第6期 曹苏群，等：最佳鉴别矢量集在无监督模式下的扩展 517 表4采用FCM和K-means分别对经PCA,OSDV和UOSDV降维的Wine数据集进行聚类所得RI值 Table 4 Rand index of FCM and K-means for reduced-dimension Wine dataset extracted by PCA,OSDV PCA OSDV UOSDV Number of features RI with RI with RI with RI with RI with RI with FCM K-means FCM K-means FCM K-means 1 0.711 0.719 0.924 0.931 0.745 0.761 2 0.711 0.719 0.910 0.910 0.754 0.754 4 0.711 0.719 0.896 0.896 0.742 0.744 6 0.711 0.685 0.969 0.961 0.707 0.698 8 0.711 0.719 0.744 0.741 0.704 0.704 10 0.711 0.719 0.715 0.717 0.707 0.707 12 0.711 0.719 0.712 0.719 0.714 0.692 800 0.90 600 0.85 400 0.80 200 0 0.70 .PC&FCM -200-9 0.65 +…OSDV&FCM …b-UOSDV&FCM -400- -e-PCA&K-means 0.60 车08*em -600 0.55 -800 5 10 15 20 25 4000-3000-2000-1000 01000 特征数 (a)RI俏和特征维数出线图 (b)采用PCA降为2维的数据散布图 0.18 -0.02 0.16 -0.04 0.14 -0.06 0.12 .-0.08 0.10 0.10 0.08 0.06 0.0%.17 -0.150.130.110.090.0 0.2 -0.10-0.080.06 -0.04 (c)采用OSDV降为2维的致据散布图 (d)采用UOSDV降为2维的数据做图 图7对Wdc数据集分别使用PCA,OSDV和UOSDV降维后分析结果 Fig.7 Results for the analysis of reduced-dimension Wdbe dataset extracted by PCA,OSDV and UOSDV 表5采用FCM和K-means分别对经PCA,OSDV和UOSDV降维的Wdbc数据集进行聚类所得RI值 Table 5 Rand index of FCM and K-means for reduced-dimension Wdbc dataset extracted by PCA,OSDV and UOSDV PCA OSDV UOSDV Number of RI with features RI with RI with RI with RI with RI with FCM K-means FCM K-means FCM K-means 1 0.750 0.750 0.935 0.932 0.863 0.863