<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[逍遥游]]></title><description><![CDATA[将思绪付诸笔端]]></description><link>https://blog.youkuaiyun.com/linuxwindowsios</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; linuxwindowsios]]></copyright><item><title><![CDATA[TensorFlow batch_dot函数理解]]></title><link>https://blog.youkuaiyun.com/linuxwindowsios/article/details/84189907</link><guid>https://blog.youkuaiyun.com/linuxwindowsios/article/details/84189907</guid><author>linuxwindowsios</author><pubDate>Sat, 17 Nov 2018 16:14:55 +0800</pubDate><description><![CDATA[batch_dot实现
载入库
import tensorflow as tf
tf.enable_eager_execution()
import keras.backend as K
import numpy as np

生成输入数据
x1 = tf.convert_to_tensor([[1,2,3],[4,5,6]])
x2 = tf.convert_to_tensor([[1,2,3]...]]></description><category></category></item><item><title><![CDATA[mac使用jupyter notebook web开发pyspark]]></title><link>https://blog.youkuaiyun.com/linuxwindowsios/article/details/80875906</link><guid>https://blog.youkuaiyun.com/linuxwindowsios/article/details/80875906</guid><author>linuxwindowsios</author><pubDate>Sun, 01 Jul 2018 17:42:25 +0800</pubDate><description><![CDATA[


