Research map / Reinforcement learning / Visual and representation RL
Accelerated Policy Learning with Parallel Differentiable Simulation
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 SHAC instead of PPO | policy learning on Humanoid MTU · Humanoid MTU | wall-clock time: over $17\times$ | Helped |
| use SHAC instead of SAC | policy learning on Humanoid MTU · Humanoid MTU | wall-clock time: $30\times$ | Helped |
| use SHAC instead of PPO | policy learning on Humanoid MTU · Humanoid MTU | sample efficiency: $382\times$ | Helped |
| use SHAC instead of SAC | policy learning on Humanoid MTU · Humanoid MTU | sample efficiency: $170\times$ | Helped |
| add a terminal critic to the short-horizon policy objective (ablation) | SHAC · CartPole Swing Up, Ant, Humanoid, and Humanoid MTU | policy learning performance | Helped |
| use short horizon length $h = 16$ or $32$ rather than the other tested lengths (ablation) | SHAC · Ant | best reward | Helped |
| increase the short horizon length $h$ (ablation) | SHAC · Ant | training time | Hurt |
| use a short horizon that is too small (ablation) | SHAC · policy learning | value estimation | Hurt |
| use a short horizon that is too large (ablation) | SHAC · policy learning | optimization landscape quality | Hurt |
Closest papers by meaning
- Accelerating Visual-Policy Learning through Parallel Differentiable Simulation (NeurIPS 2025 · similarity 0.92)
- Stabilizing Reinforcement Learning in Differentiable Multiphysics Simulation (ICLR 2025 · similarity 0.92)
- Differentiable Information Enhanced Model-Based Reinforcement Learning (AAAI 2025 · similarity 0.88)
- Diffusion Policy Policy Optimization (ICLR 2025 · similarity 0.88)
- Deep Reinforcement Learning for Robotics: A Survey of Real-World Successes (AAAI 2025 · similarity 0.88)
- Adaptive Horizon Actor-Critic for Policy Learning in Contact-Rich Differentiable Simulation (ICML 2024 · 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.