Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Maximum Entropy Population-Based Training for Zero-Shot Human-AI Coordination
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 |
|---|---|---|---|
| Train the final agent with a population-entropy-diversified partner population | MEP · Overcooked, five layouts with H_Proxy | average episode reward | Helped |
| Train using partners sampled from a diversified population | MEP · Overcooked, real-human evaluation | average episode reward | Helped |
| Add the population entropy reward (ablation) | MEP · Overcooked, five layouts with H_Proxy | average episode reward | Helped |
| Add learning-progress-based prioritized partner sampling (ablation) | MEP · Overcooked, five layouts with H_Proxy | average episode reward | Helped |
| Increase the population-entropy weight α above 0 (ablation) | MEP population training · Overcooked, five layouts | population entropy | Helped |
| Increase FCP population size from 5 to 10 (ablation) | FCP · Overcooked, H_Proxy evaluation | average episode reward | Helped |
| Use MEP rather than TrajeDi | Best response trained against each population · Matrix game | self-play and cross-play return | Helped |
Closest papers by meaning
- Diversity Is Not All You Need: Training A Robust Cooperative Agent Needs Specialist Partners (NeurIPS 2024 · similarity 0.88)
- Learning to Cooperate with Humans using Generative Agents (NeurIPS 2024 · similarity 0.88)
- Cross-environment Cooperation Enables Zero-shot Multi-agent Coordination (ICML 2025 · similarity 0.88)
- Maximum Entropy Heterogeneous-Agent Reinforcement Learning (ICLR 2024 · similarity 0.88)
- Towards Principled Unsupervised Multi-Agent Reinforcement Learning (NeurIPS 2025 · similarity 0.87)
- Learning Diverse Risk Preferences in Population-Based Self-Play (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.