Robert Jay Brown III
1
This paper is also available in postscript
and Adobe portable document format (PDF).
This paper describes an experimental computer program that could serve as one component of a computer vision system. The program is an artificial neural network learning machine implemented as a committee network of threshold logic units to function as a trainable pattern recognizer for visual images composed of a rectangular array of pixels. Each pixel contains a single number representing the gray scale value of that region of the field of view.