Please wait a minute...
高校应用数学学报
高校应用数学学报  2017, Vol. 32 Issue (3): 332-342    
    
分布式$L_{1/2}$正则化
王璞玉1, 张 海1,2∗, 曾锦山3
1. 西北大学数学学院, 陕西西安 710069
2. 中国科学院 数学与系统科学院应用数学所, 北京 100190
3. 江西师范大学 计算机信息工程学院, 江西南昌 330022
The distributed $L_{1/2}$ regularization
WANG Pu-yu1, ZHANG Hai1,2, ZENG Jin-shan3
1. School of Mathematics, Northwest University, Xi’an 710069, China
2. Academy of Mathematics and System Sciences, Chinese Academy of Sciences, Beijing 100190, China
3. College of Computer Information Engineering, Jiangxi Normal University, Nanchang 330022, China
 全文: PDF 
摘要： 研究数据集被分割并存储于不同处理器时的特征提取和变量选择问题, 其中处理器通过某种网络结构相互连接. 提出分布式$L_{1/2}$正则化方法, 基于ADMM 算法给出分布式$L_{1/2}$正则化算法, 证明了算法的收敛性. 算法通过相邻处理器之间完成信息交互, 其变量选择结果与数据集不分割时利用$L_{1/2}$正则化相同. 实验表明, 所提出的新算法有效、实用, 适合于分布式存储数据处理.
关键词： 分布式;  稀疏;  $L_{1/2}$正则化;  ADMM算法    
Abstract: This paper focuses on the feature extraction and variable selection of massive data which is divided and stored in different linked computers, and studies the distributed $L_{1/2}$ regularization. Based on Alternating Direction Method of Multipliers algorithm(ADMM), distributed $L_{1/2}$ regularization algorithm which communicates information between the neighborhood computers has been proposed and the convergence of the algorithm has been proved. The variable selection results of the approach are the same with the entire data set by using $L_{1/2}$ regularization. Numerical studies show that this method is both effective and practical which performs well in distributed data analysis.
Key words: distributed    sparse    $L_{1/2}$ regularization    ADMM algorithm
收稿日期: 2016-09-18 出版日期: 2018-04-07
:  O236  
基金资助: 国家自然科学基金(11571011)
服务　　
把本文推荐给朋友
加入引用管理器
E-mail Alert
RSS
作者相关文章　　
王璞玉
张海
曾锦山

引用本文:

王璞玉 , 张海 , 曾锦山. 分布式$L_{1/2}$正则化[J]. 高校应用数学学报, 2017, 32(3): 332-342.

WANG Pu-yu, ZHANG Hai, ZENG Jin-shan. The distributed $L_{1/2}$ regularization. Applied Mathematics A Journal of Chinese Universities, 2017, 32(3): 332-342.

链接本文:

http://www.zjujournals.com/amjcua/CN/    或    http://www.zjujournals.com/amjcua/CN/Y2017/V32/I3/332

No related articles found!