Study of neural network ensembles algorithm based CPU parallel structure

Ting Fa Xu; Jian Zhong Song; Shuang Qiao

Study of neural network ensembles algorithm based CPU parallel structure

Ting Fa Xu^*, Jian Zhong Song, Shuang Qiao

^*Corresponding author for this work

Research output: Contribution to journal › Article › peer-review

2 Citations (Scopus)

Abstract

A neural network ensemble algorithm used in automatic target recognition (ATR) was designed based on many CPU parallel structure. It is formed a same type classifier under different size and initial condition, by BP neural networks. Real-time recognition was run by the neural networks system. The result showed that statistic recognition rate was up to 95 percent for 9 type object. The processing was finished in 40 ms. The recognition rate was better than single neural networks obviously.

Original language	English
Pages (from-to)	1183-1186
Number of pages	4
Journal	Guangdianzi Jiguang/Journal of Optoelectronics Laser
Volume	14
Issue number	11
Publication status	Published - Nov 2003
Externally published	Yes

Keywords

Automatic target recognition (ATR)
Neural network ensemble
Parallel structure
Posterior probability

Cite this

@article{eaede4778791483ca65d6e4373cac7c8,

title = "Study of neural network ensembles algorithm based CPU parallel structure",

abstract = "A neural network ensemble algorithm used in automatic target recognition (ATR) was designed based on many CPU parallel structure. It is formed a same type classifier under different size and initial condition, by BP neural networks. Real-time recognition was run by the neural networks system. The result showed that statistic recognition rate was up to 95 percent for 9 type object. The processing was finished in 40 ms. The recognition rate was better than single neural networks obviously.",

keywords = "Automatic target recognition (ATR), Neural network ensemble, Parallel structure, Posterior probability",

author = "Xu, {Ting Fa} and Song, {Jian Zhong} and Shuang Qiao",

year = "2003",

month = nov,

language = "English",

volume = "14",

pages = "1183--1186",

journal = "Guangdianzi Jiguang/Journal of Optoelectronics Laser",

issn = "1005-0086",

publisher = "Tianjin University of Technology",

number = "11",

}

TY - JOUR

T1 - Study of neural network ensembles algorithm based CPU parallel structure

AU - Xu, Ting Fa

AU - Song, Jian Zhong

AU - Qiao, Shuang

PY - 2003/11

Y1 - 2003/11

N2 - A neural network ensemble algorithm used in automatic target recognition (ATR) was designed based on many CPU parallel structure. It is formed a same type classifier under different size and initial condition, by BP neural networks. Real-time recognition was run by the neural networks system. The result showed that statistic recognition rate was up to 95 percent for 9 type object. The processing was finished in 40 ms. The recognition rate was better than single neural networks obviously.

AB - A neural network ensemble algorithm used in automatic target recognition (ATR) was designed based on many CPU parallel structure. It is formed a same type classifier under different size and initial condition, by BP neural networks. Real-time recognition was run by the neural networks system. The result showed that statistic recognition rate was up to 95 percent for 9 type object. The processing was finished in 40 ms. The recognition rate was better than single neural networks obviously.

KW - Automatic target recognition (ATR)

KW - Neural network ensemble

KW - Parallel structure

KW - Posterior probability

UR - http://www.scopus.com/inward/record.url?scp=1642290746&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:1642290746

SN - 1005-0086

VL - 14

SP - 1183

EP - 1186

JO - Guangdianzi Jiguang/Journal of Optoelectronics Laser

JF - Guangdianzi Jiguang/Journal of Optoelectronics Laser

IS - 11

ER -

Study of neural network ensembles algorithm based CPU parallel structure

Abstract

Keywords

Other files and links

Fingerprint

Cite this