Maximum Neighborhood Margin Discriminant Projection for Classification

Table 4

The maximal average recognition rates (%) of each method on Musk with the corresponding standard deviations (stds) and values of dimension in the parentheses.

Methods	𝑙=50	𝑙=80	𝑙=110	𝑙=140
PCA	77.82±2.85	79.21±2.36	81.05±2.47	83.72±3.2
	(33)	(40)	(42)	(19)
LDA	77.87±2.72	79.59±3.08	81.25±2.25	83.47±3.45
	(35)	(38)	(40)	(43)
LPP	77.66±2.88	80.28±2.45	81.45±2.72	85.56±3.17
	(33)	(24)	(32)	(25)
UDP	77.63±3.37	79.81±2.63	80.94±2.31	84.18±3.83
	(33)	(38)	(42)	(24)
MNMDP	78.03 ± 3.50	81.46 ± 2.87	85.47 ± 2.25	87.91 ± 1.74
	(11)	(18)	(8)	(7)