# TensorFlow: Applying Linear Regression on MNIST Dataset

LinearRegression_MNIST

# TensorFlow Linear Regression on MNIST Dataset¶

With one hidden layer and Softmax classifier, how good can we achieve on MNIST dataset?

It turns out not too bad, 90% accuracy!

In [1]:
import sys
import time
import numpy as np
import tensorflow as tf
import matplotlib.pyplot as plt
from tensorflow.examples.tutorials.mnist import input_data
%matplotlib inline


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 [2]:
# Define parameters for linear model
learning_rate = 0.01
batch_size = 128
n_epochs = 25

# Create placeholders
X = tf.placeholder(tf.float32, [batch_size, 784], name="image")
Y = tf.placeholder(tf.float32, [batch_size, 10], name="label")

# Create weights and bias
w = tf.Variable(tf.random_normal(shape=[784, 10], stddev=0.01), name="weights")
b = tf.Variable(tf.zeros([1,10]), name='bias')

# calculate scores
logits = tf.matmul(X, w) + b

# Entropy cost function and loss
entropy = tf.nn.softmax_cross_entropy_with_logits(logits, Y)
loss = tf.reduce_mean(entropy)

# Define optimizer

In [3]:
# Run optimization and test
loss_history = []
acc_history = []
init = tf.global_variables_initializer()
with tf.Session() as sess:
sess.run(init)
n_batches = int(MNIST.train.num_examples/batch_size)
for i in range(n_epochs):
for _ in range(n_batches):
X_batch, Y_batch = MNIST.train.next_batch(batch_size)
_, loss_value = sess.run([optimizer, loss], feed_dict={X: X_batch, Y:Y_batch})
loss_history.append(loss_value)

# Check validation accuracy
n_v_batches = int(MNIST.validation.num_examples/batch_size)
total_correct_preds = 0
for j in range(n_v_batches):
X_batch, Y_batch = MNIST.validation.next_batch(batch_size)
_, loss_batch, logits_batch = sess.run([optimizer, loss, logits], feed_dict={X: X_batch, Y:Y_batch})
preds = tf.nn.softmax(logits_batch)
correct_preds = tf.equal(tf.argmax(preds, 1), tf.argmax(Y_batch, 1))
accuracy = tf.reduce_sum(tf.cast(correct_preds, tf.float32))
total_correct_preds += sess.run(accuracy)
validation_accuracy = total_correct_preds/MNIST.validation.num_examples
acc_history.append(validation_accuracy)

# Test the model
n_batches = int(MNIST.test.num_examples/batch_size)
total_correct_preds = 0
for i in range(n_batches):
X_batch, Y_batch = MNIST.test.next_batch(batch_size)
logits_batch = sess.run(logits, feed_dict={X: X_batch, Y:Y_batch})
preds = tf.nn.softmax(logits_batch)
correct_preds = tf.equal(tf.argmax(preds, 1), tf.argmax(Y_batch, 1))
accuracy = tf.reduce_sum(tf.cast(correct_preds, tf.float32))
total_correct_preds += sess.run(accuracy)

print "Test accuracy is {0}".format(total_correct_preds/MNIST.test.num_examples)

Test accuracy is 0.9102

In [4]:
plt.subplot(2,1,1)
plt.plot(loss_history, '-o', label='Loss value')
plt.title('Training Loss')
plt.xlabel('Epoch')
plt.ylabel('Loss Value')
plt.legend(ncol=2, loc='upper right')
plt.subplot(2,1,2)
plt.plot(acc_history, '-o', label='Accuracy value')
plt.title('Training Accuracy')
plt.xlabel('Epoch')
plt.ylabel('Accuracy')
plt.legend(ncol=2, loc='lower right')
plt.gcf().set_size_inches(10, 10)
plt.show()

In [ ]:



Published: February 01 2017

• category: