一文览尽谷歌 ICML 2019 成果

AI研习社

发布于 2019-06-19 14:29:43

8750

发布于 2019-06-19 14:29:43

ICML 2019结束后，大家是否意犹未尽呢？雷锋字幕组成员为你编译整理了 Google ICML 2019 成果集锦，供大家查看学习。翻译 | 汪鹏、敬爱的勇哥编辑 | 王立鱼原文链接： https://ai.googleblog.com/2019/06/google-at-icml-2019.html 注：文中所示部分录用论文为AI研习社精选论文，完整论文合集请点击阅读原文查看~

机器学习是Google的一个关键战略重点，高度活跃的团队致力于该领域中包括深度学习、经典的算法、探索理论和应用等几乎所有方面的研究。我们利用可扩展的工具和体系结构来构建机器学习系统，使我们能够解决语言、语音、翻译、音乐、视觉处理等领域的科学和工程难题。

作为一个在机器学习研究中的领导者，Google是三十六届机器学习国际会议(ICML 2019)的蓝宝石赞助商(Sapphire Sponsor)，该会议是国际机器学习领域的年度盛会，今年的会议将会于本周在加州长滩举办。将有近200名Google员工在会议上发表论文或参与到会后的研讨会中，我们期待与这个巨大的机器学习研究社区的继续合作。

如果你参加ICML 2019，希望你能来到Google展台了解那些激动人心的工作，看到我们创造性和趣味性的去解决该领域的一些最有趣的挑战，与研究人员一起讨论Google研究足球环境，AdaNet，Google机器人等等。在下面的列表中，您还可以了解更多关于Google论文的信息，这些研究结果将在ICML 2019上发表(Google的附属单位或组织用蓝色高亮显示)。

ICML 2019 委员会

Board Members include: Andrew McCallum, Corinna Cortes, Hugo Larochelle, William Cohen(Emeritus)

Senior Area Chairs include: Charles Sutton, Claudio Gentile, Corinna Cortes, Kevin Murphy, Mehryar Mohri, Nati Srebro, Samy Bengio, Surya Ganguli

Area Chairs include: Jacob Abernethy, William Cohen, Dumitru Erhan, Cho-Jui Hsieh, Chelsea Finn, Sergey Levine, Manzil Zaheer, Sergei Vassilvitskii, Boqing Gong, Been Kim, Dale Schuurmans, Danny Tarlow, Dustin Tran, Hanie Sedghi, Honglak Lee, Jasper Snoek, Lihong Li, Minmin Chen, Mohammad Norouzi, Nicolas Le Roux, Phil Long, Sanmi Koyejo, Timnit Gebru, Vitaly Feldman, Satyen Kale, Katherine Heller, Hossein Mobahi, Amir Globerson, Ilya Tolstikhin, Marco Cuturi, Sebastian Nowozin, Amin Karbasi, Ohad Shamir, Graham Taylor

录用论文

Challenging Common Assumptions in the Unsupervised Learning of Disentangled Representations

作者：Francesco Locatello, Stefan Bauer, Mario Lučić, Gunnar Rätsch, Sylvain Gelly, Bernhard Schölkopf,Olivier Bachem(Recipient of an ICML 2019 Best Paper Award)
论文地址：http://proceedings.mlr.press/v97/locatello19a/locatello19a.pdf

Learning to Groove with Inverse Sequence Transformations

作者：Jon Gillick, Adam Roberts, Jesse Engel, Douglas Eck, David Bamman
论文地址：http://proceedings.mlr.press/v97/gillick19a/gillick19a.pdf

Metric-Optimized Example Weights

作者：Sen Zhao, Mahdi Milani Fard, Harikrishna Narasimhan, Maya Gupta
论文地址：http://proceedings.mlr.press/v97/zhao19b/zhao19b.pdf

HOList: An Environment for Machine Learning of Higher Order Logic Theorem Proving

作者：Kshitij Bansal, Sarah Loos, Markus Rabe, Christian Szegedy, Stewart Wilcox
论文地址：http://proceedings.mlr.press/v97/bansal19a/bansal19a.pdf

Learning to Clear the Market

作者：Weiran Shen, Sebastien Lahaie, Renato Paes Leme
论文地址：http://proceedings.mlr.press/v97/shen19b/shen19b.pdf

Shape Constraints for Set Functions

作者：Andrew Cotter, Maya Gupta, Heinrich Jiang, Erez Louidor, James Muller, Tamann Narayan, Serena Wang, Tao Zhu
论文地址：http://proceedings.mlr.press/v97/cotter19a/cotter19a.pdf

