Optimizing prediction of security data using feature selection and ensembling

Abstract

Network security is becoming increasingly difficult in today's hyperconnected environment and network traffic and infrastructure must be protected since attacks on businesses are increasing. Anomaly based Intrusion Detection System models identify anomalies as a deviation from the expected behavior. With Machine Learning, the system can learn patterns of normal behavior across environments and applications and it offers the ability to find complex correlations in large amounts of data for detecting attacks. Machine learning algorithms working on large datasets with multi attack types increase computational time and create a problem in decision making. In this work, an Intrusion Detection System model with ensemble feature selection technique is developed to reduce large-scale datasets and improve feature selection and accuracy prediction of the model using ensemble machine learning algorithms