Abstract
Most adversarial attacks on deep reinforcement learning (DRL) assume white-box access to the victim policy, which rarely holds in practice. This paper studies transfer-based black-box attacks on DRL: the attacker crafts observation perturbations on a white-box surrogate agent and feeds them to an unknown victim. We formulate the attack as return minimization under a per-step perturbation budget. We first show that transplanting transferable image-classification attacks (FGSM, MI-FGSM, and NI-FGSM) with a per-step objective yields perturbations that transfer but are no stronger than random noise of the same budget. We then propose a trajectory-level attack that optimizes a sequence of perturbations over a receding horizon through a differentiable model of the environment and a temperature-smoothed surrogate policy, with the same optimizers. On CartPole-v1 with ten DQN and DDQN agents and 100 surrogate--victim pairs, the trajectory-level attack outperforms per-step attacks and random noise in the white-box, cross-model, and cross-algorithm settings.
Keywords
Publication details
- Journal
- Not available
- Open access
- Green open access
Cite this article
APA 7
Li, Z., Yao, R., Zeng, Y., & Gao, X. (2026). Boosting Transferable Adversarial Attacks against Deep Reinforcement Learning. https://omanscience.com/en/articles/boosting-transferable-adversarial-attacks-against-deep-reinforcement-learning
MLA 9
Li, Zexin, et al. "Boosting Transferable Adversarial Attacks against Deep Reinforcement Learning." https://omanscience.com/en/articles/boosting-transferable-adversarial-attacks-against-deep-reinforcement-learning.
Chicago (author–date)
Li, Zexin, Ruili Yao, Yiming Zeng, and Xiaoxue Gao. 2026. "Boosting Transferable Adversarial Attacks against Deep Reinforcement Learning." https://omanscience.com/en/articles/boosting-transferable-adversarial-attacks-against-deep-reinforcement-learning.
Harvard
Li, Z., Yao, R., Zeng, Y. and Gao, X. (2026) 'Boosting Transferable Adversarial Attacks against Deep Reinforcement Learning', Available at: https://omanscience.com/en/articles/boosting-transferable-adversarial-attacks-against-deep-reinforcement-learning.
Vancouver
Li Z, Yao R, Zeng Y, Gao X. Boosting Transferable Adversarial Attacks against Deep Reinforcement Learning. https://omanscience.com/en/articles/boosting-transferable-adversarial-attacks-against-deep-reinforcement-learning
IEEE
Z. Li, R. Yao, Y. Zeng, and X. Gao, "Boosting Transferable Adversarial Attacks against Deep Reinforcement Learning," https://omanscience.com/en/articles/boosting-transferable-adversarial-attacks-against-deep-reinforcement-learning.