COMPETITIVE PRINCIPAL COMPONENT ANALYSIS FOR LOCALLY STATIONARY TIME-SERIES

A new unsupervised algorithm is proposed that performs competitive pri ncipal component analysis (PCA) of a time series. A set of expert PCA networks compete, through the mixture of experts (MOE) formalism, on t he basis of their ability to reconstruct the original signal. The resu lting network finds an optimal projection of the input onto a reduced dimensional space as a function of the input and, hence, of time. As a byproduct, the time series is both segmented and identified according to stationary regions. Examples showing the performance of the algori thm are included.