2-Class Problem with patternnet

Hi, I have a problem using the NN toolbox a neural network shall be trained to recognize a two class problem. I used the default settings ( dividerand , 10 hidden neurons, divide radio 0.7, 0.15, 0.15) and my input is a 9xn matrix and my target is a 2xn matrix ([1; 0]for class one and [0; 1] for class two for each sample), where n=1012. the ratio of the classes are about 50:50 .this is the confusion matrix

This is the code that i used :

 rng('default');
x = patientInputs;
t = patientTargets ;
inputs=mapminmax(x);
targets=t;
size(inputs);
 trainFcn = 'trainscg';  % Scaled conjugate gradient backpropagation.
% Create a Pattern Recognition Network
hiddenLayerSize =10;
net = patternnet(hiddenLayerSize);
net.inputs{1}.processFcns = {'removeconstantrows','mapminmax'};
net.outputs{2}.processFcns = {'removeconstantrows','mapminmax'};
net.divideFcn = 'dividerand';  % Divide data randomly
net.divideMode = 'sample';  % Divide up every sample
net.divideParam.trainRatio = 70/100;
net.divideParam.valRatio = 15/100;
net.divideParam.testRatio = 15/100;
net.performFcn = 'mse';  % Cross-Entropy

 % Choose Plot Functions
 % For a list of all plot functions type: help nnplot
net.plotFcns = {'plotperform','plottrainstate','ploterrhist', ...
    'plotconfusion', 'plotroc'};

   net.trainParam.max_fail =55;
     net.trainParam.min_grad=1e-10;
     net.trainParam.show=10;
     net.trainParam.lr=0.01;
     net.trainParam.epochs=90;
     net.trainParam.goal=0.001;
  % Train the Network

[net,tr] = train(net,inputs,targets);

 y = net(inputs);
e = gsubtract(targets,y);
performance = perform(net,targets,y)
tind = vec2ind(targets);
yind = vec2ind(y);
percentErrors = sum(tind ~= yind)/numel(tind);

% Recalculate Training, Validation and Test Performance
trainTargets = t .* tr.trainMask{1};
valTargets = t .* tr.valMask{1};
testTargets = t .* tr.testMask{1};
trainPerformance = perform(net,trainTargets,y)
valPerformance = perform(net,valTargets,y)
testPerformance = perform(net,testTargets,y)

 % View the Network
view(net)

Can anyone tell me how to solve this problem and please go easy on me because newbie in matla and neural network .

Thanks

Answer 1

[Greg Heath]

In order to insure stability with respect to changes in operating conditions I recommend

 MINIMIZING THE NUMBER OF HIDDEN NODES

 subject to the normalized mean-square-error constraint

 NMSE = mse(error)/mean(var(target',1) < 0.01   % i.e., Rsquare >= 0.99

Although I derived this for regression, it works extremely well for classificaton.

I have zillions of posted examples in both the NEWSGROUP and ANSWERS. Try searching using

          greg patternnet

Hope this helps.

Thank you for formally accepting my answer

Greg

[afef]

when i tried to minimize the number of hiddrn nodes less than 10 i got bad performance. Conerning the use of NMSE sorry i don't understand why should i use it ? when i was looking for your examples in NEWSGROUP i tried to modify my coe to be like this

 x = patientInputs;
t = patientTargets;
 N=1012
 I=9
 O=2
[ I N ] = size(x) 
 [ O N ] = size(t) 
 Ntrn   = N-2*round(0.15*N)  %   Ntrn=708
 Ntrneq = Ntrn*O        %   Ntrneq=1416    
    %For a robust design desire Ntrneq >> Nw or
   H=10
    Hub = -1+ceil( (Ntrneq-O) / (I+O+1)) %   Hub =117
    Nw = (I+1)*H+(H+1)*O       % Number of unknown weights =122    
   %H << Hub = -1+ceil( (Ntrneq-O) / (I+O+1)) 
 Ntrials = 10
 rng(0)
 j=0
 for h =round([Hub/10, Hub/2, Hub])
    j  = j+1
    h  = h   %12
    Nw            = (I+1)*h+(h+1)*O  %  Nw=146 
    Ndof          = Ntrneq-Nw       %  Ndof=1270
    net           = patternnet(h);
    net.divideFcn = 'dividerand';           % 'dividetrain'
    for i = 1:Ntrials
        net                          = configure(net,x,t);
        [ net tr outputs regerrors ] = train(net,x,t);
        assignedclasses              = vec2ind(outputs);
        classerr                     = assignedclasses~=trueclasses;
        Nerr(i,j)                    = sum(classerr);
        % FrErr                      = Fraction of Errors (Nerr/N) 
        [FrErr(i,j),CM,IND,ROC]  = confusion(t,outputs);
        FN(i,j) = mean(ROC(:,1));    % Fraction of False Negatives
        TN(i,j) = mean(ROC(:,2)) ;   % Fraction of True Negatives
        TP(i,j) = mean(ROC(:,3));    % Fraction of True Positives
    end
 end
 PctErr=100*Nerr/N

But i got this error message Undefined function or variable 'trueclasses'. and the confusion matrix show a bad accuracy so please tell me what should i do?