- All neural network processing of the data is the average value close to 0, the standard deviation is close to 1, so that the convergence can be guaranteed as soon as the neural network
- batchNormalization means subtracting the average from the data, then divided by the standard deviation of the operation, like a layer added to the active function of an input front
Why should batchNormalization?
Guess you like
Origin www.cnblogs.com/rise0111/p/11619769.html
Ranking