Research map / Reinforcement learning / Policy gradient and actor-critic
Trust Region-Guided Proximal Policy Optimization
Yuhui Wang, Hao He, Xiaoyang Tan, Yaozhong Gan
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 constant PPO clipping with trust-region-guided adaptive clipping | PPO · Humanoid | timesteps to threshold (5000): 7241 -> 4653 | Improved |
| Replace constant PPO clipping with trust-region-guided adaptive clipping | PPO · Reacher | average reward, last 40% episodes: -6.7 -> -7.9 | Hurt |
| Replace constant PPO clipping with trust-region-guided adaptive clipping | PPO · Swimmer | timesteps to threshold (90): 564 -> 353.0 | Improved |
| Replace constant PPO clipping with trust-region-guided adaptive clipping | PPO · HalfCheetah | average reward, last 40% episodes: 4600.2 -> 4986.1 | Improved |
| Replace constant PPO clipping with trust-region-guided adaptive clipping | PPO · Hopper | average reward, last 40% episodes: 2848.9 -> 3200.5 | Improved |
| Replace constant PPO clipping with trust-region-guided adaptive clipping | PPO · Walker2d | average reward, last 40% episodes: 3276.2 -> 3886.8 | Improved |
| Use trust-region-guided adaptive clipping instead of PPO clipping | PPO · continuous-armed bandit | trials trapped in local optima: 20% -> almost all trials found the optimal solution | Improved |
| Use trust-region-guided adaptive clipping instead of PPO clipping | PPO · discrete-armed bandit | trials trapped in local optima: 30% -> almost all trials found the optimal solution | Improved |
| Increase PPO clipping parameter from epsilon=0.2 to epsilon=0.6 | PPO · MuJoCo benchmark tasks | task performance | Hurt |
| Add adaptive clipping rather than use constant PPO clipping | PPO · MuJoCo benchmark tasks | policy entropy | Improved |
Closest papers by meaning
- Simple Policy Optimization (ICML 2025 · similarity 0.93)
- DPO Meets PPO: Reinforced Token Optimization for RLHF (ICML 2025 · similarity 0.89)
- PPO in the Fisher-Rao geometry (NeurIPS 2026 · similarity 0.88)
- Colored Noise in PPO: Improved Exploration and Performance through Correlated Action Sampling (AAAI 2024 · similarity 0.87)
- Reflective Policy Optimization (ICML 2024 · similarity 0.87)
- Absolute Policy Optimization: Enhancing Lower Probability Bound of Performance with High Confidence (ICML 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.
