Bayesian analysis in expert systems

摘要: We review recent developments in applying Bayesian probabilistic and statistical ideas to expert systems. Using a real, moderately complex, medical example we illustrate how qualitative quantitative knowledge can be represented within directed graphical model, generally known as belief network this context. Exact inference on individual cases is possible using general propagation procedure. When data series of are available, techniques used for updating the original subjective inputs, present set diagnostics identifying conflicts between prior specification. A model comparison procedure explored, number links made with mainstream methods. Details given use Dirichlet distributions learning about parameters process transforming junction tree basis efficient computation.