041380e-01 7.17639208e-01 -7.37365186e-01
8.96527290e-01 2.67764330e-01 -5.89248687e-02 -6.96161091e-01
-9.28680360e-01 1.93515178e-02 -6.32318079e-01 -4.40411836e-01
8.58198524e-01 -8.78020465e-01 -6.93639696e-01 7.08898187e-01
-1.15454412e+00 -4.14291263e-01 -2.55254716e-01 3.05956542e-01
-8.77312541e-01 4.05668944e-01 -6.85126483e-01 -3.54396552e-01
-2.65149117e-01 -1.20581341e+00 1.09435546e+00 3.99924636e-01
-2.57718712e-01 -1.13591731e+00 -2.97460079e-01 -2.01706707e-01
4.15893853e-01 3.79662246e-01 9.70992267e-01 3.03061604e-01
-4.69802409e-01 1.93364277e-01 -3.87403667e-01 3.33206445e-01
5.72067857e-01 7.55946860e-02 -7.63905644e-01 -4.58423734e-01
-5.69316745e-02 -4.68202114e-01 6.61544979e-01 2.47138157e-01
1.33210063e+00 -3.82915616e-01 -2.55408548e-02 9.11871850e-01
1.23126423e+00 2.85712302e-01 6.87732637e-01 2.42732331e-01
1.11338687e+00 6.08238220e-01 5.33544838e-01 2.87899166e-01
1.29770651e-01 8.30301225e-01 1.56798899e-01 -3.52894396e-01
-3.44800562e-01 -5.28881371e-01 4.34624404e-01 8.91259849e-01
1.75400786e-02 -1.41680554e-01 -2.00093105e-01 2.77636379e-01
-1.85620594e+00 -9.30553138e-01 2.78517157e-01 1.90554523e+00
1.10226333e+00 4.36101183e-02 4.49201353e-02 -1.00751780e-01
6.74218237e-01 1.00428075e-01 -6.67584598e-01 -3.27968001e-01
-1.68288797e-01 7.43478090e-02 -3.52971070e-02 3.64805460e-01
-9.41274703e-01 1.14968729e+00 5.96883726e+00 8.46288979e-01
-1.48657739e+00 -1.55442148e-01 7.12191939e-01 -6.30037338e-02
-5.30396640e-01 -2.51036078e-01 -2.50570804e-01 6.28140807e-01
2.88024396e-01 -2.18152657e-01 7.03532636e-01 6.48242593e-01
1.91916808e-01 2.12393016e-01 -1.24641120e+00 1.11452627e+00
-1.63457859e-02 -1.42789519e+00 5.65870702e-01 -2.51545608e-01
1.02783084e+00 -6.42452180e-01 4.21027780e-01 2.32621282e-01
1.34761781e-01 -9.98923659e-01 1.07676840e+00 6.91584527e-01
1.34422767e+00 -7.91217923e-01 3.36331606e-01 -8.33131094e-03
-1.24774837e+00 3.48007232e-02 -3.09597045e-01 -8.35964456e-02
2.68159270e-01 1.07264221e-01 -9.56270695e-02 6.17416859e-01
4.93732482e-01 1.30482030e+00 -5.10965645e-01 7.56465316e-01
-1.72664374e-01 2.78573811e-01 1.22632742e-01 1.14958629e-01
3.44620109e-01 -3.46092612e-01 4.05126244e-01 1.12400699e+00
4.04006183e-01 1.43372342e-01 2.26871967e-01 8.68691564e-01
-2.12927297e-01 1.43013656e-01 -7.73468792e-01 9.94968265e-02
3.51217836e-01 1.16816807e+00 -4.57800210e-01 -3.25963110e-01
-4.80705768e-01 1.34034872e+00 4.41829175e-01 5.58570206e-01
-7.68021584e-01 -3.29928577e-01 7.82986104e-01 -1.08728863e-01
2.12685555e-01 -2.74873823e-01 -2.72848666e-01 -1.35942721e+00
-1.15386127e-02 -9.44771111e-01 1.62773188e-02 -9.67524290e-01
-1.31739640e+00 9.06123936e-01 -2.75583088e-01 -1.63777947e+00
3.54296230e-02 -6.16117775e-01 -8.10406744e-01 7.93996394e-01
-1.43155479e+00 -1.36760187e+00 -6.32181227e-01 9.69308496e-01
7.38847554e-01 -2.94565797e-01 6.27497435e-01 2.98603028e-01
-3.71110260e-01 8.00964594e-01 -5.69414534e-03 3.70471179e-02
9.26472366e-01 -1.13247883e+00 2.50045747e-01 7.17159927e-01
-2.74672471e-02 4.56367314e-01 6.07842088e-01 -5.61762810e-01
-1.52285242e+00 -1.44092417e+00 1.89721599e-01 -1.90834388e-01
4.80480969e-01 -2.95893937e-01 -9.97040868e-01 4.70003009e-01
4.31594908e-01 1.06251001e-01 3.23054075e-01 -3.11814249e-01
-3.08957011e-01 -2.08074793e-01 -8.44082415e-01 8.82384479e-01
1.28476441e+00 -5.50142884e-01 -2.85932273e-01 1.97807029e-01
8.15175414e-01 -5.32421768e-01 -8.31203699e-01 2.20487520e-01
5.93806803e-01 -8.10980558e-01 7.70665705e-01 -7.14113235e-01
9.90007579e-01 -3.89455527e-01 -2.00118959e-01 -1.55117500e+00
-5.87423861e-01 -7.35107422e-01 9.78472829e-02 1.36878216e+00
4.12888937e-02 -2.20233753e-01 4.32767719e-01 4.15460527e-01
-1.72551453e-01 -6.85255826e-01 -6.37648463e-01 -5.41962385e-01
-1.74066294e-02 -8.65439698e-02 5.98782599e-01 9.37010407e-01
-2.17733428e-01 5.47200739e-01 -8.50743890e-01 -1.74839944e-01
3.45237374e-01 1.90527558e-01 8.35451424e-01 -8.52193117e-01
-2.69191206e-01 -5.04351377e-01 -3.57861102e-01 -1.26180148e+00
2.83359885e-01 -7.69847572e-01 4.20500413e-02 -1.19752645e+00
1.49676919e-01 -5.39980948e-01 -1.34977043e-01 1.90397888e-01
-4.89753693e-01 4.63365167e-01 4.63617206e-01 2.64172703e-01
-6.26365960e-01 1.04064310e+00 2.18822193e+00 -3.19987051e-02
-1.80746660e-01 -1.63450480e-01 -7.34683275e-01 4.31211144e-01
4.88045484e-01 2.79384572e-02 -6.67140901e-01 -6.92591727e-01
-1.08573496e-01 2.50251174e-01 2.51751840e-01 -8.83260369e-01
-5.44174910e-02 -5.05794644e-01 7.39446342e-01 -2.11456567e-01
2.90838152e-01 -7.54564047e-01 2.65574127e-01 2.90120035e-01
-5.01607776e-01 1.66577280e-01 1.45772040e-01 6.94349527e-01
-3.80485475e-01 1.40367806e-01 1.07345724e+00 -2.13349611e-02
-7.25926220e-01 6.54175460e-01 -1.69757143e-01 -7.34509677e-02
1.09672940e+00 -2.66173095e-01 5.56176007e-02 -6.39626861e-01
-5.19935668e-01 3.06791544e-01 6.96700990e-01 8.33924353e-01
7.59878457e-01 -1.88862753e+00 -6.18269563e-01 4.13924992e-01
1.33455783e-01 8.75623003e-02 3.64898980e-01 4.76074457e-01
-5.54506600e-01 5.08891344e-02 -7.95830309e-01 -7.38411486e-01
-9.76937532e-01 6.73353970e-01 2.53753126e-01 1.44138202e-01
-7.88963377e-01 7.43214130e-01 8.63893509e-01 -1.44637719e-01
2.04442754e-01 -2.47538179e-01 -2.62187421e-01 -2.06779435e-01
6.87953532e-01 -4.02253158e-02 -3.76860857e-01 -5.75256646e-01
2.25011185e-02 8.34094882e-01 -8.25304314e-02 -7.81064108e-02
1.34424388e+00 -3.69789124e-01 -1.16037779e-01 4.25174832e-01
1.30411053e+00 -1.38513222e-01 -1.93708766e+00 -4.25759763e-01
-3.23857605e-01 -6.99974120e-01 -1.13674998e-01 -4.53179806e-01
-1.37970662e+00 8.42609704e-01 4.32689637e-01 -2.08809808e-01
1.35488510e+00 -3.19526285e-01 8.47622037e-01 -5.46517931e-02
2.40340412e-01 -8.65692317e-01 7.59326577e-01 4.66746926e-01
1.15853071e+00 -9.53305423e-01 -3.55430752e-01 -4.31246847e-01
-9.61716294e-01 1.23327148e+00 9.07946825e-01 -4.15978849e-01
4.69955862e-01 -2.76345089e-02 -1.21474508e-02 2.97432467e-02
-5.90857387e-01 2.13911012e-01 4.76818353e-01 6.54437959e-01
3.75289500e-01 -1.07863992e-01 -5.85004501e-02 3.98166418e-01
-1.13865487e-01 7.21071213e-02 3.81620646e-01 6.91151679e-01
-2.48929709e-01 -1.03718042e+00 -7.96251073e-02 4.26283002e-01
-1.81231648e-01 4.38216403e-02 -8.62173960e-02 4.19943303e-01
3.62299643e-02 6.22213483e-01 1.03228495e-01 -5.59613287e-01
3.90008867e-01 -3.54727179e-01 6.39796674e-01 -6.30298138e-01
-4.58123416e-01 2.29779407e-01 -2.53126085e-01 -5.17535031e-01
-4.80767727e-01 -5.45011997e-01 -9.63403642e-01 -4.13864076e-01
-4.17067297e-02 6.53168978e-03 1.85751364e-01 9.33827400e-01
2.73041308e-01 8.00786555e-01 1.03246093e+00 -1.06452727e+00
-1.42680481e-01 -8.24386954e-01 -6.08037114e-01 8.08648109e-01
3.40545088e-01 -8.18290889e-01 1.22471079e-01 5.82461357e-01] | [11.310545921325684, -0.7805736064910889] |
3c0c96ef-02be-4063-93c2-4c5cc6aaaf80 | dont-give-me-the-details-just-the-summary | 1808.08745 | null | http://arxiv.org/abs/1808.08745v1 | http://arxiv.org/pdf/1808.08745v1.pdf | Don't Give Me the Details, Just the Summary! Topic-Aware Convolutional Neural Networks for Extreme Summarization | We introduce extreme summarization, a new single-document summarization task
which does not favor extractive strategies and calls for an abstractive
modeling approach. The idea is to create a short, one-sentence news summary
answering the question "What is the article about?". We collect a real-world,
large-scale dataset for this task by harvesting online articles from the
British Broadcasting Corporation (BBC). We propose a novel abstractive model
which is conditioned on the article's topics and based entirely on
convolutional neural networks. We demonstrate experimentally that this
architecture captures long-range dependencies in a document and recognizes
pertinent content, outperforming an oracle extractive system and
state-of-the-art abstractive approaches when evaluated automatically and by
humans. | ['Shay B. Cohen', 'Mirella Lapata', 'Shashi Narayan'] | 2018-08-27 | dont-give-me-the-details-just-the-summary-1 | https://aclanthology.org/D18-1206 | https://aclanthology.org/D18-1206.pdf | emnlp-2018-10 | ['extreme-summarization'] | ['natural-language-processing'] | [ 5.44915795e-01 6.74510717e-01 -2.44312361e-01 -3.77324313e-01
-1.41552389e+00 -6.82149231e-01 8.70523095e-01 5.29598534e-01
-6.27651155e-01 1.02893126e+00 1.23711574e+00 -3.80848795e-01
-4.30826247e-02 -4.97829050e-01 -1.12217605e+00 -6.05785698e-02
3.29109468e-02 8.03827643e-01 -5.89973154e-03 -4.02767897e-01
6.37855709e-01 1.60708889e-01 -1.05803013e+00 9.03893352e-01
9.38084841e-01 6.78588390e-01 1.53770372e-01 1.30880034e+00
-3.52276981e-01 1.13220251e+00 -1.28420150e+00 -7.83202231e-01
-1.57449454e-01 -6.61273718e-01 -1.27489042e+00 -4.83789928e-02
9.50242460e-01 -6.36235952e-01 -3.86893094e-01 8.54310155e-01
4.40457851e-01 -3.90050784e-02 8.02954674e-01 -4.45075244e-01
-8.95759940e-01 1.47616065e+00 -3.01712543e-01 6.38689101e-01
5.90329766e-01 -2.18745932e-01 1.54401910e+00 -6.72354281e-01
8.78598094e-01 9.13173616e-01 5.94061017e-01 5.52286088e-01
-1.04406834e+00 -5.98850176e-02 2.84751922e-01 -2.95434118e-04
-5.67565024e-01 -6.74244165e-01 7.95978248e-01 6.23060204e-03
1.59088111e+00 7.43800819e-01 5.41508019e-01 1.48910916e+00
6.55031681e-01 1.29804480e+00 3.84555131e-01 -5.40697455e-01
9.42731276e-02 -1.62929267e-01 6.28725410e-01 5.01220882e-01
6.04058385e-01 -7.13437319e-01 -8.41341019e-01 -2.39729092e-01
-2.47780815e-01 -3.50435436e-01 -5.02867341e-01 3.21591169e-01
-1.28303921e+00 7.43833721e-01 1.26566425e-01 2.31779888e-01
-8.93956423e-01 2.91209906e-01 6.63447738e-01 4.31612402e-01
9.41356301e-01 1.01328647e+00 -5.42448878e-01 -1.20538697e-01
-1.53016758e+00 6.67593837e-01 1.53606594e+00 1.15802562e+00
3.11182022e-01 -3.59918550e-02 -5.36046624e-01 6.42121732e-01
-2.21141934e-01 4.99970824e-01 5.55169463e-01 -9.89799857e-01
8.64459455e-01 3.83020639e-01 2.54187465e-01 -7.82837868e-01
-5.12739956e-01 -9.28809762e-01 -8.39667201e-01 -5.18450856e-01
-2.81883597e-01 -4.65871632e-01 -7.28664994e-01 1.33920252e+00
-3.23848456e-01 -3.66235286e-01 3.81165117e-01 4.93549079e-01
1.38207042e+00 1.02059233e+00 -3.60021323e-01 -5.44761002e-01
1.41846228e+00 -1.25881219e+00 -8.93650532e-01 -5.60757458e-01
3.66453946e-01 -7.03017414e-01 5.84892213e-01 4.77459729e-01
-1.58862066e+00 -3.02309364e-01 -1.43327093e+00 -7.26698160e-01
-2.73220599e-01 4.05050755e-01 2.41402477e-01 1.47118479e-01
-1.19084501e+00 5.99175274e-01 -5.82644045e-01 -4.21532273e-01
3.72316450e-01 1.10255070e-01 -3.10997158e-01 1.65490150e-01
-1.04398227e+00 1.13201940e+00 6.38390183e-01 -1.09247658e-02
-7.73761153e-01 -6.32278144e-01 -7.07042098e-01 5.84243894e-01
5.90718389e-01 -1.24659181e+00 1.81197906e+00 -7.44376779e-01
-1.52398002e+00 7.49921679e-01 -2.93933809e-01 -1.22501469e+00
3.56002271e-01 -9.01289821e-01 -2.96012223e-01 6.11282945e-01
2.68362910e-01 4.42134678e-01 8.24081719e-01 -1.07250881e+00
-6.54013932e-01 -2.77421713e-01 1.97749078e-01 2.26322845e-01
-2.17744708e-01 2.17932835e-01 -3.32627803e-01 -8.21247697e-01
-2.49116987e-01 -7.08467841e-01 -7.87180942e-03 -8.73985648e-01
-1.18888581e+00 -2.50895709e-01 4.91361022e-01 -1.30674374e+00
1.41574824e+00 -1.54161143e+00 4.29353148e-01 -3.98649454e-01
4.08584058e-01 1.03951979e-03 -2.22606927e-01 9.37513351e-01
2.16042757e-01 3.38416517e-01 -4.06013459e-01 -5.19138932e-01
3.39636385e-01 3.80915403e-02 -1.04460573e+00 4.55980748e-02
3.94559860e-01 1.10969603e+00 -8.11324000e-01 -5.50316811e-01
-4.51335281e-01 -1.59852028e-01 -5.07454216e-01 1.52444676e-01
-5.76776862e-01 -2.02149808e-01 -4.84116495e-01 2.77743340e-01
3.70641530e-01 -1.28993586e-01 9.90532413e-02 -8.83236155e-02
-1.06532626e-01 7.34828055e-01 -3.02451611e-01 1.80951571e+00
-2.08750159e-01 1.08625245e+00 6.97700903e-02 -1.03865504e+00
6.62635028e-01 2.75696248e-01 1.18894905e-01 -4.59246159e-01
6.69046044e-02 3.83021057e-01 -3.93851787e-01 -5.22030175e-01
1.45794487e+00 1.38229147e-01 -6.57025993e-01 6.52451336e-01
5.58933437e-01 -2.93660641e-01 5.92038393e-01 9.64594901e-01
1.50404263e+00 -1.39130697e-01 5.70590079e-01 -4.46593791e-01
3.18539649e-01 2.65522093e-01 2.33549461e-01 1.41955864e+00
4.51514542e-01 7.51806796e-01 8.12060416e-01 -3.69187474e-01
-1.01696515e+00 -9.10444081e-01 2.83728063e-01 1.13827240e+00
-2.46319145e-01 -7.58027554e-01 -9.50741827e-01 -8.96368682e-01
-1.34182945e-01 1.42902946e+00 -7.69109368e-01 -6.66845739e-02
-9.36030149e-01 -4.28663820e-01 6.65455937e-01 4.39366251e-01
3.59871835e-01 -1.12471712e+00 -6.73068821e-01 3.77033353e-01
-6.93998277e-01 -1.04942358e+00 -5.59390068e-01 3.41637880e-01
-7.56813705e-01 -6.24230266e-01 -8.15764070e-01 -6.59340858e-01
2.44839013e-01 -9.87748280e-02 1.67856336e+00 -1.60268649e-01
1.58702806e-01 4.66963023e-01 -3.69869947e-01 -8.65834534e-01
-8.39554906e-01 9.33323979e-01 -3.70450199e-01 -2.02828452e-01
3.12208205e-01 -4.95778292e-01 -3.07182997e-01 -7.84602880e-01
-1.02025628e+00 2.08124593e-01 1.09966767e+00 8.30279589e-01
1.78713024e-01 -3.16933483e-01 1.01485991e+00 -1.35850656e+00
1.40536094e+00 -3.86838198e-01 -1.33784786e-02 4.99752969e-01
-3.41936648e-01 3.57037604e-01 7.67486334e-01 3.40205128e-03
-1.16085291e+00 -3.67230386e-01 -2.25134626e-01 2.30395615e-01
-4.41668145e-02 9.48963046e-01 8.68691057e-02 9.25571322e-01
7.26601422e-01 5.88942766e-01 -2.77356803e-01 -7.11677670e-01
6.65379643e-01 9.44337547e-01 1.12410843e+00 -3.89461815e-01
4.15068090e-01 3.15320224e-01 -5.22944510e-01 -1.08152139e+00
-1.44580531e+00 -5.34713864e-01 -4.31454927e-01 1.61732230e-02
8.63167346e-01 -8.98709774e-01 -1.57422960e-01 1.78648919e-01
-1.92615342e+00 1.75338015e-01 -5.98162055e-01 2.06350148e-01
-5.91590703e-01 4.78248656e-01 -6.62053227e-01 -3.91958356e-01
-1.22146201e+00 -3.98226261e-01 1.27438080e+00 2.86745995e-01
-5.36996007e-01 -6.43441975e-01 1.75240055e-01 3.36720347e-01
4.61266249e-01 4.00196239e-02 9.54257369e-01 -1.35401392e+00
-3.11409384e-01 -4.57416445e-01 -4.46911678e-02 4.96808916e-01
-3.13170463e-01 3.56162749e-02 -6.48065507e-01 -4.08671647e-02
2.13903666e-01 -3.46724004e-01 1.55185020e+00 4.45263028e-01
8.23483825e-01 -1.13827693e+00 -1.85827121e-01 1.12720653e-01
1.08509123e+00 -4.49183643e-01 5.36347091e-01 2.69876331e-01
3.37112069e-01 6.26427352e-01 -1.17651746e-01 3.44455808e-01
5.09964585e-01 1.75186276e-01 3.46162282e-02 1.88214883e-01
-2.57108361e-01 -4.35805380e-01 4.41750228e-01 1.39092600e+00
-1.95591431e-03 -9.06654894e-01 -5.83853841e-01 6.91865623e-01
-1.99545193e+00 -1.37433052e+00 3.38676870e-02 1.45603657e+00
1.06107020e+00 5.31927407e-01 -2.91799814e-01 -4.38077241e-01
3.64913583e-01 6.37423456e-01 -3.75897735e-01 -9.14712727e-01
-4.15437102e-01 3.24143618e-01 3.79166156e-01 4.63695258e-01
-1.18796039e+00 7.10910320e-01 6.79893446e+00 6.65207148e-01
-8.25433195e-01 -9.26850215e-02 3.78399611e-01 -3.08367461e-01
-5.25960684e-01 -3.29953022e-02 -7.89934337e-01 2.37847045e-01
1.28452420e+00 -7.40677893e-01 -1.72622707e-02 8.00356150e-01
6.50730804e-02 -1.54565781e-01 -1.21287906e+00 3.61041784e-01
8.63094628e-01 -1.98264980e+00 5.35952151e-01 -1.14360586e-01
7.19608903e-01 3.41833383e-01 -3.90199482e-01 6.55109942e-01
4.03202444e-01 -7.98604429e-01 1.01861358e+00 8.87470543e-01
3.69676739e-01 -5.98430991e-01 9.36467767e-01 6.66211069e-01
-3.90948296e-01 -9.16456431e-03 -4.43940014e-01 -3.99610959e-02
4.39715922e-01 6.51864767e-01 -7.32180357e-01 9.18032467e-01
4.19704914e-01 7.01784849e-01 -8.76185596e-01 7.81262279e-01
-4.61003184e-01 8.27921033e-01 -1.43166214e-01 -4.37076747e-01
4.82283741e-01 9.27662179e-02 1.00345504e+00 1.85017025e+00
1.76868439e-01 1.83727127e-02 -7.73087442e-02 7.87442148e-01
-9.16234434e-01 6.59630150e-02 -5.85408688e-01 -2.86778212e-01
1.08189665e-01 1.10621715e+00 -5.18904150e-01 -9.50855136e-01
-1.77089974e-01 1.10563612e+00 4.50899303e-01 4.24796700e-01
-5.12712240e-01 -8.30413282e-01 -3.14347118e-01 -2.85395771e-01
6.87741756e-01 -5.37306033e-02 -2.44175851e-01 -1.47552943e+00
3.42922866e-01 -1.09511840e+00 2.82701939e-01 -8.29553604e-01
-1.13791180e+00 8.69449675e-01 1.24136582e-01 -8.56493771e-01
-6.50444329e-01 -2.44076550e-01 -8.90258968e-01 5.20059705e-01
-1.37321186e+00 -9.65634584e-01 1.10741474e-01 -1.44496322e-01
1.09183812e+00 -2.06218079e-01 9.01158571e-01 -2.90215999e-01
-4.45811540e-01 1.50466308e-01 3.45072120e-01 1.81164160e-01
4.41610247e-01 -1.50837743e+00 7.48770416e-01 1.09769571e+00
5.12553394e-01 7.57091105e-01 1.24307573e+00 -5.60348392e-01
-1.44712734e+00 -9.50777352e-01 1.52005434e+00 -6.15818381e-01
6.17290378e-01 -3.46764922e-01 -7.45065153e-01 9.19103861e-01
1.33828938e+00 -8.09569120e-01 6.59802377e-01 2.63709724e-01
-3.44007760e-01 -2.59670690e-02 -5.70706129e-01 5.45687675e-01
7.92617023e-01 -5.29336572e-01 -1.67823577e+00 6.53364420e-01
1.15578628e+00 -3.29220176e-01 -4.39691544e-01 1.99678302e-01
4.60987955e-01 -6.55566573e-01 4.92562085e-01 -1.12361658e+00
1.18032634e+00 2.72758037e-01 -2.57814042e-02 -1.51853573e+00
-1.41375154e-01 -8.25273216e-01 -7.04031169e-01 9.93999124e-01
8.68841708e-01 -3.62783372e-01 5.07781386e-01 2.28618503e-01
-6.58798635e-01 -5.95187306e-01 -7.46047795e-01 -5.13651907e-01
1.75000682e-01 -9.94698331e-02 3.43222290e-01 3.67585421e-01
1.98790312e-01 1.19566011e+00 -3.59111607e-01 -1.54559091e-01
3.82536381e-01 4.47451234e-01 8.10034573e-01 -1.08252347e+00
-3.01031291e-01 -6.77870333e-01 6.50590286e-02 -1.52825058e+00
3.91702503e-01 -8.20635676e-01 2.46162385e-01 -2.30523992e+00
4.56005812e-01 6.83454096e-01 -4.40219864e-02 1.14555642e-01
-1.56954691e-01 -2.00174540e-01 -1.47329923e-02 1.98030263e-01
-1.11303782e+00 6.61654949e-01 6.80476785e-01 -6.49232388e-01
1.30734593e-01 9.69877020e-02 -1.26060057e+00 6.65486753e-01
6.13763869e-01 -5.57916105e-01 -1.37498736e-01 -7.77061939e-01
4.87998873e-01 1.87685505e-01 7.63059855e-02 -8.32200646e-01
6.40476048e-01 2.77893692e-01 2.84379780e-01 -1.16556251e+00
7.96445310e-02 -1.20752499e-01 -4.44129556e-01 2.19302163e-01
-1.07156396e+00 3.00297320e-01 1.20655470e-01 7.29886293e-01
-3.87875736e-01 -5.13216794e-01 2.23702397e-02 -3.79891574e-01
-2.20574677e-01 -1.93091124e-01 -6.43612027e-01 4.79241371e-01
2.78460115e-01 3.04380327e-01 -8.58071327e-01 -7.60291338e-01
-3.90717804e-01 1.76455826e-01 2.85156965e-02 3.16178024e-01
5.98601699e-01 -6.64372563e-01 -1.53624296e+00 -2.72276759e-01
-4.79855575e-02 -1.22309580e-01 1.06162623e-01 6.72373474e-01
-7.65496373e-01 8.91592264e-01 1.37861967e-01 -7.88670704e-02
-9.20208693e-01 4.06576216e-01 -1.64345667e-01 -7.64459550e-01
-9.62452888e-01 7.06805110e-01 -2.00280041e-01 -1.56196371e-01
1.28589436e-01 -6.31505668e-01 -3.32246244e-01 3.68887872e-01
7.01619148e-01 1.69114783e-01 3.84966969e-01 -2.94690847e-01
-9.54277664e-02 -1.28689304e-01 -6.72623098e-01 -3.59690875e-01
1.59586906e+00 -1.17432676e-01 -3.15588444e-01 4.42492902e-01
1.17632306e+00 4.30926949e-01 -6.76537991e-01 -1.59100622e-01
4.38816160e-01 2.46014267e-01 2.16259807e-02 -1.02977204e+00
-3.07271034e-01 6.16791487e-01 -5.65028846e-01 7.43253529e-01
9.36381042e-01 2.21060857e-01 1.09262657e+00 1.24419570e+00
-2.60224223e-01 -1.34562814e+00 3.70280333e-02 8.00024629e-01
1.58747077e+00 -8.58037233e-01 5.42827725e-01 2.06832647e-01
-6.61899567e-01 1.33239162e+00 9.60864797e-02 -3.65086675e-01
1.73856720e-01 1.60184398e-01 -1.65989578e-01 -5.06909728e-01
-1.25992143e+00 2.23415494e-01 5.20873904e-01 1.28518030e-01
2.35183239e-01 8.04888623e-05 -4.86334294e-01 8.87208641e-01
-7.06487715e-01 -1.97103977e-01 1.08008063e+00 1.02054596e+00
-7.83192813e-01 -5.29995382e-01 2.12914661e-01 8.74153793e-01
-8.66568148e-01 -3.91481638e-01 -1.02179635e+00 6.06176853e-01
-5.46766996e-01 9.05863941e-01 -1.26573527e-02 -1.18741882e-03
4.04956192e-01 2.56435484e-01 2.99601942e-01 -8.30944002e-01
-9.45476174e-01 -1.06608659e-01 8.34865570e-01 -1.03313908e-01
-3.09280455e-01 -4.72136259e-01 -9.18877900e-01 -1.88648939e-01
-1.96574539e-01 6.11651003e-01 7.57051468e-01 1.00381374e+00
7.05513120e-01 7.41175592e-01 4.46721494e-01 -6.92860425e-01
-1.02599645e+00 -1.47421229e+00 -2.47965470e-01 1.61080435e-01
6.34156585e-01 2.79413760e-01 -2.55780250e-01 2.58573800e-01] | [12.49569320678711, 9.588583946228027] |
669f0773-c476-4e33-846b-8ee9e08bf14c | weakly-supervised-fine-grained-image | 1504.04943 | null | http://arxiv.org/abs/1504.04943v1 | http://arxiv.org/pdf/1504.04943v1.pdf | Weakly Supervised Fine-Grained Image Categorization | In this paper, we categorize fine-grained images without using any object /
part annotation neither in the training nor in the testing stage, a step
towards making it suitable for deployments. Fine-grained image categorization
aims to classify objects with subtle distinctions. Most existing works heavily
rely on object / part detectors to build the correspondence between object
parts by using object or object part annotations inside training images. The
need for expensive object annotations prevents the wide usage of these methods.
Instead, we propose to select useful parts from multi-scale part proposals in
objects, and use them to compute a global image representation for
categorization. This is specially designed for the annotation-free fine-grained
categorization task, because useful parts have shown to play an important role
in existing annotation-dependent works but accurate part detectors can be
hardly acquired. With the proposed image representation, we can further detect
and visualize the key (most discriminative) parts in objects of different
classes. In the experiment, the proposed annotation-free method achieves better
accuracy than that of state-of-the-art annotation-free and most existing
annotation-dependent methods on two challenging datasets, which shows that it
is not always necessary to use accurate object / part annotations in
fine-grained image categorization. | ['Minh N. Do', 'Viet-Anh Nguyen', 'Xiu-Shen Wei', 'Jianfei Cai', 'Jianxin Wu', 'Yu Zhang', 'Jiangbo Lu'] | 2015-04-20 | null | null | null | null | ['image-categorization'] | ['computer-vision'] | [ 4.29898314e-02 -1.32147625e-01 -9.23053324e-02 -4.31024760e-01
-4.33675706e-01 -7.17128217e-01 7.28891671e-01 6.16722763e-01
-5.13746738e-01 4.78082240e-01 -1.47551402e-01 8.95657986e-02
-3.70971024e-01 -7.87535787e-01 -5.00662565e-01 -8.00503671e-01
1.48229167e-01 7.13291228e-01 8.44773352e-01 -1.66511893e-01
3.75250518e-01 7.38564909e-01 -2.05810070e+00 3.27193558e-01
6.31365538e-01 1.31936073e+00 4.71332699e-01 3.62181842e-01
-5.64094901e-01 5.17928302e-01 -8.54207933e-01 -1.63457260e-01
2.87697107e-01 -1.70768961e-01 -1.01393485e+00 4.96888787e-01
4.61529583e-01 -8.14455077e-02 2.42573157e-01 1.23679590e+00
2.37350911e-01 7.69672124e-03 8.73582542e-01 -1.26142323e+00
-3.55551749e-01 4.74494547e-01 -4.66589093e-01 2.49564037e-01
-1.51753873e-01 -2.26875186e-01 7.23580718e-01 -7.35340893e-01
3.69543076e-01 1.26122248e+00 4.74547893e-01 3.94454509e-01
-1.01638103e+00 -5.89670241e-01 4.77474660e-01 1.95379466e-01
-1.75246096e+00 -2.73331881e-01 6.92549288e-01 -6.86933994e-01
4.65029180e-01 4.42787915e-01 4.49858725e-01 4.57996130e-01
-1.17067710e-01 5.45973361e-01 1.27883542e+00 -4.84025270e-01
3.24247211e-01 3.57541203e-01 6.05476320e-01 6.55124903e-01
4.13833261e-01 -3.67782414e-01 4.46240343e-02 -7.16968551e-02
6.93569541e-01 4.34739083e-01 -1.41041100e-01 -5.99706829e-01
-1.24099839e+00 6.38719797e-01 8.36323977e-01 7.59306252e-01
-4.17350471e-01 -2.18036752e-02 4.24439967e-01 9.69766155e-02
4.16031897e-01 3.34016651e-01 -3.51285070e-01 2.68448830e-01
-1.04131937e+00 1.44202998e-02 3.06025386e-01 8.30321491e-01
1.05155599e+00 -3.67197096e-01 -4.49114710e-01 9.60221231e-01
3.19974989e-01 -5.61944060e-02 7.41661608e-01 -4.76252407e-01
1.16101697e-01 1.17606735e+00 1.59098089e-01 -7.11621284e-01
-4.39248681e-01 -5.60680568e-01 -9.70762730e-01 3.38767767e-01
6.17138386e-01 5.57387114e-01 -1.19765735e+00 1.18359327e+00
5.61608493e-01 -2.02311859e-01 -2.81120330e-01 9.69088376e-01
8.29340219e-01 2.40949348e-01 3.32818747e-01 -1.01028313e-03
1.90113306e+00 -1.08313739e+00 -2.72103339e-01 -1.30954772e-01
4.00808811e-01 -9.20993865e-01 1.18735671e+00 3.32200468e-01
-6.12575114e-01 -9.72350955e-01 -1.04665935e+00 1.39645249e-01
-8.40962410e-01 4.46014941e-01 6.14480019e-01 6.28253996e-01
-7.06878126e-01 5.32387137e-01 -5.68351209e-01 -4.78248328e-01
5.16151786e-01 2.53776401e-01 -6.63838148e-01 -8.47174004e-02
-7.14497387e-01 7.22682178e-01 7.36003995e-01 9.30590928e-02
-9.06042159e-01 -2.62123674e-01 -6.07041419e-01 2.50512391e-01
3.26959252e-01 -5.02127826e-01 9.64219868e-01 -1.08499753e+00
-9.09205496e-01 1.07978916e+00 1.18302427e-01 -2.35373467e-01
4.26439434e-01 2.96279769e-02 -1.87378943e-01 2.78688401e-01
2.53771156e-01 8.50619972e-01 1.19957602e+00 -1.44135749e+00
-9.16452706e-01 -4.16701734e-01 5.37970126e-01 -6.28314540e-02
-4.20919567e-01 -3.77662182e-02 -7.26630986e-01 -7.82729924e-01
2.92758375e-01 -7.88507640e-01 -1.00439779e-01 1.13386564e-01
-4.39191788e-01 -6.82054758e-01 9.65102017e-01 -1.77930057e-01
1.01528680e+00 -1.99203181e+00 -2.36506388e-01 -1.04500107e-01
3.25447291e-01 3.61558914e-01 -7.24688079e-03 3.60475034e-01
2.18586326e-02 1.58577874e-01 6.78687915e-02 -3.01162839e-01
1.03367269e-01 5.43519035e-02 2.60077138e-02 4.06420618e-01
2.02353433e-01 6.87403977e-01 -5.06305933e-01 -8.42643797e-01
5.56173265e-01 3.44417989e-01 -1.92166388e-01 1.53997242e-01
-1.41532615e-01 2.72167981e-01 -7.21007109e-01 7.04221666e-01
6.81606114e-01 -2.45755047e-01 -1.00454986e-01 -5.33900917e-01
-1.01214580e-01 -1.17728084e-01 -1.33608437e+00 1.28535962e+00
-5.39855123e-01 2.44688690e-01 1.66753158e-02 -1.24994111e+00
9.23032761e-01 5.94478399e-02 1.45765498e-01 -6.32857263e-01
3.25606883e-01 2.06825405e-01 1.32416248e-01 -2.39977673e-01
4.76868391e-01 -1.46522284e-01 -2.67207086e-01 4.30935025e-01
1.62253097e-01 -1.25026941e-01 5.57851017e-01 4.67178412e-03
6.54356122e-01 3.02383360e-02 5.08225858e-01 -5.84129512e-01
8.03614736e-01 1.09689526e-01 2.22610354e-01 7.94178903e-01
-2.21513018e-01 7.74783432e-01 -2.03272905e-02 -4.13739473e-01
-9.16215897e-01 -5.96034706e-01 -4.13093358e-01 1.45300007e+00
4.58479166e-01 -3.69687170e-01 -1.00779927e+00 -1.09387136e+00
-4.38702181e-02 1.01939395e-01 -8.47610831e-01 6.98758289e-02
-2.79922634e-01 -6.63643241e-01 3.56143266e-01 5.59623361e-01
6.55846536e-01 -1.07194161e+00 -7.13161111e-01 4.06272449e-02
-1.97392493e-01 -7.83612132e-01 -3.21185499e-01 5.49557865e-01
-7.17721224e-01 -1.28879142e+00 -1.00757766e+00 -9.06381965e-01
9.33185101e-01 7.14184046e-01 1.06802666e+00 4.19708610e-01
-4.90113705e-01 9.64212194e-02 -6.65220320e-01 -3.73937070e-01
-2.63264835e-01 2.38535944e-02 -1.02407478e-01 3.50752145e-01
1.40983716e-01 -2.53843635e-01 -7.61994123e-01 8.39545965e-01
-9.85170841e-01 -1.38996050e-01 9.11659718e-01 9.08247709e-01
7.46073842e-01 5.36859334e-01 4.25026178e-01 -8.39758456e-01
2.07380489e-01 -4.64158617e-02 -6.48132563e-01 4.36156034e-01
-4.61002767e-01 2.61580765e-01 7.79871643e-01 -4.38846052e-01
-8.03549707e-01 1.07671078e-02 -2.08614796e-01 -4.23495024e-01
-7.12055445e-01 -1.11493044e-01 -3.61938953e-01 -2.06480816e-01
6.18029833e-01 1.94084838e-01 -5.38576365e-01 -1.01080346e+00
2.75970221e-01 1.00854886e+00 2.98615366e-01 -5.07818282e-01
7.48470962e-01 4.83868003e-01 -1.26527935e-01 -6.98175251e-01
-8.97510886e-01 -1.02873850e+00 -1.03753304e+00 1.04226843e-01
8.42181802e-01 -7.72518575e-01 -6.12567544e-01 3.58516961e-01
-8.03646147e-01 -1.32731289e-01 -4.79978204e-01 1.43529654e-01
-3.20441514e-01 4.40842599e-01 -1.84936672e-01 -6.02953136e-01
-3.01081598e-01 -1.13505101e+00 1.67025959e+00 2.45269224e-01
4.47690524e-02 -6.90172255e-01 -4.18535620e-01 3.37434173e-01
4.43076760e-01 -9.90839303e-02 8.50778818e-01 -7.08503783e-01
-4.13729012e-01 -4.52885509e-01 -5.70466638e-01 3.03792536e-01
2.72118002e-01 -8.19921196e-02 -1.10653687e+00 -2.52073437e-01
-2.48572826e-01 -3.56230676e-01 9.69088495e-01 1.04116641e-01
1.45203173e+00 -2.45908558e-01 -5.67366958e-01 2.49526158e-01
1.46717930e+00 2.18539182e-02 4.09593433e-01 3.37615550e-01
7.17664659e-01 7.27695882e-01 1.07876456e+00 2.82159418e-01
8.03385973e-02 8.83125663e-01 4.84563291e-01 -7.04822764e-02
-3.82075638e-01 -9.00976434e-02 -2.48939291e-01 2.26002932e-01
-2.05365404e-01 -3.29318166e-01 -6.28114343e-01 5.02536297e-01
-1.73679376e+00 -7.44601250e-01 -1.87535197e-01 2.12474179e+00
6.11022234e-01 3.89339775e-01 2.36596242e-01 6.08081698e-01
8.93743873e-01 -6.93330243e-02 -2.42552593e-01 -1.33516900e-02
9.10689607e-02 2.47868687e-01 3.87980670e-01 1.51586216e-02
-1.59959364e+00 9.27056491e-01 5.06583691e+00 1.40802205e+00
-1.16296566e+00 3.18400443e-01 5.17678261e-01 4.73836601e-01
2.24945426e-01 -1.18375443e-01 -9.24392819e-01 3.64929438e-01
4.67324525e-01 2.66692996e-01 -1.54843494e-01 1.22805238e+00
-1.99978836e-02 -3.87108594e-01 -9.19270873e-01 1.06917310e+00
-1.90344565e-02 -1.14841044e+00 2.74639111e-02 -1.20909601e-01
4.46318805e-01 -4.53887731e-01 -4.76304471e-01 1.97756559e-01
-5.51992506e-02 -8.74587774e-01 1.16741133e+00 2.75280744e-01
7.49920309e-01 -4.85625386e-01 9.99543846e-01 6.03673995e-01
-1.59391427e+00 -1.49745360e-01 -8.05758893e-01 -1.97422728e-02
-1.58936754e-01 5.01472056e-01 -7.72236228e-01 5.22766829e-01
8.77772152e-01 2.49585986e-01 -9.27101374e-01 1.31623423e+00
-1.51830032e-01 5.35989046e-01 -1.72812715e-01 -1.30076027e-02
3.34897965e-01 9.05824155e-02 1.10895606e-02 1.27269399e+00
7.70277157e-02 -3.44024971e-02 5.78730047e-01 4.34278637e-01
-1.41962040e-02 2.18636438e-01 -1.22723624e-01 -1.44663192e-02
1.57117173e-01 1.86740994e+00 -1.62456799e+00 -6.06947720e-01
-1.24079406e-01 1.06971908e+00 2.32958674e-01 -1.08972058e-01
-5.86882651e-01 -5.59202194e-01 5.19835413e-01 3.79843503e-01
5.40850699e-01 -8.74030665e-02 4.53436747e-03 -9.66486454e-01
-9.71283540e-02 -5.81456482e-01 5.98888934e-01 -6.62943006e-01
-1.28161764e+00 9.65635419e-01 1.39822334e-01 -1.47301984e+00
-1.09388873e-01 -5.85754514e-01 -3.93130422e-01 7.28409529e-01
-1.36254764e+00 -1.69160187e+00 -8.31262708e-01 7.35656738e-01
7.54425645e-01 2.91659385e-01 7.90120006e-01 4.57902402e-01
-3.48547310e-01 4.63452637e-01 -1.14355184e-01 1.15506671e-01
7.05485404e-01 -1.38622248e+00 1.32193729e-01 7.00588405e-01
3.93530965e-01 6.81662202e-01 6.47793114e-01 -4.13322628e-01
-8.65358770e-01 -1.31556666e+00 4.89816785e-01 -3.76125783e-01
3.31392139e-01 -6.09011829e-01 -8.53037775e-01 1.68640256e-01
-1.09471038e-01 4.92664516e-01 2.60489851e-01 -6.59903362e-02
-5.31440556e-01 -2.33925685e-01 -1.23264074e+00 2.44394764e-02
9.60065901e-01 -2.70757705e-01 -7.11062968e-01 4.15032923e-01
4.96492893e-01 1.68656424e-01 -6.44574285e-01 4.53273594e-01
5.34858167e-01 -9.78731275e-01 1.07564580e+00 -3.85082334e-01
-1.79777607e-01 -8.14089954e-01 -1.17184594e-01 -1.04723454e+00
-5.54305375e-01 1.20997056e-01 2.69333661e-01 1.45143735e+00
-1.79739729e-01 -4.87335920e-01 6.81367874e-01 6.77059684e-03
-2.88201094e-01 -3.60387087e-01 -6.34281576e-01 -9.39240217e-01
-1.12526394e-01 -2.07203642e-01 6.35446548e-01 6.37171924e-01
-5.18185139e-01 2.86860466e-01 5.50402068e-02 1.78203419e-01
5.31049371e-01 7.87096083e-01 8.36137295e-01 -1.69383848e+00
-2.09156662e-01 -6.41239226e-01 -9.00738835e-01 -7.88841128e-01
-9.37914923e-02 -6.11443579e-01 2.46595904e-01 -1.78084624e+00
4.58240330e-01 -1.10428476e+00 -3.92193228e-01 8.07098031e-01
-2.82894462e-01 1.21339309e+00 2.97343850e-01 5.22710860e-01
-8.14970016e-01 2.31887847e-01 1.03943586e+00 -5.38313687e-01
1.49535805e-01 1.91876769e-01 -7.08545864e-01 9.00840878e-01
6.69947743e-01 -3.96188438e-01 -3.03855836e-01 1.76068813e-01
-4.40739810e-01 -5.12323201e-01 4.47273165e-01 -1.18840289e+00
-1.89430546e-02 -1.57210678e-01 5.34443140e-01 -7.29400277e-01
3.54463667e-01 -1.02380514e+00 2.18480736e-01 4.51169819e-01
8.79843831e-02 -3.73890340e-01 1.66673630e-01 3.14882904e-01
-4.84497905e-01 -5.59035003e-01 1.15156960e+00 -3.76371562e-01
-1.16344881e+00 1.72449186e-01 -9.53816101e-02 -1.37884676e-01
1.16924024e+00 -3.19977343e-01 -2.89967090e-01 1.60033390e-01
-7.56890535e-01 -2.43759722e-01 6.62540138e-01 4.25629020e-01
2.35052392e-01 -1.11995173e+00 -3.54575515e-01 2.10768059e-01
6.97945356e-01 -8.09775442e-02 5.46245098e-01 7.72582471e-01
-4.33501601e-01 6.28196180e-01 -4.25650060e-01 -7.47403860e-01
-1.61386549e+00 1.11638677e+00 2.12866023e-01 -1.89928070e-01
-5.00295937e-01 8.16513598e-01 8.41053605e-01 -2.14850128e-01
1.53412566e-01 -5.55117607e-01 -5.62135458e-01 3.58195156e-01
5.83900034e-01 1.42039865e-01 5.17831802e-01 -8.32123637e-01
-5.76724112e-01 9.57500696e-01 -8.66233706e-02 4.38621730e-01
1.05177736e+00 -2.69013196e-01 7.25044832e-02 3.07699263e-01
1.02271891e+00 1.35741666e-01 -1.03704619e+00 -6.72777444e-02
9.60715674e-03 -5.02439737e-01 -8.87929797e-02 -7.91585863e-01
-8.32703590e-01 1.23752141e+00 8.70543122e-01 5.98024130e-01
1.26485312e+00 4.27079648e-01 3.28490317e-01 1.45819917e-01
7.56501317e-01 -7.14998007e-01 -4.85255457e-02 2.19449878e-01
8.21606874e-01 -1.25152612e+00 1.17531605e-01 -6.99495912e-01
-3.14017743e-01 1.10895026e+00 5.50814927e-01 -6.43340051e-02
6.56355619e-01 1.03270501e-01 7.84400553e-02 -1.53184772e-01
-2.79118776e-01 -7.36965477e-01 7.21804619e-01 5.20463824e-01
2.21954733e-01 1.33711919e-01 -3.59379858e-01 5.81515074e-01
8.97285417e-02 -3.39752078e-01 1.29435286e-01 8.62412572e-01
-9.28751469e-01 -1.37311912e+00 -6.20693445e-01 5.41045308e-01
-4.30891961e-01 1.46336257e-01 -4.78361815e-01 9.37977433e-01
7.12928414e-01 9.22371745e-01 2.20556691e-01 -1.42281398e-01
2.68474162e-01 6.42982721e-02 5.20369887e-01 -7.15295792e-01
-7.86334991e-01 1.96588904e-01 -1.33012041e-01 -5.50151348e-01
-5.19848347e-01 -2.77773291e-01 -1.11130524e+00 7.76153579e-02
-6.83596730e-01 4.50929135e-01 7.90139198e-01 8.88494074e-01
2.31915191e-01 6.47635579e-01 4.80649084e-01 -1.24374545e+00
-2.67944008e-01 -1.12315571e+00 -7.27612436e-01 5.48762619e-01
1.21087275e-01 -1.15527666e+00 -2.11818829e-01 -2.26579630e-03] | [9.60189437866211, 1.9727554321289062] |
d17776cb-157b-493e-827d-c8404a0c9100 | custom-edit-text-guided-image-editing-with | 2305.15779 | null | https://arxiv.org/abs/2305.15779v1 | https://arxiv.org/pdf/2305.15779v1.pdf | Custom-Edit: Text-Guided Image Editing with Customized Diffusion Models | Text-to-image diffusion models can generate diverse, high-fidelity images based on user-provided text prompts. Recent research has extended these models to support text-guided image editing. While text guidance is an intuitive editing interface for users, it often fails to ensure the precise concept conveyed by users. To address this issue, we propose Custom-Edit, in which we (i) customize a diffusion model with a few reference images and then (ii) perform text-guided editing. Our key discovery is that customizing only language-relevant parameters with augmented prompts improves reference similarity significantly while maintaining source similarity. Moreover, we provide our recipe for each customization and editing process. We compare popular customization methods and validate our findings on two editing methods using various datasets. | ['Sungroh Yoon', 'Junho Kim', 'Yunji Kim', 'Yunjey Choi', 'Jooyoung Choi'] | 2023-05-25 | null | null | null | null | ['text-guided-image-editing'] | ['computer-vision'] | [ 4.74501044e-01 -3.53308022e-01 5.31193390e-02 -4.48660672e-01
-5.12387156e-01 -8.94182861e-01 9.42358315e-01 2.46272326e-01
-4.25338328e-01 3.14140916e-01 4.27912623e-01 -4.14955795e-01
3.88868339e-02 -4.76657301e-01 -2.63864636e-01 -1.64171889e-01
3.65302652e-01 1.24643765e-01 2.96035945e-01 -2.17995405e-01
7.63683915e-01 5.51214576e-01 -1.21060061e+00 3.86079788e-01
1.24466443e+00 4.00688827e-01 6.38560474e-01 9.81300652e-01
-2.90630281e-01 7.34280169e-01 -7.49523044e-01 -3.71151149e-01
2.38369375e-01 -6.02562070e-01 -4.81306940e-01 2.20909745e-01
7.06590950e-01 -7.18437016e-01 -1.15427636e-01 8.67552638e-01
7.14247525e-01 3.40639442e-01 6.84976697e-01 -1.09747398e+00
-1.40123606e+00 3.81890595e-01 -8.28378677e-01 4.23429668e-01
7.45220423e-01 3.37473661e-01 5.02875507e-01 -8.94784570e-01
1.15316486e+00 1.21216583e+00 5.87855458e-01 6.20963573e-01
-1.56258202e+00 -6.71863675e-01 2.41400093e-01 -2.41759345e-01
-1.35379767e+00 -7.08387554e-01 6.28470302e-01 -7.82263756e-01
7.72579432e-01 3.60226095e-01 6.44189954e-01 1.05206609e+00
3.45445573e-01 3.91056955e-01 1.14066648e+00 -4.04315084e-01
-1.91091187e-02 2.88403928e-01 -1.91816255e-01 6.51598155e-01
-1.03629455e-01 9.30277109e-02 -5.53655624e-01 -2.03352168e-01
1.26101410e+00 -6.21902719e-02 -3.61271858e-01 2.69482587e-03
-1.51196802e+00 7.37305939e-01 -6.91592041e-03 1.56829640e-01
-4.44503784e-01 1.16915062e-01 3.07498604e-01 5.69312930e-01
8.12010288e-01 5.67986846e-01 2.03306302e-02 -2.70885110e-01
-1.25907934e+00 2.16858298e-01 5.18342316e-01 1.09887958e+00
7.22839236e-01 9.59423184e-03 -7.71527231e-01 1.07831740e+00
9.56111774e-02 4.31447983e-01 3.43472958e-01 -1.23054934e+00
1.67031303e-01 1.28170371e-01 2.91909128e-01 -1.26072025e+00
-9.76502597e-02 -1.02753438e-01 -6.66421711e-01 4.38159078e-01
1.94462627e-01 -1.82982594e-01 -9.44287479e-01 1.56587768e+00
8.78968015e-02 -2.28889465e-01 -4.67908770e-01 7.52891123e-01
6.77871883e-01 6.90854907e-01 3.21682751e-01 -7.97725320e-02
1.12154579e+00 -1.08407760e+00 -7.33309627e-01 6.37808889e-02
4.37656701e-01 -1.31896663e+00 1.62500632e+00 1.34459972e-01
-1.33453012e+00 -5.72455406e-01 -6.81270003e-01 -3.26309383e-01
-8.01398978e-02 2.14398101e-01 3.32660526e-01 5.98491788e-01
-1.58725333e+00 6.63353086e-01 -5.57581007e-01 -7.44536400e-01
2.17150658e-01 3.17741707e-02 -2.38015011e-01 -9.32656974e-02
-7.42724180e-01 8.30774605e-01 -2.81483024e-01 -5.40564954e-01
-6.16058350e-01 -1.00104642e+00 -5.53041816e-01 -3.07387352e-01
1.19971655e-01 -1.17705584e+00 1.36017990e+00 -1.24368966e+00
-1.72246432e+00 8.29143822e-01 -4.16412175e-01 4.35279235e-02
9.38916326e-01 -1.14330843e-01 -2.74942577e-01 3.19434792e-01
2.93516546e-01 1.12712419e+00 1.19391060e+00 -1.58879066e+00
-4.16996390e-01 1.25143841e-01 1.19203553e-01 4.35150534e-01
-4.53648418e-01 3.49642187e-01 -9.61508393e-01 -1.20204711e+00
-2.13229060e-01 -8.00798237e-01 -2.76970536e-01 7.11927712e-01
-4.36611712e-01 3.04066032e-01 9.68021393e-01 -7.77583241e-01
1.45004761e+00 -2.26846266e+00 -1.85733601e-01 3.06177050e-01
5.73055029e-01 -1.26950562e-01 -4.97610956e-01 6.88836455e-01
6.53567091e-02 4.33687121e-01 -1.29302114e-01 -6.39032185e-01
-2.14906886e-01 2.23012753e-02 -2.82887310e-01 1.15197301e-01
-7.78194144e-02 9.03273106e-01 -8.59902740e-01 -6.87177360e-01
4.12749141e-01 7.35417485e-01 -7.91785836e-01 7.42684454e-02
-7.92455897e-02 6.52767479e-01 -2.70280659e-01 3.36782873e-01
8.00672829e-01 -4.50014293e-01 -9.95790660e-02 -3.92255127e-01
-2.88171947e-01 -1.38369918e-01 -1.01543260e+00 1.74739993e+00
-5.51445007e-01 9.67285812e-01 1.21398196e-01 -6.94336146e-02
7.43833363e-01 4.57624272e-02 4.67202246e-01 -6.95686996e-01
-3.34815472e-01 -1.54815912e-01 -2.56336987e-01 -4.90604937e-01
9.24938202e-01 6.09703595e-03 4.82789695e-01 1.02714515e+00
-4.48600113e-01 -4.50175285e-01 3.71923745e-01 7.75917888e-01
8.47772002e-01 1.24876499e-01 8.20744187e-02 -3.61802280e-01
1.20266460e-01 -8.88949167e-03 -2.66796015e-02 1.16782939e+00
-4.93666641e-02 9.47561622e-01 1.66723326e-01 -4.35432866e-02
-1.32590222e+00 -1.01830268e+00 2.07022756e-01 1.30980468e+00
2.53598630e-01 -6.65184975e-01 -7.65499830e-01 -3.89381140e-01
-1.47919431e-01 8.74381244e-01 -6.80745602e-01 2.29050666e-01
-1.01971403e-01 -3.59210759e-01 4.46609169e-01 3.74328554e-01
3.67678314e-01 -8.62737954e-01 -7.85359561e-01 1.87386721e-01
-1.75186887e-01 -8.23743463e-01 -1.49394000e+00 -4.87734228e-01
-7.74028182e-01 -7.28069305e-01 -1.16799891e+00 -5.60534596e-01
1.22066081e+00 6.52834892e-01 1.15611196e+00 4.78313833e-01
-1.20591417e-01 9.25383687e-01 -3.33551675e-01 4.53224666e-02
-6.94291890e-01 -2.04554945e-01 -2.21510977e-01 -1.62847325e-01
-2.92939544e-01 -4.80881304e-01 -8.75219703e-01 4.35451657e-01
-1.28343058e+00 6.79616809e-01 3.02907318e-01 5.36162198e-01
5.82226872e-01 -2.57356673e-01 2.23405242e-01 -8.71980190e-01
1.54479349e+00 -7.29585439e-02 -2.60361850e-01 3.44986975e-01
-7.50504375e-01 -6.15645535e-02 5.25597870e-01 -7.58476079e-01
-1.28861451e+00 -1.22588113e-01 6.24825284e-02 -3.25367987e-01
6.31598234e-02 4.34119552e-01 5.32500446e-01 -3.80200773e-01
8.70625496e-01 2.53744304e-01 -6.16976619e-02 -2.59184808e-01
7.02522218e-01 3.66126239e-01 5.84140778e-01 -8.31164360e-01
8.10082436e-01 5.12743831e-01 -6.07724011e-01 -9.11567569e-01
-9.39028114e-02 -2.49758996e-02 -6.15027428e-01 -5.02145410e-01
7.86832929e-01 -6.76428199e-01 -3.20816934e-01 4.22228336e-01
-1.29000282e+00 -8.44179273e-01 -2.73938715e-01 2.21020311e-01
-3.82204920e-01 6.24238372e-01 -6.25339806e-01 -5.09850144e-01
-2.49155402e-01 -1.29415917e+00 1.00874591e+00 1.68662488e-01
-6.73796594e-01 -1.20581830e+00 -4.45294492e-02 -7.06851035e-02
9.34264362e-01 4.01343219e-02 8.01261425e-01 -3.69107537e-02
-4.93835360e-01 9.18764099e-02 -6.00471795e-01 -9.75620896e-02
3.78065050e-01 5.27322054e-01 -5.78309536e-01 -2.65574604e-01
-3.81493628e-01 5.56695573e-02 5.68820357e-01 4.72291380e-01
1.34364283e+00 -1.20430864e-01 -4.02517736e-01 6.33092165e-01
1.13042510e+00 1.84738845e-01 6.06677949e-01 2.61672318e-01
5.25015712e-01 2.80105919e-01 2.78579146e-01 6.40404940e-01
4.15144622e-01 6.46036506e-01 -3.33960682e-01 -2.70144790e-01
-3.44523787e-01 -4.89770561e-01 1.90491945e-01 7.66542614e-01
-1.37845296e-02 -3.98353398e-01 -7.23610997e-01 4.18950081e-01
-1.51318288e+00 -9.45496380e-01 -2.75017917e-02 1.98669815e+00
1.00827861e+00 -1.25729278e-01 -2.06521451e-02 -3.99992377e-01
7.11637676e-01 2.54520237e-01 -6.24169707e-01 -2.58851171e-01
-8.25862437e-02 -1.48199230e-01 3.26718837e-01 9.39905524e-01
-4.80272472e-01 1.01753604e+00 7.42822409e+00 7.10577726e-01
-1.39037156e+00 1.80565089e-01 6.80742145e-01 -1.80505127e-01
-9.99942660e-01 4.02212143e-02 -2.87031502e-01 4.61329699e-01
3.27870160e-01 -5.55702329e-01 6.76401019e-01 3.46538723e-01
8.63772869e-01 -4.61975664e-01 -1.07153213e+00 1.06799304e+00
2.60319322e-01 -1.54577231e+00 5.29776692e-01 -4.56595011e-02
8.42024982e-01 -3.56113821e-01 3.51027817e-01 -2.08836272e-01
5.57852745e-01 -9.49135900e-01 9.01406467e-01 6.97476387e-01
1.24388158e+00 -2.55273253e-01 -2.12275118e-01 9.02624615e-03
-1.00610948e+00 3.37437004e-01 -1.29508957e-01 1.28866091e-01
6.15301132e-01 6.18515611e-01 -5.01211882e-01 1.24305382e-01
5.52178085e-01 6.17742896e-01 -8.71694386e-01 9.59127307e-01
-9.24906954e-02 3.86928558e-01 -2.85388023e-01 2.23370671e-01
-7.16572255e-02 -3.92562568e-01 4.63790715e-01 1.54206109e+00
6.82537675e-01 2.91659027e-01 8.90922919e-02 1.09033620e+00
1.85394213e-01 2.15454191e-01 -6.03781044e-01 -1.67452544e-01
6.44182265e-01 1.13991618e+00 -9.71076190e-01 -5.22099435e-01
-3.40527326e-01 1.83283389e+00 8.20156708e-02 8.07905972e-01
-7.41163552e-01 -4.97731030e-01 5.51503956e-01 3.76804590e-01
1.68490410e-02 -7.02163100e-01 -4.45364535e-01 -1.05980837e+00
-2.46814638e-01 -7.28784442e-01 -5.29018454e-02 -1.39616466e+00
-1.26478708e+00 4.95722234e-01 1.69374526e-01 -8.20149541e-01
-1.24308644e-02 -1.90996915e-01 -8.13844502e-01 1.10440767e+00
-1.23104119e+00 -1.00478339e+00 -6.17733359e-01 7.97913551e-01
7.78522968e-01 1.66168943e-01 5.04563510e-01 4.82839257e-01
-3.31371635e-01 6.57022357e-01 9.11489874e-02 -2.15473533e-01
1.30308843e+00 -1.06493378e+00 6.73330426e-01 6.78926706e-01
-3.11389156e-02 1.07676947e+00 8.61155689e-01 -1.00631881e+00
-1.01941550e+00 -9.02205825e-01 8.32788825e-01 -5.22901237e-01
4.21121836e-01 -3.79988588e-02 -8.24407220e-01 7.04234719e-01
5.96692264e-01 -2.89184004e-01 5.59869170e-01 -3.86682570e-01
-2.07519293e-01 3.09251845e-01 -1.09384298e+00 1.15793824e+00
1.24120748e+00 -7.25502133e-01 -9.81173888e-02 2.20679328e-01
5.53150654e-01 -5.93006551e-01 -8.08740973e-01 -2.14453891e-01
5.95157266e-01 -1.03531671e+00 9.51301575e-01 -1.11671075e-01
5.58044255e-01 -3.91282856e-01 1.49306148e-01 -1.68414342e+00
-5.41104019e-01 -9.53706086e-01 2.84334570e-01 1.27185893e+00
3.04562479e-01 -5.29962361e-01 3.28706443e-01 9.62825119e-01
9.61372405e-02 -2.95580626e-01 -1.92886159e-01 -5.15752673e-01
-1.00766897e-01 -4.73179519e-01 5.61619937e-01 1.12162530e+00
-9.43256170e-02 -2.96898969e-02 -4.44462478e-01 -1.69401556e-01
4.30421680e-01 -3.44098985e-01 9.35100555e-01 -9.06020522e-01
-5.12673371e-02 -6.92442000e-01 2.79619396e-01 -1.56132150e+00
-2.07939714e-01 -6.28132701e-01 -1.64080620e-01 -1.79428637e+00
3.00544262e-01 -6.46232486e-01 2.14733630e-01 3.17341983e-01
-3.49085271e-01 4.22945887e-01 3.43522370e-01 5.66608787e-01
-3.87308776e-01 3.06912363e-01 1.61869049e+00 -1.80859063e-02
-3.77555102e-01 -4.28000599e-01 -1.00968957e+00 3.51542085e-01
7.03954995e-01 -2.53515095e-01 -8.38137031e-01 -8.91466260e-01
2.88040936e-01 -1.30778641e-01 4.16548640e-01 -6.94320500e-01
4.17522520e-01 -5.34686446e-01 4.85924333e-01 -3.55937332e-01
1.18793296e-02 -5.37285745e-01 3.37942600e-01 1.12742580e-01
-6.55863464e-01 5.88384032e-01 2.83086300e-01 3.77047211e-01
1.36884183e-01 -1.77053005e-01 7.00520217e-01 -2.98495796e-02
-6.24834716e-01 3.08910489e-01 -7.56813884e-01 1.88297480e-02
9.01239038e-01 -5.78548431e-01 -2.97915876e-01 -9.72988427e-01
-5.89064479e-01 -9.07878764e-03 1.08722758e+00 3.93853039e-01
7.21443057e-01 -1.26368129e+00 -7.45072126e-01 3.83812308e-01
1.21048830e-01 -4.21780050e-01 3.88490319e-01 6.78240836e-01
-5.83856702e-01 -9.45452303e-02 -1.68333530e-01 -5.57493806e-01
-1.34398079e+00 3.59838068e-01 3.31284493e-01 1.55282617e-01
-8.13386261e-01 7.05494225e-01 3.69481325e-01 -2.28245109e-01
-1.83659643e-02 -2.93684781e-01 3.67237255e-02 -1.20271422e-01
5.82078457e-01 1.71950340e-01 -2.92176098e-01 -4.77839857e-01
-8.58536921e-03 8.62691104e-01 -2.61155039e-01 -7.96431422e-01
8.67239296e-01 -7.99300849e-01 8.95631537e-02 1.61344722e-01
7.95172334e-01 4.53740418e-01 -1.56593072e+00 -9.61797088e-02
-5.65494120e-01 -8.72169197e-01 4.97101620e-02 -9.73886132e-01
-1.04879344e+00 7.26061106e-01 5.35778105e-01 8.41238052e-02
1.02857184e+00 -2.59488583e-01 6.28401637e-01 1.04347922e-01
-7.63046220e-02 -1.12101114e+00 4.45342928e-01 2.71097898e-01
1.15838838e+00 -1.02658367e+00 1.20474622e-01 -3.73285621e-01
-9.41623032e-01 9.79725242e-01 6.96533024e-01 2.36262769e-01
5.92209339e-01 3.46225739e-01 4.93752658e-01 -2.15364650e-01
-7.28186011e-01 3.01658422e-01 2.63304114e-01 9.88993049e-01
8.73066306e-01 -3.17935288e-01 -4.27820444e-01 2.18763314e-02
-1.00330532e-01 2.60685116e-01 5.90375543e-01 7.77733326e-01
-1.95052639e-01 -1.08595335e+00 -5.37255287e-01 5.46040297e-01
-5.38432226e-02 -3.86987835e-01 -3.85763347e-01 4.61748749e-01
-2.24985242e-01 1.11192894e+00 1.16291508e-01 -2.33682007e-01
2.88789988e-01 -3.44790936e-01 5.47034860e-01 -5.34247756e-01
-7.12175369e-01 -2.35504545e-02 -1.60548747e-01 -4.48285401e-01
-2.67076284e-01 -5.42201161e-01 -9.36924934e-01 -6.71055794e-01
-1.14220381e-01 -1.15882590e-01 5.84491432e-01 6.07037365e-01
8.07633579e-01 5.47479749e-01 3.53523165e-01 -1.20246506e+00
-8.24412480e-02 -7.74746895e-01 -3.06442589e-01 7.45054126e-01
2.21183226e-01 -4.44530308e-01 -9.12227184e-02 5.25732875e-01] | [11.354877471923828, -0.3734437823295593] |
73087b15-e38e-44c5-a9ab-a6553ca7b172 | generation-augmented-retrieval-for-open | 2009.08553 | null | https://arxiv.org/abs/2009.08553v4 | https://arxiv.org/pdf/2009.08553v4.pdf | Generation-Augmented Retrieval for Open-domain Question Answering | We propose Generation-Augmented Retrieval (GAR) for answering open-domain questions, which augments a query through text generation of heuristically discovered relevant contexts without external resources as supervision. We demonstrate that the generated contexts substantially enrich the semantics of the queries and GAR with sparse representations (BM25) achieves comparable or better performance than state-of-the-art dense retrieval methods such as DPR. We show that generating diverse contexts for a query is beneficial as fusing their results consistently yields better retrieval accuracy. Moreover, as sparse and dense representations are often complementary, GAR can be easily combined with DPR to achieve even better performance. GAR achieves state-of-the-art performance on Natural Questions and TriviaQA datasets under the extractive QA setup when equipped with an extractive reader, and consistently outperforms other retrieval methods when the same generative reader is used. | ['Yelong Shen', 'Weizhu Chen', 'Yuning Mao', 'Xiaodong Liu', 'Jianfeng Gao', 'Jiawei Han', 'Pengcheng He'] | 2020-09-17 | null | https://aclanthology.org/2021.acl-long.316 | https://aclanthology.org/2021.acl-long.316.pdf | acl-2021-5 | ['triviaqa'] | ['miscellaneous'] | [ 1.83322892e-01 3.20045173e-01 -6.93325698e-02 9.50599555e-04
-2.27170181e+00 -9.08945084e-01 9.81020153e-01 -6.02243431e-02
-3.81343096e-01 8.40783596e-01 8.87921393e-01 -1.72587037e-02
-2.00512245e-01 -8.16315770e-01 -8.43439579e-01 -2.71946818e-01
3.70007426e-01 1.28433180e+00 1.09204382e-01 -6.85075402e-01
2.06580848e-01 7.97946937e-03 -1.52732408e+00 6.57956362e-01
1.06648958e+00 6.92949951e-01 3.26388955e-01 7.24779129e-01
-4.01584595e-01 8.00857365e-01 -6.84554815e-01 -7.79030085e-01
2.70139247e-01 -3.14007342e-01 -1.35562873e+00 -2.32161552e-01
7.20975518e-01 -6.16980612e-01 -7.03592956e-01 6.20453477e-01
6.98726594e-01 5.64162254e-01 7.11388886e-01 -4.66614902e-01
-1.44801700e+00 9.12377894e-01 -8.86200443e-02 1.85917556e-01
9.60278392e-01 -2.46322211e-02 1.76617873e+00 -1.11608028e+00
1.04109085e+00 1.36734939e+00 7.05778301e-02 6.72649145e-01
-1.24935782e+00 -2.27509066e-01 -1.31429583e-01 2.96309978e-01
-1.38117075e+00 -4.30452496e-01 5.70115387e-01 3.36589098e-01
1.16345716e+00 3.10402483e-01 9.86216888e-02 1.30676186e+00
-4.24523681e-01 1.37756836e+00 7.03206301e-01 -6.94816649e-01
5.97856008e-02 -1.51944071e-01 2.18942314e-01 5.66254437e-01
3.59531455e-02 -2.84270614e-01 -4.64234114e-01 -5.95310688e-01
3.44549626e-01 2.20979583e-02 -5.01235366e-01 9.10549089e-02
-1.16917098e+00 1.21502459e+00 4.48768824e-01 2.52969831e-01
-6.72991812e-01 1.35622457e-01 2.04123616e-01 3.84662837e-01
4.11776125e-01 1.27110887e+00 -4.83385235e-01 -7.23697916e-02
-1.02442873e+00 8.34365308e-01 9.15143073e-01 1.20423341e+00
8.95591140e-01 -3.51947397e-01 -8.70231271e-01 1.06640112e+00
7.91122392e-03 8.52325439e-01 6.44836426e-01 -1.33057940e+00
6.26475871e-01 4.62316930e-01 2.82335132e-01 -6.21319711e-01
2.76401609e-01 -4.50397015e-01 -3.51326674e-01 -8.62214625e-01
2.26445511e-01 -7.72948191e-02 -1.10003877e+00 1.68551040e+00
1.02647573e-01 -2.23916203e-01 6.31422222e-01 9.06050920e-01
1.18062103e+00 1.02104878e+00 -3.10645159e-03 1.14408679e-01
1.52057862e+00 -1.15441406e+00 -8.61456513e-01 -4.29833174e-01
6.36571169e-01 -7.00054765e-01 1.32846189e+00 1.42136142e-01
-1.25208795e+00 -2.35339105e-01 -5.34361959e-01 -9.08682525e-01
-4.26488757e-01 1.72634229e-01 4.99787509e-01 6.94531873e-02
-1.10757530e+00 3.40077519e-01 -3.22501779e-01 -2.53609478e-01
1.98955730e-01 -1.53742041e-02 -2.37898245e-01 -9.35956061e-01
-1.51998603e+00 8.43550742e-01 2.50763148e-01 -1.86953008e-01
-1.21424067e+00 -8.70699823e-01 -9.73037899e-01 2.85690248e-01
7.75430262e-01 -1.22327411e+00 1.57163954e+00 -2.93287903e-01
-1.13606322e+00 7.93143630e-01 -5.32567620e-01 -5.73438466e-01
8.19701981e-03 -7.92825818e-01 -1.31244600e-01 7.68690586e-01
3.69726241e-01 9.35488105e-01 8.96174490e-01 -1.06336689e+00
-9.27091464e-02 -3.62948239e-01 5.60262203e-01 4.81418580e-01
-1.63460806e-01 -4.68299352e-02 -7.14672983e-01 -6.36836410e-01
1.80235133e-03 -7.96842456e-01 -1.05386667e-01 -5.98394394e-01
-4.70354319e-01 -7.11978018e-01 6.77599132e-01 -1.00000358e+00
1.17805731e+00 -1.76728976e+00 3.86384338e-01 -9.91744772e-02
2.30828077e-01 1.52421013e-01 -5.74082613e-01 9.44585681e-01
6.34597003e-01 1.70520514e-01 -1.11521676e-01 -4.50683802e-01
3.64634216e-01 4.07236725e-01 -1.02960360e+00 -3.65232766e-01
4.22474742e-01 1.41334152e+00 -1.23771322e+00 -5.64177096e-01
-3.94594312e-01 4.53230947e-01 -6.61840498e-01 4.87516016e-01
-8.07712138e-01 -2.24025436e-02 -8.24422002e-01 7.68464267e-01
1.57706648e-01 -6.09002411e-01 8.25429708e-02 4.55305316e-02
8.38345587e-01 7.91658044e-01 -5.73386133e-01 2.13455820e+00
-6.95894718e-01 3.96653533e-01 -1.15975000e-01 -4.33705389e-01
7.56178498e-01 4.23097730e-01 -1.41264722e-01 -7.85066962e-01
-3.95628393e-01 3.14077497e-01 -5.64943671e-01 -4.08294410e-01
1.34986126e+00 1.39664620e-01 -3.22298735e-01 7.88851500e-01
4.59947079e-01 -5.53472996e-01 3.54329973e-01 1.14638209e+00
1.47534466e+00 6.65159971e-02 -1.14877306e-01 3.88585217e-02
2.08482519e-01 1.71770692e-01 1.62466779e-01 1.26935112e+00
5.38231969e-01 7.63856530e-01 1.42527878e-01 -6.18723929e-02
-7.97814906e-01 -1.09800005e+00 1.78064615e-01 1.54675400e+00
-1.82796400e-02 -6.84288740e-01 -4.56634074e-01 -9.31665957e-01
1.39158234e-01 9.06937540e-01 -6.99091792e-01 -5.93616702e-02
-6.07214034e-01 -2.24222228e-01 7.78917313e-01 5.92493355e-01
3.52934122e-01 -1.20911765e+00 -2.07074225e-01 1.94107011e-01
-6.49175823e-01 -1.19333434e+00 -4.19646859e-01 -8.66794139e-02
-7.82051861e-01 -8.50797474e-01 -1.07719636e+00 -5.78126132e-01
5.13604283e-01 4.43355262e-01 1.91360176e+00 1.80307895e-01
-1.27076283e-01 8.68193686e-01 -8.59890819e-01 -4.80304137e-02
-3.73518109e-01 4.84154522e-01 -4.53736007e-01 -5.58993816e-01
4.14222836e-01 -3.08376044e-01 -6.78803563e-01 -9.83106866e-02
-1.16064239e+00 -3.67876351e-01 6.79557741e-01 1.16858923e+00
6.80491388e-01 -6.01505280e-01 8.20133984e-01 -1.05909979e+00
1.08841646e+00 -6.15211487e-01 -2.65374511e-01 6.94556475e-01
-4.33902860e-01 5.81655681e-01 3.81657004e-01 -2.84739494e-01
-1.31572747e+00 -5.63790321e-01 -3.24481651e-02 -5.76896846e-01
2.36995853e-02 6.48038030e-01 6.41244724e-02 3.93776655e-01
1.05664694e+00 3.10890406e-01 -3.66368383e-01 -6.78707719e-01
1.13818049e+00 5.66283286e-01 5.23944914e-01 -1.09354234e+00
7.45267153e-01 2.29419768e-01 -3.92970920e-01 -3.77478600e-01
-1.29462826e+00 -6.68212295e-01 -4.72767726e-02 3.27210873e-01
6.41992211e-01 -1.10094273e+00 7.11453184e-02 -2.93656170e-01
-1.20812106e+00 -9.11251605e-02 -6.24692738e-01 7.72380689e-03
-3.99971873e-01 3.08556467e-01 -8.68782818e-01 -6.56436443e-01
-9.89209473e-01 -9.35151219e-01 1.82282054e+00 2.37910867e-01
-1.67741567e-01 -7.99399376e-01 1.87890410e-01 9.45762575e-01
4.39658999e-01 -2.58835286e-01 9.56480682e-01 -1.12542653e+00
-9.63976085e-01 -2.42595688e-01 -2.24964291e-01 8.94875899e-02
-8.47101212e-02 -4.29662257e-01 -1.17667270e+00 -1.58139125e-01
-4.05899912e-01 -1.15016985e+00 1.19524717e+00 -5.07693999e-02
1.02643108e+00 -7.12984800e-01 -2.20173672e-01 4.09853496e-02
1.25084531e+00 -2.70999521e-01 7.84395397e-01 4.00074981e-02
4.60241944e-01 5.42463481e-01 5.27165711e-01 1.81184217e-01
4.39796358e-01 4.06996101e-01 6.19964600e-02 2.80275971e-01
-3.00551623e-01 -5.94707549e-01 1.98863804e-01 6.50626719e-01
2.17511803e-01 -3.89036149e-01 -7.57400095e-01 9.53102469e-01
-1.62932122e+00 -1.01976645e+00 5.23109436e-01 1.85527551e+00
1.28362155e+00 -3.90765011e-01 -1.84900969e-01 -3.29317480e-01
3.24531376e-01 2.80189663e-01 -3.80968392e-01 -7.83145353e-02
-2.61149317e-01 8.11764956e-01 -5.70090935e-02 5.51747680e-01
-5.93850553e-01 1.24595928e+00 6.87841463e+00 1.02317393e+00
-3.57586771e-01 1.97704136e-01 3.47012937e-01 -2.75818139e-01
-9.73589420e-01 1.00072101e-01 -9.25579667e-01 2.46892385e-02
8.36435318e-01 -2.30818644e-01 5.44955432e-01 8.70979071e-01
-5.77390194e-01 2.31707450e-02 -1.17394435e+00 7.58478642e-01
5.32225490e-01 -1.48844099e+00 6.48323238e-01 -1.03678098e-02
8.32927644e-01 1.48098767e-01 1.35405794e-01 9.55654740e-01
7.20537126e-01 -9.67604876e-01 2.23049477e-01 4.92171884e-01
5.95369518e-01 -5.29683232e-01 8.08247089e-01 3.29715431e-01
-6.03583157e-01 -1.23175815e-01 -6.18368745e-01 2.90145040e-01
1.91351742e-01 4.20221955e-01 -1.06839669e+00 8.51698935e-01
4.02045846e-01 2.12804481e-01 -7.94924796e-01 5.26478827e-01
-5.31566024e-01 6.16848230e-01 -3.77430916e-01 -1.86442778e-01
5.26036382e-01 1.29880548e-01 5.38782835e-01 1.14982271e+00
1.76821500e-01 3.39108318e-01 8.49241242e-02 1.11933672e+00
-7.54100740e-01 2.06669286e-01 -7.73441195e-01 -4.02886927e-01
7.41318107e-01 1.17162478e+00 4.35297750e-02 -8.50906014e-01
-1.52085796e-01 9.92268264e-01 5.79339743e-01 5.67840934e-01
-1.98445380e-01 -5.42698085e-01 2.73894310e-01 -1.65827736e-01
5.15183747e-01 1.16559073e-01 2.90495455e-01 -1.47801816e+00
1.61579981e-01 -1.33659101e+00 8.63263369e-01 -1.16400743e+00
-1.57605124e+00 6.63440704e-01 1.05318256e-01 -7.77883589e-01
-1.02153695e+00 -9.29665044e-02 -9.90783572e-02 9.58489537e-01
-1.76534927e+00 -1.23435414e+00 -5.08566685e-02 7.42958128e-01
8.12613308e-01 -8.12438205e-02 1.16520977e+00 -5.07509634e-02
-2.48591360e-02 5.59263587e-01 2.46849451e-02 2.93704897e-01
7.96983063e-01 -1.46078384e+00 3.44312221e-01 8.06899130e-01
8.00137043e-01 1.04321969e+00 4.64442194e-01 -7.25912452e-01
-1.86531115e+00 -9.42057967e-01 1.05174100e+00 -8.24075699e-01
6.36373341e-01 -2.25209460e-01 -1.09023297e+00 6.80402100e-01
6.88658059e-01 -8.95970836e-02 6.80331349e-01 4.70661908e-01
-7.67586648e-01 1.33630112e-01 -1.00387442e+00 5.53083003e-01
8.81606340e-01 -1.23120868e+00 -1.60686064e+00 8.52906942e-01
1.32855618e+00 -5.59248686e-01 -7.29906261e-01 2.42433190e-01
1.60189226e-01 -3.38543594e-01 1.06378126e+00 -9.41240668e-01
6.18372500e-01 -1.24169029e-02 -5.34447193e-01 -1.26711357e+00
-1.58495128e-01 -8.31853032e-01 -6.59436285e-01 1.08090019e+00
6.49943233e-01 -3.50918502e-01 5.30792654e-01 7.87072897e-01
-1.31105706e-01 -7.10619390e-01 -6.95125163e-01 -6.53390586e-01
1.39194548e-01 -1.34935662e-01 7.83362925e-01 7.62765825e-01
-6.83881938e-02 7.77650177e-01 2.84770820e-02 8.78770798e-02
4.26767200e-01 4.69815612e-01 7.54518628e-01 -8.92762125e-01
-5.47029436e-01 2.42514834e-02 8.75717252e-02 -1.43218207e+00
5.32324672e-01 -9.40773964e-01 7.79023021e-02 -1.79733026e+00
2.54293829e-01 -3.17088664e-01 -8.51023644e-02 4.72203612e-01
-6.37870848e-01 2.99249530e-01 9.56821367e-02 2.27801591e-01
-1.15704334e+00 8.69880855e-01 1.27426577e+00 -2.90989906e-01
-8.42994675e-02 -5.68120122e-01 -1.23131204e+00 3.04183513e-02
2.35998705e-01 -3.02121162e-01 -6.10287547e-01 -9.81574059e-01
4.29541409e-01 2.66485989e-01 2.90928125e-01 -5.08895814e-01
2.46743038e-01 3.63834888e-01 2.42617071e-01 -5.87454379e-01
6.48581862e-01 -2.74150431e-01 -3.88245463e-01 -2.52322525e-01
-8.88176382e-01 1.80687621e-01 4.64423560e-02 7.42586851e-01
-5.08720815e-01 -4.12155777e-01 -3.06158178e-02 -5.30259013e-01
-4.31934178e-01 1.19605616e-01 -8.27090591e-02 8.63661826e-01
8.54846612e-02 3.63319218e-01 -7.55608380e-01 -8.41624320e-01
-5.02030909e-01 3.93719941e-01 2.38716960e-01 2.39141524e-01
7.34639704e-01 -1.29973948e+00 -9.46595132e-01 -3.07114273e-01
3.80738705e-01 1.66571304e-01 2.34589547e-01 1.11887418e-01
-4.25568521e-02 7.72195458e-01 5.87299287e-01 -3.58411938e-01
-8.93359900e-01 5.11567593e-01 -1.22282960e-01 -9.06813025e-01
-5.50811350e-01 9.24759448e-01 3.33064161e-02 -4.27999526e-01
4.35820818e-02 -1.14612117e-01 -7.75529295e-02 1.07648499e-01
8.47963691e-01 1.14021309e-01 2.87636429e-01 -1.80152506e-01
5.64190969e-02 1.72347948e-01 -6.25667453e-01 -5.16606987e-01
1.29022336e+00 1.62555929e-02 -5.13762236e-02 -5.06610982e-02
1.09494090e+00 2.01983526e-01 -7.52638519e-01 -7.74725199e-01
-6.18612133e-02 -4.53848392e-01 2.41883427e-01 -1.19139552e+00
-6.28369927e-01 7.07903206e-01 -8.77809748e-02 1.20488986e-01
1.07860279e+00 5.11381507e-01 1.28463590e+00 1.28975594e+00
4.60550070e-01 -8.40209723e-01 4.79559541e-01 7.18466163e-01
1.23668778e+00 -1.20879829e+00 -3.87715623e-02 -8.57910588e-02
-8.11470807e-01 7.49811351e-01 3.95115465e-01 -2.00680673e-01
6.40593097e-02 -3.22372526e-01 -7.65272975e-02 -4.38019663e-01
-1.13696933e+00 -5.79817832e-01 7.16313303e-01 5.37187338e-01
2.40851328e-01 -1.91027984e-01 -3.37704234e-02 6.62467420e-01
-3.35794121e-01 -1.68502554e-01 2.92853594e-01 1.13498545e+00
-2.74821609e-01 -1.08300829e+00 -1.88351154e-01 5.61831415e-01
-7.06522167e-01 -7.78532028e-01 -6.48793519e-01 6.52037323e-01
-6.10109091e-01 9.98310864e-01 -1.54771671e-01 6.78177252e-02
7.39462823e-02 6.15903437e-01 6.15242004e-01 -1.02316201e+00
-7.42403567e-01 -1.14025347e-01 5.25843203e-01 -7.16049194e-01
-2.36196414e-01 -4.42180246e-01 -9.36630189e-01 1.64434820e-01
-4.98649687e-01 7.23204970e-01 3.44822139e-01 9.73496795e-01
9.79790330e-01 3.59830284e-03 3.86169612e-01 -3.95270854e-01
-9.33882892e-01 -1.26372099e+00 -3.47195238e-01 5.93395889e-01
2.57095367e-01 -4.39648688e-01 -4.99252707e-01 -6.62185699e-02] | [11.379488945007324, 7.79613733291626] |
9d3e7daf-0cd2-4d0d-8c6a-2f38149dc058 | evaluating-diverse-knowledge-sources-for | 2208.09554 | null | https://arxiv.org/abs/2208.09554v3 | https://arxiv.org/pdf/2208.09554v3.pdf | Integrating Diverse Knowledge Sources for Online One-shot Learning of Novel Tasks | Autonomous agents are able to draw on a wide variety of potential sources of task knowledge; however current approaches invariably focus on only one or two. Here we investigate the challenges and impact of exploiting diverse knowledge sources to learn online, in one-shot, new tasks for a simulated office mobile robot. The resulting agent, developed in the Soar cognitive architecture, uses the following sources of domain and task knowledge: interaction with the environment, task execution and search knowledge, human natural language instruction, and responses retrieved from a large language model (GPT-3). We explore the distinct contributions of these knowledge sources and evaluate the performance of different combinations in terms of learning correct task knowledge and human workload. Results show that an agent's online integration of diverse knowledge sources improves one-shot task learning overall, reducing human feedback needed for rapid and reliable task learning. | ['John E. Laird', 'Peter Lindes', 'Robert E. Wray', 'James R. Kirk'] | 2022-08-19 | null | null | null | null | ['one-shot-learning'] | ['methodology'] | [ 2.37705737e-01 2.46911664e-02 2.58049350e-02 -9.20576602e-02
-5.60100555e-01 -7.71621466e-01 7.89283872e-01 3.38202000e-01
-8.25908303e-01 8.74522626e-01 3.05325501e-02 -3.16075265e-01
-6.11499846e-01 -4.68478858e-01 -4.16856349e-01 -2.33710364e-01
-2.12969538e-03 8.16154003e-01 6.77800238e-01 -6.88245714e-01
6.06244326e-01 4.55587715e-01 -1.92249596e+00 1.27687246e-01
9.59233820e-01 5.59949219e-01 1.03032315e+00 7.22789228e-01
-1.45110533e-01 1.48599088e+00 -7.83743620e-01 2.96807736e-01
2.52228409e-01 4.54263203e-03 -1.05658448e+00 -1.14975162e-01
-2.18709961e-01 -3.47018123e-01 -3.45975254e-03 6.93279386e-01
6.96418941e-01 5.50922394e-01 4.03784484e-01 -1.05439401e+00
-3.89764071e-01 2.69553155e-01 1.59147516e-01 4.62867200e-01
6.88773036e-01 5.28714240e-01 2.16065869e-01 -6.71613634e-01
7.43269801e-01 1.18538499e+00 4.56133574e-01 4.37917799e-01
-8.13404262e-01 -1.92221403e-01 8.56707990e-02 4.03547704e-01
-1.02340734e+00 -6.66957319e-01 2.99241424e-01 -5.64921319e-01
1.64633703e+00 -2.77119458e-01 3.77999425e-01 1.11872613e+00
2.44451180e-01 6.40574634e-01 1.49465716e+00 -8.11277509e-01
3.57231379e-01 5.28526366e-01 3.36286366e-01 7.78370917e-01
9.17046741e-02 4.43121672e-01 -1.25617933e+00 -1.73024893e-01
5.26111245e-01 -2.32083187e-01 -6.65762722e-02 -6.13227069e-01
-1.34146047e+00 3.89555424e-01 1.65138766e-02 3.66834670e-01
-6.62755430e-01 -2.19739884e-01 3.54750127e-01 7.52067447e-01
-7.99753964e-02 9.21119809e-01 -7.25828826e-01 -6.57320499e-01
7.37584159e-02 1.91378117e-01 1.23746693e+00 1.16364300e+00
1.13886738e+00 5.71812578e-02 -2.91653007e-01 8.42162311e-01
1.50788769e-01 7.82196760e-01 8.41928899e-01 -1.32550704e+00
3.39115024e-01 7.55417824e-01 6.01363182e-01 -7.10457623e-01
-6.97465241e-01 -1.25874162e-01 3.87703866e-01 4.19617832e-01
3.62917960e-01 -2.51740485e-01 -8.52640033e-01 1.33283579e+00
4.03096497e-01 -4.01552320e-01 4.47991073e-01 8.13440681e-01
5.03504515e-01 2.30333060e-01 3.02478582e-01 -4.60731313e-02
1.35186172e+00 -1.19967055e+00 -6.46808445e-01 -7.70221531e-01
7.61654735e-01 -6.53608263e-01 1.02758646e+00 4.55139458e-01
-1.10448265e+00 -6.72986746e-01 -9.42025244e-01 -3.47082391e-02
-8.59154522e-01 -1.09506100e-01 3.92096758e-01 3.77054513e-01
-1.29023623e+00 3.17175806e-01 -6.44073427e-01 -5.42439818e-01
-2.29568020e-01 3.55670244e-01 -1.66142002e-01 -4.21136320e-01
-1.06620252e+00 1.73779154e+00 6.28481030e-01 -1.35761276e-01
-1.25392199e+00 -3.34253877e-01 -7.28223562e-01 4.43790033e-02
9.63434100e-01 -8.72464418e-01 1.84462261e+00 -7.87588775e-01
-1.88566375e+00 4.75153178e-01 -5.41800633e-02 -2.77242094e-01
2.24895656e-01 -3.38866949e-01 2.28301082e-02 1.25541225e-01
9.48342308e-02 4.60432827e-01 6.28358543e-01 -1.34974313e+00
-9.62980211e-01 -4.33839828e-01 2.48567700e-01 6.90751731e-01
-5.49439667e-03 -1.85530894e-02 1.07268892e-01 6.53281212e-02
-4.30091858e-01 -1.12252235e+00 -2.26928350e-02 -7.49686539e-01
4.38552111e-01 -3.32275182e-01 4.00393635e-01 -5.41324377e-01
7.81209767e-01 -1.90423000e+00 1.35033697e-01 1.73408702e-01
1.25223652e-01 3.50658655e-01 -4.86289769e-01 6.88311100e-01
5.30252099e-01 -1.73785925e-01 2.98288852e-01 7.91313723e-02
2.40121722e-01 3.86435896e-01 1.14997976e-01 -4.41120207e-01
-6.95406944e-02 1.00449741e+00 -1.26747489e+00 -3.15206498e-01
2.10586011e-01 3.38911504e-01 -3.13615911e-02 3.14473182e-01
-3.66167217e-01 3.21923047e-01 -6.97852015e-01 3.18294793e-01
-1.88296810e-01 -1.40468419e-01 4.60151494e-01 5.87255657e-01
-2.42488414e-01 4.97659951e-01 -8.58649790e-01 1.89143610e+00
-8.42808366e-01 3.24227989e-01 2.24560559e-01 -4.71205682e-01
9.52201903e-01 4.17414933e-01 9.69410390e-02 -1.27694952e+00
5.60047664e-02 3.22669804e-01 4.04436618e-01 -8.19143653e-01
4.83481258e-01 2.96513319e-01 1.13817334e-01 8.07465136e-01
2.36518905e-01 -2.14312561e-02 2.86293626e-01 2.13689953e-01
1.54382133e+00 1.67542800e-01 7.22812116e-01 -2.29239538e-01
1.54772475e-01 6.95757926e-01 -2.64701601e-02 1.16250896e+00
-5.06088674e-01 -3.64865601e-01 -3.58172618e-02 -4.67256844e-01
-5.53153336e-01 -8.38072240e-01 5.40006816e-01 2.06412864e+00
1.06409870e-01 -2.61262208e-01 -6.41868711e-01 -4.50549215e-01
-8.16125944e-02 1.06608641e+00 -2.10914791e-01 -4.41303670e-01
-5.25061905e-01 -2.14674860e-01 6.43733963e-02 3.61373097e-01
2.85060912e-01 -1.78979039e+00 -1.61121845e+00 3.27139735e-01
-2.78568655e-01 -1.00833082e+00 -2.06711456e-01 5.21707356e-01
-6.98329926e-01 -1.08424807e+00 -2.87300259e-01 -8.00297439e-01
4.15160477e-01 8.05822015e-01 1.40350497e+00 5.85181192e-02
-2.43255943e-01 1.19097710e+00 -5.17564654e-01 -8.33873212e-01
-7.28747606e-01 1.86387360e-01 2.06076890e-01 -6.51454210e-01
7.37348199e-01 -3.81001860e-01 -1.02604441e-01 2.42079496e-01
-4.63008821e-01 2.99909245e-02 1.14797604e+00 7.72486150e-01
-1.37048289e-02 7.32773840e-02 7.58666456e-01 -8.28910530e-01
1.30302668e+00 -5.28364599e-01 -4.65118855e-01 4.95294392e-01
-8.98569107e-01 2.64504224e-01 9.28307250e-02 -4.66891080e-01
-1.41853869e+00 6.21105656e-02 6.39916778e-01 -1.37281314e-01
-5.63061535e-01 8.16789150e-01 3.28717619e-01 -2.70077050e-01
9.76784527e-01 5.28191209e-01 1.26557469e-01 -2.46133924e-01
2.96257854e-01 6.20540261e-01 3.41674417e-01 -1.00795722e+00
3.08388144e-01 -2.23897740e-01 -4.96409774e-01 -6.71439469e-01
-5.72596431e-01 -6.67127550e-01 -5.40683210e-01 -3.43588948e-01
5.33883929e-01 -8.97515595e-01 -1.06544769e+00 4.11005050e-01
-1.11025512e+00 -1.08384061e+00 -2.33300999e-01 4.19634700e-01
-8.22904289e-01 -1.76011831e-01 -3.47559392e-01 -9.20195639e-01
-1.02846816e-01 -1.18171382e+00 7.40995526e-01 3.88038009e-01
-3.42556715e-01 -8.40190709e-01 7.66132921e-02 5.31229973e-01
9.06348586e-01 -4.31429833e-01 9.46592987e-01 -1.08526766e+00
-8.03548098e-01 2.14708876e-03 -1.33250207e-01 -1.05703630e-01
1.02219045e-01 -9.79425490e-01 -8.68474364e-01 -3.05179805e-01
1.89624146e-01 -8.17182839e-01 2.47473195e-01 -1.32323295e-01
2.04380408e-01 -2.37014726e-01 -2.93147504e-01 -4.27798510e-01
1.22729206e+00 5.09659231e-01 1.58318251e-01 7.43052661e-01
2.87154526e-01 8.45496893e-01 7.50481248e-01 2.69662261e-01
7.24604845e-01 6.30256712e-01 3.25306207e-02 6.11412466e-01
-1.63360327e-01 -2.28983536e-02 6.12238050e-01 7.43601203e-01
-3.75932157e-01 1.68740436e-01 -1.44199562e+00 6.14028394e-01
-2.05240130e+00 -6.64856732e-01 5.75399399e-01 2.11231327e+00
9.37640846e-01 1.92120716e-01 2.24938504e-02 -3.54681611e-01
1.49419591e-01 -3.10870081e-01 -8.16902936e-01 -3.23056698e-01
4.44042057e-01 9.48378816e-02 3.36470217e-01 6.06797159e-01
-4.55282390e-01 9.97636735e-01 6.81971264e+00 4.19365585e-01
-6.72446251e-01 3.65561575e-01 -2.01575994e-01 -9.27334800e-02
2.69568264e-01 -3.68256003e-01 -7.72789240e-01 1.63197652e-01
1.31951010e+00 -4.04303223e-01 9.20979559e-01 1.04056561e+00
7.60990009e-02 -7.58692324e-01 -1.04192185e+00 5.08196712e-01
2.36103505e-01 -8.14847231e-01 -2.30766207e-01 -1.14878692e-01
5.93579888e-01 4.33490694e-01 -2.04676330e-01 8.35421801e-01
9.85153973e-01 -8.34433138e-01 6.37011826e-01 7.90615559e-01
1.01193160e-01 -4.75597858e-01 6.74273372e-01 1.06369472e+00
-6.82575226e-01 -4.10191357e-01 1.07339628e-01 -4.63906348e-01
-9.48850140e-02 -3.02550644e-01 -1.35952270e+00 4.94069189e-01
5.93913555e-01 1.06096387e-01 -7.15662479e-01 6.55582011e-01
-1.81317985e-01 6.02742843e-03 -6.83727935e-02 -3.66698682e-01
-2.55414303e-02 1.78953305e-01 4.12776351e-01 8.90715480e-01
8.94150287e-02 4.30097103e-01 7.16437697e-01 3.88658822e-01
3.69877666e-01 -1.56856433e-01 -9.41216528e-01 -3.31979953e-02
9.54974771e-01 9.00427818e-01 -4.91495162e-01 -5.59481442e-01
-3.97701561e-01 6.45638049e-01 6.59319818e-01 6.65429175e-01
-1.65680721e-01 -3.37071627e-01 3.25462312e-01 -4.98032309e-02
6.34458810e-02 -5.61032712e-01 -3.49990241e-02 -5.17330468e-01
9.70569327e-02 -1.41793334e+00 6.42881542e-02 -1.23971510e+00
-9.38827097e-01 5.36212444e-01 2.19219252e-01 -5.75651050e-01
-6.32636905e-01 -8.43639374e-01 3.50868888e-02 9.91384685e-01
-1.76755321e+00 -7.42646396e-01 -6.09095156e-01 5.48138857e-01
8.54692936e-01 -5.53787053e-01 1.15814602e+00 -1.84415132e-01
-1.22592531e-01 -7.07248822e-02 -2.40526810e-01 -4.40223873e-01
7.65860319e-01 -1.09205902e+00 6.40677512e-02 2.89284229e-01
-2.21019283e-01 8.92661572e-01 5.88201821e-01 -8.83817196e-01
-1.80492318e+00 -5.57715178e-01 6.98668361e-01 -9.48815405e-01
5.57622313e-01 -3.97732444e-02 -9.03899431e-01 8.71291339e-01
4.47123766e-01 -5.05947351e-01 4.26972806e-01 2.26856217e-01
-3.42758387e-01 2.25620165e-01 -9.02028501e-01 5.34676790e-01
1.13638890e+00 -6.34132504e-01 -1.06120837e+00 3.02530229e-01
7.43171930e-01 -3.71152788e-01 -6.50396883e-01 1.50766104e-01
4.04088974e-01 -8.89152050e-01 9.75294530e-01 -5.45734704e-01
9.04938057e-02 -2.81043112e-01 2.40627844e-02 -1.63317060e+00
-4.04482991e-01 -5.64819753e-01 -2.05915555e-01 4.89984244e-01
4.20367122e-01 -9.10731256e-01 1.85798660e-01 8.95595729e-01
-3.39917213e-01 -3.03976566e-01 -5.77240467e-01 -8.21417987e-01
-5.15045404e-01 -8.20010379e-02 3.22323412e-01 7.83785582e-01
3.72846007e-01 8.81716430e-01 -4.61186767e-02 -4.59230989e-02
2.54434556e-01 -1.87206224e-01 9.06232715e-01 -1.50655091e+00
-1.13853149e-01 -3.49868685e-01 5.43103814e-02 -7.89016664e-01
1.68283999e-01 -7.08095908e-01 6.15077436e-01 -1.63901234e+00
1.49733871e-01 -4.98818636e-01 -3.51153821e-01 6.69753253e-01
-1.38112530e-01 -5.90488970e-01 4.24212456e-01 3.07670206e-01
-1.15245259e+00 1.65659785e-01 1.19704282e+00 2.50929385e-01
-6.18570030e-01 -3.77924472e-01 -5.49203515e-01 7.88988769e-01
7.36769915e-01 -3.07127625e-01 -6.78608179e-01 -7.18187869e-01
2.88946331e-01 1.79315135e-01 2.22950995e-01 -1.22388589e+00
9.61861253e-01 -3.38236332e-01 1.34668022e-01 4.89274822e-02
4.39068645e-01 -9.96591449e-01 -4.73675802e-02 3.60955596e-01
-4.55535263e-01 3.46842498e-01 5.38253069e-01 7.31062233e-01
5.32580949e-02 -3.64871383e-01 2.70035565e-01 -8.12267423e-01
-1.40529060e+00 -5.53818107e-01 -8.56186330e-01 -4.01458479e-02
1.08035088e+00 -3.17503601e-01 -8.23636353e-01 -1.64514512e-01
-7.65659213e-01 4.45958406e-01 3.59455973e-01 6.68333352e-01
4.15841520e-01 -7.25331128e-01 -4.12588030e-01 1.70554549e-01
2.82346308e-01 -1.78190544e-01 7.88751803e-03 7.03742027e-01
-1.87323287e-01 8.05641592e-01 -7.06270397e-01 -2.34829590e-01
-9.18499351e-01 6.74922705e-01 1.81818441e-01 -2.42637858e-01
-2.01532796e-01 3.99904639e-01 -3.86191905e-02 -6.67308390e-01
3.57865840e-01 3.61466818e-02 -4.44765657e-01 -5.94461747e-02
6.05819762e-01 5.05691946e-01 2.66374201e-01 -2.53338724e-01
-2.09176674e-01 2.98751444e-01 -1.79966629e-01 -3.78800511e-01
8.74841571e-01 -3.22694182e-01 1.33859992e-01 8.31394076e-01
2.64884114e-01 -1.93506226e-01 -1.03004086e+00 -6.22685671e-01
5.48857808e-01 -2.67365873e-01 -1.51188016e-01 -1.59452879e+00
-1.09780528e-01 5.53196549e-01 5.25362015e-01 1.45769358e-01
8.50857615e-01 -6.67747706e-02 3.81131798e-01 1.35783184e+00
1.17677402e+00 -1.41274154e+00 5.81741691e-01 9.96335268e-01
7.30493009e-01 -1.40296960e+00 -8.96553695e-02 -1.91593498e-01
-7.92721093e-01 9.70347822e-01 9.38107073e-01 3.18920553e-01
2.77532816e-01 -1.41228344e-02 4.95155543e-01 -3.82918417e-01
-1.25672972e+00 -5.27799249e-01 -5.42724431e-02 9.52620327e-01
1.89465120e-01 -1.75236747e-01 8.27543586e-02 4.71382886e-01
-3.48614007e-02 4.03569758e-01 5.21850407e-01 1.69834554e+00
-8.66564095e-01 -8.12687337e-01 -2.37857670e-01 2.82538384e-01
1.06595255e-01 3.94145697e-02 -5.35079539e-01 6.46676719e-01
-1.15147457e-01 1.11872673e+00 -3.62217903e-01 -2.99995430e-02
7.37195432e-01 7.19202101e-01 5.31715274e-01 -9.81529117e-01
-7.27756143e-01 -3.01135123e-01 3.58519763e-01 -7.14606047e-01
-4.32512850e-01 -4.82836068e-01 -1.21277952e+00 7.14180470e-02
-1.21551216e-01 8.61680955e-02 8.14818799e-01 1.05591893e+00
7.92312860e-01 7.75245070e-01 -3.93800400e-02 -8.93880129e-01
-8.83120835e-01 -1.18817472e+00 -2.50723064e-02 1.65204555e-01
2.17537105e-01 -1.00472498e+00 -5.15140034e-02 3.49985391e-01] | [4.231500148773193, 1.163480281829834] |
ee701544-9d3b-44f7-b0ee-1ee49814f2ba | recurrent-network-models-for-human-dynamics | 1508.00271 | null | http://arxiv.org/abs/1508.00271v2 | http://arxiv.org/pdf/1508.00271v2.pdf | Recurrent Network Models for Human Dynamics | We propose the Encoder-Recurrent-Decoder (ERD) model for recognition and
prediction of human body pose in videos and motion capture. The ERD model is a
recurrent neural network that incorporates nonlinear encoder and decoder
networks before and after recurrent layers. We test instantiations of ERD
architectures in the tasks of motion capture (mocap) generation, body pose
labeling and body pose forecasting in videos. Our model handles mocap training
data across multiple subjects and activity domains, and synthesizes novel
motions while avoid drifting for long periods of time. For human pose labeling,
ERD outperforms a per frame body part detector by resolving left-right body
part confusions. For video pose forecasting, ERD predicts body joint
displacements across a temporal horizon of 400ms and outperforms a first order
motion model based on optical flow. ERDs extend previous Long Short Term Memory
(LSTM) models in the literature to jointly learn representations and their
dynamics. Our experiments show such representation learning is crucial for both
labeling and prediction in space-time. We find this is a distinguishing feature
between the spatio-temporal visual domain in comparison to 1D text, speech or
handwriting, where straightforward hard coded representations have shown
excellent results when directly combined with recurrent units. | ['Sergey Levine', 'Panna Felsen', 'Katerina Fragkiadaki', 'Jitendra Malik'] | 2015-08-02 | recurrent-network-models-for-human-dynamics-1 | http://openaccess.thecvf.com/content_iccv_2015/html/Fragkiadaki_Recurrent_Network_Models_ICCV_2015_paper.html | http://openaccess.thecvf.com/content_iccv_2015/papers/Fragkiadaki_Recurrent_Network_Models_ICCV_2015_paper.pdf | iccv-2015-12 | ['human-pose-forecasting', 'human-dynamics'] | ['computer-vision', 'computer-vision'] | [ 1.13994151e-01 1.83230594e-01 -4.72976685e-01 -2.88148765e-02
-5.18369257e-01 -2.84203768e-01 6.50836408e-01 -6.46941125e-01
-2.83697367e-01 4.51627791e-01 9.08990681e-01 3.49216402e-01
2.31481388e-01 -1.91648141e-01 -8.57491255e-01 -5.67316949e-01
-4.32377905e-01 3.47084045e-01 1.62304074e-01 -2.02229053e-01
-2.88189113e-01 3.91012549e-01 -1.40931356e+00 4.30005908e-01
-2.98070759e-01 7.90878773e-01 -2.22480278e-02 1.37878323e+00
5.96477509e-01 1.37688744e+00 -4.46419924e-01 -1.62633464e-01
1.89077586e-01 -6.05047643e-01 -9.06884611e-01 3.22481275e-01
7.44843245e-01 -4.95509058e-01 -1.02972484e+00 1.53005034e-01
1.00295830e+00 3.81305456e-01 7.78406203e-01 -9.13955092e-01
-4.94069606e-01 4.28714193e-02 -3.21011573e-01 3.49924982e-01
8.28875005e-01 4.45482314e-01 6.49614394e-01 -7.88751841e-01
1.22060251e+00 1.47874427e+00 1.13920796e+00 1.16833663e+00
-1.03608179e+00 -4.01159413e-02 2.07342580e-01 6.13656193e-02
-1.04426372e+00 -6.23556316e-01 5.71084201e-01 -9.35200214e-01
1.38016343e+00 -1.15018442e-01 9.51766789e-01 1.83036053e+00
7.01838791e-01 1.17093098e+00 -9.86131653e-03 -9.02936012e-02
-2.31588483e-01 -8.77864718e-01 -2.89988846e-01 9.15435493e-01
-1.66537777e-01 3.01065624e-01 -9.39513326e-01 2.85656989e-01
1.25610828e+00 1.34091839e-01 -1.66171625e-01 -3.43086571e-01
-1.61526608e+00 4.83427405e-01 2.65831679e-01 1.38280705e-01
-5.21413207e-01 9.30620968e-01 6.42466486e-01 2.97513455e-01
3.65707994e-01 1.47730544e-01 -3.19254845e-01 -3.38204205e-01
-1.22154713e+00 5.43159962e-01 6.20742559e-01 8.15647781e-01
-2.09651142e-01 4.65805560e-01 -6.26348138e-01 6.45259261e-01
3.41240734e-01 5.50528467e-01 1.02109265e+00 -1.21806586e+00
4.77527678e-01 8.12846236e-03 1.42411917e-01 -1.04678857e+00
-8.51450026e-01 -2.52467960e-01 -8.52768004e-01 1.24455392e-01
3.51014584e-01 -3.49128664e-01 -1.14049470e+00 1.82053888e+00
2.32355759e-01 2.30112091e-01 6.55964166e-02 1.26197481e+00
1.01768565e+00 6.76823914e-01 1.28069103e-01 8.18900857e-03
1.12796438e+00 -1.15867853e+00 -8.37425470e-01 -4.84836936e-01
6.95870697e-01 -4.00034517e-01 3.75820279e-01 1.22141302e-01
-1.34323740e+00 -9.98861015e-01 -8.81277263e-01 -4.13840204e-01
1.14055902e-01 3.51081043e-01 3.06871325e-01 1.90498918e-01
-9.94744837e-01 1.00112391e+00 -1.43648326e+00 -5.94441235e-01
-2.08804291e-02 4.67657655e-01 -5.28086066e-01 3.64750296e-01
-1.09436309e+00 9.04313922e-01 4.26911600e-02 5.48666894e-01
-1.11646724e+00 -1.09745339e-01 -1.37491262e+00 -4.72754091e-01
4.44721654e-02 -1.28788388e+00 1.51384759e+00 -8.36307824e-01
-1.69670546e+00 1.00129509e+00 -1.58064231e-01 -8.90750289e-01
7.62980282e-01 -7.34049916e-01 -3.46151203e-01 1.93510339e-01
-1.63905025e-01 1.20580113e+00 1.24646485e+00 -4.02397990e-01
-1.78019434e-01 -1.75618201e-01 -3.58193308e-01 5.28942227e-01
2.26679400e-01 -1.29640207e-01 -5.73986650e-01 -1.06378400e+00
3.58612984e-02 -1.31936657e+00 -2.97085136e-01 1.35840207e-01
-2.78220505e-01 -1.58010256e-02 6.92586541e-01 -1.09650648e+00
1.20748031e+00 -1.83727825e+00 7.49781013e-01 -3.54907781e-01
1.00749627e-01 1.18404843e-01 -1.66140839e-01 2.68881202e-01
-2.30835766e-01 -3.86224866e-01 1.84788793e-01 -6.02226138e-01
-7.61451479e-03 3.96486670e-01 -2.02299640e-01 7.58980632e-01
1.28643051e-01 1.34818757e+00 -6.41006887e-01 -4.52631295e-01
4.12967861e-01 6.64136589e-01 -6.39629006e-01 3.41798455e-01
-3.62057775e-01 6.16674781e-01 -1.49725350e-02 7.25476444e-01
-2.98340350e-01 -4.37407643e-01 4.45296988e-02 -1.93341866e-01
2.09033102e-01 2.39362806e-01 -1.03164434e+00 2.16626620e+00
-2.13677466e-01 8.55646610e-01 -2.84454942e-01 -6.09742343e-01
5.93447506e-01 7.81546831e-01 9.46582913e-01 -5.63321948e-01
1.30587742e-01 -1.27601206e-01 -1.86713368e-01 -8.73808563e-01
5.51238537e-01 -9.66600701e-02 -2.31277779e-01 2.45479405e-01
4.17615920e-01 2.91388303e-01 9.31852236e-02 -2.89332241e-01
1.14638913e+00 9.59497035e-01 1.54652700e-01 3.65102947e-01
2.68583447e-01 -1.50717676e-01 6.02567613e-01 5.53134203e-01
-3.33086222e-01 1.15747309e+00 2.31583655e-01 -9.05338168e-01
-1.08371711e+00 -9.63986456e-01 5.06347597e-01 1.37842798e+00
-1.66991144e-01 -4.95062798e-01 -4.58852351e-01 -2.99543023e-01
-9.19325128e-02 -9.79007259e-02 -7.80476809e-01 -2.62573212e-01
-1.21049881e+00 -4.12126899e-01 8.27114820e-01 1.06615365e+00
1.76565692e-01 -1.44728696e+00 -1.18829858e+00 5.15600801e-01
-3.95087689e-01 -1.19162607e+00 -8.28328550e-01 3.20466384e-02
-1.03286827e+00 -8.71141613e-01 -1.33918023e+00 -9.11649048e-01
1.76557809e-01 -4.51065481e-01 1.15973830e+00 -3.78671139e-01
-5.89461923e-01 6.94610536e-01 -1.15865849e-01 2.20762298e-01
-3.12634379e-01 5.44410152e-03 4.57607687e-01 -2.52489954e-01
-4.99419235e-02 -4.91627544e-01 -8.28345954e-01 3.41795713e-01
-4.23762858e-01 2.65444010e-01 4.61631536e-01 8.94387126e-01
5.64934492e-01 -9.19353008e-01 1.35193288e-01 -1.75635606e-01
2.84319073e-01 -4.12525266e-01 1.19842067e-02 -7.10429205e-03
3.55570354e-02 3.27273905e-01 8.98829997e-02 -7.08035946e-01
-7.24726319e-01 6.15065277e-01 -3.89264911e-01 -1.05198157e+00
-1.03416264e-01 1.67252973e-01 4.05428916e-01 2.87335753e-01
6.83741271e-01 2.62465209e-01 1.79853469e-01 -5.98119199e-01
3.68848056e-01 -3.56968604e-02 1.17084527e+00 -2.68600762e-01
2.76642829e-01 3.76531243e-01 1.46923155e-01 -7.72660792e-01
-6.02574408e-01 -4.07216579e-01 -1.08659005e+00 -6.01106942e-01
1.35965753e+00 -1.29456615e+00 -8.25797141e-01 5.34260392e-01
-1.24293768e+00 -6.91727281e-01 -5.47890782e-01 7.42939234e-01
-1.19735467e+00 1.35567203e-01 -1.11785996e+00 -6.82081580e-01
-3.90744537e-01 -7.54613996e-01 1.55937588e+00 -1.37577891e-01
-8.28461289e-01 -1.11114657e+00 4.48832870e-01 9.75021720e-02
-1.21440299e-01 8.44121516e-01 1.48982629e-01 -1.00039110e-01
-2.25001052e-01 -2.72057980e-01 5.46720803e-01 1.13124050e-01
-1.38082445e-01 -1.90592825e-01 -6.81224465e-01 -3.55577409e-01
-3.18677098e-01 -5.41548550e-01 1.01630688e+00 1.00992823e+00
7.98900604e-01 -4.62333590e-01 -2.07926691e-01 7.69381464e-01
7.35660791e-01 -6.88033253e-02 6.84317052e-01 8.70503932e-02
1.04882693e+00 5.82927823e-01 3.90316993e-01 4.46250200e-01
3.65847856e-01 1.01070094e+00 2.07932398e-01 -6.41405359e-02
-5.81373990e-01 -5.98844409e-01 1.12147546e+00 7.42113650e-01
-6.06890917e-01 -1.54137135e-01 -8.60792577e-01 5.54686248e-01
-2.01562953e+00 -1.43710053e+00 1.78252131e-01 1.92676175e+00
5.72878659e-01 1.30001590e-01 7.46116936e-01 -7.01477239e-03
5.17307341e-01 5.16236007e-01 -6.53606117e-01 -4.70645040e-01
9.29863006e-03 -5.03792986e-02 4.78871673e-01 3.99528801e-01
-1.37887180e+00 9.05904949e-01 7.22068739e+00 2.38407686e-01
-1.15030718e+00 -1.54783055e-02 2.75983512e-01 -6.00015819e-01
4.03780490e-01 -6.12796903e-01 -8.42870176e-01 2.28355587e-01
1.13103855e+00 3.76327753e-01 -5.01003563e-02 7.07744479e-01
3.29969972e-01 3.09667081e-01 -1.40170944e+00 1.34578252e+00
4.07302588e-01 -1.33855903e+00 -6.32493049e-02 4.47570905e-02
6.66227579e-01 2.28584662e-01 6.18998446e-02 3.29055786e-01
1.18377700e-01 -1.27372861e+00 9.60232556e-01 1.08641171e+00
9.38178182e-01 -2.46137366e-01 3.36458057e-01 4.67275620e-01
-1.39874148e+00 -2.09787399e-01 2.72568148e-02 -2.23976523e-01
5.81022799e-01 -1.63670912e-01 -6.36992276e-01 1.30007491e-01
5.84371388e-01 1.39067471e+00 -5.05195379e-01 7.73982346e-01
-9.24672186e-02 4.13042188e-01 -2.36476809e-01 4.25487250e-01
2.57941455e-01 3.04100305e-01 7.87015021e-01 1.43435788e+00
2.63985336e-01 -1.39409497e-01 2.96418041e-01 3.21330458e-01
9.52048972e-02 -3.19048226e-01 -7.82247007e-01 -6.89275712e-02
-3.72020662e-01 6.64658308e-01 -3.75309795e-01 -3.14723670e-01
-1.87085271e-01 1.41802669e+00 -8.83160457e-02 3.13304454e-01
-9.12979066e-01 1.06710143e-01 7.05520749e-01 3.66852194e-01
4.36351746e-01 -5.64375460e-01 1.62214622e-01 -1.32842875e+00
-8.07452053e-02 -5.72191298e-01 6.39184892e-01 -8.32476199e-01
-8.91954601e-01 3.53219360e-01 -2.22145207e-02 -1.59425712e+00
-1.42787957e+00 -6.05249882e-01 -3.18434775e-01 3.60047609e-01
-6.57593846e-01 -1.29932475e+00 -4.91307229e-02 7.03142941e-01
1.05427122e+00 -2.95473766e-02 6.86514318e-01 2.48104870e-01
-5.25110424e-01 3.10159504e-01 -3.28924924e-01 5.56019306e-01
6.44849062e-01 -1.05847633e+00 9.06999052e-01 6.92896068e-01
3.70056748e-01 2.76066691e-01 8.05497408e-01 -8.46012712e-01
-1.30577230e+00 -1.18228519e+00 9.74088252e-01 -8.26134086e-01
1.78590953e-01 -3.39466512e-01 -4.79122609e-01 1.09335351e+00
-1.28915220e-01 4.00049776e-01 3.34114999e-01 -1.67862579e-01
-1.55378236e-02 4.03845131e-01 -5.52790999e-01 5.83446324e-01
1.51942122e+00 -6.01557553e-01 -8.19603086e-01 3.37557882e-01
4.79863495e-01 -9.79601085e-01 -9.24327254e-01 5.71320593e-01
1.20334530e+00 -7.18254685e-01 1.27033937e+00 -8.51397455e-01
7.17464089e-01 -7.86029249e-02 9.04658530e-03 -7.99301207e-01
-4.68247414e-01 -8.25648665e-01 -9.76491749e-01 3.95053238e-01
3.19014639e-02 3.18720549e-01 1.06505799e+00 3.14567149e-01
-1.62281245e-01 -6.60054624e-01 -9.11947370e-01 -8.19728553e-01
-3.67796794e-02 -3.63969773e-01 -1.15605026e-01 4.28634346e-01
-2.15111956e-01 3.44284356e-01 -1.27355993e+00 -2.07769185e-01
3.56777430e-01 4.51635309e-02 8.97926569e-01 -7.90560663e-01
-5.41875184e-01 -2.80189216e-01 -9.01908875e-01 -1.55651689e+00
1.76190957e-01 -6.08375192e-01 2.15842336e-01 -1.62128282e+00
-1.59488425e-01 6.09143794e-01 1.57810360e-01 5.66130698e-01
2.54320294e-01 4.66273397e-01 3.72424334e-01 3.42334002e-01
-7.70975530e-01 5.17537177e-01 1.33566940e+00 -1.59178033e-01
-4.27092522e-01 1.83353201e-01 2.84436524e-01 9.93841410e-01
2.92821825e-01 -3.28624606e-01 -2.12360725e-01 -5.04855871e-01
1.92120388e-01 7.95385540e-01 8.32314670e-01 -1.57775664e+00
2.09305391e-01 1.21079385e-01 9.87194061e-01 -7.83760965e-01
7.36856520e-01 -3.37801099e-01 4.69605893e-01 9.10881639e-01
-5.80963254e-01 2.64675707e-01 -2.56971139e-02 6.71333611e-01
-1.86842740e-01 3.81791145e-01 4.67138022e-01 -2.74253577e-01
-1.08644390e+00 4.91617084e-01 -7.46081471e-01 6.29262552e-02
6.38945997e-01 -6.52751565e-01 1.88588038e-01 -6.66135907e-01
-1.69100237e+00 1.06280394e-01 2.06963196e-01 9.13023412e-01
6.49752557e-01 -1.60334778e+00 -6.32017374e-01 1.29912376e-01
-2.55192779e-02 -2.21003801e-01 4.21680868e-01 8.78515363e-01
-5.88523746e-01 5.81022918e-01 -4.64910090e-01 -9.28525507e-01
-1.33068335e+00 3.08204591e-01 7.79357195e-01 -2.75600821e-01
-1.19723010e+00 8.34058344e-01 1.98001862e-02 -1.49021089e-01
5.01279294e-01 -5.02269030e-01 -3.50868523e-01 8.01007673e-02
2.96548635e-01 4.47226226e-01 -2.87135661e-01 -1.18685520e+00
-3.47279996e-01 8.01263928e-01 4.40776378e-01 -3.06812167e-01
1.14991689e+00 -2.64212608e-01 6.11361742e-01 8.32251668e-01
1.15943515e+00 -4.47030336e-01 -1.79898965e+00 -1.96020063e-02
-6.19364418e-02 2.18325499e-02 -4.09705967e-01 -4.89522517e-01
-9.42120731e-01 1.00313544e+00 6.95005238e-01 -5.24955511e-01
7.78268933e-01 5.14782742e-02 1.08559108e+00 3.74085724e-01
-3.63094383e-03 -1.27565765e+00 5.75468957e-01 7.69512057e-01
1.15848482e+00 -9.99624908e-01 -3.24408822e-02 2.78519601e-01
-9.12641585e-01 1.22576106e+00 6.90382302e-01 -3.72047812e-01
3.93509924e-01 7.64706731e-02 -1.13519117e-01 -2.54627705e-01
-1.05906165e+00 -3.19815040e-01 7.90059865e-01 6.06770277e-01
6.92944109e-01 -2.00180054e-01 1.15899906e-01 4.02453989e-01
-2.06010252e-01 2.17122167e-01 2.34782115e-01 1.02029169e+00
-1.28038824e-01 -7.23383367e-01 -3.91647965e-01 1.03961907e-01
-4.95835871e-01 4.13912684e-01 -2.21591517e-01 6.55800998e-01
2.65318960e-01 3.48629087e-01 2.36075804e-01 -5.40196955e-01
2.62597978e-01 2.95771986e-01 8.46390665e-01 -5.73381722e-01
-5.44155717e-01 2.47321516e-01 2.66073734e-01 -1.07894337e+00
-7.01664209e-01 -8.12904358e-01 -1.24036717e+00 3.29692289e-02
1.85919344e-01 -4.55452055e-01 2.88850546e-01 9.45995569e-01
2.38244489e-01 7.38267362e-01 3.05450819e-02 -1.51141155e+00
-3.70529175e-01 -9.94626582e-01 -3.90746891e-01 5.42652309e-01
8.79171789e-01 -7.75056660e-01 1.70465991e-01 7.77277410e-01] | [7.182224750518799, -0.28494909405708313] |
3de03ea8-7929-479c-b187-72d201a40f4a | on-the-difficulty-of-intersection-checking | 2305.09901 | null | https://arxiv.org/abs/2305.09901v2 | https://arxiv.org/pdf/2305.09901v2.pdf | On the Difficulty of Intersection Checking with Polynomial Zonotopes | Polynomial zonotopes, a non-convex set representation, have a wide range of applications from real-time motion planning and control in robotics, to reachability analysis of nonlinear systems and safety shielding in reinforcement learning. Despite this widespread use, a frequently overlooked difficulty associated with polynomial zonotopes is intersection checking. Determining whether the reachable set, represented as a polynomial zonotope, intersects an unsafe set is not straightforward. In fact, we show that this fundamental operation is NP-hard, even for a simple class of polynomial zonotopes. The standard method for intersection checking with polynomial zonotopes is a two-part algorithm that overapproximates a polynomial zonotope with a regular zonotope and then, if the overapproximation error is deemed too large, splits the set and recursively tries again. Beyond the possible need for a large number of splits, we identify two sources of concern related to this algorithm: (1) overapproximating a polynomial zonotope with a zonotope has unbounded error, and (2) after splitting a polynomial zonotope, the overapproximation error can actually increase. Taken together, this implies there may be a possibility that the algorithm does not always terminate.We perform a rigorous analysis of the method and detail necessary conditions for the union of overapproximations to provably converge to the original polynomial zonotope. | ['Yifan Sun', 'Stanley Bak', 'Ertai Luo', 'Yushen Huang'] | 2023-05-17 | null | null | null | null | ['motion-planning'] | ['robots'] | [ 2.02569097e-01 5.04416108e-01 -1.77631721e-01 4.29729342e-01
-6.23775959e-01 -1.01429164e+00 2.46094659e-01 3.01678449e-01
-1.45078838e-01 1.12256205e+00 -4.65399951e-01 -8.46937358e-01
-5.11262953e-01 -9.14155781e-01 -9.55686152e-01 -7.90813506e-01
-6.20804429e-01 8.07691574e-01 6.03286445e-01 -3.38822126e-01
2.15527311e-01 7.78712213e-01 -1.39216185e+00 -4.07396287e-01
1.16017175e+00 6.19610965e-01 -1.67471141e-01 7.77298927e-01
3.36188912e-01 2.54793108e-01 -6.00016594e-01 2.46100239e-02
6.06013358e-01 -1.09903887e-01 -1.11287022e+00 5.46278134e-02
1.58062130e-01 -1.78595230e-01 2.05573529e-01 1.43642533e+00
-2.12480366e-01 3.40786964e-01 5.74371040e-01 -1.90687680e+00
3.75621468e-01 3.08543921e-01 -4.38558459e-01 -1.20863296e-01
2.23617852e-01 1.17075518e-01 8.33989620e-01 -1.91717759e-01
4.62256879e-01 1.07555437e+00 6.80240750e-01 2.75384665e-01
-1.35830522e+00 -5.42260706e-01 -4.21522558e-02 -3.37204814e-01
-1.58870685e+00 1.64333269e-01 1.19621947e-01 -4.59941953e-01
9.02387917e-01 7.42591679e-01 7.56314278e-01 3.60908825e-03
6.62664354e-01 2.32599810e-01 1.06689727e+00 -1.62413463e-01
3.44189346e-01 -1.21590726e-01 -1.72789559e-01 9.76396263e-01
6.01827323e-01 5.61594069e-01 3.38489532e-01 -4.10560101e-01
9.11706865e-01 -1.17895998e-01 -4.63906020e-01 -7.33253777e-01
-1.24286270e+00 8.51903200e-01 4.00709957e-01 1.53508335e-01
-5.45903035e-02 3.84168893e-01 2.47013822e-01 3.84458572e-01
-4.17170450e-02 9.82652783e-01 -2.79346466e-01 -2.77676463e-01
-7.51069963e-01 7.86068380e-01 1.01851559e+00 9.71597016e-01
8.27164650e-01 3.35751250e-02 3.89534086e-01 -1.03890680e-01
-1.42277449e-01 2.84721613e-01 -2.52028555e-01 -8.54450405e-01
4.47642058e-01 5.09593308e-01 4.53892618e-01 -9.90402400e-01
-5.91246068e-01 -1.60823807e-01 -6.76222742e-01 1.02574956e+00
8.60418260e-01 -2.48528451e-01 -4.57729697e-01 1.65992165e+00
5.10067821e-01 8.06032494e-02 1.03723124e-01 1.07334900e+00
-4.45020586e-01 8.13212872e-01 -2.75268376e-01 -4.60750669e-01
1.00588274e+00 -7.20050454e-01 -2.47808918e-01 5.41092120e-02
8.06320250e-01 -4.30875987e-01 7.87321031e-01 6.73871160e-01
-1.08511424e+00 7.03865141e-02 -1.26571143e+00 3.49211872e-01
-9.43395793e-02 -5.31905651e-01 5.76714098e-01 3.61134917e-01
-1.00794220e+00 7.46638536e-01 -8.35838318e-01 -1.12290140e-02
-1.63237348e-01 6.58644259e-01 -4.48303103e-01 1.46230340e-01
-1.12590110e+00 1.11593783e+00 6.23816133e-01 -7.05695823e-02
-8.83794546e-01 -4.49063867e-01 -8.43863785e-01 3.09804380e-01
1.22558856e+00 -2.04824239e-01 1.13012409e+00 -8.03880453e-01
-1.23941278e+00 1.55986249e-01 8.10892880e-03 -6.97007835e-01
7.24697292e-01 1.63987190e-01 -7.25520402e-02 9.13164839e-02
1.25439033e-01 1.10989325e-01 6.36651576e-01 -1.00852370e+00
-6.60974503e-01 -2.22764641e-01 7.00857282e-01 4.31044877e-01
4.22296941e-01 -2.83657998e-01 6.64469600e-02 -2.54105777e-01
2.04746529e-01 -1.29982674e+00 -7.69646883e-01 5.48921376e-02
-6.34391069e-01 -3.39443028e-01 4.71351445e-01 7.19212964e-02
1.17231357e+00 -1.95635676e+00 2.20922470e-01 6.58418059e-01
2.38325059e-01 5.34521602e-02 2.28014156e-01 6.52472734e-01
-1.72699109e-01 3.22827607e-01 -4.30547476e-01 4.49576765e-01
3.38116996e-02 2.35966474e-01 -4.75476831e-01 8.96806061e-01
1.92403466e-01 4.82079238e-01 -1.09226286e+00 -1.43862784e-01
2.91297227e-01 -2.31780291e-01 -8.63682151e-01 -3.38134170e-01
-4.11712915e-01 1.06540762e-01 -4.45256710e-01 2.02269942e-01
7.55706370e-01 1.75231233e-01 -6.45587593e-02 2.43869036e-01
-5.18590450e-01 2.96880245e-01 -1.53303289e+00 9.11641538e-01
-1.59396484e-01 4.85143453e-01 3.13507169e-01 -9.20136511e-01
4.08988565e-01 1.87240824e-01 6.28183067e-01 1.20853111e-02
2.16831654e-01 3.83758038e-01 -7.28140399e-02 -1.21927790e-01
6.57039821e-01 -6.94728255e-01 -3.81085396e-01 2.92270601e-01
-5.46289623e-01 -7.31235206e-01 1.38143867e-01 -9.48707536e-02
1.09885681e+00 -6.94793463e-02 4.70496178e-01 -7.39064217e-01
6.84390068e-01 7.05787003e-01 6.18571162e-01 4.08196986e-01
-2.88178682e-01 2.42197484e-01 9.99885678e-01 -1.56154290e-01
-8.65646720e-01 -9.26759541e-01 -8.77361074e-02 4.12560403e-01
6.54692650e-01 -4.05218422e-01 -8.24217021e-01 -4.69365388e-01
1.26715273e-01 7.25903213e-01 -5.52257180e-01 -2.73783833e-01
-4.76821303e-01 -4.01503853e-02 5.51088274e-01 3.00013274e-01
2.37177476e-01 -4.25571114e-01 -1.03553641e+00 1.83639750e-01
-2.45576482e-02 -7.52991080e-01 -6.52550995e-01 3.59642357e-01
-8.09851170e-01 -1.56455576e+00 -1.52492151e-01 -5.68859160e-01
8.91702712e-01 3.45443666e-01 7.53345549e-01 2.47519463e-01
1.45393431e-01 4.90078032e-02 2.18090132e-01 -3.49737138e-01
-4.88377601e-01 -2.50555336e-01 2.18372583e-01 -5.63763559e-01
-2.58750111e-01 1.10077960e-02 -1.40238374e-01 6.00498557e-01
-8.55991125e-01 -8.73632506e-02 1.19892411e-01 5.99594951e-01
8.74386132e-01 1.03070939e+00 2.60793477e-01 -6.17030203e-01
5.70088923e-01 -4.27155465e-01 -1.51656806e+00 2.51742173e-02
-3.17836583e-01 2.13369891e-01 7.75983036e-01 -5.41630268e-01
-1.54375657e-01 2.57506430e-01 2.49445438e-01 -6.03533983e-01
1.31405964e-01 8.49362493e-01 -5.04896194e-02 -2.34221488e-01
6.03620112e-01 -1.03487223e-01 2.95319438e-01 2.40155503e-01
1.64498791e-01 6.57888055e-02 5.79383850e-01 -5.00978529e-01
1.11536014e+00 4.84502316e-01 5.80364823e-01 -6.45683944e-01
-2.87132949e-01 -3.76371682e-01 -1.45960972e-01 -3.74528944e-01
6.66011691e-01 -4.08574253e-01 -1.23483086e+00 -8.91600773e-02
-9.95447874e-01 -4.53003883e-01 -3.72005910e-01 2.87303805e-01
-8.96062791e-01 3.80005598e-01 -2.31268518e-02 -1.14270616e+00
2.81739533e-01 -1.40869820e+00 7.19227135e-01 -8.29112623e-03
-3.04589540e-01 -7.32180715e-01 1.82475187e-02 -7.61963353e-02
1.48445681e-01 7.60692418e-01 9.21067953e-01 -4.90467250e-01
-6.55392289e-01 -4.42531884e-01 2.12237582e-01 -1.79460675e-01
4.69601201e-03 1.21399172e-01 -1.23204522e-01 -5.21425843e-01
1.10353120e-01 -7.37488866e-02 5.32469675e-02 2.21018478e-01
8.43114078e-01 -7.70808399e-01 -4.39445853e-01 2.57320516e-02
1.57196426e+00 2.97595292e-01 3.64265084e-01 4.53708261e-01
2.32890189e-01 6.70486927e-01 9.38551903e-01 1.31541669e-01
1.96278676e-01 5.45977056e-01 9.86098766e-01 2.02293277e-01
7.02861071e-01 8.14370159e-03 3.17734808e-01 -1.70449796e-03
1.53045515e-02 6.45613251e-03 -9.77961063e-01 5.89733183e-01
-1.75278139e+00 -9.22535896e-01 -2.27581277e-01 2.78634119e+00
7.23515332e-01 3.71967047e-01 4.99223292e-01 4.51232076e-01
8.39405954e-01 -3.51974010e-01 -5.25501013e-01 -9.01490688e-01
2.79587448e-01 -2.31823906e-01 1.06544816e+00 9.57998872e-01
-7.90200531e-01 8.26677859e-01 6.32746744e+00 7.29118407e-01
-1.14912021e+00 -2.49293506e-01 1.19261533e-01 1.68349564e-01
-3.05277109e-01 3.36404353e-01 -3.91834885e-01 1.49829522e-01
5.54803014e-01 -7.16520131e-01 8.67404103e-01 9.96881962e-01
1.01042502e-01 -5.28869510e-01 -1.09675336e+00 2.97766089e-01
-3.96057367e-01 -1.10348034e+00 -4.09347713e-01 3.95627379e-01
7.89623499e-01 -2.83954024e-01 -7.52976537e-02 2.33052790e-01
5.01546562e-01 -1.15369713e+00 8.02572548e-01 -1.11967608e-01
6.49530113e-01 -1.34633839e+00 4.46295887e-01 7.55566478e-01
-1.19727528e+00 -4.89328653e-02 -2.30640754e-01 -3.01914960e-01
2.16225222e-01 3.29772413e-01 -8.95840466e-01 7.07265139e-01
1.38465077e-01 -1.16774943e-02 2.74641454e-01 1.48435378e+00
-1.82046518e-01 1.83543429e-01 -8.05850863e-01 -1.29419774e-01
5.46969831e-01 -5.00757933e-01 1.00536215e+00 6.41695440e-01
1.67293206e-01 2.76251614e-01 2.82839060e-01 9.28636551e-01
5.30974925e-01 -3.41865867e-01 -8.91583443e-01 -2.11010389e-02
4.86397803e-01 7.45725691e-01 -8.23315263e-01 -1.94641367e-01
-1.64804190e-01 4.70322907e-01 2.08753590e-02 4.14787650e-01
-1.06482017e+00 -5.17952263e-01 7.85346508e-01 1.67778835e-01
5.47873676e-02 -5.61693549e-01 -3.90681416e-01 -8.15126181e-01
-1.36278510e-01 -7.79984355e-01 3.08252156e-01 -5.90858817e-01
-5.29459774e-01 6.24651372e-01 4.15602118e-01 -1.40311706e+00
-5.69272697e-01 -2.76535958e-01 -4.28078830e-01 8.18453968e-01
-1.11180270e+00 -3.32641542e-01 7.66579434e-02 5.38052142e-01
3.49238515e-01 5.51666141e-01 6.33781433e-01 -1.15427956e-01
-2.89614677e-01 1.82384148e-01 -1.46090239e-01 -4.82547343e-01
4.03967127e-02 -1.30547607e+00 -7.16453344e-02 1.12633908e+00
-3.59317571e-01 5.68518281e-01 1.26547194e+00 -7.30326772e-01
-1.65252078e+00 -1.30681765e+00 6.95877433e-01 -2.15352789e-01
9.32844937e-01 -3.96412387e-02 -8.33004713e-01 8.36044788e-01
-3.41512114e-01 -2.25608855e-01 -2.44961947e-01 -3.26624006e-01
2.42088176e-03 2.43736699e-01 -1.33685124e+00 1.09329617e+00
6.99860811e-01 -1.46628708e-01 -3.99121255e-01 3.65974069e-01
7.09289193e-01 -7.36361861e-01 -6.94598973e-01 6.98408961e-01
2.79461235e-01 -7.07335413e-01 6.53243661e-01 -4.33383256e-01
6.98256344e-02 -9.26635385e-01 1.57322079e-01 -1.38968623e+00
-1.32844046e-01 -1.06738138e+00 4.55125161e-02 4.47403997e-01
2.65293986e-01 -9.28141475e-01 7.86051631e-01 6.54068291e-01
-3.00843567e-01 -1.14477074e+00 -1.33317447e+00 -1.30410445e+00
4.86833751e-01 -4.08088654e-01 6.58305764e-01 8.26413393e-01
6.74956381e-01 6.70429021e-02 -3.86687145e-02 6.02241099e-01
2.86367446e-01 1.34755939e-01 6.41778708e-01 -1.12911415e+00
-1.82348013e-01 -7.96785951e-01 -3.55559766e-01 -9.86039340e-01
3.12918462e-02 -6.47674680e-01 5.15085220e-01 -1.41752005e+00
-5.33037126e-01 -9.09904420e-01 1.92320034e-01 4.22493935e-01
9.07405242e-02 -1.90129891e-01 1.62873924e-01 9.10640210e-02
-3.30653191e-01 1.97938919e-01 1.42935050e+00 -1.98616851e-02
-3.70502323e-01 3.97389561e-01 -3.88863951e-01 8.29912841e-01
7.76083112e-01 -3.94807994e-01 -6.60301089e-01 3.37362587e-02
4.92142975e-01 6.50519669e-01 2.96317309e-01 -1.00197518e+00
9.69650224e-02 -7.70721138e-01 -7.76246309e-01 -6.12574637e-01
2.45627731e-01 -1.26403427e+00 5.26127398e-01 1.00198114e+00
-6.51927143e-02 3.55763465e-01 4.56627965e-01 6.55682921e-01
8.71079937e-02 -4.71406788e-01 7.69287229e-01 2.73374557e-01
-2.08156794e-01 2.00875938e-01 -9.53832388e-01 -1.58496484e-01
1.44887066e+00 -6.29456758e-01 -1.86911404e-01 -3.17536116e-01
-4.58887249e-01 6.28619432e-01 8.92623842e-01 4.95474786e-02
3.97058010e-01 -1.03933561e+00 -9.15008485e-02 1.04453780e-01
-1.95425853e-01 4.47886407e-01 -2.25475430e-01 9.80716407e-01
-7.55922556e-01 5.20235717e-01 6.78261369e-02 -4.21720356e-01
-1.20833504e+00 8.57684612e-01 6.34021580e-01 -2.05868036e-01
-5.04875183e-01 5.41315198e-01 1.94663659e-01 -1.70723259e-01
1.81274667e-01 -9.99999225e-01 -1.61429998e-02 -4.48288828e-01
1.90995559e-01 5.54790139e-01 -7.51288682e-02 -5.75042963e-01
-4.40823287e-01 4.79273975e-01 3.53932947e-01 -4.06385601e-01
9.66995955e-01 -3.34165292e-03 -1.88243344e-01 2.42017955e-01
7.26168513e-01 1.28028572e-01 -1.24506295e+00 4.37934220e-01
-2.84321550e-02 -3.43035519e-01 -3.76007229e-01 -4.60032910e-01
-6.29629612e-01 4.72179621e-01 -6.38518259e-02 7.52815247e-01
1.06361389e+00 -2.04494059e-01 5.70432901e-01 5.08932590e-01
8.79104495e-01 -9.31444526e-01 -3.83462101e-01 8.67097437e-01
9.08599973e-01 -6.22601390e-01 7.12808296e-02 -9.23436761e-01
-6.32148683e-01 1.18298829e+00 6.36246741e-01 -5.46921194e-01
3.54667842e-01 3.61872852e-01 -6.00868225e-01 -6.29530773e-02
-6.58097863e-01 -2.92506199e-02 4.34686095e-02 2.65495926e-01
-3.02019507e-01 2.19975188e-01 -4.65492725e-01 2.53936082e-01
-6.06033683e-01 -5.22482172e-02 1.02470267e+00 9.11059022e-01
-7.79196739e-01 -6.41043067e-01 -6.83773816e-01 2.10599780e-01
-3.14962894e-01 1.90204039e-01 -2.49413997e-02 1.08425558e+00
7.52196088e-02 8.90098393e-01 2.65287589e-02 -2.45438144e-01
2.97697484e-01 -4.33282673e-01 5.73397934e-01 -6.52280390e-01
-3.86775970e-01 5.90368994e-02 2.72493362e-01 -6.67540491e-01
-9.06083137e-02 -5.70849121e-01 -1.91890812e+00 -5.67695975e-01
-8.42113674e-01 4.85685289e-01 4.45278198e-01 1.12425077e+00
-3.40696215e-03 2.47360721e-01 4.81299728e-01 -5.96576512e-01
-7.20291495e-01 -3.12098235e-01 -4.38102543e-01 -3.13723117e-01
6.09683335e-01 -9.01556849e-01 -5.54780126e-01 -5.47022760e-01] | [4.872641086578369, 1.9602086544036865] |
81b47360-0a42-408d-b895-385543d7f6a8 | distilling-large-vision-language-model-with | 2307.03135 | null | https://arxiv.org/abs/2307.03135v1 | https://arxiv.org/pdf/2307.03135v1.pdf | Distilling Large Vision-Language Model with Out-of-Distribution Generalizability | Large vision-language models have achieved outstanding performance, but their size and computational requirements make their deployment on resource-constrained devices and time-sensitive tasks impractical. Model distillation, the process of creating smaller, faster models that maintain the performance of larger models, is a promising direction towards the solution. This paper investigates the distillation of visual representations in large teacher vision-language models into lightweight student models using a small- or mid-scale dataset. Notably, this study focuses on open-vocabulary out-of-distribution (OOD) generalization, a challenging problem that has been overlooked in previous model distillation literature. We propose two principles from vision and language modality perspectives to enhance student's OOD generalization: (1) by better imitating teacher's visual representation space, and carefully promoting better coherence in vision-language alignment with the teacher; (2) by enriching the teacher's language representations with informative and finegrained semantic attributes to effectively distinguish between different labels. We propose several metrics and conduct extensive experiments to investigate their techniques. The results demonstrate significant improvements in zero-shot and few-shot student performance on open-vocabulary out-of-distribution classification, highlighting the effectiveness of our proposed approaches. Our code will be released at https://github.com/xuanlinli17/large_vlm_distillation_ood | ['Hao Su', 'Zhuowen Tu', 'Zhan Ling', 'Minghua Liu', 'Yunhao Fang', 'Xuanlin Li'] | 2023-07-06 | null | null | null | null | ['zero-shot-transfer-image-classification', 'few-shot-image-classification', 'zero-shot-learning', 'prompt-engineering', 'natural-language-understanding'] | ['computer-vision', 'computer-vision', 'methodology', 'natural-language-processing', 'natural-language-processing'] | [ 8.39050412e-02 7.36791268e-02 -4.04534131e-01 -3.58184159e-01
-6.87245905e-01 -6.31017148e-01 5.87719142e-01 2.84113914e-01
-4.23029572e-01 2.98578650e-01 2.08430111e-01 -4.81788188e-01
5.56687228e-02 -5.43107748e-01 -5.21410286e-01 -5.94968379e-01
5.27819455e-01 4.99627590e-01 2.62529939e-01 -9.75560322e-02
-6.54480010e-02 4.75929350e-01 -1.90648937e+00 1.06065229e-01
1.07581127e+00 8.24963033e-01 4.99959975e-01 7.39872754e-01
-4.74147171e-01 8.16287756e-01 -5.82343519e-01 -3.58964860e-01
1.01668663e-01 -3.01279664e-01 -6.72304749e-01 2.12480947e-01
1.22107804e+00 -3.02536875e-01 -3.34880829e-01 1.15150845e+00
7.97352195e-01 4.32201236e-01 8.33095670e-01 -1.34434867e+00
-1.34877825e+00 5.41621745e-01 -8.10526431e-01 3.96387219e-01
8.39152634e-02 2.44827658e-01 8.62488449e-01 -1.00763154e+00
3.91315877e-01 1.38391411e+00 4.38800424e-01 9.38202679e-01
-1.47891116e+00 -9.50121224e-01 3.30684960e-01 3.15986335e-01
-1.48161495e+00 -3.30423653e-01 6.71772003e-01 -7.57696390e-01
7.40233123e-01 2.12652653e-01 6.89895034e-01 1.24017239e+00
-4.57373887e-01 1.12062800e+00 1.26094663e+00 -7.26568401e-01
-7.25350678e-02 5.46065450e-01 5.45397639e-01 7.62973249e-01
3.56367350e-01 -4.90811020e-02 -5.26520908e-01 1.60772964e-01
4.21436280e-01 1.24427304e-01 -3.06408167e-01 -7.05863535e-01
-7.88349628e-01 9.92297649e-01 4.61727440e-01 1.48484111e-01
1.23875707e-01 -3.31801316e-03 3.11872095e-01 3.32439125e-01
6.71301901e-01 3.51595253e-01 -1.68077320e-01 -1.08688086e-01
-1.00971377e+00 3.07791699e-02 3.41821611e-01 1.26229072e+00
6.69225395e-01 2.37376362e-01 -4.32614148e-01 1.15856516e+00
3.78325105e-01 5.97142577e-01 8.71290326e-01 -6.00368023e-01
2.09201992e-01 5.99622667e-01 -4.48513478e-01 -4.91227329e-01
-6.28515780e-02 -4.75067794e-01 -5.97718835e-01 1.65343881e-01
2.96317250e-01 1.45494565e-01 -1.21351230e+00 1.71330869e+00
5.97585618e-01 5.09577751e-01 2.06342027e-01 5.54693460e-01
1.66250646e+00 4.97150034e-01 4.10142958e-01 -8.30132216e-02
1.51155913e+00 -1.46353137e+00 -5.86955428e-01 -2.50562906e-01
8.84219468e-01 -8.56397688e-01 1.42314589e+00 2.59528190e-01
-9.59887385e-01 -8.40340078e-01 -8.47145498e-01 -4.65258569e-01
-6.20253563e-01 4.65226918e-02 4.70291406e-01 6.49806023e-01
-1.04777288e+00 2.07018163e-02 -4.29319084e-01 -3.62345219e-01
7.76382029e-01 -2.49168370e-02 -1.91832170e-01 -2.40570724e-01
-8.53886247e-01 9.66659904e-01 2.48967648e-01 -5.72525680e-01
-1.21721959e+00 -1.00276029e+00 -9.90944207e-01 1.58223033e-01
2.69796342e-01 -5.76327503e-01 1.53846085e+00 -8.30347359e-01
-1.11395931e+00 1.18458247e+00 6.71539977e-02 -3.15928996e-01
3.64121407e-01 -1.56144379e-02 -1.58255443e-01 -9.65559483e-03
3.15786595e-03 1.14464986e+00 9.19958830e-01 -1.23577142e+00
-6.84749722e-01 -4.35220987e-01 2.49865815e-01 4.02067691e-01
-9.16915953e-01 -2.12470055e-01 -4.73576456e-01 -7.56317139e-01
-7.29547739e-02 -8.81730258e-01 -1.07184201e-02 8.25811028e-02
-1.00378126e-01 -7.72928178e-01 8.28953207e-01 -3.01087826e-01
1.35711348e+00 -2.07703853e+00 -3.65992822e-02 -3.20562720e-01
6.65161431e-01 6.97796881e-01 -4.71470743e-01 1.27765328e-01
1.82045233e-02 -4.21944633e-02 3.21066707e-01 -4.43263143e-01
-2.22888850e-02 2.42260620e-01 -4.21017915e-01 7.68649802e-02
-1.95512295e-01 1.03291965e+00 -9.59623694e-01 -7.03452766e-01
3.04038674e-01 5.00571847e-01 -4.66902882e-01 4.09539551e-01
-1.27398178e-01 2.30622664e-01 -3.17366511e-01 7.22934186e-01
5.15058756e-01 -2.67400175e-01 -1.87078327e-01 -1.21540815e-01
7.36894682e-02 1.34405062e-01 -9.49304998e-01 1.64647269e+00
-6.02259338e-01 9.22174513e-01 -1.11654878e-01 -1.02718198e+00
9.44961727e-01 9.82481837e-02 2.21600473e-01 -7.07799077e-01
1.87894285e-01 -1.79457754e-01 -9.05601904e-02 -5.80711067e-01
5.97343922e-01 -5.35333641e-02 8.87509808e-02 4.72405434e-01
5.10179698e-01 -3.92403752e-01 1.83182076e-01 3.46103549e-01
4.77653205e-01 8.69822502e-03 3.32227618e-01 -2.28972405e-01
2.49304280e-01 -8.05329755e-02 1.63451001e-01 8.43211591e-01
-5.45116484e-01 3.19096237e-01 -4.77401279e-02 -1.49378002e-01
-5.92990637e-01 -1.19194663e+00 -2.08056256e-01 1.73293424e+00
1.27954036e-01 -3.72707367e-01 -6.84577584e-01 -7.87582338e-01
2.15743691e-01 8.81368697e-01 -5.02911925e-01 -3.29870135e-01
-1.15415238e-01 -2.31883198e-01 6.90070033e-01 7.31903970e-01
2.06517503e-01 -6.50403440e-01 -4.64848757e-01 -1.28571719e-01
4.88754772e-02 -9.71264839e-01 -5.27583897e-01 1.67173997e-01
-7.39574492e-01 -9.35262740e-01 -9.07171726e-01 -1.13646185e+00
7.65769720e-01 9.91781235e-01 1.21915340e+00 1.47498697e-02
-4.74173218e-01 8.18200409e-01 -2.77780145e-01 -1.03143311e+00
-3.56410056e-01 -1.55750751e-01 2.83102781e-01 -4.04354751e-01
9.24405873e-01 -3.14348608e-01 -2.84911782e-01 -4.49458212e-02
-8.57974172e-01 3.14818084e-01 3.59576434e-01 8.77918482e-01
6.64896190e-01 -2.36360699e-01 3.10323775e-01 -7.87730694e-01
8.55992973e-01 -3.31206352e-01 -5.86366057e-01 5.46333730e-01
-8.71017933e-01 -5.55459708e-02 2.99937010e-01 -1.07429147e+00
-9.03070390e-01 -2.48854399e-01 2.51302514e-02 -1.18231320e+00
-3.02671522e-01 1.51333913e-01 5.31064086e-02 -2.02996463e-01
5.83134949e-01 3.48040104e-01 -4.32759635e-02 -5.67699194e-01
8.28945935e-01 9.27989066e-01 3.80759031e-01 -7.16460645e-01
8.10479403e-01 2.45616436e-01 -4.58837211e-01 -1.09418774e+00
-1.21045101e+00 -6.45976007e-01 -6.41886413e-01 -2.18638927e-01
8.64535213e-01 -1.39672959e+00 -5.25715828e-01 3.28416675e-01
-7.80566871e-01 -4.05994862e-01 -5.78467190e-01 5.48203886e-01
-2.09935769e-01 2.35797212e-01 -3.62241149e-01 -6.89489484e-01
-2.80735075e-01 -1.35536158e+00 8.34974945e-01 7.50565290e-01
-9.67242271e-02 -1.15896857e+00 1.58752218e-01 9.49130356e-01
2.67123789e-01 -4.84567344e-01 8.41842175e-01 -1.11747098e+00
-2.58747756e-01 1.32370099e-01 -3.16034824e-01 5.17732203e-01
-6.45742416e-02 -4.07746546e-02 -1.43941319e+00 -5.29003501e-01
