Abstract: Since the first “guest stars” appeared in the night sky thousands of years ago, discoveries of astronomical transients have been largely serendipitous. Now, in the big data era of synoptic surveys, nightly streams of hundreds of thousands (and soon millions) of time-domain alerts require a new stratagem to maximize transient discovery and the study of rare or poorly understood transient phenomena. First, we present a rigorous study into the development of a low-latency algorithm for classification and the construction of the largest low-redshift, multi-band, and multi-survey data release of supernovae (SNe) ever---the first data release of the Young Supernova Experiment (YSE DR1). This comprehensive multi-survey data release combines Pan-STARRS1 griz and Zwicky Transient Facility (ZTF) gr photometry of 1975 transients with host–galaxy associations, redshifts, spectroscopic and/or photometric classifications, and additional data products from 2019 November 24 to 2021 December 20. When validating our photometric classifier on 472 spectroscopically classified YSE DR1 SNe, we achieve 82% accuracy across three SN classes (SNe Ia, II, Ib/Ic) and 90% accuracy across two SN classes (SNe Ia, core-collapse SNe). Next, we develop Lightcurve Anomaly Identification and Similarity Search (LAISS), a pipeline for “automating serendipity” within the nightly ZTF Alert Stream via the ANTARES alert broker. LAISS leverages statistical light curve and host galaxy features to flag interesting and rare astronomical transients in real-time to expand our samples of rare SN classes and transients uncommonly found in particular host galaxy environments. Tuned for high purity such that each transient flagged is a likely bonafide anomaly, the number of candidates tagged per night is low (~1-5 objects), enabling an expert to follow-up a manageable list of transients given limited spectroscopic follow-up resources. Lastly, we demonstrate the power of approximate similarity search methods to find analogs of transients with similar light curve evolution and host galaxy environments. In total, we have reported the discovery of ~1000 transients and have flagged ~100 bonafide anomalous transients. These methods will transition us into the era of the Vera C. Rubin Observatory's Legacy Survey of Space and Time (LSST), systematically processing large volume data streams for low-latency transient discovery, anomaly detection, classification, and similarity searches.
