Mobile Malware Detection using Anomaly Based Machine Learning Classifier Techniques

Abstract

Mobile phones are a significant component of people's life and are progressively engaged in these technologies. Increasing customer numbers encourages the hackers to make malware. In addition, the security of sensitive data is regarded lightly on mobile devices. Based on current approaches, recent malware changes fast and thus become more difficult to detect. In this paper an alternative solution to detect malware using anomaly-based classifier is proposed. Among the variety of machine learning classifiers to classify the latest Android malwares, a novel mixed kernel function incorporated with improved support vector machine is proposed. In processing the categories selected are general information, data content, time and connection information among various network functions. The experimentation is performed on MalGenome dataset. Upon implementation of proposed mixed kernel SVM method, the obtained results of performance achieved 96.89% of accuracy, which is more effective compared with existing models.