Tensorflow - Implement for a Softmax Regression Model on MNIST.

Coding according to TensorFlow 官方文档中文版

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

 ''' Intro. for this python file.
 Objective:
     Implement for a Softmax Regression Model on MNIST.
 Operating Environment:
     python = 3.6.4
     tensorflow = 1.5.0
 '''

 # Set a placeholder. We hope arbitrary number of images could be input to this model.
 x = tf.placeholder("float", [None, 784])

 # Set weight/bias variables. Their initial values could be set Randomly.
 W = tf.Variable(tf.zeros([784, 10]))
 b = tf.Variable(tf.zeros([10]))

 # Model implementation
 y = tf.nn.softmax(tf.matmul(x, W) + b)

 # Set a placeholder 'y_' to accept the ground-truth values.
 y_ = tf.placeholder("float", [None, 10])

 # Calculate cross-entropy
 cross_entropy = -tf.reduce_sum(y_ * tf.log(y))

 # Train Softmax Regression Model
 train_step = tf.train.GradientDescentOptimizer(0.01).minimize(cross_entropy)

 # Initialize variables
 # init = tf.initialize_all_variables()  # Warning
 init = tf.global_variables_initializer()

 # Launch the graph in a session.
 sess = tf.Session()
 sess.run(init)

 for i in range(1000):
     batch_xs, batch_ys = mnist.train.next_batch(100)    # Grabbing 100 batch data points from training data randomly.
     sess.run(train_step, feed_dict={x: batch_xs, y_: batch_ys})

 # Model Evaluation
 ''' tf.argmax(input, axis=None, name=None, dimension=None, output_type=tf.int64)
 Explanation:
     Returns the index with the largest value across axes of a tensor.
     test = np.array([[1, 2, 3], [2, 3, 4], [5, 4, 3], [8, 7, 2]])
     np.argmax(test, 0)      # output：array([3, 3, 1])
     np.argmax(test, 1)      # output：array([2, 2, 0, 0])
 Returns:
     A Tensor of type output_type.
 '''

 # correct_prediction = tf.equal(tf.arg_max(y, 1), tf.arg_max(y_, 1))    # Warning
 correct_prediction = tf.equal(tf.argmax(y, axis=1), tf.argmax(y_, axis=1))
 accuracy = tf.reduce_mean(tf.cast(correct_prediction, "float"))
 print(sess.run(accuracy, feed_dict={x: mnist.test.images, y_: mnist.test.labels}))

 # The result is around 0.91.