When calculating mean and std of the dataset, do we calculate it on the whole dataset including the test set or without it?
Also is normalization applied on the test/validation set or do they use something different like batch normalization?
Thank you