【摘要】 <正> 1.引言神经网络模拟人脑神经功能,用大量简单关系连接来表示复杂的函数关系,具有很多特殊优点,已在系统建模、模式识别、图像处理、知识发现和控制等领域得到了广泛的应用。但当使用前向神经网络,如多层感知器(MLP)和径向基函数(RBF)网络时,需要依赖一些试探法去选择最优的网络尺寸和参数,此外这些网络在学习后通常是静止的,不具备增量学习能力(在更多还原

【Abstract】 Adaptive resonance theory offers a network solution of the stability-plasticity dilemma that exists in conventional artificial neural networks by simulating perception and cognition of humans. Fuzzy adaptive resonance theory map synthesizes fuzzy set theory and adaptive resonance theory ,holds a number of desirable properties. This paper surveys its development course principre ,application area and the newest results,forecasts the directions of the future development.更多还原