Low Entropy Communication in Multi-Agent Reinforcement Learning

Communication in multi-agent reinforcement learning has been drawing attention recently for its significant role in cooperation. However, multi-agent systems may suffer from limitations on communication resources and thus need efficient communication techniques in real-world scenarios. According to...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Yu, Lebin, Qiu, Yunbo, Wang, Qiexiang, Zhang, Xudong, Wang, Jian
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator Yu, Lebin
Qiu, Yunbo
Wang, Qiexiang
Zhang, Xudong
Wang, Jian
description Communication in multi-agent reinforcement learning has been drawing attention recently for its significant role in cooperation. However, multi-agent systems may suffer from limitations on communication resources and thus need efficient communication techniques in real-world scenarios. According to the Shannon-Hartley theorem, messages to be transmitted reliably in worse channels require lower entropy. Therefore, we aim to reduce message entropy in multi-agent communication. A fundamental challenge is that the gradients of entropy are either 0 or infinity, disabling gradient-based methods. To handle it, we propose a pseudo gradient descent scheme, which reduces entropy by adjusting the distributions of messages wisely. We conduct experiments on two base communication frameworks with six environment settings and find that our scheme can reduce message entropy by up to 90% with nearly no loss of cooperation performance.
doi_str_mv 10.48550/arxiv.2302.05055
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2302_05055</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2302_05055</sourcerecordid><originalsourceid>FETCH-LOGICAL-a675-778ce66a58763498ce1c624ea008d3610e5fc3fe5725bb4e8c94681376ebb7f83</originalsourceid><addsrcrecordid>eNotz7FqwzAUhWEtGYqTB-hUvYAd2dKVlKkEkzQBl0LxbmT1KghiKShO27x9a7fT4V8OfIQ8lqwQGoCtTfr2n0XFWVUwYAAP5LmJX3QXxhQvd1rHYbgFb83oY6A-0NfbefT59oRhpO_og4vJ4jBVgyYFH05LsnDmfMXV_2ak3e_a-pA3by_HetvkRirIldIWpTSgleRi8xullZVAw5j-4LJkCM5yh6Aq6HuB2m6E1CVXEvteOc0z8vR3Owu6S_KDSfduknSzhP8AslJDMA</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Low Entropy Communication in Multi-Agent Reinforcement Learning</title><source>arXiv.org</source><creator>Yu, Lebin ; Qiu, Yunbo ; Wang, Qiexiang ; Zhang, Xudong ; Wang, Jian</creator><creatorcontrib>Yu, Lebin ; Qiu, Yunbo ; Wang, Qiexiang ; Zhang, Xudong ; Wang, Jian</creatorcontrib><description>Communication in multi-agent reinforcement learning has been drawing attention recently for its significant role in cooperation. However, multi-agent systems may suffer from limitations on communication resources and thus need efficient communication techniques in real-world scenarios. According to the Shannon-Hartley theorem, messages to be transmitted reliably in worse channels require lower entropy. Therefore, we aim to reduce message entropy in multi-agent communication. A fundamental challenge is that the gradients of entropy are either 0 or infinity, disabling gradient-based methods. To handle it, we propose a pseudo gradient descent scheme, which reduces entropy by adjusting the distributions of messages wisely. We conduct experiments on two base communication frameworks with six environment settings and find that our scheme can reduce message entropy by up to 90% with nearly no loss of cooperation performance.</description><identifier>DOI: 10.48550/arxiv.2302.05055</identifier><language>eng</language><subject>Computer Science - Learning ; Computer Science - Multiagent Systems</subject><creationdate>2023-02</creationdate><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,777,882</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2302.05055$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2302.05055$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Yu, Lebin</creatorcontrib><creatorcontrib>Qiu, Yunbo</creatorcontrib><creatorcontrib>Wang, Qiexiang</creatorcontrib><creatorcontrib>Zhang, Xudong</creatorcontrib><creatorcontrib>Wang, Jian</creatorcontrib><title>Low Entropy Communication in Multi-Agent Reinforcement Learning</title><description>Communication in multi-agent reinforcement learning has been drawing attention recently for