Anomaly Detection@DLR-SC

Kurzfassung

The large dataset growth in aerospace applications necessitates for advanced automatic evaluation algorithms. In this talk, we focus on machine learning algorithms to detect anomalous events. First, we start with a survey on the general problem setting and describe popular algorithms in this field (e.g. density-based anomaly detection, neural network-based approaches). Then, we use these algorithms to detect anomalies in datasets from combustion experiments, earth observations and aeromedical tests that have been recently performed at the German Aerospace Center (DLR).