Imitation learning (IL) aims to learn a policy from expert demonstrations that minimizes the discrepancy between the learner and expert behaviors. Various imitation learning algorithms have been proposed with different pre-determined divergences to quantify the discrepancy. This naturally gives rise to the following question: Given a set of expert demonstrations, which divergence can recover the expert policy more accurately with higher data efficiency? In this work, we propose f-GAIL – a new generative adversarial imitation learning model – that automatically learns a discrepancy measure from the f-divergence family as well as a policy capable of producing expert-like behaviors. Compared with IL baselines with various predefined divergence measures, f-GAIL learns better policies with higher data efficiency in six physics-based control tasks.
Xin Zhang (Worcester Polytechnic Institute)
Yanhua Li ("Worcester Polytechnic Institute, USA")
Ziming Zhang (Worcester Polytechnic Institute)
Zhi-Li Zhang (University of Minnesota)
Professor Zhang is McKnight Distinguished University Professor and Qwest Chair Professor at Department of Computer Science & Engineering, University of Minnesota. He received his B.S. degree in Computer Science from Nanjing University, China, and his M.S. and Ph.D. degrees in Computer Science from the University of Massachusetts, Amherst. Dr. Zhang’s research interests lie broadly in computer communication and networks, Internet technology, multimedia and emerging applications. His past research was centered on the analysis, design and development of scalable Internet QoS solutions to support performance-demanding multimedia applications. His current research focuses on building highly scalable, resilient and secure Internet and cyber-physical systems & infrastructures and developing mechanisms to enhance Internet service availability, reliability and security, and on developing next generation, service-oriented, manageable Internet architectures to provide better support for creation, deployment, operations and management of value-added smart services and underlying networks, including mobile, cloud and content delivery services and networks. Dr. Zhang has published more than 150 journal, conference and workshop papers. Dr. Zhang has received several honors for his research, and he is co-recipient of a number of Best Paper Awards. He is a Fellow of IEEE.
More from the Same Authors
2021 Poster: SBO-RNN: Reformulating Recurrent Neural Networks via Stochastic Bilevel Optimization »
Ziming Zhang · Yun Yue · Guojun Wu · Yanhua Li · Haichong Zhang
2020 Poster: Adaptive Reduced Rank Regression »
Qiong Wu · Felix MF Wong · Yanhua Li · Zhenming Liu · Varun Kanade
2017 Poster: Hunt For The Unique, Stable, Sparse And Fast Feature Learning On Graphs »
Saurabh Verma · Zhi-Li Zhang