Adaptive computation of the Symmetric Nonnegative Matrix Factorization (NMF)
For practitioners using SymNMF in data analysis, this work offers an improved heuristic for parameter selection, but it is incremental.
The paper proposes an adaptive heuristic for the penalizing parameter in Symmetric Nonnegative Matrix Factorization (SymNMF), showing effectiveness through extensive experiments.
Nonnegative Matrix Factorization (NMF), first proposed in 1994 for data analysis, has received successively much attention in a great variety of contexts such as data mining, text clustering, computer vision, bioinformatics, etc. In this paper the case of a symmetric matrix is considered and the symmetric nonnegative matrix factorization (SymNMF) is obtained by using a penalized nonsymmetric minimization problem. Instead of letting the penalizing parameter increase according to an a priori fixed rule, as suggested in literature, we propose a heuristic approach based on an adaptive technique. Extensive experimentation shows that the proposed algorithm is effective.