TrustPay Analytics: Real-Time Fraud Detection Using Behavioral Intelligence in UPI Systems

Abstract

The rapid growth of digital payment systems has increased the demand for reliable fraud detection mechanisms to ensure secure financial transactions. Traditionally, fraud detection relied on manual verification, rule-based systems, and basic statistical techniques. These conventional approaches required significant human effort, consumed time, and failed to detect complex fraud patterns in largescale and dynamic transaction data. As digital transactions expanded, these methods became inefficient and lacked real-time decision-making capability. The proposed system presents an intelligent webbased solution for multi-aspect fraud detection in Unified Payments Interface (UPI) transactions using advanced Machine Learning (ML) and Deep Learning (DL) techniques. The system is developed using the Flask web framework and integrates Natural Language Processing (NLP) for handling textual data. Sentence Bidirectional Encoder Representations from Transformers (SBERT) is used for feature extraction, generating contextual embeddings from transaction data. These embeddings are utilized to train multiple classification models, including Gaussian Naive Bayes (GNB), Bernoulli Naive Bayes (BNB), Multinomial Naive Bayes (MNB), and the proposed Histogram-based Gradient Boosting Classifier (HGB). The system performs both binary classification for fraud detection and multi-class classification for transaction types. The proposed HGB model achieves high accuracy with strong Precision, Recall, and F1-Score metrics, outperforming traditional models. The system enables users to upload datasets through a secure login interface and generates real-time predictions with downloadable outputs, ensuring scalability, automation, and improved accuracy in financial fraud detection.