Theory-inspired machine learning—towards a synergy between knowledge and data

• DOI: 10.1007/s40194-022-01270-z
• Publication Date: 2022-04-25T11:02:58Z
• Authors: Johannes G. Hoffer; Andreas B. Ofner; Franz M. Rohrhofer; Mario Lovrić; Roman Kern; Stefanie Lindstaedt; Bernhard C. Geiger
• Keywords: Digitization

Abstract Most engineering domains abound with models derived from first principles that have beenproven to be effective for decades. These are not only a valuable source of knowledge, but they also form the basis simulations. The recent trend digitization has complemented these data in all forms and variants, such as process monitoring time series, measured material characteristics, stored production parameters. Theory-inspired machine learning combines available data, reaping benefits established knowledge capabilities modern, data-driven approaches. Compared purely physics- or models, resulting theory-inspired often more accurate less complex, extrapolate better, allow faster model training inference. In this short survey, we introduce discuss several prominent approaches show how were applied fields welding, joining, additive manufacturing, metal forming.