mlsanityData to 7 Oct 2026

Self-Supervised Learning Disentangled Group Representation as Feature

arXiv 2110.15255 ↗First posted 28 Oct 2021cs.CV cs.LGcited by 47

Not on the map. Why?

Abstract (arXiv)

A good visual representation is an inference map from observations (images) to features (vectors) that faithfully reflects the hidden modularized generative factors (semantics). In this paper, we formulate the notion of "good" representation from a group-theoretic view using Higgins' definition of disentangled representation, and show that existing Self-Supervised Learning (SSL) only disentangles simple augmentation features such as rotation and colorization, thus unable to modularize the remaining semantics. To break the limitation, we propose an iterative SSL algorithm: Iterative Partition-based Invariant Risk Minimization (IP-IRM), which successfully grounds the abstract semantics and the group acting on them into concrete contrastive learning. At each iteration, IP-IRM first partitions the training samples into two subsets that correspond to an entangled group element. Then, it minimizes a subset-invariant contrastive loss, where the invariance guarantees to disentangle the group element. We prove that IP-IRM converges to a fully disentangled representation and show its effectiveness on various benchmarks. Codes are available at https://github.com/Wangt-CN/IP-IRM.

Key facts. Cited by 47 papers in the map (data through 7 Oct 2026). Builds on 64 of its 106 references that are in the map. First posted on arXiv 28 Oct 2021.

Builds on

64 of 106 references.

References · 64

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
  2. 2020Language Models are Few-Shot Learnerscited by 32,199
  3. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,267
  4. 2020A Simple Framework for Contrastive Learning of Visual Representationscited by 11,979
  5. 2018Representation Learning with Contrastive Predictive Codingcited by 7,662
  6. 2019Momentum Contrast for Unsupervised Visual Representation Learningcited by 7,482
  7. 2017mixup: Beyond Empirical Risk Minimizationcited by 5,155
  8. 2020Bootstrap your own latent: A new approach to self-supervised Learningcited by 4,712
  9. 2013Auto-Encoding Variational Bayescited by 3,980
  10. 2015Domain-Adversarial Training of Neural Networkscited by 3,023
  11. 2020Exploring Simple Siamese Representation Learningcited by 2,853
  12. 2020Unsupervised Learning of Visual Features by Contrasting Cluster Assignmentscited by 2,845
38 more
  1. 2015Learning Deep Features for Discriminative Localizationcited by 2,430
  2. 2020Improved Baselines with Momentum Contrastive Learningcited by 2,282
  3. 2013Describing Textures in the Wildcited by 2,110
  4. 2018Unsupervised Feature Learning via Non-Parametric Instance-level Discriminationcited by 1,838
  5. 2019Invariant Risk Minimizationcited by 1,832
  6. 2013Fine-Grained Visual Classification of Aircraftcited by 1,701
  7. 2018Unsupervised Representation Learning by Predicting Image Rotationscited by 1,586
  8. 2016Unsupervised Learning of Visual Representations by Solving Jigsaw Puzzlescited by 1,379
  9. 2019Contrastive Multiview Codingcited by 1,297
  10. 2018Learning deep representations by mutual information estimation and maximizationcited by 1,265
  11. 2015Unsupervised Visual Representation Learning by Context Predictioncited by 1,062
  12. 2014Striving for Simplicity: The All Convolutional Netcited by 983
  13. 2019Adversarial Examples Are Not Bugs, They Are Featurescited by 950
  14. 2018Challenging Common Assumptions in the Unsupervised Learning of Disentangled Representationscited by 929
  15. 2016InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Netscited by 858
  16. 2020What Makes for Good Views for Contrastive Learning?cited by 843
  17. 2019Self-Supervised Learning of Pretext-Invariant Representationscited by 806
  18. 2019Learning Representations by Maximizing Mutual Information Across Viewscited by 733
  19. 2019Data-Efficient Image Recognition with Contrastive Predictive Codingcited by 672
  20. 2020Prototypical Contrastive Learning of Unsupervised Representationscited by 636
  21. 2020Contrastive Learning with Hard Negative Samplescited by 619
  22. 2018Disentangling by Factorisingcited by 614
  23. 2018Do Better ImageNet Models Transfer Better?cited by 560
  24. 2018Spherical CNNscited by 452
  25. 2020Hard Negative Mixing for Contrastive Learningcited by 399
  26. 2020Debiased Contrastive Learningcited by 354
  27. 2018Understanding disentangling in β-VAEcited by 332
  28. 2018Towards a Definition of Disentangled Representationscited by 294
  29. 2021Understanding self-supervised Learning Dynamics without Contrastive Pairscited by 254
  30. 2019Unsupervised Embedding Learning via Invariant and Spreading Instance Featurecited by 231
  31. 2020What Should Not Be Contrastive in Contrastive Learningcited by 224
  32. 2020How Well Do Self-Supervised Models Transfer?cited by 213
  33. 2018Isolating Sources of Disentanglement in Variational Autoencoderscited by 167
  34. 2015A Large-Scale Car Dataset for Fine-Grained Categorization and Verificationcited by 129
  35. 2018Learning to Decompose and Disentangle Representations for Video Predictioncited by 120
  36. 2019Deep Scale-spaces: Equivariance Over Scalecited by 113
  37. 2020Interventional Few-Shot Learningcited by 112
  38. 2019Scale-Equivariant Steerable Networkscited by 110

50 most cited of 64 All 64 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2022: 15; 2023: 17; 2024: 10; 2025: 4; 2026: 1.

