The paper analyses the two methods in detail and deduces the relation of PCA to SVD when they are used in dimension reduction.

• 在模式识别的过程中,进行特征的选择有两个目标:丢弃一些对分类贡献不大的特征;