本篇blog主要以code+markdown的形式介紹tf這本實戰書。（建議使用jupyter來學習）

第三章 TF入門學習

• 3.1 TF計算模型——計算圖

• 3.2 TF資料模型——張量

• 3.3 TF執行模型——會話

• 3.4 Google遊樂場——TF神經網路實現

3.1 TF計算模型——計算圖

1.一般用tf簡單替代TensorFlow模組名稱，簡單加法運算的計算圖。（tf中常量需用用constant進行定義）

import tensorflow as tf
a = tf.constant([1.0, 2.0], name = "a") 
b = tf.constant([2.0, 3.0], name = "b")

result = a+ b
 

2.定義兩個不同的圖

import tensorflow as tf

g1 = tf.Graph()
with g1.as_default():
    v = tf.get_variable("v", [1], initializer = tf.zeros_initializer()) # 設定初始值為0

g2 = tf.Graph()
with g2.as_default():
    v = tf.get_variable("v", [1], initializer = tf.ones_initializer())  # 設定初始值為1
    
with tf.Session(graph = g1) as sess:
    tf.global_variables_initializer().run()
    with tf.variable_scope("", reuse=True):
        print(sess.run(tf.get_variable("v")))

with tf.Session(graph = g2) as sess:
    tf.global_variables_initializer().run()
    with tf.variable_scope("", reuse=True):
        print(sess.run(tf.get_variable("v")))
 

3.2 TF資料模型——張量

import tensorflow as tf
a = tf.constant([1.0, 2.0], name="a")
b = tf.constant([2.0, 3.0], name="b")
result = a + b
print(result)

sess = tf.InteractiveSession ()
print(result.eval())
sess.close()

3.3 TF執行模型——會話

# 建立一個會話。
sess = tf.Session()

# 使用會話得到之前計算的結果。
print(sess.run(result))

# 關閉會話使得本次執行中使用到的資源可以被釋放。
sess.close()

# 使用with statement 來建立會話
with tf.Session() as sess:
    print(sess.run(result))

# 指定預設會話
sess = tf.Session()
with sess.as_default():
     print(result.eval())

# 下面的兩個命令有相同的功能。
print(sess.run(result))
print(result.eval(session=sess))

# 使用tf.InteractiveSession構建會話
sess = tf.InteractiveSession ()
print(result.eval())
sess.close()

# 通過ConfigProto配置會話

config=tf.ConfigProto(allow_soft_placement=True, log_device_placement=True)
sess1 = tf.InteractiveSession(config=config)
sess2 = tf.Session(config=config)
 

3.4 Google遊樂場——TF神經網路實現

# 3.4 神經網路
#  三層簡單神經網路
# 定義變數
w1 = tf.Variable(tf.random_normal(shape = [2,3], stddev = 1, seed = 1))
w2 = tf.Variable(tf.random_normal(shape = [3,1], stddev = 1, seed = 1))
x = tf.constant([[0.7,0.9]])

# 前向傳播
a = tf.matmul(x , w1)
y = tf.matmul(a, w2)

# 呼叫會話執行
sess = tf.Session()
sess.run(w1.initializer)
sess.run(w2.initializer)  
print(sess.run(y))  
sess.close()

• 使用tf.placeholder()定義x, 並使用tf.global_variables_initializer()來初始化所有的變數

# 呼叫placeholder
x = tf.placeholder(tf.float32, shape = (1,2),  name = "input")
a = tf.matmul(x , w1)
y = tf.matmul(a, w2)

# 使用tf.global_variables_initializer()來初始化所有的變數
sess = tf.Session()
init_op = tf.global_variables_initializer()  
sess.run(init_op)
sess.run(y, feed_dict = {x:[[0.7,0.9]]})

• 增加多個輸入

# 增加多個輸入
x = tf.placeholder(tf.float32, shape = (3,2),  name = "input")
a = tf.matmul(x , w1)
y = tf.matmul(a, w2)

#使用tf.global_variables_initializer()來初始化所有的變數
with tf.Session() as sess:
    init_op = tf.global_variables_initializer()
    sess.run(init_op)
    
    print(sess.run(y, feed_dict={x: [[0.7,0.9],[0.1,0.4],[0.5,0.8]]}))

3.5 完整神經網路demo

import tensorflow as tf
from numpy.random import RandomState

# 定義神經網路的引數，輸入和輸出節點
batch_size = 8
w1 = tf.Variable(tf.random_normal(shape = [2,3], seed = 1, stddev = 1))
w2 = tf.Variable(tf.random_normal(shape = [3,1], seed = 1, stddev = 1))
x = tf.placeholder(tf.float32, shape = (None, 2), name = "x-input")
y_ = tf.placeholder(tf.float32, shape = (None, 1), name = "y-input") # 真實值

# 定義前向傳播過程，損失函式及反向傳播演算法
a = tf.matmul(x, w1)
y = tf.matmul(a, w2) # 預測值

cross_entropy = -tf.reduce_mean(y_ * tf.log(tf.clip_by_value(y, 1e-10, 1.0))
                               + (1 - y_) * tf.log(tf.clip_by_value(1-y, 1e-10, 1.0)))
# 用adam進行學習
train_step = tf.train.AdamOptimizer(0.001).minimize(cross_entropy)

# 生成模擬資料集
rdm = RandomState(1)
X = rdm.rand(128,2)
Y = [[int(x1+x2 < 1)] for (x1, x2) in X]

# 建立一個會話來執行TensorFlow程式
with tf.Session() as sess:
    init_op = tf.global_variables_initializer()
    sess.run(init_op)
    
    # 輸出目前（未經訓練）的引數取值。
    print("w1:", sess.run(w1))
    print("w2:", sess.run(w2))
    print("\n")
    
    # 訓練模型。
    STEPS = 5000
    for i in range(STEPS):
        start = (i*batch_size) % 128
        end = (i*batch_size) % 128 + batch_size
        sess.run(train_step, feed_dict={x: X[start:end], y_: Y[start:end]})
        if i % 1000 == 0:
            total_cross_entropy = sess.run(cross_entropy, feed_dict={x: X, y_: Y})
            print("After %d training step(s), cross entropy on all data is %g" % (i, total_cross_entropy))
    
    # 輸出訓練後的引數取值。
    print("\n")
    print("w1:", sess.run(w1))
    print("w2:", sess.run(w2))