Filters








875 Hits in 5.9 sec

NAT: Neural Architecture Transformer for Accurate and Compact Architectures [article]

Yong Guo, Yin Zheng, Mingkui Tan, Qi Chen, Jian Chen, Peilin Zhao, Junzhou Huang
<span title="2020-01-13">2020</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
To make the problem feasible, we cast the optimization problem into a Markov decision process (MDP) and seek to learn a Neural Architecture Transformer (NAT) to replace the redundant operations with the  ...  ., CIFAR-10 and ImageNet, demonstrate that the transformed architecture by NAT significantly outperforms both its original form and those architectures optimized by existing methods.  ...  Algorithm 1 Training method for Neural Architecture Transformer (NAT).  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/1910.14488v5">arXiv:1910.14488v5</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/pzlz67xaongo7noorun3imj6pi">fatcat:pzlz67xaongo7noorun3imj6pi</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20200828215753/https://arxiv.org/pdf/1910.14488v5.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/00/ca/00ca8c9d62b29e5c5b643a5e1fde8e3a0de9518f.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/1910.14488v5" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

An Explicit Local and Global Representation Disentanglement Framework with Applications in Deep Clustering and Unsupervised Object Detection [article]

Rujikorn Charakorn, Yuttapong Thawornwattana, Sirawaj Itthipuripat, Nick Pawlowski, Poramate Manoonpong, Nat Dilokthanakul
<span title="2020-02-24">2020</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
The code for our experiments is at https://github.com/51616/split-vae .  ...  We show that the framework can effectively disentangle local and global information within these models leads to improved representation, with better clustering and unsupervised object detection benchmarks  ...  Nat Dilokthanakul and Rujikorn Chanrakorn are the main contributors and contributes equally to this work.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2001.08957v2">arXiv:2001.08957v2</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/2qjbomhn5zdntdojaq5y4jivqe">fatcat:2qjbomhn5zdntdojaq5y4jivqe</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20200321153121/https://arxiv.org/pdf/2001.08957v2.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2001.08957v2" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

The FIRST Classifier: compact and extended radio galaxy classification using deep Convolutional Neural Networks

Wathela Alhassan, A R Taylor, Mattia Vaccari
<span title="2018-07-27">2018</span> <i title="Oxford University Press (OUP)"> <a target="_blank" rel="noopener" href="https://fatcat.wiki/container/qjdxzupqgjennounydm4zfdlem" style="color: black;">Monthly notices of the Royal Astronomical Society</a> </i> &nbsp;
Our model achieved an overall accuracy of 97% and a recall of 98%, 100%, 98% and 93% for Compact, BENT, FRI and FRII galaxies respectively.  ...  The current version of the FIRST classifier is able to predict the morphological class for a single source or for a list of sources as Compact or Extended (FRI, FRII and BENT).  ...  a trained deep Convolutional Neural Network (CNN) model to generate accurate and robust predictions.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener noreferrer" href="https://doi.org/10.1093/mnras/sty2038">doi:10.1093/mnras/sty2038</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/6mqhd4ipkfeljlmz7p2dtpwbru">fatcat:6mqhd4ipkfeljlmz7p2dtpwbru</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20200930044354/https://arxiv.org/pdf/1807.10380v1.pdf" title="fulltext PDF download [not primary version]" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <span style="color: #f43e3e;">&#10033;</span> <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/42/8b/428bf4eb9a80a382fdcfeb93eda71bd494a2845f.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener noreferrer" href="https://doi.org/10.1093/mnras/sty2038"> <button class="ui left aligned compact blue labeled icon button serp-button"> <i class="external alternate icon"></i> oup.com </button> </a>

Morphological classification of compact and extended radio galaxies using convolutional neural networks and data augmentation techniques

