Online Fraudulent Transaction Detection Through Machine Learning and Deep Learning Algorithms

Abstract: The virtual world has led to a rise in credit card use in the modern era, but misuse and fraud of credit cards have also increased dramatically. It is necessary to identify the many kinds of credit card fraud. Such frauds cause significant financial losses for both the business and the cardholder. Determining whether or not a specific transaction is fraudulent is the primary goal. A high false alarm rate, a shift in the nature of fraud, access to public data, and a large class imbalance are all necessary for detecting fraud. It acknowledges the challenges posed by imbalanced data and explores a range of machine learning and deep learning algorithms. The study focuses on convolutional neural networks (CNNs) and their architectural variations to enhance fraud detection. Through empirical analysis, it achieves impressive results, outperforming existing methods with high accuracy, F1-score, precision, and AUC values. The research also emphasizes the importance of minimizing false negatives. Ultimately, the proposed deep learning model offers a promising solution for real-world credit card fraud detection.

Keywords: credit card fraud, machine learning, deep learning, CNN