1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
|
<span style="float:right;">[[source]](https://github.com/keras-team/keras/blob/master/keras/layers/normalization.py#L16)</span>
### BatchNormalization
```python
keras.layers.BatchNormalization(axis=-1, momentum=0.99, epsilon=0.001, center=True, scale=True, beta_initializer='zeros', gamma_initializer='ones', moving_mean_initializer='zeros', moving_variance_initializer='ones', beta_regularizer=None, gamma_regularizer=None, beta_constraint=None, gamma_constraint=None)
```
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.
__Arguments__
- __axis__: Integer, the axis that should be normalized
(typically the features axis).
For instance, after a `Conv2D` layer with
`data_format="channels_first"`,
set `axis=1` in `BatchNormalization`.
- __momentum__: Momentum for the moving mean and the moving variance.
- __epsilon__: Small float added to variance to avoid dividing by zero.
- __center__: If True, add offset of `beta` to normalized tensor.
If False, `beta` is ignored.
- __scale__: If True, multiply by `gamma`.
If False, `gamma` is not used.
When the next layer is linear (also e.g. `nn.relu`),
this can be disabled since the scaling
will be done by the next layer.
- __beta_initializer__: Initializer for the beta weight.
- __gamma_initializer__: Initializer for the gamma weight.
- __moving_mean_initializer__: Initializer for the moving mean.
- __moving_variance_initializer__: Initializer for the moving variance.
- __beta_regularizer__: Optional regularizer for the beta weight.
- __gamma_regularizer__: Optional regularizer for the gamma weight.
- __beta_constraint__: Optional constraint for the beta weight.
- __gamma_constraint__: Optional constraint for the gamma weight.
__Input shape__
Arbitrary. Use the keyword argument `input_shape`
(tuple of integers, does not include the samples axis)
when using this layer as the first layer in a model.
__Output shape__
Same shape as input.
__References__
- [Batch Normalization: Accelerating Deep Network Training by
Reducing Internal Covariate Shift](https://arxiv.org/abs/1502.03167)
|
