改进原型模式重构的协同识别算法及应用  

Arithmetic and Implication of Synergetic Pattern-Recognition Based on Modified Prototype Reconstruction

作　　者：田晓东[1] 刘忠[1] 

机构地区：[1]海军工程大学电子工程学院指挥自动化系,湖北武汉430033

出　　处：《青岛大学学报（工程技术版）》2005年第3期85-89,共5页Journal of Qingdao University(Engineering & Technology Edition)

摘　　要：针对协同学模式识别中原型模式重构问题,通过分析当前协同模式识别算法研究的现状,对基于信息反馈叠加的原型模式重构算法进行了改进,重新设计了反馈样本选择的准则和反馈时机,从而在一定程度上解决了原型模式重构的信息“饱和”问题。将该方法应用于水下目标的识别中,仿真结果表明该算法有效地提高了模式识别的识别率和可靠性,在本文的仿真算例中训练样本识别率从改进前的96%提高到改进后的100%,而测试样本的识别率也从改进前的88%提高到91.7%。而且识别的稳定性有了较大的提高。Aiming at the problem of reconstruction of prototype patterns in synergetic pattern-recognition and analyzing current synergetic pattern-recognition arithmetic, the method of reconstruction of prototype based on information-feedback was improved the criterions of selecting the feedback style and feedback occasion were redesigned. Then to some extended, the information saturation was avoided. This new method was applied to underwater target recognition, the precise and robustness of the recognition were both improved. Simulation results indicated that the advanced method of prototype reconstruction was effective. In this simulation example, the recognition rate of training pattern changed from 96% to 100%, which was caused by the modified arithmetic, at the same time, the recognition rate of testing pattern changed from 88% to 91.7%. Moreover, the reliability of recognition was enhanced.

关 键 词：信息反馈 原型模式 协同模式识别 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]