Research map / Multi-agent systems and games / Cooperative multi-agent RL
Learning to Communicate with Deep 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 |
|---|---|---|---|
| Use differentiable communication instead of Q-learning communication | Multi-agent communication learning · Switch riddle (n=4) | normalized reward | Helped |
| Use differentiable communication instead of Q-learning communication | Multi-agent communication learning · Multi-Step MNIST | normalized reward | Helped |
| Use differentiable communication instead of Q-learning communication | Multi-agent communication learning · Colour-Digit MNIST | normalized reward | Helped |
| Add parameter sharing across agents (ablation) | DIAL and RIAL · Switch riddle (n=3) | normalized reward | Helped |
| Add parameter sharing across agents (ablation) | DIAL and RIAL · Switch riddle (n=4) | normalized reward | Helped |
| Add parameter sharing across agents (ablation) | DIAL and RIAL · Multi-Step MNIST | normalized reward | Helped |
| Add parameter sharing across agents (ablation) | DIAL and RIAL · Colour-Digit MNIST | normalized reward | Helped |
| Remove communication (ablation) | Shared-parameter NoComm baseline · Switch riddle (n=3) | normalized reward | Hurt |
| Remove communication (ablation) | Shared-parameter NoComm baseline · Multi-Step MNIST | normalized reward | Hurt |
| Remove communication (ablation) | Shared-parameter NoComm baseline · Colour-Digit MNIST | normalized reward | Hurt |
| Remove noise from the differentiable communication channel (ablation) | DIAL · Switch riddle | communication protocol learning | Hurt |
| Use DIAL without parameter sharing (ablation) | DIAL · Switch riddle (n=4) | normalized reward | Mixed |
Closest papers by meaning
- Learning Multi-Agent Communication through Structured Attentive Reasoning (NeurIPS 2020 · similarity 0.91)
- Learning Multi-Agent Communication with Contrastive Learning (ICLR 2024 · similarity 0.89)
- Multi-Agent Reinforcement Learning with Communication-Constrained Priors (NeurIPS 2025 · similarity 0.88)
- Efficient Communication in Multi-Agent Reinforcement Learning with Implicit Consensus Generation (AAAI 2025 · similarity 0.87)
- Expressive Multi-Agent Communication via Identity-Aware Learning (AAAI 2024 · similarity 0.87)
- Learning Multi-Agent Communication from Graph Modeling Perspective (ICLR 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.