这是学习TensorFlow过程中记录的一点笔记，在此分享一下。

TensorFlow简介

TensorFlow是一个端到端机器学习开源平台。它有一个全面、灵活的工具、库和社区资源生态系统，可以让研究人员推送最先进的ML，让开发人员轻松构建和部署基于ML的应用程序。

TensorFlow最初是由谷歌机器智能研究机构谷歌大脑团队的研究人员和工程师开发的，用于进行机器学习和深度神经网络研究。该系统的通用性足以适用于各种各样的其他领域。

TensorFlow提供稳定的Python和c++ API，以及为其他语言提供非保证向后兼容的API。

github
中文官网

语法入门：

图片笔记来自一个群友的分享

安装

安装当前发布版本，包括cuda的GPU支持(Ubuntu和Windows):

pip install tensorflow

安装cpu版本

pip install tensorflow-cpu

教程

一个小例子

import tensorflow as tf
import numpy as np

# create data
x_data = np.random.rand(100).astype(np.float32)
y_data = x_data*0.1 + 0.3
pr
### create tensorflow structure start ###
Weights = tf.Variable(tf.random.uniform([1], -1.0, 1.0)) #权重
biases = tf.Variable(tf.zeros([1]))   #偏置

y = Weights*x_data + biases #预测值

loss = tf.reduce_mean(tf.square(y-y_data))#预测值与真实值的偏差
optimizer = tf.train.GradientDescentOptimizer(0.5)#使用梯度下降给最优法
train = optimizer.minimize(loss)
### create tensorflow structure end ###

sess = tf.Session()
# tf.initialize_all_variables() no long valid from
# 2017-03-02 if using tensorflow >= 0.12
if int((tf.__version__).split('.')[1]) < 12 and int((tf.__version__).split('.')[0]) < 1:
    init = tf.initialize_all_variables()# 初始化所有变量 版本低于0.2
else:
    init = tf.global_variables_initializer()# 其他版本使用初始化方法
sess.run(init) #开始执行初始化值 

for step in range(201):
    sess.run(train) #开始训练数据
    if step % 20 == 0:
        print(step, sess.run(Weights), sess.run(biases))

0 [0.8153867] [-0.11029217]
20 [0.30153248] [0.19439411]
40 [0.15564139] [0.27084312]
60 [0.11536213] [0.29195005]
80 [0.10424136] [0.29777747]
100 [0.10117102] [0.29938638]
120 [0.10032333] [0.2998306]
140 [0.10008927] [0.29995322]
160 [0.10002465] [0.29998708]
180 [0.10000681] [0.29999644]
200 [0.10000189] [0.29999903]

Session会话控制

import tensorflow as tf
# 创建两个矩阵相乘
matrix1 = tf.constant([[3,3]])# 一行两列
matrix2 = tf.constant([[2],[2]])# 两行一列
product = tf.matmul(matrix1,matrix2) #multiply 相当于np.dot 
#method 1
sess = tf.Session() #创建一个会话session
result = sess.run(product)# 执行我们上面定义的操作
print(result)
sess.close()# 要记得把session关闭
# method 2
with tf.Session () as sess:
    result2 = sess.run(product)
    print(result2)
    

Variable变量

import tensorflow as tf

state = tf.Variable(0,name='counter')#创建一个用来计数的变量
#可以使用 print(state.name)访问到state的名字
one = tf.constant(1) #定义一个常量 1

new_value = tf.add(state,one)# 把one 添加到state里面
updata = tf.assign(state,new_value)#把new_value加载到state上

init = tf.global_variables_initializer()

with tf.Session() as sess:#开启一个会话
    sess.run(init)#执行初始化操作
    for _ in range(3):# __做变量名代表不需这个值 只读不写
        sess.run(updata)#执行更新操作
        print(sess.run(state))# sess 是一个指针 只有run一下state 才能出来state的结果

