TensorFlow 一

栏目: 数据库 · 发布时间: 4年前

内容简介：最近事有点杂,并且还很多这次为数模 不用matlab 这次决定使用TensorFlow 额 用点新鲜东东 matlab 太老了 仅仅用作数模 ....数模需要使用 两个 一个是基础 一个是波士顿房价的预测 对于TensorFlow就学习这两个

前言

最近事有点杂,并且还很多

这次为数模 不用matlab 这次决定使用TensorFlow 额 用点新鲜东东 matlab 太老了 仅仅用作数模 ....

数模需要使用 两个 一个是基础 一个是波士顿房价的预测 对于TensorFlow就学习这两个

课程 https://mooc.study.163.com/course/2001396000#/info

配置Pycharm

过程过

(venv) ➜  untitled2

Hello world

import tensorflow as tf

# 创建一个常运算 作为节点加入默认计算图中
hello = tf.constant("hello world")

# 创建绘画
sess = tf.Session()

# 获取解雇
print(sess.run(hello))

一些概念

Tensor 张量 对应数据结构 多维数组

Flow 流 计算模型 多维数组 通过计算转化的模型 称为流

总结 tensorflow 通过计算图表达计算的编程系统 计算 计算图上的节点 节点之间描述了两个张量之间的关系

数据流图

计算图为有向图 由以下内容构成

一组节点 节点代表操作 为一种运算

一组有向边 边代表节点之间的关系 数据传递和控制依赖

有两种边

常规边 值传递

特殊边 节点控制

图

import tensorflow as tf

# 计算图
node1 = tf.constant(3.0, tf.float32, name="node1")
node2 = tf.constant(4.0, tf.float32, name="node2")
node3 = tf.add(node1, node2)

print(node3)

TensorFlow 一 

Tensor("Add:0", shape=(), dtype=float32)

输出张量结构

会话

import tensorflow as tf

# 计算图
node1 = tf.constant(3.0, tf.float32, name="node1")
node2 = tf.constant(4.0, tf.float32, name="node2")
node3 = tf.add(node1, node2)

print(node3)

# 建立对话
sess = tf.Session()

# 运行结果
print(sess.run(node1))
# 关闭session
sess.close()
/home/ming/PycharmProjects/untitled2/venv/bin/python /home/ming/PycharmProjects/untitled2/index2.py
2019-05-13 20:33:49.636022: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
Tensor("Add:0", shape=(), dtype=float32)
2019-05-13 20:33:49.640973: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 3493455000 Hz
2019-05-13 20:33:49.641242: I tensorflow/compiler/xla/service/service.cc:150] XLA service 0x1b09c80 executing computations on platform Host. Devices:
2019-05-13 20:33:49.641274: I tensorflow/compiler/xla/service/service.cc:158]   StreamExecutor device (0): <undefined>, <undefined>
3.0
/home/ming/PycharmProjects/untitled2/venv/bin/python /home/ming/PycharmProjects/untitled2/index2.py
Tensor("Add:0", shape=(), dtype=float32)
2019-05-13 20:34:29.977653: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
2019-05-13 20:34:29.983183: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 3493455000 Hz
2019-05-13 20:34:29.983552: I tensorflow/compiler/xla/service/service.cc:150] XLA service 0x1d8cc00 executing computations on platform Host. Devices:
2019-05-13 20:34:29.983602: I tensorflow/compiler/xla/service/service.cc:158]   StreamExecutor device (0): <undefined>, <undefined>
7.0

Process finished with exit code 0

张量

张量 多维数组

零阶张量 标量

一阶 向量 一维数组

n阶 n维数组

张量保存的计算过程

张量属性

Tensor("Add:0", shape=(), dtype=float32)

Add 节点名称 第几个节点输出

shape 形状 shape=() 表示标量

dtype 类型

张量形状

TensorFlow 一 

import tensorflow as tf

# 计算图
test1 = tf.constant([[[1,2,3], [2,3,4]],
                    [[23,34,5], [23,44,5]]]
                     ,name="test1")

print(test1)
Tensor("test1:0", shape=(2, 2, 3), dtype=int32)

外层两个 次之 2个 最里面3个

阶

维度

获取张量的元素

类比数组

import tensorflow as tf

# 计算图
test1 = tf.constant([[[1,2,3], [2,3,4]],
                    [[23,34,5], [23,44,5]]]
                     ,name="test1")

print(test1)

sess = tf.Session()
print(sess.run(test1)[1,1,0])
/home/ming/PycharmProjects/untitled2/venv/bin/python /home/ming/PycharmProjects/untitled2/index2.py
2019-05-13 20:47:22.263391: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
Tensor("test1:0", shape=(2, 2, 3), dtype=int32)
2019-05-13 20:47:22.268317: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 3493455000 Hz
2019-05-13 20:47:22.268869: I tensorflow/compiler/xla/service/service.cc:150] XLA service 0x14fdb30 executing computations on platform Host. Devices:
2019-05-13 20:47:22.268921: I tensorflow/compiler/xla/service/service.cc:158]   StreamExecutor device (0): <undefined>, <undefined>
23

