<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[Sherry_up的博客]]></title><description><![CDATA[]]></description><link>https://blog.youkuaiyun.com/Sherry_up</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; Sherry_up]]></copyright><item><title><![CDATA[纠错： git push error “fatal: Unable to find remote helper for 'https'"]]></title><link>https://blog.youkuaiyun.com/Sherry_up/article/details/52463893</link><guid>https://blog.youkuaiyun.com/Sherry_up/article/details/52463893</guid><author>Sherry_up</author><pubDate>Wed, 07 Sep 2016 22:13:51 +0800</pubDate><description><![CDATA[git push error “fatal: Unable to find remote helper for ‘https’”]]></description><category></category></item><item><title><![CDATA[Average loss decreases with batches increasing]]></title><link>https://blog.youkuaiyun.com/Sherry_up/article/details/52278494</link><guid>https://blog.youkuaiyun.com/Sherry_up/article/details/52278494</guid><author>Sherry_up</author><pubDate>Mon, 22 Aug 2016 17:09:29 +0800</pubDate><description><![CDATA[用Tensorflow训练深度学习网络时，average loss for each batch会随着batch size 大小而成反比的变化，本来是该保持稳定的。经过一系列的检查，找到了原因：在使用Tensorflow API的时候，落了一个参数。 
错误的代码如下，tf.reduce_sum(var, index), index表示要减少的维度，如果为none的话，方法会把所有的元素相加。正确的]]></description><category></category></item><item><title><![CDATA[How to convert a list of tensors into a tensor]]></title><link>https://blog.youkuaiyun.com/Sherry_up/article/details/52169318</link><guid>https://blog.youkuaiyun.com/Sherry_up/article/details/52169318</guid><author>Sherry_up</author><pubDate>Wed, 10 Aug 2016 11:04:36 +0800</pubDate><description><![CDATA[Convert a list of tensors into a tensor using the method of tf.convert_to_tensor or tf.pack().]]></description><category></category></item><item><title><![CDATA[关于用深度学习做answer selection的论文]]></title><link>https://blog.youkuaiyun.com/Sherry_up/article/details/52103378</link><guid>https://blog.youkuaiyun.com/Sherry_up/article/details/52103378</guid><author>Sherry_up</author><pubDate>Wed, 03 Aug 2016 19:01:33 +0800</pubDate><description><![CDATA[分享了几篇关于用深度学习做answer selection的论文，并对论文进行总结。]]></description><category></category></item><item><title><![CDATA[纠错： 深度学习模型优化时快速收敛]]></title><link>https://blog.youkuaiyun.com/Sherry_up/article/details/51996571</link><guid>https://blog.youkuaiyun.com/Sherry_up/article/details/51996571</guid><author>Sherry_up</author><pubDate>Fri, 22 Jul 2016 16:58:38 +0800</pubDate><description><![CDATA[最近在做问答系统，用CNN深度学习模型分别对问题、正答案和负答案提取特征，得到各自的特征向量。正答案与问题之间的距离比负答案与问题之间的距离要近，距离用夹角的cos值来表示，目标函数是由此确定的。模型出自《Applying Deep Learning To Answer Selection: A Study And An Open Task》，如图1所示。]]></description><category></category></item><item><title><![CDATA[TensorFlow的安装]]></title><link>https://blog.youkuaiyun.com/Sherry_up/article/details/51598141</link><guid>https://blog.youkuaiyun.com/Sherry_up/article/details/51598141</guid><author>Sherry_up</author><pubDate>Mon, 06 Jun 2016 20:55:33 +0800</pubDate><description><![CDATA[TensorFlow的安装方式有多种，包括Anaconda安装，方法如下：
1. 安装Anaconda
2. 创建一个叫做tensorflow的conda环境


