Noisy fine-tuning with gradient or model clipping on retained data provably removes the influence of forget data, with guarantees that need no smoothness or convexity assumptions.
Hypothesis testing interpretations and renyi differential privacy
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Certified Unlearning for Neural Networks
Noisy fine-tuning with gradient or model clipping on retained data provably removes the influence of forget data, with guarantees that need no smoothness or convexity assumptions.