Federated Learning with Blockchain-Enhanced Machine Unlearning: A Trustworthy Approach

TL;DR

This work addresses the privacy challenge of data deletion in federated learning (FL) for IoT by integrating blockchain to provide verifiable, immutable certification of machine unlearning. The proposed framework uses smart contracts and differential privacy to automate and audit unlearning while maintaining FL efficiency, with a system model that includes distinct training and unlearning clients and an immutable blockchain ledger. A case study in smart healthcare demonstrates secure registration, model management, and precise unlearning, supported by a privacy- and security-centric analysis and a performance evaluation on MNIST and CIFAR-10. Results show targeted unlearning with minimal impact on remaining accuracy, while blockchain overhead remains manageable, indicating practical applicability in privacy-conscious, distributed IoT deployments.

Abstract

With the growing need to comply with privacy regulations and respond to user data deletion requests, integrating machine unlearning into IoT-based federated learning has become imperative. Traditional unlearning methods, however, often lack verifiable mechanisms, leading to challenges in establishing trust. This paper delves into the innovative integration of blockchain technology with federated learning to surmount these obstacles. Blockchain fortifies the unlearning process through its inherent qualities of immutability, transparency, and robust security. It facilitates verifiable certification, harmonizes security with privacy, and sustains system efficiency. We introduce a framework that melds blockchain with federated learning, thereby ensuring an immutable record of unlearning requests and actions. This strategy not only bolsters the trustworthiness and integrity of the federated learning model but also adeptly addresses efficiency and security challenges typical in IoT environments. Our key contributions encompass a certification mechanism for the unlearning process, the enhancement of data security and privacy, and the optimization of data management to ensure system responsiveness in IoT scenarios.

Figures (25)

• Figure 1: Overview and process of our proposed system. (1) Client and agent register. (2) Global model update. (3) Model updating process. (4) Model aggregation process. (5) Machine unlearning in blockchain network
• Figure 2: Specific Client loss comparison in 200 epochs
• Figure 3: Specific Client accuracy comparison in 200 epochs
• Figure 4: Loss comparison of normal learning and with unlearning in 200 epochs
• Figure 5: Testing accuracy comparison of normal learning and with unlearning in 200 epochs