Research map / Multi-agent systems and games / Cooperative multi-agent RL
Communication Learning via Backpropagation in Discrete Channels with Unknown Noise
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 discrete differentiable communication (randomized encoder/decoder) instead of reinforced communication learning (RCL) | 2-agent MARL with 40-bit channel · Hidden-goal path-finding, noise-free | mean episode length/reward | Helped |
| use noise-tolerant encrypted discrete differentiable communication instead of RCL | 2-agent MARL with bit-flip channel noise · Hidden-goal path-finding, noisy channel | mean episode reward | Helped |
| use discrete (16-level) messages instead of real-valued messages | differentiable communication · Hidden-goal path-finding, noise-free | learning speed | No clear effect |
| use discrete (16-level) messages instead of real-valued messages | differentiable communication · Coordinated search, noise-free | mean episode reward | Hurt |
| add discrete differentiable communication versus no communication | 2-agent MARL · Coordinated search, noise-free | mean episode reward | Helped |
| add discrete differentiable communication instead of RCL | 2-agent MARL · Coordinated search, noise-free | mean episode reward | Helped |
| noisy channel with encryption approach versus RCL | 2-agent MARL · Coordinated search, noisy channel | mean episode reward | No clear effect |
| reduce quantization levels per element from 16 to 8 or 4 (10 elements) (ablation) | discrete differentiable communication · Hidden-goal path-finding | final reward | No clear effect |
| reduce message to 10 elements with 2 levels, or 5 elements with 2 levels (ablation) | discrete differentiable communication · Hidden-goal path-finding | final reward | Hurt |
| reduce message to 2 elements with 2 levels (below required 6.64 bits) (ablation) | discrete differentiable communication · Hidden-goal path-finding | final reward | Hurt |
Closest papers by meaning
- Multi-Agent Reinforcement Learning with Communication-Constrained Priors (NeurIPS 2025 · similarity 0.90)
- RGMComm: Return Gap Minimization via Discrete Communications in Multi-Agent Reinforcement Learning (AAAI 2024 · similarity 0.88)
- Learning Multi-Agent Communication with Contrastive Learning (ICLR 2024 · similarity 0.87)
- Learning Efficient and Robust Multi-Agent Communication via Graph Information Bottleneck (AAAI 2024 · similarity 0.87)
- Exponential Topology-enabled Scalable Communication in Multi-agent Reinforcement Learning (ICLR 2025 · similarity 0.87)
- Robust Communicative Multi-Agent Reinforcement Learning with Active Defense (AAAI 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.