Optimal Computing Budget Allocation to Select the Non-dominated Systems - a Large Deviations Perspective

Juxin Li, Weizhi Liu, Giulia Pedrielli, Loo Hay Lee, Ek Peng Chew

Research output: Contribution to journalArticle

4 Citations (Scopus)

Abstract

We consider the optimal computing budget allocation problem to select the non-dominated systems on finite sets under a stochastic multi-objective Ranking &amp; Selection setting. This problem has been addressed in the settings of correct selection guarantee when all the systems have normally distributed objectives with no correlation within solutions. We revisit this problem from a large deviation perspective and present a mathematically robust formulation that maximizes the lower bound of the rate function of the probability of false selection <formula><tex>$(P(FS))$</tex></formula>. The proposed formulation allows general distributions with sampling correlations across performance measures. Three budget allocation strategies are proposed. One is guaranteed to attain the global optimum of the lower bound of the rate function but has high computational cost. Therefore, a heuristic is proposed to save computational resources. Finally, for the case of normally distributed objectives, a computationally efficient procedure is proposed. Numerical experiments illustrate the significant improvements of the proposed strategies over others regarding the rate function of <formula><tex>$P(FS)$</tex></formula>.

Original languageEnglish (US)
JournalIEEE Transactions on Automatic Control
DOIs
StateAccepted/In press - Dec 1 2017

Fingerprint

Sampling
Costs
Experiments

Keywords

  • Computational modeling
  • Correlation
  • Gaussian distribution
  • large deviation theory
  • multi-objective optimization
  • optimal computing budget allocation
  • Optimization
  • Pareto optimality
  • ranking &amp; selection
  • Resource management
  • simulation optimization
  • Upper bound

ASJC Scopus subject areas

  • Control and Systems Engineering
  • Computer Science Applications
  • Electrical and Electronic Engineering

Cite this

Optimal Computing Budget Allocation to Select the Non-dominated Systems - a Large Deviations Perspective. / Li, Juxin; Liu, Weizhi; Pedrielli, Giulia; Lee, Loo Hay; Chew, Ek Peng.

In: IEEE Transactions on Automatic Control, 01.12.2017.

Research output: Contribution to journalArticle

@article{854d31063ce24856a1ba0d7b1ef2f5cd,
title = "Optimal Computing Budget Allocation to Select the Non-dominated Systems - a Large Deviations Perspective",
abstract = "We consider the optimal computing budget allocation problem to select the non-dominated systems on finite sets under a stochastic multi-objective Ranking & Selection setting. This problem has been addressed in the settings of correct selection guarantee when all the systems have normally distributed objectives with no correlation within solutions. We revisit this problem from a large deviation perspective and present a mathematically robust formulation that maximizes the lower bound of the rate function of the probability of false selection $(P(FS))$. The proposed formulation allows general distributions with sampling correlations across performance measures. Three budget allocation strategies are proposed. One is guaranteed to attain the global optimum of the lower bound of the rate function but has high computational cost. Therefore, a heuristic is proposed to save computational resources. Finally, for the case of normally distributed objectives, a computationally efficient procedure is proposed. Numerical experiments illustrate the significant improvements of the proposed strategies over others regarding the rate function of $P(FS)$.",
keywords = "Computational modeling, Correlation, Gaussian distribution, large deviation theory, multi-objective optimization, optimal computing budget allocation, Optimization, Pareto optimality, ranking &amp; selection, Resource management, simulation optimization, Upper bound",
author = "Juxin Li and Weizhi Liu and Giulia Pedrielli and Lee, {Loo Hay} and Chew, {Ek Peng}",
year = "2017",
month = "12",
day = "1",
doi = "10.1109/TAC.2017.2779603",
language = "English (US)",
journal = "IEEE Transactions on Automatic Control",
issn = "0018-9286",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - JOUR

T1 - Optimal Computing Budget Allocation to Select the Non-dominated Systems - a Large Deviations Perspective

AU - Li, Juxin

AU - Liu, Weizhi

AU - Pedrielli, Giulia

AU - Lee, Loo Hay

AU - Chew, Ek Peng

PY - 2017/12/1

Y1 - 2017/12/1

N2 - We consider the optimal computing budget allocation problem to select the non-dominated systems on finite sets under a stochastic multi-objective Ranking & Selection setting. This problem has been addressed in the settings of correct selection guarantee when all the systems have normally distributed objectives with no correlation within solutions. We revisit this problem from a large deviation perspective and present a mathematically robust formulation that maximizes the lower bound of the rate function of the probability of false selection $(P(FS))$. The proposed formulation allows general distributions with sampling correlations across performance measures. Three budget allocation strategies are proposed. One is guaranteed to attain the global optimum of the lower bound of the rate function but has high computational cost. Therefore, a heuristic is proposed to save computational resources. Finally, for the case of normally distributed objectives, a computationally efficient procedure is proposed. Numerical experiments illustrate the significant improvements of the proposed strategies over others regarding the rate function of $P(FS)$.

AB - We consider the optimal computing budget allocation problem to select the non-dominated systems on finite sets under a stochastic multi-objective Ranking & Selection setting. This problem has been addressed in the settings of correct selection guarantee when all the systems have normally distributed objectives with no correlation within solutions. We revisit this problem from a large deviation perspective and present a mathematically robust formulation that maximizes the lower bound of the rate function of the probability of false selection $(P(FS))$. The proposed formulation allows general distributions with sampling correlations across performance measures. Three budget allocation strategies are proposed. One is guaranteed to attain the global optimum of the lower bound of the rate function but has high computational cost. Therefore, a heuristic is proposed to save computational resources. Finally, for the case of normally distributed objectives, a computationally efficient procedure is proposed. Numerical experiments illustrate the significant improvements of the proposed strategies over others regarding the rate function of $P(FS)$.

KW - Computational modeling

KW - Correlation

KW - Gaussian distribution

KW - large deviation theory

KW - multi-objective optimization

KW - optimal computing budget allocation

KW - Optimization

KW - Pareto optimality

KW - ranking &amp; selection

KW - Resource management

KW - simulation optimization

KW - Upper bound

UR - http://www.scopus.com/inward/record.url?scp=85037585297&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85037585297&partnerID=8YFLogxK

U2 - 10.1109/TAC.2017.2779603

DO - 10.1109/TAC.2017.2779603

M3 - Article

AN - SCOPUS:85037585297

JO - IEEE Transactions on Automatic Control

JF - IEEE Transactions on Automatic Control

SN - 0018-9286

ER -