Prediction of phase equilibrium properties for complicated macromolecular systems by HGALM neural networks

Xuezhong He; Xiangping Zhang; Soujiang Zhang; Jindun Liu; Chunshan Li

doi:10.1016/j.fluid.2005.09.017

Prediction of phase equilibrium properties for complicated macromolecular systems by HGALM neural networks

Xuezhong He, Xiangping Zhang, Soujiang Zhang^*, Jindun Liu, Chunshan Li

^*Corresponding author for this work

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

14 Scopus citations

Abstract

Traditional error back propagation is a widely used training algorithm for feed forward neural networks (FFNNs). However, it generally encounters two problems of slow learning rate and relative low accuracy. In this work, a hybrid genetic algorithm combined with modified Levenberg-Marquardt algorithm (HGALM) was proposed for training FFNNs to improve the accuracy and decrease the time depletion comparing to the traditional EBP algorithm. The FFNNs based on HGALM were used to predict the binodal curve of water-DMAc-PSf system and protein solubility in lysozyme-NaCl-H₂O system. The results would be used for guiding experimental researches in preparation of asymmetry polymer membrane and optimization of protein crystal process.

Original language	English
Pages (from-to)	52-57
Number of pages	6
Journal	Fluid Phase Equilibria
Volume	238
Issue number	1
DOIs	https://doi.org/10.1016/j.fluid.2005.09.017
State	Published - 25 Nov 2005
Externally published	Yes

Keywords

Feed forward neural networks
Genetic algorithm
Polymer system
Prediction
Protein system

Access to Document

10.1016/j.fluid.2005.09.017

Cite this

@article{75835aa3366447518809226c617d340f,

title = "Prediction of phase equilibrium properties for complicated macromolecular systems by HGALM neural networks",

abstract = "Traditional error back propagation is a widely used training algorithm for feed forward neural networks (FFNNs). However, it generally encounters two problems of slow learning rate and relative low accuracy. In this work, a hybrid genetic algorithm combined with modified Levenberg-Marquardt algorithm (HGALM) was proposed for training FFNNs to improve the accuracy and decrease the time depletion comparing to the traditional EBP algorithm. The FFNNs based on HGALM were used to predict the binodal curve of water-DMAc-PSf system and protein solubility in lysozyme-NaCl-H2O system. The results would be used for guiding experimental researches in preparation of asymmetry polymer membrane and optimization of protein crystal process.",

keywords = "Feed forward neural networks, Genetic algorithm, Polymer system, Prediction, Protein system",

author = "Xuezhong He and Xiangping Zhang and Soujiang Zhang and Jindun Liu and Chunshan Li",

note = "Funding Information: This work was supported by National Scientific Funding of China (NSFC) (No. 20276073).",

year = "2005",

month = nov,

day = "25",

doi = "10.1016/j.fluid.2005.09.017",

language = "英语",

volume = "238",

pages = "52--57",

journal = "Fluid Phase Equilibria",

issn = "0378-3812",

publisher = "Elsevier",

number = "1",

}

TY - JOUR

T1 - Prediction of phase equilibrium properties for complicated macromolecular systems by HGALM neural networks

AU - He, Xuezhong

AU - Zhang, Xiangping

AU - Zhang, Soujiang

AU - Liu, Jindun

AU - Li, Chunshan

N1 - Funding Information: This work was supported by National Scientific Funding of China (NSFC) (No. 20276073).

PY - 2005/11/25

Y1 - 2005/11/25

N2 - Traditional error back propagation is a widely used training algorithm for feed forward neural networks (FFNNs). However, it generally encounters two problems of slow learning rate and relative low accuracy. In this work, a hybrid genetic algorithm combined with modified Levenberg-Marquardt algorithm (HGALM) was proposed for training FFNNs to improve the accuracy and decrease the time depletion comparing to the traditional EBP algorithm. The FFNNs based on HGALM were used to predict the binodal curve of water-DMAc-PSf system and protein solubility in lysozyme-NaCl-H2O system. The results would be used for guiding experimental researches in preparation of asymmetry polymer membrane and optimization of protein crystal process.

AB - Traditional error back propagation is a widely used training algorithm for feed forward neural networks (FFNNs). However, it generally encounters two problems of slow learning rate and relative low accuracy. In this work, a hybrid genetic algorithm combined with modified Levenberg-Marquardt algorithm (HGALM) was proposed for training FFNNs to improve the accuracy and decrease the time depletion comparing to the traditional EBP algorithm. The FFNNs based on HGALM were used to predict the binodal curve of water-DMAc-PSf system and protein solubility in lysozyme-NaCl-H2O system. The results would be used for guiding experimental researches in preparation of asymmetry polymer membrane and optimization of protein crystal process.

KW - Feed forward neural networks

KW - Genetic algorithm

KW - Polymer system

KW - Prediction

KW - Protein system

UR - http://www.scopus.com/inward/record.url?scp=27644528955&partnerID=8YFLogxK

U2 - 10.1016/j.fluid.2005.09.017

DO - 10.1016/j.fluid.2005.09.017

M3 - 文章

AN - SCOPUS:27644528955

SN - 0378-3812

VL - 238

SP - 52

EP - 57

JO - Fluid Phase Equilibria

JF - Fluid Phase Equilibria

IS - 1

ER -

Prediction of phase equilibrium properties for complicated macromolecular systems by HGALM neural networks

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this