AG 1, AG 2, AG 3, INET, AG 4, AG 5, D6, SWS, RG1, MMCI   Info

AG Audience

English

30 Minutes

As machine learning systems increasingly process sensitive data in critical domains, systematically assessing and mitigating privacy risks is essential. This talk explores quantitative approaches to evaluating privacy vulnerabilities in ML models, with a focus on membership and attribute inference attacks. I will present the MLPrivacy Meter, a framework that measures privacy risks and supports regulatory compliance by providing metrics to evaluate vulnerabilities across different models and training methods. I will also discuss open challenges in privacy risk quantification and propose directions for developing more robust and secure machine learning systems, laying the groundwork for future advancements in trustworthy systems.

Ina Geisler

+49 681 9325 1802

--email hidden

Zoom

passcode not visible

logged in users only