Estimation Of Distribution Algorithm Based On Boosting

Estimation of Distribution Algorithms(EDAs) is a new kind Evolutionary Algorithms in the calculation domain. It is a combination of statistical learning theory and stochastic optimized algorithms. EDAs does not have the tradition evolutionary operators, but uew the probability model to describe the distribution of candidate solutions, then sample according to the probability model, produces the new population, carries on repeatedly to realizes population's evolution. Therefore, the core of EDAs is to estimate the probability distribution of the solution space. In this article, we proposed a new EDA based on Boosting algorithm, discussioned the discrete and continuous problems separately, selected independent distribution as weak learners, used the Boosting algorithm, estimated the probability distribution of the superior solutions, finally used the linear combination of independent distribution to approximate the probability distribution. By numerical results, in some complex function optimization problems, the new algorithm is better than UMDA or UMDAc.