Textbook of AI

Abstract. Introduces batch normalisation, which normalises layer inputs over each mini-batch to have zero mean and unit variance. The technique dramatically accelerates deep network training and is originally motivated by the goal of reducing 'internal covariate shift'.