一.sessoion
session会话:
指定到某个神经网络
#encoding:utf-8
import tensorflow as tf
import numpy as np
##定义一个向量##
##运算两个矩阵##
martrix1=tf.constant([[3,3]])
martrix2=tf.constant([[2],[2]])
product=tf.matmul(matrix1,matrix2)
##np.dot()相当于##
##seeeion用来运行结果##
sess=tf.Session()
result=sess.run(product)
print(result)
sess.close()二.variable
使用变量实现一个简单的计数器
#encoding:utf-8
import tensorflow as tf
state=tf.Variable(0,name='counter')
print(state.name)
#tf.constant()#定义一个常量
#state=tf.Variable(0)
one=tf.constant(1)#常数为1
new_value=tf.add(state,one)
update=tf.assign(state,new_value)#new_value赋值给statue
init=tf.initialize_all_variables()#初始化所有的变量
#运算
with tf.Session() as sess:
sess.run(init)# run and active激活rUn步骤
for _ in range(3):
sess.run(update)
print(sess.run(state))结果：
counter:0
WARNING:tensorflow:From /home/sulei/tensorflow-workspace/variable.py:10: 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.
2017-06-27 11:23:49.419560: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.1 instructions, but these are available on your machine and could speed up CPU computations.
2017-06-27 11:23:49.419613: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.2 instructions, but these are available on your machine and could speed up CPU computations.
2017-06-27 11:23:49.419631: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX instructions, but these are available on your machine and could speed up CPU computations.
1
2
3三.Feeds的传入值:
tf.placeholder() 为这些操作创建占位符
外部对神经网络的传值
错现的错误
AttributeError: module ‘tensorflow’ has no attribute ‘mul’
解决的方法：
将mul改成multiply，则可以使用
import tensorflow as tf
input1=tf.placeholder(tf.float32)
input2=tf.placeholder(tf.float32)
ouput=tf.multiply(input1,input2)#multyiply：乘法
with tf.Session() as sess:
print(sess.run(ouput,feed_dict={input1:[7.],input2:[2.]}))
#字典的函数/usr/bin/python3.5 /home/sulei/tensorflow-workspace/feeds.py
2017-06-27 13:43:13.138963: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.1 instructions, but these are available on your machine and could speed up CPU computations.
2017-06-27 13:43:13.139033: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.2 instructions, but these are available on your machine and could speed up CPU computations.
2017-06-27 13:43:13.139056: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX instructions, but these are available on your machine and could speed up CPU computations.
[ 14.]
Process finished with exit code 0
四、激励函数
激励方程：
使某一部分的神经元先激励
Activate Function的使用方法：参考链接
五.定义激励函数和添加一个神经层以及建造神经网络
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])) #定义一个权重
biases=tf.Variable(tf.zeros([1,out_size])+0.1)
Wx_plus_b=tf.matmul(inputs,Weights)+biases#预测出来的值
if activation_function is None:
outputs=Wx_plus_b
else:
outputs=activation_function(Wx_plus_b)
return outputs
x_data=np.linspace(-1,1,300)[:,np.newaxis]
noise=np.random.normal(0,0.05,x_data.shape)#噪点
y_data=np.square(x_data)-0.5+noise
xs=tf.placeholder(tf.float32,[None,1])
ys=tf.placeholder(tf.float32,[None,1])
#输入层/隐藏层/输出层
l1=add_layer(xs,1,10,activation_function=tf.nn.relu)
prediction=add_layer(l1,10,1,activation_function=None)
loss=tf.reduce_mean(tf.reduce_sum(tf.square(ys-prediction),
reduction_indices=[1]))
#顺年,练习的步骤
train_step=tf.train.GradientDescentOptimizer(0.1).minimize(loss)
init=tf.initialize_all_variables()
sess=tf.Session()
sess.run(init)
for i in range(1000):
sess.run(train_step,feed_dict={xs:x_data,ys:y_data})
if i%50:
print(sess.run(loss,feed_dict={xs:x_data,ys:y_data}))
结果：
loss的值会越来越小六.可视化
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])) #定义一个权重
biases=tf.Variable(tf.zeros([1,out_size])+0.1)
Wx_plus_b=tf.matmul(inputs,Weights)+biases#预测出来的值
if activation_function is None:
outputs=Wx_plus_b
else:
outputs=activation_function(Wx_plus_b)
return outputs
