ABSTRACT
The technique of using neural networks is described as connectionism.
Each node in a neural network may have several inputs, each of which has
an associated weighting. The node performs a simple computation on its input
values, which are single integers or real numbers, to produce a single
numerical value as its output. The output from a node can either form an input
to other nodes or be part of the output from the network as a whole. The
overall effect is that a neural network generates a pattern of numbers at its
outputs in response to a pattern of numbers at its inputs. These patterns of
numbers are one-dimensional arrays known as vectors, e.g., (0.1, 1.0, 0.2).
