| The power of archives in the era of machine learning |
| Ashish Mahabal |
|
Caltech, USA
|
| Astronomy archives have been continually growing larger with diverse sky surveys covering different parts of the vast parameter space of observability. Better imaging capabilities have meant that it has become possible to discover fast changing and fast moving objects more easily, as the recent discoveries of FRBs and a Vatira indicate. The drive to these real-time discoveries and required rapid follow-up have in turn resulted in better and well-tuned machinery. We will provide an overview of the Zwicky Transient Facility (ZTF) Data Release 4 (DR4) and show how machine learning can work wonders with real-time data as well as large archives with examples from asteroids, to variables, to extra-galactic objects. While machine learning has flourished spectacularly we will also caution on using it blindly and discuss some precautions to take as we move towards even bigger discovery spaces. |