Batch normalization layer (Ioffe and Szegedy, 2014).
Normalize the activations of the previous layer at each batch,
i.e. applies a transformation that maintains the mean activation
close to 0 and the activation standard deviation close to 1.
Input shape:
Arbitrary. Use the keyword argument inputShape (Array of integers, does
not include the sample axis) when calling the constructor of this class,
if this layer is used as a first layer in a model.
Batch normalization layer (Ioffe and Szegedy, 2014).
Normalize the activations of the previous layer at each batch, i.e. applies a transformation that maintains the mean activation close to 0 and the activation standard deviation close to 1.
Input shape: Arbitrary. Use the keyword argument
inputShape(Array of integers, does not include the sample axis) when calling the constructor of this class, if this layer is used as a first layer in a model.Output shape: Same shape as input.
References: