Author:
A.Arockia Helen Sushma, S.Madhu Bala, J.Margrate Sneka, N.Pavithra, M.Pooja
Published in
Journal of Science Technology and Research
( Volume 6, Issue 1 )
Page No: 1 - 8
Volume 6, Issue 1
Article Type: Scholarly Article
Published Date: May, 2025
Published by: Journal of Science Technology and Research (JSTAR)
Abstract
This project aims to develop a real-time fraud detection system for UPI transactions using Machine Learning, specifically SMOTE and XGBoost. Since fraudulent transactions are rare compared to legitimate ones, SMOTE is used to balance the dataset by generating synthetic fraud samples, improving model accuracy. XGBoost, a powerful gradient boosting algorithm, is implemented to classify transactions as fraudulent or legitimate based on patterns in user behavior and transaction details. The system is designed to detect fraud instantly, helping prevent unauthorized payments by triggering alerts or blocking suspicious transactions. By continuously learning from new fraud patterns, this approach enhances the security of digital payments and reduces financial fraud risks effectively.
Keywords
UPI, Financial Fraud, SMOTE, XGBoost, Machine Learning, Real-Time Detection, Data Imbalance
References

1. Yash Patil et al., UPI Fraud Detection Using Machine Learning, IJSREM, 2024.
2. S. Jagadeesan et al., UPI Fraud Detection Using ML, IJARCCE, 2024.
3. Miss Sayalee S. Bodade, Review on UPI Fraud Detection, IJNRD, 2023.
4. Vitthal B. Kamble et al., Enhancing UPI Fraud Detection, IJRASET, 2025.
5. G. D. Clifford et al., “AF classification from a short single lead ECG recording: the PhysioNet/Computing in Cardiology Challenge 2017,” in 2017 Computing in Cardiology (CinC), Rennes, France, pp. 1–4.
6. O. Yildirim, U. B. Baloglu, R. S. Tan, E. J. Ciaccio, and U. R. Acharya, “A new approach for arrhythmia classification using deep coded features and LSTM networks,” Computer Methods and Programs in Biomedicine, vol. 176, pp. 121–133, 2019.
7. G. B. Moody and R. G. Mark, “The impact of the MIT-BIH Arrhythmia Database,” IEEE Eng. Med. Biol. Mag., vol. 20, no. 3, pp. 45–50, 2001.
8. A. Y. Hannun et al., “Cardiologist-level arrhythmia detection and classification in ambulatory electrocardiograms using a deep neural network,” Nature Medicine, vol. 25, no. 1, pp. 65–69, 2019.
9. J. Xie, R. Zhu, Y. Xu, and Y. Zhou, “A robust arrhythmia classification method based on LSTM and wavelet denoising,” IEEE Access, vol. 7, pp. 184001–184012, 2019.
10. K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), Las Vegas, NV, USA, pp. 770–778, 2016.
11. A. A. Alsharabi, H. Wang, and L. Ye, “A deep learning-based approach for ECG signal denoising and arrhythmia classification,” Computer Methods and Programs in Biomedicine, vol. 213, p. 106523, 2022.
12. R. K. Tripathy, U. R. Acharya, and D. Bhattacharyya, “Use of features from RR-time series and ECG signal for automated classification of cardiac arrhythmias using decision tree,” Journal of Medical Systems, vol. 41, no. 11, pp. 1–13, 2017.
13. D. Banerjee, H. E. Michelis, and R. R. Khanduja, “Real-time arrhythmia detection using machine learning and cloud computing,” in 2020 IEEE 44th Annual Computers, Software, and Applications Conference (COMPSAC), pp. 690–695.
14. H. Chen, J. Wu, and X. Lin, “ECG heartbeat classification based on ensemble deep learning approach,” in 2019 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), pp. 2715–2719.
15. Z. Zhao, M. Zhang, and Y. Zhou, “ECG feature extraction and classification for arrhythmia detection using wavelet transform and hybrid neural networks,” IEEE Access, vol. 7, pp. 104078–104088, 2019.
16. P. de Chazal and R. B. Reilly, “A patient-adapting heartbeat classifier using ECG morphology and heartbeat interval features,” IEEE Transactions on Biomedical Engineering, vol. 53, no. 12, pp. 2535–2543, 2006.

ML-POWERED UPI FRAUD

ML-POWERED UPI FRAUD XGBoost (Extreme Gradient Boosting) is chosen for its efficiency, scalability, and ability to
handle large datasets while preventing overfitting. The model is designed to analyze
transactional patterns, detect anomalies, and classify transactions as either legitimate or
fraudulent in real time.
Upon detecting fraudulent transactions, the system can trigger alerts, block unauthorized
payments, and apply additional verification measures to prevent losses.
II. LITERATURE SURVEY
Existing systems rely on static rules and basic verification methods, which result in high false
positives and negatives. Recent studies suggest leveraging algorithms like Random Forest, SVM,
and Neural Networks can provide more adaptable fraud detection. These studies also highlight
the importance of addressing class imbalance in financial datasets.

ML-POWERED UPI FRAUD


III. PROBLEM DEFINITION
Rule-based fraud detection systems lack adaptability, generate high false alarm rates, and
cannot detect unseen fraud types. Moreover, the scarcity of fraud examples in datasets limits
traditional models from learning complex fraudulent behavior patterns.
IV. PROPOSED SYSTEM
The proposed model uses:
SMOTE (Synthetic Minority Over-sampling Technique): to artificially balance the training data
by generating synthetic samples of the minority class (fraud).
XGBoost (Extreme Gradient Boosting): to accurately classify transactions by learning from
transaction patterns and user behavior.
Additional algorithms like Random Forest and SVM are considered for comparative
performance evaluation.

ML-POWERED UPI FRAUD

Download

Indexed In