Empirical Evaluation of Noise Influence on Supervised Machine Learning Algorithms Using Intrusion Detection Datasets

Optimizing the detection of intrusions is becoming more crucial due to continuously rising rates and ferocity cyber threats attacks. One popular methods optimize accuracy intrusion systems (IDSs) by employing machine learning (ML) techniques. However, there are many factors that affect ML-based IDSs. these noise, which can be in form mislabelled instances, outliers, or extreme values. Determining extent effect noise helps design build robust This paper empirically examines on IDSs conducting a wide set different experiments. The used ML algorithms decision tree (DT), random forest (RF), support vector (SVM), artificial neural networks (ANNs), Naïve Bayes (NB). In addition, experiments conducted two widely datasets, NSL-KDD UNSW-NB15. Moreover, also investigates use as base classifiers with ensembles methods, bagging boosting. detailed results findings illustrated discussed this paper.