wolffd@0: % ---
wolffd@0: % class cvpartition_trunctrain
wolffd@0: % NOTE: this is a fake cvpartition double for 
wolffd@0: % using cvpartitions in truncated-training size experiments
wolffd@0: % ---
wolffd@0: classdef cvpartition_trunctrain
wolffd@0:   
wolffd@0: properties (Hidden)
wolffd@0: 
wolffd@0:     mtest;
wolffd@0:     mtraining;
wolffd@0: end
wolffd@0: properties
wolffd@0:     N;
wolffd@0:     NumTestSets;
wolffd@0:     TrainSize;
wolffd@0:     TestSize;
wolffd@0: end
wolffd@0:     
wolffd@0:         
wolffd@0: methods
wolffd@0:     
wolffd@0: % ---
wolffd@0: % constuctor: directly calculates the truncated testset
wolffd@0: % ---
wolffd@0: function P = cvpartition_trunctrain(Pin, perctrain)
wolffd@0:   
wolffd@0:     P.N = Pin.N;
wolffd@0:     P.NumTestSets = Pin.NumTestSets;
wolffd@0:     
wolffd@0:     for i = 1:Pin.NumTestSets
wolffd@0: 
wolffd@0:         % copy testing data
wolffd@0:         P.TestSize(i) = Pin.TestSize(i);
wolffd@0:         P.mtest{i} = Pin.test(i);
wolffd@0: 
wolffd@0:         % calculate new training size
wolffd@0:         P.TrainSize(i) = ceil(perctrain * Pin.TrainSize(i));
wolffd@0: 
wolffd@0:         % get actual training indices
wolffd@0:         idx = find(Pin.training(i));
wolffd@0:     
wolffd@0:         % ---
wolffd@0:         % TODO: save the permutation in a global variable,
wolffd@0:         % tomake the same smaller set available
wolffd@0:         % for all further experiments.
wolffd@0:         % moreover, it would be great if the smaller training sets
wolffd@0:         % are subsets of the bigger ones
wolffd@0:         % ---
wolffd@0:         tokeep = randperm(numel(idx));
wolffd@0:         tokeep = tokeep(1:P.TrainSize(i));
wolffd@0:         
wolffd@0:         % get indices to keep
wolffd@0:         idx = idx(tokeep);
wolffd@0: 
wolffd@0:         % build truncated training set
wolffd@0:         P.mtraining{i} = false(P.N, 1);
wolffd@0:         P.mtraining{i}(idx) = true;
wolffd@0:     end
wolffd@0: end   
wolffd@0: 
wolffd@0: function out = test(P, i)
wolffd@0: 
wolffd@0:     out = P.mtest{i};
wolffd@0: end
wolffd@0: 
wolffd@0: function out = training(P, i)
wolffd@0: 
wolffd@0:     out = P.mtraining{i};
wolffd@0: end
wolffd@0: 
wolffd@0: end  
wolffd@0: end