Accelerating Safe Reinforcement Learning with Constraint-mismatched Baseline Policies

Tsung Yen Yang, Justinian Rosca, Karthik Narasimhan, Peter J. Ramadge

Research output: Chapter in Book/Report/Conference proceedingConference contribution

8 Scopus citations

Abstract

We consider the problem of reinforcement learning when provided with (1) a baseline control policy and (2) a set of constraints that the learner must satisfy. The baseline policy can arise from demonstration data or a teacher agent and may provide useful cues for learning, but it might also be sub-optimal for the task at hand, and is not guaranteed to satisfy the specified constraints, which might encode safety, fairness or other application-specific requirements. In order to safely learn from baseline policies, we propose an iterative policy optimization algorithm that alternates between maximizing expected return on the task, minimizing distance to the baseline policy, and projecting the policy onto the constraint-satisfying set. We analyze our algorithm theoretically and provide a finite-time convergence guarantee. In our experiments on five different control tasks, our algorithm consistently outperforms several state-of-the-art baselines, achieving 10 times fewer constraint violations and 40% higher reward on average.

Original languageEnglish (US)
Title of host publicationProceedings of the 38th International Conference on Machine Learning, ICML 2021
PublisherML Research Press
Pages11795-11807
Number of pages13
ISBN (Electronic)9781713845065
StatePublished - 2021
Event38th International Conference on Machine Learning, ICML 2021 - Virtual, Online
Duration: Jul 18 2021Jul 24 2021

Publication series

NameProceedings of Machine Learning Research
Volume139
ISSN (Electronic)2640-3498

Conference

Conference38th International Conference on Machine Learning, ICML 2021
CityVirtual, Online
Period7/18/217/24/21

All Science Journal Classification (ASJC) codes

  • Artificial Intelligence
  • Software
  • Control and Systems Engineering
  • Statistics and Probability

Fingerprint

Dive into the research topics of 'Accelerating Safe Reinforcement Learning with Constraint-mismatched Baseline Policies'. Together they form a unique fingerprint.

Cite this