-2.74612874e-01 -9.92612600e-01 9.35105741e-01 3.49190950e-01
1.43727016e+00 -7.49437585e-02 -2.32333586e-01 6.63038671e-01
1.15254903e+00 -5.61433733e-02 2.13083252e-03 1.28222421e-01
1.09565401e+00 5.57970047e-01 6.06373072e-01 1.74678192e-01
7.14983582e-01 6.29342377e-01 2.90838301e-01 -1.30506307e-01
-7.68742979e-01 -5.06869495e-01 2.90738195e-01 1.08927631e+00
2.19182938e-01 -2.35779732e-01 -9.80633855e-01 7.87211239e-01
-1.47725797e+00 -7.89882839e-01 7.16056302e-02 2.01995468e+00
9.94853258e-01 -1.19497553e-01 4.17280383e-02 -2.84680337e-01
4.79749531e-01 2.42574751e-01 -5.06851733e-01 -6.25299394e-01
1.67497680e-01 1.91229492e-01 2.57181436e-01 4.70054001e-01
-9.30860758e-01 1.20511901e+00 5.52122641e+00 1.35920143e+00
-1.17845547e+00 4.29125071e-01 5.18336713e-01 -3.73677522e-01
-4.41318870e-01 -3.55149657e-01 -1.39300621e+00 2.97533184e-01
7.68554986e-01 -1.46136373e-01 9.92114767e-02 1.05934334e+00
-4.28431891e-02 6.99860603e-02 -1.18031633e+00 1.22047162e+00
4.25812125e-01 -1.20398843e+00 4.02192146e-01 3.34222540e-02
1.01529264e+00 1.03046253e-01 2.85210699e-01 9.15910721e-01
5.56552768e-01 -1.07499182e+00 6.12724662e-01 2.67486125e-01
8.58172953e-01 -4.97282565e-01 2.17689425e-01 3.95088017e-01
-1.16311634e+00 -2.49092039e-02 -5.46007395e-01 5.39489873e-02
-3.53596032e-01 1.90926388e-01 -1.00393903e+00 -8.80126134e-02
8.00368607e-01 6.11972988e-01 -9.78147149e-01 7.80470371e-01
-2.93199956e-01 6.97603703e-01 6.21109828e-02 -8.94513354e-02
4.48269457e-01 -1.57563239e-01 3.54454875e-01 1.17995870e+00
1.27660513e-01 -7.55432629e-05 5.12814641e-01 7.76228607e-01
-2.58005470e-01 8.13306347e-02 -8.42808068e-01 -1.92476913e-01
9.04345989e-01 1.27910733e+00 -3.52786124e-01 -6.56397820e-01
-8.16995740e-01 5.26090443e-01 6.52602851e-01 2.38094509e-01
-8.54685903e-01 -1.83751196e-01 9.65485036e-01 -1.01918153e-01
2.66794175e-01 1.95788238e-02 -1.23923697e-01 -1.20447123e+00
-3.28916728e-01 -9.54074204e-01 5.58736861e-01 -9.46831882e-01
-1.30619192e+00 2.67552823e-01 1.06524058e-01 -1.13558626e+00
2.21879855e-02 -5.31441510e-01 -5.52826822e-01 6.94935739e-01
-1.65991080e+00 -1.43618238e+00 -4.50835615e-01 6.86372042e-01
1.12226140e+00 -3.40150654e-01 7.85373211e-01 1.92950100e-01
-6.65231228e-01 1.10341728e+00 2.04332918e-01 5.49691729e-02
7.83163846e-01 -1.22891140e+00 1.42187355e-02 6.96465909e-01
6.20058239e-01 3.82435650e-01 6.13387704e-01 -3.40550989e-01
-1.26035404e+00 -1.09142280e+00 8.17458868e-01 -6.10857904e-01
8.23600769e-01 -3.45953196e-01 -1.01230526e+00 7.31111765e-01
2.65256941e-01 9.51808542e-02 9.42400634e-01 5.23476973e-02
-6.79548800e-01 -2.50729658e-02 -1.09782660e+00 6.53529763e-01
9.44149911e-01 -7.37144709e-01 -8.89947176e-01 3.62368673e-01
8.87315631e-01 -3.45317751e-01 -8.91108155e-01 2.08265230e-01
4.65429872e-01 -6.87031269e-01 1.26294506e+00 -9.27758932e-01
2.63299376e-01 9.75503922e-02 -1.62273526e-01 -1.38294601e+00
-3.34409326e-01 -5.33250757e-02 -4.22591805e-01 1.49309015e+00
4.60661128e-02 -3.17390472e-01 6.51929140e-01 4.44111288e-01
-1.92670554e-01 -1.05264914e+00 -6.12328887e-01 -8.72189224e-01
3.22852641e-01 -4.99423683e-01 4.83687758e-01 1.28908646e+00
-2.37486675e-01 5.68372726e-01 -6.24600612e-02 1.16806835e-01
5.73501766e-01 2.24292457e-01 9.83557642e-01 -1.20123482e+00
-1.21215470e-01 -6.66036367e-01 -3.90574306e-01 -1.31452763e+00
3.21622103e-01 -1.19456100e+00 -2.46798426e-01 -1.54176998e+00
4.68349904e-01 -4.72291172e-01 -3.34521919e-01 5.87016761e-01
-1.83159739e-01 4.00612831e-01 3.58611673e-01 7.48345628e-02
-7.19125748e-01 5.80630600e-01 1.48945653e+00 -6.10903561e-01
-6.40004054e-02 -1.17571495e-01 -8.17400515e-01 8.73800039e-01
6.12737179e-01 -3.65023673e-01 -9.19550002e-01 -6.05068326e-01
-4.70771015e-01 -3.40463966e-01 2.33730853e-01 -8.30792487e-01
1.46484792e-01 -3.78497750e-01 2.96318024e-01 -4.74620789e-01
4.83507693e-01 -7.27960765e-01 -5.70868015e-01 4.05089736e-01
-6.37343943e-01 -7.59832710e-02 3.42887253e-01 5.02213538e-01
-1.10510856e-01 -5.19219816e-01 1.03256452e+00 -6.87327981e-02
-1.17892420e+00 3.34583312e-01 -1.95567504e-01 5.63254714e-01
1.33069086e+00 -3.39716882e-01 -5.64833105e-01 -2.06101388e-01
-7.07162976e-01 4.41245407e-01 3.90752286e-01 8.72587919e-01
6.98269844e-01 -1.35326064e+00 -5.92838049e-01 4.91244286e-01
5.43098092e-01 5.16556501e-02 3.21803361e-01 7.14721382e-01
-2.99691230e-01 4.24189270e-01 -1.00102164e-01 -9.21831429e-01
-1.94853389e+00 7.18861222e-01 1.76121831e-01 -1.55964121e-01
-4.68749553e-01 1.45660424e+00 6.29283249e-01 -6.27029836e-01
6.93492651e-01 -3.18883598e-01 -4.15977567e-01 4.87012655e-01
7.35163808e-01 4.09847260e-01 -2.21042991e-01 -7.44330704e-01
-2.85291839e-02 5.20735443e-01 -3.31235349e-01 3.35081995e-01
1.11328924e+00 -3.49911660e-01 3.09097767e-01 4.83525395e-01
1.02989829e+00 -1.04777165e-01 -1.09894943e+00 -5.31218290e-01
-2.21070975e-01 -4.21665251e-01 3.07485402e-01 -6.16244197e-01
-9.48039532e-01 1.33997011e+00 9.53697562e-01 4.86027859e-02
9.06995535e-01 3.63143265e-01 6.00090384e-01 3.40792447e-01
3.31020020e-02 -9.93504405e-01 3.39767307e-01 5.88245571e-01
5.05140960e-01 -1.62717748e+00 5.73199131e-02 -1.02645434e-01
-7.53908336e-01 7.99106419e-01 1.26116812e+00 3.68247837e-01
6.10400975e-01 -5.42871132e-02 3.47970694e-01 -1.90083355e-01
-8.80809784e-01 -5.25873125e-01 7.58378327e-01 7.55257905e-01
5.00270367e-01 3.49862576e-01 1.06017798e-01 4.83811736e-01
-1.99146286e-01 -3.31075400e-01 5.44351161e-01 7.63290167e-01
-8.53530586e-01 -8.52103055e-01 -4.04834330e-01 5.01599610e-01
-6.07115496e-03 -2.85776973e-01 -3.48995805e-01 7.73726165e-01
3.42360795e-01 9.51001704e-01 1.53055489e-01 -2.95011908e-01
2.15451792e-01 3.55301499e-01 6.64662182e-01 -9.66639638e-01
-3.87044877e-01 -8.27741846e-02 -2.35558346e-01 -2.76797175e-01
-2.05402762e-01 -2.20057592e-01 -9.87386763e-01 -1.67545795e-01
-4.23464149e-01 -1.48409558e-02 6.00881755e-01 7.41638064e-01
2.49051839e-01 5.77261686e-01 2.64803320e-01 -7.12066352e-01
-9.18097675e-01 -1.03822351e+00 -4.91882116e-01 4.88639981e-01
3.11581820e-01 -9.29244280e-01 -2.27873728e-01 -6.79716095e-02] | [10.058477401733398, 2.14921236038208] |
e30408bf-6fd9-45ff-a791-67f0cc3fce02 | deep-reinforcement-learning-for-unknown | 2009.06847 | null | https://arxiv.org/abs/2009.06847v2 | https://arxiv.org/pdf/2009.06847v2.pdf | Toward Deep Supervised Anomaly Detection: Reinforcement Learning from Partially Labeled Anomaly Data | We consider the problem of anomaly detection with a small set of partially labeled anomaly examples and a large-scale unlabeled dataset. This is a common scenario in many important applications. Existing related methods either exclusively fit the limited anomaly examples that typically do not span the entire set of anomalies, or proceed with unsupervised learning from the unlabeled data. We propose here instead a deep reinforcement learning-based approach that enables an end-to-end optimization of the detection of both labeled and unlabeled anomalies. This approach learns the known abnormality by automatically interacting with an anomaly-biased simulation environment, while continuously extending the learned abnormality to novel classes of anomaly (i.e., unknown anomalies) by actively exploring possible anomalies in the unlabeled data. This is achieved by jointly optimizing the exploitation of the small labeled anomaly data and the exploration of the rare unlabeled anomalies. Extensive experiments on 48 real-world datasets show that our model significantly outperforms five state-of-the-art competing methods. | ['Anton Van Den Hengel', 'Chunhua Shen', 'Longbing Cao', 'Guansong Pang'] | 2020-09-15 | null | null | null | null | ['supervised-anomaly-detection'] | ['computer-vision'] | [ 1.27934381e-01 1.52108252e-01 2.26650074e-01 -5.40650904e-01
-9.77588832e-01 -4.34911251e-01 5.31982481e-01 4.07801151e-01
-2.28015140e-01 5.51124513e-01 -3.48105401e-01 -4.55868095e-01
2.33752709e-02 -4.40377325e-01 -7.71175683e-01 -7.32124865e-01
-5.12582600e-01 1.00371802e+00 2.56239325e-01 -8.26116651e-03
4.60527211e-01 5.13634205e-01 -1.60016787e+00 -3.01453561e-01
1.19135118e+00 1.12199450e+00 -5.53529739e-01 5.46606004e-01
-2.46200368e-01 5.75761437e-01 -5.55062950e-01 9.75325629e-02
5.29376209e-01 -4.06138808e-01 -4.59636539e-01 7.08959281e-01
4.38562632e-01 -5.25182545e-01 2.52328943e-02 1.28354800e+00
2.14752033e-01 3.26352626e-01 2.73447901e-01 -1.55545747e+00
-2.74666697e-01 1.61994264e-01 -8.82794142e-01 7.46606290e-01
1.65451393e-01 4.99184102e-01 7.09878683e-01 -7.82030404e-01
3.14093977e-01 9.44621086e-01 3.18489790e-01 5.62749803e-01
-1.24071360e+00 -4.71462876e-01 7.35469401e-01 1.12951525e-01
-1.03051901e+00 -1.53639629e-01 9.07367051e-01 -3.30671102e-01
8.34663212e-01 -6.75769104e-03 4.91023570e-01 1.12012112e+00
-4.11106646e-02 5.60924709e-01 7.05587924e-01 -2.40184173e-01
7.81922102e-01 -1.95278928e-01 2.29210377e-01 6.15505040e-01
3.35395724e-01 5.14410734e-01 -1.63755655e-01 -6.60105050e-01
2.75507927e-01 4.46333766e-01 1.05106868e-01 -8.16895008e-01
-8.24103177e-01 7.45172024e-01 -5.99336773e-02 -1.07977778e-01
-5.16859472e-01 -1.38636366e-01 4.01283026e-01 6.90852940e-01
5.40376246e-01 5.14879763e-01 -6.07656121e-01 -1.29371047e-01
-6.27212465e-01 3.83684009e-01 7.17842758e-01 4.87006336e-01
1.01295030e+00 6.57016337e-01 9.37377512e-02 5.50585449e-01
4.01403397e-01 3.57734233e-01 5.22251427e-01 -6.28430307e-01
3.69267672e-01 7.74081707e-01 4.61454630e-01 -2.16598466e-01
-2.86903769e-01 -4.24314708e-01 -3.20015699e-01 5.06292760e-01
7.84238160e-01 -6.85190856e-01 -1.36497498e+00 1.85780203e+00
7.24489391e-01 7.25412071e-01 1.65727735e-01 7.04835415e-01
-1.37026444e-01 4.30927485e-01 -7.09783211e-02 -2.28921428e-01
4.87791389e-01 -9.26354587e-01 -4.57532257e-01 -6.01269305e-01
7.22320259e-01 -2.67791510e-01 9.45483923e-01 4.65131879e-01
-8.78118098e-01 -2.00683951e-01 -7.94539273e-01 8.97985935e-01
-3.85286331e-01 -3.31542790e-01 2.21270815e-01 3.99014920e-01
-7.65615165e-01 5.69820642e-01 -1.19430161e+00 -3.88053417e-01
4.63031530e-01 3.25118840e-01 -9.80846509e-02 -3.81758176e-02
-7.56863356e-01 3.53724480e-01 4.26185995e-01 1.11297965e-01
-1.43336928e+00 -5.02233922e-01 -9.87051666e-01 -2.15297174e-02
1.03351390e+00 8.82133469e-02 1.37525177e+00 -1.28547585e+00
-1.19029677e+00 6.79865837e-01 -4.02426906e-03 -6.09191835e-01
5.48888624e-01 -5.60615361e-01 -7.22076416e-01 -1.16178207e-01
7.48289376e-02 1.17330067e-01 1.08884966e+00 -1.28163326e+00
-7.69779205e-01 -5.73444486e-01 -3.09733063e-01 3.28529924e-02
-1.18112296e-01 -2.52628177e-01 -4.56605367e-02 -6.01687014e-01
3.16051155e-01 -9.03482199e-01 -7.21349716e-01 -1.18518434e-01
-6.25605404e-01 5.72864227e-02 1.35716116e+00 -2.23836318e-01
9.25437927e-01 -2.24980164e+00 -1.91789314e-01 7.44849205e-01
1.20202027e-01 2.20510319e-01 -2.26672009e-01 4.30379719e-01
-3.67330313e-01 -1.78548753e-01 -6.91412032e-01 -2.81034172e-01
-1.56974718e-01 4.14080113e-01 -7.27912962e-01 6.71759546e-01
5.96233904e-01 5.60681522e-01 -1.12475193e+00 -7.08947107e-02
1.00054823e-01 -4.72727627e-01 -7.37750828e-01 7.08637238e-01
-5.99917948e-01 1.16272593e+00 -8.49560499e-01 1.05547893e+00
5.97018838e-01 -2.26944566e-01 -1.10181518e-01 9.50471222e-01
1.16861820e-01 -3.40484262e-01 -1.33190906e+00 1.41851282e+00
-2.39644200e-04 1.37429014e-01 -1.17021032e-01 -1.42706203e+00
9.19329345e-01 1.38302431e-01 7.40754128e-01 -4.44381028e-01
-1.18056148e-01 3.93989533e-01 3.06260705e-01 -6.07565701e-01
-1.98647324e-02 3.55300158e-02 -9.63139348e-03 8.53424370e-01
1.51087761e-01 1.55567393e-01 3.22401114e-02 9.96572245e-03
1.41875792e+00 2.79809106e-02 2.74558753e-01 -8.42553750e-02
5.56959808e-01 -8.87802765e-02 8.23932111e-01 1.00344789e+00
-5.47055244e-01 4.52644348e-01 7.57051170e-01 -8.20953548e-01
-9.62738693e-01 -1.38589609e+00 4.76644635e-02 8.63535583e-01
1.46995023e-01 1.31614327e-01 -6.07937753e-01 -1.26531804e+00
7.43257403e-02 8.70072663e-01 -7.86021531e-01 -3.93997431e-01
-5.13304651e-01 -8.56849432e-01 1.05770908e-01 4.28097397e-01
1.36720374e-01 -1.39244044e+00 -7.51728952e-01 2.08549783e-01
2.90078282e-01 -7.72373438e-01 -1.46817699e-01 4.80033010e-01
-9.82327282e-01 -1.18166649e+00 -1.56751350e-01 -4.66713905e-01
1.05666339e+00 -3.55934113e-01 1.08027971e+00 2.54472286e-01
-4.59735781e-01 5.53202510e-01 -3.48163515e-01 -6.17046058e-01
-3.48509580e-01 -4.29870576e-01 2.70684451e-01 4.71688658e-01
5.75526893e-01 -7.13357627e-01 -2.26368472e-01 2.62360007e-01
-1.06158376e+00 -9.13160503e-01 3.77025366e-01 1.21616018e+00
8.09939563e-01 -9.66787562e-02 8.93354356e-01 -1.15635478e+00
3.31970781e-01 -9.71740007e-01 -1.02101064e+00 4.94952612e-02
-7.47141838e-01 2.32058004e-01 7.03931153e-01 -6.62417054e-01
-8.85116339e-01 5.73969595e-02 4.40996280e-03 -9.29039717e-01
-7.76042223e-01 3.05206805e-01 -1.32899955e-01 1.23920329e-01
6.49855494e-01 2.80058533e-01 7.66281039e-02 -3.59603912e-01
-5.35125323e-02 1.22908853e-01 5.13551295e-01 -7.00849831e-01
1.01278687e+00 5.29687464e-01 -1.39834091e-01 -5.97226501e-01
-7.95593202e-01 -5.78559935e-01 -4.92502064e-01 -8.83297324e-02
3.79376173e-01 -5.87063491e-01 -9.51829553e-02 6.08639896e-01
-4.94136661e-01 -4.16526169e-01 -8.82003307e-01 3.28314811e-01
-5.10430276e-01 5.59440136e-01 -1.73040971e-01 -1.02719426e+00
-3.55571806e-02 -1.23496962e+00 1.18857515e+00 1.27569154e-01
-7.12751821e-02 -9.40569222e-01 6.06759787e-01 -2.84648538e-01
3.15467626e-01 4.92634445e-01 9.23509061e-01 -1.86705959e+00
-5.16126335e-01 -6.91362858e-01 1.95279762e-01 2.67469138e-01
5.60559481e-02 4.88115437e-02 -9.56705630e-01 -4.29290742e-01
4.71982360e-02 -6.50321662e-01 6.12633526e-01 1.75103486e-01
1.62963474e+00 -2.06205308e-01 -2.84557223e-01 4.60053086e-01
1.21901929e+00 4.38534826e-01 3.87864828e-01 3.25149804e-01
4.97002125e-01 5.18157184e-01 9.41865265e-01 8.00182581e-01
-1.72175899e-01 1.47816077e-01 1.04601884e+00 -4.36980603e-03
9.81175661e-01 -7.26207644e-02 2.30671480e-01 1.83842853e-01
4.88920450e-01 -3.17831427e-01 -1.17842567e+00 8.19398403e-01
-2.03942561e+00 -7.91376114e-01 4.39375013e-01 2.48659301e+00
3.26011837e-01 5.51618218e-01 2.89460689e-01 1.43687710e-01
7.58254051e-01 1.19013153e-01 -1.37969303e+00 -4.52963710e-01
2.57579476e-01 1.20017223e-01 3.82937826e-02 3.14887911e-01
-1.33582270e+00 6.52357459e-01 6.17793989e+00 2.03777954e-01
-1.04157305e+00 -4.01432842e-01 7.83883750e-01 -9.91320908e-02
-1.68116495e-01 4.70980769e-03 -3.55908394e-01 4.04293567e-01
1.13952529e+00 -1.00217663e-01 3.23700905e-01 1.09753799e+00
1.60266981e-02 -2.29253948e-01 -1.39368904e+00 5.08193910e-01
-1.81357026e-01 -7.25621521e-01 -3.63276042e-02 1.12578869e-01
9.38608468e-01 2.75838554e-01 1.92069843e-01 5.10304034e-01
4.53013569e-01 -7.40302563e-01 3.83252263e-01 4.25174803e-01
4.15361077e-01 -7.12740362e-01 7.04688847e-01 5.20668328e-01
-6.81934118e-01 -5.57237625e-01 2.03841669e-03 4.44726786e-03
-2.52505410e-02 4.54711199e-01 -6.11200035e-01 8.57571065e-02
7.11178243e-01 7.29605198e-01 -5.61832488e-01 1.16842330e+00
2.44707125e-03 8.96875322e-01 -5.52794456e-01 4.12432104e-01
6.23429477e-01 -3.95662695e-01 1.00894415e+00 5.78106165e-01
3.48332375e-01 -1.49321809e-01 7.57462204e-01 8.88134897e-01
1.73841342e-01 3.71624716e-02 -8.68168890e-01 -2.19851807e-01
3.25361073e-01 1.01624620e+00 -7.68213987e-01 -2.59002358e-01
-6.05468929e-01 6.80090249e-01 3.31390232e-01 5.92709959e-01
-6.57763064e-01 -1.25094071e-01 6.33329511e-01 -1.28584191e-01
4.27032799e-01 3.40684980e-01 -6.50236979e-02 -1.21359432e+00
2.09677234e-01 -1.01041996e+00 7.62042820e-01 -1.94835469e-01
-1.57901371e+00 5.26362896e-01 -1.02736332e-01 -1.59976423e+00
-7.77506411e-01 -3.35739464e-01 -1.37367857e+00 7.00623095e-01
-1.41174257e+00 -5.06676853e-01 -1.09135367e-01 5.15709698e-01
6.96179926e-01 -4.76230770e-01 7.92711437e-01 -5.96141331e-02
-9.23627555e-01 5.05953491e-01 3.96217376e-01 4.75029461e-02
6.70697927e-01 -1.60434234e+00 6.99324906e-01 1.25145030e+00
1.86639324e-01 9.73616466e-02 7.06446946e-01 -8.14899147e-01
-9.18740988e-01 -1.28950369e+00 1.12496428e-02 -4.26137716e-01
8.92696559e-01 -1.83815539e-01 -1.49660456e+00 1.02098715e+00
-2.25710705e-01 8.31620038e-01 5.96106768e-01 -1.86956957e-01
-2.63995230e-01 1.46070734e-01 -1.49375999e+00 5.74050486e-01
7.24843204e-01 -5.91972955e-02 -5.33696353e-01 2.77960777e-01
4.33763623e-01 -4.77159172e-01 -1.70506164e-01 4.96180773e-01
-1.19027840e-02 -9.47829068e-01 5.96699238e-01 -1.26631212e+00
1.88286424e-01 -2.10948065e-01 -2.55360901e-02 -1.51429749e+00
3.84265095e-01 -5.84482431e-01 -6.33363008e-01 7.63748229e-01
5.00130415e-01 -1.01070201e+00 9.67950344e-01 5.61896026e-01
-2.00792983e-01 -1.00167656e+00 -1.02653098e+00 -6.77170575e-01
-3.55020203e-02 -4.91577715e-01 5.72588384e-01 8.93927276e-01
-2.00339600e-01 -3.56617063e-01 -1.53999925e-01 7.54445374e-01
8.92274380e-01 3.02207172e-01 6.43725693e-01 -1.41885281e+00
-3.13288063e-01 -1.02714896e-01 -5.73028028e-01 -4.79477435e-01
3.61172199e-01 -3.49158257e-01 3.36994439e-01 -5.76923311e-01
-1.71948373e-01 -2.63720900e-01 -6.35868490e-01 3.91166180e-01
-4.88131672e-01 -2.48588443e-01 -6.42927706e-01 -4.82768333e-03
-9.65917110e-01 6.24882638e-01 7.53549099e-01 7.56764114e-02
-3.11368376e-01 4.71770972e-01 -4.19923872e-01 1.05685127e+00
8.80259454e-01 -5.63598514e-01 -4.38854456e-01 -1.25096023e-01
-3.30001026e-01 9.32816789e-02 2.38495603e-01 -9.67643738e-01
-1.03065334e-01 -4.61188644e-01 3.50336313e-01 -5.80040693e-01
-7.72973290e-03 -9.72111225e-01 -5.11892557e-01 5.41127980e-01
-4.21380758e-01 3.52559805e-01 3.26578528e-01 1.23154879e+00
-2.72385806e-01 -7.90318474e-02 8.76482010e-01 -1.24280192e-01
-9.25244153e-01 5.48841655e-01 -3.95186990e-01 6.67043269e-01
1.41476715e+00 -1.32768974e-01 6.93311775e-03 -5.22314847e-01
-9.39640999e-01 7.56601334e-01 5.20646751e-01 3.68855000e-01
6.40788138e-01 -9.98446763e-01 -5.21414280e-01 8.15162897e-01
4.94198084e-01 2.76836067e-01 1.30303085e-01 5.04080951e-01
-1.78138152e-01 -2.17958689e-01 -1.86393499e-01 -9.49942291e-01
-7.11719751e-01 8.63312542e-01 5.37791431e-01 -5.17143726e-01
-8.18147838e-01 6.52594864e-01 2.41662696e-01 -6.92207754e-01
5.47383547e-01 6.24690950e-03 -8.79490301e-02 -4.40474272e-01
4.88311023e-01 4.23134118e-01 7.32608736e-02 -3.12945038e-01
-1.66092381e-01 2.19775200e-01 -3.28253061e-01 8.36040732e-03
1.24346387e+00 -5.18772714e-02 1.82083786e-01 6.48110688e-01
8.32873344e-01 -1.10752918e-01 -1.59619129e+00 -5.27317762e-01
5.36193073e-01 -5.59870481e-01 -4.64504004e-01 -7.22203255e-01
-1.30620539e+00 8.09058845e-01 8.45156014e-01 1.98087215e-01
1.25898147e+00 -1.10333063e-01 4.90428299e-01 7.69185007e-01
2.28419423e-01 -1.04485071e+00 5.17348111e-01 4.75685209e-01
5.00225067e-01 -1.71446574e+00 -3.82862121e-01 9.47035179e-02
-8.69362175e-01 9.40352380e-01 1.31675422e+00 -5.16140878e-01
7.93003261e-01 2.94496268e-01 3.38822693e-01 -3.75867695e-01
-8.96164715e-01 -1.74919173e-01 -1.30623505e-02 4.77655917e-01
-3.09319317e-01 -3.03216487e-01 2.69031018e-01 4.37603861e-01
4.51406986e-01 -4.43031788e-01 6.16757929e-01 1.44295442e+00
-5.02876759e-01 -8.93489182e-01 -5.40663481e-01 9.79662240e-01
-6.14099681e-01 2.60364741e-01 -3.83711487e-01 6.55553639e-01
-6.09209687e-02 6.93354845e-01 4.06638861e-01 5.65311275e-02
3.63905817e-01 5.19361913e-01 -2.00638056e-01 -7.99473107e-01
-1.87182143e-01 1.99870601e-01 -3.32948416e-01 -8.94670665e-01
1.41593114e-01 -1.01862538e+00 -1.12633920e+00 3.06564540e-01
-3.12567562e-01 2.55928844e-01 1.93724930e-01 1.23767078e+00
2.15773091e-01 5.23864031e-01 1.00434494e+00 -8.05825710e-01
-1.06481028e+00 -7.78929532e-01 -8.34298790e-01 6.62453055e-01
9.43182945e-01 -7.63357460e-01 -7.08092690e-01 -2.80640364e-01] | [7.612226963043213, 2.387421131134033] |
a1663058-deb6-4a2c-ac21-bd52ed5f46c4 | deep-networks-with-internal-selective | 1407.3068 | null | http://arxiv.org/abs/1407.3068v2 | http://arxiv.org/pdf/1407.3068v2.pdf | Deep Networks with Internal Selective Attention through Feedback Connections | Traditional convolutional neural networks (CNN) are stationary and
feedforward. They neither change their parameters during evaluation nor use
feedback from higher to lower layers. Real brains, however, do. So does our
Deep Attention Selective Network (dasNet) architecture. DasNets feedback
structure can dynamically alter its convolutional filter sensitivities during
classification. It harnesses the power of sequential processing to improve
classification performance, by allowing the network to iteratively focus its
internal attention on some of its convolutional filters. Feedback is trained
through direct policy search in a huge million-dimensional parameter space,
through scalable natural evolution strategies (SNES). On the CIFAR-10 and
CIFAR-100 datasets, dasNet outperforms the previous state-of-the-art model. | ['Marijn Stollenga', 'Juergen Schmidhuber', 'Jonathan Masci', 'Faustino Gomez'] | 2014-07-11 | deep-networks-with-internal-selective-1 | http://papers.nips.cc/paper/5276-deep-networks-with-internal-selective-attention-through-feedback-connections | http://papers.nips.cc/paper/5276-deep-networks-with-internal-selective-attention-through-feedback-connections.pdf | neurips-2014-12 | ['deep-attention', 'deep-attention'] | ['computer-vision', 'natural-language-processing'] | [ 8.96374807e-02 1.65625766e-01 -2.05824412e-02 -1.39100373e-01
1.92057505e-01 -6.19986057e-01 5.30741453e-01 -3.93086642e-01
-9.12385285e-01 7.89929032e-01 1.97618499e-01 -2.47033983e-01
-4.89199832e-02 -5.28151870e-01 -6.62924290e-01 -7.28484869e-01
-1.35400087e-01 2.94136345e-01 6.62768781e-01 -3.53899688e-01
1.87212467e-01 6.55338645e-01 -1.32252073e+00 3.27534616e-01
4.97254133e-01 1.01864874e+00 2.57268727e-01 1.09348226e+00
1.31944120e-01 9.17913139e-01 -7.76145399e-01 -3.24871391e-01
3.15089911e-01 -4.39930141e-01 -9.80940819e-01 -4.58444923e-01
3.47677052e-01 -1.27795935e-01 -5.91746867e-01 1.14647090e+00
5.04417658e-01 3.14083993e-01 3.34503949e-01 -9.42806542e-01
-1.16112554e+00 8.85741174e-01 -5.78710996e-03 1.02341092e+00
-4.76037264e-01 7.27704823e-01 8.41680229e-01 -6.96702600e-01
3.48820418e-01 1.39146805e+00 6.09112561e-01 1.07473385e+00
-1.24934936e+00 -6.66664541e-01 6.64309382e-01 -5.50358631e-02
-8.67699683e-01 -4.21669304e-01 3.87824923e-01 -4.20712471e-01
1.60297084e+00 2.02020090e-02 1.07155108e+00 1.20495462e+00
4.40828294e-01 8.37908626e-01 5.50097644e-01 -3.42751384e-01
3.88089955e-01 -1.20921180e-01 2.52052248e-02 7.77627409e-01
1.88563336e-02 4.13252890e-01 -3.88088405e-01 -6.54347986e-02
1.11163855e+00 -5.25795408e-02 -4.15531069e-01 -4.72994009e-03
-8.96897733e-01 7.82564521e-01 8.78126681e-01 2.56111056e-01
-4.92155313e-01 8.09152484e-01 2.40168542e-01 7.21288264e-01
1.93022192e-01 1.27715421e+00 -1.01636302e+00 -5.98127991e-02
-5.54568470e-01 1.09968357e-01 5.42397857e-01 5.68878114e-01
6.86411738e-01 5.54037809e-01 -4.77814138e-01 7.13893831e-01
2.17955783e-01 1.00515783e-01 9.18237209e-01 -1.15155315e+00
-5.23360893e-02 6.34412050e-01 -3.28343153e-01 -5.75070918e-01
-4.47013497e-01 -1.01578283e+00 -6.99065208e-01 5.04076481e-01
1.46777257e-01 -7.66716063e-01 -1.19327426e+00 1.80560327e+00
-1.47415131e-01 1.14046782e-01 1.40017658e-01 8.11413407e-01
4.79071230e-01 2.90452480e-01 1.57253653e-01 1.86304361e-01
8.82434547e-01 -1.13661504e+00 -3.33254248e-01 -6.77088737e-01
2.52651662e-01 -2.22125456e-01 9.73843336e-01 3.47512215e-01
-1.40547132e+00 -4.90524650e-01 -1.11807954e+00 1.79605216e-01
-4.27371383e-01 7.29056969e-02 9.47345674e-01 4.17405099e-01
-1.54826474e+00 7.73406506e-01 -1.03748512e+00 -7.61604160e-02
9.29227352e-01 8.79496157e-01 1.81515142e-01 3.63702357e-01
-1.26214504e+00 8.49091709e-01 5.22715271e-01 1.22081213e-01
-1.16907656e+00 -1.05992246e+00 -3.72082502e-01 5.11158764e-01
2.39228278e-01 -9.56707418e-01 1.76869404e+00 -1.63041031e+00
-1.86932230e+00 3.18068415e-01 2.71850079e-01 -8.89373958e-01
3.47933829e-01 -3.72736990e-01 -1.34837747e-01 9.51444656e-02
-6.75046742e-01 1.33689761e+00 1.01988983e+00 -7.72830248e-01
-5.93092263e-01 1.89621329e-01 2.05752432e-01 -1.72899202e-01
-6.54038906e-01 1.47812337e-01 -2.94500560e-01 -6.86355829e-01
-1.65251702e-01 -9.18402016e-01 -4.22271252e-01 7.26289973e-02
-1.00342548e-02 -3.64290148e-01 8.43733072e-01 -8.27449262e-02
1.15093160e+00 -2.04078579e+00 3.88724506e-01 -2.89695952e-02
2.90742546e-01 7.77271986e-01 -2.84684092e-01 -1.93797052e-01
-1.45943448e-01 3.81689578e-01 -1.46041483e-01 -1.29558057e-01
-1.92962289e-01 2.79779017e-01 -1.25031576e-01 9.92869213e-02
8.19903135e-01 1.29284132e+00 -1.03292382e+00 -7.60742500e-02
-1.76447108e-01 4.74453151e-01 -1.10744631e+00 -8.11792389e-02
-5.15707135e-01 1.33812338e-01 -2.27957025e-01 4.07453090e-01
1.79571733e-01 -5.11154711e-01 -1.11570112e-01 1.88682064e-01
-2.15469003e-01 1.11287951e-01 -4.92728412e-01 1.45384336e+00
-2.94059068e-01 1.10043395e+00 -1.53999165e-01 -8.17242444e-01
6.92869425e-01 1.92950115e-01 1.13804653e-01 -8.11117887e-01
4.19805318e-01 -1.41304150e-01 6.71804130e-01 -2.01487780e-01
1.51467726e-01 4.95275587e-01 2.84140438e-01 2.78078586e-01
5.26609361e-01 1.98290616e-01 1.29131347e-01 1.22309417e-01
1.46080935e+00 -3.06972802e-01 -4.58994247e-02 -4.60106254e-01
3.30915034e-01 -9.97674540e-02 6.96774006e-01 9.81775999e-01
-1.65736929e-01 3.91963869e-01 6.78843319e-01 -7.31583834e-01
-1.00694847e+00 -8.55498731e-01 1.12662844e-01 1.46948326e+00
-5.30383885e-01 -3.24251167e-02 -8.85550439e-01 -6.47407830e-01
5.41697741e-02 5.63232422e-01 -1.20514345e+00 -8.36587906e-01
-5.98870158e-01 -6.06481552e-01 6.80596828e-01 7.46597886e-01
8.30219269e-01 -1.50122523e+00 -1.04526842e+00 3.68097723e-01
6.52424634e-01 -4.57239956e-01 -4.01528060e-01 7.43143320e-01
-9.57066655e-01 -8.71929765e-01 -6.17423773e-01 -6.97808206e-01
6.17196202e-01 -2.25156099e-01 1.33603907e+00 3.08373481e-01
-3.78450185e-01 1.50613561e-01 -9.98189226e-02 -6.84721231e-01
-2.37386256e-01 5.77720046e-01 -6.28779829e-02 -2.33083174e-01
2.44823918e-01 -5.77925384e-01 -6.44202113e-01 3.60769629e-01
-6.44214332e-01 -1.68186530e-01 7.87126660e-01 1.10566831e+00
3.41491193e-01 -1.02860719e-01 6.20463431e-01 -8.15834463e-01
8.74821067e-01 -4.18924421e-01 -8.57117712e-01 3.39569718e-01
-5.58167458e-01 4.21181768e-01 6.30972326e-01 -9.32743013e-01
-1.02267575e+00 4.57068998e-03 2.28941008e-01 -9.99832153e-01
2.72035986e-01 2.48684838e-01 3.83256137e-01 -2.97282428e-01
1.01183832e+00 2.27916278e-02 -3.39574635e-01 -4.01111275e-01
9.50617492e-02 7.85758197e-02 8.59935343e-01 -2.70672739e-01
4.02824163e-01 8.16819221e-02 -4.15328413e-01 -4.78519052e-01
-7.45386422e-01 1.30845666e-01 -5.97440362e-01 -2.18287930e-01
8.25699747e-01 -6.13232851e-01 -9.55164075e-01 5.33622205e-01
-1.04777908e+00 -1.02070487e+00 -5.90574086e-01 3.51931095e-01
-1.41277701e-01 -8.36275816e-01 -7.87870109e-01 -5.64505935e-01
-3.65807503e-01 -1.03695595e+00 2.28609100e-01 7.56229997e-01
-2.43098095e-01 -1.13157642e+00 1.28057525e-02 -6.32827222e-01
1.01587415e+00 -3.00100625e-01 8.51573169e-01 -4.77577269e-01
-4.84901369e-01 2.93831468e-01 -1.49486363e-01 5.41184008e-01
-4.21924219e-02 3.36623043e-01 -9.51415896e-01 -4.87986565e-01
-4.30753112e-01 -5.06206274e-01 1.47245061e+00 5.18132329e-01
1.51088071e+00 -4.31114823e-01 -1.69712722e-01 1.00790012e+00
1.06470942e+00 2.94511616e-01 5.11204720e-01 4.49162483e-01
2.42017597e-01 1.10173583e-01 -2.25213245e-01 1.60592526e-01
-1.74554467e-01 -2.29356028e-02 7.24387109e-01 -1.08059764e-01
-3.26019615e-01 1.12108007e-01 2.40961164e-01 4.93045926e-01
-1.25871986e-01 -3.54612678e-01 -1.02879214e+00 4.06550109e-01
-1.88909876e+00 -8.53769004e-01 4.35653955e-01 1.56075144e+00
8.07883143e-01 5.41559339e-01 -2.26690546e-01 -2.46034324e-01
3.76202196e-01 -1.88606709e-01 -1.33616745e+00 -7.08806992e-01
-3.20679218e-01 4.50239986e-01 8.00625443e-01 1.72185093e-01
-1.00308371e+00 1.18616366e+00 7.62559891e+00 1.01882547e-01
-1.52126658e+00 6.15310781e-02 7.78004766e-01 -6.26598358e-01
-1.09521046e-01 -3.65136951e-01 -7.63745666e-01 3.20717633e-01
1.09140384e+00 -3.86745818e-02 7.93898046e-01 8.60405624e-01
-3.43889087e-01 3.02716736e-02 -9.41996694e-01 7.42325783e-01
-2.32385695e-01 -1.85792100e+00 -7.01984391e-02 -3.86629365e-02
1.12777841e+00 8.07185054e-01 4.76250678e-01 5.58585405e-01
1.30170035e+00 -1.12448323e+00 7.22410798e-01 8.58071268e-01
6.64825618e-01 -8.76906633e-01 5.92646539e-01 1.58595785e-01
-5.16103983e-01 -7.90960193e-01 -7.34128416e-01 -1.05322063e-01
-4.30846959e-01 6.46590739e-02 -7.43039489e-01 -5.79372287e-01
1.08695042e+00 4.05320108e-01 -9.56324279e-01 1.23192799e+00
-3.94846439e-01 9.28133607e-01 -1.17423102e-01 -3.85377139e-01
6.42654538e-01 4.18059051e-01 4.13153201e-01 1.21341670e+00
-1.35981962e-01 2.70649672e-01 -1.78148910e-01 9.46524680e-01
-6.43269360e-01 -6.76471531e-01 2.35417206e-02 -2.47697845e-01
5.89287758e-01 1.01980960e+00 -6.12963736e-01 -3.44623685e-01
-1.60021976e-01 7.90976226e-01 7.03423440e-01 6.10359550e-01
-5.22910416e-01 -4.87556726e-01 1.03772926e+00 -2.81042486e-01
7.41107881e-01 -1.11249574e-01 -1.50287777e-01 -8.38106394e-01
-5.99080324e-01 -7.88073480e-01 2.58848637e-01 -7.31823564e-01
-1.05674958e+00 8.49315643e-01 -3.26196223e-01 -4.02454913e-01
-2.11827293e-01 -6.85069919e-01 -7.84526944e-01 8.24155450e-01
-1.49882317e+00 -4.33040917e-01 -1.52148288e-02 7.76214659e-01
6.37122512e-01 -5.78064263e-01 7.25481212e-01 -1.06736207e-02
-7.39239156e-01 8.59143555e-01 4.99648862e-02 3.94465387e-01
2.90585279e-01 -1.25866127e+00 8.50356460e-01 7.11578548e-01
-1.70720726e-01 7.52464771e-01 4.78052944e-01 -4.31242973e-01
-1.10027301e+00 -1.13952219e+00 2.82560915e-01 -9.19280648e-02
6.98780417e-01 -3.67124796e-01 -1.03003800e+00 7.00057149e-01
4.24754649e-01 3.64966720e-01 2.47344926e-01 9.27387699e-02
-3.77398491e-01 -4.01317813e-02 -9.34376419e-01 7.02233016e-01
1.20119917e+00 -2.64654338e-01 -4.34287757e-01 3.83812822e-02
1.17727792e+00 -3.39436203e-01 -4.00180101e-01 7.72252381e-02
5.24258077e-01 -6.83477879e-01 7.73905694e-01 -1.31020069e+00
3.07321876e-01 1.13285191e-01 1.40315831e-01 -1.92252815e+00
-9.22499895e-01 -9.68569398e-01 -3.59232128e-01 6.42068744e-01
6.91826642e-01 -8.97408128e-01 8.76928926e-01 4.31613326e-01
-3.29779893e-01 -8.46388757e-01 -5.72487950e-01 -4.70479131e-01
2.74230868e-01 -1.07528284e-01 7.50405908e-01 7.21971154e-01
-3.19360375e-01 2.83218443e-01 1.87772095e-01 1.59558188e-02
1.97029889e-01 -6.07632399e-01 1.53273419e-01 -1.41690147e+00
-5.56007206e-01 -1.04975545e+00 -2.79783607e-01 -8.81580889e-01
9.49562192e-02 -6.21546805e-01 -4.68450859e-02 -1.04935610e+00
-2.24902719e-01 -4.19748485e-01 -8.40729594e-01 8.20527613e-01
2.36268551e-03 8.70017335e-02 3.21448296e-01 8.24953988e-02
-6.06413007e-01 4.84077960e-01 1.30930376e+00 -2.66122729e-01
-2.50579625e-01 -5.93571179e-02 -8.72343540e-01 6.76854789e-01
1.08143747e+00 -5.05593359e-01 -5.21193027e-01 -9.68566895e-01
6.20225012e-01 -3.50415230e-01 4.51930672e-01 -1.26181984e+00
7.45834649e-01 -1.18996747e-01 1.00234985e+00 -1.06286041e-01
2.70968899e-02 -3.72402072e-01 -2.47098759e-01 8.57826889e-01
-8.90849650e-01 5.07545114e-01 4.92812127e-01 3.41931105e-01
1.11581661e-01 -7.77716115e-02 1.08836758e+00 -4.86506104e-01
-7.05344617e-01 4.08953398e-01 -7.14766085e-01 1.44541889e-01
6.30840421e-01 -6.88790530e-02 -5.92292666e-01 -2.55275853e-02
-8.08514059e-01 5.23543239e-01 1.81547880e-01 5.09599149e-01
5.29292107e-01 -1.02485216e+00 -6.52108848e-01 4.32147324e-01
-3.45847011e-01 1.13202929e-01 -7.54416957e-02 4.56331521e-01
-3.89619648e-01 3.94647866e-01 -4.61326748e-01 -5.00498235e-01
-8.43662918e-01 4.25872594e-01 1.04829502e+00 5.29974811e-02
-2.88455099e-01 1.71245241e+00 5.46532646e-02 -2.03225061e-01
4.83943909e-01 -4.47498649e-01 -2.91880757e-01 -9.56642032e-02
7.59935021e-01 1.07411221e-01 2.67053604e-01 2.40778297e-01
-1.70965075e-01 -1.16599984e-01 -4.01510984e-01 -8.44549611e-02
1.65435266e+00 2.66413420e-01 -1.70671139e-02 2.22164050e-01
1.21310484e+00 -7.18824863e-01 -1.91536295e+00 -4.43741977e-01
-9.67892110e-02 -7.48162046e-02 6.46724701e-01 -1.25905919e+00
-1.45303500e+00 9.16809678e-01 5.50222099e-01 6.55792840e-03
1.09650981e+00 -8.60090628e-02 3.68006080e-01 9.67715621e-01
-1.37451813e-01 -1.24027812e+00 5.88128924e-01 1.19018269e+00
9.84554589e-01 -9.10442531e-01 -3.11032087e-01 5.46549559e-01
-4.63039070e-01 1.27089536e+00 1.02674317e+00 -5.07041872e-01
8.17357242e-01 5.82035661e-01 -7.47085512e-02 -3.86980355e-01
-1.65597725e+00 -1.30057679e-02 1.14881940e-01 5.13586164e-01
1.85332283e-01 -2.86325276e-01 5.03671229e-01 2.34826595e-01
-3.83530021e-01 -8.90211910e-02 2.47450843e-01 9.57863510e-01
-7.33493805e-01 -6.45909309e-01 -9.31141526e-02 5.92192888e-01
-4.62022543e-01 -2.84499049e-01 -6.23968422e-01 6.51476920e-01
7.76620284e-02 5.68587720e-01 6.09645486e-01 -2.91746348e-01
1.75628200e-01 -1.50357150e-02 4.50018317e-01 -4.84828413e-01
-1.23282743e+00 -2.29581088e-01 -4.30633754e-01 -5.68104029e-01
-1.65520664e-02 -8.52452278e-01 -1.24263096e+00 -2.67538667e-01
-3.09243917e-01 1.44384280e-01 6.75835311e-01 6.84846640e-01
3.40862989e-01 1.40962565e+00 4.49024439e-01 -8.25420022e-01
-6.96053863e-01 -8.93722653e-01 4.34315391e-02 -2.55922526e-01
7.44578779e-01 -6.11348033e-01 -3.36987019e-01 -1.95039973e-01] | [8.667489051818848, 3.1673424243927] |
176ea388-f093-4aef-aa1c-570b50011979 | woad-weakly-supervised-online-action | 2006.03732 | null | https://arxiv.org/abs/2006.03732v2 | https://arxiv.org/pdf/2006.03732v2.pdf | WOAD: Weakly Supervised Online Action Detection in Untrimmed Videos | Online action detection in untrimmed videos aims to identify an action as it happens, which makes it very important for real-time applications. Previous methods rely on tedious annotations of temporal action boundaries for training, which hinders the scalability of online action detection systems. We propose WOAD, a weakly supervised framework that can be trained using only video-class labels. WOAD contains two jointly-trained modules, i.e., temporal proposal generator (TPG) and online action recognizer (OAR). Supervised by video-class labels, TPG works offline and targets at accurately mining pseudo frame-level labels for OAR. With the supervisory signals from TPG, OAR learns to conduct action detection in an online fashion. Experimental results on THUMOS'14, ActivityNet1.2 and ActivityNet1.3 show that our weakly-supervised method largely outperforms weakly-supervised baselines and achieves comparable performance to the previous strongly-supervised methods. Beyond that, WOAD is flexible to leverage strong supervision when it is available. When strongly supervised, our method obtains the state-of-the-art results in the tasks of both online per-frame action recognition and online detection of action start. | ['ran Xu', 'Richard Socher', 'Mingfei Gao', 'Yingbo Zhou', 'Caiming Xiong'] | 2020-06-05 | null | http://openaccess.thecvf.com//content/CVPR2021/html/Gao_WOAD_Weakly_Supervised_Online_Action_Detection_in_Untrimmed_Videos_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Gao_WOAD_Weakly_Supervised_Online_Action_Detection_in_Untrimmed_Videos_CVPR_2021_paper.pdf | cvpr-2021-1 | ['online-action-detection'] | ['computer-vision'] | [ 4.81831849e-01 4.42373417e-02 -1.07091177e+00 -1.29263416e-01
-9.37813401e-01 -4.51234639e-01 5.89464724e-01 -3.21261704e-01
-3.68306577e-01 3.92231703e-01 5.54304898e-01 -8.99012461e-02
5.48215449e-01 -2.65307158e-01 -6.65254295e-01 -6.00055456e-01
-4.27600503e-01 1.18404195e-01 7.71274328e-01 2.61436552e-01
-3.39891738e-03 1.11920021e-01 -1.47370100e+00 7.66687930e-01
2.45807484e-01 1.16739559e+00 -1.13029005e-02 9.77011502e-01
4.27068204e-01 1.87958252e+00 -2.73818314e-01 -1.89811029e-02
4.39025462e-01 -7.94708669e-01 -8.82679999e-01 6.63267910e-01
3.91782701e-01 -9.26223576e-01 -7.95902848e-01 5.57019413e-01
1.92151561e-01 1.21860400e-01 3.54806155e-01 -1.52853346e+00
-1.35814548e-01 4.80870366e-01 -3.99787903e-01 5.59416711e-01
6.17006958e-01 6.94451034e-01 1.19276798e+00 -7.88624525e-01
7.09366322e-01 1.15457618e+00 3.97110790e-01 7.63782024e-01
-9.49726105e-01 -3.98347616e-01 5.58638036e-01 4.17299509e-01
-1.08727038e+00 -8.49409759e-01 4.76703614e-01 -4.73554045e-01
1.13576627e+00 -6.23147562e-02 6.22687578e-01 1.39959395e+00
-7.29095712e-02 1.57551634e+00 7.76679873e-01 -2.21699461e-01
3.24336946e-01 -5.97155392e-01 -3.23180288e-01 7.67668664e-01
-3.22456717e-01 6.92255571e-02 -8.73678625e-01 6.14483282e-02
9.80581343e-01 1.70617044e-01 -6.20464534e-02 -2.31765121e-01
-1.58880866e+00 3.95777613e-01 -1.04775969e-02 1.19032018e-01
-4.49809343e-01 6.78692698e-01 7.68206656e-01 1.74105272e-01
5.52065611e-01 3.26281115e-02 -4.97285426e-01 -8.99517953e-01
-8.65018666e-01 -6.44494891e-02 6.00165665e-01 1.08216381e+00
3.31777632e-01 1.08540781e-01 -5.35764873e-01 5.09070218e-01
5.25691696e-02 2.93341190e-01 5.13954461e-01 -1.54610372e+00
5.90014637e-01 6.31490111e-01 3.05223972e-01 -4.33910072e-01
-1.59315482e-01 1.20450020e-01 -3.16891640e-01 1.11626089e-01
5.31067729e-01 -1.12088650e-01 -9.45352137e-01 1.53993392e+00
4.36945260e-01 6.99257195e-01 3.86227556e-02 9.61288810e-01
4.96164203e-01 5.85111558e-01 3.16252738e-01 -2.74826169e-01
1.18617642e+00 -1.45125675e+00 -6.80347204e-01 -5.03637791e-01
9.53848362e-01 -4.05223221e-01 9.79027271e-01 3.93816411e-01
-1.05438757e+00 -5.15487731e-01 -6.96651876e-01 4.58205156e-02
2.14176357e-01 5.30002356e-01 7.12998211e-01 1.99263960e-01
-9.04214561e-01 5.51117539e-01 -1.48166144e+00 -3.72842491e-01
8.15768778e-01 6.87583312e-02 -3.36554974e-01 -5.54111861e-02
-7.97091067e-01 4.75673825e-01 4.05411690e-01 -8.19028392e-02
-1.66104114e+00 -2.11413771e-01 -9.98480380e-01 -1.96158886e-01
1.02207220e+00 -1.87188432e-01 1.87570238e+00 -1.43184364e+00
-1.69288898e+00 8.59436095e-01 -2.50553936e-01 -8.88210535e-01
7.68643498e-01 -5.07116437e-01 -4.13948327e-01 7.89577007e-01
2.77152807e-01 8.16286385e-01 1.00348377e+00 -4.84208226e-01
-1.05037153e+00 1.19964048e-01 3.11158955e-01 3.10268819e-01
-1.90223739e-01 3.25749785e-01 -7.94533610e-01 -6.95822239e-01
-6.08878359e-02 -1.02071512e+00 -6.52758703e-02 2.42635295e-01
-1.58636034e-01 -6.41705155e-01 9.32531655e-01 -5.84571242e-01
1.27473080e+00 -2.36316991e+00 -2.16770813e-01 -3.69080842e-01
1.07639633e-01 2.49154374e-01 -2.07461506e-01 3.85231405e-01
-8.95569250e-02 -2.77720541e-01 1.63827077e-01 -2.74247378e-01
-3.65062468e-02 3.17671627e-01 -2.85367578e-01 7.06807196e-01
3.75647098e-01 9.98179495e-01 -1.33923495e+00 -7.69012570e-01
2.98346072e-01 -2.91226897e-02 -5.33514559e-01 4.71175969e-01
-5.92410266e-01 4.78091776e-01 -5.93857646e-01 1.09683943e+00
-1.99050799e-01 -5.68451762e-01 2.28480428e-01 2.00160127e-02
5.75934164e-02 4.66913909e-01 -1.02919042e+00 1.79212439e+00
-2.13398933e-01 7.23740876e-01 -1.58084005e-01 -1.01368940e+00
3.25667828e-01 6.74334764e-01 1.03258014e+00 -5.91735899e-01
-9.40698832e-02 -6.61534071e-03 -1.44542277e-01 -6.81959093e-01
5.84153049e-02 3.39350730e-01 -6.51482120e-02 6.89000666e-01
3.26115280e-01 5.54830611e-01 5.45659125e-01 4.60404068e-01
1.83870053e+00 7.77659237e-01 4.55749631e-01 3.37647647e-01
4.00348246e-01 3.29670198e-02 8.93268764e-01 7.29777813e-01
-6.77410364e-01 5.47727287e-01 6.26298726e-01 -5.50565481e-01
-7.35474765e-01 -8.89087260e-01 4.42325771e-01 1.81841993e+00
8.44553187e-02 -6.64712489e-01 -6.68489754e-01 -1.19597888e+00
-1.38492122e-01 2.41359562e-01 -5.38612843e-01 -3.56702581e-02
-7.61128366e-01 -4.13014181e-02 5.57913125e-01 1.01800871e+00
6.66165948e-01 -1.39695573e+00 -7.37753749e-01 3.57024729e-01
-4.20635432e-01 -1.77185404e+00 -9.53630984e-01 -7.59190088e-03
-9.12399650e-01 -1.36555135e+00 -5.23257077e-01 -5.87021232e-01
5.33976912e-01 4.73629266e-01 1.08109105e+00 -1.34904701e-02
-1.93372473e-01 7.42396533e-01 -7.51834869e-01 -1.09996103e-01
-3.77950311e-01 -2.40741447e-01 9.33360457e-02 4.36184049e-01
5.54431200e-01 -3.64480406e-01 -8.25288475e-01 7.59080589e-01
-6.59737349e-01 1.55538574e-01 6.81085885e-01 4.58547682e-01
6.34928286e-01 -1.98057026e-01 5.44416964e-01 -6.19476676e-01
-2.21385837e-01 -3.00851226e-01 -4.23504174e-01 9.78963897e-02
-2.53763527e-01 -1.54081404e-01 4.91830230e-01 -7.24616647e-01
-9.70477581e-01 6.18417680e-01 2.00352862e-01 -7.13630795e-01
-3.33335787e-01 -7.01633021e-02 -1.13067217e-01 2.73333728e-01
5.99047899e-01 3.59585196e-01 -1.74099281e-01 -3.52797389e-01
2.54400223e-01 5.72876811e-01 6.83185816e-01 -3.07969213e-01
4.89724129e-01 8.03408861e-01 -3.80973995e-01 -5.63057721e-01
-1.30106091e+00 -9.79637980e-01 -7.21292436e-01 -6.65568769e-01
1.11289310e+00 -1.46503305e+00 -5.94898283e-01 6.55831873e-01
-8.95232916e-01 -9.44685519e-01 -5.39082706e-01 6.52008355e-01
-1.05055094e+00 3.36999387e-01 -8.99190843e-01 -7.53902018e-01
-5.66498935e-02 -8.47956479e-01 1.41594696e+00 -1.67649657e-01
-3.37832123e-01 -7.38552690e-01 -5.52661419e-02 7.01298594e-01
-1.42694533e-01 1.52169257e-01 -8.97720829e-02 -6.03468657e-01
-8.37022483e-01 -2.75981039e-01 6.33338019e-02 6.07783258e-01
1.58277556e-01 -1.61130324e-01 -9.14643645e-01 -1.84477866e-01
-3.16934437e-01 -9.04493630e-01 9.39525783e-01 3.27317119e-01
1.32517862e+00 -4.29850549e-01 -1.88451007e-01 3.24176937e-01
7.47177064e-01 1.14760466e-01 7.05246210e-01 1.47006780e-01
5.99382758e-01 1.40079990e-01 1.21968913e+00 6.92407310e-01
2.61217982e-01 8.29375088e-01 3.74332339e-01 -4.95247133e-02
-3.43523979e-01 -5.74954271e-01 1.18633974e+00 3.78508180e-01
-4.24950182e-01 -1.19060241e-01 -6.00738049e-01 5.17884910e-01
-2.31779766e+00 -1.39442241e+00 1.66235000e-01 1.97012234e+00
9.74117994e-01 3.19745600e-01 6.48544788e-01 -2.47194152e-02
6.06130898e-01 5.61037540e-01 -8.04456890e-01 9.96527374e-02
1.88428625e-01 -1.48716554e-01 5.79236805e-01 6.13151491e-02
-1.67149639e+00 1.13964844e+00 6.14326525e+00 8.11328948e-01
-6.98869586e-01 3.92592877e-01 6.23946309e-01 -4.78069425e-01
6.18944705e-01 1.30211085e-01 -8.54892015e-01 5.67578673e-01
9.70632374e-01 2.80996621e-01 2.52186328e-01 1.04586387e+00
7.43211031e-01 -4.02882248e-01 -1.49086809e+00 1.05673361e+00
1.84438437e-01 -1.21180582e+00 -2.62589335e-01 -3.80776860e-02
7.47311413e-01 2.93043643e-01 -4.27360952e-01 5.44251978e-01
5.08953512e-01 -7.20141590e-01 8.68769467e-01 2.46794835e-01
7.34349668e-01 -2.01400235e-01 3.69859725e-01 5.30071259e-01
-1.46965384e+00 -3.57205600e-01 9.91846621e-02 -3.07303071e-01
4.53053892e-01 2.98499823e-01 -7.41023660e-01 -4.44819145e-02
5.67583084e-01 1.46992123e+00 -4.23082471e-01 7.35942125e-01
-7.00605750e-01 1.09602952e+00 -1.30487561e-01 2.92878330e-01
3.60933363e-01 1.50024533e-01 3.27346176e-01 1.12939060e+00
-6.52222410e-02 4.40077484e-01 9.22025621e-01 2.30264559e-01
-1.94765672e-01 -2.18781099e-01 -4.59845126e-01 -4.52053070e-01
1.69299543e-01 1.12906909e+00 -8.31120670e-01 -6.98169410e-01
-7.20672011e-01 1.21384919e+00 4.84873243e-02 2.63439059e-01
-1.09005201e+00 2.21323937e-01 5.01776576e-01 3.08149755e-01
4.37418699e-01 -3.07961196e-01 3.87101680e-01 -1.44525909e+00
1.40783191e-01 -1.03852212e+00 8.16894054e-01 -7.87883162e-01
-9.47137296e-01 3.37071307e-02 -8.96718130e-02 -1.81355357e+00
-4.17876095e-01 -4.55674320e-01 -6.56442523e-01 -1.04639187e-01
-1.19722402e+00 -1.15108073e+00 -2.93179750e-01 7.63485610e-01
1.17876458e+00 -7.10232556e-03 4.11095232e-01 2.73735762e-01
-6.96671844e-01 3.06289434e-01 -5.19866228e-01 6.60935819e-01
8.57596278e-01 -1.13247263e+00 3.21356148e-01 1.04497659e+00
5.44101477e-01 -1.85534388e-01 3.10673326e-01 -7.26695299e-01
-1.55444574e+00 -1.34560394e+00 6.08378947e-01 -6.03486359e-01
9.42245483e-01 -3.96208614e-01 -4.68225211e-01 1.01679492e+00
-8.21961388e-02 5.69847405e-01 5.11712015e-01 -2.28623793e-01
-3.42461199e-01 -8.66445079e-02 -5.04341066e-01 5.23128867e-01
1.61369336e+00 -6.73174679e-01 -4.89676267e-01 1.00258327e+00
6.08474910e-01 -4.43679243e-01 -6.81027293e-01 2.01940402e-01
4.55440223e-01 -8.48068118e-01 8.25933635e-01 -8.17296684e-01
4.78160918e-01 -3.79572332e-01 1.45761967e-01 -5.65980732e-01
-8.86995494e-02 -9.43332195e-01 -7.91445851e-01 8.20491374e-01
1.69198990e-01 -2.19407663e-01 9.58387077e-01 4.04658288e-01
-1.17059618e-01 -6.29087567e-01 -8.76178503e-01 -1.11401916e+00
-8.55760872e-01 -6.19934201e-01 -8.67291763e-02 6.11998200e-01
1.89404085e-01 2.60931820e-01 -6.38909638e-01 -8.66527495e-04
4.17287022e-01 4.71782759e-02 9.75511372e-01 -5.21114588e-01
-6.80899322e-01 -9.97985378e-02 -6.94449961e-01 -1.64450002e+00
1.78850695e-01 -5.35168052e-01 3.63884807e-01 -1.20441663e+00
2.63355076e-01 6.04210086e-02 -3.72353703e-01 1.03325558e+00
3.77177559e-02 4.26358074e-01 1.37274623e-01 4.03726161e-01
-1.67827833e+00 4.79149401e-01 1.03917098e+00 5.41362399e-03
-2.33920470e-01 1.16530605e-01 -1.69367000e-01 1.09986782e+00
6.02460623e-01 -4.34721500e-01 -4.26133364e-01 -2.43771434e-01
-2.37011865e-01 2.00447917e-01 5.62511325e-01 -1.21205938e+00
2.45653003e-01 -4.60436702e-01 1.22366227e-01 -4.92583841e-01
2.71555394e-01 -5.31578183e-01 -3.68615985e-01 3.76558721e-01
-5.41301906e-01 -2.22937927e-01 -3.23607773e-01 9.02225614e-01
-2.46900946e-01 1.50241107e-01 6.68132782e-01 -2.59658188e-01
-1.24800336e+00 5.73790789e-01 -6.74632907e-01 4.02216345e-01
1.33071756e+00 -3.00763458e-01 -1.57214835e-01 -6.37735963e-01
-8.35941732e-01 3.77090812e-01 2.39065111e-01 5.12626290e-01
5.20090818e-01 -1.32476270e+00 -6.09622836e-01 -2.41043270e-01
4.41692561e-01 -2.32268959e-01 1.53346762e-01 1.18264711e+00
-2.09581628e-01 2.20350146e-01 1.86742872e-01 -7.79451132e-01
-1.34983516e+00 4.59136069e-01 1.06508628e-01 -3.56944919e-01
-9.05873954e-01 6.27081335e-01 1.75680608e-01 2.46342704e-01
6.22733533e-01 -3.26839447e-01 -3.03392317e-02 -9.18967351e-02
8.32604885e-01 3.98012966e-01 -3.12208593e-01 -6.20369792e-01
-3.34507525e-01 3.51526290e-02 4.84974347e-02 -1.45001844e-01
1.24144423e+00 5.00797220e-02 3.09448391e-01 5.18913209e-01
9.38203871e-01 -3.99307221e-01 -2.08228564e+00 -3.41224909e-01
1.56227499e-01 -6.13713443e-01 -1.11182697e-01 -6.32747412e-01
-9.86453950e-01 6.27249002e-01 3.45413774e-01 -2.23116413e-01
1.23735118e+00 3.29223126e-01 9.98680055e-01 4.85725135e-01
6.02376521e-01 -1.53202581e+00 8.28102469e-01 4.33904320e-01
5.40201128e-01 -1.42229521e+00 2.20386740e-02 -3.13623190e-01
-8.98037434e-01 9.23125327e-01 8.12363386e-01 -1.58030570e-01
3.10069859e-01 1.62422955e-01 -3.98396701e-02 -4.43571918e-02
-1.19330919e+00 -4.63550448e-01 2.03803837e-01 2.87603408e-01
2.01658070e-01 -5.10593466e-02 -6.55054152e-02 3.25962573e-01
6.99869096e-01 4.68338639e-01 4.57565308e-01 1.40038204e+00
-5.11477530e-01 -9.69138861e-01 -3.28839161e-02 4.68748212e-01
-5.39211810e-01 1.11218162e-01 -3.97623658e-01 5.59116542e-01
2.31110658e-02 1.02926433e+00 7.05357641e-02 -3.01523894e-01
1.40705451e-01 7.40372464e-02 3.65030736e-01 -8.68960202e-01
-2.58033574e-01 2.97826409e-01 5.00187278e-01 -1.47692108e+00
-9.62801039e-01 -8.79408300e-01 -1.31186450e+00 1.75734580e-01
-1.34999245e-01 -1.96201518e-01 -8.59200656e-02 1.07008410e+00
4.56035256e-01 2.52158195e-01 7.23402381e-01 -8.62806976e-01
-5.79113841e-01 -9.05270159e-01 -4.75743890e-01 6.44313574e-01
3.14479530e-01 -6.38075948e-01 -3.04806143e-01 7.64427066e-01] | [8.461669921875, 0.6005592346191406] |
1e84d047-b691-440a-95e9-312809a7cf9d | fast-diffusion-sampler-for-inverse-problems | 2303.05754 | null | https://arxiv.org/abs/2303.05754v1 | https://arxiv.org/pdf/2303.05754v1.pdf | Fast Diffusion Sampler for Inverse Problems by Geometric Decomposition | Diffusion models have shown exceptional performance in solving inverse problems. However, one major limitation is the slow inference time. While faster diffusion samplers have been developed for unconditional sampling, there has been limited research on conditional sampling in the context of inverse problems. In this study, we propose a novel and efficient diffusion sampling strategy that employs the geometric decomposition of diffusion sampling. Specifically, we discover that the samples generated from diffusion models can be decomposed into two orthogonal components: a ``denoised" component obtained by projecting the sample onto the clean data manifold, and a ``noise" component that induces a transition to the next lower-level noisy manifold with the addition of stochastic noise. Furthermore, we prove that, under some conditions on the clean data manifold, the conjugate gradient update for imposing conditioning from the denoised signal belongs to the clean manifold, resulting in a much faster and more accurate diffusion sampling. Our method is applicable regardless of the parameterization and setting (i.e., VE, VP). Notably, we achieve state-of-the-art reconstruction quality on challenging real-world medical inverse imaging problems, including multi-coil MRI reconstruction and 3D CT reconstruction. Moreover, our proposed method achieves more than 80 times faster inference time than the previous state-of-the-art method. | ['Jong Chul Ye', 'Suhyeon Lee', 'Hyungjin Chung'] | 2023-03-10 | null | null | null | null | ['mri-reconstruction'] | ['computer-vision'] | [ 4.70092624e-01 7.49145225e-02 7.68104121e-02 -5.84100857e-02
-1.03934932e+00 -5.84981069e-02 5.24216890e-01 -2.13158533e-01
-2.85376281e-01 7.87141502e-01 4.66995716e-01 -1.02196828e-01
-3.08522284e-01 -6.35489762e-01 -6.67713583e-01 -1.23671830e+00
8.56691226e-02 5.94650686e-01 -1.69610307e-02 1.98981091e-01
-5.77764288e-02 1.57730147e-01 -8.40913832e-01 -1.91777661e-01
1.19781470e+00 6.11336708e-01 3.99623573e-01 4.57897842e-01
2.03228191e-01 5.65624535e-01 -2.71485478e-01 5.84334321e-02
1.03973649e-01 -9.04114127e-01 -7.02444971e-01 2.35162199e-01
-9.96259674e-02 -3.99977654e-01 -3.66256297e-01 1.39353490e+00
7.09520161e-01 3.10990214e-01 8.35018873e-01 -5.75711489e-01
-6.49566293e-01 5.43846548e-01 -8.56209457e-01 1.90827265e-01
8.07823613e-02 -1.24719562e-02 4.85789001e-01 -9.25988734e-01
8.57189059e-01 9.45980072e-01 5.05095363e-01 6.51925027e-01
-1.72707367e+00 -4.78167474e-01 -4.15108278e-02 4.40192856e-02
-1.17445087e+00 -5.41606426e-01 8.68978024e-01 -5.11043727e-01
3.33654553e-01 9.81519967e-02 5.35192847e-01 1.26533961e+00
3.06904763e-01 7.95461595e-01 1.56537843e+00 -2.86749512e-01
6.57848775e-01 6.23887405e-02 6.27794787e-02 5.35641849e-01
1.62745759e-01 -3.04404330e-02 -3.71305317e-01 -3.83431882e-01
9.38748538e-01 -1.07114062e-01 -6.58787608e-01 -4.48355973e-01
-1.35750830e+00 9.13183689e-01 2.64412165e-01 3.92802119e-01
-9.62324202e-01 -5.81907257e-02 6.62444532e-02 -2.73046624e-02
8.04685175e-01 1.45276971e-02 1.81177016e-02 8.39063451e-02
-1.11341202e+00 9.78940725e-02 7.19773591e-01 6.29093766e-01
5.19058347e-01 -3.63280973e-03 -1.41479924e-01 8.50686550e-01
5.18331707e-01 8.52055073e-01 3.07751238e-01 -9.27305162e-01
3.54811162e-01 -3.22534442e-01 9.00725052e-02 -8.49601686e-01
-2.57541984e-01 -7.65030742e-01 -1.46376622e+00 -1.62228514e-02
4.91131842e-01 -2.68230289e-01 -7.30633438e-01 2.02285314e+00
6.96464300e-01 5.67962766e-01 -4.83266301e-02 1.06204498e+00
2.57226944e-01 6.06499255e-01 -1.35095239e-01 -5.97811520e-01
1.25898266e+00 -5.65287232e-01 -1.12702680e+00 -1.50555894e-01
3.00615102e-01 -7.48262346e-01 5.24168372e-01 5.92812777e-01
-1.33902323e+00 -2.42253661e-01 -1.00828493e+00 9.04992148e-02
4.85024005e-01 -1.36626303e-01 3.61946523e-01 5.92633307e-01
-8.81719530e-01 7.42714345e-01 -1.13199162e+00 -8.03670064e-02
4.23605442e-01 -1.22159712e-01 -1.35911614e-01 -5.52438080e-01
-1.03564727e+00 8.42566431e-01 -3.57032865e-01 2.03129813e-01
-1.28159630e+00 -8.48988295e-01 -5.35141051e-01 -3.98940772e-01
2.75336534e-01 -9.34900761e-01 8.43234241e-01 -6.20119452e-01
-1.61115634e+00 4.67813492e-01 -4.75407660e-01 -3.46820980e-01
8.45048845e-01 -1.05102174e-01 -2.34609798e-01 3.90650988e-01
4.32151973e-01 1.57527894e-01 1.34574640e+00 -1.27313888e+00
4.50668186e-02 -7.38156378e-01 -4.63590026e-01 1.94283485e-01
-7.05592707e-02 -3.66344601e-01 -3.48453075e-01 -8.10650349e-01
4.75481510e-01 -1.11395943e+00 -5.46635151e-01 1.34465313e-02
-3.96583796e-01 9.03745741e-02 3.30115110e-01 -8.84199023e-01
1.09826100e+00 -2.32253528e+00 7.09485471e-01 3.50710899e-01
4.94525015e-01 -1.37661904e-01 1.66533992e-01 1.48582637e-01
-1.46643311e-01 -2.04416871e-01 -8.60806406e-01 -5.14670074e-01
-1.96413428e-01 8.20169449e-02 -2.27056116e-01 1.04259276e+00
-1.99279472e-01 6.17785156e-01 -1.17204547e+00 -3.71344417e-01
1.20247282e-01 8.04081678e-01 -5.95243216e-01 4.82775345e-02
2.40467131e-01 1.15378082e+00 -6.44822657e-01 3.22397985e-02
9.04035389e-01 -3.40274572e-01 3.09896737e-01 -2.38663509e-01
1.30595550e-01 6.32912517e-02 -1.33448780e+00 2.00862503e+00
-5.08184910e-01 2.93935925e-01 6.46068931e-01 -1.07299531e+00
5.56007266e-01 5.50155997e-01 5.49632668e-01 -4.31366473e-01
-6.79978682e-03 3.64772528e-01 -6.30308017e-02 -4.15108263e-01
1.22038208e-01 -7.34354258e-01 3.80791456e-01 7.22205579e-01
-1.04658075e-01 -2.14690700e-01 -5.63542992e-02 2.81905770e-01
9.64757442e-01 8.23074281e-02 8.65748003e-02 -6.42594874e-01
3.27808857e-01 -4.41449851e-01 4.63765085e-01 9.12032008e-01
-1.94221899e-01 6.50479555e-01 1.91998601e-01 2.01680213e-01
-8.62045288e-01 -1.30009615e+00 -5.62462509e-01 4.51327622e-01
6.98394254e-02 -9.98041127e-03 -1.15225160e+00 -3.56503367e-01
-3.66984636e-01 8.14291596e-01 -6.90786362e-01 -1.55400997e-02
-6.64873123e-01 -1.27891743e+00 2.18019024e-01 1.68623731e-01
4.69398201e-01 -7.02305973e-01 -2.31309295e-01 4.48895007e-01
-6.03742480e-01 -8.59783828e-01 -7.32259631e-01 -6.48001209e-02
-1.33439350e+00 -8.16026628e-01 -1.15586317e+00 -3.87154400e-01
9.22016144e-01 2.96058774e-01 7.55032718e-01 -3.32463950e-01
-6.48606271e-02 3.52518886e-01 2.45675221e-02 7.44360983e-02
-3.84719253e-01 -1.59622952e-01 4.86682951e-02 4.97291863e-01
-1.55395314e-01 -7.18924105e-01 -9.53024805e-01 1.89534009e-01
-1.02926219e+00 1.38099879e-01 5.29160082e-01 9.41518784e-01
7.86446929e-01 1.40746251e-01 5.78003943e-01 -8.63512337e-01
7.64953673e-01 -8.20641994e-01 -1.43490553e-01 -2.20216423e-01
-5.58813214e-01 4.84695822e-01 4.31359857e-01 -5.40874898e-01
-1.28491068e+00 -1.07126892e-01 -3.65960747e-01 -3.26297730e-01
7.62043223e-02 4.92678940e-01 8.79454985e-02 1.78440794e-01
7.19342411e-01 4.35422301e-01 2.56124705e-01 -6.23372555e-01
5.10783374e-01 4.19835269e-01 2.89894849e-01 -7.41367280e-01
7.03252316e-01 1.16233063e+00 1.47452146e-01 -1.17609763e+00
-7.69539297e-01 -1.88856974e-01 -5.04493833e-01 -2.17223376e-01
1.07115710e+00 -7.05455005e-01 -2.49804720e-01 5.68933129e-01
-8.65998685e-01 -2.68513858e-01 -4.67167348e-01 8.09637964e-01
-6.23205066e-01 7.00114310e-01 -7.74759531e-01 -8.34153175e-01
-3.55654359e-01 -1.39952910e+00 8.24259818e-01 -2.10643217e-01
-2.30919898e-01 -1.14460230e+00 2.42007867e-01 2.68950701e-01
5.11161506e-01 2.09663123e-01 8.03053021e-01 1.04754209e-03
-5.46606541e-01 -4.92138267e-02 7.30778500e-02 4.31200922e-01
7.53600076e-02 -5.43237746e-01 -7.23918974e-01 -4.50123906e-01
9.22565222e-01 8.55836496e-02 8.72689426e-01 8.68510127e-01
7.94963181e-01 -1.51602104e-01 -4.62315619e-01 6.27208471e-01
1.22755754e+00 -7.06461370e-02 6.53432310e-01 -1.55289859e-01
4.71776932e-01 4.83848512e-01 1.65883824e-01 3.13014030e-01
2.36322820e-01 4.71541166e-01 -2.27592885e-02 -3.55380848e-02
-3.29516411e-01 -1.39124081e-01 1.66632652e-01 1.40886807e+00
-4.10095192e-02 1.89427450e-01 -6.22502446e-01 7.60396898e-01
-1.63366020e+00 -8.23900342e-01 -4.67335224e-01 2.35904860e+00
1.01074219e+00 6.34671422e-03 -2.11440563e-01 1.59398824e-01
6.28031492e-01 2.04204798e-01 -6.68294847e-01 3.26646417e-01
-3.35455849e-03 3.93341482e-01 3.99911225e-01 8.10412824e-01
-7.77310789e-01 2.64078796e-01 6.19588327e+00 9.57900941e-01
-8.38986695e-01 6.53753519e-01 5.38230360e-01 -6.26602620e-02
-4.77576256e-01 -8.12565014e-02 -2.34595448e-01 5.28204620e-01
7.37184286e-01 -1.20814834e-02 6.52366519e-01 4.76910710e-01
3.82452786e-01 -2.95333803e-01 -8.28866839e-01 8.41547966e-01
8.93799867e-03 -1.04762375e+00 -1.55357152e-01 2.98199505e-01
9.41555321e-01 1.16128646e-01 1.09714568e-01 -2.76777089e-01
2.93054819e-01 -7.35462070e-01 6.61290526e-01 7.03802705e-01
7.46397972e-01 -5.28138578e-01 3.27960253e-01 6.06416881e-01
-8.28068674e-01 3.53361934e-01 -1.98264375e-01 1.95473656e-01
6.20269597e-01 1.32146835e+00 -3.59837383e-01 6.31063223e-01
4.96433407e-01 7.43608534e-01 1.50314450e-01 6.72595918e-01
-3.80530775e-01 7.89359748e-01 -3.58851612e-01 3.64203095e-01
2.46139243e-03 -8.22921038e-01 8.62749755e-01 9.31340158e-01
3.55951607e-01 2.83444643e-01 -1.45921648e-01 1.05486810e+00
8.96636844e-02 -1.62823588e-01 -4.21330571e-01 3.75887096e-01
2.39056274e-01 1.12810230e+00 -7.26946533e-01 -4.58341688e-01
-1.86605141e-01 1.15374541e+00 2.03948826e-01 6.56008363e-01
-7.59846687e-01 5.84346242e-03 5.87467194e-01 8.95155817e-02
3.87105674e-01 -3.07197660e-01 -4.80008602e-01 -1.33749938e+00
-4.55221906e-03 -8.26971829e-01 7.93592036e-02 -4.63506818e-01
-1.51451194e+00 4.61681008e-01 -5.77578135e-02 -8.55616450e-01
-9.63337868e-02 -1.16420723e-01 -3.42742682e-01 1.14340484e+00
-1.15008700e+00 -4.37570542e-01 -8.22038129e-02 5.84136903e-01
3.29678208e-01 4.63174611e-01 7.38919795e-01 4.96476561e-01
-4.46923882e-01 -2.84138080e-02 6.54519618e-01 -1.52507678e-01
4.84023035e-01 -1.13759935e+00 1.68179318e-01 8.64743233e-01
-1.94076940e-01 8.20910096e-01 8.15192163e-01 -7.89449275e-01
-1.57680869e+00 -8.77085805e-01 4.53602582e-01 -3.81230056e-01
6.90988302e-01 -3.07234019e-01 -1.03159881e+00 5.65903604e-01
1.32745355e-01 -1.29336081e-02 5.28679788e-01 -1.82149932e-01
4.23180796e-02 7.34421536e-02 -1.32067943e+00 5.48406124e-01
1.04276645e+00 -6.11097515e-01 -5.50642669e-01 4.10063595e-01
1.89594552e-01 -4.07382846e-01 -1.12704611e+00 2.50366896e-01
4.08921957e-01 -8.36967885e-01 1.09066963e+00 -1.41277894e-01
3.64848465e-01 -2.82972425e-01 -1.47603704e-02 -1.58929920e+00
-3.88913631e-01 -7.35547721e-01 -2.61967212e-01 8.62241924e-01
2.39219487e-01 -8.69756341e-01 5.13928711e-01 2.79375464e-01
6.83173835e-02 -7.70259202e-01 -1.19436860e+00 -6.02307975e-01
3.36854070e-01 -5.15655160e-01 9.53746065e-02 9.12944794e-01
-2.13767543e-01 5.01047969e-01 -5.76351702e-01 1.05627604e-01
1.39157355e+00 -5.46343550e-02 3.97677213e-01 -8.84468198e-01
-6.00398719e-01 -2.15674654e-01 2.85282582e-01 -1.55340362e+00
-1.07630707e-01 -9.96459961e-01 2.68012375e-01 -1.65462267e+00
3.77311945e-01 -5.24294555e-01 -1.27940536e-01 -4.16759342e-01
-3.75331610e-01 1.65213913e-01 -1.44250765e-02 4.82109725e-01
-8.01087990e-02 7.82608032e-01 1.67037964e+00 -1.41746579e-02
1.47517277e-02 6.06304258e-02 -5.86793244e-01 6.13440692e-01
4.86731887e-01 -7.24314928e-01 -4.28674787e-01 -5.13108015e-01
6.77111931e-03 4.17194098e-01 1.70453370e-01 -7.78861165e-01
2.20111251e-01 1.40825406e-01 2.20861211e-01 -3.63260925e-01
4.01829720e-01 -5.67587495e-01 4.70783293e-01 5.71372032e-01
-1.90055162e-01 -4.22547549e-01 -2.18830481e-01 9.44097638e-01
1.54415071e-01 -4.30535734e-01 1.05800784e+00 -1.96455643e-01
3.17664415e-01 3.34491223e-01 -6.00465119e-01 2.78987914e-01
6.71146452e-01 1.86798766e-01 1.25307620e-01 -4.27279383e-01
-9.60249662e-01 -1.76835448e-01 3.17038000e-01 -3.99310216e-02
5.87153018e-01 -1.33571827e+00 -9.49417949e-01 2.32187599e-01
-5.24532139e-01 1.20607316e-01 6.35733783e-01 1.49832177e+00
-1.63523138e-01 9.92668793e-02 2.88070381e-01 -8.07641864e-01
-6.46958888e-01 5.49183249e-01 2.98126042e-01 -4.15417165e-01
-1.04289651e+00 6.17674828e-01 4.92028445e-01 -3.39558154e-01
-1.08390667e-01 -2.74605185e-01 2.55477965e-01 -2.95688640e-02
6.13393307e-01 7.23151863e-01 -1.25784695e-01 -6.22195184e-01
-2.68037081e-01 6.69560552e-01 1.37093440e-01 -5.90840816e-01
1.38599849e+00 -2.71273911e-01 -2.88147330e-01 4.61603016e-01
1.22111833e+00 1.39395043e-01 -1.47625947e+00 -4.65254754e-01
-2.14553133e-01 -3.76345724e-01 3.99545670e-01 -4.43918824e-01
-1.09000516e+00 8.40946555e-01 5.75594723e-01 3.24274689e-01
1.01467776e+00 5.23151979e-02 7.71231353e-01 -1.68099940e-01
4.83627468e-01 -9.46832061e-01 -6.89973086e-02 1.18287891e-01
9.66800392e-01 -9.02942657e-01 1.62495211e-01 -3.87895703e-01
-4.02522773e-01 5.23057222e-01 -2.57360458e-01 -3.07683617e-01
9.34383690e-01 9.14689675e-02 -1.76436320e-01 -2.95692652e-01
-3.59494328e-01 1.93169713e-01 1.76186398e-01 6.26225591e-01
2.81367302e-01 2.17598364e-01 -3.99114192e-01 3.49184275e-01
1.00803666e-01 8.74778405e-02 4.45394158e-01 7.21801877e-01
-1.80873871e-01 -7.96548843e-01 -5.69399893e-01 4.01903600e-01
-5.40859640e-01 -2.38430977e-01 1.93669006e-01 4.27518308e-01
-2.81380892e-01 9.39707041e-01 -2.38558620e-01 2.91579098e-01
2.08554938e-01 -3.95312756e-02 7.72800982e-01 -3.82369936e-01
-3.93937379e-02 3.63699108e-01 -1.90828875e-01 -4.79656607e-01
-5.90120435e-01 -1.09900272e+00 -1.04485703e+00 -2.63386786e-01
-4.87015009e-01 2.51986206e-01 7.41827846e-01 1.07297206e+00
3.40473741e-01 6.14573956e-01 5.19316971e-01 -9.37284648e-01
-7.53467739e-01 -1.19391906e+00 -7.70681441e-01 5.33074796e-01
4.35711414e-01 -5.89456916e-01 -5.87133169e-01 -9.41793099e-02] | [11.916077613830566, -2.3972079753875732] |
96a1e929-0888-42b1-a624-71e7a23a75cc | skill-based-multi-objective-reinforcement | 2203.10033 | null | https://arxiv.org/abs/2203.10033v1 | https://arxiv.org/pdf/2203.10033v1.pdf | Skill-based Multi-objective Reinforcement Learning of Industrial Robot Tasks with Planning and Knowledge Integration | In modern industrial settings with small batch sizes it should be easy to set up a robot system for a new task. Strategies exist, e.g. the use of skills, but when it comes to handling forces and torques, these systems often fall short. We introduce an approach that provides a combination of task-level planning with targeted learning of scenario-specific parameters for skill-based systems. We propose the following pipeline: (1) the user provides a task goal in the planning language PDDL, (2) a plan (i.e., a sequence of skills) is generated and the learnable parameters of the skills are automatically identified. An operator then chooses (3) reward functions and hyperparameters for the learning process. Two aspects of our methodology are critical: (a) learning is tightly integrated with a knowledge framework to support symbolic planning and to provide priors for learning, (b) using multi-objective optimization. This can help to balance key performance indicators (KPIs) such as safety and task performance since they can often affect each other. We adopt a multi-objective Bayesian optimization approach and learn entirely in simulation. We demonstrate the efficacy and versatility of our approach by learning skill parameters for two different contact-rich tasks. We show their successful execution on a real 7-DOF KUKA-iiwa manipulator and outperform the manual parameterization by human robot operators. | ['Volker Krueger', 'Luigi Nardi', 'Konstantinos Chatzilygeroudis', 'Faseeh Ahmad', 'Matthias Mayr'] | 2022-03-18 | null | null | null | null | ['multi-objective-reinforcement-learning'] | ['methodology'] | [ 2.83462971e-01 2.88813174e-01 -1.23546734e-01 -1.03542283e-01
-7.96339691e-01 -6.42804205e-01 4.08468485e-01 7.09075928e-02
-4.17408317e-01 9.19867158e-01 -1.98357552e-01 -4.18671548e-01
-7.25704730e-01 -4.60495472e-01 -8.43123913e-01 -7.09780991e-01
-3.36040229e-01 1.12738681e+00 4.32835132e-01 -3.70968461e-01
4.97974008e-01 6.38005376e-01 -1.57238662e+00 -2.54045218e-01
7.82215655e-01 7.43774951e-01 8.85144591e-01 8.20569515e-01
4.00478750e-01 6.57881141e-01 -5.03974497e-01 3.03947330e-01
5.27799964e-01 -7.55211115e-02 -9.77497220e-01 1.62628829e-01
-3.50405663e-01 -2.81953335e-01 2.78633565e-01 7.51473904e-01
4.05716091e-01 4.31441218e-01 6.07905149e-01 -1.38869965e+00
4.48339462e-01 7.14492798e-01 -2.44686022e-01 -2.62852043e-01
3.79691213e-01 7.55626321e-01 6.36587977e-01 -1.96693510e-01
4.84784931e-01 1.28470731e+00 2.97560930e-01 1.18890136e-01
-1.25479758e+00 -3.32873940e-01 6.32919744e-02 1.27496151e-02
-1.39080286e+00 -1.54780746e-01 4.99191076e-01 -7.02524543e-01
1.07078743e+00 -2.77142823e-01 5.67663074e-01 9.62485313e-01
6.52051151e-01 4.17087406e-01 9.34462547e-01 -5.93130469e-01
4.52989101e-01 9.40368101e-02 -3.01923782e-01 6.90615654e-01
-1.14519158e-02 5.14846802e-01 -3.57010096e-01 -5.76076247e-02
8.80676508e-01 -1.71816006e-01 1.41730547e-01 -6.67774498e-01
-1.35348034e+00 6.52775884e-01 1.04341954e-01 1.31676301e-01
-6.69702649e-01 4.68144000e-01 3.99424791e-01 3.89598161e-01
-3.40494841e-01 1.02370620e+00 -8.00173283e-01 -3.15317482e-01
-4.87213910e-01 7.71146953e-01 1.04277217e+00 1.24868524e+00
8.41595173e-01 -8.11576694e-02 -2.32276365e-01 5.84732115e-01
2.82128960e-01 2.46187508e-01 -3.28480353e-04 -1.40908182e+00
4.82558012e-01 3.21938574e-01 6.76044524e-01 -5.56106567e-01
-7.42962360e-01 -1.14163741e-01 -2.95959003e-02 8.63994777e-01
5.07586122e-01 -4.31525409e-01 -9.11995769e-01 1.50015306e+00
4.12410855e-01 -2.80044787e-02 1.39604032e-01 1.00775278e+00
-1.24177009e-01 5.23694038e-01 1.52797282e-01 -1.82801902e-01
1.31979549e+00 -6.89264297e-01 -6.79475248e-01 -5.64120889e-01
5.24007440e-01 -6.93456054e-01 1.15999293e+00 7.68969297e-01
-1.23523140e+00 -6.59597456e-01 -1.08344412e+00 4.56370205e-01
-1.22094512e-01 9.16310027e-02 6.28023446e-01 3.91789019e-01
-7.06405938e-01 1.09151351e+00 -1.11856866e+00 -1.76371768e-01
-1.63103063e-02 8.23756337e-01 2.49398202e-02 1.44440532e-01
-1.02164555e+00 1.50863934e+00 9.80489433e-01 -2.20487323e-02
-1.52011311e+00 -3.83226514e-01 -8.58671665e-01 -2.92032752e-02
1.16275573e+00 -5.98469019e-01 1.66895843e+00 -7.11304009e-01
-2.15949702e+00 1.43149778e-01 5.45940161e-01 -3.03815842e-01
3.82077962e-01 -4.74577159e-01 1.62586108e-01 4.20797057e-02
4.63400744e-02 6.45154238e-01 9.55337107e-01 -1.43204629e+00
-7.84648061e-01 3.43402661e-02 4.65818137e-01 3.97082806e-01
1.94626585e-01 9.01664607e-03 -2.96114057e-01 -4.81615365e-02
-1.77329138e-01 -1.26115191e+00 -5.62864602e-01 -4.66308534e-01
-3.75490993e-01 -1.69462800e-01 3.27778906e-01 -5.78664064e-01
9.38723564e-01 -1.79588056e+00 6.83052540e-01 4.96764094e-01
-2.54959971e-01 -1.65641278e-01 3.64243649e-02 5.75929701e-01
7.66903609e-02 -3.77335966e-01 -1.23766080e-01 4.00815755e-02
2.72951990e-01 4.08569753e-01 -9.23148841e-02 2.64610946e-01
2.97156483e-01 5.50900817e-01 -1.15282714e+00 -5.62677085e-01
4.47719425e-01 -5.40108383e-02 -5.67711651e-01 4.30930436e-01
-6.54724598e-01 7.43437827e-01 -7.50722766e-01 4.70750660e-01
-3.43601033e-02 8.38124305e-02 5.12390137e-01 -5.75829372e-02
-2.74853170e-01 3.59873533e-01 -1.49145615e+00 1.68505311e+00
-8.08434963e-01 -2.15379193e-01 3.10339034e-01 -8.46902370e-01
8.55872869e-01 2.95374572e-01 4.99137670e-01 -5.54063506e-02
4.62056309e-01 2.24149302e-01 2.50956893e-01 -6.98784173e-01
4.18719202e-01 -1.43714830e-01 -4.00286376e-01 3.91586930e-01
3.13685149e-01 -1.09291470e+00 3.37302446e-01 -2.29852706e-01
1.21816337e+00 9.42521453e-01 6.01835072e-01 -3.38804513e-01
3.16097707e-01 4.07049745e-01 4.27271843e-01 6.30293489e-01
1.48820370e-01 1.14872575e-01 7.06959724e-01 -9.57994759e-02
-1.08875585e+00 -7.64453590e-01 2.06400543e-01 1.05351675e+00
7.82568827e-02 -2.52862155e-01 -5.84562778e-01 -5.65911889e-01
-1.10086994e-02 1.11313605e+00 -2.83605874e-01 -2.46694624e-01
-6.46174729e-01 -1.32187873e-01 4.49807607e-02 5.89044094e-01
-7.50243515e-02 -1.15589249e+00 -1.19040418e+00 5.13133168e-01
2.91591316e-01 -9.01869535e-01 -3.85402411e-01 8.19773734e-01
-9.20774519e-01 -1.09061515e+00 -2.38939613e-01 -6.27788544e-01
5.84699273e-01 -2.44813100e-01 1.06038284e+00 -1.04418918e-01
-1.23423032e-01 4.69153106e-01 -1.87148646e-01 -6.80893838e-01
-8.04621518e-01 1.21200234e-01 2.17414156e-01 -6.18109941e-01
-3.65005165e-01 -5.31626999e-01 -2.07126081e-01 5.85708916e-01
-5.11804640e-01 3.30219343e-02 1.00083005e+00 8.44090044e-01
6.55321836e-01 5.28600633e-01 4.24737930e-01 -6.93780243e-01
8.70271087e-01 -1.65048108e-01 -1.27028239e+00 2.46081978e-01
-6.89331710e-01 3.00843805e-01 4.79360193e-01 -7.64507949e-01
-1.10959840e+00 6.99640751e-01 1.01797909e-01 -4.59012836e-01
-2.83578664e-01 7.13800251e-01 -2.25296512e-01 8.47485438e-02
6.04601204e-01 -3.31851512e-01 2.48628721e-01 -9.45633948e-02
3.08348745e-01 4.13329929e-01 2.96950191e-01 -1.21974945e+00
8.49663019e-01 -2.52891093e-01 2.64860958e-01 -4.29009914e-01
-5.39185464e-01 -2.18610004e-01 -8.87590230e-01 -4.09088165e-01
6.72668874e-01 -6.58232331e-01 -1.16754305e+00 1.52431741e-01
-1.05523264e+00 -9.91999626e-01 -3.76863480e-01 5.79496324e-01
-1.31840348e+00 -6.87905401e-02 -2.15946525e-01 -1.00068641e+00
1.88744932e-01 -1.60617173e+00 1.12918699e+00 1.72495201e-01
-6.12342715e-01 -7.49589384e-01 -5.76560833e-02 1.88467324e-01
1.78418323e-01 2.66946346e-01 8.75704348e-01 -5.86126864e-01
-5.61486363e-01 -1.94561720e-01 3.35945904e-01 2.27676317e-01
4.12223674e-02 -2.31957704e-01 -5.89961708e-01 -2.69791543e-01
-5.72525449e-02 -6.08473778e-01 -1.48757368e-01 3.51521313e-01
9.26890135e-01 -1.46899998e-01 -5.43605924e-01 2.19059382e-02
1.11433220e+00 4.37552392e-01 4.48426455e-01 5.52984297e-01
4.97633845e-01 8.70423257e-01 1.36554825e+00 4.29689139e-01
1.40780479e-01 1.15394938e+00 6.57296956e-01 5.22151172e-01
3.83941203e-01 -5.56830689e-02 5.64030588e-01 2.86653668e-01
-3.45853597e-01 2.31096566e-01 -1.12952745e+00 3.44620854e-01
-2.06106687e+00 -6.32936239e-01 2.62493283e-01 2.31944633e+00
1.06970084e+00 6.00353360e-01 3.90339106e-01 2.82879751e-02
4.15401369e-01 -4.95609164e-01 -7.14776695e-01 -2.94481874e-01
9.27701235e-01 2.45740429e-01 7.38547504e-01 6.90397620e-01
-9.25290346e-01 1.02326763e+00 5.77718735e+00 5.89617908e-01
-5.90554416e-01 -6.40687123e-02 6.71652183e-02 -3.79604064e-02
2.24285945e-01 2.42390439e-01 -8.84912848e-01 1.67034745e-01
1.08081067e+00 -1.70086563e-01 7.81897068e-01 1.12768710e+00
4.19897527e-01 -5.46420753e-01 -1.35076714e+00 4.68870968e-01
-4.04083073e-01 -8.80296528e-01 -6.39767528e-01 -4.92761247e-02
4.19312000e-01 -3.63845259e-01 -4.76390064e-01 5.87355077e-01
7.18581617e-01 -1.02206659e+00 1.00278282e+00 5.29087663e-01
4.21766758e-01 -1.04838145e+00 7.70707488e-01 6.79299295e-01
-8.57889771e-01 -3.77151370e-01 -6.64582923e-02 -1.83891375e-02
5.14421165e-01 2.98374563e-01 -1.32580411e+00 6.82269812e-01
3.73987526e-01 2.64134705e-01 2.44812458e-04 9.67441559e-01
-5.65896749e-01 2.92367041e-01 -5.06975889e-01 -2.81389356e-01
1.53673440e-01 -2.27539346e-01 6.00508392e-01 7.88856328e-01
2.88949102e-01 -6.36768937e-02 7.34230518e-01 9.40613449e-01
8.10868979e-01 -3.34224761e-01 -5.22144794e-01 -1.03270628e-01
4.55291331e-01 1.21724641e+00 -9.37407911e-01 9.14390013e-02
2.08605379e-01 5.03488183e-01 1.22536793e-01 1.85058221e-01
-8.34874749e-01 -4.11413342e-01 5.84069848e-01 -1.76939517e-02
2.73992062e-01 -6.56979978e-01 -1.95640504e-01 -3.23868454e-01
-1.87803119e-01 -1.11871326e+00 4.40507978e-02 -8.72325480e-01
-7.68527448e-01 2.21561998e-01 7.12918103e-01 -1.26743472e+00
-8.40989232e-01 -9.35770512e-01 -3.30408573e-01 9.87106085e-01
-1.10825050e+00 -6.76659584e-01 -1.27990171e-01 3.14199060e-01
6.81228399e-01 -1.58569604e-01 7.76610911e-01 -9.73134562e-02
-3.50522161e-01 -1.87145665e-01 -5.51555276e-01 -5.43548405e-01
5.28457880e-01 -1.32314956e+00 -3.13613750e-02 4.05360043e-01
-4.90780264e-01 6.05942428e-01 1.19722962e+00 -9.18000042e-01
-1.71704996e+00 -6.22590959e-01 3.21750969e-01 -5.43410063e-01
8.21211636e-01 -2.40874752e-01 -5.24913132e-01 7.86278427e-01
-7.35454559e-02 -5.77333331e-01 4.08386774e-02 1.71777308e-01
5.45974076e-01 4.34425436e-02 -9.83117223e-01 7.02636242e-01
8.34642112e-01 3.41436453e-02 -7.72611797e-01 6.65628195e-01
7.37611592e-01 -9.68943775e-01 -1.01189899e+00 4.17896062e-01
4.58869070e-01 -5.19245267e-01 9.31787550e-01 -4.67962861e-01
1.25038788e-01 -6.09894037e-01 1.16025381e-01 -1.68492448e+00
-2.93877602e-01 -9.61431444e-01 -1.71754301e-01 8.50868523e-01
4.48335439e-01 -4.01360422e-01 3.73280048e-01 5.65339208e-01
-4.27291602e-01 -8.85188103e-01 -6.62543893e-01 -9.94859993e-01
-2.72724271e-01 -5.73429585e-01 4.86022741e-01 3.30616593e-01
1.25019953e-01 5.27079046e-01 -2.47657552e-01 5.00429511e-01
3.48826617e-01 -5.38329408e-02 9.12851930e-01 -1.13584781e+00
-6.49222136e-01 -2.74312735e-01 4.22582440e-02 -7.39973545e-01
2.07628131e-01 -4.97776568e-01 7.74928808e-01 -1.52019608e+00
-2.90767401e-01 -7.74338663e-01 -2.05362611e-03 7.50266373e-01
7.01698884e-02 -8.56576681e-01 1.49418905e-01 7.56519735e-02
-4.82185304e-01 2.29988128e-01 1.32452452e+00 1.29402325e-01
-6.74680591e-01 4.45642203e-01 -3.74251336e-01 7.71493375e-01
9.14855421e-01 -4.91706729e-01 -6.72894478e-01 -1.31869718e-01
3.01602870e-01 4.70783710e-01 2.91073918e-01 -1.11548114e+00
1.22359686e-01 -5.82166374e-01 7.35712573e-02 -3.20373476e-01
2.82986224e-01 -1.21837437e+00 3.74755740e-01 6.81753099e-01
-2.29992941e-01 9.65537950e-02 3.14554960e-01 4.31987077e-01
1.84472322e-01 -6.78996265e-01 5.57508469e-01 -1.72542050e-01
-6.79399908e-01 -1.88857153e-01 -5.13701200e-01 -3.45657468e-01
1.31204951e+00 -1.35050312e-01 2.02819809e-01 -1.98587433e-01
-8.18793178e-01 7.51447499e-01 1.90406770e-01 4.75805879e-01
2.74014235e-01 -7.93281734e-01 -3.12059790e-01 -5.24079874e-02
1.13422327e-01 2.25174546e-01 -1.46903336e-01 8.50996792e-01
-3.36023718e-01 2.09645510e-01 -4.16409642e-01 -6.19301736e-01
-8.73329639e-01 4.65593785e-01 3.18053007e-01 -5.32549500e-01
-5.09009600e-01 4.80779171e-01 -1.43824682e-01 -5.76202095e-01
3.22784990e-01 -3.79765004e-01 -2.07507372e-01 -2.84999646e-02
2.46030644e-01 3.97817016e-01 1.05140112e-01 -3.77389565e-02
-3.54336292e-01 3.35257113e-01 1.58549473e-01 -4.60020214e-01
1.43864584e+00 1.99276492e-01 1.60367042e-01 5.70994794e-01
2.02206880e-01 -3.68768096e-01 -1.86978412e+00 2.21808821e-01
3.11916679e-01 -2.81036645e-01 5.26076742e-02 -1.06936705e+00
-5.26816845e-01 3.85763079e-01 4.53171462e-01 9.48708951e-02
9.09284532e-01 7.19810370e-03 3.05814594e-01 7.18810380e-01
1.05925989e+00 -1.54038966e+00 3.56620312e-01 7.34955907e-01
1.08516085e+00 -8.36463571e-01 2.50469863e-01 -3.94973606e-01
-9.14458811e-01 1.17112446e+00 6.42770052e-01 -5.95163070e-02
3.28071773e-01 7.29342878e-01 -7.13672768e-03 -3.19574565e-01
-6.42227471e-01 -3.39622021e-01 2.09529147e-01 6.88585937e-01
-7.81166879e-03 -1.90116093e-02 -9.10489187e-02 5.50643504e-01
-1.77945465e-01 1.73620462e-01 3.80008608e-01 1.41423321e+00
-6.61213219e-01 -1.32092404e+00 -6.46164536e-01 3.30905586e-01
-2.19670739e-02 4.13806707e-01 1.06033079e-01 1.02031839e+00
7.78300986e-02 8.53110850e-01 -3.79388779e-01 -4.05025005e-01
7.42943943e-01 2.16593921e-01 9.26522613e-01 -1.27035189e+00
-6.82413459e-01 1.30373433e-01 5.44543564e-01 -8.04212511e-01
-7.37975463e-02 -9.67539310e-01 -1.50999773e+00 3.34160715e-01
-3.42383385e-01 1.68029610e-02 8.21775913e-01 1.11797011e+00
1.19791344e-01 1.08690500e+00 3.49091381e-01 -1.62245548e+00
-9.83070791e-01 -8.84845078e-01 -5.23521006e-01 -2.12898552e-02
-1.02499865e-01 -1.36449277e+00 -1.48173958e-01 6.55028448e-02] | [4.633107662200928, 1.4901834726333618] |
985f1e64-28c6-4132-a543-6ba3f2cf73ba | collaborative-multi-object-tracking-with | 2303.14346 | null | https://arxiv.org/abs/2303.14346v1 | https://arxiv.org/pdf/2303.14346v1.pdf | Collaborative Multi-Object Tracking with Conformal Uncertainty Propagation | Object detection and multiple object tracking (MOT) are essential components of self-driving systems. Accurate detection and uncertainty quantification are both critical for onboard modules, such as perception, prediction, and planning, to improve the safety and robustness of autonomous vehicles. Collaborative object detection (COD) has been proposed to improve detection accuracy and reduce uncertainty by leveraging the viewpoints of multiple agents. However, little attention has been paid on how to leverage the uncertainty quantification from COD to enhance MOT performance. In this paper, as the first attempt, we design the uncertainty propagation framework to address this challenge, called MOT-CUP. Our framework first quantifies the uncertainty of COD through direct modeling and conformal prediction, and propogates this uncertainty information during the motion prediction and association steps. MOT-CUP is designed to work with different collaborative object detectors and baseline MOT algorithms. We evaluate MOT-CUP on V2X-Sim, a comprehensive collaborative perception dataset, and demonstrate a 2% improvement in accuracy and a 2.67X reduction in uncertainty compared to the baselines, e.g., SORT and ByteTrack. MOT-CUP demonstrates the importance of uncertainty quantification in both COD and MOT, and provides the first attempt to improve the accuracy and reduce the uncertainty in MOT based on COD through uncertainty propogation. | ['Fei Miao', 'Caiwen Ding', 'Chen Feng', 'Zhili Zhang', 'Yiming Li', 'Songyang Han', 'Sanbao Su'] | 2023-03-25 | null | null | null | null | ['motion-prediction', 'multiple-object-tracking'] | ['computer-vision', 'computer-vision'] | [-1.66952327e-01 3.96757424e-02 -7.73443207e-02 -4.56042945e-01
-8.11275005e-01 -4.73249346e-01 7.92443871e-01 2.59370863e-01
-5.40374637e-01 4.34928417e-01 -6.41153678e-02 -1.64595410e-01
-8.20770636e-02 -7.38381326e-01 -7.70212650e-01 -3.23156059e-01
-1.61616877e-01 7.09859550e-01 1.07594013e+00 -1.96500301e-01
6.45802468e-02 1.95847943e-01 -1.80412209e+00 5.21686301e-02
9.32366371e-01 1.33804572e+00 2.43720144e-01 7.59756923e-01
1.87932566e-01 7.68775165e-01 -6.10760391e-01 -6.73268512e-02
3.36883575e-01 1.49654165e-01 -2.64735962e-03 -5.11609793e-01
4.93201584e-01 -5.57127237e-01 -3.66800725e-01 8.04592311e-01
3.86836708e-01 1.28502324e-01 5.88800192e-01 -1.89514399e+00
-1.47527680e-01 7.98030078e-01 -5.75950801e-01 2.92759717e-01
-1.35610640e-01 4.89240438e-01 8.23907852e-01 -6.61252975e-01
2.38737851e-01 1.47892392e+00 7.67103791e-01 5.43398440e-01
-9.48688567e-01 -9.27269816e-01 2.15196401e-01 4.90458161e-01
-1.45129251e+00 -6.24714494e-01 3.58119428e-01 -5.86217642e-01
9.32679653e-01 2.01776028e-01 3.32322478e-01 7.20499337e-01
5.67227244e-01 9.68325198e-01 7.69688427e-01 9.40034464e-02
4.95881349e-01 1.87828645e-01 1.93671450e-01 5.01197457e-01
4.83020782e-01 7.05857575e-01 -7.25762725e-01 1.30318120e-01
3.63653712e-02 -3.02944034e-01 2.11798131e-01 -3.66645724e-01
-9.50101435e-01 5.56483984e-01 6.03975594e-01 -3.93738091e-01
-2.87745502e-02 7.54167795e-01 1.19249903e-01 -2.18663201e-01
4.08802778e-02 2.92668194e-01 -9.20508206e-02 -5.09195030e-01
-6.65275156e-01 2.20085263e-01 5.42285562e-01 1.27821648e+00
6.96872234e-01 -6.46854714e-02 -4.62287217e-01 3.52302313e-01
8.47972631e-01 1.16561759e+00 -2.37765968e-01 -1.28574646e+00
3.15515906e-01 5.49113393e-01 3.07964683e-01 -8.93279850e-01
-5.70383847e-01 -1.15894958e-01 -1.88400432e-01 8.46215427e-01
-1.99720543e-02 -1.49122640e-01 -9.63005483e-01 1.78674340e+00
5.08050561e-01 3.61712128e-01 1.00882389e-01 8.76812577e-01
8.43140721e-01 4.57761377e-01 -1.91099867e-02 2.12863311e-01
1.11061180e+00 -9.43760395e-01 -6.74990356e-01 -6.53738856e-01
4.02810454e-01 -4.72078234e-01 4.01361197e-01 2.60533065e-01
-7.26647854e-01 -4.88850951e-01 -1.46183217e+00 6.82452247e-02
-1.10734455e-01 -2.22688079e-01 3.88033628e-01 6.89522743e-01
-7.57704020e-01 1.93490878e-01 -1.37924302e+00 -2.60938168e-01
7.13246047e-01 4.36596096e-01 2.13144511e-01 -1.70631588e-01
-8.25964808e-01 1.07052970e+00 2.11509973e-01 -2.16420695e-01
-1.24332941e+00 -7.77161777e-01 -7.62770474e-01 -2.52127111e-01
6.59686804e-01 -5.43584287e-01 1.28848934e+00 -3.80287059e-02
-1.34376526e+00 -5.82356490e-02 -1.23356871e-01 -9.62058246e-01
5.75549185e-01 -3.03976089e-01 -1.81152835e-01 -2.04145849e-01
2.05621943e-01 1.23083389e+00 6.53965771e-01 -1.49391878e+00
-1.15834105e+00 -3.04419369e-01 1.51427642e-01 9.57219005e-02
6.50575832e-02 -2.12229759e-01 -7.87256300e-01 -2.82135457e-02
1.86701473e-02 -1.16686702e+00 -1.65732801e-01 2.13696688e-01
-2.81165168e-02 -2.45947823e-01 1.23491478e+00 -2.04635412e-01
9.50656235e-01 -2.24012804e+00 -1.90065727e-01 3.59656289e-03
5.77842832e-01 1.07805096e-01 -1.95758119e-02 5.14730699e-02
1.05301118e+00 -7.37743452e-02 -1.30182467e-02 -7.72782326e-01
3.96855205e-01 4.70167547e-01 -2.09019199e-01 3.46968621e-01
2.97089547e-01 1.07348728e+00 -9.84870791e-01 -4.94109869e-01
5.18570483e-01 3.37493241e-01 -7.39534318e-01 1.01504512e-02
-7.05758631e-01 2.62612015e-01 -4.40248191e-01 6.60983562e-01
9.46849763e-01 -6.99362904e-02 -1.57165036e-01 -2.75283307e-01
-2.55367279e-01 3.13763678e-01 -1.21322501e+00 1.27308393e+00
-2.26477996e-01 9.72252071e-01 1.95060462e-01 -1.67153716e-01
7.31818259e-01 -2.93605089e-01 5.35320878e-01 -7.91117370e-01
3.77746344e-01 3.37738101e-03 7.09702745e-02 -1.49409041e-01
1.03235567e+00 4.47072864e-01 -2.67103523e-01 1.32352188e-01
-4.50259626e-01 -3.06495219e-01 2.08907619e-01 6.55956507e-01
1.37482584e+00 1.29057437e-01 -6.55959249e-02 6.98641967e-03
-1.93017591e-02 3.50667834e-01 8.92451227e-01 9.72990096e-01
-7.19427109e-01 3.34471017e-01 7.09630176e-02 1.75712705e-01
-5.76549888e-01 -1.17359507e+00 -2.93529451e-01 8.18016231e-01
9.97813702e-01 -6.00169539e-01 -3.90365154e-01 -7.02369690e-01
6.47736192e-01 9.22751486e-01 -5.08123636e-01 -3.74939948e-01
-2.96715498e-01 -5.89362442e-01 6.37904346e-01 9.16420996e-01
6.67823434e-01 -3.38230759e-01 -1.08851004e+00 2.21523881e-01
-1.07007906e-01 -1.33444393e+00 -3.27648789e-01 1.14521392e-01
-2.77358323e-01 -1.00681937e+00 3.49616230e-01 7.96765313e-02
1.48793340e-01 6.27158046e-01 8.05058181e-01 1.52391894e-02
-9.39875469e-02 6.07923150e-01 -3.69247258e-01 -9.30781126e-01
-4.97201324e-01 -4.20487612e-01 5.02938569e-01 -3.10232818e-01
4.08392817e-01 -2.65782744e-01 -4.91234839e-01 7.69818187e-01
-5.51652670e-01 -4.02164124e-02 7.32261121e-01 4.75835711e-01
6.13811851e-01 2.97083370e-02 1.67641446e-01 -5.60314953e-02
1.43203378e-01 -5.78281283e-01 -1.15696490e+00 -7.34688938e-02
-9.62429821e-01 1.85176149e-01 -7.90915862e-02 -2.81802773e-01
-8.66968513e-01 5.97785152e-02 1.94501042e-01 -5.19959271e-01
2.14268491e-01 1.31626993e-01 -1.07432865e-01 -3.41111004e-01
7.12613583e-01 -3.65229636e-01 -3.62873152e-02 -2.89914124e-02
3.98225188e-01 7.20206082e-01 6.03915215e-01 -4.10654426e-01
7.25309074e-01 7.96572685e-01 1.75912037e-01 -5.29912174e-01
-4.42899525e-01 -5.02329290e-01 -8.13467801e-02 -4.07294929e-01
8.37885976e-01 -1.12565529e+00 -9.60791707e-01 2.28580698e-01
-1.06109357e+00 -4.04378325e-01 -3.25322479e-01 6.48662567e-01
-3.96430194e-01 1.27797335e-01 -1.00072987e-01 -1.08963633e+00
-5.77536486e-02 -1.39123523e+00 1.17307270e+00 3.67114365e-01
2.32887775e-01 -4.01688039e-01 -9.84639302e-02 4.89307016e-01
5.26407063e-01 -1.38971098e-02 1.40381545e-01 -3.65706712e-01
-1.50031638e+00 -4.47837599e-02 -3.99900764e-01 -2.55016685e-02
-2.89049745e-01 1.23042911e-01 -1.05878615e+00 -2.75986433e-01
-4.49059308e-01 -1.15657680e-01 1.27581465e+00 4.48060423e-01
3.94561410e-01 2.31651917e-01 -8.61637533e-01 4.19203401e-01
1.07690918e+00 3.74167889e-01 4.22542244e-01 3.06992292e-01
5.44377744e-01 3.45320880e-01 1.13908195e+00 5.07456481e-01
1.01284993e+00 1.04940343e+00 1.02210009e+00 6.35748446e-01
-5.64822495e-01 -1.38858825e-01 5.35680830e-01 3.78848255e-01
4.47535336e-01 -5.05007625e-01 -1.02156734e+00 4.84800458e-01
-2.17107725e+00 -7.68857121e-01 -2.82603711e-01 1.96839035e+00
3.14863890e-01 4.94587034e-01 -7.23995343e-02 -3.19566518e-01
4.54310209e-01 7.86263794e-02 -7.68644869e-01 1.14557989e-01
1.14780836e-01 -6.15245938e-01 9.00663733e-01 8.66456568e-01
-1.02784514e+00 9.80871201e-01 5.62298393e+00 8.08396041e-01
-7.92038143e-01 1.46005765e-01 8.61775801e-02 -4.63157326e-01
-2.00049236e-01 2.58277655e-01 -1.50153899e+00 4.04989779e-01
9.77725625e-01 -1.36481300e-01 3.57010990e-01 9.59695399e-01
1.57599851e-01 -6.11433446e-01 -1.30168307e+00 7.50377417e-01
2.51995604e-02 -1.40847981e+00 -5.54238141e-01 8.53169113e-02
6.90202713e-01 6.71624422e-01 1.53790861e-01 6.81490123e-01
8.37091684e-01 -7.44679630e-01 1.16016197e+00 5.05772710e-01
3.31345469e-01 -5.65454721e-01 8.08265626e-01 3.80800158e-01
-1.43267775e+00 -3.36235225e-01 -3.59213054e-01 4.36038449e-02
4.22450870e-01 5.46852112e-01 -1.24205279e+00 2.94418067e-01
9.23107266e-01 5.18208742e-01 -4.88995522e-01 1.22164845e+00
-4.90017682e-02 4.35364544e-01 -9.15650547e-01 -3.14409286e-01
-1.22794304e-02 3.03871721e-01 1.10972261e+00 7.91230440e-01
1.51005685e-01 1.48038575e-02 4.17800397e-01 1.03431666e+00
3.18599157e-02 -6.75006390e-01 -2.58183509e-01 1.20150492e-01
1.08298326e+00 1.05455959e+00 -3.80191714e-01 -3.24960649e-01
-2.77264774e-01 2.59204239e-01 1.64960623e-02 -6.98072044e-03
-1.38394535e+00 -6.35519549e-02 1.14348733e+00 2.51773112e-02
5.39397597e-01 -4.81542140e-01 -5.22737384e-01 -5.56974828e-01
-4.49426519e-03 -4.17857558e-01 1.88050717e-02 -7.30113387e-01
-7.71928430e-01 5.42619467e-01 3.02744299e-01 -1.27139401e+00
-1.95932612e-01 -4.07498747e-01 -2.57503927e-01 4.92562890e-01
-1.55096292e+00 -1.20332193e+00 -7.50644267e-01 6.09940253e-02
4.06979084e-01 -7.06449058e-03 1.35327876e-01 1.46884337e-01
-4.95774090e-01 5.77130318e-01 1.26292497e-01 -5.16616702e-01
7.08203793e-01 -1.06210637e+00 4.72155243e-01 1.07707357e+00
-9.96131971e-02 1.62150890e-01 8.38120520e-01 -1.02960718e+00
-1.68934572e+00 -1.37229598e+00 2.04210177e-01 -1.09654260e+00
6.48892641e-01 -3.60779285e-01 -3.57714176e-01 5.47719896e-01
-1.32550225e-01 2.32827932e-01 8.99917930e-02 -1.58676907e-01
-3.11678648e-01 -3.52555454e-01 -1.15547431e+00 5.47918081e-01
1.01590121e+00 -1.51090205e-01 -4.48534578e-01 3.27333175e-02
1.19095683e+00 -6.74953759e-01 -6.68716967e-01 7.75863469e-01
7.46863902e-01 -1.02246284e+00 9.07283723e-01 1.15689069e-01
-2.18081042e-01 -9.29977238e-01 -6.16361618e-01 -1.04525423e+00
-2.69691795e-01 -3.83128017e-01 -4.84899133e-01 1.20653057e+00
4.93066907e-01 -6.09615624e-01 8.11601162e-01 6.34325027e-01
-6.65487647e-01 -3.68673265e-01 -1.09289336e+00 -1.06838250e+00
-4.46666896e-01 -1.14531696e+00 7.10243642e-01 1.23330198e-01
-3.70093286e-01 1.22337736e-01 -1.54295757e-01 8.87594044e-01
9.71998751e-01 -7.14202970e-02 1.23050308e+00 -1.13561940e+00
-2.54965216e-01 -3.55986804e-01 -7.91415155e-01 -1.39294159e+00
-3.53137136e-01 -5.04593432e-01 7.49141634e-01 -1.60584009e+00
2.78898031e-02 -9.84674096e-01 -2.02698410e-01 3.75433266e-01
-1.42066538e-01 1.98492527e-01 5.22172034e-01 3.58007163e-01
-1.24834478e+00 7.67787516e-01 7.34997869e-01 -4.94120032e-01
-2.99954534e-01 -1.48214754e-02 -5.40811777e-01 5.98769367e-01
5.92015624e-01 -5.85840583e-01 -4.64404166e-01 -6.87364936e-01
1.33148253e-01 -2.16952071e-01 5.18420160e-01 -1.62109947e+00
7.98579335e-01 -3.24216515e-01 -2.17437022e-03 -1.25467169e+00
7.82162309e-01 -9.78392661e-01 8.05230588e-02 4.12224710e-01
4.84276973e-02 -2.77167380e-01 7.81156301e-01 9.94935930e-01
2.59503514e-01 1.32285744e-01 5.83910704e-01 5.14716625e-01
-1.26920736e+00 2.33350560e-01 -2.87405610e-01 -5.60548380e-02
1.35257483e+00 -1.19581833e-01 -7.63583124e-01 -2.15507478e-01
-2.06970394e-01 8.94122183e-01 5.26236355e-01 6.56069517e-01
7.56155133e-01 -1.31264055e+00 -6.18614435e-01 9.61369798e-02
5.45786619e-01 2.47064173e-01 1.38473898e-01 1.16493499e+00
-7.99216703e-02 4.86263424e-01 5.14026247e-02 -1.30698442e+00
-1.27737236e+00 3.90640229e-01 3.17212284e-01 1.67968169e-01
-2.95607448e-01 1.04475164e+00 1.22389920e-01 -1.22857131e-01
4.39835638e-01 -6.17677510e-01 1.78296305e-02 -3.83856237e-01
6.40631258e-01 5.93687892e-01 1.20539190e-02 -6.14046991e-01
-7.14859128e-01 2.88291156e-01 -9.01690349e-02 -2.14295715e-01
7.80786455e-01 -5.51458001e-01 3.55138779e-01 3.08630496e-01
5.29854059e-01 2.27790073e-01 -1.97619772e+00 -6.36157915e-02
-3.98505628e-02 -5.93988299e-01 5.47827065e-01 -9.44040954e-01
-6.34980559e-01 5.55096388e-01 1.00215709e+00 -2.29090564e-02
5.65701067e-01 2.69377351e-01 6.89823925e-01 4.88820463e-01
9.08360839e-01 -1.06597257e+00 1.70422345e-01 6.02263570e-01
6.52704358e-01 -1.66977239e+00 6.87480345e-02 -3.92468065e-01
-4.97612089e-01 4.82897192e-01 1.08791840e+00 2.88572639e-01
4.64651585e-01 8.55493724e-01 -6.31449968e-02 -1.08678266e-01
-1.09136820e+00 -3.78026515e-01 3.30967188e-01 7.54768670e-01
-4.52821344e-01 2.12085336e-01 1.63876444e-01 6.78634465e-01
1.44019648e-01 -5.72651774e-02 3.35935742e-01 1.18635643e+00
-9.69937146e-01 -6.81722105e-01 -5.32715142e-01 4.89707917e-01
2.44056702e-01 3.72609161e-02 -1.70727924e-01 6.34379923e-01
3.32208574e-01 1.57634401e+00 1.53424725e-01 -1.10861123e+00
2.40997419e-01 -6.04733765e-01 3.22670072e-01 -3.94758850e-01
-6.67007118e-02 -2.69402564e-01 3.16962421e-01 -9.25624430e-01
-5.74859381e-02 -7.47768819e-01 -1.72472286e+00 -4.13310230e-01
-5.96297801e-01 -7.53644556e-02 1.12690127e+00 9.36455131e-01
7.94889271e-01 9.14740264e-01 3.90116811e-01 -8.77911925e-01
-6.20917439e-01 -6.60362005e-01 -5.16210869e-02 -2.20117614e-01
4.50443506e-01 -1.23173833e+00 -5.15056014e-01 -5.18397748e-01] | [6.661075592041016, -2.1271770000457764] |
a0ce96d6-38e2-419d-8852-b3fc246b4d70 | stylenat-giving-each-head-a-new-perspective | 2211.05770 | null | https://arxiv.org/abs/2211.05770v1 | https://arxiv.org/pdf/2211.05770v1.pdf | StyleNAT: Giving Each Head a New Perspective | Image generation has been a long sought-after but challenging task, and performing the generation task in an efficient manner is similarly difficult. Often researchers attempt to create a "one size fits all" generator, where there are few differences in the parameter space for drastically different datasets. Herein, we present a new transformer-based framework, dubbed StyleNAT, targeting high-quality image generation with superior efficiency and flexibility. At the core of our model, is a carefully designed framework that partitions attention heads to capture local and global information, which is achieved through using Neighborhood Attention (NA). With different heads able to pay attention to varying receptive fields, the model is able to better combine this information, and adapt, in a highly flexible manner, to the data at hand. StyleNAT attains a new SOTA FID score on FFHQ-256 with 2.046, beating prior arts with convolutional models such as StyleGAN-XL and transformers such as HIT and StyleSwin, and a new transformer SOTA on FFHQ-1024 with an FID score of 4.174. These results show a 6.4% improvement on FFHQ-256 scores when compared to StyleGAN-XL with a 28% reduction in the number of parameters and 56% improvement in sampling throughput. Code and models will be open-sourced at https://github.com/SHI-Labs/StyleNAT . | ['Humphrey Shi', 'Zhangyang Wang', 'Xingqian Xu', 'Ali Hassani', 'Steven Walton'] | 2022-11-10 | null | null | null | null | ['face-generation'] | ['computer-vision'] | [ 1.71459526e-01 1.68602895e-02 1.10017814e-01 -6.81909025e-02
-1.01299465e+00 -6.35290504e-01 6.85800016e-01 -5.05290985e-01
-1.64777443e-01 6.43180013e-01 3.87600720e-01 -3.52669328e-01
1.07105426e-01 -9.90891397e-01 -7.45016873e-01 -6.95631623e-01
2.90932536e-01 3.39300543e-01 -9.83361006e-02 -4.66894537e-01
5.80242239e-02 3.87894154e-01 -1.36256170e+00 2.47141153e-01
6.52821541e-01 8.94797146e-01 3.98661822e-01 9.43740666e-01
1.80681199e-01 6.78322256e-01 -6.90892637e-01 -5.15284836e-01
5.01731277e-01 -5.96292078e-01 -8.16748559e-01 -5.48559166e-02
6.46408260e-01 -3.27895284e-01 -3.79151851e-01 5.65867186e-01
9.43365991e-01 -8.52986351e-02 3.14971447e-01 -1.02341843e+00
-8.03644896e-01 7.38125741e-01 -5.04932702e-01 7.49808690e-03
9.52504277e-02 5.94381571e-01 1.04204488e+00 -8.49356890e-01
5.29311538e-01 1.20678902e+00 6.07066035e-01 7.47374773e-01
-1.41120541e+00 -9.10901308e-01 -2.36708209e-01 -7.20671639e-02
-1.46387994e+00 -6.42397761e-01 3.05748612e-01 -2.17471734e-01
8.12297642e-01 4.22928005e-01 6.73945248e-01 1.27685201e+00
2.70021230e-01 5.66026092e-01 1.01726067e+00 -2.89059073e-01
1.09257586e-01 -1.61643624e-01 -5.01089871e-01 5.24964988e-01
9.18646455e-02 7.18625784e-02 -4.04661953e-01 1.75039679e-01
1.05712974e+00 -1.18353911e-01 -2.97448397e-01 3.48724355e-03
-1.30315924e+00 8.77611339e-01 7.12959468e-01 3.09117198e-01
-3.85307044e-01 6.96429729e-01 1.15830489e-01 1.41058326e-01
2.73561627e-01 8.18991244e-01 -3.61465305e-01 -2.16723666e-01
-1.05818641e+00 5.03771007e-01 4.49280649e-01 1.12418950e+00
7.65749454e-01 3.20724428e-01 -7.41467297e-01 5.96714675e-01
1.41039416e-02 7.68975079e-01 4.49572325e-01 -1.04468572e+00
2.52803922e-01 2.86471277e-01 -1.21558301e-01 -6.88672066e-01
-1.81512430e-01 -8.65099192e-01 -1.02135503e+00 2.95499861e-01
2.13617429e-01 -2.92966723e-01 -1.34552026e+00 1.88483119e+00
1.11700352e-02 7.32104629e-02 -1.69285312e-01 6.39230311e-01
7.22462356e-01 9.16146040e-01 -4.47034203e-02 2.89914966e-01
1.54877436e+00 -1.16582119e+00 -4.24356788e-01 -2.57597119e-01
4.52477396e-01 -9.90583241e-01 1.33071005e+00 4.45411295e-01
-1.35924757e+00 -7.09335268e-01 -9.30370033e-01 -2.84378737e-01
-2.31202543e-01 7.96223953e-02 4.68763232e-01 7.92902768e-01
-1.58340847e+00 4.11253601e-01 -5.13606131e-01 -2.92974621e-01
6.07951641e-01 5.03388226e-01 -1.63176581e-01 -1.77064434e-01
-1.04073393e+00 5.84483266e-01 1.73138559e-01 -1.55536011e-01
-9.20149982e-01 -9.91813600e-01 -6.82022691e-01 2.20744431e-01
3.26897383e-01 -1.13333809e+00 1.32596684e+00 -7.20848203e-01
-1.75762570e+00 5.54222703e-01 -1.02240689e-01 -5.94144285e-01
5.41070580e-01 -1.43422171e-01 -2.27903903e-01 -5.82214631e-02
1.30611598e-01 1.21535540e+00 8.85546684e-01 -1.02623987e+00
-5.16533256e-01 1.44313090e-03 1.16021812e-01 1.46146014e-01
-2.93718189e-01 -9.45120528e-02 -7.23286211e-01 -8.83512914e-01
-3.67564917e-01 -1.08784199e+00 -3.63543838e-01 -2.66230762e-01
-3.85507643e-01 -2.90556755e-02 7.63159275e-01 -5.80372512e-01
1.23766184e+00 -2.02459860e+00 1.90538610e-03 1.09939605e-01
4.07237232e-01 4.99430716e-01 -4.50152159e-01 4.65202779e-01
5.41573465e-02 4.39321458e-01 -2.15410173e-01 -4.58612263e-01
2.19583437e-02 -4.32925783e-02 -2.88463861e-01 -2.27390183e-03
3.10499489e-01 1.27370214e+00 -7.45040536e-01 -8.61923471e-02
8.85112137e-02 6.19047999e-01 -8.94850910e-01 1.89877108e-01
-2.01892003e-01 3.93975735e-01 -2.24368423e-01 5.21736443e-01
7.15888441e-01 -5.25952220e-01 9.39097255e-02 -1.82216629e-01
-1.44858092e-01 1.34049803e-01 -9.90644813e-01 1.80383968e+00
-7.17601180e-01 7.28516400e-01 -7.03208009e-03 -4.56044823e-01
8.65567625e-01 3.19631279e-01 2.74658859e-01 -9.81644034e-01
1.58555880e-01 1.68683484e-01 8.92873779e-02 -4.96065356e-02
7.84879982e-01 6.86195269e-02 -2.31847927e-01 4.63888168e-01
2.90772170e-01 -3.11512947e-01 3.73374581e-01 3.21350753e-01
1.32889485e+00 8.62137377e-02 3.18093970e-02 -2.84245759e-01
1.50065288e-01 -2.58400857e-01 3.38491321e-01 8.40916574e-01
2.41855308e-01 1.06641340e+00 3.29558969e-01 -3.43166202e-01
-1.36470532e+00 -9.89046574e-01 7.62627795e-02 9.65774179e-01
-1.75730303e-01 -7.06465304e-01 -9.54455376e-01 -4.66237664e-01
-2.40023315e-01 6.48634732e-01 -6.66935146e-01 -1.73131794e-01
-6.75704479e-01 -6.03336036e-01 7.62441456e-01 2.97300041e-01
8.68867934e-01 -1.27851725e+00 -6.53629899e-01 1.62062794e-01
-1.21227503e-01 -9.33479369e-01 -7.45775521e-01 -6.67338772e-03
-5.07713258e-01 -5.77490866e-01 -1.02859843e+00 -5.51308274e-01
3.77534509e-01 2.59210408e-01 1.36456251e+00 2.29153112e-02
-2.77093351e-01 -2.51527205e-02 -3.42206627e-01 -4.57483768e-01
-3.26601595e-01 5.30872345e-01 -3.59412491e-01 3.31465714e-02
-2.07692310e-01 -4.84246194e-01 -8.84066880e-01 2.77948707e-01
-1.15488648e+00 2.97762096e-01 9.65539575e-01 9.06666875e-01
5.00560045e-01 -2.13591114e-01 4.81579810e-01 -9.27259386e-01
4.64504689e-01 -5.18281937e-01 -4.04000700e-01 -3.54794711e-02
-6.93437636e-01 9.60767642e-02 8.20483208e-01 -2.01331526e-01
-8.94602358e-01 -6.40064031e-02 -3.32823038e-01 -5.25258183e-01
-7.54081905e-02 1.66988701e-01 -2.84265637e-01 -4.60901074e-02
7.21405745e-01 2.73848265e-01 -8.24233983e-03 -3.81084710e-01
5.60682178e-01 4.49172586e-01 4.53678906e-01 -4.67065871e-01
8.72810900e-01 2.42545575e-01 -1.67854339e-01 -5.98260522e-01
-6.60213351e-01 -5.50214201e-02 -1.25958631e-02 1.35402858e-01
7.84179032e-01 -1.06403828e+00 -4.22508121e-01 7.14438140e-01
-8.60884070e-01 -9.25966918e-01 -6.11519516e-01 8.67708400e-02
-6.16529226e-01 -9.15773734e-02 -5.08886456e-01 -2.91680425e-01
-7.04183459e-01 -1.30333042e+00 1.20906174e+00 3.07911456e-01
-2.20681950e-01 -7.67368674e-01 2.91274656e-02 4.30550873e-01
9.59362566e-01 1.34779096e-01 5.53592861e-01 -2.23368317e-01
-7.65720546e-01 1.99125543e-01 -3.53174746e-01 2.93282479e-01
1.48326144e-01 -1.10536650e-01 -1.01431847e+00 -5.33479810e-01
-3.87755215e-01 -1.54409766e-01 9.33009505e-01 4.49797839e-01
1.19119561e+00 -4.28451508e-01 -1.48826703e-01 1.04185569e+00
1.47456479e+00 1.66645631e-01 1.05869401e+00 1.16875447e-01
7.74022341e-01 1.28853172e-01 2.30685726e-01 5.19875526e-01
4.06733930e-01 8.93792152e-01 4.12838310e-01 -4.26461369e-01
-6.63229883e-01 -3.30801070e-01 3.14241767e-01 5.94164312e-01
-1.10113129e-01 -7.93871403e-01 -7.79747784e-01 5.66841662e-01
-1.63054025e+00 -1.02408171e+00 4.47371118e-02 2.05443144e+00
9.95989323e-01 -4.20358479e-02 1.43505156e-01 8.91281944e-03
5.90919733e-01 1.89865381e-01 -4.38004732e-01 -4.27052557e-01
-1.90398604e-01 8.21495771e-01 6.12677634e-01 3.45798224e-01
-9.19447184e-01 1.12406707e+00 6.18846178e+00 1.23499858e+00
-1.32600892e+00 -3.05312667e-02 9.05230582e-01 -2.35157743e-01
-4.29541439e-01 4.93748598e-02 -7.46631622e-01 7.03156888e-01
1.18851745e+00 -9.97075364e-02 6.51494801e-01 4.28102434e-01
2.02081636e-01 1.33100897e-01 -7.83317149e-01 9.37075734e-01
-7.51655956e-04 -1.67287207e+00 2.97878355e-01 4.52357352e-01
9.62293327e-01 -1.90530885e-02 4.82042611e-01 3.46773475e-01
5.47876477e-01 -1.33496714e+00 9.50875044e-01 4.90825117e-01
1.19422114e+00 -7.69129515e-01 5.98119974e-01 2.14115810e-02
-1.15629089e+00 -1.46193117e-01 -2.17152759e-01 1.20749567e-02
7.00645968e-02 6.51293039e-01 -9.24570203e-01 5.12921691e-01
8.14979196e-01 5.20896912e-01 -6.66195929e-01 9.45501387e-01
-1.24851160e-01 6.39682531e-01 -1.93625227e-01 1.10358335e-01
3.11416507e-01 8.90677124e-02 3.90969902e-01 1.24202383e+00
8.36423099e-01 1.71877127e-02 -1.59297645e-01 1.05539203e+00
-3.72833788e-01 -4.99915443e-02 -5.96811414e-01 1.65049985e-01
5.17121196e-01 1.45998752e+00 -4.93557423e-01 -4.49554235e-01
3.13637219e-02 1.04042196e+00 5.52426204e-02 2.96371341e-01
-1.05487645e+00 -5.41750193e-01 6.49431169e-01 3.45185071e-01
6.33494973e-01 -1.32330820e-01 -2.60568172e-01 -8.58215570e-01
-2.38920033e-01 -1.18507493e+00 7.70481676e-02 -9.16353345e-01
-8.69748950e-01 8.77002954e-01 -1.96501210e-01 -1.04742920e+00
-2.15632811e-01 -3.41742814e-01 -6.18717372e-01 1.06073713e+00
-1.35730457e+00 -1.50269151e+00 -5.15891314e-01 6.12608254e-01
6.56211138e-01 -1.06827870e-01 7.29788125e-01 3.93060446e-01
-5.47992110e-01 1.04846275e+00 -1.59417707e-02 -5.35018258e-02
7.92102754e-01 -1.21977127e+00 9.00834978e-01 7.68217087e-01
1.52751833e-01 5.60925245e-01 5.06832004e-01 -3.63557726e-01
-1.43101609e+00 -1.38274252e+00 8.19855750e-01 -3.93553466e-01
2.78818309e-01 -4.84909981e-01 -5.11264443e-01 5.14172554e-01
6.26089513e-01 -3.81871685e-02 3.65851492e-01 -2.58973271e-01
-3.84656489e-01 -2.09550858e-01 -1.13711417e+00 7.38351941e-01
1.24172306e+00 -2.18409911e-01 2.36953571e-01 1.13305867e-01
9.07820880e-01 -5.15796781e-01 -7.31740415e-01 2.15368092e-01
5.24143457e-01 -1.13398027e+00 9.84354734e-01 -1.23167969e-01
6.09718382e-01 -3.25392157e-01 -1.19988717e-01 -1.46684790e+00
-8.62097859e-01 -1.11005783e+00 4.33520041e-02 1.22212195e+00
4.63963866e-01 -6.41561568e-01 7.88913667e-01 6.06656931e-02
-2.98790544e-01 -7.87981510e-01 -6.06094122e-01 -6.29638553e-01
1.14084922e-01 -1.01202056e-01 9.72512841e-01 5.33626199e-01
-6.28836036e-01 6.50091529e-01 -5.39053380e-01 -1.46871522e-01
3.94515246e-01 -3.67976539e-02 9.97572303e-01 -8.53725612e-01
-4.28405434e-01 -6.01647913e-01 -1.33074507e-01 -1.05646288e+00
-3.63668114e-01 -8.23931575e-01 -1.83741376e-02 -1.43924093e+00
1.21293455e-01 -6.56600595e-01 -1.04798511e-01 6.42696500e-01
-1.21616550e-01 8.86020422e-01 5.80247819e-01 9.99415219e-02
-3.52108508e-01 5.76465905e-01 1.48199952e+00 -1.04751274e-01
-1.90734267e-01 -1.18655756e-01 -1.17791831e+00 2.47566730e-01
1.13752735e+00 -1.66868731e-01 -4.83998924e-01 -7.07641125e-01
1.75672293e-01 -5.31751141e-02 3.98796380e-01 -1.42360759e+00
-4.80268970e-02 3.00701614e-02 4.44459528e-01 -2.10252643e-01
4.28211689e-01 -4.66370702e-01 5.80911279e-01 3.67520064e-01
-2.31879756e-01 3.01294923e-01 2.58149534e-01 2.49680206e-01
4.13260534e-02 1.54494941e-01 7.80131340e-01 -2.73716599e-01
-4.97637302e-01 5.99145055e-01 -2.38512695e-01 1.19639978e-01
8.90866399e-01 -1.72638014e-01 -5.04201651e-01 -6.00239992e-01
-4.11764055e-01 1.99322812e-02 5.71112096e-01 3.71818304e-01
4.82709289e-01 -1.45514262e+00 -1.06467819e+00 4.31879103e-01
-7.76566640e-02 1.07990026e-01 4.96885508e-01 5.35538733e-01
-5.76170981e-01 4.03490931e-01 -3.90070826e-01 -4.00798351e-01
-9.14670289e-01 2.68139273e-01 1.92473337e-01 -6.54816091e-01
-3.88845652e-01 1.05728638e+00 3.83859098e-01 -3.26969177e-01
-2.84338504e-01 -2.20898703e-01 3.76439244e-02 -8.61047953e-02
5.32536864e-01 4.06224519e-01 1.39671400e-01 -4.97975439e-01
-2.98975259e-02 5.93568742e-01 -2.14191675e-02 -1.90901056e-01
1.22967899e+00 9.58300848e-03 6.78443387e-02 -2.34410554e-01
1.03945744e+00 2.95968682e-01 -1.48446715e+00 -6.93701282e-02
-5.42682648e-01 -4.37898338e-01 1.25732884e-01 -1.01971197e+00
-1.43551242e+00 8.09397280e-01 6.23706460e-01 2.30008706e-01
1.34481573e+00 -1.11324258e-01 1.18005359e+00 -1.45363793e-01
2.60892630e-01 -6.82642519e-01 2.69730866e-01 5.55462122e-01
1.04172873e+00 -7.22384512e-01 -1.87139869e-01 -7.60412887e-02
-5.66051841e-01 6.74167931e-01 6.66885018e-01 -1.49928451e-01
1.91930860e-01 3.73976171e-01 -6.33341596e-02 -2.37984415e-02
-8.19864452e-01 -1.73522994e-01 1.46830529e-01 4.77871746e-01
4.39301729e-01 1.64163470e-01 8.07225332e-02 3.28683227e-01
-5.93974650e-01 -6.07275292e-02 4.67351019e-01 6.59349144e-01
-2.40103066e-01 -1.29934657e+00 -3.29755008e-01 3.99708569e-01
-5.69949031e-01 -4.18968499e-01 -3.90087813e-02 6.31854951e-01
3.39646518e-01 9.02044237e-01 2.22502843e-01 -6.04092658e-01
2.41030917e-01 -3.00799817e-01 3.65274936e-01 -5.81990659e-01
-8.69872332e-01 1.42953485e-01 -3.25473472e-02 -6.74308062e-01
-1.00924820e-01 -3.61001223e-01 -9.62429345e-01 -7.09471524e-01
-1.07223399e-01 -4.29373607e-02 3.94905061e-01 4.35649574e-01
9.12541568e-01 7.71071136e-01 7.38152623e-01 -1.00619626e+00
-3.31147939e-01 -8.66198540e-01 -3.27409774e-01 2.05934137e-01
2.78299809e-01 -1.67993024e-01 2.58038566e-02 6.04898557e-02] | [11.511700630187988, -0.44549763202667236] |
90e99467-1e6b-4a12-986e-8e1bd6bd7227 | hand-priming-in-object-localization-for | 2002.12557 | null | https://arxiv.org/abs/2002.12557v1 | https://arxiv.org/pdf/2002.12557v1.pdf | Hand-Priming in Object Localization for Assistive Egocentric Vision | Egocentric vision holds great promises for increasing access to visual information and improving the quality of life for people with visual impairments, with object recognition being one of the daily challenges for this population. While we strive to improve recognition performance, it remains difficult to identify which object is of interest to the user; the object may not even be included in the frame due to challenges in camera aiming without visual feedback. Also, gaze information, commonly used to infer the area of interest in egocentric vision, is often not dependable. However, blind users often tend to include their hand either interacting with the object that they wish to recognize or simply placing it in proximity for better camera aiming. We propose localization models that leverage the presence of the hand as the contextual information for priming the center area of the object of interest. In our approach, hand segmentation is fed to either the entire localization network or its last convolutional layers. Using egocentric datasets from sighted and blind individuals, we show that the hand-priming achieves higher precision than other approaches, such as fine-tuning, multi-class, and multi-task learning, which also encode hand-object interactions in localization. | ['Hernisa Kacorri', 'Kyungjun Lee', 'Abhinav Shrivastava'] | 2020-02-28 | null | null | null | null | ['hand-segmentation'] | ['computer-vision'] | [-1.38449922e-01 -1.65398479e-01 -2.60327786e-01 -2.48840228e-01
-4.84760970e-01 -6.77928805e-01 1.03625402e-01 -8.31868649e-02
-7.00890779e-01 3.36610764e-01 4.87376451e-01 -8.00153762e-02
-1.26703784e-01 -1.84762150e-01 -6.38481855e-01 -7.44719148e-01
4.79256660e-01 1.87315717e-01 2.14988306e-01 2.32047141e-01
6.22228861e-01 5.80264091e-01 -1.72053385e+00 3.20907205e-01
7.56289482e-01 8.87042522e-01 5.50824165e-01 7.42920578e-01
1.58000872e-01 5.28218985e-01 -4.38553393e-01 -5.98920062e-02
3.00038755e-01 -3.43464375e-01 -7.86886275e-01 1.76620871e-01
1.11274993e+00 -8.77810478e-01 -2.64517576e-01 1.14838040e+00
7.88448274e-01 1.00492768e-01 5.53054631e-01 -1.07871461e+00
-4.92836326e-01 8.39701071e-02 -9.12896991e-01 5.71195483e-01
2.64577121e-01 4.40577537e-01 9.79012311e-01 -5.71993291e-01
3.68253469e-01 1.07310879e+00 5.06931357e-02 5.96011579e-01
-8.37682247e-01 -4.58652049e-01 5.90055406e-01 5.49494922e-01
-1.20255744e+00 -7.49651849e-01 5.40326178e-01 -7.02638209e-01
8.62755716e-01 4.97097429e-03 8.53104711e-01 9.87709343e-01
-2.55258083e-01 9.68716919e-01 8.84622514e-01 -5.77960193e-01
1.15089417e-01 -6.05873801e-02 2.67595768e-01 4.69897777e-01
3.36962610e-01 -4.38681245e-02 -5.64955592e-01 1.58105955e-01
1.10490644e+00 3.82286340e-01 -7.63459325e-01 -4.47224468e-01
-1.32437420e+00 3.41544300e-01 8.04449677e-01 2.48479605e-01
-6.41235411e-01 2.06103340e-01 -1.17300704e-01 -2.18586639e-01
1.70675233e-01 2.93420374e-01 -3.02088499e-01 -3.32642525e-01
-7.24579990e-01 2.69770563e-01 1.80753544e-01 9.57567573e-01
5.03059804e-01 -4.91241068e-01 -6.32481217e-01 5.50767660e-01
4.06194806e-01 2.70031095e-01 4.57076550e-01 -8.41584921e-01
5.96428812e-01 9.12518024e-01 5.23125768e-01 -2.58065432e-01
-5.61735988e-01 -4.32401419e-01 -2.02957898e-01 6.68543458e-01
1.05987644e+00 -2.35016674e-01 -1.07029569e+00 1.83402824e+00
1.75594911e-01 -1.54753894e-01 -4.23329353e-01 1.57350540e+00
3.58800292e-01 -1.23759016e-01 3.19114476e-01 1.05956852e-01
1.65686989e+00 -1.11742508e+00 -4.84336078e-01 -6.38500333e-01
4.38632339e-01 -5.86558163e-01 1.19225967e+00 2.64929712e-01
-8.77591491e-01 -2.74696320e-01 -5.31404972e-01 -2.30604470e-01
-1.58557221e-01 5.17001629e-01 7.35363841e-01 6.11428261e-01
-1.16267717e+00 1.26553163e-01 -7.39995599e-01 -7.38136470e-01
6.86848164e-01 3.96378279e-01 -3.53351176e-01 -1.41974524e-01
-3.50819647e-01 7.78633177e-01 1.27355486e-01 2.29299515e-02
-6.50380731e-01 -6.49054229e-01 -6.29943669e-01 2.03656688e-01
1.06262341e-01 -9.98520792e-01 1.20059121e+00 -9.96701539e-01
-1.06168151e+00 9.25644994e-01 -8.97307575e-01 -5.86600304e-02
5.95101357e-01 -5.78885376e-01 1.85585767e-02 2.67733157e-01
2.54809767e-01 7.64361382e-01 1.37245655e+00 -1.03776741e+00
-9.77704942e-01 -1.04294157e+00 5.34890056e-01 5.58463633e-01
-2.35427886e-01 2.43166283e-01 -5.38424551e-01 -2.82562912e-01
3.46879691e-01 -7.61622310e-01 1.64800197e-01 4.92679507e-01
-5.61791182e-01 -5.68099856e-01 9.36256886e-01 -6.79703534e-01
8.46623600e-01 -2.22558069e+00 6.82511926e-02 -9.54026803e-02
6.13071561e-01 2.63929516e-01 2.44659156e-01 -1.17011875e-01
-1.59159556e-01 -1.34524927e-01 2.09195226e-01 -3.29105407e-01
-3.77145529e-01 -2.63181359e-01 -2.23416016e-01 6.38633907e-01
-5.71110137e-02 9.27456737e-01 -8.92175376e-01 -2.29372516e-01
3.16597760e-01 5.01810014e-01 -5.45763910e-01 2.98856765e-01
-7.17519224e-02 7.12904036e-01 -3.16850901e-01 8.40791643e-01
4.59275872e-01 -5.61627030e-01 -2.75449574e-01 -2.51819372e-01
-1.01649992e-01 7.85417408e-02 -1.04361534e+00 1.63495517e+00
-3.69993240e-01 7.48787701e-01 2.55607022e-03 -3.67838770e-01
1.68132670e-02 2.17335343e-01 2.44365245e-01 -6.08083069e-01
1.63341999e-01 -5.42296432e-02 2.70879000e-01 -7.94710875e-01
1.98666438e-01 4.48332697e-01 5.64693987e-01 6.29294515e-01
-2.34616652e-01 5.17009616e-01 2.65398156e-02 -1.13553829e-01
8.66248131e-01 6.17928207e-02 2.43344024e-01 1.69097632e-01
3.23464841e-01 -2.89610505e-01 1.35205895e-01 6.10210598e-01
-6.01324618e-01 9.05735850e-01 4.93321776e-01 -9.04882029e-02
-9.59315062e-01 -1.06796241e+00 -4.98391092e-02 1.32142663e+00
2.42334187e-01 9.66424569e-02 -8.42559338e-01 -5.67373812e-01
-2.78826691e-02 5.49286425e-01 -6.23519480e-01 1.64557055e-01
-3.33316505e-01 -1.37342796e-01 6.45845830e-02 8.25613379e-01
5.40155649e-01 -9.90169704e-01 -1.00857508e+00 -3.10005337e-01
-2.57915407e-01 -9.30979550e-01 -9.84632194e-01 -2.13685423e-01
-5.54062784e-01 -1.47394347e+00 -1.27676153e+00 -8.38133693e-01
1.15304267e+00 8.40079427e-01 5.25077105e-01 -1.58977568e-01
-3.36367548e-01 8.15366745e-01 -1.71443447e-01 -3.20378453e-01
5.02541363e-01 5.22670373e-02 1.55486926e-01 2.84833997e-01
5.70241511e-01 -4.55944479e-01 -1.31560576e+00 2.46738970e-01
-2.10052758e-01 -1.26667693e-01 6.73434258e-01 6.68859065e-01
3.16961616e-01 -3.07922840e-01 4.25338410e-02 -2.28527546e-01
4.19952661e-01 -7.95618072e-02 -5.51756501e-01 2.81829268e-01
-3.80702227e-01 -1.33555708e-02 3.97935463e-03 -5.98049939e-01
-7.94741631e-01 2.16336370e-01 3.67889524e-01 -7.19983160e-01
-3.32385361e-01 -2.52791017e-01 -2.29081243e-01 -2.90618956e-01
7.14512289e-01 1.62202735e-02 7.67862191e-03 -3.50661933e-01
3.83490473e-01 9.51625407e-01 5.84358692e-01 -2.07018778e-01
3.40807557e-01 6.81864202e-01 -3.47343653e-01 -7.68196583e-01
-9.32447553e-01 -9.95769620e-01 -7.31465638e-01 -3.38845760e-01
9.93165255e-01 -8.89615297e-01 -1.24732316e+00 6.25681579e-01
-1.47815263e+00 -1.58130452e-01 -1.39977522e-02 7.26747155e-01
-4.23218310e-01 1.80125892e-01 1.31457776e-01 -9.70693111e-01
-2.44029701e-01 -1.43724322e+00 1.25826919e+00 6.29168212e-01
-2.36590147e-01 -5.33607960e-01 -4.66150790e-01 4.82958615e-01
2.73405135e-01 -3.71686965e-01 8.74323547e-01 -3.02404463e-01
-8.69453847e-01 -3.30167383e-01 -7.12803185e-01 -7.80588686e-02