# Python 2.7
$ conda create -n tensorflow python=2.7

3. 激活conda环境，用pip在该环境里安装TensorFlow

$ source activate t]]></description><category></category></item><item><title><![CDATA[word2vec使用说明]]></title><link>https://blog.youkuaiyun.com/Sherry_up/article/details/51180757</link><guid>https://blog.youkuaiyun.com/Sherry_up/article/details/51180757</guid><author>Sherry_up</author><pubDate>Mon, 18 Apr 2016 16:43:42 +0800</pubDate><description><![CDATA[Google的word2vec官网：https://code.google.com/p/word2vec/下载下来的Demo源码文件共有如下几个： 
word2vec – Revision 41: /trunk 
… 
LICENSE //Apache LICENSE 
README.txt //工具使用说明 
compute-accuracy.c 
demo-analogy.sh  //]]></description><category></category></item><item><title><![CDATA[机器学习 笔记（三）	到底选择哪种机器学习算法来分类]]></title><link>https://blog.youkuaiyun.com/Sherry_up/article/details/51137275</link><guid>https://blog.youkuaiyun.com/Sherry_up/article/details/51137275</guid><author>Sherry_up</author><pubDate>Tue, 12 Apr 2016 21:16:06 +0800</pubDate><description><![CDATA[刚刚学了Andrew Ng在Coursera上的公开课，感觉在学习的过程中，有一些疑惑，尤其是各种算法之间的联系，以及在模型训练中遇到的普遍性的注意点，这里做一个回顾和总结。
到底选择哪种机器学习算法？]]></description><category></category></item><item><title><![CDATA[机器学习 笔记（二）	Multi-class classification]]></title><link>https://blog.youkuaiyun.com/Sherry_up/article/details/51137202</link><guid>https://blog.youkuaiyun.com/Sherry_up/article/details/51137202</guid><author>Sherry_up</author><pubDate>Tue, 12 Apr 2016 21:08:05 +0800</pubDate><description><![CDATA[刚刚学了Andrew Ng在Coursera上的公开课，感觉在学习的过程中，有一些疑惑，尤其是各种算法之间的联系，以及在模型训练中遇到的普遍性的注意点，这里做一个回顾和总结。
多类问题（Multi-class classification problem）]]></description><category></category></item><item><title><![CDATA[机器学习 笔记（一）	Non-linear classification]]></title><link>https://blog.youkuaiyun.com/Sherry_up/article/details/51136898</link><guid>https://blog.youkuaiyun.com/Sherry_up/article/details/51136898</guid><author>Sherry_up</author><pubDate>Tue, 12 Apr 2016 20:43:45 +0800</pubDate><description><![CDATA[刚刚学了Andrew Ng在Coursera上的公开课，感觉在学习的过程中，有一些疑惑，尤其是各种算法之间的联系，以及在模型训练中遇到的普遍性的注意点，这里做一个回顾和总结。
非线性分类问题（    Non-linear classification）可以用逻辑回归算法Logistic regression algorithm来解决这个问题。]]></description><category></category></item><item><title><![CDATA[machine learning：计算error时lambda取值]]></title><link>https://blog.youkuaiyun.com/Sherry_up/article/details/50848264</link><guid>https://blog.youkuaiyun.com/Sherry_up/article/details/50848264</guid><author>Sherry_up</author><pubDate>Thu, 10 Mar 2016 17:06:12 +0800</pubDate><description><![CDATA[machine learning week6_ex5：众所周知，在线性回归里，lamda是用于regularization的，防止over fitting的问题。]]></description><category></category></item><item><title><![CDATA[ML课后练习问题：fmincg: nonconformant arguments]]></title><link>https://blog.youkuaiyun.com/Sherry_up/article/details/50807440</link><guid>https://blog.youkuaiyun.com/Sherry_up/article/details/50807440</guid><author>Sherry_up</author><pubDate>Sat, 05 Mar 2016 10:12:32 +0800</pubDate><description><![CDATA[Andrew Ng的公开课练习ex3，遇到如下问题： 
error: fmincg: operator +: nonconformant arguments (op1 is 401*10, op2 is 4010*1)。 
 
fmincg进行“＋”操作运算时，参数的维度冲突。原因是：写lr CostFunction的时候出错，导致矩阵维度不匹配。 
在oneVsAll文件里面，针对每个分类器，依次]]></description><category></category></item><item><title><![CDATA[selector的使用]]></title><link>https://blog.youkuaiyun.com/Sherry_up/article/details/50630217</link><guid>https://blog.youkuaiyun.com/Sherry_up/article/details/50630217</guid><author>Sherry_up</author><pubDate>Wed, 03 Feb 2016 16:45:19 +0800</pubDate><description><![CDATA[用selector可以让控件根据状态的不同呈现不同的形态。例如按钮选中、不选中，显示不同的背景图片。如果有多个这样的selector，文件会比较多，这时候也可以用java代码实现，用**StateListDrawable**类。]]></description><category></category></item><item><title><![CDATA[android 控件布局]]></title><link>https://blog.youkuaiyun.com/Sherry_up/article/details/50619927</link><guid>https://blog.youkuaiyun.com/Sherry_up/article/details/50619927</guid><author>Sherry_up</author><pubDate>Mon, 01 Feb 2016 17:50:45 +0800</pubDate><description><![CDATA[http://blog.youkuaiyun.com/penglijiang/article/details/7775642




android 控件布局(顶部、底部......) 

