<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[xpersistence的专栏]]></title><description><![CDATA[]]></description><link>https://blog.youkuaiyun.com/xpersistence</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; xpersistence]]></copyright><item><title><![CDATA[Reading Note : Gibbs Sampling for the Uninitiated]]></title><link>https://blog.youkuaiyun.com/xpersistence/article/details/8869770</link><guid>https://blog.youkuaiyun.com/xpersistence/article/details/8869770</guid><author>xpersistence</author><pubDate>Tue, 30 Apr 2013 15:44:53 +0800</pubDate><description><![CDATA[这里为了增加访问量，具体请看原文：http://www.xperseverance.net/blogs/2013/03/1682/






这几天较空总算把《Gibbs Sampling for the Uninitiated》看明白了点，看完这个对其他模型的 Gibbs Sampling 感觉应该是不会有大问题了，之后可以再仔细看看《Parameter estimation]]></description><category></category></item><item><title><![CDATA[Reading Note : Parameter estimation for text analysis]]></title><link>https://blog.youkuaiyun.com/xpersistence/article/details/8869765</link><guid>https://blog.youkuaiyun.com/xpersistence/article/details/8869765</guid><author>xpersistence</author><pubDate>Tue, 30 Apr 2013 15:42:52 +0800</pubDate><description><![CDATA[这里为了增加访问量，请看原文http://www.xperseverance.net/blogs/2013/03/1744/

伟大的Parameter estimation for text analysis！当把这篇看的差不多的时候，也就到了LDA基础知识终结的时刻了，意味着LDA基础模型的基本了解完成了。所以对该模型的学习告一段落，下一阶段就是了解LDA无穷无尽的变种，不过那些不是很有用]]></description><category></category></item><item><title><![CDATA[The Dirichlet Distribution 狄利克雷分布 (PRML 2.2.1)]]></title><link>https://blog.youkuaiyun.com/xpersistence/article/details/7400299</link><guid>https://blog.youkuaiyun.com/xpersistence/article/details/7400299</guid><author>xpersistence</author><pubDate>Tue, 27 Mar 2012 20:10:38 +0800</pubDate><description><![CDATA[http://www.xperseverance.net/blogs/2012/03/510/


        Dirichlet分布可以看做是分布之上的分布。如何理解这句话，我们可以先举个例子：假设我们有一个骰子，其有六面，分别为{1,2,3,4,5,6}。现在我们做了10000次投掷的实验，得到的实验结果是六面分别出现了{2000,2000,2000,2000,1000,1000}次]]></description><category></category></item><item><title><![CDATA[当前学习内容]]></title><link>https://blog.youkuaiyun.com/xpersistence/article/details/7333268</link><guid>https://blog.youkuaiyun.com/xpersistence/article/details/7333268</guid><author>xpersistence</author><pubDate>Thu, 08 Mar 2012 16:08:25 +0800</pubDate><description><![CDATA[1. LDA
    1.1 LSI
                1.1.1 EM 算法 2012@3@7  (PRML中的EM算法以混合高斯分布为例子)
                        1.1.1.1 » 混合高斯分布  2012@3@7
  (需要了解高斯分布，转至2.1.1)
2. PR&ML与统计学习
        2.1 高斯分布]]></description><category></category></item><item><title><![CDATA[PRML Chapter 9. Mixture Models and EM]]></title><link>https://blog.youkuaiyun.com/xpersistence/article/details/7329576</link><guid>https://blog.youkuaiyun.com/xpersistence/article/details/7329576</guid><author>xpersistence</author><pubDate>Wed, 07 Mar 2012 16:48:25 +0800</pubDate><description><![CDATA[http://www.xperseverance.net/blogs/2012/03/14/

今天从网上搜的 EM 算法的原始论文叫做 Maximum likelihood from incomplete data via the EM algorithm，下次仔细学习的时候可以看看，最近比较功利，就不弄得那么明白了。（2012@3@21）
9.1 K-means Clustering]]></description><category></category></item><item><title><![CDATA[网络课程专辑]]></title><link>https://blog.youkuaiyun.com/xpersistence/article/details/7327699</link><guid>https://blog.youkuaiyun.com/xpersistence/article/details/7327699</guid><author>xpersistence</author><pubDate>Wed, 07 Mar 2012 10:13:52 +0800</pubDate><description><![CDATA[http://its.buffalo.edu/services/capture/links/spr12/CSE674.html]]></description><category></category></item><item><title><![CDATA[LDA论文导读]]></title><link>https://blog.youkuaiyun.com/xpersistence/article/details/7327529</link><guid>https://blog.youkuaiyun.com/xpersistence/article/details/7327529</guid><author>xpersistence</author><pubDate>Wed, 07 Mar 2012 09:46:40 +0800</pubDate><description><![CDATA[这个人的水平还真不是一般的强，佩服啊
转载自： http://bbs.byr.cn/#!article/PR_AI/2530?p=1


