Dortmund 2021 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

AKPIK: Arbeitskreis Physik, moderne Informationstechnologie und Künstliche Intelligenz

AKPIK 2: AKPIK II: Deep Learning

AKPIK 2.2: Vortrag

Mittwoch, 17. März 2021, 16:15–16:30, AKPIKa

Deep Learning Based Analysis Approaches in Radio Interferometry — •Kevin Schmidt, Felix Geyer, Stefan Fröse, and Paul-Simon Blomenkamp — TU Dortmund, Dortmund, Germany

Radio interferometry enables studying our universe at the highest resolutions. The used telescope arrays collect information about the observed sky in Fourier space. Analyzing the measured sample allows the reconstruction of the source images. As the amount of available antennas in a radio interferometer array is limited, the measured Fourier space always remains incomplete. By directly applying the inverse Fourier transformation to the measured data sample, noisy artifacts dominate the reconstructed image.

The radionets project aims to reconstruct the incomplete data samples with Deep Learning based analysis approaches. To train Deep Learning models, suitable Monte Carlo data sets with known ground truths are essential. Therefore, a procedure to simulate observations of radio galaxies with radio interferometers is implemented. This talk gives an overview of the developed simulation chain and the general reconstruction idea.