x_data=np.linspace(-1,1,300)[:,np.newaxis]
noise=np.random.normal(0,0.05,x_data.shape)#噪点
y_data=np.square(x_data)-0.5+noise
xs=tf.placeholder(tf.float32,[None,1])
ys=tf.placeholder(tf.float32,[None,1])
#输入层/隐藏层/输出层
l1=add_layer(xs,1,10,activation_function=tf.nn.relu)
prediction=add_layer(l1,10,1,activation_function=None)
loss=tf.reduce_mean(tf.reduce_sum(tf.square(ys-prediction),
reduction_indices=[1]))
#顺年,练习的步骤
train_step=tf.train.GradientDescentOptimizer(0.1).minimize(loss)
init=tf.initialize_all_variables()
sess=tf.Session()
sess.run(init)
fig=plt.figure()
ax=fig.add_subplot(1,1,1)#连续性
ax.scatter(x_data,y_data)
plt.show()
for i in range(1000):
sess.run(train_step,feed_dict={xs:x_data,ys:y_data})
if i%50:
print(sess.run(loss,feed_dict={xs:x_data,ys:y_data}))
#prediction_value=sess.run(prediction,feed_dict={xs:x_data})
#lines=ax.plot(x_data,prediction_value,'r-',lw=5)
#ax.lines.remove()
动态的可视化的模型趋近：
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])) #定义一个权重
biases=tf.Variable(tf.zeros([1,out_size])+0.1)
Wx_plus_b=tf.matmul(inputs,Weights)+biases#预测出来的值
if activation_function is None:
outputs=Wx_plus_b
else:
outputs=activation_function(Wx_plus_b)
return outputs
x_data=np.linspace(-1,1,300)[:,np.newaxis]
noise=np.random.normal(0,0.05,x_data.shape)#噪点
y_data=np.square(x_data)-0.5+noise
xs=tf.placeholder(tf.float32,[None,1])
ys=tf.placeholder(tf.float32,[None,1])
#输入层/隐藏层/输出层
l1=add_layer(xs,1,10,activation_function=tf.nn.relu)
prediction=add_layer(l1,10,1,activation_function=None)
loss=tf.reduce_mean(tf.reduce_sum(tf.square(ys-prediction),
reduction_indices=[1]))
#顺年,练习的步骤
train_step=tf.train.GradientDescentOptimizer(0.1).minimize(loss)
init=tf.initialize_all_variables()
sess=tf.Session()
sess.run(init)
fig=plt.figure()
ax=fig.add_subplot(1,1,1)#连续性
ax.scatter(x_data,y_data)
plt.ion()#连续显示图形,不暂停
plt.show()
for i in range(1000):
sess.run(train_step,feed_dict={xs:x_data,ys:y_data})
if i%50:
#print(sess.run(loss,feed_dict={xs:x_data,ys:y_data}))
try:
ax.lines.remove(lines[0]) # 抹除曲线
except Exception:
pass
prediction_value=sess.run(prediction,feed_dict={xs:x_data})
lines=ax.plot(x_data,prediction_value,'r-',lw=5)#曲线的形式
plt.pause(0.1)#暂停0.1s
Tensorboard 可视化好帮手
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt #导入可视化模块
def add_layer(inputs, in_size, out_size, activation_function=None):
#activation_function=None表示没有激活函数 相当于是线性模型
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')#偏置最好不为零所以加了0.1
with tf.name_scope('Wx_plus_b'):
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
# 定义神经网络输入的placeholder
with tf.name_scope('inputs'):
xs = tf.placeholder(tf.float32,[None,1], name = "x_input")
#None表示可以输入任意的数据。因为x_data是300x1的矩阵，所以这里为[None,1]
ys = tf.placeholder(tf.float32,[None,1], name = 'y_input')
#隐藏层layer1 输入节点1，输出节点10
l1 = add_layer(xs, 1, 10, activation_function = tf.nn.relu)
#预测的时候输入时隐藏层的输入l1,输入节点10，输出为y_data 有1个节点
prediction = add_layer(l1, 10, 1, activation_function = None)
#计算损失
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)
init = tf.global_variables_initializer()
#tf.initialize_all_variables()运行的时候会提示你现在的新名称是tf.global_variables_initializer()
sess = tf.Session()
writer = tf.summary.FileWriter("/home/sulei/tensorflow-workspace/logs/",sess.graph)
#writer = tf.summary.FileWriter("/*****/",sess.graph)
# 先加载到一个文件中，然后再加载到浏览器里观看，双引号里标出存放这个文件的路径
# 莫烦视频中是tf.train.SummaryWriter()，我运行报错后发现这个函数名称被改为了tf.summary.FileWriter()
sess.run(init)之后在events.out.tfevents.1499002324的文件夹之下使用如下的命令
sulei@sulei:~/tensorflow-workspace/logs$ tensorbord --logdir='/home/sulei/tensorflow-workspace/logs/'
展示的结果：
sulei@sulei:~/tensorflow-workspace/logs$ tensorboard --logdir='/home/sulei/tenrflow-workspace/logs/'
Starting TensorBoard 54 at http://sulei:6006
(Press CTRL+C to quit)
打开链接：http://sulei:6006