Talk:Graphical model
 | This article is rated Start-class on Wikipedia's content assessment scale. It is of interest to the following WikiProjects: | ||||||||||||||||||||
| |||||||||||||||||||||
.jpg){kind=small}
.svg){kind=small} | Daily pageviews of this article
A graph should have been displayed here but graphs are temporarily disabled. Until they are enabled again, visit the interactive graph at pageviews.wmcloud.org |
First diagram edit
The first diagram shows a cyclic directed PGM. Which kind of PGM is this supposed to be?
This corresponds to none of the described models. (It could be a sort of causal model.). I suggest that we put in a Bayesian Network, then its moral graph (which is an undirected model, namely a Markov random field), it's factor graph and it's clique (join) tree (which is also an undirected model).
Untitled edit
This page is rehashing of bishop's chapter on graphical models. In reality, a graphical model is just a model that consists of a graph. Three models that involve graphs are Markov networks, Bayesian networks, and Factor graphs. I suggest deleting this page, or replacing it with a sentence that says exactly this.MisterSheik 22:44, 27 March 2007 (UTC)
Constraint Networks edit
What about discrete graphical models, like Constraint Networks?128.195.57.188 (talk) 21:08, 7 July 2010 (UTC)
Bayesian Network Example error? edit
Under the Bayesian Network section, it says "If the network structure of the model is a directed acyclic graph, ..." and then later "For example, the graphical model in the Figure shown above ...".
The example model has C <--> D , and which is I believe a cycle, making the graph not a DAG.
(I don't know anything about Bayesian Networks, but the description on this page seems to exclude the cited example?) — Preceding unsigned comment added by 99.163.133.102 (talk) 03:01, 24 July 2013 (UTC)
Solving Graphical Models and Model Selection edit
This article describes the structure of graphic models, but not the numerical simulation. The structure (ie. directed acyclic graph) is useful to a computer scientist, but not so much to the statistician or applied scientist.
"There are many different kinds of graphical models, but the two most popular ones are based on directed acyclic graphs (also called “Bayesian networks”) and on undirected graphs (also called “Markov random fields”)." page 1 http://www.cs.ubc.ca/~murphyk/Software/bnsw.pdf
BTW, my cite is an out of date (2007) article, but it does briefly describes the software and algorithms used to simulate models.
There is also the model selection issue, which is addressed towards the end of the same citation.
"In addition to inference about states and parameters, there is much interest (especially in the systems biology community) in inference about the graph structure itself." page 2-3 http://www.cs.ubc.ca/~murphyk/Software/bnsw.pdf Jim.Callahan,Orlando (talk) 17:37, 24 January 2014 (UTC)