Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Liu, Wei, Chatzi, Eleni, Lai, Zhilu
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2509.19830
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866915652696014848
author Liu, Wei
Chatzi, Eleni
Lai, Zhilu
author_facet Liu, Wei
Chatzi, Eleni
Lai, Zhilu
contents Kolmogorov-Arnold Networks (KANs) offer a structured and interpretable framework for multivariate function approximation by composing univariate transformations through additive or multiplicative aggregation. This paper establishes theoretical convergence guarantees for KANs when the univariate components are represented by B-splines. We prove that both additive and hybrid additive-multiplicative KANs attain the minimax-optimal convergence rate $O(n^{-2r/(2r+1)})$ for functions in Sobolev spaces of smoothness $r$. We further derive guidelines for selecting the optimal number of knots in the B-splines. The theory is supported by simulation studies that confirm the predicted convergence rates. These results provide a theoretical foundation for using KANs in nonparametric regression and highlight their potential as a structured alternative to existing methods.
format Preprint
id arxiv_https___arxiv_org_abs_2509_19830
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle On the Rate of Convergence of Kolmogorov-Arnold Network Regression Estimators
Liu, Wei
Chatzi, Eleni
Lai, Zhilu
Machine Learning
Artificial Intelligence
Kolmogorov-Arnold Networks (KANs) offer a structured and interpretable framework for multivariate function approximation by composing univariate transformations through additive or multiplicative aggregation. This paper establishes theoretical convergence guarantees for KANs when the univariate components are represented by B-splines. We prove that both additive and hybrid additive-multiplicative KANs attain the minimax-optimal convergence rate $O(n^{-2r/(2r+1)})$ for functions in Sobolev spaces of smoothness $r$. We further derive guidelines for selecting the optimal number of knots in the B-splines. The theory is supported by simulation studies that confirm the predicted convergence rates. These results provide a theoretical foundation for using KANs in nonparametric regression and highlight their potential as a structured alternative to existing methods.
title On the Rate of Convergence of Kolmogorov-Arnold Network Regression Estimators
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2509.19830