Handle URI:
http://hdl.handle.net/10754/624108
Title:
Fast Bayesian optimal experimental design and its applications
Authors:
Long, Quan ( 0000-0002-0329-9437 )
Abstract:
We summarize our Laplace method and multilevel method of accelerating the computation of the expected information gain in a Bayesian Optimal Experimental Design (OED). Laplace method is a widely-used method to approximate an integration in statistics. We analyze this method in the context of optimal Bayesian experimental design and extend this method from the classical scenario, where a single dominant mode of the parameters can be completely-determined by the experiment, to the scenarios where a non-informative parametric manifold exists. We show that by carrying out this approximation the estimation of the expected Kullback-Leibler divergence can be significantly accelerated. While Laplace method requires a concentration of measure, multi-level Monte Carlo method can be used to tackle the problem when there is a lack of measure concentration. We show some initial results on this approach. The developed methodologies have been applied to various sensor deployment problems, e.g., impedance tomography and seismic source inversion.
KAUST Department:
Computer, Electrical and Mathematical Sciences & Engineering (CEMSE)
Conference/Event name:
Advances in Uncertainty Quantification Methods, Algorithms and Applications (UQAW 2015)
Issue Date:
7-Jan-2015
Type:
Presentation
Additional Links:
http://mediasite.kaust.edu.sa/Mediasite/Play/cbd53a4e142948ac95c3951b449037de1d?catalog=ca65101c-a4eb-4057-9444-45f799bd9c52
Appears in Collections:
Presentations; Conference on Advances in Uncertainty Quantification Methods, Algorithms and Applications (UQAW 2015)

Full metadata record

DC FieldValue Language
dc.contributor.authorLong, Quanen
dc.date.accessioned2017-06-05T08:35:48Z-
dc.date.available2017-06-05T08:35:48Z-
dc.date.issued2015-01-07-
dc.identifier.urihttp://hdl.handle.net/10754/624108-
dc.description.abstractWe summarize our Laplace method and multilevel method of accelerating the computation of the expected information gain in a Bayesian Optimal Experimental Design (OED). Laplace method is a widely-used method to approximate an integration in statistics. We analyze this method in the context of optimal Bayesian experimental design and extend this method from the classical scenario, where a single dominant mode of the parameters can be completely-determined by the experiment, to the scenarios where a non-informative parametric manifold exists. We show that by carrying out this approximation the estimation of the expected Kullback-Leibler divergence can be significantly accelerated. While Laplace method requires a concentration of measure, multi-level Monte Carlo method can be used to tackle the problem when there is a lack of measure concentration. We show some initial results on this approach. The developed methodologies have been applied to various sensor deployment problems, e.g., impedance tomography and seismic source inversion.en
dc.relation.urlhttp://mediasite.kaust.edu.sa/Mediasite/Play/cbd53a4e142948ac95c3951b449037de1d?catalog=ca65101c-a4eb-4057-9444-45f799bd9c52en
dc.titleFast Bayesian optimal experimental design and its applicationsen
dc.typePresentationen
dc.contributor.departmentComputer, Electrical and Mathematical Sciences & Engineering (CEMSE)en
dc.conference.dateJanuary 6-9, 2015en
dc.conference.nameAdvances in Uncertainty Quantification Methods, Algorithms and Applications (UQAW 2015)en
dc.conference.locationKAUSTen
kaust.authorLong, Quanen
All Items in KAUST are protected by copyright, with all rights reserved, unless otherwise indicated.