Genetic Programming Discovers Efficient Learning Rules
for the Hidden and Output Layers of Feedforward Neural
Networks
A. Radi, and R. Poli. Genetic Programming, Proceedings of EuroGP'99, volume 1598 of LNCS, page 120--134. Goteborg, Sweden, Springer-Verlag, (26-27 May 1999)
Abstract
The learning method is critical for obtaining good
generalisation in neural networks with limited training
data. The Standard BackPropagation (SBP ) training
algorithm suffers from several problems such as
sensitivity to the initial conditions and very slow
convergence. The aim of this work is to use Genetic
Programming (GP) to discover new supervised learning
algorithms which can overcome some of these problems.
In previous research a new learning algorithm for the
output layer has been discovered using GP. By comparing
this with SBP on different problems better performance
was demonstrated. This paper shows that GP can also
discover better learning algorithms for the hidden
layers to be used in conjunction with the algorithm
previously discovered. Comparing these with SBP on
different problems we show they provide better
performance. This study indicates that there exist many
supervised learning algorithms better than SBP and that
GP can be used to discover them.
%0 Conference Paper
%1 radi:1999:GPdelrholffNN
%A Radi, Amr
%A Poli, Riccardo
%B Genetic Programming, Proceedings of EuroGP'99
%C Goteborg, Sweden
%D 1999
%E Poli, Riccardo
%E Nordin, Peter
%E Langdon, William B.
%E Fogarty, Terence C.
%I Springer-Verlag
%K algorithms, genetic programming
%P 120--134
%T Genetic Programming Discovers Efficient Learning Rules
for the Hidden and Output Layers of Feedforward Neural
Networks
%U http://www.springerlink.com/openurl.asp?genre=article&issn=0302-9743&volume=1598&spage=120
%V 1598
%X The learning method is critical for obtaining good
generalisation in neural networks with limited training
data. The Standard BackPropagation (SBP ) training
algorithm suffers from several problems such as
sensitivity to the initial conditions and very slow
convergence. The aim of this work is to use Genetic
Programming (GP) to discover new supervised learning
algorithms which can overcome some of these problems.
In previous research a new learning algorithm for the
output layer has been discovered using GP. By comparing
this with SBP on different problems better performance
was demonstrated. This paper shows that GP can also
discover better learning algorithms for the hidden
layers to be used in conjunction with the algorithm
previously discovered. Comparing these with SBP on
different problems we show they provide better
performance. This study indicates that there exist many
supervised learning algorithms better than SBP and that
GP can be used to discover them.
%@ 3-540-65899-8
@inproceedings{radi:1999:GPdelrholffNN,
abstract = {The learning method is critical for obtaining good
generalisation in neural networks with limited training
data. The Standard BackPropagation (SBP ) training
algorithm suffers from several problems such as
sensitivity to the initial conditions and very slow
convergence. The aim of this work is to use Genetic
Programming (GP) to discover new supervised learning
algorithms which can overcome some of these problems.
In previous research a new learning algorithm for the
output layer has been discovered using GP. By comparing
this with SBP on different problems better performance
was demonstrated. This paper shows that GP can also
discover better learning algorithms for the hidden
layers to be used in conjunction with the algorithm
previously discovered. Comparing these with SBP on
different problems we show they provide better
performance. This study indicates that there exist many
supervised learning algorithms better than SBP and that
GP can be used to discover them.},
added-at = {2008-06-19T17:46:40.000+0200},
address = {Goteborg, Sweden},
author = {Radi, Amr and Poli, Riccardo},
biburl = {https://www.bibsonomy.org/bibtex/2146a27b67e2ac9091b6ccdbf034d1bcc/brazovayeye},
booktitle = {Genetic Programming, Proceedings of EuroGP'99},
editor = {Poli, Riccardo and Nordin, Peter and Langdon, William B. and Fogarty, Terence C.},
interhash = {f9b4134c9d9607bdda5ad5957d4e88e9},
intrahash = {146a27b67e2ac9091b6ccdbf034d1bcc},
isbn = {3-540-65899-8},
keywords = {algorithms, genetic programming},
month = {26-27 May},
notes = {EuroGP'99, part of \cite{poli:1999:GP}},
organisation = {EvoNet},
pages = {120--134},
publisher = {Springer-Verlag},
publisher_address = {Berlin},
series = {LNCS},
timestamp = {2008-06-19T17:50:01.000+0200},
title = {Genetic Programming Discovers Efficient Learning Rules
for the Hidden and Output Layers of Feedforward Neural
Networks},
url = {http://www.springerlink.com/openurl.asp?genre=article&issn=0302-9743&volume=1598&spage=120},
volume = 1598,
year = 1999
}