Can Diffusion Models Disentangle? A Theoretical Perspective

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wang, Liming, Mirza, Muhammad Jehanzeb, Gong, Yishu, Gong, Yuan, Zhang, Jiaqi, Tracey, Brian H., Placek, Katerina, Vilela, Marco, Glass, James R.
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911176327168000
author Wang, Liming
Mirza, Muhammad Jehanzeb
Gong, Yishu
Gong, Yuan
Zhang, Jiaqi
Tracey, Brian H.
Placek, Katerina
Vilela, Marco
Glass, James R.
author_facet Wang, Liming
Mirza, Muhammad Jehanzeb
Gong, Yishu
Gong, Yuan
Zhang, Jiaqi
Tracey, Brian H.
Placek, Katerina
Vilela, Marco
Glass, James R.
contents This paper presents a novel theoretical framework for understanding how diffusion models can learn disentangled representations. Within this framework, we establish identifiability conditions for general disentangled latent variable models, analyze training dynamics, and derive sample complexity bounds for disentangled latent subspace models. To validate our theory, we conduct disentanglement experiments across diverse tasks and modalities, including subspace recovery in latent subspace Gaussian mixture models, image colorization, image denoising, and voice conversion for speech classification. Additionally, our experiments show that training strategies inspired by our theory, such as style guidance regularization, consistently enhance disentanglement performance.
format Preprint
id arxiv_https___arxiv_org_abs_2504_00220
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Can Diffusion Models Disentangle? A Theoretical Perspective
Wang, Liming
Mirza, Muhammad Jehanzeb
Gong, Yishu
Gong, Yuan
Zhang, Jiaqi
Tracey, Brian H.
Placek, Katerina
Vilela, Marco
Glass, James R.
Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
This paper presents a novel theoretical framework for understanding how diffusion models can learn disentangled representations. Within this framework, we establish identifiability conditions for general disentangled latent variable models, analyze training dynamics, and derive sample complexity bounds for disentangled latent subspace models. To validate our theory, we conduct disentanglement experiments across diverse tasks and modalities, including subspace recovery in latent subspace Gaussian mixture models, image colorization, image denoising, and voice conversion for speech classification. Additionally, our experiments show that training strategies inspired by our theory, such as style guidance regularization, consistently enhance disentanglement performance.
title Can Diffusion Models Disentangle? A Theoretical Perspective
topic Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2504.00220