Tensorflow中梯度下降法更新引數值
阿新 • • 發佈:2019-01-10
tf.train.GradientDescentOptimizer(0.01).minimize(cross_entropy)
TensorFlow經過使用梯度下降法對損失函式中的變數進行修改值,預設修改tf.Variable(tf.zeros([784,10]))
為Variable的引數。
train_step = tf.train.GradientDescentOptimizer(0.01).minimize(cross_entropy,var_list=[w,b])
也可以使用var_list引數來定義更新那些引數的值
#匯入Minst資料集 import input_data mnist = input_data.read_data_sets("data",one_hot=True) #匯入tensorflow庫 import tensorflow as tf #輸入變數,把28*28的圖片變成一維陣列(丟失結構資訊) x = tf.placeholder("float",[None,784]) #權重矩陣,把28*28=784的一維輸入,變成0-9這10個數字的輸出 w = tf.Variable(tf.zeros([784,10])) #偏置 b = tf.Variable(tf.zeros([10])) #核心運算,其實就是softmax(x*w+b) y = tf.nn.softmax(tf.matmul(x,w) + b) #這個是訓練集的正確結果 y_ = tf.placeholder("float",[None,10]) #交叉熵,作為損失函式 cross_entropy = -tf.reduce_sum(y_ * tf.log(y)) #梯度下降演算法,最小化交叉熵 train_step = tf.train.GradientDescentOptimizer(0.01).minimize(cross_entropy) #初始化,在run之前必須進行的 init = tf.initialize_all_variables() #建立session以便運算 sess = tf.Session() sess.run(init) #迭代1000次 for i in range(1000): #獲取訓練資料集的圖片輸入和正確表示數字 batch_xs, batch_ys = mnist.train.next_batch(100) #執行剛才建立的梯度下降演算法,x賦值為圖片輸入,y_賦值為正確的表示數字 sess.run(train_step,feed_dict = {x:batch_xs, y_: batch_ys}) #tf.argmax獲取最大值的索引。比較運算後的結果和本身結果是否相同。 #這步的結果應該是[1,1,1,1,1,1,1,1,0,1...........1,1,0,1]這種形式。 #1代表正確,0代表錯誤 correct_prediction = tf.equal(tf.argmax(y,1), tf.argmax(y_,1)) #tf.cast先將資料轉換成float,防止求平均不準確。 #tf.reduce_mean由於只有一個引數,就是上面那個陣列的平均值。 accuracy = tf.reduce_mean(tf.cast(correct_prediction,"float")) #輸出 print(sess.run(accuracy,feed_dict={x:mnist.test.images,y_: mnist.test.labels}))
計算結果如下
"C:\Program Files\Anaconda3\python.exe" D:/pycharmprogram/tensorflow_learn/softmax_learn/softmax_learn.py Extracting data\train-images-idx3-ubyte.gz Extracting data\train-labels-idx1-ubyte.gz Extracting data\t10k-images-idx3-ubyte.gz Extracting data\t10k-labels-idx1-ubyte.gz WARNING:tensorflow:From C:\Program Files\Anaconda3\lib\site-packages\tensorflow\python\util\tf_should_use.py:175: 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. 2018-05-14 15:49:45.866600: W C:\tf_jenkins\home\workspace\rel-win\M\windows\PY\35\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. 2018-05-14 15:49:45.866600: W C:\tf_jenkins\home\workspace\rel-win\M\windows\PY\35\tensorflow\core\platform\cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX2 instructions, but these are available on your machine and could speed up CPU computations. 0.9163 Process finished with exit code 0
如果限制,只更新引數W檢視效果
"C:\Program Files\Anaconda3\python.exe" D:/pycharmprogram/tensorflow_learn/softmax_learn/softmax_learn.py Extracting data\train-images-idx3-ubyte.gz Extracting data\train-labels-idx1-ubyte.gz Extracting data\t10k-images-idx3-ubyte.gz Extracting data\t10k-labels-idx1-ubyte.gz WARNING:tensorflow:From C:\Program Files\Anaconda3\lib\site-packages\tensorflow\python\util\tf_should_use.py:175: 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. 2018-05-14 15:51:08.543600: W C:\tf_jenkins\home\workspace\rel-win\M\windows\PY\35\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. 2018-05-14 15:51:08.544600: W C:\tf_jenkins\home\workspace\rel-win\M\windows\PY\35\tensorflow\core\platform\cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX2 instructions, but these are available on your machine and could speed up CPU computations. 0.9187 Process finished with exit code 0
可以看出只修改W對結果影響不大,如果設定只修改b
#匯入Minst資料集
import input_data
mnist = input_data.read_data_sets("data",one_hot=True)
#匯入tensorflow庫
import tensorflow as tf
#輸入變數,把28*28的圖片變成一維陣列(丟失結構資訊)
x = tf.placeholder("float",[None,784])
#權重矩陣,把28*28=784的一維輸入,變成0-9這10個數字的輸出
w = tf.Variable(tf.zeros([784,10]))
#偏置
b = tf.Variable(tf.zeros([10]))
#核心運算,其實就是softmax(x*w+b)
y = tf.nn.softmax(tf.matmul(x,w) + b)
#這個是訓練集的正確結果
y_ = tf.placeholder("float",[None,10])
#交叉熵,作為損失函式
cross_entropy = -tf.reduce_sum(y_ * tf.log(y))
#梯度下降演算法,最小化交叉熵
train_step = tf.train.GradientDescentOptimizer(0.01).minimize(cross_entropy,var_list=[b])
#初始化,在run之前必須進行的
init = tf.initialize_all_variables()
#建立session以便運算
sess = tf.Session()
sess.run(init)
#迭代1000次
for i in range(1000):
#獲取訓練資料集的圖片輸入和正確表示數字
batch_xs, batch_ys = mnist.train.next_batch(100)
#執行剛才建立的梯度下降演算法,x賦值為圖片輸入,y_賦值為正確的表示數字
sess.run(train_step,feed_dict = {x:batch_xs, y_: batch_ys})
#tf.argmax獲取最大值的索引。比較運算後的結果和本身結果是否相同。
#這步的結果應該是[1,1,1,1,1,1,1,1,0,1...........1,1,0,1]這種形式。
#1代表正確,0代表錯誤
correct_prediction = tf.equal(tf.argmax(y,1), tf.argmax(y_,1))
#tf.cast先將資料轉換成float,防止求平均不準確。
#tf.reduce_mean由於只有一個引數,就是上面那個陣列的平均值。
accuracy = tf.reduce_mean(tf.cast(correct_prediction,"float"))
#輸出
print(sess.run(accuracy,feed_dict={x:mnist.test.images,y_: mnist.test.labels}))
計算結果:
"C:\Program Files\Anaconda3\python.exe" D:/pycharmprogram/tensorflow_learn/softmax_learn/softmax_learn.py
Extracting data\train-images-idx3-ubyte.gz
Extracting data\train-labels-idx1-ubyte.gz
Extracting data\t10k-images-idx3-ubyte.gz
Extracting data\t10k-labels-idx1-ubyte.gz
WARNING:tensorflow:From C:\Program Files\Anaconda3\lib\site-packages\tensorflow\python\util\tf_should_use.py:175: 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.
2018-05-14 15:52:04.483600: W C:\tf_jenkins\home\workspace\rel-win\M\windows\PY\35\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.
2018-05-14 15:52:04.483600: W C:\tf_jenkins\home\workspace\rel-win\M\windows\PY\35\tensorflow\core\platform\cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX2 instructions, but these are available on your machine and could speed up CPU computations.
0.1135
Process finished with exit code 0
如果只更新b那麼對效果影響很大。