3.63841593e-01 -3.85212064e-01 -1.34798646e+00 -3.00886244e-01
-3.34764063e-01 -5.81738986e-02 7.38388717e-01 8.81018221e-01
1.16810012e+00 -1.84972063e-01 -6.16752446e-01 6.52676582e-01
1.02851677e+00 6.36580735e-02 6.52549922e-01 2.98087239e-01
8.20671916e-01 7.81823337e-01 2.02364311e-01 2.75492996e-01
6.96525156e-01 7.45227158e-01 6.98659718e-01 8.12808722e-02
-3.23009908e-01 -6.49244562e-02 8.63666609e-02 -3.65564466e-01
-3.70385736e-01 -1.14872895e-01 -9.42318618e-01 7.54967868e-01
-1.76849675e+00 -9.23765779e-01 -7.03533143e-02 2.61300182e+00
4.23835218e-01 -3.11303020e-01 2.77035892e-01 4.73579019e-02
9.05753016e-01 -1.36738837e-01 -1.10422921e+00 3.26376855e-01
-2.19844934e-02 -1.38294414e-01 6.13879859e-01 4.24714446e-01
-1.13230348e+00 1.03750956e+00 5.71887302e+00 -1.25583832e-03
-1.24955034e+00 1.02527604e-01 1.89827338e-01 -5.38446188e-01
3.34315300e-01 2.28346922e-02 -1.00733471e+00 4.08371538e-01
2.74306476e-01 2.52550542e-01 7.66856015e-01 8.46326947e-01
4.01541471e-01 -5.76014996e-01 -1.44160342e+00 1.53014672e+00
2.49706045e-01 -7.21946001e-01 -2.27280051e-01 2.73392260e-01
2.90091306e-01 1.92982405e-01 2.81606078e-01 -3.82620752e-01
-1.17147639e-01 -1.00182223e+00 7.24339604e-01 6.38564348e-01
7.49400735e-01 -3.23990494e-01 2.43087277e-01 3.28772902e-01
-1.04824650e+00 -5.76704979e-01 -1.07518479e-01 -1.11050829e-01
-5.49851730e-02 -8.11034665e-02 -9.01484489e-01 -2.99013108e-01
1.03812039e+00 4.76012409e-01 -7.79713035e-01 1.71673858e+00
-7.01396585e-01 1.73145995e-01 -2.65715063e-01 8.19669217e-02
-1.01636037e-01 1.01319617e-02 7.74737239e-01 5.42495310e-01
2.15864256e-01 -1.49996975e-03 -2.13385135e-01 9.04425621e-01
1.71276197e-01 9.17488337e-02 -4.99988914e-01 8.01513568e-02
3.54696363e-01 9.70144153e-01 -3.57239693e-01 1.10659659e-01
-5.08504510e-01 1.26334047e+00 4.23115343e-01 8.44798684e-01
-2.88990796e-01 -4.97022003e-01 1.04481614e+00 3.59267026e-01
2.99022049e-01 -3.79820198e-01 -3.96452367e-01 -9.99941587e-01
5.22629082e-01 -5.15124679e-01 4.12851065e-01 -1.25849223e+00
-8.44313264e-01 4.12841201e-01 -5.41461706e-01 -1.41085863e+00
-3.12977582e-01 -9.54788208e-01 -5.33483267e-01 1.40128112e+00
-1.38208735e+00 -1.45282948e+00 -9.01634812e-01 8.78520846e-01
3.54584992e-01 -6.14728443e-02 5.96259177e-01 6.05767891e-02
-5.06666541e-01 7.51726449e-01 -1.93073392e-01 3.58906597e-01
1.00801694e+00 -1.19991755e+00 6.04799055e-02 9.99047697e-01
7.65655339e-02 9.30152595e-01 3.48361999e-01 -4.96517420e-01
-1.40220284e+00 -7.34064579e-01 5.98382652e-01 -8.92211497e-01
3.08388591e-01 -1.81325972e-01 -6.12189114e-01 7.52372682e-01
-8.59023333e-02 6.58735484e-02 3.85836452e-01 2.19540000e-01
-4.72327262e-01 -1.14818707e-01 -1.02136290e+00 8.53940785e-01
1.31154406e+00 -8.18885505e-01 -5.33372760e-01 4.86642033e-01
4.21199650e-01 -4.33365434e-01 -2.71761030e-01 -1.22179933e-01
7.73387969e-01 -8.91735256e-01 9.28492367e-01 -8.14787924e-01
8.85308255e-03 -3.92616838e-01 -7.27911964e-02 -1.10877752e+00
-2.04528257e-01 -3.52996826e-01 -2.81141937e-01 1.01377332e+00
2.06502363e-01 -7.42328465e-01 8.42066824e-01 9.48358238e-01
2.71802157e-01 -2.38742635e-01 -9.55424249e-01 -4.53592360e-01
-3.33112925e-01 -4.75748241e-01 3.65939558e-01 3.00055891e-01
1.31664559e-01 1.88523546e-01 -6.80887178e-02 6.44237816e-01
7.91224301e-01 8.20194632e-02 8.04284751e-01 -1.20722640e+00
1.78693347e-02 -7.46468306e-01 -7.47715950e-01 -1.37619019e+00
-1.82854682e-02 -5.70395470e-01 -1.03623234e-01 -1.92412269e+00
-2.09473688e-02 -1.86303765e-01 -2.14257732e-01 5.75168848e-01
-2.64955312e-01 2.48136725e-02 -1.17080197e-01 5.57390273e-01
-3.78224313e-01 1.34028077e-01 1.52882111e+00 -2.23772526e-01
-5.19198954e-01 4.34686571e-01 -8.98223698e-01 6.50216103e-01
5.10014534e-01 3.87160443e-02 -4.22874242e-01 -6.47926033e-01
8.80934000e-02 -3.64058167e-02 8.91727328e-01 -9.22228932e-01
8.94237161e-01 -1.12148132e-02 6.99076355e-01 -5.94262660e-01
5.15548229e-01 -8.72213781e-01 -5.10548949e-01 1.86262295e-01
-7.15314299e-02 -3.12183797e-01 2.96960752e-02 5.91247022e-01
2.17951417e-01 -2.52607733e-01 8.33007276e-01 -1.73462927e-01
-7.87311494e-01 5.07489681e-01 -2.59482693e-02 1.29904542e-02
1.07932305e+00 -3.34050953e-01 -5.46797991e-01 -4.01239574e-01
-7.28434861e-01 3.24750781e-01 6.11733615e-01 5.78659356e-01
5.25872171e-01 -7.70454645e-01 -3.98729771e-01 3.22529733e-01
5.49032331e-01 1.29557535e-01 2.89764822e-01 1.13422382e+00
-1.28857300e-01 5.98156929e-01 -9.19306502e-02 -9.09129083e-01
-1.21949351e+00 6.02147102e-01 6.90883934e-01 6.29836917e-01
-8.59294474e-01 9.79630828e-01 4.75012273e-01 2.86794156e-01
6.94788873e-01 -3.90641034e-01 -4.07268077e-01 1.63311869e-01
9.65151668e-01 3.83865297e-01 -8.15463141e-02 -6.07496262e-01
-2.82681882e-01 7.88334489e-01 -3.65528017e-01 -1.06927663e-01
9.44184065e-01 -3.95167768e-01 -1.66619748e-01 1.84567764e-01
5.80072105e-01 -1.25601143e-01 -1.66162372e+00 -4.26373720e-01
-3.21299434e-01 -8.88066232e-01 2.02963248e-01 -8.93453360e-01
-1.00634587e+00 1.16464019e+00 1.09756732e+00 -2.42477655e-01
9.78281200e-01 3.02431554e-01 3.05596709e-01 4.68035549e-01
7.01898456e-01 -8.33741903e-01 7.68047199e-02 5.60284331e-02
9.12110090e-01 -1.54078913e+00 -3.55644375e-01 -1.36861295e-01
-5.65121531e-01 1.01365960e+00 1.03060901e+00 2.71671116e-01
5.41605651e-01 -2.46505871e-01 1.76894754e-01 -1.27265394e-01
3.30946408e-02 -7.85945237e-01 4.15173292e-01 9.35958862e-01
1.44108966e-01 -6.13985993e-02 3.31378281e-01 3.87818366e-01
-9.70266461e-02 -2.53086209e-01 8.68221000e-02 7.15069652e-01
-4.81025100e-01 -7.43634343e-01 -4.75764394e-01 8.27164292e-01
-2.18909979e-01 -1.27432525e-01 -1.99821994e-01 3.95175785e-01
7.14059845e-02 8.95059943e-01 2.44883999e-01 5.29022776e-02
3.60175520e-01 2.43063703e-01 7.85857081e-01 -6.49477780e-01
-8.98582414e-02 -8.30131024e-02 -4.38522458e-01 -5.96448183e-01
-2.97100455e-01 -8.23500931e-01 -1.05982161e+00 1.37387469e-01
-2.09265724e-01 -4.53159034e-01 7.27029979e-01 1.08864713e+00
3.39963466e-01 3.52534056e-01 2.23284319e-01 -1.08176756e+00
-2.37528712e-01 -1.16378188e+00 -5.60426831e-01 5.67165576e-02
7.22945511e-01 -1.02066505e+00 -3.34361136e-01 -7.63176680e-02] | [14.070630073547363, 0.053048595786094666] |
8d6ffed0-be6b-414f-baa8-348f6d7b7d42 | unsupervised-alignment-based-iterative | 2005.01218 | null | https://arxiv.org/abs/2005.01218v1 | https://arxiv.org/pdf/2005.01218v1.pdf | Unsupervised Alignment-based Iterative Evidence Retrieval for Multi-hop Question Answering | Evidence retrieval is a critical stage of question answering (QA), necessary not only to improve performance, but also to explain the decisions of the corresponding QA method. We introduce a simple, fast, and unsupervised iterative evidence retrieval method, which relies on three ideas: (a) an unsupervised alignment approach to soft-align questions and answers with justification sentences using only GloVe embeddings, (b) an iterative process that reformulates queries focusing on terms that are not covered by existing justifications, which (c) a stopping criterion that terminates retrieval when the terms in the given question and candidate answers are covered by the retrieved justifications. Despite its simplicity, our approach outperforms all the previous methods (including supervised methods) on the evidence selection task on two datasets: MultiRC and QASC. When these evidence sentences are fed into a RoBERTa answer classification component, we achieve state-of-the-art QA performance on these two datasets. | ['Vikas Yadav', 'Mihai Surdeanu', 'Steven Bethard'] | 2020-05-04 | unsupervised-alignment-based-iterative-1 | https://aclanthology.org/2020.acl-main.414 | https://aclanthology.org/2020.acl-main.414.pdf | acl-2020-6 | ['multi-hop-question-answering'] | ['knowledge-base'] | [ 3.18702906e-01 6.62724450e-02 -2.67501622e-01 -4.41935360e-01
-1.52257586e+00 -8.12797844e-01 8.08792830e-01 7.96072423e-01
-5.16865075e-01 6.26216173e-01 6.11771464e-01 -5.89778185e-01
-5.20386755e-01 -6.34919167e-01 -6.18682921e-01 -3.40599149e-01
4.00516033e-01 1.07293022e+00 6.08024836e-01 -5.01155198e-01
6.89410448e-01 1.51904023e-04 -1.71424413e+00 4.74693537e-01
1.14332056e+00 1.06133544e+00 -1.80580884e-01 8.91889453e-01
-5.79800367e-01 1.21735561e+00 -5.90014040e-01 -8.23779643e-01
-2.03233480e-01 -5.08719802e-01 -1.47007394e+00 -3.00704986e-01
9.64350820e-01 -5.34865409e-02 -5.74822836e-02 8.62982392e-01
3.62944335e-01 1.73649609e-01 8.33293438e-01 -6.54122710e-01
-8.61562431e-01 4.94768351e-01 1.02138847e-01 3.76387417e-01
7.35640347e-01 -1.06307276e-01 1.73406518e+00 -1.04712188e+00
6.37323797e-01 9.72906709e-01 2.89645344e-01 3.25304508e-01
-8.63706410e-01 -7.28728995e-02 -2.45587364e-01 7.87519753e-01
-1.01154470e+00 -3.74279916e-01 5.76167881e-01 -2.33805448e-01
1.04176366e+00 5.22250950e-01 3.40876073e-01 6.32087052e-01
-2.37897830e-03 6.45039260e-01 9.97646391e-01 -8.48809898e-01
4.26211655e-01 1.79314896e-01 8.69419038e-01 8.81539702e-01
-6.12563901e-02 -2.73338169e-01 -5.20060956e-01 -5.85434198e-01
-2.03694493e-01 -4.69327331e-01 -3.99308741e-01 -1.30489916e-01
-8.93318832e-01 1.25253105e+00 3.68905991e-01 2.48091936e-01
-5.37422895e-01 1.40561491e-01 2.15720013e-01 6.48853362e-01
2.12583214e-01 8.92618299e-01 -5.38242042e-01 -6.11299351e-02
-7.61146545e-01 4.47167248e-01 9.95658398e-01 6.27822459e-01
8.39895308e-01 -6.60577238e-01 -5.94334543e-01 8.81796181e-01
4.72875446e-01 6.28057480e-01 5.44424355e-01 -1.06627309e+00
6.87773705e-01 1.07603145e+00 1.97712019e-01 -8.13526452e-01
-1.07899375e-01 -4.08880323e-01 -1.38320342e-01 -3.82217616e-01
2.72372335e-01 3.08617473e-01 -8.73363733e-01 1.40112758e+00
4.39369947e-01 -2.76708603e-01 3.17956239e-01 9.54209745e-01
1.39272797e+00 6.18142366e-01 -1.11265875e-01 -5.18493587e-03
1.78899598e+00 -1.29353392e+00 -9.71442878e-01 -3.36542279e-01
7.41099477e-01 -9.95938718e-01 1.31699181e+00 5.89536270e-03
-1.12399197e+00 -5.65631032e-01 -1.17888403e+00 -5.03533304e-01
-5.03303468e-01 1.05145965e-02 4.92520928e-01 5.82908034e-01
-1.00017691e+00 1.57406881e-01 1.04683653e-01 -1.82958081e-01
9.07089785e-02 1.09898731e-01 -1.94064215e-01 -5.69544673e-01
-1.44295943e+00 1.29684544e+00 6.90062419e-02 -2.42197245e-01
-7.41529286e-01 -3.92321289e-01 -1.02920115e+00 2.53499836e-01
6.17195129e-01 -9.81850505e-01 1.48170125e+00 -4.33790684e-01
-1.22484410e+00 8.67272735e-01 -5.54810762e-01 -5.59372604e-01
-7.10583385e-03 -5.08610189e-01 -4.34542835e-01 4.94462192e-01
3.06550413e-01 5.10975957e-01 8.65763903e-01 -1.00121009e+00
-4.45147097e-01 -2.65614867e-01 2.79737383e-01 3.80574316e-01
-1.59606278e-01 1.43169746e-01 -7.69410312e-01 -2.86012441e-01
3.08625668e-01 -6.29561067e-01 1.08355418e-01 -3.53513569e-01
-2.68770754e-01 -9.25320387e-01 4.66857910e-01 -8.30118239e-01
1.36654878e+00 -1.73008609e+00 3.39078218e-01 2.04392821e-01
-3.93886119e-02 2.49613807e-01 -3.87490571e-01 5.95910251e-01
1.45999447e-01 1.62209004e-01 -3.88553858e-01 -2.79877484e-01
2.11843207e-01 3.38457435e-01 -8.78591716e-01 6.45218417e-02
3.57185841e-01 1.14322853e+00 -9.62960482e-01 -8.08814168e-01
-1.07518218e-01 -4.84910011e-02 -4.71276462e-01 5.56288838e-01
-6.01709247e-01 1.57971144e-01 -4.39061433e-01 8.77633452e-01
1.86475471e-01 -5.55500805e-01 -9.23312902e-02 -1.01313487e-01
2.33263850e-01 1.16204214e+00 -8.72881651e-01 1.57743216e+00
-3.15036952e-01 4.65023607e-01 -4.19332147e-01 -8.55614543e-01
7.96633124e-01 5.25214314e-01 3.53131182e-02 -9.29616570e-01
-5.08664809e-02 6.63327992e-01 -1.51682526e-01 -8.43525767e-01
7.41344512e-01 1.48290679e-01 -3.32392901e-02 6.76551819e-01
1.06043532e-01 -6.92155778e-01 5.88604808e-01 5.52548110e-01
1.36179173e+00 -1.64416432e-01 1.44868389e-01 -1.96101014e-02
9.46829855e-01 4.12838876e-01 5.82336076e-02 1.03455627e+00
1.40797243e-01 6.18579686e-01 1.48476988e-01 -1.93467528e-01
-6.52198792e-01 -1.02064562e+00 -2.26281881e-01 1.03002584e+00
9.64639038e-02 -6.14441633e-01 -4.46113497e-01 -1.10451114e+00
5.40332794e-02 9.54450428e-01 -6.92464769e-01 -4.92843166e-02
-6.61931694e-01 -2.24354059e-01 3.64368111e-01 3.25504988e-01
4.28706199e-01 -1.14239967e+00 -5.01323879e-01 1.85904235e-01
-7.23744392e-01 -8.85721385e-01 -3.49783808e-01 2.53264010e-01
-8.03479970e-01 -1.46702087e+00 -4.69293594e-01 -7.33074248e-01
5.70496678e-01 2.53100455e-01 1.54368007e+00 7.77590394e-01
4.30773534e-02 7.87476718e-01 -5.82641423e-01 -1.59847274e-01
-4.37018692e-01 -1.63824540e-02 -4.05488372e-01 -3.05107981e-01
5.49829066e-01 8.81203413e-02 -6.11111999e-01 2.83698469e-01
-1.01263785e+00 -4.81967330e-01 5.29830515e-01 9.26360786e-01
6.26011908e-01 -3.56171221e-01 6.44475162e-01 -7.99296141e-01
9.84884202e-01 -4.37606901e-01 -4.38901424e-01 8.52982223e-01
-8.30267251e-01 3.71207088e-01 2.70363808e-01 -3.40587348e-02
-8.60459864e-01 -6.53130293e-01 -4.83015865e-01 1.48013175e-01
1.28299281e-01 9.14403200e-01 3.15100074e-01 1.84322506e-01
8.75187874e-01 -2.24547088e-02 -2.67996222e-01 -4.79521602e-01
7.87433445e-01 8.02007496e-01 4.92815137e-01 -5.91288567e-01
8.10692549e-01 2.72512347e-01 -2.22017676e-01 -4.13376421e-01
-1.35306585e+00 -9.88042772e-01 -3.52536261e-01 -1.82379261e-01
7.99134791e-01 -5.48543334e-01 -2.51407266e-01 -2.70883352e-01
-1.28454363e+00 2.77894229e-01 -4.69974101e-01 4.32979435e-01
-3.33862901e-01 5.43965459e-01 -5.23623466e-01 -7.00751364e-01
-5.58239937e-01 -1.05116320e+00 9.60142076e-01 3.37434560e-01
-5.23371756e-01 -7.95296669e-01 5.17010868e-01 1.28783262e+00
4.74724948e-01 -5.47168732e-01 1.54512823e+00 -1.08651006e+00
-6.32257283e-01 -4.32367802e-01 -1.61361396e-01 4.26315457e-01
-2.33253818e-02 -5.25584407e-02 -8.12795758e-01 3.23605575e-02
2.02649385e-01 -1.01459646e+00 9.48793292e-01 -2.48857364e-02
7.61273921e-01 -3.30768853e-01 4.35356572e-02 -7.48884454e-02
1.23916590e+00 -2.00984299e-01 7.00796902e-01 5.82288921e-01
7.83364251e-02 8.80372226e-01 9.08376098e-01 -3.03094804e-01
6.03319287e-01 6.67717636e-01 5.32195449e-01 3.00761372e-01
-4.18560177e-01 -8.07024390e-02 3.18988413e-01 1.26407814e+00
3.18497449e-01 -3.61003309e-01 -8.25758755e-01 1.02611780e+00
-1.92041135e+00 -9.06481743e-01 -5.96650124e-01 2.17001319e+00
1.08615136e+00 -3.25645953e-02 -4.31211501e-01 1.57608837e-01
2.72637278e-01 1.31571114e-01 -3.66918266e-01 -5.98686337e-01
-2.97048122e-01 8.50664914e-01 -2.31424913e-01 6.25935555e-01
-8.84231150e-01 7.50229836e-01 6.59539604e+00 7.01023400e-01
-2.91160882e-01 2.24988818e-01 3.49117696e-01 2.43203998e-01
-7.90048242e-01 3.46385509e-01 -6.75560236e-01 7.91431665e-02
9.21247363e-01 1.53684407e-01 2.09688559e-01 6.74573541e-01
-4.98209149e-01 -2.98522949e-01 -1.06012344e+00 3.34295392e-01
5.82562149e-01 -1.55351174e+00 1.95002839e-01 -4.06818151e-01
7.65955508e-01 1.51172765e-02 -7.86879584e-02 7.91011095e-01
3.11154664e-01 -1.11737037e+00 4.46422815e-01 3.05897146e-01
3.35867941e-01 -4.73241895e-01 1.20760047e+00 3.33518535e-01
-7.29788542e-01 -1.20880276e-01 -4.38270569e-01 2.18386069e-01
2.31074885e-01 5.01302838e-01 -7.56126225e-01 7.15428770e-01
7.12636888e-01 1.60207093e-01 -8.91218781e-01 1.07159638e+00
-1.14488983e+00 9.53267217e-01 -2.00904742e-01 -3.87874305e-01
3.91081959e-01 -4.83628176e-02 5.77379525e-01 1.12143898e+00
-3.93761843e-02 -1.32701367e-01 -3.05227309e-01 7.26541698e-01
-3.25465947e-01 3.64417166e-01 -1.70568258e-01 -9.78225544e-02
5.36773682e-01 1.07365942e+00 -1.20679125e-01 -5.92835605e-01
-3.36927265e-01 7.51619756e-01 4.80307788e-01 2.05423355e-01
-5.57518303e-01 -5.62710047e-01 1.25984341e-01 -4.26728517e-01
5.15135884e-01 1.30370036e-01 -1.34758845e-01 -1.08557820e+00
4.72502351e-01 -1.22006691e+00 9.75338042e-01 -9.70686078e-01
-1.40090942e+00 5.32060444e-01 -1.11066431e-01 -1.02424634e+00
-5.95286608e-01 -4.61472601e-01 -5.44387221e-01 9.08801377e-01
-2.14826584e+00 -8.39081466e-01 -2.17371002e-01 4.35153306e-01
5.15085578e-01 -4.17075269e-02 1.05627799e+00 3.10591400e-01
-4.10491563e-02 2.64662236e-01 1.97418942e-03 -2.25387830e-02
7.82698393e-01 -1.40624511e+00 3.09211239e-02 7.79454589e-01
8.03647041e-01 9.90913391e-01 5.82813740e-01 -5.18296182e-01
-1.32777941e+00 -5.34357846e-01 1.72202182e+00 -9.20804143e-01
7.06076443e-01 1.07137010e-01 -1.28887224e+00 2.75720954e-01
7.41344213e-01 -4.97371882e-01 9.67619717e-01 3.86034369e-01
-6.36028051e-01 1.06250152e-01 -9.55599010e-01 5.55305779e-01
3.65829051e-01 -9.49772596e-01 -1.73275769e+00 6.63957298e-01
8.96540880e-01 -3.00259173e-01 -6.11454189e-01 6.54221296e-01
3.55245113e-01 -5.65324068e-01 1.01632822e+00 -9.14549232e-01
6.46268666e-01 -5.79001784e-01 -3.19878161e-01 -1.05457449e+00
7.40447342e-02 -2.83627987e-01 -5.05854487e-01 1.00560760e+00
8.68398964e-01 -4.51246291e-01 4.56706464e-01 4.56938446e-01
-1.70855150e-01 -9.54460800e-01 -1.24226999e+00 -4.10391301e-01
-1.21125363e-01 -3.85623962e-01 5.92341959e-01 8.18292320e-01
-1.14212982e-01 7.01611161e-01 1.39789686e-01 4.54508588e-02
3.64419967e-01 6.46692276e-01 6.41486287e-01 -9.58208919e-01
-2.03720272e-01 -3.08878481e-01 1.15098201e-01 -1.56006670e+00
-1.28450850e-02 -9.03926611e-01 2.30090067e-01 -1.99257195e+00
2.15121940e-01 -4.02228266e-01 -3.54479879e-01 2.94332474e-01
-5.90500057e-01 1.20886192e-01 -3.05048198e-01 3.50458592e-01
-8.88347030e-01 7.31892824e-01 8.27937365e-01 -4.08615917e-01
2.47786969e-01 -2.50923812e-01 -6.96510553e-01 4.99878049e-01
4.90904897e-01 -6.65212631e-01 -3.84481132e-01 -6.97896600e-01
7.90686011e-01 -9.32703540e-02 3.86233032e-01 -6.35756254e-01
5.25385082e-01 7.92368427e-02 -1.10427752e-01 -9.87757802e-01
4.20590758e-01 -6.23345852e-01 -5.34643114e-01 4.00555968e-01
-6.35835230e-01 3.62034827e-01 -5.56680970e-02 5.52422166e-01
-5.61869919e-01 -1.12870288e+00 3.37333858e-01 1.12278186e-01
-4.96021926e-01 -2.43448585e-01 -3.30205917e-01 4.88300174e-01
2.32542530e-01 6.72434717e-02 -1.01481044e+00 -5.47849059e-01
-1.63625732e-01 5.99677861e-01 1.52649090e-01 4.89194900e-01
8.45898867e-01 -1.29152775e+00 -9.65653419e-01 -2.61796504e-01
6.18514717e-01 -3.44405212e-02 3.73102315e-02 7.73880661e-01
-4.20722485e-01 8.91371608e-01 6.20217502e-01 -5.53527415e-01
-1.39850974e+00 1.50838450e-01 -2.20762636e-03 -7.52562523e-01
-1.66064039e-01 1.09667838e+00 -5.65000713e-01 -7.02763021e-01
2.37439394e-01 -1.81195050e-01 -7.94484258e-01 1.47526458e-01
5.04907489e-01 1.95999723e-02 4.85338092e-01 -4.33941573e-01
-3.40259284e-01 6.22977018e-01 -2.13367060e-01 -3.48041058e-01
1.13269627e+00 -6.33555790e-03 -4.75062340e-01 3.58607948e-01
1.05234492e+00 4.01950151e-01 -2.82224268e-01 -5.83360732e-01
4.31970000e-01 -4.77747560e-01 -9.57067031e-03 -1.13999224e+00
-4.00249839e-01 7.97828913e-01 4.94567990e-01 2.90673435e-01
8.65641177e-01 4.65304762e-01 8.52263927e-01 9.19107974e-01
1.10408917e-01 -1.17571664e+00 5.76720059e-01 7.73328424e-01
1.11763060e+00 -1.24408805e+00 1.69175342e-01 -1.47416726e-01
-4.77528721e-01 1.06844783e+00 3.35747361e-01 2.30180129e-01
7.64257908e-02 -6.31070137e-01 2.06794292e-01 -8.85734141e-01
-1.04124308e+00 -5.15278995e-01 9.95382845e-01 3.37980911e-02
3.06125164e-01 -4.14694130e-01 -6.22324109e-01 4.08105820e-01
-1.66383818e-01 -3.00347924e-01 7.65658543e-02 1.01860893e+00
-6.96606100e-01 -1.20014477e+00 -2.38864377e-01 5.05090117e-01
-3.54586005e-01 -4.81538087e-01 -7.91739941e-01 5.86219132e-01
-2.36956716e-01 1.48679042e+00 -1.07169457e-01 -1.23010889e-01
3.77360404e-01 5.17694294e-01 3.63947272e-01 -8.30070317e-01
-1.00980544e+00 -5.24745524e-01 6.46180809e-01 -3.13837349e-01
-4.91442382e-01 -4.76988822e-01 -1.11671209e+00 1.77331939e-01
-8.30662608e-01 8.19027007e-01 8.42020214e-01 1.33143961e+00
4.00817454e-01 2.38465697e-01 3.76316071e-01 3.93273860e-01
-8.72189164e-01 -1.17811847e+00 1.79536328e-01 6.80525243e-01
1.82919517e-01 -4.50918883e-01 -6.48665488e-01 -3.32867831e-01] | [11.13028335571289, 7.958475112915039] |
49b5f6a9-b7be-4f22-b946-7208d6ad5a00 | simplex-autoencoders | 2301.06489 | null | https://arxiv.org/abs/2301.06489v1 | https://arxiv.org/pdf/2301.06489v1.pdf | Simplex Autoencoders | Synthetic data generation is increasingly important due to privacy concerns. While Autoencoder-based approaches have been widely used for this purpose, sampling from their latent spaces can be challenging. Mixture models are currently the most efficient way to sample from these spaces. In this work, we propose a new approach that models the latent space of an Autoencoder as a simplex, allowing for a novel heuristic for determining the number of components in the mixture model. This heuristic is independent of the number of classes and produces comparable results. We also introduce a sampling method based on probability mass functions, taking advantage of the compactness of the latent space. We evaluate our approaches on a synthetic dataset and demonstrate their performance on three benchmark datasets: MNIST, CIFAR-10, and Celeba. Our approach achieves an image generation FID of 4.29, 13.55, and 11.90 on the MNIST, CIFAR-10, and Celeba datasets, respectively. The best AE FID results to date on those datasets are respectively 6.3, 85.3 and 35.6 we hence substantially improve those figures (the lower is the FID the better). However, AEs are not the best performing algorithms on the concerned datasets and all FID records are currently held by GANs. While we do not perform better than GANs on CIFAR and Celeba we do manage to squeeze-out a non-negligible improvement (of 0.21) over the current GAN-held record for the MNIST dataset. | ['David Naccache', 'Aymene Mohammed Bouayed'] | 2023-01-16 | null | null | null | null | ['synthetic-data-generation', 'synthetic-data-generation'] | ['medical', 'miscellaneous'] | [-1.74643975e-02 1.32544652e-01 4.06362960e-04 -1.52917072e-01
-9.42914128e-01 -5.44787586e-01 8.05860043e-01 -1.99524850e-01
-6.81626737e-01 1.12773037e+00 -6.28923811e-03 -8.30055773e-02
1.07913718e-01 -8.78923953e-01 -8.00015688e-01 -9.78318691e-01
7.19885007e-02 5.99538743e-01 -1.53919801e-01 1.67895988e-01
-2.25095123e-01 3.64381284e-01 -1.37979019e+00 9.32447091e-02
8.57779324e-01 9.96950090e-01 -2.17185423e-01 5.54705143e-01
2.85407633e-01 4.88547772e-01 -9.48085129e-01 -6.47473276e-01
5.46204269e-01 -5.48406184e-01 -5.14929175e-01 1.26579195e-01
3.07428956e-01 -3.82560283e-01 -1.71250194e-01 9.06338036e-01
5.09609520e-01 1.01840943e-01 9.10403848e-01 -1.42248178e+00
-3.16695154e-01 7.48203635e-01 -5.42736709e-01 -1.85958326e-01
-2.33364105e-01 -5.84623441e-02 7.85158873e-01 -4.29586560e-01
4.45619404e-01 1.04955423e+00 5.05865633e-01 5.04492104e-01
-1.38206387e+00 -9.48234499e-01 -1.45220935e-01 -1.29839346e-01
-1.56757283e+00 -5.15829802e-01 5.14955461e-01 -3.67497861e-01
5.65273166e-01 2.12218627e-01 6.15876138e-01 1.40183663e+00
-1.98892429e-01 7.38627672e-01 1.39054906e+00 -4.94752586e-01
6.54829562e-01 4.42893416e-01 -1.39166430e-01 3.09491813e-01
4.10845697e-01 7.53045902e-02 -3.41183096e-01 -4.86033976e-01
7.95184433e-01 -1.50232375e-01 -3.55546445e-01 -4.38900083e-01
-1.04608297e+00 1.25786591e+00 1.71896935e-01 6.34170249e-02
-4.95199591e-01 4.46777135e-01 2.18020126e-01 9.37232226e-02
5.46291828e-01 3.18489254e-01 -1.77725628e-01 -2.18970090e-01
-1.30555689e+00 3.56463462e-01 7.70953774e-01 9.33556259e-01
5.45957744e-01 2.49911129e-01 -1.97795257e-01 8.29958975e-01
1.67654634e-01 5.10019541e-01 5.07284105e-01 -8.72688949e-01
2.04690322e-01 2.36041080e-02 1.92599744e-01 -6.91129863e-01
5.26289828e-02 -6.62658513e-01 -9.88563001e-01 3.45555186e-01
7.00876474e-01 -3.47877294e-01 -1.11979508e+00 1.84269142e+00
5.52916750e-02 1.73271880e-01 3.41826797e-01 5.55435002e-01
4.83776987e-01 7.84257829e-01 -8.15133080e-02 -7.01292530e-02
1.41389096e+00 -8.72335792e-01 -6.28333688e-01 -1.07173450e-01
1.76063642e-01 -5.78255773e-01 9.40338314e-01 7.02251792e-01
-9.49239373e-01 -3.30246180e-01 -1.18130922e+00 2.32919201e-01
-2.35809803e-01 5.14068127e-01 7.11667299e-01 9.95435119e-01
-1.07015622e+00 4.63579893e-01 -9.85368967e-01 -9.85335857e-02
7.29168117e-01 3.31171662e-01 -3.62971991e-01 5.09024858e-02
-9.66549814e-01 5.50974250e-01 5.24066448e-01 -2.97323048e-01
-9.19617712e-01 -4.41150427e-01 -4.79145169e-01 1.54312715e-01
8.57416466e-02 -5.78272760e-01 9.93261397e-01 -8.85965526e-01
-1.61315250e+00 6.25432551e-01 1.19768433e-01 -1.08047748e+00
8.04034710e-01 -4.73494641e-02 -3.35605383e-01 3.82925570e-02
-2.03296840e-01 1.08235204e+00 1.05737519e+00 -1.30785632e+00
-4.00891602e-01 1.56854764e-02 -2.22067200e-02 -1.33508638e-01
-4.10712957e-01 -2.33360499e-01 -3.47311854e-01 -8.56809378e-01
-1.18436448e-01 -1.29828477e+00 -1.30217537e-01 -2.47295782e-01
-6.22156084e-01 9.16631594e-02 4.73436564e-01 -5.12433290e-01
9.31887984e-01 -2.28603864e+00 -2.07828388e-01 2.76470214e-01
3.82217430e-02 2.41185293e-01 1.09108664e-01 1.88941598e-01
-1.94240287e-01 1.24422930e-01 -3.63953888e-01 -7.14448929e-01
-1.33202234e-02 1.21513493e-01 -4.30961519e-01 4.26499069e-01
-1.53806165e-01 6.08256519e-01 -4.91891623e-01 6.55508693e-03
-1.82305370e-02 8.26314747e-01 -6.85283005e-01 -6.24642149e-03
-9.86744165e-02 1.56594336e-01 1.85265783e-02 2.33376816e-01
7.99614608e-01 -1.80978164e-01 4.10442092e-02 -1.06076397e-01
1.51160285e-01 7.12099150e-02 -1.24167275e+00 1.46503842e+00
-3.38841051e-01 7.60251522e-01 -2.09447384e-01 -6.55951083e-01
1.06414378e+00 4.97808188e-01 1.80116698e-01 -2.01320857e-01
1.67510837e-01 9.53059196e-02 5.68117686e-02 1.96826264e-01
4.01779085e-01 -2.08993971e-01 -2.24429537e-02 5.55242300e-01
7.42957145e-02 7.79354014e-03 2.34182805e-01 9.30060297e-02
8.46869648e-01 -2.18712717e-01 2.08282456e-01 -3.38242739e-01
8.44313651e-02 -2.84565091e-01 3.23033363e-01 6.38854563e-01
6.08482882e-02 8.97488713e-01 6.40535235e-01 -2.74650812e-01
-1.06868076e+00 -1.10745013e+00 -1.95956588e-01 3.54032695e-01
-3.95110875e-01 -4.98708069e-01 -1.24535167e+00 -6.91359997e-01
-3.04071546e-01 1.17761397e+00 -6.01334572e-01 -9.13362801e-02
-4.00403142e-02 -1.13366520e+00 8.02040040e-01 3.66526276e-01
6.63135588e-01 -9.34592366e-01 -5.70037425e-01 5.30579537e-02
-2.18600616e-01 -1.08834398e+00 -3.05383682e-01 1.20917007e-01
-8.25708628e-01 -6.15531385e-01 -9.19346094e-01 -2.15916723e-01
6.96234941e-01 -1.57340989e-01 9.75564182e-01 -5.00985265e-01
-9.62571576e-02 1.41105235e-01 -1.99321136e-01 -5.78549743e-01
-6.36557281e-01 2.65504658e-01 1.42686799e-01 1.81822658e-01
1.89648226e-01 -7.71255791e-01 -5.10961115e-01 1.63871601e-01
-9.32598710e-01 4.55058664e-02 4.33214158e-01 9.58730817e-01
7.19584107e-01 2.51158446e-01 4.44372177e-01 -1.05859256e+00
5.37637830e-01 -4.58769351e-01 -7.78681278e-01 -3.95173468e-02
-7.14729846e-01 1.11962512e-01 5.75027287e-01 -5.39567530e-01
-8.11749518e-01 1.10290505e-01 -8.18317980e-02 -6.55671418e-01
-4.07248884e-01 1.11816280e-01 -8.26093704e-02 2.49509797e-01
7.21575975e-01 1.31891891e-01 -1.16165742e-01 -4.45355535e-01
3.68859559e-01 6.78046048e-01 5.03062069e-01 -3.45777452e-01
7.51065791e-01 6.66613638e-01 -2.84113139e-01 -6.48509562e-01
-5.47796011e-01 1.74780563e-02 -7.08678290e-02 3.05634052e-01
8.11068475e-01 -1.17578769e+00 -5.92467964e-01 5.07774293e-01
-8.79531980e-01 -2.06145108e-01 -6.82472587e-01 6.75776184e-01
-5.99255204e-01 6.12955168e-02 -6.34308577e-01 -8.20279360e-01
-6.14126861e-01 -1.25215244e+00 7.69685268e-01 9.03494507e-02
-3.02507818e-01 -7.23506093e-01 -5.12969168e-03 2.76080161e-01
4.39030737e-01 4.44909900e-01 6.66852355e-01 -7.82711089e-01
-5.61847329e-01 -3.27694714e-01 -1.13088088e-02 6.17528558e-01
1.90831542e-01 -1.41863823e-02 -1.42717290e+00 -4.71635520e-01
8.32899287e-02 -2.70617872e-01 9.33033705e-01 4.49428141e-01
1.29074275e+00 -4.16558504e-01 -1.71872750e-01 6.89229250e-01
1.28261828e+00 3.74290824e-01 9.78269041e-01 1.77945599e-01
4.16355669e-01 3.03080320e-01 3.42979759e-01 4.94240403e-01
2.13868901e-01 5.60537279e-01 4.19357210e-01 -1.08645126e-01
-1.39422892e-02 -2.64258206e-01 4.71168756e-01 5.85302472e-01
-8.04105327e-02 -5.18904746e-01 -6.92318797e-01 5.26428699e-01
-1.35501420e+00 -8.99208724e-01 1.23604856e-01 2.47055793e+00
8.81876051e-01 2.67392218e-01 2.59106964e-01 3.25766295e-01
4.65509087e-01 -5.54710031e-02 -3.82914931e-01 -2.65685081e-01
-1.40985861e-01 5.19129395e-01 6.83246136e-01 2.79920906e-01
-1.19574177e+00 6.82740569e-01 6.29274797e+00 1.16487682e+00
-1.16893780e+00 2.28528708e-01 1.09911096e+00 -3.07256043e-01
-2.30198562e-01 -5.01347855e-02 -8.13471913e-01 6.88600898e-01
1.09068596e+00 6.35559037e-02 6.16835356e-01 9.89088058e-01
-2.27072656e-01 -1.36178181e-01 -1.08812845e+00 1.30556917e+00
-2.54093297e-02 -1.27085447e+00 -1.07182018e-01 4.09407347e-01
8.31880987e-01 5.52278236e-02 3.00752580e-01 3.08086008e-01
4.02758121e-01 -1.39617193e+00 8.44197869e-01 3.14575613e-01
8.94612908e-01 -1.20950758e+00 6.95265472e-01 3.53740990e-01
-4.90606755e-01 2.52968699e-01 -4.81831729e-01 2.85980910e-01
-7.50958025e-02 1.00885713e+00 -8.91696274e-01 3.08975279e-01
6.48604155e-01 1.97424889e-01 -4.74423528e-01 8.99780869e-01
-1.74592018e-01 9.93800104e-01 -6.87402308e-01 7.01870620e-02
7.33228549e-02 -2.59684652e-01 4.24151897e-01 9.37383413e-01
5.75678945e-01 -2.52558112e-01 -2.43748188e-01 1.07052994e+00
-3.26892376e-01 -7.61043057e-02 -4.55552727e-01 -1.32142559e-01
5.18961012e-01 1.30793917e+00 -7.71170735e-01 -2.62178004e-01
-1.01038046e-01 9.48106587e-01 1.27411246e-01 3.90894294e-01
-1.06195045e+00 -2.11328343e-01 7.16585755e-01 -3.65185514e-02
5.76536238e-01 -9.58940387e-02 -5.48459552e-02 -1.15299857e+00
-8.04690924e-03 -1.14879358e+00 3.43332738e-01 -5.65837443e-01
-1.17446196e+00 9.99581873e-01 2.44124964e-01 -1.20737076e+00
-5.72975695e-01 -3.30671906e-01 -2.95979440e-01 8.98790359e-01
-1.07131791e+00 -9.16481078e-01 -8.95932093e-02 3.83027554e-01
1.71153098e-01 -3.86321366e-01 1.16485977e+00 2.00323313e-01
-3.93208444e-01 1.01425862e+00 4.84541178e-01 1.01630427e-01
5.61653078e-01 -1.10039651e+00 5.53745329e-01 7.44760394e-01
6.19471371e-01 5.47564089e-01 6.29052699e-01 -2.43420020e-01
-8.24817896e-01 -9.74078119e-01 3.38081151e-01 -3.42595041e-01
1.48448363e-01 -7.07547605e-01 -5.98438442e-01 8.05122554e-01
3.42740983e-01 -2.64440831e-02 7.12225914e-01 -7.16835782e-02
-3.53681803e-01 -7.34306797e-02 -1.37739062e+00 5.80232799e-01
6.04316235e-01 -3.10518652e-01 1.44018149e-02 2.21903965e-01
4.71420735e-01 -3.74875247e-01 -8.96474481e-01 1.45485550e-01
3.76578480e-01 -1.14622080e+00 9.00316358e-01 -9.38328356e-02
3.10268313e-01 -3.35848361e-01 -2.29400307e-01 -1.63056719e+00
2.98916432e-03 -6.06456101e-01 -2.26319268e-01 1.52812350e+00
4.11633790e-01 -8.45557749e-01 1.07371414e+00 4.69121963e-01
4.65465665e-01 -8.48047733e-01 -9.19519186e-01 -8.25652242e-01
1.27444580e-01 -4.37400818e-01 8.07009459e-01 9.30801988e-01
-6.25527978e-01 2.45926380e-01 -6.07316673e-01 -1.43206775e-01
8.18265378e-01 2.40117200e-02 9.41873312e-01 -9.86689866e-01
-6.51216745e-01 -2.21596465e-01 -3.02655041e-01 -7.90659010e-01
1.80458613e-02 -8.67388070e-01 -2.48804539e-01 -1.09856069e+00
1.29466102e-01 -7.03928113e-01 -1.71511069e-01 5.59567988e-01
1.51338339e-01 4.09280598e-01 2.24368677e-01 5.71330860e-02
-3.88846244e-03 7.25396693e-01 7.46793628e-01 -3.74406613e-02
3.79431099e-02 1.35142177e-01 -7.84407079e-01 6.71341062e-01
1.17591143e+00 -6.43070042e-01 -7.09101915e-01 -2.23931164e-01
-9.67476442e-02 -2.06670061e-01 3.77951503e-01 -1.25846386e+00
-1.04992852e-01 2.59370029e-01 5.95431268e-01 -4.77177501e-01
6.47093713e-01 -7.82522619e-01 7.16252983e-01 3.23540449e-01
-2.71772474e-01 -2.48546124e-01 3.51545513e-01 5.42866349e-01
-1.36089861e-01 -2.39345983e-01 9.23249483e-01 7.47917891e-02
-5.65117337e-02 3.17633182e-01 -2.97795743e-01 7.36470893e-02
9.43937600e-01 -6.83554783e-02 -1.15345202e-01 -7.55932331e-01
-5.25124729e-01 -2.38538146e-01 5.43540776e-01 2.05190480e-01
3.31446588e-01 -1.40410209e+00 -7.97998309e-01 4.17862713e-01
-7.51764923e-02 9.79833528e-02 1.59996338e-02 4.98973519e-01
-4.34004933e-01 2.25264847e-01 -1.57205120e-01 -4.82902527e-01
-1.05752802e+00 4.56114113e-01 2.98735321e-01 -5.49817443e-01
-4.34325814e-01 9.43599761e-01 2.73553610e-01 -2.86034495e-01
1.81309119e-01 -3.34706336e-01 6.05883962e-03 3.04939002e-02
4.40258116e-01 4.73821729e-01 2.39660218e-01 -4.59666669e-01
-1.13857746e-01 1.65573470e-02 -1.57476142e-01 -4.66357559e-01
1.32077992e+00 2.44937107e-01 6.71528354e-02 3.06383789e-01
1.10010457e+00 3.06194156e-01 -1.30977118e+00 1.01702131e-01
-3.67840737e-01 -4.01724488e-01 -7.81396627e-02 -7.50979066e-01
-1.36156261e+00 8.54285061e-01 8.38941872e-01 1.88320681e-01
1.27617002e+00 -2.40102828e-01 7.64227211e-01 6.74697980e-02
3.98554623e-01 -6.84125781e-01 -2.64066935e-01 2.11289376e-01
7.56758749e-01 -8.79198909e-01 4.56245989e-02 -2.68244863e-01
-6.51837528e-01 7.34806418e-01 3.52109790e-01 -1.07419349e-01
4.51031208e-01 4.74133015e-01 -7.78170452e-02 1.01165287e-01
-7.56585002e-01 2.79498875e-01 1.38536289e-01 5.13460279e-01
2.47713402e-01 2.67326355e-01 -4.70659062e-02 6.83024108e-01
-6.75994694e-01 -2.42079403e-02 6.29340768e-01 5.85909784e-01
1.20980419e-01 -1.21894276e+00 -5.51687419e-01 5.94644785e-01
-7.23356903e-01 -2.12063462e-01 -2.77028177e-02 7.48431206e-01
2.32545733e-02 7.76863575e-01 1.18491799e-01 -2.90808141e-01
3.10589150e-02 7.43874088e-02 4.86526132e-01 -2.70913333e-01
-4.88565505e-01 1.93428576e-01 7.49865994e-02 -1.87102243e-01
-2.84078598e-01 -7.26488709e-01 -9.99475121e-01 -4.62234348e-01
-4.40588742e-01 2.71868378e-01 7.80436337e-01 5.54168880e-01
3.72395337e-01 3.75112355e-01 3.80465895e-01 -5.41022956e-01
-5.64517200e-01 -9.21007156e-01 -6.62392437e-01 2.94759810e-01
3.81498225e-02 -5.63934863e-01 -4.46475208e-01 9.65161063e-03] | [11.461160659790039, -0.005925709847360849] |
9c05881a-5fd0-4713-8cf5-362ee59d18e8 | leveraging-characteristics-of-the-output | 2305.17000 | null | https://arxiv.org/abs/2305.17000v1 | https://arxiv.org/pdf/2305.17000v1.pdf | Leveraging characteristics of the output probability distribution for identifying adversarial audio examples | Adversarial attacks represent a security threat to machine learning based automatic speech recognition (ASR) systems. To prevent such attacks we propose an adversarial example detection strategy applicable to any ASR system that predicts a probability distribution over output tokens in each time step. We measure a set of characteristics of this distribution: the median, maximum, and minimum over the output probabilities, the entropy, and the Jensen-Shannon divergence of the distributions of subsequent time steps. Then, we fit a Gaussian distribution to the characteristics observed for benign data. By computing the likelihood of incoming new audio we can distinguish malicious inputs from samples from clean data with an area under the receiving operator characteristic (AUROC) higher than 0.99, which drops to 0.98 for less-quality audio. To assess the robustness of our method we build adaptive attacks. This reduces the AUROC to 0.96 but results in more noisy adversarial clips. | ['Asja Fischer', 'Dorothea Kolossa', 'Matías P. Pizarro B.'] | 2023-05-26 | null | null | null | null | ['automatic-speech-recognition'] | ['speech'] | [ 4.30040061e-01 3.40597689e-01 3.13982338e-01 -8.33457410e-02
-1.27944326e+00 -1.00459659e+00 6.38672888e-01 3.47376227e-01
-2.94532567e-01 4.24849480e-01 1.40786842e-01 -5.52709520e-01
1.52773395e-01 -4.47025359e-01 -6.21786356e-01 -6.45819724e-01
-4.67990935e-01 -5.85852452e-02 2.82061964e-01 -7.74963871e-02
-6.57597855e-02 6.87608957e-01 -1.20705736e+00 4.56162155e-01
3.62350374e-01 1.15338838e+00 -4.47198451e-01 1.17066157e+00
4.03169453e-01 5.60027063e-01 -1.51260710e+00 -5.07112026e-01
4.57693964e-01 -3.66302818e-01 -3.61245900e-01 -2.58923233e-01
1.29185081e-01 -2.89649457e-01 -8.14785004e-01 1.44128776e+00
4.19729352e-01 1.38252944e-01 6.21309042e-01 -1.27275229e+00
-3.62833261e-01 6.99854672e-01 -5.48440479e-02 3.13662887e-01
6.23618960e-01 3.19370210e-01 7.47952282e-01 -2.71619529e-01
2.50415564e-01 1.07931602e+00 4.91793334e-01 5.58371902e-01
-1.24952233e+00 -7.57481098e-01 -6.97985515e-02 -1.18865162e-01
-1.46831417e+00 -6.83759630e-01 5.96154809e-01 -1.79710597e-01
6.38640702e-01 6.26773953e-01 2.74135530e-01 1.20232761e+00
3.37657034e-01 4.71873939e-01 7.49903142e-01 -3.14867228e-01
4.57023263e-01 3.35334271e-01 -4.17776912e-01 1.52068332e-01
-2.38458499e-01 2.35919803e-01 -4.47995663e-01 -6.30764902e-01
3.99086401e-02 -2.56823033e-01 -2.66357154e-01 4.10176218e-01
-9.04011846e-01 5.09315491e-01 2.79647205e-02 3.08106124e-01
-3.34167063e-01 1.13967225e-01 5.17407775e-01 7.07427979e-01
3.03469449e-01 4.15374875e-01 -2.93133736e-01 -2.21097454e-01
-7.44456053e-01 7.78971612e-02 9.26092923e-01 5.66870689e-01
2.41508126e-01 5.25183678e-01 -2.33673632e-01 5.47730088e-01
1.45076886e-01 8.69517386e-01 4.37395990e-01 -8.15137565e-01
3.19936544e-01 -1.02055974e-01 1.42006138e-02 -7.95913458e-01
1.89778388e-01 -4.30093795e-01 -4.43448246e-01 2.07644492e-01
6.00207508e-01 -3.31736386e-01 -7.51414359e-01 1.54306412e+00
-2.12820888e-01 5.80355451e-02 4.48486567e-01 5.46030223e-01
1.85677439e-01 8.24729145e-01 -5.61846457e-02 -3.49948794e-01
8.50419641e-01 -2.78164595e-01 -6.49610102e-01 -1.66855648e-01
3.01096886e-01 -8.76567602e-01 9.78216827e-01 5.67597330e-01
-9.19916689e-01 -2.61796296e-01 -1.18416870e+00 7.85816431e-01
-2.55514890e-01 -3.51099908e-01 -1.09076299e-01 1.15839863e+00
-8.10248792e-01 6.18427396e-01 -5.80688834e-01 4.68700171e-01
-2.87283827e-02 1.86634421e-01 -3.32786471e-01 2.13747710e-01
-1.41281092e+00 4.90851313e-01 2.31882051e-01 -2.63867795e-01
-1.02537704e+00 -6.32303178e-01 -6.61198199e-01 2.43371025e-01
9.69719216e-02 3.47042650e-01 1.43236351e+00 -1.05679810e+00
-1.41897869e+00 5.26838183e-01 2.18112841e-01 -7.72956491e-01
5.11813760e-01 7.23828822e-02 -1.04765248e+00 2.23675027e-01
-4.99553949e-01 4.22275253e-02 1.26268101e+00 -1.13693523e+00
-3.28836381e-01 -1.52277574e-01 -1.57320857e-01 -1.93211496e-01
-3.96788150e-01 1.53862357e-01 -1.67357828e-02 -8.81572187e-01
-4.90177423e-02 -9.95977223e-01 -1.72283441e-01 -3.44990075e-01
-5.12439609e-01 3.11675459e-01 7.06924021e-01 -8.45236480e-01
1.31138456e+00 -2.64574790e+00 -5.81140220e-01 7.02047169e-01
-1.46688774e-01 3.61992717e-01 -8.97255242e-02 3.02064896e-01
-1.92175627e-01 2.52061635e-01 -5.12487069e-02 7.46261775e-02
8.95688236e-02 -1.70295492e-01 -8.50692034e-01 6.16611898e-01
5.34001328e-02 1.58372372e-01 -8.06620777e-01 1.47617564e-01
-1.23419687e-02 4.07476723e-01 -5.25161028e-01 3.96367967e-01
-1.68211013e-01 -4.78262827e-02 -2.03430466e-03 4.01598126e-01
3.36313456e-01 4.93187159e-01 3.51539142e-02 1.87593609e-01
3.01223755e-01 5.10986984e-01 -1.08022761e+00 7.17769146e-01
-2.84523100e-01 9.80247200e-01 1.19794488e-01 -6.83254838e-01
1.32302427e+00 7.33518004e-01 2.66762078e-01 -3.80514443e-01
1.94117010e-01 8.28667879e-02 3.63970190e-01 -4.73941714e-02
2.53958434e-01 -4.84540798e-02 -3.56738567e-01 3.47569406e-01
1.02553703e-02 -3.96234632e-01 -3.79603207e-01 4.06087995e-01
1.39230788e+00 -8.22706342e-01 3.28771360e-02 1.29724786e-01
4.08458680e-01 -5.90505540e-01 2.27918655e-01 8.80985379e-01
-3.73277366e-01 5.17976046e-01 6.55674398e-01 9.33756679e-02
-1.01736212e+00 -1.45763302e+00 3.65646109e-02 7.52053440e-01
-2.73085684e-01 -4.48602468e-01 -8.45042467e-01 -6.24834776e-01
1.90345198e-02 1.08896053e+00 -1.94388971e-01 -6.62951887e-01
-2.60604948e-01 -2.54284322e-01 1.18785703e+00 1.23491503e-01
-9.96675715e-02 -9.47546721e-01 -7.61454105e-02 4.41675223e-02
7.15478361e-02 -1.31015980e+00 -7.03058183e-01 2.91460693e-01
-2.15619460e-01 -6.46128416e-01 -3.10287803e-01 -3.38906646e-01
2.57658750e-01 -3.39151710e-01 5.79362631e-01 -3.16282153e-01
-1.86602160e-01 5.07391572e-01 -3.33002120e-01 -6.35949969e-01
-1.50742424e+00 -4.25038517e-01 4.45391983e-01 2.73423284e-01
5.27802967e-02 -5.08107364e-01 -2.59636343e-01 2.31416523e-01
-8.91191125e-01 -8.46424222e-01 1.73120484e-01 3.90064806e-01
3.96541357e-01 4.77067947e-01 6.79626346e-01 -2.85150141e-01
8.77376556e-01 -5.33414364e-01 -5.33031583e-01 1.77260488e-01
-4.21848148e-01 -6.19671419e-02 1.15129495e+00 -1.00802195e+00
-5.81177890e-01 -3.26672234e-02 -2.64352262e-01 -7.34067857e-01
-2.88863033e-01 1.03145033e-01 -3.42697710e-01 1.04687937e-01
8.37101638e-01 3.88625979e-01 5.17618246e-02 -1.82400215e-02
2.20738366e-01 1.20875347e+00 9.03140485e-01 -2.41121247e-01
1.04128885e+00 1.67513549e-01 -3.49059105e-01 -1.04866171e+00
-4.00179058e-01 -2.22360268e-01 -4.39624377e-02 -3.80327433e-01
3.03154349e-01 -4.69475657e-01 -7.66311467e-01 3.80830824e-01
-8.76043439e-01 -6.52138665e-02 -4.21605080e-01 5.86513340e-01
-5.41891396e-01 4.33677375e-01 -5.68364561e-01 -1.33413994e+00
-4.69680667e-01 -8.90030563e-01 6.40789628e-01 -1.77781861e-02
-4.94420797e-01 -6.82075977e-01 -5.47822192e-02 3.81312333e-02
3.15455079e-01 2.70302922e-01 6.16798878e-01 -1.41392493e+00
-2.03021429e-02 -9.89632249e-01 4.98527974e-01 9.33830976e-01
1.64475590e-01 3.63334358e-01 -1.28865767e+00 -3.43440920e-01
2.49596894e-01 -5.40605076e-02 4.39392507e-01 2.23456118e-02
1.15730739e+00 -8.28930676e-01 3.31283249e-02 1.93049163e-01
8.01262259e-01 8.49213958e-01 7.65494049e-01 7.28025213e-02
8.71760398e-02 3.41481984e-01 4.30712312e-01 4.99110281e-01
-5.79644620e-01 6.01549983e-01 2.63972610e-01 3.56160998e-01
3.27971846e-01 -3.56525809e-01 1.12442875e+00 5.83225489e-01
5.75179219e-01 -4.60699022e-01 -9.39398706e-01 5.37592709e-01
-1.04282176e+00 -1.40896523e+00 4.54790056e-01 2.47597909e+00
9.08254743e-01 8.58705044e-01 2.57632613e-01 8.09242606e-01
7.99427927e-01 2.64175773e-01 -4.14941192e-01 -1.00536454e+00
-4.73858081e-02 3.16784680e-01 5.76176167e-01 5.69601238e-01
-1.01722455e+00 5.88974833e-01 6.63221359e+00 9.77985919e-01
-1.15126204e+00 -3.47771972e-01 5.51034987e-01 -1.37952611e-01
-2.18372867e-01 -3.03648055e-01 -2.33037636e-01 7.00060964e-01
1.64350581e+00 -5.77692449e-01 6.63385391e-01 9.17027473e-01
1.54039130e-01 2.75534064e-01 -1.00009179e+00 7.46653140e-01
1.52036063e-02 -8.51941407e-01 -5.75494878e-02 3.16161215e-02
2.52712488e-01 -5.34673333e-02 5.87456346e-01 2.07643539e-01
3.48026335e-01 -1.09179974e+00 7.69453883e-01 3.16468865e-01
9.01463032e-01 -9.34301674e-01 5.63993394e-01 4.09854621e-01
-8.54958296e-01 -2.53019422e-01 -4.95944805e-02 1.77674085e-01
-6.43437877e-02 4.99059349e-01 -1.32602358e+00 -6.61551505e-02
4.01038498e-01 -2.13888511e-01 -1.58448458e-01 7.69419849e-01
-3.58191915e-02 1.00284803e+00 -3.72164667e-01 2.59644818e-02
3.81205305e-02 2.32328385e-01 1.00827384e+00 1.38259065e+00
2.13362083e-01 -5.80647551e-02 -7.53068784e-03 5.01126885e-01
-2.53532439e-01 -5.71680032e-02 -7.32388794e-01 -3.15936625e-01
1.00127304e+00 6.80555224e-01 -3.15336853e-01 -2.44586095e-01
9.81662422e-03 8.70583832e-01 -3.65156591e-01 2.91400790e-01
-6.74599648e-01 -7.47510314e-01 8.36277306e-01 4.01404910e-02
-1.99703267e-03 -1.40432715e-01 -1.07641622e-01 -7.01809764e-01
1.14773393e-01 -1.31432211e+00 3.79235923e-01 -3.95146400e-01
-1.04736114e+00 6.84282780e-01 -2.74470717e-01 -1.31197751e+00
-6.24033213e-01 -1.74560830e-01 -7.46075928e-01 8.32926631e-01
-6.75384104e-01 -2.04703376e-01 3.27450514e-01 5.19415379e-01
3.68888736e-01 -5.37892342e-01 1.00427175e+00 1.41491726e-01
-1.99067786e-01 1.27795589e+00 1.52377293e-01 4.37641323e-01
4.88654375e-01 -1.16649592e+00 7.98825979e-01 9.57109630e-01
3.48326147e-01 3.70218426e-01 1.06400883e+00 -4.48273361e-01
-1.01109755e+00 -9.93313909e-01 5.43882132e-01 -3.33294958e-01
8.08247685e-01 -3.30748379e-01 -1.06371331e+00 4.50610936e-01
-1.44605204e-01 -6.96493760e-02 6.29096627e-01 -3.32655162e-01
-8.11247349e-01 -1.18896134e-01 -1.41782856e+00 5.56588829e-01
1.77246690e-01 -1.04904532e+00 -4.16324556e-01 1.20047823e-01
8.25511694e-01 -2.19050378e-01 -9.98400569e-01 1.85191557e-01
6.32506728e-01 -9.26601589e-01 8.38196397e-01 -5.43369532e-01
-1.96148567e-02 -2.60743976e-01 -5.81304431e-01 -1.39172590e+00
1.77804068e-01 -1.22096992e+00 -7.99946114e-02 1.15117061e+00
8.01831067e-01 -5.20114660e-01 4.86812025e-01 5.28425813e-01
3.36946584e-02 -4.39967632e-01 -1.08672333e+00 -1.10237849e+00
2.90334545e-04 -8.86563182e-01 4.59373832e-01 7.43720472e-01
2.62133211e-01 -2.92043656e-01 -3.31418812e-01 7.14518309e-01
5.00345647e-01 -5.98636091e-01 4.68461812e-01 -6.98864639e-01
-4.30597305e-01 -3.25034887e-01 -7.60691702e-01 -5.49727261e-01
1.56524226e-01 -7.58684337e-01 1.85633466e-01 -6.45878911e-01
-3.25231940e-01 -1.75289512e-01 -4.76749897e-01 2.46836558e-01
5.46651930e-02 7.43366778e-02 3.49471152e-01 -1.83749665e-02
-2.47335941e-01 2.53458261e-01 3.50105733e-01 -4.09080476e-01
-2.72210270e-01 4.71246541e-01 -1.78168595e-01 7.80202091e-01
1.13002038e+00 -8.04586232e-01 -1.70051143e-01 2.60418504e-01
-1.56026646e-01 3.30790102e-01 7.89027214e-02 -1.25994432e+00
-1.19460709e-01 -5.37248664e-02 1.47937179e-01 -2.38867924e-01
2.88866848e-01 -8.17908943e-01 8.47542062e-02 5.60198009e-01
-8.08130324e-01 -1.25478387e-01 9.31709930e-02 5.60284317e-01
-3.15528750e-01 -1.30972117e-01 1.01326215e+00 3.33479613e-01
2.65078098e-02 -1.05351239e-01 -8.65624726e-01 6.53936565e-02
8.84511411e-01 -4.93838787e-02 -9.61858109e-02 -9.54305887e-01
-7.59651661e-01 -3.38162988e-01 2.67526001e-01 5.13956189e-01
7.39459038e-01 -9.99088585e-01 -6.99155211e-01 4.60716903e-01
7.99611304e-03 -6.68316901e-01 -2.71626133e-02 2.32420191e-01
-2.55822748e-01 1.04790188e-01 2.38121733e-01 -1.90978676e-01
-1.54240429e+00 4.96174961e-01 5.58519244e-01 1.15310796e-01
-3.70312899e-01 5.03821611e-01 -2.77906537e-01 -1.39771281e-02
6.87916994e-01 -2.73552798e-02 1.57959312e-01 -2.67917544e-01
6.87822521e-01 3.79819930e-01 2.50318944e-01 -5.88454604e-01
-3.44350159e-01 -1.65280089e-01 6.24096394e-02 -4.96540219e-01
7.26208925e-01 2.14240402e-01 3.55990648e-01 3.78646761e-01
1.41528022e+00 5.83341181e-01 -9.57471550e-01 -5.52478805e-03
-5.95852509e-02 -4.03654307e-01 -1.00500181e-01 -7.01453269e-01
-7.13114142e-01 5.99274695e-01 7.38338649e-01 9.69018817e-01
1.00606537e+00 -8.26876909e-02 8.51672173e-01 3.31165493e-01
-6.85819937e-03 -8.73541415e-01 5.25712930e-02 5.59860468e-01
7.93340921e-01 -5.97373426e-01 -2.91458666e-01 1.42278880e-01
-8.27074528e-01 1.00373232e+00 1.13233112e-01 -2.83796024e-02
8.01367462e-01 6.65768206e-01 4.67235416e-01 3.85251313e-01
-8.08405697e-01 2.34172940e-01 2.42105737e-01 7.06916511e-01
1.14795631e-02 2.71573365e-01 1.69481903e-01 6.29434764e-01
-7.31411755e-01 -6.32787347e-01 7.78363109e-01 6.10895932e-01
-4.29047942e-01 -6.97620392e-01 -6.79183483e-01 3.50547582e-01
-9.30696547e-01 -4.06234078e-02 -6.81055605e-01 2.02520669e-01
-4.48771626e-01 1.40844893e+00 2.10093200e-01 -9.12328124e-01
4.68954206e-01 2.54556149e-01 -5.28495573e-02 -2.72409111e-01
-5.45050323e-01 3.05456221e-02 2.10648715e-01 -2.17082381e-01
4.12371337e-01 -5.10782421e-01 -1.21736753e+00 -3.62736255e-01
-2.63246208e-01 2.53928512e-01 8.86719584e-01 6.27470434e-01
9.04076546e-02 4.37654823e-01 1.24752450e+00 -5.17588437e-01
-1.04046643e+00 -8.57674241e-01 -6.41234100e-01 5.21605730e-01
7.79209912e-01 1.67346865e-01 -9.29803312e-01 -1.40763307e-02] | [14.00976848602295, 5.805953502655029] |
b1fd6326-9b34-4145-b84a-7a9277c5995d | adversarially-robust-clustering-with | 2306.09977 | null | https://arxiv.org/abs/2306.09977v1 | https://arxiv.org/pdf/2306.09977v1.pdf | Adversarially robust clustering with optimality guarantees | We consider the problem of clustering data points coming from sub-Gaussian mixtures. Existing methods that provably achieve the optimal mislabeling error, such as the Lloyd algorithm, are usually vulnerable to outliers. In contrast, clustering methods seemingly robust to adversarial perturbations are not known to satisfy the optimal statistical guarantees. We propose a simple algorithm that obtains the optimal mislabeling rate even when we allow adversarial outliers to be present. Our algorithm achieves the optimal error rate in constant iterations when a weak initialization condition is satisfied. In the absence of outliers, in fixed dimensions, our theoretical guarantees are similar to that of the Lloyd algorithm. Extensive experiments on various simulated data sets are conducted to support the theoretical guarantees of our method. | ['Sanjeev Kulkarni', 'Kun Yang', 'Soham Jana'] | 2023-06-16 | null | null | null | null | ['clustering'] | ['methodology'] | [-3.48160088e-01 4.46776003e-02 -1.19919732e-01 -2.66106069e-01
-1.08328247e+00 -8.95995677e-01 4.40929443e-01 2.73106277e-01
-4.18186039e-01 5.54394901e-01 -2.60867894e-01 -1.76083356e-01
-7.17871711e-02 -4.81489956e-01 -1.03067100e+00 -9.64383066e-01
-1.73990682e-01 5.83870709e-01 1.83410510e-01 2.97690392e-01
2.23736823e-01 6.75079226e-01 -1.09775662e+00 -5.30386388e-01
1.13639855e+00 7.17024505e-01 -6.27623498e-01 7.03785360e-01
2.71903366e-01 4.05895472e-01 -8.60409915e-01 -2.55636871e-01
8.40478003e-01 -3.90357822e-01 -4.27999794e-01 6.41168892e-01
6.25098169e-01 -1.99524239e-01 -4.43707526e-01 1.60621715e+00
4.53030348e-01 4.05805469e-01 7.01707542e-01 -1.90761483e+00
-8.64191175e-01 5.79644978e-01 -6.20670140e-01 -2.52807699e-02
2.01950893e-01 -4.99202274e-02 5.92715919e-01 -7.48163164e-01
1.95801526e-01 1.19860804e+00 6.87694788e-01 6.15502298e-01
-1.36288059e+00 -7.41180599e-01 2.38317996e-01 -1.69378966e-01
-1.63966978e+00 -3.73727739e-01 7.41467655e-01 -3.08424711e-01
6.84457943e-02 2.73564517e-01 -1.58730775e-01 8.28346193e-01
-1.50710836e-01 7.69042552e-01 1.00647283e+00 -1.10185146e-01
7.81700075e-01 5.16986661e-02 1.59635350e-01 3.69011432e-01
7.27593005e-01 2.43747473e-01 -1.36367410e-01 -6.85360074e-01
4.71833318e-01 2.94233650e-01 -2.13141441e-01 -6.53046191e-01
-9.99275029e-01 9.01347041e-01 3.47626507e-01 -6.84592426e-02
-4.14236933e-02 3.27455908e-01 1.46385759e-01 4.62648243e-01
4.65325087e-01 3.24087113e-01 -2.63173312e-01 2.24573255e-01
-9.99292076e-01 3.12497467e-01 8.89640868e-01 1.37210560e+00
4.63977516e-01 1.66169971e-01 1.66492611e-01 4.14435863e-01
3.53364170e-01 8.04789543e-01 3.48721653e-01 -1.33244395e+00
2.23642364e-01 2.52049565e-01 5.86843371e-01 -9.93724406e-01
-1.45775497e-01 -2.21043080e-01 -8.88630688e-01 4.61793333e-01
6.00298822e-01 -3.80956739e-01 -8.61517251e-01 1.78748727e+00
6.25344396e-01 7.31155515e-01 1.44581527e-01 9.06618595e-01
2.13692337e-01 3.52587700e-01 -4.27899361e-01 -4.90599185e-01
6.70659065e-01 -5.52773893e-01 -7.28119910e-01 -1.26334920e-03
4.89608467e-01 -8.45527172e-01 9.35863137e-01 2.63531804e-01
-1.05295253e+00 -1.97407633e-01 -1.19403577e+00 3.95401299e-01
-5.53850196e-02 -6.37800932e-01 2.13806033e-01 1.01565337e+00
-7.11524844e-01 5.08327365e-01 -1.02608550e+00 -2.75266439e-01
3.52984369e-01 4.41042215e-01 -3.14722002e-01 -9.82505921e-03
-6.16470575e-01 4.31858450e-01 3.79032373e-01 -1.35300443e-01
-9.93596494e-01 -5.32698750e-01 -6.31972671e-01 -2.77502954e-01
5.12072682e-01 -3.77786905e-01 1.19173217e+00 -7.42519975e-01
-1.17128718e+00 6.96536839e-01 -2.44097814e-01 -6.87409937e-01
8.50291669e-01 -2.74863631e-01 -2.23178476e-01 1.78615004e-01
1.12219594e-01 6.99859187e-02 1.03585017e+00 -1.82302523e+00
-5.52523792e-01 -4.29731339e-01 -1.06940344e-01 3.23674530e-02
-2.40866527e-01 5.35630854e-03 -2.76229441e-01 -7.24525511e-01
4.32321638e-01 -1.25655532e+00 -7.27809489e-01 4.72682454e-02
-7.76264429e-01 -5.19436318e-03 9.64158654e-01 1.26989424e-01
7.22416162e-01 -2.32670903e+00 -2.22636402e-01 7.03848720e-01
2.31806651e-01 2.34586559e-02 2.57186353e-01 1.69336468e-01
-9.88422614e-03 3.84108365e-01 -3.16909522e-01 -5.62188923e-01
3.60526830e-01 4.43823248e-01 -5.17063081e-01 1.25595617e+00
-4.97293472e-01 3.74474883e-01 -8.82794142e-01 -3.23043734e-01
1.02485396e-01 -1.96091324e-01 -7.06284761e-01 3.86500150e-01
7.15144947e-02 2.77105540e-01 -3.07435930e-01 6.26541138e-01
1.05624270e+00 -1.67783588e-01 -1.21418789e-01 2.47164845e-01
4.40251291e-01 -5.03490269e-01 -1.65189779e+00 1.16538525e+00
4.58570383e-02 3.02037716e-01 4.15362954e-01 -8.50725114e-01
8.10529888e-01 4.39031333e-01 5.31077623e-01 2.30860502e-01
3.91421407e-01 2.50598401e-01 -2.82962561e-01 2.19874550e-03
3.13696027e-01 -4.37600285e-01 -3.73048186e-01 6.68029785e-01
-2.93355048e-01 -1.44510418e-01 -1.35091245e-01 3.58191252e-01
1.02213442e+00 -5.03189623e-01 2.05112007e-02 -2.45749384e-01
2.35775590e-01 -1.65007353e-01 8.86745632e-01 1.18732035e+00
-5.66223085e-01 8.70587051e-01 2.23720104e-01 -9.67218056e-02
-1.06797981e+00 -1.48959625e+00 -1.30750537e-01 5.55457473e-01
5.16130328e-01 -1.58056810e-01 -9.21133459e-01 -7.56924987e-01
3.44357222e-01 5.90783894e-01 -5.44970155e-01 -2.74506956e-01
2.69024953e-04 -8.69675398e-01 7.73522258e-01 2.66193569e-01
3.49785089e-01 -4.03959274e-01 5.73950149e-02 6.61742091e-02
2.36870334e-01 -1.32468009e+00 -8.23844731e-01 -1.79877616e-02
-9.08816159e-01 -1.24978435e+00 -3.48062932e-01 -6.50477946e-01
1.17750430e+00 3.29534739e-01 5.94470441e-01 1.97515547e-01
1.55571565e-01 4.86500531e-01 -1.81868464e-01 -3.59753281e-01
-7.89952457e-01 -2.67307997e-01 7.54069924e-01 -2.19479552e-03
4.34021801e-01 -5.72342217e-01 -4.12049055e-01 5.62842846e-01
-1.30554473e+00 -8.81986856e-01 3.54745723e-02 6.73799872e-01
6.10946894e-01 5.21359801e-01 5.16698003e-01 -9.92458820e-01
5.50677598e-01 -6.48187757e-01 -7.92637408e-01 -7.19541460e-02
-7.94745088e-01 -2.21946388e-02 1.06835175e+00 -7.40601838e-01
-4.72270221e-01 1.00388899e-01 8.67025405e-02 -9.58792150e-01
-6.79304600e-01 -1.47634268e-01 -4.91381168e-01 -1.68644622e-01
8.16731751e-01 -9.27843526e-02 3.99102177e-03 -4.11019266e-01
6.74555361e-01 6.98553205e-01 1.02316082e+00 -9.70967591e-01
1.51631963e+00 9.21832860e-01 6.11496232e-02 -5.30096769e-01
-9.11118686e-01 -5.57910860e-01 -5.95022202e-01 8.05905908e-02
4.18014705e-01 -6.15872741e-01 -7.88840950e-01 5.16602278e-01
-6.90179706e-01 1.24148866e-02 -4.61467922e-01 3.91518027e-01
-8.73005152e-01 8.99183393e-01 -5.54132998e-01 -9.62781131e-01
1.55138811e-02 -1.07183361e+00 5.73870659e-01 1.50162252e-02
-4.32286486e-02 -1.07060230e+00 -1.00439057e-01 1.16909206e-01
-6.64553344e-02 7.16404796e-01 4.56395626e-01 -1.20107102e+00
-4.79967088e-01 -6.31300807e-01 2.26007581e-01 3.72312337e-01
3.29732984e-01 1.41925111e-01 -7.99617290e-01 -6.46743417e-01
2.06404239e-01 -6.44006431e-02 4.15243506e-01 2.99605191e-01
1.18193948e+00 -7.31814623e-01 -1.53271288e-01 8.50956798e-01
1.29967475e+00 1.70180231e-01 3.43414515e-01 3.42185497e-01
5.91152787e-01 8.85259882e-02 6.21727347e-01 6.59893751e-01
-8.02288353e-02 2.38487199e-01 6.44840837e-01 4.02918980e-02
5.00927925e-01 -3.44267607e-01 3.40757638e-01 5.59519410e-01
4.47377324e-01 -2.84955919e-01 -7.34502673e-01 6.35009050e-01
-2.04930234e+00 -9.65996146e-01 -8.72341841e-02 2.60859919e+00
7.66930342e-01 3.66330743e-01 1.99084505e-01 4.19301689e-01
9.29742813e-01 -6.14732504e-02 -6.93959892e-01 -2.86742687e-01
-2.37912446e-01 -2.71115541e-01 9.27383721e-01 7.46713936e-01
-1.33060455e+00 8.85581374e-01 7.65634775e+00 6.76555634e-01
-4.95672941e-01 1.33922368e-01 5.26243687e-01 -2.63925403e-01
-1.40725777e-01 1.85133427e-01 -6.11782014e-01 6.26064241e-01
9.06960607e-01 -5.29298842e-01 4.40671444e-01 1.06751812e+00
1.97969183e-01 1.39588386e-01 -1.06032324e+00 9.71131086e-01
2.36363053e-01 -9.02114332e-01 -1.25387430e-01 4.13548015e-02
1.07128167e+00 -1.47984594e-01 2.69066751e-01 -2.19992831e-01
1.15307641e+00 -8.69859815e-01 5.48454642e-01 1.38981000e-01
2.85858244e-01 -1.15957534e+00 6.26243412e-01 6.16083384e-01
-5.27679920e-01 -8.00296143e-02 -6.37769520e-01 2.08535507e-01
2.49155477e-01 8.69868457e-01 -7.25761533e-01 3.05339575e-01
5.78986764e-01 4.17067409e-01 -3.86370510e-01 1.31747127e+00
-1.85415134e-01 8.39248121e-01 -9.16876972e-01 3.77204418e-01
1.38877138e-01 -3.26464236e-01 9.67719018e-01 8.14348042e-01
3.24047565e-01 -3.48219252e-03 5.26959300e-01 6.08168006e-01
-3.84628356e-01 5.33719361e-02 -7.57360816e-01 2.80508697e-01
9.55732465e-01 7.52292812e-01 -7.90758431e-01 -3.08200538e-01
-2.01581866e-01 1.16823304e+00 1.46135435e-01 5.85723162e-01
-8.43874931e-01 -3.70634794e-01 1.00875175e+00 -4.09107246e-02
1.90703720e-01 -4.40129608e-01 -4.32688653e-01 -1.06218362e+00
2.54279561e-02 -9.38850462e-01 6.23171568e-01 -3.91496420e-01
-1.79865599e+00 1.62325799e-01 -1.87051594e-01 -1.37469912e+00
-1.95292868e-02 -3.31061840e-01 -6.40242219e-01 2.34233230e-01
-7.86165416e-01 -6.83266044e-01 -6.38930406e-03 9.01745856e-01
8.86998326e-02 -1.93763122e-01 6.85471296e-01 1.46135436e-02
-7.05308855e-01 9.72246468e-01 7.77897060e-01 1.43122599e-01
1.05829942e+00 -1.58880985e+00 -2.96546496e-04 1.50014281e+00
1.72880247e-01 7.63721585e-01 1.12417662e+00 -4.97498959e-01
-1.05496323e+00 -1.30349207e+00 3.12169433e-01 -6.03111267e-01
8.48109007e-01 -2.20191389e-01 -8.45257282e-01 1.08048522e+00
-1.70546696e-01 3.02755415e-01 9.37589765e-01 -5.93918450e-02
-5.80459714e-01 -3.59043404e-02 -1.61173069e+00 7.48000085e-01
7.56232560e-01 -3.50150585e-01 -6.97489917e-01 7.02757239e-01
8.24825644e-01 -4.17268991e-01 -1.00630748e+00 1.40174821e-01
-1.12508737e-01 -7.30416358e-01 1.01345527e+00 -8.81166637e-01
-2.43761048e-01 -5.99942207e-01 -4.60768729e-01 -1.34627450e+00
-1.69471696e-01 -1.19091129e+00 -3.27025145e-01 1.17161453e+00
1.08130731e-01 -7.65179098e-01 8.31507266e-01 9.60704803e-01
2.03263015e-01 -1.53115958e-01 -1.23802865e+00 -1.42656338e+00
3.92052859e-01 -6.58316791e-01 7.94817030e-01 1.08400786e+00
2.39242055e-02 -2.59113759e-01 -5.42876661e-01 9.49500382e-01
1.27160072e+00 9.26958993e-02 1.14402819e+00 -1.12586033e+00
-1.02712810e-01 -3.89386833e-01 -6.16639733e-01 -1.03525472e+00
5.30708373e-01 -7.91070879e-01 3.23038071e-01 -1.05617309e+00
2.30801478e-02 -6.51279747e-01 -3.49721730e-01 1.98581859e-01
-4.97401744e-01 2.42404968e-01 2.23991290e-01 3.74480486e-01
-9.15816665e-01 4.67296064e-01 7.09507406e-01 2.71098241e-02
-8.19700360e-02 3.89505029e-01 -7.98411369e-01 1.12150693e+00
9.24126685e-01 -8.15258324e-01 -3.72403502e-01 -9.60185900e-02
-2.29941234e-01 -3.34844738e-01 1.31652802e-01 -9.88852561e-01
4.78013426e-01 -4.27649707e-01 1.00364424e-01 -5.77456713e-01
1.45787999e-01 -1.26443839e+00 3.69774550e-02 4.14380282e-01
-2.67809242e-01 6.37791082e-02 -1.20779701e-01 1.02086723e+00
9.87627208e-02 -2.57910788e-01 1.12173784e+00 3.11635643e-01
-4.44614561e-03 6.87628508e-01 -5.48409164e-01 4.60021049e-01
1.47339714e+00 2.70820595e-02 -2.44511604e-01 -7.52864122e-01
-9.05604243e-01 3.53509665e-01 1.09113407e+00 1.35013655e-01
5.61264038e-01 -1.57144916e+00 -6.66999817e-01 1.26902461e-01
-3.43987718e-02 2.07071841e-01 -2.15381056e-01 6.02357805e-01
-3.82271230e-01 -1.16090216e-01 4.12633598e-01 -4.93682355e-01
-1.22485971e+00 1.27367890e+00 3.79101932e-01 3.09277534e-01
-5.23222208e-01 6.50021076e-01 -1.75612986e-01 -5.71665645e-01
4.48229223e-01 -1.06604353e-01 7.41418421e-01 -4.46853369e-01
5.43537796e-01 6.80089772e-01 -1.88853100e-01 -6.67644978e-01
-2.82537550e-01 2.43012100e-01 8.27496797e-02 -2.21659034e-01
9.47865784e-01 -3.37896734e-01 3.09567936e-02 4.37543601e-01
1.14476526e+00 4.69664007e-01 -1.24626338e+00 -4.32215333e-01
1.22088708e-01 -9.18712616e-01 -3.65180224e-01 9.22574848e-03
-1.15847576e+00 5.35511613e-01 4.49852973e-01 4.55306113e-01
9.12542820e-01 1.20923063e-02 8.22639287e-01 5.33290505e-01
5.03958881e-01 -1.11338854e+00 -8.05991963e-02 1.23131253e-01
3.75853449e-01 -1.23442900e+00 -8.09931569e-03 -3.51845622e-01
-4.56390232e-01 6.26628041e-01 4.71014619e-01 -7.01201439e-01
8.40599179e-01 5.51196635e-01 3.51474315e-01 2.09847793e-01
-4.94845122e-01 -4.11326140e-02 -8.09677765e-02 7.48763680e-01
-2.67185926e-01 1.70127839e-01 2.27379147e-03 5.37637115e-01
-2.68177658e-01 -3.44542533e-01 8.95714462e-01 8.94476056e-01
-6.87965333e-01 -9.13615644e-01 -1.13753748e+00 2.80002266e-01
-8.81804705e-01 2.92341888e-01 -3.74001503e-01 5.83385527e-01
-2.01315776e-01 1.42163694e+00 4.21172492e-02 -1.38092697e-01
2.92504698e-01 1.66779477e-02 1.06469072e-01 -3.38331580e-01
-1.49878442e-01 1.76782340e-01 -5.28904140e-01 -5.53040981e-01
-4.11223412e-01 -7.62792110e-01 -1.45316195e+00 -9.37473416e-01
-3.52902412e-01 5.26884317e-01 1.60869718e-01 8.88034523e-01
1.00610241e-01 -1.05984420e-01 1.02747571e+00 -3.78526807e-01
-1.00010252e+00 -5.70408523e-01 -1.05181527e+00 9.19996381e-01
4.85873938e-01 -4.59480554e-01 -1.03923237e+00 1.29394248e-01] | [5.859010219573975, 7.483976364135742] |
3d768155-ba92-4d69-ba42-a69b2cd8ce42 | next-best-view-regression-using-a-3d | 2101.09397 | null | https://arxiv.org/abs/2101.09397v1 | https://arxiv.org/pdf/2101.09397v1.pdf | Next-best-view Regression using a 3D Convolutional Neural Network | Automated three-dimensional (3D) object reconstruction is the task of building a geometric representation of a physical object by means of sensing its surface. Even though new single view reconstruction techniques can predict the surface, they lead to incomplete models, specially, for non commons objects such as antique objects or art sculptures. Therefore, to achieve the task's goals, it is essential to automatically determine the locations where the sensor will be placed so that the surface will be completely observed. This problem is known as the next-best-view problem. In this paper, we propose a data-driven approach to address the problem. The proposed approach trains a 3D convolutional neural network (3D CNN) with previous reconstructions in order to regress the \btxt{position of the} next-best-view. To the best of our knowledge, this is one of the first works that directly infers the next-best-view in a continuous space using a data-driven approach for the 3D object reconstruction task. We have validated the proposed approach making use of two groups of experiments. In the first group, several variants of the proposed architecture are analyzed. Predicted next-best-views were observed to be closely positioned to the ground truth. In the second group of experiments, the proposed approach is requested to reconstruct several unseen objects, namely, objects not considered by the 3D CNN during training nor validation. Coverage percentages of up to 90 \% were observed. With respect to current state-of-the-art methods, the proposed approach improves the performance of previous next-best-view classification approaches and it is quite fast in running time (3 frames per second), given that it does not compute the expensive ray tracing required by previous information metrics. | ['Rafael Murrieta-Cid', 'Enrique Sucar', 'Israel Becerra', 'David Troncoso', 'J. Irving Vasquez-Gomez'] | 2021-01-23 | null | null | null | null | ['3d-object-reconstruction'] | ['computer-vision'] | [ 3.59224111e-01 3.97714049e-01 2.27055520e-01 -3.91158670e-01
-6.02007627e-01 -3.43554467e-01 6.76891744e-01 5.86574301e-02
-7.15481192e-02 3.19487512e-01 -3.79113287e-01 -2.24091649e-01
-2.00145289e-01 -9.97925460e-01 -1.02728426e+00 -4.35046554e-01
1.61262870e-01 9.43532646e-01 6.65327728e-01 -3.18212174e-02
3.87858152e-01 1.14413953e+00 -2.01641273e+00 3.69185925e-01
1.47419095e-01 1.48258173e+00 4.75787163e-01 5.83417058e-01
-3.41220163e-02 4.96002525e-01 -1.49421990e-01 2.57330239e-02
3.15132737e-01 -1.81863695e-01 -5.81638694e-01 3.76601696e-01
5.31346619e-01 -5.18192351e-01 6.54982179e-02 7.18606889e-01
1.53785810e-01 -1.68901026e-01 7.63898551e-01 -7.93077230e-01
2.26670243e-02 4.63381857e-02 -3.83628428e-01 -8.84731710e-02
4.86030728e-01 -2.13611975e-01 7.77795792e-01 -1.09739625e+00
8.50781500e-01 9.59789813e-01 5.03151119e-01 5.19179344e-01
-1.10471082e+00 -2.20492318e-01 1.31153464e-01 1.28873587e-02
-1.13923264e+00 -3.20682228e-01 1.29005075e+00 -6.80197597e-01
7.36688793e-01 1.77319542e-01 6.14219725e-01 1.06286454e+00
1.64753556e-01 5.79699755e-01 1.29354274e+00 -5.72595835e-01
4.25832927e-01 2.15639189e-01 -4.85010557e-02 7.66363502e-01
1.23851009e-01 1.64883763e-01 -3.40318978e-01 6.31452128e-02
8.05038095e-01 1.89259931e-01 -2.83823729e-01 -8.78214359e-01
-1.05577588e+00 4.68373626e-01 4.57559377e-01 5.19321620e-01
-5.86692572e-01 8.95459652e-02 -1.17289394e-01 -1.63156986e-01
6.96073830e-01 2.03473061e-01 -3.97054672e-01 1.37683958e-01
-8.53481412e-01 1.86730698e-01 7.71215439e-01 7.99764335e-01
8.32971394e-01 -2.69384999e-02 4.07089621e-01 3.91860545e-01
4.21559751e-01 4.55359071e-01 4.17946875e-02 -6.81668103e-01
5.34963906e-01 7.97679305e-01 3.93061489e-01 -9.32819128e-01
-5.87313890e-01 -4.66132998e-01 -6.10650897e-01 6.76521301e-01
4.48287845e-01 2.24310860e-01 -9.23380792e-01 1.15304995e+00
5.76566279e-01 1.08548760e-01 -4.78430130e-02 9.58391726e-01
6.85532153e-01 4.59202200e-01 -7.56860137e-01 -1.37791574e-01
1.16109455e+00 -4.93671328e-01 -2.31976062e-01 -1.71293765e-01
1.71038046e-01 -7.50007093e-01 6.91397548e-01 8.07294726e-01
-1.14184260e+00 -7.56474495e-01 -1.15954363e+00 3.73926401e-01
-2.28087440e-01 7.32345879e-02 2.65528798e-01 3.86588871e-01
-7.81806350e-01 7.93850243e-01 -1.04651141e+00 -2.96334982e-01
4.67800438e-01 1.83107942e-01 -4.25907165e-01 -2.37864181e-01
-5.47553062e-01 8.89144063e-01 3.35534394e-01 1.88898727e-01
-1.21804619e+00 -5.16433775e-01 -5.08345604e-01 -5.59464134e-02
4.60546851e-01 -5.85645080e-01 1.06897557e+00 -7.89705932e-01
-1.49545586e+00 9.76938307e-01 -7.17376173e-02 -2.50306875e-01
7.19147325e-01 -2.62475222e-01 -1.95543855e-01 2.31201172e-01
-1.49277285e-01 1.84429973e-01 7.72128940e-01 -1.98839819e+00
-5.38332999e-01 -8.08744967e-01 3.08852375e-01 1.74111519e-02
1.11251935e-01 -5.31606078e-01 -5.32956004e-01 -2.58403689e-01
7.81636059e-01 -8.80230904e-01 -1.03725359e-01 2.37627000e-01
-4.28363353e-01 4.00315262e-02 9.09558833e-01 -4.72709417e-01
4.94367748e-01 -1.96592700e+00 2.30751187e-01 2.22869888e-01
2.16135290e-02 1.09307587e-01 3.27657968e-01 4.44856435e-01
2.68753599e-02 -1.33929312e-01 -2.03477100e-01 -7.11638868e-01
-3.08937609e-01 1.63024053e-01 -1.03875190e-01 6.97314501e-01
-2.07208142e-01 4.43467200e-01 -4.29954559e-01 -2.65550375e-01
6.72243297e-01 7.01779783e-01 -3.16919535e-01 4.19611096e-01
-6.81921780e-01 7.04422891e-01 -6.23690486e-01 5.70912778e-01
6.94399595e-01 -2.15292841e-01 -2.33741496e-02 -2.72853702e-01
-2.28930727e-01 9.12839994e-02 -1.37813914e+00 1.79507422e+00
-6.93664432e-01 2.63185978e-01 5.98958917e-02 -1.09445000e+00
1.17891705e+00 4.68340546e-01 7.75428951e-01 -6.31634772e-01
1.82703391e-01 4.39256847e-01 -4.61769640e-01 -4.69750494e-01
1.86272219e-01 -3.73638749e-01 2.02138394e-01 4.09646332e-01
-2.09638029e-01 -2.37623841e-01 -5.35268962e-01 -4.57332402e-01
8.19369853e-01 5.94782233e-01 3.99002403e-01 3.50241251e-02
6.04939818e-01 -4.45019118e-02 1.87331006e-01 4.33082640e-01
3.98249090e-01 9.50100541e-01 1.26944199e-01 -7.51402259e-01
-1.03376222e+00 -1.03242910e+00 -1.34730980e-01 2.71722257e-01
4.43066597e-01 2.10911766e-01 -7.11711407e-01 -7.27321923e-01
-1.11227803e-01 7.03451693e-01 -5.97653091e-01 2.89356560e-01
-6.57314599e-01 -1.28749669e-01 5.05800024e-02 3.28630477e-01
3.50758165e-01 -1.04464316e+00 -1.23317480e+00 2.15940922e-01
2.91496478e-02 -1.27976930e+00 4.16074783e-01 1.94289953e-01
-1.22098303e+00 -1.39323974e+00 -6.70315564e-01 -4.16683435e-01
6.52658522e-01 1.79903731e-01 1.20016491e+00 -3.83377671e-02
6.80179894e-02 5.12411594e-01 -4.75047916e-01 -5.69851398e-01
-5.28125048e-01 -2.72065736e-02 -1.35814667e-01 2.83251792e-01
-5.13991192e-02 -8.59008551e-01 -7.17397809e-01 4.76909608e-01
-8.97718728e-01 1.64756730e-01 5.21195889e-01 5.29280543e-01
1.10203266e+00 1.61488876e-01 1.79658115e-01 -9.06767488e-01
-2.66521633e-01 -2.83678919e-01 -9.98347104e-01 1.53376982e-01
-4.86284226e-01 1.09728508e-01 8.03855896e-01 -9.96535793e-02
-1.04911077e+00 4.95585501e-01 -3.46829146e-01 -7.26693273e-01
-6.91207290e-01 1.89099357e-01 -3.17386061e-01 3.47527154e-02
5.01473129e-01 3.31703991e-01 -2.30980992e-01 -8.85365546e-01
-7.29902238e-02 5.14370799e-01 2.00988919e-01 -2.29285836e-01
7.51485884e-01 9.69113648e-01 4.11634505e-01 -8.26614022e-01
-1.09585845e+00 -3.95297557e-01 -9.08937335e-01 -6.20523870e-01
9.26691115e-01 -5.79925358e-01 -7.48357892e-01 3.28353763e-01
-1.32328224e+00 -1.23124331e-01 -2.57673323e-01 6.20269716e-01
-8.87641788e-01 2.09555656e-01 5.85947633e-02 -1.06525242e+00
-9.26968008e-02 -1.08161020e+00 1.34994090e+00 -1.63504109e-01
1.75863445e-01 -8.11393142e-01 -1.05504684e-01 5.29913783e-01
5.69904111e-02 6.39911711e-01 8.31999958e-01 -4.61073935e-01
-9.77818012e-01 -2.28460148e-01 1.57431245e-01 3.27330828e-01
-1.97111014e-02 -3.26215804e-01 -1.34019136e+00 -1.68262854e-01
4.91334975e-01 -5.71057722e-02 5.87903202e-01 3.25099796e-01
1.20621419e+00 8.84729263e-04 -4.72885132e-01 4.76943552e-01
1.88291562e+00 3.37231934e-01 5.19456923e-01 2.31637731e-01
5.96673906e-01 5.22243321e-01 8.30462217e-01 5.34220099e-01
2.40225121e-01 1.16061628e+00 1.28153634e+00 1.14143465e-03
-1.14733890e-01 -2.98203319e-01 9.38597172e-02 5.30962884e-01
-2.81584620e-01 -5.53107023e-01 -8.54935646e-01 4.44756776e-01
-1.43395495e+00 -8.01037192e-01 -3.55206460e-01 2.27682233e+00
-6.71613216e-02 5.64623415e-01 -1.72883898e-01 5.47469139e-01
2.38293812e-01 1.60405248e-01 -5.72123349e-01 -1.99514434e-01
3.48936021e-01 2.15767547e-01 3.28648746e-01 5.89723647e-01
-9.24744189e-01 4.07621920e-01 4.74778748e+00 3.70630682e-01
-1.21009588e+00 1.03800379e-01 3.61944407e-01 1.85817391e-01
-3.47229898e-01 7.12508801e-03 -8.96496177e-01 1.68510079e-01
4.60015506e-01 7.75176942e-01 2.15422213e-01 1.02953458e+00
1.04348280e-01 -3.88474673e-01 -1.48427618e+00 9.67312753e-01
4.32272434e-01 -1.24086821e+00 -1.94666222e-01 1.89526215e-01
6.19939387e-01 1.53168905e-04 -2.66087323e-01 -3.59013438e-01
-2.35534474e-01 -8.54270339e-01 1.01791680e+00 9.42645609e-01
6.83003962e-01 -5.91571391e-01 8.30112338e-01 1.00224674e+00
-1.12217510e+00 1.36760965e-01 -7.24646151e-02 3.92574258e-02
4.23741877e-01 8.17902803e-01 -1.01330078e+00 8.74085844e-01
6.06094301e-01 4.63421017e-01 -3.04177135e-01 7.51555622e-01
-1.70863699e-02 4.03333038e-01 -3.84147584e-01 6.46290705e-02
8.41477588e-02 1.00634642e-01 6.60096645e-01 5.77699244e-01
6.44502342e-01 1.11454800e-01 1.14574403e-01 9.25757289e-01
1.62758455e-01 -1.82939395e-01 -9.41522598e-01 5.41672468e-01
5.08621968e-02 8.33015442e-01 -9.76989746e-01 -1.24687143e-02
-2.85254002e-01 8.32378507e-01 2.50350505e-01 3.24495658e-02
-6.76695883e-01 1.50059134e-01 3.93839777e-02 7.88372159e-01
6.84648275e-01 -2.65356511e-01 -2.99060941e-01 -8.50285292e-01
3.69489223e-01 -3.53994370e-01 6.28411770e-02 -9.87331331e-01
-1.07522750e+00 7.27374792e-01 5.89140356e-02 -1.61292362e+00
-3.22242260e-01 -6.91018760e-01 -1.16081508e-02 6.51804507e-01
-1.46189761e+00 -1.22261167e+00 -5.14792562e-01 4.21385169e-01
7.78713465e-01 1.87694468e-02 9.93515551e-01 1.84171483e-01
1.71098396e-01 -4.45847809e-02 1.89115088e-02 -2.60268748e-01
1.73898563e-02 -9.25388038e-01 1.02068223e-01 6.52012408e-01
4.03501540e-01 -1.45409286e-01 7.23663092e-01 -5.32539845e-01
-1.45828342e+00 -8.58063161e-01 6.59838915e-01 -4.87232625e-01
1.79951191e-02 -5.31700432e-01 -7.28819251e-01 6.54484153e-01
-4.29078698e-01 3.51987332e-01 -3.25241475e-03 -2.25468844e-01
-3.19720656e-02 -3.46890777e-01 -1.21867764e+00 5.92327490e-02
1.02029276e+00 -3.03724527e-01 -4.90123034e-01 1.69290945e-01
2.66597837e-01 -6.75076127e-01 -9.29847240e-01 6.51472330e-01
5.75267553e-01 -1.58010280e+00 1.03752410e+00 -8.20544511e-02
5.47320604e-01 -2.87715524e-01 -4.72599149e-01 -1.09220016e+00
1.87008634e-01 -4.16168869e-02 -3.97503167e-01 8.95077646e-01
2.05620319e-01 -3.78407240e-01 1.16279817e+00 6.81392252e-02
-4.00734365e-01 -1.00109351e+00 -1.16157866e+00 -6.56679273e-01
-3.34446639e-01 -7.43634224e-01 5.25422454e-01 5.65473974e-01
-8.71702552e-01 9.45220962e-02 -2.42768332e-01 6.86115026e-01
7.18640566e-01 6.90257251e-01 9.13305998e-01 -1.58006895e+00
-4.97458994e-01 -1.93503611e-02 -7.06082582e-01 -1.26245856e+00
-8.94886330e-02 -7.24256575e-01 5.22093587e-02 -1.75901914e+00
-2.03057081e-01 -6.55245125e-01 7.89568108e-03 -1.62344370e-02
5.23245573e-01 1.64482623e-01 3.51641849e-02 1.27207655e-02
-2.75044084e-01 3.77841860e-01 1.31181240e+00 1.43269062e-01
-6.91995099e-02 4.99160975e-01 -1.35659561e-01 1.00843930e+00
5.30941784e-01 -5.35214126e-01 -4.26310390e-01 -4.90489155e-01
2.09255159e-01 6.62708104e-01 7.35869765e-01 -1.39477456e+00
1.19974499e-03 -5.68842962e-02 4.07439858e-01 -1.26159656e+00
9.67359543e-01 -1.57398212e+00 5.78318596e-01 4.68818635e-01
-1.97761990e-02 -1.79991305e-01 -1.28019631e-01 5.80690384e-01
-8.27924609e-02 -3.72209519e-01 9.27343369e-01 -4.70361501e-01
-5.08173943e-01 1.83605626e-01 -3.74987610e-02 -3.75088602e-01
1.33260262e+00 -6.21211231e-01 2.36739919e-01 -2.89312273e-01
-8.50432396e-01 -3.96406233e-01 4.88867253e-01 2.16438010e-01
9.80826139e-01 -1.05047011e+00 -5.16202450e-01 2.95772940e-01
1.65569320e-01 6.05836689e-01 2.99598217e-01 5.45787156e-01
-6.10518932e-01 4.55605298e-01 -1.21559724e-02 -1.00388360e+00
-1.20959783e+00 7.56069303e-01 5.50794125e-01 -2.11033151e-01
-9.69196200e-01 4.71590698e-01 1.22858740e-01 -3.51438940e-01
2.15442300e-01 -3.97013813e-01 -2.22927317e-01 -1.35272026e-01
1.18100092e-01 3.48988056e-01 5.16757667e-01 -7.59573638e-01
-1.86755136e-01 9.16532457e-01 2.91379005e-01 -3.44431140e-02
1.70621228e+00 5.68535961e-02 2.03180388e-01 7.98784554e-01
1.17655540e+00 -1.29019925e-02 -1.23661578e+00 -7.15754777e-02
-9.22575444e-02 -4.78969038e-01 1.67146195e-02 -7.31693447e-01
-1.11456287e+00 1.15001798e+00 8.94118547e-01 2.40274146e-01
1.01351047e+00 2.87798643e-01 4.60304528e-01 2.73007095e-01
9.07655358e-01 -6.73500419e-01 1.14345476e-01 1.83842763e-01
9.03739870e-01 -1.11530972e+00 1.80923715e-01 -5.72045565e-01
-9.73954499e-02 1.22772336e+00 5.85955918e-01 -1.26027226e-01
9.17056143e-01 1.13260388e-01 -1.61384493e-01 -7.42613196e-01
-3.47974807e-01 5.30434726e-03 3.12676013e-01 4.69752043e-01
5.78226410e-02 -1.10717202e-02 2.01708630e-01 2.75535166e-01
-7.10717812e-02 7.07096606e-02 3.69841754e-01 8.51263642e-01
-4.19004947e-01 -8.63408267e-01 -5.90901017e-01 2.51909733e-01
-3.10775191e-01 5.17174244e-01 -2.66491026e-01 1.00160408e+00
3.72604340e-01 7.49197125e-01 3.02047301e-02 -2.70029008e-01
7.56468534e-01 -8.13989639e-02 5.86852849e-01 -6.10231817e-01
-2.64748454e-01 3.87785621e-02 3.36644314e-02 -7.20520556e-01
-8.01821589e-01 -6.40206218e-01 -1.13476336e+00 2.10261434e-01
-4.08790261e-01 -2.95798540e-01 9.28900719e-01 1.01075554e+00
1.05967782e-01 4.19572234e-01 9.62579966e-01 -1.26303077e+00
-3.97664309e-01 -7.08843172e-01 -5.46582162e-01 4.63771522e-01
2.88161039e-01 -9.28518534e-01 -1.66954666e-01 -4.55425158e-02] | [8.028343200683594, -2.8297412395477295] |
b76eb1d4-262a-46a1-82cf-6e38bdb1fb38 | neighborhood-normalization-for-robust | null | null | http://openaccess.thecvf.com//content/CVPR2021/html/Liu_Neighborhood_Normalization_for_Robust_Geometric_Feature_Learning_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Liu_Neighborhood_Normalization_for_Robust_Geometric_Feature_Learning_CVPR_2021_paper.pdf | Neighborhood Normalization for Robust Geometric Feature Learning | Extracting geometric features from 3D models is a common first step in applications such as 3D registration, tracking, and scene flow estimation. Many hand-crafted and learning-based methods aim to produce consistent and distinguishable geometric features for 3D models with partial overlap. These methods work well in cases where the point density and scale of the overlapping 3D objects are similar, but struggle in applications where 3D data are obtained independently with unknown global scale and scene overlap. Unfortunately, instances of this resolution mismatch are common in practice, e.g., when aligning data from multiple sensors. In this work, we introduce a new normalization technique, Batch-Neighborhood Normalization, aiming to improve robustness to mean-std variation of local feature distributions that presumably can happen in samples with varying point density. We empirically demonstrate that the presented normalization method's performance compares favorably to comparison methods in indoor and outdoor environments, and on a clinical dataset, on common point registration benchmarks in both standard and, particularly, resolution-mismatch settings. The source code and clinical dataset are available at https://github.com/lppllppl920/NeighborhoodNormalization-Pytorch. | ['Mathias Unberath', 'Russell H. Taylor', 'Gregory D. Hager', 'Masaru Ishii', 'Ayushi Sinha', 'Benjamin D. Killeen', 'Xingtong Liu'] | 2021-06-19 | null | null | null | cvpr-2021-1 | ['scene-flow-estimation'] | ['computer-vision'] | [ 3.87770385e-02 -2.99658179e-01 -1.05750360e-01 -1.96410596e-01
-8.96200776e-01 -4.58594292e-01 5.33240438e-01 7.18187094e-01
-3.87567252e-01 4.88349676e-01 2.05582663e-01 -8.59079808e-02
-2.09679022e-01 -5.02943635e-01 -4.80558932e-01 -6.79979861e-01
-2.55359411e-01 7.38869667e-01 3.09372157e-01 -7.65041187e-02
9.28516015e-02 1.08642209e+00 -1.25726604e+00 -7.81874582e-02
5.69710493e-01 7.41076231e-01 4.62774485e-02 7.01612413e-01
-3.06601916e-02 1.86925679e-01 -6.28315032e-01 7.28199556e-02
5.51024258e-01 -2.07311586e-01 -4.83072579e-01 1.14229806e-01
7.25794375e-01 -3.70095558e-02 -2.72812009e-01 7.90975869e-01
7.91475058e-01 2.26406306e-01 7.94011414e-01 -1.06233835e+00
1.39115050e-01 -1.81291506e-01 -6.43199503e-01 2.81661302e-01
4.90149379e-01 1.37967035e-01 3.63429815e-01 -8.53867412e-01
7.13011205e-01 1.09554601e+00 1.16329598e+00 2.64701545e-01
-1.35076499e+00 -3.92932802e-01 -1.90227076e-01 -1.39505312e-01
-1.67983592e+00 -3.83552700e-01 6.85809374e-01 -5.99989116e-01
6.90227568e-01 3.61951649e-01 4.24184531e-01 6.87953472e-01
5.44862866e-01 2.42601484e-01 8.96439672e-01 -2.73983300e-01
1.84742004e-01 9.84349698e-02 -9.99304354e-02 3.30109388e-01
5.00131965e-01 8.46141353e-02 -2.95499086e-01 -5.73556960e-01
1.22471774e+00 2.09692135e-01 -4.31367308e-01 -8.88048887e-01
-1.52506733e+00 6.11155808e-01 4.99664903e-01 4.70323801e-01
-4.18100446e-01 -3.86837460e-02 2.23316655e-01 1.20605782e-01
5.03248513e-01 3.37866932e-01 -2.57373661e-01 -2.35135108e-01
-8.38404357e-01 3.37580860e-01 6.14259064e-01 1.17906260e+00
5.44344246e-01 -3.94123554e-01 -7.13368505e-02 5.10761321e-01
1.54426694e-01 6.75564229e-01 3.96561295e-01 -8.28407645e-01
3.33491981e-01 4.32018727e-01 2.15446830e-01 -1.15379953e+00
-9.00828063e-01 -4.26741421e-01 -1.07244265e+00 3.27124834e-01
7.71641135e-01 1.65699482e-01 -8.23789001e-01 1.45859027e+00
7.35958040e-01 5.43303788e-01 -1.60088822e-01 9.84222353e-01
9.08913553e-01 8.67616832e-02 -2.09100395e-01 -2.71591723e-01
1.15315640e+00 -3.38158816e-01 -5.50742447e-01 -3.00694592e-02
7.33653605e-01 -1.24739516e+00 8.62665176e-01 5.02798595e-02
-1.14992082e+00 -4.36406672e-01 -7.23385036e-01 1.21197112e-01
-1.71923533e-01 -2.90676326e-01 2.03180864e-01 6.27492070e-01
-8.60231757e-01 7.28335023e-01 -1.12439167e+00 -6.33319497e-01
4.90325034e-01 4.89875078e-01 -4.89958644e-01 -8.29374641e-02
-6.35424972e-01 1.01888394e+00 -1.50771871e-01 2.64498107e-02
-3.29252392e-01 -1.03213108e+00 -7.87655771e-01 -3.90798777e-01
1.75178185e-01 -8.66409361e-01 9.54027891e-01 -2.33505562e-01
-1.06863892e+00 1.02604544e+00 -2.72012234e-01 -1.95606217e-01
8.16709757e-01 -2.48580992e-01 -2.24432155e-01 -9.73190218e-02
4.64886986e-02 2.75094926e-01 5.06343961e-01 -1.12415290e+00
-2.27385044e-01 -6.26739025e-01 -3.14486623e-01 3.48790646e-01
3.09311926e-01 -8.24338794e-02 -3.75494301e-01 -5.13330102e-01
4.97203261e-01 -8.84990275e-01 -6.93549156e-01 3.25524837e-01
-3.83986235e-01 2.00514421e-01 5.76494157e-01 -8.38194415e-02
7.09489763e-01 -2.13885522e+00 -1.58637971e-01 6.43086076e-01
3.75511467e-01 -4.67758588e-02 4.64763567e-02 2.35495657e-01
-7.30925500e-02 -2.28310034e-01 -2.30721876e-01 -4.29537177e-01
-2.89215475e-01 1.31647736e-01 1.96661517e-01 1.09681118e+00
-3.58911455e-02 5.43240368e-01 -1.01971114e+00 -5.61997294e-01
6.34872496e-01 6.79806828e-01 -4.53780293e-01 3.47512327e-02
2.89248347e-01 1.02422488e+00 -4.57911104e-01 6.07128084e-01
9.04510975e-01 -4.20500308e-01 -1.31389529e-01 -4.18425530e-01
-1.42982647e-01 1.52936891e-01 -1.69902027e+00 2.04219484e+00
-4.03766364e-01 4.61587995e-01 -8.97821859e-02 -7.73883522e-01
8.68427932e-01 2.97072530e-01 1.10961795e+00 -2.63045311e-01
2.90503114e-01 3.73361409e-01 1.15729764e-03 -2.39638180e-01
3.99488419e-01 -1.98696449e-01 6.02749959e-02 1.36803105e-01
-2.55824864e-01 -6.35500014e-01 -2.30215952e-01 -1.16654702e-01
1.24602973e+00 -2.05336168e-01 8.00412536e-01 -2.69267172e-01
5.06087542e-01 9.18152854e-02 3.80887359e-01 8.46722245e-01
-4.38265711e-01 1.12663329e+00 2.65025169e-01 -4.45607930e-01
-1.10930932e+00 -1.33391142e+00 -7.67013550e-01 2.73180287e-02
6.52081609e-01 -3.23584646e-01 -3.40519935e-01 -5.02837896e-01
1.50364935e-01 2.75646836e-01 -4.47085232e-01 2.57993527e-02
-7.87320077e-01 -7.11032450e-01 2.29861960e-01 4.36753422e-01
-7.22754896e-02 -4.83291239e-01 -8.46062124e-01 3.83560538e-01
5.48609458e-02 -1.33218932e+00 -5.32521129e-01 8.63486752e-02
-1.26602519e+00 -1.29855323e+00 -6.96083724e-01 -4.03309464e-01
8.28580976e-01 3.01422805e-01 1.23748934e+00 -3.33230607e-02
-6.55627489e-01 5.99812984e-01 -2.70953357e-01 -3.87734860e-01
-3.00880581e-01 2.77712569e-02 3.43449444e-01 -2.96791643e-01
3.22093338e-01 -5.73402643e-01 -7.77720511e-01 6.84396565e-01
-6.84833825e-01 -1.97041824e-01 2.51087368e-01 7.56788671e-01
9.35609818e-01 -2.86615968e-01 -4.00009006e-02 -5.65174103e-01
3.71708035e-01 -3.39701980e-01 -7.09665895e-01 4.41523595e-03
-1.00729518e-01 -5.58987595e-02 2.54328460e-01 -5.68020165e-01
-4.75914747e-01 2.25582018e-01 -1.16320997e-01 -7.09627390e-01
-5.91175139e-01 -4.93025891e-02 -1.93171129e-02 -3.35809380e-01
1.08669043e+00 -9.61640254e-02 3.88309717e-01 -4.28162605e-01
1.42276391e-01 4.16043997e-01 4.33417201e-01 -4.84094441e-01
8.60965908e-01 9.35433269e-01 4.75083590e-01 -1.02475142e+00
-3.54109019e-01 -9.11567509e-01 -1.03136539e+00 -1.05721772e-01
7.19306052e-01 -6.41738892e-01 -3.99941891e-01 3.60953689e-01
-1.13838971e+00 -2.73328543e-01 -7.41211712e-01 9.51988816e-01
-8.02741468e-01 1.43973514e-01 -6.44362867e-02 -4.90738004e-01
-1.15277395e-01 -1.39101994e+00 1.30607331e+00 2.14543775e-01
-4.05226111e-01 -1.33971608e+00 3.79605412e-01 -1.92999914e-01
5.02147377e-01 5.80531836e-01 4.52998668e-01 -7.35814691e-01
-5.17737210e-01 -2.90446848e-01 4.63200882e-02 -7.11639300e-02
6.69484019e-01 -2.84571946e-01 -8.92233133e-01 -2.86889434e-01
9.75143611e-02 2.75339216e-01 1.96774557e-01 9.45026219e-01
8.72451842e-01 1.67131126e-01 -5.80649674e-01 6.43423140e-01
1.32871926e+00 -4.31503020e-02 4.45644677e-01 1.41497523e-01
6.34865463e-01 5.45036137e-01 7.25764215e-01 6.03359997e-01
-8.55170563e-03 1.04746008e+00 4.02471572e-01 -3.48850429e-01
-3.32856178e-01 2.53431320e-01 -3.23488355e-01 5.44910252e-01
-1.21548235e-01 1.47864208e-01 -1.06555629e+00 3.91942173e-01
-1.72688484e+00 -6.94042802e-01 -2.49552771e-01 2.65556502e+00
4.48512107e-01 8.40320513e-02 1.17439188e-01 -5.60298637e-02
6.80358887e-01 -9.08091217e-02 -5.15421927e-01 1.11328363e-01
-3.59301791e-02 2.62574404e-01 8.48902106e-01 5.86293221e-01
-1.24622476e+00 3.42434704e-01 6.01577806e+00 5.63942850e-01
-1.10881817e+00 2.46206105e-01 4.88764346e-01 -3.13944280e-01
3.72826904e-02 -2.49426410e-01 -8.68409991e-01 1.75754696e-01
7.48363018e-01 -5.96104935e-02 -1.59523576e-01 4.95504707e-01
4.33316439e-01 -1.32176489e-01 -1.17531145e+00 1.29045296e+00
1.28085703e-01 -1.36558366e+00 -4.51776654e-01 2.28631169e-01
4.94686663e-01 3.63452643e-01 -1.03024937e-01 -2.80958742e-01
-1.13130689e-01 -9.42458332e-01 2.53343940e-01 5.40105104e-01
7.96840072e-01 -4.35922354e-01 9.27666485e-01 2.15472192e-01
-1.26580727e+00 4.43999499e-01 -3.09211075e-01 3.63767534e-01
4.21425849e-01 8.20203900e-01 -9.93972480e-01 5.77543557e-01
5.97296298e-01 7.20100880e-01 -3.38606209e-01 1.66422975e+00
2.62730688e-01 -2.40785684e-02 -7.86766231e-01 2.86236465e-01
-6.46394193e-02 -1.37919828e-01 9.80935097e-01 1.13695109e+00
5.04266083e-01 6.74074739e-02 6.89691603e-02 5.88371217e-01
2.96104729e-01 2.88743317e-01 -7.64325619e-01 6.94160879e-01
4.95411247e-01 1.16421676e+00 -9.68829036e-01 -1.12249032e-01
-5.59514821e-01 5.59507310e-01 -1.62414655e-01 1.70487717e-01
-8.88467193e-01 -1.85963377e-01 9.84979510e-01 7.90111601e-01
3.36325616e-02 -4.69630361e-01 -3.92613113e-01 -1.08419251e+00
1.09848417e-01 -5.43875098e-01 4.88535672e-01 -4.63413298e-01
-1.28533602e+00 5.59861839e-01 4.19050902e-01 -1.83468080e+00
-1.84935555e-01 -6.21660769e-01 -5.43483377e-01 7.83997357e-01
-1.29068112e+00 -8.07574034e-01 -6.35596156e-01 7.80737698e-01
3.14217597e-01 1.47431657e-01 7.02270210e-01 4.17082787e-01
2.62660757e-02 5.12221336e-01 2.45740786e-01 -4.59780879e-02
9.93106902e-01 -1.08423722e+00 3.36176068e-01 4.87473488e-01
1.70296833e-01 4.68271494e-01 7.73671389e-01 -5.66775382e-01
-1.18862104e+00 -9.86221731e-01 5.84009409e-01 -6.79991901e-01
4.96512651e-01 -3.34955424e-01 -8.94349635e-01 4.48353380e-01
-3.64436328e-01 5.69958091e-01 6.46937668e-01 8.01970623e-03
7.85774440e-02 -9.94278416e-02 -1.51834142e+00 3.98580432e-01
1.07008386e+00 -1.14912428e-01 -5.22791505e-01 5.45737863e-01
2.01169729e-01 -1.22411406e+00 -1.31316996e+00 6.07358098e-01
5.48947275e-01 -8.67133856e-01 1.41218746e+00 -2.16666222e-01
-2.49191642e-01 -4.90424067e-01 -2.52610594e-01 -1.28268087e+00
-1.92940977e-04 -5.99175453e-01 8.68976414e-02 6.77358389e-01
2.11807445e-01 -8.05083573e-01 9.04758334e-01 5.43187559e-01
-2.54785329e-01 -5.56925714e-01 -1.38922095e+00 -1.06226850e+00
4.22178209e-02 -5.12980402e-01 5.91244340e-01 9.92590785e-01
-2.04365790e-01 -2.28759289e-01 9.40193236e-02 3.48609567e-01
8.31237137e-01 1.34545770e-02 1.13708365e+00 -1.27768338e+00
-5.42209260e-02 -4.47520465e-01 -9.91346419e-01 -9.60094273e-01
-3.35539877e-01 -7.38899171e-01 -4.37386073e-02 -1.35089886e+00
-1.83942154e-01 -8.58276725e-01 -4.45518419e-02 1.46247521e-01
5.40730320e-02 2.45232686e-01 6.91597834e-02 3.90130252e-01
-3.11554968e-01 2.00380728e-01 1.41444314e+00 1.85042843e-01
-6.67102873e-01 3.29742938e-01 -1.49127349e-01 7.18864918e-01
8.33425105e-01 -5.59179544e-01 -2.48727277e-01 -1.63787052e-01
-2.74837881e-01 8.33455622e-02 5.09500921e-01 -1.24646878e+00
3.45765650e-01 -2.25769654e-01 3.93550634e-01 -8.26653600e-01
5.07220924e-01 -1.12734008e+00 4.19832170e-01 4.09613580e-01
-6.51270896e-03 2.29231700e-01 4.17836308e-01 2.55888641e-01
-3.59677449e-02 -8.12600851e-02 9.96293187e-01 -1.02475464e-01
-2.33587563e-01 6.10942304e-01 1.16552770e-01 2.56061882e-01
1.07791555e+00 -5.98895073e-01 -1.42437845e-01 -2.68319488e-01
-7.89850533e-01 -4.24397662e-02 7.15344310e-01 3.51708293e-01
7.02914059e-01 -1.46046007e+00 -8.58849883e-01 3.83227944e-01
2.53401488e-01 4.86346304e-01 2.66019464e-01 1.51765311e+00
-6.35072887e-01 3.09748918e-01 8.98033567e-03 -1.30212438e+00
-1.31991911e+00 2.32253626e-01 6.39915586e-01 -2.17790365e-01
-8.67501557e-01 4.89415526e-01 2.65279144e-01 -4.58443820e-01
1.30945832e-01 -7.36070514e-01 8.90015960e-02 -2.39786550e-01
4.22498673e-01 3.95635933e-01 4.52404916e-01 -9.16135371e-01
-6.29311740e-01 1.03654575e+00 1.99097749e-02 2.56013781e-01
1.07783604e+00 -4.68974859e-02 2.77037948e-01 3.97593141e-01
1.20646942e+00 1.72233924e-01 -1.08804286e+00 -3.50733638e-01
-1.63674369e-01 -7.88021207e-01 -1.38331071e-01 -1.18146390e-01
-8.94803762e-01 6.35338366e-01 1.11139071e+00 4.43824902e-02
9.61278260e-01 3.57918650e-01 3.83148313e-01 -2.52147261e-02
3.16521794e-01 -5.38827956e-01 -2.85529375e-01 3.42249364e-01
7.32350826e-01 -1.33416796e+00 3.52217257e-01 -5.80503106e-01
-2.15464637e-01 1.03332198e+00 3.42554271e-01 -1.81406111e-01
9.32477653e-01 4.77392852e-01 1.64069787e-01 -3.11899662e-01
-1.89436063e-01 -7.27245584e-02 4.51655596e-01 8.97128284e-01
4.79697526e-01 2.37208288e-02 -3.82787623e-02 -1.40621379e-01
-2.00218901e-01 -2.03012660e-01 3.93415391e-01 1.08928907e+00
-1.29962757e-01 -1.04708242e+00 -9.62953150e-01 4.90690917e-01
-2.45886937e-01 8.43702257e-02 1.51859641e-01 1.19563890e+00
-3.76923867e-02 4.64668036e-01 6.28872097e-01 -3.07424627e-02
8.47101986e-01 -4.03371125e-01 6.72744572e-01 -5.68011403e-01
-4.75671858e-01 3.82694781e-01 -2.23964900e-01 -9.70729232e-01
-6.76417172e-01 -1.00042748e+00 -1.22313952e+00 -1.35687396e-01
-1.60347521e-01 9.34041862e-04 8.05539727e-01 6.61165714e-01
3.50497901e-01 3.09456855e-01 5.12809575e-01 -1.07537818e+00
-3.07969093e-01 -7.07633853e-01 -5.77513635e-01 6.83455646e-01
5.33797085e-01 -9.14563656e-01 -4.84925777e-01 -9.66764987e-02] | [7.725114822387695, -2.8373353481292725] |
6c5df2b1-a72b-4aa3-b7e5-f96d7098148e | a-unified-approach-to-controlling-implicit | 2306.13853 | null | https://arxiv.org/abs/2306.13853v1 | https://arxiv.org/pdf/2306.13853v1.pdf | A Unified Approach to Controlling Implicit Regularization via Mirror Descent | Inspired by the remarkable success of deep neural networks, there has been significant interest in understanding the generalization performance of overparameterized models. Substantial efforts have been invested in characterizing how optimization algorithms impact generalization through their "preferred" solutions, a phenomenon commonly referred to as implicit regularization. In particular, it has been argued that gradient descent (GD) induces an implicit $\ell_2$-norm regularization in regression and classification problems. However, the implicit regularization of different algorithms are confined to either a specific geometry or a particular class of learning problems, indicating a gap in a general approach for controlling the implicit regularization. To address this, we present a unified approach using mirror descent (MD), a notable generalization of GD, to control implicit regularization in both regression and classification settings. More specifically, we show that MD with the general class of homogeneous potential functions converges in direction to a generalized maximum-margin solution for linear classification problems, thereby answering a long-standing question in the classification setting. Further, we show that MD can be implemented efficiently and under suitable conditions, enjoys fast convergence. Through comprehensive experiments, we demonstrate that MD is a versatile method to produce learned models with different regularizers, which in turn have different generalization performances. | ['Navid Azizan', 'Kwangjun Ahn', 'Khashayar Gatmiry', 'Haoyuan Sun'] | 2023-06-24 | null | null | null | null | ['classification-1'] | ['methodology'] | [ 8.47842470e-02 1.48775041e-01 -4.44386005e-01 -3.66679758e-01
-6.84486151e-01 -3.84669870e-01 3.81618142e-01 4.18099426e-02
-3.83617938e-01 8.89710009e-01 -2.45668623e-03 -2.01794490e-01
-4.07662779e-01 -6.71306372e-01 -6.88699722e-01 -9.93350208e-01
1.17760628e-01 9.17738602e-02 -3.25939357e-01 -4.70313281e-01
4.60073799e-01 8.29391897e-01 -1.17733729e+00 -1.52412355e-01
1.04960275e+00 9.56018806e-01 -1.55516386e-01 3.78513664e-01
7.58647248e-02 3.97123069e-01 -2.89790094e-01 -3.45360309e-01
4.47409809e-01 -3.66701722e-01 -9.79365528e-01 9.22426507e-02
6.10134542e-01 2.93380797e-01 -4.00581986e-01 1.07282090e+00
4.38761860e-01 5.14601350e-01 8.10927927e-01 -1.03043103e+00
-7.79160082e-01 3.81578237e-01 -4.81497079e-01 2.55213559e-01
-1.42775953e-01 -1.14420494e-02 1.16248393e+00 -9.25911307e-01
4.01062697e-01 9.96237814e-01 9.31426764e-01 7.30121613e-01
-1.72599065e+00 -3.30530882e-01 4.36200231e-01 -1.81942686e-01
-1.35963070e+00 -1.95472807e-01 8.65025103e-01 -6.00294948e-01
4.73216593e-01 3.36929768e-01 4.43801969e-01 8.85567963e-01
4.25405987e-02 6.64482236e-01 1.07629883e+00 -6.44560635e-01
2.24891812e-01 4.96610701e-01 5.66877723e-01 7.39273071e-01
3.13601226e-01 -9.17168290e-05 -3.24780434e-01 -1.44792497e-01
7.61367261e-01 -2.09252179e-01 -7.49995589e-01 -6.14786446e-01
-7.83906043e-01 1.26612389e+00 6.12165987e-01 2.74913669e-01
-2.25361735e-02 4.68854886e-03 2.79491156e-01 4.03370112e-01
6.65639162e-01 9.24035013e-01 -4.68029320e-01 2.21585110e-01
-5.56671143e-01 5.62662423e-01 9.75159287e-01 7.47951865e-01
1.00378478e+00 1.91644609e-01 1.58337861e-01 1.02730393e+00
-2.65918002e-02 1.39453992e-01 3.39627475e-01 -7.58543253e-01
4.00015265e-01 5.46918035e-01 4.10822965e-02 -1.03077614e+00
-6.39805675e-01 -1.08495605e+00 -1.16633713e+00 3.21877003e-01
6.30179226e-01 -9.51407254e-02 -5.52348554e-01 2.25313902e+00
1.92236096e-01 -1.29649878e-01 -1.03416041e-01 9.60814297e-01
3.44909221e-01 3.60165119e-01 9.74533632e-02 -1.84162408e-01
6.47330225e-01 -8.11726987e-01 -2.49907002e-01 -1.11729890e-01
1.12197888e+00 -4.28436756e-01 1.32985413e+00 2.79950619e-01
-1.06530273e+00 -2.89193779e-01 -1.19218266e+00 -1.63416490e-01
-3.82637262e-01 3.67626771e-02 6.40385807e-01 5.84216058e-01
-9.82135832e-01 1.05692530e+00 -5.68006158e-01 -3.68048429e-01
3.66299480e-01 4.46428001e-01 -2.97215492e-01 1.67538509e-01
-1.06003141e+00 9.79852200e-01 1.71043128e-01 3.38402331e-01
-3.68320107e-01 -1.12900412e+00 -7.92601407e-01 -1.44975021e-01
2.01460615e-01 -7.55576670e-01 9.81339812e-01 -1.18725574e+00
-1.57163727e+00 1.09140468e+00 -9.46381539e-02 -4.04014111e-01
7.29339182e-01 -2.30143070e-01 -2.11536157e-04 -2.90655643e-01
-7.30153620e-02 3.73347908e-01 8.13641727e-01 -1.16624475e+00
-2.18198881e-01 -5.96599996e-01 2.75038838e-01 2.42045417e-01
-5.21447122e-01 -2.32280403e-01 3.33061069e-02 -8.19895506e-01
2.18687668e-01 -1.07817805e+00 -5.81303775e-01 -6.12954833e-02
-4.13951844e-01 -2.75812984e-01 5.01409829e-01 -2.47447491e-01
1.20280874e+00 -2.00394869e+00 7.26822555e-01 3.88794243e-01
2.92147011e-01 3.21137875e-01 -8.45547095e-02 3.00970852e-01
-3.18387270e-01 3.82313341e-01 -7.48342037e-01 -3.05504709e-01
1.89152043e-02 3.23606908e-01 -6.47628546e-01 8.46106470e-01
9.32050049e-02 7.88104355e-01 -5.95444202e-01 2.39228066e-02
-7.85494521e-02 5.11157691e-01 -8.01649213e-01 -8.48664492e-02
-1.26004383e-01 8.32733691e-01 -7.40178645e-01 2.19646066e-01
7.94477880e-01 -4.06302243e-01 -4.21715192e-02 4.50848602e-03
-1.41372368e-01 2.41114721e-01 -1.12899423e+00 1.65856814e+00
-5.71412504e-01 7.66211867e-01 3.38195443e-01 -1.79555833e+00
7.79842257e-01 -8.24333951e-02 3.72757107e-01 -4.05141354e-01
7.13766739e-02 4.93873864e-01 -3.07142794e-01 -3.48639131e-01
3.16837400e-01 -3.77119064e-01 1.05761431e-01 2.74232149e-01
-1.30629048e-01 -1.46975741e-01 8.05973858e-02 -2.21148014e-01
7.36884713e-01 1.64265502e-02 1.47607133e-01 -7.71263480e-01
7.06326485e-01 4.90129516e-02 4.63301629e-01 8.58258843e-01
-4.62795980e-02 6.13234818e-01 5.33528388e-01 -5.56071579e-01
-8.81775796e-01 -9.36333656e-01 -7.25353301e-01 1.06738997e+00
1.41571909e-01 -1.70628369e-01 -8.26468289e-01 -4.95096296e-01
1.88930362e-01 5.35713792e-01 -8.71774852e-01 -3.30733895e-01
-9.50633883e-01 -1.20642924e+00 4.36089754e-01 3.85712028e-01
4.87863511e-01 -7.12427735e-01 -1.26950473e-01 5.73807545e-02
1.87803701e-01 -9.15162742e-01 -3.76650482e-01 4.38488305e-01
-1.22739625e+00 -1.17710423e+00 -1.00597835e+00 -8.01172554e-01
7.68641412e-01 1.20140150e-01 1.20884657e+00 1.44160748e-01
-3.42195541e-01 4.49637264e-01 -1.63059384e-01 -5.90395965e-02
-2.31250793e-01 5.57121634e-01 1.88581333e-01 4.30785529e-02
1.30299285e-01 -7.53102601e-01 -6.20769918e-01 4.19801354e-01
-9.53911960e-01 -1.70498490e-01 3.24628502e-01 8.18695307e-01
4.62583125e-01 -3.23910624e-01 6.23326600e-01 -9.55007434e-01
8.03365588e-01 -4.95530874e-01 -6.83300614e-01 1.69408977e-01
-7.95728803e-01 3.78179878e-01 7.64132261e-01 -4.02805597e-01
-7.61113882e-01 -2.19360918e-01 -2.33770460e-01 -1.96836710e-01
2.45218240e-02 6.85265303e-01 9.78989713e-03 -6.14207268e-01
1.00209951e+00 1.34719834e-01 -1.77961513e-02 -5.83569705e-01
4.42910492e-01 2.51631051e-01 1.16804436e-01 -1.00325894e+00
8.61343861e-01 5.11358202e-01 5.01429796e-01 -1.20009553e+00
-1.30188239e+00 -1.84356511e-01 -7.47111440e-01 2.45038822e-01
7.12123632e-01 -4.79640245e-01 -4.11243320e-01 3.57835919e-01
-9.36616182e-01 -6.56228721e-01 -4.51248527e-01 5.93366981e-01
-6.80827558e-01 3.24645609e-01 -4.97512877e-01 -6.08237624e-01
-7.75877535e-02 -1.22264242e+00 6.38247728e-01 1.66415408e-01
-7.91813284e-02 -1.53853822e+00 1.56628385e-01 -1.99031364e-02
4.95763749e-01 2.23666325e-01 1.12337029e+00 -4.69015270e-01
-4.51648444e-01 -1.16971709e-01 -2.12649569e-01 6.01977289e-01
-1.56472088e-04 -2.47835726e-01 -8.74507904e-01 -3.23843271e-01
3.54815781e-01 -3.34456533e-01 1.04397869e+00 6.35507047e-01
1.42941833e+00 -2.05580950e-01 -3.11289072e-01 1.25668132e+00
1.42000997e+00 -2.49445230e-01 4.10420209e-01 4.33723807e-01
6.33304000e-01 7.29336500e-01 1.09823197e-01 1.57256916e-01
-1.52734026e-01 8.32954586e-01 2.04070300e-01 -2.79698044e-01
1.88163519e-01 -5.75039256e-03 -6.53239787e-02 7.16979742e-01
-2.72065908e-01 2.43918449e-01 -5.71193039e-01 6.98508322e-02
-1.89630699e+00 -7.54455745e-01 -1.77802339e-01 2.43008184e+00
9.26510036e-01 -3.29943597e-02 3.94462310e-02 1.46908686e-01
6.33418739e-01 1.52763918e-01 -7.59534717e-01 -4.10884559e-01
-4.47527736e-01 3.96226943e-01 5.93453705e-01 7.54861534e-01
-1.07893753e+00 7.95920312e-01 6.91989231e+00 9.64907646e-01
-1.41453409e+00 -1.78717449e-02 8.40131342e-01 1.56389549e-01
-3.04961592e-01 -1.01293400e-01 -8.92579973e-01 1.00244381e-01
3.97427976e-01 -3.02842796e-01 4.08925712e-01 1.01120365e+00
2.05055714e-01 2.41786972e-01 -1.22548687e+00 1.00848150e+00
-1.19060613e-01 -1.51608264e+00 5.84730357e-02 1.50385737e-01
8.74327123e-01 -1.26765668e-01 5.11860192e-01 3.38221073e-01
-1.91497188e-02 -1.25213206e+00 3.68565679e-01 2.87456572e-01
7.51606643e-01 -6.87258184e-01 2.03520417e-01 3.27255011e-01
-1.01324606e+00 -4.31151353e-02 -7.03680992e-01 -6.49599582e-02
-1.71446711e-01 6.61606610e-01 -2.92876810e-01 3.86041433e-01
4.47319001e-01 7.47743011e-01 -2.42638528e-01 9.36837375e-01
-1.28987685e-01 5.09043753e-01 -2.25953385e-01 4.48872112e-02
3.72432739e-01 -8.01358104e-01 7.12067127e-01 1.27065217e+00
8.53411853e-02 1.23727426e-01 -3.38977166e-02 1.11091006e+00
-2.84739763e-01 3.71692836e-01 -6.23268962e-01 1.94349959e-01
-1.58379644e-01 1.18057811e+00 -5.38986504e-01 1.28775761e-01
-2.59164482e-01 6.11176550e-01 9.09017205e-01 7.35982418e-01
-7.00868428e-01 -2.95374215e-01 8.79769206e-01 2.08241299e-01
1.78767487e-01 -4.11339879e-01 -5.94674945e-01 -1.48677325e+00
2.46131614e-01 -5.53666174e-01 2.59328574e-01 -3.39284480e-01
-1.38312268e+00 5.31630516e-01 -6.27900064e-02 -8.35178673e-01
1.85584888e-01 -1.03224277e+00 -5.85497260e-01 1.02581573e+00
-1.61267579e+00 -5.89083910e-01 -3.59417886e-01 5.26589811e-01
3.71795535e-01 3.56344506e-02 6.90228939e-01 2.72361428e-01
-7.61891246e-01 8.27062488e-01 4.23581362e-01 1.18920408e-01
5.22103250e-01 -1.39207947e+00 -1.26206204e-02 4.73817497e-01
6.59169108e-02 7.84577012e-01 7.87119567e-01 -1.47750601e-01
-1.34297955e+00 -1.07418883e+00 4.58644986e-01 -2.71118909e-01
6.54622674e-01 -2.36348584e-01 -1.16142881e+00 7.53288090e-01
-1.01915523e-01 7.92428330e-02 5.98442614e-01 3.83658290e-01
-3.76091927e-01 -1.09401576e-01 -9.52669203e-01 8.58476162e-01
1.26204622e+00 -4.11456972e-01 -3.71277452e-01 5.41027963e-01
5.54862797e-01 -3.50747049e-01 -7.61691689e-01 5.89207411e-01
3.06873411e-01 -9.50499117e-01 1.01599693e+00 -1.12232363e+00
3.81481230e-01 2.14710400e-01 -1.48237675e-01 -1.47772837e+00
-2.94382066e-01 -8.01254392e-01 7.60582909e-02 6.86970234e-01
7.00444579e-01 -1.11134434e+00 8.79766822e-01 7.63595998e-01
-3.43376070e-01 -1.34045172e+00 -8.48438323e-01 -7.87369907e-01
9.27172601e-01 -3.68260682e-01 5.10000065e-02 1.04333472e+00
-1.23870172e-01 1.79219633e-01 -1.88530922e-01 3.42607424e-02
4.97166455e-01 1.10671423e-01 6.72791302e-01 -1.35290062e+00
-2.27270797e-01 -8.97473097e-01 -3.47275317e-01 -1.58118784e+00
6.05535388e-01 -1.24211204e+00 -3.35171908e-01 -9.91268039e-01
-5.11005968e-02 -1.00163972e+00 -2.74498820e-01 -3.88163025e-03
-1.30614713e-01 1.51097849e-01 -2.52730787e-01 4.44449574e-01
-1.55460924e-01 7.41835833e-01 1.33979380e+00 8.22540279e-03
-4.51360613e-01 3.23970765e-01 -7.96045482e-01 8.79468560e-01
8.97785246e-01 -3.31174731e-01 -3.17030936e-01 -4.12287503e-01
5.81503570e-01 -2.96921700e-01 3.10421616e-01 -7.56752074e-01
5.59160337e-02 -2.45605007e-01 1.32057130e-01 2.46663410e-02
3.30462307e-01 -3.50731879e-01 -3.92391294e-01 3.04479510e-01
-8.66896033e-01 -1.57380700e-01 1.05869062e-01 5.41457653e-01
-1.25783682e-01 -6.28175616e-01 1.20494688e+00 -1.21200636e-01
-2.57556826e-01 6.44818246e-01 -1.62206203e-01 5.96859872e-01
7.67142713e-01 -8.92856494e-02 -7.05853030e-02 -2.08562106e-01
-9.59529757e-01 1.21522155e-02 4.24648583e-01 -1.62672661e-02
3.22301626e-01 -1.20717633e+00 -7.44310379e-01 1.61011115e-01
-4.25325558e-02 8.65416154e-02 -3.09826527e-02 1.21609282e+00
-4.40763175e-01 4.81831521e-01 2.41408780e-01 -6.46319568e-01
-7.23816454e-01 4.20305371e-01 8.82768273e-01 -2.94412673e-01
-7.71572948e-01 1.08346426e+00 6.36348665e-01 -5.89151919e-01
2.07955971e-01 -3.22559565e-01 7.44414330e-02 -1.18644424e-01
2.18672931e-01 4.87058520e-01 8.65656734e-02 -4.91891712e-01
-6.78104535e-02 9.08866584e-01 -1.02997445e-01 4.67829295e-02
1.30662751e+00 1.07542418e-01 -1.80913642e-01 3.88924688e-01
1.58731866e+00 1.08276196e-01 -1.28869402e+00 -3.68929952e-01
3.45250666e-02 -2.42202014e-01 -7.44574741e-02 -2.91652203e-01
-1.19519162e+00 8.98494959e-01 1.67177826e-01 6.07223034e-01
8.66353154e-01 -9.55561399e-02 3.86469841e-01 6.87914133e-01
1.16390683e-01 -1.12001121e+00 -7.84844160e-03 6.23375654e-01
1.18906701e+00 -1.33254051e+00 -3.51975970e-02 -5.22536814e-01
-2.89894223e-01 1.28370440e+00 4.68154967e-01 -5.45414329e-01
9.28594053e-01 -1.06525362e-01 -7.17331097e-02 5.09854755e-04
-3.54757577e-01 3.94324362e-02 4.48326319e-01 3.68956298e-01
5.95718324e-01 -2.04852715e-01 -4.27687377e-01 3.19708526e-01
-2.99878299e-01 -1.94903523e-01 3.23691666e-01 6.35477662e-01
-3.95168960e-01 -1.10609388e+00 -1.19453833e-01 2.82532662e-01
-4.62316573e-01 4.91544232e-02 -3.12842548e-01 1.11992860e+00
-1.59725368e-01 7.35626757e-01 -2.42617428e-01 -7.30077401e-02
2.11811885e-01 1.31703783e-02 7.54453599e-01 -5.95471740e-01
-4.36797649e-01 -4.27932858e-01 -2.14417085e-01 -4.27170366e-01
-3.08287621e-01 -3.95407736e-01 -1.00332212e+00 -2.74067163e-01
-3.44680905e-01 4.37407017e-01 6.07994735e-01 1.12428510e+00
2.25398019e-01 1.96695179e-01 8.40213776e-01 -9.88694489e-01
-1.06051743e+00 -7.94902802e-01 -6.99531734e-01 5.15074193e-01
4.72992301e-01 -8.48967075e-01 -1.03740096e+00 -4.06309426e-01] | [7.884871006011963, 3.772325038909912] |
d30abc10-cbe3-4bd1-95c2-274e572a9d7c | ndt-transformer-large-scale-3d-point-cloud | 2103.12292 | null | https://arxiv.org/abs/2103.12292v1 | https://arxiv.org/pdf/2103.12292v1.pdf | NDT-Transformer: Large-Scale 3D Point Cloud Localisation using the Normal Distribution Transform Representation | 3D point cloud-based place recognition is highly demanded by autonomous driving in GPS-challenged environments and serves as an essential component (i.e. loop-closure detection) in lidar-based SLAM systems. This paper proposes a novel approach, named NDT-Transformer, for realtime and large-scale place recognition using 3D point clouds. Specifically, a 3D Normal Distribution Transform (NDT) representation is employed to condense the raw, dense 3D point cloud as probabilistic distributions (NDT cells) to provide the geometrical shape description. Then a novel NDT-Transformer network learns a global descriptor from a set of 3D NDT cell representations. Benefiting from the NDT representation and NDT-Transformer network, the learned global descriptors are enriched with both geometrical and contextual information. Finally, descriptor retrieval is achieved using a query-database for place recognition. Compared to the state-of-the-art methods, the proposed approach achieves an improvement of 7.52% on average top 1 recall and 2.73% on average top 1% recall on the Oxford Robotcar benchmark. | ['Li Sun', 'Tom Duckett', 'Yang Gao', 'Songzhi Su', 'Daniel Adolfsson', 'Cheng Zhao', 'Zhicheng Zhou'] | 2021-03-23 | null | null | null | null | ['loop-closure-detection'] | ['computer-vision'] | [-2.08133355e-01 -3.06723922e-01 -2.72287548e-01 -5.73839962e-01
-1.06947923e+00 -5.34817874e-01 8.26994359e-01 5.65851629e-01
-5.53928316e-01 3.22553933e-01 -1.97774410e-01 -8.55504423e-02
-2.92323321e-01 -1.12034178e+00 -1.02787507e+00 -6.45817101e-01
1.10907473e-01 9.55588639e-01 3.28408509e-01 -1.62483782e-01
4.97391433e-01 1.20245278e+00 -1.80004668e+00 -3.96375626e-01
8.13782573e-01 1.25254834e+00 2.69097745e-01 2.36923806e-02
-3.69696081e-01 1.23443089e-01 -2.69681692e-01 2.02011198e-01
3.12507957e-01 3.54934216e-01 -1.74671039e-01 -3.09593707e-01
4.91212279e-01 -1.33682728e-01 -3.81015986e-01 1.05778313e+00
3.91350120e-01 4.51838642e-01 8.80741537e-01 -1.36858368e+00
-4.57238346e-01 -3.66409630e-01 -4.91454750e-01 -1.98199585e-01
2.92870253e-01 -1.35332540e-01 7.65372396e-01 -1.27689040e+00
7.14514554e-01 1.39092040e+00 6.00727379e-01 1.32424599e-02
-1.10305035e+00 -7.81467140e-01 7.13045076e-02 2.81420439e-01
-1.95351052e+00 -2.30237320e-01 8.28684688e-01 -3.12163085e-01
1.07360160e+00 5.09339757e-02 7.62721419e-01 4.92807120e-01
4.89808559e-01 6.85020089e-01 8.96562457e-01 -8.47754185e-04
3.35837692e-01 -1.47628654e-02 -1.30496025e-01 6.58160746e-01
3.59913766e-01 2.15306893e-01 -6.14235282e-01 -3.03470761e-01
7.78947711e-01 5.73819458e-01 2.51883686e-01 -1.04717815e+00
-1.18833232e+00 8.00904095e-01 1.00177801e+00 1.32899415e-02
-6.40151441e-01 2.60106713e-01 2.62065709e-01 -2.81614978e-02
4.90627825e-01 1.07023992e-01 -1.15706362e-01 -1.50754765e-01
-8.14616859e-01 6.72393501e-01 6.00064695e-01 1.47900832e+00
1.36721730e+00 -2.56324023e-01 -5.00209592e-02 7.19549596e-01
5.67240298e-01 1.23644733e+00 1.52787879e-01 -7.19638944e-01
3.50397110e-01 8.08639407e-01 1.64706394e-01 -1.31327033e+00
-3.35106045e-01 -4.83896673e-01 -7.67289162e-01 1.22896925e-01
-1.81605846e-01 6.13676429e-01 -9.88072515e-01 1.36820877e+00
5.21563053e-01 2.47966141e-01 7.90382996e-02 8.96114886e-01
7.27350235e-01 7.18642771e-01 -2.11549625e-01 4.50342029e-01
1.14475238e+00 -6.58464611e-01 -3.37495744e-01 -2.62812883e-01
5.92795730e-01 -6.76726818e-01 3.91120493e-01 7.91000128e-02
-5.51493585e-01 -4.24958974e-01 -1.11129022e+00 -3.47573698e-01
-5.66939056e-01 1.22452736e-01 3.15339118e-01 1.94640532e-01
-9.59764361e-01 2.30243713e-01 -9.85304773e-01 -5.03895640e-01
4.47516024e-01 4.41351891e-01 -6.47353470e-01 -7.54022479e-01
-7.32087553e-01 1.11158192e+00 1.89910769e-01 1.09041438e-01
-9.32979286e-01 -4.92165625e-01 -1.25579011e+00 -1.61869183e-01
1.50703922e-01 -5.19425213e-01 9.15526688e-01 4.09952462e-01
-1.14780486e+00 8.96575630e-01 -6.39636576e-01 -5.35264075e-01
2.60066360e-01 -3.03830415e-01 -1.33331046e-01 2.20729206e-02
5.00688851e-01 7.18060195e-01 6.53064728e-01 -1.27146339e+00
-7.57771969e-01 -7.53587902e-01 -3.76227617e-01 3.63986343e-01
3.82743359e-01 -5.06630838e-01 -4.76370245e-01 -6.37157485e-02
9.42759693e-01 -1.28873992e+00 -2.08423644e-01 1.79502174e-01
-2.26714686e-01 -5.65772235e-01 1.00104678e+00 -6.59216195e-02
4.74202693e-01 -2.21637416e+00 -3.53558473e-02 4.22872722e-01
2.55096167e-01 -5.80773875e-02 -7.05739930e-02 5.25603831e-01
3.35566252e-01 -2.95452535e-01 -1.35924920e-01 -6.11036718e-01
4.10546303e-01 4.29920226e-01 -3.59798640e-01 8.74412537e-01
3.47060114e-01 8.76787245e-01 -1.05594540e+00 -3.14677030e-01
7.79502034e-01 7.54054606e-01 -4.48340893e-01 5.96747827e-03
8.06130543e-02 3.45133096e-01 -8.06513131e-01 8.46917152e-01
1.15830624e+00 2.97620595e-01 -3.94544691e-01 -3.42280744e-03
-5.60704350e-01 2.64523327e-01 -9.52267826e-01 2.12046075e+00
-4.39407080e-01 4.86243188e-01 -2.99549729e-01 -7.59096801e-01
1.72892892e+00 -1.92842647e-01 4.18628544e-01 -8.93229783e-01
9.03353933e-03 6.88295364e-01 -5.64672291e-01 1.75515100e-01
8.43556523e-01 -1.18838111e-02 -5.79085648e-01 -1.70318633e-01
8.14694017e-02 -6.48487270e-01 -4.28805321e-01 -4.57164831e-02
1.11090052e+00 1.30765930e-01 1.72015235e-01 -3.41531605e-01
6.27421379e-01 3.09301794e-01 5.99740684e-01 6.29586220e-01
-1.64765522e-01 5.54466724e-01 1.26303867e-01 -5.41720092e-01
-1.00322950e+00 -1.19525409e+00 -2.03559145e-01 3.34496021e-01
6.77289307e-01 -2.85432220e-01 -1.13276176e-01 -2.72004277e-01
6.59250081e-01 6.14078879e-01 -3.48701090e-01 -3.61882895e-01
-4.53381807e-01 9.10355225e-02 4.16104645e-01 4.86903340e-01
5.89041591e-01 -6.35438085e-01 -6.54784024e-01 2.18593016e-01
8.81193504e-02 -1.10363841e+00 -1.92940548e-01 3.60569537e-01
-8.68497074e-01 -7.62983441e-01 -5.31447291e-01 -8.59754980e-01
6.51841462e-01 7.31795490e-01 6.45423472e-01 -2.49521360e-01
-5.99847175e-02 2.34336957e-01 -2.72709906e-01 -5.41176796e-01
1.41957372e-01 1.33736983e-01 2.66554683e-01 -1.02653489e-01
8.86662781e-01 -6.92630231e-01 -5.02986968e-01 3.03462774e-01
-5.55011570e-01 -3.44711840e-01 8.98748815e-01 8.31860006e-01
1.28182733e+00 9.68597084e-03 1.01710960e-01 -4.14589167e-01
2.67943174e-01 -5.37118256e-01 -1.02867115e+00 -1.81942090e-01
-5.68842113e-01 3.53271633e-01 1.56573668e-01 -6.50242493e-02
-4.85235602e-01 3.95215929e-01 -1.09915495e-01 -9.74849343e-01
-3.22180688e-01 5.48796356e-01 -1.81958169e-01 -4.14769918e-01
4.42195565e-01 5.80901444e-01 1.11662827e-01 -5.92329860e-01
4.65921640e-01 6.08303666e-01 6.38059974e-01 -5.33180535e-01
1.13559222e+00 6.49302483e-01 3.75236928e-01 -7.12464929e-01
-4.28480953e-01 -1.09835601e+00 -9.24072742e-01 1.11768924e-01
7.12086499e-01 -1.33625197e+00 -6.41686559e-01 3.88953954e-01
-1.31346977e+00 1.70738384e-01 -2.34224588e-01 6.00320458e-01
-6.96661055e-01 -3.57297659e-02 -4.44192961e-02 -8.61237824e-01
-1.64466277e-01 -1.14714038e+00 1.72020471e+00 2.03937933e-01
2.73936838e-01 -5.12909710e-01 2.87317127e-01 -3.14387009e-02
3.67033213e-01 4.30540353e-01 6.93127096e-01 -6.96077764e-01
-1.13850653e+00 -7.45709717e-01 -4.12719399e-01 1.79159753e-02
-8.68309569e-03 -5.30478179e-01 -7.79155076e-01 -2.17257783e-01
-8.94657522e-02 2.16834638e-02 8.39983404e-01 1.99460343e-01
6.05654657e-01 2.12911293e-01 -6.72961414e-01 7.20467031e-01
1.60627854e+00 2.31013298e-01 4.49856758e-01 4.92166460e-01
7.32319534e-01 3.53637069e-01 1.11482930e+00 4.52294827e-01
8.12426984e-01 6.87213361e-01 7.41201222e-01 1.67787552e-01
