Frequency Fitness Assignment

Created by W.Langdon from gp-bibliography.bib Revision:1.3872

  author =       "Thomas Weise and Mingxu Wan and Pu Wang and 
                 Ke Tang and Alexandre Devert and Xin Yao",
  journal =      "IEEE Transactions on Evolutionary Computation",
  title =        "Frequency Fitness Assignment",
  year =         "2014",
  volume =       "18",
  number =       "2",
  month =        apr,
  pages =        "226--243",
  keywords =     "genetic algorithms, genetic programming, Combinatorial
                 Optimisation, Diversity, Fitness Assignment, Frequency,
                 Numerical Optimisation",
  DOI =          "doi:10.1109/TEVC.2013.2251885",
  ISSN =         "1089-778X",
  size =         "18 pages",
  abstract =     "Metaheuristic optimisation procedures such as
                 Evolutionary Algorithms are usually driven by an
                 objective function which rates the quality of a
                 candidate solution. However, it is not clear in
                 practice whether an objective function adequately
                 rewards intermediate solutions on the path to the
                 global optimum and it may exhibit deceptiveness,
                 epistasis, neutrality, ruggedness, and a lack of
                 causality. In this paper, we introduce the Frequency
                 Fitness H, subject to minimisation, that rates how
                 often solutions with the same objective value have been
                 discovered so far. The ideas behind this method are
                 that good solutions are hard to find and that if an
                 algorithm gets stuck at a local optimum, the frequency
                 of the objective values of the surrounding solutions
                 will increase over time, which will eventually allow it
                 to leave that region again. We substitute a Frequency
                 Fitness Assignment process (FFA) for the objective
                 function into several different optimisation
                 algorithms. We conduct a comprehensive set of
                 experiments: the synthesis of algorithms with Genetic
                 Programming (GP), the solution of MAX-3SAT problems
                 with Genetic Algorithms, classification with Memetic
                 Genetic Programming, and numerical optimisation with a
                 (1+1) Evolution Strategy, in order to verify the
                 utility of FFA. Given that they have no access to the
                 original objective function at all, it is surprising
                 that for some problems (e.g., the algorithm synthesis
                 task) the FFA-based algorithm variants perform
                 significantly better. However, this cannot be
                 guaranteed for all tested problems. We thus also
                 analyse scenarios where algorithms using FFA do not
                 perform better or even worse than with the original
                 objective functions.",
  notes =        "Also known as \cite{6476662}",

Genetic Programming entries for Thomas Weise Mingxu Wan Pu Wang Ke Tang Alexandre Devert Xin Yao