Dimensionality Reduction with Sparse Locality for Principal Component Analysis

Algorithm 1

The procedure for solving equation (25).

	Input: matrix ;
	Step 1: Select parameters: , ρ, ɛ;
	Step 2: Optimization
	Initialize:  = 0, N0 = 0, k = 0, ɛ = 0.8.
	While no convergence do
	(1) Update N using
	(2) Update using equation (25);
	(3) Update k = k + 1;
	(4) Check the convergence condition:
	;
	End and
	Output: k-dimensional vector .