Global convergence conditions in maximum likelihood estimation

Yiqun Zou; William P. Heath

doi:10.1080/00207179.2012.658085

Global convergence conditions in maximum likelihood estimation

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

Abstract

Maximum likelihood estimation has been widely applied in system identification because of consistency, its asymptotic efficiency and sufficiency. However, gradient-based optimisation of the likelihood function might end up in local convergence. In this article we derive various new non-local-minimum conditions in both open and closed-loop system when the noise distribution is a Gaussian process. Here we consider different model structures, in particular ARARMAX, BJ and OE models. © 2012 Taylor & Francis.

Original language	English
Pages (from-to)	475-490
Number of pages	15
Journal	International Journal of Control
Volume	85
Issue number	5
DOIs	https://doi.org/10.1080/00207179.2012.658085
Publication status	Published - 1 May 2012

Keywords

asymptotic efficiency and sufficiency
consistency
global/local convergence
MLE
non-local-minimum conditions
optimisation

Access to Document

10.1080/00207179.2012.658085

Cite this

@article{c611948b1453415693c1986532b72c52,

title = "Global convergence conditions in maximum likelihood estimation",

abstract = "Maximum likelihood estimation has been widely applied in system identification because of consistency, its asymptotic efficiency and sufficiency. However, gradient-based optimisation of the likelihood function might end up in local convergence. In this article we derive various new non-local-minimum conditions in both open and closed-loop system when the noise distribution is a Gaussian process. Here we consider different model structures, in particular ARARMAX, BJ and OE models. {\textcopyright} 2012 Taylor & Francis.",

keywords = "asymptotic efficiency and sufficiency, consistency, global/local convergence, MLE, non-local-minimum conditions, optimisation",

author = "Yiqun Zou and Heath, {William P.}",

year = "2012",

month = may,

day = "1",

doi = "10.1080/00207179.2012.658085",

language = "English",

volume = "85",

pages = "475--490",

journal = "International Journal of Control",

issn = "1366-5820",

publisher = "Taylor & Francis",

number = "5",

}

TY - JOUR

T1 - Global convergence conditions in maximum likelihood estimation

AU - Zou, Yiqun

AU - Heath, William P.

PY - 2012/5/1

Y1 - 2012/5/1

N2 - Maximum likelihood estimation has been widely applied in system identification because of consistency, its asymptotic efficiency and sufficiency. However, gradient-based optimisation of the likelihood function might end up in local convergence. In this article we derive various new non-local-minimum conditions in both open and closed-loop system when the noise distribution is a Gaussian process. Here we consider different model structures, in particular ARARMAX, BJ and OE models. © 2012 Taylor & Francis.

AB - Maximum likelihood estimation has been widely applied in system identification because of consistency, its asymptotic efficiency and sufficiency. However, gradient-based optimisation of the likelihood function might end up in local convergence. In this article we derive various new non-local-minimum conditions in both open and closed-loop system when the noise distribution is a Gaussian process. Here we consider different model structures, in particular ARARMAX, BJ and OE models. © 2012 Taylor & Francis.

KW - asymptotic efficiency and sufficiency

KW - consistency

KW - global/local convergence

KW - MLE

KW - non-local-minimum conditions

KW - optimisation

U2 - 10.1080/00207179.2012.658085

DO - 10.1080/00207179.2012.658085

M3 - Article

SN - 1366-5820

VL - 85

SP - 475

EP - 490

JO - International Journal of Control

JF - International Journal of Control

IS - 5

ER -

Global convergence conditions in maximum likelihood estimation

Abstract

Keywords

Access to Document

Fingerprint

Cite this