本站所有资源均为高质量资源,各种姿势下载。
本文旨在从因子分析的角度出发,探讨如何解决基因表达谱分析中的问题。针对独立成分分析方法在求解过程中的不稳定性,提出了一种基于选择性独立成分分析的DNA微阵列数据集成分类器。具体来说,我们首先对基因表达水平的重构误差进行了分析,选择了部分重构误差较小的独立成分进行样本重构。接着,我们基于这些重构后的样本,同时训练了多个支持向量机基分类器。最后,我们选择了部分分类正确率较高的基分类器进行最大投票,以得到最终结果。通过在三个常用测试集上的验证,我们证明了本文设计方法的有效性。
In summary, this paper focuses on addressing issues in gene expression profile analysis through the lens of factor analysis. To tackle the instability of independent component analysis in the solving process, we proposed a DNA microarray data integrated classifier based on selective independent component analysis. Specifically, we analyzed the reconstruction error of gene expression levels, selected independent components with smaller reconstruction errors, and reconstructed samples based on these components. Then, we simultaneously trained multiple support vector machine base classifiers using these reconstructed samples. Finally, we selected some base classifiers with relatively higher accuracy rates to perform maximum voting and obtain the final results. By verifying our proposed method on three commonly used test sets, we demonstrate the effectiveness of our approach.