From Click to Checkout: Deep Learning for Real-Time Fraud Detection in E-Payment Systems

Abstract

The rapid expansion of e-commerce has been paralleled by a significant increase in electronic payment (e-payment) transactions, bringing forth pressing challenges in maintaining transactional security. This paper addresses the critical issue of e-payment fraud in e-commerce by leveraging deep learning techniques for real-time fraud detection. With the growing sophistication of fraudulent activities, traditional rule-based fraud detection systems are proving inadequate, necessitating more advanced and adaptable solutions. This study proposes a deep learning model, specifically designed to enhance e-payment security by efficiently identifying fraudulent transactions. The model addresses key challenges such as class imbalance in transaction data and the need for real-time processing capabilities. Through a comprehensive methodology involving data preprocessing, model architecture design, training, and evaluation, the paper demonstrates the effectiveness of deep learning in detecting complex fraud patterns with high accuracy. The findings highlight the potential of deep learning to significantly improve the security of e-payment systems in e-commerce, thereby bolstering consumer trust and the overall integrity of online transactions. This research contributes to the evolving landscape of e-commerce security, offering insights and directions for future advancements in fraud detection technologies.