Batch Normalization Mechanism

Medium

neural-networks Google Meta NVIDIA

Batch normalization normalizes activations within a mini-batch during training. What problem does this primarily address, and what happens at inference time?

Multiple Choice

It addresses internal covariate shift; at inference it uses running averages of mean and variance computed during training It prevents exploding gradients; at inference the normalization is removed entirely It reduces the need for dropout; at inference it applies double normalization for stability It ensures all weights remain small; at inference it freezes the learned scale and shift to zero

Correct!

Incorrect — try again next time!

Upgrade to Pro to unlock solutions and explanations

Batch Normalization Mechanism

Multiple Choice

Related problems

Deep-dive guides