Cloud-Based Federated Learning System for Financial Fraud Detection

Madduri, Rajesh Reddy (2025) Cloud-Based Federated Learning System for Financial Fraud Detection. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (2MB) | Preview
Preview	PDF (Configuration Manual) Download (249kB) | Preview

Abstract

This paper introduces a Docker-based federated learning model to conduct financial fraud detection without compromising data privacy at the expense of effective detection. As opposed to conventional centralized systems which imply risks in terms of both privacy and regulation, the suggested solution lies in the containerized edge computing with federated learning, in which financial institutions can join efforts to comprehend fraudulent activity without exchanging raw data on transactions. The orchestration applied by the system is Docker Compose with 3 edge client containers each running 12,000 transactions locally and participating in a global fitting model via secure aggregation and d-privacy (d=1.2). The combination of SHAP and LIME delivers explainable AI on the edge in real-time, by making interpretable decisions in under 300 milliseconds per transaction. Experiments prove that exactly 93.34 per cent of recalls were obtained in fraud detection with 82.06 per cent AUC-ROC focused on capturing fraud rather than reducing false positive. The implementation is able to maintain (1.20, 1e-05 )-differential privacy guarantees and within four federated learning rounds, the client participation achieved is 100 percent. The proposed integration scheme S3 AWS allows distributing models without any centralization of the data, confirming the benefits of using privacy-preserving federated learning in situations of compliance regulatory control and operational needs of detecting financial fraud.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Gupta, Shaguna UNSPECIFIED
Subjects:	H Social Sciences > HG Finance T Technology > T Technology (General) > Information Technology > Cloud computing Q Science > QA Mathematics > Computer software > Computer Security T Technology > T Technology (General) > Information Technology > Computer software > Computer Security H Social Sciences > HV Social pathology. Social and public welfare > Criminology > Crimes and Offences > Cyber Crime Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Cloud Computing
Depositing User:	Ciara O'Brien
Date Deposited:	30 Mar 2026 10:01
Last Modified:	30 Mar 2026 10:01
URI:	https://norma.ncirl.ie/id/eprint/9241

Actions (login required)

View Item