Machine Learning Models with Neutrosophic Numbers for Network Anomaly Detection and Security Defense Technology

Hussein S Al-Khazraji; Ahmed M. Alkhamees; Humam M Al-Doori; Ahmed A. Metwaly; Mohamed eassa; Ahmed Abdelhafeez; Ahmed S. Salama; Ahmad M. Nagm

Authors

Hussein S Al-Khazraji Department of Electrical Power Engineering Technologies, Al-Hussein University College, Karbala, Iraq
Ahmed M. Alkhamees College of Health and Medical Technologies / Department of Anesthesia Technologies, Ahl Al Bayt University Karbala, Iraq, Iraq
Humam M Al-Doori Department of Computer Engineering Techniques, Al-Yarmok University College Diyala, Iraq
Ahmed A. Metwaly Department of Computer Science, Faculty of Computers and Informatics, Zagazig University, Zagazig 44519, Egypt
Mohamed eassa Computer Science Department, Faculty of Information Systems and Computer Science, October 6 University, Giza, 12585, Egypt; Applied Science Research Center. Applied Science Private University, Amman, Jordan
Ahmed Abdelhafeez Computer Science Department, Faculty of Information Systems and Computer Science, October 6 University, Giza, 12585, Egypt; Applied Science Research Center. Applied Science Private University, Amman, Jordan
Ahmed S. Salama Department of Computer Engineering and Electronics, Cairo Higher Institute for Engineering, Computer Science and Management, New Cairo, Egypt
Ahmad M. Nagm Department of Computer Engineering and Electronics, Cairo Higher Institute for Engineering, Computer Science and Management, New Cairo, Egypt

Keywords:

Neutrosophic Number; Security; Network Anomaly Detection; Cybersecurity; Uncertainty.

Abstract

In the dynamic world of cybersecurity, strong solutions are necessary to safeguard
intricate network systems. By looking at network anomaly detection and security protection, this
study investigates how machine learning (ML) might increase digital infrastructure security. We
assess how well critical ML approaches, such as ensemble approaches and supervised learning,
identify anomalies and lessen risks. The examination of ML-based systems integration into
comprehensive security frameworks places a strong emphasis on real-time monitoring and
adaptive responses. Examples from real-world situations highlight how crucial ML is to
improving network security. After, we apply different ML models to the real-world dataset. Then
we use the single-valued Neutrosophic numbers (SVNNs) methodology to evaluate these ML
models and select the best one. We use the multi-criteria decision-making (MCDM) approach to
obtain the criteria weights and rank the ML models using the EDAS method. The results show
that the random forest model is the best ML model under different evaluation matrices.

DOI: 10.5281/zenodo.15122434