Volume 7 Number 1 (Jan. 2012)
Home > Archive > 2012 > Volume 7 Number 1 (Jan. 2012) >
JCP 2012 Vol.7(1): 187-195 ISSN: 1796-203X
doi: 10.4304/jcp.7.1.187-195

Tree Based Orthogonal Least Squares Regression with Repeated Weighted Boosting Search

Lihua Fu1, Hongwei Li1, Meng Zhang2
1School of Mathematics and Physics, China University of Geosciences, Wuhan 430074, China
2Department of Computer Science, Central China Normal University, Wuhan 430079, China


Abstract—Orthogonal Least Squares Regression (OLSR) selects each regressor by repeated weighted boosting search (RWBS). This kind of OLSR is known to be capable of producing a much sparser model than many other kernel methods. With the aid of tree structure search, this paper is to construct an even sparser regression model in the framework of OLSR with RWBS. When RWBS being used to solve the optimization at each regression stage, OLSR is extended by keeping the k ( k >1)excellent regressors, which minimize the modeling MSE, rather than only choose the best one at each iteration. In this way, the next regressor will be searched in k subspaces instead of in only one subspace as the conventional method. Furthermore we propose a subtree search to decrease experimental time complexity, by specifying the total number of children in every tree depth. The new schemes are shown to outperform the traditional method in the applications, such as component detection, sparse representation for ECG signal and 2-d time series modeling. Besides, experimental results also indicate that subtree based algorithm is with much lower time complexity than tree based one.

Index Terms—Orthogonal least squares, repeated weighted boosting search, tree structure search.

[PDF]

Cite: Lihua, Fu, Hongwei Li, Meng Zhang, "Tree Based Orthogonal Least Squares Regression with Repeated Weighted Boosting Search," Journal of Computers vol. 7, no. 1, pp. 187-195, 2012.

General Information

ISSN: 1796-203X
Abbreviated Title: J.Comput.
Frequency: Bimonthly
Editor-in-Chief: Prof. Liansheng Tan
Executive Editor: Ms. Nina Lee
Abstracting/ Indexing: DBLP, EBSCO,  ProQuest, INSPEC, ULRICH's Periodicals Directory, WorldCat,etc
E-mail: jcp@iap.org
  • Nov 14, 2019 News!

    Vol 14, No 11 has been published with online version   [Click]

  • Mar 20, 2020 News!

    Vol 15, No 2 has been published with online version   [Click]

  • Dec 16, 2019 News!

    Vol 14, No 12 has been published with online version   [Click]

  • Sep 16, 2019 News!

    Vol 14, No 9 has been published with online version   [Click]

  • Aug 16, 2019 News!

    Vol 14, No 8 has been published with online version   [Click]

  • Read more>>