UVIP: Model-Free Approach to Evaluate Reinforcement Learning Algorithms
This provides a tool for researchers and practitioners to compare RL algorithms more reliably, though it is incremental as it builds on existing martingale approaches.
The paper tackles the problem of evaluating reinforcement learning algorithms by introducing UVIP, a model-free method that estimates the suboptimality gap and constructs confidence intervals for the optimal value function, demonstrating performance on benchmark RL problems.
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 $V^π$ 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 ({\sf UVIP}) that allows us to estimate the suboptimality gap $V^{\star}(x) - V^π(x)$ from above and to construct confidence intervals for \(V^\star\). Our approach relies on upper bounds to the solution of the Bellman optimality equation via the martingale approach. We provide theoretical guarantees for {\sf UVIP} under general assumptions and illustrate its performance on a number of benchmark RL problems.