first code

Author: kvfrans

input_data.py

@@ -0,0 +1,162 @@
+"""Functions for downloading and reading MNIST data."""
+import gzip
+import os
+import urllib
+import numpy
+SOURCE_URL = 'http://yann.lecun.com/exdb/mnist/'
+
+
+def maybe_download(filename, work_directory):
+    """Download the data from Yann's website, unless it's already here."""
+    if not os.path.exists(work_directory):
+        os.mkdir(work_directory)
+    filepath = os.path.join(work_directory, filename)
+    if not os.path.exists(filepath):
+        filepath, _ = urllib.urlretrieve(SOURCE_URL + filename, filepath)
+        statinfo = os.stat(filepath)
+        print 'Succesfully downloaded', filename, statinfo.st_size, 'bytes.'
+    return filepath
+
+
+def _read32(bytestream):
+    dt = numpy.dtype(numpy.uint32).newbyteorder('>')
+    return numpy.frombuffer(bytestream.read(4), dtype=dt)
+
+
+def extract_images(filename):
+    """Extract the images into a 4D uint8 numpy array [index, y, x, depth]."""
+    print 'Extracting', filename
+    with gzip.open(filename) as bytestream:
+        magic = _read32(bytestream)
+        if magic != 2051:
+            raise ValueError(
+                'Invalid magic number %d in MNIST image file: %s' %
+                (magic, filename))
+        num_images = _read32(bytestream)
+        rows = _read32(bytestream)
+        cols = _read32(bytestream)
+        buf = bytestream.read(rows * cols * num_images)
+        data = numpy.frombuffer(buf, dtype=numpy.uint8)
+        data = data.reshape(num_images, rows, cols, 1)
+        return data
+
+
+def dense_to_one_hot(labels_dense, num_classes=10):
+    """Convert class labels from scalars to one-hot vectors."""
+    num_labels = labels_dense.shape[0]
+    index_offset = numpy.arange(num_labels) * num_classes
+    labels_one_hot = numpy.zeros((num_labels, num_classes))
+    labels_one_hot.flat[index_offset + labels_dense.ravel()] = 1
+    return labels_one_hot
+
+
+def extract_labels(filename, one_hot=False):
+    """Extract the labels into a 1D uint8 numpy array [index]."""
+    print 'Extracting', filename
+    with gzip.open(filename) as bytestream:
+        magic = _read32(bytestream)
+        if magic != 2049:
+            raise ValueError(
+                'Invalid magic number %d in MNIST label file: %s' %
+                (magic, filename))
+        num_items = _read32(bytestream)
+        buf = bytestream.read(num_items)
+        labels = numpy.frombuffer(buf, dtype=numpy.uint8)
+        if one_hot:
+            return dense_to_one_hot(labels)
+        return labels
+
+
+class DataSet(object):
+    def __init__(self, images, labels, fake_data=False):
+        if fake_data:
+            self._num_examples = 10000
+        else:
+            assert images.shape[0] == labels.shape[0], (
+                "images.shape: %s labels.shape: %s" % (images.shape,
+                                                       labels.shape))
+            self._num_examples = images.shape[0]
+            # Convert shape from [num examples, rows, columns, depth]
+            # to [num examples, rows*columns] (assuming depth == 1)
+            assert images.shape[3] == 1
+            images = images.reshape(images.shape[0],
