tf.keras.backend.normalize_batch_in_training

TensorFlow 2 version View source on GitHub

Computes mean and std for batch then apply batch_normalization on batch.

tf.keras.backend.normalize_batch_in_training(
    x, gamma, beta, reduction_axes, epsilon=0.001
)

x Input tensor or variable.
gamma Tensor by which to scale the input.
beta Tensor with which to center the input.
reduction_axes iterable of integers, axes over which to normalize.
epsilon Fuzz factor.

A tuple length of 3, (normalized_tensor, mean, variance).