Coverage for /pythoncovmergedfiles/medio/medio/usr/local/lib/python3.8/site-packages/keras/src/layers/pooling/global_max_pooling1d.py: 80%

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 max pooling 1D layer.""" 

16 

17 

18from keras.src import backend 

19from keras.src.layers.pooling.base_global_pooling1d import GlobalPooling1D 

20 

21# isort: off 

22from tensorflow.python.util.tf_export import keras_export 

23 

24 

25@keras_export("keras.layers.GlobalMaxPooling1D", "keras.layers.GlobalMaxPool1D") 

26class GlobalMaxPooling1D(GlobalPooling1D): 

27 """Global max pooling operation for 1D temporal data. 

28 

29 Downsamples the input representation by taking the maximum value over 

30 the time dimension. 

31 

32 For example: 

33 

34 >>> x = tf.constant([[1., 2., 3.], [4., 5., 6.], [7., 8., 9.]]) 

35 >>> x = tf.reshape(x, [3, 3, 1]) 

36 >>> x 

37 <tf.Tensor: shape=(3, 3, 1), dtype=float32, numpy= 

38 array([[[1.], [2.], [3.]], 

39 [[4.], [5.], [6.]], 

40 [[7.], [8.], [9.]]], dtype=float32)> 

41 >>> max_pool_1d = tf.keras.layers.GlobalMaxPooling1D() 

42 >>> max_pool_1d(x) 

43 <tf.Tensor: shape=(3, 1), dtype=float32, numpy= 

44 array([[3.], 

45 [6.], 

46 [9.], dtype=float32)> 

47 

48 Args: 

49 data_format: A string, 

50 one of `channels_last` (default) or `channels_first`. 

51 The ordering of the dimensions in the inputs. 

52 `channels_last` corresponds to inputs with shape 

53 `(batch, steps, features)` while `channels_first` 

54 corresponds to inputs with shape 

55 `(batch, features, steps)`. 

56 keepdims: A boolean, whether to keep the temporal dimension or not. 

57 If `keepdims` is `False` (default), the rank of the tensor is reduced 

58 for spatial dimensions. 

59 If `keepdims` is `True`, the temporal dimension are retained with 

60 length 1. 

61 The behavior is the same as for `tf.reduce_max` or `np.max`. 

62 

63 Input shape: 

64 - If `data_format='channels_last'`: 

65 3D tensor with shape: 

66 `(batch_size, steps, features)` 

67 - If `data_format='channels_first'`: 

68 3D tensor with shape: 

69 `(batch_size, features, steps)` 

70 

71 Output shape: 

72 - If `keepdims`=False: 

73 2D tensor with shape `(batch_size, features)`. 

74 - If `keepdims`=True: 

75 - If `data_format='channels_last'`: 

76 3D tensor with shape `(batch_size, 1, features)` 

77 - If `data_format='channels_first'`: 

78 3D tensor with shape `(batch_size, features, 1)` 

79 """ 

80 

81 def call(self, inputs): 

82 steps_axis = 1 if self.data_format == "channels_last" else 2 

83 return backend.max(inputs, axis=steps_axis, keepdims=self.keepdims) 

84 

85 

86# Alias 

87 

88GlobalMaxPool1D = GlobalMaxPooling1D 

89