EUVIMEDEUVIMED entdecken
Crossref

Reward Poisoning Attacks on Offline Multi-Agent Reinforcement Learning

Young Wu, Jeremy McMahan, Xiaojin Zhu, Qiaomin Xie

Proceedings of the AAAI Conference on Artificial Intelligence · 26.06.2023 · 37 (9) · 10426-10434

DOI: 10.1609/aaai.v37i9.26240 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 26.06.2023. Verfasst von Young Wu und 3 weiteren Autor:innen. Die Arbeit wurde bisher 8-mal zitiert (Crossref).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

In offline multi-agent reinforcement learning (MARL), agents estimate policies from a given dataset. We study reward-poisoning attacks in this setting where an exogenous attacker modifies the rewards in the dataset before the agents see the dataset. The attacker wants to guide each agent into a nefarious target policy while minimizing the Lp norm of the reward modification. Unlike attacks on single-agent RL, we show that the attacker can install the target policy as a Markov Perfect Dominant Strategy Equilibrium (MPDSE), which rational agents are guaranteed to follow. This attack can be significantly cheaper than separate single-agent attacks. We show that the attack works on various MARL agents including uncertainty-aware learners, and we exhibit linear programs to efficiently solve the attack problem. We also study the relationship between the structure of the datasets and the minimal attack cost. Our work paves the way for studying defense in offline MARL.

Zitieren

APA

Wu, Y., McMahan, J., Zhu, X., & Xie, Q. (2023). Reward Poisoning Attacks on Offline Multi-Agent Reinforcement Learning. Proceedings of the AAAI Conference on Artificial Intelligence, 37(9), 10426-10434. https://doi.org/10.1609/aaai.v37i9.26240

Vancouver

Wu Y, McMahan J, Zhu X, Xie Q. Reward Poisoning Attacks on Offline Multi-Agent Reinforcement Learning. Proceedings of the AAAI Conference on Artificial Intelligence. 2023;37(9):10426-10434. doi:10.1609/aaai.v37i9.26240

BibTeX
@article{Wu2023,
  author = {Wu, Young and McMahan, Jeremy and Zhu, Xiaojin and Xie, Qiaomin},
  title = {Reward Poisoning Attacks on Offline Multi-Agent Reinforcement Learning},
  journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
  year = {2023},
  volume = {37},
  number = {9},
  pages = {10426--10434},
  doi = {10.1609/aaai.v37i9.26240}
}
Kostenlos starten →