Oral
Improved Sample Complexity for Incremental Autonomous Exploration in MDPs
Jean Tarbouriech · Matteo Pirotta · Michal Valko · Alessandro Lazaric
Orals & Spotlights: Reinforcement Learning
Abstract:
We investigate the exploration of an unknown environment when no reward function is provided. Building on the incremental exploration setting introduced by Lim and Auer [1], we define the objective of learning the set of -optimal goal-conditioned policies attaining all states that are incrementally reachable within steps (in expectation) from a reference state . In this paper, we introduce a novel model-based approach that interleaves discovering new states from and improving the accuracy of a model estimate that is used to compute goal-conditioned policies to reach newly discovered states. The resulting algorithm, DisCo, achieves a sample complexity scaling as , where is the number of actions, is the number of states that are incrementally reachable from in steps, and is the branching factor of the dynamics over such states. This improves over the algorithm proposed in [1] in both and at the cost of an extra factor, which is small in most environments of interest. Furthermore, DisCo is the first algorithm that can return an -optimal policy for any cost-sensitive shortest-path problem defined on the -reachable states with minimum cost . Finally, we report preliminary empirical results confirming our theoretical findings.
Chat is not available.