基于PCA的参数化间隔双子支持向量机及其在手写体识别上的应用

PCA Based on Parametric-Margin Twin Support Vector Machines and Its Application on Handwriting Recognition

改进了基于参数化间隔的双子支持向量机算法的预处理过程，在数据预处理阶段使用了主成分分析法对数据进行降维，提出了基于主成分分析的参数化间隔双子支持向量机，从而加快了整个算法的训练速度．公共数据库上的实验结果显示了该算法的优秀分类能力，对高维数据集的降维效果也比较成功．最后，将这种算法应用到手写体数字识别技术上，实验结果显示出该算法较好的分类性能．

The pretreatment process of twin support vector machine based on Parametric-margin has been improved, principal component analysis to reduce the dimensionality of the data is used in the data preprocessing stage. The twin support vector machines based on principal component analysis speeds up the training process. Experiments on public available datasets show excellent classification ability. Dimensionality reduction of the high-dimensional dataset is also more successful. Finally, this algorithm is applied on handwritten numeral recognition. The experimental results show the effective of our algorithm.