47 papers.

Sort:

By area:

  1. 2022Discovering Invariant Rationales for Graph Neural Networkscited by 186
  2. 2023Change is Hard: A Closer Look at Subpopulation Shiftcited by 119
  3. 2022Disentangled Representation Learningcited by 111
  4. 2023Equivariant Similarity for Vision-Language Foundation Modelscited by 51
  5. 2022Discover and Mitigate Unknown Biases with Debiasing Alternate Networkscited by 47
  6. 2024Exploring Diffusion Time-steps for Unsupervised Representation Learningcited by 36
  7. 2022Identifiability of Label Noise Transition Matrixcited by 34
  8. 2022Invariant Feature Learning for Generalized Long-Tailed Classificationcited by 34
  9. 2024LTGC: Long-tail Recognition via Leveraging LLMs-driven Generated Contentcited by 33
  10. 2023Text-Video Retrieval with Disentangled Conceptualization and Set-to-Set Alignmentcited by 31
  11. 2025Selftok: Discrete Visual Tokens of Autoregression, by Diffusion, and for Reasoningcited by 31
  12. 2022Symmetry-Based Representations for Artificial and Biological General Intelligencecited by 26
35 more
  1. 2022Identifying Hard Noise in Long-Tailed Sample Distributioncited by 19
  2. 2022Respecting Transfer Gap in Knowledge Distillationcited by 18
  3. 2022Equivariance and Invariance Inductive Bias for Learning from Insufficient Datacited by 14
  4. 2025MoAlign: Motion-Centric Representation Alignment for Video Diffusion Modelscited by 13
  5. 2023Make the U in UDA Matter: Invariant Consistency Learning for Unsupervised Domain Adaptationcited by 12
  6. 2023Simple Disentanglement of Style and Content in Visual Representationscited by 10
  7. 2023ArCL: Enhancing Contrastive Learning with Augmentation-Robust Representationscited by 10
  8. 2023Sample and Predict Your Latent: Modality-free Sequential Disentanglement via Contrastive Estimationcited by 10
  9. 2023Environment-Invariant Curriculum Relation Learning for Fine-Grained Scene Graph Generationcited by 9
  10. 2024Contrastive representations of high-dimensional, structured treatmentscited by 7
  11. 2022Class Is Invariant to Context and Vice Versa: On Learning Invariance for Out-Of-Distribution Generalizationcited by 6
  12. 20233D Generative Model Latent Disentanglement via Local Eigenprojectioncited by 6
  13. 2023Invariant Feature Regularization for Fair Face Recognitioncited by 6
  14. 2023Learning Geometry-aware Representations by Sketchingcited by 5
  15. 2023Distortion-Disentangled Contrastive Learningcited by 4
  16. 2023Quantitatively Measuring and Contrastively Exploring Heterogeneity for Domain Generalizationcited by 4
  17. 2023A Category-theoretical Meta-analysis of Definitions of Disentanglementcited by 3
  18. 2024Disentangling Masked Autoencoders for Unsupervised Domain Generalizationcited by 3
  19. 2024Latent Space Disentanglement in Diffusion Transformers Enables Precise Zero-shot Semantic Editingcited by 3
  20. 2025Identifying Metric Structures of Deep Latent Variable Modelscited by 3
  21. 2022Conditional Generative Models for Simulation of EMG During Naturalistic Movementscited by 2
  22. 2023Compositionally Equivariant Representation Learningcited by 2
  23. 2023Algebras of actions in an agent's representations of the worldcited by 2
  24. 2024Views Can Be Deceiving: Improved SSL Through Feature Space Augmentationcited by 2
  25. 2022Equivariant Disentangled Transformation for Domain Generalization under Combination Shiftcited by 1
  26. 2024Boosting Few-Shot Learning with Disentangled Self-Supervised Learning and Meta-Learning for Medical Image Classificationcited by 1
  27. 2024Latent Space Disentanglement in Diffusion Transformers Enables Zero-shot Fine-grained Semantic Editingcited by 1
  28. 2024Exploiting Supervised Poison Vulnerability to Strengthen Self-Supervised Defensecited by 1
  29. 2025STM3: Mixture of Multiscale Mamba for Long-Term Spatio-Temporal Time-Series Predictioncited by 1
  30. 2022Invariance-adapted decomposition and Lasso-type contrastive learningcited by 0
  31. 2022Commutativity and Disentanglement from the Manifold Perspectivecited by 0
  32. 2022A Survey on Causal Representation Learning and Future Work for Medical Image Analysiscited by 0
  33. 2023Domain Invariant Learning for Gaussian Processes and Bayesian Explorationcited by 0
  34. 2024Independence Constrained Disentangled Representation Learning from Epistemological Perspectivecited by 0
  35. 2026Contrastive-Augmented Flow Matching for Style-Content Disentanglementcited by 0