Center for Advanced Intelligence Project

Speaker:
Dr. Adi Shamir (Weizmann Institute, Israel)
https://en.wikipedia.org/wiki/Adi_Shamir

Title:
A Simple Explanation for the Mysterious Existence of Adversarial Examples with Small Hamming Distance

Abstract:
The existence of adversarial examples in which tiny changes in the input can fool well trained neural networks has many applications and implications in object recognition, autonomous driving, cybersecurity, etc. However, it is still far from being understood why such examples exist, and which parameters determine the number of input coordinates one has to change in order to mislead the network.
In this talk I will describe a simple mathematical framework which enables us to think about this problem from a fresh perspective, turning the existence of adversarial examples from a baffling phenomenon into a natural consequence of the geometry of R^n with the $L_0$ (Hamming) metric, which can be quantitatively analyzed. An interesting consequence of our analysis is to show that many proposed techniques to immunize deep neural networks against adversarial attacks are unlikely to succeed.