Adversarial Examples as an Input-Fault Tolerance Problem
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
ckpt
dataset
.gitignore
LICENSE
README.md
adef_on_svhn_funcs.py
advex_input_fault_tolerance_nips18_secml.pdf
cnn_model_pytorch.py
deformation.py
ft_plot.py
ft_utils.py
svhn_advex_fault_tolerance_pytorch.ipynb
svhn_generate_plots_for_paper.ipynb

README.md

Adversarial Examples as an Input-Fault Tolerance Problem

Fully self-contained source code release including pre-trained checkpoints and a test set for reproducing all figures and experiments in the paper.

The initial release is in PyTorch, with a TensorFlow notebook coming soon.

The main dependencies are

  • python 3.6.x
  • pytorch (tested with 0.4.1, v0.4.0 is known to not work)
  • numpy, skimage, matplotlib