1.神经网络搭建八股
2.iris代码复现
3.mnist数据集
4.训练mnist数据集
5.fashion数据集
搭建神经网络八股六步法:
1.import:导入模块
2.train,test:告知训练集和测试集
3.model=tf.keras.models.Sequential:在Sequential中搭建网络,走一遍前向传播
4.model.compile:在compile()中配置训练方法,选择优化器,选择损失函数,选择评测指标
5.model.fit:在fit()中训练,训练集和测试集喂入网络,告知每个batch是多少,要迭代多少epochs
6.model.summary:打印网络的结构和参数统计
model=tf.keras.models.Sequential([网络结构]) #描述各层网络
拉直层:tf.keras.layers.Flatten() 不含计算,把输入特征拉直成一维数组
全连接层:tf.keras.layers.Dense(神经元个数,activation=“激活函数”,kernel_regularizer=哪种正则化)
activation(字符串给出)可选:relu,softmax,sigmoid,tanh
kernel_regularizer可选:tf.keras.regularizer.l1()、tf.keras.regularizer.l2()
卷积层:tf.keras.layers.Conv2D(…)
lstm层:tf.keras.layers.LSTM()
model.compile(optimizer=优化器,loss=损失函数,metrics=[“准确率”])
optimizer可选:
'sgd’或tf.kears.optimizers.SGD(lr=学习率,momentum=动量参数)
'adagrad’或tf.kears.optimizers.Adagrad(lr=学习率)
'adadelta’或tf.kears.optimizers.Adadelta(lr=学习率)
'adam’或tf.kears.optimizers.Adam(lr=学习率,beta_1=0.9,beta_2=0.999)
loss可选:
'mse’或tf.kears.losses.MeanSquaredError()
'sparse_categorical_crossentropy’或tf.kears.losses.SparseCategoricalCrossentropy(from_logits=False)
Metrics可选:
‘accuracy’,‘categorical_accuracy’(y和y_是独热编码),‘sparse_categorical_accuracy’(y_是数值,y是独热编码)
model.fit(训练集的输入特征,训练集的标签,batch_size= ,epochs= ,validation_data=(测试集的输入特征,测试集的标签),validation_split=从训练集划分多少比例给测试集,validation_freq=多少次epoch使用测试集验证一次)
validation_data和validation_split二选一
model.summary():打印网络结构
自定义网络结构搭建神经网络六步法:
1.import:导入模块
2.train,test:告知训练集和测试集
3.class MyModel(Model),model=MyModel() (自定义网络结构)
4.model.compile:在compile()中配置训练方法,选择优化器,选择损失函数,选择评测指标
5.model.fit:在fit()中训练,训练集和测试集喂入网络,告知每个batch是多少,要迭代多少epochs
6.model.summary:打印网络的结构和参数统计
class MyModel(Model):
def __init__(self):
super(MyModel, self).__init__()
# 定义网络结构块
def call(self,x):
# 调用网络结构块,实现前向传播
return y
model = MyModel() # 实例化一个model
如
class MyModel(Model):
def __init__(self):
super(MyModel, self).__init__()
self.d1 = Dense(3) # d1是这一层起的名字,每一层用self.引导
def call(self,x):
y = self.d1(x) # 调用网络结构,实现一次前向传播
return y
model = MyModel() # 实例化一个model
案例:使用mnist数据集进行识别
六步法:
import tensorflow as tf
fashion = tf.keras.datasets.fashion_mnist
(x_train, y_train),(x_test, y_test) = fashion.load_data()
x_train, x_test = x_train / 255.0, x_test / 255.0
model = tf.keras.models.Sequential([
tf.keras.layers.Flatten(),
tf.keras.layers.Dense(128, activation='relu'),
tf.keras.layers.Dense(10, activation='softmax')
])
model.compile(optimizer='adam',
loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=False),
metrics=['sparse_categorical_accuracy'])
model.fit(x_train, y_train, batch_size=32, epochs=5, validation_data=(x_test, y_test), validation_freq=1)
model.summary()
自定义网络结构六步法:
import tensorflow as tf
from tensorflow.keras.layers import Flatten, Dense
from tensorflow.keras.models import Model
(x_train, y_train), (x_test, y_test) = tf.keras.datasets.fashion_mnist.load_data()
x_train, x_test = x_train/255.0, x_test/255.0
class FashionMedol(Model):
def __init__(self):
super(FashionMedol, self).__init__()
self.flatten = Flatten()
self.dense1 = Dense(128, activation='relu')
self.dense2 = Dense(10, activation='softmax')
def call(self, x):
x = self.flatten(x)
x = self.dense1(x)
y = self.dense2(x)
return y
model = FashionMedol()
model.compile(optimizer='adam',
loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=False),
metrics=['sparse_categorical_accuracy'])
model.fit(x_train, y_train, batch_size=32, epochs=5, validation_data=(x_test, y_test), validation_freq=1)
model.summary()