search

gongzhitaao / adversarial-classifier

It turns out that adversarial and clean data are not twins, not at all.
https://arxiv.org/abs/1704.04960
MIT License
20 stars 6 forks source link

where is the from attacks.fgsm import fgsm #1

Closed EmotionalXX closed 4 years ago

EmotionalXX commented 4 years ago

hello，sir： the line:from attacks.fgsm import fgsm I can't find this file for attacks.fgsm,can you help me?

gongzhitaao commented 4 years ago

Hi, sorry for the confusion, the attacks folder contains code from https://github.com/gongzhitaao/tensorflow-adversarial/tree/master/attacks

I was too lazy to create s git submodule.

EmotionalXX commented 4 years ago

oh,thanks! but there is still something wrong,do you know how to deal with it? image

gongzhitaao commented 4 years ago

What tf version you have? If I remember correctly, I was using tf.1.3 or tf 1.4 which are really old. I doubt the code is runnable as it is. Unfortunately I'm not planning to upgrade the code to tf 1.15 or tf2.

But the general idea behind the code is rather simple, it would probably take ~1h to rewrite the code if you really want to reproduce it.

Hope this helps :smile:

EmotionalXX commented 4 years ago

What tf version you have? If I remember correctly, I was using tf.1.3 or tf 1.4 which are really old. I doubt the code is runnable as it is. Unfortunately I'm not planning to upgrade the code to tf 1.15 or tf2.

But the general idea behind the code is rather simple, it would probably take ~1h to rewrite the code if you really want to reproduce it.

Hope this helps 😄

hello,Zhitao I uesd tf 1.13 for this examination,but I can not run for it.and i have tried many times. because this is important and first for me to make adversarial samples detection,I really want to make it.Maybe I need your help when you are free.

gongzhitaao commented 4 years ago

tf. 3 is definitely not compatible with tf 1.13. So many things changed. You may need to rewrite part of the code or use https://github.com/tensorflow/cleverhans since it is still maintained AFAIK Good luck :smile: