Cyber-Attack Mitigation in Cloud-Fog Environment Using an Ensemble Machine Learning Model

Abstract

Since the use of Cloud technologies has spread exponentially in the world, the use of Network intrusion detection system has become a field of vital importance in Cyber Security: with the endless growth of network traffic and the spread of new methods of attack, this type of technology has become a must that cloud environments cannot afford to ignore. The proposed approach of this work is based on machine learning and anomaly detection techniques highlights how the deep learning approach turns out to be the best weapon to identify and isolate this type of malicious attacks, surpassing in precision and accuracy approaches of pattern recognition and anomaly detection approaches more traditional like Support Vector Machine (SVM) or Decision Tree (DT). The obtained values of accuracy, precision and recall let us understand on which classes the model is able to be further improved, increasing even more already excellent values of predictions and instead underline the classes in which the model need of being improved with training data more distributed in classes that are performing below the average.