Extreme Point Pursuit -- Part II: Further Error Bound Analysis and Applications

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Liu, Junbin, Liu, Ya, Ma, Wing-Kin, Shao, Mingjie, So, Anthony Man-Cho
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866917832067907584
author Liu, Junbin
Liu, Ya
Ma, Wing-Kin
Shao, Mingjie
So, Anthony Man-Cho
author_facet Liu, Junbin
Liu, Ya
Ma, Wing-Kin
Shao, Mingjie
So, Anthony Man-Cho
contents In the first part of this study, a convex-constrained penalized formulation was studied for a class of constant modulus (CM) problems. In particular, the error bound techniques were shown to play a vital role in providing exact penalization results. In this second part of the study, we continue our error bound analysis for the cases of partial permutation matrices, size-constrained assignment matrices and non-negative semi-orthogonal matrices. We develop new error bounds and penalized formulations for these three cases, and the new formulations possess good structures for building computationally efficient algorithms. Moreover, we provide numerical results to demonstrate our framework in a variety of applications such as the densest k-subgraph problem, graph matching, size-constrained clustering, non-negative orthogonal matrix factorization and sparse fair principal component analysis.
format Preprint
id arxiv_https___arxiv_org_abs_2403_06513
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Extreme Point Pursuit -- Part II: Further Error Bound Analysis and Applications
Liu, Junbin
Liu, Ya
Ma, Wing-Kin
Shao, Mingjie
So, Anthony Man-Cho
Signal Processing
Optimization and Control
In the first part of this study, a convex-constrained penalized formulation was studied for a class of constant modulus (CM) problems. In particular, the error bound techniques were shown to play a vital role in providing exact penalization results. In this second part of the study, we continue our error bound analysis for the cases of partial permutation matrices, size-constrained assignment matrices and non-negative semi-orthogonal matrices. We develop new error bounds and penalized formulations for these three cases, and the new formulations possess good structures for building computationally efficient algorithms. Moreover, we provide numerical results to demonstrate our framework in a variety of applications such as the densest k-subgraph problem, graph matching, size-constrained clustering, non-negative orthogonal matrix factorization and sparse fair principal component analysis.
title Extreme Point Pursuit -- Part II: Further Error Bound Analysis and Applications
topic Signal Processing
Optimization and Control
url https://arxiv.org/abs/2403.06513