PE-MA: Parameter-Efficient Co-Evolution of Multi-Agent Systems

2025-06-13

AI Summary

To address high communication overhead and insufficient agent-level personalization in multi-agent collaborative learning, this paper proposes PE-MA, a parameter-efficient multi-agent co-evolution framework. Methodologically, PE-MA introduces a novel co-optimization mechanism integrating agent-specific lightweight adapters with neighborhood-shared adapters, enabling simultaneous global coordination and local adaptation under heterogeneity. It theoretically guarantees optimal convergence rate of O(1/√(NK)). The framework incorporates graph-structured modeling, asynchronous local updates, and efficient global aggregation, substantially reducing both communication and parameter overhead. Empirical evaluation on collaborative reasoning benchmarks demonstrates that PE-MA improves accuracy and personalization score by 12.7% and 23.4%, respectively, over state-of-the-art baselines—validating its effectiveness in unifying lightweight design, agent-level personalization, and scalability.

Technology Category

• Multiagent Systems: Multiagent Learning
• Search and Optimization: Learning to Search
• Machine Learning: Evolutionary Learning

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Multi-Agent Systems have recently emerged as a promising paradigm for collaborative reasoning and solving complex tasks. However, the design of collaborative learning algorithms in multi-agent systems faces several challenges, including high communication overhead and insufficient agent-level personalization. In this paper, we propose PE-MA (Parameter-Efficient Multi-Agent Co-Evolution), a novel collaboration framework that supports efficient, scalable, and personalized co-evolution in multi-agent systems. In PE-MA, each agent maintains a lightweight personalized adapter to support agent-specific behavior, while a shared adapter is collaboratively optimized across neighboring agents. This design balances global coordination with local adaptation under heterogeneous environments. We achieve an asymptotically optimal convergence rate of O( 1/(NK)^(1/2) ), where N is the number of agents and K the local update steps.

Problem

• Reducing communication overhead in multi-agent systems
• Enhancing agent-level personalization in collaborative learning
• Balancing global coordination with local adaptation

Innovation

• Lightweight personalized adapter for agent-specific behavior
• Shared adapter collaboratively optimized across agents
• Balances global coordination with local adaptation