Mercurial > hg > camir-aes2014
view core/magnatagatune/makro_cupaper12_get_significance.m @ 0:e9a9cd732c1e tip
first hg version after svn
author | wolffd |
---|---|
date | Tue, 10 Feb 2015 15:05:51 +0000 |
parents | |
children |
line wrap: on
line source
% makro_get_significance_cupaper %% % ------------------------------ Algorithms Compared id-sampling % --- 'mlr vs euclidean, mlr all average feat, ID-sampling' file1 = 'runlog_0b506247a68167addf97fcb0296650eb_1a58077f1232c33b787b661039df107d_finalresults'; % euclidean average feat - TODO: which sampling is this ? file2 = 'runlog_0b506247a68167addf97fcb0296650eb_1a9788c5e14f30e23ed2a05dbf513c9f_finalresults'; % --- % NOTE: all 4 test-set runs have exactly the same result % --- run1 = -1; run2 = -1; mode = 'join_datasets'; % '', 'avgbase', 'join_datasets' [p, med, avg] = test_generic_significance_signrank(file1,run1,file2,run1,0,mode); fprintf('p = %3.4f, median = %3.4f, avg = %3.4f \n\n',p,med,avg); %% % --- 'SVM vs euclidean, SVM all average feat, ID-sampling' file1 = 'runlog_0b506247a68167addf97fcb0296650eb_1841892e9df07039bbe4c3a55d11026a_finalresults'; % euclidean average feat - TODO: which sampling is this ? file2 = 'runlog_0b506247a68167addf97fcb0296650eb_1a9788c5e14f30e23ed2a05dbf513c9f_finalresults'; [p, med, avg] = test_generic_significance_signrank(file1,run1,file2,run2,0,mode); fprintf('p = %3.4f, median = %3.4f, avg = %3.4f \n\n',p,med,avg); %% % ------------------------------ Algorithms Compared TD-sampling % --- 'mlr unclustered vs euclidean unclustered, mlr all average feat TD-sampling' file1 = 'runlog_0b506247a68167addf97fcb0296650eb_c364cb0803822d55f2940656c44b184d_finalresults'; % euclidean average feat TD-sampling file2 = 'runlog_0b506247a68167addf97fcb0296650eb_72303bfa642aad872665dee7a3b1e28c_finalresults'; [p, med, avg] = test_generic_significance_signrank(file1,run1,file2,run2,0,mode); fprintf('p = %3.4f, median = %3.4f, avg = %3.4f \n\n',p,med,avg); %% % --- 'SVM unclustered vs euclidean unclustered, SVM all average feat TD-sampling' file1 = 'runlog_0b506247a68167addf97fcb0296650eb_3434f0534fa910b26bbf927c65a7fb74_finalresults'; % euclidean average feat TD-sampling file2 = 'runlog_0b506247a68167addf97fcb0296650eb_72303bfa642aad872665dee7a3b1e28c_finalresults'; % --- [p, med, avg] = test_generic_significance_signrank(file1,run1,file2,run2,0,mode); fprintf('p = %3.4f, median = %3.4f, avg = %3.4f \n\n',p,med,avg); %% % ------------------------------ Features Compared 'All 12-dim PCA Features compared' % --- namePCA12 = {'AcousticPCA12', ... 'Slaney08PCA12', ... 'TimbrePCA12', ... 'ChromaPCA12', ... 'AllPCA12', ... 'GenrePCA12'}; % this is the index how the feature types appear in the paper paperidx = [4 3 2 6 1 5]; filePCA12 = {'runlog_a18bd2111694ac59c9ba0a6810121796_1841892e9df07039bbe4c3a55d11026a_finalresults', ... 'runlog_37e47c187886f73ec9a7d8dc24a84a52_1841892e9df07039bbe4c3a55d11026a_finalresults', ... 'runlog_4c6787b403a07f5faf1ec26e891da4fa_1841892e9df07039bbe4c3a55d11026a_finalresults', ... 'runlog_c5566f74e6a0d00b50f5eea05fdacfee_1841892e9df07039bbe4c3a55d11026a_finalresults', ... 'runlog_e2c22696e7af9e7eea1fa1fd10a1f785_1841892e9df07039bbe4c3a55d11026a_finalresults', ... 