Non-Negative Matrix Factorization based on Projected Nonlinear Conjugate Gradient Algorithm
W Wang, Xianchun Zou · Surrey Research Insight Open Access (The University of Surrey) · 2008
The popular multiplicative algorithms in non-negative ma-trix factorization (NMF) are known to have slow conver-gence. Several algorithms have been proposed to improve the convergence of iterative algorithms in NMF, such as the projected gradient algorithms. However, these algo-rithms also suffer a common problem, that is, a previously exploited descent direction may be searched again in sub-sequent iterations which potentially leads to slow conver-gence of these algorithms. In this paper, we propose a projected non-linear conjugate gradient algorithm using orthogonal searching directions at each iteration which ensures each descent direction is different from others. The algorithm is shown to have better convergence per-formance as compared with both multiplicative algorithms and the projected gradient algorithms.