快速自适应共振理论网络  

The Fast Adaptive Resonance Theory Network

作　　者：彭小萍[1,2] 林小竹[1] 王嵩[1] 

机构地区：[1]北京石油化工学院信息工程学院,北京102617 [2]北京化工大学信息技术学院,北京100029

出　　处：《北京石油化工学院学报》2012年第3期17-23,共7页Journal of Beijing Institute of Petrochemical Technology

摘　　要：基于实验心理学理论,按记忆强度分组自适应共振理论网络聚类,将模式识别过程划分成若干按序进行的子过程。通过合理设置网络参数,在保证准确识别的前提下,尽可能地避免了自适应共振网络所固有的遍历匹配运算。实验表明,新网络的模式识别性能与原网络相当,而计算效率却得到了显著提升。In this paper, based on the experimental psychology, the network clusters have been grouped according to their level of memory strength, so the pattern recognition process can be divided into several sub processes that are carried out in sequence. Through a reasonable setting of network parameters, while ensuring the accurate recognition of the premise, the new network avoids the full matching operation inherent in the adaptive resonance networks as far as possible. Experiments show that the new network＇s pattern recognition performance is equivalent to the original network, but its calculating efficiency has been significantly improved.

关 键 词：自适应共振理论 记忆强度 快速 模式识别 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]