Franziska Boenisch

A new method improves privacy and accuracy in data-driven models.

Introducing new methods to improve forgetting processes in contrastive learning models.

Examining how SSL models memorize data points and its implications.