回答編集履歴

1

fix pooling layer

2023/02/14 12:39

投稿

ps_aux_grep
ps_aux_grep

スコア1579

test CHANGED
@@ -2,7 +2,7 @@
2
2
 
3
3
  [ResNet](https://arxiv.org/abs/1512.03385)スタイル.ResNet18と同等.
4
4
  ```Python
5
- from keras.layers import BatchNormalization, Activation, Add, Dropout
5
+ from keras.layers import BatchNormalization, Activation, Add, Dropout, MaxPooling1D
6
6
  def ConvBNAct(filters, kernel_size, activation = "swish"):
7
7
  def apply(inputs):
8
8
  x = Conv1D(filters, kernel_size, padding = "same", kernel_initializer = "he_uniform")(inputs)
@@ -25,12 +25,16 @@
25
25
 
26
26
  x = Concatenate(axis = -1)([inputs1, inputs2, inputs3])
27
27
  x = ResBlock(64, 7)(x)
28
+ x = MaxPooling1D()(x)
28
29
  x = ResBlock(64, 5)(x)
29
30
  x = ResBlock(64, 3)(x)
31
+ x = MaxPooling1D()(x)
30
32
  x = ResBlock(128, 3)(x)
31
33
  x = ResBlock(128, 3)(x)
34
+ x = MaxPooling1D()(x)
32
35
  x = ResBlock(256, 3)(x)
33
36
  x = ResBlock(256, 3)(x)
37
+ x = MaxPooling1D()(x)
34
38
  x = ResBlock(512, 3)(x)
35
39
  x = ResBlock(512, 3)(x)
36
40
  x = GlobalAveragePooling1D()(x)