贝叶斯优化算法在多目标优化问题中的应用

贝叶斯优化算法是近年来在进化算法领域兴起的一种新兴算法，用贝叶斯网络概率模型来显式地反映变量之间的依赖关系及可行解的分布，更符合实际问题的本质，在众多领域获得应用。针对多目标优化问题，在Pareto优化概念的基础上，用非占先排序及拥挤距离的方法来选择群体，形成解决多目标优化算法的Pareto贝叶斯优化算法，实验结果表明，Pareto贝叶斯优化算法要优于经典多目标优化算法NSGA-II。

pplication of Bayesian Optimization Algorithm in Multiobjective Problems

Bayesian Optimization Algorithms (BOA) are new paradigms for evolutionary computation. The techniques for modeling multivariate data by Bayesian networks are more essential in nature, and used in many fields. To deal with multiobjective problems, based on Pareto optimality concept, BOA is reformed in selection mechanism, and non-dominated sorting and crowding distance are used to get a new Pareto BOA. The experiment shows that the Pareto BOA outperforms the NSGA-II, which deals with many multiobjective problems successfully.