Scholarlink

With cyber-attacks evolving rapidly in sophistication and posing a significant danger to networks, Traditional Signature-based Intrusion Detection Systems (IDS) struggle to keep pace with these advancements. This research investigates machine learning and clustering algorithms for IDS with the CSE-CIC-IDS2018 dataset, established by the Canadian Institute for Cybersecurity (CIC). We conducted two experiments in this study. In the first experiment, we performed binary classification, comparing various machine learning (ML) algorithms, namely, k Nearest Neighbours (kNN), Support Vector Machines (SVM), Bernoulli Naive Bayes (BNB), Logistic Regression (LR), Decision Trees (DT), and Random Forest (RF). In addition to this, we explored the effectiveness of clustering techniques like K-Means Clustering (KMC), Density-Based Spatial Clustering of Applications with Noise (DBSCAN), Fuzzy C-Means (FCM), Intuitionistic Fuzzy C-Means (IFCM), and Picture Fuzzy Clustering (PFC) in identifying anomalies and compare them against the ML algorithms for binary classification. For the second experiment, we analyse the performance of kNN, SVM, Gaussian Naive Bayes (GNB), LR, DT, and RF for their potential in multi-class intrusion detection for classifying different modern attack types within the CSE-CIC-IDS2018. The results indicate that, for binary classification, RF gives the maximum accuracy of 95.4% among the ML techniques and FCM gives the maximum accuracy of 57.9% among the Fuzzy Clustering techniques. For multi-class classification, DT and RF both achieve the highest accuracy of 92.6%.

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

Performance Comparison of Machine Learning and Fuzzy Clustering Techniques for Anomaly-Based Intrusion Detection System

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Madhav Mahajan,
  • Kashan Akram,
  • Kshitij Gupta,
  • Sumedha Seniaray

摘要

With cyber-attacks evolving rapidly in sophistication and posing a significant danger to networks, Traditional Signature-based Intrusion Detection Systems (IDS) struggle to keep pace with these advancements. This research investigates machine learning and clustering algorithms for IDS with the CSE-CIC-IDS2018 dataset, established by the Canadian Institute for Cybersecurity (CIC). We conducted two experiments in this study. In the first experiment, we performed binary classification, comparing various machine learning (ML) algorithms, namely, k Nearest Neighbours (kNN), Support Vector Machines (SVM), Bernoulli Naive Bayes (BNB), Logistic Regression (LR), Decision Trees (DT), and Random Forest (RF). In addition to this, we explored the effectiveness of clustering techniques like K-Means Clustering (KMC), Density-Based Spatial Clustering of Applications with Noise (DBSCAN), Fuzzy C-Means (FCM), Intuitionistic Fuzzy C-Means (IFCM), and Picture Fuzzy Clustering (PFC) in identifying anomalies and compare them against the ML algorithms for binary classification. For the second experiment, we analyse the performance of kNN, SVM, Gaussian Naive Bayes (GNB), LR, DT, and RF for their potential in multi-class intrusion detection for classifying different modern attack types within the CSE-CIC-IDS2018. The results indicate that, for binary classification, RF gives the maximum accuracy of 95.4% among the ML techniques and FCM gives the maximum accuracy of 57.9% among the Fuzzy Clustering techniques. For multi-class classification, DT and RF both achieve the highest accuracy of 92.6%.