On predicting the optimal number of hidden nodes

Alan Thomas; Miltiadis Petridis; Simon Walters; Mohammad Malekshahi Gheytassi; Robert Morgan

doi:10.1109/CSCI.2015.33

On predicting the optimal number of hidden nodes

Alan Thomas, Miltiadis Petridis, Simon Walters, Mohammad Malekshahi Gheytassi, Robert Morgan

Research output: Chapter in Book/Conference proceeding with ISSN or ISBN › Conference contribution with ISSN or ISBN › peer-review

Abstract

Abstract- Determining the optimal number of hidden nodes isthe most challenging aspect of Artificial Neural Network(ANN) design. To date, there are still no reliable methods ofdetermining this a priori, as it depends on so many domainspecificfactors. Current methods which take these intoaccount, such as exhaustive search, growing and pruning andevolutionary algorithms are not only inexact, but alsoextremely time consuming - in some cases prohibitively so. Anovel approach embodied in a system called Heurix isintroduced. This rapidly predicts the optimal number ofhidden nodes from a small number of sample topologies. It canbe configured to favour speed (low complexity), accuracy, or abalance between the two. Single hidden layer feedforwardnetworks (SLFNs) can be built twenty times faster, and with ageneralisation error of as little as 0.4% greater than thosefound by exhaustive search.

Original language	English
Title of host publication	2015 international conference on computational science and computational intelligence
Place of Publication	Las Vegas
Publisher	CPS/IEEE
Pages	565-570
Number of pages	6
ISBN (Print)	9781467397957
DOIs	https://doi.org/10.1109/CSCI.2015.33
Publication status	Published - 8 Dec 2015
Event	2015 international conference on computational science and computational intelligence - Las Vegas, 7-9 December 2015 Duration: 8 Dec 2015 → …

Conference

Conference	2015 international conference on computational science and computational intelligence
Period	8/12/15 → …

Access to Document

10.1109/CSCI.2015.33Licence: Unspecified

Cite this

@inproceedings{1320a0efc1274be693bf6eb61e5fb960,

title = "On predicting the optimal number of hidden nodes",

abstract = "Abstract- Determining the optimal number of hidden nodes isthe most challenging aspect of Artificial Neural Network(ANN) design. To date, there are still no reliable methods ofdetermining this a priori, as it depends on so many domainspecificfactors. Current methods which take these intoaccount, such as exhaustive search, growing and pruning andevolutionary algorithms are not only inexact, but alsoextremely time consuming - in some cases prohibitively so. Anovel approach embodied in a system called Heurix isintroduced. This rapidly predicts the optimal number ofhidden nodes from a small number of sample topologies. It canbe configured to favour speed (low complexity), accuracy, or abalance between the two. Single hidden layer feedforwardnetworks (SLFNs) can be built twenty times faster, and with ageneralisation error of as little as 0.4% greater than thosefound by exhaustive search.",

author = "Alan Thomas and Miltiadis Petridis and Simon Walters and {Malekshahi Gheytassi}, Mohammad and Robert Morgan",

year = "2015",

month = dec,

day = "8",

doi = "10.1109/CSCI.2015.33",

language = "English",

isbn = "9781467397957",

pages = "565--570",

booktitle = "2015 international conference on computational science and computational intelligence",

publisher = "CPS/IEEE",

note = "2015 international conference on computational science and computational intelligence ; Conference date: 08-12-2015",

}

Thomas, A, Petridis, M, Walters, S , Malekshahi Gheytassi, M & Morgan, R 2015, On predicting the optimal number of hidden nodes. in 2015 international conference on computational science and computational intelligence. CPS/IEEE, Las Vegas, pp. 565-570, 2015 international conference on computational science and computational intelligence, 8/12/15. https://doi.org/10.1109/CSCI.2015.33

TY - GEN

T1 - On predicting the optimal number of hidden nodes

AU - Thomas, Alan

AU - Petridis, Miltiadis

AU - Walters, Simon

AU - Malekshahi Gheytassi, Mohammad

AU - Morgan, Robert

PY - 2015/12/8

Y1 - 2015/12/8

N2 - Abstract- Determining the optimal number of hidden nodes isthe most challenging aspect of Artificial Neural Network(ANN) design. To date, there are still no reliable methods ofdetermining this a priori, as it depends on so many domainspecificfactors. Current methods which take these intoaccount, such as exhaustive search, growing and pruning andevolutionary algorithms are not only inexact, but alsoextremely time consuming - in some cases prohibitively so. Anovel approach embodied in a system called Heurix isintroduced. This rapidly predicts the optimal number ofhidden nodes from a small number of sample topologies. It canbe configured to favour speed (low complexity), accuracy, or abalance between the two. Single hidden layer feedforwardnetworks (SLFNs) can be built twenty times faster, and with ageneralisation error of as little as 0.4% greater than thosefound by exhaustive search.

AB - Abstract- Determining the optimal number of hidden nodes isthe most challenging aspect of Artificial Neural Network(ANN) design. To date, there are still no reliable methods ofdetermining this a priori, as it depends on so many domainspecificfactors. Current methods which take these intoaccount, such as exhaustive search, growing and pruning andevolutionary algorithms are not only inexact, but alsoextremely time consuming - in some cases prohibitively so. Anovel approach embodied in a system called Heurix isintroduced. This rapidly predicts the optimal number ofhidden nodes from a small number of sample topologies. It canbe configured to favour speed (low complexity), accuracy, or abalance between the two. Single hidden layer feedforwardnetworks (SLFNs) can be built twenty times faster, and with ageneralisation error of as little as 0.4% greater than thosefound by exhaustive search.

U2 - 10.1109/CSCI.2015.33

DO - 10.1109/CSCI.2015.33

M3 - Conference contribution with ISSN or ISBN

SN - 9781467397957

SP - 565

EP - 570

BT - 2015 international conference on computational science and computational intelligence

PB - CPS/IEEE

CY - Las Vegas

T2 - 2015 international conference on computational science and computational intelligence

Y2 - 8 December 2015

ER -

On predicting the optimal number of hidden nodes

Abstract

Conference

Access to Document

Fingerprint

Cite this