Research map / Reinforcement learning / Policy gradient and actor-critic
Trust-PCL: An Off-Policy Trust Region Method for Continuous Control
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 |
|---|---|---|---|
| Use Trust-PCL instead of TRPO (ours) | continuous-control policy optimization · HalfCheetah | best average reward in first 10M steps: 4343.6 -> 7057.1 | Helped |
| Use Trust-PCL instead of TRPO (ours) | continuous-control policy optimization · Swimmer | best average reward in first 10M steps: 288.1 -> 297.0 | Helped |
| Use Trust-PCL instead of TRPO (ours) | continuous-control policy optimization · Hopper | best average reward in first 10M steps: 3516.7 -> 3804.9 | Helped |
| Use Trust-PCL instead of TRPO (ours) | continuous-control policy optimization · Walker2d | best average reward in first 10M steps: 2838.4 -> 5027.2 | Helped |
| Use Trust-PCL instead of TRPO (ours) | continuous-control policy optimization · Ant | best average reward in first 10M steps: 4347.5 -> 6104.2 | Helped |
| Add a relative-entropy trust region to standard PCL (ablation) | PCL on the evaluated continuous-control tasks · continuous-control tasks | training stability | Helped |
| Increase the trust-region size epsilon (ablation) | Trust-PCL · continuous-control tasks | training stability | Hurt |
| Train with off-policy replay and collect fewer environment steps per update (ablation) | Trust-PCL · continuous-control tasks | sample efficiency | Helped |
| Increase entropy regularization above zero (ablation) | Trust-PCL · evaluated continuous-control tasks | reward | No clear effect |
Closest papers by meaning
- An Implicit Trust Region Approach to Behavior Regularized Offline Reinforcement Learning (AAAI 2024 · similarity 0.90)
- Embedding Safety into RL: A New Take on Trust Region Methods (ICML 2025 · similarity 0.90)
- Trust Region Q Adjoint Matching (NeurIPS 2026 · similarity 0.88)
- Simple Policy Optimization (ICML 2025 · similarity 0.88)
- Trust Region Reward Optimization and Proximal Inverse Reward Optimization Algorithm (NeurIPS 2025 · similarity 0.88)
- Learning to Trust Bellman Updates: Selective State-Adaptive Regularization for Offline RL (ICML 2025 · 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.