A CASCADE NEURAL-NETWORK FOR BLIND SIGNAL EXTRACTION WITHOUT SPURIOUSEQUILIBRIA

We present a cascade neural network for blind source extraction. We pr opose a family of unconstrained optimization criteria, from which we d erive a learning rule that can extract a single source signal from a l inear mixture of source signals. To prevent the newly extracted source signal from being extracted again in the next processing unit, we pro pose another unconstrained optimization criterion that uses knowledge of this signal. From this criterion, we then derive a learning rule th at deflates from the mixture the newly extracted signal. By virtue of blind extraction and deflation processing, the presented cascade neura l network can cope with a practical case where the number of mixed sig nals is equal to or larger than the number of sources, with the number of sources not known in advance. We prove analytically that the propo sed criteria both for blind extraction and deflation processing have n o spurious equilibria. In addition, the proposed criteria do not requi re whitening of mixed signals. We also demonstrate the validity and pe rformance of the presented neural network by computer simulation exper iments.