Aishan Liu

A new technique exposes vulnerabilities in advanced AI systems combining images and text.

New benchmark assesses gender bias in AI models related to job roles.

A database to combat backdoor defects in deep learning models.