自组织特性图形的检测理论与应用

人工神经网络模型已研究多年，已使得在信息处理领域中具有类似人的性能，本文将Ｋｏｈｏｎｅｎ自组织特征映射的学习规则进行了修改，以降低拓扑邻域边界上的模糊性，而后，用它的联想存贮功能可以实现输入统计过程的特征存贮，以达到检测的目的。本文也讨论了多维检测的数学机理，作为它的一个结果，可以得出高精度的检测性能。

The Detection Theory of Self-Organizing Feature Map and Its Application

Artificial neural net models have been studied for years in the hope of achieving human like performance in the field of information processing.In the paper the learning rule of Kohonen self-organizition feature map is modified in order to decrease the fuzziness on the edges of topological neighbours.Then with its associative memory function ,we can realize the memory of the features of the input stochastic process, conseqently the detection can be performed. We also describe the mathematical mechanisms of multi-dimentional detcction, as its result high-accuracy performance can be derived.