Loading article...
Loading article...
Stream/Area:
-
Subject:
-
Paper ID:
IJSREM68688
Volume:
10 |
11 |
Nov 2026
Pages:
1-3
Publication Frequency:
Monthly (12 issues per year)
License:International Journal of Scientific Research in Engineering and Management follows the Creative Commons Attribution 4.0 (CC BY 4.0) License, allowing free use and distribution with proper attribution.
Master of Computer Application
Jaihind Institute of Management and Research, Kuran, Pune
Master of Computer Application
Jaihind Institute of Management and Research, Kuran, Pune
Rapid expansion across 24/7 digital settlement networks has introduced formidable financial hazards, leaving banking rails vulnerable to sophisticated card-absent swindles, credential hijacking, and fabricated synthetic profiles. Legacy static conditional filters struggle under severe operational friction, triggering unmanageable false alarms while failing to intercept shifting illicit stratagems. In this study, we propose and validate an industrial-grade Machine Learning Based Fraud Detection System architected to satisfy rigorous sub-50 ms clearing gateway service agreements. Addressing the acute distributional distortion of payment logs—where fraudulent operations constitute fewer than two occurrences per thousand events—we institute a coordinated rebalancing scheme coupling Borderline-SMOTE targeted synthesis with Tomek Links boundary pruning, optimized under an asymmetric focal penalty function. Our multi-tiered decision framework merges compiled gradient tree ensembles with unsupervised reconstruction error thresholds, tested rigorously over 284,807 European cardholder logs alongside the IEEE-CIS benchmark archive. The experimental pipeline delivers an Area Under the Receiver Operating Characteristic Curve (ROC-AUC) of 0.984 and a Precision-Recall AUC (PR-AUC) of 0.886, capturing 90.82% of illegitimate transfers while suppressing false customer flags to 0.061%. Simulated high-volume transaction stress trials up to 10,000 TPS confirm a robust P99 roundtrip latency of 24.8 ms. Finally, local feature attributions via TreeSHAP ensure compliance with Reserve Bank of India (RBI) risk oversight standards, PCI-DSS requirements, and national data privacy mandates.
KEYWORDS
This investigation designed, implemented, and empirically benchmarked an industrial-grade Machine Learning Based Fraud Detection System. By uniting targeted Borderline-SMOTE synthesis, Tomek Links topological refinement, and an asymmetric focal objective, our framework overcomes the persistent trade-off between intercepting rare fraud events and minimizing legitimate customer friction. Empirical tests across benchmark card records prove high discriminative precision (0.984 ROC-AUC, 0.886 PR-AUC) while sustaining 10,000 TPS throughput at 24.8 ms P99 turnaround. Future development directions will explore: (1) privacy-preserving Federated Learning across banking consortiums to collaboratively train anti-fraud models without centralizing confidential cardholder records; and (2) Self-Supervised Graph Transformers to uncover distributed mule account syndicates across interbank networks in real time.
[1] Nilson Report, "Global Payment Card Fraud Losses Reach $32 Billion," Issue 1209, pp. 1-9, 2024. [2] F. Carcillo, Y. A. Le Borgne, O. Caelen, and G. Bontempi, "Streaming active learning, chickening out, and fraud detection," Information Sciences, vol. 547, pp. 248-264, 2021. [3] A. Dal Pozzolo, O. Caelen, R. A. Johnson, and G. Bontempi, "Calibrating probability with undersampling for unbalanced classification," in IEEE SSCI, 2015, pp. 159-166. [4] S. Bhattacharyya, S. Jha, K. Tharakunnel, and J. C. Westland, "Data mining for credit card fraud: A comparative study," Decision Support Systems, vol. 50, no. 3, pp. 602-613, 2018. [5] Z. Wang, et al., "Semi-supervised graph neural networks for financial fraud detection," in Proc. IEEE ICDM, 2022, pp. 598-607. [6] F. Carcillo, et al., "Scarff: a framework for addressing drift and class imbalance in fraud detection," Knowledge-Based Systems, vol. 188, p. 104865, 2020. [7] IEEE Computational Intelligence Society, "IEEE-CIS Fraud Detection Benchmark Dataset," Kaggle Competitions, Tech. Rep., 2019. [8] S. M. Lundberg and S. I. Lee, "A unified approach to interpreting model predictions," in NeurIPS, vol. 30, 2017, pp. 4765-4774. [9] Reserve Bank of India (RBI), "Master Direction on Digital Payment Security Controls," RBI/2020-21/74, Tech. Rep., 2021. [10] PCI Security Standards Council, "Payment Card Industry Data Security Standard (PCI-DSS) v4.0," Tech. Rep., 2022.
Publication Date
11/5/2026
Peer Review Type
Double-Blind Peer Review
Review Rounds: 2
Documents
Article Views
1
Total views since publication
PDF Downloads
5
Total downloads since publication
Geographic Reach
5
Digital Preservation: This article is permanently archived on Zenodo and complies with all data preservation standards across major indexing platforms.
Plagiarism Screening
This manuscript has been checked for plagiarism using TurnitinPlagiarismChecker software.
Conflict of Interest
The authors declare no conflict of interest related to this research work.
Funding Information
No funding information available for this research.
Related research in the same subject area
Loading similar articles...