Research map / Multi-agent systems and games / Cooperative multi-agent RL
Value Propagation for Decentralized Networked 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 decentralized value propagation instead of MA-AC | Cooperative Navigation · Cooperative Navigation | cumulative reward | Helped |
| Use decentralized value propagation instead of PCL without communication | Cooperative Navigation · Cooperative Navigation | learning speed | Helped |
| Use decentralized value propagation instead of independent Q-learning | Cooperative Navigation · Cooperative Navigation | cumulative reward | Helped |
| Use partially observed actor state (ablation) | Value propagation · Cooperative Navigation | cumulative reward | Hurt |
| Use partially observed actor state rather than PCL without communication with full state (ablation) | Value propagation · Cooperative Navigation | cumulative reward | Helped |
| Set η to 0 (pure primal formulation) (ablation) | Value propagation · Random MDP | cumulative reward | Hurt |
| Set η to 1 (ablation) | Value propagation · Random MDP | variance | Hurt |
| Use intermediate η values 0.01 or 0.1 (ablation) | Value propagation · Random MDP | cumulative reward | Helped |
| Use value propagation rather than centralized PCL | Random MDP · Random MDP | convergence speed | Hurt |
| Use multi-step temporal consistency with k>1 rather than k=1 (ablation) | Value propagation · Value propagation experiments | performance | Helped |
| Use decentralized Adam updates (ablation) | Value propagation · Practical training | performance | Helped |
Closest papers by meaning
- Fully Decentralized Multi-Agent Reinforcement Learning with Networked Agents (ICML 2018 · similarity 0.93)
- Taming Communication and Sample Complexities in Decentralized Policy Evaluation for Cooperative Multi-Agent Reinforcement Learning (NeurIPS 2021 · similarity 0.89)
- Finite-Time Global Optimality Convergence in Deep Neural Actor-Critic Methods for Decentralized Multi-Agent Reinforcement Learning (ICML 2025 · similarity 0.89)
- Value Aggregation with Uncertainty in Online Decentralized MARL (ICML 2026 · similarity 0.89)
- Bayesian Ego-graph inference for Networked Multi-Agent Reinforcement Learning (NeurIPS 2025 · similarity 0.87)
- Value-Based Deep Multi-Agent Reinforcement Learning with Dynamic Sparse Training (NeurIPS 2024 · similarity 0.87)
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.