Manifold-valued data analysis of networks and shapes

Tools

Severn, Katie (2020) Manifold-valued data analysis of networks and shapes. PhD thesis, University of Nottingham.

[thumbnail of KatieSevernApprovedThesis.pdf]
Preview
 PDF (Thesis - as examined) - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader
Available under Licence Creative Commons Attribution.
Download (16MB) | Preview

Abstract

This thesis is concerned with the study of manifold-valued data analysis. Manifold-valued data is a type of multivariate data that lies on a manifold as opposed to a Euclidean space. We seek to develop analogue classical multivariate analysis methods, which are appropriate for Euclidean data, for data that lie on particular manifolds. A manifold we particularly focus on is the manifold of graph Laplacians.

Graph Laplacians can represent networks and for the majority of this thesis we focus on the statistical analysis of samples of networks by identifying networks with their graph Laplacian matrices. We develop a general framework for extrinsic statistical analysis of samples of networks by this representation. For the graph Laplacians we define metrics, embeddings, tangent spaces, and a projection from Euclidean space to the space of graph Laplacians. This framework provides a way of computing means, performing principal component analysis and regression, carrying out hypothesis tests, such as for testing for equality of means between two samples of networks, and classifying networks. We will demonstrate these methods on many different network datasets, including networks derived from text and neuroimaging data.

We also briefly consider another well studied type of manifold-valued data, namely shape data, comparing three commonly used tangent coordinates used in shape analysis and explaining the difference between them and why they may not all be suitable to always use.

Item Type: Thesis (University of Nottingham only) (PhD)
Supervisors: Dryden, Ian
Preston, Simon
Keywords: manifold-valued data analysis, multivariate data
Subjects: Q Science > QA Mathematics > QA299 Analysis
Faculties/Schools: UK Campuses > Faculty of Science > School of Mathematical Sciences
Item ID: 65862
Depositing User: Severn, Katie
Date Deposited: 19 Jul 2021 13:26
Last Modified: 19 Jul 2021 13:26
URI: https://eprints.nottingham.ac.uk/id/eprint/65862

Actions (Archive Staff Only)

Edit View Edit View