An Effective Model for Binary and Multi-classification Based on RFE and XGBoost Methods in Intrusion Detection System

Incursion detection is an important network security tool that monitors and identifies any network intrusion. The present context is insufficient to manage network security threats, which with Internet use are growing fast. However, the analytical data generated regularly by computer networks are often extremely large size. This poses a significant problem for IDSs, who must look at all aspects of the data to find invasive trends. In this paper, an effective model has been presented for binary and multiclassification problems based on RFE and XGBoost methods in an intrusion detection system. We used a method called RFE to pick important features, and then we used a classifier called XGBoost to sort things into categories. We tested all of this on a dataset called NSL-KDD. These experiments have been performed with features (i.e., 9 and 5) and without features. The findings showed improved precision in the five and binary categorization with a reduced false alarm rate. For the 5 and binary categorization accordingly, the suggested method obtained an accuracy of 98, 75, and 100%.