HDI Lab seminar UVIP: Model Free Approach to Evaluate Reinforcement Learning Algorithms

(презентация на русском языке) Policy evaluation is an important instrument for the comparison of different algorithms in Reinforcement Learning (RL). Yet even a precise knowledge of the value function Vpi corresponding to a policy pi does not provide reliable information on how far is the policy pi from the optimal one. We present a novel modelfree uppervalue iteration procedure (UVIP) that allows us to estimate the suboptimality gap V(x) Vpi(x) from above and to construct confidence intervals for V. Our approach relies on upper bounds to the solution of the Bellman optimality equation via the martingale approach. We provide theoretical guarantees for UVIP under general assumptions and illustrate its performance on a number of benchmark RL problems. The talk is based on our recent work with Denis Belomestny, Ilya Levin, Eric Moulines, Alexey Naumov and Veronika Zorina. Speaker: Sergey Samsonov, Lecturer at the Big Data and Information Retrieval School. June 1, 2021 HDI Lab: