Matematičeskoe modelirovanie, Tome 10 (1998) no. 3, pp. 117-124
Citer cet article
V. V. Ivanov; B. Purevdorj; I. V. Puzynin. Second-order learning methods for a multilayer perceptron. Matematičeskoe modelirovanie, Tome 10 (1998) no. 3, pp. 117-124. http://geodesic.mathdoc.fr/item/MM_1998_10_3_a8/
@article{MM_1998_10_3_a8,
author = {V. V. Ivanov and B. Purevdorj and I. V. Puzynin},
title = {Second-order learning methods for a~multilayer perceptron},
journal = {Matemati\v{c}eskoe modelirovanie},
pages = {117--124},
year = {1998},
volume = {10},
number = {3},
language = {ru},
url = {http://geodesic.mathdoc.fr/item/MM_1998_10_3_a8/}
}
TY - JOUR
AU - V. V. Ivanov
AU - B. Purevdorj
AU - I. V. Puzynin
TI - Second-order learning methods for a multilayer perceptron
JO - Matematičeskoe modelirovanie
PY - 1998
SP - 117
EP - 124
VL - 10
IS - 3
UR - http://geodesic.mathdoc.fr/item/MM_1998_10_3_a8/
LA - ru
ID - MM_1998_10_3_a8
ER -
%0 Journal Article
%A V. V. Ivanov
%A B. Purevdorj
%A I. V. Puzynin
%T Second-order learning methods for a multilayer perceptron
%J Matematičeskoe modelirovanie
%D 1998
%P 117-124
%V 10
%N 3
%U http://geodesic.mathdoc.fr/item/MM_1998_10_3_a8/
%G ru
%F MM_1998_10_3_a8
First-and second-order learning methods for feed-forward multilayer networks are studied. Newtontype and quasi-Newton algorithms are considered and compared with commonly used backpropagation algorithm. It is shown that, although second-order algorithms reguire enhanced computer facilities, they provide better convergence and simplicity in usage.