1
2
3

placeholder 传入值

input1 = tf.placeholder(tf.float32)# 也可以在参数中加上[2,2]定义输入值的结构二行二列
input2 = tf.placeholder(tf.float32)
output = tf.multiply(input1,input2)

with tf.Session() as sess:
    print(sess.run(output,feed_dict={input1:[7.],input2:[2.]}))#要以字典的形式传入输入值

[14.]

activation funtion激励函数

tf.nn.elu

tf.nn.softplus

tf.nn.softsign

tf.nn.dropout

建造一个神经网络

import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
#添加层 add_layer
def add_layer(inputs,in_size,out_size,activation_function=None):
    Weights = tf.Variable(tf.random_normal([in_size,out_size]))# 矩阵 大小为输入值size 行 输出值size 列
    biases = tf.Variable(tf.zeros([1,out_size])+0.1)#一般偏置项不为0 所以给初始值0.11
    Wx_plus_b = tf.matmul(inputs,Weights)+ biases#预测值 还没有被激活的值存储在则会个变量里面
    if activation_function is None:
        outputs = Wx_plus_b# 当af为None的时候 保持现状 即一个线性的激活函数
    else:
        outputs = activation_function(Wx_plus_b)# 如果af不为None 就把当前值传入已有的激活函数
    return outputs
#make up some real data
x_data = np.linspace(-1,1,300)[:,np.newaxis]#添加一个维度 300行的数据 一列
noise = np.random.normal(0,0.05,x_data.shape)#添加一个噪点 使数据更真实
y_data = np.square(x_data)- 0.5+ noise

#define placeholder for inputs to network
xs = tf.placeholder(tf.float32,[None,1])#None表示输入的个数不限
ys = tf.placeholder(tf.float32,[None,1])
#add hidden layer
l1 = add_layer(xs,1,10,activation_function=tf.nn.relu)
#add output layer
prediction = add_layer(l1,10,1,activation_function=None)
#the error between prediction and real data
# 求的是均方误差
loss = tf.reduce_mean(tf.reduce_sum(tf.square(ys-prediction),reduction_indices=[1]))
train_step = tf.train.GradientDescentOptimizer(0.1).minimize(loss)
#important step
inint = tf.global_variables_initializer()
sess = tf.Session()
sess.run(inint)
#plot the real data
fig = plt.figure()
ax = fig.add_subplot(1,1,1)
ax.scatter(x_data,y_data)
# show()函数会把图片显示后 让程序暂停  如果不想让程序暂停 可以加上ion（）函数
plt.ion()
plt.show()
for i in range(1000):
    # start training
    sess.run(train_step,feed_dict={xs:x_data,ys:y_data})
    if i%50==0:
        #to see the step improvement
        #print(sess.run(loss,feed_dict={xs:x_data,ys:y_data}))
        # to visualize the result and improvement
        try:
            ax.lines.remove(lines[0])
        except Exception:
            pass
        #plot the prediction
        prediction_value = sess.run(prediction,feed_dict={xs:x_data})
        lines = ax.plot(x_data,prediction_value,'r-',lw=5)
        plt.pause(0.1)

优化器 optimizer

新手的话用GradientDescentOptimizer 就足够了

tensorboard 可视化助手

例子1

import tensorflow as tf
def add_layer(inputs, in_size, out_size, activation_function=None):
    # add one more layer and return the output of this layer
    with tf.name_scope('layer'):
        with tf.name_scope('weights'):
            Weights = tf.Variable(tf.random_normal([in_size, out_size]), name='W')
        with tf.name_scope('biases'):
            biases = tf.Variable(tf.zeros([1, out_size]) + 0.1, name='b')
        with tf.name_scope('Wx_plus_b'):
            Wx_plus_b = tf.add(tf.matmul(inputs, Weights), biases)
        if activation_function is None:
            outputs = Wx_plus_b
        else:
            outputs = activation_function(Wx_plus_b, )
        return outputs


# define placeholder for inputs to network
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')

# add hidden layer
l1 = add_layer(xs, 1, 10, activation_function=tf.nn.relu)
# add output layer
prediction = add_layer(l1, 10, 1, activation_function=None)

