Orthogonal projection weights in dimension reduction based on Partial Least Squares

  • Authors:

  • Xue-/Qiang Zeng;Guo-/Zheng Li;Mary Qu Yang;Geng-/Feng Wu;Jack Y. Yang

  • Affiliations:

  • School of Computer Science and Engineering, Shanghai University, Shanghai 200072, PR China/ Computer Center, Nanchang University, Nanchang 330006, China.;Department of Control Science and Engineering, Tongji University, Shanghai 201804, PR China.;National Human Genome Research Institute, National Institutes of Health, US Department of Health and Human Services, Bethesda, MD 20852, USA/ Oak Ridge, D.O.E., USA.;School of Computer Science and Engineering, Shanghai University, Shanghai 200072, PR China.;Department of Radiation Oncology, Massachusetts General Hospital and Harvard Medical School, Harvard University, Boston, Massachusetts 02114, USA

  • Venue:
  • International Journal of Computational Intelligence in Bioinformatics and Systems Biology
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

Dimension reduction is important during the analysis of gene expression microarray data, because the high dimensionality in the data set hurts the generalisation performance of classifiers. Partial Least Squares Based Dimension Reduction (PLSDR) is a frequently used method, since it is specialised in handling high dimensional data set and leads to satisfying classification performance. However, the previous works exist an ambiguous usage of projection weights in PLSDR. To assure the orthogonality of projected components, the usually used project weights are nonorthogonal. Here, we propose to use orthogonal project weights for PLSDR. Experimental results on four microarray data sets show our proposed orthogonal project weights are better than the previous used to help improve the generalisation performance of classifiers.