Investigation of the Impact of Bad Learning on Blind Source Separation

Abstract

Blind source separation is one of the more challenging problems in signal processing where mixture of sources, sounds or any type of signals are to be separated and located without any prior information about these sources and the way they have been mixed. Any blind source separation algorithm consists of a contrast or cost function plus an adaptation or learning rule which is used to minimize or maximize the cost function so as to perform separation. This typically used in human being brains to distinguish between the different voices and scenes. Here we introduce an approach that explains some strange natural phenomena and relate their occurrence to what is called bad learning in the process of blind source separation.