Graph Neural Networks for Network Intrusion Detection: An IP Behavioral Analysis Perspective
Seon‐Woo Lee, Ju Young Lee, Tae Jin Lee · 2024
In the real-world network environment, there is a coexistence of immense traffic and various complex and persistent attacks. A new approach is desperately needed to effectively address these challenges. This paper proposes the introduction of artificial intelligence technology, specifically Graph Neural Networks (GNNs), which are advantageous for analyzing and understanding complex interactions within network traffic. This approach captures relational information for accurate analysis, thereby enhancing the detection capabilities of network security systems. Moreover, the robustness of this approach in various scenarios has been experimentally demonstrated by artificially increasing packet volume and testing the model.