Representation Learning Based on Autoencoder and Deep Adaptive Clustering for Image Clustering

Table 2

Clustering performance and comparison, ACC (%) and NMI (%) and ARI (%) on all datasets.

Data sets	Fashion-MNIST			Cifar-10			Cifar-100			STL-10
Metric	NMI	ARI	ACC	NMI	ARI	ACC	NMI	ARI	ACC	NMI	ARI	ACC
K-means [35]	18.55	14.37	22.89	8.71	4.87	22.89	8.39	2.80	12.97	12.45	6.08	19.20
SSC [36]	13.35	14.23	20.81	10.28	8.53	24.67	9.01	2.18	13.60	9.78	4.79	15.88
GLWTDN [37]	20.31	18.72	25.74	23.93	16.89	31.35	10.04	4.76	16.45	24.96	16.10	30.30
CILR [15]	25.68	20.71	30.10	11.05	6.12	25.83	9.43	4.53	15.54	13.98	6.52	24.52
GMVAE [16]	25.68	16.07	30.10	28.47	18.37	33.64	13.66	4.92	15.84	23.65	15.72	31.77
CatGAN [27]	26.31	17.11	32.24	26.46	17.57	31.52	12.0	4.53	15.10	21.00	13.90	29.84
JULE [17]	43.79	33.99	47.78	19.23	13.77	27.15	10.26	3.27	13.67	18.15	16.43	27.69
DEC [24]	60.54	58.43	65.25	25.68	16.07	30.10	13.58	4.95	18.52	27.60	18.61	35.90
DAC [20]	65.38	62.41	70.52	43.79	33.99	47.78	21.24	11.00	25.52	40.33	28.64	48.18
Ours	70.66	69.43	75.21	46.39	42.25	52.25	25.46	16.10	30.47	44.31	33.08	52.12

Most of the results are excerpted from [20]. The best and second-best results are marked in bold and underlined, respectively.