机器学习的概念

有监督学习(Supervised Learning)：

\[x, y\]
输入高维向量（Vector）并同时输入其标签（Label），通过建模（Modeling），使得机器能够自动在模型中计算最为合适的参数值（Parameter）。最终，使得我们获得一个参数确定的模型，并在未来输入新的高维向量时可以预测其Label值，使得Label的误差在所允许的范围之内。

无监督学习(Non-Supervised Learning)：

\[x\]
输入高维向量（Vector）并，通过建模（Modeling），使得机器能够计算出模型中相应的待定参数值（Parameter）。

典型的监督学习

贝叶斯概率

概率是一种事情发生可能性的描述.

基于统计的认知

1.1 重要原则：
- 结论来源于观测
- 定量描述优于定性描述
- 结论条件明确，叙述简洁，有好的泛化性
- 高度自洽性，高度可验证性

1.2 古典概型（静态概型）
- 试验只有有限个基本结果
- 试验的每个基本结果出现的可能性是一样的

1.3 基于统计的概率

在一定条件下，重复做\(n\)次试验（对一个事情观测\(n\)次），一个事件\(A\)发生的概率\(P(A)\)

\[P(A) = \frac{A发生次数}{总观测次数}\]

\(A与B正相关, P(A)<P(A|B)\)
\(A与B负相关, P(A)>P(A|B)\)

朴素贝叶斯分类器

维基百科:所有朴素贝叶斯分类器都假定样本每个特征与其他特征都不相关。

一个普通的规则就是选出最有可能的那个：这就是大家熟知的最大后验概率（MAP）决策准则。
\[classify\{f_1,\dots,f_n\} = \underset{c}{\operatorname{argmax}} \ p(C=c) \displaystyle\prod_{i=1}^n p(F_i=f_i\vert C=c)\]

高斯朴素贝叶斯

如果要处理的是连续数据一种通常的假设是这些连续数值为高斯分布。

例如，假设训练集中有一个连续属性，\(x\)。我们首先对数据根据类别分类，然后计算每个类别中\(x\)的均值和方差。令\(\mu_c\)表示为\(x\)在''c''类上的均值，令\(\sigma^2_c\)为 \(x\)在''c''类上的方差。在给定类中某个值的概率，\(P(x=v|c)\)，可以通过将\(v\)表示为均值为\(\mu_c\)方差为\(\sigma^2_c\)正态分布计算出来。如下，
\[ P(x=v|c)=\tfrac{1}{\sqrt{2\pi\sigma^2_c}}\,e^{ -\frac{(v-\mu_c)^2}{2\sigma^2_c} } \]
处理连续数值问题的另一种常用的技术是通过离散化连续数值的方法。通常，当训练样本数量较少或者是精确的分布已知时，通过概率分布的方法是一种更好的选择。在大量样本的情形下离散化的方法表现更优，因为大量的样本可以学习到数据的分布。由于朴素贝叶斯是一种典型的用到大量样本的方法（越大计算量的模型可以产生越高的分类精确度），所以朴素贝叶斯方法都用到离散化方法，而不是概率分布估计的方法。

from sklearn.naive_bayes import GaussianNB
# 0：晴 1：阴 2：降水 3：多云
data_table =[["date", "weather"],[1, 0],[2, 1],[3, 2],
             [4, 1],[5, 2],[6, 0],[7, 0],[8, 3],[9, 1],[10, 1]]
#当天的天气
X = [[0], [1], [2], [1], [2], [0], [0], [3], [1]]
#当天的天气对应后一天的天气
y = [1, 2, 1, 2, 0, 0, 3, 1, 1]
#现在把训练数据和对应的分类放入分类器中进行训练
clf = GaussianNB().fit(X, y)
p = [[1]]
print(clf.predict(p))

[2]

hello.py

import tensorflow as tf
hello = tf.constant('Hello, TensorFlow!')
sess = tf.Session()
print(sess.run(hello))
a = tf.constant(10)
b = tf.constant(32)
print(sess.run(a + b))

b'Hello, TensorFlow!'
42

实验指导：

实现一个线性回归的例子

# ex-01.py
import tensorflow as tf
import numpy as np

#create some training data
x_data = np.random.rand(100).astype(np.float32)
y_data = x_data*1 + 3

print ("x_data:")
print (x_data)

print ("y_data:")
print( y_data)

#create the weights and bias variables
weights = tf.Variable(tf.random_uniform([1], -1.0, 1.0))
print ("weights before initializing:")
print (weights)

biases = tf.Variable(tf.zeros([1]))
print ("bias before initializing:")
print (biases)

#predict (fit) value
y = weights*x_data + biases

#loss function
loss = tf.reduce_mean(tf.square(y - y_data))

#optimizer definition
optimizer = tf.train.GradientDescentOptimizer(0.1)

#train definition
train = optimizer.minimize(loss)

#initialiing the variables
init = tf.initialize_all_variables()

#session definition and active
sess = tf.Session()
sess.run(init)

#train the model
for step in range(501):
    sess.run(train)
    if step % 10 == 0:
        print (step,sess.run(weights),sess.run(biases))

