Skip to content
/ fgsm Public

Implementation for the Fast Gradient Sign Method for generating adversarial examples

Notifications You must be signed in to change notification settings

pratik18v/fgsm

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Fast Gradient Sign Method

This is the implementation (in PyTorch) of the method proposed in the paper: Explaining and Harnessing Adversarial Examples, for generating adversarial examples. The implementation is over the MNIST dataset.

Results

Accuracy of the network w/o adversarial attack on the 10000 test images: 97 %

Accuracy of the network with adversarial attack on the 10000 test images: 14 %

Number of misclassified examples (as compared to clean predictions): 8374/10000

Please check the iPython notebook for visualization of the results.

About

Implementation for the Fast Gradient Sign Method for generating adversarial examples

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published