Measurement-Aligned Sampling for Inverse Problem

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Zhang, Shaorong, Brekelmans, Rob, Wu, Yunshu, Steeg, Greg Ver
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866912626702811136
author Zhang, Shaorong
Brekelmans, Rob
Wu, Yunshu
Steeg, Greg Ver
author_facet Zhang, Shaorong
Brekelmans, Rob
Wu, Yunshu
Steeg, Greg Ver
contents Diffusion models provide a powerful way to incorporate complex prior information for solving inverse problems. However, existing methods struggle to correctly incorporate guidance from conflicting signals in the prior and measurement, and often failed to maximizing the consistency to the measurement, especially in the challenging setting of non-Gaussian or unknown noise. To address these issues, we propose Measurement-Aligned Sampling (MAS), a novel framework for linear inverse problem solving that flexibly balances prior and measurement information. MAS unifies and extends existing approaches such as DDNM, TMPD, while generalizing to handle both known Gaussian noise and unknown or non-Gaussian noise types. Extensive experiments demonstrate that MAS consistently outperforms state-of-the-art methods across a variety of tasks, while maintaining relatively low computational cost.
format Preprint
id arxiv_https___arxiv_org_abs_2506_11893
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Measurement-Aligned Sampling for Inverse Problem
Zhang, Shaorong
Brekelmans, Rob
Wu, Yunshu
Steeg, Greg Ver
Machine Learning
Diffusion models provide a powerful way to incorporate complex prior information for solving inverse problems. However, existing methods struggle to correctly incorporate guidance from conflicting signals in the prior and measurement, and often failed to maximizing the consistency to the measurement, especially in the challenging setting of non-Gaussian or unknown noise. To address these issues, we propose Measurement-Aligned Sampling (MAS), a novel framework for linear inverse problem solving that flexibly balances prior and measurement information. MAS unifies and extends existing approaches such as DDNM, TMPD, while generalizing to handle both known Gaussian noise and unknown or non-Gaussian noise types. Extensive experiments demonstrate that MAS consistently outperforms state-of-the-art methods across a variety of tasks, while maintaining relatively low computational cost.
title Measurement-Aligned Sampling for Inverse Problem
topic Machine Learning
url https://arxiv.org/abs/2506.11893