Functional networks

Abstract

Functional networks were introduced by Castillo [22] as a powerful alternative to ANN. Unlike Neural Networks (Adeli and Huang, 1995) [1], Functional Networks use domain knowledge in addition to data knowledge. The network's initial topology is derived based on the modeling of the properties of the real world. Once this topology is available, functional equations allow one to obtain a much simpler equivalent topology. Although functional networks also can deal with data only, the class of problems where functional networks are most convenient are the classes where the two sources of knowledge both about domain and data are available. In this chapter, functional networks is applied a) for identification of rocks and b) hot extrusion of steel. © 2009 Springer-Verlag Berlin Heidelberg.