keras顺序模型与函数式模型

• keras提供两种搭建模型的方式：
  
• 
  

• 1. 顺序模型（也可看做是函数式模型的一种特殊情况）
  
• 2. 函数式模型
  

[tr]两种方式优点缺点[/tr]

顺序模型	单输入单输出，搭建比较简单，是多个网络层的线性堆叠，不发生分叉，编译速度快	不能定义复杂模型
函数式模型	灵活，层与层之间可以任意连接可以定义复杂模型（如多输出模型、有向无环图，或具有共享层的模型）	搭建复杂，编译速度慢

顺序模型也可看做是函数式模型的一种特殊情况。
以下以mnist为例分别进行说明：
1 顺序模型Sequential

顺序模型也即一层叠一层，所以模型不发生分叉，单输入单输出，这意味着顺序模型不支持共享图层或多输入多输出模型。
第0讲顺序模型搭建mnist的代码如下：
方式1：采用model.add 一层层添加
import tensorflow as tf
import tensorflow.keras as keras
​
print(keras.__version__)
​
(x_train, y_train), (x_valid, y_valid) = keras.datasets.mnist.load_data()
assert x_train.shape == (60000, 28, 28)
assert x_valid.shape == (10000, 28, 28)
assert y_train.shape == (60000,)
assert y_valid.shape == (10000,)
​
# step1: use sequential
model = keras.models.Sequential()
​
# step2: add layer
model.add(keras.layers.Flatten(input_shape=(x_train.shape[1], x_train.shape[2])))
model.add(keras.layers.Dense(units=784, activation="relu", input_dim=784))
model.add(keras.layers.Dense(units=10, activation="softmax"))
​
# step3: compile model
model.compile(optimizer="Adam", loss='sparse_categorical_crossentropy', metrics=['accuracy'])
​
print("model:")
model.summary()
​
# step4: train
model.fit(x_train, y_train, batch_size=64, epochs=5)
​
# step5: evaluate model
model.Evaluate (x_valid, y_valid)
​
# save model
model.save('keras_mnist.h5')方式2：初始化Sequential时同时定义层
import tensorflow as tf
import tensorflow.keras as keras
​
print(keras.__version__)
​
(x_train, y_train), (x_valid, y_valid) = keras.datasets.mnist.load_data()
assert x_train.shape == (60000, 28, 28)
assert x_valid.shape == (10000, 28, 28)
assert y_train.shape == (60000,)
assert y_valid.shape == (10000,)
​
# step1, step2: Init sequential
model = keras.models.Sequential([
    keras.layers.Flatten(input_shape=(x_train.shape[1], x_train.shape[2])),
    keras.layers.Dense(units=784, activation="relu", input_dim=784),
    keras.layers.Dense(units=10, activation="softmax")
])
​
# step3: compile model
model.compile(optimizer="Adam", loss='sparse_categorical_crossentropy', metrics=['accuracy'])
​
print("model:")
model.summary()
​
# step4: train
model.fit(x_train, y_train, batch_size=64, epochs=5)
​
# step5: evaluate model
model.evaluate(x_valid, y_valid)
​
# save model
model.save('keras_mnist.h5')2 函数式API模型

顺序模型允许逐层创建模型以解决许多问题，但是有时我们需要创建更加复杂的模型，比如具有共享图层或多输入多输出的模型，这时我们可以使用keras提供的函数式API模型，它定义了每一层的输入输出，像函数调用一样，用上一层的输出作为这一层的入参，这一层的输出作为函数返回值，这就是函数式API的由来，（这一点不同与顺序模型，顺序模型每层的输入输出是一定的，不需要专门指定）。
同样的，利用函数式API模型搭建与之前等价的mnist模型，代码如下：
import tensorflow as tf
import tensorflow.keras as keras
​
print(keras.__version__)
​
(x_train, y_train), (x_valid, y_valid) = keras.datasets.mnist.load_data()
assert x_train.shape == (60000, 28, 28)
assert x_valid.shape == (10000, 28, 28)
assert y_train.shape == (60000,)
assert y_valid.shape == (10000,)
​
# step1: Model structure
# 层的实例是可调用的，它以张量为参数，并且返回一个张量
input = keras.Input(shape=(x_train.shape[1], x_train.shape[2]))
x = keras.layers.Flatten()(input)
x = keras.layers.Dense(units=784, activation="relu", input_dim=784)(x)
output = keras.layers.Dense(units=10, activation="softmax")(x)
​
# step2: define Model
model = keras.Model(inputs=input, outputs=output)
​
# step3: compile model
model.compile(optimizer="Adam", loss='sparse_categorical_crossentropy', metrics=['accuracy'])
​
print("model:")
model.summary()
​
# step4: train
model.fit(x_train, y_train, batch_size=64, epochs=5)
​
# step5: evaluate model
model.evaluate(x_valid, y_valid)
​
# save model
model.save('keras_mnist.h5')由这个例子可见：函数式API可以搭建顺序模型Sequential，所以顺序模型也可看做是函数式模型的一种特殊情况。
还有另外一种Model子类的搭建方法，这是利用python call方法的一种写法，我觉得只是写法上的区别，本质也是函数式API，代码如下：
import tensorflow as tf
import tensorflow.keras as keras
​
print(keras.__version__)
​
(x_train, y_train), (x_valid, y_valid) = keras.datasets.mnist.load_data()
assert x_train.shape == (60000, 28, 28)
assert x_valid.shape == (10000, 28, 28)
assert y_train.shape == (60000,)
assert y_valid.shape == (10000,)
​
# step1: Model structure
class MnistModel(tf.keras.Model):
    def __init__(self):
        super(MnistModel, self).__init__()
        self.x0 = keras.layers.Flatten()
        self.x1 = keras.layers.Dense(units=784, activation="relu", input_dim=784)
        self.x2 = keras.layers.Dense(units=10, activation="softmax")
​
    def call(self, input):
        x = self.x0(input)
        x = self.x1(x)
        x = self.x2(x)
        return x
​
# step2: define Model
model = MnistModel()
​
# step3: compile model
model.compile(optimizer="Adam", loss='sparse_categorical_crossentropy', metrics=['accuracy'])
​
# step4: train
model.fit(x_train, y_train, batch_size=64, epochs=5)
​
# 要放到model.fit之后
print("model:")
model.summary()
​
# step5: evaluate model
model.evaluate(x_valid, y_valid)
​
# save model （不能使用）
# model.save('keras_mnist.h5')采用model子类方法有几个问题需要注意：1. 采用model子类方法时，model.summary方法需要放到model.fit之后才能调用，否则报如下错误：
ValueError: This model has not yet been built. Build the model first by calling `build()` or by calling the model on a batch of data.2. 采用model子类方法时，调用model.save会失败：
NotImplementedError: Saving the model to HDF5 format requires the model to be a Functional model or a Sequential model. It does not work for subclassed models, because such models are defined via the body of a Python method, which isn't safely serializable. Consider saving to the Tensorflow SavedModel format (by setting save_format="tf") or using `save_weights`.