"Dirichlet process"의 두 판 사이의 차이
ph
잔글 |
잔글 |
||
| 5번째 줄: | 5번째 줄: | ||
The Dirichlet process can be seen as the infinite-dimensional generalization of the [[Dirichlet distribution]]. | The Dirichlet process can be seen as the infinite-dimensional generalization of the [[Dirichlet distribution]]. | ||
| + | |||
| + | Chinese Restaurant Process의 visualization[http://topicmodels.west.uni-koblenz.de/ckling/tmt/crp.html?parameters=0.5&dp=1#]을 보면 정말 기가 막힌데, 설명도 좋다. | ||
| + | <blockquote>samples (from the base measure H) yields a random sample of the Dirichlet process DP(0.5,H)</blockquote> | ||
2017년 6월 20일 (화) 18:14 판
Almost all is from wiki.
Dirichlet process is a probability distribution whose range is itself a set of probability distributions.
(‘process is a distribution’) distribution의 distribution을 (혼동을 막기 위해) 그냥 process로 쓰는 모양.
The Dirichlet process can be seen as the infinite-dimensional generalization of the Dirichlet distribution.
Chinese Restaurant Process의 visualization[1]을 보면 정말 기가 막힌데, 설명도 좋다.
samples (from the base measure H) yields a random sample of the Dirichlet process DP(0.5,H)