One-step Diffusion Models with Bregman Density Ratio Matching

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhu, Yuanzhi, Tsonis, Eleftherios, Degeorge, Lucas, Kalogeiton, Vicky
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918163934871552
author Zhu, Yuanzhi
Tsonis, Eleftherios
Degeorge, Lucas
Kalogeiton, Vicky
author_facet Zhu, Yuanzhi
Tsonis, Eleftherios
Degeorge, Lucas
Kalogeiton, Vicky
contents Diffusion and flow models achieve high generative quality but remain computationally expensive due to slow multi-step sampling. Distillation methods accelerate them by training fast student generators, yet most existing objectives lack a unified theoretical foundation. In this work, we propose Di-Bregman, a compact framework that formulates diffusion distillation as Bregman divergence-based density-ratio matching. This convex-analytic view connects several existing objectives through a common lens. Experiments on CIFAR-10 and text-to-image generation demonstrate that Di-Bregman achieves improved one-step FID over reverse-KL distillation and maintains high visual fidelity compared to the teacher model. Our results highlight Bregman density-ratio matching as a practical and theoretically-grounded route toward efficient one-step diffusion generation.
format Preprint
id arxiv_https___arxiv_org_abs_2510_16983
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle One-step Diffusion Models with Bregman Density Ratio Matching
Zhu, Yuanzhi
Tsonis, Eleftherios
Degeorge, Lucas
Kalogeiton, Vicky
Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
Diffusion and flow models achieve high generative quality but remain computationally expensive due to slow multi-step sampling. Distillation methods accelerate them by training fast student generators, yet most existing objectives lack a unified theoretical foundation. In this work, we propose Di-Bregman, a compact framework that formulates diffusion distillation as Bregman divergence-based density-ratio matching. This convex-analytic view connects several existing objectives through a common lens. Experiments on CIFAR-10 and text-to-image generation demonstrate that Di-Bregman achieves improved one-step FID over reverse-KL distillation and maintains high visual fidelity compared to the teacher model. Our results highlight Bregman density-ratio matching as a practical and theoretically-grounded route toward efficient one-step diffusion generation.
title One-step Diffusion Models with Bregman Density Ratio Matching
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2510.16983