Skip to content

Latest commit

 

History

History
executable file
·
14 lines (14 loc) · 1.08 KB

File metadata and controls

executable file
·
14 lines (14 loc) · 1.08 KB

Android-malware-prediction-using-deep-learning-methods

I am prepare to build up a deep learning model to automatically
predict if a android apk is a malware.
(1)At the begining I am tring to pick out the permissions from Androidmanifext.xml
step ONE:decompile.py
use apktool to decompile the target apk,result in folder apk;
step TWO:general_permissions_vector.py
extract the permissions used in the target apk.then use
all permmisions from https://developer.android.com/reference/android/Manifest.permission.html to general a vector stands for this apk;
step THREE:extract_api.py
extract all the system call from smali.
step FOUR:extract_permission_used.py
tips:the mapping between android apis to permissions is generated from pscout
paper link http://www.eecg.toronto.edu/%7Elie/papers/PScout-CCS2012-web.pdf