Abstract

Optical surveys are probing our changing sky ever more efficiently, with the Zwicky Transient Facility (ZTF) issuing a million alerts every night. The volume, and more importantly, the rate of these alerts is driving the community to using data science techniques for characterization and categorization. I will describe how we are integrating these into the ANTARES (https://antares.noao.edu/), operating on the real-time alert stream from ZTF . I will describe the infrastructure, machine learning stages and public interface, and how you can use the system for time-domain discovery. But supervised machine learning techniques are only as good as their training sets, which remain woefully biased and incomplete. I will describe the operations of the Young Supernova Experiment (YSE), and how we are using this survey to discover rare and unusual transients early to augment what we know about the time-domain sky to prepare us for the upcoming Large Synoptic Survey Telescope (LSST).