Securing federated learning with blockchain: a systematic literature review

Area of Science:

• Computer Science
• Artificial Intelligence
• Cybersecurity

Background:

• Federated learning (FL) enables collaborative model training without sharing raw data, preserving user privacy.
• Traditional FL systems encounter challenges like data privacy disclosure, unreliable parameter uploads, and high communication costs.
• Decentralized technologies, particularly blockchain, present a potential solution to these inherent FL limitations.

Purpose of the Study:

• To conduct a systematic literature review on the integration of blockchain technology within federated learning frameworks.
• To analyze existing federated learning problems addressable by blockchain integration.
• To explore the characteristics and advancements of blockchain-based federated learning approaches.

Main Methods:

• Systematic literature review methodology was employed.
• Careful screening of relevant studies was performed to identify key research contributions.
• Analysis focused on security and privacy attacks in FL, blockchain solutions, and blockchain-based FL characteristics.

Main Results:

• Blockchain integration can effectively address security and privacy vulnerabilities in federated learning.
• Blockchain-based FL enhances system performance by eliminating the need for a centralized server.
• Key aspects studied include security, privacy, record-keeping, rewards, verification, and accountability in blockchain-FL systems.

Conclusions:

• The integration of blockchain technology offers a robust solution to enhance the security, privacy, and reliability of federated learning.
• Further research is needed to address open issues and guide the development of advanced blockchain-based FL systems.
• Blockchain-based FL presents a promising direction for future decentralized machine learning applications.