A Review on Feature Selection Techniques For Intrusion Detection System

Abstract

Feature Engineering plays a significant role in the development of a Classifier Model based on Machine Learning. It aids in the reduction of dataset dimensions, training time, and computation costs; while improving the model's performance and detection accuracy. The most popular method for lowering the dimensionality of a dataset is to employ feature selection. The bigger the dataset's dimensions, the longer the Machine Learning model will need to handle (train and test) it. Those systems fall under the Intrusion Detection System category, and a system designed to monitor the data transitions and network traffic for unauthorized activities and takes corrective steps, such as generating an alert or warning, to prevent such acts (IDS). This paper discusses some recent methods used to reduce dataset dimensions to optimize performance.

Keywords: Feature Selection, Information Gain, Intrusion Detection, Dimensionality Reduction

JEL Classification: C02