一种新的FART分类器

提出了一类基于贴近度理论的模糊ART神经网络模型，简称为CBFART（Closeness Based Fuzzy ART）模型，将模糊数学中的贴近度（Closeness)和择近原则（Closest Principle)概念与自适应共振理论（ART）相结合，形成了一种新的网络模型。该模型的学习以匹配－委托循环为特点，网络分类遵循择近原则，补码编码、匹配－委托和快速委托－慢性重编码方案相结合，保证了网络学习的收敛性和稳定性，并可以做到一次性学习、提高了学习速度。文中对高维样本进行分类仿真，给出了仿真结果，分析表明该模型具有良好的聚类特性，能够稳定地对高维样本进行分类。

A New Cluster Based Fuzzy ART Model

A new fuzzy ART neural network model based on closeness theory, called CBFART for short, is introduced in this paper. The new network model is formed by incorporating the two concepts of fuzzy set theory, closeness and closest principles, with adaptive resonance theory (ART). The learning of the model is characterized by matching-consigning cycle, and the classification of patterns in the network conforms to the closest principle. Coding complement, matching-consigning, and fast consigning - slow recoding procedure work together to make sure that learning of the network is converging and stable. The above three elements also make one shot learning practicable, so as to improve the learning speed of the network. The concrete algorithm of the model and the result of simulation are given in the paper, and the analysis shows that the model is good in clustering performance.