Self-Attention Generative Adversarial Networks

作者：Han Zhang, Ian Goodfellow, Dimitris Metaxas, Augustus Odena
论文地址：http://proceedings.mlr.press/v97/zhang19d/zhang19d.pdf

High-Fidelity Image Generation With Fewer Labels

作者：Mario Lučić, Michael Tschannen, Marvin Ritter, Xiaohua Zhai, Olivier Bachem, Sylvain Gelly
论文地址：http://proceedings.mlr.press/v97/lucic19a/lucic19a.pdf
Learning Optimal Linear Regularizers
作者：Matthew Streeter
论文地址：http://proceedings.mlr.press/v97/streeter19a/streeter19a.pdf

DeepMDP: Learning Continuous Latent Space Models for Representation Learning

作者：Carles Gelada, Saurabh Kumar, Jacob Buckman, Ofir Nachum, Marc G. Bellemare
论文地址：http://proceedings.mlr.press/v97/gelada19a/gelada19a.pdf

kernelPSI: a Post-Selection Inference Framework for Nonlinear Variable Selection

作者：Lotfi Slim, Clément Chatelain, Chloe-Agathe Azencott, Jean-Philippe Vert
论文地址：http://proceedings.mlr.press/v97/slim19a/slim19a.pdf

Learning from a Learner

作者：Alexis Jacq, Matthieu Geist, Ana Paiva, Olivier Pietquin
论文地址：http://proceedings.mlr.press/v97/jacq19a/jacq19a.pdf

Rate Distortion For Model Compression:From Theory To Practice

作者：Weihao Gao, Yu-Han Liu, Chong Wang, Sewoong Oh
论文地址：http://proceedings.mlr.press/v97/gao19c/gao19c.pdf

An Investigation into Neural Net Optimization via Hessian Eigenvalue Density

作者：Behrooz Ghorbani, Shankar Krishnan, Ying Xiao
论文地址：http://proceedings.mlr.press/v97/ghorbani19b/ghorbani19b.pdf

Graph Matching Networks for Learning the Similarity of Graph Structured Objects

作者：Yujia Li, Chenjie Gu, Thomas Dullien, Oriol Vinyals, Pushmeet Kohli
论文地址：http://proceedings.mlr.press/v97/li19d/li19d.pdf

Subspace Robust Wasserstein Distances

作者：François-Pierre Paty, Marco Cuturi
论文地址：http://proceedings.mlr.press/v97/paty19a/paty19a.pdf

Training Well-Generalizing Classifiers for Fairness Metrics and Other Data-Dependent Constraints

作者：Andrew Cotter, Maya Gupta, Heinrich Jiang, Nathan Srebro, Karthik Sridharan, Serena Wang, Blake Woodworth, Seungil You
论文地址：http://proceedings.mlr.press/v97/cotter19b/cotter19b.pdf

The Effect of Network Width on Stochastic Gradient Descent and Generalization: an Empirical Study

作者：Daniel Park, Jascha Sohl-Dickstein, Quoc Le, Samuel Smith
论文地址：http://proceedings.mlr.press/v97/park19b/park19b.pdf

A Theory of Regularized Markov Decision Processes

作者：Matthieu Geist, Bruno Scherrer, Olivier Pietquin
论文地址：http://proceedings.mlr.press/v97/geist19a/geist19a.pdf

Area Attention

作者：Yang Li, Łukasz Kaiser, Samy Bengio, Si Si
论文地址：http://proceedings.mlr.press/v97/li19e/li19e.pdf

EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks

作者：Mingxing Tan, Quoc Le
论文地址：http://proceedings.mlr.press/v97/tan19a/tan19a.pdf

Static Automatic Batching In TensorFlow

作者：Ashish Agarwal
论文地址：http://proceedings.mlr.press/v97/agarwal19a/agarwal19a.pdf

The Evolved Transformer

作者：David So, Quoc Le, Chen Liang
论文地址：http://proceedings.mlr.press/v97/so19a/so19a.pdf

Policy Certificates: Towards Accountable Reinforcement Learning

作者：Christoph Dann, Lihong Li, Wei Wei, Emma Brunskill
论文地址：http://proceedings.mlr.press/v97/dann19a/dann19a.pdf

