BriefGPT.xyz
Sep, 2018
是否压缩：理解对抗攻击和神经网络压缩之间的相互作用
To compress or not to compress: Understanding the Interactions between Adversarial Attacks and Neural Network Compression
HTML
PDF
Yiren Zhao, Ilia Shumailov, Robert Mullins, Ross Anderson
TL;DR
通过实验，探究了压缩版和未压缩版深度神经网络对抗样本的传递性，并发现折叠模型生成的对抗样本对未压缩模型也很有效，但量化模型的传递性与整数精度高度敏感。
Abstract
As
deep neural networks
(DNNs) become widely used, pruned and
quantised models
are becoming ubiquitous on edge devices; such compressed DNNs are popular for lowering computational requirements. Meanwhile, recent
→