Batch normalization

Batch normalization is a technique used in training neural networks that standardizes the inputs to a layer for each mini-batch. This helps to stabilize the learning process and improve the speed of convergence during training.