Secure Data In Cloud Using Machine Learning Based On Intrusion Detection System

Abstract

Huge amounts of network traffic are produced daily due to the introduction of new technologies like cloud computing and big data, and the intrusion detection system must dynamically gather and evaluate the data produced by the incoming traffic. However, not all features in a huge dataset help describe the traffic, so limiting and choosing only a few suitable features may increase the intrusion detection system's speed and accuracy. Based on the detection rate each feature has established throughout the selection process, a feature selection method has been developed in this study to remove irrelevant features and identify the features that would help enhance the detection rate. Recursive feature reduction was used in association with a decision tree-based classifier to accomplish that goal, and the appropriate relevant features were later found. This methodology was utilized in this research, along with scikit-learn, a machine learning toolkit built in Python, to analyze the NSL-KDD dataset, an upgraded version of the KDD 1999 dataset. This method allowed for the identification of appropriate features  within the dataset, improving accuracy rates. Machine learning uses the C4.5 classification decision tree method, which is based on the ID3 algorithm. The assumption that feature selection considerably enhances classifier performance is supported by these findings.