Measuring the Leakage of a Black-Box Using Machine Learning - Giovanni Cherubin
Explore a 54-minute talk by Giovanni Cherubin from the Alan Turing Institute on measuring the leakage of a black-box system using machine learning techniques. Delve into the intersection of statistical learning theory, privacy-preserving protocols, and anonymity attacks. Learn about Website Fingerprinting attacks, a major class of traffic analysis threats, and discover practical methods for providing security guarantees to anonymity protocols. Gain insights into improving privacy and anonymity defenses against machine learning-based attacks, including Membership Inference attacks. Understand how Cherubin's background in Classical Studies, Mechatronics, and Computer Engineering informs his approach to confident methods in Machine Learning, such as Conformal Prediction.