Research map / Multi-agent systems and games / Cooperative multi-agent RL
Communication-Efficient Actor-Critic Methods for Homogeneous Markov Games
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 |
|---|---|---|---|
| Replace all-to-all communication with learned sparse observation-action and parameter communication | Decentralized actor-critic on MPE · MPE tasks | policy performance: 50% | Mixed |
| Use learned communication instead of random communication at the same communication cost | Decentralized actor-critic · MPE tasks | policy performance | Helped |
| Allow observation-action and parameter communication instead of using no communication | Decentralized actor-critic · MPE tasks | policy performance | Helped |
| Use learned bandit parameter consensus instead of random parameter consensus (ablation) | Decentralized actor-critic with learned observation-action communication · Predator-and-Prey (N=15) | policy performance | Helped |
| Use learned bandit parameter consensus instead of Rule-based parameter consensus (ablation) | Decentralized actor-critic with learned observation-action communication · Predator-and-Prey (N=15) | policy performance | Helped |
| Use learned bandit parameter consensus instead of Full-Communication parameter consensus (ablation) | Decentralized actor-critic with learned observation-action communication · Predator-and-Prey (N=15) | policy performance | Helped |
| Use learned bandit parameter consensus instead of random parameter consensus (ablation) | Decentralized actor-critic with learned observation-action communication · Cooperative Push (N=15) | policy performance | No clear effect |
| Use learned bandit parameter consensus instead of random parameter consensus (ablation) | Decentralized actor-critic with learned observation-action communication · Cooperative Navigation (N=15) | policy performance | No clear effect |
| Add policy parameter consensus (ablation) | Communication-efficient decentralized actor-critic · MPE tasks | policy performance | Helped |
| Add policy parameter consensus (ablation) | All-to-all communication actor-critic · Predator-and-Prey | policy performance | Hurt |
| Use rule-based parameter consensus instead of random parameter consensus (ablation) | Decentralized actor-critic · MPE tasks | policy performance | No clear effect |
Closest papers by meaning
- Finite-Time Convergence and Sample Complexity of Multi-Agent Actor-Critic Reinforcement Learning with Average Reward (ICLR 2022 · similarity 0.92)
- Revisiting Some Common Practices in Cooperative Multi-Agent Reinforcement Learning (ICML 2022 · similarity 0.92)
- Taming Communication and Sample Complexities in Decentralized Policy Evaluation for Cooperative Multi-Agent Reinforcement Learning (NeurIPS 2021 · similarity 0.92)
- Fully Decentralized Multi-Agent Reinforcement Learning with Networked Agents (ICML 2018 · similarity 0.91)
- More Centralized Training, Still Decentralized Execution: Multi-Agent Conditional Policy Factorization (ICLR 2023 · similarity 0.91)
- Finite-Time Global Optimality Convergence in Deep Neural Actor-Critic Methods for Decentralized Multi-Agent Reinforcement Learning (ICML 2025 · similarity 0.89)
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.