中文摘要：

本文提出一种模糊核超球感知器（FKHP）学习方法,并介绍了一种基于FKHP这种学习方法的模糊分类模型.模型构建的基本思想是首先选择适当的核函数,将训练模式从输入空间映射到高维特征空间;然后,在特征空间中,利用提出的模糊核超球感知器学习算法,为每一类训练模式找一个覆盖该类别的训练模式的超球;将每个超球,看作为一个模糊划分,以超球中心和半径为参数,定义超圆锥体的隶属函数,并为之建立一条IF-THEN分类规则;最后,以超球半径作为规则的调整参数,进行规则的优化调整.本文介绍了模型的结构、分类规则产生算法以及规则的调整策略.

英文摘要：

Fuzzy classification is an important application of fuzzy set theory and has been widely applied in many fields. Fuzzy classification rules are widely considered a well-suited representation of classification knowledge. This paper introduces a fuzzy classification model based on the proposed fuzzy kernel hyperball perceptron （FKHP） learning method. This classification model uses kernel function and perception method to automatically generate fuzzy partition and automatically create fuzzy classification rule. In constructing this model, firstly the patterns in the initial input space are mapped to high dimensional feature space by selecting a suitable kernel function. Then in the feature space, the hyperball which covers all training patterns of a class is founded for every class by the proposed FKHP algorithm. A hyperball is regarded as a fuzzy partition and a hyper-cone membership function is defined regarding the center and radius of the hyperball as parameters. An IF-THEN rule is created for a fuzzy partition, and a hyper-cone membership function is defined for a fuzzy partition. Finally, considering the possibility that each hyperball has folded regions, rules are tuned regarding the hyperball＇ s radius as tuning parameter. The model structure, the rule generated algorithm and the rule tunning policy are introduced in the paper. Since this classification model uses the theory of fuzzy set,kernel method and perception, the learning rate of rules is fast and the astringency of learning is fine, and the rules are strongly interpretable. This classification model is called FKHPBFCM（a FKHP-based fuzzy classification model）. Experiments with the data sets of standard machine leaning database evaluate the performances of this model with comparison to experiment results of the methods of kernel hyperball perceptron（KHP） and support vector machine（SVM）, and experiment results show this model has the faster classification training rate, better astringency and higher recognition rate.