Abstract
The most popular algorithms for Nonnegative Matrix Factorization (NMF) belong to a class of multiplicative Lee-Seung algorithms which have usually relative low complexity but are characterized by slow-convergence and the risk of getting stuck to in local minima. In this paper, we present and compare the performance of additive algorithms based on three different variations of a projected gradient approach. Additionally, we discuss a novel multilayer approach to NMF algorithms combined with multi-start initializations procedure, which in general, considerably improves the performance of all the NMF algorithms. We demonstrate that this approach (the multilayer system with projected gradient algorithms) can usually give much better performance than standard multiplicative algorithms, especially, if data are ill-conditioned, badly-scaled, and/or a number of observations is only slightly greater than a number of nonnegative hidden components. Our new implementations of NMF are demonstrated with the simulations performed for Blind Source Separation (BSS) data.
References
Mar 24, 1998·Journal of the Society for Gynecologic Investigation·P B Hoyer
Nov 5, 1999·Nature·D D Lee, H S Seung
Mar 16, 2004·Proceedings of the National Academy of Sciences of the United States of America·Jean-Philippe BrunetJill P Mesirov
Dec 4, 2004·IEEE Transactions on Medical Imaging·Paul SajdaLucas C Parra
Mar 1, 2006·BMC Bioinformatics·Pedro Carmona-SaezAlberto Pascual-Montano
Mar 11, 2006·IEEE Transactions on Pattern Analysis and Machine Intelligence·Alberto Pascual-MontanoRoberto D Pascual-Marqui
Aug 25, 2007·Neural Computation·Chih-Jen Lin
Jan 1, 1993·IEEE Transactions on Medical Imaging·L Kaufman
Citations
Jul 17, 2008·Computational Intelligence and Neuroscience·Rafal Zdunek, Andrzej Cichocki
May 9, 2009·Computational Intelligence and Neuroscience·M W SpratlingR Kompass
Nov 6, 2014·Computational and Mathematical Methods in Medicine·Deng-ao LiJin Zhang
Jun 9, 2009·Neural Networks : the Official Journal of the International Neural Network Society·Hojjat Adeli, Ashif Panakkat
Jun 19, 2010·IEEE Transactions on Pattern Analysis and Machine Intelligence·Reza Farrahi Moghaddam, Mohamed Cheriet
Oct 20, 2011·IEEE Transactions on Neural Networks·Zhirong Yang, Erkki Oja
Aug 29, 2014·International Journal of Neural Systems·Fengyu CongAndrzej Cichocki
Mar 8, 2012·PloS One·Kenji WatanabeTakio Kurita
Aug 21, 2010·International Journal of Neural Systems·Fengyu CongTapani Ristaniemi
Dec 30, 2009·International Journal of Neural Systems·Md AsaduzzamanKazuyuki Murase
Nov 29, 2012·International Journal of Neural Systems·Fengyu CongAndrzej Cichocki
Jul 18, 2019·Human Brain Mapping·Gareth BallMarc L Seal
Jun 12, 2019·PloS One·Andrej ČoparMarinka Zitnik