Neural network inputconnect automatically set to 0 after "train" was called

Hi,

I'm using neural network toolbox and having some problems. Here's the code

            net = patternnet([]);
            net = configure(net, trainF, trainL);
            net.divideFcn = '';
            net.performParam.regularization = lambda;
            net.layers{1}.transferFcn = 'softmax';
            net.trainFcn = 'traingd';
            net.trainParam.epochs = 10000;
            net.trainParam.show = 500;
            net.trainParam.lr = alpha;
            net.view
            net = train(net, trainF, trainL);

The problem is that after running, the IW is empty. I tried to print out the 'inputconnect' after each line was executed, and the result suggested that the inputconnect became 0 after 'train' was called. So before training, the network looked like this, which is what I want

After training, it became like this

I'd be very grateful if anyone can help figure out how to fix it, thanks

[Greg Heath]

If you are going to post patternnet code

1. Make sure it runs when cut and pasted into the command line

2. Show the results of running it on either

   a. The iris_dataset in the help and doc documentation

      help patternnet
      doc patternnet

    b. Another MATLAB classification dataset example obtained from

       help nndatasets
       doc nndatasets

3. Fix your code as suggested in 2 and repost.

Hope this helps.

Greg