机器学习：【9】函数式api实践（多输入模型）

函数式API简介：

不同于Sequential结构的网络，函数式api可以自行设计网络结构。

核心思想在于可以调用，每一层都可以用调用来实现。

#本代码基于tensorflow2.0
import tensorflow as tf
from tensorflow import keras
import matplotlib.pyplot as plt
%matplotlib inline

fashion_mnist = keras.datasets.fashion_mnist

#1 导入数据
(train_images, train_labels),(test_images, test_labels) = fashion_mnist.load_data()

train_images = train_images / 255
test_images = test_images / 255

#2 建立模型
#2.1 设置输入
input = keras.Input(shape=(28, 28)) #不用设置第一维

#2.2 构建模型
x = keras.layers.Flatten()(input) #调用Flatten层
x = keras.layers.Dense(32, activation = 'relu')(x) #
x = keras.layers.Dropout(0.5)(x)
x = keras.layers.Dense(64, activation = 'relu')(x)
output = keras.layers.Dense(10, activation = 'softmax')(x)
#相当于反复调用上一步的输出，一层一层往下套

model = keras.Model(inputs = input, outputs = output)
model.summary()

#3 模型编译
model.compile(optimizer='adam',
              loss = 'sparse_categorical_crossentropy',
              metrics=['accuracy'])

#4 模型训练
model.fit(train_images, 
          train_labels,  
          epochs=30,
          validation_data=(test_images, test_labels))

 简单实现多输入模型(部分代码)：

input1 = keras.Input(shape = (28, 28))
input2 = keras.Input(shape = (28, 28))

x1 = keras.layers.Flatten()(input1)
x2 = keras.layers.Flatten()(input2)

x = keras.layers.concatenate([x1,x2]) # 这一行用来将两个输入合成

x = keras.layers.Dense(32, activation = 'relu')(x)

output = keras.layers.Dense(1, activation = 'sigmoid')(x)

model = keras.Model(inputs=[input1, input2], outputs = output)

model.summary()