wolffd@0: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0: % Experiment code templat                                                 %
wolffd@0: % Project: sub-euclidean distance for music similarity                    %
wolffd@0: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0: %% Load features
wolffd@0: feature_file = 'rel_music_raw_features.mat';
wolffd@0: vars = whos('-file', feature_file);
wolffd@0: A = load(feature_file,vars(1).name,vars(2).name,vars(3).name,vars(4).name);
wolffd@0: raw_features = A.(vars(1).name);
wolffd@0: indices      = A.(vars(2).name);
wolffd@0: tst_inx      = A.(vars(3).name);
wolffd@0: trn_inx      = A.(vars(4).name);
wolffd@0: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0: % Define directory to save parameters & results
wolffd@0: % dir    = '/home/funzi/Documents/';
wolffd@0: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0: dmr    = [0 5 10 20 30 50];    % dimension reduction by PCA
wolffd@0: ws     = [0 5 10 20 30 50 70]; % window size
wolffd@0: % parameters of rbm (if it is used for extraction)
wolffd@0: hidNum = [30 50 100 500];
wolffd@0: lr_1   = [0.05 0.1 0.5];
wolffd@0: lr_2   = [0.1 0.5 0.7];
wolffd@0: mmt    = [0.02 0.05 0.1];
wolffd@0: cost   = [0.00002 0.01 0.1];
wolffd@0: 
wolffd@0: %% Select parameters (if grid-search is not applied)
wolffd@0: di  = 1;
wolffd@0: wi  = 1;
wolffd@0: hi  = 1;
wolffd@0: l1i = 1;
wolffd@0: l2i = 1;
wolffd@0: mi  = 1;
wolffd@0: ci  = 1;
wolffd@0: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0: % If grid search is define
wolffd@0: % log_file = strcat(dir,'exp_.mat');
wolffd@0: % inx = resume_from_grid(log_file,8);
wolffd@0: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0: %% Feature extraction
wolffd@0: EXT_TYPE = 2;
wolffd@0: switch (EXT_TYPE)
wolffd@0:     case 1  % Using PCA
wolffd@0:         assert(~exist('OCTAVE_VERSION'),'This script cannot run in octave');
wolffd@0:         coeff = princomp(raw_features);
wolffd@0:         coeff = coeff(:,1:end-dmr(di));  % Change value of dmr(di) to reduce the dimensionality
wolffd@0:         features = raw_features*coeff;
wolffd@0:           % normalizing
wolffd@0:         mm = minmax(features')';
wolffd@0:         inn= (find(mm(1,:)~=mm(2,:)));
wolffd@0:         mm = mm(:,inn);
wolffd@0:         features = features(:,inn);
wolffd@0:         features = (features-repmat(mm(1,:),size(features,1),1))./(repmat(mm(2,:),size(features,1),1)-repmat(mm(1,:),size(features,1),1));
wolffd@0:     case 2  % Using rbm
wolffd@0:         conf.hidNum = hidNum(hi);
wolffd@0:         conf.eNum   = 100;
wolffd@0:         conf.sNum   = size(raw_features,1);
wolffd@0:         conf.bNum   = 1;
wolffd@0:         conf.gNum   = 1;
wolffd@0:         conf.params = [lr_1(l1i) lr_2(l2i) mmt(mi) cost(ci)];
wolffd@0:         conf.N    = 50;
wolffd@0:         conf.MAX_INC = 10;
wolffd@0:         W1 = zeros(0,0);
wolffd@0:         [W1 vB1 hB1] = training_rbm_(conf,W1,raw_features);
wolffd@0:         features = raw_features*W1 + repmat(hB1,conf.sNum,1);
wolffd@0: end
wolffd@0: 
wolffd@0: %% Sub-euclidean computation
wolffd@0: num_case = size(trn_inx,1);
wolffd@0: trnd_12 = cell(1,num_case);
wolffd@0: trnd_13 = cell(1,num_case);
wolffd@0: tstd_12 = cell(1,num_case);
wolffd@0: tstd_13 = cell(1,num_case);
wolffd@0: 
wolffd@0: w = ws(wi);
wolffd@0: 
