A communication-efficient multi-agent actor-critic algorithm for distributed reinforcement learning

Yixuan Lin; Kaiqing Zhang; Zhuoran Yang; Zhaoran Wang; Tamer Başar; Romeil Sandhu; Ji Liu

A communication-efficient multi-agent actor-critic algorithm for distributed reinforcement learning

Yixuan Lin, Kaiqing Zhang, Zhuoran Yang, Zhaoran Wang, Tamer Başar, Romeil Sandhu, Ji Liu

Industrial Engineering and Management Sciences

Research output: Contribution to journal › Article › peer-review

Abstract

This paper considers a distributed reinforcement learning problem in which a network of multiple agents aim to cooperatively maximize the globally averaged return through communication with only local neighbors. A randomized communication-efficient multi-agent actor-critic algorithm is proposed for possibly unidirectional communication relationships depicted by a directed graph. It is shown that the algorithm can solve the problem for strongly connected graphs by allowing each agent to transmit only two scalar-valued variables at one time.

Original language	English (US)
Journal	Unknown Journal
State	Published - Jul 5 2019

ASJC Scopus subject areas

General

Fingerprint Dive into the research topics of 'A communication-efficient multi-agent actor-critic algorithm for distributed reinforcement learning'. Together they form a unique fingerprint.

Cite this

@article{e626ab2ced4e4599b8a0a565755dd6c5,

title = "A communication-efficient multi-agent actor-critic algorithm for distributed reinforcement learning",

abstract = "This paper considers a distributed reinforcement learning problem in which a network of multiple agents aim to cooperatively maximize the globally averaged return through communication with only local neighbors. A randomized communication-efficient multi-agent actor-critic algorithm is proposed for possibly unidirectional communication relationships depicted by a directed graph. It is shown that the algorithm can solve the problem for strongly connected graphs by allowing each agent to transmit only two scalar-valued variables at one time.",

author = "Yixuan Lin and Kaiqing Zhang and Zhuoran Yang and Zhaoran Wang and Tamer Ba{\c s}ar and Romeil Sandhu and Ji Liu",

year = "2019",

month = jul,

day = "5",

language = "English (US)",

journal = "Free Radical Biology and Medicine",

issn = "0891-5849",

publisher = "Elsevier Inc.",

}

TY - JOUR

T1 - A communication-efficient multi-agent actor-critic algorithm for distributed reinforcement learning

AU - Lin, Yixuan

AU - Zhang, Kaiqing

AU - Yang, Zhuoran

AU - Wang, Zhaoran

AU - Başar, Tamer

AU - Sandhu, Romeil

AU - Liu, Ji

PY - 2019/7/5

Y1 - 2019/7/5

N2 - This paper considers a distributed reinforcement learning problem in which a network of multiple agents aim to cooperatively maximize the globally averaged return through communication with only local neighbors. A randomized communication-efficient multi-agent actor-critic algorithm is proposed for possibly unidirectional communication relationships depicted by a directed graph. It is shown that the algorithm can solve the problem for strongly connected graphs by allowing each agent to transmit only two scalar-valued variables at one time.

AB - This paper considers a distributed reinforcement learning problem in which a network of multiple agents aim to cooperatively maximize the globally averaged return through communication with only local neighbors. A randomized communication-efficient multi-agent actor-critic algorithm is proposed for possibly unidirectional communication relationships depicted by a directed graph. It is shown that the algorithm can solve the problem for strongly connected graphs by allowing each agent to transmit only two scalar-valued variables at one time.

UR - http://www.scopus.com/inward/record.url?scp=85093392918&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85093392918&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:85093392918

JO - Free Radical Biology and Medicine

JF - Free Radical Biology and Medicine

SN - 0891-5849

ER -