x_data:
[ 0.89402199  0.10053575  0.55701882  0.53727293  0.36683112  0.44201213
  0.42864946  0.33498201  0.55696607  0.40338814  0.99002725  0.23393948
  0.40767717  0.20491761  0.10732751  0.64065552  0.95823038  0.36961049
  0.43446437  0.05964461  0.39571118  0.36884421  0.92716092  0.85938799
  0.61906868  0.28850925  0.33652243  0.34363723  0.19535853  0.12228432
  0.87395531  0.94348276  0.58827281  0.42699504  0.68471229  0.49138424
  0.60810924  0.59445798  0.48714778  0.84556079  0.71666372  0.73018974
  0.45137393  0.89610088  0.23163974  0.57911086  0.59919071  0.14642832
  0.16170129  0.93183321  0.03960106  0.02113333  0.26829427  0.05438332
  0.26300624  0.25510544  0.47461808  0.42821345  0.13820758  0.61761445
  0.95918888  0.57477629  0.80746269  0.9087519   0.29300612  0.2298919
  0.47188538  0.76722085  0.5297811   0.20937359  0.86967862  0.11050695
  0.41093281  0.43827012  0.63228905  0.29376149  0.35848793  0.44383112
  0.11717488  0.82729691  0.06545471  0.2652818   0.00641522  0.54356593
  0.69357717  0.9950943   0.72574335  0.77202976  0.6673649   0.33663997
  0.93204492  0.86043173  0.88384038  0.27974007  0.33629051  0.64907324
  0.77879322  0.63549143  0.0160666   0.56776083]
y_data:
[ 3.89402199  3.10053587  3.55701876  3.53727293  3.36683106  3.44201207
  3.42864943  3.33498192  3.55696607  3.40338802  3.99002719  3.23393941
  3.40767717  3.20491767  3.10732746  3.64065552  3.9582305   3.36961055
  3.43446445  3.0596447   3.39571118  3.36884427  3.92716098  3.85938787
  3.61906862  3.28850937  3.33652234  3.34363723  3.19535851  3.12228441
  3.87395525  3.94348288  3.58827281  3.42699504  3.68471241  3.49138427
  3.60810924  3.5944581   3.48714781  3.84556079  3.71666384  3.7301898
  3.45137405  3.896101    3.23163986  3.57911086  3.59919071  3.14642835
  3.1617012   3.93183327  3.03960109  3.02113342  3.26829433  3.05438328
  3.26300621  3.2551055   3.47461796  3.42821336  3.13820767  3.61761451
  3.95918894  3.57477617  3.80746269  3.90875196  3.29300618  3.22989178
  3.47188544  3.76722097  3.5297811   3.20937347  3.8696785   3.11050701
  3.41093278  3.43827009  3.63228893  3.29376149  3.35848784  3.44383121
  3.11717486  3.82729697  3.06545472  3.26528168  3.00641513  3.54356599
  3.69357729  3.9950943   3.72574329  3.77202988  3.66736484  3.33663988
  3.93204498  3.86043167  3.88384032  3.2797401   3.3362906   3.64907312
  3.77879333  3.63549137  3.01606655  3.56776094]
weights before initializing:
<tf.Variable 'Variable:0' shape=(1,) dtype=float32_ref>
bias before initializing:
<tf.Variable 'Variable_1:0' shape=(1,) dtype=float32_ref>
WARNING:tensorflow:From C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\util\tf_should_use.py:170: initialize_all_variables (from tensorflow.python.ops.variables) is deprecated and will be removed after 2017-03-02.
Instructions for updating:
Use `tf.global_variables_initializer` instead.
0 [ 0.90101707] [ 0.64157164]
10 [ 1.73624611] [ 2.48117089]
20 [ 1.6977222] [ 2.62674093]
30 [ 1.62073231] [ 2.67402244]
40 [ 1.55015469] [ 2.7114203]
50 [ 1.48748767] [ 2.74431014]
60 [ 1.43195248] [ 2.77343965]
70 [ 1.3827436] [ 2.79925013]
80 [ 1.33914053] [ 2.82212019]
90 [ 1.30050492] [ 2.84238434]
100 [ 1.266271] [ 2.86034036]
110 [ 1.23593676] [ 2.87625074]
120 [ 1.2090584] [ 2.8903482]
130 [ 1.18524206] [ 2.9028399]
140 [ 1.16413891] [ 2.91390872]
150 [ 1.14543998] [ 2.92371631]
160 [ 1.12887108] [ 2.93240666]
170 [ 1.11418998] [ 2.94010711]
180 [ 1.10118115] [ 2.94693041]
190 [ 1.08965421] [ 2.95297623]
200 [ 1.07944047] [ 2.95833325]
210 [ 1.07039058] [ 2.96307993]
220 [ 1.06237149] [ 2.96728611]
230 [ 1.0552659] [ 2.97101283]
240 [ 1.04896998] [ 2.97431517]
250 [ 1.04339111] [ 2.97724128]
260 [ 1.03844786] [ 2.97983384]
270 [ 1.03406775] [ 2.98213148]
280 [ 1.03018677] [ 2.9841671]
290 [ 1.0267477] [ 2.98597097]
300 [ 1.02370059] [ 2.98756886]
310 [ 1.02100062] [ 2.98898506]
320 [ 1.01860821] [ 2.99023986]
330 [ 1.01648819] [ 2.99135184]
340 [ 1.01460981] [ 2.99233723]
350 [ 1.01294541] [ 2.99321008]
360 [ 1.01147056] [ 2.99398375]
370 [ 1.01016378] [ 2.99466896]
380 [ 1.00900602] [ 2.99527621]
390 [ 1.00797999] [ 2.99581456]
400 [ 1.00707078] [ 2.9962914]
410 [ 1.00626528] [ 2.99671388]
420 [ 1.00555134] [ 2.99708843]
430 [ 1.00491893] [ 2.99742007]
440 [ 1.00435853] [ 2.99771404]
450 [ 1.0038619] [ 2.9979744]
460 [ 1.00342214] [ 2.99820518]
470 [ 1.00303233] [ 2.99840927]
480 [ 1.00268698] [ 2.99859095]
490 [ 1.00238085] [ 2.9987514]
500 [ 1.00210965] [ 2.9988935]

实现一个矩阵乘法计算

# ex-02.py
import tensorflow as tf

matrix1 = tf.constant([[3,3]])

matrix2 = tf.constant([[1],
                        [2]])

#matrix multiply
product = tf.matmul(matrix1, matrix2)

#method1
#sess = tf.Session()
#result = sess.run(product)
#print "result:",result
#sess.close()

#method2
with tf.Session() as sess:
    result2 = sess.run(product)
    print ("result2:",result2)

result2: [[9]]

常量/变量/赋值

# ex-03.py
import tensorflow as tf

state = tf.Variable(10, name='counter')

print( state.name)

one = tf.constant(1)

new_value = tf.add(state, one)

update = tf.assign(state, new_value)

#this initializing is very important
init = tf.global_variables_initializer()

with tf.Session() as sess:
    sess.run(init)
    for x in range(10):
        sess.run(update)
    # must do sess.run to operate the variables
        print(sess.run(state))

counter_1:0
11
12
13
14
15
16
17
18
19
20

Placeholder用法

# ex-04.py
import tensorflow as tf

input1 = tf.placeholder(tf.float32)
input2 = tf.placeholder(tf.float32)

output = input1 * input2

with tf.Session() as sess:
    print(sess.run(output, feed_dict={input1: 4, input2: 2}))
    print(sess.run(output, feed_dict={input1: [4,2], input2: [2,7]}))

8.0
[  8.  14.]

