Please use this identifier to cite or link to this item:
|Publication type:||Conference paper|
|Type of review:||Peer review (publication)|
|Title:||Uncertainty informed anomaly scores with deep learning : robust fault detection with limited data|
Goren Huber, Lilach
|Proceedings:||Proceedings of the 7th European Conference of the Prognostics and Health Management Society 2022|
|Editors of the parent work:||Do, Phuc|
|Conference details:||7th European PHM, Turin, Italy, 6-8 July 2022|
|Series:||PHM Society European Conference|
|Publisher / Ed. Institution:||PHM Society|
|Publisher / Ed. Institution:||State College|
|Subjects:||Deep learning; Anomaly detection; Predictive maintenance; Wind turbines; Uncertainty quantification; Condition based maintenance; Machine learning|
|Subject (DDC):||006: Special computer methods|
|Abstract:||Quantifying the predictive uncertainty of a model is an important ingredient in data-driven decision making. Uncertainty quantification has been gaining interest especially for deep learning models, which are often hard to justify or explain. Various techniques for deep learning based uncertainty estimates have been developed primarily for image classification and segmentation, but also for regression and forecasting tasks. Uncertainty quantification for anomaly detection tasks is still rather limited for image data and has not yet been demonstrated for machine fault detection in PHM applications. In this paper we suggest an approach to derive an uncertainty-informed anomaly score for regression models trained with normal data only. The score is derived using a deep ensemble of probabilistic neural networks for uncertainty quantification. Using an example of wind-turbine fault detection, we demonstrate the superiority of the uncertainty-informed anomaly score over the conventional score. The advantage is particularly clear in an "out-of-distribution" scenario, in which the model is trained with limited data which does not represent all normal regimes that are observed during model deployment.|
|Further description:||Best Paper Award Lizenzangabe: CC BY 3.0 United States|
|Fulltext version:||Published version|
|License (according to publishing contract):||CC BY 3.0: Attribution 3.0 Unported|
|Departement:||School of Engineering|
|Organisational Unit:||Institute of Data Analysis and Process Design (IDP)|
|Published as part of the ZHAW project:||Machine Learning Based Fault Detection for Wind Turbines|
|Appears in collections:||Publikationen School of Engineering|
Files in This Item:
|2022_Zgraggen-Pizza-Goren-Huber_Uncertainty-informed_ProPHME22.pdf||3.56 MB||Adobe PDF|
Show full item record
Zgraggen, J., Pizza, G., & Goren Huber, L. (2022). Uncertainty informed anomaly scores with deep learning : robust fault detection with limited data [Conference paper]. In P. Do, G. Michau, & C. Ezhilarasu (Eds.), Proceedings of the 7th European Conference of the Prognostics and Health Management Society 2022 (Vol. 7, Issue 1, pp. 530–540). PHM Society. https://doi.org/10.36001/phme.2022.v7i1.3342
Zgraggen, J., Pizza, G. and Goren Huber, L. (2022) ‘Uncertainty informed anomaly scores with deep learning : robust fault detection with limited data’, in P. Do, G. Michau, and C. Ezhilarasu (eds) Proceedings of the 7th European Conference of the Prognostics and Health Management Society 2022. State College: PHM Society, pp. 530–540. Available at: https://doi.org/10.36001/phme.2022.v7i1.3342.
J. Zgraggen, G. Pizza, and L. Goren Huber, “Uncertainty informed anomaly scores with deep learning : robust fault detection with limited data,” in Proceedings of the 7th European Conference of the Prognostics and Health Management Society 2022, Jul. 2022, vol. 7, no. 1, pp. 530–540. doi: 10.36001/phme.2022.v7i1.3342.
Zgraggen, Jannik, et al. “Uncertainty Informed Anomaly Scores with Deep Learning : Robust Fault Detection with Limited Data.” Proceedings of the 7th European Conference of the Prognostics and Health Management Society 2022, edited by Phuc Do et al., vol. 7, no. 1, PHM Society, 2022, pp. 530–40, https://doi.org/10.36001/phme.2022.v7i1.3342.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.