One-Class Classification with Extreme Learning Machine

Table 2

The value of measure with standard deviations (in parentheses) for a number of one-class classifiers. Twenty trials have been conducted for each dataset.

Dataset	SPECTF heart	Arrhythmia	Sonar	Liver
Classifier
Naive Parzen	41.7 (4.2)	61.8 (1.1)	46.8 (2.2)	41.5 (0.9)
Parzen	39.3 (1.7)	63.7 (1.2)	49.8 (2.9)	40.7 (1.4)
-means	38.3 (4.7)	63.7 (1.7)	53.2 (3.2)	41.7 (1.4)
1-NN	31.8 (2.6)	59.2 (1.5)	60.4 (2.2)	41.3 (1.3)
-NN	34.7 (1.2)	62.4 (0.9)	55.3 (1.3)	42.0 (1.2)
Autoencoder	39.3 (3.4)	64.8 (1.6)	50.6 (2.4)	42.2 (1.7)
PCA	NaN^1	26.3 (5.3)	37.2 (8.3)	41.1 (1.3)
MST	33.7 (1.7)	62.4 (0.8)	56.7 (1.8)	42.1 (1.1)
-centers	36.4 (2.9)	62.8 (1.2)	53.3 (2.3)	41.6 (1.3)
SVDD	38.9 (4.7)	60.5 (4.8)	51.2 (5.8)	40.6 (3.1)
MPM	31.1 (8.7)	51.9 (5.0)	44.6 (6.3)	40.7 (2.0)
LPDD	38.3 (3.9)	63.8 (2.0)	52.2 (4.2)	40.7 (1.6)
SVM	38.1 (6.4)	63.4 (1.9)	53.6 (3.1)	40.5 (2.4)
ELM	42.6 (1.8)	63.6 (1.6)	54.2 (3.5)	43.0 (1.6)

None of target data is recalled.