SecurePayFL: Collaborative Intelligence Framework for Cross-Border Fraud Detection Through Privacy-Preserving Federated Learning (Published)
This article presents SecurePayFL, a privacy-preserving federated learning framework designed to enable collaborative fraud detection in financial institutions without compromising sensitive customer data. The article addresses the fundamental challenge that while collaborative data sharing significantly enhances fraud detection capabilities, it risks violating stringent data protection regulations such as GDPR and CCPA. SecurePayFL implements sophisticated cryptographic protocols including homomorphic encryption and differential privacy techniques to secure model updates while maintaining regulatory compliance. Through a comprehensive evaluation involving fifteen financial institutions across seven Asian countries, the framework demonstrates substantial improvements in fraud detection accuracy, particularly for cross-border fraud patterns, while maintaining strict data sovereignty. The article details the architecture, implementation methodology, performance analysis, and regulatory considerations of this novel approach, establishing a new paradigm for secure financial intelligence sharing that balances effective fraud detection with robust privacy protection.
Keywords: collaborative intelligence, cross-border security, federated learning, financial fraud detection, privacy-preserving machine learning