基于主成分分析的SMO文本分类

利用SMO进行文本分类的核心问题是特征的选择问题,特征选择涉及到哪些特征和选择的特征维度问题。针对以上问题,介绍一种基于主成分分析和信息增益相结合的数据集样本降维的方法,并在此基础上对序贯最小优化算法进行改进,提出降维序贯最小优化(P-SOM)算法。P-SMO算法去掉了冗余维。实验结果证明,该方法提高SMO算法的性能,缩短支持向量机的训练时间,提高支持向量机的分类精度。

SMO Classification Text Based on PCA

Using SMO for text classification is a core problem of selection,and the characteristics selection involves which characteristic to choose and dimension problem of the characteristics which chosen.According to these problems,proposes a data sets of sample dimension reduction method based on the combination of principal component analysis and information gain,and on this basis,improves the SMO algorithm and proposes the P-SOM which can remove the redundant.The experimental result shows that it can improve the performance of SMO and shorten the training time of SVM,improve the classification accuracy of SVM.