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

Table 4

Clustering performance, ACC (%) and NMI (%) and ARI (%), with different number of clusters.

Method	DAC			Ours
Clusters	NMI	ARI	ACC	NMI	ARI	ACC
Cifar-10
5	44.89	35.58	49.35	48.12	42.53	52.33
6	44.35	35.03	48.27	47.82	42.02	51.12
7	44.21	34.69	48.05	47.39	41.79	51.02
8	44.02	34.52	47.93	46.64	41.52	50.51
9	43.92	34.21	47.88	46.52	41.27	50.32
10	43.79	33.99	47.78	46.39	40.25	50.25
Cifar-100
10	22.16	11.89	30.25	26.67	21.72	35.33
12	22.05	11.68	29.58	26.03	21.47	35.10
14	21.97	11.56	28.37	25.81	21.32	34.27
16	21.88	11.35	27.25	25.72	21.04	33.43
18	21.65	11.21	26.54	25.67	20.53	31.25
20	21.24	11.00	25.52	25.46	20.10	30.47
STL-10
5	41.43	29.68	50.30	45.01	34.82	53.35
6	41.22	29.35	50.12	44.96	34.51	52.98
7	41.05	29.06	49.26	44.72	34.03	52.17
8	40.76	28.96	48.37	44.56	33.87	51.55
9	40.55	28.72	48.27	44.33	33.53	51.23
10	40.33	28.64	48.18	44.31	33.08	51.12