CASIA OpenIR  > 复杂系统认知与决策实验室  > 听觉模型与认知计算
M3: Modularization for Multi-task and Multi-agent Offline Pre-training
Meng Linghui1,2; Ruan Jingqing1,3; Xiong Xuantang1,2; Li Xiyun1,3; Zhang Xi1; Xing Dengpeng1,2; Xu Bo1,2
2023-05
Conference NameProceedings of the 2023 International Conference on Autonomous Agents and Multiagent Systems
Conference Date2023.5.29-2023.6.2
Conference PlaceLondon, United Kingdom
Abstract

Learning a multi-task policy is crucial in multi-agent reinforcement learning (MARL). Recent work has focused on learning in the context of online multi-task reinforcement learning, where a policy is jointly trained from scratch, aiming to generalize well to few-shot or even zero-shot tasks. However, existing online methods require tremendous interactions and are therefore unsuitable for environments where interactions are expensive. In this work, we novelly introduce the modularization for multi-task and multi-agent offline pre-training (M3) to learn high-level transferable policy representations. We claim that the discrete policy representation is critical for multi-task offline learning and accordingly leverage contexts as a task prompt to enhance the adaptability of pre-trained models to various tasks. To disentangle multiple agents of variation under heterogeneous and non-stationary properties even though they receive the same task, we employ an agent-invariant VQ-VAE to identify each of the multiple agents. We encapsulate the pre-trained model as part of an online MARL algorithm and fine-tune it to improve generalization. We also theoretically analyze the generalization error of our method. We test the proposed method on the challenging StarCraft Multi-Agent Challenge (SMAC) tasks, and empirical results show that it can achieve supreme performance in few-shot or even zero-shot settings across multiple tasks over state-of-the-art MARL methods.

Sub direction classification多智能体系统
planning direction of the national heavy laboratory多智能体决策
Paper associated data
Document Type会议论文
Identifierhttp://ir.ia.ac.cn/handle/173211/57333
Collection复杂系统认知与决策实验室_听觉模型与认知计算
Corresponding AuthorXing Dengpeng; Xu Bo
Affiliation1.Institute of Automation, Chinese Academy of Sciences
2.School of Artificial Intelligence, University of Chinese Academy of Sciences
3.School of Future Technology, University of Chinese Academy of Sciences
First Author AffilicationInstitute of Automation, Chinese Academy of Sciences
Corresponding Author AffilicationInstitute of Automation, Chinese Academy of Sciences
Recommended Citation
GB/T 7714
Meng Linghui,Ruan Jingqing,Xiong Xuantang,et al. M3: Modularization for Multi-task and Multi-agent Offline Pre-training[C],2023.
Files in This Item: Download All
File Name/Size DocType Version Access License
m3_full_paper.pdf(1302KB)会议论文 开放获取CC BY-NC-SAView Download
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[Meng Linghui]'s Articles
[Ruan Jingqing]'s Articles
[Xiong Xuantang]'s Articles
Baidu academic
Similar articles in Baidu academic
[Meng Linghui]'s Articles
[Ruan Jingqing]'s Articles
[Xiong Xuantang]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[Meng Linghui]'s Articles
[Ruan Jingqing]'s Articles
[Xiong Xuantang]'s Articles
Terms of Use
No data!
Social Bookmark/Share
File name: m3_full_paper.pdf
Format: Adobe PDF
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.