Research map / Multi-agent systems and games / Cooperative multi-agent RL
Decentralized Natural Policy Gradient with Variance Reduction for Collaborative Multi-Agent Reinforcement Learning
Jinchi Chen, Jie Feng, Weiguo Gao, Ke Wei
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 variance reduction (momentum) to decentralized NPG (ablation) | MDNPG · cooperative navigation and multi-task GridWorld (ring) | average return | Improved |
| add gradient tracking to MDNPG (ablation) | MDNPG · cooperative navigation and multi-task GridWorld (ring) | average return | Improved |
| train a shared multi-task policy instead of training each agent separately (ablation) | MDNPG · Multi-task GridWorld | average return: 10.43 -> -88.37 | Improved |
| use MDNPG instead of MDPGT | decentralized policy gradient · Multi-task GridWorld sum | average return: -56.98 -> 10.43 | Improved |
| use MDNPG instead of PG with entropy regularization | policy gradient · Multi-task GridWorld sum | average return: -28.1 -> 10.43 | Improved |
| use momentum-based NPG instead of PPO, PG with momentum, SRVR-NPG | single-agent NPG · GridWorld and MountainCar | average return | Improved |
| use MDNPG instead of MDPGT and value propagation across ring, fully-connected, bi-partite topologies | MDNPG · cooperative navigation (5 agents) | average return | Improved |
| vary momentum parameter beta (0.2 vs others) (ablation) | MDNPG · cooperative navigation and multi-task GridWorld (ring) | average return | No clear effect |
Closest papers by meaning
- MDPGT: Momentum-Based Decentralized Policy Gradient Tracking (AAAI 2022 · similarity 0.94)
- Decentralized Policy Gradient Descent Ascent for Safe Multi-Agent Reinforcement Learning (AAAI 2021 · similarity 0.89)
- Federated Natural Policy Gradient and Actor Critic Methods for Multi-task Reinforcement Learning (NeurIPS 2024 · similarity 0.88)
- Provably Fast Convergence of Independent Natural Policy Gradient for Markov Potential Games (NeurIPS 2023 · similarity 0.88)
- Momentum for the Win: Collaborative Federated Reinforcement Learning across Heterogeneous Environments (ICML 2024 · similarity 0.88)
- Natural Policy Gradient as Doubly Smoothed Policy Iteration: A Bellman-Operator Framework (NeurIPS 2026 · 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.
