Cooperation Graph

Cooperation graphs represent a novel approach in multi-agent reinforcement learning (MARL), aiming to improve the efficiency and interpretability of cooperative behavior in complex tasks, particularly those with sparse rewards. Current research focuses on developing hierarchical graph structures, such as extensible or bipartite graphs, and associated algorithms that dynamically adjust the graph topology to optimize agent collaboration. These models leverage the graph structure to guide agent actions, often outperforming traditional MARL methods by facilitating efficient knowledge sharing and the integration of pre-existing cooperative knowledge, leading to improved performance in challenging multi-agent scenarios. This approach holds significant promise for advancing MARL's applicability to real-world problems requiring coordinated action among multiple agents.

Papers

March 26, 2024

Self-Clustering Hierarchical Multi-Agent Reinforcement Learning with Extensible Cooperation Graph
Qingxu Fu, Tenghai Qiu, Jianqiang Yi, Zhiqiang Pu, Xiaolin Ai
Multi Agent Reinforcement Learning Multi Agent Problem Self Supervised Clustering Cooperation Graph

June 16, 2023

Structured Cooperative Learning with Graphical Model Priors
Shuangtong Li, Tianyi Zhou, Xinmei Tian, Dacheng Tao
Graphical Model Mutual Learning Cooperative Learning Graphical Model Prior Cooperation Graph

August 5, 2022

A Cooperation Graph Approach for Multiagent Sparse Reward Reinforcement Learning
Qingxu Fu, Tenghai Qiu, Zhiqiang Pu, Jianqiang Yi, Wanmai Yuan
Sparse Reward Multi Agent Task Cooperation Graph

Cooperation Graph

Papers

Self-Clustering Hierarchical Multi-Agent Reinforcement Learning with Extensible Cooperation Graph

Structured Cooperative Learning with Graphical Model Priors

A Cooperation Graph Approach for Multiagent Sparse Reward Reinforcement Learning