Preventing Premature Convergence in a Simple EDA Via Global Step Size Setting

  • Petr Pošík
Part of the Lecture Notes in Computer Science book series (LNCS, volume 5199)


When a simple real-valued estimation of distribution algorithm (EDA) with Gaussian model and maximum likelihood estimation of parameters is used, it converges prematurely even on the slope of the fitness function. The simplest way of preventing premature convergence by multiplying the variance estimate by a constant factor k each generation is studied. Recent works have shown that when increasing the dimensionality of the search space, such an algorithm becomes very quickly unable to traverse the slope and focus to the optimum at the same time. In this paper it is shown that when isotropic distributions with Gaussian or Cauchy distributed norms are used, the simple constant setting of k is able to ensure a reasonable behaviour of the EDA on the slope and in the valley of the fitness function at the same time.


Sphere Function Premature Convergence Cauchy Distribution Isotropic Distribution Distribution Algorithm 
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.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Larrañaga, P., Lozano, J.A. (eds.): Estimation of Distribution Algorithms. GENA. Kluwer Academic Publishers, Dordrecht (2002)zbMATHGoogle Scholar
  2. 2.
    Larrañaga, P., Lozano, J.A., Bengoetxea, E.: Estimation of distribution algorithms based on multivariate normal distributions and gaussian networks. Technical Report KZZA-IK-1-01, Dept. of Computer Science and Artificial Intelligence, University of Basque Country (2001)Google Scholar
  3. 3.
    Bosman, P.A.N., Thierens, D.: Expanding from discrete to continuous estimation of distribution algorithms: The IDEA. In: PPSN VI: Proceedings of the 6th International Conference on Parallel Problem Solving from Nature, London, UK, pp. 767–776. Springer, Heidelberg (2000)Google Scholar
  4. 4.
    Rudlof, S., Köppen, M.: Stochastic hill climbing by vectors of normal distributions. In: First Online Workshop on Soft Computing, Nagoya, Japan (1996)Google Scholar
  5. 5.
    Ahn, C.W., Ramakrishna, R.S., Goldberg, D.E.: Real-coded bayesian optimization algorithm: Bringing the strength of BOA into the continuous world. In: Deb, K., Poli, R., Banzhaf, W., Beyer, H.-G., Burke, E.K., Darwan, P.J., Dasgupta, D., Floreano, D., Foster, J.A., Harman, M., Holland, O., Lanzi, P.L., Spector, L., Tettamanzi, A., Thierens, D., Tyrrell, A.M. (eds.) GECCO 2004. LNCS, vol. 3102, pp. 840–851. Springer, Heidelberg (2004)CrossRefGoogle Scholar
  6. 6.
    Yuan, B., Gallagher, M.: On the importance of diversity maintenance in estimation of distribution algorithms. In: Beyer, H.G., O’Reilly, U.M. (eds.) Proceedings of the Genetic and Evolutionary Computation Conference GECCO 2005, vol. 1, pp. 719–726. ACM Press, New York (2005)CrossRefGoogle Scholar
  7. 7.
    Ocenasek, J., Kern, S., Hansen, N., Koumoutsakos, P.: A mixed bayesian optimization algorithm with variance adaptation. In: Yao, X., Burke, E.K., Lozano, J.A., Smith, J., Merelo-Guervós, J.J., Bullinaria, J.A., Rowe, J.E., Tiňo, P., Kabán, A., Schwefel, H.-P. (eds.) PPSN 2004. LNCS, vol. 3242, pp. 352–361. Springer, Heidelberg (2004)CrossRefGoogle Scholar
  8. 8.
    Grahl, J., Minner, S., Rothlauf, F.: Behaviour of UMDAc with truncation selection on monotonous functions. In: IEEE Congress on Evolutionary Computation, CEC 2005, vol. 3, pp. 2553–2559 (2005)Google Scholar
  9. 9.
    Gonzales, C., Lozano, J., Larranaga, P.: Mathematical modelling of UMDAc algorithm with tournament selection. International Journal of Approximate Reasoning 31(3), 313–340 (2002)CrossRefMathSciNetGoogle Scholar
  10. 10.
    Grahl, J., Bosman, P.A.N., Rothlauf, F.: The correlation-triggered adaptive variance scaling IDEA. In: Proceedings of the 8th annual conference on Genetic and Evolutionary Computation Conference - GECCO 2006, pp. 397–404. ACM Press, New York (2006)Google Scholar
  11. 11.
    Bosman, P.A.N., Grahl, J., Rothlauf, F.: SDR: A better trigger for adaptive variance scaling in normal EDAs. In: GECCO 2007: Proceedings of the 9th annual conference on Genetic and Evolutionary Computation, pp. 492–499. ACM Press, New York (2007)CrossRefGoogle Scholar
  12. 12.
    Yuan, B., Gallagher, M.: A mathematical modelling technique for the analysis of the dynamics of a simple continuous EDA. In: IEEE Congress on Evolutionary Computation, CEC 2006, Vancouver, Canada, pp. 1585–1591. IEEE Press, Los Alamitos (2006)Google Scholar
  13. 13.
    Pošík, P.: Gaussian EDA and truncation selection: Setting limits for sustainable progress. In: IEEE SMC International Conference on Distributed Human-Machine Systems, DHMS 2008, Athens, Greece. IEEE, Los Alamitos (2008)Google Scholar
  14. 14.
    Pošík, P.: Truncation selection and gaussian EDA: Bounds for sustainable progress in high-dimensional spaces. In: Giacobini, M., Brabazon, A., Cagnoni, S., Di Caro, G.A., Drechsler, R., Ekárt, A., Esparcia-Alcázar, A.I., Farooq, M., Fink, A., McCormack, J., O’Neill, M., Romero, J., Rothlauf, F., Squillero, G., Uyar, A.Ş., Yang, S. (eds.) EvoWorkshops 2008. LNCS, vol. 4974, pp. 525–534. Springer, Heidelberg (2008)CrossRefGoogle Scholar
  15. 15.
    Beyer, H.G., Deb, K.: On self-adaptive features in real-parameter evolutionary algorithms. IEEE Trans. on Evol. Comp. 5(3), 250–270 (2001)CrossRefGoogle Scholar
  16. 16.
    Grahl, J., Bosman, P.A.N., Minner, S.: Convergence phases, variance trajectories, and runtime analysis of continuous EDAs. In: GECCO 2007: Proceedings of the 9th annual conference on Genetic and evolutionary computation, pp. 516–522. ACM Press, New York (2007)Google Scholar
  17. 17.
    Rudolph, G.: Local convergence rates of simple evolutionary algorithms with cauchy mutations. IEEE Transactions on Evolutionary Computation 1, 249–258 (1997)CrossRefGoogle Scholar
  18. 18.
    Obuchowicz, A.: Multidimensional mutations in evolutionary algorithms based on real-valued representation. Int. J. Systems Science 34(7), 469–483 (2003)CrossRefzbMATHMathSciNetGoogle Scholar
  19. 19.
    Hansen, N., Gemperle, F., Auger, A., Koumoutsakos, P.: When do heavy-tail distributions help? In: Runarsson, T.P., Beyer, H.-G., Burke, E.K., Merelo-Guervós, J.J., Whitley, L.D., Yao, X. (eds.) PPSN 2006. LNCS, vol. 4193, pp. 62–71. Springer, Heidelberg (2006)CrossRefGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2008

Authors and Affiliations

  • Petr Pošík
    • 1
  1. 1.Faculty of Electrical Engineering, Department of CyberneticsCzech Technical University in PraguePrague 6Czech Republic

Personalised recommendations