In [6]:

    

'''Softmax linear regression'''


import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data


    

In [7]:

    

mnist = input_data.read_data_sets("MNIST_data/", one_hot = True)


    

    




Extracting MNIST_data/train-images-idx3-ubyte.gz
Extracting MNIST_data/train-labels-idx1-ubyte.gz
Extracting MNIST_data/t10k-images-idx3-ubyte.gz
Extracting MNIST_data/t10k-labels-idx1-ubyte.gz

In [8]:

    

#will hold the input data
x = tf.placeholder(tf.float32, [None, 784])


    

In [10]:

    

#Parameters (Variables) to be tuned
W = tf.Variable(tf.zeros([784, 10]))
b = tf.Variable(tf.zeros([10]))


    

In [12]:

    

#create the model
y = tf.nn.softmax(tf.matmul(x, W) + b)


    

In [13]:

    

#placeholder will hold the input data
y_ = tf.placeholder(tf.float32, [None, 10])


    

In [19]:

    

#define the cross entroy cost function
cross_entropy = tf.reduce_mean(-tf.reduce_sum(y_ * tf.log(y), reduction_indices=[1]))


    

In [21]:

    

# define the optimization algorithm to be used
train_step = tf.train.GradientDescentOptimizer(0.5).minimize(cross_entropy)


    

In [24]:

    

# launch the model in interactive session
sess = tf.InteractiveSession()


    

In [25]:

    

tf.global_variables_initializer().run()


    

In [38]:

    

# run the training step 1000 times!
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})


    

In [30]:

    

#define the predictions tensor
correct_predictions = tf.equal(tf.argmax(y, 1), tf.argmax(y_, 1))
#define the accuracy
accuracy = tf.reduce_mean(tf.cast(correct_predictions, tf.float32))


    

In [39]:

    

print(sess.run(accuracy, feed_dict={x: mnist.test.images, y_: mnist.test.labels}))


    

    




0.922

In [ ]: