CaltechAUTHORS
  A Caltech Library Service

Batch Policy Learning under Constraints

Le, Hoang M. and Voloshin, Cameron and Yue, Yisong (2019) Batch Policy Learning under Constraints. . (Unpublished) http://resolver.caltech.edu/CaltechAUTHORS:20190327-085845627

[img] PDF - Submitted Version
See Usage Policy.

2259Kb

Use this Persistent URL to link to this item: http://resolver.caltech.edu/CaltechAUTHORS:20190327-085845627

Abstract

When learning policies for real-world domains, two important questions arise: (i) how to efficiently use pre-collected off-policy, non-optimal behavior data; and (ii) how to mediate among different competing objectives and constraints. We thus 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 multiple constraints such as lane keeping and smooth driving. We also show experimentally that our OPE method outperforms other popular OPE techniques on a standalone basis, especially in a high-dimensional setting.


Item Type:Report or Paper (Discussion Paper)
Related URLs:
URLURL TypeDescription
http://arxiv.org/abs/1903.08738arXivDiscussion Paper
ORCID:
AuthorORCID
Yue, Yisong0000-0001-9127-1989
Record Number:CaltechAUTHORS:20190327-085845627
Persistent URL:http://resolver.caltech.edu/CaltechAUTHORS:20190327-085845627
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:94191
Collection:CaltechAUTHORS
Deposited By: George Porter
Deposited On:27 Mar 2019 22:19
Last Modified:27 Mar 2019 22:19

Repository Staff Only: item control page