What is Batch Normalization and why is it used?
Updated May 15, 2026
Short answer
Batch Normalization stabilizes and accelerates neural network training.
Deep explanation
It normalizes layer inputs to zero mean and unit variance, then scales and shifts them using learnable parameters. This reduces internal covariate shift and allows higher learning rates.
Real-world example
Used in ResNet and EfficientNet for stable training.
Common mistakes
- Using batch norm incorrectly during inference mode.
Follow-up questions
- What is internal covariate shift?
- Why does batch norm improve generalization?