Learning behavior and temporary minima of two-layer neural networks

Anne J. Annema, Klaas Hoen, Klaas Hoen, Hans Wallinga

Research output: Contribution to journalArticleAcademicpeer-review

5 Citations (Scopus)
66 Downloads (Pure)

Abstract

This paper presents a mathematical analysis of the occurrence of temporary minima during training of a single-output, two-layer neural network, with learning according to the back-propagation algorithm. A new vector decomposition method is introduced, which simplifies the mathematical analysis of learning of neural networks considerably. The analysis shows that temporary minima are inherent to multilayer networks learning. A number of numerical results illustrate the analytical conclusions.
Original languageEnglish
Pages (from-to)1387-1403
JournalNeural networks
Volume7
Issue number9
DOIs
Publication statusPublished - 1994

Fingerprint

Learning
Neural networks
Backpropagation algorithms
Multilayers
Decomposition

Keywords

  • IR-15176
  • METIS-112050
  • Temporary minimum
  • Pattern classification
  • Learning
  • Neural Networks
  • Multilayer perceptron
  • Back propagation

Cite this

Annema, Anne J. ; Hoen, Klaas ; Hoen, Klaas ; Wallinga, Hans. / Learning behavior and temporary minima of two-layer neural networks. In: Neural networks. 1994 ; Vol. 7, No. 9. pp. 1387-1403.
@article{f47958519bd146e38217eec901a01472,
title = "Learning behavior and temporary minima of two-layer neural networks",
abstract = "This paper presents a mathematical analysis of the occurrence of temporary minima during training of a single-output, two-layer neural network, with learning according to the back-propagation algorithm. A new vector decomposition method is introduced, which simplifies the mathematical analysis of learning of neural networks considerably. The analysis shows that temporary minima are inherent to multilayer networks learning. A number of numerical results illustrate the analytical conclusions.",
keywords = "IR-15176, METIS-112050, Temporary minimum, Pattern classification, Learning, Neural Networks, Multilayer perceptron, Back propagation",
author = "Annema, {Anne J.} and Klaas Hoen and Klaas Hoen and Hans Wallinga",
year = "1994",
doi = "10.1016/0893-6080(94)90087-6",
language = "English",
volume = "7",
pages = "1387--1403",
journal = "Neural networks",
issn = "0893-6080",
publisher = "Elsevier",
number = "9",

}

Learning behavior and temporary minima of two-layer neural networks. / Annema, Anne J.; Hoen, Klaas; Hoen, Klaas; Wallinga, Hans.

In: Neural networks, Vol. 7, No. 9, 1994, p. 1387-1403.

Research output: Contribution to journalArticleAcademicpeer-review

TY - JOUR

T1 - Learning behavior and temporary minima of two-layer neural networks

AU - Annema, Anne J.

AU - Hoen, Klaas

AU - Hoen, Klaas

AU - Wallinga, Hans

PY - 1994

Y1 - 1994

N2 - This paper presents a mathematical analysis of the occurrence of temporary minima during training of a single-output, two-layer neural network, with learning according to the back-propagation algorithm. A new vector decomposition method is introduced, which simplifies the mathematical analysis of learning of neural networks considerably. The analysis shows that temporary minima are inherent to multilayer networks learning. A number of numerical results illustrate the analytical conclusions.

AB - This paper presents a mathematical analysis of the occurrence of temporary minima during training of a single-output, two-layer neural network, with learning according to the back-propagation algorithm. A new vector decomposition method is introduced, which simplifies the mathematical analysis of learning of neural networks considerably. The analysis shows that temporary minima are inherent to multilayer networks learning. A number of numerical results illustrate the analytical conclusions.

KW - IR-15176

KW - METIS-112050

KW - Temporary minimum

KW - Pattern classification

KW - Learning

KW - Neural Networks

KW - Multilayer perceptron

KW - Back propagation

U2 - 10.1016/0893-6080(94)90087-6

DO - 10.1016/0893-6080(94)90087-6

M3 - Article

VL - 7

SP - 1387

EP - 1403

JO - Neural networks

JF - Neural networks

SN - 0893-6080

IS - 9

ER -