Softmax Regression with MNIST

本文通过搭建Softmax Regression，并用MNIST数据集进行训练以及测试，介绍tensorflow的最基础使用方式。

MNIST数据集介绍以及Softmax回归介绍参考：http://wiki.jikexueyuan.com/project/tensorflow-zh/tutorials/mnist_beginners.html

1. MNIST数据集导入

   通过调用read_data_sets()，第一个参数填MNIST数据集存储路径，函数会自动判断当前路径下是否下载好数据，是否需要重新下载。

   import tensorflow.examples.tutorials.mnist.input_data as input_data
   mnist = input_data.read_data_sets("MNIST_data",one_hot=True)

2. Softmax回归模型搭建

   
   # Create the model
   
   
   #通过操作符号变量创建一个可交互的操作单元
   
   x = tf.placeholder(dtype=tf.float32, shape=[None, 784])
   
   #权重值和偏置量的创建
   
   w = tf.Variable(tf.zeros(shape=[784, 10]))
   b = tf.Variable(tf.zeros(shape=[10]))
   
   #Softmax模型创建
   
   y = tf.matmul(x, w) + b;
   
   # Define loss and optimizer
   
   y_ = tf.placeholder(tf.float32, [None, 10])

3. 训练模型存储

   
   #模型启动
   
   sess = tf.InteractiveSession()
   saver=tf.train.Saver()
   def train():
     #交叉熵计算
     cross_entropy = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=y_, logits=y))
     #执行反向传播
     train_step = tf.train.GradientDescentOptimizer(0.5).minimize(cross_entropy)
     tf.global_variables_initializer().run()
     # Train
     for _ in range(1000):
       batch_xs, batch_ys = mnist.train.next_batch(100)
       sess.run(train_step, feed_dict={x: batch_xs, y_: batch_ys})
   
   train()
   
   #模型存储，默认存储路径为工程同目录下文件夹
   
   saver.save(sess,save_path='./model/mnistmodel.ckpt')

4. 训练模型载入

   载入模型时，必须先完整还原网络结构的所有参数

   import tensorflow as tf
   import numpy as np
   import tensorflow.examples.tutorials.mnist.input_data as input_data
   
   mnist = input_data.read_data_sets("MNIST_data",one_hot=True)
   myGraph = tf.Graph()
   
   #还原网络结构
   
   x = tf.placeholder(dtype=tf.float32, shape=[None, 784])
   w = tf.Variable(tf.zeros(shape=[784, 10]))
   b = tf.Variable(tf.zeros(shape=[10]))
   y = tf.matmul(x, w) + b;
   
   # Define loss and optimizer
   
   y_ = tf.placeholder(tf.float32, [None, 10])
   
   
   # Test trained model
   
   
   #提取变量
   
   saver = tf.train.Saver()
   with tf.Session() as sess:
       saver.restore(sess,'model/mnistmodel.ckpt')
       print('Weight:\n',sess.run(w))
       print('biases:\n',sess.run(b))
       #test
       correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(y_, 1))
       accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
       print(sess.run(accuracy, feed_dict={x: mnist.test.images, y_: mnist.test.labels}))
   _: mnist.test.labels}))