研读笔记-A study of the effect of JP

2019-08-28  本文已影响0人  不想掉队的小布丁

文章链接:http://arxiv.org/pdf/1608.00853

摘要:文中评估JPG压缩对对抗性图像分类的影响。对于小幅度的FGSM扰动,我们发现JPG压缩常常在很大程度上逆转分类精度的下降,但并不总是如此。随着扰动大小的增加,仅通过JPG重新压缩不足以扭转FGSM的影响

in any challenging high-dimensional classification task where the inputs naturally live in (or near) a complex lower-dimensional data subspace, adversarial examples will lie outside this data subspace, taking advantage of the fact that the training objective for the neural network is essentially agnostic to the network’s behavior outside the data subspace.

Neural networks classifiers work well, in part, due to their strong inductive biases. But this same bias means that a neural network may report strong predictions beyond the data subspace where there is

扩展内容

jpg 压缩算法代码示例:
https://github.com/richgel999/jpeg-compressor

上一篇 下一篇

猜你喜欢

热点阅读