Evaluating the efficacy of decision tree-based machine learning in classifying intrusive behaviour of network users
International Journal of Advanced Technology and Engineering Exploration · 2024
Machine learning (ML) techniques have pervaded into many application areas wherever human-like behavior is mandated, be it learning, decisionmaking, prediction, and/or classification.Typically, ML techniques fall into four categories, namely, supervised, unsupervised, semi-supervised, and reinforcement learning [1].While the supervised ML attempts to link the input attributes with a target attribute [2], the unsupervised ML derives conclusions from input data in the absence of any labelled data [3].On the other hand, semi-supervised ML combines features of supervised and unsupervised techniques to build classification models [4]. *Author for correspondenceReinforcement learning relies on a trial-and-error method, wherein a particular action is taken based on some input data, if the action turns out to be acceptable, a reward point is granted, and if the action does not yield the expected result, then the system learns that such an action would not be effective in future [5].The world-wide acceptance of internet as the most popular communication medium for all kinds of online activities is quite evident in the present times.Along with many of its benefits there has been security challenges due to growing cases of cyberattacks.Though several approaches like data encryption, user authentication, access control, firewalls, etc. have been tried to prevent attacks but none of the approaches have succeeded in providing