Topics in two-sample testing

Abstract/Contents

Abstract

Driven by recent advances in the collection of biological data, many such studies draw from heterogeneous datasources. We develop an idea of Jerome Friedman's to conduct two-sample testing using supervised learning procedures. In special cases, this technique generalizes the randomization t-test, for which an asymptotic normality result is known. Using Stein's method of exchangeable pairs, we produce Berry--Esseen-type bounds for the permutation t-statistic for the purpose of statistical inference. We demonstrate the use of kernel methods in two-sample testing on non-vectorial data (text and images), and apply multiple kernel learning (MKL) to the heterogeneous data domain. We show that these techniques can effectively synthesize signals from multiple datasources and produce interpretable weights that highlight the role of each component.

Description

Type of resource	text
Form	electronic; electronic resource; remote
Extent	1 online resource.
Publication date	2013
Issuance	monographic
Language	English

Creators/Contributors

Associated with	Ray, Nelson C
Associated with	Stanford University, Department of Statistics.
Primary advisor	Friedman, J. H. (Jerome H.)
Primary advisor	Holmes, Susan, 1954-
Thesis advisor	Friedman, J. H. (Jerome H.)
Thesis advisor	Holmes, Susan, 1954-
Thesis advisor	Diaconis, Persi
Thesis advisor	Efron, Bradley
Advisor	Diaconis, Persi
Advisor	Efron, Bradley

Subjects

Genre	Theses

Bibliographic information

Statement of responsibility	Nelson C. Ray.
Note	Submitted to the Department of Statistics.
Thesis	Thesis (Ph.D.)--Stanford University, 2013.
Location	electronic resource

Access conditions

Copyright

© 2013 by Nelson Chan Ray

License

This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).

Also listed in

View in SearchWorks

Loading usage metrics...