Machine Learning Based Intrusion Detection for Virtualized Infrastructures

Abstract

Recent times have seen a steady shift of technology from traditional software models to the cloud. The substantial growth in the number of applications using cloud based infrastructures calls for the need of security mechanisms for their protection. Intrusion detection systems are one of the most suitable security solutions for protecting cloud based environments. Although there are several approaches to intrusion detection, such as signature-based and anomaly-based, machine learning (ML) based approaches have emerged as a recent interest and research area. With their robust learning models, and data centric approach, ML based security solutions for cloud environments have been proven effective. Attack features are extracted from network and application logs. Attack presence is confirmed by performing Machine learning techniques such as logistic regression and belief propagation. Performance measures such as average detection time is used to evaluate the performance of the approach.