Spectral Nonlinearly Embedded Clustering Algorithm

Table 2

Performance comparison of clustering accuracy using KM, SC, SEC, KSEC, and ESEC for the in-sample clustering on ten datasets.

Dataset	KM	SC	SEC	KSEC	ESEC
Iris	82.1 ± 1.9	83.3 ± 1.9	68.9 ± 1.5	83.3 ± 1.8	90.4 ± 2.3
Glass	54.9 ± 0.9	58.9 ± 1.3	59.3 ± 1.6	59.8 ± 2.1	72.2 ± 1.6
Wine	57.1 ± 1.8	57.9 ± 1.9	44.9 ± 1.8	60.1 ± 2.4	61.8 ± 1.8
WPBC	63.6 ± 1.9	74.2 ± 1.4	77.2 ± 1.4	77.2 ± 1.3	74.1 ± 1.2
SpectfHeart	61.6 ± 3.1	78.7 ± 3.7	55.4 ± 1.3	79.8 ± 3.9	76.7 ± 2.5
USPS	63.7 ± 1.2	61.3 ± 2.4	65.1 ± 1.7	61.9 ± 2.5	65.9 ± 2.6
Yale	37.8 ± 1.5	35.9 ± 1.7	30.9 ± 3.4	37.3 ± 1.6	40.8 ± 1.9
COIL20	67.0 ± 3.4	76.2 ± 3.3	67.6 ± 1.4	78.8 ± 2.8	77.0 ± 2.4
Isolet	66.6 ± 0.9	65.4 ± 2.6	68.7 ± 1.8	76.4 ± 2.2	70.6 ± 2.6
ORL	59.3 ± 1.5	58.2 ± 2.9	50.9 ± 3.8	61.4 ± 1.9	60.7 ± 2.7