摘要
贝叶斯网络的结构学习是数据挖掘与知识发现领域的主要研究技术之一,能从大量数据中寻找隐含的概率依赖关系和知识表达模型,对复杂决策任务的建模与求解提供支持,具有重要的研究意义。文章通过分析结构学习方法(K2和MCMC算法)的基本思想,将两种算法的优点和模型平均的思路结合起来,提出一种改进的贝叶斯网络结构学习算法。仿真实验证明该算法解决了K2和MCMC算法的缺陷,可以在无先验知识的情况下以较快的收敛速度获得较正确、稳定的模型结构。
Bayesian networks structure learning is one of main research techniques in the field of data mining and knowledge discovering, which can find underlying probabilistic dependence relationships between variables and knowledge expression model from a great deal of data, and support modeling and resolving for complex decision-making tasks, so that it has an import research signification. According to analyzing classical Structure Learning methods （K2 and MCMC algorithms）, an improved Bayesian networks Structure Learning algorithm was proposed combined with the merits of above two algorithms and the idea of model averaging. Experiment results show that the proposed algorithm can cover shortages of K2 and MCMC algorithms and can quickly achieve a comparative correct and steady model structure without priori knowledge.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2008年第17期4613-4617,共5页
Journal of System Simulation
基金
重庆市科委科技计划攻关重大项目(cstc2006aa7024)
重庆市自然科学基金项目(cstc2006bb2190)。
