Scholarlink

Ensuring network security is a challenging endeavor. One aspect of cybersecurity instrument that monitors network traffic for indications of harmful actions, unauthorized entry, or breaches of safeguarding is known as Intrusion Detection System (IDS). The primary objective of an IDS is to promptly address security incidents to protect network resources and computer systems. Various detection methods that rely on signatures, anomalies, and behavior using machine learning have been proposed. In this research, we have enhanced the Naive Bayes algorithm to identify network breaches by utilizing XGBoost, an extreme gradient boosting machine learning technique. For our experiments, we employed two benchmark datasets: NSL-KDD and UNSW-NB15. Our experiment of ensemble technique with ANOVA F-test gives better result regarding accuracy in both datasets than single classifier Naive Bayes. Furthermore, our experiments demonstrate that the accuracy of the ensemble technique is approximately 95% (DoS), 99% (Probe), 97% (R2L), 99% (U2R) in NSL-KDD and 93% in UNSW-NB15.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhancing Intrusion Detection with Ensemble Learning: Naive Bayes and XGBoost on Network Traffic Data

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Moumita Shib,
  • Md. Kamrul Hassan Khan,
  • Suhrid Talukder,
  • A. R. M. Mahamudul Hasan Rana,
  • Ratnadip Kuri,
  • Md. Hasnat Riaz,
  • Md. Rahid Parvez

摘要

Ensuring network security is a challenging endeavor. One aspect of cybersecurity instrument that monitors network traffic for indications of harmful actions, unauthorized entry, or breaches of safeguarding is known as Intrusion Detection System (IDS). The primary objective of an IDS is to promptly address security incidents to protect network resources and computer systems. Various detection methods that rely on signatures, anomalies, and behavior using machine learning have been proposed. In this research, we have enhanced the Naive Bayes algorithm to identify network breaches by utilizing XGBoost, an extreme gradient boosting machine learning technique. For our experiments, we employed two benchmark datasets: NSL-KDD and UNSW-NB15. Our experiment of ensemble technique with ANOVA F-test gives better result regarding accuracy in both datasets than single classifier Naive Bayes. Furthermore, our experiments demonstrate that the accuracy of the ensemble technique is approximately 95% (DoS), 99% (Probe), 97% (R2L), 99% (U2R) in NSL-KDD and 93% in UNSW-NB15.