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

Download:

Current browse context:

cs.LG
< prev | next >
new | recent | 2201

Change to browse by:

cs
math
math.FA

References & Citations

DBLP - CS Bibliography

listing | bibtex

Bookmark

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

Computer Science > Machine Learning

Title: Deep Neural Network Approximation For Hölder Functions

Authors: Ahmed Abdeljawad
(Submitted on 11 Jan 2022 (v1), last revised 14 Aug 2022 (this version, v2))
Abstract: In this work, we explore the approximation capability of deep Rectified Quadratic Unit neural networks for H\"older-regular functions, with respect to the uniform norm. We find that theoretical approximation heavily depends on the selected activation function in the neural network.
Comments: This is the second version. Several changes have been performed. 22 pages, 2 figures
Subjects: Machine Learning (cs.LG); Functional Analysis (math.FA)
MSC classes: Primary: 68T07, 46E35 Secondary: 41A30, 46E30
Cite as: arXiv:2201.03747 [cs.LG]
  (or arXiv:2201.03747v2 [cs.LG] for this version)

Submission history

From: Ahmed Abdeljawad [view email]
[v1] Tue, 11 Jan 2022 02:26:55 GMT (25kb)
[v2] Sun, 14 Aug 2022 17:09:05 GMT (25kb)
Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)

Link back to: arXiv, form interface, contact.