A survey on machine learning-based static analysis of Android malware

Android is the working framework of choice for most smartphones in the advertise. This platform's security has continuously been a critical concern, particularly since it grants clients to download apps from unsubstantiated sources. With the everyday generation and dispatch of thousands of apps, the highlight has moved towards malware discovery utilizing Machine Learning (ML), as restricted to customary discovery strategies. In spite of various scholarly and commercial endeavors, making an effective and reliable malware classification strategy remains an overwhelming assignment. Thus, different datasets for malware investigation have been made and shared over the past decade. These datasets might incorporate static features like API calls, Goals, and authorizations, or dynamic features such as logcat mistakes, shared memory, and framework calls. This paper points to empower analysts to secure in-depth information in the field and to distinguish potential future investigate and improvement directions. Dynamic examination demonstrates to be more vigorous against code muddling. Whereas both double and multi-classification have been investigated in later investigate, the last mentioned offers profitable experiences into the nature of malware. Given that each malware variation capacities in an unexpected way, deciding its category seem help in its prevention.

کلیدواژه ها:

android malware ، machine learning ، classification ، deep learning.