'runlog_efc6e5e9c56291cd1744092a1c59a293_1841892e9df07039bbe4c3a55d11026a_finalresults'}; namePCA12 = namePCA12(paperidx); filePCA12 = filePCA12(paperidx); p = zeros(numel(namePCA12),numel(namePCA12)); med = zeros(numel(namePCA12),numel(namePCA12)); avg = zeros(numel(namePCA12),numel(namePCA12)); for i=1:numel(name) for j = 1:i-1 [p(i,j), med(i,j), avg(i,j)] = test_generic_significance_signrank(filePCA12{i},-1,filePCA12{j},-1,0,mode); end % p(i,i) = 0.5; end % p = p + p'; imagesc(p); colormap(hot) axis xy set(gca,'XTick',1:numel(namePCA12), 'XTickLabel', namePCA12); set(gca,'YTick',1:numel(namePCA12), 'YTickLabel', namePCA12); matrix2latex(p,'%1.3f') %% 'All 52-dim PCA Features compared' % --- namePCA52 = {'GenrePCA52', ... 'AllPCA52', ... 'TimbrePCA52', ... 'AcousticPCA52', ... 'ChromaPCA52' ... }; % this is the index how the feature types appear in the paper paperidx = [5 3 1 4 2]; filePCA52 = {'runlog_3cbf4759cf58af0728aaab0b5f2660e3_1841892e9df07039bbe4c3a55d11026a_finalresults', ... 'runlog_7d5fafec0dc504215acc8cb7a9202a56_1841892e9df07039bbe4c3a55d11026a_finalresults', ... 'runlog_a3c2c0a5742a42fd54497e69b8f44e8d_1841892e9df07039bbe4c3a55d11026a_finalresults', ... 'runlog_c7164074206998aa184538bedcfdcf2f_1841892e9df07039bbe4c3a55d11026a_finalresults', ... 'runlog_efbf7c8e75ae154c2f192acd08fbdcbc_1841892e9df07039bbe4c3a55d11026a_finalresults' ... }; namePCA52 = namePCA52(paperidx); filePCA52 = filePCA52(paperidx); p = zeros(numel(namePCA52),numel(namePCA52)); med = zeros(numel(namePCA52),numel(namePCA52)); avg = zeros(numel(namePCA52),numel(namePCA52)); for i=1:numel(name) for j = 1:i-1 [p(i,j), med(i,j), avg(i,j)] = test_generic_significance_signrank(... filePCA52{i},run1,filePCA52{j},run2,0,mode); end % p(i,i) = 0.5; end % p = p + p'; imagesc(p); colormap(hot) axis xy set(gca,'XTick',1:numel(namePCA52), 'XTickLabel', namePCA52); set(gca,'YTick',1:numel(namePCA52), 'YTickLabel', namePCA52); matrix2latex(p,'%1.3f') %% 'All full-dim Features compared (SVM)' % --- name = {'4Chroma', ... '1Chroma', ... '4Timbre', ... '1Timbre', ... '1Acoustic', ... 'Genre', ... 'Slaney08', ... '1All', ... '4All', ... '4Acoustic' ... }; % this is the index how the feature types appear in the paper % Features & Chroma(1/4) & Timbre(1/4) & Slaney08 & Genre & Comb. Audio (1/4) & Comb. All(1/4) \\ paperidx = [2 1 4 3 7 6 5 10 8 9]; file = {'runlog_20a2f6a0f20f488e9386ebb8c5026fcf_85c439d9ef3d135936e7645ebc0efe36_finalresults', ... 'runlog_20a2f6a0f20f488e9386ebb8c5026fcf_85c439d9ef3d135936e7645ebc0efe36_finalresults', ... 'runlog_3154f36c34c18f60218c5d3f0c0b5931_85c439d9ef3d135936e7645ebc0efe36_finalresults', ... 'runlog_3154f36c34c18f60218c5d3f0c0b5931_85c439d9ef3d135936e7645ebc0efe36_finalresults', ... 'runlog_31981d48dd0d25564ef3c2b3ca650b3b_1841892e9df07039bbe4c3a55d11026a_finalresults', ... 'runlog_37867d3b5bd4c74b7b548732b80fb947_85c439d9ef3d135936e7645ebc0efe36_finalresults', ... 'runlog_f52d37439805ac4edc70b0432281abc3_85c439d9ef3d135936e7645ebc0efe36_finalresults', ... 'runlog_0b506247a68167addf97fcb0296650eb_1841892e9df07039bbe4c3a55d11026a_finalresults', ... 'runlog_800d97be9ef6274dc3bbe6b9be2406a6_1a58077f1232c33b787b661039df107d_finalresults', ... 