The main idea of the new-proposed model (ESOM) consists in adaptively approaching the centers of hierarchically organized data clusters. Simultaneously, for each found sub-cluster, mutually independent components should be found.