|
- clear;
- clc;
- path = '*';
- addpath(genpath(path));
- dataName = 'wdbc'
- load(['datasets\',dataName,'_Kmatrix'],'KH','Y');
- numclass = length(unique(Y));
- Y(Y<1) = numclass;
- numker = size(KH,3);
- num = size(KH,1);
- KH = kcenter(KH);
- KH = knorm(KH);
- epsionset = [0.05:0.05:0.5];
- respath = [path];
- for ie =1:length(epsionset)
- for iter = 1 : 10
- fprintf('%s: missing_ratio: %d ,iter: %d\n',dataName,epsionset(ie)*100,iter);
- load([path,'incompleteKernelDatasets\',dataName,'\',dataName,'_missingRatio_',num2str(epsionset(ie)),...
- '_iter_',num2str(iter),'.mat'],'S');
- qnorm = 2;
-
- iseedset = [0:19];
- lambdaset9 = 2.^[-15:1:15];
- res_allmean9 = zeros(4,length(lambdaset9),length(iseedset));
- res_allstd9 = zeros(4,length(lambdaset9),length(iseedset));
- Sigma_all9 = zeros(numker,length(lambdaset9),length(iseedset));
- for is = 1:length(iseedset)
- s=RandStream('mt19937ar','Seed',iseedset(is));
- RandStream.setGlobalStream(s);
- tic
- for il=1:length(lambdaset9)
- [H_normalized9,C9,WP9,Sigma_all9(:,il,is),obj9] = OS_LF_IMVC_alg(KH,S,numclass,lambdaset9(il));
- [res_allmean9(:,il,is),res_allstd9(:,il,is)] = myNMIACCV2(H_normalized9,Y,numclass);
- end
- [~,max_idx]=max(mean(res_allmean9(:,:,is)),[],'all','linear');
- res_mean9(is,:) = res_allmean9(:,max_idx,is);
- Sigma99(is,:) = Sigma_all9(:,max_idx,is);
- timecost9(is) = toc;
- end
- res_mean(:,9) = mean(res_mean9);
- res_std(:,9) = std(res_mean9);
- timecost(9) = mean(timecost9);
-
-
- end
- end
|