Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Considering Reliability of Deep Learning Function to Boost Data Suitability and Anomaly Detection

: Gauerhof, L.; Hagiwara, Y.; Schorn, C.; Trapp, M.


Vieira, M. ; Institute of Electrical and Electronics Engineers -IEEE-; IEEE Computer Society:
IEEE 31th International Symposium on Software Reliability Engineering Workshops, ISSREW 2020. Proceedings : 12-15 October 2020, Coimbra, Portugal
Piscataway, NJ: IEEE, 2020
ISBN: 978-1-7281-7735-9
ISBN: 978-1-7281-7736-6
ISBN: 978-1-7281-9870-5
International Symposium on Software Reliability Engineering (ISSRE) <31, 2020, Online>
International Workshop on Reliability and Security Data Analysis (RSDA) <5, 2020, Online>
Conference Paper
Fraunhofer IKS ()

The increased demand of Deep Neural Networks (DNNs) in safety-critical systems, such as autonomous vehicles, leads to increasing importance of training data suitability. Firstly, we focus on how to extract the relevant data content for ensuring DNN reliability. Then, we identify error categories and propose mitigation measures with emphasis on data suitability. Despite all efforts to boost data suitability, not all possible variations of a real application can be identified. Hence, we analyse the case of unknown out-of-distribution data. In this case, we suggest to complement data suitability with online anomaly detection using FACER that supervises the behaviour of the DNN.