wolffd@0: % w = subspace window size
wolffd@0: if w == 0 % trnd_12 = d(a,b) , trnd_13= d(a,c) 
wolffd@0:   for i = 1:num_case % over all cross-validation folds (num_case)
wolffd@0:     [trnd_12{i} trnd_13{i}] = simple_dist(trn_inx{i},features,indices);
wolffd@0:     [tstd_12{i} tstd_13{i}] = simple_dist(tst_inx{i},features,indices);
wolffd@0:   end
wolffd@0: else
wolffd@0:   for i = 1:num_case % for w > 1
wolffd@0:     [trnd_12{i} trnd_13{i}] = conv_euclidean_dist(trn_inx{i},features,indices,w,1);    %% normalize is better than no normalize  
wolffd@0:     [tstd_12{i} tstd_13{i}] = conv_euclidean_dist(tst_inx{i},features,indices,w,1);
wolffd@0:   end
wolffd@0: end
wolffd@0: %% Data preparation
wolffd@0: trn_dat1 = cell(1,num_case);
wolffd@0: trn_dat2 = cell(1,num_case);
wolffd@0: tst_dat1 = cell(1,num_case);
wolffd@0: tst_dat2 = cell(1,num_case);
wolffd@0: 
wolffd@0: for i=1:num_case 
wolffd@0:  %=> Compute hypothesis
wolffd@0:  trn_dat1{i}  = trnd_13{i} - trnd_12{i};
wolffd@0:  trn_dat2{i}  = trnd_12{i} - trnd_13{i};
wolffd@0:  tst_dat1{i}   = tstd_13{i} - tstd_12{i};
wolffd@0:  tst_dat2{i}   = tstd_12{i} - tstd_13{i};
wolffd@0:  
wolffd@0:  % => Normalize using logistic (lost the range)
wolffd@0: %  trn_dat1{i} = logistic(trn_dat1{i});
wolffd@0: %  trn_dat2{i} = logistic(trn_dat2{i});
wolffd@0: %  tst_dat1{i} = logistic(tst_dat1{i});
wolffd@0: %  tst_dat2{i} = logistic(tst_dat2{i});
wolffd@0:  
wolffd@0:  %=> Normalize using min-max
wolffd@0: %  mm = minmax(trn_dat1{i}')';
wolffd@0: %  inn= find(mm(1,:)~=mm(2,:));
wolffd@0: %  mm = mm(:,inn);    
wolffd@0: %  trn_dat1{i} =
wolffd@0: %  (trn_dat1{i}(:,inn)-repmat(mm(1,:),size(trn_dat1{i},1),1))./repmat(mm(2,:)-mm(1,:),size(trn_dat1{i},1),1);
wolffd@0: %  mm = minmax(tst_dat1{i}')';
wolffd@0: %  inn= find(mm(1,:)~=mm(2,:));
wolffd@0: %  mm = mm(:,inn);    
wolffd@0: %  tst_dat1{i} = (tst_dat1{i}(:,inn)-repmat(mm(1,:),size(tst_dat1{i},1),1))./repmat(mm(2,:)-mm(1,:),size(tst_dat1{i},1),1);
wolffd@0: %  
wolffd@0: %  mm = minmax(trn_dat2{i}')';
wolffd@0: %  inn= find(mm(1,:)~=mm(2,:));
wolffd@0: %  mm = mm(:,inn);    
wolffd@0: %  trn_dat2{i} = (trn_dat2{i}(:,inn)-repmat(mm(1,:),size(trn_dat2{i},1),1))./repmat(mm(2,:)-mm(1,:),size(trn_dat2{i},1),1);
wolffd@0: 
wolffd@0: %  mm = minmax(tst_dat2{i}')';
wolffd@0: %  inn= find(mm(1,:)~=mm(2,:));
wolffd@0: %  mm = mm(:,inn);    
wolffd@0: %  tst_dat2{i} = (tst_dat2{i}(:,inn)-repmat(mm(1,:),size(tst_dat2{i},1),1))./repmat(mm(2,:)-mm(1,:),size(tst_dat2{i},1),1);
wolffd@0: 
wolffd@0:  % => normalize from [-1 1] to [0 1]
wolffd@0:  trn_dat1{i} = (trn_dat1{i}+1)/2;
wolffd@0:  trn_dat2{i} = (trn_dat2{i}+1)/2;
wolffd@0:  tst_dat1{i} = (tst_dat1{i}+1)/2;
wolffd@0:  tst_dat2{i} = (tst_dat2{i}+1)/2;
wolffd@0: end
wolffd@0: correct = 0;   % correct rate
wolffd@0: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0: %% CODE HERE                               %%
wolffd@0: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0: 
wolffd@0: fprintf('Correct = %f\n',correct);
wolffd@0: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0: % Using the logging function to save paramters
wolffd@0: % and the result for plotting or in grid search
wolffd@0: % logging(log_file,[i1 i2 i3 i4 i5 correct]);
wolffd@0: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
wolffd@0: 
wolffd@0: clear;