Homogenous Ensemble Learning for Denial of Service Attack Detection

Abstract

Network hacking has become more resource-intensive in recent years, particularly in firms and organizations that rely on the internet, such as Amazon, because the hacker’s goal is to prohibit the user from using the network's resources whether the target is offensive or pecuniary. As a result, several approaches for detecting intrusion and network penetration have arisen, utilizing various techniques such as artificial intelligence processes, machine learning, and deep learning to discriminate between authorized and illegitimate users. This paper discusses the diverse methods of ensemble learning as a part of machine-learning techniques and Ensemble-learning techniques used in different datasets to identify denial of service attacks. Demonstrates a variety of machine-learning algorithms, including Random Forest, Bagging, and Boosting, these are used to discover various sorts of attacks in the NSL_KDD dataset Bagging algorithms had the maximum detection accuracy of 99 %, while AdaBoost methods had the lowest accuracy of 93 %. As a result, the same approaches were used on the same dataset, but exclusively to detect DoS attacks. Bagging, RF, and eXtreme gradient boost (XGB) algorithms had the maximum detection accuracy of 99 %, while AdaBoost methods had the lowest accuracy of 37 %.