Modified BSS algorithms including prior statistical information about mixing matrix
Jorge Igual, Luis Vergara · 2002
In blind source separation (BSS) no prior knowledge is considered. However, due to inherent indeterminations of the problem, some arbitrary normalizing conditions are imposed on the sources or on the recovering matrix. We present a related problem: when we have some prior information about any of the elements of the mixing matrix, and how traditional solutions can be modified incorporating this information to obtain new estimators.