# ex-05.py
import tensorflow as tf
import numpy as np

def add_layer(inputs, in_size, out_size, activation_function=None):
    weights = tf.Variable(tf.random_normal([in_size, out_size]))
    biases = tf.Variable(tf.zeros([1,out_size]))
    wx_b = tf.matmul(inputs, weights) + biases
    if activation_function is None:
        outputs = wx_b
    else:
        outputs = activation_function(wx_b)
    return outputs

#make some input value
x_data = np.linspace(-1, 1, 300)[:, np.newaxis]
noise = np.random.normal(0, 0.05, x_data.shape)
y_data = np.square(x_data) + noise

xs = tf.placeholder(tf.float32, [None,1])
ys = tf.placeholder(tf.float32, [None,1])

layer1 = add_layer(xs, 1, 10, activation_function=tf.nn.sigmoid)
prediction = add_layer(layer1, 10, 1, activation_function=None)

loss = tf.reduce_mean(tf.reduce_sum(tf.square(ys - prediction), reduction_indices=[1]))

train = tf.train.GradientDescentOptimizer(0.01).minimize(loss)

init = tf.global_variables_initializer()

sess = tf.Session()
sess.run(init)

for x in range(1000):
    sess.run(train, feed_dict = { xs: x_data, ys: y_data})
    if x % 10==0:
        print (sess.run(loss, feed_dict={xs:x_data, ys: y_data}))

可视化

tensorboard --logdir='logs/'

注意：不能在notebook直接运行。

ex-06-vis.py

import time
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt

def add_layer(inputs, in_size, out_size, activation_function=None):
    weights = tf.Variable(tf.random_normal([in_size, out_size]))
    biases = tf.Variable(tf.zeros([1,out_size]))
    wx_b = tf.matmul(inputs, weights) + biases
    if activation_function is None:
        outputs = wx_b
    else:
        outputs = activation_function(wx_b)
    return outputs

#make some input value
x_data = np.linspace(-1, 1, 300)[:, np.newaxis]
noise = np.random.normal(0, 0.05, x_data.shape)
y_data = np.square(x_data) + noise

xs = tf.placeholder(tf.float32, [None,1])
ys = tf.placeholder(tf.float32, [None,1])

layer1 = add_layer(xs, 1, 10, activation_function=tf.nn.sigmoid)
prediction = add_layer(layer1, 10, 1, activation_function=None)

loss = tf.reduce_mean(tf.reduce_sum(tf.square(ys - prediction), reduction_indices=[1]))

train = tf.train.GradientDescentOptimizer(0.1).minimize(loss)

init = tf.global_variables_initializer()

sess = tf.Session()
sess.run(init)

#added coding
fig = plt.figure()
ax = fig.add_subplot(1,1,1)
ax.scatter(x_data, y_data)
plt.ion()
plt.show()

for x in range(1000):
    sess.run(train, feed_dict = { xs: x_data, ys: y_data})
    if x % 10==0:
        # print (sess.run(loss, feed_dict={xs:x_data, ys: y_data}))
        
        try:
            ax.lines.remove(lines[0])
        except Exception:
            pass
        prediction_value = sess.run(prediction, feed_dict={xs:x_data})
        lines = ax.plot(x_data, prediction_value, 'r-', lw=5)
        plt.pause(0.1)

import time
for i in [1,2,34,5]:
    print(i)
    time.sleep(0.1)

ex-09-tb.py

# from ex-05
import tensorflow as tf
import numpy as np

def add_layer(inputs, in_size, out_size, n_layer, activation_function=None):
    layer_name = 'layer%s' % n_layer
    with tf.name_scope('layer'):
        with tf.name_scope('weights'):
            weights = tf.Variable(tf.random_normal([in_size, out_size]), name='W')
            tf.summary.histogram(layer_name + '/weights', weights)
        with tf.name_scope('bias'):
            biases = tf.Variable(tf.zeros([1,out_size]), name='b')
            tf.summary.histogram(layer_name + '/biases', biases)
        with tf.name_scope('wx_plus_b'):
            wx_b = tf.add(tf.matmul(inputs, weights), biases)
        if activation_function is None:
            outputs = wx_b
        else:
            outputs = activation_function(wx_b, name='output')
            tf.summary.histogram(layer_name + '/outputs', outputs)
    return outputs

#make some input value
x_data = np.linspace(-1, 1, 300)[:, np.newaxis]
noise = np.random.normal(0, 0.05, x_data.shape)
y_data = np.square(x_data) + noise

with tf.name_scope('inputs'):
    xs = tf.placeholder(tf.float32, [None,1], name='x_input')
    ys = tf.placeholder(tf.float32, [None,1], name='y_input')

layer1 = add_layer(xs, 1, 10, n_layer=1, activation_function=tf.nn.sigmoid)
prediction = add_layer(layer1, 10, 1, n_layer=2, activation_function=None)

with tf.name_scope('loss'):
    loss = tf.reduce_mean(tf.reduce_sum(tf.square(ys - prediction), reduction_indices=[1]))
    tf.summary.scalar('loss', loss)

with tf.name_scope('train'):
    train = tf.train.GradientDescentOptimizer(0.1).minimize(loss)

init = tf.global_variables_initializer()

sess = tf.Session()
sess.run(init)

merged = tf.summary.merge_all()

writer = tf.summary.FileWriter("logs/", sess.graph)

for x in range(50000):
    sess.run(train, feed_dict = { xs: x_data, ys: y_data})
    if x % 50==0:
        sess.run(loss, feed_dict={xs:x_data, ys:y_data})
        result = sess.run(merged, feed_dict={xs:x_data, ys:y_data})
        writer.add_summary(result, x)

import tensorflow as tf

a = tf.constant(5, name="input_a")
b = tf.constant(3, name="input_b")
c = tf.multiply(a, b, name="mul_c")
d = tf.add(a, b, name="add_d")
e = tf.add(c, d, name="add_e")

sess = tf.Session()
sess.run(e)

writer = tf.summary.FileWriter("E:/tensorflow/graph", tf.get_default_graph())
writer.close()

ex-10.py

import tensorflow as tf
import gzip
import numpy
import collections
from tensorflow.python.framework import random_seed
from tensorflow.python.framework import dtypes
#from tensorflow.examples.tutorials.mnist import input_data

