Principal component extraction using recursive least squares learning method

A new approach is introduced for the recursive computation of the principal components of a vector stochastic process. The neurons of a single layer perceptron are sequentially trained using a recursive least square type algorithm to extract the principal components of the input process. The approach provides a recursive way to determine the variance associated with each principal component. The proof for convergence is provided as well. Simulation results on an image compression problem are presented and a discussion on the performance of the algorithm is given