Saved in:
Bibliographic Details
Main Authors: Lu, Jian, Huang, Xiaohuang
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2510.18388
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912662696230912
author Lu, Jian
Huang, Xiaohuang
author_facet Lu, Jian
Huang, Xiaohuang
contents This paper investigates the approximation properties of shallow neural networks with activation functions that are powers of exponential functions. It focuses on the dependence of the approximation rate on the dimension and the smoothness of the function being approximated within the Barron function space. We examine the approximation rates of ReLU$^{k}$ activation functions, proving that the optimal rate cannot be achieved under $\ell^{1}$-bounded coefficients or insufficient smoothness conditions. We also establish optimal approximation rates in various norms for functions in Barron spaces and Sobolev spaces, confirming the curse of dimensionality. Our results clarify the limits of shallow neural networks' approximation capabilities and offer insights into the selection of activation functions and network structures.
format Preprint
id arxiv_https___arxiv_org_abs_2510_18388
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Approximation Rates of Shallow Neural Networks: Barron Spaces, Activation Functions and Optimality Analysis
Lu, Jian
Huang, Xiaohuang
Machine Learning
41A46
This paper investigates the approximation properties of shallow neural networks with activation functions that are powers of exponential functions. It focuses on the dependence of the approximation rate on the dimension and the smoothness of the function being approximated within the Barron function space. We examine the approximation rates of ReLU$^{k}$ activation functions, proving that the optimal rate cannot be achieved under $\ell^{1}$-bounded coefficients or insufficient smoothness conditions. We also establish optimal approximation rates in various norms for functions in Barron spaces and Sobolev spaces, confirming the curse of dimensionality. Our results clarify the limits of shallow neural networks' approximation capabilities and offer insights into the selection of activation functions and network structures.
title Approximation Rates of Shallow Neural Networks: Barron Spaces, Activation Functions and Optimality Analysis
topic Machine Learning
41A46
url https://arxiv.org/abs/2510.18388