Signal Separation using Non-negative Matrix Factorization Based on R1-norm
W Kider, Mohamed Abd Elaziz · 2012
Abstract: Nonnegative Matrix Factorization (NMF) based methods have found use in the context of blind source separation, semi-supervised, and unsupervised learning. These techniques require the use of a suitable cost function to determine the optimal factorization, and most work has focused on the use of least square formulation which is prone to large noise and outliers. In this paper we developed robust NMF algorithm using R1-norm which exhibit stability and robustness w.r.t. large noises. This algorithm is as efficient as the algorithms for least square formulations, avoiding the significant computational complexities routinely associated with R1-norm formulations. The experimental show that R1-NMF can effectively separate the observed that contain outliers better than standard NMF. [ W. kider and M. E. Abd El Aziz. Signal Separation using Non-negative Matrix Factorization Based on R1norm.