arXiv cs.MA· Dominik Jehle, Tom Zehle·· 13 小时前AI 评分31
用贝叶斯进化优化设计和调优多智能体系统
Design and Tuning of Multi-Agent Systems Using Bayesian Evolutionary Optimization
AI 导读
研究者提出贝叶斯进化优化框架，可自动发现任务特定的多智能体系统拓扑，用进化算子生成多样化工作流图，并以图感知高斯过程与 Expected Improvement 将昂贵评估集中在有潜力的候选上。
正文 · 原文
Abstract:Designing effective multi-agent systems (MAS) remains a challenging and resource-intensive task. We propose a Bayesian Evolutionary Optimization framework for automatically discovering task-specific MAS topologies. The approach combines evolutionary operators to generate diverse workflow graphs with Bayesian optimization to focus cost intensive evaluations on promising candidates. In contrast to LLM-based candidate selection, our approach uses a graph-aware Gaussian Process and Expected Improvement to provide an explicit selection criterion during evolutionary search. We evaluate the method on BigCodeBench for code generation and on a tabular preprocessing benchmark using TabPFN. On BigCodeBench, evolved architectures achieve the highest overall performance among the evaluated agent systems. On temporal tabular datasets, optimized MAS preprocessing improves over the default TabPFN model on several tasks, while remaining competitive on IID datasets. Overall, the results demonstrate the potential of Bayesian selection for guiding evolutionary MAS search under expensive evaluation budgets.
| Subjects: | Machine Learning (cs.LG); Multiagent Systems (cs.MA) |
| Cite as: | arXiv:2610.29276 [cs.LG] |
| (or arXiv:2610.29276v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.29276 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Tom Zehle [view email]
[v1]
Wed, 21 Oct 2026 06:31:27 UTC (672 KB)
来源：arXiv cs.MA · arxiv.org