One still open question in the area of research of multi-layer feedforward neural networks is concerning the number of neurons in its hidden layer(s). Especially in real life applications, this problem is often solved by heuristic methods. In this work an effective way to dynamically determine the number of hidden units in a three-layer feedforward neural network for function approximation is proposed.
@article{urn:eudml:doc:39023,
title = {An effective way to generate neural network structures for function approximation.},
journal = {Mathware and Soft Computing},
volume = {1},
year = {1994},
pages = {139-161},
language = {en},
url = {http://dml.mathdoc.fr/item/urn:eudml:doc:39023}
}
Bastian, Andreas. An effective way to generate neural network structures for function approximation.. Mathware and Soft Computing, Tome 1 (1994) pp. 139-161. http://gdmltest.u-ga.fr/item/urn:eudml:doc:39023/