An Explainable Hybrid Machine Learning Framework for UPI QR and Collect Request Fraud Detection
International Journal of Interdisciplinary and Multidisciplinary Research · 15.06.2026 · 11 (03) · 159-184
DOI: 10.54121/2021111521 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Interdisciplinary and Multidisciplinary Research, online erschienen am 15.06.2026. Verfasst von R Gayathri und 1 weiteren Autor:innen.
Abstract (Originalsprache)
The Unified Payments Interface (UPI) has emerged as a dominant digital payment platform in India, processing over 10 billion transactions monthly and increasingly attracting fraud attacks such as PAY/RECEIVE manipulation, QR code tampering, and social engineering. Conventional rule-based fraud detection systems are limited in their ability to capture evolving behavioral, contextual, and semantic fraud patterns. This paper presents an explainable hybrid fraud detection framework for UPI QR transactions that performs multi-dimensional analysis of user behavior and transaction context. The proposed framework integrates behavioral decision features, contextual device and call-state indicators, semantic text manipulation signals, and a dynamic QR reputation scoring mechanism that adapts over time. A hybrid architecture combining rule-based screening, Isolation Forest anomaly detection, and XGBoost classification is employed, with SHAP used to provide model interpretability. The framework operates real-time deployment, while delivering transparent and human-interpretable explanations that address trust and accountability requirements in financial fraud detection systems.
Zitieren
APA
Gayathri, R., & D, R. (2026). An Explainable Hybrid Machine Learning Framework for UPI QR and Collect Request Fraud Detection. International Journal of Interdisciplinary and Multidisciplinary Research, 11(03), 159-184. https://doi.org/10.54121/2021111521
Vancouver
Gayathri R, D R. An Explainable Hybrid Machine Learning Framework for UPI QR and Collect Request Fraud Detection. International Journal of Interdisciplinary and Multidisciplinary Research. 2026;11(03):159-184. doi:10.54121/2021111521
BibTeX
@article{Gayathri2026,
author = {Gayathri, R and D, Rejitha},
title = {An Explainable Hybrid Machine Learning Framework for UPI QR and Collect Request Fraud Detection},
journal = {International Journal of Interdisciplinary and Multidisciplinary Research},
year = {2026},
volume = {11},
number = {03},
pages = {159--184},
doi = {10.54121/2021111521}
}
EUVIMED