@nanmeng
2016-05-19T12:46:02.000000Z
字数 1871
阅读 1193
notes
Probabilistic_Graphical_Models
Q: how do you parameterize the undirect graphs?
unnormalized measure:
An explanation of the question above is what we shown below:
and are very likely to agree with each other, and and are very likely to agree with each other while and are likely not to agree with each other, thus and are not likely to agree with each other so
edges and each edge has values, so totally the answer is , but in total we have parameters.
Parameters:
: partition function change the unnormalized measure to probability distribution
(Notice: normalized product of factors)
- However, we cannot read the factorization from the graph.
Since sometimes we recompute the relevant features many times, we may not need a model that good. Thus, the correct independent assumptions, add a bunch of edges to capture the correlations.
CRF is just like Gibbs distribution.
partition function that is a function of X, which means for any given X I'm going to have the sum of all the Y's that correspond to that X and then I'm going to construct an additional distribution for Y over Y for any given X.
CRF is related to Logistic Model:
- Thus, the logistic model is a very simple kind model of CRF !!!
Examples:
Theorem:
Converse therom
but only hold for positive distribution . that is
minimal I-map is not the best tools for capturing structure in a distribution.
Perfect Map
Example:
The last one is the I-map but it can only capture one of the independece relationship of original network not all.
Example:
Uniqueness
Example:
Examples:
Example:
Example: