Unsupervised machine learning for transient discovery in deeper, wider, faster light curves. Issue 3 (9th September 2020)

• Record Type: Journal Article
• Authors: Webb, Sara; Lochner, Michelle; Muthukrishna, Daniel; Cooke, Jeff; Flynn, Chris; Mahabal, Ashish; Goode, Simon; Andreoni, Igor; Pritchard, Tyler; Abbott, Timothy M C
• Is Part Of: Monthly notices of the Royal Astronomical Society. Volume 498:Issue 3(2020)
• Page Start: 3077
• Page End: 3094
• Publication Date: 2020-09-09
• Subjects: methods: data analysis -- methods: observational -- techniques: photometric; Astronomy -- Periodicals; Periodicals; 520.5
• DOI: 10.1093/mnras/staa2395
• Languages: English
• ISSNs: 0035-8711

Abstract

ABSTRACT: Identification of anomalous light curves within time-domain surveys is often challenging. In addition, with the growing number of wide-field surveys and the volume of data produced exceeding astronomers' ability for manual evaluation, outlier and anomaly detection is becoming vital for transient science. We present an unsupervised method for transient discovery using a clustering technique and the astronomaly package. As proof of concept, we evaluate 85 553 min-cadenced light curves collected over two ∼1.5 h periods as part of the Deeper, Wider, Faster program, using two different telescope dithering strategies. By combining the clustering technique HDBSCAN with the isolation forest anomaly detection algorithm via the visual interface of astronomaly, we are able to rapidly isolate anomalous sources for further analysis. We successfully recover the known variable sources, across a range of catalogues from within the fields, and find a further seven uncatalogued variables and two stellar flare events, including a rarely observed ultrafast flare (∼5 min) from a likely M-dwarf.