Factor analysis as data matrix decomposition: A new approach for quasi-sphering in noisy ICA

Abstract

In this paper, a new approach for quasi-sphering in noisy ICA by means of exploratory factor analysis (EFA) is introduced. The EFA model is considered as a novel form of data matrix decomposition. By factoring the data matrix, estimates for all EFA model parameters are obtained simultaneously. After the preprocessing, an existing ICA algorithm can be used to rotate the sphered factor scores towards indpendence. An application to climate data is presented to illustrate the proposed approach. © Springer-Verlag Berlin Heidelberg 2009.