Robustness under latent distribution shift remains challenging in partially observable reinforcement learning. We formalize a focused setting where an adversary selects a hidden initial latent distribution before the episode, termed an adversarial latent-initial-state POMDP. Theoretically, we prove a latent minimax principle, characterize worst-case defender distributions, and derive approximate best-response certificates with finite-sample guarantees, providing formal meaning to empirical training diagnostics. Empirically, using a Battleship benchmark, we demonstrate that targeted exposure to shifted latent distributions reduces average robustness gaps between Spread and Uniform distributions from 10.3 to 3.1 shots at equal budget. Furthermore, iterative best-response training exhibits budget-sensitive behavior entirely consistent with our approximate certificate theory. Ultimately, we show that for latent-initial-state problems, our framework yields precise diagnostic principles and confirms that structured adversarial exposure effectively mitigates worst-case vulnerabilities.

翻译：暂无翻译