节点文献

基于模糊规则的插值推理算法综述

Fuzzy Rule-Based Interpolative Reasoning:A Survey

  • 推荐 CAJ下载
  • PDF下载
  • 不支持迅雷等下载工具,请取消加速工具后下载。

【作者】 李方轶李映杨静

【Author】 LI Fang-Yi;LI Ying;YANG Jing;School of Computer Science,Northwestern Polytechnical University;National Engineering Laboratory for Integrated Aero-Space-Ground-Ocean Big Data Application Technology;Shaanxi Provincial Key Laboratory of Speech &Image Information Processing;

【通讯作者】 李映;

【机构】 西北工业大学计算机学院空天地一体化大数据技术及应用国家工程实验室陕西省语音与图像信息处理重点实验室

【摘要】 基于模糊规则的近似推理系统是在模糊集和模糊逻辑理论上建立的,在数学、工程、计算机等科学领域得到了迅速的发展.它作为解决建模和推理中不精确和模糊问题的有效工具,利用模糊if-then规则完成推理任务.传统推理机制只能使用稠密的规则库进行推理,其能获得有效结论的前提是保证任何输入的观测值都能与现有的模糊规则相匹配.模糊规则插值(Fuzzy Rule Interpolation,FRI)技术能够有效地利用不完备(或稀疏)的规则库实现推理,为没有规则匹配的观测值通过规则插值估计出其结论.本文系统地综述了基于模糊规则的插值推理技术,着重介绍基于α-截集和基于中间规则的两大类插值方法,通过代表性算法揭示了这两类方法的基本思想.同时,还简要介绍了模糊插值推理系统的其他方法和规则插值算法的实际应用.此外,本文汇总了11个常用的评价指标,并对典型算法进行了系统的比较和讨论.最后对未来的研究工作进行了展望.

【Abstract】 Fuzzy rule-based approximate reasoning system,established with support of the fuzzy set theory and fuzzy logic,has gained rapid developments in a variety of scientific areas,including mathematics,engineering,and computer science.It works by the use of a set of fuzzy if-then rules,as an effective tool to address the issues of imprecision and vagueness in modelling and reasoning.The conventional reasoning mechanism can only perform inference with dense rule bases where any input observation is able to match the existing fuzzy rules.Fuzzy rule interpolation(FRI)facilitates fuzzy rule-based inference system to make reasoning when incomplete(or sparse)rule bases are available,where an estimation is able to be made by computing an interpolated consequent for the observation which matches no rules.This paper systematically reviews the fuzzy interpolative reasoning technique where FRI is involved.In particular,the existing methodologies of FRI are generically categorised into two groups,which are theα-cut based interpolation and the intermediate rule based interpolation,respectively.The main body of this survey reviews each of the two groups of method,where individual representative approaches are described in detail to demonstrate the basic idea of their implementations of FRI.Also,other alternative FRI methods and practical applications of fuzzy interpolative reasoning systems are briefly outlined.In addition,the most commonly used evaluation criteria over FRI algorithms are collected and presented,supported by the comparison and discussion among the typical methods.This paper finally points out potential work of future study in this area.

【基金】 国家自然科学基金面上项目(61871460);科技部新一代人工智能重大专项(2020AAA0106900);陕西省重点研发计划(2021KWZ03)资助~~
  • 【文献出处】 计算机学报 ,Chinese Journal of Computers , 编辑部邮箱 ,2022年08期
  • 【分类号】TP18
  • 【下载频次】47
节点文献中: