Please use this identifier to cite or link to this item: https://scholarbank.nus.edu.sg/handle/10635/170805
Title: ON SOLVING THE REAL-TIME DECISION PROBLEM BY SIMULATION ANALYTICS AND PARALLELIZATION ACCELERATION
Authors: JIN XIAO
Keywords: Simulation Optimization, Simulation Analytics, Ranking and Selection, Discrete Optimization via Simulation, OCBA, Large Deviation
Issue Date: 26-Sep-2019
Citation: JIN XIAO (2019-09-26). ON SOLVING THE REAL-TIME DECISION PROBLEM BY SIMULATION ANALYTICS AND PARALLELIZATION ACCELERATION. ScholarBank@NUS Repository.
Abstract: In this thesis, we focus on solving the real-time decision problem. Since decision problems in real-time are highly contextual and only allow a short period to react, solving the problem by online optimization would be very challenging. Therefore, we propose a new framework called "Simulation Analytics" that learns the relation between the context and the corresponding optimal decision in advance through simulation observations, and then implement the learned rule online. To maximize the efficiency of this framework, we focus on the budget allocation problem. There are five main contributions of this work. 1) The first one is to define Simulation Analytics. 2) The second contribution is to propose an efficient/easy-to-implement sequential allocation algorithm in a pure black-box assumption, where the number of scenarios is limited and can fully be covered by simulation. 3) The third contribution is to propose an efficient allocation scheme where the number of scenarios is infinite. 4) The fourth contribution is to accelerate the solving of simulation optimization problems by parallelization of the simulation observations under a single run condition. 5) The final contribution is the adoption of Sample Average Approximation in estimating PCS that drives the budget allocation in a sequential allocation setting.
URI: https://scholarbank.nus.edu.sg/handle/10635/170805
Appears in Collections:Ph.D Theses (Open)

Show full item record
Files in This Item:
File Description SizeFormatAccess SettingsVersion 
JinX.pdf1.52 MBAdobe PDF

OPEN

NoneView/Download

Page view(s)

25
checked on Jun 11, 2021

Download(s)

12
checked on Jun 11, 2021

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.