CNN网络编译和训练
代码：

将数据扩充维度，以适应CNN模型

X_train=x_train.reshape(60000,28,28,1)
X_test=x_test.reshape(10000,28,28,1)
model.compile(optimizer=tf.train.AdamOptimizer(),loss="categorical_crossentropy",metrics=['accuracy'])
model.fit(x=X_train,y=y_train,epochs=5,batch_size=128)
输出：
Epoch 1/5
55000/55000 [==============================] - 49s 899us/sample - loss: 0.2107 - acc: 0.9348
Epoch 2/5
55000/55000 [==============================] - 48s 877us/sample - loss: 0.0793 - acc: 0.9763
Epoch 3/5
55000/55000 [==============================] - 52s 938us/sample - loss: 0.0617 - acc: 0.9815
Epoch 4/5
55000/55000 [==============================] - 48s 867us/sample - loss: 0.0501 - acc: 0.9846
Epoch 5/5
55000/55000 [==============================] - 50s 901us/sample - loss: 0.0452 - acc: 0.9862

在训练时，网络训练数据只迭代了5次，可以再增加网络迭代次数，自行尝试看效果如何。