Research on Intelligent Network Intrusion Detection Based on Gray and Statistical Features
Hong Chen, Sisi Liu, Xiaoshu Yuan, Feng Li, Yuanyuan Huang · 2024
This paper presents an intelligent network intrusion detection method based on gray and statistical features to enhance network security. As information technology advances, network attacks have become increasingly severe. Artificial intelligence, particularly deep learning, shows significant potential in this area. The method constructs a deep learning model by analyzing statistical characteristics of network traffic data and applying gray image processing, enabling effective detection of network attacks. Research indicates that advancements in deep learning for image recognition offer new insights for network defense. By calculating statistics such as mean, variance, and standard deviation, the model extracts important information about data distribution. Experimental results demonstrate that this method achieves high accuracy on the CICIDS2017 dataset, outperforming traditional convolutional neural networks by 7 percentage points. Additionally, the image conversion method using statistical features surpasses the original byte stream conversion in accuracy and F1 score, highlighting its effectiveness in network intrusion detection.