Linear Discriminative Learning for Image Classification

Table 3

Image classification results on MNIST dataset. The parameter value is fixed for all methods. Eigen and random projection is used to get different feature spaces.

Dimension		150	200	250	300	350
Eigenfaces	CRC	75.71%	74%	72.57%	70.42%	70.42%
	SRC	74.85%	73.28%	70.57%	70.28%	66.85%
	ALWCR	77.28%	75.14%	74%	71.71%	74.05%
Randomfaces	CRC	73%	71.14%	70.28%	68.28%	66%
	SRC	70%	68.42%	69.28%	63.71%	59.14%
	ALWCR	75.71%	73.28%	72.28%	69%	67.14%