TensorFlow学习心得

自定义

#定量
y_hat = tf.constant(36,name="y_hat")
#变量
loss = tf.Variable((y-y_hat)**2,name="loss" )  
#占位符
x=tf.placeholder(tf.int64,name="x")
tf.Session().run(x*2,feed_dict={x:3})
#转码
tf.cast(变量,dtype=tf.float32)
#矩阵惩罚
tf.matmul(矩阵A，矩阵B)

基本操作

tf.transpose：转置

input_data = tf.constant([[1, 2, 3], [4, 5, 6]])
print(sess.run(tf.transpose(input_data)))
# [[1 4]
#  [2 5]
#  [3 6]]

tf.reduce_mean：求平均值（降维）

x = [[1,2,3],
      [1,2,3]]
 
xx = tf.cast(x,tf.float32)
 
mean_all = tf.reduce_mean(xx, keep_dims=False)
mean_0 = tf.reduce_mean(xx, axis=0, keep_dims=False)
mean_1 = tf.reduce_mean(xx, axis=1, keep_dims=False)
 
 
with tf.Session() as sess:
    m_a,m_0,m_1 = sess.run([mean_all, mean_0, mean_1])
 
print m_a    # output: 2.0
print m_0    # output: [ 1.  2.  3.]
print m_1    #output:  [ 2.  2.]

tf.argmax:取最大值的index

tf.argmax(input,axis)
test = np.array([[1, 2, 3], [2, 3, 4], [5, 4, 3], [8, 7, 2]])
#axis=0时比较每一列的元素
test[0] = array([1, 2, 3])
test[1] = array([2, 3, 4])
test[2] = array([5, 4, 3])
test[3] = array([8, 7, 2])
# output   :    [3, 3, 1]     

#axis=1的时候，将每一行最大元素所在的索引记录下来，最后返回每一行最大元素所在的索引数组。
test[0] = array([1, 2, 3])  #2
test[1] = array([2, 3, 4])  #2
test[2] = array([5, 4, 3])  #0
test[3] = array([8, 7, 2])  #0

tf.placeholder：占位符

#初始化
X = tf.placeholder(tf.float32, [n_x, None], name="X")
#填入
tf.Session().run(X,feed_dict={X:X_train})

tensorboard:可视化图

#初始化Tensorboard
summary_writer=tf.summary.FileWriter("summary")
summary_writer.add_graph(sess.graph)
merge=tf.summary.merge_all()

#过程
_,epoch_cost,summary=sess.run([optmizer,cost,merge],feed_dict={X:X_train,Y:Y_train})
summary_writer.add_summary(summary, epoch)


#保存TensorBoard
summary_writer.close()

tf.variable_scope:变量域

with tf.variable_scope(""):

tf.train.Saver:管理模型的保存与加载

saver = tf.train.Saver()
#保存
saver=tf.train.Saver(sess,"model/")
# 恢复模型参数
saver.restore(sess, 'model/')