GRADIENT CRITERION METHOD FOR NEURAL NETWORKS AND APPLICATION TO TARGETED MARKETING

摘要

The present invention is drawn to a unique application of the MaximumLikelihood statistical method to commercial neural network technologies. Thepresent invention utilizes the specific nature of the output in targetmarketing problems and makes it possible to produce more accurate andpredictive results by minimizing a gradient criterion to produce model weightsto get the maximum likelihood result. It is best used on "noisy" data and whenone is interested in determining a distribution's overall accuracy, or bestgeneral description of reality.