During neural network training, can gradients leak sensitive information in case training data fed is encrypted (homomorphic)?
What is the effect of training a neural network with randomly generated fake data that satisfies certain constraints?
How many unique angles of an object do you need in your image training set in order to correctly classify it?
Given a dataset of people with and without cancer, should I split it into training and test datasets such that the same person is not in both?
Only top scored, non community-wiki answers of a minimum length are eligible