摘要
针对贝叶斯推断问题研究了广义Bartlett图的一类新模型。运用无向图的可分解覆盖算法,对长度大于等于4的圈和完全二分图G=K_(m,m)(m大于等于3)进行研究。研究结果表明,长度大于等于4的圈属于广义Bartlett图,完全二分图G=K_(m,m)(m大于等于3)不属于广义Bartlett图。通过引入广义Bartlett图的概念和可分解覆盖算法,避免了在不可分解图上进行贝叶斯推断时模型选择的困难。
A new graphical models-Generalized Bartlett graphs are studied for Bayesian inference problems.The circle with length greater than or equal to 4 and the complete bipartite graph G=Km,m(mgreater than or equal to 3) are studied by the decomposable covering algorithm of undirected graphs.The circle with length greater than or equal to 4 belongs to the generalized Bartlett graph,and the complete bipartite graph G=Km,m(mgreater than or equal to 3) does not belong to the generalized Bartlett graph.By introducing the concept of generalized Bartlett graph and decomposable covering algorithm,avoiding the difficulty of model selection when Bayesian inference is performed on graphs.
出处
《青岛大学学报(自然科学版)》
CAS
2017年第3期11-14,共4页
Journal of Qingdao University(Natural Science Edition)
基金
山东省自然科学基金(批准号:ZR2016AM29)资助