A Powerful Ensemble Learning Approach for Improving Network Intrusion Detection System (NIDS)

Intrusion Detection Systems (IDSs) take a leading role in detecting and monitoring external and internal cyberattacks over all internet technologies. However, as the data is enormously increasing annually on the internet, several advanced and unknown intrusions are also dramatically increasing. Hence, the task of an intrusion detection system will get more challenging face to the current security concerns. Machine learning algorithms can have an important potential in developing and updating the performance of an IDS, particularly the ensemble learning that has received a special attention in recent decades. Based on the latter, this paper proposed an effective approach to maximize the detection accuracy and deal with the limitations of an IDS. Various powerful machine learning classifiers have been considered in this contribution, by designing and comparing the accuracy of 5 different ensemble learning models with the selection of 10 important features that have a direct impact on the target variable. The experiments have been conducted on the NSL-KDD data set and the outcomings show that our proposed method provides a strong network security and a robust performance, compared to other existing results in terms of accuracy, precision, and recall, as the accuracy of all the five models proposed is higher than ninety-nine per cent.