Active Learning for Communication Structure Optimization in LLM-Based Multi-Agent Systems

2026-05-07

AI Summary

This work addresses the instability and training-set sensitivity of existing methods for optimizing communication structures in multi-agent systems with large language models, which typically rely on random task sampling under limited training budgets. To overcome these limitations, the authors propose a task selection framework grounded in ensemble information theory. The approach estimates the information gain of candidate tasks with respect to the communication graph’s parameter distribution using ensemble Kalman inversion, and integrates embedding-based representative sampling, surrogate modeling, and batch Thompson sampling to enable efficient active optimization in black-box, noisy environments. Experimental results demonstrate that the proposed method significantly improves communication structure optimization both in standard settings and under adversarial agent attacks, while simultaneously reducing computational overhead.

Abstract

Optimizing the communication structure of large language model based multi-agent systems (LLM-MAS) has been shown to improve downstream performance and reduce token usage. Existing methods typically rely on randomly sampled training tasks. However, tasks may differ substantially in difficulty and domain, and thus they are not equally informative for updating communication structure, making optimization under limited training budgets often unstable and highly sensitive to the particular training set. To actively identify the most valuable tasks for communication-structure optimization, we propose an ensemble-based information-theoretic task selection framework. The proposed method estimates task informativeness by how much a candidate task changes the distribution over graph parameters, using ensemble Kalman inversion as an efficient and derivative-free approximation of the corresponding Bayesian update. The resulting estimator is especially suitable for black-box and noisy multi-agent systems. To enhance scalability, we construct a compact candidate pool through embedding-based representative selection and combine the informative selection with surrogate modeling and batch Thompson sampling. We validate our method in both benign settings and settings with agent attacks, demonstrating its effectiveness for communication-structure optimization under constrained computational budgets.

Problem

• Active Learning
• Communication Structure Optimization
• LLM-Based Multi-Agent Systems
• Task Selection
• Training Budget

Innovation

• active learning
• communication structure optimization
• ensemble Kalman inversion
• multi-agent systems
• information-theoretic task selection