Detecting anomalous transients in meertrap data
Loading...
Date
2024
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universty of the Western Cape
Abstract
In an era distinguished by significant technological progress, the prevalence of large and complex datasets characterizes the "big data" era across various disciplines. With improved telescopes being built aimed at generating datasets of unprecedented volumes, there is incredible potential for discovery. The MeerKAT radio telescope in South Africa has proven to be an excellent telescope to search for fast radio transients such as pulsars and fast radio bursts (FRBs). MeerTRAP (more TRAnsients and Pulsars), which commensally uses MeerKAT to search for fast radio transients, detects tens of thousands of candidate objects daily (on average), although the vast majority are not of astrophysical origin. Automated techniques such as machine learning are routinely used to identify targeted astrophysical transients. However, an emerging application of machine learning is to aid the detection of unidentified or rare sources, referred to as anomalies.
Description
Keywords
MeerTRAP, Radio frequency interference, South Africa, Machine learning, Dimension Reduction