一个基于相关存储的模式识别方法  

A Method of Pattern Recognition Based on Associative Matrix

作　　者：周景洲[1] 

机构地区：[1]深圳大学信息工程学院,广东深圳518060

出　　处：《电子学报》2002年第12A期2146-2148,共3页Acta Electronica Sinica

摘　　要：应用相关的存储技术实现模式识别可缩减存储大小,但由此带来的饱和问题严重影响系统正常工作,为解决此问题,本文在分析现有系统的基础上给出了基于一系列逻辑操作的识别技术.该技术的实质是用一形式化的方法增加输入模式间的线性无关性.通过论证与实例验证,该方法是有效而实用的.To reduce memory size and to solve the saturation problem arising from using the associative matrix in the pattern recognition process, on the basis of analyzing the conventional system, we introduce some developments based on logical operations. The aim of these operations is to enhance the linear independence between the input patterns. By presenting examples, we can illustrate that these methods are more efficient and pragmatic.

关 键 词：模式识别 神经网络 相关存储 逻辑操作 布尔神经网络 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]