early stopping neural network:: why data divided in two three subset???

dear

as i know from wiki and book by Simon hakin that early stopping use two type sets of data the first one for training and second one for validation test so which may ask

why their are three curves in performance plot when we use matlab train function ??? the curve name train error,validation and test error ???

Answer 1

[Greg Heath]

Data = Design + Non-design

Design = Train(estimate weights) + Validation(Stop training when MSEval goes thru a minimum).

Non-design = Test(Obtain UNBIASED generalization estimate of performance on unseen non-design data).

[mangood UK]

!!!!!!