Research map / Reinforcement learning / Policy gradient and actor-critic
Simple random search of static linear policies is competitive for reinforcement learning
Horia Mania, Aurelia Guy, Benjamin Recht
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 online state normalization (V1 -> V2) (ablation) | ARS · Hopper-v1 | episodes to threshold: 89493 -> 3013 | Improved |
| add online state normalization (V1 -> V2) (ablation) | ARS · Humanoid-v1 | episodes to threshold: N/A -> 142600 | Improved |
| add online state normalization (V1 -> V2) (ablation) | ARS · Walker2d-v1 | episodes to threshold: 392000 -> 89600 | Improved |
| use only top-b directions (V2 -> V2-t) (ablation) | ARS · Walker2d-v1 | episodes to threshold: 89600 -> 24000 | Improved |
| use only top-b directions (V2 -> V2-t) (ablation) | ARS · Ant-v1 | episodes to threshold: 60533 -> 20800 | Improved |
| use only top-b directions (V2 -> V2-t) (ablation) | ARS · Humanoid-v1 | episodes to threshold: 142600 -> 142600 | No clear effect |
| use only top-b directions (V1 -> V1-t) (ablation) | ARS · Hopper-v1 | episodes to threshold: 89493 -> 51840 | Improved |
| ARS linear policy instead of TRPO neural policy | MuJoCo locomotion · Walker2d-v1 | episodes to threshold: 14250 -> 24000 | Hurt |
| ARS linear policy instead of TRPO neural policy | MuJoCo locomotion · HalfCheetah-v1 | episodes to threshold: 4250 -> 1707 | Improved |
| evaluate over 100 random seeds instead of 3 | ARS · MuJoCo locomotion | success rate | Hurt |
Closest papers by meaning
- Jump-Start Reinforcement Learning (ICML 2023 · similarity 0.89)
- PWM: Policy Learning with Multi-Task World Models (ICLR 2025 · similarity 0.88)
- Overcoming Slow Decision Frequencies in Continuous Control: Model-Based Sequence Reinforcement Learning for Model-Free Control (ICLR 2025 · similarity 0.88)
- OCEAN-MBRL: Offline Conservative Exploration for Model-Based Offline Reinforcement Learning (AAAI 2024 · similarity 0.88)
- Towards General-Purpose Model-Free Reinforcement Learning (ICLR 2025 · similarity 0.88)
- MAMBA: an Effective World Model Approach for Meta-Reinforcement Learning (ICLR 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.
