Scholarlink

<p>The growing Internet of Communications provides different services and infrastructures to ensure privacy. However, the ever-evolving nature of security threats, such as Trojans, viruses, and other destructive attacks, necessitates a robust response. Ransomware, one of the deadliest data intrusion viruses that target legitimate network servers, is becoming increasingly difficult to identify due to inadequate feature identification and classification models. This leads to significant inaccuracies in identifying intruders by the IDS system. We proposed a novel and advanced Machine Learning (ML) model for identifying ransomware intrusion detection systems to address these challenges. This research introduces Relative Spectral Scaling Feature Selection (RS2FS) combined with a Cascaded Support Vector Machine (CSVM) and Adaptive Network-Based Fuzzy Inference System (ANFIS), which significantly improves ransomware intrusion detection and accuracy performance. The C-score normalization method is initially applied to eliminate null values, followed by the utilization of the Transmission-Intensive Behavior Rate (TIBR) method to determine the feature marginal rate of intrusion features. Subsequently, the RS2FS method is utilized for feature selection based on the behavioral rate. The proposed ANFIS classifier is implemented to detect Ransomware Intrusion, resulting in a simulation with an impressive detection accuracy of 95.5%, sensitivity performance of 94.9%, and specificity performance of 96.7%, outperforming existing methods.</p>

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 ınternet security: a novel ML approach for intrusion detection using RS2FS and cascaded SVM/ANFIS

  • 期刊论文

  • 发表日期：

    2025年1月24日

查看全文
  • S. Gomathi,
  • K. Anitha Kumari

摘要

The growing Internet of Communications provides different services and infrastructures to ensure privacy. However, the ever-evolving nature of security threats, such as Trojans, viruses, and other destructive attacks, necessitates a robust response. Ransomware, one of the deadliest data intrusion viruses that target legitimate network servers, is becoming increasingly difficult to identify due to inadequate feature identification and classification models. This leads to significant inaccuracies in identifying intruders by the IDS system. We proposed a novel and advanced Machine Learning (ML) model for identifying ransomware intrusion detection systems to address these challenges. This research introduces Relative Spectral Scaling Feature Selection (RS2FS) combined with a Cascaded Support Vector Machine (CSVM) and Adaptive Network-Based Fuzzy Inference System (ANFIS), which significantly improves ransomware intrusion detection and accuracy performance. The C-score normalization method is initially applied to eliminate null values, followed by the utilization of the Transmission-Intensive Behavior Rate (TIBR) method to determine the feature marginal rate of intrusion features. Subsequently, the RS2FS method is utilized for feature selection based on the behavioral rate. The proposed ANFIS classifier is implemented to detect Ransomware Intrusion, resulting in a simulation with an impressive detection accuracy of 95.5%, sensitivity performance of 94.9%, and specificity performance of 96.7%, outperforming existing methods.