Parameter learning in MTE networks using incomplete data

Antonio Fernández, Helge Langseth, Thomas Dyhre Nielsen, Antonio Salmerón

Publikation: Konferencebidrag uden forlag/tidsskriftPaper uden forlag/tidsskriftForskningpeer review

4 Citationer (Scopus)
173 Downloads (Pure)

Resumé

Bayesian networks with mixtures of truncated exponentials (MTEs) are gaining popularity as a flexible modelling framework for hybrid domains. MTEs support efficient and exact inference algorithms, but estimating an MTE from data has turned out to be a difficult task. Current methods suffer from a considerable computational burden as well as the inability to handle missing values in the training data. In this paper we describe an EM-based algorithm for learning the maximum likelihood parameters of an MTE network when confronted with incomplete data. In order to overcome the computational difficulties we make certain distributional assumptions about the domain being modeled, thus focusing on a subclass of the general class of MTE networks. Preliminary empirical results indicate that the proposed method offers results that are inline with intuition.
OriginalsprogEngelsk
Publikationsdato2010
StatusUdgivet - 2010
BegivenhedThe Fifth European Workshop on Probabilistic Graphical Models - Helsinki, Finland
Varighed: 12 sep. 201015 sep. 2010

Konference

KonferenceThe Fifth European Workshop on Probabilistic Graphical Models
LandFinland
ByHelsinki
Periode12/09/201015/09/2010

Fingerprint

Bayesian networks
Maximum likelihood

Citer dette

Fernández, A., Langseth, H., Nielsen, T. D., & Salmerón, A. (2010). Parameter learning in MTE networks using incomplete data. Afhandling præsenteret på The Fifth European Workshop on Probabilistic Graphical Models, Helsinki, Finland.
Fernández, Antonio ; Langseth, Helge ; Nielsen, Thomas Dyhre ; Salmerón, Antonio. / Parameter learning in MTE networks using incomplete data. Afhandling præsenteret på The Fifth European Workshop on Probabilistic Graphical Models, Helsinki, Finland.
@conference{27ca3eae142d400f8c4548036f95698a,
title = "Parameter learning in MTE networks using incomplete data",
abstract = "Bayesian networks with mixtures of truncated exponentials (MTEs) are gaining popularity as a flexible modelling framework for hybrid domains. MTEs support efficient and exact inference algorithms, but estimating an MTE from data has turned out to be a difficult task. Current methods suffer from a considerable computational burden as well as the inability to handle missing values in the training data. In this paper we describe an EM-based algorithm for learning the maximum likelihood parameters of an MTE network when confronted with incomplete data. In order to overcome the computational difficulties we make certain distributional assumptions about the domain being modeled, thus focusing on a subclass of the general class of MTE networks. Preliminary empirical results indicate that the proposed method offers results that are inline with intuition.",
author = "Antonio Fern{\'a}ndez and Helge Langseth and Nielsen, {Thomas Dyhre} and Antonio Salmer{\'o}n",
note = "Publisher: HIIT Publications; The Fifth European Workshop on Probabilistic Graphical Models ; Conference date: 12-09-2010 Through 15-09-2010",
year = "2010",
language = "English",

}

Fernández, A, Langseth, H, Nielsen, TD & Salmerón, A 2010, 'Parameter learning in MTE networks using incomplete data' Paper fremlagt ved, Helsinki, Finland, 12/09/2010 - 15/09/2010, .

Parameter learning in MTE networks using incomplete data. / Fernández, Antonio; Langseth, Helge; Nielsen, Thomas Dyhre; Salmerón, Antonio.

2010. Afhandling præsenteret på The Fifth European Workshop on Probabilistic Graphical Models, Helsinki, Finland.

Publikation: Konferencebidrag uden forlag/tidsskriftPaper uden forlag/tidsskriftForskningpeer review

TY - CONF

T1 - Parameter learning in MTE networks using incomplete data

AU - Fernández, Antonio

AU - Langseth, Helge

AU - Nielsen, Thomas Dyhre

AU - Salmerón, Antonio

N1 - Publisher: HIIT Publications

PY - 2010

Y1 - 2010

N2 - Bayesian networks with mixtures of truncated exponentials (MTEs) are gaining popularity as a flexible modelling framework for hybrid domains. MTEs support efficient and exact inference algorithms, but estimating an MTE from data has turned out to be a difficult task. Current methods suffer from a considerable computational burden as well as the inability to handle missing values in the training data. In this paper we describe an EM-based algorithm for learning the maximum likelihood parameters of an MTE network when confronted with incomplete data. In order to overcome the computational difficulties we make certain distributional assumptions about the domain being modeled, thus focusing on a subclass of the general class of MTE networks. Preliminary empirical results indicate that the proposed method offers results that are inline with intuition.

AB - Bayesian networks with mixtures of truncated exponentials (MTEs) are gaining popularity as a flexible modelling framework for hybrid domains. MTEs support efficient and exact inference algorithms, but estimating an MTE from data has turned out to be a difficult task. Current methods suffer from a considerable computational burden as well as the inability to handle missing values in the training data. In this paper we describe an EM-based algorithm for learning the maximum likelihood parameters of an MTE network when confronted with incomplete data. In order to overcome the computational difficulties we make certain distributional assumptions about the domain being modeled, thus focusing on a subclass of the general class of MTE networks. Preliminary empirical results indicate that the proposed method offers results that are inline with intuition.

M3 - Paper without publisher/journal

ER -

Fernández A, Langseth H, Nielsen TD, Salmerón A. Parameter learning in MTE networks using incomplete data. 2010. Afhandling præsenteret på The Fifth European Workshop on Probabilistic Graphical Models, Helsinki, Finland.