ABSTRACT
This chapter reviews the general advantages of artificial neural networks (ANNs) which have motivated their use in practical applications. It explains two alternative definitions (computer hardware oriented and brain oriented) of an ANN, and provides an overview of the computational tasks that various classes of ANNs can perform. The advantages include: (i) access to existing sixth-generation computer hardware with huge price–performance advantages; (ii) links to brain-like intelligence; (iii) ease of use; (iv) superior approximation of nonlinear functions; (v) advantages of learning over tweaking, including learning off-line to be adaptive on-line (in control); (vi) availability of many specific designs providing nonlinear generalizations of many familiar algorithms. Among the algorithms and applications are those for image and speech preprocessing, function maximization or minimization, feature extraction, pattern classification, function approximation, identification and control of dynamical systems, data compression, and so on.
