A neurocomposition method for extraction of principal components of stochastic processes

A neurocomputation method with an APEX (adaptive principal components extraction) algorithm has recently been proposed by S.Y. Kung and K.I. Diamantaras (1990). In the present work, an improved method with an algorithm called OPEX is presented. It was shown by simulation that OPEX is more robust and has a shorter convergence time than APEX when small eigenvalues are present and the autocorrelation matrix of the input process is ill conditioned.<
>