carlini nn_robust_attacks issues

carlini / nn_robust_attacks

Robust evasion attacks against neural network to find adversarial examples

BSD 2-Clause "Simplified" License

776 stars 229 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

L2 regularization term is squared. Why here specifically? Which impact?

#52 ego-thales opened 5 months ago
0
GZip error

#51 ananyag16 opened 1 year ago
2
question for self.newimg in l2_attack

#50 lzclzclzc12 opened 1 year ago
0
What are the keras and tensorflow imported in the code？

#49 xiaolinjieer opened 2 years ago
0
why 10000 in your code,what's the meaning?Thanks!!!

#48 liuyishoua opened 3 years ago
2
Create my_describe

#47 Omid-Nejati opened 3 years ago
0
What version of tensorflow + keras?

#46 kevjunlee closed 2 years ago
1
Unsuccessful TensorSliceReader constructor

#45 le000043 opened 3 years ago
1
Unable to run train_models.py

#44 le000043 closed 3 years ago
2
Unable to open file

#43 keerthi4399 opened 3 years ago
1
L2 untargeted attack not working?!

#42 as791 closed 3 years ago
2
Little mistake in the doc of the attack method

#41 Amioplk closed 3 years ago
0
I want to attack my own model training by tensorflow2.0.

#40 Madanfeng closed 3 years ago
2
L_inf always fails if abort_early is False

#39 samuelemarro opened 4 years ago
0
Any adversarial attack that sustains after resize attack

#38 BalaMallikarjuna-G closed 3 years ago
1
Low validation accuracy of CIFAR

#37 HaiQW opened 4 years ago
2
TODO

#36 HaiQW closed 4 years ago
0
Misleading printing?

#35 FairyOnIce opened 4 years ago
0
no boxmin and boxmax in L_0 and L_inf

#34 CNOCycle opened 4 years ago
0
modifier always equals zero

#33 overflocat closed 3 years ago
0
About the settings for imagenet

#32 lith0613 closed 5 years ago
3
How to control the pixel number to be noised ?

#31 lith0613 closed 5 years ago
2
About the hyper parameters for cifar and mnist

#30 lith0613 closed 5 years ago
4
How can I load the pretrained pb file of tensorflow format into the project?

#29 lith0613 closed 5 years ago
0
Can I run this on a Mac OS?

#28 d4mster closed 5 years ago
1
Modified prev init value to avoid early breaking

#27 jeromerony closed 5 years ago
0
why I only change the weight of the attacked model, the examples are not adversarial any more.

#26 xieyi318 closed 5 years ago
3
question for l2 distance in l2_attack

#25 ZekunStevenZhang closed 5 years ago
2
when i run trains_model.py i am getting this error....i am beginner,if any help would be greatfull.

#24 veenah closed 5 years ago
12
The performance of l2_attack on Pytorch

#23 zjysteven closed 5 years ago
5
Setting confidence=0 produces different adversarial accuracy between different runs

#22 ForeverZyh closed 3 years ago
7
Unable to generate l2 attack examples

#21 ericcheng09 closed 5 years ago
5
l0 attack: Potential Bug

#20 ankur6ue closed 6 years ago
1
Possibility of "fixing" the number of pixels to be modified - L0 attack for MNIST

#19 VishaalMK closed 6 years ago
3
l0 attack can't work correctly

#18 meet-cjli closed 6 years ago
0
L_infinite implementation problem

#17 hanwei0912 closed 6 years ago
3
l0 not implemented correctly

#16 gehuangyi20 closed 6 years ago
4
li attack clarification

#15 gehuangyi20 closed 6 years ago
2
Unable to reproduce L0 attack on MNIST

#14 VishaalMK closed 6 years ago
2
Made the print compatible with Python 2.

#13 mzweilin closed 6 years ago
0
Update and rename README to README.md

#12 iamgroot42 closed 6 years ago
0
Correct confidence comparison for untargeted attacks

#11 wh0 closed 6 years ago
1
Make range of box constraints configurable

#10 wh0 closed 6 years ago
0
Parameter setting for L_0 and L_inf attack

#9 ariewahyu closed 6 years ago
3
Readme, GPU utilization improvement

#8 iamgroot42 closed 6 years ago
2
Why the label of attack images is same with the original images?

#7 shenqixiaojiang closed 6 years ago
2
Graphdef cannot be larger than 2Gb in tensorflow when using the InceptionModel

#6 shenqixiaojiang closed 6 years ago
0
the method for getting the cifar10 data in the setup_cifar.py was modified

#5 shenqixiaojiang closed 3 years ago
3
Contribute implementation to Foolbox

#4 wielandbrendel closed 6 years ago
1
why the adversarial perturbations was damaged by saving the adversarial samples with scipy.misc.

#3 shenqixiaojiang closed 6 years ago
3