Skip to main content
U.S. flag

An official website of the United States government

Official websites use .gov
A .gov website belongs to an official government organization in the United States.

Secure .gov websites use HTTPS
A lock ( ) or https:// means you’ve safely connected to the .gov website. Share sensitive information only on official, secure websites.

ACMD Seminar: Representing and Classifying Data with Optimal Transport

Gustavo Kunde Rohde
Professor, Biomedical Engineering and Electrical and Computer Engineering, University of Virginia

Tuesday, February 20, 2024, 3:00-4:00 PM ET (1:00-2:00 PM MT)

A video of this talk will be made available to NIST staff in the Math channel on NISTube, which is accessible from the NIST internal home page. It will be taken down from NISTube after 12 months at which point it can be requested by emailing the ACMD Seminar Chair.

Abstract: Numerous physical phenomena are related to transporting an object, distribution, or quantity over time or space. Think of tracking and recognizing an object under different shifts and poses, brushing ink on paper to write a digit, estimating time delay, frequency or other signal parameters from a traveling wave in Radar, estimating source location from acoustic waves being propagated in air or materials, modeling photon distribution propagation in turbulence, modeling and recognizing cells, tissues, and organs under mass transport phenomena, to name a few. These are all examples of signal and image data analysis problems which must take into account movement and transport phenomena. In this talk I will describe a general purpose classification problem statement for signal and image classes for data emanating from transport processes. We proceed to show such a problem has a closed form (non iterative) global solution that is simple to calculate. The solution is based on a new mathematical representation method (denoted transport transform) that is invertible and greatly simplifies pattern recognition by rendering signal classes emanating from transport processes convex in transform space - yielding simple closed form machine learning algorithms. We compare this new solution to existing machine learning methods (including deep learning) and demonstrate its superiority in terms of accuracy, computational simplicity, robustness, and interpretability on numerous signal and image classification problems.

Bio: Gustavo K. Rohde is a professor of Biomedical Engineering, and Electrical and Computer Engineering at the University of Virginia. He is/has been an editorial board member for the IEEE Transactions on Image Processing, Cytometry part A, BMC Bioinformatics, IEEE Journal of Biomedical and Health Informatics, IEEE Signal Processing Letters, and Sampling Theory, Signal Processing, and Data Analysis. He was program co-chair for IEEE ISBI 21, and regular member of the BDMA NIH study section. He was an ISAC council member from 2014-2016. His research interests are in mathematical modeling, signal/image processing, pattern recognition and machine learning.

Host: Gunay Dogan

Note: This talk will be recorded to provide access to NIST staff and associates who could not be present to the time of the seminar. The recording will be made available in the Math channel on NISTube, which is accessible only on the NIST internal network. This recording could be released to the public through a Freedom of Information Act (FOIA) request. Do not discuss or visually present any sensitive (CUI/PII/BII) material. Ensure that no inappropriate material or any minors are contained within the background of any recording. (To facilitate this, we request that cameras of attendees are muted except when asking questions.)

Note: Visitors from outside NIST must contact Meliza Lane at least 24 hours in advance.

Created January 25, 2024, Updated June 11, 2024