A Novel Support Vector Machine with Globality-Locality Preserving

Table 2

The effects of trade-off parameter 𝜆 on classification performance; let 𝜎₁=𝜎₂=2.

𝜆	0.2	0.4	0.8	1	1.2	1.4	1.6	2
𝐶=0.1	0.8048	0.8048	0.7849	0.8008	0.8088	0.7928	0.7928	0.8008
𝑘1	40	30	30	10	50	20	90	10
𝐶=1	0.9441	0.7849	0.8167	0.8127	0.7968	0.7968	0.8008	0.8008
𝑘1	10	10	20	20	10	30	10	10
𝐶=10	0.9203	0.8964	0.9004	0.9203	0.8845	0.8964	0.8884	0.8645
𝑘1	30	30	10	20	10	10	10	10
𝐶=100	0.9283	0.9442	0.9243	0.9482	0.9203	0.9323	0.9283	0.9163
𝑘1	50	20	20	20	30	20	20	10