Bayesian method for learning graphical models with incompletely categorical data | |
Geng, Z ; He, YB ; Wang, XL ; Zhao, Q | |
2003 | |
关键词 | Bayesian learning graphical model Hyper Markov law incomplete data posterior mean CONTINGENCY-TABLES EXPERT-SYSTEMS MISSING DATA PROBABILITIES |
英文摘要 | A Bayesian method for learning probabilities under graphical models with incompletely categorical data is discussed, a partial-augmentation formula and a recursive formula of posterior means for graphical models are proposed. Instead of augmenting incompletely observed data to complete data, this partial-augmentation formula only partially augments incomplete data such that augmented data have a monotone data pattern. This recursive formula describes how posterior means are updated by using incomplete data one at a time. Based on this recursive formula, the EM algorithm can be applied to calculate approximate posterior means. This approximate method can be easily applied to Bayesian learning for graphical models. (C) 2003 Elsevier B.V. All rights reserved.; Computer Science, Interdisciplinary Applications; Statistics & Probability; SCI(E); EI; 0; ARTICLE; 1-2; 175-192; 44 |
语种 | 英语 |
出处 | EI ; SCI |
出版者 | computational statistics data analysis |
内容类型 | 其他 |
源URL | [http://hdl.handle.net/20.500.11897/158010] ![]() |
专题 | 数学科学学院 |
推荐引用方式 GB/T 7714 | Geng, Z,He, YB,Wang, XL,et al. Bayesian method for learning graphical models with incompletely categorical data. 2003-01-01. |
个性服务 |
查看访问统计 |
相关权益政策 |
暂无数据 |
收藏/分享 |
除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。
修改评论