Abstract: Predicting Astronomical Transients
Each night starting in 2023, the Large Synoptic Survey Telescope (LSST) will send out alerts for millions of astrophysical transients, including supernovae, gamma ray bursts, and other rare events. Due to the high volume of alerts and fast-paced nature of transient events, most of these transients will fade away before they can even be classified and observed. Our vision is to predict transient type based on the host galaxy’s properties in order to preemptively assign a transient type upon an initial observation of the LSST, allowing scientists to rapidly prioritize and observe interesting events. We apply machine learning techniques on the largest, most complete database of known transients and corresponding host galaxy features. Our challenges include disparate data, feature measurement errors, severe class imbalance, and data bias.