Neural network-based blind source separation for information recovery

Abstract EN

In this paper we implement the theory of Blind Source Separation and design a neural network that extracts the individual unknown independent source signals out of a given linear mixtures of them using the Kullback- Libeler divergence.

The convergence behavior of the network is demonstrated by presenting experimental results of separating three mixed signals and then five mixed signals.

This technique could be used to extract the information hidden in the signal mixture.