To be or not to be stable, that is the question: understanding neural networks for inverse problems

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Evangelista, Davide, Nagy, James, Morotti, Elena, Piccolomini, Elena Loli
Format: Preprint
Published: 2022
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929235352879104
author Evangelista, Davide
Nagy, James
Morotti, Elena
Piccolomini, Elena Loli
author_facet Evangelista, Davide
Nagy, James
Morotti, Elena
Piccolomini, Elena Loli
contents The solution of linear inverse problems arising, for example, in signal and image processing is a challenging problem since the ill-conditioning amplifies, in the solution, the noise present in the data. Recently introduced algorithms based on deep learning overwhelm the more traditional model-based approaches in performance, but they typically suffer from instability with respect to data perturbation. In this paper, we theoretically analyze the trade-off between stability and accuracy of neural networks, when used to solve linear imaging inverse problems for not under-determined cases. Moreover, we propose different supervised and unsupervised solutions to increase the network stability and maintain a good accuracy, by means of regularization properties inherited from a model-based iterative scheme during the network training and pre-processing stabilizing operator in the neural networks. Extensive numerical experiments on image deblurring confirm the theoretical results and the effectiveness of the proposed deep learning-based approaches to handle noise on the data.
format Preprint
id arxiv_https___arxiv_org_abs_2211_13692
institution arXiv
publishDate 2022
record_format arxiv
spellingShingle To be or not to be stable, that is the question: understanding neural networks for inverse problems
Evangelista, Davide
Nagy, James
Morotti, Elena
Piccolomini, Elena Loli
Numerical Analysis
Machine Learning
65K10, 68T07, 68U10
The solution of linear inverse problems arising, for example, in signal and image processing is a challenging problem since the ill-conditioning amplifies, in the solution, the noise present in the data. Recently introduced algorithms based on deep learning overwhelm the more traditional model-based approaches in performance, but they typically suffer from instability with respect to data perturbation. In this paper, we theoretically analyze the trade-off between stability and accuracy of neural networks, when used to solve linear imaging inverse problems for not under-determined cases. Moreover, we propose different supervised and unsupervised solutions to increase the network stability and maintain a good accuracy, by means of regularization properties inherited from a model-based iterative scheme during the network training and pre-processing stabilizing operator in the neural networks. Extensive numerical experiments on image deblurring confirm the theoretical results and the effectiveness of the proposed deep learning-based approaches to handle noise on the data.
title To be or not to be stable, that is the question: understanding neural networks for inverse problems
topic Numerical Analysis
Machine Learning
65K10, 68T07, 68U10
url https://arxiv.org/abs/2211.13692