Viera Maslej-Krešňáková, Khadija El Bouchefry, Peter Butka
<span title="2021-05-17">2021</span> <i title="Oxford University Press (OUP)"> <a target="_blank" rel="noopener" href="https://fatcat.wiki/container/qjdxzupqgjennounydm4zfdlem" style="color: black;">Monthly notices of the Royal Astronomical Society</a> </i> &nbsp;
The proposed architecture comprises three parallel blocks of convolutional layers combined and processed for final classification by two feed-forward layers.  ...  Compact (COMPT).  ...  APVV-16-0213 and Slovak VEGA research grant No. 1/0685/21. We thank the anonymous referee for the comments and suggestions that have improved the manuscript considerably.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener noreferrer" href="https://doi.org/10.1093/mnras/stab1400">doi:10.1093/mnras/stab1400</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/y7zlr4dc4ve3tdnx35fsueh7wa">fatcat:y7zlr4dc4ve3tdnx35fsueh7wa</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20210707013823/https://arxiv.org/pdf/2107.00385v1.pdf" title="fulltext PDF download [not primary version]" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <span style="color: #f43e3e;">&#10033;</span> <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/89/9a/899a014c6897d50fc76785c49bc9d7c0c1ec7174.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener noreferrer" href="https://doi.org/10.1093/mnras/stab1400"> <button class="ui left aligned compact blue labeled icon button serp-button"> <i class="external alternate icon"></i> oup.com </button> </a>

PeriodNet: A non-autoregressive waveform generation model with a structure separating periodic and aperiodic components [article]

Yukiya Hono, Shinji Takaki, Kei Hashimoto, Keiichiro Oura, Yoshihiko Nankaku, Keiichi Tokuda
<span title="2021-02-15">2021</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
We propose PeriodNet, a non-autoregressive (non-AR) waveform generation model with a new model structure for modeling periodic and aperiodic components in speech waveforms.  ...  To address this issue, we propose a parallel model and a series model structure separating periodic and aperiodic components.  ...  In fact, it is difficult for a neural vocoder to generate a speech waveform with accurate pitches outside the range of the training data.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2102.07786v1">arXiv:2102.07786v1</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/l7j5d5ow7jbcfnmdade3athhpa">fatcat:l7j5d5ow7jbcfnmdade3athhpa</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20210218015934/https://arxiv.org/pdf/2102.07786v1.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/3c/1e/3c1e33ceb0067c2ea0a64ceccdf15ee60675ad92.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2102.07786v1" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

Attention-based generative models for de novo molecular design

Orion Dollar, Nisarg Joshi, David Beck, Jim Pfaendtner
<span title="">2021</span> <i title="Royal Society of Chemistry (RSC)"> <a target="_blank" rel="noopener" href="https://fatcat.wiki/container/lnaynun4fzdepmirohmumo7whu" style="color: black;">Chemical Science</a> </i> &nbsp;
Attention mechanisms have led to many breakthroughs in sequential data modeling but have yet to be incorporated into any generative algorithms for molecular design. Here we explore the impact of...  ...  networks (RNN), RNN+Attention and Transformer 71 VAE architectures for the purpose of molecular generation follows.  ...  All models were trained for 100 epochs unless stated otherwise. 398 399 Neural Network Architecture 400 401 As the size of the contextual embedding is significantly larger for the two attention-based architectures  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener noreferrer" href="https://doi.org/10.1039/d1sc01050f">doi:10.1039/d1sc01050f</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/7epm2vtmxja2hj4k3lmnd3aste">fatcat:7epm2vtmxja2hj4k3lmnd3aste</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20210516113020/https://pubs.rsc.org/en/content/articlepdf/2021/sc/d1sc01050f" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/30/76/3076d3c1781cba34141f4c762a01786aed8a5b31.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener noreferrer" href="https://doi.org/10.1039/d1sc01050f"> <button class="ui left aligned compact blue labeled icon button serp-button"> <i class="unlock alternate icon" style="background-color: #fb971f;"></i> Publisher / doi.org </button> </a>

Low-rank passthrough neural networks [article]

Antonio Valerio Miceli Barone
<span title="2018-07-09">2018</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
In this work we extend this line of research, proposing effective, low-rank and low-rank plus diagonal matrix parametrizations for Passthrough Networks which exploit this decoupling property, reducing  ...  Various common deep learning architectures, such as LSTMs, GRUs, Resnets and Highway Networks, employ state passthrough connections that support training with high feed-forward depth or recurrence over  ...  CONCLUSIONS AND FUTURE WORK We proposed low-dimensional parametrizations for passthrough neural networks based on low-rank or low-rank plus diagonal decompositions of the n × n matrices that occur in the  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/1603.03116v3">arXiv:1603.03116v3</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/a3lcxaba6zgizcswobpl6n3lhy">fatcat:a3lcxaba6zgizcswobpl6n3lhy</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20200925071315/https://arxiv.org/pdf/1603.03116v3.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/8b/b1/8bb1915a741e49f63180347ab8d35876cc6a9094.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/1603.03116v3" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

