Selecting the tuning parameter in penalized Gaussian graphical models

Risultato della ricerca: Article

Abstract

Penalized inference of Gaussian graphical models is a way to assess the conditional independence structure in multivariate problems. In this setting, the conditional independence structure, corresponding to a graph, is related to the choice of the tuning parameter, which determines the model complexity or degrees of freedom. There has been little research on the degrees of freedom for penalized Gaussian graphical models. In this paper, we propose an estimator of the degrees of freedom in l1-penalized Gaussian graphical models. Specically, we derive an estimator inspired by the generalized information criterion and propose to use this estimator as the bias term for two information criteria. We called these tuning parameter selectors GAIC and GBIC. These selectors can be used to choose the tuning parameter, i.e. the optimal tuning parameter is the minimizer of GAIC or GBIC. A simulation study shows that GAIC tends to improve the performance of both AIC-type and CV-type model selectors, in terms of estimation quality (entropy loss function) in the high-dimensional setting. Moreover, GBIC model selector improves the performance of both BIC-type and CV-type model selectors, in terms of support recovery (Fscore). A data analysis shows that GBIC selects a tuning parameter that produces a sparser graph with respect to BIC and a CV-type model selector (KLCV).
Lingua originaleEnglish
pagine (da-a)559-569
Numero di pagine11
RivistaStatistics and Computing
Volume29
Stato di pubblicazionePublished - 2019

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computational Theory and Mathematics
  • Statistics, Probability and Uncertainty
  • Statistics and Probability

Cita questo

Selecting the tuning parameter in penalized Gaussian graphical models. /.

In: Statistics and Computing, Vol. 29, 2019, pag. 559-569.

Risultato della ricerca: Article

@article{0c39fddba8a4456c87521a75a9fd6c7b,
title = "Selecting the tuning parameter in penalized Gaussian graphical models",
abstract = "Penalized inference of Gaussian graphical models is a way to assess the conditional independence structure in multivariate problems. In this setting, the conditional independence structure, corresponding to a graph, is related to the choice of the tuning parameter, which determines the model complexity or degrees of freedom. There has been little research on the degrees of freedom for penalized Gaussian graphical models. In this paper, we propose an estimator of the degrees of freedom in l1-penalized Gaussian graphical models. Specically, we derive an estimator inspired by the generalized information criterion and propose to use this estimator as the bias term for two information criteria. We called these tuning parameter selectors GAIC and GBIC. These selectors can be used to choose the tuning parameter, i.e. the optimal tuning parameter is the minimizer of GAIC or GBIC. A simulation study shows that GAIC tends to improve the performance of both AIC-type and CV-type model selectors, in terms of estimation quality (entropy loss function) in the high-dimensional setting. Moreover, GBIC model selector improves the performance of both BIC-type and CV-type model selectors, in terms of support recovery (Fscore). A data analysis shows that GBIC selects a tuning parameter that produces a sparser graph with respect to BIC and a CV-type model selector (KLCV).",
author = "Antonino Abbruzzo",
year = "2019",
language = "English",
volume = "29",
pages = "559--569",
journal = "Statistics and Computing",
issn = "0960-3174",
publisher = "Springer Netherlands",

}

TY - JOUR

T1 - Selecting the tuning parameter in penalized Gaussian graphical models

AU - Abbruzzo, Antonino

PY - 2019

Y1 - 2019

N2 - Penalized inference of Gaussian graphical models is a way to assess the conditional independence structure in multivariate problems. In this setting, the conditional independence structure, corresponding to a graph, is related to the choice of the tuning parameter, which determines the model complexity or degrees of freedom. There has been little research on the degrees of freedom for penalized Gaussian graphical models. In this paper, we propose an estimator of the degrees of freedom in l1-penalized Gaussian graphical models. Specically, we derive an estimator inspired by the generalized information criterion and propose to use this estimator as the bias term for two information criteria. We called these tuning parameter selectors GAIC and GBIC. These selectors can be used to choose the tuning parameter, i.e. the optimal tuning parameter is the minimizer of GAIC or GBIC. A simulation study shows that GAIC tends to improve the performance of both AIC-type and CV-type model selectors, in terms of estimation quality (entropy loss function) in the high-dimensional setting. Moreover, GBIC model selector improves the performance of both BIC-type and CV-type model selectors, in terms of support recovery (Fscore). A data analysis shows that GBIC selects a tuning parameter that produces a sparser graph with respect to BIC and a CV-type model selector (KLCV).

AB - Penalized inference of Gaussian graphical models is a way to assess the conditional independence structure in multivariate problems. In this setting, the conditional independence structure, corresponding to a graph, is related to the choice of the tuning parameter, which determines the model complexity or degrees of freedom. There has been little research on the degrees of freedom for penalized Gaussian graphical models. In this paper, we propose an estimator of the degrees of freedom in l1-penalized Gaussian graphical models. Specically, we derive an estimator inspired by the generalized information criterion and propose to use this estimator as the bias term for two information criteria. We called these tuning parameter selectors GAIC and GBIC. These selectors can be used to choose the tuning parameter, i.e. the optimal tuning parameter is the minimizer of GAIC or GBIC. A simulation study shows that GAIC tends to improve the performance of both AIC-type and CV-type model selectors, in terms of estimation quality (entropy loss function) in the high-dimensional setting. Moreover, GBIC model selector improves the performance of both BIC-type and CV-type model selectors, in terms of support recovery (Fscore). A data analysis shows that GBIC selects a tuning parameter that produces a sparser graph with respect to BIC and a CV-type model selector (KLCV).

UR - http://hdl.handle.net/10447/351495

UR - https://link.springer.com/article/10.1007%2Fs11222-018-9823-5

M3 - Article

VL - 29

SP - 559

EP - 569

JO - Statistics and Computing

JF - Statistics and Computing

SN - 0960-3174

ER -