search

IBM / Autozoom-Attack

Codes for reproducing query-efficient black-box attacks in “AutoZOOM: Autoencoder-based Zeroth Order Optimization Method for Attacking Black-box Neural Networks” ​​​​, published at AAAI 2019
https://arxiv.org/abs/1805.11770
Apache License 2.0
56 stars 22 forks source link

Question on the code #1

Closed dssrgu closed 5 years ago

dssrgu commented 5 years ago

Hi, I'm trying to understand your code and reproduce the results. While reading through the code, I found that 'o_bestattack' at 'blackbox_attack.py' line 238 first initializes as a non-tanhed image but gets a tanhed image after attack success. Why is it not initialized as a tanhed image? Am I missing something? Thanks.

chunchentu commented 5 years ago

Hi,

The o_bestattack is a variable for storing the best attack results. We initialize it with the original image and this would not influence the attack. The class attribute USE_TANH will determine if tanh function will be applied on the image or not.

Bests Chun-Chen