How is the performance function evaluated for a closed loop NARX network ?. Is the same training set used in every epoch of the training ?

How is the performance function evaluated for training a closed loop NARX network?. Is it evaluated recursively using the previous predictions?. If so how is the training set divided?. Would it not always have the same initial conditions?.

Answer 1

[Greg Heath]

Search both NEWREADER ad ANSWERS using

 greg narx
 greg narxnet

Hope this helps.

*Thank you for formally accepting this answer.

Greg