Saved in:
Bibliographic Details
Main Authors: Li, Dapeng, Lou, Na, Zhang, Bin, Xu, Zhiwei, Fan, Guoliang
Format: Preprint
Published: 2023
Subjects:
Online Access:https://arxiv.org/abs/2312.09009
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913715667861504
author Li, Dapeng
Lou, Na
Zhang, Bin
Xu, Zhiwei
Fan, Guoliang
author_facet Li, Dapeng
Lou, Na
Zhang, Bin
Xu, Zhiwei
Fan, Guoliang
contents Parameter sharing, as an important technique in multi-agent systems, can effectively solve the scalability issue in large-scale agent problems. However, the effectiveness of parameter sharing largely depends on the environment setting. When agents have different identities or tasks, naive parameter sharing makes it difficult to generate sufficiently differentiated strategies for agents. Inspired by research pertaining to the brain in biology, we propose a novel parameter sharing method. It maps each type of agent to different regions within a shared network based on their identity, resulting in distinct subnetworks. Therefore, our method can increase the diversity of strategies among different agents without introducing additional training parameters. Through experiments conducted in multiple environments, our method has shown better performance than other parameter sharing methods.
format Preprint
id arxiv_https___arxiv_org_abs_2312_09009
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Adaptive parameter sharing for multi-agent reinforcement learning
Li, Dapeng
Lou, Na
Zhang, Bin
Xu, Zhiwei
Fan, Guoliang
Artificial Intelligence
Parameter sharing, as an important technique in multi-agent systems, can effectively solve the scalability issue in large-scale agent problems. However, the effectiveness of parameter sharing largely depends on the environment setting. When agents have different identities or tasks, naive parameter sharing makes it difficult to generate sufficiently differentiated strategies for agents. Inspired by research pertaining to the brain in biology, we propose a novel parameter sharing method. It maps each type of agent to different regions within a shared network based on their identity, resulting in distinct subnetworks. Therefore, our method can increase the diversity of strategies among different agents without introducing additional training parameters. Through experiments conducted in multiple environments, our method has shown better performance than other parameter sharing methods.
title Adaptive parameter sharing for multi-agent reinforcement learning
topic Artificial Intelligence
url https://arxiv.org/abs/2312.09009