Data processing system

Claim 1

A network for detecting a pattern of image pixel densities comprising: i) a first layer of neurons receiving image pixel densities as input signals, each neuron of the first layer comprising an electronic apparatus producing an affirmative output signal when a sum of input signals exceeds a threshold; ii) a second layer of neurons receiving outputs from neurons of the first neural layer as inputs, each neuron of the second layer comprising an electronic apparatus producing an affirmative output signal when a sum of adaptably weighted input signals exceeds a threshold; and iii) at least one subsequent layer of neurons receiving outputs from neurons of a preceding layer as inputs, each neuron of each subsequent layer comprising an electronic apparatus producing an affirmative output signal when a sum of adaptably weighted input signals exceeds a threshold; wherein the number of layers equals an abstraction degree of the pattern of image pixel densities, when: the pattern of image pixel densities is characterized as terms of pixel values combined by operations including AND, OR and equivalents thereof; a combination of pixel values by AND and OR operations is characterized as an intermediate term having an abstraction level of one; the combination of first and second intermediate terms by AND and OR operations is characterized as a third intermediate term having an abstraction level which is one greater than the highest abstraction level of the first and second intermediate terms; and the abstraction level of the pattern of pixel densities is the abstraction level of a term characterizing the pattern.