一个基于模糊神经网络的模式分类系统

目前，基于神经网络的分类系统在许多领域得到了越来越广泛的应用。但是，该系统大多采用的是离线自适应机制，即神经网络需学习新的分类知识时，要重新训练神经网络，从而大大增加神经网络的训练时间；对于重叠分类，一般是构成一个贝叶斯分类器。然而，贝叶斯分类器的构成需要关于分类数据的概率密度函数的先验知识，而这些知识常常在模式分类前是难以获得的。为了解决这些问题，文中根据模糊集合理论，提出了一种基于模糊神经网络

A PATTERN CLASSIFICATION SYSTEM BASED ON FUZZY NEURAL NETWORK

At present, pattern classification systems based on neural network are being widely used in many fields. However these systems utilize the off line adaptation, i.e., each time new information is added to the systems, it requires a complete retaining of the systems with both the old and the new information. As such, the off line adaptation can lead to increasingly longer training time. For the overlapping classes, the most prevalent method of minimizing misclassification is the construction of a Bayes classifier. Unfortunately, to build a Bayes classifier requires knowledge of the underlying probability density function for each class. This is the information that is quite unavailable. In order to solve these problems, according to the fuzzy set theory, a kind of pattern classification method based on fuzzy neural network is presented in the paper. This method combines fuzzy logic with neural network. The neural network consisting of different kinds of nerons carries out the logic operations AND, OR and MATCH widely applied in fuzzy sets so as to increase the on line adaptation, overlapping classes, learning efficiency and interpretation ability of neural network. Experiment results show that this method is useful and applicable, and has better classification efficiency and classification availability than other pattern classification methods.