mlsanityData to 7 Oct 2026

DisDiff: Unsupervised Disentanglement of Diffusion Probabilistic Models

arXiv 2301.13721 ↗First posted 31 Jan 2023cs.CVcited by 30

On the map: Diffusion models → Diffusion foundations and sampling

Abstract (arXiv)

Targeting to understand the underlying explainable factors behind observations and modeling the conditional generation process on these factors, we connect disentangled representation learning to Diffusion Probabilistic Models (DPMs) to take advantage of the remarkable modeling ability of DPMs. We propose a new task, disentanglement of (DPMs): given a pre-trained DPM, without any annotations of the factors, the task is to automatically discover the inherent factors behind the observations and disentangle the gradient fields of DPM into sub-gradient fields, each conditioned on the representation of each discovered factor. With disentangled DPMs, those inherent factors can be automatically discovered, explicitly represented, and clearly injected into the diffusion process via the sub-gradient fields. To tackle this task, we devise an unsupervised approach named DisDiff, achieving disentangled representation learning in the framework of DPMs. Extensive experiments on synthetic and real-world datasets demonstrate the effectiveness of DisDiff.

Key facts. Cited by 30 papers in the map (data through 7 Oct 2026). Placed in Diffusion foundations and sampling, one of 14 directions in Diffusion models. Builds on 38 of its 50 references that are in the map. First posted on arXiv 31 Jan 2023.

Builds on

38 of 50 references.

Closest ancestors · 9

Same direction, newest first

  1. 2022Unsupervised Representation Learning from Pre-trained Diffusion Probabilistic Models26 Dec
  2. 2022Diffusion Models already have a Semantic Latent Space20 Oct
  3. 2022Classifier-Free Diffusion Guidance26 Jul
  4. 2021Diffusion Autoencoders: Toward a Meaningful and Decodable Representation30 Nov
  5. 2021ILVR: Conditioning Method for Denoising Diffusion Probabilistic Models6 Aug
  6. 2021Diffusion Models Beat GANs on Image Synthesis11 May
  7. 2020Score-Based Generative Modeling through Stochastic Differential Equations26 Nov
  8. 2020Denoising Diffusion Implicit Models6 Oct
  9. 2020Denoising Diffusion Probabilistic Models19 Jun

Other references · 29

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
  2. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,005
  3. 2022Photorealistic Text-to-Image Diffusion Models with Deep Language Understandingcited by 6,322
  4. 2022Generative Adversarial Networkscited by 5,603
  5. 2019Generative Modeling by Estimating Gradients of the Data Distributioncited by 4,067
  6. 2014Deep Learning Face Attributes in the Wildcited by 3,679
  7. 2021SDEdit: Guided Image Synthesis and Editing with Stochastic Differential Equationscited by 1,903
  8. 2022RePaint: Inpainting using Denoising Diffusion Probabilistic Modelscited by 1,419
  9. 2022Imagic: Text-Based Real Image Editing with Diffusion Modelscited by 1,172
  10. 2018Challenging Common Assumptions in the Unsupervised Learning of Disentangled Representationscited by 929
  11. 2021Blended Diffusion for Text-driven Editing of Natural Imagescited by 924
  12. 2021DiffusionCLIP: Text-Guided Diffusion Models for Robust Image Manipulationcited by 640
17 more
  1. 2018Disentangling by Factorisingcited by 614
  2. 2020GANSpace: Discovering Interpretable GAN Controlscited by 612
  3. 2022Improving Diffusion Models for Inverse Problems using Manifold Constraintscited by 501
  4. 2018Towards a Definition of Disentangled Representationscited by 294
  5. 2020Unsupervised Discovery of Interpretable Directions in the GAN Latent Spacecited by 288
  6. 2020CLUB: A Contrastive Log-ratio Upper Bound of Mutual Informationcited by 233
  7. 2021More Control for Free! Image Synthesis with Semantic Diffusion Guidancecited by 211
  8. 2021High-Fidelity GAN Inversion for Image Attribute Editingcited by 197
  9. 2022Uncovering the Disentanglement Capability in Text-to-Image Diffusion Modelscited by 116
  10. 2019On the Transfer of Inductive Bias from Simulation to the Real World: a New Disentanglement Datasetcited by 100
  11. 2020Adversarial score matching and improved sampling for image generationcited by 95
  12. 2021Unsupervised Learning of Compositional Energy Conceptscited by 71
  13. 2021Learning Disentangled Representation by Exploiting Pretrained Generative Models: A Contrastive Learning Viewcited by 34
  14. 2021Towards Building A Group-based Unsupervised Representation Disentanglement Frameworkcited by 22
  15. 2022NashAE: Disentangling Representations through Adversarial Covariance Minimizationcited by 13
  16. 2022An Empirical Study on Disentanglement of Negative-free Contrastive Learningcited by 12
  17. 2020Improving the Reconstruction of Disentangled Representation Learners via Multi-Stage Modelingcited by 5

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 3; 2024: 10; 2025: 11; 2026: 6.

