支持向量机在模式识别中的应用  

Application of Support Vector Machine(SVM) in Pattern Recognition

作　　者：沈明华[1] 肖立[1] 王飞行[1] 

机构地区：[1]国防科技大学电子科学与工程学院,长沙410073

出　　处：《电讯技术》2006年第4期9-12,共4页Telecommunication Engineering

基　　金：国家自然科学基金资助项目(60572138)

摘　　要：针对传统神经网络存在网络结构难于确定、过学习以及局部极小等问题,研究了基于支持向量机(SVM)的模式识别问题。通过对棋盘这种典型非线性二值问题的分类研究,分析了支持向量机的分类与泛化能力,支持向量机在分类和泛化能力方面远远优于传统神经网络。最后将支持向量机用于对两类飞机目标的分类识别,通过多组蒙特卡罗试验,获得了较好的识别结果。支持向量机在目标识别中有巨大潜力和广阔前景。Aiming at the problems such as difficult determination of net structure, over fitting and local minimization of traditional neural networks, the support vector machine （SVM） applied to pattern recognition is studied. By investigating the chessboard classification, which is typical of nonlinear two - value problem, the generalization ability of SVM is analyzed. SVM is more powerful than traditional neural network in the aspect of classification and generalization. Finally two kinds of airplanes are recognized based on SVM, with many Monte -Carlo experiments good classification results are achieved. SVM has huge potentials and good prospect in the area of target recognition.

关 键 词：模式识别 支持向量机 径向基函数 泛化能力 目标识别 

分 类 号：TN957.5[电子电信—信号与信息处理]