Abstract:The universal approximation theorem establishes that neural networks can approximate any continuous function on a compact set. Later works in approximation theory provide quantitative approximation rates for ReLU networks on the class of $\alpha$-Hölder functions $f: [0,1]^N \to \mathbb{R}$. The goal of this paper is to provide similar quantitative approximation results in the context of group equivariant learning, where the learned $\alpha$-Hölder function is known to obey certain group symmetries. While there has been much interest in the literature in understanding the universal approximation properties of equivariant models, very few quantitative approximation results are known for equivariant models.
In this paper, we bridge this gap by deriving quantitative approximation rates for several prominent group-equivariant and invariant architectures. The architectures that we consider include: the permutation-invariant Deep Sets architecture; the permutation-equivariant Sumformer and Transformer architectures; joint invariance to permutations and rigid motions using invariant networks based on frame averaging; and general bi-Lipschitz invariant models. Overall, we show that equally-sized ReLU MLPs and equivariant architectures are equally expressive over equivariant functions. Thus, hard-coding equivariance does not result in a loss of expressivity or approximation power in these models.
| Subjects: | Machine Learning (cs.LG); Numerical Analysis (math.NA) |
| MSC classes: | 68T07, 41A25 |
| Cite as: | arXiv:2602.20370 [cs.LG] |
| (or arXiv:2602.20370v2 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2602.20370 arXiv-issued DOI via DataCite |
From: Jonathan Siegel [view email]
[v1]
Mon, 23 Feb 2026 21:17:46 UTC (55 KB)
[v2]
Thu, 16 Apr 2026 16:10:04 UTC (69 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。