Saved in:
Bibliographic Details
Main Authors: Jahn, Tim, Kirilin, Mikhail
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2506.14558
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909651555057664
author Jahn, Tim
Kirilin, Mikhail
author_facet Jahn, Tim
Kirilin, Mikhail
contents In this article, we rigorously establish the consistency of generalized cross-validation as a parameter-choice rule for solving inverse problems. We prove that the index chosen by leave-one-out GCV achieves a non-asymptotic, order-optimal error bound with high probability for polynomially ill-posed compact operators. Hereby it is remarkable that the unknown true solution need not satisfy a self-similarity condition, which is generally needed for other heuristic parameter choice rules. We quantify the rate and demonstrate convergence numerically on integral equation test cases, including image deblurring and CT reconstruction.
format Preprint
id arxiv_https___arxiv_org_abs_2506_14558
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Convergence of generalized cross-validation with applications to ill-posed integral equations
Jahn, Tim
Kirilin, Mikhail
Numerical Analysis
In this article, we rigorously establish the consistency of generalized cross-validation as a parameter-choice rule for solving inverse problems. We prove that the index chosen by leave-one-out GCV achieves a non-asymptotic, order-optimal error bound with high probability for polynomially ill-posed compact operators. Hereby it is remarkable that the unknown true solution need not satisfy a self-similarity condition, which is generally needed for other heuristic parameter choice rules. We quantify the rate and demonstrate convergence numerically on integral equation test cases, including image deblurring and CT reconstruction.
title Convergence of generalized cross-validation with applications to ill-posed integral equations
topic Numerical Analysis
url https://arxiv.org/abs/2506.14558