android:layout_above 将该控件的底部至于给定ID的控件之上

android:layout_below 将该控件的顶部至于给定ID的控件之下

android:layout_toL]]></description><category></category></item><item><title><![CDATA[如何定义自己的控件]]></title><link>https://blog.youkuaiyun.com/Sherry_up/article/details/50607059</link><guid>https://blog.youkuaiyun.com/Sherry_up/article/details/50607059</guid><author>Sherry_up</author><pubDate>Fri, 29 Jan 2016 14:30:42 +0800</pubDate><description><![CDATA[以定义一个ImageView为例。1.定义一个layout.xml文件<?xml version="1.0" encoding="utf-8"?>
<LinearLayout xmlns:android="http://schemas.android.com/apk/res/android"
    android:orientation="vertical" android:layout_widt]]></description><category></category></item><item><title><![CDATA[在activity之间传递handler（三）]]></title><link>https://blog.youkuaiyun.com/Sherry_up/article/details/50504297</link><guid>https://blog.youkuaiyun.com/Sherry_up/article/details/50504297</guid><author>Sherry_up</author><pubDate>Tue, 12 Jan 2016 15:04:09 +0800</pubDate><description><![CDATA[用Broacast的方法在两个activity之间传递数据]]></description><category></category></item><item><title><![CDATA[在activity之间传递handler（二）]]></title><link>https://blog.youkuaiyun.com/Sherry_up/article/details/50503008</link><guid>https://blog.youkuaiyun.com/Sherry_up/article/details/50503008</guid><author>Sherry_up</author><pubDate>Tue, 12 Jan 2016 11:39:05 +0800</pubDate><description><![CDATA[用定义application的方式传递handler保持代码的独立性和隐密性]]></description><category></category></item><item><title><![CDATA[在activity之间传递handler（一）]]></title><link>https://blog.youkuaiyun.com/Sherry_up/article/details/50502855</link><guid>https://blog.youkuaiyun.com/Sherry_up/article/details/50502855</guid><author>Sherry_up</author><pubDate>Tue, 12 Jan 2016 11:22:34 +0800</pubDate><description><![CDATA[最近在研究基于安卓系统的语音识别和蓝牙通讯，过程中涉及到多个activity的交互。activity之间的交互数据可以由Intent完成，也可以借用Handler。如果用intent的话，一定要finish当前的activity UI，另一个activity响应，数据才得以传递，此时UI势必改变。而用handler是在一个activity中控制另一个activity活动的好方法。在activity之]]></description><category></category></item><item><title><![CDATA[Android studio JNI（NDK） 开发]]></title><link>https://blog.youkuaiyun.com/Sherry_up/article/details/50501697</link><guid>https://blog.youkuaiyun.com/Sherry_up/article/details/50501697</guid><author>Sherry_up</author><pubDate>Tue, 12 Jan 2016 09:21:56 +0800</pubDate><description><![CDATA[JNI是JAVA NATIVE INTERFACE的缩小，它允许Java代码与用其它语言写的代码之间进行交互，例如典型的C或C++语言。Java是上层语言，而C或C++是底层语言（用来实现本地方法），所以JNI是起中间件的作用，将上层（交互层）和下层（驱动层）联系起来。步骤在上层用JAVA申明本地函数，在底层用C或者C++实现本地函数。]]></description><category></category></item><item><title><![CDATA[在Ubuntu上安装JDK和Eclipse]]></title><link>https://blog.youkuaiyun.com/Sherry_up/article/details/50501609</link><guid>https://blog.youkuaiyun.com/Sherry_up/article/details/50501609</guid><author>Sherry_up</author><pubDate>Tue, 12 Jan 2016 09:07:36 +0800</pubDate><description><![CDATA[在虚拟机VM上的Ubuntu里安装JDK和Eclipse，本文包含安装步骤和总结。]]></description><category></category></item></channel></rss>