Addressing common misinterpretations of KART and UAT in neural network literature

Détail du document

Identifiant

oai:arXiv.org:2408.16389

Sujet

Computer Science - Machine Learnin... Computer Science - Neural and Evol... 26B40, 41A30, 41A63, 68T05

Auteur

Ismailov, Vugar

Catégorie

Computer Science

Année

2024

Date de référencement

18/12/2024

Mots clés

uat kart network

Métrique

Résumé

This note addresses the Kolmogorov-Arnold Representation Theorem (KART) and the Universal Approximation Theorem (UAT), focusing on their common misinterpretations in some papers related to neural network approximation.

Our remarks aim to support a more accurate understanding of KART and UAT among neural network specialists.

;Comment: 10 pages; a section, two theorems and several references added

Ismailov, Vugar, 2024, Addressing common misinterpretations of KART and UAT in neural network literature

Document

Ouvrir

Source