Autonomous Hidden Node Determination using Dynamic Expansion & Contraction Approach

D. Young; L. M. Cheng

doi:10.1109/SIPNN.1994.344880

Autonomous Hidden Node Determination using Dynamic Expansion & Contraction Approach

Department of Electrical Engineering

Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 32 - Refereed conference paper (with host publication) › peer-review

1 Citation (Scopus)

Abstract

One of the major limitation of using the back-propagation neural network and its variants in real applications are that the number of hidden nodes is unknown. It is usually estimated by trial-and-error and thus it is inefficient. The paper proposes an algorithm to determine the number of hidden nodes based on the input data. The dynamic expansion and contraction approach (DECA), which comes from dynamic programming, is used to determine the optimal number of hidden nodes. The object function minimises the number of hidden nodes while the constraints are a pre-defined error. A short interval of train/test interleaving is used to minimise the learning time and avoid over-training the network. The algorithm is applicable to the neural network used for function approximation as well as pattern classification. © 1994 IEEE.

Original language	English
Title of host publication	ISSIPNN 1994 - 1994 International Symposium on Speech, Image Processing and Neural Networks, Proceedings
Publisher	IEEE
Pages	421-424
ISBN (Print)	078031865X, 9780780318656
DOIs	https://doi.org/10.1109/SIPNN.1994.344880
Publication status	Published - Apr 1994
Event	1994 International Symposium on Speech, Image Processing and Neural Networks (ISSIPNN 1994) - Hong Kong, China Duration: 13 Apr 1994 → 16 Apr 1994

Conference

Conference	1994 International Symposium on Speech, Image Processing and Neural Networks (ISSIPNN 1994)
Place	China
City	Hong Kong
Period	13/04/94 → 16/04/94

Access Output

10.1109/SIPNN.1994.344880

Other Access Options

Check CityUHK Library

Permanent Link

Right click to copy link

Cite this

@inproceedings{8efc2d2d984345f499104c44348098e6,

title = "Autonomous Hidden Node Determination using Dynamic Expansion & Contraction Approach",

abstract = "One of the major limitation of using the back-propagation neural network and its variants in real applications are that the number of hidden nodes is unknown. It is usually estimated by trial-and-error and thus it is inefficient. The paper proposes an algorithm to determine the number of hidden nodes based on the input data. The dynamic expansion and contraction approach (DECA), which comes from dynamic programming, is used to determine the optimal number of hidden nodes. The object function minimises the number of hidden nodes while the constraints are a pre-defined error. A short interval of train/test interleaving is used to minimise the learning time and avoid over-training the network. The algorithm is applicable to the neural network used for function approximation as well as pattern classification. {\textcopyright} 1994 IEEE.",

author = "D. Young and Cheng, {L. M.}",

year = "1994",

month = apr,

doi = "10.1109/SIPNN.1994.344880",

language = "English",

isbn = "078031865X",

pages = "421--424",

booktitle = "ISSIPNN 1994 - 1994 International Symposium on Speech, Image Processing and Neural Networks, Proceedings",

publisher = "IEEE",

address = "United States",

note = "1994 International Symposium on Speech, Image Processing and Neural Networks (ISSIPNN 1994) ; Conference date: 13-04-1994 Through 16-04-1994",

}

Young, D & Cheng, LM 1994, Autonomous Hidden Node Determination using Dynamic Expansion & Contraction Approach. in ISSIPNN 1994 - 1994 International Symposium on Speech, Image Processing and Neural Networks, Proceedings., 344880, IEEE, pp. 421-424, 1994 International Symposium on Speech, Image Processing and Neural Networks (ISSIPNN 1994), Hong Kong, China, 13/04/94. https://doi.org/10.1109/SIPNN.1994.344880

Autonomous Hidden Node Determination using Dynamic Expansion & Contraction Approach. / Young, D.; Cheng, L. M.
ISSIPNN 1994 - 1994 International Symposium on Speech, Image Processing and Neural Networks, Proceedings. IEEE, 1994. p. 421-424 344880.

Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 32 - Refereed conference paper (with host publication) › peer-review

TY - GEN

T1 - Autonomous Hidden Node Determination using Dynamic Expansion & Contraction Approach

AU - Young, D.

AU - Cheng, L. M.

PY - 1994/4

Y1 - 1994/4

N2 - One of the major limitation of using the back-propagation neural network and its variants in real applications are that the number of hidden nodes is unknown. It is usually estimated by trial-and-error and thus it is inefficient. The paper proposes an algorithm to determine the number of hidden nodes based on the input data. The dynamic expansion and contraction approach (DECA), which comes from dynamic programming, is used to determine the optimal number of hidden nodes. The object function minimises the number of hidden nodes while the constraints are a pre-defined error. A short interval of train/test interleaving is used to minimise the learning time and avoid over-training the network. The algorithm is applicable to the neural network used for function approximation as well as pattern classification. © 1994 IEEE.

AB - One of the major limitation of using the back-propagation neural network and its variants in real applications are that the number of hidden nodes is unknown. It is usually estimated by trial-and-error and thus it is inefficient. The paper proposes an algorithm to determine the number of hidden nodes based on the input data. The dynamic expansion and contraction approach (DECA), which comes from dynamic programming, is used to determine the optimal number of hidden nodes. The object function minimises the number of hidden nodes while the constraints are a pre-defined error. A short interval of train/test interleaving is used to minimise the learning time and avoid over-training the network. The algorithm is applicable to the neural network used for function approximation as well as pattern classification. © 1994 IEEE.

UR - http://www.scopus.com/inward/record.url?scp=85046631510&partnerID=8YFLogxK

UR - https://www.scopus.com/record/pubmetrics.uri?eid=2-s2.0-85046631510&origin=recordpage

U2 - 10.1109/SIPNN.1994.344880

DO - 10.1109/SIPNN.1994.344880

M3 - RGC 32 - Refereed conference paper (with host publication)

SN - 078031865X

SN - 9780780318656

SP - 421

EP - 424

BT - ISSIPNN 1994 - 1994 International Symposium on Speech, Image Processing and Neural Networks, Proceedings

PB - IEEE

T2 - 1994 International Symposium on Speech, Image Processing and Neural Networks (ISSIPNN 1994)

Y2 - 13 April 1994 through 16 April 1994

ER -

Autonomous Hidden Node Determination using Dynamic Expansion & Contraction Approach

Abstract

Conference

Access Output

Other Access Options

Permanent Link

Other Files and Links

Fingerprint

Cite this