Independent component analysis
ICA attempts to decompose into independent non-Gaussian components. ICA finds the independent components by maximizing the statistical independence of the estimated components.
The two broadest definitions of independence for ICA are:
- Minimization of Mutual information
- Maximization of non-Gaussianity
Algorithms
- infomax
- FastICA
- JADE
- kernel-ICA
- MELODIC (Multivariate Exploratory Linear Optimized Decomposition into Independent Components)
Application
A typical application of ICA is the “cocktail party problem”, where the underlying speech signal are separated from a sample data consisting of people talking simultaneously in a room since it maximize non-Gaussianity. ICA performs a blind source separation by exploiting the independence and
non-Gaussianity of the original sources. 14.7 Independent Component Analysis and Explor.
ICA has become an important tool in the study of brain dynamics (Brain Wave). ICA has become a widely used method for extracting functional brain networks (regions with significant correlated signal) in the brain during rest and task.
