Testing correct model specification using extreme learning machines

Jin Seo Cho, Halbert White

Research output: Contribution to journalArticle

6 Citations (Scopus)

Abstract

Testing the correct model specification hypothesis for artificial neural network (ANN) models of the conditional mean is not standard. The traditional Wald, Lagrange multiplier, and quasi-likelihood ratio statistics weakly converge to functions of Gaussian processes, rather than to convenient chi-squared distributions. Also, their large-sample null distributions are problem dependent, limiting applicability. We overcome this challenge by applying functional regression methods of Cho et al. [8] to extreme learning machines (ELM). The Wald ELM (WELM) test statistic proposed here is easy to compute and has a large-sample standard chi-squared distribution under the null hypothesis of correct specification. We provide associated theory for time-series data and affirm our theory with some Monte Carlo experiments.

Original languageEnglish
Pages (from-to)2552-2565
Number of pages14
JournalNeurocomputing
Volume74
Issue number16
DOIs
Publication statusPublished - 2011 Sep 1

Fingerprint

Learning systems
Statistics
Specifications
Neural Networks (Computer)
Lagrange multipliers
Testing
Time series
Neural networks
Experiments
Machine Learning

All Science Journal Classification (ASJC) codes

  • Computer Science Applications
  • Cognitive Neuroscience
  • Artificial Intelligence

Cite this

Cho, Jin Seo ; White, Halbert. / Testing correct model specification using extreme learning machines. In: Neurocomputing. 2011 ; Vol. 74, No. 16. pp. 2552-2565.
@article{b5fe12e585ce471782d646e8a13e580b,
title = "Testing correct model specification using extreme learning machines",
abstract = "Testing the correct model specification hypothesis for artificial neural network (ANN) models of the conditional mean is not standard. The traditional Wald, Lagrange multiplier, and quasi-likelihood ratio statistics weakly converge to functions of Gaussian processes, rather than to convenient chi-squared distributions. Also, their large-sample null distributions are problem dependent, limiting applicability. We overcome this challenge by applying functional regression methods of Cho et al. [8] to extreme learning machines (ELM). The Wald ELM (WELM) test statistic proposed here is easy to compute and has a large-sample standard chi-squared distribution under the null hypothesis of correct specification. We provide associated theory for time-series data and affirm our theory with some Monte Carlo experiments.",
author = "Cho, {Jin Seo} and Halbert White",
year = "2011",
month = "9",
day = "1",
doi = "10.1016/j.neucom.2010.11.031",
language = "English",
volume = "74",
pages = "2552--2565",
journal = "Review of Economic Dynamics",
issn = "1094-2025",
publisher = "Academic Press Inc.",
number = "16",

}

Testing correct model specification using extreme learning machines. / Cho, Jin Seo; White, Halbert.

In: Neurocomputing, Vol. 74, No. 16, 01.09.2011, p. 2552-2565.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Testing correct model specification using extreme learning machines

AU - Cho, Jin Seo

AU - White, Halbert

PY - 2011/9/1

Y1 - 2011/9/1

N2 - Testing the correct model specification hypothesis for artificial neural network (ANN) models of the conditional mean is not standard. The traditional Wald, Lagrange multiplier, and quasi-likelihood ratio statistics weakly converge to functions of Gaussian processes, rather than to convenient chi-squared distributions. Also, their large-sample null distributions are problem dependent, limiting applicability. We overcome this challenge by applying functional regression methods of Cho et al. [8] to extreme learning machines (ELM). The Wald ELM (WELM) test statistic proposed here is easy to compute and has a large-sample standard chi-squared distribution under the null hypothesis of correct specification. We provide associated theory for time-series data and affirm our theory with some Monte Carlo experiments.

AB - Testing the correct model specification hypothesis for artificial neural network (ANN) models of the conditional mean is not standard. The traditional Wald, Lagrange multiplier, and quasi-likelihood ratio statistics weakly converge to functions of Gaussian processes, rather than to convenient chi-squared distributions. Also, their large-sample null distributions are problem dependent, limiting applicability. We overcome this challenge by applying functional regression methods of Cho et al. [8] to extreme learning machines (ELM). The Wald ELM (WELM) test statistic proposed here is easy to compute and has a large-sample standard chi-squared distribution under the null hypothesis of correct specification. We provide associated theory for time-series data and affirm our theory with some Monte Carlo experiments.

UR - http://www.scopus.com/inward/record.url?scp=80053626716&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=80053626716&partnerID=8YFLogxK

U2 - 10.1016/j.neucom.2010.11.031

DO - 10.1016/j.neucom.2010.11.031

M3 - Article

AN - SCOPUS:80053626716

VL - 74

SP - 2552

EP - 2565

JO - Review of Economic Dynamics

JF - Review of Economic Dynamics

SN - 1094-2025

IS - 16

ER -