(Translated by https://www.hiragana.jp/)
[1502.03167v2] Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift