Towards Efficient Machine Unlearning Based on Explainable Techniques
- Publication Type:
- Thesis
- Issue Date:
- 2025
Open Access
Copyright Clearance Process
- Recently Added
- In Progress
- Open Access
This item is open access.
Machine learning methods have become a strong driving force in revolutionizing a wide range of applications. However, they are also bringing requests to delete training samples from models due to privacy, usability, or other entitlement requirements. Machine unlearning has emerged as a promising solution to address such deletion requests, and various methods have been explored in recent research. Despite this progress, existing machine unlearning schemes face several critical limitations: how to effectively perform unlearning in federated learning scenarios, how to achieve fine-grained unlearning at the target or feature level, and how to build a robust and trustworthy unlearning verification framework that cannot be easily bypassed. Therefore, this thesis proposes several methods to address these challenges based on explainable techniques.
Please use this identifier to cite or link to this item: