MACE improves multi-agent memory use for better task success
MACE: Memory-Agent Co-Evolution with Adaptive Memory Graphs for Multi-Agent Systems
Machine Learning
Summary
Multi-agent systems made up of large language models can plan and work together on complex tasks. The authors found that breaking down the agents’ memory into connected groups called functional units helps them remember and reuse important steps better. They made a new system called MACE that adapts how memory is organized and used based on feedback from each task. MACE helps agents choose the best parts of memory and instructions during tasks, leading to better overall task performance compared to previous methods.
What this means in practice
- •For multi-agent system developers: Enable agents to share and adapt task procedures efficiently by organizing memory into linked functional units and updating them through task outcomes.
- •For automation platform engineers: Improve automated workflows by using adaptive memory structures that help coordinating AI agents verify and repair tasks more effectively.
Authors
Kairui Yang, Minghao An, Xunkai Li, Ziheng Yi, Zekai Chen, Guangyuan He, Rong-Hua Li
Abstract
LLM-based multi-agent systems generate collaboration traces that record how agents plan tasks, verify intermediate results, and repair failures. Reusing these procedures requires preserving an action's prerequisites and the outputs needed by subsequent agents. Our empirical studies show that grouping these dependencies into functional memory units improves their retention, while connecting units increases retrieval of the units and links jointly required by a task. The preferred combination of units also changes between instructions and checklists, even when each combination's content is fixed across formats. Updating choices from the outcomes of each combination and format pairing outperforms scoring combinations and formats separately. These findings motivate MACE, a memory-agent co-evolution framework that adapts memory organization and agent memory use through execution feedback. Its MemGoG structure represents functional units as subgraphs of related conditions, actions, and outputs, connecting them through support, conflict, and repair relations. MACE Loop selects task-relevant units and relations within a memory budget and provides each agent with instructions or checklists for its current operation. It records the selected units, presentation formats, agent outputs, and task outcomes to update unit scores and relations for retrieval and inform subsequent presentation choices. Across eight benchmarks, MACE outperforms ten baselines with an average score of 81.11%, compared with 78.97% for the strongest baseline, SAGE.