Table 5.

Quantitative clustering (NMI%±std%) of different methods on six datasets.

Method	Dataset
	COIL20	GEORGIA	MNIST	ORL	PIE	USPS
NMF	70.82±2.06	60.03±1.13	45.71±2.34	82.77±1.79	85.33±1.39	38.08±2.30
GNMF	82.92±2.58	60.94±1.19	$\underset{\_}{52.02\pm 4.58}$	84.57±1.62	84.75±1.80	46.08±3.56
HNMF	87.81±1.52	61.32±1.02	49.92±4.55	84.98±1.52	84.69±1.45	46.18±2.97
HSNMF	87.65±1.60	57.74±1.04	49.26±4.52	61.52±1.09	90.56±0.67	37.08±1.69
SHNMF	72.40±2.37	60.02±1.11	45.66±2.48	83.04±1.82	85.19±1.80	38.30±2.27
HGNTR	88.24±1.43	56.07±1.67	49.42±0.58	82.53±1.64	82.81±1.81	52.33±2.30
LraHGNTR	88.15±1.29	$\underset{\_}{62.47\pm 0.91}$	49.57±4.82	82.18±1.51	75.84±2.23	50.16±2.19
HyperNTF	$\underset{\_}{88.71\pm 7.12}$	60.78±0.95	48.40±4.18	$\underset{\_}{85.27\pm 0.87}$	79.19±1.24	$\mathbf{54.07}\pm \mathbf{2.49}$
TriD	66.60±2.42	52.10±1.24	38.35±3.39	73.84±2.13	$\underset{\_}{90.98\pm 1.69}$	45.82±4.31
HNTriD	$\mathbf{91.04}\pm \mathbf{0.83}$	$\mathbf{65.29}\pm \mathbf{1.35}$	$\mathbf{53.22}\pm \mathbf{4.23}$	$\mathbf{86.44}\pm \mathbf{1.22}$	$\mathbf{94.12}\pm \mathbf{1.05}$	$\underset{\_}{53.95\pm 2.42}$

Significant values are bold.