Comprehensive Survey of Various Machine Learning Techniques to Counter Security Issues Related to Mobile Malwares

Abstract

Malware has been used to attack mobile devices since since it first appeared. The two main types of independent mobile malware attacks are mobile fraud apps and embedded hazardous apps. If one wishes to successfully fight against the cyber dangers posed by mobile malware, a detailed understanding of the permissions specified in apps and API requests is important. This study uses permission requests and API calls to build a powerful categorization model. Android applications use a wide variety of APIs, therefore we've developed three alternative categorization strategies: ambiguous, dangerous, and disruptive, to make it simpler to identify harmful apps. The findings suggest that dangerous apps employ a different set of API calls than benign ones, which demonstrates that mobile malware frequently requests detrimental permissions to access sensitive data. This article provides a thorough literature evaluation of numerous strategies for addressing android malware and associated security issues. The many techniques used to combat malware in the Android operating system are analysed in this article. According to this study, Support Vector Machine and Convolution Neural Network are the most accurate machine learning algorithms for classifying and predicting malware in the Android operating system.