'runlog_cf5a61cca09e2a3182b794b70ee1ab91_1841892e9df07039bbe4c3a55d11026a_finalresults' ... }; sets2join = {[1:4],[5:8],[1:4],[5:8],[],[],[],[],[],[]}; name = name(paperidx); file = file(paperidx); p = zeros(numel(name),numel(name)); med = zeros(numel(name),numel(name)); avg = zeros(numel(name),numel(name)); for i=1:numel(name) for j = 1:i-1 [p(i,j), med(i,j), avg(i,j)] = test_generic_significance_signrank(... file{i},sets2join{i},file{j},sets2join{j},0,mode); end % p(i,i) = 0.5; end p = flipud(p); % p = p + p'; imagesc(p); colormap(hot) axis ij set(gca,'XTick',1:numel(name), 'XTickLabel', name); set(gca,'YTick',1:numel(name), 'YTickLabel', fliplr(name)); matrix2latex(p,'%1.3f') %% 'Comparing Feature dimensions and PCA effect on combined all features' % --- % 'SVM all average feat, ID-sampling' allavg = 'runlog_0b506247a68167addf97fcb0296650eb_1841892e9df07039bbe4c3a55d11026a_finalresults'; % 'SVM 4 cluster feat, ID-sampling' allfourcluster = 'runlog_800d97be9ef6274dc3bbe6b9be2406a6_1a58077f1232c33b787b661039df107d_finalresults'; % 'SVM 12dim feat, ID-sampling' pca12 = 'runlog_e2c22696e7af9e7eea1fa1fd10a1f785_1841892e9df07039bbe4c3a55d11026a_finalresults'; % 'SVM 52dim feat, ID-sampling' pca52 = 'runlog_7d5fafec0dc504215acc8cb7a9202a56_1841892e9df07039bbe4c3a55d11026a_finalresults'; % --- % NOTE: all 4 test-set runs have exactly the same result % --- run1 = -1; run2 = -1; mode = 'join_datasets'; % '', 'avgbase', 'join_datasets' 'AllAvg vs Pca12' [p, med, avg] = test_generic_significance_signrank(allavg,run1,pca12,run2,0,mode); fprintf('p = %3.4f, median = %3.4f, avg = %3.4f \n\n',p,med,avg); 'AllAvg vs pca52' [p, med, avg] = test_generic_significance_signrank(allavg,run1,pca52,run2,0,mode); fprintf('p = %3.4f, median = %3.4f, avg = %3.4f \n\n',p,med,avg); 'Pca12 vs pca52' [p, med, avg] = test_generic_significance_signrank(pca12,run1,pca52,run2,0,mode); fprintf('p = %3.4f, median = %3.4f, avg = %3.4f \n\n',p,med,avg); %% % ------------------------------ Algorithms Weighted Training Weighted Performance Compared 'MLR t:w,e:w vs euclidean t:w,e:w' % --- file1 = 'runlog_0b506247a68167addf97fcb0296650eb_3cdcca7596fed97f87b0ec051cb8bba0_finalresults'; % euclidean baseline, file same as the unweighted above file2 = 'runlog_0b506247a68167addf97fcb0296650eb_1a9788c5e14f30e23ed2a05dbf513c9f_finalresults'; % --- % NOTE: using the "weighted" parameter here % --- [p, med, avg] = test_generic_significance_signrank(file1,run1,file2,run2, 1 ,mode); fprintf('p = %3.4f, median = %3.4f, avg = %3.4f \n\n',p,med,avg); %% 'DMLR t:w,e:w vs euclidean t:w,e:w' % --- file1 = 'runlog_0b506247a68167addf97fcb0296650eb_1cc76d534804229cbdec8b20f8b75dba_finalresults'; % euclidean baseline, file same as the unweighted above file2 = 'runlog_0b506247a68167addf97fcb0296650eb_1a9788c5e14f30e23ed2a05dbf513c9f_finalresults'; % --- % NOTE: using the "weighted" parameter here % --- [p, med, avg] = test_generic_significance_signrank(file1,run1,file2,run2, 1 ,mode); fprintf('p = %3.4f, median = %3.4f, avg = %3.4f \n\n',p,med,avg); %% 'SVM t:w,e:w vs euclidean t:w,e:w' % --- file1 = 'runlog_0b506247a68167addf97fcb0296650eb_9bd9ccddb2e4e622e2ba9826466442ba_finalresults'; % euclidean baseline, file same as the unweighted above file2 = 'runlog_0b506247a68167addf97fcb0296650eb_1a9788c5e14f30e23ed2a05dbf513c9f_finalresults'; % --- % NOTE: using the "weighted" parameter here % --- [p, med, avg] = test_generic_significance_signrank(file1,run1,file2,run2, 