Table 6. Results of the proposed SVO & SVOX methods on the USPS dataset with different values of k.

k	Reduced (ratio)		Training Speedup		Testing Speedup		Accuracy		Number of SVs		Jaccard similarity
Whole dataset	1.00		1.00		1.00		0.922		790		1.00
	SVO	SVOX	SVO	SVOX	SVO	SVOX	SVO	SVOX	SVO	SVOX	SVO	SVOX
1	0.12	0.08	12.22	40.87	1.20	4.39	0.884	0.917	629	164	0.39	0.11
2	0.13	0.10	10.51	34.48	1.13	4.18	0.905	0.918	679	176	0.40	0.12
3	0.15	0.11	9.33	31.47	1.21	4.05	0.913	0.919	690	189	0.40	0.13
4	0.17	0.13	8.25	28.34	1.05	3.88	0.917	0.919	720	190	0.41	0.13
5	0.18	0.14	7.76	27.24	1.05	4.01	0.918	0.919	725	183	0.41	0.13
10	0.25	0.21	5.26	18.01	1.08	3.83	0.917	0.920	718	194	0.46	0.14
15	0.31	0.26	4.22	14.86	1.02	3.68	0.919	0.920	724	202	0.49	0.15
20	0.36	0.31	3.59	12.86	0.99	3.77	0.920	0.920	741	202	0.50	0.15
30	0.44	0.39	2.87	10.60	1.01	3.75	0.921	0.920	746	191	0.53	0.15
40	0.51	0.46	2.41	9.06	0.98	3.66	0.920	0.921	743	190	0.57	0.16
50	0.57	0.51	2.10	7.66	1.00	3.87	0.921	0.921	750	182	0.58	0.15

Each row compares performance metrics of the methods for a different value (1 to 50) of the parameter k of the SVO & SVOX methods. The first row represents the metrics measured after training on the whole dataset. “Reduced (ratio)” is the ratio of the size of the resulting reduced dataset to the size of the whole dataset.