Research On Bayesian Network Parameter Learning Algorithm Based On DFL

Bayesian network is not sure knowledge representation model proposed after themethod of fuzzy logic, credibility and neural networks method.It is one of the importantways study uncertainty problem. At present, Bayesian network parameters of the learningmethod is mainly to accurate calculation method and the approximate solution. Precisecause data excessive fitting Precise cause data excessive fitting.The complexity of theapproximate method to become a np-hard problem solving processThis paper will dynamic fuzzy logic into bayesian network parameters in study, avoidexcessive in fitting and decrease the learning process complexity has certain effect. Thespecific work the following.(1) nalysis research is bayesian network parameters in the study, the existingproblems of the full understanding of the parameters of the statistical method of study,understanding of the point estimation and interval estimate of the advantages anddisadvantages and related reasonsAnalyse the situation of the bayesian parameter learing;(2) introduction to dynamic fuzzy set the related theory, by using the dynamic fuzzyset said bayesian network nodes in the relevant information meaning, which based ondynamic fuzzy set bayesian network knowledge representation. Objective, to express thereal real world contained in the relevant information dynamic fuzzy data.(3) based on dynamic fuzzy logic reasoning process parameters, this paper analyzesthe documents according to the parameters and more evidence reasoning process, the rulesof matching, before a piece of the membership update after. By the bayesian networkstructure characteristics, through the people’s cognitive process with the method ofreasoning to real world of causality, and combined with examples, it analyzes thefeasibility of the reasoning process(4) is given based on dynamic fuzzy logic bayesian parameter learning algorithm,introduction to dynamic fuzzy evidence bayesian parameters under study confidenceupdate. Through the confidence in a given update it under the dynamic fuzzy evidenceposteriori probability query, complete dynamic fuzzy reasoning problems and ontology bayesian reasoning problems transformation. And through an example to verify thefeasibility of the proposed algorithm(5) the dynamic fuzzy theory into the students’ learning ability and learning situationforecast of reasoning, and through the analysis of experimental results the feasibility of thismethod.The work of this paper, on the one hand, enrich the bayesian parameters the content ofstudy; On the other hand, will be based on dynamic fuzzy logic bayesian parameterlearning algorithm used in intelligent study, and prove the feasibility, to solve the providesa reference for the similar problems...