A comparison of selection, recombination, and mutation parameter importance over a set of fifteen optimization tasks

  • Authors:

  • Edwin Roger Banks;Paul Agarwal;Marshall McBride;Claudette Owens

  • Affiliations:

  • COLSA Corporation, Huntsville, AL, USA;COLSA Corporation, Huntsville, AL, USA;U.S. Army Space and Missile Defense Command, Redstone Arsenal, AL, USA;U.S. Army Space and Missile Defense Command, Redstone Arsenal, AL, USA

  • Venue:
  • Proceedings of the 11th Annual Conference Companion on Genetic and Evolutionary Computation Conference: Late Breaking Papers
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

How does one choose an initial set of parameters for an evolutionary computing algorithm? Clearly some choices are dictated by the problem itself, such as the encoding of a problem solution, or how much time is available for running the evolution. Others, however, are frequently found by trial-and-error. These may include population sizes, number of populations, type of selection, recombination and mutation rates, and a variety of other parameters. Sometimes these parameters are allowed to co-evolve along with the solutions rather than by trial-and-error. But in both cases, an initial setting is needed for each parameter. When there are hundreds of parameters to be adjusted, as in some evolutionary computation tools, one would like to just spend time adjusting those that are believed to be most important, or sensitive, and leave the rest to start with an initial default value. Thus the primary goal of this paper is to establish the relative importance of each parameter. Establishing general guidance to assist in the determination of these initial default values is another primary goal of this paper. We propose to develop this guidance by studying the solutions resulting from variations around the default starting parameters applied across fifteen different application types.