MATLAB: Neural Network k fold cross validation

neural network cross validation

I've got the code for Neural network k fold cross validation,but it isnt performing well. Do I need to initialize the network in every iteration?
Is it wrong?
please help!!
clear all;
close all;
clc;
%xlrange=input('Enter the range of data for input','s');
wineInputs=xlsread('wine.xlsx',1);
%xlrange=input('Enter the range of data for target','s');
wineTargets=xlsread('wine.xlsx',2);
%clear xlrange;
inputs = wineInputs;
targets = wineTargets;
k=10;
cvFolds = crossvalind('Kfold', size(targets,2), k);
net = patternnet(10);
for i = 1:k                                  %# for each fold
      testIdx = (cvFolds == i);               %# get indices of test instances
      trainIdx = ~testIdx  ;                  %# get indices training instances
      trInd = find(trainIdx);
      tstInd = find(testIdx);
      net.trainFcn = 'trainscg' ; 
      net.trainParam.epochs = 100;
      net.divideFcn = 'divideind'; 
      net.divideParam.trainInd=trInd;
      net.divideParam.testInd=tstInd;
  % Choose a Performance Function
  net.performFcn = 'mse';  % Mean squared error
  % Train the Network
  [net,tr] = train(net,inputs,targets);
   %# test using test instances
  outputs = net(inputs);
  errors = gsubtract(targets,outputs);
  performance = perform(net,targets,outputs);
  trainTargets = targets .* tr.trainMask{1};
  testTargets = targets  .* tr.testMask{1};
  trainPerformance = perform(net,trainTargets,outputs);
  testPerformance = perform(net,testTargets,outputs);
  test(k)=testPerformance;
  %save net
  figure, plotconfusion(targets,outputs);
  disp('reached here');
end
  accuracy=mean(test)
    % View the Network
    view(net);

Best Answer

1. Yes, the net needs to be reconfigured at the top of the for loop.
2. There is no contingency for obtaining a poor design due to an unfortunate assignment of random initial weights. Two ideas
   a. For each i of i =1:k design multiple nets differing by the assignment of random initial weights. Discard those with poor performance and average the performance of the rest.  
   b. For each k design, keep designing, evaluating and discarding nets until one satisfies an "acceptable" criterion.
Hope this helps.
Thank you for formally accepting my answer
Greg
Related Question