Neural blind separation of complex sources by extended APEX algorithm (EAPEX)
Simone Fiori, Aurelio Uncini, Francesco Piazza · 2003
Blind Source Separation by non-classical (non-quadratic) neural Principal Component Analysis has been investigated by several papers over the recent years, even if particular attention has been paid to the real-valued sources case. The aim of this work is to present an extension of the Kung-Diamantaras' APEX learning rule to non-quadratic complex optimization, and to show the new approach allows blind separation of complex-valued source signals from their linear mixtures.