一种基于SVM的多目标模糊识别方法

A Multi-Target Fuzzy Recognition Method Based on Support Vector Machine

支持矢量机是近年来在统计学习理论的基础上发展起来的一种新的模式识别方法 ,在解决小样本、非线性及高维模式识别问题中表现出许多特有的优势。本文重点分析了支持矢量机多分类问题中存在的错分、拒分现象 ,提出了一种基于支持矢量机特征空间的模糊隶属度函数。多目标识别的仿真结果表明 ,采用这种模糊隶属度函数 ,能够减少目标的错分和拒分数量 ,提高识别率。

Support vector machine(SVM) is a new pattern recognition method developed in recent years on the foundation of statistical learning theory. It wins popularity due to many attractive features and emphatical performances in the fields of nonlinear and high dimensional pattern recognition. The misclassification and the rejective classification problems in multiclass support vector machine are analyzed,and a fuzzy membership function based on SVM feature space is proposed in this paper. Simulation results of multi-target recognition show that it can reduce the number of misclassification and rejective classification targets to use this kind of fuzzy membership function,and improve recognition rate consequently.