Self-similar Epochs: Value in Arrangement

作者：Eliav Buchnik, Edith Cohen, Avinatan Hasidim, Yossi Matias

论文地址：http://proceedings.mlr.press/v97/buchnik19a/buchnik19a.pdf

The Value Function Polytope in Reinforcement Learning

作者：Robert Dadashi, Marc G. Bellemare, Adrien Ali Taiga, Nicolas Le Roux, Dale Schuurmans
论文地址：http://proceedings.mlr.press/v97/dadashi19a/dadashi19a.pdf

Adversarial Examples Are a Natural Consequence of Test Error in Noise

作者：Justin Gilmer, Nicolas Ford, Nicholas Carlini, Ekin Cubuk
论文地址：http://proceedings.mlr.press/v97/gilmer19a/gilmer19a.pdf

SOLAR: Deep Structured Representations for Model-Based Reinforcement Learning

作者：Marvin Zhang, Sharad Vikram, Laura Smith, Pieter Abbeel, Matthew Johnson, Sergey Levine
论文地址：http://proceedings.mlr.press/v97/zhang19m/zhang19m.pdf

Garbage In, Reward Out: Bootstrapping Exploration in Multi-Armed Bandits

作者：Branislav Kveton, Csaba Szepesvari, Sharan Vaswani, Zheng Wen, Tor Lattimore, Mohammad Ghavamzadeh
论文地址：http://proceedings.mlr.press/v97/kveton19a/kveton19a.pdf

Imperceptible, Robust, and Targeted Adversarial Examples for Automatic Speech Recognition

作者：Yao Qin, Nicholas Carlini, Garrison Cottrell, Ian Goodfellow, Colin Raffel
论文地址：http://proceedings.mlr.press/v97/qin19a/qin19a.pdf

Direct Uncertainty Prediction for Medical Second Opinions

作者：Maithra Raghu, Katy Blumer, Rory Sayres, Ziad Obermeyer, Bobby Kleinberg, Sendhil Mullainathan, Jon Kleinberg
论文地址：http://proceedings.mlr.press/v97/raghu19a/raghu19a.pdf

A Large-Scale Study on Regularization and Normalization in GANs

作者：Karol Kurach, Mario Lučić, Xiaohua Zhai, Marcin Michalski, Sylvain Gelly
论文地址：http://proceedings.mlr.press/v97/kurach19a/kurach19a.pdf

Learning a Compressed Sensing Measurement Matrix via Gradient Unrolling

作者：Shanshan Wu, Alex Dimakis, Sujay Sanghavi, Felix Yu, Daniel Holtmann-Rice, Dmitry Storcheus, Afshin Rostamizadeh, Sanjiv Kumar
论文地址：http://proceedings.mlr.press/v97/wu19b/wu19b.pdf

NATTACK: Learning the Distributions of Adversarial Examples for an Improved Black-Box Attack on Deep Neural Networks

作者：Yandong Li, Lijun Li, Liqiang Wang, Tong Zhang, Boqing Gong
论文地址：http://proceedings.mlr.press/v97/li19g/li19g.pdf

Distributed Weighted Matching via Randomized Composable Coresets

作者：Sepehr Assadi, Mohammad Hossein Bateni, Vahab Mirrokni
论文地址：http://proceedings.mlr.press/v97/assadi19a/assadi19a.pdf

Monge blunts Bayes: Hardness Results for Adversarial Training

作者：Zac Cranko, Aditya Menon, Richard Nock, Cheng Soon Ong, Zhan Shi, Christian Walder
论文地址：http://proceedings.mlr.press/v97/cranko19a/cranko19a.pdf

Generalized Majorization-Minimization

作者：Sobhan Naderi Parizi, Kun He, Reza Aghajani, Stan Sclaroff, Pedro Felzenszwalb
论文地址：http://proceedings.mlr.press/v97/parizi19a/parizi19a.pdf

NAS-Bench-101: Towards Reproducible Neural Architecture Search

作者：Chris Ying, Aaron Klein, Eric Christiansen, Esteban Real, Kevin Murphy, Frank Hutter
论文地址：http://proceedings.mlr.press/v97/ying19a/ying19a.pdf

Variational Russian Roulette for Deep Bayesian Nonparametrics

作者：Kai Xu, Akash Srivastava, Charles Sutton
论文地址：http://proceedings.mlr.press/v97/xu19e/xu19e.pdf

Surrogate Losses for Online Learning of Stepsizes in Stochastic Non-Convex Optimization

作者：Zhenxun Zhuang, Ashok Cutkosky, Francesco Orabona
论文地址：http://proceedings.mlr.press/v97/zhuang19a/zhuang19a.pdf