+                                    images.shape[1] * images.shape[2])
+            # Convert from [0, 255] -> [0.0, 1.0].
+            images = images.astype(numpy.float32)
+            images = numpy.multiply(images, 1.0 / 255.0)
+        self._images = images
+        self._labels = labels
+        self._epochs_completed = 0
+        self._index_in_epoch = 0
+
+    @property
+    def images(self):
+        return self._images
+
+    @property
+    def labels(self):
+        return self._labels
+
+    @property
+    def num_examples(self):
+        return self._num_examples
+
+    @property
+    def epochs_completed(self):
+        return self._epochs_completed
+
+    def next_batch(self, batch_size, fake_data=False):
+        """Return the next `batch_size` examples from this data set."""
+        if fake_data:
+            fake_image = [1.0 for _ in xrange(784)]
+            fake_label = 0
+            return [fake_image for _ in xrange(batch_size)], [
+                fake_label for _ in xrange(batch_size)]
+        start = self._index_in_epoch
+        self._index_in_epoch += batch_size
+        if self._index_in_epoch > self._num_examples:
+            # Finished epoch
+            self._epochs_completed += 1
+            # Shuffle the data
+            perm = numpy.arange(self._num_examples)
+            numpy.random.shuffle(perm)
+            self._images = self._images[perm]
+            self._labels = self._labels[perm]
+            # Start next epoch
+            start = 0
+            self._index_in_epoch = batch_size
+            assert batch_size <= self._num_examples
+        end = self._index_in_epoch
+        return self._images[start:end], self._labels[start:end]
+
+
+def read_data_sets(train_dir, fake_data=False, one_hot=False):
+    class DataSets(object):
+        pass
+    data_sets = DataSets()
+    if fake_data:
+        data_sets.train = DataSet([], [], fake_data=True)
+        data_sets.validation = DataSet([], [], fake_data=True)
+        data_sets.test = DataSet([], [], fake_data=True)
+        return data_sets
+    TRAIN_IMAGES = 'train-images-idx3-ubyte.gz'
+    TRAIN_LABELS = 'train-labels-idx1-ubyte.gz'
+    TEST_IMAGES = 't10k-images-idx3-ubyte.gz'
+    TEST_LABELS = 't10k-labels-idx1-ubyte.gz'
+    VALIDATION_SIZE = 5000
+    local_file = maybe_download(TRAIN_IMAGES, train_dir)
+    train_images = extract_images(local_file)
+    local_file = maybe_download(TRAIN_LABELS, train_dir)
+    train_labels = extract_labels(local_file, one_hot=one_hot)
+    local_file = maybe_download(TEST_IMAGES, train_dir)
+    test_images = extract_images(local_file)
+    local_file = maybe_download(TEST_LABELS, train_dir)
+    test_labels = extract_labels(local_file, one_hot=one_hot)
+    validation_images = train_images[:VALIDATION_SIZE]
+    validation_labels = train_labels[:VALIDATION_SIZE]
+    train_images = train_images[VALIDATION_SIZE:]
+    train_labels = train_labels[VALIDATION_SIZE:]
+    data_sets.train = DataSet(train_images, train_labels)
+    data_sets.validation = DataSet(validation_images, validation_labels)
+    data_sets.test = DataSet(test_images, test_labels)
+    return data_sets

