KABB: Knowledge-Aware Bayesian Bandits for Dynamic Expert Coordination in Multi-Agent Systems

Jusheng Zhang, Zimeng Huang, Yijia Fan, Ningyuan Liu, Mingyan Li, Zhuojie Yang, Jiawei Yao, Jian Wang, Keze Wang

TL;DR: LLM,Agent

Abstract: As scaling large language models faces prohibitive costs, multi-agent systems emerge as a promising alternative, though challenged by static knowledge assumptions and coordination inefficiencies. We introduce Knowledge-Aware Bayesian Bandits (KABB), a novel framework that enhances multi-agent system coordination through semantic understanding and dynamic adaptation. The framework features three key innovations: a customized knowledge distance model for deep semantic understanding, a dual-adaptation mechanism for continuous expert optimization, and a knowledge-aware Thompson Sampling strategy for efficient expert selection. Extensive evaluation demonstrates KABB achieves an optimal cost-performance balance, maintaining high performance while keeping computational demands relatively low in multi-agent coordination.

Lay Summary:

Problem. Scaling large language models and static expert ensembles incurs prohibitive compute and monetary costs.

Solution. We propose Knowledge-Aware Bayesian Bandits (KABB), which (1) computes a semantic distance to match tasks and experts, (2) continuously adapts expert skills, and (3) employs a Thompson Sampling strategy that dynamically routes each task to the best subset of experts.

Impact. On benchmark suites, KABB achieves equal or better accuracy while cutting computational expense by up to 7× compared to fixed ensembles.

Application-Driven Machine Learning: This submission is on Application-Driven Machine Learning.

Primary Area: Deep Learning->Large Language Models

Keywords: LLM, Agent

Submission Number: 774