【摘要】 作为解决神经网络学习中"稳定性/可塑性两难问题"的一种尝试,ART神经网络一直备受关注。从最初的仅仅用于处理二值输入的非监督学习网络ART1,到具有有监督学习能力的ARTMAP网络,具有一定模糊逻辑运算能力的FuzzyART网络,再到现在对于ART网络中的各种尝试,ART神经网络不断发展、改进,以便适应不同的应用场合。本文着重介绍了ART网络的基本体系结构与发展历程,对于其应用领域加以概述。更多还原

【Abstract】 As a way to overcome the stability/plasticity dilemma in neural network learning, ART neural networks have been paid much more attention. From the original unsupervised ART1 networks, which can only deal with two-value input, to ARTMAP networks with supervised learning, Fuzzy ART networks with fuzzy logic calculation, to other reformed ART network, ART networks have been developing and transforming to get the application to many fields. This paper is an introduction to the basic architecture, history, and application of ART neural networks.更多还原