Beyond Worst-case Attacks: Robust RL with Adaptive Defense via Non-dominated Policies

arXiv:2402.12673v1 Announce Type: new
Abstract: In light of the burgeoning success of reinforcement learning (RL) in diverse real-world applications, considerable focus has been directed towards ensuring RL policies are robust to adversarial attacks during test time. Current approaches largely revolve around solving a minimax problem to prepare for potential worst-case scenarios. While effective against strong attacks, these methods often compromise performance in the absence of attacks or the presence of only weak attacks. To address this, we study policy robustness …

abstract adversarial adversarial attacks applications arxiv attacks beyond case cs.lg current defense diverse focus light minimax reinforcement reinforcement learning robust success test type via world