Explainable Multi-Modal Hybrid Framework for Cross-Domain Cyber-Physical Systems Security

This research presents a multi-modal deep learningframework designed to enhance the security of Cyber-PhysicalSystems (CPS). By integrating data from sensors, network traffic,and system logs, the proposed method effectively addressesthe challenge of detecting coordinated attacks that target bothphysical and digital layers. The framework integrates domainadaptation techniques to ensure cross-domain generalizationand explainable AI methods to provide actionable insights foroperators. Comprehensive evaluations on the ToN IoT datasetdemonstrate the effectiveness of the framework, highlightingits potential for real-world applications in securing criticalinfrastructure.