AI techniques for extracting source information from Square Kilometre Array (SKA) datasets

The Square Kilometre Array (SKA) will generate extremely large datasets, containing a very high density of features that must be disentangled to classify the contributing sources and to examine radio propagation effects in the magnetised medium between the sources and the detector. The aim of the project is to develop and implement AI methods for inferring appropriate information from the six-dimensional data, initially using trial data from the SKA data challenges, but then working with real-life data from precursor projects. Attention will be paid to imperfections in the data that may cause systematic errors. Supervised and unsupervised techniques will be explored to optimise the efficiency and yield of object classification, including but not limited to: Convolutional Neural Networks, Gaussian Processes, Genetic Algorithms and Clustering methods.