下载安装spark
下载安装anaconda
路径配置
配置jupyter notebook






下载安装spark

下载地址http://spark.apache.org/downloads.html



下载安装anaconda

下载地址https://www.anaconda.com/download/#macos



路径配置



cd 
open .bash_pr...]]></description><category></category></item><item><title><![CDATA[softmax函数实现时的技巧]]></title><link>https://blog.youkuaiyun.com/linuxwindowsios/article/details/78003312</link><guid>https://blog.youkuaiyun.com/linuxwindowsios/article/details/78003312</guid><author>linuxwindowsios</author><pubDate>Sat, 16 Sep 2017 14:22:56 +0800</pubDate><description><![CDATA[在二分类任务中应用最广泛的激活函数是sigmoid，而在多分类任务中便是softmax函数，用于将实值隐射到(0-1)的区间内，通过交叉熵或者说是负对数极大似然估计作为损失函数，通过随机梯度下降、牛顿法、adagrad、adadelta等优化函数进行迭代优化。sigmoid函数公式如下：sigmoid(x)=11+exp(−x)sigmoid(x) = \frac{1}{1+exp(-x)}soft]]></description><category></category></item><item><title><![CDATA[python2.x代码版本迁移至Python3.x]]></title><link>https://blog.youkuaiyun.com/linuxwindowsios/article/details/78002697</link><guid>https://blog.youkuaiyun.com/linuxwindowsios/article/details/78002697</guid><author>linuxwindowsios</author><pubDate>Sat, 16 Sep 2017 12:34:13 +0800</pubDate><description><![CDATA[列示需要修改的内容2to3 pythonfile.py直接修改2to3 -w pythonfile.py在目录下，会出现pythonfile.py.bak的文件，是原先代码的备份文件。]]></description><category></category></item><item><title><![CDATA[Jupyter Notebook导出成pdf]]></title><link>https://blog.youkuaiyun.com/linuxwindowsios/article/details/74297891</link><guid>https://blog.youkuaiyun.com/linuxwindowsios/article/details/74297891</guid><author>linuxwindowsios</author><pubDate>Tue, 04 Jul 2017 11:17:54 +0800</pubDate><description><![CDATA[步骤1-修改latex模板文件进入 \Anaconda3\Lib\site-packages\nbconvert\templates\latex目录，找到article.tplx文件，将其替换为\documentclass{ctexart}
\setmainfont{Microsoft YaHei}步骤2-生成tex文件进入要转换的.ipynb文件所在的目录下，使用以下命令将.ipynb文件转换为.]]></description><category></category></item><item><title><![CDATA[keras修改backend的方法]]></title><link>https://blog.youkuaiyun.com/linuxwindowsios/article/details/68483618</link><guid>https://blog.youkuaiyun.com/linuxwindowsios/article/details/68483618</guid><author>linuxwindowsios</author><pubDate>Thu, 30 Mar 2017 10:07:00 +0800</pubDate><description><![CDATA[方法1在user文件夹下新建.keras文件夹，在文件夹下新建keras.json文件，json内容为：{
    "image_dim_ordering": "tf", 
    "epsilon": 1e-07, 
    "floatx": "float32", 
    "backend": "tensorflow"
}这其中的backend可以是tensorflow或者theano,当然前]]></description><category></category></item><item><title><![CDATA[tensorflow squeeze函数理解]]></title><link>https://blog.youkuaiyun.com/linuxwindowsios/article/details/68099955</link><guid>https://blog.youkuaiyun.com/linuxwindowsios/article/details/68099955</guid><author>linuxwindowsios</author><pubDate>Wed, 29 Mar 2017 20:56:16 +0800</pubDate><description><![CDATA[squeezedef squeeze(input, axis=None, name=None, squeeze_dims=None):
  # pylint: disable=redefined-builtin
  """Removes dimensions of size 1 from the shape of a tensor.  Given a tensor `input`, this ope]]></description><category></category></item><item><title><![CDATA[tensorflow clip_by_norm函数理解]]></title><link>https://blog.youkuaiyun.com/linuxwindowsios/article/details/67635867</link><guid>https://blog.youkuaiyun.com/linuxwindowsios/article/details/67635867</guid><author>linuxwindowsios</author><pubDate>Tue, 28 Mar 2017 14:07:38 +0800</pubDate><description><![CDATA[clip_by_norm这里的clip_by_norm是指对梯度进行裁剪，通过控制梯度的最大范式，防止梯度爆炸的问题，是一种比较常用的梯度规约的方式。tensorflow中的clip_by_norm示例optimizer = tf.train.AdamOptimizer(learning_rate, beta1=0.5)
grads = optimizer.compute_gradients(cos]]></description><category></category></item><item><title><![CDATA[通过图像对核函数进行直观理解]]></title><link>https://blog.youkuaiyun.com/linuxwindowsios/article/details/62422613</link><guid>https://blog.youkuaiyun.com/linuxwindowsios/article/details/62422613</guid><author>linuxwindowsios</author><pubDate>Thu, 16 Mar 2017 14:25:55 +0800</pubDate><description><![CDATA[载入库%matplotlib inline
import matplotlib.pyplot as plt
from mpl_toolkits.mplot3d import Axes3D
import numpy as np
import numpy.linalg as la几个比较著名的核函数内积def linear():
    return lambda x, y: np.inner(x, y]]></description><category></category></item><item><title><![CDATA[win10下tensorflow库gpu运行环境配置实操]]></title><link>https://blog.youkuaiyun.com/linuxwindowsios/article/details/62108276</link><guid>https://blog.youkuaiyun.com/linuxwindowsios/article/details/62108276</guid><author>linuxwindowsios</author><pubDate>Wed, 15 Mar 2017 00:10:30 +0800</pubDate><description><![CDATA[TensorFlow
NVIDIA显卡驱动
CUDA套件
cuDNN驱动
安装
测试近日，在github上闲逛时发现了一个非常有趣的一个实验，地址如下DeepLearningFlappyBird，见文知意，使用强化学习来教会鸟儿如何通关尽可能多的关卡。当然，既然是游戏那还是跑起来看一下直观。所以，把代码clone下来后便开启了如下的摸索之旅。TensorFlow对于window系统，TensorFl]]></description><category></category></item><item><title><![CDATA[Python threading模块初探]]></title><link>https://blog.youkuaiyun.com/linuxwindowsios/article/details/61656180</link><guid>https://blog.youkuaiyun.com/linuxwindowsios/article/details/61656180</guid><author>linuxwindowsios</author><pubDate>Sun, 12 Mar 2017 22:06:14 +0800</pubDate><description><![CDATA[join方法
载入库
用于打印展示的函数
不使用join方法
对线程实例列表中的最后一个线程执行join方法
对全部子线程执行join方法
总结join方法通过一段代码来直观地展示对全部子线程和子线程列表中最后一个线程调用join方法的差别。载入库import threading
import time用于打印展示的函数def myprint(value):
    for v in value:]]></description><category></category></item><item><title><![CDATA[TensorFlow学习笔记]]></title><link>https://blog.youkuaiyun.com/linuxwindowsios/article/details/60571577</link><guid>https://blog.youkuaiyun.com/linuxwindowsios/article/details/60571577</guid><author>linuxwindowsios</author><pubDate>Mon, 06 Mar 2017 10:29:56 +0800</pubDate><description><![CDATA[概念
TensorFlow计算方式
TensorFlow内建运算操作
variable标量的特殊性
参考内容概念TensorFlow计算方式TensorFlow中的计算可以被表示为一个有向图，或者叫计算图，其中每一个运算操作被表示为一个节点，连接节点的是边。每一个节点可以有任意多个输入与输出，在计算图中流动(flow)的数据是张量(tensor)，因此称为TensorFlow。
TensorFlow]]></description><category></category></item><item><title><![CDATA[logistic函数]]></title><link>https://blog.youkuaiyun.com/linuxwindowsios/article/details/60149326</link><guid>https://blog.youkuaiyun.com/linuxwindowsios/article/details/60149326</guid><author>linuxwindowsios</author><pubDate>Fri, 03 Mar 2017 21:05:52 +0800</pubDate><description><![CDATA[logistic函数logistic函数
导入包
scipy中的函数
自己定义的函数
测试
导入包>import scipy as sp
>import numpy as npscipy中的函数
>sp.special.expit(x)自己定义的函数def logistic(x):
    return 1./(1+np.exp(-x))测试
input
>sp.special.expit(1)
>]]></description><category></category></item></channel></rss>