Evolutionary Decision Tree-Based Intrusion Detection System

In this paper, an Evolutionary Decision Tree based Intrusion Detection System (IDS) has been proposed which is based on the concept of decision tree (DT) and genetic algorithm (GA). The goal of the projected IDS is to protect the computer network from the various types of cyberattacks. The main hurdles in today's decision tree-based IDS are the problem of small disjunct, preprocessing of the network logs, learning of the desired system for anomalous or signature detection, outlier handling on the training set, etc. In this research paper, the concept of interquartile range is used to preprocess the data to handle the outlier and extreme values in the training set, and then the decision tree is used to generate the DT. Further, the decision rules, which come in the category of small disjunct, who have a vital role in the accuracy are optimized using metaheuristic GA. Furthermore, the proposed system is compared with the existing IDS. The performance of the system is significant in terms of accuracy and classification error is compared to the existing systems.