Datasets = collections.namedtuple('Datasets', ['train', 'validation', 'test'])

def _read32(bytestream):
    dt = numpy.dtype(numpy.uint32).newbyteorder('>')
    return numpy.frombuffer(bytestream.read(4), dtype=dt)[0]

def dense_to_one_hot(labels_dense, num_classes):
    """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_images(f):
  """Extract the images into a 4D uint8 numpy array [index, y, x, depth].
  Args:
    f: A file object that can be passed into a gzip reader.
  Returns:
    data: A 4D uint8 numpy array [index, y, x, depth].
  Raises:
    ValueError: If the bytestream does not start with 2051.
  """
  print('Extracting', f.name)
  with gzip.GzipFile(fileobj=f) as bytestream:
    magic = _read32(bytestream)
    if magic != 2051:
      raise ValueError('Invalid magic number %d in MNIST image file: %s' %
                       (magic, f.name))
    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 extract_labels(f, one_hot=False, num_classes=10):
  """Extract the labels into a 1D uint8 numpy array [index].
  Args:
    f: A file object that can be passed into a gzip reader.
    one_hot: Does one hot encoding for the result.
    num_classes: Number of classes for the one hot encoding.
  Returns:
    labels: a 1D uint8 numpy array.
  Raises:
    ValueError: If the bystream doesn't start with 2049.
  """
  print('Extracting', f.name)
  with gzip.GzipFile(fileobj=f) as bytestream:
    magic = _read32(bytestream)
    if magic != 2049:
      raise ValueError('Invalid magic number %d in MNIST label file: %s' %
                       (magic, f.name))
    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, num_classes)
    return labels

def read_data_sets(train_dir,
                   fake_data=False,
                   one_hot=False,
                   dtype=tf.float32,
                   reshape=True,
                   validation_size=5000,
                   seed=None):
  '''
  if fake_data:
    def fake():
      return DataSet(
          [], [], fake_data=True, one_hot=one_hot, dtype=dtype, seed=seed)
  
    train = fake()
    validation = fake()
    test = fake()
    return base.Datasets(train=train, validation=validation, test=test)
  '''

  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'

  local_file = '/home/gao/dl_learning/py-example/MNIST_data/' + TRAIN_IMAGES
  with open(local_file, 'rb') as f:
    train_images = extract_images(f)

                                   
  local_file = '/home/gao/dl_learning/py-example/MNIST_data/' + TRAIN_LABELS
  with open(local_file, 'rb') as f:
    train_labels = extract_labels(f, one_hot=one_hot)

  local_file = '/home/gao/dl_learning/py-example/MNIST_data/' + TEST_IMAGES
  with open(local_file, 'rb') as f:
    test_images = extract_images(f)

  local_file = '/home/gao/dl_learning/py-example/MNIST_data/' + TEST_LABELS
     
  with open(local_file, 'rb') as f:
    test_labels = extract_labels(f, one_hot=one_hot)

  if not 0 <= validation_size <= len(train_images):
    raise ValueError(
        'Validation size should be between 0 and {}. Received: {}.'
        .format(len(train_images), validation_size))

  validation_images = train_images[:validation_size]
  validation_labels = train_labels[:validation_size]
  train_images = train_images[validation_size:]
  train_labels = train_labels[validation_size:]

  options = dict(dtype=dtype, reshape=reshape, seed=seed)
  
  train = DataSet(train_images, train_labels, **options)
  validation = DataSet(validation_images, validation_labels, **options)
  test = DataSet(test_images, test_labels, **options)
  
  return Datasets(train=train, validation=validation, test=test)

class DataSet(object):

  def __init__(self,
               images,
               labels,
               fake_data=False,
               one_hot=False,
               dtype=tf.float32,
               reshape=True,
               seed=None):
    """Construct a DataSet.
    one_hot arg is used only if fake_data is true.  `dtype` can be either
    `uint8` to leave the input as `[0, 255]`, or `float32` to rescale into
    `[0, 1]`.  Seed arg provides for convenient deterministic testing.
    """
    seed1, seed2 = random_seed.get_seed(seed)
    # If op level seed is not set, use whatever graph level seed is returned
    numpy.random.seed(seed1 if seed is None else seed2)
    dtype = dtypes.as_dtype(dtype).base_dtype
    if dtype not in (dtypes.uint8, dtypes.float32):
      raise TypeError('Invalid image dtype %r, expected uint8 or float32' %
                      dtype)
    if fake_data:
      self._num_examples = 10000
      self.one_hot = one_hot
    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)
      if reshape:
        assert images.shape[3] == 1
        images = images.reshape(images.shape[0],
                                images.shape[1] * images.shape[2])
      if dtype == dtypes.float32:
        # 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, shuffle=True):
    """Return the next `batch_size` examples from this data set."""
    if fake_data:
      fake_image = [1] * 784
      if self.one_hot:
        fake_label = [1] + [0] * 9
      else:
        fake_label = 0
      return [fake_image for _ in xrange(batch_size)], [
          fake_label for _ in xrange(batch_size)
      ]
    start = self._index_in_epoch
    # Shuffle for the first epoch
    if self._epochs_completed == 0 and start == 0 and shuffle:
      perm0 = numpy.arange(self._num_examples)
      numpy.random.shuffle(perm0)
      self._images = self.images[perm0]
      self._labels = self.labels[perm0]
    # Go to the next epoch
    if start + batch_size > self._num_examples:
      # Finished epoch
      self._epochs_completed += 1
      # Get the rest examples in this epoch
      rest_num_examples = self._num_examples - start
      images_rest_part = self._images[start:self._num_examples]
      labels_rest_part = self._labels[start:self._num_examples]
      # Shuffle the data
      if shuffle:
        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 - rest_num_examples
      end = self._index_in_epoch
      images_new_part = self._images[start:end]
      labels_new_part = self._labels[start:end]
      return numpy.concatenate((images_rest_part, images_new_part), axis=0) , numpy.concatenate((labels_rest_part, labels_new_part), axis=0)
    else:
      self._index_in_epoch += batch_size
      end = self._index_in_epoch
      return self._images[start:end], self._labels[start:end]

