Guided Selection of Accurate Belief Propagation Fixed Points

Research output: Contribution to conferencePaperResearchpeer-review

Abstract

Belief propagation (BP) and the Bethe approximation are two closely relatedconcepts that both suffer from the existence of multiple fixed points (or stationarypoints). We propose a modification of BP, termed self-guided belief propagation(SBP), that incorporates the pairwise potentials only gradually; this essentiallyselects one specific fixed point and increases the accuracy without increasing thecomputational burden. We apply SBP to various models with Ising potentials andshow that: (i) SBP is superior in terms of accuracy whenever BP converges, and (ii)SBP obtains a unique, stable, and accurate solution whenever BP does not converge.
Original languageEnglish
Number of pages6
Publication statusPublished - 14 Dec 2019
EventMachine Learning and the Physical Sciences - Vancouver, Canada
Duration: 14 Dec 201914 Dec 2019

Conference

ConferenceMachine Learning and the Physical Sciences
CountryCanada
CityVancouver
Period14/12/1914/12/19

Cite this

Knoll, C., Kulmer, F., & Pernkopf, F. (2019). Guided Selection of Accurate Belief Propagation Fixed Points. Paper presented at Machine Learning and the Physical Sciences, Vancouver, Canada.

Guided Selection of Accurate Belief Propagation Fixed Points. / Knoll, Christian; Kulmer, Florian; Pernkopf, Franz.

2019. Paper presented at Machine Learning and the Physical Sciences, Vancouver, Canada.

Research output: Contribution to conferencePaperResearchpeer-review

Knoll, C, Kulmer, F & Pernkopf, F 2019, 'Guided Selection of Accurate Belief Propagation Fixed Points' Paper presented at Machine Learning and the Physical Sciences, Vancouver, Canada, 14/12/19 - 14/12/19, .
Knoll C, Kulmer F, Pernkopf F. Guided Selection of Accurate Belief Propagation Fixed Points. 2019. Paper presented at Machine Learning and the Physical Sciences, Vancouver, Canada.
Knoll, Christian ; Kulmer, Florian ; Pernkopf, Franz. / Guided Selection of Accurate Belief Propagation Fixed Points. Paper presented at Machine Learning and the Physical Sciences, Vancouver, Canada.6 p.
@conference{0d6a540a920f439cafb9436af91e3d23,
title = "Guided Selection of Accurate Belief Propagation Fixed Points",
abstract = "Belief propagation (BP) and the Bethe approximation are two closely relatedconcepts that both suffer from the existence of multiple fixed points (or stationarypoints). We propose a modification of BP, termed self-guided belief propagation(SBP), that incorporates the pairwise potentials only gradually; this essentiallyselects one specific fixed point and increases the accuracy without increasing thecomputational burden. We apply SBP to various models with Ising potentials andshow that: (i) SBP is superior in terms of accuracy whenever BP converges, and (ii)SBP obtains a unique, stable, and accurate solution whenever BP does not converge.",
author = "Christian Knoll and Florian Kulmer and Franz Pernkopf",
year = "2019",
month = "12",
day = "14",
language = "English",
note = "Machine Learning and the Physical Sciences ; Conference date: 14-12-2019 Through 14-12-2019",

}

TY - CONF

T1 - Guided Selection of Accurate Belief Propagation Fixed Points

AU - Knoll, Christian

AU - Kulmer, Florian

AU - Pernkopf, Franz

PY - 2019/12/14

Y1 - 2019/12/14

N2 - Belief propagation (BP) and the Bethe approximation are two closely relatedconcepts that both suffer from the existence of multiple fixed points (or stationarypoints). We propose a modification of BP, termed self-guided belief propagation(SBP), that incorporates the pairwise potentials only gradually; this essentiallyselects one specific fixed point and increases the accuracy without increasing thecomputational burden. We apply SBP to various models with Ising potentials andshow that: (i) SBP is superior in terms of accuracy whenever BP converges, and (ii)SBP obtains a unique, stable, and accurate solution whenever BP does not converge.

AB - Belief propagation (BP) and the Bethe approximation are two closely relatedconcepts that both suffer from the existence of multiple fixed points (or stationarypoints). We propose a modification of BP, termed self-guided belief propagation(SBP), that incorporates the pairwise potentials only gradually; this essentiallyselects one specific fixed point and increases the accuracy without increasing thecomputational burden. We apply SBP to various models with Ising potentials andshow that: (i) SBP is superior in terms of accuracy whenever BP converges, and (ii)SBP obtains a unique, stable, and accurate solution whenever BP does not converge.

M3 - Paper

ER -