Machine Learning in Cybersecurity: Systematic Literature Review

Abstract

The technological advancement has generated an urgent need to bolster cybersecurity, addressing incidents such as Denial of Service (DOS) and Distributed Denial of Service (DDOS) attacks. Therefore, this systematic review aims to explore current technologies in cybersecurity, fortify digital security, and provide crucial insights into the latest trends, contributing to the ongoing adaptation of cybersecurity strategies. The methodology, based on the PICO strategy, organizes the search in Scopus and IEEE databases, selecting 21 publications out of a total of 308. The results underscore the intricacies of cybersecurity and the variability in the effectiveness of machine learning algorithms, highlighting the importance of meticulous tool selection. Moreover, it is noted that, on average, Decision Tree algorithms achieved a precision of 99.59% for DOS attacks, playing a pivotal role in defense against cyber threats. The conclusion emphasizes the critical need for adaptable strategies supported by efficiencies ranging from 41% to 99%, suggesting exploration of hybrid approaches and emerging challenges for the continual enhancement of cybersecurity. Additionally, a detection rate of 99.6% underscores the importance of careful tool selection, with a 32% false positive rate and 16% in metrics such as precision and recall, emphasizing the necessity for anticipation and flexibility for effective cybersecurity.