2018 10-708 (CMU) Probabilistic Graphical Models {Lecture 15} [Mean field Approximation]

 

 

 

 p_bar(x) = p(x,e)

p(x) is p(x|e)

p_bar is hard to estimate,

but p(x) is easy

and p(x) propotion p_bar

 

 

q is easier for sampling, while we cant sample from p

 

 

 

Occam's Razor???