基于TensorFlow实现卷积神经网络 2

Tensorflow建造神经网络

一、实验介绍

1.1 实验内容

1.2 实验知识点

dropout 层

Tensorboard

1.3 实验环境

二、 神经网络

2.1 神经网络是什么？

814 使用python实现深度神经网络

Google Tensorflow

神经网络 playground

playground

玩深度神经网络

2.2 神经网络究竟在干嘛？

y = a*x + b

a=0.5，b=2

x=3

0.5*3 + 2

2.3 神经网络是如何被训练的?

2.4 激励函数

三、搭建神经网络

3.1 基本流程

3.2 开始搭建

3.2.1 准备

#激活TensorFlow环境
$ cd /home/shiyanlou/tensorflow
$ source bin/activate

myNetWork.py

$ sudo gedit myNetWork.py

# -*- coding: UTF-8 -*-

import tensorflow as tf
import numpy as np

3.2.2 训练数据

Numpy API

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

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

def add_layer(inputs, in_size, out_size, activation_function=None):
#定义参数 Weights，biases
Weights = tf.Variable(tf.random_normal([in_size, out_size]))
biases = tf.Variable(tf.zeros([1, out_size]) + 0.1)
#定义拟合公式
Wx_plus_b = tf.matmul(inputs, Weights) + biases
#激励函数
if activation_function is None:
outputs = Wx_plus_b
else:
outputs = activation_function(Wx_plus_b)
return outputs

add_layer(inputs, in_size, out_size, activation_function=None)

# 添加隐藏层（hidden layer） 输入值是 xs，在隐藏层有 10 个神经元，使用ReLu激活函数
# 输入 1 维数据，经隐藏层处理输出 10 维数据
l1 = add_layer(xs, 1, 10, activation_function=tf.nn.relu)

# 添加输出层（output layer） 输入值是隐藏层 l1，在预测层输出 1 个结果
# 从隐藏层输入得到的 10 维数据，经输出层处理 输出 1 维数据
prediction = add_layer(l1, 10, 1, activation_function=None)

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

#最基本的 Optimizer：Gradient Descent，学习率=0.1
optimizer = tf.train.GradientDescentOptimizer(0.1)
train_step = optimizer.minimize(loss)

init = tf.initialize_all_variables()

Session

sess = tf.Session()
# 上面操作在 TensorFlow 中都没有产生运算，直到 sess.run 才会开始运算
sess.run(init)

# 迭代 1000 次学习，sess.run optimizer
for i in range(1000):
# training train_step 和 loss 都是由 placeholder 定义的运算，所以这里要用 feed 传入参数， placeholder 和 feed_dict 是绑定用的。
sess.run(train_step, feed_dict={xs: x_data, ys: y_data})
if i % 50 == 0:
# 观察每一步的变化
print(sess.run(loss, feed_dict={xs: x_data, ys: y_data}))

myNetWork1.py

(tensorflow) $ python myNetWork.py

3.3 改进神经网络

overfitting

ensemble

注意是暂时，对于随机梯度下降来说，由于是随机丢弃，故而每一个mini-batch都在训练不同的网络。

2^n

dropout

add layer

dropout

keep_prob

def add_layer(inputs, in_size, out_size, activation_function=None):
#定义参数 Weights，biases
Weights = tf.Variable(tf.random_normal([in_size, out_size]))
biases = tf.Variable(tf.zeros([1, out_size]) + 0.1)
#定义拟合公式
Wx_plus_b = tf.matmul(inputs, Weights) + biases
#隐含节点dropout率等于0.5的时候效果最好，即 keep_prob=0.5，原因是0.5的时候dropout随机生成的网络结构最多。
Wx_plus_b = tf.nn.dropout(Wx_plus_b, keep_prob=0.5)
#激励函数
if activation_function is None:
outputs = Wx_plus_b
else:
outputs = activation_function(Wx_plus_b)
return outputs

$ cd /home/shiyanlou/tensorflow
$ wget http://labfile.oss.aliyuncs.com/courses/893/myNetWork1.py

3.4 可视化

Tensorboard

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
# 区别：大框架，定义层 layer，里面有小部件
with tf.name_scope('layer'):
# 区别：小部件
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

# 区别：大框架，里面有 inputs x，y
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)

# 区别：定义框架 loss
with tf.name_scope('loss'):
loss = tf.reduce_mean(tf.reduce_sum(tf.square(ys - prediction),
reduction_indices=[1]))
# 区别：定义框架 train
with tf.name_scope('train'):
train_step = tf.train.GradientDescentOptimizer(0.1).minimize(loss)

sess = tf.Session()

# 区别：sess.graph 把所有框架加载到一个文件中放到文件夹"logs/"里
# 接着打开terminal，进入你存放的文件夹地址上一层，运行命令 tensorboard --logdir='logs/'
# 会返回一个地址，然后用浏览器打开这个地址，在 graph 标签栏下打开
writer = tf.train.SummaryWriter("logs/", sess.graph)
# important step
sess.run(tf.initialize_all_variables())

$ cd /home/shiyanlou/tensorflow
$ wget http://labfile.oss.aliyuncs.com/courses/893/myNetWork2.py

3.5 保存和加载

$ cd /home/shiyanlou/tensorflow
$ sudo mkdir my_net

my_net

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')

init= tf.initialize_all_variables()

saver = tf.train.Saver()

# 用 saver 将所有的 variable 保存到定义的路径
with tf.Session() as sess:
sess.run(init)
save_path = saver.save(sess, "my_net/save_net.ckpt")
print("Save to path: ", save_path)

$ cd /home/shiyanlou/tensorflow
$ wget http://labfile.oss.aliyuncs.com/courses/893/myNetWork3.py

# restore variables
# redefine the same shape and same type for your variables
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()
# 用 saver 从路径中将 save_net.ckpt 保存的 W 和 b restore 进来
with tf.Session() as sess:
saver.restore(sess, "my_net/save_net.ckpt")
print("weights:", sess.run(W))
print("biases:", sess.run(b))

$ cd /home/shiyanlou/tensorflow
$ wget http://labfile.oss.aliyuncs.com/courses/893/myNetWork4.py

四、实验总结

dropout layer

TensorBoard

卷积神经网络（CNN）

五、课后习题

卷积神经网络详解章节

六、参考链接