Neural Networks and Analog Computation

The primary motivation of this book is the need to understand the theoretical foundations of neural networks as computational devices. Underlying this need is the concept of "connectionism", which is concerned with the computational and learning capabilities of assemblies of simple processors, called artificial neural networks. Although the concept of a neural network essentially arises from biology, many engineering applications have been found through highly idealized and simplified models of neuron behaviour. Particular areas of application have been as diverse as explosives detection in airport security, signiture verification, financial and medical time series prediction, vision, speech processing, robotics, nonlinear control, and signal processing. The focus in all three models is entirely on the behaviour of networks as computers. The material should be of interest to researchers in a variety of engineering and applied sciences disciplines. In addition, it could provide the basis for a graduate level seminar in neural networks for computer science students.