An Empirical Study of Greedy Kernel Fisher Discriminants

Algorithm 1

Stagewise Greedy Kernel Fisher Discriminant Analysis.

Input: Kernel , training labels , sparsity parameter , number of bases to pick at each iteration .

(1) calculate matrix

(2) initialise

(3) for   to   do

(4)  for   to   do

(5)     (optimisation criterion)

(6)  end for

(7)  Deflate kernel matrix

(8)  calculate the projection   where is the Cholesky decomposition of and

(9) end for

(10) train Fisher Discriminant Analysis using (1) in this new projected space to find a sparse weight vector and make

  predictions using (8)

Output: final set , (sparse) weight vector , bias term