2.09531069e-01 -7.50536919e-01 5.41350665e-03 6.83292270e-01
3.23688835e-01 2.15061784e-01 -1.03731334e+00 7.55568981e-01
-2.03702974e+00 -4.67252403e-01 2.63081882e-02 2.30345559e+00
1.56738490e-01 1.41624734e-01 -2.96992719e-01 -4.08907747e-03
8.01382363e-01 2.85062343e-01 -7.67578006e-01 -4.17637192e-02
7.55952746e-02 9.09858495e-02 8.67117286e-01 3.46566916e-01
-1.04976690e+00 1.10375130e+00 4.36529446e+00 8.14178944e-01
-1.15577590e+00 1.02380197e-02 -2.30766982e-01 3.18848282e-01
-1.49565592e-01 1.09341703e-01 -1.01175475e+00 8.86067674e-02
7.52782524e-01 -2.64086396e-01 8.71240497e-02 1.21805036e+00
-1.90921761e-02 -1.20636962e-01 -9.41730976e-01 1.37309051e+00
4.92799357e-02 -1.19576454e+00 9.79565829e-02 5.76176763e-01
4.60896045e-01 5.89183807e-01 5.33717535e-02 5.17710805e-01
-7.03349039e-02 -8.46766770e-01 1.19161284e+00 5.97902417e-01
9.22921062e-01 -1.04316139e+00 9.60475266e-01 6.46115005e-01
-1.66277146e+00 5.01947142e-02 -7.34885752e-01 7.89624676e-02
9.43671986e-02 6.69522941e-01 -1.05143094e+00 7.98395693e-01
5.97939670e-01 1.13293409e+00 -3.29427093e-01 1.49961114e+00
-2.02234238e-01 -9.17841196e-02 -8.28790665e-01 -1.19062528e-01
4.33324128e-01 -2.25968719e-01 7.95445025e-01 9.54182267e-01
6.52082980e-01 -1.18383251e-01 4.26582903e-01 9.72772658e-01
-3.39614712e-02 -9.70606804e-02 -1.13836789e+00 1.85126469e-01
1.14758575e+00 1.22100091e+00 -5.58328390e-01 -2.14348510e-01
-3.57455276e-02 5.96535981e-01 2.31184408e-01 -1.04497755e-02
-4.63339984e-01 -6.19567335e-01 9.44370389e-01 2.63258852e-02
5.27469337e-01 -8.89842391e-01 6.20037727e-02 -8.77469540e-01
7.73934945e-02 -2.53624976e-01 -1.71694309e-01 -8.13343227e-01
-1.12522793e+00 5.78474939e-01 2.27548629e-01 -1.61725891e+00
-2.55449533e-01 -6.10086739e-01 -1.74248964e-01 1.15271080e+00
-2.15930033e+00 -1.18329108e+00 -6.91804945e-01 6.31851614e-01
2.13062659e-01 4.11000662e-02 8.51125181e-01 2.96133786e-01
1.55140879e-02 1.48929834e-01 3.06387156e-01 -1.18978871e-02
5.90846896e-01 -1.09080184e+00 7.00983047e-01 3.73051316e-01
1.42233595e-01 7.31626451e-01 3.02149355e-01 -7.01132119e-01
-1.89803052e+00 -1.40156460e+00 9.40057158e-01 -4.50821072e-01
3.51007670e-01 -5.65526485e-01 -7.37843633e-01 3.89208049e-01
-5.45100749e-01 4.49436665e-01 2.05588549e-01 -1.34209111e-01
-5.04281700e-01 -4.13637936e-01 -1.24511695e+00 1.24083959e-01
1.00230563e+00 -8.06606174e-01 -5.92345595e-01 1.97082624e-01
8.18673253e-01 -8.92005026e-01 -8.94695759e-01 4.99655038e-01
4.92211074e-01 -5.75337529e-01 1.12239182e+00 1.19241826e-01
-3.29876900e-01 -6.71609402e-01 -5.87687969e-01 -1.14559722e+00
-3.33241463e-01 -1.99284807e-01 2.25677997e-01 8.50242198e-01
-2.45231427e-02 -8.34643781e-01 8.30900073e-01 7.69401789e-02
-4.39407289e-01 -5.00181019e-01 -1.49420738e+00 -9.99848783e-01
-6.07069135e-02 -6.41644478e-01 9.71885979e-01 4.42384034e-01
-6.36599839e-01 1.42476007e-01 1.23423330e-01 5.44815600e-01
9.16540146e-01 2.04650402e-01 9.84274447e-01 -1.60642791e+00
7.44375765e-01 -1.81467280e-01 -1.20265067e+00 -1.52150404e+00
3.03392261e-01 -1.16353869e+00 5.19885361e-01 -1.72923350e+00
-2.86076456e-01 -7.89587975e-01 -3.95271778e-01 2.41447970e-01
5.20370603e-01 6.67638108e-02 3.51023883e-01 5.28037906e-01
-7.03645289e-01 1.05572164e+00 8.99220884e-01 -2.81695515e-01
-1.10580973e-01 -5.47938049e-04 -1.93202451e-01 2.95781523e-01
5.32769620e-01 -7.83194482e-01 -2.31050342e-01 -4.67160910e-01
-1.20642707e-01 1.81202590e-02 7.36575782e-01 -1.33726752e+00
6.29090846e-01 -4.81401198e-02 2.69741535e-01 -1.62661552e+00
8.26039433e-01 -9.70450103e-01 3.95190194e-02 3.61759037e-01
1.03495494e-01 1.89283997e-01 2.88702071e-01 9.44048762e-01
-4.75931853e-01 3.87094133e-02 4.43585187e-01 -8.37296154e-03
-9.25232768e-01 6.70734406e-01 1.79336630e-02 -4.38035488e-01
8.91981900e-01 -2.87255019e-01 -2.17670918e-01 -5.15648723e-02
-2.63969153e-01 3.46465677e-01 7.55262434e-01 5.77867806e-01
1.06846809e+00 -1.76187289e+00 -4.46548313e-01 5.67313552e-01
6.47684753e-01 8.11570525e-01 1.09270044e-01 7.44901001e-01
-6.29868090e-01 8.73580933e-01 -2.59172887e-01 -1.27428317e+00
-8.18979263e-01 3.40523064e-01 6.23863861e-02 7.21286535e-02
-8.22855473e-01 6.86328828e-01 1.83492247e-02 -8.60953212e-01
6.08468615e-02 -6.62610173e-01 8.74524787e-02 -1.47067279e-01
1.19756676e-01 1.10403284e-01 3.13263267e-01 -1.09177709e+00
-8.44700694e-01 1.02007067e+00 1.47184983e-01 -1.02970801e-01
1.46164763e+00 -1.10721879e-01 -2.35300571e-01 5.29988229e-01
1.47459245e+00 -1.75132036e-01 -1.00190210e+00 -6.13627315e-01
2.66581982e-01 -6.61391318e-01 7.61492476e-02 -2.06824034e-01
-5.91722846e-01 9.45123196e-01 6.95142806e-01 -3.19471985e-01
6.07722819e-01 1.75481826e-01 7.90840983e-01 9.38846052e-01
1.09483981e+00 -6.24453962e-01 -3.73482943e-01 9.24738407e-01
9.53602493e-01 -1.17807770e+00 2.29007769e-02 -2.30336085e-01
-3.24123234e-01 9.00437713e-01 1.57584161e-01 -6.60515487e-01
7.17000365e-01 -1.73661813e-01 -1.51904479e-01 -4.10842717e-01
-4.60889816e-01 -3.02172273e-01 4.54118997e-01 8.30673635e-01
-2.43285775e-01 1.91589236e-01 2.26686791e-01 2.60564893e-01
-2.91711122e-01 -1.16959386e-01 -1.29971161e-01 1.15886962e+00
-7.63024390e-01 -1.04588819e+00 -4.41738546e-01 2.05128223e-01
5.26556492e-01 1.74164295e-01 -6.65633082e-02 8.39960217e-01
1.31498650e-01 6.20358467e-01 2.60784239e-01 -6.32526815e-01
7.23096728e-01 5.78550762e-03 3.15739304e-01 -8.40188622e-01
6.69795554e-03 -9.48036183e-03 -3.81406456e-01 -8.52808893e-01
-1.73437700e-01 -8.33038449e-01 -1.47772300e+00 -1.08231649e-01
-2.36226961e-01 3.34512025e-01 1.27569020e+00 7.33504176e-01
7.18987167e-01 8.15779641e-02 5.48251271e-01 -1.36570871e+00
-4.54430133e-01 -7.40325809e-01 -7.59809613e-01 1.51330549e-02
5.63177049e-01 -1.23829305e+00 -1.91431209e-01 -6.43205166e-01] | [7.493706226348877, -2.1892752647399902] |
c74d4536-4ffe-4b5a-a074-49874c193a38 | universal-prototype-transport-for-zero-shot | 2203.03971 | null | https://arxiv.org/abs/2203.03971v1 | https://arxiv.org/pdf/2203.03971v1.pdf | Universal Prototype Transport for Zero-Shot Action Recognition and Localization | This work addresses the problem of recognizing action categories in videos for which no training examples are available. The current state-of-the-art enables such a zero-shot recognition by learning universal mappings from videos to a shared semantic space, either trained on large-scale seen actions or on objects. While effective, we find that universal action and object mappings are biased to their seen categories. Such biases are further amplified due to biases between seen and unseen categories in the semantic space. The compounding biases result in many unseen action categories simply never being selected during inference, hampering zero-shot progress. We seek to address this limitation and introduce universal prototype transport for zero-shot action recognition. The main idea is to re-position the semantic prototypes of unseen actions through transduction, i.e. by using the distribution of the unlabelled test set. For universal action models, we first seek to find a hyperspherical optimal transport mapping from unseen action prototypes to the set of all projected test videos. We then define a target prototype for each unseen action as the weighted Fr\'echet mean over the transport couplings. Equipped with a target prototype, we propose to re-position unseen action prototypes along the geodesic spanned by the original and target prototypes, acting as a form of semantic regularization. For universal object models, we outline a variant that defines target prototypes based on an optimal transport between unseen action prototypes and semantic object prototypes. Empirically, we show that universal prototype transport diminishes the biased selection of unseen action prototypes and boosts both universal action and object models, resulting in state-of-the-art performance for zero-shot classification and spatio-temporal localization. | ['Pascal Mettes'] | 2022-03-08 | null | null | null | null | ['zero-shot-action-recognition'] | ['computer-vision'] | [ 4.12987173e-01 2.41133258e-01 -2.61044472e-01 -1.88880503e-01
-5.02096534e-01 -3.69659454e-01 9.30681229e-01 -5.21029890e-01
-3.94129097e-01 4.91455734e-01 3.07084829e-01 3.91937256e-01
-2.45346963e-01 -7.09248066e-01 -8.98841977e-01 -1.08765578e+00
1.12209946e-01 6.91930592e-01 4.68392193e-01 -1.26474917e-01
2.86062270e-01 4.34591681e-01 -1.93191218e+00 5.49105704e-01
4.16968435e-01 8.43823493e-01 2.03220382e-01 8.34310293e-01
-1.35408923e-01 7.16865540e-01 -2.92010039e-01 -1.26365378e-01
3.26314747e-01 -9.71453965e-01 -9.47853267e-01 4.79877532e-01
4.93191987e-01 -6.56924918e-02 -4.88393813e-01 1.22348404e+00
1.73367664e-01 6.33871436e-01 1.04433906e+00 -1.37708557e+00
-1.06159818e+00 1.86885089e-01 -1.37164280e-01 1.36926770e-01
2.48745799e-01 2.43775129e-01 1.04126012e+00 -1.00147450e+00
1.01005363e+00 1.29749751e+00 4.16860074e-01 1.17939508e+00
-1.30942106e+00 -4.52432819e-02 1.48893192e-01 3.15901607e-01
-1.31099081e+00 -5.41871727e-01 6.29321635e-01 -6.37447536e-01
9.01076317e-01 3.61202896e-01 6.24888718e-01 1.40287530e+00
7.45683983e-02 7.60474920e-01 7.01084971e-01 -4.03583944e-01
5.55763960e-01 2.62625456e-01 -6.78824121e-03 7.09323287e-01
-1.23186938e-01 1.59753889e-01 -5.22976339e-01 9.69062280e-03
7.01773763e-01 2.58983791e-01 -4.38923836e-01 -9.55658734e-01
-1.26237154e+00 9.53071594e-01 4.70963418e-01 4.55708086e-01
-3.28298599e-01 2.32921585e-01 2.69631028e-01 3.05642366e-01
4.84492004e-01 4.64064926e-01 -3.61815989e-01 -1.37874167e-02
-5.56392908e-01 7.42066577e-02 6.50984704e-01 8.05392325e-01
8.01327646e-01 -7.19507039e-02 -2.95920938e-01 7.51189947e-01
4.50318418e-02 6.72022164e-01 5.51474929e-01 -1.08532834e+00
-6.71266485e-03 5.06679714e-01 1.40386000e-01 -7.59083927e-01
-8.83439034e-02 -5.75211011e-02 -4.95469868e-01 2.86654234e-01
5.81921458e-01 2.84584582e-01 -1.16674650e+00 1.93930030e+00
2.15732247e-01 4.54364508e-01 3.40367764e-01 1.15940058e+00
3.42282444e-01 6.40351772e-01 2.25328013e-01 -1.27777383e-01
9.63030517e-01 -9.72378373e-01 -2.90709227e-01 -2.21973266e-02
9.88452435e-01 -2.72842109e-01 1.22043920e+00 9.76631194e-02
-9.65254784e-01 -5.78481257e-01 -7.76748359e-01 1.25698939e-01
-5.54324746e-01 -1.21796697e-01 2.30826244e-01 3.07102203e-01
-9.05845881e-01 9.79946315e-01 -8.68196785e-01 -9.38676059e-01
4.88523960e-01 2.01814294e-01 -4.62848276e-01 -1.32813811e-01
-9.12323058e-01 9.21565473e-01 3.08136791e-01 -2.32005000e-01
-1.39120018e+00 -5.39937377e-01 -8.29608142e-01 -1.51640788e-01
4.55176473e-01 -5.95213234e-01 7.98419178e-01 -1.53051686e+00
-1.35361600e+00 8.16090047e-01 1.15149394e-02 -4.01244730e-01
3.48731786e-01 3.41400802e-02 -3.57106835e-01 4.23188061e-01
1.86957851e-01 9.64934886e-01 1.14204681e+00 -1.26904476e+00
-6.46946788e-01 -3.52588594e-01 8.58451873e-02 2.67744958e-01
-5.23666501e-01 -2.30322838e-01 -2.15266287e-01 -4.54835683e-01
2.61113435e-01 -1.06504166e+00 -1.61415525e-02 1.45645559e-01
-4.25514877e-02 -3.40312093e-01 1.13349617e+00 -2.90840715e-01
6.97447598e-01 -2.17565823e+00 8.36194992e-01 -1.34954020e-01
-1.21612556e-01 6.29773363e-02 -4.35260296e-01 5.45855388e-02
3.20593379e-02 -1.93568900e-01 -3.91344249e-01 -1.81147963e-01
3.47712785e-02 7.21863329e-01 -4.08830643e-01 8.37797403e-01
2.80927300e-01 1.07251060e+00 -1.27248514e+00 -2.81292409e-01
4.61527646e-01 5.98412752e-01 -6.01368904e-01 1.82210416e-01
-3.67583096e-01 5.63875139e-01 -4.25398409e-01 4.65104431e-01
2.06882760e-01 -2.27085948e-01 -1.41099915e-01 1.91346128e-02
1.86729372e-01 -1.19426757e-01 -9.67743456e-01 1.72288990e+00
-2.39057586e-01 3.55067074e-01 -1.34415910e-01 -1.43533373e+00
6.88326240e-01 2.43630722e-01 5.46140969e-01 -4.81187642e-01
-6.32856935e-02 2.52209336e-01 -2.05438524e-01 -7.05647886e-01
8.65303650e-02 -6.29293025e-01 1.04886346e-01 4.03537750e-01
6.34909749e-01 2.33094934e-02 2.54407339e-02 1.42884895e-01
1.01055992e+00 5.24706364e-01 -1.24807730e-01 -3.72051239e-01
6.78877831e-01 2.39360109e-02 3.07492256e-01 8.09397519e-01
-1.95994809e-01 6.55302346e-01 3.21317911e-01 -3.95592779e-01
-1.05338383e+00 -1.22448313e+00 -1.90094754e-01 1.33714294e+00
3.53902161e-01 5.18773682e-02 -8.40746641e-01 -1.09374094e+00
5.49128242e-02 7.93500066e-01 -1.02666759e+00 -5.93187332e-01
-4.49600518e-01 -3.79687995e-01 2.07614571e-01 4.48387116e-01
1.17708825e-01 -1.39463258e+00 -6.81262553e-01 5.76275848e-02
-1.46849444e-02 -7.46033311e-01 -3.76456290e-01 1.55879200e-01
-7.44142473e-01 -1.13472962e+00 -1.03065240e+00 -5.16077280e-01
8.32851946e-01 1.99788257e-01 9.05476391e-01 -1.25458986e-01
-5.13214290e-01 9.98414338e-01 -4.64474767e-01 1.29506215e-01
-4.64092821e-01 -4.42041278e-01 4.65821296e-01 5.75974166e-01
5.38971007e-01 -2.68918991e-01 -4.88222539e-01 5.70698082e-01
-9.33749616e-01 -3.24484944e-01 1.66665584e-01 9.36676085e-01
5.36631167e-01 -3.35160524e-01 5.54467916e-01 -3.38800669e-01
3.41854319e-02 -6.07223094e-01 -3.01419228e-01 3.74844551e-01
-1.51980296e-01 2.24831715e-01 6.86607063e-01 -7.83437550e-01
-9.70813155e-01 -5.52974232e-02 2.28602067e-01 -9.36748922e-01
-3.17124575e-01 -1.57366499e-01 -9.82628539e-02 -2.77699620e-01
8.66569936e-01 2.26375073e-01 6.19461797e-02 -3.84261191e-01
6.68377340e-01 5.92143059e-01 3.79939139e-01 -5.26571035e-01
4.38393474e-01 9.46683645e-01 4.66643646e-02 -1.18646383e+00
-9.85145867e-01 -6.66220903e-01 -8.40698957e-01 -4.58021611e-01
1.45036066e+00 -4.93513316e-01 -2.58437186e-01 1.32503808e-01
-9.66927469e-01 -5.18944025e-01 -1.05755568e+00 7.09259152e-01
-1.19841874e+00 3.53648305e-01 -3.50059509e-01 -8.04544091e-01
2.55111486e-01 -1.05030155e+00 1.27695596e+00 -1.61093548e-01
-1.97586298e-01 -1.10774612e+00 4.98058125e-02 1.04585096e-01
7.34223723e-02 -9.53792334e-02 7.68507361e-01 -6.98494613e-01
-6.16350114e-01 -1.58329219e-01 2.71573495e-02 6.90667927e-01
-5.56908548e-02 -3.30771416e-01 -9.92735744e-01 -2.58270472e-01
1.49699703e-01 -5.22342443e-01 1.07018256e+00 4.60635573e-01
9.00400698e-01 -6.13261573e-02 -4.61194813e-01 6.52384341e-01
1.32116008e+00 1.17316440e-01 5.31919956e-01 4.16601710e-02
7.55215764e-01 7.31315792e-01 6.17024064e-01 1.91476464e-01
-1.85478777e-01 6.62529707e-01 3.70308638e-01 1.85118958e-01
-2.27219433e-01 -3.15717459e-01 5.49615800e-01 4.99708980e-01
-2.68776506e-01 -2.33255193e-01 -6.22354925e-01 6.53153181e-01
-2.02896762e+00 -1.24163079e+00 2.30331093e-01 2.46898174e+00
2.19761282e-01 -7.12379664e-02 5.52610345e-02 -1.92596972e-01
7.33275235e-01 7.81662017e-02 -6.22994542e-01 -1.55557021e-01
-1.38583034e-01 -1.30971428e-02 4.30852324e-01 5.26225924e-01
-1.16252291e+00 9.65333819e-01 5.91813183e+00 6.89026833e-01
-9.52547014e-01 4.56374824e-01 3.33113551e-01 -2.65611291e-01
-2.21912950e-01 9.98347849e-02 -7.04807818e-01 3.59745651e-01
9.50683653e-01 -1.24774903e-01 6.05531991e-01 8.79570365e-01
-7.30829537e-02 9.21122730e-02 -1.46925855e+00 8.15173447e-01
4.48018610e-01 -1.31620610e+00 2.77016371e-01 1.41917571e-01
8.42515945e-01 1.52381867e-01 -2.04407245e-01 5.23490608e-01
1.30582556e-01 -7.37312198e-01 7.62875378e-01 8.32413316e-01
6.53349757e-01 -3.48328114e-01 3.99667323e-01 3.59512955e-01
-9.07823145e-01 -3.82574677e-01 -8.15550685e-01 5.88292554e-02
1.61499828e-01 4.64838604e-03 -2.21017569e-01 1.96708187e-01
4.98338580e-01 9.83351469e-01 -2.23934963e-01 6.76023781e-01
-2.16603628e-04 2.05163345e-01 -1.86059132e-01 2.88885403e-02
5.48080206e-01 -5.21377861e-01 8.37889373e-01 1.02751100e+00
4.52774614e-01 2.59041518e-01 3.15197825e-01 9.19365525e-01
5.06922603e-02 -3.44109815e-03 -1.00902128e+00 -1.14745937e-01
-2.81192064e-01 9.24843729e-01 -8.74200165e-01 -7.61752903e-01
-4.69073325e-01 1.28752124e+00 2.82121450e-01 6.97958350e-01
-8.81261706e-01 -1.23310290e-01 8.48609507e-01 9.39671025e-02
5.25069952e-01 4.60044295e-02 2.91197121e-01 -1.39293754e+00
-1.57965288e-01 -3.34390521e-01 4.62013185e-01 -8.52323472e-01
-1.21486688e+00 3.94034684e-01 1.71221033e-01 -1.27669597e+00
-1.66831374e-01 -7.69926310e-01 -4.28800970e-01 5.16556442e-01
-9.02172029e-01 -9.29422736e-01 -3.73321027e-02 7.04853237e-01
8.48050296e-01 -2.55232126e-01 8.06131601e-01 9.00289118e-02
-1.24057516e-01 5.38141355e-02 2.98489094e-01 -2.21742541e-01
3.97300065e-01 -1.08972943e+00 2.17548981e-02 6.51890755e-01
3.72945875e-01 3.29337955e-01 7.06306577e-01 -6.08143866e-01
-1.35974014e+00 -1.18705654e+00 6.05533123e-01 -8.71315181e-01
9.50216830e-01 -3.85622293e-01 -9.83802199e-01 7.73499846e-01
-3.05541933e-01 5.74348271e-01 2.47193784e-01 -1.42957553e-01
-2.76430637e-01 1.88452490e-02 -1.05289900e+00 6.78498089e-01
1.43028104e+00 -6.81350410e-01 -8.42139959e-01 6.99386835e-01
5.19015133e-01 2.96459407e-01 -5.89093029e-01 3.10595363e-01
4.13103610e-01 -8.90608072e-01 9.56099868e-01 -1.19941044e+00
2.00767830e-01 -2.75746554e-01 -4.29886252e-01 -1.31442893e+00
-1.85524538e-01 -3.27850312e-01 -1.70438930e-01 6.72597468e-01
1.22799784e-01 -5.89988708e-01 7.28342295e-01 3.39149505e-01
-2.51640260e-01 -5.88121891e-01 -1.12842476e+00 -1.20429337e+00
4.54586558e-02 -3.06478918e-01 2.99708471e-02 7.73830652e-01
5.80166057e-02 2.99166799e-01 -3.61211360e-01 -1.21414438e-01
9.18778956e-01 6.53281733e-02 4.10426915e-01 -9.16743934e-01
-3.55623960e-01 -3.56184751e-01 -9.02414680e-01 -9.85948741e-01
4.66081172e-01 -1.05267632e+00 2.94853896e-01 -1.28844416e+00
3.48435700e-01 -2.30747327e-01 -5.00889301e-01 2.61308223e-01
3.38099122e-01 4.65806335e-01 3.04872006e-01 2.69669324e-01
-8.67955148e-01 7.39764392e-01 1.31960690e+00 -1.58504561e-01
-2.26326495e-01 -2.55075514e-01 -2.10264951e-01 8.43330026e-01
4.61649179e-01 -4.58132654e-01 -5.49793124e-01 -2.89659709e-01
-2.86561042e-01 -5.57624400e-02 7.52357066e-01 -1.17376697e+00
-1.08123727e-01 -2.63285339e-01 2.25518912e-01 6.34633703e-03
6.20077193e-01 -6.21820211e-01 -2.87857242e-02 4.71508026e-01
-5.96121132e-01 -6.74898446e-01 -3.85538787e-01 9.24311697e-01
5.31035773e-02 -4.55806643e-01 1.14396405e+00 -3.18539441e-01
-8.50525320e-01 2.92528749e-01 -1.68468535e-01 1.04002260e-01
1.37985671e+00 -5.29753625e-01 -1.66696534e-01 -1.10711887e-01
-1.38745034e+00 -4.75913733e-02 6.22749150e-01 6.24198735e-01
6.69157267e-01 -1.38312888e+00 -4.06324476e-01 5.13510942e-01
2.54920781e-01 -5.87298870e-01 2.90806144e-01 1.01495922e+00
-4.01165932e-02 4.49840844e-01 -3.29256505e-01 -8.54288697e-01
-1.02382970e+00 9.66328442e-01 6.45890236e-01 4.48381066e-01
-9.51856256e-01 1.02686119e+00 7.44426310e-01 -4.05300796e-01
1.94940105e-01 -1.67048171e-01 4.54599084e-03 -6.61152899e-02
5.44984937e-01 4.53743279e-01 -3.10672671e-01 -1.13047123e+00
-3.08593094e-01 8.36541295e-01 2.93054283e-01 -1.04991816e-01
1.16344142e+00 -4.87856120e-02 1.21800890e-02 6.85343683e-01
1.58066106e+00 -6.18382752e-01 -1.59391797e+00 4.00487036e-02
-8.53782296e-02 -6.61179721e-01 -1.05333313e-01 -3.87697637e-01
-8.21009696e-01 9.45097923e-01 7.52413809e-01 2.89872557e-01
7.32103646e-01 6.36076927e-01 4.64904666e-01 4.29922909e-01
5.51924884e-01 -1.21219909e+00 4.17155504e-01 4.26962525e-01
8.87754917e-01 -1.21989465e+00 -3.64799172e-01 -1.72599837e-01
-8.34123969e-01 9.40334857e-01 4.40294683e-01 -3.55448633e-01
5.29072285e-01 -3.45355004e-01 -8.95675272e-02 -4.47144091e-01
-6.77260280e-01 -4.56526250e-01 3.40225339e-01 7.01585889e-01
-1.79505736e-01 1.16793618e-01 -5.13770208e-02 8.58112350e-02
4.70491916e-01 4.37542088e-02 2.88748115e-01 8.26502860e-01
-6.90862834e-01 -7.40869343e-01 -3.25324267e-01 4.70549285e-01
1.10636337e-03 2.06953108e-01 -2.05257848e-01 6.59732282e-01
2.70205319e-01 5.74958563e-01 4.75603640e-01 -2.22530782e-01
2.53163785e-01 4.20649052e-01 7.57129729e-01 -7.07406104e-01
1.30882978e-01 -7.88330510e-02 -1.88646808e-01 -7.19732761e-01
-4.45384085e-01 -7.90362418e-01 -1.30791509e+00 2.09729835e-01
-2.35957757e-01 2.16581821e-01 4.97454673e-01 1.02886283e+00
2.16097593e-01 2.24725619e-01 4.69390959e-01 -1.13424373e+00
-8.63033533e-01 -7.65228868e-01 -7.81017959e-01 9.13550735e-01
3.15150499e-01 -1.15206814e+00 -9.49055135e-01 3.21928412e-01] | [8.683539390563965, 1.0571928024291992] |
0c67c273-3ae1-417b-aeba-2f017ded23e6 | efficient-anomaly-detection-using-self | 2111.12379 | null | https://arxiv.org/abs/2111.12379v3 | https://arxiv.org/pdf/2111.12379v3.pdf | Efficient Anomaly Detection Using Self-Supervised Multi-Cue Tasks | Anomaly detection is important in many real-life applications. Recently, self-supervised learning has greatly helped deep anomaly detection by recognizing several geometric transformations. However these methods lack finer features, usually highly depend on the anomaly type, and do not perform well on fine-grained problems. To address these issues, we first introduce in this work three novel and efficient discriminative and generative tasks which have complementary strength: (i) a piece-wise jigsaw puzzle task focuses on structure cues; (ii) a tint rotation recognition is used within each piece, taking into account the colorimetry information; (iii) and a partial re-colorization task considers the image texture. In order to make the re-colorization task more object-oriented than background-oriented, we propose to include the contextual color information of the image border via an attention mechanism. We then present a new out-of-distribution detection function and highlight its better stability compared to existing methods. Along with it, we also experiment different score fusion functions. Finally, we evaluate our method on an extensive protocol composed of various anomaly types, from object anomalies, style anomalies with fine-grained classification to local anomalies with face anti-spoofing datasets. Our model significantly outperforms state-of-the-art with up to 36% relative error improvement on object anomalies and 40% on face anti-spoofing problems. | ['Aymeric Histace', 'Jean Beaudet', 'Ngoc-Son Vu', 'Loic Jezequel'] | 2021-11-24 | null | null | null | null | ['self-supervised-anomaly-detection'] | ['computer-vision'] | [ 2.09097236e-01 -5.38631558e-01 1.72879562e-01 -2.49414846e-01
-4.02443707e-01 -3.91552567e-01 7.36216366e-01 1.38563305e-01
-1.00704357e-01 3.14273447e-01 -8.73950273e-02 -1.63837880e-01
-7.90832192e-02 -8.10702324e-01 -6.37922287e-01 -9.41104233e-01
-2.44158376e-02 3.29771161e-01 2.78131664e-01 -3.87022793e-01
4.66517538e-01 8.56607676e-01 -1.64132512e+00 3.37480247e-01
1.09349048e+00 1.31251979e+00 -4.99927878e-01 3.56214374e-01
-2.55189270e-01 4.78054821e-01 -7.95462728e-01 -6.64043128e-01
2.69176990e-01 -4.16606337e-01 -5.44782877e-01 4.56610531e-01
8.11652541e-01 -2.54119009e-01 2.81214062e-02 1.39083910e+00
4.25158888e-01 -1.97181553e-02 9.11179066e-01 -1.58073878e+00
-9.46300268e-01 -3.80272721e-03 -1.20481777e+00 3.97751480e-01
3.54943842e-01 1.33754909e-01 6.99070632e-01 -8.91322434e-01
2.70959556e-01 1.52985549e+00 5.38490057e-01 4.51407075e-01
-1.01785374e+00 -8.55684817e-01 3.89434487e-01 6.94448650e-01
-1.23384881e+00 -3.36462766e-01 1.36658347e+00 -3.13562840e-01
4.84575152e-01 3.19394797e-01 3.43630552e-01 1.42813313e+00
1.67018607e-01 5.44738352e-01 1.24991632e+00 -3.39618236e-01
1.67296976e-01 -1.98834762e-01 -3.69098224e-02 1.02652466e+00
4.62070882e-01 -5.93325309e-02 -3.32511902e-01 -3.19215924e-01
5.55865824e-01 2.02255547e-01 -2.51749717e-02 -2.98952401e-01
-7.73206472e-01 7.75570154e-01 2.55214959e-01 4.67125773e-01
-3.36488903e-01 -1.72690570e-01 3.57792258e-01 3.47391516e-01
6.40253603e-01 1.40476599e-01 -2.72203833e-01 1.09962784e-01
-6.55061424e-01 5.73124811e-02 4.68497723e-01 4.30398822e-01
7.89785922e-01 4.58224148e-01 -1.92435995e-01 1.05997205e+00
3.35553259e-01 6.71288610e-01 4.37393069e-01 -5.28507769e-01
3.17340255e-01 6.85948789e-01 -3.01640898e-01 -1.60189724e+00
-5.18389285e-01 -4.35171187e-01 -1.19702911e+00 4.50497776e-01
6.63975775e-01 2.31237561e-01 -1.17731571e+00 1.50895321e+00
5.54552019e-01 3.97709817e-01 -3.99829686e-01 7.66011834e-01
4.83522803e-01 3.61453801e-01 -5.69758825e-02 -1.69843376e-01
1.53463912e+00 -8.90303016e-01 -6.14813328e-01 -1.48807773e-02
4.74339396e-01 -1.05367911e+00 9.63124096e-01 8.08466375e-01
-5.96072614e-01 -5.56447327e-01 -9.53984618e-01 2.64074475e-01
-5.52378416e-01 9.00831744e-02 5.57419300e-01 1.12793756e+00
-1.00642908e+00 4.66098547e-01 -6.28710091e-01 -5.22152305e-01
6.48254931e-01 1.89806491e-01 -5.88498533e-01 -1.36148989e-01
-6.27266824e-01 5.75619102e-01 8.09755176e-02 6.52239472e-02
-6.63952708e-01 -3.57294053e-01 -8.46283674e-01 -8.30074623e-02
4.60430205e-01 -3.37100208e-01 4.53512073e-01 -1.19422948e+00
-1.59724724e+00 9.97199774e-01 -1.40603125e-01 -1.19831152e-02
3.95683646e-01 -2.67484903e-01 -9.00369585e-01 1.58116907e-01
-2.27643084e-02 9.62372646e-02 1.35384226e+00 -1.32131612e+00
-4.81962800e-01 -8.01797807e-01 -1.34647623e-01 -3.96898091e-01
-5.46999514e-01 2.28018224e-01 -4.33810860e-01 -1.25399756e+00
3.68624002e-01 -8.00450802e-01 1.38338193e-01 5.60953319e-02
-4.93379146e-01 -2.56617516e-01 1.31236601e+00 -8.13780963e-01
1.16818833e+00 -2.18789434e+00 5.46514103e-03 4.57710028e-01
1.47517860e-01 4.90824193e-01 -4.96109456e-01 4.69172783e-02
-3.29009801e-01 1.79753751e-02 -4.90875512e-01 -3.53297025e-01
-5.78319505e-02 4.65089120e-02 -2.68224478e-01 7.23845601e-01
5.84795415e-01 5.20554185e-01 -5.41515768e-01 -5.13643026e-01
2.69281775e-01 4.00142759e-01 -7.62438118e-01 1.84787270e-02
-6.88714683e-02 5.56818247e-01 -4.77714241e-01 1.28578758e+00
1.20600522e+00 -1.01188002e-02 -4.00086492e-01 -3.68617684e-01
2.67114609e-01 -4.12750065e-01 -1.23675978e+00 1.54874098e+00
-2.84162432e-01 2.66255289e-01 1.08618185e-01 -1.21539271e+00
1.03865814e+00 -1.03126578e-01 6.92700803e-01 -8.52006912e-01
1.63773581e-01 2.00455084e-01 -8.43462627e-03 -5.27368546e-01
1.96374625e-01 2.19273761e-01 1.47380739e-01 4.31512803e-01
1.00658208e-01 3.26896757e-01 9.04863998e-02 6.36021867e-02
1.01081455e+00 1.32361203e-01 1.88013583e-01 -2.67655820e-01
9.95949328e-01 -5.02483785e-01 6.48567200e-01 5.94384968e-01
-3.90661806e-01 7.97854841e-01 6.46518469e-01 -7.08072484e-01
-7.78819323e-01 -9.24437940e-01 -1.77599058e-01 1.16572976e+00
2.27860719e-01 -2.95075983e-01 -8.87677729e-01 -1.13782299e+00
1.46884233e-01 3.95374089e-01 -8.52281868e-01 -1.85792103e-01
-6.06315672e-01 -1.29534519e+00 6.20745897e-01 3.06745261e-01
8.57111275e-01 -1.01731062e+00 4.72988337e-02 -1.58017501e-01
-1.53281957e-01 -1.11658823e+00 -4.15535063e-01 -4.07082349e-01
-6.63797259e-01 -1.31742787e+00 -5.35535514e-01 -4.71497566e-01
7.74047673e-01 2.01497421e-01 8.59931827e-01 4.02110428e-01
-6.94032907e-01 4.60558951e-01 -5.46757996e-01 -3.31601948e-01
-1.17101856e-01 -2.97726274e-01 1.91116452e-01 8.57416451e-01
5.77977598e-01 -7.00014591e-01 -5.56118429e-01 5.19000232e-01
-1.05953550e+00 -5.31311333e-01 6.23334408e-01 8.79468024e-01
3.76695037e-01 3.82924192e-02 2.94239461e-01 -7.80517638e-01
3.93721163e-01 -3.73453736e-01 -4.94132429e-01 3.10507178e-01
-4.60141748e-01 1.09840393e-01 5.21511197e-01 -4.71842200e-01
-1.12226927e+00 -7.57262036e-02 -4.19939101e-01 -4.14490223e-01
-5.13762116e-01 -2.27324784e-01 -4.92367446e-01 -4.16946292e-01
6.44681036e-01 1.67486474e-01 3.15079875e-02 -6.75563455e-01
7.67003298e-02 3.98754179e-01 4.10584360e-01 -9.13699567e-01
1.19322193e+00 7.88315654e-01 3.15764248e-01 -1.11290336e+00
-4.25341040e-01 -2.46157914e-01 -6.17569745e-01 -3.51853184e-02
8.42136979e-01 -4.63259995e-01 -7.72981048e-01 1.06715190e+00
-1.19882953e+00 9.87639129e-02 -3.61257611e-04 1.50162041e-01
-1.08973779e-01 1.03637218e+00 -5.83033621e-01 -8.02591145e-01
-1.51757345e-01 -1.12975645e+00 1.24763942e+00 1.18968621e-01
2.98149318e-01 -7.38627553e-01 2.60079522e-02 3.04253191e-01
5.30004799e-01 4.76613373e-01 9.29206371e-01 -6.95650518e-01
-4.09333259e-01 -6.46698773e-02 -5.48579276e-01 4.56293195e-01
3.86938542e-01 2.06075966e-01 -1.17472196e+00 -3.77770841e-01
3.53667289e-02 -5.83716705e-02 1.03765810e+00 -8.35313834e-03
1.73083353e+00 -3.17378670e-01 -2.08622783e-01 8.69765282e-01
1.15838683e+00 1.32596821e-01 7.36524284e-01 3.94060284e-01
1.09848404e+00 7.55222976e-01 4.32064831e-01 6.11277103e-01
7.88445547e-02 8.66829813e-01 7.26614714e-01 -1.86230317e-01
-2.14960232e-01 7.08430856e-02 4.62434590e-01 3.48280489e-01
-3.69190961e-01 -2.70169646e-01 -5.09391546e-01 1.48434371e-01
-1.63699412e+00 -1.03351724e+00 -6.48224875e-02 1.97856975e+00
2.23970085e-01 -4.95603494e-02 2.68766880e-01 3.05186361e-01
7.29100049e-01 3.51003677e-01 -4.41446543e-01 -3.35233480e-01
-4.69782859e-01 2.96987563e-01 1.67253256e-01 2.45475575e-01
-1.37769961e+00 9.97858346e-01 5.19949055e+00 1.22259796e+00
-1.23785317e+00 2.21510604e-01 7.97922194e-01 2.79859126e-01
-1.98442012e-01 -3.58969063e-01 -4.89540845e-01 7.59632111e-01
2.07665682e-01 6.39018416e-01 2.41508186e-01 7.24856138e-01
-2.07153097e-01 6.45930171e-02 -6.79262400e-01 1.17794764e+00
5.64159036e-01 -7.48137653e-01 4.09900665e-01 2.12388009e-01
5.97479343e-01 -4.24696892e-01 2.31749624e-01 5.82951047e-02
2.58875098e-02 -1.00735271e+00 5.12044489e-01 2.80744612e-01
5.39596260e-01 -8.13446522e-01 6.63767219e-01 -2.03676775e-01
-1.25005305e+00 -2.23279551e-01 -4.35677260e-01 2.00925604e-01
-1.09487921e-01 7.92566538e-01 -2.33176067e-01 7.42641449e-01
8.73432875e-01 5.10035455e-01 -9.91762757e-01 9.98189390e-01
-2.12158203e-01 5.82871735e-01 -3.81884426e-01 3.04611981e-01
1.45503625e-01 -2.33688295e-01 5.97197533e-01 1.24839377e+00
4.69619811e-01 -1.67445559e-02 1.11069620e-01 6.10497296e-01
1.11508071e-01 3.50959867e-01 -5.06316185e-01 2.00196624e-01
6.95582926e-02 1.34174562e+00 -8.18519592e-01 3.48093398e-02
-4.37755704e-01 1.33584476e+00 9.54793319e-02 4.38862979e-01
-8.38429928e-01 -3.15759093e-01 9.41748142e-01 9.77247655e-02
2.87607074e-01 -5.87394498e-02 -4.15242091e-02 -1.35070002e+00
2.13152602e-01 -1.09014606e+00 6.72743261e-01 -4.72153515e-01
-1.61706197e+00 7.37379730e-01 -2.84661025e-01 -1.18878174e+00
-1.17867343e-01 -1.07768548e+00 -8.18560839e-01 4.01466936e-01
-1.52075005e+00 -1.51777971e+00 -4.89052594e-01 8.65678728e-01
4.27150428e-01 -5.92908382e-01 6.83849931e-01 6.19549394e-01
-9.72985089e-01 9.57784414e-01 -8.54979083e-03 3.64104897e-01
9.48443770e-01 -1.04872298e+00 3.45587194e-01 1.24951231e+00
3.41513902e-01 3.62291276e-01 3.57905746e-01 -6.91804290e-01
-1.25356972e+00 -9.82969046e-01 2.44327456e-01 -3.50331098e-01
4.89900529e-01 -2.49259710e-01 -1.14411736e+00 3.58690679e-01
-6.79482147e-02 4.37717736e-01 4.55995381e-01 1.90309525e-01
-9.19455290e-01 -4.17591959e-01 -1.46308565e+00 3.55415255e-01
1.17844057e+00 -3.53744507e-01 -2.18247026e-01 1.82910234e-01
2.49776006e-01 -1.36966065e-01 -2.31888503e-01 6.32200480e-01
4.46289331e-01 -1.28656006e+00 1.10954797e+00 -6.81334376e-01
2.17643902e-01 -4.49858814e-01 -2.16265202e-01 -1.10636473e+00
-5.38371027e-01 -5.03367007e-01 -1.71604604e-01 1.41626501e+00
-3.20828594e-02 -8.71464610e-01 7.84495652e-01 -3.80427502e-02
-8.03386793e-02 -6.09985232e-01 -1.01271892e+00 -6.76709175e-01
-2.32986838e-01 -5.01666129e-01 8.01835716e-01 1.11166239e+00
-4.69438761e-01 -2.53240466e-01 -4.91627008e-01 5.44845879e-01
8.86980295e-01 2.21758075e-02 8.30628991e-01 -1.42723501e+00
-6.35539442e-02 -6.98354602e-01 -9.04790103e-01 -5.86917400e-01
1.42710119e-01 -4.85327572e-01 -4.06979471e-01 -7.58002579e-01
5.50386906e-02 -2.41491079e-01 -5.61654329e-01 5.03884614e-01
-2.67681628e-01 8.43639493e-01 -7.57750310e-03 -1.24764191e-02
-6.72373831e-01 5.91586471e-01 9.81956184e-01 -3.30768585e-01
1.89430714e-01 1.69689450e-02 -6.61339164e-01 9.93373215e-01
6.45288229e-01 -2.28631511e-01 -1.13789607e-02 -3.88939887e-01
-1.88385695e-01 -4.44473207e-01 6.09909117e-01 -1.34017563e+00
-9.52565521e-02 -1.29684865e-01 7.00236559e-01 -4.10928994e-01
2.42021441e-01 -8.88096988e-01 -2.60883808e-01 4.13655609e-01
2.52243489e-01 2.77491450e-01 2.80137569e-01 7.23204792e-01
-2.73980856e-01 3.22362781e-03 1.01035511e+00 2.28385463e-01
-8.85115206e-01 5.39726615e-01 -9.17193219e-02 -6.31081015e-02
1.07400286e+00 -2.34653726e-01 -4.50207323e-01 -1.82200924e-01
-4.96352553e-01 -2.53926933e-01 5.42899132e-01 6.89063549e-01
6.06217563e-01 -1.42677212e+00 -8.30051303e-01 7.78605580e-01
3.21024388e-01 -3.91743124e-01 5.92919588e-01 7.42414832e-01
-6.47272706e-01 9.36105102e-02 -6.20129883e-01 -7.87170231e-01
-1.41332078e+00 7.28691697e-01 1.79681316e-01 -1.64325207e-01
-4.25569922e-01 8.99677992e-01 4.59231347e-01 -1.56406671e-01
2.41754413e-01 -3.03828213e-02 -4.32014823e-01 1.17965899e-01
7.29078531e-01 6.10166490e-01 2.50864565e-01 -9.40304160e-01
-5.06901145e-01 1.12579679e+00 -1.25852734e-01 2.25864187e-01
1.14720643e+00 9.42951813e-03 -3.82868797e-01 -2.01546296e-01
1.11552656e+00 4.28139299e-01 -1.04488385e+00 -4.45783257e-01
5.60750104e-02 -7.82161832e-01 -2.71897733e-01 -6.66821599e-01
-1.50569105e+00 9.61338341e-01 1.00991046e+00 2.58956999e-01
1.45914733e+00 -2.24897236e-01 7.24307716e-01 2.33786747e-01
3.55617166e-01 -8.39117289e-01 5.94686925e-01 2.79698402e-01
8.43928635e-01 -1.53218508e+00 6.60497099e-02 -5.96440136e-01
-4.56136197e-01 1.18536389e+00 8.45874667e-01 -1.34490162e-01
6.13827109e-01 2.08490640e-02 5.71794808e-02 -2.93400556e-01
5.52813290e-03 -2.28284299e-01 6.54838264e-01 7.33069777e-01
1.60861626e-01 -1.80574149e-01 -8.23285952e-02 4.57826436e-01
5.94465174e-02 -4.98434365e-01 6.11509122e-02 6.33600771e-01
-3.50175828e-01 -1.30407429e+00 -8.46717596e-01 2.88447320e-01
-6.86585009e-01 2.27261826e-01 -4.29959595e-01 6.70457065e-01
4.36258197e-01 1.00654399e+00 1.74962237e-01 -5.16474068e-01
2.36818612e-01 1.79489832e-02 5.24765790e-01 -2.13977739e-01
-3.01951945e-01 -2.24717204e-02 -2.49939322e-01 -9.13498402e-01
-2.49043986e-01 -5.41108847e-01 -6.82809114e-01 -5.65647185e-01
-2.17538983e-01 3.20948213e-02 5.94424427e-01 1.01420808e+00
3.78817022e-01 3.97436023e-01 8.55430841e-01 -8.43828440e-01
-2.41299853e-01 -9.55559254e-01 -6.48437500e-01 8.83523822e-01
4.67598259e-01 -9.50967133e-01 -5.03058672e-01 -2.58061290e-01] | [12.839604377746582, 1.0509196519851685] |
095867b4-3f63-40fc-ac35-ac3fa9376ac2 | extracting-or-guessing-improving-faithfulness | 2210.04992 | null | https://arxiv.org/abs/2210.04992v2 | https://arxiv.org/pdf/2210.04992v2.pdf | Extracting or Guessing? Improving Faithfulness of Event Temporal Relation Extraction | In this paper, we seek to improve the faithfulness of TempRel extraction models from two perspectives. The first perspective is to extract genuinely based on contextual description. To achieve this, we propose to conduct counterfactual analysis to attenuate the effects of two significant types of training biases: the event trigger bias and the frequent label bias. We also add tense information into event representations to explicitly place an emphasis on the contextual description. The second perspective is to provide proper uncertainty estimation and abstain from extraction when no relation is described in the text. By parameterization of Dirichlet Prior over the model-predicted categorical distribution, we improve the model estimates of the correctness likelihood and make TempRel predictions more selective. We also employ temperature scaling to recalibrate the model confidence measure after bias mitigation. Through experimental analysis on MATRES, MATRES-DS, and TDDiscourse, we demonstrate that our model extracts TempRel and timelines more faithfully compared to SOTA methods, especially under distribution shifts. | ['Dan Roth', 'Muhao Chen', 'Jacob R. Gardner', 'Yuqian Deng', 'Hongming Zhang', 'Haoyu Wang'] | 2022-10-10 | null | null | null | null | ['temporal-relation-extraction', 'temporal-relation-classification'] | ['natural-language-processing', 'natural-language-processing'] | [ 9.51146260e-02 4.22119439e-01 -4.86148864e-01 -6.95099235e-01
-5.61620831e-01 -6.42791450e-01 9.87928569e-01 4.12984252e-01
-4.78310645e-01 1.03747630e+00 6.17800713e-01 -3.14445943e-01
3.49588841e-02 -9.76961672e-01 -7.69597054e-01 -3.39440763e-01
2.24000290e-01 2.98502922e-01 -5.85335828e-02 1.74325317e-01
3.66838276e-01 3.06963295e-01 -1.23736501e+00 3.12036842e-01
1.09623075e+00 6.91107392e-01 -3.24142009e-01 7.82555528e-03
1.59909725e-01 8.96257162e-01 -6.26144350e-01 -6.03919089e-01
-4.29947935e-02 -4.30139974e-02 -4.90232229e-01 -1.87144369e-01
2.00309511e-02 -4.34815675e-01 -1.11384682e-01 1.04508042e+00
2.80156136e-01 8.43943432e-02 1.08621705e+00 -1.11558187e+00
-5.59929550e-01 1.39947522e+00 -6.66463554e-01 2.46804148e-01
1.55841634e-01 1.38413668e-01 1.03747988e+00 -7.79692471e-01
6.32145405e-01 1.35353160e+00 6.86092436e-01 2.27807269e-01
-1.47309828e+00 -9.03560042e-01 5.64596295e-01 1.23531267e-01
-1.46705365e+00 -5.33284187e-01 8.95163894e-01 -4.44879323e-01
9.04675424e-01 2.39423066e-01 4.86535281e-01 1.75441241e+00
4.46715653e-01 5.85546076e-01 1.44190359e+00 -4.68501449e-01
6.08144164e-01 3.76548529e-01 3.03184330e-01 2.64043629e-01
6.91707194e-01 4.47432488e-01 -6.57179356e-01 -4.07428414e-01
6.00297928e-01 -1.01180770e-01 -2.05839217e-01 1.03346817e-01
-1.08347023e+00 8.01771700e-01 5.08686937e-02 2.30853230e-01
-4.51529562e-01 5.33408746e-02 2.80849904e-01 -1.43102974e-01
8.06015790e-01 5.43229878e-01 -5.06226778e-01 -1.05902359e-01
-1.10201526e+00 2.87997067e-01 7.76270807e-01 9.69490409e-01
7.34209657e-01 3.25405821e-02 -5.49641848e-01 6.02008104e-01
3.08520317e-01 5.43524206e-01 4.24731731e-01 -8.03724825e-01
2.65259951e-01 3.96822184e-01 4.10317689e-01 -9.70471501e-01
-2.96492457e-01 -5.92997015e-01 -7.44462252e-01 -3.00974578e-01
3.18311393e-01 -1.93783224e-01 -1.07843804e+00 1.82898188e+00
1.42106593e-01 2.54607741e-02 -6.62776902e-02 7.42912829e-01
4.28570032e-01 5.58918178e-01 5.22638261e-01 -5.65584540e-01
1.39156485e+00 -2.95782477e-01 -1.13107109e+00 -3.99641573e-01
6.37175322e-01 -5.16112089e-01 1.02118087e+00 3.10851187e-01
-8.10603499e-01 -2.04799742e-01 -1.16712797e+00 -1.29566090e-02
-2.46671334e-01 1.35764837e-01 7.98528492e-01 5.46406984e-01
-2.81873733e-01 6.82788670e-01 -1.04861188e+00 -6.12757988e-02
8.75711367e-02 -1.30982339e-01 1.06151260e-01 3.43355209e-01
-1.68209767e+00 1.15430653e+00 6.16614997e-01 -1.36851370e-01
-7.66665757e-01 -8.04684341e-01 -9.46928024e-01 2.32215658e-01
4.64393646e-01 -5.61363578e-01 1.14202011e+00 -6.41133428e-01
-1.24315464e+00 6.35738730e-01 -2.11890727e-01 -6.63667262e-01
5.05998790e-01 -2.87897646e-01 -5.17849326e-01 -1.22409143e-01
1.07125521e-01 4.32051450e-01 8.72037530e-01 -1.31289232e+00
-2.65487373e-01 -2.33157203e-01 -9.98304486e-02 2.98946407e-02
-3.08072954e-01 -2.06324667e-01 4.24478836e-02 -7.53439724e-01
1.21319033e-01 -7.87736595e-01 -1.14570335e-01 -5.45832694e-01
-6.26367092e-01 -2.63097525e-01 2.26533905e-01 -6.93964005e-01
1.69795358e+00 -2.17418480e+00 -3.01602989e-01 3.60778213e-01
4.32638079e-02 -3.14675629e-01 3.67342353e-01 3.75373751e-01
-1.07671320e-01 4.42373782e-01 -3.08403581e-01 -2.25092024e-01
1.18214443e-01 3.49666417e-01 -8.93343329e-01 4.30605292e-01
2.51591384e-01 5.27681231e-01 -6.74873412e-01 -6.64811671e-01
-1.02806846e-02 1.56848222e-01 -7.63910651e-01 9.21025202e-02
-3.21361810e-01 1.26721770e-01 -3.01554888e-01 3.62011909e-01
7.64081120e-01 -2.53209949e-01 4.39047873e-01 -4.75311369e-01
-2.40123063e-01 9.19975042e-01 -1.12666523e+00 1.37253070e+00
-4.07675326e-01 1.33223519e-01 -3.88592601e-01 -6.68100059e-01
7.83425808e-01 2.19511509e-01 2.31765836e-01 -4.49328899e-01
7.19038099e-02 -8.44705254e-02 -1.30703151e-01 -2.17556760e-01
7.55607843e-01 -3.63799334e-01 -3.46240044e-01 4.49638367e-01
-1.02621969e-03 -2.24455610e-01 5.23929717e-03 3.49568665e-01
5.15516758e-01 3.79627645e-01 5.73568523e-01 -4.84871954e-01
-9.62373540e-02 3.54060382e-02 1.06910706e+00 8.21043491e-01
-8.28289241e-02 2.43604422e-01 6.40207708e-01 -1.82466641e-01
-9.23969507e-01 -1.21578836e+00 -6.58339560e-01 8.20158124e-01
-3.96436974e-02 -6.18063569e-01 -5.21514535e-01 -9.31477010e-01
2.89346091e-02 1.72864401e+00 -7.02250719e-01 -4.03812796e-01
-3.66132647e-01 -1.04770088e+00 5.27584970e-01 6.64854288e-01
3.86551023e-01 -4.59042877e-01 -5.48629224e-01 1.74128920e-01
-5.22436738e-01 -8.77471864e-01 -3.48885268e-01 3.17325324e-01
-7.66438425e-01 -6.28660619e-01 -1.85886145e-01 1.10418126e-02
4.84953851e-01 -4.96942043e-01 1.11008239e+00 -4.92769539e-01
3.14973384e-01 1.57931656e-01 -1.33499935e-01 -6.99916124e-01
-3.89543921e-01 4.92378213e-02 1.53927878e-01 -2.69950151e-01
6.97316706e-01 -7.00198352e-01 -5.15571237e-01 2.18394369e-01
-9.07396376e-01 1.30849689e-01 3.45115572e-01 7.76703537e-01
4.81040567e-01 -4.69694249e-02 7.21373320e-01 -1.07546425e+00
4.89884645e-01 -7.16519892e-01 -4.81161445e-01 2.61049926e-01
-1.33366895e+00 3.54315132e-01 2.64750779e-01 -7.39279330e-01
-1.66698730e+00 -3.10860544e-01 1.22841336e-01 -1.36947826e-01
-3.93914059e-02 6.56950414e-01 -1.10029079e-01 8.45988333e-01
9.53237355e-01 -7.25956075e-03 -4.98269647e-01 -2.43616164e-01
4.63835359e-01 6.59338474e-01 3.02343160e-01 -1.06734729e+00
4.29848313e-01 5.32031059e-01 -3.68093133e-01 -5.34409344e-01
-1.23150170e+00 -7.25759864e-02 -3.01363498e-01 -3.68378274e-02
5.58151007e-01 -1.00443983e+00 -4.80523735e-01 -3.81700732e-02
-1.15663564e+00 -1.68334544e-01 -3.27645570e-01 7.85131037e-01
-3.63541484e-01 2.53614724e-01 -6.12742662e-01 -1.07966661e+00
-2.39922643e-01 -6.07955635e-01 9.38835800e-01 1.81247443e-01
-6.64598584e-01 -9.21932042e-01 1.43074960e-01 -1.55586064e-01
1.68849424e-01 2.00765744e-01 1.01729858e+00 -1.00357437e+00
-3.97639990e-01 4.48228158e-02 -4.57793325e-02 1.16312765e-01
9.31621790e-02 2.96994686e-01 -1.20068622e+00 -1.45223353e-03
2.78270692e-01 -1.03617452e-01 1.04474449e+00 4.11386937e-01
1.08574831e+00 -6.84846461e-01 -5.77631950e-01 2.92915016e-01
1.19944632e+00 1.23783298e-01 6.82566583e-01 7.48700127e-02
3.15402001e-01 7.06794202e-01 7.59165168e-01 7.21548021e-01
5.41698217e-01 3.70643914e-01 8.73723775e-02 2.58683980e-01
1.63855404e-01 -7.84209430e-01 3.61788005e-01 6.91227555e-01
4.30183224e-02 -2.38693386e-01 -8.99527729e-01 7.90385425e-01
-1.83460653e+00 -9.85672593e-01 -2.14482937e-02 2.18440700e+00
1.44053328e+00 4.48981822e-01 -1.86654925e-01 -3.29029374e-02
7.03063190e-01 3.15248936e-01 -3.43875229e-01 -2.02638388e-01
7.08667003e-03 -8.50376561e-02 4.61924106e-01 5.24541855e-01
-7.20350266e-01 9.37307239e-01 6.71780300e+00 7.72822440e-01
-8.41877699e-01 -3.05491611e-02 7.71645248e-01 -1.79893255e-01
-8.07997465e-01 4.26963687e-01 -1.02804613e+00 5.72982967e-01
1.15611672e+00 -3.59447509e-01 1.14320613e-01 7.28216469e-01
1.82794839e-01 -3.93200815e-01 -1.54639554e+00 5.72599232e-01
-1.89724877e-01 -1.03343368e+00 1.28035948e-01 2.19281595e-02
6.29341304e-01 -5.95821381e-01 -1.06805906e-01 5.06279647e-01
6.03712201e-01 -8.55688095e-01 1.03569496e+00 8.01241934e-01
6.24973834e-01 -7.73171365e-01 5.39338768e-01 4.73112315e-01
-7.02973783e-01 7.09653497e-02 -3.59720349e-01 -2.18876809e-01
1.30093053e-01 1.33988094e+00 -1.14886582e+00 5.08909404e-01
4.69220608e-01 5.41327000e-01 -2.36873820e-01 4.93229508e-01
-6.09747291e-01 1.02765346e+00 -4.17008132e-01 -2.16365065e-02
-2.52777070e-01 -3.34147527e-03 5.64659476e-01 1.35516512e+00
2.67737359e-01 1.43414721e-01 -6.42852932e-02 1.41976368e+00
5.14579825e-02 -2.36086369e-01 -6.95290148e-01 -1.10638380e-01
1.01239634e+00 9.57018852e-01 -4.47187811e-01 -5.41831911e-01
-9.78543311e-02 6.05048954e-01 2.63335347e-01 6.18012309e-01
-9.96860445e-01 -5.03258668e-02 3.04840177e-01 9.25498754e-02
3.19466814e-02 -1.78143874e-01 -7.08695590e-01 -1.45750487e+00
-7.64964819e-02 -6.91037476e-01 5.39730430e-01 -7.02120364e-01
-1.37194860e+00 1.52025983e-01 5.82801819e-01 -8.47227275e-01
-5.75943768e-01 -1.92446247e-01 -4.10910338e-01 8.44936430e-01
-1.31280422e+00 -9.81501639e-01 2.81486332e-01 2.42849782e-01
2.19315156e-01 4.13329393e-01 7.33395278e-01 5.62977828e-02
-5.12690663e-01 5.75517833e-01 -3.40489417e-01 1.74827389e-02
1.21920812e+00 -1.27348971e+00 1.24239750e-01 9.14523005e-01
-9.08349454e-02 1.15983260e+00 1.14386654e+00 -1.09604800e+00
-7.92336643e-01 -9.89293158e-01 1.10628784e+00 -5.94856441e-01
7.28485882e-01 -3.74185383e-01 -9.00280476e-01 1.07827199e+00
1.23817042e-01 -5.16395628e-01 6.83586717e-01 6.49995983e-01
-7.26593614e-01 -1.35728061e-01 -1.20229924e+00 7.10483491e-01
9.11667407e-01 -5.90278864e-01 -1.18082356e+00 1.45862088e-01
9.11358953e-01 -2.92725354e-01 -9.01459813e-01 5.38617134e-01
4.61741000e-01 -6.72304034e-01 6.17168069e-01 -5.70027769e-01
5.92968524e-01 -2.00143903e-01 -3.32794011e-01 -1.41382110e+00
-2.86401838e-01 -3.82859409e-01 -3.58762115e-01 1.63451660e+00
5.74287117e-01 -7.50418782e-01 3.12595725e-01 9.11142051e-01
6.90660402e-02 -3.17060411e-01 -8.36778402e-01 -6.11933887e-01
2.32716367e-01 -4.73611295e-01 8.17364395e-01 1.26152647e+00
3.38843614e-01 4.85869616e-01 -5.49647152e-01 3.39493752e-01
6.63302600e-01 2.64867008e-01 2.37699836e-01 -1.21765578e+00
-2.71692574e-01 -1.57482736e-02 3.79311055e-01 -8.15435886e-01
1.77026317e-01 -6.19017959e-01 4.31417897e-02 -1.17705977e+00
4.28624362e-01 -3.74529183e-01 -3.27783346e-01 6.45060182e-01
-1.87255561e-01 -3.43897164e-01 -2.42354438e-01 1.98471144e-01
-1.90742791e-01 6.63466811e-01 9.75068212e-01 4.06027324e-02
-1.67060211e-01 -1.62743792e-01 -9.79628444e-01 8.10107529e-01
6.85239673e-01 -6.89338624e-01 -4.61895168e-01 -7.89483115e-02
6.92259312e-01 2.31876716e-01 4.30559248e-01 -3.99192810e-01
-1.27376837e-03 -4.87695366e-01 4.24616963e-01 -6.58054173e-01
1.52729586e-01 -7.15471029e-01 9.86856893e-02 1.59593180e-01
-6.45197928e-01 -1.52375683e-01 3.71150583e-01 6.89163744e-01
9.49166864e-02 -1.13242500e-01 6.12167358e-01 -2.48784060e-03
-2.09944680e-01 -2.00999767e-01 -4.94397819e-01 -2.16628350e-02
5.88625729e-01 2.66389519e-01 -6.69764459e-01 -2.52607286e-01
-7.66143024e-01 1.23122931e-01 3.88024747e-01 2.70798564e-01
2.61611462e-01 -1.41521418e+00 -6.76298678e-01 5.59820309e-02
1.54030472e-01 -1.98771715e-01 1.61632657e-01 7.88303614e-01
2.97780365e-01 4.21851426e-01 1.29893884e-01 -5.14914691e-01
-5.83779514e-01 6.37709677e-01 1.49394259e-01 -4.54708815e-01
-3.88847649e-01 3.21767181e-01 2.23153338e-01 -1.55849829e-01
2.17132792e-01 -6.98814392e-01 -2.00798228e-01 2.18675852e-01
3.11327457e-01 2.96185195e-01 2.38589346e-02 3.06039508e-02
-5.18224120e-01 -1.75468743e-01 -3.17772299e-01 -4.67643082e-01
1.08156896e+00 -2.20713153e-01 -2.03213724e-03 7.33707607e-01
5.02256393e-01 4.15395916e-01 -1.32066274e+00 -9.78299677e-02
2.37495095e-01 -2.63023078e-01 1.51346892e-01 -1.24851847e+00
-2.80898124e-01 6.13954544e-01 3.03018093e-01 2.39952728e-01
8.67505848e-01 -1.35892347e-01 2.50010073e-01 1.18833967e-01
2.86909312e-01 -1.36615384e+00 -2.08950251e-01 4.18520480e-01
8.95754695e-01 -1.06484187e+00 2.43869171e-01 -3.29884708e-01
-7.05709755e-01 6.93516791e-01 5.60289621e-01 1.30854756e-01
5.57699621e-01 4.02398586e-01 -2.41495773e-01 -9.98902172e-02
-9.85513568e-01 1.34235799e-01 2.98014551e-01 2.84768581e-01
6.47162259e-01 2.25280270e-01 -7.48331368e-01 1.12989831e+00
-3.88830692e-01 -1.98590085e-01 6.14453971e-01 6.16405189e-01
-2.56536692e-01 -6.98139369e-01 -3.39279622e-01 4.26348031e-01
-6.02291167e-01 -4.42071527e-01 -3.34006608e-01 7.47536361e-01
-1.48209602e-01 1.02512598e+00 1.32644147e-01 -3.03127885e-01
5.75578362e-02 5.29401422e-01 3.90055716e-01 -6.37221098e-01
-3.49160463e-01 2.85766929e-01 5.97149134e-01 -3.06198925e-01
-3.37593377e-01 -6.47232592e-01 -1.16931450e+00 -3.03367436e-01
-4.55525100e-01 2.45378688e-01 6.44367695e-01 1.01968718e+00
3.97532016e-01 6.50780678e-01 4.06979144e-01 -3.37191224e-01
-1.03454590e+00 -1.22966468e+00 -4.55751091e-01 2.70359129e-01
8.73599797e-02 -7.44136393e-01 -6.31257892e-01 2.93004252e-02] | [9.873991966247559, 8.064735412597656] |
d542f93c-4b67-4cab-9797-9ebab6bfee71 | individualized-conditioning-and-negative | 2210.06368 | null | https://arxiv.org/abs/2210.06368v1 | https://arxiv.org/pdf/2210.06368v1.pdf | Individualized Conditioning and Negative Distances for Speaker Separation | Speaker separation aims to extract multiple voices from a mixed signal. In this paper, we propose two speaker-aware designs to improve the existing speaker separation solutions. The first model is a speaker conditioning network that integrates speech samples to generate individualized speaker conditions, which then provide informed guidance for a separation module to produce well-separated outputs. The second design aims to reduce non-target voices in the separated speech. To this end, we propose negative distances to penalize the appearance of any non-target voice in the channel outputs, and positive distances to drive the separated voices closer to the clean targets. We explore two different setups, weighted-sum and triplet-like, to integrate these two distances to form a combined auxiliary loss for the separation networks. Experiments conducted on LibriMix demonstrate the effectiveness of our proposed models. | ['Jundong Liu', 'Li Xu', 'Xianhui Wang', 'Charles D. Smith', 'Zhewei Wang', 'Shuyu Gong', 'Nidal Abuhajar', 'Tao Sun'] | 2022-10-12 | null | null | null | null | ['speaker-separation'] | ['speech'] | [ 2.13375628e-01 1.90932989e-01 1.52336150e-01 -4.99205351e-01
-1.04858768e+00 -4.91906703e-01 2.78841257e-01 -4.07646745e-01
1.48910359e-01 5.53815722e-01 5.60509682e-01 -2.21679822e-01
-2.24834785e-01 -1.58759207e-01 -3.90761316e-01 -9.52586889e-01
1.61975771e-01 -9.48022455e-02 -1.57503188e-01 -2.72403926e-01
-4.97422256e-02 4.33125407e-01 -1.36266363e+00 3.12084734e-01
1.33147061e+00 8.25908124e-01 2.14357361e-01 7.02080309e-01
-6.09923750e-02 5.02387881e-01 -6.86340034e-01 -1.14989758e-01
3.39811474e-01 -8.49646211e-01 -1.39251277e-01 1.14828736e-01
4.35752004e-01 -1.61983725e-02 -1.25645772e-01 1.13896680e+00
9.11254883e-01 2.46458367e-01 6.79610670e-01 -1.11420488e+00
-4.43494380e-01 1.07293332e+00 -6.22133911e-01 1.50317565e-01
1.34219617e-01 7.72089362e-02 9.12701607e-01 -8.31732929e-01
-2.74116755e-01 1.54328287e+00 4.35605377e-01 8.12717497e-01
-1.27256203e+00 -9.56787467e-01 4.28062290e-01 -1.41536370e-01
-1.23956978e+00 -1.21085739e+00 1.09648979e+00 -2.07487673e-01
4.50621128e-01 6.93287313e-01 3.68712157e-01 1.08074450e+00
-3.54372203e-01 6.89442933e-01 7.78916419e-01 -6.69445872e-01
-3.99509398e-03 5.30483425e-01 2.31252402e-01 2.10941702e-01
-1.11419052e-01 1.33267328e-01 -5.49043655e-01 -7.04198778e-02
4.18385744e-01 -2.86223263e-01 -6.94005311e-01 2.09691823e-02
-7.57552445e-01 6.31077647e-01 3.35861146e-01 5.40322483e-01
-3.76292318e-01 -2.81528383e-01 7.50979632e-02 3.71228568e-02
5.57021379e-01 3.85486335e-01 -1.12241916e-01 2.96231717e-01
-1.03516889e+00 1.65494397e-01 7.19301820e-01 9.11518931e-01
5.77675581e-01 5.36765397e-01 -6.12432659e-01 1.25517583e+00
8.43227863e-01 5.15939355e-01 5.91735661e-01 -7.48810351e-01
8.71678829e-01 2.06910178e-01 -8.59431401e-02 -7.25896001e-01
-8.49585310e-02 -8.63279223e-01 -5.98766088e-01 8.59681368e-02
7.39299133e-02 -6.82308316e-01 -9.65508819e-01 2.01633644e+00
2.55989730e-01 5.06775796e-01 3.22824836e-01 1.24259305e+00
7.93389499e-01 8.96565378e-01 -2.77255535e-01 -4.06516165e-01
9.15566802e-01 -1.24706233e+00 -1.00773203e+00 -5.85532725e-01
-5.42270876e-02 -8.59454334e-01 9.03465688e-01 3.47543210e-01
-1.15036321e+00 -6.31965220e-01 -1.25928438e+00 3.87978017e-01
2.21603766e-01 3.24848235e-01 -7.76012316e-02 9.55644369e-01
-8.59145761e-01 3.60896856e-01 -5.65313816e-01 2.18546972e-01
4.06675674e-02 2.56939918e-01 1.27798617e-01 3.52970690e-01
-1.21341872e+00 5.57438493e-01 -1.21149778e-01 5.20455122e-01
-1.03400898e+00 -6.95533156e-01 -8.45060945e-01 4.07412082e-01
2.17738301e-01 -4.14926708e-01 1.26559556e+00 -1.22744441e+00
-1.92476416e+00 3.90048206e-01 -4.37724024e-01 -2.21341357e-01
3.29610705e-01 -2.97821194e-01 -8.09062421e-01 -1.28094807e-01
-4.10752967e-02 3.25419098e-01 1.17497981e+00 -1.67505646e+00
-6.56492949e-01 -2.75718510e-01 -5.07965147e-01 5.41639447e-01
-4.77557063e-01 2.48314943e-02 -3.37372154e-01 -6.88775301e-01
1.50301486e-01 -7.38205254e-01 -2.48516679e-01 -5.16262650e-01
-9.44254279e-01 1.47032216e-01 7.18645334e-01 -9.56535280e-01
1.21169233e+00 -2.58899045e+00 3.67681146e-01 4.56569046e-01
5.69035560e-02 4.38387215e-01 -3.78647298e-01 2.23559335e-01
-2.07465529e-01 -9.70819294e-02 -1.50211290e-01 -7.87190914e-01
8.79685953e-02 -2.57645428e-01 -4.82679874e-01 2.51154780e-01
3.64809364e-01 1.94813788e-01 -3.98547649e-01 -2.97307551e-01
3.37810032e-02 6.43580198e-01 -6.92931473e-01 5.83354890e-01
1.12015247e-01 5.38098693e-01 -1.70980632e-01 2.09367722e-01
8.88238728e-01 6.90298378e-01 3.68766300e-02 -1.99147120e-01
-2.48805001e-01 7.12934554e-01 -1.43050957e+00 1.19043005e+00
-3.19274962e-01 3.81364316e-01 1.03232598e+00 -7.99591362e-01
1.03491640e+00 5.92394352e-01 1.09624125e-01 -2.52375364e-01
2.16554776e-01 2.22375393e-01 3.47724140e-01 -3.57780397e-01
2.77422160e-01 -4.44410115e-01 2.42418110e-01 -2.55863108e-02
1.10750236e-01 -1.24840699e-01 -1.12454861e-01 -3.62042229e-05
5.59417367e-01 -3.27740937e-01 -2.06608534e-01 -2.55309463e-01
8.89598131e-01 -8.47410977e-01 9.55549419e-01 4.20507222e-01
-3.65010530e-01 6.45078003e-01 2.93349952e-01 6.45833313e-01
-3.38337451e-01 -1.24342132e+00 1.87672123e-01 1.10429955e+00
1.45615175e-01 -5.06147370e-02 -9.00391757e-01 -4.17632580e-01
-1.84962422e-01 1.00522816e+00 -4.88882922e-02 -4.05294329e-01
-5.45229018e-01 -2.88878918e-01 6.01864874e-01 3.16427886e-01
1.76682934e-01 -8.27013791e-01 1.55898795e-01 5.30487821e-02
-4.00981843e-01 -6.29051387e-01 -1.11527169e+00 4.40140605e-01
-4.09651309e-01 -3.74151647e-01 -8.25362861e-01 -1.08962643e+00
5.20679712e-01 3.04366767e-01 2.72088587e-01 -4.66498911e-01
2.47655049e-01 -4.37536091e-02 -1.38005897e-01 -6.15879059e-01
-6.36811852e-01 -2.88571436e-02 5.07109940e-01 6.46809638e-01
-8.94184411e-03 -7.50964165e-01 -3.96398515e-01 3.53765696e-01
-5.80625594e-01 -1.56062946e-01 5.87017655e-01 6.01380825e-01
1.38821393e-01 1.94723204e-01 9.46895361e-01 -2.16366529e-01
1.04726660e+00 -3.85228276e-01 -2.15296566e-01 1.03603385e-01
-3.07547539e-01 8.06364268e-02 8.70055497e-01 -8.10052037e-01
-1.45112395e+00 -6.01472221e-02 -3.13567191e-01 -5.93709946e-01
-1.25050917e-01 2.88408846e-01 -8.77873838e-01 2.47710839e-01
3.86039913e-01 1.78524375e-01 8.31536725e-02 -5.58762312e-01
4.85749394e-01 1.09904468e+00 5.51695406e-01 -3.50476474e-01
8.03251147e-01 -7.52362311e-02 -6.61330998e-01 -9.52155292e-01
-5.95589638e-01 -6.02211952e-01 -2.63604790e-01 -9.41953287e-02
4.86217797e-01 -7.99896896e-01 -4.98833865e-01 4.71254289e-01
-1.02095628e+00 -2.23080978e-01 -1.52436391e-01 8.17166030e-01
-2.86181927e-01 1.90346047e-01 -3.84884864e-01 -1.39162958e+00
-5.34002602e-01 -1.22700381e+00 8.01038325e-01 6.19043887e-01
-1.86503619e-01 -6.62010372e-01 1.44499257e-01 3.80016685e-01
4.40974563e-01 -3.45468938e-01 5.28861642e-01 -8.55584800e-01
-7.33937472e-02 7.31103048e-02 2.27824122e-01 8.94127309e-01
6.00509763e-01 -6.29434213e-02 -1.33346725e+00 -3.14447224e-01
3.81253481e-01 9.92335379e-02 8.27916741e-01 6.73379064e-01
5.17807305e-01 -6.30737782e-01 -3.77369165e-01 5.74811339e-01
8.79785240e-01 5.67568839e-01 5.10557592e-01 -3.33745360e-01
6.89827085e-01 8.48320782e-01 2.11753249e-01 2.46458247e-01
-1.32834330e-01 6.27924681e-01 1.76161319e-01 -3.21799725e-01
-3.09563607e-01 -2.73889333e-01 7.26745248e-01 1.25713038e+00
3.47496450e-01 -4.94264632e-01 -4.60135132e-01 7.05665290e-01
-1.59356356e+00 -1.01696420e+00 -2.17540905e-01 2.15182376e+00
9.37200725e-01 1.78670809e-01 2.32521355e-01 2.34212309e-01
1.09010839e+00 2.02863097e-01 -4.25445825e-01 -3.36164683e-01
-1.38163820e-01 6.39912561e-02 1.20112747e-01 8.73202026e-01
-9.06430423e-01 6.27915561e-01 5.99926329e+00 7.80350208e-01
-1.34401548e+00 -1.16656728e-01 2.63078004e-01 -6.06724024e-01
-4.15202320e-01 -8.77198428e-02 -9.03590143e-01 5.15720546e-01
8.95493567e-01 -2.91455150e-01 6.45554960e-01 4.36221778e-01
6.46588683e-01 3.93338799e-01 -1.09942722e+00 8.58181596e-01
2.22586378e-01 -5.81156492e-01 -1.10605724e-01 -8.64243433e-02
4.47200894e-01 -3.90446514e-01 2.48881117e-01 3.30900609e-01
1.91217251e-02 -8.72447431e-01 1.24702466e+00 4.73745823e-01
2.93451399e-01 -8.90783072e-01 1.39068037e-01 4.81689721e-01
-1.10862112e+00 -4.10558611e-01 3.70703712e-02 1.28103912e-01
4.27144647e-01 7.16567099e-01 -8.65914524e-01 6.18936181e-01
3.44623983e-01 1.88545436e-01 -1.07606508e-01 1.04100657e+00
-3.06118309e-01 9.33708787e-01 -1.13736011e-01 1.14380196e-01
-2.20907177e-03 -3.23535442e-01 1.11376178e+00 1.14479673e+00
3.66727740e-01 -1.49659991e-01 6.65033311e-02 1.17459583e+00
3.51529196e-02 1.38153866e-01 -3.36569905e-01 1.29419595e-01
7.07926035e-01 1.35980844e+00 -2.69251391e-02 -4.56183478e-02
-9.27825794e-02 9.23367977e-01 2.65533417e-01 7.28000760e-01
-9.86551344e-01 -8.45521867e-01 8.83414090e-01 -4.11487520e-02
2.59923220e-01 1.04586883e-02 -2.92901456e-01 -8.85233939e-01
1.52936652e-01 -1.19723535e+00 -9.25106090e-03 -5.38502455e-01
-1.36270225e+00 6.83774292e-01 -2.86454827e-01 -1.13843846e+00
-1.50838152e-01 -2.64321357e-01 -1.05249655e+00 1.43017066e+00
-1.29020715e+00 -8.38863850e-01 4.61119749e-02 4.15634632e-01
5.41895807e-01 -1.95048735e-01 3.64011168e-01 6.49734139e-01
-1.05815136e+00 8.44391227e-01 1.27272293e-01 1.18792970e-02
1.04260230e+00 -9.84223545e-01 5.12976572e-02 1.06356132e+00
-5.44399843e-02 8.69227946e-01 9.28602755e-01 -3.71983826e-01
-1.02194333e+00 -1.06294274e+00 5.87732613e-01 4.01891693e-02
1.27592564e-01 -6.17910385e-01 -9.42623615e-01 5.24550080e-01
4.30650085e-01 -5.58969617e-01 8.27401459e-01 5.93432086e-03
-1.59661442e-01 -5.52422822e-01 -9.70862627e-01 7.14191914e-01
5.73579431e-01 -3.73909831e-01 -7.35104322e-01 -2.55132079e-01
8.78979504e-01 -1.16066277e-01 -2.12299734e-01 3.07710022e-01
4.09850985e-01 -7.58507729e-01 8.47829998e-01 -3.93980652e-01
1.35450557e-01 -4.54515219e-01 -1.15645923e-01 -1.97835040e+00
-5.28218150e-01 -1.01247537e+00 1.03124864e-01 2.07365108e+00
7.85652220e-01 -7.29442239e-01 3.91654462e-01 3.99418771e-01
-5.77866554e-01 -2.48707607e-01 -8.29994977e-01 -9.39256489e-01
5.26577458e-02 -1.43763497e-01 5.31960011e-01 8.08647037e-01
2.90590584e-01 8.22948277e-01 -4.55535710e-01 6.78891897e-01
5.21645427e-01 -2.15816215e-01 5.56685090e-01 -8.34514439e-01
-4.70813483e-01 -7.47786283e-01 1.71795279e-01 -1.21259618e+00
2.55885422e-01 -7.34196305e-01 5.99149704e-01 -1.41054416e+00
-1.27129257e-01 -3.68328154e-01 -3.46598804e-01 2.07841694e-01
-6.09415710e-01 -2.97124714e-01 1.93829179e-01 -2.08589941e-01
3.57201099e-02 1.00708485e+00 8.77208710e-01 -3.88878644e-01
-7.32158661e-01 4.69656378e-01 -1.25509644e+00 6.16786838e-01
7.35142589e-01 -3.95333439e-01 -6.14334464e-01 -2.83868313e-01
-5.45725346e-01 1.58433050e-01 5.54339662e-02 -1.11259234e+00
1.06826320e-01 -1.20272771e-01 5.03309444e-03 -4.72957045e-01
5.21931589e-01 -8.09794366e-01 -1.23875225e-02 2.09611490e-01
-6.64977908e-01 -8.17292094e-01 3.32233816e-01 4.39191520e-01
-2.26078495e-01 -2.93058544e-01 9.58687365e-01 4.30401802e-01
2.47231781e-01 -1.05231360e-01 -3.96739095e-01 -3.06237657e-02
8.54043663e-01 4.63254824e-02 -9.26855356e-02 -7.07676351e-01
-6.59583330e-01 4.66318965e-01 -3.25417489e-01 5.70107460e-01
4.76960719e-01 -1.47974205e+00 -9.14525449e-01 5.11254668e-01
-3.12095225e-01 -1.59733400e-01 4.05236363e-01 8.50429833e-01
1.45107031e-01 3.71933818e-01 1.45626813e-01 -3.56357485e-01
-1.58225787e+00 4.91452754e-01 6.89386606e-01 1.20179854e-01
-1.38793916e-01 1.11080313e+00 5.72105050e-01 -6.12177551e-01
6.65749311e-01 -3.12258065e-01 -3.78514022e-01 4.33125235e-02
6.30825222e-01 5.13959467e-01 -2.31017079e-02 -5.40602863e-01
-4.38694060e-01 2.88169235e-01 -4.71164584e-02 -6.42911017e-01
1.07778394e+00 -3.76755774e-01 -6.05963450e-03 5.45451283e-01
1.10985482e+00 7.77531207e-01 -1.40744781e+00 -1.45782590e-01
-2.91853368e-01 -3.02348614e-01 1.69047207e-01 -8.66594791e-01
-1.01082623e+00 8.61640036e-01 5.12530446e-01 4.32507455e-01
1.32170808e+00 -7.43896365e-02 6.49084866e-01 -1.69994190e-01
-3.06893885e-01 -9.49647546e-01 -5.83233908e-02 3.74884993e-01
1.05697405e+00 -7.02099323e-01 -5.42654514e-01 -6.05814219e-01
-7.74975836e-01 7.40128696e-01 8.94366205e-01 2.28560567e-01
4.04614776e-01 2.42093906e-01 3.64586622e-01 2.54292935e-01
-5.08546054e-01 -1.46717697e-01 5.30382812e-01 5.84257901e-01
5.63275933e-01 5.60619831e-02 -1.34394556e-01 1.33651507e+00
-2.29656726e-01 -5.42381763e-01 2.39997223e-01 5.13525665e-01
-5.46983063e-01 -1.04446352e+00 -9.80562508e-01 2.79184729e-01
-3.76263648e-01 -1.65891573e-01 -7.37467945e-01 -1.00587361e-01
7.28829131e-02 1.60076201e+00 -1.26775056e-01 -4.51715142e-01
7.47979760e-01 3.67846727e-01 1.79768041e-01 -6.59197807e-01
-5.71505249e-01 7.45051742e-01 6.23056293e-02 2.29342598e-02
-7.80578926e-02 -5.36156774e-01 -1.19645333e+00 7.23456442e-02
-7.95139730e-01 3.96489650e-01 5.15268803e-01 7.72507966e-01
3.71417522e-01 1.04499149e+00 1.10757089e+00 -1.09601581e+00
-9.15508032e-01 -1.30764306e+00 -7.92631209e-01 1.76570997e-01
8.37796509e-01 -2.43913129e-01 -8.60194921e-01 2.96708830e-02] | [14.830686569213867, 5.875107765197754] |
b0115e59-04a6-43ff-a533-acf00536a211 | a-topological-view-of-rule-learning-in-1 | null | null | https://openreview.net/forum?id=-xhk0O7iAc0 | https://openreview.net/pdf?id=-xhk0O7iAc0 | A Topological View of Rule Learning in Knowledge Graphs | Inductive relation prediction is an important learning task for knowledge graph completion. One can use the existence of rules, namely a sequence of relations, to predict the relation between two entities. Previous works view rules as paths and primarily focus on the searching of paths between entities. The space of paths is huge, and one has to sacrifice either efficiency or accuracy. In this paper, we consider rules in knowledge graphs as cycles and show that the space of cycles has a unique structure based on the theory of algebraic topology. By exploring the linear structure of the cycle space, we can improve the searching efficiency of rules. We propose to collect cycle bases that span the space of cycles. We build a novel GNN framework on the collected cycles to learn the representations of cycles, and to predict the existence/non-existence of a relation. Our method achieves state-of-the-art performance on benchmarks. | ['Chao Chen', 'Zhi Tang', 'Liangcai Gao', 'Tengfei Ma', 'Zuoyu Yan'] | 2021-09-29 | null | null | null | null | ['inductive-relation-prediction'] | ['graphs'] | [ 1.09325282e-01 2.99386173e-01 -7.01095164e-01 -1.96527429e-02
1.68626949e-01 -8.11094344e-01 4.42500263e-01 2.92162240e-01
6.12310395e-02 4.04223621e-01 9.89527181e-02 -9.53047335e-01
-4.44768876e-01 -1.52541316e+00 -9.39662516e-01 -2.25206330e-01
-6.88115180e-01 5.87059438e-01 3.66209030e-01 -1.53765976e-01
1.18611790e-02 4.56334144e-01 -1.29256845e+00 2.13089257e-01
7.93455184e-01 5.58298349e-01 -2.04559371e-01 5.22690594e-01
-2.72268772e-01 1.02959073e+00 7.07953572e-02 -5.98408997e-01
2.64441282e-01 -4.36452240e-01 -1.32372892e+00 -1.68245688e-01
4.65638302e-02 1.01697817e-01 -9.30084765e-01 9.65038240e-01
-2.14217722e-01 6.13833815e-02 4.94410276e-01 -1.30989277e+00
-5.20236254e-01 9.89298403e-01 -3.52854759e-01 1.32732809e-01
6.54006362e-01 -4.35208797e-01 1.64418304e+00 -7.39853919e-01
8.38342965e-01 9.28615510e-01 7.12801516e-01 2.16139555e-01
-1.24519515e+00 -5.49200892e-01 1.27868265e-01 6.52482688e-01
-1.72838342e+00 -2.73354590e-01 5.08596420e-01 -3.61371666e-01
1.04489660e+00 4.27451700e-01 8.21007133e-01 4.99108195e-01
4.71431352e-02 5.15143991e-01 4.38535213e-01 -7.14545608e-01
-3.87161113e-02 -3.14446509e-01 4.37440485e-01 1.37501407e+00
5.82143068e-01 4.26744446e-02 -5.23976028e-01 -8.61075297e-02
9.32146966e-01 -7.93532655e-02 -3.02592009e-01 -6.29600346e-01
-1.00108612e+00 9.07446146e-01 7.77180314e-01 2.45457083e-01
7.67656192e-02 2.11087465e-01 9.67538655e-02 4.18154240e-01
-1.82410136e-01 3.95054549e-01 -4.40841854e-01 3.09674680e-01
-3.64200264e-01 -5.17248698e-02 1.47959149e+00 1.06727314e+00
1.10310125e+00 -5.36259413e-01 2.41691455e-01 3.04197580e-01
3.17909449e-01 2.12795679e-02 -6.45813532e-03 -5.09031534e-01
4.00014579e-01 1.09950411e+00 -1.63460344e-01 -1.61033964e+00
-6.77164197e-01 -4.64822680e-01 -7.49239922e-01 -5.02473712e-01
5.55305898e-01 9.40285996e-02 -7.98397660e-01 1.72466314e+00
5.35397112e-01 5.30234694e-01 7.85330161e-02 4.78049457e-01
5.61402321e-01 6.09583199e-01 -1.72437906e-01 -2.40744203e-01
1.26183414e+00 -8.44244242e-01 -3.82295698e-01 -2.16187894e-01
1.18744147e+00 -2.40373224e-01 7.03086257e-01 4.93174940e-02
-6.79496706e-01 -1.42441407e-01 -1.04880536e+00 7.56001612e-03
-4.07178372e-01 -3.55232619e-02 1.28386176e+00 5.43873250e-01
-7.52446651e-01 7.76604712e-01 -9.64550614e-01 -2.94908941e-01
3.43728304e-01 5.23299515e-01 -4.90782887e-01 -2.83775300e-01
-1.37417686e+00 7.59832501e-01 8.43254030e-01 8.59281719e-02
-5.15931249e-01 -5.05359054e-01 -1.21547782e+00 1.90531537e-01
9.33134139e-01 -6.56130970e-01 7.37627327e-01 -5.34901440e-01
-7.00854540e-01 7.20322907e-01 -2.18805909e-01 -5.72776854e-01
-1.22278482e-02 5.97277209e-02 -7.74141610e-01 7.57931024e-02
-6.96332604e-02 6.21410385e-02 2.84852684e-01 -9.75467861e-01
-6.75525784e-01 -3.53289932e-01 5.15335679e-01 -6.41730055e-02
-3.92555922e-01 -3.50047678e-01 -9.49504852e-01 -2.05080375e-01
4.78595912e-01 -1.25759768e+00 -2.69664586e-01 -3.71346444e-01
-7.63523698e-01 -6.27530694e-01 4.83583122e-01 -4.03611869e-01
1.70459104e+00 -1.72297943e+00 -1.32293716e-01 9.60009754e-01
6.13317966e-01 1.52628690e-01 2.02020891e-02 5.14932871e-01
-1.46012068e-01 3.38507175e-01 2.99726892e-02 3.84682178e-01
-1.27801031e-01 5.39054334e-01 -5.07247925e-01 4.20765996e-01
-1.17056239e-02 1.08465278e+00 -9.26677942e-01 -4.14288253e-01
-3.26870441e-01 4.09454629e-02 -5.48229158e-01 -1.04295157e-01
-4.32571322e-01 -9.24003199e-02 -5.45475662e-01 2.89629579e-01
5.02533793e-01 -8.17338586e-01 1.12059355e+00 -1.37664795e-01
3.93055886e-01 6.70443296e-01 -1.45928669e+00 1.48043311e+00
-2.27613643e-01 2.95646161e-01 -6.26399040e-01 -1.03635919e+00
9.30599749e-01 1.16622914e-02 3.89360547e-01 -4.35771942e-01
-1.95856288e-01 1.20628603e-01 2.21309707e-01 -4.23402667e-01
2.61792660e-01 1.99488029e-01 1.57291219e-01 5.43788850e-01
-4.58085686e-02 6.75307691e-01 5.81500292e-01 5.04496157e-01
1.40756714e+00 9.43514928e-02 4.29947883e-01 -1.54461905e-01
6.12674475e-01 3.75828482e-02 7.63522685e-01 4.78907168e-01
5.39735556e-01 -1.32679448e-01 1.00108004e+00 -7.56803274e-01
-6.84055269e-01 -9.21480238e-01 1.99336648e-01 8.94482851e-01
3.50995690e-01 -1.21162844e+00 -4.12005186e-01 -9.52521861e-01
7.13024586e-02 3.26094180e-01 -6.86801016e-01 -3.99769157e-01
-8.38879645e-01 -4.84034210e-01 5.75836182e-01 4.87832487e-01
2.17266396e-01 -7.64053702e-01 -2.65772194e-01 6.92564100e-02
-3.57914567e-01 -1.34440780e+00 -5.49859941e-01 -8.68229102e-03
-8.64714384e-01 -1.87961423e+00 2.73304194e-01 -1.14836526e+00
7.56876647e-01 1.21795572e-01 1.26237798e+00 5.94250500e-01
-7.73273855e-02 2.65484303e-01 -2.84812331e-01 5.16605657e-03
-2.91138709e-01 3.18807930e-01 8.63981694e-02 -2.02801358e-03
4.87261534e-01 -8.16429198e-01 -4.08514053e-01 4.53220546e-01
-6.81397378e-01 9.07187387e-02 3.10443431e-01 4.81003135e-01
8.73100877e-01 6.58731759e-01 3.51042718e-01 -1.26366293e+00
2.49990731e-01 -4.41141248e-01 -6.28784418e-01 6.69384241e-01
-9.49882925e-01 5.44470966e-01 6.27505779e-01 -2.87323028e-01
-5.90478063e-01 3.03671271e-01 4.06572044e-01 -2.47818202e-01
2.88992107e-01 1.14110851e+00 -2.96406060e-01 -1.99491888e-01
7.25269675e-01 6.99523836e-03 -3.65569532e-01 -2.45198011e-01
7.76765585e-01 -8.97782892e-02 4.25904095e-01 -6.01186395e-01
7.66498983e-01 4.85931993e-01 6.28682554e-01 -5.85480213e-01
-9.44386184e-01 -4.55699086e-01 -5.58648109e-01 5.08227497e-02
3.16129774e-01 -6.84094846e-01 -1.11486804e+00 -3.26995671e-01
-9.82961297e-01 -1.55591562e-01 -4.85627986e-02 4.53656077e-01
-3.25710773e-01 5.59398711e-01 -5.66138387e-01 -5.32454252e-01
-2.42923424e-01 -4.54306662e-01 1.54958412e-01 5.67012578e-02
-2.93841392e-01 -1.04177928e+00 3.62501770e-01 9.17341635e-02
-3.08425486e-01 1.46918982e-01 1.37315643e+00 -7.97777653e-01
-9.88089144e-01 -2.29047135e-01 -1.50679410e-01 -3.94365847e-01
5.65662719e-02 -1.49895132e-01 -2.69428551e-01 -5.71719818e-02
-5.24593294e-01 -8.52139518e-02 1.11947560e+00 -6.86726272e-02
1.08882296e+00 -6.43047810e-01 -8.87986898e-01 6.83549821e-01
1.40231645e+00 -2.60062013e-02 8.37244391e-01 4.86833379e-02
1.02620995e+00 5.25343180e-01 3.63463134e-01 1.56154633e-01
7.49163032e-01 4.65039581e-01 1.76019371e-01 4.34573323e-01
-1.85292866e-02 -8.90308022e-01 1.14490867e-01 8.28696668e-01
-3.99380744e-01 1.77668165e-02 -1.23420799e+00 6.70666754e-01
-2.03340006e+00 -8.28193665e-01 -4.42037284e-01 2.23461485e+00
8.93655658e-01 2.16742605e-01 1.38564244e-01 2.80249923e-01
6.81584358e-01 -1.58884391e-01 -3.07435513e-01 -2.38447934e-01
4.35332656e-02 2.84149915e-01 6.49899364e-01 6.72710419e-01
-1.05171573e+00 1.03295481e+00 6.20148182e+00 6.93333507e-01
-6.83070898e-01 -3.27903748e-01 1.77107409e-01 4.01363373e-01
-5.01572490e-01 4.79903758e-01 -7.85650015e-01 -4.20180783e-02
8.97921562e-01 -4.49209899e-01 6.48235500e-01 6.84318602e-01
-4.92466301e-01 1.12547524e-01 -1.40389228e+00 8.05094898e-01
-1.37942821e-01 -1.52775538e+00 5.19560911e-02 1.87067240e-01
7.80731380e-01 -2.26473004e-01 -4.18460011e-01 2.93047011e-01
6.14281297e-01 -1.28414524e+00 9.03465673e-02 5.81586719e-01
7.78988063e-01 -9.89772558e-01 3.14170659e-01 3.05912942e-01
-1.77035594e+00 -1.90684766e-01 -3.92361790e-01 -1.91449657e-01
-1.91305950e-01 6.30609989e-01 -9.90703762e-01 9.98203516e-01
3.04632574e-01 8.89998019e-01 -4.84672129e-01 9.48503613e-01
-7.30563343e-01 7.08335996e-01 -5.25757015e-01 -1.59747582e-02
-2.83861935e-01 -3.86364013e-01 2.01386720e-01 1.08861196e+00
1.83359861e-01 4.75784600e-01 3.17365617e-01 6.46940887e-01
-5.28008938e-01 5.81221208e-02 -8.32805455e-01 -2.91015446e-01
5.63802361e-01 1.17133784e+00 -9.71177340e-01 -5.20478711e-02
-5.91156423e-01 5.25313675e-01 6.57562494e-01 2.30191171e-01
-5.79193294e-01 -3.81236821e-01 4.91212666e-01 9.19475853e-02
7.11758196e-01 -3.64093274e-01 -4.11711633e-01 -1.13728523e+00
1.57241479e-01 -7.49604642e-01 9.17311370e-01 -2.73043454e-01
-8.51979852e-01 3.44931364e-01 -3.53614122e-01 -7.59449005e-01
-3.20726307e-03 -5.07975638e-01 -6.59164786e-01 4.25120384e-01
-1.18316495e+00 -1.02246785e+00 -2.34069247e-02 8.78215969e-01
-2.63602793e-01 1.52278289e-01 1.07639384e+00 1.84647754e-01
-4.80779856e-01 6.64775968e-01 -2.46465087e-01 5.07561326e-01
4.65965867e-02 -1.11641300e+00 4.93952781e-01 8.55234265e-01
8.61130118e-01 8.60861719e-01 3.18611681e-01 -7.81320691e-01
-1.78368104e+00 -1.09186864e+00 1.25356519e+00 -4.18523163e-01
8.92858207e-01 -1.42985359e-01 -9.28450465e-01 1.16817486e+00
-4.06785876e-01 1.70520037e-01 7.69856155e-01 8.37489665e-01
-8.37802470e-01 -2.29396135e-01 -5.08181691e-01 7.71634758e-01
1.58855808e+00 -6.84408605e-01 -4.42176640e-01 2.94183016e-01
8.79213631e-01 -2.08835244e-01 -9.19645965e-01 4.43655521e-01
6.49564266e-01 -5.02003133e-01 1.05874062e+00 -1.07030892e+00
3.96337986e-01 -3.94165635e-01 1.30733564e-01 -9.38017130e-01
-4.62236285e-01 -8.20143878e-01 -1.03642523e+00 8.55741441e-01
8.48839164e-01 -4.66123104e-01 1.25403678e+00 4.24033284e-01
3.67250323e-01 -1.02920759e+00 -7.15048373e-01 -7.68241465e-01
-2.10636497e-01 -5.12079000e-01 8.34272146e-01 1.23152995e+00
6.20385468e-01 9.17258382e-01 -3.16084743e-01 4.61391330e-01
5.61398923e-01 7.49118865e-01 7.73193300e-01 -1.60984492e+00
-3.45808566e-01 -2.33624354e-01 -5.44672370e-01 -1.05071723e+00
1.79160729e-01 -1.50590646e+00 -3.33413601e-01 -1.79093254e+00
2.61085331e-01 -5.94169557e-01 -3.59432250e-01 6.96354151e-01
2.19660867e-02 -2.52269119e-01 3.98404971e-02 2.32157171e-01
-7.78765559e-01 2.70355552e-01 1.18654239e+00 9.53304768e-03
-3.38423401e-01 1.10778958e-01 -8.77735913e-01 7.12420166e-01
7.30918050e-01 -4.59343761e-01 -4.95108455e-01 -1.93777844e-01
9.87618327e-01 1.76870883e-01 1.68304488e-01 -6.16147161e-01
6.74757957e-01 -2.00313434e-01 1.75790548e-01 -4.16604489e-01
5.36859361e-03 -9.16975260e-01 3.01274955e-01 5.78895152e-01
-2.66221702e-01 -2.20816255e-01 -1.51135802e-01 9.83065784e-01
-2.41134223e-02 3.44648622e-02 1.77529350e-01 1.00758746e-01
-7.43872941e-01 6.26008153e-01 2.03769624e-01 3.53454947e-02
8.58768225e-01 2.48038191e-02 -2.73671627e-01 -3.12100261e-01
-9.29301620e-01 5.80939889e-01 1.63508505e-01 2.50968814e-01
6.97611511e-01 -1.54071999e+00 -3.55240434e-01 1.23777643e-01
1.45788401e-01 1.44403160e-01 -6.98182508e-02 7.38953829e-01
-4.26960260e-01 4.89160061e-01 2.45799854e-01 -1.36436194e-01
-1.29544020e+00 9.84762251e-01 3.29718620e-01 -7.69286573e-01
-7.55088270e-01 7.02243567e-01 1.29731759e-01 -3.84040058e-01
2.11672217e-01 -1.96594343e-01 -4.86342281e-01 -1.19262524e-01
4.02641714e-01 4.30962116e-01 -1.50684463e-02 -2.99724311e-01
-4.61912125e-01 5.19489527e-01 -2.07933635e-01 3.57465535e-01
1.17440081e+00 1.84112638e-01 -5.11290550e-01 1.18712507e-01
1.11900485e+00 4.49868403e-02 -6.79952264e-01 -7.54850030e-01
6.18578613e-01 -3.73005927e-01 -3.27099055e-01 -4.24061924e-01
-1.00070405e+00 3.09705913e-01 -1.46733120e-01 4.69503015e-01
1.03265429e+00 4.30617958e-01 7.62132943e-01 9.85269904e-01
4.09188807e-01 -8.92645538e-01 -1.44882455e-01 9.56670105e-01
4.20819253e-01 -7.30822563e-01 8.14033896e-02 -1.09993589e+00
-3.38168949e-01 1.22125483e+00 3.01884770e-01 -1.03143893e-01
1.04624188e+00 -1.03567556e-01 -7.37792134e-01 -4.63094622e-01
-8.74000609e-01 -5.85721314e-01 5.15795231e-01 4.92262572e-01
1.32983267e-01 4.28997815e-01 -4.31465924e-01 8.11859131e-01
-3.81865680e-01 6.52091205e-02 3.86213094e-01 7.53263354e-01
-4.82249230e-01 -1.40090013e+00 -6.39529526e-02 5.54428875e-01
-2.28011027e-01 -1.79157794e-01 -6.38681889e-01 5.74343026e-01
1.11808673e-01 1.04845941e+00 -1.28970161e-01 -7.81126261e-01
3.67895663e-01 8.16563740e-02 6.84012294e-01 -6.31070852e-01
2.12509423e-01 -4.83530462e-01 5.37554026e-01 -5.18222570e-01
-3.15402240e-01 -4.01918709e-01 -1.49970031e+00 -7.48300493e-01
-5.68690598e-01 3.15981448e-01 1.52652875e-01 1.05365157e+00
4.01251435e-01 3.29602301e-01 6.93870306e-01 3.17753673e-01
-9.33113098e-02 -5.09020209e-01 -7.48636782e-01 2.35244796e-01
2.57344618e-02 -5.10181069e-01 4.80029583e-02 2.71114614e-02] | [8.663936614990234, 7.688998222351074] |
520af4ab-d6fb-4d03-9cea-ddbad172af36 | saliency-weighted-convolutional-features-for | 1711.10795 | null | http://arxiv.org/abs/1711.10795v1 | http://arxiv.org/pdf/1711.10795v1.pdf | Saliency Weighted Convolutional Features for Instance Search | This work explores attention models to weight the contribution of local
convolutional representations for the instance search task. We present a
retrieval framework based on bags of local convolutional features (BLCF) that
benefits from saliency weighting to build an efficient image representation.
The use of human visual attention models (saliency) allows significant
improvements in retrieval performance without the need to conduct region
analysis or spatial verification, and without requiring any feature fine
tuning. We investigate the impact of different saliency models, finding that
higher performance on saliency benchmarks does not necessarily equate to
improved performance when used in instance search tasks. The proposed approach
outperforms the state-of-the-art on the challenging INSTRE benchmark by a large
margin, and provides similar performance on the Oxford and Paris benchmarks
compared to more complex methods that use off-the-shelf representations. The
source code used in this project is available at
https://imatge-upc.github.io/salbow/ | ["Noel E. O'Connor", 'Xavier Giro-i-Nieto', 'Kevin McGuinness', 'Eva Mohedano'] | 2017-11-29 | null | null | null | null | ['instance-search'] | ['computer-vision'] | [-1.30896568e-01 -2.36493826e-01 -3.44528824e-01 -2.21968487e-01
-1.00063503e+00 -3.41425180e-01 8.73327255e-01 4.32915658e-01
-7.54056215e-01 4.91062522e-01 3.37713271e-01 -1.08485796e-01
-3.12093496e-01 -5.98464727e-01 -7.19181299e-01 -4.18980211e-01
8.53910148e-02 3.20241421e-01 5.18090844e-01 -4.16865706e-01
6.93957150e-01 5.12292564e-01 -2.01466823e+00 4.41298097e-01
5.59108496e-01 1.21269977e+00 4.14651483e-01 5.49462497e-01
-4.70314436e-02 6.56670809e-01 -5.45140326e-01 -1.37517735e-01
3.13100547e-01 -1.09909415e-01 -1.00797570e+00 -4.88864332e-01
8.10924351e-01 -1.57606810e-01 -3.05389047e-01 7.00407445e-01
6.45958602e-01 3.31922263e-01 6.70195282e-01 -9.50313389e-01
-9.05761242e-01 2.81673163e-01 -5.21424115e-01 7.96797276e-01
4.00474578e-01 -1.66492574e-02 1.43003595e+00 -1.28517544e+00
7.07048833e-01 1.13424885e+00 4.75277692e-01 1.11680865e-01
-1.00833118e+00 -3.40659142e-01 -4.06577811e-03 5.19640684e-01
-1.48207998e+00 -3.40002865e-01 4.81850266e-01 -1.60089657e-01
1.09732950e+00 4.14600700e-01 4.74402994e-01 9.28777039e-01
1.83463290e-01 1.00674653e+00 9.47047710e-01 -4.74247605e-01
9.56429839e-02 1.69188932e-01 4.56057101e-01 6.35373056e-01
1.90734252e-01 9.10651460e-02 -6.05084002e-01 -1.17318071e-01
5.06645620e-01 2.34869957e-01 -3.38812262e-01 -5.18737435e-01
-1.27952182e+00 9.66186345e-01 1.18200958e+00 6.71037376e-01
-6.00274563e-01 4.23165321e-01 4.66412157e-01 1.81508705e-01
6.01391375e-01 6.09965146e-01 -2.84227043e-01 -3.86981368e-02
-1.31532419e+00 4.49258178e-01 2.88630277e-01 7.61101305e-01
7.46803105e-01 -2.73978233e-01 -6.96259677e-01 8.35062981e-01
1.29587993e-01 2.46574238e-01 7.19077647e-01 -7.13771880e-01
1.57036230e-01 6.33074999e-01 1.59611821e-01 -9.92129147e-01
-3.41926932e-01 -6.11095548e-01 -3.06341678e-01 4.70336936e-02
2.18017653e-01 5.04736245e-01 -1.24091470e+00 1.33816397e+00
8.10394734e-02 -2.37889439e-02 -3.21949720e-01 1.21557105e+00
1.09265840e+00 4.00227547e-01 3.05897947e-02 5.37557781e-01
1.28805768e+00 -1.32611001e+00 -4.89430904e-01 -1.67435840e-01
6.19159281e-01 -8.27338457e-01 1.17804432e+00 2.11314652e-02
-1.20719230e+00 -5.96522272e-01 -1.07534850e+00 -5.51726818e-01
-9.32406127e-01 1.68664262e-01 6.61088169e-01 3.13615620e-01
-1.43329465e+00 5.93737304e-01 -4.87490565e-01 -5.56585014e-01
5.04095793e-01 2.69948632e-01 -2.44656608e-01 -9.95918289e-02
-1.19829559e+00 1.21842015e+00 2.70366520e-01 -1.34153739e-01
-8.97311687e-01 -7.75533676e-01 -8.99948180e-01 4.31027889e-01
2.15709493e-01 -4.65006620e-01 1.13472641e+00 -8.63897502e-01
-8.66400003e-01 9.49755669e-01 -1.53993353e-01 -7.44512081e-01
4.60511059e-01 -6.68936908e-01 1.05388187e-01 3.99252027e-01
1.39640421e-01 1.08295023e+00 8.08127105e-01 -9.69011009e-01
-4.11803156e-01 -1.71997741e-01 2.98808664e-01 1.85266152e-01
-1.81235597e-01 2.12383389e-01 -7.71120727e-01 -7.08889186e-01
-2.40350917e-01 -9.42305863e-01 -7.61378631e-02 -4.70366664e-02
-1.38669804e-01 -4.61268663e-01 6.72797024e-01 -5.93981266e-01
1.10097861e+00 -2.04979277e+00 1.52432308e-01 1.54735863e-01
8.23074430e-02 3.69859129e-01 -3.16641897e-01 3.71817499e-01
-1.15524270e-01 1.75915942e-01 -7.59558976e-02 -2.65538216e-01
5.64066172e-02 -2.05280736e-01 -3.08172047e-01 5.20676136e-01
2.57431000e-01 1.30954111e+00 -9.04244125e-01 -4.65425462e-01
2.31269062e-01 6.54641569e-01 -5.49836814e-01 -1.23455502e-01
-7.21282586e-02 -5.87302782e-02 -3.71988028e-01 8.44947517e-01
4.05108780e-01 -6.03174686e-01 -3.93733591e-01 9.15506110e-02
-1.95766035e-02 4.22407299e-01 -7.89083421e-01 1.96255958e+00
-4.40910488e-01 7.05745280e-01 -2.56949067e-01 -8.99636567e-01
6.73539877e-01 1.34639936e-02 1.77038595e-01 -1.10612404e+00
1.00896470e-01 3.04393560e-01 -1.76562786e-01 -3.16397659e-02
9.31780100e-01 4.49432284e-01 -1.53019965e-01 4.06200498e-01
3.12733471e-01 1.73116028e-01 1.55998573e-01 4.65369135e-01
1.11515224e+00 1.84936345e-01 2.56515056e-01 -6.37769282e-01
5.40076911e-01 1.00398041e-01 6.60990775e-02 8.97048175e-01
-1.61067262e-01 9.40901756e-01 1.66704431e-01 -4.15957600e-01
-8.81865263e-01 -8.38702857e-01 -1.59575149e-01 1.51145697e+00
3.00656825e-01 -4.33538318e-01 -6.10933959e-01 -6.10605001e-01
2.19049886e-01 4.85980064e-01 -1.06456816e+00 -2.15056971e-01
-2.08812937e-01 -3.71203661e-01 3.71227264e-01 5.48044622e-01
4.93161589e-01 -1.46288168e+00 -1.04794478e+00 -2.37012148e-01
-5.94520383e-02 -6.77741051e-01 -5.09270489e-01 1.13308579e-01
-6.50187194e-01 -8.61780941e-01 -1.32318354e+00 -6.67195320e-01
4.40846682e-01 6.00674331e-01 1.34040701e+00 6.64328575e-01
-7.93889582e-01 6.08124077e-01 -6.08693719e-01 -5.39425254e-01
4.30463076e-01 6.21091127e-01 -3.49050045e-01 -3.07123601e-01
4.17192668e-01 8.80613327e-02 -1.13876688e+00 2.20861152e-01
-9.35582936e-01 -3.05119902e-01 5.69693565e-01 8.78104031e-01
5.34241498e-01 -7.40878761e-01 5.70993602e-01 -6.49113595e-01
6.42761290e-01 -6.46137238e-01 -4.74445432e-01 3.42386156e-01
-6.17861509e-01 1.35673478e-01 -2.43544132e-02 -5.45149632e-02
-6.15950942e-01 -3.03781241e-01 1.85271814e-01 -6.38458550e-01
-4.39456552e-02 5.20151556e-01 5.81782281e-01 -2.30881289e-01
7.51419604e-01 1.47723287e-01 -1.48525208e-01 -6.31223440e-01
3.95908922e-01 4.87901419e-01 1.91464260e-01 -2.94165134e-01
3.55522007e-01 4.72052455e-01 -1.87657937e-01 -5.86493433e-01
-7.90039361e-01 -8.52486908e-01 -5.40787697e-01 -1.79818217e-02
6.90334499e-01 -9.26641047e-01 -3.14094871e-01 7.80046955e-02
-8.25704396e-01 -2.43706033e-01 -3.46778005e-01 1.26289815e-01
-5.45711696e-01 1.76295176e-01 -4.92014021e-01 -5.30895770e-01
-5.75125277e-01 -1.25559390e+00 1.52297211e+00 3.14654440e-01
-2.72352368e-01 -7.52371132e-01 -1.55627960e-02 4.03545171e-01
9.46278870e-01 -2.50682477e-02 4.63781357e-01 -7.13315248e-01
-7.97339916e-01 -4.32795167e-01 -5.32122195e-01 -3.10657267e-02
-1.91619694e-01 -9.46502909e-02 -1.15938139e+00 -3.80158156e-01
-4.78285879e-01 -4.94716734e-01 1.51494277e+00 4.06575084e-01
1.27631021e+00 1.11155799e-02 -2.85059601e-01 4.08055067e-01
1.56224966e+00 -3.26603383e-01 8.22599173e-01 7.43612826e-01
3.76647741e-01 5.89412987e-01 9.18940723e-01 2.85137802e-01
3.28440160e-01 9.34841931e-01 5.49390733e-01 -1.75731972e-01
-3.50004345e-01 8.00902490e-03 2.09977906e-02 1.08592056e-01
-4.64446656e-02 -9.78068113e-02 -1.09805298e+00 1.19629395e+00
-2.02139688e+00 -9.20632362e-01 1.06977381e-01 2.19307017e+00
6.52870357e-01 -1.61559358e-01 9.51815024e-02 -5.62239438e-02
6.22883499e-01 3.01481992e-01 -2.05761179e-01 -5.45783699e-01
6.84889331e-02 5.50335467e-01 5.30706406e-01 4.63879496e-01
-1.18027043e+00 1.06255567e+00 6.33049536e+00 1.02963269e+00
-1.18565476e+00 2.78610438e-01 6.15758717e-01 -4.30814117e-01
-2.41940096e-01 -1.61852881e-01 -7.03527629e-01 4.45293695e-01
1.01636732e+00 -2.34118756e-02 1.97091281e-01 8.92126620e-01
-2.37194121e-01 -3.92644644e-01 -8.21041822e-01 8.26295435e-01
2.69042045e-01 -1.48340070e+00 7.98780285e-03 3.80148031e-02
6.48819625e-01 4.82898593e-01 4.11802143e-01 4.37528163e-01
-4.11624322e-03 -1.12539601e+00 7.09248543e-01 6.46443307e-01
5.26277721e-01 -6.11323118e-01 9.94688034e-01 -5.50489426e-02
-9.54230070e-01 -8.31792131e-02 -5.38291574e-01 2.15027139e-01
-2.73550987e-01 4.08177465e-01 -6.74217045e-01 3.81852120e-01
1.10862672e+00 6.42298281e-01 -1.15732145e+00 1.54905427e+00
-1.02674603e-01 2.85790831e-01 -2.52544016e-01 -1.54502168e-01
5.81645370e-01 2.26441041e-01 4.08125311e-01 1.49291503e+00
2.24741578e-01 -4.16054726e-01 -2.98070740e-02 8.14289033e-01
-2.04724789e-01 3.79808992e-01 -6.21589601e-01 1.59583583e-01
2.35128060e-01 1.24794626e+00 -6.74317420e-01 -4.08300489e-01
-3.30111682e-01 1.03923476e+00 6.03696823e-01 3.30863863e-01
-9.06826198e-01 -5.49281538e-01 5.19571543e-01 2.31022283e-01
7.81763792e-01 1.24943018e-01 -2.52828568e-01 -9.10056174e-01
4.35540974e-02 -6.02288961e-01 5.72766364e-01 -9.54538584e-01
-9.91450012e-01 5.54016173e-01 1.25909045e-01 -9.84654248e-01
-2.07086608e-01 -3.13935101e-01 -5.28214633e-01 1.04841387e+00
-1.94518340e+00 -1.14832747e+00 -4.47564363e-01 5.57887971e-01
5.63936114e-01 4.57438035e-03 7.76431382e-01 1.93529829e-01
-1.71004549e-01 7.11347163e-01 -1.06940670e-05 -5.59050925e-02
8.37463796e-01 -1.14715612e+00 3.88586253e-01 5.51201642e-01
3.86022002e-01 8.52635443e-01 4.56940651e-01 -4.35469151e-01
-1.02067220e+00 -7.99369395e-01 1.04004729e+00 -6.79221809e-01
5.36626637e-01 -2.17372432e-01 -9.53281581e-01 4.94137377e-01
4.87352371e-01 3.19372982e-01 4.46588963e-01 2.80983239e-01
-5.43322802e-01 1.22242652e-01 -1.17507565e+00 3.39716405e-01
8.44555378e-01 -6.37781322e-01 -7.69292295e-01 3.32397580e-01
6.00213051e-01 -3.10426772e-01 -7.01552331e-01 4.14893389e-01
6.15619004e-01 -9.66059387e-01 1.29479980e+00 -5.00608981e-01
5.53573430e-01 -1.48605913e-01 -1.58806443e-01 -1.05086482e+00
-5.83972692e-01 -8.69200602e-02 -4.27500121e-02 8.26333046e-01
5.71984470e-01 -5.72871625e-01 6.28618479e-01 3.70094568e-01
-4.90354002e-03 -9.57469165e-01 -9.32459533e-01 -6.87180102e-01
4.23628092e-02 -1.00337408e-01 4.67384100e-01 6.52934551e-01
-5.35976589e-02 1.11680441e-02 -1.27273440e-01 -6.25325218e-02
3.13542426e-01 3.70006084e-01 5.61059058e-01 -1.05828714e+00
3.00849248e-02 -6.64698422e-01 -7.28905678e-01 -6.15681350e-01
7.81450346e-02 -1.02255785e+00 4.24544998e-02 -1.71792507e+00
3.05405647e-01 -3.16781342e-01 -8.66939843e-01 7.25292385e-01
-3.55064511e-01 7.31916904e-01 4.78108972e-01 3.00135165e-01
-1.12483513e+00 6.24818027e-01 8.07596982e-01 -1.69424295e-01
1.48284063e-01 -3.29645395e-01 -8.44975173e-01 2.28207573e-01
9.74905193e-01 -3.29289943e-01 -1.81231618e-01 -4.10334349e-01
1.12551309e-01 -4.93768066e-01 5.68646371e-01 -1.06781185e+00
2.92840034e-01 3.82928222e-01 4.85386699e-01 -6.20251954e-01
4.94421571e-01 -7.43150294e-01 -3.20786148e-01 3.71915102e-01
-5.98034799e-01 2.53485650e-01 4.31270510e-01 4.58085656e-01
-4.12942171e-01 -3.18908066e-01 6.15316212e-01 -2.27061078e-01
-9.57357407e-01 8.59451070e-02 -7.67141059e-02 -3.45099978e-02
9.51635540e-01 -1.18541278e-01 -4.66244787e-01 -3.78734082e-01
-5.06276429e-01 2.52147108e-01 6.46114349e-01 7.57654965e-01
6.75642967e-01 -1.31466615e+00 -6.06016755e-01 5.84901385e-02
5.17585516e-01 -3.55471373e-01 8.96589160e-02 7.94533193e-01
-6.96991205e-01 1.01129758e+00 -2.29203105e-01 -6.59702957e-01
-1.27330720e+00 6.23843431e-01 1.53764442e-01 -4.36189681e-01
-4.66780543e-01 1.09074628e+00 9.66028795e-02 -1.98898360e-01
3.94765645e-01 -2.59223551e-01 -2.64607012e-01 2.37606928e-01
6.97019279e-01 2.55987316e-01 4.47504520e-01 -6.99480355e-01
-5.90262353e-01 5.08926809e-01 -5.01313865e-01 3.63826975e-02
1.36557043e+00 2.96075847e-02 7.71805421e-02 2.41642550e-01
1.29967701e+00 -3.03420931e-01 -7.82931030e-01 -2.86961049e-01
1.41118899e-01 -7.69282818e-01 5.64906120e-01 -8.94298077e-01
-1.09544265e+00 9.68677998e-01 8.29081059e-01 2.80753598e-02
8.43468666e-01 1.92030132e-01 5.79826236e-01 3.01195502e-01
4.95515436e-01 -1.04584873e+00 4.09280360e-02 6.03893816e-01
1.39956427e+00 -1.43364465e+00 2.42572978e-01 -9.97232646e-03
-8.02856505e-01 7.33708084e-01 4.58329827e-01 -6.37255073e-01
7.41261899e-01 -2.56064445e-01 -1.73050523e-01 -5.18937230e-01
-7.51839399e-01 -6.61025345e-01 7.90611565e-01 2.78230935e-01
5.35408437e-01 -1.65101603e-01 -4.20160145e-01 5.98576330e-02
5.02842255e-02 -8.47729445e-02 1.41506121e-01 1.12623501e+00
-4.42650646e-01 -7.72671759e-01 -3.57436895e-01 6.18400156e-01
-6.67063236e-01 -5.10904610e-01 -6.60386860e-01 8.47831905e-01
-2.66270012e-01 6.07965410e-01 2.84634590e-01 -1.27289474e-01
2.06283569e-01 2.33200386e-01 2.89383411e-01 -6.92815125e-01
-1.07761049e+00 -1.58157781e-01 -1.97573304e-01 -9.07150269e-01
-5.06766737e-01 -6.19871318e-01 -1.00865698e+00 -1.68422893e-01
-3.91491622e-01 1.00395150e-01 6.76272869e-01 5.31063855e-01
8.00511837e-01 5.44573367e-01 2.03263357e-01 -1.27943981e+00
-2.35719249e-01 -1.20131767e+00 -3.16551298e-01 4.97813582e-01
4.82643276e-01 -8.85026455e-01 -4.22427565e-01 -4.75490183e-01] | [10.634035110473633, 0.685482382774353] |
a3638840-3350-4306-870b-9d60f6990d5d | information-extraction-of-clinical-trial | 2006.07296 | null | https://arxiv.org/abs/2006.07296v6 | https://arxiv.org/pdf/2006.07296v6.pdf | Information Extraction of Clinical Trial Eligibility Criteria | Clinical trials predicate subject eligibility on a diversity of criteria ranging from patient demographics to food allergies. Trials post their requirements as semantically complex, unstructured free-text. Formalizing trial criteria to a computer-interpretable syntax would facilitate eligibility determination. In this paper, we investigate an information extraction (IE) approach for grounding criteria from trials in ClinicalTrials(dot)gov to a shared knowledge base. We frame the problem as a novel knowledge base population task, and implement a solution combining machine learning and context free grammar. To our knowledge, this work is the first criteria extraction system to apply attention-based conditional random field architecture for named entity recognition (NER), and word2vec embedding clustering for named entity linking (NEL). We release the resources and core components of our system on GitHub at https://github.com/facebookresearch/Clinical-Trial-Parser. Finally, we report our per module and end to end performances; we conclude that our system is competitive with Criteria2Query, which we view as the current state-of-the-art in criteria extraction. | ['Ahmed Mohamed', 'M. I. Salkola', 'Yitong Tseo', 'Freddy Abnousi', 'Anuj Kumar'] | 2020-06-12 | null | null | null | null | ['knowledge-base-population'] | ['natural-language-processing'] | [ 3.54978114e-01 3.62166345e-01 -9.83618736e-01 -4.75740790e-01
-1.44787395e+00 -8.77591431e-01 1.26559854e-01 9.85876381e-01
-7.18226790e-01 8.22846115e-01 5.91634393e-01 -1.00874627e+00
-3.24712783e-01 -4.85630393e-01 -4.95325208e-01 -1.37951295e-03
2.04011127e-02 8.11347842e-01 -4.89121536e-03 8.81466120e-02
-8.49873722e-02 3.29398215e-01 -8.73329878e-01 7.71387458e-01
8.90277624e-01 6.80830121e-01 -1.17102779e-01 6.63586140e-01
4.01015170e-02 8.32376242e-01 -2.14146867e-01 -6.45708680e-01
-2.75923610e-01 -1.74387798e-01 -1.16309893e+00 -6.94769144e-01
3.00285909e-02 1.33907139e-01 1.29006520e-01 1.00115144e+00
8.72028530e-01 -2.05265790e-01 5.63472450e-01 -7.87967503e-01
-7.66291440e-01 1.03336704e+00 8.78461376e-02 1.63612738e-02
9.13203597e-01 -8.67834017e-02 1.19309139e+00 -8.40811253e-01
1.23369479e+00 6.98529243e-01 6.22602880e-01 9.02164459e-01
-1.12387383e+00 -6.76324189e-01 1.67827487e-01 -1.05628297e-02
-1.42273128e+00 -4.72864389e-01 7.24491328e-02 -6.40080512e-01
1.56158507e+00 5.64770818e-01 3.07733476e-01 1.14115787e+00
4.14537311e-01 5.90495646e-01 9.19778705e-01 -6.02281094e-01
4.40254927e-01 4.62269457e-03 5.93949199e-01 5.56749225e-01
6.08432949e-01 1.12153664e-01 -4.62455861e-02 -7.36249208e-01
2.85952806e-01 -2.32012942e-01 -2.86892056e-01 -1.08472764e-01
-1.15963149e+00 1.00600040e+00 2.14528397e-01 1.87077835e-01
-5.92618763e-01 -5.13039716e-03 4.67165291e-01 1.54548481e-01
1.98743090e-01 4.40787375e-01 -1.01281595e+00 -3.66497561e-02
-7.73800194e-01 -6.96644606e-03 1.22905338e+00 1.21164703e+00
2.32939571e-01 -7.26917267e-01 -3.76517564e-01 4.92508203e-01
4.47909594e-01 4.03890461e-01 4.23390925e-01 -4.84863549e-01
3.88450205e-01 5.84790945e-01 2.67070010e-02 -4.23830152e-01
-8.15671265e-01 -1.35019630e-01 -3.94250125e-01 -2.48453423e-01
9.23690647e-02 -6.88682437e-01 -1.15343094e+00 1.66338146e+00
3.72394651e-01 -1.75909340e-01 -1.28604556e-02 6.19358838e-01
1.47378147e+00 -6.80020899e-02 1.17634845e+00 -1.07981294e-01
2.05810976e+00 -3.28240871e-01 -1.17197263e+00 -6.53535575e-02
1.17261255e+00 -1.02536047e+00 6.01081252e-01 2.19817400e-01
-9.84141231e-01 2.05946192e-02 -7.90748656e-01 1.71676390e-02
-7.44884789e-01 3.79051678e-02 7.65479922e-01 8.46119106e-01
-8.08992445e-01 3.93795192e-01 -1.04288697e+00 -6.06597602e-01
7.52069950e-01 5.69726169e-01 -4.56922710e-01 2.12569714e-01
-1.38247180e+00 1.00980997e+00 5.70247293e-01 -2.70934969e-01
-3.78603965e-01 -1.12750578e+00 -9.74133551e-01 -2.54869223e-01
5.58581293e-01 -1.14323652e+00 1.38651562e+00 -2.14642361e-01
-9.95463431e-01 1.11139238e+00 -1.95792779e-01 -2.54720509e-01
-7.54463300e-02 -2.01159909e-01 -8.54921937e-01 2.51453370e-02
2.65735000e-01 4.43361193e-01 -2.81830549e-01 -5.65859556e-01
-4.53709215e-01 -3.68027180e-01 -2.43034974e-01 -1.43026821e-02
-8.22689235e-02 9.02544975e-01 -4.72334594e-01 -6.27829909e-01
-3.35794210e-01 -9.46004570e-01 -6.56586230e-01 -3.53525966e-01
-6.84349835e-01 -5.11010051e-01 -2.84854770e-01 -8.40887785e-01
2.00225019e+00 -1.60933149e+00 -1.07329682e-01 2.22603902e-01
3.23152214e-01 2.36948028e-01 2.41973661e-02 5.84296405e-01
-5.88470817e-01 1.00149405e+00 -1.91567957e-01 2.35438555e-01
1.39574930e-01 1.45209748e-02 1.53180033e-01 2.96991467e-01
4.68413442e-01 1.34680295e+00 -9.67174649e-01 -8.78763497e-01
-1.42936155e-01 4.49487835e-01 -7.14211464e-01 1.05394265e-02
-4.28054869e-01 1.24686420e-01 -1.08443403e+00 7.03040838e-01
4.49432045e-01 -3.97646874e-01 6.88440979e-01 -1.66830823e-01
3.49304080e-02 7.69636512e-01 -1.11638856e+00 2.01212931e+00
-2.97766238e-01 -3.96109372e-02 -2.22366333e-01 -5.11125088e-01
3.87812018e-01 6.69400573e-01 7.21772313e-01 -3.12690824e-01
2.79643446e-01 3.42328310e-01 -7.05878437e-02 -8.40427399e-01
7.09916279e-02 6.50764480e-02 -4.17168081e-01 2.42347717e-01
1.21795975e-01 1.50642648e-01 1.02007531e-01 1.70949340e-01
1.46740448e+00 2.16103032e-01 9.90803719e-01 -8.66824687e-02
3.44952255e-01 3.35368097e-01 7.20739961e-01 6.17030203e-01
-8.41419697e-02 4.82959569e-01 2.96899945e-01 -3.61995041e-01
-5.95802486e-01 -8.20138812e-01 -8.65407944e-01 6.45375371e-01
-7.74375319e-01 -8.52951109e-01 -4.82221633e-01 -1.02178478e+00
-8.62902477e-02 6.58108771e-01 -6.70264900e-01 3.80729347e-01
-5.73190570e-01 -9.19529855e-01 7.39758015e-01 5.65127015e-01
-2.23684520e-01 -1.38834238e+00 -4.76483583e-01 6.47908568e-01
-1.19113304e-01 -1.26119173e+00 -5.80590189e-01 3.97099644e-01
-6.87311530e-01 -1.72955036e+00 -5.83065212e-01 -9.32326317e-01
2.47424364e-01 -7.84938753e-01 1.41691256e+00 -8.13206807e-02
-4.33215499e-01 3.24101627e-01 -4.45932060e-01 -8.09784830e-01
-3.54859412e-01 1.79049503e-02 -3.32799762e-01 -7.16970384e-01
1.22946692e+00 -1.46026433e-01 -9.28350031e-01 1.98236071e-02
-1.03561485e+00 -4.37950701e-01 5.43325901e-01 4.15229440e-01
1.11642575e+00 -5.82741022e-01 3.70689064e-01 -1.48355794e+00
9.61514354e-01 -5.90804875e-01 -3.69072318e-01 4.97724086e-01
-8.49716842e-01 1.02569193e-01 1.12171859e-01 -2.40848437e-01
-6.62566185e-01 2.98078626e-01 -5.46010435e-01 1.65290877e-01
-7.37991512e-01 1.04424965e+00 -2.97208220e-01 1.96826816e-01
7.59978235e-01 -5.28500557e-01 -5.53610861e-01 -5.64394534e-01
7.90563643e-01 1.00607288e+00 -2.42276303e-02 -3.72166872e-01
1.52337521e-01 2.27467954e-01 -1.76431939e-01 -1.85166597e-01
-6.15869045e-01 -9.34486568e-01 -3.70030820e-01 5.78105092e-01
1.42770362e+00 -9.43876028e-01 -8.70386899e-01 -3.47474009e-01
-1.35196567e+00 -3.71414512e-01 -4.29253638e-01 9.41350698e-01
-3.90329182e-01 2.27437332e-01 -6.62918985e-01 -2.44329393e-01
-7.33529568e-01 -9.90645468e-01 9.54146504e-01 8.11598904e-04
-7.41431594e-01 -1.12900245e+00 6.45932496e-01 3.58694606e-02
-1.13827474e-02 5.25117040e-01 9.84035671e-01 -1.37153554e+00
1.82716534e-01 -3.37057620e-01 -1.69212952e-01 -2.92200297e-01
1.16344839e-01 -2.91787565e-01 -6.99329436e-01 1.77082151e-01
-1.70585066e-01 -1.49488240e-01 6.70606196e-01 6.56575263e-01
1.06866598e+00 -2.82216787e-01 -7.83329427e-01 5.02689004e-01
1.66883910e+00 5.07066190e-01 7.03459322e-01 4.66301173e-01
4.03824121e-01 4.50261414e-01 4.03477818e-01 4.15412277e-01
4.96678531e-01 6.50952160e-01 -1.46436185e-01 -4.55590814e-01
-3.14779401e-01 -3.75145227e-02 1.39558434e-01 5.89258015e-01
6.73216283e-02 -2.21276045e-01 -1.26398730e+00 6.36873364e-01
-1.67550933e+00 -5.60845375e-01 -3.67885709e-01 2.09455228e+00
1.30372167e+00 -1.40824229e-01 -4.01259698e-02 -5.08114815e-01
6.60339892e-01 -4.90031451e-01 -3.53011936e-01 -7.97878981e-01
-1.24056205e-01 9.74613249e-01 6.20278001e-01 5.59343338e-01
-1.38669074e+00 8.07678521e-01 5.54113626e+00 9.41596210e-01
-5.33039749e-01 4.75804448e-01 4.71482128e-01 1.90109506e-01
-5.73972881e-01 1.84112445e-01 -9.48597550e-01 1.86083317e-01
1.25609660e+00 -8.76785219e-02 -2.63812423e-01 4.63039726e-01
1.96483001e-01 4.78642255e-01 -1.32679176e+00 7.30270922e-01
-6.91689998e-02 -1.42330682e+00 -1.63111851e-01 1.57517090e-01
4.07057226e-01 4.22291249e-01 -1.95858523e-01 2.04131305e-01
7.78966010e-01 -1.18138468e+00 1.78899214e-01 4.64693248e-01
1.27872217e+00 -1.28167644e-01 9.37575877e-01 -4.59868610e-01
-1.14972484e+00 2.98169464e-01 -8.96621421e-02 6.61018491e-01
1.94852009e-01 7.28710830e-01 -9.13280308e-01 1.04109836e+00
4.53321368e-01 7.51389682e-01 -6.77797496e-01 1.39366770e+00
-5.51524699e-01 6.25939608e-01 -3.72405976e-01 -1.45982206e-01
-7.07811192e-02 2.78977573e-01 5.62611163e-01 1.67497087e+00
8.61043558e-02 5.74327052e-01 3.85497510e-01 5.50181508e-01
-3.32526386e-01 6.89242661e-01 -6.25128388e-01 -2.29711756e-01
4.26439047e-01 9.30849791e-01 -6.31203115e-01 -3.26221555e-01
-6.17810607e-01 4.05085891e-01 9.66828838e-02 3.90248895e-01
-8.60318959e-01 -5.23941934e-01 3.03819001e-01 -9.91718248e-02
3.54675502e-01 3.14810276e-01 -4.20440465e-01 -1.24530041e+00
9.24087986e-02 -1.16659403e+00 1.32404661e+00 -6.03188515e-01
-1.28821218e+00 9.76852000e-01 -7.12088346e-02 -1.29034364e+00
-1.46179855e-01 -5.38967788e-01 -1.13852158e-01 8.01735878e-01
-1.44912839e+00 -1.08044362e+00 3.30964625e-01 5.95313013e-01
2.01162919e-01 2.56457943e-02 1.57129979e+00 6.37608588e-01
-5.15183687e-01 7.22428799e-01 -1.94629908e-01 4.49458063e-01
8.95208120e-01 -1.11792839e+00 1.52726367e-01 3.92403096e-01
5.12964427e-02 8.81981611e-01 2.31867373e-01 -1.21640849e+00
-1.08916485e+00 -1.14315021e+00 1.66021764e+00 -9.96971965e-01
9.11209762e-01 -2.37071410e-01 -6.14007056e-01 8.07626843e-01
6.20895088e-01 -3.37058991e-01 1.53525841e+00 5.88214755e-01
-4.00739908e-01 6.16489470e-01 -9.49818075e-01 5.75324237e-01
8.89952958e-01 -3.36680114e-01 -8.02115262e-01 7.98868477e-01
9.01510239e-01 -4.12405878e-01 -1.54124892e+00 4.78065401e-01
3.83898228e-01 1.28002509e-01 9.99516308e-01 -1.47598195e+00
2.50759214e-01 -1.33852035e-01 -6.21711202e-02 -8.45723093e-01
-3.86688709e-01 -9.18008268e-01 2.05222145e-01 9.68738317e-01
1.42661750e+00 -4.41020817e-01 7.45716929e-01 9.81406748e-01
-2.56418318e-01 -8.45232904e-01 -9.54605818e-01 -3.94248128e-01
5.02607346e-01 -8.16267788e-01 6.33111715e-01 1.30265212e+00
4.98551637e-01 4.17321593e-01 3.15032423e-01 1.35213315e-01
4.16357160e-01 6.40365258e-02 -1.25064880e-01 -1.21655393e+00
-8.93245563e-02 -4.24297363e-01 -2.06432596e-01 -2.60275900e-01
-1.09404005e-01 -1.43136239e+00 -3.76045287e-01 -2.10098696e+00
4.87247378e-01 -6.20810866e-01 -5.52419960e-01 1.10567105e+00
-3.18339437e-01 -1.77583605e-01 3.81895564e-02 8.52001086e-02
-9.41483080e-01 -8.48348811e-02 7.91749954e-01 -1.37383968e-01
-5.55832803e-01 -2.07266793e-01 -1.04641747e+00 3.53830338e-01
9.43430483e-01 -1.22176111e+00 -8.83589126e-03 -4.33852941e-01
7.84787714e-01 2.53186673e-02 1.28770862e-02 -3.75792325e-01
4.41477060e-01 -1.90819055e-01 1.50514141e-01 -2.15745687e-01
-5.12786567e-01 -7.61509240e-01 1.99202448e-01 5.20692408e-01
-5.82332313e-01 7.25474060e-02 6.40660346e-01 3.89068604e-01
5.49859600e-03 -3.62657905e-01 6.95006400e-02 -1.20678455e-01
-4.73927677e-01 4.44975257e-01 -4.96752441e-01 4.70628858e-01
6.59315765e-01 1.02169245e-01 -3.93644214e-01 1.16868980e-01
-1.44202173e+00 3.61308098e-01 8.87828246e-02 2.99068630e-01
7.08301485e-01 -1.02248228e+00 -1.17983055e+00 -2.65400559e-01
3.38021845e-01 -2.36092642e-01 2.55309660e-02 1.02920806e+00
-4.02145028e-01 1.00113904e+00 2.04281420e-01 -1.20424375e-01
-1.16926730e+00 8.96739900e-01 4.49657701e-02 -8.70095432e-01
-4.77027923e-01 5.73234200e-01 -8.07362720e-02 -5.09758651e-01
2.78717995e-01 -4.53327924e-01 -7.00968325e-01 1.33528158e-01
5.07601976e-01 -1.93067878e-01 5.42751968e-01 -1.40422001e-01
-9.11544740e-01 6.56113744e-01 -3.30029160e-01 -1.36964113e-01
1.33920753e+00 3.21308881e-01 -3.86958346e-02 -2.64760345e-01
1.31999362e+00 2.11298883e-01 1.23203814e-01 -6.62615970e-02
4.92441982e-01 3.65784138e-01 3.05384427e-01 -1.38840473e+00
-6.96774542e-01 3.95491749e-01 7.98220515e-01 -1.74370214e-01
1.15107501e+00 2.66433984e-01 3.90707672e-01 2.59097695e-01
4.26343717e-02 -1.01152444e+00 -7.00515687e-01 3.38724971e-01
7.25578725e-01 -1.19599128e+00 3.26930612e-01 -6.70328796e-01
-7.50398993e-01 9.12805259e-01 -9.37655494e-02 2.28075594e-01
1.18177986e+00 3.34084362e-01 2.08235770e-01 -7.21345544e-01
-8.22945595e-01 -6.05222285e-01 6.07111990e-01 7.56291211e-01
9.05174494e-01 3.50626647e-01 -9.07161474e-01 1.25788939e+00
2.51129150e-01 6.07803345e-01 2.11158276e-01 1.21408689e+00
1.92524523e-01 -1.89282894e+00 2.47254632e-02 5.23935854e-01
-1.15352356e+00 -9.52701092e-01 -5.33364415e-01 9.08178270e-01
1.38128191e-01 1.24397790e+00 -4.91674423e-01 -1.51904538e-01
8.11026216e-01 3.69858533e-01 1.36490256e-01 -1.10996044e+00
-1.20636642e+00 3.74641478e-01 7.32117414e-01 -7.51686931e-01
-6.30222023e-01 -6.10128820e-01 -1.37868559e+00 2.59801745e-01
-3.20054233e-01 4.40528393e-01 6.65949523e-01 4.58441049e-01
8.11689198e-01 6.56781554e-01 3.39084491e-02 5.12304485e-01
-1.90185890e-01 -7.13813365e-01 2.24211961e-02 4.47925597e-01
-8.82391930e-02 -1.73241943e-01 -5.00029791e-03 1.97061509e-01] | [8.435770988464355, 8.709287643432617] |
ab4abac0-f0fc-4a51-bd32-f31b33d4b451 | unsupervised-learning-of-object-frames-by | 1706.02932 | null | http://arxiv.org/abs/1706.02932v2 | http://arxiv.org/pdf/1706.02932v2.pdf | Unsupervised learning of object frames by dense equivariant image labelling | One of the key challenges of visual perception is to extract abstract models
of 3D objects and object categories from visual measurements, which are
affected by complex nuisance factors such as viewpoint, occlusion, motion, and
deformations. Starting from the recent idea of viewpoint factorization, we
propose a new approach that, given a large number of images of an object and no
other supervision, can extract a dense object-centric coordinate frame. This
coordinate frame is invariant to deformations of the images and comes with a
dense equivariant labelling neural network that can map image pixels to their
corresponding object coordinates. We demonstrate the applicability of this
method to simple articulated objects and deformable objects such as human
faces, learning embeddings from random synthetic transformations or optical
flow correspondences, all without any manual supervision. | ['Hakan Bilen', 'Andrea Vedaldi', 'James Thewlis'] | 2017-06-09 | unsupervised-learning-of-object-frames-by-1 | http://papers.nips.cc/paper/6686-unsupervised-learning-of-object-frames-by-dense-equivariant-image-labelling | http://papers.nips.cc/paper/6686-unsupervised-learning-of-object-frames-by-dense-equivariant-image-labelling.pdf | neurips-2017-12 | ['unsupervised-facial-landmark-detection'] | ['computer-vision'] | [ 1.85588270e-01 9.47012380e-02 6.20818697e-02 -4.56728041e-01
-7.45688975e-02 -8.35979939e-01 7.83383667e-01 -1.88515216e-01
-4.18470860e-01 2.79228896e-01 1.18578814e-01 2.02672526e-01
1.52200729e-01 -4.88137752e-01 -1.01954377e+00 -6.09188497e-01
2.52465934e-01 5.23409009e-01 1.80645585e-01 5.87752424e-02
3.18034053e-01 9.08301532e-01 -1.41174984e+00 -1.76429555e-01
1.42049685e-01 7.99793363e-01 5.55577278e-02 9.14160907e-01
1.67803481e-01 6.45244181e-01 -2.80400336e-01 -2.26613343e-01
4.94655460e-01 -7.92103857e-02 -7.75587738e-01 7.29496300e-01
8.28863084e-01 -4.62024391e-01 -4.72087413e-01 1.18430686e+00
7.30742291e-02 2.36425057e-01 7.46555626e-01 -1.25865185e+00
-9.05826449e-01 -2.75300210e-03 -2.73273647e-01 -6.21989742e-02
3.10606718e-01 2.02717900e-01 8.49136174e-01 -1.16921973e+00
1.03442013e+00 1.74484646e+00 2.59250611e-01 5.80270648e-01
-1.70072234e+00 3.08297630e-02 3.07519704e-01 2.40372002e-01
-1.20895004e+00 -5.09962201e-01 9.16964412e-01 -7.88385630e-01
4.93419051e-01 3.29029076e-02 5.66082895e-01 1.07172441e+00
2.39512287e-02 3.85332495e-01 9.34088290e-01 -4.88070279e-01
3.87097985e-01 1.33990690e-01 -2.38971990e-02 7.11734354e-01
1.02868192e-01 -6.40000328e-02 -2.92199075e-01 2.67153513e-02
1.16226697e+00 3.46811026e-01 -3.52800041e-01 -1.11391842e+00
-1.43372679e+00 7.32356071e-01 5.90398908e-01 -2.53060386e-02
-2.20686525e-01 1.62426978e-01 -2.02234134e-01 9.97458994e-02
2.00928345e-01 3.56676698e-01 -5.11742294e-01 2.12540016e-01
-2.42402866e-01 1.34958565e-01 6.54849648e-01 1.00604773e+00
1.07330871e+00 2.43801698e-02 1.56675652e-01 2.26224318e-01
4.64409262e-01 9.07544732e-01 9.55679864e-02 -1.34340692e+00
2.07207650e-01 5.76113880e-01 3.07444364e-01 -1.26620960e+00
-2.92900234e-01 -1.07830025e-01 -7.15337634e-01 6.64787352e-01
5.87501407e-01 1.31551161e-01 -9.68266785e-01 1.91640294e+00
6.10950589e-01 1.08042665e-01 -1.07163407e-01 9.01863873e-01
2.34411135e-01 2.56414413e-01 -2.91537195e-01 8.21089558e-03
1.26785159e+00 -6.80541158e-01 -4.27978039e-01 -1.62303418e-01
1.31933406e-01 -6.88577592e-01 9.11329985e-01 9.53671709e-02
-1.17949820e+00 -8.01404059e-01 -9.32277560e-01 -5.93115091e-01
-4.86349851e-01 4.14268672e-02 2.50040978e-01 3.52157116e-01
-9.42830086e-01 4.27710027e-01 -1.19019079e+00 -3.89458358e-01
3.29741746e-01 3.45493168e-01 -8.11143041e-01 -1.71295092e-01
-4.44357753e-01 1.00664544e+00 1.70078233e-01 2.43279144e-01
-1.11477911e+00 -6.39030457e-01 -1.10474312e+00 -4.29742318e-03
2.94239879e-01 -8.55927706e-01 9.46230650e-01 -9.33701515e-01
-1.50331461e+00 8.80087852e-01 -2.82340854e-01 5.34542352e-02
5.42680442e-01 -2.69814909e-01 1.48604199e-01 3.83084804e-01
-1.60519972e-01 7.84862936e-01 1.37304580e+00 -1.22254407e+00
-1.83624789e-01 -9.35717046e-01 3.06701422e-01 2.88803309e-01
-1.38616860e-01 -1.83347657e-01 -3.53031069e-01 -2.67890066e-01
2.34878153e-01 -1.17655182e+00 -4.36430067e-01 6.28358483e-01
-2.26882100e-01 1.13630958e-01 1.12575924e+00 -5.68179667e-01
1.54629394e-01 -2.11147046e+00 9.20368791e-01 1.58977568e-01
3.01129401e-01 3.27619947e-02 -3.49274755e-01 -9.26970467e-02
-4.67950739e-02 -8.44252780e-02 5.75731061e-02 -3.94599766e-01
8.44530687e-02 4.14242387e-01 -1.50467500e-01 7.84974098e-01
5.36435068e-01 1.07344735e+00 -9.27093923e-01 -6.98671117e-02
6.56544209e-01 8.08620930e-01 -7.83955574e-01 3.14581841e-01
-3.62969071e-01 9.89257932e-01 -3.76526892e-01 3.93681109e-01
6.40836596e-01 -2.16068760e-01 -6.09253980e-02 -4.29090738e-01
1.82071611e-01 1.12198070e-01 -1.38973355e+00 1.82079768e+00
-3.10468048e-01 7.24246144e-01 1.92683160e-01 -1.08561337e+00
7.11614728e-01 9.21551362e-02 4.24048215e-01 -1.75294697e-01
3.69920462e-01 -1.89147860e-01 -2.32494250e-01 -5.35892069e-01
1.67079538e-01 1.37365833e-01 2.08778411e-01 3.61322552e-01
4.47142780e-01 -3.93402427e-01 -1.75288230e-01 1.95761397e-01
8.85650933e-01 1.49402380e-01 2.44031563e-01 -5.65617196e-02
6.46821022e-01 -6.23878062e-01 4.58231896e-01 2.39597738e-01
-2.93642152e-02 7.60225952e-01 4.62758541e-01 -7.37460375e-01
-1.40208161e+00 -1.33147025e+00 -7.81316459e-02 6.43338621e-01
9.51496214e-02 -2.15959325e-02 -6.40618145e-01 -6.73543274e-01
9.88236889e-02 1.59703568e-01 -8.22941303e-01 -1.91725835e-01
-5.01998007e-01 -1.64556429e-01 -8.40444937e-02 4.95284647e-01
2.98207909e-01 -8.20299029e-01 -7.15339661e-01 -6.30979687e-02
-4.17832844e-02 -1.69080389e+00 -5.94154596e-01 -1.42593145e-01
-7.47653484e-01 -1.20146024e+00 -5.20864666e-01 -5.93660235e-01
1.18191934e+00 3.63930732e-01 8.88689458e-01 -1.36003822e-01
-5.33763230e-01 9.53112483e-01 2.20510345e-02 -2.65664816e-01
-4.13838863e-01 -2.56528407e-01 2.14972734e-01 5.50060391e-01
1.06001012e-01 -7.42660701e-01 -5.40181756e-01 3.40777725e-01
-9.82791781e-01 -1.00334343e-02 2.85090774e-01 6.09454513e-01
6.74573004e-01 -4.31132823e-01 -9.99462679e-02 -6.59403265e-01
-1.07308783e-01 -1.23895425e-02 -8.60971332e-01 2.08968192e-01
-1.41903684e-01 2.46775895e-01 2.69529819e-01 -6.43341124e-01
-6.71672821e-01 6.19482458e-01 2.18368977e-01 -8.44401538e-01
-3.74643773e-01 -7.92006925e-02 -5.79883695e-01 -3.70341390e-01
6.74448431e-01 -6.55678809e-02 2.15442628e-01 -5.07149220e-01
9.32197988e-01 2.41400599e-01 6.89545512e-01 -4.73155916e-01
1.27266335e+00 9.42166030e-01 3.29733968e-01 -1.02273715e+00
-8.30908477e-01 -2.60650009e-01 -1.61277902e+00 -1.38972864e-01
1.22155094e+00 -8.63621831e-01 -8.35751057e-01 4.66759801e-01
-1.46809494e+00 -1.87256441e-01 -4.28450167e-01 6.35845065e-01
-7.79293418e-01 4.42319632e-01 -2.64284700e-01 -3.06757480e-01
2.18030959e-01 -1.13112187e+00 1.28683829e+00 1.52150795e-01
6.47528376e-03 -1.14402223e+00 -1.88214719e-01 2.22890437e-01
1.83364496e-01 3.80216271e-01 9.86486375e-01 -2.01529264e-01
-1.07976747e+00 -1.66441634e-01 -1.03421785e-01 8.02003324e-01
4.91097242e-01 1.28328189e-01 -1.00637853e+00 -3.65789950e-01
2.59214073e-01 -2.43202358e-01 4.64953244e-01 2.88320988e-01
8.94970417e-01 -4.19972479e-01 -9.10518020e-02 7.82927930e-01
1.21725702e+00 -1.08850919e-01 2.92316735e-01 -2.76427060e-01
1.17855418e+00 6.10514641e-01 1.03649311e-01 1.44148529e-01
3.33208293e-01 7.88434565e-01 7.00658202e-01 -8.96124691e-02
-1.35022849e-01 -2.56927699e-01 3.35482270e-01 5.88319421e-01
-2.37292320e-01 2.39539757e-01 -5.25521338e-01 3.23095560e-01
-1.51445019e+00 -6.56663716e-01 2.83329517e-01 2.29547501e+00
4.33652043e-01 -7.36811012e-02 -8.80681202e-02 1.31452009e-01
5.68287551e-01 9.81423557e-02 -7.67871976e-01 -7.07472116e-02
2.54759891e-03 2.47946940e-02 3.26700658e-01 7.95624435e-01
-1.01803565e+00 9.72039223e-01 6.37597084e+00 -1.28581226e-01
-1.39873111e+00 -5.58651350e-02 2.92484224e-01 1.41064927e-01
-2.90978134e-01 2.00360537e-01 -6.73790991e-01 1.70727342e-01
4.80579853e-01 7.56658316e-02 6.21063232e-01 7.71084487e-01
-4.93282788e-02 3.33768457e-01 -1.55247879e+00 8.88008714e-01
4.74275351e-01 -1.27159584e+00 2.51602709e-01 3.07233751e-01
7.27090120e-01 -1.19965091e-01 2.80584663e-01 -2.14341506e-01
3.22157800e-01 -8.08174431e-01 7.21480548e-01 5.52727818e-01
5.08037448e-01 -1.39541999e-01 1.72440007e-01 2.80644476e-01
-7.89782286e-01 5.65545969e-02 -5.43096423e-01 -1.47462174e-01
