/A-Multimodal-Deep-Learning-Method-for-Android-Malware-Detection

Source code for Android malware analysis

Primary LanguagePython

A Multimodal Deep Leaning Method for Android Malware Detection

mnnarchitecture

Descpription

The python scripts provide some functions to detect Android malware using static features
The several static features are used in the project are used in the classification; manifest features, method API features, method opcode features, shared lib. features, string features
The main classification algorithm, the MNN(Multimodal neural network) takes the static feature vectors and perform the classification.
In addition to the MNN, we also provides other algorithms like the plain DNN(Depp neural network), the SVM, the RF.

  • FeatureVectorGeneration dir: this dir contains the scripts for generating the five feature vectors
  • Classification dir: this dir contains the scripts for the classification algorithms
  • TPA dir: this dir contains the scripts for the TPA

Requirment

  • Python 3
  • Keras (for deep learning methods)
  • Tensorflow (for SVM/RF)
  • Other libraries(Kelper lib for TPA, Numpy, scikit-learn, CUDA, etc.)