Model selection for penalized Gaussian Graphical Models

Vujacic, I; Wit, E

Risultato della ricerca: Paper

Abstract

High-dimensional data refers to the case in which the number of parameters is of one or more order greater than the sample size. Penalized Gaussian graphical models can be used to estimate the conditional independence graph in high-dimensional setting. In this setting, the crucial issue is to select the tuning parameter which regulates the sparsity of the graph. In this paper, we focus on estimating the "best" tuning parameter. We propose to select this tuning parameter by minimizing an information criterion based on the generalized information criterion and to use a stability selection approach in order to obtain a more stable graph. The performance of our method is compared with the state-of-art model selection procedures, including Akaike information criterion and Bayesian information criterion. A simulation study shows that our method performs better than the AIC, BIC.
Lingua originaleEnglish
Stato di pubblicazionePublished - 2013

Fingerprint

Parameter Tuning
Gaussian Model
Graphical Models
Model Selection
Information Criterion
Graph in graph theory
Bayesian Information Criterion
Akaike Information Criterion
Conditional Independence
Selection Procedures
High-dimensional Data
Sparsity
Sample Size
High-dimensional
Simulation Study
Estimate

Cita questo

Model selection for penalized Gaussian Graphical Models. / Vujacic, I; Wit, E.

2013.

Risultato della ricerca: Paper

@conference{bde00cfbf08a4090a0229319ba786d47,
title = "Model selection for penalized Gaussian Graphical Models",
abstract = "High-dimensional data refers to the case in which the number of parameters is of one or more order greater than the sample size. Penalized Gaussian graphical models can be used to estimate the conditional independence graph in high-dimensional setting. In this setting, the crucial issue is to select the tuning parameter which regulates the sparsity of the graph. In this paper, we focus on estimating the {"}best{"} tuning parameter. We propose to select this tuning parameter by minimizing an information criterion based on the generalized information criterion and to use a stability selection approach in order to obtain a more stable graph. The performance of our method is compared with the state-of-art model selection procedures, including Akaike information criterion and Bayesian information criterion. A simulation study shows that our method performs better than the AIC, BIC.",
author = "{Vujacic, I; Wit, E} and Angelo Mineo and Antonino Abbruzzo",
year = "2013",
language = "English",

}

TY - CONF

T1 - Model selection for penalized Gaussian Graphical Models

AU - Vujacic, I; Wit, E

AU - Mineo, Angelo

AU - Abbruzzo, Antonino

PY - 2013

Y1 - 2013

N2 - High-dimensional data refers to the case in which the number of parameters is of one or more order greater than the sample size. Penalized Gaussian graphical models can be used to estimate the conditional independence graph in high-dimensional setting. In this setting, the crucial issue is to select the tuning parameter which regulates the sparsity of the graph. In this paper, we focus on estimating the "best" tuning parameter. We propose to select this tuning parameter by minimizing an information criterion based on the generalized information criterion and to use a stability selection approach in order to obtain a more stable graph. The performance of our method is compared with the state-of-art model selection procedures, including Akaike information criterion and Bayesian information criterion. A simulation study shows that our method performs better than the AIC, BIC.

AB - High-dimensional data refers to the case in which the number of parameters is of one or more order greater than the sample size. Penalized Gaussian graphical models can be used to estimate the conditional independence graph in high-dimensional setting. In this setting, the crucial issue is to select the tuning parameter which regulates the sparsity of the graph. In this paper, we focus on estimating the "best" tuning parameter. We propose to select this tuning parameter by minimizing an information criterion based on the generalized information criterion and to use a stability selection approach in order to obtain a more stable graph. The performance of our method is compared with the state-of-art model selection procedures, including Akaike information criterion and Bayesian information criterion. A simulation study shows that our method performs better than the AIC, BIC.

UR - http://hdl.handle.net/10447/77414

M3 - Paper

ER -