摘要
提出了一种基于自适应遗传算法的特征基因选择方法,首先建立一个基于Bhattacharyya距离的基因差异度模型,根据支持向量机(SVM)分类器的分类准确率选择出一个候选特征基因子集,然后利用自适应遗传算法搜索出一组最优特征基因组合,有效避免了遗传算法早熟收敛的缺陷,提高了全局寻优能力。对结肠癌基因表达谱数据进行仿真实验,分类准确率达到了94.6%,表明方法的可行性和有效性。
proposed a feature gene selection algorithm based on adaptive Genetic Algorithm. Firstly, created a genetic difference model based on Bhattacharyya distance and choose a candidate gene set that contained 50 genes according to SVM classifier accuracy. Secondly, selected some feature genes using adaptive Genetic Algorithm avoiding premature convergence. Using this method on experiment on colon cancer gene expression data, the classification accuracy rate ar- rives at 94.6% when the feature genes are four. So the experiment results show the method is feasible and effective.
出处
《科技通报》
北大核心
2011年第2期241-245,共5页
Bulletin of Science and Technology
关键词
基因表达谱
特征基因
自适应遗传算法
支持向量机
gene expression
feature gene
adaptive genetic algorithm
support vector machine
