信任函数的逼近

THE IMPROVEMENT OF BELIEF FUNCTIONS' APPROXIMATIONS

为了解决各种不精确概率理论难以实际应用的问题,于是信任函数的逼近被提出和研究。Lowrance,etal和DavidHarmanec分别提出了信任函数概括逼近法和信任函数的五种启发式计算方法,但这些方法都未能兼顾精度要求和计算速度。本文在比较双逼近法和概括逼近法的优缺点的基础上,给出两种即兼顾精度要求又考虑计算时间的逼近方法。

In order to solve the problem that various theories about imprecise probabilities are applied difficultly,Lowrance.et al and David Harmanec presented belief function's summarization and its five heuristic methods respectively.But these methods cannot balance between the demands of accuracy and computing speed.Comparing the advantages and disadvantages of pair approximation and summarization,this article come up with two kinds of methods which cost less time than pair approximation and have higher accuracy than summarization.