I am attempting to generate confidence intervals for a neural network, according to the article here, one would bootstrap from the training set to generate new models several times, and then apply the model to the validation set n number of times to generate the relevant statistics.
Is there another way to generate confidence intervals that doesn't require retraining the network? Retraining the network is very computationally intense and requires a great deal of time-- so I would like to avoid it if at all possible.
Thanks so much!