Created
September 17, 2019 03:04
-
-
Save mwdchang/50a165bc23c55aa7c84a338e21278b08 to your computer and use it in GitHub Desktop.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| import tensorflow as tf | |
| import numpy as np | |
| import matplotlib.pyplot as plt | |
| import random | |
| from tensorflow.examples.tutorials.mnist import input_data | |
| mnist = input_data.read_data_sets("./MNIST/", one_hot=True) | |
| tf.reset_default_graph() | |
| num_inputs = 784 # 28*28 | |
| neurons_hid1 = 39 | |
| neurons_hid2 = 196 | |
| neurons_hid3 = neurons_hid1 # Decoder Begins | |
| num_outputs = num_inputs | |
| learning_rate = 0.01 | |
| X = tf.placeholder(tf.float32, shape=[None, num_inputs]) | |
| XPrime = tf.placeholder(tf.float32, shape=[None, num_inputs]) | |
| initializer = tf.variance_scaling_initializer() | |
| # Layers and architecture | |
| w1 = tf.Variable(initializer([num_inputs, neurons_hid1]), dtype=tf.float32) | |
| w2 = tf.Variable(initializer([neurons_hid1, neurons_hid2]), dtype=tf.float32) | |
| w3 = tf.Variable(initializer([neurons_hid2, neurons_hid3]), dtype=tf.float32) | |
| w4 = tf.Variable(initializer([neurons_hid3, num_outputs]), dtype=tf.float32) | |
| b1 = tf.Variable(tf.zeros(neurons_hid1)) | |
| b2 = tf.Variable(tf.zeros(neurons_hid2)) | |
| b3 = tf.Variable(tf.zeros(neurons_hid3)) | |
| b4 = tf.Variable(tf.zeros(num_outputs)) | |
| act_func = tf.nn.relu | |
| hid_layer1 = act_func(tf.matmul(X, w1) + b1) | |
| hid_layer2 = act_func(tf.matmul(hid_layer1, w2) + b2) | |
| hid_layer3 = act_func(tf.matmul(hid_layer2, w3) + b3) | |
| output_layer = tf.matmul(hid_layer3, w4) + b4 | |
| # Loss function | |
| loss = tf.reduce_mean(tf.square(output_layer - XPrime)) | |
| # loss = tf.reduce_mean(tf.square(output_layer - X)) | |
| # loss = tf.reduce_mean(tf.square(output_layer - mnist.test.images[0])) | |
| optimizer = tf.train.AdamOptimizer(learning_rate) | |
| train = optimizer.minimize(loss) | |
| # Training | |
| init = tf.global_variables_initializer() | |
| num_epochs = 5 | |
| batch_size = 150 | |
| with tf.Session() as sess: | |
| sess.run(init) | |
| for epoch in range(num_epochs): | |
| num_batches = mnist.train.num_examples // batch_size # // int division | |
| for iteration in range(num_batches): | |
| X_batch, y_batch = mnist.train.next_batch(batch_size) | |
| X2 = X_batch + (0.3 * mnist.test.images[epoch]) | |
| # X2 = X_batch + (0.8 * mnist.test.images[random.randint(1, 3)]) | |
| # X2 = X_batch + (0.5 * mnist.test.images[3]) | |
| sess.run(train, feed_dict={X: X_batch, XPrime: X2}) | |
| training_loss = loss.eval(feed_dict={X: X_batch, XPrime: X2}) | |
| print("Epoch {} Complete. Training Loss: {}".format(epoch,training_loss)) | |
| num_test_images = 10 | |
| results = output_layer.eval(feed_dict={X:mnist.test.images[:num_test_images]}) | |
| f, a = plt.subplots(2, 10, figsize=(20, 4)) | |
| for i in range(num_test_images): | |
| a[0][i].imshow(np.reshape(mnist.test.images[i], (28, 28))) | |
| a[1][i].imshow(np.reshape(results[i], (28, 28))) | |
| plt.show() | |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment