Speaker
Description
Machine learning and deep learning-based analysis techniques have recently become prominent in the scientific community. While fields such as particle physics and medical physics already successfully use deep learning methods in their analysis pipelines, these advantages still need to be explored in radio interferometry.
The radionets-project has been working on establishing deep learning-based imaging of radiointerferometric data for five years. Especially for the new generation of radio interferometers such as MeerKat and SKA, deep learning can offer enormous performance gains and improved sensitivities for the existing analysis pipelines. Recent publications have validated the general application of radionets on simulated test data sets for interferometers such as VLA, VLBA, and ALMA. The potential applications for MeerKat are currently being explored, with the goal of training a deep learning model capable of imaging actual observational data. This talk will give an overview of the first reconstructed radio skies, as seen by MeerKat, and an outlook on what deep learning can accomplish to analyze radio interferometer data in the future.