################################################################
#from here to operate the network

mnist = read_data_sets('MNIST_data', one_hot=True)

def add_layer(inputs, in_size, out_size, activation_function=None):
  weights = tf.Variable(tf.random_normal([in_size, out_size]))
  biases = tf.Variable(tf.zeros([1, out_size]) + 0.1,)
  wx_b = tf.matmul(inputs, weights) + biases
  if activation_function is None:
    outputs = wx_b
  else:
    outputs = activation_function(wx_b,)
  return outputs

xs = tf.placeholder(tf.float32, [None, 28*28])
ys = tf.placeholder(tf.float32, [None, 10])

def compute_accuracy(v_xs, v_ys):
  global prediction
  y_pre = sess.run(prediction, feed_dict={xs: v_xs})
  correct_prediction = tf.equal(tf.argmax(y_pre,1), tf.argmax(v_ys,1))
  accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
  result = sess.run(accuracy, feed_dict={xs:v_xs, ys:v_ys})
  return result

prediction = add_layer(xs, 784, 10, activation_function = tf.nn.softmax)

cross_entropy = tf.reduce_mean(-tf.reduce_sum(ys * tf.log(prediction), 
                              reduction_indices=[1]))

train_step = tf.train.GradientDescentOptimizer(0.1).minimize(cross_entropy)

sess = tf.Session()

sess.run(tf.initialize_all_variables())

for i in range(10000):
  batch_xs, batch_ys = mnist.train.next_batch(100)
  sess.run(train_step, feed_dict = {xs: batch_xs, ys: batch_ys})
  if i % 50 ==0:
    print "step:",i,", ",compute_accuracy(mnist.test.images, mnist.test.labels)

  File "<ipython-input-12-d56c6b11bc6b>", line 279
    print "step:",i,", ",compute_accuracy(mnist.test.images, mnist.test.labels)
                ^
SyntaxError: invalid syntax

# ex-10-ml.py

# import tensorflow as tf
import gzip
import numpy
import collections
from tensorflow.python.framework import random_seed
from tensorflow.python.framework import dtypes
#from tensorflow.examples.tutorials.mnist import input_data

Datasets = collections.namedtuple('Datasets', ['train', 'validation', 'test'    ])

def _read32(bytestream):
  dt = numpy.dtype(numpy.uint32).newbyteorder('>')
  return numpy.frombuffer(bytestream.read(4), dtype=dt)[0]

def dense_to_one_hot(labels_dense, num_classes):
  """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_images(f):
  """Extract the images into a 4D uint8 numpy array [index, y, x, depth].
  Args:
    f: A file object that can be passed into a gzip reader.
  Returns:
    data: A 4D uint8 numpy array [index, y, x, depth].
  Raises:
    ValueError: If the bytestream does not start with 2051.
  """
  print('Extracting', f.name)
  with gzip.GzipFile(fileobj=f) as bytestream:
    magic = _read32(bytestream)
    if magic != 2051:
      raise ValueError('Invalid magic number %d in MNIST image file: %s' %
                       (magic, f.name))
    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 extract_labels(f, one_hot=False, num_classes=10):
  """Extract the labels into a 1D uint8 numpy array [index].
  Args:
    f: A file object that can be passed into a gzip reader.
    one_hot: Does one hot encoding for the result.
    num_classes: Number of classes for the one hot encoding.
  Returns:
    labels: a 1D uint8 numpy array.
  Raises:
    ValueError: If the bystream doesn't start with 2049.
  """
  print('Extracting', f.name)
  with gzip.GzipFile(fileobj=f) as bytestream:
    magic = _read32(bytestream)
    if magic != 2049:
      raise ValueError('Invalid magic number %d in MNIST label file: %s' %
                       (magic, f.name))
    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, num_classes)
    return labels

def read_data_sets(train_dir,
                   fake_data=False,
                   one_hot=False,
                   dtype=tf.float32,
                   reshape=True,
                   validation_size=5000,
                   seed=None):
  '''
  if fake_data:
    def fake():
      return DataSet(
          [], [], fake_data=True, one_hot=one_hot, dtype=dtype, seed=seed)
  
    train = fake()
    validation = fake()
    test = fake()
    return base.Datasets(train=train, validation=validation, test=test)
  '''

  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'

  local_file = '/home/gao/dl_learning/py-example/MNIST_data/' + TRAIN_IMAGES
  with open(local_file, 'rb') as f:
    train_images = extract_images(f)

                                   
  local_file = '/home/gao/dl_learning/py-example/MNIST_data/' + TRAIN_LABELS
  with open(local_file, 'rb') as f:
    train_labels = extract_labels(f, one_hot=one_hot)

  local_file = '/home/gao/dl_learning/py-example/MNIST_data/' + TEST_IMAGES
  with open(local_file, 'rb') as f:
    test_images = extract_images(f)

  local_file = '/home/gao/dl_learning/py-example/MNIST_data/' + TEST_LABELS
     
  with open(local_file, 'rb') as f:
    test_labels = extract_labels(f, one_hot=one_hot)

  if not 0 <= validation_size <= len(train_images):
    raise ValueError(
        'Validation size should be between 0 and {}. Received: {}.'
        .format(len(train_images), validation_size))

  validation_images = train_images[:validation_size]
  validation_labels = train_labels[:validation_size]
  train_images = train_images[validation_size:]
  train_labels = train_labels[validation_size:]

  options = dict(dtype=dtype, reshape=reshape, seed=seed)
  
  train = DataSet(train_images, train_labels, **options)
  validation = DataSet(validation_images, validation_labels, **options)
  test = DataSet(test_images, test_labels, **options)
  
  return Datasets(train=train, validation=validation, test=test)

class DataSet(object):

