The Square Kilometre Array (SKA) will collect more radio data from the Universe than any telescope ever built—and cell phone signals, satellites, and television broadcasts threaten to drown it out. This project trains early-career researchers from Africa to develop new statistical and machine-learning methods that can strip this radio-frequency interference (RFI) from the telescope’s data. Without such methods, the SKA’s multi-petabyte datasets would be too contaminated to use. The research itself focuses on four technical steps—simulation, emulation, data compression, and statistical inference—that together allow astronomers to extract clean signals from noisy raw data. If successful, the work will keep the SKA operational as a tool for fundamental astronomy, helping answer open questions about galaxy formation and cosmology. The skills developed—inference from complex simulations—also transfer directly to climate modelling, epidemiology, and manufacturing, where similar noise-filtering challenges arise. The project will train two dedicated researchers and run a free summer school for 30–40 students across Africa, building lasting scientific capacity on the continent.
View original technical description
The Square Kilometre Array (SKA) is an international project with worldwide participation, to analyse radio signals from the Universe with two very large footprint telescope arrays across 9 African countries and Australia. The SKA will arguably be the largest fundamental science project ever undertaken and will open a new window on the Universe, shedding light on key unsolved problems in astronomy and cosmology. The huge volumes and sensitivity of the data from the SKA present a number of key challenges. One of the most pressing is the contaminating noise from radio-frequency interference (RFI) from the ever-growing number of cell phones, satellites, radio stations and television broadcasts. Efficiently dealing with this RFI at the multi-petabyte scale of the SKA requires rigorous new statistical and computation methods that bridge traditional statistics and cutting-edge machine learning and Artificial Intelligence. It represents a unique opportunity to build scientific capacity in Africa. The proposal is designed to contribute to this, through two main elements: the specialised training of two early-career SKA researchers, one in South Africa and one at Imperial College, focussed on recruiting from Africa; and the broader impact of training around 30-40 students from across Africa at the interface of statistics and artificial intelligence through a dedicated summer school and workshop. We aim to provide this training free to the students. The specialised research project has four main components: simulation, emulation, data compression, and statistical inference. Although this targets the SKA, the skills are broadly applicable to many areas where inference with complex simulations are important, including climate modelling, epidemiology and manufacturing. The proposers are leaders in all of the core method areas and have extensive experience in the training of junior researchers, and are ideally placed to impart this knowledge. In addition, the proposers have a proven track record of working effectively together.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know