{"title":"Toward Resilient Multi-Agent Actor-Critic Algorithms for Distributed Reinforcement Learning","authors":"Yixuan Lin, Shripad Gade, Romeil Sandhu, Ji Liu","doi":"10.23919/ACC45564.2020.9147381","DOIUrl":null,"url":null,"abstract":"This paper considers a distributed reinforcement learning problem in the presence of Byzantine agents. The system consists of a central coordinating authority called \"master agent\" and multiple computational entities called \"worker agents\". The master agent is assumed to be reliable, while, a small fraction of the workers can be Byzantine (malicious) adversaries. The workers are interested in cooperatively maximize a convex combination of the honest (non-malicious) worker agents’ long-term returns through communication between the master agent and worker agents. A distributed actor-critic algorithm is studied which makes use of entry-wise trimmed mean. The algorithm’s communication-efficiency is improved by allowing the worker agents to send only a scalar-valued variable to the master agent, instead of the entire parameter vector, at each iteration. The improved algorithm involves computing a trimmed mean over only the received scalar-valued variable. It is shown that both algorithms converge almost surely.","PeriodicalId":288450,"journal":{"name":"2020 American Control Conference (ACC)","volume":"30 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2020-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"7","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2020 American Control Conference (ACC)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.23919/ACC45564.2020.9147381","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 7
Abstract
This paper considers a distributed reinforcement learning problem in the presence of Byzantine agents. The system consists of a central coordinating authority called "master agent" and multiple computational entities called "worker agents". The master agent is assumed to be reliable, while, a small fraction of the workers can be Byzantine (malicious) adversaries. The workers are interested in cooperatively maximize a convex combination of the honest (non-malicious) worker agents’ long-term returns through communication between the master agent and worker agents. A distributed actor-critic algorithm is studied which makes use of entry-wise trimmed mean. The algorithm’s communication-efficiency is improved by allowing the worker agents to send only a scalar-valued variable to the master agent, instead of the entire parameter vector, at each iteration. The improved algorithm involves computing a trimmed mean over only the received scalar-valued variable. It is shown that both algorithms converge almost surely.