Marginal Distribution in Copula Estimation of Distribution Algorithm Based Dynamic K-S test
Hui Zhao, Lifang Wang · 2012
Estimation of distribution algorithms based on copula, and a number of different distribution functions were selected as the dimension of the marginal distribution function by applying K-S test. Dynamic K-S test, that is, dynamic adjust frequency of K-S test in different stages. In the estimation of the probability model stage, according to the actual distribution of the dominant group respectively subject to inspection of the distribution function of each dimension. In the parameter estimation stage, according to fix the copula function parameters to make the emulation experiment. According to the obedience of the different marginal distribution function to sample separately, thus increasing the diversity of the population, and improving an execution efficiency of the estimation of distribution algorithms based on copula.