Heterogeneous Information-Bottleneck Coordination Graphs for Multi-Agent Reinforcement Learning
The paper introduces Heterogeneous Information-Bottleneck Coordination Graphs (HIBCG) for multi-agent reinforcement learning. It addresses the limitations of existing sparse-graph learners by providing a theoretically grounded mechanism for edge existence and communication capacity. The proposed method enhances the learning of agent relationships and optimizes message content for task relevance.
- ▪HIBCG constructs a group-aware sparse graph with justified edge existence and message capacity.
- ▪The graph information bottleneck serves as the underlying tool for the proposed method.
- ▪The approach allows for differential edge control and follows a water-filling principle for capacity allocation.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Artificial Intelligence arXiv:2605.17393 (cs) [Submitted on 17 May 2026] Title:Heterogeneous Information-Bottleneck Coordination Graphs for Multi-Agent Reinforcement Learning Authors:Wei Duan, Junyu Xuan, En Yu, Xiaoyu Yang, Jie Lu View a PDF of the paper titled Heterogeneous Information-Bottleneck Coordination Graphs for Multi-Agent Reinforcement Learning, by Wei Duan and 4 other authors View PDF HTML (experimental) Abstract:Coordination graphs are a central abstraction in cooperative multi-agent reinforcement learning (MARL), yet existing sparse-graph learners lack a theoretically grounded mechanism to decide which edges should exist and how much information each edge should carry.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
