HELENA：基于互补拓扑联合的分层稀疏协调多智能体框架
原标题：HELENA:Hierarchical Sparse Coordination over a Union of Complementary Topologies for MAS
AI 摘要
arXiv 上发布了一篇关于多智能体系统的新论文，提出了 HELENA 框架。该框架通过蒙特卡洛树搜索和行列式点过程选择互补拓扑，构建联合图，并利用分层稀疏协调模块在每一步仅激活稀疏子图，以抑制冗余噪声传播。实验表明，HELENA 在八个基准测试上均达到最先进水平，平均提升 3.47%，在 MMLU-Pro 上提升高达 10.34%。
正文节选
Computer Science > Multiagent Systems Title:HELENA:Hierarchical Sparse Coordination over a Union of Complementary Topologies for MAS View PDF HTML (experimental) Abstract:LLM-based multi-agent systems (MAS) typically optimize a single topology, restricting reasoning to a narrow trajectory and limiting comprehensive analytical capacity. Naively merging multiple topologies into a composite graph introduces redundant noise propagation across irrelevant connections, degrading solution qu