Alert Classification for the ALeRCE Broker System: The Light Curve Classifier

Abstract We present the first version of Automatic Learning for Rapid Classification Events (ALeRCE) broker light curve classifier. ALeRCE is currently processing Zwicky Transient Facility (ZTF) alert stream, in preparation Vera C. Rubin Observatory. The classifier uses variability features computed from ZTF stream and colors obtained AllWISE photometry. apply a balanced random forest algorithm with two-level scheme where top level classifies each source as periodic, stochastic, or transient, bottom further resolves these hierarchical classes among 15 total classes. This corresponds to attempt classify multiple stochastic variables (including core- host-dominated active galactic nuclei, blazars, young stellar objects, cataclysmic variables) addition different periodic transient sources, using real data. created labeled set various public catalogs (such Catalina Surveys Gaia DR2 variable stars catalogs, Million Quasars catalog), we all objects ≥6 g -band r detections (868,371 sources 2020 June 9), providing updated classifications new alerts every day. For obtain macro-averaged precision recall scores 0.96 0.99, respectively, 0.57 0.76, respectively. Updated can be found at Explorer website ( http://alerce.online ).