A stochastic input pre-processing technique based on a process of down-sampling/up-sampling using convolution and transposed convolution layers. Defending convolutional neural network against adversarial attacks.
A Pytorch code for our paper. It includes the implementation of our defense startegy named "SIT: Stochastic Input Transformation to Defend Against Adversarial Attacks on Deep Neural Networks"
If you find this code useful in your research, please cite:
@ARTICLE{9422778,
author={Guesmi, Amira and Alouani, Ihsen and Baklouti, Mouna and Frikha, Tarek and Abid, Mohamed},
journal={IEEE Design Test},
title={SIT: Stochastic Input Transformation to Defend Against Adversarial Attacks on Deep Neural Networks},
year={2021},
volume={},
number={},
pages={1-1},
doi={10.1109/MDAT.2021.3077542}}