“…The third approach is based on clustering techniques for defining training algorithms, which are analysed in the context of classification theory. In this case the equalizer may be either a simple classifier (Georgoulakis and Theodoridis, 2000) or a radial-basis function network (RBF) (Chen, Mulgrew, and Grant, 1993;Chen, McLaughlin, and Mulgrew, 1994;Gan et al, 1999). Finally, the fourth approach employs hybrid structures, such as the combinations of the self-organized map (SOM) along with the the RBF network , as well as the combinations involving linear transversal filtering (LTE) and the SOM , also including LTE-RBF and LTE-MLP Castanié, 1998/1999).…”