Distributional Diffusion Models with Scoring Rules

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: De Bortoli, Valentin, Galashov, Alexandre, Guntupalli, J. Swaroop, Zhou, Guangyao, Murphy, Kevin, Gretton, Arthur, Doucet, Arnaud
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909676371705856
author De Bortoli, Valentin
Galashov, Alexandre
Guntupalli, J. Swaroop
Zhou, Guangyao
Murphy, Kevin
Gretton, Arthur
Doucet, Arnaud
author_facet De Bortoli, Valentin
Galashov, Alexandre
Guntupalli, J. Swaroop
Zhou, Guangyao
Murphy, Kevin
Gretton, Arthur
Doucet, Arnaud
contents Diffusion models generate high-quality synthetic data. They operate by defining a continuous-time forward process which gradually adds Gaussian noise to data until fully corrupted. The corresponding reverse process progressively "denoises" a Gaussian sample into a sample from the data distribution. However, generating high-quality outputs requires many discretization steps to obtain a faithful approximation of the reverse process. This is expensive and has motivated the development of many acceleration methods. We propose to accomplish sample generation by learning the posterior {\em distribution} of clean data samples given their noisy versions, instead of only the mean of this distribution. This allows us to sample from the probability transitions of the reverse process on a coarse time scale, significantly accelerating inference with minimal degradation of the quality of the output. This is accomplished by replacing the standard regression loss used to estimate conditional means with a scoring rule. We validate our method on image and robot trajectory generation, where we consistently outperform standard diffusion models at few discretization steps.
format Preprint
id arxiv_https___arxiv_org_abs_2502_02483
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Distributional Diffusion Models with Scoring Rules
De Bortoli, Valentin
Galashov, Alexandre
Guntupalli, J. Swaroop
Zhou, Guangyao
Murphy, Kevin
Gretton, Arthur
Doucet, Arnaud
Machine Learning
Diffusion models generate high-quality synthetic data. They operate by defining a continuous-time forward process which gradually adds Gaussian noise to data until fully corrupted. The corresponding reverse process progressively "denoises" a Gaussian sample into a sample from the data distribution. However, generating high-quality outputs requires many discretization steps to obtain a faithful approximation of the reverse process. This is expensive and has motivated the development of many acceleration methods. We propose to accomplish sample generation by learning the posterior {\em distribution} of clean data samples given their noisy versions, instead of only the mean of this distribution. This allows us to sample from the probability transitions of the reverse process on a coarse time scale, significantly accelerating inference with minimal degradation of the quality of the output. This is accomplished by replacing the standard regression loss used to estimate conditional means with a scoring rule. We validate our method on image and robot trajectory generation, where we consistently outperform standard diffusion models at few discretization steps.
title Distributional Diffusion Models with Scoring Rules
topic Machine Learning
url https://arxiv.org/abs/2502.02483