# Disentangled World Models: Learning to Transfer Semantic Knowledge from Distracting Videos for Reinforcement Learning Source: [Disentangled World Models: Learning to Transfer Semantic Knowledge from Distracting Videos for Reinforcement Learning]() Collection index: [llms.txt]() Data updated: 2026-04-08T05:00:04.658Z Authors: Qi Wang, Zhipeng Zhang, Baao Xie, Xin Jin, Yunbo Wang, Shiyu Wang, Liaomo Zheng, Xiaokang Yang, Wenjun Zeng Published: 2025-03-11T13:50:22.000Z ## Abstract Training visual reinforcement learning (RL) in practical scenarios presents a significant challenge, $\\textit{i.e.,}$ RL agents suffer from low sample efficiency in environments with variations. While various approaches have attempted to alleviate this issue by disentangled representation learning, these methods usually start learning from scratch without prior knowledge of the world. This paper, in contrast, tries to learn and understand underlying semantic variations from distracting videos via offline-to-online latent distillation and flexible disentanglement constraints. To enable effective cross-domain semantic knowledge transfer, we introduce an interpretable model-based RL framework, dubbed Disentangled World Models (DisWM). Specifically, we pretrain the action-free video prediction model offline with disentanglement regularization to extract semantic knowledge from distracting videos. The disentanglement capability of the pretrained model is then transferred to the world model through latent distillation. For finetuning in the online environment, we exploit the knowledge from the pretrained model and introduce a disentanglement constraint to the world model. During the adaptation phase, the incorporation of actions and rewards from online environment interactions enriches the diversity of the data, which in turn strengthens the disentangled representation learning. Experimental results validate the superiority of our approach on various benchmarks. ## Recorded benchmark results A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. - [Transfer Learning on DMC Reacher Easy → Cheetah Run](): Transfer Learning; DMC Reacher Easy → Cheetah Run; Episode Return: 817. - [Transfer Learning on DMC Walker Walk](): Transfer Learning; DMC Walker Walk → Humanoid Walk; Episode Return: 147. - [Robotic Manipulation on DrawerWorld DrawerClose (texture variations)](): Robotic Manipulation; DrawerWorld DrawerClose, texture variations; Success Rate: 77. - [Robotic Manipulation on DrawerWorld DrawerOpen, texture variations](): Robotic Manipulation; DrawerWorld DrawerOpen, texture variations; Success Rate: 70. ## Original sources and implementation [Original paper]() [PDF]()