# the error between prediciton and real data
with tf.name_scope('loss'):
    loss = tf.reduce_mean(tf.reduce_sum(tf.square(ys - prediction),
                                        reduction_indices=[1]))

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

sess = tf.Session()

# tf.train.SummaryWriter soon be deprecated, use following
if int((tf.__version__).split('.')[1]) < 12 and int((tf.__version__).split('.')[0]) < 1:  # tensorflow version < 0.12
    writer = tf.train.SummaryWriter('logs/', sess.graph)
else: # tensorflow version >= 0.12
    writer = tf.summary.FileWriter("logs/", sess.graph)

# tf.initialize_all_variables() no long valid from
# 2017-03-02 if using tensorflow >= 0.12
if int((tf.__version__).split('.')[1]) < 12 and int((tf.__version__).split('.')[0]) < 1:
    init = tf.initialize_all_variables()
else:
    init = tf.global_variables_initializer()
sess.run(init)

# direct to the local dir and run this in terminal:
# $ tensorboard --logdir=logs

例子2

import tensorflow as tf
import numpy as np


def add_layer(inputs, in_size, out_size, n_layer, activation_function=None):
    # add one more layer and return the output of this layer
    layer_name = 'layer%s' % n_layer
    with tf.name_scope(layer_name):
        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('biases'):
            biases = tf.Variable(tf.zeros([1, out_size]) + 0.1, name='b')
            tf.summary.histogram(layer_name + '/biases', biases)
        with tf.name_scope('Wx_plus_b'):
            Wx_plus_b = tf.add(tf.matmul(inputs, Weights), biases)
        if activation_function is None:
            outputs = Wx_plus_b
        else:
            outputs = activation_function(Wx_plus_b, )
        tf.summary.histogram(layer_name + '/outputs', outputs)
    return outputs


# Make up some real data
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) - 0.5 + noise

# define placeholder for inputs to network
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')

# add hidden layer
l1 = add_layer(xs, 1, 10, n_layer=1, activation_function=tf.nn.relu)
# add output layer
prediction = add_layer(l1, 10, 1, n_layer=2, activation_function=None)

# the error between prediciton and real data
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_step = tf.train.GradientDescentOptimizer(0.1).minimize(loss)

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

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

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

for i in range(1000):
    sess.run(train_step, feed_dict={xs: x_data, ys: y_data})
    if i % 50 == 0:
        result = sess.run(merged,
                          feed_dict={xs: x_data, ys: y_data})
        writer.add_summary(result, i)

# direct to the local dir and run this in terminal:
# $ tensorboard --logdir logs

Classification分类学习

import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
# number 1 to 10 data
#下面这行代码作用是如果没有下载数据集的话 会自动下载
mnist = input_data.read_data_sets('MNIST_data', one_hot=True)

def add_layer(inputs, in_size, out_size, activation_function=None,):
    # add one more layer and return the output of this layer
    Weights = tf.Variable(tf.random_normal([in_size, out_size]))
    biases = tf.Variable(tf.zeros([1, out_size]) + 0.1,)
    Wx_plus_b = tf.matmul(inputs, Weights) + biases
    if activation_function is None:
        outputs = Wx_plus_b
    else:
        outputs = activation_function(Wx_plus_b,)
    return outputs

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

# define placeholder for inputs to network
xs = tf.placeholder(tf.float32, [None, 784]) # 28x28
ys = tf.placeholder(tf.float32, [None, 10])

