Deep CNN training.py

# Computes softmax cross entropy between logits and labels
# Measures the probability error in discrete classification tasks
# For example, each font image is labeled with one and only one label: an image can be font SansSerif or Serif, but not both.
cross_entropy = tf.reduce_mean(
        tf.nn.softmax_cross_entropy_with_logits_v2(logits = y + 1e-50, labels = y_))

# Applies exponential decay to the learning rate!!!
learning_rate = tf.train.exponential_decay(0.05, global_step, 1000, 0.85, staircase=True)

# GradientDescentOptimizer is used to minimize loss
train_step = tf.train.GradientDescentOptimizer(
             learning_rate).minimize(cross_entropy, global_step=global_step)

# Define accuracy
correct_prediction = tf.equal(tf.argmax(y,1),
                              tf.argmax(y_,1))
accuracy = tf.reduce_mean(tf.cast(
           correct_prediction, "float
           
# Train for 5000 times
epochs = 5000
train_acc = np.zeros(epochs//10)
test_acc = np.zeros(epochs//10)
for i in tqdm(range(epochs), ascii=True):
    # Record summary data, and the accuracy
    if i % 10 == 0:  
        # Check accuracy on train set
        A = accuracy.eval(feed_dict={x: train_dataset, y_: train_labels})
        train_acc[i//10] = A

        # And now the test set
        A = accuracy.eval(feed_dict={x: test_dataset, y_: test_labels})
        test_acc[i//10] = A
        
    train_step.run(feed_dict={x: train_dataset, y_: train_labels})