Offline Reinforcement Learning with Realizability and Single-policy Concentrability
This addresses a key open problem in offline RL by relaxing strong assumptions, potentially enabling more practical applications in domains with limited data coverage.
The paper tackles the problem of achieving sample-efficient offline reinforcement learning with weak assumptions on both function classes and data coverage, and shows that a primal-dual algorithm with proper regularization achieves polynomial sample complexity under only realizability and single-policy concentrability.
Sample-efficiency guarantees for offline reinforcement learning (RL) often rely on strong assumptions on both the function classes (e.g., Bellman-completeness) and the data coverage (e.g., all-policy concentrability). Despite the recent efforts on relaxing these assumptions, existing works are only able to relax one of the two factors, leaving the strong assumption on the other factor intact. As an important open problem, can we achieve sample-efficient offline RL with weak assumptions on both factors? In this paper we answer the question in the positive. We analyze a simple algorithm based on the primal-dual formulation of MDPs, where the dual variables (discounted occupancy) are modeled using a density-ratio function against offline data. With proper regularization, we show that the algorithm enjoys polynomial sample complexity, under only realizability and single-policy concentrability. We also provide alternative analyses based on different assumptions to shed light on the nature of primal-dual algorithms for offline RL.