Designing a System Using Features Selection Based on Filter Methods and XGBoost Classifier for Detecting Intrusion in Cloud

Recently, with the accelerating growth of using cloud computing and its related services in various aspects, the difficulty of ensuring cybersecurity is always increasing. Thus, it was necessary to use an intrusion detection system based on machine learning ML as one of the best solutions to address these threats, which scans the cloud environment to detect suspicious activity. Data pre-processing emerged as the crucial concept for designing reliable, high-performance ML models as a result of the rapid growth in the volume of data being created and processed daily. Data reduction using feature selection techniques is one of the pre-processing functions. Consequently, in this paper, we seek to develop an effective IDS that has been designed using machine learning approaches, specifically the XGBoost model with a feature selection principle based on filter methods, which include information gain, gain ratio, and chi-square. Our system is tested using the NSL-KDD and Kyoto datasets. Our suggested approach has surpassed the performance of other works as it achieved these values according to the NSL-KDD and Kyoto in terms of accuracy (94.98, 96.60), F-measure (87.91, 94.70), precision (93.20, 95.12), and recall (84.37, 94.77), respectively. Our proposal also performed well compared to other studies in terms of detection rate and false alarm rate based on the classes of the NSL-KDD dataset.