Title:  Finding Needles in Wide Field Survey Haystacks with Machine Learning

Abstract: High-energy astrophysics is currently in an unprecedented era with observatories such as Chandra, XMM-Newton, and Fermi all viewing the sky over a large range of energies. Data are continuously being produced in large quantities and this amount will only continue to increase as more sensitive, wide field, multi-wavelength surveys explore the sky (e.g., eROSITA, ZTF, VCRO, SKA). For instance, the number of sources detected by eROSITA already exceeds one million objects and the survey has still not reached its full potential. Unfortunately, the X-ray data alone are often not enough to reliably classify these sources, particularly for the faint sources whose population dominates these catalogs, so additional multi-wavelength data must be used. In order to fully leverage these datasets, we need well tested methods to reliably identify interesting sources (needles) in these survey catalogs (haystacks). In this talk I will discuss our work on building a multi-wavelength machine learning pipeline to classify sources in existing X-ray source catalogs (i.e., the Chandra source Catalog, 4XMM-DR11). I will also discuss upgrades we are making to the pipeline to incorporate new time-domain datasets (e.g., ZTF, TESS) and additional sensitive multi-wavelength surveys (e.g., DECaPS, RACS). Lastly, I will outline some ideas for keeping high energy training datasets (and source catalogs) up to date as we move into an era of rapid discovery of new sources.

Astro Colloquium and 'coffee & cookies' Department gathering (3:30-3:45pm)

Please join in 538 Davey or click the link to join: https://psu.zoom.us/j/92637070419