David Hogg, “Bringing a classical-physics perspective to machine learning (and everything)”
In-person in Clark Hall, Room 110
OR
virtually over Zoom
Join Zoom Meeting:
https://wse.zoom.us/j/97055652302?pwd=dWFUUHRHS1lna2h5K0U1cEt4RDRrQT09
If you would like to meet with David W. Hogg before or after the seminar, please sign up for a time at this link:
David W. Hogg, PhD
Professor
New York University
“Bringing a classical-physics perspective to machine learning (and everything)”
Abstract: Physics was revolutionized in the 20th century after the realization that the laws should be expressed in terms of coordinate-free, geometric objects like vectors and tensors, and that the laws should respect diffeomorphism symmetries. All data are taken using physical measuring devices (such as cameras), so all data are governed by these same principles. I will show that thinking about machine-learning methods as if they were physics problems can have good effects across a wide range of data-analysis tasks and domains, even outside of the natural sciences. I’ll show some toy examples, and describe some open problems. (Work in collaboration with Soledad Villar and others at JHU.)
Biography: David W. Hogg is Professor of Physics and Data Science at New York University, and the Group Leader for Astronomical Data at the Flatiron Institute of the Simons Foundation. He works on engineering, precision measurement, and discovery in astronomical data. Current projects include searching for planets around other stars, mapping the dark matter in the Milky Way, and measuring precisely the cosmological parameters. He helps to operate and calibrate large astronomical projects including the Sloan Digital Sky Surveys and the Terra Hunting Experiment.