【深度学习】MLP,CNN,RNN基本定义与区别

学习记录

定义

MLP

multilayer perceptron(多层感知机器)。常用于处理简单逻辑数学问题。

CNN

Convolutional neural network(卷积神经网络)。适用于视频和图片的处理工作

RNN

recurrent neural network(递归神经网络)。常用于预测数据。
在RNN中，还有SimpleRNN，RNN等的神经元模型结构。如下图所示

模型框架图

以手写数据集图片为例子，分别展示三种类型的网络结构
对于MNIST的一张输入为例子，三种模型对应的输入格式如下

具体模型架构

区别

RNN与前两个输入大不同，实际的input_shape = (timesteps,input_dim)或者是时间步长的输入多维向量序列。

代码实现

MLP

import numpy as np
from tensorflow.keras.models import Sequential 
from tensorflow.keras.layers import Dense,Activation,Dropout
from tensorflow.keras.utils import to_categorical,plot_model
from tensorflow.keras.datasets import mnist

# 加载数据集
(x_train,y_train),(x_test,y_test)=mnist.load_data()
# 计算类别数量
num_labels = len(np.unique(y_train))
# 转换为独热编码
y_train = to_categorical(y_train)
y_test = to_categorical(y_test)
# 图片维度
image_size = x_train.shape[1]
input_size = image_size * image_size 

# 转换大小和归一化
x_train = np.reshape(x_train,[-1,input_size])
x_train = x_train.astype("float32")/255 
x_test = np.reshape(x_test,[-1,input_size])
x_test = x_test.astype("float32")/255

# 模型参数
batch_size = 128
hidden_units = 256
dropout = 0.45

# 模型时三层感知机，每层是relu激活函数和dropout
model = Sequential()
'''输入层'''
model.add(Dense(256,input_dim = input_size))
model.add(Activation('relu'))
model.add(Dropout(dropout))
'''隐藏层'''
model.add(Dense(hidden_units))
model.add(Activation('relu'))
model.add(Dropout(dropout))
'''输出层'''
model.add(Dense(num_labels))
# for 独热编码的输出
model.add(Activation("softmax"))
model.summary()
plot_model(model,show_shapes=True)
# 编译模型
model.compile(loss='categorical_crossentropy',optimizer='adam',metrics=['accuracy'])
# 训练模型
model.fit(x_train,y_train,epochs=20,batch_size=batch_size)
# 验证模型
_,acc = model.evaluate(x_test,y_test,batch_size=batch_size,verbose=0)
print("\nTest accuracy: %.1f%%" % (100.0 * acc))

CNN

import numpy as np
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Activation,Dense,Dropout
from tensorflow.keras.layers import Conv2D,MaxPooling2D,Flatten
from tensorflow.keras.utils import to_categorical,plot_model
from tensorflow.keras.datasets import mnist 

# 加载数据集
(x_train,y_train),(x_test,y_test)=mnist.load_data()
# 计算类别
num_labels = len(np.unique(y_train))
# 转换为独热编码
y_train = to_categorical(y_train)
y_test = to_categorical(y_test)

# 输入图片维度
image_size = x_train.shape[1]
# 更改图片尺寸和归一化
x_train = np.reshape(x_train,[-1,image_size,image_size,1])
x_test = np.reshape(x_test,[-1, image_size, image_size, 1])
x_train = x_train.astype('float32') / 255
x_test = x_test.astype('float32') / 255

# 网络参数
input_shape = (image_size,image_size,1)
batch_size = 128
kernel_size= 3
pool_size = 2 
filters = 64
dropout = 0.2 

# 构建模型
model = Sequential()
model.add(Conv2D(
    filters=filters,
    kernel_size=kernel_size,
    activation='relu',
    input_shape=input_shape))

model.add(MaxPooling2D(pool_size))
model.add(Conv2D(
    filters=filters,
    kernel_size=kernel_size,
    activation='relu'))
model.add(MaxPooling2D(pool_size))
model.add(Conv2D(
    filters=filters,
    kernel_size=kernel_size,
    activation='relu'))
model.add(Flatten())
model.add(Dropout(dropout))
model.add(Dense(num_labels))
model.add(Activation('softmax'))
model.summary()
plot_model(model, to_file='cnn-mnist.png', show_shapes=True)
model.compile(loss='categorical_crossentropy',
              optimizer='adam',
              metrics=['accuracy'])
model.fit(x_train,y_train,epochs=10,batch_size=batch_size)
_,acc = model.evaluate(x_test,y_test,batch_size=batch_size,verbose=0)
print("\nTest accuracy: %.1f%%" % (100.0 * acc))

RNN

import numpy as np
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Activation,Dense,Dropout,SimpleRNN
from tensorflow.keras.layers import Conv2D,MaxPooling2D,Flatten
from tensorflow.keras.utils import to_categorical,plot_model
from tensorflow.keras.datasets import mnist 
# 加载数据集
(x_train,y_train),(x_test,y_test)=mnist.load_data()
# 计算类别
num_labels = len(np.unique(y_train))
# 转换为独热编码
y_train = to_categorical(y_train)
y_test = to_categorical(y_test)

# 输入图片维度
image_size = x_train.shape[1]
# 更改图片尺寸和归一化
x_train = np.reshape(x_train,[-1,image_size,image_size,1])
x_test = np.reshape(x_test,[-1, image_size, image_size, 1])
x_train = x_train.astype('float32') / 255
x_test = x_test.astype('float32') / 255

# 网络参数
input_shape = (image_size, image_size)
batch_size = 128
units = 256
dropout = 0.2

# 构建模型
model = Sequential()
model.add(SimpleRNN(units=units,dropout=dropout,input_shape=input_shape))
model.add(Dense(num_labels))
model.add(Activation("softmax"))
model.summary()
plot_model(model, to_file='rnn-mnist.png', show_shapes=True)

model.compile(loss='categorical_crossentropy',optimizer='adam',metrics=['accuracy'])
model.fit(x_train,y_train,epochs=20,batch_size=batch_size)
_,acc = model.evaluate(x_test,y_test,batch_size=batch_size,verbose=0)
print("\nTest accuracy: %.1f%%" % (100.0 * acc))

参考

《Advanced Deep Learning with TensorFlow 2 and Keras》