Exponential Family of Distributions
Overview
Murphy Formulation
{% p(\textbf{x}|\theta) = \frac{1}{Z(\theta)} h(\textbf{x}) exp[\theta ^T \phi(\textbf{x})] %}
{% = h(\textbf{x}) exp[\theta ^T \phi(\textbf{x}) - A(\theta)] %}
where
{% Z(\theta) = \int h(\textbf{x})exp[\theta ^T \phi(\textbf{x})] dx %}
and
{% A(\theta) = log Z(\theta) %}
(see Murphy pg. 282)
