Intelligent System for Malware Detection (ISMD): A Hybrid Intelligent System for Efficient Malware Detection

The demand for anti-malware software products has continued to grow globally due to the increasing rate of malware attacks. Most existing antimalware solutions, such as Kaspersky, Norton, and Bitdefender, rely on signature-based detection techniques. Signature-based techniques are ineffective in detecting new and unseen malware. This research proposes a hybrid-based approach named Intelligent System for Malware Detection (ISMD), which is based on a random forest classifier. As part of the approach for developing ISMD, static and dynamic analysis methods have been developed for extracting both malware and benign features. A fine-tuned feature selection technique that combined the power and efficiency of entropy and Gini gain impurity was used for selecting important features. Next, a random forest classifier model was developed to train and classify the different malware patterns. The ISMD was tested against seven other classifiers and found to have an accuracy range of 98.89%-99.05%. The ISMD was further compared with other state-of-the-art techniques using detection accuracy, training time, and prediction time. The result showed that ISMD outperformed other state-of-the-art malware detection techniques and is suitable when training and predicting time are more important in developing an anti-malware solution.