  def __init__(self,
               images,
               labels,
               fake_data=False,
               one_hot=False,
               dtype=tf.float32,
               reshape=True,
               seed=None):
    """Construct a DataSet.
    one_hot arg is used only if fake_data is true.  `dtype` can be either
    `uint8` to leave the input as `[0, 255]`, or `float32` to rescale into
    `[0, 1]`.  Seed arg provides for convenient deterministic testing.
    """
    seed1, seed2 = random_seed.get_seed(seed)
    # If op level seed is not set, use whatever graph level seed is returned
    numpy.random.seed(seed1 if seed is None else seed2)
    dtype = dtypes.as_dtype(dtype).base_dtype
    if dtype not in (dtypes.uint8, dtypes.float32):
      raise TypeError('Invalid image dtype %r, expected uint8 or float32' %
                      dtype)
    if fake_data:
      self._num_examples = 10000
      self.one_hot = one_hot
    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)
      if reshape:
        assert images.shape[3] == 1
        images = images.reshape(images.shape[0],
                                images.shape[1] * images.shape[2])
      if dtype == dtypes.float32:
        # 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, shuffle=True):
    """Return the next `batch_size` examples from this data set."""
    if fake_data:
      fake_image = [1] * 784
      if self.one_hot:
        fake_label = [1] + [0] * 9
      else:
        fake_label = 0
      return [fake_image for _ in xrange(batch_size)], [
          fake_label for _ in xrange(batch_size)
      ]
    start = self._index_in_epoch
    # Shuffle for the first epoch
    if self._epochs_completed == 0 and start == 0 and shuffle:
      perm0 = numpy.arange(self._num_examples)
      numpy.random.shuffle(perm0)
      self._images = self.images[perm0]
      self._labels = self.labels[perm0]
    # Go to the next epoch
    if start + batch_size > self._num_examples:
      # Finished epoch
      self._epochs_completed += 1
      # Get the rest examples in this epoch
      rest_num_examples = self._num_examples - start
      images_rest_part = self._images[start:self._num_examples]
      labels_rest_part = self._labels[start:self._num_examples]
      # Shuffle the data
      if shuffle:
        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 - rest_num_examples
      end = self._index_in_epoch
      images_new_part = self._images[start:end]
      labels_new_part = self._labels[start:end]
      return numpy.concatenate((images_rest_part, images_new_part), axis=0) , numpy.concatenate((labels_rest_part, labels_new_part), axis=0)
    else:
      self._index_in_epoch += batch_size
      end = self._index_in_epoch
      return self._images[start:end], self._labels[start:end]

################################################################


mnist = read_data_sets('MNIST_data', one_hot=True)

def add_layer(inputs, in_size, out_size, activation_function=None):
  weights = tf.Variable(tf.random_normal([in_size, out_size]))
  #biases = tf.Variable(tf.zeros([1, out_size]) + 100)
  biases = tf.Variable(tf.zeros([1, out_size]) + 0.1)
  wx_b = tf.matmul(inputs, weights) + biases
  if activation_function is None:
    outputs = wx_b
  else:
    outputs = activation_function(wx_b,)
  return outputs

xs = tf.placeholder(tf.float32, [None, 28*28])
ys = tf.placeholder(tf.float32, [None, 10])

def compute_accuracy(v_xs, v_ys):
  global prediction
  y_pre = sess.run(prediction, feed_dict={xs: v_xs})
  correct_prediction = tf.equal(tf.argmax(y_pre,1), tf.argmax(v_ys,1))
  accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
  result = sess.run(accuracy, feed_dict={xs:v_xs, ys:v_ys})
  return result

#layer1 = add_layer(xs, 784, 50, activation_function = tf.nn.tanh)
#layer1 = add_layer(xs, 784, 30, activation_function = tf.nn.sigmoid)
layer1 = add_layer(xs, 784, 50, activation_function = tf.nn.relu)
layer2 = add_layer(layer1, 50, 50, activation_function = tf.nn.tanh)
layer3 = add_layer(layer2, 50, 50, activation_function = tf.nn.tanh)
layer4 = add_layer(layer3, 50, 50, activation_function = tf.nn.tanh)

prediction = add_layer(layer4, 50, 10, activation_function = tf.nn.softmax)

cross_entropy = tf.reduce_mean(-tf.reduce_sum(ys * tf.log(prediction), 
                              reduction_indices=[1]))

train_step = tf.train.GradientDescentOptimizer(0.05).minimize(cross_entropy)

sess = tf.Session()

sess.run(tf.initialize_all_variables())

for i in range(100000):
  batch_xs, batch_ys = mnist.train.next_batch(100)
  sess.run(layer1,  feed_dict = {xs: batch_xs, ys: batch_ys})
  sess.run(train_step, feed_dict = {xs: batch_xs, ys:batch_ys})
  if i % 100 ==0:
    print "step:",i,", ",compute_accuracy(mnist.test.images, mnist.test.labels)

  File "<ipython-input-13-eaeb1867aa2d>", line 286
    print "step:",i,", ",compute_accuracy(mnist.test.images, mnist.test.labels)
                ^
SyntaxError: invalid syntax

# ex-11.py
import tensorflow as tf
from sklearn.datasets import load_digits
from sklearn.cross_validation import train_test_split
from sklearn.preprocessing import LabelBinarizer

digits = load_digits()
x = digits.data
y = digits.target
#one-hot transform
y = LabelBinarizer().fit_transform(y)
x_train, x_test, y_train, y_test = train_test_split(x, y, test_size = 0.3)

def add_layer(inputs, in_size, out_size, layer_name, activation_function = None):
  weights = tf.Variable(tf.random_normal([in_size, out_size]))
  biases = tf.Variable(tf.zeros([1, out_size]) + 0.1,)
  wx_b = tf.matmul(inputs, weights) + biases

  if activation_function is None:
    outputs = wx_b
  else:
    outputs = activation_function(wx_b)
  tf.summary.histogram(layer_name + '/outputs', outputs)
  return outputs

xs = tf.placeholder(tf.float32, [None, 8*8])
ys = tf.placeholder(tf.float32, [None, 10])

layer1 = add_layer(xs, 64, 100, 'layer1', activation_function = tf.nn.tanh)
prediction = add_layer(layer1, 100, 10, 'layer2', activation_function = tf.nn.softmax)

cross_entropy = tf.reduce_mean(-tf.reduce_sum(ys * tf.log(prediction), reduction_indices=[1]))

tf.summary.scalar('loss', cross_entropy)

train_step = tf.train.GradientDescentOptimizer(0.1).minimize(cross_entropy)

sess = tf.Session()
merged = tf.summary.merge_all()

train_writer = tf.train.SummaryWriter("logs/train", sess.graph)
test_writer = tf.train.SummaryWriter("logs/test", sess.graph)

sess.run(tf.initialize_all_variables())

for i in range(1000):
  sess.run(train_step, feed_dict = {xs: x_train, ys:y_train})
  if i % 100==0:
    train_result = sess.run(merged, feed_dict={xs:x_train, ys:y_train})
    test_result = sess.run(merged, feed_dict={xs:x_test, ys:y_test})
    train_writer.add_summary(train_result, i)
    test_writer.add_summary(test_result, i)

