A Multi-population Parallel Estimation of Distribution Algorithms Based on Clayton and Gumbel Copulas

摘要

The idea of multi-population parallel strategy and the copula theory are introduced into the Estimation of Distribution Algorithm (EDA), and a new parallel EDA is proposed in this paper. In this algorithm, the population is divided into some subpopulations. Different copula is used to estimate the distribution model in each subpopulation. Two copulas, Clayton and Gumbel, are used in this paper. To estimate the distribution function is to estimate the copula and the margins. New individuals are generated according to the copula and the margins. In order to increase the diversity of the subpopulation, the elites of one subpopulation are learned by the other subpopulation. The experiments show the proposed algorithm performs better than the basic copula EDA and some classical ED As in speed and in precision.