基于保留策略的Bayesian网优化算法

提出了一种基于保留策略的Bayesian网优化算法。算法中通过学习Bayesian网络自动获取进化过程中各基因之间的依赖关系及分布描述,以便更好地指导算法的进化,并利用保留的父辈中间群体扩充学习数据集规模,解决了Bayesian网学习可靠性与较大群体规模之间的矛盾。实验表明,算法能够在有效收敛的前提下降低对群体规模的要求,具有较高的学习效率。

Bayesian Network Optimization Algorithm Based on Holding Strategy

This paper presents a Bayesian network optimization algorithm based on holding strategy.In the evolutionary process of the algorithm,Bayesian networks are learned to represent dependent relationship and distribution of genes,which are further utilized to guide the evolutionary process.And a holding strategy which holds some generations of ancestral inter population to expand the network learning dataset is employed to ensure the reliability of the network learning and to reduce the population size demanded.Experimental results demonstrate that the algorithm can warrant an efficient convergence,reduce the demand for large population,and improve learning performance.