原文的主要内容 
  
     有两种方法设计分类器： 
     1. discriminative model，就是由样本直接设计判别函数，例如SVM； 
     2. generative model，就是先从样本恢复概率模型——例如我]]></description><category></category></item><item><title><![CDATA[【JMLR'03】Latent Dirichlet Allocation (LDA) - David M.Blei]]></title><link>https://blog.youkuaiyun.com/xpersistence/article/details/7326697</link><guid>https://blog.youkuaiyun.com/xpersistence/article/details/7326697</guid><author>xpersistence</author><pubDate>Tue, 06 Mar 2012 21:56:21 +0800</pubDate><description><![CDATA[http://www.xperseverance.net/blogs/2012/03/17/


听说国外大牛都认为LDA只是很简单的模型，吾辈一听这话，只能加油了~
另外这个大牛写的LDA导读很不错：http://bbs.byr.cn/#!article/PR_AI/2530?p=1
一、预备知识：
       1. 概率密度和二项分布、多元分布，在这里
       2. 狄利]]></description><category></category></item><item><title><![CDATA[LDA Learning Structures 记录]]></title><link>https://blog.youkuaiyun.com/xpersistence/article/details/7322874</link><guid>https://blog.youkuaiyun.com/xpersistence/article/details/7322874</guid><author>xpersistence</author><pubDate>Mon, 05 Mar 2012 22:36:20 +0800</pubDate><description><![CDATA[这只是一个临时图，未来会慢慢扩充
现在的任务有三个：
1.看LDA论文
2.学习狄利克雷分布，搞懂它和多元分部之间的关系，搞懂LDA模型的各种参数之间关系以及generate的过程
3.递归至Graphical Model学习


2012@3@5



2012@3@7]]></description><category></category></item><item><title><![CDATA[PRML Chapter 8. Graphical Models  -  8.2.1 Three example graphs]]></title><link>https://blog.youkuaiyun.com/xpersistence/article/details/7322839</link><guid>https://blog.youkuaiyun.com/xpersistence/article/details/7322839</guid><author>xpersistence</author><pubDate>Mon, 05 Mar 2012 22:24:39 +0800</pubDate><description><![CDATA[8.2.1 Three example graphs
书中说明了三种基本条件独立的结构
(1) tail-to-tail   (从a，b的角度看，a和b都是c的tail)
如图 Figure 8.15


整个图表示abc的联合概率 p(a,b,c)，此时可以根据图8.15的结构推出 p(a,b,c) = p(a|c)p(b|c)p(c)
如果同时两边对c取边缘概率（marginal]]></description><category></category></item><item><title><![CDATA[总读书记录-3月]]></title><link>https://blog.youkuaiyun.com/xpersistence/article/details/7313020</link><guid>https://blog.youkuaiyun.com/xpersistence/article/details/7313020</guid><author>xpersistence</author><pubDate>Fri, 02 Mar 2012 16:54:17 +0800</pubDate><description><![CDATA[3@2
看点新概念，每天干点初中生干的活还是很爽的，虽然初中的时候看得很苦逼~