Improved Parallel Algorithms for Density-Based Network Clustering

作者：Mohsen Ghaffari, Silvio Lattanzi, Slobodan Mitrović
论文地址：http://proceedings.mlr.press/v97/ghaffari19a/ghaffari19a.pdf

The Advantages of Multiple Classes for Reducing Overfitting from Test Set Reuse

作者：Vitaly Feldman, Roy Frostig, Moritz Hardt
论文地址：http://proceedings.mlr.press/v97/feldman19a/feldman19a.pdf

Submodular Streaming in All Its Glory: Tight Approximation, Minimum Memory and Low Adaptive Complexity

作者：Ehsan Kazemi, Marko Mitrovic, Morteza Zadimoghaddam, Silvio Lattanzi, Amin Karbasi
论文地址：http://proceedings.mlr.press/v97/kazemi19a/kazemi19a.pdf

Hiring Under Uncertainty

作者：Manish Purohit, Sreenivas Gollapudi, Manish Raghavan
论文地址：http://proceedings.mlr.press/v97/purohit19a/purohit19a.pdf

A Tree-Based Method for Fast Repeated Sampling of Determinantal Point Processes

作者：Jennifer Gillenwater, Alex Kulesza, Zelda Mariet, Sergei Vassilvtiskii
论文地址：http://proceedings.mlr.press/v97/gillenwater19a/gillenwater19a.pdf

Statistics and Samples in Distributional Reinforcement Learning

作者：Mark Rowland, Robert Dadashi, Saurabh Kumar, Remi Munos, Marc G. Bellemare, Will Dabney
论文地址：http://proceedings.mlr.press/v97/rowland19a/rowland19a.pdf

Provably Efficient Maximum Entropy Exploration

作者：Elad Hazan, Sham Kakade, Karan Singh, Abby Van Soest
论文地址：http://proceedings.mlr.press/v97/hazan19a/hazan19a.pdf

Active Learning with Disagreement Graphs

作者：Corinna Cortes, Giulia DeSalvo,, Mehryar Mohri, Ningshan Zhang, Claudio Gentile
论文地址：http://proceedings.mlr.press/v97/cortes19b/cortes19b.pdf

MixHop: Higher-Order Graph Convolutional Architectures via Sparsified Neighborhood Mixing

作者：Sami Abu-El-Haija, Bryan Perozzi, Amol Kapoor, Nazanin Alipourfard, Kristina Lerman, Hrayr Harutyunyan, Greg Ver Steeg, Aram Galstyan
论文地址：http://proceedings.mlr.press/v97/abu-el-haija19a/abu-el-haija19a.pdf

Understanding the Impact of Entropy on Policy Optimization

作者：Zafarali Ahmed, Nicolas Le Roux, Mohammad Norouzi, Dale Schuurmans
论文地址：http://proceedings.mlr.press/v97/ahmed19a/ahmed19a.pdf

Matrix-Free Preconditioning in Online Learning

作者：Ashok Cutkosky, Tamas Sarlos
论文地址：http://proceedings.mlr.press/v97/cutkosky19b/cutkosky19b.pdf

State-Reification Networks: Improving Generalization by Modeling the Distribution of Hidden Representations

作者：Alex Lamb, Jonathan Binas, Anirudh Goyal, Sandeep Subramanian, Ioannis Mitliagkas, Yoshua Bengio, Michael Mozer
论文地址：http://proceedings.mlr.press/v97/lamb19a/lamb19a.pdf

Online Convex Optimization in Adversarial Markov Decision Processes

作者：Aviv Rosenberg, Yishay Mansour
论文地址：http://proceedings.mlr.press/v97/rosenberg19a/rosenberg19a.pdf

Bounding User Contributions: A Bias-Variance Trade-off in Differential Privacy

作者：Kareem Amin, Alex Kulesza, Andres Munoz Medina, Sergei Vassilvtiskii
论文地址：http://proceedings.mlr.press/v97/amin19a/amin19a.pdf

Complementary-Label Learning for Arbitrary Losses and Models

作者：Takashi Ishida, Gang Niu, Aditya Menon, Masashi Sugiyama
论文地址：http://proceedings.mlr.press/v97/ishida19a/ishida19a.pdf

Learning Latent Dynamics for Planning from Pixels

作者：Danijar Hafner, Timothy Lillicrap, Ian Fischer, Ruben Villegas, David Ha, Honglak Lee, James Davidson
论文地址：http://proceedings.mlr.press/v97/hafner19a/hafner19a.pdf

Unifying Orthogonal Monte Carlo Methods