How to use divideind and then get the validate and test accuracy in the confusion matrix?

I am creating a neural network through the command window. I have managed to create it's structure however when I try to use the 'divideind' to split the data into three data sets by index, when I try to train the network using: net = train(net,[inputs],[targets]) ; the confusion matrix will obviously only display the Training and All confusion matrix. How can I pass the other inputs (val/test) and targets (val/test) to get the results for the validation and testing in the confusion matrix?

The reason I am not using 'divideblock' or divideParam is because I am comparing different ANNs (ie different activation functions and normalized (used min-max) and un-normalized data and therefore must be sure that the data in all the neural network is the same (same index) in all sub-datasets to avoid bias results)

Thanks

Anthony

Answer 1

[Greg Heath]

Regardless of divideFcn, use all of the data at once in train. Make sure the target columns are columns of eye(c) if there are c classes