General Principles of Operations in Neuron Nets with Application to Acoustical Pattern Recognition

This report presents general principles of operations in neuron networks and is composed of two parts. One is concerned with the theoretical aspects of operations in neuron nets; the other is concerned with the application of some of these principles to the particular problem of speech recognition by artificial neurons. The term “neuron” is used without distinction for real neurons—those found in the brain—and for artificial neurons—those made from electronic components. It is possible to construct artificial neurons which are, as far as input- output relations are concerned, complete analogs of their biological counterpart (Mueller, 1958). The networks shown in the figures in this report have been assembled and tested using artificial neurons.