Guardado en:
Detalles Bibliográficos
Autores principales: Li, Zhitao, Dong, Yiqiu, Zeng, Xueying
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:https://arxiv.org/abs/2511.06235
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866911255860609024
author Li, Zhitao
Dong, Yiqiu
Zeng, Xueying
author_facet Li, Zhitao
Dong, Yiqiu
Zeng, Xueying
contents This paper presents a comprehensive analysis of hyperparameter estimation within the empirical Bayes framework (EBF) for sparse learning. By studying the influence of hyperpriors on the solution of EBF, we establish a theoretical connection between the choice of the hyperprior and the sparsity as well as the local optimality of the resulting solutions. We show that some strictly increasing hyperpriors, such as half-Laplace and half-generalized Gaussian with the power in $(0,1)$, effectively promote sparsity and improve solution stability with respect to measurement noise. Based on this analysis, we adopt a proximal alternating linearized minimization (PALM) algorithm with convergence guaranties for both convex and concave hyperpriors. Extensive numerical tests on two-dimensional image deblurring problems demonstrate that introducing appropriate hyperpriors significantly promotes the sparsity of the solution and enhances restoration accuracy. Furthermore, we illustrate the influence of the noise level and the ill-posedness of inverse problems to EBF solutions.
format Preprint
id arxiv_https___arxiv_org_abs_2511_06235
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Sparsity via Hyperpriors: A Theoretical and Algorithmic Study under Empirical Bayes Framework
Li, Zhitao
Dong, Yiqiu
Zeng, Xueying
Machine Learning
Numerical Analysis
This paper presents a comprehensive analysis of hyperparameter estimation within the empirical Bayes framework (EBF) for sparse learning. By studying the influence of hyperpriors on the solution of EBF, we establish a theoretical connection between the choice of the hyperprior and the sparsity as well as the local optimality of the resulting solutions. We show that some strictly increasing hyperpriors, such as half-Laplace and half-generalized Gaussian with the power in $(0,1)$, effectively promote sparsity and improve solution stability with respect to measurement noise. Based on this analysis, we adopt a proximal alternating linearized minimization (PALM) algorithm with convergence guaranties for both convex and concave hyperpriors. Extensive numerical tests on two-dimensional image deblurring problems demonstrate that introducing appropriate hyperpriors significantly promotes the sparsity of the solution and enhances restoration accuracy. Furthermore, we illustrate the influence of the noise level and the ill-posedness of inverse problems to EBF solutions.
title Sparsity via Hyperpriors: A Theoretical and Algorithmic Study under Empirical Bayes Framework
topic Machine Learning
Numerical Analysis
url https://arxiv.org/abs/2511.06235