Socially Responsible Machine Learning: Security, Robustness, and Beyond

Explore the critical aspects of socially responsible machine learning in this 58-minute talk presented by Chaowei Xiao at the USC Information Sciences Institute. Delve into the security threats, robustness challenges, and broader societal implications of deep learning systems. Learn about innovative approaches to enhance adversarial robustness, including purification-based methods and techniques to address covariate shift. Gain insights into the emerging challenges and opportunities presented by foundational models in the context of social responsibility. Discover how cutting-edge research in machine learning intersects with security, privacy, and real-world applications, drawing from the speaker's extensive expertise and recognition in the field.