Process finished with exit code 0

类型

自动判断

张量运算

矩阵运算

import tensorflow as tf

# 计算图
test1 = tf.constant([[[1,2,3], [2,3,4]],
                    [[23,34,5], [23,44,5]]]
                     ,name="test1")

test2 = tf.constant([[[1,2,3], [2,3,4]],
                    [[23,34,5], [23,44,5]]]
                     ,name="test2")

print(test1 + test2)

sess = tf.Session()
print(sess.run(test1 + test2)[1,1,0])

操作

计算图的节点为操作

加法操作

乘法操作

构造变量初始值为一个操作

运算操作有属性,构建图会确定下来

操作和运算设备做绑定

操作存在顺序 依赖边

计算图操作

TensorFlow 一 

import tensorflow as tf

# 清空图
tf.reset_default_graph()

# 定义变量 a
a = tf.Variable(1, name="a")
# 定义操作b = a + 1
b = tf.add(a, 1, name="b")
# 定义操作c 为 b * 4
c = tf.multiply(b, 4, name="c")
# 定义d  = c - b
d = tf.subtract(c, b, name="d")

生成图操作

import tensorflow as tf

# 清空图
tf.reset_default_graph()

# 定义变量 a
a = tf.Variable(1, name="a")
# 定义操作b = a + 1
b = tf.add(a, 1, name="b")
# 定义操作c 为 b * 4
c = tf.multiply(b, 4, name="c")
# 定义d  = c - b
d = tf.subtract(c, b, name="d")

logdir = "log1"

# 写入日子
writer = tf.summary.FileWriter(logdir, tf.get_default_graph())

writer.close();

写入当前空间下的log1文件

启动

(venv) ➜  untitled2 tensorboard --logdir ./log1/                                      
TensorBoard 1.13.1 at http://ming-pc:6006 (Press CTRL+C to quit)

TensorFlow 一

这样就可以看到生成图

TensorFlow 一

基本运算

会话

session拥有所要管理的资源

import tensorflow as tf

# 定义图
test1 = tf.constant([1 , 3,3])


# 创建会话
sess = tf.Session()

# 进行运算
print(sess.run(test1))

sess.close()
import tensorflow as tf

# 定义图
test1 = tf.constant([1 , 3,3])


# 创建会话
sess = tf.Session()

# 进行运算
try:
    print(sess.run(test1))
except:
    print("Exception")
finally:
    sess.close()

上下文管理会话

import tensorflow as tf

# 定义图
node1 = tf.constant(3.0, tf.float32, name="node1")
node2 = tf.constant(4.0, tf.float32, name="node2")
result = tf.add(node1, node2)

# 创建会话 上下文管理
with tf.Session() as sess:
    print(sess.run(result))

指定默认会话

import tensorflow as tf

# 定义图
node1 = tf.constant(3.0, tf.float32, name="node1")
node2 = tf.constant(4.0, tf.float32, name="node2")
result = tf.add(node1, node2)

sess = tf.Session()
with sess.as_default():
    print(result.eval())

常量 变量

创建常量

node2 = tf.constant(4.0, tf.float32, name="node2")

变量

import tensorflow as tf

node1 = tf.Variable(3.0, tf.float32, name="node1")
node2 = tf.Variable(4.0, tf.float32, name="node2")
result = tf.add(node1, node2, name="add")

sess = tf.Session()

# 变量初始化
init = tf.global_variables_initializer()
sess.run(init)

print(sess.run(result))

由于是静态图 所以需要对图进行赋初值

变量赋值

变量定义后无需要人工赋值,训练会自动调整变量对应的数值

import tensorflow as tf

value = tf.Variable(0, name="value")
one = tf.constant(1)
new_value = tf.add(value, one)
# new_value 赋值给value
update_value = tf.assign(value, new_value)

init = tf.global_variables_initializer()

with tf.Session() as sess:
    sess.run(init)
    for _ in range(10):
        sess.run(update_value)
        print(sess.run(value))
/home/ming/PycharmProjects/untitled2/venv/bin/python /home/ming/PycharmProjects/untitled2/index2.py
WARNING:tensorflow:From /home/ming/PycharmProjects/untitled2/venv/lib/python3.7/site-packages/tensorflow/python/framework/op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.
Instructions for updating:
Colocations handled automatically by placer.
2019-05-13 21:54:38.645047: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
2019-05-13 21:54:38.652219: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 3493455000 Hz
2019-05-13 21:54:38.652580: I tensorflow/compiler/xla/service/service.cc:150] XLA service 0x1f3ac80 executing computations on platform Host. Devices:
2019-05-13 21:54:38.652617: I tensorflow/compiler/xla/service/service.cc:158]   StreamExecutor device (0): <undefined>, <undefined>
1
2
3
4
5
6
7
8
9
10

