Keras实现手写数字识别

import numpy as   np

from  keras.datasets  import mnist

from  keras.models import   Sequential

from  keras.layers   import Conv2D,MaxPooling2D,Dropout,Flatten,Dense

(X_train,y_train),(X_test,y_test) =  mnist.load_data() #加载数据




X_train  = X_train.reshape(X_train.shape[0],28,28,1).astype('float32')
X_test = X_test.reshape(X_test.shape[0],28,28,1).astype('float32')


X_train/= 255   #由于每个像素是介于0~255的  所以除以255

X_test/=255



def tran_y(y):#手动实现one-hot编码
    y_ohe = np.zeros(10)
    y_ohe[y] = 1
    return y_ohe


y_train = np.array([
    tran_y(y_train[tmp]) for tmp in range(len(y_train))  #实现对label的one-hot编码
])


y_test = np.array([

    tran_y(y_test[i]) for i in range(len(y_test))
])



#开始搭建卷积神经网络

model = Sequential()

#添加一层卷积层，构建64个filter

model.add(
    Conv2D(
        filters=64,
        kernel_size=(3,3),
        strides=(1,1),
        padding = 'same',
        input_shape=(28,28,1),
        activation='relu'
    )
)#增加一个卷积层，卷积核数为64，大小为3*3，步长为1


model.add(MaxPooling2D(pool_size=(2,2)))
#增加一个最大池化层,大小为2*2


model.add(Dropout(0.5))
#设置DropOut层



model.add(
    Conv2D(
        filters=256,
        kernel_size=(3,3),
        strides=(1,1),
        padding='same',
        activation='relu'

    )
)

model.add(MaxPooling2D(pool_size=(2,2)))

model.add(Dropout(0.5))

model.add(Flatten())#把当前层节点展平



model.add(Dense(128,activation='relu'))

model.add(Dense(64,activation='relu'))
model.add(Dense(32,activation='relu'))
model.add(Dense(10,activation='softmax'))
#增加全连接层



model.compile(loss = 'categorical_crossentropy',optimizer='adagrad',metrics=['accuracy'])
#最后定义损失函数



model.fit(X_train,y_train,validation_data=(X_test,y_test),epochs=20,batch_size=128)
#最后放入批量样本，进行训练




scores = model.evaluate(X_test,y_test,verbose=0)