{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/the-performance-impact-of-combining-agent","title":"The Performance Impact of Combining Agent Factorization with Different Learning Algorithms for Multiagent Coordination","arxiv_id":null,"date":"2022-09-09","proceeding":"SETN 2022 9","authors":["Andreas Kallinteris","Stavros Orfanoudakis","Georgios Chalkiadakis"],"abstract":"Factorizing a multiagent system refers to partitioning the state-\r\naction space to individual agents and defining the interactions be-\r\ntween those agents. This so-called agent factorization is of much im-\r\nportance in real-world industrial settings, and is a process that can\r\nhave significant performance implications. In this work, we explore\r\nif the performance impact of agent factorization is different when\r\nusing different learning algorithms in multiagent coordination set-\r\ntings. We evaluated six different agent factorization instances—or\r\nagent definitions—in the warehouse traffic management domain,\r\ncomparing the performance of (mainly) two learning algorithms\r\nsuitable for learning coordinated multiagent policies: the Evolu-\r\ntionary Strategies (ES), and a genetic algorithm (CCEA) previously\r\nused in this setting. Our results demonstrate that different learning\r\nalgorithms are affected in different ways by alternative agent defi-\r\nnitions. Given this, we can deduce that many important multiagent\r\ncoordination problems can potentially be solved by an appropriate\r\nagent factorization in conjunction with an appropriate choice of\r\na learning algorithm. Moreover, our work shows that ES is an ef-\r\nfective learning algorithm for the warehouse traffic management\r\ndomain; while, interestingly, celebrated policy gradient methods\r\ndo not fare well in this complex real-world problem setting.","url_abs":"https://dl.acm.org/doi/abs/10.1145/3549737.3549773","url_pdf":"https://dl.acm.org/doi/abs/10.1145/3549737.3549773","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"the-performance-impact-of-combining-agent","repo_url":"https://github.com/stavrosgreece/MultiAgentLearning","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok","spdx":"GPL-3.0"}}],"tasks":[{"task_slug":"management","task_name":"Management"},{"task_slug":"policy-gradient-methods","task_name":"Policy Gradient Methods"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}