Research map / Reinforcement learning / Policy gradient and actor-critic
Model-based Reinforcement Learning with Scalable Composite Policy Gradient Estimators
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 |
|---|---|---|---|
| Replace RP gradient with TPX composite estimator (IVW at propagation node, tensor-valued k) in Dreamer | Dreamer policy gradient · DMC visual, H=15 and H=60 | normalized expected SNR / performance profile | Helped |
| Replace RP gradient with TPX composite estimator in Dreamer | Dreamer · DMC visual, H=60 | learning performance | Helped |
| Use tensor-valued k (per-dimension weights) instead of scalar k in TPX (ablation) | Dreamer TPX · DMC visual gradient accuracy ablation | normalized expected SNR | Helped |
| Include covariance terms in TPX IVW weight estimate (ablation) | Dreamer TPX · DMC visual gradient accuracy ablation | normalized expected SNR | No clear effect |
| Use IVW at latent node (TPX) instead of parameter node (TP) | PIPPS cart-pole · cart-pole replication | computation time per trajectory: 1.31 vs 1.91 for TP | Helped |
| Replace RP with TPX in Dreamer | Dreamer · DMC visual, H=15/H=60 | computation time per iteration: 1.14x at H=15, 1.4x at H=60 vs Dreamer | Hurt |
| Use LR gradient estimator instead of RP in Dreamer | Dreamer policy gradient · DMC visual, H=15 | learning performance | Hurt |
| Use TP-style composite at longer horizon vs RP in Dreamer | Dreamer · DMC visual, H=60 | learning performance | Helped |
Closest papers by meaning
- Total stochastic gradient algorithms and applications in reinforcement learning (NeurIPS 2018 · similarity 0.89)
- Parallelizing Model-based Reinforcement Learning Over the Sequence Length (NeurIPS 2024 · similarity 0.87)
- GenPO: Generative Diffusion Models Meet On-Policy Reinforcement Learning (NeurIPS 2025 · similarity 0.87)
- Diffusion-based Reinforcement Learning via Q-weighted Variational Policy Optimization (NeurIPS 2024 · similarity 0.86)
- PWM: Policy Learning with Multi-Task World Models (ICLR 2025 · similarity 0.86)
- Natural Policy Gradient as Doubly Smoothed Policy Iteration: A Bellman-Operator Framework (NeurIPS 2026 · similarity 0.86)
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.