Skip to content

Explanation-guided boosting of machine learning evasion attacks.

Notifications You must be signed in to change notification settings

um-dsp/EG-Booster

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 

Repository files navigation

EG-Booster: Explanation-Guided Booster of ML Evasion Attacks

This repository contains the source code accompanying our ACM CODASPY'22 paper EG-Booster: Explanation-Guided Booster of ML Evasion Attacks.

Used system

  • Linux
  • 6 vCPUs, 18.5 GB memory
  • GPU: 1 x NVIDIA Tesla K80

Downloading Repo

$ git clone https://github.com/EG-Booster/code.git

CIFAR10

It is highly recommended to create a new separate python3 environment:

$ python3 -m venv ./EG-CIFAR10-env

$ source EG-CIFAR10-env/bin/activate

$ cd code/CIFAR10

$ pip install -r requirements.txt

Note: to avoid any runtime and memory-related errors please adjust batch_size and num_workers in the configuration area of CIFAR10.py, according to your Hardware envirnment. Default values are batch_size = 64 and num_workers = 2

Finally, run the following:

$ python CIFAR10.py

Contact

Abderrahmen Amich: [email protected]