Selection of training samples of the neural network in fermentation industry

Yun Feng Li; Jing Qi Yuan; Yao Feng Xue

Selection of training samples of the neural network in fermentation industry

Yun Feng Li^*
, Jing Qi Yuan
, Yao Feng Xue

^*Corresponding author for this work

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

2 Scopus citations

Abstract

A new selection method, named Statistics Average Method, is proposed. With a three-layer BP network applied for total product prediction in penicillin production, the result indicates it can significantly reduce the number of training samples without affection the prediction accuracy. Compared with two other selection methods, Kohonen network selection and Kennard-Stone selection method, it is simpler and more feasible.

Original language	English
Pages (from-to)	21-24
Number of pages	4
Journal	Huagong Zidonghua Ji Yibiao/Control and Instruments in Chemical Industry
Volume	31
Issue number	6
State	Published - Dec 2004
Externally published	Yes

Keywords

Fermentation
Neural network
State prediction
Training sample

Cite this

@article{b5c644737f3147e090714e7558caf724,

title = "Selection of training samples of the neural network in fermentation industry",

abstract = "A new selection method, named Statistics Average Method, is proposed. With a three-layer BP network applied for total product prediction in penicillin production, the result indicates it can significantly reduce the number of training samples without affection the prediction accuracy. Compared with two other selection methods, Kohonen network selection and Kennard-Stone selection method, it is simpler and more feasible.",

keywords = "Fermentation, Neural network, State prediction, Training sample",

author = "Li, \{Yun Feng\} and Yuan, \{Jing Qi\} and Xue, \{Yao Feng\}",

year = "2004",

month = dec,

language = "英语",

volume = "31",

pages = "21--24",

journal = "Huagong Zidonghua Ji Yibiao/Control and Instruments in Chemical Industry",

issn = "1000-3932",

publisher = "Tianhua Huagong Jijie ji Zidonghua Yanjiu Shejiyuan",

number = "6",

}

TY - JOUR

T1 - Selection of training samples of the neural network in fermentation industry

AU - Li, Yun Feng

AU - Yuan, Jing Qi

AU - Xue, Yao Feng

PY - 2004/12

Y1 - 2004/12

N2 - A new selection method, named Statistics Average Method, is proposed. With a three-layer BP network applied for total product prediction in penicillin production, the result indicates it can significantly reduce the number of training samples without affection the prediction accuracy. Compared with two other selection methods, Kohonen network selection and Kennard-Stone selection method, it is simpler and more feasible.

AB - A new selection method, named Statistics Average Method, is proposed. With a three-layer BP network applied for total product prediction in penicillin production, the result indicates it can significantly reduce the number of training samples without affection the prediction accuracy. Compared with two other selection methods, Kohonen network selection and Kennard-Stone selection method, it is simpler and more feasible.

KW - Fermentation

KW - Neural network

KW - State prediction

KW - Training sample

UR - https://www.scopus.com/pages/publications/21144435249

M3 - 文章

AN - SCOPUS:21144435249

SN - 1000-3932

VL - 31

SP - 21

EP - 24

JO - Huagong Zidonghua Ji Yibiao/Control and Instruments in Chemical Industry

JF - Huagong Zidonghua Ji Yibiao/Control and Instruments in Chemical Industry

IS - 6

ER -

Selection of training samples of the neural network in fermentation industry

Abstract

Keywords

Other files and links

Fingerprint

Cite this