Learning to Generate with Memory [article]

Chongxuan Li, Jun Zhu, Bo Zhang
<span title="2016-05-28">2016</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
The asymmetric architecture can reduce the competition between bottom-up invariant feature extraction and top-down generation of instance details.  ...  generation, and missing value imputation.  ...  This asymmetric recognition network is sufficient for extracting invariant representations in bottom-up inference, and is compact in parameterization.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/1602.07416v2">arXiv:1602.07416v2</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/anj63ui6w5f3tfkmtn34jdb55i">fatcat:anj63ui6w5f3tfkmtn34jdb55i</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20191022110912/https://arxiv.org/pdf/1602.07416v2.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/cb/8e/cb8e8e9e648d2a65043bc1ad3959fc91ee3698d5.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/1602.07416v2" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

Limitations of Autoregressive Models and Their Alternatives [article]

Chu-Cheng Lin and Aaron Jaech and Xin Li and Matthew R. Gormley and Jason Eisner
<span title="2021-05-31">2021</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
Thus, simply training larger autoregressive language models is not a panacea for NLP.  ...  Indeed, they cannot even model them well enough to solve associated easy decision problems for which an engineer might want to consult a language model.  ...  Because this paper is in part a response to popular neural architectures, we now show thatp can in fact be computed efficiently by recurrent neural networks (RNNs) with compact parameters.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2010.11939v3">arXiv:2010.11939v3</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/byfd27lqzvfc7h5khjkihc7t4e">fatcat:byfd27lqzvfc7h5khjkihc7t4e</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20201025153249/https://arxiv.org/pdf/2010.11939v1.pdf" title="fulltext PDF download [not primary version]" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <span style="color: #f43e3e;">&#10033;</span> <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/54/bf/54bfaf3bc4067c483e8946dcaf4ecc5817e97522.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2010.11939v3" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

Real-time gravitational-wave science with neural posterior estimation [article]

Maximilian Dax, Stephen R. Green, Jonathan Gair, Jakob H. Macke, Alessandra Buonanno, Bernhard Schölkopf
<span title="2021-06-23">2021</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
This encodes the signal and noise models within millions of neural-network parameters, and enables inference for any observed data consistent with the training distribution, accounting for noise nonstationarity  ...  Using neural networks as surrogates for Bayesian posterior distributions, we analyze eight gravitational-wave events from the first LIGO-Virgo Gravitational-Wave Transient Catalog and find very close quantitative  ...  compact binary systems.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2106.12594v1">arXiv:2106.12594v1</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/4khwsg2rtrew7aikka34xq525a">fatcat:4khwsg2rtrew7aikka34xq525a</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20210626114338/https://arxiv.org/pdf/2106.12594v1.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/fe/1f/fe1f47edfec8c3e24df9adb8673a92611b62eda6.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2106.12594v1" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

Geometry-Aware Neural Rendering [article]

Josh Tobin, OpenAI Robotics, Pieter Abbeel
<span title="2019-10-28">2019</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
We extend existing neural rendering to more complex, higher dimensional scenes than previously possible.  ...  Neural rendering approaches learn an implicit 3D model by predicting what a camera would see from an arbitrary viewpoint.  ...  We introduce three datasets: Disco Humanoid, OpenAI Block, and Room-Random-Objects as a testbed for neural rendering with complex objects and high-dimensional state. 3.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/1911.04554v1">arXiv:1911.04554v1</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/wl4elqhay5d3tmnroko6culk4i">fatcat:wl4elqhay5d3tmnroko6culk4i</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20200908045047/https://arxiv.org/pdf/1911.04554v1.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/7f/d7/7fd7b1f36f2e8772798484625b36db8e01d4a053.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/1911.04554v1" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

An EM Approach to Non-autoregressive Conditional Sequence Generation [article]

