mlsanityData to 14 Oct 2026

Are Disentangled Representations Helpful for Abstract Visual Reasoning?

arXiv 1905.12506 ↗First posted 29 May 2019cs.LG cs.CV cs.NEcited by 100

Not on the map. Why?

Abstract (arXiv)

A disentangled representation encodes information about the salient factors of variation in the data independently. Although it is often argued that this representational format is useful in learning to solve many real-world down-stream tasks, there is little empirical evidence that supports this claim. In this paper, we conduct a large-scale study that investigates whether disentangled representations are more suitable for abstract reasoning tasks. Using two new tasks similar to Raven's Progressive Matrices, we evaluate the usefulness of the representations learned by 360 state-of-the-art unsupervised disentanglement models. Based on these representations, we train 3600 abstract reasoning models and observe that disentangled representations do in fact lead to better down-stream performance. In particular, they enable quicker learning using fewer samples.

Key facts. Cited by 100 papers in the map (data through 14 Oct 2026). First posted on arXiv 29 May 2019.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2021: 30; 2022: 23; 2023: 26; 2024: 14; 2025: 5; 2026: 2.

100 papers.

Sort:

By area:

  1. 2021Towards Causal Representation Learningcited by 209
  2. 2022Disentangled Representation Learningcited by 112
  3. 2023Compositional Abilities Emerge Multiplicatively: Exploring Diffusion Models on a Synthetic Taskcited by 93
  4. 2021Generalization and Robustness Implications in Object-Centric Learningcited by 81
  5. 2022GlanceNets: Interpretabile, Leak-proof Concept-based Modelscited by 73
  6. 2023Compositional Generalization from First Principlescited by 64
  7. 2022Mechanistic Mode Connectivitycited by 50
  8. 2021Self-Supervised Learning Disentangled Group Representation as Featurecited by 47
  9. 2022Synergies between Disentanglement and Sparsity: Generalization and Identifiability in Multi-Task Learningcited by 44
  10. 2021CDPAM: Contrastive learning for perceptual audio similaritycited by 42
  11. 2021A Novel Estimator of Mutual Information for Learning to Disentangle Textual Representationscited by 37
  12. 2022Deep Learning Methods for Abstract Visual Reasoning: A Survey on Raven's Progressive Matricescited by 37
38 more
  1. 2021Inspect, Understand, Overcome: A Survey of Practical Methods for AI Safetycited by 35
  2. 2024Emergence of Hidden Capabilities: Exploring Learning Dynamics in Concept Spacecited by 35
  3. 2023Text-Video Retrieval with Disentangled Conceptualization and Set-to-Set Alignmentcited by 32
  4. 2021An Image is Worth More Than a Thousand Words: Towards Disentanglement in the Wildcited by 30
  5. 2021Where and What? Examining Interpretable Disentangled Representationscited by 27
  6. 2022Symmetry-Based Representations for Artificial and Biological General Intelligencecited by 26
  7. 2024Learned feature representations are biased by complexity, learning order, position, and morecited by 23
  8. 2021Towards Building A Group-based Unsupervised Representation Disentanglement Frameworkcited by 22
  9. 2023Causal Triplet: An Open Challenge for Intervention-centric Causal Representation Learningcited by 22
  10. 2022Learning Robust Representation for Joint Grading of Ophthalmic Diseases via Adaptive Curriculum and Feature Disentanglementcited by 20
  11. 2024When does compositional structure yield compositional generalization? A kernel theorycited by 20
  12. 2025Robust Disentangled Counterfactual Learning for Physical Audiovisual Commonsense Reasoningcited by 20
  13. 2023Learning to reason over visual objectscited by 19
  14. 2021Drop, Swap, and Generate: A Self-Supervised Approach for Generating Neural Activitycited by 18
  15. 2021Scaling-up Disentanglement for Image Translationcited by 15
  16. 2022A Contrastive Objective for Learning Disentangled Representationscited by 14
  17. 2021Disentangled Representations from Non-Disentangled Modelscited by 13
  18. 2022Inductive Biases for Object-Centric Representations in the Presence of Complex Texturescited by 13
  19. 2023Towards a Unified Framework of Contrastive Learning for Disentangled Representationscited by 13
  20. 2024Fair Graph Representation Learning via Sensitive Attribute Disentanglementcited by 13
  21. 2021The Role of Pretrained Representations for the OOD Generalization of Reinforcement Learning Agentscited by 12
  22. 2022An Empirical Study on Disentanglement of Negative-free Contrastive Learningcited by 12
  23. 2022Comparing the latent space of generative modelscited by 12
  24. 2022Clustering units in neural networks: upstream vs downstream informationcited by 11
  25. 2022Lost in Latent Space: Disentangled Models and the Challenge of Combinatorial Generalisationcited by 11
  26. 2021Exploring the Latent Space of Autoencoders with Interventional Assayscited by 10
  27. 2021Raven's Progressive Matrices Completion with Latent Gaussian Process Priorscited by 9
  28. 2022Learning Disentangled Behaviour Patterns for Wearable-based Human Activity Recognitioncited by 9
  29. 2023Abstract Visual Reasoning: An Algebraic Approach for Solving Raven's Progressive Matricescited by 9
  30. 2021Disentanglement and Generalization Under Correlation Shiftscited by 8
  31. 2023DAVA: Disentangling Adversarial Variational Autoencodercited by 8
  32. 2021Be More Active! Understanding the Differences between Mean and Sampled Representations of Variational Autoencoderscited by 7
  33. 2022Multi-Viewpoint and Multi-Evaluation with Felicitous Inductive Bias Boost Machine Abstract Reasoning Abilitycited by 7
  34. 2025GAITGen: Disentangled Motion-Pathology Impaired Gait Generative Model -- Bringing Motion Generation to the Clinical Domaincited by 7
  35. 2022Disentangling Shape and Pose for Object-Centric Deep Active Inference Modelscited by 6
  36. 2023A Survey of Methods, Challenges and Perspectives in Causalitycited by 6
  37. 2023A multimodal dynamical variational autoencoder for audiovisual speech representation learningcited by 6
  38. 2023Abstracting Concept-Changing Rules for Solving Raven's Progressive Matrix Problemscited by 6

50 most cited of 100 All 100 (JSON)