Research map / Multi-agent systems and games / Cooperative multi-agent RL
ACE: Cooperative Multi-Agent Q-learning with Bidirectional Action-Dependency
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 |
|---|---|---|---|
| Use bidirectional action-dependency with sequential decision-making | Cooperative MARL · Spiders-and-Fly 5×5 | Steps gap to oracle: 0.60$\pm$0.09 -> 0.04$\pm$0.03 | Helped |
| Use bidirectional action-dependency with sequential decision-making | Cooperative MARL · Spiders-and-Fly 7×7 | Samples (M) to 100% success within 10 steps: 1.68$\pm$0.09 -> 1.01$\pm$0.06 | Helped |
| Use ACE instead of value-factorization baselines | SMAC · Hard and super-hard maps | Test win rate | Helped |
| Use ACE instead of CDS-QMIX or CDS-QPLEX | Google Research Football · academy_3_vs_1_with_keeper and academy_counterattack_hard | Win rate | Helped |
| Add interaction-aware action embeddings (ablation) | ACE · SMAC corridor and 3s5z_vs_3s6z | Win rate | Helped |
| Use sorted rather than shuffled agent order (ablation) | ACE · SMAC two evaluated maps | Win rate | No clear effect |
| Replace Q-learning with PPO on the sequentially expanded MDP (ablation) | ACE · SMAC 5m_vs_6m | Performance | No clear effect |
| Distill ACE's joint action into an individual value function for decentralized execution (ablation) | ACE-CTDE · SMAC | Performance | No clear effect |
| Transfer a policy trained on 5m_vs_6m without fine-tuning | ACE · SMAC 8m_vs_9m | Win rate: 40% | Mixed |
Closest papers by meaning
- O-MAPL: Offline Multi-agent Preference Learning (ICML 2025 · similarity 0.88)
- Backpropagation Through Agents (AAAI 2024 · similarity 0.88)
- Sequential Order Adjustment of Action Decisions for Multi-Agent Transformer (Student Abstract) (AAAI 2025 · similarity 0.88)
- Revisiting Cooperative Off-Policy Multi-Agent Reinforcement Learning (ICML 2025 · similarity 0.87)
- Inverse Factorized Soft Q-Learning for Cooperative Multi-agent Imitation Learning (NeurIPS 2024 · similarity 0.87)
- Learning Joint Behaviors with Large Variations (AAAI 2025 · similarity 0.87)
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.