In this subsection we recast the robust inference problem as a parameter
estimation problem. Consider a transformed Bayesian network
with transparent variables z'_{i}. Each transparent variable
has values 1, 2, ..., |z'_{i}|;. Suppose z'_{i} is a random
variable with distribution theta_{ij} = p(z'_{i} = j).
Call Theta the vector of all theta_{ij}.

Suppose x_{q} is queried; the objective is to find:

_{q} = a | e) = _{Theta} [p(x_{q} = a, e)/p(e)]

To solve the robust inference problem, we must maximize the posterior log-likelihood for Theta:

L(Theta) = _{q} = a, e)/p(e)]
= _{q} = a, e) -

© Fabio Cozman[Send Mail?]

Fri May 30 15:55:18 EDT 1997