Research Article

Density-Based Geometric One-Class Classifier Combined with Genetic Algorithm

Table 4

Comparison of AUC obtained from and other one-class classifiers.

ClassifierDataset
Iris_SetosaIris_VersicolorIris_VirginicaBiomedLiverBreastVehicleAbaloneSatelliteAverage/ranking
AUC × 100(%) (standard deviation)

(fit 1)100 (0.0)99.1 (0.9)98.7 (1.0)88.5 (0.9)61.3 (1.5)94.8 (0.9)94.4 (0.4)88.9 (0.5)93.5 (1.1)91.0 (0.9)/1st
(fit 2)100 (0.0)98.4 (0.8)98.1 (0.8)89.4 (1.5)59.8 (1.1)96.3 (0.7)95.4 (0.9)87.8 (0.8)91.9 (1.2)90.8 (1.0)/3rd
(fit 3)100 (0.0)98.7 (0.9)97.9 (0.9)90.5 (1.1)60.3 (1.2)95.9 (0.8)94.3 (1.1)87.7 (0.4)92.3 (1.6)90.8 (0.9)/2nd
100 (0.0)
98.5 (0.8)
96.1 (1.1)
89.6 (1.2)
61.6 (1.6)
95.1 (1.2)
92.7 (2.2)
86.7 (0.5)
91.2 (1.1)
90.2 (1.1)/5th
100 (0.0)
 = 2
98.2 (0.8)
 = 3
97.6 (0.9)
 = 1
85.2 (1.2)
 = 4
59.4 (2.3)
 = 6
96.2 (0.7)
 = 2
83.6 (2.1)
 = 7
69.0 (1.6)
 = 34
83.4 (0.9)
 = 8
85.8 (1.3)/9th
Gauss100 (0.0)99.4 (0.4)97.8 (0.6)90.0 (0.4)58.6 (0.5)82.3 (0.2)95.6 (0.1)86.1 (0.2)91.8 (0.0)89.1 (0.3)/6th
MoG100 (0.0)98.8 (0.6)97.1 (0.7)91.2 (0.9)60.7 (0.6)78.5 (1.3)95.0 (0.1)85.3 (0.5)40.0 (0.9)83.0 (0.7)/10th
Naive parzen100 (0.0)98.3 (0.6)95.4 (1.1)93.1 (0.2)61.4 (0.7)96.5 (0.4)86.9 (0.2)85.9 (0.4)97.1 (0.1)90.5 (0.4)/4th
Parzen100 (0.0)99.0 (0.3)96.5 (0.8)90.0 (1.1)59.0 (0.3)72.3 (0.5)47.1 (8.6)86.3 (0.1)53.5 (0.0)78.2 (1.2)/12th
-means100 (0.0)98.4 (1.0)95.4 (0.5)87.8 (1.2)57.8 (1.0)84.6 (3.5)88.9 (0.6)79.2 (1.1)88.6 (18.6)86.7 (3.3)/8th
-NN100 (0.0)98.3 (0.2)97.0 (0.8)89.1 (0.8)59.0 (0.9)69.4 (0.6)70.6 (8.9)86.5 (0.1)55.1 (0.0)80.6 (1.3)/11th
SOM100 (0.0)98.3 (0.5)96.6 (0.3)88.7 (0.8)59.6 (0.7)79.0 (2.3)84.6 (0.8)81.4 (0.3)96.4 (0.2)87.2 (0.6)/7th
SVDD100 (0.0)98.5 (0.1)98.1 (0.8)52.2 (0.3)59.0 (0.9)70.3 (0.5)0.3 (0.0)80.6 (0.1)19.5 (0.0)64.3 (0.3)/14th
LPDD100 (0.0)97.8 (0.5)98.6 (0.4)87.5 (0.9)56.4 (2.6)80.0 (0.5)38.6 (9.7)69.7 (0.1)52.8 (0.0)75.7 (1.7)/13th