海量数据的支持向量机优化挖掘方法  

Support vector machine based optimization mining method of massive data

作　　者：李清霞 

机构地区：[1]广东理工学院,广东肇庆526100

出　　处：《现代电子技术》2018年第6期137-140,共4页Modern Electronics Technique

基　　金：中国教师发展基金会(CTF120715);广东理工学院质量工程项目基金(JXGG2017023);广东理工学院精品资源共享课程项目基金(JPKC2016001)~~

摘　　要：传统支持向量机挖掘方法可以对领域数据实现挖掘,但在复杂多变环境下数据挖掘离散程度较大。提出海量数据的支持向量机优化挖掘方法,构造静态粒子空间,局限海量数据挖掘离散程度,形成小规模的、多簇团的粒子挖掘数据集;将单粒子挖掘数据进行离散性拟合,以多簇团粒子整合离散运算,保证挖掘计算进行周期性运行;对同轨挖掘计算进行条件约束,实现小离散程度的数据挖掘。仿真实验验证结果表明,支持向量机优化挖掘方法在复杂多变环境下具有较高的稳定性,并且挖掘离散度小、挖掘信息精度较高。The traditional data mining method based on support vector machine(SVM)can mine the domain data,but has high data mining dispersion degree in the complex and changeable environment. Therefore,an SVM-based optimization mining method of massive data is proposed to construct the static particle space,limit the data mining discrete degree,and form the small-sized and multi-cluster particle mining data sets. The discrete fitting is carried out for the single-particle mining data,and the multi-cluster particles are integrated for discrete operation to ensure the periodical operation of mining calculation. The conditional constraint is performed for the one-orbit mining calculation to realize the data mining with low discrete degree. The simulation experimental results show that the optimization mining method based on SVM has high stability in the complex and changeable environment,low mining discrete degree and high information mining accuracy.

关 键 词：海量数据 支持向量机 多簇团粒子 数据拟合 整合运算 挖掘离散 优化方法 

分 类 号：TN911-34[电子电信—通信与信息系统] TN913[电子电信—信息与通信工程]