# add output layer softmax一般是用来做分类的
prediction = add_layer(xs, 784, 10,  activation_function=tf.nn.softmax)

# the error between prediction and real data
cross_entropy = tf.reduce_mean(-tf.reduce_sum(ys * tf.log(prediction),
                                              reduction_indices=[1]))       # loss
train_step = tf.train.GradientDescentOptimizer(0.5).minimize(cross_entropy)

sess = tf.Session()
# important step
# tf.initialize_all_variables() no long valid from
# 2017-03-02 if using tensorflow >= 0.12
if int((tf.__version__).split('.')[1]) < 12 and int((tf.__version__).split('.')[0]) < 1:
    init = tf.initialize_all_variables()
else:
    init = tf.global_variables_initializer()
sess.run(init)

for i in range(1000):
    #为了提高训练效率 每次训练100个
    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(compute_accuracy(
            mnist.test.images, mnist.test.labels))

Extracting MNIST_data\train-images-idx3-ubyte.gz
Extracting MNIST_data\train-labels-idx1-ubyte.gz
Extracting MNIST_data\t10k-images-idx3-ubyte.gz
Extracting MNIST_data\t10k-labels-idx1-ubyte.gz
0.0567
0.6315
0.7266
0.7707
0.8026
0.8193
0.8267
0.8389
0.843
0.8457
0.8491
0.8543
0.8618
0.8602
0.8606
0.8634
0.8636
0.8673
0.8701
0.8721

卷积神经网络

import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
# number 1 to 10 data
mnist = input_data.read_data_sets('MNIST_data', one_hot=True)
#该函数用来计算识别的精度
def compute_accuracy(v_xs, v_ys):
    global prediction
    y_pre = sess.run(prediction, feed_dict={xs: v_xs, keep_prob: 1})
    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, keep_prob: 1})
    return result

def weight_variable(shape):
    initial = tf.truncated_normal(shape, stddev=0.1)#初始化
    return tf.Variable(initial)

def bias_variable(shape):
    initial = tf.constant(0.1, shape=shape)#偏置初始值一般为正值
    return tf.Variable(initial)

def conv2d(x, W):
    # stride [1, x_movement, y_movement, 1] 
    #对应featuremap，含义[batch,height,weight,channels],
    #因为卷积池化过程中不需要对batch channels进行操作 所以这两个值是1
    # Must have strides[0] = strides[3] = 1
    return tf.nn.conv2d(x, W, strides=[1, 1, 1, 1], padding='SAME')
    #padding 填充 当是SAME时在原图片外围填充一圈0 输出图片尺寸等于输入图片尺寸/stride
    #vaild时 （size-2*stride）/2+1
def max_pool_2x2(x):
    # stride [1, x_movement, y_movement, 1] 池化层 就是一个下采样 把图片放大
    return tf.nn.max_pool(x, ksize=[1,2,2,1], strides=[1,2,2,1], padding='SAME')
    #max_pool能更好的保存特征 
# define placeholder for inputs to network
xs = tf.placeholder(tf.float32, [None, 784])/255.   # 输入为28x28的图片
ys = tf.placeholder(tf.float32, [None, 10])#输出为表示0-9的值
keep_prob = tf.placeholder(tf.float32)
x_image = tf.reshape(xs, [-1, 28, 28, 1])#-1 表示不指定这一维大小 函数自己判断有多少张
# -1可以理解为导入数据有多少个图片  28*28*1表示图片的大小通道数
#和batch有所区别 batch_size是决定了一次训练取多少张图片
# print(x_image.shape)  # [n_samples, 28,28,1]

## conv1 layer ##
W_conv1 = weight_variable([5,5, 1,32]) # patch 5x5, in size 1, out size 32
#5*5 是patch的尺寸 1为本层输入的通道数 32是本层输出的depth 即本层卷积核的个数（filter的层数）
#卷积核 小区域加权平均的时候的权重由一个函数定义 这个函数称为卷积核 卷积核数目自己定义
b_conv1 = bias_variable([32])#每一个filter 都有一个bias
h_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) + b_conv1) # output size 28x28x32
#relu 该激励函数使输出非线性化
h_pool1 = max_pool_2x2(h_conv1)   #步长是2  所以经过池化层之后 输出为 output size 14x14x32

