Efficient learning of Bayesian networks with bounded tree-width

详细信息    查看全文

• 作者：Siqi Nie^a ; ^nies@rpi.edu ; Cassio P. de Campos^b ; ^{c.decampos@qub.ac.uk} ; Qiang Ji^a ; ^{qji@ecse.rpi.edu}
• 关键词：Bayesian network ; Structure learning ; Bounded tree-width ; Hill climbing
• 刊名：International Journal of Approximate Reasoning
• 出版年：2017
• 出版时间：January 2017
• 年：2017
• 卷：80
• 期：Complete
• 页码：412-427
• 全文大小：780 K
• 卷排序：80

文摘

This work presents a novel method for BN structure learning with bounded tree-width. The algorithm is based on a fast bijection between k-trees and the Dandelion codes. The Distance Preferable Sampling is designed to effectively cover the space of k-trees. The probabilistic hill climbing algorithm is used to obtain k-trees of high quality. Extensive experiments indicate the efficiency and effectiveness of the proposed algorithm.