Professor Ray Norris1
1Csiro/wsu, Epping, Australia
The majority of major astronomical discoveries have been unexpected rather than resulting from testing a hypothesis. Next-generation astronomical surveys therefore need to plan explicitly for this, if they are to deliver the full scientific return on investment. However, the large data volumes and complexity of next-generation telescopes typically mean that machine learning software must be developed to achieve this, rather than relying on serendipity.
The EMU (Evolutionary Map of the Universe) survey, just starting using the Australian SKA Pathfinder, is the largest ever radio continuum survey, and will detect about 70 million galaxies, compared to the 2.5 million detected over the entire history of radioastronomy. EMU therefore explores an unexplored area of parameter space, and can in principle expect to make unexpected discoveries.
To make these unexpected discoveries, which may represent the most significant scientific return from the survey, requires the development of techniques and software, and data challenges to test them. I describe the methodology and software we are developing to achieve this for EMU.
Ray Norris is an astrophysicist at CSIRO and Western Sydney University. His professional life revolves around the question of figuring out how the Universe evolved from the Big Bang to the galaxies and stars that we see around us today. To achieve this, he leads the international “Evolutionary Map of the Universe” team who use CSIRO’s new ASKAP telescope and innovative “big data” techniques to answer questions like “why do most galaxies have a black hole in their centre, and how does it affect the galaxy’s life-cycle?”. He is particularly interested in using machine learning techniques to extract the science from the data, and his holy grail is to develop a technique for making unexpected discoveries from astronomical data. As well as his mainstream astrophysical research, he is also known for his research on the astronomy of Australian Aboriginal people.