Skip to content

Realization of paper: "Generating Adversarial Malware Examples for Black-Box Attacks Based on GAN" 2017

License

Notifications You must be signed in to change notification settings

yanminglai/Malware-GAN

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

36 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Malware-GAN

This is a realization of "Generating Adversarial Malware Examples for Black-Box Attacks Based on GAN". The reaserch is done in Institute of Automation, Chinese Academy of Sciences, Summer 2018. http://english.ia.cas.cn/ with the help of my mentor XuBo.

Introduction

The idea is to use a generative adversarial network (GAN) based algorithm to generate adversarial malware examples, which are able to bypass black-box machine learning based detection models. alt text

Dependencies

Tensorflow 1.80 Keras 2.0 Cuckoo Sandbox 2.03

Dataset

The original malware sample is acquired from https://virusshare.com/, I have used 3000 malware samples and 1500 benign samples for trainning and testing(will expand further).

I used the Cuckoo Sandbox to extract API features from samples. https://cuckoo.readthedocs.io/en/2.0.3/

128 API features are selcted as dimensional vectors to be input to the neutual network (API_list.txt)

Experiments

MalGAN Architecture: generator:128+20—256—128
subsitite detector:128—256—1 maximum epoch:100

True positive rate (in percentage) on original samples and adversarial examples when MalGAN and the black-box detector are trained on the SAME training set

alt text

MLP-TPR

alt text

True positive rate (in percentage) on original samples and adversarial examples when MalGAN and the black-box detector are trained on the DIFFERENT training set

alt text

True positive rate (in percentage) on the adversarial examples after the black-box detector is retrained (RF for example)

alt text

About

Realization of paper: "Generating Adversarial Malware Examples for Black-Box Attacks Based on GAN" 2017

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages