According to the top answer over 知乎 on the question 深度学习中 Batch Normalization为什么效果好?:
顾名思义,batch normalization嘛,就是“批规范化”咯。Google在ICML文中描述的非常清晰,即在每次SGD时,通过mini-batch来对相应的activation做规范化操作,使得结果(输出信号各个维度)的均值为0,方差为1.
What does the 即 mean? Does this function as be
? In this case, is it necessary to use 即, instead of 在每次SGD时, and also why is 即 preferred to other words that mean be
, such as 是,为, etc...?