Permission-based Android Malware Detection System Using Feature Selection with Genetic Algorithm


YILDIZ O. , DOĞRU İ. A.

INTERNATIONAL JOURNAL OF SOFTWARE ENGINEERING AND KNOWLEDGE ENGINEERING, vol.29, no.2, pp.245-262, 2019 (Journal Indexed in SCI) identifier identifier

  • Publication Type: Article / Article
  • Volume: 29 Issue: 2
  • Publication Date: 2019
  • Doi Number: 10.1142/s0218194019500116
  • Title of Journal : INTERNATIONAL JOURNAL OF SOFTWARE ENGINEERING AND KNOWLEDGE ENGINEERING
  • Page Numbers: pp.245-262
  • Keywords: Android malware detection, static analysis, machine learning, support vector machines, genetic algorithm, feature selection

Abstract

As the use of smartphones increases, Android, as a Linux-based open source mobile operating system (OS), has become the most popular mobile OS in time. Due to the widespread use of Android, malware developers mostly target Android devices and users. Malware detection systems to be developed for Android devices are important for this reason. Machine learning methods are being increasingly used for detection and analysis of Android malware. This study presents a method for detecting Android malware using feature selection with genetic algorithm (GA). Three different classifier methods with different feature subsets that were selected using GA were implemented for detecting and analyzing Android malware comparatively. A combination of Support Vector Machines and a GA yielded the best accuracy result of 98.45% with the 16 selected permissions using the dataset of 1740 samples consisting of 1119 malwares and 621 benign samples.