دراسة تأثير استخلاص الميزات باستخدام نهج التعلم الجماعي للكشف عن هجمات الأمن السيبراني

Abstract

Cybersecurity has become a critical area in the digital age, as the increasing reliance on Internet-based systems and the proliferation of Internet of Things (IOT) devices expose individuals and organizations to cyberattacks. The emergence of sophisticated cyberattacks has bypassed traditional security measures, making it necessary to develop advanced tools to detect and mitigate these threats. Expert systems and machine learning algorithms are currently widely used in the field of network intrusion detection.

In this research, cybersecurity attacks included in the CSE-CIC-IDS2018 database were detected using an ensemble approach in the feature extraction process and using four classification algorithms, namely Desion Tree (DT), Random Forest (RF), Naïve Bayes (NB), and Regression Linear (RL). The research concluded that using the ensemble approach gave the best values ​​for the performance metrics Auc and F1-Score instead of using one extraction technique, and the RF classification algorithm gave the highest values ​​compared to other classification algorithms and single feature extraction techniques used in previous studies.