Neil Zhu，简书ID Not_GOD，University AI 创始人 & Chief Scientist，致力于推进世界人工智能化进程。制定并实施 UAI 中长期增长战略和目标，带领团队快速成长为人工智能领域最专业的力量。
作为行业领导者，他和UAI一起在2014年创建了TASA（中国最早的人工智能社团）, DL Center（深度学习知识中心全球价值网络），AI growth（行业智库培训）等，为中国的人工智能人才建设输送了大量的血液和养分。此外，他还参与或者举办过各类国际性的人工智能峰会和活动，产生了巨大的影响力，书写了60万字的人工智能精品技术内容，生产翻译了全球第一本深度学习入门书《神经网络与深度学习》，生产的内容被大量的专业垂直公众号和媒体转载与连载。曾经受邀为国内顶尖大学制定人工智能学习规划和教授人工智能前沿课程，均受学生和老师好评。
ICML 16-全部接受论文
ICML 2016 - 强化学习相关论文 如下：
1. Inverse Optimal Control with Deep Networks via Policy Optimization
Chelsea Finn, UC Berkeley; Sergey Levine, ; Pieter Abbeel, Berkeley
摘要：
http://arxiv.org/abs/1603.00448
Doubly Robust Off-policy Value Evaluation for Reinforcement Learning
Nan Jiang, University of Michigan; Lihong Li, Microsoft
http://arxiv.org/abs/1511.03722
Smooth Imitation Learning
Hoang Le, Caltech; Andrew Kang, ; Yisong Yue, Caltech; Peter Carr,
PAC Lower Bounds and Efficient Algorithms for The Max KK-Armed Bandit Problem
Yahel David, Technion; Nahum Shimkin, Technion
Anytime Exploration for Multi-armed Bandits using Confidence Information
Kwang-Sung Jun, UW-Madison; Robert Nowak,
The Knowledge Gradient for Sequential Decision Making with Stochastic Binary Feedbacks
Yingfei Wang, Princeton University; Chu Wang, ; Warren Powell,
https://arxiv.org/abs/1510.02354
Copeland Dueling Bandit Problem: Regret Lower Bound, Optimal Algorithm, and Computationally Efficient Algorithm
Junpei Komiyama, The University of Tokyo; Junya Honda, The University of Tokyo; Hiroshi Nakagawa, The University of Tokyo
https://arxiv.org/abs/1605.01677
Benchmarking Deep Reinforcement Learning for Continuous Control
Yan Duan, University of California, Berk; Xi Chen, University of California, Berkeley; Rein Houthooft, Ghent University; John Schulman, University of California, Berkeley; Pieter Abbeel, Berkeley
https://arxiv.org/abs/1604.06778
Cumulative Prospect Theory Meets Reinforcement Learning: Prediction and Control
Prashanth L.A., University of Maryland ; Cheng Jie, University of Maryland – College Park; Michael Fu, University of Maryland – College Park; Steve Marcus, University of Maryland – College Park; Csaba Szepesvari, Alberta
http://arxiv.org/abs/1506.02632
An optimal algorithm for the Thresholding Bandit Problem
Andrea LOCATELLI, University of Potsdam; Maurilio Gutzeit, Universität Potsdam; Alexandra Carpentier,
Sequential decision making under uncertainty: Are most decisions easy?
Ozgur Simsek, ; Simon Algorta, ; Amit Kothiyal,
Opponent Modeling in Deep Reinforcement Learning
He He, ; Jordan , ; Hal Daume, Maryland
Softened Approximate Policy Iteration for Markov Games
Julien Pérolat, Univ. Lille; Bilal Piot, Univ. Lille; Matthieu Geist, ; Bruno Scherrer, ; Olivier Pietquin, Univ. Lille, CRIStAL, UMR 9189, SequeL Team, Villeneuve d’Ascq, 59650, FRANCE
Asynchronous Methods for Deep Reinforcement Learning
Volodymyr Mnih, Google DeepMind; Adria Puigdomenech Badia, Google DeepMind; Mehdi Mirza, ; Alex Graves, Google DeepMind; Timothy Lillicrap, Google DeepMind; Tim Harley, Google DeepMind; David , ; Koray Kavukcuoglu, Google Deepmind
https://arxiv.org/abs/1602.01783
Dueling Network Architectures for Deep Reinforcement Learning
Ziyu Wang, Google Inc.; Nando de Freitas, University of Oxford; Tom Schaul, Google Inc.; Matteo Hessel, Google Deepmind; Hado van Hasselt, Google DeepMind; Marc Lanctot, Google Deepmind
http://arxiv.org/abs/1511.06581 Cited by 10
Differentially Private Policy Evaluation
Borja Balle, Lancaster University; Maziar Gomrokchi, McGill University; Doina Precup, McGill
https://arxiv.org/abs/1603.02010
Data-Efficient Off-Policy Policy Evaluation for Reinforcement Learning
Philip Thomas, CMU; Emma ,
https://arxiv.org/abs/1604.00923
Hierarchical Decision Making In Electricity Grid Management
Gal Dalal, Technion; Elad Gilboa, Technion; Shie Mannor, Technion
http://arxiv.org/abs/1603.01840
Generalization and Exploration via Randomized Value Functions
Ian Osband, Stanford; Ben , ; Zheng Wen, Adobe Research
https://arxiv.org/abs/1402.0635 Cited by 9
Scalable Discrete Sampling as a Multi-Armed Bandit Problem
Yutian Chen, University of Cambridge; Zoubin ,
摘要
Drawing a sample from a discrete distribution is one of the building components for Monte Carlo methods. Like other sampling algorithms, discrete sampling suffers from the high computational burden in large-scale inference problems. We study the problem of sampling a discrete random variable with a high degree of dependency that is typical in large-scale Bayesian inference and graphical models, and propose an efficient approximate solution with a subsampling approach. We make a novel connection between the discrete sampling and Multi-Armed Bandits problems with a finite reward population and provide three algorithms with theoretical guarantees. Empirical evaluations show the robustness and efficiency of the approximate algorithms in both synthetic and real-world large-scale problems.
http://arxiv.org/abs/1506.09039
Model-Free Imitation Learning with Policy Optimization
Jonathan Ho, Stanford; Jayesh Gupta, Stanford University; Stefano Ermon,
Improving the Efficiency of Deep Reinforcement Learning with Normalized Advantage Functions and Synthetic Experience
Shixiang Gu, University of Cambridge; Sergey Levine, Google; Timothy Lillicrap, Google DeepMind; Ilya Sutskever, OpenAI
http://arxiv.org/abs/1603.00748
Near Optimal Behavior via Approximate State Abstraction
David Abel, Brown University; David Hershkowitz, Brown University; Michael Littman,
https://cs.brown.edu/~dabel/papers/abel_approx_abstraction.pdf
Model-Free Trajectory Optimization for Reinforcement Learning of Motor Skills
Riad Akrour, TU Darmstadt; Gerhard Neumann
