Work

PropUQ-MAS: Propagation-Aware Uncertainty Quantification for LLM Multi-Agent Systems

Year: 2026

Type: conference-paper

Abstract: LLM-based multi-agent systems (MAS) solve complex tasks through communication among role-specialized agents.However, inter-agent dependencies introduce reliability risks beyond isolated agent failures.For instance, errors in intermediate messages could be inherited and amplified by downstream agents.Existing uncertainty quantification (UQ) methods mainly target isolated responses or single-agent reasoning, and therefore fail to capture uncertainty propagation in MAS.To this end, we propose PROPUQ-MAS, an error propagation-aware UQ framework that represents MAS execution as a communication-structured graph and estimates each step's reliability by combining local uncertainty with uncertainty inherited from upstream messages.Extensive experiments demonstrate that PROPUQ-MAS consistently improves UQ in MAS, with average relative gains of +6.10% in AUROC and +47.58% in PRR. 1

Source: Conference on Empirical Methods in Natural Language Processing (EMNLP)

DOI: 10.18653/2026.emnlp-main.891

Authors: Yaokun Liu, Yifan Liu, Daniel Yue Zhang, Ruichen Yao, Zelin Li, Dong Wang

Institution: University of Illinois Urbana-Champaign

Language: English

Topic: Adversarial Robustness in Machine Learning

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: bronze

Funder: National Science Foundation

Awards: HCC: Small: Crowd-Assisted Human-AI Teaming with Explanations, CHS: Small: DeepCrowd: A Crowd-assisted Deep Learning-based Disaster Scene Assessment System with Active Human-AI Interactions, FairFL-MC: A Metacognitive Calibration Intervention Powered by Fair and Private Machine Learning, CAREER: Towards Reliable and Optimized Data-Driven Cyber-Physical Systems using Human-Centric Sensing, SCC-PG: Proactive Community Health Protection Against Airborne Diseases through Intelligent and Timely Pathogen Monitoring and Communication