30 papers.

Sort:

By area:

  1. 2024Diffusion Model with Cross Attention as an Inductive Bias for Disentanglementcited by 27
  2. 2023ChatFace: Chat-Guided Real Face Editing via Diffusion Latent Space Manipulationcited by 16
  3. 2024Learning Semantic Latent Directions for Accurate and Controllable Human Motion Predictioncited by 14
  4. 2023Flow Factorized Representation Learningcited by 11
  5. 2024Disentangling representations of retinal images with generative modelscited by 8
  6. 2024Diffusion Bridge AutoEncoders for Unsupervised Representation Learningcited by 6
  7. 2024Do Diffusion Models Learn Semantically Meaningful and Efficient Representations?cited by 4
  8. 2025Modular Machine Learning: An Indispensable Path towards New-Generation Large Language Modelscited by 4
  9. 2025FreeSliders: Training-Free, Modality-Agnostic Concept Sliders for Fine-Grained Diffusion Control in Images, Audio, and Videocited by 3
  10. 2024Unsupervised Representation Learning from Sparse Transformation Analysiscited by 2
  11. 2025DRESS: Disentangled Representation-based Self-Supervised Meta-Learning for Diverse Taskscited by 2
  12. 2025A Controllable Appearance Representation for Flexible Transfer and Editingcited by 2
18 more
  1. 2025Feature-Based Lie Group Transformer for Real-World Applicationscited by 2
  2. 2026Disentangled Representation Learning via Flow Matchingcited by 2
  3. 2026StrADiff: A Structured Source-Wise Adaptive Diffusion Framework for Linear and Nonlinear Blind Source Separationcited by 2
  4. 2024Transferring disentangled representations: bridging the gap between synthetic and real imagescited by 1
  5. 2025On Designing Diffusion Autoencoders for Efficient Generation and Representation Learningcited by 1
  6. 2025Disentangled Representation Learning via Modular Compositional Biascited by 1
  7. 2023Representation Disentaglement via Regularization by Causal Identificationcited by 0
  8. 2024Independence Constrained Disentangled Representation Learning from Epistemological Perspectivecited by 0
  9. 2024Disentanglement as Identifiable Pushforward Factorisationcited by 0
  10. 2024Progressive Monitoring of Generative Model Training Evolutioncited by 0
  11. 2025Enhancing Interpretability of Sparse Latent Representations with Class Informationcited by 0
  12. 2025Disentangled representations of microscopy imagescited by 0
  13. 2025CLOAK: Contrastive Guidance for Latent Diffusion-Based Data Obfuscationcited by 0
  14. 2025Disentangled representations via score-based variational autoencoderscited by 0
  15. 2026XFACTORS: Disentangled Information Bottleneck via Contrastive Supervisioncited by 0
  16. 2026CenSynCMB: Centre Maps and Physics-Guided Synthesis for Microbleed Detectioncited by 0
  17. 2026Decafs: Disentangled Conditional adversarial Flowscited by 0
  18. 2026Learning Late, Guiding Early: Timestep-Decoupled Semantic Guidance for Fair Face Generationcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2024Disentangling Disentangled Representations: Towards Improved Latent Units via Diffusion Models
  2. 2025Disentanglement in T-space for Faster and Distributed Training of Diffusion Models with Fewer Latent-states
  3. 2025DiffSDA: Unsupervised Diffusion Sequential Disentanglement Across Modalities
  4. 2024Closed-Loop Unsupervised Representation Disentanglement with β-VAE Distillation and Diffusion Probabilistic Feedback
  5. 2023Structural Pruning for Diffusion Models
  6. 2023Diff-Instruct: A Universal Approach for Transferring Knowledge From Pre-trained Diffusion Models
  7. 2025Can Diffusion Models Disentangle? A Theoretical Perspective
  8. 2024Isometric Representation Learning for Disentangled Latent Space of Diffusion Models