PANDA：一种具有灵活编排能力的去中心化架构，用于可扩展、容错的多智能体系统
PANDA: A Decentralized Architecture with Flexible Orchestration for Scalable, Fault-Tolerant Multi-Agent Systems
浏览论文内容
中文总结 AI 辅助
PANDA提出一种去中心化多智能体架构，通过解耦通信与灵活编排（星型、链型、网格型）实现大规模扩展、容错恢复和信任治理，在HotPotQA上以8倍效率达到最先进精度并实现100%故障下任务完成。
中文摘要 AI 辅助
现有的基于LLM的多智能体系统（MAS）架构无法在规模上可靠且高效地解决多步骤任务：它们难以支持大量智能体和并发任务、容忍故障、管理智能体交互，以及适应不同任务所需的多样化规划和执行模式。我们提出了PANDA，一种去中心化架构，连接大量异构、独立管理的智能体，使它们能够发现彼此的能力，并为每个任务自组织成小型专业团队。PANDA通过将集体通信与团队通信解耦来实现扩展，允许智能体同时参与多个团队，在集体中负载均衡任务，并在每个智能体内调度并发工作。PANDA进一步将底层架构与编排策略分离，支持三种规划和执行模式（星型、链型和网格型），可根据每个任务的结构和要求进行选择。PANDA检测基础设施和编排故障，并通过围绕故障组件动态重新规划来恢复受影响的任务。最后，为了在不限制可扩展性的集中式服务的情况下提供治理，PANDA使用信任网络模型来将智能体交互限制在已建立的信任关系中。我们在HotPotQA基准上评估了PANDA，证明其可扩展到数千个智能体，在毫秒级时间内组建团队，在高达8倍效率下匹配最先进的准确性，并在现有系统失败的故障下维持100%的任务完成率。
英文摘要
Existing architectures for LLM-based multi-agent systems (MAS) cannot reliably and efficiently solve multi-step tasks at scale: they struggle to support large numbers of agents and concurrent tasks, tolerate failures, govern agent interactions, and accommodate the diverse planning and execution patterns different tasks require. We present PANDA, a decentralized architecture that connects a large collective of heterogeneous, independently administered agents, letting them discover each other's capabilities and self-organize into small specialized teams per task. PANDA scales by decoupling collective communication from team communication, allowing agents to participate in multiple teams simultaneously, load-balancing tasks across the collective, and scheduling concurrent work within each agent. PANDA further separates the underlying architecture from the orchestration strategy, supporting three planning and execution patterns (star, chain, and mesh) that can be selected according to the structure and requirements of each task. PANDA detects infrastructure and orchestration failures and recovers affected tasks by dynamically replanning around failed components. Finally, to provide governance without a centralized service that would limit scalability, PANDA uses a web-of-trust model to constrain agent interactions to established trust relationships. We evaluate PANDA on the HotPotQA benchmark, demonstrating that it scales to thousands of agents, assembles teams in milliseconds, matches state-of-the-art accuracy at up to 8x the efficiency, and sustains 100% task completion under faults where existing systems fail.
发表机构
- Khoury College of Computer Sciences（胡瑞计算机科学学院）
- Northeastern University（东北大学）
机构由 AI 辅助整理，请以论文原文为准。
