This keyword indicates number of iterations that will be used
in neural network training.
The neural network training is stopped if there is no improvement of
RMSE error for the validation set (or learning set, if no validation
option is selected in PARTITION) after ITERATIONS
iterations (this corresponds to detection of the early
stopping point) or RMSE error for the training set is less than LIMIT
value. The training is also stopped if total number of iterations in program
equals to 5*ITERATIONS.
We suggest that user perform a test run using 5-10 ensembles and SHOW
option to identify which number of iterations
is actually required for his/her task.
Default value is {1000}.