End-to-end implementation of ML-based Android malware detectors.
- DREBIN from Arp, Daniel, et al. "Drebin: Effective and explainable detection of android malware in your pocket." NDSS 2014. [paper]
- SecSVM from Demontis et al. "Yes, machine learning can be more secure! a case study on android malware detection." IEEE TDSC 2017. [paper]
- A BaseDREBIN class is also provided, allowing to easily and efficiently train any classifier on the DREBIN feature set by extending a few methods.
The implemented detectors serve as baselines for the benchmarks hosted in the Cybersecurity Use Case of the ELSA EU project. This repository should be used as a starting point to build a model. A step-by-step guide for each evaluation track can be found in this repository.
Pre-trained models (on data provided on the ELSA benchmarks website) can also be downloaded from Drive:
The downloaded files must be placed in the pretrained folder.