Android Malware Detection Using Machine Learning

Abstract

As mobile applications keep expanding, Android phones have increasingly become a target for malware attacks, which pose significant security concerns. Traditional approaches to malware detection tend to rely on signatures, which are not able to match the continuously shifting world of malware threats. The objective of this project is to design a system that takes use of machine learning (ML) technologies, specifically neural networks (NN) and support vector classifiers (SVC), for the identification of Android malware. To determine if an application is malicious or benign, our approach looks at its behaviour and many characteristics. In order to improve detection performance, we combine static and dynamic analysis, extract features using a neural network and SVC, and train the system on a carefully selected dataset. While the SVC excels at dividing applications into two groups-malicious and non-malicious. The neural network excels at seeing intricate patterns. By combining these techniques, we can reduce false positives and increase classification accuracy. According to experimental results, our hybrid model can detect new and undetected viruses with high accuracy, providing Android devices with a strong degree of protection. Notably, the solution offers real-time protection against malware attacks on mobile devices and is designed to be easily integrated into applications.