mlsanityData to 7 Oct 2026

Weakly-Supervised Disentanglement Without Compromises

arXiv 2002.02886 ↗First posted 7 Feb 2020cs.LG stat.MLcited by 237

Not on the map. Why?

Abstract (arXiv)

Intelligent agents should be able to learn useful representations by observing changes in their environment. We model such observations as pairs of non-i.i.d. images sharing at least one of the underlying factors of variation. First, we theoretically show that only knowing how many factors have changed, but not which ones, is sufficient to learn disentangled representations. Second, we provide practical algorithms that learn disentangled representations from pairs of images without requiring annotation of groups, individual factors, or the number of factors that have changed. Third, we perform a large-scale empirical study and show that such pairs of observations are sufficient to reliably learn disentangled representations on several benchmark data sets. Finally, we evaluate our learned representations and find that they are simultaneously useful on a diverse suite of tasks, including generalization under covariate shifts, fairness, and abstract reasoning. Overall, our results demonstrate that weak supervision enables learning of useful disentangled representations in realistic scenarios.

Key facts. Cited by 237 papers in the map (data through 7 Oct 2026). First posted on arXiv 7 Feb 2020.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 42; 2022: 57; 2023: 50; 2024: 46; 2025: 30; 2026: 12.

237 papers.

Sort:

By area:

  1. 2021Self-Supervised Learning with Data Augmentations Provably Isolates Content from Stylecited by 308
  2. 2021Contrastive Learning Inverts the Data Generating Processcited by 220
  3. 2021Towards Causal Representation Learningcited by 207
  4. 2021A Fine-Grained Analysis on Distribution Shiftcited by 171
  5. 2021Disentanglement via Mechanism Sparsity Regularization: A New Principle for Nonlinear ICAcited by 158
  6. 2022Weakly supervised causal representation learningcited by 140
  7. 2022Interventional Causal Representation Learningcited by 140
  8. 2022Causality Inspired Representation Learning for Domain Generalizationcited by 136
  9. 2022CITRIS: Causal Identifiability from Temporal Intervened Sequencescited by 132
  10. 2022Disentangled Representation Learningcited by 111
  11. 2023Nonparametric Identifiability of Causal Representations from Unknown Interventionscited by 105
  12. 2021Counterfactual Invariance to Spurious Correlations: Why and How to Pass Stress Testscited by 77
38 more
  1. 2021Desiderata for Representation Learning: A Causal Perspectivecited by 76
  2. 2022GlanceNets: Interpretabile, Leak-proof Concept-based Modelscited by 73
  3. 2023Multi-View Causal Representation Learning with Partial Observabilitycited by 72
  4. 2021Unsupervised Learning of Compositional Energy Conceptscited by 71
  5. 2021Causal Influence Detection for Improving Efficiency in Reinforcement Learningcited by 70
  6. 2022Identifiability of deep generative models without auxiliary informationcited by 70
  7. 2021Disentangling Identifiable Features from Noisy Data with Structured Nonlinear ICAcited by 64
  8. 2022Weakly Supervised Representation Learning with Sparse Perturbationscited by 64
  9. 2022Active Learning by Feature Mixingcited by 63
  10. 2023Not All Neuro-Symbolic Concepts Are Created Equal: Analysis and Mitigation of Reasoning Shortcutscited by 63
  11. 2021Visual Representation Learning Does Not Generalize Strongly Within the Same Domaincited by 60
  12. 2023Identifiability Results for Multimodal Contrastive Learningcited by 58
  13. 2025Projecting Assumptions: The Duality Between Sparse Autoencoders and Concept Geometrycited by 58
  14. 2021Beyond Trivial Counterfactual Explanations with Diverse Valuable Explanationscited by 56
  15. 2023Causal Component Analysiscited by 56
  16. 2023Additive Decoders for Latent Variables Identification and Cartesian-Product Extrapolationcited by 56
  17. 2021Identifiable Deep Generative Models via Sparse Decodingcited by 54
  18. 2022Causal Discovery in Heterogeneous Environments Under the Sparse Mechanism Shift Hypothesiscited by 54
  19. 2022Assaying Out-Of-Distribution Generalization in Transfer Learningcited by 53
  20. 2022Causal Representation Learning for Instantaneous and Temporal Effects in Interactive Systemscited by 50
  21. 2022Mechanistic Mode Connectivitycited by 50
  22. 2023Score-based Causal Representation Learning with Interventionscited by 49
  23. 2022Synergies between Disentanglement and Sparsity: Generalization and Identifiability in Multi-Task Learningcited by 44
  24. 2023Causal Discovery from Temporal Data: An Overview and New Perspectivescited by 43
  25. 2023Provable Compositional Generalization for Object-Centric Learningcited by 41
  26. 2023Disentanglement via Latent Quantizationcited by 40
  27. 2023Understanding Masked Autoencoders via Hierarchical Latent Variable Modelscited by 39
  28. 2023BISCUIT: Causal Representation Learning from Binary Interactionscited by 38
  29. 2021Understanding Latent Correlation-Based Multiview Learning and Self-Supervision: An Identifiability Perspectivecited by 37
  30. 2021On the Limitations of Multimodal VAEscited by 37
  31. 2022Invariant and Transportable Representations for Anti-Causal Domain Shiftscited by 37
  32. 2022Disentanglement of Correlated Factors via Hausdorff Factorized Supportcited by 37
  33. 2021Contrastively Disentangled Sequential Variational Autoencodercited by 34
  34. 2022Modeling the Data-Generating Process is Necessary for Out-of-Distribution Generalizationcited by 34
  35. 2023Measuring axiomatic soundness of counterfactual image modelscited by 34
  36. 2023General Identifiability and Achievability for Causal Representation Learningcited by 34
  37. 2022Indeterminacy in Generative Models: Characterization and Strong Identifiabilitycited by 33
  38. 2022Identifying Weight-Variant Latent Causal Modelscited by 33

50 most cited of 237 All 237 (JSON)