Keras添加网络层的N种方法

序

使用Keras搭神经网络久了，想整理一下Keras添加网络层的N种方法，以防日后忘记。
本文以mnist手写数据为例，神经网络结构使用LeNet讲解。
以下是模块的导入，数据的加载，数据的预处理。

import keras
from keras.applications.mobilenet_v2 import MobileNetV2
from keras.datasets import mnist
from keras.layers import Conv2D, MaxPool2D, Activation, Dropout, Flatten, Dense
from keras.losses import categorical_crossentropy
from keras.models import Model, Sequential, Input
from keras.optimizers import SGD
from keras.utils import to_categorical
import numpy as np
(X_train, y_train), (X_test, y_test) = mnist.load_data()
X_train = np.expand_dims(X_train, 3)
X_test = np.expand_dims(X_test, 3)
y_train = to_categorical(y_train)
y_test = to_categorical(y_test)

方法一

新建一个Sequential对象，然后使用它的add方法，一层层的去加。

model = Sequential()
model.add(Conv2D(20, (3, 3), input_shape=(X_train.shape[1], X_train.shape[2], X_train.shape[3])))
model.add(MaxPool2D((2, 2)))
model.add(Activation('tanh'))
model.add(Conv2D(30, (3, 3)))
model.add(MaxPool2D((2, 2)))
model.add(Activation('tanh'))
model.add(Dropout(0.5))
model.add(Flatten())
model.add(Dense(1000))
model.add(Dropout(0.5))
model.add(Dense(y_train.shape[1]))
model.add(Activation('softmax'))
model.compile(optimizer=SGD(learning_rate=0.01, momentum=0.9, nesterov=True), loss=categorical_crossentropy, metrics=['accuracy'])
model.fit(X_train, y_train, batch_size=1000, epochs=10)

方法二

在新建Sequential对象时，以列表的形式将每一层按顺序放到列表里，作为参数传进去。

model = Sequential([
    Conv2D(20, (3, 3), input_shape=(X_train.shape[1], X_train.shape[2], X_train.shape[3])),
    MaxPool2D((2, 2)),
    Activation('tanh'),
    Conv2D(30, (3, 3)),
    MaxPool2D((2, 2)),
    Activation('tanh'),
    Dropout(0.5),
    Flatten(),
    Dense(1000),
    Dropout(0.5),
    Dense(y_train.shape[1]),
    Activation('softmax'),
    ])
model.compile(optimizer=SGD(learning_rate=0.01, momentum=0.9, nesterov=True), loss=categorical_crossentropy, metrics=['accuracy'])
model.fit(X_train, y_train, batch_size=1000, epochs=10)

方法三

自行定义输入层，输出层，从输入层开始，前向传播到下一层，最后到达输出层。然后创建Model对象，指定inputs和outputs变量。

x_i = Input((X_train.shape[1], X_train.shape[2], X_train.shape[3]))
x = Conv2D(20, (3, 3))(x_i)
x = MaxPool2D((2, 2))(x)
x = Activation('tanh')(x)
x = Conv2D(30, (3, 3))(x)
x = MaxPool2D((2, 2))(x)
x = Activation('tanh')(x)
x = Dropout(0.5)(x)
x = Flatten()(x)
x = Dense(1000)(x)
x = Dropout(0.5)(x)
x = Dense(y_train.shape[1])(x)
pred = Activation('softmax')(x)
model = Model(inputs=x_i, outputs=pred)
model.compile(optimizer=SGD(learning_rate=0.01, momentum=0.9, nesterov=True), loss=categorical_crossentropy, metrics=['accuracy'])
model.fit(X_train, y_train, batch_size=1000, epochs=10)