Independent components analysis
Appearance
Use the PCA as a first step. Apply Whitenning of the data, and then apply an iterative algorithm.
The ICA search orthogonal components that are *statisticaly* independent between them to describe the input data matrix (composed by the features vectors).
The method was invented by Erkky Oja. And it is very used in the blind source separation, EEG analisys and FMRI analisys fields.
Related to