Accelerating the Lee-Seung Algorithm for Nonnegative Matrix Factorization
Edward F. Gonzalez, Yin Zhang · Rice Research Repository (Rice University) · 2005
Approximate nonnegative matrix factorization is an emerging technique with a wide spectrum of potential applications in data analysis. Currently, the most-used algorithms for this problem are those proposed by Lee and Seung. In this paper we present a variation of one of the Lee-Seung algorithms with a notably improved performance. We also show that algorithms of this type do not necessarily converge to local minima.