-
I train a pytorch model on cifar10 , and I use fgsm and pgd attack to test my model.
I found the accuracy in these two attack methods are much higher than "cleverhans" and don't know the reason
-
I'm seeing a small discrepancy between the number of images that successfully fooled the network when checking `success` vs. getting predictions for those images by running it through the model again.…
-
-
Hello, thanks for your excellent work and released codes.
Could you please provide your pre-trained models with/without RPT for Cifar10/100 datasets? We only have limited graphic cards while adversa…
-
For me, the Deepfool attack is not working as expected.
While I'm getting good results for FGSM and BIM, Deepfool only outputs the Gradient of the image.
The bounds + preprocessing is exactly the …
-
### Topic Suggestion
Machine Learning
### Writing sample(s):
[Blog Samples](https://www.section.io/engineering-education/authors/deewakar-chakraborty/)
### Proposal Submission
#### Propose…
-
Hi harry,
I have run the FGSM with default distance measure Linf. I can see there is PGDL2 class, however, I couldn't find FGSML2, How can I change it to L2 norm.
-
用FGSM做白盒攻击时，基于resnet的预训练模型分类标签与imagenet不符，不是用imagenet训练出来的吗？标签id与类别的对应文档在哪里呢 ？
-
Hello,
When I try to *python extract_characteristics.py -d cifar -a fgsm -r lid -k 20 -b 100*, there raises an error:
tensorflow.python.framework.errors_impl.InvalidArgumentError: conv2d_1_input:0 i…
-
Thanks for the great work.
However, I can't find the detail how the exp in fig3 of PGD and MI-FGSM are performed? what's the loss of those attacks when doing attack?