The design of RBF neural networks for solving overfitting problem

Yu Zhigang; Song Shenmin; Duan Guangren; Pei Run; Chu Wenjun

doi:10.1109/WCICA.2006.1712865

The design of RBF neural networks for solving overfitting problem

Yu Zhigang^*
, Song Shenmin
, Duan Guangren
, Pei Run
, Chu Wenjun

^*Corresponding author for this work

School of Astronautics

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

2 Scopus citations

Abstract

One of the biggest problems in designing or training RBF neural networks are the overfitting problem. The traditional design of RBF neural networks may be pursued in a variety of ways. In this paper, we present a method for the design of RBF networks to solve overfitting problem. For a practical application, frequency information is usually available for designing RBF networks by frequency domain analysis, which has a sound mathematical basis. We try to include the frequency information into the design of RBF networks, which achieve the task of approximated a function in certain frequency range and have the property of structural risk minimization. After the structure of designed network is determined, the linear weights of the output layer are the only set of adjustable parameters. The approach of design is verified by approximation cases.

Original language	English
Title of host publication	Proceedings of the World Congress on Intelligent Control and Automation (WCICA)
Pages	2752-2756
Number of pages	5
DOIs	https://doi.org/10.1109/WCICA.2006.1712865
State	Published - 2006
Event	6th World Congress on Intelligent Control and Automation, WCICA 2006 - Dalian, China Duration: 21 Jun 2006 → 23 Jun 2006

Publication series

Name	Proceedings of the World Congress on Intelligent Control and Automation (WCICA)
Volume	1

Conference

Conference	6th World Congress on Intelligent Control and Automation, WCICA 2006
Country/Territory	China
City	Dalian
Period	21/06/06 → 23/06/06

Keywords

Overfitting problem
Radial basis function
Structural risk minimization

Access to Document

10.1109/WCICA.2006.1712865

Cite this

Zhigang, Y., Shenmin, S., Guangren, D., Run, P., & Wenjun, C. (2006). The design of RBF neural networks for solving overfitting problem. In Proceedings of the World Congress on Intelligent Control and Automation (WCICA) (pp. 2752-2756). Article 1712865 (Proceedings of the World Congress on Intelligent Control and Automation (WCICA); Vol. 1). https://doi.org/10.1109/WCICA.2006.1712865

@inproceedings{40360177111648019f96fb980cf393e7,

title = "The design of RBF neural networks for solving overfitting problem",

abstract = "One of the biggest problems in designing or training RBF neural networks are the overfitting problem. The traditional design of RBF neural networks may be pursued in a variety of ways. In this paper, we present a method for the design of RBF networks to solve overfitting problem. For a practical application, frequency information is usually available for designing RBF networks by frequency domain analysis, which has a sound mathematical basis. We try to include the frequency information into the design of RBF networks, which achieve the task of approximated a function in certain frequency range and have the property of structural risk minimization. After the structure of designed network is determined, the linear weights of the output layer are the only set of adjustable parameters. The approach of design is verified by approximation cases.",

keywords = "Overfitting problem, Radial basis function, Structural risk minimization",

author = "Yu Zhigang and Song Shenmin and Duan Guangren and Pei Run and Chu Wenjun",

year = "2006",

doi = "10.1109/WCICA.2006.1712865",

language = "英语",

isbn = "1424403324",

series = "Proceedings of the World Congress on Intelligent Control and Automation (WCICA)",

pages = "2752--2756",

booktitle = "Proceedings of the World Congress on Intelligent Control and Automation (WCICA)",

note = "6th World Congress on Intelligent Control and Automation, WCICA 2006 ; Conference date: 21-06-2006 Through 23-06-2006",

}

Zhigang, Y, Shenmin, S , Guangren, D, Run, P & Wenjun, C 2006, The design of RBF neural networks for solving overfitting problem. in Proceedings of the World Congress on Intelligent Control and Automation (WCICA)., 1712865, Proceedings of the World Congress on Intelligent Control and Automation (WCICA), vol. 1, pp. 2752-2756, 6th World Congress on Intelligent Control and Automation, WCICA 2006, Dalian, China, 21/06/06. https://doi.org/10.1109/WCICA.2006.1712865

The design of RBF neural networks for solving overfitting problem. / Zhigang, Yu; Shenmin, Song ; Guangren, Duan et al.
Proceedings of the World Congress on Intelligent Control and Automation (WCICA). 2006. p. 2752-2756 1712865 (Proceedings of the World Congress on Intelligent Control and Automation (WCICA); Vol. 1).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - The design of RBF neural networks for solving overfitting problem

AU - Zhigang, Yu

AU - Shenmin, Song

AU - Guangren, Duan

AU - Run, Pei

AU - Wenjun, Chu

PY - 2006

Y1 - 2006

N2 - One of the biggest problems in designing or training RBF neural networks are the overfitting problem. The traditional design of RBF neural networks may be pursued in a variety of ways. In this paper, we present a method for the design of RBF networks to solve overfitting problem. For a practical application, frequency information is usually available for designing RBF networks by frequency domain analysis, which has a sound mathematical basis. We try to include the frequency information into the design of RBF networks, which achieve the task of approximated a function in certain frequency range and have the property of structural risk minimization. After the structure of designed network is determined, the linear weights of the output layer are the only set of adjustable parameters. The approach of design is verified by approximation cases.

AB - One of the biggest problems in designing or training RBF neural networks are the overfitting problem. The traditional design of RBF neural networks may be pursued in a variety of ways. In this paper, we present a method for the design of RBF networks to solve overfitting problem. For a practical application, frequency information is usually available for designing RBF networks by frequency domain analysis, which has a sound mathematical basis. We try to include the frequency information into the design of RBF networks, which achieve the task of approximated a function in certain frequency range and have the property of structural risk minimization. After the structure of designed network is determined, the linear weights of the output layer are the only set of adjustable parameters. The approach of design is verified by approximation cases.

KW - Overfitting problem

KW - Radial basis function

KW - Structural risk minimization

UR - https://www.scopus.com/pages/publications/34047213294

U2 - 10.1109/WCICA.2006.1712865

DO - 10.1109/WCICA.2006.1712865

M3 - 会议稿件

AN - SCOPUS:34047213294

SN - 1424403324

SN - 9781424403325

T3 - Proceedings of the World Congress on Intelligent Control and Automation (WCICA)

SP - 2752

EP - 2756

BT - Proceedings of the World Congress on Intelligent Control and Automation (WCICA)

T2 - 6th World Congress on Intelligent Control and Automation, WCICA 2006

Y2 - 21 June 2006 through 23 June 2006

ER -

The design of RBF neural networks for solving overfitting problem

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this