{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/whatever-does-not-kill-deep-reinforcement","title":"Whatever Does Not Kill Deep Reinforcement Learning, Makes It Stronger","arxiv_id":"1712.09344","date":"2017-12-23","proceeding":null,"authors":["Vahid Behzadan","Arslan Munir"],"abstract":"Recent developments have established the vulnerability of deep Reinforcement\nLearning (RL) to policy manipulation attacks via adversarial perturbations. In\nthis paper, we investigate the robustness and resilience of deep RL to\ntraining-time and test-time attacks. Through experimental results, we\ndemonstrate that under noncontiguous training-time attacks, Deep Q-Network\n(DQN) agents can recover and adapt to the adversarial conditions by reactively\nadjusting the policy. Our results also show that policies learned under\nadversarial perturbations are more robust to test-time attacks. Furthermore, we\ncompare the performance of $\\epsilon$-greedy and parameter-space noise\nexploration methods in terms of robustness and resilience against adversarial\nperturbations.","url_abs":"http://arxiv.org/abs/1712.09344v1","url_pdf":"http://arxiv.org/pdf/1712.09344v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"whatever-does-not-kill-deep-reinforcement","repo_url":"https://github.com/behzadanksu/rl-attack","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}},{"paper_slug":"whatever-does-not-kill-deep-reinforcement","repo_url":"https://github.com/behzadanksu/rlattack-dev","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}},{"paper_slug":"whatever-does-not-kill-deep-reinforcement","repo_url":"https://github.com/chenhongge/SA_DQN","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"whatever-does-not-kill-deep-reinforcement","repo_url":"https://github.com/elytopia/Info-Sec-Hammerer-and-Brunori","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"deep-reinforcement-learning","task_name":"Deep Reinforcement Learning"},{"task_slug":"reinforcement-learning","task_name":"Reinforcement Learning"},{"task_slug":"reinforcement-learning-1","task_name":"Reinforcement Learning (RL)"},{"task_slug":"reinforcement-learning-2","task_name":"reinforcement-learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1712.09344","atlas_url":"https://app.syntology.ai/?focus=1712.09344","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}