协同多智能体强化学习的鲁棒性研究

Jieyu Lin, Kristina Dzeparoska, S. Zhang, A. Leon-Garcia, Nicolas Papernot
{"title":"协同多智能体强化学习的鲁棒性研究","authors":"Jieyu Lin, Kristina Dzeparoska, S. Zhang, A. Leon-Garcia, Nicolas Papernot","doi":"10.1109/SPW50608.2020.00027","DOIUrl":null,"url":null,"abstract":"In cooperative multi-agent reinforcement learning (c-MARL), agents learn to cooperatively take actions as a team to maximize a total team reward. We analyze the robustness of c-MARL to adversaries capable of attacking one of the agents on a team. Through the ability to manipulate this agent's observations, the adversary seeks to decrease the total team reward. Attacking c-MARL is challenging for three reasons: first, it is difficult to estimate team rewards or how they are impacted by an agent mispredicting; second, models are non-differentiable; and third, the feature space is low-dimensional. Thus, we introduce a novel attack. The attacker first trains a policy network with reinforcement learning to find a wrong action it should encourage the victim agent to take. Then, the adversary uses targeted adversarial examples to force the victim to take this action. Our results on the StartCraft II multi-agent benchmark demonstrate that c-MARL teams are highly vulnerable to perturbations applied to one of their agent's observations. By attacking a single agent, our attack method has highly negative impact on the overall team reward, reducing it from 20 to 9.4. This results in the team's winning rate to go down from 98.9% to 0%.","PeriodicalId":413600,"journal":{"name":"2020 IEEE Security and Privacy Workshops (SPW)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2020-03-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"43","resultStr":"{\"title\":\"On the Robustness of Cooperative Multi-Agent Reinforcement Learning\",\"authors\":\"Jieyu Lin, Kristina Dzeparoska, S. Zhang, A. Leon-Garcia, Nicolas Papernot\",\"doi\":\"10.1109/SPW50608.2020.00027\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"In cooperative multi-agent reinforcement learning (c-MARL), agents learn to cooperatively take actions as a team to maximize a total team reward. We analyze the robustness of c-MARL to adversaries capable of attacking one of the agents on a team. Through the ability to manipulate this agent's observations, the adversary seeks to decrease the total team reward. Attacking c-MARL is challenging for three reasons: first, it is difficult to estimate team rewards or how they are impacted by an agent mispredicting; second, models are non-differentiable; and third, the feature space is low-dimensional. Thus, we introduce a novel attack. The attacker first trains a policy network with reinforcement learning to find a wrong action it should encourage the victim agent to take. Then, the adversary uses targeted adversarial examples to force the victim to take this action. Our results on the StartCraft II multi-agent benchmark demonstrate that c-MARL teams are highly vulnerable to perturbations applied to one of their agent's observations. By attacking a single agent, our attack method has highly negative impact on the overall team reward, reducing it from 20 to 9.4. This results in the team's winning rate to go down from 98.9% to 0%.\",\"PeriodicalId\":413600,\"journal\":{\"name\":\"2020 IEEE Security and Privacy Workshops (SPW)\",\"volume\":\"1 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2020-03-08\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"43\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2020 IEEE Security and Privacy Workshops (SPW)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/SPW50608.2020.00027\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2020 IEEE Security and Privacy Workshops (SPW)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/SPW50608.2020.00027","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 43

摘要

在合作多智能体强化学习(c-MARL)中,智能体学习作为一个团队合作采取行动,以最大化团队总回报。我们分析了c-MARL对能够攻击团队中的一个代理的对手的鲁棒性。通过操纵代理人的观察,对手试图减少整个团队的奖励。攻击c-MARL是具有挑战性的,原因有三个:首先,很难估计团队奖励或它们如何受到代理错误预测的影响;第二,模型不可微;第三,特征空间是低维的。因此,我们引入了一种新的攻击。攻击者首先用强化学习训练策略网络,找出一个它应该鼓励受害者代理采取的错误行动。然后,攻击者使用有针对性的敌对例子来迫使受害者采取这一行动。我们在StartCraft II多智能体基准测试上的结果表明,c-MARL团队非常容易受到应用于其智能体观察结果之一的扰动的影响。通过攻击单个智能体,我们的攻击方法对整个团队奖励产生了非常负面的影响,将其从20降低到9.4。这导致球队的胜率从98.9%下降到0%。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
On the Robustness of Cooperative Multi-Agent Reinforcement Learning
In cooperative multi-agent reinforcement learning (c-MARL), agents learn to cooperatively take actions as a team to maximize a total team reward. We analyze the robustness of c-MARL to adversaries capable of attacking one of the agents on a team. Through the ability to manipulate this agent's observations, the adversary seeks to decrease the total team reward. Attacking c-MARL is challenging for three reasons: first, it is difficult to estimate team rewards or how they are impacted by an agent mispredicting; second, models are non-differentiable; and third, the feature space is low-dimensional. Thus, we introduce a novel attack. The attacker first trains a policy network with reinforcement learning to find a wrong action it should encourage the victim agent to take. Then, the adversary uses targeted adversarial examples to force the victim to take this action. Our results on the StartCraft II multi-agent benchmark demonstrate that c-MARL teams are highly vulnerable to perturbations applied to one of their agent's observations. By attacking a single agent, our attack method has highly negative impact on the overall team reward, reducing it from 20 to 9.4. This results in the team's winning rate to go down from 98.9% to 0%.
求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
自引率
0.00%
发文量
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:604180095
Book学术官方微信