Author:
Spoor Jan Michael,Weber Jens,Ovtcharova Jivka
Abstract
AbstractAlthough many scientists strongly focus on anomaly detection in different applications and domains, there currently exists no universally accepted definition of anomalies and outliers. Using an approach based on control theory and dynamical systems, as well as a definition for anomalies as described by philosophy of science, the authors propose a generalized framework viewing anomalies as key drivers of progress for a better understanding of the dynamical systems around us. By mathematically defining anomalies and delimiting deviations within expectations from completely unforeseen instances, this paper aims to be a contribution to set up a universally accepted definition of anomalies and outliers.
Publisher
Springer International Publishing
Reference19 articles.
1. Aggarwal, C. C.: Outlier Analysis. Springer Science+Business Media, New York (2013)
2. Hodge, V. J., Austin, J.A.: Survey of outlier detection methodologies. Artif. Intell. Rev. 22, 85-126 (2004)
3. Aggarwal, C. C., Sathe, S.: Outlier Ensembles. Springer, Cham (2017)
4. Wang, X., Wang, X., Wilkes M.: New Developments in Unsupervised Outlier Detection - Algorithms and Applications. Springer, Singapore (2021)
5. Spoor, J. M., Weber, J., Ovtcharova, J.: A definition of anomalies, measurements and predictions in dynamical engineering systems for streamlined novelty detection. Accepted for the 8th International Conference on Control, Decision and Information Technologies (CoDIT), Istanbul (2022)