1.45447001e-01 2.98107356e-01 -7.27344930e-01 1.64736032e-01
4.93961930e-01 9.89550173e-01 -5.96126735e-01 5.81687033e-01
-2.78803796e-01 2.61179078e-02 -3.34614635e-01 4.00254041e-01
-9.60769802e-02 -4.07269120e-01 5.96977532e-01 6.81189835e-01
1.04010262e-01 1.82036325e-01 1.95786968e-01 9.82937157e-01
-1.05695836e-01 -1.20071605e-01 -8.53661954e-01 1.50222093e-01
1.10271782e-01 1.42029488e+00 -7.83727765e-01 -1.96424022e-01
-5.76212764e-01 9.36994135e-01 3.07272285e-01 6.58432066e-01
-5.12444258e-01 8.97594616e-02 9.66797173e-01 1.32326961e-01
6.42519653e-01 -8.61952782e-01 1.12848073e-01 -1.53322601e+00
1.60104752e-01 -6.73961639e-01 -1.07737735e-01 -9.34404492e-01
-9.68572736e-01 2.93785244e-01 -3.92730422e-02 -1.15822327e+00
-1.78793937e-01 -1.06321931e+00 -3.12113136e-01 5.61766922e-01
-1.33068955e+00 -1.24066138e+00 -5.79125285e-01 9.30943191e-01
4.53063190e-01 7.32622109e-04 6.72996104e-01 -1.47138303e-02
3.95579264e-02 1.85062006e-01 -1.13933749e-01 3.32797796e-01
4.07376260e-01 -1.28647041e+00 4.74743366e-01 7.47742355e-01
6.71457827e-01 5.35566032e-01 5.74479282e-01 3.45518477e-02
-1.74581242e+00 -1.05565464e+00 6.30370498e-01 -1.15879452e+00
5.39035857e-01 -8.93067777e-01 -7.09220231e-01 1.13782537e+00
-1.14182591e-01 7.04872072e-01 1.27711639e-01 -1.19457975e-01
-6.51655436e-01 -2.38061607e-01 -9.63881195e-01 5.75446129e-01
1.15109074e+00 -7.87805915e-01 -5.63670635e-01 3.92651498e-01
7.70592988e-01 -6.25536799e-01 -8.39313984e-01 2.33202472e-01
5.82461536e-01 -6.76588178e-01 1.37980711e+00 -9.86664236e-01
5.37553430e-02 -4.04161215e-01 -4.03059334e-01 -1.24311304e+00
-2.54036248e-01 -5.13905168e-01 8.68441388e-02 9.59254920e-01
-1.03830762e-01 -4.54008490e-01 5.86695671e-01 7.35080540e-01
2.63318509e-01 -1.36743784e-01 -9.33697999e-01 -5.81091762e-01
5.02258725e-02 -1.51684612e-01 4.71348286e-01 8.15684080e-01
-5.26055455e-01 5.86327255e-01 -4.42032754e-01 4.52679992e-01
7.63046682e-01 7.86021575e-02 1.22930384e+00 -1.52561641e+00
-1.85415834e-01 -1.02631412e-01 -1.18611014e+00 -1.15761650e+00
4.44408327e-01 -7.16521323e-01 -6.49734661e-02 -1.02721930e+00
6.20930605e-02 -8.19058642e-02 -1.01402879e-01 3.47267836e-01
7.28685930e-02 3.70220333e-01 4.25435245e-01 -3.22491527e-02
-5.49226582e-01 7.83436894e-01 1.49871099e+00 -2.50296950e-01
1.31630404e-02 9.90229473e-02 -4.07008290e-01 9.72652555e-01
3.04537028e-01 -3.13569009e-01 -4.82794583e-01 -8.34992230e-01
6.54638335e-02 -8.14496204e-02 7.08583176e-01 -6.59781337e-01
5.88300526e-02 -3.27043593e-01 6.26275003e-01 -1.04556203e-01
5.79020739e-01 -1.13505566e+00 6.90213963e-02 3.25498044e-01
-4.25282598e-01 2.15173617e-01 -1.53614074e-01 6.95380867e-01
-6.69608638e-02 6.41210601e-02 8.89979243e-01 -1.97632089e-01
-4.47403699e-01 5.53677559e-01 -1.00388266e-02 -1.65451579e-02
1.03941083e+00 -5.88157363e-02 -6.14106059e-02 -2.89415926e-01
-1.09920704e+00 -1.40015915e-01 7.26375163e-01 8.08544338e-01
5.78265965e-01 -1.39405179e+00 -5.44834793e-01 6.79117322e-01
6.29876703e-02 2.84575522e-01 -2.44568419e-02 6.48417711e-01
-3.67275059e-01 3.61172229e-01 -5.00890970e-01 -1.05127323e+00
-1.34941065e+00 9.00390208e-01 4.46183145e-01 3.81106704e-01
-7.09265590e-01 4.79813784e-01 5.30584037e-01 -5.62153637e-01
2.09689364e-01 -8.05408657e-01 4.76816073e-02 -2.40370438e-01
4.27221447e-01 7.82452300e-02 5.21702925e-03 -1.05950665e+00
-2.02727646e-01 1.16905177e+00 9.29346755e-02 -2.12029532e-01
1.23819697e+00 -3.35533619e-01 -9.70244780e-02 5.99494576e-01
1.41864443e+00 -1.00770414e-01 -1.65044987e+00 -6.23346984e-01
-9.55542699e-02 -7.88618147e-01 -2.09282219e-01 -1.02679670e-01
-1.17504966e+00 1.31129277e+00 5.98689675e-01 -1.87777039e-02
7.88523972e-01 9.38622579e-02 1.50559321e-01 5.99094212e-01
4.13028419e-01 -6.45166755e-01 5.29389560e-01 4.42607939e-01
9.99508023e-01 -1.32769001e+00 -1.97458073e-01 -3.32468092e-01
-2.29829296e-01 1.06210077e+00 3.43183994e-01 -3.32020670e-01
1.05938292e+00 -1.29044518e-01 1.78903192e-01 -2.39585917e-02
-6.48068249e-01 -4.62642908e-02 4.01175290e-01 8.60310495e-01
-1.29108459e-01 -1.44432232e-01 3.69795471e-01 -2.12408379e-01
1.37414653e-02 -2.30527043e-01 5.40018559e-01 5.76058209e-01
-2.35627368e-01 -9.54609573e-01 -1.91456214e-01 -1.97846115e-01
-2.63902605e-01 3.40488166e-01 -3.14036697e-01 7.07255065e-01
1.23544812e-01 4.65955287e-01 2.82807976e-01 -9.12642553e-02
4.38462049e-01 1.12095885e-01 9.00193810e-01 -7.16554880e-01
2.10948452e-01 9.74243879e-02 -7.08649158e-01 -7.62243211e-01
-6.99455917e-01 -9.76978600e-01 -9.89352882e-01 4.30014729e-02
1.14618510e-03 -2.94655532e-01 8.86303842e-01 1.03041053e+00
3.12981635e-01 2.23401099e-01 7.61965811e-01 -1.38120019e+00
-5.57278156e-01 -7.27325022e-01 -5.36650896e-01 8.96543562e-01
8.39844167e-01 -9.20649588e-01 -5.16839147e-01 6.37963235e-01] | [8.519933700561523, -2.440596342086792] |
4f2331f2-6970-4b72-84a5-aeb8a52a816b | gpu-accelerated-sift-aided-source | 2207.14507 | null | https://arxiv.org/abs/2207.14507v1 | https://arxiv.org/pdf/2207.14507v1.pdf | GPU-accelerated SIFT-aided source identification of stabilized videos | Video stabilization is an in-camera processing commonly applied by modern acquisition devices. While significantly improving the visual quality of the resulting videos, it has been shown that such operation typically hinders the forensic analysis of video signals. In fact, the correct identification of the acquisition source usually based on Photo Response non-Uniformity (PRNU) is subject to the estimation of the transformation applied to each frame in the stabilization phase. A number of techniques have been proposed for dealing with this problem, which however typically suffer from a high computational burden due to the grid search in the space of inversion parameters. Our work attempts to alleviate these shortcomings by exploiting the parallelization capabilities of Graphics Processing Units (GPUs), typically used for deep learning applications, in the framework of stabilised frames inversion. Moreover, we propose to exploit SIFT features {to estimate the camera momentum and} %to identify less stabilized temporal segments, thus enabling a more accurate identification analysis, and to efficiently initialize the frame-wise parameter search of consecutive frames. Experiments on a consolidated benchmark dataset confirm the effectiveness of the proposed approach in reducing the required computational time and improving the source identification accuracy. {The code is available at \url{https://github.com/AMontiB/GPU-PRNU-SIFT}}. | ['Fernando Pérez-González', "Stefano Dell'Anna", 'Giulia Boato', 'Cecilia Pasquini', 'Andrea Montibeller'] | 2022-07-29 | null | null | null | null | ['video-stabilization'] | ['computer-vision'] | [ 3.07534635e-01 -5.50134182e-01 2.98637562e-02 1.20118029e-01
-6.97765708e-01 -4.85712975e-01 3.51483911e-01 3.99397224e-01
-6.46595895e-01 4.35440868e-01 -3.85657638e-01 -1.40204892e-01
-5.02213053e-02 -4.62649673e-01 -6.72071993e-01 -9.94273901e-01
7.33817965e-02 2.83797574e-03 1.67658404e-01 2.20549569e-01
6.83094025e-01 6.81445360e-01 -1.61280394e+00 -9.43169445e-02
6.81754410e-01 1.25970089e+00 4.94761243e-02 5.51458418e-01
2.10257709e-01 7.05415010e-01 -3.47045720e-01 -4.31940973e-01
3.41988921e-01 -1.80295721e-01 -4.12751019e-01 1.22022606e-01
4.00005966e-01 -4.91826504e-01 -1.56762809e-01 1.31921387e+00
4.96576309e-01 2.34471500e-01 4.55164999e-01 -1.09495485e+00
2.20786929e-01 1.08478487e-01 -8.38774085e-01 5.50617814e-01
3.68860334e-01 1.70104370e-01 7.03645647e-01 -7.76207924e-01
4.49766785e-01 7.07733393e-01 6.69111788e-01 5.42063527e-02
-1.12871420e+00 -7.84390986e-01 -5.24406731e-01 7.55851626e-01
-1.70145953e+00 -6.85168266e-01 9.62099373e-01 -5.12704968e-01
4.66905743e-01 1.81697458e-01 5.23663759e-01 8.82915854e-01
9.70392525e-02 3.77943248e-01 8.67273033e-01 -5.11569023e-01
2.20247418e-01 -1.03063777e-01 2.02231213e-01 5.67382216e-01
3.10056359e-01 -1.38063148e-01 -6.68645144e-01 -2.13490829e-01
9.24554408e-01 -2.66331971e-01 -4.49541122e-01 -3.41850370e-01
-1.03541958e+00 7.17652440e-01 4.04935777e-02 2.85562962e-01
-5.23154557e-01 1.18030548e-01 8.15827012e-01 8.56993943e-02
3.76571149e-01 2.52271265e-01 -6.41778931e-02 -4.95312512e-01
-1.11939692e+00 2.62015616e-03 4.74745840e-01 5.96092761e-01
7.66275704e-01 2.41172612e-01 2.12935314e-01 4.86220777e-01
-7.02000335e-02 3.29692334e-01 4.77355331e-01 -1.06454694e+00
3.93243611e-01 4.01879370e-01 1.78226694e-01 -1.33441687e+00
-1.74841329e-01 -1.78753465e-01 -9.70922470e-01 2.42277086e-01
7.61251509e-01 7.97626823e-02 -2.10651442e-01 1.35562634e+00
5.81963897e-01 4.75288391e-01 -2.42289633e-01 9.24807608e-01
2.02624977e-01 4.65272337e-01 -7.01992423e-04 -3.60289484e-01
1.38491595e+00 -5.46533585e-01 -5.78072309e-01 2.23889828e-01
5.28491795e-01 -1.02485430e+00 8.87197435e-01 7.44392514e-01
-8.40699315e-01 -8.02615166e-01 -1.01182032e+00 1.55920118e-01
7.28846416e-02 5.86610734e-01 2.75434941e-01 8.36153269e-01
-7.91001618e-01 7.00805008e-01 -9.76764679e-01 -1.47799954e-01
4.02284652e-01 5.95769107e-01 -3.80554587e-01 3.07378381e-01
-8.84943128e-01 4.84407306e-01 5.04486084e-01 2.94809043e-01
-4.39612657e-01 -5.73947132e-01 -6.64765596e-01 1.01270065e-01
4.10653472e-01 -1.76939428e-01 8.26532304e-01 -1.14108074e+00
-1.41259575e+00 6.15102112e-01 -5.31686135e-02 -6.26660049e-01
7.39897847e-01 -4.00996685e-01 -3.93515872e-03 6.02308273e-01
-5.89883812e-02 1.71334594e-01 1.43326700e+00 -6.68000340e-01
-3.86656612e-01 -2.89096534e-01 -1.63841799e-01 6.88301623e-02
-6.76645041e-01 1.81188464e-01 -6.51372552e-01 -5.88763654e-01
-9.50358286e-02 -1.11510265e+00 -5.30600827e-03 -1.95988625e-01
-1.95833609e-01 1.37818024e-01 7.57377625e-01 -9.26533461e-01
1.18294132e+00 -2.23600268e+00 1.73196226e-01 2.77251333e-01
-6.10918514e-02 6.06692672e-01 4.22457576e-01 2.49639466e-01
-2.77242452e-01 -3.97427678e-01 1.05816284e-02 -3.70732456e-01
-3.57417136e-01 -4.05676454e-01 -4.56023574e-01 1.05132461e+00
-1.22753039e-01 1.52683511e-01 -5.77242494e-01 -6.75807118e-01
7.41632700e-01 5.83266914e-01 -4.56451833e-01 2.51365513e-01
1.52367428e-01 7.99114466e-01 -4.96517092e-01 4.77814496e-01
6.75106585e-01 -4.80105430e-02 2.30243117e-01 -5.54217339e-01
-2.71479875e-01 -1.63762495e-01 -1.58569753e+00 1.48591650e+00
-3.69243175e-01 6.67045772e-01 1.12398043e-01 -1.10604823e+00
7.39893436e-01 3.61462593e-01 7.91276693e-01 -4.90292370e-01
5.86345553e-01 3.76590669e-01 -1.65123135e-01 -4.38605785e-01
5.80623806e-01 4.09474134e-01 2.53670871e-01 3.71598333e-01
-1.81581393e-01 9.35865268e-02 3.17108333e-01 -1.14263020e-01
6.79166198e-01 3.67280364e-01 4.62804109e-01 -1.85214341e-01
1.09882665e+00 -2.19283149e-01 3.56394857e-01 4.08439755e-01
-1.68365970e-01 5.17319500e-01 4.62407261e-01 -5.16178191e-01
-1.15418243e+00 -4.52122062e-01 -1.98570818e-01 5.28754294e-01
2.00433806e-01 -4.61970866e-01 -1.03372622e+00 -1.76845379e-02
-4.11318004e-01 5.08525670e-01 -2.08087087e-01 -2.22710962e-03
-8.97356927e-01 -6.50162637e-01 5.67452848e-01 3.03447068e-01
4.34726864e-01 -8.16533923e-01 -1.35714531e+00 1.68778509e-01
-2.49575153e-01 -1.32903707e+00 -3.13179851e-01 -2.15969086e-01
-1.01112914e+00 -1.30132806e+00 -5.87542951e-01 -2.02535167e-01
6.09285295e-01 2.33599752e-01 5.60530186e-01 3.39206606e-01
-2.92994261e-01 3.91340345e-01 -3.16281617e-01 1.29932597e-01
-3.14451367e-01 -4.35227789e-02 3.06223601e-01 3.69110554e-01
1.56952560e-01 -5.30731678e-01 -7.17374384e-01 3.91264737e-01
-1.06216657e+00 4.10405882e-02 2.59714603e-01 7.95832276e-01
5.59540927e-01 1.96622849e-01 1.00321859e-01 -3.74112010e-01
2.64310658e-01 -1.97850630e-01 -1.08774805e+00 3.40558738e-02
-2.78393537e-01 -1.23883992e-01 8.82133424e-01 -4.19562578e-01
-7.04695106e-01 2.66958177e-01 -1.13703713e-01 -9.30291057e-01
-1.05001152e-01 1.32624462e-01 2.26974431e-02 -4.59401190e-01
3.18499029e-01 3.39847624e-01 1.08205654e-01 -2.99632758e-01
7.88683537e-03 4.62250799e-01 6.21195138e-01 -5.63008666e-01
7.47303128e-01 5.24844944e-01 3.32783967e-01 -1.25729406e+00
-2.01119632e-01 -6.17681980e-01 -7.51518607e-01 -5.01059592e-01
7.93759286e-01 -8.48667204e-01 -1.10504055e+00 7.03906775e-01
-1.19297516e+00 1.41588584e-01 2.93599386e-02 5.87260902e-01
-6.14519000e-01 1.13667130e+00 -4.11125511e-01 -9.23270941e-01
-5.56994915e-01 -1.56304181e+00 8.81608307e-01 1.98177442e-01
-2.76366323e-01 -7.48030245e-01 -1.80330530e-01 4.80204403e-01
8.48585516e-02 2.59993792e-01 6.47084534e-01 -3.31849992e-01
-6.00340307e-01 -3.84615511e-01 -9.30637047e-02 3.81584734e-01
-2.66303029e-02 1.33306861e-01 -9.09187436e-01 -5.74016213e-01
3.65346968e-01 4.23906632e-02 4.01748419e-01 5.27714610e-01
1.20046985e+00 4.00970951e-02 -4.03442085e-02 7.37281084e-01
1.49463844e+00 1.86983258e-01 6.52615666e-01 5.90484381e-01
7.72556245e-01 4.13174719e-01 9.65508282e-01 7.80213773e-01
-1.81059927e-01 1.12577546e+00 5.60225189e-01 1.79039761e-01
1.16536111e-01 2.55622059e-01 4.10418004e-01 6.54312551e-01
-3.75361264e-01 -6.95214868e-02 -7.81839073e-01 3.62956315e-01
-1.72559190e+00 -9.35890079e-01 -4.16629553e-01 2.69441366e+00
4.54819500e-01 2.37098858e-02 1.51076078e-01 5.86692810e-01
9.21090424e-01 1.92528412e-01 -3.05549473e-01 -2.00760722e-01
7.61160403e-02 2.97615796e-01 7.38034606e-01 4.97848213e-01
-1.18209636e+00 6.69247210e-01 4.09887505e+00 1.14803338e+00
-1.42378557e+00 1.50178364e-02 4.90804851e-01 -4.41115797e-02
4.45932329e-01 8.72010440e-02 -7.83457816e-01 6.84006810e-01
8.41713130e-01 -8.91615599e-02 2.76304543e-01 8.42413545e-01
4.57552433e-01 -4.92202342e-01 -8.44482183e-01 1.44513214e+00
1.69588402e-01 -1.21178246e+00 -1.45802051e-01 3.33782323e-02
1.68586180e-01 -4.60391879e-01 1.98930234e-01 -3.52626115e-01
-6.92202389e-01 -4.75860476e-01 9.13827240e-01 3.54181290e-01
8.13401401e-01 -8.80563915e-01 6.26885772e-01 1.20806694e-01
-1.25183237e+00 -1.02424867e-01 -3.57240260e-01 6.22036681e-02
2.12045670e-01 4.82155263e-01 -8.35414112e-01 6.35246098e-01
7.46847868e-01 4.62190151e-01 -5.77745020e-01 1.09874880e+00
-5.01709022e-02 5.59714198e-01 -4.12179977e-01 2.72672892e-01
-1.41080318e-03 -5.57145536e-01 8.12924802e-01 9.92619872e-01
6.08247459e-01 4.95787337e-03 -1.49813503e-01 4.15227294e-01
2.51327276e-01 3.22173476e-01 -2.42683694e-01 1.37191743e-01
2.53793359e-01 1.20234692e+00 -1.01575160e+00 -1.00521103e-01
-1.52751893e-01 9.87120509e-01 7.34185800e-04 -4.13104184e-02
-1.06035089e+00 -1.91969916e-01 5.89060247e-01 2.32890517e-01
4.91012752e-01 -5.72056234e-01 -1.82172075e-01 -1.24282849e+00
1.81363076e-01 -1.03046262e+00 3.36386025e-01 -5.02083063e-01
-5.73891461e-01 5.63848734e-01 -3.34647521e-02 -1.64274323e+00
-3.30176979e-01 -3.98428977e-01 -3.81400168e-01 6.68604314e-01
-1.33149946e+00 -8.65389228e-01 -3.13962042e-01 7.40463793e-01
7.27282166e-01 -8.51004645e-02 5.59877872e-01 6.15099072e-01
-6.94815457e-01 5.39481640e-01 2.03096181e-01 -3.22861820e-02
7.37787127e-01 -7.72961438e-01 1.43470794e-01 1.15509808e+00
6.38990402e-02 5.93771756e-01 1.02414191e+00 -5.25741577e-01
-1.55748689e+00 -5.42859674e-01 3.49395156e-01 -6.69707656e-02
8.04322660e-01 -1.38796613e-01 -8.65798771e-01 3.07492673e-01
1.61522374e-01 -1.81768134e-01 5.79002500e-01 -4.63573456e-01
5.62640466e-02 -2.45721102e-01 -8.54308724e-01 3.75964522e-01
2.87322551e-01 -5.31825244e-01 -2.05400273e-01 1.86923862e-01
1.45248147e-02 -5.51208854e-01 -7.67439067e-01 4.44995388e-02
5.75064659e-01 -1.28160894e+00 1.17735898e+00 3.35641056e-02
2.18082815e-01 -5.15458286e-01 9.60481018e-02 -6.75420284e-01
3.93197052e-02 -7.23789275e-01 -9.77789834e-02 1.31682336e+00
-3.69752973e-01 -3.74839395e-01 7.98174381e-01 4.03011918e-01
1.94192141e-01 -4.46136534e-01 -1.23834467e+00 -5.23362994e-01
-5.79403877e-01 -5.27300417e-01 1.59554109e-01 7.29866028e-01
-4.12553698e-01 -8.37504044e-02 -7.18122542e-01 5.26668370e-01
1.01439023e+00 -1.90268899e-03 9.61490691e-01 -1.01128149e+00
-4.08447683e-01 -2.05376193e-01 -6.60185158e-01 -6.71217620e-01
1.19073257e-01 -2.39988461e-01 -3.70975107e-01 -5.83152711e-01
-1.26069054e-01 -2.69236773e-01 -9.43514928e-02 5.12130447e-02
4.58380394e-02 5.90951085e-01 3.33472133e-01 5.06309807e-01
-3.53864342e-01 2.48766899e-01 5.86365104e-01 1.48435339e-01
-1.13824882e-01 1.21847495e-01 4.90480885e-02 8.76741707e-01
6.34349823e-01 -4.42263335e-01 -2.36825868e-01 -3.97676200e-01
2.20801309e-01 1.74903810e-01 6.20546043e-01 -1.30063355e+00
4.39831018e-01 2.61104256e-01 2.97665834e-01 -6.93901777e-01
5.85659564e-01 -9.63838995e-01 5.31012714e-01 5.69336772e-01
4.51589711e-02 2.49723643e-01 2.29302883e-01 3.62218678e-01
-3.22090149e-01 -5.61164260e-01 1.01561964e+00 1.76953509e-01
-8.22655559e-01 5.75584248e-02 -2.83640772e-01 -4.81726646e-01
1.01275074e+00 -4.92863685e-01 1.23016559e-01 -2.54270911e-01
-3.20662320e-01 -3.75188142e-01 7.45826006e-01 -3.65226865e-02
4.51945364e-01 -1.01453805e+00 -3.61007094e-01 1.03188761e-01
-2.60456592e-01 -3.37885708e-01 6.40581489e-01 1.35685945e+00
-1.13376594e+00 1.98873535e-01 -3.99718553e-01 -7.22496986e-01
-1.58200824e+00 5.97361803e-01 1.69063628e-01 -1.50718302e-01
-6.13095045e-01 5.95875680e-01 -1.30019605e-01 4.92378980e-01
1.46972179e-01 -1.21055543e-01 -3.84667695e-01 3.55271906e-01
5.42551219e-01 7.88837790e-01 2.74888784e-01 -1.04927123e+00
-3.53200734e-01 8.83602679e-01 -2.23657843e-02 1.51044324e-01
1.27632177e+00 -2.62689531e-01 -2.04117373e-01 4.01044916e-03
1.21325469e+00 1.20284222e-01 -1.28748095e+00 1.04167156e-01
-5.76267727e-02 -8.36263120e-01 1.84745297e-01 2.23574877e-01
-1.17227006e+00 8.58649194e-01 8.43239248e-01 1.27638623e-01
1.31172180e+00 -7.65879095e-01 8.27482224e-01 2.01579005e-01
3.95920455e-01 -1.05592704e+00 -1.50517017e-01 1.26027644e-01
4.73285288e-01 -9.89297569e-01 3.79496962e-01 -3.73762578e-01
-3.56519073e-01 1.53111672e+00 1.49583057e-01 -2.81621009e-01
2.89857537e-01 1.04303189e-01 -1.34497911e-01 8.82991776e-02
1.05125764e-02 2.43719339e-01 -7.48965219e-02 7.82116354e-02
2.38101125e-01 -3.31449300e-01 -5.69443703e-01 3.28878239e-02
3.79148088e-02 2.85983272e-03 6.15600884e-01 7.75950313e-01
2.53235251e-02 -1.17077315e+00 -8.37455988e-01 -1.44626815e-02
-6.43471062e-01 1.10214585e-02 1.66613877e-01 5.95919192e-01
5.33263162e-02 8.36403906e-01 -8.23864341e-02 -1.52471334e-01
7.82033876e-02 -1.90565318e-01 5.04840970e-01 -7.58276805e-02
-6.43122435e-01 2.68576711e-01 -2.01007053e-01 -6.60184085e-01
-6.45597637e-01 -1.03671515e+00 -7.94912755e-01 -2.93752760e-01
-3.43868762e-01 1.65895522e-01 6.47590995e-01 7.88553119e-01
1.46460608e-01 8.99085179e-02 6.91235900e-01 -1.26410806e+00
-5.34821451e-01 -5.57917356e-01 -5.48660100e-01 4.79223311e-01
1.73136771e-01 -8.06560874e-01 -4.45150137e-01 2.98083276e-01] | [12.179353713989258, 0.7736184000968933] |
4dca89c1-652c-47da-b66d-4b2bbc882da4 | some-theoretical-considerations-in-off-the | null | null | https://aclanthology.org/R15-2002 | https://aclanthology.org/R15-2002.pdf | Some Theoretical Considerations in Off-the-Shelf Text Analysis Software | null | ['Emma Franklin'] | 2015-09-01 | some-theoretical-considerations-in-off-the-1 | https://aclanthology.org/R15-2002 | https://aclanthology.org/R15-2002.pdf | ranlp-2015-9 | ['deception-detection'] | ['miscellaneous'] | [-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] | [-7.243312358856201, 3.7369399070739746] |
3c205f2e-51a8-47d7-9cc5-d49c926beb2a | supervised-pretraining-for-molecular-force | 2211.14429 | null | https://arxiv.org/abs/2211.14429v1 | https://arxiv.org/pdf/2211.14429v1.pdf | Supervised Pretraining for Molecular Force Fields and Properties Prediction | Machine learning approaches have become popular for molecular modeling tasks, including molecular force fields and properties prediction. Traditional supervised learning methods suffer from scarcity of labeled data for particular tasks, motivating the use of large-scale dataset for other relevant tasks. We propose to pretrain neural networks on a dataset of 86 millions of molecules with atom charges and 3D geometries as inputs and molecular energies as labels. Experiments show that, compared to training from scratch, fine-tuning the pretrained model can significantly improve the performance for seven molecular property prediction tasks and two force field tasks. We also demonstrate that the learned representations from the pretrained model contain adequate information about molecular structures, by showing that linear probing of the representations can predict many molecular information including atom types, interatomic distances, class of molecular scaffolds, and existence of molecular fragments. Our results show that supervised pretraining is a promising research direction in molecular modeling | ['Liang Xiang', 'Chong Wang', 'Zhirui Wang', 'Wenzhi Xiao', 'Weihao Gao', 'Xiang Gao'] | 2022-11-23 | null | null | null | null | ['molecular-property-prediction'] | ['miscellaneous'] | [ 3.51630509e-01 -1.01412781e-01 -1.02671254e+00 -7.40995765e-01
-4.85493600e-01 -3.17533791e-01 2.61921376e-01 6.32440209e-01
-4.07897592e-01 1.28163838e+00 8.84424821e-02 -5.51910043e-01
5.84172979e-02 -1.00266743e+00 -1.23715949e+00 -7.65479088e-01
-4.21437979e-01 4.69842464e-01 -6.95921779e-02 9.64921992e-03
4.71913487e-01 8.91019583e-01 -1.39016104e+00 3.51996183e-01
1.13736069e+00 8.65573585e-01 5.34061268e-02 1.67702347e-01
-6.20342009e-02 8.63010347e-01 -3.30394924e-01 -1.97432041e-01
2.65913922e-02 -1.52932152e-01 -9.03773606e-01 -4.05342817e-01
5.86151063e-01 -1.49226829e-01 -3.80555749e-01 8.56087327e-01
3.95799100e-01 4.85138893e-01 1.10701537e+00 -7.97805965e-01
-8.13623667e-01 5.66111505e-01 -3.08065981e-01 -3.16365249e-02
2.22871333e-01 1.30271614e-02 1.02477062e+00 -9.28184569e-01
6.40433490e-01 1.08429611e+00 6.54649138e-01 4.69484210e-01
-1.24352586e+00 -9.11153555e-01 2.30686497e-02 3.81485492e-01
-1.17578685e+00 -1.63795650e-01 6.79053962e-01 -5.16511440e-01
1.38705134e+00 -7.74592310e-02 2.49683484e-01 9.06931341e-01
3.58640283e-01 3.05409342e-01 5.84084809e-01 -3.03525358e-01
3.07877988e-01 -1.25880875e-02 3.21466953e-01 7.37660944e-01
4.31839764e-01 3.08134884e-01 -3.10153961e-01 -4.49536353e-01
7.96724141e-01 3.85478258e-01 -4.97147515e-02 -5.02011359e-01
-1.10193288e+00 1.07547045e+00 9.45118964e-01 5.75318187e-02
-4.27446306e-01 1.99665248e-01 2.96286196e-01 4.91500385e-02
3.94719809e-01 9.42900479e-01 -8.23786259e-01 1.18354455e-01
-3.90529841e-01 2.52036244e-01 5.87708831e-01 7.92514563e-01
1.15663588e+00 9.50433090e-02 2.91037530e-01 6.73245490e-01
-7.24491430e-03 2.08863422e-01 4.37681079e-01 -6.26806438e-01
5.22343814e-01 7.94129550e-01 2.22869795e-02 -7.02221334e-01
-4.98069376e-01 -1.00780919e-01 -9.75706458e-01 2.98481584e-02
1.26491532e-01 -2.58297324e-01 -1.08228838e+00 1.60957408e+00
1.42319635e-01 4.30341214e-02 2.63334475e-02 5.00980914e-01
1.00849545e+00 1.14328051e+00 5.83276927e-01 -4.61323589e-01
6.07002974e-01 -9.11675811e-01 -3.67609560e-01 2.29180269e-02
1.01495993e+00 -3.74524415e-01 5.19184709e-01 3.43744338e-01
-8.63781333e-01 -8.09374034e-01 -9.95891273e-01 -1.32060185e-01
-6.91351831e-01 3.41445990e-02 1.44124675e+00 4.53776598e-01
-3.30257446e-01 1.23461676e+00 -8.05004478e-01 8.03577676e-02
6.87645733e-01 9.80252147e-01 -6.83852494e-01 -1.31539464e-01
-1.07963145e+00 8.17655087e-01 7.81781912e-01 -1.97432742e-01
-7.93983698e-01 -9.33238864e-01 -1.00562167e+00 1.55272290e-01
1.14156350e-01 -4.49096084e-01 8.72688711e-01 -8.29963744e-01
-1.32978499e+00 4.78064537e-01 -1.84667394e-01 -4.86343652e-01
-2.16295347e-01 -2.58404538e-02 -4.37291563e-01 -9.03603435e-02
-2.21854091e-01 9.54356492e-01 4.86593604e-01 -9.64031935e-01
-2.36820042e-01 -3.16504925e-01 -8.84561986e-03 -8.48715380e-02
-4.12263423e-01 -1.83270544e-01 1.71567738e-01 -5.76486766e-01
4.80836853e-02 -6.75242126e-01 -7.03504622e-01 -3.64723921e-01
-4.43790466e-01 -4.31995779e-01 3.34707558e-01 -2.29492396e-01
1.04807818e+00 -1.49168777e+00 3.24510813e-01 4.54608202e-01
1.24083087e-01 3.03505838e-01 -3.49322379e-01 5.48491120e-01
-7.65358746e-01 3.01161855e-01 -1.06788248e-01 2.08629891e-01
-3.44368249e-01 9.58685502e-02 -2.89215773e-01 2.34539852e-01
1.80835336e-01 9.27128375e-01 -6.78586841e-01 -5.02128974e-02
2.57793069e-01 5.74138641e-01 -7.72909462e-01 4.88196045e-01
-5.46202600e-01 5.36567450e-01 -6.36812985e-01 6.32041156e-01
6.63144588e-01 -4.54655260e-01 1.70486301e-01 -2.09639907e-01
4.91096303e-02 5.91332257e-01 -5.73593318e-01 1.59026515e+00
-2.04195172e-01 7.53938556e-02 -6.65692568e-01 -1.16499710e+00
1.03935158e+00 6.31721020e-02 7.01404214e-01 -7.85512269e-01
-1.47713155e-01 1.94267258e-01 3.40782434e-01 -4.24794406e-01
1.25307515e-01 -3.70086402e-01 2.72724718e-01 2.88512677e-01
2.52985358e-01 2.55998522e-02 1.88406393e-01 -1.64528102e-01
6.48416758e-01 8.10644850e-02 3.07137191e-01 3.33164521e-02
5.50194621e-01 6.25701621e-02 5.27500987e-01 5.03686368e-01
3.78923446e-01 1.96937859e-01 2.87661731e-01 -1.18191421e+00
-1.11764061e+00 -5.43257833e-01 -3.79028529e-01 1.38793194e+00
-5.70934899e-02 -6.88033700e-01 -5.42535484e-01 -7.29080439e-01
2.38924354e-01 3.46983820e-01 -7.01578021e-01 -4.97372240e-01
-5.82331717e-01 -1.10771692e+00 1.02163209e-02 7.15434730e-01
-9.23347026e-02 -1.40468693e+00 3.14048678e-02 3.36069137e-01
5.26804328e-01 -8.21474671e-01 -6.96390718e-02 7.46805370e-01
-1.33184123e+00 -1.26408637e+00 -4.00957733e-01 -9.73995566e-01
8.77783716e-01 1.28244460e-01 1.04135108e+00 1.84306890e-01
-3.91871125e-01 -5.68533897e-01 -5.76389395e-02 -3.68104577e-01
-3.55331987e-01 4.05595511e-01 2.94457346e-01 -2.72884667e-01
3.52230817e-01 -8.05292130e-01 -5.45468211e-01 1.81004152e-01
-6.96317613e-01 5.23948781e-02 5.54268658e-01 8.26345742e-01
8.13056171e-01 3.39994915e-02 8.23502302e-01 -1.33313370e+00
4.82148379e-01 -3.32378864e-01 -5.12538016e-01 2.30101511e-01
-6.33483350e-01 4.00606871e-01 1.08546722e+00 -2.90849268e-01
-7.57300198e-01 2.40965977e-01 -2.40103856e-01 -1.34606957e-01
-2.48652190e-01 8.40623856e-01 -2.98658878e-01 -4.14649606e-01
7.76093900e-01 7.26068616e-02 -3.27454925e-01 -5.88944435e-01
1.97495684e-01 2.29800701e-01 3.00355762e-01 -8.65284026e-01
6.13189876e-01 7.74687976e-02 2.86902696e-01 -6.34103775e-01
-9.34807956e-01 -3.11135054e-01 -9.95907068e-01 5.82889676e-01
5.69171965e-01 -9.12271380e-01 -9.38843787e-01 9.09280851e-02
-1.09155428e+00 -2.64286488e-01 -1.80203654e-02 8.32878947e-01
-4.31361467e-01 3.10016632e-01 -5.50024033e-01 -3.48052472e-01
-3.42417508e-01 -1.22725475e+00 7.01598585e-01 1.96955651e-01
1.71781369e-02 -1.17334974e+00 2.97135085e-01 4.20675695e-01
2.04196632e-01 2.90088475e-01 1.58268452e+00 -7.81617761e-01
-6.10775888e-01 -4.01167363e-01 -6.84644952e-02 1.69965684e-01
5.17887473e-01 -1.24658290e-02 -8.18857968e-01 -2.83890188e-01
-6.94980443e-01 -7.76719868e-01 1.17139781e+00 5.34046173e-01
1.84072673e+00 -3.57521117e-01 -5.32026291e-01 1.00440586e+00
1.08894062e+00 5.49051404e-01 6.56276166e-01 1.34722516e-02
1.02427363e+00 4.12216693e-01 3.96145761e-01 2.55666018e-01
-5.17652743e-02 4.38583940e-01 5.16504705e-01 -2.96387106e-01
3.57391924e-01 -5.21049440e-01 1.58945143e-01 5.32540798e-01
-6.53808057e-01 -1.97599947e-01 -7.87137568e-01 -1.79111570e-01
-1.66543078e+00 -1.00093782e+00 2.05150470e-01 2.45075846e+00
1.11251473e+00 1.28942281e-01 2.35689521e-01 -1.60983518e-01
5.48498333e-01 -1.24019722e-03 -1.11630523e+00 -4.32213992e-01
1.26866087e-01 6.45366728e-01 5.36768794e-01 4.72145677e-01
-1.09946179e+00 1.11656964e+00 7.72370148e+00 7.00084150e-01
-1.32182705e+00 -4.00292307e-01 9.76701736e-01 2.18940794e-01
-3.02312881e-01 -7.89081678e-02 -8.96246254e-01 3.09361786e-01
1.19784522e+00 3.21140923e-02 2.79792666e-01 9.92850065e-01
1.73371375e-01 2.99287379e-01 -1.36753798e+00 8.72083664e-01
-2.08149701e-01 -1.91211045e+00 6.33261681e-01 2.68860668e-01
9.47758555e-01 -6.93742111e-02 -1.14474641e-02 3.53794873e-01
8.31955373e-02 -1.80756557e+00 -8.65085348e-02 3.71695697e-01
8.25439811e-01 -1.05739212e+00 6.95352852e-01 3.33260864e-01
-8.06321919e-01 2.96143800e-01 -8.55554283e-01 -2.59245068e-01
-2.81749725e-01 2.74703026e-01 -9.90833819e-01 3.82801265e-01
1.26932338e-01 1.15637267e+00 -6.03934705e-01 1.10936546e+00
-2.94515211e-02 7.53962338e-01 1.56717468e-02 -6.80712610e-02
3.54382284e-02 -4.75805998e-01 -3.05138201e-01 8.84532988e-01
-9.24397185e-02 6.14547133e-02 3.90207469e-01 7.99837172e-01
-5.90981424e-01 2.81391621e-01 -5.83548963e-01 -5.28346181e-01
2.85275996e-01 8.74543786e-01 -3.91896397e-01 -3.18625391e-01
-2.85887241e-01 5.18957436e-01 5.15504837e-01 4.65141565e-01
-8.18457603e-01 -3.94336909e-01 6.50454044e-01 8.00056756e-02
1.85613483e-01 -2.37572402e-01 1.71219613e-02 -1.07949495e+00
-4.38421190e-01 -8.79288137e-01 2.14512199e-01 -5.71221888e-01
-1.28286779e+00 4.41563547e-01 -2.34546617e-01 -8.92230511e-01
-1.92404538e-01 -1.20583689e+00 -6.95494056e-01 7.52562463e-01
-1.63970673e+00 -8.86453032e-01 -2.17109788e-02 3.56186569e-01
2.41832301e-01 -4.90180969e-01 1.19793344e+00 2.67217219e-01
-8.38729322e-01 5.10822654e-01 3.73319656e-01 -4.43339124e-02
8.04225802e-01 -1.18554378e+00 3.53429168e-01 1.24054439e-02
1.74993381e-01 9.38494027e-01 2.85409510e-01 -5.86219907e-01
-1.52209508e+00 -1.19577682e+00 6.31680906e-01 -1.83821544e-01
3.76823217e-01 -2.89869875e-01 -1.09816694e+00 6.05177522e-01
-2.66041517e-01 2.83543319e-01 1.17577291e+00 4.01433885e-01
-5.19860148e-01 4.19660993e-02 -9.04104710e-01 5.60537092e-02
1.06047201e+00 -4.90184486e-01 -1.81431592e-01 9.37517583e-01
7.56866097e-01 -2.83133477e-01 -1.18254340e+00 6.85160279e-01
4.11663055e-01 -6.68742597e-01 1.38933766e+00 -1.71418655e+00
6.31893516e-01 3.39853466e-02 2.72231176e-02 -1.06003368e+00
-4.87360567e-01 -4.02095824e-01 -3.63151908e-01 5.64531326e-01
8.18194807e-01 -4.47132081e-01 1.22004592e+00 7.10201204e-01
-1.27444893e-01 -9.47909236e-01 -5.05940437e-01 -4.96138483e-01
5.64006805e-01 -4.63405736e-02 6.68038607e-01 1.06072187e+00
1.14439890e-01 7.88253725e-01 -5.07630169e-01 -2.17655413e-02
2.53475904e-01 5.69431007e-01 6.81660235e-01 -1.56217670e+00
-2.32637361e-01 -1.48658812e-01 -4.38985318e-01 -1.05963361e+00
5.62486470e-01 -1.13782668e+00 -4.75817740e-01 -1.32883286e+00
4.89328146e-01 -5.03004253e-01 -4.85319853e-01 7.49796331e-01
-1.38567314e-01 6.08186871e-02 -3.57063890e-01 1.55602187e-01
-5.75312555e-01 7.66231656e-01 1.46227086e+00 -4.53253925e-01
-2.90844887e-01 9.82891396e-02 -5.92106462e-01 6.46681309e-01
6.51941776e-01 -4.43556964e-01 -3.82898271e-01 -5.38998842e-02
2.89574027e-01 -1.77561883e-02 -5.78263998e-02 -1.01649892e+00
-6.27218708e-02 -7.59553611e-01 9.10200894e-01 -4.80688125e-01
3.64298820e-01 -6.67160988e-01 -1.06940776e-01 5.19014955e-01
-6.26921773e-01 -2.90934920e-01 1.66835144e-01 5.70766211e-01
-2.41535485e-01 -3.14282835e-01 6.98002994e-01 -1.36198685e-01
-5.39551735e-01 9.45555449e-01 -8.53073299e-02 -4.32155997e-01
8.51015270e-01 -3.28684986e-01 -2.07154408e-01 -1.06878892e-01
-8.20128024e-01 9.91169587e-02 4.24743652e-01 2.44642347e-01
5.96724033e-01 -1.09990489e+00 -2.51136690e-01 3.42112064e-01
1.16077296e-01 1.37819216e-01 -8.14453932e-04 1.09382942e-01
-6.82625353e-01 7.76714683e-01 -4.57342952e-01 -2.08214432e-01
-1.15299249e+00 8.94856036e-01 3.91741246e-01 -2.73579359e-01
-1.06669016e-01 7.46158540e-01 4.61927265e-01 -6.72339082e-01
3.65311712e-01 -4.64876831e-01 -1.55365184e-01 -2.80212402e-01
5.14491439e-01 6.58395737e-02 1.93906933e-01 -4.57297504e-01
-2.44334921e-01 5.23701966e-01 -5.94319582e-01 9.27721262e-01
1.95400143e+00 6.94614708e-01 1.27621880e-02 2.34728050e-03
1.48508441e+00 -1.47187367e-01 -1.15744400e+00 -1.15361311e-01
-5.28593548e-03 -1.33980080e-01 -1.30442500e-01 -7.49799728e-01
-8.69845271e-01 1.15322828e+00 4.30017322e-01 -4.48661596e-01
6.28251553e-01 -1.78070620e-01 6.44311965e-01 1.23691297e+00
4.19582844e-01 -7.42038846e-01 1.45598158e-01 6.61654830e-01
6.26441360e-01 -1.56953192e+00 3.91544729e-01 -7.05421090e-01
-3.20379525e-01 1.43377757e+00 7.76869953e-01 -5.75681441e-02
4.82527584e-01 -1.75658986e-01 -3.80528063e-01 -3.03620934e-01
-6.56526804e-01 2.12481581e-02 5.03426611e-01 5.71693480e-01
9.75005269e-01 -3.14287245e-02 1.30241379e-01 4.29528832e-01
-1.11676700e-01 5.21537364e-02 1.26691088e-01 9.57713604e-01
-7.36934006e-01 -1.66852319e+00 -3.19010094e-02 5.27833045e-01
-3.24952960e-01 -2.82669753e-01 -6.67122304e-01 4.93726104e-01
2.46070862e-01 3.50269109e-01 -8.10243860e-02 -3.36590230e-01
5.78617193e-02 1.23505615e-01 7.71719337e-01 -9.87084746e-01
-4.42690462e-01 -3.84350806e-01 -6.79520052e-03 -2.45517880e-01
-5.88261366e-01 9.24597755e-02 -1.43814898e+00 -2.59122401e-01
-5.11255383e-01 6.76780939e-01 4.63616788e-01 7.86625922e-01
4.96486157e-01 2.71350652e-01 8.14178646e-01 -1.03677046e+00
-4.34167534e-01 -9.36539888e-01 -3.33544105e-01 6.31523311e-01
2.67995715e-01 -7.22983718e-01 -2.58561806e-03 1.22333720e-01] | [5.138836860656738, 5.760371208190918] |
2486e7cf-3eac-4fe6-9d1f-52ec878657a1 | namerec-highly-accurate-and-fine-grained | 2103.11360 | null | https://arxiv.org/abs/2103.11360v2 | https://arxiv.org/pdf/2103.11360v2.pdf | NameRec*: Highly Accurate and Fine-grained Person Name Recognition | In this paper, we introduce the NameRec* task, which aims to do highly accurate and fine-grained person name recognition. Traditional Named Entity Recognition models have good performance in recognising well-formed person names from text with consistent and complete syntax, such as news articles. However, there are rapidly growing scenarios where sentences are of incomplete syntax and names are in various forms such as user-generated contents and academic homepages. To address person name recognition in this context, we propose a fine-grained annotation scheme based on anthroponymy. To take full advantage of the fine-grained annotations, we propose a Co-guided Neural Network (CogNN) for person name recognition. CogNN fully explores the intra-sentence context and rich training signals of name forms. To better utilize the inter-sentence context and implicit relations, which are extremely essential for recognizing person names in long documents, we further propose an Inter-sentence BERT Model (IsBERT). IsBERT has an overlapped input processor, and an inter-sentence encoder with bidirectional overlapped contextual embedding learning and multi-hop inference mechanisms. To derive benefit from different documents with a diverse abundance of context, we propose an advanced Adaptive Inter-sentence BERT Model (Ada-IsBERT) to dynamically adjust the inter-sentence overlapping ratio to different documents. We conduct extensive experiments to demonstrate the superiority of the proposed methods on both academic homepages and news articles. | ['Shijie Liu', 'Yimeng Dai', 'Rui Zhang'] | 2021-03-21 | null | null | null | null | ['implicit-relations'] | ['natural-language-processing'] | [-3.74496043e-01 -3.89606804e-01 -2.56710678e-01 -5.93836725e-01
-4.34383839e-01 -5.12130439e-01 5.95522046e-01 -2.08864599e-01
-7.21024692e-01 7.93516636e-01 6.98802710e-01 -1.22943439e-01
-2.67879248e-01 -8.27095211e-01 -2.73403466e-01 -2.42521107e-01
2.87738681e-01 5.07703424e-01 -6.04640730e-02 -1.82615981e-01
2.88306564e-01 5.87809503e-01 -1.13530588e+00 2.09999189e-01
1.09996188e+00 5.33363104e-01 2.05446348e-01 4.21858728e-01
-8.77308905e-01 5.00246286e-01 -7.08964348e-01 -5.15220344e-01
5.43266945e-02 -2.60685086e-01 -7.99886644e-01 7.76691362e-03
3.84336650e-01 -2.49209851e-01 -5.27969062e-01 1.03332925e+00
7.53972471e-01 2.88976133e-01 4.93414193e-01 -8.55259836e-01
-1.20148218e+00 1.18086088e+00 -4.74341601e-01 5.40584147e-01
5.81076682e-01 1.90548114e-02 1.22056937e+00 -1.16317093e+00
4.58459914e-01 1.51795340e+00 5.17448187e-01 4.81518567e-01
-7.96344757e-01 -8.50655794e-01 5.52837372e-01 2.97390014e-01
-1.93798065e+00 -4.70067173e-01 6.14946783e-01 -2.36269861e-01
1.09779537e+00 1.88795388e-01 1.98609620e-01 1.14691794e+00
-3.27888966e-01 7.23038375e-01 8.41841340e-01 -3.36854190e-01
-2.10228041e-01 2.72521704e-01 6.42619967e-01 5.19216061e-01
3.60086679e-01 -9.63119566e-02 -6.01199090e-01 -1.39334038e-01
7.47762501e-01 2.04649284e-01 -4.22824711e-01 3.88589919e-01
-1.41499639e+00 6.50538385e-01 5.01996338e-01 6.95177078e-01
-2.94583887e-01 -1.84316128e-01 3.69832695e-01 5.62324002e-02
1.74530372e-01 5.14364839e-01 -6.07553422e-01 -7.74532184e-02
-1.00742269e+00 7.48968273e-02 9.94873047e-01 1.28032696e+00
6.30246520e-01 -3.89128760e-03 -3.99038732e-01 1.18525529e+00
3.69293094e-01 6.53630435e-01 1.00581312e+00 -3.61631691e-01
5.10275006e-01 7.54746616e-01 6.00389726e-02 -8.70631099e-01
-4.60467041e-01 -9.89037871e-01 -7.69996166e-01 -7.52700090e-01
2.22801521e-01 1.06459865e-02 -7.92921484e-01 1.73683035e+00
2.32849598e-01 2.12857604e-01 1.28783077e-01 8.46323073e-01
1.20045924e+00 6.88997805e-01 2.99847752e-01 -1.24772638e-01
1.77517462e+00 -1.04771173e+00 -7.80776739e-01 -3.19941014e-01
5.60639024e-01 -5.85195184e-01 9.27702248e-01 -2.89811760e-01
-5.04453659e-01 -6.56271219e-01 -7.59300828e-01 -4.23245162e-01
-7.41112173e-01 4.00718331e-01 3.64272952e-01 6.17950201e-01
-6.20531082e-01 2.34949395e-01 -4.20413673e-01 -5.07508397e-01
1.90897033e-01 3.61885309e-01 -3.12864035e-01 -1.94395170e-01
-1.72728872e+00 6.46193743e-01 7.96435893e-01 1.80725038e-01
-1.15340345e-01 -5.11121869e-01 -9.64243412e-01 3.98271561e-01
3.66590738e-01 -8.45650852e-01 9.48635101e-01 -7.44250059e-01
-1.13711393e+00 7.49354422e-01 -3.23214591e-01 2.20561735e-02
9.74583626e-02 -3.82593364e-01 -8.52248907e-01 -7.80212432e-02
3.79157424e-01 4.35863227e-01 3.36587906e-01 -8.12979639e-01
-5.48323810e-01 -5.18932939e-01 1.70819368e-02 3.67307812e-01
-6.15649343e-01 4.78463858e-01 -5.10301173e-01 -8.42503726e-01
6.01056628e-02 -5.82896948e-01 2.14874536e-01 -3.54213238e-01
-6.53812587e-01 -7.19603419e-01 3.03517371e-01 -7.40089178e-01
1.62010968e+00 -2.17379451e+00 -1.73246786e-01 1.30123407e-01
9.55389589e-02 2.96005517e-01 -1.31817311e-01 4.38559890e-01
1.76046088e-01 1.48771137e-01 -1.02604188e-01 -2.50246584e-01
2.64311999e-01 3.40268433e-01 -3.00422043e-01 1.02982321e-03
6.63914084e-02 9.00676489e-01 -9.41714704e-01 -7.38001227e-01
-2.20744550e-01 4.20988828e-01 -2.28468359e-01 3.53051394e-01
1.42915875e-01 1.89951435e-01 -8.70897472e-01 8.05470169e-01
5.52944243e-01 -4.41655099e-01 -2.30565947e-02 -1.40813932e-01
-2.09774047e-01 2.77718991e-01 -1.40913153e+00 1.41732550e+00
-4.24139112e-01 8.28271210e-02 -2.23315787e-02 -7.21879244e-01
1.09021890e+00 2.64134496e-01 5.33452909e-03 -6.99606240e-01
3.41146216e-02 4.85818446e-01 -1.95660070e-01 -6.02173150e-01
6.22537315e-01 -3.95156629e-02 -1.43392429e-01 4.05358315e-01
-1.41109424e-02 6.34615839e-01 3.80403310e-01 2.18038067e-01
1.05075514e+00 -2.91927904e-02 4.94327307e-01 -1.74684808e-01
9.45662200e-01 -3.66013467e-01 8.30486774e-01 8.69301856e-01
-3.46143842e-01 4.03619140e-01 5.62198423e-02 -3.75331044e-01
-7.44242072e-01 -9.23393846e-01 -3.13621730e-01 1.55448246e+00
2.60080844e-01 -3.71217668e-01 -4.02591467e-01 -7.43643582e-01
1.34515956e-01 8.09038997e-01 -2.59910494e-01 5.58274090e-02
-7.28557110e-01 -5.54285526e-01 6.88673198e-01 7.36344874e-01
7.30782330e-01 -1.37248850e+00 2.39809856e-01 4.34387207e-01
-1.75714001e-01 -1.14590800e+00 -1.05494440e+00 -7.91930780e-02
-4.01501626e-01 -9.20672655e-01 -9.90315020e-01 -1.06912673e+00
5.42614043e-01 8.52639675e-02 1.10390389e+00 -6.82575777e-02
7.75383487e-02 3.65022540e-01 -4.67457473e-01 -9.11136419e-02
1.55524880e-01 5.31408370e-01 1.86086059e-01 2.47027144e-01
9.27272975e-01 -5.40713251e-01 -4.71706778e-01 2.45156005e-01
-8.72671902e-01 -1.92958906e-01 8.95217896e-01 1.10804594e+00
2.83504158e-01 -1.78647220e-01 9.21793699e-01 -1.13588774e+00
7.55480587e-01 -7.56599903e-01 -2.16086775e-01 5.08344114e-01
-8.82075489e-01 2.15201378e-01 7.62530744e-01 -4.39860672e-01
-1.25434721e+00 -2.26909235e-01 -4.19755995e-01 -2.58370154e-02
-3.44982535e-01 7.61743724e-01 -5.04525006e-01 3.16721171e-01
3.50931019e-01 6.31445110e-01 -5.61975241e-01 -8.34509134e-01
4.81915355e-01 1.23684847e+00 6.92394555e-01 -7.63463974e-01
6.82972848e-01 -9.81938243e-02 -7.60710895e-01 -7.32430279e-01
-1.02330410e+00 -8.26712132e-01 -6.23371303e-01 2.21443474e-01
7.46725738e-01 -9.55574095e-01 -5.42848110e-01 4.23453122e-01
-1.45260191e+00 3.34203720e-01 3.65311606e-03 6.13388300e-01
1.18917331e-01 4.35291946e-01 -8.36981952e-01 -7.12832570e-01
-5.45786262e-01 -8.27382088e-01 9.44144309e-01 6.95812285e-01
-1.65665820e-01 -1.08404076e+00 -9.92929563e-02 3.65836680e-01
5.27236402e-01 -3.79164100e-01 7.82641351e-01 -1.52505243e+00
-4.60609972e-01 -7.41798058e-02 -5.88893771e-01 4.82963100e-02
1.65038332e-01 -4.94591147e-01 -7.80113757e-01 -1.44030377e-01
-2.98653215e-01 -1.03574149e-01 8.31002057e-01 -1.16442084e-01
9.96327460e-01 -5.01782477e-01 -4.88211691e-01 6.84025705e-01
1.21589231e+00 1.09987892e-01 2.02965990e-01 5.70192516e-01
1.01734984e+00 5.45392215e-01 2.64891416e-01 4.73565578e-01
6.86297774e-01 6.28658950e-01 -2.69592971e-01 2.95118362e-01
-1.82311505e-01 -2.89744884e-01 3.13366175e-01 1.17534852e+00
2.37110540e-01 -3.43921304e-01 -7.97031283e-01 6.33839428e-01
-1.62051582e+00 -1.09334064e+00 1.46060362e-01 1.73264253e+00
1.23946762e+00 4.54215333e-02 -1.36584252e-01 -3.08823049e-01
1.21445775e+00 2.60848939e-01 -5.89268148e-01 -1.38525546e-01
-3.59768033e-01 -2.13306267e-02 2.92606264e-01 2.77953327e-01
-1.02718735e+00 9.00544047e-01 5.38952875e+00 9.63464797e-01
-9.51813579e-01 3.57877575e-02 2.55479932e-01 2.49730498e-01
-4.97455835e-01 -1.35877043e-01 -1.42511523e+00 9.62158322e-01
8.29384327e-01 -3.08626115e-01 2.62861758e-01 9.44430709e-01
-8.71112347e-02 4.89649087e-01 -1.18819118e+00 1.18242800e+00
2.23043755e-01 -1.04770970e+00 1.44254997e-01 -1.21003889e-01
5.54068267e-01 -7.68172145e-02 -3.11638594e-01 6.84876144e-01
1.01425409e-01 -8.03522289e-01 5.05055010e-01 6.90031171e-01
8.18726122e-01 -5.78797042e-01 9.99641657e-01 4.41376925e-01
-1.35968471e+00 -2.51244813e-01 -2.74000525e-01 7.10753426e-02
1.88549012e-01 7.11564302e-01 -5.91241598e-01 6.47194803e-01
4.38338846e-01 8.19635451e-01 -6.19126201e-01 1.03315294e+00
-3.25326443e-01 5.12703538e-01 -2.84972042e-01 -3.81817222e-01
1.07604429e-01 -1.68425292e-02 4.09073293e-01 1.67557371e+00
2.71547019e-01 3.56242180e-01 3.86711329e-01 9.83844459e-01
-3.38878214e-01 4.24312979e-01 -3.44233125e-01 -2.76311666e-01
1.15410376e+00 1.07048512e+00 -4.58176076e-01 -5.03300488e-01
-7.88066328e-01 1.13277471e+00 4.71834153e-01 5.84529817e-01
-4.17235553e-01 -8.61546636e-01 4.69717115e-01 -2.40249205e-02
4.13914025e-01 -1.63576573e-01 -1.87340155e-01 -1.40135527e+00
1.82217598e-01 -5.07382631e-01 7.03305960e-01 -6.07105911e-01
-1.88103330e+00 6.10045433e-01 -2.08256856e-01 -8.65965784e-01
7.54907951e-02 -5.77804863e-01 -5.95923245e-01 1.02092087e+00
-1.66483915e+00 -1.44538593e+00 -1.81056678e-01 5.10642111e-01
5.85400760e-01 -3.89867395e-01 7.93083549e-01 6.59164131e-01
-9.75469291e-01 1.07492149e+00 2.41778761e-01 8.50857377e-01
9.62626398e-01 -1.22371018e+00 2.44047627e-01 8.79278004e-01
2.36779705e-01 1.25580907e+00 2.18649998e-01 -7.41532922e-01
-1.17271054e+00 -1.22005427e+00 1.59473920e+00 -3.37279677e-01
5.77578962e-01 -8.42982605e-02 -1.05592096e+00 7.79550076e-01
1.58912048e-01 5.38595952e-02 8.72304857e-01 5.39982855e-01
-5.83139718e-01 -2.85712779e-01 -9.07969773e-01 4.70810801e-01
1.26626647e+00 -8.47663462e-01 -1.27284265e+00 3.64715517e-01
9.60708082e-01 -2.49443665e-01 -8.53120863e-01 1.82412431e-01
3.70253146e-01 -5.78403294e-01 1.00239420e+00 -5.33218384e-01
1.52963012e-01 -3.40994567e-01 -1.90689042e-02 -1.13003123e+00
-6.52189195e-01 -2.28885621e-01 -1.82461858e-01 1.85720420e+00
2.38502070e-01 -8.53519380e-01 4.55096066e-01 7.75441170e-01
-3.24438691e-01 -5.88762105e-01 -9.14307714e-01 -8.37835550e-01
-1.50829583e-01 -1.14497416e-01 9.15645123e-01 1.25723994e+00
-2.13738512e-02 7.04905987e-01 -1.38584837e-01 1.72154173e-01
3.68341684e-01 2.05428988e-01 2.63811886e-01 -1.12069035e+00
-1.84533596e-01 -4.82265770e-01 -3.35340798e-01 -1.31021166e+00
5.24164975e-01 -1.07186329e+00 -7.90909156e-02 -1.51163185e+00
4.93359447e-01 -5.43052912e-01 -7.53265560e-01 6.44207597e-01
-6.49785578e-01 -1.11399554e-01 -8.31864476e-02 4.87237424e-01
-8.62442970e-01 5.78104496e-01 9.59953904e-01 -2.28634447e-01
-1.61444530e-01 -1.90601289e-01 -9.04330373e-01 3.31191897e-01
4.04868871e-01 -4.06569213e-01 -8.60252902e-02 -6.76561177e-01
1.22628555e-01 -1.20396219e-01 5.66515215e-02 -7.76179671e-01
6.12624824e-01 -1.41961485e-01 5.24343550e-01 -4.53001767e-01
-6.13303334e-02 -6.28922880e-01 -2.34200329e-01 1.58814564e-01
-5.40195942e-01 1.42145649e-01 -1.48958474e-01 5.35863400e-01
-4.74519432e-01 -3.08680773e-01 3.36332232e-01 -3.82237911e-01
-9.97801244e-01 3.97232056e-01 -2.51139514e-02 3.67906958e-01
5.43963373e-01 -1.73408940e-01 -4.05549526e-01 2.05572341e-02
-6.05514467e-01 5.27417600e-01 2.28316888e-01 5.11076450e-01
5.99082172e-01 -1.66560328e+00 -8.20986688e-01 3.10648203e-01
4.14276451e-01 -1.23187512e-01 3.53739381e-01 5.17573416e-01
-2.94194877e-01 5.81329882e-01 5.14861606e-02 -1.45931169e-01
-9.04071927e-01 5.45860052e-01 1.20324969e-01 -2.03610137e-01
-4.32618618e-01 1.03716385e+00 2.65317053e-01 -8.31250846e-01
1.31860539e-01 -6.26057154e-04 -5.77704966e-01 8.63249600e-02
8.19059134e-01 2.17294678e-01 -1.74190119e-01 -8.23959291e-01
-5.20967841e-01 4.49013382e-01 -4.59556520e-01 1.14006452e-01
1.19276845e+00 -3.59995365e-01 -4.25849438e-01 3.49127591e-01
1.10566449e+00 3.64973783e-01 -6.54231489e-01 -8.59984815e-01
3.32126915e-01 -3.78198892e-01 -9.00498480e-02 -8.44982862e-01
-7.51547873e-01 5.10604918e-01 3.55073810e-01 -2.90883947e-02
8.72813642e-01 1.32220015e-01 1.21903706e+00 6.54659808e-01
2.04773635e-01 -1.23190236e+00 -1.74171045e-01 8.65105689e-01
6.62885487e-01 -1.10578752e+00 -9.31635275e-02 -3.07939589e-01
-4.51221704e-01 1.20758915e+00 8.47827733e-01 3.47764850e-01
5.71346581e-01 -7.19106421e-02 1.82232976e-01 -6.79054111e-02
-3.34343076e-01 -2.99377829e-01 3.84800196e-01 2.98160851e-01
5.18908501e-01 7.63396546e-02 -6.04703128e-01 1.25247002e+00
-2.60903418e-01 -1.92483157e-01 2.85216153e-01 7.69143224e-01
-5.77453852e-01 -1.16435790e+00 -4.74905580e-01 5.71594357e-01
-3.76488686e-01 -4.64772165e-01 -2.64213085e-01 4.30482656e-01
3.51136386e-01 7.63828814e-01 -7.28545710e-02 -2.29148433e-01
2.53407270e-01 5.38227320e-01 -5.00268722e-03 -9.09586132e-01
-5.63070655e-01 -2.72166342e-01 1.17282137e-01 -2.34853998e-01
-2.02134401e-01 -4.64708865e-01 -1.34465170e+00 -3.26449186e-01
-3.71751904e-01 4.07869667e-01 3.93286407e-01 1.22860909e+00
5.45329750e-01 5.20740330e-01 5.93683302e-01 -2.31188983e-01
-7.85598099e-01 -1.20935249e+00 -8.88383210e-01 5.71248651e-01
1.27902567e-01 -5.80560207e-01 -2.91898727e-01 -1.53591052e-01] | [9.660387992858887, 9.281572341918945] |
b91dcc7d-64d3-4b1d-9806-8f474dbbde5e | stitching-stabilizer-two-frame-stitching | 1603.06678 | null | http://arxiv.org/abs/1603.06678v1 | http://arxiv.org/pdf/1603.06678v1.pdf | Stitching Stabilizer: Two-frame-stitching Video Stabilization for Embedded Systems | In conventional electronic video stabilization, the stabilized frame is
obtained by cropping the input frame to cancel camera shake. While a small
cropping size results in strong stabilization, it does not provide us
satisfactory results from the viewpoint of image quality, because it narrows
the angle of view. By fusing several frames, we can effectively expand the area
of input frames, and achieve strong stabilization even with a large cropping
size. Several methods for doing so have been studied. However, their
computational costs are too high for embedded systems such as smartphones.
We propose a simple, yet surprisingly effective algorithm, called the
stitching stabilizer. It stitches only two frames together with a minimal
computational cost. It can achieve real-time processes in embedded systems, for
Full HD and 30 FPS videos. To clearly show the effect, we apply it to
hyperlapse. Using several clips, we show it produces more strongly stabilized
and natural results than the existing solutions from Microsoft and Instagram. | ['Masaki Satoh'] | 2016-03-22 | null | null | null | null | ['video-stabilization'] | ['computer-vision'] | [ 1.83274135e-01 -1.78778440e-01 6.04838245e-02 5.53257018e-02
-4.68879431e-01 -6.40494585e-01 2.65015841e-01 -1.36839435e-01
-3.85800213e-01 7.11996436e-01 6.55979663e-02 -2.68889844e-01
5.79966605e-01 -4.09209162e-01 -7.60766566e-01 -7.65224516e-01
1.33340761e-01 -5.70321083e-01 8.10144603e-01 -3.09847474e-01
2.80530274e-01 2.72579283e-01 -1.62532055e+00 1.63948283e-01
7.21664071e-01 7.68517435e-01 3.21227193e-01 6.94715679e-01
2.08023563e-01 7.94229746e-01 -6.98731303e-01 -3.53346527e-01
2.93308020e-01 -4.91541713e-01 -4.02812839e-01 5.61297774e-01
5.77532291e-01 -5.75702071e-01 -1.88195884e-01 1.04990828e+00
3.55893075e-01 -1.81881115e-01 -4.22165506e-02 -1.12722600e+00
-1.64752007e-01 3.83084536e-01 -1.09216750e+00 1.86936438e-01
8.61159325e-01 2.55737733e-02 5.09555936e-01 -5.91658771e-01
7.32280850e-01 1.17720509e+00 6.26381040e-01 4.66685057e-01
-1.24905014e+00 -5.54598570e-01 -1.34649590e-01 -1.60525501e-01
-1.23822677e+00 -7.28075385e-01 4.47547138e-01 -5.85989235e-03
4.55606461e-01 4.59964544e-01 6.15636230e-01 6.70689762e-01
2.88161248e-01 5.21964788e-01 9.60930765e-01 -6.43421173e-01
-3.45995389e-02 8.20236932e-03 3.92367458e-03 6.06615365e-01
2.66440898e-01 -1.96422502e-01 -5.38508654e-01 -1.14181779e-01
1.09507799e+00 6.48056045e-02 -8.58851254e-01 -2.91120827e-01
-1.52002370e+00 4.11112040e-01 9.80685726e-02 3.07596117e-01
-4.89347875e-02 2.52259791e-01 2.34577894e-01 4.73016560e-01
3.88397455e-01 1.20764494e-01 -1.28335997e-01 -2.30762064e-01
-1.17690730e+00 1.12257898e-01 5.80009341e-01 9.61971402e-01
6.46307647e-01 1.17261946e-01 2.38066986e-01 4.99525100e-01
6.99341949e-03 6.11843407e-01 2.70710826e-01 -1.38278592e+00
3.81376177e-01 1.41035199e-01 4.01177734e-01 -1.14695883e+00
-2.04625905e-01 3.52215916e-01 -6.09321117e-01 6.49342299e-01
7.13437080e-01 -5.05661249e-01 -4.94705081e-01 1.52415085e+00
3.43061030e-01 2.14279965e-01 -5.26295379e-02 9.53530133e-01
4.62129503e-01 8.04662228e-01 -3.62188339e-01 -7.25871503e-01
1.41783583e+00 -9.69183505e-01 -1.10668993e+00 -3.50897312e-02
3.24232668e-01 -1.19269764e+00 1.10261643e+00 6.78029180e-01
-1.44075394e+00 -5.88713229e-01 -1.34301531e+00 1.32464860e-02
2.72223711e-01 1.26919240e-01 1.81363583e-01 8.05724800e-01
-1.33263659e+00 6.28406882e-01 -9.99926627e-01 -3.59927148e-01
-2.03125447e-01 4.70160276e-01 -5.83754003e-01 2.45823935e-01
-8.60341430e-01 8.32506955e-01 7.37937354e-03 -3.46677810e-01
-2.64805526e-01 -4.06898081e-01 -7.60152638e-01 2.12053925e-01
5.25228977e-01 -4.12628174e-01 1.26231742e+00 -1.26693976e+00
-1.84357285e+00 7.20779657e-01 -4.90141809e-01 -3.99694443e-01
5.93552530e-01 -6.10640466e-01 -2.24783763e-01 5.84906280e-01
-1.44600317e-01 8.04860175e-01 1.28632939e+00 -1.18107629e+00
-5.95539033e-01 7.38395378e-03 1.54701501e-01 1.28770262e-01
-5.25411367e-01 3.52114141e-01 -6.56844556e-01 -6.66022718e-01
8.87666494e-02 -1.10332596e+00 -2.37832069e-01 5.75513653e-02
-1.21315643e-01 4.96654421e-01 1.19900525e+00 -4.92918491e-01
1.42578125e+00 -2.40379786e+00 8.95583779e-02 -1.71890035e-01
2.43784294e-01 4.64398652e-01 3.07635456e-01 2.63353199e-01
-1.48013875e-01 -6.19415306e-02 1.54270902e-01 -2.54912138e-01
-5.72977126e-01 -3.28332484e-02 -3.56617063e-01 6.15312099e-01
-1.76672608e-01 3.80452603e-01 -6.17740750e-01 -5.79599380e-01
4.42824751e-01 4.91858482e-01 -6.74882233e-01 -1.73152133e-03
2.39779279e-01 2.63675958e-01 -1.08922064e-01 5.55725574e-01
9.18079853e-01 -2.87991524e-01 3.89523447e-01 -3.03599775e-01
-4.19085771e-01 -2.83150584e-01 -1.38178778e+00 1.41735268e+00
-1.26664191e-01 1.04161513e+00 3.40413690e-01 -4.39754456e-01
7.85602272e-01 4.18175191e-01 5.33469379e-01 -4.03065801e-01
2.62964696e-01 1.46709144e-01 -3.55011672e-01 -4.54153568e-01
8.50681424e-01 -1.04857907e-01 2.04987019e-01 2.74762362e-01
-4.00233030e-01 -2.70220488e-01 3.60328913e-01 3.70455235e-01
1.05175483e+00 1.12208284e-01 3.52214545e-01 -3.78825754e-01
5.70150018e-01 -3.06868762e-01 3.32264692e-01 2.24459976e-01
-3.31675351e-01 1.05070031e+00 6.52453959e-01 -4.36615229e-01
-1.22108722e+00 -8.22755098e-01 1.38548076e-01 7.34874427e-01
7.28395820e-01 -9.49083447e-01 -1.14564383e+00 -2.08087713e-01
-2.15333283e-01 4.06243727e-02 -1.62021726e-01 1.49630293e-01
-7.19106138e-01 -3.87846082e-01 3.39592874e-01 1.96264401e-01
5.44762015e-01 -7.62203276e-01 -1.21759486e+00 1.33064747e-01
-4.26870644e-01 -1.34113312e+00 -8.40720057e-01 -1.73188165e-01
-9.25748110e-01 -1.12110996e+00 -1.02275670e+00 -7.62605011e-01
6.33111238e-01 1.00448215e+00 9.74095464e-01 2.95949042e-01
5.21502569e-02 1.76155135e-01 -4.32372928e-01 2.73947921e-02
-3.62202227e-01 -4.50308114e-01 3.51945996e-01 -3.26578468e-02
-1.25864044e-01 -4.25750434e-01 -6.18392289e-01 6.32255912e-01
-9.89911616e-01 2.29484335e-01 7.68516660e-02 5.97376347e-01
2.60255814e-01 7.02874660e-02 1.58660144e-01 -5.05691469e-01
3.81967098e-01 1.89890221e-01 -7.36367047e-01 8.62149987e-03
-2.15852484e-01 -1.62301347e-01 7.53398418e-01 -5.71126640e-01
-1.01604772e+00 1.67070016e-01 1.22737490e-01 -5.19132495e-01
1.36896625e-01 -4.82180007e-02 2.72378981e-01 -2.56737381e-01
7.00517178e-01 -1.29610613e-01 4.41966742e-01 -2.62560695e-01
2.04633102e-01 5.33480048e-01 7.73820162e-01 -1.64496064e-01
7.29279935e-01 6.42046630e-01 -2.50406712e-01 -1.10921049e+00
-2.85136491e-01 -2.26712152e-01 -2.64500916e-01 -3.70228499e-01
6.80754423e-01 -1.07169199e+00 -9.70284879e-01 6.13156497e-01
-1.08045363e+00 -1.21109255e-01 1.43940933e-03 5.15128374e-01
-5.51618934e-01 8.86117101e-01 -7.21047401e-01 -4.34595406e-01
-4.73197326e-02 -1.31739008e+00 1.00424182e+00 4.57126558e-01
-3.14202398e-01 -5.80013692e-01 -7.85554480e-03 -3.24230380e-02
3.03219795e-01 3.64316642e-01 1.63114086e-01 4.18903440e-01
-5.05129576e-01 -1.01484801e-03 4.69083665e-03 3.04302990e-01
3.65568608e-01 6.35227025e-01 -7.30405927e-01 -4.96829063e-01
1.45495310e-01 -9.73593220e-02 5.39044261e-01 6.29072964e-01
7.59186566e-01 -9.06732604e-02 -3.10496271e-01 5.25001109e-01
1.49247825e+00 3.56927097e-01 9.91793692e-01 4.64543909e-01
3.33080769e-01 2.97699511e-01 8.80768299e-01 5.43761790e-01
2.83655915e-02 8.03207159e-01 4.01544064e-01 -4.28327292e-01
-3.56116779e-02 2.37723097e-01 8.72537971e-01 6.75729215e-01
-3.77660185e-01 -3.43177229e-01 -5.45653820e-01 1.91207811e-01
-1.79455602e+00 -1.10774362e+00 -4.03341949e-01 2.30917573e+00
7.99447477e-01 1.62856996e-01 2.43200555e-01 3.54477763e-01
1.10302854e+00 3.19200605e-01 -9.32084620e-02 -2.84850627e-01
-1.72422588e-01 -2.86085848e-02 5.34919381e-01 6.59522772e-01
-9.54596400e-01 8.40595245e-01 7.18704462e+00 6.59366548e-01
-1.38860607e+00 -4.85341400e-02 5.68656862e-01 -3.25616956e-01
3.89227606e-02 3.57249491e-02 -6.03856981e-01 6.89164579e-01
8.31961215e-01 -9.14386660e-02 1.10694848e-01 7.59081006e-01
6.32520854e-01 -6.97319925e-01 -7.99519956e-01 1.35050488e+00
3.86395752e-02 -1.37226665e+00 -3.31607461e-01 -5.43719232e-02
5.64978242e-01 -4.31708157e-01 -5.69968559e-02 -2.99038202e-01
-1.91380799e-01 -5.43872893e-01 9.34525371e-01 7.66328871e-02
8.46264601e-01 -7.33374000e-01 5.43738902e-01 8.80750492e-02
-1.21476877e+00 1.41582370e-01 -3.84016067e-01 -2.65720397e-01
5.49591303e-01 4.23240334e-01 -1.64708033e-01 2.10190833e-01
8.11785340e-01 4.23605174e-01 -5.97319007e-01 8.40126395e-01
-4.17029373e-02 3.00832599e-01 -4.99620944e-01 2.32438207e-01
2.57440582e-02 -3.65048707e-01 4.48747545e-01 1.00307965e+00
7.79821932e-01 4.28083360e-01 -1.26295254e-01 2.22838983e-01
2.04636559e-01 -1.00627840e-01 -6.39456749e-01 3.76270771e-01
2.85473466e-01 1.22528708e+00 -1.02766764e+00 -6.11277759e-01
-5.87640643e-01 1.20519280e+00 -3.03578526e-01 1.68973789e-01
-1.29951799e+00 -4.51497823e-01 5.01512289e-01 3.74285281e-01
3.34732771e-01 -4.63479102e-01 -7.22944215e-02 -1.63952303e+00
8.71674940e-02 -1.19274807e+00 -6.67108968e-03 -1.02010369e+00
-3.46283048e-01 6.78673744e-01 -1.66201919e-01 -1.63617396e+00
-1.71434492e-01 -3.52418840e-01 -4.68184233e-01 3.90201509e-01
-1.06191969e+00 -5.52379608e-01 -5.39171517e-01 8.97473991e-01
6.33340180e-01 2.11523384e-01 3.77745718e-01 4.56624448e-01
-4.46200013e-01 3.32586884e-01 2.06721261e-01 -4.35779661e-01
1.31378567e+00 -9.99985099e-01 2.48622045e-01 1.16500640e+00
-7.11286515e-02 5.88539362e-01 1.31219149e+00 -2.59107381e-01
-1.46360981e+00 -2.89904058e-01 7.09789038e-01 -2.83858150e-01
4.98152912e-01 -4.44106990e-03 -8.63820195e-01 5.08213580e-01
7.44956017e-01 8.03749561e-02 2.40680858e-01 -5.31457067e-01
1.26456931e-01 -1.70827314e-01 -9.55656052e-01 8.14261734e-01
5.85422635e-01 -8.95446166e-02 -4.01066810e-01 -1.90466009e-02
8.19133759e-01 -7.44463980e-01 -6.74360454e-01 8.90805200e-02
6.42196596e-01 -1.70423234e+00 9.21999931e-01 1.88376397e-01
5.78032017e-01 -5.73488414e-01 -5.06709851e-02 -9.47614312e-01
-2.56286263e-01 -1.20753336e+00 3.00255902e-02 1.25806165e+00
6.53138533e-02 -4.30533767e-01 7.79751658e-01 6.87792420e-01
1.27535775e-01 -3.06314886e-01 -6.64972484e-01 -6.79767907e-01
-4.99574095e-01 3.09979096e-02 1.96197093e-01 7.24462628e-01
5.53948581e-01 2.44983599e-01 -8.58100355e-01 1.85188442e-01
4.81630296e-01 7.05720782e-02 1.10148406e+00 -6.66619241e-01
-1.95126683e-01 2.83987988e-02 -2.55476087e-01 -1.17360628e+00
-3.53803515e-01 2.00785503e-01 -8.22640285e-02 -7.62522876e-01
5.25800623e-02 -4.21824940e-02 2.22823888e-01 2.68369824e-01
-2.83885688e-01 6.17200673e-01 5.63901424e-01 2.68121123e-01
-4.55433875e-01 6.23402111e-02 1.13410616e+00 2.95006901e-01
-3.29307973e-01 -1.61761537e-01 -4.89417404e-01 9.28121686e-01
6.18050456e-01 -1.60025477e-01 -3.71395200e-01 -3.66452664e-01
9.85578983e-04 5.40081680e-01 1.20771386e-01 -1.21810913e+00
9.70420465e-02 -8.99315476e-02 2.60029763e-01 -4.60870475e-01
3.57651949e-01 -9.83309448e-01 5.06391346e-01 7.96334684e-01
4.32499945e-02 6.05723321e-01 1.51283056e-01 2.44110122e-01
-5.05841553e-01 -2.12521523e-01 1.17347515e+00 -6.12737844e-03
-5.51151216e-01 -6.50651455e-02 -6.01365030e-01 -3.98460299e-01
1.21533847e+00 -5.19457340e-01 -4.13190544e-01 -7.09371924e-01
-5.57869017e-01 -6.99632391e-02 1.20755267e+00 1.66826501e-01
4.78532374e-01 -1.27926934e+00 -2.39179790e-01 2.47748837e-01
-5.36300838e-01 -3.31133425e-01 2.53530294e-01 1.00504792e+00
-1.25050604e+00 5.34517840e-02 -4.06001806e-01 -8.49305987e-01
-1.95841086e+00 8.05593550e-01 -1.33427709e-01 3.76478471e-02
-7.72051156e-01 5.01395702e-01 2.80646712e-01 5.89491189e-01
1.02972105e-01 -5.06242752e-01 -1.57329708e-01 1.85515761e-01
9.13397074e-01 4.04044122e-01 -1.07332833e-01 -5.06171525e-01
-3.79273146e-01 9.22865152e-01 1.36272147e-01 -3.48081321e-01
1.08242607e+00 -7.50789225e-01 -6.33303002e-02 2.49372497e-02
1.02546811e+00 6.47277892e-01 -1.65654492e+00 9.87154394e-02
-4.15841877e-01 -9.01179790e-01 -3.06825548e-01 1.02263577e-01
-1.17476869e+00 7.02683330e-01 4.98530507e-01 6.27631724e-01
1.48108149e+00 -3.93361807e-01 9.40462172e-01 6.99933022e-02
4.53905612e-01 -8.12798440e-01 3.00174564e-01 2.84685403e-01
4.79791731e-01 -1.10023201e+00 2.56509185e-01 -6.77749574e-01
-6.32014751e-01 1.39739394e+00 5.20656288e-01 -2.71310478e-01
2.53289223e-01 8.35058630e-01 2.69089282e-01 1.84959963e-01
-7.00587213e-01 7.40200803e-02 -3.21360469e-01 3.23622882e-01
5.55598438e-01 -3.30949455e-01 -5.11036813e-01 -8.16530883e-02
-7.97379985e-02 3.45029049e-02 1.28833985e+00 9.69427526e-01
-8.14779401e-01 -1.08727276e+00 -9.42002594e-01 -2.16168553e-01
-8.76399755e-01 3.81991491e-02 -6.99003115e-02 8.80010307e-01
-1.54096335e-01 1.08885264e+00 1.13559425e-01 -3.45828503e-01
1.34029329e-01 -3.48290324e-01 4.68032837e-01 -1.65343866e-01
-4.87863988e-01 6.32767975e-01 9.18045044e-02 -8.77139509e-01
-8.36573124e-01 -4.62890476e-01 -1.09327626e+00 -9.62654591e-01
-5.63758969e-01 1.04824305e-01 1.26560435e-01 4.98766810e-01
2.01199055e-01 3.56041968e-01 7.59726524e-01 -1.35818017e+00
-5.36349490e-02 -5.94527602e-01 -6.04624391e-01 3.51248384e-01
5.98895669e-01 -3.48010659e-01 -4.06202883e-01 7.73566306e-01] | [10.625197410583496, -1.3454842567443848] |
1d1ce7bc-d4f5-45a0-b472-b89abd4951db | self-supervised-dense-consistency | null | null | http://openaccess.thecvf.com//content/CVPR2022/html/Ko_Self-Supervised_Dense_Consistency_Regularization_for_Image-to-Image_Translation_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Ko_Self-Supervised_Dense_Consistency_Regularization_for_Image-to-Image_Translation_CVPR_2022_paper.pdf | Self-Supervised Dense Consistency Regularization for Image-to-Image Translation | Unsupervised image-to-image translation has gained considerable attention due to the recent impressive progress based on generative adversarial networks (GANs). In this paper, we present a simple but effective regularization technique for improving GAN-based image-to-image translation. To generate images with realistic local semantics and structures, we suggest to use an auxiliary self-supervised loss, enforcing point-wise consistency of the overlapped region between a pair of patches cropped from a single real image during training discriminators of GAN. Our experiment shows that the dense consistency regularization improves performance substantially on various image-to-image translation scenarios. It also achieves extra performance gains by using jointly with recent instance-level regularization methods. Furthermore, we verify that the proposed model captures domain-specific characteristics more effectively with only small fraction of training data. | ['Bohyung Han', 'Jinwoo Shin', 'Jae-Joon Han', 'Huijin Lee', 'Sungjoo Suh', 'Eunju Cha', 'Minsu Ko'] | 2022-01-01 | null | null | null | cvpr-2022-1 | ['unsupervised-image-to-image-translation'] | ['computer-vision'] | [ 7.78228462e-01 3.46628726e-01 -1.45842448e-01 -3.46407562e-01
-1.17884600e+00 -5.26525676e-01 8.42525423e-01 -5.08309960e-01
-2.95663085e-02 8.90705466e-01 2.35972270e-01 3.35822105e-02
4.58507985e-01 -7.23543406e-01 -1.06633818e+00 -7.47082889e-01
3.96806210e-01 3.09841931e-01 -1.80744633e-01 -2.10499763e-01
-1.37293190e-01 3.26954335e-01 -1.01465487e+00 4.67648089e-01
1.06548667e+00 9.17216003e-01 1.53056666e-01 4.67645168e-01
5.71989007e-02 8.48230243e-01 -6.36655927e-01 -4.79731202e-01
6.18577361e-01 -1.03359187e+00 -6.50711358e-01 4.35058236e-01
6.74701273e-01 -3.57802063e-01 -2.55753577e-01 9.93341267e-01
4.10009801e-01 -1.30673006e-01 6.74481690e-01 -1.22913265e+00
-9.00878549e-01 2.29241878e-01 -7.66103864e-01 -2.80001730e-01
2.84116030e-01 4.25684005e-01 7.90519476e-01 -7.89670706e-01
9.43378866e-01 1.07773936e+00 5.73612392e-01 6.52012408e-01
-1.51231980e+00 -6.49647772e-01 -8.15043598e-02 -2.11594477e-01
-1.29985559e+00 -3.69428098e-01 1.04015970e+00 -2.23368973e-01
5.76228082e-01 1.74443766e-01 3.60902697e-01 1.27628672e+00
1.20487608e-01 7.24137306e-01 1.58273554e+00 -5.64067483e-01
2.88655404e-02 1.74965456e-01 -8.48819077e-01 6.93281353e-01
-7.03388825e-02 1.51661798e-01 -3.01922381e-01 -1.09960951e-01
1.28134227e+00 -1.29204929e-01 -1.84219390e-01 -4.49309647e-01
-1.12082827e+00 8.69984210e-01 6.64132953e-01 2.49157116e-01
-2.68704385e-01 4.52257842e-01 1.06554784e-01 3.66320282e-01
6.29733622e-01 3.63479942e-01 -1.42167971e-01 1.32323861e-01
-1.08921325e+00 1.68562621e-01 3.12657386e-01 1.13853979e+00
8.70628893e-01 4.75826055e-01 -4.29200977e-01 8.35948408e-01
4.65250434e-03 7.09517896e-01 3.20264131e-01 -9.96469200e-01
5.42429864e-01 3.45448405e-01 7.71549121e-02 -8.79897535e-01
2.83585161e-01 -4.99662191e-01 -1.22731209e+00 2.10175902e-01
2.53949344e-01 -7.00106770e-02 -1.22870779e+00 2.01241446e+00
2.00351581e-01 2.33680487e-01 -1.49389915e-02 7.54436433e-01
3.97780895e-01 7.37819433e-01 1.09844021e-01 -9.26183313e-02
1.05802739e+00 -1.09547722e+00 -6.95912600e-01 -2.67505795e-01
3.04054022e-01 -9.54596579e-01 1.34080648e+00 -3.89128551e-02
-1.36045647e+00 -6.28848612e-01 -9.23994243e-01 -3.55849080e-02
-3.36621702e-02 1.18382864e-01 4.90285963e-01 5.77831268e-01
-1.12394774e+00 3.68478566e-01 -5.80192804e-01 -1.44467101e-01
7.51256704e-01 1.47139490e-01 -3.72539163e-01 -3.13517213e-01
-1.01152170e+00 5.64068675e-01 6.28004149e-02 -1.89736471e-01
-1.07002044e+00 -7.48026371e-01 -9.09158528e-01 -1.41379818e-01
1.85124338e-01 -1.11409557e+00 8.81606221e-01 -1.65679049e+00
-1.80032623e+00 1.19841576e+00 -1.58488393e-01 -4.12768275e-01
8.74818921e-01 -5.49500510e-02 -1.16276182e-01 1.94746628e-01
1.57751217e-01 1.05056679e+00 1.40304220e+00 -1.67928243e+00
-1.57032728e-01 -1.39774233e-01 -8.91380683e-02 7.44423866e-02
-2.09683761e-01 -8.96365568e-02 -4.81007814e-01 -1.24137187e+00
-1.18862368e-01 -1.14134133e+00 -2.69676059e-01 9.82334837e-02
-4.71952826e-01 3.26991051e-01 6.82368875e-01 -7.95414805e-01
6.60202324e-01 -1.99354243e+00 1.77597746e-01 9.84270275e-02
-4.15306091e-02 2.32631385e-01 -4.94172335e-01 3.88112843e-01
-9.39565077e-02 1.33161873e-01 -7.58220136e-01 -5.84688187e-01
-2.07137302e-01 3.50611985e-01 -5.59366584e-01 3.26085150e-01
5.34783959e-01 1.38182485e+00 -7.76117325e-01 -4.50864851e-01
5.70111759e-02 6.53438032e-01 -6.32372141e-01 5.63853741e-01
-3.43885243e-01 1.06705022e+00 -3.58535916e-01 4.25739646e-01
8.93197060e-01 -2.84917057e-01 1.24170154e-01 -1.25166371e-01
5.23478270e-01 -1.35722272e-02 -6.27641618e-01 2.07775068e+00
-7.09547400e-01 6.22422099e-01 6.14807801e-03 -1.05896485e+00
8.79036903e-01 1.95823029e-01 3.36556017e-01 -8.62592638e-01
-1.59991756e-02 8.47559348e-02 -2.22178146e-01 -9.83541980e-02
1.51642770e-01 -3.98604512e-01 -1.20718248e-01 4.77396101e-01
6.75164014e-02 -5.66223681e-01 -2.04972446e-01 1.61485866e-01
9.61162686e-01 5.31699836e-01 7.57026076e-02 -1.47539675e-01
4.79501903e-01 -7.96422176e-03 5.02363265e-01 6.94006085e-01
8.71921778e-02 1.21822393e+00 1.74170703e-01 -2.51197249e-01
-1.53184319e+00 -1.14255571e+00 9.90133584e-02 5.76920688e-01
1.13441505e-01 -7.37784207e-02 -1.02627969e+00 -8.83779645e-01
-3.12224746e-01 6.79241002e-01 -6.44908369e-01 -2.67489731e-01
-7.56301999e-01 -6.04633331e-01 6.55536175e-01 4.68840569e-01
8.97017777e-01 -9.96550024e-01 -1.32126898e-01 1.10836849e-02
-2.62169689e-01 -1.33729124e+00 -7.29359686e-01 -1.83636114e-01
-1.03402770e+00 -5.90824842e-01 -1.17379630e+00 -1.00880742e+00
1.14593530e+00 2.35285759e-01 1.35040903e+00 4.20416519e-02
-1.90925688e-01 3.11334312e-01 -2.75885850e-01 1.29465144e-02
-8.13410461e-01 -3.84912118e-02 -2.92841166e-01 1.47884443e-01
-4.12183255e-01 -8.13315630e-01 -6.54550433e-01 4.58519250e-01
-1.22560346e+00 5.27218521e-01 6.25242054e-01 1.14813888e+00
7.80861974e-01 -1.85433537e-01 5.03688872e-01 -1.10153568e+00
4.11027014e-01 -1.62300810e-01 -3.55495334e-01 1.96948200e-01
-5.75149953e-01 1.58328429e-01 8.51005673e-01 -4.48986918e-01
-1.28144836e+00 2.81949975e-02 3.06750443e-02 -6.77283585e-01
-1.88842535e-01 1.51557386e-01 -3.73642474e-01 -2.11036146e-01
6.45525873e-01 5.34469426e-01 8.80265981e-02 -1.56701759e-01
6.84411228e-01 3.75609785e-01 6.01501346e-01 -8.51788521e-01
1.27578127e+00 6.92206264e-01 -1.12582527e-01 -4.03964221e-01
-7.53676891e-01 9.94780138e-02 -4.57351834e-01 1.03657685e-01
7.07805037e-01 -1.22350478e+00 -1.82941049e-01 4.97425556e-01
-1.12033141e+00 -7.48188138e-01 -5.06840527e-01 9.41656381e-02
-9.05261993e-01 2.69559383e-01 -7.16746688e-01 -4.07094449e-01
-4.60910767e-01 -1.03294718e+00 1.28606272e+00 -2.28320956e-02
9.95939746e-02 -9.91050184e-01 1.40067130e-01 4.80607897e-01
6.02282226e-01 6.18355989e-01 7.12789297e-01 6.60114661e-02
-7.99169838e-01 -1.35260090e-01 -3.93910915e-01 6.58890247e-01
4.04905140e-01 -2.52955556e-01 -8.34562898e-01 -4.51916635e-01
7.49682486e-02 -4.94539678e-01 8.23892295e-01 2.18667656e-01
1.28931177e+00 -5.00378609e-01 -7.65875429e-02 9.18510735e-01
1.63916206e+00 -7.93910325e-02 1.15818393e+00 1.89050622e-02
1.05854404e+00 2.89674103e-01 4.14831221e-01 1.34264052e-01
1.03437357e-01 8.60098720e-01 2.93358654e-01 -6.20933652e-01
-5.74322522e-01 -5.97221613e-01 3.86479259e-01 7.00798273e-01
-1.99785873e-01 -2.93624014e-01 -4.50868279e-01 5.33300996e-01
-1.66173112e+00 -8.30005646e-01 2.43829399e-01 2.16439295e+00
1.13115895e+00 -9.44950283e-02 -6.94184005e-02 -3.57279330e-01
7.82315731e-01 3.33001167e-01 -5.12906969e-01 -2.37300530e-01
-4.09747779e-01 5.03047824e-01 5.96650124e-01 4.37060446e-01
-8.30845416e-01 1.15431786e+00 6.76611471e+00 1.25482070e+00
-1.10910976e+00 2.84758866e-01 1.08187318e+00 1.83669895e-01
-6.94181204e-01 2.47474257e-02 -2.56476194e-01 5.20246267e-01
5.76616347e-01 5.93835078e-02 5.16763031e-01 7.15640783e-01
1.88157350e-01 9.23077911e-02 -9.21237826e-01 9.00713384e-01
2.60861844e-01 -1.51719987e+00 3.47324908e-01 2.66521387e-02
1.52081907e+00 -3.45664203e-01 3.86105835e-01 -3.96533981e-02
3.71017158e-01 -1.30642962e+00 6.48757398e-01 2.66140968e-01
1.28901184e+00 -7.08844244e-01 5.39202154e-01 1.65831283e-01
-9.04172599e-01 3.87146950e-01 -3.00696462e-01 1.55251861e-01
2.24052608e-01 5.78649640e-01 -6.49918795e-01 6.80645883e-01
2.56743073e-01 6.31522357e-01 -3.64209831e-01 4.58635867e-01
-5.91706634e-01 7.28561401e-01 -1.38786316e-01 6.49306953e-01
2.69215763e-01 -4.80697989e-01 4.26497072e-01 1.01648724e+00
4.09058720e-01 -1.63065955e-01 4.09670174e-03 1.28082645e+00
-4.13525999e-01 2.65245000e-03 -9.05570686e-01 1.52929991e-01
1.25460729e-01 1.11644757e+00 -5.51268399e-01 -4.14638549e-01
-2.60176420e-01 1.61499369e+00 2.74824560e-01 4.78749871e-01
-1.07352614e+00 2.40588114e-02 5.45957267e-01 3.11127275e-01
4.34185326e-01 -1.82138145e-01 -5.20668149e-01 -1.25502443e+00
1.87942207e-01 -1.01213157e+00 2.61977836e-02 -8.80281031e-01
-1.33002806e+00 7.50101209e-01 -1.71817824e-01 -1.38630092e+00
-3.13649058e-01 -7.80702382e-02 -8.03664982e-01 9.10426199e-01
-1.52046895e+00 -1.69238913e+00 -4.21063602e-01 7.53324389e-01
4.58572477e-01 -2.34679505e-01 8.63968611e-01 1.87567696e-01
-3.67047310e-01 1.02133477e+00 1.88732833e-01 2.00706854e-01
9.19166327e-01 -1.00462794e+00 6.30204380e-01 8.73279929e-01
2.47899473e-01 3.92260551e-01 6.20406866e-01 -6.15960538e-01
-1.19866586e+00 -1.30400133e+00 5.72844625e-01 -3.55893701e-01
2.04297572e-01 -4.07292604e-01 -7.25181699e-01 7.35693991e-01
6.63254440e-01 1.61943510e-01 2.94431388e-01 -5.49304485e-01
-4.84474599e-01 -1.35007709e-01 -1.47840798e+00 7.02710509e-01
1.17029810e+00 -6.06533051e-01 -2.10387766e-01 4.05656904e-01
7.41278052e-01 -5.12122333e-01 -7.74645269e-01 4.47501987e-01
1.60980195e-01 -8.82856309e-01 1.13483608e+00 -4.35797632e-01
9.37321126e-01 -3.80199492e-01 -1.11717477e-01 -1.41006839e+00
-1.54712752e-01 -9.72581148e-01 2.04347044e-01 1.30283380e+00
1.41578048e-01 -4.64608848e-01 9.55927432e-01 4.24310774e-01
2.72760689e-02 -5.64150393e-01 -7.81465769e-01 -8.70824099e-01
2.89640844e-01 -4.04305235e-02 5.73365569e-01 1.02180982e+00
-5.14883876e-01 2.34133258e-01 -8.84214759e-01 -1.36259764e-01
8.53676796e-01 2.59055227e-01 1.07307971e+00 -4.21033651e-01
-5.16839027e-01 -1.61877051e-01 -4.06291515e-01 -1.23671067e+00
2.94388264e-01 -8.84173512e-01 3.86413001e-02 -1.25219834e+00
2.45161280e-01 -4.96680737e-01 -1.22703783e-01 3.99845749e-01
-2.47654885e-01 9.13338482e-01 2.07509249e-01 3.75794739e-01
-2.15240940e-01 8.15382242e-01 1.82458162e+00 -3.14763635e-01
8.09515268e-02 -2.19553396e-01 -6.24224782e-01 3.87755752e-01
7.89840698e-01 -6.33528650e-01 -5.02451062e-01 -5.65343320e-01
-4.69640940e-02 9.21078101e-02 4.89196301e-01 -8.39178681e-01
-3.42702389e-01 -3.33025217e-01 4.80378032e-01 -1.13251423e-02
3.30989033e-01 -7.37117887e-01 4.51712042e-01 3.44509065e-01
-4.37392533e-01 -1.48794800e-01 1.49609923e-01 6.21571481e-01
-4.92659122e-01 1.65605411e-01 1.11896133e+00 -1.94256634e-01
-3.53053004e-01 3.20908666e-01 1.30022064e-01 2.09882468e-01
8.70163321e-01 -1.28713742e-01 -7.25439787e-02 -7.98130095e-01
-4.56031024e-01 -2.96953499e-01 8.98607671e-01 1.62043139e-01
4.88128573e-01 -1.70845056e+00 -9.24302459e-01 3.11788976e-01
1.48898125e-01 4.38699536e-02 1.85503021e-01 6.30081475e-01
-5.89926004e-01 1.97808892e-01 -4.25488830e-01 -5.95736384e-01
-9.29571927e-01 4.46266502e-01 1.56737953e-01 -4.91158038e-01
-6.81017101e-01 8.30772996e-01 6.54763937e-01 -3.66013706e-01
-1.29829198e-01 -5.30220680e-02 5.48348010e-01 -6.60473406e-01
2.47833118e-01 -9.02379900e-02 -1.82356611e-01 -6.57832086e-01
-1.30755499e-01 7.81837165e-01 -4.41879518e-02 -2.60318547e-01
1.19522858e+00 -7.01858774e-02 2.21482292e-02 -1.01090193e-01
1.24557269e+00 1.60867795e-01 -1.78641760e+00 -3.08357805e-01
-6.05668604e-01 -6.90653145e-01 -1.32599577e-01 -8.73664439e-01
-1.45872390e+00 7.16359437e-01 5.75351357e-01 -1.89900413e-01
1.29081988e+00 -6.61135912e-02 1.08760870e+00 -1.04469091e-01
5.24997413e-01 -8.88988435e-01 3.49763244e-01 8.65910128e-02
1.02841508e+00 -1.32937968e+00 3.24925184e-02 -5.21686733e-01
-8.17023098e-01 6.16118133e-01 4.18496937e-01 -5.46985447e-01
2.32362047e-01 2.18464330e-01 1.82109863e-01 2.26289362e-01
-3.69015157e-01 -5.38898483e-02 3.45493853e-01 8.92215252e-01
3.45799416e-01 5.90211749e-02 -1.88877881e-01 1.57676518e-01
1.15828060e-01 1.67894773e-02 2.41626307e-01 7.05712616e-01
9.88934115e-02 -1.52154517e+00 -3.57194752e-01 1.06371701e-01
-4.01962191e-01 -2.82940865e-01 -3.48920524e-01 6.73723459e-01
-3.47916707e-02 6.32537663e-01 -1.51844020e-03 -1.30905673e-01
-5.49934618e-03 -1.34425819e-01 8.81856084e-01 -4.31751817e-01
-5.48150063e-01 2.64004260e-01 -2.03601912e-01 -5.69846332e-01
-5.80029786e-01 -3.79673481e-01 -1.01716352e+00 -2.75427788e-01
-4.76541221e-02 -2.02350780e-01 4.98278350e-01 7.80662596e-01
6.83011353e-01 4.61741775e-01 9.94785905e-01 -9.21947241e-01
-4.11831766e-01 -8.65463912e-01 -4.04193312e-01 9.06551301e-01
8.38333182e-03 -3.13366085e-01 -2.97488898e-01 4.81585234e-01] | [11.726645469665527, -0.4690330922603607] |
de66c584-0a89-4719-b97b-0632af5455bb | a-unified-pyramid-recurrent-network-for-video | 2211.03456 | null | https://arxiv.org/abs/2211.03456v2 | https://arxiv.org/pdf/2211.03456v2.pdf | A Unified Pyramid Recurrent Network for Video Frame Interpolation | Flow-guided synthesis provides a common framework for frame interpolation, where optical flow is estimated to guide the synthesis of intermediate frames between consecutive inputs. In this paper, we present UPR-Net, a novel Unified Pyramid Recurrent Network for frame interpolation. Cast in a flexible pyramid framework, UPR-Net exploits lightweight recurrent modules for both bi-directional flow estimation and intermediate frame synthesis. At each pyramid level, it leverages estimated bi-directional flow to generate forward-warped representations for frame synthesis; across pyramid levels, it enables iterative refinement for both optical flow and intermediate frame. In particular, we show that our iterative synthesis strategy can significantly improve the robustness of frame interpolation on large motion cases. Despite being extremely lightweight (1.7M parameters), our base version of UPR-Net achieves excellent performance on a large range of benchmarks. Code and trained models of our UPR-Net series are available at: https://github.com/srcn-ivl/UPR-Net. | ['Cheul-hee Hahm', 'Jayoon Koo', 'Youxin Chen', 'Jie Chen', 'Longhai Wu', 'Xin Jin'] | 2022-11-07 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Jin_A_Unified_Pyramid_Recurrent_Network_for_Video_Frame_Interpolation_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Jin_A_Unified_Pyramid_Recurrent_Network_for_Video_Frame_Interpolation_CVPR_2023_paper.pdf | cvpr-2023-1 | ['video-frame-interpolation'] | ['computer-vision'] | [ 2.04049516e-02 -2.03926802e-01 -3.96460116e-01 -1.42105538e-02
-7.74723589e-01 -3.84086460e-01 4.78364736e-01 -4.12954152e-01
-1.06481398e-02 7.19600379e-01 7.66309738e-01 -3.81052256e-01
5.67543924e-01 -7.70075798e-01 -8.32776964e-01 -2.04185694e-01
2.67889779e-02 -3.35439712e-01 3.53503674e-01 -2.46726915e-01
1.78885192e-01 5.80968201e-01 -1.26308596e+00 5.79701483e-01
3.64252716e-01 8.43204081e-01 9.56761464e-03 1.30344212e+00
3.82738888e-01 1.53251266e+00 -2.07476392e-01 -2.51306832e-01
4.20091152e-01 -5.26141584e-01 -1.12198699e+00 6.37998921e-04
6.16111338e-01 -1.01452494e+00 -1.04486322e+00 4.48622882e-01
5.72932243e-01 4.36477005e-01 3.40436846e-01 -1.02918220e+00
-8.00496280e-01 4.07978952e-01 -6.02041543e-01 4.01459485e-01
6.04139864e-01 6.48319840e-01 1.04847288e+00 -1.25353396e+00
9.28502619e-01 1.36097109e+00 7.87545621e-01 6.92453265e-01
-1.38261318e+00 -5.30595422e-01 6.84633851e-02 5.03089130e-02
-1.13739800e+00 -8.73161733e-01 5.91599882e-01 -4.14126843e-01
9.81515050e-01 -4.08298662e-03 7.18683362e-01 9.78474140e-01
3.17827910e-01 8.05511594e-01 6.33607745e-01 -5.13430871e-02
-8.31416622e-02 -7.68713772e-01 -1.93482861e-01 7.46764779e-01
-2.61920899e-01 1.77779600e-01 -7.61883855e-01 1.17260315e-01
1.50515556e+00 -1.13293387e-01 -5.93202353e-01 7.96633810e-02
-1.44173682e+00 5.72262406e-01 6.80360019e-01 -1.09184250e-01
-3.08758914e-01 8.03663313e-01 4.51572150e-01 2.03870967e-01
5.18477798e-01 1.14609867e-01 -2.50428438e-01 -2.56238461e-01
-1.13072443e+00 4.50049162e-01 4.70493555e-01 1.06709218e+00
1.00127971e+00 4.56459641e-01 -6.84419215e-01 5.75452089e-01
2.70798355e-01 2.77150720e-01 2.85240740e-01 -1.95012331e+00
4.54737812e-01 8.51553977e-02 4.29611534e-01 -1.10722995e+00
-4.52014916e-02 -8.79383832e-02 -9.01591420e-01 3.09214175e-01
5.52811682e-01 -3.91131371e-01 -8.80835354e-01 1.58939743e+00
1.73089504e-01 8.46055090e-01 -1.20728433e-01 8.77080798e-01
8.85917723e-01 9.62160289e-01 -4.91683148e-02 9.58095305e-03
9.96284425e-01 -1.42172635e+00 -5.20778000e-01 -5.71289919e-02
5.98307133e-01 -9.69105721e-01 8.39792490e-01 -6.89777955e-02
-1.78871691e+00 -8.31082106e-01 -7.14753866e-01 -8.12590420e-01
2.98983872e-01 2.83685396e-03 4.40067887e-01 3.18956487e-02
-1.66618490e+00 8.67722213e-01 -9.81124640e-01 1.25857964e-01
5.72966993e-01 2.38169387e-01 -2.36347809e-01 -1.40186027e-01
-1.04323602e+00 4.40294355e-01 -1.12899616e-01 3.24320763e-01
-1.00279939e+00 -1.12108052e+00 -1.11323225e+00 -1.06642924e-01
-2.16354147e-01 -1.19658399e+00 1.49827099e+00 -6.86866224e-01
-1.71281457e+00 5.14523149e-01 -7.70116150e-01 -7.76193976e-01
7.49958694e-01 -4.32003379e-01 3.46473940e-02 2.77129114e-01
1.55002341e-01 1.17804539e+00 1.03866720e+00 -9.15255070e-01
-6.46203220e-01 3.93207133e-01 2.22362444e-01 1.47621885e-01
3.57274681e-01 3.95686105e-02 -4.63193834e-01 -1.05231822e+00
-2.89955705e-01 -8.53582263e-01 -4.56161499e-01 2.26308256e-01
-1.56576976e-01 1.93595216e-02 8.65842819e-01 -7.78633296e-01
1.18797004e+00 -2.02724671e+00 1.16054960e-01 -2.77272075e-01
3.68022680e-01 2.37578899e-01 -2.14835316e-01 1.75170869e-01
-1.62948832e-01 2.71526668e-02 -2.46778622e-01 -6.27961814e-01
-3.16082656e-01 -2.18044035e-03 -6.63131177e-01 4.00087297e-01
4.23650503e-01 1.14408803e+00 -1.11284518e+00 -3.57391000e-01
5.74229836e-01 1.00253487e+00 -1.02036154e+00 2.99399525e-01
-2.18561918e-01 7.87897229e-01 -1.66138723e-01 4.25262898e-01
4.82176483e-01 -4.28602099e-01 -2.11286873e-01 -4.56664085e-01
-2.23413676e-01 5.40535390e-01 -9.24189627e-01 2.08712435e+00
-7.38818347e-01 1.10710418e+00 -1.04948737e-01 -2.00413242e-01
6.76704288e-01 4.78024036e-01 5.34711123e-01 -3.21117967e-01
9.78573412e-02 1.28422186e-01 -2.76424468e-01 1.09231204e-01
9.18151855e-01 4.71207619e-01 3.60928297e-01 4.13183063e-01
7.84892738e-02 -1.70838207e-01 4.63957012e-01 2.17609793e-01
1.18113899e+00 7.56578088e-01 1.42509356e-01 -8.13817009e-02
7.03500986e-01 -2.70864367e-01 6.87174559e-01 5.92631042e-01
-2.95198649e-01 1.15489554e+00 2.58915663e-01 -8.42875779e-01
-1.24434853e+00 -1.05716515e+00 9.66831148e-02 1.05688107e+00
3.32318526e-03 -6.65379643e-01 -4.80907351e-01 -2.51258016e-01
-1.57473221e-01 4.69182044e-01 -4.51507539e-01 2.38495961e-01
-1.12844455e+00 -1.68469787e-01 4.80192840e-01 6.74541116e-01
5.22898734e-01 -1.10269535e+00 -5.59217989e-01 4.48726416e-01
-4.43981856e-01 -1.21650815e+00 -1.09305751e+00 -5.41083872e-01
-9.31712270e-01 -9.15300250e-01 -1.03914165e+00 -7.11629093e-01
5.27500927e-01 4.58274633e-01 1.54525483e+00 3.91770452e-01
-1.29112095e-01 2.04648241e-01 -1.47420645e-01 3.78267080e-01
-5.56740999e-01 9.76839140e-02 -4.37513500e-01 1.40750349e-01
-1.82916060e-01 -8.36818516e-01 -1.23230910e+00 3.65806162e-01
-6.15234792e-01 3.76878172e-01 -4.22924422e-02 6.84417725e-01
5.25171995e-01 -7.95718133e-01 3.39722127e-01 -5.28421938e-01
3.66441369e-01 -2.89080262e-01 -5.65305114e-01 -2.19430968e-01
6.94401935e-02 -5.46814203e-02 6.88515067e-01 1.63073931e-02
-1.06751287e+00 2.57529840e-02 -4.45737988e-01 -7.62176931e-01
6.56404272e-02 -1.37038529e-01 5.25347054e-01 -1.12867162e-01
6.74640000e-01 1.48830160e-01 -1.92946360e-01 -1.25556201e-01
6.19330347e-01 2.21181393e-01 8.50651562e-01 -5.77705443e-01
7.46383905e-01 6.80863380e-01 8.35409760e-02 -5.62960327e-01
-7.58870363e-01 -1.05058149e-01 -5.28195143e-01 -3.25260997e-01
8.48202527e-01 -1.40390587e+00 -8.43733549e-01 5.47888041e-01
-1.46292126e+00 -9.64779496e-01 -4.02112424e-01 3.32907259e-01
-8.29785407e-01 2.69150525e-01 -1.28830683e+00 -3.05675477e-01
-6.25192404e-01 -1.48672879e+00 1.18906534e+00 1.20831899e-01
-4.32235271e-01 -1.15955889e+00 5.61846271e-02 1.54346213e-01
4.42448616e-01 3.25356007e-01 9.59356576e-02 3.62829298e-01
-9.33770835e-01 2.91836828e-01 -3.83319259e-01 2.00485528e-01
4.31979895e-01 4.70128685e-01 -8.47721577e-01 -3.91447365e-01
-4.12708461e-01 -3.29778671e-01 1.03863764e+00 6.40724659e-01
1.13191891e+00 -2.53897548e-01 1.32509127e-01 1.07867908e+00
1.17587733e+00 -1.02871202e-01 8.85768771e-01 1.47174001e-01
1.04117715e+00 -1.39906211e-02 3.07425559e-01 5.89545310e-01
5.31452000e-01 5.58136225e-01 1.32687628e-01 -6.10146187e-02
-7.90717065e-01 -2.95433223e-01 5.41178524e-01 7.36468375e-01
-5.33655524e-01 5.02612069e-02 -6.94582880e-01 5.97842932e-01
-1.94707882e+00 -1.17120945e+00 -9.88974273e-02 1.89485550e+00
1.04801214e+00 -1.16469609e-02 1.97386816e-01 -1.32154718e-01
7.60704279e-01 6.90349460e-01 -3.86676073e-01 -4.72125053e-01
2.87662633e-02 3.88331234e-01 5.43113947e-01 1.17898357e+00
-1.03706849e+00 1.15097129e+00 6.50936699e+00 6.19569004e-01
-1.14079857e+00 5.34685664e-02 9.55357075e-01 -2.02469751e-01
-3.99519265e-01 1.35033391e-02 -8.49010944e-01 4.75768596e-01
8.21605027e-01 -1.22340634e-01 5.12206137e-01 4.50457007e-01
7.21264303e-01 1.49136931e-01 -1.11534917e+00 9.18920994e-01
-3.76251012e-01 -2.13995934e+00 2.00925604e-01 -2.66982079e-01
1.09409428e+00 2.38457814e-01 5.53320460e-02 -3.66994254e-02
6.32613182e-01 -9.94423270e-01 6.57772124e-01 3.35743397e-01
9.46244121e-01 -7.64034688e-01 1.96685418e-01 -2.46419340e-01
-1.52511609e+00 3.18809688e-01 -3.64017516e-01 -4.08325136e-01
5.25518239e-01 5.38681686e-01 -3.71900558e-01 4.59171653e-01
5.21092594e-01 1.43401170e+00 -3.89519632e-01 7.13312984e-01
-3.41159761e-01 5.43846071e-01 -7.78985694e-02 7.44932473e-01
3.23533982e-01 -1.61255330e-01 4.52498019e-01 1.39235163e+00
3.03020597e-01 -7.77141303e-02 7.21952543e-02 8.47279191e-01
-4.38800931e-01 -2.62052864e-01 -6.84387207e-01 5.05241275e-01
5.57506621e-01 1.16964996e+00 -4.87379879e-01 -4.41897362e-01
-5.50361454e-01 9.66657102e-01 3.49157631e-01 5.57252049e-01
-9.72241998e-01 -2.04198465e-01 1.06358767e+00 8.65344182e-02
4.00525004e-01 -4.39631134e-01 -2.71097243e-01 -1.45214522e+00
-2.10858077e-01 -7.24089861e-01 7.17470869e-02 -9.94701505e-01
-7.73950577e-01 6.36466920e-01 -4.42377716e-01 -1.21603572e+00
-6.27013147e-01 -2.52696812e-01 -7.60881364e-01 1.04346323e+00
-1.63535118e+00 -1.09549057e+00 -3.28719467e-01 8.62814784e-01
9.47603762e-01 2.07709715e-01 4.93285537e-01 3.32937270e-01
-5.47664702e-01 4.58434314e-01 -3.32489312e-01 5.03829122e-01
6.99658215e-01 -9.60113108e-01 1.29807568e+00 1.24424565e+00
-6.75377771e-02 4.48940694e-01 5.85570276e-01 -5.80699980e-01
-1.31260097e+00 -1.54508126e+00 6.56704724e-01 -3.29337448e-01
7.32909799e-01 -1.09644132e-02 -7.33183205e-01 1.09622276e+00
4.54237789e-01 6.95720732e-01 1.75979361e-01 -5.39158225e-01
-4.83575314e-01 -5.83404936e-02 -7.77620077e-01 1.10073984e+00
1.06932461e+00 -6.58295393e-01 -9.51399878e-02 2.40431830e-01
1.01192546e+00 -9.70180273e-01 -1.18644619e+00 2.32187673e-01
5.01056194e-01 -1.29108703e+00 1.24939799e+00 -2.59601653e-01
1.07794929e+00 -5.70411861e-01 -5.63706756e-02 -1.01060438e+00
-5.64327419e-01 -1.39834428e+00 -5.03514647e-01 8.91795456e-01
2.82798409e-01 -4.67218935e-01 8.25934172e-01 5.73046684e-01
-2.16913953e-01 -8.56379628e-01 -6.06188178e-01 -2.67622203e-01
8.37962795e-03 -4.72548157e-01 4.23483580e-01 5.43722808e-01
-4.03743923e-01 1.93021178e-01 -6.59924626e-01 1.76303517e-02
6.36454105e-01 9.73668769e-02 9.14323866e-01 -3.12460423e-01
-4.13751751e-01 -3.97018909e-01 -2.27144599e-01 -1.57244194e+00
2.52933145e-01 -7.30868399e-01 -8.99614617e-02 -1.60689735e+00
-3.64262223e-01 -1.38480842e-01 2.32809572e-03 3.44282508e-01
-2.34851122e-01 7.85670578e-01 6.82980478e-01 3.25901389e-01
-3.42699051e-01 3.79942298e-01 1.61116230e+00 4.78723943e-02
-3.76084656e-01 -1.14211611e-01 -3.43926698e-01 7.94335902e-01
8.78189743e-01 -6.28493074e-03 -3.61380607e-01 -8.94991219e-01
-1.32912369e-02 4.81376946e-01 5.52246332e-01 -1.10107565e+00
2.09270224e-01 -7.44349649e-03 4.93453801e-01 -5.89558899e-01
5.33363283e-01 -2.78423995e-01 1.54216707e-01 4.81798142e-01
-5.34573674e-01 4.44195509e-01 1.00351997e-01 2.76950628e-01
-1.80593699e-01 2.13775411e-01 9.60070193e-01 -2.37048149e-01
-7.07839906e-01 7.16806293e-01 -2.80112475e-01 4.11172628e-01
5.97577572e-01 -9.40638036e-02 -2.39231259e-01 -6.35042071e-01
-6.69136405e-01 1.22251444e-01 5.39014637e-01 2.68389910e-01
7.46018767e-01 -1.43428946e+00 -8.94389570e-01 6.26625195e-02
-5.35334766e-01 4.26757514e-01 1.61550969e-01 6.98489547e-01
-1.11873877e+00 2.01744556e-01 -1.86928555e-01 -6.01883531e-01
-8.32304597e-01 3.65840375e-01 4.13385987e-01 -1.34428456e-01
-8.42612028e-01 9.20343399e-01 3.87066841e-01 1.77202314e-01
-4.57811635e-03 -5.61603487e-01 2.32763767e-01 -2.78431475e-01
8.47065151e-01 6.85225487e-01 -2.27550894e-01 -7.63065517e-01
-1.16590299e-01 6.52443945e-01 -3.47294882e-02 -1.14832744e-01
1.13221812e+00 -3.00740272e-01 -2.16769263e-01 2.10475832e-01
1.34447050e+00 -9.05617401e-02 -2.08061528e+00 -6.32653981e-02
-5.10259569e-01 -5.73047638e-01 -1.95994172e-02 -9.34909135e-02
-1.36334491e+00 8.02076578e-01 -6.64021522e-02 -3.13701957e-01
1.12346315e+00 -5.64981461e-01 1.20384014e+00 1.12867951e-01
3.10963094e-01 -4.91300434e-01 -8.71002451e-02 7.56075203e-01
8.04862320e-01 -9.65490162e-01 4.14739363e-02 -4.84282106e-01
-2.89978117e-01 1.24771047e+00 4.28049564e-01 -7.28112996e-01
5.54637194e-01 4.15196598e-01 -4.29560523e-03 4.41308677e-01
-1.03728008e+00 2.24591177e-02 1.34165213e-01 4.67824757e-01
9.48473215e-01 -3.60266864e-01 7.79606625e-02 -4.28610235e-01
-2.06138134e-01 5.12781858e-01 8.69675636e-01 7.15587437e-01
-1.68626577e-01 -9.59886611e-01 -2.22282276e-01 1.28315181e-01
-6.58640265e-01 -3.46445322e-01 3.28061819e-01 5.10879397e-01
-3.27536166e-01 8.57946932e-01 1.86983272e-01 -1.51075944e-01
5.00935726e-02 -3.38286012e-01 4.71763879e-01 -2.55156547e-01
-6.43583298e-01 5.95386373e-03 1.01667285e-01 -1.12609971e+00
-6.08138978e-01 -5.16845524e-01 -1.21087289e+00 -7.50409126e-01
4.66473311e-01 -2.43919939e-01 -1.52603582e-01 5.45466840e-01
6.74944580e-01 6.36748374e-01 7.78817952e-01 -1.53995204e+00
5.58905303e-02 -5.37876308e-01 6.64239749e-02 4.04315889e-01
8.42619061e-01 -1.84064373e-01 -2.13020310e-01 6.09793007e-01] | [10.68064022064209, -1.378367304801941] |
cc7d8bf1-913f-4252-9090-ca218b54d755 | query-as-context-pre-training-for-dense | 2212.09598 | null | https://arxiv.org/abs/2212.09598v2 | https://arxiv.org/pdf/2212.09598v2.pdf | Query-as-context Pre-training for Dense Passage Retrieval | Recently, methods have been developed to improve the performance of dense passage retrieval by using context-supervised pre-training. These methods simply consider two passages from the same document to be relevant, without taking into account the possibility of weakly correlated pairs. Thus, this paper proposes query-as-context pre-training, a simple yet effective pre-training technique to alleviate the issue. Query-as-context pre-training assumes that the query derived from a passage is more likely to be relevant to that passage and forms a passage-query pair. These passage-query pairs are then used in contrastive or generative context-supervised pre-training. The pre-trained models are evaluated on large-scale passage retrieval benchmarks and out-of-domain zero-shot benchmarks. Experimental results show that query-as-context pre-training brings considerable gains and meanwhile speeds up training, demonstrating its effectiveness and efficiency. Our code will be available at https://github.com/caskcsg/ir/tree/main/cotmae-qc . | ['Songlin Hu', 'Fuzheng Zhang', 'Zijia Lin', 'Wanhui Qian', 'Guangyuan Ma', 'Xing Wu'] | 2022-12-19 | null | null | null | null | ['passage-retrieval'] | ['natural-language-processing'] | [-5.25013097e-02 -5.64991832e-01 -3.89113873e-01 -2.49634042e-01
-1.67308533e+00 -5.41919231e-01 8.25833619e-01 4.83019859e-01
-4.75601166e-01 7.12393165e-01 6.32435799e-01 -1.48895569e-02
-6.45692050e-02 -8.87158394e-01 -5.36821008e-01 -4.30182487e-01
-1.16235398e-01 5.74536502e-01 5.57418227e-01 -4.36889589e-01
2.77312070e-01 -6.04543090e-02 -1.36802292e+00 6.57364488e-01
7.03407526e-01 5.10716677e-01 3.98902476e-01 8.99939120e-01
-3.25602949e-01 6.93650663e-01 -6.90892160e-01 -1.63797930e-01
6.06979467e-02 -8.32110107e-01 -1.15953648e+00 -5.07806897e-01
3.47079754e-01 -2.30609834e-01 -4.73212272e-01 6.10708058e-01
7.42618322e-01 6.52484596e-01 5.73960483e-01 -7.32475281e-01
-7.51790583e-01 5.27686894e-01 -2.79572010e-01 8.38715374e-01
7.52060175e-01 -5.04454255e-01 1.47956061e+00 -1.34161878e+00
5.34265161e-01 1.03448403e+00 2.61358738e-01 4.16817278e-01
-8.57596755e-01 -4.96779412e-01 -7.11745247e-02 3.88204008e-01
-1.61519802e+00 -3.37974489e-01 6.34354353e-01 2.66884733e-02
1.15216720e+00 5.63997507e-01 5.67020535e-01 1.02532005e+00
-1.16307229e-01 1.15702593e+00 5.11129320e-01 -7.92095482e-01
9.26892757e-02 -6.36219084e-02 3.61751407e-01 2.43982553e-01
-1.39872938e-01 -7.02315941e-02 -3.59293252e-01 -5.48317432e-01
4.44596261e-01 1.63315535e-01 -4.45348889e-01 1.29837364e-01
-1.01294756e+00 8.65556121e-01 5.67503214e-01 6.83757544e-01
-1.54559463e-01 -2.42220402e-01 4.57297117e-01 4.24472481e-01
5.36737621e-01 5.14364660e-01 -1.74738631e-01 -8.36992115e-02
-1.27872992e+00 4.69767272e-01 7.61708796e-01 1.19700813e+00
7.86123693e-01 -7.25645781e-01 -8.73251200e-01 1.17903900e+00
1.68327719e-01 4.20313329e-01 4.54172522e-01 -4.55952644e-01
5.72788656e-01 3.52856159e-01 1.69744581e-01 -7.60111868e-01
-8.73775333e-02 -4.47251648e-01 -6.06261611e-01 -8.04208636e-01
-1.24360025e-01 1.77734405e-01 -7.55342543e-01 1.50692320e+00
3.09257865e-01 4.20042604e-01 1.64371759e-01 9.56819952e-01
1.11536396e+00 1.10649753e+00 1.51145041e-01 -2.56366670e-01
1.21905637e+00 -1.40983343e+00 -5.66453099e-01 -5.70621490e-02
7.84693897e-01 -1.31065893e+00 1.42645323e+00 -2.12724745e-01
-1.06906581e+00 -5.43526351e-01 -8.10633123e-01 -5.47078192e-01
-2.94983894e-01 1.11414015e-01 1.89038247e-01 -1.30740255e-02
-8.38621199e-01 2.61055112e-01 -4.92448211e-01 -6.11054420e-01
2.73544211e-02 -1.67733669e-01 5.22007793e-02 -3.69634330e-01
-1.60201287e+00 5.38176417e-01 4.16340768e-01 -2.70114750e-01
-8.20114732e-01 -7.56549299e-01 -4.49505270e-01 3.26149702e-01
2.68029004e-01 -8.63687038e-01 1.42702401e+00 -4.84251678e-01
-1.00184023e+00 7.45890677e-01 -5.79683661e-01 -2.22414315e-01
1.98418841e-01 -6.09144151e-01 -6.50572658e-01 4.74293709e-01
3.51562142e-01 6.46091282e-01 4.86157000e-01 -1.17200768e+00
-6.25370920e-01 9.65559930e-02 2.62334764e-01 5.63799024e-01
-2.94279337e-01 3.57059836e-01 -1.23524809e+00 -7.60623872e-01
-7.85180777e-02 -9.30713117e-01 4.04510498e-02 -4.12797600e-01
-4.88508850e-01 -5.38609803e-01 8.11285436e-01 -4.40035880e-01
1.54038930e+00 -1.97031426e+00 -3.96729022e-01 5.38951382e-02
-1.89589471e-01 5.06479681e-01 -5.93679249e-01 1.10954809e+00
1.54783711e-01 1.04275055e-01 -1.68563873e-01 -5.60984373e-01
-1.41258910e-01 8.84103477e-02 -6.53901279e-01 2.27042269e-02
6.46555750e-03 9.59933102e-01 -1.26889122e+00 -8.32871139e-01
-4.32587489e-02 5.09843647e-01 -5.17745852e-01 5.30551672e-01
-4.12625939e-01 3.24409336e-01 -6.92856133e-01 3.77730638e-01
3.33395064e-01 -4.55570281e-01 -1.60216615e-01 9.96396467e-02
3.05964291e-01 8.67606461e-01 -5.17711163e-01 1.91674972e+00
-5.46122134e-01 5.98919034e-01 -5.70457816e-01 -7.60903895e-01
7.36061037e-01 4.69934702e-01 3.03069532e-01 -1.00395179e+00
-1.11627400e-01 1.10087618e-02 -4.30355966e-01 -5.10127246e-01
1.08968198e+00 3.20383906e-02 -4.65385504e-02 4.96360421e-01
4.07547280e-02 -2.01699942e-01 6.45451605e-01 7.11802065e-01
9.63450909e-01 7.16759488e-02 1.82109654e-01 -6.44176677e-02
6.26304269e-01 -1.82447061e-02 3.71246189e-01 1.02563870e+00
-1.06923794e-02 1.05653083e+00 -6.16485998e-02 2.25250736e-01
-9.05727208e-01 -1.09919381e+00 3.94650325e-02 1.30752838e+00
4.02779609e-01 -9.17590916e-01 -2.59666830e-01 -6.16473317e-01
-3.09947848e-01 7.53957331e-01 -5.02111971e-01 -2.50768006e-01
-5.21094203e-01 -3.31534445e-01 4.16924655e-01 4.08882082e-01
3.07938576e-01 -1.03804636e+00 -5.36873750e-02 2.05632612e-01
-7.62988627e-01 -9.61995363e-01 -8.03367317e-01 -2.56352037e-01
-8.12750995e-01 -9.17733073e-01 -1.21443784e+00 -9.23496366e-01
4.95669365e-01 9.04989481e-01 1.42212713e+00 5.69092572e-01
-1.66479632e-01 5.52753687e-01 -9.56811488e-01 -8.57691616e-02
-2.63990134e-01 2.11972088e-01 -2.87659854e-01 -5.09570122e-01
6.43058360e-01 -5.22326291e-01 -9.01955605e-01 3.29518944e-01
-1.04631793e+00 -3.20645839e-01 4.26813573e-01 1.02167356e+00
9.69311714e-01 -3.66224259e-01 6.50763988e-01 -6.68631732e-01
8.65819156e-01 -7.66980708e-01 -2.24935055e-01 8.30826223e-01
-4.32502627e-01 -1.77382499e-01 5.00377834e-01 -2.69440919e-01
-1.03130507e+00 -6.23078465e-01 -2.37583786e-01 -5.31947494e-01
-9.65102017e-02 9.06612933e-01 2.69432724e-01 4.67798263e-01
8.09437573e-01 4.76820737e-01 -6.16605759e-01 -4.85914648e-01
2.75324315e-01 5.98440945e-01 2.44697988e-01 -6.93687081e-01
7.00759172e-01 2.57045954e-01 -3.85432035e-01 -9.21493232e-01
-1.14202213e+00 -1.37324238e+00 -3.82246643e-01 -1.54249696e-03
4.39610481e-01 -1.12498081e+00 2.39538357e-01 -1.67319819e-01
-1.02027142e+00 -2.30099618e-01 -2.83049881e-01 5.57583869e-01
-1.83470026e-01 6.98376894e-01 -6.90128505e-01 -6.07922614e-01
-7.91634083e-01 -5.89048266e-01 1.03299034e+00 5.28676033e-01
-1.57318562e-01 -1.00768471e+00 6.18124247e-01 2.95065701e-01
3.84368837e-01 -4.57717389e-01 6.62027717e-01 -1.03778648e+00
-4.31204051e-01 -4.62227404e-01 -1.16112828e-01 9.74232331e-02
2.61345088e-01 4.75611277e-02 -9.69443679e-01 -5.70575774e-01
-3.15343976e-01 -5.25740385e-01 9.68457639e-01 5.62008470e-02
8.58061552e-01 -3.70485574e-01 -4.33313310e-01 1.76745892e-01
1.48010111e+00 -3.36575024e-02 7.30501831e-01 1.84983000e-01
3.33400965e-01 2.27844432e-01 9.42905605e-01 3.54226857e-01
4.44151491e-01 7.96030939e-01 -2.51642555e-01 -6.64534718e-02
-2.30340391e-01 -5.68560600e-01 1.30375162e-01 1.16701198e+00
1.88448861e-01 -6.12158537e-01 -7.61074722e-01 1.01033711e+00
-1.76401353e+00 -1.32162869e+00 -7.73260146e-02 2.30276155e+00
1.08338928e+00 -7.14482814e-02 2.05523551e-01 -2.04915076e-01
5.70146143e-01 1.88766450e-01 -1.15819253e-01 -6.69979304e-02
-8.07482451e-02 2.00808108e-01 -2.46040493e-01 6.74350142e-01
-1.11497700e+00 8.89949262e-01 5.58678246e+00 1.03982317e+00
-9.62502003e-01 1.60537466e-01 4.39936608e-01 -3.67578030e-01
-5.71876109e-01 1.15314491e-01 -9.58709300e-01 3.55108738e-01
8.78776789e-01 -5.39452612e-01 -1.99840590e-02 7.04340279e-01
1.49316028e-01 -2.67953306e-01 -1.07368350e+00 6.25531614e-01
2.38987416e-01 -1.24043739e+00 3.21084291e-01 -4.15629983e-01
9.60453331e-01 3.54569614e-01 -7.90362060e-02 7.99628735e-01
-2.15422735e-02 -3.58510315e-01 3.49889249e-01 5.00844181e-01
5.93263566e-01 -6.49008632e-01 7.73933887e-01 3.79987419e-01
-1.39535069e+00 2.95644641e-01 -6.12864554e-01 2.07916752e-01
1.83671415e-01 6.39828861e-01 -9.36080158e-01 9.49134588e-01
7.53464460e-01 6.72285259e-01 -6.38523698e-01 1.52180278e+00
-2.95256615e-01 8.76171887e-01 -4.19729680e-01 -2.47566774e-01
3.46758187e-01 -4.43500839e-02 6.70179129e-01 1.74510908e+00
3.74027938e-01 3.73623013e-01 3.24629813e-01 7.59004593e-01
-9.92818028e-02 4.96054620e-01 -3.76126975e-01 6.43643737e-02
7.06940114e-01 1.01233351e+00 -4.97396588e-01 -7.04498410e-01
-5.98776102e-01 1.03272271e+00 3.21412921e-01 6.27290666e-01
-6.65403187e-01 -6.04541719e-01 4.55155194e-01 -6.56469911e-02
4.19028908e-01 1.08298473e-02 3.48981053e-01 -1.36696613e+00
2.37436086e-01 -4.84297007e-01 7.74609864e-01 -7.38330007e-01
-1.49170947e+00 5.63953459e-01 1.95654720e-01 -1.66575491e+00
-4.35072720e-01 2.40955248e-01 -8.80944252e-01 9.85941648e-01
-1.89420068e+00 -9.60105836e-01 -2.35010326e-01 7.22843647e-01
7.02753246e-01 1.38915196e-01 1.04935420e+00 4.80150104e-01
-2.50157982e-01 8.23283613e-01 3.15432608e-01 3.37156206e-01
9.63817716e-01 -1.01054835e+00 2.96278596e-01 1.00737178e+00
5.76824486e-01 9.55708146e-01 4.60859925e-01 -5.23651659e-01
-1.04714620e+00 -1.05317056e+00 1.38816643e+00 -3.63082170e-01
4.33136851e-01 -2.22984016e-01 -1.24323452e+00 4.23824638e-01
4.10596758e-01 5.21392114e-02 1.06228435e+00 3.45527500e-01
-4.48692501e-01 -2.44397596e-02 -7.28203475e-01 6.29763246e-01
8.18957031e-01 -1.10725307e+00 -1.15540302e+00 6.46034479e-01
8.46816421e-01 -3.48850369e-01 -6.68010533e-01 2.90851831e-01
2.16413587e-01 -8.26988459e-01 1.35431480e+00 -4.81883854e-01
5.52935541e-01 -1.79274946e-01 -1.32832319e-01 -1.08321786e+00
-2.91955411e-01 -5.02219260e-01 -3.37661415e-01 1.34643710e+00
5.51070750e-01 -8.94824564e-02 5.03436506e-01 3.01858753e-01
-2.00710773e-01 -7.38490582e-01 -7.53889680e-01 -1.01800847e+00
2.87224531e-01 -1.12215728e-01 4.22442436e-01 9.90876257e-01
4.42328900e-01 6.56912267e-01 -9.07413438e-02 1.66219011e-01
2.78582841e-01 5.04933059e-01 8.27621579e-01 -7.23340333e-01
-2.64120668e-01 -3.47293615e-01 1.90159842e-01 -1.54341221e+00
-1.33868903e-01 -9.22621667e-01 3.02825928e-01 -1.92471910e+00
4.89019096e-01 -5.18345654e-01 -5.97083211e-01 3.05693448e-01
-7.82720625e-01 1.91685572e-01 1.42053738e-01 6.00372314e-01
-1.03724146e+00 7.82060266e-01 1.25405145e+00 -1.35691851e-01
-3.52883279e-01 7.20734075e-02 -3.92148226e-01 7.40756020e-02
8.59254897e-01 -5.70823014e-01 -7.54091918e-01 -4.30681020e-01
4.97907698e-02 1.35984182e-01 1.73057660e-01 -1.00524414e+00
3.92566204e-01 8.03302750e-02 1.78262591e-01 -9.12758529e-01
5.08005679e-01 -5.15376806e-01 -3.46388549e-01 2.50145216e-02
-7.00030863e-01 5.05027324e-02 2.65345156e-01 6.77392840e-01
-6.69441283e-01 -4.09435451e-01 3.42035949e-01 -1.20191753e-01
-7.66619742e-01 3.09518844e-01 -1.64156795e-01 5.55933118e-01
5.83010614e-01 2.68205792e-01 -4.54341948e-01 -7.70889878e-01
-4.48633343e-01 4.02497858e-01 2.87137657e-01 6.00574970e-01
7.51065493e-01 -1.33616710e+00 -9.35576677e-01 -1.21117420e-01
6.90075040e-01 -8.92211199e-02 4.23221350e-01 5.75338364e-01
-2.13926166e-01 8.39629710e-01 4.44074035e-01 -6.80414200e-01
-1.45109975e+00 6.89663887e-01 -5.89485243e-02 -6.24983430e-01
-6.81927264e-01 1.12090385e+00 3.86717655e-02 -2.49127954e-01
1.68584019e-01 -3.39556307e-01 -2.06739396e-01 -8.46301094e-02
9.14698362e-01 -3.00225858e-02 1.98175665e-02 -4.65497017e-01
-2.41003618e-01 4.80422676e-01 -5.33558071e-01 -2.71789879e-01
1.01728034e+00 -3.32004726e-01 -3.11553683e-02 3.96209866e-01
1.69415402e+00 8.39626044e-02 -6.33627117e-01 -8.17038476e-01
1.13472752e-01 -6.23466969e-01 -4.82225744e-03 -9.11637366e-01
-6.14559650e-01 8.00748885e-01 2.36529216e-01 1.89599693e-01
1.18837178e+00 1.29857779e-01 1.02453780e+00 8.30742180e-01
3.64444405e-01 -1.01741374e+00 1.92414507e-01 8.47129464e-01
1.05628824e+00 -1.26210046e+00 1.47443712e-01 -2.16707617e-01
-4.07825291e-01 7.71857619e-01 5.10210216e-01 -1.24327354e-01
7.07820594e-01 -2.50136048e-01 9.98582244e-02 -4.03069630e-02
-8.53252590e-01 -4.46777403e-01 7.42085099e-01 2.05537274e-01
6.23883545e-01 -1.68088123e-01 -5.30535817e-01 2.27705538e-01
1.68145858e-02 2.05738634e-01 1.66946977e-01 1.28747761e+00
-3.29186648e-01 -1.18463254e+00 -1.52493954e-01 3.89182538e-01
-4.21498656e-01 -6.46173239e-01 -3.27965498e-01 5.80887794e-01
-5.48252642e-01 1.13830721e+00 7.07096979e-02 -1.69938982e-01
1.25447586e-01 1.22296162e-01 4.23306108e-01 -8.96862686e-01
-6.16854668e-01 3.99313241e-01 1.87271655e-01 -3.66391718e-01
-4.97661829e-01 -4.06466603e-01 -1.30576062e+00 -1.49064166e-02
-5.37943423e-01 8.30004275e-01 1.00338437e-01 8.75930548e-01
6.19069695e-01 2.77343810e-01 8.35042059e-01 -4.66756880e-01
-2.45985463e-01 -1.21314955e+00 -2.05877393e-01 4.18685555e-01
4.03091043e-01 -2.78466672e-01 -3.80329072e-01 -1.11606136e-01] | [11.49951457977295, 7.709858417510986] |
67e1d98b-0fe6-4c62-91bf-05525ed6cc04 | faceatlasar-atlas-of-facial-acupuncture | 2111.14755 | null | https://arxiv.org/abs/2111.14755v2 | https://arxiv.org/pdf/2111.14755v2.pdf | FaceAtlasAR: Atlas of Facial Acupuncture Points in Augmented Reality | Acupuncture is a technique in which practitioners stimulate specific points on the body. Those points, called acupuncture points (or acupoints), anatomically define areas on the skin relative to specific landmarks on the body. However, mapping the acupoints to individuals could be challenging for inexperienced acupuncturists. In this project, we proposed a system to localize and visualize facial acupoints for individuals in an augmented reality (AR) context. This system combines a face alignment model and a hair segmentation model to provide dense reference points for acupoints localization in real-time (60FPS). The localization process takes the proportional bone (B-cun or skeletal) measurement method, which is commonly operated by specialists; however, in the real practice, operators sometimes find it inaccurate due to skill-related error. With this system, users, even without any skills, can locate the facial acupoints as a part of the self-training or self-treatment process. | ['Dong Zhang', 'Jurgen Schulze', 'Menghe Zhang'] | 2021-11-29 | null | null | null | null | ['face-alignment'] | ['computer-vision'] | [-8.02474022e-02 2.91477982e-02 -2.35241383e-01 -1.89786434e-01
-5.96293449e-01 -5.79933167e-01 -3.10861468e-01 -6.59600943e-02
-6.24828087e-03 5.33150911e-01 9.86090153e-02 2.02410206e-01
2.48519421e-01 -7.79610813e-01 -2.16529474e-01 -5.20290971e-01
2.55267739e-01 4.42764759e-01 -1.59416184e-01 -3.35981011e-01
4.87397134e-01 6.99906051e-01 -1.19769895e+00 -6.10691437e-04
9.40027535e-01 5.15305161e-01 -1.24940118e-02 3.23579684e-02
-3.04427028e-01 -2.25056335e-01 -3.28600526e-01 -5.59861422e-01
1.68507233e-01 -5.14152884e-01 -4.79401886e-01 -2.68454581e-01
5.06116807e-01 -3.31218928e-01 4.47915979e-02 1.17415476e+00
6.47839308e-01 -3.26634616e-01 2.78938711e-01 -8.79323184e-01
-5.73087394e-01 3.59084576e-01 -1.36325181e+00 -3.01595002e-01
7.85523236e-01 -2.46519685e-01 1.86335325e-01 -7.43248940e-01
3.26820105e-01 8.67292047e-01 1.09746146e+00 7.93141782e-01
-9.11800742e-01 -6.93955958e-01 -4.84893471e-01 -1.98245104e-02
-2.00957656e+00 -2.49742240e-01 6.78415596e-01 -2.25972265e-01
-1.03429344e-03 4.89683807e-01 1.10832095e+00 6.71373665e-01
5.63018084e-01 2.98244745e-01 8.91647637e-01 -5.86637735e-01
2.41574332e-01 1.19616151e-01 -2.33935550e-01 8.02381754e-01
4.82360944e-02 -1.38413042e-01 -5.48573792e-01 -2.46510282e-02
1.68282950e+00 3.03315103e-01 -3.88704181e-01 -1.39356136e-01
-7.60799646e-01 5.11187315e-01 8.67938221e-01 6.34800673e-01
-5.99992871e-01 2.05938250e-01 9.69776437e-02 -3.86958539e-01
7.82686099e-03 -3.89349125e-02 9.27857235e-02 -2.03111738e-01
-8.98915589e-01 -3.41864884e-01 4.38611031e-01 8.03875089e-01
4.07671899e-01 -3.07702452e-01 -2.38960072e-01 7.19598532e-01
3.70851755e-01 2.18700573e-01 3.73652846e-01 -9.72683191e-01
-1.09552547e-01 6.63752496e-01 2.43242308e-01 -1.08073342e+00
-2.99726099e-01 -1.25289559e-01 -7.61407793e-01 1.70673296e-01
4.27321613e-01 -2.87501127e-01 -7.68247783e-01 1.20673108e+00
8.82001340e-01 3.18625242e-01 -6.87897384e-01 1.37653983e+00
1.03754425e+00 2.87218839e-01 1.12232268e-01 -9.75634456e-02
1.83131516e+00 -7.45292842e-01 -1.08405924e+00 6.69463947e-02
5.74588120e-01 -9.88762438e-01 1.46812809e+00 3.18437070e-01
-1.08682024e+00 -3.99585277e-01 -4.94353265e-01 -9.90168825e-02
-6.15146384e-02 3.49957049e-01 6.92728698e-01 1.12015057e+00
-9.73285794e-01 4.44129407e-01 -8.07475328e-01 -7.14708984e-01
3.30843478e-01 5.45965850e-01 -5.20358801e-01 1.68968439e-01
-1.05740595e+00 1.07734823e+00 -2.72282302e-01 7.63832510e-01
-2.64706701e-01 -7.34713972e-01 -6.11165941e-01 4.86489870e-02
1.05616875e-01 -7.26297379e-01 1.12317085e+00 -8.26710522e-01
-1.90757215e+00 1.01535594e+00 -2.06632271e-01 2.31578812e-01
2.87715435e-01 -4.07057703e-01 -3.80654544e-01 4.84298378e-01
2.58328080e-01 6.32013857e-01 4.06962782e-01 -1.19425476e+00
-3.10900688e-01 -1.02420628e+00 1.44298136e-01 6.00093067e-01
1.14788771e-01 1.01933189e-01 -6.90377176e-01 -2.66539454e-01
5.90372384e-01 -7.00066328e-01 -3.79414648e-01 4.50148284e-01
-4.61328357e-01 -7.35609280e-03 7.87290156e-01 -9.97666895e-01
1.02826321e+00 -2.22486067e+00 -1.11863337e-01 6.01139486e-01
-8.65117684e-02 -1.43680587e-01 4.34916258e-01 3.26696128e-01
-1.48959562e-01 2.25646242e-01 2.91992158e-01 1.11261398e-01
-2.36407235e-01 1.09900022e-02 3.27278376e-01 6.29671514e-01
-7.74909675e-01 7.39426374e-01 -6.77007973e-01 -9.40522313e-01
3.15362155e-01 7.40260899e-01 -1.49879783e-01 -1.92275554e-01
5.85667133e-01 7.68455148e-01 -6.33180380e-01 1.16227984e+00
8.52356791e-01 1.90127060e-01 1.16729356e-01 -4.35979754e-01
-1.78767800e-01 -3.59226942e-01 -1.12125432e+00 2.29463935e+00
-4.84694898e-01 -1.77272469e-01 4.19035017e-01 -3.66948396e-01
8.65592301e-01 7.98669755e-01 7.93316185e-01 -5.42834580e-01
3.05356354e-01 1.08160488e-01 -4.51069683e-01 -6.90693378e-01
2.22233310e-01 -4.97762352e-01 2.33643413e-01 5.56213632e-02
-3.13442379e-01 -6.90791905e-02 -3.03104430e-01 -1.23200797e-01
3.61704826e-01 5.01756966e-01 3.82154346e-01 -1.73425630e-01
1.92966342e-01 -4.07049321e-02 1.54257789e-01 1.77354842e-01
-2.10720509e-01 6.83080733e-01 2.49430776e-01 -3.72124314e-01
-4.02870446e-01 -1.05046940e+00 -4.64981437e-01 6.38309181e-01
4.81008440e-01 -3.76624793e-01 -1.09265804e+00 -4.17819917e-01
-2.49522805e-01 3.30716878e-01 -6.66011930e-01 -1.35355489e-03
-2.36481824e-03 -2.89363023e-02 3.72559696e-01 5.25166333e-01
5.30864120e-01 -8.73179674e-01 -8.91437054e-01 -2.04558074e-02
-2.82019943e-01 -4.74311113e-01 -7.98629999e-01 -4.38502610e-01
-1.02584851e+00 -6.09034240e-01 -1.43381000e+00 -7.54066348e-01
9.64517891e-01 2.46232212e-01 6.06615126e-01 2.11681634e-01
-5.53301752e-01 4.78101909e-01 -2.55542815e-01 -4.66901153e-01
5.00121117e-01 -1.67423725e-01 -1.61140710e-01 -1.75443053e-01
5.30864060e-01 -5.01929939e-01 -9.83428299e-01 5.54535210e-01
-3.63054782e-01 -7.14846030e-02 2.97669888e-01 2.48300329e-01
7.66952455e-01 -1.49072096e-01 4.05514091e-01 -5.92130899e-01
5.41727602e-01 -9.83634740e-02 -3.32345456e-01 2.67320096e-01
1.04137398e-01 -8.15214336e-01 -4.96370159e-02 -2.03711912e-01
-8.42151523e-01 3.50756824e-01 -2.87204891e-01 -2.90694535e-01
-3.52902681e-01 5.11228502e-01 -1.02831349e-01 -6.01457775e-01
8.48116755e-01 -2.95498043e-01 2.09061608e-01 -2.84811884e-01
3.14845681e-01 9.57506299e-01 6.76776350e-01 -3.69422942e-01
3.05574596e-01 3.63488734e-01 3.88525017e-02 -6.65281177e-01
-7.53202558e-01 -5.88152468e-01 -8.98326516e-01 -6.76824152e-01
8.51725399e-01 -7.24334717e-01 -1.22855341e+00 -4.08184201e-01
-9.06938970e-01 1.17351413e-01 -1.40953615e-01 7.78286278e-01
-2.32710540e-01 4.11063135e-01 -7.05608666e-01 -8.82875264e-01
-5.98319948e-01 -7.96184599e-01 1.38250387e+00 1.03434861e+00
-4.21548069e-01 -7.94623494e-01 -1.13687679e-01 3.74589026e-01
2.42596790e-01 3.34499538e-01 4.94588405e-01 3.42159986e-01
-1.20701425e-01 -6.70572698e-01 2.82700676e-02 -3.91046971e-01
4.70450819e-01 2.25493684e-01 -8.96645904e-01 2.34876975e-01
3.08702160e-02 3.91072109e-02 -2.41307810e-01 9.31753159e-01
1.28782237e+00 1.68301519e-02 -6.10532463e-01 3.82543594e-01
1.43888164e+00 3.47111315e-01 1.09480619e+00 -7.92803895e-03
5.41862965e-01 6.36594713e-01 8.30065310e-01 1.57885149e-01
1.96933001e-01 8.01446855e-01 4.95135367e-01 -7.14305103e-01
2.32728109e-01 -2.27808729e-01 -1.31200239e-01 4.34158564e-01
-8.97622228e-01 5.71745992e-01 -8.43689620e-01 2.76498079e-01
-1.30442774e+00 -7.61359274e-01 -5.67731082e-01 2.66692281e+00
9.41463590e-01 -4.82099950e-01 3.74767274e-01 -5.83266700e-03
1.08326304e+00 -9.86094296e-01 -1.05882302e-01 -4.04908627e-01
6.99455559e-01 5.01881003e-01 3.76485735e-01 3.00966650e-01
-6.96151793e-01 6.68158591e-01 6.74042702e+00 5.39740562e-01
-1.47760916e+00 1.79560125e-01 3.78704399e-01 5.60180135e-02
-4.55521308e-02 -6.39459193e-02 -5.04378021e-01 3.55089873e-01
3.98716807e-01 9.29162502e-02 3.53807211e-01 7.87044883e-01
3.35107028e-01 -4.15804118e-01 -7.51110673e-01 1.21624160e+00
-8.91504958e-02 -9.97343242e-01 -5.16732574e-01 1.09042399e-01
2.25014254e-01 -7.04501867e-01 3.02193034e-02 -1.38424501e-01
-4.98682350e-01 -1.17890656e+00 2.74522036e-01 6.77274406e-01
1.11229730e+00 -6.86049402e-01 7.85628498e-01 2.52570957e-01
-9.97227132e-01 3.43934208e-01 -4.47806716e-01 1.51351273e-01
4.30109352e-01 2.92864114e-01 -7.11720288e-01 7.01447308e-01
7.64875054e-01 2.95442432e-01 -1.19176835e-01 1.46059692e+00
-2.23973840e-01 2.44942650e-01 -5.33234417e-01 6.82828203e-02
-1.12732269e-01 -7.31026113e-01 -1.10791512e-01 4.64269161e-01
7.75139928e-01 4.08969343e-01 -1.43065780e-01 8.08046937e-01
2.52642661e-01 6.50109768e-01 -5.51956713e-01 2.34269202e-01
3.44304919e-01 1.79413438e+00 -1.09990251e+00 1.66183636e-01
-1.36538789e-01 1.00972867e+00 -4.97225553e-01 3.09191078e-01
