Anomaly Detection @ Scale for AIoT

Anomaly detection is a must have when dealing with data, and IoT data are no exception. Despite well-known approaches – from engineering rules to graph & deep learning - it is still a challenge, It is hard to capture all the anomalies, minimize false positives, cope with the diversity of sensors and metrology issues, and deliver actionable insights at business pace. Reactivity suffers from the volume, diversity, and complexity of time series. Sounds familiar? See how we addressed these problems at large manufacturers: the approach, the method, the technology, and lessons learned.