-
-
We implemented all attacks to perform some type of internal hyperparameter-tuning. For example, FGSM loops over many step-sizes to find the minimum perturbation. At the same time, some people might wa…
-
The winning attack in last years NIPS competition:
https://github.com/tensorflow/cleverhans/blob/538bf672faeeb355bc13e08f2916eb192a22e7e7/cleverhans/attacks.py#L462
Paper:
https://github.com/te…
-
It seems that we can't use tf model as forward_model and pytorch model as backward_model at the same time, is that right?
=========================================
model = foolbox.models.Composite…
-
Hey! Wonderful library!! :)
I'm trying to generate adversarial examples for a keras model. However, I have a setup with two models, one classifier and one autoencoder, both trained independently an…
-
Hi, I was running your code (so neat and nice),
but I think fgsm attack seems not working in whitebox setting.
I ran separately adv_x to see how it looks like, and it was very clean.
I checked th…
-
I use a trained pytorch model-resnet18 in foolbox. The model is trained to classify only 2 classes, and first loaded in python. But when I use it to make adversarial examples, it reported assertion Er…
-
It seems that the perturbations computed by the FastGradientMethod (FGM) and BasicIterativeMethod attacks are sometimes equal to zero.
In a first preliminary test (not in cleverhans), this problem …
ricvo updated
6 years ago
-
Hi folks, I'm a student works on adversarial examples and new to cleverhans. I spend some time on the entry level tensorflow tutorial of MNIST with FGSM. I noticed that the MLP model is using lower le…
-
Hi,
I tried following the instructions mentioned in the [tutorial](http://foolbox.readthedocs.io/en/latest/user/tutorial.html)and [examples ](http://foolbox.readthedocs.io/en/latest/user/examples.htm…