The Library
Understanding the limits of poisoning attacks in episodic reinforcement learning
Tools
Rangi, Anshuka, Xu, Haifeng, Tran-Thanh, Long and Franceschetti, Massimo (2022) Understanding the limits of poisoning attacks in episodic reinforcement learning. In: International Joint Conference on Artificial Intelligence (IJCAI 2022), Vienna, Austria, 25-29 Jul 2022. Published in: Proceedings of the 31st International Joint Conference on Artificial Intelligence (IJCAI 2022) pp. 3394-3400. doi:10.24963/ijcai.2022/471
|
PDF
WRAP-Understanding-the-limits-of-poisoning-attacks-in-episodic-reinforcement-learning-Tran-Thanh-22.pdf - Accepted Version - Requires a PDF viewer. Download (461Kb) | Preview |
Official URL: https://doi.org/10.24963/ijcai.2022/471
Abstract
To understand the security threats to reinforcement learning (RL) algorithms, this paper studies poisoning attacks to manipulate any order-optimal learning algorithm towards a targeted policy in episodic RL and examines the potential damage of two natural types of poisoning attacks, i.e., the manipulation of reward and action. We discover that the effect of attacks crucially depend on whether the rewards are bounded or unbounded. In bounded reward settings, we show that only reward manipulation or only action manipulation cannot guarantee a successful attack. However, by combining reward and action manipulation, the adversary can manipulate any order-optimal learning algorithm to follow any targeted policy with ̃Θ(√T)total attack cost, which is order-optimal, without any knowledge of the underlying MDP.1In contrast, in unbounded reward settings, we show that reward manipulation attacks are sufficient for an adversary to success-fully manipulate any order-optimal learning algorithm to follow any targeted policy using ̃O(√T)amount of contamination. Our results reveal useful insights about what can or cannot be achieved by poisoning attacks, and are set to spur more works on the design of robust RL algorithms.
Item Type: | Conference Item (Paper) | ||||||
---|---|---|---|---|---|---|---|
Subjects: | Q Science > Q Science (General) Q Science > QA Mathematics |
||||||
Divisions: | Faculty of Science, Engineering and Medicine > Science > Computer Science | ||||||
Library of Congress Subject Headings (LCSH): | Reinforcement learning , Reinforcement learning -- Mathematical models, Mathematical optimization | ||||||
Journal or Publication Title: | Proceedings of the 31st International Joint Conference on Artificial Intelligence (IJCAI 2022) | ||||||
Publisher: | IJCAI | ||||||
Official Date: | 2022 | ||||||
Dates: |
|
||||||
Page Range: | pp. 3394-3400 | ||||||
DOI: | 10.24963/ijcai.2022/471 | ||||||
Status: | Peer Reviewed | ||||||
Publication Status: | Published | ||||||
Access rights to Published version: | Restricted or Subscription Access | ||||||
Copyright Holders: | International Joint Conferences on Artificial Intelligence | ||||||
Date of first compliant deposit: | 11 July 2022 | ||||||
Date of first compliant Open Access: | 12 July 2022 | ||||||
Conference Paper Type: | Paper | ||||||
Title of Event: | International Joint Conference on Artificial Intelligence (IJCAI 2022) | ||||||
Type of Event: | Conference | ||||||
Location of Event: | Vienna, Austria | ||||||
Date(s) of Event: | 25-29 Jul 2022 | ||||||
Related URLs: | |||||||
Open Access Version: |
Request changes or add full text files to a record
Repository staff actions (login required)
View Item |
Downloads
Downloads per month over past year