AMDwithPermission-ML
Andoird Malware Detection with different numbers of permissions using machine learning
Try to detect Android Malware with permissions only
With Machine Learning
3 folders with different numbers of permissions
- 88 permissions
- dangerous pid for 34 permissions
- sigpid for 22 permissions
packages needed:
- androguard
- numpy
- pandas
- scikit-learn
usage:
- all.py for training with train.csv and test.csv
- apkPredict.py for predict apks in dir using model file generated by all.py
- other py for merge permissions
others:
models and dataset too big so they get compressed
of course u can train ur own model
dataset need to be merged by .py
中文版
三个文件夹
- 采用88个权限的代码
- 采用34个权限的代码
- 采用22个权限的代码
需要的库文件
- androguard
- numpy
- pandas
- scikit-learn
用法
- all.py用于训练,需要提供train.csv和test.csv
- apkPredict.py用于预测,读取现有的model文件中的模型
- 其他文件用于格式化找到的数据集,可能要根据不同的数据集进行调整
其他
- 数据集和model文件太大,单独打包上传
- 不同权限个数的代码应采用不同的model