An alternative switching criterion for independent component analysis (ICA)

In solving the problem of noiseless independent component analysis (ICA) in which sources of super- and sub-Gaussian coexist in an unknown manner, one can be lead to a feasible solution using the natural gradient learning algorithm with a kind of switching criterion for the model probability distribution densities to be selected as super- or sub-Gaussians appropriately during the iterations. In this letter, an alternative switching criterion is proposed for the natural gradient learning algorithm to solve the noiseless ICA problem with both super- and sub-Gaussian sources. It is demonstrated by the experiments that this alternative switching criterion works well on the noiseless ICA problem with both super- and sub-Gaussian sources.