A comprehensive study on high performance malware classifiers based on machine learning algorithms

Main Article Content

Saifaldeen H. A., Karam H. Thanoon

Abstract

Malware as a malicious software has been developed and became an interest issue that take great
attention of the researchers and companies that delt with data security. Therefore, determining
the classes of these malwares is very important to detect further newer or modified versions that
are continuously developed. Many classifiers had been developed to implement them to build
newer detectors to that are able to secure data. This paper, as comprehensive study illustrates the
performance of the most common malwares and highest performance classifiers that were
recently proposed. The study shows that there are three proposed classifiers with highest
accuracy. They are: Random forest, Support Vector Machine and x-gradient boots. The number
of features had a great effect in classification process in which the accuracy decreased if the
features number increased. Additional techniques may enhance classification accuracy such as
New Feature Engineering.

Article Details

Section
Articles