SIMULTANEOUS EXTRACTION OF PRINCIPAL COMPONENTS
Output Variances, Yadunandana N. Rao, José Carlos Príncipe, Kenneth E. Hild · 2002
Principal Components Analysis (PCA) is an invaluable statistical tool in signal processing. In many cases, an on-line algorithm to adapt the PCA network to determine the principal projections in the input space is desire d. Algorithms proposed until now use the traditional deflation or the inflation pro cedure to determine the intermediate compo nents sequentially, after the convergence of the principal or minor comp onent is achieved. In this paper, we propose a constrain ed linear network and a robust cost function to determine any number of principal components simultaneously. The topology exploits the fact that the eigenvector matrix sought is orthonormal. A gradient-based algorithm named SIPEX-G is also presented.