Research map / Multi-agent systems and games / Cooperative multi-agent RL
Multi-Agent Game Abstraction via Graph Attention Neural Network
Yong Liu, Weixun Wang, Yujing Hu, Jianye Hao, Xingguo Chen, Yang Gao
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use GA-Comm with learned two-stage attention instead of CommNet | Communication-based MARL · Traffic Junction easy | success rate: 93.5% -> 99.7% | Improved |
| Use GA-Comm with learned two-stage attention instead of IC3Net | Communication-based MARL · Traffic Junction easy | success rate: 93.2% -> 99.7%; about 6% | Improved |
| Use GA-Comm with learned two-stage attention instead of CommNet | Communication-based MARL · Traffic Junction medium | success rate: 78.8% -> 97.6% | Improved |
| Use GA-Comm with learned two-stage attention instead of IC3Net | Communication-based MARL · Traffic Junction medium | success rate: 90.8% -> 97.6%; about 7% | Improved |
| Use GA-Comm with learned two-stage attention instead of CommNet | Communication-based MARL · Traffic Junction hard | success rate: 6.5% -> 82.3% | Improved |
| Use GA-Comm with learned two-stage attention instead of IC3Net | Communication-based MARL · Traffic Junction hard | success rate: 70.9% -> 82.3%; about 11% | Improved |
| Use GA-AC with learned two-stage attention instead of MAAC | Actor-critic MARL · Predator-Prey | average reward | Improved |
| Use GA-AC with learned two-stage attention instead of MADDPG | Actor-critic MARL · Predator-Prey | average reward | Improved |
| Use two-stage attention rather than soft-attention alone | GA-AC · Predator-Prey | learning speed | Hurt |
Closest papers by meaning
- HGAP: Boosting Permutation Invariant and Permutation Equivariant in Multi-Agent Reinforcement Learning via Graph Attention Network (ICML 2024 · similarity 0.86)
- HYGMA: Hypergraph Coordination Networks with Dynamic Grouping for Multi-Agent Reinforcement Learning (ICML 2025 · similarity 0.85)
- Learning Mean Field Games on Sparse Graphs: A Hybrid Graphex Approach (ICLR 2024 · similarity 0.85)
- Graph Diffusion for Robust Multi-Agent Coordination (ICML 2025 · similarity 0.84)
- Stability of Multi-Agent Learning in Competitive Networks: Delaying the Onset of Chaos (AAAI 2024 · similarity 0.84)
- Bridging Training and Execution via Dynamic Directed Graph-Based Communication in Cooperative Multi-Agent Systems (AAAI 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
