Bayesian model evidence for order selection and correlation testing

Leigh Andrea Johnston; Iven Michiel Yvonne Mareels; Gary Francis Egan

doi:10.1109/IEMBS.2011.6091250

Bayesian model evidence for order selection and correlation testing

Leigh Andrea Johnston, Iven Michiel Yvonne Mareels, Gary Francis Egan

Research output: Chapter in Book/Report/Conference proceeding › Conference Paper › Research › peer-review

Abstract

Model selection is a critical component of data analysis procedures, and is particularly difficult for small numbers of observations such as is typical of functional MRI datasets. In this paper we derive two Bayesian evidence-based model selection procedures that exploit the existence of an analytic form for the linear Gaussian model class. Firstly, an evidence information criterion is proposed as a model order selection procedure for auto-regressive models, outperforming the commonly employed Akaike and Bayesian information criteria in simulated data. Secondly, an evidence-based method for testing change in linear correlation between datasets is proposed, which is demonstrated to outperform both the traditional statistical test of the null hypothesis of no correlation change and the likelihood ratio test.

Original language	English
Title of host publication	Annual International Conference of the IEEE Engineering in Medicine and Biology - Proceedings
Editors	Nigel Lovell
Place of Publication	United States
Publisher	IEEE, Institute of Electrical and Electronics Engineers
Pages	5048-5051
Number of pages	4
Volume	2011
ISBN (Print)	1557-170X
DOIs	https://doi.org/10.1109/IEMBS.2011.6091250
Publication status	Published - 2011
Externally published	Yes
Event	International Conference of the IEEE Engineering in Medicine and Biology Society 2011 - Boston, United States of America Duration: 30 Aug 2011 → 3 Sept 2011 Conference number: 33rd https://ieeexplore.ieee.org/xpl/conhome/6067544/proceeding (Proceedings)

Conference

Conference	International Conference of the IEEE Engineering in Medicine and Biology Society 2011
Abbreviated title	EMBC 2011
Country/Territory	United States of America
City	Boston
Period	30/08/11 → 3/09/11
Internet address	https://ieeexplore.ieee.org/xpl/conhome/6067544/proceeding (Proceedings)

Access to Document

10.1109/IEMBS.2011.6091250

Cite this

Johnston, L. A., Mareels, I. M. Y., & Egan, G. F. (2011). Bayesian model evidence for order selection and correlation testing. In N. Lovell (Ed.), Annual International Conference of the IEEE Engineering in Medicine and Biology - Proceedings (Vol. 2011, pp. 5048-5051). [6091250] IEEE, Institute of Electrical and Electronics Engineers. https://doi.org/10.1109/IEMBS.2011.6091250

@inproceedings{794568b6e0044ec59798f16669634c8a,

title = "Bayesian model evidence for order selection and correlation testing",

abstract = "Model selection is a critical component of data analysis procedures, and is particularly difficult for small numbers of observations such as is typical of functional MRI datasets. In this paper we derive two Bayesian evidence-based model selection procedures that exploit the existence of an analytic form for the linear Gaussian model class. Firstly, an evidence information criterion is proposed as a model order selection procedure for auto-regressive models, outperforming the commonly employed Akaike and Bayesian information criteria in simulated data. Secondly, an evidence-based method for testing change in linear correlation between datasets is proposed, which is demonstrated to outperform both the traditional statistical test of the null hypothesis of no correlation change and the likelihood ratio test.",

author = "Johnston, {Leigh Andrea} and Mareels, {Iven Michiel Yvonne} and Egan, {Gary Francis}",

year = "2011",

doi = "10.1109/IEMBS.2011.6091250",

language = "English",

isbn = "1557-170X",

volume = "2011",

pages = "5048--5051",

editor = "Nigel Lovell",

booktitle = "Annual International Conference of the IEEE Engineering in Medicine and Biology - Proceedings",

publisher = "IEEE, Institute of Electrical and Electronics Engineers",

address = "United States of America",

note = "International Conference of the IEEE Engineering in Medicine and Biology Society 2011, EMBC 2011 ; Conference date: 30-08-2011 Through 03-09-2011",

url = "https://ieeexplore.ieee.org/xpl/conhome/6067544/proceeding",

}

Johnston, LA, Mareels, IMY & Egan, GF 2011, Bayesian model evidence for order selection and correlation testing. in N Lovell (ed.), Annual International Conference of the IEEE Engineering in Medicine and Biology - Proceedings. vol. 2011, 6091250, IEEE, Institute of Electrical and Electronics Engineers, United States, pp. 5048-5051, International Conference of the IEEE Engineering in Medicine and Biology Society 2011, Boston, Massachusetts, United States of America, 30/08/11. https://doi.org/10.1109/IEMBS.2011.6091250

Bayesian model evidence for order selection and correlation testing. / Johnston, Leigh Andrea; Mareels, Iven Michiel Yvonne; Egan, Gary Francis.
Annual International Conference of the IEEE Engineering in Medicine and Biology - Proceedings. ed. / Nigel Lovell. Vol. 2011 United States: IEEE, Institute of Electrical and Electronics Engineers, 2011. p. 5048-5051 6091250.

Research output: Chapter in Book/Report/Conference proceeding › Conference Paper › Research › peer-review

TY - GEN

T1 - Bayesian model evidence for order selection and correlation testing

AU - Johnston, Leigh Andrea

AU - Mareels, Iven Michiel Yvonne

AU - Egan, Gary Francis

N1 - Conference code: 33rd

PY - 2011

Y1 - 2011

N2 - Model selection is a critical component of data analysis procedures, and is particularly difficult for small numbers of observations such as is typical of functional MRI datasets. In this paper we derive two Bayesian evidence-based model selection procedures that exploit the existence of an analytic form for the linear Gaussian model class. Firstly, an evidence information criterion is proposed as a model order selection procedure for auto-regressive models, outperforming the commonly employed Akaike and Bayesian information criteria in simulated data. Secondly, an evidence-based method for testing change in linear correlation between datasets is proposed, which is demonstrated to outperform both the traditional statistical test of the null hypothesis of no correlation change and the likelihood ratio test.

AB - Model selection is a critical component of data analysis procedures, and is particularly difficult for small numbers of observations such as is typical of functional MRI datasets. In this paper we derive two Bayesian evidence-based model selection procedures that exploit the existence of an analytic form for the linear Gaussian model class. Firstly, an evidence information criterion is proposed as a model order selection procedure for auto-regressive models, outperforming the commonly employed Akaike and Bayesian information criteria in simulated data. Secondly, an evidence-based method for testing change in linear correlation between datasets is proposed, which is demonstrated to outperform both the traditional statistical test of the null hypothesis of no correlation change and the likelihood ratio test.

UR - http://www.scopus.com/inward/record.url?scp=84055190647&partnerID=8YFLogxK

U2 - 10.1109/IEMBS.2011.6091250

DO - 10.1109/IEMBS.2011.6091250

M3 - Conference Paper

SN - 1557-170X

VL - 2011

SP - 5048

EP - 5051

BT - Annual International Conference of the IEEE Engineering in Medicine and Biology - Proceedings

A2 - Lovell, Nigel

PB - IEEE, Institute of Electrical and Electronics Engineers

CY - United States

T2 - International Conference of the IEEE Engineering in Medicine and Biology Society 2011

Y2 - 30 August 2011 through 3 September 2011

ER -

Bayesian model evidence for order selection and correlation testing

Abstract

Conference

Access to Document

Other files and links

Cite this