Scholarlink

In the field of cybersecurity, Intrusion Detection System (IDS) plays an important role in securing networks as well as connected devices against malicious activities. Due to developing technologies, complications of cyber threats also increase. To tackle this complication, advanced Machine Learning (ML) techniques are used that can enhance the efficiency of IDS. This paper explores as well as analyses the performance of various ML techniques like Logistic Regression (LR), Support Vector Classifier (SVC), Naives Bayes (NB), Decision Tree (DT) and XGBoost on globally used KDDCUP99 Dataset. The evaluation parameters that are used are overall accuracy, precision and recall to explore the best suitable algorithms. In this study, XGBoost performs better compared to all the other algorithms on all evaluation parameters such as overall accuracy, precision and recall. This shows the efficiency of XGBoost in detecting patterns in complex data making it the more advanced choice for IDS. This study highlights the significance of machine learning algorithm choice in developing proactive IDS. By exploiting the robustness of ML-like Logistic Regression, Support Vector Classifier, Naives Bayes, Decision Tree and XGBoost. Organizations can enhance their cyber defence mechanism and mitigate the risk posed by evolving cyber threats.

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

Adversarial Attack Detection in Intrusion Detection System Using Machine Learning

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Sanjeev Kumar,
  • Saurabh Bharti,
  • Rajiv Singh,
  • Kapil Kumar,
  • Manju Khari

摘要

In the field of cybersecurity, Intrusion Detection System (IDS) plays an important role in securing networks as well as connected devices against malicious activities. Due to developing technologies, complications of cyber threats also increase. To tackle this complication, advanced Machine Learning (ML) techniques are used that can enhance the efficiency of IDS. This paper explores as well as analyses the performance of various ML techniques like Logistic Regression (LR), Support Vector Classifier (SVC), Naives Bayes (NB), Decision Tree (DT) and XGBoost on globally used KDDCUP99 Dataset. The evaluation parameters that are used are overall accuracy, precision and recall to explore the best suitable algorithms. In this study, XGBoost performs better compared to all the other algorithms on all evaluation parameters such as overall accuracy, precision and recall. This shows the efficiency of XGBoost in detecting patterns in complex data making it the more advanced choice for IDS. This study highlights the significance of machine learning algorithm choice in developing proactive IDS. By exploiting the robustness of ML-like Logistic Regression, Support Vector Classifier, Naives Bayes, Decision Tree and XGBoost. Organizations can enhance their cyber defence mechanism and mitigate the risk posed by evolving cyber threats.