Explore the significance of skip connections in adversarial example generation using ResNets in this 28-minute lecture from the University of Central Florida. Delve into topics such as skip gradient maps, experimental setups, threat models, and SGM attacks. Examine the comparison between SGM and robust models, investigate gamma variations in GM, and analyze single-step FGSM attacks. Gain valuable insights into the transferability of adversarial examples and their implications for deep learning security.
CAP6412 - Skip Connections Matter: On the Transferability of Adversarial Examples Generated with ResNets
University of Central Florida via YouTube
Overview
Syllabus
Introduction
Related works
Skip gradient map
Experimental setup
Threat model
SGM attacks
SGM vs robust models
GM variations in gamma
Singlestep Fgsm attacks
Conclusion
Taught by
UCF CRCV
Tags
Related Courses
-
Improving Transferability of Adversarial Examples with Input Diversity - CAP6412 Spring 2021
-
Explaining and Harnessing Adversarial Examples in Machine Learning - Spring 2021
-
Cross-Domain Transferability of Adversarial Perturbations - CAP6412 Spring 2021
-
Trust Region Based Adversarial Attack on Neural Networks
-
Adversarial Examples in Machine Learning - Crafting and Defending Against Attacks
