Utilizing Blockchain in Privacy Protection and Machine Unlearning
- Publication Type:
- Thesis
- Issue Date:
- 2025
Open Access
Copyright Clearance Process
- Recently Added
- In Progress
- Open Access
This item is open access.
The rapid advancements in blockchain technology, machine learning, and the Internet of Things (IoT) have transformed various fields, including intelligent transportation, collaborative learning, and privacy protection. However, the increasing reliance on decentralized data-driven systems has also raised significant challenges related to privacy, security, and trust, particularly in environments where sensitive information must be shared across multiple parties. This thesis investigates the integration of blockchain technology with federated learning and machine unlearning to address these challenges, providing secure and privacy-preserving mechanisms for data collaboration and model management.
The thesis first examines the role of blockchain in intelligent transportation systems, proposing a dynamic location privacy-preserving scheme that leverages blockchain’s immutability and transparency to protect vehicle location data while facilitating secure communication in transportation networks.
Building on this foundation, the research extends to federated learning and machine unlearning, where blockchain is integrated to enhance security and trust in decentralized model training. A novel framework is introduced for blockchain-enhanced federated learning, enabling multiple organizations to collaborate securely while maintaining data privacy. This framework is further extended to incorporate machine unlearning, allowing for verifiable data removal without compromising model performance or violating privacy regulations.
A key focus of this thesis is on the application of these techniques in large language models (LLMs). A novel framework, Federated TrustChain, is proposed to facilitate secure and auditable LLM training and unlearning within a federated learning setting. By leveraging blockchain, this approach ensures an immutable and verifiable record of model updates and data removal requests, enhancing trust and accountability in distributed LLM training.
Further, this research addresses cross-organizational collaboration in federated learning, where institutions must train shared models without exposing proprietary data. A blockchain-based solution is designed to support secure and efficient data sharing, integrating smart contracts and differential privacy techniques to ensure compliance with data protection regulations while enabling trustworthy federated collaboration.
The proposed frameworks are extensively evaluated through real-world-inspired case studies, including applications in healthcare and education, demonstrating their effectiveness, scalability, and practicality. The results highlight the improvements in privacy protection, security, and collaborative learning efficiency enabled by the integration of blockchain with federated learning and unlearning.
In conclusion, this thesis makes significant contributions to the field of privacy-preserving and secure collaborative learning, establishing a strong foundation for future research and real-world applications. By bridging blockchain, federated learning, and machine unlearning, this work paves the way for more secure, transparent, and privacy-aware intelligent systems.
Please use this identifier to cite or link to this item: