dualsparsegeneralfeatureslm3test4.m.svn-base

a function inside machine learning

%A script to see if dualSparseGeneralFeaturesLM3 gives the same result as
%dualSparseGeneralFeaturesLM3

rand('state',22); 
dataSet = 'linear-synthetic3';
csvFileName = sprintf('%s.data', dataSet); ;

[X, y, numExamples, numFeatures] = readCsvData(csvFileName);

X = centerData(X);
X = normalise(X);
%[X, y] = permuteData(X, y);

%Make sure the dataset isn't too large 
numExamples = min(1000, numExamples);
[X, y] = sampleData(X, y, numExamples); 

%T = rank(X); 
T = 5; 

algorithm = 'SMA'; 

if strcmp(algorithm, 'SMC')
    params.dualFeatureDirection = 'dualMaxSparseCovarianceLM';
    params.dualSparseMeasureFunction = 'sparseCovariances';
elseif  strcmp(algorithm, 'SMA')
    params.dualFeatureDirection = 'dualMaxSparseAlignmentLM';
    params.dualSparseMeasureFunction = 'sparseAlignments';
else
    error('I give up - invalid algorithm %s', algorithm);
end
params.iterations = T; 
params.kernelFunctionName = 'linearKernel'; 
params.chunkSize = 500;
params.sigma = 0.1;
params.normalise = 0;


[newTrainX, newTestX, subspaceInfo] = dualSparseGeneralFeaturesLM(X, y, X, params);

params.cacheSize = numExamples; 
[newTrainX2, newTestX2, subspaceInfo2] = dualSparseGeneralFeaturesLM3(X, y, X, params);

subspaceInfo.b
subspaceInfo2.b

norm(newTrainX - newTrainX2)
norm(newTestX - newTestX2)