Voir la notice de l'article provenant de la source Numdam
We start from the contractive functional equation proposed in [4], where it was shown that the polynomial solution of functional equation can be used to initialize a Neural Network structure, with a controlled accuracy. We propose a novel algorithm, where the functional equation is solved with a converging iterative algorithm which can be realized as a Machine Learning training method iteratively with respect to the number of layers. The proof of convergence is performed with respect to the norm. Numerical tests illustrate the theory and show that stochastic gradient descent methods can be used with good accuracy for this problem.
Després, Bruno 1
CC-BY 4.0
@article{CRMATH_2023__361_G6_1029_0,
author = {Despr\'es, Bruno},
title = {A convergent {Deep} {Learning} algorithm for approximation of polynomials},
journal = {Comptes Rendus. Math\'ematique},
pages = {1029--1040},
publisher = {Acad\'emie des sciences, Paris},
volume = {361},
number = {G6},
year = {2023},
doi = {10.5802/crmath.462},
language = {en},
url = {http://geodesic.mathdoc.fr/articles/10.5802/crmath.462/}
}
TY - JOUR AU - Després, Bruno TI - A convergent Deep Learning algorithm for approximation of polynomials JO - Comptes Rendus. Mathématique PY - 2023 SP - 1029 EP - 1040 VL - 361 IS - G6 PB - Académie des sciences, Paris UR - http://geodesic.mathdoc.fr/articles/10.5802/crmath.462/ DO - 10.5802/crmath.462 LA - en ID - CRMATH_2023__361_G6_1029_0 ER -
%0 Journal Article %A Després, Bruno %T A convergent Deep Learning algorithm for approximation of polynomials %J Comptes Rendus. Mathématique %D 2023 %P 1029-1040 %V 361 %N G6 %I Académie des sciences, Paris %U http://geodesic.mathdoc.fr/articles/10.5802/crmath.462/ %R 10.5802/crmath.462 %G en %F CRMATH_2023__361_G6_1029_0
Després, Bruno. A convergent Deep Learning algorithm for approximation of polynomials. Comptes Rendus. Mathématique, Tome 361 (2023) no. G6, pp. 1029-1040. doi: 10.5802/crmath.462
Cité par Sources :