Jorge Igual, Luis Vergara

In BSS­ICA few assumptions are considered. With respect to the mixing matrix usually nothing is supposed. We present in this paper an extension of the BSS­ICA problem: when we have a prior statistical information about the elements of the mixing matrix. We will model this information with a matrix of probability density functions (pdf) and we will analyze how traditional and new solutions to BSS­ICA problem can be adjusted to the new problem.