Research map / Reinforcement learning / Visual and representation RL
MoVie: Visual Model-Based Policy Adaptation for View Generalization
Sizhe Yang, Yanjie Ze, Huazhe Xu
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 STN blocks (0 -> 2) on low-level features (ablation) | MoVie · Cheetah run novel view | return: 254.42 -> 342.39 | Improved |
| add STN blocks (0 -> 2) on low-level features (ablation) | MoVie · Cheetah run shaking view | return: 317.66 -> 493.54 | Improved |
| use 3 STNs instead of 2 (ablation) | MoVie · Cheetah run novel view | return: 342.39 -> 309.25 | Hurt |
| use 1 STN instead of 2 (ablation) | MoVie · Cheetah run shaking view | return: 491.96 -> 493.54 | No clear effect |
| add STN to IDM (ablation) | IDM · xArm push all settings | success rate: 51 -> 59 | Improved |
| add STN to IDM (ablation) | IDM · Cup catch all settings | return: 757.84 -> 865.37 | Improved |
| add STN to IDM (ablation) | IDM · Finger spin all settings | return: 232.13 -> 301.86 | Improved |
| add STN to IDM (ablation) | IDM · Cheetah run all settings | return: 197.43 -> 209.43 | Improved |
| add STN to IDM (ablation) | IDM · xArm hammer shaking view | success rate: 35 -> 28 | Hurt |
| finetune DM end-to-end instead of freezing it (ablation) | MoVie · xArm push all settings | success rate: 64 -> 53 | Hurt |
| add STN to DM (MoVie vs DM only) (ablation) | DM · DMControl all settings | return: 520.68 -> 656.43 | Improved |
| test-time adaptation with IDM+STN instead of none | TD-MPC · DMControl all settings | return: 492.71 -> 426.76 | Hurt |
Closest papers by meaning
- Focus On What Matters: Separated Models For Visual-Based RL Generalization (NeurIPS 2024 · similarity 0.90)
- The Surprising Ineffectiveness of Pre-Trained Visual Representations for Model-Based Reinforcement Learning (NeurIPS 2024 · similarity 0.90)
- What Effects the Generalization in Visual Reinforcement Learning: Policy Consistency with Truncated Return Prediction (AAAI 2024 · similarity 0.89)
- VIGOR: Zero-Shot Visual Generalization via Latent-Space Consistency in Model-Based Reinforcement Learning (NeurIPS 2026 · similarity 0.89)
- A Simple Framework for Generalization in Visual RL under Dynamic Scene Perturbations (NeurIPS 2024 · similarity 0.88)
- Solving New Tasks by Adapting Internet Video Knowledge (ICLR 2025 · 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.
