An Ensemble Machine Learning Based Cybersecurity IntrusionDetection System (C-IDS)

Abstract

An essential component of network security is intrusion detection, which shields computer systems against attacks and illegal access. Traditional intrusion detection systems (IDS) rely on signature-based detection, which limits their ability to detect unknown complex threats. Machine learning-based methods have demonstrated encouraging outcomes in detecting unidentified harmful intrusions. An approach to developing an IDS model based on ensemble learning is presented in this research. Compared to its individual classifiers, the model presented in this research performs comparatively better overall. Lightweight ML approaches, such as QDA, NB, SGD, and DT, are used to build this ensemble model. The dataset is utilized to train and assess the performance of this suggested model as well as the individual classifiers used to construct the ensemble model. The binary class is used to evaluate the performance. In order to choose only the most pertinent features, the suggested approach additionally makes use
of a feature selection technique. The empirical findings unequivocally demonstrate that employing an ensemble classifier can be very beneficial in the field of IDS with unbalanced datasets, where misclassifications might be expensive. The suggested strategy using the Decision Tree (DT) technique beats current approaches in terms of accuracy, usually reaching 95% with improved assessment metrics according to our research using various ensemble methods. This tactic could be helpful in bolstering computer systems' and networks' defenses against new
cyberthreats