its significant role in cooperation. However, multi-agent systems may suffer from limitations on communication resources and thus need efficient communication techniques in real-world scenarios. According to the Shannon-Hartley theorem, messages to be transmitted reliably in worse channels require lower entropy. Therefore, we aim to reduce message entropy in multi-agent communication. A fundamental challenge is that the gradients of entropy are either 0 or infinity, disabling gradient-based methods. To handle it, we propose a pseudo gradient descent scheme, which reduces entropy by adjusting the distributions of messages wisely. We conduct experiments on two base communication frameworks with six environment settings and find that our scheme can reduce message entropy by up to 90% with nearly no loss of cooperation performance.</description><subject>Computer Science - Learning</subject><subject>Computer Science - Multiagent Systems</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2023</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotz7FqwzAUhWEtGYqTB-hUvYAd2dKVlKkEkzQBl0LxbmT1KghiKShO27x9a7fT4V8OfIQ8lqwQGoCtTfr2n0XFWVUwYAAP5LmJX3QXxhQvd1rHYbgFb83oY6A-0NfbefT59oRhpO_og4vJ4jBVgyYFH05LsnDmfMXV_2ak3e_a-pA3by_HetvkRirIldIWpTSgleRi8xullZVAw5j-4LJkCM5yh6Aq6HuB2m6E1CVXEvteOc0z8vR3Owu6S_KDSfduknSzhP8AslJDMA</recordid><startdate>20230210</startdate><enddate>20230210</enddate><creator>Yu, Lebin</creator><creator>Qiu, Yunbo</creator><creator>Wang, Qiexiang</creator><creator>Zhang, Xudong</creator><creator>Wang, Jian</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20230210</creationdate><title>Low Entropy Communication in Multi-Agent Reinforcement Learning</title><author>Yu, Lebin ; Qiu, Yunbo ; Wang, Qiexiang ; Zhang, Xudong ; Wang, Jian</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a675-778ce66a58763498ce1c624ea008d3610e5fc3fe5725bb4e8c94681376ebb7f83</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2023</creationdate><topic>Computer Science - Learning</topic><topic>Computer Science - Multiagent Systems</topic><toplevel>online_resources</toplevel><creatorcontrib>Yu, Lebin</creatorcontrib><creatorcontrib>Qiu, Yunbo</creatorcontrib><creatorcontrib>Wang, Qiexiang</creatorcontrib><creatorcontrib>Zhang, Xudong</creatorcontrib><creatorcontrib>Wang, Jian</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Yu, Lebin</au><au>Qiu, Yunbo</au><au>Wang, Qiexiang</au><au>Zhang, Xudong</au><au>Wang, Jian</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Low Entropy Communication in Multi-Agent Reinforcement Learning</atitle><date>2023-02-10</date><risdate>2023</risdate><abstract>Communication in multi-agent reinforcement learning has been drawing attention recently for its significant role in cooperation. However, multi-agent systems may suffer from limitations on communication resources and thus need efficient communication techniques in real-world scenarios. According to the Shannon-Hartley theorem, messages to be transmitted reliably in worse channels require lower entropy. Therefore, we aim to reduce message entropy in multi-agent communication. A fundamental challenge is that the gradients of entropy are either 0 or infinity, disabling gradient-based methods. To handle it, we propose a pseudo gradient descent scheme, which reduces entropy by adjusting the distributions of messages wisely. We conduct experiments on two base communication frameworks with six environment settings and find that our scheme can reduce message entropy by up to 90% with nearly no loss of cooperation performance.</abstract><doi>10.48550/arxiv.2302.05055</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2302.05055
ispartof
issn
language eng
recordid cdi_arxiv_primary_2302_05055
source arXiv.org
subjects Computer Science - Learning
Computer Science - Multiagent Systems
title Low Entropy Communication in Multi-Agent Reinforcement Learning
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-18T06%3A29%3A27IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Low%20Entropy%20Communication%20in%20Multi-Agent%20Reinforcement%20Learning&rft.au=Yu,%20Lebin&rft.date=2023-02-10&rft_id=info:doi/10.48550/arxiv.2302.05055&rft_dat=%3Carxiv_GOX%3E2302_05055%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true