Credit Card Fraud Transaction Detection System Using Neural Network-Based Sequence Classification Technique

Abstract

The movement towards digital era introduces centralization of information, web services, applications, and devices. The fraudster keeps an eye over ongoing transaction and forges data by using different techniques as traffic monitoring, session hijacking, phishing, and network bottleneck. In this study, the authors design a framework using deep learning algorithm to suspect the fraudulence transaction and evaluate the performance of the proposed system by updating data repositories. The neural network-based sequence classification technique is used for fraud detection of credit card transactions by including threshold value to measure the deviation of transaction. The reconstruction error (MSE) and predefined threshold value of 4.9 is used for determination of fraudulent transactions.