Learning Bayesian networks based on order graph with ancestral constraints
作者：
Highlights：
• An efficient framework is proposed for learning Bayesian Networks with ancestral constraints.
• The new framework pruned the violated nodes and suboptimal structures with the guidance of a novel revenue function.
• The accuracy of the observed network based on the proposed framework has significantly improved against some approximate approaches.
• The time and space complexity of the proposed framework are significantly reduced compared to the state-of-the-art framework.
• The new framework is robust even under the situations where errors exist with domain knowledge.
摘要
•An efficient framework is proposed for learning Bayesian Networks with ancestral constraints.•The new framework pruned the violated nodes and suboptimal structures with the guidance of a novel revenue function.•The accuracy of the observed network based on the proposed framework has significantly improved against some approximate approaches.•The time and space complexity of the proposed framework are significantly reduced compared to the state-of-the-art framework.•The new framework is robust even under the situations where errors exist with domain knowledge.
论文关键词：Bayesian network,Structure learning,Ancestral constraints
论文评审过程：Received 30 May 2020, Revised 24 August 2020, Accepted 7 October 2020, Available online 10 October 2020, Version of Record 13 November 2020.
论文官网地址：https://doi.org/10.1016/j.knosys.2020.106515
