Independent components analysis (ICA) is a mathematical method for separating a signal into its most probable additive subcomponent supposing the statistical independence of the source signals. This assumption is correct in most cases so the blind ICA separation of a mixed signal gives very good results. It is also used for signals that are not supposed to be generated by a mixing for analysis purposes.
The statistical method reveal hidden factors (independent components), presuming their non-gaussianity.
It uses principal components analysis as a first step.
Apply whitening of the data, and then apply an iterative algorithm (such as an adaptive filter).
ICA finds components that are mutually statistically independent to describe the input data matrix (composed by the features vectors). This can allow the reconstruction of source signals from several corrupted mixtures thereof.
The method is important to blind signal separation, EEG analysis and FMRI analysis.
This concept can be extended to analyse non physical signals, for instance ICA has been applied to discover discussions topics on a bag of news list archives.
Related to
