# A Sober Look at the Unsupervised Learning of Disentangled Representations and their Evaluation

arXiv: https://arxiv.org/abs/2010.14766 · HTML: https://mlsanity.com/paper/2010.14766/

## Key facts

- Cited by 52 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 27 Oct 2020.

## Built on by (most cited first, 30 of 52)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [Toward Transparent AI: A Survey on Interpreting the Inner Structures of Deep Neural Networks](https://mlsanity.com/paper/2207.13243/) (2022)
- [Generalization and Robustness Implications in Object-Centric Learning](https://mlsanity.com/paper/2107.00637/) (2021)
- [Vision Models Are More Robust And Fair When Pretrained On Uncurated Images Without Supervision](https://mlsanity.com/paper/2202.08360/) (2022)
- [Is a Caption Worth a Thousand Images? A Controlled Study for Representation Learning](https://mlsanity.com/paper/2207.07635/) (2022)
- [Disentanglement of Correlated Factors via Hausdorff Factorized Support](https://mlsanity.com/paper/2210.07347/) (2022)
- [Measuring axiomatic soundness of counterfactual image models](https://mlsanity.com/paper/2303.01274/) (2023)
- [Leveraging sparse and shared feature activations for disentangled representation learning](https://mlsanity.com/paper/2304.07939/) (2023)
- [DCI-ES: An Extended Disentanglement Framework with Connections to Identifiability](https://mlsanity.com/paper/2210.00364/) (2022)
- [A Structural Causal Model for MR Images of Multiple Sclerosis](https://mlsanity.com/paper/2103.03158/) (2021)
- [Speaker Adaptation Using Spectro-Temporal Deep Features for Dysarthric and Elderly Speech Recognition](https://mlsanity.com/paper/2202.10290/) (2022)
- [On the Identification of Temporally Causal Representation with Instantaneous Dependence](https://mlsanity.com/paper/2405.15325/) (2024)
- [Learning Causal Representations from General Environments: Identifiability and Intrinsic Ambiguity](https://mlsanity.com/paper/2311.12267/) (2023)
- [Disentangled Representations from Non-Disentangled Models](https://mlsanity.com/paper/2102.06204/) (2021)
- [Beta-VAE Reproducibility: Challenges and Extensions](https://mlsanity.com/paper/2112.14278/) (2021)
- [Diffusion Counterfactual Generation with Semantic Abduction](https://mlsanity.com/paper/2506.07883/) (2025)
- [Review of Disentanglement Approaches for Medical Applications -- Towards Solving the Gordian Knot of Generative Models in Healthcare](https://mlsanity.com/paper/2203.11132/) (2022)
- [Disentangled Generative Models for Robust Prediction of System Dynamics](https://mlsanity.com/paper/2108.11684/) (2021)
- [Disentanglement and Generalization Under Correlation Shifts](https://mlsanity.com/paper/2112.14754/) (2021)
- [Gromov-Wasserstein Autoencoders](https://mlsanity.com/paper/2209.07007/) (2022)
- [Disentangling speech from surroundings with neural embeddings](https://mlsanity.com/paper/2203.15578/) (2022)
- [Demystifying Inductive Biases for β-VAE Based Architectures](https://mlsanity.com/paper/2102.06822/) (2021)
- [Visual Referential Games Further the Emergence of Disentangled Representations](https://mlsanity.com/paper/2304.14511/) (2023)
- [Pooling Image Datasets With Multiple Covariate Shift and Imbalance](https://mlsanity.com/paper/2403.02598/) (2024)
- [Synergy Between Sufficient Changes and Sparse Mixing Procedure for Disentangled Representation Learning](https://mlsanity.com/paper/2503.00639/) (2025)
- [The Third Pillar of Causal Analysis? A Measurement Perspective on Causal Representations](https://mlsanity.com/paper/2505.17708/) (2025)
- [A mathematical theory for understanding when abstract representations emerge in neural networks](https://mlsanity.com/paper/2510.09816/) (2025)
- [Learning and controlling the source-filter representation of speech with a variational autoencoder](https://mlsanity.com/paper/2204.07075/) (2022)
- [Meta-Referential Games to Learn Compositional Learning Behaviours](https://mlsanity.com/paper/2207.08012/) (2022)
- [Modeling Complex Disease Trajectories using Deep Generative Models with Semi-Supervised Latent Processes](https://mlsanity.com/paper/2311.08149/) (2023)

## Abstract (arXiv)

The idea behind the unsupervised learning of disentangled representations is that real-world data is generated by a few explanatory factors of variation which can be recovered by unsupervised learning algorithms. In this paper, we provide a sober look at recent progress in the field and challenge some common assumptions. We first theoretically show that the unsupervised learning of disentangled representations is fundamentally impossible without inductive biases on both the models and the data. Then, we train over 14000 models covering most prominent methods and evaluation metrics in a reproducible large-scale experimental study on eight data sets. We observe that while the different methods successfully enforce properties "encouraged" by the corresponding losses, well-disentangled models seemingly cannot be identified without supervision. Furthermore, different evaluation metrics do not always agree on what should be considered "disentangled" and exhibit systematic differences in the estimation. Finally, increased disentanglement does not seem to necessarily lead to a decreased sample complexity of learning for downstream tasks. Our results suggest that future work on disentanglement learning should be explicit about the role of inductive biases and (implicit) supervision, investigate concrete benefits of enforcing disentanglement of the learned representations, and consider a reproducible experimental setup covering several data sets.