Research map / Reinforcement learning / Policy gradient and actor-critic
Value Gradient weighted Model-Based Reinforcement Learning
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 VaGraM instead of MLE for model learning with reduced model capacity | MBPO on Hopper · Hopper | return | Helped |
| Use VaGraM instead of MLE with distracting state dimensions | MBPO on Hopper · Hopper with distracting dimensions | return | Helped |
| Use VaGraM instead of MLE with sufficiently large models | MBPO on Hopper · Hopper | return | No clear effect |
| Use VaGraM instead of MBPO's MLE model loss | MBPO · HalfCheetah | return | No clear effect |
| Use VaGraM instead of MBPO's MLE model loss | MBPO · Walker | return | No clear effect |
| Use VaGraM instead of MBPO's MLE model loss | MBPO · Ant | return | Hurt |
| Add normalization to VaGraM (ablation) | VaGraM on Ant · Ant | return | Helped |
| Add normalization to VaGraM (ablation) | VaGraM on HalfCheetah · HalfCheetah | return | Hurt |
| Add normalization to VaGraM (ablation) | VaGraM on Walker2d · Walker2d | return | Hurt |
| Add an MSE term to IterVAML with λ = 0.01 (ablation) | IterVAML · Pendulum swingup | return | Helped |
| Add an MSE term to IterVAML with λ = 0.01 (ablation) | IterVAML · Cartpole balancing | return | Hurt |
| Replace VaGraM with simple inverse-value-weighted MSE (ablation) | Model learning for continuous control · Pendulum | return | Hurt |
Closest papers by meaning
- Iterative Value-Aware Model Learning (NeurIPS 2018 · similarity 0.90)
- Calibrated Value-Aware Model Learning with Probabilistic Environment Models (ICML 2025 · similarity 0.90)
- The Edge-of-Reach Problem in Offline Model-Based Reinforcement Learning (NeurIPS 2024 · similarity 0.87)
- Your Language Model is Its Own Critic: Reinforcement Learning with Value Estimation from Actor’s Internal States (NeurIPS 2026 · similarity 0.87)
- Efficient Off-Policy Learning for High-Dimensional Action Spaces (ICLR 2025 · similarity 0.87)
- Policy-shaped prediction: avoiding distractions in model-based reinforcement learning (NeurIPS 2024 · 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.