MATLAB: Can I use divideind after using a dividerand in Neural Networks

Question

Hello, sorry if my english is bad. I'll try to explain as good as possible.

Here is the thing, in my neural network, I just generated from the input matrix, random values with "dividerand", and made 3 matrices as outputs of that function. Now, how could I use that output random values for training, evaluation and testing of my neural network? Can I use "divideind" and use those three matrices to train, evaluate and test? If so, how will be the syntax? I'm using this one "net.divideParam.testInd", but I guess I am wrong.

Here's the code for the three random value matrices:

[randtrain,randval,randtest] = dividerand(InReal',0.6,0.2,0.2)

Here is what I tried to do, but it gives me an error:

net.divideFcn = 'divideind';

net.divideMode = 'sample';

net.divideParam.trainInd = (randtrain');

net.divideParam.valInd = (randval');

net.divideParam.testInd = (randtest');

Error using network/subsasgn>network_subsasgn (line 528) net.divideParam.trainInd is not a vector.

Error in network/subsasgn (line 13) net = network_subsasgn(net,subscripts,v,netname);

Error in ANN_Avanzado_Base (line 31) net.divideParam.trainInd = (randtrain');

I will appreciate any answer, thank you for your time.

Answer 1

If you did this in order to know which examples are in each division subset, it is not necessary because if you use the syntax [ net tr ] = train(net,input,target), the training record tr will contain that and other info.

My guess is that you should not have transposed randtrain.

Next time you have a question, illustrate it with one of MATLAB's neural net datasets:

 help nndatasets

HTH

Thank you for formally accepting my answer

Greg