Nonlinear gradient neural network for solving system of linear equations

Lin Xiao; Kenli Li; Zhiguo Tan; Zhijun Zhang; Bolin Liao; Ke Chen; Long Jin; Shuai Li

doi:10.1016/j.ipl.2018.10.004

Nonlinear gradient neural network for solving system of linear equations

Lin Xiao, Kenli Li, Zhiguo Tan, Zhijun Zhang, Bolin Liao, Ke Chen, Long Jin, Shuai Li

Research output: Journal article publication › Journal article › Academic research › peer-review

52 Citations (Scopus)

Abstract

For purpose of solving system of linear equations (SoLE) more efficiently, a fast convergent gradient neural network (FCGNN) model is designed and discussed in this paper. Different from the design of the conventional gradient neural network (CGNN), the design of the FCGNN model is based on a nonlinear activation function, and thus the better convergence speed can be reached. In addition, the convergence upper bound of the FCGNN model is estimated and provided in details. Simulative results validate the superiority of the FCGNN model, as compared to the CGNN model for finding SoLE.

Original language	English
Pages (from-to)	35-40
Number of pages	6
Journal	Information Processing Letters
Volume	142
DOIs	https://doi.org/10.1016/j.ipl.2018.10.004
Publication status	Published - 1 Feb 2019

Keywords

Fast convergence
Gradient neural network (GNN)
Nonlinear activation function
Performance evaluation
Systems of linear equations (SoLE)

ASJC Scopus subject areas

Theoretical Computer Science
Signal Processing
Information Systems
Computer Science Applications

More information

10.1016/j.ipl.2018.10.004

Cite this

@article{f82361aeeb484f6c84b9d0b46385d079,

title = "Nonlinear gradient neural network for solving system of linear equations",

abstract = "For purpose of solving system of linear equations (SoLE) more efficiently, a fast convergent gradient neural network (FCGNN) model is designed and discussed in this paper. Different from the design of the conventional gradient neural network (CGNN), the design of the FCGNN model is based on a nonlinear activation function, and thus the better convergence speed can be reached. In addition, the convergence upper bound of the FCGNN model is estimated and provided in details. Simulative results validate the superiority of the FCGNN model, as compared to the CGNN model for finding SoLE.",

keywords = "Fast convergence, Gradient neural network (GNN), Nonlinear activation function, Performance evaluation, Systems of linear equations (SoLE)",

author = "Lin Xiao and Kenli Li and Zhiguo Tan and Zhijun Zhang and Bolin Liao and Ke Chen and Long Jin and Shuai Li",

year = "2019",

month = feb,

day = "1",

doi = "10.1016/j.ipl.2018.10.004",

language = "English",

volume = "142",

pages = "35--40",

journal = "Information Processing Letters",

issn = "0020-0190",

publisher = "Elsevier",

}

TY - JOUR

T1 - Nonlinear gradient neural network for solving system of linear equations

AU - Xiao, Lin

AU - Li, Kenli

AU - Tan, Zhiguo

AU - Zhang, Zhijun

AU - Liao, Bolin

AU - Chen, Ke

AU - Jin, Long

AU - Li, Shuai

PY - 2019/2/1

Y1 - 2019/2/1

N2 - For purpose of solving system of linear equations (SoLE) more efficiently, a fast convergent gradient neural network (FCGNN) model is designed and discussed in this paper. Different from the design of the conventional gradient neural network (CGNN), the design of the FCGNN model is based on a nonlinear activation function, and thus the better convergence speed can be reached. In addition, the convergence upper bound of the FCGNN model is estimated and provided in details. Simulative results validate the superiority of the FCGNN model, as compared to the CGNN model for finding SoLE.

AB - For purpose of solving system of linear equations (SoLE) more efficiently, a fast convergent gradient neural network (FCGNN) model is designed and discussed in this paper. Different from the design of the conventional gradient neural network (CGNN), the design of the FCGNN model is based on a nonlinear activation function, and thus the better convergence speed can be reached. In addition, the convergence upper bound of the FCGNN model is estimated and provided in details. Simulative results validate the superiority of the FCGNN model, as compared to the CGNN model for finding SoLE.

KW - Fast convergence

KW - Gradient neural network (GNN)

KW - Nonlinear activation function

KW - Performance evaluation

KW - Systems of linear equations (SoLE)

UR - http://www.scopus.com/inward/record.url?scp=85055544077&partnerID=8YFLogxK

U2 - 10.1016/j.ipl.2018.10.004

DO - 10.1016/j.ipl.2018.10.004

M3 - Journal article

AN - SCOPUS:85055544077

SN - 0020-0190

VL - 142

SP - 35

EP - 40

JO - Information Processing Letters

JF - Information Processing Letters

ER -

Nonlinear gradient neural network for solving system of linear equations

Abstract

Keywords

ASJC Scopus subject areas

More information

Other files and links

Fingerprint

Cite this