Real Time Fraud Detection in Digital Bank Payments

Abstract

The exponential growth of digital banking, mobile wallets, and online transaction platforms has revolutionized the financial ecosystem by enabling fast, secure, and convenient cashless transactions. However, this rapid digitalization has also significantly increased the risk of financial fraud, including unauthorized transactions, identity theft, phishing attacks, and account takeovers. Traditional fraud detection systems rely heavily on static rule-based mechanisms and batch processing techniques, which are incapable of identifying complex and evolving fraud patterns in real time. This paper proposes an intelligent real-time fraud detection system that leverages Artificial Intelligence (AI) and Machine Learning (ML) techniques to analyze transaction data dynamically and detect fraudulent activities instantly. The system integrates advanced algorithms such as Random Forest for classification, Isolation Forest for anomaly detection, and Long Short-Term Memory (LSTM) networks for sequential pattern recognition. Additionally, streaming technologies are utilized to process high-velocity transaction data with minimal latency. The proposed system is evaluated using standard performance metrics such as accuracy, precision, recall, F1-score, and ROC-AUC. Experimental results demonstrate that the system achieves high detection accuracy while significantly reducing false positives and response time. This research contributes to enhancing financial security, improving customer trust, and ensuring the robustness of digital payment systems in real-world environments. The proposed model achieved 98.2% accuracy with reduced false positive rate and low transaction processing latency