hello all, in my work i use Bayesian networks to simulate the outcome of some events based on prior events. By simulate i mean i find out the probabilities of the outcomes based on the inputs and then choose randomly one of those outputs. I submitted this to UAI, and got in my review the comment that Bayesian network troubleshooting/diagnosis systems could be used predictively by injecting boundary conditions and randomly generating outputs, and that Monte Carlo diagnostic methods use simulation to achieve abduction. Can anyone point me to some reference on these specific points?
thanks in advance for your time and attention. -Jeff -- Jeferson Valadares Cofounder & Creative Director/Jynx Playware ___________________________________________ http://www.jynx.com.br +55.81.3272-4700x4729/Recife/Brasil
