We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

stat.ML

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Statistics > Machine Learning

Title: Neural network augmented inverse problems for PDEs

Abstract: In this paper we show how to augment classical methods for inverse problems with artificial neural networks. The neural network acts as a prior for the coefficient to be estimated from noisy data. Neural networks are global, smooth function approximators and as such they do not require explicit regularization of the error functional to recover smooth solutions and coefficients. We give detailed examples using the Poisson equation in 1, 2, and 3 space dimensions and show that the neural network augmentation is robust with respect to noisy and incomplete data, mesh, and geometry.
Subjects: Machine Learning (stat.ML); Numerical Analysis (math.NA)
Cite as: arXiv:1712.09685 [stat.ML]
  (or arXiv:1712.09685v2 [stat.ML] for this version)

Submission history

From: Jens Berg [view email]
[v1] Wed, 27 Dec 2017 21:05:42 GMT (8944kb,D)
[v2] Fri, 14 Sep 2018 09:17:14 GMT (3491kb,D)

Link back to: arXiv, form interface, contact.