P4 - Redundancy-orchestrated Information Fusion Exploiting Sensor Redundancy for Improved Model Robustness
- Event
- iCCC2024 - iCampµs Cottbus Conference
2024-05-14 - 2024-05-16
Cottbus - Band
- Poster
- Chapter
- Condition Monitoring
- Author(s)
- C. Holst - inIT – Institut für industrielle Informationstechnik, Technische Hochschule Ostwestfalen-Lippe, Lemgo
- Pages
- 131 - 131
- DOI
- 10.5162/iCCC2024/P4
- ISBN
- 978-3-910600-00-3
- Price
- free
Abstract
Sensor defects are a common occurrence in technical and industrial systems, manifesting as deviations in sensor measurements due to factors such as ageing, wear and tear, or environmental effects [1]. These defects introduce outliers, noise, offsets, or drift. Sensor defects significantly impact machine learning (ML) models, as data distributions increasingly deviate from the training data with the severity of the defect, making them out-of-distribution. Because ML models
are limited to the knowledge obtained from their training data, out-of-distribution data causes a decline in the accuracy of their predictions or classifications [1, 2]...