Leveraging benchmarking data for informed one-shot dynamic algorithm selection

Genetic and Evolutionary Computation Conference Companion (GECCO Companion), pp. 245–246

Abstract

A key challenge in the application of evolutionary algorithms in practice is the selection of an algorithm instance that best suits the problem at hand. What complicates this decision further is that different algorithms may be best suited for different stages of the optimization process. Dynamic algorithm selection and configuration are therefore well-researched topics in evolutionary computation. Two different settings are classically considered: hyper-heuristics and parameter control studies typically assume a setting in which the algorithm needs to be chosen and adjusted during the run, without prior information, other approaches such as hyper-parameter tuning and automated algorithm configuration assume the possibility of evaluating different configurations before making a final recommendation. In practical applications of evolutionary algorithms we are often in a middle-ground between these two settings, where one needs to decide upon the algorithm instance before the run ("oneshot" setting), but where we have (possibly lots of) data available on which we can base an informed decision.

Authors 3

1. Furong Ye Aachen Leiden Institute of Advanced Computer Science [Leiden]

   Leiden University

   Affiliation as printed

   Leiden University, Leiden, Netherlands

   Leiden Institute of Advanced Computer Science [Leiden]

2. Carola Doerr

   Centre National de la Recherche Scientifique · Sorbonne Université

   Affiliation as printed

   Sorbonne Université, Paris, France

   Centre National de la Recherche Scientifique

   Recherche Opérationnelle

   Sorbonne Université

3. Thomas Bäck Aachen Leiden Institute of Advanced Computer Science [Leiden]

   Leiden University

   Affiliation as printed

   Leiden University, Leiden, Netherlands

   Leiden Institute of Advanced Computer Science [Leiden]