DETECTION OF RANSOMWARE USING DEEP LEARNING APPROACH BASED ON ANDROID APPLICATION PERMISSIONS

Abstract

Ransomware, a type of malware commonly employed by attackers, continually enhances, updates, and spreads itself within computer systems, mobile devices and networks. The menace of ransomware malware is increasingly detrimental and perilous to the digital realm, including the domains of computers and mobile devices. The typical cause of ransomware attacks is when users inadvertently surrender their personal data to malicious individuals, who then withhold it until the victim pays a ransom. Both the number of ransomware attacks and their severity are projected to increase annually, reaching 41,000 in 2023 and 12,000 in 2019. Safeguarding against ransomware attacks poses a highly intricate and challenging task for the operator or owner of a computing device. Many the current approach to prevent and detect ransomware attacks involves studying various characteristics and employing different forms of analysis, including dynamic, static, or a combination of both. We will examine and contrast the most recent methods for identifying ransomware. This thesis aims to present a model for identifying ransomware attacks by utilizing a range of deep learning algorithms. The model will be based on the permission lists of Android applications and will incorporate various features and classifiers. The datasets will be suitable for future utilization in additional research endeavors.