International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 12 Issue: 04 | Apr 2025
p-ISSN: 2395-0072
www.irjet.net
Machine Learning-Driven Detection of UPI Fraudulent Activities Ayuti Khedekar1, Prof. Ganesh Manerkar2 1Student, Department of Information Technology and Engineering, Goa College of Engineering, Farmagudi, Goa,
India
2Assistant Professor, Department of Information Technology and Engineering, Goa College of Engineering,
Farmagudi, Goa, India ---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - The rapid adoption of Unified Payments Interface (UPI) has revolutionized digital payments in India, offering seamless and instant transactions. However, with increased usage, fraudulent activities like phishing, vishing, fake payments, and QR code scams have surged. Traditional rule-based fraud detection systems struggle to address these sophisticated tactics. This paper proposes a machine learning-based solution that integrates advanced models, including Random Forest, XGBoost, LSTM, and CNN, to detect and classify various fraud types in real-time, enhancing UPI transaction security.
including Random Forest, XGBoost, Long Short-Term Memory (LSTM), and Convolutional Neural Networks (CNN), to detect and classify different types of fraud in UPI transactions. By analyzing factors like transaction amounts, timestamps, user behavior, QR code images, and audio data from vishing attempts, the system can adapt to emerging fraud patterns for suspicious activities. The primary aim is to enhance the security of digital payments and foster trust by offering a fast, automated solution that accurately detects fraud with minimal human involvement. Traditional rule-based systems often fail to keep up with the constantly changing tactics of fraudsters, missing subtle anomalies in transaction patterns.
The system combines a hybrid approach that processes transaction data, QR code images, and audio from vishing attempts to identify fraud patterns. It is trained on a diverse dataset of labeled UPI transactions, QR codes, and voice data, enabling it to detect evolving fraud behaviors. The model’s real-time alert system ensures timely detection, with promising results in accuracy. This approach provides a scalable solution to improve the security of digital payments, with future enhancements aimed at refining detection capabilities and adapting to new fraud tactics.
1.1 OBJECTIVE The main goal of this project is to create an intelligent fraud detection system for UPI transactions that uses machine learning techniques to automatically detect and categorize different types of fraudulent activities. These include vishing (voice phishing), fake payments, impersonation, and counterfeit QR codes. The system is designed to provide real-time fraud detection and classification, ensuring that any suspicious transactions are quickly flagged for further investigation. By utilizing machine learning algorithms such as Random Forest, XGBoost, Long Short-Term Memory (LSTM), and Convolutional Neural Networks (CNN), the system will analyze transaction data, QR code images, and audio recordings to identify fraud with a high level of accuracy.
Key Words: UPI, Fraud Detection, Machine Learning, Convolutional Neural Networks (CNN), Long ShortTerm Memory (LSTM)
1.INTRODUCTION The Unified Payments Interface (UPI) has quickly become a key element of India’s digital payment ecosystem, offering users fast, secure, and affordable ways to make financial transactions. As UPI's popularity has surged, it has revolutionized how individuals and businesses manage daily payments. However, this growth has also attracted fraudsters, leading to an increase in fraudulent activities such as phishing, vishing, fake payment confirmations, and QR code scams. These evolving fraud tactics pose significant challenges to the security and reliability of UPI, affecting both users and financial institutions.
Another important objective of this project is to develop a scalable and adaptive fraud detection framework that can keep up with the ever-changing nature of fraudulent activities. As new fraud patterns emerge, the system will integrate incremental learning, enabling it to continuously update itself with new data without needing to be retrained from scratch. This flexibility ensures that the model remains effective in detecting evolving fraud tactics. Ultimately, the project aims to strengthen the security of UPI transactions, provide valuable insights to financial institutions, and support efforts to maintain trust in digital payment platforms.
This project seeks to address the growing issue of UPI fraud by implementing a machine learning-based fraud detection system. The system integrates multiple models,
© 2025, IRJET
|
Impact Factor value: 8.315
|
ISO 9001:2008 Certified Journal
|
Page 1820