Application of Synergistic Neural Networks in Data Classification
Main Article Content
Abstract
Classification is one of many tasks in Artificial Intelligence. Synergistic approach toclassification by neural networks is a recent method which can be used to improve the accuracy inclassification. This research work implemented Synergistic Neural Networks to classify Heart DiseaseData from Cleveland Data Base. It adopted two known approaches to synergistic neural networks,namely summation and selection approaches. Five different types of neural networks had beenselected in the synergy namely Multilayer Perceptron, Generallized Feedforward Networks, ModularNetworks, Radial Basis Function Networks, and Jordan networks. Comparison was made amongthe best results from individual neural networks, from summation-typed synergistic neural networksand from selection-typed synergistic neural networks.
The result from the comparison revealed that the selection-typed synergistic neural networks(maximum method) yielded the best accuracy. This work demonstrated the ability of synergisticneural networks in improving accuracy and reduces ambiguity in classification, especially whennumber of samples in training data set is limited.
Keywords : Classification / Jordan Network / Generallized Feedforward Network /Synergistic Neural Networks / Neural Networks / Modular Network /Multilayer Perceptron / Radial Basis Function Network /Cleveland Data Base / Heart Disease Data