Getting started with tensorflow

The numerical calculation is shown as a calculation diagram ,

TensorFlow The program trunk of is a diagram ,

Graph nodes are operations （ Code operation ）, There are multiple inputs and one output

The edges between nodes represent the tensor of flow between （tensors）, such as n Dimension group

 

Use the figure , Allow small and simple operations to build complex models , It will make gradient calculation simpler ----> Automatic derivation

 

h=ReLU(Wx+b)

Variables:W,b

Placeholders（ Place holder ）: The execution time will be accepted ,  x  namely input. Assign only one data type , A tensor of size

Mathematical operations： Mathematical operation node    Matrix multiplication 、 Wig 、 Activation functions, etc

 

 

import tensorflow as tf
import numpy as np

################################  Set up a map  ###################################
b = tf.Variable(tf.zeros((100,)))
W = tf.Variable(tf.random_uniform((784,100),-1,1))

x = tf.placeholder(tf.float32,(100,784))

h = tf.nn.relu(tf.matmul(x,W)+b)

#################################  Deploy the graph to the session  ###############################
#  Establish session object , Call parameters Fetches,Feeds
# Fetches:list of graph nodes which return the output of the nodes
# Feeds:dictionary mapping from graph nodes to actual values, Place holder 

#  Establish session object , Take default environment ,CPU, You can add device parameters .    The session deploys the graph to hardware ？
sess = tf.Session()
sess.run(tf.initialize_all_variables())
sess.run(h, {x:np.random.random(100,784)})

################################  Training models  ######################################

prediction = tf.nn.softmax()# The predicted value of the output 
label = tf.placeholder(tf.float32,[100,10])
cross_entropy = -tf.reduce_sum(label * tf.log(prediction),axis=1)

#  Create an optimizer for gradient descent , Learning rate setting 0.5, Minimize cross entropy cross_entropy
train_step = tf.train.GradientDescentOptimizer(0.5).minimize(cross_entropy)


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

for i in range(100):
    batch_x,batch_label = data.next_batch()
    sess.run(train_step,feed_dict={x: batch_x,label: batch_label})


with tf.variable_scope("foo"):
    v=tf.get_variable("v",shape=[1])