-1.01435769e+00 -2.50914693e-01 -1.66805014e-01 5.30510306e-01
-3.94190937e-01 6.20662346e-02 -7.31026113e-01 -7.47124076e-01
-2.61360258e-01 -7.10500717e-01 2.89536655e-01 -1.18454564e+00
-9.54452753e-01 2.34095857e-01 -3.48742902e-01 -1.44941342e+00
2.54473478e-01 -1.67121068e-01 -7.35556185e-01 1.09415460e+00
-6.40269816e-01 -1.38150024e+00 -7.67870128e-01 6.42635345e-01
1.42580852e-01 6.24793530e-01 1.23932767e+00 4.89074081e-01
-4.93028075e-01 4.59039271e-01 -3.59929323e-01 -8.73622373e-02
7.41105437e-01 -9.54904318e-01 -4.80401814e-01 1.95404753e-01
1.58350691e-01 1.01925504e+00 5.93237877e-01 -7.78057933e-01
-1.25309479e+00 -4.20180738e-01 5.54147363e-01 -2.60929137e-01
-4.39329110e-02 -1.77542362e-02 -6.28843546e-01 7.44591713e-01
3.29427034e-01 -1.37770101e-01 1.12346828e+00 3.92458588e-01
4.77209628e-01 -2.46136278e-01 -1.68943393e+00 8.83927643e-01
7.96586990e-01 -4.18198615e-01 -3.18525344e-01 5.37302554e-01
-1.49622053e-01 -1.01384866e+00 -1.03544939e+00 1.81266591e-02
1.09888947e+00 -7.55624652e-01 9.66794848e-01 -9.61723085e-03
3.30641657e-01 -4.11325634e-01 4.54753518e-01 -1.04514503e+00
-2.06311762e-01 -4.80541378e-01 6.08371437e-01 1.06273425e+00
2.74921767e-02 -4.32720602e-01 9.39214170e-01 8.24038804e-01
4.85246349e-03 -5.83819568e-01 -1.08361781e+00 -2.02627644e-01
-4.52921480e-01 2.74027973e-01 5.72810173e-01 1.05004549e+00
8.79449308e-01 3.22099090e-01 -1.59264863e-01 5.26052594e-01
3.55710179e-01 1.22476116e-01 6.67741895e-01 -1.16127026e+00
2.23802522e-01 -1.17933713e-01 -4.63251323e-01 -6.52081251e-01
-5.26443183e-01 -5.36145508e-01 -2.13442087e-01 -1.91825175e+00
-1.88259050e-01 -4.06859219e-01 1.11577451e-01 7.17049301e-01
2.54875958e-01 5.28638661e-01 -1.23751491e-01 -3.81363817e-02
1.25662327e-01 2.68818229e-01 1.79409885e+00 4.34906334e-01
-5.02375782e-01 2.39332736e-01 -8.76874506e-01 8.78771722e-01
8.64818633e-01 -4.21562701e-01 -3.83346170e-01 -3.20801169e-01
-2.71214955e-02 5.14293671e-01 3.60628098e-01 -8.61420155e-01
3.64028394e-01 -2.28837937e-01 6.90587282e-01 -7.46851981e-01
5.06689012e-01 -1.20841706e+00 6.10899687e-01 6.52391493e-01
2.75042146e-01 2.12749578e-02 9.04313102e-02 -6.74187690e-02
4.26372327e-02 -5.70687711e-01 5.94183266e-01 -4.04479772e-01
-1.22678846e-01 7.93942958e-02 -1.60529271e-01 -8.69298518e-01
1.42503881e+00 -7.00394094e-01 -1.35886827e-02 -7.30782092e-01
-1.15095460e+00 1.43391073e-01 4.56074327e-01 -2.89871484e-01
6.57440245e-01 -1.38253701e+00 -2.41159931e-01 2.12580055e-01
-2.07130119e-01 7.05839545e-02 7.66545296e-01 1.50638700e+00
-1.07241833e+00 5.56349717e-02 -6.35320783e-01 -7.50107348e-01
-1.28030264e+00 3.43110234e-01 7.44565129e-01 5.22007406e-01
-6.96050107e-01 8.44262958e-01 -6.14873171e-02 -2.38754064e-01
1.06957987e-01 -1.58650056e-01 -5.51532865e-01 -1.36997595e-01
4.29389685e-01 3.42241466e-01 -1.82999857e-02 -6.10618532e-01
-4.16699767e-01 1.17462695e+00 4.90902245e-01 -1.63441747e-01
6.95451558e-01 -2.32819706e-01 -1.88614517e-01 3.01819205e-01
3.55108082e-01 5.10583580e-01 -5.92587590e-01 3.21065664e-01
-5.19384980e-01 -9.41197395e-01 5.74340299e-02 -8.80115986e-01
-1.21269667e+00 1.06670487e+00 1.02652586e+00 1.46811396e-01
1.15324831e+00 -2.27152959e-01 5.44410884e-01 -3.63389105e-01
8.76476884e-01 -1.06027138e+00 -3.71891707e-01 -3.57686818e-01
9.84289527e-01 -5.83177686e-01 -2.07379192e-01 -1.11489034e+00
-7.43251383e-01 1.19135118e+00 7.49871671e-01 7.79866055e-02
7.59099066e-01 2.26038128e-01 4.19593841e-01 -4.82334942e-01
2.69810170e-01 2.24516019e-02 2.48405293e-01 7.23245919e-01
8.59467804e-01 1.04984730e-01 -9.56322789e-01 5.19074321e-01
-3.82606089e-01 6.92215681e-01 2.51158953e-01 8.18191767e-01
-2.59295970e-01 -8.66533875e-01 -7.85792112e-01 2.31387794e-01
-6.48905277e-01 1.92702577e-01 -1.96059063e-01 7.37743735e-01
3.16930592e-01 7.61065364e-01 7.41381496e-02 -3.20854858e-02
5.11391819e-01 -2.28045791e-01 7.91852951e-01 -6.40951514e-01
-6.83116555e-01 6.95541680e-01 -2.54521877e-01 -5.15140891e-01
-3.03531021e-01 -2.20780671e-01 -1.74621439e+00 -4.09587592e-01
-4.08730388e-01 3.89368027e-01 9.31669593e-01 6.68268621e-01
1.84377581e-01 2.48140991e-01 3.68591547e-01 -8.08061481e-01
-1.93318147e-02 -1.04583907e+00 -1.19986999e+00 -2.30898485e-02
-3.92320931e-01 -4.98649091e-01 2.74007499e-01 4.12290581e-02] | [14.232036590576172, -2.7626099586486816] |
4c6710b4-5eb1-489f-bed9-b459b4532f3b | push-the-boundary-boundary-aware-feature | 2212.12402 | null | https://arxiv.org/abs/2212.12402v1 | https://arxiv.org/pdf/2212.12402v1.pdf | Push-the-Boundary: Boundary-aware Feature Propagation for Semantic Segmentation of 3D Point Clouds | Feedforward fully convolutional neural networks currently dominate in semantic segmentation of 3D point clouds. Despite their great success, they suffer from the loss of local information at low-level layers, posing significant challenges to accurate scene segmentation and precise object boundary delineation. Prior works either address this issue by post-processing or jointly learn object boundaries to implicitly improve feature encoding of the networks. These approaches often require additional modules which are difficult to integrate into the original architecture. To improve the segmentation near object boundaries, we propose a boundary-aware feature propagation mechanism. This mechanism is achieved by exploiting a multi-task learning framework that aims to explicitly guide the boundaries to their original locations. With one shared encoder, our network outputs (i) boundary localization, (ii) prediction of directions pointing to the object's interior, and (iii) semantic segmentation, in three parallel streams. The predicted boundaries and directions are fused to propagate the learned features to refine the segmentation. We conduct extensive experiments on the S3DIS and SensatUrban datasets against various baseline methods, demonstrating that our proposed approach yields consistent improvements by reducing boundary errors. Our code is available at https://github.com/shenglandu/PushBoundary. | ['Liangliang Nan', 'Julian Kooij', 'Jantien Stoter', 'Nail Ibrahimli', 'Shenglan Du'] | 2022-12-23 | null | null | null | null | ['scene-segmentation'] | ['computer-vision'] | [ 3.82564873e-01 6.68646470e-02 -6.39169523e-03 -7.72661448e-01
-6.41648352e-01 -6.50234759e-01 4.48401749e-01 1.50854826e-01
-2.99863309e-01 3.44641298e-01 -1.63766503e-01 -2.17209622e-01
9.05431807e-02 -7.99106061e-01 -1.09920716e+00 -5.26685834e-01
4.27066647e-02 6.23016417e-01 5.84535837e-01 8.48305747e-02
4.20052826e-01 8.07739615e-01 -1.39229405e+00 2.45777592e-01
9.79201376e-01 1.33396232e+00 3.20594281e-01 6.35794640e-01
-4.57112104e-01 4.16297138e-01 -2.62335747e-01 9.12127346e-02
5.33570230e-01 4.18273658e-02 -9.05661225e-01 2.43960574e-01
4.30343002e-01 -4.65507478e-01 3.68093178e-02 9.95006144e-01
1.38462752e-01 4.55513247e-04 5.98742604e-01 -1.00218797e+00
-3.95512223e-01 4.06637788e-01 -7.75528848e-01 -1.00446597e-01
-1.63856864e-01 1.37822747e-01 1.01056015e+00 -9.28718626e-01
3.89430225e-01 1.03510618e+00 7.96397746e-01 3.19252819e-01
-1.07751596e+00 -7.13893652e-01 4.86154526e-01 -1.93153158e-01
-1.38254607e+00 -3.58347684e-01 1.02114606e+00 -6.72343671e-01
8.01963508e-01 -4.48092408e-02 4.56931442e-01 4.93510306e-01
-2.21067652e-01 8.55776668e-01 6.78043485e-01 -6.41284436e-02
1.48951679e-01 -2.46456470e-02 1.97001085e-01 5.78854501e-01
6.50354773e-02 -5.51452711e-02 -2.13624969e-01 2.60144591e-01
1.06758785e+00 7.64380991e-02 -2.11279422e-01 -5.07414937e-01
-9.03170884e-01 7.04683006e-01 1.02662718e+00 9.39852223e-02
-4.52885002e-01 3.18671107e-01 8.86865333e-02 -1.50607035e-01
6.71701193e-01 3.78266573e-01 -7.30512381e-01 2.90306032e-01
-9.86379385e-01 4.22692150e-01 4.92697626e-01 9.41820443e-01
1.21823525e+00 -3.82001549e-01 -8.49045217e-02 7.83165991e-01
5.73733747e-01 2.91566551e-01 -1.59628138e-01 -1.11136281e+00
5.71074963e-01 9.03096497e-01 2.71873295e-01 -7.89695561e-01
-5.90373456e-01 -4.79577303e-01 -4.50016618e-01 2.93270469e-01
3.39536279e-01 -1.52820468e-01 -1.45291162e+00 1.31971776e+00
6.23747051e-01 3.89037579e-01 -3.32420349e-01 1.11022055e+00
7.33637154e-01 7.45887220e-01 3.03536039e-02 5.57440162e-01
1.01595128e+00 -1.13547003e+00 -1.18946850e-01 -5.30682325e-01
5.68379223e-01 -6.98110044e-01 7.21198976e-01 9.76077318e-02
-1.06758082e+00 -5.14342010e-01 -8.96786034e-01 -3.79065275e-01
-2.00364053e-01 3.00171643e-01 6.76056564e-01 1.55269831e-01
-1.15037870e+00 6.70597374e-01 -1.01299727e+00 -1.14494031e-02
1.04449987e+00 5.12398422e-01 -2.38586515e-02 1.03697553e-01
-7.81373203e-01 4.68272626e-01 4.18004483e-01 4.17180628e-01
-7.54473209e-01 -9.05282259e-01 -9.35085773e-01 1.00790553e-01
3.98663342e-01 -6.83073699e-01 1.49323475e+00 -9.24572349e-01
-1.40571654e+00 8.72979403e-01 -2.49415845e-01 -3.92863065e-01
5.21986663e-01 -4.62819219e-01 4.17336047e-01 1.58940569e-01
1.19938582e-01 1.15728462e+00 7.07499564e-01 -1.58732462e+00
-9.47342873e-01 -4.58191544e-01 1.79677233e-02 3.00767422e-01
8.46194476e-02 -3.15919399e-01 -7.58400321e-01 -4.71710294e-01
5.25891602e-01 -6.73347592e-01 -5.73791742e-01 3.96319240e-01
-6.30246818e-01 -3.60690892e-01 9.99336064e-01 -5.06125093e-01
7.74411082e-01 -2.10827589e+00 7.65333399e-02 1.85018271e-01
2.31640294e-01 2.57679909e-01 -1.73255261e-02 -4.66733240e-02
2.06556052e-01 1.80850178e-01 -5.92279613e-01 -6.27174616e-01
-8.96194726e-02 9.45417359e-02 -2.20153287e-01 4.13557082e-01
5.26661277e-01 1.05438864e+00 -6.44143105e-01 -3.49629819e-01
5.41254282e-01 6.62668347e-01 -7.30470836e-01 2.08808109e-01
-6.36455715e-01 6.84920013e-01 -7.52992332e-01 6.66008592e-01
8.83837640e-01 -4.87401217e-01 -3.29288840e-01 -1.40678644e-01
-3.41750115e-01 4.65745866e-01 -1.14489412e+00 1.77844310e+00
-3.39122862e-01 4.43202019e-01 4.13363636e-01 -1.08682811e+00
1.05332673e+00 -3.74409966e-02 5.48203230e-01 -4.53260928e-01
2.29789227e-01 2.80693561e-01 -2.88673252e-01 -3.32854539e-01
4.37257707e-01 -4.10196967e-02 1.12819009e-01 1.78697303e-01
-1.21600516e-01 -3.63042384e-01 -6.57433271e-02 -5.16522340e-02
7.41851330e-01 5.15505910e-01 -1.57979220e-01 -7.04499707e-02
4.56296504e-01 1.60699353e-01 7.36028969e-01 5.42702973e-01
-2.82713529e-02 9.21776891e-01 4.16230172e-01 -4.47254092e-01
-1.06645322e+00 -9.71990943e-01 -3.12072188e-01 7.56413996e-01
5.41619539e-01 5.67714982e-02 -7.75044501e-01 -6.84342742e-01
2.80012369e-01 5.35317123e-01 -4.67727870e-01 1.28123268e-01
-7.20714927e-01 -2.95414686e-01 2.55173683e-01 8.62809598e-01
6.52366817e-01 -8.95848691e-01 -6.83003366e-01 1.73057869e-01
2.11152248e-02 -1.24284279e+00 -3.51845503e-01 3.55134547e-01
-9.86635923e-01 -1.01048231e+00 -7.31388688e-01 -8.36338699e-01
8.97498012e-01 2.53270060e-01 7.83056259e-01 1.85808465e-01
-9.64504480e-02 -7.05074370e-02 -2.17393667e-01 -3.85476530e-01
-3.55181806e-02 3.75617683e-01 -5.22439599e-01 -1.41682714e-01
2.91558743e-01 -4.55111086e-01 -7.86660492e-01 3.31398308e-01
-6.96789563e-01 4.59619790e-01 5.33984780e-01 5.44647813e-01
8.09162319e-01 -2.02409372e-01 3.22414696e-01 -9.30090070e-01
-2.85379253e-02 -4.57512796e-01 -1.00297296e+00 -1.27382264e-01
-3.91425163e-01 -8.13443661e-02 3.56963009e-01 1.21438578e-01
-1.00201941e+00 4.41257626e-01 -4.70128119e-01 -5.71872771e-01
-5.82211971e-01 2.56517142e-01 -1.57726154e-01 -1.69418171e-01
2.95842826e-01 -3.69554944e-02 -2.71519572e-01 -6.81891799e-01
3.04738611e-01 6.06794536e-01 4.15022701e-01 -4.38238978e-01
7.32348561e-01 5.88132024e-01 -2.26664066e-01 -7.05734074e-01
-1.10631013e+00 -6.50726318e-01 -8.96762967e-01 -1.19666561e-01
9.59696591e-01 -9.58431065e-01 -5.10577321e-01 6.84585094e-01
-1.31162822e+00 -7.90282488e-01 -2.50938922e-01 1.69246927e-01
-4.13766026e-01 -6.25076815e-02 -5.95311999e-01 -4.99950320e-01
-3.65479082e-01 -1.17703891e+00 1.51044667e+00 5.82998693e-01
5.85129075e-02 -8.90811503e-01 -2.28750065e-01 4.53320771e-01
3.83527249e-01 1.95048049e-01 7.63525784e-01 -4.20284927e-01
-9.96620536e-01 1.68513432e-02 -6.03340983e-01 3.35981101e-01
1.07942685e-01 6.79643229e-02 -1.12099648e+00 2.44646035e-02
-2.73990065e-01 -1.71229348e-01 1.10682809e+00 7.37784326e-01
1.53922725e+00 4.50462736e-02 -6.37611687e-01 1.08394516e+00
1.29259169e+00 2.43194979e-02 2.64561683e-01 2.76339054e-01
9.22646761e-01 5.18142700e-01 4.83435720e-01 3.64092499e-01
6.00877285e-01 4.50756222e-01 7.16658473e-01 -2.46107638e-01
-2.72887468e-01 -1.65425152e-01 -1.06501274e-01 3.50512683e-01
9.20506269e-02 -3.18912774e-01 -1.23443353e+00 7.08290517e-01
-1.75742722e+00 -3.32228661e-01 -2.48475358e-01 1.84538484e+00
6.97544694e-01 3.46807033e-01 -1.39602542e-01 -1.17099836e-01
6.50452375e-01 2.04102531e-01 -8.57274711e-01 -3.00098836e-01
1.46047577e-01 1.34710714e-01 5.58814943e-01 7.33997464e-01
-1.43466187e+00 1.34492707e+00 5.14944172e+00 3.97155136e-01
-1.42768443e+00 -5.51192947e-02 7.57384658e-01 -3.26361582e-02
-3.48487347e-01 7.37129152e-02 -9.96629000e-01 3.03732514e-01
4.77782965e-01 2.74354994e-01 1.55721188e-01 7.48447716e-01
1.80621684e-01 -2.15831727e-01 -9.58241403e-01 6.64635897e-01
-2.87527442e-01 -1.51238370e+00 -5.82586378e-02 -1.72405750e-01
9.04563487e-01 5.91122985e-01 -1.92817733e-01 7.95814916e-02
2.64032185e-01 -9.58736539e-01 8.99994493e-01 4.34771985e-01
5.71859121e-01 -6.22097313e-01 5.98806560e-01 4.11582589e-01
-1.28995252e+00 3.81554477e-02 -2.39864811e-01 -1.92518398e-01
3.55829328e-01 7.85921633e-01 -9.94305253e-01 3.58016461e-01
6.73983514e-01 8.81468654e-01 -3.53608251e-01 1.17541599e+00
-5.24832010e-01 5.54172158e-01 -6.20670855e-01 2.94764936e-01
4.94786859e-01 -8.80516395e-02 4.77958977e-01 9.15298045e-01
1.59941003e-01 8.74248743e-02 3.10958028e-01 1.39023137e+00
-2.00166658e-01 -1.50996879e-01 -3.34537029e-01 1.84690952e-02
4.62480128e-01 1.24447584e+00 -1.15033162e+00 -1.87130064e-01
-3.89952958e-01 8.88964415e-01 3.49716634e-01 4.73228365e-01
-8.31089258e-01 -4.19893950e-01 8.60495448e-01 2.71586716e-01
5.61854064e-01 -4.10165429e-01 -9.02666867e-01 -7.88743317e-01
8.04802179e-02 -9.81586576e-02 2.31721275e-03 -5.78808486e-01
-9.45344508e-01 3.30675870e-01 -2.40340218e-01 -7.50452340e-01
1.79755017e-01 -5.13236821e-01 -5.98476589e-01 9.23321664e-01
-1.86770761e+00 -1.30087221e+00 -5.00278890e-01 2.38885894e-01
6.27697229e-01 5.47837257e-01 3.01425397e-01 4.40778106e-01
-5.25702238e-01 2.51032233e-01 -2.39513755e-01 2.30700836e-01
3.83473456e-01 -1.26581728e+00 6.87175512e-01 7.96286464e-01
-1.11135095e-01 2.35218823e-01 2.63350427e-01 -8.38293076e-01
-8.97084892e-01 -1.45003688e+00 7.41687953e-01 -2.20464960e-01
2.47129276e-01 -5.22305965e-01 -1.13303077e+00 8.08449984e-01
-2.12750942e-01 2.20110103e-01 8.24308842e-02 -1.21262595e-02
-6.35858672e-03 2.63193771e-02 -9.35818076e-01 3.30964953e-01
1.10840321e+00 -1.90880358e-01 -3.95477772e-01 1.50792137e-01
8.35691810e-01 -6.99913681e-01 -6.28720045e-01 6.41112030e-01
2.76823759e-01 -8.71599913e-01 9.11062479e-01 -1.59132436e-01
5.84420085e-01 -5.47152519e-01 1.54166952e-01 -1.09642386e+00
-1.72432676e-01 -1.65293932e-01 2.14596391e-01 1.14938903e+00
5.48375070e-01 -5.56650400e-01 1.05101740e+00 6.63077652e-01
-6.63091779e-01 -9.83520031e-01 -7.79779434e-01 -3.01683754e-01
2.13437587e-01 -5.62428832e-01 6.65785372e-01 7.37137973e-01
-6.12890840e-01 2.13221878e-01 1.51146561e-01 5.37663102e-01
5.10123372e-01 4.89046842e-01 7.50320017e-01 -1.23285389e+00
2.07331553e-01 -6.74210668e-01 -2.23639652e-01 -1.64241862e+00
6.80462494e-02 -1.07112646e+00 5.33438265e-01 -1.96824932e+00
-8.78291726e-02 -1.00657129e+00 -2.94729210e-02 8.14735770e-01
-1.48913309e-01 2.69106030e-01 -1.72438323e-02 1.35195449e-01
-5.53020298e-01 6.51403546e-01 1.38847649e+00 1.04999952e-02
-4.15100276e-01 8.38222951e-02 -5.82746625e-01 1.02550983e+00
9.24380541e-01 -4.28499252e-01 -2.45134950e-01 -9.96837318e-01
1.34123966e-01 -2.07900226e-01 6.54084444e-01 -1.04166639e+00
3.73862535e-01 -1.26611412e-01 5.13036430e-01 -9.25970078e-01
2.98018128e-01 -8.13588500e-01 -2.24868402e-01 2.58344620e-01
-2.34335631e-01 -5.41056693e-01 3.04809749e-01 3.94177437e-01
-1.73354074e-01 -1.27800465e-01 8.31268549e-01 -8.33767205e-02
-8.27714860e-01 6.30595028e-01 1.52408974e-02 -9.46320817e-02
1.12591136e+00 -3.14057112e-01 -8.49158838e-02 1.00854225e-01
-6.50544584e-01 8.38616371e-01 7.24989057e-01 4.35539871e-01
6.22240782e-01 -8.21411014e-01 -5.41728556e-01 5.07714093e-01
-1.71843335e-01 1.18025863e+00 1.94523320e-01 7.85072267e-01
-7.64316440e-01 3.93451959e-01 8.27877000e-02 -1.10828876e+00
-7.03654945e-01 8.16233363e-03 6.73204660e-01 1.48712933e-01
-8.84956360e-01 1.32948387e+00 4.88349050e-01 -6.70323789e-01
2.44687900e-01 -6.74258709e-01 -6.71736896e-02 -4.47708108e-02
1.78926140e-01 1.09210975e-01 2.38440596e-02 -5.43508112e-01
-4.22433019e-01 8.21760833e-01 -1.89455822e-01 2.82843441e-01
1.56029856e+00 -2.08895668e-01 -3.47019359e-02 2.63603270e-01
1.07866013e+00 -3.67754996e-01 -1.95863998e+00 -2.25352213e-01
1.87831134e-01 -4.13015425e-01 2.73704350e-01 -7.33599603e-01
-1.34186709e+00 1.08476150e+00 3.72228473e-01 -1.50174156e-01
1.00081146e+00 2.59952188e-01 1.07783639e+00 2.60978397e-02
3.03600580e-01 -8.34563315e-01 -2.07124025e-01 7.12307394e-01
7.29678929e-01 -1.40195894e+00 -1.75650567e-01 -7.20334888e-01
-2.56603509e-01 1.01108348e+00 8.56422901e-01 -3.26188624e-01
7.50507891e-01 5.05518854e-01 1.18021019e-01 -3.25959653e-01
-5.16212523e-01 -2.79629827e-01 2.57205099e-01 2.65467018e-01
3.39291185e-01 -1.93757061e-02 1.00097075e-01 5.82632422e-01
6.51896223e-02 8.21667239e-02 8.12129155e-02 9.33712840e-01
-6.51692212e-01 -8.54293585e-01 -3.42421442e-01 4.93873030e-01
-2.37334311e-01 1.37239933e-01 -3.52346569e-01 5.63863754e-01
3.30291778e-01 5.65107048e-01 4.41440850e-01 -1.44127861e-01
3.14530283e-01 -6.24883920e-02 2.64917254e-01 -8.90814185e-01
-4.34413701e-01 1.88575923e-01 -1.86107635e-01 -7.37171471e-01
-2.11533949e-01 -7.29147136e-01 -1.84253466e+00 1.95038766e-01
-3.44446898e-01 -7.38318369e-04 7.73107648e-01 9.33756590e-01
5.19754887e-01 6.84101164e-01 5.86957872e-01 -1.35574341e+00
-2.70354748e-01 -7.94519961e-01 -2.74451524e-01 1.74455881e-01
4.57678199e-01 -7.36071467e-01 -1.61973849e-01 1.12324737e-01] | [8.092015266418457, -3.040123462677002] |
77ece2a3-3a82-46a6-b188-e52b74393549 | cas-net-conditional-atlas-generation-and | 2205.08239 | null | https://arxiv.org/abs/2205.08239v1 | https://arxiv.org/pdf/2205.08239v1.pdf | CAS-Net: Conditional Atlas Generation and Brain Segmentation for Fetal MRI | Fetal Magnetic Resonance Imaging (MRI) is used in prenatal diagnosis and to assess early brain development. Accurate segmentation of the different brain tissues is a vital step in several brain analysis tasks, such as cortical surface reconstruction and tissue thickness measurements. Fetal MRI scans, however, are prone to motion artifacts that can affect the correctness of both manual and automatic segmentation techniques. In this paper, we propose a novel network structure that can simultaneously generate conditional atlases and predict brain tissue segmentation, called CAS-Net. The conditional atlases provide anatomical priors that can constrain the segmentation connectivity, despite the heterogeneity of intensity values caused by motion or partial volume effects. The proposed method is trained and evaluated on 253 subjects from the developing Human Connectome Project (dHCP). The results demonstrate that the proposed method can generate conditional age-specific atlas with sharp boundary and shape variance. It also segment multi-category brain tissues for fetal MRI with a high overall Dice similarity coefficient (DSC) of $85.2\%$ for the selected 9 tissue labels. | ['Amir Alansary', 'Daniel Rueckert', 'Bernhard Kainz', 'A. David Edwards', 'Joseph Hajnal', 'Antonios Makropoulos', 'Matthew Sinclair', 'Qiang Ma', 'Liu Li'] | 2022-05-17 | null | null | null | null | ['brain-segmentation'] | ['medical'] | [ 1.97566211e-01 2.98343390e-01 1.26177743e-01 -8.41306806e-01
-2.35543177e-01 -4.78851467e-01 3.81201580e-02 1.71358176e-02
-1.96303353e-01 5.78408957e-01 -1.43755883e-01 -1.48430675e-01
-1.06215820e-01 -5.21541834e-01 -4.71041828e-01 -6.11289263e-01
-3.76054317e-01 7.76217520e-01 2.68098086e-01 5.52787662e-01
2.22309485e-01 6.77393496e-01 -9.28374529e-01 -2.49920115e-02
1.32754624e+00 7.07934737e-01 1.76154166e-01 3.36456746e-01
-2.09633455e-01 1.75406680e-01 -2.50691205e-01 -6.00341201e-01
3.07656735e-01 -4.26085413e-01 -7.81591773e-01 -1.34461343e-01
4.30417717e-01 -1.78540245e-01 1.73299491e-01 1.32612848e+00
5.24352431e-01 -1.01067424e-02 1.14601707e+00 -8.60835195e-01
-1.96301281e-01 9.41590548e-01 -1.02396834e+00 4.47742909e-01
-2.51696169e-01 -1.36525288e-01 3.49746734e-01 -7.68168807e-01
6.41303003e-01 7.94712782e-01 7.02226222e-01 7.81140268e-01
-1.15782320e+00 -1.14250481e+00 -1.35511056e-01 -9.93095189e-02
-1.48560512e+00 -4.06321317e-01 6.42054021e-01 -1.10786426e+00
4.85126168e-01 5.97533174e-02 8.30085337e-01 4.91176814e-01
4.58732575e-01 2.16634959e-01 1.16585422e+00 4.12845798e-03
1.60638958e-01 -4.71036434e-01 -1.84422314e-01 8.93683553e-01
4.25045341e-01 -3.62744689e-01 -2.47664358e-02 -1.85598377e-02
1.12542033e+00 -4.12718326e-01 -1.04311861e-01 -2.73044527e-01
-1.20178866e+00 5.34304857e-01 1.33874029e-01 4.59005296e-01
-5.09806752e-01 -3.63117978e-02 1.86080515e-01 -2.68231541e-01
7.43146598e-01 1.44236594e-01 -1.48850530e-01 1.67520374e-01
-1.58300781e+00 -2.14869350e-01 4.04414296e-01 7.22910523e-01
4.23053890e-01 1.44551545e-01 -1.14581190e-01 1.14208353e+00
5.21615565e-01 3.95194501e-01 3.94583523e-01 -1.25970864e+00
3.87167484e-01 2.61728972e-01 -4.75521535e-01 -1.11394763e+00
-8.27867210e-01 -1.27868921e-01 -1.23064697e+00 5.45994230e-02
4.79409218e-01 -6.13461435e-01 -1.13230681e+00 1.76982176e+00
7.19150484e-01 2.26834044e-01 -6.69220388e-01 9.42757368e-01
8.77522051e-01 1.55217469e-01 3.44360858e-01 -4.40405667e-01
9.38384950e-01 -4.84573662e-01 -6.57019675e-01 -1.43512547e-01
2.75199473e-01 -5.23040414e-01 3.14497739e-01 1.06707020e-02
-1.38610351e+00 -9.60995406e-02 -8.80963385e-01 3.49596620e-01
2.96123743e-01 -7.00354055e-02 6.45121574e-01 8.36839080e-01
-1.17924953e+00 5.94367445e-01 -1.41954684e+00 -2.79921889e-02
8.54050994e-01 5.06930351e-01 -4.85931724e-01 3.62072140e-02
-8.42477441e-01 1.01633322e+00 1.97615609e-01 2.51005441e-01
-7.67205179e-01 -1.08619237e+00 -6.23829961e-01 9.36596002e-03
-2.19652504e-01 -3.68016541e-01 6.78695560e-01 -7.20709741e-01
-1.38064516e+00 1.04855776e+00 3.33441347e-02 -1.47896633e-01
6.52008653e-01 2.90123940e-01 -2.89250672e-01 6.51088774e-01
2.81296670e-01 1.01577580e+00 6.02309346e-01 -9.45562780e-01
4.11103703e-02 -5.57168961e-01 -7.07377136e-01 1.23499237e-01
1.17830992e-01 4.21376228e-01 -3.11011314e-01 -6.41974390e-01
8.17819834e-01 -8.63218546e-01 -4.76735890e-01 1.54361218e-01
-4.13537472e-01 2.47772753e-01 -2.20061280e-02 -1.35605454e+00
7.46915936e-01 -1.72431707e+00 -9.56582725e-02 8.27955246e-01
5.74796081e-01 -5.25363423e-02 1.32294163e-01 -4.26518947e-01
-3.35997880e-01 2.36242518e-01 -8.61751258e-01 -1.28466287e-03
-3.74658465e-01 -2.55805224e-01 6.53025329e-01 1.01478839e+00
-1.89925686e-01 7.53211915e-01 -6.12985730e-01 -9.05800760e-01
-3.37389857e-02 5.23571193e-01 -4.91693825e-01 1.43493474e-01
2.79827893e-01 9.27149415e-01 -3.45552802e-01 4.05122340e-01
9.03045356e-01 8.68926570e-02 4.70574886e-01 -4.80449721e-02
-9.31149907e-03 -1.80269480e-01 -7.77271807e-01 1.80750668e+00
3.49263381e-03 5.66828430e-01 5.14051616e-01 -1.04699564e+00
9.42804456e-01 6.62494242e-01 9.61296022e-01 -5.03709078e-01
2.96996891e-01 2.20479041e-01 7.37158537e-01 -5.32495856e-01
-4.31340665e-01 -3.18803042e-01 6.05324686e-01 5.32460630e-01
1.57020420e-01 -2.99262017e-01 3.82666588e-01 -5.03339209e-02
7.00650573e-01 -1.28636464e-01 -1.75713629e-01 -8.87270093e-01
3.31329703e-01 -5.36528349e-01 9.76926923e-01 9.21590328e-02
-2.62793571e-01 9.18498933e-01 4.70870495e-01 -1.90296158e-01
-1.10126173e+00 -1.08617091e+00 -4.97330636e-01 4.21029449e-01
-2.71301389e-01 3.61953259e-01 -1.32512748e+00 -6.63449228e-01
-1.83657393e-01 4.42256093e-01 -5.39017022e-01 1.37595206e-01
-6.38925970e-01 -1.02862096e+00 8.26014161e-01 4.87417370e-01
2.28864014e-01 -9.32548821e-01 -5.19332230e-01 1.71211004e-01
-2.51379132e-01 -1.02277958e+00 -7.04099238e-01 -2.90249169e-01
-1.12946677e+00 -1.18904555e+00 -1.19832921e+00 -8.11099291e-01
1.33367217e+00 -4.85046387e-01 8.87687981e-01 2.52007931e-01
-4.48884755e-01 3.18122469e-02 1.82707589e-02 -1.40177310e-01
-4.91268277e-01 -1.22657657e-01 1.52473435e-01 -3.01056821e-02
-1.63163260e-01 -1.03278136e+00 -9.96224940e-01 3.11554879e-01
-6.97912216e-01 2.91978210e-01 3.26925069e-01 3.77640277e-01
5.48841834e-01 -2.70676017e-01 6.24116600e-01 -9.56486046e-01
3.74704510e-01 -6.00366831e-01 -7.48762012e-01 2.84858137e-01
-6.49703085e-01 -3.47151101e-01 9.41932052e-02 -3.27971607e-01
-1.33592558e+00 -5.32839894e-02 -3.40648353e-01 -1.27375707e-01
-3.68048519e-01 5.71911275e-01 -6.73563080e-03 -3.08576763e-01
3.30546826e-01 -2.71143354e-02 9.92479250e-02 -3.77824277e-01
1.74958501e-02 1.46630913e-01 5.74325919e-01 -6.65445387e-01
4.36838359e-01 3.20977390e-01 3.84773910e-01 -6.15701735e-01
-2.19708577e-01 -3.76071967e-02 -1.18787146e+00 -4.76998687e-01
1.13821006e+00 -4.27576900e-01 -3.17655265e-01 2.95630306e-01
-1.07767701e+00 -4.80411023e-01 4.15798992e-01 8.39762568e-01
-2.63431877e-01 5.06814539e-01 -8.24124753e-01 -4.46588695e-01
-7.60786176e-01 -1.38218892e+00 1.66167036e-01 2.73244083e-01
-2.64306903e-01 -1.23629045e+00 6.62617758e-02 3.62943351e-01
4.05503899e-01 6.64447844e-01 1.10340130e+00 -5.36605477e-01
5.15609607e-02 -8.53068009e-03 -2.24905074e-01 3.39015454e-01
2.06452370e-01 3.10011238e-01 -6.62194848e-01 -1.54422984e-01
-6.83672279e-02 2.72262454e-01 4.69753206e-01 1.03887987e+00
1.29793954e+00 -2.25009099e-02 -1.56442702e-01 8.72131467e-01
1.14625978e+00 5.87718904e-01 3.85604173e-01 -4.69123781e-01
8.10457885e-01 1.15371001e+00 1.59643561e-01 2.27074727e-01
3.64172727e-01 7.50186816e-02 2.21753746e-01 -2.64947861e-01
-2.80868232e-01 2.19206557e-01 -2.55648404e-01 1.21889961e+00
-4.51168716e-01 1.93578735e-01 -1.52218616e+00 9.06436622e-01
-1.13203049e+00 -6.61371350e-01 -4.76271272e-01 2.03215957e+00
1.16304147e+00 -1.96016297e-01 -2.73530297e-02 -2.03602970e-01
1.19292438e+00 -3.80363673e-01 -4.79995131e-01 -3.25487643e-01
2.26027861e-01 5.48684657e-01 4.31103408e-01 4.82944936e-01
-5.96017480e-01 6.61923647e-01 6.80954647e+00 4.46018696e-01
-1.18137550e+00 3.32497537e-01 1.07312226e+00 7.55558629e-03
-2.38763511e-01 -3.39456767e-01 -1.18117660e-01 6.10421717e-01
7.53425479e-01 3.46013568e-02 3.52115840e-01 4.10561919e-01
2.33939201e-01 -2.12935701e-01 -9.29068983e-01 5.99139273e-01
-1.49104863e-01 -1.25135493e+00 -2.23093659e-01 -9.38778445e-02
1.02289641e+00 1.51163593e-01 -1.26608163e-01 -4.64424431e-01
-1.55261043e-03 -1.29230905e+00 5.44757247e-01 6.89107299e-01
1.45054519e+00 -8.33289504e-01 5.46756327e-01 2.50797927e-01
-9.80057478e-01 4.52480674e-01 -1.99248940e-01 3.03781658e-01
2.49084979e-01 9.08152938e-01 -1.00400352e+00 9.89992693e-02
6.71442032e-01 1.83663115e-01 -2.56349236e-01 1.44818413e+00
-1.86192334e-01 9.05238569e-01 -2.62265503e-01 4.62634623e-01
-2.40623623e-01 -6.61634445e-01 4.02591914e-01 1.30062604e+00
4.98560756e-01 3.92249644e-01 -2.89132088e-01 1.25573754e+00
-1.07263073e-01 3.40404451e-01 -1.22078948e-01 1.56281635e-01
5.78809619e-01 1.47362781e+00 -1.50566208e+00 -1.35653555e-01
-3.67259085e-01 5.05296528e-01 2.04761088e-01 2.60762304e-01
-7.07877219e-01 -1.71389252e-01 3.12909096e-01 2.93690413e-01
-9.04172286e-02 -2.63277262e-01 -7.60240793e-01 -9.24189985e-01
-2.25395218e-01 -3.85515809e-01 1.03132047e-01 -4.29940850e-01
-9.90675390e-01 4.17531073e-01 2.95478284e-01 -5.83849251e-01
3.60413827e-02 -1.73222959e-01 -9.00384009e-01 1.01414156e+00
-1.09297013e+00 -6.94444001e-01 -8.93612429e-02 3.98197711e-01
1.70867413e-01 8.96751359e-02 5.27719736e-01 7.38301933e-01
-7.58452177e-01 6.33368373e-01 -2.51320481e-01 4.32599872e-01
4.39158410e-01 -1.23956704e+00 2.83157796e-01 9.84409750e-01
-2.30524093e-01 9.55200315e-01 4.28508788e-01 -1.07471442e+00
-5.43950260e-01 -1.10577321e+00 6.16303802e-01 1.63758904e-01
3.20569038e-01 -3.72011401e-02 -9.26895738e-01 7.10050881e-01
1.37762532e-01 2.91199028e-01 1.00285959e+00 -3.38009328e-01
1.22666284e-01 6.51778355e-02 -1.64325416e+00 4.07876819e-01
8.45404088e-01 1.40124440e-01 -2.46533886e-01 1.33356124e-01
1.86029762e-01 -6.13950014e-01 -1.39311659e+00 6.77157283e-01
7.51056910e-01 -8.02444816e-01 7.70917833e-01 -2.29550809e-01
4.55609679e-01 -1.89531129e-02 4.14631069e-01 -1.22085834e+00
-3.26793522e-01 -3.28736842e-01 1.23134814e-01 1.37411356e+00
6.15953207e-01 -4.63901460e-01 8.74792159e-01 1.25885904e+00
-2.73273796e-01 -7.36089170e-01 -1.09356856e+00 -2.47463286e-01
5.01397371e-01 -4.09254253e-01 6.09349489e-01 1.21393406e+00
-1.01860672e-01 -2.50913888e-01 1.80493608e-01 2.13721171e-01
1.01418006e+00 -2.70373076e-01 -1.37775093e-01 -1.45439768e+00
3.70250911e-01 -6.25938535e-01 -1.55812442e-01 -4.94395703e-01
2.65788406e-01 -1.19633150e+00 1.77074566e-01 -1.59399581e+00
3.78271788e-01 -7.39510536e-01 -2.37089723e-01 3.86306018e-01
-3.20470631e-02 4.39526498e-01 -2.47477591e-01 -7.32218176e-02
4.82789502e-02 1.84091911e-01 1.57227707e+00 1.03970252e-01
4.21094671e-02 -5.34893498e-02 -3.56423408e-01 1.07060122e+00
9.47517991e-01 -6.20569229e-01 -4.37880486e-01 -5.16737700e-01
-1.36436173e-03 4.11808610e-01 6.60420358e-02 -8.79060626e-01
1.01364084e-01 -2.74219096e-01 8.11821997e-01 -3.87327492e-01
-2.02458635e-01 -6.82505012e-01 2.84192979e-01 4.30807263e-01
-2.76016295e-01 1.58333197e-01 -9.02919471e-03 -9.64732170e-02
1.71499044e-01 -2.98484981e-01 1.19038606e+00 -1.48788109e-01
1.56821579e-01 8.36737335e-01 -5.51121175e-01 1.85257763e-01
8.52353752e-01 -1.28663525e-01 2.22429723e-01 -1.84517011e-01
-9.55600202e-01 2.41333410e-01 2.47856379e-01 -6.11060299e-02
7.33274102e-01 -1.10364473e+00 -9.35323060e-01 3.78975458e-02
-6.06684148e-01 3.13642949e-01 4.25769776e-01 1.49488473e+00
-1.11174273e+00 7.34585971e-02 -5.68600297e-01 -6.23131514e-01
-1.15779555e+00 -7.63036609e-02 4.43859130e-01 9.06564072e-02
-5.74436486e-01 1.17339242e+00 1.99706957e-01 -3.99924785e-01
-9.45617780e-02 -4.04982120e-01 -4.68271255e-01 2.89302156e-03
1.98726431e-01 5.88027835e-01 -3.38344611e-02 -1.00194502e+00
-3.47805291e-01 6.89932525e-01 1.85444817e-01 -1.20869108e-01
1.30970204e+00 -3.21516186e-01 -6.79889798e-01 4.11831699e-02
8.57800722e-01 -6.59953877e-02 -1.14848948e+00 2.01849863e-01
1.49538796e-02 -1.62628174e-01 1.89124495e-02 -8.21713865e-01
-1.87026477e+00 1.05851424e+00 7.23700464e-01 -2.69963127e-02
8.50380063e-01 -2.55706012e-01 7.10726976e-01 -4.53785896e-01
2.77344823e-01 -9.65929925e-01 -6.03338599e-01 5.54657243e-02
8.09508979e-01 -8.96555483e-01 7.65296724e-03 -6.97223783e-01
-5.92824280e-01 1.10246766e+00 8.78173232e-01 -1.06956866e-02
7.55701065e-01 4.24758077e-01 2.00579241e-01 -3.36536914e-01
-1.78779051e-01 4.78287369e-01 7.05385268e-01 7.30058253e-01
8.16328347e-01 1.22207552e-01 -7.99816251e-01 6.58853769e-01
-2.12400362e-01 -4.44461077e-01 2.05187082e-01 4.84472275e-01
-2.08349600e-01 -6.57750666e-01 -2.29177684e-01 8.31268668e-01
-1.10549247e+00 -1.74650028e-01 -9.57455486e-02 3.84489536e-01
2.99896061e-01 6.88487709e-01 3.70366983e-02 1.23250999e-01
-2.60676116e-01 7.65339807e-02 6.54420614e-01 -7.25053906e-01
-4.97011632e-01 9.92137492e-02 -6.23442754e-02 -4.12949502e-01
-4.23471898e-01 -9.26686645e-01 -1.76004028e+00 3.94230522e-02
-3.20166290e-01 1.24202400e-01 9.10127103e-01 1.04214978e+00
6.63912222e-02 4.87258762e-01 4.51660007e-01 -7.85162032e-01
3.95782202e-01 -1.10923123e+00 -7.17141807e-01 1.40756384e-01
-2.45550238e-02 -7.33729064e-01 -8.41383934e-02 2.88843751e-01] | [14.065080642700195, -2.414581775665283] |
9c008d97-67a9-4f64-b590-8cf819938580 | sheng-cheng-mo-xing-zai-ceng-ci-jie-gou-ji | null | null | https://aclanthology.org/2022.ccl-1.60 | https://aclanthology.org/2022.ccl-1.60.pdf | 生成模型在层次结构极限多标签文本分类中的应用(Generation Model for Hierarchical Extreme Multi-label Text Classification) | “层次结构极限多标签文本分类是自然语言处理研究领域中一个重要而又具有挑战性的课题。该任务类别标签数量巨大且自成体系,标签与标签之间还具有不同层级间的依赖关系或同层次间的相关性,这些特性进一步增加了任务难度。该文提出将层次结构极限多标签文本分类任务视为序列转换问题,将输出标签视为序列,从而可以直接从数十万标签中生成与文本相关的类别标签。通过软约束机制和词表复合映射在解码过程中利用标签之间的层次结构与相关信息。实验结果表明,该文提出的方法与基线模型相比取得了有意义的性能提升。进一步分析表明,该方法不仅可以捕获利用不同层级标签之间的上下位关系,还对极限多标签体系自身携带的噪声具有一定容错能力。” | ['Weilei Wang', 'Jianping Lu', 'Yilin Liu', 'Yansi Xiao', 'Dawang He', 'Linqing Chen'] | null | null | null | null | ccl-2022-10 | ['multi-label-text-classification', 'multi-label-text-classification'] | ['methodology', 'natural-language-processing'] | [-7.11502135e-01 -9.31421399e-01 7.62112200e-01 3.85518700e-01
3.39652479e-01 -1.33563161e+00 8.60085886e-04 6.79706812e-01
2.52305031e-01 1.53424215e+00 1.50846526e-01 -1.04321063e-01
-3.75810862e-01 -1.18680477e+00 -1.16481513e-01 -1.19538641e+00
-1.06743835e-01 1.54813337e+00 4.56896454e-01 -2.12962538e-01
6.46522820e-01 9.01963174e-01 -9.89735723e-01 2.26780951e-01
1.01226187e+00 1.13672984e+00 1.06881630e+00 3.95395905e-01
1.65994465e-01 5.39679587e-01 -1.29793501e+00 2.54385263e-01
-2.66780883e-01 2.63312370e-01 -4.93064523e-01 8.40677381e-01
9.79825616e-01 -7.12602735e-01 -1.40032899e+00 1.14651632e+00
8.49211216e-01 1.55738190e-01 8.00248086e-01 -5.69204330e-01
-6.88970864e-01 5.17314255e-01 1.81753412e-01 1.72719806e-01
-2.19329447e-01 2.55225718e-01 6.31231606e-01 -1.03560984e+00
3.11487228e-01 1.47088242e+00 1.47054875e+00 5.91854453e-01
-1.43077886e+00 -1.27570999e+00 2.34127909e-01 -5.65869138e-02
-1.78752482e+00 4.82082725e-01 4.05374587e-01 -2.89377511e-01
1.19799864e+00 1.31415737e+00 1.48043442e+00 1.67377079e+00
5.13804555e-01 1.18729448e+00 1.59718239e+00 6.60063088e-01
3.42625618e-01 1.93646550e+00 -3.01661313e-01 1.59595236e-01
1.11100554e+00 -6.90156668e-02 1.91811169e-03 9.88080129e-02
1.62113145e-01 4.14430112e-01 1.39823481e-02 4.28715199e-01
-8.84862900e-01 1.13711464e+00 2.72513658e-01 1.54895091e+00
-1.61461905e-01 -9.35105205e-01 -9.63715613e-01 4.37469631e-01
-1.46321684e-01 2.35738277e-01 4.55371916e-01 1.14449143e+00
-3.59399348e-01 -2.57944226e-01 9.72317517e-01 1.53090394e+00
9.58748281e-01 4.18169707e-01 -6.25343248e-02 8.25939417e-01
1.26400936e+00 1.21717918e+00 1.36247027e+00 -8.59242558e-01
-6.19580090e-01 1.11044550e+00 2.43873551e-01 -1.84124148e+00
-1.07462358e+00 -1.38905489e+00 -8.60762119e-01 -1.08891761e+00
2.02381015e-02 -7.42351711e-01 -7.31179833e-01 1.35814655e+00
1.51993275e-01 -4.79675949e-01 5.42435884e-01 9.68993723e-01
8.96781206e-01 1.66499794e+00 -3.14369559e-01 7.10031018e-03
1.64343417e+00 -6.88152015e-01 -6.97696507e-01 -3.49461675e-01
-1.30419576e+00 -1.67833185e+00 7.67655015e-01 -4.25288886e-01
-8.10359955e-01 -3.52010757e-01 -2.36244678e+00 5.57637930e-01
-6.44229174e-01 -2.17301428e-01 1.33644867e+00 1.21916234e+00
-1.66898191e+00 5.44025540e-01 -4.87931430e-01 -9.60336208e-01
2.04218298e-01 4.32655036e-01 1.23516284e-01 2.54815109e-02
-9.71181154e-01 1.17395830e+00 -6.10185444e-01 5.52631259e-01
-2.29048824e+00 -5.63434839e-01 -1.66371316e-02 -2.59461105e-01
9.22646165e-01 -8.25164795e-01 1.27120590e+00 -9.73896801e-01
-8.50213230e-01 6.85333490e-01 -9.88109052e-01 9.08288583e-02
6.90079272e-01 7.21125901e-01 -2.68301159e-01 4.26726520e-01
-3.70291203e-01 2.72179514e-01 7.42561460e-01 -2.56571937e+00
-6.03304029e-01 -1.72196186e+00 2.90412486e-01 -2.86746562e-01
4.16762054e-01 7.59575069e-01 3.71198505e-01 -1.84878737e-01
3.61776620e-01 -1.75765908e+00 -7.60914609e-02 -1.67153072e+00
-5.90995550e-01 3.35925341e-01 1.87530684e+00 1.87837139e-01
1.23860455e+00 -1.86238563e+00 -1.15493071e+00 7.89805114e-01
4.40561533e-01 1.01921201e+00 1.26527280e-01 2.10812998e+00
3.88525546e-01 1.00348675e+00 -1.19339538e+00 5.41152954e-01
-2.57157028e-01 4.73374464e-02 9.72334385e-01 5.59435599e-02
-8.76539722e-02 -1.69134840e-01 -2.45048389e-01 3.69777828e-01
6.32085562e-01 8.67400289e-01 6.43428922e-01 3.55307937e-01
7.40017116e-01 3.79528135e-01 -1.29188895e-01 1.31973112e+00
1.41282856e+00 -6.66335642e-01 9.08179700e-01 3.60235393e-01
-3.38837236e-01 -8.72276485e-01 -2.68248343e+00 5.85951805e-01
-1.10725254e-01 7.47062862e-01 7.89843857e-01 -8.83752346e-01
1.66133082e+00 1.37955225e+00 8.15723598e-01 -8.48904550e-02
6.21220581e-02 8.94778132e-01 2.81607717e-01 -5.62247515e-01
7.65696704e-01 -8.13020706e-01 9.65705991e-01 2.12596476e-01
-7.80075252e-01 -1.03465714e-01 5.53999662e-01 -1.87118351e-01
1.33389544e+00 -1.24756575e+00 -3.06501478e-01 -1.13869727e+00
1.44269872e+00 -3.43138427e-01 2.31915280e-01 3.48395079e-01
-8.06151032e-01 3.14322829e-01 -7.15179324e-01 4.21360970e-01
-5.74124455e-01 -1.79904830e+00 -4.37746316e-01 5.73933601e-01
-3.15669298e-01 -9.58078742e-01 -9.28256750e-01 3.61005925e-02
-2.38843337e-01 4.62354034e-01 1.33899316e-01 2.42644861e-01
-8.32224250e-01 -1.40759885e+00 -4.52624112e-01 -1.85335964e-01
1.25794530e+00 -1.62833393e+00 -4.10274506e-01 -1.82640944e-02
-4.09803301e-01 -7.88552821e-01 -4.36385602e-01 -1.64283782e-01
-1.99908245e+00 -1.52042723e+00 -7.36697763e-02 -1.60699034e+00
1.58226633e+00 1.28640413e-01 7.89060354e-01 7.20390737e-01
-6.42554641e-01 1.04778588e+00 4.86985981e-01 -1.03360522e+00
3.57146859e-01 -2.57964313e-01 1.76383629e-01 1.01650596e+00
7.25358069e-01 -4.88013715e-01 -1.45629442e+00 8.72215271e-01
-1.49318886e+00 -8.50696087e-01 3.11951339e-01 6.23821437e-01
8.04893315e-01 1.57202423e-01 -3.58336680e-02 -7.60153651e-01
1.14808202e+00 7.11130023e-01 -1.10169792e+00 6.72623992e-01
-1.31823349e+00 -7.59452641e-01 9.79779840e-01 -5.31705260e-01
-9.27583218e-01 -8.22575092e-01 -3.27238530e-01 1.95455521e-01
3.61384600e-01 -2.75864512e-01 -3.43796372e-01 2.20972210e-01
-6.07965469e-01 -4.44663279e-02 5.11324584e-01 -5.50296605e-01
-3.68379325e-01 1.39947867e+00 2.76031762e-01 -1.18706584e+00
3.47438246e-01 6.15855396e-01 1.02642871e-01 -5.74174643e-01
2.04096839e-01 -3.40393111e-02 -9.57497537e-01 7.15556860e-01
8.03165257e-01 -1.47917676e+00 -3.94184917e-01 1.06818807e+00
-3.49133849e-01 7.41063416e-01 -6.72991633e-01 1.63052917e+00
-8.14635307e-02 -6.89133853e-02 -4.98199373e-01 -9.84302044e-01
-4.19804066e-01 -2.68577147e+00 3.54502469e-01 2.25360465e+00
-2.53816191e-02 -9.50322390e-01 3.87734398e-02 5.17524242e-01
1.00231385e+00 -1.10727385e-01 1.23225462e+00 -1.32670414e+00
-1.01094162e+00 -5.53056359e-01 2.93206662e-01 1.51259160e+00
3.95049036e-01 7.16164052e-01 -4.87156883e-02 -2.45877922e-01
2.55326688e-01 8.84988010e-01 5.58921337e-01 6.98123455e-01
4.51384746e-02 -6.51446402e-01 -2.10120127e-01 2.85523593e-01
2.65639901e+00 8.07673395e-01 7.46918976e-01 1.68568945e+00
-8.77552092e-01 1.94140866e-01 3.62258404e-02 5.79532027e-01
-1.57871813e-01 1.10705543e+00 -8.35573971e-02 1.34336591e+00
-9.19561028e-01 -3.08329850e-01 8.36115122e-01 1.56388307e+00
3.54893178e-01 -8.78205597e-02 4.33007002e-01 6.26798987e-01
-2.44706655e+00 -1.82150269e+00 -1.10813649e-02 3.07638216e+00
-1.64872576e-02 -9.08361197e-01 -1.75248459e-01 5.39887071e-01
1.65846372e+00 -3.78126889e-01 -2.61123300e-01 -5.33865988e-01
1.67445149e-02 1.13589120e+00 5.66162646e-01 3.09201747e-01
-1.01336265e+00 1.62872419e-01 5.58768332e-01 9.03947532e-01
-1.38854814e+00 -1.53146544e-03 -1.03904724e-01 -7.25231707e-01
-8.67817998e-01 4.27847505e-01 -9.49118376e-01 1.65099561e+00
1.08703887e+00 -6.55857444e-01 -2.59375006e-01 3.28266919e-01
8.86684656e-01 -2.69039303e-01 -1.05041957e+00 1.46509743e+00
2.48413950e-01 -1.50016499e+00 6.65315911e-02 2.95263380e-01
1.06228995e+00 -1.51390001e-01 -1.39420778e-01 5.97867489e-01
1.10837273e-01 -1.14759970e+00 4.77882892e-01 -2.21316040e-01
-3.53137404e-01 -1.25128639e+00 1.74310529e+00 7.28735998e-02
-1.71350825e+00 -8.67523193e-01 -1.69734627e-01 -2.52800018e-01
4.07916009e-01 2.64652818e-01 -6.60430491e-01 3.57615143e-01
3.53401393e-01 -2.60945976e-01 -1.28539002e+00 2.10972714e+00
-8.02705526e-01 -3.07870120e-01 -8.03213000e-01 -2.00757718e+00
1.19336975e+00 -7.85137534e-01 9.98023272e-01 1.18300331e+00
2.50155747e-01 5.80504656e-01 4.61566061e-01 6.12016201e-01
-5.08513510e-01 1.80439591e-01 -4.37329292e-01 4.24599975e-01
9.82355475e-01 1.88661683e+00 -1.30639887e+00 -6.57492578e-01
6.77635297e-02 -1.10729861e+00 -3.08362633e-01 -5.43601274e-01
-3.96060526e-01 -8.30562055e-01 3.67769808e-01 2.83937335e-01
-1.34415820e-01 -8.09896827e-01 3.54250729e-01 -4.07960027e-01
-4.54711914e-02 -9.79523718e-01 -2.14055553e-01 -6.09890044e-01
-1.37198055e+00 2.98645437e-01 -2.34629855e-01 -1.44495094e+00
9.52657342e-01 -1.27508748e+00 -4.23188657e-01 2.85813183e-01
-5.64468876e-02 -7.28952646e-01 -8.69027674e-02 1.01706338e+00
1.07005370e+00 -2.38285214e-01 8.44448984e-01 9.54929411e-01
-7.34833062e-01 7.98364758e-01 1.70705485e+00 6.47358835e-01
2.51454324e-01 -1.54723585e+00 -1.01971900e+00 2.33028367e-01
5.56837916e-01 3.52925986e-01 -1.47918239e-02 -2.61221677e-01
-2.15712976e+00 -7.97772825e-01 7.47602463e-01 -4.66184288e-01
6.51908338e-01 3.05332057e-02 -2.13985741e-01 1.33149719e+00
4.91797477e-01 -3.48522127e-01 8.40298235e-01 -3.83758187e-01
2.45574042e-01 -1.20030642e+00 -2.11731124e+00 1.28634894e+00
7.90433228e-01 -2.38135830e-02 -6.92084134e-01 4.61367279e-01
-4.26918119e-01 -3.45133007e-01 -1.15018046e+00 8.60552371e-01
1.66332114e+00 -8.63683045e-01 6.30139887e-01 -9.89343882e-01
-6.35664105e-01 -1.24037936e-01 -1.08354127e+00 -1.31917048e+00
4.26904202e-01 -5.98819911e-01 1.07645905e+00 1.43905568e+00
1.01116323e+00 -1.24378586e+00 3.19188714e-01 1.44740307e+00
-5.64425170e-01 -6.07775450e-01 -2.13136888e+00 -1.31393862e+00
-1.34882659e-01 -9.67846289e-02 2.63709664e-01 -6.61073253e-02
-2.54837930e-01 6.06919765e-01 -5.25592081e-02 3.79478484e-02
4.16786969e-01 -3.43447685e-01 5.79569519e-01 -9.53277409e-01
1.10865259e+00 -8.22691381e-01 -3.24753106e-01 -1.02446653e-01
-8.42228711e-01 -3.36456984e-01 -1.50005209e+00 -1.95959151e+00
4.51999247e-01 -3.93225551e-01 -2.94219047e-01 -9.72138643e-01
1.68565869e+00 3.37091178e-01 8.23663354e-01 8.79685879e-01
2.49560803e-01 -5.43226898e-01 1.40725839e+00 -1.13845611e+00
-2.96940297e-01 4.46880698e-01 -2.81642824e-01 1.18296921e+00
4.15402919e-01 -1.05020106e-01 -1.73716187e-01 -7.78823435e-01
1.42453864e-01 -9.39320266e-01 -1.19429624e+00 -1.40694642e+00
6.29877031e-01 3.26033592e-01 2.61413515e-01 -8.74868110e-02
7.31819451e-01 -9.41343784e-01 6.32993102e-01 6.28670990e-01
5.25729597e-01 1.00330067e+00 -3.61832768e-01 -4.39962059e-01
-5.21868952e-02 -1.87610555e+00 9.66302991e-01 -7.43269563e-01
-5.63716233e-01 -3.73916537e-01 -5.84837615e-01 -3.28717858e-01
1.83223975e+00 -1.97869927e-01 -7.83426404e-01 7.69643247e-01
-1.19896293e+00 -6.44278944e-01 3.25601965e-01 1.21817160e+00
1.45988667e+00 -1.12498581e+00 -2.05371046e+00 1.31345749e-01
-1.09501338e+00 -1.29313335e-01 6.70464873e-01 9.00616050e-01
-8.19489121e-01 7.83074677e-01 -4.07200068e-01 -5.95219553e-01
-2.31515050e+00 8.94981503e-01 -2.10236177e-01 -6.58076763e-01
-3.59887064e-01 6.36727273e-01 -1.17028463e+00 2.03611553e-01
1.72301337e-01 -1.33497372e-01 -6.95765913e-01 4.24330205e-01
1.00003779e+00 9.08982754e-01 8.86835307e-02 -1.25766540e+00
-9.23697293e-01 1.90211427e+00 -5.45911431e-01 -2.10121453e-01
4.08392400e-01 -6.55032933e-01 -5.06357133e-01 -2.81374007e-01
8.01514864e-01 1.49517965e+00 3.76768671e-02 8.44340324e-01
-5.25989354e-01 -6.39266670e-01 -1.07669234e+00 -1.76190007e+00
-9.62000906e-01 4.73759979e-01 9.52533185e-01 -5.30108750e-01
6.63727880e-01 -7.43045926e-01 1.46155369e+00 4.04361576e-01
1.54315174e+00 -2.28398275e+00 -7.69974589e-01 6.77967370e-01
1.21868980e+00 -1.51216650e+00 3.17919731e-01 -2.32995450e-01
1.62382692e-01 1.59452355e+00 8.07541370e-01 -5.97874105e-01
1.51394343e+00 2.06385911e-01 -2.29476109e-01 -1.86177306e-02
-1.50761351e-01 -3.47347945e-01 -5.80146253e-01 1.15139115e+00
7.93504834e-01 1.99989945e-01 -1.59514201e+00 4.47725475e-01
-1.13696313e+00 4.47226018e-01 1.61393213e+00 8.54552627e-01
-8.35405886e-01 -6.12991035e-01 -1.33568537e+00 -3.74793202e-01
-7.53213584e-01 6.36407971e-01 1.81039050e-01 9.93630171e-01
5.81083894e-01 2.82139444e+00 -5.92929602e-01 -1.08169496e+00
1.78772175e+00 8.46074894e-02 8.63581121e-01 -7.65185654e-02
-1.58617628e+00 -4.15195793e-01 1.54825822e-01 8.02466869e-01
-6.69589102e-01 -2.76924431e-01 -8.28075469e-01 -1.08602905e+00
3.57035607e-01 2.04078746e+00 1.60929191e+00 3.44986111e-01
5.38549051e-02 9.80302751e-01 1.28306079e+00 -3.76174062e-01
-1.53680515e+00 -1.11111712e+00 -1.00812995e+00 -3.32686216e-01
-1.65916800e-01 -9.47982073e-01 -1.39905834e+00 -1.31943667e+00] | [-3.31616473197937, 6.907747268676758] |
2f916d60-2031-49b5-9975-2c1980353a57 | uncertainty-in-minimum-cost-multicuts-for | 2105.07469 | null | https://arxiv.org/abs/2105.07469v1 | https://arxiv.org/pdf/2105.07469v1.pdf | Uncertainty in Minimum Cost Multicuts for Image and Motion Segmentation | The minimum cost lifted multicut approach has proven practically good performance in a wide range of applications such as image decomposition, mesh segmentation, multiple object tracking, and motion segmentation. It addresses such problems in a graph-based model, where real-valued costs are assigned to the edges between entities such that the minimum cut decomposes the graph into an optimal number of segments. Driven by a probabilistic formulation of minimum cost multicuts, we provide a measure for the uncertainties of the decisions made during the optimization. We argue that access to such uncertainties is crucial for many practical applications and conduct an evaluation by means of sparsifications on three different, widely used datasets in the context of image decomposition (BSDS-500) and motion segmentation (DAVIS2016 and FBMS59) in terms of variation of information (VI) and Rand index (RI). | ['Margret Keuper', 'Amirhossein Kardoost'] | 2021-05-16 | null | null | null | null | ['motion-segmentation'] | ['computer-vision'] | [ 3.82228106e-01 4.12165135e-01 -4.53218251e-01 -2.65846550e-01
-7.61837602e-01 -5.09060502e-01 5.45595229e-01 3.68215412e-01
-2.85376042e-01 5.65943301e-01 -5.62669076e-02 -7.52905011e-02
-8.33494425e-01 -7.59523273e-01 -8.62158179e-01 -5.86845636e-01
-1.38021141e-01 9.93511558e-01 4.96474415e-01 1.76784098e-01
2.64862686e-01 4.11474407e-01 -1.33496869e+00 -3.29151936e-02
9.62701797e-01 9.67331886e-01 -6.79241261e-04 2.36535773e-01
-2.22918075e-02 -2.91669220e-02 -2.38002822e-01 -4.43333715e-01
2.88239628e-01 5.95277101e-02 -1.02417254e+00 4.28458005e-01
-3.64594869e-02 2.33593300e-01 2.68863112e-01 1.09979618e+00
1.00959532e-01 2.38802329e-01 9.49012339e-01 -1.30711067e+00
7.33953416e-02 8.28945696e-01 -1.01628196e+00 1.40727296e-01
1.83701962e-01 2.96069700e-02 1.04690695e+00 -5.94778478e-01
9.82199609e-01 1.43538058e+00 5.60041487e-01 -2.88102310e-02
-1.62928569e+00 -2.15023966e-03 1.77278906e-01 -6.60063140e-03
-1.24786031e+00 1.55399945e-02 4.31469202e-01 -7.04901814e-01
3.25828314e-01 5.42612016e-01 4.08839703e-01 7.21736073e-01
1.98803797e-01 7.33657241e-01 9.02288258e-01 -4.73149270e-01
4.07644033e-01 -1.05710551e-01 2.70496696e-01 3.23800802e-01
7.63007879e-01 -4.23299596e-02 -4.26342309e-01 -1.64833531e-01
4.26827401e-01 -4.75037903e-01 -1.88674971e-01 -6.52155578e-01
-1.10761368e+00 7.85319626e-01 2.89549261e-01 3.50274146e-01
-2.47641847e-01 1.78764760e-01 3.35514635e-01 -1.99459895e-01
5.94771802e-01 2.69054979e-01 -3.61918300e-01 2.27060422e-01
-1.08162451e+00 2.47704849e-01 5.35105288e-01 7.92383194e-01
7.22896039e-01 -2.12619096e-01 -2.61636198e-01 6.60289288e-01
6.58319354e-01 3.91758353e-01 -2.13190317e-01 -1.19458950e+00
5.93278289e-01 6.10795498e-01 -4.22070920e-02 -1.30729055e+00
-3.90427828e-01 -3.29094946e-01 -8.14088702e-01 1.34503514e-01
4.70159352e-01 3.93394148e-03 -1.00437665e+00 1.58449244e+00
6.94307268e-01 2.60015309e-01 -3.79068255e-01 8.02271783e-01
4.39675540e-01 4.65134382e-01 3.69013846e-02 -3.01561981e-01
1.12404013e+00 -3.95605177e-01 -5.19512713e-01 -1.31190330e-01
2.22744718e-01 -5.59848607e-01 6.33770704e-01 6.98464155e-01
-1.00889671e+00 -3.07260036e-01 -8.86206388e-01 2.42881194e-01
2.71894839e-02 -1.95793614e-01 4.02583331e-01 8.15821052e-01
-7.66394198e-01 6.49638593e-01 -8.56142879e-01 -1.44175619e-01
5.52996159e-01 3.87904465e-01 -8.06789920e-02 -1.17192730e-01
-8.52914929e-01 5.59594274e-01 1.71974480e-01 2.19772115e-01
-9.02412355e-01 -4.77446616e-01 -8.90162766e-01 -5.86776733e-02
6.95960581e-01 -7.02745914e-01 5.32370210e-01 -5.59303641e-01
-1.00730264e+00 8.41709614e-01 1.43005908e-01 -6.28683031e-01
8.06676507e-01 -1.14323674e-02 1.52329624e-01 2.60282785e-01
2.95677900e-01 5.69884598e-01 9.98296440e-01 -1.44603550e+00
-4.21088785e-01 -5.99963367e-01 -3.32328379e-02 9.05973930e-03
1.10063523e-01 -1.60125047e-01 -5.68744481e-01 -7.30883241e-01
3.90188962e-01 -1.02417755e+00 -6.13322020e-01 -2.75022626e-01
-7.45986700e-01 6.54440299e-02 6.07996345e-01 -6.39086604e-01
1.07168865e+00 -1.83604133e+00 7.74707735e-01 8.23832214e-01
7.73328617e-02 -2.10390776e-01 5.13605326e-02 2.69554406e-01
2.67060995e-01 4.22593176e-01 -8.38409066e-01 -4.12842780e-01
1.10608131e-01 4.28981364e-01 -6.00824505e-02 7.17416584e-01
1.28114879e-01 6.08378768e-01 -6.77483380e-01 -5.49502909e-01
3.49447459e-01 3.40319663e-01 -3.39302689e-01 -3.53068024e-01
-5.02369761e-01 4.31050628e-01 -6.99817657e-01 5.22169709e-01
8.58446062e-01 -2.61666868e-02 3.96837324e-01 -2.34615788e-01
9.86530930e-02 -2.67198652e-01 -1.65570080e+00 1.61242855e+00
-1.78003326e-01 2.81519085e-01 3.77718210e-01 -9.47146952e-01
5.65082133e-01 3.59719917e-02 9.19705749e-01 -3.05555880e-01
2.29025051e-01 -2.65512913e-02 -2.18196005e-01 -1.70571133e-01
4.32030916e-01 7.17101321e-02 -2.34527454e-01 3.74865443e-01
-1.05612949e-01 -4.92718786e-01 5.44543982e-01 4.37840015e-01
9.35099840e-01 8.79689865e-03 -4.86319400e-02 -6.93969429e-01
4.75109518e-01 1.52921900e-02 5.53864062e-01 4.16416109e-01
8.40954185e-02 8.78521144e-01 8.13588977e-01 2.30618082e-02
-6.40975714e-01 -9.01902318e-01 -4.31292921e-01 3.38967472e-01
3.69920135e-01 -3.18242788e-01 -1.03167057e+00 -6.42017722e-01
1.94412231e-01 6.20150089e-01 -7.45103061e-01 7.52950609e-02
-2.39638075e-01 -9.60958540e-01 2.48975068e-01 2.06702918e-01
-2.61875261e-02 -7.08244383e-01 -7.57013321e-01 1.11142613e-01
-1.11454621e-01 -1.26053977e+00 -2.88793772e-01 -2.81032957e-02
-1.14126539e+00 -1.37861574e+00 -5.56790709e-01 -1.33879185e-01
7.20593393e-01 -1.09632149e-01 1.26533461e+00 1.69064179e-01
-3.28198940e-01 6.89582705e-01 -3.23605776e-01 -3.84241007e-02
-3.04259628e-01 6.24678992e-02 -2.41180599e-01 2.69254804e-01
-4.45368350e-01 -3.77141148e-01 -2.64895290e-01 5.79247594e-01
-1.22833836e+00 -1.39760911e-01 1.85473725e-01 5.27018428e-01
1.02023339e+00 3.51645052e-01 2.01812074e-01 -9.01602387e-01
3.86466593e-01 -6.33936942e-01 -9.00892675e-01 3.93094867e-01
-6.87790096e-01 1.45389825e-01 6.74489662e-02 -6.63562939e-02
-9.37267661e-01 -1.03649788e-01 2.76333064e-01 -2.41849259e-01
1.28977478e-01 7.94456601e-01 -2.90491462e-01 -1.56525195e-01
3.20861340e-01 -4.04853344e-01 -2.72151381e-01 -3.70083928e-01
3.53528559e-01 1.40497148e-01 2.33585402e-01 -9.11205173e-01
6.03763640e-01 6.47890627e-01 7.07229197e-01 -7.20880985e-01
-5.12822449e-01 -4.16831523e-01 -5.43663204e-01 -4.47218567e-01
7.23965108e-01 -4.22413081e-01 -6.23466074e-01 2.79107690e-01
-9.38676059e-01 -2.25477874e-01 -4.65553671e-01 3.10525268e-01
-6.17696106e-01 5.83097160e-01 -2.58512259e-01 -8.50840807e-01
3.93541669e-03 -1.41973376e+00 1.10460210e+00 1.70144409e-01
-2.18035385e-01 -1.00681090e+00 -4.10125591e-02 6.10240817e-01
-3.23629379e-02 7.17861354e-01 9.73721862e-01 -1.32801861e-01
-7.82764792e-01 -1.45657826e-02 3.57665978e-02 2.71138251e-01
-1.98517159e-01 3.95540655e-01 -5.83589315e-01 -7.54426569e-02
-1.41604438e-01 2.16593668e-02 9.44410086e-01 1.03812599e+00
9.61658418e-01 -1.49175748e-02 -4.56705987e-01 4.65604097e-01
1.60771906e+00 -9.19478312e-02 6.98896110e-01 2.45572582e-01
6.60220563e-01 9.44462299e-01 8.10356379e-01 5.34975886e-01
3.20887953e-01 7.85827041e-01 7.86581218e-01 3.32026958e-01
-8.82206336e-02 1.00369327e-01 -2.97929309e-02 4.61691469e-01
-3.15796025e-02 -5.85537553e-01 -1.00424051e+00 7.31844604e-01
-1.94798410e+00 -5.37235200e-01 -5.69438219e-01 2.44018459e+00
3.67698699e-01 3.30818713e-01 1.47936016e-01 3.30209702e-01
8.23002577e-01 2.30145350e-01 -4.17764544e-01 -3.31379086e-01
-1.19426817e-01 7.62126967e-02 6.39100611e-01 7.53180981e-01
-8.46698403e-01 3.69235128e-01 6.37913704e+00 1.04954338e+00
-5.61018407e-01 -3.30752395e-02 1.11517322e+00 2.11197883e-01
-6.33670807e-01 1.49848834e-01 -5.50649464e-01 6.54484630e-01
6.77702546e-01 3.22966278e-02 2.84657478e-01 4.90063637e-01
1.47097141e-01 -7.25235760e-01 -7.67056108e-01 4.45593417e-01
-4.00229424e-01 -1.39086342e+00 -5.75580038e-02 2.15684146e-01
9.61215556e-01 -4.39387187e-02 1.90327868e-01 -5.21187842e-01
2.14651361e-01 -1.00258207e+00 8.66669655e-01 5.12079060e-01
5.95375121e-01 -8.30729723e-01 4.15148020e-01 2.38350943e-01
-1.16693544e+00 1.60951838e-01 -3.29405651e-03 4.47478443e-01
4.76732403e-01 1.23771405e+00 -4.90036130e-01 1.05280972e+00
5.22026300e-01 4.98649597e-01 -1.98870882e-01 1.02093279e+00
-5.29041067e-02 6.14978433e-01 -6.34312868e-01 4.89529520e-01
-3.82257327e-02 -6.66260839e-01 9.27963674e-01 8.40347886e-01
1.16224647e-01 -1.77109331e-01 2.98543386e-02 1.04683959e+00
-1.49810398e-02 -2.94914860e-02 -1.24164693e-01 1.02314651e-01
2.26237789e-01 1.16504943e+00 -1.47227788e+00 9.28882658e-02
1.53616652e-01 2.96826780e-01 -2.02439606e-01 2.60246068e-01
-7.86733329e-01 2.77830422e-01 4.23605174e-01 3.48369211e-01
4.06948864e-01 -2.76205689e-01 -6.75256193e-01 -6.86767817e-01
1.53388932e-01 -6.79070652e-01 5.29539049e-01 -3.91148299e-01
-9.72532153e-01 4.25720394e-01 4.84323323e-01 -9.18932438e-01
-2.37965748e-01 -5.04019737e-01 -4.32639182e-01 4.35995817e-01
-1.15612352e+00 -1.05266404e+00 3.86666395e-02 2.18555808e-01
4.40390348e-01 3.32560271e-01 2.13276491e-01 3.00002545e-01
-8.19907546e-01 1.55027732e-01 1.06333748e-01 -3.41818660e-01
5.17525375e-02 -1.12705731e+00 2.99305826e-01 1.02740073e+00
2.90762186e-01 2.07548887e-01 9.52768803e-01 -9.12189126e-01
-1.26168835e+00 -8.04976881e-01 3.27225059e-01 -3.49630713e-01
3.60831559e-01 4.19244543e-02 -5.78100801e-01 4.42948312e-01
-9.47081819e-02 1.96722999e-01 3.15865278e-02 -1.79175109e-01
2.11410969e-01 3.17187817e-03 -1.49748003e+00 2.09700733e-01
1.00529301e+00 9.15813074e-02 -1.20912388e-01 3.41036469e-01
5.32777846e-01 -4.13553238e-01 -1.03338158e+00 7.16932535e-01
3.35561424e-01 -1.03382039e+00 1.17705894e+00 -2.15742722e-01
3.21415335e-01 -3.03120673e-01 -1.73497066e-01 -1.05132866e+00
1.76630184e-01 -6.26558542e-01 -6.99050650e-02 1.28883028e+00
2.59663463e-01 -4.70131308e-01 7.61170685e-01 7.10197806e-01
4.97353934e-02 -9.52581942e-01 -1.28715694e+00 -7.38369048e-01
-2.71459609e-01 -6.23227775e-01 5.37390709e-01 7.00034082e-01
-6.52854979e-01 -1.30586937e-01 -1.03238165e-01 1.70096189e-01
1.07062900e+00 9.63128656e-02 4.75969315e-01 -1.39513612e+00
-2.94034600e-01 -4.04302597e-01 -4.45753813e-01 -5.68628311e-01
-2.63817161e-02 -6.63071454e-01 -8.29883739e-02 -1.55151081e+00
1.16299190e-01 -7.90508628e-01 3.45926136e-02 -5.80198839e-02
8.03540424e-02 -9.03635770e-02 1.09712236e-01 -3.13174278e-02
-5.98306119e-01 3.57743412e-01 9.98626709e-01 -1.78100884e-01
1.06272712e-01 3.33071530e-01 -4.02053386e-01 8.96166682e-01
3.21951658e-01 -8.02831113e-01 -4.09662217e-01 -2.89037168e-01
4.68874931e-01 2.60387361e-01 3.31245631e-01 -8.02398503e-01
-2.41397414e-02 -2.64396459e-01 -1.17933750e-01 -6.32432103e-01
1.91535637e-01 -9.96573567e-01 6.87729657e-01 4.36353862e-01
-1.72291651e-01 -1.37417987e-01 1.31564245e-01 7.62862742e-01
-1.52842984e-01 -6.62972510e-01 6.11617148e-01 6.80660754e-02
-5.83211780e-01 2.13616103e-01 -1.77438885e-01 3.41249913e-01
1.22612739e+00 -4.37166005e-01 6.03278354e-02 -2.21556991e-01
-8.08152556e-01 4.12676036e-01 5.77387154e-01 1.94661766e-01
3.56773436e-01 -9.12409365e-01 -5.12432337e-01 -1.35450304e-01
-2.83825785e-01 3.38926584e-01 4.51019764e-01 9.08641875e-01
-4.09929186e-01 1.69328094e-01 8.06133673e-02 -9.37367380e-01
-1.14713418e+00 1.42419294e-01 1.90268621e-01 -6.27049267e-01
-3.72915804e-01 7.63828099e-01 8.00829455e-02 -1.10330328e-01
1.22330092e-01 -6.22500777e-01 -3.48791890e-02 3.18062752e-01
-2.15164542e-01 8.96757782e-01 2.10298792e-01 -8.40512931e-01
-5.06767690e-01 8.71867120e-01 3.47682983e-01 -4.02674258e-01
1.39912724e+00 -2.53409356e-01 -1.72825456e-01 1.10087365e-01
6.94601119e-01 9.90100671e-03 -1.39284420e+00 4.85815294e-02
4.63679522e-01 -6.31809056e-01 1.17259987e-01 -6.54112160e-01
-1.33707142e+00 4.31094259e-01 3.20444971e-01 4.97302324e-01
8.38430107e-01 1.64642185e-01 6.01125836e-01 -2.73434043e-01
6.19374394e-01 -1.16568828e+00 -9.46928561e-02 9.69508886e-02
8.45149159e-01 -9.84601557e-01 4.99588311e-01 -8.95590246e-01
-6.05459154e-01 8.05333078e-01 1.62809521e-01 -2.01725774e-02
7.12043285e-01 3.56827170e-01 -4.16563600e-01 -3.41351986e-01
-3.75293791e-01 -5.93082793e-02 7.79131353e-01 4.92936254e-01
-9.40600857e-02 1.68129832e-01 -5.21239161e-01 5.63326001e-01
2.50593632e-01 -2.71307647e-01 4.70290065e-01 7.48114169e-01
-2.09805310e-01 -1.13340127e+00 -5.43998063e-01 4.37668055e-01
-4.93669301e-01 4.33184624e-01 -1.10795550e-01 8.27050209e-01
2.04075500e-01 9.18346584e-01 -2.45662406e-01 -1.23095751e-01
1.79741323e-01 -4.80508089e-01 5.35425127e-01 -6.68011785e-01
-4.60769296e-01 3.92295308e-02 1.95840001e-01 -5.03505349e-01
-7.48249352e-01 -9.16940093e-01 -1.29855454e+00 4.85384315e-02
-4.97934252e-01 1.38851359e-01 7.43130863e-01 1.08554304e+00
3.15304011e-01 5.36993444e-01 3.84303123e-01 -8.85426581e-01
-4.40557927e-01 -4.54494148e-01 -5.72849751e-01 3.49477589e-01
-1.43298328e-01 -9.20059264e-01 -4.11920279e-01 -2.01843858e-01] | [9.304021835327148, -0.2701524794101715] |
8f44c88d-d986-447b-9554-1d58924ec8b4 | punctuation-as-native-language-interference | null | null | https://aclanthology.org/C18-1293 | https://aclanthology.org/C18-1293.pdf | Punctuation as Native Language Interference | In this paper, we describe experiments designed to explore and evaluate the impact of punctuation marks on the task of native language identification. Punctuation is specific to each language, and is part of the indicators that overtly represent the manner in which each language organizes and conveys information. Our experiments are organized in various set-ups: the usual multi-class classification for individual languages, also considering classification by language groups, across different proficiency levels, topics and even cross-corpus. The results support our hypothesis that punctuation marks are persistent and robust indicators of the native language of the author, which do not diminish in influence even when a high proficiency level in a non-native language is achieved. | ['Carlo Strapparava', 'Vivi Nastase', 'Ilia Markov'] | 2018-08-01 | punctuation-as-native-language-interference-1 | https://aclanthology.org/C18-1293 | https://aclanthology.org/C18-1293.pdf | coling-2018-8 | ['cross-corpus', 'native-language-identification'] | ['computer-vision', 'natural-language-processing'] | [-4.65631634e-01 -3.83637935e-01 -5.18987834e-01 -1.76439017e-01
-5.55626214e-01 -1.01196659e+00 1.16957700e+00 7.40595341e-01
-8.23060274e-01 5.81057489e-01 5.58001041e-01 -5.96714318e-01
-8.76670331e-02 -4.63606507e-01 -3.49844307e-01 -3.11458915e-01
1.27668709e-01 1.85723022e-01 2.17192441e-01 -3.13099772e-01
5.80480993e-01 6.96025550e-01 -1.40804684e+00 1.14447795e-01
8.81220996e-01 2.22861975e-01 3.18151593e-01 4.85809982e-01
-8.55936885e-01 8.23890150e-01 -1.11814499e+00 -2.93621868e-01
-1.14730515e-01 -2.54434735e-01 -1.00801301e+00 2.67339051e-01
6.17416263e-01 4.81534243e-01 -1.07401229e-01 1.20144987e+00
-4.94677480e-03 -2.64637172e-01 8.80989730e-01 -6.26651466e-01
-5.28784513e-01 9.94239509e-01 -2.88754582e-01 4.21077847e-01
8.89374852e-01 4.76378538e-02 7.03772306e-01 -6.13069117e-01
7.61450350e-01 1.30243325e+00 6.17897272e-01 2.26907164e-01
-1.30358243e+00 -6.05539978e-01 3.69110107e-01 -3.52650076e-01
-1.57727146e+00 -4.44261104e-01 4.07090992e-01 -1.02590466e+00
4.12158489e-01 2.70474285e-01 3.68818045e-01 1.00834000e+00
1.90716833e-01 4.14567709e-01 1.65741086e+00 -9.85053480e-01
-8.01339075e-02 9.17380095e-01 6.56374693e-01 4.96840775e-01
5.33958972e-01 -3.33787769e-01 -6.69811308e-01 -3.88955697e-02
2.74148643e-01 -3.31127703e-01 -2.13304996e-01 2.00577155e-01
-1.57874262e+00 5.82352340e-01 -2.19694063e-01 1.40757108e+00
-1.34575665e-01 -5.96956789e-01 5.71290791e-01 5.99462271e-01
2.71659642e-01 7.76565552e-01 -4.54749674e-01 -3.83743435e-01
-1.32898641e+00 8.64662603e-02 1.01687562e+00 7.00493574e-01
6.15174592e-01 -6.39669523e-02 -3.07168424e-01 9.21134412e-01
1.22095689e-01 1.69701889e-01 1.10568774e+00 -4.36323941e-01
4.15268302e-01 6.61113799e-01 3.35816965e-02 -6.72577560e-01
-2.63994187e-01 -4.98206407e-01 -2.95253277e-01 1.41617626e-01
9.74395335e-01 -1.25304341e-01 -5.36413789e-01 1.75846100e+00
-1.05672158e-01 -6.47288084e-01 1.49986446e-01 2.14391112e-01
7.68282890e-01 6.62906289e-01 4.52452213e-01 -3.32973480e-01
1.72094834e+00 -5.86706698e-01 -7.63650477e-01 -1.99352968e-02
8.20436358e-01 -1.22922897e+00 1.36195719e+00 4.37351614e-01
-1.00884998e+00 -7.89119422e-01 -9.61949766e-01 5.34860790e-02
-8.26098025e-01 2.17151433e-01 2.18154296e-01 1.10026014e+00
-1.21143997e+00 5.39464712e-01 -3.27467740e-01 -4.47821081e-01
-4.14272904e-01 5.71630448e-02 -5.20217180e-01 4.12219137e-01
-1.09372675e+00 8.44119132e-01 3.95289451e-01 -4.67434615e-01
-2.88902938e-01 -4.65119153e-01 -7.93015301e-01 -1.58864021e-01
-1.22474320e-01 2.75738478e-01 1.05467701e+00 -1.11298656e+00
-1.21175706e+00 1.53062832e+00 -2.73780197e-01 -8.91536325e-02
5.26543975e-01 9.53840464e-02 -8.27746391e-01 -2.39869565e-01
5.55704236e-01 3.20697725e-01 4.01852190e-01 -1.07571948e+00
-8.53445709e-01 -2.32068017e-01 -1.08929738e-01 -6.72333241e-02
-5.25620878e-01 7.79256999e-01 -2.89444059e-01 -8.20291162e-01
2.38452218e-02 -5.10008991e-01 7.86785483e-02 -7.87479222e-01
-3.55925113e-01 -8.13199461e-01 2.94955164e-01 -7.40549862e-01
1.59702098e+00 -2.28880429e+00 -6.06243797e-02 2.12485269e-01
8.20675194e-02 1.13268733e-01 3.19762766e-01 5.94808102e-01
-4.92094755e-02 6.87181592e-01 2.36035094e-01 -4.33419496e-01
2.56813228e-01 -8.68370682e-02 -1.06962919e-01 4.63969797e-01
5.10713086e-03 4.69959199e-01 -6.42833173e-01 -7.04623938e-01
-1.64086297e-01 2.44788587e-01 7.54426345e-02 -1.18124839e-02
2.36600950e-01 3.43796462e-01 -2.71322429e-01 5.84988356e-01
2.73752630e-01 1.19685479e-01 2.21410051e-01 4.79466766e-01
-1.09649014e+00 7.55958378e-01 -1.21392369e+00 1.22967935e+00
-5.38601518e-01 9.14867580e-01 6.87901257e-03 -5.87792516e-01
9.47688699e-01 3.00923675e-01 -5.92144988e-02 -5.49804986e-01
5.39076738e-02 5.70658565e-01 2.27054536e-01 -1.48522854e-01
8.53013575e-01 9.07061994e-02 -5.89788556e-01 4.16311741e-01
-2.68366188e-02 9.10285637e-02 8.85919452e-01 1.12613305e-01
4.99009997e-01 -1.48061618e-01 5.15859663e-01 -1.06633997e+00
9.26069975e-01 -1.15033083e-01 3.09336811e-01 9.65129375e-01
-2.24252522e-01 2.40030929e-01 7.42202342e-01 -7.02407723e-03
-9.14897084e-01 -9.65667069e-01 -5.86354315e-01 1.29791522e+00
-2.20698744e-01 -4.00588393e-01 -7.78374016e-01 -5.67681849e-01
-1.29254416e-01 5.40964067e-01 -3.61059278e-01 3.11096340e-01
-4.39898044e-01 -4.45395023e-01 5.26916265e-01 2.76265759e-02
8.05468336e-02 -1.19990599e+00 -1.03869744e-01 -1.20854843e-02
1.04026541e-01 -1.23502946e+00 -5.51190257e-01 2.06561670e-01
-4.98387128e-01 -7.77813017e-01 -9.85552907e-01 -1.25143516e+00
6.18593097e-01 -5.53963073e-02 1.16451299e+00 2.28168458e-01
3.49349916e-01 5.79347014e-01 -3.71426582e-01 -3.53979498e-01
-9.12068009e-01 6.85345531e-01 2.26977766e-01 -1.15330599e-01
5.69862843e-01 -1.64067477e-01 1.56863496e-01 1.35118008e-01
-6.45186603e-01 -5.22498012e-01 3.73853683e-01 2.20125124e-01
9.73866433e-02 9.28146765e-02 5.62178046e-02 -9.99784768e-01
8.89256775e-01 -3.09240937e-01 -5.61708987e-01 2.46282950e-01
-4.97740775e-01 5.38184457e-02 6.12208486e-01 -6.41743720e-01
-6.20169520e-01 -8.45085233e-02 -1.23117015e-01 4.97183472e-01
-6.13496065e-01 5.19551933e-01 -1.53677016e-01 -2.89265156e-01
4.62557256e-01 4.11247104e-01 -3.53024811e-01 -7.05769479e-01
-2.38453984e-01 9.39023435e-01 5.71294010e-01 -6.99382484e-01
9.37700093e-01 -1.79282948e-01 -4.43341374e-01 -1.24544966e+00
-6.72682881e-01 -8.16069186e-01 -1.05517089e+00 -2.48472989e-01
7.70191967e-01 -1.00672853e+00 -5.50755441e-01 6.02342069e-01
-9.14445579e-01 -3.20779949e-01 -1.11614808e-01 6.86890483e-01
1.52251586e-01 5.73903382e-01 -4.70084220e-01 -7.54379869e-01
1.16401702e-01 -1.31255937e+00 6.67702138e-01 3.05495203e-01
-7.11388111e-01 -1.43559039e+00 7.70860678e-03 1.03364378e-01
9.58314314e-02 -2.08557816e-03 7.89681137e-01 -9.04831350e-01
-5.12693971e-02 -1.72191247e-01 1.31622583e-01 3.78310651e-01
1.99946016e-01 1.81887060e-01 -7.20704496e-01 -1.73477426e-01
-7.47580975e-02 -1.86464503e-01 6.25122190e-01 6.29146248e-02
7.14685440e-01 -2.87804782e-01 -1.10044993e-01 2.23632455e-01
1.32039511e+00 2.31914446e-02 2.06279337e-01 9.42102551e-01
4.38688368e-01 8.03205550e-01 1.11333176e-01 7.32135251e-02
3.79029274e-01 5.69464624e-01 -5.63623667e-01 -1.33802921e-01
-8.49651247e-02 -2.84495175e-01 7.62844145e-01 1.20437467e+00
3.08798254e-01 -1.36018962e-01 -1.19005299e+00 7.67144620e-01
-1.01146972e+00 -7.48338819e-01 -4.50788319e-01 2.39310360e+00
1.02475560e+00 5.31311870e-01 4.94441062e-01 3.91694725e-01
7.85817444e-01 2.08495960e-01 4.71150041e-01 -8.33104610e-01
-5.32747626e-01 1.61620796e-01 4.70517427e-01 9.57763433e-01
-9.76375282e-01 1.13189399e+00 7.64020967e+00 7.81946003e-01
-1.38818419e+00 -1.77172318e-01 5.94326138e-01 5.96598089e-01
-1.72289997e-01 -7.82347843e-02 -1.21606326e+00 6.88442528e-01
1.09407151e+00 -4.32076186e-01 4.43473458e-02 5.08076489e-01
1.24899581e-01 -3.38676602e-01 -1.05868208e+00 4.59165007e-01
-6.41624676e-04 -7.68351495e-01 -1.44759631e-02 1.75685048e-01
6.50255203e-01 -1.23723716e-01 1.47358745e-01 2.98986346e-01
2.61994749e-01 -8.12324405e-01 1.27552986e+00 2.46604592e-01
6.76751018e-01 -5.06632030e-01 3.73398095e-01 4.62182969e-01
-9.01913524e-01 1.58392861e-01 -1.12852499e-01 -4.43525970e-01
-2.74640292e-01 2.49275938e-01 -5.92753947e-01 1.90666050e-01
4.51680839e-01 1.09550111e-01 -1.01761484e+00 9.17063832e-01
-3.01657408e-01 9.95784044e-01 -6.01955876e-02 -3.21333289e-01
5.28373301e-01 -6.06144778e-02 6.60376430e-01 1.81424189e+00
2.38146558e-01 -5.21009028e-01 4.14930761e-01 4.57019746e-01
1.95202217e-01 7.20088422e-01 -2.74297446e-01 -2.68110722e-01
5.16999364e-01 1.10707152e+00 -1.08810735e+00 -5.04678845e-01
-5.98560035e-01 7.45134592e-01 2.90085584e-01 2.39903361e-01
-6.02029487e-02 -5.73391557e-01 5.68318665e-01 4.67831254e-01
2.70001907e-02 -6.11117065e-01 -3.68161321e-01 -1.09851098e+00
1.01262331e-01 -8.31056297e-01 3.32944244e-01 -9.27031562e-02
-1.06015599e+00 6.55051887e-01 -9.70279723e-02 -9.01917040e-01
-2.88172334e-01 -9.81321335e-01 -5.01208007e-01 1.23797417e+00
-1.38009596e+00 -8.00504506e-01 8.63849744e-02 3.72388989e-01
3.29719275e-01 -6.42302275e-01 8.33624661e-01 2.68935233e-01
-5.53888738e-01 8.18495750e-01 2.40891770e-01 5.68292797e-01
6.49054945e-01 -1.54714298e+00 1.95165172e-01 1.00192142e+00
5.61661422e-01 1.05595827e+00 8.76942635e-01 -4.51266021e-01
-6.79991901e-01 -4.48779672e-01 1.84100688e+00 -5.77775955e-01
1.07678139e+00 -6.06699646e-01 -9.28822815e-01 3.66586417e-01
5.40532529e-01 -5.97733319e-01 7.01682389e-01 5.63910127e-01
-3.71584803e-01 -3.18195634e-02 -7.98716664e-01 5.17912209e-01
5.95728397e-01 -8.30553412e-01 -7.36026049e-01 3.58613014e-01
5.52084208e-01 -7.54841790e-02 -9.70509589e-01 -1.70232207e-01
3.11938226e-01 -9.09069538e-01 5.51346898e-01 -4.29758161e-01
2.24045739e-01 -1.15239024e-01 1.89680398e-01 -1.08892560e+00
-5.64733505e-01 -8.76523972e-01 7.44712889e-01 1.81378973e+00
6.40588045e-01 -6.95063114e-01 5.69086730e-01 3.98605973e-01
-1.35942370e-01 -5.81487864e-02 -8.94172370e-01 -8.51556480e-01
5.39971709e-01 -2.21731260e-01 4.85599577e-01 1.29360127e+00
3.17684025e-01 3.32475096e-01 2.63044983e-01 -1.37863597e-02
2.04624414e-01 -1.19110778e-01 6.20241523e-01 -1.40364158e+00
-4.12439816e-02 -1.12515891e+00 -5.97058713e-01 -6.39255047e-01
5.82435071e-01 -9.80720818e-01 -2.70096004e-01 -1.06792521e+00
-4.68792170e-02 -3.71327311e-01 -3.08335066e-01 1.91849485e-01
-1.98471069e-01 4.90835905e-02 2.73167044e-01 2.56240308e-01
-4.84442383e-01 -2.69430399e-01 8.43952656e-01 -1.46190137e-01
-3.62520576e-01 2.90211201e-01 -9.19880271e-01 7.80546486e-01
6.58811450e-01 -3.67100626e-01 1.57168344e-01 -2.11095825e-01
8.06446448e-02 -3.09031010e-01 -2.06076294e-01 -1.20420384e+00
2.63898730e-01 -1.63766190e-01 3.85753304e-01 -2.02963755e-01
-1.55266315e-01 -5.63904464e-01 -2.33937085e-01 5.51655114e-01
-4.26168025e-01 5.41174769e-01 2.78174281e-01 -3.14824432e-01
-4.13831204e-01 -6.85313344e-01 7.02922344e-01 -2.56387442e-01
-6.84937835e-01 -1.70861304e-01 -1.00903773e+00 2.11429894e-01
7.68772542e-01 -5.57749093e-01 1.90264225e-01 -7.90565535e-02
-5.85865080e-01 4.65499563e-03 7.25014091e-01 5.15595138e-01
-2.60744631e-01 -1.08212245e+00 -8.73334765e-01 1.17411681e-01
2.01425701e-01 -8.23954105e-01 -3.21166784e-01 4.15612727e-01
-5.61281562e-01 7.81538069e-01 -1.47189096e-01 -3.53706717e-01
-1.50939095e+00 4.41054076e-01 2.99640179e-01 -4.90734905e-01
-1.59415469e-01 5.15122950e-01 1.84770659e-01 -4.21925932e-01
4.43026543e-01 -1.74181595e-01 -7.54814923e-01 4.53552783e-01
4.74877805e-01 1.70983270e-01 -7.02865496e-02 -1.17867637e+00
-2.11351693e-01 5.72011232e-01 -2.67230719e-01 -4.76497799e-01
6.77666128e-01 -2.03147948e-01 -3.92922789e-01 1.20165312e+00
9.43231165e-01 9.92555439e-01 -3.53398144e-01 -2.68124163e-01
6.02086484e-01 -1.94480285e-01 -1.08599938e-01 -7.35451519e-01
-2.92326123e-01 5.90021312e-01 3.31351548e-01 6.45836353e-01
5.90099871e-01 1.07491827e-02 1.27604142e-01 1.31376490e-01
1.79324701e-01 -1.25890768e+00 -3.34901989e-01 8.71741831e-01
6.86797261e-01 -9.81102645e-01 -1.58343781e-02 -3.14446956e-01
-3.97117466e-01 1.02122796e+00 5.63440323e-01 3.47447038e-01
6.81346595e-01 8.95906910e-02 4.04833794e-01 1.85133457e-01
-1.92119583e-01 -3.05923611e-01 4.75905985e-01 3.63322258e-01
1.22029483e+00 2.97307670e-01 -9.89385784e-01 4.63862568e-01
-7.31646121e-01 -5.79437733e-01 6.54602230e-01 7.06968367e-01
-5.70528686e-01 -1.46921611e+00 -8.70096147e-01 1.50120288e-01
-1.07938552e+00 -1.70240328e-01 -8.12806606e-01 1.02231753e+00
3.59516531e-01 8.43589127e-01 3.14871252e-01 -1.93657026e-01
2.25330114e-01 2.65830398e-01 2.22340390e-01 -6.74057364e-01
-1.16706443e+00 -5.46156317e-02 5.87079637e-02 2.44360849e-01
-3.12102318e-01 -1.23110354e+00 -9.15025949e-01 -3.78604442e-01
-2.02906895e-02 8.96530092e-01 5.87037623e-01 9.67059493e-01
-2.74033725e-01 1.54642910e-01 5.82568586e-01 -2.93573290e-01
-2.23314464e-01 -1.22452855e+00 -8.31734717e-01 3.73640448e-01
3.96842271e-01 -2.58218944e-01 -5.79194427e-01 -3.37218828e-02] | [10.444252014160156, 10.379717826843262] |
4a2a6e09-638f-496a-91cb-67f191d64d80 | deep-perceptual-enhancement-for-medical-image | null | null | https://ieeexplore.ieee.org/document/9759833 | https://ieeexplore.ieee.org/document/9759833 | Deep Perceptual Enhancement for Medical Image Analysis | Due to numerous hardware shortcomings, medical image acquisition devices are susceptible to producing low-quality (i.e., low contrast, inappropriate brightness, noisy, etc.) images. Regrettably, perceptually degraded images directly impact the diagnosis process and make the decision-making manoeuvre of medical practitioners notably complicated. This study proposes to enhance such low-quality images by incorporating end-to-end learning strategies for accelerating medical image analysis tasks. To the best concern, this is the first work in medical imaging which comprehensively tackles perceptual enhancement, including contrast correction, luminance correction, denoising, etc., with a fully convolutional deep network. The proposed network leverages residual blocks and a residual gating mechanism for diminishing visual artefacts and is guided by a multi-term objective function to perceive the perceptually plausible enhanced images. The practicability of the deep medical image enhancement method has been extensively investigated with sophisticated experiments. The experimental outcomes illustrate that the proposed method could outperform the existing enhancement methods for different medical image modalities by 5.00 to 7.00 dB in peak signal-to-noise ratio (PSNR) metrics and 4.00 to 6.00 in DeltaE metrics. Additionally, the proposed method can drastically improve the medical image analysis tasks’ performance and reveal the potentiality of such an enhancement method in real-world applications. | ['S. M. A. Sharif; Rizwan Ali Naqvi; Mithun Biswas; Woong-Kee Loh'] | 2022-10-01 | null | null | null | ieee-journal-of-biomedical-and-health-4 | ['medical-image-enhancement', 'image-enhancement', 'decision-making'] | ['computer-vision', 'computer-vision', 'reasoning'] | [ 7.25633264e-01 -2.47787312e-02 3.71830702e-01 -2.46207803e-01
-7.77893424e-01 -1.13308541e-02 2.76422143e-01 1.21724129e-01
-4.93895352e-01 5.11096358e-01 8.07991624e-02 -3.63808244e-01
-3.08346480e-01 -4.98224467e-01 -4.14788902e-01 -1.04251516e+00
-2.21343696e-01 -6.57519221e-01 -6.95503205e-02 -1.94167554e-01
2.78304011e-01 2.99343109e-01 -1.45779383e+00 3.97193104e-01
9.65611100e-01 1.27780187e+00 4.25061673e-01 8.44612837e-01
5.49105465e-01 9.48715508e-01 -5.69275677e-01 -4.61266100e-01
2.01568052e-01 -4.24531877e-01 -6.30214214e-01 3.51177216e-01
1.84366241e-01 -6.54631555e-01 -4.36450064e-01 1.38439643e+00
9.72146451e-01 8.22526962e-02 3.20940405e-01 -7.11784422e-01
-8.45866263e-01 1.53044894e-01 -8.12203467e-01 5.88311911e-01
-3.13621946e-02 4.95350182e-01 3.22855294e-01 -7.68839717e-01
3.62571061e-01 8.74025822e-01 6.14041805e-01 3.70124876e-01
-9.65544999e-01 -3.31365675e-01 -3.96834612e-01 3.27379078e-01
-1.02814662e+00 -4.95018929e-01 8.55042756e-01 -1.08940616e-01
4.85986352e-01 4.33917463e-01 2.55370080e-01 7.07452774e-01
7.94687808e-01 6.45770669e-01 1.39509177e+00 -2.42052868e-01
1.45729464e-02 -1.67030562e-02 -3.12482119e-01 6.39370501e-01
2.39824980e-01 4.26635027e-01 -1.67194963e-01 3.01226914e-01
7.53876746e-01 1.55346557e-01 -5.07106602e-01 8.72239396e-02
-1.23812544e+00 3.44659686e-01 7.34057307e-01 4.04174805e-01
-6.66870356e-01 1.05418034e-01 5.23212612e-01 2.83667862e-01
2.15102464e-01 4.73051637e-01 -1.83570191e-01 5.56219630e-02
-9.27691281e-01 -1.91688627e-01 -2.49735992e-02 4.10851777e-01
1.11947387e-01 2.56404191e-01 -4.66452420e-01 8.65822196e-01
1.60236135e-01 4.08598244e-01 5.45160949e-01 -1.00788295e+00
2.55909085e-01 1.47478878e-01 1.75083026e-01 -1.16047060e+00
-4.85611230e-01 -9.23617065e-01 -1.43912053e+00 5.76129377e-01
1.99503019e-01 -3.11280768e-02 -9.09121692e-01 1.36446571e+00
3.23371619e-01 2.40559946e-03 1.36096984e-01 1.19680333e+00
1.04437006e+00 4.65940982e-01 3.93539041e-01 -4.52767789e-01
1.54498267e+00 -8.60306501e-01 -1.16311955e+00 -9.98198017e-02
2.39596128e-01 -9.99586940e-01 1.01733017e+00 6.89723372e-01
-1.69641066e+00 -9.68226373e-01 -1.22820938e+00 -5.92127740e-02
1.16676033e-01 2.56352037e-01 6.04225993e-01 6.79722071e-01
-1.10134578e+00 8.07698190e-01 -9.03207064e-01 1.29695326e-01
6.77786827e-01 2.73425341e-01 -1.58339828e-01 -4.31069404e-01
-1.26172292e+00 7.65666664e-01 3.46629888e-01 5.40852427e-01
-8.34146500e-01 -8.74182224e-01 -7.72993147e-01 1.62097183e-03
1.14938349e-01 -8.86075795e-01 1.03307199e+00 -1.04601336e+00
-1.33752978e+00 8.17860901e-01 3.04670244e-01 -3.00589889e-01
6.39265239e-01 -1.26820430e-01 -7.82421350e-01 5.15569925e-01
-1.90134272e-01 4.64007258e-01 9.69503343e-01 -1.34572971e+00
-6.10713303e-01 -3.41372043e-01 -9.92541835e-02 3.28038186e-01
-1.75655589e-01 1.79179847e-01 -4.60519135e-01 -1.03135002e+00
1.15875453e-01 -4.63547885e-01 -3.90825063e-01 3.08661133e-01
-1.50990531e-01 5.05743921e-01 5.18907368e-01 -1.12273717e+00
1.26039624e+00 -2.31903839e+00 -3.87065113e-01 -3.99533883e-02
3.18232387e-01 4.60894644e-01 -1.85886115e-01 -1.81937039e-01
-1.10367298e-01 -2.34511122e-02 -4.45155025e-01 -5.91716282e-02
-3.12739670e-01 -1.00134954e-01 2.40196094e-01 7.78110385e-01
2.85101354e-01 9.73359644e-01 -1.00265110e+00 -5.39967120e-01
5.76083601e-01 1.00824893e+00 -4.39408362e-01 3.21819514e-01
4.88052636e-01 6.69531763e-01 -1.47334248e-01 8.16422939e-01
1.02064061e+00 -3.57499868e-01 -7.19703138e-02 -5.91938555e-01
2.53471863e-02 -1.18037499e-01 -9.08748388e-01 1.49318671e+00
-5.77489972e-01 6.50959074e-01 3.12456936e-01 -9.07625794e-01
6.00712419e-01 4.45996404e-01 5.76109231e-01 -1.24582756e+00
3.76229048e-01 2.10526690e-01 2.17445761e-01 -8.74256968e-01
3.74612957e-01 -3.59711468e-01 3.75495255e-01 -2.00921576e-02
-2.86848783e-01 1.22965261e-01 -8.57019201e-02 -1.21260948e-01
9.50815022e-01 -2.38922790e-01 3.10070932e-01 -3.75787504e-02
6.31722867e-01 -4.97602284e-01 3.00683558e-01 7.04778314e-01
-6.81663573e-01 7.06082702e-01 3.90463211e-02 -1.70441076e-01
-1.17901099e+00 -1.18763244e+00 -3.65363270e-01 7.91705370e-01
4.16797072e-01 2.19217509e-01 -5.95588923e-01 -2.25050822e-01
-5.00009000e-01 3.89013648e-01 -4.78707403e-01 -5.43140948e-01
-4.93832409e-01 -1.03340578e+00 4.47049528e-01 4.65283304e-01
8.02408278e-01 -1.24204206e+00 -9.93806839e-01 2.65366763e-01
-4.58710670e-01 -1.13019907e+00 -3.81237745e-01 -2.08253693e-02
-1.05343771e+00 -8.28920722e-01 -1.08294225e+00 -9.47466075e-01
8.05988073e-01 3.53505880e-01 1.12313128e+00 2.79803842e-01
-6.55271053e-01 4.03723381e-02 -3.00963312e-01 -2.44721413e-01
-5.08424222e-01 -6.20871484e-01 -3.55603367e-01 -2.96991672e-02
-1.06069617e-01 -3.84502381e-01 -1.45212770e+00 2.38167062e-01
-1.32121849e+00 1.56258270e-01 1.11131167e+00 1.15567517e+00
5.42811811e-01 6.40283346e-01 5.24510324e-01 -6.31038964e-01
7.19074070e-01 -2.01718494e-01 -2.24868357e-01 1.67975411e-01
-7.07522511e-01 -1.03650093e-01 5.58453441e-01 -3.63483131e-01
-1.44702578e+00 -2.39991695e-01 -3.37782234e-01 -1.04003593e-01
-2.13220179e-01 4.61809307e-01 -1.35601744e-01 -1.51836887e-01
5.32564819e-01 4.07803476e-01 6.68172631e-03 -1.54198885e-01
1.06601022e-01 5.89218020e-01 9.64902878e-01 1.59571469e-02
5.76562166e-01 5.66579342e-01 1.69480547e-01 -6.55562282e-01
-5.79914987e-01 -4.72345829e-01 -7.99577460e-02 -4.45449501e-01
8.84885073e-01 -1.03310919e+00 -7.30623484e-01 6.77361190e-01
-9.19648588e-01 -1.17816165e-01 8.38181153e-02 5.58335781e-01
-3.73421013e-01 7.78540134e-01 -9.51734006e-01 -6.76168203e-01
-7.35646486e-01 -1.29629183e+00 8.63585532e-01 2.81701744e-01
1.95192710e-01 -8.82818699e-01 -5.06377935e-01 6.09034896e-01
7.85101950e-01 4.49040800e-01 7.68779397e-01 2.47461304e-01
-3.67239118e-01 9.70257446e-02 -5.28442025e-01 7.05618799e-01
2.72068620e-01 -4.10998940e-01 -1.08639717e+00 -5.52959740e-01
4.77741361e-01 -9.64872092e-02 7.92931795e-01 9.03454065e-01
1.58393979e+00 -2.30589673e-01 1.23794638e-01 8.54757309e-01
1.57414889e+00 4.74449456e-01 1.23710704e+00 3.35575879e-01
4.23698783e-01 4.69396919e-01 6.96002781e-01 5.46250880e-01
-7.52668381e-02 3.03955406e-01 7.01346159e-01 -8.11397254e-01
-5.59160233e-01 1.91535026e-01 1.55680865e-01 8.24230969e-01
-7.65139461e-02 -1.60640180e-01 -5.26933730e-01 6.26094341e-01
-1.26012754e+00 -7.64053106e-01 -1.00877307e-01 2.07823563e+00
9.97740686e-01 7.40537718e-02 -4.76068586e-01 5.06445587e-01
7.57219255e-01 -8.95363018e-02 -6.22289062e-01 -2.49145061e-01
-1.73344776e-01 3.53019357e-01 6.00409031e-01 1.95539832e-01
-1.25549698e+00 2.22161040e-01 5.62684488e+00 8.46325755e-01
-1.27269292e+00 1.83278114e-01 1.20293665e+00 1.56264484e-01
7.34788701e-02 -5.83915174e-01 1.04551382e-01 5.43791056e-01
8.13903689e-01 2.12418616e-01 3.19464713e-01 4.42537099e-01
5.82504451e-01 -1.85229033e-01 -5.87217093e-01 1.32761514e+00
1.55560942e-02 -1.05943453e+00 -1.56059265e-01 -1.33406326e-01
8.09395134e-01 -3.75907183e-01 7.28832603e-01 -7.08810315e-02
-3.44178885e-01 -1.22931588e+00 5.05315483e-01 4.19046670e-01
9.99736130e-01 -9.42331433e-01 1.03169441e+00 -1.24102727e-01
-7.64583468e-01 -9.69101861e-02 -2.30117232e-01 1.90723866e-01
2.30683655e-01 8.46519113e-01 -4.45194393e-01 5.43155432e-01
8.22076678e-01 4.94810492e-01 -5.94750166e-01 1.23982203e+00
-1.31190762e-01 5.30477524e-01 3.20827186e-01 4.96160358e-01
2.61982918e-01 1.17345743e-01 3.95735413e-01 1.28220761e+00
4.06729609e-01 4.54674333e-01 -3.76934946e-01 5.39435267e-01
-3.87664922e-02 -5.64630218e-02 -1.02855295e-01 2.78997213e-01
-9.43180174e-02 1.30890369e+00 -7.11659849e-01 -3.29665065e-01
-1.97495252e-01 1.14069545e+00 -5.04486322e-01 4.60740447e-01
-9.29496884e-01 -5.67680240e-01 2.09167182e-01 1.67250365e-01
1.72177702e-01 1.80136949e-01 -5.42313695e-01 -8.43187034e-01
1.69213951e-01 -1.10612285e+00 2.79011220e-01 -9.48983848e-01
-1.16401422e+00 8.05200577e-01 -4.76178497e-01 -1.36932063e+00
2.54817665e-01 -6.44882560e-01 -3.73157859e-01 7.77780473e-01
-1.96525180e+00 -8.64564478e-01 -5.30852735e-01 6.72554374e-01
4.93676901e-01 1.11183375e-01 3.64380211e-01 8.23644876e-01
-3.23318571e-01 7.09681451e-01 2.47209549e-01 -1.05802238e-01
6.65219605e-01 -1.12968564e+00 8.36741179e-02 1.07090819e+00
-6.02606297e-01 4.22177404e-01 8.91920567e-01 -3.39569360e-01
-1.32972455e+00 -1.16745794e+00 4.85486865e-01 1.78658649e-01
2.81434238e-01 2.38797635e-01 -1.03276885e+00 -1.35845348e-01
4.56999332e-01 2.60260373e-01 3.80988866e-01 -5.72439969e-01
1.19137384e-01 -1.63976580e-01 -1.51281834e+00 6.69483840e-01
6.31025732e-01 -2.87944943e-01 -2.40548059e-01 9.54306200e-02
5.34022391e-01 -4.57921416e-01 -1.06915319e+00 6.21119738e-01
5.13290703e-01 -1.06342673e+00 1.26668668e+00 -1.76332906e-01
9.13309813e-01 -4.07316208e-01 7.60096684e-02 -1.18156397e+00
-4.56036150e-01 -5.16242504e-01 -4.13156562e-02 8.72336745e-01
1.83981299e-01 -3.13529015e-01 3.93651307e-01 3.79248798e-01
-2.95905650e-01 -8.91218543e-01 -7.30232835e-01 -4.03161913e-01
-1.88816875e-01 -3.77676129e-01 2.57898659e-01 7.70848095e-01
-2.62184680e-01 -1.21248774e-01 -6.63988233e-01 4.19302642e-01
8.76725435e-01 -2.55516917e-01 1.45581439e-01 -5.99720538e-01
-4.20090914e-01 -5.63587487e-01 -4.19068873e-01 -9.45743263e-01
-5.84475994e-01 -5.12286901e-01 1.52339324e-01 -1.54470277e+00
2.90565819e-01 -3.47001761e-01 -8.39119971e-01 1.03018142e-01
-6.36905909e-01 6.16693795e-01 2.91498769e-02 -2.90209148e-02
-5.74182272e-01 3.53313684e-01 1.83465552e+00 -3.91296089e-01
2.72353869e-02 -1.07659861e-01 -9.23234761e-01 5.22011578e-01
9.42429781e-01 -2.35709473e-01 -3.91722530e-01 -4.95160103e-01
1.05084904e-01 2.17101336e-01 5.56811392e-01 -1.14794326e+00
1.66162685e-01 4.71193083e-02 7.61170983e-01 -4.39870477e-01
1.16850428e-01 -8.25819254e-01 1.01811498e-01 8.29602659e-01
-4.39651042e-01 -1.88732389e-02 2.48108134e-01 3.76683533e-01
-5.13113201e-01 3.31098475e-02 1.28334117e+00 -9.72237661e-02
-8.19609940e-01 1.37473062e-01 -2.01586992e-01 -1.59738719e-01
9.10526276e-01 -4.67282057e-01 -6.91964328e-02 -2.62833655e-01
-5.39653599e-01 -1.51201695e-01 5.97290881e-02 1.16694972e-01
1.11334777e+00 -1.14212728e+00 -8.73439014e-01 8.48775506e-02
4.49301600e-02 -1.95200503e-01 9.70556080e-01 1.24125826e+00
-6.74837232e-01 5.26599772e-02 -4.82407898e-01 -6.18842900e-01
-1.35849023e+00 7.49764740e-01 4.46062803e-01 -5.04362762e-01
-6.73742414e-01 7.20120013e-01 3.22372168e-01 1.51283905e-01
2.99209982e-01 -4.22115713e-01 -1.40570402e-01 -5.10684788e-01
9.06643212e-01 4.67876434e-01 2.96970278e-01 -4.80973244e-01
-2.15715289e-01 3.42575729e-01 -1.87071636e-01 1.25625759e-01
1.31386030e+00 -5.18603921e-01 1.09066933e-01 -1.72924086e-01
1.15527189e+00 -3.23694646e-01 -1.42776072e+00 -9.39440280e-02
-3.46764982e-01 -6.90389693e-01 6.22014284e-01 -1.33970296e+00
-1.54681063e+00 9.62078333e-01 1.35590816e+00 8.47225077e-03
1.95417094e+00 -3.60793591e-01 1.05355108e+00 -2.20745653e-01
-4.50296365e-02 -1.04949796e+00 3.35863262e-01 -2.22172439e-01
8.88909519e-01 -1.65146720e+00 1.24108344e-01 -3.32622945e-01
-6.92927837e-01 9.80034769e-01 2.60098547e-01 1.82158753e-01
4.19547647e-01 2.63368577e-01 3.53875607e-01 -1.80244580e-01
-3.88470769e-01 1.03148930e-02 4.60530281e-01 6.30068600e-01
5.27018905e-01 -9.86518189e-02 -4.62117255e-01 3.61731142e-01
2.29230374e-01 1.50337696e-01 4.82059896e-01 7.66764283e-01
-3.01921815e-01 -4.97723728e-01 -6.01234674e-01 3.26699078e-01
-1.14951503e+00 -2.96601802e-01 5.23825824e-01 5.31867683e-01
3.77492368e-01 1.38142872e+00 -2.18542963e-01 -5.19564264e-02
3.75338972e-01 -8.24386954e-01 4.18873549e-01 1.21010475e-01
-7.32056081e-01 3.77558529e-01 -2.41669074e-01 -4.68130738e-01
-6.95476830e-01 -1.78816378e-01 -1.11554003e+00 -2.70815521e-01
-1.26187041e-01 -3.99139583e-01 6.51426017e-01 5.28868318e-01
1.59566894e-01 1.18287182e+00 7.07633853e-01 -5.05832911e-01
-4.87900615e-01 -8.03102493e-01 -6.11931980e-01 8.23582530e-01
6.73355162e-01 -3.33269924e-01 -1.42930597e-01 4.02874082e-01] | [13.318652153015137, -2.4281959533691406] |
6dd4f001-66f4-4c99-9c36-2d1d9e2ea728 | world-knowledge-in-multiple-choice-reading | 2211.07040 | null | https://arxiv.org/abs/2211.07040v2 | https://arxiv.org/pdf/2211.07040v2.pdf | World Knowledge in Multiple Choice Reading Comprehension | Recently it has been shown that without any access to the contextual passage, multiple choice reading comprehension (MCRC) systems are able to answer questions significantly better than random on average. These systems use their accumulated "world knowledge" to directly answer questions, rather than using information from the passage. This paper examines the possibility of exploiting this observation as a tool for test designers to ensure that the use of "world knowledge" is acceptable for a particular set of questions. We propose information-theory based metrics that enable the level of "world knowledge" exploited by systems to be assessed. Two metrics are described: the expected number of options, which measures whether a passage-free system can identify the answer a question using world knowledge; and the contextual mutual information, which measures the importance of context for a given question. We demonstrate that questions with low expected number of options, and hence answerable by the shortcut system, are often similarly answerable by humans without context. This highlights that the general knowledge 'shortcuts' could be equally used by exam candidates, and that our proposed metrics may be helpful for future test designers to monitor the quality of questions. | ['Mark Gales', 'Vatsal Raina', 'Adian Liusie'] | 2022-11-13 | null | null | null | null | ['general-knowledge'] | ['miscellaneous'] | [ 2.18919009e-01 6.46883965e-01 9.38935354e-02 -3.38967234e-01
-1.06642151e+00 -1.04206681e+00 4.14447993e-01 9.28862810e-01
-5.29525638e-01 8.56008828e-01 3.01378131e-01 -7.91248918e-01
-7.52606690e-01 -1.08099616e+00 -4.58654553e-01 1.48493260e-01
4.89065021e-01 3.41705412e-01 7.37512469e-01 -5.52510977e-01
9.16348040e-01 -4.83970298e-03 -1.79338229e+00 4.57614005e-01
1.35454297e+00 4.87597048e-01 3.69891018e-01 1.01003957e+00
-7.30367959e-01 1.06589651e+00 -9.71248984e-01 -4.15902108e-01
-2.63271749e-01 -8.84749651e-01 -1.63657618e+00 -4.16402251e-01
6.35378182e-01 1.40865343e-02 1.77046910e-01 1.11618984e+00
2.40808800e-01 3.23019236e-01 4.06601906e-01 -5.73300898e-01
-6.01370990e-01 5.63202679e-01 4.25044179e-01 5.19955516e-01
1.49896634e+00 2.83672288e-02 1.01796734e+00 -3.81705970e-01
7.27060914e-01 1.11045063e+00 2.39875287e-01 3.94879967e-01
-9.36303079e-01 -9.12785381e-02 4.07211557e-02 5.48896313e-01
-8.59075785e-01 -1.39927462e-01 2.73932964e-01 -5.95986962e-01
7.17957258e-01 7.76213408e-01 4.17853415e-01 5.93819201e-01
3.31057340e-01 3.96837443e-01 1.45188999e+00 -1.11352956e+00
4.52227175e-01 5.54976285e-01 8.90178621e-01 7.37354338e-01
3.67291003e-01 -2.71172166e-01 -7.01555789e-01 -2.09936649e-02
2.85334468e-01 -6.56884789e-01 -6.48226917e-01 -8.43663421e-03
-9.72820282e-01 6.50147915e-01 1.36449328e-03 5.42116046e-01
-1.01374492e-01 -4.46705699e-01 -2.48116925e-01 9.19594347e-01
-1.41079098e-01 1.52148962e+00 -5.44751525e-01 -6.97177231e-01
-5.68100154e-01 3.20274979e-01 1.60992086e+00 9.36464608e-01
6.03293300e-01 -7.46381998e-01 -5.38795114e-01 6.64181411e-01
4.63706285e-01 4.12347764e-01 6.75582945e-01 -9.77970779e-01
4.96421129e-01 1.20330417e+00 3.32468331e-01 -9.56060886e-01
-8.41901824e-02 -4.64308292e-01 2.62359291e-01 1.24532275e-01
8.97779882e-01 1.99861284e-02 -3.61454546e-01 1.51727772e+00
1.76622123e-01 -5.83042800e-01 3.50414962e-02 6.77457750e-01
8.78099918e-01 1.22313462e-01 -1.36041164e-01 -3.93074825e-02
1.57414448e+00 -5.31154633e-01 -1.00278485e+00 -1.29536554e-01
7.84764230e-01 -9.77737784e-01 1.53364861e+00 3.68886173e-01
-1.24080777e+00 -6.51794434e-01 -1.32065284e+00 -5.94525449e-02
-7.26299882e-01 -5.49264491e-01 -7.44128674e-02 1.07177579e+00
-9.14749622e-01 6.52756274e-01 -7.42098317e-02 -5.28831959e-01
-1.45774022e-01 4.17466015e-02 4.15545441e-02 -3.44990581e-01
-1.40972137e+00 1.27785730e+00 1.17749006e-01 -3.09673727e-01
-5.77127159e-01 -4.49050367e-01 -6.51575267e-01 2.55340189e-01
5.80239952e-01 -7.08926260e-01 1.46965873e+00 -5.96514046e-01
-1.36179471e+00 6.21592879e-01 -2.29983374e-01 7.45301545e-02
5.98188579e-01 -5.05391955e-01 -4.78391588e-01 4.02148068e-01
3.09717357e-01 7.51890987e-02 3.92638564e-01 -9.47700560e-01
-6.35770023e-01 -2.27755085e-01 5.91167867e-01 2.91811138e-01
-1.45026207e-01 -9.31842327e-02 -2.02781960e-01 4.73261364e-02
2.83199191e-01 -4.84491527e-01 9.59489867e-02 -2.62714833e-01
-2.81144917e-01 -4.60405827e-01 3.12503636e-01 -6.77940667e-01
1.67151594e+00 -1.47651780e+00 -1.61231950e-01 5.47955692e-01
2.55417585e-01 2.25404605e-01 -2.68042684e-01 5.91800392e-01
1.38428524e-01 6.19634449e-01 1.50634140e-01 7.07297444e-01
1.89789727e-01 -1.53190449e-01 1.49604529e-01 -1.75866053e-01
-1.41994851e-02 7.30097055e-01 -1.19444132e+00 -4.08021808e-01
1.37538403e-01 -1.97959617e-01 -2.84072280e-01 5.17551005e-01
-5.25334954e-01 1.07521139e-01 -5.63703597e-01 4.61547762e-01
2.79172838e-01 -3.19708556e-01 8.40021819e-02 5.33737063e-01
-3.19100507e-02 7.56291687e-01 -1.30337322e+00 1.30238008e+00
-6.08755469e-01 6.98879778e-01 -4.69008118e-01 -2.65800953e-01
1.08169258e+00 2.99928159e-01 -4.38226581e-01 -1.06350183e+00
-1.59454405e-01 2.97199637e-01 2.38458738e-01 -9.99651432e-01
4.63571906e-01 2.26172149e-01 8.61835033e-02 6.60577655e-01
-1.19884789e-01 -1.24520630e-01 5.32188535e-01 4.08068180e-01
1.53611958e+00 -1.95591804e-02 5.48334718e-01 -4.98623520e-01
8.25933337e-01 1.73598811e-01 -1.10635512e-01 1.45797372e+00
-8.13284069e-02 3.05953234e-01 6.23497128e-01 2.92801838e-02
-5.17304480e-01 -1.13095093e+00 3.71096656e-02 1.08058465e+00
2.39670694e-01 -5.77686369e-01 -1.02186513e+00 -8.50279510e-01
-2.43757069e-01 1.34979856e+00 -4.09388691e-01 -2.41526783e-01
-2.06002608e-01 1.78046018e-01 3.00139517e-01 1.79632306e-01
4.92071301e-01 -9.60973680e-01 -1.05015004e+00 3.61296475e-01
-6.39085829e-01 -5.41111052e-01 -1.98597297e-01 -7.98745751e-02
-8.08300972e-01 -1.34468520e+00 -3.10098588e-01 -5.34345388e-01
4.99302387e-01 1.41430870e-01 1.62948096e+00 7.75709271e-01
2.06246123e-01 1.15384555e+00 -8.46759915e-01 -3.37404191e-01
-5.49276114e-01 1.59215182e-01 -6.46084189e-01 -5.01509607e-01
7.88312614e-01 -1.31669089e-01 -5.71058869e-01 4.65957731e-01
-1.02368617e+00 -2.47976050e-01 4.13010627e-01 4.73699987e-01
7.09760264e-02 -1.85747162e-01 7.96599090e-01 -1.00133717e+00
1.34448242e+00 -5.80084085e-01 -2.89513588e-01 9.32875097e-01
-7.93514311e-01 4.16887939e-01 2.27439627e-01 -1.31661281e-01
-1.13451421e+00 -7.27228343e-01 -1.68438226e-01 6.19185746e-01
-3.38409901e-01 7.96391189e-01 -3.13477546e-01 -3.05817127e-01
1.07826173e+00 7.20799016e-03 -5.76537801e-03 -4.23904210e-01
1.96346402e-01 5.37072301e-01 1.38248682e-01 -8.15151334e-01
5.54819524e-01 -3.32340330e-01 -1.73449591e-01 -7.37500787e-01
-1.14699960e+00 -7.89272547e-01 -4.10473436e-01 -6.65901721e-01
8.61269534e-01 -1.90214843e-01 -8.73590887e-01 -1.89763427e-01
-1.02883768e+00 -4.00866847e-03 -4.08755153e-01 3.54342699e-01
-4.01917964e-01 2.82278866e-01 -2.15214580e-01 -9.02241230e-01
2.36462712e-01 -9.13583279e-01 3.56654584e-01 7.43016124e-01
-8.30282509e-01 -1.25827968e+00 2.98311822e-02 7.98875272e-01
5.64188898e-01 -5.03899418e-02 1.40176392e+00 -1.22944200e+00
-7.87445664e-01 -3.94410402e-01 1.28134564e-01 2.63259500e-01
-1.39028474e-03 -4.12741929e-01 -8.51386666e-01 1.59323066e-02
4.49721038e-01 -1.31611601e-01 4.03588027e-01 -1.15176342e-01
7.11460233e-01 -3.92200798e-01 -1.33308489e-02 -4.00810480e-01
1.54653156e+00 1.97611704e-01 8.51677954e-01 6.09025002e-01
1.01009615e-01 1.08018017e+00 6.29281700e-01 1.73761137e-03
2.25709811e-01 3.14091802e-01 5.47163002e-02 6.52792275e-01
-1.54790953e-01 -4.94069427e-01 1.14412390e-01 1.09900677e+00
2.67034531e-01 -3.34805608e-01 -1.10512614e+00 6.24162376e-01
-1.49518108e+00 -8.93693328e-01 -4.69934255e-01 2.44752407e+00
8.20507884e-01 4.18971270e-01 -3.23350966e-01 1.74051955e-01
4.85934436e-01 -3.86736542e-01 -1.34065822e-01 -6.03083909e-01
6.81827068e-02 5.99224031e-01 4.27276827e-02 1.01521063e+00
-3.00780892e-01 3.09416145e-01 6.81440878e+00 6.17841184e-01
-1.54670388e-01 -8.00905749e-02 2.30833247e-01 3.60186756e-01
-8.96794021e-01 4.94219840e-01 -6.71655178e-01 3.76171082e-01
1.20037186e+00 -5.65567732e-01 1.74990788e-01 5.19707978e-01
-8.55023414e-03 -7.15762556e-01 -1.28499401e+00 1.02589712e-01
3.13854665e-02 -1.17509401e+00 5.25337532e-02 -2.09213093e-01
4.78105426e-01 -7.98905015e-01 -2.03117833e-01 3.46789330e-01
2.69335151e-01 -1.02550995e+00 4.73606855e-01 1.23800898e+00
1.99995428e-01 -5.61664224e-01 8.43081474e-01 5.68728745e-01
-7.30764508e-01 -5.17234504e-02 -3.34545046e-01 -4.07571644e-01
-2.76240885e-01 3.66004527e-01 -1.04937387e+00 6.46255076e-01
2.29952037e-01 -2.04727903e-01 -1.29032516e+00 1.40040231e+00
-5.86202621e-01 7.71456778e-01 -1.14374474e-01 -8.84158254e-01
2.00233120e-03 6.99693412e-02 6.24509275e-01 1.03298628e+00
4.14899945e-01 2.28653014e-01 -2.66524125e-02 9.32624698e-01
2.54003733e-01 5.11131167e-01 -4.24784213e-01 1.01907916e-01
6.93283260e-01 7.15198219e-01 -6.04255557e-01 -3.72298568e-01
-3.77022624e-01 5.08376598e-01 1.08322099e-01 2.73762882e-01
-1.62719667e-01 -7.83476949e-01 2.64493600e-02 2.16176346e-01
-2.10159793e-01 1.15910418e-01 -4.53046143e-01 -8.10482681e-01
3.05801064e-01 -1.25257301e+00 5.37326753e-01 -9.98966753e-01
-1.11768186e+00 3.10909450e-01 3.49595770e-02 -1.09725976e+00
-2.43065372e-01 -6.91415131e-01 -6.85879409e-01 1.16479206e+00
-1.35297453e+00 -9.84757021e-02 -5.14384210e-01 2.75166571e-01
4.76461738e-01 9.79775265e-02 8.21211040e-01 -2.30374202e-01
-4.25386354e-02 4.97035205e-01 -1.64575383e-01 -1.63367465e-01
5.93367279e-01 -1.87772250e+00 -1.77793995e-01 8.33364189e-01
1.85607180e-01 1.08074570e+00 9.40228343e-01 -7.19350159e-01
-1.30901718e+00 -3.47178131e-01 1.36947775e+00 -8.87213647e-01
4.01043922e-01 2.05060422e-01 -1.37162626e+00 2.37015352e-01
5.81580579e-01 -8.55191410e-01 1.16673982e+00 3.80613983e-01
-2.74349034e-01 1.01307146e-01 -1.25390744e+00 7.13480890e-01
6.10331655e-01 -6.78147137e-01 -1.52371991e+00 3.68477285e-01
8.43022227e-01 -1.29785761e-01 -9.70414817e-01 2.62167715e-02
4.68520850e-01 -1.17924237e+00 5.02507210e-01 -6.49582922e-01
2.69902974e-01 -3.63521099e-01 -1.01019561e-01 -1.28119469e+00
-1.30589813e-01 -5.16760051e-01 -4.03782539e-02 1.15203905e+00
9.37813759e-01 -7.11247742e-01 3.61041456e-01 1.13822687e+00
6.38058260e-02 -4.26271051e-01 -7.94012487e-01 -6.33720815e-01
2.79895812e-01 -4.90121126e-01 6.55103147e-01 7.07108200e-01
5.54888725e-01 4.11190599e-01 5.30970216e-01 1.19002633e-01
2.48629212e-01 -1.79954872e-01 4.27756280e-01 -1.44611478e+00
-9.59932134e-02 -3.43640625e-01 -2.37644643e-01 -1.11091280e+00
-3.23020250e-01 -7.75523305e-01 5.71417212e-02 -1.64862823e+00
-1.04403816e-01 7.44189247e-02 -3.10481489e-01 -1.15133710e-01
-5.47194064e-01 -5.04249215e-01 2.20726907e-01 -1.65981442e-01
-8.07903707e-01 -1.36490613e-01 1.54612339e+00 1.84578896e-01
-9.59426239e-02 1.46344770e-02 -1.05249178e+00 6.79120481e-01
6.40811026e-01 -4.62543637e-01 -5.37758529e-01 -1.46182895e-01
8.77567053e-01 3.90397251e-01 3.19383591e-01 -1.21450567e+00
7.78091311e-01 -1.56977579e-01 4.29437041e-01 -3.11258674e-01
-2.33522430e-01 -6.87624097e-01 -2.13226050e-01 3.05248708e-01
-1.01374078e+00 3.52897584e-01 1.38817564e-01 4.22695071e-01
-2.81009167e-01 -1.23206782e+00 3.17394793e-01 -4.99642253e-01
-5.56723714e-01 -7.30204284e-01 -7.84261107e-01 4.59473908e-01
7.02165127e-01 -2.70913661e-01 -5.17488897e-01 -7.48188376e-01
-7.64526546e-01 3.75271678e-01 2.21535459e-01 2.20375031e-01
7.14025855e-01 -9.33717906e-01 -6.41433895e-01 -2.35379785e-02
3.54916692e-01 -5.78226984e-01 9.23530757e-02 4.49889272e-01
-6.43320203e-01 5.95007837e-01 -1.08956462e-02 -3.83590519e-01
-1.16592860e+00 1.33348972e-01 4.21670020e-01 -3.64620179e-01
-6.47290871e-02 6.72186732e-01 -5.42051256e-01 -4.42063928e-01
7.94621259e-02 -3.59800339e-01 -6.55987918e-01 2.42595226e-01
8.70301902e-01 5.52400053e-01 3.13523173e-01 2.53469944e-01
5.57607748e-02 3.83195907e-01 -2.45670304e-01 -4.52923208e-01
5.78777194e-01 -4.22273993e-01 -8.36199149e-02 7.23444700e-01
7.52658069e-01 2.88139850e-01 -4.33172703e-01 -2.19411612e-01
7.39244998e-01 -7.46913910e-01 -3.11153084e-01 -1.63330841e+00
2.58967653e-02 7.82691300e-01 5.16627789e-01 9.38614309e-01
9.57230985e-01 -1.46110356e-01 2.49237314e-01 8.84813488e-01
3.18317086e-01 -1.21829081e+00 3.70221019e-01 6.50318801e-01
8.52142811e-01 -9.68373060e-01 -2.22076938e-01 -4.32529479e-01
-3.45620632e-01 1.35874856e+00 8.38202119e-01 2.78677881e-01
3.61008495e-01 -2.16314375e-01 1.70557588e-01 -2.91061163e-01
-9.35452521e-01 -3.64537001e-01 6.44353986e-01 5.36614001e-01
6.87304080e-01 -9.25140679e-02 -6.87224269e-01 5.82839251e-01
-5.73964417e-01 -2.43207514e-02 1.03506482e+00 1.18898237e+00
-1.09500623e+00 -1.04814160e+00 -5.99355996e-01 7.56578505e-01
-4.12051737e-01 -1.09515250e-01 -8.11717331e-01 7.57257462e-01
-1.65079698e-01 1.64820230e+00 -3.14273596e-01 -3.12743522e-02
6.76704645e-01 6.55734122e-01 5.26666999e-01 -9.16385770e-01
-8.28345358e-01 -6.94169104e-01 4.17993188e-01 -3.53823245e-01
-1.98537424e-01 -4.47117001e-01 -7.37125397e-01 -9.33964103e-02
-6.90333486e-01 7.10466385e-01 3.71373564e-01 1.32749271e+00
1.72469139e-01 5.69503069e-01 2.57062256e-01 5.48390627e-01
-8.48862827e-01 -9.96394992e-01 -1.25644118e-01 4.60759163e-01
2.16387495e-01 -4.33131486e-01 -4.77187872e-01 -1.44595489e-01] | [11.467100143432617, 8.147520065307617] |
5e2554c2-d3f4-464f-ad10-e88bb1ba5f68 | new-benchmarks-for-learning-on-non | 2104.01404 | null | https://arxiv.org/abs/2104.01404v2 | https://arxiv.org/pdf/2104.01404v2.pdf | New Benchmarks for Learning on Non-Homophilous Graphs | Much data with graph structures satisfy the principle of homophily, meaning that connected nodes tend to be similar with respect to a specific attribute. As such, ubiquitous datasets for graph machine learning tasks have generally been highly homophilous, rewarding methods that leverage homophily as an inductive bias. Recent work has pointed out this particular focus, as new non-homophilous datasets have been introduced and graph representation learning models better suited for low-homophily settings have been developed. However, these datasets are small and poorly suited to truly testing the effectiveness of new methods in non-homophilous settings. We present a series of improved graph datasets with node label relationships that do not satisfy the homophily principle. Along with this, we introduce a new measure of the presence or absence of homophily that is better suited than existing measures in different regimes. We benchmark a range of simple methods and graph neural networks across our proposed datasets, drawing new insights for further research. Data and codes can be found at https://github.com/CUAI/Non-Homophily-Benchmarks. | ['Ser-Nam Lim', 'Felix Hohne', 'Xiuyu Li', 'Derek Lim'] | 2021-04-03 | null | null | null | null | ['node-classification-on-non-homophilic'] | ['graphs'] | [-1.74132697e-02 3.62004578e-01 -6.29463315e-01 -3.81318748e-01
1.54516399e-01 -6.01148605e-01 9.54355299e-01 5.83099902e-01
8.09574798e-02 7.92621315e-01 2.06138149e-01 -4.03179884e-01
-3.92344177e-01 -1.27042294e+00 -7.25790083e-01 -6.48774028e-01
-5.78778505e-01 7.29713023e-01 -1.65618099e-02 -1.69782534e-01
4.46546525e-02 3.10702205e-01 -1.06561756e+00 -1.17509052e-01
5.47115386e-01 5.32075405e-01 -2.59378076e-01 4.64112192e-01
3.00254107e-01 6.01455033e-01 -1.79864764e-01 -5.85718453e-01
4.58972842e-01 -5.02416432e-01 -8.14729512e-01 -3.39419067e-01
6.05765820e-01 3.65325660e-01 -7.22128808e-01 1.06869638e+00
5.04282653e-01 8.13990757e-02 8.91331375e-01 -1.63110185e+00
-7.30874002e-01 1.06537986e+00 -7.36913502e-01 1.25880539e-01
4.47311252e-01 -6.45588934e-02 1.79064155e+00 -4.87649530e-01
9.59281325e-01 1.26860321e+00 8.28605771e-01 1.90151319e-01
-1.33911264e+00 -7.38395035e-01 7.67615587e-02 2.46112086e-02
-1.31514823e+00 -1.26170352e-01 9.50045168e-01 -2.88050622e-01
6.10210180e-01 2.76044518e-01 8.18588793e-01 1.37238085e+00
1.79397076e-01 5.32368124e-01 1.32842302e+00 -2.94680804e-01
1.28390849e-01 1.23996042e-01 3.46876055e-01 7.91840732e-01
8.06778014e-01 1.52422920e-01 -3.92218739e-01 -2.89053589e-01
5.58050215e-01 2.76115745e-01 -2.06032008e-01 -1.02496648e+00
-9.96861756e-01 9.75390553e-01 8.52865994e-01 4.98314500e-01
-2.15505213e-01 1.01562269e-01 5.15953779e-01 8.19451392e-01
5.80802977e-01 8.41683030e-01 -8.89565870e-02 8.24851170e-02
-5.09730577e-01 -4.38251346e-03 1.22541821e+00 8.97463918e-01
9.54522491e-01 -2.52817243e-01 -8.47522542e-03 5.97599328e-01
-5.73206171e-02 1.04621485e-01 1.21154957e-01 -4.74743396e-01
2.78235137e-01 9.81811643e-01 -4.17630762e-01 -1.45930552e+00
-4.30704415e-01 -6.59821391e-01 -1.15508866e+00 -1.80288166e-01
4.07732844e-01 4.44222428e-02 -4.63923901e-01 1.88609099e+00
3.30209136e-01 3.07767630e-01 -1.57232270e-01 5.79753220e-01
8.99202764e-01 2.91881084e-01 -1.47370234e-01 -1.53635174e-01
9.28715825e-01 -8.40342402e-01 -4.58341748e-01 -9.00237355e-03
9.68552232e-01 -4.00186986e-01 1.40450478e+00 -2.70833746e-02
-8.22648823e-01 9.15800780e-02 -7.88050830e-01 2.61910170e-01
-6.49097264e-01 -7.51716316e-01 1.18349528e+00 7.58863986e-01
-1.27817690e+00 9.17558491e-01 -4.38359976e-01 -8.12304378e-01
5.29868603e-01 7.16943592e-02 -4.89198267e-01 -2.61257648e-01
-1.24804676e+00 7.85002589e-01 3.15847486e-01 -3.59341979e-01
-7.04493880e-01 -6.11157715e-01 -7.37839043e-01 1.63824171e-01
5.20381510e-01 -7.21823454e-01 5.15364170e-01 -7.93405473e-01
-9.27826107e-01 1.09352148e+00 2.95821995e-01 -6.42040312e-01
6.11349285e-01 2.38335371e-01 -3.12767893e-01 5.58361448e-02
3.01165390e-03 2.14593276e-01 5.62640548e-01 -1.30230689e+00
2.86819339e-02 -2.70408362e-01 1.93470478e-01 6.59635812e-02
-5.39823532e-01 -3.94315690e-01 -1.82534754e-01 -4.15421396e-01
-1.93819568e-01 -1.08206272e+00 -1.19958416e-01 -3.63291234e-01
-7.29888916e-01 -5.20166636e-01 6.55127048e-01 3.42432469e-01
1.32198691e+00 -1.74113202e+00 -1.04648881e-01 7.15127707e-01
8.67060363e-01 5.13650142e-02 -2.59930819e-01 1.00571728e+00
-3.12169701e-01 4.57737565e-01 -2.07402557e-01 -1.35929614e-01
5.01306839e-02 2.41801679e-01 2.82063577e-02 8.09633076e-01
-8.00459534e-02 9.35000479e-01 -1.13096404e+00 -5.03902733e-01
2.42795661e-01 2.03220546e-01 -4.78511602e-01 9.97250155e-02
3.24342623e-02 2.09867343e-01 -3.28316450e-01 8.76515269e-01
3.68737429e-01 -8.38574290e-01 4.44033772e-01 1.00959964e-01
4.87981230e-01 1.94636703e-01 -8.57153773e-01 1.18042672e+00
-3.43071431e-01 7.48471379e-01 -9.81022269e-02 -1.21203387e+00
1.04142225e+00 -3.16371545e-02 6.32507205e-01 -5.08041382e-01
3.56878430e-01 8.97410586e-02 3.84414494e-01 -1.77523181e-01
2.43276343e-01 -1.20378569e-01 1.06294051e-01 6.71795726e-01
-4.61616889e-02 -1.17664598e-01 4.47806299e-01 7.71624923e-01
1.66188788e+00 -5.44539273e-01 6.14280641e-01 -6.61731660e-01
-3.67648043e-02 -2.56864101e-01 2.87049353e-01 1.09993088e+00
-4.28306282e-01 3.72323543e-01 8.53763938e-01 -5.22732139e-01
-1.02814007e+00 -1.10621274e+00 -3.23479563e-01 8.97199392e-01
3.81587565e-01 -8.13835740e-01 -2.77632505e-01 -7.77618945e-01
3.38741750e-01 4.30952370e-01 -1.07178473e+00 -3.18389386e-01
-1.78705171e-01 -8.98433030e-01 4.55706477e-01 2.17205673e-01
7.62197226e-02 -9.32876050e-01 -5.23883104e-02 -2.37674132e-01
2.55470961e-01 -1.00719368e+00 -1.79097071e-01 2.30001211e-01
-9.09884155e-01 -1.57206118e+00 -5.68573296e-01 -6.49335206e-01
8.49527836e-01 5.36356568e-01 1.78723991e+00 5.95387936e-01
-1.24062024e-01 5.70723295e-01 -5.52716494e-01 -1.03445970e-01
-3.82137001e-01 4.21127647e-01 4.25356142e-02 -6.78690746e-02
4.47863877e-01 -1.03390419e+00 -6.88752294e-01 4.18566883e-01
-6.89731479e-01 -1.65355101e-01 4.73050833e-01 8.37518334e-01
2.61614114e-01 -3.46311145e-02 5.67854106e-01 -1.65796578e+00
8.64989936e-01 -1.13103807e+00 -2.95288920e-01 9.06365439e-02
-1.31028211e+00 -1.72526315e-01 7.83004642e-01 -1.61420092e-01
-4.92899269e-01 -3.13867718e-01 4.22626197e-01 -3.54811341e-01
7.24034458e-02 6.00926220e-01 1.71891481e-01 -1.44059807e-01
8.47485304e-01 -3.39070857e-01 1.87792405e-02 -9.33572575e-02
4.83549803e-01 3.09277147e-01 6.94013014e-02 -6.38462901e-01
8.49495947e-01 6.14921451e-01 4.04578477e-01 -8.19786787e-01
-5.35508096e-01 -5.38333058e-01 -5.34449518e-01 -2.58088648e-01
5.21591082e-02 -6.77386343e-01 -6.81437552e-01 1.26737833e-01
-5.15326738e-01 -5.58397293e-01 -3.31069469e-01 4.07867372e-01
-4.24510449e-01 5.00231981e-01 -5.63099325e-01 -6.82302833e-01
-2.83873916e-01 -6.49683714e-01 6.62591517e-01 7.03924149e-02
-5.06355047e-01 -1.65189064e+00 2.57024676e-01 1.48562957e-02
3.07640225e-01 5.94067454e-01 1.02415800e+00 -9.78983521e-01
-4.25489187e-01 -5.13081193e-01 -3.54787439e-01 -2.32052460e-01
1.63587928e-01 1.24762595e-01 -6.57291114e-01 -5.90406299e-01
-6.65702999e-01 -4.72172022e-01 1.11847210e+00 4.00677443e-01
8.34934592e-01 -2.63125896e-01 -6.11558974e-01 6.22373164e-01
1.37857628e+00 -3.43757868e-01 4.15845960e-01 2.26449847e-01
8.36707890e-01 6.96856737e-01 1.84536904e-01 3.67912710e-01
4.25584048e-01 3.81586105e-01 5.77715099e-01 -1.99278474e-01
-6.03598692e-02 -5.91857791e-01 -7.48583069e-03 9.03441727e-01
-4.08446416e-02 -7.73129225e-01 -1.16056180e+00 7.36427128e-01
-1.91430128e+00 -9.17642355e-01 -4.41021532e-01 2.06781673e+00
6.47201240e-01 1.97895125e-01 4.24392760e-01 2.04339363e-02
1.04044211e+00 5.60253203e-01 -5.00276268e-01 -3.40990603e-01
-3.76083583e-01 1.96629893e-02 5.94865918e-01 4.49502856e-01
-8.61530542e-01 1.03676629e+00 6.39859533e+00 5.59231400e-01
-8.58591795e-01 -1.53568044e-01 6.74200118e-01 -2.46089920e-02
-6.44907475e-01 4.28070813e-01 -2.74198234e-01 3.79798681e-01
8.63155067e-01 -3.24654102e-01 3.28078926e-01 9.79752600e-01
-1.66660279e-01 2.03863755e-02 -1.34701264e+00 8.75573635e-01
-7.59325922e-02 -1.34809160e+00 -3.92703973e-02 2.00265229e-01
9.64380562e-01 3.36376756e-01 -7.60543672e-03 1.40693307e-01
9.44070339e-01 -1.10887325e+00 -7.67416582e-02 2.63372511e-01
6.55580938e-01 -4.54154223e-01 6.20988965e-01 1.64991394e-02
-1.12604082e+00 1.96877688e-01 -3.48133326e-01 -3.57814461e-01
-1.00352950e-01 8.87813449e-01 -8.79893184e-01 5.27815402e-01
7.07966566e-01 1.15850639e+00 -7.45672047e-01 9.40723777e-01
-1.00581214e-01 7.57885158e-01 -3.76949012e-01 -3.36013734e-01
1.83902949e-01 -3.67496729e-01 6.13223612e-01 1.11440122e+00
1.03828929e-01 -3.53792638e-01 6.18201159e-02 7.93986917e-01
-5.25929630e-01 3.61999661e-01 -1.36051607e+00 -3.52897078e-01
7.38077343e-01 1.39993465e+00 -1.16295254e+00 -2.85063654e-01
-4.15596932e-01 4.94623423e-01 6.09778345e-01 2.38013312e-01
-4.70544666e-01 -3.74520928e-01 4.29223388e-01 4.18895692e-01
-2.48247534e-01 -8.32334533e-02 -3.56860071e-01 -1.21091747e+00
-2.69863158e-01 -7.54073262e-01 8.88911426e-01 -3.45734417e-01
-1.86732769e+00 5.33022463e-01 -4.55005690e-02 -9.26079512e-01
-1.49155423e-01 -4.73209143e-01 -9.79889274e-01 2.93489248e-01
-1.42033374e+00 -1.06399786e+00 -5.38681567e-01 6.72776341e-01
-1.84724942e-01 -2.07252741e-01 6.42495513e-01 1.61685705e-01
-5.64055681e-01 6.36362553e-01 6.99110851e-02 1.85520470e-01
8.79805565e-01 -1.54944658e+00 5.14667988e-01 6.29096329e-01
5.26765406e-01 8.25458586e-01 7.84361601e-01 -8.41883421e-01
-1.47050846e+00 -9.39226866e-01 6.99199438e-01 -6.31762981e-01
1.04549468e+00 -5.36755025e-01 -8.79241645e-01 8.27591658e-01
1.54469594e-01 3.56093347e-01 6.13448441e-01 8.44102025e-01
-5.80918312e-01 -8.12320560e-02 -9.12862062e-01 7.90508032e-01
1.66590488e+00 -6.10956669e-01 -1.06224850e-01 6.54204369e-01
2.85998195e-01 4.92036864e-02 -1.02514613e+00 4.84573573e-01
4.11174864e-01 -1.13654399e+00 8.52723181e-01 -8.01256478e-01
4.63315159e-01 1.63948059e-01 -3.20221968e-02 -1.55510557e+00
-3.54087025e-01 -8.84396076e-01 -1.48843214e-01 1.00446641e+00
5.07111609e-01 -1.02207005e+00 1.04413819e+00 3.47016245e-01
1.04767293e-01 -8.60409498e-01 -6.79104984e-01 -1.03965151e+00
2.70195603e-01 1.11654542e-01 5.42059898e-01 1.54549801e+00
4.57315624e-01 7.24137187e-01 -4.30750608e-01 -2.43694395e-01
6.01976395e-01 5.15003085e-01 1.08817232e+00 -1.57970786e+00
-8.58812705e-02 -7.94504464e-01 -5.66078603e-01 -5.34064531e-01
4.43896472e-01 -1.52198076e+00 -4.14033681e-01 -1.44379067e+00
4.74950135e-01 -6.59188509e-01 -2.65707701e-01 3.25120956e-01
-1.21405177e-01 2.30353475e-01 3.37180146e-03 3.85462552e-01
-5.99778175e-01 5.44485390e-01 1.35623252e+00 -1.15581237e-01
-1.10566385e-01 4.50515524e-02 -7.75809884e-01 5.11398256e-01
