电子学报 ›› 2001, Vol. 29 ›› Issue (4): 571-573.

• 论文 • 上一篇    下一篇

一种区间型模糊ART神经网络

赵海燕, 张友良, 曹 健   

  1. 南京理工大学制造工程学院CIMS研究所,南京 210094
  • 收稿日期:1999-11-25 修回日期:2000-12-12 出版日期:2001-04-25 发布日期:2001-04-25

An Interval-Valued ART Neural Network

ZHAO Hai-yan, ZHANG You-liang, CAO Jian   

  1. CIMS Institute,Nanjing University of Science & Technology,Nanjing 210094,China
  • Received:1999-11-25 Revised:2000-12-12 Online:2001-04-25 Published:2001-04-25

摘要: 针对分类样本特征数据取值经常为区间值的情况,本文对Carpenter的点值模糊ART进行了改进,提出了一种区间型模糊ART网络.文中阐述了区间型模糊ART网络中区间数的标准化方法、输入矢量相似度的计算方法、域值的检查方法和学习规则.最后,对机械设计中的传动形式选择进行了实例验证.

关键词: ART神经网络, 标准化方法, 相似度计算, 域值检查, 学习规则

Abstract: Because the sample values often have an interval feature.So,in this paper,we improve the ART Neural Network to an Interval-valued ART Neural Network whose inputs of the first layer and weights between the first layer and output layer belong to the interval-valued fuzzy sets I .We proposed the Normative Method,Similar Degree Computing Method,Threshold Examining Method and Learning Rule of the Interval-Valued ART.At last,an example of transmission form choice in the mechanical design is given.This example proves that our Interval-valued ART is effective.

Key words: an interval-valued ART neural network, normative method, similar degree computing method, threshold examining, learning rule

中图分类号: