|
摘要: |
非负矩阵分解试图将一个给定矩阵分解为一个非负基矩阵和一个非负系数矩阵的乘积。近年来,非负矩阵分解因其直观的解释性备受关注。交替地非负最小二乘法对于求解非负矩阵分解是非常有效的。然而,它存在全局收敛性理论上不能保证的问题。在这篇文章中,我们提出了一种修正策略,该策略可以保证极限点的存在性,且极限点是非负矩阵分解问题的一个稳定点。此外,我们还给出了增广的修正策略。 |
关键词: 非负矩阵分解 交替地非负最小二乘法 修正策略 |
DOI: |
分类号:O224 |
基金项目: |
|
A Modified Strategy in Alternating Non-negative Least Squares for Non-negative Matrix Factorization |
lixiangli
|
Abstract: |
Non-negative matrix factorization (NMF) attempts to decompose a given nonnegative matrix into the product of a non-negative basis matrix and
a non-negative coefficient matrix. In recent years, NMF has attracted much attention for its straightforward interpretability. Alternating non-negative least squares (ANLS) is a very efficient method for NMF. However, ANLS has a serious problem that its global convergence is not guaranteed in theory. In this paper, we present a modified strategy to guarantee the existence of the limit point, and the limit point is a stationary point of NMF. In addition, we give generalized modified strategies. |
Key words: Non-negative Matrix Factorization Alternating Non-negative Least Squares Modified Strategy |