1 ,mode); fprintf('p = %3.4f, median = %3.4f, avg = %3.4f \n\n',p,med,avg); %% % ------------------------------ Algorithms Weighted Training UNWEIGHTED Performance Compared 'MLR t:w,e:uw vs euclidean t:w,e:uw' % --- file1 = 'runlog_0b506247a68167addf97fcb0296650eb_3cdcca7596fed97f87b0ec051cb8bba0_finalresults'; % euclidean baseline, file same as the unweighted above file2 = 'runlog_0b506247a68167addf97fcb0296650eb_1a9788c5e14f30e23ed2a05dbf513c9f_finalresults'; % --- % NOTE: using the "weighted" parameter here % --- [p, med, avg] = test_generic_significance_signrank(file1,run1,file2,run2, 0 ,mode); fprintf('p = %3.4f, median = %3.4f, avg = %3.4f \n\n',p,med,avg); %% 'DMLR t:w,e:uw vs euclidean t:w,e:uw' % --- file1 = 'runlog_0b506247a68167addf97fcb0296650eb_1cc76d534804229cbdec8b20f8b75dba_finalresults'; % euclidean baseline, file same as the unweighted above file2 = 'runlog_0b506247a68167addf97fcb0296650eb_1a9788c5e14f30e23ed2a05dbf513c9f_finalresults'; % --- % NOTE: using the "weighted" parameter here % --- [p, med, avg] = test_generic_significance_signrank(file1,run1,file2,run2, 0 ,mode); fprintf('p = %3.4f, median = %3.4f, avg = %3.4f \n\n',p,med,avg); %% 'SVM t:w,e:uw vs euclidean t:w,e:uw' % --- file1 = 'runlog_0b506247a68167addf97fcb0296650eb_9bd9ccddb2e4e622e2ba9826466442ba_finalresults'; % euclidean baseline, file same as the unweighted above file2 = 'runlog_0b506247a68167addf97fcb0296650eb_1a9788c5e14f30e23ed2a05dbf513c9f_finalresults'; % --- % NOTE: using the "weighted" parameter here % --- [p, med, avg] = test_generic_significance_signrank(file1,run1,file2,run2, 0 ,mode); fprintf('p = %3.4f, median = %3.4f, avg = %3.4f \n\n',p,med,avg); %% % ------------------------------ Algorithms w / uw Training WEIGHTED Performance Compared 'MLR t:w,e:w vs MLR t:uw,e:w' % --- % MLR t:w,e:w file1 = 'runlog_0b506247a68167addf97fcb0296650eb_3cdcca7596fed97f87b0ec051cb8bba0_finalresults'; % MLR t:uw,e:w file2 = 'runlog_0b506247a68167addf97fcb0296650eb_1a58077f1232c33b787b661039df107d_finalresults'; % --- % NOTE: using the "weighted" parameter here % --- [p, med, avg] = test_generic_significance_signrank(file1,run1,file2,run2, 1 ,mode); fprintf('p = %3.4f, median = %3.4f, avg = %3.4f \n\n',p,med,avg); %% % --- 'SVM t:w,e:w vs SVM t:uw,e:w' % --- % SVM t:w,e:w file1 = 'runlog_0b506247a68167addf97fcb0296650eb_9bd9ccddb2e4e622e2ba9826466442ba_finalresults'; % SVM t:uw,e:w file2 = 'runlog_0b506247a68167addf97fcb0296650eb_1841892e9df07039bbe4c3a55d11026a_finalresults'; % --- % NOTE: using the "weighted" parameter here % --- [p, med, avg] = test_generic_significance_signrank(file1,run1,file2,run2, 1 ,mode); fprintf('p = %3.4f, median = %3.4f, avg = %3.4f \n\n',p,med,avg); %% % ------------------------------ Algorithms w / uw Training WEIGHTED Performance Compared 'DMLR t:w,e:w vs DMLR t:uw,e:w' % --- % DMLR t:w,e:w file1 = 'runlog_0b506247a68167addf97fcb0296650eb_1cc76d534804229cbdec8b20f8b75dba_finalresults'; % DMLR t:uw,e:w file2 = 'runlog_0b506247a68167addf97fcb0296650eb_bf08b16f453683d96ddffc31c6439730_finalresults'; % --- % NOTE: using the "weighted" parameter here % --- [p, med, avg] = test_generic_significance_signrank(file1,run1,file2,run2, 1 ,mode); fprintf('p = %3.4f, median = %3.4f, avg = %3.4f \n\n',p,med,avg);