Improved L-M algorithm for ANNs prediction of phase equilibrium in macromolecule system

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

Improved L-M algorithm for ANNs prediction of phase equilibrium in macromolecule system

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

^*Corresponding author for this work

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

3 Scopus citations

Abstract

Error back propagation (EBP) is a widely used training algorithm for feedforward neural networks (FFNNs), but low learning rate limits its applications in the networks with complex topology architecture and large patterns. In this work, two modifications on Levenberg-Marquardt algorithm for FFNNs were made. One modification was made on the objective function, while the other was the evaluation of the initial weights and biases. The modified algorithm gave a better convergence rate compared to the traditional EBP algorithm and it was less computationally intensive and required less memory. The performance of the algorithm was verified separately with polymer and protein systems. The results showed that the BP network based on modified Levenberg-Marquardt algorithm could be used to predict the binodal curve of H₂O/DMAc (N,N-dimethylacetamide) /PSf (polysulfone) system and lysozyme solubility in aqueous salt solution.

Original language	English
Pages (from-to)	393-399
Number of pages	7
Journal	Huagong Xuebao/CIESC Journal
Volume	56
Issue number	3
State	Published - Mar 2005
Externally published	Yes

Keywords

BP network
EBP algorithm
Levenberg-Marquardt algorithm
Macromolecule system

Cite this

@article{d14c67f0d4424848a4a4d19c06909b58,

title = "Improved L-M algorithm for ANNs prediction of phase equilibrium in macromolecule system",

abstract = "Error back propagation (EBP) is a widely used training algorithm for feedforward neural networks (FFNNs), but low learning rate limits its applications in the networks with complex topology architecture and large patterns. In this work, two modifications on Levenberg-Marquardt algorithm for FFNNs were made. One modification was made on the objective function, while the other was the evaluation of the initial weights and biases. The modified algorithm gave a better convergence rate compared to the traditional EBP algorithm and it was less computationally intensive and required less memory. The performance of the algorithm was verified separately with polymer and protein systems. The results showed that the BP network based on modified Levenberg-Marquardt algorithm could be used to predict the binodal curve of H2O/DMAc (N,N-dimethylacetamide) /PSf (polysulfone) system and lysozyme solubility in aqueous salt solution.",

keywords = "BP network, EBP algorithm, Levenberg-Marquardt algorithm, Macromolecule system",

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

year = "2005",

month = mar,

language = "英语",

volume = "56",

pages = "393--399",

journal = "Huagong Xuebao/CIESC Journal",

issn = "0438-1157",

publisher = "Chemical Industry Press",

number = "3",

}

TY - JOUR

T1 - Improved L-M algorithm for ANNs prediction of phase equilibrium in macromolecule system

AU - He, Xuezhong

AU - Zhang, Xiangping

AU - Zhang, Suojiang

AU - Li, Chunshan

AU - Liu, Jindun

PY - 2005/3

Y1 - 2005/3

N2 - Error back propagation (EBP) is a widely used training algorithm for feedforward neural networks (FFNNs), but low learning rate limits its applications in the networks with complex topology architecture and large patterns. In this work, two modifications on Levenberg-Marquardt algorithm for FFNNs were made. One modification was made on the objective function, while the other was the evaluation of the initial weights and biases. The modified algorithm gave a better convergence rate compared to the traditional EBP algorithm and it was less computationally intensive and required less memory. The performance of the algorithm was verified separately with polymer and protein systems. The results showed that the BP network based on modified Levenberg-Marquardt algorithm could be used to predict the binodal curve of H2O/DMAc (N,N-dimethylacetamide) /PSf (polysulfone) system and lysozyme solubility in aqueous salt solution.

AB - Error back propagation (EBP) is a widely used training algorithm for feedforward neural networks (FFNNs), but low learning rate limits its applications in the networks with complex topology architecture and large patterns. In this work, two modifications on Levenberg-Marquardt algorithm for FFNNs were made. One modification was made on the objective function, while the other was the evaluation of the initial weights and biases. The modified algorithm gave a better convergence rate compared to the traditional EBP algorithm and it was less computationally intensive and required less memory. The performance of the algorithm was verified separately with polymer and protein systems. The results showed that the BP network based on modified Levenberg-Marquardt algorithm could be used to predict the binodal curve of H2O/DMAc (N,N-dimethylacetamide) /PSf (polysulfone) system and lysozyme solubility in aqueous salt solution.

KW - BP network

KW - EBP algorithm

KW - Levenberg-Marquardt algorithm

KW - Macromolecule system

UR - http://www.scopus.com/inward/record.url?scp=17944364748&partnerID=8YFLogxK

M3 - 文章

AN - SCOPUS:17944364748

SN - 0438-1157

VL - 56

SP - 393

EP - 399

JO - Huagong Xuebao/CIESC Journal

JF - Huagong Xuebao/CIESC Journal

IS - 3

ER -

Improved L-M algorithm for ANNs prediction of phase equilibrium in macromolecule system

Abstract

Keywords

Other files and links

Fingerprint

Cite this