基于仿真的贝叶斯网络推理  

Bayesian networks Inference Based on Simulation

作　　者：黄建明[1] 

机构地区：[1]昆明理工大学计算中心,昆明650093

出　　处：《微计算机信息》2010年第28期157-159,共3页Control & Automation

摘　　要：网络推理是贝叶斯网络应用的基础。通过实例,分析了几种基于仿真的贝叶斯网络推理算法,指出它们的特点和存在的主要问题。提出一种逻辑采样算法的改进思想,避免了原算法因丢弃样本造成的计算资源浪费,有效提高了算法性能。Inference is foundation of Bayesian networks application.Several algorithms of Bayesian networks inference based on simulation were analyzed by example,and the characteristics and main defects were indicated.This paper presents an improved idea of logic sampling algorithm based on simulation,so computing resources＇ waste caused by original algorithm discards samples which was avoided,to improve the performance of algorithm effectively.

关 键 词：贝叶斯网络 近似推理 随机仿真 逻辑采样 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]