What is batch normalization in Keras?

Updated May 16, 2026

Short answer

Batch normalization normalizes layer inputs to stabilize training.

Deep explanation

It reduces internal covariate shift and speeds up convergence by normalizing activations.

Real-world example

Used in deep CNNs like ResNet for stable training.

Common mistakes

  • Placing BatchNorm after activation incorrectly.

Follow-up questions

  • Does BatchNorm improve accuracy?
  • Where to place BatchNorm?

More Keras interview questions

View all →