Tensorflow - Super Simple Neural Network
Description, Resources, Code:
Straight-forward, short and clear neural network in Tensorflow.
Ideal if you already have basic knowledge on neural nets.
Using Python 3.5 and Tensorflow 1.4
git repo incl code and images:
https://bitbucket.org/sinamackay/tens...
Useful resources:
https://www.tensorflow.org/get_starte...
https://www.tensorflow.org/get_starte...
Code:
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
mnist = input_data.read_data_sets('MNIST_data', one_hot=True)
n_input = mnist.train.images.shape[1] # 28x28 = 784
n_hidden = 500 # hidden layer n neurons
n_classes = 10 # digits 0-9
def mlp_config(n_input, n_hidden, n_classes):
x = tf.placeholder("float", [None, n_input], name='x')
y = tf.placeholder("float", [None, n_classes], name='y')
weights = {
'h1': tf.Variable(tf.random_normal([n_input, n_hidden])),
'out': tf.Variable(tf.random_normal([n_hidden, n_classes]))
}
return x, y, weights
def mlp_model(x, y, weights):
hidden = tf.nn.relu(tf.matmul(x, weights['h1']))
logits = tf.matmul(hidden, weights['out'])
pred = tf.one_hot(tf.cast(tf.argmax(logits, 1), tf.int32), depth=10)
return pred, logits
def get_loss(logits, y):
loss = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits_v2(labels=y, logits=logits))
return loss
def get_accuracy(pred, y):
correct_prediction = tf.equal(tf.argmax(pred, 1), tf.argmax(y,1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
return accuracy
def main():
x, y, weights = mlp_config(n_input, n_hidden, n_classes)
pred, logits = mlp_model(x, y, weights)
loss = get_loss(logits, y)
accuracy = get_accuracy(pred, y)
train_step = tf.train.GradientDescentOptimizer(learning_rate=.01).minimize(loss)
sess = tf.Session()
init = tf.global_variables_initializer()
sess.run(init)
for i in range(1000):
batch = mnist.train.next_batch(1000) # fetch batch of size 1000
if (i% 100) == 0:
acc = sess.run(accuracy, feed_dict={x:batch[0], y:batch[1]})
print('test accuracy at step %s: %s' % (i, acc))
else:
sess.run(train_step, feed_dict={x:batch[0], y:batch[1]})
print("Accuracy using tensorflow is: ")
print(sess.run(accuracy, feed_dict = {x: mnist.test.images, y: mnist.test.labels}))
main()