Research Article
Non Negative Matrix Factorization for Blind Source Separation
@INPROCEEDINGS{10.4108/eai.24-4-2019.2284117, author={Nabila Aoulass and Otman Chakkor}, title={Non Negative Matrix Factorization for Blind Source Separation}, proceedings={Proceedings of the Third International Conference on Computing and Wireless Communication Systems, ICCWCS 2019, April 24-25, 2019, Faculty of Sciences, Ibn Tofa\~{n}l University -K\^{e}nitra- Morocco}, publisher={EAI}, proceedings_a={ICCWCS}, year={2019}, month={5}, keywords={source separation blind source separation nmf sparsity}, doi={10.4108/eai.24-4-2019.2284117} }- Nabila Aoulass
Otman Chakkor
Year: 2019
Non Negative Matrix Factorization for Blind Source Separation
ICCWCS
EAI
DOI: 10.4108/eai.24-4-2019.2284117
Abstract
Non negative Matrix Factorization (NMF) has been a popular representation method for pattern classification problems. It tries to decompose a non negative matrix of data samples as the product of a non negative basis matrix and a non negative coefficient matrix in NMF both supervised and unsupervised mode of operations is used. Among them supervised mode outperforms well due to the use of pre-learned basis vectors corresponding to each underlying sources. In this paper NMF algorithms such as method based in the Frobinuis norm, Kullback Leibler divergence , and an extension to NMF, by incorporating sparsity. Algorithms are used to evaluate the performance of BSS in which supervised mode is used. We further illustrate the effect of hyperparameter as the rank k let the metric chooses and the initialization of decomposition matrices, on the speed of convergence of NMF algorithm.