%0 Journal Article
%T 基于样本加权的基因特征选取模型
%A 芮兰兰
%A 张洁
%A 郭少勇
%A 熊翱
%J 北京邮电大学学报
%D 2016
%R 10.13190/j.jbupt.2016.s.017
%X 摘要 针对基因表达谱数据的特点，提出了一种基于样本加权的基因特征选取模型.首先提出一种样本权重的计算方法；其次结合样本权重改进信息增益度量标准，并用其衡量基因信息量的大小，同时将基因之间信息量的重复性视为基因噪声干扰，建立未消噪和消噪的基因特征选取模型；最后结合支持向量机、逻辑回归、神经网络和决策树4种分类器，将所提模型与常见的基因选取模型进行比较分析.实验结果表明，所提选取模型在不影响分类性能的前提下，具有较好的稳定性.</br>According to the characteristics of gene expression data, a gene feature selection model based on improved information gain was put forward. The improved information gain was proposed to measure gene information quantity with sample weight and a no de-noising and de-noising gene feature selection model was established. The proposed model is compared with common gene selection model using four classifiers. Experiments validate that the proposed method can improve stability of feature selection algorithms without sacrificing predictive accuracy.
%K 特征选取
%K 信息增益
%K 样本权重
%K 噪声干扰&lt
%K /br&gt
%K Key words： feature selection information gain sample weight noise interference
%U http://journal.bupt.edu.cn/CN/abstract/abstract2906.shtml