3 papers across 3 sessions
We propose the a theoretical-sound machine unlearning evaluation framework with provable properties.
We release the largest preprocessed datasets for the Learning with Errors (LWE) problem