Saved in:
Bibliographic Details
Main Authors: Hadžić, Armin, Papez, Milan, Pevný, Tomáš
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2509.01400
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911132266004480
author Hadžić, Armin
Papez, Milan
Pevný, Tomáš
author_facet Hadžić, Armin
Papez, Milan
Pevný, Tomáš
contents Deep generative models with discrete latent space, such as the Vector-Quantized Variational Autoencoder (VQ-VAE), offer excellent data generation capabilities, but, due to the large size of their latent space, their probabilistic inference is deemed intractable. We demonstrate that the VQ-VAE can be distilled into a tractable model by selecting a subset of latent variables with high probabilities. This simple strategy is particularly efficient, especially if the VQ-VAE underutilizes its latent space, which is, indeed, very often the case. We frame the distilled model as a probabilistic circuit, and show that it preserves expressiveness of the VQ-VAE while providing tractable probabilistic inference. Experiments illustrate competitive performance in density estimation and conditional generation tasks, challenging the view of the VQ-VAE as an inherently intractable model.
format Preprint
id arxiv_https___arxiv_org_abs_2509_01400
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Distillation of a tractable model from the VQ-VAE
Hadžić, Armin
Papez, Milan
Pevný, Tomáš
Machine Learning
Deep generative models with discrete latent space, such as the Vector-Quantized Variational Autoencoder (VQ-VAE), offer excellent data generation capabilities, but, due to the large size of their latent space, their probabilistic inference is deemed intractable. We demonstrate that the VQ-VAE can be distilled into a tractable model by selecting a subset of latent variables with high probabilities. This simple strategy is particularly efficient, especially if the VQ-VAE underutilizes its latent space, which is, indeed, very often the case. We frame the distilled model as a probabilistic circuit, and show that it preserves expressiveness of the VQ-VAE while providing tractable probabilistic inference. Experiments illustrate competitive performance in density estimation and conditional generation tasks, challenging the view of the VQ-VAE as an inherently intractable model.
title Distillation of a tractable model from the VQ-VAE
topic Machine Learning
url https://arxiv.org/abs/2509.01400