Hybrid Artmap Neural Networks (hart)

In order to reduce the effect of the category proliferation phenomenon in Fuzzy ARTMAP (FAM) and in ellipsoidal ARTMAP (EAM) architectures, The genetic algorithms were used to evolve networks of both architectures called GFAM and GEAM [3][4]. The results were very promising and the category proliferation (CP) phenomenon was minimized in most of the experiments, however, the author noticed that GEAM worked better on some problems while GFAM worked better on others, this triggered the idea of a hybrid Genetic ART architecture that uses categories from both EAM and FAM architectures, this architecture was then called HART. HART evolves networks that were designed to have hyper-rectangular and hyper-ellipsoidal categories. HART was tested on 24 different datasets, the results were compared against those collected from testing FAM, EAM, GFAM and GEAM, HART performed well against all other networks and gave a great balance between accuracy and network size.