Enhancing Fraud Detection in Banking With Deep Learning: Graph Neural Networks and Autoencoders for Real-Time Credit Card Fraud Prevention

Abstract

Under the umbrella of artificial intelligence (AI), deep learning enables systems to cluster data and provide incredibly accurate results. This study explores deep learning for fraud detection, utilizing Graph Neural Networks (GNNs) and Autoencoders to enhance business practices and reduce fraudulent activities in large organizations. For real-time fraud detection, we propose Graph neural network with lambda architecture while for credit card fraud detection, we use an autoencoder, validated through case studies from two banks. The findings demonstrate that these methods effectively detect fraud with balance of precision and recall, improving the efficiency of banking systems. Python is employed for analysis, emphasizing the ability of deep learning to manage and prevent fraud in real-time on dynamic datasets. In the end, this study concludes that by using deep learning algorithms, we can control online credit card fraud detection in banks, improve the efficiency of the banking system. We can manage fraudulent activity in real-time and on dynamic datasets by utilizing deep learning algorithms, which allows for ongoing improvement of the fraud detection and prevention system.

Keywords

• Deep learning
• credit card
• fraud detection
• graph neural network
• autoencoders