def load(depth, width, height):
'''
Loading datasets
'''
# Training Set: 60,000 datapoints - Test Set: 10,000 datapoints
print("[INFO] Downloading/fetching MNIST dataset")
dataset = input_data.read_data_sets("../data", False)
# Images
# Each image is a vector of dimension 784 (1*28*28 - one channel, grayscale images)
# Values are between 0 and 1, they are already normalized
# Note: I won't use the valisation dataset to tune hyperparameters
x_train = np.concatenate((dataset.train.images, dataset.validation.images),
axis=0)
x_test = dataset.test.images
# Labels
y_train = np.concatenate((dataset.train.labels, dataset.validation.labels),
axis=0)
y_test = dataset.test.labels
x_train = x_train.reshape(
(x_train.shape[0], depth, width, height)).astype('float32')
x_test = x_test.reshape(
(x_test.shape[0], depth, width, height)).astype('float32')
return (x_train, x_test, y_train, y_test)
from sklearn.metrics import confusion_matrix
from sklearn.metrics import precision_recall_fscore_support
from lib import input_data
from lib import graph
from lib import model
from lib import corrupt
from keras import backend as K
### Keras Backend
#print(K.backend())
#print(K.image_dim_ordering(), ": ", K.image_data_format())
### Loading data
# Training Set: 60,000 datapoints - Test Set: 10,000 datapoints
print("[INFO] Downloading/fetching MNIST dataset")
dataset = input_data.read_data_sets("../data", False)
# Images
# Each image is a vector of dimension 784 (1*28*28 - one channel, grayscale images)
# Values are between 0 and 1, they are already normalized
# Note: I won't use the valisation dataset to tune hyperparameters
x_train = np.concatenate((dataset.train.images, dataset.validation.images), axis=0)
x_test = dataset.test.images
# Labels
y_train = np.concatenate((dataset.train.labels, dataset.validation.labels), axis=0)
y_test = dataset.test.labels
### Plot Sample images (4 - grayscale)
graph.plot_sample(x_train)
### Image Definition
# number of classes
# Import MNIST data
import tensorflow as tf
import lib.input_data as input_data
from tensorflow.python.framework import ops
from tensorflow.python.saved_model import builder, utils
from tensorflow.python.saved_model import signature_constants
from tensorflow.python.saved_model import signature_def_utils
from tensorflow.python.saved_model import tag_constants
mnist = input_data.read_data_sets("tmp/data/", one_hot=True)
# Set parameters
learning_rate = 0.01
training_iteration = 30
batch_size = 100
display_step = 2
# TF graph input
x = tf.placeholder(tf.float32, shape=[None, 784],
name="x") # mnist data image of shape 28*28=784
y = tf.placeholder(tf.float32,
shape=[None, 10]) # 0-9 digits recognition => 10 classes
# Create a model
# Set model weights
W = tf.Variable(tf.zeros([784, 10]))
b = tf.Variable(tf.zeros([10]))
# Construct a linear model
model = tf.nn.softmax(tf.matmul(x, W) + b, name='y') # Softmax
def run_training():
data_sets = input_data.read_data_sets('MNIST_data',
fake_data=False) # 下载数据
# 创建训练图像shape 和标记shape
with tf.Graph().as_default():
images_placeholder = tf.placeholder(tf.float32,
shape=(batch_size,
mnist.IMAGE_PIXELS)) # 输入图
labels_placeholder = tf.placeholder(tf.int32, shape=batch_size) # 标签图
print('images_placeholder 的维度为：', images_placeholder.shape)
