Research map / Reinforcement learning / Policy gradient and actor-critic
Regularizing Trajectory Optimization with Denoising Autoencoders
Rinu Boney, Norman Di Palo, Mathias Berglund, Alexander Ilin, Juho Kannala, Antti Rasmus, Harri Valpola
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 |
|---|---|---|---|
| Add DAE regularization to CEM trajectory optimization | PETS · Half-cheetah | Average Return: 10955 ± 2865 -> 12967 ± 3216 | Improved |
| Use Adam with DAE regularization instead of unregularized Adam | PETS · Half-cheetah | Average Return | Improved |
| Use DAE regularization during open-loop trajectory optimization | PETS with Adam · Half-cheetah | agreement between predicted and true rewards | Improved |
| Add DAE regularization to trajectory optimization | Model-based RL · Cartpole | learning speed | No clear effect |
| Train the dynamics model for more epochs after each episode (ablation) | Single-model end-to-end training with CEM · Motor-control tasks | early learning performance | Improved |
| Add DAE regularization (ablation) | Single-model end-to-end training with CEM · Motor-control tasks | learning progress | Improved |
| Replace DAE regularization with diagonal-Gaussian regularization (ablation) | Adam trajectory optimization · Half-cheetah | learning performance | Hurt |
| Allow optimized actions to include random noise (ablation) | DAE-regularized model-based RL · Motor-control tasks | asymptotic performance | Improved |
| Stop gradient propagation through the denoising autoencoder (ablation) | DAE-regularized trajectory optimization · Experiments | optimization performance | No clear effect |
| Use DAE regularization with Adam instead of unregularized Adam | Model-based RL · Half-cheetah end-to-end training | early learning speed | Improved |
| Use DAE regularization without adding explicit exploration | Model-based RL · Half-cheetah | asymptotic performance | Hurt |
Closest papers by meaning
- Diffusion-ES: Gradient-free Planning with Diffusion for Autonomous and Instruction-guided Driving (CVPR 2024 · similarity 0.88)
- Score Regularized Policy Optimization through Diffusion Behavior (ICLR 2024 · similarity 0.86)
- Iteratively Refined Behavior Regularization for Offline Reinforcement Learning (NeurIPS 2024 · similarity 0.86)
- Behavior-Regularized Diffusion Policy Optimization for Offline Reinforcement Learning (ICML 2025 · similarity 0.86)
- Prior-Guided Diffusion Planning for Offline Reinforcement Learning (NeurIPS 2025 · similarity 0.85)
- DiffTORI: Differentiable Trajectory Optimization for Deep Reinforcement and Imitation Learning (NeurIPS 2024 · similarity 0.85)
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.
