Efficient Nonnegative Matrix Factorization via Modified Monotone Barzilai-Borwein Method with Adaptive Step Sizes Strategy
Year: 2023
Author: Wenbo Li, Jicheng Li, Xuenian Liu
Journal of Computational Mathematics, Vol. 41 (2023), Iss. 5 : pp. 866–878
Abstract
In this paper, we develop an active set identification technique. By means of the active set technique, we present an active set adaptive monotone projected Barzilai-Borwein method (ASAMPBB) for solving nonnegative matrix factorization (NMF) based on the alternating nonnegative least squares framework, in which the Barzilai-Borwein (BB) step sizes can be adaptively picked to get meaningful convergence rate improvements. To get optimal step size, we take into account of the curvature information. In addition, the larger step size technique is exploited to accelerate convergence of the proposed method. The global convergence of the proposed method is analysed under mild assumption. Finally, the results of the numerical experiments on both synthetic and real-world datasets show that the proposed method is effective.
You do not have full access to this article.
Already a Subscriber? Sign in as an individual or via your institution
Journal Article Details
Publisher Name: Global Science Press
Language: English
DOI: https://doi.org/10.4208/jcm.2201-m2019-0145
Journal of Computational Mathematics, Vol. 41 (2023), Iss. 5 : pp. 866–878
Published online: 2023-01
AMS Subject Headings:
Copyright: COPYRIGHT: © Global Science Press
Pages: 13
Keywords: Adaptive step sizes Alternating nonnegative least squares Monotone projected Barzilai-Borwein method Active set strategy Larger step size.