Research map / Multi-agent systems and games / Cooperative multi-agent RL
Cooperative Exploration for Multi-Agent Deep Reinforcement Learning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 CMAE shared-goal exploration with restricted space selection | Q-learning on sparse-reward MPE · Pass-sparse, Secret-Room-sparse, Push-Box-sparse | final episode reward: 1.00 vs 0.00 for baselines | Helped |
| add CMAE shared-goal exploration | Q-learning on dense-reward MPE · Push-Box-dense | final episode reward: 1.38 vs 1.58 for Q-learning | Hurt |
| add CMAE shared-goal exploration | Q-learning on dense-reward MPE · Pass-dense, Secret-Room-dense | final episode reward: 5.00 vs 1.25 (Pass); 4.00 vs 1.62 (Secret-Room) | Helped |
| add count-based exploration bonus to QMIX | QMIX on sparse-reward SMAC · 3m-sparse, 2m_vs_1z-sparse | final success rate (%): 0.0 to 11.7 (3m); 0.0 to 19.8 (2m_vs_1z) | Helped |
| add CMAE shared-goal exploration to QMIX | QMIX on sparse-reward SMAC · 3m-sparse, 2m_vs_1z-sparse | final success rate (%): 47.7 vs 11.7 (3m); 44.3 vs 19.8 (2m_vs_1z) against best baseline | Helped |
| add CMAE to QMIX on dense-reward SMAC | QMIX on dense-reward SMAC · 3m-dense, 2m_vs_1z-dense, 3s_vs_5z-dense | final success rate (%): comparable to baselines (e.g., 81.3 vs 92.2/95.3 on 3s_vs_5z-dense) | No clear effect |
| remove decoupling of target and exploration policies (ablation) | CMAE on 3m-sparse · 3m-sparse | success rate (%): 47.7 to 25.4 | Hurt |
| add restricted space exploration (vs. exploring full state space directly) (ablation) | CMAE on 3m-sparse · 3m-sparse | success rate (%): 47.7 vs 9.4 without restricted spaces | Helped |
| CMAE on sparse 3s_vs_5z | CMAE · 3s_vs_5z-sparse | success rate (%): 0.0 for all methods | No clear effect |
Closest papers by meaning
- Enhancing Cooperative Multi-Agent Reinforcement Learning with State Modelling and Adversarial Exploration (ICML 2025 · similarity 0.90)
- Settling Decentralized Multi-Agent Coordinated Exploration by Novelty Sharing (AAAI 2024 · similarity 0.89)
- Randomized Exploration in Cooperative Multi-Agent Reinforcement Learning (NeurIPS 2024 · similarity 0.89)
- AIR: Unifying Individual and Collective Exploration in Cooperative Multi-Agent Reinforcement Learning (AAAI 2025 · similarity 0.88)
- Learning Joint Behaviors with Large Variations (AAAI 2025 · similarity 0.88)
- Imagine, Initialize, and Explore: An Effective Exploration Method in Multi-Agent Reinforcement Learning (AAAI 2024 · similarity 0.88)
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.