摘要
为使最小二乘支持向量机的解具有稀疏性,本文提出了一种稀疏解算法-矢量基学习．首先引入基矢量、基矢量集与矢量空间的概念,并分析新样本矢量与矢量空间的夹角,从而推导出该样本是否为基矢量的判断准则．随着新样本的到来,在线判别支持向量,使LS-SVM的支持向量具有稀疏性．提升LS-SVM动态建模的实时性,本文进一步提出用于矢量基学习的增长记忆模式递推公式．仿真分析及水处理厂的应用实例,验证了该方法的可行性和有效性．
To achieve a sparse solution for least squares support vector regression （LS-SVM）, an algorithm called vector base learning （VBL） is proposed in this paper. Firstly, the concepts of base vector （BV）, base vector set （BVS） and vector space are introduced. By calculating the angle between the new sample vector and the vector space, the criteria for determining whether the measurement vector is one of the BVS is then derived. This determination is carried out on-line for the coming new samples. This makes the solutions of LS-SVM having the feature of sparsity. To improve the modeling speed of LS-SVM, a recursive algorithm of increased memory mode for VBL algorithm is also proposed. Finally, simulation analysis and the modeling of a typical plant for water treatment clearly illustrated the validity and feasibility of the presented method.
出处
《控制理论与应用》
EI
CAS
CSCD
北大核心
2007年第1期1-5,共5页
Control Theory & Applications
基金
国家"863"计划资助项目(2003AA412110)
关键词
最小二乘支持向量机
矢量基
稀疏性
增长记忆模式
支持向量
least square support vector machine
vector base
sparsity
increased memory mode
support vector
