Skip to content

Implementing white box adversarial attacks on parameters and architecture of CNN in PyTorch

Notifications You must be signed in to change notification settings

nikhitmago/adversarial-attacks-cnn

Repository files navigation

Adversarial Attacks

Implemented 2 white box attacks on LeNet syle CNN to change predictions for the targeted and untargeted settings:

  • Fast Gradient Sign Method (FGSM)
  • Iterative Fast Gradient Sign Method (I-FGSM)

About

Implementing white box adversarial attacks on parameters and architecture of CNN in PyTorch

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published