C:\ProgramData\Anaconda3\lib\site-packages\sklearn\cross_validation.py:44: DeprecationWarning: This module was deprecated in version 0.18 in favor of the model_selection module into which all the refactored classes and functions are moved. Also note that the interface of the new CV iterators are different from that of this module. This module will be removed in 0.20.
  "This module will be removed in 0.20.", DeprecationWarning)



---------------------------------------------------------------------------

AttributeError                            Traceback (most recent call last)

<ipython-input-14-212ae64f528b> in <module>()
     39 merged = tf.summary.merge_all()
     40 
---> 41 train_writer = tf.train.SummaryWriter("logs/train", sess.graph)
     42 test_writer = tf.train.SummaryWriter("logs/test", sess.graph)
     43 


AttributeError: module 'tensorflow.python.training.training' has no attribute 'SummaryWriter'

# ex-11-dropout.py
import tensorflow as tf
from sklearn.datasets import load_digits
from sklearn.cross_validation import train_test_split
from sklearn.preprocessing import LabelBinarizer

digits = load_digits()
x = digits.data
y = digits.target
#one-hot transform
y = LabelBinarizer().fit_transform(y)
x_train, x_test, y_train, y_test = train_test_split(x, y, test_size = 0.3)

def add_layer(inputs, in_size, out_size, layer_name, activation_function = None):
  weights = tf.Variable(tf.random_normal([in_size, out_size]))
  biases = tf.Variable(tf.zeros([1, out_size]) + 0.1,)
  wx_b = tf.matmul(inputs, weights) + biases
  #drop out and keep a proportion
  wx_b = tf.nn.dropout(wx_b, keep_prob)

  if activation_function is None:
    outputs = wx_b
  else:
    outputs = activation_function(wx_b)
  tf.histogram_summary(layer_name + '/outputs', outputs)
  return outputs

keep_prob = tf.placeholder(tf.float32)
xs = tf.placeholder(tf.float32, [None, 8*8])
ys = tf.placeholder(tf.float32, [None, 10])

layer1 = add_layer(xs, 64, 100, 'layer1', activation_function = tf.nn.tanh)
prediction = add_layer(layer1, 100, 10, 'layer2', activation_function = tf.nn.softmax)

cross_entropy = tf.reduce_mean(-tf.reduce_sum(ys * tf.log(prediction), reduction_indices=[1]))

tf.scalar_summary('loss', cross_entropy)

train_step = tf.train.GradientDescentOptimizer(0.1).minimize(cross_entropy)

sess = tf.Session()
merged = tf.merge_all_summaries()

train_writer = tf.train.SummaryWriter("logs/train", sess.graph)
test_writer = tf.train.SummaryWriter("logs/test", sess.graph)

sess.run(tf.initialize_all_variables())

for i in range(1000):
  sess.run(train_step, feed_dict = {xs: x_train, ys:y_train, keep_prob:0.7})
  if i % 100==0:
    train_result = sess.run(merged, feed_dict={xs:x_train, ys:y_train, keep_prob: 0.7})
    test_result = sess.run(merged, feed_dict={xs:x_test, ys:y_test, keep_prob: 0.7})
    train_writer.add_summary(train_result, i)
    test_writer.add_summary(test_result, i)

---------------------------------------------------------------------------

AttributeError                            Traceback (most recent call last)

<ipython-input-15-482298ce8d49> in <module>()
     30 ys = tf.placeholder(tf.float32, [None, 10])
     31 
---> 32 layer1 = add_layer(xs, 64, 100, 'layer1', activation_function = tf.nn.tanh)
     33 prediction = add_layer(layer1, 100, 10, 'layer2', activation_function = tf.nn.softmax)
     34 


<ipython-input-15-482298ce8d49> in add_layer(inputs, in_size, out_size, layer_name, activation_function)
     23   else:
     24     outputs = activation_function(wx_b)
---> 25   tf.histogram_summary(layer_name + '/outputs', outputs)
     26   return outputs
     27 


AttributeError: module 'tensorflow' has no attribute 'histogram_summary'

input_data.py

# Copyright 2015 The TensorFlow Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ==============================================================================

"""Functions for downloading and reading MNIST data."""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function

import gzip
import os
import tempfile

import numpy
from six.moves import urllib
from six.moves import xrange  # pylint: disable=redefined-builtin
import tensorflow as tf
from tensorflow.contrib.learn.python.learn.datasets.mnist import read_data_sets

mnist.py

# Copyright 2016 The TensorFlow Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ==============================================================================

"""Functions for downloading and reading MNIST data."""

from __future__ import absolute_import
from __future__ import division
from __future__ import print_function

import gzip

import numpy
from six.moves import xrange  # pylint: disable=redefined-builtin

from tensorflow.contrib.learn.python.learn.datasets import base
from tensorflow.python.framework import dtypes
from tensorflow.python.framework import random_seed

# CVDF mirror of http://yann.lecun.com/exdb/mnist/
SOURCE_URL = 'https://storage.googleapis.com/cvdf-datasets/mnist/'


def _read32(bytestream):
  dt = numpy.dtype(numpy.uint32).newbyteorder('>')
  return numpy.frombuffer(bytestream.read(4), dtype=dt)[0]


def extract_images(f):
  """Extract the images into a 4D uint8 numpy array [index, y, x, depth].

  Args:
    f: A file object that can be passed into a gzip reader.

  Returns:
    data: A 4D uint8 numpy array [index, y, x, depth].

  Raises:
    ValueError: If the bytestream does not start with 2051.

  """
  print('Extracting', f.name)
  with gzip.GzipFile(fileobj=f) as bytestream:
    magic = _read32(bytestream)
    if magic != 2051:
      raise ValueError('Invalid magic number %d in MNIST image file: %s' %
                       (magic, f.name))
    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):
  """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(f, one_hot=False, num_classes=10):
  """Extract the labels into a 1D uint8 numpy array [index].

