Tampere University of Technology

TUTCRIS Research Portal

Gaussian Scale Mixture Models For Robust Linear Multivariate Regression With Missing Data

Research output: Contribution to journalArticleScientificpeer-review

Standard

Gaussian Scale Mixture Models For Robust Linear Multivariate Regression With Missing Data. / Ala-Luhtala, Juha; Piche, Robert.

In: Communications in Statistics: Simulation and Computation, 2014.

Research output: Contribution to journalArticleScientificpeer-review

Harvard

APA

Vancouver

Author

Ala-Luhtala, Juha ; Piche, Robert. / Gaussian Scale Mixture Models For Robust Linear Multivariate Regression With Missing Data. In: Communications in Statistics: Simulation and Computation. 2014.

Bibtex - Download

@article{4c603b40a499461db0756b5d1ccd0430,
title = "Gaussian Scale Mixture Models For Robust Linear Multivariate Regression With Missing Data",
abstract = "We present an algorithm for multivariate robust Bayesian linear regression with missing data. The iterative algorithm computes an approximative posterior for the model parameters based on the variational Bayes (VB) method. Compared to the EM algorithm, the VB method has the advantage that the variance for the model parameters is also computed directly by the algorithm. We consider three families of Gaussian scale mixture models for the measurements, which include as special cases the multivariate t distribution, the multivariate Laplace distribution, and the contaminated normal model. The observations can contain missing values, assuming that the missing data mechanism can be ignored. A Matlab/Octave implementation of the algorithm is presented and applied to solve threereference examples from the literature.",
author = "Juha Ala-Luhtala and Robert Piche",
note = "Online first.Accepted author version posted online 19 Jun 2014<br/>Contribution: organisation=mat,FACT1=0.25<br/>Contribution: organisation=ase,FACT2=0.75<br/>Portfolio EDEND: 2014-11-25<br/>Publisher name: Taylor & Francis",
year = "2014",
doi = "10.1080/03610918.2013.875565",
language = "English",
journal = "Communications in Statistics: Simulation and Computation",
issn = "0361-0918",
publisher = "Taylor & Francis",

}

RIS (suitable for import to EndNote) - Download

TY - JOUR

T1 - Gaussian Scale Mixture Models For Robust Linear Multivariate Regression With Missing Data

AU - Ala-Luhtala, Juha

AU - Piche, Robert

N1 - Online first.Accepted author version posted online 19 Jun 2014<br/>Contribution: organisation=mat,FACT1=0.25<br/>Contribution: organisation=ase,FACT2=0.75<br/>Portfolio EDEND: 2014-11-25<br/>Publisher name: Taylor & Francis

PY - 2014

Y1 - 2014

N2 - We present an algorithm for multivariate robust Bayesian linear regression with missing data. The iterative algorithm computes an approximative posterior for the model parameters based on the variational Bayes (VB) method. Compared to the EM algorithm, the VB method has the advantage that the variance for the model parameters is also computed directly by the algorithm. We consider three families of Gaussian scale mixture models for the measurements, which include as special cases the multivariate t distribution, the multivariate Laplace distribution, and the contaminated normal model. The observations can contain missing values, assuming that the missing data mechanism can be ignored. A Matlab/Octave implementation of the algorithm is presented and applied to solve threereference examples from the literature.

AB - We present an algorithm for multivariate robust Bayesian linear regression with missing data. The iterative algorithm computes an approximative posterior for the model parameters based on the variational Bayes (VB) method. Compared to the EM algorithm, the VB method has the advantage that the variance for the model parameters is also computed directly by the algorithm. We consider three families of Gaussian scale mixture models for the measurements, which include as special cases the multivariate t distribution, the multivariate Laplace distribution, and the contaminated normal model. The observations can contain missing values, assuming that the missing data mechanism can be ignored. A Matlab/Octave implementation of the algorithm is presented and applied to solve threereference examples from the literature.

U2 - 10.1080/03610918.2013.875565

DO - 10.1080/03610918.2013.875565

M3 - Article

JO - Communications in Statistics: Simulation and Computation

JF - Communications in Statistics: Simulation and Computation

SN - 0361-0918

ER -