Systematic Mapping of Machine Learning–Based Malware Detection Studies

The threat of computer malware and viruses is an ever-growing threat in today’s technological age. While the use of common anti-virus programs help limit the issue of having a computer infected with malware, the types and frequency of these viruses continue to rise and evolve at an uncontrollable rate. In order to combat this issue, the implementation of machine learning algorithms must continue to rise and evolve to fight and prevent a wide selection of malware. In this study, we are presenting our findings from a systematic mapping (SM) review of research in this area to determine the main factors that go into the testing and detection development of said research. These factors include the types of malware being tested, the machine learning algorithms used, among others. To determine these factors, we analyze a selection of articles from a literature database search query. This search, limited between the start of 2017 to the end of March 2022, resulted in 254 studies conducted in the scientific literature space. After conducting a multi-phase review of those studies, a subset of 28 papers were selected for further analysis. The results obtained from applying the systematic mapping process indicate that the testing of machine learning has a range of potential benefits, but has significant potential for improvement in future research.