A Folded Concave Penalty Regularized Subspace Clustering Method to Integrate Affinity and Clustering

Table 4

Performance on MNIST.

Methods	SSC	LRR	LRSC	KSSC	SSC-OMP	EDSC	S^3C	SSRSC	TLRR	KTRR	Our method
CA (%)
	78.05	64.7	63.1	74.4	73.5	78.7	60.6	79.6	68.4	80.8		81.8
	84.4	67.0	65.3	79.8	79.6	84.8	62.7	81.4	71.7	84.9		85.9
	86.8	70.3	68.0	82.8	84.8	88.4	65.8	83.2	74.5	88.8		90.3
NMI
	0.7284	0.6148	0.6094	0.7034	0.7056	0.7493	0.4333	0.7302	0.6183	0.7500		0.7508
	0.7603	0.6394	0.6339	0.7544	0.7306	0.7684	0.4354	0.7495	0.6443	0.7745		0.7865
	0.7734	0.6549	0.6407	0.7596	0.7385	0.7693	0.4426	0.7528	0.6487	0.7946		0.7905
Purity
	0.7339	0.6594	0.6283	0.7084	0.6948	0.7648	0.5245	0.7257	0.6384	0.7747		0.7806
	0.7644	0.6645	0.6366	0.7585	0.7058	0.7743	0.5433	0.7504	0.6748	0.7947		0.8019
	0.7796	0.6765	0.6444	0.7658	0.7399	0.8085	0.5574	0.7653	0.6737	0.8095		0.8145