Obey validity limits of data-driven models through topological data analysis and one-class classification

• Published in: Optimization and engineering Vol. 23; no. 2; pp. 855 - 876
• Main Authors: Schweidtmann, Artur M., Weber, Jana M., Wende, Christian, Netze, Linus, Mitsos, Alexander
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.06.2022; Springer Nature B.V
• Subjects: Clusters; Computational geometry; Control; Convexity; Data analysis; Design optimization; Domains; Engineering; Environmental Management; Financial Engineering; Homology; Mathematics; Mathematics and Statistics; Operations Research/Decision Theory; Optimization; Research Article; Support vector machines; Systems Theory; Topology; Training; Validity; Persistent homology; Topological data analysis; One-class support vector machine; Machine-learning; Deterministic global optimization

Data-driven models are becoming increasingly popular in engineering, on their own or in combination with mechanistic models. Commonly, the trained models are subsequently used in model-based optimization of design and/or operation of processes. Thus, it is critical to ensure that data-driven models...