Learning Disentangled Discrete Representations
Not on the map. Why?
Abstract (arXiv)
Recent successes in image generation, model-based reinforcement learning, and text-to-image generation have demonstrated the empirical advantages of discrete latent representations, although the reasons behind their benefits remain unclear. We explore the relationship between discrete latent spaces and disentangled representations by replacing the standard Gaussian variational autoencoder (VAE) with a tailored categorical variational autoencoder. We show that the underlying grid structure of categorical distributions mitigates the problem of rotational invariance associated with multivariate Gaussian distributions, acting as an efficient inductive prior for disentangled representations. We provide both analytical and empirical findings that demonstrate the advantages of discrete VAEs for learning disentangled representations. Furthermore, we introduce the first unsupervised model selection strategy that favors disentangled representations.
Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Builds on 33 of its 51 references that are in the map. First posted on arXiv 26 Jul 2023.
Builds on
33 of 51 references.
References · 33
- 2017Attention Is All You Needcited by 59,932
- 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,694
- 2017Neural Discrete Representation Learningcited by 4,650
- 2021Zero-Shot Text-to-Image Generationcited by 4,276
- 2016Categorical Reparameterization with Gumbel-Softmaxcited by 2,708
- 2013Estimating or Propagating Gradients Through Stochastic Neurons for Conditional Computationcited by 2,031
- 2019Generating Diverse High-Fidelity Images with VQ-VAE-2cited by 1,376
- 2023Mastering Diverse Domains through World Modelscited by 1,193
- 2016The Concrete Distribution: A Continuous Relaxation of Discrete Random Variablescited by 1,135
- 2020Mastering Atari with Discrete World Modelscited by 990
- 2018Challenging Common Assumptions in the Unsupervised Learning of Disentangled Representationscited by 929
- 2018Disentangling by Factorisingcited by 614
21 more
- 2015Adversarial Autoencoderscited by 359
- 2018Understanding disentangling in β-VAEcited by 332
- 2018Towards a Definition of Disentangled Representationscited by 294
- 2017Variational Inference of Disentangled Latent Concepts from Unlabeled Observationscited by 240
- 2020Weakly-Supervised Disentanglement Without Compromisescited by 237
- 2019Flexibly Fair Representation Learning by Disentanglementcited by 196
- 2019On the Fairness of Disentangled Representationscited by 144
- 2018Recent Advances in Autoencoder-Based Representation Learningcited by 130
- 2020Towards Nonlinear Disentanglement in Natural Data with Temporal Sparse Codingcited by 121
- 2019Spatial Broadcast Decoder: A Simple Architecture for Learning Disentangled Representations in VAEscited by 116
- 2017Backpropagation through the Void: Optimizing control variates for black-box gradient estimationcited by 104
- 2018Learning Deep Disentangled Embeddings with the F-Statistic Losscited by 101
- 2019Are Disentangled Representations Helpful for Abstract Visual Reasoning?cited by 100
- 2019On the Transfer of Inductive Bias from Simulation to the Real World: a New Disentanglement Datasetcited by 100
- 2020On Disentangled Representations Learned From Correlated Datacited by 94
- 2018Learning Disentangled Joint Continuous and Discrete Representationscited by 83
- 2019Weakly Supervised Disentanglement with Guaranteescited by 82
- 2017REBAR: Low-variance, unbiased gradient estimates for discrete latent variable modelscited by 78
- 2021Vector Quantized Models for Planningcited by 47
- 2017Learning Discrete Weights Using the Local Reparameterization Trickcited by 29
- 2019Learning Discrete and Continuous Factors of Data via Alternating Disentanglementcited by 23
Built on by
Citing papers per year: 2026: 1.
1 paper.