The code is for paper: 'Structured Adversarial Attack: Towards General Implementation and Better Interpretability' which accepted in ICLR 2019 (https://openreview.net/forum?id=BkgzniCqY7) by Kaidi Xu*, Sijia Liu*, Pu Zhao, Pin-Yu Chen, Huan Zhang, Quanfu Fan, Deniz Erdogmus, Yanzhi Wang, Xue Lin (* means equal contribution)

@inproceedings{
xu2018structured,
title={Structured Adversarial Attack:  Towards General Implementation and Better Interpretability},
author={Kaidi Xu and Sijia Liu and Pu Zhao and Pin-Yu Chen and Huan Zhang and Quanfu Fan and Deniz Erdogmus and Yanzhi Wang and Xue Lin},
booktitle={International Conference on Learning Representations},
year={2019},
url={https://openreview.net/forum?id=BkgzniCqY7},
}

To prepare the ImageNet dataset, download and unzip the following archive:

ImageNet Test Set

and put the imgs folder in ../imagesnetdata. This path can be changed in setup_inception.py.

To download the inception model:

python3 setup_inception.py

To train CIFAR10 and MNIST model: run

python3 trainmodel.py -d all

StrAttack:

run

python3 test_attack_iclr.py

You can change methods, dataset or any hyperparameter in args or add in command line.