Advancing Banking Systems with Federated Learning and a Fuzzy-Based Blockchain Framework for Secure and Efficient Transactions

Abstract

This paper presents a groundbreaking federated learning-based blockchain framework, Hy-FL, enhanced with fuzzy logic to address critical challenges in service latency, security, and privacy in distributed banking environments. The proposed hybrid approach integrates federated learning to enable secure, decentralized data processing across edge nodes, ensuring privacy compliance by eliminating the need to share raw data. Blockchain technology is employed to maintain an immutable ledger, enhancing the integrity and transparency of sensitive financial transactions. Fuzzy logic adds an adaptive layer to the system, improving decision-making in real-time transaction validations and fraud detection. The framework was experimentally validated within a decentralized banking network, where federated learning enabled partial training of local datasets and secure aggregation into a global model. Blockchain’s proof-of-work mechanism ensured robust transaction security, while fuzzy logic dynamically evaluated and mitigated transaction risks. Results demonstrated a 20% reduction in service latency, a 25% improvement in data accuracy, and a significant boost in fraud detection efficiency and cyber threat mitigation, outperforming traditional banking systems. This novel hybrid framework offers a secure, efficient, and privacy-conscious solution for modern financial services, paving the way for safer, faster, and more reliable banking operations in distributed settings. By addressing key issues such as data privacy, operational efficiency, and scalability, Hy-FL sets a strong foundation for the future of digital banking systems.