Conquering ASKAP and EMU using unsupervised machine learning

Dr Tim Galvin1



The evolutionary map of the Universe (EMU) is a key science project of the ASKAP instrument Рa radio telescope developed by CSIRO. Once completed it is expected that EMU will detect upwards of 70 million radio objects. Classifying these objects has traditionally been a manual process, where a human would inspect the complex radio morphology and apply a meaningful classification. Machine learning methods will become a necessary  component of data pipelines if we are to successfully keep pace and extract meaningful science from these large surveys. We investigate how rotationally invariant self-organising maps may be exploited to provide a unique tool to (i) classify resolved radio features that are related, (ii) identify the host galaxy producing these radio components, and (iii) detection of anomalous radio sources. By relying on an unsupervised method there is no initial set of training labels required for this approach. We explore our initial results from applying this method on sensitive EMU data.


Tim obtained his PhD from Western Sydney University and commenced his current PostDoctoral position with CSIRO Astronomy and Space Science.



AeRO is the industry association focused on eResearch in Australasia. We play a critical coordination role for our members, who are actively transforming research via Information Technology. Organisations join AeRO to advance their own capabilities and services, to collaborate and to network with peers. AeRO believes researchers and the sector significantly benefit from greater communication, coordination and sharing among the increasingly different and evolving service providers.