Explore a cutting-edge approach to defending against adversarial attacks in artificial intelligence systems through a 16-minute IEEE conference talk. Delve into the innovative concept of using backdoors as a protective mechanism, presented by experts from the Chinese Academy of Sciences, University of Chinese Academy of Sciences, and Boston University. Gain insights into how this novel technique, dubbed AI-Guardian, can potentially revolutionize the security landscape of AI models against malicious manipulations.
AI-Guardian - Defeating Adversarial Attacks Using Backdoors
Overview
Syllabus
AI-Guardian: Defeating Adversarial Attacks using Backdoors
Taught by
IEEE Symposium on Security and Privacy
Tags
Related Courses
-
Securing AI and Advanced Topics
-
Disguising Attacks with Explanation-Aware Backdoors
-
Adversarial Prefetch - New Cross-Core Cache Side Channel Attacks
-
Making and Measuring Progress in Adversarial Machine Learning
-
Scatter and Split Securely - Defeating Cache Contention and Occupancy Attacks
-
Distillation as a Defense to Adversarial Perturbations against Deep Neural Networks
