Graph neural network-based identification of vulnerable regions in spatial complex networks via virtual node model

Key Points

• Identification of vulnerable regions enhances the resilience of spatial networks, enabling better monitoring.
• A significant improvement of 25% in prediction accuracy was achieved through the virtual node model approach.
• Employing graph neural networks reveals intricate relationships in spatial complex networks via network analysis.
• Findings support the need for advanced tools in network management, particularly in monitoring critical infrastructures.