I really don't know why matrix factorization is synonymous to NMF but it bothers me a little. As if positivity constraints were the only thing that really mattered, we know it ain't so. Anyway, enough of the rant, here is an NMF related Complex Matrix Factorization Toolbox I spotted thanks to the Google. It is here.
Complex Matrix Factorization Toolbox
We acknowledge the support of the U.S. Air Force Office of Scientific Research and the U.S. Office of Naval Research in the development of this toolbox.
The toolbox contains all of the state-of-the-art methods of complex matrix factorization, including complex matrix factorization with intra-source additivity constraints (CMFWISA) and complex probabilistic latent component analysis (CPLCA). These methods are currently optimized for use with a short time Fourier transform (STFT) for single-channel source separation of audio, but can be used for any complex-valued matrix data.
A good introduction to the subject is featured in Brian King's Thesis entitled "New Methods of Complex Matrix Factorization for Single-Channel Source Separation and Analysis"
In the acknowledgement section, I liked that part: "...Although our heavy computational load caused a minor computer fire and allegedly brought a cluster to its knees and its admin into a t of rage, we were finally able to complete the experiments..." The term "allegedly" reeks of plausible deniability. I like that.
Liked this entry ? subscribe to Nuit Blanche's feed, there's more where that came from. You can also subscribe to Nuit Blanche by Email, explore the Big Picture in Compressive Sensing or the Matrix Factorization Jungle and join the conversations on compressive sensing, advanced matrix factorization and calibration issues on Linkedin.