Zhiqing Sun, Yiming Yang
<span title="2020-06-29">2020</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
In the M-step, the NAR model is updated on the new posterior and selects the training examples for the next AR model.  ...  This paper proposes a new approach that jointly optimizes both AR and NAR models in a unified Expectation-Maximization (EM) framework.  ...  Acknowledgements We thank the reviewers for their helpful comments.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2006.16378v1">arXiv:2006.16378v1</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/wpjbrofjnfh43fxp27sajafqca">fatcat:wpjbrofjnfh43fxp27sajafqca</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20200722164625/https://arxiv.org/pdf/2006.16378v1.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/3d/d6/3dd6ceabc36725fa8f8debdaa4a87ec4e35e8c22.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2006.16378v1" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

Adversarially Robust Distillation

Micah Goldblum, Liam Fowl, Soheil Feizi, Tom Goldstein
<span title="2020-04-03">2020</span> <i title="Association for the Advancement of Artificial Intelligence (AAAI)"> <a target="_blank" rel="noopener" href="https://fatcat.wiki/container/wtjcymhabjantmdtuptkk62mlq" style="color: black;">PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE</a> </i> &nbsp;
Knowledge distillation is effective for producing small, high-performance neural networks for classification, but these small networks are vulnerable to adversarial attacks.  ...  In our experiments, we find that ARD student models decisively outperform adversarially trained networks of identical architecture in terms of robust accuracy, surpassing state-of-the-art methods on standard  ...  Acknowledgments This research was generously supported by DARPA, including the GARD program, QED for RML, and the Young Faculty Award program.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener noreferrer" href="https://doi.org/10.1609/aaai.v34i04.5816">doi:10.1609/aaai.v34i04.5816</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/xxw4fvo2hfcknm3k3jcq6jt4vu">fatcat:xxw4fvo2hfcknm3k3jcq6jt4vu</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20201103211226/https://aaai.org/ojs/index.php/AAAI/article/download/5816/5672" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/8e/9e/8e9ee28ff91c3f60165591a10daef5ef91201d0f.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener noreferrer" href="https://doi.org/10.1609/aaai.v34i04.5816"> <button class="ui left aligned compact blue labeled icon button serp-button"> <i class="external alternate icon"></i> Publisher / doi.org </button> </a>

Adversarially Robust Distillation [article]

Micah Goldblum, Liam Fowl, Soheil Feizi, Tom Goldstein
<span title="2019-12-02">2019</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
Knowledge distillation is effective for producing small, high-performance neural networks for classification, but these small networks are vulnerable to adversarial attacks.  ...  In our experiments, we find that ARD student models decisively outperform adversarially trained networks of identical architecture in terms of robust accuracy, surpassing state-of-the-art methods on standard  ...  Acknowledgments This research was generously supported by DARPA, including the GARD program, QED for RML, and the Young Faculty Award program.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/1905.09747v2">arXiv:1905.09747v2</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/3ar7k3rxkzdgxnq4pprvktgroa">fatcat:3ar7k3rxkzdgxnq4pprvktgroa</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20200901104326/https://arxiv.org/pdf/1905.09747v2.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] <div class="menu fulltext-thumbnail"> <img src="https://blobs.fatcat.wiki/thumbnail/pdf/d0/07/d007f0d14f40c101f95afdce9719c411406dc721.180px.jpg" alt="fulltext thumbnail" loading="lazy"> </div> </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/1905.09747v2" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>

TraDE: Transformers for Density Estimation [article]

Rasool Fakoor, Pratik Chaudhari, Jonas Mueller, Alexander J. Smola
<span title="2020-10-14">2020</span> <i > arXiv </i> &nbsp; <span class="release-stage" >pre-print</span>
We present TraDE, a self-attention-based architecture for auto-regressive density estimation with continuous and discrete valued data.  ...  and demonstrate that TraDE works well in these scenarios.  ...  The Transformer is a neural sequence transduction model and consists of a multi-layer encoder/decoder pair. We only need the encoder for building TraDE.  ... 
<span class="external-identifiers"> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2004.02441v2">arXiv:2004.02441v2</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/ixd3hrh5uzhhljlbo7tt7m2gqi">fatcat:ixd3hrh5uzhhljlbo7tt7m2gqi</a> </span>
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20201017001612/https://arxiv.org/pdf/2004.02441v2.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] </button> </a> <a target="_blank" rel="external noopener" href="https://arxiv.org/abs/2004.02441v2" title="arxiv.org access"> <button class="ui compact blue labeled icon button serp-button"> <i class="file alternate outline icon"></i> arxiv.org </button> </a>
&laquo; Previous Showing results 1 &mdash; 15 out of 875 results