摘要
采用支持向量机方法(SVM)对上千维的基因表达数据分析时,算法的运行时间比较长。为了解决这种情况,本文采用了基于主成分分析的支持向量机(PCA-SVM)和基于核主成分分析的支持向量机(KPCA-SVM)两种算法对数据进行降维和分类,既可以整合基因数据的特征信息又可以缩短计算时间。本文比较了累计贡献率不同时两种算法的分类准确率,实验结果表明,PCA-SVM分类准确率与累计贡献率二者之间没有明确规律,KPCA-SVM分类准确率随累计贡献率的降低存在降低或者保持不变的趋势。
When the support vector machine （ SVM） method is applied in the analysis of gene expression datum with thousands of di-mensions, the running time of the algorithm is much longer. In order to solve the problem, this paper uses PCA-based SVM algorithm and KPCA-based SVM algorithm to make dimension reduction and classification on the datum, which can not only integrate the charac-teristic information of gene datum, but also shorten the calculation time. It compares the classification accuracy rate of the two algo-rithms as the accumulative contribution rate is different, the experimental results show that there is not a fixed law between PCA-SVM classification accuracy rate and accumulative contribution rate, but KPCA-SVM classification accuracy rate will decline or keep un-changeable when cumulative contribution rate declines.
出处
《长春大学学报》
2013年第12期1525-1527,1534,共4页
Journal of Changchun University
