Feature Selection Methods for Intrusion Detection Systems: A Performance Comparison
Sanjay Razdan, Himanshu Gupta, Ashish Seth · 2022 2nd Asian Conference on Innovation in Technology (ASIANCON) · 2022
Intrusion Detection Systems are used in cloud as well as in on-premises networks for detecting the intrusions. For an Intrusion Detection System, it can be computationally expensive and time consuming to process a high dimensional data to detect intrusions. Various filter as well as wrapper methods are used to select the most relevant features from the feature space for the classification. Thus, feature selection methods help to eliminate those features which do not have or have less predictive information. By using feature selection methods, we can make an Intrusion Detection System more efficient. In this paper we have selected and used four feature selection methods on NSL-KDD dataset. The reduced feature set is then used to classify the test data using Support Vector Machine. The significant outcome of this paper is the most efficient feature selection method among those discussed in this paper.