After training your model, you should check it with validation data set, which has no proportion in training step. If the accuracy for this data is really low, compare to the accuracy came from training set. We can conclude that your model was over-trained.

http://pages.cs.wisc.edu/~bolo/shipyard/neural/local.html

hi

you need to look at the curve of the error rate.

Plot the curve between the error of individual instance(Y axis) versus iteration (x axis).

by seeing the curve, if the curve is converging over time or number of iterations, then there is no overfitting else there is overfitting.

Thank you.

Hope this answer helps you a lot.

Best regards,

Indrajit Mandal

Thanks alot.