Yang Zhang gave this talk as part of the ViSP Distinguished Lecture Series (https://visp.wien/lecture-series).

Abstract: Machine learning has made tremendous progress during the past decade. While continuing to improve our daily lives, recent research shows that machine learning models are vulnerable to various privacy attacks. In this talk, I will cover our three recent works on quantifying the privacy risks of machine learning models. First, I will talk about some recent development of membership inference. Second, I will discuss the data reconstruction attacks against online learning. In the end, I will present link stealing attacks against graph neural networks.