Principles and Implementation of Artificial Neural Networks

One of the unique aspects of artificial neural networks is that they combine many of the attributes of both binary and analog circuits. Integrated circuit technology has proven to be a very successful method for realizing digital circuits with extraordinarily large numbers of components. Analog integrated circuits, on the other hand, have for the most part been restricted to relatively modest sizes. The predominance of the digital circuit is largely a result of its insensitivity to component variations, since these circuits rely primarily on the action of switches for their functionality. Analog circuits, by comparison, show a higher degree of sensitivity to the properties of the components-- amplifiers and resistors, for example-- that they comprise. For this reason, they are much more difficult to build with the same degree of size and complexity found in digital circuits.