import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
#number 1 to 10 data
mnist = input_data.read_data_sets('D:/Download/google/MNIST_data/',one_hot = True) #引用本地路径，‘\’用‘/’代替，最后加‘/’
def add_layer(inputs,in_size,out_size,activation_function = None):
Weights = tf.Variable(tf.random_normal([in_size,out_size]))
biases = tf.Variable(tf.zeros([1,out_size])+0.1)
Wx_plus_b = tf.matmul(inputs,Weights)+biases
if activation_function is None:
outputs = Wx_plus_b
else:
outputs = activation_function(Wx_plus_b)
return outputs
def compute_accuracy(v_xs,v_ys):
global prediction
y_pre = sess.run(prediction,feed_dict={xs:v_xs})
correct_prediction = tf.equal(tf.argmax(y_pre,1),tf.argmax(v_ys,1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction,tf.float32))
result = sess.run(accuracy,feed_dict={xs:v_xs,ys:v_ys})
return result
#defined placeholder for inputs to network
xs = tf.placeholder(tf.float32,[None,784])#28*28
ys = tf.placeholder(tf.float32,[None,10])#每个example有10个输出
#add output layer
prediction = add_layer(xs,784,10,activation_function = tf.nn.softmax)
#the error between prediction and real data
cross_entropy = tf.reduce_mean(-tf.reduce_sum(ys * tf.log(prediction),reduction_indices = [1]))#loss
#training
train_step = tf.train.GradientDescentOptimizer(0.1).minimize(cross_entropy)
init = tf.global_variables_initializer()
sess =tf.Session()
#important step
sess.run(init)
for i in range(1000):
batch_xs,batch_ys = mnist.train.next_batch(100)
sess.run(train_step,feed_dict={xs:batch_xs,ys:batch_ys})
if i%50 == 0:
print(compute_accuracy(mnist.test.images,mnist.test.labels))
实现的效果如图：
