Research map / Reinforcement learning / Policy gradient and actor-critic
Policy Optimization by Genetic Distillation
Tanmay Gangwani, Jian Peng
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 |
|---|---|---|---|
| Use state-space imitation-learning crossover instead of parameter-space crossover | GPO policies · HalfCheetah | average episode reward | Improved |
| Add data-sharing during policy-gradient mutation (ablation) | Base policy-optimization procedure · Ant and Walker-2D (normalized average) | normalized performance: 0.33 -> 0.83 | Improved |
| Add state-space imitation-learning crossover (ablation) | Base policy-optimization procedure · Ant and Walker-2D (normalized average) | normalized performance: 0.33 -> 0.47 | Improved |
| Use performance-fitness selection rather than random selection (ablation) | Base policy-optimization procedure · Ant and Walker-2D (normalized average) | normalized performance: 0.33 -> 0.47 | Improved |
| Combine data-sharing with performance-based parent selection (ablation) | Base policy-optimization procedure · Ant and Walker-2D (normalized average) | normalized performance: 0.83 -> 0.86 | Improved |
| Use GPO rather than the strongest policy-gradient baseline | PPO policy optimization · MuJoCo locomotion tasks | final mean episode reward | Mixed |
| Use GPO rather than the strongest policy-gradient baseline | A2C policy optimization · MuJoCo locomotion tasks | final mean episode reward | Mixed |
| Increase GPO population size | GPO · Walker2D | performance | Improved |
Closest papers by meaning
- Evolution-Guided Policy Gradient in Reinforcement Learning (NeurIPS 2018 · similarity 0.88)
- Genetic Soft Updates for Policy Evolution in Deep Reinforcement Learning (ICLR 2021 · similarity 0.88)
- DGPO: Discovering Multiple Strategies with Diversity-Guided Policy Optimization (AAAI 2024 · similarity 0.87)
- Diffusion Policy Policy Optimization (ICLR 2025 · similarity 0.87)
- Adaptive-Gradient Policy Optimization: Enhancing Policy Learning in Non-Smooth Differentiable Simulations (ICML 2024 · similarity 0.86)
- GenPO: Generative Diffusion Models Meet On-Policy Reinforcement Learning (NeurIPS 2025 · similarity 0.86)
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.
