A comparative study of machine learning models for malware detection

Malware propagation induce a serious threat to computer security, which requires continuous detection, analysis, and up-gradation of anti-virus programs. Numerous malware detection systems have been proposed however, with advanced malware’s capability of escaping from these detection techniques while remaining undetected leaves a challenging task. Machine learning (ML) methods have been commonly exploited for malware detection and analysis. In this study performance evaluation of two broad categories of ML models; tree-based models and baseline ML models is carried out. Tree-based models include AdaBoost, Gradient boosting, XGBoost, and random forest (RF) whereas baseline classifiers comprise naïve Bayes (NB), support vector classifier (SVC), k-nearest neighbors (k-NN), and decision tree (DT). The performance of these algorithms is evaluated based on performance measures such as accuracy, precision, recall, and F-score. Furthermore, a comparison of results indicates that the accuracy of RF is better among all classifiers of two categories with an accuracy of 99.96%.