3@3
我觉得这样一章一章翻下去可能没有时间，我要来倒推法，看到什么不懂的时候再去看需要看的东西



3@5
昨天玩去了，和小管子唱歌去了。今天一天啥事没干，明天开始看看Graphical Model，同时保持LDA


3@7

昨天看了图模型的独立性和D-separation，今]]></description><category></category></item><item><title><![CDATA[PRML Chapter 2. Probability Distributions]]></title><link>https://blog.youkuaiyun.com/xpersistence/article/details/7311597</link><guid>https://blog.youkuaiyun.com/xpersistence/article/details/7311597</guid><author>xpersistence</author><pubDate>Fri, 02 Mar 2012 11:08:39 +0800</pubDate><description><![CDATA[P68
conjugate priors
      In Bayesian
 probability theory, if the posterior
 distributions p(θ|x)
 are in the same family as the prior
 probability distributionp(θ),
 the prior and posterior a]]></description><category></category></item><item><title><![CDATA[Translation Modeling 学习]]></title><link>https://blog.youkuaiyun.com/xpersistence/article/details/7309020</link><guid>https://blog.youkuaiyun.com/xpersistence/article/details/7309020</guid><author>xpersistence</author><pubDate>Thu, 01 Mar 2012 11:05:26 +0800</pubDate><description><![CDATA[IBM model 3有4个参数 n t p d]]></description><category></category></item><item><title><![CDATA[研究计划]]></title><link>https://blog.youkuaiyun.com/xpersistence/article/details/7306033</link><guid>https://blog.youkuaiyun.com/xpersistence/article/details/7306033</guid><author>xpersistence</author><pubDate>Wed, 29 Feb 2012 14:56:27 +0800</pubDate><description><![CDATA[一、关注的论文方向
a. folksonomies
b. large text classification
c. wikipedia
d. key phrase extraction]]></description><category></category></item><item><title><![CDATA[New Concept English  Unit 1]]></title><link>https://blog.youkuaiyun.com/xpersistence/article/details/7295057</link><guid>https://blog.youkuaiyun.com/xpersistence/article/details/7295057</guid><author>xpersistence</author><pubDate>Sun, 26 Feb 2012 12:01:02 +0800</pubDate><description><![CDATA[虽然看paper挺溜，但是平常的英语已经不行了，从现在开始捡起初中的东西再来学习学习，哈


Lesson 1  A private conversation
turn round   vt.   转过身来
原文里是turn round，而不是turn around，但是词典里貌似一样，不知道是不是因为美式和英式不同
statement  陈述句

Lesson 2  Breakf]]></description><category></category></item><item><title><![CDATA[【COLING'2010】FactRank: RandomWalks on a Web of Facts]]></title><link>https://blog.youkuaiyun.com/xpersistence/article/details/7267844</link><guid>https://blog.youkuaiyun.com/xpersistence/article/details/7267844</guid><author>xpersistence</author><pubDate>Fri, 17 Feb 2012 14:43:28 +0800</pubDate><description><![CDATA[这篇文章的motivation是when two fact instances from two relations share the same value for a shared argument type, then the validity of both facts should be increased. Conversely, we also hypothesize that an]]></description><category></category></item><item><title><![CDATA[What’s the difference between Taxonomies and Ontologies? - Ask Dr. Search]]></title><link>https://blog.youkuaiyun.com/xpersistence/article/details/7247470</link><guid>https://blog.youkuaiyun.com/xpersistence/article/details/7247470</guid><author>xpersistence</author><pubDate>Fri, 10 Feb 2012 09:32:33 +0800</pubDate><description><![CDATA[转载自：http://www.ideaeng.com/taxonomies-ontologies-0602



What’s the difference between Taxonomies and Ontologies? - Ask Dr. Search

Last Updated Jun 2009




A Reader Asks:
What’s the diff]]></description><category></category></item><item><title><![CDATA[PRML Chapter 1. Introduction]]></title><link>https://blog.youkuaiyun.com/xpersistence/article/details/7240635</link><guid>https://blog.youkuaiyun.com/xpersistence/article/details/7240635</guid><author>xpersistence</author><pubDate>Tue, 07 Feb 2012 21:21:02 +0800</pubDate><description><![CDATA[为了防止忘记，要把每章的重要内容都记下来，从第一章开始


P 2
generalization的定义：The ability to categorize correctly new examples that differ from those used for training is known as generalization

P3
1) classification 和]]></description><category></category></item><item><title><![CDATA[PRML 读书记录]]></title><link>https://blog.youkuaiyun.com/xpersistence/article/details/7240603</link><guid>https://blog.youkuaiyun.com/xpersistence/article/details/7240603</guid><author>xpersistence</author><pubDate>Tue, 07 Feb 2012 21:12:18 +0800</pubDate><description><![CDATA[2011-11-21
今天开始啃PRML，啃完之后就NB了，希望能坚持，看多久能啃完，Mark一下为证

2011-11-23
关于 Generative Models & Discriminative Models在P43有非常形象的描述，看完之后就会知道为什么一个估计联合分布，一个估计条件分布，但是我看得还不是非常明白，在这里记录一下

2011-11-24
1.5.5 Loss]]></description><category></category></item><item><title><![CDATA[今日正式开始技术博客，O(∩_∩)O]]></title><link>https://blog.youkuaiyun.com/xpersistence/article/details/6256727</link><guid>https://blog.youkuaiyun.com/xpersistence/article/details/6256727</guid><author>xpersistence</author><pubDate>Thu, 17 Mar 2011 16:37:00 +0800</pubDate><description><![CDATA[希望自己多多努力。]]></description><category></category></item></channel></rss>