Batch Policy Learning under Constraints
Hoang Le · Cameron Voloshin · Yisong Yue

Thu Jun 13th 09:00 -- 09:20 AM @ Hall B

When learning policies for real-world domains, two important questions arise: (i) how to efficiently use existing off-line, off-policy, non-optimal behavior data; and (ii) how to mediate among different competing objectives and constraints. We study the problem of batch policy learning under multiple constraints and offer a systematic solution. We first propose a flexible meta algorithm that admits any batch reinforcement learning and online learning procedure as subroutines. We then present a specific algorithmic instantiation and provide performance guarantees for the main objective and all constraints. To certify constraint satisfaction, we propose a new and simple method for off-policy policy evaluation (OPE) and derive PAC-style bounds. Our algorithm achieves strong empirical results in different domains, including in a challenging problem of simulated car driving subject to lane keeping and smooth driving constraints. We also show experimentally that our OPE method outperforms other popular OPE techniques on a standalone basis, especially in a high-dimensional setting.

Author Information

Hoang Le (Caltech)

Hoang M. Le is a PhD Candidate in the Computing and Mathematical Sciences Department at the California Institute of Technology. He received a M.S. in Cognitive Systems and Interactive Media from the Universitat Pompeu Fabra, Barcelona, Spain, and a B.A. in Mathematics from Bucknell University in Lewisburg, PA. He is a recipient of an Amazon AI Fellowship. Hoang’s research focuses on the theory and applications of sequential decision making, with a strong focus on imitation learning. He has broad familiarity with the latest advances in imitation learning techniques and applications. His own research in imitation learning blends principled new techniques with a diverse range of application domains. In addition to popular reinforcement learning domains such as maze navigation and Atari games, his prior work on imitation learning has been applied to learning human behavior in team sports and developing automatic camera broadcasting system.

Cameron Voloshin (Caltech)
Yisong Yue (Caltech)

Yisong Yue is an assistant professor in the Computing and Mathematical Sciences Department at the California Institute of Technology. He was previously a research scientist at Disney Research. Before that, he was a postdoctoral researcher in the Machine Learning Department and the iLab at Carnegie Mellon University. He received a Ph.D. from Cornell University and a B.S. from the University of Illinois at Urbana-Champaign. Yisong's research interests lie primarily in the theory and application of statistical machine learning. He is particularly interested in developing novel methods for interactive machine learning and structured prediction. In the past, his research has been applied to information retrieval, recommender systems, text classification, learning from rich user interfaces, analyzing implicit human feedback, data-driven animation, behavior analysis, sports analytics, policy learning in robotics, and adaptive planning & allocation problems.

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors