GENERALIZED EIGENDECOMPOSITION WITH AN ONLINE LOCAL ALGORITHM

This letter presents a novel, on-line, local learning algorithm to obt ain generalized eigenvalues and their corresponding eigenvectors in de scending order with a Linear adaptive fitter. The filter is composed o f a set of forward linear projections constrained by lateral inhibitio ns. Equilibrium points of the adaptation process and their stability a re briefly analyzed. Simulations and experimental comparisons are give n to verify the validity and effectiveness of the proposed algorithm.