第6期 曹苏群，等：最佳鉴别矢量集在无监督模式下的扩展 ·515· 表2采用FCM和K-means分别对经PCA,OSDV和UOSDV降维的Inis数据集进行聚类所得RI值 Table 2 Rand index of FCM and K-means for reduced-dimension Iris dataset extracted by PCA,OSDV and UOSDV PCA OSDV UOSDV Number of RI with RI with RI with features RI with RI with RI with FCM K-means FCM K-means FCM K-means 1 0.950 0.950 0.982 0.974 0.982 0.982 2 0.942 0.950 0.966 0.957 0.957 0.957 3 0.950 0.950 0.950 0.950 0.950 0.950 0.99 0.6 0.981 0.4 0.97 0.96 0.2 0.95 0 0.94 0.2 0.93 PC&FCM -0.4 0.92 0.91 OSDV&K eans 0.6 0.9 1.0 1.5 2.5 3.0 特征数 -0.8.510-0.5 90.51015 (a)RI但I符征维数曲线图 (b)采用PCA降为2维的数据散布图 0.8 0.8r 0.6 0.6 4 0.4 0.4 0.2 0.2 0 0 0.2 -0.2 0.4 -0.4 -0.6 0.6 0.8 1.5 -1.0 -0.5 0 0.51.0 0.810 -0.5 0 r05 1.01.5 (c)采用OSDV降为2维的数据散布图 (d)采用JOSDV降为2维的数据散布图 图4对Iis数据集分别使用PCA,OSDV和UOSDV降维后分析结果 Fig.4 Results for the analysis of reduced-dimension Iris dataset extracted by PCA,OSDV and UOSDV 表3采用FCM和K-means分别对经PCA,OSDV和UOSDV降维的Pima_indians_.diabetes数据集进行聚类所得RI值 Table 3 Rand index of FCM and K-means for reduced-dimension Pima_indians diabetes dataset extracted by PCA,OSDV and UOSDV PCA OSDV UOSDV Number of RI with RI with RI with RI with RI with RI with features FCM K-means FCM K-means FCM K-means 1 0.549 0.551 0.643 0.643 0.553 0.555 2 0.550 0.551 0.626 0.625 0.549 0.552 3 0.550 0.551 0.625 0.625 0.551 0.552 4 0.550 0.551 0.616 0.617 0.554 0.553 5 0.550 0.551 0.606 0.617 0.552 0.556 6 0.550 0.551 0.591 0.596 0.556 0.553 1 0.550 0.551 0.551 0.552 0.549 0.549