Research Article

Localized Simple Multiple Kernel K-Means Clustering with Matrix-Induced Regularization

Table 3

ACC, NMI, purity, and Rand index data of localized-SimpleMKKM-matrix-induced regularization with nine comparison methods on six benchmark datasets.

DatasetAVG-KKMMKKM [10]LMKKM [12]ONKC [24]MKKM-MR [14]LKAM [22]LF-MVC [25]SimpleMKKM [20]LI-SimpleMKKM [21]Proposed

ACC (%)
FLO1751.3 ± 1.443.6 ± 1.742.7 ± 1.743.4 ± 2.057.7 ± 1.248.9 ± 0.956.7 ± 1.559.4 ± 1.559.1 ± 1.161.3±1.5
FLO10227.1 ± 0.822.4 ± 0.839.5 ± 0.740.2 ± 0.941.4 ± 0.838.4 ± 1.242.5 ± 0.844.0 ± 1.044.1±1.0
PFold29.1 ± 1.427.1 ± 1.022.4 ± 0.735.4 ± 1.534.3 ± 1.634.2 ± 1.633.3 ± 2.034.4 ± 1.935.9 ± 1.539.0±1.4
DIGIT88.8 ± 0.147.2 ± 0.647.2 ± 0.789.5 ± 0.187.4 ± 0.195.0±0.389.2 ± 0.190.3 ± 0.194.6 ± 0.194.9 ± 3.0
Cal-2534.2 ± 1.032.8 ± 0.122.6 ± 0.734.3 ± 1.234.0 ± 1.237.2 ± 1.134.6 ± 1.136.2 ± 1.236.7 ± 0.937.3±1.1
Cal-733.7 ± 0.133.3 ± 0.234.4 ± 0.146.0 ± 4.043.5 ± 3.949.4 ± 1.042.3 ± 2.739.4 ± 1.546.4 ± 0.949.8±0.8

NMI (%)
FLO1749.9 ± 0.944.3 ± 1.343.8 ± 1.142.9 ± 1.356.1 ± 0.748.1 ± 0.654.6 ± 1.057.6 ± 0.157.7 ± 0.558.9±1.0
FLO10246.0 ± 0.542.7 ± 0.256.1 ± 0.456.7 ± 0.556.9 ± 0.354.9 ± 0.458.6 ± 0.560.0 ± 0.460.1±0.4
PFold40.3 ± 1.238.1 ± 0.634.8 ± 0.644.1 ± 0.843.2 ± 1.143.7 ± 1.042.3 ± 1.244.2 ± 1.245.2 ± 1.348.0±0.9
DIGIT80.7 ± 0.248.7 ± 0.748.3 ± 0.281.7 ± 0.179.5 ± 0.189.4 ± 0.181.2 ± 0.283.3 ± 0.190.0 ± 0.190.3±1.6
Cal-2559.7 ± 0.558.6 ± 0.551.9 ± 0.359.6 ± 0.859.3 ± 0.562.0±0.659.5 ± 0.760.7 ± 0.561.4 ± 0.461.5 ± 0.5
Cal-734.9 ± 0.330.0 ± 0.330.7 ± 0.140.0 ± 1.043.1 ± 0.843.3 ± 0.240.0 ± 0.340.5 ± 0.439.2 ± 1.344.9±0.1

Purity (%)
FLO1752.3 ± 1.245.1 ± 1.444.6 ± 1.545.1 ± 1.859.2 ± 1.150.1 ± 0.657.5 ± 1.660.5 ± 1.359.7 ± 0.362.7±1.6
FLO10232.3 ± 0.627.8 ± 0.445.6 ± 0.746.3 ± 0.848.0 ± 0.644.6 ± 0.848.6 ± 0.750.3 ± 0.750.5±0.7
PFold37.3 ± 1.633.7 ± 0.931.1 ± 1.042.0 ± 1.241.2 ± 1.441.6 ± 1.340.6 ± 1.641.4 ± 1.642.5 ± 1.646.2±1.5
DIGIT88.8 ± 0.150.1 ± 0.750.2 ± 0.389.5 ± 0.187.4 ± 0.195.0±0.389.2 ± 0.190.3 ± 0.194.6 ± 0.194.9 ± 2.0
Cal-2536.2 ± 1.034.9 ± 0.124.4 ± 0.636.6 ± 1.136.1 ± 1.039.4±1.136.8 ± 1.038.2 ± 1.139.1 ± 0.839.3 ± 1.1
Cal-779.0 ± 0.276.7 ± 0.274.9 ± 0.181.2 ± 0.982.9 ± 0.483.2 ± 0.281.6 ± 0.383.3 ± 0.380.8 ± 0.783.4±0.1

Rand Index (%)
FLO1732.2 ± 1.326.3 ± 1.320.6 ± 1.135.2 ± 1.539.9 ± 1.331.6 ± 0.844.1 ± 0.441.5 ± 1.140.9 ± 0.842.6±1.4
FLO10215.5 ± 0.512.1 ± 0.524.9 ± 0.525.5 ± 0.627.2 ± 0.625.5 ± 1.028.5 ± 0.829.9 ± 0.830.0±0.8
PFold14.4 ± 1.812.1 ± 0.77.8 ± 0.418.0 ± 1.117.2 ± 1.520.1 ± 1.116.5 ± 2.017.6 ± 1.919.8 ± 1.220.8±1.5
DIGIT77.4 ± 0.231.4 ± 0.631.3 ± 0.281.7 ± 0.381.3 ± 0.190.8±2.378.0 ± 0.180.6 ± 0.288.2 ± 0.188.3 ± 2.9
Cal-2518.5 ± 0.917.3 ± 0.18.3 ± 0.618.6 ± 1.218.4 ± 0.821.5±1.018.9 ± 0.920.4 ± 0.120.1 ± 0.721.1 ± 0.8
Cal-726.5 ± 0.323.5 ± 0.320.3 ± 0.130.9 ± 1.731.5 ± 2.236.6 ± 0.529.3 ± 0.830.3 ± 0.732.2 ± 1.337.1±0.3

Bold indicates better results in comparison with other algorithms.