<![CDATA[The growing volume and complexity of network traffic have created new challenges in maintaining information security. Conventional signature-based intrusion detection systems are inadequate against modern threats, especially zero-day attacks that remain undocumented. Anomaly-based approaches using classical machine learning methods such as Support Vector Machine (SVM) show promise but still rely on manual feature engineering, which is time-consuming and requires expertise. This study proposes an anomaly detection system combining the automatic feature extraction capability of Convolutional Neural Network (CNN) with the strong classification performance of SVM. The NSL-KDD dataset is used for training, while real-time testing data are captured using Scapy. The system updates its analysis every five minutes, and detection results are presented as graphical reports and log tables sent to administrators via a Telegram Bot. Experimental results show that the hybrid CNN–SVM model achieves high accuracy and stable performance in real-time scenarios, contributing to more adaptive and intelligent intrusion detection.]]>
Copyrights © 2025
