Research map / Multi-agent systems and games / Cooperative multi-agent RL
MIXRTs: Toward Interpretable Multi-Agent Reinforcement Learning via Mixing Recurrent Soft Decision Trees
Zichuan Liu, Yuanyang Zhu, Zhi Wang, Yang Gao, Chunlin Chen
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 |
|---|---|---|---|
| add recurrent tree cells instead of RNN individual values (ablation) | MIXRTs (mixing trees) · SMAC 8m_vs_9m | test win rate (%): 78.90 -> 84.76 | Improved |
| add recurrent tree cells instead of RNN individual values (ablation) | MIXRTs (mixing trees) · SMAC 5m_vs_6m | test win rate (%): 69.17 -> 71.87 | Improved |
| add recurrent tree cells instead of RNN individual values (ablation) | MIXRTs (mixing trees) · SMAC 3s5z | test win rate (%): 87.89 -> 83.98 | Hurt |
| use mixing trees instead of VDN sum mixing (ablation) | RNN agents · SMAC 3s5z | test win rate (%): 82.03 -> 87.89 | Improved |
| use mixing trees instead of VDN sum mixing (ablation) | RNN agents · SMAC 5m_vs_6m | test win rate (%): 73.95 -> 69.17 | Hurt |
| use mixing trees instead of QTRAN mixing (ablation) | RNN agents · SMAC 3s5z | test win rate (%): 17.96 -> 87.89 | Improved |
| use mixing trees instead of QTRAN mixing (ablation) | RNN agents · SMAC 5m_vs_6m | test win rate (%): 55.46 -> 69.17 | Improved |
| mask 30% of least important features | MIXRTs · SMAC 3m | test win rate (%): 100.00 -> 48.38 | Hurt |
| mask 5% of most important features | MIXRTs · SMAC 3m | test win rate (%): 100.00 -> 83.48 | Hurt |
| mask 5% of most important features | MIXRTs · SMAC 5m_vs_6m | test win rate (%): 63.37 -> 13.37 | Hurt |
| mask 20% of least important features | MIXRTs · SMAC 2s3z | test win rate (%): 100.00 -> 70.96 | Hurt |
| increase depth of individual RTCs and mixing trees (ablation) | MIXRTs · SMAC | test win rate (%) | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- QMIX — QMIX: Monotonic Value Function Factorisation for Deep Multi-Agent Reinforcement Learning (ICML 2018) · better on 3 of 3 shared results
- QPLEX — QPLEX: Duplex Dueling Multi-Agent Q-Learning (ICLR 2021) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- MO-MIX: Multi-Objective Multi-Agent Cooperative Decision-Making With Deep Reinforcement Learning (TPAMI 2023 · similarity 0.88)
- N$\text{A}^\text{2}$Q: Neural Attention Additive Model for Interpretable Multi-Agent Q-Learning (ICML 2023 · similarity 0.87)
- Backpropagation Through Agents (AAAI 2024 · similarity 0.86)
- HyperMARL: Adaptive Hypernetworks for Multi-Agent RL (NeurIPS 2025 · similarity 0.86)
- SkillTree: Explainable Skill-Based Deep Reinforcement Learning for Long-Horizon Control Tasks (AAAI 2025 · similarity 0.86)
- Inverse Factorized Soft Q-Learning for Cooperative Multi-agent Imitation Learning (NeurIPS 2024 · similarity 0.85)
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.
