IstGPT: LLM-based Anomaly Detection for Spatial-Temporal Graph in Industrial Systems
Researchers have developed IstGPT, a novel system for detecting anomalies in industrial control systems using large language models and graph learning. This approach models complex sensor-actuator dependencies by integrating operational data, technical documents, and system diagrams to construct a spatial-temporal graph. IstGPT then employs graph neural networks to identify anomalies through reconstruction errors, outperforming 12 existing methods on nine diverse datasets. AI
IMPACT Introduces a new method for anomaly detection in industrial systems, potentially enhancing cybersecurity and operational stability.