摘要
DS合成法则计算时间复杂度是影响DS理论广泛应用的主要原因之一,介绍了几种著名的近似算法,并分析了它们的优劣性,指出了这些方法的存在的问题。给出了两种新的仿真近似算法,这两种新方法不需要预先限定焦元的个数,计算过程直观、方便,结果合理。仿真试验结果表明,这两种方法的计算结果和Bayesian方法非常接近,但不存在Bayesian方法的不合理之处。对几种算法作了进一步分析研究,给出了三种算法间的关系。
The computation complexity of the DS combination rule is one of the main reasons impacting the wider use of DS theory.Various approximation algorithms were suggested in order to overcome this difficulty. Most of these approximations aimed at reducing the number of focal elements in the belief functions involved. Two new approximation algorithms were introduced. In these new approximation algorithms, only the singleton subsets of the frame of discernment were aUowed to be focal elements. The result of simulation shows that these new approximation algorithms are very similar to Bayesian approximation, but they overcome the disadvantage of Bayesian approximation.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2007年第16期3660-3663,共4页
Journal of System Simulation
基金
国家自然科学基金重点项目(70631003)
教育部科学技术研究重点项目(107067)
教育部博士点基金(20050359006)
关键词
证据理论
合成法则
近似算法
贝叶斯算法
theory of evidence
combination rule
approximation algorithm
Bayesian algorithm
