Speaker: Eve Armstrong

Institution: New York Institute of Technology, Manhattan Campus

Title: “Predicting the Behavior of Sparsely-Sampled Systems Across Astrophysics, Neurobiology, and Epidemiology”

Abstract: Inference is a term that encompasses many techniques including machine learning and statistical data assimilation (SDA). Unlike machine learning, which harnesses predictive power from extremely large data sets, SDA is designed for sparsely sampled systems.  This is the realm of study of any realistic system in nature.  SDA was invented for numerical weather prediction, an inherently nonlinear – and chaotic – problem.  My collaborators and I have taken SDA into new fields, to inform the role of neutrinos in astrophysics, biological neuronal networks, and an epidemiological population model tailored to the coronavirus SARS-CoV-2.  We use SDA to seek solutions that are consistent with both sparse measurements and a partially-known dynamical model of the system from which those measurements arose.  The versatility of SDA across vast disciplines (and vast temporal and spatial scales) shows how these “distinct” environments possess commonalities that can inform one another.

Time/Date: 2:00 pm Thursday 15 Feb 2024

Zoom link: https://duke.zoom.us/j/92085251865?pwd=bG5Cb2ZaY2FHblpXMzlEZC92Q0dBUT09

Remote viewing location: Riddick 415