Hybrid Intelligent Modeling Technique for Data Classification

Classification is technique of data mining to Predicts the categorical or class of unseen data. It is supervised learning method. In supervised learning, class of each samples are given. It can be separated into binary classification and multiclass classification. In binary classification, two classes are used and in multiclass classification more than two classes are used. The classification of multi-class datasets is more difficult as compared to the binary data classification. In this paper, we present a hybrid technique of GA (Genetic algorithm) and ANN (Artificial Neural Network) for multiclass problem. Genetic algorithm is used to improve the performance of neural network for multiclass data classification. GA optimizes the feature and provides the weight to ANN classifier. Proposed technique classifies IRIS, LYMPHTICS, ZOO, ECOLI and WINE multiclass datasets. To demonstrate the results, all dataset taken from UCI machine learning repository and compared the accuracy, specificity, sensitivity and f-score of proposed algorithm with respect to the standard ANN algorithm.