Wide-field astrophysical surveys, big data and AI

We have many leading roles in major international sky survey projects throughout the electromagnetic spectrum, including Euclid,  Vera C,Rubin Observatory, LOFAR, Athena, JWST, ALMA, JCMT and more. These projects often create very large, very complex datasets, approaching the scale and complexity of the “Big” petabyte-scale datasets encountered by terrestrial high-technology industries. The Square Kilometre Array will generate two Exabytes per year, for example.

This has driven an rapid growth in artificial intelligence / machine learning and other data mining technologies in astronomy. We have led the application of many AI techniques to mining giant astronomical datasets, including finding strong gravitational lenses (using convnets), deconvolving and super-resolving far-infrared extragalactic images (using autoencoders), predicting multi-wavelength JWST images of high-redshift galaxies (using generative adversarial networks), and finding star-forming clumps in distant galaxies (using convnets). We collaborate with leading AI/ML computer scientists in the OU School of Computing and Communications, and the OU Knowledge Media Institute.