Implementation of Intrusion Detection System Using Various Machine Learning Approaches with Ensemble learning

Abstract

Recent years have seen an increase in advanced threat attacks, yet feature filtering-based network intrusion detection systems have a number of shortcomings that make it challenging for security managers and analysts to identify and thwart network intrusions in their organizations. Information systems are routinely protected and damage is minimized using techniques for detecting intrusions. It protects against dangers and weaknesses in real-world and virtual computer networks. Effective intrusion detection systems are now typically created using machine learning techniques. Neural networks, statistical models, rule learning, and ensemble techniques are examples of machine learning techniques for intrusion detection. Machine learning ensemble techniques are renowned for their superior performance during the learning process. For the creation of a successful intrusion detection system, a suitable ensemble technique must be investigated. In this paper, we introduced a novel ensemble method for intrusion detection in the network along with a combination of decision tree, random forest, extra tree, and XGBoost algorithms. The suggested method was created utilizing the Python programming language and aids in improving detection accuracy. Utilizing the CICIDS2017 dataset, the constructed system is evaluated based on numerous evaluation criteria, including precision, recall, and f1-score. The ensemble approach significantly raises the detection accuracy.