Projected gradient methods for nonnegative matrix factorization

被引:1186
作者
Lin, Chih-Jen [1 ]
机构
[1] Natl Taiwan Univ, Dept Comp Sci, Taipei 106, Taiwan
关键词
D O I
10.1162/neco.2007.19.10.2756
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Nonnegative matrix factorization (NMF) can be formulated as a minimization problem with bound constraints. Although bound-constrained optimization has been studied extensively in both theory and practice, so far no study has formally applied its techniques to NME In this letter, we propose two projected gradient methods for NMF, both of which exhibit strong optimization properties. We discuss efficient implementations and demonstrate that one of the proposed methods converges faster than the popular multiplicative update approach. A simple Matlab code is also provided.
引用
收藏
页码:2756 / 2779
页数:24
相关论文
共 23 条
[1]  
[Anonymous], 2005, ACCELERATING LEE SEU
[2]  
[Anonymous], ADV NEURAL INFORM PR
[3]  
Bertsekas D., 1999, NONLINEAR PROGRAMMIN
[4]   GOLDSTEIN-LEVITIN-POLYAK GRADIENT PROJECTION METHOD [J].
BERTSEKAS, DP .
IEEE TRANSACTIONS ON AUTOMATIC CONTROL, 1976, 21 (02) :174-183
[5]   Metagenes and molecular pattern discovery using matrix factorization [J].
Brunet, JP ;
Tamayo, P ;
Golub, TR ;
Mesirov, JP .
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA, 2004, 101 (12) :4164-4169
[6]  
BRUNET JP, 2004, NMF PROGRAM
[7]   PROJECTED GRADIENT METHODS FOR LINEARLY CONSTRAINED PROBLEMS [J].
CALAMAI, PH ;
MORE, JJ .
MATHEMATICAL PROGRAMMING, 1987, 39 (01) :93-116
[8]  
CHU M, 2005, OPTIMALITY COMPUTING
[9]   On the convergence of the block nonlinear Gauss-Seidel method under convex constraints [J].
Grippo, L ;
Sciandrone, M .
OPERATIONS RESEARCH LETTERS, 2000, 26 (03) :127-136
[10]  
Hoyer PO, 2004, J MACH LEARN RES, V5, P1457