基于主成分和支持向量机的降维和模式分类

Dimensionality Reduction and Pattern Classification Method Based on the SVM and PCA

首先讨论主成分分析和支持向量机的基本思想和实现过程,由于主成分分析PCA方法具备降维的功能,而支持向量机SVM方法又具有高分类准确率的优点,尝试将两者结合起来进行模式分类,最终经过实验验证获得成功。采用UCI数据库中的Wine数据库分别对PCA、SVM、主成分和支持向量机结合的模式分类这三种方法进行实验仿真和比较,并取得较为理想的结果。

The basic theory and realization of principal component analysis and supported vector machines are discussed in this paper.A dimensionality reduction and pattern classification method based on the SVM and PCA are proposed because of dimensionality reduction of PCA and high classification accuracy of SVM.The three methods are carried out through simulation experiments with database of UCI,and achieved more satisfactory results.