The Machine Learning Lab is hosting a series of talks that highlight the diverse applications of machine learning. ML+X seminars will welcome faculty from across UT Austin whose work intersects with machine learning and are held every other Friday during the semester from 3-4 pm CT. These talks will spark engaging conversation and collaboration.

On March 12, we will be joined by Stella Offner, Associate Professor of Astronomy at UT Austin.

Harnessing Machine Learning to Study the Life Cycle of Stars
Speaker: Stella Offner, PhD
Associate Professor of Astronomy at UT Austin

Friday, March 12, 2021
3:00 PM – 4:00 PM CT
Virtual: Register here: https://utexas.qualtrics.com/jfe/form/SV_d76O83GPq8m2qlU

Star formation is messy! It spans many orders of magnitude in spatial scale and involves a variety of interconnected physical processes. Forming stars announce their presence by emitting radiation and ejecting high-velocity material. However, identifying this stellar feedback is challenging, so feedback signatures have traditionally been identified “by eye” — either by astronomers or by citizen scientists. In this talk I will show that supervised convolutional neural networks (CNNs) trained using numerical simulations, provide a more reliable, quantitative and faster alternative to visual searches. I will present the results of our 3D Convolutional Approach to Structure Identification (CASI-3D) method applied to observational data to identify stellar bubbles and outflows. I will show that CASI-3D uncovers more feedback than visual searches and provides robust predictions for physical properties.

Speaker Bio: Stella Offner is an Associate Professor of Astronomy at UT Austin. She received her PhD in Physics from UC Berkeley. She was a NSF Astronomy & Astrophysics Prize Postdoctoral Fellow at the Harvard-Smithsonian Center for Astrophysics and a NASA Hubble Postdoctoral Fellow at Yale. She is a recipient of a NSF Career Award and a Cottrell Scholar Award. Her research focuses on understanding how stars like the Sun form by combining numerical simulations, observations and statistical approaches.