9.42942798e-01 -5.29436469e-01 -5.87353885e-01 1.67716950e-01
6.74038231e-01 -1.51391372e-01 3.73963892e-01 -7.54534304e-01
1.44955993e-01 -4.11389545e-02 3.04089896e-02 7.46085122e-03
7.84152225e-02 -8.57440650e-01 1.08240396e-01 5.03495932e-01
-5.75593233e-01 4.22997713e-01 -4.55721557e-01 9.22349155e-01
-5.04260287e-02 1.30350649e-01 7.45316207e-01 -2.15990618e-01
-3.57415020e-01 6.31881654e-01 1.69140607e-01 5.46151400e-01
1.14477515e+00 -2.82512128e-01 -6.45438254e-01 -7.82250345e-01
-3.64872962e-01 3.05190772e-01 8.30338478e-01 2.59460241e-01
2.70168036e-01 -1.20893002e+00 -6.80367470e-01 -1.44961134e-01
3.35740894e-01 -3.76216739e-01 -1.07649505e-01 1.00080538e+00
-3.22968721e-01 9.74748358e-02 -3.23004812e-01 -5.93083918e-01
-1.11829042e+00 6.66965663e-01 2.54475564e-01 -2.92126179e-01
-8.42658460e-01 7.30827808e-01 2.67959177e-01 -5.22024155e-01
2.65905894e-02 1.84888214e-01 -1.18554570e-02 1.31249011e-01
-1.96753412e-01 4.45968837e-01 -8.99969861e-02 -4.23154652e-01
-4.30242956e-01 1.76746458e-01 -1.05954722e-01 5.01333952e-01
1.48981643e+00 -8.60559568e-02 -1.71477124e-01 5.28395772e-01
1.27077508e+00 1.12164840e-01 -8.87567282e-01 -3.86807561e-01
1.49775743e-01 -5.74124157e-01 -2.48224631e-01 -4.94151324e-01
-1.14049864e+00 7.67817795e-01 2.34177727e-02 9.00582850e-01
7.27494240e-01 3.03389490e-01 4.80503649e-01 4.34805334e-01
2.64584124e-01 -9.32802737e-01 1.82054549e-01 3.69528383e-01
7.74087965e-01 -1.35606194e+00 2.82356977e-01 -5.83230615e-01
-4.41544771e-01 8.08574319e-01 5.26547134e-01 -5.27366221e-01
9.12047923e-01 8.83562490e-02 -2.51443624e-01 -6.52462304e-01
-8.22531521e-01 -2.59854585e-01 4.73315977e-02 6.13140404e-01
5.04535615e-01 3.94910574e-01 -4.13338751e-01 2.16530380e-03
-3.88658166e-01 -5.54960728e-01 6.54779613e-01 5.14818370e-01
2.71139015e-02 -9.48816895e-01 2.71667808e-01 9.06637132e-01
-1.50741875e-01 -2.67270356e-01 -1.13256800e+00 1.10939324e+00
-5.51452219e-01 8.14327896e-01 2.61279680e-02 -5.56863904e-01
-3.88290919e-02 -2.04499319e-01 2.91456968e-01 -5.39347351e-01
-5.04267573e-01 -5.51786959e-01 3.20035696e-01 -5.35329878e-01
-3.61885667e-01 -4.11833018e-01 -7.54133761e-01 -9.48049784e-01
-4.01209414e-01 2.59656459e-01 7.93573409e-02 4.71512854e-01
4.18392450e-01 1.27715796e-01 6.85598850e-01 -5.22013783e-01
-2.89143652e-01 -9.30020213e-01 -8.46446633e-01 9.46081102e-01
2.62939576e-02 -7.49989986e-01 -7.12963164e-01 -6.14440560e-01] | [7.009494304656982, 6.1558613777160645] |
a8be7f4b-c51e-494f-b172-b997939708ac | semiparametric-language-models-are-scalable | 2303.01421 | null | https://arxiv.org/abs/2303.01421v1 | https://arxiv.org/pdf/2303.01421v1.pdf | Semiparametric Language Models Are Scalable Continual Learners | Semiparametric language models (LMs) have shown promise in continuously learning from new text data by combining a parameterized neural LM with a growable non-parametric memory for memorizing new content. However, conventional semiparametric LMs will finally become prohibitive for computing and storing if they are applied to continual learning over streaming data, because the non-parametric memory grows linearly with the amount of data they learn from over time. To address the issue of scalability, we present a simple and intuitive approach called Selective Memorization (SeMem), which only memorizes difficult samples that the model is likely to struggle with. We demonstrate that SeMem improves the scalability of semiparametric LMs for continual learning over streaming data in two ways: (1) data-wise scalability: as the model becomes stronger through continual learning, it will encounter fewer difficult cases that need to be memorized, causing the growth of the non-parametric memory to slow down over time rather than growing at a linear rate with the size of training data; (2) model-wise scalability: SeMem allows a larger model to memorize fewer samples than its smaller counterpart because it is rarer for a larger model to encounter incomprehensible cases, resulting in a non-parametric memory that does not scale linearly with model size. We conduct extensive experiments in language modeling and downstream tasks to test SeMem's results, showing SeMem enables a semiparametric LM to be a scalable continual learner with little forgetting. | ['Houfeng Wang', 'Furu Wei', 'Si-Qing Chen', 'Tao Ge', 'Guangyue Peng'] | 2023-03-02 | null | null | null | null | ['memorization'] | ['natural-language-processing'] | [-6.06511980e-02 1.28618870e-02 -1.35692507e-01 -2.92178154e-01
-9.63236034e-01 -6.82344019e-01 2.46234775e-01 4.02203441e-01
-7.54246593e-01 8.64651680e-01 -7.53307119e-02 -5.92567027e-01
-7.31346011e-02 -8.34954739e-01 -1.01247847e+00 -3.95554185e-01
-3.95984441e-01 7.28298903e-01 3.84983689e-01 1.46495551e-01
-1.59577325e-01 2.76050925e-01 -1.52923656e+00 2.36148387e-01
9.57489908e-01 6.92830145e-01 4.25493985e-01 7.09920466e-01
-3.04224670e-01 6.52749538e-01 -5.44750988e-01 2.32311040e-02
1.05137117e-01 -2.22124234e-01 -6.63790524e-01 -4.46141474e-02
5.64219594e-01 -1.74944058e-01 -1.55218244e-01 5.92725396e-01
3.13922703e-01 4.63326752e-01 6.56540394e-01 -9.21549916e-01
-6.01041377e-01 9.69500184e-01 -6.05623186e-01 5.14550626e-01
-5.18038645e-02 1.13087341e-01 6.24815226e-01 -1.05138791e+00
2.29408875e-01 1.35829806e+00 9.99580503e-01 4.94882435e-01
-1.39569950e+00 -9.04486001e-01 5.37951708e-01 -4.43328798e-01
-1.15508282e+00 -5.34024596e-01 3.93792719e-01 -2.05624208e-01
9.91409957e-01 2.58171380e-01 8.11486483e-01 7.75106251e-01
1.62125990e-01 8.94792736e-01 1.06147552e+00 -3.04793656e-01
4.83200938e-01 3.98805559e-01 2.30421692e-01 7.51861453e-01
4.68495846e-01 -1.51104316e-01 -8.92945766e-01 -3.39897335e-01
7.86260188e-01 -5.60064986e-02 7.26436228e-02 -2.48917639e-01
-6.85011327e-01 8.14261615e-01 -9.57118422e-02 1.30631208e-01
-8.04728344e-02 8.85028020e-02 4.45317000e-01 7.17814565e-01
7.92642772e-01 5.97518146e-01 -7.22994864e-01 -5.04008949e-01
-1.47257781e+00 1.16875559e-01 9.31930304e-01 7.57771492e-01
6.60952985e-01 3.30754250e-01 1.97672710e-01 9.86115634e-01
1.59630720e-02 6.55761838e-01 1.14669490e+00 -6.05947256e-01
4.98937130e-01 5.23926616e-01 -1.39421418e-01 -4.64514703e-01
-4.88215208e-01 -4.98988032e-01 -7.33854175e-01 -1.85958058e-01
4.66978461e-01 -4.08996791e-01 -8.38241100e-01 2.04759765e+00
2.38493178e-02 7.68889338e-02 -5.31544909e-02 -1.71399098e-02
4.27890420e-01 9.41553831e-01 3.95187169e-01 -6.86825871e-01
8.03466678e-01 -7.73310006e-01 -3.05219829e-01 -5.66405594e-01
1.05682254e+00 -3.55707288e-01 1.78567111e+00 4.46936369e-01
-1.54892266e+00 -4.96306360e-01 -9.77039456e-01 6.08020369e-03
-1.93698466e-01 -3.52375031e-01 7.33839214e-01 5.83090663e-01
-1.25141907e+00 3.77864718e-01 -1.09798753e+00 -6.88067079e-02
3.11232746e-01 4.83009517e-01 -3.24452966e-02 3.14896330e-02
-1.12411690e+00 6.36161923e-01 5.71800470e-01 -2.92799056e-01
-7.85562158e-01 -1.15049112e+00 -8.18451464e-01 3.78779829e-01
2.11359397e-01 -5.26205897e-01 1.17105675e+00 -1.18538034e+00
-1.50377572e+00 3.68674964e-01 -4.69978362e-01 -6.21787906e-01
6.05357826e-01 -1.90661162e-01 -2.32467800e-01 -2.67472684e-01
-2.47027010e-01 5.72816968e-01 1.00188816e+00 -1.04526997e+00
-4.46082890e-01 -2.43252099e-01 -2.66860515e-01 3.46931547e-01
-1.03873014e+00 -4.20555711e-01 -2.71613985e-01 -7.15898216e-01
7.58689940e-02 -8.62793863e-01 3.31620835e-02 -1.24511339e-01
1.52163520e-01 -3.50433260e-01 9.19127703e-01 -4.32841122e-01
1.61312807e+00 -2.18500495e+00 -1.41703382e-01 9.81137529e-02
1.60744637e-01 4.20771986e-01 -4.87482071e-01 2.39273787e-01
1.33486509e-01 1.87999502e-01 -3.15829873e-01 -7.24563360e-01
-3.47851157e-01 1.93369985e-01 -5.69124162e-01 3.40685174e-02
1.09240264e-02 9.57233429e-01 -8.60423207e-01 -3.52030277e-01
-3.95233661e-01 1.71679065e-01 -6.01011038e-01 -1.18110171e-02
-3.35526139e-01 -9.26870704e-02 -5.62701263e-02 3.66820335e-01
6.55242085e-01 -4.93715525e-01 1.19977765e-01 5.14513791e-01
-3.60709243e-02 1.79568931e-01 -1.25963140e+00 1.24609494e+00
-6.84459984e-01 4.35207397e-01 -8.28431174e-02 -9.29294765e-01
7.64363647e-01 4.89717461e-02 1.55389696e-01 -5.03703356e-01
-4.67193097e-01 1.58503786e-01 -3.44105482e-01 -1.53202087e-01
6.43696845e-01 -5.25784731e-01 -7.85202235e-02 9.05772150e-01
1.33962572e-01 -5.45996465e-02 3.65230024e-01 2.83071369e-01
9.60010350e-01 -3.14653218e-01 -3.63326119e-03 -3.26940536e-01
3.03203374e-01 -1.45479083e-01 4.66276646e-01 1.11580002e+00
9.73238498e-02 -1.26035154e-01 2.57745951e-01 -4.23617274e-01
-1.02950060e+00 -1.42904484e+00 -8.64796713e-02 1.63367832e+00
-3.85061979e-01 -3.49649429e-01 -5.71547210e-01 -5.26701272e-01
2.71381855e-01 6.83940947e-01 -3.64600956e-01 -4.60040331e-01
-5.20753860e-01 -8.68180692e-01 5.64534783e-01 8.36546957e-01
4.26129907e-01 -9.26102340e-01 -3.46849978e-01 2.92027235e-01
2.08384812e-01 -5.27615011e-01 -7.47080922e-01 3.97725582e-01
-1.54172480e+00 -4.90245849e-01 -5.85295677e-01 -8.05152416e-01
8.91089499e-01 4.24718350e-01 1.01918650e+00 -2.40363013e-02
-1.09912887e-01 5.29191673e-01 1.29786089e-01 -5.60865760e-01
-5.52809715e-01 3.74327868e-01 4.92692739e-01 -4.08227801e-01
1.17664948e-01 -5.71511924e-01 -3.01721781e-01 2.42262017e-02
-1.16126835e+00 -3.66011523e-02 6.12918377e-01 9.63504970e-01
6.47102058e-01 3.29163760e-01 1.08105242e+00 -1.23819649e+00
9.74523425e-01 -7.35852182e-01 -6.15574360e-01 4.17644382e-01
-9.79390860e-01 3.25005412e-01 8.82252753e-01 -1.33907402e+00
-1.08574986e+00 -1.11021707e-02 3.35992306e-01 -2.64238954e-01
4.45905030e-01 9.93827879e-01 3.28634351e-01 2.41500698e-02
7.72142887e-01 7.35665858e-01 1.65097296e-01 -4.54973161e-01
3.57345015e-01 4.65977401e-01 2.37135917e-01 -8.75654042e-01
8.70804071e-01 1.91430211e-01 -1.66551813e-01 -9.08329785e-01
-9.78381932e-01 -2.22600847e-01 -4.59015518e-01 1.15677238e-01
1.96010813e-01 -1.14212549e+00 -4.19134587e-01 4.34181482e-01
-5.00046432e-01 -9.60225403e-01 -7.26811111e-01 3.99732560e-01
-4.47932929e-01 3.07055622e-01 -8.55412126e-01 -9.00045455e-01
-4.23575938e-01 -6.33735299e-01 4.83992308e-01 5.58306351e-02
-4.19491053e-01 -1.44861519e+00 1.07030734e-01 5.05593680e-02
6.81482196e-01 -5.44676840e-01 1.30441928e+00 -7.42662311e-01
-1.92134336e-01 -3.41265559e-01 4.01882827e-02 4.89384890e-01
2.43681800e-02 -2.43383199e-01 -6.88584685e-01 -8.71084750e-01
1.32345542e-01 -8.10927451e-01 1.01893115e+00 1.82143614e-01
1.30939662e+00 -6.44777298e-01 -8.54022652e-02 4.82727468e-01
1.13346076e+00 6.82859495e-02 -1.65860038e-02 4.38638031e-02
3.41243774e-01 1.82150900e-01 3.08473617e-01 4.51654583e-01
3.47119480e-01 -1.94190517e-02 -4.47618723e-01 -1.04244396e-01
4.61153425e-02 -7.34432757e-01 7.52050042e-01 1.29410112e+00
3.44692379e-01 -2.60635167e-01 -1.01480067e+00 3.41712773e-01
-1.77570701e+00 -7.89601922e-01 4.03072298e-01 2.64167261e+00
1.46149576e+00 4.67540413e-01 2.49284565e-01 9.21921246e-03
1.34996757e-01 7.02648424e-03 -9.49422240e-01 -4.06492859e-01
-1.96886912e-01 2.58762062e-01 4.50790346e-01 6.32375956e-01
-7.86855042e-01 1.04101992e+00 7.13370752e+00 1.03918290e+00
-1.44501567e+00 2.59226322e-01 9.31342721e-01 -6.43244803e-01
-4.70167071e-01 -1.55090511e-01 -1.13971412e+00 4.93877709e-01
1.60882246e+00 -5.18484652e-01 4.53948468e-01 8.85755956e-01
1.87073141e-01 -4.08755869e-01 -1.21430850e+00 7.11250544e-01
5.47137298e-02 -1.28368390e+00 4.71961707e-01 -1.84668750e-01
8.61937642e-01 4.33610566e-02 4.97058272e-01 8.08726430e-01
4.84806418e-01 -1.07919300e+00 5.76668859e-01 6.95066392e-01
9.17219996e-01 -8.08593452e-01 1.76633209e-01 1.00086343e+00
-9.16102409e-01 -4.63006765e-01 -6.34124637e-01 -7.11511448e-02
-8.45084607e-04 5.85363030e-01 -8.41491699e-01 -2.84995139e-01
4.08091247e-01 1.04692660e-01 -8.60604405e-01 8.68505180e-01
2.24305838e-01 1.26938927e+00 -6.18350983e-01 5.78605644e-02
8.64458159e-02 1.82072818e-01 2.70832539e-01 1.26191914e+00
4.70028073e-01 -6.39443472e-02 2.89129466e-01 6.63410962e-01
-2.03431472e-01 3.35468054e-01 -4.48566049e-01 -2.63783544e-01
8.79565775e-01 7.34745622e-01 -5.87295592e-01 -6.49443746e-01
-3.23368698e-01 6.48647726e-01 7.47115374e-01 4.12825137e-01
-2.69418240e-01 7.43797123e-02 -2.86817588e-02 5.45564651e-01
-8.34641457e-02 -6.63411140e-01 -4.89241064e-01 -1.09356081e+00
-1.45222306e-01 -7.23076224e-01 6.21634364e-01 -5.27991712e-01
-1.24771130e+00 4.08984751e-01 6.29279912e-02 -7.63174295e-01
-2.83329189e-01 -1.50485456e-01 -4.95081156e-01 7.71619499e-01
-1.28888297e+00 -9.23416495e-01 1.46950036e-01 4.79760557e-01
8.54912996e-01 -1.39911577e-01 9.54334795e-01 -1.39465360e-02
-5.10005713e-01 1.18674815e+00 4.25533533e-01 -3.51020664e-01
8.70204151e-01 -1.03777337e+00 1.48501724e-01 6.10364795e-01
1.39020100e-01 9.53278422e-01 5.14199495e-01 -8.72691691e-01
-1.51091325e+00 -1.19684541e+00 8.76730204e-01 -5.99583626e-01
8.18496823e-01 -5.49555898e-01 -1.34067726e+00 8.61355245e-01
-6.55100882e-01 -6.09774515e-02 9.30140257e-01 4.35079306e-01
-6.33535504e-01 -2.70659536e-01 -1.00615215e+00 4.89123076e-01
7.61909127e-01 -6.30014062e-01 -3.89628738e-01 2.93689251e-01
8.65108252e-01 -1.73420370e-01 -7.57814944e-01 2.53996342e-01
4.16233033e-01 -3.61549288e-01 7.09752619e-01 -7.43146002e-01
8.17478374e-02 1.65469244e-01 2.41846725e-01 -1.14175546e+00
-2.65270740e-01 -7.18436956e-01 -6.21032298e-01 9.61029947e-01
7.79942930e-01 -8.25139403e-01 8.55339110e-01 8.57279420e-01
2.11220771e-01 -9.33080435e-01 -8.94055009e-01 -1.23404181e+00
6.74116671e-01 -5.19458771e-01 2.80892372e-01 8.97205055e-01
1.32109657e-01 3.19787532e-01 -2.27214903e-01 -1.36414155e-01
3.05664331e-01 -1.27664238e-01 4.61407512e-01 -1.30468285e+00
-4.95449215e-01 -2.54767418e-01 1.81308225e-01 -1.45852995e+00
1.15062751e-01 -8.94352436e-01 -6.40428159e-03 -9.14698601e-01
5.88857651e-01 -7.82487452e-01 -4.18187797e-01 6.55194104e-01
-3.66369635e-01 3.23129073e-02 4.98677082e-02 4.27318245e-01
-8.49801362e-01 3.55784327e-01 9.84004021e-01 3.65468934e-02
-8.48218143e-01 2.25168034e-01 -6.65434122e-01 6.90251172e-01
5.86700022e-01 -4.48019058e-01 -9.60021675e-01 -2.87629306e-01
5.29821992e-01 2.52734125e-01 -1.73675269e-01 -8.92653227e-01
5.10035813e-01 -4.86081420e-03 5.46201646e-01 -3.76933903e-01
2.58940011e-01 -3.00421327e-01 -8.66914168e-02 4.71856356e-01
-7.01739788e-01 3.33887070e-01 6.52216971e-01 7.76254833e-01
8.08730051e-02 -2.58380890e-01 7.50156164e-01 -1.67410403e-01
-1.85009539e-01 4.83009607e-01 -7.64796495e-01 3.43522638e-01
7.47873545e-01 -2.00983465e-01 4.86019440e-02 -5.59428513e-01
-9.31747675e-01 3.23329270e-01 4.01029974e-01 2.73161262e-01
6.39377832e-01 -8.98854136e-01 -3.58578086e-01 5.17807961e-01
-2.39694864e-01 2.53848433e-01 3.60997468e-01 6.47855341e-01
3.68687548e-02 3.22791785e-01 2.48137131e-01 -3.49743783e-01
-1.17593384e+00 6.47307873e-01 1.14653021e-01 -5.32238066e-01
-4.16691005e-01 1.21358502e+00 2.56475598e-01 -3.45614582e-01
3.95412832e-01 -4.14314538e-01 2.78820932e-01 2.32736140e-01
6.99474573e-01 3.98502201e-01 -1.29652217e-01 6.24867901e-02
2.32198074e-01 1.57728642e-01 -5.48887193e-01 -1.54759213e-01
1.43810725e+00 -9.39538851e-02 -1.26052007e-01 1.09623885e+00
8.27238500e-01 2.77588125e-02 -1.61846590e+00 -6.51342452e-01
1.35091111e-01 1.52802598e-02 1.48518663e-03 -7.79848516e-01
-5.24403453e-01 7.38731027e-01 4.93931592e-01 1.70013100e-01
9.02297616e-01 -1.38351381e-01 8.76302779e-01 6.58333898e-01
3.09638232e-01 -1.29168379e+00 3.74923795e-01 8.16405892e-01
6.49740279e-01 -1.07160366e+00 -1.01015531e-02 3.02114747e-02
-4.80547696e-01 1.23470020e+00 5.61735570e-01 2.22276002e-01
9.88340855e-01 6.54628575e-01 -2.48217523e-01 1.78104907e-01
-1.12152696e+00 5.95212400e-01 2.70486802e-01 2.19015419e-01
3.72031152e-01 4.01690118e-02 -6.25022948e-02 9.05790210e-01
-4.17437762e-01 7.85506591e-02 3.84564906e-01 8.55957389e-01
-8.68104994e-01 -9.41621423e-01 1.34301167e-02 8.62751722e-01
-5.74496724e-02 -3.23306829e-01 -2.76184268e-02 5.69481254e-01
-2.54996151e-01 6.27188623e-01 3.36325109e-01 6.73213601e-02
-1.28825614e-02 5.96260428e-01 3.76318008e-01 -1.01262093e+00
-1.72676265e-01 -8.58262554e-02 -2.27091551e-01 -2.16633067e-01
2.39390910e-01 -6.97032869e-01 -1.43629742e+00 -3.81054699e-01
-2.61710167e-01 2.07489401e-01 4.34554845e-01 9.19170141e-01
2.99924403e-01 1.56689271e-01 4.43003923e-01 -4.76352662e-01
-1.24046278e+00 -9.68633413e-01 -5.22602320e-01 2.39507519e-02
3.19212317e-01 -2.55088419e-01 -5.53781629e-01 7.35756941e-03] | [9.73644733428955, 3.5431783199310303] |
7db11e03-0901-42b2-90a2-80738a942fe6 | panovpr-towards-unified-perspective-to | 2303.14095 | null | https://arxiv.org/abs/2303.14095v1 | https://arxiv.org/pdf/2303.14095v1.pdf | PanoVPR: Towards Unified Perspective-to-Equirectangular Visual Place Recognition via Sliding Windows across the Panoramic View | Visual place recognition has received increasing attention in recent years as a key technology in autonomous driving and robotics. The current mainstream approaches use either the perspective view retrieval perspective view (P2P) paradigm or the equirectangular image retrieval equirectangular image (E2E) paradigm. However, a natural and practical idea is that users only have consumer-grade pinhole cameras to obtain query perspective images and retrieve them in panoramic database images from map providers. To this end, we propose PanoVPR, a sliding-window-based perspective-to-equirectangular (P2E) visual place recognition framework, which eliminates feature truncation caused by hard cropping by sliding windows over the whole equirectangular image and computing and comparing feature descriptors between windows. In addition, this unified framework allows for directly transferring the network structure used in perspective-to-perspective (P2P) methods without modification. To facilitate training and evaluation, we derive the pitts250k-P2E dataset from the pitts250k and achieve promising results, and we also establish a P2E dataset in a real-world scenario by a mobile robot platform, which we refer to YQ360. Code and datasets will be made available at https://github.com/zafirshi/PanoVPR. | ['Kaiwei Wang', 'Yining Lin', 'Zhe Yin', 'Kailun Yang', 'Hao Shi', 'Ze Shi'] | 2023-03-24 | null | null | null | null | ['visual-place-recognition'] | ['computer-vision'] | [-2.46618986e-01 -5.64774930e-01 -3.00025135e-01 -3.50119323e-01
-4.53280360e-01 -7.16970682e-01 7.00934470e-01 -4.33811694e-01
-6.50565803e-01 2.43113399e-01 -3.75647038e-01 -2.07343698e-01
-2.59385377e-01 -9.00714874e-01 -8.50752175e-01 -6.29526258e-01
2.57208049e-01 6.91100955e-02 6.03722394e-01 -3.13565105e-01
4.76759762e-01 7.25810230e-01 -1.93504262e+00 -1.44677991e-02
5.79575300e-01 1.06946850e+00 6.40079319e-01 3.24532926e-01
3.08111399e-01 2.47146249e-01 -1.85762689e-01 -1.74387142e-01
6.74092233e-01 2.29727402e-01 -2.59956926e-01 -2.69006550e-01
5.06442547e-01 -5.29084563e-01 -6.99639678e-01 1.11595142e+00
4.75305617e-01 1.36941254e-01 3.40226978e-01 -1.80393398e+00
-6.36861563e-01 -3.24725777e-01 -7.70171404e-01 5.71841821e-02
4.35783714e-01 2.01636329e-02 6.46258712e-01 -1.42327106e+00
7.63542354e-01 1.03145158e+00 3.78737807e-01 1.55662403e-01
-4.33087826e-01 -9.60478306e-01 -1.23872347e-01 8.58878314e-01
-1.70288372e+00 -5.08396685e-01 7.34347939e-01 -2.54476875e-01
7.76813328e-01 2.13808447e-01 7.07762897e-01 7.17319429e-01
4.08712447e-01 9.69996452e-01 1.07241344e+00 -1.02568559e-01
-9.79139879e-02 1.34643689e-01 -1.12874452e-02 5.75933158e-01
1.74073964e-01 4.86790270e-01 -8.04091215e-01 8.42913017e-02
8.09400082e-01 4.48023766e-01 -5.63315868e-01 -9.05201137e-01
-1.31294656e+00 5.37848055e-01 7.58313119e-01 -2.06746712e-01
-1.95380226e-01 8.07330087e-02 2.20720917e-01 2.67081946e-01
-1.20606758e-01 8.60326290e-02 -1.46095291e-01 -1.37894988e-01
-6.04977667e-01 3.62167895e-01 2.55527556e-01 1.53646517e+00
1.01841784e+00 -1.94523200e-01 4.00831193e-01 8.42172980e-01
5.76399863e-01 9.59274352e-01 4.95469630e-01 -1.04021454e+00
7.16667891e-01 4.83021677e-01 3.41693223e-01 -1.21076107e+00
-2.08022475e-01 1.23978578e-01 -5.31553507e-01 4.69704270e-01
1.44637257e-01 2.25235432e-01 -7.77644575e-01 1.07162905e+00
2.40716830e-01 -2.26946950e-01 2.54172862e-01 1.14216423e+00
8.75331819e-01 8.58755589e-01 -4.90213722e-01 3.68055701e-01
1.50733364e+00 -1.31180310e+00 -4.95858490e-01 -5.50102055e-01
3.25912416e-01 -7.52926469e-01 9.94007707e-01 4.63671923e-01
-5.86684227e-01 -4.90840465e-01 -1.41655123e+00 -3.43652666e-01
-7.00288653e-01 5.47704518e-01 1.77981928e-01 2.79668093e-01
-1.01007187e+00 6.68172678e-03 -6.73694372e-01 -3.57771337e-01
1.53157845e-01 2.92711467e-01 -9.55545425e-01 -5.89121044e-01
-1.30437899e+00 9.59864974e-01 5.82342744e-01 4.21869248e-01
-6.60466850e-01 -3.43227267e-01 -1.14094913e+00 2.93975696e-02
4.81442302e-01 -1.08366504e-01 9.52996492e-01 -3.06282252e-01
-1.34626293e+00 7.62088537e-01 -3.67977619e-01 -2.10981086e-01
4.51769173e-01 -3.34055781e-01 -5.22339284e-01 3.73302430e-01
2.01963201e-01 9.23428595e-01 6.82051897e-01 -1.24373651e+00
-8.65077376e-01 -5.59393048e-01 9.90044512e-03 6.00647986e-01
-2.37306245e-02 -6.87527284e-02 -8.24879885e-01 -8.63896087e-02
6.37721062e-01 -1.06485713e+00 2.26999089e-01 3.70743066e-01
-1.23104639e-01 -2.51368344e-01 1.22349238e+00 -4.75732356e-01
6.07986391e-01 -2.35222888e+00 -3.32829595e-01 1.86253652e-01
6.10998943e-02 2.71952093e-01 -1.69199705e-02 5.62318563e-01
1.05619105e-02 -3.84262979e-01 4.12692465e-02 -1.08026154e-03
-1.20043144e-01 9.09646153e-02 -3.11661452e-01 5.46363652e-01
-1.49741635e-01 9.24002647e-01 -9.39408720e-01 -3.12438518e-01
5.39326310e-01 2.67304361e-01 -2.94297308e-01 -4.88416627e-02
5.54321349e-01 3.91278230e-02 -4.88990575e-01 8.18198800e-01
9.46624637e-01 1.42181307e-01 -3.03221315e-01 -8.49345699e-02
-5.57041526e-01 -4.85140532e-02 -9.10528958e-01 1.69771445e+00
-4.04335380e-01 9.95454967e-01 3.76217738e-02 -7.10733414e-01
1.25297534e+00 -6.56092316e-02 1.93054348e-01 -1.05230308e+00
6.22139079e-03 5.31550646e-01 -4.37593699e-01 -4.95957375e-01
1.07211065e+00 1.84404105e-01 -1.38047248e-01 -1.86902747e-01
-2.13626146e-01 -2.28503227e-01 1.12491466e-01 -7.13545233e-02
6.37981594e-01 2.02796042e-01 3.45805943e-01 -8.13233107e-02
6.53532088e-01 2.00861126e-01 5.60085535e-01 4.72907871e-01
-4.59325820e-01 9.82882738e-01 1.91035658e-01 -3.25424582e-01
-9.27915692e-01 -1.13059747e+00 -2.94822484e-01 5.46151578e-01
9.89770293e-01 -2.72464246e-01 -1.82678252e-01 -4.54657614e-01
-3.11472174e-02 4.58826661e-01 -2.65279531e-01 -2.01807722e-01
-4.73787218e-01 -1.14920631e-01 5.26602209e-01 5.03029108e-01
1.16550064e+00 -9.24363256e-01 -9.89946187e-01 -2.17154354e-01
-3.06468606e-01 -1.11239290e+00 -5.65917969e-01 -3.14870174e-03
-5.54354906e-01 -1.07090247e+00 -1.00816989e+00 -9.16295826e-01
4.79854286e-01 1.19287312e+00 5.70308864e-01 -3.96399111e-01
1.10891283e-01 3.58200550e-01 -3.76597106e-01 -4.06787425e-01
2.97737330e-01 -1.47775963e-01 1.14871427e-01 -1.30575955e-01
7.68945515e-01 -3.94629240e-01 -8.63983333e-01 7.54163086e-01
-7.24197209e-01 4.73281890e-02 6.85533702e-01 6.83538198e-01
6.55800760e-01 -1.59676015e-01 1.78464741e-01 -1.29944131e-01
1.97670177e-01 -3.88314635e-01 -1.15264225e+00 1.71436936e-01
-6.28740430e-01 -3.74700695e-01 4.38155711e-01 -7.04391226e-02
-6.71900272e-01 1.65507779e-01 -5.04941307e-02 -9.93673444e-01
-7.32140103e-03 4.71906513e-01 -3.39208871e-01 -2.15391457e-01
4.38214242e-01 7.16200948e-01 4.91068475e-02 -9.23743173e-02
3.16928536e-01 1.07542682e+00 7.58771837e-01 2.55019026e-04
8.36837530e-01 7.39213586e-01 -2.50593990e-01 -9.53177154e-01
-1.07239619e-01 -9.25069571e-01 -5.93441367e-01 -4.25314993e-01
7.43958414e-01 -1.16885352e+00 -8.11043322e-01 7.84039736e-01
-1.07629085e+00 8.38594045e-03 2.61668593e-01 7.63576984e-01
-5.45237005e-01 3.92381817e-01 -2.67556030e-02 -5.31565249e-01
-1.02851003e-01 -1.23904622e+00 1.13673103e+00 6.05016768e-01
4.16576743e-01 -3.91301185e-01 -2.29137495e-01 2.95028090e-01
2.14572221e-01 -2.17653230e-01 2.91766554e-01 -2.86903530e-01
-1.02810621e+00 -4.34241265e-01 -5.66967964e-01 1.41171128e-01
-1.94460750e-01 -3.06384653e-01 -9.37350631e-01 -2.90927023e-01
1.66420061e-02 -2.24817887e-01 7.46957898e-01 8.62140283e-02
8.02145243e-01 1.45499855e-01 -4.36837226e-01 7.36160100e-01
1.47786510e+00 6.58625901e-01 1.02723372e+00 8.56252015e-01
6.04752004e-01 5.33910930e-01 1.03072405e+00 1.06106125e-01
7.96028852e-01 9.15953100e-01 5.41491330e-01 2.28026062e-01
2.11086929e-01 -6.12940311e-01 4.54125583e-01 6.56802118e-01
1.86058536e-01 -2.14505896e-01 -9.96586919e-01 8.53802979e-01
-1.68654191e+00 -8.21183622e-01 -5.52531779e-02 2.24788523e+00
8.98102224e-02 -1.77570488e-02 -4.36087430e-01 6.03088364e-03
7.65549123e-01 3.20142120e-01 -5.93503177e-01 -5.09290807e-02
-7.28420960e-03 -4.32135969e-01 8.94785285e-01 3.21766287e-01
-9.40559328e-01 9.60830748e-01 4.87922239e+00 1.03002703e+00
-1.52967751e+00 6.60494249e-03 7.40113333e-02 1.07552074e-01
2.70636957e-02 3.42187583e-02 -9.41418350e-01 2.95458406e-01
4.11911011e-01 -1.48081332e-01 3.85467291e-01 1.19847655e+00
6.04285486e-02 -4.42620784e-01 -9.36090410e-01 1.58748472e+00
2.56080419e-01 -1.05806351e+00 -1.21196538e-01 9.98672768e-02
4.07796830e-01 5.86507440e-01 1.96825758e-01 2.90996432e-01
-2.19767511e-01 -6.71453178e-01 8.00065994e-01 3.12617153e-01
9.70092356e-01 -5.65558374e-01 5.76301694e-01 5.16956985e-01
-1.44615078e+00 -2.04411820e-01 -6.82879508e-01 1.56675890e-01
1.05451994e-01 9.39456150e-02 -5.72127402e-01 9.37198162e-01
1.02436185e+00 1.04534197e+00 -5.63618302e-01 1.27136958e+00
-1.60047159e-01 -2.06009656e-01 -5.18395245e-01 1.32818408e-02
2.46753544e-01 -4.51335520e-01 5.57635128e-01 4.26605910e-01
7.77976871e-01 -8.10520798e-02 -2.31511012e-01 7.04086721e-01
2.16110855e-01 -1.24831542e-01 -1.12186074e+00 3.79020870e-01
8.68649542e-01 1.30632496e+00 -7.04444766e-01 -3.07896942e-01
-5.11389852e-01 8.65442693e-01 -6.26165643e-02 4.60719556e-01
-7.54622996e-01 -9.82679009e-01 4.10766989e-01 -3.49229644e-03
4.39421445e-01 -4.13540035e-01 3.18070762e-02 -1.24014640e+00
4.46463674e-01 -6.41231120e-01 -1.42894998e-01 -1.39540410e+00
-7.89231300e-01 6.70146585e-01 3.52813333e-01 -1.85640097e+00
9.32440162e-02 -8.66222680e-01 -3.73369873e-01 8.29562843e-01
-2.00929689e+00 -9.01788235e-01 -7.99592853e-01 7.08150387e-01
5.35198569e-01 -1.85263887e-01 2.89169520e-01 4.71395582e-01
-2.60747135e-01 3.99275690e-01 4.33535784e-01 1.02128098e-02
7.95240581e-01 -7.58777440e-01 4.59451914e-01 7.08337247e-01
-4.90043610e-02 6.68118477e-01 3.51123333e-01 -2.25645050e-01
-1.52064610e+00 -1.04144502e+00 7.62792289e-01 -3.64853770e-01
2.66193211e-01 -3.43166053e-01 -6.27442241e-01 5.96344233e-01
2.05327511e-01 2.32449710e-01 -7.78158009e-02 -7.03368008e-01
-2.32838243e-01 -3.18546921e-01 -8.88598979e-01 7.88594246e-01
9.95276988e-01 -6.17464483e-01 -4.43161845e-01 3.65103036e-02
3.58358622e-01 -7.66316891e-01 -5.01377940e-01 5.24521708e-01
8.02715480e-01 -1.15738094e+00 1.00707448e+00 4.62995768e-01
3.15051526e-01 -7.69069493e-01 -2.93107241e-01 -1.05843520e+00
1.23059474e-01 -2.47944504e-01 3.99732113e-01 6.90448761e-01
1.46966770e-01 -1.29429507e+00 8.06740046e-01 2.39279315e-01
-3.17337960e-01 -7.87570596e-01 -1.10917103e+00 -8.47758710e-01
-1.97200581e-01 -4.02483135e-01 5.18020213e-01 7.08167195e-01
-5.95126534e-03 1.46372886e-02 -2.92697519e-01 5.13493180e-01
2.65768051e-01 2.42883742e-01 9.78498399e-01 -9.08035159e-01
1.79197714e-01 -9.34374332e-02 -7.19463825e-01 -1.83016729e+00
-1.25575036e-01 -8.32538486e-01 3.55970293e-01 -1.40172672e+00
-1.24801636e-01 -4.80888158e-01 -1.77302465e-01 2.09053367e-01
3.05916339e-01 4.76858020e-01 3.14335495e-01 6.71199977e-01
-4.29806530e-01 8.44008923e-01 1.29774928e+00 -6.81440905e-02
-3.71275693e-02 -8.22501406e-02 -2.31065705e-01 5.72473049e-01
8.16553593e-01 -3.32476735e-01 -7.19719708e-01 -4.78118628e-01
1.98633730e-01 2.14460284e-01 6.23258710e-01 -1.16909623e+00
6.45428658e-01 -7.12628216e-02 2.59044647e-01 -1.15080547e+00
7.48527706e-01 -9.80718911e-01 7.09139463e-03 3.25889677e-01
2.04370603e-01 6.18836641e-01 1.12799041e-01 6.74061716e-01
-6.07915580e-01 -2.63129711e-01 3.50467741e-01 1.65727045e-02
-1.35376775e+00 3.74892026e-01 -3.55140448e-01 -3.44977140e-01
1.19570220e+00 -6.33550227e-01 -7.25598514e-01 -4.30883735e-01
-7.11243376e-02 6.51019216e-01 7.46196091e-01 8.16897333e-01
1.14679956e+00 -1.55541265e+00 -2.06768736e-01 5.91198862e-01
6.25397146e-01 2.46446088e-01 3.46179128e-01 8.97429109e-01
-9.90380168e-01 8.38532388e-01 -4.91014004e-01 -8.28087628e-01
-1.29320824e+00 5.12230098e-01 3.11665386e-01 3.17590892e-01
-7.69977391e-01 6.72753334e-01 4.98549908e-01 -9.14539695e-01
-6.73787743e-02 -7.58860633e-02 -3.53485495e-01 -1.28461093e-01
3.67063433e-01 2.88958013e-01 3.44823413e-02 -9.30976987e-01
-4.54830319e-01 7.95375228e-01 -1.37598738e-01 -2.45814264e-01
9.46895957e-01 -3.52447063e-01 3.25005427e-02 3.27312231e-01
1.32566226e+00 -7.44387805e-02 -1.33382308e+00 -1.35241687e-01
-2.04264492e-01 -7.85492539e-01 -3.31944004e-02 -1.91340193e-01
-9.56593990e-01 9.83651459e-01 9.52949464e-01 -1.34339467e-01
9.57670689e-01 -2.44650990e-01 8.56348038e-01 8.38630915e-01
8.36774886e-01 -9.08809066e-01 -1.83961928e-01 5.96650481e-01
1.03520286e+00 -1.24156427e+00 -4.92441840e-02 -4.94504184e-01
-6.45169199e-01 1.18114138e+00 7.16287732e-01 -1.38855085e-01
6.63972080e-01 -2.90651202e-01 3.67200136e-01 -3.54867369e-01
-4.41364288e-01 1.05289057e-01 2.08166465e-02 6.62004173e-01
-2.60781050e-01 -1.12506241e-01 -5.43134063e-02 5.88599034e-02
-3.61036420e-01 4.44524027e-02 6.23170853e-01 1.20084894e+00
-3.50852281e-01 -8.69983733e-01 -5.15248239e-01 8.40721875e-02
2.03276008e-01 -3.04125734e-02 -1.07938075e-03 1.04458749e+00
5.16583845e-02 7.72088647e-01 2.13032756e-02 -5.76973379e-01
5.15912712e-01 -9.58405733e-02 1.73848793e-01 -2.21755043e-01
1.47717744e-01 -4.84504439e-02 -2.50168890e-01 -6.18915617e-01
-3.45926106e-01 -5.43992043e-01 -1.07997203e+00 -2.50676602e-01
-4.70516592e-01 8.73862505e-02 8.42815876e-01 6.86423361e-01
5.91476679e-01 6.82966188e-02 6.54554009e-01 -1.18809080e+00
-2.74680942e-01 -6.63402140e-01 -6.66043162e-01 -1.39257222e-01
3.25295895e-01 -9.76015985e-01 -9.90784168e-02 -2.39065170e-01] | [7.673891067504883, -2.04423189163208] |
e604345d-b012-4445-90ec-22e8d0ce9c3c | unicom-universal-and-compact-representation | 2304.05884 | null | https://arxiv.org/abs/2304.05884v1 | https://arxiv.org/pdf/2304.05884v1.pdf | Unicom: Universal and Compact Representation Learning for Image Retrieval | Modern image retrieval methods typically rely on fine-tuning pre-trained encoders to extract image-level descriptors. However, the most widely used models are pre-trained on ImageNet-1K with limited classes. The pre-trained feature representation is therefore not universal enough to generalize well to the diverse open-world classes. In this paper, we first cluster the large-scale LAION400M into one million pseudo classes based on the joint textual and visual features extracted by the CLIP model. Due to the confusion of label granularity, the automatically clustered dataset inevitably contains heavy inter-class conflict. To alleviate such conflict, we randomly select partial inter-class prototypes to construct the margin-based softmax loss. To further enhance the low-dimensional feature representation, we randomly select partial feature dimensions when calculating the similarities between embeddings and class-wise prototypes. The dual random partial selections are with respect to the class dimension and the feature dimension of the prototype matrix, making the classification conflict-robust and the feature embedding compact. Our method significantly outperforms state-of-the-art unsupervised and supervised image retrieval approaches on multiple benchmarks. The code and pre-trained models are released to facilitate future research https://github.com/deepglint/unicom. | ['Tongliang Liu', 'Jing Yang', 'Jia Guo', 'Ziyong Feng', 'Jaiwei Li', 'Kaicheng Yang', 'Jiankang Deng', 'Xiang An'] | 2023-04-12 | null | null | null | null | ['self-supervised-image-classification', 'metric-learning', 'metric-learning'] | ['computer-vision', 'computer-vision', 'methodology'] | [-1.39142096e-01 -5.58387518e-01 -6.18034065e-01 -6.26696765e-01
-9.08128858e-01 -5.53678513e-01 6.23560548e-01 6.95273876e-02
-6.54992640e-01 3.78400266e-01 1.51796937e-01 4.21308726e-02
-1.98894486e-01 -7.13577986e-01 -5.78278959e-01 -6.90671265e-01
-1.60812140e-01 1.65325642e-01 6.12610094e-02 2.02964604e-01
2.76478231e-01 2.96417743e-01 -1.88428378e+00 4.97469544e-01
6.88886464e-01 1.39956307e+00 2.86775231e-01 3.17420125e-01
-1.23558313e-01 3.15214694e-01 -4.17754740e-01 -1.88683659e-01
5.72327256e-01 -4.57434617e-02 -7.13222027e-01 2.28093609e-01
6.74316466e-01 -5.04257202e-01 -5.76685846e-01 1.18646169e+00
4.04219031e-01 1.26204506e-01 8.40804517e-01 -1.32491279e+00
-9.74291742e-01 2.63487250e-01 -5.51737785e-01 -4.01633605e-02
-3.57625261e-02 1.07686892e-01 1.35816026e+00 -1.14041519e+00
8.15219641e-01 1.02822351e+00 3.19956571e-01 2.14697152e-01
-1.02363586e+00 -8.79970133e-01 4.39972505e-02 5.14585137e-01
-1.85076761e+00 -2.38498479e-01 6.67594552e-01 -4.71577197e-01
8.87386739e-01 2.29446352e-01 3.52881253e-01 9.56791878e-01
7.23672211e-02 5.81663728e-01 8.69079053e-01 -2.59395987e-01
1.01675531e-02 3.11633229e-01 1.56654164e-01 6.44623995e-01
6.14472553e-02 -8.94601271e-03 -3.36765468e-01 -1.66857809e-01
5.54009020e-01 2.95783281e-01 -2.61137903e-01 -6.94010556e-01
-1.39388871e+00 1.09529030e+00 8.46206784e-01 3.35615635e-01
-1.54173300e-01 8.99645239e-02 4.93600041e-01 4.30766314e-01
2.99307138e-01 3.54318500e-01 -4.54341114e-01 2.04712152e-01
-1.06976819e+00 -3.19917947e-02 4.46775645e-01 1.06148386e+00
1.14941406e+00 -4.27091062e-01 -2.59157121e-01 1.11135793e+00
3.00902367e-01 3.79385740e-01 7.52227664e-01 -8.22802186e-01
2.59006590e-01 7.79576361e-01 -1.80679768e-01 -9.89153087e-01
-1.67712837e-01 -5.11845529e-01 -8.05329502e-01 -1.64041862e-01
1.10220537e-01 2.21084476e-01 -1.12166989e+00 1.51471472e+00
1.74204201e-01 -7.93857574e-02 -1.87391311e-01 1.14894283e+00
8.22405577e-01 7.96750903e-01 4.10711579e-02 1.82903513e-01
1.20785642e+00 -1.16882110e+00 -3.38043749e-01 -1.61211014e-01
7.32870638e-01 -9.15336013e-01 1.24297023e+00 7.45915920e-02
-4.63030815e-01 -6.08575106e-01 -1.19765079e+00 -1.95984006e-01
-7.41775811e-01 4.79538977e-01 5.49445510e-01 2.12327957e-01
-7.37128735e-01 3.76891345e-01 -4.50834394e-01 -2.05149621e-01
4.84385312e-01 2.72176623e-01 -6.21766806e-01 -3.21258605e-01
-1.32561278e+00 6.04211092e-01 3.75422299e-01 -1.28661573e-01
-6.58698618e-01 -4.83116806e-01 -7.97442079e-01 1.30306110e-01
2.35849157e-01 -3.50481629e-01 6.90890670e-01 -7.67675161e-01
-9.83662605e-01 1.09614015e+00 9.08180773e-02 -2.39418820e-01
2.01660722e-01 -1.51570495e-02 -2.79473186e-01 1.78809509e-01
3.08108121e-01 1.18729925e+00 8.84846210e-01 -1.05559623e+00
-4.39633191e-01 -1.36176467e-01 3.32945250e-02 1.46572068e-01
-8.87076020e-01 -1.25299424e-01 -8.09385359e-01 -6.11760736e-01
-5.92726171e-02 -1.10786343e+00 -1.02957651e-01 3.60250294e-01
-2.92277634e-01 -3.46103251e-01 5.25798082e-01 -1.01546794e-01
1.25826597e+00 -2.55244303e+00 -3.10069639e-02 2.80634969e-01
6.20911568e-02 2.15674877e-01 -6.00197613e-01 4.33173954e-01
-5.79638407e-04 -5.31988218e-04 2.83246767e-03 -2.71591365e-01
2.22799033e-01 1.55239880e-01 -4.03834820e-01 5.56171834e-01
1.33767501e-01 8.05007637e-01 -7.56636560e-01 -6.55906618e-01
3.66548955e-01 5.38026035e-01 -4.81155604e-01 9.46608335e-02
-3.74590382e-02 -1.67025864e-01 -4.63030308e-01 6.66419148e-01
7.91428447e-01 -5.40875435e-01 -9.47360620e-02 -4.39014047e-01
1.71539020e-02 2.10482448e-01 -1.11491275e+00 1.98757076e+00
-3.61616254e-01 7.47079492e-01 -3.50153863e-01 -9.58378077e-01
8.20344567e-01 -1.02510408e-01 5.69085598e-01 -7.68832266e-01
1.47245750e-01 3.41064721e-01 -1.50953084e-01 -3.71975422e-01
5.92734694e-01 3.16307962e-01 -1.18943520e-01 3.04909259e-01
2.15054214e-01 8.87631774e-02 3.08790147e-01 1.92046151e-01
8.00942540e-01 -1.55563325e-01 2.05794871e-02 -2.58727700e-01
4.29778308e-01 -9.49803218e-02 4.79686767e-01 6.27161384e-01
-2.33935103e-01 9.89518285e-01 3.90897095e-01 -5.01477897e-01
-9.21900392e-01 -9.58972573e-01 -6.78874254e-01 1.27828300e+00
3.27214628e-01 -6.90177679e-01 -6.03548825e-01 -7.90048420e-01
2.87331939e-01 2.46750310e-01 -7.37721205e-01 -3.70241463e-01
-4.10133339e-02 -6.50608599e-01 4.70286429e-01 2.79973984e-01
4.80776548e-01 -1.06993628e+00 -3.01312476e-01 -6.78646415e-02
-7.06220865e-02 -9.96994615e-01 -7.02057838e-01 2.85130858e-01
-3.77847612e-01 -1.06872940e+00 -8.60601425e-01 -1.01209557e+00
7.24817514e-01 4.85912323e-01 9.65289831e-01 2.72206664e-01
-6.13766730e-01 1.91527933e-01 -5.19608021e-01 1.88467819e-02
1.54270366e-01 4.01810914e-01 -6.34000823e-02 4.80037406e-02
6.86826527e-01 -2.79233515e-01 -9.12342846e-01 3.70299339e-01
-1.14393842e+00 -1.22911818e-01 6.36388540e-01 1.04009128e+00
7.79344380e-01 -9.09728482e-02 2.76508212e-01 -4.82224107e-01
4.57087308e-01 -5.89499712e-01 -5.09481490e-01 4.65965897e-01
-6.61111355e-01 1.13264464e-01 4.33956772e-01 -7.32947588e-01
-2.96877056e-01 5.84846102e-02 2.15862423e-01 -7.38291979e-01
2.76988819e-02 5.07400393e-01 -2.52441913e-02 -1.51423186e-01
4.86823261e-01 1.72150135e-01 -2.34406352e-01 -3.15753222e-01
4.68733370e-01 1.04030466e+00 1.34942278e-01 -4.50851023e-01
6.84105217e-01 1.96705908e-01 -3.97380054e-01 -7.56228685e-01
-7.93473184e-01 -6.81872487e-01 -4.60162461e-01 -2.46222131e-02
6.78145409e-01 -1.06305671e+00 -2.71380186e-01 4.63909149e-01
-9.04133976e-01 -2.73078412e-01 -1.34039223e-01 5.72932005e-01
-2.39607185e-01 3.42936814e-01 -6.85338676e-01 -1.86640754e-01
-3.90383035e-01 -1.43694377e+00 1.28192449e+00 2.15624750e-01
7.67301768e-02 -5.09329319e-01 4.24255431e-02 2.40296498e-01
5.66409469e-01 -2.13167772e-01 8.05440307e-01 -5.17348528e-01
-5.86410403e-01 -4.59617615e-01 -7.13461816e-01 5.59323967e-01
1.34559125e-01 1.13421947e-01 -8.44874501e-01 -4.53786910e-01
-5.14862001e-01 -7.13654160e-01 1.26617432e+00 1.30591810e-01
1.59114981e+00 -1.85491238e-02 -3.60900640e-01 7.32052505e-01
1.43288684e+00 -2.81751722e-01 5.43819368e-01 5.48361063e-01
5.67736626e-01 4.58287984e-01 6.48244977e-01 4.66722846e-01
5.03005147e-01 7.05852866e-01 3.03743780e-01 6.23361915e-02
-1.69026166e-01 -8.05739388e-02 1.25705510e-01 8.19526911e-01
4.66629058e-01 -2.14467093e-01 -8.11684728e-01 6.29675984e-01
-1.62075758e+00 -6.19450092e-01 3.54469717e-01 2.21215248e+00
1.01099515e+00 3.37830558e-02 -3.18039834e-01 -8.17600787e-02
6.75266623e-01 3.75040263e-01 -4.51156288e-01 -2.11679451e-02
-2.13581532e-01 8.51007998e-02 6.27098322e-01 3.34972441e-01
-1.34299111e+00 9.04446006e-01 5.13499594e+00 1.19134235e+00
-1.34062922e+00 -2.95101572e-02 5.49003720e-01 -3.58029962e-01
-1.90600470e-01 -4.32690009e-02 -8.12606215e-01 6.31235778e-01
6.21981621e-01 4.99400944e-02 3.67695451e-01 8.64080906e-01
-4.12168622e-01 6.91668540e-02 -1.04850352e+00 1.11945868e+00
-2.24841014e-02 -1.20745969e+00 2.31768593e-01 2.61055619e-01
8.67343187e-01 5.79446137e-01 3.13043356e-01 5.51303744e-01
-1.63324565e-01 -8.44210982e-01 5.61163783e-01 2.86780655e-01
1.08193851e+00 -5.73308825e-01 6.69831097e-01 6.11384585e-02
-1.12494886e+00 9.90296341e-03 -8.52984667e-01 2.63002247e-01
-2.94161677e-01 7.31913686e-01 -2.90633500e-01 1.78052917e-01
8.38525772e-01 9.23889518e-01 -8.35533500e-01 1.12088919e+00
6.54287189e-02 2.94792891e-01 -5.35055757e-01 1.11669667e-01
4.73265529e-01 -2.23136634e-01 1.64803833e-01 1.26678061e+00
2.17196286e-01 -3.10891539e-01 3.04690301e-01 5.40455461e-01
-4.09061044e-01 2.08629936e-01 -4.22738314e-01 -1.48555681e-01
5.50780118e-01 1.38189781e+00 -5.87039411e-01 -4.21681851e-01
-6.19477630e-01 9.74907935e-01 3.55575562e-01 4.75814462e-01
-8.61400604e-01 -6.94862962e-01 8.23834419e-01 -3.00080422e-02
4.30287421e-01 -7.21338242e-02 1.31442249e-02 -1.38467371e+00
2.11493909e-01 -7.81945050e-01 4.00768995e-01 -6.07801676e-01
-1.51080942e+00 8.30096781e-01 -1.33999377e-01 -1.52175260e+00
-8.45169798e-02 -5.98866522e-01 -1.70454189e-01 6.09389544e-01
-1.64169717e+00 -1.11726999e+00 -3.63199741e-01 4.76817548e-01
4.05261725e-01 -1.96066111e-01 9.95656252e-01 6.74310505e-01
-5.61454654e-01 9.44787443e-01 4.90623295e-01 3.29309106e-01
1.30003786e+00 -9.35624480e-01 -9.02488735e-03 4.16767061e-01
3.24384332e-01 7.14161694e-01 2.39198610e-01 -2.78646052e-01
-1.21261227e+00 -1.17812371e+00 8.52971673e-01 -1.05932020e-01
7.48321652e-01 -5.36552787e-01 -1.05940437e+00 3.89510870e-01
2.26064250e-01 6.50875688e-01 7.43339419e-01 -6.83927611e-02
-9.01021838e-01 -4.43266839e-01 -8.35163951e-01 4.97112393e-01
8.29796553e-01 -8.10858965e-01 -3.44533563e-01 4.43439990e-01
5.21644771e-01 -1.56496227e-01 -9.81307864e-01 4.83139396e-01
7.66496897e-01 -7.37778783e-01 9.79463220e-01 -4.27504897e-01
5.37891746e-01 -2.51871973e-01 -5.05132496e-01 -9.53703105e-01
-4.39286828e-01 2.92275734e-02 2.30329454e-01 1.14813972e+00
3.99561405e-01 -5.39636970e-01 5.93600988e-01 4.01656568e-01
1.84939384e-01 -1.09772134e+00 -8.20235729e-01 -6.68503463e-01
1.28664523e-01 -1.92708582e-01 5.45340061e-01 9.60122526e-01
-1.97328836e-01 1.37210906e-01 -2.60824442e-01 -7.53286481e-02
5.98483741e-01 5.18280864e-01 6.11015677e-01 -1.06895828e+00
-8.20233077e-02 -6.42395258e-01 -5.98835468e-01 -1.03703284e+00
2.47502491e-01 -1.08996868e+00 3.57952751e-02 -1.20335460e+00
4.94406283e-01 -8.09711695e-01 -8.83360863e-01 6.62781239e-01
-2.70305807e-03 6.35012865e-01 2.09060818e-01 4.84583199e-01
-9.71005440e-01 7.79181182e-01 1.13141942e+00 -4.06792402e-01
-1.76488981e-02 -4.39032018e-01 -4.14226770e-01 2.17788547e-01
8.82103384e-01 -6.57786489e-01 -3.88170242e-01 -6.06402040e-01
1.21795237e-01 -4.26222593e-01 3.83729130e-01 -8.34985018e-01
2.71152735e-01 -1.29870266e-01 4.93274927e-01 -5.24904966e-01
2.67508715e-01 -7.68900275e-01 -1.21071652e-01 1.76932231e-01
-6.85134411e-01 -3.58895659e-01 1.10324537e-02 3.98138583e-01
-5.45366049e-01 -2.48319775e-01 6.58543766e-01 8.45417082e-02
-7.35896528e-01 6.13263130e-01 -1.58003956e-01 -9.38106608e-03
8.26808035e-01 -4.49552238e-02 -3.44325781e-01 -2.13025108e-01
-2.33482763e-01 4.25959736e-01 7.31307924e-01 8.20721924e-01
5.06685197e-01 -1.53307915e+00 -4.80668217e-01 3.65306079e-01
7.18061864e-01 -1.02728069e-01 2.43536100e-01 5.47793210e-01
-3.76501143e-01 4.95498896e-01 -3.22757840e-01 -6.32856965e-01
-9.88102257e-01 6.37030065e-01 1.95212990e-01 -1.72959819e-01
-4.37371612e-01 7.54266202e-01 3.41716319e-01 -4.27657843e-01
4.37782109e-01 -5.39914444e-02 7.05706179e-02 1.90877303e-01
5.32300651e-01 -6.39349176e-03 5.68078272e-02 -6.44406557e-01
-4.78497148e-01 7.64709711e-01 -3.37006330e-01 1.65397212e-01
1.28687477e+00 -1.84786439e-01 -2.00195253e-01 2.29781389e-01
1.98897910e+00 -2.94321895e-01 -1.12606597e+00 -5.25073588e-01
-2.26485938e-01 -4.56609935e-01 2.49335155e-01 -5.36086023e-01
-1.32352400e+00 9.45185304e-01 7.87976980e-01 -1.38729542e-01
1.12519360e+00 8.16353187e-02 7.93926120e-01 5.61179101e-01
3.14579487e-01 -1.09462762e+00 6.49047196e-02 4.30343091e-01
8.87514591e-01 -1.46647906e+00 1.37498125e-01 -2.30270058e-01
-4.15071726e-01 9.72821951e-01 6.41155660e-01 -4.25871670e-01
8.65985870e-01 -1.10635564e-01 6.04832359e-02 -5.72848469e-02
-6.35591388e-01 -1.88191980e-01 7.07057655e-01 2.22906321e-01
3.19046795e-01 1.30421922e-01 -3.04190338e-01 5.74326754e-01
-9.62396935e-02 -2.86180973e-02 5.49798235e-02 7.52987683e-01
-3.08498174e-01 -1.14288592e+00 -2.57153027e-02 6.40553057e-01
-2.96875596e-01 -2.52432138e-01 -2.61375666e-01 5.62648833e-01
2.31014222e-01 5.57927608e-01 6.92018718e-02 -5.77065766e-01
1.21167414e-02 -1.00588560e-01 3.79131526e-01 -3.83748055e-01
-2.76626557e-01 6.74900599e-03 -3.46755832e-01 -6.16746724e-01
-1.17645480e-01 -1.92462921e-01 -1.01381600e+00 -1.18539482e-01
-4.60012674e-01 1.24663999e-02 6.77171350e-01 5.44904709e-01
5.70134044e-01 4.93416041e-02 7.55217969e-01 -8.76972675e-01
-6.91938579e-01 -1.04039073e+00 -4.42443311e-01 6.04300618e-01
2.68939495e-01 -8.04510534e-01 -5.38439512e-01 -1.37027249e-01] | [10.757006645202637, 0.9007059335708618] |
db5d7dbc-cde2-4418-9669-5322c8963544 | icicle-interpretable-class-incremental | 2303.07811 | null | https://arxiv.org/abs/2303.07811v1 | https://arxiv.org/pdf/2303.07811v1.pdf | ICICLE: Interpretable Class Incremental Continual Learning | Continual learning enables incremental learning of new tasks without forgetting those previously learned, resulting in positive knowledge transfer that can enhance performance on both new and old tasks. However, continual learning poses new challenges for interpretability, as the rationale behind model predictions may change over time, leading to interpretability concept drift. We address this problem by proposing Interpretable Class-InCremental LEarning (ICICLE), an exemplar-free approach that adopts a prototypical part-based approach. It consists of three crucial novelties: interpretability regularization that distills previously learned concepts while preserving user-friendly positive reasoning; proximity-based prototype initialization strategy dedicated to the fine-grained setting; and task-recency bias compensation devoted to prototypical parts. Our experimental results demonstrate that ICICLE reduces the interpretability concept drift and outperforms the existing exemplar-free methods of common class-incremental learning when applied to concept-based models. We make the code available. | ['Bartłomiej Twardowski', 'Bartosz Zieliński', 'Joost Van de Weijer', 'Dawid Rymarczyk'] | 2023-03-14 | null | null | null | null | ['class-incremental-learning'] | ['computer-vision'] | [ 5.10450721e-01 3.11889797e-01 -3.42184603e-01 -3.77461821e-01
-3.60122204e-01 -6.44855201e-01 7.39208996e-01 3.87969643e-01
-6.11028910e-01 9.42911863e-01 4.08184119e-02 -2.94471264e-01
-7.05345452e-01 -2.88751930e-01 -9.12323713e-01 -4.83241767e-01
-6.76753595e-02 1.03487015e+00 2.47030839e-01 -2.99780816e-01
3.22373807e-01 3.97372752e-01 -1.83132946e+00 6.16680682e-01
1.45078659e+00 9.20271993e-01 3.02638680e-01 3.36846620e-01
-1.67464465e-01 6.41151726e-01 -5.74060380e-01 -4.51818287e-01
2.40285739e-01 3.23348911e-03 -8.94690037e-01 -4.95917872e-02
7.47918844e-01 6.90439064e-03 3.25454026e-01 6.45560443e-01
1.53958440e-01 4.11157817e-01 8.87745857e-01 -1.50266397e+00
-1.27250636e+00 6.35237336e-01 -3.74195755e-01 4.31850106e-01
9.44157839e-02 2.26886019e-01 9.92915869e-01 -1.22505260e+00
5.24658263e-01 1.39173496e+00 1.05260742e+00 8.10563207e-01
-1.47515523e+00 -4.32107389e-01 1.02426302e+00 5.67559242e-01
-1.13088322e+00 -2.10354418e-01 8.20258975e-01 -5.29542327e-01
9.20974195e-01 2.85048306e-01 9.19057846e-01 1.34085321e+00
1.15684353e-01 9.29026842e-01 1.10964811e+00 -5.78503549e-01
6.03474617e-01 3.00769776e-01 6.38164639e-01 3.85128826e-01
6.21586800e-01 1.37519777e-01 -5.19223928e-01 -8.03669393e-02
4.49784219e-01 4.55423206e-01 -1.98736340e-01 -8.24862123e-01
-1.03431404e+00 4.05456036e-01 6.46754205e-01 1.68885827e-01
-3.86978507e-01 4.22069840e-02 3.84319097e-01 5.74723244e-01
4.39327955e-01 8.20412874e-01 -8.54796708e-01 8.70914981e-02
-5.23420036e-01 4.51268703e-01 4.10392672e-01 1.09870934e+00
8.46041381e-01 -2.37834416e-02 -6.53604090e-01 8.20357859e-01
-6.74628792e-03 4.82941538e-01 8.35208118e-01 -7.54663229e-01
3.88288617e-01 8.15340042e-01 5.91062941e-02 -3.96152139e-01
-3.94838691e-01 -6.71391428e-01 -6.62532866e-01 1.39110208e-01
2.19273493e-01 1.35718480e-01 -1.19122517e+00 1.68838644e+00
3.93397093e-01 1.21909961e-01 1.45615056e-01 5.31080365e-01
4.29897904e-01 1.22407623e-01 4.04247493e-01 -2.39385158e-01
1.25012052e+00 -1.21587932e+00 -6.84515059e-01 -5.39697945e-01
2.87442356e-01 -2.30211154e-01 1.75450945e+00 3.49827915e-01
-6.97486460e-01 -8.52554381e-01 -1.07449472e+00 -5.20380698e-02
-6.75180137e-01 -2.14526400e-01 8.24885190e-01 7.42867351e-01
-1.04189110e+00 4.25614506e-01 -5.87892294e-01 -1.30904362e-01
7.58981884e-01 3.62405390e-01 3.96267883e-02 -2.16098547e-01
-1.15618670e+00 9.59716976e-01 8.09442401e-01 1.29800942e-03
-8.73641014e-01 -1.24192870e+00 -6.38065517e-01 1.03174776e-01
6.87566459e-01 -1.04711354e+00 1.41572642e+00 -1.45309103e+00
-1.12451982e+00 4.69767451e-01 -1.97425663e-01 -6.40151203e-01
7.48324215e-01 -5.84889114e-01 -2.49305680e-01 -3.38335961e-01
7.58391321e-02 9.04802978e-01 1.30103445e+00 -1.55662858e+00
-6.81487501e-01 -3.06818008e-01 2.89318323e-01 5.03071606e-01
-6.85811341e-01 -9.81978297e-01 -9.39949006e-02 -7.97516942e-01
7.30533674e-02 -9.92039204e-01 6.77192211e-02 7.98848420e-02
-1.85700934e-02 -4.81112033e-01 8.59179080e-01 -2.40692019e-01
1.15439999e+00 -1.97203076e+00 1.41774118e-01 -1.51175648e-01
3.63016605e-01 3.50318402e-01 -3.25598061e-01 -9.32575911e-02
-3.89917612e-01 7.63247684e-02 -4.03944671e-01 -3.52340907e-01
-7.40507245e-02 3.17184478e-01 -3.89523715e-01 -3.12391698e-01
4.39626694e-01 1.13372552e+00 -1.38573325e+00 -3.51066403e-02
-3.97524126e-02 1.45505980e-01 -6.08767152e-01 4.75908034e-02
-5.27000725e-01 1.51248857e-01 -1.44241959e-01 6.52408719e-01
7.09165454e-01 -1.41846105e-01 1.60358250e-01 -2.53604539e-02
1.28652483e-01 -1.45277664e-01 -8.98485839e-01 1.68291605e+00
-6.18740201e-01 3.69910359e-01 -6.25854731e-01 -8.94224286e-01
6.80839837e-01 9.93692689e-03 -3.35151516e-02 -6.06297612e-01
-4.05281693e-01 1.27751529e-01 3.86850834e-02 -5.42008877e-01
6.38621032e-01 -3.30262393e-01 1.38941988e-01 2.90987521e-01
6.90838471e-02 -4.89327274e-02 1.18813977e-01 1.85067236e-01
9.41095412e-01 3.43834698e-01 5.01388013e-01 -3.25832158e-01
4.22951162e-01 -1.53984092e-02 6.41924798e-01 1.24007225e+00
-3.38832915e-01 5.28573871e-01 2.21079648e-01 -9.00458872e-01
-8.20957124e-01 -1.39097583e+00 -1.51502430e-01 1.41004133e+00
1.46246785e-02 -1.77064240e-01 -3.82912904e-01 -1.22918320e+00
4.72067744e-01 1.09890497e+00 -1.11317289e+00 -6.61260366e-01
-4.57285434e-01 -7.98257053e-01 -7.41292983e-02 8.30883920e-01
2.84924835e-01 -1.06969118e+00 -4.19663459e-01 9.22932476e-02
-1.34075716e-01 -5.55422962e-01 -3.96016479e-01 4.27129507e-01
-1.34030735e+00 -1.20185423e+00 -6.47026181e-01 -5.51727712e-01
1.06255341e+00 5.11370957e-01 1.38028598e+00 1.32962480e-01
-1.67479888e-01 8.29858422e-01 -3.63684118e-01 -7.45274723e-01
-2.29914188e-01 2.08992243e-01 3.06542099e-01 -1.54279852e-02
4.22301084e-01 -4.47485864e-01 -4.60475832e-01 -1.52442819e-02
-9.62465584e-01 1.93378944e-02 8.11490595e-01 1.22037685e+00
5.92575252e-01 -5.69295473e-02 1.04387987e+00 -1.63620698e+00
9.62011099e-01 -4.61635530e-01 -2.37393185e-01 7.98682034e-01
-1.43066788e+00 3.56562674e-01 6.96310163e-01 -8.10943902e-01
-1.59955025e+00 8.59991983e-02 4.11387682e-01 -4.32122320e-01
5.06840162e-02 3.29168707e-01 6.95966408e-02 1.20189227e-01
1.23921084e+00 1.89022705e-01 -1.65068254e-01 -3.43895435e-01
6.61947250e-01 3.06895226e-01 5.14452338e-01 -8.76648903e-01
8.65227163e-01 3.71490568e-01 -3.34857047e-01 -4.86516118e-01
-1.19235444e+00 -3.96458387e-01 -1.06888974e+00 -1.87871575e-01
2.33805299e-01 -1.03008318e+00 -4.52454478e-01 3.15126657e-01
-1.09883130e+00 -3.92628908e-01 -9.31225061e-01 1.49578407e-01
-4.80490029e-01 1.92429572e-01 -5.19997515e-02 -8.15809488e-01
-4.01683658e-01 -7.67570376e-01 9.33327317e-01 1.18707336e-01
-3.87518376e-01 -1.31456268e+00 1.84677709e-02 3.33985955e-01
4.75535572e-01 -1.95875652e-02 1.25296736e+00 -7.78731883e-01
-4.89481956e-01 -7.59120956e-02 -7.09983408e-02 4.76937085e-01
4.47641239e-02 -4.46240753e-01 -1.21201122e+00 -4.73299146e-01
5.84666617e-02 -4.14850175e-01 1.32165039e+00 2.16535479e-01
1.21059930e+00 -5.22223711e-01 -3.64600778e-01 3.13080966e-01
1.20565486e+00 1.36962533e-01 4.27563012e-01 5.43382823e-01
8.47728729e-01 3.79344225e-01 7.57765889e-01 3.38079631e-01
4.36981827e-01 3.56377929e-01 1.99843615e-01 1.44554943e-01
-2.99869448e-01 -1.85666814e-01 1.38460830e-01 3.12332869e-01
-3.51642936e-01 -2.37498973e-02 -8.82447779e-01 5.48178375e-01
-2.17491746e+00 -9.20201540e-01 1.39445677e-01 2.33440924e+00
1.00728452e+00 5.20632446e-01 -1.16511427e-01 5.27117662e-02
5.84395111e-01 -2.11134493e-01 -1.04106426e+00 -1.61377966e-01
-2.67511457e-02 2.94975843e-02 -2.47955080e-02 5.77224791e-01
-1.01280284e+00 8.46258640e-01 6.22565603e+00 6.37524903e-01
-7.18608677e-01 3.92344832e-01 6.60261393e-01 -1.25754640e-01
-5.98704338e-01 5.42318262e-03 -8.38933408e-01 2.32213333e-01
6.87188745e-01 -4.28083330e-01 3.39244038e-01 1.33007669e+00
-4.06286605e-02 -7.70567879e-02 -1.55747259e+00 8.20550799e-01
2.70461053e-01 -1.19758785e+00 4.62483287e-01 -5.15625298e-01
1.07606435e+00 -3.85809392e-01 3.36836815e-01 8.94847393e-01
9.70843285e-02 -5.29993773e-01 8.92263651e-01 7.90036082e-01
7.06188738e-01 -5.19301772e-01 5.05721569e-01 2.34767348e-01
-1.00064218e+00 -6.83486462e-01 -4.60279346e-01 -1.42942324e-01
-2.59254426e-01 7.34304726e-01 -1.22073734e+00 2.81618118e-01
6.42827868e-01 7.40518868e-01 -1.20803976e+00 1.10364580e+00
-4.47091132e-01 4.40861464e-01 1.40626892e-01 1.17276177e-01
-2.48317812e-02 3.27959985e-01 5.36649406e-01 9.38897371e-01
8.67761821e-02 -2.39957094e-01 7.17698112e-02 8.96021187e-01
-9.49450880e-02 -3.06579381e-01 -4.75086272e-01 4.60537136e-01
8.03448558e-01 1.05378234e+00 -6.14801586e-01 -6.26665890e-01
-1.90983471e-02 1.19467270e+00 7.70535171e-01 5.83009958e-01
-6.62692964e-01 2.55813766e-02 3.45052093e-01 3.41068983e-01
2.36177653e-01 1.38619766e-01 -4.37316924e-01 -1.14843166e+00
2.56778657e-01 -7.48522639e-01 6.91589832e-01 -6.85738921e-01
-1.65238285e+00 5.60929418e-01 3.22364479e-01 -1.14418006e+00
-1.67730495e-01 -6.27067447e-01 -5.13847470e-01 5.35785139e-01
-1.73406482e+00 -1.20579922e+00 -6.03125572e-01 3.29690278e-01
9.69844460e-01 -2.63298959e-01 7.79233754e-01 -7.34400675e-02
-3.78234774e-01 8.55809152e-01 4.18218493e-01 -6.19782567e-01
8.81147504e-01 -1.63338315e+00 4.71968740e-01 5.23631692e-01
-7.00063482e-02 9.76418495e-01 4.24237847e-01 -8.35955203e-01
-1.06151855e+00 -1.36989772e+00 7.55440712e-01 -9.72431779e-01
3.41136128e-01 -4.62144971e-01 -1.25079238e+00 8.10544670e-01
-2.06865892e-01 -1.01027720e-01 6.57202482e-01 7.67298400e-01
-6.21510804e-01 -5.99147022e-01 -1.03237355e+00 6.57031417e-01
1.40261340e+00 -3.03792179e-01 -1.08444083e+00 5.18027723e-01
9.27075982e-01 -2.06594512e-01 -4.55261707e-01 3.86343956e-01
5.57326734e-01 -8.30650926e-01 1.14420938e+00 -7.01405823e-01
3.13328905e-03 -3.08388025e-01 4.94699568e-01 -1.59345484e+00
-8.49591374e-01 -3.72089028e-01 -6.34047985e-01 9.22552466e-01
5.63730776e-01 -7.89704919e-01 5.96716762e-01 7.17779517e-01
-4.94850367e-01 -4.97797966e-01 -8.31101537e-01 -1.28909945e+00
5.31818680e-02 -3.11528146e-01 6.29918575e-01 1.00270700e+00
-4.20960635e-02 8.11382949e-01 -9.29088369e-02 -1.66391537e-01
6.38561070e-01 -5.79381697e-02 4.01971281e-01 -1.82806480e+00
-2.06797823e-01 -3.49662662e-01 -1.18349552e-01 -7.49318779e-01
2.26883754e-01 -9.33717728e-01 1.00861467e-01 -1.21982551e+00
4.06584829e-01 -7.72757053e-01 -6.36243403e-01 8.08340549e-01
-8.40184152e-01 -1.31516993e-01 7.46580139e-02 4.45952207e-01
-1.02732038e+00 6.37148321e-01 1.27799582e+00 -4.32761371e-01
-4.80771273e-01 1.41630366e-01 -9.60485697e-01 4.84873176e-01
5.48914969e-01 -5.01060009e-01 -1.15172505e+00 -4.56869543e-01
2.61593044e-01 -7.04134345e-01 4.26300287e-01 -1.34143305e+00
1.88775793e-01 -1.75859481e-01 7.15844810e-01 -3.26652616e-01
1.12031065e-01 -8.79637957e-01 1.31112352e-01 6.44114435e-01
-5.95276594e-01 1.20383397e-01 4.57179368e-01 1.24771905e+00
1.58112928e-01 -3.93437475e-01 6.45834327e-01 -2.62955278e-01
-1.08515167e+00 1.72751561e-01 -1.94171324e-01 2.62286454e-01
1.08125758e+00 -3.44992936e-01 -4.32872504e-01 1.48791686e-01
-8.30138922e-01 2.62901694e-01 2.59140551e-01 7.81203985e-01
6.43844962e-01 -1.39260602e+00 -5.65015376e-01 1.96181878e-01
6.40490234e-01 1.53292090e-01 4.02168900e-01 5.45303643e-01
-5.34783714e-02 5.17167151e-01 -3.08919460e-01 -6.19238377e-01
-8.88566315e-01 9.41000223e-01 2.28021383e-01 -3.26817453e-01
-4.53415334e-01 9.65657473e-01 4.79120195e-01 -6.53247058e-01
4.14387584e-01 -4.79407609e-01 -2.77775049e-01 1.77804112e-01
5.82684875e-01 4.54009056e-01 3.78533691e-01 1.06393054e-01
-1.87165603e-01 1.66814715e-01 -7.75774777e-01 1.77222908e-01
1.26626050e+00 -8.82811025e-02 9.41588283e-02 8.68286908e-01
5.86254597e-01 -7.77359724e-01 -1.59123147e+00 -3.90719444e-01
2.35212922e-01 -4.10797745e-01 -3.33044946e-01 -1.32860112e+00
-4.25592244e-01 7.56081522e-01 9.43282902e-01 -2.66835894e-02
1.11965001e+00 -3.50392818e-01 2.51097471e-01 9.80908811e-01
2.75421441e-01 -1.43633282e+00 6.19146287e-01 7.26814687e-01
1.19786394e+00 -1.33794534e+00 1.25711426e-01 -2.64350891e-01
-6.08877480e-01 1.05005538e+00 1.06412733e+00 3.42010111e-01
5.65232396e-01 -3.25179815e-01 -2.01992154e-01 2.93838307e-02
-9.91595447e-01 2.69197710e-02 6.04140401e-01 1.07533300e+00
2.99376279e-01 3.83954234e-02 -2.31197774e-01 7.94453919e-01
1.30548090e-01 -2.15978990e-03 3.11881602e-01 1.02362561e+00
-4.46997225e-01 -9.84763384e-01 -3.49740297e-01 6.29554927e-01
3.79939854e-01 -4.40296471e-01 -5.03681958e-01 7.01217413e-01
5.06270170e-01 3.59310567e-01 1.73997238e-01 -9.13832635e-02
5.77092469e-01 5.81337035e-01 5.63416362e-01 -9.10436690e-01
-5.73357165e-01 -5.17973185e-01 -1.49155736e-01 -7.28222057e-02
-1.40057996e-01 -7.63874769e-01 -1.19577944e+00 7.17296824e-02
-3.83525759e-01 7.11702034e-02 2.10828498e-01 9.95573521e-01
6.81950390e-01 8.08435440e-01 3.67782503e-01 -7.49127567e-01
-7.44406462e-01 -1.08326554e+00 -2.56320029e-01 7.09334254e-01
4.42639589e-01 -1.06768322e+00 -2.88958132e-01 3.04267168e-01] | [9.793540000915527, 3.4179952144622803] |
0b67809a-4b31-45c5-8e01-c2d495933a37 | trajectory-poisson-multi-bernoulli-mixture | 2306.16890 | null | https://arxiv.org/abs/2306.16890v1 | https://arxiv.org/pdf/2306.16890v1.pdf | Trajectory Poisson multi-Bernoulli mixture filter for traffic monitoring using a drone | This paper proposes a multi-object tracking (MOT) algorithm for traffic monitoring using a drone equipped with optical and thermal cameras. Object detections on the images are obtained using a neural network for each type of camera. The cameras are modelled as direction-of-arrival (DOA) sensors. Each DOA detection follows a von-Mises Fisher distribution, whose mean direction is obtain by projecting a vehicle position on the ground to the camera. We then use the trajectory Poisson multi-Bernoulli mixture filter (TPMBM), which is a Bayesian MOT algorithm, to optimally estimate the set of vehicle trajectories. We have also developed a parameter estimation algorithm for the measurement model. We have tested the accuracy of the resulting TPMBM filter in synthetic and experimental data sets. | ['Jimin Xiao', 'Ángel F. García-Fernández'] | 2023-06-29 | null | null | null | null | ['object-tracking', 'multi-object-tracking'] | ['computer-vision', 'computer-vision'] | [-6.51737228e-02 -5.04910827e-01 1.01135120e-01 -2.56633520e-01
-4.41704780e-01 -4.23616886e-01 6.88205183e-01 -4.44065750e-01
-7.65492618e-01 5.30519843e-01 -6.16937101e-01 -1.03369892e-01
-5.17469011e-02 -7.23926902e-01 -9.48256373e-01 -9.00861681e-01
6.71525300e-02 8.63829374e-01 8.04698110e-01 3.74798596e-01
1.12345435e-01 7.46716261e-01 -1.51270282e+00 -1.46128476e-01
9.04516354e-02 8.54239225e-01 3.71584564e-01 1.43501663e+00
2.78734058e-01 5.14745831e-01 -6.49845123e-01 -3.91696990e-01
3.66816342e-01 1.04290858e-01 1.34130046e-01 1.03290938e-01
5.52525163e-01 -4.64694262e-01 -4.06486124e-01 1.24333453e+00
-1.15498915e-01 3.10350865e-01 1.16618395e+00 -1.69184983e+00
1.37280196e-01 -1.92349091e-01 -6.34721577e-01 6.58117115e-01
-4.99633290e-02 3.06391954e-01 2.57245451e-01 -6.84192657e-01
2.35066995e-01 1.48709738e+00 6.36552274e-01 5.28742015e-01
-8.10549974e-01 -5.41819334e-01 -3.26271981e-01 3.03215355e-01
-1.72387135e+00 -2.51694649e-01 3.56397301e-01 -7.36882150e-01
7.08034039e-01 1.11969262e-01 5.21119058e-01 9.05713797e-01
9.84189630e-01 5.50722003e-01 9.04319406e-01 -1.78410798e-01
1.37010351e-01 3.57677430e-01 2.91443199e-01 6.30600452e-01
6.28161609e-01 4.66900378e-01 -8.47287849e-02 -2.58612990e-01
6.57346547e-01 3.83675933e-01 5.24412394e-01 -4.91023883e-02
-7.95070231e-01 6.97887242e-01 -2.32534423e-01 -9.67537835e-02
-6.40738010e-01 8.27751815e-01 -3.18602443e-01 -1.89555809e-01
1.31635070e-01 -6.21549845e-01 2.84126937e-01 1.38443604e-01
-9.10002410e-01 3.05049002e-01 7.73939848e-01 1.21680760e+00
8.87434542e-01 -1.23094879e-02 -7.57016987e-02 1.21644065e-01
1.13848722e+00 1.44968259e+00 -2.78911650e-01 -1.15249336e+00
9.80677977e-02 1.31274223e-01 5.50079465e-01 -1.05162907e+00
-2.08574325e-01 -1.14461720e-01 -6.19440913e-01 6.68561578e-01
4.72158730e-01 -4.23381001e-01 -7.88957417e-01 1.16273999e+00
4.78678405e-01 8.61485064e-01 8.91647339e-02 6.91638768e-01
3.02722663e-01 1.04662776e+00 -7.02160299e-02 -2.58466452e-01
1.22076762e+00 -5.73770821e-01 -7.99659669e-01 -1.52957380e-01
2.41407022e-01 -6.99036300e-01 -3.05783927e-01 6.06949806e-01
-8.05945933e-01 -6.66974127e-01 -7.02245831e-01 6.85095012e-01
-2.98646867e-01 1.28950641e-01 -1.74412847e-01 9.73854482e-01
-9.64072049e-01 1.35678872e-01 -1.17404044e+00 -2.96521872e-01
2.00301230e-01 3.66275996e-01 -2.84125153e-02 -2.04102188e-01
-5.77108443e-01 1.03809392e+00 4.00591820e-01 2.11128399e-01
-1.61068630e+00 -1.15739822e-01 -5.09517312e-01 -2.65953660e-01
3.35659385e-01 -6.77670717e-01 8.60970080e-01 -5.87653100e-01
-1.22241342e+00 5.48654735e-01 -5.73887348e-01 -7.29479969e-01
4.80493873e-01 -3.09153974e-01 -6.74527764e-01 2.78482914e-01
-1.06886156e-01 5.80470741e-01 1.17197001e+00 -1.45378125e+00
-1.08448672e+00 -2.16469496e-01 -4.08390611e-01 -2.65592873e-01
2.79896289e-01 2.12022066e-01 -2.67494440e-01 -1.14725016e-01
-3.56956482e-01 -1.21130240e+00 -2.47118592e-01 -2.39755064e-01
-2.64612973e-01 -1.88409314e-01 1.18056440e+00 -3.20401400e-01
9.43481266e-01 -2.16201067e+00 -7.68782869e-02 3.14379483e-01
1.20991819e-01 1.74431890e-01 1.30209059e-01 3.64809364e-01
6.30873382e-01 -4.70597535e-01 1.16105355e-01 -5.30986249e-01
-3.32092464e-01 3.17524135e-01 -9.73770097e-02 9.96300161e-01
-1.52191326e-01 4.77073759e-01 -7.66312301e-01 -5.68519175e-01
7.03481615e-01 5.41764975e-01 -1.74618199e-01 1.26664683e-01
-1.38534695e-01 1.89952627e-01 -3.66714567e-01 3.14269304e-01
1.09378040e+00 2.57783145e-01 -4.96261299e-01 6.84914663e-02
-3.62076402e-01 -6.41016901e-01 -1.63938332e+00 6.15034282e-01
-1.60146162e-01 9.38777745e-01 1.96500689e-01 -5.43636322e-01
9.23882961e-01 3.97232145e-01 7.04381466e-01 2.29879588e-01
4.08326834e-01 -1.09116599e-01 -9.80557874e-02 -5.73142886e-01
7.22228348e-01 -2.05049843e-01 1.89251870e-01 1.90323919e-01
2.95296729e-01 1.24909259e-01 2.24706665e-01 1.35703221e-01
1.21124828e+00 -2.33986363e-01 -1.72174964e-02 -1.65506184e-01
5.05385458e-01 1.97575018e-01 5.15181839e-01 1.12569642e+00
-4.20080185e-01 2.57031918e-01 -2.21609905e-01 -4.84832168e-01
-9.28612769e-01 -1.40427661e+00 -4.18520160e-02 3.72619629e-01
3.71202856e-01 4.65646796e-02 -6.51929319e-01 -1.62287861e-01
9.87302661e-02 9.99184549e-01 -2.75549531e-01 -1.17976077e-01
-4.81877416e-01 -8.89903069e-01 4.03312147e-01 2.34717056e-01
3.34527373e-01 -6.13071144e-01 -7.50074565e-01 4.80975240e-01
1.61959782e-01 -1.36059916e+00 -1.14370137e-01 -1.84600815e-01
-5.86692452e-01 -1.13855898e+00 -3.52689892e-01 -2.22324342e-01
5.77473402e-01 6.93626940e-01 6.41287386e-01 -2.93946832e-01
-1.27157524e-01 1.09469056e+00 -4.52560037e-02 -7.69244671e-01
-6.92446768e-01 -7.55759656e-01 4.69150424e-01 6.38132811e-01
8.88526678e-01 -6.69356883e-02 -3.41663480e-01 6.11994326e-01
-6.21849418e-01 -4.72203344e-01 4.97148365e-01 1.67160645e-01
4.55261320e-01 4.55643684e-01 -1.04689393e-02 -1.83952302e-01
1.10673264e-01 -6.34475350e-01 -1.43009973e+00 1.55929927e-04
6.70995377e-03 -4.40782785e-01 1.07720293e-01 -5.61797380e-01
-8.25307369e-01 3.75467837e-01 2.30364934e-01 -1.13206184e+00
-7.64351010e-01 -8.69795457e-02 1.60427988e-01 -2.91912884e-01
2.30461970e-01 2.23136827e-01 4.15736996e-03 -1.16313472e-01
1.56803280e-01 7.92434812e-01 5.68865120e-01 -4.36441079e-02
1.00613248e+00 8.71102512e-01 6.52683020e-01 -1.59945035e+00
-4.46296245e-01 -1.03792596e+00 -6.97277844e-01 -1.22667134e+00
1.36950839e+00 -9.99252141e-01 -1.23607039e+00 8.71658504e-01
-1.43905139e+00 1.88689469e-03 3.21369827e-01 1.10444963e+00
-4.84154373e-01 2.29123637e-01 -2.43927166e-01 -1.68886471e+00
4.95402932e-01 -1.17597485e+00 1.12874722e+00 4.18601125e-01
3.40008646e-01 -1.26955569e+00 4.14526194e-01 3.94115418e-01
7.70528987e-02 5.19081280e-02 -1.39729017e-02 -2.95583606e-01
-1.32937086e+00 -7.47559786e-01 1.48270711e-01 4.32574779e-01
-2.73753136e-01 5.02325237e-01 -9.45734322e-01 -3.11721504e-01
2.10759565e-01 5.18755019e-01 9.20414627e-01 1.11148703e+00
4.77027655e-01 -4.29092385e-02 -9.45943236e-01 3.68986011e-01
1.66515923e+00 4.70459759e-01 3.65141839e-01 9.28828791e-02
7.46302605e-01 2.71771461e-01 8.17247510e-01 3.40068132e-01
3.85097176e-01 6.88706398e-01 8.24710548e-01 5.92345297e-01
2.51751125e-01 2.39510074e-01 7.39460230e-01 2.37705007e-01
-2.13645250e-01 -9.66258168e-01 -9.41567183e-01 5.19850492e-01
-1.75530291e+00 -1.38643861e+00 -8.33859503e-01 2.30219984e+00
-3.87325674e-01 3.05726469e-01 6.00794435e-01 -3.19058567e-01
1.17710865e+00 -2.21243471e-01 3.01156528e-02 3.76477167e-02
1.94753021e-01 -4.82296318e-01 1.29791784e+00 7.69887507e-01
-1.24954093e+00 5.99992394e-01 6.47270823e+00 7.59615242e-01
-6.15404069e-01 3.32934886e-01 -1.00447550e-01 -3.06113455e-02
4.04438347e-01 -8.45690221e-02 -1.86586297e+00 7.52993822e-01
1.50461721e+00 1.59391627e-01 2.45366424e-01 6.37632430e-01
2.97789901e-01 -6.64983571e-01 -7.25650132e-01 9.20229971e-01
1.91612139e-01 -1.01563358e+00 -2.49972135e-01 4.47154254e-01
5.73839188e-01 4.22605038e-01 3.94505411e-02 -5.30773103e-02
6.32421136e-01 -4.14047629e-01 7.85446584e-01 1.15423584e+00
2.82562524e-02 -7.36509085e-01 7.51174390e-01 7.57801950e-01
-1.23485434e+00 -2.31163174e-01 -6.28313661e-01 1.04218356e-01
7.84466028e-01 4.67041552e-01 -9.40036237e-01 2.02835575e-01
6.22889042e-01 5.25278270e-01 -3.21438223e-01 1.59121406e+00
3.04968923e-01 7.40647316e-01 -9.46255445e-01 -4.05652016e-01
4.50426638e-01 -6.37530804e-01 1.38087380e+00 1.17603290e+00
6.38653338e-01 -7.71129876e-02 3.61282825e-01 8.38043988e-01
4.42548275e-01 -7.11141527e-01 -9.91436541e-01 3.08659583e-01
5.66663504e-01 1.30428469e+00 -9.36269462e-01 -4.96461600e-01
-5.40024638e-01 3.32313359e-01 -4.75962877e-01 3.91045213e-01
-1.02879119e+00 1.36200741e-01 5.83627224e-01 9.71068367e-02
6.20997846e-01 -4.92967218e-01 3.77521843e-01 -7.39801466e-01
-3.61267895e-01 7.00469688e-02 1.84270486e-01 -8.18409920e-01
-1.01099348e+00 3.20342213e-01 6.37497604e-01 -1.42413461e+00
-1.07863873e-01 -9.31593776e-01 -7.18707323e-01 7.10084856e-01
-1.13652909e+00 -1.00586677e+00 -9.93385836e-02 4.40771401e-01
3.09794873e-01 -3.91032815e-01 8.98843035e-02 3.33393544e-01
-5.80924690e-01 -1.31089523e-01 4.13759887e-01 5.72307296e-02
2.50675797e-01 -9.75862741e-01 4.01974209e-02 1.09251785e+00
5.52532785e-02 1.59213796e-01 9.94014502e-01 -9.87180471e-01
-1.55854070e+00 -1.40384054e+00 3.28172088e-01 -1.00944245e+00
5.90649009e-01 -2.80754179e-01 -5.63657999e-01 1.10720801e+00
2.08869502e-01 1.25978306e-01 2.53502578e-01 -5.76635778e-01
4.96958703e-01 -2.28267446e-01 -1.13805079e+00 -1.60124004e-02
1.95097461e-01 -1.22774139e-01 -5.61232746e-01 4.08094764e-01
1.24440938e-01 -9.96448174e-02 -5.62385976e-01 1.14000430e-02
4.67691541e-01 -8.10503006e-01 9.59197581e-01 -1.81856975e-01
-4.69345033e-01 -6.82517171e-01 -2.80805022e-01 -1.05284059e+00
-4.12945926e-01 -4.20183897e-01 -1.25723317e-01 1.09429967e+00
1.22817857e-02 -5.33087611e-01 5.87389946e-01 1.77639425e-02
-2.51887664e-02 6.92229792e-02 -1.32011473e+00 -1.13690650e+00
-3.36067557e-01 -8.05976510e-01 1.03296489e-01 5.17175868e-02
-7.89535820e-01 1.81615412e-01 -6.52994215e-01 1.07860374e+00
1.44185805e+00 -5.92719615e-01 1.04604793e+00 -1.39663613e+00
-9.71716642e-02 -1.29445240e-01 -9.20139015e-01 -9.53747094e-01
1.10174276e-01 -3.40109766e-01 3.93973172e-01 -9.97500300e-01
1.94816664e-01 -2.51615848e-02 -6.30269870e-02 -5.30654490e-01
1.12476043e-01 9.11166593e-02 1.64132625e-01 4.61561307e-02
-8.71116519e-01 9.09245238e-02 5.34039438e-01 -6.00793920e-02
1.59910306e-01 7.99533308e-01 5.94319105e-01 9.18919146e-01
4.55639601e-01 -1.11095095e+00 9.80952531e-02 -2.25880101e-01
6.42453432e-02 3.70929211e-01 9.72293735e-01 -1.58980536e+00
6.06848836e-01 -2.70069599e-01 2.03809768e-01 -1.49565578e+00
8.95753920e-01 -1.33959615e+00 6.82630837e-01 7.07252383e-01
3.52458447e-01 3.05236012e-01 1.12420827e-01 1.07988691e+00
1.01818338e-01 -6.58240438e-01 1.04443765e+00 -1.67680219e-01
-7.01886237e-01 3.63007993e-01 -1.19880104e+00 -5.04328907e-01
1.63680387e+00 -3.53070647e-01 -9.69189927e-02 -2.50140578e-01
-5.95920444e-01 2.70440757e-01 1.66423768e-01 1.08764090e-01
6.32367969e-01 -1.25114048e+00 -6.99183106e-01 1.60358131e-01
-1.88052595e-01 -2.36086845e-01 3.09165537e-01 9.14565563e-01
-5.49495280e-01 5.18548310e-01 -8.18037614e-02 -1.21080673e+00
-1.43643308e+00 8.40782821e-01 5.58104157e-01 1.77478120e-01
-1.40386805e-01 5.98834991e-01 -8.61999467e-02 6.13703877e-02
1.28842434e-02 -1.40332818e-01 -3.10911566e-01 -1.60309166e-01
7.28942037e-01 1.08585572e+00 -5.24389207e-01 -1.26537371e+00
-4.21612591e-01 7.62331426e-01 2.05913186e-01 -5.26506662e-01
1.06030679e+00 -3.78975749e-01 8.11125413e-02 7.65703261e-01
6.54422998e-01 -1.23592392e-01 -1.43255901e+00 1.20347187e-01
-3.61044765e-01 -5.12465477e-01 3.83176655e-01 -1.62283704e-01
-7.28414536e-01 9.73600268e-01 9.37058449e-01 3.70824695e-01
4.46121156e-01 -6.25048727e-02 3.96656930e-01 4.93547469e-01
5.54489136e-01 -8.02250087e-01 -1.84688047e-01 3.98228139e-01
1.51213363e-01 -1.14684975e+00 -2.26422548e-01 -1.88580781e-01
-3.65138352e-01 1.26155400e+00 4.08208609e-01 -5.67083538e-01
1.11740208e+00 5.24921656e-01 -4.33938690e-02 -2.39356995e-01
-7.11344719e-01 -3.91284347e-01 1.44585609e-01 7.97200322e-01
-5.43747604e-01 2.58797139e-01 4.75416929e-01 -2.33218491e-01
3.44890952e-01 1.30412593e-01 7.32389092e-01 8.00850153e-01
-9.04755056e-01 -5.47948003e-01 -1.09455884e+00 2.51442820e-01
-3.48507971e-01 4.28341508e-01 1.66198805e-01 7.58097351e-01
1.76004350e-01 1.25823379e+00 5.14370203e-01 -4.01065409e-01
9.23212320e-02 -1.58765718e-01 5.25919318e-01 -3.86955589e-01
-1.33631721e-01 3.37903529e-01 -2.92714149e-01 -2.00292692e-01
-4.73013222e-01 -1.20990312e+00 -7.39867747e-01 -3.21055323e-01
-6.81038201e-01 -1.48405069e-02 7.80892551e-01 1.20729208e+00
-5.41952960e-02 3.53976548e-01 6.28162384e-01 -9.49056268e-01
-3.26972336e-01 -1.03631985e+00 -7.91572511e-01 -2.79341340e-01
5.46354651e-01 -1.03355801e+00 -5.73874116e-01 1.53744787e-01] | [6.578239917755127, -2.0402681827545166] |
223e0d36-30cf-42fd-96ab-400906c57d85 | prior-networks-for-detection-of-adversarial | 1812.02575 | null | http://arxiv.org/abs/1812.02575v1 | http://arxiv.org/pdf/1812.02575v1.pdf | Prior Networks for Detection of Adversarial Attacks | Adversarial examples are considered a serious issue for safety critical
applications of AI, such as finance, autonomous vehicle control and medicinal
applications. Though significant work has resulted in increased robustness of
systems to these attacks, systems are still vulnerable to well-crafted attacks.
To address this problem, several adversarial attack detection methods have been
proposed. However, a system can still be vulnerable to adversarial samples that
are designed to specifically evade these detection methods. One recent
detection scheme that has shown good performance is based on uncertainty
estimates derived from Monte-Carlo dropout ensembles. Prior Networks, a new
method of estimating predictive uncertainty, has been shown to outperform
Monte-Carlo dropout on a range of tasks. One of the advantages of this approach
is that the behaviour of a Prior Network can be explicitly tuned to, for
example, predict high uncertainty in regions where there are no training data
samples. In this work, Prior Networks are applied to adversarial attack
detection using measures of uncertainty in a similar fashion to Monte-Carlo
Dropout. Detection based on measures of uncertainty derived from DNNs and
Monte-Carlo dropout ensembles are used as a baseline. Prior Networks are shown
to significantly out-perform these baseline approaches over a range of
adversarial attacks in both detection of whitebox and blackbox configurations.
Even when the adversarial attacks are constructed with full knowledge of the
detection mechanism, it is shown to be highly challenging to successfully
generate an adversarial sample. | ['Andrey Malinin', 'Mark Gales'] | 2018-12-06 | null | https://openreview.net/forum?id=H1gh_sC9tm | https://openreview.net/pdf?id=H1gh_sC9tm | null | ['adversarial-attack-detection', 'adversarial-attack-detection'] | ['computer-vision', 'knowledge-base'] | [ 2.77186960e-01 1.68573007e-01 8.87178350e-03 -3.17889363e-01
-1.09095442e+00 -7.70396531e-01 8.67930472e-01 4.90331016e-02
-4.44813371e-01 1.09338999e+00 -1.81498408e-01 -2.32986420e-01
2.33952910e-01 -9.06908214e-01 -1.17483628e+00 -6.94540441e-01
-3.68175516e-03 5.34848392e-01 3.20365191e-01 -1.90597847e-01
9.79372263e-02 7.49913692e-01 -1.04479837e+00 2.76804686e-01
7.65316606e-01 5.71945429e-01 -5.16458273e-01 7.53426373e-01
2.80263871e-01 5.51082790e-01 -1.09342074e+00 -4.53865618e-01
2.82849222e-01 -4.49047118e-01 -3.59098047e-01 -4.49607074e-01
5.08338511e-01 -5.89607716e-01 -4.95774657e-01 1.24843884e+00
6.10244632e-01 9.47895274e-02 8.65201831e-01 -1.26222622e+00
-3.06383282e-01 5.88194430e-01 2.23991219e-02 1.83363497e-01
8.54752585e-02 7.92156875e-01 6.49598122e-01 -4.34187263e-01
4.02534127e-01 1.62022400e+00 6.25788152e-01 9.54175353e-01
-1.63244903e+00 -9.41748321e-01 -1.46341145e-01 -3.34952444e-01
-9.92188871e-01 -4.50813115e-01 2.97795564e-01 -4.42751616e-01
9.68422055e-01 -2.59123966e-02 1.48163229e-01 1.69106793e+00
6.71663165e-01 6.35588408e-01 9.74316537e-01 1.11678734e-01
7.16234744e-01 2.93244839e-01 -2.46763542e-01 2.24928081e-01
4.25040513e-01 8.20045471e-01 -1.56282201e-01 -4.14583057e-01
5.53975523e-01 -1.33702695e-01 -1.50157228e-01 -1.23331420e-01
-4.12558585e-01 1.12466037e+00 6.07604742e-01 5.32565638e-02
-1.24284588e-01 6.84248209e-01 6.07988000e-01 2.63920009e-01
3.36451322e-01 1.04259515e+00 -2.68617749e-01 -4.87952083e-02
-8.50651622e-01 5.76910079e-01 9.76452529e-01 5.86130857e-01
3.44787389e-01 7.44736135e-01 -5.49146056e-01 2.56171376e-01
1.10367283e-01 4.45461959e-01 9.49925110e-02 -7.31432140e-01
3.71447414e-01 1.50055572e-01 2.79686809e-01 -6.24091089e-01
-1.88139930e-01 -4.58701640e-01 -6.29374981e-01 1.06603551e+00
7.74971545e-01 -5.65381706e-01 -1.38650668e+00 1.84987879e+00
1.03652120e-01 3.33861530e-01 3.97238553e-01 6.07129753e-01
4.85385388e-01 5.59409618e-01 2.72102356e-01 3.71256709e-01
9.61081266e-01 -2.99793214e-01 -5.96660733e-01 -3.90388817e-01
2.96283722e-01 -4.54858959e-01 7.17396140e-01 4.87734318e-01
-1.09717226e+00 -1.65930778e-01 -1.50445259e+00 5.25435686e-01
-6.55671656e-01 -5.73185503e-01 3.73995543e-01 1.28958499e+00
-5.45191407e-01 1.08092558e+00 -1.00654149e+00 5.04836291e-02
9.87779021e-01 6.08288407e-01 -1.53586939e-01 -5.46158776e-02
-1.76398838e+00 1.28905356e+00 4.81281996e-01 -1.82032436e-01
-1.87964702e+00 -8.99872065e-01 -7.22055018e-01 3.64025868e-02
2.09909439e-01 -5.59889078e-01 1.01803362e+00 -1.07934141e+00
-1.52627587e+00 4.18252289e-01 5.30552685e-01 -1.17355144e+00
9.17041540e-01 -2.64531642e-01 -9.12954882e-02 6.62303269e-02
-3.97814602e-01 7.39588380e-01 1.16534746e+00 -1.02062738e+00
-1.98814645e-01 -9.46835205e-02 2.75241166e-01 -4.37449254e-02
-1.50083289e-01 2.52181858e-01 1.35302961e-01 -7.01247394e-01
-7.95304477e-01 -8.69462907e-01 -5.06670833e-01 -5.82437366e-02
-4.90028620e-01 2.60294706e-01 8.62862110e-01 -3.60917807e-01
7.49238849e-01 -1.88748074e+00 -9.64609534e-02 2.33147442e-01
-8.61507002e-03 8.44012558e-01 1.56320501e-02 2.45145097e-01
-9.20671895e-02 4.90758747e-01 -5.72915077e-01 -4.83373962e-02
1.11607246e-01 2.28030697e-01 -5.85892200e-01 5.34453332e-01
9.38114405e-01 7.10523665e-01 -8.46062839e-01 9.47597772e-02
2.48824775e-01 7.81002522e-01 -6.42537773e-01 2.45076671e-01
-6.35770679e-01 6.53372109e-01 -4.53448266e-01 3.17969292e-01
4.72394019e-01 3.39020252e-01 -3.64036858e-01 2.94466078e-01
5.45079052e-01 1.04635336e-01 -9.56570804e-01 1.05769932e+00
-2.92289734e-01 5.05886078e-01 -6.59776255e-02 -7.64707267e-01
7.11055934e-01 5.51144242e-01 -2.42321324e-02 -1.36918247e-01
4.71292257e-01 2.11841404e-01 3.90197217e-01 2.97924709e-02
7.91538805e-02 -3.98661315e-01 -3.37230623e-01 6.82256967e-02
2.80585736e-02 -6.73358977e-01 3.40044424e-02 3.22324753e-01
1.56155837e+00 1.63283601e-01 -7.37319216e-02 1.19019235e-02
2.90615916e-01 3.20265628e-02 7.26821721e-01 1.03932095e+00
-3.98567080e-01 7.00236976e-01 8.73932421e-01 -1.89410269e-01
-1.33553576e+00 -1.29197896e+00 -4.20423746e-01 5.23233294e-01
-1.48274466e-01 -7.97875896e-02 -1.08226967e+00 -7.95317948e-01
2.80819923e-01 1.18639207e+00 -6.85610592e-01 -8.25887561e-01
-3.59320939e-01 -1.12587821e+00 1.16571856e+00 6.67871177e-01
3.05187047e-01 -1.13821316e+00 -4.19106394e-01 3.30511808e-01
6.64353490e-01 -1.02224338e+00 -1.40348241e-01 4.54067975e-01
-7.49192417e-01 -9.40598667e-01 -5.15054643e-01 -7.00890943e-02
6.22823358e-01 -7.26479650e-01 1.09031940e+00 -1.53687730e-01
-5.61656594e-01 1.14414133e-01 -1.04316317e-01 -7.53480196e-01
-1.14426064e+00 -1.70992687e-01 1.45932242e-01 -2.28089631e-01
2.53596723e-01 -5.45791805e-01 -4.10970867e-01 2.38530129e-01
-1.24121654e+00 -7.34252155e-01 5.42523205e-01 1.13169861e+00
3.17371994e-01 1.15884192e-01 9.31737840e-01 -1.20851922e+00
7.79910207e-01 -6.10509157e-01 -9.28131402e-01 -1.19467147e-01
-4.67225194e-01 3.90134096e-01 9.81019258e-01 -6.77774549e-01
-7.84930289e-01 8.71843249e-02 -4.16151911e-01 -7.07556605e-01
-3.17233890e-01 5.85865714e-02 -4.35377300e-01 -2.22790569e-01
1.18486655e+00 -3.99202943e-01 4.04326431e-02 5.06650843e-02
3.39134187e-01 2.77297944e-01 3.26626509e-01 -6.45378113e-01
9.81048465e-01 2.38982916e-01 1.76452443e-01 -5.20011246e-01
-7.47809231e-01 3.31275135e-01 -1.09181836e-01 6.91973865e-02
8.87930691e-01 -7.18842387e-01 -6.04764998e-01 4.93219256e-01
-9.04769719e-01 -3.98418307e-01 -3.75425220e-01 4.00800735e-01
-5.08697033e-01 5.89243807e-02 -6.74261212e-01 -9.44393575e-01
-1.83688968e-01 -1.50278473e+00 6.82004452e-01 3.82003456e-01
-2.59234816e-01 -1.04377973e+00 -8.98962915e-02 -3.82455289e-02
5.90727031e-01 8.15294087e-01 7.75713742e-01 -1.16339242e+00
-5.57461381e-01 -7.52951086e-01 1.65210873e-01 7.64952004e-01
-1.88335985e-01 1.33370683e-01 -1.34121108e+00 -4.49161649e-01
-7.86632970e-02 -7.16467202e-01 1.12043202e+00 4.30654883e-01
8.38980317e-01 -1.97326869e-01 -1.61859065e-01 4.47121441e-01
1.22373486e+00 2.30983645e-01 8.61652672e-01 1.08788311e-01
4.67335880e-01 3.24293613e-01 3.25643778e-01 6.17075525e-02
-5.17864525e-01 4.61636424e-01 9.56998050e-01 4.70392592e-02
-2.85736304e-02 -1.67186216e-01 8.31030667e-01 -3.99645925e-01
5.48780143e-01 -4.11645025e-01 -7.64924049e-01 2.28431150e-01
-1.58275580e+00 -1.15998280e+00 1.16983622e-01 2.47663236e+00
9.51834738e-01 9.42266464e-01 9.77544785e-02 -2.84546204e-02
7.27222502e-01 -1.22937873e-01 -1.10075772e+00 -7.96525538e-01
-1.03283241e-01 5.40835381e-01 8.18322003e-01 6.09826028e-01
-1.40642321e+00 1.11361408e+00 6.60515976e+00 8.66243362e-01
-7.88596451e-01 -5.32265753e-04 9.28174615e-01 -2.48974398e-01
-4.80882935e-02 1.09786794e-01 -1.05736601e+00 5.74988186e-01
1.40645397e+00 -4.34702747e-02 3.60983431e-01 7.52574563e-01
1.49298713e-01 -3.99361625e-02 -1.39795768e+00 3.08720767e-01
-7.99644068e-02 -1.21354079e+00 -1.26453312e-02 -5.20923138e-02
8.60745072e-01 -4.11337465e-02 5.23002028e-01 5.91962159e-01
1.08955383e+00 -1.54219747e+00 4.10416007e-01 4.09840643e-01
6.09439731e-01 -1.15766299e+00 8.31346273e-01 2.56581515e-01
-3.93839508e-01 6.13621697e-02 -4.11925972e-01 2.05000684e-01
3.26179974e-02 5.94882309e-01 -1.01327550e+00 1.86902434e-01
3.00079495e-01 1.71318799e-01 -3.04967046e-01 1.06139696e+00
-6.09370112e-01 8.42161596e-01 -5.64594090e-01 -2.26725526e-02
5.07753074e-01 1.82907283e-01 8.10410082e-01 1.04317343e+00
-6.29307404e-02 -3.52407932e-01 4.82939929e-02 1.31966829e+00
-2.67032862e-01 -5.34637153e-01 -8.76494884e-01 -2.37194821e-01
4.53381360e-01 8.74947131e-01 -4.52891290e-01 -2.80903995e-01
5.55080958e-02 7.59768724e-01 1.81085706e-01 3.54937643e-01
-1.15383923e+00 -6.80763602e-01 7.60493696e-01 -1.04262710e-01
2.86049277e-01 4.87275422e-02 -2.64284402e-01 -7.94504225e-01
-3.24849308e-01 -1.14329839e+00 2.57702053e-01 -4.65800643e-01
-1.44059849e+00 5.96677661e-01 2.28287075e-02 -8.71701956e-01
-5.33504426e-01 -7.81983495e-01 -9.11406517e-01 1.15123904e+00
-1.11104953e+00 -7.33768702e-01 3.34163070e-01 3.01298052e-01
3.79242152e-01 -3.36066693e-01 8.45537543e-01 8.88345912e-02
-8.40231419e-01 9.14170742e-01 3.55007946e-01 1.52121678e-01
8.04717958e-01 -1.36489236e+00 6.92929447e-01 1.13024926e+00
-1.11350432e-01 4.12357360e-01 1.15355837e+00 -1.00150442e+00
-1.16119039e+00 -1.41921210e+00 -1.34343226e-02 -8.40114713e-01
6.73332632e-01 -5.61123490e-01 -9.80248332e-01 5.51766634e-01
-3.23008820e-02 1.40344441e-01 3.23495746e-01 -2.01699868e-01
-4.31136608e-01 7.35011101e-02 -1.69785166e+00 6.94315851e-01
4.02323723e-01 -3.80881965e-01 -5.61456323e-01 4.19496447e-01
5.54658353e-01 -5.45001328e-01 -7.21000016e-01 3.97796273e-01
3.24847728e-01 -7.67886579e-01 1.07019174e+00 -9.45867956e-01
3.61356556e-01 -2.67205000e-01 1.44876197e-01 -1.61088657e+00
2.01804549e-01 -9.53090250e-01 -1.69829711e-01 1.27670443e+00
6.78008199e-01 -7.91437745e-01 8.68084073e-01 9.67966318e-01
-4.91918102e-02 -5.45406103e-01 -9.08532023e-01 -9.94793177e-01
7.73499131e-01 -3.49781036e-01 3.07341635e-01 5.35432696e-01
-4.27238643e-01 1.38667747e-01 -4.02158260e-01 4.19784665e-01
7.11957991e-01 -7.28880882e-01 7.38618612e-01 -8.95884216e-01
-4.33535695e-01 -5.92363477e-01 -7.19367027e-01 -3.02671254e-01
2.72017807e-01 -7.95129418e-01 2.58117527e-01 -1.05330098e+00
-1.43302277e-01 -2.94048548e-01 -2.61798441e-01 3.17880481e-01
-4.31671560e-01 2.51959085e-01 2.00036354e-03 -1.16404317e-01
-6.80001304e-02 4.35361594e-01 7.28356838e-01 -2.97966063e-01
-1.86328348e-02 2.38431603e-01 -4.59001064e-01 7.86263943e-01
9.81171072e-01 -9.68506694e-01 -3.56626660e-01 2.42279787e-02
4.45839614e-02 -1.62761256e-01 4.83929574e-01 -1.31442547e+00
-6.08438216e-02 -1.98620297e-02 6.12082064e-01 -1.29928142e-01
5.14146209e-01 -6.30012512e-01 2.77161635e-02 7.86371827e-01
-4.72617477e-01 -3.43851000e-01 6.51476026e-01 8.15026700e-01
1.00991838e-02 -4.79140341e-01 1.27037644e+00 -1.48620293e-01
-1.20709211e-01 3.51793230e-01 -5.74171841e-01 4.40287828e-01
1.20128846e+00 -4.26515229e-02 -2.58435398e-01 -3.84507239e-01
-8.55179727e-01 3.43262285e-01 4.63140875e-01 1.19787022e-01
5.75122416e-01 -1.10929275e+00 -8.37469459e-01 9.46609527e-02
-2.09451169e-01 -1.16030678e-01 8.23965855e-03 3.59236956e-01
-4.87533718e-01 9.41364765e-02 -2.56781727e-01 -4.30215806e-01
-9.09052312e-01 7.04128683e-01 6.83809578e-01 -4.30845231e-01
-3.62586141e-01 8.67447674e-01 2.21984804e-01 -1.64295465e-01
5.04646599e-01 -9.68807638e-02 4.06697169e-02 -3.10723960e-01
5.22784352e-01 1.33268774e-01 1.23748800e-03 -1.51223317e-01
-1.68105438e-01 -1.27347019e-02 -2.90173203e-01 -1.56060129e-01
1.13182914e+00 7.22105801e-01 5.86395562e-01 2.07699418e-01
8.76158416e-01 -2.79999942e-01 -1.76850820e+00 2.19972923e-01
-4.16389853e-02 -3.78824532e-01 6.30492717e-02 -9.79221225e-01
-9.62929010e-01 1.22171462e+00 7.28008091e-01 1.94966774e-02
7.00351119e-01 -5.62064350e-01 5.73137224e-01 3.92253876e-01
7.44203329e-02 -9.36395645e-01 8.14204961e-02 4.21592623e-01
7.60305405e-01 -1.23531544e+00 1.36667555e-02 5.10443300e-02
-6.10230029e-01 8.65195572e-01 7.48113871e-01 -7.40606129e-01
4.37617153e-01 6.96147978e-01 -2.33387217e-01 1.06622770e-01
-6.96015596e-01 2.32141539e-02 9.98054445e-02 7.60450065e-01
1.60554305e-01 -1.54571772e-01 2.08507732e-01 3.18547904e-01
1.61166579e-01 -2.68869996e-01 6.77423775e-01 8.70849192e-01
-2.72283673e-01 -1.17783237e+00 -5.24373651e-01 5.29127061e-01
-9.84706342e-01 -2.43915617e-01 -4.57265407e-01 7.18591034e-01
-2.52853662e-01 8.61492276e-01 -2.14750499e-01 -1.89195901e-01
3.15275371e-01 2.30826721e-01 4.30577755e-01 -5.95688462e-01
-1.21186578e+00 -3.35575372e-01 2.63035476e-01 -5.00045478e-01
2.60690689e-01 -6.20369256e-01 -1.17988861e+00 -2.46963158e-01
-4.18355316e-01 4.85163890e-02 5.58446050e-01 8.19794118e-01
1.88293621e-01 8.72976184e-01 3.64737123e-01 -8.65640044e-01
-1.17984402e+00 -7.66923249e-01 -3.42486113e-01 5.53920090e-01
3.17640185e-01 -7.87656724e-01 -7.28186965e-01 -3.08755964e-01] | [5.647232532501221, 7.856144905090332] |
6b99c2d5-9af4-4ef6-81ff-e27cd069dbe2 | automatic-skin-lesion-segmentation-using-deep | 1807.06466 | null | http://arxiv.org/abs/1807.06466v1 | http://arxiv.org/pdf/1807.06466v1.pdf | Automatic Skin Lesion Segmentation Using Deep Fully Convolutional Networks | This paper summarizes our method and validation results for the ISIC
Challenge 2018 - Skin Lesion Analysis Towards Melanoma Detection - Task 1:
Lesion Segmentation | ['Hongming Xu', 'Tae Hyun Hwang'] | 2018-07-17 | null | null | null | null | ['skin-lesion-segmentation'] | ['medical'] | [ 1.02169394e+00 1.29649537e-02 -6.90934122e-01 1.33312494e-01
-1.35453868e+00 -5.32573164e-01 6.50741339e-01 7.89962932e-02
-6.05744958e-01 4.21840668e-01 -1.34663165e-01 -6.66431487e-01
3.31978768e-01 -2.84893792e-02 1.95076335e-02 -9.20407057e-01
1.51909301e-02 -1.60157621e-01 5.39524138e-01 1.41423447e-02
1.13800496e-01 1.42689422e-01 -6.30969465e-01 9.10619915e-01
9.11962569e-01 8.34205866e-01 -1.78867027e-01 1.59121180e+00
2.48645946e-01 2.18468651e-01 -3.49213034e-01 -7.81400442e-01
-3.20205055e-02 -3.69967371e-01 -1.28007412e+00 2.39220127e-01
1.24286902e+00 7.23435879e-02 -7.72716701e-02 1.31696272e+00
5.87530196e-01 -7.60628164e-01 1.32402682e+00 -5.15258133e-01
8.73499140e-02 2.27899402e-01 -9.40908253e-01 6.37522459e-01
2.55320847e-01 4.91772234e-01 2.19462171e-01 -4.50114489e-01
1.19616067e+00 5.41056395e-01 1.41225576e+00 1.36412525e+00
-8.79451871e-01 2.60945797e-01 -1.24186799e-01 3.60442281e-01
-1.22821271e+00 -1.63138002e-01 1.06420115e-01 -6.40858471e-01
1.04695368e+00 1.21136785e+00 1.02097976e+00 1.59708798e+00
2.06709862e-01 1.35659719e+00 1.51041806e+00 -5.53994477e-01
7.30025098e-02 5.96420690e-02 5.78247070e-01 9.45216954e-01
4.17150408e-01 -1.52679607e-01 -1.34553701e-01 -1.78072929e-01
7.37535417e-01 -8.23939383e-01 -4.11317647e-02 4.61209029e-01
-9.24541414e-01 4.32724714e-01 5.59507728e-01 -2.08721571e-02
-3.43996823e-01 3.70878607e-01 9.42942977e-01 -1.06887944e-01
6.67464316e-01 1.84825376e-01 1.28612354e-01 2.18942463e-01
-1.08541965e+00 1.78641323e-02 4.03338134e-01 4.33473349e-01
-3.71775657e-01 -3.12855154e-01 -6.08678997e-01 1.08255374e+00
4.39029872e-01 4.45138544e-01 4.22242761e-01 -4.23563629e-01
2.65197545e-01 5.45551240e-01 -3.48353356e-01 -1.32004758e-02
-7.53897309e-01 -2.56387651e-01 -8.15968990e-01 1.95756689e-01
7.05090046e-01 -4.89500374e-01 -2.07670927e+00 4.82896388e-01
2.70716399e-01 1.08252130e-01 -1.01880476e-01 9.64628696e-01
1.12246454e+00 -1.82521701e-01 8.06938171e-01 6.27235994e-02
1.64489985e+00 -1.23190391e+00 -4.79635179e-01 -1.30165905e-01
1.04784477e+00 -7.81011939e-01 6.04925156e-01 2.04529464e-01
-1.02191174e+00 6.46248981e-02 -9.17201161e-01 1.13674410e-01
-7.37892568e-01 7.93612599e-01 8.93379092e-01 1.53341401e+00
-1.20240784e+00 -2.16841400e-01 -8.56102526e-01 -1.08272707e+00
9.92162108e-01 1.37154534e-01 -4.64693099e-01 1.44101912e-02
-7.56088793e-01 1.11920190e+00 1.22260585e-01 -8.41908436e-03
-9.22558904e-01 -1.12266445e+00 -4.61273581e-01 -1.16115844e+00
-3.18925008e-02 -6.50561094e-01 1.35542142e+00 -1.02920556e+00
-1.39987659e+00 1.72648907e+00 -4.08864886e-01 -1.01735842e+00
6.52090251e-01 1.77410349e-01 -8.98282528e-01 6.08687639e-01
-1.51803866e-01 1.11271667e+00 9.59894001e-01 -1.18924057e+00
-1.02726197e+00 -2.79795617e-01 -2.58325309e-01 1.09057374e-01
4.40820009e-02 2.23828137e-01 -6.41029477e-01 -4.96010214e-01
-3.29398990e-01 -9.48355734e-01 -6.43813014e-01 8.98931026e-02
-1.10107589e+00 -1.71705857e-01 3.75580966e-01 -1.11862552e+00
1.29611039e+00 -1.56777954e+00 -1.87512442e-01 7.11097956e-01
2.09446073e-01 6.91989124e-01 -4.61302072e-01 -2.62268573e-01
-3.74724209e-01 7.45310068e-01 -2.63852358e-01 -3.95492733e-01
-2.97641993e-01 -5.11531174e-01 3.08511108e-01 5.35989821e-01
2.01014772e-01 1.28180695e+00 -4.99594718e-01 -8.87930512e-01
5.00606835e-01 1.73579425e-01 3.97920549e-01 -5.16213298e-01
-4.84141037e-02 -1.58726156e-01 2.45588738e-02 1.45415235e+00
9.79387343e-01 2.05262914e-01 3.59844137e-03 -5.78577399e-01
-3.69913541e-02 -5.90595484e-01 -5.32059073e-01 1.57575202e+00
-2.83516526e-01 8.03976893e-01 4.27750766e-01 -8.09414461e-02
-2.29829624e-01 5.63234508e-01 5.19043505e-01 -4.62466627e-01
1.99064668e-02 3.82008553e-01 1.65126752e-02 -8.23958218e-01
-2.31826100e-02 1.40624776e-01 3.51845175e-01 -2.26112068e-01
1.21269949e-01 -2.66511708e-01 7.00157702e-01 2.31650293e-01
1.05027485e+00 -3.15642118e-01 4.22062606e-01 -4.28940743e-01
9.91123021e-01 5.83216906e-01 8.29607993e-03 5.10136485e-01
-9.71896350e-01 8.76455903e-01 6.72738433e-01 -2.54476458e-01
-3.87520492e-01 -1.34454560e+00 -6.97163701e-01 4.90090340e-01
-6.75694197e-02 -2.20917016e-01 -1.57047045e+00 -1.50604534e+00
3.36384661e-02 3.74256074e-01 -1.71806550e+00 4.26046878e-01
-1.67773709e-01 -1.31355834e+00 1.25573194e+00 3.84445727e-01
2.32797712e-01 -5.64615846e-01 -1.34525836e-01 -4.04464781e-01
1.54650062e-01 -1.10033953e+00 -5.28179884e-01 -1.37988523e-01
-5.47203720e-01 -2.13579202e+00 -1.55734253e+00 -1.16293538e+00
1.16705036e+00 2.57385243e-02 6.34086013e-01 1.89701453e-01
-1.70632386e+00 9.89984035e-01 -2.45182551e-02 -8.54095995e-01
-4.44473922e-01 3.16741675e-01 -5.10826707e-01 6.66891858e-02
6.13465130e-01 6.35102808e-01 -7.50426412e-01 3.09627764e-02
-8.06707859e-01 1.96593404e-01 6.52515531e-01 5.22789657e-01
7.06923783e-01 -5.31276107e-01 5.40585339e-01 -1.17896199e+00
7.44013369e-01 4.39239293e-02 -1.30624443e-01 9.96500373e-01
-6.02329858e-02 -1.15639579e+00 -1.74995005e-01 -6.37804866e-02
-1.33980513e+00 6.14092708e-01 -6.32372916e-01 5.14923036e-01
-7.00655520e-01 -8.66508633e-02 6.63031459e-01 -1.05078626e+00
1.17882693e+00 2.21827522e-01 1.91609770e-01 3.03070410e-03
2.06445500e-01 7.98519790e-01 5.93148470e-01 1.09710850e-01
4.59858179e-01 6.07101560e-01 3.31782758e-01 -1.14679718e+00
-9.75338638e-01 -1.16198313e+00 -1.08386874e+00 -7.54871428e-01
1.39056313e+00 -7.24262238e-01 5.25108501e-02 1.22333658e+00
-8.86895061e-01 -7.76093721e-01 -3.07505637e-01 -2.24558655e-02
-2.69829869e-01 3.80973697e-01 -8.52728069e-01 -6.97502553e-01
-9.92607534e-01 -1.10524583e+00 1.36225367e+00 7.25659966e-01
-2.75114179e-01 -1.82828021e+00 4.46722478e-01 4.54291761e-01
3.42684418e-01 7.54505396e-01 5.49037337e-01 -1.09896570e-01
3.88330370e-01 -7.68158853e-01 -3.53020668e-01 3.16375017e-01
-1.75575335e-02 6.61200285e-01 -9.58688498e-01 -6.07438646e-02
-1.20300317e+00 -4.37704265e-01 1.71486640e+00 8.08603227e-01
1.36606634e+00 6.32461548e-01 -8.11956048e-01 8.17897737e-01
1.50784314e+00 -4.11089391e-01 7.71672070e-01 5.24707325e-02
5.39195478e-01 5.62382400e-01 3.68599057e-01 -2.95357943e-01
-6.86780140e-02 4.01622355e-02 5.22349179e-01 -7.96074688e-01
-1.29382801e+00 2.94470713e-02 -2.17001319e-01 -7.53849819e-02
-5.92761636e-01 -1.26352102e-01 -1.23896682e+00 9.53640640e-01
-1.15831244e+00 -2.93328464e-01 -1.04967988e+00 1.66954446e+00
7.02645421e-01 -2.28597060e-01 5.78558207e-01 -2.37887412e-01
8.59721065e-01 -1.68827660e-02 -4.19089109e-01 -2.97399789e-01
-4.15889055e-01 4.23144609e-01 7.33528912e-01 4.73966539e-01
-1.94130492e+00 1.29562199e+00 8.78526402e+00 1.53885102e+00
-1.08883524e+00 2.73183316e-01 7.27242827e-01 1.46968633e-01
-2.72269584e-02 -6.42594695e-01 -7.78865457e-01 1.86810032e-01
7.31835783e-01 5.44770837e-01 -2.32176498e-01 1.50629669e-01
-3.76378953e-01 -7.38074839e-01 -5.65086305e-01 6.41064703e-01
1.91785216e-01 -1.76010656e+00 2.58354358e-02 1.35190055e-01
1.01971579e+00 3.79858352e-03 5.91322780e-01 -1.25790769e-02
-3.49287480e-01 -1.19509721e+00 -1.20492585e-01 7.99235880e-01
1.64016199e+00 -3.67817789e-01 1.13136590e+00 -3.75976473e-01
-7.87547886e-01 3.26354951e-01 -1.12699203e-01 1.05439210e+00
-4.24805023e-02 1.94996044e-01 -1.21236837e+00 4.73181039e-01
3.15320417e-02 8.62417340e-01 -1.69750333e+00 1.73645175e+00
-1.86669171e-01 1.44038796e+00 -7.87794441e-02 -2.14302843e-03
3.71567816e-01 3.33291173e-01 6.06773734e-01 2.00481343e+00
-3.30133170e-01 -3.27475429e-01 -4.02947187e-01 1.25213072e-01
3.13424468e-01 3.11216235e-01 -2.43393257e-02 -8.23925957e-02
-2.16868758e-01 2.02851152e+00 -1.01125205e+00 -2.09640652e-01
-2.19492242e-01 1.08559453e+00 -3.09443116e-01 5.44154108e-01
-7.19734073e-01 -3.49704415e-01 2.11936310e-01 8.04802403e-02
-4.76601422e-01 4.12157983e-01 -1.01786947e+00 -6.75024092e-01
-5.69284916e-01 -5.84519327e-01 9.11786258e-01 -3.26848477e-01
-1.60118759e+00 1.49428874e-01 -4.77908880e-01 -6.14120543e-01
3.50799799e-01 -1.44693112e+00 -1.41342247e+00 7.94335127e-01
-1.87523067e+00 -2.04645944e+00 -4.21025693e-01 5.22954702e-01
5.88746488e-01 -5.80432475e-01 1.08555698e+00 -2.93661773e-01
-1.14611149e+00 1.34005523e+00 -6.94634840e-02 1.13818303e-01
8.25275421e-01 -1.87168705e+00 7.80968785e-01 7.39078522e-01
-4.30760056e-01 -5.09282127e-02 -2.47743100e-01 -7.69781590e-01
-1.20665109e+00 -1.25594103e+00 8.56535658e-02 -6.32933915e-01
1.11242115e+00 8.20508003e-02 -1.41736358e-01 1.92184120e-01
5.10337114e-01 -1.16849877e-01 1.15067577e+00 5.62885664e-02
-2.27827743e-01 2.83377886e-01 -1.53529787e+00 8.53286088e-01
6.94844186e-01 -4.38368589e-01 1.69554144e-01 1.04933369e+00
-6.29259571e-02 -5.79790831e-01 -1.05179143e+00 6.14078164e-01
5.35408616e-01 -1.72738597e-01 1.29103172e+00 -1.02548003e+00
4.00279731e-01 8.63869265e-02 3.59500319e-01 -1.03748155e+00
-1.04402311e-01 -7.01468527e-01 -9.90042165e-02 4.52199429e-01
7.73381829e-01 -2.92524934e-01 1.31163526e+00 3.20709229e-01
-2.23142684e-01 -1.01440787e+00 -1.05136907e+00 -2.92209595e-01
7.00644970e-01 -4.10223275e-01 -3.86862725e-01 5.48948884e-01
-4.98675462e-03 -3.25343698e-01 5.82780361e-01 1.45568982e-01
9.88433242e-01 -6.14973247e-01 2.39385039e-01 -5.96409440e-01
4.29048479e-01 -1.18115497e+00 -4.36021864e-01 -1.27293944e-01
-5.98739535e-02 -9.83972013e-01 -2.46739686e-01 -1.74916101e+00
3.29575598e-01 1.22826733e-01 -2.82042503e-01 5.71238339e-01
-7.26100147e-01 9.90488529e-01 -3.54348011e-02 -3.58150750e-01
-8.97705615e-01 -8.70396197e-01 1.33076870e+00 -5.58795691e-01
3.02127451e-01 4.06690001e-01 -4.59656388e-01 6.99806035e-01
8.02477658e-01 1.92015827e-01 -4.10007517e-04 -8.80816802e-02
-2.51822412e-01 -3.33591670e-01 9.34368372e-01 -9.60047483e-01
3.50239009e-01 -2.44850263e-01 8.18015397e-01 -7.22173572e-01
1.41353577e-01 -2.27862358e-01 -6.77450716e-01 9.02432561e-01
-3.59543234e-01 -9.22952831e-01 4.95676041e-01 4.81503993e-01
1.50127426e-01 -5.65703273e-01 1.24902034e+00 -1.05006965e-02
-1.01063025e+00 2.39776284e-01 -9.50942218e-01 -8.78978670e-02
1.64157283e+00 -3.87060732e-01 -9.59814548e-01 4.95974898e-01
-1.34075296e+00 6.23860180e-01 2.04841807e-01 6.12874106e-02
6.08809710e-01 -8.78112376e-01 -1.35601807e+00 -4.60869111e-02
5.02269208e-01 -5.65584779e-01 7.62550771e-01 1.48057473e+00
-1.05863440e+00 8.65662932e-01 -9.73477587e-02 -5.78761220e-01
-1.88305032e+00 -1.75698012e-01 1.15248239e+00 -5.80249369e-01
-1.77947387e-01 1.57014596e+00 -1.61386609e-01 -4.05010432e-01
2.33299151e-01 7.81893730e-02 -4.98926282e-01 -7.11262152e-02
7.27651775e-01 6.67919517e-01 2.15462506e-01 -2.77656257e-01
-4.45937157e-01 6.08883560e-01 -5.70916295e-01 1.35838166e-01
3.12115014e-01 2.94185936e-01 -3.13644677e-01 -1.00678824e-01
8.38456452e-01 -4.21821535e-01 -8.60450327e-01 4.72952485e-01
1.02149226e-01 -3.38573419e-02 3.06123167e-01 -1.97903907e+00
-7.28247702e-01 8.86989176e-01 1.34370971e+00 1.42633289e-01
1.04611468e+00 -2.67453581e-01 8.86467218e-01 6.06327318e-02
-2.57110834e-01 -1.53112662e+00 -4.53627884e-01 1.51688665e-01
6.59452677e-01 -1.66180086e+00 9.07850564e-02 -1.17690802e+00
-1.03663385e+00 1.46208906e+00 8.17749858e-01 -1.36327296e-01
5.77296615e-01 8.05716217e-02 5.78587949e-01 -2.36428469e-01
-5.33769608e-01 -8.02966714e-01 1.16306520e+00 1.50960112e+00
4.16921943e-01 8.27987313e-01 -4.03988957e-01 2.47531950e-01
5.13773620e-01 -5.13424575e-01 3.97119403e-01 3.72763067e-01
-2.51263231e-01 -1.09402382e+00 6.83185086e-02 9.89512861e-01
-5.69532990e-01 -1.41602054e-01 -1.59649003e+00 1.04319036e+00
3.79356921e-01 6.08071685e-01 -2.32270941e-01 -1.05336919e-01
1.02699891e-01 9.17591900e-02 8.89995635e-01 -5.68435431e-01
-9.79486644e-01 1.30336151e-01 5.63770175e-01 -5.14893234e-01
-4.59303021e-01 -8.29744935e-01 -9.60876405e-01 2.86761254e-01
3.93338036e-03 -2.35039696e-01 1.31411719e+00 5.48503935e-01
-1.03745639e-01 6.14843428e-01 2.54887938e-01 -1.46925986e-01
-2.94492513e-01 -1.02933514e+00 -6.33145452e-01 -5.49535546e-03
1.78439334e-01 1.03439555e-01 -1.64559349e-01 4.49803174e-01] | [15.802346229553223, -3.0565314292907715] |
80bf8305-ebc8-4525-9204-d8f4f2bb924d | eegeyenet-a-simultaneous | 2111.05100 | null | https://arxiv.org/abs/2111.05100v2 | https://arxiv.org/pdf/2111.05100v2.pdf | EEGEyeNet: a Simultaneous Electroencephalography and Eye-tracking Dataset and Benchmark for Eye Movement Prediction | We present a new dataset and benchmark with the goal of advancing research in the intersection of brain activities and eye movements. Our dataset, EEGEyeNet, consists of simultaneous Electroencephalography (EEG) and Eye-tracking (ET) recordings from 356 different subjects collected from three different experimental paradigms. Using this dataset, we also propose a benchmark to evaluate gaze prediction from EEG measurements. The benchmark consists of three tasks with an increasing level of difficulty: left-right, angle-amplitude and absolute position. We run extensive experiments on this benchmark in order to provide solid baselines, both based on classical machine learning models and on large neural networks. We release our complete code and data and provide a simple and easy-to-use interface to evaluate new methods. | ['Martyna Beata Płomecka', 'Nicolas Langer', 'Roger Wattenhofer', 'Victor Gillioz', 'Lukas Wolf', 'Damián Pascual', 'Ard Kastrati'] | 2021-11-06 | null | null | null | null | ['eye-tracking'] | ['computer-vision'] | [ 7.38711283e-02 -4.15003270e-01 1.33820429e-01 -6.08403444e-01
-1.37642965e-01 -2.85837322e-01 4.05945718e-01 -2.26284504e-01
-6.44208729e-01 9.48589623e-01 -1.45873457e-01 -9.85670462e-02
-7.68498629e-02 1.34579331e-01 -5.88575184e-01 -5.99263966e-01
-3.90556425e-01 -2.81883657e-01 2.04842299e-01 -5.47089847e-03
5.82084179e-01 2.51601845e-01 -1.95296562e+00 1.27089933e-01
7.67280161e-01 1.29861939e+00 1.30993545e-01 3.70774657e-01
7.39857078e-01 4.32134271e-01 -8.52522612e-01 -5.37952371e-02
1.57185063e-01 -4.02908891e-01 -5.11594951e-01 -3.39564234e-01
5.05103886e-01 1.02632821e-01 -2.59809107e-01 1.02594793e+00
8.56320202e-01 -2.86116414e-02 4.59153920e-01 -1.80197084e+00
-6.03797615e-01 8.51223841e-02 -7.07666218e-01 9.08943772e-01
7.69318938e-01 4.73381549e-01 6.80792511e-01 -6.90163791e-01
4.32213724e-01 5.13425350e-01 3.71655673e-01 6.63568139e-01
-1.17839813e+00 -1.11221051e+00 3.76134440e-02 8.67329240e-01
-1.51537943e+00 -7.27601051e-01 6.60501182e-01 -3.45739067e-01
1.14412975e+00 2.61408806e-01 9.45107818e-01 1.75938940e+00
4.62501913e-01 5.64147055e-01 1.68936956e+00 -2.13987857e-01
2.18982652e-01 2.21499994e-01 6.31288707e-01 1.87096849e-01
2.69209355e-01 1.65187955e-01 -1.27624857e+00 9.14095566e-02
5.62331319e-01 -2.43361548e-01 -9.88486648e-01 -2.66705364e-01
-1.39293051e+00 1.44042790e-01 3.84983718e-01 1.96320251e-01
-4.81679857e-01 -1.76653057e-01 -1.10161081e-01 5.39357901e-01
2.67865330e-01 5.53606808e-01 -5.59857309e-01 -6.36364400e-01
-8.59487653e-01 2.47136697e-01 9.33079779e-01 1.05926311e+00
4.77116197e-01 -2.45424122e-01 -3.17338765e-01 5.60346723e-01
2.02346697e-01 3.45970750e-01 6.29581213e-01 -6.40141726e-01
4.60155398e-01 3.11915755e-01 1.89174071e-01 -8.43438625e-01
-8.76570046e-01 -2.66429096e-01 -4.09764856e-01 2.96919882e-01
4.85851645e-01 -3.82833421e-01 -6.05024278e-01 1.62460101e+00
-1.19486280e-01 4.50367779e-01 -4.83933508e-01 1.04887950e+00
8.92690182e-01 1.35681584e-01 -6.22861013e-02 -3.63292038e-01
1.69026160e+00 -8.04224789e-01 -8.47599030e-01 -3.33417445e-01
3.81961346e-01 -2.43216515e-01 1.29284763e+00 8.59595478e-01
-1.15477133e+00 -2.86942601e-01 -1.13015366e+00 5.65978400e-02
-5.73199451e-01 7.07846880e-02 4.00496274e-01 6.99012101e-01
-1.22760248e+00 4.62153882e-01 -8.45120490e-01 -4.19214249e-01
4.15318996e-01 7.14147389e-01 -4.71788466e-01 6.73660755e-01
-8.62607062e-01 1.11715353e+00 9.54348445e-02 -9.44455788e-02
-4.27275866e-01 -8.90898883e-01 -6.32863522e-01 -5.46182320e-02
2.16295645e-01 -3.42738599e-01 1.09594727e+00 -7.78839052e-01
-1.68589056e+00 9.90734994e-01 -4.83641267e-01 -2.10596532e-01
2.58553594e-01 -2.53923774e-01 -6.92237437e-01 3.71137634e-02
-4.75092024e-01 7.48865843e-01 7.46717572e-01 -6.96066082e-01
-5.30793548e-01 -7.11550295e-01 -1.64227366e-01 -3.84047925e-02
-4.97658491e-01 5.71089506e-01 -3.06122154e-01 -4.08299059e-01
-4.86617655e-01 -9.50772345e-01 6.66892409e-01 -2.47526601e-01
-4.65047896e-01 -2.49887571e-01 3.61333758e-01 -5.19488871e-01
1.44166124e+00 -1.96402740e+00 2.61104017e-01 9.79551300e-02
6.99282169e-01 -3.07426136e-02 -9.21905935e-02 -1.88911840e-01
-6.07563496e-01 -1.26844451e-01 4.65171337e-02 -4.48722035e-01
1.33124068e-01 -5.65821767e-01 -3.42142098e-02 6.65370464e-01
2.44615413e-03 9.71033216e-01 -6.40980184e-01 -9.81264040e-02
-3.03515382e-02 4.00104433e-01 -4.02765095e-01 1.09332837e-01
2.71902233e-01 7.25625396e-01 1.63992509e-01 7.36908913e-01
4.81787533e-01 -4.41526413e-01 -4.02239621e-01 -1.58967242e-01
-2.25185245e-01 3.91435564e-01 -9.04735148e-01 1.66889393e+00
-5.67324534e-02 1.39321160e+00 -4.03240234e-01 -3.93024683e-01
6.25527859e-01 1.89930096e-01 3.27804565e-01 -1.02691400e+00
6.96774244e-01 -2.16885924e-01 3.71484369e-01 -6.25459135e-01
2.30238870e-01 6.36986732e-01 3.85489255e-01 7.05608189e-01
2.92059034e-01 3.04801434e-01 3.88306111e-01 -3.03252846e-01
1.18339312e+00 9.69561934e-02 4.92014647e-01 -5.40385604e-01
4.07373637e-01 -6.12583935e-01 8.53281245e-02 3.16806614e-01
-5.67023396e-01 3.39825094e-01 7.92981327e-01 -3.07950109e-01
-5.32353878e-01 -8.06940675e-01 -4.79383469e-01 1.16219938e+00
3.00021231e-01 -7.92760611e-01 -1.13664472e+00 -2.96507597e-01
-2.65481949e-01 8.42393994e-01 -1.03093946e+00 -1.29983574e-01
5.34080453e-02 -9.93974805e-01 2.06549749e-01 5.48543274e-01
3.11380357e-01 -1.31723368e+00 -1.01888800e+00 -5.90120196e-01
-1.11197911e-01 -1.30365789e+00 -4.76232231e-01 2.82134712e-01
-4.28914815e-01 -1.32565725e+00 -5.67127705e-01 -4.36698318e-01
5.27717829e-01 1.73845246e-01 1.06882358e+00 3.38900462e-02
-2.20013514e-01 2.73382097e-01 -1.64811328e-01 -1.04524493e+00
5.45844734e-01 1.11723863e-01 5.39503992e-01 1.04723379e-01
1.19432986e+00 -7.45968997e-01 -8.63758206e-01 7.12789059e-01
-3.64799738e-01 4.12525833e-02 3.30930144e-01 3.51997435e-01
2.87862152e-01 -6.39422894e-01 2.69111842e-01 -1.97203442e-01
1.03908741e+00 -4.84203547e-01 -9.04180825e-01 3.41612816e-01
-7.16505766e-01 -1.16579860e-01 2.31354520e-01 -5.66936910e-01
-5.52303612e-01 -4.00789469e-01 1.49612665e-01 -3.43413562e-01
-5.54971516e-01 1.53911039e-01 8.88521448e-02 -5.76128602e-01
8.93252194e-01 1.24359831e-01 -3.33880812e-01 -1.97244242e-01
-1.20233312e-01 8.86374533e-01 6.34666324e-01 -5.57648242e-02
1.86637059e-01 9.05186459e-02 -1.05909571e-01 -7.84434199e-01
-4.71713960e-01 -2.04645857e-01 -9.09691989e-01 -2.84965217e-01
6.80826485e-01 -7.90797412e-01 -1.32052410e+00 7.26889729e-01
-1.13975298e+00 -4.85115141e-01 4.01148140e-01 8.17329347e-01
-5.30864596e-01 -1.16377942e-01 -1.83203846e-01 -5.76248109e-01
-4.18083996e-01 -1.21252918e+00 8.72325301e-01 4.89488095e-01
-5.30754447e-01 -7.35476315e-01 2.20533013e-01 -5.48505224e-02
2.96524256e-01 1.19064309e-01 2.48242021e-01 -7.52266049e-01
-3.81414056e-01 -4.85876836e-02 -2.79795051e-01 3.34715247e-02
1.86416153e-02 5.43752834e-02 -1.27968538e+00 -3.91966343e-01
1.59022480e-01 -3.50745201e-01 4.75488365e-01 4.30747926e-01
1.51056373e+00 3.56864125e-01 -4.85029817e-01 1.07950604e+00
1.07150209e+00 1.78296417e-01 8.94698381e-01 5.37232161e-01
3.82434160e-01 4.37204510e-01 -5.72255664e-02 4.98831511e-01
5.85552096e-01 7.59531498e-01 2.27698520e-01 2.55921036e-01
2.58538514e-01 4.27130789e-01 2.85216838e-01 4.18991566e-01
-6.78773403e-01 -3.59917849e-01 -1.01188254e+00 1.20258294e-01
-1.43705368e+00 -7.80280054e-01 -1.81476846e-01 2.46411848e+00
6.64302409e-01 9.68400538e-02 4.95673537e-01 1.58583254e-01
5.13394415e-01 -1.80164620e-01 -6.14877760e-01 -7.01554120e-02
1.64762318e-01 4.86008286e-01 1.89101115e-01 -1.15654521e-01
-8.08551967e-01 5.09231031e-01 7.56274509e+00 2.99419373e-01
-1.51400995e+00 3.76308076e-02 2.10426614e-01 -8.03571701e-01
3.34227264e-01 -6.54495716e-01 -9.93037283e-01 9.94059920e-01
1.47224808e+00 -3.03261131e-01 9.75721240e-01 2.96426117e-01
1.98459700e-01 -5.56995094e-01 -1.56057680e+00 1.74130070e+00
5.05662918e-01 -8.63779545e-01 -7.18964338e-01 1.36175901e-01
3.39733362e-01 5.97859919e-01 1.63406804e-01 1.19819112e-01
-3.26649278e-01 -1.07907581e+00 7.39979565e-01 7.97889411e-01
9.04136837e-01 -3.81858110e-01 2.97129750e-01 1.82994083e-01
-8.15735340e-01 -3.40367734e-01 1.15069887e-02 -2.08027005e-01
-1.12985171e-01 -3.69727314e-02 -1.96229830e-01 1.97310925e-01
1.32318020e+00 9.85926151e-01 -1.08758223e+00 1.53090107e+00
-5.01855373e-01 3.15215081e-01 -2.86069483e-01 -4.49692965e-01
-3.39062452e-01 -7.20652789e-02 4.86764312e-01 8.58438671e-01
3.07377011e-01 1.77830204e-01 -7.22204566e-01 1.22083259e+00
-1.68198615e-01 -1.04123667e-01 -3.86573076e-01 1.82757258e-01
5.17024457e-01 1.45906389e+00 -5.51734090e-01 7.52753839e-02
-6.32850707e-01 7.54022598e-01 4.19184238e-01 6.38936579e-01
-1.20521724e+00 -8.45088840e-01 8.11685860e-01 -1.24945387e-01
-3.20650712e-02 -2.37040371e-01 -3.20256591e-01 -1.24837053e+00
2.35946700e-01 -6.97550833e-01 -6.39972091e-02 -1.41901922e+00
-1.01139796e+00 9.72936332e-01 3.03308427e-01 -1.16651309e+00
-3.67066771e-01 -9.64212120e-01 -5.81353128e-01 1.01891291e+00
-1.67644584e+00 -4.51383650e-01 -1.03103518e+00 1.01245701e+00
1.32745370e-01 -1.80351034e-01 6.34306788e-01 3.20241362e-01
-1.01040721e+00 8.15929413e-01 -3.07262719e-01 -1.65166765e-01
1.13439977e+00 -1.16800809e+00 4.97608691e-01 6.03361905e-01
2.43166968e-01 8.02965879e-01 6.84656918e-01 -6.15845919e-02
-1.22657168e+00 -5.14017582e-01 6.12311721e-01 -1.11807787e+00
8.18825006e-01 -8.67285907e-01 -8.14677715e-01 8.88571203e-01
6.92838848e-01 5.07599041e-02 9.01849031e-01 2.01717868e-01
9.68014598e-02 -9.64949355e-02 -8.18368435e-01 4.98443246e-01
1.33553112e+00 -5.09916127e-01 -8.46096814e-01 5.51561192e-02
2.01826647e-01 -6.22795939e-01 -8.15617323e-01 2.72284806e-01
7.43193150e-01 -1.28304005e+00 5.74581385e-01 -4.76017386e-01
5.66408709e-02 3.29676736e-03 3.68630648e-01 -1.79213297e+00
-2.61572778e-01 -8.71969342e-01 -4.68243957e-01 5.84746242e-01
3.02234143e-01 -9.95940566e-01 4.33881670e-01 7.15731084e-01
1.43132463e-01 -9.24784184e-01 -6.72120571e-01 -6.75701320e-01
-2.35118672e-01 -7.02498078e-01 8.43548715e-01 4.74841177e-01
5.60995519e-01 4.63995785e-01 -9.23952162e-02 -1.80166326e-02
4.79284137e-01 -2.30442345e-01 7.90658474e-01 -1.57752132e+00
2.90663630e-01 -5.87344646e-01 -6.82088315e-01 -9.98152554e-01
1.98469102e-01 -4.73316282e-01 -9.74375904e-02 -9.11324859e-01
2.15181708e-01 3.95005614e-01 -6.67760789e-01 6.38356984e-01
-1.13556109e-01 5.21729112e-01 1.74804113e-03 9.51333568e-02
-8.37692678e-01 3.58615547e-01 9.87915754e-01 1.71701625e-01
-3.36645782e-01 -1.75833348e-02 -6.79761887e-01 6.61923468e-01
6.81050718e-01 -3.44360024e-01 -4.40691203e-01 -3.90853643e-01
3.00119251e-01 -4.32444423e-01 5.99200070e-01 -1.38534403e+00
7.25682616e-01 2.65292287e-01 6.83032453e-01 -4.13835287e-01
4.03463274e-01 -6.72254264e-01 -3.38759422e-01 -1.91432297e-01
-2.20674157e-01 4.11975116e-01 3.82791460e-01 3.40068668e-01
1.06664307e-01 2.34145284e-01 6.39056325e-01 4.34391916e-01
-8.79747987e-01 3.89271826e-01 -5.95933124e-02 2.01116234e-01
1.49541891e+00 -5.97069919e-01 -6.94730043e-01 -2.76504964e-01
-7.99086869e-01 4.02280092e-01 5.12754261e-01 6.69390023e-01
4.19113398e-01 -1.03789520e+00 -4.33912188e-01 8.28829229e-01
5.47611356e-01 -8.24752331e-01 -8.82470757e-02 1.64799583e+00
-8.96785501e-03 7.06614971e-01 -9.25743580e-01 -8.21292758e-01
-1.44561350e+00 5.75736523e-01 4.92671400e-01 5.85340977e-01
-3.37912053e-01 8.66898477e-01 2.33639777e-02 2.58587360e-01
6.29228771e-01 -7.45170236e-01 -5.92078447e-01 2.38431133e-02
1.11397576e+00 4.32540536e-01 5.29759467e-01 -4.63690102e-01
-4.48417574e-01 3.24247152e-01 2.70561785e-01 9.92315561e-02
1.31110346e+00 -2.58363515e-01 -1.15118600e-01 6.33867741e-01
8.96260917e-01 -2.97756940e-01 -1.28095901e+00 9.11172330e-02
-9.40485522e-02 -5.26087821e-01 2.01963380e-01 -9.57955837e-01
-1.09303820e+00 9.44589198e-01 9.24108684e-01 2.89516747e-01
1.23268473e+00 -1.50320560e-01 3.81745428e-01 5.61743200e-01
6.91111147e-01 -1.00823128e+00 -1.18277185e-01 3.12853009e-01
9.14466262e-01 -1.30846632e+00 -4.37193960e-02 -3.63189168e-02
-6.30262852e-01 9.35221851e-01 9.40246105e-01 3.93681973e-02
9.69602108e-01 3.74127984e-01 -8.90422985e-02 -4.33558613e-01
-1.02878046e+00 -9.20782387e-02 8.15095723e-01 7.39220500e-01
5.16450763e-01 -2.47428119e-01 -1.45221308e-01 9.33800519e-01
