Paper Info
Title
Discrete optimization via simulation using Gaussian Markov random fields
Abstract
We construct a discrete optimization via simulation (DOvS) procedure using discrete Gaussian Markov random fields (GMRFs). Gaussian random fields (GRFs) are used in DOvS to balance exploration and exploitation. They enable computation of the expected improvement (EI) due to running the simulation to evaluate a feasible point of the optimization problem. Existing methods use GRFs with a continuous domain, which leads to dense covariance matrices, and therefore can be ill-suited for large-scale problems due to slow and ill-conditioned numerical computations. The use of GMRFs leads to sparse precision matrices, on which several sparse matrix techniques can be applied. To allocate the simulation effort throughout the procedure, we introduce a new EI criterion that incorporates the uncertainty in stochastic simulation by treating the value at the current optimal point as a random variable.
Year
DOI
Venue
2014
10.1109/WSC.2014.7020208
WSC '14: Winter Simulation Conference Savannah Georgia December, 2014
Keywords
Field
DocType
Gaussian processes,Markov processes,covariance matrices,numerical analysis,optimisation,random processes,sparse matrices,DOvS procedure,EI criterion,GMRF,continuous domain,dense covariance matrices,discrete Gaussian Markov random fields,discrete optimization-via-simulation procedure,expected improvement computation,exploitation ability,exploration ability,ill-conditioned numerical computations,large-scale problems,optimal point,random variable,sparse precision matrices,stochastic simulation,uncertainty analysis
Stochastic simulation,Mathematical optimization,Random variable,Random field,Computer science,Discrete optimization,Gaussian,Optimization problem,Sparse matrix,Covariance
Conference
ISSN
ISBN
Citations 
0891-7736
978-1-4673-9741-4
1
PageRank 
References 
Authors
0.37
8
3
Name
Order
Citations
PageRank
Peter Salemi1102.69
Barry L. Nelson21876257.62
Jeremy Staum37613.25