Research map / Reinforcement learning / Policy gradient and actor-critic
Deep Reinforcement Learning in a Handful of Trials using Probabilistic Dynamics Models
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 probabilistic neural networks with bootstrapped ensembles and trajectory sampling | model-based control with neural network dynamics · Half-cheetah | samples to reach model-free asymptotic performance: 8 and 125 times fewer samples than Soft Actor Critic and Proximal Policy Optimization respectively | Helped |
| include both aleatoric uncertainty and epistemic uncertainty in the dynamics model (ablation) | PETS · MuJoCo control benchmarks | final performance | Helped |
| add probabilistic output distributions to the ensemble model (ablation) | PETS · MuJoCo control benchmarks | final performance | Helped |
| add an ensemble to a probabilistic neural network (ablation) | PETS · MuJoCo control benchmarks | final performance | Helped |
| use trajectory sampling rather than deterministic expectation propagation (ablation) | PETS · MuJoCo control benchmarks | final performance | Helped |
| use Gaussian moment matching for state propagation (ablation) | PETS · Cartpole | final performance | Helped |
| use Gaussian moment matching for state propagation (ablation) | PETS · Half-cheetah | final performance | Hurt |
| use CEM instead of uniform random shooting (ablation) | MPC action selection · Half-cheetah | action optimization performance | Helped |
| bound probabilistic-network output variance by its training-data range (ablation) | probabilistic dynamics model · MuJoCo control benchmarks | control performance | Helped |
| model aleatoric uncertainty with probabilistic networks (ablation) | PETS dynamics model · stochastic systems with action noise | control performance | Helped |
Closest papers by meaning
- Sample Efficient Deep Reinforcement Learning via Uncertainty Estimation (ICLR 2022 · similarity 0.91)
- Deep Model-Based Reinforcement Learning via Estimated Uncertainty and Conservative Policy Optimization (AAAI 2020 · similarity 0.89)
- Sampling-Based Safe Reinforcement Learning (NeurIPS 2026 · similarity 0.88)
- Deterministic Uncertainty Propagation for Improved Model-Based Offline Reinforcement Learning (NeurIPS 2024 · similarity 0.88)
- SUMO: Search-Based Uncertainty Estimation for Model-Based Offline Reinforcement Learning (AAAI 2025 · similarity 0.88)
- Adversarial Diffusion for Robust Reinforcement Learning (NeurIPS 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.