AI-Driven Fraud Detection for UPI Transactions

AI-Driven Fraud Detection for UPI Transactions

T. VIJAYA SREE REDDY  A. VIKAS REDDY  K. VILOCHAN GOUD  E.N. VINATHA REDDY  S. VINAY KUMAR REDDY

School of Engineering, Department of AI&ML, Malla Reddy University, Hyderabad – 500043, India.

Abstract
The rapid adoption of digital payments through the Unified Payments Interface (UPI) has heightened the risk of fraudulent activities, necessitating advanced security measures. This document presents an AI-driven fraud detection system that leverages machine learning

techniques to enhance the security of UPI transactions. Central to the system is the Random Forest Classifier, an ensemble learning method known for its high accuracy, which is augmented by the Synthetic Minority Oversampling Technique (SMOTE) to effectively address issues of

class imbalance in the dataset. The model has been trained on a comprehensive dataset, and its effectiveness is evaluated through key performance metrics such as precision, recall, and

F1-score.

A significant feature of this system is its capability for real-time fraud predictions, which enables users and administrators to swiftly respond to suspicious activities, thereby mitigating potential losses. The adaptability of the system allows it to evolve in response to emerging fraud tactics, significantly minimizing false positives that often disrupt legitimate transactions. By improving fraud detection accuracy through solid AI methodologies, this solution promotes trust and security within digital payment ecosystems—a critical factor as reliance on these technologies

continues to grow exponentially.