Dispersion-Based Population Initialization

Reliable execution and analysis of an evolutionary algorithm (EA) normally requires many runs to provide reasonable assurance that stochastic effects have been properly considered. One of the first stochastic influences on the behavior of an EA is the population initialization. This has been recognized as a potentially serious problem to the performance of EAs but little progress has been made in improving the situation. Using a better population initialization algorithm would not be expected to improve the many-run average performance of an EA, but instead, it would be expected to reduce the variance of the results, without loss of average performance. This would provide researchers the opportunity to reliably examine their experimental results while requiring fewer EA runs for an appropriate statistical sample. This paper uses recent advances in the measurement and control of a population's dispersion in a search space to present a novel algorithm for better population initialization. Experimental verification of the usefulness of the new technique is provided.

[1]  Joseph O'Rourke,et al.  Handbook of Discrete and Computational Geometry, Second Edition , 1997 .

[2]  Harald Niederreiter,et al.  Implementation and tests of low-discrepancy sequences , 1992, TOMC.

[3]  Gert Vegter,et al.  In handbook of discrete and computational geometry , 1997 .

[4]  Kenji Shimada,et al.  Bubble mesh: automated triangular meshing of non-manifold geometry by sphere packing , 1995, SMA '95.

[5]  Gregory A. Kabatiansky,et al.  Sphere packing and coding theory , 2004, Handbook of Discrete and Computational Geometry, 2nd Ed..

[6]  Ronald W. Morrison,et al.  Designing Evolutionary Algorithms for Dynamic Environments , 2004, Natural Computing Series.

[7]  Jürgen Branke,et al.  Evolutionary Optimization in Dynamic Environments , 2001, Genetic Algorithms and Evolutionary Computation.

[8]  L FoxBennett,et al.  Implementation and tests of low-discrepancy sequences , 1992 .

[9]  H. Niederreiter.  Low-discrepancy and low-dispersion sequences , 1988 .

[10]  Robert F. Tichy,et al.  Sequences, Discrepancies and Applications , 1997 .

[11]  József Beck,et al.  Geometric Discrepancy Theory Anduniform Distribution , 2004, Handbook of Discrete and Computational Geometry, 2nd Ed..

[12]  A. Owen.  Monte Carlo Variance of Scrambled Net Quadrature , 1997 .

[13]  Marc Schoenauer,et al.  Alternative Random Initialization in Genetic Algorithms , 1997, ICGA.

[14]  Chuen-Tsai Sun,et al.  An educational genetic algorithms learning tool , 2001, IEEE Trans. Educ..