Research map / Reinforcement learning / Visual and representation RL
Disentangled World Models: Learning to Transfer Semantic Knowledge from Distracting Videos for Reinforcement Learning
Qi Wang, Zhipeng Zhang, Baao Xie, Xin Jin, Yunbo Wang, Shiyu Wang, Liaomo Zheng, Xiaokang Yang, Wenjun Zeng
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 cross-domain latent distillation (ablation) | DisWM · DMC Walker Walk -> Cheetah Run | episode return | Improved |
| add disentanglement constraints in pretraining and finetuning (ablation) | DisWM · DMC Walker Walk -> Cheetah Run | episode return | Improved |
| increase disentanglement beta too much (ablation) | DisWM · DMC Cheetah Run -> Walker Walk | episode return | Hurt |
| decrease disentanglement beta too much (ablation) | DisWM · DMC Cheetah Run -> Walker Walk | episode return | Hurt |
| raise latent distillation weight eta too high (ablation) | DisWM · DMC | episode return | Hurt |
| use DisWM pretraining on varied source video domains versus no pretraining (ablation) | DisWM · DMC Cartpole Swingup | episode return | Improved |
| shrink latent dimension of beta-VAE too small (ablation) | DisWM · DMC Cheetah Run -> Walker Walk | episode return | Hurt |
| use DisWM instead of ContextWM | ContextWM · DrawerWorld DrawerClose | success rate (%): 37 -> 77 | Improved |
| use DisWM instead of DreamerV3 | DreamerV3 · DMC Reacher Easy -> Cheetah Run | episode return: 662 -> 817 | Improved |
Closest papers by meaning
- VideoWorld: Exploring Knowledge Learning from Unlabeled Videos (CVPR 2025 · similarity 0.87)
- Disentangling Representations through Multi-task Learning (ICLR 2025 · similarity 0.87)
- Zero Shot Generalization of Vision-Based RL Without Data Augmentation (ICML 2025 · similarity 0.86)
- Zero-shot World Models via Search in Memory (NeurIPS 2025 · similarity 0.85)
- Towards Principled Representation Learning from Videos for Reinforcement Learning (ICLR 2024 · similarity 0.85)
- Find, Fix, Reason: Context Repair for Video Reasoning (ICML 2026 · 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.
