Saved in:
Bibliographic Details
Main Authors: Rozet, François, Andry, Gérôme, Lanusse, François, Louppe, Gilles
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2405.13712
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914127146909696
author Rozet, François
Andry, Gérôme
Lanusse, François
Louppe, Gilles
author_facet Rozet, François
Andry, Gérôme
Lanusse, François
Louppe, Gilles
contents Diffusion models recently proved to be remarkable priors for Bayesian inverse problems. However, training these models typically requires access to large amounts of clean data, which could prove difficult in some settings. In this work, we present DiEM, a novel method based on the expectation-maximization algorithm for training diffusion models from incomplete and noisy observations only. Unlike previous works, DiEM leads to proper diffusion models, which is crucial for downstream tasks. As part of our methods, we propose and motivate an improved posterior sampling scheme for unconditional diffusion models. We present empirical evidence supporting the effectiveness of our approach.
format Preprint
id arxiv_https___arxiv_org_abs_2405_13712
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Learning Diffusion Priors from Observations by Expectation Maximization
Rozet, François
Andry, Gérôme
Lanusse, François
Louppe, Gilles
Machine Learning
Diffusion models recently proved to be remarkable priors for Bayesian inverse problems. However, training these models typically requires access to large amounts of clean data, which could prove difficult in some settings. In this work, we present DiEM, a novel method based on the expectation-maximization algorithm for training diffusion models from incomplete and noisy observations only. Unlike previous works, DiEM leads to proper diffusion models, which is crucial for downstream tasks. As part of our methods, we propose and motivate an improved posterior sampling scheme for unconditional diffusion models. We present empirical evidence supporting the effectiveness of our approach.
title Learning Diffusion Priors from Observations by Expectation Maximization
topic Machine Learning
url https://arxiv.org/abs/2405.13712