EML Trees Are Universal Approximators

DOI10.48550/arXiv.2606.23179MaRDI QIDQ7269908arXivFDO

Authors Joe Germany, Elie Abdo, Joseph Bakarji

Publication date 22 June 2026

Published in arXiv

Tldr This paper analyzes tree structures formed from Exp minus log functions to prove they achieve universal approximation capabilities for continuous mathematical functions within Sobolev spaces by explicitly mimicking polynomials. Furthermore the authors propose a learning algorithm equipped with fitting parameters to demonstrate practical feasibility in solving optimization problems using this grounded theoretical framework effectively.

Full work available at URL https://doi.org/10.48550/arXiv.2606.23179

arXiv classification

cs.LGcs.NEcs.SCmath.NA

This page was built for publication: EML Trees Are Universal Approximators

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7269908)