Efficient Classification for Ultra High Dimensional Variable Selection Öffentlichkeit

Qu, Kexin (2015)

Permanent URL: https://etd.library.emory.edu/concern/etds/37720d14d?locale=de
Published

Abstract

Rapid advances in technologies have demonstrated great needs for ultra-high dimensional data analysis in neuroimaging studies. Our work is motivated by the Autism Brain Imaging Data Exchange ( ABIDE) study, where scientist are interested to identify important biomarkers for early detection of the autism spectrum disorder ( ASD) using high resolution brain images that include hundreds of thousands voxels. However, most existing methods are not feasible to deal with such problems due to extensive computational cost coming as well model complexity. In our work, we propose a new spatial variable selection screening (SVSS) method which includes two components: 1) independent screening using each voxel as a predicator and 2) search for other predicators among neighbors based on spatial dependence. Our approach is computationally feasible and efficient; and it takes full advantage of using spatial configuration of the predicators without additional effort on building complex models. Applied to the resting state functional magnetic resonance imaging ( R-fMRI) data in the ABIDE study, our methods identify voxel-level imaging biomarkers highly predictive of the ASD. Extensive simulations also show that our method achieve better performance in predication as well as variable selection compared to the widely used SIS method.

Table of Contents

1 Introduction 1

2 Method 4

2.1 Step1:Screening 5

2.2 Step2: Variable Selection Incorporating Spatial Dependence 5

3 Simulation Studies 7

4 Application 8

5 Discussion 9

References 11

Appendices 13

Tables 13

Figures 14

About this Master's Thesis

Rights statement
  • Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.
School
Department
Subfield / Discipline
Degree
Submission
Language
  • English
Research Field
Stichwort
Committee Chair / Thesis Advisor
Committee Members
Partnering Agencies
Zuletzt geändert

Primary PDF

Thumbnail Title Date Uploaded Actions
Efficient Classification for Ultra High Dimensional Variable Selection () 2018-08-28 12:52:20 -0400

Supplemental Files

Thumbnail Title Date Uploaded Actions
3_17.R () 2018-08-28 12:52:34 -0400