Research map / Reinforcement learning / Visual and representation RL
Proximal Distilled Evolutionary Reinforcement Learning
Cristian Bodnar, Ben Day, Pietro Lió
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 |
|---|---|---|---|
| combine Q-filtered distillation crossover with proximal mutations | ERL (PDERL) · Swimmer-v2, HalfCheetah-v2, Hopper-v2, Walker2D-v2, Ant-v2 | mean reward: 337; 13522; 3397; 5184; 6845 | Improved |
| add proximal mutations to Q-filtered distillation crossover (ablation) | DERL · Walker2D-v2 and Ant-v2 | mean reward: 4050 -> 5184; 4911 -> 6845 | Improved |
| add proximal mutations to Q-filtered distillation crossover (ablation) | DERL · HalfCheetah-v2 and Hopper-v2 | mean reward: 11362 -> 13522; 2869 -> 3397 | Mixed |
| add Q-filtered distillation crossover to proximal mutations (ablation) | PERL · Walker2D-v2 and Ant-v2 | mean reward: 3364 -> 5184; 4528 -> 6845 | Improved |
| add Q-filtered distillation crossover to proximal mutations (ablation) | PERL · HalfCheetah-v2 and Hopper-v2 | mean reward: 13668 -> 13522; 3497 -> 3397 | Hurt |
| replace n-point crossover with Q-filtered behaviour distillation crossover (ablation) | ERL genetic population · Ant | offspring fitness relative to best parent | Improved |
| replace Gaussian mutation with proximal mutation (ablation) | ERL genetic population · Ant | offspring total reward: -187 -> 5496 | Improved |
| use distance-based rather than greedy parent selection (ablation) | DERL and PDERL · Walker2D-v2 | mean reward | Hurt |
| use distance-based rather than greedy parent selection (ablation) | PDERL · Hopper-v2 | mean reward | Improved |
| increase proximal mutation magnitude (ablation) | PDERL · policy behaviour | behaviour change | Improved |
Closest papers by meaning
- ERL-TD: Evolutionary Reinforcement Learning Enhanced with Truncated Variance and Distillation Mutation (AAAI 2024 · similarity 0.90)
- Two-Stage Evolutionary Reinforcement Learning for Enhancing Exploration and Exploitation (AAAI 2024 · similarity 0.89)
- Evolutionary Reinforcement Learning with Parameterized Action Primitives for Diverse Manipulation Tasks (AAAI 2025 · similarity 0.88)
- EvoRainbow: Combining Improvements in Evolutionary Reinforcement Learning for Policy Search (ICML 2024 · similarity 0.87)
- Value-Evolutionary-Based Reinforcement Learning (ICML 2024 · similarity 0.86)
- ConfigX: Modular Configuration for Evolutionary Algorithms via Multitask Reinforcement Learning (AAAI 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.
