一种改进的支持向量机模式分类方法

An ameliorated pattern classification method of SVM

支持向量机(Support Vector Machine,SVM)是在统计学习理论基础上发展起来的一种新的机器学习方法,已成为目前研究的热点,并在模式识别领域有了广泛的应用。首先分析了支持向量机原理,随后引入一种改进的径向基核函数,在此基础上,提出了一种改进核函数的SVM模式分类方法。与基于IRIS数据,进行了计算机仿真实验,与基于模糊k-近邻的模式分类仿真结果比较,结果表明改进的SVM方法分类性能比模糊k-近邻算法(Fuzzyk-Nearest Neighbor,FKNN)的分类性能更好,运算时间更短,更易于实时实现。

Support Vector Machine （SVM） is a new machine learning technique developed based on statistical learning theory, and it is attracting increasing attentions. For machine learning tasks involving pattern classification, SVM has become an increasingly popular tool. The theory of SVM is studied at first, then an ameliorated RBF kernel function is presented, based on which an improved kernel function pattern classification method of SVM is put forward. Finally, simulation is made based on the IRIS data and the result is compared with the pattern classification result of FKNN, which shows that the ameliorated SVM outperforms the FKNN, with shorter operation time and is more suitable for real-time implementation.