Credit Card Fraud Detection Using State Art of Machine Learning and Deep Learning

Abstract

The proliferation of digital financial transactions has heightened the urgency for robust credit card fraud detection mechanisms. Traditional rule-based systems are increasingly inadequate against sophisticated fraudulent activities. A comprehensive analysis of state-of-the-art machine learning (ML) and deep learning (DL) methodologies applied in credit card fraud detection. In this research, we examine models, including RF, SVM, XGBoost, LSTM, and Autoencoders, to determine their ability to detect fraudulent transactions. A hybrid ensemble framework has been proposed to increase the detection accuracy and reduce the false positives. The performance metrics used for evaluation are accuracy, precision, recall, F1-score, and AUC-ROC. The study presents that the DL models, especially LSTM and Autoencoders, demonstrate better performance in capturing complex fraud patterns. This research contributes to the development of adaptive, real-time fraud detection systems to ensure financial security.