Enhancing security in financial transactions: a novel blockchain-based federated learning framework for detecting...

Area of Science:

• Financial Technology (Fintech)
• Cybersecurity
• Data Privacy

Background:

• Fintech leverages technology for efficient financial services, but data breaches pose significant risks.
• Traditional data sharing for machine learning (ML) models compromises customer privacy.
• Protecting sensitive financial data is crucial for maintaining customer trust and regulatory compliance.

Purpose of the Study:

• To propose a novel framework combining federated learning (FL) and blockchain for enhanced financial security.
• To safeguard consumers against fraudulent transactions while preserving data privacy.
• To enable collaborative ML model training without inter-institutional data exchange.

Main Methods:

• Implementation of a federated learning (FL) framework utilizing multiple machine learning (ML) models.
• Integration of blockchain technology to provide a secure and transparent platform for FL.
• Local model training on individual institutions' customer data, followed by aggregation on a central server.

Main Results:

• The proposed framework effectively protects consumers against fraudulent transactions.
• Customer privacy is preserved as no private customer data is exchanged or stored between institutions.
• Blockchain ensures an immutable and auditable record of data exchanges and model training processes.

Conclusions:

• Combining federated learning (FL) with blockchain offers a robust solution for secure and private data collaboration in fintech.
• This approach significantly mitigates the risks associated with data breaches in the financial sector.
• The framework facilitates the development of advanced fraud detection models while upholding stringent data privacy standards.