作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2010, Vol. 36 ›› Issue (8): 1-3. doi: 10.3969/j.issn.1000-3428.2010.08.001

• 博士论文 •    下一篇

基于局部相关性的L2Boosting算法

赵秀丽1,赵俊龙2   

  1. (1. 中国人民大学统计学院,北京 100872;2. 北京航空航天大学数学系数学、信息与行为教育部重点实验室,北京 100191)
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2010-04-20 发布日期:2010-04-20

L2Boosting Algorithm Based on Local Correlation

ZHAO Xiu-li1, ZHAO Jun-long2   

  1. (1. School of Statistic, Renmin University of China, Beijing 100872;2. Key Lab of Mathematics, Information and Behavior of Ministry of Education, Department of Mathematics, Beihang University, Beijing 100191)
  • Received:1900-01-01 Revised:1900-01-01 Online:2010-04-20 Published:2010-04-20

摘要: 利用充分降维的思想对L2Boosting算法进行改进,提出基于局部相关性的L2Boosting(LCBoosting)算法。在每次迭代中,该算法根据响应变量与协变量的局部相关性充分提取信息,得到响应变量的线性组合来参与Boosting迭代,无须逐个分析所有变量。模拟结果表明,与L2Boosting算法相比,LCBoosting算法收敛速度快、预测精度高。

关键词: L2Boosting算法, 充分降维, 局部相关性

Abstract: This paper improves L2Boosting algorithm by using the idea of sufficient dimension reduction and proposes a Local Correlation based L2Boosting(LCBoosting) algorithm. In each iteration, LCBoosting algorithm extracts information sufficiently according to the local correlation between response variable and covariant, gets a linear combination of response variable to join the Boosting iteration, and avoids analyzing all the variables one by one. Simulation results show that compared with L2Boosting algorithm, LCBoosting has better performance on the convergence rate and prediction precision.

Key words: L2Boosting algorithm, sufficient dimension reduction, local correlation

中图分类号: