Model
Sequential()- 서브 클래싱(Subclassing)
- 함수형 API
함수형 API
- 가장 권장되는 방법
- 모델을 복잡하고 유연하게 구성 가능
- 다중 입출력을 다룰 수 있음
1
2
3
4
5
6
7
8
inputs = Input(shape=(28, 28, 1))
x = Flatten(input_shape=(28, 28, 1))(inputs)
x = Dense(units=300, activation='relu')(x)
x = Dense(units=100, activation='relu')(x)
x = Dense(units=10, activation='softmax')(x)
model = Model(inputs=inputs, outputs=x)
model.summary()
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Model: "model"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
input_1 (InputLayer) [(None, 28, 28, 1)] 0
flatten (Flatten) (None, 784) 0
dense (Dense) (None, 300) 235500
dense_1 (Dense) (None, 100) 30100
dense_2 (Dense) (None, 10) 1010
=================================================================
Total params: 266,610
Trainable params: 266,610
Non-trainable params: 0
_________________________________________________________________
1
plot_model(model)
복잡한 모델 구성 방법
1
from tensorflow.keras.layers import Concatenate
1
2
3
4
5
6
7
8
9
10
11
input_layer = Input(shape=(28, 28))
hidden1 = Dense(units=100, activation='relu')(input_layer)
hidden2 = Dense(units=30, activation='relu')(hidden1)
# input_layer와 hidden2를 concat
concat = Concatenate()([input_layer, hidden2])
# 최종적으로 output은 dense layer로 하나만 고름
output = Dense(units=1)(concat)
model = Model(inputs=[input_layer], outputs=[output])
model.summary()
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Model: "model_1"
__________________________________________________________________________________________________
Layer (type) Output Shape Param # Connected to
==================================================================================================
input_2 (InputLayer) [(None, 28, 28)] 0 []
dense_3 (Dense) (None, 28, 100) 2900 ['input_2[0][0]']
dense_4 (Dense) (None, 28, 30) 3030 ['dense_3[0][0]']
concatenate (Concatenate) (None, 28, 58) 0 ['input_2[0][0]',
'dense_4[0][0]']
dense_5 (Dense) (None, 28, 1) 59 ['concatenate[0][0]']
==================================================================================================
Total params: 5,989
Trainable params: 5,989
Non-trainable params: 0
__________________________________________________________________________________________________
1
plot_model(model)
여러 개의 Input을 가지는 모델
1
2
3
4
5
6
7
8
9
10
11
12
13
14
# 입력층
input_1 = Input(shape=(10, 10), name='input_1')
input_2 = Input(shape=(10, 28), name='input_2')
# 은닉층
hidden1 = Dense(units=100, activation='relu')(input_2)
hidden2 = Dense(units=10, activation='relu')(hidden1)
# input_1과 hidden2를 Concatenate
concat = Concatenate()([input_1, hidden2])
output = Dense(units=1, activation='sigmoid', name='output')(concat)
model = Model(inputs=[input_1, input_2], outputs=[output])
model.summary()
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Model: "model_2"
__________________________________________________________________________________________________
Layer (type) Output Shape Param # Connected to
==================================================================================================
input_2 (InputLayer) [(None, 10, 28)] 0 []
dense_6 (Dense) (None, 10, 100) 2900 ['input_2[0][0]']
input_1 (InputLayer) [(None, 10, 10)] 0 []
dense_7 (Dense) (None, 10, 10) 1010 ['dense_6[0][0]']
concatenate_1 (Concatenate) (None, 10, 20) 0 ['input_1[0][0]',
'dense_7[0][0]']
output (Dense) (None, 10, 1) 21 ['concatenate_1[0][0]']
==================================================================================================
Total params: 3,931
Trainable params: 3,931
Non-trainable params: 0
__________________________________________________________________________________________________
1
2
# 입력은 두 개인데 출력은 한 개
plot_model(model)
여러 개의 출력을 가지는 모델
1
2
3
4
5
6
7
8
9
10
11
12
13
# 입력층
input_ = Input(shape=(10, 10), name='input')
# 은닉층
hidden1 = Dense(units=100, activation='relu')(input_)
hidden2 = Dense(units=10, activation='relu')(hidden1)
# 출력층
output = Dense(units=1, activation='sigmoid', name='main_output')(hidden2)
sub_out = Dense(units=1, name='sum_output')(hidden2)
model = Model(inputs=[input_], outputs=[output, sub_out])
model.summary()
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Model: "model_3"
__________________________________________________________________________________________________
Layer (type) Output Shape Param # Connected to
==================================================================================================
input (InputLayer) [(None, 10, 10)] 0 []
dense_8 (Dense) (None, 10, 100) 1100 ['input[0][0]']
dense_9 (Dense) (None, 10, 10) 1010 ['dense_8[0][0]']
main_output (Dense) (None, 10, 1) 11 ['dense_9[0][0]']
sum_output (Dense) (None, 10, 1) 11 ['dense_9[0][0]']
==================================================================================================
Total params: 2,132
Trainable params: 2,132
Non-trainable params: 0
__________________________________________________________________________________________________
1
2
# 입력은 하나인데 출력은 두 개
plot_model(model)
1
2
3
4
5
6
7
8
9
10
11
12
13
input_1 = Input(shape=(10, 10), name='input_1')
input_2 = Input(shape=(10, 28), name='input_2')
hidden1 = Dense(units=100, activation='relu')(input_2)
hidden2 = Dense(units=10, activation='relu')(hidden1)
concat = Concatenate()([input_1, hidden2])
output = Dense(units=1, activation='sigmoid', name='main_output')(concat)
sub_out = Dense(units=1, name='sum_output')(hidden2)
model = Model(inputs=[input_1, input_2], outputs=[output, sub_out])
model.summary()
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Model: "model_6"
__________________________________________________________________________________________________
Layer (type) Output Shape Param # Connected to
==================================================================================================
input_2 (InputLayer) [(None, 10, 28)] 0 []
dense_14 (Dense) (None, 10, 100) 2900 ['input_2[0][0]']
input_1 (InputLayer) [(None, 10, 10)] 0 []
dense_15 (Dense) (None, 10, 10) 1010 ['dense_14[0][0]']
concatenate_4 (Concatenate) (None, 10, 20) 0 ['input_1[0][0]',
'dense_15[0][0]']
main_output (Dense) (None, 10, 1) 21 ['concatenate_4[0][0]']
sum_output (Dense) (None, 10, 1) 11 ['dense_15[0][0]']
==================================================================================================
Total params: 3,942
Trainable params: 3,942
Non-trainable params: 0
__________________________________________________________________________________________________
1
plot_model(model)
