Adversarial Examples Are Not Bugs, They Are Features

Yannic Kilcher

adversarial examples adversarial samples artificial intelligence classifier convolutional neural networks deep learning deep neural network geometry machine learning mit neural network pgd projected gradient descent research robust classifier robust features robust network security vulnerabiliby

description

Abstract: Adversarial examples have attracted significant attention in machine learning, but the reasons for their existence and pervasiveness remain unclear. We demonstrate that adversarial examples can be directly attributed to the presence of non-robust features: features derived from patterns in the data distribution that are highly predictive, yet brittle and incomprehensible to humans. After capturing these features within a theoretical framework, we establish their widespread existence in standard datasets. Finally, we present a simple setting where we can rigorously tie the phenomena we observe in practice to a misalignment between the (human-specified) notion of robustness and the inherent geometry of the data.

Authors: Andrew Ilyas, Shibani Santurkar, Dimitris Tsipras, Logan Engstrom, Brandon Tran, Aleksander Madry

https://arxiv.org/abs/1905.02175 ... https://www.youtube.com/watch?v=hMO6rbMAPew

created

2021-03-01

staked

0.0 LBC

license

Copyrighted (contact publisher)

File size

112421223 Bytes