Modeling by Neural Networks

Abstract

Up to now we have considered a number of concepts needed for the development of devices capable of automatic modeling of natural phenomena from the physical description of nature. This approach has already led us to adaptive network systems which consist of a number of relatively simple information processing units. We have called these formal neurons because their adaptive properties are similar to those of biological neurons. The aim of this chapter is to explain this similarity in more detail by showing how the adaptation of biological neurons and their networks can be described by dynamical models. Because the corresponding field of research is very broad, we present only those fundamental properties which are of importance for the modeling of natural laws. For the other topics, the reader is advised to consult the literature cited in the bibliography of this chapter.