Authors : Hemant Singh; Shree Bejon Sarkar Bappy

Volume/Issue : Volume 10 - 2025, Issue 11 - November

Google Scholar : https://tinyurl.com/mn635nwb

Scribd : https://tinyurl.com/2wfmfap3

DOI : https://doi.org/10.38124/ijisrt/25nov856

Note : A published paper may take 4-5 working days from the publication date to appear in PlumX Metrics, Semantic Scholar, and ResearchGate.

Abstract : Financial fraud remains a major issue for modern banks and financial institutions, leading to billions of dollars in losses each year. Existing fraud detection systems, which depend on rule-based methods or static machine learning models, often have difficulty keeping up with changing fraud tactics, protecting data privacy, and offering clear explanations for their decisions. This paper introduces a new AI-based approach for detecting financial fraud that combines graph neural networks (GNNs) with sequential deep learning models. This combination helps in understanding both the relationships between entities and the timing of transactions. To handle privacy issues, the system uses federated learning, allowing different financial institutions to work together on training models without sharing sensitive data. An explainable AI (XAI) component is also included to provide clear and understandable reasons for fraud alerts, which helps with meeting regulatory requirements and building user confidence. The model is tested on standard datasets and simulated fraud scenarios, showing better performance in terms of accuracy, resilience against changes in fraud patterns, fewer false alarms, and cost efficiency compared to traditional methods. This study offers a scalable, transparent, and privacy-focused solution for real-time fraud detection within financial systems.

Keywords : Financial Fraud Detection, Artificial Intelligence (AI), Graph Neural Networks (GNNs), Federated Learning, Explainable AI (XAI), Privacy-Preserving Machine Learning, Transaction Anomaly Detection, Cost-Sensitive Learning, Real-Time Fraud Detection.

References :

1. R. J. Bolton and D. J. Hand wrote an article called "Statistical fraud detection: A review" which was published in the journal Statistical Science, volume 17, issue 3, pages 235 to 249 in the year 2002.
2. A. C. Bahnsen, D. Aouada, and B. Ottersten wrote an article titled "Example-dependent cost-sensitive logistic regression for credit card fraud detection" which appeared in Expert Systems with Applications in 2016.
3. J. Jurgovsky and others wrote an article named "Sequence classification for credit-card fraud detection" which was published in Expert Systems with Applications in 2018.
4. A. Dal Pozzolo and others wrote "Credit card fraud detection: A realistic modeling and a novel learning strategy" which was published in the IEEE Transactions on Neural Networks and Learning Systems in 2015.
5. S. Hochreiter and J. Schmidhuber wrote an article titled "Long short-term memory" which was published in Neural Computation, volume 9, issue 8, pages 1735 to 1780 in 1997.
6. J. Wang and others wrote an article called "Fraud detection with graph neural networks" which was presented at the 28th ACM International Conference on Information and Knowledge Management (CIKM) in 2019.
7. J. Konečný and others wrote "Federated learning: Strategies for improving communication efficiency" which was presented at the NIPS Workshop on Private Multi-Party Machine Learning in 2016.
8. S. M. Lundberg and S. I. Lee wrote an article titled "A unified approach to interpreting model predictions" which was presented at the Advances in Neural Information Processing Systems (NeurIPS) conference in 2017.
9. R. Guidotti and others wrote a survey titled "A survey of methods for explaining black box models" which was published in ACM Computing Surveys, volume 51, issue 5, pages 1 to 42 in 2018.
10. T. Chen and C. Guestrin wrote an article called "XGBoost: A scalable tree boosting system" which was presented at the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD) in 2016, pages 785 to 794.
11. D. P. Kingma and M. Welling wrote an article titled "Auto-encoding variational Bayes" which was presented at the 2nd International Conference on Learning Representations (ICLR) in 2014.
12. A. Vaswani and others wrote an article titled "Attention is all you need" which was presented at the Advances in Neural Information Processing Systems (NeurIPS) conference in 2017, pages 5998 to 6008.
13. J. Wang, X. Cui, Y. Zhang, Z. Pei, W. Zhu, and S. Yang wrote an article called "Fraud detection with graph neural networks" which was presented at the 28th ACM International Conference on Information and Knowledge Management (CIKM) in 2019, pages 2265 to 2273.
14. J. Konečný and others wrote "Federated learning: Strategies for improving communication efficiency" which was presented at the NIPS Workshop on Private Multi-Party Machine Learning in 2016.
15. S. M. Lundberg and S. I. Lee wrote "A unified approach to interpreting model predictions" which was presented at the Advances in Neural Information Processing Systems (NeurIPS) conference in 2017.
16. J. Wang, X. Cui, Y. Zhang, Z. Pei, W. Zhu, and S. Yang wrote "Fraud detection with graph neural networks" which was presented at the 28th ACM International Conference on Information and Knowledge Management (CIKM) in 2019, pages 2265 to 2273.
17. D. P. Kingma and M. Welling wrote "Auto-encoding variational Bayes" which was presented at the 2nd International Conference on Learning Representations (ICLR) in 2014.
18. A. Vaswani and others wrote "Attention is all you need" which was presented at the Advances in Neural Information Processing Systems (NeurIPS) conference in 2017, pages 5998 to 6008.
19. R. Guidotti and others wrote "A survey of methods for explaining black box models" which was published in ACM Computing Surveys, volume 51, issue 5, pages 1 to 42 in 2018.
20. T. Chen and C. Guestrin wrote "XGBoost: A scalable tree boosting system" which was presented at the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD) in 2016, pages 785 to 794.
21. A. Dal Pozzolo, O. Caelen, R. Johnson, and G. Bontempi wrote an article titled "Calibrating probability with undersampling for unbalanced classification" which was presented at the IEEE Symposium Series on Computational Intelligence in 2015, pages 159 to 166.
22. F. Fiore, A. De Santis, F. Perla, P. Zanetti, and F. Palmieri wrote an article titled "Using generative adversarial networks for improving classification effectiveness in credit card fraud detection" which was published in Information Sciences, volume 479, pages 448 to 455 in 2019.
23. B. A. Amini, R. S. Jalili, and H. R. Rabiee wrote an article called "Cost-sensitive learning for credit card fraud detection: A comparative study" which was published in Expert Systems with Applications, volume 134, pages 249 to 262 in 2019.
24. S. K. Jha and P. B. Patil wrote an article titled "Adaptive anomaly detection for credit card fraud using deep learning" which was published in IEEE Access, volume 8, pages 192343 to 192355 in 2020.
25. X. Zhang, X. Li, and Y. Wang wrote an article called "Graph-based credit card fraud detection: A comprehensive review" which was published in IEEE Transactions on Computational Social Systems, volume 7, issue 5, pages 1134 to 1146 in 2020.
26. S. Nakamoto wrote a white paper titled "Bitcoin: A peer-to-peer electronic cash system" in 2008.
27. Y. Li, J. Wang, and X. Zhang wrote an article titled "Continual learning approaches for fraud detection in dynamic financial environments" which was published in IEEE Transactions on Knowledge and Data Engineering, volume 34, issue 6, pages 2847 to 2858 in 2022.