Evaluating Machine Learning Models for Intrusion Detection: A Performance-Based Comparison

The use of machine learning (ML) approach for network intrusion detection (NID) via a software-defined network (SDN) has received a lot of interest during the past decade. The extensive advancement of ML methods, the accessibility of enormous data, and the variety of data analysis strategies allowed us to construct a reliable system that can identify many types of network attacks. This research demonstrates how an ML-dependent NID system inspects network data and quickly identifies malicious activities. To conduct this research, the KDD 99 dataset was used. This study utilized various supervised ML algorithms, including Multinomial Naive Bayes (MNB), Random Forest (RF), Support Vector Machine (SVM), Decision Tree (DT), Gradient Boosting (GB), Logistic Regression (LR), and KNearest Neighbor (KNN) algorithms, along with a deep learning (DL) algorithm, Artificial Neural Network (ANN), after selecting the most pertinent features associated with intrusion. Our primary objective during this endeavor is to achieve the utmost precision while minimizing the duration of training and testing time. In comparison to other applicable algorithms, the DT classifier achieves a higher level of performance, reaching 99.45% accuracy with a training duration of 2.963 and a testing duration of 0.046 seconds.