深度学习入门--手写数字识别（Keras）

89 阅读 0 评论 59 点赞

我是靠谱客的博主缓慢冷风，这篇文章主要介绍深度学习入门--手写数字识别（Keras），现在分享给大家，希望可以做个参考。

复制代码

1
2
3
4
5
6
7
8
9
10
11
12
13
#导入所需要的包
import keras
from keras.datasets import mnist
from keras.layers import Dense 
from keras.models import Sequential
from keras.optimizers import SGD

#下载数据集
(x_train,y_train),(x_test,y_test) = mnist.load_data()

#查看数据的size
print(x_train.shape, y_train.shape)
print(x_test.shape, y_test.shape)

此时输出的结果为：

复制代码

1
2
3
4
#随机查看一下数据内容
import matplotlib.pyplot as plt
im = plt.imshow(x_train[0],cmap="gray")
plt.show()

输出为：

复制代码

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
#向量化图片，将28*28转化成1*784
x_train = x_train.reshape(60000,784)
x_test = x_test.reshape(10000,784)


#数据归一化
x_train = x_train/255
x_test = x_test/255


#将标签值y进行独热编码 keras.utils.to_categorical（）
y_train = keras.utils.to_categorical(y_train,10)
y_test = keras.utils.to_categorical(y_test,10)

#开始构建模型
model = Sequential()
model.add(Dense(512, activation='relu', input_shape=(784,)))
model.add(Dense(256, activation='relu'))
model.add(Dense(10, activation='softmax'))


#查看模型情况
model.summary()

输出的模型概况为：

复制代码

1
2
3
4
5
#编译神经网络
model.compile(optimizer=SGD(), loss='categorical_crossentropy', metrics=['accuracy'])

#训练网络
model.fit(x_train, y_train, batch_size=64, epochs=10, validation_data=(x_test, y_test))

执行结果：

复制代码

1
2
3
4
#查看最终结果
score = model.evaluate(x_test, y_test)
print("LOSS:",score[0])
print("ACCU:",score[1])

输出结果为：

完整版程序：

复制代码

#导入所需要的包
import keras
from keras.datasets import mnist
from keras.layers import Dense 
from keras.models import Sequential
from keras.optimizers import SGD

#下载数据集
(x_train,y_train),(x_test,y_test) = mnist.load_data()

#查看数据的size
#此部分可省略
print(x_train.shape, y_train.shape)
print(x_test.shape, y_test.shape)

#随机查看一下图片
#此部分可省略
import matplotlib.pyplot as plt
im = plt.imshow(x_train[0],cmap="gray")
plt.show()

#向量化图片，将28*28转化成1*784
x_train = x_train.reshape(60000,784)
x_test = x_test.reshape(10000,784)

#数据归一化
x_train = x_train/255
x_test = x_test/255

#将标签值y进行独热编码 keras.utils.to_categorical（）
y_train = keras.utils.to_categorical(y_train,10)
y_test = keras.utils.to_categorical(y_test,10)

#开始构建模型
model = Sequential()
model.add(Dense(512, activation='relu', input_shape=(784,)))
model.add(Dense(256, activation='relu'))
model.add(Dense(10, activation='softmax'))

#查看模型情况
#此部分可省略
model.summary()

#编译神经网络
model.compile(optimizer=SGD(), loss='categorical_crossentropy', metrics=['accuracy'])

#训练网络
model.fit(x_train, y_train, batch_size=64, epochs=10, validation_data=(x_test, y_test))

#查看结果
score = model.evaluate(x_test, y_test)
print("LOSS:",score[0])
print("ACCU:",score[1])

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
#导入所需要的包
import keras
from keras.datasets import mnist
from keras.layers import Dense 
from keras.models import Sequential
from keras.optimizers import SGD

#下载数据集
(x_train,y_train),(x_test,y_test) = mnist.load_data()

#查看数据的size
#此部分可省略
print(x_train.shape, y_train.shape)
print(x_test.shape, y_test.shape)


#随机查看一下图片
#此部分可省略
import matplotlib.pyplot as plt
im = plt.imshow(x_train[0],cmap="gray")
plt.show()


#向量化图片，将28*28转化成1*784
x_train = x_train.reshape(60000,784)
x_test = x_test.reshape(10000,784)


#数据归一化
x_train = x_train/255
x_test = x_test/255


#将标签值y进行独热编码 keras.utils.to_categorical（）
y_train = keras.utils.to_categorical(y_train,10)
y_test = keras.utils.to_categorical(y_test,10)

#开始构建模型
model = Sequential()
model.add(Dense(512, activation='relu', input_shape=(784,)))
model.add(Dense(256, activation='relu'))
model.add(Dense(10, activation='softmax'))


#查看模型情况
#此部分可省略
model.summary()


#编译神经网络
model.compile(optimizer=SGD(), loss='categorical_crossentropy', metrics=['accuracy'])


#训练网络
model.fit(x_train, y_train, batch_size=64, epochs=10, validation_data=(x_test, y_test))

#查看结果
score = model.evaluate(x_test, y_test)
print("LOSS:",score[0])
print("ACCU:",score[1])