Rethinking Adversarial Attacks In Reinforcement Learning From Policy Distribution Perspective

Abstract

Deep Reinforcement Learning (DRL) suffers from uncertainties and inaccuracies in the observation signal in realworld applications. Adversarial attack is an effective method for evaluating the robustness of DRL agents. However, existing attack methods targeting individual sampled actions have limited impacts on the overall policy distribution, particularly in continuous action spaces. To address these limitations, we propose the Distribution-Aware Projected Gradient Descent attack (DAPGD). DAPGD uses distribution similarity as the gradient perturbation input to attack the policy network, which leverages the entire policy distribution rather than relying on individual samples. We utilize the Bhattacharyya distance in DAPGD to measure policy similarity, enabling sensitive detection of subtle but critical differences between probability distributions. Our experiment results demonstrate that DAPGD achieves SOTA results compared to the baselines in three robot navigation tasks, achieving an

Stats

• citations5
• S2 citations—
• github stars0
• HF likes0
• heat score5.84
• arxiv keyduan2025rethinking

Related papers

• Adversarial Policies: Attacking Deep Reinforcement Learning (2019)0.00
• Real-time Adversarial Perturbations Against Deep Reinforcement Learning Policies: Attacks And Defenses (2021)0.00
• Attacking And Defending Deep Reinforcement Learning Policies (2022)0.00
• Adversary Agnostic Robust Deep Reinforcement Learning (2020)6.77
• Robust Deep Reinforcement Learning Through Adversarial Attacks And Training : A Survey (2024)0.00
• Online Robust Policy Learning In The Presence Of Unknown Adversaries (2018)0.00
• Regret-based Defense In Adversarial Reinforcement Learning (2023)0.00
• Query-based Targeted Action-space Adversarial Policies On Deep Reinforcement Learning Agents (2020)0.00