Enhanced Real-Time Detection of UPI Fraud using Advanced Machine Learning Models

M. Ram Kumar 1, T. Anitha 2, P. Harshitha 3, T. Mohan reddy krishna4, Swayam das 5
1,2,3,4,5,6 Computer Science and Information Technology, Siddharth Institute of Engineering & Technology

Abstract-Digital payment systems such as the Unified Payments Interface (UPI) have significantly transformed financial transactions by enabling instant, secure, and convenient money transfers. However, the rapid growth of digital transactions has also increased the risk offraudulent activities. Traditional rule-based fraud detection systems often fail to identify complex and evolving fraud patterns in real time. Therefore, there is a need for intelligent and automated fraud detection systems that can analyze large volumes of transaction data efficiently.This project proposes an advanced fraud detection framework that utilizes machine learning algorithms such as Random Forest, Extra Trees,CatBoost, and LightGBM to detect suspicious UPI transactions. The system analyzes transaction attributes  such as transaction type, amount, and account balance details to identify abnormal patterns. A web-based interface developed using Flask allows users to inputtransaction data and obtain real-time fraud predictions. The proposed model improves fraud detection accuracy and reduces false positives, thereby enhancing the securityand reliability of digital payment systems.