For the past 40 years or so, interest in the use of (artificial neural networks has been motivated by the recognition that the human brain operates in a manner that is entirely different from a conventional digital computer. A neural network is made up of an interconnection of a large number of nonlinear computation units known as neurons, which operate in a highly parallel fashion. Interest in the use of neural networks was reignited in the 1980s largely due to (1) the popularization of the back-propagation algorithm as a tool for the training of multilayer perceptrons, and (2) the use of attractor neural networks (exemplified by the Hopfield model) as content-addressable memories and optimization networks. For a historical account of neural networks, the reader is referred to Cowan (1990) and Haykin (1994).