Enhancing Transient Classification with LSST and 7-Dimensional Telescope: A Machine Learning Approach Using Single Epoch SEDs (Paek)
Type: Talk
Session: Machine Learning and Artificial Intelligence
Author: Gregory Sung Hak Paek
Abstract: The forthcoming the era of Rubin/LSST is poised to detect a vast number of transient alerts including rare events like kilonovae, necessitating follow-up observation to classify and study their embeded physics. Previous initiatives, such as the Photometric Rubin/LSST Astronomical Time-series Classification Challenge (PLAsTiCC), demonstrated robust classification using light curves but faced challenges due to the time delays inherent in collecting multi-epoch data, which can lead to the loss of critical information on transient evolution. This study proposes a machine learning-based classifier utilizing single epoch spectral energy distributions (SEDs) obtained through six optical broadband filters covering wide range of wavelength (ugrizy) in LSST, assuming strategies for deep drilling field and proposed Target-of-Opportunity observations. The classifier will be trained on both observational and simulated data including kilonovae and supernovae to effectively classify the type of transients of interset. Furthermore, we will explore the integration of the 7-Dimensional Telescope (7DT), which employs low-resolution photometric SEDs from 40 medium-band filters within a wide field-of-view. This integration promises to provide deeper insights into not only transient properties and significantly enhance classification accuracy by combining high signal-to-noise photometric data from LSST. This synergistic approach is expected to enhance the efficiency of follow-up observation of transient of interest among alerts from LSST, and consequently minimize the loss of valuable transient evolution data to better understand the physics.