UVIP: Model-Free Approach to Evaluate Reinforcement Learning Algorithms

Journal of Optimization Theory and Applications. 2026. Vol. 208. Article 89.

Belomestny D., Levin I., Naumov A., Samsonov S.

Policy evaluation is an important instrument for the comparison of different algorithms in Reinforcement Learning (RL). However, even a precise knowledge of the value function Vπ corresponding to a policy π does not provide reliable information on how far the policy π is from the optimal one. We present a novel model-free upper value iteration procedure (UVIP) that allows us to estimate the suboptimality gap V(x) − Vπ (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. Communicated by Alexander Vladimirovich Gasnikov

Keywords: policy evaluation, policy error, confidence intervals for optimal value function, Reinforcement learning, Model-free Algorithm