Android Malware Detection and Classification Using Machine Learning Algorithm

Abstract

The cyber security approach that is being offered in this project addresses the growing dangers that rogue applications that target mobile devices are posting. With several forms of malware, such adware, spyware, and ransomware, multiplying quickly, these threats to Android users throughout the world are getting worse. This paper aims to create a thorough classification framework that uses both static and dynamic information to detect Android malware effectively in response. Our method seeks to provide a comprehensive understanding of malware traits and actions by fusing dynamic runtime behavior monitoring with static analysis of APK files. To develop a powerful classification model that can reliably classify various kinds of Android malware by utilizing machine learning algorithms such as Gradient Boosted Trees (GBT) and Ridge Classifier. APK files' metadata, permissions, and code structure are extracted using static analysis, but runtime behaviors including API calls, network traffic, and system interactions are captured using dynamic analysis. Our suggested methodology shows promising results in terms of categorization accuracy, precision, recall, and F1-score after comprehensive testing and evaluation on real-world Android malware datasets. A thorough understanding of malware behavior is made possible by the combination of static and dynamic features, which makes proactive threat detection and mitigation techniques in mobile security easier to implement. Persistently exploiting gullible people with false links, URL phishing is a cyber threat that can result in financial loss, theft of identities, and data breaches. The objective of this work is to create and deploy a strong defense against URL phishing assaults and mobile security procedures against new and emerging Android malware vulnerabilities.