Enabling new discoveries in radio data with machine learning

MeerKAT’s incredible sensitivity and resolution has opened up a new window on the universe, leading to an astonishing number of new scientific discoveries. However, as datasets from MeerKAT and other radio telescopes grow in size and complexity, it is becoming increasingly difficult to solve the “needle in a haystack” problem and find rare or new phenomena. In this talk, I will outline Astronomaly, a general anomaly detection framework which combines the experience and intuition of human scientists with the raw processing power of machine learning. I will also highlight some of Astronomaly’s recent discoveries including SAURON, a candidate Odd Radio Circle detected in the MeerKAT Galaxy Cluster Legacy Survey data. The discovery of SAURON, and other unusual objects, in this small but rich dataset demonstrates the potential of machine learning to enable scientific discoveries that might otherwise be missed.