Mixtures of truncated basis functions

Helge Langseth, Thomas Dyhre Nielsen, Rafael Rumí, Antonio Salmerón

Research output: Contribution to journalJournal articleResearchpeer-review

50 Citations (Scopus)
489 Downloads (Pure)

Abstract

In this paper we propose a framework, called mixtures of truncated basis functions (MoTBFs), for representing general hybrid Bayesian networks. The proposed framework generalizes both the mixture of truncated exponentials (MTEs) framework and the mixture of polynomials (MoPs) framework. Similar to MTEs and MoPs, MoTBFs are defined so that the potentials are closed under combination and marginalization, which ensures that inference in MoTBF networks can be performed efficiently using the Shafer-Shenoy architecture.
Based on a generalized Fourier series approximation, we devise a method for efficiently approximating an arbitrary density function using the MoTBF framework. The transla- tion method is more flexible than existing MTE or MoP-based methods, and it supports an online/anytime tradeoff between the accuracy and the complexity of the approxima- tion. Experimental results show that the approximations obtained are either comparable or significantly better than the approximations obtained using existing methods.
Original languageEnglish
JournalInternational Journal of Approximate Reasoning
Volume53
Issue number2
Pages (from-to)212-227
ISSN0888-613X
DOIs
Publication statusPublished - 2012

Fingerprint

Basis Functions
Approximation
Polynomials
Polynomial
Fourier series
Bayesian networks
Bayesian Networks
Density Function
Probability density function
Trade-offs
Framework
Closed
Generalise
Experimental Results
Arbitrary

Cite this

Langseth, Helge ; Nielsen, Thomas Dyhre ; Rumí, Rafael ; Salmerón, Antonio. / Mixtures of truncated basis functions. In: International Journal of Approximate Reasoning. 2012 ; Vol. 53, No. 2. pp. 212-227.
@article{bed7b36883544a56baaf37f01c34904b,
title = "Mixtures of truncated basis functions",
abstract = "In this paper we propose a framework, called mixtures of truncated basis functions (MoTBFs), for representing general hybrid Bayesian networks. The proposed framework generalizes both the mixture of truncated exponentials (MTEs) framework and the mixture of polynomials (MoPs) framework. Similar to MTEs and MoPs, MoTBFs are defined so that the potentials are closed under combination and marginalization, which ensures that inference in MoTBF networks can be performed efficiently using the Shafer-Shenoy architecture.Based on a generalized Fourier series approximation, we devise a method for efficiently approximating an arbitrary density function using the MoTBF framework. The transla- tion method is more flexible than existing MTE or MoP-based methods, and it supports an online/anytime tradeoff between the accuracy and the complexity of the approxima- tion. Experimental results show that the approximations obtained are either comparable or significantly better than the approximations obtained using existing methods.",
author = "Helge Langseth and Nielsen, {Thomas Dyhre} and Rafael Rum{\'i} and Antonio Salmer{\'o}n",
year = "2012",
doi = "10.1016/j.ijar.2011.10.004",
language = "English",
volume = "53",
pages = "212--227",
journal = "International Journal of Approximate Reasoning",
issn = "0888-613X",
publisher = "Elsevier",
number = "2",

}

Mixtures of truncated basis functions. / Langseth, Helge; Nielsen, Thomas Dyhre; Rumí, Rafael; Salmerón, Antonio.

In: International Journal of Approximate Reasoning, Vol. 53, No. 2, 2012, p. 212-227.

Research output: Contribution to journalJournal articleResearchpeer-review

TY - JOUR

T1 - Mixtures of truncated basis functions

AU - Langseth, Helge

AU - Nielsen, Thomas Dyhre

AU - Rumí, Rafael

AU - Salmerón, Antonio

PY - 2012

Y1 - 2012

N2 - In this paper we propose a framework, called mixtures of truncated basis functions (MoTBFs), for representing general hybrid Bayesian networks. The proposed framework generalizes both the mixture of truncated exponentials (MTEs) framework and the mixture of polynomials (MoPs) framework. Similar to MTEs and MoPs, MoTBFs are defined so that the potentials are closed under combination and marginalization, which ensures that inference in MoTBF networks can be performed efficiently using the Shafer-Shenoy architecture.Based on a generalized Fourier series approximation, we devise a method for efficiently approximating an arbitrary density function using the MoTBF framework. The transla- tion method is more flexible than existing MTE or MoP-based methods, and it supports an online/anytime tradeoff between the accuracy and the complexity of the approxima- tion. Experimental results show that the approximations obtained are either comparable or significantly better than the approximations obtained using existing methods.

AB - In this paper we propose a framework, called mixtures of truncated basis functions (MoTBFs), for representing general hybrid Bayesian networks. The proposed framework generalizes both the mixture of truncated exponentials (MTEs) framework and the mixture of polynomials (MoPs) framework. Similar to MTEs and MoPs, MoTBFs are defined so that the potentials are closed under combination and marginalization, which ensures that inference in MoTBF networks can be performed efficiently using the Shafer-Shenoy architecture.Based on a generalized Fourier series approximation, we devise a method for efficiently approximating an arbitrary density function using the MoTBF framework. The transla- tion method is more flexible than existing MTE or MoP-based methods, and it supports an online/anytime tradeoff between the accuracy and the complexity of the approxima- tion. Experimental results show that the approximations obtained are either comparable or significantly better than the approximations obtained using existing methods.

UR - http://www.scopus.com/inward/record.url?scp=84855417418&partnerID=8YFLogxK

U2 - 10.1016/j.ijar.2011.10.004

DO - 10.1016/j.ijar.2011.10.004

M3 - Journal article

VL - 53

SP - 212

EP - 227

JO - International Journal of Approximate Reasoning

JF - International Journal of Approximate Reasoning

SN - 0888-613X

IS - 2

ER -