Provably Efficient Model-Free Algorithms for Non-stationary CMDPs
This addresses the challenge of safe and efficient reinforcement learning in dynamic environments with constraints, which is crucial for real-world applications like robotics and autonomous systems, representing a significant but incremental advance over stationary CMDP methods.
The paper tackles the problem of model-free reinforcement learning in non-stationary constrained Markov Decision Processes, where reward, utility, and transition functions vary over time, and proposes the first model-free algorithms with sublinear regret and zero constraint violation for tabular and linear settings, matching best results for stationary cases when the variation budget is known.
We study model-free reinforcement learning (RL) algorithms in episodic non-stationary constrained Markov Decision Processes (CMDPs), in which an agent aims to maximize the expected cumulative reward subject to a cumulative constraint on the expected utility (cost). In the non-stationary environment, reward, utility functions, and transition kernels can vary arbitrarily over time as long as the cumulative variations do not exceed certain variation budgets. We propose the first model-free, simulator-free RL algorithms with sublinear regret and zero constraint violation for non-stationary CMDPs in both tabular and linear function approximation settings with provable performance guarantees. Our results on regret bound and constraint violation for the tabular case match the corresponding best results for stationary CMDPs when the total budget is known. Additionally, we present a general framework for addressing the well-known challenges associated with analyzing non-stationary CMDPs, without requiring prior knowledge of the variation budget. We apply the approach for both tabular and linear approximation settings.