{"id":"arxiv/2602.06374","url":"https://arxiv.org/abs/2602.06374","tags":["math.FA","cs.LG"],"score":0,"title":"A Multiplicative Neural Network Architecture: Locality and Regularity of Approximation","domain":"math.FA","source":"arxiv","authors":["Hee-Sun Choi","Beom-Seok Han"],"reviews":0,"subject":"math.FA","summary":"We introduce a multiplicative neural network architecture in which multiplicative interactions constitute the fundamental representation, rather than appearing as auxiliary components within an additive model. We establish a universal approximation theorem for this architecture and analyze its approximation properties in terms of locality and regularity in Bessel potential spaces. To complement the theoretical results, we conduct numerical experiments on representative targets exhibiting sharp transition layers or pointwise loss of higher-order regularity. The experiments focus on the…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-02-06","reactions":{}}