- 1、本文档共9页,可阅读全部内容。
- 2、原创力文档(book118)网站文档一经付费(服务费),不意味着购买了该文档的版权,仅供个人/单位学习、研究之用,不得用于商业用途,未经授权,严禁复制、发行、汇编、翻译或者网络传播等,侵权必究。
- 3、本站所有内容均由合作方或网友上传,本站不对文档的完整性、权威性及其观点立场正确性做任何保证或承诺!文档内容仅供研究参考,付费前请自行鉴别。如您付费,意味着您自己接受本站规则且自行承担风险,本站不退款、不进行额外附加服务;查看《如何避免下载的几个坑》。如果您已付费下载过本站文档,您可以点击 这里二次下载。
- 4、如文档侵犯商业秘密、侵犯著作权、侵犯人身权等,请点击“版权申诉”(推荐),也可以打举报电话:400-050-0827(电话支持时间:9:00-18:30)。
- 5、该文档为VIP文档,如果想要下载,成为VIP会员后,下载免费。
- 6、成为VIP后,下载本文档将扣除1次下载权益。下载后,不支持退款、换文档。如有疑问请联系我们。
- 7、成为VIP后,您将拥有八大权益,权益包括:VIP文档下载权益、阅读免打扰、文档格式转换、高级专利检索、专属身份标志、高级客服、多端互通、版权登记。
- 8、VIP文档为合作方或网友上传,每下载1次, 网站将根据用户上传文档的质量评分、类型等,对文档贡献者给予高额补贴、流量扶持。如果你也想贡献VIP文档。上传文档
查看更多
Computational Sampling: Why, God, Why?
Friday, September 07, 2012
2:13 PM
• How to generate samples from an arbitrary probability distribution is a
significant concern for those who want to use Bayesian methods
• If we want to summarize the posterior density of some relevant quantity,
we have to be able to show and work with it
• This is easy if the posterior is an analytically closed‐form density… but many
Bayesian posteriors are not
○ Significant advantage of Bayesian methods: very complex models can
be specified
○ ...but not as a simple likelihood with a conjugate prior
• We need to find a way to get information about the shape of posterior
densities for which we cant write down a direct function
3 - Bayesian Inference II - Computation Page 1
Monte Carlo Integration
Friday, September 07, 2012
5:02 PM
• Start with a more basic idea: how do we determine:
the area under a density inside of an interval , if we cant analytically
solve the integral (i.e., the cumulative density)?
• One idea: Monte Carlo integration
Draw a lot of samples of from
○
○ Add up the number of samples that are inside of , and divide by
the total number of samples
So… how do we sample from
• ?
• This is the subject of a massive amount of study. We will cover just a few
simple methods.
3 - Bayesian Inference II - Computation Page 2
The Accept‐Reject Sampling Algorithm
Friday, September 07, 2012
5:06 PM
1) Create a proposal density, ; this should be analytically closed (
您可能关注的文档
- A negative report on the `near optimality' of the max-expect文献资料.pdf
- Induced pluripotent stem cells pastpresentand future文献.pdf
- Inexact newton methods文献.pdf
- A Probabilistic and RIPless Theory of Compressed Sensing文献资料.pdf
- Alberto Giubilini and Francesca Minerva - After-birth abortion why should the baby live文献资料.pdf
- An Accelerator for Double Precision Floating Point Operations文献资料.pdf
- An Aloha Protocol for Multihop文献资料.PDF
- AnAnalytical ModelforaGPUArchitecturewithMemorylevel andThreadlevelParallelismAwareness文献资料.pdf
- annotating and searching web tables using entities,types and relationships文献资料.pdf
- jep.25.1.49文献.pdf
文档评论(0)