Research map / Multi-agent systems and games / Cooperative multi-agent RL
Two Heads are Better Than One: A Simple Exploration Framework for Efficient Multi-Agent Reinforcement Learning
Jiahui Li, Kun Kuang, Baoxiang Wang, Xingchen Li, Fei Wu, Jun Xiao, Long Chen
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 |
|---|---|---|---|
| Add COIN (influence + curiosity intrinsic rewards) to QMIX/QPLEX | QMIX, QPLEX · SMAC 3s5z_vs_3s6z, 6h_vs_8z, corridor | test win rate | Improved |
| Add COIN to baselines on Aloha | QMIX, QPLEX · MACO Aloha | mean transmissions: nearly 20 mean transmissions | Improved |
| Add other exploration methods (EITI, CIExplore, EMC) to baselines on Aloha | QMIX/QPLEX · MACO Aloha | mean transmissions | No clear effect |
| Add COIN on Gather | QMIX, QPLEX · MACO Gather | convergence speed / win rate | Mixed |
| Add COIN to QPLEX on Hallway | QPLEX · MACO Hallway | mean win rate: 60% improvement | Improved |
| Add any exploration method to QMIX on Hallway | QMIX · MACO Hallway | mean win rate | No clear effect |
| Add COIN to QMIX in sparse-reward football | QMIX · GRF academy 3 vs 1 with keeper | mean score per episode: nearly 0.2 mean scores per episode | Improved |
| Add EMC curiosity-based exploration to QPLEX | QPLEX · GRF academy 3 vs 1 with keeper | score | Hurt |
| Add only influence-based or only curiosity-based reward to QMIX (ablation) | QMIX · GRF academy 3 vs 1 with keeper | mean score: decreased by nearly 0.5 mean scores | Hurt |
| Add curiosity-based reward to QMIX (ablation) | QMIX · SMAC 3s5z_vs_3s6z | test win rate | Hurt |
| Add influence-based reward to QMIX (ablation) | QMIX · SMAC 3s5z_vs_3s6z | test win rate | Improved |
| Add either influence or curiosity reward to QMIX (ablation) | QMIX · SMAC corridor | test win rate | Improved |
Closest papers by meaning
- Settling Decentralized Multi-Agent Coordinated Exploration by Novelty Sharing (AAAI 2024 · similarity 0.90)
- Randomized Exploration in Cooperative Multi-Agent Reinforcement Learning (NeurIPS 2024 · similarity 0.90)
- Individual Contributions as Intrinsic Exploration Scaffolds for Multi-agent Reinforcement Learning (ICML 2024 · similarity 0.89)
- Enhancing Cooperative Multi-Agent Reinforcement Learning with State Modelling and Adversarial Exploration (ICML 2025 · similarity 0.89)
- The Horizon Threshold in Cooperative Multi-Agent Reward-Free Exploration (NeurIPS 2026 · similarity 0.89)
- Measuring Mutual Policy Divergence for Multi-Agent Sequential Exploration (NeurIPS 2024 · similarity 0.89)
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.
