cifar10 knn to much accuracy ?

1 view (last 30 days)
michael scheinfeild
michael scheinfeild on 7 Nov 2017
i classify cifar10 for first group with knn and receive 100 percent accuracy i think it should happen
imgSetTrain=imageSet(fullfile(pathcifar,'cifar10Train'), 'recursive');
imgSetTest=imageSet(fullfile(pathcifar,'cifar10Test'), 'recursive');
classNames = { 'airplane' , 'automobile' , 'bird',...
'cat' , 'deer' , 'dog' , 'frog',...
'horse' , 'ship' , 'truck'};
ActDet =[];
knn=5;
for( k=1)%length(imgSetTest))
disp(k)
curSet=imgSetTest(k);
detectClass=zeros(curSet.Count,1);
for(ji=1:curSet.Count)
imgcur=gpuArray(read(curSet,ji));
% run on train
AllDist=[]; AllGroups=[];
for( kr=1:length(imgSetTrain))
kr
curSetTrain=imgSetTest(kr);
for(jir=1:curSetTrain.Count)
imgTrain=gpuArray(read(curSetTrain,jir));
AllGroups=[AllGroups;kr];
distImg=abs(imgcur-imgTrain);
distImg = sum(distImg(:))/numel(distImg);
AllDist=[AllDist ;distImg];
end
end
[dataSel indexSel]=sort(AllDist) ;
selGroups = AllGroups(indexSel(1:knn));
[js jsh]=hist(selGroups,[1:10]);
[kjs kjsi]=max(js);
detectClass(ji)=kjsi;
ActDet = [ActDet; k detectClass(ji)]
end
end

Answers (0)

Categories

Find more on Statistics and Machine Learning Toolbox in Help Center and File Exchange

Products

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!