Hybrid Machine Learning Model for UPI QR Code Fraud Detection and Secure Online Payment Authentication

Abstract

With the rise of Unified Payments Interface (UPI), QR code-based payments and digital banking services, the online payment scenario has undergone a sea change, making monetary transactions fast, simple and cashless. However, digital transformation has also fuelled cyber frauds such as fake QR codes, phishing, identity theft, account takeovers, social engineering and payment frauds. Current fraud detection methods usually rely on fixed rule-based systems that cannot adapt to changing fraud trends and often generate high false positive rates, thereby impacting transaction safety and user experience. To address these issues, this paper proposes a Hybrid Machine Learning Model for UPI QR Code Fraud Detection and Secure Online Payment Authentication by combining Convolutional Neural Networks (CNN), Bidirectional Long Short-Term Memory (BiLSTM) and Extreme Gradient Boosting (XGBoost). The proposed framework uses extensive information pre-processing, feature engineering and feature selection techniques for improving the quality of transaction data before classification. CNN is used to extract transaction features. BiLSTM is used to capture sequential behavioural patterns and temporal dependencies. Deep feature representations are fused to classify fraud accurately by XGBoost. The system also has an adaptive multi-factor authentication mechanism based on the predicted fraud probability, which involves OTP verification, biometric authentication, PIN validation and trusted device verification to improve the security of the transaction. Experimental results show that the proposed hybrid model outperforms traditional machine learning and deep learning models on multiple evaluation metrics, including Balanced Accuracy, Matthews Correlation Coefficient (MCC), Cohen’s Kappa Score, Log Loss and Average Precision (AP). The results validate the proposed approach to improve the accuracy of fraud detection, reduce false alarms, increase the reliability of authentication and provide a scalable, intelligent and secure framework to protect UPI QR code transactions from evolving cyber risks. The proposed system combines advanced artificial intelligence techniques and adaptive security mechanisms for real-time fraud detection to build a reliable digital payment ecosystem.