ClouDens: Operational Context-Aware Anomaly Detection for Large-scale Cloud System Monitoring
ClouDens uses telemetry-log context to group signals and model service dependencies before looking for anomalies.
The paper studies IBM Cloud Console telemetry logs, where sparse, high-dimensional service signals make anomaly detection hard. Its framework builds domain-guided subsets, creates a context-aware dependency graph, and applies spatio-temporal graph neural networks for forecasting-based detection. On the IBM Cloud Telemetry Dataset, it reports higher NAB scores than a GRU-based model for count-based telemetry features, meaning earlier and more accurate detection in that setting. ArXiv · AI/CL/LG's note
The paper studies IBM Cloud Console telemetry logs, where sparse, high-dimensional service signals make anomaly detection hard. Its framework builds domain-guided subsets, creates a context-aware dependency graph, and applies spatio-temporal graph neural networks for forecasting-based detection. On the IBM Cloud Telemetry Dataset, it reports higher NAB scores than a GRU-based model for count-based telemetry features, meaning earlier and more accurate detection in that setting. ArXiv · AI/CL/LG's note
score 4