Journal of Systems Engineering and Electronics ›› 2021, Vol. 32 ›› Issue (4): 854-872.doi: 10.23919/JSEE.2021.000074

• SYSTEMS ENGINEERING • Previous Articles     Next Articles

Causal constraint pruning for exact learning of Bayesian network structure

Xiangyuan TAN1(), Xiaoguang GAO1,*(), Chuchao HE1,2(), Zidong WANG1()   

  1. 1 School of Electronics and Information, Northwestern Polytechnical University, Xi’an 710129, China
    2 School of Electronics and Information Engineering, Xi’an Technological University, Xi’an 710021, China
  • Received:2020-06-24 Online:2021-08-18 Published:2021-09-30
  • Contact: Xiaoguang GAO E-mail:tanxy2017@mail.nwpu.edu.cn;cxg2012@nwpu.edu.cn;xomrssh@163.com;nwpu_wzd@mail.nwpu.edu.cn
  • About author:|TAN Xiangyuan was born in 1995. He received his B.E. degree from Northwestern Polytechnical University, China, 2017. He is now a Ph.D. candidate in the School of Electronics and Information, Northwestern Polytechnical University. His research interest focuses on Bayesian network learning, especially on structure learning. E-mail: tanxy2017@mail.nwpu.edu.cn||GAO Xiaoguang was born in 1957. She received her Ph.D. degree from Northwestern Polytechnical University, Xi’an, China in 1989. She is currently a professor in the School of Electronics and Information, Northwestern Polytechnical University. She is the deputy director of the Automatic Control Specialized Committee of China Ordnance Industry Association, and a specialized committee member of China Aviation Society of Weapon System and Photoelectric Technology of China Astronautical Society. Her research interests include probabilistic graphical models, deep learning, reinforcement learning, advanced control theory and its application in complex systems, attack defense confrontation and effectiveness evaluation of integrated avionics systems, and aviation fire control and operational effectiveness analysis. E-mail: cxg2012@nwpu.edu.cn||HE Chuchao was born in 1992. He received his Ph.D. degree from the Department of System Engineering, Northwestern Polytechnical University, China, 2020. He works in the School of Electronics and?Information?Engineering, Xi’an Technological University. His research interest is Bayesian network learning, especially on structure learning. E-mail: xomrssh@163.com||WANG Zidong was born in 1997. He received his B.E. degree from Northwestern Polytechnical University, China, 2019. He is now a Ph.D. candidate in the School of Electronics and Information, Northwestern Polytechnical University. His research interest is Bayesian network learning, especially on structure learning. E-mail: nwpu_wzd@mail.nwpu.edu.cn
  • Supported by:
    This work was supported by the National Natural Science Foundation of China (61573285)

Abstract:

How to improve the efficiency of exact learning of the Bayesian network structure is a challenging issue. In this paper, four different causal constraints algorithms are added into score calculations to prune possible parent sets, improving state-of-the-art learning algorithms’ efficiency. Experimental results indicate that exact learning algorithms can significantly improve the efficiency with only a slight loss of accuracy. Under causal constraints, these exact learning algorithms can prune about 70% possible parent sets and reduce about 60% running time while only losing no more than 2% accuracy on average. Additionally, with sufficient samples, exact learning algorithms with causal constraints can also obtain the optimal network. In general, adding max-min parents and children constraints has better results in terms of efficiency and accuracy among these four causal constraints algorithms.

Key words: Bayesian network, structure learning, exact learning algorithm, causal constraint