Principal Component Analysis (PCA) for High-Dimensional Data’s Feature Extraction

Abstract

One of the well-knownmethods for reducing the dimensionality of data in unsupervised learning is Principal Component Analysis (PCA). It reduces information loss while improving interpretability. It facilitates the identification of a dataset's most important attributes and makes data visualization in two and three dimensions simple. In this research, we used PCA and SVM (Support Vector Machine) to present a novel model for IDS (Intrusion Detection System). The suggested IDS paradigm has improved IDS accuracy in less time for testing and training.The suggested model's outcomes for dimensionality reduction techniques and classification strategies that employ SVM as the classifier. This classifier is used to evaluate the effectiveness of dimension reduction, and the PCA method is used to compare the detection time and accuracy.  The model is assessed, and five distinct kernels are used to run concurrent numerical simulations for an intrusion detection system that use SVM and PCA. Furthermore, a comparative study of the suggested IDS is carried out with respect to time, the pace at which network efficiency increases and simultaneously the error rates were reduced.