16
0

Embedding Inequalities for Barron-type Spaces

Abstract

An important problem in machine learning theory is to understand the approximation and generalization properties of two-layer neural networks in high dimensions. To this end, researchers have introduced the Barron space Bs(Ω)\mathcal{B}_s(\Omega) and the spectral Barron space Fs(Ω)\mathcal{F}_s(\Omega), where the index s[0,)s\in [0,\infty) indicates the smoothness of functions within these spaces and ΩRd\Omega\subset\mathbb{R}^d denotes the input domain. However, the precise relationship between the two types of Barron spaces remains unclear. In this paper, we establish a continuous embedding between them as implied by the following inequality: for any δ(0,1),sN+\delta\in (0,1), s\in \mathbb{N}^{+} and f:ΩRf: \Omega \mapsto\mathbb{R}, it holds that \[ \delta \|f\|_{\mathcal{F}_{s-\delta}(\Omega)}\lesssim_s \|f\|_{\mathcal{B}_s(\Omega)}\lesssim_s \|f\|_{\mathcal{F}_{s+1}(\Omega)}. \] Importantly, the constants do not depend on the input dimension dd, suggesting that the embedding is effective in high dimensions. Moreover, we also show that the lower and upper bound are both tight.

View on arXiv
Comments on this paper