A Network Intrusion Detection Model Based on Multi-Layer Feature Extraction
Xueting Yao, Yue Zhou, Hong Wang · 2025
This paper presents a network intrusion detection model based on multi-layer feature extraction. With the CIC-IDS 2017 open dataset serving as experimental data, the dense connection mechanism of OSA blocks is incorporated on the basis of the convolutional neural network to conduct local extraction of underlying features. The cost sensitive matrix method is adopted to address the issue of low detection model efficacy resulting from dataset imbalance. The outcomes demonstrate that the model is significantly enhanced in extracting model features and improving detection results.