Privacy preservation model for scalable and reliable anomaly detection in heterogeneous cloud data using machine learning algorithms

Abstract

Modern business and other services depend on cloud infrastructure to provide scalability and reliability into their operations. These operations involve storing and retrieving a huge volume of data that can transform the human life cycle, business, education and other relevant fields that can change the world in significant ways. Intrusion detection is the process of identifying the intrusions in cloud infrastructure, which violates the security policies and standards. The detection process is based on the assumptions of the behaviours related to the non-legitimate users whose actions are abnormal than the normal one; which facilitates the detection of non-authorized activities. Detecting and handling such anomalies leads to expected system reliability as they are detected and handled before the damage occurs. The machine learning algorithms act in such a way that observations are collected from historical data, to predict the future course of action from the knowledge the system has gained, which is useful to detect the abnormal behaviour. newlineThe main focus of this thesis is to explore the machine learning algorithms, which have been used to detect the anomalies in realistic scenarios. More specially, one of the goals of this work is to develop a secure anomaly detection framework with better performance in terms of execution time and newline