AI Paper Atlas

Research map / Multi-agent systems and games / Cooperative multi-agent RL

Cooperative Heterogeneous Deep Reinforcement Learning

Han Zheng, Pengfei Wei, Jing Jiang, Guodong Long, Qinghua Lu, Chengqi Zhang

NeurIPS 2020 · Multi-agent systems and games › Cooperative multi-agent RL · approach: Decentralized actor-critic MARL · 3 citations (OpenAlex)

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 changedWhereMeasuredEffect
combine SAC, PPO, and CEM with cooperative exploration and local-global memoryCSPC · Humanoid-v2max average return: 5142$\pm$133 -> 5412$\pm$239Improved
combine SAC, PPO, and CEM with cooperative exploration and local-global memoryCSPC · Ant-v2max average return: 3766$\pm$2359 -> 5337$\pm$220Improved
combine SAC, PPO, and CEM with cooperative exploration and local-global memoryCSPC · Walker2d-v2max average return: 4222$\pm$290 -> 5317$\pm$256Improved
combine SAC, PPO, and CEM with cooperative exploration and local-global memoryCSPC · Hopper-v2max average return: 3558 $\pm$139 -> 3619$\pm$52Improved
combine SAC, PPO, and CEM with cooperative exploration and local-global memoryCSPC · Swimmer-v2max average return: 274$\pm$118 -> 261$\pm$117Hurt
add cooperative exploration through policy transfer (ablation)CSPC · Walker2d-v2final average performanceImproved
add local memory replay (ablation)CSPC · Walker2d-v2final average performanceImproved
add global memory replay (ablation)CSPC · Walker2d-v2final average performanceImproved
remove PPO (ablation)CSPC · Walker2d-v2final average performanceHurt
remove CEM (ablation)CSPC · Walker2d-v2final average performanceHurt
remove SAC (ablation)CSPC · Walker2d-v2final average performanceHurt
use three SAC agents with cooperative exploration and local-global memory instead of heterogeneous agents (ablation)CHDRL · Walker2d-v2final average performanceHurt

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy