Android Malware Classification using LSTMs
Sequence classification of complex flows for detection of malicious intent
The static analysis of Android APKs can reveal flows of information that may or may not be indicative of malicious intent. “Complex flows are mechanisms that capture the usage of sensitive mobile resources, while revealing the structure of and relationships within these usages”. The author’s of this paper suggest that these flows contain certain that could be mined(ideally using some sort of sequence classifier).
This project demonstrates the usage of LSTM modules trained using categorical cross entropy loss performing with an 80% test accuracy, albeit being trained only on a small corpus along with design changes to handle computational contraints.

https://github.com/AshVijay/Android-Malware-Classification —