Hands-On Lab: Robustness in Image Recognition (Part 2)

GOTO Amsterdam 2019

An intriguing property of Deep Neural Network image classifiers is that, while they are normally highly accurate, they are vulnerable to so-called adversarial examples. Adversarial examples are inputs which have deliberately been modified to produce a desired response by a DNN. If your cat is accidentially recognized as an ambulance that may be amusing, but what if your autonomous car doesn't recognize a stop sign because of a couple of stickers? In this workshop we will look at typical adversarial attack methods and ways to mitigate this threat using the IBM Adversarial Robustness Toolbox.

This is a 2 parts Hands-on Lab, starting at 14:15 in Verwey kamer.