Research map / Multi-agent systems and games / Cooperative multi-agent RL
Heterogeneous Skill Learning for Multi-agent Tasks
Yuntao Liu, Yuan Li, Xinhai Xu, Yong Dou, Donghong Liu
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 |
|---|---|---|---|
| Use HSL (skill representation, skill selector, skill-based policy learning) instead of CDS | HSL · SMAC 6h_vs_8z | test win rate: around 10% | Improved |
| Use HSL instead of CDS | HSL · SMAC 5m_vs_6m | test win rate: around 20% | Improved |
| Use HSL instead of CDS on a large-scale map (many agents) | HSL · SMAC 27m_vs_30m | test win rate | Improved |
| Use HSL instead of RODE in heterogeneous-agent maps (MMM2, 3s5z_vs_3s6z) | HSL · SMAC MMM2 and 3s5z_vs_3s6z | test win rate | Improved |
| Use HSL instead of HSD (skill discovery from local observation only) | HSL · SMAC Super Hard maps | test win rate | Improved |
| Use HSL instead of QMIX/QPLEX/role-based methods | HSL · GRF 3 vs 1 with keeper, Hard Counter-attack, Corner | test win rate | Improved |
| Use HSL instead of CDS with many agents | HSL · GRF Corner | test win rate | Improved |
| Use HSL instead of role-based methods and baselines | HSL · GoBigger 3 vs 3 and 3 vs 3 with thorn | test win rate | Improved |
| Add the skill selector instead of random skill assignment (ablation) | HSL · SMAC 6h_vs_8z, GRF Hard Counter-attack, GoBigger 3 vs 3 | test win rate | Improved |
| Add the learned skill representation instead of raw states for skill assignment (ablation) | HSL · SMAC 6h_vs_8z, GRF Hard Counter-attack, GoBigger 3 vs 3 | test win rate | Improved |
| Add the intrinsic (entropy) reward to the skill selector objective (ablation) | HSL · GRF Hard Counter-attack and GoBigger 3 vs 3 | test win rate | Improved |
| Add the intrinsic reward to the skill selector (ablation) | HSL · SMAC 6h_vs_8z | test win rate | No clear effect |
Closest papers by meaning
- Hierarchical Multi-Agent Skill Discovery (NeurIPS 2023 · similarity 0.90)
- Learning Generalizable Skills from Offline Multi-Task Data for Multi-Agent Cooperation (ICLR 2025 · similarity 0.89)
- Maximum Entropy Heterogeneous-Agent Reinforcement Learning (ICLR 2024 · similarity 0.87)
- Disentangled Unsupervised Skill Discovery for Efficient Hierarchical Reinforcement Learning (NeurIPS 2024 · similarity 0.87)
- Robust Policy Learning via Offline Skill Diffusion (AAAI 2024 · similarity 0.87)
- Acquiring Diverse Skills using Curriculum Reinforcement Learning with Mixture of Experts (ICML 2024 · similarity 0.86)
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.
