Using the Hopfield neural network as a classifier by storing class representatives

The main contribution of this report is the suggestion, development and trial of another way for using the Hopfield network for classification. Rather than storing individual members of the training sets, a method for storing representatives of the sets is considered. Representatives of sets are stored by calculating a connection matrix such that all the elements in a training set are attracted to members of the same training set. An arbitrary element is then classified by the class of its attractor if the attractor is a member of one of the original training sets. The elements themselves are not required to be attractors. The method is successfully applied to artificially generated classes and to the classification of cervical cells for cancer detection. The algorithms are coded in a language called APL for rapid prototyping. The language limits the size of the networks, however, when implemented on a microcomputer. Given these constraints the execution times were nevertheless small due to the few number of epochs required to store the sets