Hi,
I was wondering if anyone could help me answer these questions I have.
1. When we run the NN, how do we know how many weights were optimized?
2. In earlier JMP platforms there was early stopping to regularize the networks and avoid overfitting. In JMP 14, I think we still have early stoping, but we can also add K-fold cross validation to help limit this too. Is that correct? Do these also determine when the network stops after x number of epochs? Do we have any other objective stopping criterion?
3. How sensitive is the model to missing data that is imputed using a normal distribution?
4. If we have multiple outputs, does the network build different models or is it more multi-task?
Thank you for your time.