## conv2 layer ##
W_conv2 = weight_variable([5,5, 32, 64]) # patch 5x5, in size 32, out size 64
b_conv2 = bias_variable([64])
h_conv2 = tf.nn.relu(conv2d(h_pool1, W_conv2) + b_conv2) # output size 14x14x64
h_pool2 = max_pool_2x2(h_conv2)                                         # output size 7x7x64

## fc1 layer ##
W_fc1 = weight_variable([7*7*64, 1024])
b_fc1 = bias_variable([1024])
# [n_samples, 7, 7, 64] ->> [n_samples, 7*7*64]
h_pool2_flat = tf.reshape(h_pool2, [-1, 7*7*64])
h_fc1 = tf.nn.relu(tf.matmul(h_pool2_flat, W_fc1) + b_fc1)
h_fc1_drop = tf.nn.dropout(h_fc1, keep_prob)# 防止过拟合

## fc2 layer ##
W_fc2 = weight_variable([1024, 10])
b_fc2 = bias_variable([10])
prediction = tf.nn.softmax(tf.matmul(h_fc1_drop, W_fc2) + b_fc2)


# the error between prediction and real data
cross_entropy = tf.reduce_mean(-tf.reduce_sum(ys * tf.log(prediction),
                                              reduction_indices=[1]))       # loss
train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)

sess = tf.Session()
# important step
# tf.initialize_all_variables() no long valid from
# 2017-03-02 if using tensorflow >= 0.12
if int((tf.__version__).split('.')[1]) < 12 and int((tf.__version__).split('.')[0]) < 1:
    init = tf.initialize_all_variables()
else:
    init = tf.global_variables_initializer()
sess.run(init)

for i in range(1000):
    batch_xs, batch_ys = mnist.train.next_batch(100)
    sess.run(train_step, feed_dict={xs: batch_xs, ys: batch_ys, keep_prob: 0.5})
    if i % 50 == 0:
        print(compute_accuracy(
            mnist.test.images[:1000], mnist.test.labels[:1000]))

Extracting MNIST_data\train-images-idx3-ubyte.gz
Extracting MNIST_data\train-labels-idx1-ubyte.gz
Extracting MNIST_data\t10k-images-idx3-ubyte.gz
Extracting MNIST_data\t10k-labels-idx1-ubyte.gz
0.077
0.761
0.867
0.902
0.913
0.922
0.934
0.938
0.939
0.952
0.958
0.949
0.958
0.966
0.963
0.968
0.969
0.967
0.972
0.973

Saver 保存读取

保存

import tensorflow as tf
import numpy as np
#Save to file
#remember to define the same dtype and shape when restore
W = tf.Variable([[1,2,3],[3,4,5]], dtype=tf.float32, name='weights')
b = tf.Variable([[1,2,3]], dtype=tf.float32, name='biases')

if int((tf.__version__).split('.')[1]) < 12 and int((tf.__version__).split('.')[0]) < 1:
    init = tf.initialize_all_variables()
else:
    init = tf.global_variables_initializer()

saver = tf.train.Saver()

with tf.Session() as sess:
    sess.run(init)
    save_path = saver.save(sess, "my_net/save_net.ckpt")
    print('save to path',save_path)

save to path my_net/save_net.ckpt

读取

import tensorflow as tf
import numpy as np
#restore variables
# redefine the same shape and same type for your variables
tf.reset_default_graph()# Clears the default graph stack and resets the global default graph
W = tf.Variable(np.arange(6).reshape((2, 3)), dtype=tf.float32, name="weights")
b = tf.Variable(np.arange(3).reshape((1, 3)), dtype=tf.float32, name="biases")

# not need init step

saver = tf.train.Saver()
with tf.Session() as sess:
    saver.restore(sess, "my_net/save_net.ckpt")
    print("weights:", sess.run(W))
    print("biases:", sess.run(b))

INFO:tensorflow:Restoring parameters from my_net/save_net.ckpt
weights: [[1. 2. 3.]
 [3. 4. 5.]]
biases: [[1. 2. 3.]]

参考：https://www.bilibili.com/video/BV1Lx411j7ws