Machine learningML & AI Methods

Towards Scalable Exact Machine Unlearning Using Parameter-Efficient Fine-Tuning

The article presents Sequence-aware Sharded Sliced Training (S3T), a new framework for machine unlearning that improves system deletion capabilities with minimal impact on model performance, proving more effective than other methods.