Re: Bayesian Networks and Belief Functions

Bruce D'Ambrosio (dambrosi@CS.ORST.EDU)
Tue, 08 Jun 1999 07:47:56 -0700

Rolf -

Wonderful example, thank you. But I am confused: If Bel is to be a
lower bound, how can you assume A1 and A2 are independent? Shouldn't a
lower bound assume the worst case interaction between A1 and A2? If
not, what does Bel mean? And if we do assume worst case, wouldn't that
be that A1 and A2 are mutually contradictory (P(A1 And A2) = 0)? This
would give a Bel(R) = 0, no?

tnx - Bruce