Structure Learning Algorithm Based Bayesian Network

Bayesian network is a design model. It can be used as an effective method to represent causal information. It has been widely used in risk assessment, fault diagnosis, decision-making system, gene sequence analysis and bio-medical and many other fields.In this paper advantages of Bayesian network theory and classical data mining efficiency advantages combined together.Then Bayesian network structure learning and Bayesian Network Classifier and Application can be worked.The detailed information is described below: A new frequent itemsets mining algorithm is proposed. The algorithm significantly reduced the number of data sets traverse. The results show that the efficiency is better than traditional Apriori algorithm in high Mivacurium scattered data set. After the heuristic-based frequent item sets of double counting, a new learning algorithm for Bayesian network is proposed. The results show that the new algorithm is superior in efficiency in scattered data sets with high Mivacurium.By introducing genetic algorithm to construction of Bayesian classifier,we propose a constrained Bayesian network classifier construct algorithm based on genetic algorithm-GBA.This algorithm adopts genetic algorithm to learn structure and reduces the complexity of learning the structure of Bayesian network. As far as this Classifier`s tructure learning, the fitness function based on logarithm likelihood is designed.The code scheme of network structure, and the corresponding genetic operators are designed either. Experimental results show that GBA algorithm performs well when the relationship between attributes of a data set is relatively complicated.A new Bayesian model selection for clustering algorithms is proposed. Model parameters for the learning algorithm, we give two different Bayesian estimation strategies: maximum a posteriori estimation and conditional expectation estimation. After test, the accuracy rate is better than similar algorithms.The traditional model is improved with MCMC theory. And it is used to discusse the principles and steps in detail for dealing with images. Finally, we use new methods to Deal with CT images that is seeds of horse chestnut Experiments show that the new method have the same effect as traditional methods, but with fewer iterations and simple cores of jump.