Novel adaptive generalized principal component analysis algorithm based on Hebbian rule

In order to adaptively estimate the generalized principal component from input signals,a novel generalized principal component analysis algorithm was proposed based on the Hebbian linear neuron model.Since the autocorrelation matrices of the signals were estimated directly from the sampled data at the current time,the proposed algorithm had low computation complexity.Trough analyzing all of the equilibrium points by Lyapunov method,it is proven that if and only if the weight vector in the neuron had the same direction with the generalized principal component,the proposed algorithm attains the convergence status.Simulation results shows that compared with some same type algorithms,the proposed algorithm has faster convergence speed.