Representation Disentanglement

Follow this topic Ask about this topic

Representation disentanglement aims to make different properties—such as content and style, identity and expression, or task-relevant and domain-specific signals—independently accessible in a model’s features. Common approaches include variational objectives, orthogonality and independence constraints, adversarial or contrastive learning, sparse autoencoders, and architectures with separate feature streams or subspaces. The methods are used in settings such as controllable image and speech generation, domain generalization, privacy and fairness, and mechanistic interpretability, with evaluation through probes, leakage tests, reconstruction quality, and downstream performance. A central challenge is that factors may be correlated or not uniquely identifiable, while suppressing unwanted information can also reduce useful task information or reconstruction fidelity.

Latest papers

98 papers9 papers in the last four weeks, up 29% on the four weeks before. 0.1% of all new papers.

Automated Disentangling Analysis of Skin Colour for Lesion Images
Feb 22, 2026Wenbo Yang, Eman Rezk, Walaa M. Moursi +1Disentangled Representation Learning Representation Disentanglement
University of Waterloo, Waterloo, ON, Canada, N2L 3G1
CRAFT-LoRA: Content-Style Personalization via Rank-Constrained Adaptation and Training-Free Fusion
Feb 21, 2026Yu Li, Yujun Cai, Chi ZhangImage Generation T2I Diffusion Models
AGI Lab, Westlake University · George Washington University · The University of Queensland
Structure-Level Disentangled Diffusion for Few-Shot Chinese Font Generation
Feb 21, 2026Jie Li, Suorong Yang, Jian Zhao +1Few-Shot Learning Conditional Diffusion Models
State Key Laboratory for Novel Software Technology, Nanjing University, China · School of Artificial Intelligence, Nanjing University, China · Department of Computer Science and Technology, Nanjing University, China +1
Recommender system in X inadvertently profiles ideological positions of users
Feb 2, 2026Paul Bouchaud, Pedro RamaciottiRecommender Systems Representation Disentanglement
Complex Systems Institute of Paris Ile-de-France CNRS, Paris, France. · m´edialab, Sciences Po, Paris, France. · Learning Planet Institute, CY Cergy Paris University, Paris, France.
Navigating Ideation Space: Decomposed Conceptual Representations for Positioning Scientific Ideas
Jan 13, 2026Minqian Liu, Yuexi Shen, Dawei Zhou +1Scientific Idea Evaluation Scientific Document Retrieval
Virginia Tech · University of California, Santa Barbara · University of California, Davis
Learning Domain- and Class-Disentangled Prototypes for Domain-Generalized EEG Emotion Recognition
Sep 1, 2025Guangli Li, Canbiao Wu, Zhehao Zhou +3Cross-Subject EEG Decoding Prototype-Based Classification
School of Biological Science and Medical Engineering, Hunan University of Technology, Zhuzhou 412008, China · School of Biomedical Engineering, Health Science Center, Shenzhen University, Shenzhen 518060, China · Guangdong Provincial Key Laboratory of Biomedical Measurements and Ultrasound Imaging, Shenzhen 518060, China +1
Sparse but Wrong: Incorrect L0 Leads to Incorrect Features in Sparse Autoencoders
Aug 22, 2025David Chanin, Adrià Garriga-AlonsoActivation Sparsity LLM Interpretability
University College London · MATS Research
Post-hoc Disentanglement of Textual and Acoustic Features in Pre-trained Speech Encoders
Oct 3, 2024Hosein Mohebbi, Grzegorz Chrupała, Willem Zuidema +2Representation Disentanglement Disentangled Representation Learning
CSAI, Tilburg University · ILLC, University of Amsterdam · ILCC, University of Edinburgh

Feb 22, 2026J/K move · Enter open · S save

Automated Disentangling Analysis of Skin Colour for Lesion Images

Wenbo Yang, Eman Rezk, Walaa M. Moursi, Zhou Wang

University of Waterloo, Waterloo, ON, Canada, N2L 3G1

Disentangled Representation Learning Representation Disentanglement Dermatology Counterfactual Data Augmentation

Open paper AskPDF arXiv

Machine-learning models applied to skin images often have degraded performance when the skin colour captured in images (SCCI) differs between training and deployment. These discrepancies arise from a combination of entangled environmental factors (e.g., illumination, camera settings) and intrinsic factors (e.g., skin tone) that cannot be accurately described by a single "skin tone" scalar -- a simplification commonly adopted by prior work. To mitigate such colour mismatches, we propose a skin-colour disentangling framework that adapts disentanglement-by-compression to learn a structured, manipulable latent space for SCCI from unlabelled dermatology images. To prevent information leakage that hinders proper learning of dark colour features, we introduce a randomized, mostly monotonic decolourization mapping. To suppress unintended colour shifts of localized patterns (e.g., ink marks, scars) during colour manipulation, we further propose a geometry-aligned post-processing step. Together, these components enable faithful counterfactual editing and answering an essential question: "What would this skin condition look like under a different SCCI?", as well as direct colour transfer between images and controlled traversal along physically meaningful directions (e.g., blood perfusion, camera white balance), enabling educational visualization of skin conditions under varying SCCI. We demonstrate that dataset-level augmentation and colour normalization based on our framework achieve competitive lesion classification performance. Ultimately, our work promotes equitable diagnosis through creating diverse training datasets that include different skin tones and image-capturing conditions.