Help with Multilayer Perceptron algorithm code.

I'm trying to write MATLAB code for Multilayer Perceptron. My dataset is a bunch of images and I'm trying to use MLP for segmenting some objects. I'm giving the extracted features as input. Presently, the network has just one hidden layer and I have tried varying number of nodes in the hidden layer. I tried varying my learning rate from 0.1 to 0.9. The error is oscillating between 1000 and 3000. I'm not sure where the mistake is. Can someone verify my code?

function model = MLP_train(fea,OV,rate,NoIter,hidden1,visualize)
%
% INPUTS:
%       fea     -> Input feature vector
%       OV      -> Target Vector
%       rate    -> Learning rate
%       NoIter  -> No of iterations/epochs
%       hidden1 -> No of hidden layers
% OUTPUT:
%       model   -> learned model

[m,n] = size(fea);

%Adding bias
bias = ones(m,1);
fea = [fea bias];

NoOfInputNodes = n + 1;
NoOfHiddenLayerNodes = NoOfInputNodes;

%Generate Random Weights
weightsInputHidden = (randn(NoOfInputNodes,NoOfHiddenLayerNodes) - 0.5)/10;
weightsHiddenOutput = (randn(1,NoOfHiddenLayerNodes) - 0.5)/10;

for iter = 1:NoIter
    randInput = randperm(m,m);
    for i = 1:m
        input = fea(randInput(i),:);
        target = OV(randInput(i),1);

          inputHidden = tanh(input*weightsInputHidden);
          HiddenOutput = inputHidden*weightsHiddenOutput';
          error = HiddenOutput - target;

          delta = error.*rate.*inputHidden;
          weightsHiddenOutput = weightsHiddenOutput - delta;

          delta= (rate/10).*error.*weightsHiddenOutput'.*(1-(inputHidden'.^2))*input;
          weightsInputHidden = weightsInputHidden - delta';
      end
      pred = weightsHiddenOutput*tanh(fea*weightsInputHidden)';
      err(iter) =  (sum((pred' - OV).^2))^0.5;
      fprintf('Epoch no -> %d; Error = %d\n',iter,err(iter));
      if err(iter) < 0.001
          fprintf('Converged at epoch: %d\n',iter);
          break;
      end
  end

if visualize == 1
    plot(1:iter,err);
    title('Error plot');
end

model.weightsInputHidden = weightsInputHidden;
model.weightsHiddenOutpu = weightsHiddenOutput;
end

Thank you.

[Angel Esqueda]

Could you solve it? Thanks in advance