Unlearning with Fisher Masking
This addresses the need for effective data revocation in machine learning models for users and developers, offering a novel approach with incremental improvements over existing methods.
The paper tackles the problem of machine unlearning by proposing a Fisher masking strategy to improve data removal and retain performance on remaining data, achieving almost complete unlearning without fine-tuning and maintaining most performance across various datasets and networks.
Machine unlearning aims to revoke some training data after learning in response to requests from users, model developers, and administrators. Most previous methods are based on direct fine-tuning, which may neither remove data completely nor retain full performances on the remain data. In this work, we find that, by first masking some important parameters before fine-tuning, the performances of unlearning could be significantly improved. We propose a new masking strategy tailored to unlearning based on Fisher information. Experiments on various datasets and network structures show the effectiveness of the method: without any fine-tuning, the proposed Fisher masking could unlearn almost completely while maintaining most of the performance on the remain data. It also exhibits stronger stability compared to other unlearning baselines