English

Integral Representations of Sobolev Spaces via ReLU$^k$ Activation Function and Optimal Error Estimates for Linearized Networks

Numerical Analysis 2025-05-13 v2 Numerical Analysis

Abstract

This paper presents two main theoretical results concerning shallow neural networks with ReLUk^k activation functions. We establish a novel integral representation for Sobolev spaces, showing that every function in Hd+2k+12(Ω)H^{\frac{d+2k+1}{2}}(\Omega) can be expressed as an L2L^2-weighted integral of ReLUk^k ridge functions over the unit sphere. This result mirrors the known representation of Barron spaces and highlights a fundamental connection between Sobolev regularity and neural network representations. Moreover, we prove that linearized shallow networks -- constructed by fixed inner parameters and optimizing only the linear coefficients -- achieve optimal approximation rates O(n122k+12d)O(n^{-\frac{1}{2}-\frac{2k+1}{2d}}) in Sobolev spaces.

Keywords

Cite

@article{arxiv.2505.00351,
  title  = {Integral Representations of Sobolev Spaces via ReLU$^k$ Activation Function and Optimal Error Estimates for Linearized Networks},
  author = {Xinliang Liu and Tong Mao and Jinchao Xu},
  journal= {arXiv preprint arXiv:2505.00351},
  year   = {2025}
}