On the Need for Random Baseline Comparisons in Metaheuristic Search

Date
2018-01-03
Authors
Soper, Daniel
Contributor
Advisor
Department
Instructor
Depositor
Speaker
Researcher
Consultant
Interviewer
Annotator
Journal Title
Journal ISSN
Volume Title
Publisher
Volume
Number/Issue
Starting Page
Ending Page
Alternative Title
Abstract
A wide variety of organizations now regularly rely on established metaheuristic search algorithms in order to find solutions to otherwise intractable optimization problems. Unfortunately, neither the developers of these algorithms nor the organizations that rely on them typically assess the algorithms’ performance against a baseline random search strategy, opting instead to compare a specific algorithm’s performance against that of other metaheuristic search algorithms. This paper reveals the folly of such behavior, and shows by means of an optimization case study that simple random or nearly random search algorithms can, in certain circumstances, substantially outperform several of the most widely used metaheuristic search algorithms in finding solutions to optimization problems. The implications of the observed results for both organizations and researchers are presented and discussed.
Description
Keywords
Intelligent Decision Support for Logistics and Supply Chain Management, Metaheuristic search, Optimization, Random search
Citation
Extent
10 pages
Format
Geographic Location
Time Period
Related To
Proceedings of the 51st Hawaii International Conference on System Sciences
Table of Contents
Rights
Attribution-NonCommercial-NoDerivatives 4.0 International
Rights Holder
Local Contexts
Email libraryada-l@lists.hawaii.edu if you need this content in ADA-compliant format.