Robust-Android-Malware-Detection-Against-Adversarial-Example-Attacks

Features Used in This Paper

This paper uses a total of 379 features collected from .apk file. There are 126 action features, 106 API features and 147 permission features and these features are shown in list_total_actions.txt, list_total_apis.txt and list_total_permissions.txt respectively.

The code & tools in feature_extract can be used to extract features for our model.

Source Code of Our Model

VAE2_ms_su.py, MLP_util.py, VAE_util.py and utilities.py are the source files of our model. main.py is an example of using our model. The key points are:

• Using model = VAE2_ms_su.VAE_SU(...) to build the model
• Using model.train_mlp(...) to train the model
• Using model.get_predict_(...) to get prediction from the model