Research map / Multi-agent systems and games / Cooperative multi-agent RL
Gossip-based Actor-Learner Architectures for Deep Reinforcement Learning
Mahmoud Assran, Joshua Romoff, Nicolas Ballas, Joelle Pineau, Michael Rabbat
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 |
|---|---|---|---|
| replace synchronous AllReduce gradient averaging with asynchronous gossip | A2C · Atari BeamRider (40M steps) | game score: 9829 -> 10188 | Improved |
| replace AllReduce with gossip | A2C · Atari Breakout (40M steps) | game score: 495 -> 690 | Improved |
| replace AllReduce with gossip | A2C · Atari Seaquest (40M steps) | game score: 1894 -> 1892 | No clear effect |
| replace AllReduce with gossip | A2C · Atari SpaceInvaders (40M steps) | game score: 3021 -> 3074 | No clear effect |
| use gossip with many simulators | A2C · six Atari games, 600 runs | percentage of successful runs | Improved |
| use gossip in place of exact gradient averaging | A2C · Atari | sample efficiency | No clear effect |
| use gossip in place of AllReduce | A2C · GPU hardware utilization | GPU utilization: 10-fold improvement | Improved |
| use GALA-A2C at 25M steps instead of IMPALA at 50M | GALA-A2C · Atari Qbert | game score: 18902 -> 18810 | Mixed |
Closest papers by meaning
- Cleanba: A Reproducible and Efficient Distributed Reinforcement Learning Platform (ICLR 2024 · similarity 0.86)
- IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures (ICML 2018 · similarity 0.86)
- SRL: Scaling Distributed Reinforcement Learning to Over Ten Thousand Cores (ICLR 2024 · similarity 0.85)
- Cheaper and Faster: Distributed Deep Reinforcement Learning with Serverless Computing (AAAI 2024 · similarity 0.84)
- A Generalist Hanabi Agent (ICLR 2025 · similarity 0.84)
- Single-Loop Federated Actor-Critic across Heterogeneous Environments (AAAI 2025 · similarity 0.84)
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.
