Skip to content
master
Switch branches/tags
Go to file
Code

Latest commit

 

Git stats

Files

Permalink
Failed to load latest commit information.

README.md

ZO-minmax

Code for paper:

Sijia Liu* , Songtao Lu* , Xiangyi Chen* , Yao Feng* , Kaidi Xu* , Abdullah Al-Dujaili* , Minyi Hong and Una-May O’Reilly, "Min-Max Optimization without Gradients: Convergence and Applications to Adversarial ML", (* Equal Contribution)

Prerequisites

The code is tested with python3.7 and TensorFlow v1.13. Please use miniConda to manage your Python environments. The following packages are required:

tensorflow-gpu>=1.13.0
scipy=1.1.0

Black-box ensemble evasion attack (adversarial example)

To download the pre-trained models:

python3 setup_imagenet.py

To prepare the ImageNet dataset, download and unzip the following archive:

ImageNet Test Set

and put the imgs folder in ../imagesnetdata.

ZO-minmax attack:

run

python3 Main_minmax_universal.py --minmax=1

You can change methods or any hyperparameter in args.

ZO-Finite-Sum case:

run

python3 Main_minmax_universal.py --minmax=0

Black-box poisoning attack against logistic regression model

run

python Main_poison_attack.py
python Main_poison_attack_change_lambda.py

Compare stable point

run

python Main_comparison_STABLEOPT.py

About

Zeroth-order Min-max Optimization

Resources

Releases

No releases published

Packages

No packages published

Languages