EvolveMAS: A Unified Self-Evolution Framework for Decentralized Multi-Agent Systems

DOI10.48550/ARXIV.2610.34450MaRDI QIDQ7408888arXivFDO

Authors Bingbing Xu, Jiawei Wang, An Zhang, Tat Seng Chua, Yalun Wu, Qiwei Ma, Dan Zhang, Long Bai, Boyang Wang, Jiatong Li, Zheqi Lv, Shaozu Yuan, Peiying Wang, Tat-Seng Chua

Publication date 25 October 2026

Published in arXiv

Tldr This paper proposes EvolveMAS, a unified self-evolution framework for decentralized multi-agent systems that enables test-time adaptation and reliable experience reuse across different interaction settings through two complementary processes: Within-Episode Online Adaptation and Cross-Episode Knowledge Evolution.

Full work available at URL https://doi.org/10.48550/ARXIV.2610.34450

arXiv classification

cs.AIcs.MA

Cites work

• Multi-agent reinforcement learning: a selective overview of theories and algorithms
• Working Memory

This page was built for publication: EvolveMAS: A Unified Self-Evolution Framework for Decentralized Multi-Agent Systems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7408888)