primalgeneralfeatures.m.svn-base

a function inside machine learning

function [newTrainX, newTestX, u, p] = primalGeneralFeatures(trainX, trainY, testX, T, featureDirection)
% Compute features based on the General Feature Extraction Framework for a 
% set of training examples and label. 
%
% Usage: [newTrainX, newTestX, u, p] = primalGeneralFeatures(trainX, trainY, testX, T, featureDirection)
% Inputs/Outputs:  
%   trainX - an (l x n) matrix whose rows are the training inputs
%   trainY - (l x m) containing the corresponding output vectors
%   testX - an (l x n) matrix whose rows are the test inputs
%   T - the number of iterations to be performed
%   featureDirection - a string specifying a function which computes a
%       projection direction on centered matrices trainX and trainY at each
%       iteration
%
%   newTrainX - the new training data matrix
%   newTestX - the new test data matrix
%   u - the matrix with columns composed of the projection directions
%   p - the matrix with columns composed of the p's
%
% Copyright (C) 2006 Charanpal Dhanjal 

% This library is free software; you can redistribute it and/or
% modify it under the terms of the GNU Lesser General Public
% License as published by the Free Software Foundation; either
% version 2.1 of the License, or (at your option) any later version.
% 
% This library is distributed in the hope that it will be useful,
% but WITHOUT ANY WARRANTY; without even the implied warranty of
% MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
% Lesser General Public License for more details.
% 
% You should have received a copy of the GNU Lesser General Public
% License along with this library; if not, write to the Free Software
% Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301
% USA

if (nargin ~= 5)
    fprintf('%s\n', help('primalGeneralFeatures'));
    error('Incorrect number of inputs - see above usage instructions.');
end

Xj = trainX; 

numExamples = size(trainX, 1); 
numFeatures = size(trainX, 2);

u = ones(numFeatures, T);
t = ones(numExamples, T); 

%Compute the projection directions 
for j=1:T
	u(:, j) = feval(featureDirection, Xj, trainY); 
    
    t(:, j) = Xj*u(:,j);
    p(:,j) = Xj'*t(:, j)/(t(:, j)'*t(:, j));
    
    Xj = Xj - t(:, j)*p(:,j)';  
end

%Compute new features on training and test data 
Z = u/(p'*u); 

newTrainX = trainX*Z; 
newTestX = testX*Z;