Genetic Algorithm Based Over-Sampling with DNN in Classifying the Imbalanced Data Distribution Problem

Objective: Data imbalance exists in many real-life applications. In the imbalanced datasets, minority class data creates a wrong inference during classification that leads to more misclassification. More research has been done past solve this issue, but as of now there is no global working solution found do efficient classification. After analyzing various existing literatures, it proposed minimize misclassification through genetic based oversampling and deep neural network (DNN) classifier. Method: method synthetic samples are generated on algorithm. Initial populations for algorithm using Gaussian weight initialization technique fittest individual from population selected by Euclidean distance further processing generate double size dataset classified with DNN. Findings: The performance oversampled training DNN Classifier compared C4.5 Support Vector Machine (SVM) classifiers classifier outperforms other two classifiers. SMOTE ADASYN considered comparison. It approach approaches. also proved experiment reduced statistically significant comparatively better. Novelty: generation initialization, sample selection measure, reduce novelty work. Keywords: Genetic algorithm; Gauss initialization; SMOTE; ADASYN; Imbalanced data; Classification