Poster: Human-in-the-Loop Anomaly Detection in Industrial Data Streams.

The detection of anomalies in an industrial setting remains an important and open challenge for most manufacturing companies. The potential benefits from the utilization of an anomaly detection system are substantial, as deviations from normal operating conditions can cause downtimes, quailty issues or safety hazards. The main requirements for an anomaly detection system include the selection of the machine learning model applicable to streaming data, providing the explanations of the model’s decision and participation of human operator in the learning process of the model. We have proposed the anomaly detection system, which addresses the above challenges and is applicable in industrial environment.