print('lables_placeholder 的维度为:', labels_placeholder.shape)
# 第一层
with tf.name_scope('hidden1'):
weight = tf.Variable(tf.truncated_normal(
[mnist.IMAGE_PIXELS, mnist.HIDDEN1_UNITS],
stddev=1.0 / math.sqrt(float(mnist.IMAGE_PIXELS))),
name='weights') # 初始化随机
biases = tf.Variable(tf.zeros([mnist.HIDDEN1_UNITS]),
name='biases')
hidden1 = tf.nn.relu(
tf.matmul(images_placeholder, weight) + biases) # 第一层的计算方法
print('hidden1 的维度为:', hidden1.shape)
# 第二层
with tf.name_scope('hidden2'):
weight = tf.Variable(tf.truncated_normal(
[mnist.HIDDEN1_UNITS, mnist.HIDDEN2_UNITS],
stddev=1.0 / math.sqrt(float(mnist.HIDDEN1_UNITS))),
name='weight')
biases = tf.Variable(tf.zeros([mnist.HIDDEN2_UNITS]),
name='biases')
hidden2 = tf.nn.relu(tf.matmul(hidden1, weight) + biases)
print('hidden2 的维度为:', hidden2.shape)
# 输出层
with tf.name_scope('softmax_linear'):
weight = tf.Variable(tf.truncated_normal(
[mnist.HIDDEN2_UNITS, mnist.NUM_CLASSES],
stddev=1.0 / math.sqrt(float(mnist.HIDDEN2_UNITS))),
name='weight')
biases = tf.Variable(tf.zeros([mnist.NUM_CLASSES]), name='biases')
logits = tf.matmul(hidden2, weight) + biases
print('logits 的维度为:', logits.shape)
labels = tf.to_int64(labels_placeholder) # 标签一致化
print('labels 维度:', labels.shape)
loss = tf.losses.sparse_softmax_cross_entropy(labels=labels,
logits=logits)
# 根据稀疏表示的label和输出层数据计算损失 http://blog.csdn.net/cnki_ok/article/details/70243960
print('loss 的维度为:', loss.shape)
tf.summary.scalar('loss', loss) # 可视化
optimizer = tf.train.GradientDescentOptimizer(
mnist.LEARNING_RATE) # 设置下降参数
global_step = tf.Variable(0, name='global_step', trainable=False)
train_op = optimizer.minimize(loss, global_step=global_step) # 训练目标
correct = tf.nn.in_top_k(logits, labels_placeholder, 1)
eval_correct = tf.reduce_sum(tf.cast(correct, tf.int32)) # 正确率获取图
summary = tf.summary.merge_all()
init = tf.global_variables_initializer()
saver = tf.train.Saver()
sess = tf.Session()
summary_writer = tf.summary.FileWriter(log_dir, sess.graph)
sess.run(init)
for step in range(2000):
start_time = time.time()
batch = data_sets.train.next_batch(batch_size, fake_data=False)
_, loss_value = sess.run([train_op, loss],
feed_dict={
images_placeholder: batch[0],
labels_placeholder: batch[1]
})
duration = time.time() - start_time
if step % 100 == 0:
print('Step %d: loss = %.2f (%.3f sec)' %
(step, loss_value, duration))
summary_str = sess.run(summary,
feed_dict={
images_placeholder: batch[0],
labels_placeholder: batch[1]
})
summary_writer.add_summary(summary_str, step)
summary_writer.flush()
if (step + 1) % 1000 == 0 or (step + 1) == 2000:
checkout_file = os.path.join(log_dir, 'model.ckpt')
saver.save(sess, checkout_file, global_step=step)
print('Training Data Eval:')
do_eval(sess, eval_correct, images_placeholder,
labels_placeholder, data_sets.train)
print('Validation Data Eval:')
do_eval(sess, eval_correct, images_placeholder,
labels_placeholder, data_sets.validation)
# Evaluate against the test set.
print('Test Data Eval:')
do_eval(sess, eval_correct, images_placeholder,
labels_placeholder, data_sets.test)
def run_training():
"""Train MNIST for a number of steps."""
# Get the sets of images and labels for training, validation, and
# test on MNIST.
#data_sets = input_data.read_data_sets(FLAGS.input_data_dir, FLAGS.fake_data)
