Explore a 15-minute conference talk that delves into a novel approach for purifying backdoors in deep learning models. Learn about the Self Attention Distillation technique proposed by researchers from Wuhan University, Zhejiang University, and Xi'an Jiaotong University. Discover how this method, named REDEEM MYSELF, addresses the critical issue of backdoor attacks in deep learning systems. Gain insights into the potential applications and implications of this purification technique for enhancing the security and reliability of AI models.
REDEEM MYSELF - Purifying Backdoors in Deep Learning Models Using Self Attention Distillation
Overview
Syllabus
REDEEM MYSELF: Purifying Backdoors in Deep Learning Models using Self Attention Distillation
Taught by
IEEE Symposium on Security and Privacy
Tags
Related Courses
-
Distillation as a Defense to Adversarial Perturbations against Deep Neural Networks
-
AI-Guardian - Defeating Adversarial Attacks Using Backdoors
-
BadEncoder - Backdoor Attacks to Pre-trained Encoders in Self-Supervised Learning
-
Selective Amnesia - On Efficient, High-Fidelity and Blind Suppression of Backdoor Effects in Trojaned Machine Learning Models
-
RAB - Provable Robustness Against Backdoor Attacks
-
Spectral-DP - Differentially Private Deep Learning through Spectral Perturbation and Filtering
