Distance dependent Chinese Restaurant Processes
Here is a note of Distance dependent Chinese Restaurant Processes
文章链接http://pan.baidu.com/s/1dEk7ZA5
1. Distance dependent CRPs
In the traditional CRP ,the probability of a customer sitting at a table is computed from the number of other customers already sitting at that table.
Now we introduce the distance dependent CRP, the seating plan probability is described in terms of the probability of a customer sitting with each of the other customers .
let denote the i th customer assignment ,the index of the customer with whom the i th customer is sitting ,let denote the distance measurement between customers i and j , let D denote the set of all distance measurements between all customers ,and let be a decay function .
Notice that the customer assignments do not depend on other customer assignment , only the distances between customers.
This distribution is determined by the nature of the distance measurements and the decay function .For many sets of distance measurements ,the resulting distribution over partition is no longer exchangeable ;this is an appropriate distribution to use when exchangeability is not a reasonable assumption.
2.The decay function:
In general the decay function mediates how distances between customers affect the resulting distribution over partitions .Function f is non-increasing , takes non-negative finite values ,and satisfies f(∞)=0。 (衰减函数的性质)
3. Sequential CRPs and the traditional CRP
A sequential CRP is constructed by assuming that dij=∞ for those j>i ,and this guarantees that no customer can be assigned to a later customer.And when f(d)=1 for d≠∞ and dij<∞ for j<i, the sequential CRP is can re-express the traditional CRP.
NOTICE : although these models are the same ,the corresponding Gibbs samplers are different .(why ?)
4. Marginal invariance:
The traditional CRP is marginally invariant : Marginalizing over a particular customer gives the same probability distribution as if that customer were not included in the model at all .But the DDCRP does not have this property ,and this paper gives us two example of the relevant property of DDCRPS.
Language modeling : a fully observed model
Mixture modeling: a mixture model
5. Relationship to dependent Dirichlet processes (DDP):(they are both infinite clustering model that models dependencies between the latent component assignments of the data )
The first difference is that the dependent Dirichlet process mixture use the truncations of the stick-breaking representation for approximate posterior inference ,in CONTRAST, the ddCRP mixtures are amenable to Gibbs sampling algorithms . Another difference is that the spirit behind them ,in the DDP, data are drawn from distributions that are similar to distributions of nearby data,and the particular values of the nearby data impose softer constraints than those in the ddCRP.(区分ddCRP与贝叶斯非参数模型)
Distance dependent Chinese Restaurant Processes的更多相关文章
- URAL 1962 In Chinese Restaurant 数学
In Chinese Restaurant 题目连接: http://acm.hust.edu.cn/vjudge/contest/123332#problem/B Description When ...
- Distance Dependent Infinite Latent Feature Model 阅读笔记1
阅读文献:Distance Dependent Infinite Latent Feature Model 作者:Samuel J.Gershman ,Peter I.Frazier ,and Dav ...
- 中国餐馆过程(Chinese restaurant process)
也就是说假设空桌子有a0个人,然后顾客选择桌子的概率和桌子上人数成正比. 性质: 改变用户的排列方式,桌子的排列方式,概率不变换.
- Marginalize
在David M.Blei 的Distance Dependent Chinese Restaurant Processes 中提到:DDCRP 的一个重要性质,也是和dependent DP 的一个 ...
- 100 Most Popular Machine Learning Video Talks
100 Most Popular Machine Learning Video Talks 26971 views, 1:00:45, Gaussian Process Basics, David ...
- ICLR 2013 International Conference on Learning Representations深度学习论文papers
ICLR 2013 International Conference on Learning Representations May 02 - 04, 2013, Scottsdale, Arizon ...
- 关于LDA的文章
转:http://www.zhizhihu.com/html/y2011/3228.html l Theory n Introduction u Unsupervised learning by ...
- Bayesian machine learning
from: http://www.metacademy.org/roadmaps/rgrosse/bayesian_machine_learning Created by: Roger Grosse( ...
- R Language
向量定义:x1 = c(1,2,3); x2 = c(1:100) 类型显示:mode(x1) 向量长度:length(x2) 向量元素显示:x1[c(1,2,3)] 多维向量:multi-dimen ...
随机推荐
- Python 进程、线程、协程的介绍与使用
一.必备的理论基础 二.操作系统发展史 三.进程理论 四.线程理论 五.协程 一.必备的理论基础 操作系统理论: 操作系统是一个协调\管理\控制计算机硬件资源与应用软件资源的控制程序 操作系统的两大功 ...
- Python 元类的基本介绍及使用
一.什么是元类 二.如何使用元类 (一)不依赖class关键字创建一个自定义类 (二)自定义元类控制类的产生 (三)自定义元类控制类的调用 一.什么是元类 在python中,一切皆对象,而对象都是由类 ...
- docker容器内存占用过高(例如mysql)
简介 该文章适用于配置低,特别是内存低的服务器,在用容器部署服务时有可能会因为容器占用内存过高导致服务挂掉时参考解决(不是运行在容器里的话,也是可以修改mysql的配置文件限制内存占用) 最近用doc ...
- git基础命令的使用-附命令汇总
本文从https://www.liaoxuefeng.com/wiki/896043488029600学习,边学习边记录. git命令:创建版本库:git init : 把当前目录变成Git可以管理的 ...
- Python第一天哇
iDLE的清除方法ctrl+: 当然,你首先要把网上百度到那个文件按照步骤加上去啦 我百度的=-=:https://www.cnblogs.com/stuqx/p/7291933.html Pyth ...
- PostgreSQL基础操作
1. 查看版本信息 1.1 查看客户端版本信息 黑窗口中输入:psql --version(有两条横线) 没有配置全局的环境变量时,就只能在PostgreSQL安装目录的bin目录中打开黑窗口执行该命 ...
- 使用vue-baidu-map解析geojson
这是后台给我的gejson: {"type":"FeatureCollection","features":[{"type&quo ...
- Java环境准备
电脑重装系统了,所以需要重新配置环境变量. 首先必备工具:jak.eclipse.maven.tomcat 首先配置Java运行环境. 在系统环境变量中新建变量JAVA_HOME:jdk所在的路径,P ...
- 交换机广播风暴,STP生成树协议,端口聚合
交换机(工作在数据链路层)具有学习功能: 一台刚重启的交换机上的mac地址表为空,根据数据包的来源,目的地来学习MAC地址与端口的映射关系映射关系,对于MAC地址表之中已有的就不管了,对未知端 ...
- 手算CRC及其实现
前言: 这篇文章主要讲的是如何手算CRC以及运用CRC,更侧重方法的步骤,对原理方面不做探讨. 方法也是按照我个人理解的来,所以并不专业~ 一些搬过来的代码我也修改了一下下 如果想了解原理的可参考资料 ...