Print  Close

A Wasserstein-type Distance for Gaussian Mixtures on Vector Bundles

Abstract Number:

2545 

Submission Type:

Contributed Abstract 

Contributed Abstract Type:

Paper 

Participants:

Michael Wilson (1), Tom Needham (2), Chiwoo Park (3), Suprateek Kundu (1), Anuj Srivastava (4)

Institutions:

(1) N/A, N/A, (2) Florida State Board of Administration, N/A, (3) Florida A&M - Florida State University College of Engineering, N/A, (4) Florida State University, N/A

Co-Author(s):

Tom Needham  
Florida State Board of Administration
Chiwoo Park  
Florida A&M - Florida State University College of Engineering
Suprateek Kundu  
N/A
Anuj Srivastava  
Florida State University

First Author:

Michael Wilson  
N/A

Presenting Author:

Michael Wilson  
N/A

Abstract Text:

A Wasserstein-type distance is an Optimal Transport Distance where the set of admissible couplings is restricted in some way. Here, we focus on an Optimal Transport Distance between Gaussian Mixtures, where the admissible couplings are restricted to be Gaussian Mixtures as well. Building on previous work, we extend this Wasserstein-type Distance for Gaussian Mixtures to Gaussian Mixtures defined on trivial vector bundles. We present applications of Wasserstein-type distances for Gaussian Mixtures to the shape analysis of recordings of a nano-particle manufacturing process, and to building machine learning classifiers on Diffusion Tensor MRI data.

Keywords:

Optimal Transport|Differential Geometry|Statistical Shape Analysis| | |

Sponsors:

Section on Statistics in Imaging

Tracks:

Brain Imaging

Can this be considered for alternate subtype?

Yes

Are you interested in volunteering to serve as a session chair?

No

I have read and understand that JSM participants must abide by the Participant Guidelines.

Yes

I understand that JSM participants must register and pay the appropriate registration fee by June 1, 2024. The registration fee is non-refundable.

I understand