Automated probabilistic classification of transients and variables

Abstract

There is an increasing number of large, digital, synoptic sky surveys, in which repeated observations are obtained over large areas of the sky in multiple epochs. Likewise, there is a growth in the number of (often automated or robotic) follow-up facilities with varied capabilities in terms of instruments, depth, cadence, wavelengths, etc., most of which are geared toward some specific astrophysical phenomenon. As the number of detected transient events grows, an automated, probabilistic classification of the detected variables and transients becomes increasingly important, so that an optimal use can be made of follow-up facilities, without unnecessary duplication of effort. We describe a methodology now under development for a prototype event classification system; it involves Bayesian and Machine Learning classifiers, automated incorporation of feedback from follow-up observations, and discriminated or directed follow-up requests. This type of methodology may be essential for the massive synoptic sky surveys in the future.

First Results From the Catalina Real-Time Transient Survey

“…Thus, even limited photometric follow-up of transients from surveys such as LSST can provide very large numbers of these important cosmological distance indicators. However, we are also making steady progress to tackle the classification problem using advanced mathematical and statistical methodologies (see, e.g., Mahabal et al 2008b). As the survey progresses, we will also optimize transient discovery by employing machine learning techniques (Borne 2008).…”

VAST: An ASKAP Survey for Variables and Slow Transients

“…Considerable insight into the classification problem can be gained from the current efforts on classification of transients in optical synoptic sky surveys (Bloom et al 2008;Donalek et al 2008;Mahabal et al 2008aMahabal et al , 2008bMahabal et al , 2010Mahabal et al , 2011Djorgovski et al 2011). We experimented with several approaches, using the data from the Palomar-Quest and Catalina Real-Time Transient Surveys, as well as selected data sets from the literature.…”

The LSST Data Mining Research Agenda

“…This might lead to a petascale data mining compute engine that runs in parallel alongside the data archive -to test every possible N-point correlation, multi-parameter association, and classification rule. In addition to such a "batch discovery machine", a rapid-response data mining engine (i.e., classification broker) is needed in order to produce and distribute scientifically robust near-real-time classifications of astronomical sources, events, objects, or event host objects (e.g., we need the redshift of the host galaxy in order to interpret and classify a supernova accurately) [23,24,25]. These classifications are derived from integrating and mining data, information, and knowledge from multiple distributed VOaccessible data repositories, robotic telescopes, and astronomical alert networks worldwide.…”

“…These classifications are derived from integrating and mining data, information, and knowledge from multiple distributed VOaccessible data repositories, robotic telescopes, and astronomical alert networks worldwide. Incoming event alert data will be subjected to a suite of machine learning (ML) algorithms for event classification, outlier detection, object characterization, and novelty discovery [18,23,24,25,26,27]. Probabilistic ML models will produce rank-ordered lists, to guide follow-up observations on the 10-100K alertable astronomical events that will be identified each night by the LSST sky survey alone.…”