A fast computational Gauss–Seidel type iPALM algorithm using an incremental aggregated gradient strategy for weakly convex composite optimization problems with application in image processing

In this paper, we propose a Gauss–Seidel type inertial proximal alternating linearized minimization method with incremental aggregated gradient (IAG-GiPALM) for solving a class of nonconvex and nonsmooth composite optimization problems, whose objective function is the sum of a finite number of smoot...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Journal of computational and applied mathematics Jg. 474; S. 116973
Hauptverfasser: Jia, Zehui, Hou, Junru, Dong, Ping, Liu, Zhiyu
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Elsevier B.V 01.03.2026
Schlagworte:
Incremental aggregated proximal method
Kurdyka-Łojasiewicz property
Nonconvex optimization
Proximal alternating linearized minimization
Proximal alternating linearized minimization
Nonconvex optimization
Kurdyka-Łojasiewicz property
Incremental aggregated proximal method
ISSN:0377-0427
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:In this paper, we propose a Gauss–Seidel type inertial proximal alternating linearized minimization method with incremental aggregated gradient (IAG-GiPALM) for solving a class of nonconvex and nonsmooth composite optimization problems, whose objective function is the sum of a finite number of smooth nonconvex functions and nonsmooth weakly convex functions. This new algorithm inherits the advantages of the Gauss–Seidel type inertial proximal alternating linearized minimization method (GiPALM) and the incremental aggregated proximal method. Under some mild conditions, we prove that any limit point of the sequence generated by IAG-GiPALM is a critical point of the optimization problems. Moreover, we establish the global convergence and convergence rate of the algorithm under the Kurdyka-Łojasiewicz property. In addition, some numerical results are conducted to demonstrate the efficiency of the new method.
ISSN:0377-0427
DOI:10.1016/j.cam.2025.116973