Why should data be globally normalized and locally normalized at the same time?

torch / tutorials

A series of machine learning tutorials for Torch7

622 stars 313 forks source link

Open sunshineatnoon opened 8 years ago

sunshineatnoon commented 8 years ago

In /2-supervised/1_data.lua, the trainData and testData are first normalized globally and then locally, why do we need to normalize data twice?