{"id":"arxiv/2602.12482","url":"https://arxiv.org/abs/2602.12482","tags":["cs.LG","math.CA"],"score":0,"title":"Geometric separation and constructive universal approximation with two hidden layers","domain":"cs.LG","source":"arxiv","authors":["Chanyoung Sung"],"reviews":0,"subject":"cs.LG","summary":"We give a geometric construction of neural networks that separate disjoint compact subsets of $\\Bbb R^n$, and use it to obtain a constructive universal approximation theorem. Specifically, we show that networks with two hidden layers and either a sigmoidal activation (i.e., strictly monotone bounded continuous) or the ReLU activation can approximate any real-valued continuous function on an arbitrary compact set $K\\subset\\Bbb R^n$ to any prescribed accuracy in the uniform norm. For finite $K$, the construction simplifies and yields a sharp depth-2 (single hidden layer) approximation result.","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-02-12","reactions":{}}