MATLAB: Neural Network train/val/test

neural networkvalidation

I need to develop a neural network to classify some objects. One of the parameters of a neural networks are the train/validation/test ratios. I was thinking to train the network using 70% of the labelled examples and the remaining 30% to the testing party.

Is the validation set a subset of the whole set or training set only? Do you suggest to do a 100% validation rate?

Best Answer

I suggest using the defaults. See

 help patternnet
 doc patternnet

Related Solutions

MATLAB: How to train a feedfordward neural network with error weights

 Use the command window
 help mse
 doc mse

Hope this helps.

Thank you for formally accepting my answer

Greg

MATLAB: Is the mse of the testing set in the neural network computed simultaneously with the mse of the training and validation sets

Yes.