ConceptGradient-Based Machine Unlearning
Gradient-Based Machine Unlearning
A class of machine unlearning methods that adjust model parameters through gradient optimization to forget specific data while preserving the influence of the remaining data. It is commonly applied in privacy protection and data deletion scenarios.
Timeline (last 90 days)
Oct 30
Gradient-based machine unlearning methods typically achieve unlearning by optimizing a forget loss and a retain loss, but similar loss values do not necessarily imply similar output-space behavior.
Unverified50%