An Empirical Analysis of Likelihood-Weighting Simulation on a Large, Multiply-Connected Belief Network
This addresses incremental improvements in simulation methods for medical diagnosis, specifically for researchers in probabilistic reasoning.
The study tackled the convergence of likelihood-weighting algorithms on a complex medical belief network, finding that Markov blanket scoring and self-importance sampling significantly improved convergence in two difficult diagnostic cases.
We analyzed the convergence properties of likelihood- weighting algorithms on a two-level, multiply connected, belief-network representation of the QMR knowledge base of internal medicine. Specifically, on two difficult diagnostic cases, we examined the effects of Markov blanket scoring, importance sampling, demonstrating that the Markov blanket scoring and self-importance sampling significantly improve the convergence of the simulation on our model.