A SELF-ORGANIZING NETWORK FOR MIXED CATEGORY PERCEPTION

A neural network model capable of self-organizing in presence of multi ple or mixed categories is presented. A certainty factor is derived ab out the decision on how well the features (due to single or mixed cate gories) have been interpreted by the network. One part of the model, t he, monitor, controls the performance of the other part, the, categori zer in the self-organization process. The network automatically adjust s the number of nodes in the hidden and output layers, depending on th e nature of overlap between the patterns from different categories. Ma thematical derivations of the bounds on the number of nodes have been presented. The capability of the model is demonstrated experimentally both on one-dimensional binary strings and visual patterns.