/fgsm-attack

Implementation of targeted and untargeted Fast Gradient Sign Method

Primary LanguageJupyter Notebook

FGSM-attack

Implementation of the targeted and untargeted Fast Gradient Sign Method attack [1] and a MNIST CNN classifier that is used to demonstrate the attack.

I implemented the MNIST CNN classifier and the FGSM attack to get familiar with pytorch.

Reproduce

Check out fgsm_attack.ipynb and run the notebook.

Results

Targeted

Targeted FGSM attack on MNIST classifier with target class 8

Untargeted

Untargeted FGSM attack on MNIST classifier

[1] Goodfellow, Ian J., Jonathon Shlens, and Christian Szegedy. "Explaining and harnessing adversarial examples." arXiv preprint arXiv:1412.6572 (2014).