Research map / Multi-agent systems and games / Cooperative multi-agent RL
LDSA: Learning Dynamic Subtask Assignment in Cooperative Multi-Agent Reinforcement Learning
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 |
|---|---|---|---|
| add both regularizers (subtask-difference and selection-smoothing) (ablation) | LDSA · SMAC corridor, 5m_vs_6m, 2c_vs_64zg | test win rate | Helped |
| add subtask-difference regularizer L_phi (vs smoothing regularizer L_h) (ablation) | LDSA · SMAC ablation scenarios | test win rate | Helped |
| add subtask decoder generating policy parameters (ablation) | LDSA · SMAC ablation scenarios | test win rate | Helped |
| ability-based subtask selection instead of random selection (ablation) | LDSA · SMAC ablation scenarios | test win rate | Helped |
| increase QMIX parameter count to match LDSA (ablation) | QMIX · SMAC ablation scenarios | test win rate | No clear effect |
| increase number of subtasks k beyond 4 (ablation) | LDSA · SMAC corridor | test win rate | No clear effect |
| sample one subtask per agent instead of mixing all subtask policies (ablation) | LDSA · SMAC corridor | test win rate | Helped |
| compute selection from trajectory-subtask similarity instead of directly estimating from trajectories (ablation) | LDSA · SMAC corridor | test win rate | Helped |
| use LDSA instead of QMIX | QMIX · SMAC Easy scenarios | test win rate | No clear effect |
Closest papers by meaning
- Automata-Conditioned Cooperative Multi-Agent Reinforcement Learning (ICML 2026 · similarity 0.90)
- HyperMARL: Adaptive Hypernetworks for Multi-Agent RL (NeurIPS 2025 · similarity 0.89)
- Cooperative Policy Agreement: Learning Diverse Policy for Offline MARL (AAAI 2025 · similarity 0.88)
- Multi-agent cooperation through learning-aware policy gradients (ICLR 2025 · similarity 0.88)
- Learning Joint Behaviors with Large Variations (AAAI 2025 · similarity 0.88)
- Learning Generalizable Skills from Offline Multi-Task Data for Multi-Agent Cooperation (ICLR 2025 · similarity 0.88)
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.