  Args:
    f: A file object that can be passed into a gzip reader.
    one_hot: Does one hot encoding for the result.
    num_classes: Number of classes for the one hot encoding.

  Returns:
    labels: a 1D uint8 numpy array.

  Raises:
    ValueError: If the bystream doesn't start with 2049.
  """
  print('Extracting', f.name)
  with gzip.GzipFile(fileobj=f) as bytestream:
    magic = _read32(bytestream)
    if magic != 2049:
      raise ValueError('Invalid magic number %d in MNIST label file: %s' %
                       (magic, f.name))
    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, num_classes)
    return labels


class DataSet(object):

  def __init__(self,
               images,
               labels,
               fake_data=False,
               one_hot=False,
               dtype=dtypes.float32,
               reshape=True,
               seed=None):
    """Construct a DataSet.
    one_hot arg is used only if fake_data is true.  `dtype` can be either
    `uint8` to leave the input as `[0, 255]`, or `float32` to rescale into
    `[0, 1]`.  Seed arg provides for convenient deterministic testing.
    """
    seed1, seed2 = random_seed.get_seed(seed)
    # If op level seed is not set, use whatever graph level seed is returned
    numpy.random.seed(seed1 if seed is None else seed2)
    dtype = dtypes.as_dtype(dtype).base_dtype
    if dtype not in (dtypes.uint8, dtypes.float32):
      raise TypeError('Invalid image dtype %r, expected uint8 or float32' %
                      dtype)
    if fake_data:
      self._num_examples = 10000
      self.one_hot = one_hot
    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)
      if reshape:
        assert images.shape[3] == 1
        images = images.reshape(images.shape[0],
                                images.shape[1] * images.shape[2])
      if dtype == dtypes.float32:
        # 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, shuffle=True):
    """Return the next `batch_size` examples from this data set."""
    if fake_data:
      fake_image = [1] * 784
      if self.one_hot:
        fake_label = [1] + [0] * 9
      else:
        fake_label = 0
      return [fake_image for _ in xrange(batch_size)], [
          fake_label for _ in xrange(batch_size)
      ]
    start = self._index_in_epoch
    # Shuffle for the first epoch
    if self._epochs_completed == 0 and start == 0 and shuffle:
      perm0 = numpy.arange(self._num_examples)
      numpy.random.shuffle(perm0)
      self._images = self.images[perm0]
      self._labels = self.labels[perm0]
    # Go to the next epoch
    if start + batch_size > self._num_examples:
      # Finished epoch
      self._epochs_completed += 1
      # Get the rest examples in this epoch
      rest_num_examples = self._num_examples - start
      images_rest_part = self._images[start:self._num_examples]
      labels_rest_part = self._labels[start:self._num_examples]
      # Shuffle the data
      if shuffle:
        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 - rest_num_examples
      end = self._index_in_epoch
      images_new_part = self._images[start:end]
      labels_new_part = self._labels[start:end]
      return numpy.concatenate((images_rest_part, images_new_part), axis=0) , numpy.concatenate((labels_rest_part, labels_new_part), axis=0)
    else:
      self._index_in_epoch += batch_size
      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,
                   dtype=dtypes.float32,
                   reshape=True,
                   validation_size=5000,
                   seed=None):
  if fake_data:

    def fake():
      return DataSet(
          [], [], fake_data=True, one_hot=one_hot, dtype=dtype, seed=seed)

    train = fake()
    validation = fake()
    test = fake()
    return base.Datasets(train=train, validation=validation, test=test)

  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'

  local_file = base.maybe_download(TRAIN_IMAGES, train_dir,
                                   SOURCE_URL + TRAIN_IMAGES)
  with open(local_file, 'rb') as f:
    train_images = extract_images(f)

  local_file = base.maybe_download(TRAIN_LABELS, train_dir,
                                   SOURCE_URL + TRAIN_LABELS)
  with open(local_file, 'rb') as f:
    train_labels = extract_labels(f, one_hot=one_hot)

  local_file = base.maybe_download(TEST_IMAGES, train_dir,
                                   SOURCE_URL + TEST_IMAGES)
  with open(local_file, 'rb') as f:
    test_images = extract_images(f)

  local_file = base.maybe_download(TEST_LABELS, train_dir,
                                   SOURCE_URL + TEST_LABELS)
  with open(local_file, 'rb') as f:
    test_labels = extract_labels(f, one_hot=one_hot)

  if not 0 <= validation_size <= len(train_images):
    raise ValueError(
        'Validation size should be between 0 and {}. Received: {}.'
        .format(len(train_images), validation_size))

  validation_images = train_images[:validation_size]
  validation_labels = train_labels[:validation_size]
  train_images = train_images[validation_size:]
  train_labels = train_labels[validation_size:]

  
  options = dict(dtype=dtype, reshape=reshape, seed=seed)
  
  train = DataSet(train_images, train_labels, **options)
  validation = DataSet(validation_images, validation_labels, **options)
  test = DataSet(test_images, test_labels, **options)
  
  return base.Datasets(train=train, validation=validation, test=test)


def load_mnist(train_dir='MNIST-data'):
  return read_data_sets(train_dir)

探寻有趣之事！

软培

机器学习的概念

有监督学习(Supervised Learning)：

无监督学习(Non-Supervised Learning)：

典型的监督学习

贝叶斯概率

朴素贝叶斯分类器

高斯朴素贝叶斯

hello.py

实验指导：

实现一个线性回归的例子

实现一个矩阵乘法计算

常量/变量/赋值

Placeholder用法

可视化

ex-06-vis.py

ex-09-tb.py

ex-10.py

input_data.py

mnist.py

热门文章

最新文章

相关电子书

软培

机器学习的概念

有 监 督 学 习(Supervised Learning)：

无 监 督 学 习(Non-Supervised Learning)：

典型的监督学习

贝叶斯概率

朴素贝叶斯分类器

高斯朴素贝叶斯

hello.py

实验指导：

实现一个线性回归的例子

实现一个矩阵乘法计算

常量/变量/赋值

Placeholder用法

可视化

ex-06-vis.py

ex-09-tb.py

ex-10.py

input_data.py

mnist.py

热门文章

最新文章

相关电子书

有监督学习(Supervised Learning)：

无监督学习(Non-Supervised Learning)：