Abstract:Internet of things (IoT) has been playing an important role in many sectors, such as smart cities, smart agriculture, smart healthcare, and smart manufacturing. However, IoT devices are highly vulnerable to cyber-attacks, which may result in security breaches and data leakages. To effectively prevent these attacks, a variety of machine learning-based network intrusion detection methods for IoT networks have been developed, which often rely on either feature extraction or feature selection techniques for reducing the dimension of input data before being fed into machine learning models. This aims to make the detection complexity low enough for real-time operations, which is particularly vital in any intrusion detection systems. This paper provides a comprehensive comparison between these two feature reduction methods of intrusion detection in terms of various performance metrics, namely, precision rate, recall rate, detection accuracy, as well as runtime complexity, in the presence of the modern UNSW-NB15 dataset as well as both binary and multiclass classification. For example, in general, the feature selection method not only provides better detection performance but also lower training and inference time compared to its feature extraction counterpart, especially when the number of reduced features K increases. However, the feature extraction method is much more reliable than its selection counterpart, particularly when K is very small, such as K = 4. Additionally, feature extraction is less sensitive to changing the number of reduced features K than feature selection, and this holds true for both binary and multiclass classifications. Based on this comparison, we provide a useful guideline for selecting a suitable intrusion detection type for each specific scenario, as detailed in Tab. 14 at the end of Section IV.

An intrusion detection method based on feature selection and support vector machine

Two-level Feature Selection Method Based on SVM for Intrusion Detection

Intrusion Detection System Based on Feature Selection and Support Vector Machine

An Intrusion Detection Feature Selection Method Based on Improved Mutual Information

Intrusion Detection System Based on Choosing Characters and Weighting Characters

A Combined Feature Selection Method Based on Clustering in Intrusion Detection

Building lightweight intrusion detection system using wrapper-based feature selection mechanisms

SSA-CFS:An Effective Feature Selection Method for Intrusion Detection System

A Method to Extract Network Intrusion Detection Feature

An Intrusion Detection Method Based on Principal Component Analysis and Decision Tree

A FEATURE SELECTION ALGORITHM DESIGN AND ITS IMPLEMENTA-TION IN INTRUSION DETECTION SYSTEM

Feature selection for intrusion detection systems

A Research and Application of Feature Selection Based on KNN and Tabu Search Algorithm in the Intrusion Detection

A Hybrid Intrusion Detection with Decision Tree for Feature Selection

Feature Selection Based on Rough Set and Modified Genetic Algorithm for Intrusion Detection

Intrusion Detection System Based on RF-SVM Model Optimized with Feature Selection

Machine Learning-Based Intrusion Detection: Feature Selection versus Feature Extraction

Fs-Svm Based Intrusion Detection Feature Selection And Classification

Relevant Feature Selection Model Using Data Mining for Intrusion Detection System

An Improved Feature Selection Algorithm Based on MAHALANOBIS Distance for Network Intrusion Detection

An adaptive network intrusion detection method based on PCA and support vector machines