data_sets = input_data.read_data_sets('MNIST_data', fake_data=False) # 下载数据
# Tell TensorFlow that the model will be built into the default Graph.
with tf.Graph().as_default():
images_placeholder = tf.placeholder(tf.float32, shape=(FLAGS.batch_size, mnist.IMAGE_PIXELS))
labels_placeholder = tf.placeholder(tf.int32, shape=(FLAGS.batch_size))
with tf.name_scope('hidden1'):
weights = tf.Variable(
tf.truncated_normal([mnist.IMAGE_PIXELS, mnist.HIDDEN1_UNITS],
stddev=1.0 / math.sqrt(float(mnist.IMAGE_PIXELS))),
name='weights')
biases = tf.Variable(tf.zeros([mnist.HIDDEN1_UNITS]),
name='biases')
hidden1 = tf.nn.relu(tf.matmul(images_placeholder, weights) + biases)
# Hidden 2
with tf.name_scope('hidden2'):
weights = tf.Variable(
tf.truncated_normal([FLAGS.hidden1, mnist.HIDDEN2_UNITS],
stddev=1.0 / math.sqrt(float(mnist.HIDDEN2_UNITS))),
name='weights')
biases = tf.Variable(tf.zeros([mnist.HIDDEN2_UNITS]),
name='biases')
hidden2 = tf.nn.relu(tf.matmul(hidden1, weights) + biases)
# Linear
with tf.name_scope('softmax_linear'):
weights = tf.Variable(
tf.truncated_normal([mnist.HIDDEN2_UNITS, mnist.NUM_CLASSES],
stddev=1.0 / math.sqrt(float(mnist.HIDDEN2_UNITS))),
name='weights')
biases = tf.Variable(tf.zeros([mnist.NUM_CLASSES]),
name='biases')
logits = tf.matmul(hidden2, weights) + biases
labels = tf.to_int64(labels_placeholder)
loss = tf.losses.sparse_softmax_cross_entropy(labels=labels, logits=logits)
tf.summary.scalar('loss', loss)
optimizer = tf.train.GradientDescentOptimizer(mnist.LEARNING_RATE)
global_step = tf.Variable(0, name='global_step', trainable=False)
train_op = optimizer.minimize(loss, global_step=global_step)
correct = tf.nn.in_top_k(logits, labels_placeholder, 1)
eval_correct = tf.reduce_sum(tf.cast(correct, tf.int32))
summary = tf.summary.merge_all()
# Add the variable initializer Op.
init = tf.global_variables_initializer()
# Create a saver for writing training checkpoints.
saver = tf.train.Saver()
# Create a session for running Ops on the Graph.
sess = tf.Session()
# Instantiate a SummaryWriter to output summaries and the Graph.
summary_writer = tf.summary.FileWriter(FLAGS.log_dir, sess.graph)
# And then after everything is built:
# Run the Op to initialize the variables.
sess.run(init)
# Start the training loop.
for step in xrange(FLAGS.max_steps):
start_time = time.time()
# Fill a feed dictionary with the actual set of images and labels
# for this particular training step.
'''
feed_dict = fill_feed_dict(data_sets.train,
images_placeholder,
labels_placeholder)
'''
batch = data_sets.train.next_batch(batch_size=100, fake_data=False)
# Run one step of the model. The return values are the activations
# from the `train_op` (which is discarded) and the `loss` Op. To
# inspect the values of your Ops or variables, you may include them
# in the list passed to sess.run() and the value tensors will be
# returned in the tuple from the call.
print('logits 的维度是', logits.shape)
print('loss 的维度是', loss.shape)
print('feed_dict im的维度是', images_placeholder.shape)
print('feed_dict im的维度是', labels_placeholder.shape)
_, loss_value = sess.run([train_op, loss],
feed_dict={images_placeholder: batch[0],
labels_placeholder: batch[1]})
duration = time.time() - start_time
# Write the summaries and print an overview fairly often.
if step % 100 == 0:
# Print status to stdout.
print('Step %d: loss = %.2f (%.3f sec)' % (step, loss_value, duration))
# Update the events file.
summary_str = sess.run(summary, feed_dict={images_placeholder: batch[0],
labels_placeholder: batch[1]})
summary_writer.add_summary(summary_str, step)
summary_writer.flush()
# Save a checkpoint and evaluate the model periodically.
if (step + 1) % 1000 == 0 or (step + 1) == FLAGS.max_steps:
checkpoint_file = os.path.join(FLAGS.log_dir, 'model.ckpt')
saver.save(sess, checkpoint_file, global_step=step)
# Evaluate against the training set.
print('Training Data Eval:')
do_eval(sess,
eval_correct,
images_placeholder,
labels_placeholder,
data_sets.train)
# Evaluate against the validation set.
print('Validation Data Eval:')
do_eval(sess,
eval_correct,
images_placeholder,
labels_placeholder,
data_sets.validation)
# Evaluate against the test set.
print('Test Data Eval:')
do_eval(sess,
eval_correct,
images_placeholder,
labels_placeholder,
data_sets.test)