Imbalanced data Classification in Credit Card Fraudulent Activities Detection using Multi-Class Neural Network

Credit card fraud detection is an important research in the recent days because each and every day different fraud activities are increasing. The different techniques are introduced to detect and alert the credit card fraudulent activities. The various artificial intelligence, data mining and machine learning techniques are used to classification and prediction of credit card system. The main drawback of the previous methods such as data is imbalanced, accuracy and false positive rate. In this paper, proposed a new deep learning-based algorithm called Multi-Class Neural Network (MCNN) for prediction of credit card detection. The Multi-Class Neural Network techniques address the imbalance data and misclassification issues in face to face and E-commerce transaction. The proposed work consists of three steps such as pre-processing, texture feature extraction and multi-class natural network. The MCNN consists of trainer node, hidden layer, learning rate, learning iteration and random seed for repeating the process. The implementation of proposed work dataset is classified into three ways such as fraud, No Fraud and invalid record. Using the texture feature extraction and MCNN the prediction and classification and accuracy rates are increased in terms of fraud transaction (0), non-fraud transaction (1) and invalid records (2).