Adaptive evolutionary algorithms and extensions to the hyflex hyper-heuristic framework

  • Authors:
  • Gabriela Ochoa;James Walker;Matthew Hyde;Tim Curtois

  • Affiliations:
  • Department of Computing Science and Mathematics, University of Stirling, UK;School of Computer Science, University of Nottingham, UK;School of Computer Science, University of Nottingham, UK;School of Computer Science, University of Nottingham, UK

  • Venue:
  • PPSN'12 Proceedings of the 12th international conference on Parallel Problem Solving from Nature - Volume Part II
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

HyFlex is a recently proposed software framework for implementing hyper-heuristics and domain-independent heuristic optimisation algorithms [13]. Although it was originally designed to implement hyper-heuristics, it provides a population and a set of move operators of different types. This enable the implementation of adaptive versions of other heuristics such as evolutionary algorithms and iterated local search. The contributions of this article are twofold. First, a number of extensions to the HyFlex framework are proposed and implemented that enable the design of more effective adaptive heuristics. Second, it is demonstrated that adaptive evolutionary algorithms can be implemented within the framework, and that the use of crossover and a diversity metric produced improved results, including a new best-known solution, on the studied vehicle routing problem.