TensorFlow: tf.ConfigProto() and Session

转载链接 TensorFlow中的小知识：tf.ConfigProto()与Session

Also make one heretensorflowsmall neural network,代码如下：

#Complete neural network
import tensorflow as tf  #导入tensorflow模块
import numpy as np       #导入numpy模块
BATCH_SIZE = 8 #Indicates how many sets of data are fed to the neural network at a time
seed = 23455  #It outputs the same random number every time
 
#基于seed产生随机数
rng = np.random.RandomState(seed)
#随机数返回32行2列的矩阵,表示32组 数据作为输入
X = rng.rand(32,2)
#从X这个32行2列的矩阵中取出一行 Judge if less than1 给Y赋值1,如果不小于1 给Y赋值0
#as a label for the dataset（正确答案）,数据标注
Y = [[int(x0 + x1 < 1)] for (x0,x1) in X]
print('X:\n',X)
print('Y:\n',Y)
 
#1.定义神经网络的输入、参数和输出,Two features are entered herex,输出一个标签y_
x = tf.placeholder(tf.float32,shape=(None,2))    #利用占位符placeholder,输入float32的 A tensor representation of one row and two columns 输入n组特征
y_ = tf.placeholder(tf.float32,shape=(None,1))    #利用占位符placeholder,输入float32的 A tensor representation of one row and two columns 输入n组标签
 
w1 = tf.Variable(tf.random_normal([2,3],stddev=1,seed=1))    #第一层网络,A matrix of normally distributed random numbers with two rows and three columns,标准差为1,随机种子为1
w2 = tf.Variable(tf.random_normal([3,1],stddev=1,seed=1))    #第一层网络,A matrix of normally distributed random numbers with three rows and one column,标准差为1,随机种子为1
 
#2.定义前向传播过程
a = tf.matmul(x,w1)       #xthrough the first layer of the network,Matrix weighted multiplication
y = tf.matmul(a,w2)       #through the second layer of the network,Matrix weighted multiplication,That is, the output of the last layer
 
#3.Define the loss function and backpropagation method
loss = tf.reduce_mean(tf.square(y-y_))
train_step = tf.train.GradientDescentOptimizer(0.001).minimize(loss)  #梯度下降优化器  
 
#4.生成会话,训练STEPS轮
sess = tf.Session()
init_op = tf.global_variables_initializer()
sess.run(init_op)
#Output the current untrained parameter values
#print('w1:',sess.run(w1))
#print('w2:',sess.run(w2))
 
#训练模型
STEPS = 5000   #训练5000轮
for i in range(STEPS):
    start = (i*BATCH_SIZE) % 32
    end = start + BATCH_SIZE
    sess.run(train_step,feed_dict={x: X[start:end],y_: Y[start:end]})       #Take the corresponding data from the initial dataset and labels,feed into the neural network
    if i % 500 == 0:            #每500轮打印一次loss值
        total_loss = sess.run(loss,feed_dict={x:X,y_:Y})
        print('经过%d轮训练,lossAll values ​​are：%g'%(i,total_loss))
 
#输出训练后的参数取值
print('输出训练后的参数取值：')
print('w1:\n',sess.run(w1))
print('w2:\n',sess.run(w2))

输出如下：

X:
 [[0.83494319 0.11482951]
 [0.66899751 0.46594987]
 [0.60181666 0.58838408]
 [0.31836656 0.20502072]
 [0.87043944 0.02679395]
 [0.41539811 0.43938369]
 [0.68635684 0.24833404]
 [0.97315228 0.68541849]
 [0.03081617 0.89479913]
 [0.24665715 0.28584862]
 [0.31375667 0.47718349]
 [0.56689254 0.77079148]
 [0.7321604  0.35828963]
 [0.15724842 0.94294584]
 [0.34933722 0.84634483]
 [0.50304053 0.81299619]
 [0.23869886 0.9895604 ]
 [0.4636501  0.32531094]
 [0.36510487 0.97365522]
 [0.73350238 0.83833013]
 [0.61810158 0.12580353]
 [0.59274817 0.18779828]
 [0.87150299 0.34679501]
 [0.25883219 0.50002932]
 [0.75690948 0.83429824]
 [0.29316649 0.05646578]
 [0.10409134 0.88235166]
 [0.06727785 0.57784761]
 [0.38492705 0.48384792]
 [0.69234428 0.19687348]
 [0.42783492 0.73416985]
 [0.09696069 0.04883936]]
Y:
 [[1], [0], [0], [1], [1], [1], [1], [0], [1], [1], [1], [0], [0], [0], [0], [0], [0], [1], [0], [0], [1], [1], [0], [1], [0], [1], [1], [1], [1], [1], [0], [1]]
经过0轮训练,lossAll values ​​are：5.13118
经过500轮训练,lossAll values ​​are：0.429111
经过1000轮训练,lossAll values ​​are：0.409789
经过1500轮训练,lossAll values ​​are：0.399923
经过2000轮训练,lossAll values ​​are：0.394146
经过2500轮训练,lossAll values ​​are：0.390597
经过3000轮训练,lossAll values ​​are：0.388336
经过3500轮训练,lossAll values ​​are：0.386855
经过4000轮训练,lossAll values ​​are：0.385863
经过4500轮训练,lossAll values ​​are：0.385186
输出训练后的参数取值：
w1:
 [[-0.69597054  0.8599247   0.0933773 ]
 [-2.3418374  -0.12466972  0.5866561 ]]
w2:
 [[-0.07531644]
 [ 0.8627887 ]
 [-0.05937821]]

总结：

神经网络的实现过程（八股）：准备、前传、反传、迭代

准备：
        import 相关模块
        定义常量
        
        生成数据集

前向传播
        定义输入
        x =
        y_ =
        
        定义（网络层）参数：
        w1 =
        w2 =
        
        Define the network inference calculation process
        a =
        y =
        
反向传播
        定义损失函数
        loss =
        
        反向传播方法
        train_step =

生成会话,训练STEPS轮,迭代
        sess = tf.Session()
        init_op = tf.global_variables_initializer()
        sess.run(init_op)
       

        #训练模型
        STEPS = 5000   #训练5000轮
        for i in range(STEPS):
            start =
            end =
            sess.run(train_step,feed_dict)