Into the danger zone: stable extrapolation in high-dimensional function and operator learning

DOI10.48550/arXiv.2609.36709MaRDI QIDQ7362009zbMATH OpenarXivFDO

Authors Ben Adcock, Simone Brugiapaglia, Xuemeng Wang

Publication date 29 September 2026

Published in arXiv

Tldr We study regression problems with out-of-distribution generalization by identifying classes of holomorphic functions and operators for which extrapolation is stable at algebraic rates even in the presence of large distribution shifts due to a 'blessing of high dimensionality'.

Full work available at URL https://doi.org/10.48550/arXiv.2609.36709

Mathematics Subject Classification ID

Approximation by polynomials (41A10) Multidimensional problems (41A63) Spaces of differentiable or holomorphic functions on infinite-dimensional spaces (46E50) Computational learning theory (68Q32) Artificial neural networks and deep learning (68T07)

arXiv classification

cs.LGmath.NAstat.ML

This page was built for publication: Into the danger zone: stable extrapolation in high-dimensional function and operator learning

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7362009)