Process finished with exit code 0

运行流程

import tensorflow as tf

# 清除计算图
tf.reset_default_graph()

value = tf.Variable(0, name="value")
one = tf.constant(1)
new_value = tf.add(value, one)
# new_value 赋值给value
update_value = tf.assign(value, new_value)

init = tf.global_variables_initializer()

with tf.Session() as sess:
    sess.run(init)
    for _ in range(10):
        sess.run(update_value)
        print(sess.run(value))

logdir = "log"

# 写日志
writer = tf.summary.FileWriter(logdir, tf.get_default_graph())
writer.close()

TensorFlow 一

变量依赖于控制init

占位符

运行的时候需要输入 %

x = tf.placeholder(tf.fload32, [2, 3], name="tx")

提交数据

import tensorflow as tf

a = tf.placeholder(tf.float32, name="a")
b = tf.placeholder(tf.float32, name="b")
c = tf.multiply(a, b, name="c")

init = tf.global_variables_initializer()

with tf.Session() as sess:
    sess.run(init)
    result = sess.run(c, feed_dict={a:10.0, b:2})
    print(result)
/home/ming/PycharmProjects/untitled2/venv/bin/python /home/ming/PycharmProjects/untitled2/index2.py
2019-05-13 22:16:29.402834: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
2019-05-13 22:16:29.407957: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 3493455000 Hz
2019-05-13 22:16:29.408692: I tensorflow/compiler/xla/service/service.cc:150] XLA service 0x26e9c80 executing computations on platform Host. Devices:
2019-05-13 22:16:29.408751: I tensorflow/compiler/xla/service/service.cc:158]   StreamExecutor device (0): <undefined>, <undefined>
20.0

Process finished with exit code 0

使用字典提交数据

import tensorflow as tf

a = tf.placeholder(tf.float32, name="a")
b = tf.placeholder(tf.float32, name="b")
c = tf.multiply(a, b, name="c")
d = tf.multiply(a, b, name="d")


init = tf.global_variables_initializer()

with tf.Session() as sess:
    sess.run(init)
    result = sess.run([c,d], feed_dict={a:[23.34,53.23], b:[23.45,243.5]})
    print(result)
    print(result[0])
/home/ming/PycharmProjects/untitled2/venv/bin/python /home/ming/PycharmProjects/untitled2/index2.py
2019-05-13 22:18:28.478916: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
2019-05-13 22:18:28.483828: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 3493455000 Hz
2019-05-13 22:18:28.484084: I tensorflow/compiler/xla/service/service.cc:150] XLA service 0x1359c80 executing computations on platform Host. Devices:
2019-05-13 22:18:28.484115: I tensorflow/compiler/xla/service/service.cc:158]   StreamExecutor device (0): <undefined>, <undefined>
[array([  547.323, 12961.505], dtype=float32), array([  547.323, 12961.505], dtype=float32)]
[  547.323 12961.505]

Process finished with exit code 0

可视化应用

使用的是TensorBoard

从日志中读取数据


以上就是本文的全部内容，希望对大家的学习有所帮助，也希望大家多多支持 码农网

查看所有标签

猜你喜欢:

本站部分资源来源于网络，本站转载出于传递更多信息之目的，版权归原作者或者来源机构所有，如转载稿涉及版权问题，请联系我们。

码农书籍
亮剑.NET

亮剑.NET

2009-3 / 55.00元

《亮剑.NET:SharePoint Server 2007开发实战》共分为8章，详细讲解了SharePoint上常见的开发任务，讲述了各种开发场景下需要了解的知识，并提供了丰富的实例。《亮剑.NET:SharePoint Server 2007开发实战》第1章为基础知识，讲述SharePoint的基本概念，基本的对象模型，代码编写注意事项，并讲解了一个集开发和部署打包为一体的项目结构的创建；第2......一起来看看 《亮剑.NET》 这本书的介绍吧!

码农工具
在线进制转换器
在线进制转换器

各进制数互转换器

HTML 编码/解码
HTML 编码/解码

HTML 编码/解码

Markdown 在线编辑器
Markdown 在线编辑器

Markdown 在线编辑器

  • New
  • 文章
  • 话题
  • 教程
    • 关注 码农网 公众号
    版权所有，保留一切权利！© 2018-2024 码农网 粤ICP备17054400号-3