Improved Techniques for Blind Source Separation
Yongjian Zhao · Advances in computer science research · 2015
In many practical applications such as biomedical signal processing, it is often desirable to extract one or a few source signals instead of all signals.The classical FASTICA algorithm can extract a source signal which has the maximum negentropy of all signals.However, the extracted signal is not necessarily the desired one.To address these problems above, a constraint is introduced to a negentropy based cost function.As a result, a constrained method is proposed which can extract a desired signal from its mixture exclusively.