Advertisement

An Empirical Study of Population and Non-population Based Search Strategies for Optimizing a Combinatorical Problem

  • Antti Autere
Conference paper

Abstract

The performance of several algorithms for optimizing a combinatorical problem is compared empirically. The following topics are studied: • Can population based search strategies find good solutions faster than non-population ones?
  • • Are strategies that recombine two individuals to form new solutions better than those who use one individual only?

  • • Does the size of the population affect on the performance?

  • • Does the way how individuals are selected from the population affect on the performance?

The test problem used in these experiments is knapsack. 5000 variations of the test problem were solved. The number of the test function evaluations was recorded in the simulations. Statistical data was obtained in the form of cumulative frequencies and average values.

Based on these simulations the short answers to the four questions above are: yes; not necessarily; yes, but it depends on the selection strategy; and yes.

Keywords

Genetic Algorithm Test Problem Random Search Crossover Operation Uniform Crossover 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. [BH91]
    Thomas Back and Frank Hoffmeister. Extended selection mechanisms in genetic algorithms. In Richard K. Belew and Lashon B. Booker, editors, Fourth International Conference on Genetic Algorithms, San Diego, July 1991.Google Scholar
  2. [Gol89a]
    David E. Goldberg. Genetic Algorithms in Search Optimization and Machine Learning. Addison-Wesley, 1989.Google Scholar
  3. [Gol89b]
    David E. Goldberg. Sizing populations for serial and parallel genetic algorithms. In J. David Schaffer, editor, Third International Conferenceon Genetic Algorithms, George Mason University, June 1989.Google Scholar
  4. [KGV83]
    S. Kirkpatric, C. D. Gelatt, and M. P. Vecchi. Optimization by simulated annealing. Science, 220 (4598), 1983.Google Scholar
  5. [Man91]
    Vittorio Maniezzo. The rudes and the shrewds: an experimental comparison of several evolutionary heutistic applied to the qad problem. Technical Report 91-042, Politecnico di Milano, 1991.Google Scholar
  6. [RND77]
    Edward M. Reingold, Jurg Nievergelt, and Narsing Deo. Combinatorical Algorithms: Theory and Practice. Prentice-Hall, 1977.Google Scholar
  7. [SJ91]
    William M. Spears and Kenneth A. De Jong. On the virtues of parameterized uniform crossover. In Richard K. Belew and Lashon B. Booker, editors, Fourth International Conference on Genetic Algorithms, San Diego, July 1991.Google Scholar
  8. [Whi89]
    Darrel Whitley. The genitor algorithm and selection pressure: Why rank-based allocation of reproductive trials is best. In J. David Schaffer, editor, Third International Conference on Genetic Algorithms, George Mason University, June 1989.Google Scholar

Copyright information

© Springer-Verlag/Wien 1993

Authors and Affiliations

  • Antti Autere
    • 1
  1. 1.Department of Computer ScienceHelsinki University of TechnologyEspooFinland

Personalised recommendations