Research map / Multi-agent systems and games / Cooperative multi-agent RL
E-MAPP: Efficient Multi-Agent Reinforcement Learning with Parallel Program Guidance
Can Chang, Ni Mu, Jiajun Wu, Ling Pan, Huazhe Xu
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 |
|---|---|---|---|
| Guide agents with parallel programs and learned task allocation | E-MAPP · Overcooked seen tasks in novel maps, Hard | completion rate: 0.00% -> 56.3% | Improved |
| Guide agents with parallel programs and learned task allocation | E-MAPP · Overcooked unseen tasks in novel maps, Hard | completion rate: 0.0% -> 43.7% | Improved |
| Add a feasibility predictor (ablation) | E-MAPP · Overcooked hard tasks | completion rate: 38.5% -> 56.3% | Improved |
| Add a feasibility predictor (ablation) | E-MAPP · Overcooked hard tasks | average score: 1.42 ± 0.56 -> 1.58 ± 0.60 | Improved |
| Add a reachability predictor (ablation) | E-MAPP · Overcooked hard tasks | average timesteps: 23.37 -> 17.6 | Improved |
| Add a reachability predictor (ablation) | E-MAPP · Overcooked hard tasks | completion rate: 43.8% -> 56.3% | Improved |
| Add a cost-to-go predictor (ablation) | E-MAPP · Overcooked hard tasks | average timesteps: 20.42 -> 17.6 | Improved |
| Add a cost-to-go predictor (ablation) | E-MAPP · Overcooked hard tasks | completion rate: 52.0% -> 56.3% | Improved |
| Use parallel rather than sequential programs (ablation) | E-MAPP · Overcooked hard tasks | average timesteps: 20.33 -> 17.6 | Improved |
| Use parallel rather than sequential programs (ablation) | E-MAPP · Overcooked hard tasks | completion rate: 48.8% -> 56.3% | Improved |
| Use E-MAPP in a partially observable setting instead of its original full-observation setting | E-MAPP · Overcooked hard tasks | completion rate: 56.3% -> 27.1% | Hurt |
| Use E-MAPP rather than MAPPO | Overcooked hard tasks · Overcooked hard tasks | average score: 0.59± 0.27 -> 1.58 ± 0.60 | Improved |
Closest papers by meaning
- Scalable Multi-Agent Reinforcement Learning through Intelligent Information Aggregation (ICML 2023 · similarity 0.88)
- Efficient Episodic Memory Utilization of Cooperative Multi-Agent Reinforcement Learning (ICLR 2024 · similarity 0.88)
- Learning to Share: Selective Memory for Efficient Parallel Agentic Systems (ICML 2026 · similarity 0.88)
- Efficient Multi-agent Reinforcement Learning by Planning (ICLR 2024 · similarity 0.87)
- POGEMA: A Benchmark Platform for Cooperative Multi-Agent Pathfinding (ICLR 2025 · similarity 0.87)
- Learn to Follow: Decentralized Lifelong Multi-Agent Pathfinding via Planning and Learning (AAAI 2024 · 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.
