skip to main
|
skip to sidebar
Feb 20, 2009
Gibbs Sampling 经验
p(x)=p(x,y)/p(y,z)
若 |x| 很小,只需要知道正比的概率 p(x,y)
但这只是一方面。在 Gibbs sampling 中,经常的情况是,p(x,y) 可以算,但很繁琐,而 p(y,z) 有很多因子可以和 p(x,y) 约掉。所以,不一定要约掉 p(y,z)。
即 p(x,y) 有些公共(不变)部分,通过联合 p(y,z) 的推导中消去。
0 comments:
Post a Comment
Newer Post
Older Post
Home
Subscribe to:
Post Comments (Atom)
Labels
algorithms
(7)
android
(3)
blackberry
(22)
blogger
(8)
c/c++
(25)
c#
(8)
computer vision
(30)
cplusplus
(25)
css
(16)
database
(4)
django
(20)
gs2
(3)
html
(10)
IC
(3)
image processing
(7)
java
(5)
javascript
(3)
knowledge
(65)
latex
(15)
lifehacks
(20)
linux
(33)
machine learning
(57)
machine vision
(4)
math
(22)
matlab
(17)
mobile
(5)
note
(9)
opencv
(4)
optimization
(13)
photography
(7)
photoshop
(2)
programming
(6)
python
(64)
R
(4)
resources
(3)
software
(31)
trouble shooting
(69)
tutorial
(83)
video coding
(9)
visual studio
(8)
WebApp
(21)
wxwidgets
(5)
Blog Archive
►
2013
(1)
►
March
(1)
►
2012
(44)
►
December
(6)
►
November
(7)
►
October
(1)
►
September
(2)
►
July
(6)
►
June
(5)
►
May
(5)
►
April
(2)
►
March
(3)
►
February
(3)
►
January
(4)
►
2011
(100)
►
December
(9)
►
November
(4)
►
September
(6)
►
July
(10)
►
June
(31)
►
May
(11)
►
April
(12)
►
March
(2)
►
January
(15)
►
2010
(110)
►
December
(25)
►
November
(15)
►
October
(10)
►
September
(6)
►
August
(6)
►
July
(5)
►
June
(8)
►
May
(7)
►
April
(3)
►
March
(19)
►
February
(1)
►
January
(5)
▼
2009
(222)
►
December
(3)
►
November
(4)
►
October
(8)
►
September
(28)
►
August
(19)
►
July
(10)
►
June
(25)
►
May
(34)
►
April
(40)
►
March
(36)
▼
February
(9)
c# 内存管理
c# 和 c++ 区别
c#
Outlook, google, blackberry calendar, 日历
Firefox 收藏夹导入 IE
collapsed gibbs sampling 的疑问
Gibbs Sampling 经验
时间,时区
Latent dirichlet allocation
►
January
(6)
►
2008
(17)
►
December
(15)
►
November
(2)
My List
My Life
My Space
0 comments:
Post a Comment