I have read on some forum that there are training methods, where one is to train NN on whole training data at once, another one is to train NN in parts like the first from 1 to 1000 samples, then 1 to 2000 samples and so on. The third is to train NN in parts like first from 1 to 1000 samples, then 1000 to 2000 samples and so on. I want to ask that whether all these methods are correct and which to use when?

More Aamir Nawaz's questions See All
Similar questions and discussions