一种基于模糊感知器的模糊神经分类器

A Neuro-fuzzy Classifier Based on Fuzzy Perceptron Model

隶属函数应能客观反映模式的不确定性分布情况 ,模糊神经分类器中为每个输入特征划分相同数量的模糊变量显然是不合理的 .针对这一问题 ,根据模糊积分的思想 ,本文通过分析隶属函数的变化曲线来确定模糊密度 ,以此为依据为输入特征划分合适的模糊变量 。

For a neuro_fuzzy classifier, it's unreasonable to partition the fuzzy sets of the same number for the different input features. In order to address this issue, fuzzy density is used here to partition the input feature reasonably. Iris data sets are used to illustrate the effectiveness of the proposed scheme.