HAO Colloquium - Michael Kirk, ASTRA / NASA GSFC

At the Intersection of Machine Learning and Heliophysics

SDO has produced an unprecedented amount of information about the sun’s atmosphere and photospheric magnetic field. Yet this wealth of science data presents its own problem: how do we use the entire 20 PB SDO archive to drive discovery science? In addition, how do we prepare for the next generation of solar imagers such as DKIST?  The still-emerging field of data science can provide exciting new tools for scientists to ask critical questions of the archive such as Artificial Intelligence (AI) and Machine Learning (ML) techniques, but these methods also have significant pitfalls and dead ends. Fundamentally through shaping a science question into a data science structure can yield novel, subtle, and reproduceable results. This talk will explore some of the early victories, regular pitfalls, and a vision into the near future where we will develop our science questions alongside the archives to get the most out of every pixel.

Date and time: 
Wednesday, May 26, 2021 - 2:00pm to 3:00pm