FEDERATED LEARNING FOR CYBER THREAT DETECTION IN DIGITAL BANKING SYSTEMS

Abstract

This research explores Federated Learning for cyber threat detection in digital banking systems using secondary qualitative data analysis. Literature review reveals that FL enables collaborative, privacypreserving model training across distributed banking nodes, enhancing detection of malware, phishing, and transactional anomalies. Key themes identified include applications of FL in cybersecurity, benefits and limitations, and operational implementation strategies. Findings indicate that FL improves privacy, regulatory compliance, and threat intelligence but faces challenges such as data heterogeneity, communication latency, and interpretability. The study emphasizes the importance of adaptive strategies, secure aggregation, and phased deployment to maximize effectiveness and informs future research on empirical validation and cross-institution collaboration.