BriefGPT.xyz
Sep, 2018
对抗样本是否不可避免？
Are adversarial examples inevitable?
HTML
PDF
Ali Shafahi, W. Ronny Huang, Christoph Studer, Soheil Feizi, Tom Goldstein
TL;DR
本文从理论研究角度分析识别器的易受攻击性，识别了对抗例子的的根本限制。通过实验探讨理论在现实世界的问题中的应用以及图片复杂性对分类器抗对抗性的限制。
Abstract
A wide range of
defenses
have been proposed to harden
neural networks
against
adversarial attacks
. However, a pattern has emerged in which
→