Method of Bayesian Network Parameter Learning Base on Improved Artificial Fish Swarm Algorithm

摘要：

Bayesian network is an effective model to solve uncertainty problem. Parameter learning is an important step for building a Bayesian network, and its performance directly affects the networks accuracy. In this paper, artificial fish swarm algorithm is introduced into the parameter learning of Bayesian network composed of Noisy-Or and NoisyAnd nodes, and the global search capability is also improved by genetic algorithm. The experimental results show that the improved artificial fish swarm algorithm can learn the parameter better, with the characteristic of rapid optimization speed, good global convergence and insensitivity to initial value.

关键词： Bayesian network parameter learning Artificial fish swarm algorithm genetic algorithm Noisy-And Noisy-Or

作者: Yan Wang Liguo Zhang

作者单位: Dept. of Computer North China Electric Power University Baoding, China College of Information Science & Technology Agricultural University of Hebei Baoding, China

会议类型: 国际会议

会议名称: 2010 International Conference on Bio-inspried System and Signal Processing(2010 IEEE生物系统与信号处理国际会议 ICBSSP 2010)

会议地点: 厦门

会议语种:英文

页码: 147-149

在线出版日期: 2010-10-26（万方平台首次上网日期，不代表论文的发表时间）