MiniRep: Robust Reputation-Based Aggregation for Multi-Agent Debate
作者: Jiaming Zhang, Yuwan Liu, Yue Huang, Sisi Duan
分类: cs.AI
发布日期: 2026-09-30
💡 一句话要点
提出MiniRep以解决多智能体辩论中的信誉评估问题
🎯 匹配领域: 支柱九：具身大模型 (Embodied Foundation Models)
关键词: 信誉评估 多智能体系统 聚合方法 恶意智能体 大型语言模型
📋 核心要点
- 现有信誉评估方法无法可靠预测智能体在新任务中的表现，尤其在恶意智能体的影响下更为明显。
- MiniRep通过评估智能体当前任务表现和历史信誉，设计了一种新的聚合机制，防止相似响应的智能体影响决策。
- 实验结果显示，MiniRep在多种攻击条件下均优于传统的聚合方法，尤其在MATH任务中表现突出。
📝 摘要（中文）
随着大型语言模型驱动的自主智能体迅速发展，建立可信的合作机制变得愈发重要。现有的信誉评估方法往往无法准确预测智能体在新任务中的表现，尤其是在恶意智能体的影响下。本文提出MiniRep，一个针对多智能体辩论的信誉基础聚合系统，能够有效评估智能体的行为和信誉，防止相似响应的智能体主导最终决策。实验结果表明，MiniRep在多种任务和攻击条件下均优于传统聚合方法和信誉基础方法。
🔬 方法详解
问题定义：本文旨在解决多智能体辩论中信誉评估的不足，特别是在恶意智能体影响下的决策准确性问题。现有方法往往依赖于历史表现，无法适应新的任务环境。
核心思路：MiniRep的核心思路是结合当前任务表现与历史信誉进行综合评估，同时设计机制防止相似响应的智能体主导最终决策。这样的设计旨在提高系统的鲁棒性和准确性。
技术框架：MiniRep的整体架构包括信誉评估模块、行为评估模块和聚合决策模块。信誉评估模块根据历史行为计算信誉分数，行为评估模块则关注当前任务的表现，聚合决策模块负责整合各智能体的输出。
关键创新：MiniRep的主要创新在于构建了一个攻击分类法，涵盖信誉系统攻击和软件测试变异操作，能够有效识别和应对恶意智能体的影响。这一方法与传统的信誉评估方法有本质区别。
关键设计：在设计上，MiniRep设置了多个参数以调节信誉评估的权重，并采用了特定的损失函数来优化聚合决策的准确性。此外，网络结构经过精心设计，以确保在多种攻击条件下的鲁棒性。
🖼️ 关键图片
📊 实验亮点
实验结果显示，MiniRep在MATH任务中无论在攻击条件下还是正常情况下均优于传统的聚合方法和信誉基础方法。在异构10智能体设置下，MiniRep在28种攻击条件下均表现出色，展现出显著的性能提升。
🎯 应用场景
MiniRep的研究成果在多个领域具有潜在应用价值，包括智能客服、自动化决策系统以及多智能体协作环境。通过提高智能体之间的信誉评估能力，MiniRep能够促进更可信的智能体合作，增强系统的整体性能和安全性。未来，该方法有望推动智能体技术在更复杂场景中的应用。
📄 摘要（原文）
Autonomous agents powered by large language models (LLMs) are rapidly evolving into an open agentic ecosystem. To support trustworthy collaboration, industry initiatives increasingly assess agent reputation from past behavior and provide performance leaderboards. However, reputation derived from past performance may not reliably predict an agent's behavior on new tasks, particularly when malicious agents can adapt their behavior and influence other agents during collaboration. We study reputation in multi-agent debate (MAD), where multiple agents answer the same query, debate to improve their answers, and aggregate them into a final output. We present MiniRep, a reputation-based aggregation system for MAD under malicious agents. To ground our threat model in established research, we construct an attack taxonomy drawing on reputation-system attacks and software-testing mutation operators, covering strategic exploitation of reputation and subtle corruption of agent proposals. Guided by this taxonomy, MiniRep evaluates agents based on both their behavior on the current task and their reputation over time, while preventing groups of agents with highly similar responses from dominating the final decision. We assess MiniRep across diverse tasks, LLM-agent compositions, corruption placements, and attack types drawn from our taxonomy. Our experimental results show that, MiniRep outperforms both conventional MAD aggregation and conventional reputation-based approaches on MATH no matter being attacked or not. Also, under a heterogeneous 10-agent setting on MATH, MiniRep outperforms all baselines in all 28 attack conditions.