-5.48814833e-01 6.12525761e-01 3.65781307e-01 6.85710430e-01
2.38634422e-02 -5.71888804e-01 -3.19888681e-01 7.29328334e-01
-3.61463755e-01 -1.65835753e-01 -4.97561604e-01 9.38121378e-01
-2.94325948e-02 9.78168905e-01 3.32427740e-01 -7.31284142e-01
5.39979041e-01 1.57297567e-01 9.88003194e-01 -3.69008809e-01
-3.28131229e-01 -3.63380045e-01 -2.78347284e-01 -9.85420227e-01
-4.71589744e-01 -7.20235348e-01 -8.03299844e-01 -4.13168430e-01
-2.65212804e-01 -2.46258765e-01 6.71808422e-01 9.80460107e-01
6.84119165e-01 6.69613183e-01 2.28631064e-01 -1.22284448e+00
-1.50293976e-01 -1.38023889e+00 -9.91338909e-01 2.99718350e-01
4.55285132e-01 -1.25181317e+00 -4.80016410e-01 3.20826955e-02] | [14.113465309143066, 0.17096258699893951] |
67a6952f-ad03-4f96-874b-18fdc63d165b | on-conditioning-the-input-noise-for | 2205.03859 | null | https://arxiv.org/abs/2205.03859v1 | https://arxiv.org/pdf/2205.03859v1.pdf | On Conditioning the Input Noise for Controlled Image Generation with Diffusion Models | Conditional image generation has paved the way for several breakthroughs in image editing, generating stock photos and 3-D object generation. This continues to be a significant area of interest with the rise of new state-of-the-art methods that are based on diffusion models. However, diffusion models provide very little control over the generated image, which led to subsequent works exploring techniques like classifier guidance, that provides a way to trade off diversity with fidelity. In this work, we explore techniques to condition diffusion models with carefully crafted input noise artifacts. This allows generation of images conditioned on semantic attributes. This is different from existing approaches that input Gaussian noise and further introduce conditioning at the diffusion model's inference step. Our experiments over several examples and conditional settings show the potential of our approach. | ['Vineeth N. Balasubramanian', 'Balaji Krishnamurthy', 'Siddharth Ramesh', 'Ayush Chopra', 'Surgan Jandial', 'Vedant Singh'] | 2022-05-08 | null | null | null | null | ['conditional-image-generation'] | ['computer-vision'] | [ 5.94259560e-01 3.82974207e-01 3.12209390e-02 -3.51554483e-01
-5.58095336e-01 -4.69354063e-01 1.18536532e+00 1.37620391e-02
-3.42413872e-01 6.32030964e-01 3.10058743e-01 -3.17688167e-01
-7.72199258e-02 -1.01456106e+00 -4.61250007e-01 -8.46143365e-01
1.20250858e-01 4.89626229e-01 3.55051458e-01 -1.31573245e-01
5.47093511e-01 4.89086717e-01 -1.58345461e+00 3.27018350e-01
9.07873154e-01 7.27800190e-01 2.61619866e-01 8.60716462e-01
-3.72479618e-01 8.79122138e-01 -6.59348190e-01 -4.97635931e-01
2.82028854e-01 -7.29178846e-01 -5.61625838e-01 4.65719789e-01
3.65734965e-01 -1.24744073e-01 -1.06539682e-01 1.06931639e+00
4.98678386e-01 7.00254515e-02 8.91513944e-01 -1.20285439e+00
-1.12280333e+00 5.24301648e-01 -6.35162830e-01 7.59027526e-02
2.09552720e-01 3.00130934e-01 7.07343459e-01 -5.55418849e-01
9.87939060e-01 1.43314075e+00 4.76629645e-01 7.44745910e-01
-1.63200116e+00 -3.82665783e-01 1.87123984e-01 5.34712896e-02
-1.13463616e+00 -2.95701742e-01 6.30723774e-01 -5.55211306e-01
8.41265857e-01 1.43574074e-01 6.30121708e-01 1.21277690e+00
1.33829087e-01 9.16928113e-01 1.58370042e+00 -7.84887135e-01
3.84672314e-01 4.81296003e-01 -3.51085901e-01 5.73645175e-01
1.45659447e-01 3.49645674e-01 -3.34597588e-01 -8.18386599e-02
9.97877121e-01 -3.37849200e-01 -1.78816110e-01 -5.96972585e-01
-1.12610173e+00 1.00156069e+00 2.29142934e-01 1.61978334e-01
-3.23284000e-01 2.74955153e-01 -5.68753458e-04 1.72945097e-01
7.58640349e-01 5.14521420e-01 3.91206704e-02 -1.24316014e-01
-1.22167945e+00 5.43212771e-01 7.45433331e-01 9.37485337e-01
7.54148126e-01 1.06506661e-01 -4.17464256e-01 8.38367581e-01
3.24843407e-01 2.66033351e-01 1.39254004e-01 -1.34384406e+00
1.15071796e-01 5.02150893e-01 7.32745528e-02 -7.24247754e-01
1.42509669e-01 -2.18860552e-01 -5.78666329e-01 8.87011051e-01
4.86959368e-01 -1.24121450e-01 -1.55563712e+00 1.60825789e+00
2.60839075e-01 5.66567704e-02 1.64408702e-02 6.36753380e-01
-9.38014500e-03 6.34707808e-01 5.56357428e-02 1.05512001e-01
8.04435909e-01 -8.53169799e-01 -6.41738057e-01 -2.07400322e-02
3.63539755e-01 -9.36911464e-01 9.81278419e-01 5.28377533e-01
-1.16670966e+00 -2.85860151e-01 -9.56813216e-01 7.07019120e-02
-4.29826796e-01 -2.06329972e-01 8.25345397e-01 1.06233525e+00
-1.16685414e+00 8.06638360e-01 -9.49730098e-01 -4.41883475e-01
8.63170624e-01 2.29212433e-01 -8.71767178e-02 -3.41002017e-01
-9.11514163e-01 1.03541160e+00 5.81922308e-02 -1.87730566e-01
-9.56943631e-01 -5.50217032e-01 -6.01593137e-01 -2.69034922e-01
4.74111706e-01 -1.05217612e+00 1.17414927e+00 -1.25784707e+00
-1.66829646e+00 7.42395580e-01 6.85736686e-02 -5.53305328e-01
9.92464721e-01 -4.94797230e-02 -5.76820150e-02 -7.28763193e-02
-5.21902665e-02 1.13477123e+00 1.09807754e+00 -1.54786026e+00
-5.77221036e-01 -1.44352466e-01 2.14626163e-01 1.64735958e-01
-2.76974887e-01 -9.36557502e-02 -4.45017427e-01 -7.95291662e-01
-2.24719003e-01 -1.07144094e+00 -6.93046868e-01 4.28602606e-01
-4.43081945e-01 -8.93876702e-03 7.28646874e-01 -1.69515312e-01
9.52592194e-01 -2.02018809e+00 1.57265499e-01 3.85236830e-01
3.54087539e-02 1.91728041e-01 -1.04229130e-01 5.64136922e-01
3.02491277e-01 4.93496895e-01 -6.47754371e-01 -6.04741096e-01
-3.15640878e-04 3.97239506e-01 -2.10492522e-01 2.23495916e-01
5.02225280e-01 6.27706409e-01 -8.40218365e-01 -3.70782793e-01
3.04214180e-01 8.56726825e-01 -8.00050378e-01 8.51720273e-02
-4.91580486e-01 5.55551827e-01 -3.11056644e-01 4.22661394e-01
5.63726068e-01 -1.21394776e-01 -1.65795222e-01 1.83865651e-01
-1.81311339e-01 -2.55760439e-02 -1.32012057e+00 1.73600030e+00
-2.77910829e-01 6.52331233e-01 -7.24627227e-02 -6.80513978e-01
7.89722085e-01 9.97947231e-02 2.43409947e-01 -3.71715218e-01
1.30808756e-01 1.00106791e-01 4.77672108e-02 -3.48663181e-01
5.04706502e-01 -4.09532309e-01 4.17037487e-01 4.45555151e-01
3.44950259e-02 -5.98303974e-01 3.31531107e-01 3.78103584e-01
9.40691352e-01 5.48754692e-01 -1.10361755e-01 -2.64156818e-01
1.09459661e-01 1.91729754e-01 2.54844725e-01 9.53615963e-01
1.53356731e-01 1.05532575e+00 3.55333656e-01 -1.87328368e-01
-1.31131530e+00 -9.23890769e-01 -8.25150609e-02 4.31934059e-01
-1.22100357e-02 -2.77530968e-01 -1.12537229e+00 -5.89203715e-01
-5.09400852e-02 1.05644989e+00 -9.18624401e-01 -3.56975570e-02
-2.31185228e-01 -8.20177853e-01 4.47794378e-01 5.68265319e-01
3.08871448e-01 -1.03311789e+00 -6.08053148e-01 2.66156107e-01
2.13676125e-01 -8.84205341e-01 -2.14322761e-01 2.48602152e-01
-9.64980483e-01 -7.08553910e-01 -1.15953422e+00 -3.73537123e-01
9.12914276e-01 -5.28378896e-02 1.01965988e+00 -1.37175051e-02
-4.81028080e-01 5.57977200e-01 -3.29535604e-01 -6.36869133e-01
-7.68883824e-01 9.10450518e-03 -1.97632134e-01 1.36593670e-01
1.24284543e-01 -6.64112568e-01 -6.18765831e-01 1.70447141e-01
-1.28561842e+00 3.26879501e-01 6.61224544e-01 6.91263199e-01
6.52421176e-01 9.21622962e-02 2.20258623e-01 -1.20095754e+00
8.63344073e-01 -3.47693503e-01 -5.84065437e-01 1.04993150e-01
-8.98407936e-01 3.74539256e-01 1.96739346e-01 -6.58449769e-01
-1.40414286e+00 1.75917223e-01 -5.70433661e-02 -4.85722244e-01
-2.85900503e-01 3.33340555e-01 -1.65767968e-03 8.88409987e-02
9.45782959e-01 -3.91009524e-02 2.33317301e-01 -2.92680591e-01
8.52545381e-01 4.70693111e-01 3.52818131e-01 -6.66759312e-01
5.90102673e-01 7.68057048e-01 4.14147135e-03 -8.04775357e-01
-4.10810560e-01 5.64312041e-02 -5.55613518e-01 -1.17301509e-01
9.03273940e-01 -5.77633202e-01 1.76777299e-02 4.44714934e-01
-9.83464241e-01 -7.04914093e-01 -6.07157350e-01 2.99217701e-01
-7.69377768e-01 1.40254378e-01 -4.11658287e-01 -1.14335310e+00
2.35744908e-01 -1.29323614e+00 9.13878441e-01 1.03840016e-01
-4.09102052e-01 -1.03345156e+00 2.18191184e-02 1.29906520e-01
6.59899175e-01 1.39965996e-01 7.87275851e-01 -1.41576558e-01
-1.00007439e+00 -1.53324202e-01 -2.14618415e-01 6.04097247e-01
3.03492360e-02 1.74905747e-01 -1.06765258e+00 8.03928301e-02
-1.82629049e-01 -1.90637037e-01 1.07844543e+00 4.27282393e-01
9.50983822e-01 -1.31374463e-01 -4.88843381e-01 5.15990078e-01
1.34359491e+00 1.34156227e-01 1.05539632e+00 2.65095294e-01
5.69950700e-01 6.90713763e-01 3.46833646e-01 3.19145054e-01
1.60976931e-01 4.93260086e-01 2.57677227e-01 -2.43025690e-01
-5.84040105e-01 -4.39834893e-01 1.00074664e-01 1.55166879e-01
-1.36289269e-01 -5.22274852e-01 -8.20112586e-01 6.90093577e-01
-1.76661777e+00 -1.16067088e+00 -6.76185265e-02 2.07533455e+00
8.03382218e-01 1.93407297e-01 -6.32223859e-02 2.27754921e-01
6.04941666e-01 5.56025794e-03 -4.54163790e-01 -2.84951538e-01
-1.84262231e-01 2.65577644e-01 5.90596080e-01 7.72683918e-01
-8.18246663e-01 9.87716198e-01 6.87198019e+00 8.74615669e-01
-9.05794919e-01 -1.93789572e-01 8.81502032e-01 8.96505341e-02
-7.40332127e-01 3.48871946e-01 -7.74101615e-01 3.78022611e-01
5.21597624e-01 6.34227097e-02 4.55850989e-01 7.08174050e-01
2.40439177e-01 -5.71989238e-01 -9.40253258e-01 6.36193514e-01
2.32580706e-01 -1.42838287e+00 2.34275252e-01 2.83182681e-01
1.06348920e+00 -2.34232664e-01 2.88111567e-01 -1.78925186e-01
8.46879363e-01 -1.03534460e+00 7.50617981e-01 6.72620714e-01
6.10368907e-01 -5.30601799e-01 2.39600748e-01 3.07901949e-01
-4.88131613e-01 4.21716720e-02 -2.38650754e-01 2.58472748e-02
4.58869278e-01 8.02788556e-01 -8.59339356e-01 1.98023751e-01
6.11434460e-01 3.81315529e-01 -4.36855018e-01 1.00801897e+00
-3.75615090e-01 6.24280572e-01 -3.94769371e-01 3.24485451e-03
2.01979548e-01 -2.01426744e-01 5.03422976e-01 1.26861012e+00
5.31555355e-01 -1.83921643e-02 2.90757120e-02 1.20490766e+00
1.64730519e-01 7.60573137e-04 -8.28401387e-01 -1.38370141e-01
1.90897182e-01 1.02478719e+00 -1.16606462e+00 -4.99552250e-01
-2.55744010e-01 1.39120138e+00 1.42248288e-01 3.84021670e-01
-6.42559648e-01 -1.52646616e-01 4.13376272e-01 4.60450500e-01
5.25887370e-01 -4.35920775e-01 -4.55785751e-01 -9.26651359e-01
-2.85751969e-01 -8.64984155e-01 1.46302313e-01 -8.07573676e-01
-1.40984750e+00 5.12455761e-01 1.20920621e-01 -8.10277820e-01
-2.54324794e-01 -4.06755626e-01 -4.18492496e-01 9.20636952e-01
-1.44617021e+00 -1.11017573e+00 -1.22085139e-01 4.50900316e-01
6.23217523e-01 7.00731501e-02 7.59921134e-01 4.94646020e-02
4.99744480e-03 2.29213804e-01 -1.10076986e-01 -2.58376271e-01
7.94948518e-01 -1.43908846e+00 5.91439247e-01 8.71754348e-01
4.07924473e-01 7.23179400e-01 8.39500606e-01 -8.23807120e-01
-1.00026679e+00 -7.98102677e-01 6.30353868e-01 -6.48990154e-01
3.65982443e-01 -2.66970128e-01 -6.82429433e-01 4.14705276e-01
4.22525167e-01 -2.61600018e-01 4.87278461e-01 -1.40323028e-01
-2.15136439e-01 1.47931695e-01 -1.09786510e+00 7.99181461e-01
1.08652699e+00 -3.28670442e-01 -1.12452701e-01 2.96137985e-02
3.43978524e-01 -2.93257654e-01 -4.44550037e-01 2.30684549e-01
3.76615644e-01 -1.22490752e+00 8.34195971e-01 -1.51213497e-01
3.68627161e-01 -3.96308184e-01 6.43927138e-03 -1.62023282e+00
-2.69072741e-01 -7.38631964e-01 1.17747568e-01 1.34819281e+00
6.72235668e-01 -2.80391127e-01 8.86760235e-01 1.01302743e+00
-1.20304022e-02 -5.31614721e-01 -3.44116658e-01 -7.16507792e-01
1.41431421e-01 -5.76508224e-01 4.09401089e-01 7.54555702e-01
-4.07470614e-01 1.93838671e-01 -3.56935978e-01 -2.42054462e-01
6.03417754e-01 -1.96464986e-01 8.67843628e-01 -1.07369900e+00
-4.81410742e-01 -7.57199109e-01 -3.87718350e-01 -1.21490371e+00
-2.06964090e-01 -6.39792621e-01 1.40150592e-01 -1.81621480e+00
-2.07762048e-02 -7.39273846e-01 4.95131090e-02 2.88743407e-01
-3.25060070e-01 3.60581994e-01 2.02922344e-01 1.20107885e-02
-1.15136482e-01 3.92186075e-01 1.34172225e+00 -1.41333565e-01
-2.44997740e-01 -6.88312948e-02 -8.68066430e-01 5.93166769e-01
8.95176530e-01 -5.35345197e-01 -7.82230198e-01 -6.41744792e-01
2.36555427e-01 -3.99826676e-01 2.96323955e-01 -1.01613975e+00
2.02128932e-01 -2.97038138e-01 4.77150768e-01 -1.20329641e-01
4.24423307e-01 -6.84003711e-01 2.62880325e-01 1.22138046e-01
-5.20842075e-01 -1.05685376e-01 1.06032956e-02 7.61636138e-01
-6.07483685e-02 -2.24244282e-01 8.69695842e-01 -3.44992310e-01
-5.99861443e-01 1.81905717e-01 -4.99995083e-01 -1.20930478e-01
1.06247437e+00 -5.44548392e-01 4.22850773e-02 -6.08342767e-01
-7.67821908e-01 -1.19695522e-01 7.11718380e-01 5.40448546e-01
5.54478765e-01 -1.02229607e+00 -7.07377553e-01 2.88374901e-01
-8.84192809e-02 -2.97127068e-02 3.25809009e-02 5.58605850e-01
-3.01481545e-01 -1.08508348e-01 -5.33768944e-02 -6.09855473e-01
-1.09334123e+00 7.01598823e-01 1.99915618e-01 -1.93208512e-02
-5.76873183e-01 9.26836848e-01 8.97472948e-02 -6.19068146e-02
1.65936545e-01 -1.72524765e-01 1.13574885e-01 1.95829640e-03
4.80130762e-01 2.60201067e-01 -2.04551131e-01 -3.20623457e-01
5.94301932e-02 3.55891049e-01 -1.17124818e-01 -7.69641101e-01
1.14030159e+00 -9.20550153e-02 1.76048592e-01 4.22786653e-01
7.23877251e-01 1.63395301e-01 -1.72309017e+00 1.23715237e-01
-1.00265369e-01 -7.57906258e-01 2.28527665e-01 -1.01126909e+00
-9.92813468e-01 1.02593136e+00 6.93847835e-01 3.11846972e-01
9.14617360e-01 -7.65989274e-02 3.33190024e-01 1.71731226e-02
3.23991358e-01 -1.23849034e+00 2.17764288e-01 1.39389664e-01
6.70225918e-01 -1.12999988e+00 1.02441050e-01 -5.01353323e-01
-9.04451370e-01 8.50279927e-01 3.69368851e-01 -2.48696476e-01
6.70988917e-01 6.02865636e-01 3.55284512e-01 -6.55237436e-02
-5.68795085e-01 -4.37480271e-01 -5.53686917e-02 9.99193072e-01
4.45389569e-01 -1.12030305e-01 -2.84763962e-01 -8.57427344e-02
5.18579502e-03 2.76421607e-01 5.38054168e-01 1.12305617e+00
-3.66084546e-01 -1.68925059e+00 -3.52816731e-01 3.92106652e-01
-5.05778909e-01 -2.43942782e-01 -4.76200521e-01 6.78240001e-01
1.86150357e-01 9.45383847e-01 -1.11767255e-01 -7.55941123e-02
1.58430368e-01 1.28201991e-01 7.61494994e-01 -6.33351207e-01
-3.40225041e-01 1.26406163e-01 5.53780086e-02 -3.65161061e-01
-5.57376027e-01 -8.10060322e-01 -1.01227355e+00 -1.69578955e-01
-6.70633256e-01 4.88246642e-02 8.77463818e-01 6.58230960e-01
5.19582152e-01 3.72908294e-01 2.63595432e-01 -1.05215716e+00
-3.40356201e-01 -7.80237317e-01 -4.04576242e-01 4.92119312e-01
7.62000009e-02 -6.15602314e-01 -1.96473986e-01 4.94220346e-01] | [11.324429512023926, -0.21012060344219208] |
a498ba85-4ae0-4e9f-b8d2-11a0a3db240f | learning-an-effective-context-response | 2009.06265 | null | https://arxiv.org/abs/2009.06265v1 | https://arxiv.org/pdf/2009.06265v1.pdf | Learning an Effective Context-Response Matching Model with Self-Supervised Tasks for Retrieval-based Dialogues | Building an intelligent dialogue system with the ability to select a proper response according to a multi-turn context is a great challenging task. Existing studies focus on building a context-response matching model with various neural architectures or PLMs and typically learning with a single response prediction task. These approaches overlook many potential training signals contained in dialogue data, which might be beneficial for context understanding and produce better features for response prediction. Besides, the response retrieved from existing dialogue systems supervised by the conventional way still faces some critical challenges, including incoherence and inconsistency. To address these issues, in this paper, we propose learning a context-response matching model with auxiliary self-supervised tasks designed for the dialogue data based on pre-trained language models. Specifically, we introduce four self-supervised tasks including next session prediction, utterance restoration, incoherence detection and consistency discrimination, and jointly train the PLM-based response selection model with these auxiliary tasks in a multi-task manner. By this means, the auxiliary tasks can guide the learning of the matching model to achieve a better local optimum and select a more proper response. Experiment results on two benchmarks indicate that the proposed auxiliary self-supervised tasks bring significant improvement for multi-turn response selection in retrieval-based dialogues, and our model achieves new state-of-the-art results on both datasets. | ['Rui Yan', 'Chongyang Tao', 'Dongyan Zhao', 'Daxin Jiang', 'Xueliang Zhao', 'Ruijian Xu'] | 2020-09-14 | null | null | null | null | ['conversational-response-selection'] | ['natural-language-processing'] | [ 4.55646634e-01 -1.67419598e-01 -3.14045131e-01 -8.86611938e-01
-1.10169697e+00 -1.98374912e-01 6.87561929e-01 1.38925925e-01
-2.66812593e-01 6.71657324e-01 6.40636981e-01 -7.47611523e-02
-9.58851948e-02 -3.52725416e-01 2.84040179e-02 -6.63601816e-01
3.26812059e-01 7.59840012e-01 2.17363179e-01 -8.86285484e-01
6.73011065e-01 -2.29603454e-01 -1.20045698e+00 8.99728000e-01
1.11656797e+00 1.07968271e+00 4.98358011e-01 8.16946983e-01
-3.49173635e-01 1.04994273e+00 -6.98528886e-01 -2.15900200e-03
-2.83038914e-01 -9.05590236e-01 -1.28494012e+00 -1.69377744e-01
5.25447214e-03 -3.31371427e-01 -1.03615664e-01 5.82639456e-01
7.98712969e-01 5.90461493e-01 4.90761042e-01 -8.32593024e-01
-5.30524194e-01 6.81723297e-01 -1.36448979e-01 1.16299838e-01
8.71609390e-01 2.14312628e-01 1.14863873e+00 -8.78474176e-01
3.24315786e-01 1.67407882e+00 3.71439248e-01 8.93214047e-01
-1.23352599e+00 -4.13836122e-01 3.19611132e-01 4.60094273e-01
-8.53026807e-01 -8.12016904e-01 1.03572345e+00 1.61710218e-01
1.05907488e+00 5.29826880e-01 -3.05380626e-03 1.25354791e+00
6.66702762e-02 1.25035954e+00 1.12682807e+00 -4.73897040e-01
-5.19839153e-02 4.19342756e-01 3.01143080e-01 3.08541566e-01
-1.02538550e+00 -2.57240564e-01 -7.46603012e-01 -1.93981290e-01
6.25874028e-02 -2.22398043e-01 -4.58282322e-01 1.48126721e-01
-1.07863688e+00 9.63657022e-01 2.20744029e-01 1.96542889e-01
-1.77613631e-01 -5.39789617e-01 7.67597079e-01 8.86794269e-01
6.84099197e-01 4.71435547e-01 -6.98073864e-01 -2.34630108e-01
-4.47437912e-01 2.27504492e-01 1.01093817e+00 7.63059795e-01
6.66835308e-01 -1.61469072e-01 -6.72708988e-01 1.59355056e+00
3.49591494e-01 2.29998395e-01 8.32303584e-01 -7.11710989e-01
8.11621428e-01 7.07160413e-01 -9.23159719e-02 -1.04317784e+00
-7.23430276e-01 -1.38130218e-01 -9.85433936e-01 -5.66003680e-01
2.43781447e-01 -2.03095466e-01 -2.63268739e-01 1.75157082e+00
4.28169459e-01 -2.15411365e-01 4.17600304e-01 1.05961013e+00
1.21264553e+00 8.45024884e-01 -3.43261845e-02 -5.36075354e-01
1.09794343e+00 -1.40337026e+00 -1.00609159e+00 -3.56954336e-01
9.36221182e-01 -1.04187953e+00 1.39242864e+00 3.04246962e-01
-8.62432718e-01 -6.78450108e-01 -7.29439199e-01 -2.31985897e-02
3.83191034e-02 2.85958976e-01 3.21476698e-01 2.15131178e-01
-8.16204309e-01 2.86221713e-01 1.89914498e-02 -2.84481198e-01
-3.13311994e-01 2.91560411e-01 -1.38199329e-01 1.24711022e-02
-1.69368839e+00 1.18993843e+00 1.64125219e-01 3.12246323e-01
-5.23530722e-01 -1.53184950e-01 -7.25094318e-01 -1.29536718e-01
6.45761371e-01 -3.27955484e-01 1.57211328e+00 -9.51019406e-01
-2.15385389e+00 6.46966040e-01 -3.03902328e-01 -3.06793243e-01
2.30214566e-01 -2.23584138e-02 -4.98748004e-01 2.83557712e-03
9.95336659e-03 4.21748728e-01 7.13905036e-01 -1.12200391e+00
-8.11317086e-01 -3.96111995e-01 -5.36754802e-02 7.06799865e-01
-3.03567559e-01 1.09836236e-01 -2.66221136e-01 -2.93122202e-01
1.71164081e-01 -8.17592323e-01 -1.52563676e-02 -8.35997462e-01
-4.49215055e-01 -9.44370270e-01 8.16955268e-01 -5.95633686e-01
1.41542208e+00 -1.73861372e+00 2.73431897e-01 -1.72669232e-01
-4.24381979e-02 2.95542181e-01 -5.51532626e-01 6.32130444e-01
2.39466190e-01 -2.61445940e-01 -2.93379687e-02 -5.23562729e-01
-4.61022519e-02 -3.40498760e-02 -3.22542906e-01 1.79693341e-01
1.93672225e-01 6.06276631e-01 -8.71622026e-01 -4.46458310e-01
1.53719738e-01 -6.18417412e-02 -4.43064362e-01 9.17892754e-01
-4.10688341e-01 9.29033220e-01 -6.10487998e-01 4.88917530e-01
3.69769722e-01 -1.29928470e-01 4.24884409e-01 -1.17069200e-01
1.49145335e-01 9.44976330e-01 -9.80117142e-01 1.70468271e+00
-8.63155782e-01 4.14293021e-01 2.84734160e-01 -1.24622774e+00
1.37678790e+00 2.71792918e-01 2.10208908e-01 -1.28219283e+00
3.38065811e-03 1.76417649e-01 1.92369655e-01 -7.69132078e-01
9.33283448e-01 4.42763381e-02 -3.52063239e-01 8.49823952e-01
2.92411763e-02 -1.36828914e-01 -1.67105511e-01 2.57962167e-01
8.17486048e-01 -5.45314327e-02 1.85522437e-01 3.79682034e-02
1.04447103e+00 -4.52030003e-02 5.63935399e-01 8.08846056e-01
-2.57437259e-01 4.39290822e-01 3.08148205e-01 -3.63675505e-01
-5.43141127e-01 -2.74767488e-01 6.81982748e-03 1.82472920e+00
3.28728735e-01 -2.53836274e-01 -4.79525298e-01 -8.41889977e-01
-4.18844461e-01 6.92570150e-01 -2.76387811e-01 -4.25774574e-01
-7.32568622e-01 -7.33854651e-01 4.11390245e-01 1.83376774e-01
6.15554094e-01 -1.41852570e+00 -8.74588042e-02 4.68136221e-01
-8.69341731e-01 -8.97243023e-01 -6.55563116e-01 3.50738913e-01
-6.65539443e-01 -9.01209235e-01 -3.62158477e-01 -9.59338605e-01
3.04376274e-01 5.30416012e-01 1.20738995e+00 4.37393188e-01
3.36109519e-01 2.43463680e-01 -7.06906617e-01 1.02068983e-01
-8.56487334e-01 3.74280781e-01 6.34737089e-02 8.31256583e-02
4.99904186e-01 -1.72279507e-01 -4.85799789e-01 7.02118814e-01
-4.73185807e-01 2.49333531e-02 3.12744737e-01 1.55809963e+00
2.82360107e-01 -3.43776464e-01 1.29197788e+00 -9.49243665e-01
1.38019300e+00 -5.44542432e-01 6.45798221e-02 6.41149163e-01
-8.05246234e-01 1.07974097e-01 9.03332829e-01 -4.79868174e-01
-1.57472181e+00 -1.93495452e-01 -3.27058911e-01 3.33082378e-01
-2.00123146e-01 4.84374762e-01 -2.55760998e-01 1.42543435e-01
6.02608085e-01 4.53493834e-01 1.02163009e-01 -4.70956624e-01
3.45652312e-01 1.31108248e+00 3.02351713e-01 -7.14586735e-01
-8.73018578e-02 -1.22642964e-01 -6.40369952e-01 -5.52630126e-01
-8.77889335e-01 -7.74125695e-01 -6.00030422e-01 -2.57432491e-01
3.88698816e-01 -6.14005923e-01 -9.14724648e-01 4.99439001e-01
-1.27758074e+00 -5.01001716e-01 3.08806807e-01 2.85978943e-01
-6.44165158e-01 5.39397061e-01 -8.20091903e-01 -1.00842822e+00
-7.15786457e-01 -1.26273215e+00 8.00343513e-01 4.05950665e-01
-5.10348141e-01 -1.08627141e+00 2.56991208e-01 7.20745325e-01
6.14349127e-01 -6.60088420e-01 1.03522182e+00 -1.30994022e+00
-1.01855971e-01 4.34138700e-02 -4.33468893e-02 2.61015475e-01
1.73131883e-01 -4.31518734e-01 -9.75834131e-01 -1.85552478e-01
5.00630379e-01 -1.12188065e+00 8.59556079e-01 -3.93796451e-02
9.68216836e-01 -5.54440439e-01 -5.25320098e-02 1.81978852e-01
7.52280533e-01 4.16685373e-01 3.55912715e-01 2.75820792e-01
3.19333464e-01 1.09609270e+00 1.18843830e+00 5.13047338e-01
8.18634152e-01 9.16097879e-01 2.34644592e-01 1.33622736e-01
1.53000951e-01 -1.86494231e-01 2.34739885e-01 1.24335957e+00
5.81765890e-01 -4.11994398e-01 -6.18239462e-01 3.34355831e-01
-2.29691982e+00 -7.40400672e-01 -1.71316788e-01 2.09396267e+00
1.43758321e+00 -2.29388047e-02 5.77075258e-02 -1.64550647e-01
6.74077749e-01 4.16298747e-01 -5.88515937e-01 -6.96191311e-01
-2.04524234e-01 -2.03946397e-01 -2.99106479e-01 6.09818161e-01
-1.00129116e+00 1.03851485e+00 5.50714111e+00 9.09401000e-01
-1.16088533e+00 6.57996088e-02 8.21345806e-01 2.02864245e-01
-2.59698153e-01 -2.99003050e-02 -9.50531006e-01 4.33152258e-01
8.91162753e-01 1.08366840e-01 4.23928946e-01 6.42304182e-01
4.16866153e-01 -3.49457502e-01 -1.12308085e+00 7.50379920e-01
3.62852812e-01 -1.00551045e+00 -1.11054532e-01 -6.07194364e-01
5.01576126e-01 -3.59179765e-01 -1.48436189e-01 8.00760329e-01
3.07558477e-01 -9.52772796e-01 6.25338256e-02 6.42321885e-01
3.76951158e-01 -6.65817022e-01 1.00463104e+00 7.65300632e-01
-7.85124838e-01 -1.62104607e-01 -4.09181535e-01 -8.64234641e-02
2.50659753e-02 -2.73547368e-03 -1.07303393e+00 4.82023805e-01
3.24366748e-01 5.50906003e-01 -4.85769957e-01 6.44587338e-01
-1.44048676e-01 5.26520431e-01 -7.35252127e-02 -5.16193688e-01
2.21463278e-01 -1.65377975e-01 4.54226702e-01 1.28808641e+00
-2.34091237e-01 2.16939136e-01 5.38174629e-01 4.97862488e-01
-6.57774508e-02 4.73387092e-01 -1.79684892e-01 3.10456038e-01
7.89905846e-01 1.29728949e+00 -2.70295471e-01 -1.61378473e-01
-2.57700533e-01 9.41522717e-01 4.81962323e-01 2.09224492e-01
-4.66861963e-01 -2.96929628e-01 2.77824700e-01 -4.03398395e-01
-3.30775589e-01 2.77482539e-01 -3.81519139e-01 -1.00975549e+00
-7.97605608e-03 -1.33902419e+00 5.20566285e-01 -4.49384809e-01
-1.39030528e+00 7.14750051e-01 -2.91718423e-01 -1.24400616e+00
-8.90834808e-01 -8.89544189e-02 -8.69103372e-01 1.02624989e+00
-1.55007172e+00 -1.02181482e+00 -1.23719752e-01 6.97612047e-01
1.33387089e+00 -4.58861172e-01 1.14436066e+00 4.94875610e-02
-6.14060462e-01 8.17779064e-01 1.97476149e-01 -1.79551002e-02
1.40806150e+00 -1.17808354e+00 -4.44930121e-02 3.21123481e-01
-2.79907912e-01 6.49941564e-01 5.53716660e-01 -3.87477428e-01
-1.40633106e+00 -6.62559927e-01 1.00698376e+00 -1.59217343e-01
3.88342708e-01 -1.02592155e-01 -1.20360720e+00 1.72953993e-01
4.65867192e-01 -4.99274641e-01 8.04769397e-01 5.63293457e-01
-7.01873517e-03 -1.93795428e-01 -8.46409857e-01 5.14348328e-01
6.58777714e-01 -7.85475791e-01 -7.41553068e-01 6.22902274e-01
7.43795514e-01 -6.20974541e-01 -6.42303526e-01 3.35687727e-01
3.92894715e-01 -9.49143887e-01 7.28073478e-01 -8.68756711e-01
4.26333129e-01 1.63608402e-01 -3.66082415e-02 -1.63196266e+00
-8.66279379e-02 -6.53133631e-01 5.61354123e-02 1.45516348e+00
3.93458098e-01 -3.97731006e-01 4.00819004e-01 7.56280899e-01
-2.73123056e-01 -9.69580710e-01 -9.12941217e-01 -1.44583315e-01
-7.52706407e-03 -1.68937385e-01 4.03341800e-01 1.06539309e+00
4.65096027e-01 1.20106685e+00 -8.66351545e-01 -4.54756021e-01
4.93189394e-02 4.42059010e-01 1.00855660e+00 -1.00752389e+00
-1.75236672e-01 -4.79518890e-01 3.93127948e-01 -1.83868670e+00
5.32417059e-01 -6.38915896e-01 5.82199931e-01 -1.36621141e+00
1.38305306e-01 -6.84760630e-01 7.26716965e-02 2.45120049e-01
-5.60030341e-01 -3.55634540e-01 2.82842368e-02 4.40928429e-01
-1.16124439e+00 9.33889151e-01 1.21336508e+00 -3.89207840e-01
-6.29963696e-01 5.53359568e-01 -6.83423936e-01 4.66870934e-01
8.38222980e-01 -4.31325108e-01 -5.39356887e-01 -2.87404001e-01
-5.99585064e-02 9.21258867e-01 -9.51932073e-02 -3.41031134e-01
6.57160044e-01 -3.39724213e-01 5.10501117e-02 -7.04386830e-01
3.90939444e-01 -4.12521034e-01 -6.83064938e-01 -2.74030515e-03
-1.17513490e+00 -1.27088845e-01 -1.64110854e-01 6.08519256e-01
-4.10043955e-01 -5.02982974e-01 5.75365365e-01 -1.59930646e-01
-7.94242680e-01 -1.37512282e-01 -4.44363534e-01 2.41610408e-01
3.25270891e-01 1.05104461e-01 -5.46250880e-01 -8.49978745e-01
-5.99733710e-01 8.72535050e-01 -1.16569899e-01 7.94855118e-01
8.26258242e-01 -1.14715075e+00 -8.58630657e-01 5.82859181e-02
2.43712112e-01 -1.43199265e-02 5.94901860e-01 8.53103757e-01
2.99602598e-01 5.54522455e-01 -1.67422742e-02 -8.00788939e-01
-1.48600960e+00 1.63434789e-01 3.80151629e-01 -6.40075445e-01
-2.62795001e-01 8.56232762e-01 -5.39262593e-02 -1.19905984e+00
4.92522418e-01 1.01255231e-01 -7.99693525e-01 2.10416064e-01
6.19617820e-01 1.12019107e-01 1.49512902e-01 -5.99799454e-01
-1.74309656e-01 4.15218398e-02 -4.31559712e-01 -2.82479208e-02
1.05018520e+00 -7.40377009e-01 -1.94751456e-01 4.84855652e-01
1.24733853e+00 -1.70413449e-01 -1.04182100e+00 -9.78422523e-01
1.88314840e-01 -3.38410556e-01 -2.18290552e-01 -1.03797054e+00
-6.74501896e-01 9.11620140e-01 2.94707537e-01 4.43136036e-01
1.17435443e+00 -1.70095995e-01 9.51086283e-01 9.60328937e-01
8.91415179e-02 -1.67947364e+00 6.51318133e-01 1.10425007e+00
1.11980343e+00 -1.75574160e+00 -2.63311148e-01 -3.04129515e-02
-1.17110240e+00 1.29865396e+00 1.12683296e+00 2.84119040e-01
2.66750842e-01 -2.15612486e-01 4.28425878e-01 4.31328863e-02
-1.29911160e+00 -7.46295676e-02 2.54642993e-01 4.43247348e-01
7.46688426e-01 -7.58232325e-02 -5.57724655e-01 8.62866223e-01
9.88401175e-02 -5.52832663e-01 3.47390413e-01 6.85271323e-01
-5.91952801e-01 -1.33007812e+00 -1.83487803e-01 4.16672081e-01
-5.82345650e-02 -1.26716867e-01 -7.60001242e-01 2.01521635e-01
-6.34857714e-01 1.60675633e+00 -2.29116574e-01 -8.33388507e-01
3.84140283e-01 2.96858966e-01 -7.78356045e-02 -7.31037557e-01
-1.06864488e+00 1.78139746e-01 5.79937518e-01 -3.73464316e-01
-5.19098818e-01 -2.76685208e-01 -1.16682684e+00 -1.63330197e-01
-6.55199766e-01 3.63540918e-01 3.42835933e-01 1.27796352e+00
2.81982213e-01 4.66766536e-01 1.32776678e+00 -5.46125412e-01
-1.21486914e+00 -1.53413808e+00 -1.20647840e-01 4.16109800e-01
1.97019801e-01 -3.68022084e-01 -2.20194727e-01 -4.52959776e-01] | [12.628416061401367, 7.741860866546631] |
c5292166-b874-4161-92fc-4ca0b9f8e002 | ancient-chinese-word-segmentation-and-part-of-1 | 2303.01912 | null | https://arxiv.org/abs/2303.01912v2 | https://arxiv.org/pdf/2303.01912v2.pdf | Ancient Chinese Word Segmentation and Part-of-Speech Tagging Using Distant Supervision | Ancient Chinese word segmentation (WSG) and part-of-speech tagging (POS) are important to study ancient Chinese, but the amount of ancient Chinese WSG and POS tagging data is still rare. In this paper, we propose a novel augmentation method of ancient Chinese WSG and POS tagging data using distant supervision over parallel corpus. However, there are still mislabeled and unlabeled ancient Chinese words inevitably in distant supervision. To address this problem, we take advantage of the memorization effects of deep neural networks and a small amount of annotated data to get a model with much knowledge and a little noise, and then we use this model to relabel the ancient Chinese sentences in parallel corpus. Experiments show that the model trained over the relabeled data outperforms the model trained over the data generated from distant supervision and the annotated data. Our code is available at https://github.com/farlit/ACDS. | ['Piji Li', 'Shuo Feng'] | 2023-03-03 | null | null | null | null | ['memorization', 'part-of-speech-tagging', 'chinese-word-segmentation'] | ['natural-language-processing', 'natural-language-processing', 'natural-language-processing'] | [-1.28312290e-01 2.48697456e-02 -8.42682272e-02 -6.04730964e-01
-6.69176102e-01 -5.30991495e-01 2.00417086e-01 -1.05108313e-01
-9.85219121e-01 9.75110292e-01 4.64487284e-01 -4.30104673e-01
7.20898509e-01 -6.40760243e-01 -6.13748491e-01 -5.07487774e-01
1.00503370e-01 4.44650501e-01 5.38720667e-01 -2.94965059e-01
2.67264336e-01 5.37789892e-04 -6.28360689e-01 1.24819562e-01
1.26481795e+00 4.90452379e-01 9.09755707e-01 3.04278165e-01
-5.33803999e-01 6.14425838e-01 -8.11143577e-01 -2.62421072e-01
1.68183848e-01 -5.28773546e-01 -1.06906009e+00 -1.74200431e-01
-1.42032132e-01 -2.17381209e-01 -1.69438913e-01 1.34197187e+00
6.73714995e-01 2.13034779e-01 5.05465381e-02 -3.48936498e-01
-1.20211112e+00 1.33172643e+00 -3.55078667e-01 4.98217314e-01
-8.97975862e-02 -2.28972808e-01 8.79907727e-01 -7.83655047e-01
5.96327841e-01 9.42982614e-01 4.96116489e-01 9.70313549e-01
-6.10172272e-01 -6.57286406e-01 5.08163691e-01 9.12574306e-02
-1.38570976e+00 -1.93070412e-01 5.45630574e-01 1.00002706e-01
8.18963587e-01 -1.45652398e-01 7.49882698e-01 1.01637757e+00
-4.24069688e-02 1.31067121e+00 1.08675301e+00 -7.13679135e-01
2.06324738e-03 -2.67495960e-02 5.30499399e-01 5.07605195e-01
3.43900323e-02 -3.21308643e-01 -1.38716400e-01 1.95353568e-01
6.57867610e-01 6.63454533e-02 -4.09131557e-01 7.73382306e-01
-1.01679349e+00 6.88057780e-01 3.14590007e-01 8.96076918e-01
-3.64766642e-02 -1.08612314e-01 2.33405784e-01 1.22109028e-02
7.71810591e-01 4.53657120e-01 -9.86431956e-01 -2.39133909e-01
-7.95364559e-01 -3.79845411e-01 8.22202325e-01 1.16782749e+00
8.84508312e-01 1.58054471e-01 9.96460170e-02 1.08917916e+00
2.52256334e-01 6.83660209e-01 9.91982341e-01 -5.37067890e-01
5.11200368e-01 3.20999831e-01 -1.98697358e-01 -2.40326419e-01
-2.24901304e-01 -1.96331024e-01 -5.19192874e-01 -4.87444907e-01
1.23879835e-01 -7.63016403e-01 -1.46202552e+00 1.58719933e+00
5.06986901e-02 8.70784894e-02 1.60989031e-01 8.14941883e-01
5.98856330e-01 1.05139029e+00 3.90776128e-01 -9.19391438e-02
1.33452690e+00 -1.26709390e+00 -9.78736043e-01 -8.08662474e-01
8.29500258e-01 -1.05078888e+00 1.13158536e+00 1.76122591e-01
-8.87047052e-01 -5.76364577e-01 -9.24701750e-01 -2.71835238e-01
-4.59604561e-01 3.10803473e-01 4.14844483e-01 4.47539717e-01
-8.69733393e-01 7.48441815e-01 -1.17842472e+00 -3.05423319e-01
4.04165924e-01 3.28090996e-01 -8.11979026e-02 -3.55946481e-01
-1.56168354e+00 8.27814043e-01 1.02576840e+00 3.33737046e-01
-5.65809906e-01 -1.37111202e-01 -7.72718549e-01 -1.44649014e-01
3.84676009e-01 7.47661814e-02 1.38654637e+00 -1.16441798e+00
-1.47964573e+00 5.53329468e-01 2.55566649e-02 -2.29938954e-01
3.96572798e-02 -7.90991127e-01 -5.47519922e-01 -1.89821631e-01
2.86945313e-01 5.89680374e-01 7.30664507e-02 -8.63076389e-01
-7.52711356e-01 -4.55225110e-01 -3.03459644e-01 2.07380623e-01
-2.73388147e-01 3.27094436e-01 -8.33397567e-01 -7.53946126e-01
8.59175846e-02 -8.33886266e-01 -7.22870111e-01 -8.33760023e-01
-3.56885582e-01 -4.43511635e-01 6.26987457e-01 -1.19303501e+00
1.27476335e+00 -2.19872427e+00 -4.57686037e-01 -1.32907212e-01
-4.29278880e-01 8.37198555e-01 -3.47179383e-01 2.30222747e-01
1.25120848e-01 4.64096099e-01 -4.79729831e-01 -4.11163330e-01
-3.39118093e-01 7.85005867e-01 -7.63694867e-02 9.09750015e-02
4.02015269e-01 8.64561677e-01 -1.13650572e+00 -7.17320561e-01
-2.99674273e-01 1.69681042e-01 -1.34138435e-01 3.65592629e-01
-3.97879660e-01 5.33892989e-01 -8.58638108e-01 7.21076667e-01
8.19597125e-01 -8.20320845e-02 4.37501580e-01 3.47484261e-01
-2.43678242e-01 6.83724940e-01 -8.56857777e-01 1.96530247e+00
-3.10438961e-01 2.32064068e-01 -2.77015150e-01 -9.53795969e-01
1.02745557e+00 3.28728825e-01 -1.50005147e-01 -8.80671144e-01
3.96092117e-01 5.16073406e-01 8.66852924e-02 -5.81231952e-01
6.70378029e-01 -2.71448731e-01 -1.37576401e-01 3.95333022e-01
3.21656048e-01 3.37320954e-01 2.76920348e-01 1.22236006e-01
1.05073464e+00 3.92145067e-01 1.92602158e-01 -1.87737197e-01
3.46548378e-01 3.09320420e-01 1.23586512e+00 5.29222190e-01
-5.64579777e-02 7.20180929e-01 1.15682259e-01 -2.39349604e-01
-1.05339932e+00 -7.08608985e-01 -1.30380481e-01 1.06764817e+00
1.11177951e-01 -5.31705916e-01 -8.11208427e-01 -9.53666866e-01
-6.74235046e-01 6.61182344e-01 -4.19012159e-01 1.72602072e-01
-1.26967323e+00 -9.79973316e-01 6.93389893e-01 9.82273042e-01
7.05351651e-01 -1.44510293e+00 9.65492949e-02 4.63808268e-01
-3.98542881e-01 -1.12254393e+00 -6.22634232e-01 4.38231111e-01
-9.31368411e-01 -7.69080639e-01 -9.28538799e-01 -1.33156347e+00
8.54972661e-01 9.13569983e-03 9.34184611e-01 3.62569898e-01
2.28791893e-01 -3.51793498e-01 -1.04000556e+00 -5.09294569e-01
-3.70156109e-01 3.08139920e-01 -1.75577402e-01 -4.50124949e-01
8.36413205e-01 -2.86395282e-01 -4.06640112e-01 5.59287071e-02
-9.46191967e-01 -1.32018859e-02 8.61720443e-01 8.98181081e-01
5.76395094e-01 -2.77510852e-01 4.06217456e-01 -1.38468516e+00
3.64529878e-01 -4.93415415e-01 -5.84668159e-01 2.90030658e-01
-3.89304876e-01 1.42387107e-01 7.26677120e-01 -5.64243674e-01
-1.56916189e+00 4.67460528e-02 -8.18530798e-01 2.23789155e-01
-2.79680550e-01 8.25373530e-01 -4.17461365e-01 4.19576317e-01
4.22382772e-01 2.53211081e-01 -4.90123063e-01 -1.12036014e+00
3.35867852e-01 1.07495821e+00 3.34566385e-01 -7.23494172e-01
3.15407455e-01 1.10642493e-01 -6.82608068e-01 -7.00488746e-01
-1.17709994e+00 -4.62545991e-01 -9.79752541e-01 4.39639062e-01
1.19220090e+00 -9.78917301e-01 1.57841682e-01 5.63890457e-01
-1.33672976e+00 -6.58565521e-01 -1.24263503e-01 6.81165874e-01
3.07507724e-01 7.17751384e-01 -1.12697661e+00 -6.32964015e-01
-3.96921843e-01 -7.31046498e-01 6.50966287e-01 5.90183079e-01
1.41107291e-01 -9.48736966e-01 2.70519167e-01 1.44717366e-01
9.12908986e-02 -3.27488244e-01 6.52945697e-01 -1.24190617e+00
-4.52746242e-01 -2.74380714e-01 1.65222790e-02 8.30004752e-01
2.06975564e-01 -2.06090435e-01 -7.83273518e-01 1.14744594e-02
4.03357968e-02 -3.35265517e-01 9.42453027e-01 1.61732629e-01
8.64397585e-01 -1.30840287e-01 -2.52137333e-01 2.79592842e-01
1.40784860e+00 6.37027025e-01 6.12663865e-01 3.89070362e-01
8.39191258e-01 5.07992983e-01 8.54730308e-01 1.02975532e-01
2.96510667e-01 2.35598385e-02 -2.49590814e-01 3.94096412e-03
-9.11861807e-02 -3.77506256e-01 4.24267173e-01 1.63583016e+00
-2.28155375e-01 -3.09944481e-01 -1.26340139e+00 9.18201506e-01
-1.79679644e+00 -3.75879377e-01 -2.31607884e-01 1.85005641e+00
1.28811467e+00 1.87956795e-01 -4.41394895e-01 -3.69030148e-01
9.20529783e-01 1.05991766e-01 -1.48520404e-02 -2.19380200e-01
-2.60302901e-01 5.66347182e-01 6.28074586e-01 5.19725978e-01
-1.02536416e+00 1.55116820e+00 5.53500700e+00 1.07934010e+00
-1.02583992e+00 4.70500201e-01 5.90029120e-01 5.06887317e-01
-3.08072388e-01 4.35045570e-01 -1.20454574e+00 6.58993542e-01
1.11268699e+00 3.04232538e-01 2.63427924e-02 8.54421020e-01
-7.20609054e-02 -7.06696734e-02 -6.10352039e-01 3.02416205e-01
1.57846764e-01 -1.03508794e+00 -1.62028074e-01 -1.45490244e-01
8.48707616e-01 7.15732813e-01 -5.67116320e-01 4.43297982e-01
8.21715236e-01 -6.58804297e-01 5.24928868e-01 2.01040447e-01
4.81494755e-01 -4.27487075e-01 1.18806279e+00 6.09436333e-01
-1.10445058e+00 2.36492723e-01 -8.11387300e-01 -9.47016254e-02
4.98869419e-01 5.54012060e-01 -6.58225656e-01 5.91916919e-01
6.98721051e-01 8.38238657e-01 -5.57071447e-01 8.92218590e-01
-1.01053154e+00 1.15580177e+00 -3.34941417e-01 -3.48578036e-01
4.98167038e-01 -3.76076072e-01 -1.91522557e-02 1.39361382e+00
3.92642826e-01 5.44133008e-01 3.59843850e-01 4.95311439e-01
-1.09895200e-01 2.86708623e-01 -1.56108022e-01 -2.85271734e-01
5.59520781e-01 1.09666193e+00 -9.26818132e-01 -5.96147478e-01
-7.33188868e-01 1.19584692e+00 5.94384909e-01 3.65292639e-01
-5.35598278e-01 -5.31032145e-01 1.14798851e-01 -2.62248904e-01
3.96549970e-01 -3.95642817e-01 -3.09341520e-01 -1.50773978e+00
-4.08135876e-02 -4.63587850e-01 5.41285276e-01 -6.57861769e-01
-1.36034250e+00 9.49130774e-01 -3.13675344e-01 -9.53451574e-01
2.61417240e-01 -7.37322390e-01 -7.97586381e-01 1.03646660e+00
-1.75716615e+00 -1.18912053e+00 1.13600507e-01 3.45473200e-01
7.44160116e-01 -2.09250022e-02 7.52723455e-01 5.56440294e-01
-8.20167303e-01 2.48046473e-01 3.09945315e-01 8.73926520e-01
7.71166801e-01 -1.44661117e+00 8.02304864e-01 1.19760203e+00
1.49796441e-01 7.16906130e-01 1.44044384e-01 -1.15640903e+00
-5.97231746e-01 -1.03036690e+00 1.37501299e+00 -2.28046477e-01
6.92486525e-01 -3.83006155e-01 -1.26293778e+00 1.07228529e+00
4.60994244e-01 3.38259637e-02 9.09508228e-01 2.03851521e-01
7.49121308e-02 1.48761168e-01 -5.51863432e-01 5.84524751e-01
8.97774041e-01 -1.79008976e-01 -1.19877267e+00 2.33369038e-01
9.66916323e-01 -3.02431852e-01 -6.31196797e-01 1.77515477e-01
9.77959558e-02 -3.02955061e-01 4.23055947e-01 -4.64357823e-01
3.67337257e-01 -3.74302715e-01 -4.19859476e-02 -1.30363178e+00
-2.74433732e-01 -3.55923593e-01 6.23393178e-01 1.54655552e+00
9.12531912e-01 -4.74616110e-01 8.24335039e-01 5.06695390e-01
-8.19193065e-01 -3.87026757e-01 -5.66517949e-01 -8.34095597e-01
4.46985155e-01 -1.60816848e-01 4.85308915e-01 1.20610619e+00
7.21858367e-02 6.38264120e-01 -4.20459747e-01 1.13272883e-01
4.43214513e-02 -5.41589633e-02 1.59096569e-01 -9.40170348e-01
-1.81212038e-01 1.61320537e-01 1.14176124e-01 -1.38247466e+00
1.22765854e-01 -8.25524688e-01 6.60046518e-01 -1.57535720e+00
1.08241558e-01 -6.39378548e-01 -4.38589394e-01 8.38504195e-01
-7.71226466e-01 1.61100283e-01 1.60636663e-01 3.12485307e-01
-7.24560499e-01 6.74546003e-01 1.33379602e+00 2.69258738e-01
-2.00105295e-01 -1.22644909e-01 -5.29849052e-01 7.79402196e-01
1.11772680e+00 -8.43570769e-01 1.24287821e-01 -9.84197438e-01
8.38900357e-02 -2.15946987e-01 -4.01719183e-01 -6.84308410e-01
2.38095284e-01 -1.88468039e-01 5.38705170e-01 -5.23026586e-01
2.99897064e-02 -5.65219283e-01 -4.44090754e-01 4.64920640e-01
5.67499325e-02 4.12658826e-02 4.13692892e-02 3.56795937e-01
-3.55405927e-01 -6.77581251e-01 6.19517207e-01 -7.59616375e-01
-1.15342033e+00 2.18964875e-01 -4.99618769e-01 3.75581771e-01
5.04253566e-01 9.80502367e-02 -5.10558844e-01 1.88822642e-01
-7.28062630e-01 5.04025578e-01 1.76555142e-01 2.53276706e-01
3.83656681e-01 -1.19188929e+00 -6.88601077e-01 1.45473167e-01
-1.47724494e-01 4.34178293e-01 1.49763033e-01 4.61413860e-01
-8.50539088e-01 1.80813283e-01 -2.98003256e-01 -5.02557047e-02
-9.34715033e-01 4.63562101e-01 -9.30541083e-02 -3.68785709e-01
-5.19995213e-01 1.07076454e+00 -5.28558530e-02 -7.31742024e-01
-3.99206914e-02 -2.95353293e-01 -4.92104560e-01 -2.38003179e-01
4.65357184e-01 -2.75033284e-02 -1.30421922e-01 -4.80767846e-01
-2.28356376e-01 4.22147721e-01 -3.68130267e-01 -1.17058776e-01
1.48240495e+00 -2.70615220e-01 -3.17844540e-01 3.97333622e-01
9.20769572e-01 2.86562383e-01 -1.08764637e+00 -5.34038961e-01
4.71749157e-01 -2.36114949e-01 -3.22393775e-01 -8.38791966e-01
-9.55208004e-01 1.03418446e+00 2.37667769e-01 -1.59829855e-01
9.58146095e-01 2.25105166e-01 1.29571414e+00 5.56512594e-01
2.16087550e-01 -1.50441742e+00 -9.98733416e-02 1.07322752e+00
2.77119547e-01 -1.30522978e+00 -3.07872981e-01 -3.89645755e-01
-9.50995743e-01 9.28960681e-01 8.55144083e-01 -2.73471862e-01
9.31407094e-01 2.65447766e-01 7.10251689e-01 1.33800894e-01
-2.42392197e-01 -5.00546217e-01 -1.31403282e-01 5.67121983e-01
6.41653419e-01 -1.41081307e-02 -9.19803202e-01 1.14938581e+00
-1.70371175e-01 -9.34949331e-03 5.51908195e-01 1.35894513e+00
-7.09381938e-01 -1.78734899e+00 -8.96431357e-02 1.85735747e-01
-1.01359773e+00 -7.07777202e-01 -2.89222956e-01 6.52536273e-01
3.86520535e-01 8.75689387e-01 -1.97794102e-02 -1.73732072e-01
7.22790277e-03 2.34861702e-01 3.24180163e-02 -1.11218417e+00
-5.12693107e-01 5.77428877e-01 3.26249480e-01 -7.37912878e-02
-6.66348875e-01 -4.48386341e-01 -1.61139500e+00 1.47768691e-01
-6.11223340e-01 7.38992393e-01 4.60368007e-01 1.27560663e+00
-6.42704368e-02 3.14786315e-01 4.01052117e-01 -5.20781696e-01
-2.06763953e-01 -1.42253530e+00 -9.11431313e-01 1.97791204e-01
4.77571897e-02 5.78432493e-02 -1.15808912e-01 1.85702473e-01] | [9.960715293884277, 10.088488578796387] |
da38cdc6-9221-424c-9289-ee181cd7999a | analysis-and-tuning-of-a-voice-assistant | 2106.11759 | null | https://arxiv.org/abs/2106.11759v1 | https://arxiv.org/pdf/2106.11759v1.pdf | Analysis and Tuning of a Voice Assistant System for Dysfluent Speech | Dysfluencies and variations in speech pronunciation can severely degrade speech recognition performance, and for many individuals with moderate-to-severe speech disorders, voice operated systems do not work. Current speech recognition systems are trained primarily with data from fluent speakers and as a consequence do not generalize well to speech with dysfluencies such as sound or word repetitions, sound prolongations, or audible blocks. The focus of this work is on quantitative analysis of a consumer speech recognition system on individuals who stutter and production-oriented approaches for improving performance for common voice assistant tasks (i.e., "what is the weather?"). At baseline, this system introduces a significant number of insertion and substitution errors resulting in intended speech Word Error Rates (isWER) that are 13.64\% worse (absolute) for individuals with fluency disorders. We show that by simply tuning the decoding parameters in an existing hybrid speech recognition system one can improve isWER by 24\% (relative) for individuals with fluency disorders. Tuning these parameters translates to 3.6\% better domain recognition and 1.7\% better intent recognition relative to the default setup for the 18 study participants across all stuttering severities. | ['Jefferey Bigham', 'Sachin Kajarekar', 'Panayiotis Georgiou', 'Shrinath Thelapurath', 'Ashwini Palekar', 'Darren Botten', 'Sarah Wu', 'Lauren Tooley', 'Colin Lea', 'Zifang Huang', 'Vikramjit Mitra'] | 2021-06-18 | null | null | null | null | ['intent-recognition'] | ['natural-language-processing'] | [ 2.07670197e-01 2.14192450e-01 -8.02587792e-02 -2.98345327e-01
-1.05686080e+00 -4.25733298e-01 2.19524086e-01 -2.99268931e-01
-4.85094279e-01 5.66389084e-01 7.57624149e-01 -7.09238708e-01
1.83031693e-01 1.28280120e-02 3.60463886e-03 -1.39875770e-01
2.68289924e-01 5.36251009e-01 -6.59544617e-02 -3.45283300e-01
-2.28927642e-01 2.97510117e-01 -1.26640904e+00 2.90326208e-01
1.07134938e+00 3.60463947e-01 3.71121466e-01 8.71828139e-01
4.42227349e-02 4.14061576e-01 -1.17995453e+00 -1.20026179e-01
-4.35200818e-02 -5.79612672e-01 -9.29063380e-01 3.14225823e-01
4.42621887e-01 -6.71622515e-01 -5.68402946e-01 8.77237201e-01
9.21498775e-01 3.16970944e-01 2.53088653e-01 -5.59928119e-01
-5.78998387e-01 8.10460687e-01 3.19314450e-01 7.94264436e-01
6.70915008e-01 6.74702942e-01 6.89079821e-01 -7.07983494e-01
2.00669110e-01 1.47163785e+00 5.90723276e-01 9.52876151e-01
-1.43622673e+00 -4.90828782e-01 1.95653036e-01 3.76682170e-02
-1.14828527e+00 -1.43971467e+00 1.52971730e-01 -3.61432463e-01
2.01156378e+00 5.57952285e-01 5.58560610e-01 1.20418453e+00
-2.53040284e-01 4.69213814e-01 7.07441032e-01 -4.40795571e-01
-7.18866591e-04 1.93097323e-01 5.82411647e-01 2.10078910e-01
-1.52926728e-01 4.16195065e-01 -6.52243257e-01 1.27311245e-01
2.76600838e-01 -4.88189578e-01 -6.48521185e-01 7.90904403e-01
-1.13168180e+00 5.16004205e-01 -1.19964398e-01 5.60435832e-01
-3.16352934e-01 -1.81549281e-01 4.77265120e-01 7.02381611e-01
4.28123802e-01 4.07932699e-01 -5.80126822e-01 -8.50758076e-01
-8.70362461e-01 -1.84682920e-03 7.87589967e-01 8.09907734e-01
-6.61735684e-02 7.33428299e-01 -3.81886870e-01 1.67633533e+00
5.79269007e-02 6.78238928e-01 1.05378234e+00 -6.27603352e-01
6.95905030e-01 1.21105224e-01 -1.39881363e-02 -3.00506353e-01
-5.26511192e-01 -5.67343473e-01 -2.26895571e-01 1.19155226e-02
4.84729052e-01 -3.61680955e-01 -1.23396790e+00 1.84257841e+00
-3.28907579e-01 -4.73445386e-01 1.42665356e-01 7.14009881e-01
6.79381847e-01 6.68747246e-01 1.96211740e-01 -6.17405415e-01
1.23918343e+00 -8.83745849e-01 -1.16032696e+00 -7.65292227e-01
8.32512021e-01 -9.37602758e-01 1.72783220e+00 5.41316092e-01
-1.35093939e+00 -4.11598921e-01 -8.12488616e-01 1.34186134e-01
-6.97938427e-02 2.96592098e-02 8.97738561e-02 1.42172432e+00
-1.37063336e+00 3.11671346e-01 -6.97680712e-01 -4.59285825e-01
-4.90060449e-02 4.87047821e-01 -2.66532719e-01 8.41263309e-02
-1.30376589e+00 1.26802301e+00 2.79225051e-01 -3.94160777e-01
-5.59947371e-01 -8.15533638e-01 -7.99497545e-01 9.69622657e-03
7.63766374e-03 -2.47115806e-01 1.83378291e+00 -7.70325840e-01
-1.68896663e+00 7.00332761e-01 -3.67831290e-01 -4.32595879e-01
3.03223729e-01 -3.39470059e-01 -1.21172619e+00 -4.15379964e-02
-1.64631933e-01 2.46620089e-01 5.93029320e-01 -5.80903172e-01
-5.57147861e-01 -3.59641492e-01 -5.73302269e-01 4.80146676e-01
-3.34419668e-01 6.88452542e-01 1.57946095e-01 -9.30599213e-01
-8.02970529e-02 -8.43563795e-01 3.48118573e-01 -6.17355943e-01
-3.49507809e-01 -2.93110877e-01 6.10723436e-01 -1.49455893e+00
1.51729298e+00 -2.23923683e+00 -1.47619411e-01 -8.35580751e-02
-2.34604552e-01 9.46375966e-01 1.13457339e-02 1.75125822e-01
-1.95154473e-01 3.75876635e-01 -1.54627159e-01 -4.07902062e-01
6.68797344e-02 2.04934627e-01 -1.63223892e-01 3.03705901e-01
-5.18567003e-02 5.24971128e-01 -6.70875728e-01 2.37293035e-01
4.64761287e-01 3.25368404e-01 -6.72204614e-01 2.70339400e-01
8.66577625e-02 3.97315502e-01 4.47567403e-01 5.15371919e-01
2.27949575e-01 4.55837131e-01 -7.08982795e-02 5.76352298e-01
-1.53656021e-01 1.24067354e+00 -6.68566883e-01 1.28153765e+00
-7.06744492e-01 6.30813181e-01 4.12522495e-01 -6.78682923e-01
6.61081612e-01 7.64554918e-01 -1.27490565e-01 -6.15624487e-01
-2.81516612e-02 5.75358927e-01 7.94300139e-01 -5.60589969e-01
2.00197697e-01 -5.49715400e-01 1.92353949e-01 1.39261574e-01
-3.56375091e-02 -3.27501982e-01 -1.39244169e-01 -2.28439137e-01
1.36358047e+00 -9.59722757e-01 2.26165175e-01 -2.47604832e-01
3.75717163e-01 -2.27442220e-01 3.32615823e-01 7.78628111e-01
-5.43903232e-01 5.17408788e-01 1.32763349e-02 1.59635723e-01
-8.89846742e-01 -1.21201229e+00 -1.98917180e-01 1.22247314e+00
-7.20757008e-01 -5.90741456e-01 -1.01216757e+00 -1.55888245e-01
-2.01289833e-01 1.43985236e+00 3.56748044e-01 -4.62694943e-01
-6.47331417e-01 -3.08048218e-01 9.45118248e-01 5.96853971e-01
3.15552384e-01 -1.36566782e+00 1.76697999e-01 4.31887209e-01
-3.01423043e-01 -1.09387207e+00 -9.59346771e-01 9.76216123e-02
-4.86075103e-01 -4.17212784e-01 -8.33060265e-01 -8.05755019e-01
2.27421392e-02 1.21526554e-01 8.21445167e-01 4.16451991e-02
-1.65044554e-02 3.39974165e-01 -4.93565261e-01 -1.25059381e-01
-1.06102419e+00 3.70476209e-02 7.36334026e-01 -5.73563933e-01
4.09480989e-01 -4.68515009e-01 -2.17804685e-01 3.64867419e-01
-4.43849206e-01 -3.28575939e-01 3.28162134e-01 1.00240588e+00
-8.21199864e-02 -1.08695127e-01 8.76811743e-01 -4.25505340e-01
1.20330465e+00 -3.09940457e-01 -3.64563707e-03 -3.42963524e-02
-7.31284261e-01 -2.53278822e-01 6.92684352e-01 -6.52565300e-01
-9.42425489e-01 -2.13715404e-01 -9.07638133e-01 -6.70305490e-02
-6.55432522e-01 1.64409176e-01 -4.31853920e-01 3.93523008e-01
8.09711158e-01 3.34342450e-01 3.18078518e-01 -7.44972944e-01
2.62281775e-01 1.54358184e+00 7.21333683e-01 -2.09081341e-02
2.27044970e-01 -4.27773982e-01 -1.19703162e+00 -1.60270178e+00
-2.37360019e-02 -5.39115846e-01 -5.54209873e-02 8.99468884e-02
5.93871713e-01 -7.73881793e-01 -4.39514309e-01 8.33467543e-01
-1.11790884e+00 -6.59750283e-01 -2.84019738e-01 8.23072314e-01
-6.59492612e-01 3.15981239e-01 -5.02701879e-01 -1.01353681e+00
-4.77677137e-01 -1.50117290e+00 6.85456455e-01 -1.73968077e-01
-8.60951066e-01 -5.25118351e-01 -1.82633385e-01 5.83098888e-01
7.73396790e-01 -8.05513322e-01 8.71734202e-01 -1.05667126e+00
3.86790067e-01 2.43149251e-02 2.95070589e-01 1.02482677e+00
6.05088949e-01 -4.60184723e-01 -8.42744768e-01 -3.76268595e-01
9.16608721e-02 -4.86805961e-02 5.50871611e-01 4.29975390e-01
6.29127145e-01 -6.93067908e-01 -5.54554909e-02 3.12650084e-01
3.53796095e-01 8.71486485e-01 5.70664465e-01 -2.58653432e-01
4.36938167e-01 6.29040301e-01 -5.38908280e-02 1.61360577e-01
-2.66925525e-02 1.16049135e+00 -3.38481158e-01 3.65596890e-01
-8.94346774e-01 -1.35375544e-01 1.07298183e+00 1.07223713e+00
3.66572976e-01 -5.74011624e-01 -1.08321536e+00 7.69137383e-01
-1.07509017e+00 -9.90622282e-01 -1.04560442e-01 2.33567262e+00
9.95163560e-01 1.36222377e-01 5.13922393e-01 4.47061032e-01
1.13194156e+00 -1.08153401e-02 -5.44530690e-01 -8.87442529e-01
3.97397317e-02 4.47694272e-01 1.64520204e-01 1.00039399e+00
-4.92336899e-01 1.25705302e+00 7.19582701e+00 7.79647768e-01
-1.14580810e+00 3.50899607e-01 3.38520586e-01 -4.78725523e-01
-6.59477487e-02 -3.20895880e-01 -7.88401663e-01 7.35330999e-01
1.76174796e+00 -1.94470555e-01 1.07935619e+00 6.46057785e-01
6.83462441e-01 2.66089976e-01 -9.10341501e-01 1.25545871e+00
9.31788534e-02 -6.65635824e-01 -2.51489282e-01 1.47544490e-02
1.19183913e-01 1.90786123e-01 1.51577935e-01 4.38196123e-01
1.40288025e-01 -1.15538847e+00 8.25773418e-01 1.48850113e-01
1.21265864e+00 -5.29337227e-01 2.54749358e-01 3.42358291e-01
-4.57205385e-01 -1.63427994e-01 9.31140333e-02 -2.92380393e-01
4.26456988e-01 4.32017684e-01 -1.41031289e+00 -3.37161928e-01
4.32184339e-01 3.15374956e-02 -9.85269770e-02 7.89747179e-01
-2.41103590e-01 1.27203178e+00 -5.92307627e-01 -9.39007178e-02
-1.74502373e-01 4.11695033e-01 9.42261875e-01 1.38882840e+00
5.50475776e-01 1.87281862e-01 -2.25006759e-01 4.80239540e-01
5.61919510e-02 1.35347590e-01 -6.14202797e-01 -4.19042885e-01
7.97476530e-01 3.57664675e-01 -5.07142060e-02 -2.75208861e-01
-3.23997468e-01 1.17512441e+00 1.18962869e-01 3.84343922e-01
-2.06824362e-01 -4.51430529e-01 1.13153028e+00 2.47933596e-01
-1.57401800e-01 -3.38459611e-01 -3.34909499e-01 -8.58011961e-01
1.27586648e-01 -1.42888784e+00 1.17613487e-01 -5.27911186e-01
-1.02206266e+00 7.30422497e-01 -2.74410456e-01 -6.88924968e-01
-8.23423028e-01 -7.13315487e-01 -5.89959800e-01 1.27249753e+00
-7.53043890e-01 -4.56745952e-01 2.61657059e-01 4.62449342e-01
1.15475178e+00 -5.68766057e-01 9.01699066e-01 4.19006497e-01
-6.55575871e-01 1.00959897e+00 -1.28500149e-01 -1.86218858e-01
5.82277417e-01 -1.16965532e+00 8.95351410e-01 9.80069876e-01
3.04153822e-02 7.35678792e-01 8.23954582e-01 -6.24428809e-01
-9.21009362e-01 -8.69313657e-01 1.26546979e+00 -2.16548577e-01
4.74666744e-01 -1.92389950e-01 -8.41437399e-01 6.10375106e-01
1.77735031e-01 -5.00097692e-01 6.40100121e-01 2.42480159e-01
-5.49758039e-02 2.09383462e-02 -1.29925144e+00 7.78699517e-01
1.51429439e+00 -8.01514804e-01 -8.92725468e-01 5.88135839e-01
1.02436006e+00 -1.91568077e-01 -5.93909383e-01 1.13755763e-01
2.75475860e-01 -7.67486334e-01 7.73746550e-01 -7.60191500e-01
-2.51863509e-01 1.66471064e-01 -2.46175334e-01 -2.03899145e+00
-3.45422417e-01 -1.13048697e+00 9.67121348e-02 1.24875963e+00
6.07537329e-01 -9.17679310e-01 1.39113348e-02 7.60851026e-01
-8.74711871e-01 5.45586161e-02 -1.26315796e+00 -1.14491141e+00
3.95108402e-01 -7.78671682e-01 4.62422520e-01 3.71979058e-01
6.46194875e-01 2.75354564e-01 -1.36005506e-01 1.81508705e-01
-2.19304711e-01 -9.60317075e-01 2.09460482e-01 -6.79066360e-01
-3.03729236e-01 -8.28846455e-01 -4.72770780e-01 -1.08962452e+00
2.70511001e-01 -9.89977956e-01 3.52947742e-01 -1.25244641e+00
-5.23737729e-01 -2.86124229e-01 1.14062339e-01 6.03018582e-01
-1.65039286e-01 -3.32292467e-01 2.67605543e-01 -5.92941931e-03
3.20525050e-01 4.68513578e-01 6.48478746e-01 -2.26904601e-01
-6.15611851e-01 4.21954513e-01 -7.82909036e-01 5.82174420e-01
1.10735738e+00 -1.88470110e-01 -5.80096185e-01 -4.68289465e-01
-6.43490553e-01 3.60826164e-01 7.11757094e-02 -1.15846300e+00
-9.15924236e-02 3.37512903e-02 -1.92076117e-01 -1.28012791e-01
4.35828567e-01 -2.29194522e-01 3.25242355e-02 5.39842010e-01
-4.91322994e-01 8.03442895e-02 4.03458476e-01 -2.68146638e-02
1.27759963e-01 -3.88362497e-01 8.39257836e-01 -1.93715066e-01
-3.46426278e-01 -2.64435589e-01 -1.09328306e+00 2.85959333e-01
3.89711559e-01 -3.79532613e-02 -5.41372597e-01 -7.30931401e-01
-1.16162038e+00 -3.10358405e-01 1.53634056e-01 6.95571899e-01
4.92895842e-01 -1.04222178e+00 -7.98860490e-01 4.22367036e-01
-3.19726206e-02 -6.28037512e-01 2.18578912e-02 7.41820157e-01
-2.87401527e-01 7.90510058e-01 1.52770504e-01 2.42614038e-02
-1.43845642e+00 1.86241895e-01 7.05793619e-01 3.25660974e-01
-6.74030542e-01 1.11123669e+00 -2.24675015e-01 -4.14135605e-01
5.19289196e-01 -6.54034376e-01 2.38570452e-01 -1.85099557e-01
7.90779889e-01 6.55279875e-01 7.45471537e-01 -8.83688152e-01
-5.20548224e-01 -2.11308122e-01 -4.28144544e-01 -6.57313585e-01
7.29287922e-01 -2.30123341e-01 4.22025263e-01 2.99055368e-01
9.30741787e-01 2.41440997e-01 -7.01042056e-01 8.29097256e-02
-2.44122326e-01 -2.09524050e-01 4.08298373e-01 -1.36403906e+00
-6.97245300e-01 8.32111597e-01 8.20094824e-01 4.28578407e-01
8.42733145e-01 1.29781112e-01 1.04796994e+00 2.58950561e-01
1.39792100e-01 -1.16731155e+00 -3.15228134e-01 7.33182847e-01
1.17004049e+00 -7.49255478e-01 -8.00053477e-01 -3.13249290e-01
-7.29994476e-01 7.56364226e-01 3.57643723e-01 3.80807161e-01
4.73054230e-01 3.21761101e-01 2.26868004e-01 1.86207607e-01
-5.46628237e-01 -5.33944249e-01 3.03545478e-03 9.76609647e-01
5.91648042e-01 5.79882383e-01 -6.32729411e-01 8.21042895e-01
-8.83080065e-01 -2.96059340e-01 3.29351842e-01 4.24118072e-01
-6.83820307e-01 -1.07098484e+00 -5.29066086e-01 7.45054781e-01
-3.81370723e-01 -4.71634626e-01 -5.23562193e-01 3.24600160e-01
-4.38736416e-02 1.72774434e+00 1.51872560e-01 -5.60831249e-01
7.82978892e-01 7.38318682e-01 1.10406332e-01 -9.72667575e-01
-7.71399915e-01 6.82297349e-02 6.77619696e-01 -3.13643426e-01
3.70692462e-01 -1.15883231e+00 -1.29837418e+00 -2.52069920e-01
-3.28580976e-01 -1.37250423e-01 5.74970484e-01 1.09356844e+00
2.61132628e-01 6.39421046e-01 2.32003137e-01 -3.30978990e-01
-9.74751830e-01 -1.60313225e+00 -6.34870887e-01 1.51161760e-01
6.73990667e-01 -4.62262452e-01 -7.78614879e-01 -5.55626005e-02] | [14.466484069824219, 6.458902835845947] |
46031044-9086-4e2e-966f-1c5db8d3f75d | context-aware-configuration-and-management-of | 2307.03126 | null | https://arxiv.org/abs/2307.03126v1 | https://arxiv.org/pdf/2307.03126v1.pdf | Context-Aware Configuration and Management of WiFi Direct Groups for Real Opportunistic Networks | Wi-Fi Direct is a promising technology for the support of device-to-device communications (D2D) on commercial mobile devices. However, the standard as-it-is is not sufficient to support the real deployment of networking solutions entirely based on D2D such as opportunistic networks. In fact, WiFi Direct presents some characteristics that could limit the autonomous creation of D2D connections among users' personal devices. Specifically, the standard explicitly requires the user's authorization to establish a connection between two or more devices, and it provides a limited support for inter-group communication. In some cases, this might lead to the creation of isolated groups of nodes which cannot communicate among each other. In this paper, we propose a novel middleware-layer protocol for the efficient configuration and management of WiFi Direct groups (WiFi Direct Group Manager, WFD-GM) to enable autonomous connections and inter-group communication. This enables opportunistic networks in real conditions (e.g., variable mobility and network size). WFD-GM defines a context function that takes into account heterogeneous parameters for the creation of the best group configuration in a specific time window, including an index of nodes' stability and power levels. We evaluate the protocol performances by simulating three reference scenarios including different mobility models, geographical areas and number of nodes. Simulations are also supported by experimental results related to the evaluation in a real testbed of the involved context parameters. We compare WFD-GM with the state-of-the-art solutions and we show that it performs significantly better than a Baseline approach in scenarios with medium/low mobility, and it is comparable with it in case of high mobility, without introducing additional overhead. | ['Franca Delmastro', 'Mattia Giovanni Campana', 'Valerio Arnaboldi'] | 2023-07-06 | null | null | null | null | ['management'] | ['miscellaneous'] | [-4.09171171e-02 3.55284750e-01 -3.11946809e-01 -6.04780875e-02
1.30368829e-01 -5.51413774e-01 7.06172526e-01 7.29298145e-02
-2.94076353e-01 1.02101541e+00 -3.34854454e-01 -8.34623992e-01
-7.13982046e-01 -1.14855456e+00 -1.81108907e-01 -6.95809543e-01
-4.47097063e-01 5.84414423e-01 8.27222884e-01 -1.03738405e-01
-7.77660683e-02 7.78582811e-01 -1.76430380e+00 -3.27062905e-01
5.17578542e-01 1.08099163e+00 4.80213523e-01 4.15428370e-01
-4.46246654e-01 6.15456887e-02 -9.32931364e-01 2.48440415e-01
1.09269403e-01 -5.93528859e-02 -6.10363007e-01 -8.82426351e-02
-4.62465733e-01 -3.99263442e-01 8.34154040e-02 2.61729240e-01
7.89133072e-01 -4.42114286e-02 4.11384791e-01 -1.50590527e+00
3.92966807e-01 4.72502679e-01 -1.33317903e-01 2.26593956e-01
7.66154587e-01 -1.31757706e-01 1.78133875e-01 -1.45041704e-01
9.94765818e-01 8.11281443e-01 2.89458096e-01 4.59429890e-01
-9.94366229e-01 -6.52458847e-01 1.60753474e-01 -5.49276453e-03
-1.62134600e+00 -5.93381882e-01 3.75146866e-01 -1.38824627e-01
6.76858842e-01 5.25635362e-01 5.85278034e-01 1.06159508e+00
-1.16665952e-01 -5.33723086e-02 5.19988716e-01 -4.63345736e-01
5.02467871e-01 4.16914582e-01 -1.15946993e-01 -8.72353539e-02
6.92266524e-01 -1.12303205e-01 -1.54282555e-01 -3.28401715e-01
8.41407955e-01 -4.68968928e-01 -4.22649920e-01 -4.81662989e-01
-1.08561385e+00 1.33234575e-01 4.34427559e-02 1.21476650e+00
-6.29582763e-01 -9.65433717e-02 5.49568608e-03 4.59461719e-01
-3.76123749e-02 -5.75759560e-02 -2.78155267e-01 -6.43995643e-01
-8.00155580e-01 8.65641311e-02 1.13701093e+00 1.24737585e+00
3.60796839e-01 -4.79929686e-01 1.14055254e-01 5.27939022e-01
6.38328433e-01 5.28178871e-01 9.75711420e-02 -8.36021960e-01
3.73417050e-01 2.23984838e-01 5.65011501e-01 -9.05055761e-01
-8.61198843e-01 -5.67162275e-01 -8.55281532e-01 3.66565824e-01
4.49606568e-01 -4.96977419e-01 -1.83018237e-01 1.75057435e+00
5.20615280e-01 4.28293765e-01 1.29382849e-01 3.62977594e-01
6.67619944e-01 4.86354262e-01 -9.96905267e-02 -6.97396040e-01
1.05584085e+00 -1.92098722e-01 -1.12361443e+00 1.30941048e-01
5.55176497e-01 -7.69506395e-01 6.22555256e-01 2.50176132e-01
-1.20210016e+00 -4.45835620e-01 -9.97246861e-01 7.95846581e-01
-3.97698790e-01 -1.03743359e-01 1.52410209e-01 1.33073652e+00
-1.20344412e+00 3.50136131e-01 -6.37219250e-01 -1.00669599e+00
-1.02415025e-01 1.05076981e+00 -1.58837780e-01 3.42616946e-01
-1.07883489e+00 4.90210712e-01 2.22436100e-01 -1.41525835e-01
-2.79359698e-01 -2.62266099e-01 -4.97033894e-02 3.91748697e-02
2.73752928e-01 -7.10549772e-01 8.89173508e-01 -4.08810228e-01
-1.74835443e+00 4.82021689e-01 -9.56768617e-02 -2.01847136e-01
7.44736016e-01 2.92083979e-01 -1.03939545e+00 3.29135247e-02
6.26390651e-02 2.20804647e-01 5.35042696e-02 -1.34404731e+00
-8.83851886e-01 4.50222790e-02 6.90008044e-01 -7.09694996e-02
-2.97204673e-01 -2.28605330e-01 -8.60340834e-01 -6.63777664e-02
-4.97406535e-02 -1.11872220e+00 -8.11253637e-02 -1.67546600e-01
-4.24763173e-01 -9.21630263e-02 8.38989794e-01 2.28565231e-01
1.60743642e+00 -2.10545754e+00 -2.21939608e-01 8.84658873e-01
-3.03388499e-02 3.57547671e-01 1.68090478e-01 4.21617001e-01
4.55254078e-01 3.75039756e-01 2.89659083e-01 -4.20115411e-01
2.31334157e-02 4.47171926e-01 5.95424831e-01 1.84257939e-01
-8.21001172e-01 -5.10988757e-02 -7.24220514e-01 -3.39767188e-01
4.77829427e-01 5.06022394e-01 -2.96701491e-01 9.91291255e-02
2.36322582e-01 6.33825004e-01 -7.43962109e-01 2.46895373e-01
1.25130892e+00 6.73786644e-03 8.41964066e-01 -2.82102264e-02
-6.62133574e-01 2.70606399e-01 -1.71444690e+00 1.39136696e+00
-8.08291256e-01 1.22529685e-01 4.90520597e-01 -8.25981915e-01
9.48867023e-01 7.34792233e-01 7.87722230e-01 -4.96535391e-01
4.17942911e-01 6.29600704e-01 -8.37077349e-02 -5.83976626e-01
4.11587954e-02 4.06400889e-01 3.27842422e-02 4.42348927e-01
-1.71882987e-01 4.05209869e-01 3.02050203e-01 -8.55369195e-02
1.21765816e+00 -1.08856067e-01 2.61609793e-01 -3.98229003e-01
9.33486879e-01 -7.57798791e-01 2.37579077e-01 9.77679968e-01
-1.11071048e-02 -2.99470890e-02 3.47656518e-01 2.07268775e-01
-2.60045022e-01 -1.13354945e+00 -2.37478435e-01 4.72471207e-01
7.96028256e-01 -4.38739657e-01 -5.66611171e-01 -6.48098767e-01
-1.23660356e-01 6.93776906e-01 -1.56788588e-01 1.63353175e-01
-3.84515107e-01 -5.52476048e-01 5.74501932e-01 -3.41494590e-01
6.25131726e-01 -6.29656255e-01 -5.34018636e-01 7.33772159e-01
-1.58628285e-01 -1.44792712e+00 1.45678386e-01 -8.47795010e-02
-7.29502201e-01 -9.68748569e-01 -5.54170132e-01 -6.07312620e-01
2.99564660e-01 5.03843486e-01 9.21774924e-01 3.48413497e-01
6.23331785e-01 4.68434602e-01 -4.07650560e-01 -1.25549808e-01
-5.15086055e-01 7.11578429e-01 1.08450808e-01 1.13013476e-01
2.67309040e-01 -1.13626325e+00 -5.81594110e-01 1.18218720e+00
-6.33291602e-01 -3.41517180e-01 3.15117031e-01 -2.42479697e-01
3.55226427e-01 3.20506722e-01 8.35795879e-01 -6.19337440e-01
6.10342264e-01 -9.57897961e-01 -5.02281010e-01 -2.33483315e-03
-3.96301121e-01 -4.06338483e-01 3.65706235e-01 -3.02044213e-01
-8.92663836e-01 -4.17856663e-01 -6.09595716e-01 3.42775583e-01
-5.31071007e-01 2.00768277e-01 -7.53720045e-01 -2.86352515e-01
5.01953423e-01 -3.19690168e-01 -1.86746046e-01 -8.08930457e-01
-1.52939662e-01 1.15193307e+00 -1.53733073e-02 -3.59463483e-01
8.87570977e-01 5.89031816e-01 2.20078826e-01 -1.15850580e+00
2.12048620e-01 -3.64734471e-01 -3.96139085e-01 -4.29117799e-01
5.81801951e-01 -6.54711366e-01 -9.47532833e-01 3.09848428e-01
-1.02357423e+00 -3.75194728e-01 1.06594577e-01 6.17062747e-01
-1.41552314e-01 1.79369450e-01 2.52147317e-02 -7.65299499e-01
1.57058120e-01 -1.07903349e+00 4.83759969e-01 4.10657763e-01
-2.88467497e-01 -1.05289817e+00 -2.34914958e-01 -9.27024856e-02
9.01557148e-01 4.15441543e-01 5.02421796e-01 -2.85990000e-01
-4.89700556e-01 -9.26768407e-02 5.22047691e-02 -2.67498910e-01
5.10132313e-01 -7.64829442e-02 -5.98410666e-01 -1.66729122e-01
-3.44848901e-01 8.01811635e-01 -2.97937632e-01 5.70679367e-01
8.44670951e-01 5.35903685e-02 -1.10492051e+00 3.07950824e-01
1.31754661e+00 5.71278811e-01 8.50111604e-01 6.12960339e-01
9.41726193e-02 4.90999848e-01 8.63678634e-01 5.66742599e-01
3.26420933e-01 1.41837263e+00 7.71056771e-01 -1.38523370e-01
-7.70457238e-02 2.66454905e-01 1.76627398e-01 1.99220642e-01
-4.64238077e-01 -1.11404216e+00 -3.75966460e-01 3.00601900e-01
-1.63406992e+00 -8.28631580e-01 -5.86203039e-01 2.58234525e+00
-4.76691648e-02 8.52860630e-01 5.48224509e-01 5.88132083e-01
9.32305634e-01 -1.02960750e-01 -1.98079310e-02 -4.69301611e-01
2.91791912e-02 1.25539720e-01 6.97822094e-01 6.91430986e-01
-6.94570720e-01 4.00045216e-01 5.42972946e+00 6.88299596e-01
-1.08515096e+00 4.50098813e-01 -2.61985123e-01 -7.86693245e-02
-3.92428190e-01 -2.55777806e-01 -8.05344284e-01 7.53519535e-01
1.11168826e+00 -1.35567695e-01 7.37050474e-02 3.96981478e-01
1.01800919e+00 -2.31761426e-01 -5.18881619e-01 7.87505269e-01
-7.24169612e-01 -1.30389369e+00 -2.47208863e-01 6.84871197e-01
4.27858323e-01 -5.27274758e-02 -5.19638538e-01 -3.48686725e-01
-4.67656761e-01 -2.64492393e-01 5.83617985e-01 2.82309234e-01
8.18287194e-01 -8.71728778e-01 9.77243900e-01 3.43492270e-01
-1.52565944e+00 -3.81072946e-02 5.53770028e-02 -1.45396795e-02
8.61386657e-01 8.86516929e-01 -4.87357080e-01 8.66706252e-01
4.83751953e-01 2.27642342e-01 -2.92503648e-02 1.28596568e+00
-3.42139800e-04 3.33761662e-01 -9.47153628e-01 -1.69855624e-01
9.42108482e-02 -8.15217793e-02 8.40102136e-01 9.30577278e-01
9.46850955e-01 -1.51481345e-01 -3.18909585e-01 3.57721955e-01
2.06513122e-01 2.08420113e-01 -5.85358679e-01 6.06586933e-01
1.18366909e+00 8.95816684e-01 -8.32641244e-01 -1.79545149e-01
-4.43843871e-01 6.50593460e-01 -7.16026247e-01 3.22763234e-01
-8.40754569e-01 -6.60277843e-01 1.00984299e+00 9.06238139e-01
3.42208713e-01 -6.42040431e-01 2.21447617e-01 -5.09588838e-01
-4.59398217e-02 -3.31278086e-01 -3.78917865e-02 -2.82502532e-01
-4.76624697e-01 6.71771944e-01 6.51327968e-02 -1.54807413e+00
-4.04207945e-01 -2.01491997e-01 -6.45638824e-01 5.73019624e-01
-1.40408051e+00 -7.51768410e-01 -6.09770715e-01 8.30561697e-01
-5.81444018e-02 -6.51106462e-02 9.62645233e-01 1.08335114e+00
-1.19464964e-01 7.07283139e-01 2.21523002e-01 -7.24541068e-01
4.12237287e-01 -8.32499564e-01 6.84763119e-02 6.35856807e-01
-5.08984029e-01 5.49614727e-01 7.64586389e-01 -4.45554256e-01
-9.60275114e-01 -5.57899475e-01 1.12758780e+00 3.40056211e-01
1.82686865e-01 -5.27338326e-01 -3.79101455e-01 3.76625746e-01
-9.26344097e-02 -6.75720051e-02 7.04559982e-01 2.50506699e-02
8.26792121e-01 -4.09401327e-01 -1.56970572e+00 5.05561471e-01
1.57255495e+00 -2.33620126e-02 2.88214654e-01 6.06093407e-02
3.88389468e-01 -1.84202358e-01 -1.09277725e+00 2.13525951e-01
7.28852153e-01 -1.33919871e+00 1.04321456e+00 4.55364883e-01
-8.79295349e-01 -7.86848962e-01 -9.06422809e-02 -8.32273960e-01
-1.56956434e-01 -9.94274735e-01 1.55580998e-03 1.76019120e+00
4.71610546e-01 -7.33928561e-01 6.32373452e-01 3.02965760e-01
1.63085803e-01 -2.83642828e-01 -1.34402120e+00 -1.04945171e+00
-5.09448111e-01 -6.42917216e-01 1.24339020e+00 6.06059372e-01
-1.48044661e-01 6.66872710e-02 -1.84345677e-01 5.79746187e-01
3.24325919e-01 -6.11780822e-01 1.18775868e+00 -1.59181511e+00
-2.22782299e-01 -3.59037429e-01 -6.72743559e-01 -1.33196378e+00
-1.89810678e-01 -3.16915423e-01 -5.17882288e-01 -1.74817312e+00
-1.07909763e+00 -1.35721922e+00 -7.67494366e-02 -1.21880984e-02
8.20028543e-01 6.36504740e-02 4.66719121e-02 6.65813535e-02
-3.43456805e-01 -1.16916150e-01 9.10127461e-01 3.50499958e-01
-8.00421953e-01 9.23867583e-01 -3.22645962e-01 4.77418691e-01
1.02609479e+00 -3.80389154e-01 -7.54053950e-01 -1.30255073e-01
5.46174794e-02 2.10445389e-01 4.71359231e-02 -1.59923267e+00
1.40736982e-01 -6.71945289e-02 -3.14436644e-01 -2.06355631e-01
2.91473567e-01 -1.58043194e+00 8.74441147e-01 6.68251336e-01
2.69489586e-01 -4.03056949e-01 6.27046078e-02 3.93217742e-01
2.99152970e-01 -2.18867958e-01 5.16406417e-01 4.25834686e-01
-4.80423152e-01 6.82368428e-02 -1.13899326e+00 -7.39688993e-01
1.32006145e+00 -5.83040833e-01 -1.04916550e-01 -5.91048360e-01
-1.06054735e+00 1.45840451e-01 5.90438783e-01 4.36190188e-01
-8.49878415e-02 -1.13961852e+00 -4.84847240e-02 -8.50701034e-02
-8.26544315e-02 -4.94536906e-01 2.45889261e-01 8.75447750e-01
-3.88865232e-01 3.30947429e-01 -3.62586677e-01 -4.89202708e-01
-1.55512464e+00 1.47823498e-01 6.23567164e-01 -2.18768552e-01
-1.91942945e-01 2.42565677e-01 -4.22788918e-01 -1.86545402e-01
3.48960489e-01 -2.40784049e-01 -7.13772953e-01 6.98395520e-02
4.60427165e-01 6.54945195e-01 2.13058695e-01 -6.45257175e-01
-7.43751347e-01 8.28776062e-01 7.09748507e-01 -3.14344049e-01
1.04933500e+00 -9.72159445e-01 1.66116893e-01 9.69679356e-02
8.88316989e-01 4.64564860e-01 -5.48339009e-01 1.84762880e-01
1.78457964e-02 -3.26571316e-01 6.12804517e-02 -5.72680473e-01
-1.01767015e+00 2.14538321e-01 1.14975798e+00 9.01841521e-01
1.06449378e+00 -8.80481005e-02 5.96401274e-01 6.62311763e-02
1.10206091e+00 -7.57468879e-01 -5.70448637e-01 6.86494038e-02
2.56205976e-01 -5.08045495e-01 -2.62936771e-01 -8.89144242e-01
2.57132292e-01 1.21368337e+00 3.18144828e-01 3.51708412e-01
1.27434039e+00 4.40287441e-01 3.02896295e-02 -1.18133672e-01
-2.99515933e-01 -5.51943421e-01 -3.94527346e-01 1.10830927e+00
2.70761073e-01 1.89412739e-02 -9.67771471e-01 4.63987380e-01
-1.31675694e-02 5.22837520e-01 5.96093357e-01 1.03669262e+00
-5.37867904e-01 -1.99004877e+00 -4.11135882e-01 3.79164338e-01
-3.84043813e-01 6.96251214e-01 1.89680636e-01 1.11402822e+00
9.22190249e-01 1.64888871e+00 1.54959604e-01 -4.86069798e-01
6.13534689e-01 -3.76438916e-01 3.40171367e-01 -2.14067623e-01
-3.66559178e-01 6.09406456e-02 7.50934184e-01 -6.02879465e-01
-8.56773734e-01 -6.26834095e-01 -1.36658537e+00 -7.01494038e-01
-3.24987441e-01 4.51249659e-01 9.56046760e-01 9.28165734e-01
5.88178992e-01 5.75811148e-01 7.31801331e-01 -9.54259813e-01
6.18962228e-01 -6.44733548e-01 -9.42209601e-01 6.72292635e-02
1.75788358e-01 -1.09754050e+00 -6.00106955e-01 -5.70368230e-01] | [6.201981544494629, 1.3398306369781494] |
b92751af-4b1c-460d-b4e9-a9642644b0be | deep-structured-learning-for-mass | 1410.7454 | null | http://arxiv.org/abs/1410.7454v2 | http://arxiv.org/pdf/1410.7454v2.pdf | Deep Structured learning for mass segmentation from Mammograms | In this paper, we present a novel method for the segmentation of breast
masses from mammograms exploring structured and deep learning. Specifically,
using structured support vector machine (SSVM), we formulate a model that
combines different types of potential functions, including one that classifies
image regions using deep learning. Our main goal with this work is to show the
accuracy and efficiency improvements that these relatively new techniques can
provide for the segmentation of breast masses from mammograms. We also propose
an easily reproducible quantitative analysis to as- sess the performance of
breast mass segmentation methodologies based on widely accepted accuracy and
running time measurements on public datasets, which will facilitate further
comparisons for this segmentation problem. In particular, we use two publicly
available datasets (DDSM-BCRP and INbreast) and propose the computa- tion of
the running time taken for the methodology to produce a mass segmentation given
an input image and the use of the Dice index to quantitatively measure the
segmentation accuracy. For both databases, we show that our proposed
methodology produces competitive results in terms of accuracy and running time. | ['Gustavo Carneiro', 'Andrew P. Bradley', 'Neeraj Dhungel'] | 2014-10-27 | null | null | null | null | ['mass-segmentation-from-mammograms'] | ['medical'] | [ 3.23712349e-01 3.39339435e-01 -1.66540816e-01 -8.41133773e-01
-9.16545510e-01 -2.16664582e-01 4.93613362e-01 6.53818190e-01
-5.26414037e-01 4.57004905e-01 -2.99816012e-01 -5.84088922e-01
-3.27784151e-01 -9.46600020e-01 -6.33244932e-01 -7.47927606e-01
-3.24410945e-01 8.51784825e-01 2.65720814e-01 1.00879580e-01
1.32984981e-01 7.61400521e-01 -1.09689987e+00 2.87421525e-01
7.38721848e-01 1.39297390e+00 9.52412263e-02 7.35891283e-01
-2.99265534e-02 7.43530929e-01 -3.32447857e-01 -5.71904421e-01
2.72966802e-01 -5.04782140e-01 -1.19488215e+00 3.59832197e-01
3.21680635e-01 -3.23801488e-01 -1.61218777e-01 9.03062761e-01
3.38705659e-01 -4.19003665e-01 1.09280014e+00 -8.98775995e-01
-1.07191034e-01 7.07362771e-01 -6.58106208e-01 4.81249303e-01
-8.20234492e-02 -1.44970417e-01 6.49724185e-01 -4.86704826e-01
6.52177036e-01 9.53396261e-01 8.91923249e-01 2.99585968e-01
-1.09789109e+00 -2.44597912e-01 -5.19679010e-01 1.64697364e-01
-1.15898430e+00 -3.17857742e-01 3.16925287e-01 -5.33119917e-01
3.95149350e-01 5.97989798e-01 7.15202570e-01 3.60360175e-01
5.59393466e-01 9.54495370e-01 1.14338350e+00 -5.60451806e-01
5.83762050e-01 3.51007245e-02 3.69363189e-01 9.78427410e-01
4.75773215e-01 -1.30175963e-01 1.55579150e-01 -2.23610193e-01
5.12373984e-01 -1.06248476e-01 1.18056163e-01 -6.26967251e-01
-1.14925301e+00 1.00293183e+00 5.64562619e-01 5.93264163e-01
-4.04641211e-01 9.05926377e-02 5.27280927e-01 -4.21202742e-02
6.84836626e-01 1.96696505e-01 -2.07001477e-01 3.21186393e-01
-1.68395317e+00 1.70022577e-01 8.71336997e-01 3.86572212e-01
3.46693963e-01 -4.39220995e-01 -1.99416026e-01 5.89522243e-01
4.52880114e-01 5.20630479e-01 6.64474845e-01 -7.73031592e-01
1.49196073e-01 5.63850820e-01 -1.49825484e-01 -1.28468001e+00
-9.87204075e-01 -2.43361607e-01 -1.01764941e+00 2.58722622e-02
6.19818926e-01 6.75864220e-02 -1.24193215e+00 1.07607067e+00
3.63902718e-01 -2.66484976e-01 6.71076626e-02 6.60408556e-01
1.01692188e+00 3.63783091e-01 1.40855089e-01 -2.24226311e-01
1.20369244e+00 -7.24416673e-01 -5.93975902e-01 2.06895143e-01
9.73646522e-01 -3.34589124e-01 3.14767301e-01 3.66838694e-01
-1.34922016e+00 -4.83443081e-01 -1.11439419e+00 1.49062410e-01
-2.81712562e-01 3.77548426e-01 6.31731153e-01 7.30644584e-01
-1.03757167e+00 8.51509631e-01 -1.63969743e+00 -4.06749099e-01
8.69731903e-01 5.20720005e-01 -1.78909495e-01 1.81750208e-01
-6.64051592e-01 9.46335375e-01 5.20828605e-01 1.61158200e-02
-8.27831268e-01 -5.29419005e-01 -8.79928052e-01 -7.62075111e-02
8.40758011e-02 -3.53211433e-01 1.29107177e+00 -1.05034661e+00
-9.88799930e-01 1.42310262e+00 1.58665150e-01 -1.08700144e+00
9.30992126e-01 2.97020316e-01 5.41910976e-02 5.94665825e-01
6.53555095e-02 9.29333270e-01 4.81853753e-01 -1.12227392e+00
-5.72774291e-01 -5.74285388e-01 -5.11002481e-01 -2.20542058e-01
-2.04060346e-01 -8.35072529e-03 -3.32849145e-01 -4.62175190e-01
5.85040689e-01 -7.98143268e-01 -5.12682974e-01 2.35967353e-01
-3.65869075e-01 4.33205850e-02 6.16930664e-01 -1.05620778e+00
9.03614163e-01 -1.92062211e+00 1.49919629e-01 4.93976593e-01
4.42152470e-01 1.93064570e-01 1.56683639e-01 -1.04650848e-01
1.16514750e-01 1.83682144e-01 -8.28616679e-01 -2.35862732e-01
-1.74879432e-01 2.04260528e-01 4.56992269e-01 8.21297944e-01
2.15270534e-01 1.24444413e+00 -6.92266464e-01 -9.83607829e-01
2.86154151e-01 2.38408938e-01 -2.16066137e-01 -7.56872147e-02
-8.48400295e-02 1.20344989e-01 -2.68302292e-01 7.09699690e-01
9.05833483e-01 -4.48616356e-01 1.92645013e-01 -1.29721284e-01
2.33509347e-01 -3.00129622e-01 -8.44692647e-01 1.44733357e+00
-1.59548402e-01 5.43836594e-01 8.54853988e-02 -1.48750830e+00
9.10547197e-01 4.05745693e-02 9.57634211e-01 -5.83615482e-01
4.89652127e-01 4.74511713e-01 7.45218098e-02 -4.63758260e-01
2.03008652e-01 -1.46180078e-01 2.65322059e-01 4.47550833e-01
2.33996093e-01 -3.68715614e-01 4.60875511e-01 2.06036717e-02
1.10706544e+00 -2.35131398e-01 5.73106945e-01 -7.41304100e-01
5.22734463e-01 3.51245314e-01 -6.08498901e-02 5.98976910e-01
-3.76111478e-01 6.28568649e-01 7.04097331e-01 -5.13058782e-01
-9.47709680e-01 -8.47694039e-01 -6.14318311e-01 6.58802032e-01
9.02159512e-02 1.29445776e-01 -9.41212118e-01 -9.29451108e-01
1.10747404e-01 3.69965792e-01 -9.42095578e-01 1.12678319e-01
-6.44752562e-01 -1.34006488e+00 4.59781975e-01 6.09315574e-01
5.65532684e-01 -9.82108355e-01 -1.09369993e+00 5.68737760e-02
-1.18459798e-02 -8.27459276e-01 1.18975408e-01 4.57973689e-01
-1.24076164e+00 -1.22153366e+00 -1.17495835e+00 -8.69642794e-01
7.07567930e-01 -2.72318035e-01 1.17228866e+00 2.89552480e-01
-6.93118691e-01 1.09559581e-01 -3.42037946e-01 -3.68138283e-01
-8.91817987e-01 2.20278844e-01 -5.76793849e-01 -1.08960122e-01
2.57770121e-01 3.36356610e-02 -6.72720730e-01 2.65577286e-01
-1.20932400e+00 2.27713585e-01 6.82340145e-01 6.48917139e-01
8.52215290e-01 2.28038318e-02 3.91158134e-01 -8.68918121e-01
2.27501035e-01 -7.40937114e-01 -5.78266978e-01 1.86392680e-01
-5.46405911e-01 -1.23243243e-01 1.88264381e-02 -1.18133761e-02
-6.50230408e-01 1.87285528e-01 -3.57554048e-01 1.52871445e-01
-1.13389157e-01 6.31244540e-01 2.80543536e-01 -2.91548193e-01
6.87253654e-01 2.26301208e-01 2.22112760e-01 -3.41760486e-01
5.83358146e-02 6.41958356e-01 6.51188612e-01 -2.79464237e-02
3.57411355e-01 8.22610199e-01 3.71987849e-01 -5.64545691e-01
-6.53143764e-01 -4.85416144e-01 -9.01583076e-01 -1.13869935e-01
1.10133469e+00 -3.50727320e-01 -2.22522825e-01 4.72664028e-01
-7.58669913e-01 -3.03806633e-01 -3.97217572e-01 3.43220890e-01
-7.98883498e-01 5.13866305e-01 -6.86092675e-01 -5.42935073e-01
-6.04812622e-01 -1.31665802e+00 1.22876406e+00 8.12852457e-02
-5.85910603e-02 -1.19199646e+00 -7.58827776e-02 4.18124735e-01
4.71058220e-01 5.98621368e-01 8.59414220e-01 -1.04003394e+00
-1.06454909e-01 -3.58126670e-01 -5.02150118e-01 3.96874964e-01
5.68410829e-02 -1.70482304e-02 -6.24874830e-01 -2.95978934e-01
1.76012963e-01 -2.14945376e-01 1.06313014e+00 1.04599655e+00
1.57122910e+00 -2.49204808e-04 -6.43910885e-01 6.92815542e-01
1.48947096e+00 2.79678553e-01 5.93462348e-01 4.72400039e-01
3.09812337e-01 5.47965109e-01 6.28790379e-01 2.86222935e-01
9.28360373e-02 3.02555144e-01 5.02210081e-01 -6.26341343e-01
-1.38349414e-01 2.59872437e-01 -3.62674594e-01 4.76054728e-01
4.77460818e-03 -6.16985410e-02 -1.24576199e+00 6.03187025e-01
-1.61923981e+00 -6.06584907e-01 -1.04649261e-01 1.84329879e+00
7.40290761e-01 9.88244042e-02 2.98321694e-01 4.42949116e-01
5.04978240e-01 -9.25390124e-02 -3.58894408e-01 -3.04333657e-01
8.87787044e-02 4.84359562e-01 7.99409211e-01 2.75493383e-01
-1.47709036e+00 4.35678750e-01 7.17453575e+00 8.02403152e-01
-1.25958622e+00 1.02225937e-01 1.34894919e+00 1.36823267e-01
1.01701751e-01 -5.91505527e-01 -5.28297901e-01 3.16543072e-01
1.05529892e+00 1.71954383e-03 -3.44875127e-01 7.67998993e-01
-2.91318796e-03 -4.25796360e-01 -1.23683679e+00 6.49895906e-01
2.04232018e-02 -1.46581626e+00 -1.94482803e-01 6.64122105e-02
7.80501783e-01 8.65550190e-02 1.19122555e-02 -1.05602160e-01
-1.26742050e-01 -1.25181437e+00 5.45865476e-01 4.63706762e-01
6.11903787e-01 -5.95346868e-01 1.23478007e+00 2.55703300e-01
-6.87009394e-01 1.53175384e-01 -2.27237225e-01 4.66040999e-01
-1.57749787e-01 6.93018854e-01 -1.15691626e+00 6.16680562e-01
5.38005054e-01 4.57774132e-01 -9.69405770e-01 1.41005504e+00
5.16770840e-01 6.39861047e-01 -3.04989934e-01 -1.02197103e-01
3.68571788e-01 -2.16386542e-02 8.55416730e-02 1.44247890e+00
3.58167499e-01 -1.52492791e-01 1.67040065e-01 9.13917243e-01
-2.03224458e-02 4.89175200e-01 -1.29557282e-01 -8.90299007e-02
-1.86430290e-01 1.36700463e+00 -1.48208344e+00 -4.56177473e-01
-9.84131470e-02 7.45362759e-01 -1.11811392e-01 -1.96222275e-01
-7.95621097e-01 -1.55866116e-01 -2.08349526e-01 2.84003973e-01
3.02948922e-01 1.00580394e-01 -5.39632618e-01 -6.09204054e-01
-1.56449210e-02 -6.66094065e-01 4.28460002e-01 -3.45654041e-01
-1.02007484e+00 5.38294971e-01 3.69118929e-01 -6.92096829e-01
-1.71842083e-01 -6.82907462e-01 -3.16398948e-01 4.83618200e-01
-1.39219189e+00 -9.83648539e-01 -3.10361415e-01 7.21500367e-02
4.43881541e-01 -1.15681879e-01 5.98170996e-01 7.72131309e-02
-2.37972602e-01 5.23627102e-01 3.11602235e-01 1.94819748e-01
1.34243995e-01 -1.40344524e+00 3.18416446e-01 4.25376326e-01
7.94119462e-02 -2.04035699e-01 5.90745211e-01 -5.75122774e-01
-9.81729209e-01 -1.07202327e+00 6.21038496e-01 -2.52032608e-01
4.90965486e-01 8.50789025e-02 -7.51877964e-01 5.05605817e-01
-1.13140531e-01 2.47403219e-01 5.72234988e-01 -5.05520046e-01
4.08974320e-01 1.50669947e-01 -1.69124496e+00 -1.44988783e-02
5.41385353e-01 2.55810529e-01 -5.18689096e-01 6.03487313e-01
2.76688218e-01 -5.02423048e-01 -1.00947654e+00 9.43658054e-01
4.01359737e-01 -1.11541843e+00 7.52932072e-01 -3.29750597e-01
7.19487011e-01 2.48428375e-01 -1.97326601e-01 -8.99398804e-01
6.47256076e-02 -4.32628542e-02 7.10752513e-03 5.24683416e-01
4.36306000e-01 -3.82109106e-01 1.04780185e+00 2.57711202e-01
1.03427395e-01 -1.31807017e+00 -9.89515960e-01 -5.90327561e-01
5.78436852e-01 -2.46956512e-01 3.71468991e-01 5.66307425e-01
-3.13148320e-01 -5.52567363e-01 3.45757544e-01 -1.42519742e-01
6.33431494e-01 1.28165796e-01 3.62679482e-01 -1.25900447e+00
-1.49602279e-01 -5.33765316e-01 -9.20693934e-01 -6.21698380e-01
-4.75083888e-02 -1.13054597e+00 8.10637474e-02 -1.68103278e+00
6.55855238e-01 -5.71195364e-01 -2.79126644e-01 3.46470207e-01
4.92574163e-02 4.42319661e-01 -3.16009820e-02 1.78220004e-01
-6.16181552e-01 2.57677902e-02 1.37368703e+00 -3.45435172e-01
-5.96748553e-02 2.41430759e-01 -3.74267310e-01 8.78741980e-01
8.69434655e-01 -5.09362638e-01 3.03750057e-02 -2.31599227e-01
-1.28785476e-01 3.10697496e-01 4.53181535e-01 -1.23756194e+00
-1.42182231e-01 4.86210734e-02 7.18038797e-01 -9.27256525e-01
1.65744592e-02 -6.03597999e-01 -1.05388463e-01 1.06705332e+00
-6.01074278e-01 -3.30824614e-01 3.21245939e-01 4.03341025e-01
-1.74419671e-01 -6.18905663e-01 1.15844142e+00 -2.22350433e-01
-2.55221635e-01 2.54558533e-01 -3.35847795e-01 -1.75385684e-01
1.31828761e+00 -1.93498224e-01 1.90500081e-01 -1.24940731e-01
-1.00319493e+00 1.65669098e-01 3.35991263e-01 -1.20965429e-01
4.00557369e-01 -1.09107828e+00 -1.04066169e+00 1.43304244e-01
-8.65808949e-02 -1.52107596e-01 -3.28135490e-02 1.20831299e+00
-1.26661706e+00 5.13289630e-01 -3.50981116e-01 -9.69858527e-01
-1.52784169e+00 4.18897271e-01 7.80196428e-01 -5.93528211e-01
-6.29109979e-01 9.12692726e-01 -6.06209524e-02 -3.44604731e-01
3.02385747e-01 -7.68151999e-01 -1.74333751e-01 7.17764115e-03
3.65112722e-01 3.99633914e-01 5.26102841e-01 -6.53971434e-01
-3.32227468e-01 2.46657550e-01 -3.25500406e-02 -7.08500203e-03
1.30707037e+00 1.70988292e-01 -2.39142388e-01 1.95055574e-01
1.42089868e+00 -5.63853800e-01 -7.01199114e-01 -7.23303854e-02
2.48082802e-01 -2.06425533e-01 4.05891269e-01 -8.27108443e-01
-1.41284919e+00 8.45061243e-01 1.24227762e+00 5.63828707e-01
1.17899692e+00 3.28920156e-01 7.62159467e-01 3.20300221e-01
1.00239545e-01 -8.93750012e-01 -2.38099068e-01 -1.31363392e-01
6.81814194e-01 -1.67896831e+00 2.31012404e-01 -4.36276406e-01
-3.77182007e-01 1.29774570e+00 2.13317037e-01 -1.09630190e-01
8.91288221e-01 3.15610051e-01 2.70612221e-02 -5.08150339e-01
-1.06300578e-01 -1.01486489e-01 4.92114395e-01 4.18920010e-01
5.08640587e-01 3.09621036e-01 -9.00401294e-01 4.34278905e-01
-1.64530486e-01 7.19790012e-02 2.47420669e-01 1.06403148e+00
-6.91010952e-01 -7.02420115e-01 -4.28735405e-01 9.50363457e-01
-7.91694939e-01 3.75868559e-01 -5.06023705e-01 8.95205379e-01
1.93297386e-01 7.07572520e-01 1.51153103e-01 5.86425550e-02
-2.30377182e-01 -1.11491062e-01 6.37184024e-01 -4.84402806e-01
-4.70913440e-01 7.04579242e-03 -1.60851777e-01 -3.36371869e-01
-6.33427382e-01 -6.09254777e-01 -1.39360809e+00 7.04624802e-02
-5.03702164e-01 -1.13592045e-02 9.60840940e-01 9.90097284e-01
-2.45101184e-01 5.30679226e-01 4.70965773e-01 -9.48104143e-01
-8.03398371e-01 -9.51409757e-01 -7.00182378e-01 4.37567681e-01
1.96232229e-01 -4.40040737e-01 -2.96213359e-01 5.80605939e-02] | [14.982802391052246, -2.4754631519317627] |
d5e58f7d-5bb9-4d5d-a601-df2c36b9f613 | early-diagnosis-of-lung-cancer-using-computer | 2107.12205 | null | https://arxiv.org/abs/2107.12205v1 | https://arxiv.org/pdf/2107.12205v1.pdf | Early Diagnosis of Lung Cancer Using Computer Aided Detection via Lung Segmentation Approach | Lung cancer begins in the lungs and leading to the reason of cancer demise amid population in the creation. According to the American Cancer Society, which estimates about 27% of the deaths because of cancer. In the early phase of its evolution, lung cancer does not cause any symptoms usually. Many of the patients have been diagnosed in a developed phase where symptoms become more prominent, that results in poor curative treatment and high mortality rate. Computer Aided Detection systems are used to achieve greater accuracies for the lung cancer diagnosis. In this research exertion, we proposed a novel methodology for lung Segmentation on the basis of Fuzzy C-Means Clustering, Adaptive Thresholding, and Segmentation of Active Contour Model. The experimental results are analysed and presented. | ['Chaitra K M', 'Mustafa Basthikodi', 'Ananth Prabhu G', 'Abhir Bhandary'] | 2021-07-23 | null | null | null | null | ['lung-cancer-diagnosis'] | ['medical'] | [-3.31098102e-02 1.49288222e-01 -2.48518437e-01 2.10798398e-01
-2.54311651e-01 -1.19270287e-01 4.65255469e-01 3.96634877e-01
-5.64380169e-01 7.22277820e-01 6.12905696e-02 -2.39304110e-01
8.48447625e-03 -9.22713280e-01 2.11028770e-01 -8.10658455e-01
3.33963066e-01 8.85527611e-01 6.02604151e-01 1.86639443e-01
8.66666958e-02 9.24669266e-01 -1.00389266e+00 -2.12965921e-01
9.18370187e-01 5.50281584e-01 3.38078856e-01 6.92170143e-01
-4.73735750e-01 4.81280506e-01 -3.34307432e-01 1.17360495e-01
2.69258618e-02 -7.29988635e-01 -6.74208462e-01 1.27056077e-01
-3.02215785e-01 -3.10661644e-01 -3.12964544e-02 8.15348566e-01
2.28499398e-01 -1.21214539e-01 1.15911770e+00 -8.84294093e-01
1.14193104e-01 9.74766985e-02 -4.31974858e-01 2.08199695e-01
-3.51335853e-02 -4.17704098e-02 3.74575108e-01 -8.54842007e-01
2.18534023e-01 7.65287042e-01 6.81697071e-01 8.50505769e-01
-6.32833242e-01 -4.61953193e-01 -6.69549227e-01 1.32676885e-01
-1.46854055e+00 -8.01889077e-02 3.60667855e-01 -5.27719975e-01
5.11277437e-01 4.31911051e-01 9.79564071e-01 1.43737674e-01
5.41378498e-01 4.54659492e-01 1.04640090e+00 -5.50399482e-01
2.90736198e-01 3.05205256e-01 6.23028651e-02 8.15713704e-01
8.47868085e-01 2.25342754e-02 4.33830410e-01 -2.34376520e-01
7.30376542e-01 4.88162637e-01 -1.34624854e-01 -6.38297722e-02
-9.40982163e-01 7.34857857e-01 3.14998418e-01 1.08662188e+00
-5.74912071e-01 8.98816213e-02 1.96776196e-01 -3.80752772e-01
-1.00856647e-02 -1.72255561e-01 -1.80603430e-01 -7.97482580e-02
-1.25621426e+00 -1.66905344e-01 7.81565547e-01 4.23461765e-01
2.10274249e-01 -1.94541633e-01 -8.68909657e-02 3.25620353e-01
5.48007727e-01 7.22089112e-01 7.71852016e-01 -5.24688721e-01
-7.74196535e-02 9.71902490e-01 -2.25191712e-01 -6.58664584e-01
-3.85028571e-01 -3.43234688e-01 -9.59203005e-01 3.46408725e-01
4.51651156e-01 1.30313843e-01 -1.10712993e+00 9.96707141e-01
1.96146980e-01 -6.10749498e-02 1.92416124e-02 4.52227920e-01
5.42972624e-01 5.88389575e-01 4.68479425e-01 -6.05553091e-01
1.23638856e+00 -5.20532370e-01 -7.86283970e-01 3.21861744e-01
6.40140831e-01 -5.25918603e-01 7.56488919e-01 2.41273910e-01
-8.69669199e-01 -3.39337021e-01 -6.38174057e-01 3.65903109e-01
-2.32194930e-01 1.50638511e-02 2.58050859e-01 8.71928096e-01
-8.57888639e-01 3.03962976e-01 -9.37259495e-01 -1.06242621e+00
6.11680448e-01 3.65969121e-01 -7.67304376e-02 2.29924873e-01
-7.78460026e-01 8.14743638e-01 5.40705204e-01 -9.60669965e-02
-6.41350687e-01 -2.83064276e-01 4.33202088e-02 2.26785988e-02
2.44522378e-01 -7.58064806e-01 1.03735483e+00 -6.46280289e-01
-9.70611691e-01 8.42774451e-01 -1.97625846e-01 -2.75936127e-01
6.64312005e-01 7.28276819e-02 -1.76161930e-01 3.50512147e-01
-6.91889375e-02 3.25012773e-01 4.83768672e-01 -1.03605735e+00
-9.37649429e-01 -3.11439306e-01 -8.02747667e-01 2.55688667e-01
-4.08063650e-01 -8.15903246e-02 -5.63047826e-01 -4.04947370e-01
1.17265463e-01 -1.00673378e+00 -5.80618739e-01 2.48735175e-02
-2.26355389e-01 -3.25203151e-01 1.22836506e+00 -6.03087664e-01
1.42807817e+00 -1.98699534e+00 -1.45656586e-01 4.80007529e-01
3.98816049e-01 3.21952879e-01 1.02513576e+00 2.70788372e-01
1.25968546e-01 5.62467515e-01 -6.51291311e-01 4.12418634e-01
-5.25131345e-01 3.87412101e-01 2.43958592e-01 2.69853115e-01
-2.69440383e-01 4.17416930e-01 -6.78574443e-01 -1.39396513e+00
1.94112450e-01 2.23531649e-01 -1.29334629e-01 2.48880073e-01
1.24951050e-01 3.65810066e-01 -9.38151538e-01 8.52500498e-01
3.86688977e-01 -9.22254696e-02 -5.55913486e-02 1.41140655e-01
-3.30224037e-01 -4.46461201e-01 -6.51869059e-01 8.42481494e-01
5.29563241e-03 2.71212399e-01 -7.21972063e-02 -7.64351130e-01
5.52809775e-01 1.00501156e+00 9.73461926e-01 -1.16908468e-01
3.75728130e-01 4.99150991e-01 1.20651521e-01 -6.79193199e-01
1.25073362e-02 -4.30312157e-01 3.00981909e-01 8.77366439e-02
-6.23186409e-01 -4.96637732e-01 3.02077055e-01 2.85689533e-01
1.07257998e+00 -3.39685977e-01 8.41425776e-01 -3.05835575e-01
8.70401621e-01 5.43343544e-01 3.80105853e-01 2.26389721e-01
-4.01981235e-01 4.75724936e-01 2.13168263e-01 1.10003561e-01
-8.77425015e-01 -1.23382008e+00 -4.97878045e-01 1.64644763e-01
-2.62410045e-01 3.05073172e-01 -1.01363599e+00 -7.58885920e-01
-2.18312204e-01 6.15235388e-01 -2.37242237e-01 8.43938440e-02
-5.94975233e-01 -7.27884591e-01 4.73849684e-01 2.32138693e-01
1.06598341e+00 -1.09336853e+00 -7.05760062e-01 2.68680573e-01
1.56482950e-01 -3.54335457e-01 -1.35458142e-01 8.30306262e-02
-1.62673903e+00 -1.18229616e+00 -1.21669316e+00 -7.27599561e-01
8.08477938e-01 1.56382576e-01 8.08065772e-01 7.08367944e-01
-7.55406737e-01 5.05724669e-01 -6.58206493e-02 -4.21395272e-01
-7.68299997e-01 8.93400908e-02 -2.53797859e-01 -6.84931040e-01
3.09034109e-01 -2.99539596e-01 -4.82374430e-01 1.16411120e-01
-8.25257421e-01 -1.62858933e-01 9.81095433e-01 3.49892378e-01
5.37460685e-01 6.97983027e-01 4.27413046e-01 -1.19243157e+00
3.94488633e-01 -5.75818837e-01 -2.35489056e-01 1.59269646e-01
-9.44670796e-01 -3.14035535e-01 4.61092502e-01 1.64909586e-01
-1.18321347e+00 1.98178694e-01 -1.15828756e-02 -6.09826632e-02
-4.93269265e-01 1.83921739e-01 -1.45877553e-02 1.75268441e-01
5.95640421e-01 1.94354624e-01 -8.08648542e-02 -2.31197298e-01
-1.97279423e-01 7.77193487e-01 4.66440916e-01 7.00716227e-02
1.03697670e+00 3.97578001e-01 4.84465927e-01 -1.25326037e+00
-3.89297456e-01 -9.25084352e-01 -8.13077569e-01 -5.49140751e-01
1.43876493e+00 -4.66974139e-01 -1.40133157e-01 4.29172158e-01
-7.41895914e-01 -3.22147384e-02 -2.55042434e-01 8.20481420e-01
-2.31063768e-01 4.47475284e-01 -3.81432950e-01 -1.01760995e+00
-5.12956023e-01 -4.26335394e-01 1.48175657e-01 5.88048339e-01
-2.30492011e-01 -1.32770276e+00 2.34185576e-01 2.36659318e-01
3.68882120e-01 3.23997885e-01 1.13074577e+00 -5.91892302e-01
-5.35053968e-01 -3.42686802e-01 -7.82969072e-02 1.37141004e-01
3.43059331e-01 3.51405352e-01 -5.74650288e-01 -9.10656061e-03
2.97008067e-01 1.79417357e-01 8.12276006e-01 4.54232216e-01
8.71382833e-01 -3.70384455e-02 -9.93123710e-01 6.29496872e-02
1.63773942e+00 9.35359180e-01 7.99031079e-01 2.27371585e-02
5.13158143e-01 4.15632695e-01 7.33097136e-01 1.39445886e-01
-2.01944187e-01 2.01781243e-02 3.90451968e-01 -3.87336761e-01
-1.86060131e-01 1.46566495e-01 -1.15984723e-01 9.46715891e-01
-4.66680825e-01 -4.78943825e-01 -1.33111775e+00 6.24650538e-01
-1.30325866e+00 -1.21943617e+00 -7.24958897e-01 2.19447637e+00
7.32741237e-01 3.56199205e-01 1.93930507e-01 4.74524140e-01
8.21224093e-01 -4.55094993e-01 -8.95526931e-02 -8.51304561e-04
4.59029168e-01 2.33384907e-01 7.43400872e-01 4.54255342e-01
-8.48144293e-01 4.34393942e-01 6.61675930e+00 7.00400651e-01
-1.16681707e+00 1.86354786e-01 4.91074562e-01 4.31011975e-01
-7.99645558e-02 2.18795776e-01 -6.61209583e-01 4.92521733e-01
5.51470101e-01 -4.04767543e-01 -2.92565405e-01 7.09260821e-01
4.63954449e-01 -7.25254416e-01 -4.78169113e-01 5.64079046e-01
-2.19456136e-01 -6.80292130e-01 -1.37048796e-01 3.46494645e-01
4.89827722e-01 -3.32541734e-01 -2.23765671e-01 2.31592581e-02
-2.54200011e-01 -1.01131237e+00 1.35503069e-01 8.55771899e-01
7.52431214e-01 -7.01155126e-01 8.03191304e-01 7.29089797e-01
-1.18220997e+00 4.53628711e-02 -1.88310727e-01 2.99482375e-01
2.58788057e-02 6.19301319e-01 -1.49949622e+00 1.86562687e-01
4.02053952e-01 1.48772970e-01 -5.94199419e-01 1.36259937e+00
6.71793595e-02 1.14175463e+00 -6.20757818e-01 -3.68442327e-01
-1.07298687e-01 -3.73750448e-01 6.28106952e-01 1.00750005e+00
5.53312719e-01 4.68893915e-01 -6.85617514e-03 7.49798000e-01
2.33211458e-01 4.79448438e-01 -5.41410923e-01 -2.33136222e-01
3.53698373e-01 1.23414004e+00 -1.42500877e+00 -5.40928125e-01
-3.82615089e-01 8.64000022e-01 -3.98573577e-01 -1.40272498e-01
-9.33180511e-01 -2.21200973e-01 -3.52289796e-01 8.12653005e-01
-2.22391605e-01 5.08547239e-02 -3.93688172e-01 -5.28341234e-01
-6.79749072e-01 -1.23323612e-01 4.63098824e-01 -2.09461302e-01
-5.65221071e-01 2.60451764e-01 2.96385109e-01 -9.53254402e-01
-1.18416458e-01 -3.27939481e-01 -1.19888937e+00 6.40796006e-01
-7.72153795e-01 -7.66830564e-01 -3.79942507e-01 4.30289149e-01
5.35282373e-01 -1.57908276e-01 7.91888833e-01 2.87684500e-01
-5.74219227e-01 -1.32756919e-01 2.55239815e-01 -2.94834822e-02
5.19055605e-01 -1.40024412e+00 -4.64485914e-01 6.40927970e-01
-6.05988801e-01 1.19418383e-01 5.53857267e-01 -1.10553920e+00
-6.28953934e-01 -8.13217819e-01 1.08709550e+00 -3.64374965e-02
2.01608315e-01 3.84392142e-01 -7.77507007e-01 1.35851815e-01
2.94841342e-02 -1.47568971e-01 5.53671241e-01 -6.26991928e-01
7.82727301e-01 9.23276246e-02 -1.35013175e+00 3.93769562e-01
4.10649806e-01 -1.20450243e-01 -5.34647822e-01 3.49172503e-01
-5.10073788e-02 2.24993929e-01 -7.52350688e-01 2.70705700e-01
4.38926637e-01 -1.21433437e+00 6.79063618e-01 2.07734257e-01
-6.17595911e-02 -2.85972387e-01 3.27081442e-01 -6.63836122e-01
-5.90918839e-01 7.79509172e-02 4.84469801e-01 1.24728560e+00
5.22388220e-01 -4.15409416e-01 1.31374526e+00 4.70564753e-01
4.78074141e-02 -6.74949050e-01 -8.50319564e-01 -4.21734542e-01
1.53084666e-01 1.59775615e-01 -1.82680130e-01 7.21997142e-01
-4.04768139e-01 -4.11836505e-02 4.53966588e-01 -3.35617125e-01
8.11638713e-01 -3.84991884e-01 3.13968986e-01 -1.61449218e+00
-1.33224055e-02 -4.98770982e-01 -2.86864787e-01 -1.96322352e-01
-4.87151742e-01 -6.62891030e-01 8.31263699e-03 -2.07822013e+00
4.30821925e-01 -4.46271926e-01 -1.62613526e-01 3.82828414e-01
-1.75253496e-01 1.73726261e-01 -2.79273391e-02 6.77951038e-01
-7.74366409e-02 -2.50616502e-02 1.22839558e+00 1.08686902e-01
-2.70311177e-01 4.07133281e-01 -1.29011378e-01 1.04785359e+00
1.15284121e+00 -6.41921282e-01 -4.71953392e-01 5.02574444e-01
-1.61742985e-01 3.26501250e-01 -3.82262282e-02 -1.54751503e+00
7.67423585e-02 -5.65745652e-01 5.82128406e-01 -9.11800981e-01
-6.96764290e-02 -1.59462190e+00 6.37827396e-01 1.38205779e+00
1.23129599e-01 -4.97465044e-01 -1.80678874e-01 5.58793664e-01
-7.88528398e-02 -7.99005568e-01 1.18322086e+00 -3.14055234e-01
-3.37424070e-01 1.34047568e-01 -9.88918364e-01 -1.19918823e-01
1.56123638e+00 -7.52134323e-01 2.49379143e-01 -1.52503833e-01
-6.70765817e-01 -2.45322417e-02 6.14133894e-01 -4.91285831e-01
4.21056777e-01 -1.15716767e+00 -5.32566845e-01 -2.06987873e-01
-2.86651284e-01 1.89893648e-01 1.16496988e-01 1.30910003e+00
-1.32778716e+00 4.49183643e-01 -1.87322900e-01 -3.86489689e-01
-1.69100428e+00 5.04443765e-01 5.41041553e-01 -5.29290676e-01
-4.32438225e-01 4.90412980e-01 2.69343287e-01 3.60664576e-01
-3.26760113e-02 -3.81151497e-01 -7.11067021e-01 -1.19816177e-01
-1.56997144e-01 9.55213487e-01 -2.98116684e-01 -5.59811532e-01
-3.49956870e-01 7.27357030e-01 9.71350372e-02 -2.50218123e-01
5.78292668e-01 7.76486695e-02 -3.51596087e-01 4.83173043e-01
8.50595772e-01 5.63240588e-01 -5.71546972e-01 4.09624547e-01
2.04968721e-01 -1.18531071e-01 2.38765717e-01 -7.75525630e-01
-9.18211639e-01 7.06804037e-01 8.61354768e-01 4.70645636e-01
1.27197981e+00 -2.13242117e-02 9.23045754e-01 3.99555415e-01
7.76768569e-03 -1.10255933e+00 -2.80732363e-01 2.29509696e-01
4.64658767e-01 -9.35037196e-01 2.94032514e-01 -6.41297460e-01
-4.25900102e-01 1.41506493e+00 4.13611859e-01 -1.10512510e-01
1.02739835e+00 4.48307455e-01 -4.48571146e-03 -1.58559054e-01
-4.77278501e-01 -1.51217729e-01 2.60487590e-02 4.63128597e-01
7.65481174e-01 8.18456486e-02 -9.53341365e-01 3.93333048e-01
1.63527012e-01 2.59029120e-01 1.73094586e-01 1.30406988e+00
-1.44319284e+00 -1.09170473e+00 -1.02239859e+00 6.82095587e-01
-9.09927070e-01 2.41512299e-01 -6.66602015e-01 1.20696890e+00
5.33998013e-01 6.43240333e-01 1.00014053e-01 2.92133957e-01
6.22521639e-02 3.02173823e-01 3.31752032e-01 -3.62201899e-01
-5.07656753e-01 2.70452291e-01 -1.93248272e-01 1.21290311e-01
-4.72348660e-01 -6.70127630e-01 -2.33795547e+00 -2.10070148e-01
-5.17503560e-01 7.00030923e-01 7.21864760e-01 6.51946604e-01
-5.78843236e-01 3.56327802e-01 5.38086593e-01 -1.20187365e-01
-2.57457197e-01 -1.08586490e+00 -9.20096815e-01 1.05832450e-01
-1.64699927e-01 -3.37281466e-01 -6.52943194e-01 2.32401133e-01] | [15.313376426696777, -2.390078067779541] |
54656492-b9e1-496f-b200-06dce3d105af | zero-shot-cross-lingual-abstractive-sentence | null | null | https://aclanthology.org/P19-1305 | https://aclanthology.org/P19-1305.pdf | Zero-Shot Cross-Lingual Abstractive Sentence Summarization through Teaching Generation and Attention | Abstractive Sentence Summarization (ASSUM) targets at grasping the core idea of the source sentence and presenting it as the summary. It is extensively studied using statistical models or neural models based on the large-scale monolingual source-summary parallel corpus. But there is no cross-lingual parallel corpus, whose source sentence language is different to the summary language, to directly train a cross-lingual ASSUM system. We propose to solve this zero-shot problem by using resource-rich monolingual ASSUM system to teach zero-shot cross-lingual ASSUM system on both summary word generation and attention. This teaching process is along with a back-translation process which simulates source-summary pairs. Experiments on cross-lingual ASSUM task show that our proposed method is significantly better than pipeline baselines and previous works, and greatly enhances the cross-lingual performances closer to the monolingual performances. | ['Xiangyu Duan', 'Mingming Yin', 'Weihua Luo', 'Min Zhang', 'Boxing Chen'] | 2019-07-01 | null | null | null | acl-2019-7 | ['abstractive-sentence-summarization'] | ['natural-language-processing'] | [ 1.32252350e-01 2.33549058e-01 -2.55910575e-01 -3.73682559e-01
-1.57438588e+00 -5.81460536e-01 8.25621545e-01 6.09399714e-02
-3.79708230e-01 8.55488539e-01 8.45346272e-01 -4.41673070e-01
6.94843888e-01 -3.74595761e-01 -9.53662872e-01 -2.55602479e-01
4.46486056e-01 5.05915523e-01 1.07430339e-01 -6.14760518e-01
2.14450255e-01 -2.80728996e-01 -8.62528384e-01 7.56279826e-01
1.52383375e+00 2.05357000e-01 8.42685819e-01 6.20279431e-01
-4.50183243e-01 7.15820849e-01 -9.88769948e-01 -7.20516980e-01
-1.04042493e-01 -1.03092539e+00 -9.37461555e-01 -1.19390182e-01
9.50749755e-01 -3.34089875e-01 -1.82524458e-01 1.23602605e+00
6.53181136e-01 3.31640728e-02 5.98196387e-01 -6.82841480e-01
-1.35254812e+00 1.47236216e+00 -4.51986760e-01 3.81335855e-01
4.32130426e-01 1.38883382e-01 1.07138920e+00 -9.42814827e-01
6.42434835e-01 1.43940556e+00 4.43610847e-01 6.87444210e-01
-8.98147166e-01 -3.55167955e-01 2.60152131e-01 2.11113960e-01
-9.14543450e-01 -6.73683465e-01 7.06296980e-01 -5.62165007e-02
1.23083246e+00 1.45565942e-01 5.28515577e-01 1.35444927e+00
5.08785903e-01 1.25917280e+00 9.12039042e-01 -4.52038139e-01
-1.47878081e-01 1.61886841e-01 4.86963362e-01 5.98752260e-01
8.70354176e-02 -3.85977536e-01 -8.10045660e-01 2.99992383e-01
3.10692400e-01 -5.34166694e-01 -4.05443758e-01 2.98207343e-01
-1.57196712e+00 6.17393732e-01 2.00200647e-01 3.90164673e-01
-3.27480137e-01 -7.94673860e-02 8.08960497e-01 5.00066221e-01
7.10811019e-01 6.11342013e-01 -4.51495349e-01 7.33648539e-02
-1.10594797e+00 9.79236215e-02 7.12543488e-01 1.44191396e+00
4.67045993e-01 4.09986317e-01 -6.73705041e-01 9.13155854e-01
-9.61926207e-02 7.83473372e-01 9.40379024e-01 -6.21070623e-01
1.05442679e+00 3.35662007e-01 -2.31971204e-01 -4.07017380e-01
4.01084609e-02 -5.49647987e-01 -8.87144268e-01 -6.09008849e-01
2.25965381e-02 -2.99357623e-01 -3.54368895e-01 1.77377558e+00
-1.04773156e-01 3.89290266e-02 7.84722686e-01 6.66335762e-01
1.39708328e+00 1.13829672e+00 -9.32910442e-02 -4.46175605e-01
1.42094243e+00 -1.65661693e+00 -9.23960865e-01 -5.80091953e-01
6.47343218e-01 -9.18291390e-01 1.53174543e+00 9.23466161e-02
-1.44580686e+00 -8.10101748e-01 -1.14965129e+00 -6.84027195e-01
-4.17832099e-02 6.20353639e-01 3.54387939e-01 1.70124054e-01
-1.18127143e+00 4.82505500e-01 -7.75176823e-01 -5.16666114e-01
7.03168139e-02 -2.49645680e-01 -3.53234410e-01 -1.02075830e-01
-1.33283687e+00 1.14801824e+00 6.26117408e-01 -4.27819192e-02
-8.42482150e-01 -9.09484863e-01 -1.15534842e+00 1.58671349e-01
1.75891966e-01 -9.36823130e-01 1.72557402e+00 -1.00170124e+00
-1.87283790e+00 7.52713084e-01 -4.25372064e-01 -5.30244708e-01
3.05513829e-01 -4.95773137e-01 -1.51700720e-01 8.10190588e-02
6.39298737e-01 6.93093240e-01 4.45307165e-01 -9.11992908e-01
-5.21785140e-01 -9.55639333e-02 -2.26454258e-01 6.37262583e-01
-2.04788789e-01 1.41686931e-01 -3.08070958e-01 -7.44053066e-01
-2.27759868e-01 -5.07041574e-01 -9.24857929e-02 -9.67414558e-01
-6.49636984e-01 -5.73873222e-01 3.45690548e-01 -1.13330030e+00
1.13185370e+00 -1.68086350e+00 2.24330589e-01 -8.93883228e-01
-3.60215843e-01 3.73887330e-01 -5.02632618e-01 9.98272121e-01
9.42043141e-02 -8.20819512e-02 -2.83167034e-01 -6.76745594e-01
-5.88678047e-02 -1.18578061e-01 -7.32728064e-01 -1.17733078e-02
3.89973134e-01 1.28128827e+00 -1.32027495e+00 -5.18459797e-01
-1.72646284e-01 8.53072032e-02 -3.86901528e-01 4.62599099e-01
-3.59706014e-01 4.58729833e-01 -1.64707735e-01 1.31464630e-01
3.39612305e-01 9.07249451e-02 3.20557117e-01 -8.33971649e-02
-2.14959174e-01 1.08648014e+00 -2.75465608e-01 2.49530125e+00
-6.54586792e-01 6.26571119e-01 -3.34577680e-01 -8.90830517e-01
9.01402652e-01 5.09492815e-01 -3.29499364e-01 -7.07823277e-01
1.65284112e-01 3.46133381e-01 8.75774398e-02 -5.50914466e-01
9.45122600e-01 -1.41783759e-01 -4.29712147e-01 6.67375624e-01
4.31942463e-01 -3.69629920e-01 5.28230488e-01 6.93292141e-01
8.28790188e-01 3.49730432e-01 5.74061394e-01 -6.40365422e-01
5.98081887e-01 2.16794237e-01 5.25955439e-01 6.75772607e-01
1.75211385e-01 6.84928238e-01 6.13443971e-01 1.18166782e-01
-9.72829998e-01 -1.29026961e+00 3.25681597e-01 9.19988394e-01
2.65426245e-02 -6.33914948e-01 -1.14976096e+00 -8.00046444e-01
-4.34321135e-01 1.30450475e+00 -2.61070430e-01 -2.08358467e-01
-8.14836562e-01 -3.44276458e-01 6.16020918e-01 5.27911484e-01
6.45061016e-01 -1.20927119e+00 -1.10651903e-01 3.72793883e-01
-7.29487658e-01 -1.16298223e+00 -1.05807543e+00 -2.52188414e-01
-8.21215928e-01 -6.30224526e-01 -8.34398746e-01 -1.17658806e+00
5.26896298e-01 3.83369476e-01 1.35065138e+00 -3.08482558e-01
4.09378439e-01 -9.04933959e-02 -2.72235811e-01 -3.96643013e-01
-9.20270264e-01 5.61016083e-01 7.54941702e-02 -3.02829772e-01
3.75901341e-01 -3.77032369e-01 -1.92345306e-01 -5.04609108e-01
-5.32263696e-01 6.69658542e-01 8.77399445e-01 8.50968361e-01
3.38507384e-01 -5.63333273e-01 1.22858572e+00 -1.03551185e+00
1.14548171e+00 -3.76386046e-01 -2.31343791e-01 5.08570254e-01
-1.06543474e-01 2.09952071e-01 1.01698613e+00 -2.61547834e-01
-1.30331612e+00 -4.26831245e-01 -2.21514210e-01 -4.80596647e-02
5.31391380e-03 6.60966575e-01 -3.92931223e-01 8.81518841e-01
4.98804450e-01 7.73100793e-01 -1.67375281e-01 -5.16150117e-01
5.63266575e-01 8.64419341e-01 8.97509873e-01 -5.75927913e-01
4.30123150e-01 -1.37742937e-01 -7.01063097e-01 -9.22046959e-01
-1.31566918e+00 -2.16077641e-01 -8.95222425e-01 1.06983975e-01
8.62931550e-01 -1.21645832e+00 -2.09199935e-01 3.88681293e-01
-1.89747167e+00 -4.38909978e-01 -2.72699952e-01 4.73632216e-01
-5.66156209e-01 4.99959499e-01 -8.52791727e-01 -3.28390807e-01
-8.37383926e-01 -1.17977190e+00 1.28543556e+00 4.76822518e-02
-2.95319527e-01 -1.05547857e+00 2.99547583e-01 4.03007627e-01
3.26398045e-01 -3.54193449e-01 9.45184648e-01 -7.39449799e-01
-4.00229007e-01 1.28236935e-01 2.62255245e-03 7.58105397e-01
1.67345062e-01 -2.31112346e-01 -6.35562897e-01 -2.38979727e-01
1.95594534e-01 -4.15328979e-01 9.15775239e-01 1.78141594e-01
6.36328042e-01 -6.46304309e-01 9.19268951e-02 2.37343222e-01
1.10781884e+00 -1.03698090e-01 4.46343631e-01 -1.00244664e-01
5.99561572e-01 8.08705509e-01 4.86140251e-01 -3.34589243e-01
1.15944314e+00 3.63109559e-01 -3.66352648e-01 3.18014063e-02
-6.46752834e-01 -6.52947903e-01 1.22291780e+00 2.19491696e+00
4.71836209e-01 -2.71023959e-01 -7.09657609e-01 9.10576344e-01
-1.87056148e+00 -1.08665752e+00 -4.12642747e-01 1.96528602e+00
1.33307111e+00 -2.69341078e-02 -2.22617030e-01 -6.23625636e-01
6.25889003e-01 3.56441826e-01 -2.91178644e-01 -6.75789773e-01
-3.15754652e-01 -4.44874726e-02 4.00979146e-02 7.45537460e-01
-7.67261505e-01 1.61263251e+00 5.95012569e+00 1.02350748e+00
-1.18612444e+00 4.14327025e-01 3.04032654e-01 -4.68886942e-02
-4.51173574e-01 -3.39516252e-02 -1.04986858e+00 4.43418890e-01
1.28431749e+00 -7.64742017e-01 9.84198377e-02 6.82107270e-01
2.41766393e-01 -4.57105115e-02 -1.34893823e+00 6.01074040e-01
7.27279603e-01 -1.44915068e+00 4.48076159e-01 -5.53112209e-01
1.05402446e+00 2.68359989e-01 -3.06247294e-01 8.13369155e-01
4.98263448e-01 -6.60716534e-01 8.11748981e-01 2.28214607e-01
8.55855584e-01 -7.03355610e-01 7.83600092e-01 7.12434113e-01
-1.01963949e+00 6.22509599e-01 -5.58562875e-01 -1.07880935e-01
4.01230782e-01 1.31753057e-01 -8.01328838e-01 8.72579038e-01
1.61790535e-01 1.09217656e+00 -7.45148718e-01 5.14968216e-01
-8.14723969e-01 7.92231441e-01 2.88703054e-01 -2.73015290e-01
4.23322201e-01 -2.76447028e-01 6.14058614e-01 1.47509491e+00
3.29369634e-01 -1.76302306e-02 2.62587696e-01 8.83467376e-01
-2.96242356e-01 6.20026827e-01 -8.17146003e-01 -1.05279312e-01
3.42278600e-01 1.12466323e+00 -2.38316804e-01 -9.33667302e-01
-5.71581185e-01 1.35248280e+00 7.05178976e-01 3.54927748e-01
-4.63950723e-01 -4.86728698e-01 2.80359358e-01 -2.19417289e-01
-1.45218344e-02 -2.67163903e-01 -3.55355024e-01 -1.71922290e+00
3.72606777e-02 -1.08875501e+00 -1.54742539e-01 -9.79866564e-01
-1.53789699e+00 1.00536633e+00 -3.43660712e-02 -1.11352766e+00
-4.06662375e-01 -2.23936662e-01 -1.15927970e+00 1.14581788e+00
-1.61437953e+00 -1.46926868e+00 1.98637068e-01 3.40295583e-01
1.44185913e+00 -4.83950406e-01 7.65148044e-01 7.18303621e-02
-7.17970669e-01 6.20944917e-01 4.65193652e-02 2.22002510e-02
1.02207446e+00 -1.44412422e+00 9.51887429e-01 1.33909678e+00
1.36536717e-01 8.87741864e-01 6.29619062e-01 -8.52206171e-01
-1.26127183e+00 -1.30163109e+00 1.52697039e+00 -4.70994145e-01
8.55365753e-01 -5.65378010e-01 -1.04284465e+00 9.59012508e-01
1.36015618e+00 -7.08638310e-01 6.00739658e-01 8.77928957e-02
-8.77926573e-02 3.21447812e-02 -5.23619652e-01 1.01941621e+00
9.98572052e-01 -6.09294832e-01 -1.39639771e+00 5.89416087e-01
1.27880907e+00 -3.60921681e-01 -5.31387389e-01 8.07983056e-02
6.25714809e-02 -6.60306811e-01 4.39831734e-01 -7.03119516e-01
1.09460521e+00 -1.04962245e-01 -2.14318167e-02 -1.97343075e+00
-7.76373297e-02 -6.74689770e-01 1.75393701e-01 1.66075754e+00
4.53798980e-01 -5.35580575e-01 1.47321463e-01 -2.70937264e-01
-1.01281810e+00 -5.82676649e-01 -7.19082654e-01 -8.87358785e-01
5.85826755e-01 -1.30663797e-01 4.92490858e-01 8.86507750e-01
3.84305060e-01 1.32317531e+00 -1.52174532e-01 1.17021084e-01
5.73829889e-01 2.09695786e-01 9.09480870e-01 -6.51768506e-01
-2.44333103e-01 -6.32051408e-01 3.47407341e-01 -1.44477034e+00
7.65802622e-01 -1.58583379e+00 2.96877474e-01 -1.99525583e+00
4.27304536e-01 1.16372861e-01 9.53354463e-02 2.72812963e-01
-7.19729125e-01 -1.40969962e-01 1.40707523e-01 1.36529163e-01
-6.58153951e-01 1.05788553e+00 1.37771511e+00 -1.60374388e-01
-7.82321114e-03 -3.15017074e-01 -9.72638190e-01 5.45007169e-01
6.79110408e-01 -3.99780601e-01 -4.91511673e-01 -9.95587468e-01
-1.75983131e-01 4.41847384e-01 -1.17365502e-01 -7.19693482e-01
3.56801808e-01 -3.05791339e-03 -1.54919967e-01 -9.90897417e-01
-5.65108545e-02 8.01734626e-03 -5.42034626e-01 3.24346930e-01
-7.05397606e-01 3.83946508e-01 2.13396758e-01 2.11909965e-01
-4.59405750e-01 -3.09679627e-01 5.42169094e-01 -3.29249293e-01
-3.81228328e-01 -2.45578829e-02 -3.91307503e-01 4.30927634e-01
6.68262184e-01 2.63418108e-01 -7.52197802e-01 -2.92521596e-01
-2.39821404e-01 4.58130211e-01 2.91254580e-01 7.80485690e-01
3.78740609e-01 -1.42362535e+00 -1.40643561e+00 1.12728521e-01
7.19520077e-02 -5.15106469e-02 3.15356046e-01 8.43407512e-01
-4.81308758e-01 7.69088805e-01 -1.27291545e-01 -3.89282405e-01
-1.00634515e+00 3.98712069e-01 1.46134719e-01 -4.21330452e-01
-6.48397624e-01 8.70708644e-01 4.20608848e-01 -7.85497904e-01
-1.06771462e-01 -3.59784007e-01 -1.06382892e-01 2.37130877e-02
8.54805291e-01 2.07795516e-01 6.10526092e-02 -5.55805981e-01
-3.28291468e-02 2.95306385e-01 -3.85407507e-01 -2.66382098e-01
1.17031288e+00 -3.75998735e-01 -4.36476201e-01 9.08260584e-01
1.06313181e+00 2.78471559e-01 -9.74142134e-01 -4.93478090e-01
7.23570734e-02 6.77450523e-02 -1.82395935e-01 -7.68753111e-01
-5.98113179e-01 1.24540579e+00 -2.95907855e-01 -1.42110661e-01
6.86889887e-01 -5.65770874e-03 1.24554312e+00 5.22055805e-01
2.95625627e-01 -1.02678668e+00 1.93376601e-01 1.00547445e+00
1.43422198e+00 -1.32862186e+00 -2.55859405e-01 -3.11696082e-01
-1.10738266e+00 9.60446298e-01 8.37281108e-01 -4.36809689e-01
-1.15749404e-01 -6.06319029e-03 2.68219411e-01 4.80280817e-02
-1.20370853e+00 1.91235058e-02 4.40545022e-01 3.47354174e-01
8.16189587e-01 5.04885726e-02 -6.95967972e-01 1.03576589e+00
-9.07605708e-01 -4.55113679e-01 7.99209237e-01 4.79831874e-01
-4.27921951e-01 -1.18327308e+00 -2.68083345e-02 2.38668263e-01
-4.83216524e-01 -8.11878562e-01 -4.80889261e-01 5.86062551e-01
-3.13098073e-01 9.56082642e-01 1.62012383e-01 2.03972682e-02
3.70704025e-01 2.13455811e-01 5.44578493e-01 -1.26924753e+00
-9.92651999e-01 4.67413589e-02 1.97426453e-01 -1.99749202e-01
7.10667372e-02 -7.71363497e-01 -1.16607857e+00 -2.29148358e-01
-2.35681646e-02 3.80392671e-01 6.22424960e-01 1.13856399e+00
3.36487889e-01 8.96549165e-01 5.16467214e-01 -8.16902041e-01
-7.42654383e-01 -1.48836136e+00 -1.81339696e-01 8.97127613e-02
2.29669660e-01 1.56167448e-01 -2.22309642e-02 1.64006114e-01] | [12.273290634155273, 9.509532928466797] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.