Project title: Galaxy modelling in next generation radio surveys with Machine Learning
PI: Matt Jarvis
Radio astronomy is on the brink of groundbreaking scientific advances with the construction of the global €2 billion project, the Square Kilometre Array (SKA) radio telescope. From understanding the formation of the first light sources, to searching for the origins of life, the SKA will tackle some of the most fundamental scientific questions of our time. However, reaching science goals will be inhibited by a lack software capable of handling large data volumes. Proven to automate and accelerate data analysis, AI can handle large images and extract physical properties from the data itself. AI is an active area of research in radio astronomy, being applied to data reduction and object detection and classification. I will use AI to extract astronomical objects from SKA data at scale, using proven image segmentation methods tailored specifically for astronomical observations. I developed the ML-powered source-finding software, ContinUNet, that has proven highly effective and efficient on SKA precursor data. This Fellowship would enable further innovations: modifying ContinUNet for detection and parameter estimation of HI galaxies as well as continuum sources. With SKA science verification data becoming available in 2026, it is crucial to develop innovative tools now to enable scientific breakthroughs and advance the field.
