FastICA

FastICA is an efficient and popular algorithm for independent component analysis invented by Aapo Hyvärinen at Helsinki University of Technology. FastICA seeks an orthogonal rotation of prewhitened data, through a fixed-point iteration scheme. Algorithm maximizes measure of non-Gaussianity of rotated components. Non-gaussianity serves as proxy for statistical independence. Fast ICA can also be alternatively derived as an approximative Newton iteration. It can also use a similar strategy for other types of independent signals. It's called FastICA for independent source analysis. It is a very strong condition and requires infinite data to verify. Use this article to help you understand a concept. You need to use the following text to help with understanding FastICA.