黑马深度学习和CV入门2——深度学习

01.神经网络、优化方法与正则化

激活函数

import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt

x = np.linspace(-10,10,1000)
y = tf.nn.sigmoid(x)
plt.plot(x,y)
plt.grid()
plt.show()

import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt

x = np.linspace(-10,10,1000)
y = tf.nn.tanh(x)
plt.plot(x,y)
plt.grid()
plt.show()

import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt

x = np.linspace(-10,10,1000)
y = tf.nn.relu(x)
plt.plot(x,y)
plt.grid()
plt.show()

参数初始化

神经网络的搭建

import tensorflow as tf
import tensorflow.keras as keras
import tensorflow.keras.layers as layers


model = keras.Sequential([
    #第一个隐层
    layers.Dense(3,activation="relu",kernel_initializer="he_normal",name="layer1",input_shape=(3,)),
    #第二个隐层
    layers.Dense(2,activation="relu",kernel_initializer="he_normal",name="layer2"),
    #输出层
    layers.Dense(2,activation="relu",kernel_initializer="he_normal",name="layer3")
])
model.summary()

利用function API构建神经网络

import tensorflow as tf
import tensorflow.keras as keras
import tensorflow.keras.layers as layers


#定义模型的输入
inputs = keras.Input(shape=(3,),name='input')
#第一个隐层
x = layers.Dense(3,activation="relu",name="layer1")(inputs)
#第二个隐层
x = layers.Dense(2,activation="relu",name="layer2")(x)
#输出层
outputs = layers.Dense(2,activation="sigmoid",name="layer3")(x)
#创建模型
model = keras.Model(inputs=inputs,outputs=outputs,name="Functional API Model")

model.summary()

通过model的子类构建神经网络

import tensorflow as tf
import tensorflow.keras as keras
import tensorflow.keras.layers as layers



#定义一个model的子类
class MyModel(keras.Model):
    #定义网络的层结构
    def __init__(self):
        super(MyModel,self).__init__()
        #第一个隐层
        self.layer1 = layers.Dense(3,activation="relu",name="layer1")
        #第二个隐层
        self.layer2 = layers.Dense(2,activation="relu",name="layer2")
        #输出层
        self.layer3 = layers.Dense(2,activation="sigmoid",name="layer3")
    #定义网络的前向传播
    def call(self,inputs):
        x = self.layer1(inputs)
        x = self.layer2(x)
        outputs = self.layer3(x)
        return outputs
#实例化模型
model = MyModel()
#设置输入
x = tf.ones((1,3))
y = model(x)
model.summary()

神经网络的优缺点

常见的损失函数

交叉熵损失函数

import tensorflow as tf

#设置真实值和预测值
y_true = [[0,1,0],[0,0,1]]
y_pre = [[0.05,0.9,0.05],[0.05,0.05,0.9]]
#实例化交叉熵损失
cce=tf.keras.losses.CategoricalCrossentropy()
#计算损失结果
print(cce(y_true,y_pre))

二分类的交叉熵损失函数

import tensorflow as tf

#设置真实值和预测值
y_true = [[0],[1]]
y_pre = [[0.1],[0.9]]
#实例化交叉熵损失
bce=tf.keras.losses.BinaryCrossentropy()
#计算损失结果
print(bce(y_true,y_pre))

MAE损失(L1 LOSS)

import tensorflow as tf

#设置真实值和预测值
y_true = [[0],[1]]
y_pre = [[1.],[0.]]
#实例化MAE损失
mae = tf.keras.losses.MeanAbsoluteError()
mae(y_true,y_pre)

MSE损失(L2 LOSS)

smoth L1损失

梯度下降算法

反向传播算法(BP算法)

动量梯度下降算法

AdaGrad

RMSprop

Adam算法

学习率退火

深度学习正则化

L1、L2正则化


Dropout正则化


提前停止