基于限制型粒子群优化的贝叶斯网络结构学习  

Bayesian network structure learning based on restricted particle swarm optimization

作　　者：邸若海[1] 高晓光[1] 

机构地区：[1]西北工业大学电子信息学院,陕西西安710129

出　　处：《系统工程与电子技术》2011年第11期2423-2427,共5页Systems Engineering and Electronics

基　　金：国家自然科学基金(60774064)资助课题

摘　　要：贝叶斯网络结构学习是数据挖掘与知识发现领域的主要研究技术之一,在网络结构的搜索空间相对较大的情况下,已提出的相关算法往往都会存在算法收敛速度慢、学习到的结果准确性较差的缺陷。提出一种信息论结合粒子群优化的算法,利用互信息限制粒子的初始化,使得粒子群优化算法能在较短的时间内收敛,应用ASIA网络作为仿真模型,并与K2算法比较。实验结果表明,提出的算法能够快速、准确地得到贝叶斯网络结构。The Bayesian network structure learning is one of the main research technologies in the field of data mining and knowledge discovery, while the search space of the network structure is relatively bigger, some proposed algorithms have some defects that the convergent speed is slow and the accuracy is poor. A kind of information theory combining particle swarm optimization algorithm is put forward, which uses mutual information to limit particle initialization, and makes the particle swarm optimization algorithm converge in a relatively short period of time, then an ASIA network is applied as the simulation model and the proposed algorithm is compared with K2 algorithm. Experimental results show that the proposed algorithm can rapidly and accurately get Bayesian network structures.

关 键 词：贝叶斯网络 结构学习 互信息 粒子群优化 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]