Estimates on the generalization error of physics-informed neural networks for approximating PDEs
IMA Journal of Numerical Analysis · 02.02.2023 · 43 (1) · 1-43
DOI: 10.1093/imanum/drab093 ↗Kurz gefasst
Fachartikel in der Zeitschrift IMA Journal of Numerical Analysis, online erschienen am 14.01.2022. Verfasst von Siddhartha Mishra und 1 weiteren Autor:innen. Die Arbeit wurde bisher 238-mal zitiert (Crossref) und nennt 46 Literaturangaben. Gefördert von: European Research Council Consolidator.
Abstract (Originalsprache)
Physics-informed neural networks (PINNs) have recently been widely used for robust and accurate approximation of partial differential equations (PDEs). We provide upper bounds on the generalization error of PINNs approximating solutions of the forward problem for PDEs. An abstract formalism is introduced and stability properties of the underlying PDE are leveraged to derive an estimate for the generalization error in terms of the training error and number of training samples. This abstract framework is illustrated with several examples of nonlinear PDEs. Numerical experiments, validating the proposed theory, are also presented.
Förderung
- European Research Council Consolidator
Zitieren
APA
Mishra, S., & Molinaro, R. (2023). Estimates on the generalization error of physics-informed neural networks for approximating PDEs. IMA Journal of Numerical Analysis, 43(1), 1-43. https://doi.org/10.1093/imanum/drab093
Vancouver
Mishra S, Molinaro R. Estimates on the generalization error of physics-informed neural networks for approximating PDEs. IMA Journal of Numerical Analysis. 2023;43(1):1-43. doi:10.1093/imanum/drab093
BibTeX
@article{Mishra2023,
author = {Mishra, Siddhartha and Molinaro, Roberto},
title = {Estimates on the generalization error of physics-informed neural networks for approximating PDEs},
journal = {IMA Journal of Numerical Analysis},
year = {2023},
volume = {43},
number = {1},
pages = {1--43},
doi = {10.1093/imanum/drab093}
}
EUVIMED