›› 2016, Vol. 29 ›› Issue (6): 37-.

• 论文 • 上一篇    下一篇

相关向量机分类方法的应用研究

顾健   

  1. (上海理工大学 光电信息与计算机工程学院,上海 200093)
  • 出版日期:2016-06-15 发布日期:2016-06-22

Application Research on Classification Method of Relevance Vector Machine

GU Jian   

  1. (School of OpticalElectrical and Computer Engineering,University of Shanghai for Science and Technology, Shanghai 200093, China)
  • Online:2016-06-15 Published:2016-06-22
  • About author:顾健(1991-),男,硕士研究生。研究方向:电力系统安全状态评估。

摘要:

针对相关向量机在训练数据集较大时模型训练的时间复杂度较高这一问题。结合MichalE.Tipping提出的快速边际似然算法,文中建立了相关向量机算法改进前后的电力系统安全状态评估模型。以IEEE 30节点系统测试为例,采用相同的训练、测试数据和相同的Matlab仿真环境进行仿真。通过对比仿真测试结果,发现改进后的算法在具备高精度和稀疏度的同时大幅缩短了运行时间,有效地解决了当训练的数据集较大时时间复杂度高的问题。

关键词: 相关向量机, Matlab, 快速边际似然算法

Abstract:

Aiming at the problem of relevance vector machine that when the training data set is large,the training time of the model becomes more complicated.Bsaed on the Quick Marginal Likelihood Algorithm proposed by MichalE.Tipping.,we establishe the power system security state assessment model of improved and old RVM Algorithm.By comparing the test results,The improved algorithm not only has high accuracy and sparsity but also greatly shortens the running time.Effectively solve the problem that when the training data set is large,the training time of the model becomes more complicated.

Key words: relevance vector machine, Matlab, quick marginal likelihood algorithm

中图分类号: 

  • TP306.1