There exists a neural network that does not make avoidable mistakes

Andrew Ronald Gallant, Halbert White

Research output: Contribution to conferencePaper

73 Citations (Scopus)

Abstract

The authors show that a multiple-input, single-output, single-hidden-layer feedforward network with (known) hardwired connections from input to hidden layer, monotone squashing at the hidden layer and no squashing at the output embeds as a special case a so-called Fourier network, which yields a Fourier series approximation properties of Fourier series representations. In particular, approximation to any desired accuracy of any square integrable function can be achieved by such a network, using sufficiently many hidden units. In this sense, such networks do not make avoidable mistakes.

Original languageEnglish (US)
Pages657-664
Number of pages8
StatePublished - Dec 1 1988

Fingerprint

Fourier series
Neural networks

All Science Journal Classification (ASJC) codes

  • Engineering(all)

Cite this

@conference{488e16176ec44eb9bb5daa5229194e84,
title = "There exists a neural network that does not make avoidable mistakes",
abstract = "The authors show that a multiple-input, single-output, single-hidden-layer feedforward network with (known) hardwired connections from input to hidden layer, monotone squashing at the hidden layer and no squashing at the output embeds as a special case a so-called Fourier network, which yields a Fourier series approximation properties of Fourier series representations. In particular, approximation to any desired accuracy of any square integrable function can be achieved by such a network, using sufficiently many hidden units. In this sense, such networks do not make avoidable mistakes.",
author = "Gallant, {Andrew Ronald} and Halbert White",
year = "1988",
month = "12",
day = "1",
language = "English (US)",
pages = "657--664",

}

There exists a neural network that does not make avoidable mistakes. / Gallant, Andrew Ronald; White, Halbert.

1988. 657-664.

Research output: Contribution to conferencePaper

TY - CONF

T1 - There exists a neural network that does not make avoidable mistakes

AU - Gallant, Andrew Ronald

AU - White, Halbert

PY - 1988/12/1

Y1 - 1988/12/1

N2 - The authors show that a multiple-input, single-output, single-hidden-layer feedforward network with (known) hardwired connections from input to hidden layer, monotone squashing at the hidden layer and no squashing at the output embeds as a special case a so-called Fourier network, which yields a Fourier series approximation properties of Fourier series representations. In particular, approximation to any desired accuracy of any square integrable function can be achieved by such a network, using sufficiently many hidden units. In this sense, such networks do not make avoidable mistakes.

AB - The authors show that a multiple-input, single-output, single-hidden-layer feedforward network with (known) hardwired connections from input to hidden layer, monotone squashing at the hidden layer and no squashing at the output embeds as a special case a so-called Fourier network, which yields a Fourier series approximation properties of Fourier series representations. In particular, approximation to any desired accuracy of any square integrable function can be achieved by such a network, using sufficiently many hidden units. In this sense, such networks do not make avoidable mistakes.

UR - http://www.scopus.com/inward/record.url?scp=0024124325&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0024124325&partnerID=8YFLogxK

M3 - Paper

AN - SCOPUS:0024124325

SP - 657

EP - 664

ER -