School of Mathematics & Statistics

Statistical Process Monitoring of Networks and Artificial Intelligence

 

This research area focuses on monitoring high-dimensional, dynamic systems—ranging from engineered networks to deep learning models—using statistical process control techniques. The group develops monitoring schemes tailored to neural networks, network autoregressive processes, and graph-based data. Applications include anomaly detection in AI pipelines, surveillance of financial networks, and quality control in manufacturing settings. Emphasis is placed on real-time performance, robustness under nonstationarity, and interpretability of alerts. Recent contributions explore control charts for neural network residuals, online detection of structural breaks, and scalable monitoring in spatial and spatiotemporal domains.