Research map / Reinforcement learning / Policy gradient and actor-critic
Evolution-Guided Policy Gradient in Reinforcement Learning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 EA population search with DDPG learning and periodically insert the RL actor into the population | ERL · Hard Inverted Double Pendulum | episodes to solve: $10000$ vs. $22000$ episodes | Helped |
| Combine EA population search with DDPG learning | ERL · Mujoco continuous-control benchmarks | learning performance | Helped |
| Use ERL instead of EA | Mujoco continuous-control benchmarks · All except Swimmer | learning performance | Helped |
| Use ERL instead of EA | Swimmer · Swimmer | learning performance | No clear effect |
| Use ERL instead of PPO | Mujoco continuous-control benchmarks · Four of six benchmarks | learning performance | Helped |
| Use ERL instead of PPO | Hopper and Walker2D · Hopper and Walker2D | sample efficiency | Hurt |
| Remove the selection operator (ablation) | ERL · HalfCheetah and Swimmer | normalized learning performance: $\sim$80\% | Hurt |
| Add episode-wide fitness-based selection | ERL · Hard Inverted Double Pendulum | solves task | Helped |
| Add periodic synchronization of the RL actor into the EA population | ERL · Mujoco continuous-control benchmarks | learning performance | Mixed |
Closest papers by meaning
- ERL-TD: Evolutionary Reinforcement Learning Enhanced with Truncated Variance and Distillation Mutation (AAAI 2024 · similarity 0.92)
- EvoRainbow: Combining Improvements in Evolutionary Reinforcement Learning for Policy Search (ICML 2024 · similarity 0.90)
- Value-Evolutionary-Based Reinforcement Learning (ICML 2024 · similarity 0.89)
- 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 0.89)
- Learning to Reuse Policies in State Evolvable Environments (ICML 2025 · similarity 0.88)
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.