Using Machine Learning Models to Detect Different Intrusion on NSL-KDD

Abstract

While the network brings great social and economic benefits to mankind, the security situation of the network is becoming increasingly severe, and various forms of network attacks occur frequently. This paper uses Python to train machine learning model to improve the processing efficiency of intrusion detection system. By comparing five machine learning models such as SGD Classifier, Ridge Classifier, Decision Tree classifier, Random Forest Classifier, Extra Tree Classifier, the best machine learning model suitable for intrusion detection system is found out. In the experiment, feature selection is used to filter the features of the data. The recursion method was used to eliminate the irrelevant features and the NSL-KDD data set was used to identify the relevant features, which greatly improved the accuracy and reliability of the model. The experimental results show that Random Forest Classifier and Extra Tree Classifier perform well, and the extra tree model can still guarantee high stability and accuracy when dealing with difficult problems. The application of these two models is helpful to build a better intrusion detection system.