摘要
本文基于模糊Hopfield网络研究提出了一种神经元模糊识别系统(Neuro-Fuzzy Recognition System,简称NFR系统,或NFRs)。其核心是(N+1)阶模糊Hopfield网络和NFR聚类核。通过(N+1)阶模糊Hopfield网络中的N阶子网络,对样本模式进行模糊聚类,学习样本模式中隐含的模糊聚类结构知识,形成NFR聚类核。基于NFR聚类核形成的知识结构,(N+1)阶模糊Hopfield网络对由待识别模式和样本模式构成的模式集合进行模糊聚类运算。NFRs可对模式空间的模式进行分类和识别,并依样本模式将其划分为等价类。论文对NFRs的性能进行了理论分析和示例研究,结果显示,NFRs具有良好的特性。
In this paper, a neuro-fuzzy recognition system (NFRs for short) is proposed based on fuzzy Hopfield neural network. The NFRs is mainly composed of a so-called NFR core and a fuzzy Hopfield network of (N + 1) th order. The NFR core is formed by acquiring the cluster structure knowledge along with the clustering of pattern samples on the fuzzy Hopfield network. Based on the NFR core and the fuzzy Hopfield network of (N + 1) th order, clustering operation is performed on the pattern set containing both the input pattern and the samples. The NFRs is therefore able to recognize patterns in the pattern space and classify them into equivalent classes. The NFRs is investigated theoretically and the results and an illustrative example show that the NFRs holds favorable properties.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
2000年第4期387-394,共8页
Pattern Recognition and Artificial Intelligence
基金
国家自然科学基金
北京市自然科学基金
