基于先验约束的网络群体意图变分推理预测方法及装置
By constructing a spatiotemporal interaction graph and node feature matrix, and combining a network security knowledge base and a spatiotemporal graph variational reasoning model, the problem of insufficient multi-source network attack identification capability in existing technologies is solved, and stable and information-rich feature representation and accurate identification of complex collaborative attack behaviors are achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- QIANYUAN NATIONAL LABORATORY
- Filing Date
- 2026-06-10
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies struggle to effectively capture the relationships between network entities when facing multi-source, multi-stage, and cross-temporal cyberattacks, and their detection accuracy decreases when data is sparse or attack behavior is disguised.
By constructing a spatiotemporal interaction graph and node feature matrix, prior feature vectors from a cybersecurity knowledge base are introduced. Node feature representation is then performed using a spatiotemporal graph variational inference model. KL divergence regularization constraints are employed to generate probability distributions and latent feature representations. Finally, a self-attention mechanism is used to aggregate global spatiotemporal dependencies.
It significantly improves detection accuracy and noise resistance in situations with sparse data or disguised attacks, and enables accurate identification of multi-stage collaborative attack behaviors across time and entities.
Smart Images
Figure CN122419976A_ABST