Abstract
Anomaly intrusion detection technologies are essential for network and computer security as the threat gets more serious yearly. Ensemble learning techniques are promising machine learning methods in anomaly detection that aim to produce multiple models and combine their output in a specific manner to obtain a perfect attack detection. However, it’s still difficult to choose an appropriate ensemble method for a particular dataset. This research is conducted on entry-disciplinary concept in which the knowledge is transferred between network security and machine learning. Thus, the problematic of anomaly detection in network traffic is considered, and two novel ensemble methods for anomaly detection is presented. In both methods, the decision rule (henceforth, Rule-set) which is extracted from two different families of classifiers Naïve Bayes and decision tree J48 will be used as an ensemble constitute classifiers. In the first method, a set of Rule Evaluation Metrics (henceforth, REMs) extracted from Rule-sets will be used for combining classifiers and solving rules conflict whenever occurred. While in the second method the paper presents a novel stacking approach as follows: a cover property of Rule-sets will be utilized to re-encode training instances and produce metadata set that is used for training a meta-level classifier which produces the ultimate result. The evaluation of the proposed methods will be conducted on CICIDS2017 dataset in a term of detection rate, execution time, false alarm rate, accuracy, and other interesting measures. The experimental results show attests their superiority of accuracy that reaches 99.8630% and 99.8642% for first and second methods respectively and lower execution time for both methods especially for the second proposed method, which is (0.25 s).
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.