Heterogeneous Temporal Graph Neural Networks in Cybersecurity

This study explores HT-GNNs in cybersecurity, tackling challenges through temporal attention and event pattern encoding. Leveraging the THEIA dataset, it introduces novel architecture design and fusion methodologies. Key findings include enhanced anomaly detection performance, critical pattern analysis, and scalability evaluation. The research offers modular and precise detection methods, laying a foundation for advancements in cybersecurity graph-based analytics.