This is the project page of our paper:
Learning to Learn Transferable Attack
Fang, S., Li, J., Lin, X., & Ji, R.
AAAI 2022. arxiv
images/dev_dataset.csvcontains only URLs of the images. Download these images and put them intoimages- Generate adverial examples and the results will be saved in
output- ResNet-50 as the source model:
python main.py --source-model resnet50 - DenseNet-121 as the source model:
python main.py --source-model densenet121
- ResNet-50 as the source model:
- Download the adversarially trained models in here
- To evaluate success rate
- Attack naturally trained models:
python eval/evaluate_NT_trained.py --adv-dir output - Attack adversarially trained models:
python eval/evaluate_AT_trained.py --adv-dir output
- Attack naturally trained models:
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent