Dr. Xiaochen Yang
12:00 21 August 2024
University of Glasgow | School of Mathematics and Statistics | Website
Safeguarding machine learning: from black-box threats to certified robustness.
Adversarial attacks involve manipulating inputs to deceive machine learning models, comprising their reliability. This talk will discuss the transferability of such attacks, indicating the fundamental vulnerabilities in the underlying principles of machine learning algorithms, and present a certified robustness method to enhance model security.
