摘要
文章在概率主成分分析分布估计算法的基础上提出了一种基于小生境的分布估计算法。将选出的最优个体集合随机划分为两部分,分别用概率主成分分析模型进行分布估计,并产生新个体。然后利用分布参数自动地调节小生境的参数,将产生的新个体融合到小生境当中。试验结果表明,该算法能够有效地防止早熟收敛,可以较大的提高算法的全局搜索效率。
Based on the estimation of distribution algorithms (EDAs) with pmbabilistic principal component analysis (PPCA), a kind of EDAs based on Niobe and PPCA is proposed .in this paper. We first Randomly divide the set of pmmis.ing solutions into two sub-sets and use PPCA model for esch sub-set to estimate the information of distribution and to generate new poputions, Then we use the parameters of distribution to adjust the parameter the niobe which fuse new poputions in it. Experimental results demonstrate that the new algorithm can effectively avoid premature and greatly improve efficiency of global search.
出处
《微电子学与计算机》
CSCD
北大核心
2006年第7期204-207,共4页
Microelectronics & Computer
关键词
进化计算
小生境算法
分布估计算法
概率主成分分析
Evolutionary computation, Niche algorithm, Estimation of distribution algorithm, Probabilistic principal component analysis