一种基于邻域互信息最大化和粒子群优化的特征基因选择方法  

Efficient Gene Selection Technique Based on Maximum Neighborhood Mutual Information and Particle Swarm Optimization

作　　者：徐天贺[1] 马媛媛 徐久成 

机构地区：[1]河南师范大学计算机与信息工程学院,河南新乡453007 [2]智慧商务与物联网技术河南省工程实验室,河南新乡453007 [3]河南省高校计算智能与数据挖掘工程技术研究中心,河南新乡453007

出　　处：《小型微型计算机系统》2016年第8期1775-1779,共5页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61370169;61402153;60873104)资助;河南省科技攻关重点项目(142102210056)资助;新乡市重点科技攻关计划项目(ZG13004)资助

摘　　要：针对基因表达谱数据高维度、低样本和数值型的特点,为了选出对分类有用的特征基因,提出了一种基于邻域互信息最大化和粒子群优化的特征基因选择方法.该方法首先基于邻域互信息最大化对基因进行排序,然后选择出排序在前且冗余较少的基因构成初选特征基因集合,最后用初选特征基因集合对粒子群优化算法的部分种群进行初始化,能较快的搜寻到较优的特征基因子集.实验结果表明,该算法可快速有效地选择特征基因,并获得较高的分类精度.According to the characteristics of microarray gene expression data in high dimension, low sample and continuous type, in order to select feature gene which are useful in classification, put forward an efficient gene selection technique based on maximum neighborhood mutual information and particle swarm optimization. The algorithm firstly sorts of genetic characteristics based on the maximum neighborhood mutual information, and then selects and sorts in the former and less redundant genes constitute the primary feature gene set, finally using the primary feature gene set to initialize population particle swarm optimization algorithm, fast search to the optimal feature subset of genes. The experimental results show that, the algorithm can quickly and effectively select feature gene, and obtains higher classification accuracy.

关 键 词：邻域互信息最大化 粒子群优化算法 基因选择 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]