This repo is initial implementation for the paper "Undetectable Adversarial Examples based on Microscopical Regularization"

How to run

adversarial attack for 'best case' mode

python main_ut.py

adversarial attack for 'average case' mode

python main_at.py

Tips

You can download the dataset from:
https://drive.google.com/file/d/13N7Y2KeEaK4l-KwaDSw7PiK5S5-6T9EQ/view?usp=sharing