arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2601.21742cs.AIcs.CLcs.MA
收藏

知识上下文学习:在基于大语言模型的多智能体系统中以正确方式建立信任

Epistemic Context Learning: Building Trust the Right Way in LLM-Based Multi-Agent Systems

National University of Singapore(国立新加坡大学);Singapore University of Technology(新加坡科技学院);Nanyang Technological University(南洋理工大学);Duke University(杜克大学);University of Waterloo(滑铁卢大学);Microsoft(微软公司);Peking University(北京大学);University of Pennsylvania(宾夕法尼亚大学)

Ruiwen Zhou, Maojia Song, Xiaobao Wu, Sitao Cheng, Xunjian Yin, Yuxi Xie, Zhuoqun Hao, Wenyue Hua, Liangming Pan, Soujanya Poria, Min-Yen Kan

AI总结:

本文提出Epistemic Context Learning(ECL),通过历史交互构建同伴资料以提升多智能体系统中信任建模的准确性,使小型模型在性能上超越大模型,并在多种配置中表现出良好的泛化能力。

摘要:

在多智能体(MA)系统中,个体智能体往往缺乏鲁棒性,倾向于盲目服从误导性的同伴。我们证明这种弱点源于顺从性和对同伴可靠性评估能力不足。为了解决这一问题,我们首先正式化了历史感知参考的学习问题,将同伴的历史互动作为额外输入,使智能体在不确定时能够估计同伴可靠性并学习来自可信同伴的知识。这将任务从评估同伴推理质量转变为基于交互历史估计同伴可靠性。我们随后开发了知识上下文学习(ECL):一种推理框架,其预测基于显式构建的同伴资料。我们进一步通过强化学习使用辅助奖励优化ECL。我们的实验表明,ECL使小型模型如Qwen 3-4B能够准确识别可靠同伴,从而在与历史无关的基线模型(Qwen 3-30B)的性能上高出8倍。ECL还使前沿模型达到接近完美(100%)的性能。我们证明ECL在各种MA配置中具有良好的泛化能力,并发现LLMs在信任建模方面表现良好,揭示了信任建模准确性和最终答案质量之间的强相关性。

英文摘要:

Individual agents in multi-agent (MA) systems often lack robustness, tending to blindly conform to misleading peers. We show this weakness stems from both sycophancy and inadequate ability to evaluate peer reliability. To address this, we first formalize the learning problem of history-aware reference, introducing the historical interactions of peers as additional input, so that agents can estimate peer reliability and learn from trustworthy peers when uncertain. This shifts the task from evaluating peer reasoning quality to estimating peer reliability based on interaction history. We then develop Epistemic Context Learning (ECL): a reasoning framework that conditions predictions on explicitly-built peer profiles from history. We further optimize ECL by reinforcement learning using auxiliary rewards. Our experiments reveal that our ECL enables small models like Qwen 3-4B to outperform a history-agnostic baseline 8x its size (Qwen 3-30B) by accurately identifying reliable peers. ECL also boosts frontier models to near-perfect (100%) performance. We show that ECL generalizes well to various MA configurations and we find that trust is modeled well by LLMs, revealing a strong correlation in trust modeling accuracy and final answer quality.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