main.py

@@ -0,0 +1,105 @@
+import tensorflow as tf
+import numpy as np
+import input_data
+import matplotlib.pyplot as plt
+import os
+from scipy.misc import imsave as ims
+
+mnist = input_data.read_data_sets("MNIST_data/", one_hot=True)
+n_samples = mnist.train.num_examples
+
+n_hidden = 500
+n_z = 20
+batchsize = 100
+
+def merge(images, size):
+    h, w = images.shape[1], images.shape[2]
+    img = np.zeros((h * size[0], w * size[1]))
+
+    for idx, image in enumerate(images):
+        i = idx % size[1]
+        j = idx / size[1]
+        img[j*h:j*h+h, i*w:i*w+w] = image
+
+    return img
+
+# encoder
+def recognition(input_images):
+    with tf.variable_scope("recognition"):
+        w1 = tf.get_variable("w1",[784,n_hidden])
+        b1 = tf.get_variable("b1",[n_hidden])
+        w2 = tf.get_variable("w2",[n_hidden,n_hidden])
+        b2 = tf.get_variable("b2",[n_hidden])
+        w_mean = tf.get_variable("w_mean",[n_hidden,n_z])
+        b_mean = tf.get_variable("b_mean",[n_z])
+        w_stddev = tf.get_variable("w_stddev",[n_hidden,n_z])
+        b_stddev = tf.get_variable("b_stddev",[n_z])
+
+    h1 = tf.nn.sigmoid(tf.matmul(input_images,w1) + b1)
+    h2 = tf.nn.sigmoid(tf.matmul(h1,w2) + b2)
+    o_mean = tf.matmul(h2,w_mean) + b_mean
+    o_stddev = tf.matmul(h2,w_stddev) + b_stddev
+    return o_mean, o_stddev
+
+# decoder
+def generation(z):
+    with tf.variable_scope("generation"):
+        w1 = tf.get_variable("w1",[n_z,n_hidden])
+        b1 = tf.get_variable("b1",[n_hidden])
+        w2 = tf.get_variable("w2",[n_hidden,n_hidden])
+        b2 = tf.get_variable("b2",[n_hidden])
+        w_image = tf.get_variable("w_image",[n_hidden,784])
+        b_image = tf.get_variable("b_image",[784])
+
+    h1 = tf.nn.sigmoid(tf.matmul(z,w1) + b1)
+    h2 = tf.nn.sigmoid(tf.matmul(h1,w2) + b2)
+    o_image = tf.nn.sigmoid(tf.matmul(h2,w_image) + b_image)
+    return o_image
+
+images = tf.placeholder(tf.float32, [None, 784])
+# instead of mapping directly to z, map a gaussian over z, parameterized by mean/stddev
+# important: z_stddev contains log(standard_deviation^2).
+z_mean, z_stddev = recognition(images)
+print z_mean.get_shape()
+
+# unit guassian
+samples = tf.random_normal([batchsize,n_z],0,1,dtype=tf.float32)
+guessed_z = z_mean + (z_stddev * samples)
+
+generated_images = generation(guessed_z)
+
+# -log of p(x|z)
+generation_loss = -tf.reduce_sum(images * tf.log(1e-10 + generated_images) + (1-images) * tf.log(1e-10 + 1 - generated_images),1)
+
+# we want real p(z) to be unit guassian
+# the KL divergence loss between real p(z) and q(z|x)
+# q(z|x) is the recognition network
+# latent_loss = 0.5 * tf.reduce_sum(tf.square(z_mean) + tf.exp(z_stddev) - z_stddev - 1,1)
+latent_loss = 0.5 * tf.reduce_sum(tf.square(z_mean) + tf.square(z_stddev) - tf.log(tf.square(z_stddev)) - 1,1)
+
+cost = tf.reduce_mean(generation_loss + latent_loss)
+optimizer = tf.train.AdamOptimizer(0.001).minimize(cost)
+
+train = True
+visualization = mnist.train.next_batch(batchsize)[0]
+reshaped_vis = visualization.reshape(batchsize,28,28)
+ims("results/base.jpg",merge(reshaped_vis[:64],[8,8]))
+# train
+saver = tf.train.Saver(max_to_keep=2)
+with tf.Session() as sess:
+    if train:
+        sess.run(tf.initialize_all_variables())
+        for epoch in range(10):
+            for idx in range(int(n_samples / batchsize)):
+                batch = mnist.train.next_batch(batchsize)[0]
+                _, real_cost = sess.run((optimizer, cost), feed_dict={images: batch})
+                # dumb hack to print cost every epoch
+                if idx % (n_samples - 3) == 0:
+                    print "%d: %f" % (epoch, real_cost)
+                    saver.save(sess, os.getcwd()+"/training/train",global_step=epoch)
+                    generated_test = sess.run(generated_images, feed_dict={images: visualization})
+                    generated_test = generated_test.reshape(batchsize,28,28)
+                    ims("results/"+str(epoch)+".jpg",merge(generated_test[:64],[8,8]))
+    else:
+        saver.restore(sess, tf.train.latest_checkpoint(os.getcwd()+"/training/"))
+        batch = mnist.train.next_batch(batchsize)[0]