Using Cascade Generalization and Neural Networks to Select Cryotherapy method for Warts

In this paper, complementary neural networks are applied to cascade generalization. Complementary neural networks comprise two neural networks trained to predict truth and falsity values. Two levels of cascade generalization are implemented in this paper. Two approaches are proposed. First, a neural network is trained in the base level whereas complementary neural networks are trained in meta level of cascade generalization. Second, complementary neural networks are trained in both levels of cascade generalization. The proposed methods are used to select cryotherapy method for wart treatment. The cryotherapy data set is obtained from UCI machine learning repository. Ten-fold cross validation is used in the experiment. The proposed approach gives 98.89% accuracy which higher than the existing methods which are cascade generalization and stacked generalization.