Files

PDF

HOW TO CITE

A Comparative Evaluation of Machine Learning Algorithms for Network Instrusion Detection
Authors (Affiliation): Kavalla Prasanna (NFSU), Ujjaval Patel (National Forensic Sciences University)
Abstract:

Network intrusion detection systems (NIDS) are essential for cyber security since they help find and prevent malicious activities in the network traffic. This research uses a dataset from Kaggle and various machine learning methods to achieve a more effective NIDS. For analyzing dataset, an initial phase of the research a Recursive Feature Elimination (RFE) algorithm was used to identify the most significant features for the Random Forest Classifier. Further, Optuna has been used to find the best model and adjust its settings when utilizing three types of classifiers: Decision Tree, K-Nearest Neighbors (KNN) and Logistic Regression. All these algorithms are used to detect malicious activities for enhancing network security, and improving threat detection strategy. An exhaustive comparative evaluation demonstrates that Decision Tree algorithms outperform other algorithms for the effective network intrusion detection in adaptive manner with much higher accuracy.

Keywords: Network Intrusion Detection, Machine Learning, Feature Selection, Hyperparameter Optimization, Logistic Regression, K-Nearest Neighbors, Decision Tree, Optuna.
Vol & Issue: VOL.3, ISSUE No.1, June 2024