Finite-Time Bounds for Distributionally Robust TD Learning with Linear Function Approximation
This work addresses a key gap in theoretical guarantees for robust RL algorithms, providing foundational convergence results for policy evaluation with function approximation, which is incremental but closes an important theoretical gap.
The paper tackles the problem of robust temporal-difference learning with linear function approximation in reinforcement learning under model uncertainties, establishing an O~(1/ε²) sample complexity to achieve an ε-accurate value estimate.
Distributionally robust reinforcement learning (DRRL) focuses on designing policies that achieve good performance under model uncertainties. In particular, we are interested in maximizing the worst-case long-term discounted reward, where the data for RL comes from a nominal model while the deployed environment can deviate from the nominal model within a prescribed uncertainty set. Existing convergence guarantees for robust temporal-difference (TD) learning for policy evaluation are limited to tabular MDPs or are dependent on restrictive discount-factor assumptions when function approximation is used. We present the first robust TD learning with linear function approximation, where robustness is measured with respect to the total-variation distance and Wasserstein-l distance uncertainty set. Additionally, our algorithm is both model-free and does not require generative access to the MDP. Our algorithm combines a two-time-scale stochastic-approximation update with an outer-loop target-network update. We establish an $\tilde{O}(1/ε^2)$ sample complexity to obtain an $ε$-accurate value estimate. Our results close a key gap between the empirical success of robust RL algorithms and the non-asymptotic guarantees enjoyed by their non-robust counterparts. The key ideas in the paper also extend in a relatively straightforward fashion to robust Q-learning with function approximation.