MATLAB: Classification error at each epoch

neural network classification epochs errors

Hi there, I am performing a classification problem using Neural Network tool. I did the following:
hiddenLayerSize = 10;
net = patternnet(hiddenLayerSize);
net.inputs{1}.processFcns = {'removeconstantrows','mapminmax'};
net.outputs{2}.processFcns = {'removeconstantrows','mapminmax'};
net.divideParam.trainRatio = 70/100; net.divideParam.valRatio = 15/100; net.divideParam.testRatio = 15/100;
net.trainFcn = 'trainscg'; % Scaled conjugate gradient
net.performFcn = 'mse'; % Mean squared error
net.plotFcns = {'plotperform','plottrainstate','ploterrhist', … 'plotregression', 'plotfit'};
I then trained the network
[net,tr] = train(net,inputs,targets);
and got some results. From tr information, I can extract classification hit/miss for each class, as well as mse at each epoch. However, is it possible to extract, at each epoch, percentage of hit/miss for each sub-class?
How mse is estimated at each time step?
Is hit/miss percentage for each class explicitly computed at each time step?
Best Regards, Alessandro

Best Answer

Classification error at each epoch Asked by alessandro on 19 Jan 2016 at 8:45 Latest activity Edited by alessandro on 19 Jan 2016 at 8:45 Hi there, I am performing a classification problem using Neural Network tool. I did the following:
GEH1 = 'I deleted default statements'
 net = patternnet;
 net.performFcn = 'mse'; % Mean squared error
 ..[net,tr] = train(net,inputs,targets);
and got some results. From tr information, I can extract classification hit/miss for each class, as well as mse at each epoch. However, is it possible to extract, at each epoch, percentage of hit/miss for each sub-class?
GEH2 ='YES. The locations of the trn/val/tst subclass indices are in tr'
How mse is estimated at each time step?
GEH3 = 'Isn't it obvious?'
Is hit/miss percentage for each class explicitly computed at each time step?
GEH4 = ' No'
Hope this helps
Thank you for formally accepting my answer
Greg
Related Question