Analyzing the Efficacy of Machine Learning Algorithms on Intrusion Detection Systems

Abstract

Internet security has been a problem for businesses around the world. Encryption, authentication, and virtual private networks have been used to safeguard the network infrastructure and communications over the whole process of data protection. Intrusion detection systems (IDSs) are an advancement in network security that safeguards organizational data. System detecting intrusions into computer networks is known as IDS. Throughout their life, information cannot be guaranteed to be secured. An IDS's task is to find if there is any danger or security breach. The IDS identifies deliberate attempts by authorized users or by third parties to take advantage of security flaws as well as actual abuse. In this study, the authors used classifiers such as decision trees, support vector machines, Naive Bayes, random forests, and logistic regression. The authors also used machine learning algorithms to calculate accuracy, precision, recall, and false positives. They can conclude that this model suggested decision trees with the highest accuracy of 82.3%.