Optimization of feed-forward neural networks configuration used for bridge condition rating approximation

Roszilah Hamid, Khairullah Yusuf, Abdul Khalim Abdul Rashid

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

Abstract

The paper presents result from experiments on network architecture and transfer functions configuration in the feed-forward neural networks (FFNNs) applied to bridge condition rating approximation. Trial and error approach is done on three layers feed-forward neural network by varying the number of neurons in hidden layer. Levenberg-Marquardt training algorithm (trainlm) and sigmoid transfer function are applied in FFNN to investigate the best configuration to be used for bridge condition rating. Mean square error (MSE) and correlation coefficient (R) are used to measure the network performance. The results indicated that the configuration of FFNN with thirty-one neurons in hidden layer using tangent-sigmoid (tansig) transfer function in output layer have produced the best MSE and R than other configurations.

Original languageEnglish
Title of host publicationInternational Conference on Engineering Mechanics, Structures, Engineering Geology, International Conference on Geography and Geology - Proceedings
Pages408-412
Number of pages5
Publication statusPublished - 2010
Event3rd WSEAS International Conference on Engineering Mechanics, Structures, Engineering Geology, EMESEG'10, International Conference on Geography and Geology 2010, WORLDGEO'10 - Corfu Island
Duration: 22 Jul 201024 Jul 2010

Other

Other3rd WSEAS International Conference on Engineering Mechanics, Structures, Engineering Geology, EMESEG'10, International Conference on Geography and Geology 2010, WORLDGEO'10
CityCorfu Island
Period22/7/1024/7/10

Fingerprint

transfer function
experiment

Keywords

  • Bridge condition rating
  • Feed-forward neural networks
  • Transfer function

ASJC Scopus subject areas

  • Geology

Cite this

Hamid, R., Yusuf, K., & Rashid, A. K. A. (2010). Optimization of feed-forward neural networks configuration used for bridge condition rating approximation. In International Conference on Engineering Mechanics, Structures, Engineering Geology, International Conference on Geography and Geology - Proceedings (pp. 408-412)

Optimization of feed-forward neural networks configuration used for bridge condition rating approximation. / Hamid, Roszilah; Yusuf, Khairullah; Rashid, Abdul Khalim Abdul.

International Conference on Engineering Mechanics, Structures, Engineering Geology, International Conference on Geography and Geology - Proceedings. 2010. p. 408-412.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Hamid, R, Yusuf, K & Rashid, AKA 2010, Optimization of feed-forward neural networks configuration used for bridge condition rating approximation. in International Conference on Engineering Mechanics, Structures, Engineering Geology, International Conference on Geography and Geology - Proceedings. pp. 408-412, 3rd WSEAS International Conference on Engineering Mechanics, Structures, Engineering Geology, EMESEG'10, International Conference on Geography and Geology 2010, WORLDGEO'10, Corfu Island, 22/7/10.
Hamid R, Yusuf K, Rashid AKA. Optimization of feed-forward neural networks configuration used for bridge condition rating approximation. In International Conference on Engineering Mechanics, Structures, Engineering Geology, International Conference on Geography and Geology - Proceedings. 2010. p. 408-412
Hamid, Roszilah ; Yusuf, Khairullah ; Rashid, Abdul Khalim Abdul. / Optimization of feed-forward neural networks configuration used for bridge condition rating approximation. International Conference on Engineering Mechanics, Structures, Engineering Geology, International Conference on Geography and Geology - Proceedings. 2010. pp. 408-412
@inproceedings{59b0a2d9a2ff4e9892d4efd6c7967a1f,
title = "Optimization of feed-forward neural networks configuration used for bridge condition rating approximation",
abstract = "The paper presents result from experiments on network architecture and transfer functions configuration in the feed-forward neural networks (FFNNs) applied to bridge condition rating approximation. Trial and error approach is done on three layers feed-forward neural network by varying the number of neurons in hidden layer. Levenberg-Marquardt training algorithm (trainlm) and sigmoid transfer function are applied in FFNN to investigate the best configuration to be used for bridge condition rating. Mean square error (MSE) and correlation coefficient (R) are used to measure the network performance. The results indicated that the configuration of FFNN with thirty-one neurons in hidden layer using tangent-sigmoid (tansig) transfer function in output layer have produced the best MSE and R than other configurations.",
keywords = "Bridge condition rating, Feed-forward neural networks, Transfer function",
author = "Roszilah Hamid and Khairullah Yusuf and Rashid, {Abdul Khalim Abdul}",
year = "2010",
language = "English",
isbn = "9789604742035",
pages = "408--412",
booktitle = "International Conference on Engineering Mechanics, Structures, Engineering Geology, International Conference on Geography and Geology - Proceedings",

}

TY - GEN

T1 - Optimization of feed-forward neural networks configuration used for bridge condition rating approximation

AU - Hamid, Roszilah

AU - Yusuf, Khairullah

AU - Rashid, Abdul Khalim Abdul

PY - 2010

Y1 - 2010

N2 - The paper presents result from experiments on network architecture and transfer functions configuration in the feed-forward neural networks (FFNNs) applied to bridge condition rating approximation. Trial and error approach is done on three layers feed-forward neural network by varying the number of neurons in hidden layer. Levenberg-Marquardt training algorithm (trainlm) and sigmoid transfer function are applied in FFNN to investigate the best configuration to be used for bridge condition rating. Mean square error (MSE) and correlation coefficient (R) are used to measure the network performance. The results indicated that the configuration of FFNN with thirty-one neurons in hidden layer using tangent-sigmoid (tansig) transfer function in output layer have produced the best MSE and R than other configurations.

AB - The paper presents result from experiments on network architecture and transfer functions configuration in the feed-forward neural networks (FFNNs) applied to bridge condition rating approximation. Trial and error approach is done on three layers feed-forward neural network by varying the number of neurons in hidden layer. Levenberg-Marquardt training algorithm (trainlm) and sigmoid transfer function are applied in FFNN to investigate the best configuration to be used for bridge condition rating. Mean square error (MSE) and correlation coefficient (R) are used to measure the network performance. The results indicated that the configuration of FFNN with thirty-one neurons in hidden layer using tangent-sigmoid (tansig) transfer function in output layer have produced the best MSE and R than other configurations.

KW - Bridge condition rating

KW - Feed-forward neural networks

KW - Transfer function

UR - http://www.scopus.com/inward/record.url?scp=79958788259&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79958788259&partnerID=8YFLogxK

M3 - Conference contribution

SN - 9789604742035

SP - 408

EP - 412

BT - International Conference on Engineering Mechanics, Structures, Engineering Geology, International Conference on Geography and Geology - Proceedings

ER -