Coverage for /pythoncovmergedfiles/medio/medio/usr/local/lib/python3.8/site-packages/keras/src/layers/pooling/global_average_pooling1d.py: 52%
21 statements
« prev ^ index » next coverage.py v7.4.0, created at 2024-01-03 07:57 +0000
« prev ^ index » next coverage.py v7.4.0, created at 2024-01-03 07:57 +0000
1# Copyright 2015 The TensorFlow Authors. All Rights Reserved.
2#
3# Licensed under the Apache License, Version 2.0 (the "License");
4# you may not use this file except in compliance with the License.
5# You may obtain a copy of the License at
6#
7# http://www.apache.org/licenses/LICENSE-2.0
8#
9# Unless required by applicable law or agreed to in writing, software
10# distributed under the License is distributed on an "AS IS" BASIS,
11# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12# See the License for the specific language governing permissions and
13# limitations under the License.
14# ==============================================================================
15"""Global average pooling 1D layer."""
18import tensorflow.compat.v2 as tf
20from keras.src import backend
21from keras.src.layers.pooling.base_global_pooling1d import GlobalPooling1D
23# isort: off
24from tensorflow.python.util.tf_export import keras_export
27@keras_export(
28 "keras.layers.GlobalAveragePooling1D", "keras.layers.GlobalAvgPool1D"
29)
30class GlobalAveragePooling1D(GlobalPooling1D):
31 """Global average pooling operation for temporal data.
33 Examples:
35 >>> input_shape = (2, 3, 4)
36 >>> x = tf.random.normal(input_shape)
37 >>> y = tf.keras.layers.GlobalAveragePooling1D()(x)
38 >>> print(y.shape)
39 (2, 4)
41 Args:
42 data_format: A string,
43 one of `channels_last` (default) or `channels_first`.
44 The ordering of the dimensions in the inputs.
45 `channels_last` corresponds to inputs with shape
46 `(batch, steps, features)` while `channels_first`
47 corresponds to inputs with shape
48 `(batch, features, steps)`.
49 keepdims: A boolean, whether to keep the temporal dimension or not.
50 If `keepdims` is `False` (default), the rank of the tensor is reduced
51 for spatial dimensions.
52 If `keepdims` is `True`, the temporal dimension are retained with
53 length 1.
54 The behavior is the same as for `tf.reduce_mean` or `np.mean`.
56 Call arguments:
57 inputs: A 3D tensor.
58 mask: Binary tensor of shape `(batch_size, steps)` indicating whether
59 a given step should be masked (excluded from the average).
61 Input shape:
62 - If `data_format='channels_last'`:
63 3D tensor with shape:
64 `(batch_size, steps, features)`
65 - If `data_format='channels_first'`:
66 3D tensor with shape:
67 `(batch_size, features, steps)`
69 Output shape:
70 - If `keepdims`=False:
71 2D tensor with shape `(batch_size, features)`.
72 - If `keepdims`=True:
73 - If `data_format='channels_last'`:
74 3D tensor with shape `(batch_size, 1, features)`
75 - If `data_format='channels_first'`:
76 3D tensor with shape `(batch_size, features, 1)`
77 """
79 def __init__(self, data_format="channels_last", **kwargs):
80 super().__init__(data_format=data_format, **kwargs)
81 self.supports_masking = True
83 def call(self, inputs, mask=None):
84 steps_axis = 1 if self.data_format == "channels_last" else 2
85 if mask is not None:
86 mask = tf.cast(mask, inputs[0].dtype)
87 mask = tf.expand_dims(
88 mask, 2 if self.data_format == "channels_last" else 1
89 )
90 inputs *= mask
91 return backend.sum(
92 inputs, axis=steps_axis, keepdims=self.keepdims
93 ) / tf.reduce_sum(mask, axis=steps_axis, keepdims=self.keepdims)
94 else:
95 return backend.mean(inputs, axis=steps_axis, keepdims=self.keepdims)
97 def compute_mask(self, inputs, mask=None):
98 return None
101# Alias
103GlobalAvgPool1D = GlobalAveragePooling1D