Comparison of gradient based training algorithms for multilayer perceptrons

George Irwin; Gordon Lightbody; Sean McLoone

Comparison of gradient based training algorithms for multilayer perceptrons

George Irwin
, Gordon Lightbody
, Sean McLoone

Queen's University Belfast

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

Abstract

The training speeds of Batch Backpropagation using steepest descent, Conjugate Gradient and Quasi-Newton algorithms for a feedforward neural network are compared. Results illustrating the advantages of the Hessian-based techniques are given and issues affecting speed discussed.

Original language	English
Pages (from-to)	11/1-11/6
Journal	IEE Colloquium (Digest)
Issue number	136
Publication status	Published - 1 Jan 1994
Externally published	Yes
Event	Proceedings of the IEE Colloquium on Advances in Neural Networks for Control and Systems - London, UK Duration: 25 May 1994 → 27 May 1994

Cite this

@article{9181ff49e48140ab948dbb46ec6419b4,

title = "Comparison of gradient based training algorithms for multilayer perceptrons",

abstract = "The training speeds of Batch Backpropagation using steepest descent, Conjugate Gradient and Quasi-Newton algorithms for a feedforward neural network are compared. Results illustrating the advantages of the Hessian-based techniques are given and issues affecting speed discussed.",

author = "George Irwin and Gordon Lightbody and Sean McLoone",

year = "1994",

month = jan,

day = "1",

language = "English",

pages = "11/1--11/6",

journal = "IEE Colloquium (Digest)",

issn = "0963-3308",

publisher = "Institute of Electrical Engineers",

number = "136",

note = "Proceedings of the IEE Colloquium on Advances in Neural Networks for Control and Systems ; Conference date: 25-05-1994 Through 27-05-1994",

}

TY - JOUR

T1 - Comparison of gradient based training algorithms for multilayer perceptrons

AU - Irwin, George

AU - Lightbody, Gordon

AU - McLoone, Sean

PY - 1994/1/1

Y1 - 1994/1/1

N2 - The training speeds of Batch Backpropagation using steepest descent, Conjugate Gradient and Quasi-Newton algorithms for a feedforward neural network are compared. Results illustrating the advantages of the Hessian-based techniques are given and issues affecting speed discussed.

AB - The training speeds of Batch Backpropagation using steepest descent, Conjugate Gradient and Quasi-Newton algorithms for a feedforward neural network are compared. Results illustrating the advantages of the Hessian-based techniques are given and issues affecting speed discussed.

UR - https://www.scopus.com/pages/publications/0028780064

M3 - Article

AN - SCOPUS:0028780064

SN - 0963-3308

SP - 11/1-11/6

JO - IEE Colloquium (Digest)

JF - IEE Colloquium (Digest)

IS - 136

T2 - Proceedings of the IEE Colloquium on Advances in Neural Networks for Control and Systems

Y2 - 25 May 1994 through 27 May 1994

ER -

Comparison of gradient based training algorithms for multilayer perceptrons

Abstract

Other files and links

Fingerprint

Cite this